Large Igneous

Geophysical Monograph Series

including
IUGG Volumes

MauriceEwingVolumes
Mineral Physics
 Geophysical Monograph Series

67 High-PressureResearch:
Applicationto Earthand 83 NonlinearDynamicsand Predictability
of Geophysical
PlanetarySciences(Mineral Physics
Volume3) Yasuhiko Phenomena (IUGG Volume 18) William L Newman,
$yonoandMurli H. Manghnani(Eds.) Andrei Gabrielov, and Donald L. Turcotte(Eds.)
68 MicrowaveRemoteSensingof SeaIce FrankCarsey, I]4 SolarSystem
Plasmas
in SpaceandTime J.Burch,
J.H.
RogerBarry,Josefino
Comiso,D. AndrewRothrock,
Robert Waite,Jr. (Eds.)
Shuchman,W. TerryTucker,WilfordWeeks,and Dale
Winebrenner (Eds.) 85 The PolarOceansandTheir Rolein Shapingthe Global
Environment O. M. Johannessen,R. D. Muench, andS. E.
69 SeaLevelChanges:Determinationand Effects(IUGG Overland (Eds.)
Volume11) P. L. Woodworth,D. T. Pugh,J. G.
DeRonde, R. G. Warrick, and J. Hannah (Eds.) 86 SpacePlasmas:
CouplingBetweenSmallandMedium
ScaleProcessesMahaAshour-Abdalla,
TomChang,and
70 Synthesis
of Results
from ScientificDrillingin the Indian PaulDusenbery(Eds.)
Ocean RobertA. Duncan, David K. Rea, RobertB. Kidd,
UlrichvonRad,andJeffreyK. Weissel(Eds.) 87 The UpperMesosphere and LowerThermosphere:
A
Reviewof Experiment
andTheory R.M. Johnson
andT. L.
71 Mantle Flowand Melt Generationat Mid-OceanRidges Killeen (Eds.)
JasonPhippsMorgan,DonnaK.Blackman, andJohnM. Sinton
(Eds.) 88 ActiveMarginsand MarginalBasinsof the Western
Pacific BrianTaylorandJamesNatland(Eds.)
72 Dynamicsof Earth'sDeep Interiorand EarthRotation
(IUGG Volume12) Jean-Louis LeMou•l, D.E.Staylie, 89 NaturalandAnthropogenic
Influences
in Fluvial
and ThomasHerring(Eds.) GeomorphologyJohnE.Costa,AndrewJ.Miller,Kenneth
W. Potter,and PeterR. Wilcock (Eds.)
73 Environmental Effectson SpacecraftPositioning
and
Trajectories(IUGG Volume13) A. VallanceJones(Ed.) 90 Physics
oftheMagnetopause
PaulSong,
B.U.(•.
Sonnerup,andM.F. Thomsen(Eds.)
74 Evolution of the Earth and Planets (IUGG Volume 14)
E. Takahashi,RaymondJeanloz,and DavidRubie(Eds.) 91 SeafloorHydrothermalSystems: Physical,Chemical,
75 InteractionsBetweenGlobalClimateSubsystems:
The Leg- Biological,and GeologicalInteractions SusanE.
acy of Hann (IUGG Volume15) G.A. McBeanandM. Humphris,RobertA. Zierenberg,LaurenS.Mullineaux,
and Richard E. Thomson (Eds.)
Hantel (Eds.)
76 RelatingGeophysical
Structures
and Processes:
TheJef- 92 MaunaLoaRevealed:
Structure,Composition,
History,
freysVolume(IUGG Volume16) K. Aki and R. and Hazards J. M. Rhodes and John P. Lockwood
Dmowska (Eds.) (Eds.)

77 The MesozoicPacific:Geology,Tectonics,andVolcanism 9:3Cross-Scale

Couplingin SpacePlasmasJamesL. Horwitz,
MalcolmS. Pringle,William W. Sager,WilliamV. Sliter, NagendraSingh,andJamesL. Burch(Eds.)
and Seth Stein (Eds.)
94 Double-Diffusive Convection Alan Brandt and H.J.S.
78 ClimateChangein ContinentalIsotopicRecords P.K. Fernando (Eds.)
Swart, K. C. Lohmann,J. McKenzie, and S. Savin(Eds.)
95 EarthProcesses:
Readingthe IsotopicCode AsishBasu
79 TheTornado:Its Structure,Dynamics,Prediction,and and Stan Hart (Eds.)
Hazards C. Church,D. Burgess,C. Doswell,R. Davies-Jones
(Eds.) 96 SubductionTopto Bottom GrayE.Bebout,David
Scholl,StephenKirby,andJohnPlatt(Eds.)
80 AuroralPlasmaDynamics R. L. Lysak(Ed.)
97 Radiation Belts:Modelsand Standards J.F. Lemaire, D.
81 SolarWind Sourcesof Magnetospheric
Ultra-Low Heynderickx,and D. N. Baker(Eds.)
FrequencyWaves M. J. Engebretson,
K. Takahashi,
and
M. Scholer (Eds.) 98 MagneticStorms BruceT. Tsurutani,
WalterD.
Gonzalez, YohsukeKamide, and John K. Arballo (Eds.)
82 Gravimetry
andSpaceTechniques
Appliedto Geodynamics
andOcean Dynamics(IUGG Volume17) BobE.Schutz, 99 CoronalMassEjections Nancy Crooker,JoAnnJoselyn,
Allen Anderson,ClaudeFroidevaux,and Michael Parke(Eds.) andJoanFeynman
 Geophysical Monograph 100

Large Igneous Provinces

Continental,Oceanic,and Planetary
Flood Volcanism

JohnJ.Mahoney
Millard F. Coffin

Editors

AmericanGeophysical
Publishedunderthe aegis of the AGU BooksBoard

Cover.Mapshowing globaldistribution
oflargeigneousprovinces (areasinbright
red,fromM. Coffin'sglobalcompilation). Imageproduced byAndrewGoodliffe
andFernandoMartinez(SOEST,Universityof Hawai'i) usingGMT (P. Wessel
andW. H. F. Smith,Freesoftwarehelpsmapanddisplaydata,EosTrans.AGU,
72, 441, 1991). The datasetcontainscombinedglobalseafloortopography(W.
H. F. Smith and D. T. Sandwell,Global seafloortopographyfrom satellite
altimetryand shipdepthsoundings, Science,277, 1956-1962,1997) and land
topography derivedfromtheUSGS30 arc-second digitalelevationmodel.

Library of Congress Cataloging-in-Publication Data

Large igneous provinces: continental,oceanic, and planetary flood
volcanism/John J. Mahoney, Millard F. Coffin, editors.
p. cm. -- (Geophysical Monograph ; 100)
Includes bibliographicalreferences.
ISBN 0-87590-082-8
1. Flood basalts. 2. Volcanism. I. Mahoney, John J., 1952-.
II. Coffin, Millard F., 1955-. III. Series.
QE462.B3L37 1997
522' .26--dc21 97-43026
CIP

ISBN 0-87590-082-8
ISSN 0065-8448

Copyright1997 by the AmericanGeophysicalUnion

2000 Florida Avenue, N.W.
Washington,DC 20009

Figures,tables,andshortexcerptsmaybereprintedin scientificbooksandjournals
if the sourceis properlycited.

Authorizationto photocopyitemsfor internalor personaluse,or the internalor

personaluseof specificclients,is grantedby the AmericanGeophysicalUnion
for librariesandotherusersregisteredwith theCopyrightClearanceCenter(CCC)
Transactional ReportingService,providedthatthebasefeeof $1.50percopyplus
$0.35 perpageis paiddirectlyto CCC, 222 RosewoodDr., Danvers,MA 01923.
0065-8448/97/$01.50+0.35.
This consentdoesnot extendto other kinds of copying,suchas copyingfor
creatingnew collectiveworksfor resale.The reproduction of multiplecopiesand
the use of full articlesor the use of extracts,includingfiguresand tables,for
commercialpurposesrequirespermissionfrom AGU.

Printed in the United States of

 CONTENTS

Preface

John]. MahoneyandMillardœ.Coffin ....................................... ix

The Columbia River Flood Basalt Province: Current Status
Peter R. Hooper ............................................... 1

Evolutionof the RedSeaVolcanicMargin,WesternYemen

Martin Menzies,Joel Baker,Gilles Chazot, and Mohamed Al'Kadasi .................... 29

The North AtlanticIgneousProvince

A.D. Saunders, ]. G. Fitton,A. C. Kerr,M. ]. Norry,andR. W. Kent ..................... 45
CretaceousBasaltsin Madagascar
andthe TransitionBetweenPlumeand
ContinentalLithosphereMantleSources
MichaelStorey,JohnJ. Mahoney,andAndrewD. Saunders......................... 95
The Caribbean-ColombianCretaceousIgneousProvince:
The InternalAnatomyof an OceanicPlateau
AndrewC. Kerr,JohnTarney,GiselleF. Marriner,AlvaroNivia,andAndrewD. Saunders ........ 123
RajmahalBasalts,
EasternIndia:MantleSourcesandMelt Distribution at a VolcanicRiftedMargin
W. Kent,A.D. Saunders,P. D. Kempton,and N. C. Ghose ........................ 145
TheOntongJavaPlateau
CiiveR. Neal,JohnJ.Mahoney,LorenW. Kroenke,RobertA. Duncan,andMichaelG. Petterson .... 183
The Paran•-Etendeka Province
David W. Peate .............................................. 217

StratigraphyandAgeof KarooBasalts of Lesothoand Implicationsfor

CorrelationsWithin the KarooIgneousProvince
J. $. Marsh,P. R. Hooper,J. Rehacek,R. A. Duncan,andA. R. Duncan .................. 247
SiberianTraps
Muku! $harma ............................................... 273

Giant RadiatingDykeSwarms:
TheirUsein IdentifyingPre-Mesozoic
LargeIgneousProvinces
andMantlePlumes
Richard E. Ernst and Kenneth L. Buchan .................................. 297

Plume/LithosphereInteractionin the Generationof ContinentalandOceanicFloodBasalts:

Chemicaland IsotopicConstraints
John C. Lassiterand DonaldJ. DePao!o .................................. 335

FloodBasaltsandMagmaticNi, Cu, and PGESulphideMineralization:

Comparative Geochemistry
of the Noril'sk(SiberianTraps)andWestGreenlandSequences
PeerC. Lightfoot
andChrisJ. Hawkesworth ................................ 357
Emplacementof ContinentalFloodBasaltLavaFlows
StephenSelf,ThorvaldurThordarson,
andLaszloKeszthelyi........................ 381
LargeIgneousProvinces:
A PlanetaryPerspective
James W. Head III and Millard F. Coffin ..................................
 PREFACE
Continentalflood basalts,volcanic passivemargins, and
oceanicplateausrepresentthe largestknown volcanic epi-
sodeson ourplanet,yet theyarenot easilyexplainedby plate
tectonics.Indeed,somearelikely to recordperiodswhenthe
outward transferof material and energy from the Earth's
interior operatedin a significantlydifferent mode than at
present.In recentyears,interestin large-scalemafic magma-
tism hassurgedashigh-precision geochronological, detailed
geochemical, andincreasingly sophisticatedgeophysical data
have becomeavailablefor many provinces.However, the
sheer amount of recent material, often in the form of detailed
collaborativeresearchprojects,canoverwhelmnewcomersto
the field andexpertsalike asthe literaturecontinuesto grow
dramatically.The needfor an up-to-datereview volumeon a
sizable subsetof the major continentaland oceanicflood
basaltprovinces,termedlargeigneousprovinces,wasrecog-
nizedby the Commissionon Large-VolumeBasalticProv-
inces (International Association of Volcanology and
Chemistryof the Earth's Interior), and the co-editorswere
chargedwith organizingand implementingsucha volume.
We hopethatthis volumewill be valuableto researchers and
graduate studentsworldwide, particularly to petrologists,
geochemists,geochronologists, geodynamicists, and plate-
tectonics specialists;it may also interest planetologists,
oceanographers, andatmospheric scientists.
Nearly a decadehaspassedsincethe publicationof Conti-
nental Flood Basalts(edited by J.D. Macdougall;Kluwer,
1988), the volumethat comesclosestto beinga predecessor
of this one. Fundamentalchangesin understanding of large
igneousprovinceshaveoccurredin thelasttenyears,includ-
ing (1) the generalacceptance of oceanicplateausand vol-
canic passive margins as the submarinecounterpartsof
continentalflood basalts,(2) the realizationthat the major
large igneousprovincesare amongthe very largestigneous
eventson Earth in the last severalhundredmillion years,(3)
the wide applicationof the starting-plume hypothesisfor the
originof manylargeigneousprovinces,(4) a growingunder-
standingof therole suchprovincesplayin continentalgrowth,
and (5) the recognitionthat similareventshavebeenimpor-
tant on the other terrestrialplanets.More recent data on
continentalflood basaltprovincesandthe growingdatabase
for volcanicpassivemarginsandoceanicplateausemphasize
theneedfor a new bookprovidinga broadperspective on the
stateof knowledgein thisfield.
Syntheses of recentand earlierwork, combinedwith new
resultsandinterpretations, arepresented herefor manyof the
major provinces.Most of thesechaptersincludeup-to- date
reviewsof geologicsetting,age and agedistribution,petrol-
ogyandgeochemistry, petrogenesis,
mantlesources, probable
causes,and post-emplacement evolution. Several chapters
include discussionsof possibly related climatic, oceano-
graphic,or biosphericeffects,as well as effectson crustal
evolution.Readerswill quicklyseethat"onefits all" doesnot
applyto largeigneousprovinces,at leastat our currentlevel
of understanding.For example,althoughmost authorsnow
agreeon the importanceof plumeinvolvementin the larger
provinces,theroleinferredfor theplumecanbe quitediffer-
ent,evenwithindifferentpartsof a singleprovince(e.g.,from
beingthe principalsourceof magmato a heatsourcedriving
lithosphericmelting to the "carder"of a halo of entrained
mantlebroadlylike that feedingmid-oceanridges).
Otherpapersfocuson the extensivedike swarmsthatmay
be the only remainingrecordof older flood basaltevents
going back well into the Precambrian,on the geochemical
differences(and their possiblecauses)betweencontinental
and oceanic flood basalts, on sulfide mineralization in flood
basaltsystems,on large igneousprovinceselsewherein the
solarsystem,andon the physicalmechanismof flood-basalt
lava flow emplacement, whichrecentwork suggests may be
quitedifferentthanthoughtpreviously.
Numerouscurrentand plannedresearchprojectspromise
thatlargeigneousprovinceswill remainan activeandfruitful
areaof scientificendeavorwell into the next century.As the
work presentedin this volumedemonstrates, studiesof con-
tinentalandoceaniclargeigneousprovincesarecomplemen-
tary: on the continents,abundantsamplesare availablefor
petrological,geochemical,geochronological, volcanological,
andotherstudies,whereasin theoceanssamplingis difficult
but the lithosphericsettingis relativelyeasyto characterize
and geophysicalimaging is quite effective. Nevertheless,
study of flood basalt magmatismremainsvery much at a
reconnaissance stagein comparisonto suchphenomenaas
mid-oceanridgeandarcmagmatism.Giventhepotentialthat
largeigneousprovinceshavefor illuminatingmantledynam-
ics,aswell asfor effectingglobalenvironmental change,we
hope that the paperspresentedhere will stimulatenew re-
search endeavors.
We are grateful to the many reviewers (see acknow-
ledgmentsin individual papersfor names) who provided
prompt,in-depthcritical evaluationsof the papers.The sup-
port of the InternationalAssociationof Volcanology and
Chemistryof the Earth's Interior (IAVCEI), in
Wally Johnson,is appreciatedin producingthisvolume.We tions.Without their cheerfulexpertise,both of us would no
also wish to thank all the lead authorsfor harnessingthe doubthavelong sincewiggled out of editingthisbook.
intellect and energy of their co-authors,and for (mostly)
meetingthe deadlinesimposedby stemeditors.We are par- JohnJ. Mahoney
Universityof Hawaii
ticularlyindebtedto Diane Hendersonandthe SOEST Publi-
cationsstafffor their invaluableassistance,
includingseveral Millard F. Coffin
roundsof copy editing,formatting,and advicewith illustra- University of Texasat
 The Columbia River Flood Basalt Province: Current Status

Peter R. Hooper

Departmentof Geology,Washington
State University,Pullman, Washington

The ColumbiaRiver flood basaltprovinceis smallerby an order of magnitude

than the Deccan,Karoo,ParanS,and Siberiancontinentalflood basaltprovinces.
Its smaller size, relative youth (17-6 Ma), excellent exposure, and easy
accessibilityhave alloweddevelopmentof a flow-by-flow stratigraphyin which
many flows can be tracedacrossthe ColumbiaPlateau,often linked directly to
their strongly oriented feeder dikes in the southeastquadrant. The detailed
stratigraphyprovidesa preciserecordof the changesin magmacompositionand
volume with time and demonstratesmore clearly here than in other provinces
that singlefissureeruptionshad volumesin excessof 2,000 km3 and flowed
acrossthe plateau for distancesup to 600 km with negligible changesin
chemicalor mineralogicalcomposition.
Currentevidencesuggeststhat the ColumbiaRiver flood basaltsresultedfrom
impingementof a small mantleplume,the Yellowstonehotspot,on the baseof
the lithospherenear the Nevada-Oregon-Idaho border at 16.5 Ma and that the
main focus of eruptionthen moved rapidly north to the Washington-Oregon-
Idaho border from where the main eruptionsoccurred.The rapid northerly
translationof the main eruptiveactivity may have beencontrolledby weakened
or thinnedzonesin the lithosphere.The few earliestflows have typical mantle
plume compositionsand the last, small-volumeflows are contaminatedby
continental crest. In between, the great majority of flows carry a strong
lithosphericsignature,the sourceof which remains controversial-•eitheran
enrichedcontinentallithosphericmantle or assimilatedcontinentalcrest. The
physicalnatureand rate of magmaeruptionare also controversial.Recentwork
suggestsflows grew by internalinjectionratherthan by turbulentsurfaceflow
and this has been used to imply significantly lower eruption rates than
previouslyenvisaged.However,the chemicaland mineralogicalhomogeneityof
singleColumbiaRiver basaltflows acrossmany hundredsof kilometersimplies
that eruptionandflow rateswere still exceptionallyhigh.

INTRODUCTION

LargeIgneousProvinces:Continental,Oceanic,andPlanetary The ColumbiaRiver BasaltGroup (CRBG) forms a large

Flood Volcanism intermontaneplateau of 164,000 km2 lying between the
GeophysicalMonograph100 CascadeRange and the Rocky Mountains in southeast
Copyright1997 by the AmericanGeophysicalUnion Washington, west-central Idaho, and northeast
2 COLUMBIA RIVER FLOOD BASALT PROVINCE

(Figure 1). Basalt flows dip gently west (<2ø) in the
northernpart and are deformedin the southby mild east-
west folds (includingthe Yakima fold belt on the west side
of the PascoBasin) which are associatedwith WNW-ESE
andNNE-SSW strike-slipfaults.
The plateau contains more than 300 individual basalt
flows with an averagevolumeof 500-600 km3per flow and
a total volume of about 175,000 km3 [Tolan et al., 1989].
The earliest eruptionsof CRBG-related tholeiitesmay be
those of the Steens Basalt close to the Nevada/Oregon

.......... 120ø...... 115 ø

i i

! I
i i
48 ø
i
I
I I
i ß

Washington

46 ø

Portland ..

John.Day
' Basin

44 ø
Malheur

Boise

0
I 7151510
km Oregon . SteensMt.
i •
50 100
mi [
42ø [Idaho
_ _ _

Figure 1o Map of the ColumbiaPlateau,showingthelimitsof the ColumbiaRiverbasaltsascurrentlyrecognized

and
thefeederdikesystems. CJmarkstheouterboundaries of theChiefJosephdikeswarmfeedingtheClarkston (Imnaha,
GrandeRondeandWanapum)andSaddleMountainsBasalt.The GrandeRonde(GR) andCornucopia (C) swarmsare
dikeconcentrationswithintheCJ.M istheMonumentdikeswarmwhichfedthePictureGorgeBasaltin theJohnDay
basin.BMU is theBlue Mountains
 HOOPER 3

border which erupted about 17.0-16.5 Ma [Rytuba and plume-related CRBG magmatism south of the Columbia
McKee, 1984; Carlson and Hart, 1987; Swisheret at., 1990; Plateau suggeststhat the model of White and McKenzie
Lees, 1994;Zobacket at., 1994; Camp, 1995;Hooper et at., [1989, 1995], in which both a mantle plume and active
1995c]. If so, the eruptionsthenmovedrapidlynorthward, lithospheric extension are required to provide the
the main volume of magmaeruptingbetween16.5 and 15.5 exceptionallylargemelt volumes,needssomemodification.
Ma from stronglyNNW-oriented fissures(Chief Joseph Current evidence suggeststhat mantle plumes are the
dike swarm) which crossed the eastern end of the primary causeof continentalflood basalteruptions,abetted
Washington-Oregon border (Figure 1). Magmatic activity by the right tectonicenvironmentin which weakenedzones
then decreased,finally endingaround6.0 Ma [Totanet at., or thinspotsin the lithosphere[Thompsonand Gibson,
1989; Baksi, 1989] (Figure 2). 1991] are available to increasethe melt volume. The form
After pioneermappingat the turnof the century[Russell, and locationof the lithosphericthinspotsmay determinethe
1893, 1901; Merriam, 1901; Smith, 1901, 1903a,b; exact location of the eruptions,sometimesfunneling the
Lindgren, 1901], progresswas hamperedby the apparent magmaaway from the centerof the underlyingplume.The
similarityof the manyflowsexposedin the deepcanyonsof lithosphericextensiontypically associatedwith flood basalt
the Columbia, Snake, and Salmon Rivers. Washington eruptions,and which sometimesleads to plate separation,
[ 1922] analyzedmajorelementsandcomparedthe tholeiitic follows as a consequenceof this plume and flood basalt
compositions of the flows to thoseof other classicflood activity.
basaltprovincessuchas the Indian Deccan.Later, Fuller
PHYSICAL EVOLUTION
[ 1931] describedmanyof the physicalaspectsof the CRBs,
includingtheremarkablyintricatemixturesof basalt,altered
glass, and sediment (palagonite/pillowcomplexesand Eruptive Sequence
"pepperites") formedwhen the CRB lavasran into lakes
createdby the dammingof the largeriversby earlierflows. The SteensBasalt, a 1000-m-thick successionon Steens
Modem researchon the CRBG was initiatedby Watersand Mountain, Oregon (Figure 3), may be the earliest
his studentsin the 1960s [ Waters,1961]. manifestationof the Columbia River flood basalt [Carlson
The problemof distinguishing individualflowshasbeen and Hart, 1988]. Despite obviouslithologicaldifferences,
largely overcomeby applicationof magnetic polarity such as thinner flows and more variable mineralogies,the
measurements and increasinglyprecisechemicalanalyses. sequence of flows on Steens Mountain is coarsely
Campbelland Runcorn[ 1951] demonstrated that magnetic plagioclase-phyricand chemically similar to the Imnaha
Basalt on the Columbia Plateau and to the lower basalt
polarity varied in somebasalt sections,and more recent
work has shown that individual flows possessunique flows at Malheur Gorge (Figure 3). At the southend of the
magneticpropertieswhich may be used as a reliable Pueblo Mountains, south of SteensMountain, the base of
correlatingtool [Wellset al., 1989]. Use of majorelementthe SteensBasaltsequenceincludespicriticbasaltmembers
analysesin conjunctionwith carefullymeasuredsections (J. Evans, personal communication,1995)of cumulate
permittedWatersandhis students[e.g.Schmincke, 1967b]origin. These earliestCRBG-type eruptionsare part of a
to developthe first outlinesof a flow stratigraphy.Tracelong linear zone extendingfrom possiblyas far southas
elementanalysiswas first used to characterizeflows by Roberts Mountain in Nevada [Zoback et at., 1994] north
Osawaand Gores[ 1970] and hasbecomethe mostreliable throughSteensMountainto Pullman(Washington),a zone
methodof correlatingindividualflows acrossthe province that may have beendisplacedby WNW trendingstrike-slip
faults associatedwith later basin and range extension
[e.g.,Beesonet at., 1985;Reidelet at., 1989;Hooperet at.,
1995b]o [Lawrence, 1976; Hart and Carlson, 1987].
Improved magnetic and analytical techniqueshave The SteensBasalt is composedentirely of Imnaha-like
allowed flow-by-flow mapping, leading to relatively basalt,the lower half magneticallyreversed,the upperhalf
accurate estimates of the aerial extent and volume of normal [Mankinen et al., 1987]. At Malheur Gorge,
of manyfeeder dominantlyImnahaBasaltof unknownmagneticpolarityis
individualeruptionsandto the identification
dikes. This knowledge, and an increasinglymature overlainby a thin sequence of GrandeRondeBasalt.Farther
understandingof the tectonic settingof this small but north, alongthe southernedgeof the previouslyrecognized
otherwisetypical flood basaltprovince,makesthe CRBG Columbia River basalt province (southernmargin of the
importantin the developmentand evaluationof models Wallowa Mountains and at Squaw Butte a few kilometers
which attempt to explain the associationbetweenlarge northwestof Boise,Idaho) only a few flows of magnetically
igneous provinces, mantle plumes, and lithospheric reversed Imnaha Basalt underlie a thick sequence of
thinning. Currentwork on the earliestmanifestations of magneticallynormalImnahaBasaltwhich is overlainby
4 COLUMBIA RIVER FLOOD BASALT PROVINCE

GROUP
SU!•
GROUP
FORMATION MEMBER ISOTOPIC
MAGNETIC
AGE POLARITY
, , ,
(m.y.)
LOWER MONUMENTAL MEMBER 6 N

BASALT OF LOWER MONUMENTAL

--"I•ASALT
OFTAMMANY
C.
ICE HARBOR MEMBER 8.5
Basalt o! Goose Island N
Basalt of Martindale R

Basalt ol Basin City ........

N
BUFORD MEMBER R

ELEPHANT
MOUNTAIN
MEMBER 10.5 R,T
SADDLE POMONA
MEMBER 12 R
EQUA3:Z'•=L
MEMER '' N
MOUNTAINS wEISS
EN
I•'I•'•:S'
RIDGEMEMBER
Basalt of SlipperyCreek .. N
BASALT Basalt
ofTenmile
Creek ....
N
Basalt of Lewiston
.............
Orchards N
Basalt of Cloverland N
.....

I't ASOTIN
MEMBER 13

• Basalt
ofHuntzinger N
O WILBUR
CREEKMEMBER
CI• Basalt
ol.L..apw.
ai N
• Basalt
ofWahluke ,
N
UMATILLA
MEMBER
• Basalt
ofSillusi N
'"• Basalt
ofUmatilla N
,,• PRIESTRAPIDSMEMBER !4.5
• Basaltof Lolo R

IT Basalt
ofRosaliB R
LU .oz"A'
EBE. T,R
IT' 03 WANAPUM SHUMAKER
CREEK
MEMBER N
, ,

'"• FRENCHMAN
SPRINGS
MEMBER
<1:: 133 BASALT
• B_as.
allofLyons
Ferry N
• Basalt
ofSenfinal
Gap N
:•) Z Basalt
ofSand
H.o.
l10.w 15.3 N
• O
• Basalt
ofSilver
Basalt Falls
ofGinkgo N•E
E
O 03 Basalt
ofPalouse
Falls E
•' LooKINGGLASS
MEMBER N
t'l" ECKLER DODGE
MEMBER
,--I MOUNTAIN Basalt
ofDodge N
O BASALT Basalt
ofRobinette
Mountain N
15.6

GRANDE N2

BASALT
R1
16.5
T

IMNAHA No
BASALT

z• 175 Ro

Figure 2. Stratigraphic
tableof the ColumbiaRiver Basalt
 HOOPER 5

Vvashington
i
118 •
I
I
magmatismmovednorthwardwith time [Camp, 1995] from
Pullman*
I?'Iøscøw SteensMountain to the Washington-Oregon border.Present
radiometric dates do not supportthis, but the northward
ston migration may have been too rapid, in hundreds of
thousandsrather than in millions of years,to be detectedby
current absolute age dating techniques. NNW to N-S
trendingfeeder dikes are presenton SteensMountain and
what appearsto be a large vent complexof ImnahaBasalt
occurson Lookout Mountain (Oregon) at the southernend
of the Cornucopiadike swarm(LM, Figure3 andFigure4),
Wallowa
•t,
which is composedprimarily of Imnahaand GrandeRonde
L.G. -[' feeder dikes.

The phyric Imnaha flows were succeeded,conformably

take,IIII %*
and without an apparenttime gap, by the aphyricflows of
Grande Ronde Basalt. The petrographic change was
Is• II accompaniedby a sharp increase in silica (Figure 5).
•'John
Da5 Grande Ronde Basalt makes up 85% by volume of the
CRBG andwas eruptedin lessthan onemillion years(16.5-
Basalt of 15.6 Ma) [Tolan et al., 1989]. These flows are basaltic
Malheur
andesiteswith a silica range between 53.0 and 57.5%
Burltq•
.}untur:
(Figure 5). Chemicaldifferencesbetweenthe many Grande
WSR[ Ronde flows are small and they lack phenocrysts,making
their individual identification particularly difficult. The
flows do, however, exhibit three changes in magnetic
polarity which can be recordedin the field by a simple,
portable, fluxgate magnetometer and confirmed, where
necessary, by more complete paleomagnetic charac-
terization using drilled cores.The Grande Ronde Basalt is
.,_, I
!
thereforedivided into four magnetostratigraphic units from
bottomto top,R], N], R2andN2,whichhavebeenmapped
across the province [Figure 2; Choiniere and Swanson,
?':'---- ...... 1979; Hooper et al., 1979; Swansonet al., 1979, 1980,
118 ø
I
116•
! 1981; Swansonand Wright, 1983;Reidelet al., 1989].
Figure 3. Aerial extent of the Imnaha Basalt, southeastern
Most Grande Ronde flows were fed by membersof the
Columbia Plateau. LZ=suture zone between older cratonic Grande Ronde and Cornucopia dike swarms (zones of
lithosphereto north and east and accretedterranesto the south concentrateddikes within the Chief Josephdike swarm,
and west; SD-Sevin Devils Mts; LF=Limekiln Fault; Figure 1) in the most southeasterlycomer of the province,
FB=Farewell Bend; PM=Pedro Mt; LM = area of Imnaha vent but many Grand RondeN2 flows were derivedfrom dikes
scoria at Lookout Mountain; short solid lines represent the fartherwest [Tolan et al., 1989]. While Imnahaflows filled
approximatepositionsof Imnahaand GrandeRondedikessouth the deep canyonsthat predate the CRBG eruption and
of the Wallowa Mountains.OWL=Olympic-WallowaLineament; almostreachedthe peaksof the SevenDevils and Wallowa
WSRP-western Snake River Plain; R=Riggins graben;LG = La
Mountains (Figure 3) [Hooper and Camp, 1981], the
GrandeandBaker Grabens(14.-13 Ma). W = Weisergrabens.O-
I = Oregon-Idaho graben. Note that the field of SteensBasalt GrandeRondelavaseruptedover the newly developedflat
shown is significantly smaller than indicated by Hart and plains of Imnaha Basalt. As a result, the Grande Ronde
Carlson [1987]; it probablyextendsmuchfartherto the west, in lavas flowed farther west down-slope and each flow is
particular,but its preciselimitsarepoorlyconstrained. significantlythinnerthan most individualImnahaflows. A
few of the oldest Grande Ronde flows are present as far
increasinglythick (northwards)sequenceof GrandeRonde south as Farewell Bend and Malheur Gorge (Oregon)
Basalt. Still farther north, no Imnaha flows with reverse (Figure3), but increasingly youngerGrandeRondeunitsare
magneticpolarity are present.Assumingthis to be the same foundonly fartherand fartherto the northwest,creatingan
reversedto normalupwardpolaritybreak,thenthe centerof "off-lap" effect [Hooperand Camp, 1981]. The
6 COLUMBIA RIVER FLOOD BASALT PROVINCE

. •.•':'.•'i'•:"
ß '" '"': .: :+%
.: ....
.
:'::::i'"
•:;:.
.... ;:-: .,..
'!.,:..
' :....... --;;
....*.......
' "'•.•" ..•:.s,ß
-. ,...•::..:.
'...•

..........
ß
:....•....
.........
•,.• ;•
.......... •::...:::.-:..;
................. ::•:•
..........
:'• ............... •
;-• ........
.:,•%•.,::.; ....
.•':•:::-:½.•':'
....;7':'•*;;• •-• •:,;:;;i•;7%:.;;;•.;& e '•'' "•-'--"--'-----"•"' '--'--'"---'-'•'""-""--"--'"'----'"••"-'""'••••

.,•-' •.-.';-•
:• ...............
- •-•'-....."' .- ---;'•
=...2.. . -
...... -.'..:•.'
,'-:........................
> '?.::• . ........
•;;•;½.•
......
:•:.-•; '•.;•
........... :::..•-3.•....;;
..•
...:•., ................
..,..•
........... •.., • •. -•,•.-..:...½•-•.•••••.:•••-•*:, ,•½½-m,,
..-½',..:-•--•
..... -.• -- - .....•'"' '• ...... • ...... •' •'•""•••••' -.•' '•%.;';•.:•..
•. ß•

Figure 4. The Cornucopiadike swarmcrossingpalegranitesandlimestones

of the Wallowa Mountainshorst.

R2 and all N2 Grande Ronde flows eruptedonly on the sedimentis overlainby an extraordinary3500 m of CRBG
northwestern(downthrow)side of the northeasttrending [Reidel et al., 1989], almost three times the thicknessof
Limekiln Fault (Figure 3), which formedtowardsthe end of basaltthat accumulatedaroundthe feeder dikes supplying
the R2 magnetostratigraphic episode.It remainsunclearto theseflows to the southeast.The PascoBasinhasan abrupt
what extentthis southeast to northwestoffiap of flowswas a N-S trending eastern edge against the gently westward
consequence of the continuingnorthwardmigrationof the dippingPalouseslopeand is probablybestregardedas a N-
feederdike system[Camp, 1995] or of the continuingrise S orientedgrabenor rift. It was clearly a topographiclow
of the southeast comer of the Columbia Plateau. But it is which acted as a catchmentarea throughoutthe Tertiary,
evident that a continuously regenerating southeast- first for sediment and then for successiveeruptions of
northwestslopewas createdacrosswhich new lava flowed CRBG. Presenttopographyin boththe southeast part of the
towardsthe deepeningPascoBasin(Figure 1) [Hooperand Columbia Plateau and the PascoBasin suggeststhat this
Camp, 1981]. differential vertical movement acrossthe CRB province,
The apparentlycontinuousrise of the southeastcomer of including the steady rise of the batholithicrocks in the
the province is a facet of the rise of this part of Idaho southeastand the depressionof the Pasco Basin, has
documented throughout the Tertiary [Axelrod, 1968]. continuedto the presentday.
Correlation of the uplifted blocks with late Jurassicand At the heightof the CRBG eruptions,R2 and N2 flows of
Cretaceousgraniteintrusionssuggests that uplift was due to Grande Ronde Basalt not only filled the PascoBasin but
the steadyisostaticrise of thesebodies[Hooperand Camp, overflowed into the lower Columbia River channel and
1981]. That this relativeuplift continuedinto post-Miocene continuedthrough the Columbia Gorge acrossthe rising
time is dramatically illustratedin the Wallowa and Seven Cascade arc to the Pacific Ocean [Tolan et al., 1989].
Devils Mountains, where the basal flows of Imnaha basalt These are remarkabledistances(typically 300 to 600 km)
are now exposed on granite near the tops of those for individual eruptionsto cover and most workers have
mountains,many hundredsof metersabovethe sameflows accepted in the past that the formation of these flows
in the surroundinglow ground. required very high eruption rates [Shaw and Swanson,
The Pasco Basin, at the low end of this continuously 1970].
developingeast-westslope,is filled with 3000 to 7000 m of During the N• and R2 eruptionof GrandeRondeflows in
sedimentof Eocene age or older [Campbell, 1989]. The the southeastof the ColumbiaPlateau,separateeruptions
 HOOPER 7

8O
Saddle Mts

70

6O
Wanapum Y• x

5O ß
ß Eckler Mts

xxx
Grande
Ronde
x xx

x•

x
2O

Imnaha (N)

10 mmm ß
ß mm
ß
ß ß

0
ß ß Imnaha(R)
I I I mI
48 53 58 I 0 I 5 20 0 2

SIO2% A!203% TiO2 %

&&

x
x
x x x

xx
x
x
x xx
x

x
xx
xx
x

m,,,,

mmm
!

I I I I I

0 200 400 0 500 1•)00 .7 0.71 0.72

Zr ppm Ba ppm 87Sr/86S r

Figure 5. ColumbiaRiver BasaltGroup. Chemicaland isotopicparametersplottedagainststratigraphicpositionfor

major flows, illustratingmajor chemicaldifferencesbetweensubgroupsand formations[Hooper, 1988a; Hooper and
Hawkesworth,
8 COLUMBIA RIVER FLOOD BASALT PROVINCE

Picture Gorge Basalt (PGB) [Bailey, 1989a,b] from the [Watkins and Baksi, 1974]. The three formationscan be
Monument dike swarm occurredin the John Day Basin clearly distinguishedon the basisof phenocrystassemblage
(Figure 1) [Fruchter and Baldwin, 1975]. This basin was and major element composition(e.g., TiO2/MgO plots
structurallyisolatedfrom the restof the CRBG provinceby [Bailey, 1989a]). Across a low saddle in the Blue
the NE-SW trending Blue Mountains anticlinal uplift Mountains uplift, normal polarity PGB flows interfinger
(Figure 1). with N• Grande Ronde basaltswhich are cappedby R2
The PGB subgroupdiffers from the Imnaha and Grande Grande Ronde basalts.Radiometricages (16.5-15.6 Ma)
Ronde Basaltsin being more chemicallyprimitive (Figure [Baksi, 1989] confirm the correlationof the PGB with the
6a). The PGB alsohasdistinctivelylower concentrationsof N•-R2 magnetostratigraphic zones of the Grande Ronde
Sr (Figure 6b) and different incompatibletrace element Basalt(Figure 2).
ratioswhich link it more obviouslyto the contemporaneous The Picture Gorge Basaltsubgroupis lessthan 1.5% by
basin and range eruptionsto the southrather than to the volumeof the whole CRBG, but it mimicsthe sequenceof
main sequenceof the CRBG on the Columbia Plateau the physicalevolutionof the main CRBG eruption[Bailey,
(Figure 6c) [Hooper and Hawkesworth,1993]. The PGB 1989a]. The early plagioclase-phyricTwickenhamBasalt
subgroupis divided into three formations[Bentley and filled in deep canyonsof the earliertopography,as did the
Cockerham, 1973; Swansonet al., 1979; Bailey, 1989a]: similarlyphyric ImnahaBasaltto the east.The Twickenham
from base to top, the coarsely plagioclase-phyric Basaltwas followed by much thinnerand more widespread
TwickenhamBasalt(normalmagneticpolarity);the aphyric aphyric to sparselyphyric flows of Monument Mountain
to sparsely'phyric Monument Mountain Basalt (normal Basalt, equivalentto the GrandeRonde Basalt. The PGB
polarity); and the predominantlyphyric Dayville Basalt endedin the smaller,predominantlyphyric and chemically
(changing from normal to reversed magnetic polarity) more variable flows of Dayville Basalt in much the same

70 ''

•j• o o o
o
a
65-
oO o o o

60-
ß i ßß o 0 O0 O
c• o._. ßPicture
Gorge
[] Imnaha
AB
x o Steens

27,, X o
55-

c• 50- 4,.,Xo oøo••ø

. XGrande
Ronde
45-
00 •
o
•0

40-
' ,. oo
o
35- X: X 0 0

)X

30

40 90 140 190 240 290

Zr ppm
Figure 6. ChemicaldifferencesbetweenSteens,Imnaha(Rock Creek(RC) and AmericanBar (AB) types),Grande
Ronde,and PictureGorgeBasalts.(a) Mg# (Mg/(Mg+Fe2)vs Zr; (b) Sr vs SiO2;(c) Nb/Zr vs Zr/Y. [Hooper,1988a;
Hooper and Hawkesworth,1993
 HOOPER 9

600 - - x Grande Ronde

o

o o o Steens

ß Picture Gorge
500 '
[] Imnaha RC

[] Imnaha AB

400 '

x x

x
x
300

200 '

lOO

48 49 50 $1 52 53 54 55 56 57 58

SIO2%
0.12

0.1

[] Imnaha RC
0.08
ß Imnaha AB

x Grande Ronde
0.06
o Steens

Picture Gorge
0.04

0.02

0 I 2 3 4 5 6 7
10 COLUMBIA RIVER FLOOD BASALT PROVINCE

way as the main eruption to the east ended in the small,
chemically variable Saddle Mountains Basalt. Although
clearly from different sources,the similaritiesin thesetwo
sequences suggest that many aspects of the flow
petrography, including the presence or absence of
phenocrysts, were controlledby topographicfeaturesrather
than by more fundamentaldifferencesin magmasourceor
evolutionarypaths. Largeplagioclasephenocrysts appearto
correlate with locally thickened canyon-filling flows,
suggestingthat the phenocrystsgrew rapidly after eruption,
rather than by the more normally assumedintratelluric
crystallizationin deep magma chambersbefore eruption
[Swansonand Wright, 1981; Wrightet al., 1989].
In the main CRBG eruptions farther east, the large
volumesof GrandeRondeBasaltwere followedby a lull in
magmatism. Only small eruptionsof relatively primitive
Eckler Mountain Basalt occurredalong the easternend of
the Oregon-Washington border. Theselocalizedflows,the
single flow of RobinetteMountain and the many Dodge
flows, are typically underlain,and sometimesoverlain,by
saprolite horizons that correlate with saprolite and/or
sedimentaryunitsin otherpartsof the ColumbiaPlateau,all
implyinga significantdecreasein volcanicactivity.
A return to intensemagmaticactivity in the form of very
largeeruptionsof WanapumBasaltresumedthereafter(15-
14.5 Ma [Tolan et al., 1989]), albeit with magma
significantly more enriched in iron, titanium, and
phosphorus and, relativeto GrandeRondeBasalt,depleted
in silica (Figures5 and 7a). The Wanapumbeganwith the
small, relatively evolved Lookingglassflow, followed by
the manylargeflowsof the FrenchmanSpringsMember,all
eruptedfrom the westernmargin of the Chief Josephdike
swarm [Taubeneck, 1970]. The subsequentShumaker
Creek, Powatka,and Roza Members [Figure 2; Hooper and

Imnaha, GRB, Ekler Mt & Wanapum

TpO•l•e
* a

o.g--
ß Wanapum Basalt

'-.. Tel•x•x
0.7--
TiO2 "Gap' -..
imnaha, Grande Ronde
& Eckler Mt Basalts
0.5-'
x ß

imnaha
0.3--

0.1 I I I I ! I .........

0.5 I 1.5 2 2.5 3 3.5

TIO2%

Figure 7. P2Os/TiO2plotsfor (a) Imnaha,GrandeRonde,EcklerMountainandWanapum flows.GRB=Grande Rond

Basalt;Tgfs=FieldSpringsflow of the GRB; Ter=Robinette Mt flow (EcklerMt); Ted=Dodgeflows (EcklerMt)
Tel=Lookingglassflow; Tfs=Frenchman SpringsMember; Tsh=Shumaker Creek Member; Tpow=Powatkaflow;
Tr=RozaMemberandTpr=PriestRapidsMember(seeFigure2). Note similarrangesfor GrandeRonde(crosses) and
Imnaha (squares)Basalts.(b) SaddleMountain flows. Tu=Umatilla;Ta=Asotin;Tp=Pomona;Tem=Elephant
Mountain;Tesq=Esquatzel; Tw=WilburCreek;Tws, Twc, Twt, and Twl=the SlipperyCreek,Cloverland,Tenmile
Creek,and LewistonOrchardsflowsrespectively of the Weissenfels
RidgeMember;Tig, Tim, and Tib = the Goose
Island, Martindale, and Basin City flows, respectively,of the Ice Harbor Member; Tn=Eden; and Tim=Lower
MonumentalMember. The SaddleMountainsflows coverthe samerangeasthe lower flows of the CRBG but extend
to moreextremevaluesof P205 [Hooper,1988a;Hooperand Hawkesworth,
 HOOPER 11

Saddle Mountains Flows

b
ßd
T•g
In

i m

- Tib
Twl ß • -L•- Tim

• eeTIm Twt,•/••
Twc(•8
'X•"
Tw Ttmelllll
• ¸ •
Tws Tesq
Tern
Ta coooo Tb
m•l• Tp
I I

0.5 1.5 2 2.5 3 3.5

Ti02%

Figure ?. (continued)

Swanson,1990] were eruptedfrom the centerof the Chief process which requires higher than normal mantle
Josephdike swarm, to be followed by flows of the Priest temperatures. These featuresare well documentedon the
Rapids Member erupted from the eastem edge of that Columbia Plateau. Although the CRBG is substantially
swarm in west-centralIdaho. The three largestmembersof smallerthan otherclassiccontinentalflood basalts(an order
Wanapum Basalt (Frenchman Springs, Roza, and Priest of magnitudelessthanthe estimatedoriginalvolumesof the
Rapids Members) are chemically very similar, but are Deccan, Paran•, Karoo, and Siberian Trap provinces),
interposed with the smaller and much more siliceous 149,000 km3 of Grande Ronde Basalt (85% of the total
Lookingglass, Shumaker Creek, and Powatka flows CRBG) was eruptedin lessthan one million years[Tolan et
(Figures2 and $). al., 1989]. Individual eruptionsare remarkably homo-
The CRBG eruptionrate declinedbetween 14.5 and 6.0 geneousand may containmore than 700 km3 of magma
Ma during the eruption of the Saddle Mountains Basalt [Shaw and Swanson, 1970; Tolan et al., 1989]; volumes
(Figure 7b). These mainly small flows (exceptingthe greater than 2000 km3 have been suggestedfor some
Pomona) filled valleys and canyonscreated in the older individualGrandeRondeflows [Reidelet al., 1989].
sheetflows of the CRBG by tectonicdeformationand river The Pomona flow, one of the largest in the Saddle
erosion. Saddle Mountains Basalt flows are characterized Mountains Basalt, has an estimated volume of more than
by diverse compositions and by consistently more 700 km3 and can be traced for almost 600 km from its
radiogenicSt-isotopesignatures (Figure$). source in west-central Idaho. It flowed down the Snake
River canyon and, like some of the earlier GrandeRonde
Flow Volumeand Homogeneity flows, flooded the PascoBasin, overflowed into the lower
Columbia River channel, and plunged deep into wet
Unusually large individual sheet-flowsare the hallmark sedimentsoff the Pacific coast [Beesonet al., 1979; Pfaff
of a continentalflood basalt province and all continental and Beeson, 1989; Wells et al., 1989; Neim et al., 1994].
flood basalt provinces consist of exceptionally large This may be the longestterrestriallava flow documented
volumesof tholeiitic basaltextrudedover a shortperiod, a and it showsno discerniblevariation in its major or
12 COLUMBIA RIVER FLOOD BASALT PROVINCE

elementcompositionthroughoutits length [Hooper, 1982, would not be detectedin a magma body eruptingover a
1988a]. Such large volumes of homogeneousmagma period of many years. However, the eruptionrates of the
availableat onetime suggests
unusuallyrapid eruptionrates CRBG magmasremain unresolved.
from very large magmareservoirs.As there is no evidence
of collapsestructuresaccompanyingthe hugeeruptions,the Flow Morphology
reservoirs probably formed near the base of the crust
[Swansonand Wright,1981; Wrightet al., 1989]. A striking feature of the sheet-flowsof the Columbia
The only rationalefor the formationof sheet-flowsrather Plateau is the frequent development of well-defined
than shield volcanoesin continentalflood basaltprovinces colonnades and entablatures with vesiculartops [Figure8].
is the eruption of an unusually large volume of low Such simple flow structuresare lesscommonor evenrare in
viscositymagma in a shorttime. Justhow shorta time has otherprovinces (Deccan, Karoo, Paranti)wheremostof the
becomea matter for debate. Big eruptionslike thoseof the flows are compound. (The term "compound flows" is used
Roza Member were still largely liquid when they ran into here to imply massive units built from many small lava
water and chilled as glasssome300 km from their feeding flows or toes, eachwith pipe vesicles developed along their
fissures [Swanson et al., 1975]. Most of their large base and oxidized ropy upper surfaces;Figure 9; [Walker,
plagioclasephenocrystsgrew after eruption,implying little 1968]). The entablatureof a typical simple flow on the
loss of heat over large distances. Recently, Ho and Columbia Plateauis finer grainedand more glassythan the
Cashman [1995] have shown that the Ginkgo flow
(Frenchman Springs Member, Figure 2) lost only 20øC
while traveling across550 km of plateau (0.036ø/km) and
that the proportion of phenocrysts (15%) remained
unchangedover this distance. Shaw and Swanson[1970]
envisagedlava fronts 50 m high, perhaps 100 km long,
moving acrossthe very gently slopingColumbiaPlateauat
3 to 5 km per hour [Hooper, 1982]. However, in analogyto
the observedgrowthof flows on Hawaii, Hon et al. [1994]
and Self et al. [this volume] arguedthat the CRBG flows
were fed internallyin a successionof large lobes,eachnew
magma pulse raisingthe older flow crust,and that flow in
sucha processmay have beensignificantlylessrapid than
the turbulent flow originally envisaged by Shaw and
Swanson[ 1970]. Flow growthby inflationseemsinevitable.
This must have been the case for the Pomona flow, which
moved down a narrow steep-walledcanyon and which
would surely have formed a solidifiedroof to becomea
well-insulatedlava tube. A similar mechanismis probable
for sheet-flowscrossingthe plateausurface,but it remains
unclearjust how fast flows that were fed in this manner
could grow. It seemsimprobablethat the detailed,small-
scalemodels of Hawaii can be expandedby severalorders
of magnitudeto fit the muchhighereruptionratesapparent
on the Columbia Plateau[Hon et al., 1994]. By assuminga
lower viscosity and smaller initial flow thickness,later
inflated, and the rate of crustalgrowth derived from the
Hawaii study,Self et al. (this volume) concludethat CRBG
flows coveringareasof 700 to 2000 km2 were formedover
many years ratherthan the many days or weeks implied in
the Shaw and Swanson model. The remarkable
Figure 8. Photoof a typical"simple"flow [Walker, 1968] on the
homogeneityof individualflows is more easilyreconciled Columbia Plateau, showing the colonnade at the bottom,
with the more rapideruptionmodelsbecauseit is difficult to entablatureabove, and a vesicularand scoracioustop, Grand
understandhow some evidence of crystal fractionation Coulee,
 HOOPER 13

:•::• ßd'ii:•
•*

•.•::.' '•:%. ..'.:"• •'""-•:::'::;::•.•:'•::.::

......
•::..•-?'::
•'•?'•:•::•?'::----:,•-•-:
•'•:'7"•- •:-:-'•.z •': .:: .•:_:•.:•:•.,-•: •.:•::•::•-•::•.•::•.a
.•-.--•--.:.-.•.::..•?:.•..•
:?,•:::,.•::.:-:•:
..... ------::..:::,....?'-•:•-..
. '•-::•:i•).•:•
•.?•:.•-•
:•-;:•....- .,,•:• •:?• •. -•,
•::•::•
•'•"•:.A•
•.• •.;•:•
.......
.•..•.•
.....
• ... "'•'::"•,•::
4 ?:•,,:,•:. .. • .,•' :::,.•..'-:'•: --•:•::•:::•'•
•..•.•:•:•...•.:•"•:•
• •':.?•:•:•:?•:•
•:•:•.. •."•::.
'.
' ":' '•?'•, ' .... '"..... :• •.•',•':
-::,:,•'•e':•:.•,,
.•z,•..',::'.*
•.'--•:•:•'.•...,:•:•,..:.,
., .•:,•:•.•'•,
,,.
:•,,,,.•.•?,•,
,.•:•::•....,,,...
'. • • ,:•'-,'•-:--:.•, •:::•:•
............ ...
.•.• ..

...... ..•:• ..::..2:.:•,.

........ ß

,•.%•.?:....•.....:::.:::•.:...?•.•
,..•...•....,..••..,•
.......
.::. ...
......
•

Figure9. Lavalobesin a typicalcompoundflow[Walker,1968],K•oo basalts,

Lesotho,
southern
Africa.Eachlobe
tendsto havepipevesiclesat the b•e andan oxidizedropytop. Successive
lobes•e chemicallyidentical,many
makingup a singlecompound flow.Thesestructures
•e r•e ontheColumbiaPlateau.

colonnade. Long and Wood [1986] suggestedthat wet sedimentmay be complex,leadingto intricatemixtures
entablaturesformed when fresh lava was coveredby water of small black glassfragmentsevenly mixed throughthe
which penetratedcracksin the congealinglava, causing pale sediment(pepperRe). Often, solid massesof the
rapidcoolingfromthe surfacedownward.In thismodelthe younger basalt burrowedbeneaththe less dense and
usuallysharpbreakbetweenentablature and colonnadeis unconsolidated older sedimentto form invasive(as opposed
the line where the cooling surfacerising from the base to intrusive)flows [Byerly and Swanson,1978]. Recent
meetsthatdescending fromthetop. Certainlythe formation work alongthe Pacificcoast[Beesonet al., 1979;Neimand
of lakes on top of theseflows must have been common, Neim, 1985;Pfaff and Beeson,1989;Neim et al., 1994] has
becauseflow after flow obstructedand dammed the large detailed extreme examplesof this invasive mechanism.
riverscrossing the plateau.SeeSelfet al. (thisvolume)for a Many large CRBG flows (includingthe Pomona)swept
more detaileddiscussion of the morphologyof CRBG flows downthe ColumbiaRiver valley and invadedwet sediment
and alternativemodelsfor their origin. as dikes and sills for hundreds of meters offshore. Careful
mappinghas demonstrated the continuityof CRBG flows
Pillow-PalagoniteComplexes
andInvasiveFlows on the plateauand flows throughthe ColumbiaGorge,so
establishingbeyondreasonable doubtthatthe coastaldikes
The presenceof manylakesandriverson the Columbia and sills are invasive into the wet sediments and are not
Plateau during the CRBG eruptionsis evident from local feederdikes as originallyinterpreted[Snavelyet al.,
numerous examplesof pillow-palagonite
complexesformed 19731.
when basalt lava invadedwater and wet sediment(Figure
1O) [Fuller,1931;Schmincke,
1967a;Swanson
and Wright, Basalt Textures
1981]. Almost any road cut acrossthe plateau in
southeasternWashingtonexposesthese complexes. In The great majority of flows on the ColumbiaPlateau
many examplesthe elongationand slopeof the pillows belong to the dominantGrande Ronde Basalt, which is
the aphyric, despiteits evolved chemicalsignature. Other
createthe equivalentof foresetbeds, demonstrating
directionof magmaflow. Reactionsbetweenhot lava and formationsare largelyphyric,with plagioclasealways
14 COLUMBIA RIVER FLOOD BASALT PROVINCE

Figure10. Pillow-palagonite
complex,
Asotin
Grade,
southeast
Washington.
Suchcomplexes
arecommon
across
the
Columbia
Plateau,
formedwhenthelavasflowedintoriverbedsandlakescreated
by thedamming
of riversby
previousflows.

dominantphenocryst phase,followedby olivine, then thenaugite)observedis nottheassemblage

responsible
for
augite. There is no simplecorrelation betweenthe fractional
crystallization
in magmareservoirs
at thebaseof
proportion andabundance of phenocrystsandsuchindices the crust and should not be used uncritically as the
of fractionationasMg# (Mg/{Mg+Fe2}) or Zr abundance. assemblage
to testfractionation
modelsin theCRBG.
Porphyritic textures
aretraditionally
regardedastheslow
intratelluric
growthof phenocrysts in reservoirs
belowthe Linear VentSystems
surface,surrounded by the finer grained,more rapidly
cooled,matrix phasesthat formedas the residualliquid The earliestwork on the ColumbiaPlateausuggested
that
the CRBG magmaeruptedfrom fissures. A detailed
chilled on extrusion. However, an alternativescenariohas
beendemonstrated [Lofgren,1980]in which account
experimentally of the highlyoriented
NNW-SSEfeederdikesand
phenocrystsandgroundmass growat differingratesin a smallventsfor manyflowswasprovided
by Swanson et al.
singlecoolingenvironment.
Two linesof evidencesuggest [1975] and this work hasbeenaugmentedby numerous
more detailedstudiesin the southeastern
that this alternative mechanism has been effective on the part of the plateau
Columbia Plateau:the correlationof porphyritictextures [e.g.,HooperandSwanson, 1990;ReidelandTolan,1992;
withunusually flowsin theImnahaand Hooper et al., 1995b]. Linear vent systemsfeeding
thickcanyon-filling
Twickenham Basalts, and the fewer, smaller phenocrysts individual eruptionsor sequentialeruptionsof nearly
created in the Roza flow when it plunged into water identicalcomposition, suchas the Roza and Ice Harbor
hundredsof kilometersfrom its eruptioncenter[Wright et systems, arefrom50 km to over200 km longandonlya
al., 1989]. It is, therefore,possiblethat the typical few kilometerswide with a minimumlength-to-width ratio
phenocryst assemblage (plagioclase,followedby olivine, of 15:1. Vents and conesof varioussizes, complete
 HOOPER 15

scoria, ash layers, cinders, and spatter can be studied most controversialaspectsof the origin and evolutionof the
between Clarkston and Pomeroy (Washington)[Swanson CRBG. The age of enrichmentin crust or mantle will be
and Wright, 1978; Swansonet al., 1980; Hooper et al., reflectedin the isotopicsignature.
1995b].
Sources
CHEMICAL VARIATION AND
PETROGENETIC MODELS Before the effects of partial melting and crystal
fractionationprocessescan be adequatelyassessedit is
General necessaryto characterizethe source componentsand to
identify groupsof flows that couldbe derivedfrom single
Most continentalflood basaltstend toward higher silica sources by partial melting and/or crystal fractionation.
and potassiumconcentrations than typical oceanicbasalts. Waters[ 1961], using the few major elementanalysesthen
However, the high silica content of the Grande Ronde available, pointed out that partial melting and/or fractional
basaltic andesites (53.0 to 57.5 wt% SiO2, normalized crystallizationprocessesalone were unable to create the
without volatiles) is unusual and, in the context of chemical differences observed in the CRBG and that

continentalflood basalts,is rivaled only by the TasmanJan different sources were required. Thirty-five years later,with
dolerites[e.g., Hergt et al., 1989]. The scarcityof picritic the benefit of thousands of major and trace element analyses
compositionsand the lack of rhyolitic units in the CRBG and a comprehensivecoverage of isotopic ratios [e.g.,
alsocontrastwith mostotherflood basaltprovinces. Carlson, 1984; Church, 1985; Hooper, 1988a;Smith, 1992;
Major units within the CRBG (subgroupsand formations; Wright et al., 1989; Hooper and Hawkesworth, 1993;
Figure 2) are defined by differencesin field appearance, Brandon et al., 1993; Chamberlain and Lambert, 1994;
petrography,and major element compositions(Figure 5) Lambert et al., 1995], the precisenature of those sources
[ Waters, 1961; Wright et al., 1973]. Theseunits also differ remainsonly partially resolved.Most workerswould agree
in their relativetraceelementabundances and isotopicratios that three or four different mantle sources,and a variable
in a mannerindicativeof their derivationfrom a varietyof crustalcomponent,can be identifiedwith varying degrees
sourcecomponentsin eitherthe mantleor the crust[Carlson of clarity.
et al., 1981; Carlson, 1984; Hooper and Hawkesworth, Carlson [1984] and Carlson and Hart [1988] identified
1993; Brandon et al., 1993]. The consistentlybasaltic three mantle sourcecomponents on the basisof Sr, Nd, and
character of all CRBG flows implies that they were Pb isotopic arrays, including an enriched subcontinental
ultimately derived from the partial melting of a mantle lithospheric mantle source for the more primitive flows of
source.Most ColumbiaRiver basaltsare relatively evolved the Saddle Mountains Basalt, similar to the sourceinvoked
and differ from oceanic basalts in that their large ion for the Snake River tholeiites by Leeman [ 1975] and others.
lithophile (LIL) and high field strength(HFS) incompatible In addition, Carlson and Hart [1988] recognized a
trace elementsare decoupled(as shownby negativeTa-Nb lithospheric component in the majority of the flows and, on
anomalieson a mid-oceanridge basaltnormalizedelement the basisof high K/P ratiosand variable8•80 contentwith
diagram). This trace element pattern is typical of little variation in 878r/86Sr,argued that this lithospheric
subduction-relatedmagmas and many continentalcrustal component was probably derived from crustal
rocks and indicates a "lithospheric"(mantle or crustal) contaminationaccompaniedby crystal fractionation,most
source componentsimilar to, but generally less extreme obviously in the Grande Ronde and Saddle Mountains
than, that characteristicof subduction-relatedcalc-alkaline flows.

sequences[e.g., Prestvik and Goles, 1985]. LIL/HFS
element decoupling is generally attributed to the LIL
enrichmentof a previouslydepletedmantle wedge by the
products of hydration from the subducting slab [e.g.,
Pearce, 1983]. Sucha signaturecouldhave been acquired
by the CRBG magmaseitherthroughcrustalcontamination
or by partialmelting of an oldersubcontinental
lithospheric
mantle (SCLM) enrichedin LILs in an earlier subduction
process.In all probability,both these sourcecomponents
were involved, but a clear distinctionbetween them has not
been found and their relativeimportanceremainsone of the
On the basis of an expandedisotopicdata base for the
Imnaha,GrandeRonde,and WanapumBasalts,Hooper and
Hawkesworth [1993] identified a similar isotopic
distribution, but interpreted the source components
differently.Carlson•s (1984) first sourcecomponent(C-I;
87Sr/86Sr <0.7035; •Nd>+6.5; 2ø6Pb/2ø4Pb=18.8; 2ø?pb/
204Pb=15.51;208pb/204Pb=38.3; and 8•80 =5.6) is found
primarily in the Picture Gorge Basalt and in other basalts
eruptedover a large part of the OregonPlateau,including
the ubiquitoushigh aluminaolivinetholelites(HAOTs) and
the SteensBasalt. Carlson and Hart [ 1988] derived
16 COLUMBIA RIVER FLOOD BASALT PROVINCE

0.5132
from a depletedmantle which they associated with ocean
islandbasaltsand explainedthe lithosphericsignatureby a 0.5130

small addition of sediment to the source. In contrast, 0.5128

Hooper and Hawkesworth[1993] emphasizedthe strong

Bulk Earth

0.5126
decouplingof the LIL from the HFS incompatibletrace
elementsin theserocks.They ascribedthe decouplingto a 0.5124

depleted mantle source subsequently enriched by 0.5122 •( x

Saddle Mountains
subduction-related fluids to create an enriched 0.5120
X
X

subcontinentallithosphericmantle componentassociated
0.5118
with the Blue Mountains accreted terranes, a source too i x 87Sr/O65r
I I i i i I I
young to displaymuch variationin their depleted-mantle- 0.5116 I

0.702 0,704 0.706 0.708 0.710 0.712 0.714 0.716

type isotopic signature.This source was activated by
decompressional meltingduringbasinandrangeextension
and it has been recognizedover most of the Basin and
Range Province [e.g., Fitton et al., 1988; Hooper et al., 19.4 _ 206pb/204pb
1995a]. Both Carlson and Hart [1988] and Hooper and 19.2
Hawkesworth[1993] emphasizedthe distinctness of this
19.0
sourcefrom that responsiblefor the main ColumbiaRiver
basalts. 18.8

x X
Carlsun's [1984] second (C-2) mantle source is 18.6 -

dominated by the relatively restricted array of Imnaha 18.4 -

Basalt isotopicratios (87Sr/86Sr=0.704; SNd=+4.5; 2ø6Pb/ 18.2 -

x
2ø4Pb=19.09; 2ø7Pb/2ø4Pb =15.65; 2ø8pb/2ø4Pb =38.7; and 18.0 -

$•80 >6.0). This is interpretedby Carlson as a mantle 17,8 -

source enriched by a small percentage of sediment. I7.6 ! i i

87Sr/86Sr

However, on the basis of some of the earlier Imnaha flows O.703 0.705 0.707 0.709 0.711

lacking the LIL/HFS decoupling and showing overall

similarities to Hawaiian tholeiites, Hooper and Figure 11. Radiogenicisotopiccompositionof the CRBG [after
Hawkesworth[1993] related this componentto an ocean Hooper and Hawkesworth,1993]. Triangles= PictureGorgeand
high alumina olivine tholeiites(HAUT) from NE Oregon;open
island or mantle plume-relatedmantle source(Figure 11)
squares= Imnaha; filled squares= GrandeRunde;diamonds=
and used it specificallyas evidencethat the CRBG was a
Wanapum;crosses= SaddleMountainsBasalts.
consequence of mantle plume activity [see also Brandon
and Gules, 1988].
The isotopicarray formed by the Grande Runde and
WanapumBasaltswas interpretedby both Carlson[1984] equilibrium amongstphenocrystphases,for example,and
andHooperand Hawkesworth[ 1993] as a mixingarrayof lack of correlation between degree of evolution and the
which one end member is the Imnaha Basalt (Carlsun'sC- abundanceand compositionof the phenocrystassemblage)
2). The otherend memberenvisagedby Carlsonand Hart evidence is lacking. Both Carlson and Hart [1988] and
[1988] is the same as for the more primitive Saddle Hooper and Hawkesworth[1993] agreedthat the Saddle
MountainsBasalt(C-3; 87Sr/86Sr =0.7075; SNd=-5; 2ø6Pb/ MountainsBasaltis probablyderivedfrom SCLM enriched
2ø4pb=18.09-19.11; /5]80 <+6), a subcontinentallitho- in the late Archaen with some degree of crustal
sphericmantleenrichedin late Archaentime. Hooperand contamination, the latter most evident in flows like the
Hawkesworth[1993] also interpretedthis end memberas Wilbur Creek with unusuallyhigh 878r/86Sr(0.713) values
SCLM, but one enrichedin the mid-Proterozoic anddistinct and eruptedfrom fissureson or to the eastof the boundary
from the SCLM responsiblefor the significantlymore separating the accreted terranes and the older North
radiogenicSaddleMountainsBasalt.Unlike Carlsonand American plate.
Hart [1988], Hooper and Hawkesworth[1993] explained In a detailedisotopicstudyof the PictureGorge Basalt,
the lithosphericsignatureof the Grande Runde and Brandon et al. [1993] found some variation in the Sr and
WanapumBasaltsas dueprimarilyto thisenrichedSCLM Nd isotoperatios,the latter (but not the former) of which
componentwithout resort to crustalcontamination, for correlatedwith silica content.Assumingsilica contentwas
which field (crustal xenoliths) and petrographic(lack of an index of crystal fractionation,they interpreted

correlationasevidenceof crystalfractionation
accompanied Partial Melting, CrystalFractionation,
by crustalassimilation,so agreeingwith earlierstudiesby and Magma Mixing
Bailey [1989b] and Carlson [1984]. In the Hooper and
Hawkesworth[ 1993] scenario,however,the use of silica as The chemical variation causedby mixing of different
an index of fractionationwas not convincing;the silica sourcecomponents,the dominantprocessin the formation
content would vary with both crustal assimilation and
increasingadditionsof an enrichedSCLM component.
Other isotopicstudiesincludethoseof Church [1985; Pb
isotopes],Smith's [1992] discussionof the origin of the
CRBG flows along the Pacific coastof Oregon,and two
studies [Chamberlain and Lambert, 1994; Lambert et al.,
1995] which discussedthe isotopicvariationsin the Grande
RondeBasalt.Churchsuggested a depletedoceanicmantle
sourcecontaminatedby subduction-derived fluids to form a
hydrated peridotite, a concept echoed by Hooper and
Hawkesworth [1993] using geochemicalevidence. Smith
emphasized the need for iron-rich (pyroxene-rich)
subcontinentallithosphereas sourcesfor most CRBG flows
[seealso Wrightet al., 1989], includingthe GrandeRonde
Basalt, but denied any associationbetweenImnaha Basalt
(Carlson'sC-2) and a plume-relatedsourcebecause,he
claimed,sucha component wasavailableovera largepart
of the PacificNorthwest.Chamberlainand Lambert[ 1994]
and Lambertet al. [1995] arguedfor at leastsix separate
source components:a plume component,two depleted
MORB-like sourcecomponents slightlycontaminated with
sediment,two local crustalcomponents,and a sixth,more
complexcrustalcomponent.
In summary,the scatterof isotopicvaluesrequiresthat
the CRBG is the product of many different source
components.
The evolvednatureof eventhe mostprimitive
flows requireseither that the mantle-derived magmas
underwent substantialfractionationin the crustbeforeany
flows eruptedor that the mantlesourcewas unusually
enrichedin iron [Wrightet al., 1989;Smith,1992].Beyond
thesebasicpointsthere appearsto be little consensus as to
the natureof the mantlesourcesandthe degreeto whichthe
nearlyubiquitouslithosphericsignature,as displayedby the
decouplingof LIL from HFS incompatibletraceelements,is
the result of partial melting of an enrichedMORB-type
source or of crustal contamination.No clear way to
distinguishbetween these two possibilitieshas yet been
identified. Theoretically, the relative values of /5180 and
87Sr/86Srshould provide such a distinction[dames, 1981].
When appliedto the GrandeRondeBasalt,for example,the
data supportthe crustalcontaminationmodel [Carlson and
Hart, 1988]. Unfortunately, the /5180 values are only
marginallyabove acceptablemantlevaluesand all the data
are from whole rocks, notoriouslyprone to increasesin
these values by small degrees of alteration, and so this
evidenceis also suspect.
HOOPER 17
of the Grande Ronde Basalt, makes it difficult to evaluate
the effectsof partialmeltingandfractionalcrystallizationin
those rocks. However, in the Imnaha Basalt the more
uniform LIL/HFS element and isotopicratios permit a
singlesourcemodelasa first approximation
for their origin.
The chemical variation in the Imnaha Basalt can be
explained by a combinationof partial melting, crystal
fractionation,and magma mixing. The differencebetween
the Imnaha and Grande Ronde Basalts is illustrated in
Figure 12; the tight positivecorrelationbetweenP andZr in
the Imnaha Basalt, as required in a system controlled
primarily by partial melting and/or crystal fractionation,
contrastswith the wider scatterof the Grande Ronde data,
which reflectmixing.
Chemical variation within the Imnaha Basalt is
considerable[Hooper et al., 1984]. Two or three discrete
sequencesof Imnaha Basalt (Figure 13a,b), each with a
significantrange in HFS element concentrations,differ in
their Sr contentand Mg# at the sameZr abundance,and in
suchHFS/HFS ratiosas Zr/Y. The variationssuggestthat
eachsequencewas formedfrom a differentparentalmagma
and that each parental magma could representdifferent
degreesof partial melting from a commonsource.Variation
within each sequence is consistent with fractional
crystallization of a gabbroic assemblage(plagioclase:
augite: orthopyroxene 16:10:4 for the AB flows),
augmentedby recharge,in magma chambersin the lower
crust [Hooper, 1988b; Hooper and Hawkesworth,1993].
Although the processinvolved many cycles of 30-50%
fractionationwith 40-50% recharge,there is little increase
in silicaandlittle evidenceof a negativeeuropiumanomaly.
A similar distinction between partial melting and
subsequentgabbrofractionationcan be made for the Eckler
Mountain Basalt [Hooper et al., 1995b]. In the Saddle
MountainsBasalt,the detailedstudyof the Ice Harborflows
by Helz [1973, 1976] and Helz and Wright [1982]
demonstratedthat crystal fractionationplayed little part in
the formationof thoseflows whereasthe largevariationsin
isotopicratiosin all SaddleMountainBasaltflows implied
that the chemical variation is principally a function of
multiple(includingcrustal)sources.
In additionto the role of the mixing of partialmeltsfrom
differentsources(GrandeRondeBasalt)and mixing in the
form of rechargeof magmareservoirswhich seemsto have
been a normal part of the gabbro fractionationprocess
(Imnaha and Eckler Mountain Basalts),there are also
18 COLUMBIA RIVER FLOOD BASALT PROVINCE

Imnaha & Grande Ronde Type Sections

0.7

0.6

0.5

xX'x Xxx•k•
XX
xx x
0.4

x X x Grande Ronde basalt

0.3 x
xX X
[]
[] Imnaha Basalt

0.2 '

0.1

100 150 200 250 300 350

Zr ppm

Figure 12. P205 versusZr for Imnaha and Grande Ronde Basalt, illustratingthe tight positive correlationin the
Imnaha (filled squares),as requiredby simplepartialmelting+ fractionalcrystallizationand the greaterscatterin the
Grande Ronde Basalt (crosses)in which the chemicalvariation is ascribedin large part to mixing of two source
components.

examplesof magmamixing at later stagesin the evolution
of CRBG magmas.In a provincewhere large volumesof
magma eruptedover a short time, magmamixing during
eruptionseemsprobable. Late magmamixing has been
identified where the chemical differences between flows are
clear, as between the Wilbur Creek and Asotin flows which
mix to form the Lapwai or Huntzingerflow [Hooper, 1985;
Reidel and Fecht, 1987], or between the two flows of the
Umatilla Member (Saddle MountainsBasalt).However, in
the more common scenarioin which two magmasare of
similar composition,mixing is difficult to demonstrate.
Thus, mixing betweenthe many rapidly eruptedGrande
Ronde flows seems probable, but has not yet been
demonstrated becauseof their similarcompositions.
Carlson and Hart [1988)] explained the chemical and
isotopic variationswithin the Grande Ronde Basalt by a
combinationof crystal fractionationand assimilationof a
potassium-richupper crustalcomponent(AFC). In contrast
Helz and Wright [1982] and Wright et al. [1989] argued
against significant plagioclase-dominated (gabbroic)
fractionation for the Grande Ronde. Hooper and
Hawkesworth [1993] demonstrated a close positive
correlation between Rb/Zr and Rb/Sr in the Grande Ronde
Basalt which could not be explained by gabbro
fractionation.They calculatedthat the maximumpossible
gabbro fractionationwithin the Grande Ronde was limited
to 10%. This makes it difficult to acceptthe Carlson and
Hart [1988] AFC model, in which assimilationof crustal
rocksmustbe offsetby substantialfractionation.Mixing of
mantle-derivedbasalticmagma with a crustalmelt would
still be a feasibleexplanation,but one might then expectto
see some indication of this silicic crustal melt on
 HOOPER 19

70- reservoirs at or near the base of the crust. Clear evidence of

variable partial melting and gabbro fractionationcan be
demonstrated in the Imnaha and Eckler Mountain Basalts
60-
but appearsto be largelyabsentin the GrandeRondeBasalt.
Partial melting, crystal fractionation,and source mixing
50- processeswere probably augmentedby the late mixing of
magmasjust prior to and duringeruption.
40-

TECTONIC SETTING
Zr ppm
30 , • • , I • t, • ,, i
100 125 150 175 200 225 250 275 The main volume of the CRBG eruptedthroughNNW-
SSE orientedfissuresthat now form the Chief Josephdike
38O
Srppm •'--- •---'-I. --- " b. swarm (Figures 1 and 3). The dikes are confined to the
340 ,• F.C.
southeastquadrant of the Columbia Plateau and extend
P.M. almost as far south as Farewell Bend on the Snake River
300 (Figure 3; P.R. Hooper, unpublishedmapping, 1981;
Howard Brookes, pers. comm. 1995). Circumstantial
evidence[Camp, 1995; Hooper et al., 1995c] suggeststhat
x• Zr ppm the Steens Basalt may represent the earliest CRBG
220 1 ! . ! ! I I ! !
100 120 140 160 180 200 220 240 260 eruptions.If so, this earliest outburst(16.5 Ma) was of
Nb/Zr C.
smallerbut more frequenteruptionsthan the main fissure-
0.10 -
fed CRBG eruptionsto the north,and so formedthe Steens
Mountain shield volcanic edifice, rather than a true flood
basaltplateau.
The Chief Josephdike swarmis specificallyconfinedto
0.O6 - the relatively thin lithosphericwedge of the accretedBlue
Mountains oceanicterrane that forms a deep embayment
into the older and thicker North Americanplate (Figure 3)
Zr/Y
[e.g., Vallier, 1995]. The suturethat forms the boundary
0.02 ' • -- betweenthesetwo contrastinglithospheres (SZ in Figure 3)
2 4 6
is markedby a stronglyfoliatedandshearedzoneseparating
Figure 13o ImnahaBasalt.Plotsto illustratebothpartialmelting strongly contrastingrock types [Snee et al., 1995] with
(P.M.) and gabbrofractionation(F.C.). (a) Mg# vs Zr (datafrom different 87Sr/86Srsignaturesand intruded by ultramafic
Hooper et al. [1984]). (b) Sr vs Zr (averageflow analysesfrom lenses[Armstronget al., 1977;Fleck and Criss, 1985;Mohl
Hooper [1988b]); Nb/Zr vs Zr/Y (data from Hooper and and Thiessen,1995]. The thinner oceaniclithospherewest
Hawkesworth[1993]). Envelopesenclosedatapointsbelievedto
of this suturehasbeenparticularlyproneto furtherthinning
have formed by gabbro fractionation from a single parent
magma. Data points outsidethese envelopesare interpretedas
during subsequenteast-westextensionfrom the Eoceneto
resultingfrom mixing betweenmagmasderived from different the present[e.g., Hooper et al., 1995a].
parents. Deformation during and after the CRBG eruption is
largely confined to that part of the Columbia Plateau
surface.Such silicic magmasare specificallyabsentfrom overlyingthe accretedterranes.The basaltsare undeformed
the CRBG. north of the suturezone in easternWashington(Figure 3)
In summary,the most fundamentalchemicaland isotopic where they lie on the old cratonic crust. To the south,
divisions within the CRBG result from different mantle and where they overlie the thinner, less competent,accreted
crustal source components. Within these divisions, terranes,they are folded and faulted (Figure 3) [Reidel et
however, smaller sequenceswith relatively coherent al., 1994; Hooper et al., 1995b]. Thickness and
isotopicand trace elementratios can be isolated. Each of competencyof the lithospherethus appearto have played
thesesequences may be derivedfrom a commonsourceand major roles in determiningboth where the basaltserupted
their chemicalvariationmay result from a combinationof andhow they were subsequently deformed.
different degreesof partial melting and crustal(gabbroic) The east-westtrendingYakima fold belt westof the Pasco
fractional crystallization (accompaniedby recharge) in Basin(Figure 1) and similarfoldsto the east,grew
20 COLUMBIA RIVER FLOOD BASALT PROVINCE

and after the eruptions[Hooperand Camp, 1981; Reidel, South of the OWL significantly greater east-west
1984; Price and Watkinson,1989; Hooper et al., 1995b]. extensionand lithosphericthinningare demonstrated by the
The highly orientedNNW-SSE feeder dikes of the Chief development of graben, horst, and complex pull-apart
Josephdike swarm testify to the consistentENE-WSW structures [Gehrels, 1981; Hooper and Conrey, 1989;
extension throughout the eruptive period (17-6 Ma). Mann, 1989;Mann and Meyer, 1993]. Extensionappearsto
Strike-slipstrain with minor translationtook place along have occurred sporadically;the La Grande, Baker and
steeply inclined NNE-SSW (left lateral) and WNW-ESE Weiser grabensformed between 13 and 14.5 Ma [Bailey,
(right lateral)faults,includingthe Hite andassociated
faults 1990] and farther south, current evidence suggeststhat
[Hooperet al., 1995b]andfaultsparallelwith the Olympic- extensionincreasedat about 15 Ma [Hooperet al., 1995c].
Wallowa Lineament(OWL) [Hooper and Conrey, 1989; East-westfolds of this age are not obvious.Here the strain
Mann and Meyer, 1993]. The same strain pattern ellipsoid has c•-I vertical and c•-3 WSW-ENE horizontal,
representedby these structuresfrom 17 to 6 Ma can be with the NNW-SSE horizontalaxis representingc•-2. Prior
measureddirectly on the ColumbiaPlateautoday [Kim et to 14.5 Ma, when the east-west Aldridge Mountains
al., 1986]. anticlineand parallel folds in northeastOregonmay have
The amount and timing of displacementon the OWL formed,the strainpatternwasprobablymoresimilarto that
remainscontroversial[seeMann and Meyer, 1993; Reidel north of the OWL.
and Tolan, 1994]. Much of the translationmay have taken The simplicityof this structuralmodeland its application
placeprior to the CRBG eruptionsin a mannersimilarto the over 164,000km2for ten million yearsare unusualandmay
50 km or more left-lateral displacementon the Hite Fault reflect the relatively simple geology of the Columbia
prior to WanapumBasalteruption(Figure3) [Reidelet al., Plateauwhere, in the deformedsouthernpart, oceaniccrust
1994; Sobczyk,1994; Hooper et al., 1995b]. What is clear is topped by a thin veneer of uniform basalt flows.
from these and other detailed analyses [Reidel, 1984; Superimposed on this long-livedregionalstrainpatternis
Anderson,1987] is that the strainpatternobservedacross the isostaticrise of the Idaho batholith and other granitic
the wholeColumbiaPlateauis consistent with thepatternof bodiesthroughoutthe Tertiary [e.g., Axelrod, 1968]. The
east-west extension that created the Basin and Range rise of the granitic rocks in the east and the continued
Province and that the orientationof this patternhas not deepeningof the Pasco Basin to the west resultedin a
changedover the last 17 million years.If the 25ø clockwise constantly evolving east-to-westslope down which the
rotationof northeasternWashington[Fox and Beck, 1985] CRBG lavas flowed.
is restored,the type andorientationof strainappearsto have Given this regional setting of thinned and attenuated
changed little since the Eocene over the whole Pacific lithosphere throughwhichthe CRBG waserupted,therehas
Northwest.For a contraryview, seeBarrashet al. [ 1983]. been discussionas to the genesisof the CRBG eruption.
North of the OWL, where the amount of extension and Was the eruptiona consequence of extension,in the form of
lithosphericthinning was minimal [Taubeneck,1970], the back-arcspreading300 km to 400 km behindthe active
NNW-SSE extensionalfissures(dikes) and east-westfolds Cascadevolcanic arc [e.g., Carlson, 1984; Carlson and
dominate,with WNW-ESE right-lateralandNNE-SSW left- Hart, 1988; Smith, 1992]? Or was the eruptiondue to the
lateral strike-slipfaulting forming a conjugatepair on a Yellowstonehotspot,whose trail acrossthe easternSnake
regional scale.On a strain ellipsoid,c•-I is NNW-SSE and River Plain placesit beneaththe McDermitt Craterson the
horizontal.Near the surface,with minimal crustalthickness, southend of the Steens-PuebloMountain ridge at 16.5 Ma
c•-3 is vertical, so c•-2 is WSW-ENE horizontal, the at the beginningof the CRBG eruption(Figure 3) [e.g.,
configurationrequiredto createthe east-westfolds.Deeper Hooper, 1984; Brandon and Goles, 1988; White and
into the crust,the vertical componentincreases, to become McKenzie, 1989, 1995; Richards et al., 1989; Carlson,
c•-2 and the WSW-ENE horizontal axis of the strain 1991;Draper, 1991;Geistand Richards,1993;Hooperand
ellipsoidbecomesc•-3, the configurationrequiredto create Hawkesworth,1993; Camp, 1995]?
conjugateWNW-ESE (right lateral) and NNW-SSE (left There are obvious problems in attributingthe CRBG
lateral) conjugatestrike slip faults. Towardthe baseof the eruptionto a mantle plume in the form of the Yellowstone
crust, in the vicinity of prospectivemagma reservoirs,the hotspot, in addition to our embarrassingignoranceas to
vertical pressure increased to equal the NNW-SSE what a hotspotactually is. First, the main eruptionof the
horizontalpressure(=c•-l), with the WSW-ENE horizontal CRBG occurred300 km to 400 km north of the supposed
axis as c•-3, the geometrythat would createvertical NNW- hotspottrack and, second,the vast majority of the CRBG
SSE extensionalfissures,a plane of weaknesswhich was magmahas a lithospheric,ratherthan an asthenospheric or
used by the magma in forcing its way to the surfacefrom potentially hotspot-related, geochemical and isotopic
reservoirs at the base of the crust. signature.Nevertheless,the associationof hotspots
 HOOPER 21

continentalflood basaltsin generalis well established

[e.g., magmatism
changed
significantly
[Hooper,1990;Hooper
Duncan, 1978; Morgan, !981; Macdougall, 1988; White and Hawkesworth,1993].
and McKenzie,1989; Campbelland Griffiths,1990;Duncan Given that the modelof Whiteand McKenzie[1989,
and Richards,1991; Basu et al., 1993] and the geographic 1995]requiresa thinnerthannormallithosphere,
in addition
displacement of the plume from the centerof eruptionmay to thehightemperaturesof themantleplume,to accountfor
simplyreflectthe ability of the risingmagmato seekout the the exceptionallylarge volumesof melt involved in flood
weakestzonesin the continentallithosphere[Thompsonand basalt provinces, those authors emphasized contempo-
Gibson, 1991]. raneous volcanism and extension. However, the field
Recentwork along the southernmargin of the Columbia evidence on the Columbia Plateau, the Deccan, and the
River province [Pierce and Morgan, 1992; Zobak et al., SiberianTraps indicatesthat extensionalstrainwas absent
1994; Camp, 1995] suggestedthat the first magmatic immediately before and during the main flood basalt
manifestationof the hotspot erupted as a south-to-north eruptions[Hooper, 1990; Renneand Basu, 1991]. The same
linearbelt of Imnaha-likebasaltadjacentto the westernside is true for the Karoo province [Marsh and Eales, 1984;
of the lithosphericboundary(SZ in Figure 3), possibly Marsh et al., this volume] and, apparently, the Paranti
following a seriesof older (Eocene?)riffs formed by the [Turner et al, 1996]. Flood basalteruptionsseemto act as a
thinningof the extendingoceaniclithosphereaway from the catalystfor extension,leadingto modestgrabenstructures
older craton. In CampUs [1995] model,the plumeheadwas in someprovinces(CRBG, SiberianTraps) and facilitating
distortedparallel to the overriding,more competent,craton plate separationin others(Karoo, Deccan,Paranti,and NE
[seealsoGeist and Richards,1993] and the rapid ascentof Atlantic) 3-5 m.y. afterthe main magmatism.
the magmawas facilitatedalongthe weak boundariesin the We may conclude(1) that the large melt volumes of
accretedcollage,the "thinspots"of Thompsonand Gibson continentalflood basalt provincesrequire both a mantle
[1991]. Supportingevidencefor this model comes from plume and weakened or thinned lithosphere,(2) that the
recentrecognitionthat the "UnnamedIgneousComplex"of eruptions through the lithospheredirectly or indirectly
Kirtleman [1973] in the Malheur River Gorge of eastern trigger furtherextensionwhich may leadto plate separation,
Oregon (Figure 3) is a southernextensionof Imnaha and and (3) that this subsequentextensionis accompaniedby
Grande Ronde Basalts and that the tholeiites of Steens magmatismwhich is of small volume and of more varied
Mountain are chemically similar to the earliest Imnaha composition(calc-alkalicto bimodal and alkalic) than flood
flows [Evans, 1990; Hooper et al., 1995c]. basalt volcanism.Both a weakenedcontinentallithosphere
Using similar arguments,the subsequent overridingof the and a mantle plume are essentialto the formation of a
hotspotby the westward motion of the thicker and more continental flood basalt province, but the extension and
competentNorth American plate, would have funneledthe possibleplate separation,which are so often associated with
magmaalongthe ancientandweakertectonicboundarythat flood basalt volcanism, follow and appear to be the
underlies the eastern Snake River Plain and which consequence ratherthan the causeof mantleplume activity,
convenientlyparallelsthe approximatetrace of the hotspot in the manneroriginally advocatedby Morgan [ 1981].
beneath the plate. For an alternativeexplanationof the
suddenshift in the center of the hotspotmagmatismsee CONCLUSIONS
Geistand Richards(1993).
Whiteand McKenzie [ 1989] arguedpersuasivelythat the The three major tectonic factors most pertinent to the
large volume and rapid eruption rates of the CRBG and origin of the CRBG are (1) the Yellowstonehotspot;(2)
other continentalflood basaltsrequirea mantle hotspotin back-arc spreading and consequent thinning of the
the literal sense.The chemical and isotopicsimilarity of continentallithosphere;and (3) physical proximity of the
early (RC type Imnaha) flows to Hawaiian tholeiites flood basalt eruptionto the tectonicboundarybetweenthe
[Hooper and Hawkesworth,1993] supportsthe presenceof thinner, denser,oceaniclithosphereof the accretedterranes
a mantleplume. However, the CRBG eruptionclearlytook and the thicker, more competentlithosphereof the old
place in an environmentwhere east-westextensionhad been North American plate, recently welded and thickened
well establishedfor 20 million years or more [e.g., White, furtherby intrusionof the Idaho batholith.
1992]. Extensionimmediatelyadjacentto the centerof the Almost all CRBG flows were eruptedfrom fissuresin the
CRBG eruptionwas minimal (perhapslessthan 1%), but thinner accretedterranesand the first eruptionsappear to
when the amount of extension on the south side of the OWL have followed a narrow north-southzone, possibly a pre-
increasedabout 15 Ma to create typical basin and range Miocene graben, adjacentto the old crustalsuture.Feeder
faults and grabens,the physicaland chemicalnature of the dike orientationis NNW in all cases(Nevada Riff,
22 COLUMBIA RIVER FLOOD BASALT PROVINCE

Mountain, and the Chief Josephdike swarm).This suggests Hawkesworth,1993]. There is evidencefor andagainstboth
that the presentnorth-southto locally NNE alignmentof these models. In contrast, the Imnaha and the Eckler
thesezonesresultedfrom subsequent increasesin basinand Mountain Basaltscan both be interpretedas evolving from
rangeeast-westextensionto the south,controlledby right- their own singlesourceby a combinationof partialmelting
lateraldisplacementalongWNW megashears, as arguedby and gabbrofractionationprocesses.Magma mixing during
Lawrence [1976] and supportedby more detailed work or just prior to eruptioncan be demonstrated in the Saddle
[Hooper and Conrey, 1989; Mann, 1989; Mann and Meyer, MountainsBasaltin at leasttwo instancesand suchmagma
1993]. mixing may have been a common occurrenceduring the
The large volumes of tholeiite magma require the high largeeruptionsof GrandeRondeBasalt.
mantle temperaturesof a hotspot and the similarities There appearslittle immediateprospectof resolvingthe
between the earliest Imnaha Basalt and Hawaiian tholeiites relativesignificanceof the rolesof the crustandan enriched
supportthe derivationof theseearlymagmasfrom a mantle subcontinentallithosphericmantle in the generationof the
plume. CRBG. Oxygenisotopedataon the individualphasesof the
In contrast,the continentallithosphericsignatureapparent phyricPictureGorgeBasaltmight help to confirmthe data
in all but these few early CRBG flows suggestseither currentlyavailableon wholerock samples.In contrast,there
significantcrustalcontamination and/orthe entrainmentof is little doubt that a detailedstudyof the relationsbetween
enriched subcontinentallithosphericmantle (SCLM) into structures andmagmatismin the little-knownareaof eastern
the plume head. The lithosphericcontributionvaried with Oregonat the southernlimits of the ColumbiaRiver basalt
position and time and divides the CRBG into three province would enhance our understandingof the
fundamentalsubgroups,facts which appearto this author relationshipsbetween continental flood basalts, mantle
more readily explainedby the incorporationof threedistinct plumes,and lithosphericextension.Suchan investigation,
types of SCLM than by contaminationby three different employing chemical-stratigraphicmapping and paleo-
compositionsof the samecrust. First, the Imnaha and the magnetictechniques,has the potentialto resolvesome of
GrandeRondeBasaltsappearto have beenderiveddirectly thesefundamentalproblemsat relativelylow cost.A fuller
from the mantle plume entraining and mixing with a understandingof the relative timing of extensional
lithosphericcomponentwhich was enrichedby subduction structuresand volcanismin easternOregoncould also help
processesin the mid-Proterozoic. Second, the Picture solve the contentious problems associatedwith dis-
Gorge Basalt, which erupted further west with a volume placementalong such regional featuresas the Olympic
almost two orders of magnitudeless than the combined Wallowa Lineament and the Brothers fault zone.
Imnaha and Grande Ronde Basalts,may have been derived The apparent conflict between White and McKenzie's
by decompressionalmelting from the more recently [1995] model and field evidencefor the relativetiming of
enriched SCLM below the Mesozoic accreted terranes the flood basalteruptionsand extensioncan be resolvedby
during basin and range extension.And third, the later employingthe Thompsonand Gibson [1991] conceptof
Saddle Mountains Basalt erupted as the North American thinspotsin the continentallithosphere. It is the presenceof
craton overrodethe hotspot,so melting an SCLM at the a weak or thin lithosphere,not contemporaneous extension,
base of the North American plate, enriched in the late whichappearscriticalto the White andMcKenziethesis.In
Archaen. the caseof the Columbia River basalts,suchthinspotswere
These mantle-generatedmagmas appear to have been almostcertainlyavailablebecauseof earlierextension.Thin
storedin large reservoirsnear the base of the crust from lithosphere,combinedwith the high temperaturesof the
where they were periodicallyeruptedrapidly to the surface plume, could have created the unusually large melt
through fissures whose strong NNW-SSE orientation volumes. The geologic evidence indicatesthat extension
reflects the prevailing regional ENE-WSW extensional increasedfollowing the huge tholeiitic eruptions,and as
strain. The exceptionallylarge individual eruptionsmay extension increased the nature of the magmatism also
have been triggered by infusion of a new batch of melt changedto that more typical of extensionalcontinental
enteringthe deepmagmareservoirsfrom the mantle. The terranes.The relative timing of extensionand volcanism
compositionalvariations in the dominant Grande Ronde derived from the field evidence on the Columbia Plateau
Basalt have been explained either by a combinationof and severalother flood basaltprovincespointsto the mantle
crystalfractionationand crustalassimilation[Carlsonand plume and consequentflood basalt eruptionacting as a
Hart, 1988] or by mixing of plume-derivedmagma and stimulantto increasedextension,which in some provinces
variable proportions of an SCLM, with only minor has led to the separationof continents.The detailed
fractionation and crustal assimilation [Hooper and knowledgeavailablefor the ColumbiaPlateausuggests

continentalbreakupwas a potentialconsequence
and not a
precursorof flood basaltmagmatism.
Acknowledgments. I am particularlygratefulfor the careful and Brandon, A.D., and G. G. Goles, A Miocene subcontinental
thoughtfulreviews of an earlier manuscriptby Anita Grunder,
Alan Brandon,and JohnMahoneywhichcontributed to significant
improvements.I would alsolike to acknowledgethe many yearsof
encouragement and help from colleagues in the Geology Brandon, A. D., P. R. Hooper, G. G. Goles, and Ro Lambert,
Departmentand the GeoanalyticalLaboratoryat WashingtonState
University.
REFERENCES
Anderson,J. L., The structural
geologyandagesof deformation of
a portionof the southwest ColumbiaPlateau,Washington and
Oregon, Ph.D. thesis,283 pp., Univ. SouthernCalifornia, Los
Angeles, 1987.
Armstrong, R. L., W. H. Taubeneck,and P.O. Hales, Rb/Sr and
K/Ar geochronometryof the Mesozoic granite rocks and their
Sr isotope composition,Oregon, Washingtonand Idaho, Geol.
Soc. Am. Bull., 88, 397-411, 1977.
Axelrod, D. I., Tertiary florasandtopographichistoryof the Snake
River basin, Idaho, Geol. Soc.Am. Bull., 79, 713-733, 1968.
Bailey, D. G., Geochemistryandpetrogenesisof Miocenevolcanic
rocksin the PowderRiver VolcanicField, northeasternOregon,
Ph.D. thesis,341 pp., WashingtonStateUniv., Pullman, 1990.
Bailey, M. M., Revisions to stratigraphicnomenclatureof the
Picture Gorge Basalt Subgroup,ColumbiaRiver Basalt group,
in Volcanism and Tectonism in the Columbia River Flood-
Basalt Province, Spec. Pap. 239, edited by S. P. Reidel and P.
R. Hooper, pp. 67-84, GeologicalSocietyof America, Boulder,
CO, 1989a.
Bailey, M. M., Evidencefor magmarechargeand assimilationin
the Picture Gorge Basalt Subgroup, Columbia River Basalt
Group, in Volcanism and Tectonismin the Columbia River
Flood-Basalt Province, Spec.Pap. 239, edited by S. P. Reidel
and P. R. Hooper, pp. 343-356, GeologicalSocietyof America,
Boulder, CO, 1989b.
Baksi,A. K., Reevaluationof the timing anddurationof extrusion
of the Imnaha, Picture Gorge and Grande Ronde Basalts,
Columbia River Basalt Group, in Volcanismand Tectonismin
the Columbia River Flood-Basalt Province, Spec. Pap. 239,
edited by S. P. Reidel and P. R. Hooper, pp. 105-112,
GeologicalSocietyof America,Boulder,CO, 1989.
Barrash, W., J. G. Bond, and R. Venkatakrishnan, Structural
evolutionof the ColumbiaPlateauin Washingtonand Oregon,
Am. d. Sci., 283, 897-935, 1983.
Basu, A. R., P. R. Renne, D. K. DasGupta,F. Teichmann, R. J.
Poreda,
Earlyandlatealkaliigneous
pulses
anda high3He
Plume origin for the Deccan flood basalts,Science,261, 902-
905, 1993.
Beeson, M. H., K. R. Fecht, S. P. Reidel, and T. L. Tolan,
Regionalcorrelationswithin the FrenchmanSpringsmemberof
the ColumbiaRiver BasaltGroup:New insightsinto the Middle
Miocenetectonicsof northwestern Oregon,Oreg. Geol., 47, 87-
96, 1985.
Beeson,M. H., R. Perttu,and J. Perttu,The origin of the Miocene
basalts of coastal Oregon and Washington: An alternative
hypothesis,Oreg. Geol., 41, 159-166, 1979.
HOOPER 23
Bentley, R. D., and R. S. Cockerham,The stratigraphyof the
Picture Gorge Basalt (PGB), north centralOregon, Geol. Soc.
Am. A bstr. with Programs,5, 9, 1973.
plume in the Pacific Northwest: Geochemicalevidence, Earth
Planet. ScioLett., 88, 273-283, 1988.
Evaluating crustal contamination in continental basalts: the
isotopic composition of the Picture Gorge Basalt of the
Columbia River Basalt Group, Contrib. Mineral. Petrol., 114,
452-464, 1993.
Byerly, G., and D. A. Swanson,Invasive Columbia River basalt
flows along the northwestmargin of the Columbia Plateau,
north-centralWashington,Geol.Soc.Am. Abstr.Programs,10,
98, 1978.
Camp,V. E., Mid-Miocenepropagationof the Yellowstonemantle
plume head beneaththe Columbia River basalt sourceregion,
Geology,23, 435-438, 1995.
Campbell, C. D., and S. I. Runcorn, Magnetizationof the
Columbia River basalt in Washington and north Oregon, d.
Geophys.Res.,61, 449-458, 1951.
Campbell,I. H., and R. W. Griffiths, Implicationsof mantleplume
structure for the evolution of flood basalts, Earth Planet. Sci.
Lett., 99, 79-93, 1990.
Campbell, N. P., Structural and stratigraphicinterpretationof
rocks under the Yakima fold belt, Columbia Basin, based on
recent surface mapping and well data, in Volcanism and
Tectonismin the Columbia River Flood-BasaltProvince,Spec.
Pap. 239, editedby S. P. Reidel andP. R. Hooper,pp. 209-222,
GeologicalSocietyof America,Boulder,CO, 1989.
Carlson, R. W., Isotopic constraintson Columbia River flood
basalt genesis and the nature of the subcontinentalmantle.
Geochim. Cosmochim.Acta, 48, 2357-2372, 1984.
Carlson,R. W., Physicaland chemicalevidenceon the causeand
source characteristicsof flood basalt volcanism, Aust. d. Earth
Sci., 38, 525-544, 1991.
Carlson,R. W., and W. K. Hart, Crustal genesison the Oregon
Plateau,d. Geophys.Res.,92, 6191-6206, 1987.
Carlson, R. W., and W. K. Hart, Flood basalt volcanism in the
Pacific Northwestern United States, in Continental Flood
Basalts, edited by J. D. Macdougall, pp. 35-61, Kluwer
Academic Publishers,Dordrecht, 1988.
Carlson, R. W., G. W. Lugmair, and J. D. Macdougall, Columbia
River volcanism:the questionof mantleheterogeneityor crustal
contamination, Geochim. Cosmochim. Acta, 45, 2483-2499,
1981.
Chamberlain,V. E., and R. St. J. Lambert,Lead isotopesand the
sourcesof the Columbia River Basalt Group, d. Geophys.Res.,
99, 11805-11818, 1994.
Choiniere,S. R., and D. A. Swanson,Magnetostratigraphy
and
correlation of Miocene basalts of the northwestern American
coast and Columbia Plateau, Am. d. Sci., 279, 755-777, 1979.
Church,S. E., Geneticinterpretation
of lead-isotope
datafromthe
ColumbiaRiverBasaltgroup,Oregon,Washington andIdaho,
Geol. Soc. Bull., 96, 676-690, 1985.
Draper,D. S., Late Cenozoicbimodalmagmatism
in the
24 COLUMBIA RIVER FLOOD BASALT PROVINCE

Basinand RangeProvinceof southeastern

Oregon,J. Volcanol. Hooper, P. R., Physicaland chemicalconstraintson the evolution
Geotherm. Res., 47, 299-328, 1991. of the ColumbiaRiverbasalt,Geology,12, 495-499, 1984.
Duncan, R. A., Geochronologyof basaltsfrom the Nintyeast Hooper,P. R., A caseof simplemagmamixing in the Columbia
Ridge and continentaldispersionin the easternIndianOcean,J. River Basalt Group: the Wilbur Creek, Lapwai and Asotin
Volcanol. Geotherm.Res., 4, 283-305, 1978. flows, SaddleMountainsFormation, Contrib. Mineral. Petrol.,
Duncan, R. A., and M. A. Richards,Hotspots,mantle plumes, 91, 66-73, 1985.
flood depositsand true polar wander,Rev. Geophys.,29, 31, Hooper, P. R., The Columbia River Basalt, in ContinentalFlood
1991. Basalts, edited by J. D. Macdougall, pp. 1-33, Kluwer
Evans, J. G., Geology and mineral resourcesof the Jonesboro AcademicPublishers,Dordrecht, 1988a.
quadrangle,Malheur County, Oregon,Oreg. Dept. Geol. Min. Hooper, P. R., Crystalfractionationand recharge(RFC) in the
Ind., GMS-66, scale 1:24,000, 1990. American Bar flows of the Imnaha Basalt, Columbia River
Fitton, J. G., D. James,P. D. Kempton,D. S. Omerod,and W. P. basaltgroup,J. Petrol.,29, 1097-1118, 1988b.
Leeman,The role of lithosphericmantlein the generationof late Hooper,P. R., The timingof crustalextensionandthe eruptionof
Cenozoic basic magmas in the western United States, in continentalflood basalts,Nature, 345, 246-249, 1990.
Oceanic and Continental Lithosphere.' Similarities and Hooper,P. R., andV. E. Camp,Deformationof the southeast part
Differences,Spec. Vol., editedby I. A. Menziesand K. G. Cox, of the ColumbiaPlateau,Geology,9, 323-328, 1981.
pp. 331-350, Journalof Petrology,Oxford, 1988. Hooper,P. R., andR. M. Conrey,A modelfor thetectonicsetting
Fleck, R. J., and R. E. Criss, Strontiumand oxygenisotope of the Columbia River Basalt eruptions,in Volcanismand
variationsin Mesozoic and tertiary plutonsof centralIdaho, Tectonismin the ColumbiaRiver Flood-BasaltProvince,Spec.
Contrib. Mineral. Petrol., 90, 291-308, 1985. Pap. 239, editedby S. P. ReidelandP. R. Hooper,pp. 293-306,
Fox, K. F., and M. E. Beck, Paleomagneticresultsfrom Eocene GeologicalSocietyof America,Boulder,CO, 1989.
volcanicrocksfrom northeastern Washingtonand the Tertiary Hooper,P. R., and C. J. Hawkesworth,Isotopicand geochemical
tectonicsof the Pacific Northwest. Tectonics,4, 323-341, 1985. constraintson the origin and evolutionof the ColumbiaRiver
Fruchter, J. S., and S. F. Baldwin, Correlationsbetween dikes of basalt,J. Petrol., 34, 1203-1246, 1993.
the Monumentswarm,centralOregonandPictureGorgeBasalt Hooper, P. R., and D. A. Swanson,The Columbia River Basalt
flows, Geol. Soc.Am. Bull., 86, 514-516, 1975. group and associatedvolcanic rocks of the Blue Mountains
Fuller, R. E., The aqueouschilling of basaltic lava on the Province,U.S. Geol. Surv.Prof Pap. 1437, 63-99, 1990.
Columbia River Plateau,Am. J. Sci., 21, 281-300, 1931. Hooper, P. R., C. R. Knowles, and N. D. Watkins,
Gehrels,G. E., The geologyof the westernhalf of the La Grande Magnetostratigraphyof the Imnaha and GrandeRonde Basalts
basin, northeasternOregon, M.S. thesis,97 pp., Univ. of in the southeastpart of the ColumbiaPlateau,Am. J. Sci, 279,
SouthernCalifornia,Los Angeles,1981. 737-745, 1979.
Geist,D., and M.A. Richards,Origin of the ColumbiaPlateauand Hooper, P. R., W. D. Kleck, C. R. Knowles, S. P. Reidel, andR. L.
Snake River Plain: Deflection of the Yellowstoneplume, Thiessen,Imnaha Basalt, ColumbiaRiver Basalt Group,J.
Geology,21, 789-792, 1993. Petrol., 25, 473-500, 1984.
Hart, W. K., and R. W. Carlson,Tectoniccontrolson magma
Hooper,P. R., D. G. Bailey, and G. A. McCarley Holder,Tertiary
genesis and evolution in the northwesternUnited States, J.
calc-alkalinemagmatismassociated with lithosphericextension
Volcanol. Geotherm.Res., 32, 119-135, 1987.
in the Pacific Northwest,J. Geophys.Res., 100, 10, 303-10,
Helz, R. T., Phaserelationsof basaltsin their meltingrangeat 319, 1995a.
PH20=5 kb as a functionof oxygenfugacity,J. Petrol, 14, 249-
302, 1973. Hooper, P. R., B. A. Gillespie, and M. E. Ross, The Eckler
Mountain Basalts and associatedflows, Columbia River Basalt
Helz, R. T., Phaserelationsof basaltsin their meltingrangesat
group, Can. J. Earth Sci., 32, 410-423,1995b.
PH20=5 kb. Part II. Melt Compositions,J. Petrol., 17, 139-193,
1976. Hooper, P. R., C. J. Hawkesworth, K. Lees, M. Francis, J.
Helz, R. T., andT. L. Wright, Inferredpetrologyof the sourceof Johnston, and B. Binger, The southern extension of the
the Yakima Basalt subgroup,(abstract),IAVCEI Meeting, Columbia River basalts:Tectonic implications(abstract),Eos
Iceland, 1982. Trans.AGU, 76, 46, Fall Meeting Suppl.,F698, 1995c.
Hergt, J. M., B. W. Chappell,M. T. McCulloch,I. McDougall, James,D. E., The combineduseof oxygenandradiogenic
isotopes
and A. R. Chivas,J. Petrol., 30, 841-883, 1989. as indicatorsof crustalcontamination,Ann. Rev. Earth. Planet.
Ho, A., and K. V. Cashman,Geothermometry of the Ginkgoflow, Sci. 9, 311-344, 1981.
ColumbiaRiver Basalt group, Eos Trans.AGU, 76, 46, Fall Kim, K., S. A. Dischler,J. R. Aggson,andM.P. Hardy,The state
Meeting Suppl.,p. F679, 1995. of in situ stressesdeterminedby hydraulicfracturingat the
Hon, K., J. Kauahikaua, R. Denlinger, and K. Mackay, Hanford site, Richland, Washington, Rockwell Hanford
Emplacement and inflation of Pahoehoe sheet flows - OperationsReport RHO-B W-ST-73-P, 1986.
Observations and measurements of active lava flows on Kilauea Kittleman,L. R., Guide to the geologyof the Owyheeregionof
Volcano,Hawaii, Geol. Soc.Am. Bull., 106, 351-370, 1994. Oregon,Univ. Oreg.Mus.Nat. Hist., Bull., 21, 45 pp., 1973.
Hooper, P. R., The Columbia River basalts,Science,215, 1463- Lambert, R. St J., V. E. Chamberlain,and J.G. Holland, Ferro-
1468, 1982. andesitesin the GrandeRondeBasalt:their composition

significancein studiesof the origin of the ColumbiaRiver
Basaltgroup,Can. •I. Earth Sci., 32, 424-436, 1995.
Lawrence,R. D., Strike-slipfaultingterminatesBasinand Range
provincein Oregon,Geol.Soc.Am.Bull., 87, 846-850,1976.
Leeman,W. P. Radiogenictracersappliedto basaltgenesisin the
SnakeRiver Plain-YellowstoneNational Park region- evidence Neim, A. R., and W. A. Neim, Oil and investigationof the Astoria
for a 2.7 B.Y. old upper mantle keel. Geol. Soc. Am. Abstr.
Prog., 7, 1165, 1975.
Lees, K. R., Magmaticand tectonicchangesthroughtime in the
Neogenic volcanic rocks of the Vale area, Oregon,North Osawa, M., and G. G. Goles, Trace element abundances in
WesternUSA. Ph.D. thesis,284 pp., Open University,Milton
Keynes, U.K., 1994.
Lindgren,W., The gold belt of the Blue Mountainsof Oregon,
U.S. Geol. Surv. 22ndAnn. Rept., 551-776, 1901.
Lofgren,G. Experimental studiesonthedynamiccrystallization of Pearce, J. A., The role of subcontinentallithospherein magma
silicatemelts,in Physicsof MagmaticProcesses, editedby R.
B. Hargraves, pp. 487-551, Princeton University Press,
Princeton, N.J., 1980.
Long, P. E., and B. J. Wood, Structures, texturesand cooling Pfaff, V. J., and M. H. Beeson, Miocene basalt near Astoria,
histories of Columbia River basalt flows, Geol. Soc. Am. Bull.,
97, 1144-1155, 1986.
Macdougall,J. D., Continentalflood basaltsandMORB: A brief
discussionof similaritiesand differencesin their petrogenesis,
in ContinentalFlood Basalts, edited by J. D. Macdougall, pp.
331-341, Kluwer Academic Publishers,Dordrecht, 1988.
Mankinen, E. A., E. E. Larson, C. H. Gromme, M. Prevot, and R.
S. Coe, The SteensMountain (Oregon) geomagneticpolarity
transition 3. Its regional significance.or. Geophys.Res., 92,
8057-8076, 1987.
Mann, G. M., Seismicity and late Cenozoic faulting in the
Brownlee Dam area---Oregon-Idaho:A preliminary Report,
U.S. Geol. Surv. Open-FileReport,89-429, 46 pp., 1989.
Mann, G. M., and C. E. Meyer, Late Cenozoic structureand
correlation to seismicity along the Olympic-Wallowa
Lineament, northwesternUnited States, Geol. Soc. Am. Bull.,
105, 853-871, 1993.
Marsh, J. S., and H. V. Eales, The chemistryand petrogenesisof
igneousrocksof theKarooCentralArea,southern
Africa, Geol.
Soc. S. Afr. Spec.Publ. 13, 27-68, 1984.
Merriam, J. C., A contributionto the geology of the John Day
basin,California Univ. Dept. Geol. SciencesBull., 2, 269-314,
1901.
Mohl, G. B., and R. L. Thiessen,Gravity studiesof an islandarc -
continentsuturein west-centralIdaho and adjacentWashington, Reidel,S. P., andT. L. Tolan,Eruptionandemplacement
in Geologyof the Blue MountainsRegion of Oregon,Idaho,
and Washington:Petrology and TectonicEvolution of Pre-
TertiaryRocksof the Blue MountainRegion,Prof. Pap. 1438,
edited by T. L. Vallier and H. C. Brooks, pp. 497-516, U.S.
GeologicalSm:vey,Washington, D.C., 1995.
Morgan,W. J., Hotspottracksandthe openingof theAtlanticand
Indian Oceans,in The Sea, Vol. 7, edited by C. Emiliani, pp.
443-487, Wiley-Interscience,New York, 1981.
Neim, A. R., andW. A. Neim, Oil andinvestigation of the Astoria
basin,Clastopandnorthernmost Tillamookcounties,northwest
Oregon,Oil andGasInvestigations, 14, Oreg.Dept. Geol.Min.
Ind., Oil Gas. Invest. 14, 1985.
Neim A. R., B. K. McKnight, and H. J. Meyer, Sedimentary,
HOOPER 25
volcanic,and tectonicframeworkof forearcbasinsand the Mist
gas field, northwestOregon,in GeologicPt'eldTrips in the
Pacific Northwest, edited by D. A. Swansonand R. A.
Haugerud,Annual Meeting, IF 1-42, GeologicalSocietyof
America, Boulder, CO, 1994.
basin,Clastopandnorthernmost Tillamookcounties,northwest
Oregon,Oil andGasInvestigations, 14, Oreg.Dept. Geol.Min.
Ind., Oil Gas. Invest. 14, 1985.
Columbia River basalts,in Proceedings,2nd Columbia River
Basalt Symposium,editedby E. H. Gilmour and D. Stradling,
pp. 55-71, EasternWashingtonState College Press,Cheney,
1970.
genesisat active continentalmargins. In ContinentalBasalts
and Mantle Xenoliths,editedby C. J. Hawkesworth,and M. J.
Norry, pp. 230-249, Shiva,Norwich, 1983.
Oregon; Geophysicalevidencefor Columbia Plateau origin,
Geol. Soc.Am. Spec.Paper in Volcanismand Tectonismin the
ColumbiaRiver Flood-BasaltProvince,Spec.Pap. 239, edited
by S. P. Reidel and P. R. Hooper, pp. 143-156, Geological
Societyof America, Boulder,CO, 1989.
Pierce,K. L., andL. A. Morgan,The trackof the Yellowstonehot
spot:Volcanism,faultinganduplift,Geol.Soc.Am.Mem.179,
1-53, 1992.
Prestvik, T., and G. Goles, Commentson petrogenesisand the
tectonicsettingof the ColumbiaRiver basalts,Earth Planet.Sci
Lett., 72, 65-73, 1985.
Price, E. H., and A. J. Watkinson,Structuralgeometryand strain
distributionwithin easternUmtanum fold ridge, south-central
Washington,in Volcanism and Tectonism in theColumbiaRiver
Flood-BasaltProvince,Spec.Pap. 239, editedby S. P. Reidel
andP. R. Hooper,pp. 265-282,GeologicalSocietyof America,
Boulder, CO, 1989.
Reidel, S. P., The Saddle Mountains - The evolution of an
anticline in the Yakima fold belt, Am. • Sci., 284, 942-978,
1984.
Reidel, S. P., and K. R. Fecht,The Huntzingerflow; evidenceof
surfacemixingof theColumbiaRiverbasaltanditspetrogenetic
implications,Geol.Soc.Am.Bull., 98, 664-677, 1987.
of flood
basalt: An example from the large-volume Teepee Butte
Member, Columbia River Basalt Group, Geol. Soc. Am. Bull.,
104, 1650-1671, 1992.
Reidel, S. P., and T. L. Tolan, Late Cenozoic structureand
correlation to seismicity along the Olympic-Wallowa
Lineament, northwestern United States:Discussionand reply,
Geol. Soc. Am. Bull., 106. 1634-1638, 1994.
Reidel, S. P., T. L. Tolan, P. R. Hooper, M. H. Beeson,K. R.
Fecht, R. B. Bentley,and R. L. Anderson,The GrandeRonde
Basalt,ColumbiaRiver basaltgroup;stratigraphic descriptions
and correlations in Washington, Oregon and Idaho, in
Volcanism and Tectonism in the Columbia River Flood-Basalt
Province,Spec.Pap. 239, editedby S. P. Reidel and P.
26 COLUMBIA RIVER FLOOD BASALT PROVINCE

Hooper, pp. 21-54, Geological Society of America, Boulder, Basalts, in Proceedings, Geological Society of America,
CO, 1989. Cordilleran SectionMeeting,Pullman,1978 (guide).
Reidel, S. P., N. P. Campbell,K. R. Fecht, and K. A. Lindsey, Swanson,D. A., and T. L. Wright, Guide to geologic field trip
Late Cenozoic structure and stratigraphy of south-central betweenLewiston, Idaho, and Kimberly, Oregon, emphasizing
Washington,in Regional Geology of WashingtonState, Bull. the Columbia River Basaltgroup,in Guidesto Some Volcanic
80, pp. 159-180, WashingtonDepartmentof NaturalResources, Terranes in Washington, Idaho, Oregon, and Northern
Division of GeologyandEarthResources, Olympia, 1994. California, Circ. 838, editedby D. A. Johnstonand J. Nolan,
Renne, P. R., and A. R. Basu, Rapid eruptionof SiberianTraps pp. 1-16, U.S. GeologicalSurvey,Washington,D.C., 1981.
flood basaltsat the Permo-Triassic boundary,Science,253, 176- Swanson,D. A., and T. L. Wright, Geologicmap of the Wenaha
179, 1991. TucannonWilderness,Washingtonand Oregon, scale 1:48,000,
Richards, M. A., R. A. Duncan, and V. E. Courtillot, Flood basalts U.S. Geol. Surv.Misc. Field StudiesMap MF-1536, 1983.
and hotspottracks:Plume headsand tales, Science,246, 103- Swanson,D. A., T. L. Wright, andR. T. Helz, Linear vent systems
107, 1989. and estimatedratesof magmaproductionand eruptionfor the
Russell,I. C., A geologicalreconnaissance in centralWashington, Yakima Basalt on the Columbia Plateau,Am. d. Sci., 275, 877-
U.S. Geol. Surv.Bull., 108, 108 pp., 1893. 905, 1975.
Russell,I. C., Geology and water resourcesof Nez PerceCounty, Swanson,D. A., T. L. Wright, P. R. Hooper, and R. D. Bentley,
Idaho, U.S. Geol. Surv. Water-Supply Paper, 53, 54 pp., 1901. Revisionsin stratigraphicnomenclatureof the Columbia River
Rytuba, J. J., and E. H. McKee, Peralkaline ash-flow tuffs and BasaltGroup, U.S. Geol. SurveyBull. 1457-G, 59 pp., 1979.
calderasof the McDermitt volcanicfield, southeastOregonand Swanson,D. A., T. L. Wright, V. E. Camp, J. N. Gardner, R. T.
north centralNevada,d. Geophys.Res.,89, 8616-8628, 1984. Helz, S. M. Price, S. P. Reidel, and M. E. Ross, Reconnaissance
Schmincke, H. -U. Fused tuff and peppetitesin south-central geologicalmap of the ColumbiaRiver Basalt Group, Pullman
Washington,Geol. Soc.Am. Bull., 78, 319-330, 1967a. and Walla Walla quadrangles,southeastWashington and
Schmincke,H. -U, Stratigraphyand petrographyof four upper adjacentIdaho, scale 1:250,000, U.S. Geol. SurvoMisc. Inves.
Yakima Basalt flows in south-centralWashington,Geol. Soc. Ser• Map 1-1139, 1980o
Am. Bull., 78, 1385-1422, 1967b. Swanson,D. A., JoLoAnderson,V. E. Camp,P. R. Hooper, W. H.
Shaw,H. R., and D. A. Swanson,Eruptionand flow ratesof flood Taubeneck,and T. L. Wright, Reconnaissance geologicalmap
basalts,in Proceedings,2nd ColumbiaRiverBasaltSymposium, of the Columbia River Basalt group, northern Oregon and
editedby E. H. Gilmour and D. Stradling,pp. 271-299, Eastem western Idaho, scale 1:250,000, U.S. Geol. Surv. Open-File
WashingtonStateCollegePress,Cheney,1970. Report 81-797, 1981.
Smith, A.D., Back-arc convection model for Columbia River Swisher,
C. C., J. A. Ach,andW. K. Hart,Laserfusion4øAr/39Ar
basaltgenesis,Tectonophysics, 207, 269-285, 1992. datingof the type SteensMountainBasalt,southeastern Oregon
Smith,G. O., Geologyandwaterresources of a portionof Yakima and the age of the Steens geomagneticpolarity transition
County,Washington,U.S. Geol.Surv. WaterSupply-Paper,55, (abstract),Eos Trans.AGU, 71, 1296, 1990.
68 pp., 1901. Taubeneck, W. H., Dikes of the Columbia River basalt in
Smith, G. O., Description of the Ellensberg quadrangle, northeastern Oregon, western Idaho and southeastern
Washington,U.S. Geol. Surv. Geol. Atlas, Folio 86, 7 pp., Washington, in Proceedings, 2nd Columbia River Basalt
1903a. Symposium,editedby E. H. Gilmour and D. Stradling,pp. 73-
Smith, G. O., Geologyand physiography of centralWashington, 96, EasternWashingtonStateCollegePress,Cheney, 1970.
U.S. Geol. Surv.Prof Pap. 19, 1-39, 1903b. Thompson, R. N., and S. A. Gibson, Subcontinentalmantle
Snavely, P. D., N. S. MacLeod, and H. C. Wagner, Miocene plumes, hotspotsand pre-existingthinspots,d. Geophys.Res.,
tholeiitic basaltsof coastalOregonand Washingtonand their 148, 973-977, 1991.
relations to coeval basaltsof the Columbia Plateau, Geol. Soc. Tolan, T. L., S. P. Reidel, M. H. Beeson, J. L. Anderson, K. R.
Am. Bull., 84, 387-424, 1973. Fecht, and D. A. Swanson, Revisions to the estimatesof the
Snee,L. W., K. Lund, J. F. Sutter,D. E. Balcer, and K. V. Evans, aerial extent and volume of the ColumbiaRiver Basalt Group,
An 40Ar/39Archronicleof the tectonicdevelopmentof the in Volcanism and Tectonism in the Columbia River Flood-
SalmonRiver suturezone, westernIdaho, in Geologyof the Basalt Province,Spec.Pap. 239, editedby S. P. Reidel and P.
Blue MountainsRegion of Oregon,Idaho, and Washington: R. Hooper, pp. 1-20, Geological Society of America, Boulder,
Petrologyand TectonicEvolutionof Pre-TertiaryRocksof the CO, 1989.
Blue Mountain Region,Prof Pap. 1438, editedby T. L. Vallier
Turner, S., C. Hawkesworth,K. Gallagher,K. Stewart, D.Peate,
and H. C. Brooks, pp. 359-414, U.S. Geological Survey, and M. Mantovani, Mantle plumes,flood basalts,and thermal
Washington,D.C., 1995. models for melt generationbeneathcontinents:Assessmentof
Sobczyk, S. M., Crustalthicknessand structuresof the Columbia conductive heating model and applicationto the Paranti,d.
Plateau using geophysicalmethods,unpubl. Ph.D. thesis,208 Geophys.Res., 101, 11,503-11,518, 1996
pp., WashingtonStateUniversity,Pullman, 1994. Vallier, T. L., Petrology of the pre-Tertiaryigneousrocks in the
Swanson,D. A., and T. L. Wright, Field guide to field trip Blue mountainsprovince of Oregon, Idaho and Washington:
between Pasco and Pullman, Washington, emphasizing Implicationsfor the geologicevolutionof a complexislandarc,
stratigraphy,vent areasand intra-canyonflows of the Yakima in Geology of the Blue MountainsRegion of Oregon,

and Washington.'Petrology and TectonicEvolution of Pre-
Tertiary Rocks of the Blue Mountain Region, Prof. Pap. 1438,
edited by T. L. Vallier and H. C. Brooks, pp. 125-510, U.S.
GeologicalSurvey,Washington,D.C., 1995.
Walker, G. P. L., Compoundand simple lava flows and flood
basalts,Bull. l/olcanol., 35, 579-590, 1968.
Washington,H. S., Deccantraps and other plateaubasalts,Geol.
Soc. Am. Bull., 33, 765-804, 1922.
Waters, A. C., Stratigraphic and lithologic variations in the
Columbia River basalt,Am. d. Sci., 259, 583-611, 1961.
Watkins, N. D., and A. K. Baksi, Magnetostratigraphyand
oroclinal folding of the ColumbiaRiver, Steensand Owyhee
basaltsin Oregon,Washingtonand Idaho,Am. d. Sci., 274, 148-
189, 1974.
Wells, R. E., R. W. Simpson,R. D. Bentley,M. H. Beeson,M. T.
Mangan, and T. L. Wright, Correlationof Miocene flows of the
Columbia River Basalt group from the centralColumbia River
Plateau to the coastof Oregon and Washington,in Volcanism
and Tectonism in the Columbia River Flood-Basalt Province,
Spec. Pap. 239, edited by S. P. Reidel and P. R. Hooper, pp.
113-130, GeologicalSocietyof America,Boulder,CO, 1989.
HOOPER 27
White, R. S., Magmatism during and after continentalbreak-up,
Geol. Soc. LondonSpec.Publ., 68, 1-16, 1992.
White, R. S., and D. P. McKenzie, Magmatismat rift zones:The
generationof volcaniccontinentalmarginsand flood basalts.d.
Geophys.Res., 94, 7685-7729, 1989.
White, R. S., and D. P. McKenzie, Mantle plumes and flood
basalts.d. Geophys.Res.,100, 17543-17586,1995.
Wright, T. L., M. J. Grolier, and D. A. Swanson, Chemical
variation related to the stratigraphyof the Columbia River
basalt, Geol. Soc. Am. Bull., 84, 371-386, 1973.
Wright, T. L., M. Mangan,andD. A. Swanson,Chemicaldatafor
flows and feeder dikes of the Yakima Basalt subgroup,
ColumbiaRiver BasaltGroup,Washington,
OregonandIdaho,
andtheir bearingon a petrogenetic
model.U.S.Geol.Survey
Bull. 1821, 71 pp., 1989.
Zoback, M. L., E. H. McKee, R. J. Blakely, and G. A. Thompson,
The northernNevada rift: Regionaltectono-magmatic relations
and middle Miocene stressdirection, Geol. Soc. Am. Bull., 106,
371-382,
 Evolution of the Red SeaVolcanic Margin, WesternYemen

Martin Menzies, Joel Baker, Gilles Chazot,• and Mohamed Al'Kadasi 2

Departmentof Geology,RoyalHolloway Universityof London,Egham,Surrey,UnitedKingdom

The temporal relationshipsbetween rifting processesassociatedwith the

breakupof the Afro-Arabian Plate and the openingof the Red Sea have been
studiedon the upliftedvolcanicmarginin westernYemen. Excellentexposure
allows for the application of absolute and relative dating techniques.
Magmatism:
40Ar/39Ar
datingof feldspars
in thevolcanicrocksindicates
that
(a) volcanismbeganat 29-31 Ma, (b) a changefrom basicto silicic volcanism
occurredat ca. 29 Ma, (c) largevolumevolcanismlasteduntil 26.5 Ma, and (d)
eruptionratesdecreasedwith time. Exhumation:fissiontrack (FT) analysesof
apatitesfrom basementmetamorphicand sedimentaryrocks reveal that the
volcanicmarginwas rapidlyand deeplyexhumed,at the earliest,in the Oligo-
Miocene. This result is consistentwith a major erosionalbreak in the volcanic
stratigraphy,
bracketed
by40Ar/39Ar
datingashavingformedbetween
26 and19
Ma. Extension:field evidenceindicatesthat extensionwas largelypost-volcanic
becausethe volcanicstratigraphyis devoid of major faults. In highly extended
terranes
40Ar/39Ar
agesof "hanging
wall"volcanic
rocksandapatiteFT agesof
"footwall" basement rocks demonstrate that extension occurred in the late
Oligocene-earlyMiocene. Surfaceuplift: field evidencefor surfaceuplift may
exist in the changefrom marineto continentalsedimentation found in the pre-
volcanicsedimentary
rocks(>31 Ma). Exhumation,
the presumed
response
to
surfaceuplift, occurredsome6 m.y. later, allowingthat amountof time for the
volcanicmarginto be uplifted by >2-3 km. Integrationof field and laboratory
data reveals that the Red Sea volcanic margin evolved in response to
contemporaneous surface uplift and volcanism that predated extension and
exhumationby as muchas 5 m.y.

•Alsoat theEcoleNormaleSuperieure,
Lyon,France,andat INTRODUCTION
BRGM, Orleans, France
2AlsoattheUniversity
of Sana'
a, Sana'a,
Yemen. In the last twenty years many different rifling models
have been formulatedto explain the openingof the Red
Sea [e.g., McGuire and Bohannon, 1989; Menzies eta!.,
LargeIgneousProvinces:
Continental,
Oceanic,andPlanetary 1992; Davison et al., 1994] (Figure 1). The young riff
Flood Volcanism mountainsof Yemen provide a rare opportunityto study
GeophysicalMonograph100 extensiveonshoreexposureswhich encapsulate the rifting
Copyright1997by the AmericanGeophysical
Union processand, ultimately, ocean basin formation.
29
30 EVOLUTION OF RED SEA VOLCANIC MARGIN, WESTERN YEMEN

40øN 40øN

500 km

30 ø

.6 cm/yr 2 cm/yr
20 ø

MEN

ß. . crust
extended
continental
\x l•i '::'-•
,I oceaniccrust
• lateCenozoic
provinces volcanic
< 15 Ma
early.Cenozoic volcanic
:::111
I provinces
ß > 15 Ma

I I Phanerøzøic
I:i:i:i:i:i:il
Precambrian
basement
• Zagros Fold Belt
'•---'•transform
faults
• extensional rift trend
,• thrust belt
outline of Afro-Arabian
'.'.- 30OE:::::...... .:-:-..
•domal u lift

Figure1. TheYemenlargeigneous
province
located
on thesouthern
RedSeamargin[afterBakeret al., 1996a].
Ratesof platemovementand/orseafloorspreading
are
 MENZIES ET AL. 31

papershave focusedon the flood volcanismof the Yemen [Al'Subbary,1995]. Fieldworkprovidesan importantbasis
Highlands [e.g., Chiesa et al., 1983a,b; Manetti et al., for establishing the relativetiming of rift processes,but a
1991; Chazot and Bertrand, 1993; Baker et al., 1996a,b] more precise chronology is dependentupon absolute
which is believedto have contributions from deep-seated datingof boththe volcanicrocks(4øAr/39Ar) andthe main
mantleplumesor shallowerasthenospheric sources[e.g., period of crustal cooling (apatite fission track [FT]
Almond,1986; Camp and Roobol,1992]. Anotheraspect analyses).
of the evolution of the rifted margin that has resultedin
considerable controversyis the origin of the Afro-Arabian
dome [Cloos, 1939; Bohannonet al., 1989; Dixon et aL, MAGMATISM
1989] and its relationshipto upwellingmantle,volcanism,
surfaceuplift, andmountainbuilding. The Yemenmargin The exposedvolcanicstratigraphyis dominated,at the
hasclearlyundergonesurfaceuplift and exhumationin an base,by a thicksequence of basaltsoverlainby an equally
arid climatebut how theserelate,in time andspace,to the thick sequenceof rhyolites-ignimbrites (Figure 2). The
volcanicactivity and crustalextensionon the marginsis basaltsthickento the west and southand the uppermost
unknown. Whetheror not rifted continentalmarginsare sectionis dominatedby more silicic compositionsof
plate driven (passiverift) or plume driven (active rift) is rhyoliticairfall, ignimbriticpyroclasticdeposits,rhyolitic
one aspectof rift margin developmentthat continuesto lava flows, and volcaniclasticunits.
intrigueearth scientists. Although basalticmagmatismwas widespreadon the
In this paper, the methodologyused to unravel the southern Red Sea margin (Figure 1), it was not
timing of surface uplift, magmatism,exhumation,and contemporaneous with extension,as there are no syn-
extension
is summarised.
Onthesouthern
RedSeamargin volcanic extensional faults within the volcanic
in the Yemen Republic,Precambrianbasementis overlain stratigraphy. Minor normal faults have been observed.
by a cover of Mesozoic-Cenozoic limestones and Furthermore, there is little or no evidence for erosional
sandstones(-1000-2000 m) and Tertiary basalts and breaksthat would mark syn-volcanicexhumationwith the
rhyolites(-2500 m). Limited fossilevidenceindicatesthat exceptionof one near the top of the volcanicstratigraphy
the pre-volcanicsandstonesare Palaeocene-Eocene in age [Baker et al., 1996a]. The presenceof graniticintrusions

East (rift shoulder)

West
(rift
margin) •
1500m
•
C• >1000
m
proximal
mafic
•
andrhyolitic
pyroclastic
x •..... majorignimbrite
flowunits
[---I welded andunwelded
pyroclasticrocks
airfall
• basaltflows
-== +
intrusive % rhyoliticpyroclastic
.- 1000 m
graniteand % rocks; minor intercalated
basalt flows

• syenite
.c-

._

•'• ..
ß- 500
m 0.5-20mthick
ß?
• . basalt
flows
unconformity?

31-29 Ma
om

Tawilah Formation-
cross-bedded fluviatile
15ø23'30"N 15"24'52"N 15ø29'20" N
43v27'22"E 44ø09'52"E 44 ø25'08"E sandstone and siltstone
1050 rn asl 2300 rn asl 2370 rn asl

1.Wadi Lahima/Jabal 2. Sana'a 3. North-east

Hufash section section Sana'a section

Figure 2. Simplifiedvolcanostratigraphy
of the Yemen LIP [Baker, 1995; Baker et al., 1996b] along a west to east
traversefromthe Red Seacoastto Sana'a(Fig. 1). Note the lateralthinningof the LIP fromthe rift margin(west)to
the rift shoulder(east). In addition,note the changefrom basalticvolcanismat the baseto bimodal volcanismand
eventuallysilicic volcanismat the
32 EVOLUTION OF RED SEA VOLCANIC MARGIN, WESTERN YEMEN

at the surface within the upper part of the volcanic

stratigraphy(Figure 2) warrantsattentionbecausegranitic •: 1600 ===....J56
.....
::::•
(plateau
........
age) southwestern
Yemen
intrusionsnormally crystallizeat depthsof at least 1000 m ...........
•::::::•':•:':•::•:h-....
....

[Blakeyet al., 1994]. ß? 1200

• J40 (plateau age)
basalt +
4øAr/39Ar
Dating '• 800
rhyolite
•..... J14B(plateauage)
.......

The 4øAr/39Ar
datingof silicatephasesseparated
from .--.
iJ• 400 basalt
volcanic rocks defines a tightly constrainedsuite of late J73 (plateau age) ,--,,
Oligoceneages [Baker et al., 1994b, 1996a]. The silicic o MNY355
&388(isochron
i
ages)': ß i ß i ß

units are ideal for 4øAr/39Ardatingin that they contain

JB136 (step age) ::..•
potassium-richphases(sanidine,anorthoclase, amphibole, :•;i•::,i:;i•i•i::•::i:;
.......

J iha na (so ut h
and biotite). Whereas all the mineral samplesanalysed 150 :•:•'½•iili
ilia. Sana'a)
yieldedplateauages,whole-rockbasalticsamplesyielded basalt+ .;i:.:.'.•?B133
(plateau
age)
no plateau ages and are therefore of limited value in 1 oo
rhyolite
Yemen geochronologicalstudies. Mineral plateau and basalt
isochron ages, in the lowermost basaltic section, range
50
from 29.2 to 30.9 Ma [Baker et aL, 1996a] and, although
older basal flow agesare found in the southernpart of the
volcanicprovince,the youngestbasalflow agescomefrom ß i ß i ß JB124
(step
age)
-:-?
i ß i

the northeast(Figure 3). A progressive decreasein age is 600 ._,•._•_JB177._(iso_chro_n

a9e)_=19.87_+.07Ma
observedupwardsthroughthe volcanicstratigraphy [Baker unconformit• northeast
• .................
!.JB186 (step age) Sana'a
et al., 1996a] (Figure 3), but the magnitudeof this decrease
differs significantlyfrom that previouslyreportedfrom K- 400

Ar data [Civetta et al., 1978]. In the northeast,the oldest

rhyolites are ca. 29 Ma in age and the uppermostunits .c• JB171 (stepage)
have ages> 26 Ma. Baker et al. [1996a] evaluatedwhole- 2OO
• JB163 (plateau age)
rock 4øAr/39Arstep-heatingspectraand discussedthe basalt +

problemswith such data. Major problemsrelate to post- rhyolite JB148 (plateauage)

'l'

crystallisationloss of radiogenicAr from the groundmass

of fine-grainedandglassyrocks,andexcess
4øArand39Ar 800

recoil in somesamples.The 4øAr/39Ardataconstrain

the E I JB22711(plateauage)
.......
western
Sana'a
periodof volcanismto 30.9-26.5 Ma. In the southernpart • JB232 (plateauage)
• 600
of the Yemen large igneousprovince(LIP), Zumboet al.
,_

[ 1995] reportedagesof 28.9 and26.5 Ma for two basaltsat ß--• 400

':'i JB22811(plateau age)
the bottomand the top of a volcanicsection,respectively. ! JB221 (plateauage)
.:.

Thesedatareiteratethe conclusionsof Baker et al. [ 1996a]

.:.

03 200 basalt +
that the uppermost volcanic units within the
.--.

rhyolite
volcanostratigraphy have agesof 26.5 Ma. However,the o
JB216 (plateauage)
lowermost flow in the section studiedby Zumbo et al.
[ 1995] is by no meansbasalas it coincidesin age with the 600 JB335 (stepage) •
E Sana'a
basaltic-silicicperiod in northernYemen (i.e., 29-26 Ma). ...

•' JB326
(plateau
age)•
Zumboet al. [ 1995] alsoobtained4øAr/39Ar
plateauages • JB276(plateau
age)
on dikescuttingthe LIP, andtheserangefrom 25.4 to 16.1 ß• 4oo basalt+
J:: rhyolite
Ma, suggestingthat LIP formation may have been active .,_

for a longer period than is indicated by the volcano- 13. JB261 (plateauage)•
m 200 basalt
stratigraphycurrentlypreservedin this LIP. In summary: •
.--.
JB279 (plateauage) •
ß• JB281 (isochronage)
(a) Basalticmagmatismbeganthroughoutthe area 31-
29 m.y. ago and lastedfor a maximum of 2 m.y. with a •
0 , , i
?
,

cumulative thickness of 1000-1500 metres, 26 27 28 29 30

(b) A switch from basaltic to bimodal magmatism 4øAr/39Ar

ages
occurred at ca. 29 Ma,
 MENZIES ET AL. 33

EXTENSION
(c) Basalticand silicicmagmatismlastedfrom ca. 29-26
m.y. agowith a cumulativethicknessof 1000metres,
(d) The youngestexposedflows in the LIP have agesof Detailed mapping [e.g., Davison et al., 1994; Al'Kadasi,
26.9-26.5 Ma, 1995; At'Subbary, 1990, 1995; Baker, 1995] along
(e) Plutonicand hypabyssalactivity continuedfrom ca. traverses perpendicularto the present volcanic margins
25-16 Ma and one could argue that LIP formationmust helped constrain the timing of extension. In the Tihama
have accompaniedsuch activity, albeit at a reduced Plain east of A1 Hudaydah (Figure 4), domino-faultblocks
eruptionrate, contain basement metamorphic rocks (Precambrian)
(f) A hiatus in volcanismoccurredbetween26 and 19 overlain by a limestone-sandstone sequence(Jurassic-
Ma and is evidentas an erosionalunconformitywithin the Tertiary) which in turn is overlainby basaltsand rhyolites
volcanicsequence. (Oligocene) containing contemporaneous granites(Oligo-
Miocene?). Fieldwork reveals that extensionis largely
K-At Dating post-volcanic[Menzieset at., 1992; Davisonet at., 1994].
Volcanic rocks (e.g., basalts,basanites,and basaltic
trachyandesites) at the base of the flood basalt-rhyolite Pre-LIP SedimentsaryRocksand Hanging Wall Basalts
sequence in westernYemenweredatedusingconventional
K-Ar techniques.Sixty-threewhole-rockanalysesandone The 4øAr/39Ar
dates[Bakeret al., 1996a]for thevolcanic
hornblendeanalysisgave a range in K-Ar datesof 14.4- rocksallow evaluationof the timing of extension. Within
63.6 Ma [,4l'Kadasi, 1995], similar to that obtainedfrom the pre-volcanicsedimentaryrocks(>31 Ma) there are no
several published studies of the Yemen-EthiopianLIP angular unconformitiesto indicate extensionalepisodes
[e.g., Civettaet al., 1978; Manetti et al., 1991]. However, before the initiation of the LIP. Furthermore, within the
the applicability of these ages to primary petrogenetic volcanicstratigraphy(31-26 Ma) there is little or no fault
processes must be carefullyconsidered becauseof (1) the control on the distribution of the volcanic rocks, and no
large rangein radiogenic4øAr,(2) the highly variable angular unconformities,or faults, have been recognised
weight loss on ignition (0-6 %), and (3) the unequivocal within the volcanic stratigraphy, providing the most
evidencefor high-level contaminationprocesses[Baker et importantconclusionthat widespreadextensionmusthave
al., 1994a; 1996b]. largely postdatedthe Yemen LIP (i.e., <26 Ma). The
The 4øAr/39Ardata allow the K-Ar analysesto be hanging wall of rotatedfault blocks found west of Sana'a
screened,a processthat revealsthat the majority of the K- contains late Oligocene lava flows overlying Cenozoic-
Mesozoic sediments and in some cases Precambrian
Ar datesare of little value (J. Baker et al., Evaluatingthe
usefulnessof whole-rock K-Ar dating in a young large basement. This providesa maximum age for the onsetof
igneous province, ms. in progress).Whereas the most extension,placing the brittle extensionof the pre-existent
useful K-Ar dates are forthcoming from sampleswith geologyat <26 Ma.
>80% radiogenic4øAr,the screened
K-Ar age rangeis'
consistentwith, but lacks the resolutionof, 4øAr/39Ar Ptutonic rocks
analysesfrom Yemen [Baker et al., 1996a]. From the
4øAr/39Ar andK-Ar database it is apparent
thatmanyof the Plutonic rocks that intruded the western and southern
publishedK-Ar analysesfor the Yemen flood basalts(and
coastof Yemen are mainly alkaline to peralkalineA-type
probablyothervolcanicfieldsmarginalto the Red Sea) are
granitesof possiblemantle origin [Chazot and Bertran&
erroneous andthat on the basisof 4øAr/39Arplateauages 1995; Btakey et at., 1994]. Various types of granites
flood volcanismlasteda mere 5 m.y. (31-26 Ma), at best
[Blakey et at., 1994] within the basalt-rhyolitesequence
half the time scalequotedin the literatureon the basisof
were unroofed as a consequenceof exhumation. More
K-Ar dates.
importantly, intrusive granites are an integral part of
domino-faultblockson the volcanicmargin,a factorthat is
importantas the age of the granitesis consideredto be
Figure3. The4øAr/39Ar
chronostratigraphy
of theYemenLIP different from that of the volcanic rocks used to constrain
[Baker et al., 1996a]. The sectionsof the Yemen LIP that were
studiedoccuralong a west to easttraversefrom the Red Sea rift the "maximum"age of the onsetof extension.
marginto Sana'a(rift shoulder). Along thesesectionsvolcanism Zumboet al. [ 1995]reportedtwo 4øAr/39Ar
plateauages
beganat 31-29 Ma, the switchfrom basalticto silicicvolcanism of 21.4 and 22.3 Ma for a plutonicbody in the southern
occurredat ca. 29 Ma and the youngestpreservedvolcanicunits part of Yemen, and Btakeyet at. [1994] reporteda Rb-Sr
are 26.5 Ma in age. age of ca. 25 Ma. In many casesthe granitesare
34 EVOLUTION OF RED SEA VOLCANIC MARGIN, WESTERN YEMEN

oN

16

15

14 o s:.25. ..[-•
125 •,.•.•
.........

.,.....,,•
";/--
-.-:
13
FTAg-e(Ma)
225;................
.
......
32.5 .......

ADEN .'•375-
..........................
! ............ !..............................
.l..... I............................................................
I .................. I................................................
42 43 44 45 46 47 48 49 øE
Figure4. Contoured
apatitefissiontrackagesontheRedSeaandGulfof Aden[Menzies
et aL, 1997] for igneous,
metamorphicand sedimentarybasementrocksthat underliethe Yemen LIP. The contoursfor 25 Ma and 175 Ma are
pickedoutasthinblacklines.Notethatapatite
fission
trackages_<25
Ma (withlongmeantracklengths,
i.e.,> 14gm)
occuronlyontheRedSeamarginwherethecoolingeventwasrapid[Menzies
et al., 1997].

youngerthan the volcanicrocks[Blakeyet aL, 1994] and reactivation of pre-existentlineamentsof Gondwanaor

theirintrusion intotheYemenLIP occurred duringa major Jurassicage.
magmaticevent that occurredalong the coastof Saudi ApatiteFT analyses[Menzieset aL, 1997] canprovide
Arabia between21 and24 Ma [Sebai 1989;Fdraudet al., additionalinformationon the timing of extensionif one
1991; $ebai et al., 1991] and eruptionof the Harrat As acceptsthat, in part of the Red Sea and Gulf of Aden
Sirat and Harrat Hadan in Saudi Arabia around 31-22 Ma margins,tectonicdenudation (drivenby extension) may
and28-27 Ma, respectively[DuBrayet al., 1991;$ebai et havebeenoneof themosteffectivemeansof coolingthe
aL, 1991]. Thesedatapointto a main,widespread period crust.While erosional denudation playsan importantrole,
of extension between24 and21 Ma [Fdraudet al., 1991; rapidcooling(asis evidentin thetracklengthdata)would
Sebaiet al., 1991]. bebestachieved by tectonicprocesses.Detailedmapping
of the Red Sea [Davisonet al., 1994] and Gulf of Aden
.marginsin Yemen demonstrates
a coincidencebetweenthe
Footwall Basement Rocks
presenceof a belt of domino-faultblocksproximalto the
present-daymarginsand the thermal domaincharacterised
The zone of Oligo-Mioceneexhumationdefinedon the by rapidexhumation
[Yellandet al., 1994]. Aspects
of the
basisof apatiteFT analyses[Menzieset aL, 1997] is apatite FT analysesreflect tectonic, as well as erosional
restrictedin distribution.It is proximalto boththe Red denudation,particularlywhere the sampleswere taken
Sea volcanicmargin and the Gulf of Aden non-volcanic from the footwallexposures
of fault blocks. As with the
margin,andit extendsintotheRamlatAs SabataynGraben 4øAr/39Ar
data,theapatite
FT analyses
constrain
thetiming
(Figure 4), a Jurassicriff filled with >5000 m of sediment of extensionas havingbegun<26 m.y. ago. This is
[Davison et aL, 1994]. Exhumationappearsto be compatible with what has been deduced from other
inextricably linked to areas of crustal extensionand/or geologicalconsiderations
[Davisonet al.,
 MENZIES ET AL. 35

EXHUMATION Ramlat As SabataynGraben (Figure 4) [Menzies et al.,

1997]. We prefer to interpretthe apatiteFT analysesto
Fieldwork [Menzieset al., 1992, 1994a,b;Davison et al., mean that regionsproximal to the Gulf of Aden and Red
1994; Baker, 1995] reveals exhumational/erosional breaks Sea margins were rapidly exhumed in Oligocene and
nearthe top of the Yemen LIP, beforethe eruptionof more Miocene times, respectively, and that the apatite FT
recent volcanic rocks. Apatite FT analyseson samples analyses record that history. The youngestapatiteFT ages
collectedfrom the Red Sea volcanic margin [Menzieset of 16-18 Ma imply that this cooling episode spanned
al., 1997] highlight an area immediatelyadjacentto the several million years. Furthermore,the lack of interflow
sediments indicates that exhumation must have been
Red Seamargin with apatiteFT ages< 25 Ma (Figure 4)
and track length distributionsconsistentwith very rapid minimal during the main episode of magmatism (31-26
exhumation. Elevated Precambrianbasement"highs" on Ma) as surface uplift proceeded. However, as magmatism
the rift shoulderhave apatiteFT ages>> 100 Ma (Figure4) waned and extension became widespread exhumation
and track lengthdistributionsindicativeof a complexpre- became more pronounced. This is supportedby the
rift history. An interveningarea along the Red Sea and presenceof minor sedimentaryhorizonstoward the top of
Gulf of Aden margins,and inlandalongthe Balhaf graben, the volcanostratigraphy and an erosionalunconformityat
has apatite FT ages of 25-100 Ma and track length the top of the volcanic stratigraphynortheastof $ana'a
[Baker et at, 1996b].
distributionsindicative of rapid exhumation(Figure 4).
Modeling of the samplepopulationwith the longesttrack
lengths (>14.0 •tm) provides the most robust way of MISS1NG VOLCANOSTRATIGRAPHY?
estimatingthe timing of the most recent cooling event
[Menzieset al., 1997]. While thereis a significantspread Although -2500 m of volcanic rocks exist on the
in the timing of maximumtemperatures prior to cooling, uplifted rift shouldersin westernYemen (Figure 2), this
age and length best-fit paths for five samplesthat were may be only part of the original stratigraphy[Menzieset
modelledclusterat and around31 Ma. Modeling of two a/., 1992]. Many aspectsof the field and laboratorydata
older samplesrevealsthat the timing of the latestcooling leadus to this conclusion.Volatile-rich,amphibole-bearing
from temperaturesin excessof 100øC for both samples granites intruded into the Oligocene volcanic rocks are
was around28-30 Ma. Considerationof all the apatiteFT now exposedat the surface,requiring, almost certainly,
data[Menziese! al., 1997] indicatesthat25 Ma is an upper removal of 1000-1500 m of cover rocks by denudational
limit on the timing of the most recentsignificantcooling processes.4øAr/39Ar andRb-Sr datingof theseplutonic
episode. Takentogether, 4øAr/39ArandapatiteFT analyses rockspointsto intrusionbetween25 and 21 Ma, in marked
indicatethat crustalcoolingexhumationwas synchronous contrastto the age of the volcanic stratigraphy(31-26
with LIP formation. If one assumesthat surfaceuplift, in Ma). In addition,dike swarmscuttingthe LIP indicatethat
part, generatedrelief and drove exhumation,then surface magma transportby crack propagationthough the LIP
uplift must have begun >30 m.y. ago, before eruptionof continuedfor 10 m.y. from 26 to 16 Ma [Zumbo et al.,
the YemenLIP andbeforemajor extension. 1995], again in contrast to the age of the preserved
One could arguethat the samplesutilisedfor apatiteFT volcanicstratigraphy.Althoughunconformitiesare absent
analysesproximal to the Red Sea margin were buried throughoutmost of the middle and lower volcanic section,
under2-3 km of volcanicrock and consequently any pre- there is a conspicuousunconformityin the uppermost
existentrecord of pre-volcanicexhumation(if it existed) volcanicstratigraphy. 4øAr/39Ar chronostratigraphy
[Baker
waseffectivelyobliterated.The basementand sedimentary et al., 1996a] of the flows above and below this
sampleswere annealed and subsequentcrustal cooling unconformityindicatesa time gap of 7 m.y. between26
triggeredby crustalextensionand tectonicdenudationin and 19 Ma. Ironically,this time gap corresponds with the
the late Oligoceneand Miocene"set"the fissiontrack ages main periodof injectionof plutonicrocksand dikesalong
in this time period. In other words, it would have been the Red Sea margin indicatingthat the exhumationthat
highly unlikely that "older" FT, and thus exhumational removedpart of the volcanostratigraphy may have been
ages, could have survived. Therefore the apatite FT exacerbatedby an importanttectonicchangearoundthis
analysesare biasedby the very processes with which they time.
are associated(i.e., volcanism). Apatite FT analysesindicatedthat the main period of
While we acknowledgethatthis is a possibility,it fails to exhumationwas <30 Ma (Figure 4) and that exhumation
explainwhy an identicalthermalhistoryexistson the Gulf became more active and widespreadat <26 Ma. This
of Aden non-volcanic margin and within the Jurassic happensto be the period of extensionthat relatesto
36 EVOLUTION OF RED SEA VOLCANIC MARGIN, WESTERN YEMEN

openingof the proto-Gulf of Aden [Court#lotet al., 1987]
and formation of the earliest oceanic crust at ca. 20 Ma
[Sahota et al., 1995]. Such a major tectonicoccurrence
may relate to plate movement, so that the Yemen LIP
movedaway from the plume(i.e., Afar plume)but magma
productionfrom shallowmagmachamberscontinuedfor 5
m.y. or more. Alternatively, rifting in the Gulf of Aden
may have channeled plume-derived material eastward
away from the Yemen LIP. However, once the eruption
ratesdecreased[Baker et al., 1996a] and the constructional
aspectsof LIP formationabated,extensionled to collapse
of the margin and contempormleous exhumationunroofed
the margin. One couldspeculatethat about1500 m of the
volcanic section is missing, and was eroded during a
periodof exhumationthat wastriggeredby the openingof
the Gulf of Aden. Partial support for this idea is
forthcomingfrom sedimentbudgetanalysis(I. Davisonet rocks (low 87Sr/86Srand high 143Nd/144Nd
al., pers. comm.) to explain the sedimentaryrocks that compositionssimilar to upper and lower crustal rocks
underlie the Tihama Plain.
CRUST-MANTLE PROCESSES
Geochemistrycan be used to understandcrust-mantle
processesand the provenanceof individual batchesof
magma that constitutethe Yemen LIP. We are uniquely
placed in the southernRed Sea to assigna provenanceto
the LIP because of information available from several
studies:
(a) The asthenosphere or depletedupper mantle from
studiesof MORB eruptedalongthe Gulf of Aden ridge;
(b) A mantleplume from studiesof continentalvolcanic
rockseruptedin an areathat, on the basisof heat flow and
seismic tomography, may be a site of a deep mantle
structure(i.e., the Afar plume);
(c) The lithosphericmantle from the studyof ultramafic
xenoliths derived from the shallow Arabian mantle; and
(d) The continentalcrust from studiesof igneousand
metamorphicrocksexposedat the surfaceandentrainedby
volcanic rocks erupted through crust of Archaean, (<< 29 Ma).
Proterozoic,and Phanerozoicage.
It is widely acceptedthat Cenozoicvolcanismassociated
asthenosphere.From these observations,the chemical
compositionof the magmas is related to mixing of
differentmantle sources. However, a very importantnon-
oceanic isotopic component also exists, which is
tentativelyrelated to melting of the lithosphericmantle
[Chazot, 1993; Chazot and Bertrand, 1993, 1995] or to
contaminationof the magmas by the continentalcrust
[Baker et al., 1996b]. Baker et al. [1996b] demonstrated
that, in most cases, throughout 2500 m of volcanic
stratigraphy,the volcanic rocks are contaminatedwith
crust (Figure 5a,b). In some casesthis includeshighly
magnesianmagmas which contain zoned clinopyroxene
phenocrysts that recordthe contamination process[Baker
et al., 1994a]. The isotopic data (Figure 5a,b) on the
volcanicrocks from the Yemen LIP define a range in Sr
andNd isotoperatiosfrom compositions similarto oceanic
ratios) to
(high 878r/86Sr
and low 143Nd/144Ndratios). Correlations
between878r/86Srratiosand indicesof fractionation
(e.g.,
SiO2 and Fe/Fe+Mg ratio) support a model of crustal
contamination during fractionation in crustal magma
chambers[Bakeret al., 1994, 1996b](Figure5c).
In the case of the Yemen LIP, elementaland isotopic
data [Baker et al., 1994, 1996a,b; Chazot and Bertrand
1993, 1995] define a temporal change in magma
source/storagethat, with time, changes from initial
eruption of deep mantle-derived basic melts that are
variably contaminatedwith lower crust, to eruption of
silicicmeltscontaminatedwith uppercrustthroughstorage
in shallow crustal reservoirs(Figure 5). This change is
consistentwith the involvementof a mantleplume(equiv.
Afar) and asthenosphere (equiv.Gulf of Aden) earlyin the
historyof the LIP (31-29 Ma), pondingand underplating
of basic magmasat the Moho and formationof shallow
magma chambers after 2 m.y. of LIP formation, and
caldera-formingeruptionslate in the evolutionof the LIP
SURFACE UPLIFT

with LIP formationin Ethiopiaand Yemen [Chiesaet al.,

1989] was related to the impingementof a large mantle Withtheuseof 4øAr/39Ar andapatiteFT analyses we can
plume head beneath the Afro-Arabian lithosphere[e.g., quantify the onset and terminationof magmatism,the
I/Vhite et al., 1987; I/Vhite and McKenzie, 1989]. periodof peak exhumationon the rifted margin,and the
Volcanism in Djibouti [Vidal et al., 1991; Deniel et al., relativetimingof extension.4øAr/39Ar data(Figure3) and
1994] is characterised by a "HIMU" isotopicandelemental apatiteFT analyses(Figure 4) reveal a brief period of
signaturethat may be that of the Afar mantle plume. Oligocenemagmatismand a protractedperiod of Oligo-
Moreover,geochemicalstudiesof basaltseruptedalongthe Miocene exhumation/extension. While we are left with
Red Sea [Eissenet al., 1989; IZolkeret al., 1993] and the little or no informationaboutthe timing of surfaceuplift,
Gulf of Aden [Schillinget al., 1992] spreadingaxeshave onsetof surfaceuplift in the Oligocenewouldbe expected
helped to define the isotopic compositionof the local if it contributedto Oligo-Miocene
 MENZIES ET AL. 37

Evidencefor surfaceuplift, or base-levelchanges,may mountainsreaching3600 m high. Vital informationabout

be quite subtle, especially if it was a transientfeature surface uplift may be found in (1) shallow marine
immediatelyprior to the onsetof volcanism. Much of this mudstoneswithin the volcano-sedimentary stratigraphy,
evidenceis to be found in the pre-LIP sediments. The (2) palaeocurrent directions within the pre-volcanic
Yemen LIP is underlainby fluvial and marine sediments sandstones, (3) any temporal palaeoenvironmental
(Figure 6) overlain by thick paleosols[Al'Kadasi, 1995; changes, and (4) palaeoaltitudedata locked in plant
Al'Subbary, 1995; Baker et al., 1996a,b]. Palaeo- remainswithinthe sedimentary andvolcanicstratigraphies.
environmental interpretations indicate a nearshore With regardto the latter, England and Molnar [1990]
depositionalenvironmentof very low relief (Figure 7), in stressed that one of the least controversial measures of
contrast to the present topography of the rift-margin surfaceuplift may be containedwithin the palaeoaltitude
information that can be inferred from seasonality
0.51295 measurements in plant speciesexistingover a wide range
0 of altitudes. Although in westernYemen plant remains
occur within the pre-volcanicsedimentsand the volcanic

••0 000
0.51290 •,
stratigraphy,our presentknowledgeof the extentof plant
o •]
O.51285 o
o
A speciationindicatesthat their use is not conduciveto the
determinationof palaeoaltitudes.
rI r• []

0 5128O
[] o o• ß
Shallow marine mudstones ?

0.51275

In additionto marine sedimentsoccurringimmediately

0.51270 ! ! i
beneaththe LIP, fossiliferousmudstonesexistover a range
0 7036 0.7039 0.7042 0.7045 0.7048 of altitudesat the sediment-volcanicrock contact(Figure
6) and within the volcanic stratigraphy. They are useful
0.51295 indicatorsof surface uplift, provided they are marine in

area
enlarged origin. The fossils in many of these sedimentsdo not
C> above

0.51285 allow one to distinguishbetweenlagoonal(near sea level)

upper crustal
or lacustrine (any altitude) environments. However, a
contarmnation
precise indication of the provenanceof these sediments
• rhyohtes
z 0.51275

comesfrom a comparisonof the 87Sr/86Sr ratio of the

O.51265
lower crustal
contamination sedimentswith that of contemporaneous
seawater. Fossils
andespies in sedimentsassociatedwith the Yemen LIP have a range
0.51255
of 87Sr/86Sr
isotopesdifferentfrom that of Tertiarysea
water, but similar to the Yemen volcanic rocks [Baker,
1995]. Thesefossils/sediments are thereforemore likely to
0.51245
O.7036 0.7046 0.7056 0.7066 be freshwaterin origin, derivedfrom surfacerunoff. Their
S7Sr/S6Sr
origin in a freshwater lacustrine environment is also

0 7056

Figure 5. (a) Isotopicvariationwithin the YemenLIP [Bakeret

range
ofvalues
for. * al., 1996b] convergeson a "mantle" compositionwith low
andes•bcsamples- 0 7038-0 7062 //I
0.7051
87Sr/86Sr
ratioandhigh143Nd/143Nd
ratios[Baker,
1995;Baker

rhyohtic
sarnples-
0.7042-0.7054
iI/
et al., 1996b]similarto thatobserved
in oceanicrocks. (b) The
spreadin 87Sr/86Sr
and 143Nd/143Nd
isotopicratios in the
O.7O46 o volcanic rocks of the Yemen LIP is believed to result from
contaminationwith upper and lower crustal rocks. (c)
0 7041
Correlationsbetweenisotoperatios and indicesof igneous
fractionation(e.g., SiO2) indicatethat assimilationof crustal
rocks occurredat the sametime as fractionationin magma
0.7036
chambersat shallowlevels in the crust. The rhyoliteshave a
!

4O 5O 55 higherinput of uppercrustalmaterial,perhapsdue to residence

andformationin crustalmagmachambers.The leastfractionated
SiO2 (wt.%) volcanicrock(lowestSiO2)is the least
38 EVOLUTION OF RED SEA VOLCANIC MARGIN, WESTERN YEMEN

• • •,Lithology
Depositional
Environment ß • •, Lithology
Environment

::::::::::::::::::::::::::::::
" ;:'t•Claystones
and Siltstones

•-:-:1Sandy
mudstones
ß, ;•• Ironconcretions

• •: .•
;•• Cross-bedded
Medium fine
sandstone
gr.sst.
• • ;8'.• Channel
conglom.
2oo'•]:?•'½'•:'•'•?::!::?•:;::j
• • • ::•'•Coarse
gr.
sst
• o • Conglomeratic
sst.
- c • • • Burrows
o • O• BenthicForaminefira
• • 0 0

.... • 0

:.:-:.:.:.:.:.:.:.:.•
•
-':':':':':':':':':':
:.:.:.:.:.:.:.:.:.:.
•• •
...-.-.-.'.-.'.-.'.'
• --
O
E

• • • • • BASALTIC
LAVAS

• o• •• :..:..!•.••z • o
._ z ',:':,':'l
0 ½ <0
• Z
w .-
Z E -'.'.'.'. O•
....
½ 0 ....
z

•50.... •--- • .-- • • O

'

-- -7•Z•• • - • --
• •0 • •J I

0- •r 25
AMRAN GP c s f mc pB
 MENZIES ET AL. 39

N
ASIR

I
I
!
!

Fluvial and
Shallow marine
sands
SANA'A ':i:!:!:!:!:
HUDYDAH

;
I AL BAYDA

!
!

""" "" '"" """

'? marine sands
....
:-- Shallow marine
limestone, shales
.............
.............
& sands ..................

Cretaceous ;::::::::::::::::::::::::::::::-•:::MUKALLA :-.... '•'"•'-'

Shoreline
..........................

.............................................

• ,• -•_-•.•-•-•-• *• ........

Jurassic
Limestone
SOMALIA
'f.'.••Precambrian
Basement 200 km

GREAT
W TIHAMAPLAIN ESCARPMENT RIFT MOUNTAINS OF N.W. YEMEN SANA'A PLATEAU E

I li)Graniticrnounta,ns
ii)Centralrota,e.dfaul,
iii)Easternhighvolcancplateau
•-.....forming
Alluvia,
tans >1
aBajada I -.. II De
,n,•
d*' •.•
Drainage
blockterra,n
....... ,-........
I.... Erosion ..__.
Y Scarp
Centripetal
Drainage

• _.. I Antecedent
Drainage i • Jabal
Nabi
Height/m I Shuyab 3660m
Sana'a
30004
Red
Sea
Coastline Internal Trap Iopography
_,.•..
'<*•,•.__. Pv!,i•ce•-ReCneen!r3OOO
;11;/• Jabal Drainage
j•
•ii•*•
',•.•.•
Volcanic
Co
es
!
• Basins •,.•½ • •-•-,•.,._.,• l
• ...........................
•..., .....
ß
..............................
.•..2000
River
terraces
f.
.-•.½
•'::•':'"•
............
-•'"'
•"-?-.--•-.... %•..:..,•
.............-•,=½
• .............. •..
' DomesDunes •a•,blo½•
Salt Jabal •.,-.-
' '• ;;:
............................
i:"O'"'":•"•
...... .....................
?•i:••ooo
0 • "•.......................................................................................................................................................................................................................... sea
level

Figure 7. Palaeoenvironmental interpretations

of the Tawilah sandstones(which underliethe Yemen LIP) [Al'Subbary,
1990; 1995] indicate a west to east paleoslopeand a Cretaceousshoreline(top figure). The overall topographyis a
gentlewestto eastdippingpalaeoslope whichshouldbe contrasted with the post-LIPtopography(bottomfigure). The
latterreachesa maximumheightof 3660 rn with considerablerelief westwardto the Tihama Plain. Surfaceuplift of
severalthousandmetreshasoccurredsincedepositionof the sedimentsbut beforeextensionand collapseof the margin
in the late Oligoceneto early Miocene [Davisonet al., 1994; Menzieset al., 1994a,
40 EVOLUTION OF RED SEA VOLCANIC MARGIN, WESTERN YEMEN

consistentwith the subaerialaspectof the Yemen LIP and One of the major problemsis to establishhow much
their very localisedoccurrenceas lenticularbodies. time is lockedup in the formationof the pre-LIP sediments
andthe paleosols.We havepreciseageinformationon the
Sandstones overlyingvolcanicrocks(onsetat 29-31 Ma) but the age
information from the sedimentsin which the paleosols
Although palaeocurrentdirectionshave been used to formed is less precise. The Medj-Zir Formation is
define domal surfaceuplift effectivelyin other LIPs [e.g., believedto be Paleocenein age [Al'Subbary,1995] and, if
Cox, 1989], it shouldbe notedthat erosionalprocesses on a correct,this could mean that 30 m.y. may have separated
broad low-relief dome (-1-1.5 km maximum elevation formationof the uppermostpre-volcanicsediments andthe
over areasof-2 x 106km2 [Whiteand McKenzie,1989]) firstbasalticeruptions.
would not be highly energetic. Thereforethe erosional
responseto doming may be very subtleand largely lost, Palaeoenvironmental Change
particularly in an extended volcanic margin that is
subsequently brokenup into rotated,domino-faultblocks. An overallpalaeoenvironmental changeis evidentin the
A detailed study of the pre-volcanicCretaceous-Tertiary pre-LIP (Palaeocene-Eocene) sedimentsandthe Oligocene
sandstones [Al'Subbary 1990, 1995] demonstrated a volcanicrocks. A temporalchangecan be summarisedas
southwestprovenancefor the sandstonesin present-day follows, althoughsteps(a) through(d) have no accurate
Sudanand/or Ethiopia,as well as the presenceof a west- age control:(a) Cretaceous-Tertiary - depositionof fluvial
eastpalaeoslope(Figure 7) which is basedon a continental sandstones on a peneplain(low-relief continent)(Figure
aspectto the sedimentsin westernYemen and a marine 6); (b) Cretaceous-Tertiary - development of marine
aspect to the sediments in eastern Yemen. The sedimentsnear the top of the fluvial sandstonesequence
palaeocurrentdata [Al'Subbary, 1995] do not reveal a (shallow marine, continental shelf); (c) Paleocene -
patternconsistent with domaluplift. However,comparison deposition of fluvial sandstonesand developmentof
of the pre-volcanic topographywith the present-day paleosols(low relief continent)(Figure 6); (d) Paleocene-
topographyemphasises (Figure7) the topographic change localised ponding of lacustrine deposits (low-relief
andthe needfor surfaceuplift. continent);(e) Oligocene- eruptionof thousands of metres
of subaerialvolcanicrocks(continent)(Figures2 and3).
Paleosols Given our concernsaboutthe time gap in the pre-LIP
sedimentstwo interpretationsare possible.
The nature of the sedimentaryrocks underlying the (1) The lack of any disturbancebetween the rnarine,
Oligocenevolcanicrocksis pivotalto establishing subtle fluvial, paleosol, and lacustrinefacies strongly indicates
changesin the palaeoenvironment. Detailedwork on these that changesin base level were slight and on the order of
sedimentsby Al'Kadasi [1995] and Al'Subbary [1995] tensof metres. In fact, the developmentof thick paleosols
indicates (1) widespreaddevelopmentof paleosolsor reflectstectonicstability,sedimentstarvationand passage
ferricretedclasticsedimentsimmediatelybeneaththe basal of time. All of these could be used to argue against
volcanicunits (Figure 6), (2) a considerable variationin significantsurfaceuplift. The baselevel changeobserved
the thicknessof the paleosolsfrom 5-70 m, and (3) that at the top of the pre-LIP sedimentscouldhave happened
fermginouspaleosoldevelopment is moremarkedtoward anytime betweenca. 60 Ma and31 Ma. Therecouldbe a
the top of the pre-LIP fluvial to shallowmarinesuccession major time break betweenthe sedimentsand the LIP, on
[Al'Subbary, 1990, 1995]. the order of tens of millions of years, in which casethe
Summerfield[1991] statedthat low local relief was an change in palaeoenvironmentis not relevant to the
essentialfactor in duricrust(paleosol/ferricrete)formation volcanicmarginhistoryinvolvingplumeimpingement.
and thus a lack of intense denudation and sediment (2) Alternatively, these facies changes could be
starvation were vital prerequisites.We concur with interpretedas indicative of a temporalchangefrom an
SummerfieM [1991] and argue that the widespread aqueous(i.e., shallowmarine/continental margin)to a sub-
occurrenceof paleosolsindicatesthat sedimentstarvation aerial environment (i.e., emergent continent). On a
affectedthe whole areaof studyafter a period(Cretaceous- developing volcanic margin this could representthe
Tertiary) of sandstonedeposition [Al'Subbary, 1990, geologicalexpressionof the early stagesof broadsurface
1995]. Moreover, sedimentstarvationcontinuedfor the uplift. Initially this wouldhavebeenon the orderof tens
next 6 m.y. when about3000 m of volcanicrockswere of metres but it eventually accountedfor >2000 m of
eruptedwith few interveningsedimentary units. surface
 MENZIES ET AL. 41

We proposethat initiation of surfaceuplift occurred finally to Dr. AbdulkarimAl'Subbary (Universityof Sana'a)for

prior to (>31 Ma) the onset of LIP formation and must continualassistance with field researchin Yemen. Victor Camp,
have continuedthroughoutthe period of peak LIP John Mahoney, and John Pallister are thanked for their
magmatismuntil 26 Ma. The protractedperiod of constructive
commentswhich helpedimprovepresentationof our
ideas. Gary Nichols' input regarding fluvial sedimentsis
exhumationwas the erosionalresponseto surfaceuplift
appreciated.
exacerbatedby tectonicdenudation,once extensionled to
collapseof the margin some 26 m.y. ago. This is
consistentwith the apatite FT analyseson the basement REFERENCES
rocks.

Al'Kadasi,M., Temporaland spatialevolutionof the basalflows

SUMMARY of the Yemen Volcanic Group. Unpublisheddoctoralthesis,
Royal Holloway Universityof London,285 pp., 1995.
Applicationof absolute datingtechniques (i.e.,4øAr/39Ar Almond, D.C., Geologicalevolutionof the Afro-Arabiandome,
andapatiteFT analyses)to the YemenLIP indicates Tectonophysics,131,301-332, 1986.
(a) initiationof magmatismat 31 Ma, A1 Subbary,A. A., Stratigraphicand sedimentological
studiesof
(b) rapidexhumationof the LIP in the Oligo-Miocene, the Tawilah Group, A1-Ghirasarea, northeastSana'a,Yemen
(c) extensionin the Miocene(<26 Ma), Arab Republic,UnpublishedMSc thesis,Universityof Sana'a,
184 pp., 1990.
(d) contemporaneous extensionand plutonic/hypabyssal
Al'Subbary, A. A., The sedimentologyand stratigraphyof the
activity (<26Ma). Cretaceous-earlyTertiary Tawilah group, western Yemen.
Unfortunately, no unequivocalevidencecan be found Unpublisheddoctoralthesis, Royal Holloway University of
for the effectsof surfaceuplift in the sedimentaryor the London, 153 pp., 1995.
volcanic record other than a very general palaeo- Baker, J., Stratigraphy,40Ar/39Argeochronologyand geo-
environmental change. While erosional denudation chemistry of Cenozoic volcanism in Yemen. Unpublished
requires surfaceuplift to enhancetopographicrelief, in doctoralthesis,Royal Holloway University of London, 386
many instancestectonic denudationdriven by extension pp., 1995.
can more effectively cool the crust. Extension and Baker, J., C. Macpherson,D. Mattey, M. Menzies, and M.
Thirlwall, Laser fluorination oxygen isotope analysis of
exhumationmay relate to widespreadextension(<26 Ma)
phenocryst phases from Oligo-Miocene flood basalts and
due to the openingof the Gulf of Aden. Exhumationwas
Quaternaryintraplatebasaltsin Yemen. U.S. Geol. Surv. Circ.
minimal for 5 m.y. duringeruptionof the YemenLIP (31- 107, 18, 1994a.
26 Ma), but accelerated around 26 Ma, the time of Baker, J., M. Menzies, L. Snee, and M. Thirlwall, Stratigraphy,
widespreadextensionandplutonicactivity. 4øAr/39Ar geochronology andgeochemistry of floodvolcanism
We concludefrom an integrationof field observations in Yemen. Mineral. Mag., 58A, 42-43, 1994b.
and absolute dating that continental rifling and LIP Baker, J., L. Snee,and M. A. Menzies,A brief Oligoceneperiod
formationin the southernRed Sea occurredin responseto of flood volcanismin Yemen; implicationsfor the durationand
surfaceuplift and flood volcanismwith minimal extension, rate of continentalflood volcanismat the Afro-Arabian triple
junction. Earth Planet. Sci. Lett., 138, 39-55, 1996a.
followed by a protracted period of exhumation and
Baker, J.A.,M. F. Thirlwall, andM. A. Menzies, Sr-Nd-Pb
and
extension.This modelrequiresthat previousmodelsbased trace element evidence for crustal contamination of a mantle
on preliminary apatite FT analysesand K-Ar data be plume: Oligocene flood volcanism in western Yemen,
revised [Menzieset al., 1992, Davison et al., 1994] and is Geochim. Cosmochim.Acta, 60, 2559-2581, 1996b.
consistentwith the ideasof Richardset al. [ 1989]. Blakey, S., M. A. Menzies,and M. F. Thirlwall, Geochemistryof
Tertiary within-plate(A-type) granitoidsat a passivemargin,
Acknowledgments.Geologicalresearchin Yemen would not southern Red Sea (abstract), International Association of
have been possible without the enthusiasmof Jim McGrath Volcanology and Chemistry of the Earth's Interior, Ankara,
(British Council)and Dr. Al'Kirby (Universityof Sana'a). We 1994.
acknowledge financial support from the British Council Bohannon, R. G., C. W. Naeser, D. L. Schmidt, and R. A.
(Al'Kadasi & Baker), the EuropeanUnion (Chazot), CNRS- Zimmerman, The timing of uplift, volcanism and rifting
INSU (Chazot), the NERC (Menzies and Hurford), the peripheral to the Red Sea: A case for passiverifting, d.
Departmentof Geology, Royal Holloway Universityof London Geophys.Res.,94, 1683-1701, 1989.
(Baker & Al'Kadasi)and the Arab League(Al'Kadasi). British Camp, V. E., and M. J. Roobol, Upwelling asthenosphere
Petroleumand the Royal Society are thankedfor supportinga beneath western Arabia and its regional implications,d.
major expeditionto Yemen in 1991. This work, undertakenover Geophys.Res.,97, 15,255-15,271,1992.
the last five years, would have been impossiblewithout the Chazot,G., Evolution geochimiquedu magmatismeCenozoique
logisticalsupportprovidedby the Universityof Sana'a. Thanks au Yemen: interactionsentre le rift Mer Rouge - Aden et
42 EVOLUTION OF RED SEA VOLCANIC MARGIN, WESTERN YEMEN

point chaudAfar, Unpublisheddoctoralthesis,Univ. Claude McGuire, A. V., and R. G. Bohannon,Timing of mantle

Bernard,Lyon I., 159 pp., 1993.
Chazot, G., and H. Bertrand, Mantle sources and magma-
continental crust interactions during early Red Sea-Aden
rifting in southernYemen: elementaland Sr, Nd, Pb isotopic
evidence,d. Geophys.Res.,98, 1819-1835, 1993.
Chazot, G., and H. Bertrand, Genesisof silicic magmasduring
Tertiary continentalrifting in Yemen, Lithos,36, 69-84, 1995.
Chiesa, S., L. La Volpe, L. Lirer, and G. Orsi, Geologicaland
structuraloutlineof Yemenplateau:YemenArab Republic,N.
rib. Geol. Paleont. Mh., 11, 641-656, 1983a.
upwelling:evidencefor a passiveoriginfor theRed Searift, d.
Geophys.Res.,94, 1677-1682, 1989.
Menzies, M. A., J. Baker, D. Bosence,C. Dart, I. Davison,A.
Hurford, M. Al'Kadasi,K. McClay, G. Nichols,A. Al'Subbary,
andA. Yelland, The timingof magmatism, uplift and crustal
extension: preliminary observationsfrom Yemen, in
Magmatismand the Causesof ContinentalBreak-up,Spec.
Publ. 68, edited by B.C. Storey,T. Alabaster,and R. J.
Pankhurst,pp. 293-304, The GeologicalSociety,London,
1992.

Chiesa, S., L. La Volpe, L. Lirer, and G. Orsi, Geology of the Menzies,M. A., A. Yelland, J. Baker, S. Blakey, G. Chazot,M.
Dhamar-Rada'volcanic field, Yemen Arab Republic, N. rib. Al'Kadasi, and C. Rundle, Evolutionof the Red Sea volcanic
Geol. Paleont. Mh., 8, 481-494, 1983b. margin- a multi-isotopicapproach,U.S. Geol. Surv. Circ.,
Chiesa, S., L. Civetta, M. De Fino, L. La Volpe, and G. Orsi, 1107, 216, 1994a.
The Yemen Trap Series:genesisand evolutionof a continental Menzies,M.A., M. A1-Kadasi,S. A1-Khirbash,A. A1-Subbary,J.
flood basalt province,d. Volcanol. Geotherm.Res., 36, 337- Baker,S. Blakey,D. Bosence,I. Davison,C. Dart, L. Owen,K.
350, 1989. McClay,G. Nichols,A. Yelland,andF. Watchom,Geologyof
Civetta, L., L. La Volpe, and L. Lirer, K-Ar agesof the Yemen theRepublicof Yemen,in TheGeologyandMineralResources
plateau,d. Volcanol.Geotherm.Res., 4, 307-314, 1978. of Yemen,editedby D. A. McCombe,G. L. Fernette,andA. J.
Cloos, H., Hebung-Spaltung-Vulkanismus, Geol. Rundsch.,30, Alawi, Tech. Rep., pp. 21-48, Ministry of Oil and Mineral
405-527, 1939. Resources,GeologicalSurveyandMineralsExploration Board,
Courtillot, V., R. Armijo, and P. Tapponnier, Kinematicsof the Yemen Mineral SectorProject, 1994b.
Sinai triple junction and a two-phasemodel of Arabia-Africa Menzies,M. A., K. Gallagher,A. Yelland, and A. J. Hurford,
rifting, in ContinentalExtensionalTectonics,Spec.Publ., 28, Volcanic and non-volcanicrifted marginsof the Red Sea and
edited by M.P. Coward, J. F. Dewey, and P. L. Hancock,pp. Gulf of Aden: Crustalcoolingandmarginevolutionin Yemen,
559-573, The GeologicalSociety,London, 1987. Geochim.Cosmochim.Acta, in press,1997.
Cox, K. G., The role of mantleplumesin the developmentof Richards,M. A., R. A. Duncan, and V. E. Courtillot, Flood
continentaldrainagepatterns,Nature, 342, 873-877, 1989. basaltsand hot-spottracks- plume headsand tails, Science,
Davison, I., M. A1-Kadasi, S. A1-Khirbash, A. A1-Subbary,J. 246, 105-107, 1989.
Baker, S. Blakey, D. Bosence,C. Dart, L. Owen, M. Menzies, Sahota,G., P. Stules,and K. Gerdes,Evolutionof the Gulf of
K. McClay, G. Nichols, and A. Yelland, GeologicalEvolution Aden and implications
for the developmentof the Red Sea.
of the southernRed Searift margin- Republicof Yemen. Geol. "RiffSedimentation
andTectonics in theRedSea-GulfofAden
Soc. Am. Bull., 106, 1474-1493, 1994. Region",Sana'aUniversity,Yemen,56, 1995.
Deniel, C., P. Vidal, C. Coulon, P. J. Vellutini, and P. Piguet, Schilling,J. G., R. H. Kingsley,B. B. Hanan,andB. L. McCully,
Temporal evolution of mantle sources during continental Nd-Sr-Pbisotopicvariationsalongthe Gulf of Aden:evidence
rifting: the volcanismof Djibouti (Afar), d. Geophys.Res.,99, for Afar mantle plume-continental lithosphereinteraction,d.
2853-2869, 1994. Geophys.Res.,97, 10,927-10,966,1992.
Dixon, T. H., E. R. Ivins, and B. J. Franklin, Topographicand Sebai,A., V. Zumbo, G. Feraud,H. Bertrand,A. G. Hussain,G.
volcanic asymmetry around the Red Sea: constraintson rift Giannerini,
andR. Campredon,
4øAr/39Ar
datingof alkaline
models. Tectonics, 8, 1193-1216, 1989. andtholeiiticmagmatism
of SaudiArabiarelatedto the early
DuBray, E. A., D. B. Stoeser, and E. H. McKee, Age and RedSearifting,EarthPlanet.Sci.Lett.,104,473-487,1991.
petrologyof the Tertiary As Saratvolcanicfield, southwestern Sebai,A., Datation4øAr/39Ar du magmatisme lie aux stades
SaudiArabia, Tectonophysics, 198, 155-180, 1991. precocesde l'ouverturedes rifts continentaux:exemplesde
Eissen,J.P., T. Juteau,J. L. Joron, B. Dupre, E. Humler, and A. l'AtlantiqueCentralet de la Mer Rouge,Unpublisheddoctoral
Al'Mukhamedov,Petrologyand geochemistryof basaltsfrom thesis, Univ. Nice-SophiaAntipolis.,1989.
Red Sea axial rift at 18ø North, d. Petrol., 30, 791-839, 1989. Summerfield,M. A., Sub-aerialdenudationof passivemargins:
England,P., and P. Molnar, Surfaceuplift, uplift of rocks,and regionalelevationversuslocalreliefmodels,EarthPlanet.Sci.
exhumationof rocks,Geology,18, 1173-1177, 1990. Lett., 102, 460-469, 1991.
F6raud,
G., V. Zumbo,A. Sebai,andH. Bertrand, age Vidal, P., C. Deniel, P. J. Vellutini, P. Piguet,C. Coulon,J.
4øAr/39Ar
and duration of tholeiitic magmatism related to the early Vincent,andJ. Audin, Changesof mantlesources in the course
openingof the Red Sea rift, Geophys.Res. Lett., 18, 195-198, of a rift evolution:theAfar case,Geophys.
Res.Lett.,18, 1913-
1991. 1916, 1991.
Manetti, P., G. Capaldi, S. Chiesa,L. Civetta, S. Conticelli,M. Volker,F., M. T. McCulloch,andR. Altherr,Submarine
basalts
Gasparon,L. La Volpe, andG. Orsi, Magmatismof the eastern from the Red Sea:new Pb, Sr andNd isotopicdata,Geophys.
Red Seamarginin the northernpart of Yemen from Oligocene Res. Lett., 20, 927-930, 1993.
to present,Tectonophysics, 198, 181-202, 1991. White, R., and D. McKenzie, Magmatismat rift zones:

generationof volcaniccontinentalmarginsand flood basalts,J.
Geophys.Res.,94, 7685-7729, 1989.
White, R. S., G. D. Spence,S. R. Fowler, D. P. McKenzie, G. K.
Westbrook,and A. N. Bowen, Magmatism at rifted continental
margins,Nature, 330, 439-444, 1987.
Yelland, A., M. A. Menzies, and A. Hurford, Exhumation of the
Red Sea and Gulf of Aden rift-flanks: fission track contrasts
within the volcanic and non-volcanicmarginsof Yemen, U.S.
Geol. Surv. Circular, 1107, 360, 1994o
MENZIES ET AL. 43
Zumbo,V., G. Feraud,H. Bertrand,andG. Chazot,4øAr/3UAr
chronology of the tertiary magmatic activity in Southern
Yemen during the early Red Sea- Aden rifting, d. Volcanol.
Geotherm. Res., 65, 265-279, 1995.
Mohamed Al'Kadasi, Joel Baker, Gilles Chazot, and Martin
Menzies,Departmentof Geology,Royal Holloway University of
London,Egham,Surrey,TW20 OEX,
 The North Atlantic IgneousProvince
A.D. Saunders

Departmentof Geology,Universityof Leicester,Leicester,UnitedKingdom

J. G. Fitton

Departmentof Geologyand Geophysics,

Universityof Edinburgh,l/Vest
MainsRoad,Edinburgh,UnitedKingdom

A. C. Kerr, M. J. Norry, and R. W. Kent

Departmentof Geology,Universityof Leicester,Leicester,UnitedKingdom

The North AtlanticIgneousProvinceextendsfrom easternCanadato the BritishIsles,a pre-

drift distance of almost 2000 km. The igneous rocks are predominantlybasaltic, but
differentiates
and anatecticmeltsare alsorepresented. Two majorphasesof igneousactivity
can be discerned.Phase1 beganabout62 m.y. agowith continent-based magmatismin Baffin
Island, W and SE Greenland,the British Isles, and possiblycentralE Greenland(the Lower
Basalts around Kangerlussuaq).Phase2 began about 56 m.y. ago and is representedby
seaward-dipping reflectorsequences
(SDRS) alongthe continentalmargins,the Main Series
basaltsin centralE Greenland,the Greenland-Faeroes Ridge, and Iceland.Contaminationby
continentalcrustwas prevalentduringPhase1 but also occurredduringPhase2, especially
duringthe formationof the earlySDRS.Althoughit is unnecessary to involvethe continental
lithospheremantle in the formationof Phase1 or Phase2 magmas,it is not possibleto
completelyexcludeit. We arguethatthe Icelandplumeplayeda pivotalrole in the formation
of theNorth AtlanticIgneousProvincebecause(1) the simultaneous andwidespreadinitiation
of activityrequiresa majorthermaleventin the mantle;(2) someof the magmasassociated
with Phase1 werehighlymagnesian, indicatingthattheliquidsand,by implication,the mantle
sourceregionswereunusuallyhot;(3) the SDRSwereemplacedsubaerially or into shallow
water, indicatingbuoyantsupportby the mantle during rifling and breakup;and (4) the
isotopicandcompositional diversityrecordedin present-dayIcelandicbasaltsis observedin
many of the Palaeocenesequences, after crustalcontaminationand pressureof melt
segregation are takeninto account.The widespread and simultaneous activityof Phase1
activityrequiresan abnormallyhighmantleflux rate.Thismaybe associated with the arrival
of a start-upplume,but alternativelyit representsthe arrival of a pulseof hot mantle,
followinga periodof weak plumeactivityduringthe Cretaceous. In eitherscenario,the
igneousactivityappearsto havebeenfocusedalonglinesof weakness in the lithosphere.
Phase2 activityis closelylinkedto continental
breakup.Forcedmantleconvection, causedby
hot mantleflowingintothe developing riff zonesduringcontinentbreakup,may haveled to
the very high magmaproduction rates(two to threetimesthoseobservedin present-day
Iceland)thatformedtheSDRSandassociated deepcrustalintrusions.
1. INTRODUCTION

Large IgneousProvinces:Continental,Oceanic,andPlanetary
Flood Volcanism
The widespread magmatism that occurred throughout
the North Atlantic region during Palaeogene times
GeophysicalMonograph100
Copyright1997 by the AmericanGeophysicalUnion constitutesone of the first large igneousprovinces(LIPs)
46 THE NORTH ATLANTIC IGNEOUS PROVINCE
to be recognised.Giekie [ 1880] notedthe similarityof the
basaltic successionsin the British Isles, the Faeroes and
Iceland, and the idea of a broadly related, originally
contiguousprovince was further developedby Holmes
[1918]. His Brito-Arctic Province encompassed the now
dispersedbasalticlava piles and associateddifferentiates
and plutonic rocks found in NW Europe, the Faeroes,
Iceland, and East and West Greenland.In 1935, Richey
introducedthe term Thulean Province, and recently the
province has incorporatedthe igneous rocks found in
eastern Canada and along the margins of the North
Atlantic basin (Figure 1). In this paper, becausewe are
focusingon the developmentand influenceof the Iceland
thermal anomalythroughoutTertiary to Recenttimes,we
usethe generalterm North Atlantic IgneousProvince.
The North Atlantic Igneous Province includes the
basaltic and picritic lavas of Baffin Island and West
Greenland;the >5-km-thick sequences of continentalflood
basaltsin the ScoresbySund-BlossevilleKyst region of
East Greenland;the seaward-dipping reflectorsassociated
with the Greenland and NW European volcanic rifted
margins; the aseismicridges of the Greenland-Iceland-
Faeroes Ridges; and, of course, Iceland itself. Indeed,
Iceland provides a possiblemodel for the formation of
oceanic plateaus,which are the most enigmaticof LIPs
[e.g., Saunders et al., 1996]. In addition to the large
volume of extmsive rocks, the North Atlantic Igneous
Province also boastsan impressivediversityof intrusive
bodies.
The region has a long historyof explorationand study,
both academicand commercial,ensuringa large database
of geological information. Recently, interestin the region
has grown, with a succession of OceanDrilling Program
(ODP) legs(104, 152, and 163) on the Voting Plateauand
the SE Greenland margin; seismic profiling associated
with the Faeroes-Iceland Ridge Experiment (FIRE)
[Staples et al., 1996]; seismic profiling and on-land
investigationsalongthe Greenlandmarginsby the Danish
LithosphereCentre [Larsen et al., 1995]; and detailed
petrologicaland geophysicalinvestigations of the Iceland
neovolcaniczones and the ReykjanesRidge. Becauseof
the diversity of its magmaticproductsand the extensive
informationpresentlyavailable,the North AtlanticIgneous
Province provides an excellent opportunity to study
severalfacetsof LIP development,includingthe duration
of magmatism,the interactionbetween lithosphereand
sub-lithospheric mantle, and the role of a mantleplumein
LIP formation.
Consideredindividually,the onshorebasaltsequences
of the North Atlantic IgneousProvincecoversmallerareas
than many continentalflood basalts.However, when we
includethe offshoresequences, suchas the basaltsalong
the continent-ocean transitionand on the major offshore
plateaus(e.g., the Voting and Rockall Plateaus),the total
areaincreases to about1.3 x 106km2 [Eldholmand Grue,
1994]. This is comparablewith the estimatedoriginalarea
of the Deccan Provincein India [e.g., Mahoney, 1988].
Inclusion of the thickened crest of the Greenland-Faeroes
Ridge and Iceland makes the North Atlantic Igneous
Province one of the largestin the world [Macdougall,
1988].
An outstandingproblemof LIP formationis the extent
to which mantle plumesare involved in their formation.
Many workersnow acceptthat plumeshave a pivotalrole
in the formationof continentalfloodbasalts[e.g.,Morgan,
1971, 1972; Whiteand McKenzie,1989], oceanicplateaus
and aseismicridges[e.g., Vink, 1984; Coffinand Eldholm,
1994], and volcanic rifted margins [e.g., White et al.,
1987; Coffin and Eldholm, 1994]. The short duration of
the bulk of the magmatism associated with some
individual LIPs (e.g., Deccan [Courtillot et al., 1988];
North Atlantic [White, 1989]) implies an event of almost
cataclysmicproportions.No consensusexists, however,
whetherLIPs representthe arrival of a plumeheadat the
base of the lithosphere,with consequentoutburstof
magma [Richards et al., 1989; Griffiths and Campbell,
1990; Campbelland Griffiths,1990], or whethera plume
head growsmore slowly('incubates'),and the magmatic
releaseis due to extensionof the overlyinglithosphere
[e.g., White and McKenzie, 1989; Kent et al., 1992] (see
discussionby Saunderset al. [ 1992]).
Otherworkershaverelegatedthe role of mantleplumes
to insignificance or dismissed themaltogether.
Andersonet
al. [ 1992], for example,invoked'hotcells'--broadregions
of the mantlewhich have low seismicvelocityandwhich
by implicationare hotter than adjacentregions.In this
model, the buildup of heat is accomplishedby thermal
blanketingby the overlying lithosphere,and adjacent
regions are cooled by subduction.The geochemical
characteristics of hotspot-related
magmasmay be provided
by delamination of the continental lithosphere[e.g.,
McKenzieand O'Nions, 1983;AllOgreand Turcotte,1985;
Smith, 1993], thus eliminatingthe requirementfor plumes
connectedto an exotic,deep-seated mantlesource.Mutter
et al. [1988] and Zehnder et al. [1990] dispensed
altogetherwith the requirementfor thermal anomalies,
suggestingthat the excessmagmatismassociatedwith
volcanicrifted marginswas a result of enhancedmantle
convection,triggeredby the thermalcontrastbetweenold,
cold lithosphereand the warmer asthenosphere. Such a
processis necessary, they argued,wherethereis no clear
evidencefor a mantle plume. Which (if any) of
 SAUNDERS ET AL. 47

• Onshore Svalbard
Basalt flows & sills
(Spitsbergen)
;..• Offshore
Seaward-Dipping
•--• Reflector
Sequences
1 914 Drill Site

/ / •'0m.y.
70øN Svarten-
hoek •
60 m.y.
Nuussuaq
Wollaston

Disko Holdwith'
50 m.y. oreo
Scoresby

Biosseville

Kangerlussuaq 142
Greenland
SE Greenland Ridge
ß

/ 988
914-917,
989, 990 4o7i
oø 408
60øN
Ridge
-.
ß
ß' Faeroes
ß .,,.

ß::: ':i:i:ie':

ofiB
552-555 2,• ' ' '
, '"!Hatton
Z• ß
'"'
:: Bank '
,,, British Tertiary Igneous
Province
ZONE
50øN

30øW 15øW oo

Figure 1. Map of theNorthAtlanticshowingthe extentof theNorthAtlanticIgneousProvince,andthe locationsof

the main areasdiscussed in thetext (Baffin Islandnot shown)[modifiedafterLarsenet al., 1994]. Estimatedpositions
of the ancestralIcelandplumearetakenfromLawvetandMiiller [1994] (dashedline between60 and70 Ma indicates
uncertaintyaboutexistenceof the plumepriorto 62 m.y. ago).Alternativelocationsof plumeaxisare shownby W
(Whiteand McKenzie[1989], at anomaly24 time) andBr (Brooks[1973b]ca. 60 m.y. ago). Smallfilled circles
indicate location of DSDP and ODP basementsites.JMR: Jan Mayen Ridge; JMFZ: Jan Mayen Fracture Zone; Far-
Shet Basin: Faeroes-Shetland Basin.

models is correct has serious implications for An important aspect of the North Atlantic Igneous
understandingthe thermalstructureof the mantle and its Province is the relative ease with which we may gain
evolutionthroughtime. We shallreturnto thesemodelsin information about the present-dayhotspot. Many LIPs
Section 7. have been related to an extant hotspot (e.g.,
48 THE NORTH ATLANTIC IGNEOUS PROVINCE
R6union; Kerguelen Plateau-Isles Kerguelen; Paranti-
Tristan da Cuhna; Madagascar-Marion/PrinceEdward
Islands), but in all cases the plume thought to be
responsiblefor the formation of the LIP is now located
beneaththick lithosphere.Consequently,the volume and
composition of the magmas are restricted by the
lithosphericlid to moderate-to high-pressure melting[e.g.,
Ellam, 1992]. In the case of Iceland, however, the
proposedplume is ascendingvery close to the Mid-
Atlantic Ridge (MAR), and the mantle is thus melting at
least partly in the spinel field. It is thereforepossibleto
samplelow-pressuremelts from the presentday hotspot.
Furthermore,samplingis not restrictedto a focusedpoint,
as on most oceanislands,but is possibleacrossvirtually
the entire width of the thermal anomaly. As we shall
describe below, this has provided an opportunity to
investigatenot only the thermalstructureof the plume,but
its compositionalstructureaswell.
As a working hypothesis,we follow severalprevious
workers in acceptingthat a mantle plume is presently
located beneath easternIceland, and that this is the gun,
still firing, not merely smoking,responsiblefor much of
the activity of the North Atlantic IgneousProvince[e.g.,
Whiteand McKenzie, 1989, 1995]. In Section7 we discuss
other possiblehypotheses.These are the main questions,
commonto many LIPs, that we seek to addressin this
paper:
ßWhat are the age anddurationof the magmatism?
ßHow was the compositionof magmacontrolledby the
lithosphereand/or by more deep-seatedprocesses(for
example,a plume or a hotcell)?
ß What doesthe magmatismtell us aboutthe evolution
and structureof the sublithosphericmantle beneaththe
North Atlantic?
ß Did the arrival of a hot plumeinitiatemagmatismand
lead to continentalbreakup,or did lithosphericthinning
releasemagmafrom a pre-existingthermalanomaly?Was
this anomalycreatedby a slowly incubatingplume or a
hotcell?
2. TIME SCALE USED IN THIS CHAPTER
Much of the igneousactivity reportedin this chapter
occurredin the Cenozoic,mostly during Palaeoceneand
Eocene times. Several time scalesare used in the literature,
namelythoseof Hadand et al. [1982, 1990], Candeand
Kent [1992, 1995], and Berggren et al. [1985, 1995].
There are considerable differences between some of these
time scales,so it is importantto standardiseto one of
them. Cande and Kent [1992] presenteda detailedrevised
Cenozoicmagnetochronology basedon an evaluationof
seafloormagneticanomaliesin the SouthAtlantic and fast
spreadingregionsin the Indian and Pacific Oceans.They
useda Cretaceous-Palaeogene boundarycalibrationage of
66 m.y., basedon high-precision
laserfusion4øAr-39Ar
sanidinedates.Subsequentstudieshave shown,however,
that this age may be too old, and a consensusage of 65
m.y. has emerged for the Cretaceous-Palaeogene
boundary. Gradstein et al. [1994] also used the 65 m.y.
age as an anchorpoint for their Mesozoictime scale.This
date was subsequentlyused to revise the geomagnetic
polarity time scale[Candeand Kent, 1995], which formed
the basis of the geochronologyand chronostratigraphy
usedby Berggrenet al. [1995]. We usethe Berggrenet al.
[ 1995] time scalethroughoutthis chapter(Figure2).
As a prefaceto Sections3 and 4 we stressthat thereare
few high-precisiondates for many regions of the North
Atlantic IgneousProvince(althoughit may be one of the
better-datedLIPs). We have tried to restrictour discussion
to 4øAr-39Ar
ages(step-degassing
of whole-rocksamples
or laser fusion of feldsparsfrom volcanic rocks). K-Ar
ages are less reliable [e.g., Fitch et al., 1988], although
suchdatesare oftenthe only onesavailable.
3. THE TECTONIC DEVELOPMENT OF THE
NORTH ATLANTIC BASINS
For the purposesof this review we shall focuson the
area of the North Atlantic to the north of latitude 50øN,
including the continental borderlands of NW Europe,
Greenland and NE Canada (approximately the region
shown on Figure 1). Srivastava and Tapscott [1986]
provided a chronologyof the tectonicdevelopmentof the
North Atlantic region. Rifting successivelypropagated
northward, initially into the Labrador Sea and Rockall
Trough, and then into the NE Atlantic. In someinstances
the line of eventualseparationwas guidedby pre-existing
anisotropy in the lithosphere,but the thermal swell
associatedwith the ancestral Iceland plume may have
exertedan importantcontrolduring the Palaeogene[Hill,
1991].
A series of basins characterisesthe pre-Cenozoic
palaeogeographyof the North Atlantic region. These
basinswere partly precursorsto plate separation,but they
were also remnantsof basinsthat had been developing
throughoutthe Mesozoic.In the British Isles,the Tertiary
lavas erupted onto a varied substrate with a low
topography, probably developed during the Upper
Cretaceous[e.g., Watson, 1985]. Many of the eruptive
centresare either in, or along the margins of, Mesozoic
basinssittingon Proterozoicor Archaeancrust[Thompson
and Gibson, 1991]. Deep sedimentarybasinsoccur
SAUNDERS ET AL.
50 THE NORTH ATLANTIC IGNEOUS PROVINCE
the East Greenland margin (for example, the Jameson
Land Basin) although,like in the British Isles, lavaswere
eruptedthroughcrustof differentagesand composition.In
West Greenland, a complex graben systemwas already
developed in the Mesozoic, between the future Baffin
Island and Greenland [Henderson et al., 1976]. These
grabencontainthick sequencesof Cretaceoussediments
and perhaps Jurassic sediments as well. Subsidence
continued after the start of volcanism in the Palaeocene.
Seafloor spreadingwas underwayin the region to the
north of the Azores-GibraltarFractureZone by 126 Ma,
slowly propagatingnorthwardsso that by the Cenomanian
(Anomaly M0-34, 95 Ma), seafloorspreadingwas active
in all regions south of the Charlie Gibbs FractureZone
(-53øN). Farther north, rifting had alreadybegunin the
region of the futureRockall Troughduringthe Aptian (ca.
118 Ma), beforedevelopinginto seafloorspreadingduring
the Albian or Cenomanian[Robertset al., 1981]. Seafloor
spreadingbetweenCanadaand Greenlandwas preceded
by a longperiodof slow extension[Chalmers,1991].
Seafloor spreadingwas underway in the Atlantic in all
regions southof Greenlandby Campaniantimes (ca. 84
Ma). Spreadingin the Rockall Troughstoppedcompletely
during the Campanian or Maastrichtian [Kristoffersen,
1978; Roberts et al., 1981]. There is some debate about
when seafloor spreadingbegan in the Labrador Sea.
Srivastava[ 1978] introducedthe commonlyacceptedplate
kinematic model, suggestingthat the earliestidentifiable
anomaly in the northernLabrador Sea is anomaly31. In
Roest and Srivastava's [1989] recent revision, however,
the oldest preserved anomaly is 33 (ca. 80 Ma), but
Chalmers[1991] suggestedthat in the northernLabrador
Sea spreadingprobablybegan as late as the Palaeocene
(anomaly 27). Seafloor spreading slowed down
considerablyby 50 Ma, and had stoppedby 36 Ma. The
main axis of spreadingmigrated into the NE Atlantic,
between Europe and Greenland,during the early Eocene,
Chron24 (C24) time [ Vogtand Avery, 1974].
Prior to about 62 Ma there is little evidenceof large-
volume magmatismin the North Atlantic region. There is
no evidence that the Cretaceous ocean crust is thicker than
normal (6-8 km). Anton Dohrn [Joneset al., 1994] and
RosemaryBank [Hitchen and Richie, 1993; Morton et al.,
1995], Late Cretaceousseamountsat the northernend of
the Rockall Trough and a sill of tentativeCampanianage
in the Faeroe-ShetlandBasin [Fitch et al., 1988] indicate
sporadicoutburstsof pre-Cenozoicmagmaticactivity in
the region, over and abovethat associatedwith rifling and
extensionin the Rockall Trough. The preciseextent and
causeof this Late Cretaceousactivity, and its relationship
to the laterPalaeogeneoutbursts,are presentlyunknown.
After the Cretaceous,however, the style of magmatism
and sedimentation changed dramatically. Broadly
speaking,two episodesor phasesof magmatismoccurred
duringthe Palaeogene[e.g., Whiteand McKenzie,1989].
An episodeof continent-based magmatismat 62-58 Ma
(Phase1) producedlargevolumesof basaltandassociated
intrusive rocks in the future eastern Canada (Baffm
Island), West Greenland,SE Greenland,the BritishIsles,
and possiblyparts 'of the central East Greenland
successions. This activityprecededthe main breakupand
separationof Greenlandfrom NW Europeby about4 m.y.
Local uplift aroundseveralbasins,such as the northern
North Sea Basin and the Faeroe-Shetland Basin, is
indicated by the input of clastic sedimentsinto what
previously had been low-energy depositional
environments [Anderton, 1993]. This uplift began in
Danian times, reachedits peak in the Thanetian(58-55
Ma) and was followed in someareasby rapid subsidence
[e.g.,Englandet al., 1993; Turnerand Scrutton,1993]
Plate separationand ocean crust formation, which
beganat approximatelyC24r time (56-53.5 Ma) [Vogtand
Avery, 1974], was accompaniedby the formationof thick,
seaward-dippingreflector sequences(SDRS) along much
of the E Greenlandand NW Europeanmargins(Phase2).
Larsen and JakobsdSttir[1988] have estimated,on the
evidenceprovidedby seismicprofilesacrossthe SDRS in
EastGreenland,that duringbreakup,volcanicproductivity
along the North Atlantic rift zone was almostthreetimes
higherthan in present-dayIceland.Eruptionratesduring
the initial stagesof plate separationwere higherthan any
subsequently recordedin the North Atlanticbasin.Where
recovered,most of the lavas of the SDRS were eruptedin
a subaerialor shallow aqueousenvironment,indicating
dynamicsupportof the plate margins[Hinz, 1981; Cliff et
al., 1995]. Anomalouslythick basalticcrust(up to 35 km
thick [Bott, 1983; Whiteet al., 1995; Stapleset al., 1996])
continuedto be producedbetween Greenlandand the
Faeroesto form the Greenland-lceland-Faeroes Ridgesand
Iceland itself.
Plate tectonic reconstructionsplace the centre of the
Iceland hotspot beneath central Greenland during the
period60-50 Ma, althoughthereis someuncertainty about
its precise location. Lawvet and Miiller [1994], on the
basis of relative plate motions, located the plume axis
close to the centre of Greenland between 60 and 50 Ma
(Figure 1), whereasBrooks [1973b], Brooksand Nielsen
[1982b] and Whiteand McKenzie [1989] placedit farther
to the east, close to Kangerlussuaq,during the time of
eruptionof the EastGreenlandbasalts.Thereis, however,
little evidenceof a hotspot'track' until the formationof
the Greenland-Faeroes Ridge, but this may simply
 SAUNDERS ET AL. 51

the lack of surfaceexpressionof magmatismdue to the

Anomalies 26 to 24
thick cratoniclithospherebeneathcentralGreenland.The
Phase 1 magmatismoccurredover a widespreadarea, 76 ø
Vering
some2000 km across(outlineby the bold dash-dotline in Extent of Phase 1
magmatism
Figure 3), requiring a broad thermal anomaly [White,
1988], and perhaps associatedwith a startup plume
[Griffiths and Campbell, 1990; Campbelland Griffiths,
1990]. The Phase2 magmatismwas initially concentrated 68 ø
along the lines of plate separation,but still requireda
broadthermalanomalyto accountfor the dynamicsupport BTIP
of the plate marginsandthe huge,widespreadvolumesof
melt [Whiteet al., 1987]. We returnto this point in Section
7. Magmatismcontinuedlongerin EastGreenlandthanin 60 ø

the British Isles, reflectingthe proximity of the ancestral

Icelandplume afterplateseparation. Possible thermal
extent of 2000rn

4. THE NORTH ATLANTIC IGNEOUS PROVINCE: axisymmetricplume

52 ø
A REVIEW OF AGE, DURATION, AND MAIN
CHARACTERISTICS OF THE MAGMATISM I Flows
&sills,(58-62
Phase
continental 1
Ma)

K-X--X] MajorSDRS Phase 2

Reviews of the magmatism in the North Atlantic _.':'•
.-':_
ß Flows& sills (<56 Ma)
31 251 25 \
IgneousProvinceare providedby Dickin [1988], Mussett
et al. [1988], Upton [1988], Whiteand McKenzie [1989], 68 ø 60 ø 52 ø 44 ø 360

Larsenet al. [ 1994], andRitchieand Hitchen [ 1996]. Each

Figure 3. Tectonic reconstructionat Chron 24 time (ca. 55 Ma),
of the following subsectionswill briefly considerthe but including the magmaticactivity from C26 throughC24 time.
geographicalextent of the sub-province,the age and Location of continental blocks and plate boundaries from
duration of the magmatic activity, the magmatic Srivastava and Tapscott [1986]; estimated positions of the
productivity,andprovidea summaryof the geochemistry ancestralIceland plume axis at 60 and 55 Ma from Lawvet and
of the main igneousunits.We concentrateon the basaltic Miiller [1994]. Note that White and McKenzie [1989] place the
lavas,becausethey are oftenthe mostvisibleexpressions axis of the plume at 55 Ma closerto the centralEast Greenland
volcanic successions. Ornamentation of the volcanic successions
of magmaticactivity and the most extensivelysampled
and studied. is basedon whetherthe activity is Phase1 or Phase2, ratherthan
continent- or ocean-based. The bold dash-dot line indicates the
Lavas were eruptedin three differenttectonicenviron-
extent of the early (Phase 1) magmatism.The dotted circle is
ments: (1) onto continentalcrust, forming continental drawn assumingan axisymmetricplume head centred on the
flood basalts;(2) along the continentalmargins,where Lawvet and Miiller [ 1994] 55 Ma positionof the plumeaxis.
they form the off-lapping lava piles of the SDRS in a
transitionalor fully oceanicsetting;and (3) in a truly
oceanic setting along the Greenland-Iceland-Faeroes excessmagmatism, over and above that at normal mid-
Ridges.This differencein eruptivesettingis reflectedin oceanridges. Secondly,the basaltsfrom Iceland and the
variationsin composition,the continentalflood basalts adjacentridge segmentsprovide compositionaldata about
oftenshowingevidenceof modificationby the continental the underlyingplume. They thus providereferencepoints
lithosphere,and in timing, the continentalsequences on Sr, Nd and Pb isotope, and chemical variation
generallybeingolderthanthe SDRS. diagrams. Thirdly, gravity and topographic (and
bathymetric) data acrossthe island and adjacentseafloor
4.1. Iceland help to constrainthe thermalstructureof the plume.
The geologyof Icelandcanbe dividedinto axial rift and
For the purposeof this review,Icelandis importantfor marginal zones [e.g., Saemundsson,1986]. The axial rift
three reasons.Firstly, the crustalstructureand magmatic zonesor neovolcaniczones,definedon the basisthat they
plumbing systemsprovide potentialanaloguesfor the are younger than the BrunhesChron (0.7 Ma), are the
formationof PalaeogeneSDRS alongthe marginsof the main regions of crustal formation. dakobsson[1972] has
Atlantic Basin. Both environmentsare characterisedby estimatedthat the volcanic dischargein the
52 THE NORTH ATLANTIC IGNEOUS PROVINCE
zonesis in the orderof 40 km3/1000y (approximately1.0
x 10-4 km3 per km of rift per yearfor a 400-km-long
rift
zone), with higher effusion rates near the centre of the
island.Nearly 90% of the lavas are tholeiiticbasalt,with
small amounts of acid and intermediate rocks. Both central
andfissureeruptionsarefound.Zonesof limitedextension
occur as branchesto the main rift zones in three regions.
One of these is the Snaefellsnes Peninsula, where the
averagedischargerate of lavas is lower than in the main
rift zones [dakobsson,1972], and where the lavas are
mainly transitional to alkalic in composition,consistent
with small-degreemelts derivedfrom greaterdepthsthan
the tholeiites of the main rift zones.
The marginal zones,representingcrustmore than 0.7
m.y. old, make up more thantwo-thirdsof the areaof the
island. The oldest crust in Iceland is found in the northwest
part of the island,where basaltsof up to 15 m.y. old have
been recovered [Moorbath et al., 1968; McDougall et al.,
1984]. In easternIcelandthe oldestcrustis approximately
13-14 m.y. old [Watkinsand Walker, 1977].
The Icelandic crust is approximately25-30 km thick
[Bjarnassonet al., 1993; Bott, 1983; White et al., 1995,
1996], and at leastthe upper 10 km compriseslava flows.
The flows dip gentlytowardsextantandextinctrift zones,
the dips increasing in the vicinity of the axial zones
[Walker, 1960]. The metamorphicgradeincreasesthrough
the lava pile, suchthat in the Tertiary sequences it i'snot
uncommonfor the flows exposedon mountaintopsto be
unmetamorphosed,whereas flows from valley floors
preserve zeolite or lower greenschistgrades [Walker,
1960]. In the main sequencesof Tertiary lavas, the
geothermalgradientis estimatedto be approximately40-
60øC/km; in the axial zonesthis locally risesto 100øC/km
[Pdlmason and $aemundson, 1974; Pdlmason et al.,
19781.
Pdlmason's[1980, 1986] modelprobablybestdescribes
crustal formation in Iceland, albeit with modifications
suggested by Menke and Levin [ 1994]. In this model,the
excessmagmatismalongthe axial zone causesoverloading
of the lithosphere,leadingto subsidence of the crust.Lava
flows eruptedclosestto the axis will subsidethe greatest
amount and be buried beneathsubsequentflows. These
rift-proximal flows subsideand form the deeperportions
of the Icelandic crust and undergo progressivezeolite,
amphiboliteandgranulitefaciesmetamorphism. The distal
fringes of the largest flows escapingthe rift zone,
however, avoid burial and remain close to the surface.
Smaller flows unable to escapethe axial rift will not
remain at the surface.
The modelpredictsthatthe axial zoneis underlainby a
partialmeltzoneat about12 km depth,consistent
with low
resistivityvaluesat depthsgreaterthan about12 km under
the axial zone [Beblo and Bj•irnsson, 1980]. Recent
geophysicalstudies,however,suggestthat the lower crust
may be cooler, even in the axial zone [Menke and Levin,
1994]. It is possiblethat there is far greaterheat lossvia
hydrothermalcirculationfrom high-levelmagmabodiesin
the axial zone,thanpredictedby the Phlmasonmodel.
As pointed out by Iresland [1983] and Hardarsonand
Fitton [1994], the basaltspreservedin the surficiallava
sequences of the marginalzoneswill providea preferential
sampling of the larger eruptive units. Material erupted
solelyin the rift axiswill not be preservedfor near-surface
sampling.This is important,becauseit is evidentthat the
spectrumof compositions seenin the neovolcaniczonesis
far wider than in the Tertiary lavas on Iceland. Basalts
from the neovolcaniczones range from light-rare-earth-
element (REE)-enriched to strongly light-REE-depleted
(Figure 4), and Zr/Nb rangesfrom 5 to 35. In comparison,
the Tertiary sequences have a much more restrictedrange
of Zr/Nb (9+2 [Hardarson and Fitton, 1994]), and the
REE showuniform, slightlight-REEenrichment.
Schilling et al. [1982] suggestedthat this variation is
due to changesin the compositionof the mantlesource(a
greater involvement of light-REE-depletedmid-ocean
ridge basalt(MORB)-like mantlein more recentIcelandic
eruptions),but it is more likely that the earlier Tertiary
sequences result from homogenising of small,
compositionally varied,magmabatchesin largechambers.
The two tectonic environments on Iceland, axial and
marginalzones,thusprovidedifferenttypesof information
about the Icelandic sourceand the magmatic plumbing
systems.
The Phlmason model also adequately explains the
morphology and formation of PalaeogeneSDRS, the
offiappingappearanceof SDRS being similar to the rift-
dipping structureof the Icelandiclava piles [e.g., Gibson
and Gibbs, 1987; Larsen and Jakobsd6ttir, 1988]. By
analogywith the Icelandicsetting,the accessible portions
of the PalaeogeneSDRS may not provide detailsof the
full rangeof availablemagmatypes;lavaseruptedcloseto
the rift axis will have subsidedto depthswell beyondthe
reachof present-daydrilling.
4.1.1. Temperatureof the Iceland plume. Although
there is generalagreementthat the thick crustassociated
with the Greenland-Iceland-Faeroes Ridge is the resultof
melting of hot mantle, there is no consensusaboutthe
precisetemperatureof the mantlebeneathIceland.On the
one hand, White et al. [1995] suggestedthat the excess
temperaturein the core of the plume is of the order of
200øC, with a disk-shapedthermal anomaly extending
some 1350 km from the plume axis. In this model,
 SAUNDERS ET AL. 53

100 --
--

--
Iceland Neovolcanic Zones
passivelyin responseto plate separation.Although this is
--

valid in the caseof normal oceancrust,it is unlikely to be

.50-- ß QzTholeiite, ERZ true for Iceland, where part of the uplift of the island is
• , QzThole.!ite,
ERZ
--

thoughtto be causedby dynamic uplift from the plume.

Consequently,the mantle flow rates throughthe melting
zone are likely to be enhancedconvectively,producing
greater volumes of magma for a given potential
temperatureand implying that the mantle is cooler than
U) -
predictedby the passiveupwellingmodel. Indeed, Whiteet
al. [1995] noted that the predictedvolume of melt from
• ? OITholeiite,
ERZ
REE inversionmodelling of data from Krafla, closeto the
x,,cc ß O}T.
holeiite,
ERZ
axis of the plume, is only 16 km. This is significantlyless
• ; pPiicCrriitt;
' EERRz
Z than the observed25 km crustal thickness,and implies
I I I I I Sm
La Ce Pr NdI EuI Gd
I ;b Dyi Hoi Eri Tm
i Ybi Lui forced convection.
lOO Ribe et al. [1995] applied numericalmodels for plume
buoyancy,flow, and melting to derive their temperature
contrasts.When they usedSchilling's [ 1991] estimatesof
the plume flux (1390 kg s-l) and excesstemperature

Iceland
Kolbeinsey
ReykJanes Ridge
I - Ridge o Mohns
• oo Ridge
^ o.•t•I• • j:• Knipovitch
- q:}o3'"oc•• •5 u'-',O o Ridge

0.1
RbBa hNbTa K LaCeSrNd PSmZrHfEuTi b YYbLu
87sr/SSsr
14
Figure 4. Chondrite-normalisedREE data, and bulk-silicate
Earth (BSE)-normalisedtrace element data, for lavas from the
Iceland neovolcaniczones. Data source:Hdmond et al. [1993]. 12
BSE normalising values from McDonough and Sun [1995].
Chondritevaluesfrom Nakamura [1974]. o
10-

plume has a narrow vertical conduit, some 50-100 km in

diameter[e.g., Vogt, 1983; White, 1988], and the disk-like
head cools progressivelyto ambient temperaturesat its o 09
margins. Ribe et al. [1995], on the other hand, modelled 6
the samesystemassuminga diffuse zone, at least 600 km 50 55 60 65 70 75 80
in diameter,of warm (AT = 90øC),upwellingmantle.
Latitude ('N)
Three direct observations provide the key input
parametersfor thesemodels.Theseare the thicknessof the Figure5. Variationof 87Sr/86Sr
andfractionation-corrected
Na
crust,the positiveresidualdepthanomalyof zero-agecrust and Fe along the Mid-Atlantic Ridge, in the vicinity of Iceland.
Note the decrease in 87Sr/86Srto MORB-like values at about
along the MAR and on Iceland, and the compositionof
60øN, whereasFEB.0-remainselevateduntil about 52øN. If high
eruptedbasalts(see Figure 5). White et al. [1995] used a
combination of these observations to derive their model.
FEB.0(and low NAB.0)reflectsdeeper,more extensivemeltingas
suggestedby Klein and Langmuir [1987], this diagram implies
The volume of melt (or crustalthickness)is relatedto the that the thermalsignatureof the plumeis more extensivethanthe
temperatureof the mantle source [McKenzie and Bickle, isotopicsignature.La/Yb showsa similar distributionto that of
1988]. However, the parametersused by White et al. 87Sr/86Sr.
Data sources:
Fe andNa fromKlein and Langmuir
[1995] assume that the mantle decompressesand melts [1987];87Sr/86Sr
fromHart etal.
54 THE NORTH ATLANTIC IGNEOUS PROVINCE
(AT=263øC), the predictedtopographyabove the plume
axiswas 10 km, corresponding to a crustalthicknessof 68
km. As they said,"this far exceedsthe crustalthicknessof
20-34 km predicted for the same AT by purely passive
upwelling [McKenzie and Bickle, 1988; Klein and
Langmuir,1987]." Their preferredmodelrequireda plume
with a AT of 93øC, a buoyancyflux of 1370 kg s-1, a
volumeflux of 193 m3s-1, anda meanupwellingvelocity
of 2 cm y-1. The predictedtopographic anomalymore
closely resembledthe observedone, althougha cooler
plumewould probablyhave givenan evenbetterfit.
It is difficult to assessthe Ribe et al. [1995] and White
e! al. [1995] models using independentchemicalcriteria.
Recently, however, Wolfe et al. [1997] presentedthe
resultsof a regionalbroadbandseismicexperimentwhich
indicated a cylinder of low velocity material beneath
Iceland.Their datasuggestthatthe cylinderhasa diameter
of approximately300 km at between 100 and 400 km
depth, which is wider than the 'narrow stem' model of
Whiteet al. [ 1995]. Excesstemperaturemay be asmuchas
200-300øC in the model of Wolfe et al. [1997], but
predictionof temperaturefrom P and S wavesis fraught
with uncertainty.
The models mentioned here provide possible end-
member conditions;in reality, intermediatetemperatures
and fluxes probably pertain. The likelihood of forced
convectionmeansthat the Whiteet al. [1995] and Wolfeet
al. [ 1997] estimatesof mantletemperature aretoo high. It
also means that the existing REE inversionmodels are
inaccuratefor the plume-ridgeenvironment.Melt volumes
will be increasedover thosein purelypassiveupwelling
modelsby a factor approximatelyequivalentto (rate of
mantleupwelling)/(rateof plate separation).In Icelandthis
factor will be >1 and, during the initial stagesof plate
breakup above a mantle plume, may be >>1. The
excessivemelt thicknesses predictedby numericalstudies
may be alleviated if some of the plume material flows
alongthe adjacentmid-oceanridge,but it is difficultto see
how effective this mechanismwould be in removing
excessthermalenergy.It is alsolikely thatthe temperature
and plume flux have varied slightlywith time. V-shaped
ridgesin the oceancrustto the southof Icelandimply
short-livedincreasesin magma productionrates, which
equateto transienttemperatureincreasesof about30øC if
the decompression is entirelypassive[Whiteet al., 1995],
lessif thereis a componentof forcedconvection.
4.1.2. The composition of the Iceland plume. An
extensiveliteratureexistson the compositionof Icelandic
basalts and differentiates [e.g., dakobsson, 1972;
dakobssonet al., 1978; Imsland, 1983; Steinthorssonet al.,
1985; Oskarssonet al., 1985; Hdmondet al., 1993]. There
is a wide rangeof compositions, from stronglylight-REE-
depleted picrites to moderately light-REE-enriched
tholeiites,alkali basalts,trachytesandrhyolites(Figure4).
Multielementplots showpositiveNb anomalies,a feature
of most ocean island basalts[e.g., Tarney et al., 1980;
Hofmann et al., 1986; Saunderset al., 1988]. The more
primitive basaltsand picrites show positive Ba and Sr
anomaliesthat may be due to crustalassimilation[e.g.,
Hdmondet al., 1993] but which alternativelymay be a
source characteristic.
It hasbeenlong recognisedthat the Icelandicbasaltsare
samplinga 'depleted'mantlesource;see,for example,the
end- 87Sr/86Sr
diagram(Figure6A) [e.g.,O'Nionset al.,
1977; Zindler et al., 1979; Hdmondet al., 1993]. Data for
many picrites and tholeiites from the neovolcaniczones
overlap with North Atlantic mid-ocean ridge basalt
(MORB) compositions,
but othersamples,includingalkali
basalts from the SnaefellsnesPeninsula and elsewhere,
extendthe Icelanddatafield to higherS7Sr/S6Sr
andlower
end values.Like many other oceanislands,therefore,the
Icelandbasaltsdo not definea singleisotopiccomposition
but form an array in isotopespace[e.g., Sun and dahn,
1975; Zindler and Hart, 1986; Furman et al., 1995]. To
producethe low end values by modificationof a melt
from a MORB source(for example,by assimilationof
Icelandic crust) requires a componentmuch older than
Iceland, so the mantle source beneath Iceland must be
isotopicallyheterogeneous [Hdmondet al., 1993].
Trace element abundancesare controlledin part by
fractionation processes(partial melting and fractional
crystallisation)but incompatibleelementratiosprobably
reflect more closely the compositionof their source.
Nonetheless, if the amountof partialmeltingis sufficiently
small,eventhosetraceelementswith very similarand low
bulk distribution coefficients in mantle lherzolite will be
fractionatedfrom one another.Similarly, melt extraction
will leave a residue with drastically changedelement
ratios, so remeltingthis residuecan, in theory,produce
liquids substantiallydepletedin the more incompatible
elements.This was the principle behind the dynamic
partial melting models suggestedby Wood [1979b] and
Elliott et al. [1991] to accountfor incompatible-element
depletedbasaltsand picrites in Iceland. Despite these
caveats,however, there is a clear correlationbetweentrace
elementabundancesand isotoperatiosin Icelandicbasalts
(for example,Zr/Nb and end, Figure 7A), implyingthat
the range of incompatible element patternscannot be
producedby partialmeltingprocesses alone.
The implicationis that the Icelandplume comprisesa
'depleted'end-memberand a 'less-depleted' end-member,
relative to estimatesof primordial mantle. These
 SAUNDERS ET AL. 55

12
.......... Faeroes (LREE-depletedUpper Series)

10 '""""•
MORB' :.. SEGreenland Hatton
nk

Hatton
Bank
--.-.-..-
•.•:::i•:::•:•:
••
Faeroes
(LREE-

Iceland
QT
....
• AB
•
Skye (PreshalMore)

Vering Plateau
A (upper
Series)
i i i
o
0.7020 0.7025 0.7030 0.7035 0.7040
2o
Atlantic MORB

lO -• Figure
6A Vering
Plateau
(Lower Series)
•Mull Plateau
Lavas
Faeroes

-lO
(Sample
K1
)
z -20

-30
•;::::'--.:.-::::-'.•
'-:-'-•
SE Greenland
Lava
Main ,rnphi
Series

- (Site 917,
LowerSeries) Felsic Veinsin
Granulites

-B
-40

Archean Granufites
-50 I I I I

0.702 0.704 0.706 0.708 0.710 0.712

87Sr/86Srt
Figure6. endt versus
87Sr/86Srt
forbasalts
andrelated
rocksfromtheNorthAtlanticIgneous
Province.
A: expanded
partof the diagramshowingthe variationin Icelandicbasaltsandpierites(PB - picriticbasalts;OT - olivinetholeiites;
QT- quartztholeiites;AB - alkalibasalts),variousSDRSbasalts(SE Greenland,HattonBankandthe UpperSeriesat
Site 642, Voring Plateau),basaltsfrom the FaeroeIslands('Faeroeslight-REE-enriched basalts'are from the Lower,
Middle andUpperSeriesof theFaeroePlateauLavaGroup),andSkye(PreshalMore basalts).'SE Greenland'basalts
are from Site 918, and selectedsamplesfrom the UpperSeriesat Site 917. B. Main diagramillustratingthe wide
variationof gNdand87Sr/86Srin theNorthAtlanticIgneous Province,
andmostlyreflectingthe effectsof crustal
contamination.Faeroes
sampleK1 is a highlycontaminatedbasaltfromtheUpperSeriesof theFPLG.VoringPlateau
(Lower Series)rocksare predominantly dacites.Qualitativecontamination
vectorsfor amphiboliteand granulite
crustalcontaminationare shown.Data sources:Atlantic MORB -/to et al. [1987]; Faeroes- Garidpyet al. [1983];
Iceland- Hdmondet al. [1993] and references
therein;Voring Plateau-Taylor and Morton [1989]; HattonBank -
Macintyre
andHamilton
[1984];MullPlateau
lavasandArchaean
felsicveinsfromgranulite
- Kerr[1995a];Skye
lavas- Dickin [.1981],Thompson et al. [1972, 1980, 1982, 1984];SE Greenland- Fitton et al. [1996a]. Samplesage-
correctedto approximate ageof emplacement wheredataallow;otherwise present-dayvaluesareplotted.The shiftdue
to agecorrectionis smallcomparedto the overallrangeof valueson these
56 THE NORTH ATLANTIC IGNEOUS PROVINCE

40 0.025

Iceland
35 _ (Neovolcanic
Zones)

3O -

.......
•wu • •' •ateau
VeringPlateau
(/R • •'""••... -"••
' ' "I :.:::••:.,:,•?
roes(2)
25 0.04
- Lavas • "-' •. •:•••••:..•. •••• -
• ':'":•,••'• SE Greenland

N
- -=• •---•
•• • ii•::•ii•ii•,;,••,•;•'•
:::=::::::•
r', (Site
918) _
0.05
15

10 0.1

5
rou,rou, ß Zr/Nb -
0.2

A GreenlandGreenland
0 I
0 2
9O
.......
I I
4 6
• inTe•iarybasalts
• Iceland '
I I
8 10
I

80

70
0.015

60

SEGreenland Feeroes
(1)
0.02
(Site917,LS) Feeroes
Skye(2) I z
-• SEGreenland • (Sample
K1)• I IM,,R
N 40
I 0
(Site
917,
70
MS)50 •
__.-:.•
:::......-::._
-'---:'..:i:i:._ / I
......... 3o -•.-•.•..-.-...:.:.•.....:...••o 5
I12,• Feeroes (2)
-
0.03
N

30
Felsic
Veins in
........
"'"'":'"'"""•"•'ilililililililililililiiiiii'ii:'::'::'"'•,,•Mull
PlateauLa•as •
Iceland
- 0.04
20 -Granulit i• .......
"•e (3) • - 0.05

es,•o
Skye
(4)Vering
•
Vering
10 70Plateau(LS)50 Plateau -0.1
-B o ......... (us)
SE Greenland (Site 917, LS',
0 I I I I I
-50 -40 -30 -20 -10 0 10 2O

sNdt

Figure7. Zr/NbversessNdin lavasfromtheNorthAtlanticIgneous Province.

(A) Detailshowingthevariationfor
sampleswithend > 0, andtheclearoverlapbetweenIcelandandsomebasaltsuitesfromtheNorthAtlanticIgneous
Province.
Notethepositivecorrelation
of between
Zr/Nbandend in Icelandic
basalts.Groups1 and2 referto basalts
andrelatedrocksfrom the LowerBasalts,Kangerlussuaq, centralEastGreenland[Holm, 1988]. (B) Plot showingthe
full scaleof variation.Note the subhorizontal
dispersion
indicativeof mixingbetweena high end sublithospheric
sourceand a low-eNd contaminant.Two bulk assimilationlines are shown,betweenArchaeanfelsic veins and two
end-membersof the Icelandarray(figuresgivepercentage
of contaminant).
Faeroes
(1) refersto light-REE-depleted
basalts
theUpperSeries of theFaeroePlateauLavaGroup;Faeroes
(2) represents
theremainder of theFaeroes
basalts,
except forsample K1, a highlycontaminated
basalt, whichisshown separately.
'Skye1': Preshal
Morebasalts; 'Skye
2, 3 and4': SkyeMainLavaSeries. Datasources: asfor Figure6, plusWestGreenland
- Holmet al. [1992;1993];SE
Greenland (Sites917 and918) fromFittonet al. [1996a,
b]; HattonBankZr/Nbvalues- BrodieandFitton[1996].
Rangeof TertiaryIcelandbasaltcompositions fromHardarson andFitton[1994].FieldforMORBassembled from
dataof Ito et al. [ 1987] andunpublished
dataof the

membersare approximatelydelineatedby the picritesand
alkali basalts,respectively,at least in isotopicterms.The
depletedend-membersuperficiallyresemblesMORB but
Thirlwall et al. [1994] have shown that Icelandic basalts
andpicriteshavehigher2øSpb/2ø4pb at a given2ø7pb/2ø4pb These complexeswill not be discussedin detail here;
than North Atlantic MORB. Fitton et al. [1996b, 1997]
have also shown from Zr-Nb-Y correlations that the and Thompson[1982]. Basaltdyke swarms,oftenfocused
depleted Iceland end-member is different from MORB
(see Section 7). It would appear, therefore, that the
depletedend-memberis an intrinsicpart of the Iceland
plume, as suggestedby Hdmondet al. [ 1993], Thirlwall et
al. [1994], and Kerr et al. [1995b], and not entrainedor
advected MORB mantle.
There are several implicationsof this two-component
mantlemodelthat have a bearingon studiesof otherparts
of the North Atlantic Igneous Province, if the ancestral
Iceland plume was indeed responsiblefor much of the
magmatism.
ß It is necessaryto considerthe Icelandplume not as a
singlepoint in isotopeor elementspacebut as a spectrum
of compositions(e.g., Figure 7). Both 'depleted• and 'less- 'erodedbasalwrecks' of large centralTertiary volcanoes.
depleted'basaltsmay be derivedfrom the plume.
ß The 'less-depleted'end-membermay be associated
with a mantle lithology that has a lower meltingpoint than much more extensiveregion of Tertiary volcanicactivity
the depletedend-member;for example as veins [Wood,
1979a;Tarneyet al., 1980] or streaks[Fitton and Dunlop,
1985] in more depletedperidotite.The proportionof 'less-
depleted' to 'depleted' componentscontributingto the
melt will be a functionof the conditionsof meltingaswell
as the compositionof the source.Deeper, small-degree
meltingwill enhancethe effectsof the 'less-depleted' end- types were identified by Bailey et al. [1924]: the Non-
member,as suggestedby Hdmondet al. [ 1993] and Hards
et al. [1995] for the Icelandic alkali basalts. More
extensivemeltingwill homogenise the system.
ß The extent of melting will tend, on average,to be
greatestin the axial rift zonesnear the axis of the plume,
where the hottest mantle is allowed to decompressthe
most.Melt segregationwill preferentiallyextractthe more
fusible material and progressivelychange the bulk
compositionof the source.This processmay producea
radialchemicaland isotopicgradientin the plumesource.
We shall return to thesepointsin Section7, but they
need to be borne in mind for the remainder of Section 4.
4.2. British Tertiary IgneousProvince
A recordof Palaeogeneigneousactivity is preservedin
NW Scotland (including the classic areas of
Ardnamurchan,Skye, Mull, Rum, Eigg, Muck, Canna,and
Arran), Ireland (Antrim and the Moume Mountains),
Lundy, and offshoreregionssuchas the Rockall Plateau,
SAUNDERS ET AL. 57
Irish Sea and the Faeroe-ShetlandBasin. Tertiary
volcanism in western Britain tends to be focused around
central intrusive complexes, which comprise a wide
varietyof igneousrocksrangingfromperidotiteto granite.
excellentreviewscan be found in Emeleus[1982, 1991]
towardsthe complexes,indicatelocalcrustaldilationof up
to 25% [Speight et al., 1982] and have cross-cutting
relationshipsthat show emplacementat different times
throughoutthe province'shistory.
The igneous rocks of the British Tertiary Igneous
Province,or HebrideanProvince,have a long historyof
investigation.In the latter half of the nineteenthcentury,
discussionof the British Tertiary IgneousProvincewas
dominatedby Archibald Giekie and John Judd. Giekie
[1867, 1888] believed that the lavas erupted through
fissures,which are now representedby the great dyke
swarms.Alternatively, Judd [1874, 1889] proposedthat
the igneous complexes of Western Scotland were the
During this period Giekie [ 1880] was the first to recognise
that the British Tertiary province was only a part of a
encompassing the Faeroesand Iceland.
The publicationof the Mull memoir of the Geological
Surveyof Scotland[Bailey et al., 1924] was a milestonein
the historyof igneouspetrology.It established for the first
time the conceptof 'magma types' and 'magma series',
basedon the igneousrocks of Mull. Two main magma
PorphyriticCentralType and the PlateauType. Thesetwo
typeslaterbecameknownasthe tholeiitic[Kennedy,1933]
and alkali olivine basalt [Tilley, 1950; Tilley and Muir,
1962] types,respectively.
The relationshipbetweenthesetwo magmatypeswas
debatedby authorssuch as Bowen [1928] and Kennedy
[ 1933], and the British Tertiary IgneousProvincefeatured
prominentlyin thesediscussions. In his studyof Hawaiian
and Hebridean lavas, Wager [1956] proposed that
tholeiiticlavaswere the resultof partialmeltingof a layer
of peridotite at a high structurallevel, and that alkali
basaltswere derivedby partialmeltingat a muchdeeper
level. Wager, therefore,proposeddepthof melting as a
criticalfactorin determiningmagmatype,a conceptwhich
is now central to models of magma genesis and
composition.
Like the lavas in East Greenland, those of the British
Tertiary Volcanic Province were erupted onto a varied
surfaceof Precambrianto Cretaceousage. Most of the
landscapewas erodedto low relief by Late
58 THE NORTH ATLANTIC IGNEOUS PROVINCE
times. Thin sedimentary deposits of that age are not
uncommon,althoughthere is good evidencethat in some
areas even this cover had been removed or thinned before
the lavas were erupted. The extensivedevelopmentof
clay-with-flintsdeposits(for example,in Antrim, Northern
Ireland) implies a period of subaerialweathering and
erosion during Late Cretaceous/EarlyPalaeocenetimes
[e.g., Wilson and Manning, 1978]. Pre-basaltTertiary
sedimentsare localised and scarce,and the environmentof
depositionvaries considerably.Sedimentsassociatedwith
the Eigg Lava Formation [Emeleus,1997], for example,
are generallylow-energydepositsassociatedwith sluggish
streams and shallow freshwater lakes. Terrestrial
sedimentscontaininglignite, leaf-bedsand sandstones
are
occasionallyinterbeddedwith lavas on Skye, Mull and
elsewhere[Richey, 1935]. The slightlyyoungersediments
associatedwith the Canna Lava Formationand the Skye
Main Lava Series are high energy conglomeratesand
sedimentarybreccias,possiblyrelatedto the development
of the Rum CentralComplex,a substantialvolcanicedifice
at that time. On Mull there is a mudstone at the base of the
lava pile, which may representa lateritisedtuff [Bailey et
al., 1924], and the basal Tertiary sedimentscontaina few
metresof sandstonewith grainsof aeolianorigin [Bailey,
1924].
The diverse substrate is reflected by the style of
eruption of the earliest magmas. Most of the British
Tertiary IgneousProvincelavas were eruptedsubaerially
on dry land, but local occurrencesof pillow lavas, vitric
tuffs and hyaloclastites(for example, at the base of the
Skye Main Lava Series [Andersonand Dunham, 1966])
are consistentwith emplacementinto shallowwater. There
is no clear indicationfrom the terrestrialdepositswhether
stronguplift began significantlyprior to basalteruption.
Palaeogene uplift of the order of 300 to 1000 m is
recordedby apatite fission tracks in sedimentsfrom the
East Irish Sea Basin [Hardman et al., 1993]. This uplift
appearsto have been accompanied by a heatingeventand
emplacementof the FleetwoodDyke Group.The latter is a
suite of doleritesthat form part of the Irish Sea dyke
swarm[Kirton and Donato, 1985], andwhichhaveyielded
K-Ar dates of 65.5+1.0 and 61.5+0.8 Ma [Arter and
Fagin, 1993]. Post-Cretaceous uplift andtilting of western
Scotlandis recordedin the stratigraphy of the Inner Moray
Firth Basin in easternScotland[Underhill, 1991] and this
couldbe linked with the developmentof the proto-Iceland
thermal anomaly and associatedrifting in the west
[Thomsonand Underhill, 1993]. Thermal uplift may also
be responsiblefor other uplift events(for example,the
uplift and denudationof up to 3 km of sedimentsfrom
partsof northernBritain [Lewiset al., 1992]).
4.2.1. Skye. The Skye lava successioncoversan areaof
some 1500 km2, has a cumulativethicknessof-1200 m
[England, 1994], and has been divided into three magma
types:the Skye Main Lava Series,the Fairy Bridgemagma
type, and the PreshalMore magma type. The Skye Main
Lava Series is composedpredominantlyof transitional
alkali basalts,mostlywith Lan/Ndn-< 1 andNdn/Ybn = 3-5
[Thompsonet al., 1972, 1980] (Figure 8), althoughsome
20% of the lava pile contains intercalatedhawaiites-
mugearites-benmoreites-trachytes [Andersonand Dunham,
1966; England, 1994]. (Note that Lan/Ndn refers to
chondrite-normalised La/Nd.) The Fairy Bridge magma
type is found occasionallywithin the upper half of the
Skye lava pile and comprisesbasalts with flat REE
patternsbut otherwisehasmajor elementchemistrysimilar
to the Skye Main Lava Series [Thompsonet al., 1980;
Scarrow, 1992]. This type is also well representedin the
Skye dyke swarm [Mattey et al., 1977]. The PreshalMore
magma type is representedby severalbasalticflows near
the top of the Skye lava succession, by occasionaldykes,
and by some intrusionsin the Cuillin intrusivecomplex
[Thompson,1982]. The magma type is characterised by a
tholeiitic major element chemistry with light-REE-
depletedpatterns[Essonet al., 1975; Mattey et al., 1977;
Thompsonet al., 1980] (Figure 8).
4.2.2. Mull. The Mull lava succession covers an area of
840 km2 onthe Islandof Mull andthe adjoiningmainland
area of Morvern. Bailey et al. [1924] and Emeleus[1991]
estimatedthat the lava successionhas an approximate
aggregate thickness of 1800 m. Kerr [1994, 1995a]
identifiedthree magma types within the Mull succession.
These three magma types are broadly similar to those
identifiedon Skye, the differencebeing that on Mull a
successive relationshipbetweenthe threetypescanbe seen
clearly. The earliest lavas, the Mull Plateau Group, are
transitionaltholeiitic-alkalicpicritic-basalts
(up to 15 wt.%
MgO) to hawaiiteswith similar REE patternsto the Skye
Main Lava Series (Figure 8). Trachytesand benmoreites
near the top of the Mull Plateau Group accountfor less
than 5% of the total preservedlava volume. The Coire
Gorm magmatype which overliesthe trachytesof the Mull
Plateau Group comprises transitional basalts with
chondritic REE profiles, like the Skye Fairy Bridge
magma type. The Central Mull tholeiitesare the youngest
lava type found on Mull; they possessflat to LREE-
depletedREE patternsand are compositionallysimilar to
the Preshal More lavas from Skye (Figure 8). Like the
equivalentbasaltsfrom the Skye Main Lava Series,some
of the Mull Plateau Basalts have characteristic trace
elementsignatures,includinghigh Ba/Rb and Ba/Nb, and
low end values(Figures 5 and 8), consistentwith
 SAUNDERS ET AL. 59

100 _
_

Mull Basalts Mull Basalts

50•

o 10--
--

c'• -
F: 5--
03 - Central Mull
Tholeiites

,• []•Nd(t
)= +7.5 Tholelites
o o •Nd(t
)=-9.3
1 I I I I I I I I I I I I I I I

Skye Basalts Skye Basalts

Skye Main Lava
• Skye
Ma.
inLava

.• 10--
• -
E -
03 5
rj') - Preshal More
--

Tholelites
,r-'- Preshal More
o o •Nd(t
)=+8.6 Tholelites
[] []•Nd(t
)--7.0
I I I I I SmEu
La Ce Pr NdI I Gd
I T'DDy
I YI ErI TmYb
I I LuI RbBaTh Nb Ta K LaCe Sr Nd P Sm Zr Hf Eu Ti Tb Y Yb Lu

Figure 8. Chondrite-normalisedREE andbulk silicateEarth-normalised

traceelementpatternsfor basaltsfrom the
twomainlavasuccessions onMull andSkye,Scotland. Datasources: Mull - Kerr [1995a],Kerret al. [1995a];Skye-
Thompson et aL [1972,1980,1982,1984].CentralMull tholeiiteandPreshalMorebasaltpatterns areaverage values;
Mull PlateauGroupbasalts arerepresented
by samplesBR24andBR8;andSkyeMainLavaSeriesis represented by
samplesSK906 and SK976. end valuescalculatedat t=60 Ma.

lationof ArchaeanLewisiangranulites(seeSection6). 4.2.4. Northern Ireland. The Antrim Lavasrepresentthe

4.2.3. TheSmall Isles.The thin lava sequences
foundon
Eigg, Muck, Rum, and Canna are composedof two
formations,the Eigg Lava andthe CannaLava Formations
[Emeleus, 1997]. The Eigg Lava Formation, erupted
before the emplacementof the Rum Central Complex,
consistsof transitional olivine basalts and, low in the
succession,several flows of mugearite.Feldspar-phyric
basaltichawaiitesoccur near the top. The Canna Lava
Formationwas eruptedafter the Rum Central Complex
andcomprises four memberswith rocktypesrangingfrom
olivine basalt to hawaiite and icelandite.
chemistryof the lavas are providedby Emeleus[1985,
1997]. The most dramaticmanifestationof the Tertiary
activity in this area is, however, the Rum Central
Complex,the remnantsof a cyclicallylayeredmagma
chamberoriginallyderivedfrom a picritic magma[e.g.,
Emeleus,1987; Younget al., 1988].
largestremnantof the British Tertiary IgneousProvince,
coveran areaof 3500 km2, andachievea thickness
of up
to 800 m. The lava successionconsistsof three formations,
the Lower, Middle and Upper Formations[Old, 1975;
Wilson and Manning, 1978]. The Middle Formation
includes the basalts of the Giant's Causeway (the
CausewayMember). The basaltsof the Upper and Lower
Formations, although transitional basalts, are more
tholeiitic than the basaltsof the Mull PlateauGroup and
the Skye Main Lava Series[Lyle, 1980; 1985]. Basaltsof
Details of the the Lower Formation, and most of those of the Upper
Formation, are light-REE-enrichedwith convex-upwards
patterns [Barrat and Nesbitt, 1996]. The basalts of the
CausewayMember show great variation in the degreeof
light-REE enrichment(Lan/Ybnrangesfrom <0.6 to >3)
and in isotoperatios (e.g., gNdt-60 ranges from -11 to
+8.5), probably caused by assimilation of
60 THE NORTH ATLANTIC IGNEOUS PROVINCE
amounts of Dalradian crust of Late Proterozoic-Lower
Palaeozoicage [Wallace et al., 1994; Barrat and Nesbitt,
1996]. The light-REE-depletedbasaltsfrom the Causeway
Member resemble the Preshal More and Central Mull
Tholeiites.
4.2.5. Age of the British Tertiary IgneousProvince.For
detailedaccountsof the age of the BritishTertiary Igneous
Province, the reader is referred to Mussettet al. [1988],
Dickin [1988] and Ritchie and Hitchen [1996]. The oldest
agesof any reliability for the Britishprovinceare 63.0+3.4
and 63.3+1.8 Ma (4øAr-39Arages) for reversely
magnetisedbasalticlavas from Eigg and Muck [Dagley
and Mussett, 1986] (Figure 2). These ageshave recently
beenconfirmed
andrefinedby step-heating
4øAr-39Ar
ages
(62.8+0.6 and 62.4+0.6 Ma) on sanidinesfrom tuffs
intercalatedwith the Eigg Lava Formation[Pearsonet al.,
1996] Theseare amongthe oldestagesfor the entireNorth
Atlantic IgneousProvince.A K-Ar age of 81.+4 Ma for a
doleritesill from the Faeroe-Shetland IntrusiveComplex
(Well 219/28-2 [Fitch et al., 1988]) requiresconfirmation
by 4øAr-39Ar
methods;the bulk of the complexgives
cooling ages of 55-53 Ma [Hitchen and Richie, 1993].
Lavas with reversed magnetic polarity from the older
PlateauGroupon Mull give an 4øAr-39Ar
ageof 60+0.5
Ma [Mussett,1986]; the Centre3 Graniteon Mull givesan
Rb-Sr age of 58.2+1.3 Ma [Walsh et al., 1979]; and the
centres on Ardnamurchan,which again have reversed
polarity, give an Rb-Sr age of 60.0ñ 1.7 Ma [Walshet al.,
1979]. The Western Granophyre on Rum (reversed
polarity) has an Rb-Sr age of 59.8+0.4 Ma [Mussett,
1984]. No reliable ages are available for the lavas from
Skye or Arran, althoughit is possiblethat the Main Series
lavas from Skye are coeval with the Canna Lava
Formation on Rum (61.4+0.4 Ma [Mussett,1984; Mussett
et al., 1988; Bell and Williamson, 1994]. Note, however,
that pebbles of the Rum granophyre occur in
conglomeratesbetweenthe lavas on Canna, and the lavas
sit unconformably on the granophyre [Black, 1952],
implyingthat eitherthe 61.4 Ma age for the basaltis too
old, or the 59.8 Ma agefor the granophyreis tooyoung.
Most of the Palaeogeneactivity in Ireland occurred
duringa reversedpolarity event.Dingle Dyke (ca. 59 Ma),
Blind Rock Dyke (61.7ñ0.5 Ma), and the Carlingford
Granophyre(60.9-+0.5 Ma) are among the oldest dated
events[Thompson,1986]. Few reliable agesare available
for the Antrim Lavas (58.3-+1.1 to 61.0+0.6 Ma
[Thompson,1986; Wallaceet al., 1994]).
The majority of the datesare thereforeconsistentwith
emplacementof magmas during C26r (57.95-60.9 Ma,
accordingto the time scale of Berggrenet al. [1995]),
althoughit is possiblethat someactivity (e.g., on Muck
and Eigg) occurred during C27r. Not all of the dated
materialfalls into this neatpattern,however.The Northern
Granite of Arran, for example',has an Rb-Sr age of
60.3.+0.8Ma (and a virtually identical4øAr-39Ar
age)
[Evans et al., 1973; Dickin et al., 1981] but shows a
normal magnetic polarity. Coire Uaigneich Granite on
Skye (59.3.+0.4 Ma) is also magneticallynormal. It is
unclearif thesebodieswere emplacedduringC27n or 26n.
In some areas, activity continuedfor severalmillion
years after the main event; for example, Beinn an
DubhaichGranite,Skye (53.5.+0.4Ma [Dickin, 1981]), the
Loch Ba Felsite, Mull (56.5-+1 to 58.2-+1.3Ma [Mussett,
1986; Walshet al., 1979]), the Sgurr of Eigg pitchstone
obsidian(52.1-+0.5Ma [Dickin and dones,1983]), andthe
later granites in the Mourne Mountains (53.3-+0.6Ma
[Thompsonet al., 1987]). The Lundy Granite, the most
southerly known part of the North Atlantic Igneous
Province,has an Rb-Sr age of 54+4 Ma [Hamptonand
Taylor, 1983]. It is intruded by dykes that are
predominantlymagneticallyreversed,and that havebeen
datedby 4øAr-39Ar
at 56.4-+0.3Ma [Mussett
et al., 1976].
4.3. Ofj•hore UK
An importantrecordof widespreadPalaeogene volcanic
and intrusiveactivity is preservedon the continentalshelf
and basinsaroundthe British Isles. Indeed, the volume of
material far exceedsthat preservedon the mainland,and
the extensive commercial exploration that has been
underwayfor the lastthreedecadeshasensuredreasonable
samplingdensityand seismiccorrelations.A review of the
igneousactivityto the northwestof the UK is providedby
Ritchie and Hitchen [ 1996], who allocatedthe activityto
sevenmajor categories,basedon location,age, structure
and genetic relationship:(i) the Faeroe Plateau Lava
Group (FPLG, see Section 4.7), (ii) the North Rockall
Trough - Hebrides Lavas Group; (iii) central igneous
complexes,suchas St Kilda, Rockall and Erlend;(iv) the
Faeroe-ShetlandIntrusive Complex; (v) volcaniclastic
deposits;(vi) the Minch region; and (vii) the Wyville-
ThomsonRidge.
The FPLG (see Section 4.7) and the North Rockall
Trough - HebridesLavas Group are both part of a much
more widespread,in partsdiscontinuous, subcropof lavas
and sills that extend from the southern end of the Rockall
Plateauto the Voring Plateau(Figure 1). As discussedin
the previous section, the age of the FPLG is poorly
constrainedbut it was probably emplacedduringC26r-n
and C24r times.On the basisof K-Ar and biostratigraphic
ages,the North Rockall Trough - HebridesLavas Group
activity spans63 to 50 Ma [Ritchie and Hitchen,

but this may narrowwhen 4øAr-39Ar
measurements
are
made. Most of the activity associatedwith the Faeroe-
ShetlandIntrusiveComplex,a belt of intrusiverocksthat
coversan area of approximately
40,000 km2, occurred
between 55 and 53 Ma [Hitchen and Ritchie, 1987; Fitch
et al., 1988; Ritchie and Hitchen, 1996]. Earlier activity
recordedin this complex, such as the 80-m.y.-old basalt
sill recoveredfrom Well 219/28-2 [Fitch et al., 1988] may
be related to the Cretaceousmagmatismin the Rockall
Trough(e.g., RosemaryBank [Hitchenand Ritchie, 1993;
Morton et al., 1995] and Anton Dohrn Seamount[Joneset
al., 1994]).
Volcaniclastic deposits are common in offshore
boreholes,and provide importantcon:elationhorizonsin
the North Sea and to the northwest of Britain. Because
they canoftenbe datedbiostratigraphically, theyprovidea
usefulmeansof datingmajor volcanicepisodes.Knox and
Morton [1983, 1988] showedthat there were two distinct
phasesof volcaniclasticsedimentationin the North Sea
Basin, the first during nannofossilzonesNP5-6 (60-57.5
Ma on the Berggrenet al. [1995] time scale),the second
and mostvoluminousduringzonesNP9 to 13 (56-50 Ma)
(Figure2). Basic,acid, tholeiiticand alkalinevarietiesare
found. The most extensive horizons occur in the Balder
Formation(NP10), which are predominantlyderivedfrom
Fe-Ti-rich basalt precursors[Knox and Morton, 1988;
Morton and Evans, 1987]. Eruptive (source)centreswere
broadlyto the west of the BritishIslesand in the Faeroe-
Greenlandregion during the first phaseof activity, but
appearto have been restrictedto the Faeroe-Greenland
areaduringthe secondphase[KnoxandMorton, 1988].
4.4. WestGreenlandand Baffin Island
Basalts and picrites of Palaeocene age crop out
extensivelyin centralWestGreenlandandin the regionof
CapeDyer on Baffin Island [Clarke and Pedersen,1976;
Clarke, 1977; Larsen et al., 1992]. The two successions
were probably contemporaneous and contiguous.The
West Greenland lavas cover an area of approximately
55,000km2 [ClarkeandPedersen,1976],andthe lavapile
may exceed 5 km in thicknesson Ubekendt Ejland,
althoughfaulting makes accuratedeterminationof the
thicknessdifficult [Larsen, 1977]. There are insufficient
thicknessand age data to make an accurateassessment of
their volumeor their eruptionrates.
The West Greenland lavas have been divided into three
lithostratigraphical
units by Hald and Pedersen[1975].
These are, from old to young: the Vaigat Formation,
consistingof lavas and hyaloclastites,mostly of picritic
composition; the Maligfit Formation,which is dominated
SAUNDERS ET AL. 61
by feldspar-phyrictholeiitic basalts; and the Hareoen
Formation, consistingmostly of olivine-phyric basalts.
Most of the lavas were erupted subaerially, but the
presenceof hyaloclastites,brecciasand marine mudstones
in the Vaigat Formation implies eruption into shallow
seawater during the earliest stages of activity. The
outcropsat Cape Dyer and along the Baffin coastto the
north are generally much thinner than those in West
Greenland.
Beckinsaleet al. [1974] publisheda Rb-Sr isochronof
67+5 Ma for a small intrusionemplacedinto picrites at
Ubekendt Ejland. Parrott [1976, in Clarke et al., 1983]
arguedthat the bulk of the activity on Ubekendt Ejland
occurredbetween60 and 56 Ma, on the basisof 4øAr-39Ar
dates.The bulk of the West Greenlandlavasare reversely
magnetised [e.g., Larsen et al., 1992], apart from a
sequenceof normal polarity lavasin the lower part of the
Vaigat Formation which Athavale and $harma [1975]
tentatively correlated with C25n. On the basis of
palynological data [Piasecki et al., 1992], however, the
bulk of the Vaigat and Maligfit Formationsappearto have
been erupted during C26r, suggestingthat the normal
event in the Vaigat Formationis C27n. In earlier studies,
dinoflagellatesrecovered from mudstone on Nuussuaq
suggestedthat the earliest lavas of the Vaigat Formation
correspondto nannoplanktonzone NP3 [Jurgensenand
Mikkelsen, 1974; Larsen et al., 1992], althoughit is now
thoughtthat this assignmentmay be too old (L.M. Larsen,
pers.comm.).Thishasbeenconfirmed
by recent4øAr-39Ar
dates of 60-60.5 Ma for the oldest West Greenland basalts
[Storey et al., 1996]. On the basis of the latest age
determinations and palynologicaldata,therefore,the main
phase of magmatismin West Greenlandwas essentially
contemporaneous with the earliest activity of the British
Tertiary IgneousProvinceand SE Greenland(Figure2).
Seafloorspreadingat a transectat 57-62øN, 600 km to
the southeastof Disko, probablybeganduringanomaly27
time [Chalmers, 1991], although Roest and Srivastava
[1989] have argued for earlier initiation of seafloor
spreading (anomaly 33: Late Cretaceous). Detailed
information about the structure of the seafloor between
Cape Dyer and Disko is not available to the authors.
Nonetheless,it appears,on the basisof the interpretation
of Chalmers [1991], that the main pulse of flood basalt
magmatismin the Disko areaand the initiationof seafloor
spreading at 57-62øN were approximately contempo-
raneous.This is an importantpoint,becauseif it couldbe
shownthat the onshoremagmatismpre-datedthe seafloor
spreading,then it would demonstratethat the thermal
anomaly was a pre-existingfeature. Conversely,if the
flood basalts and picrites substantiallypostdate
62 THE NORTH ATLANTIC IGNEOUS PROVINCE
seafloor spreading,it would indicate the arrival of the
thermalanomalybeneaththisregion.
Onset of seafloor spreadingin the Palaeoceneis also
suggested by the sedimentson the Labradormarginand in
the Disko-Nuussuaqarea [Chalmers,1991]. No significant
unconformity is seen in sedimentsof Campanian age
(C33), as would be expected if rifting and seafloor
spreadinghad occurredat this time. There is, however,a
major hiatus in the Danian in the Labrador margin
sequences [Balkwill, 1987] and at the top of the
Cretaceousin the Disko-Nuussuaqarea [Hendersonet al.,
1976], consistentwith footwall uplift and possibly the
effects of a mantle thermal anomaly. The main eruptive
centresappear to have been seawardof both the Baffin
Island and West Greenland successions[Upton, 1988],
implying that what is now a grabenstructurein the Davis
Strait was a structuralhigh duringthe DanJan.Southeast-
directedsyn- and post-volcanictilting of Disko, and lava
thicknessesthat indicate flow from the west, also indicate
syn-magmaticuplift in the region of Davis Strait [Larsen
and Pedersen, 1990, 1992].
An unusual feature of the West Greenland and Baffin
Island lavasis the high proportionof picrites,between30
and 50% of the total lava pile [e.g., Clarke, 1970; Clarke
and Upton, 1971; Clarke and Pealersen,1976; Francis,
1985; Pedersen, 1985; Holm et al., 1993; Gill et al., 1992;
Larsen et al., 1992], which is substantiallygreaterthan in
either East Greenland (-15% of the Lower Basalts; see
below) or the British Tertiary Igneous Province. The
compositionof the parental liquids responsiblefor the
picriteshas been the subjectof considerabledebate;were
they primary high-MgO liquids[e.g., Clarke, 1970; Clarke
and O'Hara, 1979], or did the liquids undergo olivine
accumulation[Hart and Davis, 1978]? Analysed olivines
in lavasfrom the Vaigat Formationhave forsteritecontents
in excess of Fo92 [Pedersen, 1985] which implies
equilibriumliquid MgO contentsof about 19%. Gill et al.
[1992] used this figure to estimatepotentialtemperatures
in the mantle sourcebetween 1540 and 1600øC,assuming
a depth of melt segregationequivalent to 2.0 GPa, and
anhydrous melting. This implies an excess mantle
temperatureof between 240 and 300øC, a surprisingly
high figure given the distalnatureof theselavasin relation
to the proposedplume axis [Gill et al., 1992; Chalmerset
al., 1995] (see Section7).
The lavas on Baffin Island are almost exclusively
picrites or olivine tholeiites [Francis, 1985; Robillard et
al., 1992]. Robillard et al. [1992] identified two
compositionaltypesthat are stratigraphicallyinterbedded:
(i) a light-REE-depletedsuite, with La,/Sm,-0.6-0.7 and
875r/86Srpresent
day(pd)
(unleached)
0.7031-0.7032
and(ii) a
suitethat showsslight light-REE enrichment(Lan/Smn-1-
1.2) and slightly more radiogenic Sr isotope ratios
(0.7032-0.7039). They comparedthese two suites with
depleted,normal (N) -MORB, and enriched(E) -MORB,
respectively,and argued that they are derived from two
distinctmantle sourceson the peripheryof the ancestral
Iceland plume. Picrites from the lowermostparts of the
successionson Disko overlap with present-dayNorth
Atlantic
MORB(87Sr/86Sr(pd)
0.7030-0.7036
andeNdt=60
+7.3 to +10.1), although most other picrites from
Ubekendt Ejland and Svartenhuk Halvo overlap with
MORB and with basaltsfrom Iceland (eNdt:60 >+3.4 and
87Sr/86Sr(pd)
>0.7031)[Holmetal., 1993].
The West Greenland and Baffin picrites and olivine
basaltsprovide evidenceof rapid ascentof magma, with
minimal interactionwith the crustor storagein long-lived
magma bodies [e.g., Upton, 1988]. Nonetheless,
contaminationof picritic and basalticmagmaby shaleand
sandstonein assimilation-fractionalcrystallisation(AFC)-
type processeswas demonstratedby Pedersen and
Pedersen[1987], who analyseda rangeof lava typesfrom
the Vaigat and Maligfit Formations.Rhyolites from the
successionsappear to represent anatectic crustal melts
rather than the products of fractional crystallisation.
Igneousactivity continuedin the WestGreenlandarea,but
at decliningrates,throughthe Palaeogene.Alkaline lavas
of the Erqufi Formation and a suite of lamprophyresof
Oligoceneagewere emplacedon UbekendtEjland [Parrot
and Reynolds,1975; Larsen, 1977].
4.5. Central East Greenland
The on-land portion of the East Greenlandmagmatic
provincestretchesfrom Kap GustavHolm in the southto
ShannonIsland in the north, a distanceof some 1200 km.
Figure 1 showsthe locationsof the main outcrops.The
mostnortherlyoutcrops,aroundWollastonForland,Hold
with Hope, and Shannon Island, will be describedin
subsection 4.6. Wager [1934], Brooks [1973a], Deer
[1976], Noe-Nygaard [1974, 1976], Upton [1988], and
Larsenet al. [1989] providedkey descriptions.
The province is dominated volumetrically by basalt
lavaswhich comprisethe spectacularlandscapein central
East Greenland, along the Biosseville Kyst between
Kangerlussuaqand Scoresby Sund, where the lava
sequences may be as much as 7 km thick and individual
flowsmay havevolumesof up to 300 km3 [Nielsenand
Brooks, 1981; Larsen et al., 1989]. The total volume of
extrusive material preservedin central East Greenland
between Kangerlussuaq and Scoresby Sund is
approximately160,000km3 [Nielsenand Brooks,

Larsen et al., 1989]. A further 10,000 km3 may have
coveredJamesonLand and areasto the north, and up to
60,000km3 of basaltmayhaveremainedon the conjugate
plate boundary following plate separation.This gives a
total volumeof approximately
230,000 km3. In addition,
the region is characterisedby later intrusive centres,the
most famous of which is the gabbro-granophyre
Skaergaard intrusion [Wager and Deer, 1939], and
extensivedyke swarmsalongthe coastalmargin.
The basalts of central East Greenland have been divided
into two series, the Lower Basalts and the Main Series,
comprisinga total of 10 formations,some of which are
tentatively assignedas lateral equivalents[Larsen et al.,
1989]. The oldest preservedpart of the sequence,the
Lower Lavas or Lower Basalts, is exposed near
KangerlussuaqFjord, at the southernend of the main
outcrop.The Lower Basaltshave an estimatedthicknessof
1.5 km and approximately15% of the lavasare high-MgO
basalts or picrites [Nielsen et al., 1981; Brooks and
Nielsen, 1982a,b; Fram and Lesher, 1996]. The basalts
have undergone considerable secondary alteration, in
places to greenschist grade, making direct age
determinationdifficult. The Main Series (or Plateau)
Basalts,exposedalong the BiossevilleKyst and around
ScoresbySund, are predominantlytholeiites. Although
Mg-rich varietiesoccur,none are aspicritic asthosefound
in the Lower Basalts.High-Si varietiesof basaltoccurnear
the bottom of the sequenceand in at least one flow near
the top of the Main Series in the SkraentemeFormation
[e.g.,Larsen et al., 1989].
Where the base of the lava pile is exposed,the basalts
sit either on Palaeogene sediments or lap onto the
Precambrian basement in the west and north. Most of the
lavaswere eruptedsubaerially,throughfissures,although
some of the earliest flows, those in the Vandsfaldsdalen
Formation,were eruptedin a shallowmarine environment
[Soper et at., 1976b], resulting in thick hyaloclastite
deposits.Larsen and Watt [1985] believed that Mesozoic
sedimentsunderliethe entire easternhalf of the lava pile.
Facieschangesindicatea shallowingof the marine basin
in the Danian [Soper et at., 1976b]. This is succeededby
an unconformity passing up into coarse sands and
volcanogenic sediments of the basal part of the
Vandsfaldsdalen Formation. There is no indication in the
sedimentsof strong uplift in this region, or in adjacent
sourceareas,prior to Daniantimes(pre-65 Ma).
There is uncertaintyaboutthe age of the basaltsfrom
centralEast Greenland.The main constraintsare provided
by microfossilsand palaeomagnetism, neither of which
provide an absoluteage. All of the basalts,includingthe
Lower Basalts,were eruptedduring a period of reversed
SAUNDERS ET AL. 63
polarity [e.g., Tarling, 1967; Soper et al., 1976b]. A
marine dinofiagellate (,,tpectodiniumhomomorphum)is
found in shalesinterbeddedwith hyaloclastites at the base
of the Lower Basaltsin the Kangerlussuaq region[Soperet
al., 1976a,b]. Previous studieshave suggestedthat the
dinofiagellatehas a range from mid-Thanetianto early
Bartonian(ca. 56-40 Ma on the time scaleof Berggrenet
at. [1995]) and that its base correspondsto the base of
nannoplankton zone NP9 (the so-called 'Base
ApectodiniumDatum') [Powell, 1988], implying that the
basaltswere erupted during C24r [e.g., Berggren et at.,
1985]. However, this biostratigraphiccontrol should be
used with caution (D. Jolley, pers. comm.). Several
occurrencesof Apectodiniumspp. have been reportedat
stratigraphic levels below the 'Base Apectodinium
Datum', and/t. homomorphumoccursin strataof Danian
age in the Maureen Formation of the central North Sea
[Thomas, 1996]. Given these uncertainties, the Lower
Basaltsmay have been eruptedduring either C26r, C25r,
or C24r. Noble et at. [ 1988] suggested
that the Main Series
basaltsalong the BiossevilleKyst were eruptedbetween
53 and 57 Ma (K-Ar dates).Hansen et at. [1989] obtained
an incremental
heating4øAr-39Ar
ageof 56.7+4.3Ma for a
basaltfrom the Main Seriesin the ScoresbySundarea,and
Storeyet at. [1996] suggested, on the basisof 4øAr-39Ar
data, that the basaltsof the Lower Seriesmay be 59-60
m.y. old.
The basalts from the Scoresby Sund region are
predominantly tholeiites that underwent extensive
differentiationin mid- to upper-crustalmagmachambers,
resultingin the productionof high-TiO2 ferrobasaltsor
titano-tholeiites.
The depthof fractionationis indicatedby
the displacementof the basaltsto the low-clinopyroxene
sideof the 1-atmosphere cotecticon the normativeot-di-hy
triangle of Thompson[ 1982], consistentwith fractionation
at about 0.35 GPa (11 km) [Larsen et al., 1989]. That
these magma chamberswere open is indicatedby the
cyclicaleruptionof high Ti, Fe tholeiitesand, indeed,by
the large volumesof individualflows [Hogg et al., 1989;
Brookset at., 1991]. Evacuationof a realisticportionof a
magma body (•1% [O'Hara and Mathews, 1981]) to
produceflows with a typical volumeof 20 to 60 km3
requiresa chamberof up to 12,000 km3 [Larsenet al.,
1989]. As pointed out by Larsen et at. [1989], a sill-like
elliptical chamberwith dimensions150 x 30 x 5 km would
fit easily within the area of the postulatedfeeder dyke
swarms(-200 x 30 km).
There are few publishedisotopeor comprehensive trace
element data for the basalts from central East Greenland.
The bulk of the data are for the Lower Basalts at
Kangerlussuaq[Holm, !988; Gill et at., 1988]. Larsen
64 THE NORTH ATLANTIC IGNEOUS PROVINCE

1 oo
al. [1989] published large amountsof major and some
trace element data for the Main Series basalts, and Holm A ]•_ Central East
[1988] included isotopic and trace element analysesfor • $0 Greenland
eight Main Series basalts. The Flado Dykes, emplaced
immediatelyto the southof Kangerlussuaq, are considered
to be the hypabyssalequivalentsof the Lower Basalts,and
at the time of writing they providethe mostcompletetrace • •o
element data set for the central East Greenland basalts
[Gill et al., 1988].
Values
of87Sr/86Sr(pd)
andgNd(pd)
range
from0.7032to
0.7094 and +7.8 to -5.8, respectively,for the entirecentral
'•o_ :' ":::
East Greenlandbasaltprovince [Carter et al., 1979; Holm, !::.:.:.:.:.
Basalts.:.:.:.:.::'::i:' '"'
1988; Larsen et al., 1989]. The lowest 87Sr/86Sr
and
highest gNd values overlap those of present-dayratios RbBaThNbTa K LaCeSrNd PSmZrHfEuTi b YYbLu
eruptedin the active rift zonesof lceland(e.g., Figure 7);
most of the ScoresbySund basaltsfall within the range o Prince of Wales Mts
0.7034-0.7038. Therefore, the bulk of the basalts have ß Flado Dyke (GGU267901B), Pic-Ank. Series
isotopic signatures suggesting derivation from a [] Flado Dyke (GGU267906), TholeiiteSeries
sublithosphericsource,althoughthe (limited) Pb isotope ß Flado Dyke (GGU267909), TholeiiteSeries
ß Lower Series Basalt (GM20332)
data suggest some contaminationof the magmas with
unradiogenicPb [Holm, 1988]. The greatestisotopicrange
is in the Vandsfaldsdalen Formation in the Lower Basalts, Figure 9. Chondrite-normalisedREE and bulk silicate Earth-
but high 87Sr/86Sr,
high-SiO2basaltsare foundelsewhere normalisedtraceelementpatternsfor basaltsfrom Kangerlussuaq
in the succession;for example, in the Skraenterne (Flado Dykes and Lower SeriesBasalts) and Prince of Wales
Formation [Larsen et al., 1989]. Similar, high-SiO2 and Mountains(averagevalue). Data sources:Kangerlussuaq - Gill et
al. [ 1988]; Princeof Wales Mountains- Hogg et al. [ 1989]. Field
high 87Sr/86Sr
basaltsare found in West Greenland
for Icelandtakenfrom Figure4.
[Pedersen, 1985; Pedersen and Pedersen, 1987] and the
Faeroes[Hald and Waagstein,1983; Gari•py et al., 1983].
The Lower Basalts include picrites and ankaramites, parameterssuch as La/Yb and Zr/Nb, Holm [1988] also
equivalentsof which are also found in the Flado dykes recognised two groups in the Lower Basalts, which
[Holm, 1988; Gill et al., 1988]. Isotopedata are available broadly correspond to this bipartite division into
foronlyoneankaramite
(87Sr/86Sr(pd)
= 0.7064)andthree 'enriched'and 'depleted'types.Both groupsshowa range
picrites(0.7031-0.7044), which have valuesoverlapping of isotopevalues(e.g., end: Figure 7), whichHolm [1988]
thoseof lower-Mg lavasin the Lower Basalts. attributed to mixing between an Icelandic-typemantle
Many of the picrites and ankaramitesof the Lower sourceand old continentallithosphericmantle. Although
Basalts and Flado Dykes have high abundancesof the there can be little doubt that crustal contamination was
highly incompatibleelements(e.g., Rb to Ce on Figure9), involved in producingthe high 87Sr/86Sr,
high-SiO2
and steepREE profiles [Gill et al., 1988]. Sometholeiites characteristicsof some basalts from East Greenland, we
of the Lower Basalts sharethis characteristic,but tend to believe that the evidence for the involvement of
have lower La•/Yb, ratios.The majorityof the tholelitesof continentallithosphericmantle is far from clear-cut,and
the Lower Basaltshave flatter REE patternsand show returnto this point in Section6.
relative depletion of the highly incompatibleelements
(e.g., samplesGGU267909 and GM20332 on Figure 9). 4.6. Northeast Greenland
The limited trace element data for the Main Series Basalts
[Holm, 1988; Larsen et al., 1989] suggestthat they share Scatteredoutcropsof Tertiary basaltscropout alongthe
the characteristics
of the 'depleted'tholeiitesof the Lower East Greenland margin between 72ø and 76øN. In the
Basalts(for example, similar Zr/Y ratios and similar trace Gauss-Halvo- Hold with Hope region,the lava succession
element patternsfrom Rb to Ti), althoughthe absolute is -800 m thick [Upton et al., 1980, 1995] and has been
abundancesof incompatible elements may be high, dividedinto LowerandUpperSeries.A recent4øAr-39Ar
especiallyin the evolved ferrobasaltsand titano-tholeiites date on a basalnephelinitefrom Hold with Hope gives an
from ScoresbySund [Larsenet al., 1989]. On the basisof eruptionage of 58.7 + 1.4 Ma [Upton et al., 1995].
 SAUNDERS ET AL. 65

lOO
dykeswhichpostdatethe Upper Seriesgive agesof 56.7 +
0.7 and 56.6 + 1.9 Ma, respectively.Upton et al. [1995]
50-- • Wollaston
.Forland
argued that these dates, combined with the available
palaeomagnetic evidence,indicatethat the Lower Series
and the earliestpart of the Upper Seriesbelongto C24r,
with succeeding magnetically normal polarity lavas
belongingto Subchron24n.3. However, it is more likely 10'•
that eruptionoccurredduringC26r-26nor 25r-25n on the
basisof the revisedtime scaleof Berggrenet al. [ 1995]. A o3
o3
5
o Basalt(228237),Woll.Forland(œNd t = +8.2)
latesheetfromthe MyggbuktaComplexgives4øAr-39Ar
ß Basalt(228230),H w H, LowerSeries(œNd t = +7.6)
age of 32.7 + 2.9 Ma [Upton et al., 1995]. The causesof
'" Basalticandesite(227913),H w H, L. Series(œNd t = -4.4)
this late magmatic event are unclear, but Upton et al. ß Basalt(227965),H w H UpperSeries(œNd t = -2.7)
[1995] related it to minor tectonicadjustmentsalong the
1
continentalmargin. I I I I I Sm
La Ce Pr NdI EuI Gd
I ; b DyI HoI ErI Tm
i YbI LuI
The Lower Series basalts have been correlated with lOO --

lavasfrom the upperpart of the succession recoveredfrom --

Wollaston Forland
Hole 642E on the Voring Plateau(see below). They are 50--
1• -
and Hold with Hope
mildly light-REE-enrichedquartz tholeiites (Lan/Ybn=2 t't:i _

[Thirlwall et al., 1994], Figure 10), with Nd, Sr and Pb

isotopicratios similar to Icelandictholelites.Pb isotopes
suggestthat the basalts of the Lower Series were not
.• 10•
contaminatedby continentalcrust. The lavas may have
beeneruptedin an oceanicsetting,a suggestion supported • 5--
by the moderateLan/Ybnratiosthat imply shallowmelting ID_
E
_

and melt segregation[Thirlwall et al., 1994]. The Upper CO -

Series lavas, on the other hand, are regardedas higher-

pressuremelts of a similar source and have variable
amounts of contamination. Like the lavas from
ScoresbySund-Kangerlussuaq region,the basalunitswere
erupted into shallow water, but the remainderof the
basalts were subaerial.
Two periods of Tertiary magmatismhave also been
identifiedat Trail113,approximately150 km southof Hold
with Hope [Price et al., 1996]. The first comprises
a series
of tholeiitic sills, which have an emplacementage of
approximately
54 Ma (4øAr-39Ar
stepheating).Thesecond
comprisessmallervolumesof alkalicdykesandtwo large
syenitecomplexes,emplacedat about 36 Ma and of a
similarageto the MyggbuktaDykes.
4. 7. Faeroe Islands
The Faeroe Islands are the exposedpart of the much
larger Faeroes Block, a fragment of continentalcrust
cappedby a thick sequence of basaltlavaflows,the Faeroe
PlateauLava Group (FPLG) [Casten, 1973; Castenand
Nielsen, 1975; Bott et al., 1974; HaM and Waagstein,
1984; Ritchie and Hitchen, 1996]. To the northwest,the
crust becomesentirely oceanicin structure,as it grades
into the Faeroes-Iceland
Ridge. Southwestwards, the block
boundsthe Rockall Plateau, a fragmentof Precambrian
the
I I I I I KI LaCeSrNd
RbBaThNbTa I I I I PSmZr[-ifEuTi
I I I I I I ; b YYbLu
I I I
Figure 10. Chondrite-normalisedREE and bulk silicate Earth-
normalisedtrace elementpatternsfor basaltsand basalticandesite
from WollastonForland, and Lower and Upper Series,Hold with
Hope, NE Greenland.Data source:Thirlwall et al. [1994]. Note
the similarity of the patternsof the high end Lower Serieslava
with the basaltsfrom the Upper Seriesfrom the Voring Plateau
(Figure 13). end calculatedassumingt=50 Ma.
lithospherethat has a capping of Palaeocenebasaltsand
Cenozoic sediments [Roberts, 1975]. The Faeroe Island
lavascoveran areaof approximately
1400km2 on land,
but extend considerablyfarther than this offshore. The
lavas have an exposedthicknessof about 3 km, but the
basehasnot beenrecovered,evenby drilling to a depthof
2 km in the Lopra-1 borehole and to 660 m in the
Vestamanna-1 borehole. Therefore, the entire sequence
exceeds5 km in thickness[HaM and Waagstein,1984;
Waagsteinand Hald, 1984; Waagstein,1988].
The FPLG has been divided into three series, Lower,
Middle, and Upper, on the basisof minor unconformities
[Noe-Nygaard and Rasmussen,1968]. All of the lavas
were eruptedsubaerially,and progressiveuplift
66 THE NORTH ATLANTIC IGNEOUS PROVINCE

lOO
of the islandsresultedin gentle tilting towardsthe east.
There was a hiatus between the Lower and Middle Series Faeroe Plateau Lava Group
Upper Series
lavas,markedby a 10-metre-thicklayer of claysand coal.
The latter has been datedas late Palaeocene[Lurid, 1983]. J3•••'•=• Uid•
eSeries .
There are presentlyno reliableradiometricagesavailable
for the FPLG. All of the Middle andUpper Serieslavasare
magneticallyreversed,but the Lower Seriescontainsat o lO

least two normal polarity events [Nielsen, 1983].

Waagstein[1988], on the basis of comparisonswith
marinemagneticanomalydata,proposedthat the lavasof r• Upper
Series,
Sample
K-1
(œNcl
t=-10)
UpperSeries,MORB-type(œNd
t > +9)
the Lower Serieswere emplacedduring the period early
26r to early 24r, and that the Middle and Upper Series
were emplacedduring C24r.
All of the analysedFPLG are tholeiites.Thosefrom the La Ce Pr Nd Sm Eu Gd b Dy Y Er Tm Yb Lu
Lower Series are silica-oversaturated,whereas the Middle 100 -

and Upper Seriesbasaltsare olivine tholeiites.A perhaps Faeroe Plateau Lava Group
more fundamentalchangein compositionoccursnear the
boundarybetween the Middle and Upper Series,where
there is a change from entirely light-REE-enrichedto a .,.e_,
© 10--
03 -
mixture of light-REE-depletedand light-REE-enriched • -
-

tholeiites[Noe-Nygaardand Rasmussen,1968; Schilling (/-)

,_•
-
-

and Noe-Nygaard, 1974;Bollingberget al., 1975; Gari•py :::3 _

ß

et al., 1983] (Figure 11). Some of this variationin REE •. 1--

patternsmay be due to contaminationof magmasby
continentalcrust (e.g., sampleK-l), but it is likely that
variations in the compositionof the sourceor depth of
melting also played an importantrole (see Sections6 and
7) [Schillingand Noe-Nygaard, 1974; Wood,1979a,b].
4.8. Seaward-DippingReflectorSequences
Seaward-dippingreflector sequences(SDRS) have a
characteristic,off-lapping architectureon seismicprofiles.
Their volcanic nature is now firmly establishedby deep-
sea drilling [Robertseta/., 1984; Eldholmet al., 1987;
Larsen et al., 1994]. The SDRS in any one region may be
up to 6 km thick, and they are complementedby thick
sequencesof basic intrusivesequencesat lower to middle
crustal level [e.g., White el al., 1987]. Furthermore,the
majority of the lavas, where sampled, were erupted
subaeriallyor in shallow water, testifying to substantial
supportof the marginduringrifting andplateseparation.
In addition to seismic reflection and refraction studies,
the seaward-dippingreflector sequencesin the North
Atlantic have been the target of six Deep Sea Drilling
Project (DSDP) and Ocean Drilling Program (ODP)
cruises:Legs 48 and 81 on the Rockall Plateau;Legs 38
and 104 on the Voring Plateau;and Legs 152 and 163 on
the SE Greenlandmargin.
4.8.1. Rockall Plateau/Hatton Bank (DSDP Legs 48 and
81). Leg 48 failed to reach igneousbasementdue to
E -
03 - UpperSeries,MORB-type(•Ndt > +9)
0.1
I 1 I I I KI LaCeSrNd
RbBaThNbTa I i I I PSmZrHfEuTi
I I I I I I ; b YYbLu
i I I
Figure 11. Chondrite-normalisedREE and bulk silicateEarth-
normalisedtrace element patterns for basaltsfrom the Faeroe
Islands. Average values for the Upper (excluding high-•Nd,
MORB-like depleted basalts), Middle and Lower Series are
shown.High •Nd, MORB-like basaltsfrom the Upper Seriesare
shown as a separate average value. Sample K-1 is a highly
contaminatedbasalt from the Upper Series, and is plotted on
Figures 6 and 7. Data source: Gariipy et al. [1983]. •Nd
calculatedassumingt=60 Ma.
technicaldifficulties. During Leg 81, however, basaltic
basement was reached at all of the drilled sites (552
through 555) [Robertset al., 1984]. Sites 552 and 553
were located on the main SDRS; Site 555 on the flanks of
Hatton Bank on the most 'landward' or easternpart of the
SDRS; and Site 554 was on the westernedgeof the SDRS.
Penetration of the basaltic basement was in excess of 100
m at three of the four sites.
The basaltsfrom the upperpart of the sequenceat Site
555 (the most landward of the four sites) are reversely
magnetisedand gave K-Ar agesof 52.3+1.7 and 54.5+2.0
Ma [Macintyre and Hamilton, 1984],
 SAUNDERS ET AL. 67

10--

lO

Q) -
o 552 Average
ß 553 Average (D - ß 553 Average
-o -
Hatton Bank a 554Average (.)O• -
_ Hatton Bank • 554Average
o ß 555 Average ß 555 Average
(D 1-- rJ')o.1
I I I i I Sm
I EuI Gd
I ;b DyI YI ErI Tm
I YbI LuI I I I NbTa
I I KI LaCe
i I SrI NdI PI SmI ZrI HfI EuI TiI ½bYi YbLu
i i
--

RbBaTh
_

La Ce Pr Nd
:• -
rn -
• -
(D _

E•10
lO

ß 918-8B
I -
a 918-12B a 918-12B
ß 918-12B ß 918-12B
ß 918-13B ß 918-13B

SE Greenland Margin ,, 918-15 SE Green/and Margin ,, 918-15

0.1
I ! I I I Sm
La Ce Pr NdI EuI Gd
I ;b DyI YI ErI Tm
I YbI LuI RbBaThNbTa K LaCeSrNd PSmZrHfEuTi b YYbLu

Figure 12. Chondrite-normalised REE and bulk silicateEarth-normalised trace elementpatternsfor basaltsfrom
Hatton Bank on the SW Rockall Plateau,and SE Greenland.Data for Hatton Bank representaveragesby ODP Site
(552 through555). Data for SE Greenlandrepresentindividualsamples.Data sources:HattonBank - Merriman et al.
[ 1988], with revisedNb valuesfrom Brodie and Fitton [ 1996]; SE Greenland- Fitton et al. [ 1996b].

corresponding
to C24r.Basaltsamples
analysed
4øAr-39Ar
by Sinton and Duncan [1996] all show disturbedage
spectra,althoughresultsfor two lavasfrom Site 555 which
indicate an eruption age of 57.6+1.3 and 57.1+5.6 Ma,
respectively,are consideredto be reliable; these ages
would correspondto C25r. These basalts lie above
sedimentsbelonging to nannofossilZone NP9 (upper
Palaeocene)[Backman,1984], so althoughthe K-Ar ageis
consistent
with the biostratigraphic
data, the 4øAr-39Ar
ages are significantly older. The NP9 sedimentsare
predominantly volcanogenic and were succeeded by
sedimentsdeposited in a brackish, intertidal lagoonal
environment. The basalts at Sites 552 and
eruptedin a shallowmarine environment,whereasthoseat
Site 553 were eruptedsubaerially.The basaltsat Site 552
are overlain by sedimentsof NPll and possibly NP10
(early Eocene) age, and there are abundant tuffs
interbeddedwith the overlying sediments.Basaltsat Sites
553 and 554 are at leastNP 11 (early Eocene)age.
The basalts recovered from Hatton Bank are all
tholeiites that show strong depletion of the light REE
(Figure 12), similar to parts of the Upper Lavas of the
Faeroesor the PreshalMore basaltsfrom Skye [doron et
al., 1984; Merriman et al., 1988]. Despitethe stronglight
REE depletion, however, it is apparent from the Pb
isotopesthat the basaltsfrom Site 553 have sufferedminor
contamination
by materialwith low 2ø6pb/2ø4pb
(<17),
but MORB-like 2ø7pb/2ø4pb,
possiblymid-Proterozoic,
Laxfordian continentalcrust [Morton and Taylor, 1987;
Merriman et al., 1988]. Such material has been dredged
from the Rockall Bank [Miller et al., 1973; Morton and
554 were Taylor, 1991].
4.8.2. Voting Margin.' DSDP Leg 38 and ODP Leg 104.
Three sites were drilled seaward of the Voring Plateau
escarpmentduring Leg 38 [Talwani et al., 1976]. Short
sectionsof altered basalt were recovered at all three sites,
although the importance of these rocks was not fully
realised until Hinz [1981] proposed that the
68 THE NORTH ATLANTIC IGNEOUS PROVINCE

50--- Voring Plateau Voring Plateau

(Site 642) (Site 642)

5-- o Average 7 to 7.5% MgO

ß Average 7.5 to 8% MgO
a Average 8 to 9% MgO
ß Average 9 to 10% MgO
Upper Series Upper Series
I I I I I I I I i I I I I I I

50;

10--

o Average47-50%SiO2
ß Average51-55%SiO2
a Average56-60%SiO2
ß Average
61-67%SiO2
Lower Series Lower Series

I I I I I Sm
La Ce Pr NdI EuI Gd
I lb DyI Yi ErI Tm
I YbI LuI RbBaTh NbTa K LaCeSr Nd P SmZr Hf Eu 'ri Tb Y YbLu

Figure 13. Chondrite-normalised REE and bulk silicateEarth-normalised trace elementpatternsfor basaltsand
evolvedlava compositions from the Voring Plateau(ODP Leg 104, Site 642). Samplesare averageson the basisof
MgO content(for the Upper Seriesbasalts)and SiO2 content(for the Lower Seriesbasaltsanddacites).Data sources:
Vierecket al. [1988, 1989].

representaccumulationsof basaltic lavas. A 900-m-thick
Eocenevolcanicsectionwas drilled on the Voting Plateau
at Hole 642E during Leg 104 [Eldholmet al., 1987]. The
volcanic successionwas divided into Upper and Lower
Series,separatedby about 7 m of estuarine,volcaniclastic
sediments.All of the lavas recoveredfrom the Voting
Plateauwere eruptedin a subaerialenvironment,except
for the somepillow basaltsfrom Site643 on the outeredge
of the SDRS [Eldholmet al., 1987].
The basaltsfrom the Upper Seriesat Site 642 include
tholeiiteswith flat to slightlylight-REE-enriched
profiles,
and they are compositionallysimilar to basaltsfrom the
ReykjanesRidge [Viereck et al., 1988] (Figure 13). The
gNdt=60 values (+6 to +7.5) are slightly lower than in
Icelandic basalts of equivalent Zr/Nb (Figure 7). The
Lower Series drilled at Site 642 comprisesbasalts,
intermediate rocks and 13 peraluminous, cordierite-
bearing dacite flows. The dacitesshow stronglight-REE
enrichment, low abundancesof Nb, Sr, P, Eu and Ti
(Figure13), elevated87Sr/86Sr
(0.7088to 0.7118)andlow
gNdt=60(-7 to -9) (Figure 6). These data are consistent
with assimilation of substantial amounts of metaluminous
continentalbasementor overlying sediments[Viereck et
al., 1988].
The lavas of the Lower Series are all normally
magnetised,whereas those of the Upper Series are
reverselymagnetised.They havebeententativelyassigned
to Chrons25n-24r [SchOnharting and Abrahamsen,1989]
but without radiometricage data this assignmentmust be
considered as speculative. Various workers have
suggested,on the basis of geochemicalcriteria, that the
basaltsfrom the Voting margincorrelatewith basaltsfrom
the conjugatemargin at Hold with Hope and Wollaston
Forland[Schilling,1976; Vierecket al., 1988; Uptonet al.,
1995]. Becausethe age of the NE Greenlandbasaltsis well
constrained,an implicationof this correlation,if correct,

that the Upper Seriesbasaltsfrom Site 642 are no younger
than C25r.
4.8.3. SE GreenlandMargin: ODP Legs 152 and 163.
A transect of five basement sites was drilled at 63øN on
the SE Greenland margin during Legs 152 and 163
[Larsen et al., 1994; Duncan et al., 1996]. This margin
was chosen for study becauseof its relatively simple
structure,detailedseismiccoverage,and the presenceof a
150-km-wide subcropof SDRS. The occurrencein the
distal portion of the SDRS of a magnetic anomaly
correspondingto C24n indicated that the oldest basalts
from the sequencewere at leastC24r.
The most successfulpenetrationand recovery was at
Hole 917A, close to the inner part of the SDRS on the
continental shelf, where 91 extrusive units were identified
from the 749-m-deephole. The sequencewas dividedinto
three series [Larsen et al., 1994]. Lavas from the Lower
Series include basaltsthat show significantlithospheric
contamination,but somehave preserveda sublithospheric
chemical signatureand resembletholeiitesfrom Iceland
[Fitton et al., 1996b] (these can be seen on Figure 7B).
The Middle Serieslavas,which includedacites,are clearly
contaminatedby crust[Fitton et al., 1996a,b;L.M. Larsen
et al., 1996]. The Upper Series,which exhibitsmuch less
contamination,contains magnesianbasalts and picrites
[Fram et al., 1996]. The Upper Seriesis separatedfrom
the Middle Seriesby a thin sedimentary horizon,implying
a marked hiatus in eruptive activity. All of the lavas
recovered at Site 917 were erupted in a subaerial
environment,and are reverselymagnetised.On the basis
of 4øAr-39Ar
ages,the basaltsof the Lower and Middle
Serieswere eruptedapproximately61-62 m.y. ago (C27r
accordingto the time scale of Berggren et al. [1995])
[Sintonand Duncan, 1996; Werneret al., 1996]. There are
no radiometric ages for the lavas of the Upper Series,
which meansthat the age of picriticmagmatismat this site
is not preciselyconstrained.
Basaltswere also recoveredat Site 989, inboard of Site
917; Sites915 and 990, approximately3 km to the SE of
Site 917; and at Site 918, located closeto the centreof the
outcropof the dippingreflectors.All of the basalticflows
recoveredfrom Site 918 appear,on the basisof Sr, Nd and
Pb isotopes,to be uncontaminatedby continentalcrustand
showstronglight-REE depletion(Figure 12) [Fitton et al.,
1996a,b]. In many respectsthey resemble the basalts
recovered from Hatton Bank, and the data for both areas
are plottedon Figure 12 for direct comparison.Like those
from Site 917, the basalts recovered at Site 918 were
erupted in a subaerial setting, and most are reversely
magnetised.The uppermostunit at Site 918 providesa
reliable4øAr-39Arage of 52 Ma [Sintonand Duncan,
SAUNDERS ET AL. 69
1996]. This unit may be a sill, so the underlyingbasaltic
flows may be significantlyolder,but the seafloormagnetic
lineationssuggestthat the bulk of the SDRS were erupted
duringC24r [e.g.,Larsenand dakobsd6ttir,1988].
4.8.4. Deep seismicprofiling of the continentalmargins.
An importantfeatureof the volcanicrifted marginsof the
North Atlantic is the presenceof thick prismsof material
with high seismicvelocities(7.3-7.4 km s-1) at lower
crustaldepths.These prisms,which have been observed
under the Rockall [White et al., 1987; Fowler et al., 1989;
Morgan et al., 1989; Barton and White, 1995], Voring
[Mutter et al., 1982; Skogseidand Eldholm,1988;Mutter
and Zehnder, 1988], SE Greenland [Larsen and
dakobsd6ttir,1988], and Lofoten [Goldschmidt-Rokita et
al., 1994] margins,have been interpretedas underplated
olivine gabbro or as continental crust with a high
proportionof gabbroicintrusions;they can be considered
as the plutonic equivalents,perhapswith a substantial
cumulate component,of the SDRS. The thicknessof the
underplatedprisms varies considerably,ranging from a
few kilometresat the distalregionsof the province(e.g., 5
km at the Edoras Bank, on the southeastflank of the
Rockall Plateau[Bartonand White, 1995]), to more than
10 km on the Hatton Bank and Voring margins [e.g.,
White et al., 1987; Mutter and Zehnder, 1988; Mutter et
al., 1988]. The large volumes of underplatedmaterial
stronglyinfluencecalculationsof total magmaproduction.
The increasein the thicknessof the igneousrocksalong
the Europeanmargin as the Faeroes-IcelandRidge is
approached implies either an increase in potential
temperature and/or an increase in mantle convection
towards the ancestral plume axis [Barton and White,
1995].
4.9. SummaryStatement
The activity associatedwith the North Atlantic Igneous
Provinceappearsto have begunin the Early Palaeocene,
approximately62 m.y. ago, duringmagneticreversal27r
or 26r (Figure2). Earlier,possiblyCretaceous,magmatism
is recordedon the. Anton Dohm Seamount,Rosemary
Bank, and the Faeroes-Shetland Basin, but their
relationship with the later activity is unclear. The
possibilitythat the North Atlantic IgneousProvincehad
precursormagmatism substantiallybefore 62 Ma cannot
be precluded,althoughthe volumesare likely to havebeen
small.
The available age data indicate two main phasesof
igneous activity within the North Atlantic Igneous
Province as a whole. Phase 1, recordedin the terrestrial,
continent-basedlava sequencesin West Greenland,
70 THE NORTH ATLANTIC IGNEOUS PROVINCE
Greenland, NE Greenland, and the British Isles, occurred
during C26r (and possiblybeganduringC27r) and lasted
for about 4 m.y., from 62 to 58 Ma. The Lower Series
lavas from central East Greenlandmay also belong to
Phase 1. Many Phase 1 basalts show evidence of
contaminationas they ascendedthrough the continental
lithosphere(see Section 6), although within individual
subprovincesit is not unusual to find lavas that have
undergone minimal contamination(e.g., some of the
basaltson Skye and Mull). This first phaseof activitywas
alsocharacterised by frequenteruptionof picritic magmas,
implying a hot mantle source,a point we returnto below.
Residualactivity continuedin severalregions.
Phase2 began at about 56 Ma, in the late Palaeocene,
and is linked with the breakup of the NE Atlantic,
formation of the bulk of the SDRS along the continental
margins, and the eruption of at least the Main Series
basaltsin centralEast Greenland.The Phase2 activity was
1ocalisedalongthe linesof continentalbreakupand, in the
case of the central East Greenland flood basalts, close to
the ancestralplume axis. The timing of emplacementwas
therefore controlled in part by the timing of plate
separation.The formation of the SDRS was essentially
diachronous.Phase 2 continues at the present day in
Iceland but with (1) more focusedactivity and (2) lower
magma productionrates than accompaniedplate breakup
during the Palaeocene/Eocene. Many of the magmas
associatedwith Phase2 are uncontaminated by continental
lithosphere,presumablybecausethey were eruptedduring
or after plate ruptureand separation.
We emphasise, however, that the age of the
emplacementof the SDRS is poorly constrained.For
example, the only SDRS that have been drilled to
underlying basement (Site 917 on the SE Greenland
margin) revealedan unexpectedseriesof olderbasalts(26r
insteadof the anticipated24r). This resultdoesnot negate
the observationsin the previousparagraph,becausethere
is no reasonwhy the lower SDRS in a regionare not pre-
breakup continental flood basalts, preserved by the
subsequentrifting, subsidenceand burial. It is likely,
however, that as more data become available, Phases1 and
2 will becomecontiguous,but the presentdatasuggestthat
there was a short hiatus between the two. A further
anomaly that concernsus is the age of the central East
Greenlandbasalts.It is not clear why eruption did not
begin in 26r or 27r times. The ancestralplume axis was
closerto Kangerlussuaq than either SE Greenlandor NW
Britain at 62-58 Ma, whether either Lawvet and MUller's
[ 1994] or White and McKenzie's [ 1989] positioningof the
hotspotis used.We suggestthat the Lower Basaltsof the
Kangerlussuaq areamay in fact be olderthan indicatedby
the publishedbiostratigraphicevidence.Confirmingthis
by radiometricmet_hods will be difficult, giventhe extent
of metamorphism of these lavas, but we note that
preliminaryage determinations supportan early periodof
eruption[Storeyet al., 1996].
5. RATES OF MAGMA PRODUCTION
AND CRUSTAL GENERATION
A characteristicof many LIPs is the transientnatureof
the bulk of the magmatism.It has long been recognised
that the rate of magma productionin the North Atlantic
increasedduring breakupand then sharplydeclined,but
these rates are only order-of-magnitudeestimates.This
impressionsimplyreflectsthe uncertainties in the volumes
of magma involved, and in the timing and duration of
emplacement.Roberts et al. [1984] estimatedthe total
volumeof Palaeocene to earlyEocenebasaltto be 2 x 106
km3,whereasWhiteet al. [ 1987]suggesteda totaligneous
volume, including eruptedmaterial and additionsto the
deepercrust,of between5 x 106 and 1 x 107 km3. The
most accurateestimatescurrentlyavailableare probably
those of Eldholm and Grue [1994]. Their estimateof the
total crustalvolume is 6.6 x 106 km3, closeto the lower
estimate of White et al. [1987] and, by their own
admission,a conservativefigure. This volume comprises
the SDRS and deep crustal prisms associatedwith the
Atlantic margins and the East Greenland basalts but
excludesthe magmatismassociatedwith the Greenland-
Iceland-FaeroesRidge and the Phase 1 activity in West
Greenland,Baffin Bay, and the BritishIsles,which clearly
predatesthe breakupmagmatismby severalmillion years.
It also excludesdykes and material that may have been
erodedor buried beneaththe Greenlandice cap.
Calculating eruption rates and duration of activity is
fraughtwith uncertainty.The problemis compounded by a
paucityof age data. On a broad scale,Eldholmand Grue
[1994] estimated the mean extrusion rate to have been
about0.6 km3/yr,or 2.3 x 10-4 km• per km of rift/yr,
assumingsteadymagma productionover a 3 m.y. period
alonga 2560 km rift zone. Thesefiguresrefer only to the
volcanic portions of the margins, and they increase
dramatically if the bulk of the magma was produced
duringthe earlieststagesof breakup;if two-thirdsof the
basalts were emplaced within 1.0 or 0.5 m.y. after
breakup,the initial eruptionrateswould be 2 to 4 times
higher [Eldholm et al., 1987; Eldholm and Grue, 1994].
Larsenet al. [ 1989] calculatedthe effusionratesin central
EastGreenland to be 2 x 10-4km3perkm of rift/yr, similar
to the figures calculatedby Nielsen and Brooks [1981].
Similarly, Larsen and dakobsd6ttir[1988] estimated

eruption
ratesof 4.5-4.6x 10-4km3 perkmof rift/yrwere
involved in the formation of the SDRS along the East
Greenland margin. For comparison, Ptilmason [1986]
estimatedthat the eruptionratesin Icelandare of the order
of 1.33x 10-4km3perkmofrift/yr,witha maximum
near
the centreof the islandof 2.5 x 10-4 km3 perkm of rift/yr
[Jakobsson,1972].
An alternative way of evaluating the magmatic
productivityis to considerthe total outputper kilometreof
rift axis, including lavas and plutonic rocks. If the
Icelandic crust has an averagethicknessof 25 km [Bott,
1983; White et al., 1995], then the mean magmatic
productivityoverthe last 15 m.y. hasbeen4 to 5 x 10-4
km3 perkm of rift/yr.By comparison,if theentireigneous
crustof the North Atlanticmargins(6.3 x 106 km3) was
emplacedin 3 m.y., during C24r, then the total magmatic
productivity
wasontheorderof 8.2 x 10-4km3 perkm of
rift/yr. If the bulk of the magmatismoccurredduringthe
initial 1 m.y. after breakup, then this figure would be
higher still. Along-marginvariationsin crustalthickness
also imply that total magmatic productivity varied as
functionof distancefrom the plume axis [e.g., Barton and
White, 1995].
Clearly, far more data are requiredto verify and refine
theseestimates;in particular,deep crustalseismicdata to
ascertainthe volumesof deepcrustalintrusivebodies,and
reliable agesfor the SDRS are needed.At 63øN on the SE
Greenlandmargin, recent drilling and age determinations
[Larsen et al., 1994; Sinton and Duncan, 1996] have
revealedthat a substantialportion of the lowermostSDRS
were eruptedearlier than C24r, implying that the effusion
rates are lower than those originally estimatedby Larsen
and JakobsdtSttir [1988]. Nonetheless,there appearsto be
a consistent message that the magmatic productivity
during breakupwas 2 to 3 times greaterthan on present-
day Iceland.
6. THE INFLUENCE OF THE CONTINENTAL
LITHOSPHERE
The volumes and compositionof magmas reflect a
complex interplay between the lithosphere and the
underlyingasthenosphere. The lithospherenot only serves
to truncatethe low-pressureregionsof any meltingcolumn
(even to the extent of precludingmelting altogether)but
also may modify magmasby contaminationand crustal-
level fractionation. Questions that are central to flood
basaltformationincludethe following. What is the role of
the continentallithospherein flood basalt genesis?Is it
simply a physical control of the melt zones in the
underlyingasthenosphere, or doesthe lithospheremelt to
SAUNDERS ET AL. 71
producesignificantvolumesof magma?To what extent
doespre-existingtopographyat the baseof the lithosphere
influencethe channellingof hot, buoyantplume material,
or how do thinspots in the lithosphereinfluence the
location and volume of magmatism?To what degree are
magmas contaminated as they ascend through the
lithosphere(especiallyby continentalcrust)?What is the
extent of medium- to low-pressure fractionation in
intracrustal magma chambers, especially in controlling
eruptionof picrites?
6.1. Crustal Contamination
Many lavasassociatedwith North Atlantic activity were
erupted through and onto continental crust, so crustal
contamination of magmas is expected. Furthermore, a
substantialportion of the province has been erupted
through lithosphereof Archaean or Proterozoicage (for
example, all of the Scottishprovince north of the Great
Glen Fault), with distinctive trace element and isotopic
characteristics.Here, we briefly review evidence for
contamination,beginningwith the BritishTertiary Igneous
Province,becauseit hasreceivedthe greatestattention.
Crustal contaminationof asthenosphere-derived Skye
and Mull lavas has been postulatedby many authors
includingMoorbath and Bell [ 1965], Moorbath and Welke
[1969], Carter et al. [ 1978], Moorbath and Thompson
[1980], Dickin [1981, 1988], Thompsonet al. [1982],
Thirlwall and Jones [1983], and Kerr et al. [1995a]. Of
interest here is the identification of the contaminants and
the mechanism of their incorporationin the magmas.
Carter et al. [ 1978] identified two fundamentallydifferent
contaminantsby using Sr and Nd isotope data. The
premiseof theirargument
wasthatthetwo contaminants,
granulite- and amphibolite-facies rocks, have very
different isotopic characteristics,the granulite gneiss in
particularhavinglow 87Sr/86Sr
ratiosreflectinglow time-
integratedRb/Sr ratios.They concludedthat in generalthe
basic lavas from Skye, Mull and the Small Isles were
contaminated by granulite-facies crust of Lewisian
(Archaean)age, whereasthe Skye granitescontaina large
componentof amphibolite-faciesLewisian crust. The low
Rb and Th and high Ba contentof many basaltsfrom the
British Tertiary IgneousProvincecan be explainedby the
incorporationof small amountsof Lewisiangranulite.This
is illustratedby Figures6, 7, and 8. Note in particularthe
displacementto low end values,but at constantZr/Nb, of
the Mull and Skye suiteson Figure7. Most putativecrustal
contaminantsexert relatively minor influence on Zr/Nb,
whereasPrecambriancrustalrocks,in particularArchaean
granulitesand amphibolites,have dramaticallylower
72 THE NORTH ATLANTIC IGNEOUS PROVINCE
valuesthan magmasderivedfrom sublithospheric
sources.
Similar displacementsare also shown by magmas from
central East Greenland, the Faeroes, and the SE Greenland
and Voring PlateauSDRS Lower Serieslavas.
Subsequentstudies have refined this model using a
variety of isotope and trace element tracers and
discriminants.Dickin [1981] showedthat the magmasof
the Skye Main Lava Serieswere contaminatedprimarily
by granulite-facies crust, but that the magmas that
producedthe succeedingPreshalMore basaltsand Cuillin
gabbros were contaminatedby amphibolite-faciescrust.
One interpretationis that the site of crustal assimilation
migratedupwardsthroughthe crustwith time. Other parts
of the Hebridean provincehave also been 'fingerprinted'
in this way. Minor pitchstoneand felsite intrusionsin the
Eigg lava pile show contamination by granulite-type
material [Dickin and dories,1983], andthe layeredgabbros
of the Rum intrusion show evidence of upper crustal
contamination [Palacz, 1984]. Thompsonet al. [1986]
identified possible contaminationfrom Moine schistsin
several basal tholeiites from SW Mull, and Wallace et al.
[1994] recently demonstratedthat the crustalcontaminant
of the Tertiary basaltsin Northern Ireland is probablyof
Proterozoic age, reflecting the younger age of the
basementbelow that region.
Crustal assimilation models of different
sophisticationhave been proposed.These range from the
combined assimilation-fractionalcrystallisation (AFC)
model of DePaolo [1981] to those involving partial
melting of gneissesof intermediate composition[e.g.,
Thirlwall and dories,1983] or the melting of more fusible
components in the Lewisian basement [Moorbath and
Thompson,1980; Dickin, 1981; Thompsonet al., 1982].
Selective extraction of melts from the crust appears
necessarybecauseof the frequentabsenceof a correlation
betweenthe degreeof silica saturationand 87Sr/86Sr
or
end. Using combinedNd and Ce isotopedata for Skye
basalts,Dickin et al. [1987] arguedthat the contaminant
precursors were granitic sheets, selectively extracted
during the assimilation process, rather than partially
melted intermediategneisses.Note, however,that Reiners
et al. [1995] have suggestedthat under certainconditions
AFC-type processescan lead to significantcontamination
of relatively unfractionatedbasalticmelts.
Thirlwall and dories [1983] and Kerr et al. [1995a]
notedthat within the Mull PlateauGroup and Skye Main
Lava Seriesthe mostmagnesian(i.e., hottest)lavasare the
most contaminated. This observation led
[ 1995a] to suggestthat the hottestmagmashad undergone
assimilation of continental crust during their turbulent
ascentin thin sheet-likemagma conduits,with the more
evolved magma being too cool to assimilatemuch crust.
The pattern of contaminationobservedin many of the
Faeroese basalts is similar to that recorded by British
Tertiary IgneousProvincelavas.SampleK-1 of Garidpyet
al. [1983], recovered from the Upper Series, has
abnormallyhigh 87Sr/86Sr
and 2øSpbfiø4pb and low end,
2ø6pb/2ø4pb,
and2ø7pb/2ø4pbratios(Figures6, 7, and 12),
consistentwith contaminationof the parentalmagmaby
Precambrianamphibolitegneiss.Less contaminatedlavas
from the Faeroesfall within the 'depleted'quadranton the
Nd-Sr mantle array (Figure 6), but many of thesebasalts
haveanomalously
high2øSpb/2ø4pb
ratios,againsuggesting
assimilationof high Pb/Nd crustalmaterial.Somesamples
from the Lower, Middle, and Upper Seriesfall on the so-
called oceanic mantle arrays on the Nd-Sr and Pb-Pb
diagrams suggesting that these samples suffered
insignificantcontamination.
Other suites that show evidence of contamination that
has been attributed to assimilation of continental crust
includethe following:
ß Contamination by Precambrian, possibly mid-
Proterozoic
Laxfordiancrust,to producelow 2ø6pb/2ø4pb
values in SDRS basalts from Site 553 on the Rockall
Plateau[Morton and Taylor, 1987].
ßEvidenceof contamination by andbulk assimilationof
levels of granulite and amphibolite gneissesin the basalts and
dacitesrecoveredat Site 917, especiallyin the Lower and
Middle Series,on the SE Greenlandmargin[Larsenet al.,
1994; Fitton et al. , 1995; 1996a,b].
ß Bulk assimilationof crust to produceperaluminous,
cordierite-bearing,
high 87Sr/86Sr
and low end dacitesin
the Lower Series at Site 642 on the Voring margin
[Vierecket al., 1988] (Figures6, 7, and 13) andto produce
dacites recovered from Well 163/6-1A on the Rockall
Plateau[Morton et al., 1988].
ß Upper Series lavas from Hold with Hope (some
containing more than 10% MgO) which have high
2ø7pb/2ø4pb
and2ø8pb/2ø4Pb[Thirlwall
et al., 1994].
ßHigh-SiO2,high87Sr/86Sr
basaltsin theplateaubasalts
from ScoresbySund [Larsenet al., 1989].
ß Selective contamination of basaltic magma by
Precambrianbasement in some dykes along the East
Greenlandmargin [Blichert-Tofiet al., 1992].
ß Evidence of extensivecontaminationof picritic and
basalticmagmas of the Vaigat and Malig•t Formations,
West Greenland,by shalesand sandstones[Pedersenand
Pedersen,1987]. In the more extremeexamples,this has
Kerr et al. producedrhyolites [Pedersenand Pedersen, 1987] and
nativemagmaticiron [e.g., Pedersen,1981].
The main conclusion to be drawn from these studies is
that crustal contaminationwas virtually a
 SAUNDERS ET AL. 73

process,
at leastduring
Phase
1 activity,
butalsoduring Weaverand Tarney[1983]arguedthatevidencefor
someof the Phase2 activity.An implication
is thatthe crustal
contamination
in manyPhanerozoic
floodbasalts
is
contamination
needsto be identifiedand accountedfor ambiguous.
The crustandthe underlying
continental
before
anyattempt ismadetoevaluate
thecomposition
of lithosphere
candevelop
together,
andmayinherit
similar
thesublithospheric
source,
or estimates
theconditions
of traceelementandisotopic
signatures.
Theclearadvantage
meltingby, for example,
REE inversion
methods[e.g., of invoking
thecontinental
lithosphere
isthatit explains
White and McKenzie, 1995]. the absenceof a correlationbetweenSiO2contentand,for
Althoughit may be arguedthat someof the example, PborSrisotopes. This"problem" can,however,
contaminationmayhaveoccurred asthemagmas ascended be circumventedby usingselectivecontamination models,
throughthe continental lithospheric
mantle, a potential oropen-system magma chambers, as mentioned above.
reservoir
for long-termdevelopment of radiogenic Sr and
unradiogenic Nd (Pb may be eitherradiogenic or 6.2. TheLithosphere
asa Mechanical Filter
unradiogenic),thereis littleevidence for thisbeinga
majorprocess in theNorthAtlanticIgneous Province. Not onlymaythe crustcontaminate basalticmagmas,
Holm[1988]argued thatthelow•Nd,high87Sr/86Sr lavas but it cancauseascending magmas to stall,eitherat the
Moho or at shallower discontinuities,and form magma
belonging to the LowerBasaltsfrom Kangerlussuaq
represent mixing betweenan incompatible-element- chambers. Thompsonet al. [1972]demonstrated thatthe
enriched component in ancientcontinental lithospheric lavasfromtheSkyeMainLavaSeriescouldbeeitherHy-
mantle,andmeltsfroman Icelandic-type
source(hence or Ne-normative,
implyingthatthelavashadfractionated
producing thehorizontal arrays onFigure 7).Herejected to theirpresentcompositions at pressuresof-1.0 GPa,
the suggestion of contamination of the magmas by wherethelow-pressuresilica-saturation
thermalbarrieris
granulite- or amphibolite-dominated crustpartlyon the not operational.
Thompson [1982]later showed thatthe
basisthat the Rb/Nb ratiosof the lavasdid not corroborate SkyeMainLavaSeries basaltsplottedalongthe0.9 GPa
the isotopesystematics. However,the lavas have olivine-plagioclase-clinopyroxene cotectic, possibly
undergone greenschistgrade metamorphism, rendering the reflecting
processes
operating
inMohomagma chambers.
Rb abundances unreliable. Themajorityof thebasalts
alongtheBiosseville Kyst
The involvement of hydrouscontinental lithospheric and near ScoresbySundwere eruptedto the eastof the
mantlein basaltproduction is neverthelessan attractive Caledonianfront,
whereastheKangerlussuaq regionliesto
the west of this belt. In addition, whereas the
option. Several workers [e.g., Gallagher and
Hawkesworth, 1992] have arguedthat the continental Kangerlussuaq regionhas small, generallyshallow,
lithosphericmantleundergoes wholesale meltingduring MesozoicandearlyTertiarybasinsoverlyingthe stable
extension-drivendecompression. Thompson andMorrison Proterozoicand Archaeanbasement,the region to the
[1988]andKerr [1993]haveproposed thatthemantle north of Scoresby Sundis characterised by a deepcrustal
lithospherehasalsocontaminated someof theTertiary basin, and relatively thin continentalcrust--theJameson
basaltsof SkyeandMull. Dickin[1981]argued thatthe Land Basin [Larsen and Marcussen, 1992]. This crustal
extremely low2ø6pb/2ø4pb and2ø7pb/2ø4pb end-member in anisotropy may have controlledthe natureof theerupted
somebasaltscouldnot be generated in the mantle;the magmas. Lavasequences fromnearScoresby Sundshow
mostlikelysource of suchmaterial, he argued,wouldbe repeated development of FeTi basalts,consistentwith
Lewisiangranulite (i.e.,ancient,
U-poorrock).However, fractionation and mixing in upper- and mid-crustallevel
small-degree meltspercolating fromthe asthenospheremagmachambers [Larsenet al., 1989;Brookset al.,
mayleadto modification of thelithospheric mantle, for 1991 ]. Seismic reflectiondatafortheJameson LandBasin
example, selective
enrichment of incompatible elements indicate not only that thebasin sitson topof anomalously
andhydrous phases [e.g.,Foleyet al., 1987;McKenzie, thinbasement (6-9 km), but alsothatthereareextensive
1989;,,trndtandChristensen, 1992].Giventhatmanyof sill and dyke complexes withinthe basin[Larsenand
theTertiarymagmas wereemplaced intolithospherethat Marcussen, 1992]. These may bethemid-to upper-crustal
had been essentially unaffected by tectonismsince magma bodies that supplied thethicklavasequences tothe
Proterozoicand in someareasArchaeantimes,therehad Scoresby
Sundregionvia lateralinjection,although
beena long'time for the lithospheric
mantleto have geochemical
comparisons
areneeded to corroborate
this
becomemodifiedin thisway.It is questionable,
however, suggestion.
whethervery low U/Pb ratioscan be generated
and LavasfromtheKangerlussuaq
LowerBasalts,
however,
maintainedwithinthe continental
lithospheric
mantle. appear
to havehada different
eruption
history.
Not
74 THE NORTH ATLANTIC IGNEOUS PROVINCE

are a high proportion of the lavas highly magnesian, 16 [ [ [ [ [ [

Lithosphere
implying more or lessdirect ascentthroughthe crust,but Thicknoss

also more evolvedmagmasshow strongevidenceof high P•oli

pressure(-0.9 GPa) pyroxenefractionation,presumablyat
Kangerlussuaq
the base of the crust [Fram and Lesher, 1996], and similar

'4iI
to thatreportedin the SkyeMain Lava Seriesof the British
Tertiary Igneous Province. Crustal structurehas clearly
British
Isles
• ':'"'":'"'":""•-..'"iij
/ Score
by
..::iii:.::..

played an importantrole in the emplacementhistoryof the

Greenland basalts.
Once magma chambershave becomeestablished,either
at the base of or within the crust or beneath oceanic
spreadingcentres,they will act as effective densitytraps,

hindering the eruption of high-densitypicritic magmas
[e.g., Huppert and Sparks, 1980; Stolper and 14/alker,
1980]. This may explainwhy suchmagmasare frequently
restricted to the initial episodesof continentalbreakup
(e.g., in West Greenland, the early activity in SE
Greenland,the Lower Basaltsin Kangerlussuaq, andparts
of the British Tertiary provincesuch as Rum) before the
establishmentof steady-state magmabodies[Fitton et al.,
1995]. The alternativepossibilityis that the temperatureof
the mantle sourcehas changedwith time, a point returned
to in Section 7.
The lithosphere also acts as a lid, capping and
truncating any melting events in the underlying
asthenosphere[e.g., McKenzie and Bickle, 1988; Ellam,
1992; Saunderset al., 1992; Fram and Lesher, 1993]. At
its most extreme,thick lithospherewill curtailall melting,
but the introduction of hot mantle may raise the
temperatureof the mantleto the point wherethe solidusis
reached and melting begins. The variation in mantle
temperatureand lithospherethicknesscan produce four
'end-member'conditions.(i) Thick lithosphere(> 150 kin)
and ambient temperaturemantle, where melting will be
restrictedto small-degree,incompatible-element-enriched,
hydrousmelts. Most of these melts probably never reach
the surface,being 'frozen' in the overlying lithospheric
mantle. (ii) Thick lithosphereand hot mantle (excess
potentialtemperature,
ATp> 100øC)causemeltingto be
restrictedto high pressureøThis will producehigh-MgO,
high-FeO and incompatibleelement-enriched liquidswith
stronglyfractionatedREE patternsbecauseof the retention
of the heavy REE in garnet. (iii) Zero-thickness
lithosphereand normal mantle temperaturesleading to a
systemdominatedby shallow-levelmelting and producing
normal ocean crust. (iv) Zero-thicknesslithosphereand
elevated mantle temperature; like condition (ii), the
melting will begin at greaterdepththan (i) and (iii), thus
producing some melts with high FeO contents and
fractionatedREE patterns.Melting continuesto a much
higher level, however,resultingin very large volumesof
L
Rooks,
8lReykj•nes-
• •. -
0.5 1.0 1.5 2.0 2.5 3.0
Fractionation-corrected •02
Figure 14. Fractionation-corrected total iron (FeO) versus
fractionation-corrected TiO2 in basaltsfrom the North Atlantic
IgneousProvince. Arrows indicatequalitativelythe effect of
changingthe thicknessof the lithosphere(i.e. changingthe
pressureat the top of the meltingzone)andchangingthe pressure
at whichmeltingbegins(whichis a functionof temperature and
volatile contentof the source).From Fram and Lesher[1993].
liquid and dilution of the high-pressuresignature.Thick
oceaniccrustresults(Iceland).
The "lid effect" is evident in the compositionof many
North Atlantic basalts[e.g., Fram and Lesher, 1993]. The
various REE and multielement plots presentedin this
paper reveal the fractionationof light from heavy REE,
andY and Ti from the more incompatibleelements.This is
likely to result from melting in the presenceof garnet,
which preferentiallyretainsthe heavy REE (e.g., Tm, Yb
and Lu) and Y and Sc. Fram and Lesher[1993] attributed
correlationsbetween FeO and TiO2 and between Dy/Yb
and TiO2 to lithosphere-induced effects (Figure 14). By
usingonly primitivebasaltsto minimisethe effectsof low
pressurefractionation,all of their datawere thencorrected
for any slight fractionation by numerically adding
equilibrium olivine to bring the Mg number (i.e.,
Mg/Mg+Fe2+) of the rocksto 0.70. Thesefractionation-
correctedTiO2 and FeO values show a clear positive

This trend is the oppositeto what is seenin regionally
averagedMAR and Icelandic basalts,where there is an
inverse correlation between fractionation-corrected Fe and
moderately incompatibleelementssuch as Na20 [Klein
and Langmuir, 1987] and TiO2. The iron contentof the
primarymelts increases with the meanpressureof melting
[Langmuirand Hanson,1980;Klein and Langmuir,1987],
but the content of Ti and Na decreaseswith increasing
mean extent of melting [Klein and Langmuir, 1987].
DecreasingNa and increasingFe are indeed observedin
MAR basalts as Iceland is approached, presumably
reflecting increasing depth and extent of melting
associatedwith the Iceland thermal anomaly (Figure 5).
Why then do the data for basaltsfrom the entireNorth
Atlantic Igneous Province show positive correlations
betweenFeO, TiO2 and Dy/Yb? Fram and Lesher[ 1993]
argued that the lithosphererestrictsmantle upwelling,
effectively increasing the mean pressure of melting.
Sampleswith high fractionation-corrected TiO2, FeO and
Dy/Yb are thus derivedfrom meltsthat retainedthe high
pressuremelting signature.In a generalsense,the values
of TiO2, FeO and Dy/Yb decreasewith time, suchthat the
youngerbasaltshavethe lower values.This, they argued,
is due to progressivethinningof the lithosphereassociated
with continentalbreakupand developmentof an oceanic
system.
Similar argumentscan be presentedusing Sc and Zr
(Figure 15). Scandiumis an elementthat is partitionedinto
clinopyroxene (DSc(2.aGPa)--0.51)andgamet(DSc(2.aGPa)= SE Greenlandmargin [Fram et al., 1996] where the
2.27) (where D=mineral-meltpartitioncoefficient)[Ulmer,
1989]; it behavesin a similar fashionto the heavy REE.
During low-pressure meltingin the field of spinelstability, provinceis associatedwith the Central Complexes(where
Sc/Zr of the melt increaseswith increasing extent of
melting, reflecting the progressiveremoval of pyroxene
(Figure 15, curves 1 and 2). During melting of garnet
peridotite,however, Sc is retainedby garnetin the source
until all of the garnetis melted,eitherisobaricallyor, more
likely, as the melting matrix ascendsabove the garnet-
spinel transition.The basaltsfrom the SDRS along the
Hatton Bank and SE Greenlandmarginsshow the highest
concentrationsof Sc; along with basaltsfrom the Upper
Series in the Faeroesand Voring Plateau, they have Sc
concentrationsas high as or higher than MORB. The Sc
datathus corroboratethe REE data,which imply that all of
these basalts were generatedby large-degreemelts at
shallow levels (curve 1 on Figure 15). Even these
conditionsare insufficientalone to producethe high Sc
abundancesfound in some SDRS basalts; the melts must
have undergone substantialfractionation of plagioclase
and olivine (e.g., curve 1a). Note that Tertiary basaltsfrom
Mull, central East Greenland, and Iceland have lower Sc
SAUNDERS ET AL. 75
abundancesand lower Sc/Zr, implying that the 'average'
depthof melting is greaterthan duringthe formationof the
SDRS.
This change in magma composition, apparently
reflectingthe shift from high (garnetpresent)to moderate
(garnet minor or absent)pressureof melting, is observed
locally in at least three areas. Ellam [1992] and Kerr
[1994] suggestedthat the heavy-REE-depletedpatterns
observedin Skye Main Lava Seriesand the Mull Plateau
Group (Figure 8) were generatedbelow a lithospheric'lid'
which was at least 60-80 km thick and restrictedmelting
to a sourceregion containinggarnet.Later, more extensive
melting of a similar source beneath thinner lithosphere
(<60 km), entirely within the spinel lherzolite stability
field, resultedin the relatively flat heavy REE patternsof
the Fairy Bridge/Coire Gorm and PreshalMore/Central
Mull Tholeiite types. A similar shift in chemistry is
observed in the Antrim Lavas (when passingfrom the
Lower Formationto the CausewayMember of the Middle
Formation [Barrat and Nesbitt, 1996]). In a third site-
specificstudy,Fram et al. [ 1996] also calculatedthat the
mean pressureof melting decreased,and the extent of
melting increased,during the rifting and breakupof the SE
Greenlandmargin.
An implication of Kerr's [1994] model is that the
lithospherebelow the Hebrideshad to thin by at least 15
km in 2 to 3 m.y. It is difficult to see how this thinning
could be achievedby lithospherestretchingalone, unlike
the
magmatismand breakup were more or less in the same
place. Most of the crustalextensionin the British Tertiary
locally it may be up to 25%; for example, on Skye
[Speight et al., 1982]), but the regional extension is
generally much less (-•1% [England, 1988]).
Consequently,Kerr [ 1994] and Barrat and Nesbitt [ 1996]
have suggestedthat the lower lithospherewas erodedby a
strongly advectingplume mantle. Removal of 15 km of
lithosphereby hot plume mantle would, however, cause
substantialsurfaceuplift (R. England,pers.comm., 1996),
of which there is no evidence. Long-distance,lateral
transportof magmasto the Skye and Mull fieldscannotbe
completely eliminated, and neither can changingsource
compositions.
7. THE ROLE OF THE ICELAND PLUME
7.1. ThermalAspects
The voluminous magmatism that accompaniedplate
separationwas mostly short-lived and often restricted
76 THE NORTH ATLANTIC IGNEOUS PROVINCE

6O
SE Hatton Bank
Greenland Vering Plateau (Upper Series)
Margin 1
5O 2%
•% 2
Faeroes
Prince of Wales Mts

4O 0.9 •
Iceland
2O%
Faeroes
Scoresby
Mull Sund
(Central
Tholeiites)

2O
2O%
10%
'• 2%
'
Mull (Coire Gorrn
Basalts)

10 Archean Felsic
Mull (Plateau Basalts)
• Rocks
ß
ß.........?•..-.,•.½..............•..•½•½.
..,..,...,.....,...,.•,.•,..,..,.. • • ". I I I

50 100 150 200 250 300

Zr (ppm)
Figure 15. Sc versusZr for North AtlanticIgneousProvincebasalts.The field of Archaeanfelsicrocks(mainly veins
in granulites[Kerr, 1995a]) suggests that contamination hasplayedan insignificantrole in controllingthe abundances
of Sc and Zr in thesesamples.Data sourcesas for Figures6 and 7, with additionaldatafrom Fitton et al. [1996b] for
the SE Greenlandmarginbasalts,andBrodie andFitton [ 1996] for HattonBank basalts.Icelandbasaltdatafrom A.D.
Saunders,J. G. Fitton andB. S. Hardarson(unpublished neutronactivation(Sc) andX-ray fluorescence (Zr) analyses).
Modellingcurvesassumefractionalmeltingof spinelperidotite(curvesI and2) andgarnetperidotite(curve3). Open
circles:'depleted'mantle;filled circlesandtriangles:50:50 mixtureof 'depleted'and 'primitive' mantle(assumesan
equalcontributionfrom eachmantletype duringmelting).Compositions and definitionsof 'primitive' and 'depleted'
mantle, mineral proportionsand partition coefficientsfrom McKenzie and O'Nions [1991]. Line l a represents
evolutionby fractionalcrystallisation of a liquid derivedby 20% meltingof 'depleted'spinelperidotite.Fractionating
phasesassumedto be plagioclaseplus olivine (Sc and Zr will each have partitioncoefficientsapproachingzero).
Fractions of liquid remaining are shown. Inclusion of clinopyroxenein the crystallisingassemblagewill lead to
decreasingSc/Zr ratiosin the liquid.

no more than one polarity event in any one locality.
Shortly after plate separationthe excessivemagmatism
was restrictedto the Greenland-Faeroes Ridge,and latterly
to Iceland, on the axis of the present-dayplume. There is
strongevidencefrom seafloorbathymetry[Haigh, 1973],
from the major element compositionof basaltserupted
along the MAR south of Iceland [Klein and Langmuir,
1987], and from the thicknessof the ocean crust around
Iceland [White et al., 1995] that the thermal effectsof the
present-dayIceland plume extend beyond the boundaries
of Iceland itself [e.g., Vogt, 1983]. Furthermore,it can be
shown that the thermal effects have a more widespread
extentand influencethanthe compositional(traceelement,
isotopic)effects(Figure 5). Interestingly,it would appear
that the thermal influenceof the Icelandplume may be as
widespread,at least to the south of Iceland, as it was
duringearly Tertiary times.
Recentactivity on Icelandprovidesa usefulmodelwith
which to explorefurtherthe Palaeocenemagmatismalong
the margins of the North Atlantic. Certainly, the likely
presenceof a plume beneath Iceland (see Section4.1)
lends powerful supportto thosewho advocatea plume-
type origin for the bulk of the North Atlantic Igneous
Province. However, there are several anomalies that have
to be accountedfor if we follow the plumeparadigm.
Firstly, as discussedin Section 5, the rate of

generationduring the formationof the SDRS was up to
three times greaterthan in present-dayIceland. If the rate
of magmatism was simply a function of magma
temperature and extension (decompression) [e.g.,
McKenzie and Bickle, 1988], then this would imply that
during the Eocene large volumes of the plume were
substantiallyhotter than at the presentday. Is there any
independentevidence for this? Secondly, some basalts
from the SDRS (e.g., from the SE Greenlandmargin and
Hatton Bank), and the more distalon-shoreportionsof the
North Atlantic IgneousProvince(West and SE Greenland;
the British Tertiary Igneous Province) are at least
superficiallysimilar to MORB and thus appearto have
beenderivedfrom a depletedmantlesource,[e.g.,Joron et
al., 1984; Rob#lard et al., 1992; Holm et al., 1993; Kerr et
al., 1995b; Fitton et al., 1996b, 1997]. How could this
occur if a plume was responsiblefor transportingthe
thermal energy to the site of magma generation?Before
we addressthese two points, it is worth considering
alternativesto the plumemodel.
The excessmagmatismthat occurredduringthe initial
stagesof plate separationmay be a result of enhanced
convectionin the underlyingasthenosphere [Mutter et al.,
1988]. Fundamentalto this model is the nature of the
initial plate separation;onerequirementis a sharpthermal
boundarylayer at the edgesof the separatingplates.The
asthenosphere adjacentto this boundaryor 'edge' cools,
thus setting up a local convection cell beneath the
developingrift axis. Increasedflux of mantle throughthe
melt zone results in increasedmagma generationand the
formation of a volcanic rifted margin. Developing plate
boundariesthat do not have this marked 'edge' shouldnot
develop the necessaryconvection systems and remain
essentially'non-volcanic'(e.g., Biscaymargin).
Although enhancedconvectionmay be an important
method of generatinglarge volumesof magma from hot
mantle (seebelow), it is difficult to seehow it can work in
ambient asthenosphere
(potential temperature,Tp,
-1300øC). The convectionsystemrequiresheat extraction
from the downwelling limb of the cell into the adjacent,
cool lithosphere.This effect, coupled with energy loss
during melting, will tend to be self-limiting, and it is
unlikely that any more melt will be producedthan during
normal seafloor spreading, once plate separationhas
occurred. Numerical models [e.g., Boutilier and Keen,
1995] tend to supportthis prediction.On the otherhand,if
the mantle beneaththe lithosphereis hotter than ambient,
then enhanced convection could occur as the hot mantle
flows from beneaththe lithosphereand is channelledinto
the developingrift axis.
Knowledge about the temperature and thermal
SAUNDERS ET AL. 77
evolutionof the sublithospheric
mantle during Palaeocene
and Eocene times is crucial to these discussions. The
volume of melt generatedat a rift zone is dependenton the
amount of decompression(controlled by lithosphere
thicknessand amountof extension)and the temperatureof
the underlying asthenosphere[McKenzie and Bickle,
1988]. Estimatesaboutthe thermal stateof the mantle may
therefore be obtained from the volume of lavas and
intrusive rocks, providing we have an approximate
understandingof the melting conditions(anhydrousor
hydrous;passiveor active?) and the compositionof the
source material. Alternatively, we may assess the
temperaturedirectly from the lavasthemselves.
The most cited evidence for the existence of hotter-
than-ambientmantle is the occurrenceof picritic magmas.
If it can be shown that eruptedliquids had a high MgO
content(either by direct analysisof aphyric lavas or by
extrapolationfrom equilibriumolivine compositions)and
were anhydrous, then this implies elevated melt
temperatures.Whether or not picrites can be used as
temperature probes is, however, less clear. Although
picritesare found associatedwith continentalflood basalts
[e.g., Campbell and Griffiths, 1990], and have thus been
linked with anomalouslyhot mantle, their frequencyof
eruption in these and other settings (e.g., mid-ocean
ridges)may be a functionof 'eruptivepotential'as much
as actualfrequencyof production:picritesare densemelts
which will tend to stall in basalticmagma chambers[e.g.,
$tolper and Walker, 1980].
In the North Atlantic IgneousProvince,picritic parental
magmashave been inferredfor Baffin Island [e.g., Clarke
and Pedersen, 1976; Francis, 1985], West Greenland
[Pedersen,1985; Gill et al., 1992], centralEast Greenland
[Nielsen et al., 1981; Gill et al., 1988; Fram and Lesher,
1996], SE Greenland [Larsen et al., 1994; Fram et al.,
1996; L.M. Larsen et al., 1996; Thy et al., 1996], the
British Tertiary Igneous Province (especially the Rum
complex) [e.g., Kent, 1995], and Iceland [e.g., Elliott et
al., 1991]. Parentalmagmasof the Skye Main Lava Series
and the Mull PlateauGroup may have containedas much
as 15% MgO [Kerr, 1995a;Scarrowand Cox, 1995]. The
potentialtemperatureof the upwellingmantlethat would
have produced high-MgO basalts in Mull has been
estimatedto have been 1420-1460øC [Kerr, 1995a], some
120-160øC hotter than ambient. The Rum liquids would
have required an even hotter source. Olivines from a
peridotiticdyke on Ard Nev, Rum [McClurg, 1982], may
have been in equilibriumwith a dry melt containing18-20
% MgO, at a temperatureof 1420-1460øC (corresponding
to an approximatesourcetemperatureof 1540øCat 2 GPa)
[Kent, 1995]. Equivalent potential temperaturesfor
78 THE NORTH ATLANTIC IGNEOUS PROVINCE
sourceof the West Greenlandpicritesmay have been as
high as 1540-1600øC [Gill et al., 1992].
A full evaluation of the Palaeocenepicrites is long
overdue, but two observationsare pertinent. (i) The
majority of the picrites were erupted during the early
stages of the North Atlantic Igneous Province, during
times of initial rifting and prior to breakup. With the
possibleexceptionof the picritesrecoveredat Site 917 on
the SE Greenlandmargin (presentlyundated),no high-
MgO liquidshavebeenfoundin the SDRS associated with
the Phase2 magmatism.As pointedout above,this should
be treated with caution, however, because steady-state
magmabodiesmay have filtered the densepicrite liquids.
(ii) The picrites are not restricted in spaceto any one
locality that may, for example,have corresponded with a
hot plume stem [Campbell and Griffiths, 1990] (see
discussionby Larsen et al. [1992] and Chalmerset al.
[ 1995]). The eruptionsitesare scatteredfrom Baffin Island
in the west to the British Tertiary Provincein the east,a
distanceof some2000 km (Figure 3).
Although appealmay be made to the developmentof a
hotcell [Anderson et al., 1992] beneath Baffin Island,
Greenland, and NW Europe to explain the wide
distribution of the Palaeocenemagmatism,we find this
explanationunsatisfactoryfor severalreasons.Firstly, the
hotcell hypothesisadvocatesthermal blanketingby the
lithosphere,and coolingby adjacentsubductionzones,to
generatehorizontalthermalgradientsin the asthenosphere.
The time span required to develop high potential
temperaturessuggestedby picritic magmasis likely to be
severalmillions if not tens of millions of years, if indeed
suchhigh temperatures couldever be achieved.Suchslow
build-up of heat is not consistentwith the absenceof
widespread magmatism in developing Cretaceousrifts
suchas the LabradorSea-DavisStrait or Rockall Trough.
Secondly,the early phaseof magmatismappearsto have
occurredsimultaneouslyover a widespreadarea. This is
again inconsistentwith the long period of asthenosphere
heating. Thirdly, the presence of a thermal, gravity,
chemical, and topographicanomaly beneathpresent-day
Iceland lends powerful circumstantialsupport that a
similaranomalyexistedin the past.
The uplift recordedin the DanJansedimentsin central
East Greenland and in the Forties Field provides strong
corroborative evidence for the development of a
widespreadthermalanomalyandresultantbuoyantsupport
of the lithosphere.The more distal effectsof this anomaly
are recorded in the subsidencehistory of the North Sea
Basin,where there is evidencefor a transientthermalpulse
between 65 and 52 Ma [Nadin et al., 1995]. The
simultaneousoccurrenceof magmatismin West Greenland
and the British Isles implies phenomenallyhigh mantle
flux rates if the plume was axisymmetric,indicatingthe
arrival of a large plume head or blob. These flux rates
could be reducedsomewhatif the plume was channelled
by pre-existinglithospherictopography,or ascendedin a
sheet-like form rather than as a tube. Whatever model is
invoked, for the thermal anomaly to be emplaced so
quickly and over sucha wide area impliesthat it had low
viscosity and, by implication, was very hot. This could
account for the restrictionof high-MgO liquids to the
Phase 1 activity; later magmatismwas derived from a
systemthat had alreadybegunto cool (Figure 16).
What could be responsiblefor the burst of magmatism
that producedthe SDRS and the complementaryplutonic,
deep crustal prisms? Plate rupture and separationand
decompressionmelting of the hot asthenosphere [White
and McKenzie, 1989; Barton and White, 1995] appearto
offer the best explanation. Cliff et al. [1995], using
subsidencecurves constrainedby sedimentologicaland
biostratigraphicdata, estimated that the excessmantle
temperatureat the Hatton Bank margin was of the orderof
75 to 100øC at the time of rifting. The equivalentAT at
Site 643 on the Voting marginwas between30 and 75øC,
and 110-190øC at Site 918 on the SE Greenlandmargin.
Their subsidence model predicted a peak mantle
temperatureat 42 Ma, ratherthan at the time of breakup,at
Site 918. The temperatureexcessesare lower than the
estimatesderived from high-MgO liquids and provide
supportfor the idea that the systemhad begun to cool
down by the time of platebreakup.
Enhancementof melt productionby activeupwellingof
hot mantle from beneaththe separatingplates may also
contributeto the very large volumesof melt. As the plates
separated,hot mantlewould tend to flow back towardsthe
developingNorth Atlantic rift axis and generate,by forced
convection,the large volumesof magma associatedwith
the SDRS. (This mechanismwould be in additionto any
mantle channellingdirectly along the rift axis from the
plume, but which would be unable to melt if it was
travelling essentiallyhorizontally and isobarically.)We
have already mentioned, in Section 4.1, that forced
convectionis likely in Iceland and reducesthe modelled
temperature of the plume. Note that the hot mantle
associatedwith the developingplume head would cool
rapidly, not only by heat lossthroughconductioninto the
overlyinglithosphere,but alsothroughmagmaproduction.
Supply of new, hot mantlecould not have continuedat the
initial high flux rates.
It is interestingto speculatethat mantlebackflowto the
developingrift axes also providesa mechanismwhereby
the North Sea Basin returned to its normal

Labrador Sea
Opening Mantle
Phase 1 Plume
(62-58 m.y.)
ß Hot mantle plume
ß Mainlycontinent-basedmagmatism
at widespreadlocalities
ß Picritesand high-Mg basaltscommon
Phase 2 Mantle
Plume
(56 m.y. on)
ß Cooler plume
ß Large volumes of melt
producedalong lines of plate separation
ß Plume flows back to rift zones?
Figure 16. Schematicillustration of the developmentof the
Iceland plume during Palaeoceneand Eocene times. Phase 1.
Rapiddevelopment
of a widespread
plume
headanderuption
of
basaltsand picritesover a large area. Sites of eruptionmay be
controlledby pre-existingthinspots(BTIP) or a developingrift
system (B, WG, SEG). Abbreviations:NEG. - Northeast
Greenland; BTIP- British Tertiary Igneous Province; SEG-
Southeast Greenland; WG - West Greenland; B - Baffin Island.
Phase 2. This phase is associatedwith the main rifting and
breakup event, and subsequentactivity forming the Faeroes-
GreenlandRidge and Iceland. By the time of plate separation,
plume flux rates had diminished,the plume head may have
cooled, and activity was focused along the developingrifted
margins. Abbreviations:V - Voting Plateau; G - central East
Greenland; SEG - SoutheastGreenland margin; HB - Hatton
Bank margin.
curve only 7 or 8 m.y. after the thermalpulsearrived;the
hot mantle reachedits maximum extent by about 55 Ma,
then partially retreatedas the main North Atlantic Basin
opened.
The absenceof any obviousplume activity in the North
Atlantic regionprior to 62 Ma suggests that the plumewas
not presentin the region prior to that time, or the plume
was very weak, or the region had not migrated into the
plume's sphereof influence. The plume may have been
SAUNDERS ET AL. 79
presentto the northwestof Greenland in the Cretaceous.
The Alpha Ridge, a region of anomalouslythick crust in
the Arctic Basin,and Cretaceousflood basaltsin the high
Arctic on Axel Heiberg Island indicate some form of
plume activity [Lawvet and Miiller, 1994; Tarduno,1996].
There are, however, at present no geochemicaldata or
radiometricage dateswith which to confirm or refute the
involvement of the ancestral Iceland plume in the
formationof thesemagmaticsubprovinces.
If the mantle flux rates predictedfor the Palaeogene
sequenceshad been operatingthroughoutthe Cretaceous,
we would expect to find voluminousmagmatismin the
Labrador Sea/Davis Strait regions where basin develop-
ment had been operating on and off throughout the
Mesozoic. The only magmatismin this area appearsto be
that associatedwith the Disko-Baffin Island sequences,all
of which are of Palaeocene age, and which erupted
suddenlyat about60-61 Ma. Phase1 magmatismappears,
therefore,to be associatedwith a suddenburst of plume
activity. It is not possibleto say, at present,whether this
activity representsthe arrival and impact of a new plume
[e.g., Richardset aL, 1989; Campbelland Griffiths, 1990],
or a sudden increasein the flux rate of what previously
had been a relatively weak plume system.Although there
is little doubtthat both phasesof activity exploitedlinesof
weakness(e.g., thinspots)in the lithosphere,the location
and natureof the Phase2 magmatismwas controlledto a
far greaterdegreeby the developingrift axesand linesof
plate separation.During Phase 1, magmatismappearsto
havebeenan activeresponseto the plume;.duringPhase2,
which continuesto the presentday, magmatismhasbeena
subtle (and changing) interplay between a passive
upwelling in response to plate separationand active
convection.
7.2. CompositionalStructureof theIceland Plume
Plumes are thought to originate from a thermal
boundarylayer deepwithin the Earth. The preciselocation
of thisboundarylayer is disputedandmay not be the same
for all plumes. Equally contentiousis the role of the
boundary layer in isolating material of different
compositions.At the one extreme,it couldbe arguedthat
plume materialis derivedfrom a boundarylayer that is
compositionallyidentical to the material into which it
ascends.The plumewouldthusbe distinguished solelyon
the basis of its thermal signature (T, rates of melt
generation,buoyancyand gravity effects). However, we
may expect any eruptedmagmasto be compositionally
different(for example,in terms of major elementsand
many trace elementratios) from melts derived from
80 THE NORTH ATLANTIC IGNEOUS PROVINCE
arnbient mantle, simply because the geometry of the
melting column would be affected by an increasein
potential temperature. If both the plume and the
surrounding asthenospheric mantle are internally
homogeneousand have the same bulk composition,we
may reasonably expect, however, the mantle and the
erupted productsto have the same Sr, Nd, Pb, and He
isotopicsignatures.
At the other extreme, the plume may originate in a
boundarylayer that is compositionallydifferentfrom the
asthenosphere into which it rises,or the plumemay entrain
mantle material as it ascends(especiallyduring its initial
stages)[e.g., Griffiths and Campbell, 1990]. In either of
these cases, there is no reason to assume that the
surroundingambient asthenosphereand the plume have
the same isotopic compositions.Indeed, it is the long-
recognisedisotopic diversity of magmas from oceanic
islands that has led to complex, often contradictory,
chemodynamic models of Earth structure, involving
plumes originating from isotopicallydistinct, long-lived
reservoirs[e.g., Zindler and Hart, 1986].
The North Atlantic offers a uniqueopportunityto study
the compositionalstructureof a large mantle plume. Not
only is there the juxtaposition of the plume and a
spreadingridge, thus providing material from the entire
width of the plume head, but there is also excellent
preservationof materialfrom the Palaeogene to the present
day. It is generallyacceptedthat the axis of the plume is
presently located beneath eastern Iceland and that the
thermal effects of this plume extend to approximately
50øN on the MAR [e.g., Whiteet al., 1995, and references
therein].Schillinget al. [1973] and Hart et al. [1973] were
the first to report the geochemicalgradient along the
Reykjanes Ridge - MAR. Various compositional
parameters,such as 87Sr/86Sr,
2ø6pb/2ø4pb,
and La/Sm,
correlate inversely with water depth, and these workers
interpretedthe correlationsas a resultof mixing between
an 'enriched' Iceland plume and a 'depleted' asthenos-
phere(MORB mantle).As noted above,the extentof the
thermal effects of the plume is greater than that of the
elemental and isotopic effects, implying some form of
decouplingwithin the plume structure.The distal part of
the plume, between62ø and 54øN on the MAR, appearsto
comprise hotter-than-ambient mantle (with high
fractionation-correctedFeO values [Klein and Langmuir,
1987]), but which also possesses MORB-like isotopeand
trace element ratios.
As mentioned in Section 4.1, the Iceland plume is
isotopically heterogeneous, producing a range of
'depleted' and 'less depleted' magma types. This
'depleted' end-memberis not the same as the MORB
source,and to minimise confusionwe call it the depleted
plume end-member.Thirlwall et al. [1994] and Thirlwall
[1995] noted that all analysedbasaltsfrom Iceland have
distinctlyhigher2øSpb/2ø4pb
at a given 2ø7pb/2ø4pb
than
North Atlantic MORB. The least crustally contaminated
Palaeocenebasaltsfrom Hold with Hope also show this
characteristic,leading Thirlwall et al. [1994] to suggest
that the plume can be 'fingerprinted'.On the basisof these
data Thirlwall [ 1995], Harris et al. [ 1995] and Kerr et al.
[1995b] also suggestedthat the depleted end-member
found in Icelandic basalts originates from within the
plume. Thus any model of the Iceland plume structure
needsto considerthree components:(i) the depletedand
(ii) the less depletedplume end-members,and (iii) the
heated, adjacent asthenosphere(usually the MORB
source). Can we detect all three componentsin the
Palaeogene basalts?
The incompatible-element-depleted nature of uncon-
taminatedHebridean,Hatton Bank, and FaeroesTertiary
lavashas long beenrecognised[Carter et al., 1979; Wood,
1979b; Morrison et al., 1980; Ridley, 1973; Thompsonet
al., 1980; Garidpy et al., 1983] (see, for example,Figures
8, 11, and 12). Morrison et al. [1980] suggestedthat a
small-degreemelt fraction had been extractedfrom the
mantle sourceregion of the Hebridean lavas during the
Permo-Carboniferous, and that remelting of this
Palaeozoic residuum produced the depleted Tertiary
basalts.Subsequentanalysisof the Nd isotoperatios of
British Tertiary IgneousProvince basalts[Thirlwall and
Jones, 1983; Kerr et al., 1995a] has shown, however, that
someuncontaminatedlavaspossesseNdt=60ashigh as+9.
Thus, the Hebridean basalts have a long history of
depletion which is inconsistentwith a single-stage
depletionevent as recentas the Permo-Carboniferous. The
sameproblemexistsfor Wood's[1979b] dynamicmelting
model, which advocatedthat the melting and extraction
episodeswere closein time.
Although data for many basalticsuitesplot to the low-
end side of the Iceland array on Figure 7, indicating
variableamountsof lithosphericcontamination, many data
plot on the array or projectback to it. The implicationof
thesedata is that the majority of the primary meltswere
originallyderivedfrom sourceswith end andZr/Nb lying
on or closeto the lceland-MORB array. Basaltsfrom the
Faeroes, West Greenland, and Site 918 on the SE
Greenlandmargin encompass much of the Icelandarray.
The Mull Plateau Lavas and the Group 1 basaltsfrom
central East Greenlandmay have tapped sublithospheric
mantle with high Zr/Nb (-25) and end (-+9) whereas
basaltsbelongingto Group2 from centralEastGreenland
tappedless depletedmantle with a lower Zr/Nb

and end (-+7). The data shownon Figure 7 are consistent
with the ultimate derivationof the majority of the basalts
from mantlewith an Icelandicsignature.
The two exceptionsare the light-REE-depletedbasalts
from Hutton Bank and from the Fieroes Upper Series.
Both setsof lavas have high Zr/Nb, far abovethe Iceland
array. Thesebasaltsappearto have been derivedfrom hot
MORB mantle (the elevatedtemperaturesbeing necessary
for the formationof the SDRS andthe dynamicsupportof
the margin during rifting). The Hutton Bank basaltsare
'among the farthest from any putative plume axis,
suggestingthat the plume had a MORB-like annulusin
Eocenetimes,much as it appearsto do today [Fitton et al.,
1996b, 1997]. This may be overly simplistic,however,
becausethe Fieroes basalts were erupted closer to the
ancestralplume axis, implying that MORB mantle was
partially entrainedwithin the body of the plume, and not
just formingan annularrim.
These observationsconfirm the work by Fitton et al.
[1996b, 1997], who have deviseda discriminantbetween
Icelandic basalts and MORB using Nb/Y versus Zr/Y
(Figure 17). This diagramrelieson the unusualdistribution
of Nb in MORB and ocean island basalts. In ocean island
basalts,Nb is invariablyenrichedrelativeto Zr. When data
for basaltsfrom the Iceland neovolcaniczonesare plotted
on this diagram, they fall within a tightly defined array.
Data for IcelandicTertiary basaltsform a tight clusterbut
are still within the array.All normalMORB for which we
havereliabledatafall belowthe array(enriched,or E-type,
MORB projectfrom normalMORB towardsthe array,but
are not plotted here). The discriminantis less sensitive
than Pb isotopesto the effects of crustal contamination,
becausecrust and MORB plot on the same side of the
array(i.e., towardslow Nb/Y), andthe Nb/Pb ratio of most
continentalcrustis significantlylower than that of mantle-
derivedliquids.
The diagramdemonstrates that basaltsfrom Site 918 on
the SE Greenlandmarginplot at the 'depleted'end of the
Iceland array, despite being light-REE-depletedand
otherwiseresemblingnormal MORB. This is the same
conclusionas that drawn from Figure 7. Light-REE-
depletedbasaltsfrom the Upper Serieson the Fieroes and
from Hutton Bank fall within the MORB field, however,
suggestingthat the mantle sourcefor thesebasaltsis not
Icelandic.Otherbasaltsfrom the Fieroe Islandsplot in or
closeto the Iceland field; the slight displacement below
the lower part of the array shownby someof the Middle
Seriesbasaltsis probablya resultof crustalcontamination.
Light-REE-depleted samples from West Greenland, a
regionalsoon the peripheryof the Palaeocene plume,also
plot below the Icelandarray, indicatingderivationfrom a
SAUNDERS ET AL. 81
1ø
I
/ Middle
Series)
e• ß$) eye• ?•• •'• SEGreenland
I (Site
918)
• • • Hutton
Bank
I Fieroes (depleted
0.01 Upper
Series)
Zr/Y
Figure 17. Nb/Y versus Zr/Y in basalts from the Iceland
neovolcaniczones,normal (N-) MORB, SE Greenlandmargin,
Hutton Bank, and the Fieroes Islands. Modified after, and data
from, Fitton et al. [1996b, 1997] and Garidpy et al. [1983].
'Depleted' and 'enriched' refer to basalts from the Iceland
neovolcanic zones that have low Zr/Y and high Zr/Y,
respectively.Note that the Iceland data define a tight array,
whereasN-MORB fall below this array, despitehaving similar
REE profiles to depletedIcelandicpicrites(see, for example,
Figure 4). PM - primordialmantle,from Sun and McDonough
[]989].
MORB-like source[Fitton eta/., 1997b,c]
The data for the Icelandicand the Palaeogenesequences
provide strong circumstantialsupportfor the involvement
of a heterogeneousmantle plume in the formation of the
North Atlantic IgneousProvince.Heating of MORB-like
asthenospherealone will not produce the diversity of
isotopiccompositions,when the effectsof the lithosphere
(pressurecontrol and contamination)are accountedfor.
Although far more data are required to constrain
speculationsabout the nature of the plume, it would
appear,on the basisof the availabledata,that the plume is
intrinsicallyheterogeneous, with both 'less depleted'and
'depleted'components. Either or bothof thesecomponents
may have originatedfrom the lower mantle [e.g., Hart et
al., 1992]. Furthermore,the plume has an annulusof hot,
depletedMORB mantle that appearsto have been the
sourceof magmaseruptedon the peripheryof the plume
[Fitton et al., 1997]. Whether such an annulusexists in
other plumes is difficult to ascertain;normally, such
material would be trapped beneath lithosphereand be
unableto readily
82 THE NORTH ATLANTIC IGNEOUS PROVINCE
8. MAIN CONCLUSIONS AND SUGGESTIONS
FOR FUTURE WORK
ß The North Atlantic IgneousProvince is one of the
largest LIPs, occurring in places as far apart as Baffin
Island and West Greenland in the west, the British Isles in
the east, and Hold with Hope in NE Greenlandin the
north, a distance of more than 2000 km. The total volume
of Palaeoceneand Eocenemagmasprobablyexceeded6 x
106km3 andmaybe asmuchas 10 x 106km3.
ß The bulk of the activity occurredin two main phases.
Phase 1 began about 62 m.y. ago, and lastedfor 2 to 4
m.y; most of the magmas associatedwith Phase 1 were
eruptedthroughand onto continentalcrust.Phase2 began
at about 56 Ma with the onset of plate breakup and
separationand continuesat the presenttime in Iceland.No
rocksdatedby 4øAr-39Ar
methodsand with reliableage
spectrahave agesgreaterthan about62 Ma.
ß The main product of the activity is tholeiitic basalt,
althoughalkali basaltsare common.Differentiatedrocks
are frequently found on continentalareasand along the
rifted margins.
ß The lavas eruptedon Iceland, especiallythosein the
neovolcaniczones, attest to the heterogeneous nature of
the underlying mantle. This heterogeneityappearsto be
replicatedthroughoutthe province.Just as the Icelandic
basaltsdo not originate from a simple, one-component
sourcebut originate from depletedand enrichedmantle
end-members,so the Palaeogenebasaltsare derivedfrom a
spectrum of sublithosphericmantle compositions.The
compositionof the sublithosphericmelt dependson the
conditionsof melting and may have been influencedby
the distancefrom the ancestralplume axis.
ß There is good evidence that the lithospherehas
strongly influencedthe compositionof the eruptedrocks,
both by truncating the melting column and by
contaminatingmany of the magmasas they ascendedto
the surface. The continental crust has been the main
contaminant, but the evidence for involvement of
continentallithosphericmantle is equivocal.The fact that
either could be involved as a contaminant, however,
suggeststhat caution shouldbe exercisedwhen inversion
modellingof REE data is undertaken.
ß The thick oceanic crust associated with Iceland
the Greenland-FaeroesRidge is convincingevidencethat a
thermal anomaly, presumably a mantle plume, resides
beneathIceland. The precisetemperatureof the plume is
not known; estimatesrange from about 70ø to 200øC
above ambient temperature. Estimatesbased solely on
passiveupwellingof mantlein responseto plateseparation
must be treated with caution, because forced convection
will reduce the temperature required to produce the
calculated melt volumes.
ß We envisage that the activity of Phase 1 was
associated with a transient,very hot, low-viscosityplume.
This would account for the simultaneouseruption of
basaltsand picrites in Greenland,Canadaand the British
Isles(a distanceof more than 2000 km). Very high mantle
flux rates were probably associatedwith this transient
event,which may representthe arrival of a new plume, or
a suddenincreasein flux rates of a pre-existingweaker
plume. Channelling of the hot mantle into pre-existing
thinspotsmay have restricted magmatismto localised
areas.
ß Phase2 magmatismwas linked with plate separation,
andmay well have tappedmantlematerialthathad already
lost substantial heat via conduction and via the melt
generationassociated
with Phase1. The bulk of the SDRS
and mid- and lower-crustal level intrusive rocks of the
volcanic rifled margins were producedduring this time.
Magma supplyrates were possibly2 to 3 times that in
present day Iceland, suggestingthat forced convection
operatedduring the time of initial rifting. Flow of the
plume mantle into the rift axis from beneaththe adjacent
subcontinentalareasmay have been an importantprocess,
becausethis would have enhancedthe convectionprocess.
ßThe plumemay havebeencompositionally zoned.The
recoveryof light-REE-depletedbasaltsfrom HattonBank
on the southeasttip of the Rockall Plateau,the furthest
point from the Palaeocene/Eocene plume axis, suggests
that the plumemay havehad an annulusof heatedMORB-
type mantle. This MORB-type mantle was hotter than
ambient, given the large volumes of melt to form the
SDRS, and the dynamic supportof the margin (the lavas
were eruptedcloseto or abovesea-level).
ßAlthoughthere is a largeamountof publisheddataand
text on the North Atlantic IgneousProvince,thereare still
several important gaps in our knowledge that make
definitive assessment difficult. Reliable agesfor the basalt
successionsin central East Greenland, the FaeroesIslands,
Voring Plateau,Skye, Mull and otherScottishIslandsare
absent.Further recovery of samplesfrom the SDRS in
transectscloserto the ancestralplumeaxis(abortedduring
ODP Leg 163) is requiredto determinethe compositional
and gradientsin the Palaeoceneplume. Detailedstudiesof the
Cretaceous-Tertiarysedimentationhistory of the basins
offshoreof Europemay help constrainthe uplift historyof
the region and date more preciselythe fluctuationsin the
flux of the plume.Despitethesegaps,however,the North
Atlantic Igneous Province offers a virtually unique
opportunity to understandthe interactionsbetween a
mantle plume and the lithosphere,and to understand

thermal and compositionalhistoryof the plume itself. It is
among the best, if not the most accessible,of natural
laboratories.
Acknowledgments. In sucha paper,it is not possibleto give
full creditto all workers or to representtheir ideaswith complete
accuracy. We hope that any omissions or inaccuraciesare
consideredas no more than accidental.We have been helped by
thoughtfulandthorough(in somecasesvery detailed)reviewsby
Rick Carlson, Henry Emeleus,Richard England, Lotte Melchior
Larsen, Chip Lesher, John Mahoney, Dan McKenzie, and
Matthew Thirlwall. Researchin the North Atlantic region has
been supportedin part by the NERC, UK (Grants GST/02/673
JGF and ADS, and GT5/F/92/GS/4 to RWK).
REFERENCES
All•gre, C. J., and D. L. Turcotte, Geodynamicmixing in the
mesosphereboundarylayer and the origin of oceanicislands,
Geophys.Res. Lett., 12, 207-210, 1985.
Anderson,D. L., Y.-S. Zhang, and T. Tanimoto, Plume heads,
continental lithosphere, flood basalts and tomography, in
Magmatism and the Causes of Continental Break-up, Spec.
Publ., 68, edited by B.C. Storey, T. Alabaster and R. J.
Pankhurst, pp. 99-124, The Geological Society, London,
1992.
Anderson,F. W., and K. C. Dunham, The Geology of Northern
Skye,HMSO, Edinburgh,1966.
Anderton, R., Sedimentation and basin evolution in the
Paleogeneof the Northern North Sea and Faeroe-Shetland
Basin, in PetroleumGeologyof NorthwestEurope, editedby
J. R. Parker,p 31 The GeologicalSociety,London,1993.
Arndt, N. T., and U. Christensen,Role of lithosphericmantle in
continentalvolcanism:thermalandgeochemicalconstraints, d.
Geophys.Res., 97, 10,967-10,981, 1992.
Arter, G., and S. W. Fagin, The Fleetwood Dyke and the
Tynwald fault zone, Block 113.27, East Irish Sea Basin, in
Petroleum Geology of Northwest Europe, edited by J. R.
Parker,pp. 835-843, The GeologicalSociety,London, 1993.
Athavale, R. N., and P. V. Sharma,Palaeomagneticresults on
early Tertiary lava flows from West Greenland and their
bearing on the evolution history of Baffin Bay-LabradorSea
region., Can. d. Earth Sci., 12, 1-18, 1975.
Backman, J., Cenozoic calcareousnannofossilbiostratigraphy
from the northeasternAtlantic Ocean - Deep Sea Drilling
ProjectLeg 81, Init. Repts.Deep Sea Drill. Proj., 81, 403-428,
1984.
Bailey, E. B., The desertshoresof the Chalk Seas, Geol. Mag.,
61, 102-116, 1924.
Bailey, E. B., Co T. Clough, W. B. Wright, J. E. Richey, and G.
V. Wilson, Tertiary and post-Tertiarygeology of Mull, Loch
Aline and Oban, Geol. Surv. G. B. Mem. Geol. Surv. Scotl.,
1924.
Balkwill, H. R., Labrador Basin: structural and stratigraphic
style, in SedimentaryBasinsand Basin-FormingMechanisms,
Mere. 12, edited by C. Beaumontand A. J. Tankard, pp. 17-
43, CanadianSocietyof PetroleumGeologists,Calgary, 1987.
SAUNDERS ET AL. 83
Barrat, J. A., and R. W. Nesbitt, Geochemistryof the Tertiary
volcanism of Northern Ireland, Chem. Geol., 129, 15-38,
1996.
Barton, A. J., and R. S. White, The Edoras Bank margin;
continentalbreak-up in the presenceof a mantle plume, d.
Geol. Soc. Lond., 152, 971-974, 1995.
Beblo, M., and A. Bj6rnsson,A model of electrical resistivity
beneath NE-Iceland, correlation with temperature, d.
Geophys.,47, 184-190, 1980.
Beckinsale, R. D., R. N. Thompson, and J. J. Durham,
Petrogenetic
significance
of initial87Sr/86Sr
ratiosintheNorth
Atlantic Tertiary igneousprovincein the light of Rb-Sr, K-Ar
and180 abundance
studies
of theSarqfita
qfiqfiintrusive
complex, Ubekendt Island, West Greenland, d. Petrol., 15,
525-538, 1974.
Bell, B. R., and I. T. Williamson, Picritic basalts from the
Palaeocenelava field of west-centralSkye, Scotland;evidence
for parental magma compositions,Min. Mag., 58, 347-356,
1994.
Berggren, W. A., D. V. Kent, and J. J. Flynn, Jurassicto
Paleogene: Part 2. Paleogene geochronology and
chronostratigraphy,in The Chronology of the Geological
Record,Spec.Publ. 1O, editedby N.J. Snelling,pp. 141-195,
The GeologicalSociety,London, 1985.
Berggren,W. A., D. V. Kent, C. C. SwisherIII, andM.-P. Aubry,
A revisedCenozoicgeochronology and chronostratigraphy,in
Geochronology, Time Scales and Global Stratigraphic
Correlation, Spec.Publ. 54, editedby W. A. Berggren,D. V.
Kent, M.-P. Aubry and J. Hardenbol,pp. 129-212, (SEPM)
Societyfor SedimentaryGeology,Tulsa,OK, 1995.
Bjarnasson,
I. T., W. Menke,•. G. F16venz,
andD. Caress,
Tomographic image of the mid-Atlantic plate boundary in
southwesternIceland,d. Geophys.Res., 98, 6607-6622, 1993.
Black, G. P., The agerelationshipof the granophyreandbasaltof
Orval, Isle of Rhum, Geol. Mag., 89, 106-112, 1952.
Blichert-Toft, J., C. E. Lesher, and M. T. Rosing, Selectively
contaminated magmas of the Tertiary East Greenland
macrodikecomplex, Contrib. Mineral. Petrol., 110, 154-172,
1992.
Bollingberg, H., C. K. Brooks, and A. Noe-Nygaard, Trace
element variations in Faeroese basalts and their possible
relationshipsto oceanfloor spreadinghistory,Bull. Geol. Soc.
Denmark, 24, 55-60, 1975.
Bott, M.P. H., The crust beneaththe Iceland-FaeroeRidge, in
Structureand Developmentof the Greenland-Scotland Ridge.
New Methods and Concepts, edited by M. H. P. Bott, S.
Saxov, M. Talwani and J. Thiede, pp. 63-75, Plenum Press,
New York and London, 1983.
Bott, M. H., J. Sunderland, U. Casten, and S. Saxov, Evidence
for continental crust beneath the Faeroe Islands, Nature, 248,
202-204, 1974.
Boutilier, R. R., and C. E. Keen, Models of small scale
convection and melting below divergent plate boundaries
(abstract),Eos Trans.AGU, F546, 1995.
Bowen, N. L., The Evolution of the Igneous Rocks. Princeton
University Press,1928.
Brodie, J., and J. G. Fitton, Compositionof basalticlavas
84 THE NORTH ATLANTIC IGNEOUS PROVINCE
the seaward-dippingreflector sequencerecovered during
DSDP Leg 81 (Hatton Bank), Proc. Ocean Drill. Prog., Sci.
Results,152, in press,1996.
Brooks, C. K., Tertiary of Greenland- a volcanic and plutonic
recordof continentalbreak-up,Mern.Arner.Ass.Petrol. Geol.,
19, 150-160, 1973a.
Brooks,C. K., Rifting and doming in southerneastGreenland,
Nature, 244, 23-25, 1973b.
Brooks, C. K., and T. F. D. Nielsen, The Phanerozoic
developmentof the Kangerdlugssuaqarea, East Greenland,
Medd. Gronl., Geosci.,9, 30 pp., 1982a.
Brooks, C. K., and T. D. F. Nielsen, The E Greenlandcontinental
margin: a transition between oceanic and continental
magmatism,d. Geol. Soc.Lond., 139, 265-275, 1982b.
Brooks,C. K., L. M. Larsen,and T. F. D. Nielsen, Importanceof
iron-rich tholeiitic magmas at divergent plate margins: a
reappraisal,Geology,19, 269-272, 1991.
Campbell, I. H., and R. W. Griffiths, Implications of mantle
plume structure for the evolution of flood basalts, Earth
Planet. Sci. Lett., 99, 79-93, 1990.
Cande, S.C., and D. V. Kent, A new geomagneticpolaritytime
scalefor the Late Cretaceousand Cenozoic,d. Geophys.Res.,
97, 13,917-13,951, 1992.
Cande, S.C., and D. V. Kent, Revised calibration of the
geomagneticpolarity timescalefor the late Cretaceousand
Cenzoic,d. Geophys.Res., 100, 6093-6095, 1995.
Carter, S. R., N.M. Evensen, P. J. Hamilton, and R. K. O'Nions,
Neodymium and strontium isotope evidence for crustal
contamination of continental volcanics, Science, 202, 743-
747, 1978.
Carter, S. R., N.M. Evensen, P. J. Hamilton, and R. K. O'Nions,
Basalt magma sources during the opening of the North
Atlantic, Nature, 281, 28-31, 1979.
Casten, U., The crust beneath the Faeroe Islands, Nature, 241,
83-84, 1973.
Casten, U., and P. H. Nielsen, Faeroe Islands- a microcontinental
fragment?d. Geophys.,41,357-366, 1975.
Chalmers, J. A., New evidence on the structure of the Labrador
SeWGreenlandcontinentalmargin, d. Geol. Soc. Lond., 148,
899-908, 1991.
Chalmers,J. A., L. M. Larsen, and A. K. Pedersen,Widespread
Palaeocene volcanism around the northern North Atlantic
Labrador Sea: evidencefor a large, hot, early plume head,d.
Geol. Soc. Lond., 152, 965-969, 1995.
Clarke, D. B., Tertiary basaltsof Baffin Bay: Possibleprimary
magma from the mantle, Contrib. Mineral. Petrol., 25, 203-
224, 1970.
Clarke, D. B., The Tertiary volcanic province of Baffin Bay,
Geol. Ass. Canada, Spec.Paper, 16, 445-460, 1977.
Clarke, D. B., and M. J. O'Hara, Nickel, and the existence of
high-MgO liquids in nature,Earth Planet. Sci. Lett., 44, 153-
158, 1979.
Clarke, D. B., and A. K. Pedersen,Tertiary volcanicprovinceof
Baffin Bay, in Geologyof Greenland,editedby A. Escherand
W. S. Watts, pp. 365-385, GeologicalSurvey of Greenland,
Copenhagen,1976.
Clarke, D. B. and B. G. J. Upton, Tertiary basaltsof Baffin
Island: field relationsand tectonicsetting,Can. d. Earth Sci.,
8, 248-258, 1971.
Clarke, D. B., G. K. Muecke, and G. Pe-Piper,The lamprophyres
of Ubekendt Island, West Greenland: productsof renewed
partial melting or extremedifferentiation?,Contrib.Mineral.
Petrol., 83, 117-127, 1983.
Clift, P. D., J. Turner, and Ocean Drilling Program Leg 152
ScientificParty, Dynamic supportby the Icelandicplume and
verticaltectonicsof the northeastAtlantic continentalmargins,
J. Geophys.Res., 100, 24,473-24, 846, 1995.
Coffin, M. F. and O. Eldholm, Large igneousprovinces:crustal
structure, dimensions, and external consequences,Rev.
Geophys.,32, 1-36, 1994.
Courtillot, V., G. F6raud, H. Maluski, D. Vandamme, M. G.
Moreau, and J. Besse, Deccan flood basalts and the
Cretaceous/Tertiaryboundary,Nature, 333, 843-846, 1988.
Dagley, P., and A. E. Mussett,Palaeomagnetism andradiometric
dating of the British Tertiary Igneous Province:Muck and
Eigg, Geophys.d. R. Astron.Soc.,85, 221-242, 1986.
Deer, W. A., Tertiary igneousrocks betweenScoresbySundand
Kap Gustav Holm, East Greenland,in Geologyof Greenland,
editedby A. Escherand W. S. Watt, pp. 404-429, Geological
Surveyof Greenland,Copenhagen,1976.
DePaolo, D. J., Trace element and isotopiceffectsof combined
whole rock assimilationand fractional crystallization,Earth
Planet. Sci. Lett., 53, 189-202, 1981.
Dickin, A. P., Isotope geochemistryof Tertiary igneousrocks
from the Isle of Skye, N.W. Scotland,d. Petrol., 22, 155-189,
1981.
Dickin, A. P., The North Atlantic Tertiary Province, in
ContinentalFlood Basalts, edited by J. D. Macdougall,pp.
111-149, Kluwer Academic Publishers, Dordrecht,
Netherlands, 1988.
Dickin, A. P., and N. W. Jones,Isotopicevidencefor the age and
origin of pitchstonesand felsites,Isle of Eigg, NW Scotland,
d. Geol. Soc. Lond., 140, 691-700, 1983.
Dickin, A. P., S. Moorbath and W. H.J., Isotope,trace element
and major element geochemistryof Tertiary igneousrocks,
Isle of Arran, Scotland,Trans. Roy. Soc. Edin., 72, 159-170,
1981.
Dickin, A. P., N. W. Jones, M. F. Thirlwall, and R. N.
and Thompson,A Ce/Nd isotope study of crustalcontamination
processesaffectingPalaeocenemagmasin Skye, Northwest
Scotland, Contrib. Mineral. Petrol., 96, 455-464, 1987.
Duncan, R. A., H. C. Larsen, J. Allen, et al., Proc. Ocean Drill.
Prog., Init. Repts.,163, 296 pp., 1996.
Eldholm, O., J. Theide, and E. Taylor, Proc. OceanDrill. Prog.,
Init. Repts., 104, 783 pp., 1987.
Eldholm, O., and K. Grue, North Atlantic volcanic margins:
dimensionsand productionrates,d. Geophys.Res., 99, 2955-
2968, 1994.
Ellam, R.M., Lithosphericthicknessas a control on basalt
geochemistry,Geology,20, 153-156, 1992.
Elliott, T. R., C. J. Hawkesworth, and K. Gr6nvold, Dynamic
meltingof the Icelandplume,Nature, 351,201-206, 1991.
Emeleus,C. H., The centralcomplexes,in IgneousRocksof the
British Isles, edited by D.S. Sutherland,pp. 369-414,

Wiley & SonsLtd., 1982.
Emeleus,C. H., The Tertiary lavas and sedimentsof north-west
Rhum, Inner Hebrides,Geol. Mag., 122, 419-437, 1985.
Emeleus, C. H., The Rhum layered complex, Inner Hebrides,
Scotland,in Origins of IgneousLayering,editedby I. Parsons,
pp. 263-286, Reidel, Dordrecht, 1987.
Emeleus,C. H., Tertiary igneousactivity,in Geologyof Scotland,
editedby G. Y. Craig, pp. 455-502, The GeologicalSociety,
London, 1991.
Emeleus, C. H., Geology of Rum and the adjacent Islands,
Memoir, British GeologicalSurvey,in press,1997.
England,R. W., The early Tertiary stressregime in NW Britain:
evidence from the patterns of volcanic activity, in Early
Tertiary Volcanismand the Openingof the NE Atlantic, Spec.
Publ. 39, edited by A. C. Morton and L. M. Parson,pp. 381-
389, The GeologicalSociety,London, 1988.
England, R. W., The structureof the Skye lava field, Scott. d.
Geol., 30, 33-37, 1994.
England,R. W., R. W. H. Butler, and D. H. W. Hutton, The role
of Paleocenemagmatismin the Tertiary evolutionof basinson
the NW seaboard,in PetroleumGeologyof NorthwestEurope,
pp. 97-105, The GeologicalSociety,London, 1993.
Esson,J., A. C. Dunham, and R. N. Thompson,Low alkali, high
Ca, olivine tholeiite lavas from the Isle of Skye, Scotland,d.
Petrol., 16, 488-497, 1975.
Evans,A. L., F. J. Fitch, and J. A. Millar, Potassium-argon age
determinationson some British Tertiary igneous rocks, d.
Geol. Soc. Lond., 129, 419-443, 1973.
Fitch, F. J., G. L. Heard, and J. A. Miller, Basalticmagmatismof
late Cretaceousand Palaeogeneage recordedin wells NNE of
the Shetlands,in Early Tertiary Volcanismand the Openingof
theNE Atlantic, Spec.Publ. 39, editedby A. C. Morton and L.
M. Parson, pp. 253-262, The Geological Society, London,
1988.
Fitton, J. G., and H. M. Dunlop, The Cameroon line, West
Africa, and its bearingon the origin of oceanicand continental
alkali basalt,Earth Planet. Sci. Lett., 72, 23-38, 1985.
Fitton, J. G., A.D. Saunders,L. M. Larsen, M. S. Fram, A.
Demant, C. Sinton, and Leg 152 Scientific Party, Magma
sourcesand plumbing systemsduring break-up of the SE
Greenlandmargin: resultsfrom ODP Leg 152, or. Geol. Soc.
Lond., 152, 985-990, 1995.
Fitton, J. G., B. S. Hardarson,R. M. Ellam, and G. Rogers,Sr-,
Nd-, and Pb-isotopiccompositionof volcanicrocks from the
southeastGreenland margin at 63øN: temporal variation in
crustal contaminationduring continentalbreak-up, in Proc.
OceanDrill. Prog., Sci. Results,152, in press,1996a.
Fitton, J. G., A.D. Saunders,L. M. Larsen, B. S. Hardarson,and
M. J. Norry, Volcanic rocksfrom the East Greenlandmargin
at 63øN: composition,petrogenesisand mantle sources,in
Proc. OceanDrill. Prog., Sci. Results,152, in press,1996b.
Fitton, J. G., A.D. Saunders,M. J. Norry, and B. S. Hardarson,
Thermal and chemicalstructureof the Icelandplume, Earth
Planet. Sci. Lett., in press,1997.
Foley, S. F., G. Venturelli, D. H. Green, and L. Toscani, The
ultrapotassic rocks: characteristics, classification, and
constraintsfrom petrogeneticmodels,Earth-Sci.Rev., 24, 84-
SAUNDERS ET AL. 85
134, 1987.
Fowler, S. R., R. S. White, G. D. Spence,and G. D. Westbrook,
The Hatton Bank continentalmargin II. Structurefrom two-
ship expanding spread profiles, Geophys.d., 96, 295-309,
1989.
Fram, M., and C. E. Lesher, Geochemical constraintson mantle
melting during creation of the North Atlantic basin, Nature,
363, 712-715, 1993.
Fram, M. S., and C. E. Lesher, Polybaric differentiation of
magmas of the East Greenland early Tertiary flood basalt
province,•/. Petrol., in press,1996.
Fram, M. S., C. E. Lesher, and A.M. Volpe, Mantle melting
systematics: the transition from continental to oceanic
volcanismon the southeastGreenlandmargin, in Proc. Ocean
Drill. Prog., Sci. Results,152, in press,1996.
Francis, D., The Baffin Bay lavas and the value of picrites as
analoguesof primary magmas,Contrib. Mineral. Petrol., 89,
144-155, 1985.
Furman, T., F. Frey, and K.-H. Park, The scale of source
heterogeneitybeneaththe Easternneovolcaniczone, Iceland,
d. Geol. Soc. Lond., 152, 997-1002, 1995.
Gallagher,K., and C. J. Hawkesworth,Dehydrationmeltingand
the generationof continentalflood basalts,Nature, 358, 57-
59, 1992.
Gari6py,C., J. Ludden,and C. Brooks,Isotopicandtraceelement
constraintson the genesis of the Faeroe lava pile, Earth
Planet. Sci. Lett., 63, 257-272, 1983.
Gibson,I. L., and A.D. Gibbs,Accretionaryvolcanicprocesses
and the crustalstructureof Iceland, Tectonophysics,
133, 57-
64, 1987.
Giekie, A., On the Tertiary volcanicrocks of the British Islands,
Geol. Mag., 4, 316-319, 1867.
Giekie, A., The lava fields of North-WesternEurope.Nature, 23,
3-5, 1880.
Giekie, A., The history of volcanic action during the Tertiary
period in the British Isles, Trans.Roy. Soc.Edin., 35, 21-184,
1888.
Gill, R. C. O., T. F. D. Nielsen,C. K. Brooks,and G. A. Ingram,
Tertiary volcanism in the Kangerdlugssuaqregion, E
Greenland:trace-elementgeochemistryof the Lower Basalts
and tholeiitic dyke swarms,in Early Tertiary Volcanismand
the Openingof the NE Atlantic, Spec.Publ. 39, editedby A.
C. Morton and L. M. Parson,pp. 161-179, The Geological
Society,London, 1988.
Gill, R. C. O., A. K. Pedersen,and J. G. Larsen,Tertiarypicrites
from West Greenland:melting at the peripheryof a plume?in
Magmatism and the Causesof ContinentalBreak-up, Spec.
Publ. 68, edited by B.C. Storey, T. Alabaster and R. J.
Pankhurst, pp. 335-348, The Geological Society, London,
1992.
Goldschmidt-Rokita, A., K. F. J. Hansch, H. B. Hirschleber, T.
Iwasaki, T. Kanazawa, H. Shimarura,and M. A. Sellevol, The
ocean/continenttransitionalong a profile throughthe Lofoten
Basin, northernNorway, Mar. Geophys.Res., 16, 201-224,
1994.
Gradstein,F. M., F. P. Agterberg,J. G. Ogg, J. Hardelbol,P. van
Veen, J. Thierry, and Z. Huang, A Mesozoic time scale, o
86 THE NORTH ATLANTIC IGNEOUS PROVINCE
Geophys.Res.,99, 24,051-24,074, 1994.
Griffiths, R. W., and I. H. Campbell, Stirring and structurein
mantle startingplumes, Earth Planet. Sci. Lett., 99, 66-78,
1990.
Haigh, B. I. R., North Atlantic oceanictopographyand lateral
variationsin the uppermantle,Geophys.d. R. ,4stroh.Soc.,33,
405-420, 1973.
Hald, N., and A. K. Pedersen,Lithostratigraphyof the early
Tertiary volcanic rocks of central West Greenland,Rapp.
Gronlands Geol. Unders., 69, 17-24, 1975.
Hald, N., and R. Waagstein, Silicic basaltsfrom the Faeroe
Islands: Evidence of crustal contamination, in Structure and
Developmentof the Greenland-Scotland
Ridge,editedby M.
H. P. Bott, S. Saxov, M. Talwani and J. Thiede, pp. 343-349,
Plenum Press,New York, 1983.
Hald, N., and R. Waagstein,Lithology and chemistryof a 2-km
sequenceof Lower Tertiary tholeiitic lavas drilled on
Suduroy•Faeroe Islands (Lopra-1), in The Deep Drilling
Project 1980-1981 in the Faeroe Islands, edited by O.
Berthelsen,A. Noe-Nygaard and J. Rasmussen,pp. 15-38,
ForoyaFr6dskaparfelag,T6rshavn,1984.
Hampton,C. M., and P. N. Taylor, The age and natureof the
basement of southern Britain: evidence from Sr and Pb
isotopesin some English granites,d. Geol. Soc. Lond., 140,
499-509, 1983.
Hansen,
H.,Rex,D.C.,P.G.Guise,
andC.K.Brooks,
40Ar/39Ar
ages on Tertiary East Greenland basaltsfrom the Scoresby
Sundarea(abstract),Eos Trans.AGU, 74, 625, 1989.
Hardarson, B. S., and J. G. Fitton, Geochemical variation in the
Tertiary basaltsof Iceland,Min. Mag., 58,4,372-373, 1994.
Hardman, M., J. Buchanan, P. Herrington, and A. Carr,
Geochemical modelling of the East Irish Sea Basin: its
influence on predicting hydrocarbontype and quality, in
Petroleum Geology of Northwest Europe, edited by J. R.
Parker,pp. 809-821, The GeologicalSociety,London, 1993.
Hards, V. L., P. D. Kempton, and R. N. Thompson, The
heterogeneous Iceland plume: new insightsfrom the alkaline
basalts of the Snaefell volcanic centre, d. Geol. Soc. Lond.,
152, 1003-1009, 1995.
Harland, W. B., A. V. Cox, P. G. Llewellyn, C. A. G. Pickton,A.
G. Smith, and R. Walters, ,4 Geologic Time Scale, 131 pp.,
CambridgeUniversit3,Press,Cambridge,1982.
Harland, W. B., R. L. Armstrong,A. V. Cox, L. E. Craig, A. G.
Smith, and D. G. Smith,,4 GeologicTimeScale 1989, 263 pp.,
CambridgeUniversityPress,Cambridge,1990.
Hart, S. R., and K. E. Davis, Nickel partitioningbetweenolivine
and silicate melt, Earth Planet. Sci. Lett., 40, 203-219, 1978.
Hart, S. R., J.-G. Schilling, and J. L. Powell, Basaltsfrom Iceland
and along the Reykjanes Ridge: Sr isotope geochemistry,
Nature, 246, 104-107, 1973.
Hart, S. R., E. H. Hauri, L. A. Oschmann, and J. A. Whitehead,
Mantle plumes and entrainment:isotopicevidence,Science,
256, 517-520, 1992.
H6mond, C., N. T. Arndt, and A. W. Hofmann, The
heterogeneousIceland plume: Nd-Sr-O and trace element
constraints,
d. Geophys.Res.,98, 15,833-15850,1993.
Henderson, G., A. Rosenkrantz, and E. J. Schiener, Cretaceous-
Tertiary sedimentaryrocks of West Greenland,in Geologyof
Greenland,editedby A. Escherand W. S. Watt, pp. 340-362,
GeologicalSurveyof Greenland,Copenhagen,1976.
Hill, R. I., Starting plumes and continentalbreak-up,Earth
Planet. ScioLett., 104, 398-416, 1991.
Hinz, K., A hypothesisof terrestrialcatastrophes.
Wedgesof very
thick oceanwarddipping layers beneathpassivecontinental
margins- their origin and palaeoenvironmental significance,
Geol. dahrb., E22, 3-28, 1981.
Hitchen, K., and J. D. Ritchie, Geological review of the West
Shetlandarea, in Petroleum Geology of North WestEurope,
editedby J. Brooks and K. W. Glennie,pp. 737-749, Graham
and Trotman, London, 1987.
Hitchen,K., and J. D. Ritchie, New K-Ar ages,and a provisional
chronology,•for the offshore part of the British Tertiary
IgneousProvince,Scott.d. Geol., 29, 73-85, 1993.
Hofmann, A. W., K. P. Jochum, M. Seufert, and W. M. White,
Nb and Pb in oceanic basalts: new constraints on mantle
evolution, Earth Planet. Sci. Lett., 79, 33-45, 1986.
Hogg, A. J., J. J. Fawcett,J. Gittins, and M.P. Gorton,Cyclical
variation in composition in continental tholeiites of East
Greenland, Can. d. Earth Sci., 26, 534-543, 1989.
Holm, P.M., Nd, Sr and Pb isotopegeochemistryof the Lower
Lavas, E Greenland Tertiary Igneous Province, in Early
Tertiary Volcanismand the Opening of the NE ,4tlantic,Spec.
Publ. 39, editedby A. C. Morton and L. M. Parson,pp. 181-
195, The GeologicalSociety,London, 1988.
Holm, P.M., N. Hald, and T. F. D. Nielsen, Contrasts in
compositionand evolution of Tertiary CFBs between West
and East Greenland and their relations to the establishment of
the Icelandicmantle plume, in Magmatismand the Causesof
ContinentalBreak-up, Spec. Publ. 68, editedby B.C. Storey,
T. AlabasterandR. J. Pankhurst,pp. 349-364, The Geological
Society,London, 1992.
Holm, P.M., R. C. O. Gill, A. K. Pedersen,J. G. Larsen,N. Hald,
T. F. D. Nielsen, and M. F. Thirlwall, The Tertiarypicritesof
West Greenland: contributions from 'Icelandic' and other
sources,Earth Planet. Sci. Lett., 115, 227-244, 1993.
Holmes, A., The basalticrocks of the Arctic region,Min. Mag.,
18, 180-222, 1918.
Huppert, H. E., and R. S. J. Sparks,The fluid dynamicsof a
basalticmagma chamberreplenishedby influx of hot, dense
ultrabasic magma, Contrib. Mineral. Petrol., 75, 279-289,
1980.
Imsland,P., Icelandandthe oceanfloor. Comparison
of chemical
characteristicsof the magmatic rocks and some volcanic
features, Contrib. Mineral. Petrol., 83, 31-37, 1983.
Ito, E., W. M. White, and C. G6pel, The O, Sr, Nd andPb isotope
geochemistryofMORB, Chem.Geol., 62, 157-176, 1987.
Jakobsson,S. P., Chemistry and distributionpatternsof Recent
volcanic rocks in Iceland, Lithos, 5, 365-386, 1972.
Jakobsson,S. P., J. J6nnson,and F. Shido, Petrologyof the
westernReykjanesPeninsula,Iceland,d. Petrol., 19, 669-705,
1978.
Jones,E. J. W., R. Siddall, M. F. Thirlwall, P. N. Chroston,and
A. J. Lloyd, Anton Dohrn seamountand the evolutionof the
RockallTrough,Ocean.Acta, 17, 237-247,

Joron, J.-L., H. Bougault, R. C. Maury, M. Bohn, and A.
Desprairies,Strongly depleted tholeiites from the Rockall
Plateaumargin,North Atlantic:geochemistry andmineralogy,
Init. Repts.Deep SeaDrill. Proj., 81, 783-794, 1984.
Judd, J. W., On the ancient volcanoesof the Highlands and the
relationsof their productsto the Mesozoicstrata,Q. J. Geol.
Soc. Lond., 30, 220-302, 1874.
Judd, J. W., The Tertiary volcanoes of the Western Isles of
Scotland,Q. J. Geol. Soc. Lond., 45, 187-218, 1889.
J•rgensen,T., and N. Mikkelsen, Coccolithsfrom volcanic
sediments(Danian) in Nugssuaq,West Greenland,Bull. Geol.
Soc. Denmark, 23, 225-230, 1974.
Kennedy, W. Q., Trends of differentiationin basaltic magmas,
Am. J. Sci. 25, 239-256, 1933.
Kent, R. W., Magnesian basaltsfrom the Hebrides, Scotland:
Chemicalcompositionand relationshipto the Icelandplume,
J. Geol. SocoLond., 152, 979-983, 1995.
Kent, R. W., M. Storey, and A.D. Saunders,Large igneous
provinces:sites of plume impact or plume incubation?
Geology,20, 891-894, 1992.
Kerr, A. C., Elemental evidence for an enriched small-fraction
melt input into Tertiary Mull basalts,Western Scotland,J.
Geol. Soc. Lond., 150, 763-769, 1993.
Kerr, A• C., Lithospheric thinning during the evolution of
continentalLarge IgneousProvinces:a case study from the
North Atlantic Tertiary Province, Geology, 22, 1027-1030,
1994.
Kerr, A. C., The geochemistryof the Mull-Morvern Tertiary lava
succession,NW Scotland: an assessmentof mantle sources
during plume-relatedvolcanism, Chem. Geol., 122, 43-58,
1995a.
Kerr, A. C., The melting processes
and compositionof the North
Atlantic (Iceland) plume:geochemicalevidencefrom the early
Tertiary lavas,J. Geol. Soc.Lond., 152, 975-978, 1995b.
Kerr, A. C., P. D. Kempton, and R. N. Thompson, Crustal
assimilationduring turbulent magma ascent [ATA]; new
isotopicevidencefrom the Mull Tertiary lava succession,
N.W. Scotland, Contrib. Mineral. Petrol., 119, 142-154,
1995a.
Kerr, A. C., A.D. Saunders,V. L. Hards, J. Tarney, and N.H.
Berry, Depleted mantle plume geochemicalsignatures:No
paradoxfor plumetheories,Geology,23, 843-846, 1995b.
Kirton, S. R., and J. A. Donato, Some buried Tertiary dykes of
Britain and surrounding waters deduced by magnetic
modellingand seismicreflectionmethods,J. Geol. Soc.Lond.,
142, 1047-1057, 1985.
Klein, E. M., and C. H. Langmuir, Global correlationof ocean
ridge basaltchemistrywith axial depthand crustalthickness,
J. Geophys.Res., 92, 8089-8115, 1987.
Knox, R. W. O., and A. C. Morton, Stratigraphicaldistributionof
Early Palaeogenepyroclasticdepositsin the North Sea Basin,
Proc. Yorks. Geol. Soc., 44, 355-363, 1983.
Knox, R. W., and A. C. Morton, The record of early Tertiary N
Atlantic volcanism in sediments of the North Sea Basin, in
Early Tertiary Volcanismand the Openingof the NE Atlantic,
Spec.Publ. 39, edited by A. C. Morton and L. M. Parson,pp.
407-419, The Geological Society,London, 1988.
SAUNDERS ET AL. 87
Kristoffersen,Y., Sea-floor spreadingand the early openingof
the North Atlantic, Earth Planet. Sci. Lett., 38, 273-290, 1978.
Langmuir, C. H., and G. N. Hanson, An evaluationof major
elementheterogeneityin the mantle sourcesof basalts,Phil.
Trans. R. Soc. Lond., A297, 383-407, 1980.
Larsen, H. C., and C. Marcussen, Sill-intrusion, flood basalt
emplacementand deep crustalstructureof the ScoresbySund
region, East Greenland, in Magmatism and the Causes of
ContinentalBreak-up,Spec.Publ. 68, editedby B.C. Storey,
T. Alabasterand R. J. Pankhurst,pp. 365-386, The Geological
Society, London, 1992.
Larsen,H. C., and S. Jakobsd6ttir,Distribution,crustalproperties
and significanceof seawards-dippingsub-basement reflectors
off E Greenland,in Early Tertiary Volcanismand the Opening
of the NE Atlantic, Spec.Publ. 39, editedby A. C. Morton and
L. M. Parson,pp. 95-114, The Geological Society, London,
1988.
Larsen,H. C., C. K. Brooks, J. R. Hopper, T. Dahl-Jensen,A. K.
Pedersen,T. F. D. Nielsen, and field parties, The Tertiary
openingof the North Atlantic: DLC investigationsalong the
eastcoastof Greenland,Rapp. GronlandsGeol. Unders.,165,
106-115, 1995.
Larsen, H. C., Ao D. Saunders,P. D. Clift, et al., Proc. Ocean
Drill. Prog., Init. Repts.,152, 977 pp., 1994.
Larsen,J. G., Transitionfrom low potassiumolivine tholeiitesto
alkali basaltson Ubekendt Ejland, Medd. Gron., 200, 1-42,
1977.
Larsen, L. M., and A. K. Pedersen,Volcanic marker horizons in
the Maligfit Formation on Disko and Nfigssuaq, and
implicationsfor the developmentof the southernpart of the
West Greenlandbasin in the early Tertiary, Rapp. Gronlands
Geol. Unders., 148, 65-73, 1990.
Larsen, L. M., and A. K. Pedersen,Volcanic marker horizons in
the upper part of the Maligfit Formation on easternDisko and
Nuussuaq,Tertiary of West Greenland:syn- to post-volcanic
basin movements,Rapp. Gronlands Geol. Unders., 155, 85-
93, 1992.
Larsen,L. M., and S. W. Watt, Episodicvolcanismduring break-
up of the North Atlantic: evidence from the East Greenland
plateaubasalts,Earth Planet. Sci. Lett., 73, 105-116, 1985.
Larsen, L. M., J. G. Fitton, and M. S. Fram, Volcanic rocks of the
southeastGreenlandmargin in comparisonwith otherpartsof
the North Atlantic Tertiary igneousprovince,in Proc. Ocean
Drill. Prog., Sci. Results,152, in press,1996.
Larsen,L. M., W. S. Watt, and M. Watt, Geology and petrology
of the Lower Tertiary plateau basaltsof the ScoresbySund
region, East Greenland,Bull. Geol. Surv. Green., 157, 1-164,
1989.
Larsen, L. M., A. K. Pedersen,G. K. Pedersen,and S. Piasecki,
Timing and durationof Early Tertiary volcanismin the North
Atlantic: new evidencefrom West Greenland,in Magmatism
and the Causes of Continental Break-up, Spec. Publ. 68,
editedby B.C. Storey,T. Alabasterand R. J. Pankhurst,pp.
321-333, The Geological Society,London, 1992.
Lawver, L. A., and R. D. M•iller, Iceland hotspottrack, Geology,
22, 311-314, 1994.
Lewis, C. A., P. F. Green, A. Carter, and A. J. Hurford,
88 THE NORTH ATLANTIC IGNEOUS PROVINCE
K/T palaeotemperatures throughoutnorthwestEngland:three
kilometresof Tertiary erosion?Earth Planet. Sci. Lett., 112,
132-145, 1992.
Lund, J., Biostratigraphyof interbasalticcoals,in Structureand
Developmentof the Greenland-Scotland Ridge.New Methods
and Concepts,editedby M. H. P. Bott, S. Saxov,M. Talwani,
and J. Thiede, pp. 417-423, Plenum Press,New York and
London, 1983.
Lyle, P., A petrologicaland geochemicalstudyof the Tertiary
basalticrocks of NE Ireland, or. Earth Sci. R. Dub. Soc, 2, 137-
152, 1980.
Lyle, P., The petrogenesis
of Tertiary basalticand intermediate
lavas of NE Ireland, Scott. d. Geol, 21, 71-84, 1985.
Macdougall, J. D. (Ed.), ContinentalFlood Basalts, 341 pp.,
Kluwer Academic Publishers,Dordrecht,Netherlands, 1988.
Macintyre,R. M. and P. J. Hamilton, Isotopicgeochemistry of
lavas from Sites 553 and 555, Init. Repts. Deep Sea Drill.
Proj., 81, 775-781, 1984.
Mahoney, J. J., Deccan Traps, in ContinentalFlood Basalts,
editedby J. D. Macdougall,pp. 151-194, Kluwer Academic
Publishers,Dordrecht, Netherlands, 1988.
Mattey, D. P., I. L. Gibson,G. F. Marriner,andR. N. Thompson,
The diagnosticgeochemistry,relative abundanceand spatial
distributionof high-Ca,low-alkalitholeiitedykesin the lower
Tertiary regional swarm of the Isle of Skye, NW Scotland,
Min. Mag., 41,273-285, 1977.
McClurg, J. E., Petrologyand evolutionof the northernpart of
the Rhum ultrabasic complex, Unpublished PhD thesis,
University of Edinburgh, 1982.
McDonough,W. F. and S.-s. Sun,The compositionof the Earth,
Chem. Geol., 120, 223-253, 1995.
McDougall, I., L. Kristjansson and K. Saemundsson,
Magnetostratigraphyand geochronologyof NW-Iceland, or.
Geophys.Res.,89, 7029-7060, 1984.
McKenzie, D., Some remarks on the movement of small melt
fractions in the mantle, Earth Planet. Sci. Lett., 95, 53-72,
1989.
McKenzie, D., and M. J. Bickle, The volume and compositionof
melt generatedby extensionof the lithosphere,d. Petrol., 29,
625-679, 1988.
McKenzie, D., and R. K. O'Nions, Partial melt distributionsfrom
inversion of rare earth element concentrations,J. Petrol., 32,
1021-1091, 1991.
McKenzie, D., and R. K. O'Nions, Mantle reservoirsand ocean
island basalts,Nature, 301, 229-231, 1983.
Menke, W. and V. Levin, Cold crust in a hot spot, G. Res. Lett.,
21, 1967-1970, 1994.
Merriman, R. J., P. N. Taylor and A. C. Morton, Petrochemistry
and isotopegeochemistryof early Palaeogene basaltsforming
the dipping reflector sequenceSW of Rockall Plateau,NE
Atlantic, in Early Tertiary Volcanismand the Openingof the
NE Atlantic, Spec.Publ. 39, editedby A. C. Morton and L. M.
Parson,pp. 123-134, The GeologicalSociety,London,1988.
Miller, J. A., D. H. Matthews, and D. G. Roberts, Rocks of
Grenvilleagefrom Rockall Bank,Nature, 246, 61, 1973.
Moorbath, S., and J. D. Bell, Strontium isotope abundance
studies and Rb-Sr age determinationson Tertiary Igneous
rocksfrom the Isle of Skye,NW Scotland,or.Petrol., 6, 37-66,
1965.
Moorbath, S., and R. N. Thompson, Strontium isotope
geochemistry and petrogenesis
of the earlyTertiarylava pile
of the Isle of Skye, Scotland,and other basicrocksof the
British Tertiary Province: an example of magma-crust
interaction,d. Petrol., 21,295-321, 1980.
Moorbath, S., and H. Welke, Lead isotope studieson igneous
rocksfrom the Isle of Skye, NW Scotland,Earth Planet. Sci.
Lett., 5, 217-230, 1969.
Moorbath, S., H. Sigurdssonand R. Goodwin,K-Ar ages of
oldestexposedrocks in Iceland, Earth Planet. Sci. Lett., 4,
197-205, 1968.
Morgan,J. V., P. J. Barton,and R. S. White, The HattonBank
continentalmargin III, Structurefrom wide-angleOBS and
multichannelseismicrefractionprofiles,Geophys.d. Int., 98,
367-384, 1989.
Morgan, W. J., Convectionplumesin the lower mantle,Nature,
230, 42-43, 1971.
Morgan,W. J., Platemotionsanddeepmantleconvection,
Geol.
Soc. Am. Mem., 132, 7-22, 1972.
Morrison, M. A., R. N. Thompson,I. L. Gibson, and G. F.
Marriner, Lateral chemical heterogeneityin the Palaeocene
upper mantle beneath the ScottishHebrides,Phil. Trans.
Royal Soc.Lond., A297, 229-244, 1980.
Morton, A. C., and J. A. Evans, Geochemistryof basalticash
beds from the Fur Formation, Island of Fur, Denmark, Bull.
Geol. Soc. Denmark, 37, 1-9, 1987.
Morton, A. C., and P. N. Taylor, Lead isotopeevidencefor the
structureof the Rockall dipping-reflectorpassivemargin,
Nature, 326, 381-383, 1987.
Morton, A. C., and P. N. Taylor, Geochemicaland isotopic
constraintson the nature and age of basementrocks from
Rockall Bank, NE Atlantic, or. Geol. Soc. Lond., 148, 630-634,
1991.
Morton, A. C., J. E. Dixon, J. G. Fitton, R. M. Macintyre, D. K.
Smythe and P. N. Taylor, Early Tertiary volcanicrocks in
Well 163/6-1A, Rockall Trough, in Early Tertiary Volcanism
and the Openingof the NE Atlantic, Spec.Publ. 39, editedby
A. C. Morton and L. M. Parson,pp. 293-308, The Geological
Society,London, 1988.
Morton, A. C., K. Hitchen, J. D. Ritchie, N.M. Hine, M.
Whitehouse, and S. G. Carter, Late Cretaceousbasaltsfrom
Rosemary Bank, northern Rockall Trough, or. Geol. Soc.
Lond., 152, 947-952, 1995.
Mussett,A. E., Time and durationof Tertiary igneousactivity of
Rhum and adjacentareas,Scott.d. Geol., 20, 273-279, 1984.
Mussett,A. E., 4øAr/39Ar
step-heating
agesof the Tertiary
igneousrocks of Mull, Scotland,or. Geol. Soc. Lond., 143,
887-896, 1986.
Mussett,A. E., P. Dagley, and M. Eckford, The BritishTertiary
IgneousProvince:palaeomagnetism andagesof dykes,Lundy
Island, Bristol Channel, Geophys.d. R. Astron.Soc., 46, 595-
603, 1976.
Mussett,A. E., P. Dagley, and R. R. Skelhorn,Time andduration
of activity in the British Tertiary igneousprovince,in Early
Tertiary Volcanismand the Openingof the NE Atlantic,

Publ. 39, edited by A. C. Morton and L. M. Parson,pp. 337-
348, The Geological Society,London, 1988.
Mutter, J. C., and C. M. Zehnder, Deep crustal structureand
magmaticprocesses:the inceptionof seafloorspreadingin the
Norwegian-Greenlandsea, in Early Tertiary Volcanismand
the Openingof the NE Atlantic, Spec.Publ. 39, editedby A. C.
Morton and L. M. Parson,pp. 34-38, The GeologicalSociety,
London, 1988.
Mutter, J. C., W. R. Buck, and C. M. Zehnder,Convectivepartial
melting. 1. A model for the formation of thick basaltic
sequences during the initiation of spreading,J. Geophys.Res.,
B93, 1031-1048, 1988.
Mutter, J. C., M. Talwani, and P. L. Stoffa, Origin of seaward-
dipping reflectorsin oceaniccrustoff the Norwegian margin
by "subaerial seafloor spreading," Geology, 10, 353-357,
1982.
Nadin, P. A., N.J. Kusznir, and J. Toth, Transientregionaluplift
in the Early Tertiary of the northern North Sea and the
developmentof the Iceland plume, J. Geol. Soc. Lond., 152,
953-958, 1995.
Nakamura,N., Determinationof REE, Ba, Fe, Mg, Na and K in
carbonaceousand ordinary chondrites,Geochim.Cosmochim.
Acta, 38, 757-775, 1974.
Nielsen, P. H., Geology and crustal structure of the Faeroe
Islands - a review, in Structure and Development of the
Greenland-Scotland Ridge. New Methods and Concepts,
editedby M. H. P. Bott, S. Saxov,M. Talwani, and J. Thiede,
pp. 77-86, Plenumpress,New York andLondon, 1983.
Nielsen, T. F. D., and C. K. Brooks, The E Greenland rifted
continentalmargin: an examinationof the coastalflexure, J.
Geol. Soc. Lond., 138, 559-568, 1981.
Nielsen, T. F. D., N.J. Soper,C. K. Brooks,A.M. Faller, A. C.
Higgins,and D. W. Matthews,The pre-basalticsedimentsand
the lower basalts at Kangerdlugssuaq,East Greenland: their
stratigraphy, lithology, palaeomagnetismand petrology,
Medd. Gron., Geosci.,6, 25 pp., 1981.
Noble, R. H., R. M. Macintyre, and P. E. Brown, Age constraints
on Atlantic evolution:timing of magmaticactivity alongthe E
Greenland continental margin, in Early Tertiary Volcanism
and the Opening of the NE Atlantic, Spec.Publ. 39, editedby
A. C. Morton and L. M. Parson,pp. 210-214, The Geological
Society,London, 1988.
Noe-Nygaard,A., Cenozoic to Recent volcanismin and around
the North Atlantic Basin, in The OceanBasinsand Margins,
edited by A. Naim and F. G. Stehli, pp. 391-443, Plenum
Press,New York, 1974.
Noe-Nygaard, A., Tertiary igneousrocks betweenShannonand
ScoresbySund, East Greenland, in Geology of Greenland,
editedby A. Escherand W. S. Watt, pp. 386-402, Geological
Surveyof Greenland,Copenhagen,1976.
Noe-Nygaard,A., and J. Rasmussen,Petrologyof a 3000-metre
sequenceof basaltic lavas in the Faeroe Islands, Lithos, 1,
268-304, 1968.
Old, R. A., The age and field relationshipsof the TardreeTertiary
rhyolite complex,CountyAntrim, N, Ireland,Bull. Geol. Surv.
G. B., 51, 21-40, 1975.
O'Hara, M. J., and R. E. Mathews, Geochemical evolution in an
SAUNDERS ET AL. 89
advancing, periodically replenished, periodically tapped,
continuously fractionated magma chamber, J. Geol. Soc.
Lond., 138, 237-277, 1981.
O'Nions, R. K., P. J. Hamilton and N.M. Evensen, Variations in
143Nd/144Nd
and 87Sr/86Sr
ratiosin oceanicbasalts,Earth
Planet. Sci. Lett., 34, 13-22, 1977.
Oskarsson,N., S. Steinthorsson,and G. E. Sigvaldason,Iceland
geochemical anomaly: origin, volcanotectonics,chemical
fractionationand isotopeevolution of the crust,J. Geophys.
Res., 90, 10,011-10,025, 1985.
Palacz, Z. A., Isotopic and geochemical evidence for the
evolutionof a cyclical unit in the Rhum intrusion,north-west
Scotland, Nature, 307, 618-620, 1984.
Pillmason, G., A continuum model of crustal generationin
Iceland:kinematicaspects,J. Geophys.,47, 7-18, 1980.
Pillmason, G., Model of crustal formation in Iceland, and
applicationto submarinemid-oceanridges, in The Western
North Atlantic Region,GNAM Ser., editedby P. R. Vogt and
B. E. Tucholke, pp. 87-97, Geological Society of America,
Boulder, CO, 1986.
Pillmason, G., and K. Saemundsson,Iceland in relation to the
Mid-Atlantic Ridge, Ann. Rev. Earth Planet. Sci., 2, 25-50,
1974.
Pillmason, G., S. Arnorsson, I. B. Fridleifsson, H.
Kristmannsd6ttir, K. Saemundsson, V. Stefansson, B.
Steingrimsson, J. Tomasson,and L. Kristjansson,The Iceland
crust:evidencefrom drillhole data on structureand processes,
in Deep Drilling Resultsin the Atlantic Ocean; Ocean Crust,
Maurice Ewing Ser. 3, edited by M. Talwin, C. G. Harrison,
and D. E. Hayes,pp. 43-65, AGU, Washington,D.C., 1978.
Parrott,
R. J.E., 4øAr/39Ar
datingonLabrador
Seavolcanics
and
their relation to sea-floor spreading, M.Sc., Dalhousie
University, Halifax, 1976.
Parrott, R. J. E. and P. H. Reynolds, Argon 40/argon 39
geochronology: age determinations of basalts from the
Labrador Sea area, Geol. Soc. Amer. Bull., 7, 835, 1975.
Pearson, D. G., C. H. Emeleus and S. P. Kelley, Precise
4øAr/39Ar
agesfor the initiationof igneous
activityin the
Small Isles,Inner Hebridesand implicationsfor the timing of
magmatismin the British Tertiary Volcanic Province,J. Geol.
Soc. Lond., 153, 815-818, 1996.
Pedersen,A. K., Armacolite-bearingFe-Ti oxide assemblages in
graphite-equilibratedsalic volcanic rocks with native iron
from Disko, central West Greenland, Contrib. Mineral.
Petrol., 77, 307-324, 1981.
Pedersen, A. K., Reaction between picritic magma and
continentalcrust:early Tertiary silicic basaltsand magnesian
andesites from Disko, West Greenland, Bull. Geol. Soc.
Greenland, 152, 126 pp., 1985.
Pedersen, A. K. and S. Pedersen, Sr isotope chemistry of
contaminated Tertiary volcanic rocks from Disko, central
West Greenland, Bull. Geol. Soc. Denmark, 36, 315-336,
1987.
Piasecki, S., L. M. Larsen, A. K. Pedersen and G. K. Pedersen,
Palynostratigraphyof the lower Tertiary volcanicsand marine
clasticsedimentsin the southernpart of the West Greenland
Basin: implications for the timing and duration of
90 THE NORTH ATLANTIC IGNEOUS PROVINCE
volcanism,Rapp. GronlandsGeol. Unders.,154, 13-31, 1992.
Powell, A. J., A modified dinoflagellate cyst biozonation for
latest Palaeocene and earliest Eocene sediments
central North Sea, Rev. Palaeobot. Palaen., 56, 327-344,
1988.
Price, S., J. Brodie, A. Witham, and R.W. Kent, Mid-Tertiary
rifting and magmatismin the Trail113 region,East Greenland.
d. Geol. Soc. Lond., in press,1996.
Reiners, P. W., B. K. Nelson, and M. S. Ghiorso, Assimilation of
felsic crust by basalticmagma:thermal limits and extentsof
crustalcontaminationof mantle-derivedmelts, Geology,23,
563-566, 1995.
Ribe, N.M., U. R. Christensen,and J. Theissing,The dynamics
of plume-ridge interactions,1: ridge-centredplumes, Earth
Planet. Sci. Lett., 134, 155-168, 1995.
Richards, M. A., R. A. Duncan, and V. E. Courtillot, Flood
basaltsand hot-spottracks: plume headsand tails, Science,
246, 103-107, 1989.
Richey, J. E., Scotland.' the Tertiary Volcanic Districts, His
Majesty's StationeryOffice, Edinburgh,120 pp., 1935.
Ridley, W. I., The petrology of volcanic rocks from the Small
Isles of Inverness-shire,
Report of the Instituteof Geological
Sciences, 1973.
Ritchie, J. D. and K. Hitchen, Early Paleogeneoffshoreigneous
activity to the northwestof the UK and its relationshipto the
North Atlantic Igneous Province, in Correlation of the Early
Paleogenein NorthwestEurope, Spec.Publ. 101, editedby R.
W. O. Knox, R. M. Corfield and R. E. Dunay, pp. 63-78, The
Geological Society,London, 1996.
Roberts, D. G., Marine geology of the Rockall Plateau and
Trough, Phil. Trans. Roy. Soc.London.,A278, 447-509, 1975.
Roberts,D. G., D. G. Masson, and P. R. Miles, Age and structure
of the southernRockall Trough: new evidence,Earth Planet.
Sci. Lett., 52, 115-128, 1981.
Roberts,D. G., and D. Schnitker, et al., Init. Repts. Deep Sea
Drill. Proj., 81,923 pp., 1984.
Robillard, I., D., Francis and J. N. Ludden, The relationship
between E- and N-type magmas in the Baffin Bay lavas,
Contrib. Mineral. Petrol., 112, 230-241, 1992.
Roest, W. R., and S. P. Srivastava,Sea-floor spreadingin the
Labrador Sea: a new reconstruction,Geology, 17, 1000-1003,
1989•
Saemundsson, K., Subaerial volcanism in the western North
Atlantic, in The WesternNorth Atlantic Region, edited by P.
R. Vogt and Be E• Tucholke, pp. 69-86, Geol. Soc. Amer.,
1986.
Saunders,A.D., M. J. Norry, and J. Tarney, Origin of MORB
and chemically-depleted mantle reservoirs: trace element
constraints, in Oceanic and Continental Lithosphere:
Similarities and Differences,edited by M. A. Menzies and K.
G. Cox, pp. 415-445, Journal of Petrology Special Issue,
Oxford University Press,1988.
Saunders,A.D., M. Storey, R. W. Kent, and M. J. Norry,
Consequences of plume-lithosphere interactions, in
Magmatism and the Causes of ContinentalBreak-up, Spec.
Publ. 68, edited by B.C. Storey, T. Alabaster and R. J.
Pankhurst,pp. 41-60, The GeologicalSociety,London, 1992.
Saunders,A.D., J. Tarney, A. C. Kerr, and R. W. Kent, The
formationand fate of large oceanicigneousprovinces,Lithos,
from the 37, 81-95, 1996.
Scarrow,J. H., Petrogenesisof the Tertiary lavas,Isle of Skye,
N.W. Scotland, Unpublished D. Phil. thesis, University of
Oxford, 1992.
Scarrow, J. H., and K. G. Cox, Basalts generated by
decompressive adiabaticmelting of a mantleplume - a case
studyfrom the Isle of Skye, NW Scotland,d. Petrol, 36, 3-22,
1995.
Schilling, J.-G., Iceland mantle plume: geochemicalstudy of
ReykjanesRidge, 242, 565-571, 1973.
Schilling,J.-G., Rare-earth,Sc, Cr, Fe, Co, andNa abundancesin
DSDP Leg 38 basementbasalts:some additionalevidenceon
the evolutionof the Thulean Volcanic Province,Init. Repts.
Deep Sea Drill. Proj., 38, 741-750, 1976.
Schilling,J.-G., Fluxes and excesstemperatures of mantleplumes
inferred from their interaction with migrating mid-ocean
ridges,Nature, 352, 397-403, 1991.
Schilling,J.-G., and A. Noe-Nygaard,Faeroe-Iceland plume:rare
earth evidence,Earth Planet. Sci. Lett., 24, 1-14, 1974.
Schilling,J.-G., P.S. Meyer, and R. H. Kingsley,Evolution of
the Icelandhotspot,Nature, 296, 313-320, 1982.
Sch6nharting,G. and N. Abrahamsen,Paleomagnetism of the
volcanic sequencein Hole 642E, ODP Leg 104, Voring
Plateau, and correlation with Early Tertiary basaltsin the
North Atlantic, in Proc. Ocean Drill. Prog., Sci. Results,104,
911-920, 1989.
Sinton,C. W. andR. A. Duncan,4øAr-39Ar
agesof lavasfrom
the southeastGreenland margin, ODP Leg 152 and the
Rockall Plateau,DSDP Leg 81, in Proc. OceanDrill. Prog.,
Sci. Results,152. in press,1996.
Skogseid,J. and O. Eldholm, Early Cainozoic evolution of the
Norwegian volcanic passive margin and the formation of
marginalhighs, in Early Tertiary Volcanismand the Opening
of theNE Atlantic, Spec.Publ. 39, editedby A. C. Morton and
L. M. Parson, pp. 49-56, The Geological Society, London,
1988.
Smith, A.D., The continentalmantle as a sourcefor hotspot
volcanism, Terra Nova, 5, 452-460, 1993.
Soper, N.J., C. Downie, A. C. Higgins, and L. I. Costa,
Biostratigraphic ages of Tertiary basalts on the east
Greenlandcontinentalmargin and their relationshipto plate
separationin the northeastAtlantic, Earth Planet. Sci. Lett.,
32, 149-157, 1976a.
Soper,N.J., A. C. Higgins, C. Downie, D. W. Matthews and P.
E. Brown, Late Cretaceous- early Tertiary stratigraphyof the
Kangerdlugssuaq area,eastGreenland,andthe ageof opening
of the north-east Atlantic, d. Geol. Soc. Lond., 132, 85-104,
1976b.
Speight,J. M., R. R. Skelhorn,T. Sloan,and R. J. Knaap,The
dyke swarms of Scotland, in Igneous Rocks of the British
Isles,editedby D. S. Sutherland,pp. 449-459, JohnWiley and
Sons Ltd., 1982.
Srivastava,S. P., Evolutionof the LabradorSeaand its bearing
on the early evolution of the North Atlantic, d. R. Astron.Soc.,
52, 313-357,

Srivastava, S. P., and C. R. Tapscott, Plate kinematics of the
North Atlantic, in The WesternNorth Atlantic Region, GNAM
Ser., edited by P. R. Vogt and B. E. Tucholke, pp. 379-404,
GeologicalSocietyof America,Boulder,CO, 1986.
Staples,R. K., R. S. White, B. Brandsd6ttir,W. H. Menke, P. K.
H. Maguire, J. McBride, and J. Smallwood, Faero-Iceland
Ridge Experiment- 1. The crustalstructureof north-eastern
Iceland,d. Geophys.Res.,in press,1996.
Steinthorsson, S., N. Oskarsson,and G. E. Sigvaldason,Origin of
alkali basaltsin Iceland: a plate tectonicmodel, d. Geophys.
Res., 90, 10,027-10,042, 1985.
Stolper, E., and D. Walker, Melt density and the average
compositionof basalt, Contrib. Mineral. Petrol., 74, 7-12,
1980.
Storey, M., R. A. Duncan, H. C. Larsen, A. K. Pedersen,R.
Waagstein,L. M. Larsen,C. Tegner and C. A. Lesher,Impact
and rapid flow of the Iceland plumebeneathGreenlandat 61
Ma (abstract),Eos Trans.AGU, 77, F839, 1996.
Sun, S.-S., and B.-m. Jahn,Lead and strontiumisotopesin post-
glacialbasaltsfrom Iceland,Nature, 255, 527-530, 1975.
Sun, S.-s., and W. F. McDonough, Chemical and isotopic
systematicsof oceanic basalts: implications for mantle
compositionand processes,Magmatismin the Ocean Basins,
Spec.Publ. 41, editedby A.D. Saundersand M. J. Norry, pp.
313-345, The GeologicalSociety,London,1989.
Talwani, M., and G. Udintsev et al., Init. Repts.Deep Sea Drill.
Proj., 38, 1256 pp., 1976.
Tarduno, J. A., Arctic flood basalt volcanism: examining the
hypothesis of Cretaceous activity at the Iceland hotspot
(abstract),Eos Trans.AGU, 77, F844, 1996.
Tarling, D. H., The palaeomagneticpropertiesof someTertiary
lavas from East Greenland, Earth Planet. Sci. Lett., 3, 81-88,
1967.
Tarney, J., D. A. Wood, A.D. Saunders,J. R. Cann, and J. Varet,
Nature of mantleheterogeneityin the North Atlantic:evidence
from deep sea drilling, Phil Trans. Roy. Soc. Lond., A297,
179-202, 1980.
Taylor, P. N., and A. C. Morton, Sr, Nd, and Pb isotope
geochemistryof the upper and lower volcanic seriesat Site
642, Proc. Ocean Drill. Prog., Sci. Results, 104, 429-435,
1989.
Thirlwall, M. F., Generationof Pb isotopiccharacteristics
of the
Icelandplume,J. Geol. Soc.Lond., 152, 991-996, 1995o
Thirlwall, M. F., and N. W. Jones,Isotope geochemistryand
contaminationmechanicsof Tertiary lavas from Skye, in
ContinentalBasalts and Mantle Xenoliths, edited by C. J.
Hawkesworthand M. J. Norry, pp. 186-208, Shiva,Nantwich,
1983.
Thirlwall, M. F., B. G. J. Upton, and C. Jenkins, Interaction
between continentallithosphereand the Iceland plume - Sr-
Nd-Pb isotope geochemistry of Tertiary basalts, NE
Greenland,J. PetroL, 35, 839-879, 1994.
Thomas, J., The occurrence of the dinofiagellate cyst
Apectodinium(Costa and Downie 1976) Lentin and Williams
1977 in the Moray and Montrose Groups (Danian to
Thanetian)of the UK centralNorth Sea, in Correlationof the
Early Paleogenein Nor&west Europe,Spec.Publ. 101, edited
SAUNDERS ET AL. 91
by R. W. O. Knox, R. M. Corfield, and R. E. Dunay, pp. 115-
120, The Geological Society,London, 1996.
Thompson,P., Dating the British Tertiary IgneousProvince in
Irelandby the 40Ar/39Ar
stepwise
degassing
method,
unpublishedPhD thesis,University of Liverpool, 1986.
Thompson,
P., A. E. Mussett
andP. Dagley,Revised
4øAr-39Ar
age for granitesof the Mourne Mountains, Ireland, Scott. d.
Geol., 23, 215-220, 1987.
Thompson,R. N., Magmatism of the British Tertiary Volcanic
Province, Scott. d. Geol., 18, 49-107, 1982.
Thompson, R. N. and S. A. Gibson, Subcontinentalmantle
plumes, hotspots and pre-existing thinspots,d. Geol. Soc.
Lond., 148, 973-977, 1991.
Thompson, R. N., and M. A. Morrison, Asthenosphericand
lower-lithospheric mantle contributions to continental
extensionalmagmatism:an example from the British Tertiary
Province, Chem. Geol., 68, 1-15, 1988.
Thompson,R. N., J. Esson,and A. C. Dunham, Major element
chemicalvariation in the Eocenelavas of the Isle of Skye, d.
Petrol., 13, 219-253, 1972.
Thompson, R. N., A. P. Dickin, I. L. Gibson, and M. A.
Morrison, Elementalfingerprintsof isotopiccontaminationof
Hebridean Palaeocenemantle-derived magmas by Archaean
sial, Contrib. Mineral. Petrol., 79, 159-168, 1982.
Thompson,R. N., M. A. Morrison, G. L. Hendry, and S. J. Parry,
An assessment of the relative roles of crust and mantle in
magma genesis:an elementalapproach,Phil. Trans. R. Soc.
Lond., A310, 549-590, 1984.
Thompson,R. N., I. L. Gibson, G. F. Marriner, D. P. Mattey, and
M. A. Morrison, Trace-elementevidenceof multistagemantle
fusion and polybaric fractional crystallisation in the
Palaeocenelavas of Skye, N W Scotland,d. Petrol., 21, 265-
293, 1980.
Thompson, R. N., M. A. Morrison, A. P. Dickin, I. L. Gibson,
and R. S. Harmon, Two contrasted styles of interaction
betweenbasicmagmascontinentalcrustin the British Tertiary
Volcanic Province,d. Geophys.Res.,91, 5985-5997, 1986.
Thomson,K. and J. R. Underhill, Controls on the development
and evolution of structuralstyles in the Inner Moray Firth
Basin, in PetroleumGeologyof NorthwestEurope, editedby
J. R. Parker, pp. 1167-1178, The GeologicalSociety,London,
1993.
Thy, P., C. E. Lesher, and M. S. Fram, Low pressure
experimental constraintson the evolution of basaltic lavas
from Site 917, southeastGreenland continentalmargin, in
Proc. OceanDrill. Prog., Sci. Results,152, in press,1996.
Tilley, C. E., Some aspectsof magmatic evolution, Q. d. Geol.
Soc. Lond., 106, 37-61, 1950.
Tilley, Co E., and I.D. Muir, The Hebridean Plateau Magma
Type, Trans. Edin. Geol. Soc. 19, 208-215, 1962.
Turner, J. D. and R. A. Scrutton, Subsidencepatterns in the
western margin basins: evidence from the Faeroe-Shetland
Basin, in PetroleumGeologyof NorthwestEurope, editedby
J. R. Parker, pp. 975-983, The Geological Society, London,
1993.
Ulmer, P., Partitioning of high-field strengthelementsamong
olivine, pyroxene, garnet and calc-alkaline
92 THE NORTH ATLANTIC IGNEOUS PROVINCE
experimentalresults and application,Carnegie Inst. Wash.
¾rbook,]988-]989, 42-47, 1989.
Underhill, J. R., Implications of Mesozoic-Recent basin
developmentin the western Inner Moray Firth, UK, Mar.
Petrol. Geol., 8, 359-369, 1991.
Upton, B. G. J., History of Tertiary igneousactivity in the N
Atlantic borderlands,in Early Tertiary Volcanismand the
Opening of the NE Atlantic, Spec. Publ. 39, edited by A. C.
Morton and L. M. Parson, pp.429-453, The Geological
Society,London, 1988.
Upton, B. G. J., C. H. Emeleus,and N. Hald, Tertiary volcanism
between74ø and 76øN, NE Greenland:Hold with Hope and
GaussHalvo, J. Geol. Soc. Lond., ]37, 491-508, 1980.
Upton, B. G. J., C. H. Emeleus, D.C. Rex, and M. F. Thirlwall,
Early Tertiary magmatism in NE Greenland,J. Geol. Soc.
Lond., 152, 959-964, 1995.
Viereck, L. G., J. Hertogen,L. M. Parson,A. C. Morton, D.
Love, and I. L. Gibson, Chemicalstratigraphyand petrology
of the Voring plateau tholeiitic lava and interlayered
volcaniclastic sediments at ODP Hole 624E, Proc. Ocean
Drill. Prog., Sci. Results,I04, 367-396, 1989.
Viereck, L. G., P. N. Taylor, L. M. Parson,A. C. Morton, J.
Hertogen, I. L. Gibson, and O. S. Party, Origin of the
Palaeogene Voring Plateau volcanic sequence, in Early
Tertiary Volcanismand the Openingof the NE Atlantic, Spec.
Publ. 39, editedby A. C. Morton and L. M. Parson,pp. 69-83,
The Geological Society,London, 1988.
Vink, G. E., A hotspotmodel for Icelandandthe Voring Plateau,
J. Geophys.Res., 89, 9949-9959, 1984.
Vogt, P. R., The Iceland mantle plume: statusof the hypothesis
after a decadeof new work, in Structureand Developmentof
the Greenland-ScotlandRidge. New Methodsand Concepts,
editedby M. H. P. Bott, S. Saxov, M. Talwani and J. Thiede,
pp. 191-213, Plenum Press,New York and London, 1983.
Vogt, P. R., and O. E. Avery, Detailed magneticsurveysin the
north-eastAtlantic and Labrador Sea, J. Geophys.Res., 79,
363-389, 1974.
Waagstein, R., Structure,compositionand age of the Faeroe
basaltplateau,in Early Tertiary Volcanismand the Opening
of the NE Atlantic, Spec.Publ. 39, editedby A. C. Morton and
L. M. Parson,pp. 225-238, The GeologicalSociety,London,
1988.
Waagstein,R., and N. Hald, Structureand petrographyof the 660
m lava sequencein the Vestamanna-I drill hole, lower and
middle basalt series, Faeroe Islands, in The Deep Drilling
Project 1980-]981 in the Faeroe Islands, edited by O.
Berthelesen,A. Noe-Nygaard and J. Rasmussen,pp. 39-70,
Foroya Fr6dskaparfelag,T6rshavn, 1984.
Wager, L. R., Geological investigationsin East Greenland.Part
1. General geology from Angmagsalikto Kap Dalton, Medd.
Gron., 105, 46 pp., 1934.
Wager, L. R., A chemical definition of fractionationstagesas a
basisfor comparisonof Hawaiian, Hebrideanand other basic
lavas. Geochim. Cosmochim.Acta, 9, 217-248, 1956.
Wager, L. R., and W. A. Deer, Geologicalinvestigationsin East
Greenland. III. The petrology of the Skaergaardintrusion,
Kangerdlugssuaq, Medd. Gron., ]05, 353 pp., 1939.
Walker, G. P. L., Zeolite zonesand dyke distributionin relation
to the structureof basaltsin eastemIceland,d. Geol., 68, 515-
528, 1960.
Wallace, J. M., R. M. Ellam, I. G. Meighan, P. Lyle, and N. W.
Rogers, Sr Isotope data for the Tertiary lavas of Northem
Ireland: Evidencefor open systempetrogenesis, d. Geol. Soc.
Lond., 151,869-877, 1994.
Walsh,J. N., R. D. Beckinsale,R. R. Skelhom,andR. S. Thorpe,
Geochemistryand petrogenesis of Tertiarygraniticrocksfrom
the Island of Mull, northwest Scotland, Contrib. Mineral.
Petrol., 71, 99-116, 1979.
Watkins, N. D., and G. P. L. Walker, Magnetostratigraphy
of
easternIceland, Am. d. Sci., 277, 513-584, 1977.
Watson, J., Northern Scotland as an Atlantic-North Sea divide, d.
Geol. Soc. Lond., 142, 221-244, 1985.
Weaver, B. L., and J. Tarney, Chemistryof the subcontinental
mantle: inferencesfrom Archaeanand Proterozoicdykes and
continental flood basalts, in Continental Basalts and Mantle
Xenoliths,edited by C. J. Hawkesworthand M. J. Norry, pp.
158-185, Shiva, Nantwich, Cheshire,U.K., 1983.
Werner, R., P. van den Bogaard, C. Lacasse, and H.-U.
Schmincke, Chemical composition, age, and source of
volcaniclastic sediments from Sites 917 and 918, in Proc.
OceanDrill. Prog., Sci. Results,152, in press,1996.
White, R. S., A hot-spotmodel for the early Tertiary volcanism
in the N Atlantic, in Early Tertiary Volcanismand the
Opening of the NE Atlantic, Spec.Publ. 39, editedby A. C.
Morton and L. M. Parson,pp. 3-13, The GeologicalSociety,
London, 1988.
White, R. S., Igneousoutburstsand massextinctions,Eos Trans.
AGU, 70(46), 1480-1483, 1989.
White, R. S., and D. McKenzie, Magmatismat rift zones:the
generationof volcanic continentalmarginsand flood basalts,
d. Geophys.Res., 94, 7685-7729, 1989.
White, R. S., and D. McKenzie, Mantle plumesandflood basalts,
d. Geophys.Res., 100, 17,543-17,585, 1995.
White, R. S., J. W. Bown, and J. R. Smallwood,The temperature
of the Iceland plume and origin of outward-propagating V-
shapedridges,d. Geol. Soc.Lond., 152, 1039-1045, 1995.
White, R. S., G. D. Spence,S. R. Fowler, D. P. McKenzie, G. K.
Westbrook, and A. N. Bowen, Magmatism at rifted
continentalmargins,Nature, 330, 439-444, 1987.
White, R. S., J. H. McBride, P. K. H. Maguire, B. Brandsd6ttir,
W. Menke, T. A. Minshull, K. R. Richardson, J. R.
Smallwood, R. K. Staples, and F. Group, FIRE: Faeroe-
Iceland Ridge Experiment,Eos Trans. AGU, 77, 197, 200-
201, 1996.
Wilson, H. E. and P. I. Manning, Geology of the Causeway
Coast,Memoir of the GeologicalSocietyof NorthernIreland,
Belfast(H.M.S.O.), 72 pp., 1978.
Wolfe, Co J., I. Th. Bjamason, J. C. VanDecar, and S.C.
Solomon, Seismic structureof the Iceland mantle plume,
Nature, 385, 245-247, 1997.
Wood, D. A., A variably veined suboceanicupper mantle -
genetic significance for mid-ocean ridge basalts from
geochemicalevidence,Geology,7, 499-503, 1979a.
Wood, D. A., Dynamic partial melting: its applicationsto

petrogeneses of basaltseruptedin Iceland,the FaeroeIslands,
the Isle of Skye (Scotland) and the TroodosMassif (Cyprus),
Geochim. Cosmochim.Acta, 43, 1031-1046, 1979b.
Young, I. M., R. C. Greenwood,and C. H. Donaldson,Formation
of the EasternLayered Series of the Rhum Complex, north-
west Scotland, Can. Mineral., 26, 225-233, 1988.
Zehnder, C. M., J. C. Mutter, and P. H. Buhl, Deep seismicand
geochemical constraints on the nature of rift-induced
magmatism during break-up of the North Atlantic,
Tectonophysics, 173, 545-565, 1990.
Zindler, A., and S. R. Hart, Chemical Geodynamics,Ann. Rev.
Earth Plan. Sci., 14, 493-571, 1986.
SAUNDERS ET AL. 93
Zindler, A., S. R. Hart, and F. A. Frey, Nd and Sr isotoperatios
and rare earth element abundancesin Reykjanes Peninsula
basalts:evidence for mantle heterogeneitybeneath Iceland,
Earth Planet. Sci. Lett., 45, 249-262, 1979.
R. W. Kent, A. C. Kerr, M. J. Norry, and A.D. Saunders,
Departmentof Geology, University of Leicester,Leicester,LE1
7RH, United Kingdom.
J. G. Fitton, Department of Geology and Geophysics,
Universityof Edinburgh,West Mains Road,Edinburgh,
EH9 3JW, United
 CretaceousBasaltsin Madagascarandthe TransitionBetween
Plume and ContinentalLithosphereMantle Sources
Michael Storey
DanishLithosphere
Centre,Copenhagen,
Denmark

JohnJ. Mahoney
Schoolof OceanandEarthScienceand Technology,
UniversityofHawaii, Honolulu,Hawaii

Andrew D. Saunders

Department
of Geology,University
ofLeicester,
Leicester,UnitedKingdom

Isotopicdataon Cretaceous basaltsfrom threetransectsalongthe 1500-km

lengthof the rifled easternmarginof Madagascar revealsystematic along-axis
variationsthatsuggest theMarionplumewasan importantsourceof meltsat the
southernend of the rifled margin. The relative melt contributionfrom the
lithosphericmantleanda normalmid-oceanridgebasaltmantlesourceappears
to have increasednorthwardalongthe continentalriff away from the estimated
plumecenter.Data on basaltflowsfromthe Volcande l'Androymassifandon
the associated Ejeda-Bekilydike swarm,situatedinlandfrom the rifted margin
in the southof the islandbut abovethe postulatedfocalpointof the plume,show
signaturesattributedto the continentalmantle. These systematicspatial
geochemical variationsplaceboundson the composition and thermalstructure
of the Marion mantleplume at the time of the formationof the Madagascar
igneousprovincesome88 millionyearsago.

INTRODUCTION ridge basalts(MORB) or oceanislandbasalts(OIB) [e.g.,

The traceof mantleplumeactivityon the oceancrustcan
sometimes be backtracked to continental volcanic
margins composedof flood basaltsand seaward-dipping
reflectors [e.g., Hinz, 1981; Morgan, 1981; White and
McKenzie, 1989]. Volcanic rocks eruptedin suchsettings
show a spectrumof elementaland isotopiccompositions
that rangewell beyondthoseobservedin eithermid-ocean
Large IgneousProvinces:Continental,Oceanic,and Planetary
Flood Volcanism
GeophysicalMonograph100
Copyright1997 by the AmericanGeophysicalUnion
Hawkesworthet al., 1990; Saunderset al., 1992]. Some of
rifted
these geochemical features have been interpreted to
indicate that the continental mantle lithosphere is an
importantsite of melt generationin the formationof flood
basaltprovinces[e.g., Hawkesworthet al., 1990; Hergt et
al., 1991]. A different view is that the observedcompo-
sitions arise through contamination of asthenosphere-
derived melts by the continentalmantle and crust [e.g.,
Carlson et al., 1981; McKenzie and Bickle, 1988; Ellam
and Cox, 1991; Arndt and Christensen, 1992]. If the
continentalmantleis essentiallyanhydrousthen it is likely
that only small amountsof melt can be generatedwithin
the lithosphere[Arndt and Christensen,1992; Farnetani
96 CRETACEOUS MADAGASCAR BASALTS
and Richards, 1994]. On the otherhand, if portionsof the
continentalmantle are hydrated then the lower solidus
temperature,comparedto dry peridotite,will increasethe
possibility of partial melting of the lithosphereduring
rifling [e.g., Gallagher and Hawkesworth,1992; Harry
and Leeman, 1995; McKenzie and OWions, 1995]. Geo-
thermometry and geobarometrystudies of continental
mantle nodule suitesindicatethat the majority of nodules
last equilibratedat pressuresand temperaturescloseto the
wet solidus[seeMcKenzieand OWions, 1995]. If partsof
the continental lithospheric mantle have been
metasomatised by fluids or small-degreemelts originating
from the MORB-source mantle or from subduction of
oceaniclithosphere,thentheseregionsarepotentialsitesof
melt generation(in addition to the asthenosphere) along
rifled margins. Which type of source predominates
dependson the interplay of factors such as the major
elementcompositionof the mantlelithosphere,the volume
of hydrated mantle present, the thickness of the
lithosphere,andthe potentialtemperatureof the underlying
asthenosphere.
The island of Madagascarprovides an opportunityto
evaluate the relative roles of different mantle sources to
flood basalt magmatism related to continentalbreakup.
The island was the site of widespread voluminous
magmatismduringthe rifling of Madagascarfrom India at
ca. 88 Ma [Storeyet al., 1995]. Plate reconstructions
place
the Marion (Prince Edward) hotspotcloseto the southern
tip of the continental rift of Madagascarat this time
(Figure 1). The rock types mainly consistof basalt flows
and dikes, but also include rhyolite flows and rarer
microgranitesand microgabbrointrusions.In this paperwe
describethe settingof Madagascarin the Late Cretaceous
and the nature and timing of the igneous activity. In
particular,we focus on publishedand new chemicaland
isotopicdataon basaltsfrom threetransectsalongthe 1500
km lengthof the rifled easternmargin.
REGIONAL SETTING OF MADAGASCAR
Madagascar(Figure 1) is the largestislandin the Indian
Ocean, separated from the African mainland by the
Mozambique Channel and to the north by the Western
Somali and Comores Basins. To the east of the island the
water depth increasesrapidly to •5 km in the Mascarene
Basin. Whereasthe Western Somali Basin is floored by
oceaniccrust [Coffin et al., 1986], the crustalstructureof
the Mozambique Channel is less clear; it may be partly
continental[e.g., Mougenot et al., 1986]. A pronounced
aseismicridge, the MadagascarPlateau, runs from the
southerntip of the island toward the SouthwestIndian
Ridge (SWIR) [e.g., Sinha et al., 1981]. This plateau
representsthe trace of the Marion hotspot,which is now
situated beneath Marion and Prince Edward Islands on the
AntarcticPlate southof the SWIR (Figure 1).
The island is the product of two major rifling events
associatedwith the progressivebreakupof the Gondwana
supercontinent. The first was the separation of
Madagascar/Indiafrom Africa, which beganwith seafloor
spreadingin the Western Somali and Comores Basins
duringBajociantimes (•180 Ma) [Coffin and Rabinowitz,
1988], although earlier rifling may have provided
important depocentersfor Karoo, Permo-Triassicand
Early Jurassicsedimentsin the Morondava,Majunga, and
Diego Basins (Figure 2) [Nicholsand Daly, 1989]. The
pre-dispersallocation of Madagascar against mainland
Africa has been the subject of considerable debate.
However, from seafloormagneticanomaliesand fracture
zone patterns, it is now generally acceptedthat prior to
Middle Jurassictimes it lay adjacent to the coast of
Somalia,Kenya, and Tanzania [Smithand Hallam, 1970;
McElhinnyet al., 1976; Scruttonet al., 1981; Coffin et al.,
1986; Coffin and Rabinowitz 1988; de Wit et al., 1988]
ratherthan againstthe coastof Mozambique[Green, 1972;
Flores, 1984] or as a stationaryisland in a fixed location
relative to Africa [Dixey, 1960]. Jurassic and Early
Cretaceoussouthwardmotion of Madagascarrelative to
Africa was accommodated by a major transformfault, the
Davie Ridge [Coffin and Rabinowitz,1987; Bassiasand
Leclaire, 1990]. This motion resultedin the sedimentary
basinsof westernMadagascarbeingfaultedagainstslivers
of older ocean crust to form the Mozambique Channel.
Madagascar/Indiawas stablerelativeto Africa by the time
of the Albian/Aptianboundary(112 Ma).
The secondrifling event occurredin the Late Cretaceous
as seafloorspreadingin the MascareneBasin resultedin
the separation of Greater India from Madagascar.
Regardingthe pre-dispersallocationof Madagascaragainst
India, Crawford [1978] speculatedthat the Proterozoic
Narmada-Son lineament of central-west India can be traced
into northernMadagascar.Katz and Premoli [ 1979] linked
the Bhavani lineament in southern India with either the
Itremo or Ranotsara lineaments in central eastern
Madagascar,and Agrawal et al. [1992] suggestedthat
Madagascarwas part of the Dharwarcratonof India prior
to fragmentation.A detaileddiscussionof this subjecthas
beengivenby Windleyet al. [ 1994].
The oldestmagneticlineationsrecognizedin the southern
MascareneBasin are anomalies33, 33r, and 34 [Dyment,
1991]. They trend to the northwestand are displacedby a
series of northeast-southwest-trending fracture zones
(Figure 1). Anomaly 34 (•84 Ma) varies from 50 km to
200 km from the continental edge of southeast
Madagascar,as determinedfrom satellite-derived
 STOREY ET AL. 9'/

40 ø 50 ø 60 ø 70 ø

Figure 1. Madagascarand principal structuresin the SouthwestIndian Ocean.Magnetic lineationsin the Mascarene
and Madagascarbasinsare from Dyment [1991]. The Marion hotspottrack was calculatedusingthe model of Miiller et
al. [1993]. The hypothetical88 to 120 Ma track is shownby the faint dashedline. The longitudinalerror in the model
for the post-84 Ma hotspottrack is of the order of severalhundredkilometers.The uncertaintiesincreasefor the older
portion of the track. Inset is a plate reconstructionfor •88 Ma showingrelative positionsof Africa, India, and
Madagascarand the paleoridgesystem:MascareneRidge (MR), SouthwestIndian Ridge (SWIR), SoutheastIndian
Ridge (SEIR). Shadedcirclehasa radiusof approximately1000 km and illustratesthe possiblecenter(M) and lateral
extentof the Marion plumetop. From Storeyet al. [ 1995].

data. The linearity of the rifted east coast margin is
suggestive of strike-slipfaultingpriorto the openingof the
MascareneBasin [Dyment, 1991]. Such faulting would
have provideda zone of weaknessduringthe breakupof
Madagascarand India, as would the stronganisotropyin
the basementfabric [Boastand Nairn, 1982].
Calculationsbasedon the plate motion model of Miiller
et al. [1993] placethe Marion hotspotabout100 km south
of Madagascararoundthe time of continentalbreakupwith
India [Storey et al., 1995]. It is also noteworthythat the
postulated 88 to 120 Ma track of the Marion hotspot
closely parallels the rifted easternmargin of Madagascar
(Figure 1). Whether the Marion hotspotexistedbefore 88
Ma is unknown,althoughit has beenspeculatedthat it was
the source of the earlier Karoo flood basalts of southern
Africa [e.g., Morgan, 1981]. If the plume did exist
98 CRETACEOUS MADAGASCAR BASALTS

Sedimentary rocks of the Diego,

Majunga& MorondavaBasins Massif Diego Suarez
Diego
• Eocene Nosy
B• Basin

•5• Danian Ampasindava

Peninsula
;.'-'• Cretaceous
•
•
Carboniferous-Jurassic
(marine
asin
&continental) Majunga •Cap Est

CapStAndr• U-'"•'•

Maintirano Tamatave

{\

(a)

Morondava
Basin ,-•-. ,Nosy Varika
Mananjary
Fianarantsoa

ii:,i"i;'!i:'
i' Volcanic
Provinces
,•_.

ß _•
:i_,,.'-_.,
:, Cenozoic-Recent
{• Farafan-
Tuh
:.:;?:i:.i
;.i•
:!•11gana
/ Cretaceous
• "'• ?: Basement
//•-•
Major
dike
swarms
.....

...

..'i••..i•
.:..;::!..;
............
ort

Dauphin.•+-•Pan-African
granites

....
ß....
;'•"
Volcan
de
1•4ndroy ':• Archmean and
Ejeda-Bekily
Dike Swarm
.::.....•
: Proterozoic

Figure 2. (a) Geologicalsketchmapof Madagascarbasedon Besairie[1964]. Cretaceous volcanicrocksmainly crop

out alongthe easternrifted margin,at Volcande l'Androyin the south,and in the Majungaand Morondavabasinsin
the west. (b) Map of the Moho topographyof Madagascar(contoursare in km), showing the crustal thickness
decreasingprogressivelyfrom the centralhighlandsto the easternrifted margin[from Fourno and Roussel,1994]. (c)
The drainagepattern of Madagascar:the dottedlines are eastward-draining rivers and the solid lines are westward-
drainingrivers;note the easterlypositionof the main watersheddivide.

before 88 Ma, then it could have played an activerole in
the breakup of Madagascarand India by providing a
preferred path for propagationof the MascareneRidge.
The triggerfor breakupitselfmay havebeenthe northward
propagationof the MascareneRidge as a consequence of
the •88 Ma captureof the spreadingridge systemto the
south by the Marion hotspot, as the trailing edge of
Madagascanlithospherepassednorth of the plume axis
[Storeyet aT., 1995].
Unlike the earlier breakup with Africa, prodigious
amountsof magmawere eruptedovermuchof Madagascar
in association
with this rifting event. Betweenthe opening
of the MascareneBasin and anomaly34 time, the plume
alsoseemsto have interactedwith the triplejunctionof the
SWIR, SoutheastIndian Ridge and nascent Mascarene
Ridge, as indicated by the presenceof large oceanic
plateaus[GosTinet aT., 1980; Sinha et aT., 1981] flankedby
anomaly 34 (Figure 1). The Conrad Rise and the

and southernpart of the MadagascarPlateauare conjugate
with respect to the SWIR; the northeasternpart of the
MadagascarPlateauand the bathymetrichigh on the west
side of the southerntip of India [Laughton, 1975] are
conjugate with respect to the Mascarene Ridge, as
suggestedby the structural trends of these features
[Dyment,1991]. Collectively,the Madagascarflood basalts
andtheseoceanicplateausconstitutea singlelargeigneous
province [Coffin and Eldholm, 1994] with a probable
original
areaofmorethan1 x 106km2.
SUMMARY OF THE GEOLOGY OF MADAGASCAR
Main PhysiographicFeatures
The physiographyof the islandreflectsfairly accurately
its underlyinggeology.The mountainousbackboneof the
island, rising to 2638 m in the Ankaratra Massif, is
composedmainly of Precambrianbasementwith local
Cenozoic volcanic provinces(Figure 2). To the east, a
narrow coastalplain and a remarkablystraightcoastline
are definedby Mesozoicand Cenozoicfaulting,whereasto
the west a seriesof embaymentsmark the onshoreoutcrops
of the MorondavaandMajungasedimentary basins.These
basins,which began developingin the Carboniferousand
Permian, continuedas importantmarine and continental
depocentersthroughoutthe Jurassicand Cretaceousand
into the Tertiary [e.g.,Besairieand Collignon, 1972]. They
deepentoward the west and northwest.The smallerDiego
Basin is in the far north of the island.
The island has a remarkablephysiographicasymmetry,
with the main watersheddivide runningcloseto the eastern
margin(Figure2c). Eastwardfrom the divide,the land sur-
face descendsrapidly acrossa seriesof fault scarpsto the
narrow easterncoastalplain, a narrow continentalshelf,
and into the deep Mascarene Basin. Westward, the
topographygradually drops toward the Morondava and
Majunga basins,althoughthere is a set of cuestasformed
from basinward-dipping, resistant sedimentary and
volcanic rocks. It is temptingto suggestthat the physio-
graphic asymmetryresultsfrom uplift associatedwith the
Cretaceousmagmaticunderplatingand rifting (cf. Paranti
and Deccan [Cox, 1989]).
Precambrian Basement
The Precambrianbasementof Madagascarmakes up
nearly two-thirds of the island [Besairie, 1964; Besairie,
1967]. The Precambrianterrainis dividedinto two major
blocks by the NW-SE Bongolavo-Ranotsara shear zone
[see Windleyet al., 1994]. North of this zone the basement
is mainly granitic or gneissic, whereas south of it
supracrustalrocks predominate.However, the Androyan
STOREY ET AL. 99
Systemto the southof the shearzone is characterizedby
granulites and migmatites and is consideredto be Late
Archeanto Early Proterozoicin age [Paquetteet al., 1994;
Windley et al., 1994]. Large areas of the Madagascan
Precambriancrustwere reworkedby the Pan-Africanevent
around 550 Ma [Windley et al., 1994]. East-westprofiles
of Bouguer anomaliesacrossthe Precambrianbasementof
Madagascarreveal rapid shallowingof the Moho from a
crustal domain of normal thickness (35-40 km) in the
center of the island to a thicknessof 25-27 km along the
eastcoast(Figure2b) [Fournoand Roussel,1994].
CretaceousMagmatism
The Cretaceous volcanic and intrusive rocks of
Madagascarcrop out semicontinuously alongthe 1500-km
lengthof the eastcoast,which marksthe rifted margin,and
in the Majunga and Morondava basins in western
Madagascar(Figure 2). The rocks include basalt flows,
dikes, and some rhyolite flows. Along the rifted margin,
the flows lie mainly upon the Precambrian basement,
whereas most of the dikes are coast-parallel.The lavas
generally form shallowly seaward-dippingpiles. Whether
any seaward-dippingbasaltreflectorsare presentoffshore
is unknown;however,the narrownessand steepness of the
continentalmargin appearto precludethe existenceof a
largevolcanicwedgeof the sortthattypifiesothervolcanic
rifted margins, such as East Greenland [e.g., Larsen and
dakobsd6ttir,1988]. Severalportionsof the rifted eastern
margin of Madagascarare devoid of volcanic rocks but
this may be a functionof a lack of preservationratherthan
a reflection of the magmatic activity. For example, lavas
are foundinfrequentlyin the regionnorthof Tamatave,but
dikesare very common.Elsewhere,the basaltis reducedto
thick piles of saprolite and laterite covered by heavy
vegetation.
The Volcan de l'Androy complex in southern
Madagascarcontainsthe thickestsequenceof Cretaceous
volcanicrocks exposedon the island;this massifforms an
oval-shapedoutcropsome50 km wide by 90 km long and
consistsof interbeddedflows of basalt and rhyolite, with
microgranite intrusions exposed at the northern and
western margins of the complex (Figure 3) [Battistini,
1959]. The massif sits within a much wider basin formed
by headwatererosionby the MandrareRiver system.This
massif is located more than 100 km west of the continental
rift in an area that has apparentlyundergonelittle crustal
extension.The total thicknessof the flows may be more
than 2000 m [Battistini, 1959]. The lowest flows lie
unconformably on Archean granitic gneisses of the
Androyan System. Substantialerosionhas left a seriesof
rhyolite-cappedtable mountain remnants (e.g.,
100 CRETACEOUS MADAGASCAR BASALTS

•-•Major
Faults
L
Basement I
[• Structures
I
•--•DikesI
Dip& Strike
of•
[•]Bedding
/

oTranomaro
:::..:.'.'...•
Rhyolite (Upper)
..2• Basalt
(Main
Series)
• Rhyolite
(Lower)
Basalt
(RimSeries)
I 10km I [•-• Microgranite
•,• Rhyolite

Figure 3. Left: Geologicalmap of the Volcan de l'Androy basalt-rhyolitelava massif.Vohitsimbe is a prominent

rhyolite-cappedtable mountainin the centerof the massifiThe dip arrowsindicatethat the complexhas a saucer-
shapedstructure.Map modifiedafter Battistini[1959] usingthe Landsatimageshownat right.

simbe), and it is unclear how much material has been
strippedfrom the top of theseplateaus.In many areas,the
lowest flows are rhyolite, forming an annular outcrop
pattern,but there are local outcropsof underlyingbasalt
that we have termedthe "rim series"(Figure 3). Above the
lower rhyolitesis a layer of flat-lying or gently inwardly
dipping tholeiitic basalt flows, interbeddedwith and
overlainby the upperrhyolites.The proportionof rhyolite
to basalt increasesmarkedly up section,the topmostunits
formingcolumnar-jointed rhyoliteflows up to 50 m thick.
The flows appearto havebeenderivedlocally,as indicated
by microgranites that partly encirclethe massifandwhich
are compositionallyidenticalto the lower rhyolites;the
overall form resemblesa large caldera,althoughthere are
no obviousring faults,which arenormallyassociated with
such structures.To the west of the Volcan de l'Androy
massif is the Ejeda-Bekily dike swarm, which trends
toward the massif (Figure 2). These dikes are unusual
amongthe Cretaceous rocksin thatthey arepredominantly
nepheline-normative[Dostalet al., 1992].
Basalt flows, interbedded with Upper Cretaceous
sedimentaryrocksin the Majunga and Morondavabasins,
crop out for 700 km in the northwestand 200 km in the
southwest.Sedimentationbeganduring the Carboniferous
in the Morondava Basin, but was slightly later (late
Permian)in the Majunga Basin [Besairie,1966; Boastand
Nairn, 1982]. Sedimentationalternatedbetween marine
and non-marine, reflecting the proximity of the basin
margins.In the MorondavaBasin,the largestof the basins,
igneousrocksare mostprominentin the north and south.
Lava flows (predominantlybasaltic)make up a pile as
muchas 100 m thick, but 30 m is more common.Rhyolite
flows and microgabbrointrusionsare present,and a dike
swarm crops out adjacent to the basalt flows in the
northernpart of the basin (Figure 2). The Majunga Basin
extends along the west coast of Madagascarfrom the
southernside of the AmpasindavaPeninsulato Cap St.
Andr6 [Boastand Nairn, 1982] (Figure 2). Basalticlavas
wereeruptedovera wide areaof theMajungaBasinduring
the Turonian and in the north of the basin are overlain by
sedimentswith an Upper Turonianmarinefauna.The lavas
share many of the features of those found in the
Morondava Basin, although they are compositionally
different. They are widespread,averageapproximately50
m in total thickness(up to 200 m in someareas),and are
predominantly
 STOREY ET AL. 101

Small outliers of basalt, reported as Cretaceousby

I i I
44 ø 46 ø 48ø iOø

Besairie [1964], rest on Precambrian basement in north-

central Madagascar (Figure 2). These outliers are
important because,if they are Cretaceous,they indicate _12
ø ,•'

that the extent of the magmatismwas far greaterthan that

presently preserved along the coastal margins. They
14 F Sambava
' [] '
suggestthat basalt flows travelledacrossthe island,from
the eastern rift zone into the western basins.
Majun.•ga•/•-•
7
TIMING OF CRETACEOUS MAGMATISM IN
18ø
MADAGASCAR AND THE BREAKUP FROM INDIA • [• Antananarivo
/'.'•'•'•j
Tamatave
Althoughthe Cretaceousflood basaltsof Madagascarcan
be relatedto the track of the Marion hotspot(Figure 1), the
precise timing of volcanism was poorly known until
recently. K-Ar datesrange from 31 to 97 Ma [Dostal et al.,
22ø
• • /?'•Mananja•
' •
1992; Storetvedt et al., 1992], whereas paleontological ,
evidence[Besairieand Collignon,1972; Boastand Nairn,
1982] suggests a Maastrichtian (65.4 to 71.3 Ma
[Obradovich,1993]) to Turonian(88.7 to 93.3 Ma) age Ejeda-Bekil•
rangefor the igneousactivity.An 40Ar-39Ardatingstudy Dike Swarm

[Storeyet al., 1995], however,showedthatthe Madagascar Fort

Dauphin
4øAr/39Ar AGES
Cretaceousvolcanicprovinceformed over no more than 6 o
• i 2o•km 75Ma 80 85 90 95
m.y., a much shorter interval than indicated by the
i .... i .... i .... i .... i

publishedK-Ar dates. This study demonstratedthat the
volcanic rocks of the 1500-km-long rifted margin show
virtually no statistically significant differences in age
(Figure 4); the weightedmean of the isochronagesis 87.6
+ 0.6 Ma. Two tholeiitic basalt flows from the Majunga
Basin gave ages of 87.6 + 2.9 Ma and 88.5 + 1.3 Ma,
respectively,within errorof the age determinations for the
rifted margin rocks. A rhyolite dike intrudedinto the base
of the Volcan de l'Androy lavas has an age of 86.3 + 1.9
Ma. An age of 84.4 + 0.4 Ma was given by sanidine
phenocrystsfrom a rhyolite flow at the top of Volcan de
l'Androy, which is slightly youngerthan the mean of the
rifted margin and Majunga Basindates.Similar youngages
were also shown by a basalt sampleof the Ejeda-Bekily
dike swarm(84.8 + 1.3 Ma), and by a basaltflow from the
southwestpart of MorondavaBasin(84.5 + 0.7 Ma). These
datasuggestthat Madagascanvolcanismceasedfirst in the
north [Storeyet al., 1995].
BecausebreakupbetweenMadagascarand GreaterIndia
occurred at the time of the CretaceousQuiet Zone, the
precise age of the rift-to-drift transition cannot be
determinedfrom seafloormagneticlineations.However,
seafloor-spreadingwas clearly well organized in the
southernMascareneBasinshortlybefore84 Ma, the age of
anomaly 34 [Harland et al., 1989]. The establishmentof
regular seafloor spreadingin the MascareneBasin thus
appears to have coincided with the eruption of the
Madagascarflood basaltprovince.
Figure 4. The outcrop pattern of Madagascar Cretaceous
volcanic rocks (black) with shadedboxes showing our three
sampletransectsalongthe easternrifted margin,centeredaround
Mananjary (south), Tamatave (central) and Sambava(north),
respectively.
Also shownare4øAr-39Ar
agedeterminations
on the
Cretaceousvolcanic rocks. Sample symbols:squares= eastern
rifted margin; circles = Majunga and Morondava basins;
diamonds- Volcan de l'Androy and Ejeda-Bekily dike swarm.
The errorbarsare+ 1s. From Storeyet al. [ 1995].
GEOCHEMISTRY OF THE CRETACEOUS
MAGMAS OF MADAGASCAR
Geochemical studies of the Cretaceous volcanic rocks of
Madagascarare few. A reconnaissance isotopicstudywas
carriedout by Mahoneyet al. [ 1991] on samplesfrom the
Ejeda-Bekilydike swarmand from the flows eastof Tulear
and along the southeastcoastnear Mananjary.Major and
trace elementdata for the samesampleset were given by
Dostal et al. [ 1992]. Here, we review this work andpresent
resultsof our new geochemicalinvestigations.
Samples
Samplesfor our presentstudywere collectedfrom three
transectsencompassing most of the lengthof the volcanic
rifted margin of easternMadagascar(Figure 4). From
south to north they are the Mananjary transect,
102 CRETACEOUS MADAGASCAR BASALTS
Tamatavetransect,and the Sambavatransect.Petrographic
descriptionsof a subset of the samplesare given in
Appendix 1. In addition,we also sampledthe Volcan de
l'Androy massif and the Ejeda-Bekily dike swarm,
essentiallyformingonetransectperpendicular to the rifted
margin(J. Mahoneyet al., ms. in prep.).
Becauseof limited exposureit is not possibleto state
with certaintythe environmentof emplacement,although
the presenceof vesicularblockssuggests that mostunits
were probably emplaced as thick flows rather than
intrusivesheets.Around Tamatave,the sampledunitswere
predominantlydikes. Texturesrange from fine-grained,
almostaphaniticbasaltsto ophiticdoleritesand, in some
cases(MAN 90-17), accumulativegabbros.
The majority of the Mananjary samplesare coarse-
grainedbasaltsor doleriteswith significantlylessthan 5%
phenocrystsand are thus likely to approximateliquid
compositions. Somedoleritescontaintraceor small(<5%)
amountsof partiallyresorbedplagioclase.However,many
fine-grained basalts contain small amounts of
clinopyroxene microphenocrysts in additionto plagioclase
phenocrysts, implyingthat similarphasesmay be present
in the doleritesbut cannotbe readily distinguished. A few
samplescontainolivinemicrophenocrysts, now completely
replacedby clays. Clinopyroxeneis often purplishto
brown in thin section,reflectingthe high Ti andFe content
of many samples(e.g., MAN 90-80). Somesamplesshow
no visible signsof alteration(e.g., MAN 90-20), which is
alsoreflectedin their low total weight losson ignition.The
majority of samples,however,showpatchyreplacementof
the mesostasis by yellow-brownor green-brownclaysand,
rarely, chlorite.The majority of the primarysilicatephases
are unaffectedby alteration.Vesicles, where present,are
often filled with clays, chlorite or, in one sample(MAN
90-15), prehniteand pumpellyite.
All of the Tamatavesamples,with the exceptionof TAM
92-43 which may be from a flow, were taken from dikes.
The samplesare aphyric,and either pale-greenor brown
clays replace the mesostasis.Calcite is a common
secondaryphase.
The samplesfrom the Sambavatransectwere collected
from dikes and flows. Like the samplesfrom Mananjary,
they exhibit a range of textures,but a commontype is an
"ophimottled" texture, where pyroxene, mostly
clinopyroxene but also some orthopyroxene,forms
glomero-ophiticpatches.Samplesmay be aphyric (e.g.,
SAM 92-1) or plagioclase-phyric(SAM 92-3). Clinopy-
roxene is a microphenocrystphase in some fine-grained
samples (e.g., SAM 92-20D). Alteration is usually
restrictedto replacementof mesostasis
by brownclays.No
calcite is observed,and some samples(e.g., SAM 92-3)
showno visible signsof alteration.
Analytical Methods
X ray fluorescence(XRF) measurementsand instru-
mental neutron activation analysis (INAA) of the
Madagascarsampleswere carriedout at the Universityof
Leicester.For XRF analysis,sampleswere crushedin an
agateshatterbox.For determinationof the trace elements
V, Cr, Ba, Nb, Zr, Y, Sr, Rb, and Ni the rock powderwas
made into 46-mm-diameterpowder briquettesby adding
severaldropsof a 7% solutionof polyvinylalcoholto 15 g
of rockpowderandsubjecting the mixtureto a pressure
of
15 tons in a steeldie. Analysiswas carriedout usinga W-
anodetube on a PhilipsPW 1400 X ray spectrometer. For
major-element analysis, rock powders were dried
overnight at 120øC before determiningweight loss on
ignition(LOI) at 800øC.Glassdiskswere madeby fusing
(at •1200øC) a mixtureof 1 g of ignitedrock powderand5
g of lithium metaborate/lithium tetraborateflux (Johnson
Matthey Spectroflux100B). Analysis of the glassdisks
was carried out using a Rh-anodetube on an ARL 8420+
X ray spectrometer.
Rare-earthelements(REE) and Ta, Th, Hf, Sc, and Co
concentrationswere determinedby instrumentalneutron
activationon 0.2-g splits of the Madagascarsamplesand
also of five lavas from Marion and Prince Edward islands,
and Funk Seamount(a volcano of the Marion group).
Sampleswere irradiatedfor approximately30 hoursin a
thermalneutronflux of 1 x 1012n cm-2 in theImperial
CollegeReactorCentreand activitiesof the radioisotopes
countedon EG & G ORTECcoaxialand loaxialhigh-purity
germanium detectors. Reactor flux variation was
monitoredand correctedby interspersingwafer-thin iron
foils between the samples.Elemental abundanceswere
determinedby referenceto the standardAC(II) [Pottset
al., 1981]. Becauseof the low abundancesof Sc and Co in
AC(II), the standard JB-la [Ando et al., 1987] was
included for Sc and Co determinations. The standards
BOB-1 and JB-la (excludingSc and Co in the latter)were
run as "unknowns" in each sample batch to monitor
analytical uncertainty.A subsetof our resultsappearsin
Tables 1 and 2.
Sr, Nd, and Pb isotopic determinations,as well as
isotope-dilutionabundancemeasurements,were carried
out at the Open University,England,and at the University
of Hawaii. At the Open University, rock chips(<1 mm),
ratherthan powders,were selectedfor Pb-isotopicanalysis
to avoid contaminationintroducedthrough the crushing
procedure.To removepossiblesurfacecontamination, the
rock chips were cleaned for 10 minutes in an ultrasonic
bath usingultrapure6M HC1 followed by ultrapureH20.
This procedurewas repeated at least twice. Dissolution
was by a HF-HNO3 mixture, the residuebeing
STOREY ET AL.
104 CRETACEOUS MADAGASCAR BASALTS
 STOREY ET AL. 105

0 00000 •
106 CRETACEOUS MADAGASCAR BASALTS

TABLE 2. Major andTraceElementContentsof VolcanicRocksfrom Funk Seamount

and Marion and Prince Edward Islands

Location Marion Funk Seamount Prince Edward

Sample WJM-50 WJM-49 WJM-21 D1-B1 D1-E PREI-13

SiO2 46.9 48.5 49.5 44.7 44.9
TiO2 3.44 3.36 2.74 4.05 4.01
A1203 16.48 15.13 16.55 16.03 16.10
FeO a 11.97 12.33 11.05 13.59 13.53
MnO 0.17 0.18 0.19 0.16 0.15
MgO 5.11 5.53 3.94 5.59 5.88
CaO 10.29 9.90 7.22 8.09 8.05
Na20 3.32 3.11 4.29 4.65 4.37
K20 1.30 1.18 1.86 1.66 1.69
P205 0.54 0.51 0.83 1.10 1.08
LOI 0.81 0.65 0.61 0.79 0.77
H20- 0.03 0.05 0.05 0.13 0.14
Total 100.36 100.43 98.84 100.53 100.71

XRF (ppm)
V 288 264 104 187 175
Cr 32 29 3 73 73
Ba 278 256 448 391 374
Nb 41 36 57 59
Zr 251 250 356 329 390
Y 30 32 42 38 27
Sr 614 507 777 1099 1074
Rb 22 20 33 26 29
Ni 41 <2 108 49 41
Sc 23 25 14 13 14
Cu 58 11 445 18 17
Zn 108 111 338 141 147
Co 42 47 20 48 44

INAA (ppm)
La 32.0 27.4 48.1 43.9 45.2 16.4
Ce 65.4 59.2 95.8 95.0 98.3 37.1
Nd 35.4 34.0 48.7 56.2 58.8 24.6
Sm 7.8 7.5 10.9 13.0 13.0 6.1
Eu 2.6 2.7 3.8 4.4 4.5 2.7
Gd
Tb 1.04 1.09 1.44 1.65 1.65 0.91
Yb 2.26 2.32 2.94 1.92 2.00 1.92
Lu 0.32 0.32 0.38 0.23 0.24 0.27
Ta 2.40 2.05 3.33 2.87 3.07 2.71
Th 3.27 2.83 4.48 3.29 3.59 1.56
Hf 6.17 6.23 8.63 8.73 9.29 8.46
Sc 24.6 27.1 15.6 14.1 13.6 21.2
Co 48 47 23 47 47 32

INAA analyseswere madeduringthis study.The XRF traceandmajorelementdataare

from Reid and le Roex [1988] (Funk Seamount)andMahoneyet al. [1992] (Marion).
a All Fe as
 STOREY ET AL. 107

to nitratesand then chlorides.Pb was separatedusing a typicallyhavehigherSiO2, as well as a strongdepletionin

microcolumn technique for which the total procedural Nb and Ta relative to other highly incompatibleelements
blankwaslessthan0.5 ngg-1.Sr andNd wereseparated (Figures5 and 6). There is no obviousspatialor temporal
using conventional ion-exchange procedures. Isotopic distinction between the location of the Fe-Ti series basalts
compositionswere measuredon a MAT 261 multicollector and these low Fe-Ti basaltsalong the east coastand the
mass-spectrometer.A VG Sector multicollector mass- two types are occasionallyinterlayered.The low Fe-Ti
spectrometer was used at the University of Hawaii basalts showelevated andwidelyranginginitial87Sr/86Sr
following sample preparationand analytical procedures and low gNdvalues(M. Storeyet al., unpubl.data) similar
outlined by Mahoney et al. [1991]. For interlaboratory to those of the Morondava Basin in southwestern
comparison, standard rock BCR-1 was run at both the Madagascar[Mahoneyet al., 1991]. The MorondavaBasin
Open University and the University of Hawaii. Isotopic tholeiites have been interpretedas being variably but
compositionsof a subsetof samplesare given in Table 3 highly contaminated by ancientcrustalmaterial,broadly
and the U, Pb, and Th isotope-dilutiondata in Table 1. like that affecting the Busheand PoladpurFormationsof
Also shownin Table 3 are isotopicvaluesage-corrected to the later (66 Ma) DeccanTraps in India [Mahoneyet al.,
88 Ma on the basisof measuredparent-daughter ratiosand 1991]. Likewise, we consider the compositionaland
published 4øAr-39Ar ages[Storeyet al., 1995].For the isotopicfeaturesof most of the low Fe-Ti basaltsto be an
Open University Pb isotopicdata, the age correctionwas indicationthat they have experiencedvariable amountsof
derivedfrom the Th/Pb ratio and assumingTh/U = 4. The crustal contamination.
University of Hawaii resultswere age-corrected usingU, Lastly, we recognizea rare groupof high Mg-Ti basalts
Th, Pb, Sr, Rb, Sm, and Nd isotope-dilutiondata (not from the Mananjarytransect(Figure 5). Thesebasaltsare
shownin Table 1 for Nd, Sm, Rb, or Sr) determinedon the strongly enriched in the light REE, resemblingMarion
samesamplesplitsanalyzedfor isotoperatios. hotspot lavas. Three of the samples analysed have a
pronouncedBa spike in their primitive-mantle-normalized
East Coast Basalts and Dikes patterns (Figure 6). Two features common to both the
Marion lavasandthe Mananjaryhigh Mg-Ti basaltsare a
Chemical results. The great majority of data for rocks steepslopebetweenthe middleandheavyREE (Figure6)
from the rifled easternmarginof Madagascarfall on a low- and low Sc abundances (<25 ppm; Tables 1 and 2). These
pressure fractionation trend of strong Fe- and Ti- observationsindicateresidualgarnetduringmelting.
enrichmentwith decreasingMgO (Fe-Ti series), which
showsa maximumbetween4 and6 wt.% MgO (Figure5). Isotopes.New isotopicdata for the east coastsamples
The Fe-Ti enrichmenttrendis mostpronouncedin samples that lack chemicalsigns of probablesignificantcrustal
from the Mananjary and Tamatave transects. It is contamination(e.g., high SiO2 and La/Nb) are given in
accompaniedby a slight decrease in SiO2 and near- Table 3. In Pb isotopicspace(Figure 7) most of the data
constant CaO/A1203, with plagioclase fractionation plot above the Northern HemisphereReferenceLine of
indicated by varying degreesof Sr and Eu depletion Hart [1984], a generalfeatureof the Indian Oceanmantle
(Figure 6). The lavas and dikes of the Fe-Ti seriesshow domain [e.g., Dupr• and All•gre, 1983; Hamelin et al.,
strong compositional similarities to evolved iron-rich, 1986; Price et al., 1986; Mahoneyet al., 1989, 1992]. The
silica-poor liquids described for the East Greenland basaltsof the southernmost (Mananjary)transecton the
volcanic rifled margin [Brookset al., 1991] but contrast eastcoastfall mainly into two isotopicallydistinctgroups,
with recent alkalic lavas from Funk Seamount and Marion which correlatebroadlywith major elementcompositions.
Island, where low CaO/A1203 suggeststhat fractionation Data for the more evolvedmembers(TIO2>3 wt.%) of the
occurredat higher pressures[seeReid and le Roex, 1988]. MananjaryFe-Ti serieslavasmostlyplot to theright of the
The riff-relatedFe-Ti basaltsof Madagascarare enriched geochron and show a positive correlation between
in the light REE (Figure 6). In primitive-mantle- 2ø6pb/2ø4pb
andboth2ø7pb/2ø4pb
and2øspb/2ø4pb
(Figure
normalized trace element plots of samples from the 7). Thehighest
2ø6pb/2ø4pb
ratios(•18.6)arecomparable
Tamatave and Sambava transects, the presence of a to thosemeasured
in recentMarionhotspotlavas,although
pronounced positive Ba spike is the most notable the latterare characterized
by slightlylower2ø7pb/2ø4pb
differencefrom the MananjaryFe-Ti seriespatterns. and 2øspb/2ø4pb
ratios(Figure7). Curiously,
the less
A number of basalts,particularly from the Mananjary evolvedmembers(TIO2<2 wt.%)) of the MananjaryFe-Ti
and Sambava transects, fall below the main Fe-Ti series basalts exhibit a much narrower range in
enrichment, low-pressure fractionation trend and also 2ø6pb/2ø4pb
ratios(17.7-17.9),substantially
lower
108 CRETACEOUS MADAGASCAR BASALTS

TABLE 3. Sr,Nd, andPb IsotopicCompositions

of RocksfromtheRifled EasternMarginof Madagascar
Transect 87Sr/86Sr 87Sr/86Sr 143Nd/laaNd •Nd 2ø6pb/2øapb
2ø7pb/2øapb
2øspb/2øapb
2ø6pb/2øapb
2ø7pb/2øapb
2øspb/2øapb
Sample measured t= 88 Ma measured t-88 Ma measured measured measured t=88 Ma t=88 Ma t--88 Ma

Mananjary
Mananjary high-Mg-Ti rocks
MAN90-45 0.70379 0.70366 0.513023 +8.1 17.946 15.548 37.712 17.794 15.541 37.513
MAN90-47 0.70380 0.70370 0.512847 +4.7 17.822 15.580 37.889 17.699 15.574 37.727
MAN90-85 a 0.70338 0.70336 0.512754 +2.7 16.921 15.175 37.084 16.805 15.169 36.977
MAN90-86 0.70424 0.70422 0.512720 +2.2 16.918 15.153 37.677 16.765 15.146 37.478

Mananjary Fe-Ti series (Ti02>3 wt.%)

MAN90-1B 0.70357 0.70347 0.512925 +5.9 18.448 15.578 38.813 18.256 15.569 38.561
MAN90-6 0.70363 0.70351 0.512930 +6.0 18.762 15.607 38.928 18.549 15.596 38.649
MAN90-8 0.70603 0.70596 15.891 15.013 37.017 15.823 15.010 36.928
MAN90-31 0.70355 0.70345 18.820 15.605 38.856 18.615 15.595 38.587
MAN90-43 0.70356 0.70340 0.512929 +6.0 18.807 15.588 38.764 18.619 15.579 38.517
MAN90-49 0.70308 0.70295 0.513000 +7.4 18.348 15.506 38.141 18.144 15.496 37.873
MAN90-57 0.70377 0.70354 0.512894 +5.4 18.695 15.615 38.899 18.492 15.605 38.633
MAN90-63 0.70325 0.70313 18.260 15.544 38.086 18.086 15.536 37.858
MAN90-66 0.70348 0.70339 0.512913 +5.6 18.574 15.572 38.591 18.422 15.565 38.392
MAN90-71 0.70349 0.70336 17.967 15.483 38.153 17.776 15.474 37.902
MAN90-80 0.70347 0.70330 0.512957 +6.5 18.389 15.552 38.616 18.182 15.542 38.345
MAN90-81 0.70359 0.70347 18.421 15.582 38.636 18.218 15.572 38.370

Mananjary Fe- Ti series (Ti02<2 wt.%)

MAN90-2 0.70366 0.70356 0.512994 +7.1 18.012 15.479 37.971 17.874 15.473 37.788
MAN90-15 0.70350 0.70335 17.847 15.485 37.963 17.782 15.481 37.878
MAN90-16 a 0.70355 0.70354 0.512919 +5.5 17.892 15.517 37.957 17.794 15.512 37.840
MAN90-17 0.70333 0.70330 17.982 15.571 38.254 17.831 15.564 38.057
MAN90-20 0.70491 0.70489 0.512880 +5.0 16.876 15.558 37.614 16.834 15.556 37.559
MAN90-22/2 0.70438 0.70431 0.512903 +5.3 17.991 15.510 38.006 17.844 15.503 37.813
MAN90-77 0.70360 0.70357 0.512986 +6.9 18.041 15.487 38.051 17.864 15.478 37.818

Tamatave

TAM92-6 a 0.70577 0.70569 0.512685 +1.3 16.718 15.160 37.980 16.702 15.159 37.956
TAM92-27 a 0.70431 0.70418 0.512900 +5.4 17.606 15.416 37.912 17.454 15.409 37.748
TAM92-30 a 0.70499 0.70478 0.512851 +4.5 17.767 15.426 38.123 17.372 15.407 37.656
TAM92-43 a 0.70446 0.70435 0.512847 +4.4 17.515 15.421 37.846 17.336 15.412 37.632

Sambava

SAM92-1 a 0.70425 0.70420 0.512700 +1.5 17.305 15.496 38.078 17.192 15.491 37.984
SAM92-3 a 0.70386 0.70382 0.512752 +2.6 17.120 15.391 37.738 16.987 15.385 37.567
SAM92-10 a 0.70340 0.70338 0.512891 +5.1 17.645 15.476 37.884 17.421 15.465 37.633
SAM92-16 a 0.70346 0.70339 0.512885 +5.0 17.718 15.506 38.007 17.539 15.497 37.807
SAM92-20D a 0.70385 0.70374 0.512787 +3.3 17.328 15.433 37.877 17.196 15.427 37.711
SAM92-33B a 0.70682 0.70681 0.512353 -5.3 18.173 15.782 38.852 18.158 15.781 38.815

USGS standard

BCR-I 0.70498 0.512627 18.816 15.640 38.741

BCR-I a 0.70502 0.512633 18.807 15.621 38.678

a Analyzed at the University of Hawaii.

Pb isotopicratiosarecorrected
for fractionation
usingtheNBS 981 standard
valuesof Todtet al. [1996].Errorson theHawaii dataare_<+0.000012
(0.2 • units)for 143Nd/144Nd,
+0.00002for 87Sr/86Sr,
+0.010for2ø6pb/2ø4pb
and2ø7pb/2ø4Pb,
and+ 0.032for 2øsPb/2ø•Pb.
Maximumerrorsonthe
OpenUniversity dataarecomparable,
exceptforsamples
MAN90-22/2,
-45,and-66,whichhaveerrors
_<+0.000021
for
 STOREY ET AL. 109

1.2 seen in Marion hotspot lavas. Basaltsfrom the Tamatave

and Sambavatransects,as well as the Mananjary high Mg-
1.0 Ti basalts,are characterized
by low 2ø6pb/2ø4pb
ratios,
with the majority having values between 16.6 and 17.8.
0.8 oooø• •a_ Theselow2ø6pb/2ø4pb basalts
formtwopopulations onthe
Sambava
0.6 Tamatave
basisof their2ø7pb/2ø4pb
ratios.Interestingly,
thegroupof
Mananjary low 2ø6pb/2ø4pb basaltswhichhave high 2ø7pb/2ø4pb
0.4
Mananjary ratios overlapswith data for MORB from the anomalous
(high-Mg-Ti) SWIR ridge segmentbetween39ø and 41øE [Mahoney et
0.2
6 al., 1992].
In 206pb/204pb-eNd
and 206pb/204pb-87Sr/86Sr
isotope
spacethe evolved membersof the Mananjary Fe-Ti series
basaltsshow an oppositetrend from the other east coast
basalts,with initial end decreasingfrom +7.4 to +5.4 with
increasing
2ø6pb/2ø4pb.
Thisarray(TrendI) spans
a range
ß
of values from those for normal MORB from the 32-39øE

o segmentof the SWIR to Marion hotspotcompositions

Ooo (Figure8). Initial 87Sr/86Sr
increases
from 0.7031 to
0.7036,overlapping withthehigh87Sr/86Sr
regionof the
Marionhotspot field.These87Sr/86Sr
ratiosarelessthan
the lowest values measured for basalts from R6union and

•8•- Mauritius
forclosely
comparable
2ø6pb/2ø4pb.
The Sambava and Tamatave basalts and the other
Mananjary samples, including the high Mg-Ti basalts,
141 form a loosely defined array (Trend II in Figure 8) in
ß

oo 06) o which end decreases

and 87Sr/86Srincreaseswith
ß
decreasing
2ø6pb/2ø4pb.
A similartrendis shownby
Oo
o•P=
ß o
isotopicdatafor the Ejeda-Bekilydike swarm[Mahoneyet
al., 1991]. At thehigh2ø6pb/2ø4pb
endof thearray,Trend
56
I! approaches,but does not quite reach, the values of
normal MORB from the SWIR (32-39øE). At the low
2ø6pb/2ø4pb
end, the array partly overlapswith the
52t anomalousMORB compositionsreported for the SWIR
o%ø--•
0

0 ridge segmentbetween 39ø and 41øE [Mahoney et al.,

0
48 • 1992].
46

Origin of TrendI

Trend I is consistent with derivation of the evolved

Mananjary Fe-Ti seriesbasaltsfrom mixtures of Marion
4.0
plume mantle and Indian MORB mantle, the latter
3.5 isotopicallysimilar to the sourceof MORB from the 32-
3.0
39øEsegment of the SWIR.Thevariationof 2ø6pb/2ø4pb
with 87Sr/86Sr
andendis suggestiveof two-component
2.5
0
2.0 Figure 5. Selectedmajor elementoxidesand ratios versusMgO
0o for basaltsfrom the easternrifted margin. The solid heavy line

1.5
1.0
10 8 6 4
representsthe low-pressure (0.5 kb) liquid line of descent
calculatedusingthe programMELTS [Ghiorsoand Sack, 1995].
Note that for a given MgO content,the Mananjary basaltson
MgO (wt%) average have lower Na20 than those from Tamatave and
110 CRETACEOUS MADAGASCAR BASALTS

lOO

100
I Mananjary
high-Mg-
Ti
basalts
Tamatave basalts

50

10-- • '
5 m TAM92-3 ':;I

•
_
o MAN90-45
i MAN90-47
•e v TAM92-6 • TAM92-21B
o TAM92-9 • TAM92-27 "•
ß MAN90-85 ß TAM92-12 ß TAM92-43
• MAN90-86 , TAM92-16 + TAM92-49
1 • t i i i • i • i i i I i i i i •

100 - 100 --
_

50 -- 50-- Sambava basalts

Mananjary Fe-Ti basalts _

100_ 1O0

50
•• Mananjary
high-S/,
low
Fe-Ti
basaltsa ,/1!:•L•:-•:•.Marion/s/and
and
Funk
Seamount
50 •• ,?'W,,',%•,
, ./x o.'"'•-, -:.. •_

' -' 'i,• •- ,

lO

5 n WJM-21

•MAN90-35
ß WJM-49
,• WJM-50
u MAN90-40 ß D1-B1
ß MAN90-41
1-]--
ß D1-E
I I I I I I I [ '--T-F-I--F-r I- I 1 ]' I I
RbBaThNbTaLaCeSrNd P SmZr HfEu TiTb YYbLu RbBaThNbTaLaCeSrNd P SmZr HfEu Ti b YYbLu

Figure 6. Primitive-mantle-normalized incompatibleelement data for basaltsfrom the easternrifted margin of

Madagascarand lavasof Marion IslandandFunk Seamount(normalizingvaluesfrom Sunand McDonough[1989]).
Note the pronouncedBa spike shownby the high-Mg-Ti basaltsfrom Mananjaryand also most basaltsfrom the
Tamataveand Sambavatransects.Many of the high-Fe-Ti basaltsfrom Mananjaryshow pronouncedSr and Eu
anomalies,indicatingthat they have sufferedsignificantamountsof plagioclasefractionation,whereasothersare
clearlyaccumulative in plagioclase.
The high-Si,low-Fe-TibasaltsfromMananjaryarestronglydepletedin Nb andTa
relativeto otherhighly incompatibleelements,consistent
with significantcontamination
by continental
 STOREY ET AL. 111

38.5

• Sambava
• Tamatave

_ .,.•ROZET
39.0
• Mananjary
(high-Mg-Ti
rocks) t-88Ma ..
ß Mananjary
(Fe-Ti
series;
Ti02>
3wt%)
• 38.5 _ 0 Mananjary
(Fe-Ti
series;
Ti02<2 wt%) ß
ß•::'•?:'•UNION-
'-- MAURITIUS
/ ...... MARION
t=88 Ma
• 38.0

-Bekily dikes
O4 37.5

SWIR 39-41 ø E

37.0

36.5 I I I I I I I

15.7

..... CROZET
• r=uu Ma ..•

15.6
_ Ejeda-Bekily
dike
15.5
o l:, ::::::::::::::::::::::::
SWIR 39-41 ø E
15.4

15.3

15.2

15.1

15.0 ß I I I I I I
15.5 16.0 16.5 17.0 17.5 18.0 18.5 19.0 19.5

2o6pb/2O4pb
Figure 7. Initial (88 Ma) 206pb/204pb
versus207pb/204pb
and208pb/204pb
for basaltsfromtheeastern
riftedmarginof
Madagascarshowingdatafrom Table 3 andMahoneyet al. [ 1991]. Note that althoughsomeof the high-Fe-Ti basalts
from Mananjaryhave206pb/204pb
ratioswhichoverlapwith thoseof Marion,they are alsocharacterized
by higher
2ø7pb/2ø4pb
and2øSpb/204pb
ratios.ThefieldfortheEjeda-Bekily
dikeswarmis fromMahoneyet al. [1991]; thefields
for the SWIR (39-41øE) and the Marion R6union-Mauritiusand Crozethotspotsare from Mahoneyet al. [1992] and
referencestherein.Also shownis the the estimatedrangein Pb isotopiccompositionof the Marion and R6union
plumesat 88 Ma, assumingg valuesfor the respectivesourcesof 10 and 15, and r = 3.3 [seePeng and Mahoney,
1995]. NorthernHemisphereReferenceLine (NHRL) is from Hart [1984].

mixing [e.g.,Langmuiret al., 1978;Barling and Goldstein, well with isotopic composition.Secondly, these basalts
1990] betweena high-eNdcomponentwith low Nd/Pb and havehigher2ø7pb/2ø4pb
andhigher2øspb/2ø4pb
thanthose
Sr/Pb, and a low-sNd componentwith high Nd/Pb and fromtheMarionhotspotat comparable
2ø6pb/2ø4pb.
One
Sr/Pb. However, there are two discrepancies.The first is explanationis that minor amountsof continentalPb, with
that the variation in Nd/Pb and Sr/Pb ratios of Trend I high 2ø7pb/2ø4pband 2øspb/2ø4pbratios, were
basalts is small and these elemental ratios do not correlate incorporatedinto thesemagmasduringascentthrough
 112 CRETACEOUS MADAGASCAR BASALTS

2ø7pb/2ø4pband 2ø8pb/2ø4pb
without significantly

ßSambava
(high-Mg-Ti
rocks)
(> Tamarave
$ Mananjary

ß Mananjary
(Fe-Ti
seriesTiO2>
- O Mananjary
(Fe-Ti
3wt%)
seriesTiO
2<2wtø/o)
changing
the2ø6pb/2ø4pb
composition

which have high 2ø7pb/2ø4pb

is demonstrated
by
the stronglycontaminatedlavas in southernMadagascar,
and 2øspb/2ø4pb
for
2ø6pb/2ø4pb
ratios rangingbetween•18 and 18.8
[Mahoney et al., 1991; ms. in prep.]. An alternative
explanation is that the early Marion plume activity
/
/ I • • -.".l.-k. involved the melting out of a componentwith slightly
i' o 0 I •ß
%•:::•)Trend
-:.. higher2ø7pb/2ø4pb
and 2øspb/2ø4pb
than seenin the
Io ßß •.:..."I-|i•'::-•-\
•us/r/us present-dayhotspot.If this were the case it would make
/ •=• Marionmore conformablewith otherIndianOceanhotspot
CROZET
islands such as R6union and Crozet.

OriginoftheLow2ø6pb/
2ø4pb
Basalts
(Trend
II)
Trend II
-
TrendII showsdecreasing 2ø6pb/2ø4pb
with increasing
Ejeda-Bekilydikes
87Sr/86Sr,
contrasting
with the general
trendshownby
MORBandOIB of increasing 2ø6pb/2ø4pb
withincreasing
87Sr/86Sr(Figure8). Low 2ø6pb/2ø4pb is commonly
observedin lamproitesand continentalflood basalts,and
' SWIR
39-41
øE thereis evidencethat it can be a compositional featureof
bothcontinentallithospheric mantle[e.g.,Hawkesworthet
al., 1990] and lower crust [e.g., Moorbath and Welke,
i ! I i i i i

0.706 _ (•----.-• SWIR39-41

øE 1969].The low 2ø6pb/2ø4pb basaltsfromthe eastcoastof
Madagascaroffer no uniqueexplanationfor their origin,
\ Trend II
althoughthe generallack of strongelementalevidencefor
0.705 - \ \ ••'r•:••jeda-Bekilydikes
o \ •..'.•%.•..% REu~/o~-
crustalcontaminationin the samplesanalysedsuggests
the sourceof the unradiogenic
that
Pb is more likely to be the
mantlelithosphere.
Thepresence
of bothlow 2ø7pb/2ø4pb
0.704 -
and high 2ø7pb/2ø4pb
types implies continental
componentsof different ages;a single-stagePb growth
t•- 11&A,
TrendI model suggeststhe involvement of Early to Middle
•.• "" •o //•,•' • .... Proterozoic lithosphere in the origin of the high
0.703 - ---- 2ø7pb/2ø4pb-low 2ø6pb/2ø4pb basaltsfrom the Tamatave
and Sambavatransects,whereasthe low 2ø7pb/2ø4pb -
/' .J Atlantic, 2ø6pb/2ø4pb
basalts
indicate
a roleforArchean
lithosphere.
0.702 -
•.:.•-.... / Pacific Trend II can thus be viewed as being broadlyconsistent
MORB
i
with mixing of low 2ø6pb/2ø4pb
Madagascan
mantle-
15.5 16.o I
16.5 17.0 I I.
17 5 18.0 I 18.5 I I
19.0 19.5
lithosphere-derived
components of differentageswith an
2o6pb/2O4pb Indian Ocean normal-MORB mantle component(and
possiblyminor Marion hotspotinfluence).
Figure8. Initial SNdand87Sr/86Sr versus206pb/204pb for basalts Mantlewithhigh2ø7pb/2ø4pb
- low2ø6pb/2ø4pb
isotopic
from the easternrifted marginof Madagascar. The fieldsfor the compositionsis being tapped today along the SWIR
SWIR (32-39øE) and Atlantic and Pacific MORB are from between39ø and 41øE (Figures7 and 8). A proposed
Mahoneyet al. [ 1992] and referencestherein.Also shownare the
originfor theselow 2ø6pb/2ø4pb
MORBsis that they
estimatedrange in isotopic compositionof the Marion and
represent the remnants of ancient continental mantle,
R6unionplumesat 88 Ma. OtherdatasourcesareasFigure7.
thermally eroded by the Marion plume from Indo-
Madagascarin the middle Cretaceous[Mahoney et al.,
lithosphere.The presencein Madagascarof continental 1992]. The presenceof isotopicallysimilarbasaltsfrom the
lithospherewith the necessaryisotopiccompositionto lift Tamataveand Sambavatransectssupportsthis
 STOREY ET AL. 113

Volcande 191ndroy Zr, and Hf, flatterREE profiles[(LaJYb)N- 3] and very

pronounced,negative Ba, Eu, Sr, P, and Ti anomalies.
The Volcan de l'Androy massif is a bimodal basalt- SiO2 rangesfrom 71 to 76 wt.%. GroupII rhyolitesare
rhyolite associationwith volumetrically minor hybrid
magmas. The basalts are generally more evolved than 20
18i ß
AVolcan
do
I'Androy
[I
those of the east coast, althoughsome sparselyolivine-
phyric high-MgO (11-12 wt.%) lavas are present. The
Ejeda-Bekdy
d•kes East
coast
basalts
Fe-Ti enrichmenttrond..'

Volcan de l'Androy basalts form two compositionally

distinct groups (Groups ! and II). Group I (the most
abundant)has lower total iron, TiO2, and Nb than the
Group II basalts, at given MgO contents(Figure 9). In
mantle-normalizeddiagrams the Group I basalts show
pronouncednegative Nb, Ta, and Sr anomaliesand are
light-REE enriched(Figure 10). The Sr anomaliesmay be
a result of plagioclasefractionation,as Sr/Nd decreases
with decreasingMgO. The Group I magmas,which have
low initialendandveryhighinitial87Sr/86Sr
(>0.71;J. East
coast
basalts
ß o
Mahoney et al., ms. in prep.) almostcertainlyassimilated Fe- Ti enrichment trend

continental crust.
The Group II basalts,which are mainly found in the "rim
Volcan
deI•ndroy_•ß ./ ,,
series"(Figure 3) on the easternside of the massif, lack
Nb, Ta, and Sr anomalies (Figure 10). Their mantle-
normalizedpatternsshow strongsimilaritieswith thoseof
Group
IIbasalts
OIB, indicating that small-degree,alkalic mantle melts
were available during the formation of the Volcan de •.' Volcande I'Androy
Group I basalts
l'Androycomplex.The 87Sr/86Sr
ratiosof the GroupII
basaltsrangefrom 0.7059 to 0.7064,whichis substantially 100

lessthanthosefor GroupI but significantlyhigherthanfor 90

Marionhotspotbasalts(Figure8). The Pb isotopesare also
8o
distinctfrom those of the Marion hotspot,with much
higher2ø8pb/2ø4pb
ratios(>39;J. Mahoney
et al.,ms.in 70

prep.). This result leadsus to proposethat the Group II •- 60

basalts represent small-degreealkalic melts from the c•. 50
Volcande/'Androy
continental lithosphericmantle, as suggestedfor the z 40 GroupII basalts
associatedEjeda-Bekily dike swarm [Mahoney e! al.,
30
1991].
The absenceof eutaxitictexturesin the thick rhyolites 20

that are especiallyprominentin the upper part of the 10

sequencesuggeststhat these are lava flows rather than 0 Group_

I basalts
pyroclasticdeposits,despitehavingaspectratiosin excess 14 12 10 8 6 4

of 50:1. Most of the rhyolitesare phenocryst-poor.

As with
thebasalts,two distinctsetsof rhyolitesarepresent.Group
I rhyolites,like their basaltcounterparts,have relatively
low abundancesof Nb, Ta, Zr, and Hf. (La/Yb)N (N =
primitive-mantle-normalized)ratios are about 8, and these
rhyolites have moderatenegative Eu anomalies(Figure
10). SiO2 rangesfrom 65 to 80 wt.%, the highestvalues
probably reflecting alteration (Figure 11). The
microgranitessurroundingthe massifare compositionally
identicalto the GroupI rhyolites.The GroupII rhyolites,
on the other hand, have much higher contentsof Nb, Ta,
MgO (wt.%)
Figure 9. Fe203, TiO 2 and Nb versusMgO in basaltsfrom the
Volcan de l'Androy complex and the associatedEjeda-Bekily
dike swarm in the south of Madagascar.Data for Volcan de
l'Androy are from our work and will be presentedin full
elsewhere(J. Mahoney et al., ms. in prep.). The data for the
Ejeda-Bekily dikes are from Dostal et al. [1992] and A.D.
Saunderset al. (unpub.data).Note that the Volcan de l'Androy
basaltsfall into two compositionally
distinctgroups.Shownfor
comparisonis the compositionalrangeexhibitedby basaltsfrom
the easternrifted margin(dashed
114 CRETACEOUS MADAGASCAR BASALTS

1000 •-
¸ AND90-1 1000 Volcan
deI•ndroy c•AND90-6
_

Volcande I•ndroy Group I rhyolites ß AND90-17

Group
I basalts ß AND90-2
_

.,•_..,•,
- • AND90-71
_

[] AND90-7
_

•, ß AND90-8 100 "•]i•'• • ßAND90-72

100 •-
....
• ,, AND90-13
_

,'•• •. • AND90-60
.•...-.*• :• • : AND90-•
_

__

10

10-•

000 --
- Volcande It4ndroy c• AND90-75 1000
-• ! Volcan
Groupde I•Androy
II rhyolites ,, AND90-34
.......
- GroupII basalts ß AND90-76
• •. r•AND90-43
_

[] AND90-78

100 --
ß AND90-82
100 •,•..•---• ,• ßAND90-53
_

10

1 • - .....

1 ' 0.1 3 •- i i • i i• •1 i i i T--•7•

Rb Ba Th Nb Ta La Ce Sr Nd P Sm Zr Hf Eu Ti Tb Y Yb Lu RbBaTh Nb Ta La CeSr Nd P SmZr Hf Eu Ti Tb Y Yb Lu

Figure 10. Primitive-mantle-normalized incompatibleelementdatafor GroupsI and II basaltsand rhyolitesof Volcan

de l'Androy. Note the generalsimilarity betweenthe Group II basaltsand the Ejeda-Bekily dikes (shownby shaded
field; data from Dostal et al. [ 1992]).

presenton Vohitsimbe, where they are found interbedded [1991] noted that some of the Ejeda-Bekily dikes have
with both Group I rhyolitesand basalts.Group II basalts similarisotopiccompositions
to the low 2ø6pb/2ø4pb
are found in nearbydikes.Severalof the GroupII rhyolites MORBs found on the SWIR to the southof Madagascar.
are mildly peralkaline.The relativelylow concentrations of
Ba, light- and middle-REE, Eu, Sr, P, and Ti are consistent Majunga and Morondava Basins
with extraction of K-feldspar, plagioclase,apatite and
titanomagnetite,+ amphibole.These rhyolitesare almost Isotopeand trace element data for a suite of tholeiites
certainlyrelated to the Group II basaltsby differentiation from the southern part of the Morondava Basin were
(Figure 11), althoughisotopicdata (J. Mahoneyet al., ms. reportedby Mahoneyet aL [1991] andDostal et aL [1992].
in prep.) also indicate some crustal assimilation. The rockshave similarcompositions to the GroupI basalts
Conversely,the Group I rhyolitesmay be differentiatesof of Volcan de l'Androy, with negative Nb, Ta, Sr, and P
the Group I basaltsor somemixture of basaltand crustal anomalies,
high initial 87Sr/86Sr
(>0.71), and low •Nd
melts; the Nb and Zr abundancesin the Group I rhyolites (<-6), rather similar to the Bushe Formationof the Deccan
(27-67 ppm and >500 ppm, respectively)are too high for Traps in India [Mahoney et al., 1991]. The isotopicdata
them to be pure crustalmelts. andvariationsin REE ratiosare entirelyconsistent with the
magmashaving assimilatedsialiccrust.
Ejeda-Bekily Dike Swarm The compositional similarity between the Volcan de
l'Androy and southernMorondava Basin basalts (Figure
These dikes (Figure 2) have been discussedextensively 12, top) suggeststhat lavas may well have been able to
by Mahoney et al. [ 1991] and Dostal et al. [ 1992]. They travel across the island and into the western basins. The
havelowinitial•Ndand206pb/204pb
ratiosandareunusual distancefrom Volcan de l'Androy to the region southeast
among the Madagascar basalts in being nepheline- of Tulear is about 200 km. The northernmostoutcropsin
normative.In terms of their trace elements,they are similar the Tulear area would have necessitated flows some 300-
to the Group II basaltsof Volcan de l'Androy but have 350 km long. These are not excessivedistanceswhen
higher Ba/Ta and Ba/La ratios (Figure 10). Mahoney et al. comparedwith large flood basaltflows, suchas those
 STOREY ET AL. 115

250
ß Voh•s•mbeRhyol•tes
(c)
Ot•r AndroyRhyol•tes
200
ß AndroyBasalts

150 GroupII Rhyol,es

100

ß Gm• II R•es

60 65 70 75 80 85 0 200 400 600 800 1000 12• 1400 1600

SiO2 (wt%) Zr (ppm)

250 1.2

200 ½•-½
ßß ......... •----•-• (b) 1.0

C•/e• Group
IIRl-•olites 0.8

150
lOO I-' 0.4
,
GroupII Rl-•olites
0.2

o I••.•"• 0.0

0 5 10 15 20 25 30 35 40 0 50 100 150 200 250

FeO(t)/MgO Nb (ppm)

Figure 11. Compositionalvariationsin rhyolitesfrom Volcan de l'Androy.(a) Nb vs. SiO2, illustratingthe marked
differencebetweenthe GroupI and II rhyolites.Note that both typesof rhyolite are interbeddedwith basaltsin the
Vohitsimbelava succession.(b) FeO(t)/MgO versusNb, showingthe relationshipbetweenthe GroupI andII rhyolites,
and the Group I and II basalts(FeO(t)=total Fe expressedas FeO). (c) Zr versus Zr/Nb. The very high Zr
concentrationsof the GroupII rhyolitessuggestthatthey arethe productsof largeamountsof crystalfractionationof a
basalticparent.(d) Nb versusTh/Nb. Note the trendsto high Th/Nb in both the basaltsand rhyolitesof Group I,
indicatingcrustalcontamination.

the Columbia River province [e.g., Tolan et al., 1989;
Hooper,this volume].
The basaltsof the Majunga Basin, in the northwestof
Madagascar,are compositionallydifferentfrom the Tulear
basalts.They show greater diversity althoughthis may
simply reflect the wider area sampled.As a whole, they
straddlethe normativequartz-,olivine-tholeiitedivide,but
in termsof trace elementscanbe dividedinto threegroups:
a groupwith large negativeNb and Ta anomalies,a group
with convex-upwardmantle-normalizedpatternsand low
abundancesof heavy REE, and a group with convex-
upward mantle-normalized patterns and greater
abundancesof heavy REE (Figure 12) [cf. Melluso et al.,
1997]. The last two groupscan be related by different
pressuresof melting; the first groupmay have undergone
some crustal assimilation.
DISCUSSION
Marion Plume Mantle in Fe-Ti BasaltPetrogenesis
Despite the evidence that mantle plumes play an
importantrole in flood basaltformation,few geochemical
studieshave providedunequivocalevidencethat hotspots
contributematerially to the magmatism.Exceptionsare
isotopicstudieson the northwesternDeccan Traps [Peng
and Mahoney,1995] andEastGreenlandTertiarytholeiites
[Thirlwall et al., 1994] which respectivelyrevealed the
presence of R6union and Iceland mantle plume
componentsin the eruptedmagmas.The Mananjary Fe-Ti
basaltsalso provide good isotopicevidencefor a plume-
type component. Trend I (Figure 8) can largely be
accountedfor by mixing between a mantle end-member
somewhatsimilar to that being tapped today at Marion,
Prince Edward, and Funk volcanoes,and a high eNd-low
87Sr/86Sr
component
similarto normalSWIR MORB
mantle.
That mantlerathersimilar in isotopiccompositionto the
present-dayMarion hotspotmagmaswas importantat 88
Ma in Madagascaris of relevancein understanding the Pb
isotopicevolution of Indian Ocean mantle plumes.It has
beenproposed
thatlarge2ø6pb/2ø4pb
variations
alongthe
NinetyeastRidge, presumedto be the 82-38 Ma track of
the Kerguelen plume, are essentially an aging effect
resulting
froma highit (238U/2ø4pb)
valuein theplume
source[Class et al., 1993]. This idea was disputedfor the
R6unionhotspotcaseby Peng and Mahoney[ 1995] on
116 CRETACEOUS MADAGASCAR BASALTS

TulearBasalts - TUL90-2 increasingdistancealongthe rift axisawayfromtheplume

,•, /• (Morondava
Basin)
ßTUL90-7 center,becauseof the decreasingpotentialtemperatureof
e.._ •', •• ;;-% rj AND90-2
ß • ,/•.'• the sublithospheric mantle(Figure 13). If hydratedportions
of the continentalmantle are also undergoingmelting in
responseto extension and conductive heating by the
10--
plume, then the amount of melt producedmay be less
.,,:,• sensitive to changes in potential temperature of the
asthenosphere beneatha continentalrift (Figure 13). In this
scenario, the proportion of melt generated from the
lithosphericmantle relative to melt producedfrom the
plume should become higher with increasingdistance
I I I [ I ', I I I I •I----T--3 T •
alongthe rift axis away from the plume center.
Likewise,asthe degreeto whichthe lithosphere hasbeen
Majunga
Basalts MJ92-2 thinned by extensiondecreasesinland from the riff, the
(Majunga
Basin) ßß IVU92-7
MJ92-6
E 50 ß MJ92-11
extent of melting within the plume will become
•: MAN90-6 increasinglyrestricted by the "lid" effect and melts
generatedwithin the continentalmantle may predominate.
/
! •' []
, • .•,
_ ., _
• •e,•-ß,, .;,•.ß--• If the lithosphereis thicker than about 130 km, then most
/ik,.,
-•' ._.•_.•'•'•--.•,'•',,•,-[]_ ',•,-•, of the melt will be generatedby "wet" continentalmantle
as the "dry" plume solidusis not crossed(Figure 13). Do
the MadagascarCretaceousvolcanic rocks fit with these
predictions?Here we consider the significanceof the
regional geochemicalvariationsin the basaltsalong the
riftedmarginandinlandat Volcan de l'Androy.
Using the plate motion model of Miiller et al. [1993],
RbBa•NbTa •Pb SrNd P SmZr •Eu • • Y YbLu
Storeyet al. [ 1995] placedthe Marion hotspotwithin about
100 km south of Madagascar around the time of
continentalbreakup with India. They suggestedthat,
Figure 12. Primitive-mantle-normalizedincompatibleelement
datafor somebasaltsof the Majunga and Morondavabasins(A. within the uncertaintyof the plate reconstruction, Volcan
D. Saunderset al., unpub.data). Note the similarityof the Tulear de l'Androy marked the focal point of the Marion plume at
patternsandsomeGroupI basaltsfrom Volcande l'Androy (e.g., 88 Ma. The isotopic evidence for a Marion plume
AND90-2). Likewise, the patternsof the Majungabasaltsshow componentin the Mananjary Fe-Ti basaltsis consistent
similaritieswith the eastcoastFe-Ti basalts(e.g.,MAN90-6). with the plume being situatednear the southernend of the
continentalriff. In contrast, all of the basaltsfrom the two
basis of isotopic data for basalts and picrites of the more northerly transects(Tamatave and Sambava)have
northwest Deccan Traps. There, the Pb isotopic low 2ø6pb/2ø4pb
ratios,reflecting
significant
continental
compositionof the plume appearsto have changedlittle lithosphericinfluences.For comparableMgO contents,the
since the Late Cretaceous.The MadagascarFe-Ti basalts Tamataveand Sambavabasaltshave higherNa20 contents
alsodo not requirelargechanges
in 2ø6pb/2ø4pb
in the (Figure 5), which may indicatea decreasein the amountof
Marion plume in the last 88 m.y. The implicationis that melting northward along the rift axis, althoughsource
the R6union and Marion plumes have much smaller it heterogeneityis another possibleexplanationfor the Na
variation.
values(<15) thanrequiredby the Classet al. [1993] model
(it > 30). This is supportedby measuredit values in The total absenceof Marion-like isotopiccompositions
R6union subalkalic basalts, which are •20, in general in the basaltsof the centraland northernpart of the eastern
[Pengand Mahoney, 1995]. rifted margin suggests that plume melts were only
available in the south of the island, or that the plume
Significanceof Spatial CompositionalVariationsin the signaturewas maskedby lithosphericand/orMORB-type
Basalts mantle components.The isotopic data indicate that the
sourcesof the basaltsin the north are mainly a mixtureof
A prediction from dry-melting models for a mantle continentalmantle lithosphereand Indian MORB-source
plume that underliesa zone of continentalextensionis that mantle. Supportfor a greaterinvolvementof lithospheric
the amountof melting shoulddecreasesystematically with mantle in the north is also providedby the positive
 STOREY ET AL. 117

Extension

• ernperature
!.ø•
) •ø

Melts
predominantly
from Rift
asthenosphereaxis .........
__
,,, ,,,,

Wet
•ølidUS12;0
oC
158,.0
oC200

50

I Melting
oflithosphere
mantle
I byplume
heating
anddecompression -•e•,
....::::-'---- 100
krn
. ',,,garnet
- 21 150

200

! .
lOO
krn
Meltingof lithospheremantle
150
byplume
heating Lithosphere
I'd• øc20O

Asthenosphere

Figure13. Sketchillustrating
howtheproportion of mcl!froma "dry"mantle plume, relative
tomeltderived from
"wet"lithospheric
mantle,mayvarybeneath a developingcontinental
rift.If thelithosphereisthicker
thanabout130
km,thenmostmeltwillbegeneratedwithinthelithospherebecausethe"dry"plumesolidus (potential
temperatureof
.•1580øC)is notcrossed.
However,withtimecontinued extensionandthinning of thelithospherewill resultin
enhanced dccomprcssional
melting
of theplume, such thatplume-derived meltswill showanincreasing dominance
overthosederivedfromthe continental
mantle.In additionto thistemporalvariation,therealsomay be a spatial
variation
alongtheriftaxiswitha decrease
intheproportion
ofplume-derived
meltswithincreasing
distance
fromthe
plumecenter.[3istheoriginal lithosphere
thickness
divided
bytheextended thickness.
Plumeisotherms
arefrom
Watson andMcKenzie [1991].Solidus
curvesandthegarnct-spincl
transition
deptharefromMcKenzie andBickle
[ 1988] andMcKenzieand O'Nions[ 1995].

anomalies
shownby the Sambava
andTamatave
basaltsin mixing between asthenosphere-derived magmas and
primitive-mantle-normalized
diagrams(Figure 6). For magmas(high Ba/Nb) derived from the continental
example,highBa/Nbis observed
in lateCenozoicbasalts lithosphericmantle [Fitton et al., 1988]. The east-coast
eruptedin an extensional
environment in the western rifted margin basaltsof Madagascarmay record the
United States[e.g., Fitton et aL, 1988]. Someof these transitionbetweena plume-dominated sourcein the south
Cenozoic lavas exhibit a negative correlationbetween and a mixture of continentallithosphericmantle and
143Nd/144Nd and La/Nb, a Featureinterpretedto indicate MORB mantle sources in the
118 CRETACEOUS MADAGASCAR BASALTS

0.8
ß Volcande I'Androy(GroupII basalts)
Figure 14 showsa primitive-mantle-normalized
ratio of a
t lO
A Ejeda-Bekilydikes lightREEto a middleREE(Ce/Sm(N))plottedagainsta
ß Sambava • Tamatave
ratioof a heavyREE to a middleREE (Yb/Sm(N))for
Spinel
Iherzolite
melts •
ß
Mananjary(high-Mg-Tirocks)
Mananjary(Fe-Tiseries;TiO2>3wt%) basaltsof the easternrifted margin, Volcan de l'Androy,
•• 5 ' :ß:::•i'::;
.....::lower O Mananjary(Fe-Tiseries;TiO2<2
wt%) and the Ejeda-Bekily dike swarm. Sampleswith obvious
0.6
0 >' signsof crustalcontamination(e.g., high SiO2 and La/Nb)
ee ,:'•,'•. 2 :::::::7:
are not plotted. Also shown are the calculated
compositions of aggregatedmelts producedby fractional
:upper
crust meltingof garnetlherzoliteand spinellherzolite.Because
the ratio on each axis has a common denominator, trends
0.4
causedby mixing of meltsproducedfrom garnetlherzolite
with melts producedfrom spinellherzolitewill be linear
[e.g., Thirlwall et al., 1994], as suggestedby the steep
array defined by the Fe-Ti basaltsand the high Mg-Ti
0.2
Marion - Prince Edward
basaltsfrom the rifted margin.Interestingly,this array also
•.10 overlaps compositions of Marion Island and Funk
• '•.....•::FunkSeamount
Seamountlavas, perhaps suggestingthat the potential
Garnet Iherzolite melts temperature of the Marion plume has not changed
significantlywith time. The lowermelt productivityof the
0
0
i
1
i
2
i
3
i
4 5
present-day Marion hotspotmay resultfrom a combination
of its intraplatesetting[cf. Ellam, 1992] and a lower flux
Ce/Sm(N)
of plume material. The interceptson the meltingcurves
Figure 14. The variationof the light to middleREE (Ce/Sm)and couldsuggestthat the Fe-Ti basaltsformedby mixing of
heavyto middle REE (Yb/Sm) ratiosfor MadagascarCretaceous meltsproducedby =4% melting of garnetlherzoliteand
volcanic rocks, normalized(N) to primitive-mantlevalues. The =10% meltingof spinellherzolite.The Volcande l'Androy
compositionsof aggregatedmeltsproducedby differentdegrees GroupII basaltsshowevidencefor residualgarnetin their
of melting of a spinellherzoliteand garnetlherzolitesourceare source,and they appearto have beenproducedby lower
shown. The sourceand melting modesfor spinel lherzolite are extentsof melting than the east coast lavas. The Ejeda
from Kelemenet al. [ 1992], the sourcemodefor garnetlherzolite Bekily dikes define two compositionalgroups.One group
is from S. Bernstein et al. (Plume-related, post-rift basaltic
appearsto representsmall amounts(=3%) of melting of
magmatismalong the East Greenlandmargin,ms. submittedto
Earth Planet. Sci. Lett.) andthe garnetlherzolitemeltingmodeis
spinellherzolite;the othergroupsuggests contributions of
from Walter and Presnall [1994]. The meltingmodelis basedon melt producedby small-degreemelting of both garnet
the fractionalmelting equationsof Gast [1968] andShaw [ 1970], lherzoliteand spinellherzolitesources.Thesecalculations
andpartitioncoefficientdatafrom Kelemenet al. [ 1993], Shimizu neglectthe sourceheterogeneityindicatedby the rangein
and Kushiro [1975], and Hart and Dunn [1993]. The Sr, Nd, and Pb isotopes.
compositionof the model mantle is one of slightREE depletion In summary,the REE variationsare consistentwith the
(Ce/Sm(N)
= 0.83).All of theMadagascar
datashownarefrom largestextentof melting occurringalongthe riftedmargin,
our work, except for some additional analysesof the Ejeda- in responseto extensionof the Madagascanlithosphere
Bekily dikes from Dostal et al. [1992]. Continental crust
and mantle decompression. The lower Na20 contentsof
compositionsare from Taylor and McLennan[1985].
the Mananjary Fe-Ti basaltsmay also suggestan increase
in mantletemperatures and degreeof meltingto the south.
In addition to these along-rift-axisvariations,isotopic Paradoxically, despite likely higher sublithospheric
dataon the Group II basaltflows of Volcan de l'Androy(J. potential temperatures, smaller melt fractions were
Mahoney et al., ms. in prep.) and the associatedEjeda- generatedat Volcan de l'Androy, which is presumedto
Bekily dike swarm (Figure 8), situatedinland from the mark the center of the Marion plume at 88 Ma. Isotopic
rifted margin in the south of the island but above the and elemental data on the Volcan de l'Androy Group II
postulatedfocal point of the plume, are dominatedby basalts and the Ejeda Bekily dikes suggestthat they
componentsfrom the continentalmantle. The lack of representsmall-degreemeltsof spinellherzoliteandgarnet
Marion hotspot compositionsin the Volcan de l'Androy lherzoliteof the continentallithosphere.
basaltsand the Ejeda-Bekily dikes may be causedby the The observed spatial variations in isotopic and REE
thickness of the lithosphererestricting the amount of compositionsof the basalts that form the Madagascar
meltingof the plume. Cretaceous volcanicprovincebetterfit with the notion

were producedfrom a smallerthermaland compositional
mantle anomaly than the large starting-plume head
envisagedfor the origin of somefloodbasaltprovinces.
APPENDIX 1 (PETROGRAPHIC DESCRIPTIONS)
MAN 90-lB. Granularto subophiticbasalt.Trace quantitiesof
brown smectite pseudomorphs, possibly after olivine.
Abundant titanomagnetite.Clinopyroxeneis a pink-brown
color,reflectingthe high Ti contentof the sample.
MAN 90-2. Fine-grained granular-texturedbasalt. Contains
approximately3% partiallyresorbedand fracturedplagioclase
phenocrysts.Alteration is restrictedto localizedpatchesof
greenclaysin the groundmass.
MAN 90-6. Fine-grained, sparsely plagioclase-and clinopy-
roxene-phyric basalt. The majority of the plagioclase
phenocrystsare found in glomerocrystswith subordinate
amountsof clinopyroxene.Isolatedphenocrystsof plagioclase
are partiallyresorbed.The groundmass comprisesplagioclase
and granularclinopyroxeneand titanomagnetite. Alterationis
.slight,restrictedto dark-brownclay pseudomorphs,
possibly
afterolivine (<1% of rock).
MAN 90-8. A very fine grained,nearlyaphyric,almostaphanitic
roqk. Contains plagioclase,equant opaque grains, and
clinopyroxene.
MAN 90-15. Aphyric basalt,with possibleinfilling of vesicles
with prehniteand pumpellyite.
MAN 90-16. Very fine-grainedbasalt,with approximately5%
clinopyroxenephenocrysts. The rock showslittle evidencefor
secondaryalteration.
MAN 90-17. Olivine gabbro.This is a plagioclaseorthocumulate,
with strongly colored (brown) intercumulusclinopyroxene.
Abundant(•5%) well-roundedolivine crystals.Alterationis
slightandrestrictedto localisedformationof clay.
MAN 90-20. Strongly plagioclase-phyric basalt with trace of
clinopyroxene.Probably partially accumulative.No visible
signof alteration.
MAN 90-22/2. Vesicular, variolitic, aphyric basalt.Vesiclesare
filled with very pale-greenchlorite/smectite.
MAN 90-31. Sparselyplagioclase-phyric dolerite, with brown
clays replacing the mesostasis. Some clays may be
pseudomorphing olivine.
MAN 90-43, -49. Aphyric dolerite. Yellow-brown clays form
patches in mesostasis.Trace amounts of partially resorbed
plagioclasephenocrysts.
MAN 90-45,-47. Olivine-clinopyroxene-phyric basalts.
MAN 90-57. Aphyric dolerite.
MAN 90-63,-71. Sparselyplagioclase-phyricdolerites.Green-
brown smectiteis presentin mesostasis.
MAN 90-66. Medium- to coarse-grainedplagioclase-rich
dolerite. Subophitic texture between clinopyroxene and
plagioclase; •,3% opaque minerals, probably mostly
titanomagnetite. Several patches of mesostasisinterstitial
between plagioclase and clinopyroxene are replaced by
yellow-greenclay (•2% of rock).
MAN 90-77. Plagioclase-richdolerite with an "ophimottled"
appearancecaused by abundant ophitic glomerocrystsof
plagioclase and clinopyroxene. The areas around the
STOREY ET AL. 119
glomerocrysts comprise plagioclase, clinopyroxene,
titanomagnetiteand an altered mesostasis(now yellow-green
clay, •4% of rock).
MAN 90-80. Plagioclase-,titanaugite-phyricbasalt.Groundmass
is hyalopilitic, with abundantopaques.No visible signs of
alteration.
MAN 90-81. Aphyric dolerite, with dark-green smectite in
patchesof mesostasis.
MAN 90-85. Sparsely clinopyroxene- and plagioclase-
microphyricbasalt.Unusualopaque-rimmedocelli with cores
filled with an unidentified,low-birefringencemineral.
MAN 90-86. Medium-grainedclinopyroxene-microphyric basalt.
Chlorite-filled vesicles.
TAM 92-6. Aphyric, plagioclase-richdolerite, with abundant
titanomagnetite.Mesostasisreplacedby pale green smectite,
and calcite.
TAM 92-27. Coarse-grained,aphyric basalt; 5-10% brown
smectiteclaysreplacemesostasis.
TAM 92-30A. Thin section shows segregation of mafic
(clinopyroxene)and felsic (plagioclase)phases.Alteration is
pervasive,with green-coloredclays, calcite and occasionally
quartz. The plagioclaseis fractured and partially altered to
clays.
TAM 92-43. Aphyric, coarse-grainedbasalt.Brown clays replace
the mesostasis.
SAM 92-1. Aphyric, fine-grained, granular-texturedbasalt.
Brown clay replacesthe mesostasis.
SAM 92-3. Plagioclase-phyricbasalt, with •10% plagioclase
phenocrysts.The groundmass hasan ophimottledtexture,with
clinopyroxeneforming glomero-ophiticclusters.No alteration
is visible.
SAM 92-10, -16. An aphyric basalt, although clinopyroxene
forms glomero-ophitic clusters, giving an ophimottled
appearance.Some of the pyroxenemay be orthopyroxeneor
low-Ca pyroxene. Orange-brownclays have replacedpart of
the mesostasis.
SAM 92-20D. Fine-grained basalt. Abundant pseudomorphsof
pale-greenclay may be replacingmicrophenocrysts
of olivine
and/or clinopyroxene. The iron-rich mesostasisis almost
opaque.
SAM 92-33B. Plagioclase-phyric basalt. The groundmass
plagioclase is strongly flow-aligned. The groundmassalso
contains granules of clinopyroxene, olivine, and
orthopyroxene.
Acknowledgments.Field work in Madagascar was made
possible through A. Razafiniparany (University of
Antananarivo).A. Randriamanantenasoa and M. Rajaoherinirina
provided field support.Anton le Roex donatedthe Marion and
Funk samples.Chris Hawkesworthprovidedgenerousaccessto
the Open University isotope laboratory. Peter Van Calsteren,
Janet Hergt and Mabs Klunker gave analytical advice and
assistance.Nick Marsh is thankedfor carryingout XRF analyses
of the Madagascarrocks,and. StefanBernsteinfor discussionon
modelling of REE data. Rob Ellam, Godfrey Fitton, and
120 CRETACEOUS MADAGASCAR BASALTS

Harris critically reviewed the manuscript. This work was structurein the WesternSomali Basin, Geophys.d. R. Astron.
supportedby the NERC, NSF, and Danish National Research Soc., 86, 331-369, 1986.
Council. Cox, K. G., The role of mantle plumes in the developmentof
continentaldrainagepatterns,Nature, 342, 873-876, 1989.
REFERENCES Crawford, A. R., Narmada-Son lineament of India traced into
Madagascar,d. Geol. $oc. Ind., 19, 144-153, 1978.
Agrawal, P. K., O. P. Pandey, and J. G. Negi, Madagascar:a
de Wit, M. J., M. Jeffrey, H. Bregh, and L. Nicolaysen,
continental fragment of the paleosuperDharwar craton of
Geological Map of Sectionsof GondwanaReconstructedto
India, Geology,20, 543-546, 1992.
their Disposition ca 150 Ma, Univ. Witwatersrand,
Ando, A., N. Mita, and S. Terashima, 1986 values for fifteen GSJ
Johannesburg, 1988.
rock reference samples, "Igneous rock series", Geostand.
Dixey, F., The geologyand geomorphology of Madagascarand a
Newslett., 11, 159-166, 1987.
comparisonwith easternAfrica, d. Geol. $oc. Lond., 116, 255-
Arndt, N. T., and U. Christensen,The role of lithosphericmantle
268, 1960.
in continental flood volcanism- thermal and geochemical
Dostal, J., C. Dupuy, C. Nicollet, and J. M. Cantagrel,
constraints,dr.Geophys.Res.,97, 10,967-10,981, 1992.
Geochemistryand petrogenesisof Upper Cretaceousbasaltic
Barling, J., and S. L. Goldstein,Extreme isotopicvariationsin
Heard Island lavas and the nature of mantle reservoirs,Nature,
rocks from southern Malagasy, Chem. Geol., 97, 199-218,
1992.
348, 59-62, 1990.
Dupr6, B., and C. J. All•gre, Pb-Sr isotopevariation in Indian
Bassias,Y., and L. Leclaire,The Davie Ridge in the Mozambique
Oceanbasaltsand mixing phenomena,Nature, 303, 142-146,
Channel: Crystalline basementand intraplatemagmatism,N.
1983.
jb. Geol. Palaont. Mh., 2, 67-90, 1990.
Dyment, J., Structureet 6volution de la lithospher•reoc6anique
Battistini, R., La structuredu massif volcaniquede l'Androy
dansl'oc6anIndien: Apport desanomaliesMagn6tiques,Th•se
(Madagascar),Bull. Geol. Soc.France, 7, 187-191, 1959.
doct.,374 pp., Univ. Louis Pasteur,Strasbourg,France,1991.
Besairie,H., Geologicalmap of Madagascar(3 Sheets).Serv.
Ellam, R. M., Lithospheric thickness as a control on basalt
Gdol.Madagascar, 1964.
geochemistry,Geology,20, 153-156, 1992.
Besairie,H., Report 172, Serv. Geol. Madagascar,1966.
Ellam, R. M., and K. G. Cox, An interpretationof Karoo picrite
Besairie, H., The Precambrian of Madagascar, in The
basaltsin termsof interactionbetweenasthenospheric magmas
Precambrian, edited by K. Rankama, pp. 133-142, Wiley,
and the mantle lithosphere,Earth Planet. $ci. Lett., 105, 330-
London, 1967.
342, 1991.
Besairie,H., and M. Collignon, G6ologiede Madagascar- I, les
Farnetani,C. G., and M. A. Richards,Numerical investigationsof
terrains s6dimentaires,Ann. Gdol. Madagascar, 35, 463 pp.,
1972. the mantle plume initiation model for flood basaltevents,d.
Geophys.Res., 99, 13,813-13,833, 1994.
Boast, J., and A. E. M. Nairn, An outline of the geology of
Fitton,J. G., G. James,P. D. Kempton,D. S. Ormerod,and W. P.
Madagascar,in The Ocean Basinsand Margins, vol. 6, edited
Leeman,The role of lithosphericmantle in the generationof
by A. E. M. Nairn and F. G. Stehli, pp. 649-696, Plenum,New
Late Cenozoic basic magmasin the westernUnited States,d.
York, 19820
Petrol., Spec. Vol., 331-351, 1988.
Brooks, C. K., L. M. Larsen, and T. F. D. Nielsen, Importanceof
Fitton, J. G., A.D. Saunders,L. M. Larsen, B. S. Hardarson,and
iron-rich tholeiitic magmas at divergent plate margins: A
M. J. Norry, Volcanic rocks from the southeastGreenland
reappraisal,Geology,19, 269-272, 1991.
Carlson,R. W., G. W. Lugmair, and J. D. Macdougall,Columbia
margin
at63øN:composition,
petrogenesis
andmantle
sources,
in Proc. OceanDrill Prog., Sci. Results,152, editedby H. C.
River volcanism: the question of mantle heterogeneityor
Larsen,A.D. Saundersand S. Wise, OceanDrilling Program,
crustal contamination, Geochim. Cosmochim.Acta, 45, 2483-
2499, 1981.
College Station,Tex., in press,1997..
Flores, G., The SE Africa triple junction and the drift of
Class,C., S. L. Goldstein,S. J. G. Galer, and D. Weis, Young
Madagascar,d. Petrol. Geol., 7, 403-418, 1984.
formationage of a mantleplume source,Nature, 362, 715-721,
1993. Fourno, J.P., and J. Roussel, Imaging the Moho depth in
Madagascarthroughthe inversionof gravity data:geodynamic
Coffin, M. F., and O. Eldholm, Large igneousprovinces:Crustal
implications,Terra Nova, 6, 512-519, 1994.
structure, dimensions, and external consequences,Rev.
Gallagher,K., and C. J. Hawkesworth,Dehydrationmelting and
Geophys.,32, 1-36, 1994.
the generationof continentalflood basalts,Nature, 358, 57-59,
Coffin, M. F., and P. D. Rabinowitz, Reconstruction of
1992.
Madagascarand Africa- evidence from the Davie fracture-
Gast, P. W., Trace element fractionationand the origin of
zone and Western Somali Basin,d. Geophys.Res., 92, 9385-
9406, 1987.
tholeiitic and alkaline magma types, Geochim.Cosmochim.
Acta., 32, 1057-1068, 1968.
Coffin, M. F., and P. D. Rabinowitz,Evolutionof the conjugate
Ghiorso, M. S., and R. O. Sack, Chemical mass transfer in
East African-Madagascanmargins and the western Somali
magmatic processes,IV. A revised and intemally consistent
Basin, Geol. Soc.Am. Spec.Pap., 226, 78 pp., 1988.
Coffin, M. F., P. D. Rabinowitz, and R. E. Houtz, Crustal
thermodynamicmodel for the interpolationand extrapolation
of liquid-solid equilibria in magmatic systemsat

temperaturesand pressures,Contrib. Mineral. Petrol., 119,
197-212, 1995.
Goslin, J. J., J. S6goufin,R. Schlich,and R. L. Fisher,Submarine
topographyand shallow structureof the MadagascarRidge,
Bull. Geol. Soc. Am., 91, 741-753, 1980.
Green,A. G., Sea floor spreadingin the MozambiqueChannel,
Nature, 236, 19-21, 1972.
Hamelin,B., B. Dupr6,andC. J. Allt:gre,Pb-Sr-Ndisotopicdata
of IndianOceanridges:new evidenceof largescalemapping
of mantle heterogeneities,Earth Planet. Sci. Lett., 76, 288-
298, 1986.
Harland,W. B., R. L. Armstrong,A. V. Cox, L. E. Craig,A. G.
Smith, and D. G. Smith,,4 GeologicTime Scale, 263 pp.,
CambridgeUniversityPress,Cambridge,1989.
Harry, D. L., and W. P. Leeman, Partial melting of melt
metasomatised subcontinentalmantle and the magmasource
potential of the lower lithosphere,d. Geophys.Res., 100,
10,255-10,269, 1995.
Hart, S. R., A large scale isotopic mantle anomaly in the
SouthernHemisphere,Nature, 309, 753-757, 1984.
Hart, S. R., and T. Dunn, Experimentalcpx/meltpartitioningof
24 trace elements,Contrib. Mineral. Petrol., 113, 1-8, 1993.
Hawkesworth,C. J., P. D. Kempton,N. W. Rogers,R. W. Ellam,
and P. W. Van Calsteren,Continentalmantlelithosphere,and
shallowlevelenrichmentprocesses in theEarth'smantle,Earth
Planet. Sci. Lett., 96, 256-268, 1990.
Hergt, J. M., D. W. Peate, and C. J. Hawkesworth, The
petrogenesisof Mesozoic Gondwana 1ow-Ti flood basalts,
Earth Planet. Sci. Lett., 105, 134-148, 1991.
Hinz, K., A hypothesison terrestrialcatastrophes.
Wedgesof
very thick oceanward dipping layers beneath passive
continentalmargins,Geol. d., E22, 3-28, 1981.
Katz, M. B., and C. Premoli, India and Madagascar in
Gondwanalandbased on matching Precambrianlineaments,
Nature, 279, 312-315, 1979.
Kelemen, P. B., H. J. B. Dick, and J. E. Quick, Formation of
harzburgiteby pervasive melt/rock reaction in the upper
mantle, Nature, 358, 635-641, 1992.
Kelemen,P. B., N. Shimizu,and T. Dunn, Relativedepletionof
niobium in some arc magmas and the continental crust:
Partitioningof K, Nb, La and Ce during melt/rockreactionin
the uppermantle,Earth Planet. Sci. Lett., 120, 111-134, 1993.
Larsen,H. C., and S. Jakobsd6ttir,Distribution,crustalproperties
and significanceof seawards-dipping sub-basement reflectors
of E Greenland,in Early Tertiary Volcanismand the Opening
of the NE ,4tlantic,Spec.Publ. 39, editedby A. C. Morton and
L. M. Parson,pp. 95-114, The GeologicalSociety,London,
1988.
Langmuir,C. H., R. D. Vocke Jr., G. N. Hanson,and S. R. Hart,
A general mixing equationwith applicationsto Icelandic
basalts,Earth Planet. Sci. Lett., 37, 380-392, 1978.
Laughton, A. S., General BathymetricChart of the Oceans,
CanadianHydrographicService,sheet5-05, 1975.
Mahoney, J. J., J. H. Natland, W. M. White, R. Poreda, S. H.
Bloomer,R. L. Fisher, and A. N. Baxter, Isotopicand
geochemical
provincesof the westernIndianOceanspreading
centers,d. Geophys.Res.,94, 4033-4052, 1989.
STOREY ET AL. 121
Mahoney,J., C. Nicollet, and C. Dupuy, Madagascarbasalts:
tracking oceanic and continental sources,Earth Planet. Sci.
Lett., 104, 350-363, 1991.
Mahoney,J. J., A. P. le Roex,Z. X. Peng,R. L. Fisher,andJ. H.
Natland,Southwestern limitsof Indianoceanridgemantleand
theoriginof low2ø6pb/2ø4pb
mid-ocean
ridgebasalt:
isotope
systematics
of thecentralSouthwest
IndianRidge(17-50øE),
d. Geophys.Res., 97, 19,771-19,790, 1992.
McElhinny, M. W., B. J. J. Embleton,L. Daly, and J.P. Pozzi,
Palaeomagnetic evidencefor the locationof Madagascarin
Gondwanaland,Geology,4, 455-458, 1976.
McKenzie,D., andM. J. Bickle,The volumeandcomposition of
melt generatedby extensionof the lithosphere,d. Petrol.,29,
625-679, 1988.
McKenzie,D., and R. K. O'Nions,The sourceregionsof ocean
islandbasalts,d. Petrol., 36, 133-159, 1995.
Melluso,L., V. Morra,P. Brotzu,A. Razafiniparany, V. Ratrimo,
and D. Razafimahatratra,Geochemistryand Sr-isotopic
compositionof the Late Cretaceous flood basaltsequenceof
northern Madagascar: petrogenetic and geodynamic
implications, d. AJ?.Earth Sci.,24, in press,1997.
Moorbath,S., and H. Welke, Lead isotopestudieson igneous
rocks from the Isle of Skye, northwesternScotland,Earth
Planet Sci. Lett., 5, 217-230, 1969.
Morgan, W. J., Hotspottracksand the openingof the Atlantic
and IndianOceans,in TheSea, vol. 7, editedby C. Emiliani,
pp. 443-487, Wiley, New York, 1981.
Mougenot, D., M. Recq, P. Virlogeux, C. Lepvrier, Seaward
extensionof the East-Africanrift, Nature, 321, 599-603, 1986.
Mtiller, R. D., J.-Y. Royer,and L. A. Lawver,Revisedplate
motionsrelativeto the hotspotsfrom combinedAtlantic and
IndianOceanhotspottracks,Geology,21, 275-278,1993.
Nichols,G. J., andM. C. Daly, Sedimentationin an intracratonic
extensionalbasin:the Karoo of the centralMorondavaBasin,
Madagascar,Geol.Mag., 126, 339-354, 1989.
Obradovich,J. D., A Cretaceous time scale,in Evolutionof the
WesternInterior Basin, Spec.Pap. 39, editedby W. G. E.
Caldwell, pp. 379-396, GeologicalAssociationof Canada,
Waterloo, ON, 1993.
Paquette,J. L., A. N6d61ec,B. Moine, andM. Makotondrazafy,
U-Pb, singlezirconPb-evaporation,andSm-Ndisotopicstudy
of a granulitedomainin SE Madagascar,d. Geol., 102, 523-
538, 1994.
Peng, Z. X., and J. J. Mahoney, Drillhole lavas from the
northwesternDeccan Traps, and the evolution of R6union
hotspotmantle,Earth Planet Sci. Lett., 134, 169-185, 1995.
Potts,P. J., O. Williams-Thorpe,and J. S. Watson,Determination
of the rare-earth element abundances in 29 international rock
standards
by instrumentalneutronactivationanalysis:Critical
appraisalof calibrationerrors, Chem. Geol., 34, 331-352,
1981.
Price,R. C., A. K. Kennedy,M. Riggs-Sneeringer, and F. A.
Frey, Geochemistry of basaltsfrom the IndianOceantriple
junction: Implication for the generationand evolution of
Indian Oceanridge basalts,Earth Planet. Sci. Lett., 78, 279-
296, 1986.
Reid,A.M., andA. P. le Roex,Kaersutite-bearing
xenoliths
122 CRETACEOUS MADAGASCAR BASALTS
megacrystsin volcanic rocks from Funk Seamount in the
southwestIndian Ocean,Mineral. Mag., 52, 359-370, 1988.
Saunders,A.D., M. Storey, R. W. Kent, and M. J. Norry,
Consequences of plume-lithosphere interactions, in
Magmatism and the Causes of ContinentalBreak-up, Spec.
Publ. 68, edited by B.C. Storey, T. Alabaster,and R. J.
Pankhurst,pp. 41-60, The GeologicalSociety,London,1992.
Scrutton, R. A., W. B. Heptonstall, and J. H. Peacock,
Constraintson the motion of Madagascarwith respectto
Africa, Mar. Geol., 43, 1-20, 1981.
Shaw, D. M., Trace element fractionation during anatexis,
Geochim. Cosmochim.Acta., 34, 237-243, .1970.
Shimizu, N., and Kushiro, I., The partitioningof rare earth
elements between garnet and liquid at high pressures:
preliminary experiments,Geophys.Res. Lett., 2, 414-416,
1975.
Sinha, M. C., K. E. Louden, and B. Parsons,The crustalstructure
of the MadagascarRidge,Geophys.J. R. Astron.Soc.,66, 351-
377, 1981.
Smith, A. G., and A. Hallam, The fit of the southerncontinents,
Nature, 225, 139-144, 1970.
Storetvedt, K. M., J. G. Mitchell, M. C. Abranches, S. Maaloe,
and G. Robin, The coast-parallel dolerite dykes of east
Madagascar;age of intrusion, remagnetizationand tectonic
aspects,J. Afr. Earth Sci., 15, 237-249, 1992.
Storey,M., J. J. Mahoney, A.D. Saunders,R. A. Duncan,S. P.
Kelley, andM. F. Coffin, Timing of hot-spotrelatedvolcanism
and the breakupof Madagascarand India, Science,267, 852-
855, 1995.
Sun, S.-s. and W. F. McDonough, Chemical and isotopic
systematics of oceanic basalts: implications for mantle
composition and processes,in Magmatism in the Ocean
Basins, Spec. Publ. 42, edited by A.D. Saundersand M. J.
Norry, pp. 313-345, The GeologicalSociety,London, 1989.
Taylor, S. R., and S. M. McLennan, The Continental Crust: Its
Compositionand Evolution,312 pp., Blackwell,Oxford, 1985.
Thirlwall, M. F., B. G. J. Upton, and C. Jenkins,Interaction
betweencontinentallithosphereandthe Icelandplume; Sr-Nd-
Pb isotopegeochemistryof Tertiary basalts,NE Greenland,dr.
Petrol., 35, 839-879, 1994.
Todt, W., R. A. Cliff, A. Hanser, and A. W. Hofmann,
Evaluation
of a 2ø2Pb-2øsPb
doublespikefor high-precision
lead isotopeanalysis,in Earth Processes: Readingthe Isotopic
Code, Geophys.Monogr. Ser., vol. 95, editedby A. Basu and
S. R. Hart, pp. 429-437, AGU, Washington,D.C., 1996.
Tolan, T. L., and S. P. Reidel, Structuremap of a portion of the
Columbia river flood-basalt province, in Volcanism and
Tectonismin the ColumbiaRiver Flood-BasaltProvince,Spec.
Pap. 239, edited by S. P. Riedel and P. R. Hooper, enclosed
map, GeologicalSocietyof America,Boulder,Colo., 1989.
Walter, M. J., and D.C. Presnall,Melting behaviourof simplified
lherzolite in the systemCaO-MgO-A1203-SiO2-Na20from 7
to 35 kbar, dr.Petrol., 35, 329-359, 1994.
Watson, S., and D. P. McKenzie, Melt generationby plumes:a
studyof Hawaiian volcanism,d. Petrol., 32, 1991.
White, R. S., and D. P. McKenzie, Magmatismat rift zones:the
generationof volcaniccontinentalmarginsandflood basalts,d.
Geophys.Res., 94, 7685-7730, 1989.
Windley, B. F., A. Razafiniparany,T. Razakamanana,and D.
Ackermand, Tectonic framework of the Precambrian of
Madagascar and its Gondwana connections:a review and
reappraisal,Geol. Rundsch.,83, 642-659, 1994.
John J. Mahoney, School of Ocean and Earth Science and
Technology,University of Hawaii, 2525 Correa Road, Honolulu,
Hawaii, 96822, USA.
Andrew D. Saunders,Departmentof Geology,University of
Leicester, University Road, Leicester, LE1 7RH, United
Kingdom.
Michael Storey, Danish LithosphereCentre, Oster Voldgade
10, DK-1350 Copenhagen K, Denmark(e-mail:
 The Caribbean-ColombianCretaceousIgneousProvince'
The InternalAnatomy of an OceanicPlateau
ß 2
Andrew
C.Kerr• John
Tarney
• Giselle
F.Marrlner,
Alvaro
Nivia3,Andrew
D Saunders
4

The Late CretaceousCaribbean-Colombianigneousprovince is one of the

world's best-exposedexamples of a plume-derivedoceanic plateau. The
buoyancyof the plateau(resultingfrom residualheat and thick crust) kept it
from being totally subductedas it moved eastwardwith the Farallon Plate from
its site of generationin the easternPacific and encountered a destructiveplate
margin. In effect, the plateaumakesup much of the CaribbeanPlate; it is well
exposed around its margins, but more so in accreted terranes in western
Colombia (including the well-known Gorgona komatiites and Bolivar
mafic/ultramaficcumulates). Compositionally,the lavas of the plateau form
three groups:(a) basalts,picrites,and komatiiteswith light-rare-earth-element
(LREE)-depletedchondrite-normalised patterns;(b) basaltswith LREE-enriched
patterns;and (c) basaltswith essentiallyflat REE patterns(the most dominant
type) similar to many of the basaltsfrom the Ontong Java Plateau. These three
typesdemonstrate the heterogeneous natureof the mantleplume sourceregion.
The picritesandthe komatiitesseemto lie nearerthe baseof the plateauthanthe
more homogeneousbasalts;thus, the more MgO-rich melts may have been
eruptedbeforelargemagmachambershad a chanceto develop. A reconstructed
crustal crosssectionthroughthe plateau consistsof dunitic and pyroxenitic
cumulatesnear the basewhich are overlainby layeredolivine-richgabbrosand
more isotropic gabbros. The lowermost eruptive sequence comprises
compositionallyheterogeneous picrites/komatiitesoverlainby more homogene-
ous pillow basalts. Spectacularhornblende-plagioclase veins cut the Bolivar
assemblageandthesemay representlocal partial meltsof the p|ateau'sbaseas it
was thrusted onto the continent. Subduction-related batholiths and extrusive
rocksfoundaroundthe marginof the provinceare of two distinctages;one suite
representspre-p|ateau collision-related volcanism whereas the other suite,
slightlyyoungerthanthe plateau,may be associated with obduction.

•Department
of Geology,
University
of Leicester,
University 1. INTRODUCTION
Road, Leicester LE 1 7RH, UK
2Department
of Geology,
RoyalHolloway
University
ofLondon, Oceanic large igneousprovincessuch as the Ontong
Egham,Surrey,TW20 0EX, UK Java Plateau, the Nauru Basin, Manihiki Plateau, and the
3Ingeominas-
Regional
Pacifico,
AA9724,
Cali,Colombia Kerguelen Plateau, with 10-40 km thick crustal sections
4Department
of Geology,
University
of Leicester,
University
Road, Leicester LE 1 7RH, UK
[e.g., Coffin and Eldholm, 1994], are vast outpourings
of
basalticmagma(oftenwith eruptedvolumesof basalt>4 x
LargeIgneousProvinces:Continental,Oceanic,andPlanetary 106km3).Despitetheirobvious
volumetric
importance
we
Flood Volcanism
still know relatively little about their structure and
GeophysicalMonograph100
composition.Several of these plateausand basinshave
Copyright1997 by the AmericanGeophysicalUnion
124 THE CARIBBEAN-COLOMBIAN PLATEAU
been sampled by the Deep Sea Drilling Project/Ocean
Drilling Program (DSDP/ODP); however, the greatest
depthto which any of them hasbeen drilled is 700 m into
the Nauru Basinby DSDP legs61 and 89 [e.g., Saunders,
1985]. Becausetheseover-thickenedareasof oceaniccrust
tend to be more buoyantthan normal oceancrust(due to
both thermal and densitydifferences)[Burkeet al., 1978;
Nur and BenAvraham, 1982; Cloos,1993], they arenot so
easily subducted.These oceanicplateaushave, therefore,
the potentialto be obductedonto continentalmargins[Ben
Avraham et al., 1981], thus increasingboth the possibility
of their preservationand accessibilityfor detailed study
[e.g., Saunderset al., 1996].
One such area where it is believed that an oceanic
plateauhasbeenobductedontothe marginof a continentis
northwestern South A•nerica [Marriner and Millward,
1984; Millward et al., 1984; Aspdenet al., 1987; Kerr et
al., 1996a]. This review will show that the accreted
volcanicsuccessions in ColombiaandEcuador,alongwith
the thickened oceanic crust of the Caribbean Plate, and
other smaller obducted basaltic fragments around the
marginsof the Caribbean(Figure 1) are the remnantsof a
Late Cretaceous oceanic plateau. The processes of
accretion, imbrication, and tectonic uplift of the
Caribbean-Colombian oceanic plateau have locally
exposedits lower reaches,providing informationon the
internal structureand compositionof one of these vast
oceanicplateaus.
2. STRUCTURE AND TECTONICS OF
THE CARIBBEAN REGION
The preserved water-coveredarea of the Caribbean-
ColombianCretaceousIgneousProvince(CCCIP) is about
6 x 105km2;however,because
a significant
portionof the
plateau appears to have accreted onto the western
continentalmargin of Colombia and Ecuador(and some
may have subcreted),the oceanicplateaumay originally
have beenmore than twice this size [Burke, 1988].
The seismicrefractionstudiesinitiatedby Edgar et al.
[ 1971], which were integratedwith gravitydataby Caseet
al. [1990] and summarisedby Donnelly [1994], revealthe
anomalouslythickenednature of the oceaniccrust in the
submergedportion of the CaribbeanPlate. The crustof the
Caribbeanvariesfrom-8 to >20 km thick [seeFigure 3.5
of Donnelly, 1989], well in excessof the 6-7 km of oceanic
crust produced at a normal mid-ocean ridge. This
thickenedoceaniccrust was drilled in severalplacesby
DSDP Leg 15. From the resultsof this drilling, Donnelly
[1973] and Donnelly et al. [1973] proposedthat a large
flood basalteventhad occurredin the Caribbeanregionin
the Late Cretaceous. This 'flood basalt event' is now
recognisedas the remnant of a major oceanic plateau
[Burke et al., 1978; Duncan and Hargraves, 1984;
Donnelly et al., 1990; Hill, 1993; Kerr et al., 1996a],
which Duncan and Hargraves [1984] and Klaver [1987]
proposedhad formed as a consequence of upriseof a deep
mantleplume.
Throughoutthe history of geologicalresearchin the
Caribbeanregiontherehasbeenconsiderable controversy
as to whether the Caribbean Plate formed in situ or
whether it was transportedto its presentpositionfrom a
westerly (Pacific) direction. The former view is an older
one [e.g., Schubert,1935;MeyerhoffandMeyerhoff,1972]
and has receivedrelatively little attentionin recentyears,
and most modern authors accept the premise that the
componentsof the CaribbeanPlate have had a highly
mobile history [Burke et al., 1984; Pindell, 1990].
Nevertheless,within this model, opinionsdiffer as to the
timing and extent of movements [e.g., Burke, 1988;
Donnelly, 1989; Pindell and Barrett, 1990]. However,the
origin of the Caribbean Plate in the easternPacific as an
oceanicplateauappearsto be generallyaccepted.
Using a fixed hotspot reference frame, Duncan and
Hargraves [1984] and Hill [1993] suggestedthat the
CCCIP was producedby melting duringthe initial 'plume
head'phaseof the Galfipagoshotspot.Eastwardmovement
of the FarallonPlate in the Late Cretaceous-EarlyTertiary
broughtthe northernpart of the plateauinto the continental
gap which had opened up between North and South
America sincethe Jurassic.The eastwardmoving plateau
appearsto have beentoo buoyant(due to remnantheat and
crustalthickness)to be subducted[Burkeet al., 1978; Hill,
1993]. This 'clogging' of the subductionzone led to a
'flip' in the directionof subductionfrom eastto west, and
the Atlantic Plate beganto be consumedby subductionas
opposed to the Farallon/CaribbeanPlate [Lebr6n and
Perfit, 1994]. The northernpart of the CaribbeanPlate is
boundedby a seriesof approximatelyeast-westtrending
strike-slip faults, whereas its southernmargin is a broad
complexzone of convergenceand right lateral strike-slip
faulting [Pindell et al., 1988; Ladd et al., 1990]. This
tectonicactivity has exhumedand exposedportionsof the
oceanicplateaucrustat the marginsof the CaribbeanPlate.
In contrast, the southern part of the Caribbean-
Colombianplateaubeganto interactwith the northwestern
continental margin of South America. In this region,
dockingof the plateauwith the continentalso resultedin
the jamming of the subductionzone, but rather than
subductionflip this led to progressive,westwardback-
steppingof the subductionzone,leadingto the formation
of accretedoceanicplateauterranesalongthe
 KERR ET AL. 125

•...............
Cretaceous
lavas

Subduction
Zones '"

Spreading
........ Cuba .....................
Ridges

El Tambor Gp ..... • •. •D•[ss'e•61•m• '''

.......... 152
©:I•S'D•
kiie•
'.•4•........
'"""" .. "'..- .........
ß ..........
- :,, 151©
....
153
ß ' ' ' ß1 Curacao
Aruba
..... ' Costa ............ Bonaire

Villa de Cura

.................. Azuero ' . . • t,... Venezuela -"

Gorgona '

• Colombia • ß ß
............... • o-.: o
a"'' ß,
ß't... ß
' - .,•,

........... •alapagos
.........
........... Islands .........
"" '-'-' 400 km
I'"' I , I

Figure 1. Map showingmain basalticexposuresof the Caribbean-Colombian CretaceousIgneousProvince,after

Donnelly et al. [1990] and Iturralde-Vinent[1994]. Numbereddots:locationsof DSDP sitesfrom which basaltswere
recovered.Galfipagosmay representthe remnanthotspotmarkingthe site of earlier plateauformation,but this is not
certain.

marginof SouthAmerica[e.g.,Kerr et al., 1996a].
Recently, it has been reported that long-wavelength
magneticanomaliesover the Venezuelanand Colombian
basinsin the CaribbeandisplayNE-SW and E-W linear
patterns [Hall, 1995]. These anomalies have been
interpretedasbeingthe resultof an Early Cretaceous
phase
of seafloorspreadingat the Farallon-Pacific-Phoenix
triple
junction, at which the CCCIP may have formedat -90 Ma
[Hall, 1995].
The CCCIP is, for the most part, composedof pillow
lavas and massive flows or shallow sills of basalt/diabase
with occasionaloccurrencesof more picritic lavas.The age
of CCCIP has been the subject of some debate in the
literature.Pre-1990 knowledgeof the age of the province
was based on fossils found in associated sediments and on
K-Ar dates(seeFigures2 and 3 of Donnellyet al. [ 1990]).
However, it is likely that seafloor alterationand zeolite-to
greenschist-facies metamorphismhave resultedin Ar loss,
and it is therefore doubtful if any of the K-Ar age
determinationscan be trusted. Although Jurassicfossil
ageshave been cited for parts of the CCCIP (Costa Rica,
Venezuela and Puerto Rico), the tectonic relationship
betweenthe fossil-bearingsedimentsand the basaltsis by
no means clear [Donnelly et al., 1990; Donnelly, 1994].
Sediments overlying the submerged basalts in the
Caribbean Sea, drilled by DSDP Leg 15, place a
biostratigraphicage limit of 88-83 Ma on the cessationof
igneousactivity in the province.(All agesare basedon
126 THE CARIBBEAN-COLOMBIAN PLATEAU
time-scaleof Hadand et al. [ 1990]).
Recent4øAr-39Ar
datingof CCCIP basaltsfrom, Haiti,
Curacao, Isla Gorgona in Colombia, and the Nicoya
Peninsulain Costa Rica [Sintonand Duncan, 1992; Sinton
et al., 1993] has shown that the ages of all thesebasalts
group around 87-90 Ma. These results suggestthat the
bulk of volcanic activity associatedwith the CCCIP
occurredover a relatively short period. Nevertheless,as
will be shown below, there still is some evidence to
suggestthat in severalareasof the Caribbeanthe volcanic
activity may be older, and there is an obviousneed for
more high-precisionradiometricdatingin the province.
3. FIELD RELATIONS AND GEOCHEMISTRY
OF THE CCCIP
3.1. Curacao and Aruba
The islandsof Curacaoand Aruba, located70 km north
of the Venezuelan coast (Figure 1), contain some of the
best-preservedsectionsof the CCCIP. The Curacao lava
successionis more than 5 km thick [Klaver, 1987], with
abundantpillow lavas. The lower half of the exposed
successionis mostly composedof picrites and olivine
basalts.The picritescan containas muchas 31 wt% MgO
[Beets et al., 1982]. Kerr et al. [1996c] have shown that
these high-MgO picrites are olivine cumulatesand have
calculated,usingthe maximum Fo contentof the olivines,
that the primary magmasof the Curacaopicritescontained
16-18 wt% MgO. The upper exposedhalf of the Curacao
lava succession consistsof more-evolved,plagioclase-and
clinopyroxene-phyric pillow basalts,with doleritesillsand
occasionalreworkedhyaloclastites.
There is only one known intercalationof pelagic
sedimentswithin the Curacao lava succession[Klaver,
1987]. This observationimplies that the Curacao lavas
were eruptedover a relativelyshorttime span.Suchrapid
emplacementis consistentwith a mantleplume origin for
the Curacaolava succession. The pelagicsedimentswithin
the top half of the lava successionconsistof siliceous
shalesand limestones[Klaver, 1987]. Wieldmann[1978]
reportedthe occurrenceof fossil ammonitesof Mid-Albian
(100-105 Ma) agewithinthesepelagicsediments. This age
is significantly older than the 88-90 Ma found using
4øAr-39Armethods[C. Sinton, pers. comm. 1994].
However, referenceto kVieldmann[1978] revealsthat the
preservation state of the Curacao ammonites is "rather
poor, due to low grade metamorphismand deformation"
and "very few (ammonites)are complete,but even then
distorted.Generallyonly small fragmentsare preserved."
Given this information, it is conceivable that, due to
misidentification,the ammoniteagesare incorrectand the
ageof the Curacaolavasuccession
is, asthe4øAr-39Ar
data
suggest,88-90 Ma.
Kerr et al. [1996c] have shownthat the picrites of the
Curacao successionare related to the basaltsby simple
fractionalcrystallisationinvolvingolivine, clinopyroxene,
and plagioclase(Figure 2) and that for the most part the
Curacao lavas possessnearly constantNd isotopicratios
(Figure 3b), and incompatibletraceelementratios(Figure
3a) with chondritic rare-earth element (REE) patterns.
These features suggestthat either the inferred plume
sourceregion of the Curacaolavas was homogeneousor
that melts from a heterogeneous source became
homogenisedduringmeltingor in magmachambersduring
ascentto the surface.The elevated87Sr/86Sr
of theCuracao
basalts can be explained by contaminationof picritic
magmas with highly altered oceanic crust along with
concomitantfractionalcrystallisation[Kerr et al. 1996c].
The Aruba lava successionhas been less intensely
studiedthan the Curacao lavas. The basaltic sequenceis
about 3 km thick [Beets et a/.,1984] and is intruded and
metamorphosedby the 85-90 Ma [Priern et al., 1986]
tonalitic/gabbroicAruba batholith. The lava succession
comprises pillow basalts and dolerite sills, with
intercalations of volcaniclastic sediments (containing
Turonian [90.5-88.5 Ma] ammonites) and some well
sortedconglomerates [Beetset al., 1984]. The chemistryof
the Aruba lavas is similar to that of the Curacaobasalts,
except for the presence of more evolved ferrobasalts.
Plagioclaseand clinopyroxeneare the main fractionating
phases.The MgO contentof the lavasvariesfrom 10 wt%
for the basalts(which occurnearerthe baseof the exposed
section)to 4 wt% for the ferrobasalts(locatednearerthe
top of the lava pile [Beetset al., 1984]). As they have
chondriticREE patternsand similar incompatibleelement
trace element ratios (Figure 4a) to the Curacaolavas, this
suggeststhat they were derived from a similar, or the
same, mantle source region [Beets et al., 1984; Klaver,
1987].
Beetset al. [1984] initially concludedthat the Curacao
and Aruba lava successions were part of an island arc
succession.However, as Klaver [1987] and Kerr et al.
[1996c] have pointedout, the presenceof (non-boninitic)
high-MgO lavas,the absenceof a subductiontrace element
signature,and the chemical similarity to Pacific oceanic
plateau lavas (Figure 3a) provide strong evidencefor a
mantleplume origin for theselavas.
3.2. I/enezuela
Althoughthe coastalborderlandsof Venezuelapreserve
somebasalticand ultramaficrocks [Oxburgh, 1966; Beets
et al., 1984], which are possibly part of the
 KERR ET AL. 127

16"''
16--

øSo 12-

=
oo
O¸

I I I I 4
10 20 30 40 0 lO 20 3O 4O
MgO MgO
1800 28 ''

oo 24 -
1200
2O

16 -
6OO
12-
oo
0 ', 8
0 10 20 30 40 0 10 20 30 40

MgO MgO
Figure 2. Plots of A1203, CaO, Ni, and Y vs MgO for Curacaobasaltsand picrites.Also shownare calculated
fractionationtrends,with ticks at 4 wt% crystallisationintervals.After Kerr et al. [ 1996c].

[Donnelly et al., 1990], many of the rocks have been
extensively tectonised and metamorphosed to high-
pressure/low-temperature equivalents[Beetset al., 1984].
As a resultof this tectonismandmetamorphism,the age of
these basic rock associations from the Venezuelan coastal
regionhasnot beenfully resolved.As reviewedby Beetset
al. [ 1984], K-Ar agesso far obtainedrange from 107 to 65
Ma. It is generally accepted that these basaltic and
ultramafic volcanic rocks are an allochthonous terrane
obductedfrom the north in the Late Cretaceousor Early
Tertiary [Beetset al., 1984]. Palaeomagneticstudieshave
shown that the whole belt was rotated 90 ø clockwise
duringobduction[Stearnset al., 1982].
The Villa de Cura Groupis a 4- to 5-km-thick,250-km-
long belt of metamorphic volcanic and volcaniclastic
rocks, 50 km south of the Venezuelan coast. This
approximatelyE-W trending unit is composedof both
mafic and more siliceousrocks. Donnelly et al. [1990]
subdividedthesevolcanicrocksinto two chemicalgroups:
a more evolved series of subduction-related lavas and
tuffs, and a series of basalts and ultramafic rocks with
transitional(T-)MORB characteristics. However, extensive
tectonisationmeans that the interrelationshipsbetween
thesetwo groupsare unclear.The mafic, T-MORB lavas
are found chiefly in two formationswithin the Villa de
Cura Group:the Tiara andthe E1Carmenformations.
The basalts of the Tiara Formation are underlain by
sedimentscontainingAlbian (97-112 Ma) radiolari,a[Beck
et al., 1984; Donnellyet al., 1990], thus placinga
maximum age of 112 Ma on the overlying basalts.The
lavas of the Tiara and the E1 Carmen Formationsmostly
possess relatively flat chondrite-normalised[Sun and
McDonough, 1989] REE patterns(with LaN/SmN< 1 and
TbN/YbN> 1, althoughone of the Tiara lavashasLaN/YbN=
5 [Beetset al., 1984]). Some of the lavas from the 1200-m-
thick E1 Carmen Formation contain up to 12 wt% MgO
and have been classifiedas 'picrites' by Beetset al. [ 1984],
though they are more correctly classified as high-MgO
picritic basalts. Donnelly [1989] suggestedthat the E1
Carmen formation may be correlated with the Curat;ao
picrites. Furthermore,Donnelly et al. [1990] noted that
basaltsof the Tiara and E1 Carmen formationshave a very
similar chemistryto the picritesandbasaltsof Curat;aoand
Aruba.
Several small ultramafic units also crop out in close
proximity to the Villa de Cura Group, and it is highly
likely that they are genetically related to the spatially
associatedbasalts[Donnelly et al., 1990]. One of the most
noteworthyof theseexposuresis the Tinaquillo peridotite
[e.g., MacKenzie, 1960]. A small exposure of
128 THE CARIBBEAN-COLOMBIAN PLATEAU

1985]. A K-Ar whole rock age determinationon a Sans

ß
=c
•
100
1a, -I•ax
Plateau
averages
ß Kerguelen
ß Ontong
• Nauru
Java
Basin
Souci basalt yielded an age of 87+4 Ma [Wadge and
MacDonald, 1985]. Althoughthis placesthe formation
within the age rangeof otherCaribbeanoceanicbasalts,
the altered nature of the SansSouci basaltsrequiresthat
•, 10 ='•-•\"•
=
ß Manihiki the age be treatedwith caution.Wadgeand MacDonald
[1985] proposed that the Sans Souci Formation is

ß Aruba Lava Formation ,

1 I I I I I I I I I 2.0- o Cura9aoLavaFormation
Th K Nb La Ce Sr Nd P Zr Ti Y a Bonairo arc lavas
12

3,I"•org
ona
komatiites

M•••llan
R
1.0'
lO- Gorgona
D-basalts
Gorgona
E-basalts

OntongJava
Plateau
0.4
Nauru Basin

ß Cura•;aopicrites
-- 13 Cura•;aobasalts
;;•DHaiti basalts 0.2
o - I I I
o.7o25 0.7030 0.7035 0.7040 0.7045

(87Sr/86Sr)i
0.1 !

Figure 3. (a). Comparison of trace element compositionsof

0.6 1.0 2.0 3.0 5.0 7.0
Curaqaobasaltsand picriteswith averagecompositionsof other
ocean plateaus: Kerguelen, Ontong Java, Nauru Basin, and Hf
Manihiki Plateau (normalisedto primitive mantle valuesof Sun __
and McDonough[1989]), (b) Plot of (gNd)i vs (87Sr/86Sr)ifor b= n Jamaica
lavas from the CCCIP along with lavas from CretaceousPacific
oceanicplateaus.Data sourcesreferencedby Kerr et al. [1996c]. ß Cura(;:ao
Leg 15 Site 151
* DSDPLeg15
gabbro,dolerite,andperidotiteon the Paraguanfi
peninsula
has been describedby Martin-Bellizia and de Arozena
[1972].

3.3. Trinidad
I I I
The Sans Souci formation in northeastern Trinidad is 15 20 25
representedby a 1000-m-thick sequenceof basaltic
Zr/Y
pyroelasticrocks, basalts,gabbrosand minor terrigenous
sedimentary rocks [Wadge and MacDonald, 1985]. Figure 4. (a) Plot of Th vs Hf comparingtholeiiticplateaulavas
Compositionally, theselavasresemblenormal(N-) MORB from Aruba and Cura{;aowith Cretaceousarc lavasfrom Bonaire;
tholeiites,and (with one exception)the analysedbasalts afterBeetset al. [1984]. (b) Plot of Ti/P vs Zr/Y for basaltsfrom
have low levels (<10 x primitive mantle values) of Jamaica[Wadgeet al. 1982], Cura{;ao[Kerr et al., 1996b] and
incompatibletrace elements [Wadge and MacDonald, CaribbeanDSDP Leg 15 (G. F. Marriner,unpublished

separated from therestof Trinidadby a seriesof strike-slip
faults, and they further suggestedthat the basaltsare part
of the CCCIP that was juxtaposedalong the northern
marginof SouthAmericain LateCretaceous times.
3.4. Jamaica
The 40-km2 Bath-Dunrobin Formation, eastern
Jamaica,consistsof an inlier of highly deformedbasalts,
dolerites,and isotropicgabbros[Wadgeet al., 1982].Thin
sedimentaryintercalationsin this-2.5-km-thick ophiolitic
assemblage containmostlyCampanian(83-74 Ma) fossils,
althougholder,possiblyreworked,fossilshavebeenfound
within the succession
[Wadgeet al., 1982]. The chemistry
of the Bath-Dunrobinbasaltsas reportedby Wadgeet al.
[1982] is MORB-like and very similar to the Aruba and
Cura•;aobasalts,as well as to some of the basaltsfrom
DSDP Leg 15 (Figure 4b). Despite these similarities
Wadgeet al. [1982] suggested oceaniccrustcomprising
the Bath-Dunrobinophiolitehad been obductedfrom the
north after formation in the Yucatan Basin. However, as
Donnelly [1994] pointedout, there is no strongevidence
that the Bath-Dunrobin Formation was not derived from
the south,and thusit is probablypart of the CCCIP.
3.5. DSDP Leg 15 Drill Sites
Drilling during DSDP Leg 15 encounteredbasaltsor
doleritesat five sites (Figure 1) in the central Caribbean
[Donnelly et al., 1973]. As noted above, the
biostratigraphicage of the basaltsand doleritesat sites
146, 150, 151, and 153 is late Turonian (88 Ma); however,
at the most westerly site (152), an early Campanian(83
Ma) age has been assigned[Donnellyet al., 1973]. This
resultsuggeststhat, in the west of the province,volcanism
continuedfor up to 5 m.y. after activityhad ceasedin the
east. The basalts are all plagioclase-phyrictholelites
[Benceet al., 1975]. Donnelly et al. [1973] and Benceet
al. [1975] reportedmajor and rare earth elementdata for
the Leg 15 basalts,and we have recentlyanalyseda suite
of samplesfrom the five DSDP Leg 15 sitesfor a rangeof
major and trace elements.The new chemical data are
presentedin Figure 5.
Both Donnelly et al. [1973] and Bence et al. [1975]
subdividedthe lavas and sills into two chemicalgroups:
thosewith slightly depletedlight REE patterns(Las/Sm$
<1) and TiO2 < 1.8 wt% (Holes 146, 150, 152, and 153);
and thosewith Las/SmN >2 and TiO2 >2.0 wt% (Hole 151)
(Figure 5b). Thesetwo groupscan alsobe observedin the
new data set (Figure 5). The new trace elementdata also
reveal that the basaltsfrom Site 151 possesshigherNb/Zr
KERR ET AL. 129
and Ti/Y ratiosthan the rest of Leg 15 basalts(Figure 5e),
consistentwith derivation from a mantle source region
with garnetstablein the residue.
Bence et al. [1975] suggestedthat the basalts from
Holes 146, 150, 152, and 153 could be related to each
other by fractional crystallisation of olivine and
plagioclase,reflectedin decreasingCa and Ni with lower
MgO/[MgO + Fe203(t)]. However, lower incompatible
element contents and lower Nb/Zr ratios reveal that the
basalts from Site 152 cannot be related to those from sites
146, 150, and 153 by fractionalcrystallisationalone. The
Site 152 basaltsmay thereforebe the resultof eithermore
extensive melting of the same source region which
producedthe 146, 150 and 153 basaltsor melting of a
separatemantle sourcethat was slightly more depletedin
incompatibleelements.
Bence et al. [1975] proposedthat the basalts from
DSDP Leg 15 were derived from two different parental
liquids. However, the new trace elementdata (Figure 5e)
suggestthat three differentparentalliquidsmay have been
involved. As will be shown below, such variation has also
beenfoundat othersitesthroughoutthe province.
3.6. Hispaniola
Hispaniola contains two main belts of Cretaceous
basaltic rocks: the Dumisseau Formation in the southwest
and the more central Duarte and Siete Cabezas Formations.
Smaller, poorly studied,ophiolitic bodiesare found along
the north coastof the island [Donnelly et al., 1990]. The
Dumisseau Formation in southwestern Haiti consists of a
1500-m-thick sequence of interbedded pillowed and
massive basalts, dolerites, pelagic limestones,turbidites
and siliceous siltstones. Gabbroic intrusions and dolerite
dykes occasionallycut the sequence.Maurrasse et al.
[1979] reported palaeontological evidence that the
sedimentsassociated with the upperbasaltsof the sequence
are early Campanian(74 Ma) to late Santonian(83 Ma) in
age, and fossils from the lower stratigraphiclevels are
mostly of Coniacianand Turonian(86-90 Ma) age. These
latteragesaremorein accordwithrecent4øAr-39Ar
dating
of basaltsof the DumisseauFormation,which spanthe age
range 87-90 Ma [Sintonand Duncan, 1992].
Sen et al. [ 1988], in a comprehensive
assessmentof the
chemistryof the Dumisseaubasalts,showedthat the major
elementcompositionsof the lavasare very similarto those
of many of the basaltsdrilled duringDSDP Leg 15. Most
of the samplesare basaltic(<10 wt% MgO); however, a
picrite (22.7 wt% MgO) was alsodiscoverednear the base
of the succession.Becausethe olivines in this picrite are
not highly magnesian(Fo87), it is likely the rock is
130 THE CARIBBEAN-COLOMBIAN PLATEAU

25-- 3.3 •
a. b.
20--- 2.8

15-- • 2.3

10-- I- 1.8-

5 - 1.3-
[] [] []
[]
I•rrß
rß•.n
0 0.8 i i I
0.3 0.35 0.40 0.45 0.30 0.35 0.40 0.45

MgO/Fe203(t)+MgO MgO/Fe203(t)+MgO
14-
C. [] [] d.
120-
12-

10-
._ 80-
Z

40-
.

0 I I I I I
0.30 0.35 0.40 0.45 0.30 0.35 0.40 0.45

MgO/Fe203(t)+MgO MgO/Fe203(t)+MgO

600-
e. Colombian
• X ß

[]
Site 146

Site 150
500-- basalts
* Site 151
iti
O Site 152

400
-- ß A Site 153

3øøT •A••e•e
•j•,, Gorgona ß •- --•',-, enriched
ß Cura9ao

200-•-J
ø basalts _ } Gorgona
komatiites,
"'• N-MORB

I/' -•-
/ - picrites& depleted E-MORB
100 It....._.....•.•
I I basaltsI I X OIB
0 0.05 0.10 0.15 0.20

Nb/Zr

Figure 5. (a-d) Plotsof Nb, TiO2,Ni, andCaO vs MgO/MgO + Fe203(t)for DSDP Leg 15 basalts(t = total iron), (e)
Ti/Y vs Nb/Zr showingdatapointsfor Leg 15 basalts(G. F. Marriner,unpubl.data),Curacaolavas [Kerr et al.,
1996c],alongwith datafieldsfor lavasfromGorgonaIsland[Kerr et al., 1996b],westernColombia[Nivia, 1987]and
the DumisseauFormation,Haiti [Senet al., 1988].Plottedcompositionsfor averagenormal(N-MORB) and enriched
(E-MORB) mid-oceanridgebasaltandfor averageoceanislandbasalt(OIB) afterSunandMcDonough

olivine cumulate.Unlike the volcanicarc successions
(the
Los RanchosFormation)in northernHaiti [Donnellyet al.
1990], the trace element characteristicsof the Dumisseau
basaltsindicatethat they are of ocean-flooraffinity.
Most of the basalts are similar to those found at Site
151 in that they possess>2.0 wt% TiO2 and LaN/SmN >1
(Figure 6); however, Sen et al. [1988] discoveredseveral
basaltsat the base of the successionthat have TiO2 < 1.8
wt% and LaN/SmN< 1, valuesthat are similar to those in
basaltsfrom otherLeg 15 sites.This evidencesupportsthe
contention that the Dumisseau basalts are an obducted
section of the CCCIP [Sen et al., 1988]. Considerationof
radiogenic isotope and incompatibletrace element ratios
supportsa heterogeneous sourceregion for the Dumisseau
basalts containing at least two components[Sen et al.,
1988], one with enriched and one with depleted
incompatible trace element contents (relative to Bulk
Earth). Sen et al. [ 1988] alsonotedthat the Sr, Nd, and Pb
isotopic compositionsof the Dumisseaubasalts overlap
with thoseof the Gal/tpagoshotspot(Figure 3b) and they
usedthis evidenceto supporta Pacificplume-relatedorigin
for the Caribbean Plate.
The Duarte Formation in central Hispaniola is
composed predominantly of schists and amphibolites
[Donnelly et al., 1990]. Donnelly and Rogers [1978]
suggestedthat the unmetamorphosed basalts and cherts
(containingSantonian(83-86.5 Ma) radiolarians;Donnelly
et al. [ 1990]) of the SieteCabezasFormationrepresentan
oceanicbasalt sequence.The presenceof both high- and
low-TiO2 basalts in the Siete Cabezas and Duarte
Formations[Donnelly and Rogers, 1980; Donnelly et al.,
1990] that are similar to basaltsfrom both DSDP Leg 15
and southernHaiti, led Donnelly et al. [1990] to propose
that these rocks are a 'metamorphosedand tectonised
allochthon of Caribbean Cretaceous basaltic rocks.'
3.7. Central America
Several igneous assemblagesare exposed along the
Pacific coastof CostaRica on the Nicoya and SantaElena
peninsulas(Figure 1) and are collectivelyknown as the
Nicoya Complex. The Nicoya Complex has been studied
intensively by numerous authors (see Donnelly et al.
[1990] for a review). It consistsof a lower section
composedmostly of massivebasalts,small gabbroicand
plagiograniteplutons,with some pillow basaltsand very
little sediment,and an uppersectionconsistingof massive
and pillowed basalt and picritic basalts, intrusions,and
minor sediments [Donnelly et al., 1990]. The whole
complexis stronglytectonisedand Bourgoiset al. [1984]
suggested that it was emplacedin a seriesof nappes.
KERR ET AL. 131
lOO
Site
151
basalt
50
• 20
lO - Site 146; 150; 152; 153
Caribbean basalts
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
Figure 6. Chondrite-normalised rare earth elementpatternsfor
Dumisseau lavas (solid lines) [after Sen et al., 1988] in
comparisonwith basaltsfrom DSDP Leg 15, Sites 146, 150, 152,
and 153 (shadedarea) and basaltfrom DSDP Site 151 (dashed
line). DSDP datafrom G. F. Marriner(unpublished).
Most previous workers [e.g., Bourgois et al., 1984;
Wildberg,1984] have subscribed to the view that the lower
part of the Nicoya Complexis stratigraphically
overlainby
a sequenceof Jurassiccherts. However, this view has
recently been challengedby Donnelly [1994], who has
pointed out that the lower part of the Nicoya Complex
consistsmostlyof intrusiverocksandnowhereliesdirectly
below a depositional contact with the chert, thus
suggestingthat the Nicoya complex may intrude the
Jurassic
cherts.Recent4øAr-39Ar
dateson severalNicoya
basalts[Sinton et al., 1993] have yielded early Coniacian
(86.5-88.5 Ma) ages;thesenew age data supportthe view
of Donnelly [1994] that the Nicoya Complex is
significantlyyoungerthan Jurassic.
Wildberg [1984] and Berrangd and Thorpe [1988]
presentedcomprehensive geochemicaldata for the Nicoya
and SantaElena complexes.The basaltsand picritic basalts
were classified as low-potassium quartz and olivine
tholeiites(5-8 wt% MgO), havingmostlyflat REE patterns
and low levelsof incompatibletraceelements(Figure 7).
In many respectstheir chemistryis similar to that of the
basalts from Cura{;aoand from DSDP Leg 15 Sites.
Wildberg[1984] suggested that the Nicoya Complexwas
composedmostly of non-subduction-related oceaniccrust,
and he linked someof the volcanismin the upperNicoya
Complex to that of the Caribbean oceanic plateau.
Wildberg [1984] proposedthat the remaining
132 THE CARIBBEAN-COLOMBIAN PLATEAU
rocks in the upper Nicoya Complex were generatedin a
primitive island arc, althoughthis interpretationhas been
questioned[Donnellyet al., 1990;Donnelly, 1994].
Berrangdand Thorpe[ 1988] suggested thatthe Nicoya
complex was an obductedsegmentof oceaniccrustthat
had formed in a back-arcbasin. However, a multi-element
(primitive mantle normalised)diagram,using Berrang6
and Thorpe's data (Figure 7), shows that the Nicoya
basaltsdo not have subduction-related characteristics
(i.e.,
they have primitive mantlenormalisedratiosKN/NbNand
LaN/NbN << 1) and are very similar to oceanicplateau
basaltsfrom Curacao.Thus, from a considerationof basalt
compositions
andnew4øAr-39Ar
ages,it wouldappearthat
the Nicoya Complex representsanotherobductedportion
of the CCCIP.
Metamorphosedmafic rocks (the E1 Tambor Group)
crop out in several places in easternGuatemala[e.g.,
Donnelly et al., 1990] and thesehave also been sampled
duringdrilling off the westerncoastof Guatemala(DSDP
legs67 and 84). Biostratigraphical
agesfor the El Tambor
Group range from lowermost Cretaceousto Cenomanian
[Donnelly et al., 1990]. Although Donnelly et al. [1990]
suggested that the metamorphosed mafic rocks of
Guatemala were part of the CCCIP, their older ages
comparedwith most of the provinceallow the possibility
they could representoceaniccrust in existencebefore the
CCCIP.
Wildberg [1984] noted the presenceof basaltsvery
similarto thoseof the Nicoya Complexin the Azueroand
Sona peninsulas in western Panama. However, little is
known of the chemistryof thesebasalts.Much of eastern
Panamaeastof the canal(the Darien region)is composed
of Pre-Tertiary basaltic rocks [Escalante, 1990]. Cherts
containingCampanianradiolariaoverlie the basaltsaround
the Golfo de San Miguel [Bandy and Casey, 1973], thus
placinga youngerage limit on the complex.Goosenset al.
[1977] showedthat basaltsfrom the Golfo de SanMiguel
have chemicalsignaturessimilarto oceanbasalts;they are
probablypart of the CCCIP.
3.8. Cuba
The geology of Cuba is complex and appears to
representa seriesof accretedterranesof both continental
and oceanic material, most of which are younger than
Jurassic [Iturralde-Vinent, 1994]. Although extensive
outcropsof Upper Cretaceousvolcanicand intrusiverocks
can be found in Cuba, many of these include andesites,
dacites, rhyolites, tuffs, and tholeiitic to calc-alkaline
granitoids,all with arc-like chemistry[Iturralde-Vinent,
1994]. Underlying these arc lavas is an ophiolitic
200
IO0
•x R202
x
I I I I I I I I I I
Rb K Nb La Ce Sr Nd P Zr Ti Y
Figure 7. Multi-elementdiagram,showingdata for lavasfrom
the Nicoya Complex,CostaRica. SampleR202 is significantly
enrichedin incompatibleelementscomparedwith otherbasalts
fromthe Complex(datafromBerrangdand Thorpe[1988]).
sequence,the basalts of which may be as young as
Turonian. Iturralde-Vinent [1994] has interpretedthis
ophioliteas having formed in a back-arcbasin;however,
sparseand incompletechemicalanalysesof thesebasalts
mean that this theory cannot be rigorously tested.
Alternatively,the ophiolitecouldrepresentanotherpart of
the CCCIP.
3.9. Colombia and Ecuador
Although it is generally acceptedthat many of the
outcropsof Cretaceousbasaltin the Caribbeanand Central
America are part of one large igneousprovince,it is not
widely realisedthat thisprovincealsoextendsinto western
Colombia and Ecuador [Marriner and Millward, 1984;
Millward et al., 1984; Kerr et al., 1996a]. The Romeral
fault (Figure 8) marksa majorterraneboundaryin western
Colombia. To the east of the fault, Bouguer gravity
anomaliesare stronglynegative(-220 mgal) [Caseet al.,
1973], confirming geological observations that the
basementis composedof continentalcrust.In contrast,to
the west of the Romeral fault, Bouguer anomaliesare
stronglypositive (+135 mgal) [Case et al., 1971]. Thus,
the westernColombian-Ecuadorian Andes are composed
of high-densitymaterialand, as will be shownbelow,this
materialis undoubtedlyoceanicin origin.
The mafic rocksof westernColombiacropout in three
accretedbelts which trend approximatelyN-S; they
 KERR ET AL. 133

Key Post-Cretaceous chemistryand also occur as mixed magmasand breccias

[cf. McGarvie, 1984].
:•'•
., Pacific
coast
basalts
The volcanic successionon the island of Gorgona
Western Cordillera (Figure 8) appearsto be a continuationof the basaltic
basalts, gabbros, sequencesof Pacific Cordillera [e.g., McGeary and Ben
• ultramafics
(dark
shading)and Avraham, 1986] and as suchhasbeenproposedto be part
sediments
of the CCCIP [Donnelly et al., 1990; Storey et al., 1991;
Central Cordillera
Kerr et al., 1996b]. Gorgona has stimulatedinterest in
•"-'• sediments
basalts and ,Bolivar-Rio
recentyearsas the locationof the only known Phanerozoic
,Frio
spinilex-texturedkomatiites[e.g., Aitken and Echeverrœa,
Granites 1984; Kerr et al., 1996b]. Picrites, picritic tuff breccias,
I
basalts,and gabbrosare also presenton this small island
1 O0 km
(2.5 x 8 km). The geochemicalsignatures of thesevolcanic
South
America rocks imply that their mantle sourceregion was markedly
Gorgona heterogeneous, containingcomponentsboth depletedand
enrichedrelative to chondrites(Figures 9a and 9b). The
occurrence of picrites and komatiites with calculated
parental magma MgO contentsof 18-21 wt% requires
Precambrian &
Palaeozoic temperaturessignificantlyaboveambientmantle and more
Pacific Ocean metamorphic consistentwith decompressionin a mantle plume [e.g.,
basement
........

Storeyet al., 1991; Kerr et al., 1996b].

There has been considerable discussion in the literature

Figure 8. Simplified map of northwesternSouthAmerica (after
Nivia [1987]), showing the three separateaccreted terranes
containingCretaceousbasalts.
from east to west, the Central, Western, and Pacific
(Serrania de Baud6) Cordilleras (Figure 8). The three
cordilleras are composed of fault-bounded slices of
pillowed and massive basalt, and dolerite sills [see
Goosenset al., 1977; Martinet and Millwar& 1984; Nivia,
1987]. These major thrust/faultslicesare occasionallyup
to 15 km in width, althoughmore often they are <10 km
wide. Several occurrencesof picriteshave been reported
amongthe volcanic assemblages of the Central Cordillera
[Spadea et al., 1989]. These picrites are geochemically
similarto the ultramaficlavasfrom GorgonaIsland[Aitken
and Echeverrœa,1984; Kerr et al., 1996b].
Fault-bounded lenses of metasediments (slates,
siltstones,cherts,and sandstoneturbidites),dippingsteeply
eastwards, sometimes separate the basaltic slices,
particularly in the WesternCordillera where the volcanic
sequenceis thickest. Some gabbroicplugs and batholiths
have intruded the basalts,along with several ultramafic
complexes; however, very few dikes cut the basaltic
sequences.We note one exception,around the region of
Vijes, 30 km north of Call in southernColombia, where
felsite dykescut the basalticsequence.Thesefelsitesalso
are found in breccialayersand are mixed with basalt.Kerr
et al. [1996a] have suggestedthat these felsitesmay be
analogousto Icelandic felsites, which display a similar
as to the origin of the Cretaceousbasaltsof the Central,
Western, and Pacific Cordilleras.Most authorsagree that
the basalt successions are allochthonous and have been
accretedontothe marginof northwesternSouthAmericain
the Late Cretaceousor Early Tertiary [e.g., McCourt et al.,
1984; Millward et al., 1984; Bourgois et al., 1987].
However, whereas Barrero [1979] and McCourt et al.
[ 1984] advocatedan island-arc-related origin, Millward et
al. [1984], Nivia [1987], and Kerr et al. [1996a] have
proposed that the Cretaceousbasic igneous rocks of
Colombiaare the remnantsof an oceanicplateauand have
demonstratedthat the geochemistryof the Cretaceous
Colombianbasaltsis inconsistentwith an arc-relatedorigin
for the rocksof the province.Figure 10a [from Kerr et al.,
1996a] showsthat the Cretaceousbasaltsof Colombia are
markedlydifferentboth from Recentarc-derivedvolcanic
rocks from Colombia and from Cretaceous arc volcanic
rocks from Bonaire, Netherlands Antilles. Indeed, the
geochemistryof the basaltsof the Central and Western
Cordilleras(Figure 10b) is very similar to that of Pacific
oceanicplateausandof thebasaltsof Cura{;ao, Netherlands
Antilles (Figure 3a).
The relative agesof basaltsfrom the three cordilleras
are also a matter of debate, becausethe altered nature and
low potassiumcontentsof the lavasmeanthat K-Ar ages,
althoughabundantin the literature(compiled by Nivia
[1987]), spanquite a wide rangeand are of questionable
reliability. Two proposalshave been made regardingthe
relative agesand the timing of the accretionof eachof
134 THE CARIBBEAN-COLOMBIAN PLATEAU

1000 , mineralagesfor the Bugabatholithwhichrangefrom 69-

ß Enriched-basalt am 113 Ma. As a result,they concluded,somewhatarbitrarily,
ß Depleted-basalt that the Bugabatholithwas 99 Ma and thereforethat the
[] Komatiite basalts of the Amaime formation had to be older than this.

/x Picrite However, becausethe Amaime formationsensustrictois

only in faulted contactwith the Buga batholith,the
evidence that the Amaime basalts are older than 100 Ma is
not unequivocal. The ageof CentralCordillerabasaltshas
still to be resolvedsatisfactorily.It is, however,possible
thatthesebasaltsmaybe olderthanTuronianandsomight
IO0 ........ I ........ I ........ be part of an earlier oceanicplateauwhich was also
O.O01 0.01 0.1 1 obductedontothe Colombiancontinentalmargin.
Nb/Zr
0.7
lOO al Cretaceous Colombian lavas

I Gorgona
REE
patternsb. Recent
Colombian
volcanic
rocks
Bonaire arc volcanic rocks
(Washikemba
Formation)

lO 0.3--
,.
L•!i•a __

o,T,,, ,
B
I
I I I I I I I I I

o 1.0 1.5

La/Y
LaCePrNd SmEuGdTbDy•o •r T; •b Lu
lOO
bl
Figure 9. Summaryof critical featuresof Gorgonabasaltsand Pacific plateaus
komatiites, emphasisingrange of 'enriched' and 'depleted'
coinpositions.(a) Ti/Y vs Nb/Zr. (b) Chondrite-normalised
rare m
earth patterns(normalisingvalues after Sun and McDonough •: lO
[1989]). The shadedarea in (b) encompasses mostof the more- o
enrichedGorgonabasalts.After Kerr et al. [1996b]. i

o
three volcanicbelts:(a) the three belts representa single o 1
magmaticprovincewhichdevelopedin the lateCretaceous
and has subsequentlyaccretedonto the northwestern
marginof SouthAmerica[Goosens et al., 1977;Spadeaet
al., 1989], and (b) the basaltsof eachcordillerarepresent 0.1
three different volcanic provinces,which successively Th K Nb La Ce Sr Nd P Zr Ti Y
accretedon to the continentalmargin [Aspdenet al., 1987;
McCourt et al., 1984]. Figure 10. (a) Plot of Nb/Y vs La/Y showing Cretaceous
The basalts of the Central Cordillera (the Amaime Amaime and Volcanic Series basalts from Colombia [Nivia,
1987], Cretaceousarc-relatedbasaltsfrom Bonaire and Recent
Formation)have not beendatedradiometrically,
nor have
volcanics from Colombia [Martinet and Millwar& 1984]. (b)
any fossilsbeenfoundin the intercalated
sediments.The Comparisonof multi-elementpatternsof Colombianbasalts
Bugatonaliticbatholithintrudessomeamphibolites
which (dashedfield) with basaltsfrom Pacific oceanicplateaus(solid
have been assumedto be part of the same basaltic lines); data from Nivia [1987] and G. F. Marriner and A.D.
sequenceas the Amaime Formation[McCourt et al., Saunders(unpublisheddata). N-MORB normalisationvalues
1984]. McCourt et al. [1984] reportedK-Ar and Rb-Sr from Sun and McDonough[
 KERR ET AL. 135

The age of the WesternCordillera is constrainedbetter Thin

sediments
by Turonianto Coniacian(86-90 Ma) age ammonitesand
microfossils found in sediments intercalated with the
Homogeneous
basalts [Barrero, 1979; Bourgois et al., 1985]. The pillow lavas,
biostratigraphicalevidence from intercalatedsediments intrusive
suggeststhat the basalts of the Pacific Cordillera are sheets, gabbro
stocks
Coniacian to Santonian (83-88 Ma) in age [Gansser,
1973].Recent4øAr-39Ar
datingof basalts
andgabbros
from Heterogeneous
GorgonaIsland,the proposedoff-shorecontinuationof the komatiites,
PacificCordillera,hasyieldedagesof 86-88 Ma [Sintonet 10 picritesand
basalts
al., 1993]. It appearsthen that the mafic rocks of the
Western Cordillera and of Serrania de Baud6 are the same
age (Turonian-Coniacian) as most of the basalts in the Isotropic
Caribbeanregion. This, in conjunctionwith evidenceof gabbros,
gabbronorites
their similar chemistry (Figure 10b) and tectonic
considerations,suggeststhat these Colombianbasaltsare Layered
obductedsegmentsof the Caribbeanoceanicplateau. gabbros
The Bolivar-Rio Frio Ultramafic Complex in west Moho
Dunite with
centralColombia (Figure 8) is a 30-km-long, 5- to 10-km- 2O
bands of
wide body that trends NNE-SSW. This mafic-ultramafic Iherzolite,
complex forms part of the easternmostportion of the pyroxenites&
Western Cordillera, with fault slicesof basalt and sediment olivine
gabbronorites
to the west. To the south, the complex is composed
primarily of rhythmicallylayered,and in placestectonised,
norites,olivine norites,and gabbronorites, which underlie
a horizon of isotropic gabbronorites[Barrero, 1979; Figure 11. Hypothetical cross-sectionthrough the Caribbean
Bourgoiset al., 1982; Nivia, 1996]. The gabbrosof the oceanicplateau,basedon field and petrologicalstudiesof the
Bolivar Complex have incompatible trace element BolivarComplexandotheraccretedsegments of the CCCIP.
compositions that are similarto thoseof the basaltsof the
Western Cordillera [Nivia, 1996], suggestingthat the 11). This cross section is consistent with a recent
Bolivar Complex could be genetically related to the geophysicalmodel for Ontong Java [Farnetani et al.,
basalticplateausequences. 1996] andis similarto modelsof Archeanoceanicplateau
Toward its northernend, the complexis composedof structure[cf. Kent et al. 1996].
thick layers(-10 m) of serpentinised duniteseparatedby Severalpoints may be made in relationto Figure 11.
thinner (up to 1 m) bands of lherzolites, olivine Firstly, the dominance of horizontal basalt flows and
websterites,and olivine gabbronorites[Nivia, 1996]. This intrusivesheetsand sills,as opposedto the sheeteddykes
rock associationwas originally interpretedby Barrero of a "typical" ophiolite,suggests that the magmasupply
[ 1979] as a concentricallyzonedbody.Alternatively,Nivia greatlyexceededthe extensionnecessaryto accommodate
et al. [1992] presentedevidencethat the complex is it. Secondly,the layeredgabbroswereprobablyformedby
composedof faulted imbricatedblocksof oceanicLayer 3, the accumulationof crystals at the base of magma
with the ultramafic rocks representingits lower layers. chambers, whereasthe isotropicgabbroswereproducedby
These ultramafic rocks are overlain successivelyby the in situ crystallisationof magma chamberliquids.
cumulus gabbros and an upper horizon of isotropic Finally, the dunites and lherzolites possesscumulate
gabbros.Following our field and petrologicalstudieson textures that may representpart of the upper mantle
the complex,we concurwith the conclusions of Nivia et al. sequenceof the plateau. '
[ 1992] andsuggestthatthe BolivarComplexrepresents the The Bolivar Complex is intruded by a suite of
lower layers of the obducted Caribbean-Colombian plagioclase-hornblende pegmatiteveins,which can be up
oceanicplateau. These field observationson the Western to 2 m thick. As well as plagioclaseand hornblende
Cordillera, especiallythe better exposuresnear Bolivar, phenocrysts,the veins containvariable amountsof quartz
have enabledus to constructa hypotheticalcrosssection and dunite, peridotite and gabbro xenoliths.Unlike the.
throughthe Caribbean-Colombian oceanicplateau(Figure gabbros of the Bolivar Complex, the veins have
136 THE CARIBBEAN-COLOMBIAN PLATEAU

subduction-relatedgeochemicalsignature(La/Nb >>1); 1.6

they thereforeappearto be unrelatedto the formationof
the oceanic plateau. Nivia [1996] has proposedthat the
veins were formed during fluid-inducedpartial melting at ß Colombia
the base of the plateau, as it was obductedonto the
continentalmargin. 0.8- ß Gorgona
. o Cura(;ao
4. COMPOSITIONAL VARIATION WITHIN THE 0.4-- ß Haiti
CARIBBEAN-COLOMBIAN OCEANIC PLATEAU .
,• Costa Rica
ß Leg 15
o i i I
The preceding discussion emphasises that the o 10 20 30 4O
Caribbeanoceanicplateau comprisesa wide diversityof
MgO
basaltic magma compositions.It is logical to group the
lavas and intrusive rocks on the basis of chemical
differences that cannot be accountedfor by fractional .
crystallisation and must therefore result from mantle ß

heterogeneitiesor be causedby differing partial melting ß ß

processes.Three such types may broadly be identified A•AeA•

within the CCCIP (Figures5e and 12):
(a) tholeiitic lavas and sills having flat chondrite-
normalisedREE patterns(LaN/NdN = 0.9-1.1; SmN/YbN=
1--1.5) and Ti/Y - 250-325 and Nb/Zr = 0.05-0.1.
Examplesare the basaltsand picritesof the Curacaolava
successionand the basaltsof western Colombia. However, I I

despite the chondritic REE patterns,the Curacao lavas 1.0 1.5 2.0

possesspositive (eNd)•values (+6 to +7), which indicate LaN/Nd

that their mantle sourceregionhashad a long-termhistory
of light REE depletion,and has only beenenrichedin the
light REE relatively recently [Kerr et al., 1996c]. In terms
of their trace elements,these basaltsare compositionally
indistinguishablefrom those recovered from Pacific
oceanicplateaus(Figure 3).
(b) A secondgroup is representedby tholeiiticlight-REE-
depleted(LaN/NdN < 0.9) lavas and intrusiverocks,with
Ti/Y = 120-280 and Nb/Zr <0.05, and (eUd)i>+8. The
picrites, komatiites, and light-REE-depletedbasalts of
Gorgona Island; several of the Dumisseau lavas; and
basalts from DSDP Leg 15 (Site 152) are the main
representativesof this magma type. The Gorgonabasalts
are less light-REE-depleted than the komatiites and
picrites,thus precludinga simple fractional-crystallisation
geneticrelationship.Kerr et al. [ 1996b] haveproposedthat
the depleted basalts may have mixed with some more
enrichedGorgonamagmasen routeto the surface.
(c) Mildly alkalic basaltswith La•/Nd• >1.1; SmN/Yb•>2;
Ti/Y >350 andNb/Zr >0.1. Thesebasaltsare chiefly found
in Hispaniola, Costa Rica, Gorgona,and DSDP Leg 15,
Site 151. It is significantthat the areaswhere thesebasalts
occur appear to be at or near the edge of the CCCIP,
suggestingthat they may represent lesser degrees of
melting at the cooler periphery of the plume head.
N
Figure 12. Summary of compositionsof Colombian/Caribbean
Cretaceousplateaubasalts,(a) LaN/NdN vs %MgO; (b) LaN/NdN
vs SmN/YbN.
Alternatively thesebasaltsmay representlate-stagemelts
from the plume as it cooleddown.
Several authors have suggested that the CCCIP
representsthe melting of the initial 'plume head' phaseof
the Galfipagoshotspot[e.g., Duncanand Hargraves,1984;
Hill, 1993]. Although there is significant evidence to
suggestthat the voluminousmagmatismof the province
was causedby the melting of a large plume head,there is
little direct evidenceto supportthe conclusionthat it was
the Galfipagos hotspot [Kerr et al., 1996b]. The
geochemicalevidencefor the CCCIP being formedby the
ancestralGalfipagosplume head is also equivocal.Figure
3b shows that the Sr and Nd isotopic compositionof
present Galfipagoslavas overlaps substantiallywith the
compositionof lavasfrom the CCCIP. However,giventhat
the present-dayGalfipagoslavascover quite a wide range
from•Nd-- q-10to +4 and87Sr/86Sr
= 0.7025to 0.7035,this
overlap may just be coincidental; thus, whereas the
geochemicalevidenceis not inconsistent with a Galfipagos
hotspotoriginfor the CCCIP, it is not proofof sucha

Whether the CCCIP was formed by the Galfipagos
hotspotor anotherplume, it is obviousfrom the chemical
diversity of rocks exposedat the surfacethat the plume
sourceregion from which the CCCIP was derived was
markedly heterogeneous.Becausemost of the range of
chemicaland isotopiccompositions seenwithin the CCCIP
has been found within the small (8 x 3 km) island of
Gorgona,this heterogeneitymustbe on a relativelysmall
scale[Kerr et al., 19961:.].The evidencesuggests that there
are essentially two plume componentsmaking up this
heterogeneity,one with depletedand anotherwith more-
enrichedincompatibletrace elementcontentsand isotopic
ratios(relative to Bulk Earth). Kerr et al. [1995] and N. T.
Amdt et al. (ms. submittedto EPSL, 1996) have proposed
that the plume that producedthe CCCIP consistedof
enriched and more fusible streaksset in a depletedand
more refractorymatrix, which was also an integralpart of
the plume. The enrichedplume components probablyonly
accountfor <10% of the plumesourceregion(N. T. Arndt
et al., ms. submittedto EPSL, 1996); nevertheless,these
components,becauseof their enrichednature,can exert a
significant control on the chemistry of the erupted
magmas.
Previous drilling into the upper sectionsof several
Pacific oceanic plateaus has yielded basalts with
predominantly fiat incompatible trace element patterns
(e.g., Figures3a and 10b). In termsof isotopicratios,lavas
from individualplateaushave a more restrictedrangethan
the CCCIP. However, when taken togetherthey displaya
greaterdegreeof isotopicheterogeneity(Figure 3b). The
CCCIPmwhose deeper sectionshave been exposed by
obduction and erosion--is characterisedby basalts and
picritesthat possessa relativelywide rangeof enrichedand
depletedtrace elementcompositions and of isotopicratios
(e.g., Figures5, 6, 9, and 12). Is it possiblethen that the
lower parts of oceanicplateaushave more heterogeneous
basaltsand more picrites than the upper sections?If so,
then what mechanisms could be responsible for the
increasing chemical homogeneity and more evolved
basalticcompositions towardsthe top of a plateau?
The picrites of Curacao occur near the base of the
exposedlava succession [Kerr et al., 1996c].Additionally,
many continental flood basalt provinces (e.g., West
Greenland, the Siberian Traps, and the Deccan Traps)
preservesequencesthat have picritesnear their base [see
Campbell and Griffiths, 1992; Kerr et al., 1995, for
reviews]. Although there are exceptions(e.g., Deccan),
thesepicritesmostlytend to be more incompatible-element
depletedthan their associatedbasalts[Kerr et al., 1995].
Thus,the evidencefrom continentalflood basaltprovinces
and the CCCIP suggeststhat the early stagesof large
KERR ET AL. 137
igneous province formation may be characterisedby
picrites and basalts with heterogeneouschemical and
isotopicsignatures.
Considerationshould be given as to whether these
featurescould be attributedto magmaticplumbing.In the
initial stagesof volcanic activity, magma chamberswill
probablybe relatively small and poorly developedwithin
the cool and rigid lithosphere.There is thus a greater
chance that individual isotopically distinct picritic and
basalticmagmabatcheswill be ableto eruptwithoutbeing
'trapped' and homogenised in magma chambers.
Moreover, they may interact and become contaminated
with alteredoceaniccrust.In contrast,as the large igneous
province continues to develop, magma chamberswill
becomelarger and longer-livedin a crust/lithosphere that
is rheologically weaker. Also, in the later stages of
magmatismthe crustwill be substantially thicker,making
it more likely that denserpicritic magmaswill pond and
fractionate.Magma chamber trapping of picritic melts
would result in olivine fractionation, so reducing the
chanceof picrite eruption.Thus, althoughchemicallyand
isotopicallydistinctivepicrites and basaltsare probably
still formedduringmelt generationin theselater stagesof
plateau development,there is more likelihood of such
magmabatchesbeing trappedby large magmachambers
and undergoingextensive mixing and homogenisation,
thereby masking any mantle-derived heterogeneities.
Figures12 and 5e supportthesemagmamixing proposals
in that the three types (outlinedat the beginningof this
section)do not define distinctgroupsbut rather form a
broadcontinuumfrom depletedlavasto the more-enriched
lavas.
An alternative possibility is that the mantle plume
source region may have become more uniform in
compositionand cooler, so producingbasaltic primary
melts rather than picritic ones. Becauseplume heads
shouldentrainsurroundingmaterialduringascentthrough
the lower and, to a lesserextent, the upper mantle [e.g.,
Griffithsand Campbell,1990],theymay evolvetowardan
approximateaverage mantle composition[Stein and
Hofmann, 1994]. However, during the initial stage of
plumeimpact,particularlyasthe rapidlyuprisinghot axial
zone of the plume impingesbeneaththe lithosphere, there
may be extensiveinteractionwith the upperasthenosphere
and lower lithosphere,picking up any heterogeneities in
that region.Finally, becauseplumesare a dynamicmelt-
generatingsystem,there may be significantcompositional
heterogeneitiesresulting from the different depths of
continuedmelt extraction(N. T. Arndt et al., ms. submitted
to EPSL, 1996), but thesewill only be evidentin the early
stagesof plateaudevelopment,beforethe development
138 THE CARIBBEAN-COLOMBIAN PLATEAU
large magma chambersthat trap individually distinct
magmabatches.
There are several possiblereasonswhy the deeper
levelsof otherobductedplateausand, in particular,Ontong
Java are not exposed.The exposedsectionsof other
oceanicplateaus(particularlyOntongJava [e.g., Neal et
al., this volume]) are composedalmostentirelyof pillowed
and massive basalts with intrusive dolerite sheets,
representing uppercrustallevels(< 4 km). The reasonfor
this differencewith the CCCIP may relateto the ageof the
plateauat the time of imbricationand accretion.Ontong
Java formed in two major plateau-buildingepisodes,at
N120 and-90 Ma [Mahoney et al., 1993] and was
obducted over the North Solomon Trench at 20 Ma
[Kroenke,1974; Wells, 1989], i.e., at least70 m.y. after its
formation. In contrast, the Caribbean-Colombian oceanic
plateau appears to have been obducted in the Late
Cretaceous(Early Tertiary?) [Burke, 1988], probablyno
morethan 15 to 25 m.y. after formation.Consequently, the
Caribbean-Colombian plateau may still have been
relativelywarm andmorebuoyantthanthemucholderand
cooler plateaus such 3s Ontong Java and Wrangellia
[Lassiter et al., 1995] at the time of obduction.This
additionalthermal buoyancy of the CCCIP may be one
reasonwhy deepersectionsof this plateauwere obducted.
Anotherreasonwhy deepcrustalsequences havenot (yet)
beenexposedin the SolomonIslandsmay resultfrom the
greatercrustalthicknessof the OntongJava (25-42 km;
[Cooperet al., 1986;Furomotoet al., 1976]) as compared
to the Caribbean-Colombianoceanic plateau (8-20 km
[Edgar et al., 1971]).
The plume model of Campbellet al. [1989] predicts
that the picritic and thus the hottestmagmasassociated
with the initial plume head phaseof magmatismwill be
restrictedto a relativelynarrowregion(150 km diameter)
at the axis of the plume, over the so-calledtail, with the
productionof lower MgO magmas(< 12 wt%) in the
cooler outer zonesof the plume head. Althoughthe pre-
tectonicgeometryof the provinceis difficultto determine,
the presenceof 88-90 Ma picriticandpicriticbasaltlavas
throughout the CCCIP (Gorgona,CostaRica,Cura{;ao, and
in southern Colombia) would appear to require
anomalouslyhot mantle (>200 øC above ambientupper
mantletemperatures) within a large part of the proposed
plume head and thereforeseemsto be inconsistent with
standardplume models [e.g., Campbell et al., 1989;
Saunderset al., this volume].
Onepossibleexplanationis thattheplumeheadwasnot
the conventionallyassumedmushroomshape.Both Sleep
et al. [1988] and White [1992] suggested
that plumeheads
can also take the form of tabular upwellingblobs of hot
material. If so, the hot ascendingmantle might occupya
broadarea,notjust a narrowaxial conduit.
The tectonic setting and structureof the lithosphere
above the ascendingplume head could also play an
important role [cf. Thompson and Gibson, 1993] in
determining whether or not hot mantle capable of
generatingpicritesis presentbelow a region.In section2,
we notedthat the CCCIP may have formedat a spreading-
ridge triple junction. The presenceof a triple junction in
the lithosphereabove the impinging plume head could
potentially provide a way for hot plume material to
become channelledrapidly along the spreadingridges to
the more distal parts of the plume head, for instance,
propagatingeastwardsinto the Caribbeanregion.This may
have allowed eruption of picrites far from the perceived
centreof the plume.
5. PLATEAU OBDUCTION
The abundantmafic volcanicsequences throughoutthe
length of westernColombia are arrangedin coast-parallel
lenses alternating with graywacke-shalemetasediments
[see Nivia, 1996; also Bourgois et al., 1982] that are
mostly stronglydeformedand vary in metamorphicgrade
from chlorite-pumpellyitefacies phyllites to schistsand
occasionalmigmatiticgneisses.Along the easternmargin
of the mafic volcanic belt, marked by the RomeralFault,
local exposuresof blueschistsand ultramaficrocks (now
talcose and serpentinised) form lenses within the
metasediments. These relationshipsare best accountedfor
by subduction-accretionprocessesat an active margin
(lateral crustal accretion). Indeed, deep crustalxenoliths
broughtup in the Mercaderesvolcanicdiatreme[Weberet
al., 1995], which liesjust eastof the easternmarginof the
mafic volcanic zone in southwesternColombia, include
mixed high-grademafic and metasedimentary rock types
that are consistent with subduction-accretion. The essence
of subduction-accretion is that continent-derived
graywacke-shalesedimentis tectonicallymixed with up-
standingmafic structureson the oceanfloor (arcs,aseismic
ridges,plateaus,etc.) that are scrapedoff the subducting
oceanslab [cfi, SengOtand Okuogullari, 1991; von Heune
and Scholl, 1993]. It is interestingto considerthe fate of
oceanicplateausin this environment.Becausesubduction
of old, cold lithosphereis the dominantdriving force of
plate tectonics[e.g., Carlson et al., 1983], it is assumed
that warm, buoyantplateauswill behavedifferentlywhen
they reach an active margin [Cloos, 1993; Saunderset al.,
1996]. To what extent are they obductedor underplated
[Kimura and Mukai, 1991]?
Undoubtedly, the sheer areal extent of the

volcanic complex in Colombia and the Caribbeanimplies
that significantamountswere not subducted.However,this
material is dominantly hydrothermallyalteredbasaltsand
basaltic sheets, and it is only locally that the deeper,
fresherpartsof the plateaucrustare seen(e.g., the Bolivar
UltramaficComplex).Decol16ment zonesare importantin
accretionaryterranes [e.g., Moore, 1989]; Kirnura and
Ludden [1995] have made the point that the decol16ment
zone in normal oceancrustprobablyoccursat the baseof
the hydrothermal circulation cells, where rheologically
weak altered rock is underlain by fresh dolerites and
gabbros.The upper zonesare peeled off and imbricated
with the developingsedimentaryaccretionarywedge (as
observedin westernColombia)whereasthe fresherbasalts
and gabbrosare subductedor underplated.
Of course, another, deeper, rheologicallyweak zone
lies beneaththe Moho in oceanic lithosphere,which is
essentiallytemperature-dependent (> 1000 øC [Nicolasand
Violette, 1982]), and is usually the zone along which
ophiolitecomplexesare obductedand alongwhich thereis
normally metamorphismand melting as the hot ophiolitic
sheetis thrustedover continentalmarginsediments[Searle
and Malpas, 1982]. It is probablethatthe Bolivar Complex
with its layered and bandedperidotites,pyroxenites,and
gabbrosrepresentsthe deeper,hotter part of the plateau
that was underthrustedby the deformedmetasedimentsof
the accretionarywedge. If so, the abundanthornblende-
pyroxene pegmatites may result from hydrous partial
melting of the peridotiresand gabbrosas fluids expelled
from the underthrusted sediments entered the base of the
thrust sheet. Caution is necessarybecause it has been
shown that ophiolites (e.g., Oman ophiolite) have an
abundance of gabbroic veins just beneath the Moho
[Boudierand Nicolas, 1995], which is roughlywhere they
occurat Bolivar. In contrast,whereasthe former are 'dry'
gabbroic veins, a majority of the veins at Bolivar are
spectacular coarse hornblende-anorthosite pegmatites,
transitionalto hornblende-biotitetonalites,and some (the
earlier) are quite stronglydeformed.Suchveins are found
wherever these deep sectionsare exposedin Colombia
[Bourgois et al., 1982; Tistl et al., 1994]. If they are
associated with fluid entry into the baseof the imbricating
plateau,theypotentiallyoffer a meansof datingthe time of
obductionand/or subcretionof an oceanicplateau at a
continentalmargin. However, a K-Ar age of 108 ñ 16 Ma
on hornblendepegmatitesfrom Bolivar [Barrero, 1979]
and K-Ar hornblendeages clusteringaround-•20 Ma for
the CondotoComplexsouthwestof Medellin [Tistl et al.,
1994] have been interpreted as primary intrusive
emplacementagesratherthan tectonicemplacementages
andneedto be supplemented
by 4øAr-39Ar
agesto resolve
these alternatives.
KERR ET AL. 139
Figure 13 illustratesa possibletectonic model for
plateau emplacement,with imbricationat two levels.
Plateaushave a potentially longer residencetime at the
Earth's surfacethan normal oceaniccrust [Kroenke, 1974;
Cloos, 1993; Saunders et al., 1996] and may be more
difficult to subduct. However, the evidence from the
Caribbean-Colombian plateausuggests
that (in comparison
with supra-subduction
zone ophiolites)it is the upperpart
that is imbricated and obducted,whereasthe deeperzones
are subductedor tectonicallyunderplatedbeneathactive
continentalmarginsand only occasionallyobducted.This
beginsto resolvethe apparentparadox,identifiedby Burke
[1988], that obducted plateaus are rather rare in the
geological record: mostly it may be only the upper,
volcanic sections of plateaus that are imbricated and
obducted, as may be the case with some Precambrian
greenstonebelts [e.g., Kusky and Kidd, 1992]. What
happensto the deeper parts of plateausis a matter for
speculation,but if underplatedbeneathcontinentalmargins
and melted, they could make a major contributionto
crustalgrowth, adding material with relatively primitive
isotopic characteristics[e.g., Tarney and Jones, 1994;
Abbottand Mooney, 1995].
6. ARC-DERIVED BASALTS ASSOCIATED
WITH THE CCCIP
The marginsof the Caribbeanare markedby numerous
occurrencesof arc-derivedlavas and equivalentintrusive
rocks [e.g., Donnelly et al., 1990; Donnelly, 1994]. The
lavasaretypicallyislandarctholeiites(calledthe primitive
islandarc seriesby Donnelly et al. [1990]), and the main
exposuresare found in Hispaniola,Cuba, Puerto Rico,
Venezuela,Ecuador,Bonaire,partsof the LesserAntilles,
andthe Quebradagrande complexeastof the Romeralfault
in Colombia. Nowhere are these arc lavas found to be
interlayeredwith the Caribbeanoceanicplateaubasalts,so
whereasthere is a closespatialrelationship,the natureof
the relationshipneedsto be clarified.The arc-derivedlavas
andthe basaltsof the CCCIP overlapin age,but the island
arc rocksappearto be slightlyolder(Cenomanian-Albian;
90-112 Ma) than the plateau-derivedbasalts.Donnelly
[ 1994] proposedthat the islandarc basaltsformedon top
of the plateau at its margin. However, until this is
confirmedby careful dating, we considerit more likely
that the island arc tholeiites are the volcanic products
associated with the pre-plateau-collision oceanic/
continental arc in the Caribbean-Colombianregion and
have been accretedontothe edgeof the plateauduringthe
early stagesof obduction.
Numerous calc-alkaline tonalitic batholiths of Late
Cretaceousage (< 90 Ma) with some associated
140 THE CARIBBEAN-COLOMBIAN PLATEAU

(a) OCEANIC
PLATEAU __•-Hydrothermal
Base TurbiditesDelta
Rivers

Fresllgabbro + + ..•,+.++ + '.........

, + + .+ + + + ++++.+•+. _+.+ ++++. +•+. +-+.• ...... +++++ •
i+. ++ + ++.++ _+_:_+.-L+.2+•
=•.++._++..+•+•• •
MOHO•-• -- • •--•-- Platedepressed
Cumulates
&Layered
Peridotite1000øC
(ductile)

L .---' Detachments
[
Potential

Bolivar Metasedirnents
gabbros & / x
(b) peridotites
/ \ M_afi..c.
lavas
k4- ' : ,; • i i
%_'• 4- ' i t I! i]; '; •" I"l';'

' ß =.... , i'll

' ....... ' ' :'' i ,;t It i,ii,I
I ,:111
i :i,1
.... !•!i:t
:t
' '1,,
,,,1i i1,11:
'x,xamphio,tes
',,,',,, Thrusts", "-.', ' ......
' ',,,,", ' '"• Underplated
-... Plateau?
""-...

Figure13. Schematic obduction-imbrication

modelfortheCaribbean-Colombian plateau.
(a) Thick(>15 km)plateau
crust formed in close proximityto the continentalmargin of northwesternSouthAmerica.The marginwas
accretionary,
comprising scraped-off
turbiditemetasediments,
amphibolites,
etc.Two potentialdetachment zonesin
theapproaching plateauareshown,oneatthebaseof thehydrothermalcirculation
zone,theothermuchdeeper, within
themantlelithospherebelowtheplateau.(b) Whentheplateaureached thesubductionzone,onlythedeeperzonesof
theplateauweresubducted. Theupperlayersarescraped-off
andimbricated alongthetwopossible detachments
and
accretedontothe continentalmargin.Seetext for furtherdetails.

alkaline extrusives are found around relatively close proximity to the western margin of
the Caribbean
marginsand in Central America [Donnellyet al., 1990] northern South America. One possibilityis that if the
andappearto largelypostdate theislandarctholeiiteevent. ascendingplume was channelledalong an eastward-
The inceptionof this calc-alkalinephaseof magmatism propagating rift linked to the separatingNorth and South
appearsto coincidebothspatiallyandtemporallywith the American Plates as a result of Atlantic opening,then the
formationof the CCCIP, perhapspenecontemporaneously. time gap between the formation of the (hot) oceanic
Donnelly et al. [1990] suggested that the changefrom plateau and its interaction with the westward-moving
island arc tholeiites to calc-alkalinemagmatismmay be continentscouldbe quite short.
relatedto 'a major changein plate movements.'On the
basisof significantnew chemicaldata from Hispaniola, 7. CONCLUSIONS
Lebr6nandPerfit [ 1994] suggested thatthetransitionfrom
island arc tholeiites to calc-alkaline volcanism in the 1. The CCCIP represents the dismembered remnantsof
Caribbeanwas causedby the partial subductionof the an oceanicplateauthat formedin the easternPacificOcean
buoyantCaribbeanoceanicplateauchokingthesubduction at-87-90 Ma. Shortlyafter its formationthe northernhalf
zone,thuspromotingreversalof subduction polarity.Our of the plateau was pushed between North and South
observationson Aruba suggesta close link between America to form the bulk of the Caribbean Plate. As the
tonaliteemplacement and the deformation/imbrication of southernpart of the plateau approachedthe continental
theplateauagainstthe SouthAmericancontinental margin. margin of northwesternSouthAmerica,it wastoo buoyant
The age similaritybetweenthe formationof the CCCIP to totally subductand so its topmostpartsaccretedontothe
and the inceptionof voluminoustonaliticmagmatismis continental margin.
intriguingand suggests that,if the modelof Lebr6nand 2. The lavas that occur nearerthe baseof the exposed
Perfit [1994] is correct,the CCCIP probablyformedin CCCIP tend to be (a) more picritic and (b)

compositionallymore heterogeneousthan basalts from
nearer the top of the succession,which possessnear-
chondritic REE patterns and are quite uniform in
composition.This observationmight be explainedby the
developmentof larger magmachambersin the later stages
of plateauevolution,which couldact as a trap for picritic
magmasin which heterogeneous picritic magmascould
both fractionateand become homogenised.The inferred
poorlydevelopedmagmachambers in the earlierstagesof
plateau evolution mean that heterogeneouslavas and
picritesare more likely to reachthe surface.
3oThe basalts,picrites,and komatiitesof the province
require a markedly heterogeneous plume sourceregion,
containingat leasttwo components.
4. The gabbros,norites,gabbronorites,dunites,olivine
websterites, and lherzolites of the Bolivar Ultramafic
Complex provide a glimpse of the compositionof the
deeper,subvolcanicsectionsof an oceanicplateau.From
the Bolivar Complex and other accreted terranes in
Colombia,a probablesectionthroughthe deeperpartsof
an oceanicplateauhasbeenconstructed. From the top, this
section consists of homogeneousbasalts, followed by
heterogeneouspicrites and komatiites, and isotropic
gabbro.This isotropicgabbrolies abovelayeredgabbro,
with dunite containing bands of lherzolite, olivine
websterite,and olivine gabbronoritesat still lower levels.
5. Island arc tholeiites (both of similar age and older
than the CCCIP) are found around the margins of the
Caribbeanand to the eastof the accretedplateausequences
in Colombia.It is likely that theserepresentthe remnants
of the pre-plateau-collision volcanicarc. The presenceof
primitivetonaliticbatholiths(< 90 Ma) closelyassociated
with the CaribbeanPlateauand plateauimbricationmay
recorda possiblereversalof subduction polaritycausedby
the attemptedobductionof the plateau.
6. Continueddetailedstudyof the CCCIP couldtell us
muchmore aboutthe internalstructureof oceanicplateaus,
as well as the fate of plateausand their interactionwith
continentalmargins.The resultsshouldhave an important
bearing on understandingPrecambriangreenstonebelts
andprimitivecrustalgrowthmechanisms, aswell asonthe
evolutionof the Caribbeanregion itself. There is a great
needfor detailed,reliablegeochronology in theregion.
Acknowledgments. We are gratefulto Leon Pors,Carmabi
Institute,Curaqao,ArmandoCuret and Vivi Ruiz, Aruba, and
Manuel Itturalde-Vinente,Havana, Cuba, for all their help with
the fieldwork, and Dirk Beets,GerardKlaver, andNick Arndt for
much advice and discussion.RosemaryHickey-Vargas,Bob
Duncan, and John Mahoney are thanked for constructive
criticismof the manuscript.This work was supportedby Grants
GR9/583A & GR3/8934 from the Natural Environment Research
Council.
KERR ET AL. 141
REFERENCES
Abbott, D., and W. Mooney, The structuraland geochemical
evolution of the continental crust- support for the oceanic
plateaumodel of continentalgrowth,Rev. Geophys.,33, 231-
242, 1995.
Aitken, B. G., and L. M. Echeverria,Petrologyand geochemistry
of komatiites from Gorgona Island Colombia, Contrib.
Mineral. Petrol., 86, 94-105, 1984.
Aspden, J. A., W. J. McCourt, and M. Brook, Geometrical
control of subduction-relatedmagmatism:the Mesozoic and
Cenozoic plutonic history of westernColombia,d. Geol. Soc.
London, 144, 893-905, 1987.
Bandy, O. L., and R. E. Casey, Reflector horizons and
paleobathymetrichistory,easternPanama,Geol. Soc.Am. Bull.,
84, 3081-3086, 1973.
Barrero, D., Geology of the Central westernCordillera, west of
Buga and Roldanillo, Colombia,Publ. Especial,Ingeominas,4,
1-75, 1979.
Beck, C. M., D. Girard, and P. DeWeber, Le "Volcano-
s6dimentaire du Rio Guare": Un 616ment de la nappe
ophiolitique de Lomo de Hierro, cha6ne Cara6be
V6n•zu61ienne,Paris, Academyof Science,ComptesRendusde
Seances(D), 299, 337-342, 1984.
Beets, D. J., G. Th. Klaver, F. F. Beunk, G. Kieft, and P.
Maaskant,Picritesasparentalmagmaof MORB-type tholeiites,
Nature, 296, 341-343, 1982.
Beets, D. J., W. V. Maresch, G. Th. Klaver, A. Mottana, R.
Bocchio, F. F. Beunk, and H. P. Monen, Magmatic rock series
and high-pressuremetamorphismas constraintson the tectonic
history of the southernCaribbean,Geol. Soc. Am. Mere., 162,
95-130, 1984.
Ben Avraham, Z., A. Nur, D. Jones, and A. Cox, Continental
accretion: from oceanic plateaus to allochthonousterranes,
Science,213, 47-54, 1981.
Bence, A. E., J. J. Papike, and R. A. Ayuso, The petrology of
submarinebasaltsfrom the centralCaribbean:DSDP Leg 15, d.
Geophys.Res.,80, 4775-4804, 1975.
Berrang6,J.P., and R. S. Thorpe, The geology,geochemistryand
emplacement of the Cretaceous-Tertiaryophiolitic Nicoya
complex of the Osa Peninsula, southern Costa Rica,
Tectonophysics, 147, 193-220, 1988.
Boudier, F., and A. Nicolas, Nature of the Moho transition zone
in the Oman ophiolite,d. Petrol., 36, 777-796, 1995.
Bourgois, J., B. Calle, J. Tournon, and J-F. Toussaint, The
Andean ophiolitic megastructureson the Buga-Buenaventura
transverse (Western Cordillera-Valle Colombia),
Tectonophysics, 82, 207-229, 1982.
Bourgois, J., J. Azema, P.O. Baumgartner, J. Tournon, A.
Desmet, and J. Aubouin, The geologic history of the
Caribbean-Cocosplate boundarywith specialreferenceto the
Nicoya complex(CostaRica) and DSDP results(Legs 67 and
84 off Guatemala): a synthesis,Tectonophysics,108, 1-32,
1984.
Bourgois, J., J.-F. Toussaint,H. Gonzales,A. Orrego, and B.
Calle, Les ophiolites des Andes de Colombia, Evolution
structural et signification geodinamic, in Geodynamic
142 THE CARIBBEAN-COLOMBIAN PLATEAU
Caraibbes Symposium,edited by A. Mascle, pp. 475-493,
Technip-Paris,1985.
Bourgois,J., J-F. Toussaint,H. Gonzales,J. Az6ma, B. Calle, A.
Desmet, L. A. Murcia, A. P. Acevedo, E. Parra, and J. Tournon,
Geologicalhistory of the Cretaceousophioliticcomplexesof
northwestern South America (Colombian Andes),
Tectonophysics,143, 307-327, 1987.
Burke, K., Tectonic evolution of the Caribbean,Ann. Rev. Earth
Planet. Sci., 16, 201-230, 1988.
Burke, K., C. Cooper, J. F. Dewey, P. Mann, and J. L. Pindell,
Caribbeantectonicsand relativeplate motions,Geol. Soc.Am.
Mem., 162, 31-63, 1984.
Burke, K., P. J. Fox, and M. C. Seng6r,Buoyantoceanfloor and
the origin of the Caribbean,d. Geophys.Res., 83, 3949-3954,
1978.
Campbell, I. H., R. W. Griffiths, and R I. Hill, Melting in an
Archcan mantle plume: headsit's basalts,tails it's komatiites,
Nature, 339, 697-699, 1989.
Campbell,I. H., and R. W. Griffiths, The changingnatureof
mantle hotspotsthrough time: Implicationsfor the chemical
evolutionof the mantle,d. Geol., 92, 497-523, 1992.
Carlson, R. L., T. W. C. Hilde, and S. Uyeda, The driving
mechanismof plate tectonics:relationto ageof the lithosphere
at trenches,Geophys.Res.Lett., 10, 297-300, 1983.
Case,J. E., S. L. G. Duran, A. Lopez, and W. R. Moore, Tectonic
investigationson westernColombia and easternPanama,Geol.
Soc. Am. Bull., 82, 2685-2712, 1971.
Case, J. E., J. Barnes, G. Paris, I. H. Gonzales, and A. Vina,
Trans-Andeangeophysicalprofile, southernColombia, Geol.
Soc. Am. Bull., 84, 2895-2904, 1973.
Case, J. E., W. D. MacDonald, and P. J. Fox, Caribbean crustal
provinces;seismicand gravity evidence,in The Geology of
North America, vol. H.' The Caribbean Region, edited by G.
Dengo and J. E. Case, pp. 15-36, Geological Society of
America, Boulder, CO, 1990.
Cloos, M., Lithospheric buoyancy and collision orogenesis:
Subductionof oceanic plateaus,continentalmargins, island
arcs, spreadingridges and sea mounts, Geol. Soc. Am. Bull.,
105, 715-737, 1993.
Coffin, M. F., and O. Eldholm,Large igneousprovinces:crustal
structure, dimensions and external consequences,Rev.
Geophys,32, 1-36, 1994.
Cooper,A. K., M. S. Marlow, and T. R. Bruns,Deep structureof
the central and southernSolomonIslandsregion: implications
for tectonic origin, Circum-Pacific Councilfor Energy and
Mineral Resources,Earth ScienceSeries, 4, 137-175, 1986.
Donnelly, T. W., Late Cretaceousbasaltsfrom the Caribbean, a
possibleflood basaltprovinceof vastsize,Eos Trans.AGU, 54,
1973.
Donnelly, T. W., Geologichistory of the Caribbeanand Central
America, in The Geology of North America. vol. A: An
Overview,edited by A. W. Bally and A. R. Palmer, pp. 299-
321, GeologicalSocietyof America,Boulder,CO, 1989.
Donnelly, T. W., The CaribbeanCretaceousbasaltassociation:A
vast igneousprovince that includesthe Nicoya Complex of
CostaRica, Profil, (Universityof StuttgarO7, 17-45, 1994.
Donnelly, T. W., D. Beets,M. J. Carr, T. Jackson,G. Klaver, J.
Lewis,R. Maury, H. Schellenkens,
A. L. Smith,G. Wadge,and
D. Westercamp,History and tectonic setting of Caribbean
magmatism,in The Geology of North America, Vol. H: The
CaribbeanRegion,editedby G. DengoandJ. E. Case,pp. 339-
374, GeologicalSocietyof America, Boulder,CO, 1990.
Donnelly, T. W., W. Melson, R. Kay, and J. J. W. Rogers,
Basaltsand doleritesof late Cretaceousage from the central
Caribbean,Init. Repts. Deep Sea Drill. Proj., 15, 989-1012,
1973.
Donnelly,T. W., and J. J. W. Rogers,The distributionof igneous
rocksthroughoutthe Caribbean,Geol. Mijnbouw,5 7, 151-162,
1978.
Duncan,R. A and R. B. Hargraves,Platetectonicevolutionof the
Caribbean region in the mantle reference frame, in The
Caribbean-South American Plate Boundary and Regional
Tectonics,Mem. 162, editedby W. E. Bonini, R. B. Hargraves,
and R. Shagam,pp. 81-93, Geological Society of America,
Boulder, CO, 1984.
Edgar,N. T., J. I. Ewing, and J. Hennion, Seismicrefractionand
reflection in the Caribbean Sea, Am. Assoc.Petrol. Geol., 55,
833-870, 1971.
Escalante,G., The Geology of SouthernCentral America and
westernColombia, in: The Geologyof North America, 17ol.H:
The CaribbeanRegion,editedby G. DengoandJ. E. Case,pp.
201-230, GeologicalSocietyof America,Boulder,CO, 1990.
Farnetani,C. G., M. A. Richards,andM. S. Ghiorso,Petrological
modelsof magma evolution and deep crustalstructurebeneath
hotspotsand flood basaltprovinces,Earth Planet.Sci. Lett. (in
press)
Furumoto,A. S., J.P. Webb, M. E. Odegard,andD. M. Hussong,
Seismic studies of the Ontong Java Plateau, 1970,
Tectonophysics, 34, 71-90, 1976.
Gansser, A., Facts and theories on the Andes, d. Geol. Soc.
London, 129, 93-131, 1973.
Goossens,P. J., W. I. Rose, and D. Flores,Geochemistryof the
tholeiitesof the basicigneouscomplexof NW SouthAmerica,
Geol. Soc. Am. Bull., 88, 1711-1720, 1977.
Griffiths, R. W., and I. H. Campbell, Stirring and structurein
mantle starting plumes, Earth Planet. Sci. Lett., 99, 66-78,
1990.
Hall, S. A., Oceanic basementof the Caribbean basins,Geol. Soc.
Am. Abstractswith Programs,27, p. A153, 1995.
Harland,W. B., A. L. Armstrong,A. V. Cox, L. E. Craig, A. G.
Smithand D. G. Smith,A GeologicTimeScale1989, 263 pp.,
CambridgeUniv. Press,Cambridge,1990.
Hill, R. I., Mantle plumes and continentaltectonics,Lithos, 30,
193-206, 1993.
lturralde-Vinent,M. A., Cuban geology:A new plate tectonic
synthesis,d. Petroleum Geol., 17, 39-70, 1994.
Kent, R. W., B. S. Hardarson,A.D. Saunders,and M. Storey,
Plateaus ancient and modern: Geochemical and
sedimentological perspectives on Archaean oceanic
magmatism,Lithos, 37, 129-142, 1996.
Kerr, A. C., A.D. Saunders,J. Tarney, N.H. Berry, and V. L.
Hards, Depleted mantle plume geochemicalsignatures;no
paradoxfor plume theories,Geology,23, 843-846, 1995.
Kerr, A. C., J. Tarney, G. F. Marriner, A. Nivia, A.D.

and G. Th. Klaver, The geochemistryand tectonic settingof
late Cretaceous Caribbean and Colombian volcanism, d. S. Am.
Earth Sci., 1996a (in press).
Kerr, A. C., G. F. Marriner, N. T. Arndt, J. Tarney, A. Nivia, A.
D. Saunders,and Duncan,R. A., The petrogenesis of Gorgona
komatiites,picrites and basalts:new field, petrographicand
geochemicalconstraints,Lithos,37, 245-260, 1996b.
Kerr, A. C., J. Tarney, G. F. Marriner, G. Th. Klaver, A.D.
Saunders, and M. F. Thirlwall, The geochemistry and
petrogenesisof the late-Cretaceouspicrites and basalts of
Curaqao,NetherlandsAntilles: a remnantof an oceanicplateau,
Contrib. Mineral. Petrol., 124, 29-43, 1996c.
Kimura, G., and A. Mukai, Underplatedunits in an accretionary
complex:Melange of the ShimantoBelt of easternShikoku,
south-westJapan,Tectonics,10, 31-50, 1991.
Kimura, G., and J. Ludden, Peeling oceaniccrust in subduction
zones,Geology,23, 217-220, 1995.
Klaver, G. Th., The Curaqao lava formation an ophiolitic
analogue of the anomalousthick layer 2B of the mid-
Cretaceousoceanicplateausin the westernPacific and central
Caribbean, Ph.D. Thesis, University of Amsterdam, The
Netherlands, 1987.
Kroenke, L. W., Origin of continentsthrough developmentand
coalescenceof oceanicflood basaltplateaus.Eos Trans. AGU,
55, p. 443, 1974.
Kusky, T. M., and W. S. F. Kidd, Remnantsof an Archean
oceanic plateau, Belingwe greenstone belt, Zimbabwe,
Geology,43, 43-46, 1992.
Ladd, J. W., T. L. Holcombe,G. K. Westbrook,andN. T. Edgar,
Caribbean margin geology: active margins of the plate
boundary, in The Geology of North America, Vol. H.' The
CaribbeanRegion,editedby G. DengoandJ. E. Case,pp. 261-
290, GeologicalSocietyof America,Boulder,CO., 1990.
Lassiter,J. C., D. J. DePaolo, and J. J. Mahoney, Geochemistry
of the Wrangellia flood basalt province: Implications for the
role of continental and oceanic lithospherein flood basalt
genesis.d. Petrol., 36, 983-1009, 1995.
Lebr6n, M. C., and M. R. Perfit, Petrochemistryand tectonic
significanceof Cretaceousisland-arcrocks,Cordillera Central,
DominicanRepublic, Tectonophysics, 229, 69-100, 1994.
MacKenzie, D. B., High temperaturealpine-typeperidotitefrom
Venezuela, Geol. Soc. Am. Bull., 71,303-318, 1960.
Mahoney,J. J., M. Storey,R. A. Duncan,K. J. Spencer,andM.
Pringle,Geochemistryand age of the OntongJavaPlateau,in
The Mesozoic Pacific: Geology, Tectonics,and Volcanism,
Geophys.Monogr.Set., vol. 77, editedby M. S. Pringle,W. W.
Sager, W. V. Sliter, and S. Stein, pp. 233-261, AGU,
Washington,D.C., 1993.
Marriner, G. F., and D. Millward, The petrology and
geochemistryof Cretaceousto Recent volcanismin Colombia:
The magmatichistory of an accretionaryplate margin, d. Geol.
Soc. London, 141,473-486, 1984.
Martin Bellizia, C., and J. M. de Arozena, Complejoultramfifico
zonado de Tausabana-E1Rodeo, gabro zonado de Siraba-
Capuanay complejo sub-volcfinicoestratificadode SantaAna,
Paraguanfi,Estado Falc6n. Trans. 6th Caribbean Geol. Conf.,
Margarita, pp. 337-356, 1972.
KERR ET AL. 143
Maurrasse,F., J. Husler, G. Georges,R. Schmitt,andP. Damond,
Upraised Caribbean sea floor below acousticreflector "B" at
the southernpeninsulaof Haiti. Geol. Mo'nbouw,58, 71-83,
1979.
McCourt, W. J., J. A. Aspden, and M. Brook, New geological
and geochronological data from the Colombian Andes:
continentalgrowth by multiple accretion,d. Geol. Soc.London,
141,831-845, 1984.
McGarvie D. W., Torfajokull: a volcano dominatedby magma
mixing, Geology,12, 685-688, 1984
McGeary, S., and Z. Ben-Avraham,The accretionof Gorgona
Island, Colombia: multichannel seismic evidence, in
TectonostratigraphicTerranes of the Circum-Pacific Region,
edited by D. G. Howell, pp. 543-554, Circum-PacificCouncil
for Energyand Mineral Resources,Houston,TX, 1986.
Meyerhoff, A. A., and H. A. Meyerhoff, Continentaldrift. IV:
The Caribbean 'Plate', d. Geol., 80, 34-60, 1972.
Millward, D., G. F. Marriner, and A.D. Saunders,Cretaceous
tholeiitic rocks from the western Cordillera of Colombia, d.
Geol. Soc. London, 141,847-860, 1984.
Moore, J. C., Tectonicsandhydrogeologyof accretionaryprisms:
role of the decollementzone, d. Struct. Geol., 11, 95-106, 1989.
Nicolas, A., and J. F. Violette, Mantle flow at oceanicspreading
centres:models derived from ophiolites, Tectonophysics,81,
319-379, 1982.
Nivia, A. Geochemistryand origin of the Amaime and Volcanic
sequences, Southwestern Colombia, M. Phil. Dissertation,
Universityof Leicester,Leicester,England,163 pp., 1987.
Nivia, A., The Bolivar mafic-ultramaficcomplex,SW Colombia:
the baseof an obductedoceanicplateau,d. S. Am. Earth Sci. (in
press).
Nivia, A., N. Galvis, andM. Maya, Geologiade la Plancha242 -
Zarzal. Ingeominas,Bogota, 163 pp., 1992.
Nur, A., and Z. Ben Avraham, Oceanic plateaus, the frag-
mentationof continents,and mountainbuilding, d. Geophys.
Res., 87, 3644-3661, 1982.
Oxburgh,E. R., Geology and metamorphismof Cretaceousrocks
in Eastern Carabobo State, Venezuelan Coast Ranges, in
Caribbean GeologicalInvestigations,Mere. 98, editedby H. H.
Hess, pp. 241-310, Geological Society of America, Boulder,
CO, 1966.
Pindell, J. L., Geologicalargumentssuggestinga Pacific origin
for the Caribbean plate, in Transactions of the Twelfth
Caribbean Geological Conference,St. Crib, Virgin Islands,
edited by D. K. Larue and G. Draper, pp. 1-4, Miami
GeologicalSociety,Miami, FL, 1990.
Pindell, J. L., S.C. Cande, W. C. Pitman, D. B. Rowley, J. F.
Dewey, J. LaBrecque, and W. Haxby, A plate-kinematic
frameworkfor modelsof Caribbeanevolution,Tectonophysics,
155, 121-138, 1988.
Pindell, J L., and S. F. Barrett, Geologic evolution of the
Caribbeanregion:A plate tectonicperspective,in The Geology
of North America, Vol. H: The CaribbeanRegion,editedby G.
Dengo and J. E. Case, pp. 405-432, Geological Society of
America, Boulder, CO, 1990.
Priem, H. N. A., D. J. Beets, N. A. I. M. Boelrijk, and E. H.
Hebeda, On the age of the late Cretaceous
144 THE CARIBBEAN-COLOMBIAN PLATEAU
batholithon Aruba, NetherlandsAntilles (southernCaribbean
borderland).Geol.Mo'nbouw,65, 247-265, 1986.
Saunders,A.D., Geochemistryof basaltsfrom the Nauru Basin
DSDP Legs 61 and 89: Implicationsfor the origin of oceanic
flood basalts,Init. Repts.Deep Sea Drill. Proj., 89, 499-517,
1985.
Saunders,A.D., J. Tarney, A. C. Kerr, and R. W. Kent, The
formationand fate of largeigneousprovinces,Lithos,37, 81-
95, 1996.
Schubert,C., Historical Geology of the Antillean-Caribbean
Region,Hafner,New York, 811 pp., 1935.
Searle,M.P., and J. Malpas,Petrochemistry
and origin of sub-
ophiolite metamorphic and related rocks in the Oman
mountains,J. Geol. Soc. London, 139, 5-248, 1982.
Sen,G., R. Hickey-Vargas,D. G. Waggoner,and F. Maurrasse,
Geochemistry of basalts from the Dumisseau Formation,
southernHaiti: implicationsfor the origin of the Caribbeansea
crust,Earth Planet. Sci. Lett., 87, 423-437, 1988.
Seng6r, A .M.C., and A. H. Okuogullari, The role of
accretionary wedges in the growth of continents:Asiatic
examplesfrom Argandto platetectonics,Ecol. Geol. Helv., 84,
535-588, 1991o
Sinton, C. W., and R. A. Duncan, Temporal evolutionof the
Caribbean Cretaceousbasalt province, results of 40Ar-39Ar
dating,Eos Trans.AGU, 73, 532, 1992.
Sinton,C. W., R. A. Duncan,andM. Storey,40Ar-39Aragesfrom
Gorgona Island, Colombia and the Nicoya Peninsula,Costa
Rica (abstract),Eos Trans.AGU, 74, 553, 1993.
Sleep,N.H., M. A. Richards,andB. H. Hager,Onsetof mantle
plumesin the presenceof pre-existingconvection,
J. Geophys.
Res., 93, 7672-7689, 1988.
Spadea,P., A. Espinosa,andA. Orrego,High-Mg extrusiverocks
from the Romeral Zone ophiolites in the southwestern
ColombianAndes, Chem. Geol., 77, 303-321, 1989.
Stearns,C., F. J. Mauk, and R. Van der Voo, Late Cretaceousto
early Tertiary palcomagnetism of Aruba and Bonaire,
NetherlandsAntilles,J. Geophys.Res.,87, 1127-1141,1982.
Stein, M., and A. W. Hofmann, Mantle plumesand episodic
crustalgrowth. Nature, 372, 63-68, 1994.
Storey, M., J. J. Mahoney, L. W. Kroenke, and A.D. Saunders,
Are oceanicplateaussitesof komatiiteformation?Geology,19,
376-379, 1991.
Sun, S.-s, and W. F. McDonough, Chemical and isotopic
systematics of oceanic basalts: Implications for mantle
compositionsand processes,in Magmatism in the Ocean
Basins, Spec. Publ. 42, edited by A.D. Saundersand M. J.
Norry, pp. 313-345, The GeologicalSociety,London,1989.
Tarney, J., and C. E. Jones, Trace element geochemistryof
orogenicigneousrocks and crustalgrowth models,or. Geol.
Soc. Lond., 151,855-868, 1994.
Thompson, R. N., and S. A. Gibson, Sub-continentalmantle
plumes, hotspots,and pre-existingthinspots,J. Geol. Soc.
Lond., 148, 973-977, 1993.
Tistl, M., K. P. Burgath, A. H6hndorf, H. Kreuzer, R. Mufioz,
andR. Salinas,Origin and emplacement
of Tertiaryultramafic
complexes in northwest Colombia: Evidence from
geochemistryand K-Ar, Sm-Nd and Rb-Sr isotopes,Earth
Planet. Sci. Lett., 126, 41-59, 1994.
yon Huene, R., and D. W. Scholl, The return of sialic material to
the mantle indicated by terrigenousmaterial subductedat
convergentmargins,Tectonophysics, 219, 163-175, 1993.
Wadge, G., and R. MacDonald, Cretaceoustholeiites of the
northerncontinentalmarginof SouthAmerica:The SansSouci
Formation of Trinidad, d. Geol. Soc. London, 142, 297-308,
1985.
Wadge, G., T. A. Jackson,M. C. Isaacs,and T. E. Smith, The
ophioliticBath-DunrobinFormation,Jamaica:Significancefor
Cretaceousplate evolution in the northwesternCaribbean,or.
Geol. Soc. London, 139, 321-333, 1982.
Weber, M., J. Tarney, and A. C. Kerr, Crustal and mantle
xenolithsfrom volcanicbreccias,SW Colombia:Implications
for the crustalstructureof the northwesternAndes,Eos Trans.
AGU, 76, p. F376, 1995.
Wells, R. E., Origin of the oceanic basementof the Solomon
Islandsarc and its relationshipto the Ontong Java Plateau-
Insightsfrom Cenozoicplate motion models,Tectonophysics,
165, 219-235, 1989.
Wieldmann, J., Ammonites from the CuracaoLava Formation,
Curacao,Caribbean,Geol. Mo'nbouw,57, 361-364, 1978.
White, R. S., Magmatismduring and after continentalbreak-up,
in Magrnatismand the Causesof ContinentalBreak-up,Spec.
Publ. 68, edited by B. Co Storey, T. Alabaster, and R. J.
Pankhurst,p. 1-16, The GeologicalSociety,London,1992.
Wildberg,H., Der Nicoya-Komplex,CostaRica, Zentralamerika:
Magmatismusund Genese eines polygenetischenOphiolith-
Komplexes, Munstersche Forschungen zur Geologie und
Palaontologie,62, 1-123,
 RajmahalBasalts,EasternIndia: Mantle Sourcesand
Melt Distributionat a VolcanicRifted Margin
W. Kent and A.D. Saunders

Departmentof Geology,Universityof Leicester,LeicesterLEI 7RH, UnitedKingdom

P. D. Kempton

NERC IsotopeGeosciences
Laboratory,KingsleyDunhamCentre,Keyworth,NottinghamNGI2 5GG,
tJ)•itedKingdom

N. C. Ghose

Departmentof Geology,Patna University,Patna 800 005 Bihar, India

Basaltsin the RaimahalHills representthe edgeof a >200-m-thick sequence

of predominantly tholeiiticlavaseruptedin the BengalBasin,easternIndia. The
40Ar/39Ar agesof Rajmahallavasanddikesclusterat 116-113 Ma. Theseages
indicate that volcanism in easternIndia occurredwithin a longer period of
igneousactivityin westAustralia(130-100 Ma) andwas contemporaneous with
the final stagesof basalticvolcanismat OceanDrilling ProgramSites 749 and
750 on the central Kerguelen Plateau. Pb-Nd-Sr isotoperatios of the least
crustallycontaminatedRajmahalbasaltsdiffer from thoseof most Kerguelen
Plateaulavasand appearto reflect an Indian mid-oceanridge basalt(MORB)-
like sourcewith 206Pb/204Pb•17.9, 207Pb/204Pb •15.5, •Nd(t)>+5, and
87Sr/86Sr(t)
<=0.7037.The isotopic
similarities
with IndianMORB are best
explainedif Rajmahalbasaltswere generatedby decompressional meltingof
asthenospherewellingup passivelybeneaththe rifled marginof easternIndia.
Upwellingcouldhavebeeninitiatedanddrivenby viscouscouplingof MORB-
sourcemantleto flow in the conduitof the Kerguelenplume,locatedabout 1000
km to the southof easternIndia at 116 Ma. The MORB-sourcemantle lay at the
edgeof thethermalhalodefiningtheplumeboundaryandmayhavebeenheated
by the plume.We inferthatthe Rajmahalbasaltsare examplesof lavaswhich,
thoughassociated spatiallyandtemporallywith the magmaticproductsof hot
spotactivity,werederivedfromcompositionally'normal'asthenosphere.

1. INTRODUCTION

LargeIgneousProvinces:Continental,Oceanic,andPlanetary The volume of melt generatedduringcontinentalbreak-

Flood Volcanism up is enhancedsignificantly where lithosphericrifting
GeophysicalMonograph100 occurs above unusually hot asthenosphere(100-200øC
Copyright1997by theAmericanGeophysical
Union hotter than ambient mantle), in the presenceof a mantle
146 RAJMAHAL BASALTS

plume [e.g., Foucher et al., 1982; White and McKenzie, locatedbeneaththe easternConrad Rise (-53.4øS; Figure
1989, 1995]. Differencesin the compositionand thickness 2). The origin of the 85øE Ridge north of 10øN was not
of igneouscrustacrossvolcanicrifted marginsmay reflect discussed, but Mailer et al. arguedthat the Rajmahallavas
variations in the rate and duration of lithospheric were not directly producedby the Kerguelenhot spot. In
extension, mantle temperature (distance of the plume their reconstructionfor 118 Ma, the Kerguelenplume was
centerfrom the developingrift system),mantle flow rate predictedto lie 1000+400 km to the southof easternIndia.
within the plume conduit,and the easewith which melt is A further challengeto the Kerguelenplume link has come
intrudedlaterallyin the crust[e.g.,Pedersonand Ro, 1992; fromAndersonet al. [ 1992], who proposedthat Cretaceous
Keen et al., 1994]. These variations can be modeled lavas in and around the eastern Indian Ocean are
quantitativelyonly in igneousprovinceswhere exposureis manifestationsof decompressional melting above a 'hot
excellent, the age and chemical compositionof rifted cell', or large region of unusuallyhot uppermantle. This
margin basaltsare known, and seismicand boreholedata hot cell is one of several low-velocity anomaliesin the
are available to estimate the timing of lithospheric convecting mantle inferred on the basis of seismic
extension,uplift and subsidence.Unfortunately,there are tomographic experiments. The cell was suggestedby
very few provinces where all of these criteria are met Andersonet al. to be much largerthan individualpresent-
satisfactorily.For example, tholeiitic lavas on the rifted day Indian Ocean plumes, which are 'embedded' in the
marginsof continentswhich once formedpartsof eastern low-velocityanomaly.
Gondwana (India, Australia) generally are well-exposed, In additionto disagreements over which mantle plume,
but are poorly sampledand thusnot fully understood.One if any, is responsiblefor Cretaceous igneousactivityon the
of the least-knownvolcanicrifted marginsis that of eastern easternIndian margin, there is controversyover the extent
India, where the Early CretaceousRajmahal basaltscrop of plume involvement in Rajmahal magmatism.In the
out on the western margin of the Bengal Basin. In this absenceof a clear 'hot spot' signaturein Rajmahalbasalts,
paper,we presenta comprehensive chemicaland Pb-Nd-Sr Mahoneyet al. [ 1983] proposedthat the Kerguelenplume
isotopicstudy of the Rajmahalbasaltsand their associated furnishedheat, but not material, to Rajmahalvolcanism.
dikes.Our resultsare usedto assessthe petrogenesis of the However, a decadelater, ODP (Ocean Drilling Program)
basalts, the nature of mantle sources contributing to Leg 120 drilled volcanic basement on the central
volcanism, and controls on the distribution of Cretaceous Kerguelen Plateau with Nd-Sr isotopic ratios and
igneousrocksalongthe easternIndian rifted margin. incompatibletrace element abundancesnot dissimilarto
Evidencefrom Cretaceousplate tectonicreconstructions thoseof the Rajmahal basalts[Salterset al., 1992; Storey
[e.g., Davies et al., 1989; Royer and Coffin, 1992] suggests et al., 1992]. Thesechemicaland isotopicsimilaritieswere
that the Rajmahal basalts form part of a large igneous dismissedrather summarilyby Curray and Munasinghe
province that includesrifted margin basaltsin southwest [1992], who favored a completelyseparateorigin for the
Australia (Bunbury, Naturaliste Plateau), and lavas two features. This is in part due to the difficulty of
forming the central and southern Kerguelen Plateau reconcilingthe chemicaldata(which suggestsomekind of
(Figure 1). On the basisof chemicalandPb-Nd-Sr isotopic link) with the apparentdistanceof easternIndia from the
data, this igneous province has been attributed to the center of the Kerguelen plume on Cretaceous plate
activity of the Kerguelen hot spot [e.g., Mahoney et al., reconstructions [e.g., Mailer et al., 1993] (note that these
1983, 1995; Storey et al., 1989, 1992; Kent, 1991; Salters reconstructionsassume an axisymmetric plume head
et al., 1992; Frey et al., 1996]. However,this link hasbeen locatedat-49øS). Given that plate tectonicreconstructions
challengedon the basisof alternativeplatereconstructions, by Davieset al. [1989] andRoyerand Coffin[1992] imply
suggestingthat the Rajmahal basaltshave a connection a plume in close proximity to the easternIndian margin,
with the offshore 85øE Ridge (Figure 2) [Curray and this difficulty is possiblyan artifact of the particularhot
Munasinghe, 1991]. In these reconstructions, the plume spotreferenceframe employedin the Mailer et al. [1993]
responsiblefor Rajmahaland 85øE Ridgevolcanismis the reconstruction.Regardlessof which plate reconstructions
Crozet plume, currentlylocatedbeneaththe Crozet Plateau are employed, the chemical-isotopicmeasurementsfrom
at-46.2øS. A Crozet plume sourcefor the 85øERidge was Rajmahal and Kerguelen Plateau basaltsrequire further
testedby M•iller et al. [1993] using a revisedmodel for explanation.Are the similaritiesmerely accidental,or were
global plate motions relative to hot spots.M•iller et al. bothsetsof basaltsproducedfrom the samemantlesource?
found no evidencefor Early Cretaceousmagmatismabove To answerthis question,we needto considerthe Rajmahal
the Crozet plume, but matched the southernpart of the basaltsin a local context (i.e., in terms of the Mesozoic
85øE Ridge (to 10øN) with a probablynow inactiveplume evolution of the eastern Indian margin) and a
 KENT ET AL. 147
70 80 90

70 80 90 ß527'
OD•

' .,--' u

ao

•isure 1. (a) Plate tectonicreconstruction

o[ pa• o[ easternOondwanaat 119 Ma [modified [rom Ro•er •d Co•,
1992] showJn•locationso[ Mesozoicbasaltsalon• the west Australianmar•Jn. Sdected DSDP and ODP sites are
shown by dots. Bathymetficcontours[or the Ker•uden Plateauare shown at 2000 m and 4000 m, and seafloor
magneticanomaliesby thin lines. The in[e•ed nowherepa• o[ the Indian continent('Greater India') is shown by
dashedlines. Abbreviations:• - East Antarctica,Aus - WesternAustralia,Bun - Bunbu• basalts,Mad - Madagascar.
(b) Plate tectonicreconstructiono[ pa• o[ easternGondwanaat 110 Ma [modified [rom Ro•er •d Co•, 1992]
showJn•locationso[ the Rahmahal-Sylheti•neous provinceand ODP Le• 119 and Le• 120 siteson the
Plateau(KP). The presumedcentero[ the Ker•uden plume [a•er M•ho•e• et •L, 1995] is shownby an opencircle.

context(in regard to their relationshipto other Cretaceous SanthaiParganasdistrictof easternBihar, some300 km to

igneousrocksin and aroundthe easternIndian Ocean). the northwestof Calcutta(Figure 3). In any given traverse,
up to ten basalt flows are encountered,dipping gently
2. GEOLOGICAL BACKGROUND
(_<3ø) towards the east-southeast.Field studies in the
2.1. TheRajrnahal-SylhetIgneousProvince Rajmahal Hills show that the exposedlocal thicknessof
the volcanicsuccessionnowhereexceeds230 m [e.g., Ball,
The Rajmahal Hills (approximately24015' to 25ø15'N, 1877; Kent, 1992; Kent et al., 1996]. Therefore,reportsof
87020' to 87ø45'E) extend over an area of-4100 km2 in the at least twenty-five basalt flows with a
148 RAJMAHAL BASALTS

70 80
6O 90

•o
e trail

• ...........
•'"•"••- 1 Kerguelen
Plume
• 2 Crozet
Plume
7o 8o 3 Conrad
Rise
Plume
•o .... • 90 4 MarionPlume

•o•/ / • ½fNintyeast
Ridge •0
/ 43Ma
• ••_• ..... •••Ridge • / ':-:'
'••.•...•

Figure 2. (a) Plate tectonic reconstructionof the easternIndian Ocean at 84 Ma [after Royer and Coffin, 1992]
illustratingpossibleplume tracksproposedby Curray and Mttnasinghe[1991] and Mailer et al. [1993]. Locationsof
hot spots(open circles)are inferredfrom the distributionof Cenozoicseamountand oceanislandvolcanism.The
proposedtrail linking the Raimahalbasaltsto the Crozetplume followsthe line of the 85øE Ridge. The trail of the
putativeConrad Rise plume follows the 85øE Ridge to -10øN. Bathymetriccontoursfor the KerguelenPlateauand
Broken Ridge are shownat intervalsof 1000 m (rangefrom 4000 to 1000 m). 'Af' denotesAfrica; otherabbreviations
as in Figure 1. (b) Plate reconstruction
of the easternIndian Oceanat 43 Ma [after Royer and Coffin, 1992] showing
the southernpart of the Kerguelenplumetrack (NinetyeastRidge,the northernKerguelenPlateau,and BrokenRidge)
andthe locationof the Crozet Plateau.Bathymetriccontoursfor the KerguelenPlateauand BrokenRidge as in Figure
2a. The ConradRise,southernMadagascar Rise(SMR), andDel CanoRise(DCR) are oceanicplateausthatmay have
formedabovethe Marion hot
 KENT ET AL. 149

I I I
86OE 88 ø

/' ShillongPlateau Shillono .._ ,.

Rajmahal Hills -220 m
..
Tura%--.•.a..
rø ,• Nhasi.Jaint•
Kuchma
Rajmahal 15m
Kansat
Sylh
Mica Belt
332 m

\\\dK Chotanagpur
Plateau
Koderma
ß Jalangi
o>216 m
24 ø --
Dhanbad Asansol
Bolpur •, r/ Ar.
eaof
Jhar/a
T >287m Debagram ,,, • ma•n
map
Raniganj
ß • Nadia
0 km 50 • Alluvium
Cretaceous Basalts
I I I

>131
m ©Burdwan
Bengal >184 m
•
Gondwanasediments
Archean-Proterozoic
N , Basin basement rocks
Ghatal
ß d•T Dike
(Cret.,
Tertiary)
--- Icutta "' ß Borehole
....i,:! ß City. town. village
,, ..... Port Canning
I i'T "; I

Figure 3. Map of the Rahmahal-Sylhetigneousprovince,showingprincipaloutcropsof Early Cretaceousbasalt,and

locationof explorationwells drilled in the westernBengal Basin. The minimum thicknessof the Rajmahal lavas is
indicatedfor eachdrill site [afterBiswas, 1959, 1963; Khan and Azad, 1963], whereasthicknessestimatesfor lavas in
the RajmahalHills and Sylhetregionare averagesbasedon field observations[Kent, 1992; Shukla, 1992; Kent et al.,
1996]. Basalticdikesare shownschematically.

thicknessof-600 m [e.g., Baksi et aL, 1987; Baksi, 1995]
are incorrect. At localities such as at Sahibganj in the
northernRajmahal Hills (Figure 4), the lavas exhibit bole
horizons (oxidised flow tops or tuffaceous material),
attestingto their eruption in a subaerial environment.
However, such features are comparativelyrare, possibly
dueto rapid eruptionof the flows.
Althoughbasalticlavasform >95% of the total exposure
in the RajmahalHills, small (24-28 km2) rhyoliticlava
flows occupypaleovalleyscut in the basaltsat Taljhari and
Berhait (see Figure 4 for locations)[e.g., Raja Rao and
Purushottam, 1963; Deshmukhet at., 1964]. Silicic tuffs
and bentonites are relatively abundant, occurring at
intervals throughoutthe lava succession.The tuffs and
bentonites appear to be particularly common in the
northwesternpart of the RajmahalHills, where individual
depositsare usually <1 m thick, but can locally reach
thicknessesof up to 14 m [e.g., $engupta, 1988]. These
depositsoften bear the remainsof plants,whose exquisite
preservationsuggeststhat they were depositedin shallow
ephemeral lakes on the surface of the basalt flows
[Sengupta,1988].
On the westernmargin of the RajmahalHills, the basalts
overlie unconformably sedimentary rocks of the
GondwanaSupergroup(Early Permian-Early Cretaceous)
(Figures 3 and 4). The Cretaceouslavas and Gondwana
sedimentsoccupy a small, broadly north-south-trending
sub-basinon the westernflank of the larger Bengal Basin
[e.g., Sengupta, 1966; Mukhopadhyayet at., 1986]. This
sub-basin is one of several Permo-Triassic basins
developedon the easternflank of the ChotanagpurPlateau,
a cratonised mobile belt consisting of >l.l-g.y.-old
metamorphicrocks (Figure 3) [Ghose, 1983; Mazumdar,
1988]. The basementrocks are faulted down to the east,
where hydrocarbon exploration wells and geophysical
surveyscarried out during the 1950s [e.g., Biswas, 1959,
1963; Khan and Azad, 1963; Sengupta,1966] indicatethat
the Rajmahalbasaltsunderliemuch of the westernBengal
Basin. The basaltshave a maximum apparentthicknessof
332 m in the Kansat borehole, West Bengal (Figure
150 RAJMAHAL BASALTS

I I I

Mirza
Chauki Sahibganj
N •'- • •]-i•'•Ambadih
a

'• Mahadeoganjl
I •-, Dudhkol

_ 25o00,N •
..."'"i
t ejamPahar
OGogra
Hill •
Taljhari _
./ Bara-Simra
Hill
•
f' :•Gandeswari

.:•• Berhait
'.. {?
) L.• ¾'"'•=•d•/• N Barha•a
,

_ 24o45 , _

0 10 20 Hiranpur
, , , , , QQ Pakur•
km Nunpahar Quarry
+

- •
• Alluvium
+ '••
Rajmahal
Basalts -
• Dubrajpur
Fm.
Gondwana
Barakar
Fm. Supergroup
:• Talchir
Fm.
• Chotanagpur
granite-gneiss
• Rhyolite
flow
Q Samplelocality +

87ø15'E 87o30' • 87045 '

I I I

Figure
4. Mapshowingthedistribution
ofvolcanic
andsedimentary
rocks
intheRajmahal
Hillsandsurrounding
areas,
andlocations
ofsampling
sites.
Outline
ofRajmahal
basalts
basedonBall[1877];
rhyolite
flows
based
onRaja
RaoandPurushottam
[1963]andDeshmukh
etal. [1964].

[Mukhopadhyay et aL, 1986]. To the east-northeast of the to 22 km thick.Nonetheless,

Currayand Munasinghe
RajmahalHills, the lavasthin to just 15 m at theKuchma [1992]ventured
to suggest
thatthelavasequence
pinches
borehole[KhanandAzad, 1963].The southern limit of the out over a gravity high at -22 ø19'N, 88ø39'E,
lavapile is notknownwith anydegreeof certainty, dueto corresponding
approximately
to the marginof the Indian
burialof the Rajmahalbasaltsbeneaththe hugeBengal shield.Thisis consistent
with seismic
reflection
profiles
Fan,a Cretaceous-Recent sequence of fluvialsediments up acrossthe westernBengal Basin, which show that
 KENT ET AL. 151

depthto the top of the basaltpile increasesfrom -•3.1 km bole horizons,generally0.1-0.4 m thick, but occasionally
below sea level at Ghatal to >4 km below sea level at Portup to 14 m thick [Shukla, 1992]. The Sylhet lavashave a
Canning(Figure 3) [Sengupta,1966]. gentle southerlydip (2-3ø), except where overthrustby
The Rajmahal basalts reappear at 25ø13'N, 91ø2I'E, Precambrianmetamorphicrocks of the Shillong Group
where lavasequivalentin age and chemicalcompositionto (northerlydip of up to 50ø immediatelybeneaththrust
rocksin the RajmahalHills occupya 60 x 4 km east-west- planes).Basalticdikes,rangingin thicknessfrom 0.3 to 7.0
trending strip to the southwestof Shillong, Meghalaya m, are abundantwithin the lava pile and show chilled
(Figure 3) [e.g., Talukdar and Murthy, 1970; Pantulu et marginsagainstthe lavas [Talukdar and Murthy, 1970].
al., 1992; Baksi, 1995]. Thesebasalts,known as the Sylhet Field evidencenoted by Talukdar and Murthy suggests
Traps, form part of an uplifted block on the northern that at leastsomeof the dikeswere intrudedalongan east-
margin of the Sylhet Trough, Bangladesh[Johnsonand west-trendingmonoclineformed by folding of the Sylhet
Alam, 1990]. Although not proven by drilling, they are flows.
almost certainly contiguouswith basalts in the western
Bengal Basin. On this basis,we infer that the Rahmahal- 2.2. Age of the Rajmahal Basalts
Sylhet igneousprovince spannedat least 5ø of longitude
(roughly 87ø to 92øE), equivalentto an area of about2 x Radiometricage data for the Rajmahalbasaltshavebeen
105km2. Continuation of the provinceeastof 89øEposes obtainedby severalworkersand are summarised in Figure
difficulties for Curray and Munasinghe's [ 1991] Crozet 5. Most recently,three partially alteredtholeiitesfrom the
hot spot hypothesis for the Rajmahal basalts. The RajmahalHills and an alkali basaltfrom the BengalBasin
hypothesisrests,in part, on an inferredlink betweenthe were analysedby Baksi [1995] using 4øAr/39Arstep-
RajmahalHills (-87øE) and the 85øE Ridge, a structureof heating methods (see also Dalrymple and Lanphere
unknown origin lying offshore of the rifted margin of
easternIndia (Figure 2). On the other hand, the Sylhet
lavas give credence to a hot spot track linking the
Rahmahal-Sylhetprovinceto the northernmost(>82-m.y.-
old) section of the Ninetyeast Ridge, a >5000-km-long
submarine volcanic edifice believed to have formed above 20

the 'tail' of the Kerguelenplume [e.g., Duncan, 1991]. As

I-I Conventional K-Ar
emphasisedby Curray and Munasinghe[ 1991, 1992], the 0-.9. 10 ß 4øAd39Ar Total Fusion
northernmost Ninetyeast Ridge shows an eastward
• 0
curvaturebeneaththe sedimentsof the Bengal Fan, such
that were it to continuenorthwards,its landfall would be in
the vicinity of easternBangladesh(Figure 3). Thus, the • -10
Sylhet lavas lie exactly in the location predicted by _

t"h -20
northward extrapolationof the Kerguelen plume track.
This could be pure coincidence.On balance,however,it
.• -30 -
_

ß [] E] _

seemsunwise to regard the 85øE Ridge as part of the _

Rajmahal-Sylhetigneousprovince until more is known - ,-- Whole-rock , • F'spar-•

-40
aboutthe age of this edifice and its structuralrelationship
to the Indian rifted margin. ooooo
Intriguingly,the cumulativethicknessof the Sylhetlavas
is significantlygreaterthan that of the volcanicsuccession Sample
exposedin the RajmahalHills. Betweenelevenand thirty-
five tholeiitic basalt flows with a total thickness of about Figure 5. Age discordance diagram, showing published
300 m are exposed in the West Khasi Hills, south of conventional
K-Arand40Ar/39Ar
total-fusion
agesforRajmahal
whole-rock samples and feldspar separates[McDougall and
Shillong (Figure 3) [see Pantulu et al., 1992; Shukla,
McElhinny, 1970; Baksiet al., 1987, Baksi, 1995] as a percentage
1992]. The tholeiitesare overlain locally by two flows of of the best estimateof the age of the basalts(116.2+0.6 Ma
alkali basalt and approximately150 m of rhyolitic lavas [Pringle et al., 1994]). Whole-rock apparentages are shown on
and tuffs [Talukdar, 1967]. A second sequence of the left and plagioclaseapparentageson the right. Note that ages
tholeiites, some 90-150 m thick, overlies the tuffs obtainedby conventionalK-Ar are usually lower than the best
[Talukdarand Murthy, 1970]. The basaltsare separated by ageestimate,sometimesby asmuchas
152 RAJMAHAL BASALTS

[1974]). Thesesamplesgave isochronagesrangingfrom 2.3. Dike Swarms in Eastern India

116.9ñ1.1 Ma to 110.9ñ1.6 Ma (1c5errors), but did not
yield concordantage plateaus.This lack of concordance K-Ar ages for whole-rock samplesof basaltic dikes
indicatesthat the agesobtainedare not true crystallisation croppingout 70-250 km to the west and southwestof the
ages;theyreflectlossof 4øArdueto alterationand/orloss Rajmahal Hills (Figure 3) range from 110-75 Ma
of 39Arfrom alterationminerals(e.g., clays)duringsample [Agrawal and Rama, 1976]. This age range prompted
irradiation.In contrast,
themean4øAr/39Ar plateauageof a Agrawal and Ramato suggesttemporalcontinuitybetween
plagioclaseseparatefrom the southwestern RajmahalHills Rajmahal and Deccan volcanism. However, this seems
(116.2ñ0.6 Ma, lo error)reportedby Pringle et al. [1994] mostunlikelyin the light of modeminterpretations of plate
fulfils rigorous statisticalcriteria for concordance[see motionsin the Indian Ocean region (see for example,
Pringle, 1993]. This age currently provides the best Royer and Sandwell [1989]). Moreover, conventionalK-
estimatefor the timing of Rajmahalvolcanism,andis used Ar whole-rockagesfor partiallyalteredRajmahallavasare
to calculateinitial Nd-Sr isotopicratiosfor basaltsin this often younger than the most reliable estimates of
study.A majorproblemwith the 4øAr/39Ar database for crystallisationage, sometimesby up to 30% (Figure 5).
Rajmahalbasaltsis that the relativestratigraphic positions This suggeststhat agesof much lessthan 110 Ma reflect
of the samplesanalysedarenot known. variable degrees of sample alteration and/or analytical
In additionto 4øAr/39Ar agesobtainedfor lavasfromthe problems.
RajmahalHills, a tholeiiticlava and basalticdike from the A lack of temporal continuitybetween Rajmahal and
KhasiHills (Sylhet)havebeendatedby K-Ar methods[see Deccan magmatism is supportedalso by preliminary
Baksi et al., 1987; Sarkar et al., 1996]. These samples resultsfrom 4øAr/39Ar step-heatingstudiesof threedikes
gave ages of 108ñ4 Ma and 110ñ4 Ma (lc5 errors), from eastern India (R. W. Kent and M. S. Pringle,
respectively. These ages appear to reflect loss of unpublisheddata). The resultsindicatetwo distinctage
radiogenic4øArduringsamplealterationand shouldbe populations, one of Aptian age (e.g., 4øAr/39Ar
isochron
taken as minimum estimatesof the true crystallisation age of whole-rock sampleRJ 1-13-2 = 113.4ñ1.0 Ma, 1•
ages. error) and anotherof Paleoceneage (e.g., 4øAr/39Ar
Furtherageconstraints
areavailablefrompaleomagnetic isochronage of the Salmadike = 64.4ñ0.3 Ma, 1• error).
data [McDougalland McElhinny,1970;Klootwijk,1971; The formerpopulationof dikesoccursover a wide region
Poornachandra Rao et al., 1993, 1996; Sherwood and to the west and southwestof the Rajmahal Hills, whereas
BasuMallik, 1996], which showthe majorityof samples the latter appearto be confinedlargely to the Jhariaand
from the RajmahalHills and SylhetTrapsto be normally Raniganj coalfieldsof the Damodar Valley (Table 1 and
magnetized.However,basaltstowardsthe top of the lava Figure 3). A reconnaissance survey of the Chotanagpur
pile (exactpositionnot documented) at threesitesin the Plateausuggeststhat dikes of Aptian age are quite scarce
western Rajmahal Hills show transitional to reversed or elseare poorlyexposed;whereexposed,.
they occuras
magneticpolarity [Klootwijk, 1971; Sherwoodand Basu short en 6chelon topographicridges up to 8 m wide.
Mallik, 1996]. This observationis surprising,giventhatthe Paleocenedikes, distinguishedfrom Aptian dikes by a
radiometricagesobtainedfor Rajmahalbasaltssuggest that greaterabundanceof brown-weatheringFe-Ti oxides,are
they were eruptedduringthe Cretaceous Normal Polarity even lessnumerousthan the Cretaceousdikes, but are up
Superchron (118-84 Ma). In an attemptto reconcilethe to 40 m wide and extend for several tens of kilometers.

paleomagnetic and 4øAr/39Arresults,Kent et al. [1992b] In addition to basaltic dikes, subhorizontalcylindrical

suggested that the polarityreversalrecordedby Rajmahal sills of lamproite are exposedin the Jharia and Raniganj
lavas is the so-called 'ISEA' interval. The ISEA interval, coalfields[e.g., Middlemostet al., 1988; Kent et al., 1992a;
equivalentto chronM-lr of Gradsteinet al. [1994], is a Rock et al., 1992]. We undertooka detailed chemical-
brief magnetic field reversal dated by high-resolution isotopicstudyof the lamproitesin orderto determinetheir
ammoniteand foraminiferalstratigraphyat 115.0ñ0.3 Ma, suitability as contaminantsto the Rajmahal basalts.The
i.e., Upper Aptian [e.g., Tarduno,1990; Gradsteinet al., full results of our lamproite study will be reported
1994]. If our interpretationof the polarity reversalis elsewhere,but we include relevant isotopic data in the
correct,it providesthe bestminimumestimatefor the age discussion and in certain of the illustrations. In order to
of volcanismin the Rajmahal Hills. We note also that determine the age of the lamproites, we carried out
paleomagneticinclinationdata obtainedby Klootwij'k 4øAr/39Arlaser-ablationstudiesof phlogopiteseparates
[1971] suggestthat the Rajmahalbasaltswere eruptedat from two samplescollectedin the Jhariacoalfield(R. W.
-47øS, some2ø northof the presumedpresent-day latitude Kent and S. P. Kelley, unpublisheddata). These samples
of the Kerguelenhot spot[e.g.,Miiller et al., 1993]. gaveweightedmean4øAr/39Ar
agesof 116.6ñ0.8Ma
 KENT ET AL. 153

TABLE 1. SampleLocalitiesandPetrographicDetails

SampleNumber Locality Information SpecimenDescription

Early CretaceousDikes: Chotanagpur

PlateauandDamodarValley (DV), Bihar andWest Bengal
RJ 1-12-1 Quarry section,4 km eastof Domchanch,Koderma(Mica Belt) Cpx & plag-phyricmg dolerite
RJ 1-13-1 Meghatariquarry,4 km northwestof Koderma(Mica Belt)
RJ 1-13-2
RJ 1-13-3 Baurhi Kalan, NH31 Highway, 22 km northwestof Koderma(Mica Belt)
RJ 1-19-1 1 km west of Kalidaspurcolliery, RaniganjBasin,easternDV Cpx-phyricmg dolerite
RJ 1-19-2
RJ 1-20-3 2 km northeastof Asansol,RaniganjBasin,easternDV Cpx-phyricmg dolerite
RJ 1-20-4

Early TertiaryDikes:DamodarValley, Bihar andWestBengal

MaheshpurI Mohudavillage, 15 km westof Dhanbad,JhariaBasin,easternDV Aphyric mg-cgferrodolerite
MaheshpurII "
Salma 2 km northeastof Asansol,RaniganjBasin,easternDV Aphyric cg ferrodolerite

Early CretaceousBasalticLavas:RajmahalHills, SanthaiParganas,Bihar*

RJ 1-22-1 Maliparavillage, 2.5 km southwestof Durio Aphyric mg basalt
RJ 1-23-1 West flank, Gandeswari Hill, 3.5 km southwestof Lalmatia Plag-phyricfg-mg basalt
RJ 1-23-3 " Plag-phyricmg-cgbasalt
RJ 1-23-4 " Aphyric cg basalt
RJ 1-23-5 " Plag-phyricfg-mg basalt
RJ 1-23-7 " Plag-phyricmg basalt
RJ 1-25-1 North flank of Bara-Simra Hill, 0.5 km southwestof Lalmatia Plag-phyricfg basalt
RJ 1-26-2 Northwestflank of SaharpurHill (Bejam Pahar),2.6 km northof Lalmatia Plag-phyricmg basalt
RJ 1-26-7 " Plag-phyricfg basalt
RJ 1-26-9 "
RJ 1-27-5 Northeastflank of GograHill, 15 km east-northeast
of Lalmatia Plag-phyricmg basalt
RJ 1-27-6 "
RJ 1-30-1 Hiranpur, 17.5 km northwestof Pakur Aphyric fg basalt
RJ 1-30-2 Quarry, 6 km west of Pakur
RJ 1-30-3 Dhanbadvillage, 9 km westof Pakur
RJ 1-30-4 "
RJ 1-30-5 "
RJ 1-30-6 " Cpx-phyricfg basalt
RJ 1-31-1 Chandalaquarry,4 km northwestof Barharwa Aphyric fg basalt
RJ 1-31-3 "
RJ 1-31-4 " Aphyric fg trachyandesite
RJ 1-31-4A "
RJ 1-31-6 " Aphyric fg basalt
RJ 1-31-7 "
RJ 2-1-5 Nunpahar,1.8 km northeastof Ambajoravillage Piag-phyric cg bas•Llt
RJ 2-1-6 " Plag-phyric mg basalt
RJ 2-2-5 Dhudkol,2.5 km north-northwest
of Taljhari Plag-phyricmg basalt
RJ 2-2-6 "
RJ 2-5-1 Westernflank of ridge,500 m northeastof Sitalpur Plag-phyricfg basalt
RJ 2-5-2 "
RJ 2-5-3 " Plag-phyricmg-cg
154 RAJMAHAL BASALTS

TABLE 1. (continued)

SampleNumber Locality Information SpecimenDescription

EarlyCretaceous
BasalticLavas:RajmahalHills, SanthalParganas,
Bihar*

RJ 2-6-1 Northernflank of ridge, 1.5 km southof Sahibganj Plag-phyricfg basalt

RJ 2-6-2 " Plag-phyricmg basalt
RJ 2-6-3 " Aphyric mg basalt
RJ 2-6-4 " Aphyricfg basalt
RJ 2-6-5 "
RJ 2-6-6 "
RJ 2-6-7 "
RJ 2-7-1 3 km southof Mirza Chaukirailway station,15 km westof Sahibganj Aphyric fg basalt
RJ 2-7-2 "
RJ 2-7-3 " Cpx-phyricfg-mg basalt
RJ 2-7-4 Aphyricfg basalt
RJ 2-7-4A "
RJ 2-7-5 "
RJ 2-7-7A "
RJ 2-7-7B "
RJ 2-8-1 Quarry,1.5 km westof roadat Ambadiha,10 km eastof Sahibganj
RJ 2-8-2 " Plag-phyricfg basalt
RJ 2-8-3 "
RJ 2-8-4 "
RJ 2-8-5 " Aphyric fg basalt
RJ 2-9-1 0.5 km southof Mahadeoganj,5 km westof Sahibganj Aphyric mg basalt
RJ 2-9-2 " Aphyric fg basalt
RJ 2-9-3
RJ 2-9-4 "

Rhyolites:RajmahalHills, SanthalParganas,Bihar*
RJ 2-1-3 Nunpahar,1.8 km northeastof Ambajoravillage Aphyric vfg rhyolitic tuff
RJ 2-2-2 Dhudkol, 2.5 km north-northwestof Taljhari Vesicularfg rhyolite

Notes:Cpx = clinopyroxene,plag= plagioclase.(v)fg, mg, cg = (very) fine-, medium-,andcoarse-grained. *For localities

in the RajmahalHills, sampleswith the lowestsuffix(e.g.,RJ 2-9-1) correspond to the first exposedlava flow in the section.

109.1+0.7 Ma (lo errors). However, resistancefurnace 3. CRETACEOUS VOLCANISM IN AND AROUND

heatingof phlogopiteseparatedfrom the first of our two
Jharialamproites(reportedby Pringle et al. [1994]) gavea
concordant
4øAr/39Ar
isochronageof 113.5+0.5Ma (lo).
Therefore,the lamproitescould be the sameage as Aptian
basalticdikes from the Damodar Valley and Chotanagpur
Plateau,and slightly youngerthan lavas in the Rajmahal
Hills• If so, the laserablationanalysescouldreflectmixing
of once-distinctargoncompositions with differentapparent
ages,suchthat the agesare precisebut inaccurate.Work is
in progressto clarify the geologicalmeaningof thesedata.
Meanwhile, the singleresultobtainedby resistancefurnace
heatingprovidesthe bestestimateof the true crystallisation
ageof the Indian lamproites.
THE EASTERN INDIAN OCEAN
Early Cretaceousigneous activity along the west
Australianmargin and in easternIndia was precededby
differentialuplift [Kent, 1991] and threemajorphasesof
extensionalor transtensional faulting,culminatingin final
breakupof the two continents at-134 Ma [e.g.,Powellet
al.• 1988; Marshall and Lee, 1989]. Seafloor spreading
betweenIndia and Australiabeganat about 155 Ma, with
the formationof oceaniccrustin the Argo Basinoff north-
westAustralia(Figure 1) [Ogg et al., 1992]. Over a period
of at least25 m.y., and perhapsas much as 55 m.y., thick
lava sequences were eruptedalongthe marginof
 KENT ET AL. 155

Australia (Figure 1). These lava sequencesappearto be Ma of subaerialsilicic volcaniccentersalongthe northwest

youngestin the south [e.g., Pyle et al., 1995], suggesting Australian margin [Thurow and yon Rad, 1992]. If
that the main rift zone between eastern India and western volcanic activity occurred a considerable time after
Australia propagatedfrom north to south.It is not known breakup, we must appeal to a processother than rifted
whether equivalent lava sequenceswere erupted on the margin uplift and convection-enhanced melting to explain
easternIndian margin; in northeastIndia and Myanmar the origin of the basalts.Identificationof the process(es)
(Burma), the evidence is obscuredby magmatism and involved, and the mantle source of these rocks, awaits
sedimentation related to the India-Asia collision [see chemicaland isotopicinvestigations and4øAr/39Ar dating
Gibling et al., 1994]. of core samplesobtainedfrom the Australianmargin.
A key questionconcernsthe geodynamicsignificanceof Beginningshortlybefore 118 Ma (oldestbasementK-Ar
the west Australian margin lavas. Were these rocks age; Leclaire et al. [1987]), a huge oceanic plateau was
generatedabove a mantle plume(s)or do they result from constructedon juvenile oceanic crust that had formed
some aspect of continental breakup, for example, betweenIndia and Australia-Antarctica.The origin of this
convection-inducedmelting associatedwith uplift along edifice, the Kerguelen Plateau, has proved to be
the rifted margin [e.g., Buck, 1986].9Chemicaland isotopic perplexing. The northernmost portion underlying the
resultsfrom Early Cretaceous(130-100 Ma) Bunburyand KerguelenArchipelagois likely to have been producedin
NaturalistePlateaulavas,emplacedon the southwestpart two or more stagesby the Kerguelenplume. By analogy
of the margin, suggestsimilaritiesbetweenat least some with the central portion of the plateau (see below), the
Australian margin lavas and recent volcanic productsof oldestvolcanic basementon the northernsectionmay date
the Kerguelenhot spot(e.g., in termsof Lamb, Th/Ta, and from about 118 Ma. Igneouscrustprobablywas addedat
present-dayN d-Sr isotope ratios) [Storey et al., 1992; 43-39 Ma during interactionof the Kerguelenplume 'tail'
Mahoney et al., 1995; Frey et al., 1996]. However, these with the SoutheastIndian Ridge (Figure 2b) [e.g., Royer
similaritiesdo not require the Australian lavas to have a and Sandwell, 1989; Charvis et al., 1995]. The central and
mantle plume origin. A major obstacleto interpretationis southern portionsof the KerguelenPlateauhave4øAr/39Ar
crustal contamination, which has modified the Pb-Nd-Sr ages in the range 114-85 Ma [Whitechurchet al., 1992;
isotopiccompositions of thesebasalts.It is noteworthythat Pringle et al., 1994]. Thesepartsof the plateauhave been
the least crustally contaminatedBunbury and Naturaliste attributedto the plume 'head' stage of volcanism [e.g.,
Plateau lavas have initial 87Sr/86Sr ratios of-0.7039 Davies et al., 1989; Storey et al., 1989, 1992; Weis et al.,
[Mahoneyet al., 1995; Frey et al., 1996], i.e., lower (by at 1989].
least 0.0004) than those of basalts erupted on the Basaltsfrom the centralportionof the KerguelenPlateau
Kerguelen Archipelago, the most recent product of the (ODP Sites 749 and 750) have major and trace element
Kerguelen hot spot. This suggestsa contribution to compositionsconsistentwith generationat the intersection
southwestAustralian margin volcanism from a source of a plume with a spreadingridge [Salters et al., 1992;
otherthan the Kerguelenplume, most probablythe Indian Storeyet al., 1992]. In contrast,basaltsfrom ODP Site 738
mid-oceanridge basaltsource(87Sr/86Sr -0.7025-0.7048 on the southernpart of the plateau have compositions
[e.g., Mahoneyet al., 1992, andreferencestherein]). suggestingeruption through thinned continental crust
Too few chemicaldataexistto evaluatethe possibilityof [Alibert, 1991; Mahoney et al., 1995]. Although not
a hot spot influence on magmatismfarther to the north. proven,the possibilityof blocksof continentallithosphere
However, the small amount of radiometric age data for beneaththe southernand centralKerguelenPlateaushould
lavas on the northwest Australian margin provide some be taken seriously.Velocitiesof 6.6-7.1 km s-• in the
insight into the magmatic developmentof this region. lower crustbelow the RaggattBasin, southeastof Site 750,
Basaltson the ScottPlateau,the northernmostsequenceof allude to the presenceof transitionalcrustcomparablein
lavas on the west Australian margin, have K-Ar ages velocity structureto the stretchedcontinentalcrust of the
essentiallythe same as those of the oldestBunbury lavas Rockall Basin, eastern North Atlantic [Operto and
(132-128 Ma [yon Rad and Exon, 1983]). If interpreted Charvis, 1995, 1996]. The velocity data were interpreted
literally, these ages suggestthat volcanismon the Scott by Operto and Charvis [1995] as evidencefor underplated
Plateauoccurredabout 25 m.y. after continentalbreakup basalt and/or extensional shear zones akin to those
(recall that oceanic lavas in the Argo Basin have recordedby seismicstudieson the margins of the eastern
radiometricages of about 155 Ma). Additional evidence North Atlantic [e.g., White and McKenzie, 1989; Reston,
for post-breakup igneous activity comes from Early 1993]oBy analogywith Hatton Bank in the North Atlantic,
CretaceousbentonitesrecoveredduringODP Legs 122 and it is possiblethat failed attemptsat seafloorspreadingprior
123, the distributionof which impliesthe existenceat-130 to continentalbreakupallowed highly intrudedcrust
156 RAIMAHAL BASALTS

the rifled marginsof India and East Antarcticato become
detached and stranded in the oceanic realm. Although
speculativeat present,we note that a Hatton Bank-type
origin for portions of the central Kerguelen Plateau is
capableof explainingthe extraordinarilylargerangein Pb-
Nd-Sr isotoperatiosshownby KerguelenPlateautholeiites
(see Section4).
The youngestlavas from the central KerguelenPlateau
(85 Ma; ODP Site 747) are of an age similar to-88 Ma
basaltsdredgedfrom BrokenRidge (seeFigure2), a large
(1000 x 100-200 km) submarinevolcanicedificecurrently
lying some 1800 km to the north of the KerguelenPlateau
[Duncan, 1991]. Broken Ridge and Site 747 lavas
representthe f'mal stagesof KerguelenPlateauvolcanism,
the cessation of which was marked by a phase of
acceleratedseafloorspreadingin the easternIndian Ocean
and constructionof the NinetyeastRidge betweenabout82
Ma and 38 Ma (Figure 2b) [Duncan, 1991]. In common
with lavasfrom the centralKerguelenPlateau,mostbasalts
from DSDP (Deep Sea Drilling Project)and ODP siteson
the Ninetyeast Ridge have major and trace element
compositionssuggestiveof moderateto high degreesof
partial melting (10-30%) beneath juvenile oceanic
lithosphere[Frey et al., 1991; Saunderset al., 1991; Kent
and McKenzie, 1994]. Plate tectonicreconstructionsby, for
example, Royer and Sandwell [1989] suggestthat the
NinetyeastRidge formed closeto a spreadingaxis. Thus,
from the Early Cretaceousto the Mid-Tertiary, a periodof
some 75 m.y., the eastern Indian Ocean ridge system
appearsto havemaintaineda positioncloseto the centerof
the Kerguelenplume. By analogy with Iceland, this was
probably accomplishedvia a series of riff jumps or
propagationepisodesasthe Indian platedriftednorthwards
over the hot spot.
4. GEOCHEMISTRY
4.1. Samplesand Analytical Methods
crushing. Small (2-3 mm) chips of each sample were
hand-picked and ground in an agate barrel. Agate can
contain small amountsof galena, requiring a test to be
performed in order to estimate the amount of Pb
contamination of basalts crushed in the agate barrel.
Isotopic and isotope-dilutiondata obtainedfrom a set of
agate-crushedRajmahal basalt powders were compared
with measurementsof splitsof the samepowderscrushed
in a tungsten carbide mill. Elemental abundancesand
isotopicratios of Pb from the two splits were identical
within the limits of analytical error (M. Storey, pers.
comm.), suggestingthat Pb isotoperatios are not affected
significantlyby the crushingprocess.
Whole-rockpowderswere analysedfor major and trace
elements by X ray fluorescence(XRF) spectrometry.
Analyseswere carried out at the University of Leicester
using two XRF spectrometers:an ARL 8420+ and a
Philips PW1400. Details of XRF operatingconditionsand
samplepreparationproceduresare givenby Saunderset al.
[ 1991] and Storey et al. [ 1992]. Our results, along with
detailsof precisionand accuracyof the data,are presented
in Table 2. Following irradiation at the University of
London Reactor Centre, rare-earth element abundances
and concentrations of Th, Ta, Hf, and Sc for nineteen
samples were determined by instrumental neutron
activation analysis(NAA) at the University of Leicester.
Samplepreparationand analyticaltechniquesare similarto
those describedby Fitton et al. [1997]. The NAA results
appear in Table 3, together with estimatesof analytical
uncertainties. Ratios of Nb/Ta and Zr/Hf can be used to
check for consistencybetweenNAA and XRF data. The
Nb/Ta and Zr/Hf ratios of C1 chondrite are 17.6 and 36.3,
respectively [e.g., Sun and McDonough, 1989]. Of the
nineteensamplesanalysedby NAA andXRF, only two (RJ
1-20-3 and 1-26-7) have Nb/Ta <13. The non-chondritic
Nb/Ta ratios of thesesamplesmost probablyreflect errors
in the measurementof Nb. Zr/Hf ratios in our samples
range from 35-41, suggesting that the Zr and Hf
abundance data also are reliable.

The basalt and rhyolite samplesanalysedin this study Isotopic ratios of Nd and Sr were measuredfor fifteen
were collected from localities listed in Table 1 and shownRajmahal samplesand a Paleocenedike (Salma) at the
in Figures3 and 4. Care was taken to avoid lava flow tops NERC Isotope Geosciences Laboratory (NIGL),
containinganalcite, stilbite, chabazite,laumontite,agate, Keyworth. Lead isotopic data were obtained at NIGL
andchalcedony.Samplesrangefrom the very fresh(all our (thirteenRajmahalsamples)and at the IsotopeLaboratory,
Rajmahal basaltshave weight-loss-on-ignitionvalues of University of Hawaii (two samples).Analytical methodsat
<1%) to the moderatelyaltered (trachyandesites, weight NIGL were describedby Kempton and Hunter [1997];
losson ignitionof 3.6-3.7%). Specimenswere selectedfor those used in Hawaii were describedby Mahoney and
chemical analysison the basis of thin sectionslacking Spencer[1991]. Sr and Pb were run as the metal on single
zeolitesand obvioussignsof alteration.Chalcedonyand Ta and singleRe filaments,respectively,using a Finnigan
zeolitesare particularlyabundantin the rhyolite samples, MAT 262 multicollectormassspectrometer(NIGL) and a
and could not be removed in their entirety prior to VG Sectormulticollectormassspectrometer(University
KENT ET AL.
158 RAJMAHAL
KENT ET AL.
160 RAJMAHAL
 KENT ET AL. 161

TABLE 3. NeutronActivationAnalysisDatatbr RajmahalBasaltsandEarlyTertiaryDikes

Sample
RJ- La Ce Nd Sm Eu Gd Tb Yb Lu Ta Th Hf Sc

GroupI Dikes
1-19-1 7.7 18.0 12.6 3.89 1.52 5.0 0.84 2.63 0.41 0.37 1.12 3.03 34
1-20-3 7.8 17.5 12.9 3.91 1.54 4.8 0.83 2.59 0.39 0.35 1.10 2.81 36

Group I Lavas
1-26-7 7.0 18.1 14.0 4.30 1.61 5.4 0.90 2.91 0.45 0.37 0.70 3.03 39
1-27-6 7.2 18.1 13.4 4.23 1.62 5.3 0.92 2.85 0.44 0.34 0.87 3.08 39
1-30-5 6.9 16.5 13.1 3.91 1.51 4.9 0.86 2.85 0.43 0.33 0.81 2.87 42

GroupII Dikes
1-12-1 12.0 27.3 18.7 4.78 1.79 5.3 0.91 2.56 0.38 0.45 1.27 3.79 32
1-13-1 11.5 25.8 17.5 4.64 1.71 5.1 0.82 2.43 0.36 0.42 1.18 3.57 32

GroupII Lavas
1-31-4 35.0 54.8 34.8 7.76 2.53 9.5 1.42 4.40 0.67 0.63 4.02 4.78 34
2-7-1 16.8 37.7 22.1 5.38 1.84 5.7 1.02 2.99 0.44 0.54 3.40 4.33 33
2-7-2 16.6 38.1 21.8 5.41 1.83 5.7 1.02 3.06 0.46 0.53 3.22 4.31 32
2-7-3 12.0 26.1 15.5 4.05 1.50 4.1 0.79 2.31 0.34 0.37 1.97 3.07 30
2-7-4 10.8 23.4 14.8 3.70 1.41 4.1 0.73 2.01 0.30 0.38 1.38 2.86 30
2-7-4A 18.2 39.8 23.1 5.58 1.85 5.9 1.06 3.14 0.45 0.55 3.49 4.35 30
2-7-5 12.7 22.3 17.1 4.42 1.60 5.0 0.87 2.42 0.36 0.41 1.86 3.42 33
2-7-7A 10.1 23.6 14.5 3.63 1.42 3.9 0.71 1.98 0.30 0.37 1.39 2.82 28
2-7-7B 18.1 40.1 23.4 5.52 1.87 5.6 1.06 3.12 0.45 0.56 3.57 4.41 33
2-9-3 17.8 37.6 22.8 5.32 1.77 5.7 1.00 2.95 0.47 0.51 3.08 4.21 33

Early Tertiary Dikes

Mah I 14.9 36.7 25.2 6.40 2.14 7.2 1.34 3.73 0.58 0.88 1.99 5.06 43
Salma 15.2 37.6 25.6 6.53 2.18 7.2 1.38 3.72 0.58 0.89 2.02 5.13 43

Standards
BOB-1 Meas. 4.7 12.4 10.0 3.00 1.20 3.8 0.73 2.52 0.40 0.41 0.40 2.41 33
BOB-1 Rec. 4.7 13.8 10.7 3.32 1.25 4.2 0.74 2.63 0.44 0.51 0.45 2.53
BCR-1 Meas. 24.7 53.2 28.9 6.68 2.09 6.7 1.02 3.49 0.50 0.78 6.31 5.22 33
BCR-1 Rec. 24.9 53.7 28.8 6.59 1.95 6.7 1.05 3.38 0.51 0.81 5.98 4.95 33
JB-la Meas. 36.9 65.8 26.7 5.14 1.54 4.6 0.69 2.14 0.33 1.62 9.64 3.70 29
JB-la Rec. 38.1 66.1 25.5 5.07 1.47 4.5 0.69 2.10 0.32 2.00 8.80 3.48 29

Notes: All abundancesare in ppm. Precisionwas determinedfrom eight replicateanalysesof oceanfloor basalt
standardBOB-l, exceptfor Sc,whichwasdetermined from eightreplicateanalysesof theUSGS basaltstandardBCR-1.
Precisionfor eachelement,asestimatedby onestandarddeviation(in ppm)from the mean,is: La = 0.12, Ce = 0.32, Nd =
0.4, Sm = 0.12, Eu = 0.03, Gd = 0.3, Tb = 0.05, Yb = 0.08, Lu = 0.01, Ta - 0.02, Th = 0.02, Hf = 0.07, and Sc = 1.0.
Accuracycan be assessed by comparingrecommended valuesfor BOB-1 and JB-la [see Saunderset al., 1991;
Govindaraju,1994] with measuredvalues(meanof seventeenand eighteenanalyses,respectively).The standard
deviationsof the JB-1a analysesare(in ppm):La = 1.03,Ce = 2.01, Nd = 1.0, Sm= 0.13, Eu = 0.05, Gd = 0.1, Tb = 0.06,
Yb = 0.07, Lu = 0.03, Ta = 0.06, Th = 0.30, Hf = 0.12, and Sc =
162 RAJMAHAL BASALTS

i i i i i i i i i
KENT ET AL.
164 RAJMAHAL BASALTS

Hawaii).Bothspectrometers in staticmode. Hills studiedby previousworkers[e.g., Mahoneyet al.,

wereoperated
Nd was run as the metal on a double Re-Ta filament 1983;Baksiet al., 1987; Sarkar et al., 1989;Storeyet al.,
assemblyusing a VG 354 massspectrometer (NIGL) 1992; Baksi, 1995]. Elemental abundancesof Early
operatingin dynamicmode.Isotopicratiosand isotope- Cretaceous dikes from the ChotanagpurPlateau and
dilution abundancemeasurementsfor our samples are easternDamodarValley, analysedhere for the first time,
givenin Table4, togetherwith datafor standards and are very similar to those of basalticlava flows in the
procedural blanks, and estimates of analytical Raimahal Hills. This similarity is consistentwith the
uncertainties. radiometricage data discussedaboveand suggests that the
dikes and lava flows share a commonpetrogenesis.Our
4.2. LeachingExperiment analyses, togetherwith thoseof previousworkers,indicate
a wide rangein major elementabundances andratios(e.g.,
In orderto testwhetherthehighmeasured 87Sr/g6Sr
ratio
of Cretaceous dike RJ 1-19-1 is an effect of subaerial
Figure 6). The Raimahal basaltshave SiO2 and K20
contents(49.4-55.9 wt% and0.11-1.10 wt%, respectively)
alteration,an aliquantof this samplewas subjectedto a
that generally are higher than those of normal MORBs
multistepleaching procedure.About 150-200 mg of
(mid-oceanridge basalts)eruptedat present-dayspreading
powderwas placedin a sealedbeakeron a hotplateat
centersin the Indian Ocean (mostly <50 wt% SiO2 and
-120øC for 1 h in 1 ml 6M HC1 and 2 ml ultrapureH20.
<0.2 wt% K20; see, for example,Dosso et al. [1988]).
The liquid, includingsuspended particlesof the sample,
However, the lower bound on Rb abundances(range from
was then removedby pipette and the beakerreplenished
1 to 44 ppm) lies well within the range exhibited by
with fresh HC1. The sealed beaker was placed in an
normal MORBs (mean -0.6 ppm [e.g., Sun and
ultrasonicbath.This cyclewasrepeatedseveraltimesuntil
McDonough,1989]). The upperlimits of 'immobile' trace
the liquidremainedcolorless uponadditionof new HC1.
elementconcentrations,suchas thoseof Zr (72-186 ppm),
The final leachatewasremoved,the samplerinsedin ultra-
Ta (0.33-0.56 ppm), and Nb (2.1-10.5 ppm) exceedthose
pureH20 and dried.
Table 4 shows that the measured 878r/g6Srratio of the of averagenormal MORB (approximately74, 0.13 and 2.3
ppm,respectively[e.g., Sun and McDonough,1989]). The
leachedaliquantis only slightlylower than that of the
lower boundsfor Ba (61-347 ppm) and Sr (222-391 ppm)
unleached sample.This suggests that the high 878r/S6Sr
lie within the range of values for 'enriched' MORBs,
ratio of RJ 1-19-1 is not an effect of alteration,in keeping
which have mean values of 57 and 155 ppm, respectively
with the low weight losson ignition(0.57%; Table 2) and
[e.g., Sun and McDonough, 1989]. However, it is likely
lack of evidencefor alterationindicatedby petrographic
that abundancesof Ba (and other large-ion lithophile
studies. On this basis, and petrographicevidence, we
elements) in the Raimahal basalts reflect, in part, the
interprethigh relativevaluesof 878r/S6Sr in someother
mobility of these elements in low-temperaturefluids
Rajmahalbasaltsto reflectmagmaticratios.
circulating through the lava pile. For example, BaYRb
4.3. Results ratios in our Raimahal samplesrange from 4 to 70, in
contrastto near-constantBa/Rb ratios of 11-12 [Hofmann
All Early Cretaceous basalticrocksfrom our collection and White, 1983] in pristine oceanicbasalts.Magmatic
are quartz-normativetholeiites,with the exceptionof abundances of these elements could have been modified

sampleRJ 2-5-3, an olivine tholeiite(Tables 1 and 2). also during contaminationof the Raimahal basalts by
Within the RajmahalHills, it is possiblethat we sampled continental crust (seebelow).
the samelava flow on more than one occasion(recall that a The Raimahal basaltswere divided by Storey et al.
maximum of ten basalt flows are present at any one [1992] into two chemical groups(I and II) on the basisof
locality).However,we believethis is unlikely,giventhat different Ti/Zr and Zr/Y ratios for a given value of MgO.
mostflows appearto extendalongstrikeover distances of Figure 7 shows that these groups are apparentalso in our
only a few kilometers.Due to the small areal extentof data set. Raimahal Group I basalts (including samples
individualflows, it hasnot provedpossibleto correlatethe analysed by previous workers) have Ti/Zr ranging from
lavas between each of the various stratigraphicsections 82-120 and a mean Zr/Y ratio of 3.2+0.7, whereas Group
examined. II basalts have Ti/Zr ratios of 45-78 and a mean Zr/Y ratio
of 4.6+0.7. A relatively evolvedRaimahallava (RM82-10)
Major and TraceElements.Cretaceous tholeiitesamples analysedby Mahoney et al. [1983] appearsto share
analysedin this study (Tables 2 and 3) have chemical featuresof both groups(e.g., Ti/Zr = 56, Zr/Y = 3.1), asdo
compositions similar to basaltflows from the Rajmahal two "Group II" lavas from the Bengal Basin with
 KENT ET AL. 165

60
RM82-10
ß Group I
RB88/34
0_.,9
' 56 o Group II 3?. 12
(• 0
• 52
o
o • o
48 o
o
LL8
1.4 3.0

0ø8
---ø-.
o 1.0

O 0.6

0.2

0.8

oo

0¸
0.6

RM82-10 R B88/34
0.4 ,, ,, I, , ,, I, ., ,•, ,, ,'•i i, il i i I I I I i i i •, ,, I, , ,, I, ,, , I, ,, ,I , ,, ,I ,, , ,I ,i I i
4.0 6.0 8.0 10.0 .0 6.0 8.0 10.0
MgO (wt.%)
Figure6. Majorelement variations
fortheRajmahalbasalts,
versusMgOcontent.TheRajmahal basaltsaredivided
intotwogroups(I andII) onthebasisof different
Ti/ZrandZr/Yratios
(seeFigure
7).In addition
to ourdata,the
plots
include
results
obtained
previously
forRajmahal
lavas
[Mahoney
etal.,1983;
Storey
etal.,1992;Baksi,
1995].
Samples
RM82-10andRB88/34
haveanomalously
highSiO2 abundances
andlowCaO/A1203 ratios,
andareinferred
to be altered.

-100 and Zr/Y >4 [Baksi, 1995]. We suspectthat for outliersin Figure7 are RB88/34 [Storeyet aL, 1992] and
RM82-10 theseratiosareartifactsof analyticalerror,but at RJ2-7-7B (Table2). RB88/34haslow Zr (50 ppm)andon
present cannot rule out a real distinction between this the basisof major elementcomposition (e.g., Figure6),
sampleandthe RajmahalGroupI and II basalts.The two appears to be altered.RJ 2-7-7B haslow TiO2 (1.33 wt%)
anomalousBengal Basin lavas have low Zr abundances relativeto Y, possiblyreflectinganalyticalerror.
(-•90 ppm) relativeto other incompatible elementsand Concentrations of SiO2andK20 in RajmahalGroupI
non-chondritic Zr/Hf ratios(>42); thesevery likely reflect basaltsgenerallyare lowerthanthosein GroupII rocks
analytical errors in the measurementof Zr. Two other (Figure6), whereasCaO and Sc contents are
166 RAJMAHAL BASALTS

140 group minerals.Our secondsilicic sample,RJ 2-1-3, is a

. Casuarina rhyolitic tuff with 78.7 wt% SiO2 and 6.3 wt% K20. This
- RB88/34 ß Group l
specimenhas a major elementcompositionsimilarto that
120- ß ß O Group
II of a silicic tuff from the Sylhetregion [see Talukdarand
Murthy, 1970].
,,...lOO oBoB Sampleswere collectedalso from two 30-m-wide dikes
•N Av.
N-MORB of Paleoceneage (-64 Ma) in the easternDamodarValley.

•'80Av.
E-MORB
•••••'•
The dike specimens(Maheshpur,abbreviatedto 'Mah' in
Gosselin Table 2, and Salma) are characterised by low SiO2(49.2-
60 ß
O OO
RM82-10 100 _

40 ,,,, • .... • .... • .... •RJ2-7-7B i i i i _

1 2 3 4 5 6 50--

Zr/Y
_

Figure 7. Plot of Ti/Zr vs. Zr/Y, comparingRajmahal basalts

with 129-123•a lavasfromBunbury,
Western
Australia,
and
averageenriched('E-type') and normal ('N-type') MORB. See
10•
text for an explanationof outlying Rajmahaldata. 'B' denotes
Bengal Basin lavas of Baksi [1995]. Bunbury lavas are divided Q) --
• 5
on the basisof different age and chemicalcompositioninto two
magma types, termed Casuarinaand Gosselin [see Frey et al.,
1996]. MORB averagesare from Sun and McDonough[1989]. Rajmahal; RJ
• - 1-26-7
= RJ 1-27-6

z
roup I []RJ1-30-5
higher in Group I basalts(9.9-11.8 vs. 8.9-10.3 wt% and '• RbBaTh K TaNbLaCeSrNdHf ZrSmTi Tb Y
34-42 vs. 28-36 ppm, respectively). Abundancesof ;;100
compatibletrace elementsin the two groupsshow wide
ranges:for example,Group I basaltshave 31-147 ppm Ni
>e 50
and 105-673 ppm Cr, whereasGroup II basaltshave 18-
210 ppm Ni and 45-790 ppm Cr (note that the rangesof
values quoted above and below include publisheddata).
Rajmahal Group I basaltshave low Ce/Y (0.4-1.0) and lO

exhibit a wider range in Zr/Nb (14-34) when comparedto

Group II basalts(0.9-1.4 and 13-23, respectively).Group 5
o

I basalts also have low La/Ta (19-22), variable Ba/Ta

(165-584) and low to moderateTh/Ta (1.2-3.1). Low Nb
and Ta are a feature of both Rajmahalmagmatypes,but
are more obvious in Group II basaltsbecauseof higher
relative abundancesof K and La on primitive-mantle-
normalisedtrace elementplots(Figure 8).
In addition to basalt samples,we analysedtwo silicic
rocks from the northernRajmahalHills (Tables 1 and 2).
The Taljhari rhyolite flow, representedby sampleRJ 2-2-
2, is a micrporphyriticpitchstonelocally up to 21.5 rn
thick. In comparisonto two amygdule-freesamplesof the
Taljhari rhyolite analysedby Raja Rao and Purushottarn
[1963] (SiO2-67.5 wt%, K20 N3.8 wt%), our rhyolite
sample has high SiO2 (71.4 wt%) and K20 (8.3 wt%)
(Table 2). Thesevalues,togetherwith high abundances of
Rb and Ba (238 and 1045 ppm, respectively)and the light
rare-earth elements, are consistentwith the presencein
sampleRJ 2-2-2 of abundantalkali feldsparand zeolite-
I I I K• Ta• Nb• La• Cei Sr• Nd• Hf• Zr• Sm[ TiI TI
Rb Ba Th bY
Figure 8.(a) Primitive-mantle-normalisedincompatible trace
elementpatternsfor RajmahalGroup I basalts.Group I samples
have moderately flat patterns (the alteration-pronelarge-ion
lithophile elements excepted), consistentwith the total melt
fraction being dominated by melts generatedin the spinel
stabilityfield (<100 km depth).Normalisingvaluesfor primitive
mantle are those of Sun and McDonough [1989]. (b) Primitive-
mantle-normalisedincompatibleelement patternsfor Rajmahal
Group II samples show features akin to average crustally
contaminatedbasalts(shadedfield) from westernIndia (Bushe
Formation, Deccan), Madagascar(East Coast tholeiites) and
southernBrazil (GramadoFormation,ParantiBasin). Averages
were compiled from referencesof Saunders et al. [1992].
Normalisingvaluesarefrom Sunand McDonough
 KENT ET AL. 167

49.6 wt%), high Fe203* (16.3-16.4 wt%, where Fe203* = dikesshowa restricted
rangein tgNd(t
) from+0.1 to -0.4.
total Fe), high TiO2 (2.8-2.9 wt%) and high P205 (0.28- Values of 878r/86Sr(t)
for unleachedrock 13owders lie
0.29 wt%). Theseelementabundances, togetherwith low between 0.70403 and 0.70911, extending the previous
Ni and Cr concentrations(65-67 ppm and 78-84 ppm, range in initial Sr isotope ratios (0.70385•0.70828) to
respectively)are consistent with extensivefractionationof significantlyhigher values.The total range in present-day
tholeiitic magma at low pressure.Our chemical and 2ø6pb/2ø4pb
for our samplesis 17.93•18.03,2ø7pb/2ø4pb
=
isotopicresults suggestthat the Maheshpurand Salma 15.52g15.66,and2øSpb/2ø4pb
= 38.12g39.14.
dikesare very similarto ferrobasalticlavasfrom the upper Figure 9 revealsthat the two groupsof Raimahalbasalt
partof theDeccanTrapssequence at Mahabaleshwar Ghat, identified by Storey et al. [1992] cannot easily be
westernIndia (for example,sampleMB81o18 of Mahoney distinguishedusing NdoSr isotoperatios. Group I basalts
et al. [1982]). The Mahabaleshwarsectionlies about 1400 arecharacterised
by hightgNd(t
) (+5.1to +0.1) anda wide
km to the west of the Damodar Valley, whereasthe nearest rangein 878r/86Sr(t)
(0.7037-0.7084),whereasGroupII
exposedDeccanTrapslavasanda Mahabaleshwar-like sill basalts
havetgNd(t
) valuesof +0.1to -6.7 and87Sr/86Sr(t)
[e.g., Sen and Cohen, 1994] lie about 240 km to therangingfrom 0.7050 to 0.7091. In Figure 9, a Group I dike
southwestof the Damodar Valley. Given the very large
(RJ 1-19-1) and two Group II samples (RJ 1-13-1 and
size of the Maheshpurand Salmadikesandthe observed Jalangi,a samplefrom the Bengal Basin [Baksi, 1995]) lie
similarityto Mahabaleshwarferrobasalts,
it is possible
that abovethemainisotopic
array.Theserockshave87Sr/86Sr(t)
the dikesrepresent
liquidstransported
eastwards acrossthe of 0.7080-0.7084andgXd(t)
valuesof +0.1 to-0.2; they
Indian subcontinent over a distance of > 1000 km.
continue the trend from Group I sample RM82-5
[Mahoney et al., 1983] towardsmoderatelylow valuesof
IsotopicVariations.Figure9 showsthatthenewNd-Sr gXdandhigh 87Sr/86Sr. A secondgroupof sampleslying
isotoperatiosof Rajmahalbasalticlavas and dikes are offsetfrom the main Nd-Sr isotopicarray includeRJ 2-7-4,
similarto thoseof samplesanalysedpreviously[Mahoney
2-7-5, 2-7-7A, and a rhyolitic tuff (RM82-15 [Mahoney,
et aL, 1983; Storey et al., 1992; Baksi, 1995]. For our
1984]).Theserockshavemoderate valuesof 878r/86Sr(t)
samplesonly,thetotalrangein tgNd(t
) is-3-2.0
to -6.7,where (0.7045-0.7070)
andlowtgNd(t
) (-1.5to -11.1).
t- 116 Ma. Relativeto the Rajmahallavas,the Cretaceous
In Figures 10 and 11, Group I basaltshave measured
valuesof 2ø6pb/2ø4pb, 2ø7pb/2ø4pb,
and2øSpb/2ø4pb that are
+6
very similar to thoseof Group II basalts(17.93-18.03 vs.
+4
17.96-18.03, 15.52-15.60 vs. 15.56-15.66, and 38.12-
38.48 vs. 38.15-39.14, respectively).Note that we do not
+2 includein our plots or discussionthe RajmahalPb isotopic
data illustratedby Storey et al. [1992, figures 11 to 13].
Casuarina
d•
0
Their samples,collectedindependentlyof ours, appearto
have becomecontaminatedwith Pb prior to analysis.
O
_ • Figures 10 and 11 show that the new Rajmahal isotopic
Gosselin measurementsare very similar to thoseof Cretaceouslavas
from the southernKerguelen Plateau (ODP Site 738), the
(30
-6
O NaturalistePlateau,someBunburylavas,and someBroken
_ O This study (d = dike)
'•' Mahoney et al. [1983] _ Ridge lavasoIn Figure 10, the Rajmahal basaltslie just
-8
A Storey et al. [1992] _
outsidethe field of present-dayMORBs from the Indian
-10 - 0 Baksi[1995] RM82-15 - Ocean, but in Figure 11 the two data setsoverlap. Lavas
-,,,, I,,,,1•,,I .... [] .... I .... I .... - from islandsin the southernIndian Ocean,representedby
0.703 0.704 0.705 0.706 0.707 0.708 0.709 0.710 the Kerguelen Archipelago, Heard Island, and the
87Sr/s6Sr(t) McDonald
Islesoverlap
withthefieldof Rajmahal'Group
II basalts,raisingthe possibilitythat at leastsomeof our
Figure9. InitialeNd(t)versus 87Sr/S6Sr(t)
for Raimahal samples samplescouldhave been generatedby the Kerguelenhot
fromthisandpreviousstudies,compared to lavasfromBunbury,
WesternAustralia.RM82-15 is a rhyolitictuff from the Rajmahal spot (but see below). In contrast,Recentlavas from the
HillswitheNd(t)of-11.1 and87Sr/86Sr(t) of 0.70702[Mahoney, southwestIndian Ocean,represented by alkali basaltsfrom
1984]. 'Salma' is a ferrobasaltic dike of Paleocene age,analysed the Crozet Archipelago, show no overlap with the
aspartof thisstudy.Fieldsfor theBunburylavasarefromStorey Rajmahal basalts.The Rajmahal Pb isotopic data also
et al. [ 1992] andFrey et al. [1996]. generallyare unlikethoseof basaltsfrom
168 RAJMAHAL BASALTS

+12
1992d]. Furthermore, it is not at all obvious from the
- ,.• Pacific,• availablechemicaland isotopicdata that Afanasy-Nikitin
+8 "• • '"•i•-I•-" •_ N. Atlantic
seamountlavas are relatedto the Crozet plume sourceat
• .... • M•B •
Salma • "••49 • 80 Ma (see discussionby Mahoney et aL [1996]).

+4 • • '•.'•++• CrozetA.-
0 39-41• • •d •.• /
Nevertheless,until samplesare collectedfrom the Crozet
Plateau (the-70-m.y.-old volcanic edifice on which the
SWIR / + •. '....::.•/KerguelenA., CrozetArchipelagowas constructed) and the 85øERidge,

-4 •747 ' Heard

I. it will not be possibleto dismissa link between the
Raimahalbasaltsandthe Crozethot spot.
-8 5.DISCUSSION
-12 I • • • , 5.1. Evidencefor Fractionationand Contamination
of
0.712
[ * Indian
Lamproitos
- Rajmahal
Magmas
L b ßBunburyBasalt Incontrasttoolivine
tholeiites
drilled
onthecentral
0.710
/
I-'
738•
':•
ß
........
ß Naturaliste
Plateau
• 0 d + Kerguelen
-
Plateau,
-
Kerguelen
Plateau,
the
Rajmahal
basalts
cluster
along
the1

(•0'708
]_
Af-Nik*•
• ,748 40.0
•
L •'• Broken
Ridge _

! 0 747 Kerguelen
A., [ • Kerg.
A./• _
/ •'•4F--•%' '""-,E',:,
.......
.....
' Crozet
A.- •
0.704
|•-39•••½••.•_+•49(:::Z:::::::::::•
SWIR' '•U•E
•• • ") / Pacific, • - % • 39.0
0.702
/ l- , \ Most
Indian
MORB
/•N-
Atlantic
ORB •
M, •• • Heard
I. •/ •_
16.5 17.0 17.5 18.0 18.5 19.0 19.5 •O_ / 747,7/4../'•'""'""•:•":•••'
"••• • /•// /
2ø6pb/2ø4pb •o 38.oh Salrna '• "•<-Most
Indian
-
] 39-41øE• '"•// . •"• •' MORB
Figure10.(a)InitialsNd(t)vs.present-day
206Pb/204Pb,
and(b) I- _
87Sr/86Sr(t)
vs. present-day 206Pb/204Pb
for Rajmahalbasalts / •_.•// /..// / Pacific,N. Atlantic
(datafromthisstudy,symbols
asin Figure9). TheRajmahaldata ! 0 Af-Nil••(/ // MORB
370 I ...... IX'"•,, ,•U, , • • I•, s, I ,, ,
are comparedto our unpublishedresultsfor the DamodarValley 15.8

-b 738½
ø•ø• 4
lamproites and data for Bunbury [Frey et aL, 1996] and
NaturalistePlateaulavas [Mahoneyet al., 1995]. We also show
fields for the Kerguelen Plateau and Broken Ridge [Alibert, 15.7-- f• 749747,750 -
1991; Salterset al., 1992, and V. J. M. Salters,unpublisheddata; - • '•748 Crozet
A. .
Mahoneyet al., 1995], KerguelenArchipelago,Heard Island,and
McDonald Isles [Storeyet al., 1988; Gautier et al., 1990; Weiset
•15.6- Most
Indian
/•••'•::•%'
'••••;•-
'""
al., 1993; Barling et al., 1994], the Afanasy-Nikitin seamount ß
• - ,
(Af-Nik) and CrozetArchipelago[Mahoneyet al., 1996], Indian, O•15.5
Pacific, and North Atlantic MORB (referencesof Mahoneyet al. • _- Pacific,
N.Atlantic
[1989, 1992]) and 39ø-41øE SouthwestIndian Ridge (SWIR)
15.4
lavas[Mahoneyet al., 1992].

Afanasy-Nikitin seamount [see Mahoney et aL, 1996; 15.3

Sushchevskaya et al., 1997]. This seamount,togetherwith 16.5 17.0 17.5 18.0 18.5 19.0 19.5

the 85øE Ridge and Raimahal basalts,was believed by 2o6pb/2O4pb

Curray and Munasinghe[1991, 1992] to lie on the track of
the Crozet hot spot. However, unless the isotopic
compositionof the Crozetplume (as representedby alkali
basaltsfrom the Crozet Archipelago)changedmarkedly
between 116 Ma and 80 Ma, and/or componentsexternal
to the plume head were involved in producing the
Raimahal basalts,the Raimahal basaltscould not have
been generatedabovethe Crozet hot spot [cf. Kent et al.,
Figure 11.(a) Present-day 208Pb/204Pbvs. 206Pb/204Pb,and (b)
present-day 207Pb/204Pb vs. 206Pb/204Pbfor Rajmahalbasalts,
comparedto data for other lavas eruptedin and aroundthe
easternIndian Ocean.All data are measured ratiosexceptfor
thosefor basaltsfrom the Afanasy-Nikitinseamount(age-
correctedto 80 Ma [Mahoney et al., 1996]). Symbols,
abbreviations, and data sourcesare as in Figure 10. To avoid
clutter,datafor theDamodarValleylamproites arenotincluded.
 KENT ET AL. 169

atmosphere cotecticfor olivine,clinopyroxene, plagioclase basaltsshowpeaksat Sr, the oppositeof what onewould
and liquid in CIPW norm space(not shown).This expectif theserockshad fractionated a largeamountof
observation, togetherwith ourchemicalresultsin Table2, plagioclase relativeto GroupII basalts.Furthermore, in
is consistent at fairly shallow contrastto the Bunburylavas,Ga abundances
with fractionalcrystallisation in the two
levels in the crust. At low pressures,plagioclaseand groupsof Rajmahal basalt are virtually identical at
clinopyroxeneform the bulk of the crystallisingdifferent abundancesof La (Figure 12a). Plagioclase
assemblagein tholeiitic magmas [e.g., Green and segregation at low pressuresshouldleadto Ga depletionin
Ringwood, 1967],resulting in highCaO/A1203 andlow Sr residualmelts,in a manneranalogousto thebehaviorof Sr
and Sc concentrations in residualmelts. High CaO/A1203 [e.g.,Goodman,1972].The fact thatGa is not depletedin
and Sc and low Sr contentsin RajmahalGroupI basalts Group II sampleswhen comparedto Group I rocks
relativeto GroupII rocks(Figures6 and12)couldimplya suggests that the lower CaO/A1203and Sc and higherSr
higherratioof plagioclase to pyroxene in thecrystallisingcontentsof GroupII basalts(Figure 12b)reflecta process
assemblage, consonantwith fractionationof Group I other than low-pressurefractionation.One possible
magmasat pressures lowerthanthoseat whichGroupII explanation for the differencesis variable,but ratherlow,
magmas weredifferentiated. Thisexplanation wasfavored amountsof contamination by continentalcrust.Combined
by Frey et al. [1996] as a meansof accounting for assimilationand fractional crystallisationwould tend to
compositional differencesbetween-129-m.y.-oldand result in the most evolved lavas being the most
-123-m.y.-oldbasaltsfrom the BunburyBasin,Western contaminated [e.g., DePaolo, 1981'].Figures6, 12c, and
Australia (Casuarina- and Gosselin-type lavas, 12d showthat this is not true of the Rajmahalbasalts[cf.
respectively; Figures7 and12).However,in Figure8a,the Mahoneyet al., 1983]; for example,at---7 wt% MgO,
primitive-mantle-normalised patternsof RajmahalGroupI GroupII rockshave SiO2 and K20 abundances
:
that are

26

• 24
•RB88/34
ß o o% a RB88/34
c•00 •:).,-, 0 • 3

Casuarin•(•
tD 00•'
v
ci. 22
0 0 8 2 RM82-10

2O
asuarin• oø
', ø
0 Gosselin

I I , I , I , I , I , I ,
18
500
2o

b ß Group I
d
Casuarina•
oGroup
II
• 400
15
ß

E
O oo
v
d. 300
ß

2OO RB88/34

Gosselin
100 0 , I , I , I , I
' ' ' ' '
25 46 48 50 52 54 56 58 60

La (p.p.m.) SiO (wt.%)

Figure12.Plotsof (a)Lavs.Ga,(b)Lavs.Sr,(c)Sr/Zrvs.SiO2,and(d)LaYTh vs.SiO2forRajmahal basalts
from
thisworkandprevious studies
[Mahoney et al., 1983;Storeyet al., 1992;Baksi,1995].Fieldsfor Casuarina-
and
Gosselin-type
Bunbury lavas[Storey
etal., 1992;Freyetal., 1996]areshown forcomparison. Notethelocations of
dataforRajmahallavasRM82-10andRB88/34, inferred
onthebasis of majorelement
abundances (Figure6) to be
170 RAJMAHAL BASALTS

significantly higher than those of Group I basalts.This partialmelt derivedtherefrom,is in keepingwith slightly
suggests that chemicaldifferencesbetweenthe two groups highSiO2contents andhighabundances of Sr in Rajmahal
were establishedprior to extensivefractionationof low- Group II basalts(Figures 6, 8b, and 12b). Primitive-
pressurepyroxeneandplagioclase. mantle-normalised traceelementpatterns(e.g., troughsat
Few chemical analyses of Indian crustal rocks and Nb andTi in Figure8b) andPb isotopicratiosof GroupII
mantle xenoliths exist to identify the contaminant(s)of basalts (in particular, high 2ø7pb/2ø4pb relative to
Rajmahalmagmas,but key featuresof our data set allow 2ø6pb/2ø4pb) alsoareconsistentwith minorassimilation of
us to discem the broad composition of the material old siliciccrust.A likely candidatefor the contaminant is
involved. Our unpublishedisotopicresultsfor unleached the ProterozoicChotanagpur(Bengal) granite-gneiss,
samplesof DamodarValley lamproitefall roughlyin the whichformsthe bulk of surfaceexposures on the eastern
middle of the Nd-Pb and Sr-Pb isotopic arrays for Chotanagpur Plateau[Ghose,1983]. Simplecalculations
Rajmahal basalts(Figure 10), allowing us to firmly rule suggestthat in the presenceof a free fluid phase,granite-
out lamproite or the lamproite mantle source as the gneisswould undergopartial melting if heatedfrom an
contaminant. This is important, because the Indian originaltemperature of-300øC (appropriate to thecrustof
lamproitesare thoughtto be representative of metasomatic the Indianshieldat a depthof-•16 km [e.g.,Negi et al.,
vein material residing within the lowermostlithosphere 1986]) to a temperaturein excessof 800øC.Assimilation
[e.g., Middlemost et al., 1988]. If such readily fusible of small melt-fractionsof Chotanagpurgranite-gneiss
material did not mix with Rajmahal magmas,it is most couldhaveoccurred,for example,at the marginsof dikes
unlikely that cooler,pristinelithosphericperidotiteswould supplying the Rajmahal lavas. Modeling of the
have acted as a contaminant (see, for example, the contamination
processawaitstrace elementand isotopic
thermomechanical models for plume- lithosphere measurements
on samplesof the granite-gneiss.
interactionsdiscussedby Arndt and Christensen[1992]).
We infer from this that the Rajmahalcontaminantis crustal
in origin. 5.2. Depth and Degreeof Partial Melting
The Nd-Sr isotopicratios of Rajmahalbasaltssuggest
three crustal contamination vectors diverging from a In light of suggestions of a petrogeneticlink between
Rajmahaland KerguelenPlateaubasalts[e.g., Davies et
mantleend-member
with e•d y+5 and 87Sr/86Sr
50.7037
al., 1989;Storeyet al., 1992], it is pertinentto considerthe
(Figure9). The first vector,definedby dataplottingabove
melting conditionsunder which both suitesof rockswere
the main Nd-Sr isotopicarray, points towardsa crustal
produced.We adopt the fractionalmelting inversionof
end-memberwith e•qd<-0.2 and 87Sr/86Sr
y0.7084. The
McKenzie and O'Nions [1991, 1995], incorporatingthe
secondvector(main array) pointstowardsan end-member
modificationsproposedby Whiteet al. [ 1992], as a means
with e•d <-7 and 87Sr/S6Sr
y0.7091. Rajmahalbasalts of placingquantitativeboundson meltingconditions. The
defining the third vector (below the main array) were inversionutilisesaveragedconcentrations of the rare-earth
contaminatedby an end-member with e•d <-11 and elements to estimate the melt distribution as a function of
87Sr/86Sr
y0.7070. These vectorscould point to three depth,the total integratedmelt fraction,and the total melt
separate crustal end-members, each with similar Pb thickness,equivalent to the thicknessof basaltic crust
isotopic characteristics
(2ø6pb/2ø4pb-18.0,2ø7pb/2ø4pb produced.Concentrations of the majorandminorelements
y15.7, and:øSpb/2ø4pby39.1), or to a singlecontaminant are then predicted from the results of the rare-earth
that is markedly heterogeneouswith respectto end and element inversion [e.g., White and McKenzie, 1995].
87Sr/86Sr.In the absenceof radiogenicisotopedata for Assumptions aremadewith regardto startingcomposition
easternIndian crust,we cannotdistinguisheasilybetween (i.e., the mantlesourceis specifiedas havinga chemical
thesepossibilities. composition that is primitiveor depletedin lightrare-earth
If the amountof crustassimilatedby Rajmahalmagmas elementsrelativeto the bulk earth),sourcemineralogy,
was <_5%,as suggestedby limited effectson major element phase proportionsentering the melt, and mineral-melt
(e.g., SiO2, K20) concentrations,
the contaminant(s)may partitioncoefficients.Only datafor sampleswith >_6wt%
have had very low e•d (probablymuch lessthan -20) and MgO were used.The inversionswere run assumingthat
high 87Sr/86Sr (yy0.710). Archeangranitoidsin western the garnet-spineltransitionoccursbetween80 and 100 km
India have these characteristics,with e•d of about -40, depth, equivalentto a mantle potentialtemperatureof
87Sr/86Sr
of about0.820, and unusuallyhigh Sr contents, 1500øC(the potentialtemperature
is the temperature
that
sometimesin excessof 500 ppm [Lightfoot, 1985]. An mantleperidotitewouldhaveif it ascended adiabatically
to
ancientgraniticcontaminantof this type, or small-volume the surfacewithout
 KENT ET AL. 171

Figure 13a showsrare-earthelement measurementsfor illustratethe observed(dots) and predicted(solid lines)

the Rajmahalbasalts(averageof four GroupI and seven concentrations
of major and othertraceelementsusingthe
Group II samples)normalisedto estimatedconcentrations melt distributionwith depthinferredfrom the inversionfit.
in the bulk earth. Observed mean concentrations for the Figure 13d showsthe partial melt distributionrequiredto
Raimahalbasaltsareshownby heavydots,with errorbars produce this fit, with the dashed line illustrating the
indicating one standarddeviation, estimated from the inferred original melt distribution after correction for
varianceof the data.Theheavysolidline showsthebestfit 15.5% olivine fractionation(the correctionis basedon the
to the data obtained by inversion. The mantle source misfit between calculatedand observedaverageMgO and
producingthe best fit is primitivemantle(single-stagetotal Fe asFeO; seeMcKenzieand 0 'Nions[ 1991]).
meltingof a depleted sourcecompositioncannotproduce a An effect of includingGroup II basaltsin the inversion
good fit to the light rare-earth element abundancesof model is that althoughthe amountof crustaddedto these
crustallycontaminated basalts;seebelow andBrodieet al. basaltsis inferred to be modest (_<5%),the contaminant
[1994]). The fit is leastgoodfor Eu, but for all rare-earth imparts a weak light rare-earth element/heavyrare-earth
elements is well withintheerrorbars.Figures13band13c element fractionation signature that is not a result of

40

a b

'• 10

• 6

La Ce Pr Nd SmEu GdTb Dy Ho Er TmYb Lu Y GaAI CaSc V MnFe Si CoMg Cr Ni

60

40-
0.21
'• Spinel
• Garnet
[
!
-d - - - Corrected
for
fractionation
of15.5%
20-- 0.1-1---.

10-
8 -

0,0
Cs Rb K BaTh U Ta Nb P Sr Zr Hf 'lq Na 50 lOO 15o

Depth (km)
Figure13.(a)Observed andcalculatedrare-earth
element concentrations,
normalisedto bulkearth,for Rajmahal
basaltswith>6 wt% MgO (average of neutronactivation datafromthisstudy)[cf.McKenzie andO'Nions,1991,
1995].Meanvaluesfor eachelementareshownby dots.Verticalbarsillustrate onestandard deviation,
estimatedfrom
thevarianceof thedata.Theheavylineshows thebestfit tothedataobtained byfractional-melting
inversion.Results
obtainedforKerguelen PlateauODPSite747areshown forcomparison. (b)and(c)Predictedconcentrationsof major
andothertraceelements usingthefractionation-corrected
primarymeltdistribution
(paneld) inferredfromtherare-
earthelementinversion.(d)Uncorrected(heavyline)andcorrected (shortdashed
line)meltdistributions
forRajmahal
basalts,
estimated fromthe rare-earth
elementinversion fit. Themantlesourceusedin the inversions is primitive
mantle[seeMcKenzieand 0 'Nions,1991,
172 RAJMAHAL BASALTS

meltingin the mantlesource(compareFigures8a and 8b). ODP siteson the NinetyeastRidge were reportedby Kent
This is expressedin the melt distributionmodel (Figure and McKenzie [1994]. All of these lavas were probably
13d)by a low-amplitudetail at depth> 100 km, whichmay eruptedin close proximity to the Kerguelenhot spot;as
also includesmall melt-fractionsgeneratedin the garnet noted above, the Rajmahal basaltspossiblywere erupted
stability field; there is no significanteffect on the melt -1000 km away from the center of this hot spot. Our
distribution at shallower levels. If the tail is excluded, the modeling assumesthat basaltssampledat the ODP sites
resultant fit to the highly incompatible trace element (i.e., the uppermostflows) are representativeof the entire
concentrations shows a marked deterioration. This result igneouscrustalsectionin termsof their rare-earthelement
suggeststhat the inversionmodel for Rajmahalbasaltsis concentrations.This is unlikely to be true, but the
relatively insensitive to minor amounts of crustal inversionsdo at least provide a quantitativemeans of
contamination(the effect is similar to addingfurther low- comparingavailabledata.
degreemelts from the garnet stability field [cf. Brodie et Figure 14 shows the melt distributionsobtainedfor
al., 1994]), and that inclusionof Group II samplesin the Kerguelen Plateau and Loranchet Peninsula basalts,
model does not obscure conditionsof melt generation corrected for the effects of olivine fractionation. Details of
within the asthenosphere. the fractionationcorrectionapplied to lavas from each
The upper limit of melting in the inversionis-50 km, locality are given in the caption to Figure 14. The
typical of basaltsgeneratedbeneathstretchedand thinned estimated total melt fraction and total melt thickness are
continentallithosphere(see, for example,figures 13 to 17
of Whiteand McKenzie [ 1995]). The lower limit of melting • Spinel
is not clearly defined,due to uncertaintyover the amount •- Plag. --•
0.4
of melt generated in the garnet stability field. The
maximum melt fraction estimated from the inversion is
about 10% (Figure 13d) or about two-fifths less than
McKenzie and O'Nions [1991] estimatefor the average 0.3-

extent of melting for normal MORB. The total melt

thicknessgenerated,after fractionationcorrection,is 5.1 ""•oo "",, •',, / Kerg. A.
km, or some 2 km less than the average thicknessof ............. '",,
oceaniccrust [e.g., Whiteet al., 1992]. The melt thickness
estimateis muchgreaterthanthe observedthicknessof the , 749 • "%.• ',
Rajmahal-Sylhetlava pile. This discrepancycould arisein
severaldifferentways. Some of the melt couldhave been 0,1 - "747
I '"',,,
• .•.,
'"',,,
redistributedby lateralflow, consistentwith the occurrence ß I '"'X'•x'" '"'
of dikes to the west and southwestof the RajmahalHills
Ral,I ' '"x'x '"',
and on the Shillong Plateau (Figure 3). Alternatively, 0.0
basaltcouldhave been removedover the last 116 m.y. by 0 50 1 O0 150
erosion.Neither of theseprocesses are likely to accountfor
the large discrepancynotedabove.However, it is possible Depth (km)
that melt was trapped at the Moho or in the lower crust. Figure 14. Partial melt distributionsinferredfrom rare-earth
This inference is reasonablebecausethe Moho probably element inversions of data for tholeiitic basalts from the eastern
acts as a density filter for basaltic melts which have a Indian Ocean,comparedto Icelandiclavas.KerguelenPlateau
densityof-2.8 kg m-3 [e.g.,Huppertand Sparks,1980]. melt distributioncurvesare numberedaccordingto ODP site.
Unfortunately,a lack of modem seismicprofilesacrossthe 'Raj' is the fractionation-corrected
Rajmahalmelt distribution
easternIndian margin meansthat at presentwe cannottest from Figure 13d, whereas'Ice' is that obtainedfor Icelandic
basaltsby McKenzieand0 'Nions[ 1991]. Dottedlinesshowmelt
the hypothesisthat igneousunderplatinghascontributedto
distributioncurvespredictedfor isentropic
decompressionpaths
the-35-km-thick crustbeneaththe southernBengalBasin
of mantle at potentialtemperatures of 1350øC, 1450øC,and
[e.g., Brune and Singh, 1986]. 1550øC (see also fig. 10 of White and McKenzie[1995]).
In order to place the Rajmahal results in a regional Correctionsfor olivine fractionationare 9.6% (ODP Site 747),
context, we used the inversion schemealso to model ODP 13.3% (Site 749), 35.9% (Site 750), and 11.8% (Loranchet
data for the Kerguelen Plateau [Storeyet aL, 1992] and Peninsula,KerguelenArchipelago).Data sources:Rajmahal- this
datafor LoranchetPeninsulatholeiites,northernKerguelen study; Kerguelen Plateau - Storey et al. [1992]; Loranchet
Archipelago [Storey et al., 1988]. Inversion results for Peninsula- Storeyet al. [
 KENT ET AL. 173

greatestat KerguelenPlateauSite 750 (-32% and 26.2 km, reconstructions (Figure lb). Melt distributionsfor basalts
respectively,using a primitive mantle source).Resultsfor from Sites749 and747 againsuggest highmantlepotential
this site are comparableto those obtained for enriched temperatures(-1400øC) but a lower degree of melting,
MORB-like basalts from Iceland (-30% and 22.3 km, consonantwith the presenceof thicker lithosphereabove
respectively,for melting of a primitive mantle source;see the plume.Thesesitesprobablylay slightlyoff-axis(recall
McKenzie and 0 'Nions [ 1991]) and for Cretaceouslavas that lavas from Site 749 are about 114 m.y. old, whereas
from oceanic plateaus in the western Pacific (e.g., the basaltsfrom Site 747 are about 85 m.y. old). Figure 14
estimated total melt fraction and total melt thickness for shows that Eocene tholelites from the Loranchet Peninsula
the Ontong Java Plateau are 28-30% and 17-24 km, also have a melt distributionconsistentwith their being
respectively,dependingon the sourceconcentrations used produced by melting of anomalouslyhot (-1460øC)
in the inversions;see Mahoney et al. [1993]). The melt mantle. The melt distribution inferred from inversion of
fractions at Kerguelen Plateau Sites 749 and 747 are Rajmahal basalt data (Figures 13d and 14) suggestsa
somewhat smaller than for Site 750 (18% and 14%, mantle potentialtemperatureof about 1350øC(i.e., 40ø to
respectively,for a primitive mantle source).Site 749 has 50øC higher than the potential temperaturerequired to
an 'Icelandic'-type melt distribution,with the top of thegenerate normal MORBs [cf. McKenzie and O'Nions,
melt column at a depth of-35 km and the base at-120 1991]). A potential temperatureof 1350øC is consonant
km. The total melt thickness is 11.6 km. Note that the with eruptionof the Rajmahalbasaltson the peripheryof
thicknessestimatesfor igneouscrustat Sites 749 and 750 the thermalfield of the Kerguelenplume [cf. Miiller et al.,
differ considerably. However, because the rare-earth 1993].
element data used in the models are not necessarily To summarise,the Raimahal lavas appearto represent
representativeof the entire crustal sectionat each site, moderatedegree (-10%) partial melts generatedbeneath
crustalthicknessesestimatedby inversionmay not be close thinned lithosphere. A near-ridge, steady-stateplume
to true thicknesses. Indeed, seismic reflection and ('tail') origin is suggestedfor KerguelenPlateauSite 749
refractiondata (referencesof Operto and Charyis [1995, and 750 basalts, which have melt distributions akin to
1996]) currently provide no obviousindication of large those of Recent Icelandic lavas.
variations in crustal thickness between ODP sites.
It is interestingthat the resultsfor Site 749 are not 5.3. Mantle Sources
dissimilar to those we obtain for tholeiites from the
LoranchetPeninsula(Figure 14). Theselavasare believed With the exceptionof alteration-and/or contamination-
to be representativeof the shield-building stage of prone incompatible elements, the primitive-mantle-
subaerialvolcanism on the Kerguelen Archipelago [e.g., normalisedtrace element patternsof Rajmahal Group I
Storeyet al., 1988]. The total melt fractionand total melt basalts(Figure 8a) are relatively flat. Flat patternsresult
thicknesscalculatedfor the Loranchettholeiites(---18%and from the total melt fraction being dominatedby melts
6.9 km, respectively) are similar to those of normal generatedin the spinelstabilityfield, and are characteristic
MORBs. As with tholeiites from the Kerguelen Plateau, of basementsamplesfrom the centralKerguelenPlateau
meltingbeginsat a depthof about120 km, or some50 km [Salterset al., 1992] and lavasfrom oceanicplateausin the
deeperthanwould be expectedfor ambientmantlewelling western Pacific (Nauru Basin, Ontong Java, Manihiki)
up at mid-oceanridges [cf. White et al., 1992]. Deep [e.g., Saunders,1986; Mahoneyet al., 1993]. Despitetheir
melting is to be expectedif the mantle were hotter than similar trace element patterns,a few Rajmahal Group I
normal, i.e., if melting occurredin a plume (potential basalts have low Th/Ta when comparedto Kerguelen
temperatures of 1450-1550øC[ Whiteand McKenzie,1989, Plateau lavas (i.e., they are more MORB-like; see Figure
1995]). 15). The sameis true for the least-contaminated Casuarina-
Comparisonof the melt distributioncurvesinferredfor type lavas from Bunbury, Western Australia. Data for
Kerguelen Plateau ODP sites with melt distributions these rocks overlap the MORB field in Figure 15a. In
predictedfrom isentropicdecompression of asthenospheric contrast,a small number of Rajmahal Group I basalts,all
mantle (curves defined by short dashesin Figure 14) RajmahalGroup II basalts,and the Gosselin-typeBunbury
suggestsa mantle potential temperatureat Site 750 of basaltshave high Th/Ta and Ba/Ta. A likely explanationis
about 1450øC.We infer from this that Site 750 lay closeto that Ba and Th have been addedto theserocks by crustal
the centerof the Kerguelenplumeat 114 Ma; the shallow contaminationof magmas.The sameprobablyis true of
top of the melting columnsuggests a positionbeneaththe KerguelenPlateaubasaltswith high La/Ta, Th/Ta, and
axis of an oceanic ridge, in good agreementwith plate Ba/Ta [cf. Storeyet al., 1989, 1992;Mahoneyet al.,
174 RAJMAHAL BASALTS

50[ Casuarina •....

Korguolon This
Gossolin explanation
'::;'ii!i:i requires
lithosphere
were that
rafts
ofcontinen
incorporated
into
the
Kerguelen
Plateau
at
•-.•• Plat•a•c.•
h ':':':'•'••
anearly
stage
inits
development
(cf.
Section
3).In Figure 15b, KerguelenPlateaulavas form a broadly

20[ •
L MORB
•
.•......
"/""'"'"'""'•
""' •_•l]]• o'•
Av.
/ I triangular
Cont. membersfield,
with
La/Ta,high
suggestive
+of
differentmixing
La/Ta
andbetween
Ba/Ta three
ratios.
Ba/Taend-member,
most
end-
The
high
apparent
in

[ • ,.,...•..•-p-
k a
Crust
/ Rajmahal
Group
IIbasalts,
Kerguelen
Plateau
Site
738 I
basalts,and Gosselin-typeBunbury lavas, is likely to be
Proterozoiccrust [cf. Frey et al., 1996]. The second(low
La/Ta,
moderate
Ba/Ta)end-member
isexpressed
inlavas
• -- ' Kerg.A. / from the KerguelenArchipelagoand is probablya
• .... • ' ' ' ' ' ' '' • component
residing
withinthe Kerguelen
plume.The
1 lO
identityof the third (low La/Ta, low Ba/Ta) end-memberis
Th/Ta lessclear. Clues can be gatheredfrom Figure 15a, where
the leastcrustallycontaminatedKerguelenPlateaubasalts
(Site 749) have La/Ta ranging from 17 to 21 and Th/Ta
lOO
from 1.6 to 1.7. The corresponding values for average
normalMORB are 19 and 0.9, respectively[e.g., Sun and
b Gosselin
/ Most
Kerguelen
/ i?
?
McDonough, 1989]. We infer from this that the third end-
memberin KerguelenPlateaulavasis probablythe Indian
MORB source.

Are the Rajmahal basaltssimply normal-MORB-like

._1 / Av. normal 'X' ___ at:•[Y
/ MORB ,•'•;';..d,;,,
•v. Cont.
I Orus,
/ Otong
lO
100 1000
Ba/Ta
Figure 15.(a) Plot of Th/Ta vs. La/Ta for Rajmahalbasalts.Note
the trend of the Rajmahal Group I basalts and Casuarina
(Bunbury) lavastowardsthe MORB field, and the similarTh/Ta
ratios of average crust and Gosselin-type Bunbury basalts.
Symbols,data sources,and averagesas in Figure 12. The MORB
field is from references of Mahoney et al. [1993]. Average
continentalcrust is from Wedepohl[1994] and primitive mantle
(denotedby 'P') is from Sun and McDonough[1989]. (b) Plot of
Ba/Ta vs. La/Ta, showingthat RajmahalGroup II basaltsoverlap
the field of KerguelenPlateau lavas, whereasthe least crustally
contaminatedRajmahalGroup I basaltsare intermediatebetween
lavas from the Kerguelen- and Ontong Java plateaus. Most
Casuarina-typelavas also lie in this region on the diagram,
whereasGosselin-typelavas are characterised by higherLa/Ta
and Ba/Ta. Symbols as in Figure 12. Data sources:Bunbury -
Storey et al. [1992], Frey et al. [1996]; KerguelenPlateau-
Salters et al. [ 1992], Mahoney et al. [1995]; OntongJavaPlateau
- Mahoney et al. [1993]. Average continentalcrust is from
Wedepohl[1994]; averageN-type MORB and primitive mantle
(shownas 'P') are from Sun and McDonough[ 1989].
melts that have assimilated variable amounts of continental
crust(reflectedin the divisioninto GroupsI andII), or do
they containa plume component? In orderto answerthis
question, we compare our results with the Pb-Nd-Sr
isotopicrangesof Cretaceous-Tertiary flood basaltsin, and
on the marginsof, the easternIndian Ocean,andthe ranges
of Indian MORB, KerguelenArchipelago,and Afanasy-
Nikitin seamountlavas (Table 5). AnalysedRajmahal,
Bunbury,and NaturalistePlateaubasaltsgenerallydo not
have the moderate2ø6pb/2ø4pb values (18.06-18.27)
thoughtby Weiset al. [1993] to correspondto the 'pure
Kerguelenplume', nor the high measured2ø6pb/2ø4pb
values(18.04-19.11) of NinetyeastRidge lavas[Saunders
et al., 1991; Weisand Frey, 1991; Frey and Weis,1995];
age-correctingthe data would move them even further
away,by -0.1 to -0.2 in 2ø6pb/2ø4pb,
assuming
a 238U/2ø4pB
ratio of 5. Indo-Australianmarginbasaltsinterpretedabove
to be least affected by crustal contaminationalso have
measured 2ø7pb/2ø4pband 2øspb/2ø4pbvalues lower than
thoseof the putative 'pure Kerguelenplume' (by up to
0.05 in 2ø7pb/2ø4pb,andup to 1.03in 2øspb/2ø4pb;
Table5).
In view of these Pb isotopic differences and the
incompatibleelementdifferencesnoted above,we believe
it unlikely that the Rajmahal, Bunbury, and Naturaliste
Plateau basalts contain a componentderived from the
Kerguelen plumeø We cannot rule out completely this
possibility becausethe Kerguelen plume, like Hawaii,
Iceland, and other large plumes, appears to
 KENT ET AL. 175

TABLE 5. Pb, Nd, and Sr IsotopicRangesfor SelectedBasaltsfrom the EasternIndian OceanandContinentalMargins

Province Age(Ma) 2ø6pb/2ø4pb

2ø7pb/2ø4pb
2ø8pb/2ø4pbl•Nd(t
) 87
Sr/86Sr(t
)

Bunbury(Casuarina) 129 17.91-18.00 15.49-15.62 37.72-38.26 +3.5 to +0.9 0.70392-0.70506

Bunbury(Gosselin) 123 18.00-18.05 15.61-15.72 38.04-39.08 *-4.6 0.70763-0.70786

Rajmahal(Group I) 116 17.93-18.03 15.52-15.60 38.12-38.48 +5.1 to +0.1 0.70372-0.70839

Rajmahal(Group II) 116 17.96-18.03 15.56-15.66 38.15-39.14 +0.1 to-6.7 0.70498-0.70911

Naturaliste Plateau 7100 17.79-18.09 15.50-15.70 37.65-39.22 +4.9 to -12.9 0.70393-0.71302

KerguelenPlateau 7114-85 17.38-18.42 15.43-15.75 37.80-39.14 +5.2 to -9.4 0.70351-0.70984

BrokenRidge 89-63 17.98-18.47 15.50-15.65 38.40-39.27 +3.4 to -2.7 0.70391-0.70734

NinetyeastRidge 82-43 18.04-19.11 15.50-15.66 38.44-39.08 +5.7 to 0.0 0.70382-0.70600

Afanasy-Nikitin Seamount ?80 16.77-18.12 15.41-15.61 37.06-37.08 +5.5 to-8.0 0.70368-0.70662

KerguelenArchipelago 39-0 17.99-18.61 15.48-15.59 38.29-39.21 +5.7 to -2.9 0.70427-0.70598

Indian MORB 155-0 16.87-22.85 15.37-15.91 37.08-39.33 +11.3 to -4.0 0.70242-0.70562

Notes:*EpsilonNd valuescalculatedfor two Gosselin-type samplesareidentical.Pb isotopicvaluesaremeasured ratios,with

the exceptionof datafor the Afanasy-Nikitinseamount(age-corrected to 80 Ma). Isotopicvaluesof Nd and Sr are age-corrected
assumingthe agesgiven in the table, with the exceptionof data for the NaturalistePlateau(correctedto 110 Ma) and Kerguelen
PlateauODP Site 738 (present-dayvalues)[seeMahoneyet al., 1995].
Sourcesfor isotopicdata:Bunbury- Frey et al. [1996];M. Storey,unpublished data.Rajmahal- Mahoneyet al. [1983]; Storey
et al. [1992];Baksi[1995];thisstudy.NaturalistePlateau- Mahoneyet al. [ 1995].KerguelenPlateau- Davieset al. [ 1989];Weiset
al. [1989];Alibert [1991];Salterset al. [1992], andV. J. M. Salters,unpublished data;Mahoneyet al. [1995]. BrokenRidge -
Mahoneyet al. [1995]. NinetyeastRidge- Mahoneyet al. [1983]; Saunderset al. [1991]; Weisand Frey [1991]; Frey and Weis
[1995]. Afanasy-Nikitinseamount- Mahoneyet al. [1996]; Suschevskaya et al. [1997]. KerguelenArchipelago- Whiteand
Hofmann[1982];Storeyet al. [1988]; Gautieret al. [1990]; Weiset al. [1993]. IndianMORB - Mahoneyet al. [1989, 1992],and
referencestherein;Weisand Frey [ 1996].

compositionallyheterogeneous[e.g., Prey and Weis, MORBs have 2ø6pb/2ø4pb values>17.8 (see Table 5) and
1995]. the leastcrustallycontaminated of our Rajmahalsamples
Table 5 showsthat analysedRajmahal,Bunbury, and havePb isotoperatiosthatareremarkably similarto those
NaturalistePlateaubasaltsdo not have2ø6pb/2ø4pb <<17.8, of averagepresent-day IndianMORB (e.g., Figure11).
indicating that they do not containthe low 2ø6pb/2ø4pb Therefore,the sourceof Rajmahalbasaltsis likely to be
material found in some Late Jurassic-Early Cretaceous compositionally 'normal'(MORB-source) asthenosphere.
IndianMORBs [Luddenand Dionne, 1992; Weisand Prey,
1996] and some KerguelenPlateaulavas (Sites 747 and 5.4. Origin of theRahmahal-Sylhet IgneousProvince
750 [Salterset al., 1992; V. J. M. Salters,unpublished
data]). This low 2ø6pb/2ø4pb materialis believedto be On the basisof the dataand discussion presentedabove,
ancient Gondwanan lithosphere incorporatedinto the it ispossible to account fortheRajmahal andSylhetbasalts
shallow asthenosphereprior to, or during continental by decompressional meltingof moderately hot (potential
breakup[Mahoneyet al., 1989, 1992;Storeyet al., 1989, temperature about 1350øC)normal-MORB-type mantle,
1992].Importantly,the absenceof thismaterialin analysed followedby assimilation of graniticcrustandlow-pressure
Rajmahalbasaltsdoes not rule out the Indian MORB- fractionalcrystallisation. We envisionthat prior to the
sourceas the sourceof Rajmahal basalts;many Indian Early Cretaceous the Kerguelenhot spotwas
176 RAJMAHAL BASALTS

beneatheasternGondwana[e.g., Kent, 1991; Kent et al., the NinetyeastRidge (Section2.1) is thereforelikely to be

1992c;Arne, 1994]. By 130 Ma, India had brokenaway coincidental.An origin for the Rajmahalbasaltsoutsidethe
from Australia-Antarctica and drifted northwards relative Kerguelenplume accommodates Miiller et al. 's [1993]
to the hot spot[e.g.,Miiller et al., 1993]. Rajmahal-Sylhet observationthat the rifted volcanicmarginof easternIndia
volcanismoccurredat about 116 Ma by melting on the wasabout1000 km away from the centerof the Kerguelen
peripheryof a steady-state plume('tail'). Partialmeltingof hot spotat 116 Ma, withoutrecourseto a large 'hot cell'
the ridge-centeredplume tail producedthe volcanic [Andersonet al., 1992] beneatheasternGondwana.
sectionsof the southernand centralKerguelenPlateauand Our preferred origin for the Rajmahal-Sylhetigneous
BrokenRidge.Upwellingandpartialmeltingbeneaththe provincealso providesan explanationfor the occurrence
riftedmarginof easternIndiamay havebeeninitiatedand of Aptian (c. 113-110 Ma) basalticdikesto the west and
drivenby viscouscouplingof the MORB mantleto flow in southwestof the Rajmahal Hills, and to the south of
the conduitof the Kerguelenplume(Figure 16). Shillong(Figure 3). The Rajmahaland Sylhetdikesattest
Basedon the resultsof numericalmodelingof steady- to magmagenerationor melt transportup to 300 km inland
state plumes, Hauri et al. [1994] proposeda similar from the rifted margin, consistentwith a thermal flux
explanation for magmatism northof theHawaiianridge,in smallerthanthat inferredfor the Deccanigneousprovince
which the flow of ambientmantle lying at the thermal (recall our suggestionthat magma in the -64-m.y.-old
boundaryof a mantleplumebecomescoupledto, but not Salmadike may have travelled> 1000 km inlandfrom the
entrainedin, flow within the plume boundary.In Hauri et rifted margin of western India). The small number of
at's model, the ascentvelocityof mantlematerialat the crystallisation agesobtainedthus far for Rajmahaldikes
thermalboundaryof the plume is of the order of 1-10 suggestthat dike emplacementcould have occurredin
cm/y,or roughlythe sameasupwellingvelocitiesinferred easternIndia -2-3 m.y. after eruption of the Rajmahal
beneath oceanic ridges [e.g., Sparks and Parmentier, lavas.At present,we haveno evidencefor dikesolderthan
1991]. With one significantmodification (conductive 113 Ma; however,by analogywith basinevolutionon the
heatingof asthenospheric mantleat the edgeof theplume), southwestAustralian margin [e.g., Marshall and Lee,
this model can account for the occurrence of MORB-like 1989], extension or transtensionin the Bengal Basin
tholeiiticbasaltson the easternIndianmargin.The location probablycommencedin the Late Jurassicand continued
of Sylhetlavason a northward extrapolation of thetrendof well into the Early Cretaceous.Hence, we infer that

NW SE

I Rajmahal
HillsBengal
Basin Bay
ofBengal
I

I asthenosphere mantleflowcoupled plume

toplume
flow flow
-' c 800 km

Figure 16. Cartoonillustratingan origin for the Rajmahalbasaltsby passivedecompressional meltingof normal
MORB-typemantlebeneaththestretched andthinnedlithosphereof theBengalBasin.Upwellingis suggested to have
occurreddueto viscouscouplingof flow in the MORB-source mantleat theboundaryof theKerguelenplumeto flow
in the conduitof the plume.The centerof the plumewasabout1000km to the southof easternIndia at the time of
Rajmahalvolcanism,assuming a platetectonicreconstruction
similarto thatproposed
by Miiller et al.

eruptionof the Rajmahal-Sylhetlavaswas precededand
accompaniedby stretchingand thinning of the Indian
crust.Lithosphericthinningwas mostprobablya product
of extensionassociated with the final stagesof continental
breakup,ratherthanconvectiveactivityassociated with the
Kerguelenplume.
The Bunbury and NaturalistePlateaulavas could have
an origin comparableto that suggestedhere for the
Rajmahalbasalts,assumingthat their Pb isotopicratios
largely reflect contaminationby continentalcrust. The
greaterdistanceof the Scottand Exmouthplateausfrom
the presumedcenter of the Kerguelenplume makes it
unlikely that lavas on theseplateauswere producedby
coupledflow of MORB-sourcemantleto flow in the 'tail'
of the Kerguelenplume. However, other west Australian
margin basalts,notably those on the Wallaby Plateau
[Colwellet al., 1994], couldhavebeengeneratedby these
means.Unfortunately,the Wallaby Plateaulavas are not
well-characterised in terms of age and chemical and
isotopiccomposition.Therefore,the relationshipof these
rocks to the Kerguelen Plateau remains uncertain. If
MORB-source mantle was induced to well up along the
rifted continentalmarginsby viscouscouplingto flow in
the Kerguelen plume, the Wallaby Plateau basalts are
predicted to have relatively MORB-like isotopic
signatures. Closerto the plume center,the compositionof
eruptedmagmasought to reflect more clearly the plume
isotopic signature. However, as we noted above, this
simple picture is complicatedby the likelihood that
strandedblocksof continentallithosphereunderliepartsof
the central and southernKerguelen Plateau. Moreover,
there is substantialdisagreementover what constitutesthe
'pure Kerguelenplume' isotopicsignature[e.g., Class et
at., 1996; Frey and Weis, 1996]. Until suchquestionsare
resolved,perhapsby drilling of volcanicbasementon the
northern Kerguelen Plateau, it will not be possible to
understandfully the link betweenrifted marginvolcanism
andoceanicplateauformationin the early IndianOcean.
Our model for the origin of Rajmahal-Sylhetbasalts
leadsus to questionwhetherthe developmentof the Early
CretaceousIndian Ocean province was similar to that of
other large igneousprovinces.The best-knowncontinental
margin flood basalts are those of the Tertiary North
Atlanticprovince(total volume-6.6 x 106 km3 [Eldholm
and Grue, 1994]). This igneous province is usually
attributedto the initiation of the Iceland hot spot [e.g.,
references of Fitton et at., 1997; Saunders et at., this
volume], whereaswe accountfor most of the Kerguelen
Plateau, Broken Ridge, and related continental margin
volcanismby the activity of a plume tail. Key featuresof
the North Atlantic province include(1) the occurrenceof
KENT ET AL. 177
62-56-m.y.-old basaltflows and associatedintrusiverocks
alongthe strikeof the rifted marginsfor distancesof more
than 2000 km [e.g., Saunderset al., this volume];(2) an
abundanceof basaltic intrusive centerson the opposing
continentalmargins;and (3) the existenceof post-56-Ma
linear basalticridges(the Greenland-Iceland and Faeroes-
Icelandridges)leadingaway from the rifted marginsto the
Icelandic Plateau. In comparison,the-130-100-m.y.-old
Indo-Australianmargin lavas appear to be confined to
isolatedzones (Figure 1), each zone ranging from 200 to
800 km in length. As in the Tertiary North Atlantic
igneousprovince,some of thesezonescorrespondto pre-
existingsedimentarybasins(e.g., the Bengal Basin), but
others (e.g., Naturaliste Plateau) clearly do not. Basaltic
intrusive complexes are unknown, either onshore or
offshoreof easternIndia and westernAustralia.Ridgesof
anomalously thick oceaniccrustakin to thosein the North
Atlantic extend offshore from the eastern Indian and East
Antarcticrifted margins,and appear,at leastin the caseof
the northernNinetyeast Ridge, to be younger than the
Early Cretaceousigneousprovince(the GaussbergRidge,
lying to the southof the KerguelenPlateau,has not been
studiedin detail and is of uncertainage andorigin).
The absenceof exposed intrusive centers in eastern
India and southwestAustraliais puzzling,but couldbe an
artifact of limited syn- and post-rift uplift of the
continentalmargins, with a consequentlack of deep
erosionandexposureof intrusivebodies.Alternatively,the
magmaticplumbingsystemcouldhavebeendifferentfrom
thaton theNorth Atlanticmargins,preventingthe ponding
of basalticmagmain the uppercrust.At present,thereare
very few constraints.The acquisitionof modem seismic
profiles acrossthe easternIndian margin would greatly
improve our knowledge of the crustal structure and
subsidencehistory of the Bengal Basin, and facilitate a
detailedcomparisonwith otherrifted margins.
Even from a brief comparisonsuchas this, it is evident
that developmentof the eastern Indian and southwest
Australianvolcanicrifted marginswas considerablymore
protractedthan that of the North Atlantic margins,where
the bulk of igneousactivitywas confinedto a periodof-7
m.y. [Saunderset at., thisvolume].Importantly,somewest
Australianmargin lavas, includingthe Bunbury basalts,
were eruptedat least9-15 m.y. beforethe main episodeof
KerguelenPlateauvolcanism.Constructionof the volcanic
sectionsof the centralandsouthernKerguelenPlateautook
place over a period of at least 29 m.y., suggestingan
established,steady-stateplume like that beneathpresent-
day Iceland. Each of these observationsfits with a non-
plume-headorigin for the Early CretaceousIndian Ocean
igneous
178 RAJMAHAL BASALTS

6. CONCLUSIONS compositionof the Kerguelen mantle plume at 116 Ma

[contraClasset al., 1993]. The fact that they havebeen so
The Rajmahalbasaltsof easternIndia are derivedfrom used highlightsthe dangersof assumingthat all basalts
MORB-like magmasthatprobablyformedat the boundary associated with a largeigneousprovincenecessarily havea
of a mantlehot spot.Most plate tectonicreconstructionsdeep-seatedplume source.Carefulstudyof otherprovinces
suggestthat this hot spotwas the Kerguelenplume.The may yet reveal that mantle upwelling coupledto flow
4øAr/39Aragesof lava samplesfrom the RajmahalHills within plumeconduitsis a potentmeansof producinglarge
andBengalBasin,anddikesfromthe Chotanagpur Plateau volumesof basaltwith a MORB-like chemicalandisotopic
andDamodarValley (mostly116-113 Ma) pointto a short signature,well away from the plumecenter.
periodof igneousactivityon the easternIndianmargin,
coincident with volcanism on the central and southern
Kerguelen Plateau and within a longer period of Acknowledgments. We are grateful to B. Kumar, D.
magmatismalong the continentalmargin of western Mukherjee,and B. K. Trivedi for assistancein the field, andNick
Australia(-130-100 Ma). There is evidence,in the form Marsh for supervisingXRF work. John Mahoney generously
providedPb isotopicdata for two of our Rajmahalsamples.Dan
of a small number of radiometric ages for continental
McKenzie gave valuablehelp with the inversionmodeling.We
basalts and old seafloor, for a broadly north-south thankVincent Saltersand Mick Storeyfor makingavailableto us
progressionof plateau-typevolcanism on the west their unpublished isotopic data for Indian Ocean basalts.
Australianrifted margin, suggesting that plate tectonic Reviewers Fred Frey and Doug Pyle gave constructiveand
developments weretheprincipalcontrolon thetimingand insightfulcommentson the first draftof the paper;FredFrey and
sitingof melt generationalongthe main Indo-Australian JohnMahoney providedhelpful commentson the seconddraft.
rift zone (including the location of Rajmahal and Researchin eastern India was supportedby the NERC (grant
KerguelenPlateauvolcanism). GT4/89/GS/55 to R. W. Kent) and the Universityof Leicester
The Rajmahalbasaltsare bestexplainedas moderate- (researchassociateship to R. W. Kent).
degreepartialmeltsproducedbeneaththinnedcontinental
REFERENCES
lithosphere.
In this respect,they are quiteunlike contem-
poraneous
high-degree
meltsgenerated
at Kerguelen
Plateau ODP Sites 749 and 750, the elemental Agrawal, J. K., and F. A. Rama, Chronologyof Mesozoic
volcanicsof India, Proc. Indian Acad. Sci., 84, 157-179, 1976.
compositions
of whichare consistent
with a near-ridge,Alibert, C., Mineralogy and geochemistryof a basaltfrom Leg
plume
tailorigin.Weinferfrominversion
modelingof Site 119, Site 738: Implicationsfor the tectonichistoryof the most
750 datathatthemantlepotentialtemperaturecloseto the southernpart of the Kerguelen Plateau, Proc. Ocean Drill.
centerof the Early CretaceousKerguelenhot spotwas Prog., Sci. Results,119, 293-298, 1991.
about1450øC.Thepotential temperatureof asthenosphere Anderson,D. L., Y.-S. Zhang, and T. Tanimoto, Plume heads,
beneaththeIndianmargin(about1000km fromtheplume continental lithosphere, flood basalts and tomography, in
centerat the time of Rajmahalvolcanism)wasprobably Magmatismand the Causes of ContinentalBreak-up, Spec.
about 13 50øC. Publ., 68, edited by B.C. Storey, T. Alabaster, and R. J.
Pankhurst,pp. 99-124, The GeologicalSociety,London,1992.
Isotopically,
the closestmarineanalogs to the least Amdt, N. T., and U. Christensen,The r61eof lithosphericmantle
crustally
contaminated Rajmahal basaltsarepresent-day in continental flood volcanism: Thermal and geochemical
Indian MORBs. Our chemicaland isotopicresultssuggest constraints, J. Geophys.Res.,97, 10,967-10,981,1992.
thatmostRajmahal basalts,
includingdikesemplaced to Arne, D.C., Phanerozoicexhumationhistoryof northernPrince
thewestandsouthwest of theRajmahalHills,assimilated CharlesMountains(East Antarctica),Antarct. Science,6, 69-
small melt-fractionsderived from crust similar to the 84, 1994.
Chotanagpurgranite-gneiss. After contamination, A. K., Petrogenesisand timing of volcanism in the
Baksi,
Rajmahal parent magmas underwent in the Rajmahal flood basalt province, northeasternIndia, Chem.
differentiation
Geol., 121, 73-90, 1995.
uppercrust.Volumetrically smalloccurrences of silicicBaksi, A. K., T. R. Barman, D. K. Paul, and E. Farrar,
volcanic and pyroclasticrocks attestto more-extreme WidespreadEarly Cretaceousflood basaltvolcanismin eastern
differentiationof tholeiitic liquids in areas where India: Geochemicaldata from the Rajmahal-Bengal- Sylhet
conditions allowedsmallhigh-levelmagmachambers to Traps,Chem.Geol., 63, 133-141, 1987.
become established. Ball, V., Geologyof the RajmahalHills, Mem. Geol. Surv.India,
Finally,wenotethatbecause theyappearto havebeen 13, 155-248, 1877.
derivedfrom a MORB-like source,the Rajmahalbasalts Biswas,B., Subsurfacegeology of West Bengal,India, in Proc.
cannot be used to estimate the Pb-Nd-Sr isotopic Symp.Dev. PetroleumResources,United Nations
 KENT ET AL. 179

Commissionfor Asia and the Far East, Mineral Resources Eldholm, O., and K. Grue, North Atlantic volcanicmargins:
DevelopmentSeries, 1O, editedby Anon., pp. 159-161, United Dimensionsand productionrates,•. Geophys.Res., 99, 2955-
Nations, New York, 1959. 2968, 1994.
Biswas,B., Results of explorationfor petroleumin the western Fitton, J. G., A.D. Saunders,L. M. Larsen,B. S. Hardarson,and
part of the Bengal Basin, India, in Proc. 2nd Symp. Dev. M. J. Norry, VolcanicrocksfromtheEastGreenland marginat
Petroleum Resources, United Nations Economic Commission 63øN: Composition,petrogenesisand mantle sources,Proc.
for Asia and the Far East, Mineral ResourcesDevelopment OceanDrill. Prog., Sci. Results,152, in press,1997.
Series,18, editedby Anon., pp. 241-250, United Nations,New Foucher, J.-P., X. Le Pichon, and J.-C. Sibuet, The ocean-
York, 1963. continenttransitionin the uniformstretchingmodel:R61eof
Brodie, J., D. Latin, and N. White, Rare-earth element inversion partial melting in the mantle,Phil. Trans. R. Soc.London,Set.
for melt distribution:Sensitivityandapplication,d. Petrol., 35, A, 305, 27-43, 1982.
1155-1174, 1994. Frey, F. A., and D. Weis, Temporalevolutionof the Kerguelen
Brune, J. N., and D. D. Singh, Continent-likecrustalthickness plume: Geochemical evidence from-38 to 82 Ma lavas
beneaththe Bay of Bengal sediments,Bull. Seismol.Soc.Am., formingthe NinetyeastRidge, Contrib.Mineral. Petrol., 121,
76, 191-203, 1986. 18-28, 1995.
Buck, W. R., Small-scaleconvectioninducedby passiverifting: Frey, F. A., and D. Weis, Reply to Discussion
by Classet al.,
The causeof uplift of rift shoulders,Earth Planet. Sci. Lett., Contrib. MineraløPetrol., 124, 104-110, 1996.
77, 362-372, 1986. Frey, F. A., W. Bo Jones,H. Davies, and D. Weis, Geochemical
Charvis,P., M. Recq, S. Operto, and D. Brefort, Deep structure and petrologic data for basaltsfrom Sites 756, 757, and 758:
of the northernKerguelenPlateauand hotspot-relatedactivity, Implicationsfor the origin and evolutionof the Ninetyeast
Geophys.J. Int., 122, 899-924, 1995. Ridge, ProcoOcean Drill. Prog., Sci. Results,121, 611-659,
Class, C., S. L. Goldstein, S. J. G. Galer, and D. Weis, Young 1991o
formationage of a mantleplume source,Nature, 362, 715-721, Frey, FoA., N.J. McNaughton,D. R. Nelson, J. R. deLaeter,and
1993.
Ro A. Duncan, Petrogenesisof the Bunbury Basalt, Western
Class, C., S. L. Goldstein, and S. J. G. Galer, Discussion of Australia: Interaction betw.•en the Kerguelen plume and
'Temporal evolution of the Kerguelen plume: Geochemical Gondwanalithosphere?Earth Planet. Sci. Lett., 141, 163-183,
evidence from -38 to 82 Ma lavas forming the Ninetyeast 1996.
Ridge', Contrib. Mineral. Petrol., 124, 98-103, 1996. Gautier, I., D. Weis, J.-P. Mennessier, P. Vidal, A. Giret, and M.
Colwell, J. B., P. A. Symonds,and A. J. Crawford,The natureof Loubet, Petrology and geochemistry of the Kerguelen
the Wallaby (Cuvier) Plateauand other igneousprovincesof Archipelago basalts (South Indian Ocean): Evolution of the
the west Australian margin, d. Aust. Geol. Geophys.,15, 137- mantlesourcesfrom ridge to intraplateposition,Earth Planet.
156, 1994. Sci. Lett., 100, 59-76, 1990.
Curray,J. R., and T. Munasinghe,Origin of the RajmahalTraps
Ghose, N. C., Geology, tectonics, and evolution of the
and 85øE Ridge: Preliminaryreconstructions of the traceof the
Chotanagpur granite-gneiss complex, eastern India, in
Crozethotspot,Geology,19, 1237-1240, 1991.
Structureand tectonicsof Precambrianrocks of India, Recent
Curray, J. R., and T. Munasinghe,Reply to Commentby Kent et
Researchesin Geology, 10, editedby S. Sinha Roy, pp. 211-
al., Geology,20, 958-959, 1992.
247, HindusthanPublishingCo., New Delhi, 1983.
Dalrymple, G. B., andM. A. Lanphere, 4øAr/39Ar spectra of
Gibling, M. R., F. M. Gradstein,I. L. Kristiansen,J. Nagy, M.
some undisturbedterrestrialsamples,Geochim. Cosmochim.
Sarti, and J. Wiedmann,Early Cretaceousstrataof the Nepal
Acta, 38, 715-738, 1974.
Himalayas: Conjugate margins and rift volcanism during
Davies, H. L., S.-S. Sun, F. A. Frey, I. Gautier, M. T. McCulloch,
Gondwana break-up, J. Geol. Soc. London, 151, 269-290,
R. C. Price, Y. Bassias, C. T. Klootwijk, and L. Leclaire, 1994.
Basaltbasementfrom the KerguelenPlateauand the trail of a
Goodman,R. J., The distributionof Ga and Rb in coexisting
DUPAL plume, Contrib. Mineral. Petrol., 103, 457-469, 1989.
groundmassand phenocrystphasesof some basic volcanic
DePaolo,D. J., Trace elementand isotopiceffectsof combined
rocks, Geochim. Cosmochim.Acta, 36, 303-317, 1972.
wallrock assimilation and fractional crystallization, Earth
Planet. Sci. Lett., 53, 189-202, 1981. Govindaraju, K., 1994 compilationof workingvaluesand sample
Deshmukh,S.S., S.C. Chakraborty,and M. V. N. Murthy, The descriptions for 383 geostandards,GeostandardsNewsletter,
18, 1-158, 1994.
origin of the shapes and sizes of the amygdules in the
pitchstoneand basaltflows from Taljhari and Berhait, Santhal Gradstein,F. M., F. P. Agterberg,J. G. Ogg, J. Hardenbol,P. van
Parganas,Bihar, Rec. Geol. Surv.India, 93, 45-50, 1964. Veen, J. Thierry, and Z. Huang, A Mesozoic timescale, J.
Dosso,L., H. Bougault, P. Beuzart,J.-Y. Calvez, and J.-L. Joron, Geophys.Res., 99, 24,051-24,074, 1994.
The geochemicalstructureof the SoutheastIndian Ridge, Green, D. H., and A. E. Ringwood, The genesis of basaltic
Earth Planet. Sci. Lett., 88, 47-59, 1988. magmas,Contrib. Mineral. Petrol., 15, 103-190, 1967.
Duncan,R. A., The age distributionof volcanismalongaseismic Hauri,E. H., J. A. Whitehead,andS. R. Hart,Fluiddynamicand
ridgesin the easternIndian Ocean,Proc. Ocean Drill. Prog., geochemicalaspectsof entrainmentin mantle plumes,J.
Sci. Results, 121,507-517, 1991. Geophys.Res.,99, 24,275-24,300,
180 RAJMAHAL BASALTS
Hofmann, A. W., and W. M. White, Ba, Rb, and Cs in the Earth's
mantle,Zeitsch.Naturforsch.38, 256-266, 1983.
Huppert, H. E., and R. S. J. Sparks,The fluid dynamicsof a
basalticmagma chamberreplenishedby influx of hot, dense
ultrabasic magma, Contrib. Mineral. Petrol., 75, 279-289,
1980.
Johnson,S. Y., and A.M. N. Alam, Sedimentationand tectonics
of the Sylhet Trough, northeasternBangladesh,U.S. Geol.
Surv. OpenFile Rep., 90-313, 42 pp., 1990.
Keen C. E., R. C. Courtney,S. A. Dehler, andM.-C. Williamson,
Decompressionmelting at rifted margins: Comparison of
modelpredictionswith the distributionof igneousrockson the
easternCanadian margin, Earth Planet. Sci. Lett., 121, 403-
416, 1994.
Kempton,P. D., and A. G. Hunter,A Sr-Nd-Pb-Oisotopestudy
of plutonicrocksfrom MARK, ODP Leg 153: Implicationsfor
mantle heterogeneityand magma chamber processes,Proc.
OceanDrill. Prog., Sci. Results,153, in press,1997.
Kent, R. W., Lithosphericuplift in easternGondwana:Evidence
for a long-lived mantle plume system?Geology, 19, 19-23,
1991.
Kent, R. W., Plume - lithosphere interaction: Petrology of
Rajmahalcontinentalflood basaltsand associatedlamproites,
northeastIndia, Ph.D. thesis,University of Leicester,1992.
Kent, R. W., and D. P. McKenzie, Rare-earth element inversion
models for basalts associatedwith the Kerguelen mantle
plume,Mineral. Mag., 58A, 471-472, 1994.
Kent, R. W., N. C. Ghose, P. R. Paul, M. J. Hassan, and A.D.
Saunders,Coal-magmainteraction:An integratedmodel for
the emplacementof cylindrical intrusions,Geol. Mag., 129,
753-762, 1992a.
Kent, R. W., N. C. Ghose,A.D. Saunders,and M. Storey,Basalt
stratigraphyof the RajmahalHills and adjacentareas,Bihar,
northeast India (abstract), in Mesozoic Magmatism of the
EasternMargin of India, Programmeand Abstracts,editedby
N. C. Ghose,pp. 2-4, PatnaUniversity,Patna,1992b.
Kent, R. Wo, M. Storey, and A.D. Saunders,Large igneous
provinces: Sites of plume impact or plume incubation?
Geology,20, 891-894, 1992c.
Kent, R. W., M. Storey,A.D. Saunders,N. C. Ghose,andP. D.
Kempton,Commenton 'Origin of the RaimahalTraps and
85øERidge:Preliminaryreconstructions of the Crozethotspot'
by J. R. Curray and T. Munasinghe,Geology,20, 957-958,
1992d.
Kent, R. W., A.D. Saunders,M. Storey, and N. C. Ghose,
Petrologyof Early Cretaceous flood basaltsand dykesalong
the rifted volcanicmarginof easternIndia, d. Southeast
Asian
Earth Sci., 13, 95-111, 1996.
Khan,A. H., and J. Azad, The geologyof Pakistangasfields,in
Proc. 2nd Symp.Der. PetroleumResources,UnitedNations
EconomicCommission for Asia and the Far East, Mineral
ResourcesDevelopmentSeries, 18, pp. 275-282, United
Nations, New York, 1963.
Klootwijk, C. T., Palcomagnetism of the Upper Gondwana
RaimahalTraps,northeast
India, Tectonophysics,
12, 449-467,
1971.
Leclaire, L., Y. Bassias, M. Denis-Clocchiatti, H. Davies, I.
Gautier,B. Gensous,P.-J. Giannesini,J. S•goufin, M. Tesson,
and J. Wannesson, Lower Cretaceousbasalt and sediments
from the Kerguelen Plateau, Geo-Marine Lett., 7, 169-176,
1987.
Lightfoot,P. C., Isotopeand trace elementgeochemistry of the
SouthDeccan lavas, India, Ph.D. thesis,The Open University,
1985.
Ludden,J. N., and B. Dionne, The geochemistry of oceaniccrust
at the onsetof rifting in the Indian Ocean,Proc. OceanDrill.
Prog., Sci. Results,123, 791-799, 1992.
Mahoney,J. J., Isotopicand chemicalstudiesof the Deccanand
RajmahalTraps,India: Mantle sourcesandpetrogenesis, Ph.D.
thesis,University of California, SanDiego (Scripps),1984.
Mahoney, J. J., and K. J. Spencer,Isotopic evidencefor the
originof the Manihiki and OntongJavaplateaus,Earth Planet.
Sci. Lett., 104, 196-210, 1991.
Mahoney,J. J., J. D. Macdougall,G. W. Lugmair,A. V. Murali,
M. Sankar Das, and K. Gopalan, Origin of the Deccan Trap
flows at Mahabaleshwarinferredfrom Nd and Sr isotopicand
chemicalevidence, Earth Planet. Sci. Lett., 60, 47-60, 1982.
Mahoney, J. J., J. D. Macdougall, G. W. Lugmair, and K.
Gopalan,Kerguelenhot spot sourcefor RajmahalTraps and
NinetyeastRidge?Nature, 303, 385-389, 1983.
Mahoney, J. J., J. H. Natland, W. M. White, R. Poreda,S. H.
Bloomer, Ro Lo Fisher, and A. N. Baxter, Isotopic and
geochemicalprovincesof the westernIndian Oceanspreading
centers,d• Geophys.Res., 94, 4033-4052, 1989.
Mahoney, J. J., A. P. le Roex, Z. Peng, R. L. Fisher,and JoH.
Natland, Southwesternlimits of Indian Ocean Ridge mantle
andthe originof low 2ø6pb/2ø4pb
mid-ocean
ridgebasalt:
Isotope systematicsof the Central SouthwestIndian Ridge
(17ø-50øE),d• Geophys.Res.,97, 19,771-19,790,1992.
Mahoney, J. J., M. Storey, R. A. Duncan,K. J. Spencer,and M.
Pnngle, Geochemistryand age of the OntongJavaPlateau,in
The Mesozoic Pacific: Geology, Tectonicsand Volcanism,
Geophyso Monogr• Set., vol. 77, edited by M. S. Pringle, W.
Sager, W. Sliter, and S. Stein,pp. 233-261, AGU, Washington,
D.C., 1993o
Mahoney, J. J., W. B. Jones,F. A. Frey, V. J. M. Salters,D. G.
Pyle, and H. L. Davies, Geochemicalcharacteristics of lavas
from Broken Ridge, the NaturalistePlateau and southernmost
Kerguelen Plateau: Cretaceousplateau volcanism in the
southeastIndian Ocean, Chem. Geol., 120, 315-345, 1995.
Mahoney,J. J., W. M. White,B. G. J. Upton,C. R. Neal,andR.
A. Scrutton,BeyondEM-1: LavasfromAfanasy- Nikitin Rise
andthe CrozetArchipelago,IndianOcean,Geology,24, 615-
618, 1996.
Marshall,J. F., and C. S. Lee, Basinframeworkand resource
potential
of theAbrolhos
sub-basin,
in Aust.BureauMineral.
Res.,Geol. Geophys.,Yearbook1988-1989,editedby K. H.
Wolf and A. G. L. Paine,pp. 63-67, Australia,Bureauof
Mineral Resources,Canberra,1989.
Mazumdar,S. K., Crustalevolutionof the Chotanagpur
Gneissic
Complex
andtheMicaBeltof Bihar,Mem.Geol.Soc.India,8,
49-83,
 KENT ET AL. 181

McDougall, I., and M. W. McElhinny, The Rajmahal Traps of Pringle, M. S., M. Storey, and J. Wijbrans,4øAr/39Ar
India- K-Ar ages and paleomagnetism,Earth Planet. Sci. geochronology of Mid-Cretaceous Indian Ocean basalts:
Lett., 9, 371-378, 1970. Constraintson the origin of large flood basalt provinces
McKenzie, D., and R. K. O'Nions, Partial melt distributionsfrom (abstract),Eos Trans.AGU, 7,5,44, Fall MeetingSuppl.,728,
inversion of rare-earth element concentrations,d. Petrol., 32, 1994.
1021-1091, 1991. Pyle, D. G., D. M. Christie,J. J. Mahoney, and R. A. Duncan,
McKenzie, D., and R. K. O'Nions, The sourceregionsof ocean Geochemistryand geochronologyof ancientSoutheastIndian
island basalts,d. Petrol., 36, 133-159, 1995. Oceanand SouthwestPacificOceanseafloor,d. Geophys.Res.,
Middlemost, E. A. K., D. K. Paul, and I. R. Fletcher, Minette- 100, 22,261-22,282, 1995.
lamproiteassociationfrom the Indian Gondwanas,Lithos, 22, Raja Rao, C. S., and A. Purushottam,Pitchstoneflows in the
31-42, 1988. Rajmahal Hills, Santhal Parganas,Bihar, Rec. Geol. Surv.
Mukhopadhyay,M., R. K. Verma, and M. H. Ashraf, Gravity India, 91, 341-348, 1963.
field and structuresof the Rajmahal Hills: Examples of the Reston,T. J., Evidencefor extensionalshearzonesin the mantle,
Paleo-Mesozoic continental margin in eastern India, offshoreBritain,andtheir implicationsfor the extensionof the
Tectonophysics, 131,353-367, 1986. continentallithosphere,Tectonics,12, 492-506, 1993.
Mtiller, R. D., J.-Y. Royer, and L. A. Lawver, Revised plate Rock,N.M. S., B. J. Griffin, A.D. Edgar,D. K. Paul, andJ. M.
motionsrelative to the hot spotsfrom combinedAtlantic and Hergt, A spectrumof potentiallydiamondiferouslamproites
Indian Oceanhot spottracks,Geology,21,275-278, 1993. and minettes from the Jharia coalfield, eastern India, d.
Negi, J. G., O. P. Pandey,and P. K. Agrawal, Super-mobilityof Volcanol. Geotherm.Res., ,50, 55-83, 1992.
hot Indian lithosphere,Tectonophysics, 131, 147-156, 1986. Royer, J.-Y. and M. F. Coffin., Jurassicto Eoceneplate tectonic
Ogg, J. G., K. Kodama, and B. P. Wallick, Lower Cretaceous reconstructionsin the Kerguelen Plateauregion, Proc. Ocean
magnetostratigraphyand paleolatitudes off NW Australia Drill. Prog., Sci. Results,120, 917-928, 1992.
(ODP Site 765 and DSDP Site 261, Argo AbyssalPlain, and Royer, J.-Y., and D. T. Sandwell,Evolution of the easternIndian
ODP Site 766, GascoyneAbyssalPlain), Proc. OceanDrill. Ocean since the Late Cretaceous: Constraints from GEOSAT
Prog., Sci. Results,123, 523-548, 1992. altimetry,d. Geophys.Res.,94, 13,755-13,782, 1989.
Operto,S., andP. Charvis,KerguelenPlateau:A volcanicpassive Salters,V. J. M., M. Storey,J. H. Sevigny,and H. Whitechurch,
margin fragment?Geology,23, 137-140, 1995. Trace elementand isotopiccharacteristics of Kerguelen-Heard
Operto, S., and P. Charvis, Deep-structureof the southern Plateaubasalts,Proc. OceanDrill. Prog., Sci. Results,120, 55-
KerguelenPlateau(southernIndian Ocean)from ocean-bottom 62, 1992.
seismometerwide-angle seismicdata, d. Geophys.Res., 101, Sarkar, A., A. K. Datta, B.C. Poddar, V. K. Kollapuri, B. K.
25,077-25,103, 1996. Bhattacharyya,and R. Sanwal, Geochronologicalstudies of
Pantulu, G., J. D. Macdougall, K. Gopalan, and P. Mesozoicigneousrocksfrom easternIndia, d. SoutheastAsian
Krishnamurthy,Isotopic and chemicalcompositionsof Sylhet Earth Sci., 13, 77-81, 1996.
Traps basalts:Links to the RajmahalTraps and Kerguelenhot Sarkar, S.S., S. K. Nag, and S. Basu Mallik, The origin of
spot(abstract),Eos Trans.AGU, 73, 14, Fall Meeting Suppl., andesitefrom Rajmahal Traps, easternIndia: A quantitative
328, 1992.
evaluationof a fractional crystallisationmodel, d. Volcanol.
Pederson, T., and H. E. Ro, Finite-duration extension and Geotherm. Res., 37, 365-378, 1989.
decompressionmelting, Earth Planet. Sci. Lett., 113, 15-22,
1992.
Saunders, A.D., Geochemistryof basaltsfrom the Nauru Basin,
Deep SeaDrilling ProjectLegs61 and89: Implicationsfor the
PoornachandraRao, G. V. S., J. Mallikharjuna Rao, and M. V.
origin of oceanicflood basalts,Init. Repts.Deep Sea Drill.
SubbaRao, Palaeomagnetic and geochemicalcharacteristics
of
Proj., 89, 499-518, 1986.
the Rajmahal Traps, easternIndia, d. SoutheastAsian Earth
Sci., 13, 113-122, 1996. Saunders, A.D., M. Storey,I. L. Gibson,P. Leat, J. Hergt, andR.
Poornachandra Rao, G. V. b., M. V. Subba Rao, and J. N. Thompson, Chemicaland isotopicconstraints on the origin
of basaltsfrom NinetyeastRidge, Indian Ocean:Resultsfrom
Mallikharjuna Rao, Geochemical and palaeomagnetic
DSDP Legs 22 and 26 and ODP Leg 121, Proc. OceanDrill.
signaturesin Sylhet Trap flood basaltvolcanism:Preliminary
Prog., Sci. Results,121,559-590, 1991.
results,Geol. Mag., 130, 163-184, 1993.
Powell, C. McA., S. R. Roots, and J. J. Veevers, Pre-break-up Saunders,Ao D., M. Storey, R. W. Kent, and M. J. Norry,
continental extension in East Gondwanaland and the early Consequences of plume-lithosphere interactions, in
opening of the eastern Indian Ocean, Tectonophysics, 1,55, Magmatism and the Causes of Continental Break-up, Spec.
261-283, 1988. Publ. 68, edited by B.C. Storey, T. Alabaster, and R. J.
Pringle, M. S., Age-progressivevolcanism in the Musicians Pankhurst,pp. 40-60, The GeologicalSociety,London, 1992.
seamounts:A test of the hot spot hypothesisfor the Late Sen, G., and T. H. Cohen, Deccan intrusion, crustal extension,
CretaceousPacific, in The Mesozoic Pacific: Geology, doming,and the size of the Deccan- R6unionplume head,in
Tectonicsand Volcanism,Geophys.Monogr. Ser., vol. 77, Volcanism,edited by K. V. Subbarao,pp. 201-216, Wiley
editedby M. S. Pringle, W. Sager,W. Sliter, and S. Stein, pp. Eastern, New Delhi, 1994.
187-215, AGU, Washington,D.C., 1993. Sengupta,S., Geologicaland geophysicalstudiesin western
182 RAJMAHAL BASALTS

of BengalBasin,India, Bull. Am. Assoc.Petrol. Geol., 50, Todt,W., R. A. Cliff, A. Hanser,andA. W. Hofmann,2ø2pb+
1001-1017, 1966. 2øspb
doublespikefor leadisotopic
analyses,
TerraCognita,
Sengupta,S., UpperGondwanastratigraphy andpalaeobotany of 4, 209, 1984.
RaimahalHills, Bihar, India, Geol. Surv.India Mort,gr., 98, von Rad, U., and N. F. Exon, Mesozoic sedimentaryand volcanic
(PalaeontologicaIndica), 180pp., GSI, Calcutta,1988. evolution of the starved passive continental margin off
Sherwood,G. J., and S. BasuMallik, A palaeomagnetic
androck northwest Australia, in Studies in Continental Margin
magneticstudy of the northernRajmahalvolcanics,Bihar, Geology,Mem. 34, editedby J. S. Watkins, C. L. Drake, and
India, J. SoutheastAsian Earth Sci., 13, 123-131, 1996. R. E. Sheridan, pp. 253-281, American Association of
Shukla,R., Studyof SylhetTrap flows in Balat area,KhasiHills PetroleumGeologists,Tulsa, 1983.
District,Meghalaya(abstract),in MesozoicMagmatismof the Wedepohl, K. H., The compositionof the continentalcrust,
EasternMargin of India, Programme andAbstracts, editedby Mineral. Mag., 58A, 959-960, 1994.
N. C. Ghose,pp. 13, PatnaUniversity,Patna,1992. Weis, D., and F. A. Frey, Isotopegeochemistryof Ninetyeast
Sparks,D. W., andE. M. Parmentier, Melt extractionfrom the Ridge basalts:Sr, Nd, and Pb evidencefor the involvementof
mantlebeneathspreading centers,Earth Planet.Sci.Lett., 105, the Kerguelenhot spot,Proc. OceanDrill. Prog., Sci. Results,
368-277, 1991. 121, 591-610, 1991.
Storey,M., A.D. Saunders,J. Tamey, I. L. Gibson,M. J. Norry, Weis, D., and F. A. Frey, R61e of the Kerguelen plume in
M. F. Thirlwall, P. Leat, R. N. Thompson,andM. A. Menzies, generating the eastern Indian Ocean sea floor, J. Geophys.
Contamination of Indian Ocean asthenosphereby the Res., 101, 13,831-13,849, 1996.
Kerguelen- Heardmantleplume,Nature,338, 574 -576, 1989. Weis, D., Y. Bassias, I. Gautier, and J.-P. Mennessier, DUPAL
Storey,M., A.D. Saunders,J. Tarney,P. Leat, M. F. Thirlwall, anomaly in existence 115 Ma ago: Evidence from isotopic
R. N. Thompson, M. A. Menzies, and G. Marriner, studyof the KerguelenPlateau(southIndianOcean),Geochim.
Geochemicalevidencefor plume-mantleinteractionsbeneath Cosmochim. Acta, 53, 2125-2131, 1989.
Kerguelenand HeardIslands,Indian Ocean,Nature, 336, 371- Weis, D., F. A. Frey, H. Leyrit, and I. Gautier, Kerguelen
374, 1988. Archipelagorevisited:Geochemicaland isotopicstudyof the
Storey, M., R. W. Kent, A.D. Saunders,J. Hergt, V. J. M. SoutheastProvince lavas, Earth Planet. Sci. Lett., 118, 101-
Salters,H. Whitechurch,J. H. Sevigny, M. F. Thirlwall, P. 119, 1993.
Leat, N. C. Ghose,and M. Gifford, Lower Cretaceousvolcanic White, R. S., and D. McKenzie, Magmatismat rift zones:The
rocks on continental•narginsand their relationshipto the generationof volcaniccontinentalmarginsandflood basalts,or.
KerguelenPlateau,Proc. OceanDrill. Prog., Sci. Results,120, Geophys.Res., 94, 7686-7729, 1989.
33-53, 1992. White, R. S., and D. McKenzie, Mantle plumesandflood basalts,
Sun, S.-S., and W. F. McDonough, Chemical and isotopic J. Geophys.Res., 100, 17,543-17,585, 1995.
systematicsof oceanic basalts: Implications for mantle White, R. S., D. McKenzie, and R. K. O'Nions, Oceanic crustal
composition and processes,in Magmatism in the Ocean thickness from seismic measurements and rare-earth element
Basins•Spec.Publ., 42, editedby A.D. Saundersand M. J. inversions,J. Geophys.Res.,97, 19,683-19,715, 1992.
Norry, pp. 313-345, The GeologicalSociety,London,1989. White, W. M., and A. W. Hofmann, Sr and Nd isotope
Sushchevskaya, N.M., G. V. Ofchinnikova,A. Y. Borisova,B. geochemistryof oceanicbasaltsand mantleevolution,Nature,
Vo Belyaszsky•JoVasilyeva,and L. K. Levsky,Geochemical 296, 821-825, 1982.
heterogeneity of Afanasy-NikitinRise magmatism,northeast Whitechurch,H., R. Montigny, J. Sevigny,M. Storey,and V. J.
Indian Ocean(in Russian),Petrologia,in press,1997. M. Salters,1992, K-At and 4øAr/39Arages of central
Talukdar, S.C., Rhyolite and alkali basaltfrom the SylhetTraps, Kerguelen Plateau basalts, Proc. Ocean Drill. Prog., Sci.
Khasi Hills, Assam, Curt. Sci., 36, 238-239, 1967. Results 120, 71-78, 1992.
Talukdar, S.C., and M• V. N. Murthy, The Sylhet Traps, their
tectonichistory,andtheir bearingon problemsof Indianflood N. C. Ghose,Departmentof Geology,PatnaUniversity,Patna
basaltprovinces,Bull. Volcanol.,35, 602-618, 1970. 800 005, Bihar, India.
Tardun., J. A., Brief polarity reversal during the Cretaceous P. D. Kempton, NERC Isotope Geosciences Laboratory,
Normal Polarity Superchron,Geology,18, 683-686, 1990. KingsleyDunhamCentre,Keyworth,Nottingham NG12 5GG,
Thurow• J., and U. von Rad, Bentonites as tracers of earliest United Kingdom.
Cretaceouspost-break-upvolcanismoff northwestern Australia R. W. Kent and A.D. Saunders,Departmentof Geology,
(Legs 122 and 123), Proc. Ocean Drill. Prog., ScioResults, Universityof Leicester,UniversityRoad,LeicesterLE1 7RH,
123, 89-110, 1992• UnitedKingdom. E-Mail:
 The OntongJavaPlateau

CliveR.NealI John
J.Mahoney
9 2,Loren
W.Kroenke
2,Robert
A Duncan
3,andMichael
G.Petterson
4

The Alaska-sizeOntongJavaPlateau(OJP) in the southwest Pacificis the largest

of the world's large igneous provinces and formed entirely in an oceanic
environment.Limited sampling of the upper levels of basaltic basementreveals
stronglybimodalagesof •122 Ma and•90 Ma. Geochemicalsignatures indicatetwo
isotopicallydistinct,ocean-island-likemantle-sourcetypes for the 122 Ma basalts
and that the 90 Ma source was almost identical to one of the 122 Ma sources,
stronglysuggestingthat a singlemantleplume causedboth eruptiveevents.In the
125-90 Ma period,the OJP appearsto have been locatednearthe Pacific Plate Euler
polesandthusmovedlittle relativeto a postulatedhotspotat about42øS, 159øW;the
early phaseof emplacementprobablyalso occurrednear a spreadingcenter,but
substantialvolumeswere emplacedoff-axis.The easternlobeof the plateauappears
to have been rifled shortlyafter 90 Ma. Incompatibleand major elementdata are
consistentwith 20-30% polybaricpartialmeltingof a peridotitesource,beginningin
the garnetstabilityfield and continuingin the spinelfield. Most existingbasaltic
basementsamplesappearto have experienced30-45% of crystalfractionation;the
resulting cumulateswould be wehrlitic to pyroxenitic in composition,with an
averagedensity
of •3.25 g cm-3.We conclude
thatthesecumulates formmuchof
theplateau's
highvelocity(•7.6 km s-l) basalcrustallayer.Therelativelyhigh
densityof this layer appearsto havepreventedemergenceof muchof the OJP above
sea level despitea total crustalthicknessexceeding30 km. Although the deepest
levels of the crust could be eclogitized (further increasing density), post-
emplacementsubsidence of the plateauhasprobablybeentemperedby the presence
of a roughly 85-km-thick melt-depletedmantle root with a relatively low density
(_<3.25
g cm-3)for mantlematerial.
Estimates
of partialmeltingdeduced
fromthe
apparentthicknessof the mantleroot imply that the OJP formedby 17-31% partial
melting,in excellentagreementwith geochemicalmodeling.

1. INTRODUCTION
•Departmentof Civil Engineeringand GeologicalSciences,
Universityof Notre Dame, Notre Dame, Indiana Oceanic plateausare now generally recognizedas the
2School
of OceanandEarth ScienceandTechnology,University counterpartsof continentalflood basaltsandthick volcanic
of Hawaii at Manoa, Honolulu, Hawaii
sequences found at many passive plate margins.
3Collegeof Oceanicand Atmospheric
Sciences,OregonState
Collectively, such areas have been termed large igneous
University, Corvallis, Oregon
4BritishGeological
Survey,MuchisonHouse,WestMainsRoad,
provincesor LIPs [e.g., Coffin and Eldholm, 1991]. Since
Edinburgh,UnitedKingdom the mid-1980s,a consensus hasgrownamonggeochemists
and geodynamicists that most oceanicplateausare created
at hotspots;more recently,the larger plateaushave been
Large IgneousProvinces:Continental,Oceanic,andPlanetary ascribed to the initial "plume-head" stage of hotspot
Flood Volcanism development[e.g., Richardset al., 1991; Saunderset al.,
GeophysicalMonograph100 1992; Kent et al., 1992]. Many laboratory experiments,
Copyright 1997 by the AmericanGeophysicalUnion alongwithnumerical
modeling,
suggest
thatmantle
183
184 THE ONTONG JAVA PLATEAU

initiationcausedby a temperatureinstabilityat a boundary 2. PHYSICAL FEATURES AND GROSS STRUCTURE

layer resultsin a large, inflated"plume-head"trailedby a OF THE OJP
narrow feeder conduit or "tail" [e.g., Whitehead and
Luther, 1975; Olson and Singer, 1985; Griffiths and The Alaska-size OJP covers an area of more than 1.5 x
Campbell,1990;Neaveland Johnson,1991]. In plume- 106 km2 and consistsof two parts: the main, or high,
head models, emplacementof the LIP occurs in a plateauin the west andnorthandthe easternlobeor salient
geologicallyshortperiodof only a few million yearsor (Figure2). The plateausurfacerisesto depthsof •1700 m
less[e.g.,Richardset al., 1989]. Thisis a consequence
of in the centralregion of the high plateaubut lies generally
extensivedecompressional meltingwhenthe risingplume between2 and 3 km depth. The OJP is boundedto the
head, with temperaturesarguedto averageup to 300øC northwestby the Lyra Basin, to the north by the East
above those of ambient upper mantle, approachesand Mariana Basin,by the Nauru Basinto the northeastandthe
spreadsout at the baseof the lithosphere,where it may Ellice Basinto the southeast(Figure 1). Along its southern
reacha diameterof 2000 km or more [e.g., Campbelland and southwestern boundaries the OJP has collided with the
Griffiths, 1990; Hill et al., 1992; Ribe et al., 1995; Kincaid Solomon Islands arc and it now lies at the junction
et al., 1996]. For the largestplateaus,the enormoussizeof between the Pacific and Australian plates, extending
the sourcerequiredsuggests an origin of tl•e plumein the eastwardto the Vitiaz arc-trenchsystem.
lower mantle, possibly at the core-mantleboundary; a Althoughmuch of the surfaceis relativelysmooth,the
plumeheadoriginatingat the 660-kmmantlediscontinuity, top of the plateau is punctuatedby several sizable
for example,would not reachsufficientsize to accountfor seamounts, includingOntongJavaatoll (the largestatoll in
the volumeof magmarequiredto generatethe largerLIPs the world), and physiographyaroundthe marginsof the
[e.g., Coffin and Eldholm, 1993]. plateauis complicated[e.g., Kroenke,1972;Bergeret al.,
Severalmajor oceanicplateausare locatedin the Pacific 1993]. In many areas,the basementcrust is coveredwith
Oceanbasin, includingHessRise, ShatskyRise, the Mid- pelagic sediments>1 km thick, which are thickest where
Pacific Mountains, the Manihiki Plateau, and the world's the plateauis at its mostshallow[e.g.,Mayer et al., 1991].
largest,the OntongJavaPlateau(OJP) (Figure 1). Unlike Roughbasementtopography onthemarginsof the OJPhas
their continentaland continental-marginanalogs,plateaus beeninterpretedas extensionalhorstand grabenstructures
formed in intraoceanicsettingsoffer a meansof studying thatpredatemuchof the sediment cover[e.g.,Ewinget al.,
LIP sourcesand structureand testingmodelsof LIP origin 1968; Kroenke et al., 1971; Kroenke,1972; Berger and
withoutthe oftenconsiderable complicationscausedby the Johnson,1976; Hagen et al., 1993].
presence of continental lithosphere.Unfortunately, the Crustal thickness on much of the high plateau is
crustalbasementsof oceanicplateausremainvery poorly considerable even in comparisonto otherplateaus. Non-
sampled,in general,becausethey are largely submerged seismicmethods(e.g., gravitydata)providea lower limit
and buried beneath thick covers of marine sediments. Of of 25 km [Cooper et al., 1986; Sandwell and Renkin,
the plateaus in the Pacific, the OJP is by far the best- 1988], whereas seismic and combined seismic-gravity
sampled,with basementpenetrationsat Deep Sea Drilling evidence indicates crustal thicknesses in the 30-43 km
Project (DSDP) Site 289 (9 m), and Ocean Drilling range, with an estimatedaverage around 36 km [e.g.,
Program (ODP) sites 803 (26 m) and 807 (149 m). Furumoto et al., 1970, 1976; Murauchi et al., 1973;
Moreover, tectonicallyuplifted portionsof OJP basement Hussonget al., 1979; Miura et al., 1996; Richardsonand
are exposedabove sealevel in the easternSolomonIslands Okal, 1996; Gladczenkoet al., 1997]. A 36-km thickness
of SantaIsabel, Ramos,Malaita, and Ulawa (Figures2 and translatesto a volume of > 5 x 107km3 [Mahoney, 1987;
3). Some of the thickest exposureson these islands Coffin and Eldholm, 1993]. The maximumextentof OJP-
recently have been sampled in detail; studies are still related volcanism may be even greater, as the Early
underway,but the resultsnow availableprovide a much Cretaceouslavasfilling the Nauru Basinto the northeastof
better,thoughoften surprising,pictureof the OJP thanwas the OJPandsimilarlavasin the EastMarianaandPigafetta
available only a few years ago. This paper presentsa basinsto the north have been proposedto be closely
review of geochemical,geochronological, and geophysical related to the OJP [e.g., Castillo et al., 1991, 1994;
knowledgeaboutthe plateauand additionallyreportsnew Gladczenkoet al., 1997 and referencestherein]. Indeed,
resultsof a detailed geochemicalinvestigationof the top by analogywith some continentalflood basaltprovinces
2-3 km of volcanic basementexposedon the island of [see Ernst and Buchan, this volume], the lavas in these
Malaita. Particular attention is given to the origin and basins might reflect one or more giant radiating dike
effect of"hidden" cumulatesin the crustof the plateau. swarmsassociated with emplacement of the
 NEAL ET AL. 185

40 ø
N

ess Rise

30 ø
N

20 ø acific
N
Mountains

East Mariana Basin

10 ø
N

Jav• __

Stewart Basi•
Ellice ...........................
ß__Basin .................
...

ß.-:'..-•-.-.....-_...-...
......

140øE 150øE 160øE 170øE 180 ø 170øW 160øW

Figure 1. Map of the west-central

Pacificshowinglocationsof principaloceanicplateaus(shaded)andmagnetic
anomalylineations[afterNakanishi et al., 1992].

Resultsof recentlycompletedseismicstudiesof crustal and sonobuoy wide-angle reflection/refractionmea-

structurehave,asyet, beenpresentedonly in abstractform
[e.g.,Miura et al., 1996]. Nevertheless,
a simplecrustal
modelcanbe constructed frompre-1995seismicrefraction
surements,
despitethe limitationsof the data(Figure2).
Broadly, the OJP appearsto have a crustal seismic
structureresemblingthat of normalPacificoceanic
186 THE ONTONG JAVA PLATEAU

oXl C'CB
øW WQ Q X'
5.0,,----J-- I I 5.8 5.9 _--
10 - 2C •---- _

7.0 "" - •
7.0
7.0
• 20--
- (7.5)• 6.9 -
o
10 ø ....,-
N
'• :30
-
(7.5)I 7.6 ß

7.7 ,'
(7.5) 3b ''

- • ." _
40 -

, 8.6

10 ø
S

150øE 160øE 170øE 160øE

Figure 2. E-topo 5 bathymetricmap of the westernequatorialPacific Basin (centeredon the OJP and Nauru Basin,
shaded
by satellite-derived
gravityfabricilluminated
fromthenorth[afterSmithandSandwell,
1995a,
b]). The
locationsof drill sites 807, 803, 289, and 288 (the latter did not reach basement)are shown as triangles. Contours
representthe depthto the top of Layer 3A in the high plateau(contourinterval= 2 km). Deep crustalseismicrefraction
lines are labeledA-A*, B-B*, and C-C* [from Furumotoet al., 1970] and P, Q, and R [from Fururnotoet al., 1976].
Sonobuoyrefractionlines [from Hussonget al., 1979] are representedby isolateddots. Inset is compositecrustal
cross-section
X-X'. P-wavevelocities(in km s-]) are from Furumotoet al. [1970, 1976]; velocitiesin parentheses
havebeeninsertedfollowingHussonget al. [1979].

but eachlayer is abnormallythickenedby up to a factorof
five [Hussonget al., 1979]. Figure 2 showsa composite
section(XX') acrossthe highplateauwhich depictsa thick
crustallens,more than 40 km thick where it underpinsthe
center of the OJP. Such a crustal configuration is
consistentwith a structure-contour
map showingthe top of
Layer 3A on the high plateaualsopresentedin Figure2.
This is based,in part, on datafrom the refractionsurveys
of Furumotoet al. [1970, 1976], but usesonly the most
reliable, upper crustaldata, all from first arrivals; it also
uses the later sonobuoymeasurements summarizedby
Hussonget al. [1979], which showedLayer 2B velocities
to be remarkablyuniform. The upper crustal section
includesthe water column,sediments(containing
 NEAL ET AL. 187

-•
15•
øE 16•
øE 50 km
• PROVIIvo•
...©•'•C
HOISEUL
ß•ee•e •
. • S• O•e•
JMOS
S•
I LAITA
-- ".._ •
V••
• .......... ß ß m•V e
NEW
GEORGIA

o ,SNDS
• S•NDS .'
SA.C.,SO.A
Figure 3. Division of the SolomonIslandschaininto geologicallydistinctregionsof the Pacific,Volcanic,and
CentralProvinces[e.g.,Coleman,1966, 1976;ColemanandPackham,1976].

high-velocity horizons), Layer 2B (composedof dense, feature,togetherwith the OJP's crustalvelocity structure,
high velocity basalts),and Layer 2C (dolerites?). Depths possiblevery low-amplitudemagnetic anomaliesacross
to the top of Layer 3A (with P-wave velocitiesin the 6.9- the northernhalf of the high plateau,and the very sparse
7.0 km s-• range, appropriatefor high-level gabbrosor age data for plateaubasementlavasavailableat the time,
possibly mafic granulites [e.g., Rudnick and Jackson, led Hussonget al. [ 1979] to proposean origin of the OJP
1995]) range from lessthan 10 km aroundthe plateau's at an unusually active, WNW-trending, slow-spreading
edgesto more than 16 km in a broadcentralregion of the ridge over a periodof severaltensof millionsof years. An
high plateau. The general shape is an elongate, ESE- age progressionacrossthe plateau was implicit in this
WNW-trending depression,in much of which Layer 2 is interpretation.Variationson this hypothesisproposedthat
over 12 km thick. the OJP was formed at a migrating triple junction or
On the central and westernparts of the high plateau, transform during a period of ridge jumping and heavy
Layer 3B velocitiesare established to be ratherhigh at 7.6- volcanism [Winrefer, 1976; Hilde et al., 1977; Taylor,
7.7 km s-• (Figure 2). This range is appropriatefor 1978]; however, explanationsfor why volcanism was
granular gabbros[Schaeferand Ned, 1994; Farnetani et heavy were lacking. On the basisof preliminaryisotopic
al., 1995, 1996] or garnetgranulites[Nixon and Coleman, and elemental data, together with the geophysical
1978], althoughHoutz and Ewing [1976] consideredsuch evidence, Mahoney [1987] proposedthat the OJP was
velocitiesto be compatiblewith deep oceanicbasaltsor formedby a largeridge-centered or near-ridgeplume,and
serpentinizedperidotites. High sub-Moho velocities of discussedthe possibilitythat this plume was the early
8.4-8.6 km s-l, appropriatefor eclogite[Ned and Taylor, Louisville hotspot (now located at •53øS [Wesseland
1989; Rudnick and Jackson, 1995; Saunderset al., 1996] Kroenke, 1997]).
were detectedin the northwestand southwestportionsof Subsequentmajor and trace elementdata for basement
the plateau. lavas from drill sites on the OJP and from outcropsin
Malaita and Santa Isabel (summarizedbelow) were found
3. TECTONIC SETTING OF OJP EMPLACEMENT to be consistentwith the plateau having formed in the
vicinity of a spreading center (or at least on thin
The original plate tectonicsettingof the OJP is opento lithosphere)by high fractionsof partialmelting[Mahoney
some question becausewell-defined magnetic anomaly et al., 1993; Tejadaet al., 1996a]. Tardunoet al. [ 1991],
lineationsdo not appear to be presenton the plateau. Richardset al. [ 1991], and Mahoney and Spencer[ 1991]
Possiblespreading-ridge or fracture-zonefabric is subdued all favoredthe Louisvillehotspotasthe plume involvedin
and difficult to interpretunambiguously, andmostpre-OJP the origin of the OJP; however,followingRichardset al.
crust to the west and south of the plateau has been [1989], they attributed the plateau to a cataclysmic
subducted. However, block-faultingstructuresalong the outpouringof magma associatedwith the initial, plume-
easternmargin of the high plateau were interpretedby head stageof the hotspot,probablyin the early AptJan.
Andrewsand Packham [ 1975] and Hussonget al. [ 1979] Mahoney and Spencer [1991] argued that, even if not
to be parts of fracturezonestrendingroughlyNNE. This initially surfacingnear a spreadingaxis, plume
188 THE ONTONG JAVA PLATEAU
would tend to attract ridges becauseof their expected
control on riff propagation. Recently, Winterer and
Nakanishi [1995] also inferred a near-ridgeplume origin
for the OJP; however,they interpretedbathymetryand the
fabric in the new satellite-derivedgravity map of Smith
and Sandwell [1995a,b] to indicate an orientation of
fracturezonesand spreadingaxis nearly perpendicularto
that suggestedby Hussonget al. [ 1979].
The plateau geometry inferred by Winterer and
Nakanishi [1995] appearsdifficult to reconcilewith the
nearby ENE-WSW M-series Nauru magneticlineationson
the east side of the OJP [e.g., Nakanishi et al., 1992]
(Figure 1). Moreover, Taylor [1978] reported M-series
magneticanomalylineationsin the nearbyLyra Basin on
the west side of the OJP roughly parallelingthosein the
Nauru Basin on the east. A more recent aeromagnetic
survey in the Lyra Basin also revealed roughly ENE-
WSW-oriented M-series magnetic anomaly lineations(B.
Taylor, personalcommunication,1995). Our interpretation
of the combinedbathymetryand gravity map of Smithand
Sandwell [1995a,b] is that the data are consistentwith a
NNE-trending fracture-zonefabric on the high plateau.
Ratherthan providingevidencefor eithera single,brief,
cataclysmicemplacementeventor a basementageprogres-
sion, recent 40Ar-39Arages for OJP basementlavas yield
an intriguing, stronglybimodal distribution. The agesof
lavas from Sites289, 807, and 803 on the high plateau,as
well as for basementlavas from Malaita, Ramos, and Santa
Isabel, suggestthat most of the plateaumay have formed
in two relatively brief episodes,the first at 122+3 Ma, the
secondat 90+4 Ma (errorsindicatetotal rangesratherthan
weighted means) [Mahoney et al., 1993; Tejada et al.,
1996a,b; Parkinson et al., 1996]. Thus, much of the fabric
interpretedfrom bathymetryand satellite-derivedgravity
datais likely to representpreexistingoceaniccrustcovered
by widespreadplateau-basalteruptions,consistentwith
pre-122 Ma, southward-youngingM-series magnetic
anomalylineationsin the Nauru andLyra basins.Although
a spreadingcentermay well have been present,at leastin
the 122 Ma phase of eruptions,substantialmagmatism
must have occurredwell beyondthe immediatevicinity of
a ridge axis. Indeed, available geophysicalevidence
weakly favors emplacementof most OJP lavas in an off-
ridge location[Coffin and Gahagan, 1995]. As sampling
of the plateau is still very limited, the relative crustal
volumes of the •122 and •90 Ma episodes,and thus
emplacementrates,are as yet unclear. However, Tejadaet
al. [ 1996a,b] and Kroenkeand Mahoney [ 1996] suggested
that the 122 Ma episodewas significantlylarger than the
90 Ma event, hypothesizingthat the 122 Ma event
generally correspondedto the constructionof the main,
high plateauandthat the easternsalientwas the main focus
of activity at 90 Ma.
4. THE OJP AND CRETACEOUS PLATE MOTIONS
Formationof the larger oceanicplateausin the Pacific
appearsto be associated temporallywith major changesin
Pacific Plate motion. Four such changesappearto have
occurredin the Late Jurassicto Early Cretaceous,a time
when the plate was relatively small and being guidedby
the motionof adjoiningplates[Kroenkeand Sager, 1993].
Thesechanges,at about140, 125, 110, and 100 Ma, appear
to haveoccurrednearthetimesof formationof the Shatsky
Rise, OJP (early event)and Manihiki Plateau,westernand
northernHess Rise, and central Hess Rise, respectively,
along or near the divergentboundariesof the PacificPlate
[Kroenkeand Sager, 1993]. The mostpronounced change
in motion occurred at z125 Ma, between magnetic
anomalies M1 and M0 near the Barremian-Aptian
boundary [e.g., Steiner and Wallick, 1992], and was
probablyconcomitantwith the cessationof southwestward
subduction beneath northeastern Gondwana and the
beginning of northwestwardsubductionbeneathEurasia.
This event can be noted, for example, in the reversalin
magneticanomaly lineationpatternfrom M3-M1 to M1-
M3 east of the OJP near the west end of the Nova-Canton
Trough (Figure 1; see also Nakanishi and Winterer
[1996]). Associatedwith this changewas the formationof
a large portion of the OJP (i.e., the 122ñ3 Ma event) and
the Manihiki Plateau(R. A. Duncan,unpubl.data, 1993).
Unfortunately, existing data do not allow resolutionof
whetherOJP eruptionsfollowedor at leastpartlypreceded
the change. Indeed,40Ar-39Ardatesfor OJP lavasprovide
some of the best estimatesof the minimum age of
magneticreversalM0 andthe Barremian-Aptian boundary
[Pringle et al., 1992; Mahoneyet al., 1993].
Rough pre-100 Ma Pacific Plate motions can now be
determinedback to •145 Ma, using(1) the probableage
progression alongthe ShatskyRiseestimatedby Sagerand
Han [1993]; (2) paleomagneticevidenceof a changefrom
a southwardto a northwardcomponentof plate motionat
•125 Ma [Steiner and Wallick, 1992] and limited
latitudinal movement until about 90 Ma at the latitude of
the Mid-PacificMountains[TardunoandSager,1995];(3)
recentlymappedmagneticanomalylineationsin the west-
central Pacific [Nakanishiet al., 1992]; (4) charts of
central and westernPacific seamounts[Mammerickxand
Smith, 1985]; and (5) the new satellite-derivedgravity
mapsof Smithand Sandwell[1995a,b]. Figure4 (bottom)
showsthe site of the OJPat approximately125 Ma on the
Pacific Plate, far from continentalinfluences,centered
 NEAL ET AL. 189

about 42øS, 159øW. From about 125 Ma until northward trajectory, and from about 100 Ma to
approximately100 Ma the OJP appearsto have been approximately85 Ma the OJP movedsteadilynorthward
positionedvery closeto the PacificPlateEuler poles,and with the plate [Yah and Kroenke, 1993]. The plate
thus moved relatively little. At approximately100 Ma, reconstructionmodel of Yah and Kroenke [1993] shows
plate motion changedfrom a northwestwardto a more that at 90 Ma, the age of the secondmajor eruptiveevent,
the easternmargin of the OJP passesapproximatelyover
thepositionoccupiedby the centralhighplateauat 125 Ma
160 ø 180 ø
(Figure4, middle).
The locationof the plateauin the 125 Ma reconstruction
140ø •amoa 160ø of Figure 4 is at least roughly consistentwith paleo-
magneticdata for basementand basal sedimentsat Site
289, which yield a paleolatitudeof 30-35øS [Hammondet
al., 1975]; a similarpaleolatitudeis indicatedby basement
lavas at Site 807, althoughbasementthere may have been
tilted around the time of emplacement [Mayer and
Tarduno,1993]. No hotspotis knownto existtodayin the
vicinity of the triangleshownin Figure4, but this areais
one of the least surveyedin the world's oceans. This
location is •1800 km distant from the Louisville hotspot,
which, as noted above, has been suggestedby several
workers to be linked to the OJP and which has a well-
80 Ma marked seamounttrail (the Louisville Ridge or Seamount
Chain) going back to •70 Ma. Older portionsof the
Louisville hotspot trail, if they existed, have been
destroyedby the Vitiaz-Tongatrenchsystem[cf. Mahoney
and Spencer,1991]. To accommodatea Louisvillehotspot
originfor the OJP,true polarwanderof•10-15 ø since125
Ma must be invoked [e.g., Mayer and Tarduno, 1993].
Also, isotopicand elementaldata for the 122 and 90 Ma
OJPlavasare closelysimilarto eachother,consistent with
a single plume being responsiblefor both eruptive
episodes, but they are distinct,particularlyin Pb isotopic
ratios,from the 0-70 Ma lavasof the Louisville Seamount
Chain (see Section6.3.1). At present,identificationof the
hotspotassociated with the OJPremainselusive.
90 Ma
Between about 90 and 85 Ma, a major change in
Australian and Antarctic Plate motions took place as

Figure4. Bottom: approximatelocation of the OJP high

plateauat 125 Ma. Shadingrepresents postulatedextentof 122
Ma volcanism.Severalpresent-day hotspotsare shownas circles.
The trianglerepresentsthe inferredlocationof the OJP plume
centerbeneaththe crestof the high plateau. Heavy lines indicate
a possiblespreadingridge arrangement(schematicexceptat the
easternend). Middle: locationof the OJPat 90 Ma; notethat the
easternlobe of the plateau is now above the inferred plume
center. Here, shadingrepresents'thepostulatedregionof heaviest
90 Ma volcanism.Top: reconstruction for 80 Ma showingrifling
of the easternlobe (schematic).The 125 Ma reconstruction uses
125 Ma the 125-110 and 110-100 Ma Pacific Plate Euler poles of
Kroenke and Wessel[1997]; the 90 and 80 Ma reconstructions
are after Yah and Kroenke [
190 THE ONTONG JAVA PLATEAU

spreadingbeganin the TasmanBasinand in the Southwest al., 1986; Petterson, 1995; Petterson et al., 1997]. These
Pacific Basin, south of Campbell Plateau [Cande et al., islandsappearto representthe tops of several-km-thick
1989]. Followingthe •90 Ma eruptiveepisodeon the OJP, "tectonic flakes" of OJP crust thrusted onto the old forearc-
post-emplacement rifting and seafloorspreadingmay have backarc region, probably during the Pliocene [e.g.,
Kroenke, 1972; Petterson et al., 1997]. Farther south, a
occurredfor up to severalmillion yearswithin the plateau's
easternsalient,in conjunctionwith spreadingin the Ellice largethrust-sheetof OJP crustformsthe seafloorsouthof
Basin to the east (Figure 4, top). A rifted characteris Malaita [Auzendeet al., 1996], possiblyextendingto and
suggestedby the recent satellite-derivedgravity data of evenincludingpart of Makira (San Cristobal)[Pettersonet
Smith and Sandwell [1995a,b], which show that the al., 1995;Birkhold-VanDykeet al., 1996].
northernand southernridgesboundingthe bathymetriclow The Solomon Islands group has been subdividedfor
in the easternsalientresembleconjugateridges,with slight severaldecadesinto three geologicalregionsor provinces
gravity lows on their southern and northern sides, (Figure 3) [e.g., Coleman, 1966, 1976; Hackman, 1973;
respectively (Figure 2). A similar rift-margin fabric Colemanand Packham, 1976]. The easternregion,termed
extends eastward across the adjoining Ellice Basin. the Pacific Province, appears to be an uplifted,
Magnetic lineationshave been detectedin the Ellice Basin overthrusted, largelyunmetamorphosed portionof the OJP,
trending roughly east-west(D. Handschumacher,unpubl. asnotedabove[additionalreferencesincludeAndrewsand
data) but have not yet been identified;however,Duncan Packham,1975;Hughesand Turner,1977;Colemanet al.,
[1985] obtained an 40Ar-39Ardate of 82.6+1.2 Ma for a 1978; Coleman and Kroenke, 1981; Ramsay, 1982;
mid-ocean-ridge-typebasalt from the easternend of the Hopson, 1988; Petterson,1995; Tejada et al., 1996a].
Ellice Basin. Note that rifting of the easternsalient andAdjacentto the Pacific Provinceon the southwestis the
spreadingin the Ellice Basin soon after the •90 Ma OJP Central Province,which containsvariably metamorphosed
episode could explain why the OJP is not connected CretaceousandEarly Tertiaryseafloorandremnantsof the
physically to a post-plateau chain of seamounts(the northeast-facing arc sequence that grew duringthe Earlyto
"plume-tail"stageof hotspotdevelopment). Middle Tertiaryabovethe thensouthwest-plunging Pacific
Plate (prior to the arrival of the OJP from the east). The
5. RELATIONSHIP TO THE SOLOMON ISLANDS boundarybetweenthe Pacific and Central provincesis
generally submerged,but lies above sea level on Santa
Alongthe plateau'ssouthernand southwestern margins, Isabel, where it forms a fault zone termed the Kaipito-
basementtopographyreflects faulting and deformation Korigholefault system[e.g., Hawkins and Barron, 1991;
causedby the collision of the OJP with the Solomon Tejadaet al., 1996a]. Along the southwestern flank of the
Islandsarc [e.g., Kroenke, 1972; Kroenkeet al., 1986, Central Province is the Volcanic Province, an island arc
1991; Auzendeet al., 1996]. The OJP arrived at the old sequencecomposedof volcanicand intrusiverocksand
Solomonarc duringthe earlyNeogenein a "soft"docking activevolcanoes;the age of this provinceappearsto be <4
without significantdeformation,with southwest-directed Ma [e.g., Petterson, 1995]. Significantly, the plateau
subductionendingaround27-23 Ma [e.g., Colemanand appearsto be more or lessunsubductable [Cloos,1993;
Kroenke, 1981; Cooper and Taylor, 1985]. Initiationof Abbottand Mooney, 1995]. However,the post-Miocene
northeast-directed subductionbeneaththe plateauoccurred removal of a portion of the lower OJP betweenSanta
progressively
from the ESE becausecollisionof the OJP Isabel and Makira is evident from recent seismicsurveys
with the old arc was diachronous. Collision moved WNW
[Mann et al., 1996; Phinney et al., 1996; Cowleyet al.,
alongthe arc throughoutthe Neogeneas a resultof the 1996].
generalwestwardmotion of the OJP, with subduction
beginningon the plateau'ssouthside at 10-5 Ma and 6. GEOCHEMISTRY AND AGE OF BASEMENT
continuingthroughoutthe Pliocenealongthe arc [Mannet
al., 1996]. Subduction
duringthe quiescent periodbetween 6.1. Submarine Drillholes
27-23 Ma and 10-5 Ma appearsto havebeenoccurringin
theTongaandTrobriandtrenches instead.Thecollisionof The only detailed basementstratigraphyfor the entire
the Woodlark Basin spreading-ridge complexwith the plateau,prior to the recent and ongoingwork in Malaita
southwest-facingSan Cristobal forearc pushedthe arc and Santa Isabel, came from the 149-m-thick section of
northeastwardinto the OJP and producedthe Malaita ODP Site 807 and the 26 m sectionof Site 803 [Mahoney
Anticlinorium,an extensivefold belt embracingthe eastern et al., 1993]. At Site 807, the sectionhas been divided into
Solomon islands of Malaita, Ulawa, Ramos, and the severalunits(A, C-G), eachof which is a packetof low-K,
northernhalf of Santa Isabel [Kroenke, 1972; Kroenke et tholeiitic pillow lavas and massiveflows, exceptfor
 NEAL ET AL. 191

F, which consistsof a single28-m-thick flow, and Unit B [Tejada et al., 1996a;Parkinsonet al., 1996]. The Nd-Pb-
which is a 50-cm-thickinterlavalimestone[Kroenkeet al., Sr isotopiccharacteristics of the Sigana Basaltsclosely
1991; Mahoney et al., 1993]. Unit A (46 m thick) is resemble those at Site 807 and Site 803. Both Unit-A-like
isotopically (Figure5) andchemically(Figure6 andTable (hereafter"A-type") and Units-C-G-like (hereafter"C-G-
1) distinctfrom Units C-G in having lower 206pb/204pb type") isotopiccompositionsare presentand, as at Site
(18.3-18.4 vs. 18.6-18.7) and initial end (+4.8 - +5.4 vs. 803,the•90 Ma lavas aresimilar totheC-G-type [Tbjada
+5.9- +6.3), higher initial 87Sr/86Sr (0.7040-0.7041 vs. et al., 1996a]. Averageincompatibleelementsignatures
0.70339-0.70345), and slightly higher ratios of highly are againvery similarto thoseat the ODP sites(Figure6),
incompatible to moderately incompatible elements. althougha widerrangeof abundances is presentamongthe
Consistent with a plume-headorigin,the isotopicvaluesof SiganaBasalts,includingseverallavasthat appearto be
bothgroupsfall withinthe rangeof thosefor "Dupal-type" highly differentiated basalts [Tejada et al., 1996a;
hotspotislands(although few present-dayislandshave Parkinson et al., 1996].
valuescloselysimilarto thoseof the OJP lavas). Except In additionto the low-K tholeiitic SiganaBasalts,rare 90
for elementsespeciallysusceptible to seawateralteration Ma alkalic dikes, termedthe SiganaAlkalic Suite [Tejada
(e.g., Rb, K, Cs), primitive-mantle-normalized incompat- et al., 1996a], are present within the Sigana Basalt
ible elementpatternsare quite flat (Figure 6), although exposureson SantaIsabel. Despitebeingindistinguishable
some of the most incompatibleelementsare relatively in age from the younger group of tholelites, they have
depleted(e.g., Ba and Th); Li and P are also depleted high-206pb/204pb,"HIMU" or "Mangala Group"-type
whereasMo is relativelyenriched(notethat rock powders isotopic signatures quite distinct from those of the
were ground in alumina). Major and trace element tholeiites(Figure 5).
compositions
suggestboth groupsreflecthigh degreesof
melting(probably20-30%, with Units C-G representing 6.3. Malaita
slightlyhigherdegreesof meltingthanUnit A). The single
flow sampledat DSDP Site 289 is nearly identicalto the
6.3.1. O•IP Basement. Following reconnaissance
studies
Units C-G lavas. Althoughdifferingsomewhatin major
by Rickwood[1957] and Pudsey-Dawson[1960], Hughes
elementcomposition, the Site 803 lavasare closelysimilar
and Turner [1976, 1977] mapped the southernpart of
isotopically to the Units C-G basalts, with identical
Malaita, wherethey termedthe basementsectionthe Older
206pb/204Pb and initial end and only slightlyhigherinitial
Basalts. Recently, the basement lavas throughoutthe
87Sr/86Sr(0.7036-0.7037) (Figure 5). Their incompatible
island have been renamedthe Malaita Volcanic Group
element signature is also similar to that of Units C-G,
[Petterson,1995]. The 40Ar-39Aragesof samplescollected
althoughthey are slightly enrichedin the more highly
by Hughes and Turner showed them to be 122 Ma
incompatibleelements(Figure6 andTable 1).
[Mahoney et al., 1993]. All are low-K tholeiites of
The 40Ar-39Aragesfor Unit A and Units C-G lavas, as
essentiallyA-type composition,althoughrocks from the
well as the Site 289 basalt, are identical,within errors,at
much more extensive exposures in southern Malaita
•122 Ma [Mahoney et al., 1993]. However, the Site 803
exhibit somewhatgreater chemical and isotopicvariation
lavas yielded significantly younger ages of •90 Ma;
sediments above basement at Site 803 could not be used to
than the 46 m of Unit A flows at Site 807 (e.g., Figure 5)
[Mahoney, 1987; Mahoney and Spencer,1991; Tejada et
confirmthisage asthe siteis locatedin a depression which
al., 1996a].
appearsto containderivedsediments[Kroenkeet al., 1993;
The thickestexposuresof basementcrust(reaching3-4
Mahoneyet al., 1993; Sliter and Leckie, 1993]. Foraminif-
km in stratigraphicthickness[Petterson,1995; Tejada et
eral agesof thesesedimentsvary from Barremianto Lower
al., 1996b;Pettersonet al., 1997]) are in the centralpart of
Aptian [Sliter and Leckie, 1993].
Malaita, which remained unsampled until recently.
Basementin centraland northernMalaita is exposedin the
6.2. Santa Isabel coresof NW-SE-trending anticlinalto periclinalstructures
formed in responseto the collision of the OJP with the
The •90 Ma event was confirmedby 40Ar-39Arresults Australian Plate [e.g., Pettersonet al., 1997]; the major
for the tholeiiticSiganaBasalts,which form the basement outcrops of Malaita Volcanic Group rocks are in the
of the northernportion(PacificProvince)of SantaIsabel Kwaio, Kwara'ae, Fateleka,and Tombaltaareas(Figure 7).
andRamosIsland.Boththe •122 Ma and•90 Ma groups These areas are dominated by submarinebasalt flows
are present,with the 90 Ma lavas being particularly which range from <1 m to 60 m in thickness,with most
abundantin the middleand southeastern partsof the island between 4 and 12 m thick; dikes are rare in most
192 THE ONTONG JAVA PLATEAU

TABLE 1. Compositionsof Basaltsfrom Malaita, ODP Leg 130 Sites,andAverageNauru BasinBasalt

Standard Reference Materials Malaita ODP Leg 130
BIR-1 BHVO-I A-Type C-G-Type ML475 Site 807 Site 807 Site 803 Nauru

Publ. Meas. Publ. Meas. (n=48) (n=21) A-Type C-G-Type Basin

SiO2 49.92 49.25 49.68 48 24 49.88 49.09 49.7

TiO2 1.59 1.20 0.73 161 1.15 1.37 1.2

A1203 14.31 14.47 14.00 14.29 14 24 15.46 13.9

Fe203 13.71 12.69 9.84 13 52 12.26 11.59 4.0

MnO 0.20 0.21 0.16 0 20 0.20 0.16 0.2

MgO 7.11 7.75 9.99 6 66 7.64 6.25 6.8

CaO 11.29 12.35 14.52 1211 12.13 12.22 11.4
Na20 2.23 2.05 1.48 239 2.20 2.32 2.4

K20 0.14 0.14 0.08 0.29 0.15 0.47 0.1

P205 0.14 0.11 0.06 0.14 0.10 0.15 0.1

Li R3.4 3.95 P4.6 4.12 6.42 6.71 3.51

Be P0.58 0.14 PI.1 0.97 0.65 0.47 0.25
K 1139 1129 664.0
Sc R44 56.9 •31.8 36.4 44.0 46.5 57.9 46.5 49.3 48.9 47.4

V •313 476 •317 369 370 314 217 325 347 312
Cr •382.0 399.8 R289.0 340.3 96.2 133.0 446.3 156 153 252 245

Co •51.4 53.7 •45.0 49.2 51.0 50.9 38.8 50.O

Ni •166 163 •121 119 70.5 98.1 133 96 94 110 115

Cu R126 113.4 •136 135.7 107.1 140.8 128.0

Zn •71 71 •105 106 99 80 47
Ga PI6.0 15.0 •21.0 25.1 19.4 17.0 13.1
Rb il .0148 R11 9.93 1.72 1.32 0.91
Sr •108 111 •403 425 145 153 135 188 113 172 110
Y •16 15.7 •27.6 26.4 28.4 22.6 13.5 30.2 25.0 27.6 22.0
Zr P22 15.9 R179 182 94.1 69.3 29.9 98.6 64.4 79.4 49.5
Nb P2 0.7 •19 20.3 5.44 3.76 1.57 5.89 3.50 4.45 3
Mo I0.5 0.10 • 1.02 1.004 0.69 0.54 0.18 0.75 0.79 0.50
Cs P0.45 0.03 •0.13 0.16 0.02 0.01 0.01 0.19 0.04 0.18
Ba P7.7 6.94 •139 135.6 34.1 23.6 11.5 24.1 13.6 16.5
La P0.88 0.65 •15.8 15.9 5.59 3.66 1.61 5.79 3.38 4.31 3.4
Ce •2.5 2.01 R39 40.6 15.72 10.55 4.80 14.39 8.99 9.5
Pr •0.5 0.35 •5.7 5.22 2.32 1.57 0.72 2.29 1.46 1.78
Nd P2.5 2.40 •25.2 25.05 11.64 8.26 3.78 10.79 7.10 8.46 6.3
Sm •1.I 1.04 •6.2 6.22 3.56 2.67 1.30 3.36 2.39 2.88 2.6
Eu R0.54 0.51 •2.1 2.11 1.30 0.97 0.51 1.29 0.92 1.11 1.0
Gd P1.9 1.77 •6.4 6.42 4.46 3.40 1.74 4.31 3.26 3.73 3.3
Tb •0.41 0.34 •1.0 0.94 0.79 0.62 0.32 0.64 0.73 0.8
Dy R2.4 2.43 •5.2 5.23 4.93 3.90 2.05 4.94 4.06 4.45
Ho P0.5 0.50 •0.99 0.98 1.05 0.83 0.43 1.06 0.90 0.97
Er P1.8 1.67 •2.4 2.51 3.01 2.37 1.26 2.99 2.46 2.74
Tm P0.27 0.24 •0.33 0.32 0.43 0.34 0.17
Yb •1.7 1.60 R2.02 2.07 2.90 2.28 1.16 2.72 2.40 2.59 2.6
Lu •0.26 0.27 •0.29 0.28 0.43 0.34 0.17 0.41 0.35 0.38 0.4
Hf •0.58 0.58 •4.38 4.32 2.51 1.91 0.87 2.54 1.65 2.07 1.7
Ta •0.0006 0.04 R1.23 1.34 0.36 0.27 0.10 0.38 0.21 0.30 0.2
W •0.20 0.23 P0.27 0.38 0.06 0.04
Pb P3.2 3.11 P2 2.04 1.30 0.81 0.40 0.72 0.67
Th I0.89 0.02 •1.08 1.23 0.53 0.33 0.12 O.52 0.28 0.35

U I0.025 0.027 •0.42 0.43 0.14 0.09 0.03 0.14 0.12 0.21
 NEAL ET AL. 193

Our ongoingwork andthat of colleagues A. Saundersand
T. Babbs(Univ. of Leicester,UK) revealsthat broadlyC-
G-type lavasare presentbelow an A-type cap some600 m
thick ([Tejada et al., 1996b] and C. R. Neal et al.; M. L.
Tejadaet al.; T. Babbset al., manuscripts in preparation,
1997). The range of chemicalvariationis again greater
than in the 149 m of flows at Site 807 (Figures8a-d, 9),
butA- andC-G-typebasaltscanclearlybe distinguished in
incompatibleelement (e.g., Figure 6), major element
(Figures8c and 9d), and isotopic(Figure 5) diagrams.
Also, all dated basementsamplesfrom central Malaita
belongto the 122 Ma event[Tejadaet al., 1996b]. Thus,
no 90 Ma lavas have been found on the island, and A-type
lavas form a section more than 13 times thicker than at Site
807, some 1600 km away to the north. The transition
betweenA-type and C-G-type groupsis again sharp,but
unlike Site 807, where an interlavalimestoneunit (Unit B)
separatesthe two types, no correspondingsedimentary
layer is seenin Malaita. Theseresultssuggestthat the part
of the OJP now exposedon Malaita was more proximalto,
or possibly downslopefrom, the main locus of A-type
eruptive activity than Site 807. Moreover, the combined
data for Malaita, Santa Isabel, and the drill sites
demonstratethat both A- and C-G-type lavaswere erupted
over a considerableportionof the plateau[cf. Tejadaet al.,
1996a,b].
For eachtype, a voluminous,well-mixed mantle source
(relative to the scale of melting) is indicated. In the
contextof a zonedplume-headmodel, one type may better
representthe plume-source compositionand the other
better reflect average entrained mantle, or the plume
sourceitself may have containedboth types;the C-G-type
is closer to some estimates of average lower mantle
isotopic composition[Tejada et al., 1996a,b]. Intrigu-
ingly, lavas of the much more poorly sampledManihiki
Plateauto the eastof the OJP (Figure 1) are very closein
age to the •122 Ma OJP event (R. A. Duncan,unpubl.
data, 1993) and also define two isotopicgroups,one of
which is quite similar to the OJP A-type (Figure 5a,b),
possibly suggestingthat the sources of the OJP and
Manihiki Plateau were related [Mahoney and Spencer,
1991; Mahoney et al., 1993]. Indeed,Coffin and Eldholm
[1993] speculated thatthe Manihiki Plateaumay be part of
a "greater" OJP event, along with lavas filling the Nauru
andEastMariana basins. Isotopicand elementaldata(e.g.,
Figures 8 and 9) for the basin-filling basalts partially
overlapwith thosefor C-G-type OJP lavas[e.g., Tokuyama
and Batiza, 1986; Floyd, 1986; Castillo et al., 1986, 1994;
Saunders, 1986; Mahoney, 1987]. However, their 40Ar-
39Arages,at 111-115 Ma, are significantlydifferentfrom
either the 122 or 90 Ma events on the OJP, and Castillo et
al. [1994] proposedthat they reflect OJP sourcemantle
that had flowed northward and mixed variably with
ambient MORB-type asthenosphere(recall that the OJP
itself probably moved very little during this period; see
Section4 and Figure4).
The similarity in isotopic and incompatibleelement
ratiosof the 90 Ma OJPtholeiitesto the 122 Ma C-G-type
strongly suggestsboth groups shared a common plume
mantlesource.As notedabove,the plateaumay havebeen
locatedat a roughly similar geographicpositionat these
two times;however,the lack (so far) of any OJP lavaswith
intermediateages arguesagainsta smoothplume-headto
plume-tail transition after the 122 Ma event. One
possibility suggested by simple experimental and
theoreticalmodelingis that the strongbimodalityin ageof
OJP lavas representsa separationof the starting-plume
head from its conduit (tail) as it rose throughthe 660-km
mantle discontinuity,followed by formation of a second,
smaller head which rose to the base of the lithosphere
some 30 m.y. after the first [Bercovici and Mahoney,
1994]. Alternatively, plume material may have collected
beneaththe plateau more or less continuouslyafter 122
Ma, but did not erupt to the surfacein significantamounts
until a changein the stressfield of the OJP around90 Ma
providedthe necessarypathwaysfor melt egress[Tejadaet
al., 1996a;Ito and Cliff, 1996].
A quite different model was proposedby Larson and
Kincaid [1996]. They suggestedthat the 122 Ma event
reflected upward advection of a region of the 660-km
boundarylayer in responseto acceleratedsubductionof
slab material into the deep mantle. Ascent of this
boundarylayer would increasetemperaturesin the upper
mantle above the peridotite solidus and thus cause
voluminousplateau magmatism. At about the sametime,
increased amounts of slab material would encounter the

Note that all Fe is listedas Fe203. Data for Malaitanbasaltsfrom this work, for ODP basaltsfrom Mahoneyet al. [1993], and the
averagecompositionfor the Nauru Basinbasaltsare calculatedfrornBatiza [1986], Tokuyamaand Batiza [ 1986], Saunders[ 1986],
Castilloet al. [1986, 1991], andFloyd [1986, 1989]. The publishedcompositions of standardreferencebasaltsBIR-1 and BHVO-1
[Govindaraju, 1989] are shown along with our ICP-MS (inductivelycoupledplasmamass-spectrometric) determinations(Meas.)
performedduring the analysesof the Malaitan basalts.Superscripts on publishedvalues:R = recommendedvalues (i.e., most
accuratelydeterminedabundances); P = provisionalabundances; I = data are for informationonly (least accuratelydetermined
abundances).
Major elementabundances
arein wt%;all othersarein partsper
194 THE ONTONG JAVA PLATEAU

+12
• I Southern
Malaita
Pacific
MORB •F-I Sigana
CentralMalaita
Basalts
Santa
Isabel
Z• SiganaAlkalicSuite

+8 Santa
Isabel
Manihiki • Nauru
Plateau[-"% • • Basin

dr/•.••
+4

x.
xj .,teu a
0
0.702 0.703 0.704 0.705

(87Sr/86Sr)t
+12

MORB

+8

sUitniet•
'ticManihiki
Nauru

0C:•G/_•••••'•Aed
ge Mangaia
Plateau

A•.•l•
• Ridge
•b
+4

Manihiki
Plateau

317
o I I
18 19 2O

2ø6pb/2ø4pb
Figure5. Initial SNd(t)vs. (87Sr/86Sr)t(a) andpresent-day2ø6pb/204pb(b) for basement lavasof MalaitaandSanta
Isabeland90 Ma alkalicdikesof SantaIsabel[Mahoney,1987;MahoneyandSpencer,1991;Tejadaet al., 1996a;M.
Tejada et al., unpubl.data, 1995]. Fields for A-type and C-G-type/Site803 drillhole lavas are shownby heavy
outlines,thosefor Malaitanaln6ites,ManihikiPlateau,LouisvilleRidge,PacificMORB, andMangalaGroupislands
of the SouthPacificby lightoutlines[seeMahoneyet al., 1993; Tejadaet al., 1996a,for data
 NEAL ET AL. 195

lOO - i i i i i

. o

.. \
! \
!
\
!
/

lO

BHVO Std
N-Type MORB
.... OIB
a, Malaita A-Type
• Malaita C-G-Type
.... x---- ODP Site 803
.... A---- ODP Site 807 A-Type
.... m--- ODP Site 807 C-G-Type
.... o--- Isabel:SiganaBasalts

o.1
Cs Ba Th Nb K Ce Pr Sr Nd Zr Eu Gd Dy Y Er Yb
Rb W U Ta La Pb Mo P SmHf Ti Tb Li HoTm Lu
Figure6. Primitive-mantle-normalizedelement profilestbraverage
of OJPbasaltsof Malaita(thisstudy),Sites803
and807 [Mahoney et al., 1993],andSantaIsabel[Tejadaetal., 1996a]).Average
N-typeMORBandOIB patterns and
primitive-mantle
normalizingvaluesare from Sun and McDonough[1989]. BHVO-1 standardvaluesare from
Govindaraju[ 1989].

core-mantle
boundary,andthe resultinginstabilitywould a geochemical link between the OJP and the Louisville
producediapiricupwellingarguedto reachthe surfaceasa hotspothave been unsuccessful. The Nd-Pb-Sr isotopic
plumeheadsome20-30 m.y. afterthe initialmagmatic range def'medby •70-0 Ma seamountlavas along the
event. Thismodelaccounts
for the bimodalityin the OJP •4000-km-long Louisville Ridge (see Figure 5) is
basementages but implies significantisotopicand restrictedandvaluesvary little with rock type or age,age
chemicaldifferences
betweenthe 122 Ma magmatism, of underlyingoceaniccrust,or depthof melting,indicating
which would have an upper mantle and/or 660-km a long-lived,isotopicallyhomogeneous source[Chenge!
boundary-layer
source,andthe 90 Ma event,whichwould al., 1987;Hawkinset al., 1987]. Althoughinitial l•Ndand
have a sourceoriginatingat the core-mantle
boundary. 878r/86Srvaluesfor the LouisvilleRidgeoverlapwith those
Thus, the model does not explain the marked of the 122 and90 Ma OJPC-G-typebasalts(Figure5), the
compositional
similaritybetweenthe 122 Ma C-G-type differencein Pb isotoperatiosis substantial
(e.g.,0.4 to 1.0
and the 90 Ma basalts. in 2ø6pb/204Pb).It is highly unlikely to be the result of
As notedin Section4, asidefrom problemsassociated radiogenicingrowthin the plume sourcebetween90 Ma
withpre-90Ma platereconstructions,
attempts
to establish and 70 Ma, but could representa major
196 THE ONTONG JAVA PLATEAU
Tombaita
Inlier
Fateleka
Inlier
Kwara'ae
Inlier
Kwaio
Inlier
0 50 km
i..i......•
Haruta
Limestone
Alite Limestone
Kwara'ae Mudstone
OJP Basalts
Figure 7. Simplified geologicalmap of Malaita defining the
OJP basementinliers [after Petterson,1995].
changefrom the plumeheadto plumetail [Mahoneyet al.,
1993] (as postulatedto be a common occurrenceby
Campbell and Griffiths [1990]). Note, however,that the
changewould have to have occurredduringthe 20 m.y.
period for which the recordjust happensto be missing
[Mahoneyet al., 1993].
6.3.2. Late-Stage Volcanismon Malaita. Two younger
volcaniceventsare recordedin the pelagic sedimentary
sectionoverlying basementon Malaita. The first consists
of flowsandsillsof alkalicbasaltscroppingout irregularly
within Eocene limestones in southern and northern Malaita
andthe nearby islandof Ulawa. Thosein southernMalaita
were called the Younger Basaltsby Hughesand Turner
[1977], a term supersededby the name Maramasike
Volcanic Formation, which includesboth the northern and
southernexposures[Petterson, 1995]. Although they
appearat the samestratigraphiclevel, their thicknessvaries
considerablyfrom place to place, locally reaching a
maximumof •500 m. An 40Ar-39Arage of 44.2+0.2 Ma
was obtained by Tejada et al. [1996a] for one of these
basalts. Their Nd-Sr-Pb isotopicratios are distinctfrom
thoseof the basementtholeiites,and they may reflect the
passageof the OJPoverthe Samoanhotspot[Tejadaet al.,
1996a]. Probable counterpartsare seen in seismic
reflection records of the high plateau north of Malaita,
which show numerous sill-like intrusions
sedimentarysection [Kroenke, 1972]; moreover, at least
someof the large seamounts
atop the plateaumay be
related to Maramasike volcanism.
The second late-stagevolcanic event is recorded in
central Malaita as several intrusionsof aln0ite, a rare,
ultramaficalkalic magma of deep mantle origin with
affinitiesto kimberlite. With an age of 34 Ma [Davis,
1977], the aln0itessignificantlypostdatethe Maramasike
VolcanicFormation. They appearto correspond to low
degreesof meltingof sublithospheric mantle[e.g.,Nixon
and Boy& 1979; Nixon et al., 1980; Neal, 1985, 1988]
during flexural extensionof the OJP as it overrodethe
outer-trenchhigh prior to colliding with the old North
Solomontrench [Colemanand Kroenke,1981;Petterson,
1995]. Diatreme-likeintrusions,probablycorrelativewith
the alnOiteson Malaita, are evident in seismicreflection
recordsover abouta third of the high plateau[Kroenke,
1972; Nixon, 1980]. The alnOitescontaina rich and varied
suiteof mantlexenolithsandxenocrysts,whichhavebeen
studiedextensively[e.g.,Nixonand Coleman,1978;Nixon
and Boy& 1979; Neal, 1988, 1995; Neal and Davidson,
1989]. Most of the xenoliths appear to represent
lithospheric
mantleandindicatethatthispartof theplateau
(near the southernedge of the OJP) had a pre-collision
lithosphericthicknessof about120 km [Nixonand Boy&
1979]. Isotopically,most of the xenolithsare distinctfrom
the OJP basementtholeiites,with generallyhigherinitial
878r/86Sr
and more variablelgSdvalues(Figure5a) [Neal,
1985, 1988;Neal andDavidson,1989],indicatingthatthe
lithosphericmantle of the plateauhas been variably
modifiedsincethe plateauformed.
7. EVALUATION OF OJP SOURCE REGION AND
MELTING CHARACTERISTICS
7.1. A Core-MantleBoundaryOriginfor the OdPPlume?
The ultimate origin of the OJP (i.e., the sourceof the
plume) is difficult to evaluatebecausethe presumedplume
headarguablycontainedboth plume-sourceand entrained-
mantle components.However, geochemicalsignaturesof
the plumesourcecouldpersistandbe detectablein erupted
magmasif suchsignatureswere (1) of sufficientmagnitude
and markedly distinct from those of ambientmantle, and
(2) not obliteratedby melting or magmaticdifferentiation
processes.Physicalconsiderations imply that plumeheads
giving rise to the largestLIPs may originateat the core-
mantle boundary [e.g., Campbell and Griffiths, 1990;
Coffin and Eldholm, 1991, 1994]. A key chemical
"fingerprint"of a core-mantleboundaryoriginmay be the
within the unusual enrichment of siderophilictrace elements [e.g.,
Walker et al., 1995]; such enrichmentmay occur
 NEAL ET AL. 197

1.6 -
ß Malaita A-type
[] Malaita C-G-type ,x
ß ML475
1.4 /X Site 807 A-type a b
ß Site 807 C-G-type
1.2
X Site 803
o Isabel
d- DSDP Site 289 <9o
O Nauru Basin

0.6-
o
0.4-
x

o.2
I I
0.0
46 47
o
•1
48 49
-• _ •
50 51
I
_

52
I

54 7 8 9 10

SiO2(wt%) MgO (wt%)

2.2

2.0

1.8 X ß o

1.0

0.8 C d
ß
I I I I
0.6
4 5 6 7 8 9 10 9 10

MgO (wt%) MgO (wt%)

15 4.0 i

14
e 3.5-

3.0-
13

•' 2.5-
• 12 xA x
¸ x 0 2.0-

- ;•,•c•
1.5
_
e= 11
x

lO _

ß o • 1.o-

o
0.5
4 5 6 7 8 9 lO
4 5 6 7 8 9 10

MgO(wt%) MgO(wt%)
Figure8. Majorelementvariations in theMalaitanOJPbasalts (thisstudy)comparedwiththoseof DSDP Site289
[Stoeser,1975], ODP Leg 130 [Mahoneyet al., 1993], SantaIsabel[Tejadaet al., 1996a],and the Nauru Basin
[Batiza,1986; Tokuyamaand Batiza, 1986;Saunders,1986;Castilloet al., 1986, 1991;Floyd, 1986, 1989]. (a)
Classification
of thesebasaltsgenerallyaslow-K tholeiites,
with K20 values> 0.2 wt% probablyresultingfromlow-
temperaturealteration;
(b-f) majorelementvariations
usingMgO asthefractionation
198 THE ONTONG JAVA PLATEAU

ß Malaita A-Type
[] Malaita C-G-Type
ß ML475
1.8 A Site 807 A-Type 65

ß Site 807 C-G-Type

X Site 803
1.6 6O
q- DSDP Site 289
O Isabel
1.4 ¸ Nauru Basin 55

E•1.2 - _ E5o
•.•
•.•

• 1.0•- C) 45

0.8 40

0.6 a 35 b
ß
0.4 30 I I

12 14 15 16 17
7 8 9
5 6
A1/Ti
A1203
(wt%)
17
5OO I
,
x
c
d
400 - 16

•00 - x

[]
200

IO0
13

I
0
I I
2 4 5 6 7 12
5 6
AI/Ti
A1/Ti
18

20O

e 16

150
14

ß,-..• t O0
oo¸ []
'0
0 ++ (

50

8
¸

6
04'--- 5 6 7 8 9 10 1 2 3 4 5 6 7 8
MgO (wt%) A1/Ti
Figure 9. Trace-elementcomparisonof A-type and C-G-typebasaltsfrom Malaita with thosefrom Sites807, 803
and 289, Nauru Basin, and SantaIsabel(data sourcesas in Figure
 NEAL ET AL. 199

entrainment of small amounts of outermost core material ß MalaitaA-Type •x( Site803

during the initial formation of the plume or by chemical [ ] MalaitaC-G-Type () Isabel
i• ML475 (•> Nauru Basin
diffusion across the boundary [e.g., Kellogg and King,
4.5 /• Site807 Unit A ß MORB
1993]. Unlike major siderophiles(Fe, Ni, Co, etc.), which I Site 807 Units C-G
are affectedstronglyby silicatemeltingand differentiation
4.0
processes, many siderophilictrace elements,suchas Mo,
W, Au, andRe, as well asthoseof the platinumgroup(Rh,
3.5
Ru, Os, Ir, Pt, Pd), are generallyincompatiblein silicate
minerals [e.g., Newsom and Palme, 1984; Jones and
3.0
Drake, 1986; Fleet et al., 1991, 1993; Walker et al., 1993].
Thus, large degreesof partial melting and basaltic-type
2.5
magmaticdifferentiationshouldnot significantlyalter the
ratios of these elementsto other incompatible(lithophile)
2.0
elements.
Jain et al. [1995] reportedthe presenceof a significant
enrichment in W and Mo for OJP basalts from the ODP
0 10 20 30 40 50 60 70
drill sites and Malaita. These ICP-MS data were
Zr/Nb
interpretedas supportinga core-mantleboundaryoriginfor
the OJP plume source. However, our recentreanalysisof Figure 10. Variations in Zr/Y and Zr/Nb exhibitedby the OJP
the samples,using a more aggressive(between-sample) and Nauru Basin basalts(data sourcesas in Figures6 and 8) and
southernEast Pacific Rise MORB [Mahoneyet al., 1994].
wash procedure [McGinnis et al., 1996, 1997]
demonstrates that the W enrichments are an artifact of
the Zr/Y ratio varies by nearly a factor of two (Y being
instrumental memory from previous samples and moderately incompatible in spinel peridotite but
standards.In fact, the OJP basaltsactuallyexhibit slightly compatiblein garnet). Similarly, the variationsin Sm/Yb
negative anomalies on primitive-mantle-normalized are comparableto thosein La/Yb (see Figure 11). These
element diagrams (Figure 6), similar to those seen for featuressuggestthat melting encompassed both the garnet
MORB and ocean island basalts.
and spinel peridotite stability fields [cf. Mahoney et al.,
In contrastto W, however, a persistentpositive Mo
1993; Farnetani et al., 1996].
anomaly is still evident in primitive-mantle-normalized
Simple forward modeling can be used to investigate
elementpatterns(Figure 6). The observedenrichmentof
some featuresof the mantle sourceand melting process
Mo in OJPbasaltsis consistentwith a siderophile-element-
that supplied the A- and C-G-type lavas. Given their
enrichedsourceregion. Preliminaryanalysesof platinum
generally flat primitive-mantle-normalized element
group elements and Au in OJP basalts also appear
patterns(Figure 6), we took a sourcewith primitive-mantle
consistentwith a siderophile-enriched source[e.g., Jain et
incompatibleelement abundancesas one possible"end-
al., 1996], but more detailedstudy is requiredbefore any
member". On the otherhand,the initial end valuesof A-
definitive statements can be made.
type basaltsclusteraround+4 to +5, andthoseof C-G-type
lavas around +6 (Figure 5a). These values are roughly
7.2. Partial Melting of the Plume Head PeridotiteSource halfway betweenthe presumedprimitive mantlevalue of 0
andthe averageMORB valueof about+ 10, suggesting that
Significantamountsof partialmeltingof an adiabatically a mantle source with mixed elemental characteristics could
rising plume head should occur when it intersectsthe also be appropriate.For our peridotite sourcemodeling,
peridotitesolidusin the upper mantle. One noteworthy meltingwas assumedto initiatein the garnetstabilityfield,
characteristicof the OJP basementlavas is illustratedby with the bulk of melting occurringin the spinelstability
Figure 10, showingZr/Nb vs. Zr/Y. Theseelementswould field as the plume head impinged on the lithosphere.
be highly incompatiblein the fractionatingassemblage (see Initial sourcemineralogy was estimatedfrom the modal
Section 8.1) and their ratios thus insensitiveto different mineralogyof garnet,garnet-spinel,and spinelperidotites
amounts of differentiation. Relative to southern East studied by Neal [1985, 1988] to be 60% olivine, 20%
PacificRiseMORB, for example,the OJPbasaltsdisplaya orthopyroxene, 10% clinopyroxene,and 10% garnet,
very restricted range of Zr/Nb, consistentwith high melting in the proportionsof 15:30:25:30. After an initial
degreesof partial melting (Nb being highly incompatible phaseof batchor fractionalmelting in the garnetstability
andZr moderatelyincompatibleduringmelting);however, field (minimum of 5%, maximum of 15%), the mode
200 THE ONTONG JAVA PLATEAU

5.0

5%
x Gt
Peridotire BatchMelting
4.5 - ',,,,
• • Fractional
Melting
15%
GtPeridotite
Melting
15 % Gt Peridotite
• • Melting Melting
X

5% Gt Peridotite u--••, • \ \

3.0-
Melting j• •L•• "Mixed" "Mixed"
Source Source

2.5- Primitive
•/• ••• A.•"• Primitive
•
Mantle
[____•}( a Mantle [-] b
2.0
10 12 14 16 18 20 22 10 12 14 16 18 20 24

Zr/Nb
2.5 i i i

FractionalMelting Batch Melting

5% Gt Peridotire
Melting
2.0

•k•o 'øø 5%GtPeridotite

A • ,• Melting
Primitive . ß
Mantle Primitive
Man •'
I 15% otite
Melting

"Mixed"
Source
15%
GtPeridotire
Melting
C '•"Mixed"Source
d
0.5 • I t • •
0.8 1.0 1.2 1.4 •.6 •.8 0.8 1.0 1.2 1.4 1.6 1.8 2.0

(Sm/Yb)
N
Figure11. Modelingof traceelementratiosas a functionof fractional
andbatchpartialmelting. The primitive
mantlesourceis takenfrom Sun and McDonough[1989] andthe "mixed"sourceis a 50:50 mixtureof primitive
mantleandMORB source[McKenzieand 0 'Nions,1991]. McKenzieand 0 'Nions[ 1991] reportedconcentrations
for
the rare earth elementsin their estimatedMORB source,but not Zr, Nb, or Y. In estimatingconcentrations
of these
elements we assume:(i) a Nb/Laratioof 1, producing
a Nb concentration of 0.206ppm;(ii) a Zr/Nbratioof 32 which,
onthebasisof 0.206ppmNb, yieldsa Zr abundance of 6.6 ppm;and(iii) Y abundance is 10timesthatof Yb, or 3.47
ppm. Themeltingtrajectories arefor hybridmelts,andincorporate meltingin boththegarnetandspinelperidotite
stabilityfields.Twomeltingpathsareshownforeachsource: onewhichincorporates a 5%partialmeltandthesecond
a 15%partialmeltof thesource in thegarnetperidotite
stability
field. To eachof thesemeltsareprogressively added
5% increments of melt derivedfromthe spinelperidotitestabilityfield afterphasetransformation facilitatedby the
risingplumehead.Thefirstcrossoneachmeltpath(i.e.,thatfarthest fromthesourcecomposition) representsa 5% or
15%meltfromgarnetperidotite plusa 5%meltfromspinelperidotite, resultingin a minimum partialmeltrepresented
on eachof thesepathsof 10% and20%, respectively. The higherthe degreeof partialmelting,the morethe melt
composition converges onthatof thesource.Partitioncoefficients usedarefromGreenet al. [1989]andHornet al.
[ 1994]. SubscriptN means

compositionof the residuewere calculatedand subsequent
meltingwas calculatedfor a spinelperidotite,assumingall
remaininggarnetwas convertedto spinel. Melts generated
from garnetand spinelperidotitewere then mixed to give
calculatedparentalmelt compositions. The startingpoints
for the model batch and fractionalmelting trajectoriesin
Figure 11 representthe minimum- and maximum-degree
partial melts generatedin the garnetstabilityfield (5% and
15%, respectively)which havebeenmixed with a 5% melt
from spinel peridotite, assumingeither a primitive or
mixed mantle source(50% MORB source,50% primitive
mantle). The crosseson the trajectoriesrepresentmixing,
in 5% increments,of these initial compositionswith
subsequent melts generatedin the spinelstabilityfield and
formed from the residueof the initial melting. Note that
because the later melts are derived from a residue after a
phasetransformation(andhencea changein bulk partition
coefficients)the melt trajectoriesfor neither batch nor
fractional melting lead to the initial mantle source
compositions asthe degreeof partialmeltingis increased.
The partial melting trajectoriesin Figure 11 imply that
the source(s) for the OJP lavas contained a greater
proportionof primitive-typemantlethanMORB mantle,as
the bulk of the dataplot betweenthe modeltrajectories;the
model sourcesand their calculatedmelt trajectoriesdo,
however,generallybracketthe basaltdata.For simplicity,
we have not considered more complicated melting
processes (e.g., zonerefining)whichmay have operatedto
someextent,at leastin the garnetperidotitestabilityfield,
duringthe rise of the plumehead.From our modeling,the
calculatedminimum degreeof partialmeltingfor the OJP
basaltsis 15-25% (generallyfor A-type basalts)and the
maximum is 20-35% (generally for C-G-type basalts).
However, cautionis requiredin interpretingsuchmelting
estimatesbecausetheserangesarea productof variationin
the model parametersused. The variationis producedby
the choice of fractional or batch melting models, the
amountof melt derivedfrom the garnetperidotitestability
field, andthe type of sourcemelted.
The data for A- and C-G-type lavas form two groups
with only limited overlap (Figure 11). Although these
rangesare small,we believethesegroupsare real in view
of the high precision of the ICP-MS data (errors on
individual data points would generally be within the
symbolplotted). Given the isotopicdifferencesbetween
the two groups,the small but consistentdifferencesin
incompatibleelement ratios could reflect trace element
differences inherent in the respectivemantle sources.
However,they may alsoreflectdifferentamountsof partial
melting followed by open-systemevolution,with the A-
type representinglowerdegreesof melting[cf. Mahoneyet
NEAL ET AL. 201
al., 1993; Tejada et al., 1996a]. Major elementsprovide
some additionalinsight. Figure 9d showsthe A1/Ti ratio
vs. A1203; despite variable amounts of plagioclase
fractionation/accumulation,the OJP basaltsexhibit a fairly
restrictedrange in A1203 (exceptingSite 803, the bulk of
the samplescontain 13.5-14.5 wt%) and again define A-
and C-G-type groupsin terms of A1/Ti ratio. Changesin
A1/Ti will be controlledby the degreeof partial melting
and, as A1 is buffered, decreasingthe degreeof partial
melting will decreasethe AI/Ti ratio by increasingthe
relative abundanceof moderatelyincompatibleTi. Either
the individualsourcesfor A-type and C-G-typebasaltshad
distinct mineralogies (not likely from trace element
evidence)or the A-type lavasrepresenta somewhatlower
degreeof partialmeltingthanthe C-G-typebasalts.
8. CRYSTAL FRACTIONATION AND
THE HIDDEN CUMULATES
8.1. Fractionation
Xenolithsare presentin someof the 122 Ma C-G-type
basementbasaltsin centralMalaita, particularlyat deeper
stratigraphiclevels in the Kwaio inlier. These xenoliths
are gabbroic(mainly plagioclaseand clinopyroxene,with
minor olivine and spinel) and anorthositic. They provide
graphicevidencefor crystalfractionation,consistentwith
major-elementdatafor the Malaitan and otherOJPbasalts,
which show that all A-type and C-G-type lavas are
evolved; for example, whole-rock molar Mg-number (=
Mg/[Mg+Fe2+], assumingthat 75% of the total Fe is Fe2+)
is 0.50 to 0.29. Preliminary studyindicatesplagioclaseis
the predominantphasein the gabbroicxenoliths,whereas
the basalt major-element compositions suggest that
fractionation of plagioclase was comparatively minor:
althoughEu may undergoa slightdecreasetowardthe top
of the A-type (1.3 ppm -• 1.1 ppm) and C-G-type (1.2 ppm
-•0.9 ppm) portionsof the stratigraphicsection,Sr is not
depletedsignificantly(e.g., Figure 6), and A1203showsa
restrictedrange (most sampleshave 13.5-14.5 wt%) and
doesnot correlatewith Eu or Sr. Plagioclasephenocrysts
are fairly commonin the lavas,but clinopyroxeneis the
dominantphenocrystand appearsto have been a major
fractionatingphase. For example, Sc abundancesrange
from 31-59 ppmandCr contents from25-200 ppmfor A-
type and C-G-type(high-MgOsampleML475, with Cr =
446 ppm, exhibitsa cumulatetexture),with two basalts
from SantaIsabel and the 90 Ma samplesfrom Site 803
containingbetween200 and 300 ppm Cr (Figure 9b,c).
Olivine phenocrystsare rare and MgO-Ni correlations
poor,but the generallylow Ni contents
of the lavas
202 THE ONTONG JAVA PLATEAU

9e) indicateolivine was removed,probablyas an early From calculated parental magma compositions
fractionatingphase. In general,correlationsof major corresponding to 15-30% total melting, we modeleda
elementswith typicalfractionationindices,suchas MgO broadcrystallization pathway(Figure 12). The liquidus
andMg-number,areratherpoor;in part,thislikely reflects phasesused and their relative proportionswere derived
a lack of strongolivine controlon the analyzedbasaltsat from petrography(e.g., the range of clinopyroxene and
this evolutionary stage, although effects of seawater- plagioclasecompositions usedwere thosemeasuredby
alterationon the lavasand open-system magmaplumbing electronmicroprobefor phenocrystsin the Malaita flows;
networks[cf. Mahoneyet al., 1993; Tejadaet al., 1996a,b] C. R. Neal, unpubl.data,1995)andexperimental petrology
cannot be discounted. (e.g.,theknowncontraction of the olivineandplagioclase
As primaryOJPmagmacompositions areunknown,it is liquidus fieldsandexpansion of thepyroxene fieldsduring
difficult to quantify the amountof crystalfractionation; crystallization of tholeiiticmagmaunderpressure, but <20
however, estimates can be obtained by combining kbar [e.g., Yoder and Tilley, 1962; O'Hara and Yoder,
petrographicand experimentalpetrologic data. Partial 1967;BVSP, 1981]). Farnetaniet al. [1996] modeledthe
melting appearsto have been initiated in the garnet crystallization of threeexperimentally derivedmeltsfrom
stabilityfield andconcludedin the spinelstabilityfield, as spinel peridotitesunder conditionsapproximatingthose
discussed above. We estimated the major-element expected for melts from a plume head. The main
compositionof OJP parental magmas by using the crystallizingphasespredictedwere olivine, clinopyroxene,
experimentalresults (Table 2) of Falloon and Green spinel, plagioclase,and orthopyroxene. With increasing
[1988] for melts derivedfrom garnetperidotite(30 kbar) depth of crystallizationthe cumulate was predictedto
and Hirose and Kushiro [1993] for melts from spinel change from olivine-orthopyroxene gabbro, through
peridotite(20 kbar). Ratherthana singlecomposition, we troctolite and/or leucogabbro, to melanogabbro,
calculateda range of potential parental compositions pyroxenite,or clinopyroxene-norite.In our modeling,five
assuming 5%, 10%, and 15% partialmeltingin the garnet stagesof crystallizationwere assumed,eachcrystallizinga
stabilityfield, and that the calculatedmeltsthen mix with volume equal to 10% of the startingmagma: Stage 1 =
partial melts from the spinel stability field in 5% 100% olivine (Fo90);Stage2 = 70% olivine (FoBs)+ 30%
increments, the total from both fieldsnot exceeding30% spinel;Stage3 - 50% olivine (FOB0),50% clinopyroxene;
of partial melting (the maximum amountsuggestedby Stage4 = 95% clinopyroxene,5% plagioclase;Stage5 =
Mahoney et al. [1993] and Tejada et al. [1996a]). For 90% clinopyroxene,5% plagioclase,5% orthopyroxene.
example,5% of garnetperidotitemelt mixed with 20% of Figure 12 showsthe calculatedliquid evolutiontrendsin a
spinelperidotitemelt corresponds to 25% of totalmelting plot of MgO/TiO2 vs. CaO/A1203 for several different
with a 1:4 proportionof garnetand spinelperidotiteend- model parental melts; it can be seen that compositions
members. This approachyields picriticmeltswith high similar to thoseof most of the OJP basaltsare generated
MgO contentsaveraging16 wt%. primarily during Stage4 and at the beginningof Stage5.
This result is consistentwith the paucity of olivine (and
TABLE 2. Melt CompositionsDerived From Garnet-and Spinel-
orthopyroxene) phenocrysts in the lavas, and the
PeridotiteStabilityFieldsand Used in Major ElementModeling
predominance of clinopyroxene and plagioclase
Garnet
Peridotite Spinel
P•ridotite fractionationwhich hasoverprintedevidenceof the earlier
Melt• Melt2 olivine removal (e.g., Figure 9e). The modelingindicates
that most OJP lavas result from 30-45% of crystal
SiO2 45.83 47.47 fractionation and that the xenoliths seen in some flows in
TiO2 1.08 0.75 centralMalaita correspondto relatively late stagesof this
A1203 11.70 15.53 process(Stage4 or 5).
Cr203 0.42 0.21
FeO 9.04 8.51 8.2. The Hidden Cumulates

MgO 19.99 13.94

CaO 10.66 11.11
The cumulate assemblagederived from magmatic
evolution similar to that modeled above would consist

Melt compositionfrom MYP-90-40 at 35 kbar and 1600øC primarily of early olivine and later clinopyroxene,with
[from Falloon and Green, 1988]. lesseramountsof plagioclase,orthopyroxene, and spinel.
Melt compositionfrom KLB-1 at 20 kbar and 1375øC[from Corresponding rock types could include dunite,
Hirose and Kushiro, 1993]. pyroxenite,and gabbro[cf. Farnetan.iet al., 1996].
 NEAL ET AL. 203

20 I I I I I I

15%Partial
Melts 25%Partial "•
....... C
15
_ Melts
_
ß A-Type
[] C-G-Type
X Site 803
10
O Santa Isabel

20% Partial b 30%Partial d

Melts Melts
15-

10 -

5 -

o
0.65 0.70 0.75 0.80 0.85 0.90 0.95
0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.60

CaO/A1203
Figure12. Modelingof theMgO/TiO
2 andCaO/A1203variation
in OJPbasalts
bycrystalfractionation.
Parental
meltsareestimated
fromtheexperimental
results
ofFalloonandGreen[1988]formeltsderived
fromgarnetperidotite
(30kbar)andHirose andKushiro[1993]formeltsfromspinelperidotite
(20kbar).Thesecompositions
arecombined
in 5%increments
to givethetotalamountof partial
meltatthetopof eachgraph,withthemaximum amount of melt
fromgarnetperidotite
notexceeding
15%andtheresulting parentalwaysbeingahybridofthetwomeltcompositions.
Fractional
crystallization
occurs
infivestages,eachcrystallizing
a volumeequalto 10%ofthestarting
magma:Stage
1 = 100%olivine(Fo90);Stage2 = 70% olivine(Fo85)+ 30% spinel;Stage3 = 50% olivine(FOB0), 50%
clinopyroxene;
Stage4 = 95%clinopyroxene, 5%plagioclase;Stage5 = 90%clinopyroxene,5%plagioclase, 5%
orthopyroxene.

MgO contentof 30.1 wt% is calculatedfor the bulk
cumulate by using the estimatedmean amount of
fractionationand corresponding
total model fractionating
assemblage.Assuming(1) that, as above, the average
parentalmagmaMgO = 16 wt%, (2) that the 7.0 wt%
averageMgO contentof OJPsamplesis representative of
eruptedliquidson theplateauasa whole,and(3) thatthe
30.1 wt% MgO value is representative of the bulk
cumulatepile, a roughestimateof the total amountof
cumulatesextractedfrom the magmascan be made by
massbalance[e.g., Cox, 1993]'
Cp = CLXL + CcX C
whereCp,Cid,andCc arethe concentrations of an element,
in this caseMgO, in the parentalmagma,eruptedliquid,
and cumulate, respectively;and Xi• and Xc are the
respective massfractionsof eruptedliquidand
204 THE ONTONG JAVA PLATEAU

The relativemassof cumulatesto eruptedliquidsis km s-1(Figure2) [Furumotoet al., 1976]. The thickness
of this seismiclayer has been estimatedto be 9-16 km
Xc/Xk = (Ck - Cp)/ (Cp - Cc) [Furumotoet al., 1976; Hussonget al., 1979; Carlsonet
al., 1981; Coffin and Eldholm, 1993]. Over the years,the
layer hasbeenascribedto a varietyof possiblerocktypes
The ratio Xc/XI• givesan estimateof the proportionof
from, for example,garnetgranulitesto cumulategabbros
crystalcumulatesnecessary to lower the MgO contentof
to underplatedbasaltmagma(i.e., non-cumulategabbros)
the primary magmafrom 16.0 wt% to the average7.0
[Houtzand Ewing, 1976;Nixon and Coleman,1978;Neal
wt%. Using the abovevalues,the ratio is 0.64; the total,
and Taylor, 1989;Carlsonet al., 1981;SchaeferandNeal,
combined fractionatingassemblageconsistsof 49.1%
1994; Farnetani et al., 1996; Ito and Cliff, 1996].
olivine, 40% Ca-rich clinopyroxene,5.5% spinel,5.0%
Compositionsbetweengabbroand pyroxenitewouldhave
plagioclase,and 0.4% orthopyroxene.This assemblage P-wave velocities between 7.4 and 8.1 km s-• at pressures
corresponds to a wehrlitic to pyroxeniticcumulateof
of •10 kbar [e.g., Carmichael,1989]. Therefore,asnoted
density•3.25 g cm-3. In generalagreement, Cooperet al. above, mixtures of gabbroicand pyroxenitic cumulates
[ 1986] suggested a densityfor lowerOJPcrustof 3.0-3.25
produced by OJP magmatism,or their metamorphic
g cm-3, whereasGladczenkoet al. [1997] estimateda derivatives,are likely to be responsiblefor much of the
lower crustaldensityof 3.25-3.30 g cm-3;both studies
basalhigh-velocitycrustallayer.
utilizedgravitydata. Conversion of thesedensitiesto P- Significant amounts of intruded and underplated
wave velocities (Vp) indicatesthat the lower OJP crust
materialundoubtedlycontributeto the thicknessof the OJP
shouldexhibitVp >_7.8 km s-l, which is not the case(see
[Tejada et al., 1996a; Ito and C//fi, 1996] as, probably,
Figure 2). However,as notedby Fountainet al. [1994],
does some buried preexisting oceanic crust, thus
reducedaverageVp can be obtainedby havinga hetero-
complicatingsuchsimplecrustalmodels. Moreover,the
geneouslower crust. This may be achievedby the
relative amountsof magmatism(volcanism,intrusion,and
presence of a mixtureof gabbroandpyroxenite/wehrlite or
underplating)associatedwith the 122 and 90 Ma events
basaltand pyroxenite/wehrlite in the lower OJP crust,or are unknownand, as notedearlier, the possibilityof some
their metamorphicderivativesof granuliteand possibly
magmatism occurring throughoutthe 30 m.y. period
eclogite[cf. Fountainet al., 1994].
betweenthe two eruptiveeventscannotbe discounted at
As discussed earlier,the OJP may partlyhave formedat
present. Intrusionand underplatingassociated with the
a mid-oceanridge but part, and perhapsmuch, of the seamount and Maramasike volcanism on the plateau,
plateaumusthavebeenemplacedoff-axis.For simplicity, possiblycausedby the passageof the OJP over the
we havetakentwo extremes:(1) emplacement of the entire
Samoanhotspot(which took place from •60-30 Ma [e.g.,
plateauat a mid-ocean ridge(no preexisting oceaniccrust) Yah and Kroenke, 1993]), was probablyrelativelyminor
and (2) emplacement "off-axis"with a 7.1-km preexisting
but presentlyis impossible to quantify. A furthercaveatis
oceaniccrust [White et al., 1992; Coffin and Eldholm,
that the OJP lavasstudiedto daterepresentonly the upper
1993]. In addition, minimum and maximum average
levels (_<4km) of the eruptive sectionand may not be
thicknesses of the OJP are taken as 25 km and 36 km,
representative of earlier,deeperlevels,whichmaybe more
respectively(seeSection2).
picriticin composition [e.g.,Storeyet al., 1991].
Results of our "hidden cumulate" modeling are as
follows. For an off-axis emplacement(presenceof a 7.1-
km-thickpreexistingoceaniccrust),cumulatethicknesses 9. UPLIFT AND SUBSIDENCE OF THE OJP
vary between 7.0 km (assuminga 25-km total OJP
thickness)and 11.3 km (assuminga 36-km total OJP 9.1. Paleodepths
thickness), whereas corresponding erupted basalt
sequences would be 10.9 km and 17.6 km, respectively. The OJP is roughly isostaticallycompensated[e.g.,
For an on-axisemplacement (no preexisting oceaniccrust), Sandwelland MacKenzie, 1989] and most of it stands2-3
cumulate thicknesses are 9.8 km and 14.0 km, with km above the surroundingseafloortoday (althoughstill
corresponding basalticsequences of 15.2 and 22.0 km, some 1700 meters or more below sea level). Subsidence
respectively.For comparison, recallthat the thicknessof rates for normal seafloorof similar age [e.g., Stein and
seismicLayer 2 is estimatedat > 12 km over muchof the Stein, 1992] suggestthatthe plateaushouldhavesubsided
high plateau(Figure2). some 3 km since its formation. Remarkably, however,
Seismicrefractionstudiesof the highplateauidentifieda althoughthereare indications
thatthe north-central region
basalcrustallayer with high P-wavevelocityof 7.5-7.7 of the high plateauhas subsideda few hundred
 NEAL ET AL. 205

more than the northeasternmargin since the Cretaceous redepositionand submarineerosioncan be ignored,and,
[Berger et al., 1992], little evidenceis presentfor any most importantly, that subsidencefollowed the normal
significant amounts (i.e., y l km) of post-emplacement seafloortime-depthcurve) are not valid in this case. In
subsidence at the locations where basement has been view of other considerations outlinedbelow, we interpret
sampled. Rather,normalfaultsat the northernand eastern these results as further evidence that the OJP did not
margins of the OJP indicatethat the adjoining abyssal subsideas expectedof normal oceaniccrust. Therefore,
seafloorin theseregionshassubsidedsomewhatrelativeto these calculations yield only minimum emplacement
the plateau [e.g., Kroenke, 1972; DSDP Leg 30 unpubl. depths.
reflection profiles; Kroenke et al., 1986; Hagen et al., Strong evidence for non-emergenceof the OJP can be
1993]. We hypothesize that the less than expected found in the basement lavas in Santa Isabel and Malaita, as
subsidenceof the high plateau could reflect (1) partial well as at the drill sites. Lithological and chemical
thermal•and magmatic "rejuvenation"at •90 Ma and evidence suggestsemplacementnear or even below the
during the OJP's •60-30 Ma passage[e.g., •m and calcite compensationdepth (CCD) in somecases[Hughes
Kroenke, 1993] over the Samoanhotspot;(2) the near- and Turner, 1977; Kroenke et al., 1991; Hawkins and
neutralbuoyancyof the "hiddencumulates"(3.25 g cm-3is Barron, 1991; Tarduno, 1992; Saunders et al., 1993].
generallyintermediatebetweenthe densityof oceaniccrust Field evidence from the basement sectionsof Malaita, on
and upper mantle, whether fertile or melt-depleted)that the southernflank of the OJP, demonstrates that very little
would have fractionated from OJP magmas while the interflow sedimentis present;that which existsis siliceous
plateauwas elevatedin responseto the physicalpresence rather than calcareous and the sediments overlying
of the plume head (see below); (3) the buoyingeffect of basement are pelagic mudstones, suggesting that
the melt-depleted(and thereforelessdense)uppermantle emplacementoccurredbelow the •122 Ma CCD in this
keel beneaththe OJP; and/or(4) the "flexural bulge"of the area [Saunderset al., 1993; Neal et al., 1994, 1995]. An
OJP along the plateau'ssouthwestmargin (a responseto Aptian bathyal limestonelies abovebasementat Site 289
collision with the Australian Plate [e.g., Coffin e! al., [Andrewsand Packham, 1975], whereasat Site 807, on the
1996]), which would help maintainthe elevationof the northern flank of the plateau, a thin limestone is
plateau,at least in this region.Also, Ito and Cliff [1996] intercalatedbetween the A-type and C-G-type basalts,
recently suggestedthat, rather than just a 90 Ma indicating late-stage eruption depth was at least
rejuvenation,subsidencemay have been temperedby temporarily above the CCD; however, the sediments
continved underplating of the plateau for •30 m.y. immediatelyabovebasementare not calcareous [Kroenke
betweenthe eruptiveevents. e! al., 1991]. The depthof the CCD duringthe Cretaceous
If it is assumed for modeling purposes that post- is poorly known and fluctuatedwidely, but a rough
emplacementsubsidence of the plateauwas, in fact, similar estimatefor 122 Ma is •2-3 km [e.g.,Arthure! al., 1986].
to that of normal oceaniccrust, a rough "prediction"of From CO2 contentsin Site 807 glasses,Michael and
original depth of the OJP's surfacecan be obtainedfor Cornell [ 1996] estimatedthat eruptiondepthsof lavas at
each drill site on the plateau. This approachwas usedto thissitewere in the 1000m (A-typelavas)to 2500 m (C-
good effect in estimatingthe uplift associatedwith the G-type) range; note that many of the lavas may have
formation of the Kerguelen Plateau [Coffin, 1992]. flowedconsiderable distances fromtheirpointof eruption,
Applicationto the OJPyieldsresultswhich are crudebut at so that the eruptiondepthand final emplacement depth
leastgenerallyconsistentwith emplacementof the basalts may not be the same. However, these depthsand the
well below sea level (e.g., Site 288 =-482 meters, presence of limestones
at Sites807 and289 may implythat
assuminga geophysicallydeterminedsedimentthickness the axis of the OJP plume head, above which maximum
of 1090 m; Site 803 = -930 meters; Site 807 = -520 dynamicuplift would be expected,was situatedcloserto
meters). However, a largediscrepancy existsfor Site 289 the northernend of the high plateauat 122 Ma (i.e., the
inthatanemplacement
ofbasement
above
sealevel(+202 large,asyet unsampled,domalregionof shallowerdepths
meters) is indicated, which is in direct contrast with to the southwestof Site 289 [cf. Tejadae! al., 1996a]).
lithologicalevidence(i.e., bathyalAptian limestoneabove Moreover,becausethe sectionof A-type basaltsis much
basement,lack of evidence for erosional surfaces,lack of thickerat Malaita than at Site 807, the principalsite(s)of
vesicularity or oxidation in the basalt recovered,etc.). A-type eruptiveactivity may not have coincidedwith the
This indicatesthat the assumptions
used(i.e., sedimentand plume axis.
structural,conditions at each site reflect regional In conclusion, the combination of constructional
conditions, complications arising from sediment magmatismand dynamic uplift from impingementof
206 THE ONTONG JAVA PLATEAU

plume head was apparently insufficient to raise the course, if significantamountsof OJP crust formed at 90
basementsurface of much of the OJP to shallow depths, Ma, in the 122-90 Ma period, or after 90 Ma, then this
althoughpartsmay have been shallow,particularlythe as minimum estimatewould be lower still. Two examplesof
yet unsampledcrestal regions of the high plateau and dynamicuplift modelingare presentedbelow.
eastern salient. This situation contrasts with those for 9.2.1. Model 1. Experimentsby Griffiths et al. [1989]
Wrangellia [e.g., Richards et al., 1991; Lassiter et al., followed the evolutionof surfacetopographyas a buoyant
1995] and the Kerguelen Plateau [Coffin, 1992], which dropletrose througha viscousliquid. Maximum surface
contain evidence of extensive emergenceand subaerial uplift was observedwhen the leading edge of the diapir
erosion in basal sediment and basalt stratigraphy. was 0.2 diapir diameters beneath the surface and was
However, Wrangelliawas built upon an islandarc and the quantifiedas
KerguelenPlateauformed in a young,narrow oceanbasin
andmay containfragmentsof continentallithosphere[e.g., 0.25(1- X)(ApDo )
Hmax
Storey et al., 1992; Mahoney et al., 1995; Hassler and
Shimizu,1995; Opertoand Charvis,1996].
where:Hmax= maximumuplift; )•: (Pm- Pl)/(Pm- OD);Pm
9.2. Estimatesof Initial Dynamic Uplift = upper mantle density (3.3 g cm-3); Pl = lithospheric
density (3.15 g cm-3); PD = plume head density (varied
Impingementof a plumeon the lithosphere will produce between3.1 and 3.15 gcm -3 resultingin )t,valuesbetween
temporary surface uplift, even if the plume surfaces 0.75 and 1.0; seeFigure 13); Dp = Pm- PD;Do = diameter
beneatha mid-oceanridgeoRoughestimatesof the amount of sphericalplumehead.
of such uplift can be obtainedby applicationof results As expected,the maximum amountof uplift is predicted
from experimentaland theoreticalstudies[e.g., Olsonand from the largestplume head (Figure 13). However, this
Nam, 1986; Griffithset al., 1989; Hill, 1991;van Kekenet approachimpliesthat maximumdynamicuplift associated
al., 1993; Ribe, 1996]. Such estimatesrequire that the with OJP formation would barely exceed 3 km and could
volume of the plume head be known or assumed;some be much less.
limits on this volume can be obtained from the estimated 9.2.2 Model 2. Hill [1991] quantifieddynamicuplift
volume of the plateau and the degreeof partial melting above a spreading (flattening) plume head and
involvedin producingthe basalts. demonstratedthat the amountof uplift is sensitiveto the
For modelingpurposes,we use a minimum OJP areal initial thermal structureof the overlying material; this is
value of 1.5 x 106km2. (Coffin andEldholm [1994] gavea manifested by the temperaturedifference between the
largerestimateof 1.86 x 106km2, includingpossibleOJP- plumeheadandambientmantle(DT).
related sequencesin the Nauru, East Mariana, and Lyra
basins, but here we consider only the plateau itself, as E = ctATL

defined by the 4-kin depth contour.)A maximum volume

of 5.4 x 107 km3 results from using the maximum esti- Here, E = amount of surfaceuplift; ct = coefficientof
mated averagecrustalthicknessof 36 kin, togetherwith thermal expansion(3 x 10-5 øC-•); AT-- averageexcess
the assumption of no preexistingoceaniccrust(Section2). temperatureanomaly (in øC); L = thicknessof the
A minimum volume of 2.7 x 107 km3 is obtained by anomalously hot layer(flattenedplumehead).
assuming the OJP formed completely on preexisting In Figure 14, the sameestimatesof partialmeltingand
oceanic crust and taking the estimatedaverage crustal plateauandplume-headvolumewere usedas in Model 1.
thicknessformed by the OJP plume to be 17.9 km (25 km The difference is that the spherical plume head is
minus 7.1 km of preexistingcrust). Assumingfurtherthat considered to have flattened to a disc with a diameter of
all of the OJP formed in the 122 Ma event and that the either 1000, 2000, or 3000 km; the smallerthe diameterof
average degree of partial melting was as low as 15% the flattenedplume head, the greaterits thicknessand,
(typical of values estimatedfor MORB), then an upper therefore,the greaterthe surfaceuplift. Also includedare
limit on the plume-headvolume would be 3.6 x 108km3; resultsfrom assuminga maximumand minimumAT (300
for a sphericalhead,this corresponds to a diameterof 883 and 100øC, respectively)between plume and ambient
km. A "pseudominimum"plume-headvolume of 9.0 x mantle. Maximumsurfaceuplift of 4.1 km is obtainedfor
107km3, with a sphericaldiameterof 555 km, is obtained a flattenedplumeheadof minimumdiameter,AT = 300øC,
by assuming30% partial melting and using the lower andpartialmeltingof 15%. If theaveragedegreeof partial
estimateof OJPvolume [cf. Coffin and Eldholm, 1993]; of meltingis 20-30%, in the rangeof valuesconsidered
 NEAL ET AL. 207

3.5
uplift values for plateaus estimatedby Farnetani and
Richards[ 1994], who assumeda greaterAT (350øC)and a
3.0 Preexisting
large sphericalplume head diameter(800 km)). Again,
Crust (7.1 km) uplift could be much less.
25

25%O"'. Maximum
OJP 9.3. Effect of Added Oust
20 30% Volume

15 Although by itself dynamic uplift appearsincapableof

- /•'. -.:::. raisingmuch of the OJP from ridge or abyssaldepthsto
_ MinimumOJP above sea level, addition of as much as 36 km of new crust
Volume
might be assumed to cause widespread emergence.
However, overwhelmingevidenceto the contraryis seenat
a
_

all presentlysampledlocations.Thus, on average,the bulk

density of the OJP must be slightly greaterthan that of
0.0
i i

15%
normal ocean crust simply thickenedby a factor of up to
20%
3.0 25% No PreexistingCrust five or so. For example,massbalancecalculationssuggest

2.530% that up to 4 km of surfaceuplift would resultfrom simply

overthickeningoceaniccrust(densityof 2.9 g cm-3). If the
".:'.',•• MaXl'mum
OJP densityis increasedto 3.08 g cm-3(the averagedensityof
2.0 " e
the OJP we calculateon the basisof P-wave velocities),the
1.5
surfaceuplift is reducedto betweenonly 1.8 and 3.0 km,
dependingupon the thicknessassumedfor the plateau.
1.0 MinimumOJP Indeed, averagecrustaldensities(estimatedfrom published
V
P-wave velocities[e.g., Bjarnasonet al., 1993; Opertoand
Charvis, 1996; White et al., 1996] using the methodsof
0.5
b Bott [1982]) indicate that bulk crustal density increases
0.0 from Kerguelen(• 2.8 g cm-3)to Iceland(• 2.9 g cm-3)to
0.75 0.80 0.85 0.90 0.95 1.00
the OJP (• 3.1 g cm-3).Although Iceland has an average
density greater than that of Kerguelen, it is emergent
becauseof a greater crustal thickness(22-35 km [e.g.,
Figure 13. Estimates of surface uplift resulting from Bjarnason et al., 1993; White et al., 1996] vs. 20-25 km
impingementof the 122 Ma OJP plume head upon the [e.g., Operto and Charvis, 1996]) and because of the
lithosphereusing the model of Griffiths et al. [1989]. The
dynamic uplift afforded by the plume beneaththe Mid-
maximumvolumeof the plumeheadis calculated by assuming a
minimum and maximumaveragethicknessof the OJP of 25 and
Atlantic Ridge. Much of the KerguelenPlateauwas also
36 km, respectively.The size of the plume head is directly originally at shallow depths or subaerial [Coffin, 1992]
proportional to the amountof surfaceuplift. Calculations
in (a) because of plume-driven dynamic uplift and crustal
assume thatthe OJPwasbuilt upon7.1 km of preexisting oceanic thickening; as mentioned earlier, it may contain some
crust (off-ridge setting), whereas (b) assumesthat the total continentallithosphericmaterial,unlike Iceland.
thickness
of the OJPwasgenerated
by the plumeevent(ridge We conclude that the key to the initial depth of OJP
setting). Dashed and solid lines representthe minimum and emplacementis largely in the high-velocitybasal crustal
maximumOJP volume,respectivelyø The percentages represent
layer, argued here to have been formed from hidden
the assumedamountsof partial melting that generatedOJP
magmas(i.e., the lowerthe assumed degreeof melting,the larger cumulatesplus intrudedand underplatedgabbros.Also, the
the plume head required): 15% = open square;20% = open extreme thickness of the OJP provides conditions
triangle;25% = filled circle;30% = opendiamond. conduciveto granulite or, possibly,eclogitedevelopment
[Fountain et al., 1994; Rudnick and Jackson, 1995;
Saunders et al., 1996], both of which could contribute to
likely for OJP lavas[e.g., Mahoneyet al., 1993; Tejadaet temperingOJP uplift. Mafic granulitexenoliths(including
al., 1996a; Section 7.2, this paper], then the maximum a garnet-bearingvariety) are indeedpresentin the Malaitan
surfaceuplift would be between1.0 and 3.0 km, similarto alnOites(P. H. Nixon, pers. comm., 1996) at the southern
the estimate from the Griffiths et al. [1989] approach. edgeof the plateau,and the high P-wave velocitiesof 8.4-
(Note that theseresultsare also similar to the minimum 8.6 km/s just below the Moho in the northwestern
208 THE ONTONG JAVA PLATEAU

3.5 I I I l
high plateau(attributedlargely to cumulates)andpossible
3.0 _a Preexisting
_ high-densityeclogitic material indicatedby higher than
normal mantle P-wave velocitiesjust below the Moho in
2.5 _ Crust
(7.1
km)_ parts of the high plateau (see below) would argue for
subsidence of the plateau,at leastin theseareas.In orderto
maintainplateauelevationand preventdelaminationof the

_ -'1000km-
2.0

high density material, it seems likely that a region of

1.5
relatively buoyant, melt-depletedmantle residesbeneath
AT= 100øC the plateau, most probably the residue from plateau
1.0
2000 km formation. Degree of partial melting has been estimated
from the compositionof the OJP basaltsto be between20

3000
km
............
-......
-..........
- and 30% [Mahonc)' et al., 1993; Tejada et al., 1996a;
Section 7.2, this paper]. Such high degreesof partial
melting, coupledwith the volume of the OJP, requirethat a

4.0
b NoPreexisting
Crust significantmelt-depletedzone or root be present(barring
convective removal) in the upper mantle beneath the
_ _

plateau. This mantle root would cause a decreasein

3.O
_ __ o seismic velocities beneath the OJP and this has indeed
beenreportedrecentlyby Richardsonand Okal [1996].
An estimateof the depletedmantleroot thicknesscanbe
_ 1000 krn
2.0
made from the mass-balanceapproach of Marks and
AT=100øC Sandwell [1991], who used the ratio of geoid height to
1.0
topography to develop a two-layer (a layer of

3000
km
2000km ........ -......[]........... [] overthickenedcrust and a low-density, depletedmantle
root) Airy compensationmodel:
0.0
1S 20 25 30
(pc-Pw)h = (Pm-Pc)r + (pm-p•)t
% Partial Melt

Figure 14. Surfaceuplift estimatedfor impingementof the 122 Here, h = elevationof plateau above the oceanfloor (2.5
Ma OJP plume headuponthe lithosphereusingthe modelof Hill km); r- thickenedlayer at the baseof the crust,as defined
[1991] and varying the amount of assumedpartial melting by Marks and Sandwell [1991] (taken as thicknessof the
between a minimum of 15% and a maximum of 30%. Maximum cumulate layer in our model); t = thickened depleted
and minimum plume head sizes are calculatedas in Figure 13. mantle layer; Pc= OJP cumulatedensity(3.25 g cm-3);Pw
Values of AT are betweenthe plume head and ambientupper = seawaterdensity(1.025 g cm-3);Pm= mantledensity(3.3
mantle; 100øCis the assumedminimum(dashedlines)and300øC g cm-3); p• = melt-depletedmantle density(3.24 g cm-3).
is the assumedmaximum(solid lines). Solid symbols= OJP is 25
km thick; open symbols = OJP is 36 km thick. Assumed
The thicknessof the depleted mantle layer, t, can be
calculated:
flattenedplume head diameters:squares= 1000 km; triangles=
2000 km; circles= 3000 km. (a) Assumesthat the OJP was built
upon 7.1 km of pre-existing oceanic crust (off-ridge setting),
whereas (b) assumesthat the total thicknessof the OJP was
t - (Pc- Pw)
(Pm -Pl)
(1-8)h
generatedby the plume event(ridge setting).
The fraction of Airy compensationis representedby 8
and corresponds to the compensationassociatedwith the
southwestern regions of the OJP might represent cumulate layer at the base of the OJP. As cumulate
eclogitizedlowermostcrust. thicknessdependson the total material addedby the OJP
plume event(s), r varies between 7 and 14 km (Section
9.4. Melt-DepletedMantle Root to OJP 8.2). In order to maintain the mass balance, 8 varies be-
tween 0.06 and 0.12, respectively.This approachyields a
Elevation of the OJP relative to the surroundingocean thicknessof the melt-depletedmantle of between82 and
floor appearsto be persistentas the effectsof temporary 87 km.
uplift have long since dissipated. The presenceof a As a rough checkon the 82- to 87-km root thickness,the
relatively denselower crustallayer beneathmuch of the average degree of partial melting that createdit can
 NEAL ET AL. 209

estimatedby dividingthe thicknessof the OJPcrustby the steeplydescendingslabbeneaththe OJP [e.g., Cooperand
combinedthicknessof the OJP crustplus depletedmantle Taylor, 1987], there has been little time for slab-derived
root, and the result comparedwith the degreeof partial fluids to affect more than the edge of the OJP and
meltingestimatedfrom geochemicalevidence.The results substantialeclogitizationof lower crustby this mechanism
imply that the degree of partial melting required to has probablynot yet occurred(thoughit may eventually).
generate the depleted mantle root was between 17% Nevertheless,Petterson [1995], Mann et al. [1996], and
(assumingadditionof 17.9 km of materialto preexisting Petterson et al. [1997] have argued that geophysical
7.1 km thick oceancrust)and 31% (assumingadditionof evidence is consistentwith underthrustingof the leading
36 km of crust with no preexistingocean crust). These edge of the OJP beneaththe Solomonblock, even as the
valuesare in excellentagreementwith estimatesof partial upper levels are overriding it. Although superficially
melting from geochemicaldata [Mahoney et al., 1993; resemblingtrue subduction,this processis interpretedby
Tejadaet al., 1996a;Section7.2, thispaper].Furthermore, Pettersonet al. [1997] as a mechanical"wedge effect" of
our estimates of the depleted mantle root thickness the old forearcsplittingthe OJPcrustandforcingits lower
compare well with the observedregion of decreased levels downward; as such, it should be of limited
seismicvelocity beneaththe OJP [Richardsonand Okal, magnitudeand duration.
1996;W.P. Richardson,pers.comm., 1996]. Is any significantamountof eclogitepresent,then,in the
OJP? Earlier, we noted that sub-Moho P-wave velocities of
10. SUBDUCTIBILITY OF THE OJP 8.4-8.6 km s-1in parts of the high plateau(e.g., Figure 2)
might signal the presenceof eclogite immediatelybelow
the Moho. The regions where these velocities were
Assuming that subcrustal lithosphere was not
measuredare generallyfar from the Solomonarc and thus
compositionallydifferentfrom surroundingasthenosphere,
from slab-derived fluids and subduction-related stress.
Cloos [1993] concluded from buoyancy analysis that
Eclogitization in these (and possibly other) regions, if
oceanic plateaus >30 km thick would be virtually
confirmed by future deep-crustalseismic studies,must
unsubductible. Abbott and Mooney [1995], taking into
have occurred largely in the absenceof such factors,
accountthe additionalbuoying effect of a melt-depleted
following or during emplacementof the plateau'sthick
lithosphericmantle root (depletedin both garnetand Fe),
crust. Whereas substantialamounts of eclogite could
arguedthat oceaniccrustthicker than about 25 km should
eventuallyleadto delaminationof the plateau'slithosphere
be unsubductible.A critical factor in subductibilityis the
in locationsfar from a subductionzone [e.g., Saunderset
extent to which density increaseswhen lower parts of a
al., 1996], the buoying effect of a low-density, melt-
plateau's crust encounterthe high pressureregion of a
depletedroot would tend to counteractdelamination. For
subduction zone; the density increase resulting from
the present, we emphasize that the possibility of an
transformationof lower crust to eclogite can promote
eclogite layer should be consideredcarefully in future
subductionof at leastthe lower portionsof plateaucrust.
models of OJP structure,subsidence,and eventualfate.
Saunderset al. [1996] demonstratedthat if averageOJP
crustal thickness is close to 36 km the basal levels of the
11. FUTURE RESEARCH AVENUES
OJPwouldbe in the eclogitestabilityfield. However,they
arguedthat conversionof lower crust from granuliteto
eclogite facies would be long-delayedin the absenceof To evaluatethe OJP's origin, history,and consequences
small amountsof water (kinetically importantfor crystal more fully, a numberof fundamentalquestionsremain to
nucleationand transportof ions) and directedstress[e.g., be answered. To a large extentthey require (a) a much
,,lustrheim, 1987]. They suggestedthat the stressfield more comprehensivesampling on and near the OJP of
resultingfrom arrival of a plateauat a subduction
zoneand basementand later-stageigneousrocks, and (b) state-of-
the additionof fluids escapingoff a slab subsequently the-artgeophysicalstudiesof the plateau'scrustalstructure
sinkingbeneatha plateau(followingreversalof subduction and its relation to surroundingareas. Further detailed
directionaftercollision)wouldpromoteeclogiteformation samplingis required, for example, to assessthe relative
andcauseeventualsubduction of at leastthe lowerplateau contributions of the 122 Ma and 90 Ma events, how
crust. Ultimately,only thosepartsof uppercrustthat had much--if any--volcanism occurredin the periodbetween
overridden the originalforearc(e.g.,Malaita,SantaIsabel) 122 and 90 Ma, and to documenthow representativethe
mightbe preservedasidentifiableremnantsof a plateau. limitedrangeof geochemical variationin the few presently
In the caseof the OJP,northeastward
subduction began sampledsitesis of the plateau'suppercrustas a whole.
only at 10-5 Ma (see Section5). Althoughproducinga The answers,in turn, would throw light on the size
210 THE ONTONG JAVA PLATEAU

composition of the plume head(s) thought to be Geophysical(includingmagnetic)studieswill alsohelp

responsiblefor the OJP and the relation of plateau to decipher the original setting of OJP emplacement,
emplacementto plate-motionchanges.The crestsof the particularlyits proximity to a spreadingcenterand which
high plateauand easternsalient,whichmay correspond to partsof the plateaumay have been formedessentiallyon-
the locations of the plume axis at 122 and 90 Ma, axis. Refinement of plate reconstructions for the Early
respectively,are likely to be importantfor samplingthe Cretaceous Pacific is needed to better determine the
range of geochemicalcompositionspresent. Similarly, original location of the plateau and its presumedparent
drilling in block-faultedareas and possibledipping- plumehead.
reflector sequences[Kroenke, 1972] could provide a The global impacts of OJP volcanism on the surface
meansof samplingrelatively deep into the volcanicpile, environment,marine and atmosphericchemistry,and the
possibly even into the upper levels of the cumulate biosphereare very poorly understoodat presentbut may
sequencein some places(particularlyusing a ship with have been significant. For example, widespreadearly
risercapability). Suchsamplingmightwell revealwhether Aptian and early Turonian periodsof marineblack shale
early,highlymagnesianlavas,predictedto existin plateau deposition(anoxic events) are well-documented[e.g.,
crustand similar to thoseseenin stratigraphically lower $liter, 1989] and may correspond,at least in part, to OJP
parts of the CaribbeanPlateau'seruptivesequence[e.g., emplacement. Likewise, excursionsin seawater87Sr/86Sr
Storeyet al., 1991; Kerr et al., this volume],are present.to lower valuesappearto have occurredin the Aptian and
Drilling of seismically imaged diatreme-like structures Turonian [dorieset al., 1995]. Moreover, recentresearch
would determine if some are true kimberlites and could on terrestrialplant fragmentsand coal seamssuggestsa
yield lower crustaland uppermantlexenolithssimilarto suddenincreasein atmospheric•3C/12Cat the beginningof
those from the Malaitan aln6ites, which would be critical the Aptian that may largely be attributableto OJP volca-
for understandingthe nature of the lower crust and the nism [GrOckeet al., 1997]. In conjunctionwith datingand
refractory/eclogitizedupper mantle. Systematicanalysis geochemicalwork on OJP basementrocks, further studies
of highly siderophilictrace elementsin both future and of Cretaceoussedimentsdepositedon land, at different
existingsamplecollectionswill be importantfor testingthe depthson pre-OJPedificesin the oceans(the ShatskyRise,
hypothesisof a core-mantleboundaryoriginfor the plume for example), and adjacent to the plateau should help
source,and for resolvingwhich of the geochemicaltypes revealthe OJP's role in suchenvironmentalchanges.
seenthus far may better representthe plume source(vs. Acknowledgments:Many thanksgo to the Ministry of Energy,
entrainedmantle). Water, and Mineral Resources (Geology Mapping) of the
Detailed seismic studies are essential to establish a Solomon Islands; without their logistical support, constant
guidance,diplomacyin the bush, and enthusiasmfor field work,
clearerpictureof crustalstructure(both upperand lower)
this study would not have been possible. Michael Storey,
and thickness,the compositionof the lower crust,as well TadeuszGladczenko,and co-editorMike Coffin providedcritical
as the importanceof magmaticunderplatingandthe deep- reviewswhich greatlyimprovedthis contribution.Thanksalsoto
level interaction between the southern OJP and the
S. Schaefer,J. Rigert, and J. Westerink for initial direction in
Solomon block. Coupled with paleodepth data from cumulateand uplift modeling, and to N. ttulbirt, P. Wessel,and
sedimentsat siteson andadjacentto the OJP(includingthe M. Tejada. This study was supportedby NSF grantsEAR 93-
crestal regions of the high plateau and easternsalient, 02471 and ECS92-14596 to CRN, EAR 92-19664 to JJM, and
EAR 93-02472 to RAD.
which may have been at relatively shallowlevelsduring
emplacement),such informationis requiredfor a better
understanding of uplift and subsidencehistory--andof the REFERENCES
probabilitythat the OJP will ultimatelybecomeaccretedto
continentalcrustor at leastpartially subducted.Evidence Abbott, D., and W. Mooney, The structuraland geochemical
is now availablethat lower crust at the southernedge of evolution of the continentalcrust: Support for the oceanic
the plateauis underthrustingto the southwest[Mann et al., plateaumodelof continentalgrowth,Rev.Geophys.Suppl.,US
1996; Phinney et al., 1996; Cowley et al., 1996; Petterson Nat. Rept. to IUGG 1991-1994, 231-242, 1995.
Andrews, J. E., and G. H. Packhamet al., Init. Repts.Deep Sea
et al., 1997]. Whether this process,which presently
Drill. Proj., 30, 753 pp., 1975.
appearsminor, could evolve into true subductionmay
Arthur, M. A., W. E., Dean, and S. O. Schlanger,Variations in
dependlargely on the extentof lower crustaleclogitization the global carbon cycle during the Cretaceousrelated to
and on the rate at which it may be occurringat present. climate, volcanism, and changesin atmosphericCO2, in The
Future seismic investigationsshould help evaluate the Carbon Cycle and Atmospheric C02: Natural Variations
amountof eclogitization. Archeanto Present.Geophys.Monogr. Set., vol. 32, edited

E. T. Sundquist and W. S. Broecker, pp. 504-529, AGU
Washington,D.C., 1986.
Austrheim,H., Eclogitizationof lower crustalgranulitesby fluid
migration through shear zones,Earth Planet. Sci. Lett., 81,
221-232, 1987.
Auzende, J-M., L. W. Kroenke, J-Y. Collot, Y. Lafoy, and B.
Pelletier, Compressivetectonismalong the easternmargin of
Malaita island (Solomon Islands), Mar. Geophys.Res., 18,
289-304, 1996.
BVSP - BasalticVolcanismStudyProject,Basaltic Volcanismon
the Terrestrial Planets, 1286 pp., PergamonPress,1981.
Batiza, R., Trace-elementcharacteristics of Leg 61 basalts,Init.
Repts.DeepSeaDrill. Proj., 61,689-695, 1986.
Bercovici,D., and J. J. MahoneyJ. J., Doubleflood-basaltevents
and the separationof mantle-plume
headsat the 660 km
discontinuity,Science,266, 1367-1369,1994. 105, 715-737, 1993.
Berger, W. H., and T. C. Johnson,Deep sea carbonates: Coffin, M. F., Emplacementand subsidenceof Indian Ridge
Dissolution and mass wasting on Ontong Java Plateau, plateausand submarineplateaus,in Synthesisof Resultsfrom
Science, 192, 785-787, 1976. Scientific Drilling in the Indian Ocean, Geophys.Monogr.
Berger,W. H., L. W. Kroenke,L. A. Mayer,J. Backman,T. R. Ser., vol. 70, edited by R. A. Duncan, D. K. Rea, R. B. Kidd,
Janacek,L. Krissek,M. Leckie, and M. Lyle, The recordof U. von Rad, and J. K. Weissel, pp. 115-125, AGU,
OntongJavaPlateau:Main resultsof ODPLeg 130,Geol.Soc.
Am. Bull., 104, 954-972, 1992.
Berger,W. H., R. M. Leckie, T. R. Janacek,R. Stax, and F.
Takayama,Neogenecarbonate sedimentation
on OntongJava
Plateau:Highlightsand openquestions, Proc. OceanDrill.
Prog.,Sci.Results,130, 711-744, 1993.
Birkhold-VanDyke, A. L., C. R. Neal, J. C. Jain, J. J. Mahoney,
and R. A. Duncan, Multi-stage growth for the Ontong Java
Plateau(OJP)? A progressreportfrom San Cristobal(Makira),
SolomonIslands(abstract),Eos Trans.AGU, 77, Fall Meeting
Suppl.,p. 714, 1996.
Bjarnason,
I. T., W. Menke,6. G. F16venz,
andD. Caress,
Tomographic image of the mid-Atlantic plate boundary in
southwesternIceland,d. Geophys.Res.,98, 6607-6622, 1993.
Bott, M. H. P., The Interior of the Earth: Its Structure,
Constitution, and Evolution, 382 pp., Elsevier, Amsterdam,
1982.
Campbell, I. H., and R. W. Griffiths, Implications of mantle
plume structure for the evolution of flood basalts, Earth
Planet. Sci. Lett., 99, 79-93, 1990.
Cande, S.C., J. L. Labrecque,R. L. Larson,W. C. Pitman III, X.
Golovchenko,and W. F. Haxby, Magnetic Lineationsof the
World's Ocean Basins (charO, American Association of
PetroleumGeologists,Tulsa, Okla., 1989.
Carlson, R. L., N. I. Christensen,and R. P. Moore, Anomalous
crustal structuresin ocean basins:Continentalfragmentsand
oceanicplateaus,Earth Planet. Sci. Lett., 51, 171-180, 1981.
Carmichael,R. S., Practical Handbookof PhysicalPropertiesof
Rocksand Minerals, 741 pp., CRC Press,Boca Raton, Fla,
1989.
Castillo, P., R. Batiza, and R. J. Stern, Petrologyand geochem-
istry of Nauru Basin igneouscomplex:large-volume,off-ridge
eruptions of MORB-like basalt during the Cretaceous,Init.
Repts.Deep Sea Drill. Proj., 89, 555-576, 1986.
Castillo, P., R. W. Carlson, and R. Batiza, Origin of the Nauru
NEAL ET AL. 211
Basin igneous complex: Sr, Nd, and Pb isotope and REE
constraints,Earth Planet. Sci. Lett., 103, 200-213, 1991.
Castillo, P., M. S. Pringle, and R. W. Carlson, East Mariana
Basin tholeiites: Jurassic ocean crust or Cretaceous rift basalts
related to the Ontong Java plume? Earth Planet. Sci. Lett.,
123, 139-154, 1994.
Cheng, Q., K.-H. Park, J. D. Macdougall, A. Zindler, G. W.
Lugmair, J. Hawkins, P. Lonsdale,and H. Staudigel,Isotopic
evidencefor a hotspotorigin of the Louisville seamountchain,
in Seamounts,Islands,and Atolls, Geophys.Monogr.Ser., vol.
43, editedby B. Keating,P. Fryer, R. Batiza, and G. Boehlert,
pp. 283-296, AGU, Washington,D.C., 1987.
Cloos, M., Lithosphericbuoyancy and collisional orogenesis:
subductionof oceanic plateaus,continentalmargins, island
arcs, spreadingridges, and seamounts,Geol. Soc. Am. Bull.,
Washington,D.C., 1992.
Coffin, M. F., and O. Eldholm, Large Igneous Provinces:
JOI/USSAC Workshop Rept. University of Texas at Austin
Institutefor GeophysicsTech.Rept.No. 114, 79 pp., 1991.
Coffin, M. F., and O. Eldholm, Scratching the surface:
Estimating the dimensions of large igneous provinces,
Geology,21, 515-518, 1993.
Coffin, M. F., and O. Eldholm, Large igneousprovinces:crustal
structure, dimensions, and external consequences,Rev.
Geophys.,32, 1-36, 1994.
Coffin, M. F., and L. M. Gahagan,Ontong Java and Kerguelen
Plateaux: CretaceousIcelands?d. Geol. Soc. Lond., 152, 1047-
1052, 1995.
Coffin, M. F., P. Mann, T. Shipley, E. Phinney, and S. Cowley,
Structureand stratigraphyof the southernOntongJavaPlateau
(abstract),Eos Trans. AGU, 77, Fall Meeting Suppl., p. 712,
1996.
Coleman, P. J., The Solomon Islands as an island arc, Nature,
211, 1249-1251, 1966.
Coleman, P. J., A re-evaluation of the Solomon Islands as an arc
system,CCOP/SOPAC Tech.t•all., 2, 134-140, 1976.
Coleman, P. J., and L. W. Kroenke, Subduction without
volcanism in the Solomon Islands arc, Geo-Mar. Lett., 1, 129-
134, 1981.
Coleman, P. J., and G. H. Packham,The Melanesianborderlands
and India-Pacific plates boundary,Earth Sci. Rev., 12, 197-
233, 1976.
Coleman, P. J., B. McGowran, and R. W. Ramsay, New early
Tertiary ages for basal pelagites, northeastSanta Isabel,
Solomon Islands (central southwestflank, Ontong Java
Plateau),Bull. Aust.Soc.Explor. Geophys.,9, 110-114, 1978.
Cooper, P., and B. Taylor, Polarity reversalin the Solomon
Islands arc, Nature, 314, 428-430, 1985.
Cooper, P., and B. Taylor, The spatial distribution of
earthquakes,focal mechanisms,and subductedlithosphere
212 THE ONTONG JAVA PLATEAU
the SolomonIslands, in Marine Geology,Geophysics,and
Geochemistry of the Woodlark Basin-SolomonIslands,
Circum-PacificCouncilfor Energy and Mineral Resources, Floyd, P. A., Petrologyand geochemistry
Earth Sci. Ser., vol. 7, editedby B. Taylor andN. F. Exon, pp.
67-88, Circum-Pacific Council for Energy and Mineral
Resources,Houston, Tex., 1987.
Cooper,A. K., M. S. Marlow, and T. R. Bruns,Deep structureof
the centraland southernSolomonIslandsregion:Implications
for tectonic origin, in Geology and Ofj3hore Resourcesof
Pacific Island Arcs - Central and WesternSolomonIslands, Fountain, D. M., T. M. Boundy, H. Austrheim,and P. Rey,
Circum-Pacific Council for Energy & Mineral Resources,
Earth Sci. Ser., vol. 4, edited by J. G. Vedder, K. S. Pound,
and S. Q. Boundy, pp. 157-175, Circum-PacificCouncil for
Energy and Mineral Resources,Houston,Tex., 1986.
Cowley, S., P. Mann, M. Coffin, and T. Shipley, Folds and
unconformitieswithin the Central SolomonsTrough:A record
of back-thrustingrelated to post-Mioceneshallowsubduction
of the Ontong Java Plateau (abstract),Eos Trans. AGU, 77, Furumoto,A. S., J.P. Webb, M. E. Odegard,and D. M. Hussong,
Fall Meeting Suppl.,p. 712, 1996.
Cox, K. G., Continentalmagmaticunderplating,Philos. Trans.R.
Soc. Lond., A 342, 155-166, 1993.
Davis, G. L., The ages and uranium contentsof zircon from
kimberlites and associatedrocks (abstract),ExtendedAbstr.
2nd Intl. Kimberlite Conf, SantaFe, 1977.
Duncan,R. A., Radiometricagesfrom volcanicrocks along the
New Hebrides-Samoalineament,in GeologicalInvestigations
of the Northern Melanesian Borderland Circum-Pacific
Councilfor Energy and Mineral Resources,Earth Sci. Ser.,
vol. 3, edited by T. M. Brocher,pp. 67-76, Circum-Pacific
Council for Energy and Mineral Resources,Houston, Tex.,
1985.
Ewing, J., M. Ewing, T. Aitken, and W. J. Ludwig,North Pacific
sedimentlayers measuredby seismicprofiling, in The Crust
and UpperMantle of the PacificArea, Geophys.Monogr.Ser.,
vol. 12, editedby L. Knopoff, C. L. Drake, and P. J. Hart, pp. Griffiths, R. W., M. Gurnis, and G. Eitelberg, Holographic
147-173, AGU, Washington,D.C., 1968.
Falloon, T. J., and D. H. Green, Anhydrouspartial melting of
peridotitefrom 8 to 35 kb and the petrogenesis of MORB, d.
Petrol. Spec.LithosphereIssue,pp. 379-414, 1988.
Fametani,C. G., and M. A. Richards,Numericalinvestigations of
the mantle plume initiation model for flood basaltevents,d.
Geophys.Res.,99, 13,813-13,833, 1994.
Farnetani, C. G., M. A. Richards, and M. S. Ghiorso M. S.,
Modeling crustal structure in provinces affected by plume
volcanism (abstract), Eos Trans. AGU, 76, Fall Meeting
Suppl.,p. 591-592, 1995.
Farnetani,C. G., M. A. Richards,and M. S. Ghiorso,Petrological
modelsof magmaevolutionand deepcrustalstructurebeneath
hotspotsand flood basalt provinces,Earth Planet. Sci. Lett.,
143, 81-96, 1996.
Fleet, M. E., W. E. Stone, and J. H. Crocket, Partitioning of
palladium, iridium, and platinum between sulfide liquid and
basalt melt: Effects of melt composition,concentration,and
oxygen fugacity, Geochim. Cosmochim.Acta, 55, 2545-2554,
1991.
Fleet, M. E., S. L. Chryssoulis,W. E. Stone,and C. G. Weisner,
Partitioningof platinum-groupelementsandAu in the Fe-Ni-
Cu-S system:Experimentson the fractionalcrystallizationof
sulfide melt, Contrib. Mineral. Petrol., 115, 36-44, 1993.
of oceanicintraplate
sheet-flowbasalts,Nauru Basin, Deep Sea Drilling Project,
Leg 89, Init. Repts.Deep SeaDrill. Proj., 89, 471-497, 1986.
Floyd, P. A., Geochemicalfeaturesof intraplateoceanicplateau
basalts,in Magmatism in the Ocean Basins,Spec.Publ. 42,
editedby A.D. Saundersand M. J. Norry, pp. 215-230, The
GeologicalSociety,London, 1989.
Eclogite-facies shear zones•eep crustal reflectors?
Tectonophysics, 232, 411-424, 1994.
Furumoto,A. S., D. M. Hussong,J. F. Campbell,G. H. Sutton,
A. Malahoff, J. C. Rose, and G. P. Woollard, Crustal and
uppermantlestructureof the SolomonIslandsas revealedby
seismicrefractionsurvey of November-December1966, Pac.
Sci., 24, 315-332, 1970.
Seismic studies on the Ontong Java Plateau, 1970,
Tectonophysics, 34, 71-90, 1976.
Gladczenko, T. P., M. Coffin, and O. Eldholm, Crustal structure
of the Ontong Java Plateau:Modeling of new gravity and
existingseismicdata,d. Geophys.Res.,in revision,1997.
Govindaraju,K., 1989 compilationof workingvaluesandsample
descriptionfor 272 geostandards, Geostand.Newsl.,13, 1-114,
1989.
Green, T. H., S. H. Sie, C. G. Ryan, and D. R. Cousens,Proton
microprobe-determined partitioningof Nb, Ta, Zr, Sr, and Y
betweengarnet,clinopyroxene,and basalticmagmaat high
pressureandtemperature, Chem.Geol., 74, 201-216, 1989.
Griffiths, R. W., and I. H. Campbell, Stirring and structurein
mantle startingplumes,Earth PlanetoSci. Lett., 99, 66-78,
1990.
measurements of surfacetopographyin laboratorymodelsof
mantle hotspots,Geophys. d., 96, 477-495, 1989.'
Gr6cke, D. R., A. Constantine, and M. I. Bird, Palaeo-environ-
mental change recorded in stable carbon-isotopesin Early
Cretaceous plant fragments,southeastern Australia,Aust. d.
Botany,in press,1997.
Hackman, B. D., The Solomon Islands fractured arc, in The
WesternPacific: Island Arcs, Marginal Seas, Geochemistry,
edited by P. J. Coleman, pp. 179-191, University of Western
Australia Press,Nedlands, 1973.
Hagen,R. A., L. A. Mayer, D.C. Mosher,L. W. Kroenke,T. H.
Shipley, and E. L. Winterer, Basementstructureof the
northernOntongJava Plateau,Proc. OceanDrill. Prog., Sci.
Results, 130, 23-31, 1993.
Hammond, S. R., L. W. Kroenke, and F. Theyet, Northward
motionof the OntongJavaPlateaubetween110 and 30 m.y.:
A paleomagnetic investigationof DSDP Site 289. Init. Repts.
DeepSeaDrill. Proj., 30, 415-418, 1975.
Hassler,D. R., andN. Shimizu,Old peridotitexenolithsfrom the
Kerguelen
Islands(abstract),
Eos Trans.AGU, 76, Fall
MeetingSuppl.,693-694, 1995.
Hawkins, M.P., and A. J. M Barron, The geologyand

resourcesof Santa Isabel. Solomon Islands Ministry of
Natural Resources,Geol. Div. Rept.J25, 114 pp., 1991.
Hawkins, J. W., P. F. Lonsdale, and R. Batiza, Petrologic
evolution of the Louisville seamount chain, in Seamounts,
Islands, and Atolls, Geophys.Monogr. Ser., vol. 43, editedby
P. Fryer, B. Keating,R. Batiza, and G. Boehlert,pp. 235-253,
AGU, Washington,D.C., 1987.
Hilde, T. W. C., S. Uyeda, and L. W. Kroenke, Evolutionof the
westernPacific and its margin, Tectonophysics, 38, 145-165,
1977.
Hill, R. I., Starting plumes and continentalbreak-up, Earth
Planet. Sci. Lett., 104, 398-416, 1991.
Hill, R. I., I. H. Campbell, G. F. Davies, and R. W. Griffiths,
Mantle plumes and continentaltectonics,Science,256, 186-
193, 1992.
Hirose, K., and I. Kushiro, Partial melting of dry peridotitesat
high pressures:Determination of compositionsof melts
segregatedfrom peridotiteusingaggregates of diamond,Earth
Planet. Sci. Lett., 114, 477-489, 1993.
Hopson, P.M., The geology of east central Santa Isabel.
Solomon Islands Ministry of Natural Resources,Geol. Div.
Rept. J23, 1988.
CA, 1997.
Horn, I., S. F. Foley, S. E. Jackson, and G. A. Jenner,
Experimentallydeterminedpartitioningof high field stren•&
Kroenke, L. W., R. Moberly Jr., and G. R. Heath, Lithologic
interpretationof continuousreflection profiling, Deep Sea
and selectedtransitionelementsbetween spinel and basaltic
melt, ChemoGeol., 117, 193-218, 1994.
Drilling Project, Leg 7, Init. Repts.Deep Sea Drill. Proj., 7,
Houtz, R. and J. Ewing, Upper crustalstructureas a functionof
plate age,J. Geophys.Res., 81, 2490-2498, 1976.
Hughes,G. W., and C. C. Turner, Geologyof SouthernMalaita,
Honiara Min. Nat. Res. Geol. SurveyDiv., Bull. No. 2, 80 pp.,
1976.
Hughes, G. W., and C. C. Turner, UpraisedPacific oceanfloor,
southern Malaita, Solomon Islands, Geol. Soc. Am. Bull., 88,
412-424, 1977.
Hussong, D. M., L. K. Wipperman, and L. W. Kroenke, The
crustal structureof the Ontong Java and Manihiki oceanic
plateaus,J. Geophys.Res., 84, 6003-6010, 1979.
Ito, G., and P. Clift, Evidence for multi-stagedaccretionof the
Manihiki and OntongJavaPlateausfrom their verticaltectonic
histories(abstract),Eos Trans.AGU, 77, Fall Meeting Suppl.,
p. 714, 1996.
Jain, J. C., J. A. O'Neill Jr., C. R. Neal, J. J. Mahoney,andM. G.
Petterson, Siderophile elements in large igneous provinces
(LIPs): Origin of the Ontong JavaPlateauat the core-mantle
boundary? (abstract), Eos Trans. AGU, 76, Fall Meeting
Suppl.,p. 700, 1995.
Jain,J. C., C. R. Neal, J. A. O'Neill Jr., and J. J. Mahoney,Origin
of the OntongJavaPlateau(OJP) at the core-mantleboundary:
Platinum group element (PGE) and gold (Au) evidence
(abstract),Eos Trans. AGU, 77, Fall Meeting Suppl., p. 714,
1996. Washington,D.C., 1987.
Jones, C. E., H. C. Jenkyns,A. L. Coe, and S. P Hesselbo, Mahoney, J. J., and K. J. Spencer,Isotopic evidence for the
Strontium isotopic variations in Jurassic and Cretaceous origin of the Manihiki and Ontong Java oceanic plateaus,
seawater, Geochim. Cosmochim.Acta, 58, 3061-3075, 1995.
Jones, J. H., and M. J. Drake, Geochemicalconstraintson core
formation in the Earth, Nature, 322, 221-228, 1986.
Kellogg, L. H., and S. D. King, Effect of mantle plumeson the
NEAL ET AL. 213
growth of D" by reaction between the core and mantle,
Geophys.Res. Lett., 20, 379-382, 1993.
Kent, R. W., M. Storey, and A.D. Saunders,Large igneous
provinces: Sites of plume impact or plume incubation?
Geology,20, 891-894, 1992.
Kincaid, C., D. W. Sparks, and R. Detrick, The relative
importanceof plate-drivenand buoyancy-drivenflow at mid-
oceanridges,d. Geophys.Res., 101, 16,177-16,193, 1996.
Kroenke, L. W., Geology of the Ontong Java Plateau, Hawaii
Inst. Geophys.Rept., HIG- 72-5, 119 pp., 1972.
Kroenke,L. W., and J. J. Mahoney,Rifting of the OntongJava
Plateau's easternsalient and seafloor spreadingin the Ellice
Basin: Relation to the 90 Myr eruptiveepisodeon the plateau
(abstract),Eos Trans. AGU, 77, Fall Meeting Suppl., p. 713,
1996
Kroenke,L. W., andW. Sager,The formationof oceanicplateaus
on the Pacific Plate (abstract),Eos Trans. AGU, 74, Fall
Meeting Suppl.,p. 555, 1993.
Kroenke,L. W., andP. Wessel,Pacificplatemotionbetween125
and 90 Ma and the formationof the Ontong Java Plateau
(abstract),ChapmanConf. on Global Plate Motions,Marshall,
1161-1227, 1971.
Kroenke, L. W., J. Resig, and P.A. Cooper, Tectonicsof the
southeastern Solomon Islands: Formation of the Malaita
Anticlinorium,in Geologyand OffshoreResourcesof Pacific
Island Arcs--Central and WesternSolomon Islands, Earth Sci.
Set., vol. 4, edited by J. G. Vedder, K. S. Pound, and S. Q.
Boundy,pp. 109-116, Circum-PacificCouncilfor Energyand
Mineral Resources,Houston, Tex., 1986.
Kroenke, L. W., W. Berger,, T. R. Janacek,et al., Proc. Ocean
Drill. Prog., Init. Repts.130, 1240 pp., 1991.
Kroenke, L. W., W. Berger, mid Leg 130 ShipboardScientific
Party, Proc. Ocean Drill. Prog., Sci. Results, 130, 867 pp.,
1993.
Larson, R. L., and C. Kincaid, Onset of mid-Cretaceous
volcanismby elevationof the 670 km thermalboundarylayer,
Geology,24, 551-554, 1996.
Lassiter,J. C., D. J. DePaolo, and J. J. Mahoney, Geochemistry
of the Wrangellia flood basaltprovince:Implicationsfor the
role of continental and oceanic lithospherein flood basalt
genesis,d. Petrol., 36, 983-997, 1995.
Mahoney, J. J., An isotopicsurveyof Pacific oceanicplateaus:
Implicationsfor their natureorigin, in Seamounts,Islands,and
Atolls, Geophys.Monogr. Ser., vol. 43, editedby B. Keating,
P. Fryer, R. Batiza, and G. Boehlert, pp. 207-220, AGU,
Earth Planet. Sci. Lett., 104, 196-210, 1991.
Mahoney, J. J., W. B. Jones,F. A. Frey, V. J. M. Salters,D. G.
Pyle, and H. L. Davies, Geochemicalcharacteristicsof lavas
from Broken Ridge, the NaturalistePlateau,and
214 THE ONTONG JAVA PLATEAU

Kerguelen Plateau: Cretaceousplateau volcanism in the Nakanishi, M., and E. L. Winterer, Tectonic eventsof the Pacific
southeastIndian Ocean, Chem. Geol., 120, 315-345, 1995. Plate related to the formation of the Ontong Java Plateau
Mahoney,J. J., J. M. Sinton,M.D. Kurz, J. D. Macdougall,K. J. (abstract),Eos Trans. AGU, 77, Fall Meeting Suppl., p. 713,
Spencer,and G. W. Lugmalt, Isotope and trace element 1996.

characteristics of a super-fastspreadingridge: East Pacific Nakanishi,M., K. Tamaki, and K. Kobayashi,Magneticanomaly

Rise, 13-23øS,Earth Planet. Sci. Lett., 121, 173-193, 1994. lineationsfrom late Jurassicto early Cretaceousin the west-
Mahoney, J. J., M. Storey,R. A. Duncan,K. J. Spencer,andM. centralPacificOcean,Geophys.J. Int., 109, 701-719, 1992.
Pringle, Geochemistryand age of the OntongJavaPlateau,in Neal, C. R., Mantle studiesin the westernPacific and kimberlite-
The MesozoicPacific: Geology, Tectonics,and Volcanism, type intrusives,Unpubl. Ph.D. Thesis, University of Leeds,
Geophys.Monogr.Ser., vol. 77, editedby M. S. Pringle,W. UK, 365 pp., 1985.
W. Sager, W. V. Sliter, and S. Stein, pp. 233-262, AGU, Neal, C. R., The origin and compositionof metasomatic fluids
Washington,D.C., 1993. and amphibolesbeneathMalaita, SolomonIslands,J. Petrol.,
Mammerickx, J., and S. M. Smith, Bathymetry of the north- 29, 149-179, 1988.
centralPacific,Map and Chart SeriesMC-52, The Geological Neal, C. R., The relationshipbetweenmegacrystsand their host
Societyof America,Boulder,Colo. 1985. magma and identification of the mantle source region
Mann, P., L. Gahagan,M. Coffin, T. Shipley,S. Cowley, andE. (abstract),Eos Trans. AGU, 76, Fall Meeting Suppl., p. 664,
Phinney, Regional tectonic effects resulting from the 1995.
progressive East-to-Westcollisionof the OntongJavaPlateau Neal, C. R., and J.P. Davidson, An unmetasomatizedsourcefor
with the MelanesianArc System(abstract),Eos Trans.AGU, the Malaitan aln6ite(SolomonIslands):Petrogenesis
involving
77, Fall Meeting Suppl.,p. 712, 1996. zone refining, megacrystfractionation,and assimilationof
Marks, K. M., and D. T. Sandwell, Analysis of geoid height oceanic lithosphere,Geochim.Cosmochim.Acta, 53, 1975-
versus topographyfor oceanic plateausand swells using 1990, 1989.
nonbiasedlinear regression,d. Geophys.Res., 96, 8045-8055, Neal, C. R., and L. A. Taylor, Negative Ce anomaly in a
1991. peridotitexenolith from Malaita, SolomonIslands:Evidence
Mayer, L. A., andJ. A. Tarduno,Paleomagnetic
investigation
of for crustalrecyclinginto the mantleor mantlemetasomatism?,
the igneoussequence,Site 807, OntongJava Plateau,and a Geochim. Cosmochim.Acta, 53, 1035-1040, 1989.
discussionof Pacific true polar wander, Proc. Ocean Drill. Neal, C. R., J. J. Mahoney, A.D. Saunders,and T. L. Babbs,
Prog., Sci. Results,130, 51-59, 1993. Trace element characteristics of a 3.5-4 km thick section of
Mayer, L. A., T. H. Shipley,E. L. Winterer,D. Mosher,andR. OJP basalts,Malaita, Solomon Islands (abstract),Eos Trans.
A. Hagen, Seabeamand seismicreflectionsurveyson the AGU, 75, Fall Meeting Suppl.,p. 727, 1994.
OntongJavaPlateau,Proc. OceanDrill. Prog., Sci. Results, Neal, C. R., S. Schaefer, J. J. Mahoney, and M. G. Petterson,
130, 45-75, 1991. Uplift associated with large igneous province (LIP)
McGinnis, C. E., C. R. Neal, and J. C. Jain, Developmentof an petrogenesis: A case study of the Ontong Java Plateau,SW
analyticaltechniquefor the accurate& precisedetermination Pacific (abstract),Abstractswith Programs,27 No. 6, Geol.
of the high field strengthelements(HFSEs), Cs, & Mo by ICP- Soc.Am., p. A48, 1995.
MS with geologicalapplications(abstract),Eos Trans.AGU, Neavel, K. E., and A.M. Johnson,Entrainmentin composition-
77, Fall Meeting Suppl.,p. 772, 1996. ally buoyantplumes,Tectonophysics, 200, 1-15, 1991.
McGinnis, C. E., C. R. Neal, and J. C. Jain, Analyticaltechnique Newsom, H. E., and M. J. Palme,The depletionof siderophile
for the accurateand precisedeterminationof the high field elements in the Earth's mantle: New evidence from
strength elements (HFSEs) by ICP-MS with geological molybdenumand tungsten,Earth Planet. Sci. Lett., 69, 354-
applications,Geostand.Newsl.,in revision,1997. 364, 1984.
Nixon, P. H., Kimberlites in the south-westPacific, Nature 287,
McKenzie, D., and R. K. O'Nions, Partial melt distributionsfrom
inversion of rare earth element concentrations,d. Petrol., 32,718-720, 1980.
1021-1091, 1991. Nixon, P. H., and F. R. Boyd, Garnet-bearinglherzolites and
Michael, P. J., and W. C. Cornell,H20, CO2, C1,and S contents discrete nodule suites from the Malaita aln6ite, Solomon
in 122 Ma glassesfrom Ontong Java Plateau,ODP 807C: Islands, SW Pacific, and their bearing on oceanic mantle
Implicationsfor mantle and crustalprocesses(abstract),Eos compositionand geotherm,in The Mantle Sample:Inclusions
Trans.AGU, 77, Fall Meeting Suppl.,p. 714, 1996. in Kimberlites and Other Volcanics,edited by F. R. Boyd and
Miura, S., M. Shinohara,N. Takahashi, E. Araki, A. Taira, K. H. O. A. Meyer, pp. 400-423, AGU, Washington,D.C., 1979.
Suyehiro,M. Coffin, T. Shipley, and P. Mann, OBS crustal Nixon, P. H., and P. J. Coleman,Garnet-bearinglherzolitesand
structureof Ontong ,lava Plateau converginginto Solomon discrete nodule suites from the Malaita aln6ite, Solomon
Islandarc(abstract),EosTrans.AGU, 77, Fall MeetingSuppl., Islands, and their bearing on the nature and origin of the
p• 713, 1996. Ontong Java Plateau, Bull. Aust. Soc. Explor. Geophys.,9,
Murauchi, S., W. J. Ludwig, N. Den, H. Hotta, T. Asanuma,T. 103-106, 1978.
Yoshi, A. Kubotera, and K. Hagiwara, Seismic refraction Nixon, P. H., R. H. Mitchell, andNoW. Rogers,Petrogenesis
of
measurementson the Ontong Java Plateaunortheastof New aln6itic rocks from Malaita, Solomon Islands, Melanesia,
Ireland, d. Geophys.Res., 78, 8653-8663, 1973. Mineral. Mag., 43, 587-596,

O'Hara, M. J., and H. S. Yoder Jr., Formation and fractionation
of basicmagmasat high pressures,Scott.d. Geol., 3, 67-117,
1967
Olson, P., and H. A. Singer, Creepingplumes,d. Fluid Mech.,
158, 511-531, 1985.
Olson,P., and I. S. Nam, Formationof seafloorswellsby mantle
plumes,d. Geophys.Res., 91, 7181-7191, 1986.
Operto, S., and P. Charvis, Deep structure of the southern
KerguelenPlateau(southernIndianOcean)from oceanbottom
seismometerwide-angleseismicdata, d. Geophys.Res., 101,
25,077-25,103, 1996.
Parkinson,I. J., R. J. Arculus, E. McPherson,and R. A. Duncan,
Geochemistry,tectonicsand the peridotitesof the northeastern
SolomonIslands(abstract),Eos Trans.AGU, 76, Fall Meeting
Suppl.,p. 642, 1996.
Petterson,M. G., The geology of north and central Malaita,
SolomonIslands(includingimplicationsof geologicalresearch
on Makira, Savo Island, Guadalcanal, and Choiseul between
1992 & 1995), Geological Mere., 1/95, Water and Mineral
ResourcesDivision, Ministry of Energy, Water, and Mineral
Resources,Honiara, Solomon Islands, 1995.
Petterson,M. G., C. R. Neal, A.D. Saunders,T. L. Babbs, J. J.
Mahoney, and R. A. Duncan, Speculationsregardingthe
evolutionof the Ontong Java Plateau(abstract),Eos Trans.
AGU, 76, Fall Meeting Suppl.,p. 693, 1995.
Petterson,M. G., C. R. Neal, J. J. Mahoney, L. W. Kroenke, A.
D. Saunders,T. L. Babbs,R. A. Duncan, D. Tolia, B. McGrail,
and M. Barron, Structure and deformation of north and central
Malaita, Solomon Islands: Tectonic implications for the
Ontong JavaPlateau - Solomonarc collision and for the fate of
oceanicplateaus,Tectonophysics, in revision,1997.
Phinney,E., P. Mann, M. Coffin, and T. Shipley,Along-strike
variationsin the style of oceanicplateauaccretionwithin the
Malaita accretionaryprism, SolomonIslands(abstract),Eos
Trans.AGU, 77, Fall Meeting Suppl.,p. 712, 1996.
Pringle, M. S., J. D. Obradovich, and R. A. Duncan, Estimated
ages for magnetic anomaly M0 and interval ISEA, and a
minimum estimatefor the durationof the Aptian (abstract),
Eos Trans.AGU, 73, Fall Meeting Suppl.,p. 633, 1992.
Pudsey-Dawson,P.A., SouthMalaita geologicalreconnaisances, Schaefer,S., and C. R. Neal, An estimateof uplift associatedwith
1957-1958, British Solomon Islands Geol. Rec., 1957-1958,
27-31, 1960.
Ramsay, W. R. H., Crustal strain phenomenain the Solomon
Islands:Constraintsfrom field evidenceandthe relationshipto
the India-Pacificplatesboundary,Tectonophyics,87, 109-126,
1982.
Ribe, N.M., The dynamicsof plume-ridge interaction2. Off-
ridge plumes,d. Geophys.Res.,101, 16,195-16,204, 1996.
Ribe,N.M., U. R. Christensen, andJ. TheiBing,The dynamicsof
plume-ridge interaction, 1: Ridge-centeredplumes, Earth
Planet. Sci. Lett., 134, 155-168, 1995.
Richards, M. A., R. A. Duncan, and V. E. Courtillot, Flood
basaltsand hot-spot tracks: Plume heads and tails, Science,
246, 103-107, 1989.
Richards,M. A., D. L. Jones,R. A. Duncan, and D. J. DePaolo,
A mantle plume initiation model for the Wrangellia flood
basaltand otheroceanicplateaus,Science,254, 263-267, 1991.
NEAL ET AL. 215
Richardson, W. P., and E. Okal, Crustal and upper mantle
structure of the Ontong Java Plateau from surface waves:
Results from the Micronesian PASSCAL experiment
(abstract),Eos Trans. AGU, 77, Fall Meeting Suppl., p. 713,
1996.
Rickwood,F. K., Geologyof the islandof Malaita, in Geological
Reconnaissance of Parts of the Central Islands of the B.S.I.P..'
Colonial Geol. and Min. Res., 6, 300-306, 1957.
Rudnick, R. L., and I. Jackson,Measured and calculatedelastic
wave speedsin partially equilibratedmafic granulitexenoliths:
Implications for the properties of an underplatedlower
continentalcrust,J. Geophys.Res. 100, 10,211- 10,218, 1995.
Sager,W. W., and H.-C. Han, Rapid formationof the Shatsky
Rise oceanic plateau inferred from its magnetic anomaly,
Nature, 364, 610-613, 1993.
Sandwell, D. T., and K. R. MacKenzie, Geoid height versus
topographyfor oceanicplateausand swells,J. Geophys.Res.,
84, 7403-7418, 1989.
Sandwell,D. T., and M. L. Renkin, Compensationof swells and
plateausin the North Pacific: No direct evidencefor mantle
convection,J. Geophys.Res.,93, 2775-2783, 1988.
Saunders,A.D., Geochemistryof basaltsfrom the Nauru Basin,
Deep Sea Drilling Project Legs 61 and 89, Init. Repts. Deep
Sea Drill. Proj., 89, 499-518, 1986.
Saunders•A.D., M. Storey, R. W• Kent, and M. J. Norry,
Consequences of plume-lithosphere interactions, in
Magmatism and the Causes of Continental Break-Up, Spec.
Publo 68, edited by B.C. Storey, T. Alabaster, and R. J.
Pankhurst,pp. 41-59, The GeologicalSociety,London, 1992.
Saunders,A.D., T. L. Babbs,M. J. Norry, M. G. Petterson,B. A.
McGrail, J. J. Mahoney, and C. R. Neal, Depth of
emplacementof oceanicplateau basalticlavas, Ontong Java
Plateau and Malaita, Solomon Islands: Implications for the
formation of oceanicLIPs? (abstract),Eos Trans. AGU, 74,
Fall Meeting Suppl., p. 552, 1993.
Saunders,A.D., J. Tarney, A. C. Kerr, and R. W. Kent, The
formationand fate of large oceanicigneousprovinces,Lithos,
3 7, 81-89, 1996.
the Ontong Java Plateau and the importanceof cumulates
(abstract),Eos Trans. AGU, 75, Fall Meeting Suppl., p. 711,
1994.
Sliter, W. V., Aptian anoxia in the Pacific Basin, Geology,17,
909-912, 1989.
Sliter, W. V., and R. M. Leckie, Cretaceous planktonic
foraminifers and depositionalenvironmentsfrom the Ontong
JavaPlateauwith emphasison Sites803 and 807, Proc. Ocean
Drill. Prog., Sci. Results,130, 63-84, 1993.
Smith, W. H. F., and D. T Sandwell, Oceanographic
"pseudogravity"in marine gravity fields derived from
declassified
GeoõatandERS-1 altimetry(abstract),Eos Trans.
AGU, 76, Fall Meeting Suppl.,p. 151, 1995a.
Smith, W. H. F., and D. T. Sandwell,Marine gravity field from
declassifiedGeosatand ERS-1 altimetry (abstract),Eos Trans.
AGU, 76, Fall Meeting Suppl.,p. 152, 1995b.
Stein, C. A., and S. Stein, A model for the global variation
216 THE ONTONG JAVA PLATEAU

oceanic depth and heat flow with lithosphericage, Nature, Basin, SouthwestPacific, Init. Repts.Deep Sea Drill. Proj.,
359, 123-129, 1992. 61,673-687, 1986.
Steiner, M. B., and B. P. Wallick, Jurassic to Paleocene van Keken, P. E., D. A. Yuen, and A. P. van den Berg, The
paleolatitudes of the Pacific Plate derived from the effectsof shallowrheologicalboundariesin the uppermantle
paleomagnetismof the sedimentarysequencesat Sites 800, on inducingshortertime scalesof diapiric flows, Geophys.
801, and 802, Proc. OceanDrill. Prog., Sci. Results,129, 431- Res. Lett., 20, 1927-1930, 1993.
446, 1992. Walker, D., L. Norby, and J. H. Jones,Superheatingeffectson
Stoeser,D. B., Igneous rocks from Leg 30 of the Deep Sea metal-silicate partitioning of siderophileelements,Science,
Drilling Project,Init. Repts.Deep Sea Drill. Proj., 30, 410- 262, 1858-1861, 1993.
444, 1975. Walker,R. J., J. W. Morgan,andM. F. Horan,187Osenrichment
Storey,M., J. J. Mahoney, L. W. Kroenke, and A.D. Saunders, in some plumes: Evidence for core-mantle interaction?
Are oceanicplateaussites of komatiite formation?Geology, Science,269, 819-822, 1995.
19, 376-379, 1991. Wessel,P., and L. W. Kroenke, RelocatingPacific hotspotsand
Storey, M., R. Kent, A.D. Saunders,V. J. Salters,J. Hergt, H. refining absolute plate motions using a new geometric
Whitechurch,J. H. Sevigny, M. F. Thirlwall, P. Leat, N C. technique,Nature, in press,1997.
Ghose, and M. Gifford, Lower Cretaceous volcanic rocks White,R. S., J. H. McBride,P. K. H. Maguire,B. Brandsd6ttir,
along continental margins and their relationship to the W. Menke, T. A. Minshull, K. R. Richardson, J. R.
KerguelenPlateau,Proc. OceanDrill. Prog., Sci. Results,120, Smallwood, R. K. Staples,and the FIRE Working Group,
33-54, 1992. Seismic images of crust beneathIceland contributeto long-
Sun, S.-s., and W. F. McDonough,Chemicaland isotopic standingdebate,Eos Trans.AGU, 77, 197-201, 1996.
systematicsof oceanic basalts: Implications for mantle White, R. S., D. McKenzie, and R. K. O'Nions, Oceanic crustal
compositionand processes,in Magmatism in the Ocean thickness from seismic measurements and rare earth element
Basins, Spec. Publ. 42, edited by A.D. Saundersand M. J. inversions,d. Geophys.Res., 97, 19,683-19,715, 1992.
Norry, pp. 313-345, GeologicalSocietyof London,1989. Whitehead, J. A., and D. S. Luther, Dynamics of laboratory
Tarduno, J. A., Vertical and horizontal tectonics of the diapirandplumemodels,d. Geophys.Res.,80, 705-717, 1975.
CretaceousOntong Java Plateau(abstract),Eos Trans.AGU, Winterer, E. L., Bathymetry and regionaltectonicsettingof the
73, Fall Meeting Suppl.,p. 532, 1992. Line Islandschain,Init. Repts.Deep Sea Drill. Proj., 33, 731-
Tarduno,J. A., and W. W. Sager,Polar standstillof the mid- 748, 1976.
CretaceousPacific plate and its geodynamicImplications, Winterer, E. L., and M. Nakanishi, Evidence for a plume-
Science,269, 5226-5228, 1995. augmented, abandoned spreading center on Ontong Java
Tarduno,J. A., W. V. Sliter, L. W. Kroenke,M. Leckie, J. J. Plateau(abstract),Eos Trans.AGU, 76, Fall MeetingSuppl.,p.
Mahoney, R. J. Musgrave, M. Storey, and E. L. Winterer, 617, 1995.
Rapidformationof the OntongJavaPlateauby Aptianmantle Yan, C. Y., and L. W. Kroenke,A plate-tectonicreconstruction
plume volcanism,Science,254, 399-403, 1991. of the southwestPacific, 100-0 Ma, Proc. OceanDrill. Prog.,
Taylor, B., Mesozoic magneticanomaliesin the Lyra Basin Sci. Results, 130, 697-710, 1993.
(abstract),Eos Trans.AGU, 59, Fall MeetingSuppl.,p. 320, Yoder, H. S., Jr., and C. E. Tilley, Origin of basaltmagmas:An
1978. experimentalstudy of natural and syntheticrock systems,d.
Taylor, S. R., PlanetaryScience:A Lunar Perspective,481 pp., Petrol., 3, 342-532, 1962.
Lunar and PlanetaryInstitute,Houston,Tex., 1982.
Tejada, M. L. G., J. J. Mahoney, R. A. Duncan, and M.P. R. A. Duncan, College of Oceanicand AtmosphericSciences,
Hawkins,Age andgeochemistry of basementandalkalicrocks Ocean AdministrationBuilding 104, Oregon State University,
of MalaitaandSantaIsabel,SolomonIslands,southern
margin Corvallis, OR 97331.
of the OntongJavaPlateau,d. Petrol., 37, 361-394, 1996a. L. W. Kroenke and J. J. Mahoney,Schoolof Oceanand Earth
Tejada,M. L. G., J. J. Mahoney,R. A. Duncan,and C. R. Neal, Science and Technology, University of Hawaii, Honolulu, HI
Geochemistry andageof the OntongJavaPlateau(OJP)crust 96822.
in central Malaita, SolomonIslands(abstract),Eos Trans. C. R. Neal, Departmentof Civil Engineeringand Geological
AGU, 77, Fall MeetingSuppl.,p. 715, 1996b. Sciences,University of Notre Dame,Notre Dame, IN 46556.
Tokuyama,H., andR. Batiza,Chemicalcomposition of igneous M. G. Petterson,British GeologicalSurvey,MurchisonHouse,
rocksandoriginof the sill andpillow basaltcomplexof Nauru WestMains Road, Edinburgh,EH9 3LA, United
 The Paranfi-Etendeka Province

David W. Peate

Departmentof Earth Sciences,The Open University,Milton Keynes,UnitedKingdom

Stratigraphicdata and 40Ar-39Ar ages for the Early CretaceousParanti-

Etendeka flood basalts indicate that the main magmatic episode lasted for
severalm.y. (129-134 Ma) and was linked to the northwardopening of the
SouthAtlantic Ocean, but with some earlier magmatism(135-138 Ma) found
inland far from the eventualoceanicrift. The regional distributionof distinct
high-Ti/Y (Urubici, Pitanga,Paranapanema, Ribeira) and 1ow-Ti/Y (Gramado,
Esmeralda)magmatypes in the lavasand associateddyke swarmsimplies that
magma generationoccurredover a wide area and involved different mantle
sources. Low MgO contents (3-7 wt%) indicate extensive fractional
crystallisation,and uppercrustalassimilationwas importantin the evolutionof
the Gramado magmas. However, Paranti basalts that are consideredto be
uncontaminatedby crust have trace element and isotope characteristics(e.g.
Nb/La < 0.8; gNdi< 0) and major elementfeaturesthat appearto requiremantle
sourcesdistinctfrom thoseof oceanicbasalts.The minor, late-stage,Esmeralda
magmatype is an exception,requiringa componentfrom incompatible-element-
depletedasthenosphere. The role of the Tristan mantle plume appearsto have
been largely passive,with conductiveheating facilitating mobilisationof old
lithospheric
material.Significantrhyoliticeruptions(>1000 km3) that can be
correlatedacrossthe Atlantic Oceanaccompaniedthe final magmaticphasein
the southeastParanti and the Etendeka. The flood basalts post-date most
estimatesfor the Jurassic-Cretaceous
boundary,ruling out any link to a faunal
extinction.

INTRODUCTION Vandoros,1967]. The Rio GrandeRise and Walvis Ridge

in the SouthAtlantic Oceanare interpretedas representing
The extensive Paranti lava field in central South the fossil trace of the Tristanmantle plume and they con-
America and the minor Etendeka remnant in Namibia once nect the flood basalt exposuresof the Paranti and the
formed a single magmaticprovince [Erlank et al., 1984; Etendeka,respectively,to the presentmagmaticactivity of
Bellieni et al., 1984a]that was associated
with the opening the plume found on the islandsof the Tristan da Cunha
of the SouthAtlantic Ocean duringthe Early Cretaceous. group and Gough (Figure 1) [O'Connor and Duncan,
This Paranfi-Etendeka provinceranksas one of the largest 1990; Gallagher and Hawkesworth, 1994]. This associa-
extant continentallarge igneousprovinces(LIPs), with a tion has bolsteredmodelsthat stressan importantrole for
preserved
volumein excessof 1 x 106km3 [Cordaniand the Tristan plume in generatingthe flood basalts [e.g.,
Morgan, 1981; White and McKenzie, 1989; Peate et al.,
LargeIgneousProvinces:Continental,Oceanic,andPlanetary 1990].
Flood Volcanism
Beforethe 1960s,numerouspaperson the petrographx,
GeophysicalMonograph100
and general geologicalsetting of the Parantilavas were
Copyright1997 by the AmericanGeophysicalUnion
218 PARANA-ETENDEKA PROVINCE
Figure 1. Map of the South Atlantic region [after White and
McKenzie,1989]. Shallowfeatures,between4 km water depth
andsealevel, are highlightedin white.The magmaticeffectsof
the Tristanmantleplume,whosepresentlocationis markedby
the Tristanda Cunhaislandgroup,canbe tracedbackalongthe
Rio GrandeRiseandthe WalvisRidgeto the floodbasaltexpo-
suresof the Parantiand the Etendeka,respectively.
SB -- Salado
basin, CB - Colorado basin.
published,primarily by Brazilianscientists,and thesewere
reviewed by Cordani and Vandoros [1967]. Even the
earlieststudies[e.g., Baker, 1923] realisedthe true extent
of the Paranti lava field. The continental drift debate of the
1960s saw a renewed interest in the Paranti and Etendeka
lavas in the contextof the breakuphistoryof Gondwana.
K-Ar dating [reviewedby Rocha-Camposet al., 1988] and
palaeomagneticstudies[reviewedby Ernestoand Pacca,
1988] confirmed the close link between the Paranti and
Etendekalavas.Thesestudiesdemonstrated that they were
both Early Cretaceous in age and thus not
contemporaneous with the Early JurassicKaroo and Ferrar
flood basalt provinces elsewherein Gondwana. Several
geochemicalstudiesinto the petrogenesisof the Paranti
magmatism began during the 1980s, involving
collaborations between Brazilian, Italian, British, and
American universities [e.g., Bellieni et al., 1984a;
Mantovani et al., 1985a; Fodor et al., 1985a; Hawkesworth
et al., 1986]. The principal results of the joint
Italian/Braziliangroup are summarisedby Piccirillo et al.
[1988], and all the data are collated in a book [Piccirillo
and Melfi, 1988] which contains extensive reviews on
many aspects of Paranti magmatism, including pet-
rography,mineral chemistry,major and trace elementand
isotopegeochemistry,K-Ar dating,and palaeomagnetism.
Erlank et al. [ 1984] gavethe first comprehensive descrip-
tion of the Etendeka magmatismand summarisedthe
chronological,compositional, andpalaeomagnetic datathat
showedthat the Etendekasequences onceformedthe east-
ern edge of the Parantibasin.Thesesequences representa
small fragment of the Paranfi-Etendeka province,but the
excellent exposure has led to significant insights,
particularlyinto the physical volcanologyof the rhyolite
eruptions[e.g., Milner et al., 1992].
Almost a decadehaspassedsincethe lastmajor reviews
of the Paranfi-Etendekaprovince were published.In this
time, therehasbeena surgeof interestgloballyin largeig-
neousprovinces,driven in part by the developmentof sev-
eral new geodynamicalmodels for their generation[e.g.,
White and McKenzie, 1989; Richards et al., 1989]. A
wealth of new information has been obtained on the
Paranfi-Etendekaprovince on a diverse range of topics
suchas detailedlava stratigraphy[e.g., Peate et al., 1990,
1992; Milner et al., 1995b], geochemicaland petrogenetic
modelling [e.g., Turner and Hawkesworth, 1995; Peate
and Hawkesworth,1996], precise40Ar-39Ardating[e.g.,
Renne et al., 1992; Turner et al., 1994], fission-track
chronologyof continentalmargin development[Gallagher
et al., 1994], and the nature of intrusive magmatism
[Piccirillo et al., 1990; Regelous, 1993]. Thus, it is
appropriateto review what hasbeen learnedrecentlyabout
the Paranfi-Etendekaprovinceand to see how suitablethe
various current plume-relatedgeodynamicmodels are in
explainingthe principalfeaturesof the province.
GEOLOGICAL BACKGROUND
SpatialExtentof Parand-EtendekaMagmatism
The Paranti lava field covers an area of at least 1.2 x 10 6
km2 over southernBrazil, Uruguay,easternParaguayand
northernArgentina [Cordani and Vandoros,1967], and its
presentextent is boundedby the margin of the underlying
Paranti sedimentary basin (Figure 2). The three-
dimensionalstructureis well known by surfacemapping
and from numerousoil explorationboreholes(Figure 3)
[Zalan et al., 1987; Peate et al., 1992]. The thickest
preservedaccumulationof lavas (1.5-1.7 km) is in the
north, coincidentwith the deepestpart of the Parantibasin.
The extent of the lavas in northernArgentinais uncertain
becauseof overlying Quaternarysediments,but borehole
data indicatesignificantlava thicknesses(>1 km [Leinz et
al., 1968]). The averagelava thicknessoverthe provinceis
estimatedat about 0.7 km [Leinz et al., 1968]. The Paranti
lavas are spectacularlyexposed along the coastal Serra
Geral escarpmentin southeastBrazil: the Serra Geral
Formationis the formal stratigraphicname for the Paranti
lavas in Brazil. As the lavas have different stratigraphic
names in the neighbouringcountries(Argentina: Posadas
Member, Curuz6 CuatifiFormation:Paraguay:Alto Paranti
Formation:Uruguay: Arapey Formation),we will use the
general term Paranti province to encompassall
 PEATE 219

edge of Paran• I Ioverlying

youngersedimentary
lavas rocks
sedimentary
basin
!:• rhyolites

-F
AN
r•'%
•1 dykes

ES•

SOUTH AMERICA

Janeiro

Paulo

SA

NA
A,

/ \

AFRICA

aran, Etendeka

Walvis Bay

0 500km
I I
Buenos
•
Aires

Figure 2. Pre-driftGondwanareconstruction showingthe extentof the Paran/t-Etendeka magmatismin relationto the

Paranftsedimentary basinandproto-Atlanticrift [de Wit et al., 1988;Peateet al., 1992]. Dykes are concentratedin
four areas:PG = PontaGrossa,SRJ= SgoPaulo- Rio de Janeirocoast,PA = easternParaguay,andin the Etendeka
(not shown).SA = Santosbasin,CA = Camposbasin,ES = EspritoSantobasin,NA = Namibebasin,CU = Cuanza
basin,MV = Morro Vermelho.Recentpalaeomagnetic dataand40Ar-39Arages[Montes-Laueret al., 1995] indicate
thatthe isolatedbasaltexposures
closeto theBrazil-Boliviaborder(AN = Anari,TA = Tapirapug)areEarly Jurassic in
ageandnot relatedto the Paran/t-Etendekamagmatism aspreviouslysuggested. The + showspresent-day latitudeand
longitudeat 5ø
220 PARANA-ETENDEKA PROVINCE
I i
52 50 48 øW
BRAZIL
-
22
Ordowcian
basement
Atlantic
Ocean
0 250 500 km
(a)
I I I I
Figure 3. (a) Depthto basementbeneaththe Parantisedimentary
of Parantilavas(contoursat 200-m intervals);from Peateet al. [ 1992].
formationsand the associatedintrusivemagmatism.
The majority of analysedsamplesin the literatureare
from the Serra Geral region, where sectionsup to 0.8 km
thick are found. Inland from the escarpment,significantto-
pographicrelief is rarely developedon the plateau,excep-
tions being the deeply incisedvalleys of a few westward-
flowing rivers. Most of the Parantilava field is thusdevoid
of suitablesectionsfor detailedflow-by-flow stratigraphic
studies.Although the lava pile is greaterthan 1 km thick
over much of the centralParanti,surfacesamplescan only
provide informationaboutthe uppermostfew hundredme-
tres. Access to borehole samples provided Peate et al.
[1992] with a unique opportunityto look at the otherwise
inaccessibledeeperlevelsof the lava pile.
On the African Plate, the Etendeka lavas are scattered
over an area of 0.8 x 105 km2 in northwestern Namibia
[Erlank et al., 1984]. The coastalregionsof Angola have
been less well studied, but Early Cretaceoustholeiitic
basalts and rhyolites are found, at least as far north as
Luanda, in the onshoreNamibe (NA on Figure 2) and
Cuanza (CU on Figure 2) basins [Piccirillo et al., 1990;
Alberti et al., 1992]. TheseAngolanlavasprobablycovera
similar area to the Etendeka lavas.
I I I
52 50 48 øW
_
- _ ,
; ,,?
PARAGUAY
ARGENTINA
A tlan tic
Ocean
250 500 km
(b)
I
basin(contoursat 600-m intervals),(b) Isopachmap
Significantdyke swarmsare found along the Brazilian
coast between Silo Paulo and Rio de Janeiro, in eastern
Paraguay,and southof the Etendekalava field. This distri-
bution suggeststhat the lavas originally covereda much
greaterareaof at least2.0 x 106km2,nowreducedby ero-
sion. Modelling of fissiontrack data by Gallagher et al.
[1994] indicated that as much as 3 km of material could
have been eroded from the Brazilian coastalplain since
continentalbreakup, althoughnot all of this would have
been lavas.
ExtensiveEarly Cretaceouslavas are found in offshore
basinsalongthe Braziliancoast[Chang et al., 1992]. Lava
thicknessesof 600 m have been drilled in the Campos
basin (200 km east of Rio de Janeiro:CA on Figure 2)
which coversan area of 105 km2. Mizusakiet al. [1992]
showedthat these mainly subaeriallyeruptedlavas were
more similar in compositionto the Parantibasaltsthan to
mid-oceanridge basalts(MORB), and they consideredthe
Camposbasinlavasto be an easternextensionof the north-
ern Parantilava field. Fodor and Vetter [1984] found simi-
lar 'Paranti-like'basaltlavas up to 150 km offshorein the
Espirito Santobasin (500 km northeastof Rio de Janeiro:
ES on Figure2) and in the Santosbasin(200 km
 PEATE 221

in the Luis Alves craton beneath the southem Paranti basin

'basalts' 'rhyolites' and in the San Francisco craton to the northeast of the
150
basin. Early Proterozoicrocks are found in the Curitiba
massif on the northern margin of the Luis Alves craton,
and in the Transamazonian massif around the west and
,_ lOO
northwest margins of the Paranti basin. The Mid-
Proterozoic Ribeira belt underlies the central and northeast-
ern parts of the basin. The last major pre-Paranfievent
c.-. 50 across the region was the Brasiliano or Pan-African
orogeny at 750-500 Ma. The Brasiliano mobile belts are
found beneath the eastern and northwestern Paranti basin.

o In Namibia, the basementgeology beneaththe Etendeka

45 50 55 60 65 70 75

SiO2 (wt0/o)
Figure 4. SiO2 histogramillustratingthe bimodal compositionof
the Paranfi-Etendekalavas.The silica gap between61 and 63 wt%
forms a natural division into 'basalts' (black shading) and
'rhyolites' (white shading).The histogramdistributiondoes not
reflect the relative eruptedvolumesof basaltand rhyolite magma WG
due to the bias of samplingtowardsthe coastalmarginswhererhy-
olitesare more common.Terminologyfor the silicic rocksis com-
plicated because their compositionsstraddle the boundariesof
severalfields on many classificationdiagrams.As a simplification,
Erlank et al. [1984] used the term 'quartz latites' for all the
Etendekasilicic rocks,whereasBellieni et al. [1986] preferredthe
term 'rhyolite' to 'encompassall the compositionsof the Paranti AM
silicic rocks. Paran
basin
of Rio de Janeiro:SA on Figure 2). The S•,oPauloplateau,
which extends for about 400 km southeast from the Santos
basin, appearsto be underlain by attentuatedcontinental
Basement ages CM,
crust, and dredgedsamplesalso show affinities with the
Paranti lavas [Fodor and Vetter, 1984]. Significant ................
0.5-1.0Ga LA
quantitiesof Early Cretaceousvolcanic material have also
been found by seismic studiesthroughoutthe Namibian 1.8-2.7 Ga
continentalshelf but no sampleshave yet been analysed South
[Light et al., 1992]. Atlantic
Paranfi-Etendeka magmatism is strongly bimodal LA
(Figure 4), and the virtual absenceof sampleswith 60-64 DF
wt% SiO2, except locally in the Etendeka, produces a
natural division of the lavas into what are loosely termed 35øS 56"W
'basalts'and 'rhyolites'. The lava pile is dominated by
tholeiitic basalts (>90%), but significant quantities of
Figure 5. Simplified map of the basementgeologyof the Paranti
rhyolitesare found along the Brazilian continentalmargin
and in the Etendeka.
region of South America, showing the Archaean cratons and
Early Proterozoicmassifs(speckledshading:LA = Luis Alves
craton, SF = Silo Francisco craton, CM = Curitiba massif, AM =
RegionalGeology Transamazonianmassif, CG = Central Goias massif) and the
Mid-Late Proterozoicmobile belts (grey shading:RI = Ribeira
The basement beneath the Paranti basin consists of sev- belt, WG - WesternGoias belt, PG -- Paraguaibelt, DF = Dom
Feliciano belt) [adapted from Mantovani et al., 1991]. The
eral Archaean to Early Proterozoic cratonic blocks sur- PalaeozoicParantisedimentarybasin coincidesroughly with the
rounded by Mid-Late Proterozoicmobile belts (Figure 5 presentoutline of the Parantiflood basalts,and obscuresthe un-
[e.g., Mantovani et al., 1991]). Archaeanrocks are found derlyingbasement
222 PARANA-ETENDEKA PROVINCE
lavas is formed mainly by rocks of the Pan-A•can
Damara Sequence, which partially overlie Proterozoic
rocks (2.1-1.7 Ga) on the southwestcomer of the Congo
craton(Figure6b [Milner et al., 1995a]).
The Paranti basin was established in the Late Ordovician
as an intracratonicsedimentarybasin.It is alignedroughly
northeast-southwest, paralleling the Brasilianostructural
trendsin the underlyingbasement,and at its deepestpoint
reachesover 5 km below sealevel (Figure 3) [Zalan et al.,
1987; Peate et al., 1992]. The earliest, Palaeozoic, sedi-
mentsare largelymarinesiliciclastics,but the Mesozoicse-
quencesare exclusivelycontinental(lacustrineand fluvial)
sediments that culminate in the aeolian sandstones of the
JurassicBotucatuFormation. The Early Cretaceouslavas
were eruptedsubaeriallydirectly onto the Botucatusand-
stonesover virtually all of the Paranfi-Etendekaprovince,
and they only oversteponto older strata(Parantibasinand
pre-Ordovicianbasement)along the northeastmargin of
the Parantilava field and in parts of the Etendekaregion.
The aeoliansandstones persistas intercalations,
up to 160
m thick, within the Paranti and the Etendeka lava
sequences,and the more northerlyParantilavasare capped
by similar aeolian sandstonesof the Caiurfi Formation
[Rocha-Camposet al., 1988]. Thus, climatic conditions
were arid and desert-like throughoutthe period of lava
eruption. This is consistentwith the lack of palaeosol
developmentwithin the Parantilavas.It might alsoexplain
why developmentof entablature-stylejoint patternsare
absent,becausesuch features in the Columbia River basalts
of the USA have been attributed to the effects
percolatingwater during cooling of the flows [Long and
Wood, 1986].
Age of Parand-EtendekaMagmatism
Stratigraphicand sparsefossil evidencecan only con-
strain the age of the Paranfi-Etendekalavas to between
Upper Triassic and Upper Cretaceous[Rocha-Camposet
al., 1988]. Over 200 K-Ar ageshave been determinedon
Paranfi-Etendekasamples (reviewed by Erlank et al.
[1984] and Rocha-Camposet al. [1988]). K-At agesrange
from nearly400 Ma to lessthan 100 Ma, clearlyindicating
problems with both excess radiogenic argon and post-
crystallisationargonloss.Most samples(> 70%) yield ages
in the range 115-135 Ma, with a stronglydefinedmodeat
127 Ma. There is no significantage differencebetweenthe
extrusiveand intrusivephasesof magmatismbut the tech-
nique is not sufficientlypreciseto assessany systematic
variation in the age of the magmatismacrossthe province
or to estimatethe durationof magmatism.There havebeen
few detailed Rb-Sr studiesof the Paranfi-Etendekalavas,
partly due to the lack of sufficientlyporphyriticsamples
with suitablemineral phases.Mantovani et al. [1985b] re-
ported a combined Rb-Sr mineral isochron age of
135.5ñ3.2Ma from threeChapec6rhyolitesamples.In the
Etendeka,an Awahab rhyolite unit gave a Rb-Sr mineral
isochronageof 129.1+3.6Ma [Milner et al., 1995a].
Determiningthe ageof the Paranfi-Etendeka magmatism
is criticalfor evaluatingthe temporalrelationshipbetween
the volcanism and the opening of the South Atlantic
Ocean.This, in turn, would help to betterconstrainrecent
geodynamicmodelsthat attemptto link continentalrifting,
the mantleplume, and flood basaltproduction.Improved
estimatesfor the durationof magmatismwouldallow erup-
tion ratesto be inferred,which would have implicationsfor
the thermalmechanismbehindthe generationof the flood
basaltmagmas.Recentdatingeffortshave concentrated on
the potentialof the 40Ar-39Artechniqueto obtainmore
preciseage estimates[Baksi et al., 1991; Hawkesworthet
al., 1992; Renne et al., 1992, 1996a, 1996b; Turner et al.,
1994; Stewart et al., 1996].
Renne et al. [1992], in a study of the lavas of southern
Brazil, concludedthat the Parantimagmatismbegan at
133+1 Ma and lastedlessthan a million years,consistent
with the dataof Hawkesworthet al. [ 1992] for thisregion.
Fromthisresult,Renneet al. [ 1992] inferreda meanerup-
tionrateof-l.5 km3yr-1 similarin magnitude to thatesti-
mated for the Deccanprovinceof India. By contrast,in a
studycoveringthe full areal extentof the province,Turner
et al. [1994] suggested that the Parantilavaswere erupted
of over a longer interval(-10 m.y.) between137 Ma and 127
Ma, with a meaneruptionrateof-0.1 km3yr-1,an orderof
magnitudelessthan the estimateof Renne et al. [1992].
Additionalwork by the Open Universitygroup[Stewartet
al., 1996] confirmedtheir earlier conclusionof a signifi-
cant durationof magmatism.As this result conflictswith
the prevailingprejudicethat most flood basaltprovinces
were erupted in less than a few million years, it has
sparked debate, particularly about the validity of the
different analyticaltechniquesused. Turner et al. [1994]
comparedtheir preferredtechniqueof laser spot heating
with laser steppedheating for two whole-rocksamples.
One samplegave indistinguishable agesof 136 Ma from
both techniques,whereasthe other samplegave a poor
correlationof spot analysesbut produceda seemingly
good, althoughprobablymeaningless,plateauage. Renne
et al. [1996a] separatedplagioclasegrainsfrom samples
from a single region and analysedthem in two different
labs using different methods (Berkeley: laser stepped
heating:Nice: furnacesteppedheating)andthey gavecon-
cordantage ranges.Stewart et al. [1996] used a sample
from a Chapec6rhyolite unit to evaluatedifferent
 PEATE 223

39Ar techniques.Laser spot heatingof the groundmass largestgroup of complexesis in the Ponta Grossaregion
gave an isochronage of 131.8+1.4 Ma, which is indistin-
guishablefrom the plateauage obtainedfrom furnacestep
heating of a plagioclaseseparateanalysed in Canberra
(131.6+0.2 Ma), whereas laser step heating of the
plagioclaseseparategave a slightly older age (133.5+1.2
Ma isochronage).
It should be stressed that the conclusions of Renne et al.
[1992] were basedon samplesfrom just the coastalSerra
Geral escarpment,which representsa small fractionof the
overall province.Recentanalysesby Renneet al. [1996a]
on the PontaGrossadolerites(Figure2) that are inferredto
be the feedersof the northernParantilavas gave distinctly
younger ages (129.2+0.4 to 131.4+0.4 Ma) than these
workers had measured for the southern lavas, and a few
had agesas young as 120 Ma. Thus, Renne et al. [1996a]
revisedtheir estimateof the durationof the main pulse of
magmatismto about 3 m.y.. The older ages (135 to 138
and comprisesat least six discretecentres,including the
extensivelystudiedJacupirangacarbonatite[e.g., Huang et
al., 1995] whichhasa 40Ar-39Arageof 132 Ma [Renneet
al., 1993]. Other complexes of similar age occur in
southernBrazil (Anitapolis, 131 Ma [Renneet al., 1993]),
easternParaguay(127 Ma [Renne et al., 1993]), and in
Uruguay (Mariscala, 133 Ma [Stewart et al., 1996]). In
Africa, this Early Cretaceousalkaline magmatism was
concentratedin the Damaraland province of northern
'SOUTH'
MERICA' ''
PARANA ' ' '
VOLCANIC
PROVINCE
ß ßCerroCora'(•
RANGA,

Ma: n=9) of Turner et al. [1994] and Stewart et al. [1996] ß PARAGUAYAN .ANGOLAN
are found only in flows and dykes in the northern and ßPROVINCE ß ß PROVINCE

westernmarginsof the Parantilava field and at the baseof

.........

ß ' 'Sapucai•' ..........

someof the centralParantiboreholes,partsof the province ..........

not yet sampledby the Berkeleygroup.Thus, basedon the

..........

DAMARALAND- ß
available evidence, there are no valid reasonsto discount PROVINCE ß ß

these older ages, and it shouldbe emphasisedthat where

.......

......

both groupshave analysedsamplesfrom the samearea,the .......

agesare in good agreement.In the correlatedsequencesof

southeastBrazil and the Etendeka, the Berkeley group
URUGUAYAN
PROVINCE
...... AFRICA
have measuredages of 131.9+0.5 Ma to 132.9+0.6 Ma ...... 0 ..... 500' km
I I.
(n=9) [Renneet al., 1992, 1996b]andthe OpenUniversity
group found similar ages of 131.2+1.1 Ma to 132.9+2.8
,sJ':':'"'.-...":•".•'.•s•....•::'..•
x x x x x x x u x x x '..UU Km,.. x '..!• '
Ma (n=4) [Turner et al., 1994] with the exceptionof the 'l'Yt•'""'"'"'""••:':••'••'•••!ii!:.•.
/ / /
f'•'--'•.---.-'.••:..'•:::•.:•i•ijii::...•.-•_%
/ / / /
% % % % % % -& -• -• -• • • % % % % . ß
/•/ / / / - r -' / /
'[ ............... '""'"'"•'•':"':'•'":•'•:•:••;•.'•L•
." ." ." ." ." ." ." ." ." ." / ." ." ." ." ." ."
highestflow on the escarpmentthat is younger (129.4+1.3
"'""•:ii:•i•:i'":':"'""•'•:'"'""•i':::'
f.................
: '"'"'" ""'"'"'•'"'"'•;•'
x/Early
toMid-Proterozoic
basement'
/ / --20os
% ............. % .
Ma). A similar situation is found with the Ponta Grossa
•-------.-..------••.-•.-.-..•..•...:.•::•::•••5:.....!
,,, ,,, ,,, ,,, ,,, ,,, ,,, ,,, ,,, ,,, ,,, ,,, ,,, ,,,•.....•.,:•, •..:•
dolerites,althoughTurner at al. [ 1994] did find two dykes •'"'-'-'•'"•"•:•:'•:....
[-'-'"'"" - .......•.•!i•::•t,,
',, •,
'•'---'"'"'•--'-'.'-'•.:.:-:.•:•.:••'....:...i:-.,•,.,,,
ß ,,, ,,'
',, ',, •,
,,, ,,, ,,,
•,
,,,
•,
,,,
',, •,
,,, ,,,
•, •, ',,•'•:•:!:.i•[JKOrUSU
,,, •=,,----•.?.•::::.•.........••..• .......
% % % % %
with slightly older ages(134 Ma). In summary,the major-
ity of analysedsamplesgive 40Ar-39Aragesbetween129 •'
-"'"'"•'"'""•'""•:••-••••
Ma and 134 Ma, but there is alsogoodevidencefor earlier
•21os
magmatism inland (135-138 Ma [Turner et al., 1994;
Stewart et al., 1996]) and for youngermagmatismper-
sistingalong the coast(120-128 Ma [Turner et al., 1992;
Renneet al., 1996a;Stewartet al., 1996]).
....
:;:?:ii:.ii
cr,os:s:
,,,, ......
!ii:.::ii::iii::i::i
i::i:•:.:il
i::
:.::?:ii
:::.•
::iSpitzkoppe --22os
Contemporaneous
Alkaline Magmatism I I I
14OE 15OE 16OE

Several alkalic complexes,broadly contemporaneous
with the flood basaltvolcanism,were emplacedaroundthe
margin of the Parantibasin (Figure 6) and cover a wide
compositionalspectrumincludingcarbonatites,alkali gab-
bros, phonolites, syenites, and granites [Ulbrich and
Gomes, 1981; Milner et al., 1995a]. In South America, the
Figure 6. (a) Location of alkalic magmatism (circles with
crosses)contemporaneous with the Paranfi-Etendeka flood vol-
canism(dark grey shading);(b) detailedmap of the Damaraland
complexes(black shading)in relationto the Etendekalava field
(dark grey shading) and basement rocks (Early to Mid-
Proterozoic;speckledshading:Pan-African;light grey shading)
[from Milner et al.,
224 PARANA-ETENDEKA PROVINCE
Namibia and in Angola [Marsh, 1973]. The Damaraland
complexesform a northeast-trendinglinear feature along
the southernmargin of the Etendekalava field, extending
from the coast to 350 km inland (Figure 6b). They
represent high-level volcanic intrusionsand many are
inferred to be caldera-collapse structures. Intrusive
relationshipshave previouslybeen usedto suggestthat the
alkalic magmatismlargelypost-datedthe main flood basalt
eruptions, but Milner et al. [1995a] concluded that
magmatic activity in the alkalic complexes was
contemporaneous with the onsetof flood basaltvolcanism,
a relationshipalso seen in the Deccan and Siberianflood
basaltprovinces.Milner et al. [1992] have proposedfrom
detailed mapping and geochemicalcorrelationsthat the
oldest Etendeka rhyolite units were erupted from the
Messumcomplex.New dating(40Ar-39Ar,Rb/Sr)indicates
that the Damaralandcomplexesrangein age from 137 Ma
to 124 Ma. Many were active over a long interval(Messum
132 to 127 Ma, Okenyenya 129 to 123 Ma), and Cape
Crossand Paresishave 40Ar-39Aragesof 137 to 135 Ma
that are earlierthanthe main phaseof Etendekavolcanism
[Milner et al., 1995a;Renneet al., 1995b].
BASALT MAGMA TYPES
In termsof petrography,mostof the Parantilava pile can
be viewed as a homogeneous sequenceof virtuallyaphyric
tholeiitic basalts [Comin-Chiaramontiet al., 1988], but
significantcompositionalvariationsexist.Initial classifica-
tions divided the basaltsinto a low-Ti group largely re-
strictedto the southof the province,and a high-Ti group
dominantin the north [Bellieni et al., 1984a; Mantovaniet
al., 1985a]. As more data became available,the original
choiceof 2 wt% TiO2 to divide high- from low-Ti flows
appearedarbitraryand was not governedby any naturaldi-
vision in the distributionof TiO2 contentsin the basalts
(Figure 7a). Furthermore,based on other compositional
criteria, high-Ti flows found in the southare distinctfrom
the main groupin the north [Bellieniet al., 1984a].
Peate et al. [1992] attemptedto clarify the statusof dif-
ferent compositionalgroups within the Paranfi-Etendeka
lavas, with the aim of using these groupsboth to look at
the internalstratigraphyof the provinceasa whole and as a
means to simplify petrogeneticmodelling. Six magma
typeswere distinguishedon the basisof major andtraceel-
ement abundancesand ratios. Analysesof representative
samples of each magma type are listed in Table 1.
Compositionalcriteriathat allow flows to be assignedto a
particularmagmatype were deliberatelyselectedfrom ele-
mentsroutinely analysedby X-ray fluorescence(XRF) so
as to be as widely applicableas possible(Table 2). More
reliance was placed on elementssuch as the high-field-
strength(HFS) elementsthat are gen6rallyimmobile
during alterationprocesses.
Peate et al. [1992] foundthat
-90% of availablebasalticlava analyses(> 2000 with loss
on ignition(LOI) <2.5 wt%) couldbe classifiedinto oneof
the magmatypes.Someproblemsinevitablyarosebecause
the analyses were produced in several different
laboratories.
While it is useful to maintain the distinction between
low-Ti andhigh-Ti magmatypeswhendiscussing petroge-
nesis,it is unwiseto rely on the abundanceof a singleele-
ment for classificationpurposes.Thus, the magmatypes
defined by Peate et al. [1992] are groupedinto low-Ti
(Gramado, Esmeralda) and high-Ti (Urubici, Pitanga,
Paranapanema,Ribeira) varieties on the basis of a wide
range of similar compositional characteristics.The
Gramadoand Esmeraldamagmatypesare low-Ti magmas
in the senseusedby Hergt et al. [1991] to distinguisha
compositionally distinctive group of magrnas found
throughout the Mesozoic flood basalt provinces of
Gondwanathat have low Ti/Y values(<310) similarto or
less than MORB. Even though the Paranapanemaand
Ribeira magmatypesoverlapin Ti contentswith the low-
Ti Gramadoand Esmeraldamagmatypes,they will be re-
ferredto ashigh-Ti typesbecauseof their high Ti/Y values
(>310) and othercompositionalsimilaritieswith the other
high-Ti magmatypes(Figure7b).
The reliability of compositionaldata on the borehole
sampleshad to be demonstrated by Peate et al. [1992] as
only rock chippingswere available.Petrographic studyin-
dicatedthat, in a givenhole,the chipscamefrom a range
of levelswithin at leastone flow and probablysampled
severalflows. Althoughsampleswere hand-pickedin an
effort to selectchipsfrom a singlelithologicalunit and to
avoid altered fragments,analysesshowedthat most are
affectedby secondaryprocessesand have, for example,
markedly higher Na contentsthan surfacelavas.However,
there is good agreementbetween the boreholedata and
analysesof surfacelava samplesfor relatively immobile
elementssuch as Ti, Zr, and Y, and thus, using these
elements,the boreholesamplescan be reliably classified
into differentmagmatypesandcanbe usedto establish the
regionaldistributionof magmatypes[Peateet al., 1992].
Compositionand Distributionof Low-Ti Magma Types
The Gramado magma type has a distinctive trace
element signaturerelative to the high-Ti types, with a
greater relative enrichmentof large-ion-lithophile(LIL)
over HFS elementsand light rare-earthelements(LREE),
and a prominent negative Ti-anomaly on a
 PEATE 225

mantle-normaliseddiagram (Figure 8). The Esmeralda magmashave higher Ti/Zr (>60) than Gramadomagmas
patternis similarto the Gramado,exceptwith generally (Figure 7b). Both of these low-Ti magma types show
lower incompatibletrace elementabundances,
and it also significantbut variabledepletionof Nb and Ta relativeto
has a lesser degree of LREE enrichment. Esmeralda La (Nb/La0.5-0.8). Gramadosamples
havehigh878r/86Sr
i
(0.7075-0.7167) and low [•Ndi (-8 to -3) relative to bulk
I Earth values(Figure 9), whereasEsmeraldasamplesform
high-Ti: an almostlinear array displacedfrom the Gramadofield to
_ high-Ti basalts
+ Urubici lower 87Sr/86Sri(0.7046-0.7086) and higher[•Ndi (-4 to
ß Pitanga +3). They both have more radiogenic Pb isotope
[] Paranapanema compositions
(206Pb/204Pb>l
8.2) thanthe high-Timagma
+ []
* Ribeira types, but Esmeraldasampleshave a restrictedisotopic
low- Ti basalts range (206Pb/204pb=18.6-18.8)comparedto Gramado
02- ß Gramado lavas(206Pb/204Pb:
18.4-19.1:Figure10).
/x Esmeralda In southernBrazil, Gramadoflows are mainly restricted
to the coastalSerra Geral escarpment,whereasEsmeralda
flows often locally capthe lava pile and are more common
. Iow-Ti
on the centralplateau.The generalpatternof surfacedistri-
bution of these two magma types is consistentwith the
2 • 8 lO stratigraphicalrelationshipinferredfrom the centralParan•t
boreholes and Serra Geral road sections, with the
MgO (wt%) Esmeralda magmas forming a younger unit above the
lOO !
Gramadomagmas[Peate et al., 1992]. The southernmost

90- I
I [] [] •11
ß
ll
Paran/t lavas, in Argentina and Uruguay, appear to
compriseonly Gramadoflows, althoughthis may simply
ß
reflect the paucity of samples. In the Etendeka, the
80- Tafelbergbasaltsthat form mostof the main lava field are
compositionallyequivalentto the Gramadomagmatype.
N 70- Gramado flows are estimatedto compriseup to a third of
!- 60- the preservedParan/t-Etendeka lava pile, with Esmeralda
flows making up perhaps5 to 10%.
50- I
I
Compositionand Distributionof High-Ti Magma Types
40- I

Iow-Ti 'i high-Ti The 'Northern' Magma Types. Although the Ribeira,
30 i i i i i

5OO 6OO 700

Paranapanema,and Pitanga magma types span a wide
IO0 200 300 400
range in TiO2 contents(1.5-4.1 wt%), they are discussed
Ti/Y togetherbecausethey sharemany compositionalfeatures.
They have similar primitive-mantle-normalisedtrace
elementpatterns(Figure 8), differing only in the degreeof
high-Ti
18
magma [] enrichment (e.g., Ti/Y of Ribeira-360, Paranapanema
types [] [] -410, Pitanga-530). NegativeNb-Ta anomaliesrelativeto

Figure 7. (a) MgO vs. TiO2: this showsthe evolvednatureand

compositionalvariety of the Paran/t basalts. The dashed line
.: marks the approximate division between low-Ti and high-Ti
magmatypes. (b) Ti/Y vs. Ti/Zr: low-Ti magmatypes(Gramado,
Esmeralda) are distinguished from high-Ti magma types
• 12- +

(Urubici, Pitanga, Paranapanema,Ribeira) by low Ti/Y (<310).

LL +
Esmeraldamagmashave higher Ti/Zr (>60) than Gramadomag-
10 !
mas.(c) TiO2 vs. total Fe (Fe203(t)), showingthe differenthigh-
1 • 3 Ti basalt magma types. Data sources:Petrini et al. [1987],
Hawkesworthet al. [ 1988], Mantovani and Hawkesworth[ 1990],
TiO2 (wt%) Peate [ 1990], Peate and Hawkesworth[
226 PARANA-ETENDEKA PROVINCE

TABLE 1. SelectedCompositionalCharacteristics
of ParantiBasaltBagmaTypesandPreviousNomenclature

Magmatype Characteristics TiO2 Ti/Y Ti/Zr 87Sr/86Sri Previousnomenclature

Gramado low Ti/Y & Ti (south) 0.7-1.9 < 310 < 70 0.7075-0.7167 LTi(S), LPT, II, Tafelberg,
Albin
Esmeralda low Ti/Y & Ti (south) 1.1-2.3 < 310 > 60 0.7046-0.7086 LTi(S), LPT, I

Ribeira high Ti/Y, low Ti (north) 1.5-2.3 > 310 > 65 0.7055-0.7060 LTi(N), LPT
Paranapanema high Ti/Y & Ti (north) 1.7-3.2 > 330 > 65 0.7055-0.7063 HTi(N), IPT, III
Pitanga high Ti/Y & Ti (north) > 2.9 > 350 > 60 0.7055-0.7060 HTi(N), HPT

Urubici high Ti/Y & Ti (south) > 3.3 > 500 > 57 0.7048-0.7065 HTi(S), HPT, Khurnib

SeePeateet al. [1992] for additionaldetails.Equivalenttypesin theEtendekaare in italics[Duncanet al., 1988].

the LREE and LILE are developedto the sameextentin all compositionbut have a wider range that extendsto less
three magma types (Nb/La-0.64). They also have a radiogenicSr and Pb values(87Sr/86Sri=0.7048-0.7065,
restrictedrange in Sr-, Nd- and Pb-isotopecomposition 206Pb/204Pb =17.46-18.25: Figures 9 and 10). Data on
(87Sr/86Sri= 0.7055-0.7063,l•Ndi= - 1.6to -3.6,206Pb/204Pb coexisting Ti-magnetite and ilmenite indicate lower
- 17.81-18.12) relative to the other Parantimagma types, oxygenfugacityconditionsfor Urubici magmas(between
andhave beengroupedtogetheras the 'Northern'basaltson the quartz-fayalite-magnetite and magnetite-wustite
Figures9 and 10. buffers) than for Paranapanemaand Pitanga magmas
Thesethreemagmatypesalsohave a closespatialasso- (between the nickel-nickel oxide and quartz-fayalite-
ciation.They are foundthroughoutthe northernlava field, magnetitebuffers) [Bellieniet al., 1984a].
and down the western flank as far south as northern The surface extent of Urubici flows is restricted to a
Argentina [Peate et al., 1992]. Together,they compriseap- small strip (-100 km x 350 km) alongthe northeastflank
proximately half of the total preservedlava volume, with of the lava field in southernBrazil. Throughoutthis area
the Pitanga and Paranapanemamagmasprobably having they are interbeddedwith Gramadoflows [Peate et al.,
roughly equal volumeseachof-20% of the total, and the 1992]. Urubici samplesare also found near the base of
Ribeiraflows makingup -5%. Paranapanema lavascovera some of the central Parantiboreholes[Peate et al., 1992].
wide area of the northernParant region, centredon the Duncanet al. [1988] foundhigh-Ti flows (Khumibmagma
Parant River, with Pitanga lavas concentratedalong the type) in the formerly adjacentEtendekalavasof Namibia,
northeastand easternmargins.This outcroppatternis con- northof 20øS(Figure 6b) that are compositionally equiva-
sistentwith the borehole data, where Paranapanemasam- lent to the Urubici magmatype [Peate et al., 1992]. If
ples overlie Pitangasamples.Ribeira sampleshave a simi- Africa and South America are juxtaposedto their pre-
lar surface distribution [Petrini et al., 1987] to the Atlantic Ocean positions,then the Urubici and Khumib
Paranapanemasamplesin the northern Parant. However, flows are directly adjacent. The coincidenceof the
Ribeira sampleswere found only in one of the central southernlimit of their outcropindicatesthatthe Gondwana
Parant boreholes, where they were below Pitanga and plate reconstruction of de Wit et al. [1988] providesa
Paranapanema flows (Figure 11) [Peateet al., 1992]. reasonablefit to within 100 to 200 km for this region.The
The Urubici Magma Type. Like the Pitanga magma Urubici (and Khumib)magmatype is volumetricallymuch
type, thishasa high TiO2 content(>3 wt%). It sharesmany less significant than the other high-Ti magma types,
incompatible trace element features with the Pitanga comprising less than 5% of the total preservedlava
magmatype but at higherabundances (Figure8). It is read- volume.

ily distinguishedby its high Sr (>550 ppm) and low INTRUSIVE MAGMATISM
Fe203(t) (<14.5 wt%: Figure7c), andits greaterheavyrare
Ponta GrossaDyke Swarm
earthelement(HREE) fractionation(Tb/Yb Urubici-0.58,
Pitanga-0.38; Figure 12). Urubici samples partially A prominent swarm of northwest-southeast-striking
overlap the 'Northern' basalts in Sr-Nd-Pb isotope dykesintrudesthe Precambrianbasementand Paranti
 PEATE 227

TABLE 2. RepresentativeAnalysesof Paranfi-Etendeka

BasaltMagma TypesandRhyolite Subgroups

sample B448 DSM DUP DUP FEG MM FEG B980 PAR CB DUP MM PRG MM
name 06 30 38 92-23 90-90 92-92 06 1110 35 90-88 86 90-13
group ESM ESM GRA GRA AG CDS SM RIB PMA PIT URU GU GI OU
Low-Ti Basalts PalmasRhyolites High-Ti Basalts Chapec6Rhyolites

SiO2 49.10 51.10 50.89 56.14 66.95 66.88 70.67 50.07 50.13 51.27 53.01 64.27 65 57 65.70
TiO2 1.52 1.37 0.95 1.68 1.08 0.93 0.69 1.57 2.19 3.35 3.76 1.48 1 24 1.22
A1203 15.19 13.77 14.92 13.19 12.90 12.79 12.54 15.51 13.56 12.70 12.86 12.92 14 56 12.92
Fe203 13.11 13.45 10.34 13.54 6.61 6.14 4.92 13.16 14.64 14.96 12.68 7.58 758 6.52
MnO 0.20 0.21 0.17 0.19 0.11 0.11 0.06 0.20 021 0.23 0.19 0.15 0 O4 0.13

MgO 6.39 6.13 7.99 3.26 1.51 1.77 0.59 5.60 553 4.34 4.34 1.40 O85 0.13
Na20 11.52 10.73 11.61 6.89 3.47 2.89 1.35 10.62 994 8.56 8.30 2.94 1.46 3.26
CaO 2.63 2.55 2.44 2.81 3.33 2.80 2.67 2.50 2 54 3.02 2.57 3.41 2.92 3.26
K20 0.17 0.54 0.51 2.57 3.47 3.96 5.15 0.58 0 99 1.05 1.70 4.15 5.34 4.35
P205 0.16 0.16 0.16 0.24 0.32 0.27 0.20 0.20 026 0.54 0.58 0.46 0.45 0.32
Sc 37 42 40 32 18 18 16 44 38 33 29 12 .
12
V - 323 221 317 93 90 28 - 397 405 343 66 46
Cr 267 94 307 54 12 19 7 114 126 46 75 12 44
Co 46 50 44 41 19 18 13 46 52 34 36 18 16
Ni 94 58 99 16 8 8 6 93 60 20 58 7 5
Cu - 169 99 61 90 57 21 - 173 140 267 11 18
Zn - 89 72 105 76 74 82 - 92 132 107 129 106
Ga - 21 17 22 18 18 17 - 21 26 26 23 24
Rb 9 19 10 86 148 174 226 8 24 30 30 103 117 131
Sr 154 163 216 211 145 131 85 273 380 459 764 401 275 321
Y 28 29 23 44 48 40 73 25 31 40 39 71 69 66
Zr 78 100 92 220 279 258 342 110 154 252 307 638 555 595
Nb - 6 9 16 22 21 26 - 13 23 27 51 49 48
Ba 148 163 243 461 548 681 721 272 344 465 600 1002 927 1046
La 4.6 8.35 10.6 32.2 40.1 40.1 54.5 14.0 21.5 33.4 42.5 75.3 86 75.2
Ce 13.0 20.8 22.9 66.3 86.9 84.7 111 30.8 44.0 76.3 90.4 163 155 159
Nd - 14.5 12.8 35.9 40.8 40.0 52.2 19.3 25.3 48.0 54.3 83.9 82 77.7
Sm 3.1 3.97 3.18 7.94 8.53 8.03 11.1 4.1 5.25 11.6 11.6 16.6 .
15.0
Eu 1.0 1.41 1.11 1.92 1.86 1.60 1.91 1.48 1.79 2.94 3.53 4.33 3.62
Tb 0.8 0.90 0.63 1.29 1.39 1.20 1.78 0.69 0.82 1.34 1.54 2.29 1.89
Yb 2.3 2.92 2.09 4.03 4.30 3.79 6.19 2.2 2.65 3.53 2.99 5.84 5.23
Lu 0.3 0.47 0.35 0.65 0.66 0.54 0.97 0.45 0.40 0.59 0.44 0.86 0.71
Hf 2.0 2.71 2.15 5.84 7.11 6.49 8.61 2.9 3.92 7.13 7.97 15.0 13.9
Ta 0.2 0.37 0.50 1.12 1.76 1.65 2.23 0.61 0.87 1.58 1.92 3.27 3.09
Th 0.74 1.75 2.04 9.00 13.0 13.7 18.2 1.74 2.60 3.67 4.25 8.98 12.6
U 0.21 0.71 - 2.23 4.58 5.10 5.81 0.24 0.87 1.00 1.34 2.15 2.84

gSri 0.9 20.1 42.2 98.9 146a 209 292 18.3 19.7 13.3 7.1 18.7 23.4 47.3

gNdi 2.2 0.7 -3.8 -6.3 -6.9a -7.2 -7.3 -3.2 -2.8 -2.2 -3.4 -5.1 -5.4 -5.1

a isotopedatafrom sampleFEG92-39.
Datafrom:Petrini et al. [1987];Peate[1990];Albertiet al. [1992];Garlandet al. [1995];PeateandHawkesworth
[1996].All samples
haveL.O.I. (losson ignition)<l.6wt%, exceptPRG86(3.1wt%).Basaltmagmatypes:ESM=Esmeralda; GRA=Gramado; RIB=Ribeira;
PMA=Paranapanema;
PIT=Pitanga;
URU=Urubici.Rhyolitesub-groups:
AG=AnitaGaribaldi;CDS=Caxias do Sul; SM=SantaMaria;
GU=Guarapuava;
GI-Giraul; OU=Ourinhos.
gSriand gNdi valuescalculated
at 130 Ma relativeto Bulk Earth(present-day
values:
87Rb/86Sr=0.0847;87Sr/86Sr=0.7047;147Sm/144Nd=0.1967;
143Nd/144Nd=0.51264).
Isotope ratios normalised to NBS 987
(87Sr/86Sr=0.71025)
andJ&M Nd
228 PARANA-ETENDEKA PROVINCE

100--
f oUrubici the feedersto lavaseruptedtowardsthe continentalmargin
_

_
]- Pitanga and subsequentlyeroded. However, 40Ar-39Ar ages
--
/ ø Paranapanema
/ [Turner et al., 1994; Renne et al., 1996a] indicate that
• ], Ribeira many of the dolerites are of similar age to the surface
50 ,/'•% / • Esmeralda
- lavas, and thus were probably feeders to the northern
-
J Parantilava pile, althoughthere are a few coast-parallel
-- dykesthat give youngerages(120-125 Ma). Raposoand
Ernesto [1995] showed from a study of magnetic
•L10-- susceptibility anisotropy that many dykes preserve
_
evidencefor lateralflow. Fabricsindicatinga morevertical
_ flow directionare more commonin the southeast part of
5-- the swarm,perhapsindicatinga sourcecloseto the proto-
_

Atlanticrift. East-west-trending
doleriticdykesnearMorro
Th Nb La Sr P Zr Eu Tb Yb Vermelho (MV on Figure 2) in southernAngola may
Ba K Ta Ce Nd Sm Hf Ti Y representa continuationof the Ponta Grossadyke swarm
onto the African Plate prior to continental separation
Figure 8. Primitive-mantle-normalised
diagramfor averageleast
[Piccirillo et al., 1990].
evolved(>4.5 wt% MgO) samplesof eachParantibasaltmagma
type.NormalisingvaluesfromSunand McDonough[1989]. Data
sources:Peate [ 1990], Peate and Hawkesworth[ 1996].
15.8
Esmeralda -1- Northern
basal;
/• • basalts Urubici
•• basalts
_-•_ 15.7-
0
Gramado
-5 15.6-
••
•N%,•
a• basalts
-7
I rhy6/ite.'.:
, ,
•• 0.705 0.706
••• • Palmas
15.5
. ./ .......... •:••••:•••..• • rhyolites
basalts&
-10 15.4
0.05 0.20 0.25
(87Sr/86Sr)i
-Q 39.0-
Figure 9. Initial Sr- andNd-isotopiccompositionof Parantimag- Q_ [] Northern
J
mas (at 130 Ma). Data for the 1ow-Ti magmasare plottedon the
Chapec61
,, •'•-•o
ßrhyolites
J .e,•."•
,",,:"-•"
c)
main diagram,and the inset is an expandedview to highlight
featuresof the high-Ti magmasthat havea more limitedisotopic
variation. 'Northern Basalts' group combines the Pitanga, Q_ 38.5-
Paranapanemaand Ribeira magmatypes.Crossmarksbulk Earth ß Gramado
estimate. Data sources:Cordani et al. [1988], Mantovani and
Hawkesworth [1990], Peate [1990], Peate and Hawkesworth A Esmeralda
[1996]. Palmas

sedimentsin the PontaGrossaregion, and rare dykesalso

38.0-
/ o rhy.
olites
17.0 17.5 18.0 18.5 19.0 19.5
crosscutthe lava pile (Figures2 and 13) [Pi½½irilloet al.,
2o6pb/2O4pb
1990]. Most of thesedoleriteshavesimilarmajor andtrace
element compositionsto the Paranapanema lavas,with a
few similar to the Pitangalavas[Regelous,1993]. A com- Figure 10. Variation in present-dayPb isotoperatiosfor Paranti
parisonof meanpalaeomagnetic poles[Ernestoand Pa½½a, basaltsand rhyolites. Data for samplesof recentTristan plume
activity (Tristan da Cunha, Gough, Inaccessible)and from the
1988] suggested that the doleriteswere emplacedafter the Walvis Ridge plume trace are plotted for reference[Sun, 1980;
eruptionof the main preservedportionof the Parantilavas, Richardsonet al., 1982; le Roex et al., 1990; Cliff et al., 1991].
andsoPic½irilloet al. [ 1990]proposedthatthe dykeswere Symbolsand datasourcesfor Parantilavasasfor Figure
 PEATE 229

N-1 136.9'•_1.3 I.•

CB
Paranapanema
x I 1N (•S G
•.,./..:./.,•
•) '• •.. _•/-Esmeralda
132.2+1.0
•::•. -
rhyolites S
Palmas
, • •, ,.?•:•:•:•:•:?• _•.• .:: .•.:•::7:::::::•...,•,?:•::•,:....•
.......ß Z• 129.4•1.3
••/ x•/ x/ x/ / .
.....
•'• -......
",•-;¾-'<•
.....
•;<'•'
.... •4".',•........ •;.•;½;.';•:,'•,'
';";' •'•"'•:":•'•:•: •
........ -•.••.'0
•'- •.•:•:•:•:•:•:•
......
................................
...................
-............
'1323•8 ........
-- 0 ' xxxxxx ....
•:•;•;'
?•">•;,'-
•½;•'.•;•.•.,•>:::•;•'•Y•
'//•• 32
................
;6•'2"•>'"
:.•;•/••
.................
'"'•<":•7"•:'<';f•½'5';•?;:•:•:•½5 *• •'' 131
'8•1
'25';-
'::'e•
:::• 133.1•1.
•:•';•::*':;:e.........
r::
............. •' ..:.:.:::.•:,•;½::½•[;•{5•5::5:•[•:
:::::•::•:::•'.;" 32
:"•""•" 4+.......................
•;•'• 126.8•.•
":•::.';;;:;::':.:.:'.'.':;•'•:•:J55•
.....................
-•-.=-.-;---=-.•-Z.• .......
............. --'2
276+1
::::::::::::::::::::::::::::::::::::::::::::::::
• % % % % % % •.>.... .-. ß:,,,....:.x•
4• •,z.-...,+>)•x • .::::
.... % % .... % % % % ..........................................
<.,-.-.-...:.:-x<<•<<<.:.:ee::•:::::•:•:::-:-:-:-'-'-'-'""

% % % % % .... .......
ß ... • % % % % %

--1 0 200km -• •'•x N• •'/• • •ibe

vedical x140','••?•??
exaggeration ,,. . ,' ' Pltanga
ß UrubiciGramado . ,
- -2km ' •"

Figure 11. SchematicN-S crosssectionthroughthe Parantilavas [Peate et aL, 1992; Stewart et aL, 1996]. The
timelinesfor 135 Ma, 133Ma and 129 Ma (constructed
from40Ar-39Ar
agesof Stewartet al. [1996]) cut acrossthe
stratigraphicunits defined by the compositionalmagma types of Peate et al. [1992]. This would indicatethat the
magmatypesare diachronous.

wt% SiO2 and2.2-2.6 wt% TiO2, andthe two typescannot

2.5- be related by any simple petrogenetic process
-H-+ ++
[Hawkesworthet al., 1992]. Regelous [1993] concluded
++ that neither the Paraiba nor the Ubatuba dolerites had ex-
z
q- +
trusive equivalentsin the Paranti lava pile to the west.
However, it is possiblethat they fed the formerly adjacent

I--1.5- .•a•=•tllE@
!' Grarnado
+Urubic, basalticlavas that crop out in northwesternAngola south

• ,,,',',. / Esmeralda
ßPitanga/ of Luanda(Figure2). TheseAngolanlavashaveK-Ar ages
i•.,.•i;.e / Palmas o Paranapanema•
of 124-145 Ma [Piccirillo et al., 1990] but no composi-

1.0:
•"•';•"
ß[":=;'
[erhyolites
Chapec6 []rhyolites tionaldataare availablefor comparison with the dolerites.
20 SAO PAULO- ß
':SOUTH ....
La/YbN
RIO DE JANEIRO

'AMERICA ß '
ß '
ßCOASTAL
' SWARM
DYKE

Figure 12. Tb/YbN vs La/YbN for Parantibasaltsandrhyolites(N ß .

denoteschondrite-normalised). For the 1ow-Timagmas,Tb/YbN VOLCANIC

remainsfairly constantover a rangein La/YbN from 1 to 8. For PROVINCE.
the high-Ti basalt magma types, the Paranapanemamagmas ß .

overlap with the 1ow-Ti Gramadomagmas,whereasthe Pitanga

andUrubicimagmasare displacedto successively higherTb/YbN E•SYE•R•i
ßPARAGUAY
andLa/YbN. Data sourcesfor Parantilavasasfor Figure9.
ßDYKE ß ß
ßSWARM ß
AFRICA
S•o Paulo - Rio de daneiroCoastalDyke Swarm
PONTA G ROSSA
ß ß DYKE SWARM .
Numerousnortheast-southwest-trending dykes, all with
..... ETENDEKA
high-Ti compositions(Ti/Y >310 [Comin-Chiaramontiet VOLCANIC
al., 1983]), intrude the Precambrianbasementrocks along ..... PROVINCE
the coast between S•o Paulo and Rio de Janeiro States TAFEl-BERG .....
(Figure 2). The dyke compositionsdefine two distinct &-REGIONAL.....
DOLERITES
groups,termed the Paraibaand Ubatubamagmatypes by
Regelous[1993]. 40Ar-39Aragesindicatethat the dykes ......
0
i .......
500 km
I'
were emplacedbetween129 and 133 Ma, overlappingin
age with the Ponta Grossadykes and the northernParanti
Figure 13. Locationof principaldyke swarmsassociated
with the
lavas [Turner et al., 1994]. The Paraibadoleriteshave 50- Paranfi-Etendekamagmatism.Arrows indicatepossiblelinks be-
53 wt% SiO2 and 3.2-4.1 wt% TiO2 whereasthe Ubatuba tween dykes'and surfacelavas of similar age and composition
doleriteshave relatively evolvedcompositions with 54-58 [modifiedfrom Regelous,
230 PARANA-ETENDEKA PROVINCE
OtherParang IntrusiveMagmatism
A regional airborne magneticsurvey [Druecker and
Gay, 1987] revealed a significant swarm of northwest-
southeast-trendingdykes that covers most of eastern
Paraguayto the west of the Parantilava field (Figure 2).
Few of thesedykeshavebeensampledbut they seemto be
similar in compositionto the Pitangaand Paranapanema
magmatypes.One dyke gavean old 40Ar-39Arage (137
Ma), similarto otherhigh-Ti lava flows alongthe western
margin [Stewartet al., 1996].
Intrusivemagmatismis lesscommonin southernBrazil.
Occasionaldykesand sillsare foundalongthe SerraGeral
escarpment,and along the coastalmargin where most
strike approximatelyparallel to the coast.A few dykes
haveGramado-typecompositions, with Urubici-typedykes
more commonnear the coast.On the escarpmentnear Silo
Joaquim,numerousdykes and sill-like intrusionswithin
the lava pile, includinga 100-m-widedyke intrudingthe
Parantibasin sediments,all sharea remarkablyhomoge-
neous,Esmeralda-likecomposition (l•Ndi---+0.5,87Sr/86Sri
-0.7059, Ce/SmN-1.3, where N denoteschondrite-nor-
malised[PeateandHawkesworth,1996]).
Boreholeshave revealed numeroussills, 2 to 200 m in
thickness, that intrudethe Parantibasinsediments, mainly
withinthe Palaeozoicstrata.Somesillscropout at the sur-
face, north of S•o Paulo (Figure 2). The sills can reacha
combinedthickness of over 1000m in thenorthernpartsof
the basin [Bellieni et al., 1984b]. The sills are all tholeiitic
and are compositionallysimilar to the Paranti lavas
[Bellieniet al., 1984b;Regelous,1993]. Peateet al. [ 1990]
showedthatthe distribution of compositional typesamong
the sillsmirroredthe distribution of lava magmatypesat
the surface, with Gramado-like sills in the south and
PitangaandParanapanema
sillsin the north.
EtendekaIntrusiveMagmatism
Doleritic dykesand sills are abundantwithin a 100-km-
wide zone along the Namibian coast from the main
Etendekalava field to at leastas far southas WalvisBay
(Figure2), principallyintrudingthe Damarabasementand
Karoo basin sediments[Marsh et al., 1991]. The limited
geochronologicaland palaeomagneticdata indicate that
they are of similarageto the Etendekalavas[Erlank et al.,
1984]. Erlank et al. [1984] identified three suitesof do-
lerites:the relativelywidespreadTafelbergand Regional
dolerites,and the Horingbaaidolerites.The Tafelbergdo-
leriteshave identicalpetrographical and compositional at-
tributesto the Tafelberglavas,the Etendekaequivalentof
the Gramadolavas in the Paranti.The Regionaldolerites
are more mafic than the Tafelbergdolerites,but otherwise
sharemany geochemicalattributeswith them. The MORB-
like Horingbaai dolerites(l•Ndi-+8 [Hawkesworthet al.,
1984]) form thin dykes and sills that intrude basement
rocksand the overlyingEtendekalava pile in a restricted
portionof the coastalarea near Cape Cross(Figure 6b).
Erlanket al. [ 1984] presented40Ar-39Aragesof---128Ma
for the Horingbaaidolerites.
Duncanet al. [ 1939]recognised fourgeochemically dis-
tincttypeswithin the Huab sill complex,a suiteof sillsthat
intrude basement rocks and Karoo sediments in the Huab
River valley (Figure 6b). One type is similar to the
Tafelbergmagmas,but the otherthreetypesare closestin
compositionto the Horingbaai dolerites.Field evidence
suggeststhat they also postdatethe main preservedlava
pile. Early Cretaceous
dykeshavealsobeenreportedfrom
the coastalregionnear the SouthAfrican-Namibianborder
andfarthersouthnearCapeTown [Reid,1990].
RELATIONSHIP OF MAGMATISM TO OPENING OF
THE SOUTH ATLANTIC OCEAN
Asymmetryof Lava Distribution Relative
to Proto-AtlanticRiff
Severalideashave been proposedto explainthe pro-
nouncedasymmetricaldistributionof the lavas relative to
the eventuallocationof continentalseparation. O'Connor
and Duncan [1990] argued that the lava distributionre-
flectsthe positionof the Tristanplumeat the time of mag-
matism,which they placedbeneaththe centralParantiat
130 Ma. Thompsonand Gibson [ 1991] consideredthat
interactionof the plume with the baseof the lithosphere
might play an importantrole in the locationof flood basalt
volcanism.They suggestedthat the ParantiBasin was a
lithospheric'thin-spot'where material,divertedlaterally
from the main plumeaxisto the east,coulddecompress to
producemagmas. White and McKenzie [1989] linked the
generationof floodbasaltsto riftingabovea mantleplume
and inferredthat the lavaswould be eruptedfrom the rift
zone. For the Paranti-Etendekaprovince, pre-existing
topographicrelief near the Etendeka(inferred from the
oversteppingof the lavas onto basement)was considered
by White and McKenzie [1989] sufficient to prevent
widespreadflow of lavas eastwardsontothe African Plate.
Another explanation for the asymmetryis that rifting
occurredvia simple shearrather than pure shear [Peate,
1990].Harry and Sawyer[ 1992] showedthatthe develop-
ment of a horizontalpressuregradientin the lower crust
during the early stagesof extensioncould provide a
mechanismfor the lateralflow of magmaover 100-200

from the rift zone to beneath the Paranti province, thus
accountingfor the asymmetriclava distribution.
Compositionaland age correlationsbetweendykesand
lavas give an indication of where the lava flows were
eruptedfrom. In mostplacesthe dykesare similarin com-
position to the local lava flows (Figure 13). Regional
compositionalsimilaritiesbetweensills and the overlying
lavas also suggestthat the lavaswere eruptedin the same
generalregion where they now crop out. This emphasises
that the different magma types do not represent the
temporalevolutionof a single mantle sourceregion and
that magma generationoccurredover an area comparable
to the areal extent of the lavas. The absenceof any old
dykes (along-strike equivalentsof the easternParaguay
dykes:136-138 Ma) alongthe Braziliancoastsuggests that
at least some of the early magmas were generatedand
erupted at considerabledistances-westof the eventual
Atlantic riff, and that the Paraguayandykes had not
transportedmagma laterally from this rift zone. The
magma types were erupted in different parts of the
province, perhapsat different times, as the result of a
complex interplay between plume-lithosphereinteraction
and lithosphericextension.
Implicationsof Regional Stratigraphy
Stratigraphicalstudiescan reveal the internal structure
and sequentialdevelopmentof the lava pile. This knowl-
edge is critical to the understandingof how magmatic
sourcesand processesvaried during the evolutionof the
province,and it also providesa meansof determiningany
shiftsin the principal locus of magmatismthat might be
linked to regionaltectonicprocesses.
The distributionof different compositionswithin the
lava pile discussedabove indicatesthat flows of each
basalt magma type tend to form a relatively coherent
lithostratigraphical unit. By analogy with detailed
stratigraphical sequences through other flood basalt
provincessuch as the Deccan and Columbia River, Peate
et al. [1990, 1992] made the reasonableassumptionthat
these units could also be considered as
chronostratigraphicalunits. From this assumption, it
appearsthat the internal structureof the Paranglava pile
comprisesan overlappingsequenceof units dipping to-
wards the north, which suggestsa northward-migrating
sourcefor the magmatism(Figure 11). Peate et al. [1990]
suggestedthat this migration occurredin responseto the
northwardpropagationof rifting duringthe initial opening
of the South Atlantic Ocean. The 40Ar-39Ar data of Renne
et al. [ 1992, 1996a] are consistentwith sucha model, with
the southernBrazil and the Etendekalavaseruptedat 132-
PEATE 231
133 Ma and the northernlavas(as representedby the Ponta
Grossa dolerites) erupted at 129-131 Ma. Turner et al.
[1994], however, suggestedthat the earliest magmatism
was inland, in the west and north, and then became concen-
tratednearthe northward-propagating SouthAtlanticrift.
It is importantto realisethat the Paran/tmagmatypes
distinguishedby Peate et al. [1992] are definedsolelyon
compositional characteristics. They are not
stratigraphicallydefined units as is the case in the
subdivisionof several other flood basalt provinces(e.g.,
Siberia, Deccan) where continuoussectionsthrough the
lavas that can be correlated on a regional scale are more
common. The 40Ar-39Ar data of Turner et al. [1994] and
Stewartet al. [ 1996] suggest,in fact, that the magmatypes
are diachronous, with differentmagmatypesbeingerupted
simultaneously in differentplacesand over a long period
of time. This is illustratedin the crosssection(Figure 11)
where the time lines clearly crosscutthe magma type
boundaries.The oldest 40Ar-39Ar Paranftages are from
Paranapanema and Pitangasamplesin the north and west
of the lava field that, based on the borehole stratigraphy,
should actually be younger than the low-Ti magmas of
southeastBrazil. Between 133 and 132 Ma, Paranapanema
flows were being erupted in the northernhalf of the
provinceat thesametimeasGramadoandUrubiciflowsin
the southernhalf. New data (S.P. Turner, unpubl. data)
from a cored borehole in northwest Uruguay also cast
doubton the chronologicalsignificanceof the geochemical
stratigraphy asParanapanema lavasoverlieGramadolavas
in the centreof the provincebut underliethemin the south.
Duration of Magmatism
It is importantfor modellingthe geodynamicprocesses
involvedin flood basaltgenerationto knowthe durationof
magmaticactivityandwhetheror not the eruptionratevar-
ied during this interval. Although the Paran•-Etendeka
lavasappearto have beeneruptedover quitea significant
time interval (-3 m.y. [Renne et al., 1996a];-10 m.y.
[Stewartet al., 1996]), improvedestimates for the duration
of magmatismare clearly needed.The completestrati-
graphicintervalof magmatismmustbe sampled,and we
also need to ascertainthe volume of magma eruptedas a
function of time in order to assessvariationsin eruption
rate. If the magmatypes are not chronostratigraphic then
theseobjectivesare difficult to achievewithouta largere-
gionalcoverageof precise40Ar-39Arages.Stewartet al.
[ 1996] suggestthat eruptionratesincreased
with time, with
a lessvoluminousphaseat 138-133 Ma followedby the
main pulseat 133-129 Ma.
Palaeomagnetismpotentially offers an
232 PARANA-ETENDEKA PROVINCE
meansof establishingthe durationof the Paranfi-Etendeka
magmatismbecausethe frequencyof geomagnetic field re-
versals was high during the Early Cretaceous [e.g.,
Harland et al., 1990]. Palaeomagneticdata are available
from twenty stratigraphicsectionsin the Paranti[Ernesto
and Pacca, 1988], and each sectionrecordsup to four
polarityreversals.The main drawbackto establishing the
total numberof reversalsspannedby the magmatismhas
beenthe apparentlack of suitablestratigraphic markersto
correlate polarity intervals between sampled sections.
Milner et al. [1995b] used a lithostratigraphiccorrelation
based on compositionally distinct rhyolite units to
demonstratethat magmatismin the southeasternParanti
spannedat least ten different polarity reversals.Between
135 Ma and 130 Ma, the average length of a polarity
intervalwas 0.24 m.y., which suggests a durationof about
2.4 m.y. for eruptionof the 1 km of lava in this region.
This figure compareswith estimatesfrom the 40Ar-39Ar
datingstudiesof <1 m.y. [Renneet al., 1992] and-3 m.y.
[Turner et al., 1994; Stewart et al., 1996] for this part of
the province.
Future studies combining magnetostratigraphy,lava
composition,
precise40Ar-39Aragesand coredborehole
samplesoffer the besthopeof enhancingour knowledgeof
the sequential development of the Paranfi-Etendeka
province.About a quarterof the total eruptedvolumelies
in northernArgentinaand Uruguayandyet little is known
about the compositionand chronologyof these lava se-
quencesasthey are largelyburiedunderyoungersedimen-
tary rocks. Turner et al. [1994] measuredyoung ages
(-•128 Ma) on two samples,but it is uncertainwhether
theseagesare representative of the whole lava pile in this
region or whether they belong to a volumetricallyminor
laterphaseof activityas is foundfurthernorthon the Serra
Geral escarpment (128-129 Ma) and in coast-parallel
dykes in the Ponta Grossaregion (120-125 Ma [Renneet
al., 1996a]).
Rifling History of the SouthAtlantic Region
The age for the onsetof sea-floorspreadingwithin the
southernSouthAtlantic Oceanapparentlydecreased north-
wards [Austinand Uchupi, 1982]. The oldestseaflooroff
the southernAfrican coastnear Cape Town has been as-
signedeitherto magneticanomalyM13 or M9 (137 Ma or
130 Ma), whereas the earliest recognisablemagnetic
anomalyat the latitudeof the Paranfi-Etendeka provinceis
M4 (-127 Ma: referencesof Renneet al. [1992]). The age
of the earliestoceaniccrustcan only placea lower limit on
the age of continental rifling in a particular region.
However, studies of the tectonic and subsidencehistories
of both onshoreand offshorebasinsalso suggestthat the
onsetof rifling was earlierin the south[Changet al., 1992;
Light et al., 1992]. Rifting began in southernArgentina
andthe tip of SouthAfrica at 200-220 Ma, andin northern
Argentina(Coloradobasin:Figure 1) at-170 Ma [Light et
al., 1992]. The earliest extensionin the offshorebasins
alongthe Brazilianmarginwas evenyounger,at-140 Ma
[Chang et al., 1992].
The geometricalrelationshipof the PontaGrossadyke
swarm with the Brazilian coastline and associated coast-
parallel dyke swarmsis reminiscentof the classictriple-
junctiongeometry[Burkeand Dewey, 1973] linkedto do-
mal uplift and rifling abovea hotspot.For the plume-im-
pactmodel of flood basaltgeneration[e.g.,Richardset al.,
1989], the triple junction cannotmark the site of plume
impactbecausethe earliestmagmatismwas either to the
northwest[Turner et al., 1994] or to the south[Renneet
al., 1996a]. Furthermore,dykes are not orientedradially
aboutthe locationof the inferredplume head,but have a
dominantlynorthwest-southeast trend over a wide area of
the province(PontaGrossa,easternParaguay),a trendthat
can be linked to reactivation of Proterozoic basement struc-
tures.Dyke orientationsare determinedto a largeextentby
the regionalstressfield at the time of emplacement. The
Saladoand Coloradobasinsin northernArgentina(Figure
1), and the inferred Paran•t-Chacobasin deformationzone
all show a similar northwest-southeast orientation. These
featuresare all consistent
with modelsthat requiresignifi-
cant internal deformation and clockwise rotation within
centralSouthAmerica during continentalbreakupas a re-
sult of the CentralAtlantic regionremainingclosedas the
SouthAtlantic Oceanopened[Niirnbergand Mailer, 1991;
Turner et al., 1994]. Furtherwork is requiredto decipher
the exact relationshipbetweenmagmatismand extension
duringthe developmentof the Paran/t-Etendeka province
and the opening of the South Atlantic Ocean. The sub-
mergedlavasin the offshorebasinsare poorlystudiedand
yet form a key piecein thispuzzle.
The exact locationof the Tristanplumeaxis duringthe
Paran•t-Etendekamagmatism is still uncertain,although
VanDecaret al. [ 1995] have presentedintriguingevidence
for a 'fossil'plume conduit in the sublithospheric upper
mantlebeneaththe northeastParan•tlava field. They found
a low velocity anomaly, roughly cylindricalin form, 300
km acrossand extendingfrom 200 km to at least600 km
depth,which they inferredto be the thermal,andpossibly
chemical,remnantof the originalplume conduitthat sup-
plied the Tristanplume head.If this interpretationis cor-
rect, it impliesthat the uppermantleand lithosphere have
remainedcoupled since the breakup of Gondwana.The
long-term presenceof a thermal anomaly beneath
 PEATE 233

northernParantiwill alsohaveimplicationsfor post-Paranfi

magmatismanduplift in this region. Low¾i IiSh-¾i o o
SHALLOW-LEVEL PROCESSES IN
0.725-
]eGramado
+Chapec6
Urubici •0
],• Esmeralda[] Northern
[o Palmas o •.,o

[ rhyolites
ßrhyolites
d•ø
PARAN.•-ETENDEKA BASALT PETROGENESIS • 0.720-
o

Fractional Crystallisation ß %o
'• 0.715

The majority of Paranfi-Etendekalavas have low MgO

contents(3.0-6.5 wt%: Figure 7a) and low compatibleele- •0.710'
ment contents(<100 ppm Ni). Sampleswith more mafic
compositions(MgO 6.5-9.0 wt%) are volumetricallyin-
significant(<<2%), and still have compositions
(Mg# <65:
0.705- i I I
=•11
i I

Mg# is atomic ratio 100'Mg/(Mg+0.85Fetotal))that are far 45 50 55 60 65 70 75

removed from the expected compositionof melts in
equilibriumwith mantle peridotite.These characteristics
indicatethat all the magmashave undergoneextensive Figure 14. SiO2 vs. 878r/86Sri
crystalfractionation,presumablyin sill complexesnearthe The effectsof crustalassimilationare clearlyevidentin the low-
baseof the crust[e.g.,Cox, 1980].Only onepicritesample
(MgO 15 wt%) has so far been found and this is from near
the baseof the Etendekabasalts[Gibsonet al., 1997].Data
for the Paranfi-Etendeka basaltsplot on the experimentally
determined1-atm cotectic (ol+pl+cpx+liq) in a CIPW
normatire Di-O1-Hy-Qz•e diagram [Thompsonet al.,
1983], which impliesat leasta final stageof equilibration
and crystallisationat low pressuresin near-surfacemagma
reservoirs.The major-elementtrendswithin eachmagma
type are controlledprimarilyby fractionalcrystallisation
of
a predominantly gabbroic crystal extract (olivine +
clinopyroxene+ plagioclaseñ Fe-Ti oxide [Cox, 1980;
Bellieniet al., 1984a;Peate, 1990]).
Crustal Assimilation
The low-Ti basalts in the southern Paranti and the
Etendekashowsignificantvariationsin Sr-, Nd-, Pb-, and
O-isotope compositions and in highly incompatible
elementratios,indicatingthe importanceof open-system
behaviour in their evolution [Erlank et al., 1984;
Mantovani et al., 1985a; Fodor et al., 1985b;Petrini et al.,
1987; Mantovani and Hawkesworth, 1990; Peate and
Hawkesworth,1996]. An increase in 87Sr/86Sriwith
increasingSiO2 in the Gramadobasaltsis consistentwith
an assimilationand fractionalcrystallisation
(AFC) styleof
crustalcontamination,where the amountof assimilationis
linkedthermallyto the extentof crystallisation(Figure14).
Whole-rockb180 data showa similarincrease[Fodor et
al., 1985b] but thesehave been compromised by water-
rock alteration[Harris et al., 1989; Iacumin et al., 1991].
Data on mineral separatesindicatethat magmatic8180
valueswere higher(+6.3%oto +8.3%o)thantypicalmantle
SiO2 (wt%)
for Parantibasaltsand rhyolites.
Ti Gramadomagmatype whichshowsa positivecorrelation.The
broadness of the Gramadoarray is attributedto regionalvaria-
tionsin the 87Sr/86Sri
of thepre-AFCparentalmagma[Peateand
Hawkesworth,1996]. Data sourcesas for Figure7, plus Garland
et al. [1995].
values(-+5.5%o),indicativeof somecrustalinput [Harris
et al., 1989]. The contamination
processwascomplexand
cannot be represented by a single, progressively
contaminatedliquid line of descent,even on a local scale
where significant temporal variations in the rate of
contaminationare evident [Mantovani and Hawkesworth,
1990; Peate and Hawkesworth,1996].
Despite the widespread consensusfor the role of
contamination,disagreement existsover the composition
of the 'uncontaminated' parentallow-Ti magma(s).Many
studieshave suggestedthat the temporal change from
Gramadoto Esmeraldaflows simply reflectsa relatively
abruptdecreasein the extent of crustalcontaminationof a
commonparentalmagmathat had SNdi >+4 [e.g., Petrini
et al., 1987]. However, in detail, it is difficult to reconcile
all the compositionaldatawith sucha model.For example,
Gramado samplesshow a positive correlationbetween
87Sr/86SriandTh/Ta, consistent with the additionof crustal
material with high 87Sr/86Sriand Th/Ta. Esmeralda
magmas have similar Th/Ta to the less contaminated
Gramadomagmasbut at lower87Sr/86Sri. Thus,Peateand
Hawkesworth[1996] have arguedthat the Gramadoand
Esmeraldamagma types evolved from distinctparental
magmas.The parental Gramadomagma had Nb/La <0.8
and SNdi----3, and subtleregionalvariationsin 87Sr/86Sri
(0.707-0.709) prior to AFC are the principalcauseof the
broadnessof the Gramadoarray on Figure 14 [Peateand
Hawkesworth,1996]. The Esmeraldamagmatype
234 PARANA-ETENDEKA PROVINCE

1.6
significantrangesin 87Sr/86Sri
and gNdi that are reason-
ably correlatedwith incompatibleelementratios such as Tristan
plume IOPMI
La/Nd and Nb/La (e.g., Figure 15), but not with elemental
abundances. Thesevariationscan be explainedby mixing high-Ti
potassic
rocks •
betweenan incompatible-element-depleted melt with gNdi •l.0- Iow-Ti
potassic
rocks•-'•"//"1 a I•1
>+4 and a contaminatedGramado magma at relatively
Z
shallowlevels followed by fractionalcrystallisation[Peate
and Hawkesworth, 1996].
Mineral separatesfrom high-Ti Khumib (=Urubici)
0.4-
••••
•
.•///••,///
._j•l•[Low-ri
ß
•• IoNød
•Hig/ciT'
[•Esmeraldal
1
le GramadoI +Uruti
iernj
samplesin the Etendekahave •180 valuesof-•+5.9%o,
indicatingminimal crustalinteraction[Harris et al., 1989].
However, crustal assimilationdoes appearto have influ-
encedthe compositionof someUrubici magmas,despite
-20-•5-•0•-•d
i • • 10
their relatively high incompatibleelementcontents.Small Figure 15.8Ndi vs.Nb/La diagramto illustratethelow 8Ndi and
Nb/La characteristicsof the Paranti basalts, which are distinct
variationsin 87Sr/86Sri (0.7047-0.7065)arepositivelycor-
from the present-dayTristan plume and MORB compositions.
related with SiO2 (50-60 wt%) and Th/Ta (2.1-4.0), in- The positivecorrelationof 8Ndi and Nb/La for the Esmeralda
dicativeof an AFC styleof contamination (Figure 14). The magmatype indicatesthe additionof a MORB-like component.
effects of crustal contaminationappear to have been Data on Brazilianmafic potassicmagmaswhichrepresentsmall-
limited in the 'Northern' magma types (Pitanga, degreelithosphericmantlemelts are also plottedto assesstheir
potentialrole in Parantimagmatism[Gibsonet al., 1996]. The
Paranapanema,Ribeira) as they all share the same,
minor Tafelkop basaltsin Etendekaare compositionally similar
restricted, variability in Sr-Nd-Pb isotopic composition to the Tristan plume [Milner and le Roex, 1996]. Symbolsand
(Figures7 and 8), and it is notablein this contextthat these datasourcesfor Parantilavas as for Figure 7. Tristanplume data
magmatypestogethercompriseperhapshalf of the Paranti from le Roex et al. [1990] and Cliffet aL [1991]. OPM - average
eruptedmagmavolume. oceanicplumemagma[Gibsonet al., 1996].
Crystal fractionation(+ crustalassimilation),however,
is not a viable meansof explainingthe variety of basaltic plume[e.g.,Richardseta/., 1989]. The predictions of such
magmatypesin the Paranti,given the wide rangein Tie2 plume decompression modelsfit the observations of some
and incompatibletrace elementcontentsat a similarMgO floodbasaltprovincesvery well: e.g., in the Deccan,erup-
content(e.g., 1.0 to 4.0 wt% Tie2 at 5 wt% MgO: Figure tion rateswere apparentlyvery high (-1 km3 yr-]), the
7a). Thesedifferencesmusthavea deeperorigin,reflecting traceelementand isotopecharacteristics
of somerelatively
differencesin the meltingprocessand/ormultiplesources. uncontaminated basalts are similar to those of recent
volcanicsfrom the R6union plume, and they also have
MANTLE ORIGINS OF THE major elementcharacteristics(high Fe, low Si) consistent
PARAN•-ETENDEKA BASALTS with melting of fertile peridotite at depth [Peng and
Mahoney, 1995; Turner et al., 1996].
Plume Modelsfor Flood Basalt Generation However, the featuresof some flood basaltprovinces,
notably the Parant-Etendeka,are not easily reconcilable
The immensevolume of magmaassociated with a flood with modelsthat involve just decompressionalmelting of
basalt province such as the Parant-Etendekarequiresa plumemantle[e.g., Turneret al., 1996].Parantibasaltsthat
largethermalanomalywithin the mantle.This is generally are considered not to have been contaminated with
attributedto the presenceof a mantleplume [e.g., Morgan, continentalcrust have certain trace elementand isotopic
1981], especiallygiven the spatialand temporalconnec- features(i.e., low Nb/La <0.8, low •Ndi <0: Figure 15) that
tions often observedbetweenflood basaltoutcropsand the are not commonly observedin oceanicbasalts(MORB or
subsequent hotspottracesin the oceanbasins(Figure 1). In OIB) as well as major elementcompositions that require
most models, melt is generatedrelatively rapidly by de- mantle sources distinct from those of typical oceanic
compressionof plume mantle;the detailsdiffer as to how basalts[Hergt et al., 1991; Hawkesworthet al., 1992;
decompression occurs,whetherin responseto lithospheric Turnerand Hawkesworth,1995]. It is alsopossiblethat the
extension[e.g., White and McKenzie, 1989], thermal ero- overall eruption rate was significantlyless than in the
sion and thinning of the lithosphereby the plume [e.g., Deccan [Turner et al., 1994]. One major differencemight
Yuenand Fleitout, 1985], or the suddenarrival of a large be the influence of the lithosphereoverlyingthe mantle
head of plume material during the initiationof a mantle plume,andhotly debatedissuesat presentarethe

contribution of lithospheric material to flood basalt
magmatismand whetherthe lithosphericmantle can melt
to a sufficientextent to producethe observedvolumesof
flood basalt melts [e.g., Hawkesworthet al., 1988, 1992;
Ellam and Cox, 1991; Arndt and Christensen, 1992;
Saunderset al., 1992; White and McKenzie, 1995].
Plume Involvementin Parand-EtendekaMagmatism
Compositionalevidence for the involvement of the
Tristanplume in the Parant-Etendekamagmatismis mini-
mal. The only Parant-Etendekamagmasthat have similar
incompatible-trace-element and isotopiccharacteristics to
the present-dayTristan plume, as representedby lavas
from the islandsof the Tristanda Cunhagroupand Gough,
are the Tafelkop basalts and the Okenyenya alkaline
intrusions,both in the Etendeka(Figure 6b) [Milner and le
Roex, 1996]. The relatively minor Tafelkop basaltsare
found at the baseof the lava sequences in a restrictedarea
north of Messum, and they have high Ti/Y (> 500) and
Nb/La (-1.0), and Sr-Nd-Pb isotopecompositionsthat fall
within the field for the Tristanplume [Milner and le Roex,
1996]. A few tholeiitic basaltswith similarNb/La (-1.0) to
the Tafelkop basaltshave recentlybeen found in southern
UruguaynearMariscala(Figure6a) [Kirsteinet al., 1997].
The compositionof the Tristan plume has thus remained
broadlyconstantfor the last-135 m.y., and the suggestion
of Classet al. [ 1993] that differencesin isotopiccomposi-
tionsbetweenthe older Walvis Ridgebasaltsandthe recent
Tristanmagmatismresultprimarily from radioactivedecay
in the plume sourcecan be ruled out. However, it remains
difficult to distinguishthe Tristanplume signaturein pro-
posedmixing arraysinvolvingthe Parant-Etendekalavas.
Hawkesworth et al. [1992] describedsome late-stage
dykesalongthe Brazilian coastbetweenS•.oPauloand Rio
de Janeirothat also had similar trace elementand isotope
characteristicsto the modem Tristan plume. However,
recent work has shown that these dykes, called the S•.o
Sebasti•.odolerites, are in fact much younger than the
Parant-Etendekaevent, with a 40Ar-39Arage of 81 Ma
[Regelous,1993].
The only Paranti basaltsthat unequivocallycontain a
recognisablyasthenospheric componentare the late-stage
Esmeralda magmas, and yet they represent a
volumetricallyinsignificantportionof the total flood basalt
magmatism [Peate and Hawkesworth, 1996]. Esmeralda
magmasshowsignificantpositivecorrelationsof l•Ndiwith
Nb/La (Figure 15) and Sm/Ce. They indicatemixing with
an incompatible-element-depleted componentwith a trace
element signaturesimilar to MORB (Ce/SmN <1, Nb/La
>0.9) and high l•Ndi >+4 [Peateand Hawkesworth,1996].
PEATE 235
This componentthough,is unlike the compositionof the
presentTristanplume as it hashigherl•Ndi (Figure 15) and
lower208Pb/204pb
(Figure10).
LithosphericInvolvementin Parand-EtendekaMagmatism
An intriguingfeature of the main basalticvolcanismis
that high-Ti and low-Ti magmatypes form distinctgeo-
graphical regions in the northwestand southeastof the
province, respectively, which must represent either
variabledegreesof melting beneaththe provinceor spatial
heterogeneitieswithin the lithosphere. Fodor [1987]
arguedthat both types could be generatedfrom a common
mantle sourceif the degreeof melting was controlledby
proximity to the underlying Tristan mantle plume, with
high-degreelow-Ti meltsgeneratedover the plume axis in
the southand low-degreehigh-Ti melts producedon the
plume peripheryto the north. Arndt et al. [1993] instead
suggested that lithosphericthicknesswas a more important
factor because it determined the extent to which the
underlying asthenospherecould decompressand melt.
Beneatha thick lithosphericcap, melting is restrictedand
occurs in the presence of residual garnet, leading to
incompatible-element-enriched basaltswith high Ti/Y and
Tb/Yb similar to high-Ti basalts(Figures 7 and 12). For
thinnerlithosphere,melting shouldbe more extensiveand
occur within the spinel peridotite field, producingbasalts
with lower concentrationsof incompatibleelementsand
with low Ti/Y and Tb/Yb, similar to 1ow-Ti basalts.Any
incompatible-trace-element or isotopicdifferencesbetween
the high-Ti and low-Ti types or relative to typical
asthensopheric melts were simply dismissedby Fodor
[1987] and Arndt et al. [1993] as crustal contamination
effects. However, it is difficult to explain the isotopic
variations of the Parant lavas in this manner. If the Parant
mantle sourceis assumedto have a similarcompositionto
the high l•Ndi componentseen in the Esmeraldamagma
type, then the trace element and isotope differences
betweenthe 1ow-Ti Gramadomagmatype and the high-Ti
magma types would require the coincidenceof different
degreesof melting with assimilationof crust of different
composition. Furthermore, the simultaneouseruption of
high-Ti Urubici (Tb/YbN-2.3) and low-Ti Gramado
(Tb/YbN-1.3) lavason the SerraGeral escarpmentwould
imply significantvariationsin lithosphericthicknessover a
relativelysmalldistanceaccordingto the modelof Arndt et
al. [1993].
The 'Northern' magma types (Pitanga, Paranapanema,
Ribeira) show good evidencefor a gradualincreasein the
degree of melting of a similar mantle sourceat progres-
sively shallower depths. Paranapanemalavas,
236 PARANA-ETENDEKA PROVINCE
locally overlie Pitanga lavas acrossmost of the northwest
Paran/t, have lower incompatible element contents and
lower Ti/Y and Tb/Yb values, but have a similar Sr-Nd-Pb
isotopic composition [Peate, 1990]. Major element
differences(lower FEB.0and higher Sis.0: fractionation-
corrected to MgO 8.0 wt%) also suggest that the
Paranapanemamagmaswere generatedat lower pressures
thanthe Pitangamagmas[Garland et al., 1996]. Gallagher
and Hawkesworth [1994] argued that these lavas were
erupted through thick lithospherebecauseseismic data
indicate that the lithospheric thickness beneath the
northeasternpartsof the Paran/tprovinceis presently150-
200 km, although it is uncertain to what extent the
lithospheremight have rethickenedsincethe Paran/tevent
by conductivecooling. Extensionacrossthe Ponta Grossa
structurewas probablylessthan 20% [Ussamiet al. 1991].
Thus there is little evidence that significant extension
occurredin this region that couldhave producedextensive
decompressionalmelting of asthenosphericmantle. It
might be assumedfrom the large volume and similar
compositional characteristicsof the 'Northern' magma
types that they represent the best estimate for the
compositionof the dominantcomponentin the early stages
of the Tristan plume, and yet this component was
apparentlynot sampledsubsequently by plumemagmatism
after continentalbreakup. For example, Nb/La remains
constantat-0.64 in all the 'Northern'magmatypes,a value
which is lower than in MORB or Tristanplume magmas
(Figure 15). Instead, it seems that a source within the
lithosphericmantleis necessary.
Similar high-Ti and 1ow-Ti domains can be distin-
guished in the other Mesozoic Gondwananflood basalt
provinces(Karoo, Ferrar). A boundarybetweenthe two
types can be traced across the former Gondwanan
supercontinent andhasled to speculation that it represents
a majorcompositional discontinuitywithinthe lithospheric
mantle [e.g., Cox, 1988; Erlank et al., 1988; Hawkesworth
et al., 1988; Hergt et al., 1991]. Supportingevidencefor
such a lithosphericmantle boundarycomes from mafic
potassicrocks erupted around the marginsof the Paran/t
lava field duringthe Cretaceous.Thesecanbe dividedinto
two compositionallyand geographicallydistinctgroups:a
high-Ti groupfound alongthe northeastmarginand a low-
Ti groupfound adjacentto the centralParan/tlavasalong
the coastin the eastand in Paraguayto the west [Gibsonet
al., 1996]. Such magmas are widely accepted as
representingsmall-degreemeltsof the lithosphericmantle
and so Gibson et al. [1996] concludedthat the spatial
distributionof the high- and 1ow-Ti mafic potassicrocks
was analogousto the broad high-Ti/low-Ti provinciality
seen in the Paran/t lavas and reflected a major
compositionalboundary in the lithosphericmantle. They
modelled the low- and high-Ti flood basaltsas large-
degreeasthenospheric melts that had mixed with between
20% and 50% of the low- and high-Ti mafic, potassic,
lithosphere-derived melts, respectively, and then
undergone variable degrees of crustal assimilation.
However, there are severalproblemsfor sucha model. To
explain the observedrange in Ti/Y (300-700) within the
high-Ti 'Northern'magmatypes(Figure7b) wouldrequire
the additionof between10% and50% of the high-Ti mafic
potassicmelt, and yet thesemagmatypeshave relatively
homogeneous Sr-Nd-Pb isotopiccompositions that show
no systematicvariationswith Ti/Y. Furthermore,most of
the high-Ti mafic potassicrockshaveunsuitablecomposi-
tions (higher Nb/La, similar l•Ndi: Figure 15) to account
for the high-Ti flood basaltsby mixing with either MORB
or a plume-derivedmelt. Ti/Y is too high (>400) in the
low-Ti mafic potassicrocksfor themto representa suitable
low-degree melt contaminant in the low-Ti Gramado
basaltsand, in addition,the alreadylow TiO2 contentsof
primitive Gramado basalts would require any
asthenosphericmelt to have an extremely Ti-depleted
composition.Ellam and Cox [ 1991] had similardifficulties
in extendingtheir lamproite-mixingmodel for the high-Ti
Nuanetsipicritesto the low-Ti magmatypeselsewherein
the Karoo provinceof southernAfrica.
Instead,Hergt et al. [ 1991] arguedthat the distinctive
compositionalfeatures of the Gondwana low-Ti flood
basaltswere inheritedfrom an unusual,melt-depleted man-
tle sourcethat had been modified by the additionof sub-
ductedsedimentmaterial.The majorelementcompositions
of the Gramadomagmasindicatean origin from a mantle
sourcedistinctfrom that of oceanicbasalts[Hergt et al.,
1991; Turner and Hawkesworth,1995]. FeB. 0 contentsof
peridotite melts depend both on pressureand source
composition. Although Gramado samples with low
87Sr/86Sr
i (0.707-0.709) have similar FEB.0(-10.2) to
recentSouthAtlantic MORB, it is unlikely that both were
generatedat similarly shallow depths.Thus, Hergt et al.
[ 1991] and Turner and Hawkesworth [ 1995] concluded
that the Gramado magmas originated in an Fe-depleted
source,presumablydue to previousmelt extractionevents,
arguedto be in the continentalmantlelithosphere.
Uncontaminated Urubici and Khumib magmas are
characterised
by low 206Pb/204Pb
(17.4-17.8)coupledwith
high 207Pb/204Pband 208Pb/204pb,
typicalfeaturesof the
oceanic DUPAL isotopic province [Hawkesworthet al.
1986; Peate, 1990; Milner and le Roex, 1996]. Similar Pb
isotope characteristics,and broadly similar Sr and Nd
isotope compositions, are seen in the nearby
contemporaneous Jacupiranga carbonatites

pyroxenites,and oxygenisotopedata confirmthat they are
not the resultof crustalcontamination[Toyodaet al., 1994;
Huang et al., 1995]. Thus,by inference,it appearsthat the
low 206Pb/204pb
characteristic
of the Urubicimagmasis a
mantle feature. Hawkesworth et al. [1986] showed that
there was a close compositionallink betweenthe Urubici
magmasand 70 Ma samplesfrom DSDP Site 525A on the
WalvisRidgethathaveNb/La-0.8, 206pb/204pb
-17.6 and
•Ndi--3 [Richardsonet al., 1982]. It is possiblethat this
low 206Pb/204pb
materialis an intrinsicpart of the Tristan
plume that is not presentlybeing sampledby the active
plume magmatism on Tristan da Cunha or Gough.
However, if the Urubici basaltsare derived from the litho-
sphericmantle, then the presenceof basaltswith similar
trace element and isotopic features erupted within the
oceanic part of the plume track sixty million years later
might result from material having been thermally eroded
from the base of the lithosphericmantle and entrainedin
the asthenosphere[Hawkesworthet al., 1986; Milner and
le Roex, 1996]. Mahoney et al. [1992] proposeda similar
modelto explainlow 206pb/204Pb rocksin the Madagascar
province and along part of the SouthwestIndian Ridge
(39ø-41øE) that lies 400 km from the Marion plume.
LithosphericMantle Melting Model
Althoughgeochemicalevidencefor the participationof
lithosphericmantle material in someflood basaltmagmas
is now widely accepted,the extentof this involvementhas
yet to be agreed[e.g., Hawkesworthet al., 1988; Ellam and
Cox, 1991; Mahoney et al., 1992; Saunderset al., 1992;
Whiteand McKenzie, 1995]. Plume decompression models
for flood basalt magmatism predict minimal melt
generation(<<5%) within the lithosphericmantle,andthen
only in the initial stages. These models might be
inappropriatefor some flood basaltprovincesbecausethe
models assume rapid eruption rates [e.g., White and
McKenzie, 1989] and an anhydrous solidus for the
lithosphericmantle [Arndt and Christensen,1992]. Melt
generationwithin the lithosphericmantlewill be facilitated
by the presenceof just 0.3 wt% H20+CO2, which allows
melting at the lower temperaturevolatile-enrichedsolidus
[Gallagher and Hawkesworth,1992]. Turner et al. [ 1996]
developeda conductiveheating model in which melting
can take place solely within the lithosphericmantle,driven
by conductionof heat from a mantle plume incubating
beneatha lithosphericcap [cf. Saunderset al., 1992]. The
important controlling variables are the plume potential
temperature,the thickness of the mechanical boundary
layer, and the durationof heating.Becausethe time scales
of conductiveheat transferare long, this model predictsa
PEATE 237
protractedperiod of magmatism.On a time scale of-10
m.y., melting can occurwithin the lithosphericmantleover
a broad area (comparableto the size of the province)to
produce1-3 km of magma,without any melt contribution
from the underlyingplume,providedthatthe lithosphereis
>100 km thick. If the lithosphericthicknessis reducedto
<100 km by extension or thermal erosion, then
decompressional meltingwill commencewithin the plume,
leading to higher eruption rates and the eruption of
magmasdominatedby sub-lithospheric melts.Turneret al.
[1994] discussed the evidencethat, duringthe onsetof the
Paranfi-Etendeka magmatism,the amountof extensionwas
too low to permit decompressionalmelting of the
asthenosphere as the lithospherewas probablyat least 150
km thick. From the geochemicalargumentsmarshalled
above, it is clear that plume mantle similar to that
supplyingrecentTristan da Cunhaor Goughmagmasor
MORB asthenospherehad a minimal role in Paranti-
Etendekamagmatism.Thus,the conductiveheatingmodel
can account for many of the observed distinctive
characteristics of the Paranfi-Etendeka flood basalts. It is
only in the late stagesof Parantimagmatism,duringerup-
tion of the Esmeraldamagmatype, that lithosphericexten-
sionproceededto suchan extentto allow decompressional
melting of the asthenosphere.Figure 16 presentsa cartoon
summaryof the progressiveevolutionof basalticmagma-
tism within the Parantiprovinceas viewed by Garland et
al. [1996].
In the Deccanprovince,largevolumesof basaltsappear
to have been derived by decompressional melting of as-
thenosphereassociatedwith the R6union mantle plume
[e.g., Peng and Mahoney, 1995], whereas in the Paranti-
Etendeka province, the Tristan plume appearsto have
playeda largelypassiverole, solelyprovidingheatby con-
duction to produce extensive melting of lithospheric
mantle [Turner et al., 1996].
RHYOLITE MAGMA TYPES AND PETROGENESIS
The Paranfi-Etendekarhyolites cover an area of at least
17,000kin2 andtypicallyform the uppermostunitsof the
lava sequences.They can be divided into two distinct
groupson the basisof petrographyand geochemicalcom-
position(Figures 14 and 17), similarto the high-Ti/low-Ti
division of the basalts[Bellieni et al., 1986; Harris et al.,
1990]. The high-Ti rhyolites(called Chapec6rhyolitesin
the Paranti)are highly plagioclase-phyric,
with high incom-
patibleelementabundances(e.g.,Zr >500 ppm),low •180
(-+6.5%oin pyroxene)andsimilar87Sr/86Srito thehigh-Ti
basalts (0.705-0.708). In contrast, the volumetrically
dominantlow-Ti rhyolites(called Palmasrhyolitesin
238 PARANA-ETENDEKA PROVINCE

were linked to the basalts either by extensive fractional

INorthwestMigration
ofmagmatism
• Southeast
crystallisationand crustal assimilationor by melting of
Pitanga/Paranapanema Gramado Esmeralda
underplatedbasalts [Bellieni et al., 1986; Garland et al.,
138-133Ma 135-131Ma 129Ma
',,',,',,'•,'•,'•,'•,'•,L,',,L,•,•,•'•'•'•'•'•'•'•'•'•'•'•'•'• 1995].
There are some minor outcrops of rhyolites near
I Mariscala in southernUruguay (Figure 6a). Recent work
:: . ß :::::::::::. •g• ==========================
....................... . . by Kirstein et al. [ 1997] has shownthat they are composi-
tionally distinct from the Palmasand Chapec6rhyolites,
and are interpretedas lower crustalmelts. They have been

8OO
.......
':•:••"•.•.•..••.....`..``..•...``•.•.•.••••••:••••••••:•••:•:••••:••:•...••••:•:•:;•:•:•:•;•:
0:•;m;:• a • [] Chapec6 (high-TO
[] Guara•puava
ß Ourinhos
•600 Palmas (low-TO
ß Santa Maria
Figure 16. Cartoonsectionillustratinga model for the temporal ß Anita Garibaldi
evolutionof magmagenerationwithin the Paranti[Garland et al., o o Caxias do Sul
1996]. The oldest basalt unit (Pitanga) was derived from the N 400
greatestdepth in the lithosphere(90-120 km), melting relatively
fertile, anhydrous peridotite. The magma source became
gradually shallowerwith time, with the Gramado basaltunit re- o%
flectingshallow-levelmagmageneration(< 60 km), meltingrel-
200
atively refractoryand hydrousperidotite.Asthenospheric mate-
rial only becomes clearly recognisablein the youngest unit
(Esmeralda),once extensionhas proceededto the extent that de- 50 160 1•)0 260 2•0 300
compressionof the asthenosphere can take place. Rb (ppm)
125

Paranti)are virtually aphyric(<5% phenocrysts)with low Caxias

doSul • ß
incompatibleelementcontents(e.g., Zr <400 ppm), high 100- oTafelberg• ©
o Awahab
ß©
ee
•80 (•+10%oin pyroxene)andhigh 87Sr/86Sri
of 0.714-
o
0.727. Analysesof selectedParan• rhyolitesare listed in
Table 2. Calculatederuptiontemperaturesfor both types .e•ß•e•$
are unusually high for silicic magmas (950-1100øC
[Bellieni et al., 1986; Garland et al., 1995]).
The rhyolitesare restrictedto the later stagesof magma- 50- •O "•-
ß,•,,,•o •
(O • •

tism and, unlike the basaltswhich crop out almostexclu-

sively on the SouthAmerican Plate, the rhyolitesare con- 25-
centratedalongthe present-daycontinentalmargins,which
ß
indicatesa closelink with the rifling of the SouthAtlantic
i i i
Ocean.This fact, togetherwith the pronounced'silicagap' 0.50 0.75 1.00 1.25 1.50
betweenrhyolitesand basalts(Figure 4), led many workers
to assumethat the rhyoliteswere the productsof crustal TiO2 (wt%)
anatexis [e.g., Erlank et al., 1984; Hawkesworthet al.,
1988]. Harris et al. [1990] linked the regionaldistribution Figure 17. (a) Rb vs. Zr: 400 ppm Zr discriminatesbetween
high-Ti Chapec6 and low-Ti Palmasrhyolite types, and the
of the high- and 1ow-Ti rhyolites(Figure 18) to differences Chapec6subgroupsare resolvedby differentRb/Zr (Ourinhos
in lower crustalsourceregions,with the high-Ti rhyolites >0.2; Guarapuava<0.2), (b) TiO2 vs. Cu, illustratingcomposi-
generatedfrom Archaeanlower crustand the 1ow-Tirhyo- tional differencesbetweenthe low-Ti rhyolite subgroups. The
lites derived from melting Late Proterozoicmobile belt Caxias do Sul subgroupis divided into the Tafelberg and
material. However, the coincidenceof basalt and rhyolite Awahab magmasystemswhich lie, respectively,aboveand be-
low an erosionaldisconformity
in thefield [Milneret al., 1995b].
provinciality (1ow-Ti Gramado basalts and Palmas Data from Hawkesworthet al. [1988], Peate [1990], Mantovant
rhyolitesin the south;high-Ti PitangabasaltsandChapec6 and Hawkesworth[ 1990], Whittingharn[1991], Garland et al.
rhyolitesin the north) led to suggestions that the rhyolites [ 1995],andS.C. Milner andA.R. Duncan(unpubl.
 PEATE 239

measureda 40Ar-39Arageof 132.0+0.4Ma for the Giraul

Paran•. rhyolites .......
23
rhyolites,which is similarto 40Ar-39Aragesobtainedby
Palmas
sub-groups Grossa Turner et al. [1994] for the Guarapuava(131.8+1.8 Ma)
• Santa
Mafia swarm..'.24 andOurinhos(131.7+0.8 Ma and 128.7+1.1Ma) rhyolites.
1•] Anita Gafibald! Palaeomagneticdata, though, indicate a differencein the
15ffiCaxiasdo Sul relative ages of the Ourinhos (reversedpolarity) and
Chapec6 ...... 25 Guarapuava(normal polarity) subgroups[Ernesto and
sub-groups . . .

Pacca, 1988].
/ Oufinhos
/ Guarapuava 26 om
Petrogenesisof the High-Ti Rhyolites
.,•

27 •
Althoughlower crustalgranuliteshave been suggested
:: :: as a potential sourcematerial for the high-Ti rhyolites
ß ßbasalts [Bellieni et al., 1986; Harris et al., 1990], the closecoinci-
'28 dencebetweenmanycompositional featuresof the high-Ti
basaltsand Chapec6rhyolitesmake the basaltsa plausible
parental material. Garland et al. [1995] showedfrom a
29
South
comparison of traceelementandisotopecompositions that,
of the threehigh-Ti basaltmagmatypes,only the Pitanga
Atlantic
basement
'30
basaltsare a suitableparentto the Chapec6rhyolites(e.g.,
55 54 53 52 51 50 49 48
Pitangabasaltsand Chapec6rhyoliteshavesimilarTb/Yb:
Longitude(øW) Figure 12). The Chapec6rhyolitescan be modelledsatis-
factorily either by extensive fractional crystallisation
Figure 18. Distributionof rhyolitesubgroups withinthe Paranti (-70%) of a parentalPitangamagmaor by about30% par-
province[Peateet al., 1992;Garlandet al., 1995]. tial meltingof a Pitangabasaltsource[Bellieniet al., 1986;
Garland et al., 1995], but the existenceof significantgaps
datedat-128 Ma by 40Ar-39Ar andtheyalsopreserve clas- in both SiO2 and incompatibleelementcontentsbetween
sicignimbritictextures[Kirsteinet al., 1997]. the Pitanga basaltsand Chapec6rhyolites(Figure 19)
argues strongly in favour of a partial melting origin.
High-TiRhyoliteMagmaTypes Melting of underplatedbasalt,whichpondedat the baseof
the crustbecauseof the densitycontrastat the Moho [e.g.,
Peateet al. [1992] dividedthe Chapec6rhyolitesinto Cox, 1980], probablyoccursduringthe final stagesof con-
the Ourinhossubgroup (Rb/Zr >0.2, 87Sr/86Sri=0.7076- tinentalbreakupas the crustthins and decompresses. A
0.7080) and Guarapuava subgroup (Rb/Zr <0.2, similarmodelhasbeenproposed to explainthevoluminous
87Sr/86Sri-0.7055-0.7060).
TheOurinhos
subgroup
forms high-TiLebomborhyolitesin theKarooprovince[Cleverly
a limitedoutcropin the northeast
of the Parangt,
whereas et al., 1984].
the Guarapuava subgroup is moreextensive, forminga The Ourinhossubgrouphas higher878r/86Sri(0.7078
seriesof separateoutcrops alongtheeastern marginof the vs. 0.7055) and SiO2 and lower TiO2 than the more volu-
centralParangtlava field (Figure18). Smallpetrographic minous Guarapuavasubgroup.These differencescan be
differencesamongthe Guarapuava outcrops suggestthat explainedby a localisedAFC processstartingwith a
theyrepresent differentflow units.In the Africanpartof Guarapuavamagma, involving -10% upper crustal as-
theprovince,high-Tirhyolites havebeenreported fromthe similationcoupledwith-20% fractionationof an assem-
northernmost partof theEtendeka province(Samsas rhyo- blageprimarilyof plagioclase andclinopyroxene plusFe-
lites[Milner,1988])andalongthesouthern Angolancoast Ti oxides,apatite and zircon [Garland et al., 1995].
(Giraulrhyolites[Albertiet al., 1992]).Althoughthepub- Normativecompositions indicatethat Chapec6rhyolite
lishedgeochemicaldata are limited,both Ourinhos-and magmaslastequilibratedat pressures of 5-15 kb [Garland
Guarapuava-like compositionsare found within the et al., 1995]. Thus it appearsthat the magmaswere ex-
Samsusrhyolites [Harris et al., 1990], but the Giraul tractedrapidly from lower crustaldepths,withoutfurther
rhyolitesmay representa third subgroupas they have significantdifferentiation
of themagma(exceptlocallyfor
87Sr/86Sriof 0.7062-0.7074, intermediate
betweenthe the Ourinhossubgroup). The Chapec6rhyolitespreserve
Ourinhos andGuarapuava subgroups.Renneet al. [1996b] no texturalevidenceof pyroclastic activity,andeach
240 PARANA-ETENDEKA PROVINCE

75
Palmas
been correlated throughout the Etendeka. Whittingham
70-
[ 1991] did a similar studyof the Caxiasdo Sul subgroupin
Chapec• southeastBrazil, distinguishingseven distinct units, and
• ' .
showedthat stratigraphiccompositionalchangesmirrored
o•65-
thoseseenin the Etendekarhyolites.Milner et al. [ 1995b]

•-•-•
60-
G
ca

.(•-55
IHigh-Ti
-"-
'silicagap'
50 I High-TiI r::•tøw'mil
45
Pitanga
,
, 'z•, [• High-m•
o traced acrossinto the southernParantiand thus provides
Zr (ppm)
Figure 19. Zr vs. SiO2 diagramhighlightingthe differentpet-
rogeneticoriginsof the high- and low-Ti rhyolitcs.Data sources
as in Figure 16.
unit has a limited spatialextent,being <40 km long and
<100 m thick [Bellieni et al., 1986]. A few dykes of
Chapec6compositionare found within the Ponta Grossa
dyke swarm (Figure 18), and so Garland et al. [1995]
suggestedthat the Chapec6rhyolitesare dyke-fed lava
flows.
Low-Ti RhyoliteMagma Types
The Palmas rhyolites are divided into three composi-
tional subgroups(Figure 17b) [Peateet al., 1992; Garland
et al., 1995]. The Caxias do Sul subgroupis the largest
volumetricallyand is found in the southeastof the lava
field near the SerraGeral escarpmentwherethe flow units
reach a combinedthicknessof up to 400 m (Figure 18).
The Anita Garibaldi subgrouplies to the north of the
Caxiasdo Sul subgroup,along the easternmargin of the
central Paranti lava field, and field evidence showsthat it is
youngerthanthe Caxiasdo Sul subgroup. The SantaMaria
subgroup is found in two widely separatedexposures:
alongthe southernescarpment to the westof the Caxiasdo
Sul subgroupwhich it overlies,and in the centralParanti
where it capsthe highestexposures and is thusinferredto
be youngerthanthe Anita Garibaldisubgroupaswell.
The excellentexposurein the EtendekaallowedMilner
et al. [ 1992] to look at the low-Ti rhyoliteunitson a flow-
by-flow basis. Individual units are remarkablyhomoge-
neousand showno compositionalzonation.Most are simi-
lar in compositionto the Caxiasdo Sul subgroup,with the
exceptionsof the uppermostunit (similar to the Anita
Garibaldi subgroup)and some units to the north in the
Sarusasregion (similar to the Santa Maria subgroup)
[Milner, 1988]. Each unit or group of successiveunits is
geochemicallydistinctiveon the basisof petrographyand
elementabundances(especiallyTi, Fe and Cu), and have
have demonstrated a direct correlation of some individual
units between these two areas that are now on opposite
sides of the South Atlantic Ocean.
The upper and lower parts of the Etendekasequences
are separatedby an erosionaldisconformitythat can be
anotherreferencepoint for reconstructions of this part of
Gondwana.Rhyolitesbelow this disconformityhave simi-
lar petrographyand compositions and are only foundin the
southof both areas.Small, stratigraphically controlleddif-
ferencesbetweenunits suggestthat the units sequentially
tappeda single, evolving magmabody, referredto as the
Awahab magma system[Milner et al., 1995b], and they all
appear to have been erupted from the Messum structure
(Figure 6b), where compositionallysimilar quartz-mon-
zonite intrusionsare found [Milner et al., 1992]. The rhyo-
lites overlying the disconformity apparently tapped a
different magma body, the Tafelberg magma system,
whoseeruptionsite is unknown.One of theseunitsreaches
its maximum thickness close to the Namibian coast,
suggestinga sourcenow offshore. The Awahab magma
system comprisesat least five individual eruptive units,
with an estimatedtotal volumeof >8,600 km3, dispersed
over an area of >33,000 km2, and the Tafelbergmagma
systemis similar in terms of numberof eruptedunits and
magma volume [Milner et al., 1995b]. Units of the
Awahab magma system, which stratigraphicallyare the
oldestlow-Ti rhyolites,give 40Ar-39Aragesof 131.9+0.6
Ma to 132.1+0.4 Ma [Renneet al., 1996b], and unitsof the
overlyingTafelbergmagmasystemgive 40Ar-39Aragesof
132.8+1.1 Ma [Renne et al., 1992] and 130.3+ 1.2 Ma
[Stewartet al., 1996].
Individual silicic units can have volumes >1,000 km3,
unusuallyhigh aspectratios(1:200-1:2,000) and can have
travelled lateral distances>300 km [Milner et al., 1992,
1995b]. As with similar examples from the geological
record, the eruptive mechanismof theseunits has proven
controversial[e.g., Henry and Wolff, 1992]. Whittingham
[ 1991] consideredthem to be extensivelava flows, mainly
dueto the lack of texturalevidencefor an explosiveor ign-
imbritic origin. Bellieni et al. [1986] and Milner et al.
[1992], on the other hand, interpretedthem as rheoign-
imbrites,in which high temperaturesmaintainedwithin an
ignimbriteflow causereweldingof particlesand a final pe-
riod of flow that producesmany lava-like features.Milner
et al. [1992] found rare pyroclastictexturespreserved
 PEATE 241

flow margins,andGarlandet al. [ 1995] founda thin, distal PARANA-ETENDEKA FLOOD BASALTS
pyroclastic fall deposit related to the Santa Maria sub- AND MASS EXTINCTIONS ?
group. However, Manley [1995] issueda note of caution
for suchan interpretation
by suggestingthat certainpyro- The possibility of a causal link between flood basalt
clastictexturescanbe formedlocallywithin a lavaflow. eruptionsand major faunal extinctionshas been much de-
batedin the literature[e.g., Rampinoet al., 1988]. The
Permian-Triassicand the Cretaceous-Tertiary boundaries
Petrogenesisof the Low-Ti Rhyolites
mark two of the most significantmassextinctioneventsin
Earthhistory.Renneet al. [ 1995] demonstrated that erup-
If the Palmasrhyolitesare crustalmelts, the basement tion of the Siberianflood basaltprovincecoincided,within
rocks exposedin southernBrazil cannot representa uncertainty,with the Permian-Triassicboundary.There is
suitablesourcematerial becausethey have lower Nd alsoa closetemporalrelationshipbetweenthe Cretaceous-
isotoperatiosthan the rhyolites(Proterozoicmobilebelt Tertiary boundaryand the Deccanflood basalts,although
t•Nd130Ma = -8 to -19; Archaean
cratonmaterialt•Nd130Ma
= whetherthey were contemporaneous is still debated[e.g.,
-21 to -32; Palmasrhyolitest•Nd130Ma = -6 to -8). The Venkatesanet al., 1993; F•raud and Courtillot, 1994].
relatively radiogenicSr and Pb isotopiccomposition Rampino et al. [1988] correlated the eruption of the
(87Sr/86Sri=0.714
to 0.728; 206Pb/204Pb
>19.0) of the Paran•t-Etendeka flood basalts to the mass extinction event
rhyoliteswouldsuggest a relativelyincompatible-element- at the Jurassic-Cretaceous
boundary.This extinctionevent
enriched,uppercrustalsource,but their high eruption at the end of the Tithonianis one of the eightmassextinc-
temperatures (>1000øC)and major elementcomposition tion eventsrecognisedby Raup and Sepkoski[ 1984], when
requirea fairlybasicsourceakinto lowercrustalgranulites 37% of marine genera(mostly ammonites,bivalves,and
or basalts. corals)disappeared.
Severalworkershaveinsteadinvestigated
possible
frac- Most estimatesplacethe Jurassic-Cretaceous boundary
tionation links between the Gramado basalts and the at between135 Ma and 145 Ma [e.g., Renneet al., 1992],
Palmasrhyolites[Bellieniet al., 1986; Garlandet al., which is older than the 40Ar-39Aragesof most of the
1995].Any crystalfractionation mustbe accompanied by Paran•t-Etendeka lavas. Therefore, any temporal
assimilationof uppercrustal materialbecause therhyolites correlation between flood basalt eruption and mass
continuethetrendtowardshigh87Sr/86Sr, 206Pb/204Pb and extinction is, in this case, unlikely [Hawkesworthet al.,
5180 seenwithin the basalts(e.g., Figure14), and 1992; Renne et al., 1992], althoughbetter constraintson
modellingof anAFC process canproduce a goodfit to the the age of the Jurassic-Cretaceous boundaryare clearly
major and traceelementdata.The entry of significant required.It is not possibleto locatepreciselythe Jurassic-
amountsof magnetiteinto the calculatedcrystallising Cretaceousboundarywithin the Paran•tbasin sequences
assemblage at-56% SiO2 rapidlyenhances the SiO2 becauseof the essentiallyunfossiliferousnature of the
contentin the liquid over a small extent of fractionation local sediments.One solutionmight be to look at more
andcoolingandcanaccount fortheobserved silicagap.It distantandbetterdatedsequences aroundthe peripheryof
is notablethat,unlikethe Chapec6-Pitanga case,highly the Paran•t basin, perhaps in northwesternArgentina,
incompatible elements suchasZr showa nearlycontinuous where distal tephra depositsfrom the large rhyolitic
trendthroughthe Gramadobasaltsto Palmasrhyolites, eruptionsin the Etendekaand southernBrazil might be
consistent with a liquid line of descent(Figure 19). foundthatcouldprovidesuitablemarkerhorizonsdenoting
Normativecompositions suggest thatthePalmasrhyolites the time of at least part of the Paranft-Etendeka
equilibratedat lowerpressures (<5 kb) thanthe Chapec6 magmatism.
rhyolites,indicativeof shallowponding[Garlandet al., Rampinoet al. [ 1988] suggestedthat the environmental
1995]. Compositional differences between the different impactof large flood basalteventscouldbe significant.
Palmassubgroups can be explainedby regionaland Individualflows can have eruptedvolumesof at least700
temporal variationsin the extent of assimilationand exact km3 (Rosaflow, ColumbiaRiver province),whichwould
natureof thefractionating crystalassemblage.
ThePalmas
rhyolitesappearto have evolvedfrom a parentalbasalt
release
large
quantities
ofs•lhurdioxide
and
carbon
diox-
ideintotheatmosphere.
( TheParan•t
eruptions
wereof
compositionby fractional crystallisationand crustal similarmagnitude,in term• of the total volumeof magma
assimilation
in a stable,shallow-levelmagmachamber,
and erupted,to the Deccanand Siberianlava sequences. If, in-
werethenemplaced explosively at thesurface
asrheoign- deed,the flood basalteruptionsin the Deccanand Siberia
imbrites[Garlandet al., 1995]. were responsiblefor significantfaunalextinctions,
242 PARANA-ETENDEKA PROVINCE
in response to climaticchanges causedby the repeated mineralogical and petrochemical
additionof volcanogenic gasesto the atmosphere overa 944, 1986.
relativelyshorttime span(<1 m.y.), thenthe apparently Burke,K., and J. F. Dewey,Plume-generated
key indicatorsin applyingplate tectonicsto old rocks,d.
mutedenvironmental effectof the Parantieruptions
might Geol., 81,406-433, 1973.
reflecta sloweraverageeruptionrate,thusgivingthebio- Chang,H. K., R. O. Kowsmann, A.M. F. Figueiredo, andA. A.
spheremoretimeto recoverbetweensuccessive eruptions. Bender,Tectonicsandstratigraphy of the eastBrazilrift sys-
Acknowledgments.I thank Chris Hawkesworth and Marta
Mantovani for initiating my interest in the Paranfi-Etendeka
regionand for their guidanceand supportoverthe lastdecade.
Anton le Roex and Asish Basu providedconstructive reviews,
and I thank John Mahoney for his editorial commentsand ex-
tremepatience.Andy DuncanandSimonMilner kindlyprovided
unpublisheddatafrom the Etendeka.Discussionson flood basalts
with ChrisHawkesworth, Marta Mantovani,Nick Rogers,Janet
Hergt, Simon Turner, FrancesGarland, Sally Gibson,Mukul
Sharma,Kerry Gallagher,LindaKirsteinand MarcelRegelous
havebeenstimulating.This reviewwas writtenwhile I wasa re-
searchfellow in the Institutefor Environmental
Physicsat the
Universityof Heidelberg,Germany,fundedby the Heidelberg
Academy of Sciences,and I would like to thank Augusto
Manginiandthe groupfor theirhospitality.
REFERENCES
Alberti, A., E. M. Piccirillo, G. Bellieni, L. Civetta,P. Comin-
Chiaramonti,and E. A. A. Morais, Mesozoic acid volcanics
from SouthernAngola:petrology,Sr-Nd isotopecharacter-
istics, and correlation with the acid stratoid volcanic suitesof
the Paran•tbasin(south-eastern
Brazil), Eur. d. Mineral., 4,
597-604, 1992.
Arndt, N. T., and U. C. Christensen,
The role of lithospheric
mantle in the generation of continental flood basalts,d.
Geophys.Res.,97, 10967-10981, 1992.
Arndt, N. T., G. K. Czamanske, J. L. Wooden and V. A.
Fedorenko, Mantle and crustal contributions to continental
flood volcanism,Tectonophysics,223, 39-52, 1993.
Austin, J. A., and E. Uchupi, Continental-oceanic
crustaltran-
sitionoff southwestAfrica, Bull. Am. Ass.Petrol. Geol., 66,
1328-1347, 1982.
Baker,C. L., The lava fieldsof the Paran•tbasin,SouthAmerica,
d. Geol., 31, 66-79, 1923.
Baksi,A. K., R. V. Fodor,andE. Farrar,Preliminaryresultsof
4øAr-39Ar datingstudieson rocksfromthe SerraGeralflood
basalt province and the Brazilian continentalmargin
(abstract),Eos Trans.AGU, 72, 300, 1991.
Bellieni G., P. Comin-Chiaramonti, L. S. Marques,A. J. Melfi,
E. M. Piccirillo,A. J. R. Nardy,andA. Roisenberg,
High- and
1ow-Ti flood basaltsfrom the Paran•tplateau (Brazil):
petrologyand geochemicalaspectsbearingon their mantle
origin. N. db. Miner. Abh., 150, 272-306, 1984a.
Bellieni,G., P. Comin-Chiaramonti,L. S. Marques,A. J. Melfi,
E. M. Piccirillo, and D. Stolfa, Low-pressureevolutionof
basaltic sills from boreholesin the Paran•tbasin, Brazil,
TschermaksMin. Pet. Mitt., 33, 25-47, 1984b.
Bellieni,G., P. Comin-Chiaramonti,
L. S. Marques,A. J. Melfi,
A. J. R. Nardy, C. Papatrechas,
E. M. Piccirillo, A. J. R.
Nardy, A. Roisenberg,andD. Stolfa,Petrogeneticaspects of
acid and basalticlavas from the Paran•tplateau(Brazil):
aspects, d. Petrol.,27, 915-
triplejunctions:
tem: an overview,Tectonophysics,
213, 97-138, 1992.
Class,C., S. L. Goldstein,
S. G. Galer,andD. Weis,Youngfor-
mationage of a mantleplumesource,Nature, 362, 715-721,
1993.
Cleverly,R. W., P.J.Betton,andJ. W. Bristow,Geochemistry
andpetrogenesis of theLebomborhyolites,Geol.Soc.S. Aft.
Spec.Publ., 13, 171-194, 1984.
Cliff, R. A., P. E. Baker, and N.J. Mateer, Geochemistry
of
Inaccessible Island volcanics, Chem. Geol., 92, 251-260,
1991.
Comin-Chiaramonti, P., C. B. Gomes, E. M. Piccirillo, and G.
Rivalenti, High TiO2 dykesin the coastlineof Silo Pauloand
Rio de Janeiro States,N. dahrb. Miner. Abh., 146, 133-150,
1983.
Comin-Chiaramonti,P., G. Bellieni, E. M. Piccirillo, and A. J.
Melfi, Classificationand petrographyof continentalstratoid
volcanics and related intrusives from the Paranti basin
(Brazil), in MesozoicFlood Volcanismfrom theParand Basin
(Brazil): Petrogeneticand GeophysicalAspects,editedby E.
M. Piccirillo, and A. J. Melfi, pp. 47-72, IAG-USP, Silo
Paulo, 1988.
Cordani, U. G., L. Civetta, M. S. M. Mantovani, R. Petrini, K.
Kawashita,C. J. Hawkesworth,P. N. Taylor,A. Longinelli,G.
Cavazzzini, and E. M. Piccirillo, Isotope geochemistryof
flood volcanics from the Paran•t basin (Brazil), in The
MesozoicFlood Volcanismof theParand Basin:Petrogenetic
and GeophysicalAspects,editedby E. M. Piccirillo, and A. J.
Melfi, pp. 157-178, IAG-USP, SiloPaulo, 1988.
Cordani, U. G., and P. Vandoros, Basaltic rocks of the Paran•t
basin,in Problemsin Brazilian GondwanaGeology,editedby
J. J. Bigarella, R. D. Becker, and J. D. Pinto, pp. 207-231,
1967.
Cox, K. G., A model for flood basalt volcanism, d. Petrol., 21,
629-650, 1980.
Cox, K.G., The Karoo Province, in Continental Flood Basalts,
editedby J. D. McDougall,pp. 239-271,Kluwer, 1988.
de Wit, M., M. Jeffery,H. Bergh and L. Nicolaysen,Geological
mapof sectorsof Gondwanareconstructed to their disposition
at c. 150 Ma, Bernard Price Inst. Geophys.Res., Univ. of
Witwatersrand,Johannesburg (scale1:10,000,000),1988.
Druecker,M.D., and S. P. Gay, Mafic dyke swarmsassociated
with Mesozoic rifting in easternParaguay,in Mafic Dyke
Swarms, Spec. Pap. 34, edited by H. C. Halls and W. F.
Fahrig, pp. 187-193, Geological Association of Canada,
Toronto, ON, 1987.
Duncan, A. R., J. S. Marsh, S.C. Milner, and A. J. Erlank,
Distribution and petrogenesisof the basic rocks of the
Etendeka Formation of northwestern Namibia, in
GeochemicalEvolutionof the ContinentalCrust, pp. 10-19,
Poeosde Caldas, Brazil, 1988.
Duncan,A. R., S. R. Newton, C. van den Berg, and D. L. Reid,
Geochemistryand petrologyof doleritesillsin the Huab River
valley, Damaraland,north-western Namibia, Communs.geol.
Surv. Namibia, 5, 5-17,

Ellam, R. M., and K. G. Cox, An interpretationof Karoopicrite
basaltsin termsof interactionbetweenasthenospheric magmas
andthe mantlelithosphere, Earth Planet.Sci. Lett., 105, 330-
342, 1991.
Erlank, A. J., J. S. Marsh, A. R. Duncan,R. M. Miller, C. J.
Hawkesworth,P. J. Betton,andD.C. Rex, Geochemistry and
petrogenesis of the EtendekavolcanicrocksfromSouthWest
Africa/Namibia,Geol. Soc. S. Afr. Spec.Publ., 13, 195-246,
1984.
Erlank, A. J., A. R. Duncan,J. 5. Marsh, R. J. Sweeney,C. J.
Hawkesworth,S.C. Milner, R. M. Miller, and N. W. Rogers,
A laterallyextensivegeochemical
discontinuity
in the sub-
continentalGondwanalithosphere,in GeochemicalEvolution
of the ContinentalCrust,pp. 1-10, Polosde Caldas,Brazil,
1988.
Ernesto,M., and I. G. Pacca,Palaeomagnetism of the Paranti
basin flood volcanics, southernBrazil, in MesozoicFlood
Volcanism Jkomthe ParandBasin(Brazil):Petrogenetic and
GeophysicalAspects,edited by E. M. Piccirillo, and A. J.
Melfi, pp. 229-255, IAG-USP, SiloPaulo, i988.
F6raud, G. and V. Courtillot, Comment on: 'Did Deccan volcan-
ism pre-date the Cretaceous-Tertiary transition ?', Earth
Planet. Sci. Lett., 122, 259-262, 1994.
Fodor, R. V., and Vetter, S. K., Rift-zone magmatism:petrology
of basaltic rocks transitional from CFB to MORB, southeast-
ern Brazil margin, Contrib. Mineral. Petrol., 88, 307-32i,
1984.
Fodor,R. V., C. Corwin, and A. Roisenberg,Petrologyof Serra
Geral (Paranti) continentalflood basalts,southernBrazil:
crustal contamination,source material and South Atlantic
magmatismContrib.Mineral. Petrol.,91 54-65, 1985a.
Fodor, R. V., C. Corwin, and A. N. Sial, Crustalsignaturesin the
SerraGeral flood basaltprovince,southernBrazil: O- and Sr-
isotopeevidence,Geology,13, 763-765, i 985b.
Fodor, R. V., Low- and high-TiO2 flood basaltsof southern
Brazil: origin from picritic parentageand a commonmantle
source,Earth Planet. Sci. Lett., 84, 423-430, i987.
Gallagher,K., and C. J. Hawkesworth,Dehydrationmeltingand
the generation of continental
floodbasalts,Nature,358, 57-59
1992.
Gallagher,K., and C. J. Hawkesworth,Mantle plumes,conti-
nentalmagmatism andasymmetryin the SouthAtlantic,Earth
Planet. Sci. Lett., 123, 105-117, 1994.
Gallagher,K., C. J. Hawkesworth,andM. S. M. Mantovani,The
denudationhistoryof the onshorecontinentalmarginof SE
Brazil inferred from apatite fission track data, d. Geophys.
Res., 99, 18117-18145, 1994.
Garland, F. E., C. J. Hawkesworth, and M. S. M. Mantovani,
Description and petrogenesisof the Paranti rhyolites, d.
Petrol., 36, 1193-1227, 1995.
Garland,F. E., S. P. Turner and C. J. Hawkesworth,Shiftsin the
sourceof Parantibasaltsthrough time, Lithos, 37, 223-243,
1996.
Gibson, S. A., R. N. Thompson,A. P. Dickin, and O. H.
Leonardos,Mafic potassicmagmatickey to plume-lithosphere
interactions and continental flood-basalts, Earth Planet. Sci.
Lett., 14i, 325-341, 1996.
Gibson,S. A., R. N. Thompson,A. P. Dickin, J. G. Mitchell, and
S.C. Milner, Temporalvariationin magmasourcesrelatedto
the impactof the TristanPlumed. Conf.Abs.,2, 32, 1997.
Harland, W. B., R. L. Armstrong,A. V. Cox, L. E. Craig and D.
PEATE 243
G. Smith, A Geological Time Scale 1989, Cambridge
University Press,Cambridge, 1990.
Harris, C., H. S. Smith, S.C. Milner, A. J. Erlank, A. R. Duncan,
J. S. Marsh, and N. P. Ikin, Oxygen isotopegeochemistryof
the Mesozoic volcanics of the Etendeka Formation, Namibia,
Contrib. Mineral. Petrol., 102, 454-461, 1989.
Harris, C., A.M. Whittingham, S.C. Milner and R. A.
Armstrong,Oxygen isotopegeochemistryof the silicia vol-
canic rocks of the Etendeka-Paranfiprovince: source con-
straints,Geology,! 8, i i i 9-112 i, i 990.
Harry, D. L., and D. S. Sawyer, Basaltic volcanism,mantle
plumesand the mechanicsof rifting: the Parantiflood basalt
provinceof SouthAmerica,Geology,20, 207-210, 1992.
Hawkesworth,C. J., J. S. Marsh, A. R. Duncan,A. J. Erlank, and
M. J. Norry, The role of continentallithospherein the gen-
eration of the Karoo volcanic rocks' evidence from combined
Nd- and Sr-isotopestudies,Geol. Soc.S. Afr. Spec.Publ., 13,
341-354, 1984.
Hawkesworth, C. J., M. S. M. Mantovani, P. N. Taylor, and Z.
Palacz, Evidence from the Parantiof southBrazil for a conti-
nental contribution to DUPAL basalts,Nature, 322, 356-359,
1986.
Hawkesworth, C. J., M. S. M. Mantovani, and D. W. Peate,
LithosphereremobilisationduringParantiCFB magmatism,in
Oceanic and Continental Lithosphere,' Similarities and
Di,/Jerences,editedby M. A. Menzies and K. Cox, pp. 205-
223, Journalof Petrology,Oxford, 1988.
Hawkesworth, C. J., K. Gallagher, S. Kelley, M. S. M.
Mantovani, D. W. Peate, M. Regelous, and N. W. Rogers,
Parantimagmatismand the openingof the SouthAtlantic, in
Magmatismand the Causesof ContinentalBreak-up, Spec.
Publ. 68, editedby B. Storey,A. Alabaster,and R. Pankhurst,
pp. 221-240, The GeologicalSociety,London, i 992.
Henry, C. D., and J. A. Wolff, Distinguishingstrongly rheo-
morphictuffs from extensivesilicic lavas,Bull. Volcanol.,54,
171-186, 1992.
Hergt, J. M., D. W. Peate,and C. J. Hawkesworth,The petrogen-
esis of Mesozoic Gondwana low-Ti flood basalts, Earth
Planet. Sci. Lett., 105, i 34-148, i 991.
Huang, Y. M., C. J. Hawkesworth,P. W. van Calsteren,and F.
McDermott, Geochemicalcharacteristicsand origin of the
Jacupirangacarbonatites,Brazil, Chem. Geol., 119, 79-99,
i995.
Iacumin,P., E. M. Piccirillo, and A. Longinelli, Oxygen isotopic
compositionof Lower Cretaceoustholeiitesand Precambrian
basementrocks from the Paranti basin (Brazil): the role of
water-rockinteraction,Chem. Geol. (IsotopeGeoscience),86,
225-237, 1991.
Kirstein, L. A., C. J. Hawkesworth, and S. P. Turner, CFB mag-
matismin southernUruaguay:marginalto a mantleplume? d.
Conf A bs., 2, 44, 1997.
Leinz, V., A. Bartorelli,and C. A. Isotta,Contribui•io ao estudo
do magmatismobasfilticoMesoz6icda baciado Paranti,An.
Acad. bras. Ci•nc., 40, 167- i 81, 1968.
le Roex, A. P., R. A. Cliff, and B. J. I. Adair, Tristan da Cunha,
SouthAtlantic: geochemistry and petrogenesis of a basanite-
phonolitelava series,d. Petrol.,31,779-812, 1990.
Light, M.P. R., M.P. Maslanyj, andN. L. Banks,New geophys-
ical evidencefor extensionaltectonicson the divergentmargin
offshore Namibia, in Magmatism and the Causes of
ContinentalBreak-up, Spec.Publ. 68, editedby B. Storey,
244 PARANA-ETENDEKA PROVINCE
Alabaster, and R. Pankhurst,pp. 257-270, The Geological
Society,London, 1992
Long, P. E., and B. J. Wood, Structures,texturesand cooling
historiesof Columbia River basaltflows, Bull. Geol. Soc. Am.,
97, 1144-1155, 1986.
Mahoney,J. J., A. P. le Roex,Z. Peng,R. L. Fisher,•d J.H.
Natland, Southwesternlimits of Indian Ocean ridge mantle
and the origin of low 206pb/204pb mid-oceanridge basalt:
isotope systematicsof the central SouthwestIndian Ridge
(17ø-50øE),d. Geophys.Res.,97, 19,771-19,790, 1992.
Manley, C. R., How voluminousrhyolite lavas mimic rheo-
morphicignimbrites:eruptivestyles,emplacement conditions,
and formation of tuff-like features, Geology, 23, 349-352,
1995.
Mantovani, M. S. M., L. S. Marques,M. A. De Sousa,L. Civetta,
L. Atalla, and F. Innocenti, Trace element and strontiumiso-
tope constraintson the origin and evolution of Paranticon-
tinental flood basalts of Santa Catarina state, southernBrazil,
d. Petrol., 26, 187-209, 1985a.
Mantovani, M. S. M., U• G. Cordani, and A. Roisenberg,
Geoquimica isot6pica em vulcfinicasficidas da Bacia do
Parantie implica{;fesgen6ticasassociadas,
Rev. Bras. Geoc.,
15, 61-65, 1985b.
Mantovani, M. S. M., and C. J. Hawkesworth,An inversionap-
proachto assimilationand fractionalcrystallisation processes,
Contrib. Mineral. Petrol., 105, 289-302, 1990.
Mantovani, M. S. M., A. C. B.C. Vasconcellos,W. Shukowsky,
E. J. Milani, M. Basei, S. J. Hurter, and S. R. C. de Frietas,
The Brusquetransect(SA20) from the Dom Felicianobelt to
the Amazon Craton: explanatory pamphlet, Global
GeosciencesTransectProject, 1991.
Marsh, J. S., Relationshipsbetween transform directions and
alkaline igneous rock lineaments in Africa and South
America, Earth Planet. Sci. Lett., 18, 317-323, 1973.
Marsh, J. S., A. J. Erlank, and A. R. Duncan, Preliminarygeo-
chemicaldata for dolerite dykes and sills of the southernpart
of the Etendeka Igneous Province, Communs.geol. Surv.
Namibia, 7, 71-73, 1991.
Milner, S.C., The geology and geochemistryof the Etendeka
Formation quartz latites,Namibia, Ph.D. thesis,University of
Cape Town, Cape Town, SouthAfrica, 1988.
Milner, S.C., A. R. Duncan, and A. Ewart, Quartz latite
rheoignimbriteflows of the EtendekaFormation,northwestern
Namibia, Bull. Volcanol., 54, 200-219, 1992.
Milner, S.C., A. P. le Roex, and J. M. O'Connor, Age of
Mesozoic igneousrocks in northwesternNamibia, and their
relationshipto continentalbreak-up,d. Geol. Soc.Lond., 152,
97-104, 1995a.
Milner, S.C., A. R. Duncan,A.M. Whittingham,and A. Ewart,
Trans-Atlantic correlation of eruptive sequencesand indi-
vidual silicic units within the Paranfi-Etendekaigneous
province,d. Volcanol.Geotherm.Res.,69, 137-157, 1995b.
Milner, S.C., and A. P. le Roex, Isotope characteristicsof the
Okenyenya igneous complex, northwesternNamibia: con-
straintson the compositionof the early Tristanplumeand the
origin of the EM1 mantlecomponent,Earth Planet. Sci. Lett.,
141,277-291, 1996.
Mizusaki, A.M. P., R. Petrini, G. Bellieni, P. Comin-Chiramonti,
J. Dias, A. De Min, and E. M. Piccirillo, Basaltmagmatism
along the passivecontinentalmargin of SE Brazil (Campos
basin),Contrib.Mineral. Petrol., 111, 143-160, 1992.
Montes-Lauer, C. L., I. G. Pacca,A. J. Melfi, E. M. Piccirillo, G.
Bellieni, R. Petrini, and R. Rizzieri, The Anari and Tapirapui
JurassicFormations,westernBrazil: palaeo-magnetism, geo-
chemistryand geochronology,Earth Planet. Sci. Lett., 128,
357-371, 1995.
Morgan, W. J., Hotspot tracks and the opening of the Atlantic
and Indian Ocean, in The Sea: vol. 7; The Oceanic Litho-
sphere,editedby C. Emiliani, pp. 443-487, Wiley, New York,
1981.
Niirnburg, D., and R. D. Miiller, The tectonicevolution of the
SouthAtlantic from Late Jurassicto present,Tectonophysics,
191, 27-53, 1991.
O'Connor, J. M., and R. A. Duncan, Evolution of the Walvis
Ridge - Rio Grande Rise hotspotsystem:implicationsfor
African and South Americanplate motionsover plumes,d.
Geophys.Res., 95, 17,474-17,502, 1990.
Peate, D. W., Stratigraphyand petrogenesisof the Paranticon-
tinentalflood basalts,southernBrazil, Ph.D. thesis,The Open
University, Milton Keynes, 1990.
Peate, D. W., C. J. Hawkesworth, M. S. M. Mantovani, and W.
Shukovsky,Mantle plumesandflood basaltstratigraphyin the
Paranti,SouthAmerica, Geology,18, 1223-1226, 1990.
Peate, D. W., Co J. Hawkesworth, and M. S. M. Mantovani,
Chemical stratigraphyof the Parantilavas (South America):
classificationof magma types and their spatial distribution,
Bull. Volcanol., 55, 119-139, 1992.
Peate, D. W., and C. J. Hawkesworth,Lithosphericto astheno-
spheric transition in Low-Ti flood basalts from southern
Paranti,Brazil, Chem. Geol., 12 7, 1-24, 1996.
Peng, Z. X., and J. J. Mahoney, Drillhole lavas from the north-
westernDeccan Traps, and the evolution of R6union hotspot
mantle, Earth Planet. Sci. Lett., 134, 169-185, 1995.
Petrini, R., L. Civetta, E. M. Piccirillo, G. Bellieni, P. Comin-
Chiaramonti,L. S. Marques,and A. J. Melfi, Mantle hetero-
geneity and crustal contaminationin the genesisof low-Ti
continentalflood basaltsfrom the Parantiplateau(Brazil): Sr-
Nd isotopesand geochemicalevidence,d. Petrol., 28, 701-
726, 1987.
Piccirillo, E. M., and A. J. Melfi (Eds), The Mesozoic Flood
Volcanismof the Paranc•Basin: Petrogeneticand Geophysi-
cal Aspects,600 pp., IAG-USP, SiloPaulo, 1988.
Piccirillo, E. M., A. J. Melfi, P. Comin-Chiaramonti,G. Bellieni,
M. Ernesto,L. S. Marques, A. J. R. Nardy, I. G. Pacca,A.
Roisenberg,and D. Stolfa, Continentalflood volcanismfrom
the Parantibasin(Brazil), in ContinentalFlood Basalts,edited
by J. D. McDougall, pp. 195-238,Kluwer, 1988.
Piccirillo, E. M., G. Bellieni, H. Cavazzini, P. Comin-
Chiaramonti, R. Petrini, A. J. Melfi, J.P. P. Pinese, P.
Zantadeschi,and A. de Min, Lower Cretaceoustholeiitic dyke
swarms in the Ponta Grossa Arch (South East Brazil):
petrology, Sr-Nd isotopes,and genetic relationshipsfrom
Paranti flood volcanics, Chem. Geol., 89, 19-48, 1990.
Rampino, M. R., S. Self, and R. B. Stothers,Volcanic winters,
Ann. Rev. Earth Planet. Sci., 16, 73-99, 1988.
Raposo,M. I. B., and M. Ernesto,Anisotropyof magneticsus-
ceptibility in the Ponta Grossadyke swarm (Brazil) and its
relationshipwith magma flow direction,Phys. Earth Planet.
Int., 87, 183-196, 1995.
Raup, D. M., and J. J. Sepkoski,Periodicityof extinctionsin the
geologicpast,Proc. Natl. Acad. Sci. USA,81, 801-805, 1984.
Regelous,M., Geochemistryof doleritesfrom the Paranti

basalt province, southernBrazil, Ph.D. thesis, The Open
University, Milton Keynes, 1993.
Reid, D. L., The Cape Peninsuladolerite dyke swarm, South
Africa, in Mafic Dykesand EmplacementMechanisms,edited
by A. J. Parker,P. C. Rickwoodand D. H. Tucker,pp. 325-
334, Balkema, Rotterdam, 1990.
Renne, P. R., M. Ernesto, I. G. Pacca,R. S. Coe, J. M. Glen, M.
Pr•vot, and M. Perrin, The age of Parantiflood volcanism,
rifting of Gondwanaland, and the Jurassic-Cretaceous
boundary,Science,258, 975-979, i992.
Renne, P. R., D.F. Mertz, M. Ernesto,L. Marques,W. Teixeira,
H. H. Ens, and M.A. Richards,Geochronologic constraintson
magmatic and tectonic evolution of the Paranti province
(abstract),Eos Trans.AGU, 74, 553, 1993.
Renne, P. R., Z. Zichao, M. A. Richards,M. T. Black, and A. R.
Basu, Synchrony and causal relations between Permian-
Triassic boundary crises and Siberian flood volcanism,
Science, 269, 1413-1416, 1995.
Renne, P. R., K. Deckart, M. Ernesto, G. F6raud, and E. M.
Piccirillo,Age of the PontaGrossadike swarm(Brazil), and
implicationsto Paranti flood volcanism,Earth Planet. Sci.
Lett., 144, 199-212, 1996a.
Renne,P. R., J. M. Glen, S.C. Milner, andA. R. Duncan,Age of
Etendeka flood volcanism and associated intrusions
southwestern Africa, Geology,24, 659-662, 1996b.
Richards, M. A., R. A. Duncan, and V. E. Courtillot, Flood
basaltsand hotspottracks:plumeheadand tails, Science,246,
103-107, 1989.
Richardson, S. H., A. J. Erlank, A. R. Duncan, and D. L. Reid,
CorrelatedNd, Sr and Pb isotopevariationin Walvis Ridge
basalts and implications for the evolution of their mantle
source,Earth Planet. Sci. Lett., 59, 327-342, 1982.
Rocha-Campos,A. C., U. G. Cordani, K. Kawashita, H. M.
Sonaki,and I. K. Sonaki,Age of the Parantiflood volcanism,
in MesozoicFlood Volcanism
from theParandBasin(Brazil):
Petrogenetic and GeophysicalAspects, edited by E. M.
Piccirillo and A. J. Melfi, pp. 25-45, IAG-USP, Silo Paulo,
1988.
Saunders,A.D., M. Storey, R. W. Kent, and M. J. Norry,
Consequences of plume-lithosphere interactions, in
Magmatismand the Causesof ContinentalBreak-up,Spec.
Publ., 68, editedby B. Storey,A. Alabaster,and R. Pankhurst,
pp. 41-60, The GeologicalSociety,London,1992.
Stewart,K., S. Turner, S. Kelley, C. J. Hawkesworth,L. Kirstein,
andM. S. M. Mantovani,
3-D 40Ar-39Argeochronology
in
the Parantiflood basaltprovince,Earth Planet.Sci. Lett., 143,
95-110, 1996.
Sun, S.-s., Lead isotopic study of young volcanic rocks from
mid-ocean ridges, ocean islands, and island arcs, Philos.
Trans. R. Soc. London. Ser. A, 297, 409-445, 1980.
Sun, S.-s. and W. F. McDonough,Chemicaland isotopicsys-
tematicsof oceanicbasalts:implicationsfor mantlecompo-
sition and processes,in Magmatism in the Ocean Basins,
Spec.Publ. 42, editedby A.D. Saunders andM. J. Norry, pp.
313-345, The GeologicalSociety,London,1989.
Thompson,R. N., M. A. Morrison, A. P. Dickin, and G. L.
Hendry, Continentalflood basalts..... Arachnidsrule OK ?, in
ContinentalBasalts and Mantle Xenoliths, edited by C. J.
PEATE 245
Hawkesworth and M. J. Norry, pp. 158-185, Shiva Press,
Nantwich, 1983.
Thompson, R. N., and S. A. Gibson, Subcontinentalmantle
plumes, hotspotsand pre-existingthinspots,or. Geol. Soc.
Lond., 147, 973-977, 1991.
Toyoda,K., H. Horiuchi,and M. Tokonami,Dupal anomalyof
Braziliancarbonatites:geochemical correlationswith hotspots
in the SouthAtlantic and implicationsfor the mantle source,
Earth Planet. Sci. Lett., 126, 315-331, 1994.
• n I
Turner, S. P., •v•. r•ege•ous,S. Kelley', C. J. Hawkesworth,and
M. S. M. Mantovani,Magmatismandcontinental break-upin
the SouthAtlantic:high precision40Ar-39Argeochronology,
Earth Planet. Sci. Lett., 121,333-348, 1994.
Turner, S. P., and C. J. Hawkesworth, The nature of the sub-
continental mantle: constraintsfrom the major element
compositionof continentalflood basalts,Chem. Geol., 120,
295-314, 1995.
Turner, S. P., C. J. Hawkesworth,K. Gallagher,K. Stewart,D.
W. Peate, and M. S. M. Mantovani, Mantle plumes, flood
basaltsand thermalmodelsfor melt generationbeneathcon-
tinents:assessment
of a conductiveheatingmodeland ap-
plicationto the Paranti,or. Geophys.Res.,101, 11,503-11,518,
1996.
in Ulbrich, H.G.J., and C.B. Gomes, Alkaline rocks from conti-
nentalBrazil, Earth Sci. Rev., 17, 131-154, 1981.
Ussami,N., A. Kolisnyk,M. I. B. Raposo,F. J. F. Ferreira,E. C.
Molina, andM. Ernesto,Detectabilidade
magn•ticade diques
do Arco de Ponta Grossa: um estudo integrado de
magnetometriaadrea e magnetismode rocha, Rev. Brasil.
Geoc., 21,317-327, 1991.
VanDecar,JoC., D. E. James,and Mo Assump•fio,Seismicevi-
dencefor a fossil mantleplume beneathSouthAmericaand
implicationsfor plate driving forces, Nature, 378, 25-31,
1995.
Venkatesan,T. R., K. Pande,and K. Gopalan,Did Deccanvol-
canism pre-date the Cretaceous/Tertiary
transition? Earth
Planet. Sci. Lett., 119, 181-189, 1993.
White, R. S., and D. P. McKenzie,Magmatismat rift zones:the
generationof volcaniccontinentalmarginsand flood basalts,
,I. Geophys.Res.,94, 7685-7729, 1989.
White, R. S., and D. P. McKenzie, Mantle plumesand flood
basalts,or. Geophys.Res.,100, 17,543-17,585,1995.
Whittingham,A., Stratigraphy and petrogenesisof the volcanic
formationsassociated with the openingof the SouthAtlantic,
southernBrazil, Ph.D. thesis,Universityof Oxford, Oxford,
1991.
Yuen, D. A. and L. Fleitout,Thinning of lithosphereby small-
scaleconvectivedestabilisation,Nature, 313, 125-128, 1985.
Zalan, P. V., S. Wolff, J. C. J Conceicfio,M. A. M Astolfi, A. S.
Veira, V. T. Appi, O. A. Zanottoand A. Marques,Tectonics
and sedimentation of the ParantiBasin,Anaisdo symposio do
Gondwana, 35, 1987.
David W. Peate,Departmentof Earth Sciences,The Open
University,Walton Hall., Milton Keynes,MK7 6AA, United
Kingdom.(e-mail:
 Stratigraphyand Age of Karoo Basaltsof Lesothoand Implicationsfor
CorrelationsWithin the Karoo IgneousProvince
J. S. Marsh,• P. R. Hooper,
2J. Rehacek,
2R. A.Duncan,
3andA. R. Duncan
4

The Lesothoremnantcontainsthe type succession

for Karoo 10w-Ti basaltsof
centralsouthernAfrica. The 4øAr/39Ar
datingindicatesthat the sequence
was
emplacedwithin a very shortperiodat about180Ma andconsists of a monotonous
pile of compound basaltlavaflowswhichlackssignificantpalaeosols andpersistent
sedimentary intercalations. We haveusedgeochemistry to establish a stratigraphic
subdivision of the lavapile. Thin unitsof basaltflows,theMoshesh'sFord,Golden
Gate, Sani, Roma, Letele, and Wonderkop units, with diverse geochemical
characterand restrictedgeographicaldistribution,are presentat the baseof the
succession.Theseareoverlainby extensiveunitsof compositionally moreuniform
basalt,the Mafika Lisiu, Maloti, SenquandMothae units,which build the bulk of
the sequence.A singlepalaeomagnetic polarityreversaloccurswithin the lower
third of the basaltsuccessionand is consistentlylocatedwithin the Mafika Lisiu
unit. This andthe persistentandrelativelyuniformthicknessof the stratigraphic
units suggestthat the pile was constructedin a uniform mannerby eruptionof
basalt onto a generallyplanarsurfacefrom a widespreadplexusof dykes. The
stratigraphic sequence in Lesothocloselyresembles thatin the thinnersequenceof
10w-Ti basaltsof the SpringbokFlatsremnant,some400 km to the north.A thin
unit of high-Ti basaltwithin the upperpartof the SpringbokFlats sequencecan be
correlatedwith the thick high-Ti basalt suite along the riff-related Lebombo
structureon the easternmarginof the Karooprovince. This is the first established
correlationbetweenthesetwo importantoutcropsof Karoo volcanicrocksand
demonstrates that the 10w-Ti basalts of Lesotho and the cratonic interior are the
approximatetime equivalentsof the lower part of the Lebombosequence.This
conclusionhasimportantimplicationsfor modelsfor the origin of the Karoo flood
basaltprovince.

1. INTRODUCTION
•Department
of Geology,RhodesUniversity,SouthAfrica
2Department
of Geology,Washington
StateUniversity,Pullman,
Washington
Remnantsof the eruptedandintrusiveproductsof theKaroo
3Collegeof Oceanicand Atmospheric
Sciences,OregonState igneousprovince, one of the classicMesozoic continental
University, Corvallis, Oregon floodbasaltprovinces,arefoundthroughoutsouthernAfrica.
4Department
of Geological
Sciences,
Universityof CapeTown, Suchlargecontinental igneouseventsare frequentlyascribed
Rondebosch,South Africa to the rise of deep-seatedmantle plumes and, in some
instances,are thoughtto be relatedto continentalbreakup.
Plumescan provide the anomalouslyhigh temperaturesat
Large IgneousProvinces:Continental,Oceanic,andPlanetary shallowdepthnecessaryfor generationof the largevolumes
Flood Volcanism of basalticmagmaovera shortperiod,a characteristicof flood
GeophysicalMonograph100 basaltprovinces.
Copyright1997 by the AmericanGeophysicalUnion However, there is debate concerningplume-lithosphere
248 STRATIGRAPHYAND AGE OF KAROOBASALTSOF LESOTHO

interactions,melt generationin the plume, and how the

compositionsof eruptedbasaltsrelate to specificsourcesin
the plume and/orthe overlyinglithosphere.Thesequestions
I ........................
.,-.•-•C:•(
i........
:'.,"'
';.•':':'.'
ZIMBABWE
.":".•'..'•
• )
"" ...... ß..... 2.'.. ...;;7:.: "

can be addressedthrough geochemistryof the basaltsbut • r .-,I '- '-•" ",•, ß

'• •,'"'.}BOTSWANA
-'5.•
requirecomprehensive documentation of the compositionsof
the different geochemicaltypes of basalt, their volumes,
distribution,and agerelationships.
• , c.."..i'
I :: :: '"-'
•'•-"• .. •'. *--, '"'
Somerecentauthors[Morgan, 1981; Whiteand McKenzie, • I '• Marlenlal
ß r"
/
1989; Campbelland Griffiths,1990] proposeplumemodels .. [ . ', ! ',. /SPRINdBOKFLATS
for the originof the Karooprovince. The lackof a thorough Keelmanshoop
ß • .'" ,"'

knowledgeof the ages,distribution,stratigraphyand geo- •'"N

iß¾o---TANDJ
IESBERG x0O
chemicalcharacterof all the igneousproductsof the Karoo \ LESOT!
I(
provincerendertheir analysesunsatisfactory.For example,
there is considerable documentation of the typesof igneous 5, ' •// NEW
AMALFI
materials that were erupted and their broad isotopicand
geochemical featuresbut, apartfrom the centralsectionof the CapeTown
........ •""'•.-..- INSIZWA
0 1000km
Lebombomonocline,very little is known aboutthe detailsof I !

the geochemical stratigraphyin the different erosional

remnants,or how they correlatewith one another.Agesare
alsopoorly understood.In contrastto the short-livednature
of the activityin manyotherfloodbasaltsequences, available
datafor Karoovolcanismhavebeeninterpreted asindicating
severalpulses.of activityovermany tensof millionsof years
[Fitchand Miller, 1984]. The existenceof a high-Ti,low-Ti
provincialityhasbeenknownfor decades, but itsgeographical
structurehasnotbeenfully documented. Anotheroutstanding
problem is the lack of precise knowledgeregardingthe
eruption sites for the volcanic sequence. Eruption site
localitiesare more importantthan lava remnantdistributions
for understanding the emplacement of the Karooprovincein
relationto plumetectonics.
A number of detailedstudiesare currentlyunderwayto
addressthesedeficiencies.This paperdescribes resultsof a
studyaimed at establishingthe stratigraphicsequencein the
principaloutcropof the low-Ti basaltsequence, the Lesotho
remnant, where the volcanic succession,more than 1.5 km
thick,is spectacularly exposedin a deeplydissected, uplifted
plateau.The basisfor the stratigraphicsubdivisionis geo-
chemical.Compositional variations withinthebasaltsequence
are slight and we argue at length for the validity of the
stratigraphicsubdivisionbecausethey are provingcrucialin
establishing correlations amongst other low-Ti basalt
remnants. We also show that the stratigraphicsequenceis
consistentwith recentage andpalaeomagnetic data. Finally,
we discuss the widerimplicationof thenewresults;the insight
they provide into the constructionof the lava pile, andtheir
use in allowing, for the first time, a stratigraphiccorrelation
between the Lebombo and Lesotho remnants to be established.
The comprehensive compositionalcharacterization
of the low-
Ti basalt types of Lesothoprovidesthe basisfor detailed
petrogeneticmodelling, a topic to be addressedin a future
paper.
Figure 1. Map of theKarooigneousprovinceshowingdistribution
of remnantbasalt outcrops(black) and subsurface-extent (light
stippledarea) and localitiesreferredto in the text. The limit of the
CentralArea is indicatedby the dashedline. The heavily stippled
box is the areashownin detailin Figure2.
2. OVERVIEW OF THE KAROO PROVINCE
Salientfeaturesof the provincehavebeensummarised by
Ealeset al. [1984] andCox [1988] andwe mentiononly some
of the importantfeatureshere. Figure 1 showsthe distribution
of volcanic outcropsof the province,and it is importantto
emphasize that these are erosional remnants of a more
extensivecarapaceof lava which, on a wealth of circum-
stantial evidence, we believe to have covered much of
southernAfrica. Theseeruptionswere the culminationof a
long period of intracontinentalsedimentationto form the
Karoo sedimentarysequenceand its correlativesthroughout
much of southern Africa and neighbouringGondwana
continents. Areas between these erosional remnants are
intrudedby a networkof dykesandsillswhichareparticularly
well developedin the sedimentary sequences of the main and
subsidary Karoobasins(Figure2). Apart from the Botswana
andRooi Randdykeswarms,muchof the dykenetworkin the
Karoo province is diffuse, and although systematic
orientationsof dykesoccurin someareas,their development
is not intense.Althoughquantificationis difficult,thevolume
of shallow-level intrusionsin the Karoo basin may have
rivaled that of the eruptedproducts. Previousgeochemical
studies have shown the overall similarity between the
compositions of the dykesandsillsandthe associated basalts.
By considering the intrusiveandextrusivesuitestogether,the
overidingimpressionis oneof the widespreadavailability
 MARSH ET AL. 249

}ueenstown Queenstown
cOo

32 ø

100km

27ø 28 27øl 28

Figure2. Mapof dykesandsills(heavystipple) intruding

Karoosedimentary strata
in a portionof theKaroobasinsouth
of Lesotho(seeFigure1 for location).Basaltoutcropshownin lightstipplein topright-hand corner.

hugevolumes(1-2 x 106 km3) of magmabeneathsouthern and picritesand, along the Lebomboand Save-Mwenezi
Africa duringthe Karoo event. structures,
a thick sequence of rhyolite.In the Mweneziarea,
To facilitatedescription
anddiscussion,
it is convenient
to volcanismwas also accompaniedby ring complexesof
subdividethe Karoo provinceinto subareas.Eales et al. granite,gabbro,syenite,andnephelinesyenite.
[ 1984]identifieda CentralArea(Figure1) thatembracesall A furthernorth-south
geochemical subdivision of theKaroo
the Karoo volcanic remnants of the interior of southern Africa provincebasedon a high-Ti andlow-Ti provinciality within
and their associatedvast network of intrusive dykes and basalticrockswas recognizedby Cox et al. [1967]. This
sheets. The volcanic remnants included in the Central Area provincialityhasbeendocumented anddiscussed by Erlank
are the main Lesotho remnant,the volcanic sequencenear et al. [1988] and $weeneyand 14'atkeys [1990] and is now
Marientalin Namibia,theSpringbok Flatssome400 km north known to occur in all Mesozoic flood basalt provincesof
of Lesotho,andthe vastoutcropsandsubcrops of Botswana southernGondwana.Althoughthe exactboundariesbetween
as well as the intrusive suites associated with the main Karoo the two types are subjectto debate,it is clear that they
basin. We retain this groupingfor convenience, with the transcend Mesozoic tectonic boundaries and do not coincide
exceptionof Botswana.Theprincipalelements in grouping with the geographicalareasusedhere.
these different areas into a Central Area are tectonic and
petrographic;
i.e., rocktypesare overwhelmingly
tholeiitic 3. SAMPLING AND ANALYTICAL TECHNIQUES
basaltsof similarcompositions
andwereemplacedin a stable
cratonic environment. Nine serialsections(Figure3) were sampledin the deeply
Marginal to the Central Area are the thick riff-related dissected and elevated basalt plateau in and adjacentto
sequences of theTuli syncline,theSave-Mwenezi (previously Lesotho, and in each sectionattemptswere madeto sample
Sabi-Nuanetsi)area and the 700-km-long Lebombo mono- every flow. Detailed descriptionsof samplelocalitiesare
cline. Thick anddiverseassemblages of volcanicrockswere givenin the Appendix. The altitudeof eachsamplesitewas
eruptedin theseareas,includingnephelinites,tholeiiticbasalts determined from altimeterreadings(precision+5 m)
250 STRATIGRAPHY AND AGE OF KAROO BASALTS OF LESOTHO

j I
29
elements and Na were determinedon undilutedpressed

27 ii•
powder pelletsas describedby Duncanet al. [1984] andle
Roex [1985]. Loss on ignition (LOI) and H20-were
Wonderkop
o /_...•/! determined gravimetrically. Strontium isotopes were
Clocohm
ß 0 r• • Oxbow determinedat the RadiogenicIsotopeFacility,Universityof
CapeTown.Standard preparationtechniques wereemployed
/75 Mafika
Lisiu andanalyseswerecarriedout on a VG Sectormulticollector
5 .'
/ massspectrometer applyinga massfractionation
correctionof
Maseruj• Btlshmafl'sPass 868r/88Sr = 0.1194. Measured 878r/86Sr ratios have been
normalised to a value of 0.71022 for the standard SRM-987.
....

• • .Roma
{od ...... Selected results are in Tables 1-3. Some 87% of the
X• ".
• •Sem6}' collectionof 469 samplesfrom Lesothohavebeenanalyzed
at both RU and WSU. Thus our collective data contain all the
•30

elementsof two separategeochemicalstudiesof the same

L•
/
I
sequence.Table1 containsdatathatallowcomparisonof the
WSU and RU data sets. For some elements there are
• oluksnok
inconsistentdifferenceswhich are beyondsimpleanalytical
precision.Thesedifferences reflectthe influenceof a variety
-Ben MacDhui
5-'/.•' of differentmethodsof sampling,samplepreparation,and
'audes Nek
X ray analyticaltechniquesandphilosophy.Despitedetailed
-31 Ba•kly East review of our datait hasnot beenpossibleto determinethe
causesof concentrationdifferences. Nevertheless,use of the
0 1O0 km
I I two sets of data, individually or in conjunction,yields
I I I
consistentresultsin the applicationsdescribedbelow.

4. OVERVIEW OF STRATIGRAPHIC VARIATIONS

Figure 3. The basaltremnantof LesothoandEasternCapeshowing
samplingsections(labeledlines)andotherlocalities.International
boundaryshownas a dashedline. Figure 4 shows the variation of some important
compositionalparameterswith height throughthe Oxbow
were correctedto absolutevalues abovemean sea level by section,oneof themostcompletesections throughthethickest
referenceto severalknownspotheightsontopographic maps. part of the basalt sequence. This sectionhas all of the
In all sectionsexceptthe Bushman'sPassandRomasections importantfeaturesshownby otherprofilesand,for descriptive
(Figure3), severalshortcoresweredrilledat eachsamplesite. purposes, illustrates the general vertical geochemical
Thesecoreswere usedfor whole-rockchemicalanalysisand variations within the basalts. The main features are as follows.
palaeomagnetic studiesat Washington
StateUniversity(WSU) 1. The chemicalvariabilityin the lowermostflows,in terms
andfor4øAr/39Ar datingat OregonStateUniversity(OSU). In of elementabundances and incompatibleelementratios,is
addition,at the majorityof samplesites,1-2 kg wholerock largecomparedto the main overlyingsequence.
sampleswere takenby hammer,usuallywithin a few metres 2. Within the mainsequence,thereis a complexpatternof
of the coresamplesites.Theselarger samples,togetherwith variationcharacterized by severalsharpinflectionsin element
someof the coresamples,wereanalyzedat RhodesUniversity abundancesand interelement ratios. However, there is a
(RU). Samplesof SpringbokFlatsbasaltwereobtainedfrom general upward trend of subdued but increasing
four boreholecoresdrilledthroughthesequence andwerealso differentiation,with the uppermostflows showinga sharp
analyzed at RU. At WSU major and trace elementswere trendto moreevolvedbasalticcompositions (low magnesium
determinedby wave-length dispersiveX ray fluorescence number(Mg#), Ni, Cr; high Ti, P, Zr, Fe, etc.).
(XRF) on 2:1 dilutionfusionbeadsasdescribed by Hooperet 3. Dykesandotherintrusions in the sectiongenerallyhave
al. [ 1993]. The WSU datado not containdeterminations for compositional characteristics
of basaltshigherin thesequence
loss of volatiles. In addition, a subsetof sampleswas and can be correlatedon the basis of geochemistrywith
analyzedfor rare-earthelements (REE),Hf, Th, Ta, Pb,andU known units within the basalts. However, in the Oxbow
by inductivelycoupledplasmamassspectrometry, also at sectionseveraldykesandsheetshighin thevolcanicsequence
WSU. At RU majorelements weredetermined by XRF using havecompositions whichwerenot foundamongst thebasalts
the fusiontechniqueof NorfishandHutton[1969], andtrace andmay havefed flowsthathavebeenremovedby

These features confirm results from earlier, more limited
studiesof Cox and Hornung[ 1966], Marsh and Eales [ 1984]
andMarsh [ 1984]; specifically,thereis a generalizedtrendto
more evolved compositionsupward in the sequence;thin
basalticunits of diversegeochemistryare locatedat the base
of the sequence;and the main part of the sequenceexhibits
ratherlimitedcompositional variabilitythat is a challengefor
stratigraphicsubdivision. This study is aimed at a Oa!!er
explorationof the geochemicaldiversityamongstthe lower
flows and at subdividing the main sequenceof flows,
previouslytermedthe LesothoFormationby severalauthors
[Locket al., 1974; Marsh andEales, 1984].
5. THE BASIS FOR GEOCHEMICAL SUBDIVISION
The compoundnature of many of the basalt flows, the
presenceof only onepalaeomagnetic reversalnearthe baseof
the pile, and the absenceof widespreadsedimentaryor
weathering horizons leave basalt compositionas the only
reasonablebasisfor establishing
a stratigraphic
subdivisionof
the lava pile. Thus, subdivisionmakes use of changesin
geochemical parameters, specifically element ratios of
immobilehigh-field-strengthincompatibleelementsand,to a
lesser extent, element abundances,to indicate boundaries
between sequencesof flows having specific and more
uniform compositional characteristics. Assessing the
significanceof compositionaldifferencesbetweenadjacent
samplesin a stratigraphicsequenceis somewhatsubjective
and due regardhasto be givento the following:
1. The stratigraphicsubdivisionis only usefulif it canbe
usedby others;a repeatstudyshouldarriveat the sameresult.
Becauseof the subtlenatureof the variationsin the sequence,
we havereliedon bothRU andWSU datasets(i.e., effectively
two separatestudies)to establishthe subdivisions.
2. Within-flow variation. Some flows are internally
differentiated but most are not. However, all flows have
undergonelow-temperature alterationleadingto depositionof
quartz, chalcedony, and zeolite in vesiculatedtops and
bottomsof flows. Althoughwe have focussedsamplingon
the massive,amygdale-freeinteriorzonesof the flows, our
duplicate study has shown that concentrations of some
elementsmay vary with both samplesize and samplesite
within suchzones. Our studyalsoincludedmultiplesampling
(12 samples)of boththemassiveandamygdaloidalpartsof a
singleflow. We haveusedresultsof thesesecondarystudies
andthe assessment of elementmobilityin Karoobasaltsmade
by Marsh and Eales [1984] to assistus in determiningthe
possiblestratigraphicsignificanceof geochemicalvariations
betweenadjacentflows.
3. The recognitionof a givenunit in itscorrectstratigraphic
positionin severalsections.Stratigraphic subdivisiononlyhas
MARSH ET AL. 251
meaningif individual units are persistentover large areas.
Thus, the identification of a specific unit in two or more
sections,in its correct relative position, is important in
confirmingthe validity of the subdivisionø
Figures 5a-c illustrate variation of some compositional
parameterswith heightat four widely spacedsections.These
diagramsillustratethe repetitionof geochemicalvariationsin
several sections and •h
.........•,.... •s•en,•e
• • of groupsof flows-':'•
•,L,,
distinctivecompositions at more or lessthe samestratigraphic
position through all sections. This is emphasizedby the
dashedlines in Figure 5. We have employeddiagramslike
thesetogetherwith more conventionalvariationdiagramsto
subdividethe successionto producea geochemicalstratig-
raphy. The proposedsubdivisionof the volcanicsequencein
Lesotho is illustrated in Figure 6, showing stratigraphic
columnsfor each of the major sectionssampledby us.
Also shown in Figure 6 is the position of the palaeo-
magneticreversalfirst establishedby Van Zijl et al. [ 1962] at
Sani Pass and the Bushman's-Motimo Nthuze-Thaba Putsoa
sequenceof passes east of Maseru. We confirmed the
reversal's location in Sani Pass, and a French team from
UniversiteScienceset Techniquesdu Languedoc(M. Prevot,
personal communication,1993) has confirmedthe reversal
and transition at Bushman's Pass. The positions of the
reversalat Oxbow, SaniPass,Ongeluksnek,Mafika Lisiu, and
Ben McDhui are from our current study. The reversal
provides an independentcheck on the validity of the geo-
chemicalstratigraphy,andthis aspectwill be discussedlater.
The mostimportantfeatureof Figure6 is the subdivisionof
the main sequenceof lava flows. Following Lock et al.
[1974], Marsh and Eales [1984] referredinformally to this
main sequenceas the Lesotho Formation and the basalt
formingtheseflows asthe Lesothotype. We suggestthatthe
nameLesothoFormationbe retainedbut that it be recognised
that the formation is built of a numberof basalticmagma
types. We proposeto abandonthe term 'Lesothomagma
type' in favourof the new namesproposedhere. Similarly, it
is convenient to group the lowermost units of diverse
geochemicalcharacterinto a single stratigraphicunit for
which we proposethe nameBarkly EastFormation.
Below we describethe essentialfeaturesof the geochemical
stratigraphyandjustify the definition of the differentunits.
Most of our data are from sections concentrated in central and
northern Lesotho, and we found it difficult to correlate the
lowerpart of the succession
in thesenorthernsectionswith the
lower part of the sequencein the two southernsectionsat
Ongeluksnekand Ben McDhui. Thus, for convenience,the
stratigraphyof the northernsectionsis discussed
separately
from that of the southern sections and correlations are drawn
where appropriate.It is alsoconvenientto discussthe lower
units (Barkly East Formation), up to and including
252 STRATIGRAPHY AND AGE OF KAROO BASALTS OF LESOTHO

TABLE 1. Selected
Analyses
of BasaltsfromUnitsof theBarklyEastFormation
Moshesh's Ford Golden Gate Sani
BMC-1 BMC-1 ON-1 ROM-10 BUS-2 BUS-2 MI,P-186 ROM-8 SP-47 SP-47

LAB** WSU RU WSU WSU WSU RU WSU WSU WSU RU

SiO2 52.62 51.47 51.22 52.56 53.52 51.79 52.98 52.10 50.73 50.12
TiO2 15.92 15.47 16.13 15.07 14.87 14.88 13.65 14.60 14.94 14.73
A120• 1.01 0.99 0.99 1.00 1.07 1.03 0.97 0.97 1.02 1.01
FeO* 9.44 9.21 9.18 10.08 9.30 9.93 10.86 9.85 9.36 9.53
MnO 0.16 0.16 0.16 0.17 0.17 0.17 0.17 0.17 0.16 0.18
MgO 10.16 9.92 10.64 10.23 10.06 10.08 8.83 10.20 11.46 11.23
CaO 6.13 6.14 6.26 6.79 5.86 6.54 7.61 8.61 6.85 6.56
Na20 1.03 0.99 0.38 0.69 0.71 0.65 0.92 0.62 0.53 0.46
K20 2.64 2.33 3.40 2.41 2.32 2.32 2.40 2.13 2.19 2.06
P2Os 0.196 0.191 0.166 0.132 0.141 0.143 0.146 0.155 0.167 0.159
LOI 2.46 2.21 2.59
H20' 0.90 0.72 1.20

TOTAL 99.30 100.21 98.53 99.14 98.02 100.46 98.54 99.40 97.41 99.83

Ni 57 68 86 96 89 115 95 108 80 96
Cr 323 290 332 276 239 238 190 434 399 388
Sc 36 26.9 33 31 25 27.9 32 34 34 35.2
V 251 212 251 263 256 251 226 239 253 254
Ba 230 270 144 200 191 229 281 141 137 165
Rb 21 21 9 16 16 15 24 15 12 9.3
Sr 276 278 299 158 159 162 198 135 158 157
Zr 137 132 124 122 125 123 134 93 100 95
Y 27 27.4 26 27 30 30.8 27 25 24 25.7
Nb 18 20 16 4.9 4.2 3.6 7.0 4.5 3.9 2.4
Ga 19 16 21 21 23 17 18
Cu 86 68 92 93 101 102 50 87 91 96
Zn 82 81 76 99 94 87 99 82 84 86

87Sr/•6Sr 0.70594 0.70916 0.70877 0.70978 0.70877 0.70718

(S7Srff6Sr)o 0.70542 0.70840 0.70808 0.70890 0.70801 0.70674
** - analyses
fromWashington
StateUniversity
(WSU) andRhodesUniversity(RU)
* - all Fe as FeO

Wonderkopunit,separatelyfromtheoverlyingmainsequence of two or morestackedtrends(Figure7a). Thus, anyplot of

units(LesothoFormation). incompatibleelementsinvolvingP, Nb, Ta, U, andSr defines
two or more parallel to subparallelor convergingtrends
5.1. The Northern Sections (Figures8, 9). Within thesetrends,samplesfrom different
stratigraphichorizonsform distinctclusters. Someof the
5.1.1. The Barkly East Formation.Selectedanalysesof trendsdo not projecttowardsthe origin of the graphs,and
samples from the basalticunitsof thisformationaregivenin henceincompatible-elementratios,especiallythoseinvolving
Table 1. Becausethe WSU data setis more complete,these the five elements identified above, should be effective
data are used in variation diagrams illustrating the discriminatorsbetween groups of samplesfrom different
composition of theseunits. WSU data,togetherwith selected stratigraphichorizons. Figure 10 showsa numberof ratio-
duplicateanalysesfromthe RU dataset, havebeenusedto ratio plots where geochemical discriminationbetween
compileTable 1. Plotsof Mg# vs incompatible element differentunitsis clearlyillustrated.
concentrations for samplesfrom the lowerunits(Figure7) In all thesesectionsthe basalunit is the Golden Gate unit,
showstrongnegativecorrelationindicativeof considerable first recognizedby Marsh [1984] in outliersof the main
differentiation.For all elementsexceptP, Nb, Ta, Sr, andU, basalticlava pile in the northeastern Free State(previously
the datafor all unitsaretightlyclustered(Figure7 b andc). OrangeFree State). Within the geochemicalcontextof the
However,for thesefive elements, variationswithMg# consist basaltsfrom Lesotho,GoldenGate basaltsare enrichedin
 MARSH ET AL. 253

TABLE 1. (continued)
Roma Letde Wonderkop
BUS-4 ROM-3 OXB-63 OXB-63 OXB-60 SP-45 MLP-183 MLP-183 ROM-11 ROM-11
WSU WSU WSU RU WSU WSU WSU RU WSU RU

50.49 51.69 51.05 49.11 51.38 51.64 53.23 52.19 52.58 50.43
13.85 14.95 12.98 12.65 15.46 15.12 15.99 15.54 15.52 15.27
0.85 0.87 0.81 0.85 0.85 1.16 0.94 1.05 0.95 0.91
9.97 9.99 10.22 10.9 8.60 9.81 8.17 9.41 9.26 8.12
0.17 0.20 0.19 0.17 0.16 0.15 0.14 0.15 0.17 0.17
10.43 11.12 8.78 8.49 10.72 10.98 10.56 10.31 10.85 10.73
7.73 7.79 12.9 12.01 6.39 6.56 6.90 5.71 6.85 6.95
0.56 0.24 0.46 0.44 0.50 0.65 0.71 0.84 0.65 0.60
1.94 2.08 1.70 1.62 2.76 2.51 2.33 2.45 2.36 1.95
0.096 0.100 0.105 0.105 0.123 0.185 0.153 0.192 0.146 0.146
2.97 1.92 2.36
0.58 0.91 1.57

96.08 99.02 99.20 99.90 96.94 98.76 99.12 100.66 99.34 99.21

114 97 247 259 74 81 76 82 83 100

468 445 901 805 298 314 363 279 350 324
34 36 35 27.9 40 38 37 30.5 39 32.8
251 256 231 215 239 272 242 229 256 230
108 92 88 136 107 143 125 230 126 151
15 7 11 12 10 10 13 17 12 12
116 129 100 107 191 224 222 239 213 217
83 84 74 74 77 108 105 122 99 89
25 26 21 23 23 26 24 28.9 23 23.4
3.3 3.0 3.6 4.0 8.3 10.5 13 16 10.5 9.9
19 20 15 18 21 17 19
96 96 47 69 87 105 89 85 101 81
80 79 85 89 72 82 74 76 110 77

0.70994 0.70887 0.70609 0.70588

0.70893 0.70805 0.70556 0.70547

Hf, Pb, La, Th relativeto Nb, Ta, heavyREE, P, Y, U, andTi,
leadingto characteristicallyhigh Zr/Nb, Zr/Y, La/Yb, Th/Nb
andLa/Nb andlow Ti/Zr, P/Zr, andU/Pb (Figures9 and 10).
Our new datashowthatthe flowsin the upperpart of the unit
havehigherincompatible-element abundancesandLa/Yb and
Ti/Zr comparedto the lower flowsandthisformsthe basisfor
an upper and lower subdivision. This distinctioncannotbe
made for otherratios, and collectivelyall flows within this
unithavesimilarcompositions whichareeasilydistinguished
from all other units. The extent of the Golden Gate unit to the
south is unknown and awaits results of further studies. To the
northandwest,in the basaltoutliersof the Free State,thisunit
thinsandis overstepped by unitshigherin the sequence.
Overlying the Golden Gate flows at Bushman'sPassand
Romais the Romaunit (Figure6). It showsthe samerelative
enrichmentamongstincompatibleelementsas the Golden
Gate flows exceptfor distinctivelyhigh Th/Ta and Ti/Zr, and
low Zr/Y and P/Zr (Figure 10). It appearsto havea limited
distribution,
nothavingbeenfoundin sectionsto the northof
Bushman's Pass, and its extent to the southwestis unknown.
The Sani unit Overlies the Golden Gate unit at Sani Pass.
A single sample, ROM-8, from the Roma sectionhas also
been grouped into this unit. The Sani unit is similar in
compositionto the Golden Gate basalts but is clearly
distinguishedfrom them by high P/Zr andTi/Zr (Figure 10).
The extent of this unit along the easternescarpmentis still
unknown and more sections to the north and south of Sani
Passneedto be sampledøThe correlationof ROM-8 with the
Sani unit suggeststhat this unit may extendeast-westacross
Lesotho.AlthoughROM-8 sharesmanygeochemicalfeatures
with the Sani Passsamples,it hasmuchhigherTh relativeto
other incompatibleelements(Figure 9) and it is worrisome
that if the Sani unit is extensiveenoughto reach east-west
acrossLesotho,it was not found in the neighbouringsection
at Bushman'sPass(Figure 6). However, on the evidenceof
ROM-8, this unit is very thin in the west andmay havebeen
missedduringsampling.
The Letele unit, which is characterizedby high Ti/Zr
254 STRATIGRAPHY AND AGE OF KAROO BASALTSOF LESOTHO

TABLE2. Selected
analyses
(RU data)ofbasalts
fromtheMainSequence
(Lesotho
Formation)
Mafhka Lisiu Mafhka Lisiu - low Zr/Iqb
MLP-155 MLP-172 OX]343 8P-32 SP-36 ON-35 MLP-02 MLP-166 8OM-92 BMC-07

SiO• 50.22 50.69 49.94 49.60 50.04 50.13 49.74 50.79 49.66 50.08
TiO• 0.97 0.98 0.88 0.92 1.07 0.88 0.99 0.96 0.84 0.90
AI•O• 15.54 15.38 15.04 14.70 14.92 15.01 14.87 14.83 15.17 15.11
FeO* 9.83 9.60 9.44 9.71 10.04 9.76 9.38 9.79 9.01 8.49
MnO 0.17 0.17 0.16 0.17 0.18 0.17 0.16 0.18 0.15 0.15
MgO 7.36 7.01 7.29 7.48 6.59 8.56 7.33 7.87 6.39 6.33
CaO 10.21 10.96 10.34 10.61 10.53 10.78 10.15 10.32 9.86 10.79
Na•O 2.06 2.30 2.05 1.91 2.21 2.21 2.01 2.40 2.26 2.25
K•O 0.22 0.51 0.38 0.58 0.65 0.44 0.69 0.70 1.03 0.55
P•Os 0.169 0.163 0.138 0.171 0.199 0.147 0.195 0.181 0.165 0.163
LOI 3.14 1.56 2.81 3.00 2.51 1.70 2.77 1.79 3.22 3.73
0.37 1.13 1.72 1.65 0.89 0.69 1.81 0.71 1.90 1.17
TOTAL 100.25 100.45 100.19 100.50 99.83 100.47 100.08 100.53 99.65 99.69
Ba 130 192 155 165 217 172 175 227 220 203
Sc 30.6 30.8 33.7 30.4 29.5 30.4 29.9 29.6 30.9 34.6
85 79 81 76 76 73 87 75 82 82
Cu 94 87 74 90 65 80 78 120 75 84
Ni 97 87 92 80 79 107 75 116 79 96
6.2 6.4 4.2 8.9 10.6 5.3 8.5 11.2 9.3 8.5
Zr 85 75 76 82 88 69 84 95 81 74
Y 25.5 23.2 23.5 24.2 25.1 22.7 25.6 24.7 24.2 23.9
Sr 251 222 176 198 231 202 202 208 187 186
Rb 2.1 10.8 8.8 10.2 12.9 8.9 10 14.2 34 9.2
Co 44 43 44 45 44 48 45 46 41 44
Cr 214 301 345 266 289 458 195 389 274 304
V 219 245 226 236 231 231 283 225 218 252
Ce 25 24 22 29 24 19 28 31 27 24
Nd 13 13 11 13 14 10 13 15 12 12
La 11 10 8 11 13 9 12 13 11 10
S7SrfføSr 0.70657 0.70551 0.70572 0.70544 0.70529
(•7Sr•Sr)o 0.70611 0.70515 0.70534 0.70502 0.70497

Maloti LowerSenclu
MLP-148 MLP-17 OX]3-40 BMC-14 ON-04 BUS-31 BUS-40 ON-07 BMC-25 SP-13 MLP-31 OXB-21

SiO2 49.82 50.60 50.72 49.58 50.61 51.63 49.86 49.38 49.20 49.64 47.42 49.77
TiO2 0.89 0.99 0.89 0.92 1.07 0.87 1 1.07 0.92 0.94 0.93 1.03
AI20• 14.81 14.81 14.55 15.46 14.92 14.51 15.4 14.35 15.17 15.21 15.73 14.78
FeO* 9.75 10.34 9.79 9.64 10.60 9.84 10.22 10.92 10.05 10.22 9.38 10.42
MnO 0.17 0.18 0.16 0.16 0.19 0.16 0.19 0.18 0.18 0.18 0.16 0.18
MgO 7.17 6.89 6.91 7.22 5.89 6.75 6.74 6.42 6.44 6.61 6.28 6.22
CaO 10.6 10.36 10.64 10.47 10.40 10.90 10.81 10.28 10.24 10.45 10.87 10.22
N•O 1.95 2.27 2.04 1.92 2.48 2.11 2.09 1.94 2.13 2.11 1.78 2.10
K20 0.62 0.64 0.54 0.45 0.64 0.37 0.54 0.56 0.65 0.56 0.46 0.80
P2()• 0.160 0.171 0.156 0.161 0.180 0.161 0.180 0.190 0.151 0.153 0.164 0.162
LOI 2.56 1.96 2.26 2.27 1.75 2.47 2.43 2.65 3.48 2.78 5.23 2.97
H20' 2.17 1.14 1.07 1.52 0.89 0.84 1.5 1.71 1.49 1.07 0.92 1.21
TOTAl. 100.67 100.33 99.73 99.78 99.62 100.61 100.96 99.65 100.09 99.92 99.31 99.86
Ba 208 225 213 192 227 190 187 200 164 195 157 245
Sc 32.6 32.2 32.8 32.5 33.7 33.8 32.3 33.7 38.5 31.3 31.9 32.9
• 84 88 89 86 85 82 84 93 85 82 90 85
Cu 80 88 77 87 96 89 96 101 113 54 94 85
Ni 83 82 84 92 64 93 75 68 73 79 76 67
Nb 6.8 6.7 8.7 6.8 8.2 8.0 7.1 7.8 4.9 5.3 6.1 4.8
Zr 90 99 92 83 105 93 87 98.2 80 85 87 99
Y 25.4 27.6 21.9 24.7 29.5 25.6 25.5 28 23.7 25.2 24.9 26.7
Sr 177 191 193 199 196 198 200 194 206 184 167 182
Rb 11.5 13.7 6.9 4.9 14 3.4 7.2 8.2 9.9 12 10.7 18
Co 43 44 44 46 42 46 44 44 42 45 44 43
Cr 220 194 268 209 179 279 177 151 256 193 132 223
V 235 231 227 226 246 234 248 252 273 240 243 244
Ce 30 32 25 28 31 31 26 28 21 22 29 30
Nd 14 15 14 13 16 15 15 14 11 13 13 16
La 12 12 14 9 13 12 11 13 8 12 9 13
878rff6Sr 0.70625 0.70615 0.70580
(87Sr•Sr)o 0.70579 0.70589
 MARSH ET AL. 255

TABLE 2. (continued)
Mothae Oxbow
UpperSenqu
MLP-115 MLP-39 OXB-11 BMC-36 SOM-04 SP-01 OXB-02 OXB-09 MLP-61 MLP-65 OXB-22 OXB-14

SiO2 51.09 48.61 50.98 49.25 49.73 49.56 50.67 48.57 47.59 50.14 50.37 50.17
TiO2 1.02 0.99 0.96 0.96 0.99 1.06 1.46 0.76 1.20 1.31 1.55 1.55
AI20• 14.87 14.15 14.93 14.75 14.88 14.11 13.43 11.52 14.51 14.02 13.55 13.54
FeO* 10.37 10.24 10.23 9.76 10.00 10.75 12.55 11.82 11.18 11.78 13.52 13.56
MnO 0.18 0.18 0.16 0.19 0.18 0.23 0.21 0.20 0.19 0.21 0.23 0.22
MgO 6.24 7.61 6.32 7.08 6.34 6.32 5.21 11.65 5.58 5.81 5.51 5.28
CaO 10.02 10.60 10.60 10.60 10.26 10.27 9.66 9.90 10.01 10.01 10.11 10.26
Na20 1.95 1.87 2.12 1.99 2.13 1.90 2.23 1.71 2.16 2.30 2.46 2.63
K20 0.67 0.65 0.52 0.64 1.19 0.68 0.82 0.60 0.63 0.69 0.52 0.28
P20• 0.170 0.168 0.152 0.164 0.171 0.178 0.234 0.145 0.197 0.214 0.186 0.191
LOI 2.40 2.66 2.66 2.95 2.97 3.57 2.27 2.05 5.23 2.37 1.94 2.10
H20' 1.9 1.92 1.05 1.62 1.02 1.38 1.28 1.08 1.53 1.28 0.44 0.47
TOTAL 101.09 99.64 100.68 99.94 99.86 100.01 100.024 100.01 100.01 100.15 100.386 100.25
Ba 201 224 185 213 247 196 279 169 200 260 190 172
Sc 34.6 34.6 34.3 34.8 34.6 35.9 35.6 37.1 38.6 35.5 34.2 33.3
Zn 87 90 93 79 79 79 97 87 99 93 106 116
Cu 102 103 101 90 99 91 150 69 110 121 210 212
Ni 55 76 73 70 62 54 52 178 57 60 62 63
Nb 6.8 6.8 7.3 5.3 6.6 8 9.1 4.7 8.6 7.5 6 6.3
Zr 97 88 91 86 92 96 132 69 109 114 116 120
5( 26.7 25.9 24.8 26.6 25.6 27.7 34 20.7 30.4 32.9 32 32
Sr 172 165 202 179 188 174 187 139 111 184 182 184
Rb 12.3 11.6 7.7 10 29 14 21 12 11.6 11.9 13.7 11.3
Co 42 46 47 46 41 41 42 70 42 43 48 48
Cr 184 298 238 244 244 235 128 427 166 158 95 92
V 259 257 233 255 249 264 278 244 286 279 356 356
Ce 28 28 24 24 30 29 38 25 34 32 31 26
Nd 16 13 14 13 14 15 19 11 16 19 16 16
La 14 11 11 13 11 13 18 10 12 14 13 11
s7Sr
ff6ar 0.70680 0.70679 0.70666 0.70604 0.70636
(878yff6Sr)o 0.70627 0.70596 0.70618 0.70548 0.70591
* - All Fe as FeO

Sample
Sections:
MLP - Marfica
Lisiu;BMC - BenMcDhui;ON - Ongeluksnek;
SOM - Semongkong
SP - SamPass; OXB - Oxbow; BUS - Bushman'sPass.

Th/Ta, moderateP/Zr andZr/Nb, andlow Zr/Y (Figures9 and from the LesothoFormationare in Table 2. Theseanalyses
10) is foundfrom the LetelePassnorthwardsinto the basalt andthosein all the pertinentvariationdiagramsarefromthe
outliersof the northeasternFree State. Overlying all these RU dataset. Fourunitsbuildthemain sequence. One of these
typesis the Wonderkopunit (Figure6) which,in contrast to the Maloti unit, is a thin (maximum:120 m thick), distinctive
underlyingunits,is characterizedby highNb and,to a lesser group of flows occurringat about2400 m in all sections,
extent, P, U, and Th relative to most other incompatible exceptat SaniPasswherea generalupwardwarpin thebasalt
elements(Figures9 and 10). It is widespreadthroughout sequenceresultsin this unit croppingout at about2900 m.
northernandcentralLesotho,beingpresentin all the northern The Maloti unit effectivelysubdivides the LesothoFormation
sections,andit may havecorrelativesin the southernsections, intotwo. It marksa stageof distinctcompositional changein
as discussed later. In the basalt outliers of the Free State it the evolutionof the main sequencebasaltsand is discussed
oversteps underlyingunitsto formthebasalflowson Clarens first in the sectionsthatfollow. Criticalcompositional datafor
sandstones at WonderkopandClocolan(Figure3). the unitsin the LesothoFormationare summarised in Figures
In summary,the overall geochemicalstratigraphyin the 11-13. Figure11 is a composite diagramconstructed fromall
lower units of the northern sections is one of a series of units the sectionsusing the altitudeof the lowestsampleof the
(GoldenGate,Sani,Roma,andLeteleunits)characterized by Maloti unit as a datum(i.e., for eachsection,the altitudeof
low Nb relativeto Zr andotherincompatible elements.These eachsampleis normalised to the altitudeof the lowestMaloti
units have variable relative enrichmentsof P, Ti, Y and Zr, sample). Thus, the stratigraphyin Figure 6 is directly
and ratios amongst these elements serve to distinguish comparable to Figure11. The overlapin the symbolsof the
betweenthesehighZr/Nb units.Thereis thena shiftto flows different units in Figure 11 is a functionof the different
with higher Nb (and consequentlylow Zr/Nb) which thicknesses for some of the units,in particularthe overlapin
constitutethe overlyingandwidespreadWonderkoptype. the altitude of the Mothae unit at Oxbow and Mafika Lisiu
5.1.2. TheLesothoFormation. Selectedanalysesof basalts and the Senqu unit at Semongkongand Ben
256 STRATIGRAPHY AND AGE OF KAROO BASALTS OF LESOTHO

TABLE 3. Selected
Analysesof Low-Ti andHigh Ti Basaltsfromthe Springbok
Flats
Unit 1 Unit 2 Unit3
WD4-6 WD4-8 WD4-14 LB1-2 LB1-3 RTL1-46 RTL1-39 RLl-17 RLI-11 LBl-12 RL1-24

SiO2 50.12 50.32 50.73 48.95 50.29 50.57 48.16 47.32 49.79 48.54 48.46
TiO2 0.87 0.95 0.90 0.82 0.87 0.91 0.86 0.93 0.96 0.94 1.00
A12Os 14.55 14.45 14.86 14.31 14.78 14.18 14.53 15.12 15.07 14.75 15.18
FeO* 8.95 9.43 9.14 8.97 9.60 9.65 9.36 9.44 9.14 9.92 10.20
MnO 0.19 0.16 0.15 0.15 0.15 0.15 0.17 0.19 0.15 0.16 0.17
MgO 7.48 7.17 7.19 7.61 8.31 8.25 7.52 7.54 6.49 6.12 6.38
CaO 9.61 7.30 9.94 7.86 9.72 9.69 9.25 9.50 9.91 10.33 10.05
Na20 2.09 2.84 2.19 2.44 1.91 2.18 2.05 1.99 1.63 2.16 2.04
K20 0.71 1.06 0.76 0.99 0.75 0.63 0.45 0.58 0.72 0.65 0.75
P2Os 0.110 0.120 0.130 0.100 0.117 0.113 0.143 0.147 0.110 0.160 0.155
LOI 3.60 4.82 3.29 5.34 2.24 3.05 5.05 3.95 3.76 4.54 2.89
H20' 1.89 0.82 1.41 2.66 1.86 1.07 2.09 2.19 2.01 2.13 2.05
TOTAL 100.17 99.44 100.69 100.20 100.60 100.44 99.64 98.89 99.74 100.40 99.32

Ba 238 467 210 371 205 176 149 194 115 190 243
Sc 28 36 26 28 29 27 31 30 34 31 37
Zn 69 79 76 78 81 65 77 80 73 73 88
Cu 70 76 72 62 89 56 61 59 76 67 72
Ni 102 86 96 102 124 120 84 93 97 57 61
Nb 3.7 4.3 5.0 3.8 3.1 2.6 3.5 3.1 3.0 6.1 6.8
Zr 98 111 107 103 101 86 68 67 61 88 93
Y 25.4 30.4 27.7 26.2 26.4 25.0 24.0 22.8 22.2 26.4 25.6
Sr 135 225 143 213 143 142 175 152 174 209 176
Rb 16 31 19 32 17 15 7.8 8.3 15 12 16
Co 53 51 49 51 54 58 53 48 47 51 47
Cr 386 407 360 374 463 535 307 357 411 230 246
V 224 256 205 216 213 195 237 222 240 220 264

0.71228 0.71467 0.70988 0.71762 0.71220 0.70922 0.70561 0.70569 0.70617 0.70673
0.71140 0.71365 0.70890 0.71651 0.71132 0.70844 0.70528 0.70529 0.70575 0.70606
* - All Fe as FeO
RL1, LB1, WD4, andRTL1 referto boreholes
in Figure18

It is characterizedparticularlyby low Ti/Zr andto a much

lesserextentby low P/Zr comparedto mostotherLesotho

;':'
O OO O
Formationbasalts(Figures11 and 13). Thebaseof theunitis
markedby an abruptdecrease
andby a sharpincrease
in thesetwo ratiosandZr/Nb
in Zr/Y (Figure12). Samples fromthe
øo upperpart of thisunit may showTi/Zr andP/Zr ratiosthat
appeartransitional betweenvaluestypicalof theMalotiunit
2000 '-.:. . • and the overlying flows of the Senqu unit. Whethera
persistent,butthin,transitional unitseparatingtheMalotiand
Senquunitscanbeunequivocally recognized mustawaitmore
1500H' I, I, I ,I,,,,I50....60I ....70I,
80 100 120 140 50 60 70 80 5 6 7 8 9 10 detailed sampling. However, the possiblepresenceof a
transitionalunit doesnot hamperclearrecognitionof thetop
Zr Mg# Ti/Zr P/Zr
of the Maloti unit, and in this studythe transitionalsamples
Figure4. Variationof Zr andsomeinterelement
ratioswith height havebeenincorporated intothe Maloti unit.
in the basalts of the Oxbow section. Dots - lava flows; circles - 5.1.2.2. The Mafika Lisiu unit: Betweenthe Maloti and
intrusions.
Wonderkopunitsis a sequence of flows,about400 m thick,
of rather variablecomposition.Thesevariationsmay be a
Nevertheless,theseslightoverlapsdo not distractfrom the basisfor futhersubdivision of thisunit andthey are discussed
effectiveness
of Figure11 in summarising theimportantserial below.

geochemicalfeaturesof all sections

in a few simplediagrams. About 150 m below the Maloti unit is a thin unit
5.1.2.1. characterized
The Maloti Unit: This unit has constant thickness by distinctlylowZr/Nb (andhighNb) compared
including to otherMaf'ficaLisiu flows,as is apparentfromthe RU data
of a little over 100 m andis presentin all sections,
the two southernsectionsat Ongeluksnek andBen McDhui. in Figures11 and12. Thisgroupof samples is also
 MARSH ET AL. 257

TABLE 3. (continued)
Unit3 Unit4 Hil•h-Ti
RL1-27 RTL1-05 RL1-36 RL1-43 RL1-59 RL1-60 RTL1-21 RTL1-24

47.49 49.04 49.75 49.96 48.01 49.01 51.28 46.80

1.28 1.19 1.27 1.27 1.60 1.52 3.09 3.16
15.00 14.10 14.16 14.42 13.75 13.50 14.33 14.75
11.72 11.71 11.86 11.93 12.98 12.67 10.21 10.53
0.20 0.18 0.18 0.19 0.22 0.20 0.13 0.13
5.57 6.06 5.82 5.77 6.00 5.83 3.36 3.96
10.38 10.17 10.17 10.30 10.25 9.70 7.57 8.51
2.09 2.18 2.01 2.18 2.00 1.92 2.44 2.12
0.76 0.64 0.52 0.49 0.62 0.96 2.32 1.91
0.202 0.140 0.153 0.155 0.222 0.225 0.620 0.640
2.60 2.82 2.22 1.83 2.65 2.68 3.16 4.87
1.49 1.73 1.79 1.65 1.89 1.73 1.38 2.56

98.78 99.96 99.90 100.15 100.19 99.94 99.89 99.94

258 105 182 179 222 220 1049 1034

35 38 40 40 40 43 16 16
88 92 92 91 109 108 93 96
102 155 136 131 217 175 80 123
53 81 66 63 60 53 53 55
7.0 4.2 4.9 5.0 5.7 6.9 32 33
116 89 100 99 120 126 455 460
32.0 27.9 31 32 39 41 44 46
180 158 162 170 149 144 1158 1162
12 20 7.7 5.8 10 30 45 20
47 51 50 49 53 50 44 47
174 284 218 197 154 167 69 72
281 318 301 310 358 322 203 228

0.70644 0.70642 0.70597 0.70585 0.70546 0.70635 0.70534 0.70509

0.70595 0.70548 0.70562 0.70560 0.70496 0.70481 0.70505 0.70496

in theWSU data,but theirseparation from otherflowswithin thebasaltsequence.Althoughdataaresparse,the low-Zr/Nb

thisunit is lessperfect. Despitetheir distinctiveandconstant flows also separateMafika Lisiu basaltswith different Sr-
Zr/Nb ratios, this group has highly variable incompatible- isotopiccompositions.Theseaspectsarediscussed in more
elementratios includingP/Zr, Ti/Zr, and Zr/Y, and this is detail below.
importantfor the possiblecorrelationdiscussed in Section5.2. In the lower part of the Mafika Lisiu unit, a groupof
Figure 6 indicatesthat this low-Zr/Nb group lies strati- sampleswith lower P/Zr, and to a lesserextent Ti/Zr, than
graphicallyjust below the Maloti unit in the Bushman'sPass other Maf'•a Lisiu basalts occurs within the interval 50 to 100
section,unlikeat othersectionswherethe verticalseparation m abovethebaseof thisunit. Thisgroupcouldformthebasis
is muchlarger. At Bushman'sPassthe low-Zr/Nb groupis for the separationof theseflowsinto a separate basalunit to
represented by a singlesample,BUS-22 from the summitof the LesothoFormationoverlyingthe Wonderkopflows.
Bushman'sPass. Samplecollectionwasresumedsome9 km However, the compositionalrange of this putativeunit as
to the east on the Molimo-Nthuse/Thaba Putsoa Pass at the expressed by inter-element ratios,apartfromP/Zr perhaps,is
samealtitudeasthe summitof Bushman'sPass and,within a large and overlapswith data for sampleshigher in the
shortinterval,encountered the Maloti unit. In compiling sequence (Figure11). In thisregardtheapparentlydistinctive
Figure 6, no correctionwas made for the known slight P/Zr forthisgroupis emphasized in theWSU databy a small
eastwarddip of the lava sequence.A 1odip wouldresultin a group of samplesin the Mafika Lisiu section. This is not a
150-m differencein elevationbetweenthe two segments of featureof the RU data. In the absenceof clearcut, distinctive
the sectionandwouldcorrecttheapparent compression of the geochemical criteriaforthisgroupits statusasa separate unit
upperpart of the Mafika Lisiu unit in the Bushman'ssection relieson stratigraphy; i.e., samplesfromthisunithaveratios
in Figure6. In the Oxbow section,a singlenormalMafika thataredifferentfromsamples immediatelyaboveandbelow
Lisiu basaltflow is interbeddedwith the low-Zr/Nb flows, but but overlapwith thoseof sampleshigherand lower in the
in all othernorthernsectionsthe low-Zr/Nb flows appearto sequence.Thus,we includethesesamplesin theMaf'•a Lisiu
form a homogeneous of the low-Zr/Nb unitalthough
unit. The significance theyareidentifiedasseparate symbolsin several
horizonis enhancedby the possibilitythat it is closelyand of the diagrams.
consistently relatedto thepalaeomagnetic polarityreversalin 5.1.2.3 The$enqu Unit: This unit overliesthe Maloti
258 STRATIGRAPHY AND AGE OF KAROO BASALTS OF LESOTHO

unit. However,the compositions of the basaltsin thesetwo

subunitsoverlapcompletelyandtheycannotbe distinguished
independentlyof their stratigraphic
context.Formalproposals
for further subdivision of this unit must await more detailed
studies.
5.1.2.4. The Mothae unit: The uppermostbasaltflows in
themain sequenceshowmarkedgeochemical trendstowards
quite evolved compositionsover a short stratigraphic
interval. Theseevolvedbasalts,exhibitinggenerallylower
CaO, MgO, A1203 and higher FeO, TiO ,2 P • sand
I
incompatibletrace-elementabundances, havebeengrouped
0.8 1.0 1.2 1.4 1.6 1.0 1.2 1.4 0.8 1.0 1.2 1.4 0.8 1.0 1.2
into a separateunit, the Mothaeunit (Figures6, 11, and 13).
Oxbow Mafika Lisiu SaniPass Ongeluksnek The baseof thisunit is markedby abruptinflectionsin several
interelementratios and changes in element abundances
.I I I P/Zr (Figures5 and 11) in the Oxbow section,but in the Mafika
Lisiu sectionthe baseis markedby an inflectionin P/Zr only,
withothercompositional parameters
varyingcontinuously and
smoothlyacrossthisboundary.Thus,althoughincompatible-
elementratios do not distinguishthis unit, Mg# and TiO2
allow almost complete separationof these basalts from
0.5 underlyingunits(Figures11 and 13). We emphasizethatthe

.• • ?nx•
Vv abruptchangesobservedin the Oxbow sectionmay be an

4 5 6 7 8 9 10 5 6 7 8 9 10 6 8 10 6 ? 8 9 [•OXB MLP BUSSOMSP ON

1.5 W
Oxbow Mafika Lisiu SaniPass Ongeluksnek E

1.5

1.o
• I • I I Ti/Zr 1.0

2410 2330 2395 2275 2765 2480 2330

5455
0.:5
=R -R •R
• • • 9 V
Roma --
•L
iø•o•r?b
30 40 50 60 708090405060708090 50 60 70 80 50 60 70 80

BARKLY EAST F.
LESOTHO F.
Figure 5. Vertical variationsof somegeochemicalparametersin Wonderkop
Vertical
four sectionsthroughthe Lesothobasaltsequence.Dashedlines Scale in km
Mothae Letele

showthe correlationsof differentbasaltunitsrecognized

on the basis upperSenqu Roma

lower Senqu Sani

of geochemistry. Maloti Golden Gate
Mafika Lisiu Moshesh's Ford

and, like the Mafika Lisiu unit, has rather variable

compositions whichtendtowardsbeingslightlymoreevolved Figure 6. Summaryof the stratigraphicsubdivisionsfor each
in termsof Mg# andZr abundances comparedto underlying sectionbasedon geochemistry.
The datumis thebaseof the Maloti
unit which occurs in all sections. Abbreviations of sections are OXB
unitsof the LesothoFormation(Figures11-13). Interelement
- Oxbow; MLP - Mafika Lisiu Pass;BUS - Bushman'sPass;SOM -
ratios and Mg# show several abrupt changes in the
Semongkong; SP- SaniPass;ON - OngeluksnekPass;BMC - Ben
succession, andin mostsectionsthe coincidenceof an abrupt McDhui. The position of the palaeomagneticpolarity reversalis
change(of varying magnitude)of the samesensein Ti/Zr, indicatedby thebracketlabeledR. In theON andBMC sections the
P/Zr andMg# at aboutthe samestratigraphic level formsthe positionof thereversalhasnot beenpreciselyfixeddueto the wide
basisfor a possiblesubdivision
intoanupperandlowerSenqu sampling
 MARSH ET AL. 259

2O
presumedto have fed lavasnow lost dueto erosion.

15
Nb
/• Moshesh's
Ford 5.2. The Southern Sections

o?
10
.•/-••o•••ø'o/
Won
derkop
(o) southern
It hasnot beenpossibleto correlatelower unitsin the two
sectionswith thoseof the Barkly eastFormationin
thenorthernsections.As illustratedin Figure14, at the base
of both southern sections is the well-established Moshesh's
Ford unit which is widespreadin the regionaroundBarkly

_,Zr
en
Gate
(•)I-b-]
160
East(Figure 3) and in the basaltoutliersbetweenJamestown
140 and Molteno further to the southwest[Marsh and Eales,
120 1984]. Its discoveryat Ongeluksneksubstantially increases
100
ani,
Sani (x) 20
Nb []
8O
15 _ rh

1.2 ,TiO2/•x
' •C• 10

1.0
- •L.••• ••Letele (a) m I m

0.8
P205
o
0.20 -
45 50 55 60 65 70 75

Mg# 0,15 -

Figure 7. Variationof someincompatible elementswith Mg# in

unitsof the BarklyEastFormation.The symbolsindicatesamples 0.10 -
from different stratigraphicunits defined on the basis of
geochemistryas discussedin the text. 0,05 .... I .... i ....
50 100 150 2t

artefact of poor sampling density in a sequencewhere Zr

compositions arechangingsmoothlyandrapidly,andthebasis
for a non-arbitrarydefinitionof this unit may be confirmed Figure 8. Covariationof someincompatibleelementsin basalts
only by more careful sampling. from the Barkly EastFormation.Note subparallelandconverging
5.1.2.5 Intrusions: The current study was aimed at trends. Symbols as in Figure 7 indicate basaltsfrom different
sampling basalt lava flows in order to determine a stratigraphicunits.
stratigraphicsubdivision for thevolcanicpile. Intrusions
were
only sampledwhen encounteredin the samplingsections. 's Th/Ta
Generally it was possibleto classify the dykes into the L o Roma
10/ ,-x•,'"•toM4
stratigraphic scheme described above on the basis of
geochemistry, and in all casesthe classification
wasconsistent r I•, .•c øø•d•
with stratigraphy;i.e., no intrusionwas foundto belongto a
unit lower in the sequencethan the flows which it intrudes.
Oneexceptionto thiscorrelationof dykeswith flowsis a suite
of smallintrusionscollectedat about2500 m in the vicinity of
5 10 15 20 0 2 4 6 4 5 6 7 8

Nb Pb La/Yb
the Oxbow Lodge on the Oxbow section. These are
chemicallyevolved,with low Mg# anddistinctivelyhigh Fe, Figure 9. REE and other incompatibleelement variations in
Ti/Zr and Zr/Nb (Figures 11 and 13). They cannot be selectedsamplesfrom unitsof the BarklyEastFormation. Symbols
correlatedwith any known lava unit in Lesotho and are as in Figure
260 STRATIGRAPHY AND AGE OF KAROO BASALTS OF LESOTHO

70
Roma

65 _ a• Letele Golden Gate

60 i

Wonderkop
• 55

50 Moshesh's
45 Ford
I I
40
5 10 15 20 25 30 4 5 6 7 8 6 7 8

Zr/Nb P/Zr P/Zr

Figure 10. Plotsof incompatible elementratiosin basaltsfrom unitsof the Barkly EastFormation.Note the distinctive
compositions of the differentunitsmakingtheseplotsa usefuldiscrimination diagram.Symbolsin Figure7.

the knownareaover which this chemicaltypewaseruptedat samplesandthe apparentthicknessof someof theseunits.

the onset of basaltic activity. The northern limit to its In summary,simplecompositionalcorrelations
indicatethat
distributionis unknownbut it is not recognizedfrom any of basalt units which form the Lesotho Formation in northern
the northern sections. The Moshesh's Ford type is Lesothoappearto lie directlyon the Moshesh'sFordunit in
characterizedby its distinctivehigh Nb contentand by its the southern sections. The low-Nb units which form the
combinationof low Zr/Nb and Ti/Zr andhigh Zr/Y. lowermost units in the northern sections are absent.
In the Ben McDhui section, the Moshesh's Ford unit is Furthermore,providedthe correlationscanbe sustained, the
overlainby flowswith low Zr/Nb (- 8, whichis equivalentto datasuggest that flows of the low-Zr/Nb andnormalMafika
the Moshesh'sFord basalts),high andvariableP/Zr (8.5-10) Lisiu unitsinterdigitatein the southernsections,in contrastto
and moderateand constantTi/Zr (- 65). Interbeddedwith their simpler interrelationship in northernLesotho. These
theseis a singleflow (BMC-3) with similarP/Zr and Ti/Zr, conclusions are tentative and could reflect sampling
but higherZr/Nb (-13). Above is a sequenceof flowswith inadequacies.Detailedsamplingof moresections in southern
compositional characteristicstypicalof normalMafika Lisiu, Lesothois neededto establisha reliable stratigraphyfor the
Maloti, and Senqubasalts,in the correctstratigraphic
order,as lower part of the sequencein this area.
recognizedin the northernsections.Theselow-Zr/Nb flows
in the lower part of the southernsequencehave affinitiesto 5.3. Discussion

the Wonderkoptype of the northernsectionson the basisof

Zr/Nb, Zr/Y and to a lesserextent Ti/Zr, but not P/Zr. On the Figure 13 summarises someof the principaldiscriminating
otherhand,the low-Zr/Nb flows correlateperfectlyin terms geochemicalfeaturesof the LesothoFormation,andFigure15
of geochemistrywith the low-Zr/Nb Mafika Lisiu basalts. showsthe compositionsof the LesothoFormationunits in
In the lowerpartof the Ongeluksnek section,a sequence of relationto thoseof the Barkly EastFormation.An important
flows with broadly similar compositionscan also be featureof Figure 13 is the minor overlapin Ti/Zr and Zr/Y
recognized. These flows divide into a low-Zr/Nb group betweenMafika Lisiuflowsandthosehigherin the sequence,
(Zr/Nb - 8) and a high-Zr/Nb group(Zr/Nb- 12-20). The providedthe low-Zr/Nb Mafika Lisiu flows are considered
formercorrelatewith the low-Zr/Nbgroupin theBenMcDhui separately(they are shownwith a separatefilled-diamond
section and hence also with the low-Zr/Nb Mafika Lisiu flows. symbol in Figure 13). This underscores the significanceof
The other high-Zr/Nb flows are similarto the singlehigh- theselow-Zr/Nb flows. The compositional variabilityof the
Zr/Nb flow at Ben McDhui and are all compositionally Mafika Lisiu unit as a whole is alsoevidentin comparison
identicalto normalMafika Lisiu basalts.As at Ben McDhui, with the more compositionally homogeneous Senquunit.
the normalsequence of Mafika Lisiu,Maloti, andSenquunits More detailedstudiesin the Mafika Lisiu sequencecould
is developedin the overlyingsequence.Figure14 summarises result in further subdivisions within the unit. The overall
thesecorrelations,but it shouldbe stressed that the faulting increasingdegreeof fractionationin the sequenceMafika
close to our samplingsectionat Ongeluksnekmay have Lisiu-Maloti-Senqu-Mothae-Oxbowis also evident from
influencedthe relativestratigraphic positionof someof the Figure 13, butthe largechangesin Ti/Zr andZr/Y between
 MARSH ET AL. 261

1.4I Mothae

Loft
,)
v
Oxbow
Dykes

0.8

40 50 60 70 0.6 0.8 1.0 1.2 1.4 1.6 10 15 20 25 30

Mg# TiO 2 Zr/Nb

1.4

-00

0.8

A
.... [ .... [ .... [ .... I ....
40 50 60 70 80 6 7 8 9 10 3 3.5 4

Ti/Zr P/Zr Zr/Y

Figure11. Composite serial

section diagram showingvertical
compositional
variationsintheLesothoFormation. Data
fromallsample sections.Thecalculation ofnormalized
altitudeisexplained
inthetext.Thestratigraphic unitsdefined
onthebasis ofgeochemistryaredistinguished bydifferent
symbols asindicated.
In theMafikaLisiuunitthelow-Zr/Nb
samples areshown assoliddiamonds andthepossiblebasalsubunitashalf-filled
circles.Thestatusofthese unitsis
discussedin thetext. Forclarity,theinterbeddedflowsectionatthebaseof BenMcDhuiandOngeluksnek sectionsis
identifiedas"MixedUnit"(upright triangles)
ontheplot;seetextandFigure14forfull discussionof thestratigraphic
characterizationof these
262 STRATIGRAPHY AND AGE OF KAROO BASALTS OF LESOTHO

140
_ Zr
- Oxbow

120
Maloti
Senqu
Senqu
100

80

low Zr/Nb
Mafika Lisiu
60 Mafika Lisiu

40
-,,,, ,,,, ,,,,I,,,,I,,
5 10 15 0.6 0.8 1.0 1.2 1.4 1.6 0.1 0.15 0.2 0.25

Nb TiO2 P205
Figure 12. Covariationamongstsomeincompatible elementsin thebasaltsof theLesothoFormation.Note: symbolshave
beenchangedfrom thoseusedin Figure 11 as indicated,exceptfor the "mixedunit" samplesfrom Ben McDhui and
Ongeluksnek,which havebeenclassifiedintotheir geochemical typeaccordingto Figure 14. Note the tight clustering
of the dataand the distinctivecharacterof the low-Zr/Nb Mafika Lisiu flows andthe high Zr/Nb Oxbow intrusions.

The diagramspresented arepotentiallyusefuldiscrimination

40
2.5 3 3.5 4 4.5 50 60
Zr/Y
Figure 13. Interelementratio plots for main sequencebasalts.
Symbolsas in Figure 12.
Mafika Lisiu,Maloti, Senqu/Mothae,andOxbowunitsmake
it unlikely that this sequencerepresentsa normal differen-
tiationtrend involving gabbrofractionation.
The relative geochemicalhomogeneityof the Lesotho
Formation units as a group comparedto the diversityin the
Barkly East Formationis emphasizedin Figure 15. Basalt
magma types in the Barkly East Formationare not simply
related by fractionationto thoseof the LesothoFormation.
Interestingly,the datafor the Oxbow intrusionsin the Senqu
unit plot on the fringes or well away from the Lesotho
Formation cluster and suggest a possible return to a
geochemicaldiversityin the late stagesof eruption,similarto
that which characterized the flows in the lower units.
diagramsfor stratigraphicstudies. However,we emphasize
thatthe compositional differencesamongthe basaltunitsare
subtle and require data of the highest quality for their
recognition.Furthermore,althoughthe generalstratigraphic
structureof the sequence is recognizable in bothRU andWSU
data sets,problemsof small interlaboratorybiasesfor some
elementsresultin characteristicincompatible-element ratios
for any one unit being slightlydifferentin the two datasets
70 80 90
(seeTable 1 for somecomparisons betweenthetwo datasets).
Ti/Zr
Thesetwo effectshave provedtestingfor the integrityof our
geochemicalsubdivisions andmake it difficultto classifyany
randomsample,especiallyfrom the LesothoFormation,using
these diagrams. We warn against such haphazard and
uncriticaluseof our results.As in many stratigraphicstudies,
an individualsamplecannotbe divorcedfrom its contextand,
in the Lesothosuccession, this requiresthat classification
be
basedon resultsfrom a sequenceof flows.
5.4. GeochemicalStratigraphy:IsotopicEvidence
Figure 16 summarisesour Sr isotopicdata,someof which
are listedin Tables1 and2. In general,the resultsare similar
to thoseof previousstudiessummarisedby Marsh and Eales
[1984] andBristowet al. [1984], in that basaltsof the Barkly
EastFormationare isotopicallymorediversethanthoseof the
Lesotho Formation. This is also consistent with the trace
element diversity discussedabove. Marsh et al.
 MARSH ET AL. 263

Ongeluksnek includedatafrom the uppermostpart of the sequence.

Ramluckan[ 1992] recognizedfive geochemicalunitsin the
5OO
41e
40 ß
•
39ß•'••l
-.
"-.
"-. "normal"
Mafika
Ben
McDhui sectionbeing,from the baseup, the GiantsCup, Agate Vale,
Sakeng,Mkhomazana, andthePhinongunits.The GiantsCup
unit is about14 m thick and,from the analysesand descrip-
ol ""- Lisiu I I tions,appearsto consistof severalthin, highly alteredflows;
it wasnot sampledby us. The Agate Vale unit is equivalent
47. ''-•g.•.
?..C.•
b'"-....."-.... :9 to the Golden Gate unit, and the overlying Sakengunit is
equivalentto the Saniunit. The Mkhomazanaunit is alsovery
• ---,. •.,._o ...... '... ••'• ß5
2 ß• ........ - ..... "..1••1ß _ thin and may have been missedin our samplecollection.
_• Moshesh s •'ora ........ •ß
These flows are strongly altered and appear to have been
1-• ...................................
•-1 distinguishedby Ramluckan[ 1992] on thebasisof highNa20.
90 We considerNa to provide a weak basisfor geochemical
stratigraphyandare dubiousaboutthe statusof this unit. The

• "normal"
80
. Mafika
Lisiu,,,, overlying basaltsform the Phinong unit, which is largely
70 equivalentto our Mafika Lisiu unit. Althoughwe have not
madea detailedanalysisof Ramluckan'sdata,it appearsthat
60
•"1ow
Zr/Nb"
Mafika
Lisiu samplesat the baseof his Phinongunit are equivalentto our
50 Wonderkopunit andsamplesfrom the topmostflows classify
[] • [] Moshesh's
Ford asMaloti basalts.Also within the centralpart of the Phinong
40 .... I .... I ....

5 10 15 20 unit are several flows of high-Nb basaltswhich can be

correlated to the low-Zr/Nb Mafika Lisiu subunit. In
Zr/Nb
summary,the resultsof Ramluckan'swork arequitesimilarto
Figure 14. Detailed stratigraphicsectionat the base of the ours, although there are differencesin the stratigraphic
Ongeluksnek and Ben McDhui illustrating geochemical interpretationof the compositional variations.
classificationof the samplesimmediatelyoverlyingtheMoshesh's Marsh and Eales [ 1984] summarised the geochemistryof a
Ford basalts.Thesearethe "mixedunit"samplesfrom Figure11. collectionof basaltsfrom a 600-m sectionthroughthe lower
The low-Zr/Nb samples(filleddiamonds) correlatewith the field for
partof the sequenceat Naudes'Nek, some12 km southeast of
the low-Zr/Nb Mafika Lisiu subunit in the northern sections,
whereasthe interbeddedhigh-Zr/Nbbasalts(invertedtriangles) the Ben McDhui section. Apart from the basal Moshesh's
correlate with the "normal" Mafika Lisiu basalts from all sections. Ford basalts,thissectionwasregardedasbeingbuilt of flows
The palaeomagnetic polarityof the samplesis indicatedby "R"and of the Lesotho magma type; in fact, the geochemical
"N" andthe numbereddotsreferto samplelocationsandnumbers. characterizationof the Lesothotype of Marsh and Eales
[1984] andMarsh [1987] reliedon samplesfrom theNaudes'
demonstratedthat Sr (and Pb) isotopiccompositionis an Nek section. We have reanalyzedthesesamplesand their
effectivediscriminator amongstdifferentbasalttypesin the compositionscorrelate with the Mafika Lisiu basalts A
Karooprovince,and it is unlikelythatotherisotopicsystems sequenceof low Zr/Nb flows near the base of the section
would presenta generalstratigraphic picturemuchdifferent overlies Moshesh's Ford basalts as at Ben McDhui. Maloti
from that in Figure 16. Within the Mafika Lisiu unit the Sr basaltshavenot beenencountered in the section,presumably
isotope compositionsgroup around0.706 and 0.705; the becausesamplingterminatedbelowtheiroutcroplevel. Apart
higher valuesare characteristic of samplescollectedin the from the Moshesh'sFord basalts,Marsh and Eales [1984]
upper part of the unit abovethe low Zr/Nb horizon,andthe alsodocumenteda numberof chemicallydistinctunits(Kraai
lower valuesare from samplesbelowthishorizon. River, Vaalkop, Omega,and PronksbergHigh-K) within the
Barkly East Formationto the west aroundBarkly East and
5.5. Comparisonwith PreviousStudies Molteno. None canbe correlatedwith the unitsof the Barkly
East Formation in the northern sections.
Apart from the reconnaissancestudyof Marsh [ 1984],the
resultsof whichhavebeenincorporated intothisinvestigation, 6. CORRELATION BETWEEN PALAEOMAGNETIC
theonlyotherstratigraphicstudyin theLesothobasaltsis that AND GEOCHEMICAL STRATIGRAPHY
of Ramluckan[ 1992] on an 800-m sectionat Sani Pass. This
studydiffersfromoursin thatit terminated
at SaniTop(see We have determined the precise position of the
Appendix), within the Maloti unit, and thereforedoesnot palaeomagnetic
polarityreversalat theOxbow,SaniPass,
264 STRATIGRAPHY AND AGE OF KAROO BASALTS OF LESOTHO

25 ß Golden Gate

Nb • Letde P/Zr

_ •Mafika
Lisiu
x S•ni 10
20 0 Rom•

[] 0 Wonderkop

• []Moshesh's
Ford
8
0/•.....
'•-..... Mafika
Lisiu
10 Maloti•..... - •:• • Mothae
&
,,• •-',•••:. Senqu •1 II•r.I •-.•A Malot,
Oxbow
6

• aLsiu
4
40 60 80 100 8 10 12 14
Ti/Zr FeO*
o

P/Zr Zr/Nb o * *
10
x ß •

30Mafika• • •ß
[ ( } •Maloti, Senqu
\ ',:/.., ,'" &Mothae

'lowZr/Nb• ......•.--• •_ aaa a a

MafikaLisiu ' • a
I .... I .... I .... I ....

60 80 100 2 3 4 5 6

Ti/Zr Zr/Y

Figure15. Variation
diagrams
showing
thecomposition
of samples
fromtheBarklyEastFormation
in relation
to
composition
fieldsfor unitsof theLesothoFormation.

Mafika Lisiu sectionsand the approximatepositionof the 7. IMPLICATIONS FOR STRUCTURE AND
reversalat OngeluksnekandBen McDhui [Rehacek,1995]. EMPLACEMENT OF THE LAVA SEQUENCE
Althoughwe did not determinethe polarityof samples
collectedfrom the Bushman'sPasssections,we were able to The stratigraphyillustrated
in Figure6 givesinsightintothe
correlate our geochemicalsample sites with the well- overall structureof the lava sequencein Lesotho. The
establishedreversaldescribedby Van Zijl et al. [1962]. At extensiveMaloti unit is particularlyimportantin thisregard.
Semongkong and in the Mafika Lisiu (E) section,all our The present-dayaltitudeof the baseof this unit is shown
samples havenormalpolarityandthereversalpresumably lies adjacentto eachsectionin Figure6 andit is evidentthatthe
lowerin thesequence thanourlowestsample.Thepositionof elevations are lower for sectionslying within the basalt
the reversalin eachsectionis shownon Figure6. remnant(SOM andMLP-E) comparedto thoseat theedges.
In all instances the reversal occurs within the Mafika Lisiu The overallhigh elevationof the SaniPasssectionis alsoa
unit;moreprecisely, it occursbelowthelow Zr/Nb subunitat noteworthyfeature. This confirmsthe observations made
SaniPass(95 m below),Maf'hkaLisiu (65 m), andBushman's previously[e.g.,Stockley,1947]thatthepresentstructure of
Pass(92 m). At Oxbow,Ongeluksnek, andBen McDhui, the lava remnantis one of a broadbasinwith slightinward
wherelow Zr/Nb flowsareinterlayered with 'normal'Mafika dipsof the flows. The overallconstancy in thicknessof the
Lisiu flows, the reversal appearsto occur within the Maloti and otherunitssuggests thatthis structuredeveloped
interbeddedsequence (Figures6 and 14). Althoughmore subsequent to eruption,butwe cannotexcludethepossibility
detailedwork is requiredin the two southern sections,it is of somecontemporaneous subsidence.
clear that a consistentrelationship exists between our The surface onto which the earliest basalt flows were
geochemical stratigraphy andthepalaeomagnetic reversal. emplacedwasunderlainby the Clarenssandstones of
 MARSH ET AL. 265

I ' I intrusionsthat can be correlatedwith any otherunits of the

Barkly eastFormation.The problemof the lack of correlation
Oxbow Dykes
in the unitsof the Barkly EastFormationbetweennorthand
Mothae
south Lesotho awaits further detailed studies for its resolution.
The overlying, thicker geochemicalunits of the Lesotho
U. Senqu
FormationspreadacrossLesothowith more or lessconstant
L. Senqu
thickness. It should be pointed out that the thicknesses
Maloti
Mafika Lisiu
represented in Figure6 havenotbeencorrectedfor any dip of
the strata. Several sectionswere sampledover horizontal
Wonderkop
Moshesh's Ford
distancesof 15 km or greater,and a slightdip of 1ocanresult
A Letele
in over- or underestimatesof thicknessesby 100 to 200 m
xx X Sani
dependingon the relationshipbetweenthe directionof dip and
0 Roma the orientationof the section. Thus, the apparentslight
Golden Gate differences in thicknessof units between different sections,
particularlythe northernsectionsin Figure 6, are not con-
Dykes
sideredsignificant. This constancyof thicknessstrongly
suggeststhat the bulk of the lava pile was built in a uniform
0.704 0.706 0.708 0.710 0.712 manner (uniform eruption rates ?) on a generally planar
surface.Thereis nothingin the stratigraphy that suggests the
initial87Sr/S6Sr presenceof geographically focussedlavaeruptionsites;even
the restricted occurrence of some of the lower units seems to
Figure 16. Summaryof initial Sr isotoperatio datafor the different
be a resultof topographiccontrol. Indeed,our observations
units.
and availablegeologicalmaps lead us to concludethat the
flows of all unitswere fed from a diffuseplexusof dykes
fluvial and aeolian origin. This surface exhibited mild spreadthroughoutLesotho and probably the entire Karoo
topographicrelief, with gentleslopesand occasionalsteep- basin. Thatthe flow typein thebasaltpile is overwhelmingly
sidedvalleys[Locket al., 1974]. The distributionof the thin compoundin naturemay be significantin this regard. The
lower units is determinedin part by topographyand this vastsubvolcanic networkof dykesandsillsis ampleevidence
plausiblyexplainstheabsence of unitsin someof the sections. of the generalavailabilityof basalticmagmain the uppermost
On the otherhand,it is strikingthatthebasalGoldenGateunit crustallayers.
in northern Lesotho and the Moshesh's Ford unit in southern
Lesothoshowsuchextensivedevelopment(over areasup to 8. THE AGE OF KAROO VOLCANISM
40,000km2).As emphasised
above,sections
northof latitude
30øSexhibita reasonablycontinuous stratigraphy
amongstthe Figure17 summarises
new4øAr/39Ar
agedataobtainedin
lower units, which differs completelyfrom that in the two the currentstudy(R. Duncan,P Hooper, J. Marsh, and A.
southern sections and the areas farther to the southwest Duncan,unpublished information).A previousreviewof age
described by Marsh and Eales [1984]. This lack of databy Fitch and Miller [ 1984] proposedthat activityin the
correlationcould reflect the influenceof an importanttopo- Central Area was episodicover a period of 40 to 50 m.y.
graphicbarrierto lower flow distribution,of geographically starting at 190-195 Ma, and with a main phase of
separated eruptionsites,or of both. In thisregard,it isnotable emplacement at about175 to 180 Ma. That this long spanof
that the thicknessof the sequencebetweenthe baseof the activityshouldbe represented in the Lesothosequencedefies
volcanic sequenceand the Maloti unit at Ben McDhui and geologicalevidence,suchasthe lackof significantweathering
Ongeluksnekis variableandstandsin contrastto the regular horizonsandsignificantsedimentaryunitsinterbeddedin the
thicknessof this sequencein all the northernsections.This basalts,suggestinga shortperiodfor the emplacementof the
variation is evidence of the more dramatic pre-volcanic basalt sequence. The preservationof the transitionin the
topography in the southcomparedto the north,a featurealso palaeomagnetic polarityreversal,in particular,impliesa very
born out by the work of Lock et al. [1974]. Locationsof shortperiodfor the buildingof the Lesothosuccession. Much
eruptionsites may be indicatedby intrusionsof specific of the spreadin the old K/Ar agesshownin Figure 17 comes
compositional types. IntrusionsofMoshesh'sFordbasaltsare from dolerites,which may be thoughtto preservethe time
known from Ongeluksnekand some60 km to the southwest rangeof activitymore completelycomparedto a lava pile
at Blikana [Marsh and Eales, 1984], but we havenot found whichhasundergone anunknownamountof erosion.On
266 STRATIGRAPHY AND AGE OF KAROO BASALTS OF LESOTHO

New Amalfi sheet earlier age for the commencementof basalt eruptionin
Tandjiesberg
Sill southernLesothoat 190-195 Ma [Fitch and Miller, 1984].
Rb/Sr isochron
lnsizwaintrusion
An age of ca 180 Ma for the Central Area activity is
consistent with the correlations between the Lesotho/
_• Mariental
basalts SpringbokFlats remnantsand the centralLebombomade
ß • Keetmanshoop
sills
below. R. Duncan (unpublisheddata, 1995) determined
• Top 4øAr/39Ar
agesfornephelinite,
picrite,basaltandrhyolitefrom
4OAr/3•Ar__ the Lebombo. The mafic rocksyield agesof 180+2Ma and
I-•tl.9q Middle Lesothothe Jozinirhyolitesslightlyyoungeragesof about177 Ma.

•ß basalts
Theseareall withinerrorof the Rb/Srisochronageof 179+4
Base
Ma for the Jozinirhyolitesobtainedby Allsoppet al. [1984].
10-
Ages of 180 Ma have also been obtainedfrom 4øAr/39Ar
measurementson plagioclaseseparatesfrom the Kirwan
- Old K/At basalts,Antarctica(R. Duncan,unpublisheddata, 1993),
5 -- ages •
whichHarriset al. [ 1990] correlated
ongeochemical grounds
basalts and dolerites with low-Ti basaltsof southernLebombo, a correlationin
i I keepingwith their closespatialrelationship in Gondwana
I I I I I I

135 145 155 165 175 185 195 205

reconstructions.There is now overwhelmingevidence thatthe
whole Karooigneousprovincewas emplacedwithin a short
age (Ma) period at 180 Ma. Furthermore,the Kirwan basaltdataand
the preciseU-Pb ageof 181.2+0.7Ma for the Dufek intrusion
Figure 17. Comparisonof recentage determinations for Central [Minor and Mukasa, 1995] extend the area of Gondwana
Area basaltsandintrusionswith previousageslargelydeterminedby
influencedby Karoo-ageflood basaltvolcanismdeepinto
conventional
K-Ar techniques
(seetext for references).
4øAr/39Ar Antarctica.
ageswith errorbarsarewholerockplateauagesandthosewithout
are total fusion ages.
9. CORRELATIONS WITH OTHER KAROO
VOLCANIC SUCCESSIONS
other hand, dolerite intrusionsare relatively undegassedin
comparisonto lava flows and doleriteagedeterminations by 9.1. SpringbokFlats
conventionalK/Ar methodsmay be more susceptible to the
presenceof excessargon. The only othermajorremnantin the CentralArea for which
Our new age data for the basaltsin and aroundLesotho a detailedgeochemical stratigraphyhasbeenestablished is the
show a tight clusteringaround 180 Ma. Significantly,the poorly exposedSpringbokFlats remnantsome400 km due
basaltsfrom Mariental in Namibia are identicalin ageto the north of Lesotho (Figure 1). A detailed geochemicaland
Lesotho basalts, as are the dolerite sills from the isotopicinvestigation (J. Marsh,unpublishedinformation)of
Keetmanshoop areain southernNamibia. Thesenew 4øAt/ four widely spacedboreholecoresthroughthe sequence has
39Aragesareconsistent
withtheRb/Srinternalisochron
age led to the recognitionof five stratigraphic
units(seeTable 3
obtainedby Richardson[1984] for the Tandjiesbergsill in for representativeanalyses).Oneof theseunitsconsists of 30
southernNamibia (Figure 1) and are alsoconsistent
with the m of high-Ti flows in the upperpart of the sequence, which
geochemical evidence of a very close compositional are compositionallyand isotopically[Marsh et al., 1992]
correspondence
betweentheseNamibianbasaltsanddolerites identicalto the low-Fe, high-Ti-Zr basaltsat the baseof the
and those of the main Karoo basin. Data for other differen- SabieRiver Formationstudiedby Sweeneyet al. [ 1994] in the
tiated intrusionsare includedin Figure 17; theseare Rb/Sr central Lebombo. Thus, the SpringbokFlats succession
isochronagesfor the Insizwaintrusion(F. J. Kruger,personal providesa vital correlatinglink betweenthe CentralArea
communication, 1993) and the New Amalfi intrusion basaltsandthoseof the Lebombo. This is importantbecause
[Williams, 1995] (Figure 1). Thus all the new datafrom the thereare at leasttwo palaeomagnetic polarityreversalsof the
CentralArea showa tight clusteringaround180 Ma andthis senseR to N in the Lebombo successionand, aswill be shown
smallrangeis consistent with agedatafrom otherlargeflood below, the geochemicalcorrelationis criticalin determining
basaltprovinceswhichemphasize the shorttime spans(1-3 which one correlates with the reversal in the Lesotho
m.y.) for the emplacementof the bulk of their igneous succesion. Previous studies have been unable to determine
products.Specifically,the latestdataofferno supportfor an any reliablecorrelationbetweenthe two
 MARSH ET AL. 267

andisotopiccompositions [Marshet al., 1992]overlapalmost

Springbok
• WD-4
Flats
completely,althoughthe more evolvedFe-richcompositions
aremoreextensivelydevelopedin the upperSpringbokFlats
sequencethan in Lesotho. Previously,Marsh and Eales
[1984] categorizedthe SpringbokFlatsbasaltsas a separate
typedifferingfrom the Lesothobasalts.The few analysesof
SpringbokFlatsbasalton whichthisproposalwasbasedwere
•, 600
very Fe-rich and comparisonswere made with the lower,
100 km least-evolved basalts of the Lesotho Formation at the Naude's
I Units
Nek in the southernpart of the Lesothoremnant.The more
" 400 4 completedatasetprovidesa truerpictureof therelationships
andthe impression gainedis thatthe SpringbokFlatssection
high-Ti
•J 3
(max. preservedthicknessabout700 m) represents a com-
a:• 200 1
pressedversionof the Lesothosequence.
2
In detail, the basalUnit 1 in the SpringbokFlatssequence
1 showsstronggeochemical affinitiesto the basalGoldenGate
m 0
RL-1 LB-1 WD-4 RTL-1 unit in northernLesothoin havinghigh87Sr/86Sr andZr/Nb,
low P/Zr and Ti/Zr (Figures20 and 21) and similarrelative
Figure 18. Map showinglocationof the SpringbokFlatsboreholes REE abundances.An importantdifferencebetweenthe two
anda summaryof the geochemicalstratigraphydeterminedin each units is their degreeof differentiation;Unit 1 is relatively
core.
primitivewith Mg# = 60-66 whereasGoldenGatebasaltsare
considerablymore evolvedwith Mg# = 50-58. Figure 21
6OO indicatesthatUnit 1 andthe GoldenGateunit havethehighest
Zr/Y ratiosof theunitsin theirrespective successionsbutthat

400
600 SpringbokFlats
.basalts
200

400
_

40 60 80 100 2 2.5 3 3.5 4 4.5

Mg# Ti/Zr Zr/Y

Figure 19. Compositional variationswith heightin the Springbok

Flatssuccession. This is a compositediagramcontainingdatafrom
all coresplottedagainstheightabovethe baseof the lava flows. 200

Figure 18 illustratesthe stratigraphicsequencein the

SpringbokFlatsandFigures19 and20 summarise someperti-
nentgeochemicaldata. Loggingof boreholecoresprovided
no geologicalbasis(apartfrom the easyrecognitionof the
macrophenocrystic high-Ti basaltflows) for the subdivision
of the basaltflow sequence,and Figure 18 was constructed I I J • I I I J I
usinggeochemistry, employingthesamebasiccriteriafor the 0.705 0.710 0.715
recognitionof distinctgeochemicalunits as in the current
Lesothostudy. Leaving asidethe high-Ti basalts,the overall
compositionalvariationin the SpringbokFlats sequenceis
87Sr/86Sr
very similar to that in Lesotho;there is an overallupward Figure 20. Verticalvariationin initial Sr isotoperatiosthroughthe
trend towardsevolved basalts,and major and trace element SpringbokFlatssuccession. Symbolsdefinedin Figure
268 STRATIGRAPHY AND AGE OF KAROO BASALTS OF LESOTHO

P/Zr P/Zr
10
MafikaLisiu O O MafikaLisiu
--

Maloti
•O•'.•..••-•,•Oxbow •l•"•'"•Maloti
2 ørd• / ."• • Unitl
O om??" """
_ ..................
:""'
a• ßUnit4 '"•
....
,• '-•,.,.•,j•
Gate
40 60 80 100 2 3 4 5 6

Ti/Zr Zr/Y
Figure 21. Comparisonof somepertinentinterelement
ratiosin Springbok
Flatsbasaltunitswith thoseof the succession
in Lesotho. Symbolsdefinedin Figure 19.

the absoluteZr/Y ratiosdiffer. This differencemayreflectthe fashionwith time. Consideringall the evidence,we suggest
different degree of evolution (Zr/Y ratios shouldincrease that the two sequencescan be correlatedas illustratedin
slightly with fractional crystallization). In general, the Figure 23.
geochemical andstratigraphic evidencefor a correlation
of the
two is compelling,exceptthat the GoldenGate basaltsare 9.2. Lebombo
more evolvedand perhapsslightlymore contaminated (as
suggested by the highandvariableSr-isotopedata)compared The Lebombo monocline shows considerable lateral and
to Unit 1 basalts. vertical lithological variation in the volcanic sequenceas
Unit 2 basaltswith high P/Zr and Ti/Zr, low Zr/Y, and summarisedin the maps by Eales et al. [1984]. In the
similar compatible-and incompatible-element abundances northern Lebombo, sporadicnephelinitesat the base are
correlatewith the Mafika Lisiu unit. Unit 3 is compositionally
similarto the Senquunit (Figure21). The Fe-richbasaltsof SpringbokFlats
Unit 4 correlatewith the more evolved Senquand Mothae Ti/Zr Zr/Y Zr/Nb 87Sr/86Sr ]
4 - P/Zr•
units in many respects,exceptfor Ti/Zr, La/Yb and La/Sm / /
3 ß
which are similarto ratiosin the Oxbow dykes.However,the /
.'• 2 ß
Unit 4 basaltsaremoreprimitive,with lowerZr/Y, FeO*, and
TiO•_ and higher AI•_O 3 and MgO comparedto the Oxbow
I I I I I I I
dykes, but could evolve by fractionationto the Oxbow 1:::1., 56789 50 60 70 80 3 3.5 4 4.5 20 30 40 .706.708.710

compositions. Correlation between Unit 4 and Oxbow I I I I I I I I I I I I I I

ß• ox- ß
intrusions
isalsofavoured
by 87Sr/S6Sr
ratios,buttheisotopic /- / / .
evidence is not exclusive.
/
In summary,the stratigraphicunits identifiedin Lesotho ML •ß
have a closegeochemicalaffinity to thosein the Springbok GG

Flats sequence. Although correlationsare not perfect,it is Lesotho

important to emphasisethat the temporal changes in
geochemical character of the basalts, as reflected in the
Figure 22. Comparativevariationof interelementratiosand Sr-
geochemicalstratigraphy,are extremelysimilarin bothareas isotopiccompositionbetweenunitsin LesothoandSpringbokFlats.
(Figure 22). Although we hesitateto concludethat the Althoughthe actualvalueof a ratiomaynot correspond preciselyin
Springbok Flats basaltsflowed from the samefeedersthat the correlatedunits, the overall pattern of variation is striking.
suppliedthe Lesothosequence,it is clear that the magma Abbreviations: OX - Oxbow;SQ - Senqu;ML - Mafika Lisiu;GG -
supply systemsfor both sequencesevolved in an identical Golden Gate
 MARSH ET AL. 269

LESOTHO LEBOMBO Lebombosequence (Figure23) suggests thatcomparisons of

(Oxbow) 8 kra 1ow-Ti basaltsshould only consider those basaltslying above
180(2)
Mbuluzii!!iiiiiiiiii
N the correlatedhigh-Ti basaltflows. In the Lebombothis
rhyoliteiiiiiiiiiii!i includesall the low-Ti basalts,but in the SpringbokFlatsthese
includeonlytheuppermostFe-richflows;i.e., the equivalent
SPRINGBOK }ozini of the SenquandMothaPflowsandtheOxbowdykes.Figure
• FLATS rhyolite
'• 177(2) 24 demonstratesthat the Lebombo basalts are composi-

• _ø•aY•e
s-
-•• high-Ti,
Fe tionallydifferentfromthe Springbok Flatsbasalts,principally
basalt 179(1) in beingmoreevolved,i.e., richerin Fe, Ti, Zr andpoorerin
•,.e 0.5
• 1ow-Ti182(1)
basalt
MgO, A1203,etc. In generaltermstheseLebombobasalts
could have evolved from Unit 4 basaltsfrom the Springbok
Flats,but whether suchevolved basaltsexistedin the Central

--•a•
z.. •.•.:::•:•:•:
• • 181
(1) Area to providea directcorrelationis uncertain.However,as
Marsh et al. [ 1992] emphasized,the isotopiccharacterof the
• Letaba R low-Ti basaltsfrom the two areasis strikinglydifferentand,
high-Tipictires 1
on availableevidence,we suggestthat the Lebombomagma
systemdevelopedin a largely separatemanner from that

Polarity Polarity
•Age supplyingthe CentralArea.
The geochemical correlations
betweenLesotho,Springbok
betweenLesotho, Flats, and the Lebomboillustratedin Figure 23 indicatethat
Figure 23. Summaryof theproposedcorrelations
Springbok Flats,andLebombo.Thediagramispartlyschematicand thepolarityreversalat the baseof the Lesothopile correlates
not to scale,particularlywith regardto the compositeLebombo withthatin thelow-Fe,high-Tibasaltin thelowerpartof the
section. Magnetostratigraphyfrom this study and Hatgraves
(personal communication,1995). Agesfrom this study.Error on 18
agesindicatedby numberin brackets.Note the differentthicknesses
17
of mafic rockspreservedat eachlocality. - All0a • _

16

15
overlain,in sequence,by Letabapicrites,SabieRiver high-Ti
basalts,andJozinirhyolites.Southwards, thepicritesoverstep
14 - Lesotho i{
13
the nephelinitesand are, in turn, oversteppedby the basalts. - • Fiats
3
In the centralLebomboa wedgeof low-Ti basaltseparates the
Ti02 • ß ^,. •
high-Ti basaltsinto a basallow-Fe varietyand an overlying 2.5

high-Fevariety. The low-Ti basalts thickensouthwards at the 2

expenseof the high-Ti typesand completelydominatethe : • • '•"• • • Springbok

basaltsequencein the southernLebombo. In addition,in
1.5
: • ;?'• Flats
Swaziland,the Mbuluzi rhyolitesoverliethe Jozinirhyolites
and are in turn overlainby the little-knownMovene basalts.
Nowhere is the complete sequencepreserved,and the FeO* •• Springbok
- •,• Flats
succession in Figure23 is a compositecolumnshowingthe
variouslithologiesin their correctstratigraphicorder,aswell
as our preliminary palaeomagneticpolarity reversal 10
stratigraphy
and4øAr/39Ar
ages. Lesotho

Duncan et al. [ 1984] demonstrated

thatthe low-Ti basalts 8
2 4 6 8 10
of the Sabie River Formation of the southern Lebombo were
MgO
geochemicallydistinct from other Central Area basalts,
specificallythe basaltspreviouslyreferredto asthe Lesotho
Figure 24. ComparisonbetweenCentralArea 1ow-Tibasalts
type. However,in the lightof thenewage,palaeomagnetic (Lesotho,SpringbokFlats)andthosein the Lebombo(asterisks)
and geochemicaldata discussedhere, and the work of illustrating themoreevolvedcharacter of theLebombobasalts.Data
Sweeneyet al. [ 1994],thisquestiondeservesreconsideration. fromthisstudyandSweeneyet al. [ 1994].Notethealmostcomplete
Firstly, the correlationestablishedfor the high-Ti basalts overlapin composition
betweenthe Lesothobasalts(datafrom all
located at the base of the Sabie River Formation in the units)andthoseof the Springbok
270 STRATIGRAPHY AND AGE OF KAROO BASALTS OF LESOTHO

Sabie River Formation. This correlation implies that the
basalticeruptions in theLebombowerecontemporaneouswith
those in SpringbokFlats and Lesotho. However, the geo-
chemical correlationshown in Figure 23 suggeststhat the
main volumeof Lebombobasalt,specificallythe SabieRiver
low-Ti basalts,largelypostdates
themainlow-Ti emplacement
event in the Central Area.
10. CONCLUSIONS
Detailedsampling,togetherwith preciseX ray fluorescence
analysis, has allowed us to construct a geochemical
stratigraphy for the Karoo basalts in Lesotho. This
stratigraphyis consistentwith results of palaeomagnetic
polaritystudies
and4øAr/39Ar
agedeterminations.
Thinbasaltic
units with diverse geochemistry form the Barkly East
Formationat the baseof the sequenceand,althoughsomeof
theseunitsmaybefoundoverareasapproaching
40,000km2,
others are found over rather limited areas. In the sections north
of latitude30øS thereis a reasonablycontinuousstratigraphy
appearsthatthebulk of the CentralArea basaltsmay slightly
predate the basaltic volcanism (but not necessarilythe
nepheliniticor picriticextrusions)
of theLebombo.However,
theseagedifferences,
if valid,arenotresolvable
by 4øAr/39Ar
dating. In any event, apart from the high-Ti basaltin the
SpringbokFlats succession, the largegeochemical,isotopic,
andpetrographicdifferencesbetweenthe Lebombovolcanic
suite and that of the Central Area emphasizesthat the
Lebombomagmaticsystemevolvedand eruptedits volcanic
productsseparatelyfrom that in the CentralArea andLesotho
in particular.The rift environmentof the Lebomboprovided
amplescopefor tappingmagmasourcesperhapsnot available
beneaththe stablecratonicinteriorandmay alsohaveplayed
an importantrole in confining the eruptedproductsin rift
depressionsalongthe Lebomboand to the east. Available
data demonstratethat few of the flows eruptedalong the
Lebombo flowed westwards across the cratonic interior of
southern Africa.
APPENDIX

amongstthe lowerunitswhichdifferscompletelyfrom that in

the sectionsto the south.Furtherinvestigations are neededto SampleLocalities
map out the stratigraphyin the lower part of the sequencein
southernLesotho and to determinethe causeof the poor The Ben McDhui section(BMC samples)was sampled
north-south correlations between the lower units. along the jeep track that ascendsCarlisle'sHoekjust eastof
Overlying the Barkly East Formation is the Lesotho the villageof Rhodesto the weatherstationat the baseof Ben
Formation,comprisinga numberof thick units of constant Mcdhui andfrom thereup the slopeto the summitbeacon.At
thickness which can be traced throughoutthe Lesotho OngeluksnekPassthe road follows a zone of shearingand
remnant.Theseunitsbuild themainbulk of thebasaltpile and smallfaults. Sampleswerethuscollectedin threetraverses up
their flows appearto havebeenfed from a widespreadplexus a series of streams to the south of the road from the border
of dykes. The new age data indicatethat the entireLesotho postat the footof thepassto thetop. The uppermostsamples
lava pile and its correlativesup to 1300 km away (Mariental (ON-3 to 18) weretakenup the slopefrom the top of the road
in Namibia) were emplaced at 180+2 Ma, an age that is passto the summitof an unnamedpeak immediatelyto the
consistentwith Rb-Sr isochron ages from differentiated south. At Sani PasssamplesSP-1 to 20 were collectedfrom
dolerite intrusions in the Central Area mafic rocks in the the summitof the hill Kotisepholasome11 km from the top
Lebombo. There is no evidencein the new data for a long of pass,down ontothe road and then down the roadto Sani
period of episodic activity as proposedpreviously. The Flats. The remainingsamplesSP-21 to 51 were takenfrom
apparentshortdurationof the activity is entirely consistent Sani Top down the pass to the basalt-Clarenssandstone
with the lack of significantsubaerialweatheringhorizonsor contact. The Oxbow samplesuite was also collectedin two
palaeosolsin the basalt sequence,with the palaeomagnetic sections. The upper section (OXB-1 to 25) was from the
data, and with the overall close geochemicalcorrelation summit of the hill Mahlasela down the Mahlasela Pass to the
between the basalt flows and dolerite intrusions. areaaroundthe Oxbow Lodge. The remainingsampleswere
There is a stronggeochemicalsimilaritybetweenthe basalt collectedfromthetop of the MotengPassdownto the basalt-
sequencein Lesothoandthat in the SpringbokFlatssome400 Clarenssandstonecontact. The Mafika Lisiu samplesuite
km north of Lesotho. We propose specific correlations comes from a double collection. The Mafika Lisiu (E)
betweenindividualunitsin thetwo sequences andsuggest that samplesMLP-1 to 50 werecollectedstartingin thebedof the
the SpringbokFlat succession represents a compressed version MalibamatsoRiverjust northof the Village of Ha Lejoneup
of the Lesothosequence.Throughcorrelationof geochem- to theroadandthenfollowingtheroadup to thesummitof the
ically distinctivehigh-Ti basaltflowshighin theupperpart of Mafika Lisiu Pass. In the western section,MLP 100 to 186
the SpringbokFlats sequencewith the high-Ti basaltsat the and the detailed collection MLM 1 to 17 were collected from
base of the basaltic successionin the central Lebombo, it thetop of thepassdownto thebaseof the basaltsjust west
 MARSH ET AL. 271

Pitseng.MLP-60 to 65 werecollectedup thehill immediately Province:an introduction,Spec.Pub/. Geol. Soc.S. Aft., 13, 1-26,
1984.
southof the top of the pass. Samplesin the Bushman'sPass
sectionwere collectedalongthreepasses, Bushman's,Molimo Erlank, A. J., A. R. Duncan, J. S. Marsh, R. S. Sweeney,C. J.
Hawkesworth,S.C. Milner, R. McG. Miller, andN. W. Rogers,
Nthuse, and Thaba Putsoathat lie alongthe main road from
A laterally extensive geochemical discontinuity in the
Maseru eastwardsto Mantsonyane.BUS-1 to 22 were taken
subcontinentalGondwana lithosphere(abstract),Geochemical
from the bed of the Liphiring streamjust east of Nazareth Evolutionof the ContinentalCrust,Pocosde Caldas,Brazil, 1-10,
along the road to the top of Bushman'sPass,and samples 1988.
BUS-24 to 40 were collectedfromjust southof Ha Chalalisa Fitch, F. J., and J. A. Miller, Dating Karoo igneousrocks by
to the summit of Thaba Putsoa Pass. conventional K-Ar and4øAr/39Ar agespectrum methods, Spec.
The Roma Sectionwas sampledon the Semonkongroad Publ. Geol. Soc. S. Aft., 13, 247-266, 1984.
south of Roma from the Thabana-li-Mele Harris, C., J. S. Marsh, A. R. Duncan, and A. J. Erlank, The
craft centre on the
Makhalengriver northwards alongtheroadto Roma. Finally, petrogenesisof the Kirwan basaltsof Dronning Maud Land,
Antarctica, d. Petrol., 31, 341-369, 1990.
the Semongkong section,alsoa two-partsection,wassampled
Hooper, P. R., D. M. Johnson,and R. M. Conrey, Major and trace
from the summit of Thaba Putsoajust west of the Ha element analyses of rocks and minerals by automatedX-ray
Ramabanta-Semonkong roaddownontothatroad(SOM-1 to spectrometry, Open File Report, 37 pp., Washington State
18) then alongthe roadto the crossingof the Makhoalipana University, Pullman, 1993.
stream(SOM-19 to 40). SOM-82 to 100were collectedin the le Roex, A. R., Geochemistry,mineralogy,and magmaticevolution
steepgullythatdescendsfromtheroadto thebottomof theLe of the basaltic and trachytic lavas from Gough Island, South
Bihan Falls on the Maletsunyane River just south of Atlantic, d. Petrol., 26, 149-186, 1985.
Semongkong. Lock, B. E., A. L. Paverd,and T. J. Broderick,Stratigraphyof the
Karoo volcanicrocksof the Barkly Eastdistrict,Trans. Geol. Soc.
Acknowledgments.Fundsfor this researchwere providedby the S. Aft., 77, 373-374, 1974.
Marsh, J. S., Geochemistryof Karoo basaltsand dolerites in the
Foundation for Research Development,South Africa (Marsh,
northeastern OrangeFree State:recognitionandorigin of two new
Duncan)and the National ScienceFoundation(Hooper). Vehicle
costswere subsidizedby RhodesUniversity,GeologicalSurveyof Karoo basalt magma types (abstract),Geocongress'84, 91-94,
Geological Societyof SouthAfrica, Potchefstroom,1984.
Namibia, and Gold Fields of SouthAfrica. J.S.M. thanksthe Anglo
Marsh, J. S., Basaltgeochemistryand tectonicdiscriminationwithin
AmericanCorporationandS. Marshfor access to the SpringbokFlats
continentalflood basaltprovinces,d. Volcanol.Geotherm.Res.,
cores. We thank the following colleaguesfor field and laboratory
32, 35-49, 1987.
assistance:
MikeJackson,
Phoenix
Hoyle.Reviews'by
K. CoxandR.
Marsh,J. S., andH. V. Eales,Chemistryandpetrogenesis
of igneous
Fodor improvedthe manuscript.
rocksof the Karoo Centralarea,southernAfrica,Spec.Publ. Geol.
Soc. S. AJb.,13, 27-68, 1984.
REFERENCES
Marsh,J. S., R. A. Armstrong,and R. S. Sweeney,New Pb, Sr, and
Nd isotope data from the Karoo Province (abstract), 24th
Allsopp,H. L., W. I. Manton,J. W. Bristow,andA. J. Erlank, Rb-Sr Congress, 262-264, Geological Society of South Africa,
geochronologyof Karoo felsicvolcanics,Spec.Publ. Geol. Soc. Bloemfontein, 1992.
S. Aft., 13, 273-280, 1984. Minor, D. R., andS. B. Mukasa,A newU-Pb crystallization
ageand
Bristow,J. W., H. L. Allsopp,A. J. Erlank,J. S. Marsh, and R.A. isotope geochemistryof the Dufek layered mafic intrusion;
Armstrong,Strontiumisotopecharacterizationof Karoo volcanic implicationsfor the formationof the FerrarProvince(abstract),
rocks,Spec.Publ. Geol. Soc.S. Aft., 13, 295-330, 1984. Eos Trans AGU, 76, 5284, 1995.
Campbell,I. H., and R. W. Griffiths,Implicationsof mantleplume Morgan, W. J., Hotspottracksand the openingof the Atlantic and
structurefor the origin of flood basalts,Earth Planet. Sci. Lett.,
Indianoceans,in TheSea,vol. 7, editedby C. Emiliani,pp. 443-
99, 79-93, 1990. 487, Wiley Interscience,New York, 1981.
Cox, K. G., The Karoo Province, in Continental Flood Basalts, Norrish, K., and J. T. Hutton, An accurateX-ray spectrographic
edited by J. D. Macdougall,pp. 239-271, Kluwer Academic, methodfor the analysisof a wide rangeof geologicalsamples,
Dordrecht, 1988. Geochim. Cosmochim.Acta, 33, 431-453, 1969.
Cox, K. G., andG. Homung,The petrologyof the Karroo basaltsof Ramluckan,V. R., The petrologyand geochemistryof the Karoo
Basutoland,Amer. Mineral., 51, 1414-1432, 1966. sequencebasalticrocks in the Natal Drakensbergat Sani Pass,
Cox, K. G., R. MacDonald, and G. Homung, Geochemicaland MSc thesis,University of Durban-Westville,Durban, 1992.
petrologicalprovincesin the Karroo basaltsof southernAfrica, Rehacek,J., Chemicalandpaleomagnetic stratigraphy
of basaltsin
Amer. Mineral., 52, 1451-1474, 1967. northernLesotho,Karoo Province,PhD thesis,WashingtonState
Duncan,A. R., A. J. Erlank,andJ. S. Marsh,Regionalgeochemistry University, Pullman, 1995.
of theKarooIgneousProvince,Spec.Publ. Geol.Soc.S. Aft., 13, Richardson,S. R., Sr, Nd and O isotopevariationin an extensive
355-388, 1984. Karoo doleritesheet,southernNamibia, Spec.Publ. Geol. $oc. S.
Eales, H. V., J. S. Marsh, and K. G. Cox, The Karoo Igneous Afr., 13, 289-294,
272 STRATIGRAPHY AND AGE OF KAROO BASALTS
Stockley,G. M., Report on the GeologyofBasutoland,114 pp.,
Government Printer, Maseru, Basutoland, 1947.
Sweeney, R. S., and M. K. Watkeys, A possiblelink between
lithospherearchitectureand Gondwanabasalts,0r.Aj?icanEarth
$ci., 1O, 707-716, 1990.
Sweeney,R. S., A. R. Duncan,andA. J. Erlank, Geochemistry and
petrogenesisof CentralLebombobasaltsof the Karoo Igneous
Province,0r.Petrol., 37, 95-125, 1994.
Van Zijl, J. S. V., K. W. T. Graham,and A. L. Hales, The
palaeomagnetism of the Stormberglavasof SouthAfrica, I:
Evidencefor a genuinereversalof the Earth'sfield in Triassic-
Jurassictimes,Geophys.d. R. Astron.Soc.,7, 23 -29, 1962.
White, R. S., and D. P. McKenzie,Magmatismat rift zones:the
generationof volcaniccontinentalmarginsand floodbasalts,0r.
Geophys.Res.,94, 7685-7729, 1989.
Williams,C., The petrogenesis of theNew Amalfi Sheet:a highly
OF LESOTHO
differentiatedKaroo intrusion,MSc thesis,RhodesUniversity,
Grahamstown, 1995.
A.R. Duncan,Departmentof GeologicalSciences,Universityof
Cape Town, Rondebosch 7700, South Africa (e-mail -
ard•ucthpx.uct.ac.za).
R.A. Duncan,Collegeof OceanandAtmospheric Sciences,
Oregon
State University, Corvallis, OR 97331-5503, USA (e-mail -
rduncan•oce.orst.edu).
P. R HooperandJ. Rehacek,Departmentof Geology,Washington
State University, Pullman, WA 99164-2812, USA (e-mail -
prhooper•mail.wsu.edu).
J.S. Marsh, Department of Geology, Rhodes University,
Grahamstown 6140, SouthAfrica(e-mail-
 SiberianTraps
Mukul Sharma

TheLunaticAsylumof theCharlesArmsLaboratory,Divisionof GeologicalandPlanetarySciences

CaliforniaInstituteof Technology,
Pasadena,California

This paperexaminesrecentattemptsto estimatethe timing and durationof

Siberianfloodvolcanismandto modelthe petrogenesis of the SiberianTraps.
The most significantfindingsare (1) the bulk of the Siberianlavaserupted
within a periodof aboutone million yearsat 250 Ma; (2) one of the earliest
eruptions
hasa 3He/4He
ratiowhichis 12.7timestheatmospheric
value;(3) the
early-stage
lavasthatconstituteabout8% of theprovincedisplaywidelyvarying
mineralogy andchemistry, with e•qd(t)
valuesrangingfrom-10 to +7; and(4) the
voluminous late-stage lavas (•90% of the province) are remarkably
homogeneous in mineralogyand chemistry,shownearlyflat rare earthelement
patterns(Ce = 20 x primitivemantle;Ce/Yb • 2 x primitivemantleratio), have
higherTh/Ta and La/Ta ratiosthan thoseexpectedfrom meltingof primitive
mantle,anda volume-weighted •;Nd(t)• +2. Thetiminganddurationof Siberian
flood volcanismsupporta first-ordercausallink betweenthe volcanismand the
Permian-Triassic
massextinction.Assessment of the geochemicaland isotopic
datapointsto contributions
from two differentsourcesfor the SiberianTraps.
The early-stage
lavaswerederivedfrom a mantlesourcewith •qd • +8 and
containingresidualgarnet.The late-stagelavaswere derivedfrom a shallower
mantle sourcewith e•d • +4. The magmasassimilatedvariableamountsof
continentalcrust.Severalmajor outstanding
problemsregardingthe province
remain to be solved.

1. INTRODUCTION These variations are then used to evaluate the existing

The intent of this paper is to summarizeexisting
geochemical andisotopicdataandtheirinterpretationfor the
origin and petrogenesis of the SiberianTraps.A well-
characterized,
stratigraphically
controlledsuiteof rockswith
selectedgeochemical and isotopicdatais usedto showthe
existenceof systematic temporalvariationsin major and
traceelementsandin initialSr-,Nd-,andPb-isotopic ratios.
LargeIgneousProvinces:Continental,Oceanic,andPlanetary
Flood Volcanism
GeophysicalMonograph100
Copyright1997by theAmericanGeophysical
Union
modelsof the genesisof the SiberianTraps.
Several workers from the former Soviet Union have
contributedsignificantlyto understanding the geologyof the
SiberianTraps [e.g., Zolotukhinand Al'mukhamedov,1988
andreferencestherein].Most of the early Sovietstudieswere
conductedin the western and northwesternparts of the
provinceas a resultof the discoveryof largeCu-Ni sulfide
depositsin the Noril'skregion(Figure 1). A numberof such
studieswere aimedat understanding the originof the sulfide
ores.Systematicgeochemicalwork to understand the origin
and petrogenesis of the SiberianTraps was undertakenby
G.V. Nesterenko and co-workers [Nesterenkoet al., 1964,
1969a,b, 1971, 1972; Nesterenko and Frolova, 1965;
Balashov and Nesterenko, 1966; Nesterenkoand Al'mukha-
274 SIBERIAN TRAPS

2. GEOLOGIC SETTING

2.1. General

Detailsof the geologicsettingof the SiberianTrapshave

been given by Zolotukhinand Al'mukhamedov[1988 and
referencestherein],Fedorenko[ 1980, 1981], Naldrett et al.
Yakutsk [1992], Hawkesworthet al. [1995] and Fedorenkoet al.
[ 1996].The floodbasaltsoccupythe northwestern marginof
the Siberianplatform,which has been a stablecratonsince
the end of the Precambrian.It is surroundedto the northby
the Taimyr Peninsulaandto the westby the EastEuropean-
Urals block [Khain, 1985]. Within the westernpart of the
Siberianplatform,early Paleozoicdolomites,limestonesand
argillaceoussedimentsare overlainby Devoniancalcareous
o 5oo
and dolomiticmarls, dolomites,and sulfate-richevaporites
Lake Baikal ! ß ! andearlyCarboniferous shallow-waterlimestones. Theseare
km in turnunconformablyoverlainby the Carboniferous to Late
Platform
Cover (• Putorana Permian siltstone, sandstone, conglomerate, and coal
FloodBasalts (• Noril'sk measures of the Tungusskaya Series. The Siberian Traps
overlie the TungusskayaSeries sedimentsand are present
Pyroclastic
and • Maimech-Kotuimainly in the Tunguskabasin. At present,the extmsive
intrusive
intrusive(• Nizh.
Dominantly
Tunguska basalticrocksareestimated
to occupyan areaof 3.4 x l0 s
Crystalline
basement
(• Aidan
Shield km2withan average thickness
of 1 km [LurieandMasaitis,
• Anabar Massif 1964]. Fedorenkoet al. [1996] summarizedestimatesby
different workers of the original volume of the Siberian
Figure 1. Schematic map of Permian-TriassicSiberian Traps
Traps. They suggestedthat the total initial volume of the
(modifiedfrom Renneand Basu [ 1991]). lavasandintrusions
couldbemuchhigherthan2 x 106km3.
The entire volcanicsuccession lacks significantinterbedded
sedimentary rocksor paleosols.The basement throughwhich
medov, 1966; Nesterenko et at, 1972; Nesterenko and the lavaswere eruptedis not exposedin the Tunguskabasin.
Al'mukhamedov,1975; see also Masaitis et al., 1966]. It is inferred to be similar to Archean-EarlyProterozoic
However, until quite recentlyonly very sketchyoutlinesof granulite-granitegneisses,amphibolites,crystallineschists,
the geology and geochemistryof this extensiveregion quartzitesand marble exposedin the Aldan shieldto the
existed in the western scientific literature [Lurie and southeastand in the Anabar Massif to the northeast[Khain,
Masaitis, 1964; Nalivkin, 1973; DePaolo and Wasserburg, 1985]. The thicknessof the lavapile in the Tunguskabasinis
1979;BasalticVolcanismStudyProject,1981;Khain, 1985]. typicallygreaterthan 3 km in the northwestandthinsto the
Becauseof the political changesin the SovietUnion in the southeast, where the sequenceis only a few tensof meters
mid-1980s,severalgroupsin North Americaand in Europe thick [Zolotukhin and Al'mukhamedov,1988]. Individual
obtained access to a vast collection of well-characterized flows are, in general,a few tensof metersthick and extend
rock samples,especiallyfrom the Noril'sk region where over a few tens of kilometers; however, some flows are as
thick as 150 rn and can be traced for several hundred
drillholeshad been sunk to explore for Cu-Ni ores. As a
result,a substantial
amountof good-qualitygeochemical and kilometers[Zolotukhinand Al'mukhamedov,1988].
isotopicdatafor the flood basaltprovincehasbeenobtained One of the mostinteresting characteristics
of the lavapile
in the last six years. A review of the earlier work on the is the presence of largequantities
of basaltictuffswhichare
stratigraphyand geochemistryof the SiberianTraps was distributed widelybut irregularlywithinthe Tunguska basin
givenby Zolotukhinand Al'mukhamedov [ 1988]. In addition, [Zolotukhinand Al'mukhamedov,1988]. The pyroclastic
several excellent papers focussingon the geology and depositsreacha maximumthicknessof 700 rn in the center
geochemistryof the Noril'sk regionhave appearedrecently of the Tunguskabasin and thin toward the north and west
[Wooden et al., 1993; Lightfoot and Naldrett, 1994; where the lavas are well developed.To the southof the
Hawkesworthet al., 1995;Fedorenkoet al., 1996]. Tunguskabasin,the tuffsarepresentwithout
 SHARMA 275

lava.Theintrusive
facies
oftl•etraps
crop
outmainly
atthe Fedorenko et at, 1996] (Figure 1): (1) Putorana,(2)
margins of the Tunguska basin where they intrude the Noril'sk, (3) Maimecha-Kotui,and(4) NizhnyayaTunguska.
Precambrian basement, Paleozoic sediments, and the The great majority of the Putoranarocks are relatively
associatedvolcanicrocks.Within the Precambrianbasement, homogeneous, aphyric, and polyphyrictholeiitic basalts
the intrusionsare presentas thin sills or dikes;most of the [Zolotukhinand Al'mukhamedov,1988; Sharma et al.,
intrusive bodies, however, are present as thicker 1991]. The volcanismin the Noril'sk region is markedby
differentiated and undifferentiated sills in the Devonian widely varyingrock typesfrom picriticthroughtholeiiticto
sediments and to a lesser extent in the associated volcanic subalkalic basalts and basaltic andesites[Zolotukhin and
rocks [Hawkesworthet al., 1995]. Sills vary in thickness Al'mukhamedov,1988; $harma et al., 1991]. The volcanic
from a few metersup to 500 m and in someareascontribute rocks of the Maimecha-Kotuiarea are quite evolved as
to as much as 50% of the thicknessof the sedimentary- shownby a wide varietyof rocktypesthatincludepicrites,
volcanic succession[ZolotukhinandAl'mukhamedov,1988]. tholeiitic basalt, alkaline-olivine basalt, trachybasalt,
Linear dikesand dike swarms,the latterextendingto several trachyandesite, basanite,
olivinenepheliniteandmaimechite
hundredkilometers,are presentin the northernand north- [e.g.,ZolotukhinandAl'mukhamedov, 1988].TheNiztmyaya
easternmarginsof the Tunguskabasin.Althoughthe feeder Tunguskaregionis remarkableas it consists dominantlyof
dikes for individual formations are not observed, it is basaltictuffs. Figure 2 gives the currentlyunderstood
believed that the bulk of the Siberian Traps lavas were correlationof volcano-stratigraphic
sequencespresentin the
eruptedthroughlarge linearvent systems[Khain, 1985]. four regions[Sadovnikov,1981; Zolotukhinet al., 1986;
Recentwork on the northwestern margin of the Tunguska Zolotukhinand Al'mukhamedov,1988; Sharmaet al., 1991;
Basin suggeststhat magmaticactivitywas focusedalong Fedorenkoet al., 1996].
discrete lineamentsand that during the early phase of The alkalic to ultra-alkalic rocks of the Maimecha-Kotui
volcanism the centers of volcanic activity switched regioncropout on the westernslopeof the AnabarMassif,
episodically betweendifferenteruptivesites[Hawkesworth about800 km northeastof the Noril'sktype-section(Figure
et al., 1995]. 1). At present,
the stratigraphic
positionof varioussuitesof
The Siberian Traps can be divided broadly into four the Maimecha-Kotuiregion relative to those in the other
regions of fundamentallydifferent volcanic sequences regionsis not clear(Figure2). (Note that in thispaperthe
[ZolotukhinandAl'mukhamedov, 1988;$harmaet al., 1991; terms "suite" and "formation" have been used

Maimecha-Kotui Noril'sk Putorana NizhnyayaTunguska

Yambukansky(Th, TF)
Maimechinsky(MI, P)
Samoedsky (Th) " Nerakarsky
(Th) - ....
0-600 m
0-800 rn /// 500-550
m 100-130 m

Kumingsky(Th) .' . Honnamakitsky(Th) Kochechumsky(Th)

60-450 m
Delkansky(TA, ASA, P) 0-210 m
• •• 100-600
m
370-1000 (?) m Nidymsky(TF)
'" ' .'' ''1Ayansky
(Th) ' -.
Kharaelakhsky(Th, ASA)

Kugotsky(Th, ASA)
380-600 m 150-800m 200-400 m
Mokulaevsky (Th) Korvuchansky(TF)
1000 m 400-650 m 200-400 m
Pravoboyarsky(Th, TF) Morongovsky(Th, ASA)
300-600 m
Tutonchansky (TF)
250-40•._-• Nadezhdinsky(Th, ASA, TF)
25-200 m

Arydzhansky(A) 200-570 m
300-600 m Tuklonsky(Th)
0-250 m

Khakanchansky(TF) A = Alkaline basalt

10-130 rn
ASA = Alkaline to Subalkaline basalt
Gudchikhinsky(Th, P) TA = Trachyandesite
0-200 rn TF = Tuff
Th = Tholeiite
Syverminsky
(Th, ASA) MI- Mairnechite
0-240 rn
P - Picrite
Ivakinsky(ASA, Th, TF)
0-330 m

Figure2. Composite
volcano-stratigraphic
section
oftheSiberian
Traps(modified
fi'omZolotukhin
andAl'mukhamedov
[1988],Sharmaet al. [1991], andFedorenkco
et al. [
276 SIBERIAN TRAPS

interchangeablyand refer to a group of successiveflows rock chemistryand field relations,Naldrettet al. [1992] and
with similargeochemicalcharacteristics;
the formerterm is Fedorenko [ 1991] assessed whetherthe differentintrusion
widespreadin Russianliteraturewhereasthe latterhasbeen typescouldbe linkedto the specificmagmatypesin the lava
usedby Lightfootet al. [1993]).4øAr-39Ar studies indicate pile. Utilizing an extensivechemical data-setfor the
that the volcanism in the Maimecha-Kotui region was Noril'sk region,Fedorenkoet al. [1996] proposeda new
somewhatearlierthanin the Noril'skregionandit may have classificationfor the rocks exposedin the region. These
continuedthroughoutthe main-stageeruptions(seebelow; workersidentifiedeightdifferentvarietiesof primarymantle
cf. Basu et al. [1995]). The lavasof the Noril'skregionare meltswhichthey classifiedinto four primitivemagmatypes:
divided into 11 suites,with the compositethicknessof the (1) a low-Yb type with high TiO2 (thisis a characteristic of
entire sectionreaching up to 3 km [Fedorenko,1981; lavas from the lower to middle Morongovsky), (2) a
Zolotukhin and Al'mukhamedov, 1988; Fedorenko et al., moderate-Yb type with moderate TiO2 (Ivakinsky-
1989]. The Putoranastratigraphy is dividedintothreesuites Gudchikhinsky),(3) a high-Yb type with low TiO2
(maximumthickness = 1.8 km). Thesethreesuitesandtheir corresponding to upperMorongovsky-Samoedlff, and (4) a
counterparts in theNoril'skandNizhnyayaTunguska regions high-Yb type with very low TiO2 (Tuklonsky).Becauseof a
representthe main phase of the volcanicactivity in the lack of geochemicaldata suchclassification schemes have
SiberianTraps,with outpouringof > 90% of the lavas.The notbeenproposedin otherareasof the SiberianTraps.
NizhnyayaTunguskastratigraphy is dividedinto five suites
with a compositethicknessof about 1 km. The present 3. AGE OF SIBERIAN VOLCANISM
review concentrateson the geochemicaldata for the
extmsive rocks from the Putorana and Noril'sk sections.
The age of inceptionand duration of Siberianflood
volcanismare currentlytopicsof activeresearch.The moti-
2.2. Basalt Compositional Magma Typesof the Noril'sk vation for establishing the preciseage of inceptionof the
Region volcanism stems from the possibility that the massive
volcanic activity may be related to the Permo-Triassic
Extensivework in the Noril'sk regionhasshownthatnot extinction event (see below). Early geochronological
every stratigraphic break is associated with a significant investigations usingthe whole-rockK-Ar methodsuggested
changein magma chemistry[Hawkesworthet al., 1995]. that the eruption of the Siberian flood basalts was
Naldrettet al. [ 1992] reviewedthe majorandtraceelement concentrated between 235 and 220 Ma but lasted from 240
data for the Noril'sk area and suggested that five principal to 200 Ma [Zolotukhinand Al'mukhamedov,1988 and
magmatypeswereinvolvedin thegeneration of thelavasof references therein].The first4øAr-39Arinvestigation
wasby
the formationsup to and includingthe Mokulaevsky(see Baksi and Farrat [1991a], who reported whole-rock
Figure2): (1) the Ivakinskyand Syverminsky magmatype incrementalheatingdata for two lava flows from nearthe
of alkalicand subalkalicaffinity, (2) the Gudchikhinsky Ni- bottomand near the top of the volcanicsuccession. They
rich suitewhichincludespicriticbasalts,(3) theprimitivebut concludedthat the volcanismbeganat-238 Ma and lasted
Ni-depletedTuklonskysuite,which is characterized by flat for - 10 m.y. Subsequentrefinementof laboratorytechniques
rare earthelement(REE) profilesand alsoincludespicritic led Baksiand Fartar [1991b] to revisetheir estimateof the
basalts,(4) the Lower Nadezhdinskytype, which is light- age of volcanismto 244-240 Ma. Furtherresultsfrom two
REE-enrichedand has low Nd and high Sr isotoperatios, other4øAr-39Ar laboratories
showedapparent inconsistencies.
and (5) the Mokulaevskytype, which is primitive and has Renneand Basu[ 1991]analyzedwhole-rockandplagioclase
close similarities to the Tuklonsky type. In the Upper samplesfrom the IvakinskyandNerakarskyformations (see
Nadezhdinskyand the Morongovskythere are flows with Figure2) usinglaserincremental heating,and inferredthat
compositionstransitionalbetween the Mokulaevskyand the SiberianTraps eruptedover an extremelyshorttime
LowerNadezhdinsky[Hawkesworth et al., 1995].Naldrettet interval(900,000 D 800,000 years)beginningat about248
al. [1992] also identifiedfive principalgroupsamongthe Ma. Additional data were reportedby Dalrymple et al.
intrusiverocksof the Noril'sk area:(1) thoseof alkalicand [1991, 1995], who found slightlyyoungerages(245-244
subalkalicaffinity, (2) Ti-rich doleritedikesfoundonly in Ma) for the flood basaltsthan for biotitesfrom intrusions
thenortheastern partof theNoril'skregion,(3) doleritedikes cuttingthe basalts(249 ñ 1 Ma). Theseworkersconcluded
and sills found throughout the Noril'sk region, (4) thatthe plagioclasesamplesusedto datethe basaltshadlost
differentiated intrusions not related to the centers of -2% of theirradiogenic
4øArandthatthebasaltswereolder
mineralization,and(5) differentiated
intrusions
presentin the than249Ma. Thediscrepancy in the4øAr-39Ar
agesappeared
vicinityof thecentersof mineralization.On thebasisof bulk to be partlyrelatedto Ar lossor gainandpartlyto theage
 SHARMA 277

the neutronflux monitorusedby the differentlaboratories. Clearly, this interpretationneedsto be evaluatedfurtherby

An additionallimit on the ageof inceptionof the volcanism obtainingnew datafor the lowerpartof the Maimecha-Kotui
wasprovidedby ion microprobeU-Pb datafor zirconsfrom area.

theNoril'skI intrusionthatyieldeda concordiainterceptdate The Maimechinskysuite(see Figure2) from Maimecha-

of 248.0 + 3.7 Ma [Campbellet al., 1992]. The Noril'sk I Kotui has been consideredto be either youngerthan the
intrusion cuts through the lower third of the volcanic Samoedskyformation [Horan et at., 1995; Fedorenkoet at.,
sequencein the Noril'sk region. Accuratedating of this 1996] or correlatedwith the Nerakarsky or Samoedsky
intrusionwould thus give a minimum age of inceptionof formations [Zolotukhin and Al'mukhamedov, 1988]. A
volcanism.The zircondata,althoughsignificantin providing phlogopitefrom an alkalic-ultrabasicintrusion(the Gulin
independent
corroboration
of the4øAr-39Ar
results,
werestill intrusion)associated
with the Maimechinskysuitegave an
tooimprecise
to resolve
theapparent
inconsistencies
in 4øAr- 4øAr-39Ar
ageof 250.4+ 1.3Ma [Renneet at., 1994;Basuet
39Ardata from the three different laboratories.Recently, al., 1995;cf. Datrympteet al., 1995]. This resultled Basuet
Renne[ 1995]showed
that(1) the4øAr-39Ar
basaltdatafrom at. [ 1995] to infer that the terminationof alkalinevolcanism
the threelaboratories
yield consistent
ages(249-246 Ma) if in the Maimecha-Kotui area was synchronouswith the
normalizedto the same referencemonitor standardand (2) initiationof main-stagetholeiiticvolcanismto the southwest
the biotitesanalyzedby Datrympteet al. [ 1991, 1995] have in the PutoranaandNoril'sk areas.This conclusionis subject
excess4øAr.Furthermore,recalculation of the age of the to otherinterpretations
as4øAr-39Ar
datingcannotdistinguish
reference standard with the astronomically calibrated between the inception and terminationof the main-stage
geomagnetic polaritytime scale[Renneet al., 1994]givesan
volcanism.Further, becausethe Maimechinskyformation
ageof 250.0 + 1.6 Ma for the inceptionof the Siberianflood
has been consideredto be either youngeror synchronous
basalt volcanism. This conclusion is consistent w;th
with the highest formationsin the Noril'sk and Putorana
additional4øAr-39Ardataobtainedby Pringteet at. [1995]. sections,it is more likely that the terminationof alkalic
Also, Kamo et al. [1996] publisheda preciseU-Pb zircon volcanismin the Maimecha-Kotuiregion was synchronous
andbaddeleyiteage for the Noril'sk I leucogabbro
of 251.2 with the termination of main-stagetholeiitic volcanism
+ 0.3 Ma, consistent
with the adjusted
4øAr-39Ar
results.In elsewhere.
summary,it appearsthat the bulk of the SiberianTraps Additional limits on the duration of the Siberian Traps
eruptedat around250 Ma, possiblylargelywithin a period eruptionare providedby paleomagnetic data from the
of about1 m.y. The age of inceptionis within errorof the Noril'sk area [Lind et al., 1994]. Thesedata showthat the
estimatedage for the Permo-Triassic
boundary,251.1 + 3.6 lvakinsky formation has reversedpolarity, whereasthe
Ma [Claoud-Longet al., 1991, 1995] and 249.9 + 1.5 Ma youngersuitesof the sequence are characterizedby normal
[Renneet at., 1995].
polarity[Lindet al., 1994;seealsoCampbellet al., 1992].
The alkalic volcanism in the Maimecha-Kotui area may Thisobservation led Campellet at. [ 1992]to suggestthatthe
have been somewhatolder than the main stage of flood eruptionof the SiberianTraps occurredwithin 600,000
volcanism in the Putoranaand Noril'sk regions. This is years,spanning a R-N intervalduringtheIllawaraReversals,
suggested by 4øAr-39Ardatingof anolivinenephelinite from a time of rapid reversalsfrom the latestPermianto the
the lowerpart of the Maimecha-Kotuisectionwhichgavea middle Triassic.However, recentpaleomagnetic work by
plateauage of 253.3 + 2.6 Ma [Basu et al., 1995]. The Mitchell et at. [1994] castdoubton the aboveestimatesof
uncertainty associatedwith the age of volcanism in durationof the flood volcanism.If the Noril'sk stratigraphy
Maimecha-Kotuiis too largeto establishif the volcanismin represents accuratelyall stagesof Siberianvolcanism,then
this region is older than the main pulse of volcanism(= all thesamples fromthePutorana regionshouldhavenormal
250.0 + 1.6 Ma). However,Basu et al. [1995] pointedout polarity.Mitchellet al. [ 1994]foundthatthreesamples from
that the inferreduncertaintyof + 1.6 Ma for the main pulse the southernpart of the Putoranaregionhave a reversed
of volcanismis largebecauseof averagingof datafrom three polarity,suggesting that the flowsfrom thisregionmay be
different labs and that in order to make an intralaboratory younger than their counterpartsin the Noril'sk area.
comparisonthe uncertaintymay be may be reducedto +0.3 Alternatively,the Noril'sk area may in fact be a R-N-R
Ma; the latter is the calculateduncertaintyin the age of a sequence[Mitchellet at., 1994]. Anotherinterpretation of
samplefrom the Ivakinsky formationanalyzedby these the data would be that the inferred correlation between the
workers [Renne and Basu, 1991]. Basu et at. [1995] variousunits (see Figure 2) from the Noril'sk and the
concludedthat the age of 253.3 + 2.6 Ma for the olivine Putoranaregionsis not correct.A much more detailed
nephelinitefrom Maimecha-Kotui is olderthanthe initiation paleomagnetic and geochemicalstudyof the lava pile is
of the main stage of flood volcanismat 250.0 + 0.3 Ma. requiredto differentiateamongthesepossibilities and
278 SIBERIAN TRAPS

obtain an accurate estimate of the duration of the volcanism.

been attributedto the melting of metasomatizedshallow
If the inferredcorrelationbetweenthe Nerakarskyformation mantlecausedby lithosphericstretching,with or withouta
from the Putoranaregionand the Kharaelakhsky formation plume [e.g.,Hawkesworthet al., 1984, 1986; Gallagherand
fromtheNoril'skareais correct(seeFigure2), the4øAr-39ArHawkes-worth,1992]. Yet anothermodelof CFB generation
andU-Pb datingindicatethatan estimated minimumof 75% wasgivenby King andAnderson[ 1995],who suggested that
of the totalvolumeof the SiberianTrapswaseruptedwithin pull-apartof an asymmetriclithosphere(seebelow) could
- 1 m.y. generateconditionsconduciveto CFB volcanism.
Morgan [1981] proposedthat the Siberianflood basalt
4. SIBERIAN FLOOD VOLCANISM AND THEeruptionwas linked to the initiationof the Jan Mayen
PERMO-TRIASSIC MASS EXTINCTION hotspot.Accordingto his plate-tectonic reconstruction,
the
LomonosovRidge could be a part of the hotspottrack.
The Permo-Triassic mass extinction was the most Figure3 is a polarview showinglithospheric platesandthe
catastrophicin the geologicrecord,with as many as 90% of presentlocationsof the SiberianTraps, the Lomonosov
marine speciesand 70% of terrestrialvertebratefamilies Ridge,andthe JanMayenhotspot. A detailedplate-tectonic
dying out [Erwin, 1994]. Whether or not a relationship reconstruction of the openingof the Arctic Ocean is not
exists between this mass extinction and Siberian flood availableat presentto testMorgan'sproposal.Further,it is
volcanismdependsdirectlyon the relativeagesof the two likely that the LomonosovRidge is not a segmentof the
events.As mentionedabove, Campbell et al. [1992] and conjectured hotspottrack as seismic,gravityandmagnetic
Renne et al. [1995] showedthat the bulk of the Siberian data from the LomonosovRidge point to a continental
flood volcanismis synchronous with the Permo-Triassic structureunderlying1-2 km thick sediments[seedokat et
boundary.Renne et al. [1995] arguedthat the volcanism
producedsufficient stratosphericsulfate aerosolsto cause
rapid global coolingwhich in turn led to marineregression.
A short-livedvolcanicwinter was followedimmediatelyby
greenhouse conditions resulting from a buildup of Eurasian Plate
volcanogenicCO2. The environmentalextremaengendered
the massextinction.Althoughthe synchronicity of Siberian ß SiberianTraps
flood volcanismand the Permo-Triassic boundaryappears
established, the above model is not without critics. For
example,Erwin [ 1994] andreferencesthereinsuggested that
the climaticeffectsof SO2emanationsare self-limitingand
may not lead to global cooling.Furthermore,Erwin argued
[quotedby Kamo et al., 1996] that to have been a major
iCELAND
contributorto the massextinction, Siberianvolcanismshould
have somewhat preceded the extinction. The above
argumentsled Kamo et al. [1996] to suggestthat Siberian
Juan de Fuca
floodvolcanismmay havebeena majorobstruction to early Plate

Triassicbiosphericrecoverybut not the ultimatecauseof the

extinction. North Ameri
late

5. ARE THE SIBERIAN FLOOD BASALTS THE

RESULT OF A START1NG PLUME?
Plate

The productionof continentalflood basalts(CFBs) has
beenattributedto the arrivalof plumeheadsfrom the core-
mantleboundary[e.g., Richardset al., 1989; Hill, 1991].
Accordingto the starting-plumemodel, initial rapid and
voluminousCFB eruptionsare followed by decreasing
eruptionratesand,in somecases,the generationof a hotspot
track as a lithosphericplate moves over a relatively
stationarymantle plume. Alternatively,some CFBs have
Figure 3. Polarview showingthe plateboundariesandthe present
locationsof the SiberianTraps,the LomonosovRidgeandthe Jan
Mayen hotspot.The onsetof volcanismon the Siberianplatform
250 m.y. agowasproposedby Morgan [1981] to be a consequence
of the arrival of a plume currentlyactive at Jan Mayen, the
Lomonosov Ridgebeingthepostulated hotspottrackleft duringthe
opening of the Arctic Ocean. However, the plate-tectonic
reconstructionfor 250 Ma is incompletefor lack of data, and at
presentno hotspotcanbe linkedreadilyto the Siberian
 SHARMA 279

al., 1992 and referencestherein].Moreover, similarityof P- King and Anderson[ 1995] suggested that duringtectonic
wave velocity structure determined for the Lomonosov pull-apartof asymmetriclithospheres
(i.e., lithosphereswith
Ridge and that measuredalongthe outerKara and Barents two greatlyvaryingthicknessesbecauseof their relativeages
shelves,along with geomagneticanomaly patternsin the as, for example, younger continentalmargins attachedto
Nansen Basin, suggestthat the ridge is a fragment of Archeancratons)as muchas 104 km3 of magmamay be
continentalmargindetachedfrom Siberiaby normalseafloor generated. The magma production is accomplishedvia
spreadingat the Nansen Ridge [Sweeneyet al., 1982; transportof mantle material from the thicker part to the
Forsythand Mair, 1984; Wilson,1985;Jokatet al., 1992]. thinner part of the lithosphere.According to King and
Numericalmodelingof the mantleplume initiationmodel Anderson[1995], this model can explainmany featuresof
for CFB eventssuggeststhat large-scaleuplift (0.5-4 km) largeigneousprovincesincludingthe rapidturn-onandturn-
shouldoccur in the generalregion of the flood volcanism off of the volcanism,the absenceof uplift in someprovinces,
[e.g., Farnetani and Richards, 1994]. Utilizing these and the presence of many large igneous provinces at
predictions,Renne et al. [1995] suggestedthat a mantle continentalmargins adjacentto Archean cratons. All of
plume head associated with Siberianflood volcanismmight thesefeaturescharacterizethe SiberianTraps.However,the
have causedan uplift of 1-3 km over an area-500 km in volume of magma producedin Siberiais >100 times more
radius.The uplift shouldhave reacheda maximum 5-20 thanthat achievablein King and Anderson'smodel.
m.y. beforethe onsetof the volcanism.However,thereis no Recent studies have noted that the numerous dike swarms
evidence for such uplift in the general area where the and sheeted-dike-likesequencesin the Siberian Traps
Siberianflood basaltsare developedbest [e.g., Kamo et al., indicate that lava eruption occurred in an extensional
1996]. Indeed, the Middle Carboniferousto Late Permian environment[Zonenshain et al., 1990].However,despitethe
TungusskayaSerieswhich is inferred to underliemuch of extensionalconditionsand crustalthinning (8-10 km), the
the Siberian Traps contains some of the largest coal- flood volcanism was not associatedwith significant
measures in the world, with few erosional breaks. These lithosphericrifting [Zonenshainet al., 1990]. Thus, mantle
observationsled Kamo et al. [1996] to conclude that decompression resultingfrom rifting was probablynot the
although the tectonic conditions during Tungusskaya primary cause of widespreadmelting [Zonenshainet al.,
sedimentation were somewhat unstable, sedimentation 1990; seealso Whiteand McKenzie, 1995].
generally compensatedsubsidenceand the area remained Collectively, the above observationssuggestthat the
close to sea level during basin filling. Further, available SiberianTraps eruptioncannotbe linked directly either to
paleontological and geochemicaldataindicatethatthe initial lithosphericstretchingin the absenceof a plume or to
volcaniceruptionsin the Noril'sk regionmay have occurred hotspot initiation. Yet there appearsto be a consensus
underwaterin shallowlakesor lagoons[Fedorenko,1991]. supportinga plume origin among those working on the
Thus, the lack of uplift in the general region of flood SiberianTraps [e.g., $harma et al., 1991, 1992; Renneand
volcanismprovidesno evidencefor a plume underlyingthe Basu, 1991; Arndt et al., 1993; Woodenet al., 1993; Renne
Siberian platform before or during the eruptions.If the et al., 1995; Whiteand McKenzie, 1995; Hawkesworthet al.,
Siberian volcanismhad a plume origin, then the lack of 1995;Fedorenkoet al., 1996]! Two piecesof evidencehave
uplift in the westernSiberianplatformbeforeor duringthe engenderedsuch a confluenceof opinion: (1) the large
eruption of the SiberianTraps must be explained.This volume(> 2 x 106 km3) of magmaemplaced and(2) the
question requires an evaluation of models of plume- shortdurationof-1 m.y. of eruption.Additionalevidence
lithosphereinteractionto explorethe specificissue of connectinga mantle plume and the SiberianTraps comes
whetheror not extensionin the uppermantleand crustcan fromthe3He/4He analysisof theolivinenephelinitefromthe
mitigatethe need for substantial uplift [Fedorenkoet al., lowerpartof Maimecha-Kotui section
whichhasan 4øAr-
1996]. Note that the above argumentsabout a lack of 39Arage of 253.3 + 2.6 Ma [Basuet al., 1995]. Olivine
evidenceof uplift in the westernSiberianplatformrely phenocrysts of thisrockshowed 3He/4Heratiosup to 12.7
implicitlyononekey inference: theTungusskaya Seriesis as times the atmosphericratio (R^), much higher than the
extensiveas the SiberianTraps.This inferenceis basedon averagevalue for mid-oceanridgebasalts(= 8 x RA) [e.g.,
someoutcrops at themarginof thefloodbasaltprovinceand Grahamet al., 1992]. This observation indicatesthatthereis
has to be re-evaluated.For example,if the Tungusskaya a connection between Siberian flood volcanism and an
Seriesis not as extensiveas the SiberianTraps, it can be undegassed mantlesource[Basuet al., 1995], ostensibly a
speculated thatthe sedimentsandassociated coal-measuresplumefromthelowermantle[e.g.,Porcelliand Wasserburg,
weredeposited in grabensdeveloped in a regionthatwas 1995]. A similar resulthas been reportedfor someearly-
upliftedbecause of plumeactivity. erupted alkalic basalts associatedwith Deccan
280 SIBERIAN TRAPS

0 c
Sa
0
Ku

Nr & Kha
Hm & Mu
I
An & Mo
Nd ß o i

Tu
Khch

Gd
Sv
Iv

I i I I I i I

0.4 0.5 0.6 0.7 1 2 3 4 2

Mg* I•0 + Na20 (wt%) TiO2 (wt

Sa o 0 ß o

Ku o o

Nr & Kha • . 0 •
Hm & Mu
An & Mo

Nd o

Tu C• o

Khch o

Gd
Sv
Iv i o-

i I I I I I I

10 20 30 40 50 60 70 1 2 70.00 0.02 0.04 0.06 0.08 0.10

(Ce). (Ce/Yb)N Ta/La

Sa- 0

Ku- 0

Nr & Kha -
Hm & Mu -
An & Mo - II

Nd-
Tu-
Khch -

Gd- o I

Sv-
Iv_ i o, I

i I I I

20 30 60 -8 -6 -4 -2 0 2 4 6 17 18 19
SSr(t) • (t)
 SHARMA 281

magmatism[Basuet al., 1993], which was triggeredby the error bars aroundthe averagevalue for a stratigraphicunit
initiationof the R6union hotspot[e.g., Morgan, 1981]. In reflect 1• variations within that unit. Note that the data from
summary,the large volume of magma erupted,the short the correlated formations from Noril'sk and Putorana are
durationof eruptionand, consequently, the high eruption placednextto eachother.Thereappearsto be no significant
rates(-1.7 km3 yr-• [RenneandBasu,1991]),areconsistent geochemicaland isotopicdifferencebetweenthe correlated
with a starting-plumeorigin for the SiberianTraps.Further, unitsfrom the Noril'sk and the Putoranaregions(Figure4).
the high3He/4Hesignature in an earlylavaalsoindicates a Some notable features of the geochemicaland isotopic
hotspotorigin.However, whetherthe SiberianTrapsand Jan stratigraphy are listedbelow.
Mayenhotspotare relatedis not clearat present. (1) The oldestlavasof the Ivakinskyformationshowthe
lowestMg* = 0.39 (molarMg* = Mg/(Mg + 0.85*Fetot) ) and
6. TEMPORAL GEOCHEMICAL AND ISOTOPIC the highestTiO2 content(-- 2.7 wt.%); the younger,volu-
VARIATIONS minous lavas from Putoranaand Noril'sk display Mg* =
An assessment of temporalgeochemical andisotopicvari- 0.55-0.58 and TiO2 = 1.2-1.4 wt.% (Figure4a and c). None
ationsin a floodbasaltprovinceis vital in understanding the of the lavas show Mg* >0.79 that would suggest
compositional evolutionof the magmasandthe variationsin equilibrationwith mantleolivine [e.g.,Albarbde,1992]. The
sourcecomposition.Furthermore,data on these variations lavaswith the highestMg* comefrom Gudchikhinskysuite
can be combinedwith estimatesof the volume of erupted and typically have MgO >10 wt.% [Sharmaet al. 1991;
magma to establishthe relative roles of various crust and Woodenet al., 1993].
mantle reservoirs and to evaluate models of flood basalt (2) The lavas becomeless alkaline from the baseof the
genesis.Numerousmajor and trace elementdeterminations lava pile upward with decreasing(Ce)N and decreasing
as well as Sr-, Nd-, and Pb-isotopicanalysesare now (Ce/Yb)N (here the subscriptrefersto chondrite-normalized
availablefrom the SiberianTraps. Althoughmost of the values) (Figure 4b, d and e); the bulk of the SiberianTraps
samples havecomefromthenorthernandnorthwestern parts (Ayansky through Nerakarsky in the Putorana and
of the SiberianTraps(Noril'skandnorthernPutorana),they Morongovsky through Kharaelakhsky in theNoril'sk region)
are adequateto derive first-order conclusionsabout the has an average total alkali content of-2.5, (Ce)s-10 and
evolutionof magma sources.Detailedaccountsof element (Ce/Yb)• -2.
and isotopicvariationsin stratigraphically controlleddrill (3) With the exceptionof the Gudchikhinskypicrites,the
core and outcropsamplesmay be found in Lightfootet al. Ta/La ratio of the bulk of the SiberianTraps is distinctly
[1990, 1993], Sharma et al. [1991, 1992], Woodenet al. lower than the primitivemantleTa/La ratio of 0.06 [Sunand
[1993], Hawkesworthet al. [1995], and Fedorenkoet al. McDonough, 1989] (Figure4f).
[1996]. Theseworkersalso discussed detailedpetrogenetic (4) The averageesr(t) valuesof suitesthroughoutthe lava
modelsfor the SiberianTraps.For the followinganalysis,the pile are +8 to +51 (1 e unit = 1 part in 10,000relativeto the
data are obtainedfrom the above studies.In addition,some estimatedbulk-earthisotoperatio at time t; the bulk earthis
unpublishedtrace and rare earth element data from the assumed tohave(87Sr/SSSr)250m y.= 0.7045; Figure4g). Apart
Putorana are also used (A.R. Basu and M. Sharma, from the Khakanchansky basaltictuff, the meanesr(t)values
unpublished data). The completedata set may be obtained of the early lavas(IvakinskythroughTuklonsky)rangefrom
fromtheauthor.Someselected dataarepresented in Figures +20 to +30. The Nadezhdinskybasalts,which eruptedafter
4 through13. the Tuklonskypicrites,displaya sharpincreasein the mean
Figure4 illustratesstratigraphic variationsin the average esr(t)value (to +51). The bulk of the SiberianTraps lavas
valuesof selectedoxidesand in elementand isotopicratios showmeanesr(t)valuesdecreasing progressively up-section
in Noril'sk (open circles)and Putorana(filled circles).The (Figure4g).

Figure4. Stratigraphic
variations
of selectedmajorelements, traceelements,
elemental ratiosandisotopesin the Siberian
Traps.Data from the Noril'sk and Putoranaregionsare plottedas openand filled circles,respectively. Y-axis: Iv =
Ivakinsky,Sv = Syverminsky, Gd = Gudchikhinsky, Khch= Khakanchansky, Tu = Tuklonsky,Nd = Nadezhdinsky, An =
Ayansky,Mo = Morongovsky, Hm = Honnamakitsky, Mu = Mokulaevsky, Nr = Nerakarsky,Kha = Kharaelakhsky, Ku =
Kumingsky,Sa - Samoedsky. The data from An, Hm and Nr suitesof the Putoranaregionare combinedwith the
correlatedunitsMo, Mu, and Kha in the Noril'skregion.This legendwill be followedfor Figures5 through13. Only
averagesand 1, variationsareplottedfor eachof the stratigraphic
units.Them are no significantgeochemicalvariations
betweenthe correlatedunitsfrom the Putoranaand Noril'sk regions.The subscript"N" in panelsd and e refersto
chondrite-normalized
values.The verticallinesin panelsf andh showthe positionof the primitivemantle[afterSunand
McDonough,
282 SIBERIAN TRAPS

(5) The averagel•Nd(0 values(calculated

assuming
bulk 8a

earth
(•43Nd?nnNd)2s0m.y.
-- 0.512316)
show
widevariability
in Ku
Kha & Nr
the lowerpartof the lavasequence (-7 to +4; Figure4h). The Mu & Hm
bulk of the Siberianlavashave ENd(t) values-+2 (Figure M¸ & An
4h). Nd

(6) The (2ø6pb/2ønpb)t

ratiosof theentireSiberianlavapile Tu
Khch
appearto varyonlymodestlyfrom 17.3to 18.3(Figure4i). ß Gd
Geochemicalstudiesof flood basaltprovincesassociated ß Sv
with the breakupof Gondwanahaverevealedthe existence ß Iv

of two typesof basalt:a low-Ti and a high-Tibasalt[e.g.,

Bellieni et al., 1984; Marsh, 1987; Ellam and Cox, 1991;
Hergt et al., 1991]. The SiberianTrapsalso displaylarge
variationsin TiO2 contentsfor a given Mg* (Figure 5).
Lightfootet al. [1993] dividedtheultramaficandmaficlavas
intotwo broadgroups:a lowerassemblage with highTi and
an upperassemblage with low Ti. The high-Tisuitesinclude
the Ivakinsky, Syverminsky,and Gudchikhinskyand the
low-Ti suitesincludethe rest of the formations(Figure 5). 1 2 3 4 5 6
The low-Ti andhigh-Ti groupsarenot relatedby fractional
crystallizationof similarphaseassemblages from a uniform La/Sm
parentalmagmaastheyhaveoverlapping rangesin Mg* but Figure 6. Variationsin La/Sm versusGd/Yb. Data from Lightfoot
differentTiO2 contents.The distinction betweenthe high-Ti et al. [1993], Woodenet al. [1993], Hawkesworthet al. [1995], and
andlow- Ti groupsbecomesquiteclearwhenGWYbratiois M. SharmaandA.R. Basu(unpublished data).The arrowpointsto
plotted against La/Sm ratio [Lightfoot et al., 1993; the locationof one datapoint from the Ayanskyformationwhich
Hawkesworthet al., 1995] (Figure6). Figure6 showsthat, has extremelyhigh Gd/Yb ratio and lies outsidethe diagram.A
Gd/Yb ratio of 2.0 is chosenarbitrarilyto distinguish the low-Ti
andhigh-Tigroups[afterLightfootet al., 1993].
I I I I
© Sa
¸ Ku
with very few exceptions,the high-Tigroupis depletedin
[] Kha & Nr
/', Mu & Hm
heavyREE (Yb) relativeto middleREE (Gd) in comparison
4 - Low-Ti
,7 M¸ & An to the low-Ti group.Accordingly,Lightfootet al. [1993]
ß ß
ß ¸ Nd useda GWYb ratio= 2 to demarcate the two groups(Figure
ß ¸ Tu 6). The differencesin the GWYb ratio may suggest different
[] Khch
3- depthsof meltingfor the two groups(seebelow).Further,
ß Gd
the high La/Sm ratios(>3) of the Ivakinsky,Syverminsky,
High-Ti ß Sv
and Nadezhdinskysuitesmay reflect continentalcrustal
ß....AA ßß ... ß ,v contamination [e.g.,Lightfootet al., 1993;Hawkesworth et
2 ß
al., 1995] (Figure6).
Melts derivedfrom primitivemantleare expectedto have
a Th/Ta ratio of-2.3 (Figure 7a) [Sun and McDonough,
1989;seealsoWoodenet al., 1993].BecausebothTh andTa
are highlyincompatible elements,the meltsareexpected to
¸ ¸ ¸ retain their Th/Ta ratio during closed-system fractional
crystallization and evolve alongthe solidline in Figure7a.
0.3 0.4 0.5 0.6 0.7 0.8 0.9 Most of the low-Ti and high-Ti basaltsappearto have
evolvedwith Th/Ta ratiosthat are quitedifferentfrom each
Mg* other (Figure 7a). Also, the Th/Ta ratiosof the lavasare
Figure 5. Plot of TiO2 (wt%) versusMg* after Lightfootet al. much higher than expected from the closed-system
[1993]. Mg* = molar Mg/(Mg + 0.85*Fetot).The distinction crystallization of primitive-mantle-derived melts.The only
betweenthe low-Ti and high-Ti basaltsis indicatedapproximately exceptions to this rule are some picrites from the
by the solidline. Gudchikhinskyformation(Figure 7a). The high-Ti
 SHARMA 283

Log (Ta/Yb)
-1 2 -0.8 -0.4 0.0 0.4
I I I

0.4

i i i i I i ' ' I

I
ß

Th/Ta-6.7 ß 0.0

-0.4

/Ta - 3.3
I _
-0.8
Sa
¸ Ku
r• Kha & N r
Primitive Mantle
Mu & Hm Low-Ti
(Th/Ta - 2.3)
Mo & An
1 w • Nd
D a
c Tu
Khch
ß Gd
0 1 2 3 4
B Sv High-Ti
Ta, ppm A Iv

Figure 7. (a) TantalumversusTh concentrations in the Siberianlavas.The low-Ti andhigh-Timagmasevolvedwith

differentinitialTh/Taratios.Interestingly,
theTh/Taratiosfor nearlyall, including
theleastcontaminatedlavas,aremuch
higherthantheprimitivemantleTh/Taratio[e.g.,SunandMcDonough, 1989].(b) Log(Th/Yb)versus log(Ta/Yb).Most
datafor the low-Ti andhigh-Tibasalts plot alongtwo differenttrendswith slopes>45ø, indicating
(1) the derivation
of
magmasfrom sources with two differentTh/Ta ratiosand(2) thatbothtypesof basaltsarecontaminated by continental
crust,whichhashighTtgTa ratios.

displaywide variationsin Th/Ta ratioswhichareashighas4 versuslog (Th/Yb). On this diagram,magmasoriginating

for someIvakinskyand Syverminskysamplesandas low as from a common sourceand evolving in a closedsystem
2.3 for some Gudchikhinskysamples.On the other hand, definean array with a slopeof 45ø [Russelland Cherniak,
most of the low-Ti sampleshave relativelyconstantTh/Ta 1989]. In Figure7b, the low-Ti and high-Ti groupsdefine
ratiosover a rangeof Th and Ta concentrations, suggesting two distinctarrays(with someexceptions),
eachwith a slope
generationof theselavasvia fractionalcrystallizationof a >45ø. As the average upper continentalcrust may have
magmasourcewith Th/Ta-6.7 [cf. Woodenet al., 1993]. In Th/Ta as high as 10 [e.g., Condie, 1993], the observed
detail,however,the Th/Ta ratiosof bothhigh-Ti and low-Ti enhancement in the Th/Ta ratios of Siberian lavas is
groupsincreasesomewhatwith increasingTa concentration. consistentwith contaminationof magmaswith continental
Thisis illustratedin Figure7b, whichis a plot of log (Ta/Yb)
284 SIBERIAN TRAPS

elementpatternsin flood basaltprovinces incompatibleelements for each of the 11 units in the

Incompatible
have been used extensivelyto assessthe natureof magma combined Noril'sk and Putorana sequences.The trace
sourcesand contaminants.Figure8 illustrates
the primitive- element patternsare comparedwith averagesof selected
mantle-normalizedaverages of moderately to highly formationsfrom other flood basalt provincesas well as
normal mid-ocean ridge basalts (N-MORB) [Sun and
McDonough, 1989], ocean island basalts(OIB) [Sun and
McDonough,1989], and the continentallithosphericmantle
as sampledby spinel harzburgitexenoliths[McDonough,
1990] (Figure 8c). Additionally,mantle-normalized patterns
of the averageuppercontinental crust(UCC) [Condie,1993]
andBolgokhtokhskygranodioriteintrusion[Hawkesworth et
© Sa al., 1995] are given in Figure 8a; the latterwas assumed by
¸
[]
Ku
Kha & Nr
Hawkesworthet al. [1995] to reflectthe compositionof the
RbBa K ThTaLaCePbSrNd P SmZr HfEu Ti Y • Mu & Hm
Low-Ti
local continentalcontaminantin the Noril'sk region.Both
100 x7 Mo & An
high-Ti and low-Ti Siberianlavasshowhigh concentrations
b' ½Nd
o
[]
Tu
Khch
of large-ion-lithophileelements(Rb, Ba) relativeto high-
ß Gd field-strengthelements(Ta, P, and Ti) (Figure8). The trace
High-Ti
ß
ß
Sv
Iv
elementcharacteristics of the high-Ti basaltsappearto be
intermediatebetweenthoseof the Ambenali (Deccan)and
Urubici (Paranti;cf. Figures8a and 8c). None of the suites
displaythe relativelysmoothpatternsshownby average
1
OIB, the Urubici and the Ambenali, however. All Siberian
RbBa K ThTaLaCePbSrNd P SmZr HfEu Ti Y low-Ti basaltsshow(1) enrichments
in Zr and(2) depletions
100
in Ta relative to La and (3) Ti/Zr = 32-117. These
c characteristics are similar to those observed in the low-Ti
•
-•--
Urub•c•
OIB
(Paranah•h-T0 Gondwananbasalts.Such featureshave been arguedto be
• 10 •
-•-
Ambenah,
DeccanTraps
Tasmanm
(Ferrarlow-T0
distinctive of continentallithosphericmantle containing
o
z •
•
N-MORB
L•hosphenc
mantle
small amountsof subductedsediment[e.g., Hergt et al.,
(sp Iherz xenohths) 1991]. An importantdistinctionbetween low-Ti Siberian
1 lavas and their Gondwanancounterpartsis that the former
show low Rb/Ba ratios (except the Nadezhdinsky)•a
RbBa K ThTaLaCePbSrNd P SmZr HfEu Ti Y feature that could result from contamination with a
Bolgokhtokhsky granodiorite-type melt [see also
Figure 8. Primitive-mantle-normalized incompatible-element plot Hawkesworthet al., 1995]. However,in contrastto the low-
(primitive-mantlevaluesfrom Sunand McDonough,1989).For the Ti Gondwananbasalts,averagelithosphericmantle,average
sakeof clarity,the data are presentedin threedifferentpanels:(a) uppercontinental crust,or Bolgokhtokhsky granodiorite,
the
averagesof high-Ti basaltsuites,averageupper continentalcrust low-Ti Siberian lavas (except the Khakanchanskyand
(UCC) [Condie, 1993], and Bolgokhtokhsky granodiorite
Nadezhdinsky) do not show pronouncednegative Ti
[Hawkesworthet al., 1995]; (b) averagesof 1ow-Ti basaltsuites;
and(c) averageUrubicimagmatype (representing the Parantihigh- anomalies(Figure8).
Ti lavas)[Peate,this volume],oceanislandbasalts(OIB) [Sunand Sharmaet al. [ 1991, 1992] studiedrepresentative samples
McDonough, 1989], the least contaminatedunit from the Deccan from theNoril' sk andPutoranaregionsandfoundthat(1) on
Traps (Ambenali Formation) [Lightfootet al., 1990; Peng et al., an gSr-gNd diagram,a majorityof the rocksfall in the region
1994], the averageof low-Ti Ferrar basalts[Hergt et al., 1991], of OIB and (2) the bulk of the SiberianTrapshasgNd(t)-•+2
normalmid-oceanridge basalts(N-MORB) [Sunand McDonough, and gsr(t)-+7. Further work by Lightfootet al. [1993],
1989], and the averagecontinentallithosphericmantleas sampled Wooden et al. [1993] and Hawkesworthet al. [1995]
by spinel harzburgitexenoliths [McDonough, 1990]. The trace substantiatedthese observations.Figure 9 shows the
element patternsof most of the Siberian lavas are betweenthe
availableSr andNd isotopicdata.Also shownin the diagram
OIB/Paranfihigh-Ti basaltsandAmbenaliFormationpatterns.Also,
unlike their Gondwanancounterparts (e.g., low-Ti Ferrarbasalts)
are the fields of present-day PacificMORB and OIB. Note
most of the 1ow-Ti Siberian lavas do not show a pronounced that datafor 18 of 26 high-Ti basaltsdo not fall in the OIB
negativeTi anomaly.The incompatibleelementpatternsof the field. In contrast,data for most of the low-Ti basalts(the
Siberian Traps are not diagnosticof derivationfrom either a Morongovsky, Mokulaevsky, Kharaelakhskyand their
MORB-type or an OIB-typemantlesource. Putoranacounter-parts) fall in the OIB field.
 SHARMA 285

12 I|' I ' I ' • ' ! ' I ' I ' I '

\Pacific

8 -\ \ .,---, Modern Ocean

',,..•,.; ",,ß /IslandBasalts
© Sa
4
© _e ¸ Ku
- '•',, El. ', ß• u Kha & Nr
/• Mu &Hm Low-Ti
v •7 Mo &An
o -BulkEarth "i '• i •. "-. '; • Nd
¸ Tu

-4 '-'I',,,,,
,•,,
•',,,• •, rq

m
Khch

Sv High-Ti
& Iv
<•
-8

, , I , I ,
-12 [ ' • ' [
-30 -20 -10 0 10 20 30 40 50 60 70 80

:•Sr(t)
Figure9. Plotof gSr(t) versus
gNd(t) fortheSiberian floodbasalts.
Datasources:
Sharmaeta/. [1991,1992];Lightfoot
et
al. [1993];Woodenetal. [1993];Hawkesworth etal. [1995].Fieldof modemoceanislandbasalts
(OIB) isfromHart and
Staudigel[1989].ThePacificMORBfieldis defined by datafromMacdougallandLugmalt[1986],Whiteet al. [1987],
andMahoneyet al. [ 1994].

the early, volu-metricallynegligible eruptions(Ivakinsky modemOIB andMORB. Data for all the basalts,excepttwo
throughNadezhdinsky)have the extremeesrand •Ndvalues. of the Tuklonskysuite,fall well to the right of the geochron
Equally remarkable is that the voluminous late-stage in Fig 10a, indicatingtheir derivationfrom a mantlesource
eruptions (Morongovsky and Ayansky through with multi-stageevolution of the U-Pb system[see also
Kharaelakhskyand Nerakarsky)displaya relativelynarrow $harma et al., 1992]. In comparisonwith the bulk of the
rangeof •Ndvalues.Anothercuriousaspectof the datais that Siberianbasalts,most of the Gudchikhinskylavas have
the isotopicextremesare confinedto one or two suitesand higher2ø7pb/2ø4pb
for a given2ø6pb/2ø4pb.
Theroughpositive
the datado not show a continuousmixing arraybetweenthe trend shownby the low-Ti lavas appearsto be consistent
extremes. For example,the highest•Ndand the lowestesr with mixingwith a contaminant
havinga low time-integrated
valuescomefrom the Gudchikhinsky picrites.Similarly,the U/Pb ratio, possibly lower crust. Mixing with a
206 04
samplesfrom the Nadezhdinsky(and a sample from the Bolgokhtokhsky granodiorite-type
melt(with( Pb/2Pb)t=
Morongovskyand Ayansky) show the highestes• and the 16.29 and (2ø7pb/2øqPb)t-15.3) couldalsoproducethe
206 04
lowestend. observed trend[Wooden
et al., 1993].Onthe( Pb/2 Pb)tvs
208 04
The Pb isotopic compositionsof the low-Ti Siberian ( Pb/2Pb)tdiagram(Figure10b),the Siberianbasaltdata
basalts
definea roughpositivetrendonthe(2ø6pb/2ø4pb)t
vs lie slightlyabovethe NorthernHemisphereReferenceLine
(2ø7pb/2øqPb)t
plot(Figure10a).Thebulkof theSiberian
data (NHRL) [Hart, 1984] and in the fields of MORB and OIB.
lie in a clusterat the lower end of the array deftnedby Unlike the lavas from Paran•, Deccan, and Columbia
286 SIBERIAN TRAPS

15.8
compositional variationsof the MORB-typeand OIB-type
mantlereservoirsare grosslyidentified[e.g., Zindler and
15.7
q•/ Modern
OIB •
Hart, 1986], the CLM reservoir appearsto be very
-Q 15.6 heterogeneous [e.g., McDonough,1990]. The problemis
Q_
•1'
0
furthercompounded as magmasmay evolveandmay get
0,1 15.5 contaminated by continentalcrustalmaterialen routeto the
© Sa surface.In thissection,temporalandspatialvariations in the
0 15.4
/ Modem
MORB
] O
[]
Ku
Kha & Nr chemicalandisotopiccomposition of the SiberianTrapswill
A Mu & Hm
be usedto assess mantlesources, magmadifferentiation, and
Bolgokhtokhs•
Granod•ffie
v Mo &An
15.3
(16 29,15 3) O Nd crustalcontamination. Specifically,the followingquestions
, I .... I , , , , I , , , , I , , , ,
O
[]
Tu
Khch
are significant:(1) whetherthe bulk of the SiberianTraps
' ' ' I ' ' ' • I .... I .... I ' ' '•
ß Gd displaysevidenceof substantial crustalcontamination, (2)
ß
ß
Sv
Iv
whetheror not a relationshipexistsbetweenthe chemical
evolution of a lava suite and the extent to which it is contam-

-Q 38 inated,and (3) the mechanismof crustalcontamination. A

Q_

0
•1' relatedissue is whetherthe primary magmas were picriticor
C,,I basaltic[cf.BasalticVolcanism StudyProject,1981].
As explainedabove,the low-Ti andhigh-Tigroupsin the
0 37
'- - - • -••Modern MORB SiberianTrapscannotbe relatedby fractionalcrystallization
oo /.• -/
•lgokhtokh,•
/ ½' of similar phase assemblages from a uniform parental
v

•Granodiorite /

(16
29.•½ b•
magma(Figure 5) [Lightfootet al., 1993]. Further,in com-
parisonto the low-Ti group,the high-Ti groupis relatively
36
16.5 •7.0 17.5 •8.0 •8.5 •9.0 depletedin heavy REE (Figure 6), suggesting different
depthsof generation. Lightfootet al. [1993]showedthatthe
(206pb/20•Pb)t GWYb ratios of high-Ti lavas are consistentwith their
generationat garnetstabilizationdepths.In contrast,
the low-
Figure 10. Plot of (206pb/204Pb)t versus(207Pb/204Pb)t (a) and
Ti basaltswere likely generated by partialmeltingof spinel
(208Pb/204Pb)t (b) in Siberianbasalts.The datapointsfall in the
fields of OIB and MORB and trend toward the field of DUPAL peridotite.These observations may suggestmeltingof a
OIB. Most of the Siberiandatafall to the rightof the geochronat commonsourceat two different depthsto producethe
250 Ma in panel(a), indicatinga multistageU/Pb evolutionof the Siberianlavas[e.g.,Arndt et al., 1993; Woodenet al., 1993].
sourcematerial.Note also that the Bolgokhtokhsky granodiorite Alternatively,two different sourcesmay have melted to
could be a contaminant.Data sources:Sharma et al. [1992]; producethe 1ow-Ti and high-Ti groups[Lightfootet al.,
Woodenet al. [1993]. Fieldsof OIB, MORB andDUPAL OIB are 1990, 1993;Fedorenkoet al., 1996]. Figure7 showsthatthe
afterHart, [1984],ZindlerandHart [1986],lto et al. [1987],Storey least-contaminated lavas(i.e., thosewith the lowestTh/Ta)
et al. [1988], le Roexet al. [1990], Barlingand Goldstein[1990], of the low- and high-Ti groupshave two distinctTh/Ta
Dossoet al. [ 1991], Mahoneyet al., [1992, 1994], and Woodhead
ratios. As Ta and Th are not expectedto fractionate
andDevey[ 1993].NorthernHemisphere Reference
Line (NHRL) is
from Hart [ 1984]. significantlyfrom each other during moderateto high
degreesof partial melting and not at all duringfractional
crystallizationprocesses,this result suggestsderivationof
magmafrom two distinctsources.The main problemwith
provinces,none of the Siberianbasaltsfall in the modem this interpretationis that the Th/Ta ratiosdisplayedby the
DUPAL OIB field.
least-contaminatedlavas of the low-Ti and high-Ti groups
are both higher than the estimatedprimitivemantlevalue.
7. MANTLE SOURCES, CRYSTAL FRACTIONATION Thus, even theselavas may be slightlycontaminated. This
AND CONTAMINATION observationindicatesthat the following two possibilities
cannotbe ruled out: (1) the magmaswere derivedfrom a
The problem of assigningthe roles of differentmantle common sourcebut were contaminatedby two different
sources in CFBs is a non-trivial exercise as at least three contaminants and (2) the magmaswere derivedfrom two
mantle reservoirsmay be implicatedin the generationof sources and were contaminated by two different
flood basalts:(1) a MORB-type mantle, (2) an OIB-type contaminants.
mantle which may also be undegassed,and (3) the Figure11 showsthe variations
of •47Sm/]44Nd
ratioswith
continental lithospheric mantle (CLM). Whereas the SNd
(t) in the SiberianTrapsandcompares
themto a solid
 SHARMA 287

....... 1ow-Tigrouphas a volume-weightedsNa(t) value of about

+2.0 [Sharmaet al., 1992]. However,it may be arguedthat
5
Part,al
rnelt,
ng/
_ fractional xln

continentalcrustal contaminationmay have uniformly

©
¸
Sa
Ku
lowered
theSNd
values
of the1ow-Ti
group;
twofeatures
of
the low-Ti lavas stronglysupportthis argument:(1) an
•7- IBulk
Earth
[] Kha & Nr
/x Mu & Hm Low-Ti
Mo & An extremelyhigh Th/Ta ratio of-6.7, which is about three
¸
O
Nd
Tu
times higher than that of the primitive mantle, and (2) a
[] Khch Ta/La ratio of 0.04, which is about 30% lower than the
ß Gd

i
ß Sv High-Ti primitivemantlevalue of 0.06.
-10
i
ß Iv
The Ta/La ratio, in conjunctionwith the observed•Nd(t)
I• Granuhtel
values,pointsto the natureof the magmasource(s)and the
' Granite
extentof crustalcontributionin the caseof the low-Ti group
05 007 009 0.11 013 015 0.17 019 021 0.23 0.25
[see also Hawkesworthet al., 1995]. To illustratethis, a
147Sm/144Nd
simplebinarymixing calculationis usedfor two cases'(I)
Figure 11. Relationshipsbetween 147Sm/144Nd basalticmagma,Ta = 0.1 ppm,Ta/La = 0.06, Nd = 9 ppm
and SNd(t) in
and SNd-- +4; (II) basalticmagma,Ta = 0.1 ppm, Ta/La =
Siberian basalts.First-ordereffects of partial melting/fractional
crystallization(xln) and contaminationcan be assessed 0.06, Nd = 9 ppm and SNdm +8. A tonaliticcontaminant
usingthis
diagram.The high-Ti and low-Ti basaltssuggestmixing between [Condie, 1993; see also Woodenet al., 1993] is assumed,
two mantlememberswith positive•Nd valuesandtwo continental with Ta = 0.7 ppm,Ta/La = 0.02,Nd = 22 ppmandSNd '-' -
crustal contaminantswith low Sm/Nd and negative •Nd. 11. WhenSNd for themagmais +4 (caseI), additionof about
Alternatively, melting of a single mantle source, first at high
6% of thecontaminant givesriseto a mixturewith •Nd'-'+2.0
pressureand then at low pressure,couldgeneratemagmasthat, on
variablemixing with continentalmaterial,would producethe two
andTa/La = 0.04. In contrast,whenSNd for the magmais +8
trends.Data sourcesare the same as in Figure 9. See text for(caseII), no proportionof this contaminant andpostulated
discussion andfor explanationof line andarrows. magmacanreconcilethe observedSNd andTaYLa.The results
remainthe samewith otherlikely crustalcontaminants and
indicatethat if the magmahasSNd •- +8, only contaminants
that representsmixing of N-MORB, bulk-earth-type mantle with SNd values<-37 (> 3.3 Ga continental crust)wouldbe
and an averagesediment(not shown).The high-Ti and low- ableto reconcilethe observed SNd andTa/La. Recentstudies
Ti groupsdefinetwo broadarrayswith positiveslopes.The in the Aldan shieldsuggest the presence of predominantly
horizontalcomponentof the arrayssuggests variabledegrees <3.0 Ga continentalcrust and extensivereworkingof
of partial melting of the mantle source(s)and/or fractional continentalmaterialin the mid-Proterozoic [e.g.,Nutmanet
crystallization.
In comparison to the low-Ti lavas,the high- al., 1992 and referencestherein].Thus it is concludedthat
Ti basalts,in general,could have beengeneratedby lower thesourceof the low-Ti magmashadan SNd valueof- +4. It
degreesof partial melting of the mantle source.This ob- followsthat the mantlesourcesof the high-Ti and low-Ti
servationis consistent with the Gd/Yb datain Figure6. The groupsare different.Note that the above argumentsare
positivecorrelationin Figure 11 wouldresultfrom contam- basedon forwardmixing calculations that are not unique.
ination of mantle-derivedmagmas(with high Sm/Nd and Nonetheless, the conclusions appearto be the simplest
•Nd)with continentalcrustalmaterialsuchas granulite(with explanationfor the data.
low Sm/Nd and end).The two arraysin Figure 11 allow the What was the mechanismthat governedthe high Th/Ta
possibilityof magma derivationfrom two mantlereservoirs ratiosof the high-Ti andlow-Ti magmas?Was it mixingof
with •Ndvaluesof approximately+4 and+8. Further,the two variableamountsof subducted shalewithin the lithospheric
arraysallow the possibilityof more than one contaminant. mantle [Lightfootet al., 1990, 1993; Hawkesworthet al.,
The latter observation is consistent with the evolution of 1995], or was it contaminationwith two continentalcrustal
high-Ti and low-Ti groupswith two distinctTh/Ta ratios componentswith different Th/Ta ratios [Wooden et al.,
(Figure7). 1993]? This issueis difficult to resolveas the trace element
Alternatively, the data could suggestderivationof the contentsand Sr-, Nd- and Pb-isotopiccompositionsof
high-Ti groupby lowerdegreesof partialmeltingof a source subducted shaleand sometypesof continentalcrustmay not
with •Ndof +8; high degreesof partial melting of such a differ substantiallyfrom each other. Specifically,mantle
sourceat shallowerdepthscouldproducethe low-Ti basalts. containingsubductedshalemay have negativeTi, Ta, and
The main difficulty with this interpretationis that the SNd Nb anomalies,high Th/Ta ratios, radiogenicSr and Pb
valuesof noneof the low-Ti samplesexceed+4. Indeed,the isotopes,and unradiogenic Nd isotopes.The
288 SIBERIAN TRAPS

Siberian lavasdisplayall theabovecharacteristics except In summary,the trace-element andisotopicdatafor the

that they have relativelyunradiogenic Pb isotopic SiberianTrapsareconsistent with derivation of magmas
compositions, pointingto a contaminant with low time- fromtwosources. Thehigh-Timagmas weregenerated by
integrated U/Pb ratioas the mostsignificant contaminant.lowdegreesofpartial
melting ofamantle source withend --8.
Sucha contaminant is generallyconsidered to be the lower Largerdegrees
ofpartialmelting
of a magma sourcewithend
continental crust[seealsoWooden et al., 1993].Thus,the --+4produced thelow-Tilavas.Thetwo groups appear to
simplestexplanationof the datais thattherearetwo different haveevolved withtwo distinctTh/Taratiosmuchhigher
crustal contaminants for thehigh-Tiandthelow-Tigroups. thantheprimitivemantlevalue,suggesting (1) theinfluence
If it is assumedthat the contaminant is continentalcrustwith of two differentcontaminants and (2) either early
end = -11 (which is the lowest value observed in a contamination
or bufferingof Th/Taratios[Wooden et al.,
Nadezhdinsky sample;seeFigures9, 11) anda ]47Sm/144Nd1993].Therelatively unradiogenic Pb isotopic composition
ratioof 0.11, thenthe minimumageof the contaminant is of the contaminated lavas appearsto favor the lower
--109 y. If it is assumedthat the othercontaminant is continental crust as the dominant contaminant.
continentalcrustwith end= -5 (whichis the lowestvalue Althoughthe Mg* of the Siberianlavasvariesfrom 0.30
observed in an Ivakinskysample;seeFigures9, 11) anda to 0.70, the bulk of the rocksshowa ratherrestricted range
]47Sm/•44Nd ratioof 0.11,thentheminimum ageof this (-0.54 to 0.56). The Mg* is much lowerthanexpectedfor
contaminant is-450 Ma.
theprimarymagmasof MORBsandOIBs,whichmaybe as
Variationsof eSr(t) and (2ø6pb?ø4pb)t for the Siberian high as -0.81 [Albarbde,1992]. Recent measurements of
basaltsareplottedin Figure12 andshowa broadnegative primary melt inclusionsin MORB olivinesfound Mg*
trendwithsomesamples fromtheTuklonsky suitefallingto varyingfrom 0.73 to 0.78 [Sobolevand Chaussidon, 1996].
theleft of thistrend.Thesedatapointto thepresence of two The Ni contentsof primarymelt inclusionsin OIBs couldbe
contaminants withinthelow-Tigroup:onewith2ø6pb/2ø4Pb < greaterthan 600 ppm [Clagueet al., 1991]. If the Siberian
6.7,esr-20,andtheotherwith2ø6pbf2ø4pb 517.2,esr> +70. Traps basaltswere primary,they would reflectderivation
ThePbandSrisotopic compositions of theBolgokhtokhskyfrom a mantle more Fe-rich than the normal MORB source.
granodiorite are consistent
with it beinga contaminant for Indeed, a deep mantle sourcewith abnormallyhigh Fe
the Tuklonskylavas.However,the Sr isotopiccomposition contentshasbeeninferredfor the plume-derived lavasat the
of theBolgokhtokhsky granodiorite(- 0.70634)istoolowto Isle of Skye [Scarrowand Cox, 1995]. The Siberianlavas,in
achieve the Sr isotopic shift required of a suitable general,have Ni contentsvarying between 150 and 200
contaminantfor the Nadezhdinskysuite (Figure 12) ppm. Thus, comparedto a primarymelt with Mg* = 0.78
[ Woodenet al., 1993]. and Ni = 600 ppm, the Siberian lavas appearto have
undergoneabout 30% reductionin Mg* and Ni content.
Takentogether,the aboveconsiderations suggest
thatolivine
fractionation,andnotanunusualmagmasourcewithhighFe
content,is responsiblefor the low Mg*. If the parental
50 • ov liquids of the SiberianTraps had Mg* of about0.78, the
ßß ,,,{% o Ku&Nr i=
[] Kha extentof olivine fractionation
neededto reducethe Mg* to
• ......(16
w29261)
30 -Oranodmnte ßß• '•'
• ß •/•
• •\ -•' au&Hm• 0.55 is about33% (whereolivineis with Mg* = 0.90). Note
'-• 0 0 •00• w ß•A V Nd
¸ Mo&An that the above arguments assume that (1) the Ni
ß /" va [] Khch concentrationof the precursorsof the Siberianlavaswas
lO ©• • [] [] ß
ß
Gd
Sv
._

muchhigherthan observedand (b) the reductionin the Ni

[] u ß Iv content of the lavas resulted from olivine fractionation.
-10
Theseassumptions cannotbejustifiedby the presentdataset
asa plot of Mg* versusNi concentration (not shown)shows
-30 .... • .....,M,od,
e?MORB Mantle.(Sche
• ' ima?c
!, , no strongcorrelationbetweenthesetwo parameters for the
16 5 17 0 17 5 18.0 18.5 19.0
bulk of the SiberianTraps.
(2ø6pb/2ø4pb)t Figure 13 showsthe variationof end(t) versusMg*.
Figure 12. Variationsoœssr(0 and (206Pb/204Pb)t œorthe Siberian
Crustal contamination and magma evolutionappearto be
basalts.Note that the Bolgokhtokhskygranodioritccannotbca relatedin the early-erupted
high-Ti groupas indicatedby a
dominantcontaminant. The data point to a contaminant with low crudepositivecorrelationbetweenthe Mg* andend(t) values
dine-integrated U/Pb ratio,possiblylower crust[c.g, Rudnckan (Figure 13) [cf. Sharma et al., 1991]. In contrast,no such
Founain 1995; seealso Woodenetal., 1993]. correlationis seenfor the late-erupted,voluminous

©
O
Sa
Ku
E]
/•
Kha & Nr
Mu & Hm
v ? Mo&An
¸ Nd
O Tu
E] Khch
ß Gd
ß Sv
ß Iv
-lO
o 3 0.8
Figure 13. Plot showingvariationof •Nd (t) againstMg*. Dashed
line enclosesall high-Ti basaltdata. The Tuklonsky suite basalts
(enclosedby thedottedline) arecumulates.
group:with the exceptionof the Tuklonskylavas,the Mg*
values for the least contaminated and the most contaminated
low-Ti basaltsare similarto eachother[cf. Lightfootet al.,
1993; Wooden et al., 1993; Hawkesworthet al., 1995].
Woodenet al. [ 1993] notedthatthemineralogyandchemical
compositions of the low-Ti basalts(exceptthe Tuklonsky
suite) suggestthat the magmaticprecursorsof theselavas
underwentvariable olivine + plagioclase+ clinopyroxene
fractionation at crustal depths. The basalts from the
TuklonskysuitedisplayhighMg* andlow gNd values.These
lavasaremostprobablycumulates[Woodenet al., 1993]and
were evolvedby mixingwith low gNdcrustalmeltsthatwere
alsoLREE-enriched[e.g.,Penget al., 1994].
Major elementstudiesin otherfloodbasaltprovincesalso
suggest similar scenarios, with substantial magmatic
evolutionthrough gabbroicfractionation(i.e., fractionation
at lessthan5 kbar of olivine,plagioclase,andclinopyroxene
in crustalmagmachambers)[e.g.,Krishnamurthy and Cox,
1977; Cox, 1980; Basaltic VolcanismStudyProject, 1981].
Magmatic differentiationand contamination may occur in
large chambersat deep crustallevels [e.g., Woodenet al.,
1993; Peng et al., 1994] similar to thosepreservedin the
Ivrea Zone of northernItaly [e.g., Voshageet al., 1990].
Indeed,magmapondingat or nearthe crust-mantle boundary
(the Moho) appearsto be a natural consequence of the
densitycontrastbetweenpicriticmeltsandthe continentallid
[e.g., Cox, 1980; Farnetani et al., 1996 and references
therein].The two typesof contamination seenin the Siberian
Traps have also been recognizedin other flood basalt
provinces:(1) the most evolved magmas are the most
contaminated;examples come from the Columbia River
province[Carlsonet al., 1981, 1983], Paranti[Hawkesworth
et al., 1986; Petrini et al., 1987], and Ferrar Supergroup
SHARMA 289
[Menziesand Kyle, 1990]; and (2) the mostprimitivebasalts
(high Ni, Cr and Mg*) are the mostcontaminated; examples
come from the Deccan Traps [e.g., Mahoneyet al., 1982;
Devey and Cox, 1987; Mahoney, 1988] and the British
TertiaryVolcanicProvince[e.g.,Thompson et al., 1986].
8. MODELS OF GENERATION OF SIBERIAN TRAPS
High-'R
8.1 General
As discussed earlier,thereappearsto be little disputeover
the involvementof a mantleplume in the generationof the
Siberianflood volcanism.However, there is disagreement
overthe composition of the plumesource(relativelyMORB-
like mantleversusrelatively"undepleted"and undegassed,
possiblylower mantle source),relativeroles of the plume
sourceandCLM in the generation of thebulk of the Siberian
Traps,and the extentof modificationof the magmasby the
continentalcrust.The followingmodelshavebeenproposed.
(1) Melting of the CLM in response to heatingby a mantle
plume with further modification of the magma by
continentalcrust[Lightfootet al., 1990, 1993;Hawkesworth
et al., 1995]. This modelis identicalto thatproposedfor the
CFBs related to the breakup of Gondwana [e.g.,
Hawkesworthet al., 1986]. Accordingto this model, the
low-Ti basaltsof the TuklonskyandMokulaevskysuitesare
the result of melting of CLM that containedsediments
subducted previously.The lower-assemblage, high-Tibasalts
from Gudchikhinsky suiteare derivedfrom a mantleplume
similarto thosepostulatedfor oceanislandbasalts[Lightfoot
et al., 1993].
(2) Melting of a lowermantleplume(eNd-- +2 to +4) with
minimal involvement of MORB-type mantle, CLM, or
continentalcrust [Sharma et al., 1991, 1992; Basu et al.,
1995;cf. DePaoloand Wasserburg, 1979].Accordingto this
model,early eruptionsof volumetricallyinsignificantlavas
were dominated by contributionsfrom the MORB-type
mantle(e.g.,Gudchikhinsky). The bulk of the Siberianflood
basaltswere derived from decompressional melting in the
plumehead.Magmasen routeto the surfacewere modified
slightly by the continentalcrust (Morongovskythrough
Kharaelakhskyandtheir Putoranacounterparts).
(3) Melting of a deep mantle plume with large-scale
modification of the magmas in shallow-level magma
chambers. Accordingto thismodel,theuniformcomposition
of many of the low-Ti magmas reflects buffering by
continentalcrust [Arndt et al., 1993; Woodenet al., 1993].
The degree of melting, which is contingentupon the
thickness of the lithosphere, controlswhetheror not garnetis
presentas a residue.The low-degreemelts (with residual
garnetin the source)evolveto high-Ti basalts.In
290 SIBERIAN TRAPS

the meltsderivedfrom a largedegreeof meltingof mantle thatofferedby DePaoloand Wasserburg [ 1979]. Whiteand
peridotite in theabsence of garnetevolveto become thelow- McKenzie[1995] invertedthe REE patternsof the lavasto
Ti basalts. show that following the initial stagesof melting, during
which the CLM could be involved,the subsequent bulk of
8.2.Meltingof theContinental Lithospheric Mantle the Siberian Traps basaltswere plausibly formed from
melting of a plume. This model is obviouslytoo simple,
Lightfootet al. [1990, 1993] and Hawkesworthet al. however,asit reliesheavilyon REE patternsandNd isotopic
[1995]proposed thatSiberianfloodvolcanism wastheresult data, not taking into accountthe complextrace element
of meltingin the CLM engendered by a hot mantleplume. patternsdisplayedby the lavas(Figures4, 7, 8). In particular,
The principalargumentis baseduponlow Ta/La andNb/La themodeldoesnot explainthe causeof systematic variations
ratios shown by the upper-assemblage low-Ti basalts,a in Th/Ta ratiosthat are alsomuchhigherthanexpectedfrom
featurethat these authorsattributedto a CLM containing meltingof a primitivemantlesource(Figure7). As discussed
subducted sedimentary material. This modelwas criticized above, continentalcrustalcontaminationprobablyexplains
by Arndtet al. [1993] and Woodenet al. [1993], who argued the high Th/Ta and Ta/La ratios. Further, the 'nearly
that the CLM is too cold and too dry and doesnot have an primitive' endvalues of-+2 of the bulk volume of the
appropriatetrace-elementcompositionto form a major SiberianTraps can be explainedby the modestcontam-
source for the flood basalts. On the basis of the elevated
inationof OIB-typemeltswith •Ndof-- +4 (Figures9, 11).
(878r/SSSr)t, low lgNd(t),and unradiogenicPb isotopic
compositions of the low-Ti lavas, Woodenet al. [1993] 8.4. Buffering of Plume-DerivedMagmas by Continental
concluded that the low Ta/La and Nb/La ratios resulted from Crust
contamination by old continentalcrust.Whiteand McKenzie
Woodenet al. [1993] and Arndt et al. [1993] invoked
[1995] gave additionalargumentsagainstthe CLM being a
majorsourceof the SiberianTraps:(1) if the sourceof heat large-scalebuffering of plume-derivedmagmasby conti-
for generating the floodbasaltswasa hotmantleplume,then nentalcrustin shallow-levelmagmachambers to accountfor
it is very difficultto explainhow rapidmeltingin the plume (1) thenearlyuniformmajor-andtrace-element andisotopic
headcouldbe suppressed; and (2) conduction of heatin the compositions
of thelow-Tilavas,(2) high(87Sr/aasr)t
ratios,
CLM is too slow to producelargevolumesof magmain a and (3) negative Nb-Ta anomalies in the lavas. In
formulatingthe model, Woodenet al. [1993] observedthat
shortperiod.Thesearguments were examinedquantitatively
the low-Ti lavas show significantamountsof fractional
by Turneret al. [ 1996], who testedthe feasibilityof exten-
crystallizationof olivine, plagioclase,and clinopyroxene,
sive melting in the lithosphereabove a plume insteadof
and reflect evolutionin magma chambersat pressures less
within the plume itself. They concludedthat the most
than5 kbar. Woodenet al. [ 1993] modeledmagmaevolution
favorablecircumstances underwhich significantlithospheric
in chambersthat were periodicallyreplenishedand tapped,
melting would ensue without melting in the underlying
in which the magmaswere continuouslyfractionatingand
plume include a >100-km-thick, volatile-enrichedlithos-
assimilatingcrustalwall rocks.Accordingto theirmodel,the
phere (0.3 wt.% H20 q- CO2) overlying a plume with a
high-Ti and low-Ti magmaswere derived from different
potentialtem-peraturebetween 1380 and 1580 øC. Under
degreesof meltingof a plumesource.This modelis ableto
suchconditions,-1-2-km-thickfloodbasaltsmay be derived
explainmanyof the featuresobservedin the Siberianbasalts
from the CLM if the eruptionoccursover 10-15 m.y. A
as resultingfrom crustalcontamination and magmamixing,
periodof 10-15 m.y. is requiredbecauseof the extremely
low thermalconductivityof the CLM. Becausethe bulk of
including(1) the high (87Sr/86Sr)t
and low Ta/La ratios
(Figure4), (2) high andnearlyconstantTh/Ta ratios(Figure
the SiberianTrapswasmostprobablyproducedin -1 m.y., it
7), and (3) nearlyuniformmajor elementandtraceelement
is most likely that most of the lavas were derived from
compositions of the upper-stage
lavas.However,the model
meltingwithin the plume.
doesnot accountfor the existenceof two isotopically
distinct
8.3. Meltingof a LowerMantle Plume magmasources(see, e.g., Figure 11) for the high-Ti and
low-Ti lavas.

On the basisof volume-weightedNd-isotopiccomposition

of the Siberianlavas (average8Nd(t)= +1.8), which also 8.5. Considerations for a GeneralModel of the Siberian
Flood Volcanism
showeda sharpcut-off at 8Nd(t)= 0, Sharmaet al. [1991,
1992]proposed that 90% of the SiberianTrapslavaswere Any generalmodel of generationof the SiberianTraps
derived from the partial melting of a 'nearly primitive' must. account for the following key observationsand
lower-mantle-derived plume.This suggestionis identicalto
 SHARMA 291

(1) The bulk of the magmatismoccurredrapidly, within flood magmatismand its role in the earth's history. The
-1 m.y. province is enormous and we are just beginning to
(2) The flood volcanism was not accompaniedby understand the complexities[seeFedorenkoet al., 1996]. In
significantuplift or rifting. the following, an outline of some of the areasfor future
(3) At least some of the early alkaline ultramafic researchis provided.At present,the samplingis confined
magmatismpossessed 3He/4Hesignatures
indicativeof the largely to the northern and northwesternparts of the
involvement
of a high3Hesource. province. High-quality geochemical,isotopic, and geo-
(4) Incompatibleelementpatternsof the high-Ti and low- chronologicaldata from additionalsectionsin the southern
Ti groupsare not diagnosticof derivationfrom either a Putorana(e.g., Nizhnyaya Tunguskaarea) are needed to
MORB-typeor an OIB-typemantlesource. evaluatethe proposedmodelsof Siberianflood volcanism.
(5) During the initial stagesof magmatism,high-Ti lavas Further, analysesof other rock types associatedwith the
were derived from a mantle source with I•Nd,'--+8 and Siberian Traps, including kimberlites,alkalic basalts,and
containingresidualgarnet. maimechites,are required to understandthe plume-CLM
(6) The bulk of the SiberianTraps, representedby the interaction.An importantproblem intimately tied to the
voluminous late-stagelow-Ti lavas, was derived from a evolutionof someof the early lavasin the Noril' sk regionis
shallowermantlesourcewith I•Nd ,---+4. their relationship to ore mineralization.Detailed sulfur
(7) The Siberian Traps magmas underwent variable isotopeanalysisof the lavasand the associated mineralized
continentalcrustalcontamination, asindicatedby highTh/Ta intrusionsis neededto evaluatethis issue[seeHawkesworth
and low Ta/La ratios. The contaminants had low time- et al., 1995;Lightfootand Hawkesworth,thisvolume].
integratedU/Pb ratios as indicatedby the unradiogenicPb The extent to which the CLM and continental crust were
isotopicratiosof the contaminated lavas.It is possiblethat involvedin the generationof the SiberianTraps magmas
magma processingtook place in crustal reservoirsthat shouldbe evaluatedfurther.To this end, significantcontri-
underwentperiodicreplenishment, periodictapping,crystal butions have come from (1) Horan et al. [1995], who
fractionation,and wallrock assimilation. discussed the Os isotopicvariationsin someearly-erupted
At present,no modelhasbeenproposedthat would satisfy picritesand maimechitesfrom the Maimecha-Kotuiregion,
all the above observations.Fedorenko et al. [1996] and (2) Walker et al. [1994], who analyzedore-bearing
suggestedthat Anderson's [1994] model of melting of intrusionsfrom Noril'sk for Os, Nd, and Pb isotopes.
perisphereshould be investigatedto explain the high-Ti However, suchstudiespresentlycan be conductedonly on
basalts.Recently,D. L. Anderson(The helium-leadparadox, samples with low Re/Os ratios and high enough Os
ms. in process)hassuggested that an incompatible-element- abundances. This limitation leads to a focus on volumetri-
richperisphereandnot deepmantleplumessuppliesthe high cally insignificantflows whichmay not be representative of
•He/4Heto OIB magmas.This ideapresents an intriguing the bulk of the SiberianTraps.A detailedoxygenisotopic
possibilityas the perispheremay alsobe the sourceof the study on mineral separatesis neededto assessthe role of
early alkaline-ultramaficlavas from the Maimecha-Kotui crustalcontamination relativeto CLM influences[cf. Penget
areawhichdisplaya high3He/4He signature. However,the al., 1994].
ultimate source of 3He has to be the lower mantle which is Another importantissueis that of spatialprogression of
generallyconsidered to be the relativelyundegassed mantle volcanismon the Siberianplatform and its relationship,if
reservoir[seee.g. Porcelliand Wasserburg, 1995].Thusone any,with the LomonosovRidge(Figure3). A detailedplate-
must ask the questionof whetherthe two magma sources tectonicreconstruction based on precise4øAr-39Ar and
inferred in the Siberian Traps could be identified with paleomagnetic dating is needed to addressthis problem. In
perisphere(high-Ti, I•Na,'--3-8)
and plume (low-Ti, I•Sd""+4, this context, some accessible portionsof the Lomonosov
high 3He/4He).Note that in the above discussion it is Ridgeshouldalsobe sampled.
assumedthat the perisphereis a viablemantlereservoirfor
producinglarge volumesof melts. The existenceof this Acknowledgments.I would like to thank A. R. Basufor many
reservoiras a viable mantlesourceis not acceptedby many discussions on the genesisof the SiberianTraps. David Peate
geochemists. persuadedme to write thisreviewpaperandgaveme new insights
into the data. Comments from J. L. Wooden, an anonymous
reviewer, and J. J. Mahoney significantlyimprovedthis paper.
9. WHERE DO WE GO FROM HERE?
Many thanksto Don Porcelli,Tom Latourette,and Paul Asimow
for all the suggestions.This work is supported
by a grantto G. J.
Recent studieshave shownthat the SiberianTraps have Wasserburg(NASA-NAGW 3337). Division contributionNo.
the potentialto provide importantinsightsinto continental 5661
292 SIBERIAN TRAPS
REFERENCES
Albar•de, F., How deep do commonbasalticmagmasform and
differentiate?,
d. Geophys.Res.,97, 10,997-11,009,1992.
Anderson,D. L., The sublithosphericmantle as the sourceof
continental flood basalts; the case against the continental
lithosphereand plume head reservoirs,Earth Planet. Sci. Lett.,
!23, 269-280, 1994.
Amdt, N. T., G. K. Czamanske, J. L. Wooden, and V. A.
Fedorenko,Mantle and crustalcontributionsto continentalflood
volcanism,Tectonophysics, 223, 39-52, 1993.
Baksi,A. K., andE. Farrar,40Ar/39Ardatingof the SiberianTraps,
USSR: evaluationof the agesof the two majorextinctionevents
relativeto episodesof flood-basaltvolcanismin the USSR and
theDeccanTraps,India, Geology,! 9, 461-464, 1991a.
Baksi,A. K., andE. Farrar,40Ar/39Ardatingof whole-rockbasalts
(Siberian Traps) in the Tunguska and Noril'sk area, Siberia
(abstract),Eos Trans.AGU, 72, 570, 1991b.
Balashov, Y. A., and G. V. Nesterenko, Distribution of the rare
earthsin the traps of the Siberianplatform,Geochem.Int., 3,
672-679, 1966.
Barling, J. and S. L. Goldstein,Extreme isotopicvariationsin
Heard Island lavas and the nature of mantle reservoirs,Nature,
348, 59-62, 1990.
Basaltic Volcanism Study Project, Basaltic Volcanismon the
TerrestrialPlanets,1286 pp., PergamonPress,New York, 1981.
Basu,A. R., P. R. Renne,D. K. DasGupta,F. Teichmann,andR. J.
Poreda, Early and late alkali igneouspulsesand a high-3He
plumeoriginfor the Deccanfloodbasalts,Science,261, 902-905,
1993.
Basu, A. R., R. J. Poreda, P. R. Renne, F. Teichmann, Y. R.
Vasiliev, N. V. Sobolev, and B. D. Turrin, High-3He plume
origin and temporal-spatialevolution of the Siberian flood
basalts,Science,269, 822-825, 1995.
Bellieni, G., P. Brotzu, P. Comin-Chiaramonti,M. Emesto, A.
Melfi, I. G. Pacca,and E. M. Piccirillo,Floodbasaltto rhyolite
suite in southem Paranti Plateau (Brazil): palaeomagnetism,
petrogenesis and geodynamicalimplications,d. Petrol., 25, 579-
618, 1984.
Campbell,I. H., G. K. Czamanske,V. A. Fedorenko,R. I. Hill, and
V. Stepanov, Synchronismof the Siberian Traps and the
Permian-Triassic boundary,Science,258, 1760-1763, 1992.
Carlson,R. W., G. W. Lugmair, and J. D. Macdougall,Columbia
River volcanism:the questionof mantleheterogeneity or crustal
contamination,Geochim. Cosmochim.Acta, 45, 2483-2499,
1981.
Carlson,R. W., G. W. Lugmair,and J. D. Macdougall,Columbia Fedorenko,
river volcanism: the questionof mantleheterogeneity
or crustal
contamination (replyto a commentby D. J. DePaolo),Geochim.
Cosmochim.Acta, 47, 845-846, 1983. Fedorenko,V. A., andO. A. Dyuzhikov,Establishing
Clague,D. A., W. S. Weber,andJ. E. Dixon,Picriticglasses from
Hawaii, Nature, 353, 553-556, 1991.
Claou6-Long,J. C., Z. Zichao,M. Guogan,andD. Shaohua,The Fedorenko,V. A., V. M. Kuligin, G. C. Vitozhents,S. K. Mikhalev,
age of the Permian-Triassicboundary,Earth Planet. Sci. Lett.,
105, 182-190, 1991.
Claou6-Long, J. C., W. Compston, J. Roberts, and C. M. Fedorenko,V. A., P. C. Lightfoot,A. J. Naldrett,G. K. Czamanske,
Fanning,Two carboniferous ages:a comparisonof SHRIMP
zircondatingwith conventionalagesand40Ar/39Aranalysis,in
Geochronology,Time Scales, and Stratigraphic Correlation,
Spec. Publ. 54, edited by W. Bergenn, J. Hardenboland D.
Kent, pp. 1-27, (SEPM) Society for SedimentaryGeology,
Tulsa, OK, 1995.
Condie,K. C., Chemicalcompositionand evolutionof the upper
continentalcrust:contrastingresultsfrom surfacesamplesand
shales,Chem. Geol., 104, 1-37, 1993.
Cox, K. G., A model for flood basaltvulcanism,d. Petrol., 21, 629-
650, 1980.
Dalrymple,G. B., G. K. Czamanske,
M. A. Lanphere, V. Stepanov,
andV. Fedorenko,40Ar/39Aragesof samplesfromtheNoril'sk-
Talnakhore-bearingintrusions and the Siberianflood basalts,
Siberia(abstract),
Eos Trans.AGU, 72, 570, 1991.
Dalrymple, G. B., G. K. Czamanske,V. A. Fedorenko,O. N.
Simonov, M. A. Lanphere, and A. P. Likhachev, A
reconnaissance 40Ar/39Argeochronologicstudyof ore-bearing
and related rocks, Siberian Russia, Geochim. Cosmochim.Acta,
59, 2071-2083, 1995.
DePaolo,D. J. and G. J. Wasserburg,
Nd isotopesin floodbasalts
from the Siberian Platform and inferences about their mantle
sources,Proc. Natl. Acad. Sci. USA, 76, 3056-3060, 1979.
Devey, C. W., and K. G. Cox, Relationshipsbetweencrustal
contaminationand crystallisationin continentalflood basalt
magmaswith specialreferenceto the Deccan Traps of the
WesternGhats,India, Earth Planet. Sci. Lett., 84, 59-68, 1987.
Dosso,L., B. B. Hahart,H. Bougault,J. G. Schilling,and J. L.
Joron,Sr-Nd-Pb geochemicalmorphologybetween10øN and
17øN on the Mid-Atlantic Ridge-A new MORB signature,
Earth Planet. Sci. Lett, 106, 29-43, 1991.
Ellam,R. M., andK. G. Cox, An interpretation of Karoopicritic
basaltsin termsof interactionbetweenasthenospheric magmas
and the mantlelithosphere,Earth Planet. Sci. Lett., 105, 330-
342, 1991.
Erwin, D. H., The Penno-Triassicextinction,Nature, 367, 231-236,
1994.
Farnetani,C. G., andM. A. Richards,Numericalinvestigationsof
the mantle plume initiation model for flood basaltevents,d.
Geophys.Res.,99, 13,813-13,833, 1994.
Farnetani,C. G., M. A. RichardsandM. S. Ghiorso,Petrological
modelsof magmaevolutionand deepcrustalstructurebeneath
hotspotsand flood basaltprovinces,Earth Planet. Sci. Lett.,
143, 81-94, 1996.
Fedorenko, V. A., Petrochemicalseries of extrusive rocks of the
Noril'skregion,SovietGeol.Geophys., 22, 66-74, 1981.
V. A., Tectoniccontrolof magmatism andregularities
of Ni-bearinglocalitieson the north-westem Siberianplatform,
SovietGeol.Geophys., 32, 41-47, 1991.
theperiodof
latePaleozoic-earlyMesozoicvolcanismin theNoril'skregionof
the Siberianplatform,SovietGeol.Geophys., 21, 114-117,1980.
andL. V. Makeeva,Rare earthelementsin magmaticformations
in theNoril'skregion,SovietGeol.Geophys., 30, 61-71, 1989.
C. J. Hawkesworth, J. L. Wooden,andD. S. Ebel,
 SHARMA 293

of the flood-basaltsequenceat Noril'sk, North Central Siberia, Noril'sk Symposium,392 pp., Ontario Geological Survey,
Inter. Geol. Rev., 38, 99-135, 1996. Toronto, 1994.
Forsyth,D. A. and J. A. Mair, Crustalstructure of the Lomonosov Lightfoot,P. C., A. J. Naldrett,N. S. Gorbachev,W. Doherty,and
Ridgeandthe Fram andMakarovBasinsnearthe northPole,d. V. A. Fedorenko,Geochemistryof the SiberianTrap of the
Geophys.Res.,89, 473-481, 1984. Noril'sk area, USSR, with implications for the relative
Gallagher,K., and C. Hawkesworth,dehydrationmeltingand the contributionsof crust and mantle to flood basalt magmatism,
generationof continentalflood basalts,Nature,358, 57-59, 1992. Contrib. Mineral. Petrol., 104, 631-644, 1990.
Graham,D. W., W. J. Jenkins,J.-G. Schilling,G. Thompson,M.D. Lightfoot, P. C., C. J. Hawkesworth,J. Hergt, A. J. Naldrett,N. S.
Kurz, and S. E. Humphris, Helium isotopegeochemistryof Gorbachev,V. A. Fedorenko,andW. Doherty,Remobilisation of
midoceanridge basaltsfrom the south-Atlantic,Earth Planet. the continentallithosphereby a mantle plume: major- trace-
Sci. Lett., 110, 133-147, 1992. element,and Sr- Nd- and Pb-isotopeevidencefrom picriticand
Hart, S. R., A large-scale isotope anomaly in the southern tholeiitic lavas of the Noril'sk District, SiberianTrap, Russia,
hemisphere mantle,Nature, 309, 753-756, 1984. Contrib. Mineral. Petrol., 114, 171-188, 1993.
Hart, S. R., and H. Staudigel, Isotopic characterizationand Lind, E. N., S. V. Kropotov,G. K. Czamanske,S.C. Gromme,and
identification of recycledcomponents, in Crust/MantleRecycling V. A. Fedorenko,Paleomagnetism of the SiberianFlood Basalts
at Convergence Zones,editedby S. R. Hart andL. Giilen,pp. 15- of the Noril'sk area: a constrainton eruptionduration,Int. Geol.
28, Kluwer Academic Publishers,Dordrecht, 1989. Rev., 36, 1139-1150, 1994.
Hawkesworth,C. J., J. S. Marsh, J. S. Duncan, A. J. Erlank, and M. Lurie, M. L., and V. L. Masaitis,PlateauBasalts,Int. Geol. Cong.,
J. Norry, The role of continentallithosphere
in the generationof 22, Moscow, 1-12, 1964.
the Karoo volcanic rocks: evidence from combined Nd- and Sr- Macdougall,J. D. andG. W. Lugmair,Sr andNd isotopesin basalts
isotopestudies,Spec.Publ. Geol.Soc.S.AJ•.,13, 341-354, 1984. from the EastPacificRise:significance for mantleheterogeneity,
Hawkesworth,C. J., M. S. M. Mantovani, P. N. Taylor, and Z. Earth Planet. Sci. Lett., 77, 273-284, 1986.
Palacz, Evidence from the Paranti of south Brazil for a Mahoney,J. J., DeccanTraps,in ContinentalFlood Basalts,edited
contributionto Dupal basalts,Nature,322, 356-359, 1986. by J. D. Macdougall,pp. 151-194,KluwerAcademicPublishers,
Hawkesworth,C. J., C. J. Lightfoot,V. A. Fedorenko,S. Blake, A. Dordrecht, 1988.
J. Naldrett,W. Doherty, and N. S. Gorbachev,Magma differen- Mahoney,J. J., J. D. Macdougall,G. W. Lugmair,M. SankarDas,
tiation and mineralisation in the Siberian continental flood A. V. Murali, and K. Gopalan,Origin of the DeccanTrap flows
basalts,Lithos,34, 61-88, 1995. at Mahabaleshwarinferredfrom Nd and Sr isotopicandchemical
Hergt,J.M., D. W. Peate,andC. J. Hawkesworth, The petrogenesis evidence,Earth Planet. Sci. Lett., 60, 47-60, 1982.
of Mesozoic Gondwana 1ow-Ti flood basalts,Earth Planet. Sci. Mahoney,J. J., A. P. le Roex,Z. Peng,R. L. Fisher,J. H. Natland,
Lett., 105, 138-148, 1991. Southwestern limits of IndianOceanridgemantleandthe origin
Hill, R. I., Startingplumesandcontinental break-up,Earth Planet. of low 206Pb/204Pb mid-oceanridgebasalt:isotopesystematics of
Sci. Lett., 104, 398-416, 1991. the central SouthwestIndian Ridge (17ø-50ø E), d. Geophys.
Horan, M. F., R. J. Walker, V. A. Fedorenko, and G. K. Res., 97, 19,771-19,790, 1992.
Czamanske, Osmiumandneodymiumisotopicconstraints on the Mahoney,J. J., J. M. Sinton,M.D. Kurz, J. D. Macdougall,K. J.
temporaland spatialevolutionof Siberianflood basaltsources, Spencer, and G. W. Lugmair, Isotope and trace element
Geochim.Cosmochim.Acta, 59, 5159-5168, 1995. characteristics of a super-fastspreadingridge:EastPacificrise,
Ito, E., W. M. White, and C. G/Spel,The O, Sr, Nd andPb isotope 13-23øS, Earth Planet. Sci. Lett., 121, 173-193, 1994.
chemistryofMORB, Chem.Geol.,62, 157-176,1987. Marsh,J. S., Basaltgeochemistry andtectonicdiscrimination
within
Jokat, W., G. Uenzelmann-Neben, Y. Kristoffersen, and T. M. continentalflood basaltprovinces,d. Volcanol.Geotherm.Res.,
Rasmussen,Lomonosov Ridge--a double-sidedcontinental 32, 35-49, 1987.
margin,Geology,20, 887-890, 1992. Masaitis,V. L., I. I. Abramovich,D. A. Dodin,andA. A. Smyslov,
Kamo, S. L., G. K. Czamanske,and T. E. Krough,A minimumU- Uraniumin SiberianPlatformtrap rocks,Geochem.Int., 3, 392-
Pb age for Siberian flood-basalt volcanism, Geochim. 405, 1966.
Cosmochim.Acta, 60, 3505-3511, 1996. McDonough, W. F., Constraintson the compositionof the
Khain,V. E., Geologyof the USSR,272 pp., GebriiderBomtraeger, continentallithosphericmantle,Earth Planet. Sci. Lett., 101, 1-
Berlin, 1985. 18, 1990.
King,S. D.,'andD. L. Anderson, An alternative mechanism of Menzies, M. A., and P. R. Kyle, Continentalvolcanism:a crust-
flood basalt formation, Earth Planet. Sci. Lett., 136, 269-279, mantleprobe,in ContinentalMantle, editedby M. A. Menzies,
1995. pp. 157-177,OxfordUniversityPress,Oxford, 1990.
Krishnamurthy,P., and K. G. Cox, Picritebasaltsand relatedlavas Mitchell, C., J. G. Fitton, A. I. Al'mukhamedov, and A. I.
from the Deccan Traps of WesternIndia, Contrib. Mineral. Medvedev,The age and durationof flood basaltmagmatism:
Petrol., 62, 53-75, 1977. geochemical and palaeomagnetic constraintsfrom the Siberian
le Roex, A. P., R. A. Cliff, andB. J. Adair, Tristanda Cunha,South Province,Mineral. Mag., 58A, 617-618, 1994.
Atlantic:Geochemistry and petrogenesis of a basanite-phonolite Morgan,W. J., Hotspottracksandthe openingof the Atlanticand
lava series,d. Petrol., 31, 779-812, 1990. Indian Oceans,in The Sea, vol. 7, edited by C. Emiliani, pp.
Lightfoot,P. C., and A. J. Naldrett,Proceedings
of the Sudbury- 443-487,Wiley,New York,N.Y.,
294 SIBERIAN TRAPS

Naldrett,A. J., P. C. Lightfoot,V. A. Fedorenko,

W. Doherty,and Renne,P. R., andA. R. Basu,Rapideruptionof theSiberianTraps
N. S. Gorbachev,Geologyand geochemistry of intrusions
and flood basaltsat the Permo-Triassic
boundary,Science,253, 176-
floodbasaltsof theNoril'sk region,USSR,with implications for 179, 1991.
theoriginof theNi-Cu ores,Econ.Geol.,87, 975-1004,1992. Renne,P. R., A. L. Deino, R. C. Walter, B. D. Turrin, C. C.
Nalivkin, D. V., Geologyof the U.S.S.R,855 pp., Universityof Swisher,T. A. Becker,G. H. Curtis,W. D. Sharp,and A.-R.
Toronto Press,Toronto, 1973. Jaouni,Intercalibration
of astronomical
andradio-isotopictime,
Nesterenko,G. V., A. I. Al'mukhamedov,and Y. I. Belyayev, Geology,22, 783-786, 1994.
Cadmiumin the differentiationof a basicmagma,Geochem.Int., Renne,P. R., Z. Zichao,M. A. Richards,
M. T. Black,andA. R.
9, 432-437, 1972. Basu,Synchrony
andcausalrelations
between
Permian-Triassic
Nesterenko,G. V., and A. I. Al'mukhamedov,Geochemical boundarycrisesand SiberianFloodVolcanism, Science,269,
peculiaritiesof the trappeanmagmadifferentiation,
in Recent 1413-1416, 1995.
Contributions to Geochemistry edited Richards,M. A., R. A. Duncan,andV. E. Courtillot,Floodbasalts
andAnalyticalChemistry,
by A. I. Tugarinov,pp.233-242,JohnWiley,New York, 1975. andhotspot tracks:Plumeheadsandtails,Science,
246, 103-107,
Nestemko, G. V., and A. I. Al'mukhamedov,Titanium in the 1989.

pyroxenesof differentiatedtraps,Geochem.Int., 3, 767-777, Rudnick,R. L. andD. M. Fountain,Natureandcomposition of the

1966. continental crust:a lowercrustalperspective,
Rev.Geophys., 33,
Nesterenko,G. V., and L. P. Frolova,Lithium and rubidiumin 267-309, 1995.
SiberianTrap rocks, Geochem.Int., 2 (Supplement),295-305, Russell,J. K., and D. Chemiak,Theoryand applicationof Pearce
1965. element
ratiosto geochemical
dataanalysis,
315pp.,Geological
Nesterenko, G. V., N. S. Avilova, and N. P. Smimova, Rare Associationof CanadaShortCourse,Vancouver,1989.
elementsin the trapsof the Siberianplatform,Geokhimiya,I O, Sadovnikov,G. I., Correlationand origin of the volcanogenic
970-976, 1964. formations of the Tunguskabasin,NorthernAnabarregionand
Nesterenko,G. V., I. Y. Belyayev,and P. H. Phi, Silver in the Taymyr,IzvestiyaAcadem.Nauk.,9, 49-63, 1981.
evolutionof mafic rocks,Geochem.Int., 2, 162-169, 1969a. Scarrow,J. H., andK. G. Cox, Basaltsgenerated by decompressive
Nesterenko,G. V., Y. I. Belyayev,andF. Fam-Khung,Silverin the adiabaticmeltingof a mantleplume:a casestudyfromtheIsleof
evolutionofmafic rocks,Geochem.Int., 6, 119-126, 1969b. Skye,NW Scotland, J. Petrol.,36, 3-22, 1995.
Nesterenko,G. V., Y. B. Znamenskiy,A. I. Al'mukhamedov, and Sharma,M., A. R. Basu, and G. V. Nesterenko,Nd-Sr isotopes,
V. D. Tsykhanskiy,Nb. Ta, Zr, and Hf in trap magma petrochemistry, and originof the Siberianfloodbasalts,USSR,
differentiation,Geochem.Int., 8, 725-738, 1971. Geochim. Cosmochim.Acta, 55, 1183-1192, 1991.
Nesterenko,G. V., A. I. Al'mukhamedov,and Y. I. Belyayev, Sharma,M., A. R. Basu,and G. V. Nestemko,TemporalSr- Nd-
Cadmiumin the differentiationof basicmagma,Geochem.Int., and Pb-isotopicvariationsin the Siberianflood basalts:
6, 669-675, 1972. implications
for the plume-source
characteristics,
Earth Planet.
Nutman,A. P., I. V. Chemyshev,H. Baadsgaard, andA. P. Smelov, Sci. Lett., 113, 365-381, 1992.
The Aldan Shield of Siberia, USSR: the age of its Archaean Sobolev,A. V., andM. Chaussidon,H20 concentrations
in primary
components
and evidencefor widespread
reworkingin the mid- melts from supra-subduction zones and mid-oceanridges:
Proterozoic,Precamb. Res., 54, 195-210, 1992. implicationsfor H20 storageandrecyclingin themantle,Earth
Peng,Z. X., J. Mahoney,P. Hooper,C. Harris,andJ. Beane,A role Planet. Sci. Lett., 137, 45-55, 1996.
for lower continentalcrustin flood basaltgenesis?
Isotopicand Storey,M., A.D. Saunders,J. Tamey,P. Leat,M. F. Thirlwall,R.
incompatible elementstudyof the lowersix formationsof the N. Thompson, M. A. Menzies,andG. F. Marriner,Geochemical
westernDeccanTraps,Geochim.Cosmochim. Acta,58, 267-288, evidence for plume-mantleinteractions
beneathKerguelen and
1994. HeardIslands,IndianOcean,Nature,326, 371-374,1988.
Petrini,R., L. Civetta,E. M. Piccirillo,G. Bellieni,P. Comin- Sun, S.-S., and W. F. McDonough,Chemicaland isotopic
Chiaramonti,L. S. Marques,and A. J. Melfi, Mantle hetero- systematics of oceanic basalts:implicationsfor mantle
geneityand crustalcontamination in the genesis of low-Ti composition andprocesses,
in Magmatism in theOceanBasins,
continentalfloodbasaltsfrom the ParantiPlateau(Brazil):Sr-Nd Spec.?ubl.42, editedby A.D. Saunders andM. J. Norry,pp.
isotope andgeochemical evidence,d. Petrol.,28, 701-726,1987. 313-345,The GeologicalSociety,London,1989.
Porcelli,D. and G. J. Wasserburg,Masstransferof helium,neon, Sweeney, J.F., J.R. WeberandS.M. Blasco,Continental
ridgesin
argon,andxenonthrougha steady-state uppermantle,Geochim. the Arctic ocean: LOREX constraints,
Tectonophysics,
89, 217-
Cosmochim.Acta, 59, 4921-4937, 1995. 237, 1982.
Pringle,M. S., C. Mitchell, J. G. Fitton, and M. Storey, Thompson, R. N., M. A. Morrison,
A. P. Dickin,I. L. Gibson,
and
Geochronologicalconstraints
on the origin of large igneous R. S. Harmon,Two contrasting stylesof interaction between
provinces:Examplesfrom the Siberianand Kerguelen flood basicmagmasand continental crustin the BritishTertiary
basalts,D. L. Anderson,S. R. Hart, and A. W. Hofmann, VolcanicProvince,J. Geophys.Res.,91, 5985-5997,1986.
convenors,Plume 2, Terra Nostra, 3/195, Alfred-Wegner- Turner,S., C. Hawkesworth, K. Gallagher,K. Stewart,D. Peate,
Stiftung,Bonn, 120-121, 1995. andM. Mantovani,Mantle plumes,floodbasalts,andthermal
Renne,P. R., Excess40Ar in biotite and homblendefrom the modelsfor meltgenerationbeneathcontinents:
assessmentof a
Noril'sk 1 intrusion,Siberia:implicationsfor the age of the conductiveheatingmodel and applicationto the Paranti,d.
SiberianTraps,EarthPlanet.Sci.Lett.,131, 165-176,1995. Geophys.Res.,101, 11,503-11,518,
 SHARMA 295

VoshageH., A. W. Hofmann, M. Mazzucchelli, G. Rivalenti, S. Woodhead,J. D. and C. W. Devey, Geochemistryof the Pitcairn
Sinigoi, I. Raczek, G. Demarchi, Isotopicevidencefrom the seamounts, I: sourcecharacterandtemporaltrei•ds,Earth Planet.
Ivrea Zone for a hybrid lower crustformed by magmaticun- Sci. Lett., 116, 81-99, 1993.
derplating,Nature, 347, 731-736, 1990. Zindler, A., and S. Hart, Chemicalgeodynamics, Ann. Rev. Earth
Walker, R. J., J. W. Morgan, M. F. Horan, G. K. Czamanske,E. Planet. Sci., 14, 493-571, 1986.
J. Krogstad,V. A. Fedorenko,and V. E. Kunilov, Re-Os iso- Zolotukhin,V. V., andA. I. Al'mukhamedov,Trapsof the Siberian
topic evidence for an enriched-mantlesourcefor the Noril'sk- Platform, in Continental Flood Basalts, edited by J. D.
type, ore-bearing intrusions,Siberia, Geochim. Cosmochim. Macdougall,pp. 273-310, Kluwer AcademicPublishers,New
Acta, 58, 4179-4197, 1994. York, 1988.
White,R. S., andD. McKenzie,Mantleplumesandfloodbasalts,d. Zolotukhin,V. V., A.M. Vilensky,andO. A. Djuzhikov,Basaltsof
Geophys.Res., 100, 17,543-17,585, 1995. theSiberianPlatform,245 pp.,Nauka,Novosibirsk,1986.
White, W. M., A. W. Hofmann, and H. Puchelt, Isotope Zonenshain,L. P., M. I. Kuzmin, and L. M. Natapov, Geology of
geochemistry of Pacificmid-oceanridgebasalt,d. Geophys.
Res., the USSR: a platetectonicsynthesis, in Geologyof the USSR:a
92, 4881-4893, 1987. Plate TectonicSynthesis,editedby B. M. Page,p. 242, American
Wilson,J. T., Room at the top of the world,Nature,316, 768, 1985. GeophysicalUnion, Washington,D.C., 1990.
Wooden, J. L., G. K. Czamanske,V. A. Fedorenko,N. T. Amdt, C.
Chauvel,R. M. Bouse,B.-S. King,R. J.Knight,andD. F. Siems, M. Sharma, The Lunatic Asylum of the Charles Arms
Isotopic and trace-elementconstraintson mantle and crustal Laboratory,Division of Geologicaland PlanetarySciences,Mail
contributions to Siberiancontinentalflood basalts,Noril'sk area, Code 170-25, California Instituteof Technology,Pasadena,CA
Siberia,Geochim.Cosmochim. Acta, 57, 3677-3704, 1993.
 Giant RadiatingDyke Swarms'Their Use in Identifying Pre-Mesozoic
Large IgneousProvincesand Mantle Plumes

Richard E. Ernst

GeologicalSurveyof Canada, Ottawa, Ontario, Canada

Departmentof Earth Sciences,Universityof WesternOntario,London,Ontario, Canada

Kenneth L. Buchan

GeologicalSurveyof Canada, Ottawa, Ontario, Canada

The identificationof large igneousprovinces(LIPs) and associatedmantle

plumeshasbeenconfinedfor the mostpartto the MesozoicandCenozoicrecord.
With few exceptions,pre-Mesozoic LIPs havebeenpartiallyor totallydestroyed
by tectonicanderosional processes.
Here,we describea methodof identifyingthe
remnantsof pre-MesozoicLIPs and the location of paleoplumesfrom the
convergentpointsof giantradiatingdyke swarms.It is basedon the observations
from sevencasehistoriesthat giant radiatingdyke swarmsconvergetowards
knownmantleplumecentersbeneaththe volcanicaccumulations of Mesozoicand
CenozoicLIPs. We describethree additionalMesozoic and 14 pre-Mesozoic
swarmswhich likely focus on paleoplumes.Severalother candidatesare also
considered.

1. INTRODUCTION havea mantleplumeoriginincludecontinental floodbasalts,

Large igneousprovinces(LIPs) includecontinentalflood
basalts,volcanicpassivemargins,oceanicplateaus,oceanic
basinflood basalts,someseamounts and submarineridges,
and representlarge volumesof magma originatingfrom
processes otherthannormalsea-floorspreading[Coffinand
Eldholm, 1994]. They have been a focus of much recent
interestregardingtheir implicationsfor the thermalstructure
of the mantle, specificallytheir relationshipwith mantle
plumes,andtheir association with continentalbreakup[e.g.,
Morgan, 1981; Richardset al., 1989; Whiteand McKenzie,
1989;Storey,1995;Anderson,1995].LIPs that are likely to
LargeIgneousProvinces:Continental,Oceanic,andPlanetary
Flood Volcanism
GeophysicalMonograph100
Copyright1997by the AmericanGeophysical
Union
oceanicplateaus,and oceanicbasinflood basalts.Character-
isticssuggestiveof a mantleplumeoriginincludeemplace-
ment of large volumesof magma in a shorttime interval
[e.g., Coffinand Eldholm,1994], linkageto hotspottracks
[e.g., Morgan, 1981; 1983], and associatedtopographic
doming[e.g.,Cox, 1989;LeCherninant andHearnan,1989].
Coffinand Eldholm [1994] summarizedthe distributionof
LIPs. Most are of MesozoicandCenozoicage. In the pre-
Mesozoicrecord LIPs are poorly preserved,and thustheir
former locationshavebeendifficultto identify.Destruction
of pre-MesozoicLIPs has resultedfrom many processes,
including subduction,continent-continent collision, and
erosion.
In this paper, we discussthe surfacegeometryand other
characteristics
of giant dyke swarms(swarmswith lengths
> 300 km) [Ernst et al., 1995b] and their importancein
understanding mantleplumedynamicsandthe generationof
298 GIANT RADIATING DYKE SWARMS, LIPS, AND MANTLE PLUMES

plume-relatedLIPs. Althoughthe origin of most giant LIP. Furthermore,the criteriaof arrangingcontinental blocks
swarms is unclear, and small swarmshave a variety of so as to restorethe primary geometryof the swarm may
originsincludingthosethat are subduction
and continent- facilitatethe reconstruction
itself [Ernstet al., 1995a].
collision related [Ernst et al., 1995a and references In this section,dyke swarmsassociated with the major
therein], we suggestthat giant swarms which have a Cenozoic and Mesozoic continental flood basalts are
radiatinggeometry are derived from mantle plumes. A described in orderof increasingage(Figure1 andTable 1).
giantradiatingswarmconverges towardsthe formermantle A consistentpattern emerges, in which most swarms
plume centerand associatedflood basalt(or its erosional convergetowardsthe centerof topographic uplift marking
remnant). the locationof the plumehead.Theseexamplessuggestthat
In presentingradiometricages, we indicatethe uncer- dyke swarmsassociated with continentalflood basaksmay,
tainty level (lo or 20) where available.However, in many at leaston a regionalscale,form a simpleradiatingpattern
publications theuncertainty
levelsarepoorlydocumented or centeredon the mantle plume. This relationshipwill be
not reported. Provisionalnames for dyke swarms are furtherexploredin later sectionsof the paper,where giant
indicatedby a superscriptpn. radiatingdyke swarmsare utilizedto establishthe location
of flood basaltsthat have been removed by erosion or
2. CONTINENTAL FLOOD BASALTS AND THEIR deformedin laterorogenies.
ASSOCIATED DYKE SWARMS
2.2. Columbia River Event
2.1. Introduction
The ColumbiaRiver basaltgroup(Figure2) was erupted
Continentalflood basaltsand associated passivemarginsbetween17.5 and 6 Ma (K-Ar). Most of the extrusiverocks
are the only mantle-plume-related (97%) were emplacedduring the first 3.5 m.y. [Hooper,
LIPs for which a signifi-
cant amountof informationon dyke swarmsis presently 1988;Tolanet al., 1989] andarethoughtto be relatedto the
available.We will focuson continental examples,although arrival of a mantle plume centered several hundred
the possibility of using swarmsto locate oceanicmantle kilometersto the south,which subsequently migratedeast-
plumesis discussed briefly in section5.5. northeastto its presentpositionbeneathYellowstone[e.g.,
Dyke swarms are associatedwith all continentalflood Smithand Braile, 1993; Parsonset al., 1994] (Figure 2).
basalts,althoughtheir full extentand distributionare often Activity to the southalongthe Nevadarift hasbeenrelated
difficultto assessbecauseof a lack of exposure.The feeder to lithosphericrifting associatedwith the same plume
dyke systemis typicallyhiddenbeneaththe volcanicpile. [Zobacket al., 1994; Parsonset al., 1994].
Therefore,it is necessaryto rely on differentialerosionto Extensive dyke swarms are associatedwith this LIP
exposeportionsof the feeder system.Furthermore,where (Figure 2). Flows of the Columbia River Basalt group
dykeswarmshavebeenemplacedat depthanddidnotreach eruptedfrom north-northwest-trending fissuresystemsand
the paleosurface,erosionmay not have been sufficientto dykesof the 17-15 Ma (K-Ar) Chief JosephandMonument
exposethem. swarms [e.g., Swanson et al., 1975; Tolan et al., 1989;
The volcanicportionsof continentalflood basak prov- ,•tkinsonand Lambert, 1990]. Dykes associated with the
incesare typically locatedalongcratonmarginsand,hence, Nevadarift trendsouthfrom the plumecenterandare dated
are likely to be deformedand destroyedduringsubsequent at 17-14 Ma (K-Ar) [Zobacket al., 1994].
oceanclosing.Fahrig [1987] notedthat coevaldyke swarms Otherpoorlydatedswarmswhichmay alsobelongto this
extendinto the cratonwhere they are betterprotectedfrom eventincludethe SteensMountaindykesof north-northeast
laterdeformation.In addition,dyke swarmsare lesslikely to trend [Walker and MacLeod, 1991] and the dykes of the
be erodedthan flood basaltsbecausethey lie underneath the CascadeRange with a trend of about 315ø. The Cascade
lavaswhichthey feed andreachto at leastmidcrustaldepths Rangep,dykes are related to 17 to 10 Ma basalticand
[e.g., Fahrig, 1987; Ernst et al., 1995a].Thus,coevaldyke andesiticflows [unit Tbaa of Walker and MacLeod, 1991]
swarmsmay providethe best,and in many casesthe only, and, on the basis of their age, probablybelong to the
record of old continental flood basalts. Columbia River event.
In instances
wherea giantdyke swarmhasbeendismem- The ChiefJoseph,Monument,andCascade Range pndykes
bered by plate tectonicprocesses,its primary radiating radiateover an angle of 45 ø aboutthe site of presumed
pattern may only become apparentwhen continentshave plumeinitiation.The Nevadariftpndykesalsotrendtowards
beenreconstructed.Only thencanthe focusof the radiating thesamecenter.However,the SteensMountaindykesdonot
patternbe used to locatethe plume centerand associated fit theradiating
 ERNST AND BUCHAN 299

LightfootRiver
KOLA-ONEGA

Lac de
Gras
60øN
MACKENZIE ABrrIBI
Tertiary Valley
MATACI-IEWAN

Chief Joseph
30øN N-T-S
WyomingBlock-2 RED
RECONST. SEA

Tanzania-3
ß Beldl /'
0ø
Coastline

GAIRD• Ponta Grossa

I ( • •,,jeda
30øS

Phanerozoic stassI
Late Proterozoic ......
60øS
Middle Proterozoic
EarlyProterozoic

90øE 180øE 270øE

$uperiør
IF?xøffudbul'Y
'11
0ø 90øE

Figure 1. Distribution
of key dykeswarmsassociated with eventsdiscussed
in thispaper.Swarmslongerthanabout500
km are labeledin uppercase. Trianglesand lowercase namesidentifysmallerswarms. N-T-S is Narmada-Tapti-Son.
Swarms thatarediscussed aspartof a largereventareBotswana (LowerLimpopoKarooevent);BritishTertiary(North
AtlanticVolcanicProvinceeven0;ChiefJoseph (ColumbiaRiver even0;Ejeda-Bekily (Madagascareven0;Gairdner
(Willouranevent);Grenville(CentralIapetusevent);Lac de Gras(BoothRiver event);N-T-S (Deccaneven0;Ponta
Grossa(Paranfi-Etendeka even0;RedSea(Afarevent);Senneterre (UngavaBayevent);Tanzania-3 (Bukoban event)and
Whin-Midland Valley(Jutland event);WyomingBlock-2(Willouranevent).Thenorth-central portionof NorthAmerica
hasbeenenlarged atthelowerrightto betterillustrate
thelargenumberof swarms locatedthere.

2.3. Afar Event The oldeststageof activityis markedby the Yemeni LIP
(agerangeof 30-25Ma, 4øAr-a9Ar),
theEthiopianLIP
The Afar plume (Figure3), whichcausedbroadregional (about30 Ma, 4øAr?Ar)andat leastoneharrat,
Harrat
uplift [Campand Roobol,1992] and incipientrifling along Hadan(28-27Ma,4øAr-39Ar)
[Bakeretal.,1996;Hofmann
the Gulf of Aden and the Red Sea, is associatedwith the et al., 1995; Sebai et al., 1991a, respectively].All of these
onsetof separationof Africa and Arabia [e.g., Whiteand units have a transitionalto alkali basalt composition
McKenzie, 1989; Cox, 1989]. The Yemeni and Ethiopian [Coleman,1993;Campand Roobol,1989].Few dykeshave
LIPs are the main volcanicexpressionof this plume, but yet been identifiedthat are coevalwith this early activity.
additional componentsinclude about 20 volcanic fields Exceptionsare thoseassociated with HarratHadan [Sebaiet
(locally called 'harrats' in the Arabian Peninsulaand the al., 1991a], whichtrendtowardstheplumecenter(Figure3).
largedyke swarmwhich parallelsthe Red Sea [e.g., Cole- A younger
24-21Ma (4øAr-a9Ar)
tholeiitic
stageof
man, 1993]. The total volume of associated magmatismis magmatismconsistsof the major coastparallel Red Sea
greater
than0.35x 106km3[Mohr,1983;Bakeretal., 1996]. dykes,Yemeni dykesand associated plutons[Sebaiet al.,
The magmatismof thiseventis discussed
in threemain age 1991a]. Somevolcanismin Ethiopiamay alsobe of thisage
ranges,30-25 Ma, 24-21 Ma, and lessthan 5 Ma. basedon 4øAr-39Ar
dating[Hofmann
et al., 1995].
300 GIANT RADIATING DYKE SWARMS, LIPS, AND MANTLE PLUMES

TABLE1. DykeSwarms
andLargeIgneous
Provinces
(LIPs)Associated
withPlume
Arrival
a

Event Plume
Head
b AgeofPlume Component
Dyke Selected Coeval LIPs or

(sectionof paper) Arrival (Ma) Swarmsof Radiating Remnant LIPs

Pattern

Columbia River Event Yellowstone 17-14 Chief Joseph ColumbiaRiver BasaltGroup

(section 2.2) Monument CFB
? Cascade
Range
pn
Nevada rift

Afar Event Afar --• 25 Red Sea EthiopianCFB

(section 2.3) Ethiopian Yemeni CFB
Yemeni
Harrat Hadan
'?AI Ghayl
pn
North Atlantic Volcanic Iceland 61-53 E. GreenlandTertiary NAVP mafic plutonsand
Province Event W. GreenlandTertiary lavas

(section 2.4) British Tertiary E. Greenland PM

W. Greenland PM
Voring PlateauOP
Deccan Event R6union --•66 Narmada-Tapti-Son Deccan CFB
(section 2.5) Panvelpn
Mt. Girnarpn
Cambay
Graben
pn
? Ranipn

MadagascarEvent Marion 90-83 Ejeda-Bekily MadagascarCFB

(section2.6) Morondavapn N. MadagascarRidge OP
Tamaravepn

Alpha Ridge Event - 100 Hazen Straitpn Alpha Ridge OP

(section 2.9) Lightfoot
Riverpn Queen Elizabeth Islands CFB

Paranfi-Etendeka Event Tristan da 137-127 Ponta Grossa Paranti CFB

(section2.7) Cunha Santos-Rio de Janeiro

Florian6polis
? Paraguay
West Bodoquena
Serro do Caiap6
c

Etendeka Etendeka CFB

Horingbaai
Mehlberg
Cape Pennisula
Lower LimpopoEvent Bouvet? - 184 Botswana Karoo CFB

(section2.8) Marion? Lebombo ? Ferrar CFB

Orange River
Lower Zambesi Event - - 180 Rushinga Karoo CFB

(section 2.8) Cholo ? Ferrar

Gorongosa
 ERNST AND BUCHAN 301

TABLE 1. (continued)

Event PlumeHeadb Age of Plmne ComponentDyke Selected Coeval LIPs or

(sectionof paper) Arrival (Ma) Swarmsof Radiating Remnant LIPs

Pattern

Central Atlantic Event Fernando de - 200 ENA Florida volcanism

(section4.1) Noronha? Charleston

CapeVerde'?
Amapfi
Guyana
pn

Moroccopn
Taoudennipn
Liberiapn

SiberianTrap Event Jan Mayen? -248 Ebekhaya Siberian CFB

(section2.9) Maymecha
pn
Jutland Event - 300 Whin-Midland Valley Oslo rift volcanism

(section4.2) Oslo Rill Whin & Midland valley sills

Scania

Yakutsk Event -350 Chara-Sinsk

(section4.3) Vilyui-Marcha
Dzhardzhanpn
To•nporuk
pn Dzhalkan volcanics

CentralIapetusEvent - 600 Long Range Tibbit Hill volcanism

(section4.4) Grenville
Adirondack
? Southern
Appalachian
pn
? GannakouriepEvent 720 Gannakouriep
(section4.13)
Franklin Event 723-718 Franklin Natkusiak CFB

(section4.5) Thule Coronation sills

Willouran Event 800-780 Mackenzie Mountains

(section4.6) Wyoming Province
Hottah sheets

Gairdner Willouran volcanism

? Bukoban Event 1000-800 Tanzania-3

d '? Bukoban volcanism

(section4.13) Tanzania-4
d
Tanzania-5
d
.9Coastline (Bahia) - 1000 Salvador
Event Ilh6us-Oliveng,a-Calnac5
(section4.13) Itacar6

Kola-OnegaEvent 1040 Kola-Onega

(section4.7)
Abitibi Event 1140 Abitibi

(section4.8) Eye
302 GIANT RADIATING DYKE SWARMS, LIPS, AND MANTLE PLUMES

TABLE 1. (continued)

Event Plume
Head
b AgeofPlume Component
Dyke Selected Coeval LIPs or
(sectionof paper) Arrival (Ma) Swarmsof Radiating Remnant LIPs
Pattern

Keweenawan Event 1110-1085 Keweenawan CFB

(section4.8)
SudburyDyke Event 1238-1235 Sudbury
(section4.13)
Mackenzie Event 1270-1265 Mackenzie CoppermineRiver volcanics
(section4.9) Muskox Intrusion

.9 Booth River Event 2030-2023 Lac de Gras BoothRiver Complex

(section4.13)
Fort Frances Event -2076 Fort Frances '?Mille Lacs Group
(section4.10)
UngavaBay Event - 2200 Senneterre
(section4.13) Klotz
Maguire
Matachewan Event 2490-2445 Matachewan Volcanics and mafic plutons
(section4.11)
Mistassini Event --2470 Mistassini
(section4.12)
.9 McArthur Basin McArthur Basin-1pn
Event
(section4.13)
Superscriptpn indicatesprovisionalnames.
a Detailsandreferences
in text.Labelling
of LIP typesatlerCoffinandEldholm[1994]:CFB iscontinental
floodbasalt,
PM is passivemargin,andOP is oceanicplateau.NAVP is NorthAtlanticVolcanicProvince. "?" in tableindicates
uncertaintyeitherin the identificationof an eventor in the correlationof swarmsor LIPs to an event.
b ForMesozoic
andCenozoic
LIPs,theplume
isgiven
thename
ofthehotspot
thatbacktracks
tothesiteof plume
initiation.
c Dashed
linesseparate
events
presently
located
onseparate
continents.
dTanzania
3, 4 and5 aretheTanzania
III, IV andV ofHallsetal. [1987].

addition,an age of 24 +__3 Ma (K-Ar) hasbeendetermined intrusionresultingin local crustalextensionup to 50%. A1

for the majorepisodeof dykingwithin the EthiopianPlateau
[Megrueet al., 1972; p. 84 of Mohr and Zanettin,1988]. In
the southeasternportion of the Yemen LIP, Yemeni dykes
haveagesof-25.5 and 18.5-16 Ma (4øAr-39Ar)
[Zumboet
al., 1995a].
The characterof the 24-21 Ma dykeschangesacrossthe
Ad Darb fault (Figure3) [pp. 58-60 of Coleman,1993]. The
Red Seadykesnorthof the fault aremainlyindividualdykes
with widthsup to 100 m, whereasthe Yemeni dykessouth
of the fault tend to be sheeteddyke complexeswith more
widely spaceddykeshaving widthsof 0.5 to 15 m [Cole-
man, 1993]. The greatestconcentration of Yemeni dykesis
along the coastal escarpmentzone, but less concentrated
swarmslace the interior [Mohr, 1991]. The Yemeni dykes
aredividedinto subswarms, somewith veryhighdensitiesof
Ghaylp,,dykestrendat a high angleto the Yemenidykesand
areassigned a broadagerangeof Cenozoic[Kruk, 1980].
After correctionfor the openingof the Red Sea,the Red
SeaandYemenidykestrendawayfromtheplumecenterfor
a distance of nearly 2000 km [Eyal and Eyal, 1987;
Baldridgeet al., 1991]. However,as thesedykeshave ages
of 24-21 Ma, they postdatethe main LIP activityat 30-25
Ma and their distributionmust be stronglyinfluencedby
rifting alongthe Red Sea.Likewise,the 24 Ma dyke swarms
in the EthiopianPlateauhave variabletrends,which appear
morerelatedto rift faultsratherthanfocussingon the plume
center[figure3 of Mohr and Zanettin,1988].
The youngestactivity, 5-0 Ma magmatism,occurswithin
the westernportion of the EthiopianLIP [Zumboet al.,
1995b; figure 3.1 of Coleman, 1993]. In addition,

120•W
WESTERN
NORTH
AMERICA
"oMa
Sl
.-17
Ma
•"
40•N•
500 km
Figure 2. Dyke swarmsof the 17-14 Ma ColumbiaRiver-Nevada
Rift magmaticevent. Swarmsassociatedwith the arrival of the
Yellowstone plume at 17 Ma include the Chief Joseph(CJ),
Monument(M), SteensMountain(S) and Nevada riftpn (N)
swarms,andpossiblytheCascade
Range
pn(C) swarm[Tolanet al.,
1989; Walker and MacLeod, 1991; Zoback et al., 1994]. Stars
locatethe positionof the mantleplume at 17 Ma and its present
positionunderYellowstone[Parsonset al., 1994]. Distributionof
volcanicrocks(shaded)after Tolan et al. [1989] andZobacket al.
[1994]. Circle with a radius of-400 km outlinesthe region of
plume-generated uplift [Parsonset al., 1994].
harratsexhibit activity in this time range [F6raud et al.,
1987;figure3.1 of Coleman,1993].
2.4. North Atlantic Volcanic Province Event
The North Atlantic VolcanicProvince(Figure4), with a
total volume of more than 6.6 x 106 km3, consistsof
extrusiverocks,intrusivecentersandassociated dykesof the
BritishTertiaryprovincein the BritishIsles;volcanicrocks
thatextendfor 2000 km alongthe eastcoastof Greenland,in
the adjacentocean,on the Voring Plateau(off Norway),and
on the Rockall Bank; and minor accumulationsin the Disko
Island area of west Greenlandand adjacentBaffin Island
[e.g.,Dickin, 1988; Whiteand McKenzie,1989; Coffinand
Eldholm,1994].Agesfrom all threeareasfall betweenabout
61 and 53 Ma (mostly based on K-Ar ages) with the
preponderance of magmatismoccurringbetween57.5 and
54.5 Ma [White and McKenzie,1989, 1995 and references
therein]. Whiteand McKenzie[1989] proposedthat volca-
nism in all of theseareaswas associated
with an upwelling
ERNST AND BUCHAN 303
mantleplume locatedbeneaththe eastcoastof Greenland.
Plume-generateduplift occurredbetween65 and 55 Ma
[Nadin and Kusznir, 1995] and was associatedwith the
breakupandseparation of EuropeandGreenland.The tail of
the plumeis now expressed by continuing
volcanicactivity
in Iceland.
An alternativemodelwasproposedby LawverandM•iller
[1994], who computeda different track for the Iceland
hotspoton the basisof studiesof plate motionsand other
hotspots.Their 60 Ma locationfor the hotspotlies in west
centralGreenland,a distanceof nearly 1000 km from that
advocatedby White and McKenzie [1989]. Furthermore,
theysuggested thattheplumearrivedmuchearlier,possibly
at 130 Ma, beneathEllesmereIslandin northernCanada.
Dykesof the BritishTertiaryprovincehavea dominant
northwesttrend,althoughmore localradiatingdyke setsare
1 ooo km
_ R 25"N
--
R H ARABIA
.-"
\
fault
S"N
AFRICA
Figure 3. Dyke swarmsof the -25 Ma Afar magmaticevent of
Africa and Arabia in the present-dayconfigurationof the region.
Red Seadykes(R) are afterEyal and Eyal [1987], Ethiopiandykes
(E) afterMohr and Zanettin[1988], Yemeni (Y) dykesafterMohr
[ 1991], Harrat Hadan (H) dykesafter Sebai et al., [1991], andAI
Ghaylpn(A) dykes after Kruk [1980]. Shadedpatternshowsthe
distribution of volcanic rocks [White and McKenzie, 1989],
includingthe EthiopianLIP, the Yemeni LIP nearthe southerntip
of the Arabian Peninsula and smaller 'harrats' farther north. Circle
with a radiusof 1000 km depictsthe probableextentof the thermal
anomaly in the mantle surroundingthe plume [after White and
McKenzie,1989].Dottedlinesmarkrift-relatedfaultsin
304 GIANT RADIATING DYKE SWARMS, LIPS, AND MANTLE PLUMES
Baffin Is.
Disko Is.
Rockall
1000 km
Figure 4. Dyke swarmsof the-60 Ma North Atlantic Volcanic
Province.The BritishTertiaryProvincedykes(B) are after$peight
et al. [1982], and the east Greenland dykes (EG) and west
Greenland dykes (WG) after Nielsen [1987]. Distribution of
volcanism after White and McKenzie [1989] and Coffin and
Eldholm [1994]. Continental reconstmctionafter White and
McKenzie [1989]. Large star locatesplume center of White and
McKenzie [1989]; small star locatesthe proposed60 Ma hot spot
positionof Lawvet and Miiller [ 1994].
associatedwith intrusivecomplexes[Speightet al., 1982;
MacDonald et al., 1988]. Dyke swarmsaboundalongand
orientedparallel to the coastof east Greenland[Nielsen,
1987]. In west Greenland,a minor dyke set is associated
with volcanicactivity [Nielsen,1987].
In a reconstruction of' the North Atlantic (Figure 4), the
dominantdykesof the BritishTertiaryProvinceandthoseof
Greenland have overall trends which roughly converge
towardsthe plumecenterof WhiteandMcKenzie[ 1989].On
the other hand, the dykes of Greenlanddo not converge
towardsthe alternative60 Ma hotspotlocationof Lawver
and Miiller [ 1994]. Therefore,on the basisof the radiating
swarm model, we prefer the White and McKenzieplume
location.This is also the locationof the Kangerdlugssuak
triplejunctionof Burkeand Dewey[ 1973].
The British Tertiary dykes show a prominentswing in
trend from southeastto east-southeast.This change in
direction probably reflects the decreasinginfluence of
plume-generated upliftandincreasing influenceof a regional
stresscomponent[Ernstet al., 1995a,b].Similardeviation
from a simpleradiatingpatternis observedin severalother
of the largestdyke swarmsdescribed
below.
2.5. Deccan Event
The Deccanflood basaltprovincein India (Figure5) hasa
volumeof morethan8 x 106km3,andis associated with the
separationof India from the Seychelles[e.g., Mahoney,
1988; Whiteand McKenzie, 1989, 1995]. The majorityof the
magmatismwas quiterapid, 65.5+0.5 Ma (lo) [wholerock
4øAr-39Ar from Baksi,1994]andis thoughtto be relatedto
the arrival of the R6unionplume [e.g.Morgan, 1981;Storey,
1995]. The presentdrainagepatternis consistentwith the
existenceof a Late Cretaceoustopographicswell centered
abovethe plumecenter[Cox, 1989].
Locally, the distributionof dykes associatedwith the
Deccanflows can be quite complex,with radial, crosscut-
I
I
Seychelles
R i i
1 ooo km
Figure 5. Dyke swarmsof the - 66 Ma Deccanmagmaticeventof
India andthe Seychelles.Componentswarmsincludethe Narmada-
Tapti-Son(N), PanvelFlexure
pn(P), CambayGraben pn(C), and
Mt. Girnarpn (M) swarms,and probablythe Ranipn (R) swarm.
Distributionof dykesfollows Murthy [1987] andRadhakrishnaet
al. [1994]. Distribution of flood basalts (shaded), region of
anomalouslyhot mantlearoundthe plume(circle)and continental
reconstructionfor >65 Ma are after White and McKenzie [1989].
Starshowslocationof plumecenterafterBhattacharjiet al. [1994].
Diamond locatescenterof magmatismof Hooper [1990] and is
'nearthe plume centerof Whiteand McKenzie[1989]. The Kutch
regionis located
justnorthof theMt. Gimarpn

ting, curved,random,andsubparallel
patterns[Auden,1949,
Karkare and Srivastrava, 1990; Sant and Karanth, 1990;
Hooper,1990].However,severalregional-scale
swarmsare
observed. The most dramatic is the east-west-trending
swarmwhich Followsthe Narmada-Tapti-Son
Lineament,an
often-reactivated Precambrian suture zone, for more than
600 km [e.g.,Murthy, 1987;Deshmukhand Sehgal,1988].
East-trendingdykesof the Mt. Gimar areacouldbe viewed
as a continuationof the Narmada-Tapti-Sonswarm. The
dykesassociated with the Panvelflexuretrend north-south
[e.g., Deshmukhand Sehgal, 1988] and are of at leasttwo
generations:oldertholeiiticpre-flexuredykeswhichdip 60-
80ø east and younger alkaline dykes which are vertical
[Dessai and Bertran& 1995]. The Cambay graben is
paralleledby north- to northwest-trendingdykes[Bhattac-
harji, 1988].
The dykesof the Narmada-Tapti-Son lineament,Cambay
graben,Mt. Girnar area, and ?anvel flexure showa rough
convergencetowarda localitywhich hasbeeninterpretedas
a plume center mainly on the basis of rift convergence
[Bhattacharji,1988; Bhattacharjiet al., 1994] (Figure 5).
This location was also identified as the Gulf of Khambat
triplejunctionby BurkeandDewey[ 1973].
This plumecenterdiffersfrom the plumecentersuggested
by Whiteand McKenzie[ 1989] andthe centerof magmatism
proposedby Hooper [1990], bothof whichare locateda few
hundredkilometersto the south(see symbolin Figure 5).
Hooper's [1990] centerof magmatismis marked by large
randomly orientedfeeder dykes which he consideredthe
primaryfeedersfor the DeccanLIP. However,Karkare and
Srivastava[ 1990] and Bhattacharjiet al. [ 1994] identified
primary feederdykesalignedalongthe Narmada-Tapti-Son
zone. Furthermore,dyke frequencyalong the Narmada-
Tapti-Sonzone increasestowardthe plumecentershownin
Figure 5 [Bhattacharjiet al., 1994]. Additionalmain feeder
dykes were identified by Karkare and Srivastava[1990]
parallelingthe Panvelzone,andin the Kutchregion.
There clearly are complexitiesto the overall radiating
swarmpattern.There are somecrosscutting trendsthrough-
out the region and randomlyorienteddykesin local areas
[e.g., Hooper, 1990]. The complexity in the dyke
distribution may reflect the superpositionof an early
radiatingswarm, associatedwith plume head arrival, and
later dykes whose orientationis controlledby both rifting
and the changinguplift topographyas the plume flattened
againstthe lithosphere[Griffithsand Campbell,1991].
There are otherdyke swarmsof similaragethat are likely
associatedwith the Deccan event. Deccan-agedykes of
unspecifiedorientationare reportedfrom the Seychelles,
which were adjacentto India at that time [Devey and
Stephens,1991]. Also, in the Kerala region of southwest
ERNST AND BUCHAN 305
India, the northwest-trending
Ranipndykes (Figure 5) have
anageof 69 + 1 Ma (4øAr-39Ar
wholerock)[Radhakrishna
et
al., 1990, 1994].
2.6. MadagascarEvent
Widespread Cretaceousflood basalts are present in
Madagascar(Figure 6) and can be relatedto the track of the
Marion hotspotand the breakupof Madagascarand India
[Mahoneyet al., 1991;Storey,1995;Storeyet al., 1995]. On
the basisof numeroushigh-precision
4øAr-39Ar
analyses
(whole rock and feldspar)this LIP hasbeendatedat 90-83
Ma [Storeyet al., 1995].
Dyke swarmsassociated with the volcanismare mapped
in three regions[Storeyet al., 1995]. Together,the main
dyke trends in each region form a radiatingpatternwhich
fansover about65 o and convergesnearthe estimated88 Ma
locationof the Marion hotspot(Figure6). We interpretthis
focal point (large star) to mark the arrival of the Marion
plume.
BecausesouthwestIndia was adjacentto Madagascarat
that time, Storeyet al. [1995] suggestedthat somedykesof
India may be relatedto thisevent.The K-Ar wholerock age,
81 +3 Ma, Fornorth-northwest-trending Palaipndykes [e.g.,
Radhakrishnaet al., 1990; 1994] is roughlysimilarin ageto
the Madagascar LIP, and, therefore, may be related.
However,they do not fit the radiatingpatternder'reedby the
majorityof dykesof Madagascar.
2.7. Paran6-Etendeka Event
The separationof SouthAmericaandAfrica is associated
with the arrival of the Tristanda Cunhaplume at-130 Ma.
Melting in thisplumeheadis thoughtto be responsible for a
major LIP, the Parantibasalt province of easternSouth
Americawith a volumeof >1.2 x 106 km3, andthe smaller,
originally contiguous,Etendekaprovince of southwestern
Africa (Figure 7) [e.g., Storey, 1995; Whiteand McKenzie,
1989].The agerangeof theParantievent(basedonthe4øAr-
3OArmethod)hasbeenreportedas 133-131Ma [Renneet
al., 1993] and as 138-127 Ma [Turner et al., 1994; Stewart
et al., 1996].Identical4øAr-39Ar
agesof 133-131Ma arealso
reported for the Etendeka LIP [Renne et al., 1996b]. The
presentdrainagepatternis consistentwith an Early Creta-
ceous topographicswell centeredover the plume center
[Cox, 1989].
The most prominentdyke swarm of the Parantievent is
the Ponta Grossa swarm, which is well der'reed over a
distanceof about 300 km [Sial et al., 1987; Raposoand
Ernesto, 1995] and has an age of 132-129 Ma [Renneet al.
1996a]. Other smallerswarms,the Santos-Riode Janeiro
306 GIANT RADIATING DYKE SWARMS, LIPS, AND MANTLE PLUMES

(Figure 7). However, as notedby Renneet al. [1996a], the

actualsite of plume impactmay have occurredfarthersouth
MADAGASCAR becausethe age of the Ponta Grossaswarmpostdatesthe
main magmatismof the southernParan/tlavasto the south
(133-132 Ma) by a few million years.
In Africa, dyke swarms associatedwith the Etendeka

IA/DI,
4 event are poorly defined, largely becauseof difficulty in
distinguishingthem from Karoo-agedykes (section2.8).
Nevertheless,somesmall swarmsare evident.The Horing-
baai dykes, which include both north- and northwest-
trending sets [Namibia Geological Survey, 1988], have

500 km

Figure 6. Dyke swarmsof the -88 Ma Madagascarmagmatic

event. Distribution of dykes and volcanic rocks (shaded) in
Madagascarare after Stor•y et al. [1995]. E, M, and T are the
Ejeda-Bekily,Morondava pnandTamatave pndykes,respectively, of
Madagascar. The Palaipn(P) swarmof India [Radhakrishna et al.,
1994] is possiblycorrelative.The small star and circle locatethe
88-Ma centerand possibleextentof the Marion plume [Storeyet
al., 1995]. The large star locatesthe convergence point for the
dykes.Reconstruction for-88 Ma is afterStoreyet al. [1995].

Florian6polisswarms,are coastparallel and togetherwith

the PontaGrossadykesform a convergentpatternfocussed
towardsthe southeast.
Stewartet al. [ 1996]reportedagesof
133.3-129.4Ma (4øAr-39Ar)
for the Santos-Rio
de Janeiro
swarm,whereasRenneet al. [1996a] indicatedthat the coast
Figure 7. Dyke swarms of the ca. 130 Ma Paranfi-Etendeka
parallel dykes may be youngerwith agesof 126-119 Ma
magmaticevent of South America and Africa. In South America,
(4øAr-39Ar).
A thousand kilometersto thewestandnorthwest
the Paranti.swarmsincludethe Ponta Grossa(PG), Paraguay(P),
of the plume center,respectively,are the West Bodoquena Santos-Riode Janeiro (S), Florian6polis(F), West Bodoquena
andSerrodo Caiap6swarmswhichparallelthe PontaGrossa (WB), and Serrodo Caiap6 (SC) swarms.In Africa, the Etendeka-
swarmand are approximatelycontemporaneous with it [Sial agedswarmsincludethe Etendeka(E), Horingbaai(H; north-and
et al., 1987]. Finally,the largenorthwest-trending
Paraguay northwest-trending setsshown)and Cape Peninsula(CP) swarms
swarm, which is traced from aeromagneticmaps, is also andthe singleMehlberg(M) dyke. SouthAmericandyke distribu-
thoughtto belongto the Paran/tevent [Drueckerand Gay, tions are after Sial et al. [1987], Druecker and Gay [1987] and
1987] althoughdykesseemto have a slightlyolder age of Turner et al. [1994]. Informationon the Florian6polisdykes is
from P. Comin-Charamonti(1995, pers.comm.)and Piccirillo et
138-13 7 Ma [Stewartet al., 1996].
al. [1991]. Dykes of Africa are after Namibia GeologicalSurvey
Studiesof the anisotropyof magneticsusceptibilityof the [1988], Duncan et al. [1990], Reid [1990], Reid and Rex [1994],
PontaGrossaswarm indicatemainly subhorizontal magma andMubu [1995]. Distributionof volcanicrocksis after Whiteand
flow, excepttowardsits easternend wherethe inferredflow McKenzie [1989]. Continentalreconstructionfor-130 Ma after
is typically steeper[Raposoand Ernesto,1995]. Thesedata O'Connor and Duncan, [1990]. Star locatesapproximateconver-
are consistentwith a plume sourceto the eastof the swarm gencepointof
 ERNST AND BUCHAN 307

MORB-like chemistry and intrude the basal units of the Burke and Dewey [1973]. The presentdrainagepatternsin
Etendeka[Duncan et al., 1990]. A coast-parallel swarm(not southeasternAfrica are consistentwith a Jurassictopo-
shown in Figure 7) was suggested by Hawkesworthet al. graphic swell centeredon this plume center [Cox, 1989].
[1992], but its extentis unclear(D. Peate,pers.comm. 1995; Possiblecandidateplumesare the Marion hotspot[Morgan,
A. Duncan,pets.comm. 1995). However,a recentaeromag- 1981; Richards et al., 1989; Duncan and Richards, 1991]
netic interpretationrecognizedan Etendekaswarmwhich is andthe Bouvethotspot[Lawveret al., 1992;Storey,1995].
coastparallel but is located farther from the coast [Mubu,
1995]. Along the coastto the southare northwest-trending
dykes which collectively are termed the SouthernCape
dykesby Mubu [ 1995]. From southto north,they consistof
the CapePeninsuladykeswith an ageof 132 +6 Ma (2(•; K-
Ar whole rock) [Reid, 1990; Reid et al., 1991a], the
Cederbergdykes (not shown)[Hunter and Reid, 1987] of
unknownage and the Mehlbergdykewith an age of 134 +3
Ma (4øAr-39Ar
wholerock)[ReidandRex,1994]. /R
Most of the dyke swarmsof the Parantiand Etendeka
eventsconvergetowardsthe generalregion advocatedfor
the plume centreusingothercriteria.Our preferredlocation
basedon the dyke patterndiffersslightlyfrom that of other
authors[e.g., WhiteandMcKenzie,1989;Harry andSawyer,
1992] but is similarto that of the SaoPaulotriplejunctionof
Burkeand Dewey [ 1973]. The slightmisfitof the Santos-Rio
de Janeiroand.Florian6polis subswarmsmay be explained
Falkland/Malvinas
by their later emplacementalongthe riff margin.The weak Plateau
convergenceof the Paraguay,Serro do Caiap6, and Cape
Peninsulasubswarmsmay reflect the deflectionof these t I
1000 km
dykesinto a regionalstressfield outsidethe influenceof the
plume-uplift [Ernst et al. 1995b] or that plume impact ANTARCTICA
actuallyoccurredfurther south [Renneet al., 1996a]. The
northwest-trendingHodngbaai dykes fit the radiating
•%,,\•''''//
pattern.However,the role of the north-trending Horingbaai
dykesin thetectonicstoryremainsunclear. Figure 8. Dyke swarmsof the 184 Ma Karoo magmaticevent in
southeastAfrica. The Botswanadykes (B), OrangeRiver dykes
2.8. Karoo Event (O), andthe Lebombodykes(Le) [Vail, 1970; Wilsonet al., 1987;
Mubu, 1995] radiatefrom the Lower Limpopotriplejunction(large
Large parts of the Karoo magmaticprovinceof south- star)of [Cox, 1970; Burke and Dewey, 1973]. The possiblearcuate
continuationof the OrangeRiver dykesis indicatedby a dottedline
easternAfrica, and the Ferrarmagmaticprovinceof Antarc-
[Mubu, 1995]. The Limpopodykes(L) are alsoof Karoo age,but
tica were emplacedin perhapslessthan 1 million yearsat crosscutthe Botswanaswarm[Wilson,1990]. The Rushinga(lower
184Ma, on the basisof U-Pb baddeleyite/zircon
and4øAr- Zambesi,Imhamangombe) (R), Cholo (ShireHighlands)(Co) and
3OAr
dates[Encarnaci6n et al., 1996andreferences
therein]. Gorongoza(Go) swarms[Woolleyand Gatson, 1970; Cox, 1970;
The volumesof Karoo andFerrarmagmatismareat least2 x Vail, 1970; MacDonald et al., 1983; Eales et al., 1984; Wilson et
106 km3 and 0.5 x 106 km3, respectively.
The Karoo al., 1987] appearto radiatefrom the Lower Zambesitriplejunction
magmatismand Ferrar magmatismin adjacentAntarctica of Burkeand Dewey[1973]. Numerousotherswarmsincludingthe
have beenrelatedto a mantleplume (Figure 8) which was east-trendingGap (G) [Cox, 1970] and northwest-trending
associatedwith the breakupof Africa and Antarctica[e.g., Cederberg(C) dykes[Hunterand Reid, 1987;Mubu, 1995] arenot
obviouslycorrelatedwith any singlecenter.The distributionof
Whiteand McKenzie,1989]. The part of the Ferrarprovince
volcanicrocks,the Africa-Antarcticareconstruction for this period
located in the TransantarcticMountains,althoughcoeval andthe circlewhichencloses the regionof anomalously hot mantle
with the Karoo has differentcompositionand may have a aroundthe Lower Limpopo plume are after Whiteand McKenzie
subduction-relatedorigin [Encarnaci6net al., 1996]. The [1989]. The TransantarcticMountains (unlabeled) containing
Karooplumecenterof Whiteand McKenzie[ 1989] is similar related Ferrar magmatismare located near the bottom of the
to the Lower Limpopo triple junction of Cox [1970] and diagramin
308 GIANT RADIATING DYKE SWARMS, LIPS, AND MANTLE PLUMES
At least three prominentdyke swarmsof Karoo age
convergeon the Lower Limpopoplumecenter[Figure16.7
of Winalley,1984]. Theseincludethe Botswana(1500 km
long), Lebombo,and OrangeRiver swarmswhich form a
convergent patternspanning100ø of arc(Figure8). Prelimi-
naryestimates usingmeasurements of anisotropyof magnet-
ic susceptibilityindicateverticalflow at the easternend of
the Botswanaswarmand dominantlyhorizontalflow farther
west--an observation consistentwith a plume source
underlyingthe focalregion[ErnstandDuncan,1995].
Accordingto WindleyI1984; figure 16.7], other Karoe-
dge dyke swarmsconvergeon the Lower Zambesitriple
junctionof BurkeandDewey[ 1973],about700 km fromthe
LowerLirapopetriplejunction(Figure8). Theseincludethe
Rushinga,Cholo,andGorongozaswarms.Thisfocalregion
and the associatedvolcaniccenters[MacdonaMet al., 1983]
may definea secondsiteof Karoo-ageplumearrival.
There are numerousotherKaroo-agedswarmswhich are
not obviouslycorrelatedwith eitherthe Lower Limpopoor
Lower Zambesiplume centers.For example,the Limpopo
dykescrosscutthe Botswanaswarm[Wilson,1990] and so
are not relatedto the Lower Limpopoplumecenter. Sim-
ilarly, they intersectthe Gorongozaswarmandso probably
arenot relatedto the LowerZambesiplumecentereither.
2.9. Additional Cases
Dyke swarmsare associatedwith other LIPs, but their
distributionand (or) age are as yet rather poorly docu-
mented.For example,the volcanicrocksof the Alpha Ridge
oceanicplateauand nearbyQueenElizabethIslandsof the
CanadianArctic (Figure 9a) form a largeigneousprovince
which may have been producedby a plume [Embryand
Osadetz,1988]. Most of this volcanismis Cretaceousin age
(Hauterivianto early Cenomanian).Dykesof Cretaceous age
also occur over a wide area on the northem Arctic islands
and Embry and Osadetz[1988] notedthat many appearto
radiatefrom the plume.In detail,the HazenStraitpn swarm,
based largely on aeromagneticinterpretation,and the
LightfootRiver swarmmay representsubswarms of a giant
radiating dyke swarm whose focal point marks a plume
center. Precisedating is necessaryto test this model. The
tectonic relationshipof similar-ageeast-trendingSurprise
Fiordpl, dykesthatcutthe areais not clear.
A secondexampleinvolvesthe SiberianTraps(Figure9b)
whichcoveran areaof 1.5 x l06 km2 in the Siberian
plat-
form [Zolotukhinand Al'Mukhamedov,1988]. They were
apparentlyextrudedwithin a periodof 5 to 10 m.y. at -248
Ma, basedon4øAr-39Ar
dating[RenneandBasu,1991;Baksi
and Farrar, 1991]. Severaldykeswarmsareassociated
with
the Siberian Traps. Two Perme-Triassic swarms, the
800
N 100
øN Laptov
(a)'N•."•':"
"•!•••."•••••••••
'"'":'
' cean il•' Tamyr
Belt
Sea
Queen
Elizabeth
/':•:•'••'• I /:"::A•Z•?•:•:•:•:•:•::
Islands /:•"•;.•.•}::• • ::
M'
•(::'•::::::::?•?:•::•::•:
si
Figure 9. Additionalcases.(a) Dyke swarmsof the Alpha Ridge
event of northernNorth America (modified after Embry and
Osadetz[1988]). Dyke setsincludeHazenStrait pn(H), Lightfoot
River (L), and SurpriseFiordpn(S). Locationof AlphaRidgeand
Queen Elizabeth Islands volcanism from Embry and Osadetz
[1988]. (b) Dyke swarmsof the SiberianTrapsof Siberia[Krasnov
et al., 1966; Blagovyeshchenskaya,
1973; Thomshinand Okrugin,
1995]. Dyke setsincludethe Ebekhaya(E), Maymecha pn (M),
Kureykapn(KU) and Kochikhapn (KO). There are alsodeformed
Perme-Triassicdykes in the Taymyr belt to the north [Vakar,
1962]. Distributionof SiberianTrap volcanicsis after Zolotukhin
andAl'Mukhamedov[ 1988].
Ebekhayaand Maymechapl•swarms,cut basementrockseast
of the main accumulationof volcanic rocks [Blagovye-
shchenskaya,1973; Tomshinand Okrugin, 1995] and may
be part of a giant radiating pattern. Other swarmsdo not
readily fit this pattern. Kureykapndykes cut the volcanic
rocks and may be younger.Kochikhapndykesform an arc
about the focal point of the Ebekhaya and Maymechap n
swarms.They couldrepresenta giantcircumferential swarm,
locatedalong the rim of a centralmega-caldera,similarto
the mega-calderasuggestedfor the Mackenzie event (see
section4.9) [Bardgat et al., 1996]. Alternatively,they may
representa featureassociated with coronae,a type of plume
collapse feature common on Venus but not previously
identifiedon Earth(R. E. Ernstand K. L. Buchan,Identify-
ing VenusJanCoronaeon EarthUsingGiant Circumferential
Dyke Swarms,manuscriptin preparation).Circumferential
swarmshavebeenobservedon a muchsmallersizein the
GalapagesIslands[Chadwickand Dieterich,1995].
3. GIANT DYKE SWARMS
Large dyke swarms were emplaced throughoutthe
Proterozoicand the Phanerozoic(Figure 1). Severalof the
youngerswarmsare part of known continentalfloodbasalts
as outlined in the previoussection.Many more
 ERNST AND BUCHAN 309

especiallythoseof pre-Mesozoicage,arenot associated with

known continentalflood basalts.Before discussing the uses
of giant dyke swarmsin identifyingthe localitiesof mantle
plume centersand associatedLIPs (section 4), and in
understanding characteristicsof plume-generated
LIP events
(section 5), it is instructiveto summarizewhat is known
aboutgiantdyke swarms.

3.1. General Characteristics

Of a total of more than 500 diabasedyke swarmsrecently

cataloguedin the Global Mafic Dyke GIS DatabaseProject
[Ernstet al., 1996], 141 are classifiedasgiantdyke swarms
with lengths_>300km [Ernst et al., 1995b]. Of these,63
havelengths_>500km and 17 have lengths>_1000km. Giant
dyke swarmsare concentratedin North America, eastern
South America, southernAfrica, India, and Australia. This
distributionprobablyreflectsmore completeinformationon
dykesandtheir distributions in theseareasthan is available
elsewhere.
Giant dyke swarmscan be groupedinto five typesbased
on their geometry[Ernst et al., 1995b] (Figure 10). Three
types exhibit a radiatingpattern:type I has a continuous
fanningpattern,type II hasa fanningpatternsubdivided into
separatesubswarms,and type III has subswarmsof sub-
parallel dykeswhich radiate from a commonpoint. Two
types show a linear pattern:type IV has subparalleldykes
distributedover a broad zone and type V has subparallel
dykesrestrictedto a narrowzone. SubparalleltypesIV and
V could, in some cases,representdistalportionsof typesI Figure 10. Characteristicpatternsof giant dyke swarms.Swarms
and III, respectively,if the segmentobservedis outsidethe aredividedinto 5 types:I= continuous fanningpattern;II= fanning
pattern subdividedinto separatesubswarms:III= subswarmsof
regionof plume-induced uplift andsubjectonlyto a regional
subparallel
dykeswhichradiatedfrom a commonpoint;I,_¾•,
stressfield [Ernstet al., 1995a,b].
subparalleldykes distributedover a broad area; V- subparallel
Beyond a certain distance,giant radiatingswarmsmay dykes restrictedto a narrow zone. Starslocateprobablemantle
becomesubparallel,as the influenceof the regionalstress plume centersas definedby radial dyke patterns.After Ernst et al.
field exceedsthose of the plume-generated uplift [Ernst et [1995b] and reprintedwith permissionof A. A. Balkema,Rotter-
al., 1995a,b]. On the other hand, near the plume center, dam.

dykesmay exhibitvariabletrendsasa resultof beinglocated

on the relativelyflat portion(the crest)of the topographic
uplift, where there are insignificantdifferentialhorizontal
stresses.
3.2. RelationshipwithPlumes
Recent work has indicated a direct connection between the
emplacement of giantradiatingdykeswarms(typesI, II, and
Ill) and the arrival of mantle plumes at the base of the
lithosphere[LeCheminantand Heaman, 1989; Heaman,
1991; Ernst et al., 1995a,b].Evidenceincludes(1) a radiat-
ing dyke patternsuggestiveof a centrallylocatedmagma
source;(2) coevalvolcanicand plutonicrocksin the focal
regionwhich may representthe remnantsof LIPs emplaced
above a mantle plume; (3) lateral magma flow in dykes
beyondthe focal region; (4) uplift of the focal region in
responseto the arrival of the mantleplume;and (5) rapid
emplacementof someswarms.Each of thesecharacteristics
is elaborated below.
(1) The radiatingpattemand continuityof swarmsfor up
to 2500 km suggests thatthe magmasourcewas localizedin
the vicinity of the focal area [Ernst et al., 1995a]. In some
casesthe giantradiatingpatternonly becomesapparentafter
continents have been reassembledinto their pre-drift
configuration[e.g.,May, 1971;Park et al. 1995;Ernstet al.,
310 GIANT RADIATING DYKE SWARMS, LIPS, AND MANTLE PLUMES
(2) Key evidencesupportingthe presenceof a plume
centerbeneaththe focal region of radiatingswarmsis the
abundanceof coevalvolcanicrocksandplutonsin the focal
region [Fahrig, 1987]. However,it shouldbe notedthat in
somecasesthe volcanismwas displacedup to two or three
hundredkilometersfrom the center(section5.1).
(3) Studiesof the anisotropyof magneticsusceptibility
of
giant radiating dyke swarmsindicatethat, far from the
swarm'sfocus,magmaflow was subhorizontal [Greenough
and Hodych, 1990; Ernst and Baragar, 1992; Raposoand
Ernesto,1995; Ernst and Duncan, 1995]. On the otherhand,
magma flow was subverticalin the focal area [Ernst and
Baragar, 1992; Ernst and Duncan, 1995]. This patternis
consistentwith magmainjectedasthin verticalbladesfrom a
mantleplumebeneaththe focal areaandflowing laterallyto
greatdistances,as much as 2000 km or more [e.g., Ernstet
al., 1995a].
(4) Broadregionaluplift hasbeendocumented
in the focal
regionof somegiantradiatingdyke swarms[e.g.,LeChem-
inant and Heaman, 1989; Rainbird, 1993]. Becausetopo-
graphicuplift is associated
with the arrivalof mantleplumes
[e.g., Whiteand McKenzie,1989;Cox, 1989;Richardset al.,
1989; Griffithsand Campbell,1991], it followsthatuplifted
focal regionsof dyke swarmsalso mark the arrival of a
plumehead [LeCheminantand Heaman,1989]. On the other
hand,it is unclearat presentwhetherplumetails[Richardset
al., 1989] can generategiant dyke swarms.The smaller
magnitudeof volcanismassociated with plumetails suggests
that dyke swarmsassociatedwith plume tails shouldbe
muchsmallerthanthoseassociated with plumeheads.
(5) High precision U-Pb dating of baddeleyitefrom
severaldykesof the Mackenziegiantradiatingdyke swarm
(see section 4.9) has demonstratedthat emplacement
occurredwithin a few million years (1267ñ2 Ma; 20)
[LeCheminantand Heaman, 1989]. This result is consistent
with the observationsfrom the Columbia River, Deccan,
Karoo,and SiberiaLIPs for rapidemplacement in lessthana
few million years[e.g., seerefs.of Storey,1995, Whiteand
McKenzie, 1995]. However, as discussedin section 5.3,
evidencefrom otherdyke swarmsindicatesthat somedyke
eventsmay be episodicoverintervalsof up to 30 m.y.
3.3. Origin of TypeI, II, and III Swarms
The differences betweentype I, II, and III patternsmay
reflect the distributionof high-levelmagmachambersand
(or) thepresence of coevalor pre-existing
rift zones.
Baragar et al. [1996] have discussedthe distributionof
high-levelmagma chambersin the vicinity of the plume
head associatedwith the Mackenzieswarm(Figure 1l a).
They suggestthat magma accessand loci for magma
chamberdevelopmentwere providedby graben collapse
nearthe uplift core.
In this model, magma chambersgrow at a number of
centersdistributedconcentricallyaboutthe centerof uplift,
with each chamberresponsiblefor the injectionof a sub-
swarmof dykes.If the chambersare closetogetherthen no
gaps in the dyke distributionshouldbe observed(type I,
Figure 10). The classictype I exampleis the Mackenzie
dyke swarmof northernNorth America(seesection4.9). If
the chambersare widely spaced,however,dyke-poorgaps
will be presentbetweensubswarms asin typeII (Figure 10).
The best example of a giant swarm exhibitinga type II
pattern is the Matachewanswarm of North America (see
section4.11). Other type II examplesincludethe Columbia
River Basalt Group dykes (Figure 2) and the swarms
associated with the LowerLimpopocenter(Figure8).
The presenceof coevalor pre-existingrifts may causea
concentration of dykesin the vicinity of and parallelto the
rifts. This model is best suitedto type III swarms(Figure
10). In detail, Fahrig [1987] suggestedthat dykes are
associated with continentalbreakupandemplacedparallelto
(passivemargin dykes) and perpendicular to (failed arm
dykes)the rift margin (Figure l lb). Examplesof passive-
margindykesincludethe Red Sea dykesof the Afar event
(Figure 3), the east Greenlanddykesof the North Atlantic
Tertiary event (Figure 4), the Panveldykesof the Deccan
event (Figure 5), Santos-Riode Janeiroand Florian6polis
dykes of the Paranftevent (Figure 7), and the Lebombo
dykes of the Lower Lhnpopo (Karoo) event (Figure 8).
Failed-armtypesmay includethe Narmada-Tapti-Son dykes
of the Deccan(Figure5), the Botswanaswarmof the Lower
Limpopoevent (Figure 8), the Grenvilleswarmof eastern
North America (section4.4), and the Yakutsk dykes of
Siberia(section4.3).
A consequence of the Fahrig [1987] model is that most
passive-margindykes are destroyedor deformed in a
subsequentcollision, whereas failed-arm dykes survive
largely intact (Figure 1lb). This may explainwhy many
dyke swarmsare truncatedat one end by youngerorogenic
belts[Buchanand Halls, 1990].
If both widely spacedmagmachambers and coevalrifts
are present,then giant swarmsmay exhibit geometries
intermediatebetweentypesI, II, and lII. For example,the
CentralAtlanticreconstructed swarm(Figure12 andsection
4.1) appearsto exhibita rathercontinuousfanbut alsohasa
concentrationof dykesalongthe Africa-SouthAmericaand
theNorth America-Africarift margins.
4. APPLICATION OF GIANT DYKE SWARMS IN
IDENTIFYING PALEOPLUMES AND PALEO-LIPS
In this section we summarizeplume-related
 ERNST AND BUCHAN 311

1. DYKEINTRUSION

•- P'•R-•
S i•---•
Failed
arm
Figure 11. (a) Model of Baragar et al. [1996] for magmatism
relatedto plume-induceduplift as describedin the text. Reprinted
with permissionof Oxford UniversityPress.(b) Model of Fahrig
[1987] for three-stage
platetectoniccyclein the evolutionof mafic
continentaldyke swarms.Reproducedwith the permissionof the
GeologicalAssociationof Canada.
3. COLLISION
events that have been identified on the basis of their
convergingdykepattems.

4.1. Central Atlantic Event

The CentralAtlantic reconstructed swarmis an example

of a dyke swarmdismembered by plate tectonicprocesses
(Figure 12). Sincebeing first identifiedby May [1971] in a
pre-Atlanticcontinentalreconstruction, the 270ø fan has
remained the classic example of a reconstructedgiant
radiatingdyke swarm [Oliveira et al., 1990; $ebai et al.,
1991b;Bertran& 1991; Ernst et al., 1995a]. Dykes of this
swarmare tracedup to 2800 km from the plumecenter.The
North Americanportionof the swarm,knownasthe Eastern
North America dykes, is the best studiedgeochemically aL, 1991b]. The convergentpoint of the Central Atlantic
[e.g., McHone et al., 1987; de Boer et al. 1988; McHone,
reconstructedswarm and the interpretedlocation of the
1996]andthe mostpreciselydated,at 201 +_2Ma (2c•;U-Pb)mantleplume is the Blake Plateau.
[Dunning andHodych, 1990].Dates of206-196(4øAr-39Ar The plumethoughtmostlikely responsiblefor the Central
on plagioclase)have been obtainedfor related dykes in Atlantic reconstructed swarm is the Fernando de Noronha
MoroccoandAlgeria, Mali, andIberia(not shown)[Sebaiet hotspot[Hill, 1991]. The CapeVerde plumewas
312 GIANT RADIATING DYKE SWARMS, LIPS, AND MANTLE PLUMES

sedimentationin the Newark-type rift basinsof eastem

I
1000 km North America[Hill, 1991]. Sedimentary deposition
in these
basinsbeganat -230 Ma but was interruptedat 215-210
Ma, presumablyby uplift associated with the approaching
plume.Volcanismcommenced at -200 Ma. The presence of
NORTH the rift basinsprior to volcanismindicatesthat the region
E was under tensionbefore the plume arrived. The plume
AMERICA model is supportedby preliminarymagneticfabric studies
from the EasternNorth Americadykes.Thesedataindicate
M predominately subhorizontal magmaflow [Greenoughand
Hodych,1990;Ernst et al., 1995c]in dykeslocatednorthof
about41øN (i.e., distalfrom the plumecenter).
Interestingly,despite the large extent of the Central
Atlantic reconstructedswarm, the amount of volcanism in
the focal region is minor [Hill, 1991]. Some associated
volcanicrocksexist in the subsurfaceof Florida [de Boer et
al., 1988; Heatherington and Mueller, 1991] and in
Morocco [Sebei et al., 1991b; Bertrand, 1991]. Extensive
AFRICA volcanismalsooccurredalongthe entireeastcoastof North
America,the so-calledEastCoastMargin igneousprovince
[Holbrook and Kelemen, 1993]. However, these volcanic
rockshave been relatedto the transitionto oceanicrifting,
ratherthanto an initial plumeevent[e.g.Austinet al. 1990;
Holbrookand Kelemen, 1993; Holbrook et al. 1994].

4.2. Jutland Event

We definethe Jutlandevent(Figure13) onthebasisof the

convergenceof three swarms:the Whin-MidlandValley,
SOUTH AMERICA Oslo Rift and Scaniaswarms,all of age-300 Ma. The most
impressiveof these is the >400-km-long, east-trending
Whin-Midland Valley swarm in the British Isles and the
Figure 12. Dyke swarmsof the 200 Ma CentralAtlanticmagmatic adjacentNorth Sea [MacdonaMet al., 1981;Smytheet al.,
event of North America, South America, and Africa [after May, 1995].The largeWhin Valley andMidlandValley sills[e.g.,
1971; McHone et al., 1987; Oliveira et al., 1990; Sebai et al., Dunhamand Strasser-King,1982] are associated with this
1991b] in a reconstructionof Rowley [in Keppie and Dallmeyer,
swarm.In the Olso rift of Norway the oldestigneousactivity
1989]. In North America these consistof the EasternNorth consists of a swarm of north-northwest- to north-northeast-
America(E) subswarmwhich is locally crosscut(dashedlines)by
the slightly younger Charleston(C) subswarm[Ragland et al., trendingdykeswith an ageof 2974-9Ma (2F; Rb-Srmineral
1983;Smith, 1987; de Boer et al., 1988]. In Africa, the dykes[e.g.,isochron) [Suntivoli and Larsen, 1993]. In addition,
Sebaiet al., 1991b] are dividedon the basisof distributioninto the northeast-trendingKongsbergdykeswith an age >275 Ma
Morocco p"(M), Liberia p"(L), andTaoudenni p"(T) subswarms.In may be coevalwith the 300 Ma dykes[Sundvolland Larsen,
South America numerous subswarms[Sial et al., 1987; Gibbs, 1993]. The east-southeast-to southeast-trending Scania
1987;Choudhuriet al., 1991] are groupedinto the north-northwest dykes of the Tomquist Line of southern Sweden
trending Amapfi (A) and related subswarms(Cassipor6,Jari, [Gorbatschevet al., 1987; Sundvolland Larsen,1993] have
Cayenne, Apatoe),andin Guyana p"(G), the northeast-trendingessentiallythe same age (294+4 Ma; K-Ar) [Klingspor,
Cerro Bolivar, Supenaamand Tukutu River and possiblythe Rio
Trombetassubswarms[Choudhuriet al., 1991; Ernst et al., 1995b,
1976]. Accordingto Gorbatschev et al. [1987, p. 370] this
1996;Sial et al., 1987; Gibbs, 1987]. The starmarksthe focalpoint swarm (Scania)
continues into Poland, beneathyoungersedi-
of the swarms. mentary cover.
The coevalnatureof the threemagmaticeventswasnoted
by Whiteand McKenzie [1989] but was locatedfar (-2000 by Smytheet al. [1995]. They linked the Whin-Midland
km) from the focal point of the swarmat 200 Ma [Morgan, Valley and Oslo Rift swarmsinto a singlearcuateswarm
1983]. Arrival of the plume is markedby an interruptionin explainedby a "concentration of regionalhorizonal

Figure 13. Dykes of the ~300 Ma Jutland event of Europe.
Distribution of Whin-Midland Valley (WM) dykes after Smythe et
al. [1995]. Oslo Rift (0), and Scania (S) dykes are after Gorbat-
schey et al. [1987] and the Kongsberg (K) dykes are after Sundvoll
and Larsen [1993] and references therein. Star marks convergence
of Whin-Midland Valley, Oslo Rift and Scania swarms. Small star
locates center of arcuate swarm defined by Whin-Midland Valley
and Oslo Rift dykes [Smythe et al., 1995].
stresses through the Faeroe-Shetland area" to the north.
They noted that the Scania dykes did not fit this model. We
note that the arcuate dykes may defme a giant circum-
ferential swarm, and the Scania dykes, a radiating swarm,
both associated with a plume centered to the north (see
model in R. E. Ernst and K. L. Buchan, in preparation). We
prefer an alternative plume model, in which the three
swarms-Whin-Midland Valley, Oslo Rift, and Scania-
defme a giant radiating swarm focussed on a mantle plume.
This model is consistent with the triple junction model (rift-
rift-rift type) of Burke and Dewey [1973], in which conver-
gence of the Midland Valley rift, the Oslo graben, and the
Danish Trough (on the south side of the Tornquist Line)
define the Jutland triple junction.
4.3. Yakutsk Event
The Devonian-age Yakutsk swarm (Figure 14) is located
in the Siberian craton and fans over an angle of about 150°.
Subswarms are associated with rift arms of a triple junction
[Shpount and Oleinikov, 1987; Gusev and Shpount 1987].
The Vilyuy-Markha and Chara-Sinsk subswarms are located
on either side of the Vilyuy rift and do not converge towards
the proposed plume centre because of subsequent Devonian
'v' -shaped opening of the rift [figures 21.3 5 and 21.40 of
Sengor and Natal'in, 1996]. The TomporukPn subswarm and
associated Dzhalkan volcanic rocks [Levashov, 1979]
paralleled rifting which probably resulted in separation of a
ERNST AND BUCHAN 313
narrow strip of Paleozoic passive-margin deposits [pp. 554-
556 of Sengor and Natal'in, 1996].
The Yakutsk swarm is poorly dated with most K-Ar whole
rock ages falling in the range between 375 and 320 Ma.
Nevertheless, the overall radiating pattern of the swarm and
the association with triple junction rifting and oCean opening
suggest a plume origin.
4.4. Central Iapetus Event
Several late Proterozoic dyke swarms of eastern North
America may locate a mantle plume associated with the
breakup of a late Proterozoic supercontinent and the
formation of the Iapetus Ocean (Figure 15). Ocean opening
occurred along nearly the same boundary as the subsequent
Atlantic Ocean.
Three subswarms are found on the Canadian Shield. The
east-trending Grenville dykes extend over a length of about
700 km and a width of 100 km and occur along the Ottawa
Graben, a failed Iapetus rift zone [Kumarapeli et aI., 1990;
SIBERIA
ANGARA
CRATON 500km
~ ~ Probable outer limit of
D
.... Precambrian basement
sea of
Okhotsk
115"E 130~
Figure 14. Dyke swarms of the -350 Ma Yakutsk magmatic event
of Siberia [after Shpount and Oleinikov, 1987]. CS and VM locate
the Chara-Sinsk and Vilyuy-Markha subswarms [Tomshin and
Koroleva, 1990; Erinchek et al., 1995] and D and T are the
Dzhardzhan and Tomporuk subswarms [Shpount and Oleinikov,
1987; Ernst et at., 1996].· Shaded pattern locates associated paleorift
[after Shpount and Oleinikov, 1987]. The outer limit of
Precambrian basement is from Rosen et at. [1994]. Star locates
focal point of swarm assuming some closure of the Vilyuy rift.
 314 GIANT RADIATING DYKE SWARMS, LIPS, AND MANTLE PLUMES

iscentered ontheoutcropsof TibbitHill metavolcanic

rocks
withan ageof 554 +4/-2 Ma (2c•;U-Pbonzircon)[Kumar-
apeliet al., 1989].The Adirondackdykeshavean arcuate
trend which also convergestowardthe SuttonMountains
NORTH_ triplejunction[St.Seymour
andKumarapeli,
1995].These
dykeshave K-Ar agesof 588-542Ma [Isachsenet al.,

AMERICA
] 'iii-'..' 1988].The northeast-trending
LongRangedykesoccurin
Newfoundland and southeasternLabrador. Based on U-Pb
baddeleyite
andzircongeochronology,
theirageis 615+2
Hudson Ma (2c0[Kamoetal., 1989;KamoandGower,1994].
Southof the CanadianShield,coast-parallel
dykesare
found over a lengthof nearly 1500 km in Precambrian
basement
massifsin the centralandsouthernAppalachian
Mountains.They are consideredto havebeenpart of the
easternmargin of Laurentiaduringthe late-Precambrian
riftingwhichformedthe IapetusOcean[Goldbergand
Butler, 1990]. TheseSouthern
Appalachianpn
dykeshave
beenvariablymetamorphosedto lowandmedium
grades
by
theAppalachian
orogenybutthecommonnortheast
trendof

' tlantic thedykesisconsidered

currently
available
tobeprimary.
fortheSouthern
Theonlyisotopic
Appalachianpn
date
dykesis a
734 +26 Ma (1c•)Rb-Srageat the southendof the swarmin
theBakersville subswarm [Goldberget al., 1986;Goldberg
andButler,1990].Thereisevidence fortwopulses of conti-
nentalrifting in the centralandsouthern Appalachians,at
550-600 Ma and at 700-760 Ma [Aleinikoffet al., 1995;
:...::::::::::::::::::::::::::::: TolloandHutson,1996].TheBakersvilleagesuggests that
at leastsomeof theSouthern
Appalachianp,
dykesbelongto
the older event.
The Long Range,Grenville,Adirondack, and perhaps
someof the Southern Appalachianp. swarmsbelongto the
youngerrift event.The radiatingdykepatternmay indicate
thatthisriftingwascausedby a mantleplume.The focal

' ,000m
' '"':':':...
point of the dykes is locatedfarthereastthan the Sutton
Mountainstriple junction even after correctionof the latter
for 100 km of Appalachian
thrusting[St. Seymourand
Figure15.Dykeswarms of the-600 Ma CentralIapetus
magmatic Kumarapeli, 1995]. This locationis similar to that of the
eventof easternNorthAmerica, associated
withtheopeningof the Montrealtriplejunctionof BurkeandDewey[ 1973].
IapetusOcean.LongRangedykes(L) areafterFahrigand West TheagerangefortheCentrallapetus eventis quitebroad,
[1986],Kamoetal. [1989]andKamoandGower[1994];Grenville at least25 m.y. (615-590 Ma). If the 554 Ma Tibbit Hill
dykes(G) are after St. Seymourand Kumarapeli [1995];
Adirondack dykes(A) are afterSt. Seymour and Kumarapeli volcanicrocksare alsoconsidered, thenthe agerangeis
[1995] and Coishand Sinton[1992].Southern Appalachianpn extended to about60 m.y. However,it is possible thatthe
dykes(S) areafterGoldberg andButler[1990]andarerestricted
to TibbitHill rocksarenotrelatedto thearrivalof a plumebut
exposedbasement massifs.
Starlocatestheapproximateconvergent insteadrepresentsrift products
associatedwiththeincipient
point of the swarms.ShadedarealocatesTibbit Hill volcanicrocks opening of theIapetus Ocean[St.Seymour andKumarapeli,
(including
outcrop
areas
andsubsurface
extensions
interpreted
from 1995].
gravitymodelling).
Dottedpatternlocates
Phanerozoic
coverrocks. The Baltoscandian dykes,distributedalongthe western
marginof Baltica,arealsothoughtto be associated
withthe
St. SeymourandKumarapeli,1995].Theyhavebeendated openingof theIapetusOcean[Andrdasson, 1994].Themost
at 590 +2/-1 Ma (2c•;U-Pb on baddeleyite)[Kamoet al., preciselydated of theseis the Sareksubswarmin northern-
1995].Grenvilledykesconverge slightlytowards
theeastin mostSweden(68øN)with an ageof-•606 Ma (U-Pbzircon
thedirectionof theSuttonMountainstriplejunction,
which and4øAr-39Ar
micaonbaked
wallrock)
[Svenningsen,

DuringtheCaledonide collision,thepassivemargincontain-
ing the Baltoscandian dykes shortenedby hundredsof
kilometers[Andrdasson,1994]. Therefore,their original
distributionandorientation,aswell astheirrelationship
with
the Central Iapetus dykes of Laurentia,remain to be
determined.
4.5. Franklin-Natkusiak Event
Franklin-Natkusiak
magmatismconsists of the Natkusiak
volcanic rocks, Coronation sills, Franklin dykes, Thule
dykes,andothercoevalsillsanddykes(Figure16). An age
of 723+3/-2 Ma (2(x; U-Pb baddeleyite)[Hearnanet al.,
1992] is basedon six sillsand a singledyke. Otherdykes
have been correlatedby K-Ar dating and (or) paleomag-
netism[e.g.,Fahrig et al., 1965;doriesand Fahrig, 1978;
Dawes, 1991]. Franklindykesextendedfor morethan 1500
km acrossthe Arctic islandsand the adjacentmainlandof
North America and northwesternGreenland(Figure 16)
[Robertson andBaragar,1972;Fahrig,1987;Nielsen,1987;
defferson et al., 1994;Dawes,1991;Ernstet al., 1995b].
Hearnanet al. [1992] and Rainbird [1993] proposedthat a
mantleplumelocatednorthof VictoriaIslandwasresponsi-
ble for this magmatismand they providedstratigraphic
evidencefor regionaluplift abovethe plumecenter.A more
preciselocationfor this plumecenteris providedby the
convergingdykepattern.
In general,the dykesassociated with this eventhave a
fanningdistribution(Figure 16). The densestsubswarms
trend southeasterly acrossBaff'mIsland and the adjacent
mainland of North America, and also acrossnorthwest
Greenlandafter restorationto its pre-driftposition[Rowley
and Lotres,1988]. Other,lessdensesubswarms includethe
southeast-trending dykesthatare associated with Natkusiak
volcanicrockson Victoria Island,and a few north-trending
dykesonthemainlandto thesouth.
Othersmallswarms,thoughtto be of similarage,do not
fit the simpleradiatingpattern;in particular,the south-
trendingdykeson northern Baff'mIslandandsouthwesterly-
trending dykeson Somerset Island[FahrigandWest,1986].
Theexplanation of thesediscordanttrendsisnotknown.
Sill complexes of Franklinageare abundant on Victoria
Islandandthemainlandof NorthAmerica(Figure16).
4.6. Willouran Event
The Wi!louranmagmaticeventis recognizedin southern
Australia,where the -800 Ma (Sm-Nd mineral isochron)
[ZhaoandMcCulloch,1993]Gairdnerdykeswarmhasbeen
linked to a mantle plume beneaththe coevalWillouran
volcanicprovince[Zhaoet al., 1994](Figure17).
Recently,Park et al. [1995] haveproposed
that780 Ma
ERNST AND BUCHAN 315
I
500 km
I
3000E
NORTH Hudson
AMERICA
oE say 290
270 oE
Figure 16. Dyke swarmsof the-723 Ma Franklin-Natkusiak
magmaticeventof North AmericaandGreenland.Greenlandis in
the reconstructedpositionof Rowley and Lottes [1988]. Other
Arctic islandsremain in their presentpositionrelativeto North
America. Franklin dykes are found on Baffin Island and the
adjacentmainland(B) and on SomersetIsland(A; Astondykes)
[Fahrig and West,1986], on VictoriaIsland(V) [Rainbirdet al.,
1994a,b] and on the North Americanmainland(M) [Fahrig and
147est,1986; Baragar and Donaldson, 1973]. Thule dykes of
northwestGreenlandare afterDawes [1991]. Sill complexes(s) of
North Americaarefrom Fahrig [ 1987]. ShadinglocatesNatkusiak
volcanicrocks.Starlocatesplumecenterbasedon convergence of
dykes.
mafic dykesand sheetsin three widely separatedareasof
westernNorth America (the northwesternCanadianShield,
the Mackenzie Mountains in the northernCordillera, and the
WyomingProvince)representsubswarms of the Willouran
event. Ages are basedon U-Pb baddeleyitedating from
LeCherninantand Hearnan [1994] and Harlan and Premoin
Park et al. [ 1995].
Paleomagnetic
evidence[Park et al., 1995] indicatesthat
the North American subswarms have not moved relative to
oneanothersinceemplacement andthattheirroughlyradial
patternis primary.Togetherwith the Gairdnerdykes of
Australia,Park et al. [1995] proposedthattheyform a giant
radiatingdyke swarmin a reconstruction of Australiaand
Laurentia [Borg and DePaolo, 1994]. The Willouran
volcanicprovinceis locatednearthe southeastern endof the
Gairdnerswarm [Zhao et al., 1994], near the focal area of
the giantradiatingswarmafter reconstruction. The various
subswarms extend between 1200 and 2000 km from the
focal point.Park et al. [1995] suggested
thatthe
316 GIANT RADIATING DYKE SWARMS, LIPS, AND MANTLE PLUMES

4.8. Abitibi and Keweenawan Events

M
Mafic magmaticeventsassociated
with the Mid-Continent
Rift of North Americaoccurovera spanof at least65 m.y.
(Figure 19), consistingof an older 1140 Ma Abitibi event
anda younger1110-1085 Ma Keweenawanevent.
The weakly fanning Abitibi dyke swarm [Ernst and

wLAURENTIA Buchan, 1993] was emplacedat 1141+1 Ma (2c•; U-Pb

1000 krn
Figure 17. Dyke swarmsof the -780 Ma Willouran magmatic
event in Australia and westernNorth America [Park et al., 1995
andreferences therein].Continental
reconstructionis that of Borg
and DePaolo [1994]. The 780 Ma dykes of the Mackenzie
baddeleyite) [Kroghet al., 1987].It consists
mainlyof a few
very wide (up to 200 m) dykes.The largestof theseis
termedthe GreatAbitibi dykeandcanbe tracedfor nearly
700 km. On the basis of magneticfabric studiesand
longitudinal compositional variations,thatdykewasproba-
bly emplacedlaterallyfrom a sourceareato the southwest
[Ernst, 1990; Ernst and Bell, 1992]. The Abitibi swarm
focusessouth of Lake Superiorin the vicinity of the
GoodmanSwell, an areaof resetRb-Srbiotiteageswhich
arethoughtto reflectlocaluplift at 1128+20Ma [Peterman
andSims,1988].Hutchinson et al. [1990]proposed thatthe
swell representsuplift associated with a mantle plume
responsible for the youngerKeweenawan activity,whereas

30øE
I 40øE,•
Mountains(M) and the Wyoming Province(W) and inclined
Hottah(H) sheetsin the CanadianShieldof westernNorth America
are comparedwith -800 Ma Gairdner(G) dykes of southem
Australia [Park et al., 1995]. The small star marks the mantle
plumecenterproposedby Zhao et al. [1994] on the basisof the
distributionof Willouranvolcanicrocks(shaded).The largestar Barents
--I
marksthe locationof the proposedplume centerdefinedon the Sea
basisof the convergingdyke pattern[Park et al., 1995]. Present
outline of easternNorth America(dottedpattern)is shownto
indicate the orientation of Laurentia.
/
event may have been a precursorto Late Proterozoic
breakup of a supercontinent
which incorporatedboth
Australia and Laurentia.

4.7. Kola-OnegaEvent _ WhiteSea Fl'

The Kola-Onega dyke swarm of Finland and Russia
[Berkovsky and Platunova,1987;Gorbatschev et al., 1987]
with a length of 500 km and a width of 600 km, radiates
'EUROPE -
fromthe northoveran angleof about50ø (Figure18).Mert-
aneneta/. [1996] havepublishedSm-Nd isochronagesof I I
300 km
1042+50, 1013+32, and 1066+34 Ma (2c•) for the Laanila,
Ristijfirvi,and Kautokeinodykes,respectively,
all of which i
belongto the westernportionof the swarm.
Figure18. Dykesof the-1000 Ma Kola-Onega magmatic
eventof
Convergence of the Kola-Onegaswarmsuggests a plume
northernEurope[Berkovsky and Platunova,1987;Gorbatschev
et
centerto the north.Coevaldiabasesillsalongthe northern
al., 1987].L, R, andK labeltheLaanila,Ristij'&rvi,
andKautokeino
coastof the Kola Peninsula[Sinitsyn,1963, citedby Berk- dykes,respectively,for whichSm-Nddateshavebeenreported
ovskyandPlatunova,1987andby Gorbatschev et al., 1987] [Mertanenet al., 1996].Starlocates
theconvergence
pointof the
may be part of a remnantLIP.
 ERNST AND BUCHAN 317

dykes southof the lake. The Mellen-Gogebicand Central

Wisconsinswarmsmay alsobe Keweenawanin age [Green
et al., 1987;King, 1990].

4.9. Mackenzie Event

One of the largestknown dyke swarmson Earth is the

Figure 19. Dyke swarms of the 1140-1085 Ma Abitibi and
Keweenawanmagmaticevent(s)of centralNorth America. GAD
and K label the Great Abitibi and Kipling dykes,respectively,of
the 1140 Ma Abitibi dyke swarm (distributionafter Ernst and
Mackenzie swarm of northwesternNorth America (Figure
20). It coversan areaof 2.7 x 106 km2, extendsmorethan
2600 km from the inferred plume center, and fans
dramaticallyover an angleof about 100ø [Fahrig and dones,
1969; Fahrig, 1987]. Coeval flood basalts,the Coppermine
River lavas,as well asthe layeredmafic/ultramaficMuskox
intrusion,are presentnearthe focusof the swarm,although
the focusitselfis hiddenbeneathyoungercoverrocks.High-
precisionU-Pb baddeleyitedatinghasdemonstrated that the
t • •

Buchan [1993]). The minor Eye Dashwa (ED) swarm is also

/" 2500
E 2800
E
thoughtto be about 1140 Ma [Osmani, 1991]. Star locatescenter 500 km
70øN
of Goodmanswell [Petermanand Sims,1988]. Keweenawandykes
dated as or assumedto be •-1100 Ma are markedby dottedlines
[Green et al., 1987] and includePukaskwa(P), ThunderBay (or MI
Pigeon River) (T), Ely-Moose (E), Carlton County (C), Baraga
(Marquette)(B), Mellen-Gogebic(M) andCentralWisconsin(CW)
subswarms. The arcmarksthe maximumextentof the radialpattern
in the Abitibi swarm and the arrows mark the inferred orientation of
the regional stressmaximum.The Mid-Continent(Keweenawan)
rift is shaded.

Hudson
Ernst et al. [1995a] suggestedthat the swell represents
plume-generateduplift which could also be synchronous Bay
with eraplacementof the Abitibi dyke swarm.At a distance s\xx 'N
of 600 km from the centerof the swell,the Kipling dyke of
the Abitibi swarm swingsparallelto the Great Abitibi and
othernearbyAbitibi dykes.This probablymarksthe distance
at which the influenceof the regionalstressfield exceeded NORTH
the plume-generated uplift stress(Figure 19). AMERICA
To the west, the minor Eye-Dashwa swarm [Osmani,
1991] has K-At (whole rock) agesof 11324-27and 11434-27
Ma (2(•) (samplesGSC87-55 and GSC87-56 [Hunt and
500 N
Roddick,1987]) anda virtualgeomagnetic pole(ourprelimi- / I
nary unpublisheddata) similarto that of the Abitibi swarm.
Hence,the Eye-Dashwadykesmay form a subswarmof the Figure 20. Dykes of the 1270-1265 Ma Mackenzie magmatic
Abitibi swarm. event of northern North America. Coeval Coppermine River
A secondperiod of activity occurred•-30-65 m.y. later. volcanic rocks are shaded.Star marks focal point of swarm. S
Voluminous Keweenawanvolcanismand sill emplacement markslocationof coevalsills[afterFahrig and West,1986;Hulbert
et al., 1993]. There are also other related sills in the eastern
were accompaniedby a numberof dyke swarms,generally
CanadianArctic islands(coveredby the scalebar) [Ernst et al.,
alignedparallelto the rift arms [Green et al., 1987]. They 1995a]. MI is Muskox Intrusion.The inner arc marksthe transition
includethe northwest-trending Pukaskwadykeseastof Lake from vertical flow to horizontal flow in the swarm based on
Superior,the northeast-trending Thunder Bay (or Pigeon magneticfabric studies[Ernst and Baragar, 1992]. The outer arc
River), Ely-MooseandCarltonCountydykesnorthwestand marksthe extent of the purely radial pattern.Arrow indicatesthe
west of the lake, and the east-trending
Baraga(Marquette) inferredorientationof the regionalstress
318 GIANT RADIATING DYKE SWARMS, LIPS, AND MANTLE PLUMES

Mackenziedyke swarm,the Muskox intrusion,and several

sillsfar fromthe focusregionwereemplacedwithina period
I I
of lessthan 5 million yearsbeginningat 1272 Ma, with all Lake of
100 km

the dated dykes being injected at 1267+2 Ma (2•) the Woods

[LeCherninantand Hearnan,1989, 1991]. The Coppermine \\ 49øN
River lavashave not beendatedpreciselybut are correlated
with the Mackenziedykesbasedon similar compositions,
paleomagnetism, K-Ar ages,and a decreasein dyke abun-
danceupwardsin the volcanicpile, consistent with at least
some of the dykes being feeders [Gibson et al., 1987;
Baragar et al., 1996].
According to LeCherninantand Heaman [1989] the
Mackenzieswarmwas initiatedby a mantleplumeimping-
ing on the lithosphere.Their conclusionis basedon the ..... '.'.'.'.'Lake

radiatingpatternof dykesandthe stratigraphic evidencefor

uplift in the focalregionpreceding
magmainjection,andit is
supported by magneticfabricdataindicatingverticalflow in ß o

the focal region and lateral flow beyond it [Ernst and

Baragar, 1992]. Breakupalong this northernboundaryof
Laurentiaat the time of dyke emplacement[LeCherninant
and Hearnan,1989] is thoughtto have openedthe Poseidon
Ocean[Fahrig, 1987].

4.10. Fort Frances Event

The northwest-trending Fort Francesdyke swarmto the
westof Lake Superiorhasa lengthof 300 km anda width of
400 kin (Figure 21). The distributionof the Fort Frances
dykes,basedin part on aeromagnetic interpretation
[Chan-
dler, 1991], suggestsa fan of about35ø [Buchanand Halls,
1990]. U-Pb baddeleyiteagesof 2077 +4/-3 and2076 +5/-4
Ma (2c•)were reportedby Wirthet al. [1995] andBuchanet
al. [1996], respectively.
Southwickand Day [ 1983] proposedthatthe Fort Frances
swarmwas relatedto a hotspotbeneaththe Paleoproterozoic
Animikie Basinto the southeast of the exposedswarm.They
also suggestedthat the swarmwas injectedto the northwest
along late Archeanfractureswhich were reactivatedas the
failed arm of a triple junction. The Fort Francesswarm
convergesto the southeastto a postulatedmantle plume
center,consistentwith the hotspotmodel of Southwickand
Day [1983]. A rift-relatedsequenceof rocks (Mille Lacs
Group) in the lower Animikie basinincludesvolcanicunits
that may be coevalwith the Fort Francesswarm [Southwick
and Day, 1983].
4.1 1. Matachewan Event
The Matachewandyke swarm occursover an area of
2.5 x 10-skm2 in thesouthern
andcentralSuperior
Province
I 9•TM
Figure 21. Dykes of the-2076 Ma Fort Frances (or Kenora-
Kabetogama)magmaticeventof centralNorth America(modified
from figure 2 of Chandler [1991]). Younger cover rocksof the
AnimikieBasinareshownin dottedpattern.Starandarrowindicate
directionto convergence
pointof swarm.
[e.g., Halls, 1991] (Figure 22). It is composedof threesub-
swarmsseparatedby areasof few dykes.After correctionfor
later deformation[West and Ernst, 1991; Bates and Halls,
1991; Bird et al. 1996] eachsubswarmradiatesfrom a focal
regioncenteredapproximatelyon easternLake Huron.
The age of the Matachewanswarmhasbeendefinedby
U-Pb baddeleyitedatesof 2446+3 and 2473 +16/-9 Ma (2•)
[Heaman, 1995] indicatingthat it was emplacedover a
minimum of 30 m.y. However, more data are neededto
establishwhetherthe inagmatismwas episodicor continuous
over this interval.
A number of coeval volcanic and intrusive units are found
at the southeastern end of the exposedswarm [Krogh et al.,
1984; Prevec et al., 1995], not far from the swarm focusand
presumedmantle plume center [Halls and Bates, 1990].
Direct evidencefor magmaflow directionis not seenin the
Matachewanswarm, but a preferentialdirectionof dyke-
splitting(bifurcation)impliesnorthwardinjectionof magma
away from the focal regionandplume[Halls, 1982;Bisson,
 ERNST AND BUCHAN 319

Mesozoic giant dyke swarms may be associatedwith

James
NORTH paleoplumes.However, in eachcase,dyke distribution,age,
Bay I I
or tectonicsettingis still speculative.
AMERICA lOO km
The Gannakouriep swarm of western South Africa
appearsto fan from a focusto the south(Figure24a). It can
be interpretedas (1) a radiatingswarmwith a fan angleof
25ø which in the north swingsinto a regional paleostress
field orientednorth-northeast [Ernst et al., 1995b] or (2) a
linear l•orth-northeast-trending swarm which has been
reoriented into a northward trend within the deformed
Gariep belt [Ransome,1992]. Gannakouriepdykesare dated
at 717+11 Ma (Rb-Sr mineralisochron)[ReM et al., 1991b].
They may be relatedto Pan-Africanrifting that developed
just prior to the onsetof Gariepsedimentation
[Gresseand
Scheepers,1993].
The north-trendingSalvador,northwest-trendingItacar•,
and west-trending11h•us-Olivenq;a-Camacgdykesappearto
Lake definea radiatingswarmwhichcoversan angleof about80ø
Superior

W
72øW
Lake
Huron
- 54øN • NORTH
AMERICA

Figure 22. Dyke swarms of the 2470-2450 Ma Matachewan

magmaticeventof centralNorth America [Fahrig and West,1986].
after correctionfor a subsequentdeformationevent [West and
Ernst, 1991]. Main subswarmsare labelled M1, M2 and M3.
Dotted lines are N-S dykes tracked on aeromagneticmaps and
identifiedas possiblyMatachewanin the region of-48øN, 78øE
basedon paleomagnetic results[Buchanet al., 1993]. Shadedareas
in focal region representcoeval volcanic and intrusive units
[Bennett et al., 1991; Prevec et al., 1995; S. Prevec, written
communication,1995].

i i
4.12. Mistassini Event
100 km

The Mistassinidyke swarmof easternNorth Americafans - 51øN

over an angle of 30ø (Figure 23). The focal point of the Lac
swarmand possibleplumecenteris locatedsoutheast of the Mistassini
swarm within the Grenville Province. No coeval volcanics
havebeenidentifiedin the focalregion.The swarmconsists
of boththoleiiticand komatiiticsuites[Fahrig et al., 1986].
The age of the swarm is about 2470 Ma basedon U-Pb
Figure 23. Dykes of the Mistassinimagmaticeventof northeastern
baddeleyite/zircondating[Hearnan,1994].
North America (based on Fahrig and West [1986] and a new
tracing from aeromagneticshadowgrammapsproducedfrom the
4.13. Additional Cases
National AeromagneticData Base by the GeologicalSurvey of
CanadaGeophysicalData Center).Starandarrowindicatedirection
There are several additional examplesin which pre- to convergencepoint of
320 GIANT RADIATING DYKE SWARMS, LIPS, AND MANTLE PLUMES
Figure 24. Additionalcases.(a) Dykesof the •-720 Ma Gannakou-
riep event [Reid et al. 1991b; Ransome, 1992; Gresse and
Scheepers, 1993]. Starlocatesconvergent pointof swarm.(b) Dyke
swarmsof the •-1000 Ma Coastline(Bahia) magmaticevent of
Brazil [Gomes et al., 1989; Correa-Gomeset al., 1991; Correa-
Gomes and Tanner de Oliveira, 1994a,b; Correa-Gomes,1995].
Componentsubswarms are labelledas follows:S is Salvador,I is
Itacard and C is Ilhdus-Oliven•a-Camacfi.Arrows indicate flow
directionsdeterminedfrom texturalobservations by Correa-Gomes
and Tanner de Oliveira [1994a,b] and Correa-Gomes[1995]. Star
(Figure24b). Flow lineationsandotherrheological markers
in the dykesof all threeareasindicatethatmagmaflow had
an inclinationof 35-25 ø upwardsfroma sourceto theeastor
southeast.Accordingto Correa-Gomeset al. [1991] and
Correa-Gomes[1995], these dykes are associatedwith a
failed attempt at continentalbreakup between South
Americaand Africa. Isotopicdatingyieldsan agerangeof point at the coeval Booth River intrusivesuite in the
150m.y. This may indicatea complicated Kilohigok basin [LeCheminant,1994]. In northernAust-
scenarioinvolving
more than one plume event. Two Salvadordykes give ralia, the McArthur Basin dyke swarm[Tuckerand Boyd,
differentages:the first, an ageof >924 Ma (2ø7pb-2ø6pb, 1987] of unknownage fansto the southover an angleof
2.6% discordant)[Heaman,1991], and the second,agesof about 50 ø
1021ñ8Ma and 1003ñ33(4øAr-39Ar
hostrock biotiteand
dyke plagioclase)[D'Agrella-Filhoet al., 1989]. The Ilhdus
andOlivengasubswarms havepaleomagnetic directions that
differby about20ø, perhapscorrelatingwith the reportedage
difference of 65ñ45 m.y. (11hdusdykes, 1011+24 and
1012ñ24 Ma (1cy), and Olivem;a dykes 1077+24 and
1078ñ18Ma (1cy);4øAr?Ar hostrock biotite and dyke
plagioclase)[D'Agrella-Filho et al., 1990; Renne et al.,
1990].
The 1238-1235 Ma (U-Pb baddeleyite)[Krogh et al.,
1987; Dud•tset al., 1994] Sudburydykesdefine a broad,
roughly linear swarm trendingnorthwestfor a distance
greaterthan200 km in the Superiorand Southernprovinces
of the CanadianShield(Figure24c). The generalextentof
the swarm has been definedprimarily by paleomagnetic
studies[e.g.,Palmeret al., 1977].In addition,someSudbury
dykescanbe tracedsoutheast intothe GrenvilleProvinceas
increasingly deformedbodiesfor distances of at least50 km
and perhapsas much as 100 km [Bethuneand Davidxon,
1988; Dud•tset al., 1994]. Magneticfabric studies[Ernst,
1994] reveal a consistenthorizontal flow fabric and a
consistent senseof imbricationsuggesting magmaflow from
the southeast.Therefore,the magmasource(plume?)was
likely locatedto the southeast,
althoughin the absenceof
swarmconvergence a morepreciselocationcannotbe deter-
mined.
It is suggested
(K.L. Buchanet al., Paleomagnetism
and
Geochronology of Dykes of the Ungava,manuscriptin
preparation)that threewidely separated Paleoproterozoic
dyke swarms in northeastern North America represent
remnantsof a major magmaticevent (Figure 24d). The
Klotz,MaguireandSenneterre dykeshaveU-Pb agesin the
range of 2230-2210 Ma ([Buchanet al., 1993]; K. L.
Buchanet al., ms. in preparation),and convergeover an
angleof-80 øtowardsa focalpointnearUngavaBay.
Other exampleswhich may representradiatingdyke
swarmsare found in Tanzania,northernCanada,and north-
ern Australia(Figure1). In Tanzania,threedykeswarmsof
under'reedage radiate from the southernend of Lake
Victoriaover an angleof about90ø. They havebeenrelated
to the nearbyBukobanvolcanicrocks[Halls et al., 1987]
which have a poorly definedage of 1200-800Ma. In the
Slave Province of the Canadian Shield, the 2030-2023 Ma
(U-Pb baddeleyite)[LeCheminant
and van Breemen,1994]
Lac de Gras swarm fans approximately10ø from a focal

5. IMPLICATIONS OF GIANT RADIATING DYKE
SWARMS FOR LIPS AND PLUMES
The previoussectionhasoutlinedevidencesuggesting
that
pre-Mesozoicgiant radiatingdyke swarmsare remnantsof
LIPs which, as a result of erosion,have lost most or all of
their volcaniccomponent.In this sectionwe discuss(1) the
relationship betweenLIP distributions anduplift topography
as determinedfrom radiatingswarms,(2) a methodto map
the boundaryof melt generationin the plume,(3) evidence
from dyke swarmssuggesting that someLIP eventsmay be
tens of millions of years in durationrather than the <5
million yearsconsidered typicalof post-Paleozoic LIPs, (4)
lateral feeding of sills distal from the plume, and (5) the
possibility that some continentalswarms are fed from
oceanic LIPs.
5.1. Correlationof Dyke Distribution,Uplift Topography,
and LIPs
Two dyke swarm parameterscan be relatedto plume-
generateduplift: the radial-lineartransitionin the swarm
patternandthe convergence pointof the dykes.
Firstly, the location of the transitionbetweenthe radial
and the more-distal,subparallelportionsof swarmsis con-
trolled by the balancebetweenthe stresses due to plume-
generateduplift and regionalforces[McKenzieet al., 1992;
Ernst et al., 1995b]. Therefore, this transition marks the
approximateouterboundaryof the topographic uplift at the
time of dyke intrusion.Modelingby Griffithsand Campbell
[ 1991] suggests that uplift topographyvariesduringplume
ascent,but that at the time of probablepeak magmaem-
placement(about 5 m.y. after maximum uplift), the outer
boundaryof uplift is similarto that of the plume head.In
actual cases,the uplift will be subjectto additionalfactors
suchas heterogeneity in the structureand lithologyof the
lithosphereandmaybe quitecomplicated. However,we will
takethe simplemodelasa startingpoint.In the casesof the
Mackenzieand Abitibi swarms(Figures19 and 20), the
transitionoccursat about1000km from the plumecenter,
consistentwithplumemodellingof Griffithsand Campbell
[1991].
Secondly,the convergence
pointof swarmscanbe usedto
locatethe positionof maximumuplift. We have already
discussedin a generalwayhowtheconvergentdykepattern
locatesa plume center.In more detail,the locationof the
focuscanbe calculatedrigorouslyfromthe convergence of
greatcircleswhichrepresentdykeswithinthe radialportion
of the swarm [Ernst et al., 1995b]. In the case of the
Mackenzieswarm,which coversa large portionof the
ERNST AND BUCHAN 321
CanadianShield,the calculatedconvergence point basedon
10 representative dykesdistributedevenlyacrossthe swarm
is 70øN, 244øE with a 95% confidenceuncertaintyof +2ø
[Ernst et al., 1995b]. Other proposedplume centersin this
paperhave not beencalculatedrigorously.
In general,most of the volcanicrocks associatedwith a
mantleplumeare foundnearthe centerof topographic uplift.
Examples include the Afar (Figure 3), North Atlantic
VolcanicProvince(Figure4), and Deccan(Figure5) events.
However, in some cases, the maximum accumulation of
volcanic rocks is offset hundreds of kilometers from the
centerof the topographicuplift. For example,the Columbia
River basaltgroup,which resultedfrom the arrival of the
Yellowstone plume, has a maximum accumulationof
volcanicrocksnearly500 kilometersfrom the plumecenter
on the edge of the uplift (Figure 2). Becausethis event is
only 17 m.y. old, the presentuplift probablyreflectsthe
paleo-uplifi.
An older exampleof volcanismoffsetfrom the centerof
topographicuplift involves the Abitibi and Keweenawan
events (Figure 19) at 1140 and 1110 to 1085 Ma. The
inferred center of uplift associatedwith these events is
locatedto the southof Lake Superior,in the vicinity of the
GoodmanSwell [Petermanand Sims, 1988], a region of
---1130Ma Rb-Sr biotiteagesresetduringuplift. Assuming
the swell as the plume center,we note that Keweenawan
rifling andmajorvolcanismis concentrated in an arc located
---200km away from the centerof the GoodmanSwell.
Two other examplesof LIPs whose exposedvolcanic
rocksappearto be offsetfrom the plume centerincludethe
Natkusiak volcanics of the 723 Ma Franklin event and the
CoppermineRiver volcanicrocksof the 1270 Ma Macken-
zie event. In both cases the volcanic rocks are located
approximately300 km southof the plumecenter(Figures16
and 20). However, for these older events,includingthe
Abitibi and Keweenawanevents(discussedabove), it is
difficult to be sure that erosionhas not preferentially
removedthe lavasonthetopographic uplift.
Thereare a numberof possibleexplanations for offsetof
volcanismfrom the plume center.Firstly, the observed
distributionof volcanicrocksis stronglyinfluenced by the
developmentof riffs, and suchaccumulations will also be
resistantto erosion. Secondly,the presenceof local
previouslythinnedlithospere,a "thinspot"in the terminol-
ogy of Thompson and Gibson[ 1991] may facilitateaccu-
mulationof volcanicrocksin theseareas.Thirdly, in the
modelof Baragaret al. [ 1996]magmachambers arelocated
in a ring associated with centralgrabencollapse(Figure
1l a) and are displacedfrom the plumecenterby a distance
of several hundred kilometers. Volcanism from such
322 GIANT RADIATING DYKE SWARMS, LIPS, AND MANTLE
berswould alsobe offsetfrom the plumecenter,providing
a possibleexplanationfor the observedoffset.Finally, giant
circumferential dyke swarms (R. E. Ernst and K. L.
Buchan, ms. in preparation)may feed lava flows offset
from the plumecenterby hundredsof kilometers.
5.2. Location of Boundaryof Plume
The pattern of magma flow in giant radiating dyke
swarms can be related to melt generationin the plume
head. In particular,the transitionfrom verticalto horizontal
flow with increasingdistancefrom the focus(seemagnetic
fabric studiesof section3.2) is thoughtto mark the outer
boundaryof melt generationin the underlyingmantleplume
[Ernstand Baragar, 1992] and may reflectthe penetration
of only the central portion of the plume to the depths
conduciveto voluminousmelt generation.
In only three swarms--the Mackenzie, Botswana, and
the Poma Grossa--has a transition between vertical flow
andhorizontalflow beendeterminedfrom magneticfabric
studies. In the Mackenzie swarm (Figure 20), the
transition occurs within 500 km of the focus [Ernst and
Baragar, 1992]. In the case of the Botswana swarm
(Figure 8), predominantvertical flow is indicatedat the
easternend of the swarm, with lateral flow predominating
at, and presumablybeyond, a distanceof-•400 km west
of the plume center [Ernst and Duncan, 1995]. A
magneticfabric studyof the Poma Grossaswarm (Figure
7) showsdykeswith steepflow giving way to dykeswith
horizontal flow with increasingdistancefrom the plume
center [Raposoand Ernesto, 1995]. The transitionoccurs
at least200 km from the plume centerdisplayedin Figure
7.
As the radiusof the plume head after arrival and flatten-
ing againstthe lithosphere(-1000 km [e.g., White and
McKenzie, 1989]) is likely to be much larger than our
estimatesof this vertical to horizontaldyke-flow transition
boundary(_<500)km, we may speculateon what additional
factor(s)is impedingshallowplume penetrationand associ-
atedmelt generationbeyondthe transitiondistance.Helms-
taedt and Gurney[1994] suggested that the penetrationof
the plume to a shallow level can be limited by a thick
refractory lithosphericmantle root underlying adjacent
Archean crustal areas. Furthermore, because thick litho-
sphericroots appearnecessaryfor kimberlitesto be dia-
mondiferous,mapping of the boundarybetween shallow
parts of plumes and adjacentmantle roots using magma
flow trajectoriesin dykescouldhelp predictthe distribution
of diamondiferousvs. barren kimberlites [Helmstaedtand
Gurney, 1994; LeCheminantet al. , 1996].
PLUMES
5.3. Duration of LIP Events
Many plume-relatedLIPs of Mesozoicand Cenozoicage
were emplacedrapidly [see summariesof Storey, 1995;
White and McKenzie, 1995]. As noted in earlier sections,
most of the volcanism in the Deccan, Siberian, Karoo,
ColumbiaRiver, andParantiLIPs occurredin time spansof
lessthan a few m.y. In othercases,LIPs form over much
longerperiods.For example,the Icelandicplateauformed
over more than 20 m.y. However, this is an exampleof a
plumetail capturedby a slow-spreading
ridge[Oskarsson et
al., 1985].
The identificationof pre-Mesozoicpaleoplumesand
remnant LIPs from convergentdyke swarms provides
additionalexamplesfor which timing of LIP emplacement
can be assessed.Unfortunately,only a few of the pre-
Mesozoic LIPs (Table 1) have a sufficient number of
precisedatesto allow estimatesof the time spanof em-
placement.
Some pre-Mesozoiceventsdo appearto have occurred
rapidly. For example,basedon highprecisionU-Pb dating,
both the Mackenzie and Franklin events occurred in less
than5 m.y. at 1267and723 Ma, respectively.On theother
hand, other swarms seem to exhibit a much longer em-
placementhistory. The Matachewandykes and associated
mafic plutonsyield agesbetweenabout2490 and 2445 Ma.
However, more data are requiredto determinewhetherthe
magmatismwas episodicor continuousover this interval.
The Central Iapetusevent spansat least25 m.y. basedon
U-Pb agesfor the 615 Ma Long Rangedykesand the 590
Ma Grenville dykes. The -1140 Ma Abitibi and -1100
Keweenawan dykes may representanother example of
repeatedmagmaticactivityassociated with a singleplume.
A possibleexplanationfor multipleplume 'impacts'from
one plume has been offered by Bercoviciand Mahoney
[1994]. They suggested that separationof the plume head
from its tail at the 660-km-deepboundarycould result in
accumulationat that boundaryof a secondplume head. It
shouldbe notedthatthe secondplumeheadwill only appear
in the sameplace at the surfaceif the plate has not moved
substantiallyin the interval.
5.4. RemoteFeedingof Sills
In their mappingand modelingstudyof Early Tertiary
volcaniccentersin westernNorth America, Hyndmanand
Alt [1987] madea convincingcasethat dykesof a radiating
swarm may change orientationalong strike and become
laccoliths. Ernst et al. [1995a] summarized evidence
supportinga similar origin for somesill complexes

iatedwith giantradiatingdyke swarms,basedon the radiat-
ing patternsof theseswarmsand evidencefor lateral flow.
Three examplesare givenbelow.
Associatedwith the Mackenzie swarm (Figure 20) are
coeval sills within (or near) older sedimentarybasinsat
distancesof 800, 1000, 1400, and 1500 km from the pro-
posedplume center[Ernstet al., 1995a]. Given that Mac-
kenzie dykes are thoughtto have been fed throughlateral
injectionbeyond500 km [Ernst and Baragar, 1992], it is
unlikely that thesesillswere fed vertically, as verticalflow
would require the existenceof severalvery widely separ-
ated magma sourcesof identicalage. Instead, it is more
probablethat thesesills were fed from laterally emplaced
Mackenzie dykes, some of which changedorientation,
becomingsillswithin (or near) sedimentarybasins[Fahrig,
1987; Ernst et al., 1995a]. This reorientationof dykesinto
sills requiresa modificationof the stressconditionsby a
horizontalplane of weakness,or by the local load due to
sediment accumulation.
Other examplesin which coevalsillsare locatedfar from
plume centersincludethe Franklin and Parantievents.In
the caseof the Franklinevent(Figure 16), sillsat a distance
of about 800 km from the probableplume centermay be
fed in a similar fashion to the Mackenzie sills described
above. Likewise, White [1992] proposedthat Parantisills
within the sedimentsof the Parantibasinwere fed laterally
throughdykes"rootedin the rifted areafar to the east."
5.5. Implicationsfor OceanicLIPS
An importantclassof LIPs includesthe oceanicplateaus,
which may be due to mantleplumeheadsimpingingon the
baseof the oceaniclithosphere.Some of the largestexam-
ples are the Ontong Java and Kerguelenplateaus[e.g.,
Coffin and Eldholm, 1994, and referencestherein]which
may mark the arrival of mid-Cretaceoussuperplumes
[Larson, 1991].
Justas giant radiatingdyke swarmsare generatedwhen
plumesgive rise to continentalflood basalts,they might
also be expectedto occur when plumesproduceoceanic
plateaus[Ernst et al., 1995a]. On Venus, where giant
radiatingdyke swarmscut basalticrocks, the swarmsare
observedby their surfacegrabens[McKenzieet al., 1992;
Grosfilsand Head, 1994;Ernstet al., 1995a].By analogy,
giantdyke swarmson Earth shouldradiatethroughoceanic
crust away from oceanic mantle plumes. Many such
swarmsmay be concentrated parallelto spreadingridgesas
plume heads in ocean basins are likely to "capture"
spreadingridgeseven from a considerable distanceaway
[Mahoneyand Spencer,1991]. Dykes may be difficult to
ERNST AND BUCHAN 323
observein the oceanbasins,either directly or usinggeo-
physicaltechniques,becauseindividualdykesare narrow,
intrudehostrock with a similarbasalticcompositionto that
of the dykes themselves,and are typically coveredby
sediments.Whether suchswarmswould be more readily
observedwhere they have propagatedinto adjacentconti-
nentalcrust is unclear,becausetheir mode and depthof
emplacementupon reachingthe continentalcrust have not
been studied. However, if continental swarms associated
with distaloceanicplumesexist, a completeradiatingdyke
pattern would not be observedeven after reconstructing
plate positionsto their configurationat the time of oceanic
plumeemplacement.As yet, thereare no knownexamples
of giant dyke swarmsin oceanbasinsor on adjacentconti-
nentswhichcanbe linkedto oceanicplumes.
6. DISCUSSION
Giant dyke swarmsare importantin extendingthe record
of mantle plume-relatedevents and LIPs through the
Paleozoic and Proterozoic (Table 1). In order to assess
plume productionthroughtime, we will consideronly the
magmatic events in continentalareas (continentalflood
basalts)becausethe pre-MesozoicoceanicLIP record has
beenremovedby subduction.
Most continental flood basalt events that have been
identifiedto date [e.g., Coffin and Eldholm, 1994] are of
Mesozoic and Cenozoic age (Figure 25a). As described
above, severalof theseincludecoevalgiant radiatingdyke
swarmswhose fanningpatternslocate approximatecenters
of associated mantleplumes.However, giantradiatingdyke
swarms are also found throughoutthe Paleozoic and
Proterozoic(Figure 25b), over an intervalof time that is an
order of magnitudelonger than that representedby the
Mesozoic and Cenozoic. As discussedin the paper, they
were likely generatedby mantleplumes.
A great number(more than 100) other giant swarmsdo
not show an obviousradiatingpattern [Ernst et al., 1996].
Fifty-six of thesewhich are datedare shownin Figure25c.
Many of these giant swarms may representportionsof
radiatingdyke swarms.If so, they are also likely to be
relatedto mantleplumesandpaleocontinental floodbasalts.
When all datedgiant dyke swarmsand known continental
floodbasalteventsare combined(Figure25d) they indicate
a nearly continuous recordof major magmaticeventsfrom
the earliest Proterozoicto the present.Thus, in order to
fully appreciateand understand the role of mantleplumes
and continentalflood basaltsit is imperativethat the entire
record of sucheventsthroughgeologicaltime be investi-
gated.Giant radiatingswarmsprovidethe most
324 GIANT RADIATING DYKE SWARMS, LIPS, AND MANTLE PLUMES

2.5 2.0 1.5 1.0 0.5 0.0

I I I I I I I I I I I I I I I I I I I I I I I I I

(a) CFBevents

I IIII
(b) Giant
radiating
swarms
notin(a)

- II I I II I II i II I II I I
(C) Giant
swarms
notin(a)or(13)
I I I I I I
I I I I I I
:3 1
Z
- '' '
' IIIIII ' I'1 '1
I1' ' I 'I '"'
"11111 ' II I IIII
(d) Composite
of(a),(b)and(c)
-- I I I I I
I I I I I
I I I I II I

I I I I I I I I I I I I I I I I I I I I I I I I I
2.5 2.0 1.5 1.0 0.5 0.0

Age (Ga)
Figure25. Bargraphshowing
theagedistribution
of events
potentially
associated
withthearrivalof a plumeheadat the
baseof thecontinental
lithosphere.
(a)continental
floodbasalt
eventsafterCoffinandEldholm [1994],(b)giantradiating
dykeswarms exclusiveof thoseassociated
withthecontinental
floodbasalteventsof (a) (thispaper),(c) giantdyke
swarms notobservedto radiate(fromErnstet al. [1996]),and(d) composite
of (a), (b) and(c). Solidbarusedfor events
datedto within100m.y.Dashed barindicates
ageuncertaintiesbetween100and250m.y.Eventswithageuncertainties
greaterthan250 m.y. arenot shown.

record of these events. Canadareviewof thismanuscript

by BobBaragarwasappreciated.
Thoroughreviewsby Kevin Burke,Mike Wingate,andvolumeco-
Acknowledgments. We would like to thankBob Baragar,Bill editorJohnMahoneywere particularlyhelpful,especiallyin the
Davis, Henry Halls, Steve Kumarapeli,Rob Rainbird, Tony younger LIP examples.This is Geological Survey of Canada
LeCheminant, and Marie-Claude Williamson for fruitful dis- contribution # 37095 and Canadian LITHOPROBE # 787.
cussions
regarding
therelationship
of dykeswarms
to mantleplume
events. In addition, Rob Rainbird and Garth Jacksonare thanked REFERENCES
for assistancein distinguishingthe dykes associatedwith the
Franklin-Natkusiakevent, Andr6 Ciesielskiand the Geological Aleinikoff, J. N., R. E. Zartman, M. Walter, D. W. Rankin, P. T.
Surveyof CanadaGeophysicalData Centerfor informationon the Lyttle, and W. C. Burton,U-Pb agesof metarhyolites of the
Mistassini swarm, Steve Prevec for information on the volcanic Catoctin and Mount Rogersformations,central and southern
rocksandplutonsassociated
with the Matachewanswarm,andKarl Appalachians:evidencefor two pulsesof Iapetanrifting,Am. d.
Wirth for information on the Fort Frances swarm. Colin Reeves Sci., 295, 428-454, 1995.
provideda copyof thethesisby M. Mubu with somekey detailson Anderson,D. L., Lithosphere,
asthenosphere,
andperisphere,
Rev.
dykes of the Karoo events.An internalGeologicalSurveyof Geophys.,3, 125-149,

Andr6asson,P. G., The Baltoscandianmargin in Neoproterozoic-
early Palaeozoictimes. Some constraintson terranederivation
and accretion in the Arctic Scandinavian Caledonides,
Tectonophysics,
231, 1-32, 1994.
Atkinson, S.S., and R. St. J. Lambert, The Roza Member feeder
dyke system, Columbia River basalt group, USA:
Compositionalvariationand emplacement,in Mafic Dykesand
EraplacementMechanisms,edited by A. J. Parker, P. C.
Rickwood,and D. H. Tucker,pp. 447-459, Balkema,Rotterdam,
1990.
Auden, J. B., Dykes in western India: A discussionof their
relationshipswith the Deccantraps, Trans. National Inst. Sci.
India, 3, 123-157, 1949.
AustinJr.,J. A., P. L. Stoffa,J. D. Phillips,J. Oh, D. S. Sawyer,G.
M. Purdy, E. Reiter, and J. Makris, Crustal structureof the
Southeast Georgia embayment-Carolinatrough: Preliminary
resultsof a compositeseismicimageof a continentalsuture(?)
and a volcanicpassivemargin,Geology,18, 1023-1027,1990.
Baker, J., L. Snee,and M. Menzies,A brief Oligoceneperiodof
flood volcanismin Yemen: implicationsfor the durationandrate
of continental flood volcanism at the Afro-Arabian triple
junction,Earth Planet.Sci.Lett., 138,39-55, 1996.
Baksi, A. K., Geochronologicalstudieson whole-rock basalts,
Deccan Traps, India: evaluationof the timing of volcanism
relativeto the K-T boundary,Earth Planet.Sci. Lett., 121, 43-56,
1994.
Baksi,A. K., andE. Fartar,n,Ar?,•Ar
datingof the SiberianTraps,
USSR: Evaluationof the agesof the two majorextinctionevents
relativeto episodesof flood-basaltvolcanismin the USSR and
the DeccanTraps,India, Geology,19, 461-464, 1991.
Baldridge,W. S., Y. Eyal, Y. Bartov, G. Steinitz,and M. Eyal,
Miocene magmatismof Sinai relatedto the openingof the Red
Sea,Tectonophysics, 197, 181-201, 1991.
Baragar,W. R. A., and J. A. Donaldson,Coppermineand Dismal
lakesmap-areas.Geol.Surv.Can. Paper 71-39, 1973.
Baragar, W. R. A., R. E. Ernst, L. Hulbert, and T. Peterson,
Longitudinalpetrochemicalvariation in the Mackenzie dyke
swarm, NorthwesternCanadian Shield, d..Petrol., 37, 317-359,
1996.
Bates,M.P., and H. C. Halls, Broad-scaleProterozoicdeformation
of the centralSuperiorProvincerevealedby paleomagnetism
of
the 2.45 Ga Matachewandyke swarm,Can. d. Earth Sci., 28,
1780-1796, 1991.
Bennett, G., B. O. Dressier, and J. A. Robertson, Huronian
supergroupand associatedintrusive rocks, in Geology of
Ontario, Spec. Vol. 4, Part I, pp. 549-591, OntarioGeological
Survey,Toronto,ON, 1991.
Bercovici,D., and J. Mahoney,Doubleflood basaltsand plume
headseparationat the 660-kilometerdiscontinuity,
Science,266,
1367-1369, 1994.
Bertrand,H., The Mesozoictholeiiticprovinceof northwestAfrica:
a volcano-tectonicrecord of the early opening of Central
Atlantic,in Magmatismin Extensional StructuralSettings:The
Phanerozoic AJFicanPlate, editedby A. B. KampunzuandR. T.
Lubala,pp. 148-188,Springer-Verlag, 1991.
Berkovsky,A. N., and A. P. Platunova,Dyke swarmsof the East
European craton: Aeromagneticand geological evidence, in
ERNST AND BUCHAN 325
Mafic DykeSwarms,Spec.Pap. 34, editedby H. C. Halls andW.
F. Fahrig,pp. 373-377, GeologicalAssociationof Canada,1987.
Bethune,K. M., andA. Davidson,Diabasedykesandthe Grenville
Front southwestof Sudbury,Ontario,in CurrentResearch,Part
C, CanadianShield,88-IC, pp. 151-159, GeologicalSurveyof
Canada,Ottawa, ON, 1988.
Bhattacharji,S., Propagatingmafic dike swarmsin the Deccan
volcanics,hot spottrack and interplaterifting, in International
Symposium on Mafic Dykesand RelatedMagmatismin Rifling
and Intraplate Environmentswith Workshopon Mafic Dyke
Magmatismin theBalticShield,August8-13, ! 988, p. 11, IGCP-
257 TechnicalReportNumberOne, Instituteof Geology,Lund
University,Sweden,1988.
Bhattacharji,S., N. Chatterjee,J. M. Wampler,and M. Gazi, Mafic
dikesin Deccanvolcanics- Indicatorof India intraplaterifting,
crustal extension and Deccan flood basalt volcanism, in
Volcanism:RadhakrishnaVolume,editedby K. V. Subbarao, pp.
253-276, Wiley EasternLimited, 1994.
Bird, R. T., W. R. Roest, M. Pilkington,R. E. Ernst, and K. L.
Buchan, The application of digital geophysicaldata to the
restoration of crustal deformation in the Canadian Shield, in
CurrentResearch,Part C, CanadianShield,Pap. 96C, pp. 117-
124, GeologicalSurveyof Canada,Ottawa,ON, 1996.
Bisson,J. R., Quantitativeanalysisof width, spacing,trend and
bifurcationdirection of 2.6 Ga dykes in selectedareas of the
Central Superior Province. B.Sc. Thesis, 86 pp., Carleton
University,Ottawa, 1985.
Blagovyeshchenskaya, M. N., Geologicalmap of Siberianplatform
and adjoining territories, scale !:1,500,000, Ministry of
Geologyof the USSR, 1973.
Borg, S. G., andD. J. DePaolo,Laurentia,Australia,andAntarctica
as a Late Proterozoicsupercontinent: Constraintsfrom isotopic
mapping,Geology,22, 307-310, 1994.
Buchan.K. L., and H. C. Halls, Paleomagnetism of Proterozoic
mafic dyke swarmsof the Canadianshield,in Mafic Dykesand
EraplacementMechanisms,edited by A. J. Parker, P. C.
Rickwood,andD. H. Tucker,pp. 209-230, Balkema,Rotterdam,
1990.
Buchan,K. L., J. K. Mortensen,andK. D. Card,Northeast-trending
EarlyProterozoic dykesof southernSuperiorProvince:Multiple
episodes of emplacement recognized from integrated
paleomagnetismand U-Pb geochronology,Can. d. Earth Sci.,
30, 1286-1296, 1993.
Buchan,K. L., H. C. Halls, and J. K. Mortensen,Paleomagnetism,
U-Pb geochronology,and geochemistryof Marathon dykes,
SuperiorProvinceandcomparison with the Fort Francesswarm.
Can. d. Earth Sci., 33, 1583-1595, 1996.
Burke,K., and J. F. Dewey, Plume-generated triplejunctions:key
indicatorsin applyingplate tectonicsto old rocks,d. Geol., 81,
406-433, 1973.
Camp, V. E., and M. J. Roobol, The Arabian continentalalkali
basaltprovince:Part I. evolutionof HarratRahat,Kingdomof
SaudiArabia, Geol. Soc.Am. Bull., 101, 71-95, 1989.
Camp,V. E., and M. J. Roobol,Upwellingasthenosphere beneath
westernArabia and its regionalimplications, d. Geophys.Res.,
97, 15,255-15,271, 1992.
Chadwick,Jr.,W. W., andJ. H. Dieterich,Mechanical
modeling
326 GIANT RADIATING DYKE SWARMS, LIPS, AND MANTLE PLUMES
circumferentialand radial dike intrusion on Galapagos
volcanoes,J. Volcanol.Geotherm.Res.,66, 37-52, 1995.
Chandler,V. W., Aeromagneticanomalymapof Minnesota,scale
1:500,000, StateMap SeriesMap S-17, MinnesotaGeological
Survey,St. Paul,MN, 1991.
Choudhuri,A., E. P. Oliveira,and A. N. Sial, Mesozoicdyke
swarmsin northernGuianaand northernBrazil and the Cape
Verde- Fernandode Noronhaplume vortices:a synthesis, in
ExtendedAbstracts for the InternationalSymposiumon Mafic
Dykes,compiledby W. Teixeira,M. Emesto,andE. P. Oliveira,
pp. 17-22, S•o Paulo,Brazil, 1991.
Coffin, M. F., and O. Eldholm,Largeigneousprovinces:
crustal
structure,dimensions, andexternalconsequences.
Rev.Geophys.,
32, 1-36, 1994.
Coish,R. A., andC. W. Sinton,Geochemistry of maficdikesin the
Adirondackmountains: implicationsfor the constitution
of Late
Proterozoic mantle. Contrib. Mineral. Petrol., 110, 500-514,
1992.
Coleman, R. G., Geologic Evolution of the Red Sea, Oxford
UniversityPress,New York, 186pp., 1993.
Correa-Gomes, L. C., The maficdykeswarmalongthe coastlineof
Bahia State,Brazil:An attemptof continental
breakupbetween
SouthAmericaand Africa 1.0 Ga ago?(abstract)Program&
Abstractsfor theThirdInternational DykeConjkrence,
Sept.4-8,
1995,Jerusalem,Israel,p. 19, 1995.
Correa-Gomes, L. C., and M. A. F. Tannerde Oliveira,Map of
mafic dyke provinces, Bahia, Brazil, scale 1:1000000,
SICT/SME/UFBA/PPPG, 1994a.
Correa-Gomes,L. C., and M. A. F. Tannerde Oliveira,Map of
Bahia state,Brazil: major provinces,temporalevolutionand
present knowledge. Some evidencesabout upper mantle
behavior. International Symposiumon the Physics and
Chemistryof the UpperMantle,August14-19, 1994 S•o Paulo,
Brazil, 1994b.
Correa-Gomes, L. C., M. A. F. Tannerde Oliveira,H. Concieqao,
and M. B. Abram, Tectonicstylesand chemistryof the mafic
dykesin the easternpartof SaoFranciscocraton,Bahia,Brazil,
in Extended Abstracts for theInternationalSymposium onMafic
Dykes,compiledby W. Teixeira,M. Emesto,andE. P. Oliveira,
pp. 66-70, SaoPaulo,Brazil, 1991.
Cox, K. G., Tectonicsandvulcanism of theKarrooperiodandtheir
bearing on the postulatedfragmentationof Gondwanaland,in
Aj•ican Magmatismand Tectonics,editedby T. N. Clifford and
I. G. Gass,pp. 211-235,Oliver & Boyd,Edinburgh,1970.
Cox, K. G., The role of mantle plumes in the developmentof
continentaldrainagepatterns,Nature,342, 873-876, 1989.
D'Agrella-Filho,M. S., I. G. Pacca,T. C. Onstott,P. R. Renne,and
W. Teixeira,Paleomagnetism andgeochronology of maficdikes
from the regionsof Salvador,Olivenqaand Uaufi, SaoFrancisco
craton,Brazil: presentstageof the USP/PrincetonUniversity
collaboration.
BoletimIG-USP SdrieCientifica20, 1-8, 1989.
D'Agrella-Filho,M. S., I. G. Pacca,P. R. Renne,T. C. Onstott,and
W. Teixeira, Paleomagnetism of Middle Proterozoic(1.01 to
1.08 Ga) maficdykesin southeastern BahiaState- S•,oFrancisco
Craton,Brazil, Earth Planet. Sci. Lett., 101,332-348, 1990.
Dawes,P. R., GeologicalMap of GreenlandSheet5, Thule,scale
1:500,000,GeologicalSurveyof Greenland.,1991.
de Boer,J. Z., J. G. McHone,J. H. Puffer,P. C. Ragland,
andD.
Whittington, Mesozoic and Cenozoic magmatism,in The
Geologyof North America, Volume1-2, TheAtlanticContinental
Margin,U.S.,editedby R. E. Sheridan
andJ. A. Grow,pp.217-
241, GeologicalSocietyof America,Boulder,CO, 1988.
Deshmukh, S.S., andM. N. Sehgal,Mafic dykeswarmsin Deccan
volcanic provinceof Madhya Pradeshand Maharashtra,in
DeccanFloodBasalts, Mere.10, editedby K. V. Subbarao, pp.
323-340, GeologicalSocietyof India, 1988.
Dessai,A. G., andH. Bertrand, The "PanvelFlexure"alongthe
WesternIndiancontinental margin:an extensional
faultstructure
relatedto Deccanmagmatism,Tectonophysics, 241, 165-178,
1995.
Devey, C. W., and W. E. Stephens,Tholeiiticdykes in the
Seychelles
andtheoriginalspatialextentof theDeccan,d. Geol.
$oc. London, 148, 979-983, 1991.
Dickin,A. P., TheNorthAtlanticTertiaryProvince,in Continental
FloodBasalts,editedby J. D. Macdougall,pp. 111149,Kluwer
Academic Publishers,1988.
Druecker,M.D., andS. P. GayJr.,Mafic dykeswarmsassociated
with Mesozoicrifting in easternParaguay,SouthAmerica,in
MaficDykeSwarms,Spec.Pap.34, editedby H. C. HallsandW.
F. Fahrig, pp. 187-193, GeologicalAssociationof Canada,
Toronto, ON, 1987.
Dudhs,
F. 0., A. Davidson,
andK.M. Bethune,
AgeoftheSudbury
diabasedykesandtheirmetamorphismin theGrenvilleProvince,
Ontario,in RadiogenicAge and IsotopicStudies:Report8,
CurrentResearch1994-F, pp. 97-106, GeologicalSurveyof
Canada,Ottawa, ON, 1994.
Duncan,R. A., andM. A. Richards,
Hotspots, mantleplumes,flood
basalts,
andtruepolarwander,Rev.Geophys., 29, 31-50, 1991.
Duncan,A. R., R. A. Armstrong,A. J. Erlank,J. S. Marsh,andR.
T. Watkins, MORB-related dolerites associatedwith the final
phasesof Karoo flood basaltvolcanismin southernAfrica, in
Mafic Dykesand Eraplacement Mechanisms, editedby A. J.
Parker, P. C. Rickwood,and D. H. Tucker, pp. 119-129,
Balkema, Rotterdam,1990.
Dunham,A. C., and V. E. H. Strasser-King, Late Carboniferous
intrusionsof northernBritain,in IgneousRocksof the British
Isles,editedby D. S. Sutherland, pp. 277-283,JohnWiley and
Sons,Ltd., 1982.
Dunning,G., andJ.P. Hodych,U/Pb zirconandbaddeleyite ages
for the PalisadesandGettysburgsillsof the northeastern
United
States:Implications
fortheageof theTriassic/Jurassic
boundary,
Geology,18, 795-798, 1990.
Eales,H. V., J. S. Marsh,and K. G. Cox, The Karooigneous
province:An introduction,
in Petrogenesis
of the VolcanicRocks
of theKarooProvince,Spec.Publ.13,editedby A. J.Erlank,pp.
1-26, The GeologicalSocietyof SouthAfrica, Johannesburg,
1984.
Embry, A. F., and K. G. Osadetz,Stratigraphyand tectonic
significance
of Cretaceous
volcanismin QueenElizabethIslands,
CanadianArcticArchipelago,Can.J. Earth Sci.,25, 1209-1219,
1988.
Encamaci6n,J., T. H. Fleming,D. H. Elliot, and H. V. Eales,
Synchronousemplacement of FerrarandKaroodoleritesandthe
earlybreakupof Gondwana,Geology, 24, 535-538,

Erinchek, Ju. M., E. D. Milshtain, and R. S. Kontorovich, The
structure of the Middle Paleozoic Viluy-Markha dyke belt
interpreted from aeromagnetic data (Siberian Plate,
Russia)(abstract),Program & Abstracts for the Third
International Dyke ConJ•rence,Sept. 4-8, 1995, Jerusalem,
Israel, editedby A. AgnonandG. Baer,pp. 27, 1995.
Ernst,R. E., Magma flow directionsin two mafic Proterozoicdyke
swarmsof the CanadianShield: As estimatedusinganisotropy
of magneticsusceptibility
data,in Mafic DykesandEraplacement
Mechanisms,editedby A. J. Parker,P. C. Rickwood,and D. H.
Tucker,pp. 231-235, Balkema,Rotterdam,1990.
Ernst,R. E., Mappingthe magmaflow patternin the Sudburydyke
swarm using magnetic fabric analysis,in Current Research
1994-E, pp. 183-192, GeologicalSurvey of Canada,Ottawa,
ON, 1994.
Ernst,R. E., andW. R. A. Baragar,Evidencefrom magneticfabric
for the flow patternof magmain the Mackenziegiant radiating
dyke swarm,Nature, 356, 511-513, 1992.
Ernst, R. E., and K. Bell, Petrologyof the Great Abitibi dyke,
SuperiorProvince,d..Petrology,33,423-469, 1992.
Ernst,R. E., andK. L. Buchan,Paleomagnetism of theAbitibi dyke
swarm, southernSuperiorProvince, and implicationsfor the
LoganLoop, Can. d. Earth Sci.,30, 1886-1897,1993.
Ernst,R. E., andA. R. Duncan,Magma flow in the giantBotswana
dykeswarmfrom analysisof magneticfabric(abstract), Program
& Abstractsfor the Third InternationalDyke Conference,Sept.
4-8, 1995, derusalem,Israel, editedby A. Agnon and G. Baer,
pp. 30, 1995.
Emst, R. E., J. W. Head, E. Parfitt, E. Grosfils, and L. Wilson,
Giant radiatingdyke swarmson Earth and Venus, Earth Sci.
Rev., 39, 1-58, I995a.
Ernst,R. E., K. L. Buchan,and H. C. Palmer,Giant dyke swarms:
characteristics,distribution and geotectonicapplications,in
Physics and Chemistryof Dykes, edited by G. Baer and A.
Heimann,Balkema,Rotterdam,pp. 3-21, 1995b.
Ernst,R. E., A. G. Lindsey,and J. Z. de Boer, Flow patternin the
early Jurassic Higganum (Conn./Mass.) and contemporary
ChristmasCove (Maine) dikes using magneticfabric analysis
(abstract),1995 Abstractswith Programs, GSA Northeastern
Section,pp. 42, 1995c.
Ernst, R. E., K. L. Buchan,T. D. West, and H. C. Palmer, Diabase
(dolerite)dyke swarmsof the world: first edition, 1:35,000,000
map, 104 pp., Geol.Surv.Can. OpenFile 3241, 1996.
Eyal, Y., and M. Eyal, Mafic dyke swarmsin the Arabian-Nubian
Shield, Isr. d. Earth Sci., 36, 195-211, 1987.
Fahrig, W. F., The tectonicsettingsof continentalmafic dyke
swarms: Failed arm and early passivemargin in Mafic Dyke
Swarms,Spec.Pap. 34, editedby H. C. Halls andW. F. Fahrig,
pp. 331-348, GeologicalAssociationof Canada,Toronto,ON,
1987.
Fahrig, W. F., and D. L. Jones,Paleomagneticevidencefor the
extent of Mackenzie Igneous Events, Can. d. Earth Sci., 6,
679-688, 1969.
Fahrig, W. F., and T. D. West, Diabase dyke swarmsof the
Canadian shield, Map 1627A, GeologicalSurvey of Canada,
Ottawa, ON, 1986.
Fahrig,W. F., E. H. Gaucher,andA. Larochelie,Paleomagnetism
ERNST AND BUCHAN 327
of diabasedykes of the CanadianShield, Can. d. Earth Sci., 2,
278-298, 1965.
Fahrig, W. F., K. W. Christie, E. H. Chown, D. Janes,and N.
Machado,The tectonicsignificanceof somebasicdyke swarms
in the CanadianSuperiorProvincewith specialreferenceto the
geochemistryand paleomagnetismof the Mistassini swarm,
Quebec,Canada, Can. d. Earth Sci., 23,238-253, 1986.
F6raud, G., G. Giann6rini,and R. Campredon,Dyke swarmsas
paleostressindicatorsin areasadjacentto continentalcollision
zones: examples from the European and northwestArabian
plates,in Mafic Dyke Swarms,Spec.Pap. 34, editedby H. C.
Halls and W. F. Fahrig,pp. 273-278, GeologicalAssociationof
Canada, Toronto, ON, 1987.
Gibbs,A. K., Contrastingstylesof continentalmafic intrusionsin
the GuianaShield,in Mafic DykeSwarms,Spec.Pap. 34, edited
by H. C. Halls and W. F. Fahrig, pp. 457-465, Geological
Associationof Canada, Toronto, ON, 1987.
Gibson, I. L., N. S. Madhurendra,and W. F. Fahrig, The
geochemistryof the Mackenzie dyke swarm,Canadain Mafic
Dyke Swarms,Spec.Pap. 34, editedby H. C. Halls and W. F.
Fahrig,pp. 109-121,GeologicalAssociation
of Canada,Toronto,
ON, 1987.
Goldberg,S. A., andJ. R. Butler,LateProterozoicrift-relateddykes
of the southernandcentralAppalachians, easternUSA, in Mafic
Dykesand Eraplacement Mechanisms,editedby A. J. Parker,P.
C. Rickwood, and D. H. Tucker, pp. 131-144, Balkema,
Rotterdam, 1990.
Goldberg,S. A., J. R. Butler, and P. D. Fullagar,The Bakersville
dike swarm:geochronology andpetrogenesis of Late Proterozoic
basalticmagmatismin the southernAppalachianBlue Ridge,
Am. d. Sci., 286, 403-430, 1986.
Gomes, L. C. C., M. A. F. Tanner de Oliveira, and L. R. B. Leal,
Structuralfeaturesassociated
with mafic dikes:examplesfrom
the Atlantic coastalbelt of Bahia, Brazil, Boletim IG-USP, Sdrie
Cientifica,20, 21-24, 1989.
Gorbatschev,R., A. Lindh, Z. Solyom, I. Laitakari, K. Aro, S.B.
Lobach-Zhuchenko, M. S. Markov, A. I. Ivliev, and I. Bryhni,
Mafic dyke swarmsof the Baltic shield,in Mafic Dyke Swarms,
Spec. Pap. 34, edited by H. C. Halls and W. F. Fahrig, pp.
361-372, GeologicalAssociationof Canada,Toronto,ON, 1987.
Green,J. C., T. J. Bornhorst,V. W. Chandler,M. G. Mudrey, P. E.
Myers, L. J. Pesonen,and J. T. Wilband,Keweenawandykesof
the Lake Superiorregion: Evidencefor evolutionof the Middle
ProterozoicMidcontinentRift of North Americain Mafic Dyke
Swarms,Spec.Pap. 34, editedby H. C. Halls and W. F. Fahrig,
pp. 289-302, GeologicalAssociationof Canada,Toronto,ON,
1987.
Greenough,J. D., and J.P. Hodych,Evidencefor lateralmagma
injectionin the early Mesozoicdykesof easternNorth America,
in Mafic Dykesand EraplacementMechanisms,editedby A. J.
Parker,P. C. Rickwood,and D. H. Tucker, pp. 35-46, Balkema,
Rotterdam, 1990.
Gresse, P. G., and R. Scheepers,Neoproterozoicto Cambrian
(Namibian) rocks of South Africa: a geochronologicaland
geotectonicreview,d. AJ•.Earth Sci., 16, 375-393, 1993.
Griffiths, R.W., and I. H. Campbell,Interactionof mantle plume
heads with the Earth's surface and onset of
328 GIANT RADIATING DYKE SWARMS, LIPS, AND MANTLE PLUMES
convection, d..Geophys.Res.,96, 18,295-18,310,1991.
Grosfils,E. B., and J. W. Head,The globaldistribution of giant
radiatingdike swarmson Venus:implications for the global
stressstate,Geophys.Res.Lett.,21, 701-704, 1994.
Gupta,V. K., Shadedimageof totalmagnetic field of Ontario,east-
centralsheet,scale1:1 000 000, Map 2586, OntarioGeological
Survey,Toronto,ON, 1991a.
Gupta, V. K., Shadedimage of total magneticfield of Ontario,
southern sheet, scale 1:1 000 000, Map 2587, Ontario
GeologicalSurvey,Toronto,ON, 1991b.
Gusev,G. S., andB. R. Shpount,Precambrian andPaleozoicrifting
in northeastern Asia, Tectonophysics,
143,245-252, 1987.
Halls, H. C., The importanceandpotentialof maficdykeswarmsin
studiesof geodynamicprocesses, Geosci.Canada, 9, 145-154,
1982.
Halls, H. C., The Matachewandyke swarm,Canada: An early
Proterozoicmagneticfield reversal,Earth Planet. Sci. Lett., 105,
279-292, 1991.
Halls, H. C., and M.P. Bates, The evolution of the 2.45 Ga
Matachewan dyke swarm, Canada, in Mafic Dykes and
EraplacementMechanisms,edited by A. J. Parker, P. C.
Rickwood,andD. H. Tucker,pp. 237-249,Balkema,Rotterdam,
1990.
Halls, H. C., K. G. Bums, S. J. Bullock, and P.M. Batterham,
Mafic dyke swarmsof Tanzaniainterpretedfrom aeromagnetic
datain Mafic DykeSwarms,Spec.Pap. 34, editedby H. C. Halls
and W. F. Fahrig, pp. 173-186, GeologicalAssociationof
Canada,Toronto, ON, 1987.
Harry, D. L., and Sawyer, D. S., Basaltic volcanism,mantle
plumes,and the mechanicsof rifting: the Parantiflood basalt
provinceof SouthAmerica,Geology,20, 207-210, 1992.
Hawkesworth,C. J., K. Gallagher,S. Kelley, M. Mantovani,D.W.
Peate,M. Regelous,and N. W. Rogers,Parantimagmatismand
the openingof the SouthAtlantic,in Magmatismand the Causes
of ContinentalBreak-up,Spec.Publ. 68, editedby B.C. Storey,
T. Alabaster,and R. J. Pankhurst,pp. 221-240, Geological
Societyof London,1992.
Heaman,L., U-Pb datingof giantradiatingdykeswarms:potential
for global correlationof mafic magmaticevents, Extended
Abstractsfor the InternationalSymposium on Mafic Dykes,S•o
Paulo, Brazil, compiledby W. Teixeira,M. Emesto,and E. P.
Oliveira,pp. 7-9, 1991.
Heaman,L. M., 2.45 Ga global mafic magmatism:Earth'soldest
superplume?, in Eighth International Conference on
Geochronology, Cosmochronology & IsotopeGeology,Berkeley,
California,Programwith Abstracts(Berkeley,California),p.
132, U.S. Geol. Surv. Circular 1107, 1994.
Heaman, L. M., U-Pb dating of mafic rocks: past, presentand
future (abstract), Program with Abstracts Geol. Assoc.
Can./Mineral. Assoc.Can., 20, A43, 1995.
Heaman, L. M., A. N. LeCheminant, and R. H. Rainbird, Nature
andtimingof Franklinigneousevents,Canada:Implicationstbr
a Late Proterozoicmantleplumeand the break-upof Laurentia,
Earth Planet. Sci. Lett., 109, 117-131, 1992.
Heatherington, A. L., andP. A. Mueller, Geochemical
evidencefor
Triassicrifting in southwestern Florida, Tectonophysics,
188,
291-302, 1991.
Helmstaedt,H.H., and J.J. Gurney, Geotectoniccontrolson the
formationof diamondsandtheirkimberliticandlamproitichost
rocks: Applications to diamondexploration,in Proceedings
5th
InternationalKimberliteConJbrence, Araxd, Brazil 1991, v. 2,
Diamonds:Characterization, GenesisandExploration,editedby
H. O. A. Meyer,andO. H. Leonardos, pp. 236-250, 1994.
Hill, R. I., Startingplumesandcontinentalbreak-up,Earth Planet.
Sci. Lett., 104, 398-416, 1991.
Hofmann,C., G. Feraud,R. Pik, C. Coulon,G. Yirgu, D. Ayalew,
C. Deniel,andV. Courtillot,4øAF/39AF
datingof EthiopianTraps.
IUGG XXI GeneralAssembly AbstractsWeekA, A465, 1995.
Holbrook, W. S., and P. B. Kelemen,Large igneousprovinceon
the US Atlanticmarginand implicationsfor magmatism during
continentalbreakup,Nature,364, 433-436, 1993.
Holbrook, W. S., G. M. Purdy, R. E. Sheridan,L. Glover III, M.
Talwani, J. Ewing, and D. Hutchinson,Seismicstructureof the
U.S. Mid-Atlantic continentalmargin, •. Geophys.Res., 99,
17,871-17,891, 1994.
Hooper, P. R., The ColumbiaRiver Basalt,in ContinentalFlood
Basalts,editedby J. D. Macdougall,pp. 1-33, KluwerAcademic
Publishers,1988.
Hooper,P. R., The timing of crustalextensionand the eruptionof
continentalflood basalts,Nature, 345, 246-249, 1990.
Hulbert, L., B. Williamson,and R. Th6riault,Geologyof Middle
Proterozoic MacKenzie diabase suites from Saskatchewan: An
overview and their potentialto host Noril'sk-typeNi-Cu-PGE
mineralization,in Summaryof Investigations 1993, Misc.Report
93-4, editedby R. Macdonald.T. I. I. Sibbald,C. T. Harper,D.
F. Paterson, and P. Guliov, pp. 112-126, Saskatchewan
GeologicalSurvey, Saskatchewan Mineral Resources, Regina,
SK, 1993.
Hunt, P. A., andJ. C. Roddick,A compilationof K-Ar ages,report
17, in RadiogenicAge and IsotopicStudies:Report 1. Geol.
Surv.Can., Paper87-2, pp. 143-210, 1987.
Hunter, D. R., and D. L. Reid, Mafic dyke swarmsin southern
Africa, data,in Mafic DykeSwarms,Spec.Pap. 34, editedby H.
C. Halls and W. F. Fahrig,pp. 445-456, GeologicalAssociation
of Canada,Toronto, ON, 1987.
Hutchinson, D. R., R. S. White, W. F. Cannon, and K. J. Schulz,
Keweenawhot spot:Geophysicalevidencefor a 1.1 Ga mantle
plume beneaththe Midcontinentrift system,•. Geophys.Res.,
95, 10,869-10,884, 1990.
Hyndman,D. W., and D. Alt, Radial dikes,laccoliths,and gelatin
models,J. Geol., 95, 763-774, 1987.
Isachsen,Y. W., W. M. Kelly, C. Sinton,R. A. Coish,and M. T.
Heizler, Dikes of northeastAdirondackregion:introductionto
their distribution, orientation, mineralogy, chronology,
magnetism, chemistry and mystery, in New York State
Geological Association, 60th Ann. Meeting, Field Trip
Guidebook,edited by J.F. Olmsted, pp. 215-243, New York
State,Albany, 1988.
Jefferson,C. W., L. J. Hulbert, R. H. Rainbird, G. E. M. Hall, D.C.
Gr6goire,andL. I. Grinenko,Mineralresource assessment
of the
NeoproterozoicFranklin igneous events of Arctic Canada:
Comparisonwith the Permo-Triassic Noril'sk- TalnakhNi-Cu-
PGE depositsof Russia,Open File 2789, 48 pp., Geological
Surveyof Canada,Ottawa,ON,

Jones,D. L., and W. F. Fahrig, Paleomagnetismand age of the
Aston dykes and Savage Point sills of the Boothia Uplift,
Canada, Can. J. Earth Sci., 15, 1605-1612, 1978.
Kamo, S. L., and C. F. Gower, Note: U-Pb baddeleyitedating
clarifiesage of characteristic
paleomagnetic
remanenceof Long
Range dykes, southeastern Labrador,Atl. Geol., 30, 259-262,
1994.
Kamo, S. L., C. F. Gower, and T. E. Krogh, Birthdatefor the
IapetusOcean?A preciseU-Pb zirconand baddeleyiteage for
the Long Range dikes, southeastLabrador,Geology, 17, 602-
605, 1989.
Kamo, S. L., T. E. Krogh, and P.S. Kumarapeli,Age of the
Grenville dyke swarm, Ontario-Quebec:implicationsfor the
timing of Iapetanrifting, Can.J. Earth. Sci.,32, 273-280, 1995.
Karkare,S. G., andR. K. Srivastava, Regionaldyke swarmsrelated
to the DeccanTrap alkalineprovince,India, in Mafic Dykesand
Emplacement Mechanisms, edited by A. J. Parker, P. C.
Rickwood,and D. H. Tucker,pp. 335-347, Balkema,Rotterdam,
1990.
Keppie,J. D., and R. D. Dallmeyer,Tectonicmap of Pre-Mesozoic
terranes in Circum-AtlanticPhanerozoic orogens, scale 1:
5,000,000, Nova Scotia Department of Natural Resources,
Halifax, 1989.
King, E. R., Precambrianterraneof north-centralWisconsin:an
aeromagneticperspective,Can. d. Earth Sci., 27, 1472-1477,
1990.
Klingspor,I., Radiometricage-determination of basalts,dolerites
and related syenite in Skfine, southernSweden, Geol. FOren.
StockholmFOrh., 98, 195-216, 1976.
Krasnov,I. I., M. J. Lurje, andV. L. Masaitis(editors),Geologyof
the SiberianPlatform,Nedra,Moscow,USSR,447 pp., 1966.
Krogh, T. E., D. W. Davis, and F. Corfu, PreciseU-Pb zirconand
baddeleyiteagesfor the Sudburystructure,in The Geologyand
Ore Depositsof theSudburyStructure,Spec.Vol. 1, editedby E.
G. Pye, A. J. Naldrett, and P. E. Giblin, pp. 431-445, Ontario
GeologicalSurvey,Toronto,ON, 1984.
Krogh, T. E., F. Corfu, D. W. Davis, G. R. Dunning, L. M.
Heaman, S. L. Kamo, N. Machado, J. D. Greenough,and E.
Nakamura, Precise U-Pb isotopic ages of diabasedykes and
mafic to ultramaficrocksusingtraceamountsof baddeleyiteand
zircon, in Mafic Dyke Swarms,Spec.Pap. 34, edited by H. C.
Halls and W. F. Fahrig,pp. 147-152, GeologicalAssociationof
Canada, Toronto, ON, 1987.
Kruk, W., GeologicalMap of the YemenArab Republic,SheetAl
Hazm, 1:250,000, Federal Institute for Geosciencesand Natural
Resource,1980.
Kumarapeli, S. P., G. R. Dunning, H. Pintson,and J. Shaver,
Geochemistryand U-Pb zirconage of commenditicmetafelsites
of the Tibbit Hill Formation,Quebec Appalachians,Can. J.
Earth Sci., 26, 1374-1383, 1989.
Kumarapeli,S. P., K. St. Seymour,A. Fowler,andH. Pinston,The
problemof the magmasourceof a giantradiatingmafic dyke
swarmin a failed arm setting,in Mafic Dykesand Emplacement
Mechanisms,editedby A. J. Parker,P. C. Rickwood,andD. H.
Tucker,pp. 163-171,Balkema,Rotterdam,1990.
Larson,R.L., Geologicalconsequences of superplumes,
Geology,
19, 963-966, 1991.
ERNST AND BUCHAN 329
Lawver, L. A., and R. D. Mtiller, The Iceland hotspottrack,
Geology,22, 311-314, 1994.
Lawyer,L. A., L. M. Gahagan,andM. F. Coffin,The development
of paleoseaways aroundAntarctica,Antarct.Res.Ser., 56, 7-30,
1992.
LeCheminant,A. N., Proterozoicdiabasedykeswarms;Lac de
Gras and Aylmer Lake area, District of Mackenzie,Northwest
Territories,scale 1: 250,000, Open File 2975, Geological
Surveyof Canada,Ottawa,ON, 1994.
LeCheminant,A. N., and L. M. Heaman, Mackenzie igneous
events, Canada: Middle Proterozoic hotspot magmatism
associatedwith ocean opening, Earth Planet. Sci. Lett., 96,
38-48, 1989.
LeCheminant,A. N., andL. M. Heaman,U-Pb agesfor the 1.27 Ga
Mackenzie igneousevents, Canada: Support for a plume
initiationmodel(abstract),in Programwith Abstracts,16, A73,
GeologicalAssociationof Canada,Waterloo,ON, 1991.
LeCheminant,A. N., andL. M. Heaman,779 Ma maficmagmatism
in the northwestern Canadian Shield and northern Cordillera: A
new regional time-marker (abstract), in 8th International
Conferenceon Geochronology,Cosmochronology and Isotope
Geology,Program with Abstracts(Berkeley,California). U.S.
Geol. Surv. Circular 1107, pp. 197, 1994.
LeCheminant, A. N., and O. van Breemen, U-Pb ages of
Proterozoicdyke swarms,Lac de Gras area, N.W.T.: evidence
for a progressivebreak-up of an Archean supercontinent,
Program with Abstracts, 19, A62, GeologicalAssociationof
Canada, Toronto, ON, 1994.
LeCheminant, A. N., L. M. Heaman, O. van Breemen, R. E. Emst,
W. R. A. Baragar,and K. L. Buchan.Mafic magmatism,mantle
roots and kimberlitesin the Slave craton, in Searchingfor
Diamonds in Canada, Open File 3228, edited by A. N.
LeCheminant, D. G. Richardson,R. N. W. DiLabio, and K. A.
Richardson,pp. 161-169,GeologicalSurveyof Canada,Ottawa,
ON, pp. 161-169, 1996.
Levashov, K. K. Paleorifi structurein the eastem environs of the
Siberian Platform. Int. Geol. Rev., 21, 188-200, 1979.
Macdonald,R., D. Gottfried,M. J. Farrington,F. W. Brown, andN.
G. Skinner, Geochemistryof a continentaltholeiite suite: late
Palaeozoicquartz doleritedykes of Scotland,Trans. R. Soc.
Edinburgh:Earth Sci.,72, 57-74, 1981.
Macdonald,R., R. Crossley,and K. S. Waterhouse,1983. Karoo
basalts of southern Malawi and their regional petrogenetic
significance,
Mineral. Mag., 47, 281-289, 1983.
MacDonald, R., L. Wilson, R. S. Thorpe, and A. Martin,
Emplacement of the Cleveland dyke: Evidence from
geochemistry, mineralogyandphysicalmodelling,J. Petrol., 29,
559-583, 1988.
Mahoney,J. J., DeccanTraps,in ContinentalFlood Basalts,edited
by J. D. Macdougall,pp. 151-194,Kluwer AcademicPublishers,
Dordrecht, 1988.
Mahoney,J. J., and K. J. Spencer,Isotopicevidencefor the origin
of the Manihiki and OntongJavaoceanicplateaus,Earth Planet.
Sci. Lett., 104, 196-210, 1991.
Mahoney, J., C. Nicollet, and C. Dupuy, Madagascarbasalts:
trackingoceanicandcontinentalsources,
Earth Planet.Sci. Lett.,
104, 350-363,
330 GIANT RADIATING DYKE SWARMS, LIPS, AND MANTLE
May, P. R., Patternof Triassic-Jurassic
diabasedykesaroundthe
North Atlantic in contextof predrifi positionof the continents,
Geol. Soc. Am. Bull., 82, 1285-1292, 1971.
McHone, J. G., Broad-terrane Jurassic flood basalts across
northeastem North America,Geology,24, 319-322, 1996.
McHone, J. G., M. E. Ross,and J. D. Greenough,1987. Mesozoic
dyke swarmsof easternNorth America,in Mafic Dyke Swarms,
Spec. Pap. 34, edited by H. C. Halls and W. F. Fahrig, pp.
279-288, GeologicalAssociationof Canada,Toronto,ON, 1987.
McKenzie, D., J. M. McKenzie, and R. S. Saunders, Dike
emplacementon Venus and on Earth, J. Geophys.Res., 97,
15,977-15,990, 1992.
Megrue,G. H., E. Norton,and D. W. Strangway,Tectonichistory
of theEthiopianrift asdeducedby K-Ar agesandpaleomagnetic
measurements of basalticdikes,J. Geophys.Res.,77, 5744-5754,
1972.
Mertanen,S., L.J. Pesonen,and H. Huhma, Palaeomagnetismand
Sm-Nd agesof the Neoproterozoicdiabasedykesin Laanilaand
Kautokeino, northern Fennoscandia, in Precambrian Crustal
Evolution in the North Atlantic Region,Spec.Publ. 112,edited
by T. S. Brewer,pp. 331-358, The GeologicalSociety,London,
1996.
Mohr, P., Ethiopianflood basaltprovince,Nature, 303, 577-584,
1983.
Mohr, P., Structure of Yemeni Miocene dike swarms and
emplacementof coeval granite plutons, Tectonophysics,
198,
203-221, 1991.
Mohr, P., and B. Zanettin,The Ethiopianflood basaltprovince,in
ContinentalFlood Basalts,editedby J. D. Macdougall,pp. 63-
110. Kluwer Academic Publishers,Dordrecht, 1988.
Morgan,W. J., Hotspottracksandthe openingof the Atlanticand
Indian oceans,in TheSea. volume7, editedby C. Emiliani,pp.
443-487. Wiley Interscience,
New York, 1981.
Morgan, W. J., Hotspottracksand the earlyrifting of the Atlantic,
Tectonophysics,94, 123-139, 1983.
Mubu, M. S., Aeromagneticmappingand interpretationof mafic
dyke swarmsin southernAfrica, M. Sc. thesis,Dept. of Earth
ResourceSurveys,InternationalInstitutefor AerospaceSurvey
and Earth Sciences,Delft, the Netherlands,1995.
Murthy, N. G. K., Mafic dykeswarmsof the Indianshield,in Mafic
Dyke Swarms,Spec.Pap. 34, editedby H. C. Halls and W. F.
Fahrig,pp. 393-400, GeologicalAssociation of Canada,Toronto,
ON, 1987.
Nadin, P. A., and N.J. Kusznir, Palaeoceneuplift and Eocene
subsidence in the northern North Sea basin from 2D forward and
reversestratigraphic
modelling,J. Geol. Soc.London,152, 833-
848, 1995.
Namibia GeologicalSurvey (Cape Cross), 1:250,000 Geological
Series, Sheet 2013, 1988.
Nielsen,T. F. D., Mafic dyke swarmsin Greenland:A review in
Mafic DykeSwarms,Spec.Pap. 34, editedby H. C. Halls andW.
F. Fahrig, pp. 349-360, Geological Associationof Canada,
Toronto, ON, 1987.
O'Connor,J. M., andR. A. Duncan,Evolutionof the Walvisridge-
Rio Granderise hot spot system:implicationsfor African and
SouthAmericanplate motionsover plumes,J. Geophys.Res.,
95, 17,475-17,502, 1990.
PLUMES
Oliveira, E. P., J. Tamey, and X. J. Jono, Geochemistryof the
Mesozoic Amapti and Jari dyke swarms, northern Brazil:
Plume-relatedmagmatismduring the opening of the central
Atlantic, in Mafic Dykesand EraplacementMechanisms,edited
by A. J. Parker,P. C. Rickwood,andD. H. Tucker,pp. 173-183,
Balkema, Rotterdam, 1990.
Oskarsson,N., S. Steinthorsson,and G. E. Sigvaldason,Iceland
geochemical anomaly: origin, volcanotectonics,chemical
fractionationandisotopeevolutionof the crust,J. Geophys.
Res.,
90, 10,011-10,025, 1985.
Osmani, L. A., Proterozoicmafic dike swarmsin the Superior
Provinceof Ontario,in Geologyof Ontario, Spec.Vol. 4, Part 1,
editedby P. C. Thurston,H. R. Williams,R. H. Sutcliffe,andG.
M. Stot,pp. 661-681, OntarioGeologicalSurvey,Toronto,ON,
1991.
Palmer,H. C., B. A. Merz, and A. Hayatsu,The Sudburydikesof
the GrenvilleFrontregion:paleomagnetism, Petrochemistry,
and
K-Ar agestudies,Can.J. Earth Sci., 14, 1867-1887,1977.
Park, J. K., K. L. Buchan,and S.S. Harlan, A proposedgiant
radiatingdyke swarmfragmentedby the separation of Laurentia
and Australiabasedon paleomagnetism of ca. 780 Ma mafic
intrusions in western North America, Earth Planet. Sci. Lett.,
132, 129-139, 1995.
Parsons,T., G. A. Thompson,and N.H. Sleep,Mantle plume
influence on the Neogene uplift and extensionof the U.S.
westernCordillera?,Geology,22, 83-86, 1994.
Peterman, Z. E., and P. K. Sims, The Goodman swell: A
lithosphericflexure caused by crustal loading along the
midcontinentrift system,Tectonics,7, 1077-1090, 1988.
Piccirillo, E. M., G. Bellieni, R. Petrini, G. Cavazzini, P. Comin-
Chiaramonti, M. H. F. Macedo, A. J. Melfi, P. Zantedeschl,J.P.
P. Pinese,and G. Martins, Mesozoicmafic dykesand intrusives
from Brazil: petrology, geochemistryand Sr-Nd isotopes.
ExtendedAbstractsfor the InternationalSymposiumon Mafic
Dykes,S6o Paulo,Brazil, compiledby W. Teixeira,M. Emesto,
andE. P. Oliveira,pp. 15-16, 1991.
Prevec, S. A., R. S. James,R. R. Keays, and D.C. Vogel,
Constraintson the genesisof Huronian magmatismin the
Sudbury area from radiogenic isotopic and geochemical
evidence.,in The NorthernMargin of the SouthernProvinceof
the CanadianShield,ProgramandAbstracts,Can. Mineral., 33,
4, 930-932, 1995.
Radhakrishna,T., M. Joseph,P. K. Thampi, and J. G. Mitchell,
Phanerozoicmaficdykeintrusionsfrom the highgradeterrainof
southwesternIndia: K-Ar isotopeand geochemicalimplications,
in Mafic Dykesand Eraplacement Mechanisms,editedby A. J.
Parker, P. C. Rickwood, and D. H. Tucker, pp. 363-372,
Balkema, Rotterdam, 1990.
Radhakrishna, T., R. D. Dallmeyer, and M. Joseph,
Palaeomagnetism and '•Ar/q"Arvs. 3"Ar/q"Arisotopecorrelation
ages of dyke swarms in central Kerala, India: tectonic
implications,EarthPlanet.Sci.Lett., 121,213-226, 1994.
Ragland,P. C., R. D. HatcherJr., andD. Whittington,Juxtaposed
Mesozoic dike sets from the Carolinas: A preliminary
assessment, Geology,11,394-399, 1983.
Rainbird, R. H., The sedimentaryrecord of mantle plume uplift
precedingeruptionof theNeoproterozoic Natkusiakflood

d. Geol., 101,305-318, 1993.
Rainbird, R. H., D. A. Hodgson, W. Darch, and R. Lustwerk,
Bedrockandsurficialgeologyof NortheastMinto Inlier, Victoria
Island, District of Franklin, Northwest Territories, scale
1:50,000, Open File 2781, Geological Survey of Canada,
Ottawa, ON, 1994a.
Rainbird,R. H., D. A. Hodgson,andC. W. Jefferson, Bedrockand
surficialgeology,Washington Islands(NTS 78 B/5) District of
Franklin, NorthwestTerritories,scale1:50,000, OpenFile 2920,
GeologicalSurveyof Canada,Ottawa,ON, 1994b.
Ransome,I. G. D., The geodynamics,
kinematicsandgeochemistry
of the Gannakouriepdyke swarm,M.Sc. thesis,Universityof
CapeTown, 1992.
Raposo, M. I. B., and M. Ernesto,Anisotropyof magnetic
susceptibility
in the PontaGrossadyke swarm(Brazil) and its
relationshipwith magmaflow direction,Phys. Earth Planet.
Inter., 87, 183-196, 1995.
Reid, D. L., The Cape Peninsuladolerite dyke swarm, South
Africa, in Mafic DykesandEmplacementMechanisms, editedby
A. J. Parker,P. C. Rickwood,and D. H. Tucker, pp. 325-334,
Balkema, Rotterdam,1990.
Reid, D. L., and D.C. Rex, Cretaceousdykesassociated
with the
openingof the SouthAtlantic:the Mehlbergdyke, northern
Richtersveld,SouthAJ?.J. Geol.,97, 135-145, 1994.
Reid, D. L., A. J. Erlank,andD.C. Rex, Age and correlationof the
False Bay dolerite dyke swarm, south-western Cape, Cape
Province,SouthAJ?.J. Geol.,94, 155-158,1991a.
Reid, D. L., I. G. D. Ransome,T. C. Onstott,and C. J. Adams,
Time of emplacement and metamorphism of late Precambrian
mafic dykesassociated with the Pan-AfricanGariep orogeny,
southernAfrica: implications
for the ageof the Nama group,J.
AJ?.EarthSci.,13,531-541,1991b.
Renne,P. R., andA. R. Basu,Rapideruptionof the SiberianTraps
flood basaltsat the Permo-Triassic
boundary,Science,253, 176-
179, 1991.
Renne,P. R., T. C. Onstott,M. S. D'Agrella-Filho,I. G. Pacca,and
W. Teixeira,4øAr?9Ar datingof 1.0-1.1 Ga magnetizations from
the Silo FranciscoandKalaharicratons:tectonicimplicationsfor
Pan-African and Brasiliano mobile belts, Earth Planet. Sci. Lett.,
101,349-366, 1990.
Renne,P. R., D. F. Mertz, M. Emesto,L. Marques,W. Teixeira,H.
H. Ens, and M. A. Richards,Geochronologicconstraintson
magmaticand tectonicevolutionof the ParantiProvince
(abstract),
Eos Trans.AGU 1993 Fall MeetingSuppl,p. 553,
1993.
Renne,P. R., K. Deckart,M. Emesto,G. F•raud,E. M. Piccirillo,
Ageof thePontaGrossa
dikeswarm(Brazil),andimplications
to
Parantiflood volcanism,Earth Planet. Sci. Lett., 144, 199-211,
1996a.
Renne,P. R., J. M. Glen, S.C. Milner, A. R. Duncan,Age of
Etendeka flood volcanism and associated intrusions in
southwesternAfrica, Geology,24, 659-662, 1996b.
Richards,M. A., R. A. Duncan,andV. E. Courtillot,Floodbasalts
and hot spottracks:Plumeheadsandtails,Science,246, 103-
107, 1989.
Robertson,W. A., and W. R. A. Baragar,The petrologyand
palcomagnetism
of the Coronation
sills,Can.•. EarthSci.,9,
ERNST AND BUCHAN 331
123-140, 1972.
RowIcy,D. B., andA. L. Lottes,Plate-kinematic
reconstructions
of
the North Atlantic and Arctic: Late Jurassic to present,
Tectonophysics,
155, 73-120, 1988.
Rosen,O. M., K. C. Condie,L. M. Natapov,and A.D. Nozhkin,
Archcanand Early Proterozoicevolutionof the Siberiancraton:
a preliminary assessment,in Archcan Crustal Evolution,
Developmentsin PrecambrianGeology 11, edited by K. C.
Condie,pp.411-459,Elsevier,Amsterdam, 1994.
Sant, D. A., and R. V. Karanth,Emplacementof dyke swarmsin
the Lower Narmadavalley,westernIndia, in Mafic Dykesand
EmplacementMechanisms,edited by A. J. Parker, P. C.
Rickwood,and D. H. Tucker,pp. 383-389. Balkema,Rotterdam,
1990.
Sebai,A., V. Zumbo, G. F6raud,H. Bertrand,A.G. Hussain,G.
Giann6rini,and R. Campredon, 40Ar/39Ardatingof alkalineand
tholeiiticmagmatism of SaudiArabiarelatedto the earlyRed
Searifling,Earth Planet.Sci.Lett., 104,473-487, 1991a.
Sebai,A., G. Feraud,H. Bertrand,
andJ. Hanes,4øAr/3"Ar datingand
geochemistry of tholeiiticmagmatismrelatedto the early
openingof theCentralAtlanticrifl, EarthPlanet.Sci.Lett.,104,
455-472, 1991b.
Oeng6r,A.M. C., and B. A. Natal'in,Palcotectonics
of Asia:
fragmentsof a synthesis,
in The TectonicEvolutionof Asia,
editedby A. Yin andT. M. Harrison,pp. 486-640,Cambridge
UniversityPress,1996.
Shpount,
B. R., andB. V. Oleinikov,A comparison
of maficdyke
swarmsfrom the SiberianandRussianplatforms,in Mafic Dyke
Swarms,Spec.Pap. 34, editedby H. C. HallsandW. F. Fahrig,
pp. 393-400,GeologicalAssociationof Canada,Toronto,ON,
1987.
Sial, A. N., E. P. Oliveira,andA. Choudhuri,Mafic dyke swarms
of Brazil, in Mafic DykeSwarmsSpec.Pap. 34, editedby H. C.
HallsandW. F. Fahrig,pp. 379-383,Geological Associationof
Canada,Toronto, ON, 1987.
Smith.W. A., Palcomagnetic resultsfroma crosscuttingsystemof
northwestand north-southtrendingdiabasedikes in the North
CarolinaPiedmont,Tectonophysics, 136, 137-150,1987.
Smith,R. B., andL. W. Braile,The Yellowstonehotspot:physical
properties, topographic and seismicsignature, and space-time
evolution(abstract),Eos Trans.AGU 1993Fall MeetingSuppl,
p. 602, 1993..
Smythe,D. K., M. J. Russell,andA. G. Skuce,Intra-continental
rifling from the majorlate Carboniferous quartz-dolerite
dyke
swarmof NW Europe,Scott.J. Geol.,31,151-162, 1995.
Southwick,D. L., and W. C. Day, Geologyand petrologyof
Proterozoicmafic dikes, north-centralMinnesotaand western
Ontario,Can.J. Earth Sci.,20, 622-638, 1983.
Speight,
J.M., R. R. Skelhom,
T. Sloan,andR. J.Knaap,Thedyke
swarmsof Scotland,in IgneousRocksof theBritishIsles,edited
by D. S. Sutherland,pp. 449-459,Wiley and Sons,London,
U.K., 1982.
Stewart,K., S. Turner,S. Kelley, C. Hawkesworth,L. Kirstein,M.
Mantovani, 3-D, 4øAr-39Argeochronology in the Paranti
continentalflood basaltprovince,Earth Planet.Sci. Lett., 143,
95-109, 1996.
Storey,B.C., The roleof mantleplumesin continental
332 GIANT RADIATING DYKE SWARMS, LIPS, AND MANTLE
casehistoriesfrom Gondwanaland, Nature,377, 301-308, 1995.
Storey, M., J. J. Mahoney, A.D. Saunders,R. A. Duncan, S. P.
Kelley, andM. F. Coffin, Timing of hot spot-related
volcanism
and the breakupof Madagascarand India, Science,267, 852-
855, 1995.
St. Seymour,K., and P.S. Kumarapeli,Geochemistryof the
Grenvilledykeswarm: Role of plume-source mantlein magma
genesis,Contrib.Mineral. PetroL, 120, 29-41, 1995.
Sundvoll,B., andB. T. Larsen,Rb-Sr and Sm-Ndrelationships in
dyke and sill intrusionsin the Oslo Riff and relatedareas,Nor.
Geol. Unders.Bull., 425, 25-41, 1993.
Svenningsen, O.M., Extensional deformation and dyke
emplacement alongthe Late Precambrian Baltoscandian passive
margin: the Sarek dyke swarm, Arctic SwedishCaledonides
(abstract), Program & Abstractsfor the Third International
Dyke Conference,Sept. 4-8, 1995, Jerusalem,Israel, editedby
A. Agnon,andG. Baer,pp. 73, 1995.
Swanson,D. A., T. L. Wright, and R. T. Helz, Linearvent systems
and estimatedratesof magmaproductionand eruptionfor the
Yakima basalt on the Columbia Plateau, Am. J. Sci., 275,
87%905, 1975.
Thompson, R. N., andS. A. Gibson,Subcontinentalmantleplumes,
hotspotsand pre-existingthinspots,J. Geol. Soc.London,148,
973-977, 1991.
Tolan, T. L., S. P. Reidel, M. H. Beeson,J. L. Anderson,K. R.
Fecht, and D. A. Swanson,Revisionsto the estimatesof the areal
extent and volume of the Columbia River Basalt Group, in
Volcanism and Tectonism in the Columbia River Flood-Basalt
Province,Spec. Pap. 239, edited by S. P. Reidel, and P. R.
Hooper,pp. 1-20, GeologicalSocietyof America,Boulder,CO,
1989.
Tollo, R. P., andF. E. Hutson,700 Ma riff eventin theBlueRidge
provinceof Virginia: a unique time constrainton pre-Iapetan
rifling of Laurentia,Geology,24, 59-62, 1996.
Tomshin,M.D., and O. V. Koroleva,Compositedykesof the
Vilyuisk palcoriffsystem,SiberianPlatform,Yakutia, USSR in
Marie Dykes and EraplacementMechanisms,edited by A. J.
Parker, P. C. Rickwood, and D. H. Tucker, pp. 535-540,
Balkema, Rotterdam, 1990.
Tomshin,M. D, andA. V. Okrugin,Dyke swarmof alkalinebasites
in the north of the Siberianplatform,Program with Abstracts,
Third International Dyke Conjkrence,Jerusalem,Israel, edited
by A. Agnon,andG. Baer,p. 76, 1995.
Tucker, D. H., and D. M. Boyd, Dykes of Australiadetectedby
airbome magneticsurveys,in Mafic Dyke Swarms,Spec.Pap.
34, edited by H. C. Halls and W. F. Fahrig, pp. 163-172,
GeologicalAssociationof Canada,Toronto,ON, 1987.
Turner, S., M. Regelous,S. Kelley, C. Hawkesworth,and M.
Mantovani,Magmatismand continentalbreak-upin the South
Atlantic:high precision•'Ar-3"Argeochronology,Earth Planet
Sci. Lett., 121,333-348, 1994.
Vail, J. R., Tectoniccontrolsof dykesandrelatedirruptiverocksin
easternAfrica, in AJkicanMagmatismand Tectonics,editedby
T. N. Clifford, and I. G. Gass,pp. 337-354, Oliver & Boyd,
Edinburgh,1970.
Vakar,B. A., Trap formationsof the Taymyr,Petrologyof Eastern
Siberia: Volume 1, Siberian Platform and its Northern Rim,
PLUMES
editedby G. D. Afanas'ev,
pp.256-340,Academyof Sciences
of
the U.S.S.R, 1962.
Walker,G. W., andN. S. MacLeod,GeologicMap of Oregon,
scale 1:500,000, U.S. Department of the Interior, U.S.
GeologicalSurvey,Washington, D.C., 1991.
West,G. F., andR. E. Ernst,Evidencefrom aeromagnetics
on the
configurationof Matachewan dykesandthetectonicevolutionof
theKapuskasing Structural
Zone,Ontario,Canada,Can.J. Earth
Sci., 28, 1797-1811, 1991.
White,R. S., Magmatism
duringandaftercontinental
break-up,
in
Magmatismand the Causesof ContinentalBreak-Up,Spec.
Publ. 68, edited by B.C. Storey, T. Alabaster,and R. J.
Pankhurst, pp. 1-16,The GeologicalSociety,London,1992.
White, R. S., and D. McKenzie, Magmatismat rift zones:the
generationof volcaniccontinentalmarginsand flood basalts,J.
Geophys.Res.,94, 7685-7729, 1989.
White, R. S., andD. McKenzie,Mantle plumesandfloodbasalts,J.
Geophys.Res., 100, 17543-17585,1995.
Wilson, J. F., A craton and its cracks:some of the behaviourof the
Zimbabwe block from the Late Archean to the Mesozoic in
response to horizontalmovementsandthe significance of some
of its mafic dyke fracturepatterns,J. AJk.Earth Sci, 10, 483-501,
1990.
Wilson,E. D., D. L. Jones,andJ. D. Kramers,Mafic dykeswarms
of Zimbabwe,in Mafic DykeSwarms,Spec.Pap. 34, editedby
H. C. Halls and W. F. Fahrig, pp. 433-444, Geological
Associationof Canada,Toronto, ON, 1987.
Windley, B. F., The EvolvingContinents,JohnWiley and Sons,
2nd ed., 1984.
Wirth, K. R., J. D. Vervoort,and L. M. Heaman,Nd isotopic
constraints on mantle and crustal contributions to 2.08 Ga
diabasedykes of the southernSuperiorProvince(abstract),
Program & Abstractsfor the Third International Dyke
Conference,Sept. 4-8, 1995, Jerusalem,Israel, editedby A.
Agnon,andG. Baer,p. 84, 1995.
Woolley, A. R., and M. S. Garson,Petrochemicaland tectonic
relationshipof the Malawi carbonatite-alkaline
provinceandthe
Lupata-Lebombo volcanics, in AJkican Magmatism and
Tectonics,editedby T. N. Clifford,andI. G. Gass,pp. 237-262,
Oliver & Boyd, Edinburgh,1970.
Zhao,J.-X., andM. T. McCulloch,Sm-Ndmineralisochron
agesof
Late Proterozoicdyke swarms in Australia: evidencefor two
distinctiveeventsof mafic magmatismand crustalextension,
Chem. Geol., 109, 341-354, 1993.
Zhao, J.-X, M. T. McCulloch, and R. J. Korsch,Characterization
of
a plume-related
-800 Ma magmaticeventanditsimplications
for
basin formation in central-southernAustralia, Earth Planet. Sci.
Lett., 121,349-367, 1994.
Zoback,M. L., E. H. McKee,R. J. Blackely,andG. A. Thompson,
The northernNevadariff: Regionaltectono-magmatic relations
and middle Miocene stressdirection,Geol. Soc.Am. Bull., 106,
371-382, 1994.
Zolotukhin,V. V. andA. I. Al'Mukhamedov,
Trapsof theSiberian
platform, in Continental Flood Basalts, edited by J. D.
Macdougall, pp. 273-310, Kluwer Academic Publishers,
Dordrecht, 1988.
Zumbo, V., G. F6raud, H. Bertrand, and G. Chazot,
 ERNST AND BUCHAN 333

chronologyof Tertiary magmaticactivity in SouthernYemen KennethL. Buchan,GeologicalSurveyof Canada,601 Booth

duringthe early Red Sea-Adenrifting, J. Volcanol.Geotherm.
Res., 65,265-279, 1995a.
Zumbo,V., G. F6raud,P. Vellutini, P. Piguet,and J. Vincent,First
4øAr/39Ar
datingon Early Plioceneto Plio-Pleistocene magmatic
eventsof the Afar - Republicof Djibouti,J. Volcanol.Geotherm.
Res., 65, 281-295, 1995b.
St.,Ottawa,Ontario,K 1A 0E8 Canada,kbuchan•gsc.nrcan.gc.ca.
RichardE. Ernst,GeologicalSurveyof Canada,601 BoothSt.,
Ottawa, Ontario, K IA 0E8 Canada,remst•gsc.nrcan.gc.ca;
and
Departmentof Earth Sciences,University of WesternOntario,
London, Ontario, N6A 5B7
 Plume/LithosphereInteractionin the Generationof Continentaland
OceanicFlood Basalts' ChemicalandIsotopicConstraints

John C. Lassiterl and Donald J. DePaolo

Departmentof Geologyand Geophysics,UC Berkeley,Berkeley,California

The plume initiation model for flood basalt genesis predicts that melt
generationwill occur almost entirely within the plume head, beneath the
lithosphere. However, isotopic and trace element differences between
continentaland oceanicflood basalts(CFBs and OFBs, respectively)requirethe
incorporationof a lithosphericcomponentin the former. Debate persistsas to
whether the "continental" signaturespresent in many CFBs derive from
contamination of (plume-derived) melts with small volumes of highly
incompatible-element-enriched lithosphericcomponents(e.g., through crustal
assimilation)or substantialmelt generationwithin the lithosphericmantle. The
chemicaland isotopiccompositionsof OFBs and CFBs are usedto constrainthe
mantle sourcesand melting environmentresponsiblefor flood basaltgeneration
and the extent of crustalassimilation. Major and trace elementtrendsin CFBs
reflect smallerextentsof partialmeltingthan in OFBs andthe presenceof gamer
in CFB sourceregions. These observationsare consistentwith the plume
initiation model, becausethick continentallithospherewill inhibit ascentand
melt generationin sublithosphericplumes. However, melt generationfrom
refractory, Fe-poor lithosphericmantle is also indicated for several CFBs,
includingthe SiberianTraps. Depth of melt generationtypically decreaseswith
time for a given province,as indicatedby the removalof a gamersignaturein
trace element ratios (e.g., decreasingSm/Yb). These chemical and temporal
trends suggestearly melt generationwithin hydrousbut otherwiserefractory
lithosphericmantle, followed by mechanicalerosionof the lithospherethat
allows additionalascentand melt generationwithin the underlyingplume.

1Send correspondenceto current addressat: Departmentof
TerrestrialMagnetism, CarnegieInstitutionof Washington,5241
Broad BranchRoad, NW, Washington,DC 20015
Large IgneousProvinces:Continental,Oceanic,andPlanetary
Flood Volcanism
GeophysicalMonograph 100
Copyright1997 by the AmericanGeophysicalUnion
1. INTRODUCTION
The associationof hotspotsand flood basaltshas long
been recognized[e.g., Morgan, 1971]. Fluid-dynamic
models [e.g., Griffiths, 1986; Griffiths and Campbell,
1990] have produceda plausiblemodel for flood basaltsas
forming when large plume heads associatedwith new
mantleplumesascendbeneaththe lithosphere.However,
considerabledebatepersistsconcerningthe actual source
of magmasfor flood basalts. Severaltheoreticaltreatments
of plume/lithosphereinteractionhave concludedthat melt
336 PLUME/LITHOSPHERE INTERACTION IN FLOOD
generationin responseto plume ascentis largelyrestricted
to the plume itself, with little melt generationoccurring
within the lithosphericmantle [e.g.,Arndt and Christensen,
1992; Farnetani and Richards, 1994]. However, many
continentalflood basalts(CFBs) have isotopicand trace
elementsignaturesvery dissimilarto thosefound in either
oceanic flood basalts (OFBs) or ocean island basalts
(OIBs). Although crustal assimilationmay accountfor
some of the differences between continental and oceanic
flood basalts, several studies have proposedthat the
continentallitho-sphericmantle (CLM) is a major source
of CFB melts [e.g., Lightfootet al., 1990a;Gallagherand
Hawkesworth, 1992].
In this paper, we review the available chemicaland
isotopicdatafor lavasfrom severalcontinental andoceanic
flood basalt provinces and OIBs. Our purposeis to
determine the role of the lithosphericmantle in the
generationof flood basaltsand to evaluatewhetherthe
predictionsconcerningmelt generationinherent in the
plume initiationmodel are borne out. For example,do
CFBs, which form in regionswherethe lithosphereis old
andthick, reflect smallerpercentages of meltingat greater
depth than OFBs (or OIBs), as predictedby the plume
initiation model? In order to establish the role of the
lithosphericmantlein the generationof floodbasalts,it is
necessaryto distinguishchemicalsignaturesarisingfrom
crustal contaminationfrom those generatedby partial
melting of the CLM. Are the distinctivechemicaland
isotopic signaturesof CFBs explained adequatelyby
crustalcontaminationalone, or is melt generationwithin
the CLM required? We review the chemicalevidencefor
both crustalcontaminationand melt generationfrom the
CLM as well as the inherent limitationsof geochemical
studies for uniquely distinguishingbetween the two.
Althougheveryfloodbasaltprovinceis unique,ourgoalis
to illustrate several features that are common to the
evolutionof many continentaland oceanicflood basalts.
By so doing, we hope to developa genericmodel of
plume/lithosphere interactionthat is consistentwith the
available observations. Finally, by discussing the
limitationsof traditionalisotopicand traceelementstudies
for distinguishingcrustal contam-inationfrom CLM
melting,we seekto stimulatefuture researchcapableof
addressingthis long-standingproblem in flood basalt
studies.
2. MELT FRACTIONS AND DEPTH OF
MELT GENERATION
Systematicrelationshipsbetweenlithospherethickness,
depthof melting,andmelt fractionareexpected
for plume-
BASALTS
generatedbasalts. As hot mantle ascendsin upwelling
zones,it intersectsits solidusat somedepthand beginsto
partially melt. The depth at which melting begins is
largely a function of the potential temperatureof the
ascending mantle [e.g., McKenzie and Bickle, 1988],
although variations in composition, especially volatile
content, may also result in more or less fertile source
regions[e.g., Falloon and Green, 1989]. In extensionalor
rift environments,suchas at mid-oceanridges,lithosphere
is thin or absent, so there is essentiallyno barrier to
adiabatic ascent [McKenzie and Bickle, 1988]. The
thickness of the resultant melt zone and the total amount of
melt producedare directlyrelatedto potentialtemperature,
so that averagedepth and averagedegreeof melting are
stronglycorrelatedfor differentmid-oceanridge segments
[e.g., Klein and Langmuir, 1987]. In non-rifling
environments,the lithosphereactsas a mechanicalbarrier
to mantle ascent. The thickness of the melt zone and the
maximumextentof partialmeltingare thereforelimitedby
the lithospherethickness [e.g., McKenzie and O'Nions,
1995]. Consequently,plumes rising beneaththick, rigid
lithosphere should undergo less partial melting than
plumesrising underthin lithosphere,andthe averagedepth
of melting will be greater [e.g., Ellam, 1992; Saunderset
al., 1992].
The depth and extent of melting in plumes can be
estimated from analysis of both the major and trace
element compositionsof basalts. For example, with
increasing depth of origin FeOT (total Fe as FeO)
increases,and with increasingmelt fraction at a given
pressureNa20 decreasesand CaO/A1203increase[e.g.,
Hirose and Kushiro, 1993; Baker and Stolper, 1994]. For
trace elements,incompatibleelement concentrations are
approximatelyinversely pro-portionalto melt fraction.
Depth of melting is qualitativelyindicatedby the presence
or absenceof a residual-garnetsignaturein the basalts,
becausegarnetis stablein mantleperidotiteonly at depths
greaterthan-75 km [cf. Nickel, 1986].
The effects of fractional crystallization on chemical
composition must be removed in order to evaluate
chemicalsignaturesderiving from melt generation. The
majorityof lavasfrom both continentaland oceanicflood
basaltprovinceshave undergonesignificantfractionation,
evidenced
by low Mg#s [=100 x molarMg/(Mg+Fe2+)]
and MgO and Ni contents[e.g., Mahoney, 1988; Wooden
et al., 1993; Lassiter et al., 1995a]. To compare the
compositionsof lavas that have undergone different
fractionationhistories,major element compositionswere
correctedto a constant8 wt% MgO using best-fit linear
regressions,excludingpicritic samplesand sampleswith
lessthan 4 wt% MgO [e.g., Langmuiret al., 1992].
 LASSITER AND DEPAOLO 337

regression techniqueis similarto theregressions of Penget amongthe OFBs considered herebecause,althoughthese

al. [1994] and Turner and Hawkesworth[1995]. The main flood basaltswere eruptedin an oceanicenvironment,the
limitationsare the poorly definedmajor elementtrendsof preexisting arc lithospherethrough which the basalts
manyfloodbasaltsuites,whichresultin largeuncertainties
in the regressions,and the fact that fractionationtrendsare Ontong
Java I
• Caribbean
NauruBasin II
only approximatelylinear becausedifferent phasesmay 1.0 ß Plateau

•'Manihiki
Plateau
II ' • ßi ß• ß• ß
becomesaturatedat differentstagesof fractionation. The []
O
Wrangellia
SiberianTraps
results of our regressionsare generally consistentwith 0.9
!-]
A
Deccan
Traps
Columbia River
thoseof Turner and Hawkesworth[ 1995] and Peng et al. ß MORB

[1994], althoughsomedifferencesexist. For example,we O Back-arc

basins
calculatesomewhatlower FeBand higher NaB for average
Deccan lavas than do Turner and Hawkesworth [1995] •••. •;* ,'•l•creasing Fand
P'
(subscriptdenotesoxide extrapolationto 8 wt%). More
importantly,we alsoutilize incompatibleelementratiosto
assessthe depth and extent of melting. As we discuss
below, our estimates for relative depth and degree of
melting obtainedby these two independentmethodsare
•
ß*h•••Melt••fion
I
beneat
: I :
ß lithosphericlid
largelymutually consistent.
Melt generationbeneath
lithosphericlid
3.5
2.1. Major ElementRegressions

Figure 1 showscalculatedmeanNaBand (Ca/A1)8 values 3.0

plotted against FeB for several continentaland oceanic
flood basalt suites. Also shownare averageextrapolated 2.5 R Increasing
F andP --

values for two intraplatehotspots(Hawaii and R6union)
and one on-ridge hotspot(Iceland). The mid-oceanridge
basalt(MORB) array providesa referencecorresponding
to basaltsproducedwhere there is approximatelyzero-
thicknesslithosphere,but where largevariationsin mantle
potentialtemperatureresultin strongcorrelationsbetween
depth of melting and melt fraction. If plume-sourceand
MORB-source mantle have similar major element
compositions,intraplate plume-derivedmelts should be
shiftedoff the MORB array so that smallermelt fractions
(evidenced by higher NaB and lower (Ca/A1) 8) are
associatedwith a higher pressure(greaterdepth)of origin
(higherFEB),as qualitativelyshownin Figure 1.
Many CFBs have major elementcompositions consistent
with the effectsoutlinedaboveof melt generationbeneath
a lithosphericlid. With the possibleexceptionof the
Columbia River basalts (but see below) and some
Gondwana basalts, all of the OFBs, CFBs, and OIBs
examinedpossesshigh FeB comparedwith most MORB
(e.g., >10 wt%), suggestingderivationfrom mantle with
anomalouslyhigh potentialtemperature. Lavas from the
Siberian and Deccan Traps have higher NaB and lower
(Ca/A1)8for a given FeBthan mostMORB and plot at the
edge of or outside the MORB array in Figure 1. The
Wrangellia flood basalts also possessmajor element
compositionssuggestiveof melt generationbeneath a
lithosphericlid. The Wrangellia flood basaltsare unique
H
2.0
1.5 , t , I , t a I t I t I a I a
6 8 10 12 14
Fes (wt%)
Figure 1. Average (CaYA1)8and NaB versusFeB for several
continental and oceanic flood basalt suites and three ocean island
hotspots. Data for these and subsequent figuresare from the
following: OntongJavaPlateau[Mahoney,1987;Mahoneyand
Spencer,1991;Mahoneyet al., 1993];NauruBasin[Floyd, 1986;
Mahoney, 1987; Mahoney and Spencer, 1991; Castillo et al.,
1991]; Caribbean Plateau [Donnelly et al., 1973; Aitken and
Etchevera,1984; Sen et al., 1988; Kerr et al., 1996]; Manihiki
Plateau [Jacksonet al., 1976; Clague, 1976; Mahoney, 1987;
Mahoneyand Spencer,1991]; Wrangellia[Lassiteret al., 1995a];
SiberianTraps [DePaoloand Wasserburg, 1979;Lightfootet al.,
1990b;Sharrnaet al., 1991;Sharrnaet al., 1992;Lightfootet al.,
1993; Woodenet al., 1993];DeccanTraps[Mahoneyet al., 1982;
Cox and Hawkesworth, 1985; Mahoney, 1988; Lightfoot et al.,
1990a;Peng et al., 1994]; ColumbiaRiver [Carlsonet al., 1981;
Carlson, 1984; Nelson, 1983; Nelson, 1989; Hooper and
Hawkesworth,1993]; Iceland(I) [Hernondet al., 1993]; Hawaii
(H) [Rhodes,1996; Yanget al., 1996];R6union(R) [Fisket al.,
1988;Albaredeand Tamagnan,1988]. Averagesfor MORB and
back-arcbasin basaltsare from Klein and Langrnuir[1987].
ParantiandFerrarlow-Ti basaltaveragesrecalculated fromHergt
et al.
338 PLUME/LITHOSPHERE INTERACTION IN FLOOD BASALTS

erupted was more similar in terms of its thicknessand 4.0 ' I ' I ' I '
compositionto continental lithospherethan to oceanic
lithosphere. The Wrangellia flood basalts therefore 3.5

represent a hybrid flood basalt province, sharing Crustal assimilation and/or

similarities to both continental and oceanic flood basalts refractory mantle source
3.0 + J

[Lassiter et al., 1995a]. That the inferred melting

environmentsof the Wrangellia flood basaltsand some ..•+ß + Iimnahal
++

CFBs are similar is thusnot entirelyunexpected. 2.5

Most otherOFBs, includingthe OntongJavaPlateauand + ß

Nauru Basin, the Manihiki Plateau, and the Caribbean 2.0

+ ++ +
+
+

Basin, all show evidenceof high melt fractionsand depth

+
+ +
+

of origin but no effect of a lithosphericlid becausethey + +

plot well within the high-melt-fractionend of the MORB
field. This observation is consistent with previous
conclusionsand with suggestionsthat several of these
plateaus formed near ridges or triple junctions [e.g.,
Mahoney, 1987; Mahoney et al., 1993]. Icelandicbasalts
also plot within the high-melt-fractionend of the MORB-
array, as expected given Iceland's location along a
spreadingridge. In contrast,lavas from the intraplate
Hawaiian and R6unionhotspotshave higherNaB,and lavas
from R6unionhave lower (Ca/A1)8for their Fe8valuesthan
basaltswithin the MORB-array. As with many of the CFB
suites,these trends suggestmelt generationbeneaththe
lithospherewithin hot mantleplumes.
Because of the well-documentedisotopic differences
between MORB, flood basalts, and OIBs, we must
consider whether the chemical trends described
reflect variationsin the major elementcompositionof the
mantle sources rather than differences in melting
conditions. Recent studies indicate that major element
variationsare presentin the plume sourcesof someOIBs
[e.g., Hauri, 1996]. However, the shifts away from the
MORB melting array observedfor severalCFBs and for
lavas from Wrangellia, R6union, and Hawaii are unlikely
to be the result of such variations, becausemost OFBs and
on-ridge hotspot lavas from Iceland do not record such
shifts in major element compositiondespite possessing
similar Sr- and Nd-isotopic signaturesas the R6union,
Hawaii, and Wrangellialavas.
Crustal assimilation may also affect the extrapolated
major element compositionsof CFBs. For example,the
low Fe8 inferred for the Columbia River basaltsappearsin
part to be the result of crustal contamination. Figure 2
comparesthe estimatedFeBandNaBfor the Imnahaandthe
Grande Ronde suites of the Columbia River
Crustalassimilationappearsto have stronglyaffectedthe
isotopic and trace element compositionsof the Grande
Ronde basalts[e.g., Carlson et al., 1981; Nelson, 1989].
In contrast, the Imnaha basalts appear more directly
derived from a mantle plume source with less crustal
1.5 i I • I • I • ] i I , I
6 8 10 12
Fes (wt%)
Figure 2. Typical effects of crustalassimilationon flood basalt
composition
are qualitativelyillustratedby the lowerFeBvalues
of (crustally contaminated) Bushe and Grande Ronde lavas
comparedwith less contaminatedlavas from the Ambenali and
Imnaha suitesof the Deccan Traps and ColumbiaRiver Basalt
Province. The trend to lower Fe8 producedby crustal
contamination or melt generationfrom an Fe-poorsource(e.g.,
back-arcbasinlavas)is roughlyoppositethat expectedfor melt
generationat high pressurebeneatha litho-sphericlid. Symbols
for MORB andback-arclavasare asin Figure 1.
interaction[e.g., Brandon and Goles, 1988; Hooper and
Hawkesworth,1993]. Consistentwith this conclusion,the
above Imnahabasaltshave higher FeBvaluesthan GrandeRonde
basaltsandhavemajor elementcompositions that suggest
they were generatedbeneatha lithosphericlid. A similar
comparisoncan be made for the Bushe and Ambenali
suitesof the DeccanTraps. Althoughoxygenisotopicdata
for phenocrystsindicate that the Bushe and severalother
early formationsin the Deccanassimilatedcrustalmaterial,
Ambenali lavasare interpretedto more closelyrecordthe
isotopiccompositionof a sublithospheric source[e.g.,Cox
and Hawkesworth,1985; Peng et al., 1994]. As with the
GrandeRondebasalts,Peng et al. [ 1994] notedthat Bushe
lavashave lower FeBthan lesscontaminatedlavassuchas
the Ambenali. These examples illustrate that crustal
contaminationwill in general lower the inferred FeB
contentof a suite of lavas. The high FeBvaluesof CFBs
suchasthe Deccanand SiberianTrapsthereforecannotbe
explainedby this mechanism.
Melt generationfrom refractorymantlesources(e.g.,the
basalts. CLM) will have a similar effect on FeB. The low Fe
contentof many back-arclavas (Figure 1) suggests they
are derivedfrom a sourcemore refractorythan typical
MORB-sourcemantle[e.g., Falloon et al., 1994] andthus
are similar to what we expect for melts derived from
refractory lithosphere. CLM-derived melts
 LASSITER AND DEPAOLO 339

commonly possesslower Fe contentsfor two reasons.

. i . 'tl,•. .%0
,o.o, . , .
First, Fe content in primitive magmas is strongly CC • o UC
influencedby the pressureof melt generation,with Fe Meltingof o o
increasing with increasing pressure [e.g., Hirose and spinel
peridotitel
[ o o
Kushiro, 1993]. Therefore, for similar source
_

(low
F) / / n n • o o000o o -

compositions,melts from sublithosphericplumes should CLM---•

have higher Fe contentsbecauseof their greaterdepthof
origin. Also, lithosphericmantle is likely to be more
refractory on average than the underlying convecting
mantle. Melt extraction,and the accompanyingchemical
buoyancythis imparts, may be importantfor stabilizing
ancient lithosphericmantle [e.g., Jordan, 1978]. Mantle
xenoliths typically are depletedin a basaltic component;
e.g., they have lower Fe, Ca, A1, and Na relative to
primitivemantleestimates[e.g.,McDonough,1990, 1992].
As a result, melts generatedfrom refractoryCLM should
have lower Fe than melts generatedunder comparable
conditionsfrom asthenospheric mantle. Picritesfrom the
Karoo confirm that CLM-derived melts can be relatively
Fe-depleted. Ellam et at. [1992] showed that Karoo
picrites with a large lithospheric-mantle-derived
component(as indicatedby unradiogenicOs isotopes)had
lower Fe contentsthan picrites without a lithosphere-
derivedcom-ponent. Thus, crustalcontaminationandmelt
generationfrom refractorylithosphericmantleboth reduce
the inferred Fe content of parental magmas and thereby
obscure the signaturesof melt generation beneath a
lithosphericlid.
Becausethe major element signaturesof both crustal
contaminationand melt generationfrom refractorylitho-
sphereare almostoppositeto the signaturesproducedby
melt generationfrom plumesbeneaththe lithosphere,the
high NaB and low (Ca/A1)8for a given FeBrecordedby
average lavas from the Deccan and Siberian Traps and
Wrangelliarelative to most MORB or OFBs likely reflect
melt generationbeneaththe lithosphere.This conclusionis
further strengthenedfrom considerationof trace element
fractionationsin these lavas, as discussedbelow.
2.2. Trace Element Ratios
Trace elementabundanceratiosprovide an independent
meansof assessing the depthand extentof partial melting
responsiblefor generationof flood basalts. Becauserare
earthelement(REE) partitioncoefficientsin mostcommon
mantle phasesare well established,and becausethe bulk
solid/melt partitioning of heavy REE is dramatically
differentfor garnetand spinelperidotite,REE profilesare
particularlyuseful for determiningthe relative extentsof
partialmeltingandthe mineralogy(andthereforedepth)of
the sourceof flood basalts.Partialmeltingof eithergarnet
[--] •afi o o
ß ß !misseau
Fm.,Haiti.
Melting of garnet
peridotite(low F)
DMMa Upper
Noril'sk
0 1 2 3 4
Sm/Yb
Figure 3. LaJSm and Sm/Yb values for continentaland oceanic
flood basalt suites. Batch melting trends for garnet and spinel
peridotite calculated using the partition coefficients, modal
abundances,and primitive (PM) and depleted (DMM) mantle
trace element ratios of McKenzie and O'Nions [1991]. Arrows
denote the effect of decreasingmelt fraction (F). Median
compositionof CLM [McDonough, 1990], lower (LC), upper
(UC), and bulk continentalcrust (CC) [Taylor and McLennan,
1985] shownfor comparison.Symbolsare as in Figure 1.
or spinel peridotite will preferentiallyenrich the lighter
REE such as La, which are more incompatiblein most
mantle phasesthan middle REE suchas Sm. However, the
degree of enrichmentof middle REE relative to heavy
REE suchas Yb dependsgreatly on whethergarnetexists
as a residualphaseduringmelting,becausethe heavyREE
are preferentiallyretainedby garnetduringmeltingbut not
by most other mantle phases [cf. compilation of
distributioncoefficientsby McKenzieand 0 'Nions,1991].
Further-more,fractionalcrystallizationwill produceonly
modestchangesin La/Sm andSm/Yb ratioscomparedwith
variationscausedby changesin melt fraction or source
mineralogy. For the REE, the evolvednatureof many
flood basalts therefore introduces fewer uncertainties than
it did in our assessment
of major elementtrends.
The enrichmentsin La/Sm and Sm/Yb producedby
batchmeltingof spineland garnetperidotiteare shownin
Figure 3, along with La/Sm and Sm/Yb values for lavas
from several continental and oceanic flood basalt
provinces. Lavas from the Ontong Java Plateauand the
Nauru Basinhave relativelyflat REE patterns,with lower
La/Sm and Sm/Yb than most continentalflood basalts,and
appear to representhigh-degreemelts [Castillo et at.,
1991; Mahoney et at., 1993]. Lavas from the Caribbean
Plateau have variable La/Sm and Sm/Yb ratios. Basalts
from Curacao and drill sites in the central Caribbean
primarily have low Sm/Yb ratios similar to those
340 PLUME/LITHOSPHERE INTERACTION IN FLOOD BASALTS
OntongJavaand Nauru Basin lavas [Donnellyet al., 1973;
Kerr et al., 1996], but lavas from the Dumisseau
Formation,Haiti, have high Sm/Yb ratios and appearto
have beengeneratedthroughsmallerdegreesof meltingof
a garnet-bearingsource [Sen et al., 1988]. Though not
shown in Figure 3, lavas from the Manihiki Plateaualso
have nearly flat REE patterns [Jackson et al., 1976],
similarto lavasfrom OntongJavaandNauruBasin.
In contrast,most CFBs have higher La/Sm and Sm/Yb
than OFBs. The higher La/Sm and Sm/Yb in most CFBs
suggestthey are generatedthrough smaller degreesof
partial melting than most OFBs (see below for discussion
of the effects of crustalcontamination). The Wrangellia
flood basalts have lower La/Sm and Sm/Yb than most
CFBs but higher than other OFBs, which indicatesthese
basalts were generatedthrough intermediatedegreesof
partial melting. As we observedfor the major element
compositionsof the Wrangellia basalts,this is consistent
with the prevolcanichistory of Wrangellia,becausethese
flood basaltseruptedthrough preexistingarc lithosphere
which, given its age (minimum of •55 to 125 Ma at the
time of volcanism[Lassiteret al., 1995b]), was mostlikely
thicker than average oceaniclithospherebut thinner than
ancientcontinentallithosphere.
Comparedwith the oceanicplateaus,continentalflood
basaltsdisplaymuch greaterscatterin their REE profiles.
However, much of this scatterarisesfrom contaminationof
plume meltsby meltsderivedfrom continentalcrustand/or
lithosphericmantle. As with the major elementtrends,we
must examine the effects of this contamination on trace
element ratios in order to understandthe plume melting
environment. A common characteristic of this contam-
ination in many continental flood basalt suites is a
depletion in high-field-strength elements (HFSEs),
particularly Nb and Ta, relative to other similarly
incompatibleelements. For example, CFBs commonly
have sub-chondriticNb/La and Ta/La ratios, whereasmost
oceanic plume-derived basalts (OFBs and OIBs) have
higher than chondriticNb/La and Ta/La [e.g., Arndt and
Christensen,1992]. In order to examinethe conditionsof
melt generationwithin the plumes responsiblefor flood
basalt volcanism, we have attempted to filter out the
greater part of the continentalcontaminationby only
plotting in Figure 4 data for lavas with minimal Nb-Ta
depletions (e.g., Nb/La or Ta/La > 0.85 x chondritic
value). When the data are fiRered in this manner, the
scatterpresentin Figure 3 is reducedsignificantly. More
importantly,althoughthe averageLa/Sm of many CFB
suites is diminished, the Sm/Yb values are similar to the
unfiltereddata. As a result,it is muchmore apparentfrom
the shallow La/Sm-Sm/Yb trend of the fiRered data that
R
Iceland
Hawaii
0 I 2 3 4
Sm/Yb
Figure 4. La/Sm-Sm/Yb valuesfor continentalflood basaltlavas
(and Wrangellia) with minimal Nb-Ta depletions. Only samples
with Nb/La > 0.9 or La/Ta < 20 are included. Field for Iceland
lavasincludessampleswith > 5 wt% MgO. Batchmeltingcurves
are as in Figure 3. Symbolsfor Wrangellia,SiberianandDeccan
Traps,and ColumbiaRiver basaltsare as in Figure 1.
melting of garnet-bearingsourceswere importantin the
generationof theselavas. Becausegarnetis only stablein
the mantle at depths greaterthan-75 km [e.g., Nickel,
1986], the more pronouncedgarnet signaturein the REE
patterns of relatively un-contaminatedCFBs compared
with most OFBs is consistentwith the major element
indicationsthat CFBs are primarily generatedthrough
smaller degrees of partial melting of anomalouslyhot
mantlebeneatha thick lithosphericlid.
Comparisonof the major element and trace element
trendsfor individualsuitesof lavasfurtherstrengthens
this
conclusion. For example, most Columbia River basalts
have low Nb/La values and also have lower FeB values
than many other flood basalts(Figure 1). However, lavas
from the relatively uncontaminatedImnaha suite of the
Columbia River Basalt Group have high FeB and Nas
(Figure 2), which we interpret as resulting from melt
generation beneath the continental lithosphere. The
lmnaha lavas, represented in Figure 4, are also
characterizedby near-chondriticNb/La values and have
high Sm/Yb and moderateLa/Sm values that are most
readilyexplainedby meltingof a garnet-bearing source.
Data for three ocean island suites are also plotted in
Figure4. Lavasfrom the intraplateHawaiianandR6union
hotspotsare stronglyenrichedin Sm relativeto Yb, again
indicatingmelt generationfrom a garnet-bearingsource.
Icelandicbasaltsshowa considerable rangein La/Sm and
Sm/Yb, but on average have much lower Sm/Yb ratios
than Hawaiian or R6union lavas and have REE patterns
similar to those of many OFBs. Again, theseresults

consistent with the differences in major element
compositionsbetween Icelandic and intraplatehotspot
lavas. IcelandiclavashavelowerNaBfor a givenFeBthan
lavas from either Hawaii or R6union.Becausethe Hawaii,
R6union, and Iceland plumes are all characterizedby
similarisotopicsignatures(e.g., high 3He/nilebut g•d and
87Sr/86Sr values intermediate between those
MORB and inferred for the bulk earth [Condomineset al.,
1983,'Albaredeand Tamagnan,1988; Kurz et al., 1996]),
the major and trace element differences between lavas
fromthe intraplateHawaiianandR6unionhotspotsandthe
on-ridgeIcelandhotspotare unlikelyto reflect systematic
differencesin the compo-sitionsof these plumes. The
observed compositional differences between Iceland,
Hawaii, and R6unionlavasare insteadwhat are expected
giventhe tectonicsettingsof thesethreehotspots,and are
consistentwith the conclusionsof Ellam [1992], who
reporteda similar relationshipbetweenaverageCe/Yb in
oceanislandbasaltsand the age of the lithospherethrough
whichthe basaltswere erupted.
An unexpectedfinding of our analysisis that basaltq
from R6union and to a lesser extent from
apparentlyrepresentsmallerdegreesof meltingthan either
continentalor oceanic flood basalts. A comparisonof
R6union and Deccan basalts in Figures 1 and 4 is
especiallyinstructive,becauseinitiation of the R6union
plume has been linked to the formation of the Deccan
Traps[e.g., Vandammeand Courtillot,1990]. In particular,
oncethe effectsof continentalcontaminationare removed,
for instance by comparing R6union lavas with the
relatively uncon-taminatedAmbenali suite, the R6union
lavas have major element compositions that plot farther
from the MORB field and have greater incompatible
elementfractionations,indicatingderivationfrom smaller
extentsof melting [e.g., Ellam, 1992]. This may seem
surprising,becauseplume "tails" shouldbe at leastas hot
as plume "heads," as entrainmentof ambient mantle in
rising plume heads will tend to reduce the potential
temperatureof the head [Griffiths and Campbell, 1990].
Furthermore,ancientcon-tinentallithosphereis typically
several hundred kilometers thick, much thicker than
averageoceaniclithosphere(-100 km) [e.g., Chapmanand
Pollack, 1977]. As we discussin a latersection,the higher
melt fractionsinferred for many CFBs than for intraplate
hotspot basalts may reflect the erosion of continental
lithospherethat often accompanies flood basaltvolcanism.
Erosion of oceanic lithosphereby steady-stateplume
"tails" may be significantlyless pronounced,perhaps
becauseoceaniclithosphereis typicallydrier andtherefore
strongerthan continentallithosphereor becausethe time
LASSITER AND DEPAOLO 341
during which any portion of a moving lithosphericplate
residesover a narrow plume conduitis too shortto permit
significantheatingandsofteningof the lithosphericlid.
3. CRUSTAL ASSIMILATION OR MELT
GENERATION WITHIN THE CLM?
found in
The above discussionillustratesthe strongcontrolthat
the lithosphericmantle has on melting in sublithospheric
plumes. This influence over the melting environmentis
generally consistentwith the plume initiation model for
flood basaltgenesis,which predictsmelt generationto be
largelyconfinedto the plumeheadbeneaththe lithosphere
[e.g., Arndt and Christensen, 1992; Farnetani and
Richards, 1994]. However, pronouncedisotopicand trace
element differences exist between continental and oceanic
flood basalts(Figure 5). Becausehigh 3He/4Heratios in
lavasassociatedwith severalCFBs suggesta componentof
undegassed lower mantleorigin [Basuet al., 1993; Basuet
al., 1995; Dodson et al., 1996], the compositionsof deep-
seatedmantle plumes ascendingbeneaththe continents
Hawaii should not systematicallydiffer from those ascending
beneathocean basins. Therefore,the isotopicand trace
element differences between CFBs on the one hand and
OIBs and OFBs on the otherrequirethe incorporationof a
"continental"componentin the former. If theseisotopic
differences can be explained as resulting from
volumetrically small additions of material to plume-
derived melts, such as through crustal assimilation,then
there is no problem with plume model predictionsthat
meltingis confinedlargelyto the plumeinterior.However,
if a substantialfractionof magmaproductionoccurswithin
the lithosphere,we would need to modify currentmodels
of plume/lithosphereinteraction. It is important to
remember that contaminationof plume-derived basalts
from continental crust and melt generation from the
lithosphericmantle are not mutuallyexclusive,but that it is
the magnitudeof the latter that may help constrainmodels
of plume/lithosphereinteraction.
Unfortunately,resolvingthis mostbasicquestionon the
originof the isotopicandtraceelementsignatures in CFBs
has proven extremely difficult. Because of the
heterogeneityof both continentalcrustand the CLM and
the complexityof the processes involvedin partialmelting
and assimilation,most geochemicalstudieshave failed to
conclusivelyprove or disprovethe involvementof specific
components. In the following discussion,we review the
geochemicalevidence and circumstantialargumentsin
supportof both crustal and lithospheric-mantle sources
for the "continentalcomponent"in variousCFB
342 PLUME/LITHOSPHERE INTERACTION IN FLOOD BASALTS

MORB

Oceanic Plateaus Oceanic

Plateaus]

Siberian Traps Siberian Traps

Deccan Traps Deccan Traps

-15 -10 -5 0 5 10 15 0.0 0.5 !.0 1.5 2.0

•--'Nd Nb/La

Figure 5. Representativerangesfor gNd and Nb/La in differenttectonicsettings. Dashedline denoteschondritic

Nb/La ratio [Sun and McDonough, 1989]. CFBs are isotopicallymuch more heterogeneousthan either OFBs or
MORB. OFB rangesincludesamplesfrom the OntongJavaPlateau,Nauru Basin,Manihiki Plateau,CaribbeanBasin,
and Wrangellia Terrane. MORB data sources:Dosso et al. [1993]; Price et al. [1986]; Prinzhofer et al. [1989];
Michael [

3.1. Chemical Discrimination of Crust and CLM Compo-
nents in CFBs
In order to distinguishthe effectsof crustalassimilation
from assimilation or melting of the CLM, we must
examine geochemicalsignaturesthat are clearly different
in thesetwo reservoirs. Oxygen isotopedata, particularly
for phenocryst phases, show promise in this regard,
becauseO isotope variationsin both asthenospheric and
lithosphericmantle are small comparedwith the large 1so-
enrichmentscommonly found in continentalcrust. Fresh
MORB possesswell-defined b180 values of 5.7 + 0.3%o
[Ito et al., 1987]. CLM-derived xenolithsshow slightly
greaterO isotopeheterogeneity,but for the most part also
possessb180 < 6%o[e.g., Kyser, 1986; Mattey et al., 1994].
In contrast, many CFBs have elevated b180, suggesting
they have assimilated1sO-enrichedcontinentalcrust. For
example,phenocrystsin the Busheand severalother lower
stratigraphicformationsof the Deccan Traps have higher
b180 values than commonly found in uncontaminated
mantle-derived melts or xenoliths [Peng et al., 1994].
Furthermore, there is a rough positive correlation in
Deccan lavas between b180 values and other indications of
"continental" contamination such as elevated 87Sr/86Sr.
Lavas of the Grande Ronde suite from the Columbia River
Basalt Province are also characterizedby elevatedwhole-
rock b180 values [e.g., Nelson, 1983; Carlson, 1984], as
are many lavas of the Paranti,[e.g., Fodor et al., 1985;
Harris et al., 1990]. Unfortunately,relatively few detailed
studiesof stable isotopesystematicsin CFBs have been
performed. Especiallyneededare studiesexaminingb180
valuesin both whole-rockand phenocrystphases,because
O isotopes in phenocrysts are more resistant to
postmagmaticalterationthan are whole-rocks[e.g., Eiler et
al., 1996].
Osmium isotopesalso have the potentialto differentiate
crustal and CLM contributions to CFBs. Continental crust
has high Re/Os and is characterized by extremely
radiogenicOs isotopes.In contrast,numerousstudieshave
demonstrated that the CLM is characterized by
subchondritic¾osdue to long-term Re depletion [e.g.,
Walkeret al., 1989; Carlsonand Irving, 1994; Pearsonet
al., 1995]. Therefore, assimilation of crustal and CLM
componentswill have opposite effects on Os isotopes.
Unfortunately, because Os behaves as a compatible
elementduring fractionalcrystallization,Os abundancesin
the highly evolved lavastypically dominantin flood basalt
provincesare extremelylow. As a result,mostOs isotopic
studies of CFBs have concentratedon relatively rare
picriteso However, the high Os abundancesin picrites
make them much less sensitive to crustal assimilation than
LASSITER AND DEPAOLO 343
Os isotopesin basaltic lavas. Nevertheless,several Os
isotope studies have suggestedCLM additionsto flood
basalts rather than addition of continental crust. Osmium
isotope variations in picrites from the Karoo province
provide the most unambiguous evidence for a CLM
component in a flood basalt sequence. These picrites
displaya positivecorrelationbetweenOs and Nd isotopes,
so that picriteswith the strongest"continental"signatures
(e.g., low l•Nd) have the lowest ¾os. Therefore, the
continentalcomponent in the Karoo picrites must have
beenderivedfrom the (low ¾o0CLM ratherthan (high
continentalcrust [Ellam et al., 1992]. Other Os isotope
studieshave suggesteda CLM componentin picritesfrom
the North American midcontinentrift [Shirey, 1997] and
the SiberianTraps [e.g., Horan et al., 1995].
Recent improvementsin low-blank Re and Os analysis
and sample/spike equilibration methods (thus allowing
more accurate Re/Os determinations)have opened the
window for Os isotopic studies of the low-MgO (and
therefore low-Os) basaltsthat make up the bulk of most
CFB provinces. Low-Os basaltsare much more sensitive
to crustalcontaminationthan high-Os picrites,so that the
presenceor absenceof radiogenic Os isotopesin these
basaltsplacesa strongconstrainton the timing and extent
of crustal contamination. For example, basalts of the
Ferrar flood basalt province have initial ¾o•values only
slightly higher than values typically found in oceanic
plume-derivedlavas. Therefore,either thesebasaltswere
not crustally contaminatedto any significant degree, or
crustal contamination occurred when the basalts were
much more picritic so that the higher Os concentrations
bufferedthe effects of crustalassimilation[Molzahn et al.,
1996]. In contrast, the Grande Ronde basalts of the
ColumbiaRiver basaltprovincehave extremelyradiogenic
Os isotopes,whereas lavas of the Imnaha group in the
same province do not [Chesley et al., 1996]. As we
discussedearlier, the Grande Ronde basalts have been
strongly influenced by crustal contamination, but the
Imnahabasaltsappearto be essentiallyunmodifiedplume-
derived basalts. These studiesillustratethe potential for
Os isotopes to reveal the respective roles of crustal
assimilationand CLM melting in flood basaltgenesis.
In contrastto Os and O isotopes,most trace element
ratios are much less diagnosticof crustal or CLM con-
tamination. For example, many CFBs have pronounced
depletionsin HFSEs relativeto plume-relatedbasaltsfrom
oceanic settings (OIBs and OFBs), as indicated, for
example, by low Nb/La (Figure 5; see also Arndt and
Christensen[1992]). Average continentalcrust is also
stronglydepletedin HFSEs [e.g., Taylor and McLennan,
1985]. In contrast, although xenoliths derived from
344 PLUME/LITHOSPHERE INTERACTION
CLM have highly variableNb/La ratios,on averagethey
are not depleted in HFSEs [e.g., McDonough, 1990].
Theseobservations have led many researchers
to conclude
that the low Nb/La values presentin many CFB suites
require crustal assimilation rather than assimilation of
CLM-derived melts [e.g., Woodenet al., 1993;Arndt et al.,
1993; Brandon and Goles, 1995].
Unfortunately, HFSE depletions in CFBs are not as
diagnostic of crustal contamination as is commonly
assumed. The conclusion that HFSE depletions are
uniquelydiagnosticof crustalassimilationis beliedby the
existenceof CLM-derived, HFSE-depletedlavas. Many
lamproites,for example, have Nb/La < 1 [Mitchell and
Bergman,1991], as do minettesfrom the ColoradoPlateau
[Thompsonet al., 1990]. The Karoo picrites clearly
illustrate that at least some CFBs derive their
depletions from the CLM rather than continentalcrust.
These picrites possess strong negative Nb anomalies
[Ellam and Cox, 1989, 1991] that are correlatedwith ¾Os
(Figure 6). This correlationis oppositethat expectedfor
assimilation of high-¾oscrustal material and therefore
indicatesthat the extremely low Nb/La values in these
picrites and associatedbasalts must have been derived
from somecomponentor processwithin the low-¾os CLM.
A numberof factorscouldexplainwhy appropriatesources
for the Karoo picrites are not well representedin global
xenolith data bases. Mantle xenoliths may provide a
chemicallybiasedsamplingof the CLM, or Nb/La ratios
may be fractionatedduring partial melting to a greater
extent than is commonly assumed (e.g., Nb may be
retainedby trace phasessuchas ruffle at smalldegreesof
partial melting [e.g., Green and Pearson, 1987], thereby
decreasingNb/La in the melt). In any event,althoughthe
origin of the HFSE depletionsin the Karoo picritesis not
completelyunderstood,they nevertheless demonstrate that
suchdepletionscannotbe usedto rule out the presenceof
CLM-derivedcomponents in otherfloodbasaltsuites.
3.2. Melt Generationfrom the CLM.' Evidencefrom the
Siberian Traps
The SiberianTraps provide an excellentexamplefor
assessingthe relative importanceof plume, crustal,and
lithosphericmantle componentsin a CFB. Althoughthis
province has been studied extensively, little consensus
exists concerningthe nature of the mantle source(s)for
theselavasor the degreeto which the magmasassimilated
crustalmaterial. For example,Sharmaet al. [1991, 1992]
concluded that the bulk of the Siberian flood
interactedlittle with eitherthe lithosphericmantleor crust
and that their compositions closely resemble the
IN FLOOD BASALTS
12
I ' I
Crustal
contamination
Other Nuanetsi
picrites
s4
Gomakwe area
picrites
0.3 0.4 0.5 0.6 0.7 0.8
Nb/La
HFSE Figure 6. The positive correlationbetweenNb/La and ¾Os
observedfor picrites from the Karoo flood basalt province is
inconsistentwith the derivation of HFSE depletions(e.g., low
Nb/La) throughcrustalassimilation,but is consistent with mixing
betweenmeltsderivedfrom a high ¾Os, highNb/La plumesource
and a low ¾Os,low Nb/La (and Fe-poor)sourcewithinthe CLM.
Picrites sampledfrom near the Gomakwe intrusionconsistently
plot alongseparate,subparalleltrendsin variouselement-isotope
plots [Ellam et al., 1992]. Data for Karoo picrites: Ellam and
Cox [ 1989]; Ellam and Cox [ 1991]; Ellam et al. [ 1992].
compositionof their plume source. In contrast,Lightfoot
et al. [1993] proposedthat the isotopicand traceelement
signatures of lavasfrom the Noril' sk regionof the Siberian
Traps reflect the incorporationof a lithosphericmantle-
derivedmelt, whereasWoodenet at. [ 1993] proposedthat
the compositionsof theselavaswere controlledprimarily
by crustal assimilationand fractional crystallizationin
large, near-steady-state
crustalmagma chambers. In this
section we examine each of these scenarios in turn,
drawingupon data availablefrom the literature.
The Siberian Traps can be divided into three
subprovinces:the Putorana,the Meimecha-Kotui,and the
Noril'sk [e.g., Sharrna et al., 1991]. The Putorana
formationsrepresentthe volumetric bulk (-90%) of the
province[e.g., $harrna et at., 1992]. However, mostof the
detailed stratigraphic and geochemical work on the
SiberianTraps has concentratedon the earlier sequences
recoveredin drill coresand exposuresfrom the Noril'sk
region. The upper Noril'sk formationsappearto be the
chemicaland temporalequivalentsof thoseof the Putorana
[e.g., $harrnaet al., 1992] and thusreflectthe composition
of the bulk of the province. However,the heterogeneous
lower Noril'sk formations contain lavas that represent
basalts distinctgeochemicalend-members.Mixtures of theseend-
members can explain most of the geochemical
characteristicsof the upper sequences.In particular,

upper-sequencelavas have isotopic and trace element
ratios that, to first order, can be explainedby mixtures of
lavassimilarto thoseof the Gudchikhinskyand Tuklonsky
suites. The origin of these volumetrically minor lower
sequencesis therefore critical to our understandingof the
origin of the main uppersequences.
Lavas from the Siberian Traps display considerable
isotopicheterogeneity.The gNavalues,for example,range
from +6 to - 11 (Figure 5). However,Sharmaet al. [ 1991,
1992] noted that most of this heterogeneitywas restricted
to lavas erupted early in the history of the province and
that volumetrically most lavas were very homogeneous,
with gNa• +2. They concludedthat this homogeneity
requires the lavas to have been derived from a
homogeneous,nearly primitive plume source,with little
subsequentmodification. However, the conclusionthat a
homogeneous lava sequence,no matterhow large,requires
melt generation from a homogeneous(plume) mantle
sourceappearsunjustified. For example,the isotopically
homogeneous Putoranaand upperNoril'sk lavasalsohave
very homogeneousmajor elementcompositions,but these
compositionsare far removed from those of primary
mantle melts. The Mg#s in the Putorana and upper
Noril'sk vary systematicallybetween-50 and 60, yet
primarymantlemelts have much higherMg#s (>_70)[e.g.,
Roederand Eroslie,1970]. Obviously,someprocesseither
accompanyingor following crystalfractionationacted on
theselavasto producethe observedhomogeneityin major
element compositions,and there is no reasonto believe
that the same process could not also account for the
observedisotopichomogeneityo
Furthermore,most Siberian Traps lavas, includingthe
homogenousupper sequencesdiscussedby Sharma et al.
[1992], have many trace elementsignatures that are unlike
thoseobservedin otherplume-relatedlavas. For example,
most Siberian Traps lavas (including the homogeneous
upperNoril' sk section)have pronouncedHFSE depletions,
with (TaJLa)•4 and (Nb/La)•4 <1 (subscript denotes
normalizationto chondriticratio), whereasmost OIBs and
OFBs have superchondritic ratios(Figure 5; seealsoArndt
and Christensen [1992]). Thus, although we cannot
excludethe possibilitythat thesetrace elementsignatures
derive from a plume source,the inferred compositionof
the Siberian plume would necessarilybe unlike that
inferredfor any other mantleplume.
In contrastto Sharma et al. [ 1991], Woodenet al. [ 1993]
proposedthat the compositionsof most Siberian Traps
lavaswere significantlymodified from that of their initial
plume source, predominantly through assimilation of
continentalcrust in large, periodically replenishedand
tappedcrustalmagmachambers.The modelingof O'Hara
LASSITER AND DEPAOLO 345
and Mathews [1981] suggeststhat in such a system
signaturesof crustal assimilationmay become decoupled
from those of fractional crystallization as the magma
approachesa steady-statemajor element composition.
However, although the major element homogeneityof
most Putoranaand upper Noril'sk lavas, combinedwith
their evolved nature, does require some process of
homogenizationsubsequentto melt generation,this does
not directly resolvewhethercontaminationoccurredin the
lithosphericmantle or the crust. Below, we outlineseveral
observationsthat suggesta lithospheric-mantle origin for
many of thesefeaturesis more likely.
Woodenet al. [ 1993] proposedthat crustalprocessingof
Siberian magmas took place in large, quasi-steady-state
magma chambers, in part to explain the apparent
decoupling of fractional crystallization from crustal
assimilation. However, Os isotopedata for picrites from
the lower Noril'sk section suggest that any crustal
contaminationmust have occurred prior to significant
fractionalcrystallization. Most researchersconcurthat the
Gudchikhinsky picrites, which have t•Nd • -3-4 and
negligibleHFSE anomalies,likely reflect the composition
of the Siberian plume [e.g., Woodenet al., 1993]. In
contrast, the Tuklonsky picrites have t•Nd• -2 and
pronouncedHFSE depletions,yet they possessessentially
identical¾osto the Gudchikhinskypicrites [Horan et al.,
1995]. Althoughminor isotopicdifferencesexistbetween
the Tuklonsky picrites and basalts, their similar trace
elementpatternssuggestsimilar contaminationhistories
for both. Therefore,if the Tuklonskybasaltswere derived
from picritic parental liquids similar to the Tuklonsky
picrites, then contaminationmust have occurredbefore
significant fractional crystallization of these picritic
parents.However, Woodenet al. [ 1993] suggested that the
Tuklonsky picrites contain cumulusolivine and are thus
derivedfrom more evolved, low MgO (and thereforelow
Os) liquids similar to the Tuklonskybasalts. However, if
this were the case,then the similarity of Os isotopesin the
Gudchikhinsky and Tuklonsky picrites again precludes
significantmixing of radiogenic continentalcrust with
these"parental"basalticmelts, becausethe low Os content
of evolved basalts would make Os isotopes highly
sensitive to such contamination. Thus, regardlessof
whether the Tuklonsky basalts are derived from picritic
melts or the Tuklonsky picrites are derived from basaltic
melts, the "continental" isotopic and trace element
signaturesin theselavas must have been impartedbefore
the basaltsunderwentsignificantfractionalcrystallization.
The steady-statemagma chamber model also cannot
easily explain the apparent decoupling between
incompatibletrace element ratios that we expect to
346 PLUME/LITHOSPHERE INTERACTION IN FLOOD BASALTS

6
correlated if these ratios are controlled by crustal ' I ' . I ' I ' I ' I ' I ' I '

assimilation. For example, the Ta depletionsin most Crustal

o Karoo
Picrites
SiberianTraps lavas were taken by Woodenet al. [1993] 5
UC contamination
o• • Nadezh":':'•'••••••
'""-"
'•'"'"'"
'"
as evidence for crustal assimilation. However, if this were of•rAø A •'"':•••••:-'"'•/
................
the case,we would expecta positivecorrelationbetween
La/Ta and La/Sm, becauseupper continentalcrust or O
/• :...••
......
,,!•i::,i,?i'iiii:":'
:i:i:?i'?!
......
moderate-degreepartial melts of the lower continental
crustwill in generalbe enrichedin light REE. Figure 7 Gudchikhinsk ky
picrites
shows La/Ta-La/Sm variations for lavas of the Noril'sk
picrites
•••'••••.
..................................
"-.':"•••••.• --
region. SomeNoril'sk lavasdo possess the high La/Sm,
high La/Ta values predictedfor crustalassimilation. In
particular, Nadezhdinsky suite lavas, which most
• oo••me/CLM
•
O (•'"":'::'
UpperNoril'sk
, , , , ,(•9dP,utø•?
2 , ,m• g, , ,
researchersagree are crustally contaminated,possess 10 20 30 40 50
highly elevatedLa/Sm and high La/Ta. However, most La/Ta
upper Noril'sk lavas possessmodestLa/Sm values and Figure 7. La/Ta-La/Sm variationsin SiberianTrapslavas.Most
high La/Ta values. Significantly,the Tuklonskypicrites crustalrocks possesselevatedLa/Sm ratios, and partial melting
possesseven higher La/Ta values than the bulk of the will further increase LadSm. Therefore, crustal contamination
upper Noril'sk sectionbut only slightly higher La/Sm. will increaseLa/Sm as well asLa/Ta. The very high La/Ta ratios
Theselavasclearly assimilateda componentvery different of the Tuklonsky picrites combinedwith only modestLa/Sm
than that containedin the Nadezhdinskylavas. The La/Ta enrichmentssuggestthat these picrites have not experienced
and La/Sm values of most upper-sequencelavas are significant crustal contamination,but instead contain a CLM-
intermediatebetweenthose of the Tuklonskypicritesand derived, HFSE-depleted component. Lavas with high La/Sm
(e.g., La/Sm > 3), which likely have assimilatedcrustalmaterial,
the Gudchikhinskypicrites,as are the Nd- and Pb-isotopic
are restrictedto the early phasesof volcanismin the Noril'sk
ratios. Therefore,the bulk of the SiberianTraps (=upper region,and are alsocharacterized by elevated$180 [Das Sharma
Noril'sk + Putorana) may have been generated from et al., 1994]. Lavas from the upper part of the Noril'sk section
mixturesof primary melts similar to the Gudchikhinsky havenormal$180, and La/Ta-La/Sm valuesintermediatebetween
andthe Tuklonskypicrites. those of the Tuklonsky and the Gudchikhinskypicrites. These
Assimilationof upper continentalcrustcannotproduce lavas probablyresult from mixing of plume- and CLM-derived
the La/Sm-La/Ta variations in the Tuklonsky and upper- melts.

sequencelavasbecauseof the high La/Sm in uppercrustal

rocks(Figure 7) [e.g., Taylor and McLennan,1985]. It is
possiblethatbulk assimilation of sometypesof low La/Sm
mafic lower crustcould explainthis correlation,although
the combinationof high La/Ta and low La/Sm in lower
crustalxenolithsis quiterare. However,in the morelikely
scenariowhere plume-derivedlavas assimilatedpartial
melts of the lower crust, the increasein La/Sm produced
from partial melting of granuliticor amphiboliticsources
againprecludes suchassimilation asa meansto accountfor
the Tuklonsky lavas. Finally, we note that other flood
basalt sequencesthat are clearly crustallycontaminated,
such as the Grande Ronde suite of the Columbia River
basalts or the Bushe in the Deccan, do not have the low
La/Sm, high La/Ta values found in the Tuklonsky. We
therefore considerthat these correlationsmay reflect the
incorporationof a HFSE-depletedmelt derived from the
CLM.
Appropriate sourcesfor the low La/Sm-high La/Ta
Tuklonskypicrites are also rare in CLM-derived mantle
xenoliths. However, the CLM-derived Karoopicriteshave
many trace element similarities with the Tuklonsky
picrites. Figure 7 shows that the Karoo picrites have
similarLa/Ta but higher La/Sm valuesthanthe Tuklonsky
picrites and upper Noril'sk lavas. However, ratios of
highlyto moderatelyincompatibleelementssuchasLa/Sm
are strongly influenced by the extent of partial melting.
Therefore, we can envision a scenariowherein larger
extentsof partial melting of a sourcesimilarto the CLM
sourceof the Karoo picrites,perhapsaccompanied by the
removalof garnetas a residualphase,couldproducemelts
similarto the Tuklonskyserieslavas.
Oxygen isotopic data also do not support a crustal
assimilationmodel for the Tuklonskyand upper-sequence
lavas. Most SiberianTraps lavas,includingsamplesfrom
the Gudchikhinsky and Tuklonsky picrites, the upper
Noril'sk, and the Putorana, have b•80 values within the
rangeof or only slightlyhigherthan valuesfoundin fresh
MORB glasses($•80 < 6.1, with few exceptions)[Das
Sharma et al., 1994]. In contrast, lavas from the
Nadezhdinsky and lvakinsky suites have b•80 > 6.9,
suggestingthat these lavas have assimilatedhigh b180
continentalcrust. Both these suitesare characterized
 LASSITER AND DEPAOLO 347

high La/Sm and La/Ta ratios, which are also consistent ß I ' • I ' I ' I ' I ' 1

with assimilationof high-La/Sm continentalcrust. The Plume

ß melts
normalmantle$•80 found in the Tuklonskyandmostother ß
SiberianTraps lavas, in contrast,suggeststhat theselavas ß

have not assimilateda significantamount of high o CLM

crustal material. Although mafic lower crustal xenoliths

have lower average $•80 than metasediments or average ENd o
O• •øaøo
o Crustal
upper continentalcrust [e.g., Fowler and Harmon, 1990], o assimilation

mafic lower crustalcomponentsalsohave lesspronounced

HFSE depletions,thus requiring greater addition of such -10

material to account for the high La/Ta values in the

Tuklonsky picrites. The required combination of low ' I ' I ' I ' I ' I ' I * I
La/Sm, normalmantle $•80, but high La/Ta is quite rare in b
crustal rocks.
Correlations between major elements and isotopic
compositionsin Siberian Traps lavas are also difficult to
reconcile with crustal assimilation models and instead
suggest the presence of a CLM-derived, Fe-poor •Nd2 o o
component. Values of FeO and end are correlated in
oß o o
Siberian Traps lavas (Figure 8a). This correlation is
similar to that observedin picritesfrom the Karoo flood
basaltsbetween eNa,¾Os,and Fe203. As we discussed ß
ß
earlier, the associationof low ¾Osand low Fe203 in the -4 , I a I • I • I , I , I • I ,
Karoo picrites indicatesthat melts derived from the (low 8 10 12 14 16

¾Os)CLM had lower Fe than melts derived from the
underlyingplume [Ellam et al., 1992].
The correlation between FeO and eNain the Siberian
Traps is unlikely to reflect the shallow-levelevolution•of
these lavas. For example, because average upper
continental crust has low Fe (but also low Mg) [e.g.,
Taylor and McLennan, 1985], the observed correlation
between FeO and endcould in principle be producedby
assimilationof Fe-poor, loW-eNdupper continentalcrust.
However, upper continentalcrustis characterizedby high
La/Sm and so is precludedby the La/Sm-La/Ta variations
in the Tuklonsky and upper-sequencelavas as discussed
above. Figure 8b showsthe samecorrelationbetweenFeO
and end after lavas from the crustally contaminated
Nadezhdinskysuite and otherhigh La/Sm lavashave been
removed. Although bulk assimilationof lower crust is
permissible,lower crustis more mafic than uppercrustand
hasa higherFe content[e.g., Taylor and McLennan,1985;
Rudnick and Taylor, 1987]. Therefore, a very large
amount of assimilation of lower crust is required to
producethe FeO-eNdtrend in Figure 8b. Furthermore,
simple assimilationof low Mg, low Fe continentalcrust
shouldproducea positivecorrelationbetweenFe and Mg,
which is not observed. Although it is possiblethat some
convolutionof crustal assimilationcoupled with magma
chamberprocessingcould accountfor this correlation,a
simpler explanation is that these variationsresult from
FeOT (wt%)
Figure 8. a. PositivecorrelationbetweenFeOT and ENdin the
SiberianTraps. b. FeOT-•Ndcorrelationafter removalof lavas
from the Nadezhdinskysuite and other high La/Sm lavas. High
5]80 and La/Sm values in Nadezhdinskysuite lavas are
consistentwith crustalcontamination,but Tuklonsky lavashave
normalmantle5180 andlow La/Sm. Symbolsareasin Figure7.
See text for discussion.
mixing of Fe-poor, CLM-derived melts with Fe-rich,
plume-derivedmelts similar to that observedfor the Karoo
picrites.
If the correlationbetween FeO and endin the Siberian
Traps lavas doesreflect mixing of Fe-poormelts from the
CLM with Fe-rich plume-derivedmelts, then a substantial
massfraction of CLM-derived melt must be presentin the
more contaminated,low endlavas. Assimilation of small
amounts of highly incompatible-element-enriched melts
suchas lamproitesinto tholeiiticmagmascannotreproduce
the observedcorrelation. This has importantimplications
for the plume impact model of flood basaltgenesis,which
predictslittle or no melt generationfrom the CLM [Arndt
and Christensen,1992; Farnetani and Richards, 1994]. It
is therefore importantto assessthe averagemelt fraction
derived from the lithosphererather than the plume. The
Putorana formations represent-90 vol% of the entire
Siberian Traps province [Sharma et al., 1991] and
348 PLUME/LITHOSPHERE INTERACTION
-1000
• -2000 ß
ß
-3000
ß
eee ee
,•.e• e,•1• ßI
I-., , I , I ,
0.2 0.4 0.6 0.8 1.0
(CaO/AI203)
Figure 9. The decrease in Sm/Yb and increasein CaO/AI203with increasing stratigraphicheightobservedin lavas
fromtheNoril'skregionareconsistent witha temporalincreasein thedegreeof partialmeltingandremovalof garnet
as a residualphasewithin the sourceregion. Thesechangesin the meltingenvironmentwithin the mantleare
temporallycorrelatedwith a generaldeclinein 87Sr/S6Sr(withthe exception
suggestingthatthe isotopicvariations
arealsorelatedto processesoccurring
to plumemelting. Data arefrom Woodenet al. [ 1993].
providea good estimateof the averagecompositionof the
SiberianTraps. Furthermore,the Putoranalavasgenerally
plot near the high FeO-high sNa,"plume"-derivedend of
the array in Figure 8. The sNavaluesof averagePutorana
lavas (•+2) are only slightly lower than e•a valuesof the
Gud-chikhinskypicrites (•+4), which we interpret as
reflectingthe uncontaminatedSiberianplume. Therefore,
even if this shift is due entirelyto assimilationof Fe-poor
CLM-derived melts with e•a values lessthan or equal to
•-2 (the value of mostTuklonskypicrites),the fractionof
CLM-derivedmelt in theselavasis still relativelysmall,
probablylessthan 10-20%. This is still a largeamountof
CLM-generated melt comparedwith predictionsof the
plume impact model. However, melt volumes of this
magnitude may be feasible if hydrous regions with
depressed solidiare involvedin melt generation.
The temporalevolutionof the SiberianTrapsprovidesa
final line of circumstantialevidencesupporting a modelof
plume/lithosphericmantle interactionrather than one of
crustalassimilation.Figure 9 illustratesthe stratigraphic
variation in lavas of the Noril'sk region. These lavas
IN FLOOD BASALTS
ß
8
ß
ß ß
ß ß
ß
ß
, I , I , I , I(Dd&li•e
e , I I
0 1 2 3 0.704 0.706 0.708 0.710
Sm/Yb (87Sr/86
Sr)T
of lavasfromtheNadezhdinsky suite),
withinthemantle,e.g.a shiftfromCLM
record a general decreasein Sm/Yb and an increasein
CaO/A1203 with increasingstratigraphicheight, which
most likely reflectsthe removal of garnet as a residual
phasein the source(s)of the lavas [e.g., Woodenet al.,
1993] and an increase in melt fraction. The decreasein
Sm/Yb, in particular,is unlikely to reflect changesin the
amount or kind of crustal assimilation, and so this
progression reflectsa changein the meltingenvironment.
Over the same interval, however, there is a general
accompanying decline in tracers of "continental"
contamination, suchas high La/Ta or 87Sr/86Sr.Although
thesetwo trendsdo not entirelymirror one another(e.g.,
the Gudchikhinsky picrites, which have the most
pronouncedgarnet signatureof any SiberianTraps lavas,
are also the leastcontam-inated),the overall correlationis
quiteclear. Becauseone of thesetrendsreflectsa change
in melting conditionswithin the mantle, this correlation
suggests that the accompanyingdeclinein the strengthof
continentalsignaturesalso reflects a changeoccurring
within the mantle, e.g., a decreasein the amountof melt
generatedfrom the CLM relative to melt generated

the underlying plume. The fact that the inferred CLM-
derivedcomponentdecreases at the sametime the apparent
depth of melt generationdecreasesfurther suggeststhat
this shift was brought about in part by the mechanical
erosionof the CLM, which allowedthe Siberianplume to
ascend to shallower levels.
In summary,no single piece of evidenceprecludesthe
possibilitythat the compositionsof the Tuklonskypicrites
or the volumetrically dominant Putorana and upper
Noril'sk lavas were controlled by assimilationof very
specificcrustalcomponents.However,in combinationthe
availabledata do not supportsucha model. In contrast,
melt generation from a sediment-enrichedlithospheric-
mantlecomponentsimilar in many respectsto the source
of the Karoo picrites and mixing of plume- and
lithosphere-derived meltsprovidea simplermechanismfor
producing most of the isotopic and trace element
characteristics of the Siberian Traps. Crustal
contaminatiom which significantly altered the
compositions of someearly lavas(e.g., the Nadezhdinsky
suite),was primarily of secondaryimportanceduringthe
main stagesof volcanism.
4. COMMON CHARACTERISTICS OF FLOOD
BASALTS: TOWARDS A GENERAL MODEL
4.1. TemporalEvolution
The temporalincreasein melt fractionand decreasein
the depthof melting inferredfrom major andtraceelement
trendsin the SiberianTraps is a commonfeatureof many
CFBs. For example, Kerr [1994] and Fram and Lesher
[1993] inferred a temporal progressiontowards higher
degreesof partialmeltinganddecreasing depthof melting
for basalts from the North Atlantic Tertiary Province.
Basaltseruptedearly in the formationof the North Atlantic
TertiaryProvincehavehigh Sm/Yb, requiringthe presence
of garnetas a residualphasein their sourceregion. Later
basaltslack this garnet signature. Kerr [1994] suggested
that progressiveerosion of the lithospherebeneaththe
North Atlantic Tertiary Province allowed the Iceland
plumeto ascendto shallowerlevels. The resultinghigher
degreesof melting and/ora shifttowardsmeltingof spinel
peridotiteled to the removalof garnetas a residualphase.
Stewart and Rogers [1996] inferred a similar temporal
increase in melt fraction for lavas of the southern
Ethiopianrift and proposedthermomechanical erosionof
the lithosphere
by the Afar plume. Lavasat the baseof the
Keweenawan section of the North American midcontinent
rift also have significantlyhigher Sm/Yb ratiosthan most
lavashigher in the sequence,a patternthat is mirroredby
LASSITER AND DEPAOLO 349
other sequencesfrom the midcontinentrift [cf. Shirey et
al., 1994; Nicholson et al., 1997]. Again, this temporal
declinein Sm/Yb is likely the resultof decreasingdepthof
meltingwith time. Finally, a similartemporalprogression
has been suggestedfor the Deccan Traps. Peng and
Mahoney [1995] noted a north-southtrend of decreasing
ratiosof highly to moderatelyincompatibleelements(e.g.,
Nb/Y) in the westernDeccanTraps,which they suggested
might reflect progressivelithosphericthinning and an
increasein the extent of partial melting for the (younger)
southern Deccan relative to the (older) northern Deccan
lavas. Erosion of the Indian lithosphereduring Deccan
volcanism is consistent with geophysical studies that
indicate the modern Indian lithosphereis significantly
thinnerand hotterthan most othershields[e.g.,Negi et al.,
1986].
The temporal progressionfrom low to higher melt
fractions is observed both in flood basalts associated with
continental rifting (e.g., the North Atlantic Tertiary
Province)and in flood basaltsnot associatedwith rifting
(e.g., SiberianTraps) and is oftenaccompaniedby a switch
from continentalisotopic and trace element signaturesto
more plume-like signatures. As in the Siberian Traps,
theseobservationssuggestsignificantlithosphericerosion
during flood basalt volcanism, often accompaniedby
partial melting of the CLM. The inferred lithospheric
erosionoccursover time scalestoo shortto be explained
by simple conductive heating of the lithosphere.
Lithosphericerosionmay be acceleratedif partial melting
of fusible (hydrous?) domains within the lithosphere
reduces lithosphere viscosity and enhancesconvective
erosion by the plume head. This interpretation is
consistentwith the observationthat the chemicalsignatures
of contaminationare often,thoughnot always,strongestin
lavas eruptedearly within a flood basalt sequence(e.g.,
SiberianTraps [Woodenet al., 1993]; Wrangellia[Lassiter
et al., 1995a]) because progressive melting of the
lithosphereshoulddepleteany hydrousdomainspresent.
4.2. Major Element Trends
Both the Karoo flood basaltsand SiberianTraps record
the incorporation of an Fe-poor, HFSE-depleted
component. In the Karoo picrites,and most likely in the
SiberianTraps as well, this componentappearsto have
been derived from the lithospheric mantle. Similar
correlations between major element and isotopic
compositionare seen in other flood basalt provincesas
well. For example,the DeccanTraps also show evidence
for mixing of Fe-rich plume melts with Fe-poor CLM
melts. Figure 10 shows a correlationbetween FeOT
350 PLUME/LITHOSPHERE INTERACTION
lO
Deccan Traps
ß from the CLM. Hergt et al. [ 1991] notedthat lavasfrom
[] ß []
-5
ENd ß E:I
ßDD ßEl []
-10
-15
Bushe/lower
formations
-25 , I , I , I
8 10 12
FeOx (wt%)
Figure 10. FeOT-gNd in the Deccan Traps.
contaminated lavas from the Bushe and other early Deccan
formations
(opensymbols)havelowergNdfor a givenFeOT than
lavas eruptedlater in the sequence. This shift is similar to that
observedfor lavasfrom the crustallycontaminated Nadezhdinsky
suite in the SiberianTraps. However, the correlationbetween
FeOT and [:Ndpersistswhen theseformationsare removed,and
does not appear to depend upon the extent of fractional
crystallization.
[•Ndin Deccan lavas similar to what we observefor the
SiberianTraps. As with the SiberianTraps,we mustagain
assessthe degree to which the observed FeOT-gNdtrend
reflects the effects of fractional crystallizationand/or
crustalassimilationrather than variationsin primary melt
composition. In the absenceof Fe-Ti oxide precipitation,
fractionalcrystallizationwill usuallyincreaseFe contentin
basalticmagmas. An unusualfeatureof the Deccanlavas
is that the least fractionated lavas tend to be the most
contaminated [e.g., Mahoney, 1988]. Therefore, the
correlationbetweenFeOT and [•Ndcouldsimplyreflectthis
fact. However, Deccan lavas spanningonly a narrow
range in MgO contentdisplaythe sametrend as the entire
data set, regardlessof whether one focusessolely upon
high- or low-MgO samples. Crustalcontaminationalso
does not appear sufficient to explain the observed
correlation. Crustal contamination is particularly
pronouncedin lavas of the Busheand other early Deccan
formations [Cox and Hawkesworth, 1985; Peng et al.,
1994]. These early Deccan lavas have consistentlylower
[•Nd
for a given FeOTthan mostotherDeccanlavas(Figure
10), suggestingthat crustal contaminationhad a more
pronouncedeffect on isotopicandtraceelementratiosthan
on major element composition. However, when these
lavas are removed, the correlationbetweenFeOT and [•Nd
persists.
IN FLOOD BASALTS
Many flood basaltsrelatedto the breakupof Gondwana
appearto contain an even larger fraction of melt derived
the Parantiand the Ferrar have lower Fe8 than typical
MORB, and they arguedthat thesemelts were generated
DE3
from previouslymelt-depleted,but subsequently hydrated,
lithosphericmantle. Turner and Hawkesworth [1995]
reached similar conclusions using major element
regressionssimilar to those we have utilized in our
discussion of melt generation. Supportfor a lithospheric
mantle sourcefor the Parantilavas comesfrom isotopic
variationsin potassicrocks from southernBrazil. Many
,
14 16 Brazilian potassic lavas have Sr- and Nd-isotopic
characteristicssimilar to the so-called high-Ti Paranti
[Gibson et al., 1995], yet some of these lavas have
Crustally unradiogenicOs isotopes,which suggestsa lithospheric
mantleorigin [Carlsonet al., 1997]. Gibsonet al. [1995]
also noted an apparentspatial correlationbetweenhigh-
and low-Ti mafic potassic volcanic rocks and the
occurrenceof high- and low-Ti flood basalts,suggesting
thatbothhigh- and low-Ti Gondwananbasaltsmay contain
a significantCLM-derived component.RecentOs isotope
analysesof Ferrar flood basaltsalso supporta lithospheric
mantle origin over crustalcontaminationmodelsfor these
basalts[Molzahn et al., 1996].
There is less chemical evidence from the Gondwanan
flood basaltsfor substantialmelt generationwithin sub-
lithospheric plumes than for most other flood basalt
provinces. This lack of evidence for plume-generated
meltsmay reflect a greaterwater contentandthusa lower
soliduswithin the pre-rifting Gondwananlithosphereas a
result of protractedsubductionalong the supercontinent's
margins or may reflect a lower than average excess
potentialtemperaturefor the Tristanand otherGondwanan
plumespresumablyresponsiblefor triggeringlithospheric
melting. In any case, melt generationfrom the CLM
duringflood basaltgenesisappearsto be a commonfeature
of many CFBs, a fact that appearsat oddswith traditional
anhydrousplume/lithosphere interactionmodels.
5. CONCLUSIONS
The interplay of plume, CLM, and crustal processes
affectingthe chemicaland isotopiccompositions of flood
basaltsis a complex one that varies significantlyfrom
provinceto province. However,the precedingdiscussion
has highlightedseveralcommonfeaturesin the evolution
of many continental and oceanic flood basalts. Our
interpretations
of thesefeaturesare summarizedbelow.
(1) Lithosphere thickness controls flood

compositionby limiting adiabaticascentof sublithospheric
plumes. The presenceof this "lithosphericsignature"in
CFBs, but not in OFBs, indicatesthat melting initiated
within sublithospheric
plumesandthat substantial thinning
of the lithospherecausedby extensionor protractedin-
cubation are not necessary prerequisites for flood
volcanism. However, the observation that OIBs are
generatedfrom even smallerextentsof meltingthan CFBs
suggeststhat somethinning of continentallithospheredoes
occur.
(2) Melting of lithosphericmantleand mixing of CLM-
andplume-derivedmeltscan explainthe isotopicandtrace
elementtrendsobservedin many CFB suitesincludingthe
bulk of the Siberian Traps. Significantmelting of the
lithospheric mantle is in conflict with anhydrous
plume/lithosphere interaction models, and therefore
suggestsmelting of hydrousregionswithin the CLM, as
proposed by Gallagher and Hawkesworth [1992].
However,CLM-derived meltsare generallymore prevalent
during the early phases of volcanism, with later lavas
containinga greater proportionof plume-derivedmelts.
On average,the total proportionof melt generatedfrom the
CLM is less than 10-20% of that generated from
sublithosphericsources.
(3) Averagedepth of melt generationdecreases while
melt fraction increasesduring the courseof volcanismfor
severalCFBs. Furthermore,this patternis observedboth
in CFBs associatedwith extension or rifting (e.g., the
North Atlantic Tertiary Province) and where extensionis
not observed(e.g., SiberianTraps). Combinedwith the
shortdurationof volcanismmeasuredfor many CFBs, this
andthe accompanyingdeclinein fractionof CLM-derived
melt suggest progressive mechanical erosion of the
lithosphere, possibly triggered by partial melting of
discrete hydrous domains within the CLM and an
accompanying decreasein lithosphereviscosity.
The plume initiation model for flood basaltgenesisis
generally consistentwith our observations,particularly
regardingthe depth of initial melt generationand extentof
meltingin continentaland oceanicflood basalts.However,
the fraction of CLM-derived melt inferred from
analysis exceeds predictions of anhydrous
plume/lithosphere
interactionmodels. More sophisticated
models that consider compositionaldifferencesbetween
the CLM and sub-lithosphericplumes are necessaryto
accountfor thesediscrepancies.
Unambiguous differentiation of crustal and CLM-
derivedcomponents
in CFBshasprovento be exceedingly
LASSITER AND DEPAOLO 351
difficult using traditional incompatibletrace element and
isotopic arguments, as any survey of the flood basalt
literaturewill attest. We have attemptedto illustratethe
meansby which a combinationof trace elementand major
element arguments can help differentiate the various
processeswithin the plume, lithosphericmantle, and crust
that operate simultaneouslyduring flood basalt genesis.
However, future progress in this area will require
additional data from isotopic systemscapable of clearly
distinguishingcrustal and CLM components. Recent
studiesexamining oxygen and osmium isotopesin flood
basalts offer the best hope that many of the difficult
questionsconcerning the role of continental crust and
lithosphericmantle in flood basalt genesis may soon
timallybe resolved.
Acknowledgments. We thank N. Arndt and an anonymous
reviewer for their comments, R. Carlson, S. Shirey, and A.
Brandonfor useful suggestions and discussions, and J. Mahoney
for his editorial patience. This material is based upon work
supportedunder a National ScienceFoundationgraduateresearch
fellowship,The Berkeley Fellowship,NSF grant EAR 9304419,
an NSF postdoctoralfellowship, and by the CarnegieInstitution
of Washington.
REFERENCES
Aitken, B. G., and L. M. Etchevera,Petrology and geochemistry
of komatiites and tholeiites from Gorgona Island, Columbia,
Contrib. Mineral. Petrol., 86, 94-105, 1984.
Albarede, F., and V. Tamagnan, Modeling the recent
geochemicalevolution of the Piton de la Fornaise volcano,
R6union Island, 1931-1986, d. Petrol., 29, 997-1030, 1988.
Amdt, N. T., and U. Christensen,The role of lithosphericmantle
in continental flood volcanism: Thermal and geochemical
constraints,d. Geophys.Res., 97, 10,967-10,981, 1992.
Arndt, N. T., G. K. Czamanske, J. L. Wooden, and V. A.
Fedorenko, Mantle and crustal contributions to continental
flood volcanism,Tectonophysics, 223, 39-52, 1993.
Baker, M. B., and E. M. Stolper,Determiningthe compositionof
high-pressure mantle melts using diamond aggregates,
Geochim. Cosmochim. Acta., 58, 2811-2827, 1994.
Basu, A. R., P. R. Renne, D. K. DasGupta, F. Teichmann, and
R.J. Poreda,Early and late alkali igneouspulsesand a high-
3He plume origin for the Deccanflood basalts,Science,261,
902-906, 1993.
our Basu, A. R., R. J. Poreda, P. R. Renne, F. Teichmann, Y.R.
Vasilev, N.V. Sobolev, and B.D. Turrin, High-3He plume
origin and temporal-spatialevolution of the Siberian flood
basalts,Science, 269, 822-825, 1995.
Brandon, A.D., and G. G. Goles, A Miocene subcontinental
plumein the PacificNorthwest: Geochemicalevidence,Earth
Planet. Sci. Lett., 88, 273-283, 1988.
Brandon, A.D., and G. G. Goles, Assessingsubcontinental
lithosphericmantlesourcesfor basalts:Neogenevolcanism
352 PLUME/LITHOSPHERE INTERACTION IN FLOOD
the Pacific Northwest, USA as a test case, Contrib. Mineral.
Petrol., 121,364-379, 1995.
Carlson,R. W., G. W. Lugmair, andJ. D. MacDougall,Columbia
River volcanism: The question of mantle heterogeneityor
crustal contamination, Geochim. Cosmochim.Acta, 45, 2483-
2499, 1981o
Carlson, Ro W., Isotopic constraintson Columbia River flood
basalt genesisand the nature of the subcontinentalmantle,
Geochim. Cosmochim.Acta., 48, 2357-2372, 1984.
Carlson, R. W., and A. J. Irving, Depletion and enrichment
historyof subcontinental
lithosphericmantle: An Os, Sr, Nd,
and Pb isotopic study of ultramafic xenoliths from the
northwesternWyoming craton, Earth Planet. Sci. Lett., 110,
99-119, 1994.
Carlson,R. W., S. Esperan•a,and D. P. Svisero,Chemicaland
Os isotopicstudyof Cretaceouspotassicrocksfrom southern
Brazil, Contrib. Mineral. Petrol., in press,1997.
Castillo, P. R., R. W. Carlson, and R. Batiza, Origin of Nauru
basin igneouscomplex: Sr, Nd, and Pb isotopeand REE
constraints,Earth Planet. Sci. Lett., 103, 200-213, 1991.
Chapman,D. S., and H. N. Pollack, Regionalgeothermsand
lithosphericthickness,Geology,5, 265-268, 1977.
Chesley,J. T., J. Ruiz, and P. R. Hooper, Crust-mantlemixing:
implicationsbasedon the Re-Os isotopesystematics of the
Columbia River Basalt Group, EOS Trans. AGU, 77, 832,
1996.
Clague,D. A., Petrologyof basalticand gabbroicrocksdredged
from the Danger Island troughs,Manihiki Plateau,Init. Repts.
Deep Sea Drill. Proj., 33, 891-911, 1976.
Condomines,M., K. Gr6nvold, P. J. Hooker, K. Muehlenbachs,
R. K. O'Nions, N. Oskarsson,and E. R. Oxburgh, Helium,
oxygen, strontium,and neodymiumisotopicrelationshipsin
Icelandic volcanics, Earth Planet. Sci. Lett., 66, 125-136,
1983.
Cox, K. G., and C. J. Hawkesworth,Geochemicalstratigraphyof
the Deccan Traps at Mahabaleshwar,WesternGhats, India,
with implicationsfor open system magmaticprocesses,d.
Petrol., 26, 355-377, 1985.
Das Sharma,S., D. J. Patil, R. Murari, G. Gopalan,and G. V.
Nesterenko,Oxygen isotopesystematics of Siberianbasalts,d.
Geol. Soc. India, 44, 327-330, 1994.
DePaolo, D. J. and G. J. Wasserburg,Neodymium isotopesin
flood basalts t¾om the Siberian Plattbrm and inferences about
their mantle sources,Proc. Natl. Acad. Sci., 76, 3056-3060,
1979.
Dodson,A., D. J. DePaolo, and B. M. Kennedy,Noble gassesin
the Columbia River Basalt, Eos Trans. AGU, 77, 288, 1996.
Donnelly, T. W., W. Melson, R. Kay, and J. J. W. Rogers,
Basaltsand doleritesof late Cretaceousage from the central
Caribbean,Init. Repts. Deep Sea Drill. Proj., 15, 989-1011,
1973.
Dosso, L., H. Bougault, and J. -L. Joron, Geochemical
morphology of the north Mid-Atlantic Ridge, 100-240N:
Trace element-isotopecomplementarity,Earth Planet. Sci.
Lett., 120, 443-462, 1993.
Eiler, J. M., J. W. Valley, and E. M. Stolper,Oxygen isotope
BASALTS
ratiosin olivinesfrom the Hawaii ScientificDrillingProject,d.
Geophys.Res., 101, 11807-11814, 1996.
Ellam, R. M., Lithosphericthicknessas a control on basalt
geochemistry,Geology,20, 153-156, 1992.
Ellam, R. M., andK. G. Cox, A Proterozoiclithospheric
source
for Karoo magmatism:Evidencefrom the Nuanetsipicrites,
Earth Planet. Sci. Lett., 92, 207-218, 1989.
Ellam, R. M., andK. G. Cox, An interpretation of Karoopicrite
basaltsin termsof interactionbetweenasthenospheric magmas
and the mantlelithosphere,Earth Planet. Sci. Lett., 105, 330-
342, 1991.
Ellam, R. M., R. W. Carlson, and S. B. Shirey, Evidencefrom
Re-Os isotopesfor plume-lithospheremixing in Karoo flood
basaltgenesis,Nature, 359, 718-721, 1992.
Falloon, T. J., and D. H. Green, The solidusof carbonated,fertile
peridotite,Earth Planet. Sci. Lett., 94, 364-370, 1989.
Falloon,T. J., D. H. Green,andA. L. Jaques,Refractorymagmas
in back-arcbasinsettings,Mineral. Mag., 58A, 263-264, 1994.
Fametani,C. G., andM. A. Richards,Numericalinvestigations of
the mantle plume initiation model for flood basalt events,d.
Geophys.Res.,99, 13813-13834, 1994.
Fisk, M. R., B. G. J. Upton, and C. E. Ford, Geochemicaland
experimental study of the genesis of magmas of R6union
Island,Indian ocean,d. Geophys.Res.,93, 4933-4950, 1988.
Floyd, P. A., Petrologyand geochemistryof oceanicintraplate
sheet-flowbasalts,Nauru basin,Deep Sea Drilling Projectleg
89, Init. Repts.Deep Sea Drill. Proj., 89, 471-497, 1986.
Fodor, R. V., C. Corwin, and A. N. Sial, Crustalsignaturesin the
SerraGeral flood-basaltprovince,southernBrazil: O- and Sr-
isotopeevidence,Geology,13,763-765, 1985.
Fowler, M., and R. S. Harmon, The oxygenisotopecomposition
of lower crustal granulite xenoliths, in Petrrology and
Geochemistryof Granulites, edited by D. Vielzeuf, pp. 493-
506, NATO/ASI Series, Vol. X, 1990.
Fram, M. S., and C. E. Lesher, Geochemical constraints on
mantle melting during creation of the North Atlantic basin,
Nature, 363, 712-715, 1993.
Gallagher,K., andC. Hawkesworth,Dehydrationmeltingandthe
generation of continental flood basalts, Nature, 258, 57-59,
1992.
Gibson, S. A., R. N. Thompson, A. P. Dicken, and O. H.
Leonardos,High-Ti and low-Ti potassicmagmas: Key to
plume-lithosphereinteractions and continental flood-basalt
genesis,Earth Planet. Sci. Lett., 136, 149-165, 1995.
Green,T. H., andN.J. Pearson,An experimentalstudyof Nb and
Ta partitioningbetweenTi-rich mineralsand silicateliquidsat
high pressureand temperature,Geochim. Cosmochim.Acta,
51, 55-62, 1987.
Griffiths, R. W., Thermalsin extremelyviscousfluids, including
the effectsof temperature-dependentviscosity,d. Fluid Mech.,
166, 115-138, 1986.
Griffiths, R. W., and I. H. Campbell, Stirring and structurein
mantle startingplumes, Earth Planet. Sci. Lett., 99, 66-78,
1990.
Harris, C., A.M. Wittingham, S.C. Milner, and R. A.
Armstrong, Oxygen isotope geochemistry of the

volcanic rocks of the Etendeka-Paranfiprovince: Source
constraints,Geology, 18, 1119-1121, 1990.
Hauri, E. H., Major-elementvariability in the Hawaiian mantle
plume,Nature, 382, 415-419, 1996.
Hemond, C., N. T. Arndt, U. Lichtenstein,and A. W. Hofmann,
The heterogeneous Icelandplume: Nd-Sr-O isotopesandtrace
elementconstraints,• Geophys.Res.,98, 15833-15850, 1993.
Hergt, J. M., D. W. Peate, and C. J. Hawkesworth, The
petrogenesisof Mesozoic Gondwanalow-Ti flood basalts,
Earth Planet. Sci. Lett., 105, 134-148, 1991.
Hirose, K., and I. Kushiro, Partial melting of dry peridotitcat
high pressures: determinationof compositionsof melts
segregated from peridotitcusingaggregates of diamond,Earth
Planet. Sci. Lea., 114, 477-489, 1993.
Hooper,P. R., and C. J. Hawkesworth, Isotopicandgeochemical
constraintson the origin and evolutionof the ColumbiaRiver
Basalt, • Petrol., 34, 1203-1246, 1993.
Horan, M. F., R. J. Walker, V. A. Fedorenko, and G. K.
Czmanske,Osmium and neodymiumisotopicconstraintson
the temporaland spatial evolutionof Siberianflood basalt
sources,Geochirn.Cosmochim.Acta, 59, 5159-5168, 1995.
Ito, E., W. M. White, and C. G6pel, The O, Sr, Nd, and Pb
isotopegeochemistryof MORB, Chem. Geol., 62, 157-176,
1987.
Jackson,E. D., K. E. Bargar, B. P. Fabbi, and C. Heropoulos,
Petrologyof the basalticrocksdrilled on leg 33 of the Deep
Sea Drilling Project, Init. Repts. Deep Sea Drill. Proj., 33,
571-630, 1976.
Jordan, T. H., Composition and development of the
subcontinentaltectosphere,Nature, 274, 544-548, 1978.
Kerr, A. C., Lithospheric thinning during the evolution of
continentallarge igneousprovinces: A casestudy from the
North Atlantic Tertiary province, Geology, 22, 1027-1030,
1994.
Kerr, A. C., J. Tarney, G. F. Marfiner, G. T. Klaver, A.D.
Saunders, and M. F. Thirwall, The geochemistry and
pertogenesisof the late-Cretaceouspicrites and basaltsof
Cura•o, NetherlandAntilles: a remnantof an oceanicplateau,
Contrib. Mineral. Petrol., 124, 29-43, 1996.
Klein, E. M., and C. H. Langmuir,Global correlationsof ocean
ridgebasaltchemistrywith axial depthandcrustalthickness,d.
Geophys.Res.,92, 8089-8115, 1987.
Kurz, M.D., T. C. Kenna, J. C. Lassiter, and D. J. DePaolo,
Helium isotopicevolutionof Mauna Kea volcano: first results
from the 1-km drill core,d. Geophys.Res., 101, 11781-11791,
1996.
Kyser, T. K., Stable isotopevariationsin the mantle, in Stable
Isotopes in High Temperature Geological Processes,Rev.
Mineral., 16, edited by J.W. Valley, H.P. Taylor, and J.R.
O'Neil, pp. 141-146, Mineralogical Society of America,
Chelsea, MI, 1986.
Langmuir, C. H., E. M. Klein, and T. Plank, Petrological
systematicsof mid-ocean ridge basalts: constraintson melt
generationbeneath ocean ridges, in Mantle Flow and Melt
Generationat Mid-Ocean Ridges,Geophys.Monogr Ser., vol.
71, editedby J.P. Morgan, D. K. Blackman,and J. M. Sinton,
LASSITER AND DEPAOLO 353
pp. 183-280, AGU, Washington,D.C., 1992.
Lassiter,J. C., D. J. DePaolo, and J. J. Mahoney, Geochemistry
of the Wrangellia flood basaltprovince: Implicationsfor the
role of continental and oceanic lithospherein flood basalt
genesis,d. Petrol., 36, 983-1009, 1995a.
Lassiter, J. C., M. T. Silk, M. A. Richards, D. J. DePaolo, C. G.
Farnetani,and R. A. Duncan, Constraintson the origin of the
Wrangellia flood basalt province: support for the plume
impactmodel of flood basaltformation,Eos Trans.AGU, 76,
587, 1995b.
Lightfoot,P. C., C. J. Hawkesworth,C. W. Devey, N. W. Rogers,
and PD W. C. Van Calsteren, Source and differentiation of
DeccanTrap lavas: Implicationsof geochemicaland mineral
chemicalvariations,d. Petrol., 31, 1165-1200, 1990a.
Lightfoot, Po C., A. J. Naldrett, N. S. Gorbachev,W. Doherty,
andVoA. Fedorenko,Geochemistryof the SiberianTrap of the
Noril'sk area, USSR, with implications for the relative
contributionsof crust and mantle to flood basaltmagmatism,
Contrib. Mineral. Petrol., 104, 631-644, 1990b.
Lightfoot, P. C., C. J. Hawkesworth,J. Hergt, A. J. Naldrett, N.
S. Gorbachev, V. A. Fedorenko, and W. Doherty,
Remobilization of the continental lithosphereby a mantle
plume: Major-, trace-, and Sr-, Nd-, and Pb-isotopeevidence
from picritic and tholeiitic lavas of the Noril'sk district,
SiberianTrap, Russia,Contrib. Mineral. Petrol., 114, 171-188,
1993.
Mahoney, J. J., An isotopic survey of Pacific ocean plateaus:
Implicationsfor their natureand origin, in Seamounts,Islands,
and Atolls, Geophys. Monogr. Ser, vol. 43, edited by B.
Keating, P. Fryer, R. Batiza, and G. Boehlert, pp. 207-220,
AGU, Washington,D.C., 1987.
Mahoney, J. J., Deccan Traps, in Continental Flood Basalts,
edited by J. D. Macdougall, pp. 151-194, Kluwer Academic
Publishers,Boston, 1988.
Mahoney, J. J., and K. J. Spencer, Isotopic evidence for the
origin of the Manihiki and OntongJavaplateaus,Earth Planet.
Sci. Lett., 104, 196-210, 1991.
Mahoney, J. J., J. D. Macdougall, G. W. Lugmair, A. V. Murali,
M. Sankar Das, and K. Gopalan, Origin of the Deccan Trap
flows at Mahabaleshwarinferred from Nd and Sr isotopicand
chemical evidence, Earth Planet. Sci. Lett., 60, 47-60, 1982.
Mahoney, J. J., M. Storey, R. A. Duncan, K. J. Spencer,and M.
Pringle, Geochemistryand age of the OntongJavaplateau,in
The Mesozoic Pacific: Geology, Tectonics,and Volcanism,
Geophys.Monogr. Ser., vol. 77, edited by M. Pringle, W.
Sager,W. Sliter, and S. Stein,pp. 233-261, AGU, Washington,
D.C., 1993.
Mattey, D., D. Lowry, and C. Macpherson, Oxygen isotope
compositionof mantle peridotitc,Earth Planet Sci. Lett., 128,
231-241, 1994.
McDonough, W. F., Constraints on the composition of the
continentallithosphericmantle, Earth Planet. Sci. Lett., 101,
1-18, 1990.
McDonough, W. F., Chemical and isotopic systematicsof
continental lithospheric mantle, Proc. 5th Intl. Kimberlite
Conf, 478-485,
354 PLUME/LITHOSPHERE INTERACTION IN FLOOD
McKenzie,D., and M. J. Bickle, The volumeandcompositionof
melt generatedby extensionof the lithosphere,d. Petrol., 29,
625-679, 1988.
McKenzie, D., and R. K. O'Nions, Partial melt distributionsfrom
inversionof rare earth element concentrations, d. Petrol., 32,
1021-1091, 1991.
McKenzie, D., and R. K. O'Nions, The sourceregionsof ocean
islandbasalts,d. Petrol., 36, 133-159, 1995.
Michael, P., Regionally distinctivesourcesof depletedMORB:
Evidence from trace elements and H20, Earth Planet. Sci.
Lett., 131, 301-320, 1995.
Mitchell, R. H., and S.C. Bergman,Petrologyof Lamproites,
447 pp., Plenum Press,New York, 1991.
Molzahn, M., L. Reisberg,G. W6rner, Os, Sr, Nd, Pb, O isotope
and trace element data from the Ferrar flood basalts,
Antarctica: evidence for an enriched subcontinental
lithospheresource, Earth Planet. Sci. Lett., 144, 529-546,
1996.
Morgan, W. J., Convectionplumesin the lower mantle,Nature,
230, 42-43, 1971o
Negi, J. G., O. P. Pandey,and P. K. Agrawal,Super-mobilityof
hot Indian lithosphere,Tectonophysics,
131, 147-156, 1986.
Nelson, D. O., Implications of oxygen-isotopedata and trace-
element modeling for a large-scalemixing model for the
ColumbiaRiver basalt,Geology,11,248-251, 1983.
Nelson, D. O., Geochemistryof the GrandeRonde basaltof the
ColumbiaRiver BasaltGroup;a reevaluationof sourcecontrol
and assimilation effects, in Volcanism and Tectonism in the
ColumbiaRiver Flood-BasaltProvince,Spec.Pap. 239, edited
by S. P. Reidel and P. R. Hooper, pp. 333-341, Geological
Societyof America, Boulder,CO, 1989.
Nicholson, S. W., S. B. Shirey, K. J. Schulz, and J. C. Green,
Evolution of 1.1 Ga midcontinent rift basalts: rift-wide
correlation and the interaction of multiple mantle sources
duringrift development,Can. d. Earth Sci. specialIGCP 336
vol., in press,1997.
Nickel, K. G., Phaseequilibriain the systemSiO2-MgO-A1203-
CaO-Cr203 (SMACCR) and their bearing on spinel/garnet
lherzolite relationships,Neuesdahrb. Miner. Abh., 155, 259-
287, 1986.
O'Hara, M. J., and R. E. Mathews, Geochemical evolution in an
advancing, periodically replenished, periodically tapped,
continuously fractionated magma chamber, d. Geol. Soc.
London, 138, 237-277, 1981.
Pearson,D. G., R. W. Carlson,S. B. Shirey,F. R. Boyd, andP.H.
Nixon, Stabilizationof Archaeanlithosphericmantle: A Re-
Os isotopestudy of peridotitexenolithsfrom the Kaapvaal
craton,Earth Planet. Sci. Lett., 134, 341-357, 1995.
Peng, Z. X., J. Mahoney, P. Hooper, C. Harris, and J. Beane, A
role for lower continental crust in flood basalt genesis?
Isotopic and incompatible element study of the lower six
formations of the western Deccan Traps, Geochim.
Cosmochim.Acta, 58, 267-288, 1994.
Peng, Z. X., and J. J. Mahoney, Drillhole lavas from the
northwesternDeccan Traps, and the evolution of R6union
hotspotmantle, Earth Planet. Sci. Lett., 134, 169-185, 1995.
Price, R. C., A. K. Kennedy,M. Riggs-Sneeringer,
and F. A.
BASALTS
Frey, Geochemistryof basaltsfrom the Indian Oceantriple
junction: Implicationsfor the generationand evolution of
Indian Ocean ridge basalts,Earth Planet. Sci. Lett., 78, 379-
396, 1986.
Prinzhofer,
A., E. Lewin,andC. J. All•gre, Stochastic
meltingof
the marblecakemantle: Evidencefrom localstudyof theEast
PacificRise at 12050'N, Earth Planet. Sci. Lett., 92, 189-206,
1989.
Rhodes,J. M., Geochemicalstratigraphyof lava flows sampled
by the Hawaii ScientificDrilling Project,d. Geophys.Res.,
101, 11729-11746, 1996.
Roeder, P. L., and R. F. Emslie, Olivine-liquidequilibrium,
Contrib. Mineral. Petrol., 29, 275-289, 1970.
Rudnick, R. L., and S. R. Taylor, The compositionand
petrogenesisof the lower crust: a xenolithstudy,d. Geophys.
Res., 92, 13981-14005, 1987.
Saunders,A.D., M. Storey, R. W. Kent, and M. J. Norry,
Consequences of plume-lithosphere
interaction,in Magmatism
and the Causesof ContinentalBreak-up,Spec.Publ. 68, edited
by B.C. Storey,ToAlabaster,andR. J. Pankhurst, pp. 41-60,
GeologicalSocietyof London, 1992o
Sen, G., RoHickey-Vargas,D. G. Waggoner,andF. Maurrasse,
Geochemistry of basalts from the Dumisseau Formation,
southernHaiti: implicationsfor the origin of the Caribbean
Sea crust, Earth Planet. Sci. Lett., 87, 423-437, 1988.
Sharma,M., A. R. Basu, and G. V. Nesterenko,Nd-Sr isotopes,
petrochemistry,and the origin of the Siberian flood basalts,
USSR, Geochim. Cosmochim.Acta, 55, 1183-1192, 1991.
Sharma,M., A. R. Basu, and G. V. Nesterenko,Temporal Sr-,
Nd-, and Pb-isotopicvariationsin the Siberianflood basalts:
Implications for the plume-source characteristics,Earth
Planet. Sci. Lett., 113, 365-381, 1992.
Shirey, S. B., Re-Os isotopiccompositions of Midcontinentrift
systempicrites: implicationsfor plume-lithosphere interaction
and enrichedmantle sources,Can. d. Earth Sci., in press,1997.
Shirey, S. B., K. W. Klewin, J. H. Berg, and R. W. Carlson,
Temporalchangesin the sourcesof flood basalts:Isotopicand
trace element evidence from the 1100 Ma old Keweenawan
Mamainse Point Formation, Ontario, Canada, Geochim.
Cosmochim.Acta, 58, 4475-4490, 1994.
Stewart, K., and N. Rogers, Mantle plume and lithosphere
contributionsto basaltsfrom southernEthiopia,Earth Planet.
Sci. Lett., 139, 195-211, 1996.
Sun, S.-s., and W. F, McDonough, Chemical and isotopic
systematicsof oceanic basalts: implications for mantle
composition and processes,in Magmatism in the Ocean
Basins,Spec. Publ. 42, edited by A.D. Saundersand M. J.
Norry, pp. 313-345, GeologicalSocietyof London,1989.
Taylor, S. R., and S. M. McLennan, The ContinentalCrust: Its
Composition and Evolution, 312 pp., Blackwell Scientific
Publications,London, 1985.
Thompson,R. N., P. T. Leat, A. P. Dickin, M. A. Morrison,G. L.
Hendry, and S. A. Gibson, Strongly potassicmafic magmas
from lithosphericmantle sourcesduring continentalextension
and heating: Evidence from Miocene minettesof northwest
Colorado, USA, Earth Planet. Sci. Lett., 98, 139-153, 1990.
Turner, S., and C. Hawkesworth, The nature of the
 LASSITER AND DEPAOLO 355

continental mantle: Constraints from the major-element F. Siems,Isotopicand trace-elementconstraintson mantleand

compositionof continentalflood basalts,Chem. Geol., 120, crustal contributions to Siberian continental flood basalts,
295-314, 1995. Noril'sk area, Siberia, Geochim. Cosmochim. Acta, 57, 3677-
Vandamme, D, and V. Courtillot, Latitudinal evolution of the 3704, 1993.
R6union hotspotdeducedfrom palcomagneticresultsof Leg Yang, H.-J., F. A. Frey, J. M. Rhodes, and M. O. Garcia,
115, Geophys.Res.Lett., 17, 1105-1108, 1990. Evolution of Mauna Kea volcano: Inferences from lava
Walker, R. J., R. W. Carlson, S. B. Shirey, and F. R. Boyd, Os, compositionsrecoveredfrom in the Hawaii Scientific Drilling
St, Nd, and Pb isotope systematicsof southern African Project,Or.Geophys.Res., 101, 11747-11767, 1996.
peridotitcxenoliths: Implicationsfor the chemicalevolution
of subcontinental mantle, Geochim. Cosmochim. Acta, 53,
1583-1595, 1989. JoC. Lassiterand D. J. DePaolo, Berkeley Center for Isotope
Wooden, J. L., G. K. Czmanske, V. A. Fedorenko, N. T. Arndt, Geochemistry, Department of Geology and Geophysics,
C. Chauvel, R. M. Bouse,B. S. W. King, R. J. Knight, and D. Universityof California, BerkeleyCA
 FloodBasaltsandMagmaticNi, Cu, andPGE SulphideMineralization'Comparative
Geochemistryof theNoril'sk(SiberianTraps)andWestGreenlandSequences
Peter
C.Lightfoot
•
Departmentof Earth Sciences,The Universityof Toronto,Toronto,Ontario, M5S 3B1, Canada
Chris J. Hawkesworth

Departmentof Earth Sciences,The Open University,Milton Keynes,MK7 6AA, UK

The geologicalsettingsof the continentalflood basaltsat Noril'sk in the northwestern

SiberianTrapsand in West Greenlandhave a numberof parallels.Theseinclude(1) a thick
(0-3.5 km) sequenceof tholeiitic and picritic lavas; (2) the inferred presenceof a mantle
plume or hot spot;(3) comagmaticintrusions,someof which are picritic; (4) faults which
penetratethe uppermantleand actedas conduitsfor magmatism;and (5) an epicontinental
settingwith sulphidicsediments. The intrusionsat Noril'sk areknownto hostgiantdeposits
(>555 milliontonneswith 2.7 wt% Ni, 3.9 wt% Cu, 3 ppmPt, 12 ppmPd) of sulphidemin-
eralization,whereasthosein WestGreenlandcontainonly smallknownshowingsof nickel
sulphidemineralization.The lavasat bothNoril'sk andWestGreenlandexhibitan empirical
relationbetweencrustalcontaminationand depletionin Ni and Cu, and at Noril'sk suchla-
vas are highly depletedin the platinumgroup elements(PGE). Different stylesof crustal
contaminationare recognized,but the observeddepletionin Ni, Cu, and the PGE is similar
irrespectiveof the natureof the contaminant.Moreover,the main trendsare controlledby
igneouscrustalcomponents that are likely to have had low S contents.Thus, we arguethat
initial sulphursaturationoccurredlargely in responseto the increasein SiO2 (typically 5
wt%) broughtaboutby 20-25% crustalcontamination, ratherthanto the additionof signifi-
cant quantitiesof crustal-derived S. At Noril'sk, minor- and trace-elementargumentsindi-
cate that the contaminatedlavas contain<0.5% of shallow-levelevaporite-richsediments,
and so althoughsuchsedimentsmay have contributedsignificantquantitiesof S, they were
not responsiblefor the crustalcontaminationthat is associatedwith the distinctiveNi, Cu,
and PGE depletionsin the Nadezhdinskylavas.Mass-balance considerations
suggestthat as
little as 1% of the S availablein the mantle-derivedmagmasand assimilatedevaporite-rich
sedimentsat Noril'sk is locked up in the known deposits.The sulphidesat Noril'sk have
unusually high metal contents,and so they must have interactedwith large volumes of
magma.In one modelthe intrusionsactedas open-system conduits,or chonoliths,in which
the sulphideswere continuallyupgradedas more magmapassedthrough.The earlier mag-
maswerestrippedof Ni, Cu andPGE, but asthe systemevolvedthe sulphidesbecamesatu-
rated and, in somecases,isolated,and so the later magmaswere progressivelylessdepleted
in metals.In anothermodelthe sulphidesformedat muchdeeperlevelsand were eraplaced
into the intrusionsas sulphidemagmas.Whichevermodel provesto be correct,the lavasat
Noril'sk are characterisedby a progressiveupward recovery in Ni, Cu, and PGE abun-
dances(over about700 m) as the degreeof contaminationfalls, but thosein West Greenland
are not. This may in turn haveimplicationsfor explorationin otherlargeigneousprovinces.
1. INTRODUCTION
1 Presentaddress:
Inco Limited,Field Exploration
Office,
Highway 17 West, CopperCliff, Ontario,P0M 1N0, Canada
Continentalflood basalts(CFB) representmajor mag-
LargeIgneousProvinces:Continental,Oceanic,andPlanetary
Flood Volcanism matic events,and someare associated
with importantsul-
GeophysicalMonograph100 phidemineralization:
for example,theNorils'k deposits
in
Copyright1997by theAmericanGeophysical
Union the Siberian Traps, the Duluth Complex in the Ke-
358 FLOOD BASALTS AND MINERALIZATION
weenawanmidcontinentriff, andthe Insizwadepositin the
Karoo [e.g., Naldrett and Lightfoot, 1993]. Different CFB
are characterised
radiogenic-isotope
by differentmajor- andtrace-elementand
compositions,andthereis an activede-
bateboth over the extentto which onemodelis applicable
to all CFB, and whether melting is triggeredby litho-
sphericextensionor the emplacement of deep-seated man-
tle plumes [e.g., Richardset al., 1989; Whiteand McKen-
zie, 1989, 1995; Gallagher and Hawkesworth, 1994;
Turner et al., 1996]. Most CFB have relatively high in-
ferred melt generationrates and were thereforeassociated
with mantle hot spots,but the relative contributionsfrom
mantleplumesand both the crustalandmantleportionsof
the continentallithospherevary from one CFB provinceto
another [e.g., Mahoney et al., 1982; Macdougall, 1988;
Hergt et al., 1991; Turner and Hawkesworth,1995]. Such
considerations are relevantto understanding the availabil-
ity of primitive S-undersaturated magma [Keays, 1995]
and the nature of the high-level processesgoverningthe
compositionof the eruptedmagmasand the Ni, Cu, and
platinumgroup elements(PGE) mineral potentialof any
comagmaticintrusions.
Crustalcontaminationof mantle-derived meltstypically
results in elevated SiO2 and lithophile-elementabun-
dances,more "enriched"radiogenicisotopecompositions,
and in some caseshighly contaminatedmagmasthat also
exhibit pronounceddepletionsin Ni, Cu, and the PGE
[Lightfoot et al., 1990, 1993, 1994; Naldrett et al., 1992,
1995; Briigmann et al., 1993; Woodenet al., 1993; Fe-
dorenko,1994; Hawkesworthet al., 1995]. The recogni-
tion that in someareasit wasthe lavaswith strongcrustal
contaminationsignaturesthat were preferentiallydepleted
in Ni, Cu, and PGE and hence, for example,had low
Cu/Zr, led to the suggestionof a link betweenthe forma-
tion of magmatic sulphideliquids and contaminationof
flood-basaltmagmasby the continentalcrust [Naldrett et
al., 1992]. This suggestionencouragedNaldrett and
Lightfoot[1993] and LightJbotet al. [1994] to arguethat
the compositionof flood-basaltmagmascouldbe usedas
an indicationof the likelihoodthat comagmaticintrusions
are hoststo giant multibillion dollar Ni, Cu, and PGE sul-
phidemineral depositssuchas thosefoundat Noril'sk and
Talnakh in Russia(see Figure 1). Recenteffortshave fo-
cusedmore closelyon an understanding of the reasonfor
this strong empirical relation, not least becausethe data
provide the mineral explorationindustrywith improved
models for the location of mineral deposits. However,
suchmodelsare very sensitiveto whethersulphideforma-
tion is triggeredsimply by an increasein the silica contents
of the magmas [Irvine, 1975] and/or by the addition of
large amountsof crustalsulphur[e.g., Ripley, 1981]. A1-
10
'•8M•
ß
3.6NORIL'S
A•••5K•-•/o•
(•2.
KAMBALDA TALNAKH
89•2.• _BAY % •
% % JINCHUAN% •648•.20
0.1
10
%oooo. %
• 00 • 000 • 0,000
Production + R•rw• in • 0 • t
Figure 1. Plot of Ni gradeagainsttonnage(production + re-
serves)
for majorsulphidedeposits,afterNaldrettandLightfoot
[1993]. TheNoril'skdepositis halfthesizeof thatat Sudbury,
buthasa muchhigherreserveof Ni andPGE.
ternatively,sulphideformationmay be dueto a physical
controlsuchas magmatemperature, pressure,or fugacity
of sulphuror oxygen[e.g., Wendlandt,1982].
In this contribution we consider data from two CFB: the
Noril'sk region of the SiberianTraps and the Qeqer-
tarssuaq(Disko) Island and NuussuaqPeninsulaof the
West Greenland CFB, becauseboth include contaminated
tholeiiticlavaswith pronounceddepletionsin Ni and Cu
[e.g.,Pealersen,1985a;Naldrett et al., 1992, 1995;Light-
foot et al., in press;P. C. Lightfootet al., ms. in prepara-
tion]. We note that other CFB such as the Osler Volcanic
Group of the Keweenawan midcontinentriff also exhibit '
thesefeatures[Lightfootet al., 1991]. Geologicalandgeo-
chemicaldataareusedto investigate thefollowing:
(i) whether the addition of crustalsilicatematerial alone
is sufficientto triggerthe segregationof largevolumesof
sulphidefrom mafic magmas, or whether an external
sourceof sulphursuchas shaleor evaporite-rich sediment
is requiredto producegiantmagmaticsulphide deposits;
(ii) the extentto which contaminationby differentcrus-
tal componentscan be recognizedand then linked to si-
derophileandchalcophileelementdepletion;and
(iii) whetherthere are specificfeaturesof the lava che-
mostratigraphy that can be usedto identifyequilibration
betweensulphideandsilicateliquidsin thehigh-levelcon-
duits of CFB.
We first review the controlson the formationof mag-
matic sulphides,then compareaspectsof the geologyof
the WestGreenlandandSiberianCFB, andultimatelyuse
thisinformationtogetherwith geochemical datato address
the above
 LIGHTFOOT AND HAWKESWORTH 359

2. CONTROLS ON THE FORMATION

SiO2
OF MAGMATIC SULPHIDES

Tridymite
A number of importantcontrolsgovern the likelihood Tridymite + Liquid
that a mafic magma will segregateand concentrateeco- + Liquid
nomicamountsofNi-, Cu-, andPGE-richsulphides.

2.1. Sourceof Sulphur Fayalite 1200 øC

Tridymite
+ Liquid
+ 2 Liquids
Mantle-derivedmagmasare likely to containsomesul- Liquids
phur,and S may be introducedby assimilationof basement
or sediments from the continental crust. Basaltic rocks
typicallyoutgasS duringsubaerialeruption,and therefore
the measuredS contentsof basalticrocksare a poor index
of thosein the originalmagmas.Mafic intrusiverocksde-
void of sulphidecontainrarely more than 1000 ppm S,
FeO
MORB glassescontain-800 ppm [Harnlynet al., 1985], Wustite+Liquid FeS

and the pre-eruptionS contentsof Columbia River flood WEIGHT PERCENT

basaltshave been measuredat 1900 ppm [Thordarson,
1995]. Crustal rocks have vastly differing sulphurcon-
tents. For example, gneissesand granitoidsfrom West
Greenlandcontain an averageof 220 ppm S and quart-
zofeldspathicsediments700 ppm S [Shaw et al., 1976];
mineralizedand unmineralizedgreenstone-belt rockstypi-
cally contain5-25 wt% S, and <2000 ppm S, respectively
(P. C. Lightfoot, unpublisheddata, 1996); a compositeof
black shalesfrom the Nuussuaqarea of West Greenland
contains0.8% S [Pedersen,1979] andevaporiticsediments
at Noril'sk have 19.5 wt% S (P. C. Lightfoot,unpublished
data, 1996). Addition of crustalS to mafic magmasmay
be an importantcontrolfor a numberof deposits,suchasat
Noril'sk [Grinenko,1985] andDuluth [Ripley,1981].
2.2. SulphurSolubility
The solubility of sulphurin mafic magmasdependson
their silica content. Irvine [1975] demonstratedthat the
addition of silica-rich crustalmaterial to mafic magmas
changesthe proportionsof tetrahedralto octahedralsitesin
the magmaandtherebythe solubilityof sulphur. Thus,on
Figure2 [fromIrvine, 1975] the additionof SiO2 to a ho-
mogenous,sulphur-richliquid of compositionA changes
its composition to point B, whereit consists
of two liquids,
onesilicate-rich(Y) andthe othersulphide-rich (X). Iden-
tificationof crustalcontamination in maficmagmas,and in
particularthe magnitudeof the increasein SiO2 due to
contamination, is thereforeimportantto understanding the
causesof sulphideformation. Moreover,there are a num-
ber of CFB whosecompositions have beenmodifiedsig-
nificantly by crustalcontaminationprocesses and which
havea smallerthan expectedinventoryof chalcophileele-
Figure2. The 1200øCisothermof theFe-S-Osystemillustrating
howthe additionof SiO2 to a homogenous sulphur-richliquidof
compositionA resultsin a changeto compositionB, where it
consistsof two liquids,one silicate-rich(Y) and the other sul-
phide-rich(X). After Irvine [ 1975].
ments like Ni, Cu, and PGE [e.g., Naldrett et al., 1992;
Lightfootet al., 1994, in press,in preparation;
Bragmann
et al., 1993].
Overall, the solubilityof sulphurin mafic magmasis a
functionof magmatemperature, pressure,and both fO2
and fS2 (wheref = fugacity)[MacLean,1969; Wendlandt,
1982]. Thesephysicalparametersare linked to composi-
tion; for example,magmatemperatures dependon the de-
gree of differentiationand fS2 dependson the relative
abundanceof sulphurin the system. In this contribution
we are concernedwith the compositionalcontrolson S
solubility.
2.3. S-UndersaturatedMafic Magmas
If a silicateliquid is to generatea sulphidedepositwith
elevatedNi, Cu, andPGE, it is essentialthatthe liquidwas
not S-saturatedat an earlier stagein its evolution[Keays,
1995]. If magmasequilibratedwith and were separated
from sulphideearly in their evolution,then their Ni, Cu
and PGE contents would be so low that even if external
sulphurwas assimilated,any late-stage,higher-levelsul-
phideswould tend to be barrenof economicconcentrations
of ore metals. Hot, primitivemagmasappearmore likely
to be S-undersaturated because at low Fe contents the S
solubilityof a magmais muchhigherthanat high Fe
360 FLOOD BASALTS AND MINERALIZATION
tents [Wendlandt,1982]. The degreeof partial melting
also controlsthe amount of S in the magma, and Keays
[1995] for example,arguedthat S-saturated
magmasare
generatedat <25% partialmelting.
2.4. Partitioningof the Metalsand Upgradingof the Sul-
phide Liquid
Conventionalsulphur-saturation modelsfor the genera-
tion of magmaticNi, Cu, andPGE deposits[e.g.,Naldrett,
1989]proposethatthe equilibration of silicatemagmawith
immisciblemagmaticsulphideliquid resultsin the strong
depletionof Ni, Cu, and PGE in the silicateand enrich-
ment of theseelementsin the immisciblesulphideliquid.
This effect is due in part to the high partitioncoefficients
of theseelementsinto the sulphideliquid (DNi = 500-900,
DCu = 1400, Dpt = 3900, Dpd = 35,000 [Peach et al.,
1990]) but is alsoa functionof the ratio of magmato sul-
phide-the R factor of Campbelland Naldrett [1979].
More recently,the importanceof progressive upgradingof
the sulphideliquid in Ni, Cu, and PGE hasbeenaddressed
by Naldrett et al. [ 1995]. They and Rad'ko [ 1991] recog-
nized that the sulphidesat Noril'sk musthaveequilibrated
with largevolumesof silicatemagmain the conduitsof the
SiberianCFB and that the sulphideshad beenupgradedin
Ni, Cu, and PGE by continuedequilibrationwith succes-
sive magma batchespassingthroughthe conduits. The
magma,depletedin thesemetals,was then eruptedas the
lavasof the NadezhdinskyFormationof the SiberianTraps
[Naldrett et al., 1992, 1995].
3. COMPARATIVE GEOLOGY AND
GEOCHEMISTRY OF THE SIBERIAN TRAPS
AND WEST GREENLAND CFB
3.1. Large-ScaleTectonicSetting
The lavas of both the Noril'sk region of the Siberian
Trapsandthe West GreenlandCFB appearto be associated
with the emplacementof a mantleplume. In the caseof
the SiberianTraps,the 250 Ma sequenceof basaltsis lo-
cated on the Siberian Platform to the southwest of the East
Siberian lowlands, the Enisei Trough and the Khatanga
Trough [Aplonov, 1988]. The Noril'sk region is at the
northwestcomer of the SiberianTraps (Figure 3), and it
containsa remarkable diversity of rock types, including
two differentsequencesof primitive picritic lavas,one of
which has many geochemicaltraits consistentwith rela-
tively large degreesof melting of asthenospheric upper
mantle[Lightfootet al., 1993; Woodenet al., 1993]. Mor-
gan [1981] suggestedthat melt generationwas linked to
the JanMayen hot spot,whereasFedorenkoet al. [1996]
arguedthat structuraland sedimentological
evidencedoes
not supportuplift accompanyinga mantleplume.
In West Greenland,lavas in the Davis Straitregionwere
eruptedduring rifting betweenCanadaand Greenlandat
-63 Ma [Holm et al., 1993]. The lavascropout in Canada
on Baffm Island at Cape Dyer [Clarke and Pedersen,
1976; Clarke, 1991] and alongthe westcoastof Greenland
on Qeqertarssuaq (formerly Disko Island),and Nuussuaq
(theNuussuaqPeninsula)(Figure4), UbekendtIsland,and
the SvartenhukPeninsula. Holm et al. [1993] showedthat
the picriticrocksare compositionally coincidentwith Ice-
landic lavas and suggestedthat the picritic lavas of the
West GreenlandCFB were derived,at leastin part, from a
mantleplume. Lightfootet al. [in press]reportnew geo-
chemicaldata which indicatethat the lavas containup to
14-16 wt% MgO. This contrastswith at leastthe miner-
alized intrusionsat Noril'sk, which have Fo82olivinesthat
were in equilibriumwith a low-Mg tholeiiticparental
magma.
3.2. RegionalSetting
The lavas of the Noril'sk region of the SiberianTraps
and the West GreenlandCFB were both eruptedin regions
of extensive Phanerozoic sedimentation. In West Green-
land, the lavaswere eruptedonto a progradingdeltaicas-
semblageon the marginof a Proterozoicto Archeancraton
[Pedersenand Pulvertafi, 1992], whereasat Noril'sk, they
were eruptedonto an epicontinentalassemblage of shal-
low-watersedimentsand sabkhadeposits[e.g., Naldrett et
al., 1992]. At Noril'sk, Ripheiandolomites,argillites,and
limestonesof marine origin restunconformably on Lower
Proterozoicgneissesand crystallineschists.The Ripheian
sedimentsare overlain by extensiveDevoniancalcareous
and dolomitic marls, dolomites, and sulphate-rich
evaporites,and Lower Carboniferous shallow-waterlime-
stones. These rocks are unconformablyoverlain by the
Middle Carboniferousto Upper PermianTungusskaya epi-
continentallagoonaland continentalsediments, including
siltstones,sandstones,conglomerates,and coal measures
[Smirnov,1966; Glazkovskyet al., 1977; Simonov,1994].
The overlyingSiberianTrapsrepresent
>1 x 106km3 of
eruptedmagmathat peakedin the Early Triassicat 248-
250 Ma [Renne and Basu, 1991; Campbellet al., 1992;
Dalrympleet al., 1995].
In West Greenland,the subhorizontalvolcanicrocksrest
unconformablyon a ruggedpalaeosurface composedof
fault scarps,basementridges,anddeeplyerodedchannels
throughthe underlyingCretaceous-Tertiary
sediments
and
Archean-Proterozoic basement [Clarke and
 LIGHTFOOT AND HAWKESWORTH 361

Location Map

Independent States

Kharayelakh .........................
Basin ....

SG32

Talnakh
0
o o o Melkoye
c)oc)(
Norll'

•o• •a#lc andCr©taceous

,©clment
..........::......:.....:••
Kumglmky, $amoyedsky,Idokulaev•y
and KQraye4akhsky lava formatlone

.:...:....:....•
Morongov•'y,
and Nacle•
Khaki, Tuldor•y
lava fc)m'•atk)•
I
• tvak•ky,
Guclc•lc• $yvemdleJcy
and
lavQ formations

o• Cmnbdan
to UppmPermira
sediments
• Bolgokhtoksky
granltold
complex
..• Northwesttrencllng
faults
/ Mcdn northeast • north-northeast
Khant•oye • trendrig
faults
Lake
/ Idinor northeast and north-northeast
• trem:lng
faults
• Drlloorelocations

Figure3. Geologicalmapof theNoril'skregion,afterLightfootet al.

362 FLOOD BASALTS AND MINERALIZATION

I I
55 ø 54 ø

manak
Itivdl
Fault
-70.5 ø

D-93-08
Hammersdal
Complex
D-91-01

D-91

-70 ø dlukunguaq
dyke Sarqaq::•
-1 Intmsions•
D-91-05 QEQERTARSSUAQ
D-93-09{
D-93-06

-69.5 ø

Approximate locations LEGEND

of volcanic centers'
A Asuk volcanic center • Icecap
B K•gfinguaq
volcanic
center
C Va•gatvolcaniccenter Qeqertarssuaq '"'"""':•
Tholeiitic,contaminated
tholeiitic and picritic lavas
D Mal•gatvolcaniccenter
• Fault Cretaceous-Tertiary
sediments
-69 ø -:;• Precambrian
basement
50km ß Sampled section

Figure 4. Geologicalmap of the West GreenlandCFB, after Ulff-Moller [1991

 LIGHTFOOT AND HAWKESWORTH 363

1976]. On Qeqertarssuaq, a gneissridge runsnorth-south stratigraphicunitstermedthe Vaigat, Maligat, andHareOen

through the centre of the island and was an important
topographiccontrolduringthe eruptionof lavasthat acted
as a barrierto the migrationof someflows. Early volcan-
ism was dominatedby the developmentof picritic pillow
lavas and hyaloclastitebrecciasflowing onto an epiconti-
Formations[HaM and Pedersen, 1975] (Figure 6). The
divisionof the volcanicstratigraphywas basedlargelyon
lithostratigraphyand the relativepositionsof key marker
W E

••a••s
samøedsk•y
nental deltaic sequence[Pedersenand Pulvertafi, 1992].
The sedimentson Nuussuaqconsistof a <5-km-thick flu-
vial epicontinentaldeltaicsequencewith moderateto high
carboncontentand associated hydrocarbonaccumulations, ky
but without exposed evaporite sequences. Deltaic sedi- ......................
•:Age:of ............................
.....-.-..•..:•::;•:::::..------------•

mentation appearsto have progradedfrom the southeast :;::•::•:.•

•;;•?'"•"••• ueaw•y •.............
towardsthe northwest,and numerousgneissinliers on Qe- A:'g'•::"o•?"•'•:....•l
contaminated
r.....................
'M
:.-• k•'"" '•--...:E'O:w..No•
- ....'•:'•':••'••07
•::
"•"•:•:•:::";i•
.......
...............................................
::sk
.....................................................
':--:•'• ................................................
:'•'•(:••;
'"•' ! fi ow s ''•.....
qertarssuaqindicatethat the depthto the basementis shal- •:.:..:•..:..::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
........ ::::::::::::::::::::::::::::::::::::::::::::::::::
==============================================
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:::::::
.............
:_:.:_..•.:.:.:.
=============================================
lower than on Nuussuaq. .• ............ :..... .:. ---..... ...... ::::::::::::::::::::::::::
. ============================================================

•....•.•;;•:•;••••J•.•:.;J•;•;•
•.••{•"--"-'---'-----------•••:•:•:•::•::•::•::-
.............................
•'"•'••':"-"""'•"••:•••••••••!
--••E --.-.------•----•
.........
•.-.••
3.3. VolcanicStratigraphy
:::::::::::::::::::::::::::::::::::::::::::::::::::::
• Khakhanchansky
T•
In the West GreenlandCFB and at Noril'sk, up to 3500
m of picritic and tholeiitic lavascropout. The sequenceat
Noril'sk thins significantlytowardsthe southeaston the Syverminsky tholeiite

Putoranaplateauwhere only a few tens of metersof lavas

are present within a thick sequenceof tuffs. The lava
stratigraphywas establishedby Russianworkersbasedon
stratigraphic,petrological,andmajor-elementcriteria[e.g.,
Fedorenko, 1981], and these subdivisionshave been re-
tained in subsequentstudiesof the trace-elementand ra-
diogenicisotopevariations. The lavasconsistof a Lower
Sequenceof three formations,the Ivakinsky,Syverminsky,
and Gudchikhinsky,overlainby an Upper Sequenceof the
Khakhanchansky, Tuklonsky, Nadezhdinsky, Moron-
.v.:'al•ns•ysub all•ahc bas:-'alt.
Figure 5. Stratigraphyof the SiberianTrapsat Noril'sk pro-
jectedontoa west-east
section,afterLightfootet al. [1994].
Formation Member
::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

govsky, Mokulaevsky, Kharaelakhsky,Kumginsky and

SamoedskyFormations(Figure 5). The Ivakinsky consists ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

dominantly of subalkalic basalts and trachybasalts, ........................................

whereasthe overlying Syverminskyconsistsof a sequence

......................................
........................................

Ordling•ssoq
........................................

--.-..-.-.-.....-.-.-.-.-.-.-.............-.-:.-...-...-...-.._-.-........-.- •

of tholelites. The Gudchikhinskycontainsa lower unit of .......................................

tholeiiticbasaltsand an upperunit of picrites[Lightfootet :.:.:.:.:.:-::-:-:-:-:.:-:.:.:-:-..:.:.:.:.:.:.:.:-:-:-:-:-:-:-:-:-...

........................................
-:.:.:.:.:
:.:...:.:.:.:.:.:.:.:.:.:...:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:...:.:.:.:.:
........................................

:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:

al., 1990; 1993]. The rocks of the Upper Sequenceare

.................................................................................

........................................

dominantlytholeiiteswith subordinatepicritesin the upper

:::::::::::::::::::::::::::::::::::::::::
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
.....-...-.-.-.-.-.-.-.-.-.-.-.-.-...-.-.-.-.-.-.-.::........•
:::::::::::::::::::::::::::::::::::::::: :....-.-.-...-.

unit of the Tuklonsky[Lightfootet al., 1993] and andesitic

basaltsin the Kharaelakhsky[Hawkesworthet al., 1995].
The rocks of the Tuklonsky throughto the Mokulaevsky
are of principal concernin this manuscript. These rocks
occupy a considerable portion of the stratigraphy
(Tuklonsky: 2-220 m thick; Nadezhdinsky: 150-670 m
thick; Morongovsky: 230-790 m thick; Mokulaevsky:
255-1095 m thick [Lightfootet al., 1994]). Picritescom-
prise <1% of the stratigraphicthicknessin the Noril'sk w E
area [Fedorenko, 1994]. Figure 6. Stratigraphyof the West Greenlandtholeiites, after
The volcanicstratigraphyof the WestGreenlandCFB on Pedersen [1985b]. The contaminatedunits are the Asfik and
Qeqertarssuaq(Figure 4) is subdividedinto three litho- Kfigfinguaq
364 FLOOD BASALTS AND MINERALIZATION
horizons,rather than geochemicalcriteria. Thus, the Vai-
gat Formation was distinguished from the overlying
Maligat Formationon the presenceof picritic lavas,hyalo-
clastites, and a small number of basalt horizons. The
Maligat Formationconsistsof feldspar-phyrictholeiiticba-
salts, and the Harefen Formation consists of transitional
olivine-phyricbasaltsseparatedfrom the Maligat by shales
with coal seams. Pedersen[1985a] subdividedthe Vaigat
Formation into severalmemberswithin two volcanic cy-
cles:the lower cycle of-•900 m thicknessconsistsdomi-
nantly of picrites (Naujfinguitmember) with two 50- to
100-m-thicktholeiite horizons(the Asfik and Kfigfinguaq
members). The tholeiitic horizons of the Asfik contain
shalexenolithsand metallic iron [Pedersen,1978; Klock et
al., 1986], whereasthe Kfigfinguaqmember is devoid of
native iron. The Asfik and Kfigfinguaqmembersare both
highly amygdaloidal,reddishbrown colouredflow groups
which have elevated silica contents attributed to crustal
contamination[Lightfootet al., in press]. The Naujfinguit
flows were followed by the eruption of olivine-poor
tholeiiticbasaltsof the Qordlortorssuaq
memberin a wan-
ing stageof volcanicactivity [Pedersen,1985b]. The up-
per cycle of the Vaigat Formationcomprises800 m of pic-
rites of the Ordlingassoqmember,which includesthreeho-
rizons of distinctive, contaminated lavas intercalated with
the picrites[Pedersen,1985b].
On Nuussuaq,the Vaigat Formationis exposedeast of
the Itivdle Fault (Figure 4) and consistsof more than five
packages,or subunits,of contaminatedlava flows with
km of picritic rocks [e.g., Hald, 1977a, b; Lightfootet al.,
in press]. These lavas have been assignedan interim no-
menclaturebecauseit hasyet to be established whetherin-
dividual subunitscan be correlatedfrom Nuussuaqto Qe-
qertarssuaq. On Nuussuaq,the more basaltic, contami-
nated lavas are therefore in memberstermed B0 to B4, and
the picritesare in membersP0 to P6. One of the contami-
nated lava horizonssouthof the Serfatsill (Figure 4) has
associated native iron mineralization. The native iron oc-
cursas blebbyto massivecumulateswithin the lower part
of a complexflow unit at the baseof the B3 unit. This na-
tive iron resemblesthe native iron in a large boulderrest-
ing on Vaigat Formation lavas in Stordal,centralQeqer-
tarssuaq[e.g., Ulff-Moller, 1991, andreferencestherein].
The Maligat Formationis composedof thicker,massive
basalts. Both formationsoccur throughoutcentral and
western Qeqertarssuaq,west of the Itivdle Fault on Nuus-
suaq,and alsoon the eastsideof Nuussuaq(Figure4). On
Qeqertarssuaq,the Maligat Formation is subdividedinto
the Rinks Dal (-750 m thick), the NordOord,andNiaqus-
sat members[Pedersen,1975]. The Harefen Formationis
-•250 m thick and consistsof olivine porphyritictransi-
tional basaltsand is developedonly on HareOenIsland.
3.4. RegionalStructure
The lavas at Noril'sk and in the West Greenland CFB
are associatedwith major faults, someof which may have
acted as the conduitsfor magma migrationø At Noril'sk
the epicontinentalsedimentsand unconformablebasaltse-
quenceare cut by a major seriesof NNW-SSE-trending
faults which appearto have played a centralrole in con-
trollingthe surfacedistributionof differentportionsof the
lava stratigraphy[e.g., Naldrett et al., 1992]. They may
also have been the foci for many of the intrusionsin the
region,in particularthosewhich hostthe major depositsof
Ni, Cu, and PGE sulphidemineralizationat Noril'sk and
Talnakh (Figure 3). The three major, possiblymantle-
penetratingfaults are the Noril'sk-Kharaelakh,Imangda,
and North Kharaelakhsky. In the lower part of the
stratigraphy,the primitive lavasof the Gudchikhinsky are
thickest along the Noril'sk-Kharaelakh and North
Kharaelakh faults and were presumablyerupted along
thesestructures.In contrast,the lavasof the Tuklonskyare
thickestalongthe ImangdaFault andwere eruptedwell to
the east of Noril'sk. Lavas of the Nadezhdinskywere
eruptedin the Noril'sk region and centredclose to the
Noril'sk-Kharaelakh Fault. Fedorenko [1981, 1994] and
Lightfootet al. [1990, 1993, 1994] discussed the thickness
variations in the different formations of lava in the
Noril'sk Region. Naldrett et al. [1992] usedisopachdia-
gramsfor the thicknessof theseformationsto demonstrate
that magmaticactivity switchedepisodically betweendif-
ferent eruptivesites,ratherthan migratingsteadilyacross
the provinceas, for example,when a platemovesacrossa
mantleplume[e.g.,Deveyand Lightfoot,1986].
In West Greenland,the BoundaryFault cutsthroughthe
easternpart of Nuussuaqand marksthe easternextentof
Cretaceousand Tertiary sedimentswherethey abut base-
ment Proterozoicgneiss[Pulvertafi,1989]; the fault may
haveservedlocallyas a controllingstructurefor the lateral
containmentof the flood basalts(Figure 4). The Itivdle
FaultcutsacrosswesternNuussuaqanddefinesthe bound-
ary of the Maligat Formationflows on westernNuussuaq,
which are downfaultedby more than 1 km relativeto the
main Vaigat packageeastof the fault. The eruptionof the
picriticlavasin the West GreenlandCFB appearsto have
been controlledby major faults; for examplethe Vaigat
Formationpicritic lavasare believedto havebeencentred
closeto the Itivdle Fault (Figure4), but someof the pack-
agesof tholeiites(e.g., the Asfik and Kfig/tnguaq)were
eruptedin separatesub-basins away from the major fault
lines [Pedersen,1985a]. This situationcontrastswith
 LIGHTFOOT AND HAWKESWORTH 365

A C
I I I I ] I I I I I i_ ß i I •. l
• 700
• 500
o
• 300 ß ._ L_ J, ' *
o 100 Km *• '• ' LKm
n -100
'• -300
'" -500
o• -700
• -900

•-•oo
•-•=oo
:•-1500
•-1700
o::-1900
,,-,-2100
E -23OO
•T:••'•
• • 5• ••?:
?•:•• •::•::•::•::•
:'• •=:
':•-::•-:-•-•::•
•:•
•::
:::::::::::::::::::::
• ::•
• ::?:•
•::•::?:• • :•::::::•::•:;
•• ••:•......:•
::?:-•• ••[•[•==•
••[•==•==•==•;=•=•=•;
•[•[•
•••==•;•;
;•;?;=;•
=••=;•
==•
•;•[•• •=•
[•=•
•;•==•
••=•
•• •
'• -2500 Gd _• . ?• •• ..........' LGd.'""'•"•'-'•--' ..................
2..................
'i"L"i '"'•.........
• ..................
L
:::.....:::-:.:
:.:.::
::.:
:.:
:.:.:.-.: :::.:.:.:
:.:.::
:.:
:.:
:.:
:.:.:
:.::
:.:.:.:.:.:.:.:.:.:.:.:.......:.:.:.:.:.:.:.:.:
:.ß.::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
:.:.:.:_:.
..:.:.:-....:.:.:.:.:.:.:.:.:...•.•.:.:.:.:.:.:.:.:.:.:
• ...........•m
:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:.:
:.:.:.:.:.....:.:.:.:.:.
ß.....:
:.:.:....
•::-:.•:::
:•---: :.:::::::::::::::::::::::::::::::::::::::::::::::::::
•-2700"•'•,'v"s" , • ; •';'•;i , ,'v .................................................
', ................
-•",-•......
-,...................
•.-2900 0.1 1 2 3 4 5
cu•r LalSm
B D
• /I [ [ [ I I I mI I ... ,m,
m I ' ' ' . ,i, I ....
E 700F'Sm ''o . %., ø"]- Sm %.tø -
= 500I- ß •*' -] - • ß -
'•O •00 _,_ - .,:# -
=o -•00
-300
r- ... . ;_ '•." .d- __"•.' • --•-
'- I-- Kh tL 1 Kh
._=-700 . • - .• -
• -1100 • :

::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
..........
ß'"- '.-:--'-:-•;=:.•
ii:'•[[[[[::[•[•i•[•[:[:[•[::i[[i[::[[[
:::::::::::::::::::::::::::::::::
==========================================================================================================•:::::::::::•:::
.......................
:'•
.........
*:,'*:;-'-':;'::;--'::
........
:-'::::
..........
::,'",:-
........... =====================================================================================================================
;'-':-".--:•::.•:=::::::::
.........
',:...................
:-'.': :!:•:!:•:!:•:•:•:!:•:•:•:!:i:i:!:i:!:•:•:i:i:i:•:i:!:•:i:i:i:i:•:•:•:•.:•::•::•:i:i:•:i:i:i:i:i:i:i:i•:!:i:i:•::•:!:i:i:i:i:i:!:i•:i:i:i:i•:i:i:i:i:i•:i•:i:i:i:i:iii•
..............................................................
'...................................................................................
.--z•:.
............. •::'i•:i:::•:•:!:!:i:!:!:':•::•-':i:i:i:i:i:•:i:•:•:•:•:•:!:•:•:!:
::"---"•----"•,-•:•-•
........
•,.-•
...................
s-2.,oo
•:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
..'.• ........................................
•:• it•.................................................................
.....
................................... •••.....••'•:•:•i.•.'-.
::.......... ...........
::.........................
::::::::::::::::::::::
.......................
::::::::::::::::::::::::::::::::::
....................
:::::::: •,-.,:::,,.
............. ,,-----:.----•---.---.•.-
......
-.--.---,
U ........................................
*-:: -.-...--..-.-..-•
...........................................

'•-2500 ,:•..........................................................
,•-- ..........
....
......
.ff::.:-:..-.:-•,,.•.-..•:.,-:_:..•
........................................
.•:•..................................
ß
...............................
..................
.•:........................
•:.........................
..-..
....................
...,,:
.........................
•-.............................................
.•..:::::.•
Iv o
ff•.•s0011 , ,eo• , • , , ,,I , F Iv o o
, ,o , o
....o
<1: 10 100 30•5 46 47 48 49 50 51 52 53 54 55
CU (ppm) SiO• (wt %)
Figure 7. Chcmostratigraphy of a compositesectionof the SiberianTraps at Noril'sk [data from Lightfootel •.,
1993, ! 994, ms. in preparation;•oode• git•., ] 993; H•wkesworthgit•., 1995].

Noril'sk regionwhere much of the eruptionwas centredon structedfrom data for samplesof drill core (hole SG-32)
major mantle-penetratingfaults. The centre of Maligat and surface outcrop (sections 1F, 15F, and 16F) (P. C.
volcanismis poorly known and althoughit may have been Lightfoot et al., ms. in preparation).The sectionof strati-
west of Qeqertarssuaq,the presenceof many Maligat-like graphy focused on here is that between the base of the
doleritesills(Sarqaqdolerites)alongthe BoundaryFault in Tuklonskyandthe top of the MokulaevskyFormations.In
the Sarqaqvalley (Figure 4) suggeststhat at leastsomeof this stratigraphicinterval there is a marked drop in Cu
the Maligat lavas were erupted close to the Boundary content at the base of the Nadezhdinskyfollowed by a
Fault. gradualrecovery in abundancelevelsupwardsthroughthe
sectionto the Mokulaevsky(Figure 7a) [Lightfootet al.,
3.5. Chemostratigraphy of the Lavas of the Nor#'sk Re- 1990; 1993; 1994; Woodenet al., 1993]. The changesin
gion and the WestGreenlandCFB Cu contentare accompaniedby similarvariationsin Ni and
PGE contents[Lightfoot et al., 1990; Briigmann et al.,
Figure 7 showsthe compositechemostratigraphy of the 1993], and by a suddenincreasein La/Sm at the baseof
SiberianTraps at Noril'sk. The sectionhas been recon- the Nadezhdinskyfollowed by a progressivedecline
366 FLOOD BASALTS AND MINERALIZATION

A
:::::

1300 - ß ß B4 -

1200
J400
1150 ;::::
:::
:::::::
:'i::::,:
:::'"',....'/..'",:
:::::::::::::::::::::::::::::::
................................ :::
::: ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
.......
:::::::::::::::::::::::::::::::::::::::::::::::::::::
11oo ß B2 200

900 •
...........................................................................................................................................
n, 800 - 03 i•O•o•,,• O I•g•nguaq - •
900 B1 •= 700 _ o o..................................................
XS'•'• ..................
::i::i
• 600
800
• 5oo
0.1 1 10 0.1 I 10
Cu/Zr CulZr

• B
:•:•:•:::::::::::::::::•:•:::::::•:::::..•..::•:•::.•:::::::::::::•:•:::::::::•:::::::::::::::::•:::::::::::•:::::::•:::•:•:::::::•:•:::::::::::::::•:•:•:::::::::•:::::::•:•:::•:::•:::•:::•:•:•:•:•:::•::
:P::.•.':::::::::::::::=:::=:=:
.o_
1300 - & B4 _ 1700
1600

1500
1400
E 1150................................................................................................
.-.-.. i::::...................................
: ..........................................
: .........................
::'...........
.......................
j[.i...............................................................................................
.-.. '•3oo
1200
• 1100 ß B2
ß 1050 11oo - • • ==================
1000
900
800 - • " K'b•guaq - •
E 900 i:::::!!:•:::!:::::::::•:::::::::::::::i::::::::::::::::!:•::::•:!!:::::::::::::::•:::i::i:::::::::!:i::::::::::::::::i:•:::::::::::[!::i[i:::::i•:[::::[:•i:::::::i:::::i:
B1 700 :..................................................................................................
.._U..
• ......................................................................
ß• .=..........:.:.:..:............................................................................. Asuk •.•.•.•.•.•.•.•.•.•••••••.•.•••••••••••.•••••••..••••••••.••••••..••••••••••••.••••••••••.•••.•.•.•••••••••••.•...•••.•.•.•
..............................

o 850
•. ...........
•.....•.•.•.•••.•.•.•.••..•.•••..•.••••.•.•.•••••.•.•••.•••.•.•••••.•.•••.•...`•.•••.•••••.••..•••••......
............................
.....
600
!:!:iiiii:iii
i:i:i:i:i
i.i:i:i:::i:i
!:::::::::i:i:!
:::::
i:::::::::i
!i:::i:!=::!:i:i:::::!:!:!:i
::!:!:!.::i:::i:::!:::i:!:::::i:!:!:!:!:::::!:i:i:!:!:•:!:!:!:i:i:::::::::::•::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::.:::::::::::::::::::::
5oo :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::
,• 800
1 2 3 4 5 6 7 I 2 3 4 5 6 7
LalSm La/Sm

Figure 8. Chemostratigraphy of compositesectionsof the West GreenlandCFB, with datafrom Lightfootet al. [in
press]. (a) Cu/Zr versusstratigraphicposition,Nuussuaq. (b) LadSmversusstratigraphicposition,Nuussuaq. (c)
Cu/Zr versusstratigraphic
positionon Qeqertarssuaq.(d) LaYSmversusstratigraphic
positionon Qeqertarssuaq.

wards the top of the Mokulaevsky (Figure 7b). This In the West Greenland CFB on Qeqertarssuaq,a com-
changein La/Sm is accompaniedby systematicchangesin positestratigraphyhasbeendevelopedbasedon the dataof
SiO2, in large ion lithophileelementto high field strength Lightfoot et al. [in press](Figure 8). Pedersen[1985a,b]
element ratios (LILE/HFSE), and in radiogenicisotopic first demonstratedthat many of the tholeiitic lavas from
ratiosand has been explainedin termsof a progressivede- Qeqertarssuaq had been contaminatedby crustalmaterial
cline in the amountof contaminationup-section[Lightfoot and that not only was olivine in theselavasdepletedin Ni
et al., 1990, 1993; Wooden et al., 1993; Hawkesworth et but alsothat someof the whole-rockcompositions had ex-
al., 1995]. Naldrett et al. [1995] and P. C. Lightfootet al. tremely low Ni and Cu contents.He further suggested that
(ms. in preparation)ascribethe progressivechangesin the the strongNi and Cu depletionwas causedby the segrega-
metal contentsof these lavas to continuedequilibrationof tion of a magmatic sulphide liquid from the magma.
highly contaminatedTuklonsky-typemagma with a sul- Lightfootet al. [in press]demonstrate that the depletionin
phide liquid in the conduitsof the CFB, which are now Ni and Cu is associatedwith magmasthat had elevated
representedby the Noril'sk and Talnakh intrusionsand contents of SiO2, light rare earth elements and LILE,
their associatedmagmaticNi, Cu, and PGE sulphides. similar to those in the contaminated tholeiites at
 LIGHTFOOT AND HAWKESWORTH 367

(seeFigure9a). In the stratigraphic sectionshownin Fig-
ure 8, the depletionin Cu is observedto be a suddenevent,
much like that recordedat the base of the Nadezhdinskyin
Figure7a. The thickness
of the Cu-depleted
unitsis -100
m, which is lessthan the 500-m thicknessat Noril'sk (and
the 1000-msequence in the OslerGroupvolcanicsof the
Keweenawan midcontinentrift [Lightfoot et al., 1991;
Naldrett and Lightfoot,1993]). However,thereare a num-
ber of unitson Qeqertarssuaq andNuussuaqthat showCu
depletion,whichmakesthe overallvolumeof Cu-depleted
lavas-103 km3, compared
with-5 x 103km3 atNoril'sk.
One of the more significantdifferencesbetweenthe
West Greenland and Noril' sk successions is that the former
preservesno evidenceof a gradualrecoveryin the Cu
contents of the lavas either within or above the contami-
nated, low-Cu units, such as the Asfik or the Kfigfinguaq
memberson Qeqertarssuaq.The mostCu-depletedrocks
show the same low La/Sm as the Noril'sk lavas,but again
there is no gradualdecreasein La/Sm upwardsthrough
eitherthe Asfik or the Kfigfinguaqmembers(seeFigure8).
On Nuussuaq,like Qeqertarssuaq, the tholeiiticmembers
havelow Cu contentsandhigh La/Sm ratiosrelativeto the
picriticrocks[seeLightfootet al., in press],andtheyalso
showno systematic compositionalchanges within individ-
ual units. Suchdifferencesin the ratesof chemicalchange
with stratigraphicheight betweenWest Greenlandand
Noril'sk (andthe Keweenawan)presumably reflectdiffer-
encesin the dynamicsof themagmafeedersystems, which
may in turn have influencedthe extentto whichdifferent
magmabatchesequilibrated with separatedsulphides.
4. DISCUSSION
At issueis whetherdifferentprocesses
of contamination

Mr-S,n. Noril'sk

d• o o
o
o NJ

• 3
I & I 0IlV-SV•
I I
A-cO
• picritesWest Greenland
2

<;•B•L Ao ß
o øA
Maltgat
•

45 50 55
JAstik_Ktig•inguaq•o•ø••
0
• -CO
I

1
oo- ABO-B4
0• o
9

2
I

3
I

4
i

5
I

6
I

7 8

SiO2 (wt%) La/Sm

Figure 9. (a) Plot of variationsin Cu/Zr ratiosand SiO2contentsin samplesfrom the SiberianTraps [datafrom
Lightfootet al., 1990, 1993, 1994,ms. in preparation;
Woodenet al., 1993;Hawkesworth
et al., 1995], andthe West
GreenlandCFB [datafrom Lightfootet al., in press].
(b) Plotof Cu/Zr-La/Smratiosfor samples
fromtheSiberianTraps[datafromLightfootet al., 1993,ms.in prepara-
tion; Woodenet al., 1993; Hawkesworthet al., 1995], and the West GreenlandCFB [datafrom Lightfootet al. in
press].For the Siberianlavasthefollowingabbreviations areusedfor thedifferentformations:Iv - Ivakinsky;Sv -
Syverminsky; Gd - Gudchikhinsky; Tk - Tuklonsky(filledcircles);Nd - Nadezhdinsky (opencircles);Mr - Moron-
govsky;Sm- Samoedsky. FortheWestGreenland lavasthesymbols areasfollows:filledcircles- VaigatFormation
picriteson Qeqertarssuaq;filledtriangles - VaigatFormation picritesonNuussuaq; opencircles- Asfk andKfgfin-
guaqmembers; opentriangles - B0 to B4 rocksin theVaigatFormation onNuussuaq; opendiamonds - MaligatFor-
368 FLOOD BASALTS AND MINERALIZATION

and crustal end-members can be identified and the extent correlations

to which the sulphurand silica contentsof the contami-
nantsdeterminewhethera magmawill segregate magmatic
sulphide. Further considerations are whetherthe geo-
chemicalvariationsin the lava stratigraphyprovideinfor-
mationaboutthe timingof sulphidesaturation andanyup-
gradingof the siderophileelementdepositsby, for exam-
ple, continuedinteractionof the sulphideswith successive
magmabatchesin the CFB conduits.
4.1. RelationBetweenNi-Cu-PGE Depletionand Major-
Element,Trace-Element,
andIsotopicComposition
The tholeiiticlavasof the Nadezhdinsky
Formationhave
relativelylow Cu contents,
low Cu/Zr,andelevatedSiO2
and La/Sm (Figure 9 and Lightfootet al. [1994]). The
sametraits are found in the tholeiitic lavas of the Asfik,
Kfig/tnguaq,andB0-B4 unitsof the Vaigat Formation,and
portions of the MalEgat Formationin West Greenland
(theserockswill be referredto below as the CWGT, i.e.,
contaminated West Greenlandtholeiites). The negative
betweenCu/Zr andbothSiO2 andLa/Sm are
developedbetter in the CWGT than in the Nadezhdinsky
lavas, although this is in part becausethe CWGT lavas
with high Cu/Zr havesignificantly lowerSiO2 andLa/Sm
thantheir Noril'sk counterparts.The CWGT tend to have
consistentlylower Cu/Zr thanthe Nadezhdinskylavas,and
the rangeof Cu/Zr in the lattersuggests
that the degreeof
equilibrationwith sulphidewas lesscompletethan in the
CWGT. The Ivakinsky and Syverminskylavasat Noril'sk
are tholeiitic to subalkalic basalts whose low Cu/Zr ratios
may reflect small degreesof meltingin the presenceof re-
sidualsulphide.
The shiftsto high La/Sm and low Cu/Zr in the CWGT
and Nadezhdinsky are accompaniedby increasesin
87Sr/86Sro
anddecreases
in end(Figure10). TheNoril'sk
lavasdefine a broadtrend of increasingLa/Sm with de-
creasingl•Nd which was modelledby Lightfoot et al.
[ 1994] and Naldrett et al. [ 1995] in termsof contamination
of a Tuklonskyparentalmagmaby a crustalcomponent
with high 87Sr/86Sr
and La/Sm,and low l•Nd. This con-
taminationwasthenfollowedby progressive
mixingof the

10
0.25
Noril'sk Noril'sk
0.20

-2 0.15

-6

-10
ø•;•Nd 0.10

0.05
-14

10 • 0 I I

•o Picrites Astik-

ß©1• ß ß West Greenland 0.25 Kt•gdnguaq

0 West
•o o o Greenland
MaHgato0
ß 0.20
'•nd 2 trend•
••alt•at
A•A•'
-2
at> 0.15

trend
1• A A•BO-B4
o
-6 A A o

BO-B4 o
o 0.10
A AA • • o
• •A trend
2ß
-10
As•k- o ß . ß
0.05
-14 Kt•gdnguaq
o ptcrttes
oo
-18

0 2 4 6 8 0 2 4 6
La/Sm La/Sm

Figure 10. (a) Plot of ENdvalues(at 250 Ma) versusLa/Smratiosin samplesfromthe SiberianTraps[datafrom
Lightfootet al., 1990,1993,1994,ins.in preparation;
Woodenet al., 1993;Hawkesworth
et al., 1995],andtheWest
Greenland CFB[datafroinLightfoot
etal., in press].Thesymbolsandabbreviations
areasforFigure9, exceptthat1
is a lamprophyre
apparentlyassociated
withthe SiberianTraps.(b) Plotof Th/Laagainst
LaYSmratiosfor the Sibe-
rianTrapsandWestGreenland CFB,withdatasources, symbols,
andabbreviations
asforFigure
 LIGHTFOOT AND HAWKESWORTH 369

contaminatedmagma with Mokulaevsky magma in the crustal-likeSr and Nd isotopicratios (e.g., Figure 10a),
conduitsof the CFB system. Sucha model requiresthe higher SiO2 and low Ti/Zr, they have been largelyattrib-
establishmentof a large reservoirof contaminatedTuklon- uted to crustalcontaminationprocesses[Lightfootet al.,
sky magma followed by progressivereplenishingof this 1990, 1993, 1994, in press;in preparation;Natdrett et at.,
reservoirwith Mokulaevskymagma[e.g., Briigmannet at., 1992, 1995; Wooden et at., 1993; Hawkesworth et at.,
1993]. 1995; Fedorenko,1994]. Thus,in theserocksLa/Sm may
The CWGT are different from the Noril'sk lavas in that be regardedas an index of contamination,and from the
the sampleswith higher La/Sm ratioshave a rangein end above studies,the following general changesaccompany
consistent with contributions from two La-enriched end- the increase in La/Sm in both the CWGT and the
members (Figure lea). Most of the data for rocks from Nadezhdinsky:
Qeqertarssuaq
plot on a steepertrendof La/Smagainstend (i) increasesin La/Yb, Zr/Y, Th/Nb and Th/La (e.g.,
than those from Nuussuaq,except that the three samples Figure 1eb),
with the highest La/Sm ratios have relatively high 8Nd. (ii) reductionin Cu/Zr (Figure 9), Ti/Zr, P/Zr and Ta/La
This indicationof two La-enrichedcomponentsin the la- (e.g., Figure 11),
vas of the CWGT is reinforcedby the plot of Th/La against (iii) increases
in 87Sr/86Sr
o andreduction
in 8Nd(Figure
La/Sm (Figure 1eb), in that lavaswith lower 8Ndvaluesare lea),
characterisedby elevated Th/La, and rocks with higher (iv) increasesin SiO2 often with little changein Mg-
La/Sm ratios have intermediate Th/La in addition to number(see Section4.3).
slightlyhigher 8Ndvalues. On both diagrams,the CWGT Such changesare all consistentwith crustalcontamina-
data fan out from the fields for the Vaigat Formationpic- tion, but there are also differences in both the size of the
ritic lavas, and there are broad increasesin Th/La with de- changesand the element ratios between the CWGT and
creasing8Ndin both Noril'sk and West Greenland. The
least Th- and La-enriched rocks from Noril'sk still have 0.08

slightly elevated Th/La and La/Sm and low end values, Noril'sk
I
which hasled to contrastingmodelsin which suchmagmas
were derived largely from incompatible-element-enriched 0.06 - - -•-
I
sourceregionsin the mantle lithosphere[Lightfootet at., I

1993] or from the asthenospheric mantle with additional

crustalcontamination[ Woodenet at., 1993].

4.2. Nature of the CrustalComponents

in the Siberianand
0.04•
0.02
I
I
I
• Tk
/(• Nd
I
Greenland Lavas and Intrusions I
I

The nature of the crustal material incorporatedinto 0.0 • •• ß I I

mantle-derivedmagmasis importantto modelsof sulphide ß ß West Greenland
formationand specificallyto whethersulphidesaturationis picrites
• •
primarily triggeredby the addition of excessS, or by in- 0.06- - •'•- •-
creasesin silica content. Crustal componentsmay include
bulk sedimentaryor meta-igneousrocks,or meltsthereof, 0.04
and in principle may be distinguishedon the basisof se- I ß • A•BO-B4
0
lected minor-element and trace-element ratios. For exam- I Maltgat
oOd••a• •i•i•A o
ple, upper-crustalrocks tend to have elevated Rb/Ba, 0.02 [ o •• o
I

Rb/Sr,andhencewith time relativelyhigh 87Sr/86Sr;

deep • Ast•k-Kt•gdnguaq
I
crustalmelts are more likely to have been generatedin the
presenceof residualgarnetand henceto have low heavy o 2 4 6 8
rare earth elementsand higher Sr contents,and most crus- La/Sm
tal melts have relatively low P205 and TiO2 [e.g., Taylor
and McLennan, 1985; Harris et al., 1986; Hawkesworth Figure 11. Plot of Ta/La againstLa/Sm ratios for the rocks of
and Clarke, 1994]o the SiberianTraps and West Greenland. Symbols,abbreviations
Becausethe shiftsto increasingLa/Sm in the CWGT and and data sourcesas for Figure 9. The dashedlines are at the
Nadezhdinsky lavas are accompaniedby shifts to more Ta/La andLa/Sm ratiosof primitive
370 FLOOD BASALTS AND MINERALIZATION
Nadezhdinsky lavas that reflect both differencesin their
crustalcontaminantsand, more speculatively,in the proc-
essesof contaminationand fractionation(see also section
4.3). These differencesincludethose betweenaverage
Rb/Ba and Rb/Sr in the Nadezhdinskylavas(0.08 and 0.07
respectively [Lightfoot et al., 1994]) and in the CWGT
(0.16 and0.14) andthe fact thatthe CWGT haverelatively
low Sr/Zr (1.3-1.8), whereasin the NadezhdinskySr/Zr
and Sr are extremelyvariable(0.6-4.1 and 35-518 ppm)
and showno systematic trendswith initial 87Sr/86Sr
(P. C.
Lightfoot et al., ms. in preparation). Among the major
elements, Fe203 decreasesand SiO2 contentsincrease
morerapidly with decreasingMgO contentsin the Noril' sk
than in the West Greenlandpicrites. Moreover the Cu/Zr
ratios of the CWGT are more restrictedand on average
lower than thosein the Nadezhdinsky.
For the Nadezhdinsky,the combinationof elevatedZr/Y
and La/Sm, low Ti/Zr, P/Zr and Ta/La (Figure 11), and
Rb/Ba similar to the Tuklonsky[Lightfootet al., 1993] is
more consistentwith contaminationby mid- to lower
crustalmaterialthan by uppercrustalmetasediments [e.g.,
Lightfoot et al., 1994; Hawkesworthet al., 1995]. Thus,
the model of Lightfootet al. [1994] and Fedorenko[1994]
involvedcontaminationof a primitiveTuklonskylava with
8% granodioritic melt (similar in compositionto the
younger, Triassic Bolgokhtokhintrusion)and N24% sub-
sequentfractionationof olivine, plagioclase,and augitein
the ratio 35:41:24. Other models of contamination of the
Nadezhdinsky have utilised different amounts of crust
[Lightfoot et al., 1990; Woodenet al., 1993; Fedorenko,
1994], but the observedchangesin trace-elementratios
stronglysuggestthat the contaminantwas granodioriticin
composition. Such material will typically contain<0.01
wt% S, and so it would appearthat the contaminantre-
sponsiblefor the elevated LadSmratios in the Nadezhdin-
sky rocks was unlikely to be the source of significant
amounts of additional sulphur. The bulk rocks of the
Lower Talnakh intrusionshave many of the isotopicand
incompatible-element characteristics
of the Nadezhdinsky
lavas, and Hawkesworthet al. [1995] emphasisedthat
thesetoo are unlikely to be due to crustalcontamination
with sedimentary material. New datafrom P. C. Lightfoot
et al. (ms. in preparation)furtherdemonstratethat the De-
vonian evaporitesbeneaththe SiberianTrapshave high Sr
abundances(2051-2812 ppm) and 87Sr/86Sr(0.7078-
0.7084) and very low rare earth elementcontents(<10 x
chondrite). Thus, evaporitecontaminationwould resultin
large changesin Sr contentsbut have little effecton La/Sm
or end values.
In summary,the available evidencefrom Noril'sk indi-
catesthat the contaminationprocessesresponsiblefor the
Granodiorite
0.709
aporites
O.7O7
0.705
Noril'sk
0.703 i i i i
1 2 3 4 5
La/Sm
Figure 12. Plot of initial LaYSInversusSr isotoperatiosfor lavas
and intrusive rocks from Noril'sk, Siberia. The small field for
the evaporitesis simply to illustratethe observedrange in Sr
isotopiccomposition(P. C. Lightfoot et al., ms. in preparation).
Symbols for the different intrusionsare as follows: filled trian-
gles- Noril'sk-type; opentriangles- Lower Talnakh-type;circles
- dolerites;open diamonds- differentiated,unmineralizedintru-
sions[afterHawkesworthet al., 1995]. Other datafrom Lightfoot
et al. [ 1990, 1993, 1994].
elevated La/Sm, Th/La, and SiO2 and the reductionin
Ti/Zr, P/Zr, and Ta/La involveda granodioritecomponent
and took place earlierthan any additionof sulphurfrom
the evaporites. Theseare the contamination processes as-
sociatedwith the observeddepletionin Ni, Cu, and PGE
(e.g., Figures7, 9 and 10), and so it is inferredthat initial
sulphideformationwastriggeredby the resultantincreases
in SiO2 ratherthanby the additionof significantquantities
of crustalS. Hawkesworthet al. [ 1995] notedthat Sr iso-
topic ratios in the Noril'sk lavas increasedboth with in-
creasingLa/Sm and with no changein La/Sm (Figure 12),
suggestingthat the latter might reflectthe high-leveladdi-
tion of evaporite-derived material. The wide rangesof Sr
in the Nadezhdinsky(35-518 ppm), in which Sr is decou-
pled from the otherminor andtraceelements,may alsore-
flect the additionof crustalSr from the evaporites(P. C.
Lightfoot et al., ms. in preparation). However, such
changesappearto have been secondaryto the crustalcon-
taminationprocessesassociated with the markeddepletion
in the metalcontentsof the Nadezhdinskylavas.
Modelsfor the CWGT infer that, like the Nadezhdinsky,
the CWGT were generatedby crustal contaminationof
nearly picritic magmaswith 14-16 wt% MgO [Pedersen,
1985b;Lightfootet al., in press]. However,the picritesof
West Greenland have much clearer affinities with oceanic
basalts(e.g., high Ta/La and high end) and there is com-
pelling evidence for at least two crustal contamination
trends(Figure 10). Contaminationappearsto have

datedsignificantfractionationof feldsparand mafic min-
erals,becauseolivine in the CWGT on Qeqertarssuaq was
in equilibriumwith a high-Mg parentalmagma,and yet it
showsstrongdepletionin Ni [Pedersen,1985b], consistent
with equilibrationof the magma with sulphidesprior to
olivine crystallisation [e.g., Lightfoot et al., in press].
However, the picrites sampled in West Greenland show
much lessevidencefrom both major and traceelementsfor
interactionwith lithosphericmaterial than the Tuklonsky
picritessampledat Noril'sk (Figures9-11, 13). The field
relationsshow that the Qeqertarssuaq lavas were erupted
onto Archeangneissesand over a thin onlappingsequence
of epicontinentalshales. Some of the CWGT contain
fragmentsof bituminousshale [Pedersen,1979], and on
Nuussuaqthere are almost 5 km of shaleswhich contain
significanthydrocarbons[Pulvertafi, 1989].
All the CWGT are characterised by lower Ta/La, Ti/Zr,
and P/Zr, and •Nd and higher 87Sr/86Srthan the West
Greenlandpicrites(e.g., Figure 11). In detail, two trends
fan out from the fieldsof the picrites:trend 1 is to elevated
Th/La (and Th/Nb or Th/Ta) and relatively low end with
increasingLa/Sm, and trend 2 is to less elevatedTh/La,
higher end values and higher La/Yb and Zr/Y at higher
La/Sm than trend 1 (Figure 10). The crustalcomponents
responsiblefor thesetwo trendsare the subjectof con-
tinuing debate,but the high Zr/Y and La/Sm at moderate
Th/La may be more consistentwith an igneousProterozoic
metabasiccontaminantfor trend 2. In contrast,Th/La ra-
tios are not readily fractionatedby igneousprocesses,par-
ticularlyin the uppermantle,and so the high Th/La ratios
of trend 1 are stronglyindicativeof a sedimentarycon-
taminant[e.g.,Hawkesworthet al., in press]. The different
end valuesfor trends 1 and 2 are reflectedin different
modelNd agesof 2.6 Ga and 1.1 Ga for the mostcontami-
natedsampleson trends1 and 2, respectively.Pb isotopic
compositions indicatethat both crustalcontaminants had
unradiogenic Pb isotopicratios,but thatthe time-integrated
Th/U of the contaminantresponsible for trend 1 washigher
thanthatresponsible for trend2 [Lightfootet al., in press].
Spatially,trend 1 is dominatedby data for the Asfik-
Kfig/tnguaqand the Maligat rocks from Qeqertarssuaq,
whereasdata for Nuussuaqscatteracrossto trend 2, to-
getherwith just threesamplesfrom Qeqertarssuaq (Figure
10).
4.3. Modelsfor Mineralization
A number of observations bear on models of crustal
contamination and the interaction between the contami-
natedmagmasand sulphidesin Noril'sk and West Green-
land.
LIGHTFOOT AND HAWKESWORTH 371
0.8 Tk
p
0.7 As•k-
Kt2gdnguaq
0ø6
0.5
Mr-Sm
I
Mallgat
0.4 I I I
45 50 55
SiO2 (wt%)
Figure 13. Plot of Mg-numberversusSiO2 contentfor the lavas
of Noril'sk and West Greenland. Data from Lightfoot et al.
[1990, 1993, 1994, in press,ms. in preparation];Hawkesworthet
al. [1995]; Woodenet al. [1993]. Shadedfields - Noril'sk; blank
fields - West Greenland;abbreviationsas for Figure 9. In the
CWGT the lowestCu/Zr ratiostend to be developedin the rocks
that haverelativelyhigh SiO2at high Mg-number.Mg number=
Mg2+/Mg2+ + Fe2+)assuming that85% of thetotalFe is Fe2+.
(i) SiO2 more markedly with decreasingMgO in the
picrites of Noril'sk than in those from West Greenland.
This difference can be seen in the shallower trends, and at
higher SiO2, for the Noril' sk picriteson a diagramof Mg-
numbervs. SiO2 (Figure 13).
(ii) The lavas that exhibit the strongestevidence for
having equilibratedwith sulphides(i.e. low Cu/Zr in the
Nadezhdinskyand CWGT, Figure 9) are thosewhich ap-
pear to have experiencedthe largest amountsof crustal
contamination. Moreover, at least some of the contami-
nants inferred from minor- and trace-element ratios are
likely to have had high SiO2 but low S contents;for exam-
ple, granodiorite.It is inferredthat this contaminationwas
early and that the associatedsulphursaturationwas very
largelywith magmaticS.
(iii) Cu/Zr ratiosare more scatteredin the Nadezhdinsky
than in the CWGT, suggestingthat the interactionwith
sulphideswas more variablein the former.
(iv) In West Greenlandthe contaminatedunits are -100
m thick, they are sandwichedbetween relatively uncon-
taminatedbasalts(Figures 8 and 14a), and there are sharp
chemicalbreaksat the topsand bottomsof theseunits. In
contrast,at Noril'sk the rocks that are highly depletedin
Cu are-200 m thick, and abovethem thereis a gradualin-
creasein Cu/Zr and decreasein La/Sm with stratigraphic
heightin the upperunits of the Nadezhdinskyand the
372 FLOOD BASALTS AND MINERALIZATION

A WEST
GREENLAND B SIBERIAN
TRAPS

"- L 1

Low i• • High
I I

La/Sm
--• Low LeISm.........
-•- High
SWITCHING CulZr
• CulZr •
MAGMA
CHAMBER CONDUIT Contaminated
MODEL Contaminated MODEL lava flow
ß
-•.;-•.•.,.-• ...... -:.-•--.,•-•:•
;•:-•
•
•:•:;,;.•-.•:•;•:•:•:•:•.•.•:•:•:•
•• lavaflow .........
•• .................
:?:::•::•
...........
:::::::::::::::?:?:?::::
.....
' ''
P•cnt•c
flows
•
•r•
• -- :•:::::.....•
Contami
nation
by )•! .. shale or
• • .• •• :::.•:•::.•.•:•:••
•tonalitic •" '"'••'•'"""""""""""•••_
__

• .....
•::::•?.•::?:?:•::?:•?:•?::•
......
crust :•.:•
••
.................
Evapor
e
'••-•:••:•:/ hide • '"•Contaminati
(3)•
Figure 14. Sketchsectionsillustratinghow the differencesin Cu/Zr andLa/Smwith heightin the lavapilesat West
GreenlandandNoril' sk, may reflectdifferencesin theirmagmaplumbingsystems.

rongovskyand Mokulaevsky Formations(-1000 m, Fig-
ures 7 and 14b).
The variations in Mg-number with SiO2 indicate that
those rocks with low Cu/Zr (the Nadezhdinskyand the
CWGT) have SiO2 contentsthat are typically 5% higher
than those on, for example, the fractionationtrend of the
uncontaminated West Greenland picrites (Figure 13).
Such increasesin SiO2 in the Asfik and Kfig/tnguaqrocks
of West Greenland are consistent
tion of a crustalcontaminantthat had 65-70% SiO2, which
in turn can explain the observedincreasesin La/Sm and
reductionin Ta/La (Figure 11). In the contextof this dis-
cussion,an increaseof 5 wt% SiO2 appearsto have been
sufficientto move the magma compositioninto the field of
sulphursaturation(Figure 2 and Irvine [1975]), and so we
concludethat sulphursaturationcan be achievedby crustal
contaminationwithout the necessityfor the addition of
significantamountsof crustalsulphur. The picritic paren-
tal magmas were sulphur-undersaturated, or they too
would have fractionatedCu/Zr ratios. If suchpicritescan
containup to -2000 ppm dissolvedS, as reportedfor non-
picritic basaltsfrom the Wanapumof the ColumbiaRiver
Basalts[Thordarson, 1995], they are only requiredto re-
flect more than 10% partial melting of an upper mantle
sourcecontaining200 ppm S [see discussionby Keays,
1995].
with 20-25% assimila- Models for the genesisof the giantNi, Cu, and PGE sul-
phide depositsat Noril'sk highlight the importanceof
equilibrationof the sulphideliquid with largevolumesof
magma. Whereas equilibrationhas traditionallybeen hy-
pothesizedto occur in deep crustalmagma chambersby
eithersettlingof the denseimmisciblesulphidethroughthe
magma column [Naldrett et al., 1992] or batch equilibra-
tion in a deepchamber[Bragmannet al., 1993], equilibra-
tion may also occur in narrow, horizontal, open-system
magma chambers termed chonoliths [Rad'ko,

Naldrett et al., 1995; Torgashin,1994]. One versionof
sucha model for Noril'sk is illustratedin Figure 15. The
early Tuklonsky lavas sufferedrelatively little crustalcon-
tamination,and suchmagma is inferredto have been pa-
rental to the more contaminatedmagmas. Crustal con-
tamination primarily involved relatively deep-seatedgra-
nodioriticmelts, andthe resultantincreasein SiO2 was suf-
ficient to trigger sulphursaturation. Minor and trace ele-
ment data may then be used to evaluatethe likely subse-
quent contribution from the shallow-level evaporite se-
quence,in that the evaporiteshave 2000-2500 ppm Sr and
Sr/Sm of-1500 (P. C. Lightfoot et al., ms. in preparation).
Becausethe Nadezhdinskylavashave an averageSr/Sm of
-60, they cannothave accommodatedmore than 0.5% of
evaporite;nonetheless,becausethe evaporitescontain-20
wt% S, the contaminatedlavascouldhave receivedup to
1000 ppm S from the evaporitesequences.
The sulphideliquid is denseand so in this modelit could
not be readily expungedfrom the chonolith[Naldrettet al.,
1995]. Rather, the chonolithacted as an open-system
magma conduitthroughwhich successive batchesof sili-
cate magma travelled,to be eruptedat the surfaceas the
observedCFB sequence.Simplistically,we envisagethat
the initial magmasequilibratedwith the sulphideliquid
and were scavengedof Ni, Cu, and PGE to p•'oducethe
Ni-, Cu-, and PGE-poor lowermostlava flows of the
Nadezhdinsky. However, a distinctive feature of the
Noril'sk depositsis thatthe oresareunusuallyrich;for ex-
ample, Cu andNi contentsare more than twice thosein the
oresat Sudbury,Pt more than six times, and Pd more than
24 times the Sudburyore levels(Figure 1 and Lightfoot
[ 1996]). The implicationis thatthe sulphidesequilibrated
with very largeamountsof magma(2500-5000 timestheir
mass[Bragmannet al., 1993]) andthatsubsequent batches
of magmafurtherequilibratedwith the sulphideliquid as
they passedthroughthe chonolithenrouteto the surface.
As progressively more magmapassedthroughthe chono-
lith, the degreeof metal depletiondeclinedand the lavas
producedat the surfaceprogressively recoveredtheir si-
derophileelementcontents[Naldrettet al., 1995] (Figure
14). Apparently,the capacityof the sulphideliquid to
scavengeadditionalmetals decreasedwith time and some
partsof the sulphideliquidmay havebecomeisolatedfrom
the magma. This mechanismis importantas it explains
why sucha largevolumeof high-gradesulphideminerali-
zation is developedat Noril'sk, and it accountsfor the
continuousupward recoveryin the siderophile-element
abundances of the Nadezhdinsky throughMorongovsky 4.4. SulphurBudget
flows.
In detail, some aspectsstill have to be reconciledwith
suchmodels. For example,it is not entirely clear that
densemagmaticsulphidesat the baseof a chonolithcan
LIGHTFOOT AND HAWKESWORTH 373
easily react with fresh influxes of silicate magma, and
there are step-wisevariationsin Cu and Ni contentswithin
the Nadezhdinsky lavas [Fedorenko et al., 1996; P. C.
Lightfoot et al., ms. in preparation]which are not yet ex-
plained. A further concernis that in the chonolithmodel
very largevolumesof magma(>5000 km3) arerequiredto
havetravelledthroughvery small (200 m wide) intrusions.
Suchfactorswere highlightedby Czamanskeet al. [ 1995],
who suggestedthat crustal contaminationtook place at
greater depthsand that the sulphideswere subsequently
intrudedintotheirpresentpositions.
In West Greenland,the sulphide-ladengabbrosof the
HammersdalComplex and the Igdlukunguaqdyke demon-
stratethat sulphursaturationdid occur [Lightfootet al., in
press]. Furthermore,many of the gabbroicintrusionsare
located proximal to faults through which the lavas were
eruptedand thereforemay have had the opportunityto be-
come chonoliths as envisagedfor the Noril'sk systems.
The recognition of contaminatedlavas which have rela-
tively low siderophile-element contentsadds further evi-
denceto suggestthat the CWGT equilibratedwith mag-
matic sulphides. These are all featuresthat encourageex-
plorationeffortsin West Greenland. However,the absence
of a progressiverecovery in siderophile element abun-
danceswithin the CWGT (Figure 14) doesnot correspond
with the model presentedfor Noril'sk, and the lack of re-
covery may indicatethat continuedupgradingof any sul-
phides was not a feature of the West Greenlandmagma
systems. At issue,therefore, is whether such evidence
from the lavasis diagnosticof sulphideupgrading,as seen
at Noril'sk, and whether it shouldbe an importantaspect
of any exploration strategy based on basalt che-
mostratigraphy.At Noril'sk there was opportunityfor the
magmato assimilateadditionalsulphurfrom the evaporitic
sedimentsand large amountsof magma were availableto
continueto equilibratewith largevolumesof sulphide,and
theseare both conditionsfor the formation of giant depos-
its. In the West Greenland CFB, the amount of sulphur
available was moderate, although contamination with
sedimentshas been demonstrated(trend 1, Figure 10) and
shaleslocally contain 0.8 wt% S. Some lavas have been
depletedstronglyin Ni, Cu, and PGE, presumablyby in-
teractionwith magmaticsulphides,but there is no evidence
that the siderophile-element contentsof the sulphidesever
reachedthe point at which the siderophileelementabun-
dancesin the tholeiiticmagmasprogressivelyrecovered.
A critical controlon the sulphide-formingcapacityof a
magma is the amountof sulphuravailablewhich can then
interactwith silicate magma and concentratethe Ni,
374 FLOOD BASALTS AND MINERALIZATION

STAGE 1: Tk

Epicontinental sediments

Figure 15. Schematicmodelfor the evolutionof the Noril'sk system.

Stage1: Picriticandtholeiiticlavasof the TuklonskyFormationwere eruptedthroughconduitsystems100 km eastof

Noril'sk in the regionof the ImangdaFault (Figure 2).

Stage2: Eruptionof lavasswitchedto theNoril'sk Region,andinitiallyfocussed alongandto the westof theNoril' sk-
KharaelakhFault [Naldrettet al., 1992]. Contaminatedlavasof the Ndl were generatedby interactionof Tuklonsky
magmatype with granodiorites, with fractionalcrystallization.Thesemagmastravelledthroughthe chonolithand un-
derwenta secondstageof contamination involvingassimilation of evaporiticsediments.The combinationof elevated
silicacontentsand abundantsulphurresultedin the fractionationof immisciblesulphides.Thesesulphidespondedat
the baseof the chonolith,andreactedwith new batchesof Ndl magmaastheypassedthroughthe chamber.

Stage3: Migrationof Ndl magmasoverthe sulphides continued,andprogressively upgraded the sulphideliquidsin

Ni, Cu andthe PGE. Subsequent Nd2,Ndl midMr magmastravelledthroughthe conduitandasthe sulphides became
repletewith metalsandisolatedfrom the magma,the degreeof depletionof the magmasin the metalsdeclinedto pro-
ducethe observedupwardincreasein Ni, Cu, and PGE abundances with stratigraphic
height[e.g.,Naldrettet al.,
1992; Briigmannet al.. 1993].

Stage4: The Noril'sk conduitshutdownandmagmatism migratedtowardsthenortheast.The sulphideliquidsunder-

went fractionationto formthe Cu-richores[Czamanske
et al., 1992;Naldrettet al., 1995;Zienteket al.,

and PGE. Assimilationof mostcrustalrock typesis likely
to dilute ratherthan enhancethe sulphurcontentsof mafic
magmas,and so modelsfor the formationof giant sulphide
depositstypically require the introductionof sedimentary
sulphurfrom compositionallyunusualsedimentsand/or a
very efficient segregationand concentrationmechanismto
ensure the isolation of large volumes of mantle-derived
sulphurfrom the magmas.
The bestevidencefor the sourceof the sulphuris argua-
bly from sulphurisotopes,and the ores at Noril'sk typi-
cally have 534Svaluesof +8 to +14 per mil, compared
with 0 per mil in mantle-derivedrocksand +20 per mil in
the evaporitic sediments [Grinenko, 1985]. Grinenko
[1985] further noted that the barren intrusions in the
Noril'sk region tend to have sulphurisotopiccompositions
closeto mantlevaluesor displacedto slightlypositiveval-
ues (534S = -2 to +6) and tend to be locatedin clastic
sedimentsdevoid of crustal S. In contrast, the economi-
cally mineralizedintrusionshave more positive534Sand
tend to be located in crustal rocks that contain abundant
sulphatesin the form of evaporites. Grinenko [1985]
thereforesuggestedthat much of the S was derivedin situ
from the wall rocks. Using a mantlesulphurisotopiccom-
position of zero and assumingthat the evaporiteshave
534S- +20, Grinenko[ 1985]showedthat20-36% of the S
in the noneconomic mineralized intrusions could have
been from a crustal sourceand that the parentalmantle-
derivedmagmashad 1200-2300 ppm S. However,usinga
similar approachfor the mineralizedintrusions,the calcu-
lated initial S contentsof the parentalmagmaswere higher
than could have been dissolved in the observed volumes of
silicate magma. Reversing these calculationssuggested
that the crustal S in the mineralized intrusions had 534S
-+11 to +12, and Grinenko [1985] therefore concluded
that contamination of the mineralized intrusions involved
sulphur source at a deeper crustal level. The preferred
sourcewas sour gas S, becauseit was known to be com-
mon in the areaand has 534S-+ 10. However,suchcalcu-
lations predate the sulphide upgradingmodels in which
magmaticsulphidesinteractwith large volumesof mantle-
derived magma [Naldrett et al., 1995]. In thesemodels,
the massbalanceconsiderations no longerrequirethe pa-
rental magmasto have had anomalouslylarge dissolvedS
contents,
andhencethe crustalS component
canhave5348
values similar to the evaporitic sediments,rather than to
the sourgas [cf. Naldrett, 1989].
In West Greenland, mineralized intrusions which are
comagmaticwith contaminatedtholelitesfrom Nuussuaq
(the Igdlukunguaqdyke and the HammersdalComplex)
havenegative534Svaluesof-7 to -11, whichdemandsa
crustalcontribution
of S with negative534Sif the parental
LIGHTFOOT AND HAWKESWORTH 375
picritic magmashad a zero per mil signature. Interest-
ingly, one sampleof a thick, simpleflow from the Asfik
unit has 5348 = +13.6, which contrastswith most of the
otherCWGT, which have 534S= -5 to +5, and is consistent
with the additionof sedimentaryS from a shalesource.
Based on the sulphurisotopearguments,it is apparent
that crustal sulphur has contributedto mineralizationat
both Noril'sk and West Greenland. However, the extent to
which the addition of crust-derived S affected the mineral
potentialof the intrusionsstill needsto be explored.
The Ni- and Cu-depleted lavas of the Nadezhdinsky
Formationat Noril'sk have a volumeof-5,000 km3, based
on isopachmaps of their surfacedistribution[Naldrett et
al., 1992]. The available data from the Columbia River
Basalts [Thordarson, 1995], experimentalstudiesof the
dissolvedsulphur contentsof tholeiitic melts (500-2000
ppm S for melts with 5-25 wt% FeO) [Buchananand No-
lan, 1979; Haughton et al., 1974], and submarinepillow
basalts(1050-1800 ppm S in rockswith 9-12.9 wt% FeO)
[Mathez, 1976] are consistent with high-Mg mantle-
derivedmelts containing-1900 ppm S. However, basaltic
magmasmay typically have S contentsof half that value
(i.e., 950 ppm), and so the bestestimateof the amountof S
in the parentalNadezhdinskymagmasprior to saturationis
4.7 km3, which at 1.8 g/cctranslates
into 8.5 x 109tonnes
of elementalS derivedfrom the uppermantle.
The depositsat Noril'sk contain 555 million tonnes of
ore (and perhapsasmuch as 1500 Mt). Ore gradesaverage
3.9 wt% Cu, and 2.7 wt% Ni, and mineralizationis typi-
cally 50% gangue. If 50% of the rock is FeS, then the 555
million tonneswouldcontainabout17% S andthe deposits
as a whole would contain 1 x 108 tonnes of S. This is
likely to be a minimumvalue,but it indicatesthat only 1-
2% of the magmaticS availableis lockedup in the depos-
a its at Noril' sk.
The amountsof crustalS assimilatedin the magmasare
more difficult to constrain,partly becausethe lavasvary
flow by flow and partly becausesulphurisotopedataap-
pearto be in conflictwith the traceelementarguments.As
discussed above, the trace element contents of the
Nadezhdinsky lavas suggestthat no more than 0.5%
evaporite was assimilated in even the most Sr-rich
Nadezhdinskyrocks. Recent work by Czamanskeet al.
[1995] confirmsthat the geologicalevidenceat Talnakh
favoursthe replacementof the evaporite-richsedimentary
packageby theTalnakhintrusion(<10 km3)ratherthanby
the staticrupturingand dilationof the sediments.Assum-
ing that the intrusionreplaceda similar volume of sedi-
ments, the amount of S added from those sedimentscan be
estimatedcrudely. Thus, the amountof crustreplacedis
---10-20 x 109 tonnes,and if it containedan average
376 FLOOD BASALTS AND MINERALIZATION
25% S, the amountof crustalS contributedto the systemis
4 x 109 tonnes. Suchquantitiesof S are consistent with
the S isotoperatios of the magmaticsulphides[Grinenko,
1985], and they representalmosthalf of the S availablein
the magmasthemselves. The total amount of S available
in the Noril'sk systemis thereforeestimatedto be-12.5 x
109tonnes,andyet <1% appearsto havebeenscavenged
by known mineral deposits.Finally, it is importantto reit-
erate that the shallow-levelevaporite-richsedimentswere
not responsiblefor the trace element and radiogeniciso-
tope variationsthat are associatedwith the observedde-
pletionsin Cu, Ni, and PGE in the crustallycontaminated
lavasand intrusions(Figures7, 9 and 10). Rather,it is in-
ferred that the initial sulphide formation took place at
depthin responseto crustalcontamination with a granodio-
ritic component(Section 4.2, and Lightfoot et al. [1993,
1994]; Hawkesworthet al. [1995]). The unresolvedques-
tion concernsthe detailedrelationsbetweenthe sulphides
inferredto have separatedat depthand thosepresentlyob-
servedin the shallow-levelintrusionswithin evaporite-rich
sediments.
In West Greenlandshalesfrom Qeqertarssuaq have an
averageS contentof 0.8% S, and the gneisseshave an av-
erage of 82 ppm S. The geochemicaldata discussedin
section4.2 suggestthat the Asfik and Kfigfinguaqflows
were contaminatedwith up to 30% shale,implyingthat the
contribution of crustal S in these rocks is 0.3 x 8000
2400 ppm S addedto the magma. This amountis similar
to that inferred for the S contents of the mantle-derived
magmas (-1900 ppm S), indicating that roughly equal
amountsof mantle and crustallyderivedS may have been
presentin the contaminatedlavas and associatedcomag-
matic intrusions. Assumingthat the Asfik and Kfigfinguaq
have a volume of-103 km3, the amountof crustalS added
was 2 x 109 tonnes, and there must have been a total S
budgetof over 4 billion tonnesavailablefor mineralising
processes. The observation that the mineralization at
IgdlukunguaqandHammersdalhave•534S valuesof-7 to -
12 providessomesupportfor the suggestion that abouthalf
of the S was crustalin origin andthatthe restwas mantle-
derived.
In the model discussedhere, one implicationis that al-
though very large sulphidedepositsmay exist in West
Greenland,they may not be of the gradefoundat Noril' sk.
A particular challengenow will be to use PGE data for
these rocks to ascertain whether these elements, which
have much higher sulphide/silicatepartition coefficients,
recorda similar recoverypatternandto look in more detail
at the Maligat CWGT, which appearto be comagmatic
with someof the observedmineralizedintrusions.Signifi-
cantly,thesepatternsof gradualrecoveryin the Ni, Cu and
PGE contentsin the lavas are not unique to the Siberian
Traps,and good examplesthroughover 1000 m of basalt
stratigraphyhave been recordedin the Osler Volcanic
Groupin the Keweenawanof Ontario,Canada[Lightfootet
al., 1991; Naldrett and Lightfoot,1993]. The potentialof
intrusionsin the Keweenawan region should therefore
continueto intrigueexplorationistsfor sometime to come.
5. SUMMARY
There is a clear association
betweensulphideminerali-
zation and someCFB. Two of the moststrikingexamples
are the Permo-TriassicSiberianTraps at Noril'sk and the
Tertiary CFB of West Greenland,and althoughthey share
a numberof commonfeatures,there are significantdiffer-
ences. The main pointsof this discussionmay be summa-
rised as follows.
1. Present exposuresof the Siberian Traps and West
GreenlandCFB containup to 3500 m of predominantly
picritic and tholeiitic lavas. However,at Noril'sk the pic-
ritesrepresent-1% of the preservedsequence,whereasin
West Greenland they comprise >50%. In both West
Greenlandand the Siberian Traps, magmatismhas been
linked to the presenceof mantlehot spots. However,the
volumesof magmaderivedfrom asthenospheric and litho-
sphericsourcesappearto havebeendifferentin the two ar-
= eas.
2. Intrusionscontemporaneous with volcanismwere de-
velopedin boththe SiberianTrapsandin WestGreenland.
Many of theseintrusionscanbe correlatedchemicallywith
units in the associatedlava sequences[Fedorenko,1994;
Naldrett et al., 1992; Hawkesworthet al., 1995], and some
may have acted as open-systemconduits (chonoliths)
throughwhichthe lavaserupted.At Noril'sk the distribu-
tion of the intrusiontypesis linkedbroadlyto the location
of the maximumthicknessof comagmaticlava [Naldrettet
al., 1992]. A generalcoincidenceis that the mineralized
intrusionsare associatedwith major NNW-SSE-trending
mantle-penetratingfaults which appearto have acted as
conduitsfor the magmasand as loci for the intrusions. In
Siberia,the intrusionsthat carry mineralizationare associ-
ated with the Noril'sk-Kharaelakh Fault, and in West
Greenland,numerousdiabaseand gabbro intrusionsare
linked to the BoundaryFault.
3. The CFB of the Noril'sk region and West Greenland
were erupted onto epicontinentalsedimentarysequences.
At Noril'sk theseinclude evaporiticsedimentsand deeper
reservoirsof sour gas and oil. In West Greenland,the
magmatism occurred over a progradingdeltaic assem-
blage, such that magmaswere eruptedonto basement,a
shallow sequenceof hydrocarbon-richshales,or a
 LIGHTFOOT AND HAWKESWORTH 377

sequence (>5 km) of hydrocarbon-bearingshales 7. Many of the intrusionsat Noril'sk may have actedas
[Putvertafi, 1989]. chonoliths,and theseconduitscan be regardedas the sites
4. Mineralization at Noril'sk is associated with rela- where sulphidesformed and equilibratedwith successive
tively smallpicritic to gabbroicintrusions.The intrusions batchesof silicateliquid passingthroughthe CFB conduit.
contain 50-200 times more sulphidemineralizationthan The high gradesof the oresat Noril'sk are consistentwith
can have been generatedby in situ settlingof immiscible upgradingof the depositsby the equilibrationof the sul-
sulphideliquid from the silicateportionof the intrusions, phide liquids with the successivebatchesof magma. In
andso theymustreflectscavengingof S from significantly this zone-refiningprocess,a gradualrecoveryin the Ni,
larger magma volumes. In addition, the sulphidesare Cu, and PGE contentsof the later silicatemagmasis pre-
enormouslyrich in ore metals;productionplusreservesof dicted,and this recoveryis an observedfeatureof the lava
>555 million tonnes(and possibly 1000 million tonnes) sequence[Lightfootet at., 1990, 1993, 1994; Bragmannet
total 3.9 wt% Cu, 2.7 wt% Ni, 3 ppm Pt, and 12 ppm Pd, at., 1993; NaTdrett et aT., 1992, 1995]. The West Green-
which is 2-10 timeslargerthanin othergiantdepositslike land lavas contain discreteunits of heavily contaminated
Sudbury [Lightfoot, 1996]. Mineralization in West and Ni- and Cu-depletedlavas, and thesetoo can be as-
Greenland is spatially associatedwith the Hammersdal cribed to equilibrationof contaminatedmagma with sul-
Complex and the Igdlukunguaqdyke. Theseoccurrences phide liquid. However, no gradualupwardrecoveryin si-
have smallsulphidecontents,but the sulphideshavemod- derophile-element abundances is evident;rather,the top of
erate to high Ni, Cu, and PGE abundances[Utff Motter, the package is marked by the suddenadjustmentof the
1991]. systemwith the expulsionof picritic lavas. This lack of
5. Within the lava sequences at both Noril'sk and West recoverysuggests that the dynamicsof the conduitsystem
Greenland,a broadempiricalrelationexistsin whichlavas in West Greenland were quite different from those at
with highSiO2,LILE, La/Sm,Th/Nb,and87Sr/86Sr o also Noril'sk, and suchdifferencesin conduitsystemdynamics
have low Cu, Ni and Cu/Zr [e.g., Lightfoot et at., 1990, may have implicationsfor mineral exploration. Finally,
1993, in press;Woodenet at., 1993]. Thesefeaturesindi- we note that the featuresof the Noril'sk region are alsoob-
cate that the contaminatedlavas tend to have abnormally servedin the Keweenawanmidcontinentriff in a sequence
low siderophileelement concentrations, and these low of contaminatedlavas(Osler Volcanic Group) which have
concentrations have typically been ascribedto the equili- overall low Ni/MgO and an upwarddeclinein the degree
brationof the magmawith a densesulphideliquid which of contamination [Lightfoot et at., 1990; Natdrett and
was removedfrom the system. Lightfoot, 1993]. The similaritiesbetweenthis sequence
6. The minor and trace element variations in the and that at Noril'sk are particularlyencouragingfrom the
Noril'sk lavasindicatethat the contaminantwas granodio- perspectiveof mineralexploration.
ritic in composition[e.g., Lightfoot et aT., 1993, 1994;
NaTdrett et at., 1992; Hawkesworth et at., 1995]. Such Acknowledgments.The authorsare indebtedto Falconbridge
rocks have low S contents, and so we conclude that the Explorationwho supportedthe costof our programon the West
initial sulphideformationresponsible for the low Ni, Cu GreenlandCFB. We particularlyappreciatethe advice and as-
sistancefrom Kevin OlshefskyandTony Greenwith this research
and PGE contentsof the contaminatedlavaswas triggered
programand discussions with Reid Keays. We are alsograteful
by the associated increasein SiO2ratherthanthe introduc- for the encouragement and supportof Tony Naldrett,Valeri Fe-
tion of additional(crustal)S. Subsequent interactionwith dorenko, and Nick Gorbachevwith the Noril'sk aspectof this
shallow level evaporite-richsedimentscontributedaddi- study. Much of our work relieson the high-qualitydataacquired
tional S, but it was not responsible for the crustalcontami- by Will Dohertyof the OntarioGeologicalSurveyandthe Geo-
nation associatedwith the distinctivemetal depletionsin logicalSurveyof Canada. The manuscript wasimprovedby the
the Nadezhdinskylavas. Mass balance considerations constructivecommentsof Gerry Czamanske,Bruce Doe, Valeri
suggest thataslittle as 1% of the S availablein themantle- Fedorenko,andJohnMahoney. The manuscriptwaspreparedby
derivedmagmasand the assimilatedevaporite-richsedi- JanetDryden. Diagramsweredraftedby SteveJosey.
mentsat Noril'sk is lockedup in the known deposits.In REFERENCES
the West Greenland rocks, two trends of contamination
with different crustal contaminantsare recognised,and
Aplonov, S., An abortedTriassicoceanin West Siberia, Tecton-
they exhibit similar depletionsin Ni and Cu. As at ics, 7, 1103-1122, 1988.
Noril'sk, initial sulphideformationwould appearto have Bragmann,G. E., A. J. Naldrett, M. Asif, P. C. Lightfoot, and N.
been primarily controlled by the associatedincreasein S. Gorbachev,Siderophileand chalcophilemetalsas tracersof
SiO2 ratherthan by the S contentsof the crustalcontami- the evolution of the Siberian Traps in the Noril'sk Region,
nants. Russia, Geochim. Cosmochim.Acta, 57, 2001-2018,
378 FLOOD BASALTS AND MINERALIZATION
Buchanan,D. L., and J. Nolan, Solubility of sulfur and sulfide
immiscibilityin synthetictholeiiticmeltsand their relevanceto
BushveldComplexrocks,Can. Mineral., 17, 483-494, 1979.
Campbell, I. H., and A. J. Naldrett, The influence of sili-
cate:sulphideratios on the geochemistryof magmatic sul-
phides,Econ. Geol., 74, 1503-1505, 1979.
Campbell, I. H., G. K. Czamanske,V. A. Fedorenko,R. I. Hill,
V. Stepanov,and V. E. Kunilov, Synchronismof the Siberian
Traps and the Permian-Triassic boundary,Science,258, 1760-
1763, 1992.
Clarke,D. B., and A. K. Pedersen,Tertiary volcanicprovinceof
west Greenland,in Geologyof Greenland,editedby A. Escher
and W. S. Watt, pp. 365-385, Geological Survey of Green-
land, Copenhagen,1976.
Clarke, D. B., The Tertiary volcanic province of Baffin Bay,
GAC Spec.Pap., 16, 445-460, 1991.
Czamanske,G. K., V. E. Kunilov, M. L. Zientek, L. C. Cabri, A.
P. Likhachev, L. C. Calk, and R. L. Oscarson,A proton-
microprobestudyof magmaticsulphideoresfrom the Noril'sk
Talnakh district, Siberia, Can. Mineral., 30, 249-287, 1992.
Czamanske,G. K., T. E. Zen'ko, V. A. Fedorenko,L. C. Calk, J.
R. Budahn,J. H. Bullock, T. L. Fried, B.-S. W. King, and D.
F. Siems, Petrography and geochemicalcharacterizationof
ore-bearingintrusionsof the Noril'sk Type, Siberia:with dis-
cussionof their origin, Res.Geol., 18, 1-48, 1995.
Dalrymple, B. G., G. K. Czamanske,V. A. Fedorenko,O. N. Si-
monov, M. A. Lanphere, and A. P. Likhachev, A reconnais-
sance
40Ar/39Argeochronological
studyof ore-bearing
and
related rocks, Siberian Russia.Geochim.Cosmochim.Acta, 59,
2071-2083, 1995.
Devey, C. W., and P. C. Lightfoot, Volcanologicaland tectonic
control of stratigraphyand structurein the westernDeccan
Trap. Bull. Volcanol.,48, 195-207, 1986.
Fedorenko,V. A., The petrochemicalseriesof volcanicrocksof
the Noril'sk region.Geol. Geophys.,6, 78-88, 1981.
Fedorenko,V. A., Evolution of magmatismas reflectedin the
volcanicsequenceof the Noril'sk Region,in Proceedingsof
the Sudbury-Noril'skSymposium,Spec. Vol. 5, editedby P.C.
Lightfoot and A. J. Naldrett, pp. 171-184, OntarioGeological
Survey,Toronto, 1994.
Fedorenko,V. A., P. C. Lighttbot,A. J. Naldrett,G. K. Czaman-
ske• C. JoHawkesworth, J. L. Wooden, and D. S. Ebel, Petro-
genesisof the flood-basaltsequenceat Noril'sk, North Central
Siberia, Int. Geol. Rev., 38, 99-135, 1996.
Gallagher,K., and C. J. Hawkesworth,Mantle plumes,continen-
tal magmatism and asymmetry in the South Atlantic, Earth
Planet. Sci. Lett., 123, 105-117, 1994.
Glazkovsky,A. A., G. I. Gorbunov,and F. A. Sysoev,Deposits
of nickel, in Ore Depositsof the USSR,Vol. II, editedby V. I.
Smirnov,pp. 3-79, Englishtranslationby D. A. Brown, 1977.
Grinenko, L. N., Sourcesof sulfur of the nickeliferousand barren
gabbro-doleriteintrusionsof the northwestSiberianplatform,
lnt. Geol. Rev., 695-708, 1985.
Hald, N., andA. K. Pedersen,Lithostratigraphy
of the EarlyTer-
tiary volcanicrocksof centralWest Greenland,Rapp. Gron-
landsgeol. Unders.,69, 17-24, 1975.
Hald, N., Lithostratigraphy
of the Maligat and Hare6enforma-
tions, West Greenlandbasaltgroup,on Hare6enand Western
Nugssuaq,Rapp. Gronlandsgeol. Unders., 79, 9-16, 1977a.
Hald, N., Normally magnetized lower Tertiary lavas on
Nugssuaq, central West Greenland,Rapp. Gronlands geol.
Unders., 79, 5-7, 1977b.
Hamlyn, P. R., R. R. Keays, W. E. Cameron,A. J. Crawford, and
H. M. Waldron, Preciousmetals in magnesianlow-Ti lavas:
Implications for metallogenesisand sulfur saturationin pri-
mary magmas, Geochim. Cosmochim.Acta, 49, 1797-1811,
1985.
Harris, N. B. W., J. A. Pearce, and A. G. Tindle, Geochemical
characteristicsof collision-zonemagmatism,in Collision Tec-
tonics,Spec.Publ. 19, editedby M.P. CowardandA. C. Ries,
pp. 67-81, GeologicalSocietyof London, 1986.
Haughton, D. R., P. L. Roeder, and B. Skinner, Solubility in
submarinebasalt glass, J. Geophys.Res., 81, 4269-4276,
1974.
Hawkesworth,C. J., P. C. Lightfoot, V. A. Fedorenko,S. Blake,
A. J. Naldrett, W. Doherty, andN. W. Gorbachev,Magma dif-
ferentiation and mineralisation in the Siberian continental
flood basalts,Lithos, 34, 61-88, 1995.
Hawkesworth, C. J., S. P. Turner, D. Peate, F. McDermott and P.
van Calsteren, Elemental U and Th variations in island arc
rocks: implications for U-series isotopes,Chem. Geol., in
press,1997.
Hergt, J. M., D. W. Peate, and C. J. Hawkesworth,The petro-
genesisof Mesozoic Gondwanalow Ti flood basalts,Earth
Planet. Sci. Lett., 105, 134-148, 1991.
Holm, P.M., R. C. O. Gill, A. K. Pedersen,J. G. Larsen,N. Hald,
T. F. D. Nielsen and M. F. Thirwall, The tertiary picritesof
West Greenland: contributions from Icelandic and other
sources,Earth Planet. Sci. Lett., 115, 227-244, 1993.
Irvine, T. N., Crystallisationsequenceof the Muskox Intrusion
and other layered intrusions:II Origin of the chromititelayers
and similar depositsof othermagmaticores,Geochim.Cosmo-
chim. Acta, 39, 991-1020, 1975.
Keays, R. R., The role of komatiiticand picritic magmatismand
S-saturationin the formationof ore deposits,Lithos, 34, 1-18,
1995.
Klock, W., H. Palme, and H. J. Tobschall, Trace elements in
natural metallic iron from Disko Island, Greenland, Contrib.
Mineral. Petrol., 93, 273-282, 1986.
Lightfoot, P. C., A. J. Naldrett, N. S. Gorbachev,W. Doherty,
and V. A. Fedorenko,Geochemistryof the SiberianTrap of the
Noril'sk Area, USSR, with implicationsfor the relativecontri-
butions of crust and mantle to flood basalt magmatism,Con-
trib. Mineral. Petrol., 104, 631-644, 1990.
Lightfoot,P. C., G. Chai, A. E. Hodges,and D. Rowell, An up-
date report on the certificationof the OGS in-houseMRB
standard-reference materials,in Summaryof Field Work and
Other Activities,Misc. Publ. 157, pp. 231-236, Ontario Geo-
logical Survey, Toronto, 1991.
Lightfoot,P. C., C. J. Hawkesworth,J. Hergt, A. J. Naldrett,No
So Gorbachev, and Vo A. Fedorenko, Remobilisation of the
continental lithosphere by a mantle plume: major-, trace-
elementand Sr-, Nd-, and Pb-isotopeevidencefrom picritic
andtholeiitic lavasof the Noril'sk District, SiberianTrap,

sia, Contrib.Mineral. Petrol., 114, 171-188, 1993.
Lightfoot, P. C., A. J. Naldrett, N. S. Gorbachev, V. A. Fe-
dorenko,C. J. Hawkesworth,J. Hergt, and W. Doherty,Che-
mostratigraphyof SiberianTrap lavas,Noril'sk District, Rus-
sia: implicationsfor the sourceof flood basaltmagmasand
their associatedNi-Cu mineralisation,in Proceedingsof the
Sudbury-Noril'sk Symposium,Spec. IZol. 5, edited by P.C.
Lightfoot and A. J. Naldrett, pp. 283-312, OntarioGeological
Survey, Toronto, 1994.
Lightfoot, P. C., The giant nickel depositsand Sudbury and
Noril'sk (abstract),1996 Prospectorsand DevelopersAssoc.
Canada,Nickel in a Nutshell Workshop,1-12, 1996.
Lightfoot, P. C., C. J. Hawkesworth,K. Olshefsky,T. Green,W.
Doherty and R. R. Keays, Geochemistryof Tertiary tholelites
and picritesfrom Qeqertarssuaq (Disko Island) and Nuussuaq,
West Greenlandwith implicationsfor the mineral potentialof
comagmaticintrusions. Contrib. Mineral. Petrol., in press,
1997.
Macdougall, J. D. (Ed.), ContinentalFlood Basalts, 341 pp.,
Kluwer Academic Publishers,Dordrecht, 1988.
MacLean, W. H., Liquidus phase relations in the FeS-FeO-
Fe304-SiO2 systemsand their applications
in geology,Econ.
Geol., 64, 865-884, 1969.
Mahoney,J., J. D. Macdougall,G. W. Lugmair,A. V. Murali, M.
Sankar Das, and Gopalan, K., Origin of the Deccan Traps
flows at Mahabaleshwarinferredfrom Nd and Sr isotopicand
chemical evidence, Earth Planet. Sci. Lett., 60, 47-60, 1982.
Mathez, E. A., Sulfur solubility and magmaticsulfidesin subma-
rine basaltglass,J. Geophys.Res.,81, 4269-4276, 1976.
Morgan, W. J., Hotspottracksand the openingof the Atlantic
and Indian oceans,in The Sea, Vol. 7, edited by C. Emiliani,
pp. 443-487, Wiley Interscience,New York, 1981.
Naldrett,A. J., MagmaticSulfideDeposits,196 pp., Oxford Uni-
versity Press,New York, 1989.
Naldrett, A. J., P. C. Lightfoot, V. A. Fedorenko,W. Doherty,
and N. S. Gorbachev,Geologyand geochemistryof intrusions
and flood basaltsof the Noril'sk Region,USSR, with implica-
tions for the origin of the Ni-Cu ores,Econ. Geol., 87, 975-
1004, 1992.
Naldrett, A. J., and P. C. Lightfoot, Ni-Cu-PGE ores of the
Noril'sk region, Siberia:A model for giant magmaticsulfide
depositsassociated with flood basalts,in Giant Ore Deposits,
Spec.PuN. 2, editedby B. H. Whiting, C. V. Hodgsen,and R.
Mason, pp. 81-123, Society EconomicGeologists,Littleton,
CO, 1993.
Naldrett, A. J. V. A. Fedorenko, P. C. Lightfoot, N. S. Gor-
bachev, W. Doherty, M. Asif, S. Lin, and Z. Johan,A model
for the formationof the Ni-Cu-PGE depositsof the Noril'sk,
Trans.Inst. Mining Metal., 104, B 18-B36, 1995.
Peach, C. L., E. A. Mathez, and R. R. Keays, Sulphidemelt-
silicate melt distribution coefficients for noble metals and other
chalcophileelementsasdeducedfrom MORB: Implicationsfor
partial melting, Geochim. Cosmochim.Acta, 54, 3379-3389,
1990.
Pedersen,A. K., New investigationsof the native iron bearing
volcanic rocks of Disko, central West Greenland, Rapp.
Gronlandsgeol. Unders.,75, 48-51, 1975.
LIGHTFOOT AND HAWKESWORTH 379
Pedersen,A. K., Non-stoichiometric magnesianspinelsin shale
xenolithsfrom a native iron-bearingandesiteat Asfik, Disko,
central West Greenland, Contrib. Mineral. Petrol., 67, 331-
340, 1978.
Pedersen, A. Ko, A shale buchite xenolith with A1-Armalcolite
and Native Iron in a lava from Asfik, Disko, Central West
Greenland,Contrib. Mineral. Petrol., 69, 83-94, 1979.
Pedersen,A. K., Lithostratigraphy of the Tertiary Vaigat Forma-
tion on Disko, centralWestGreenland,Rapp. Gronlandsgeol.
Unders.,124, 30 pp., 1985a.
Pedersen,A. K., Reactionbetweenpicritemagmaandcontinental
crust: early Tertiary silicic basalts and magnesianandesites
from Disko, West Greenland,Gronlandsgeol. Unders.Bull.,
15, 126 pp, 1985b.
Pedersen,A. K., and Pulvertaft, T. C. R., The nonmarine Creta-
ceous of the West Greenland Basin, onshore West Greenland,
Cret. Res., 13, 1-10, 1992.
Pulvertafi,T. C. R., Reinvestigationof the Cretaceousboundary
fault in Sarqaqdalen,Nugssuaq,centralWest Greenland,Rapp.
Gronlandsgeol. Unders., 145, 28-32, 1989.
Rad'ko, V. V., Model of dynamic differentiation of intrusive
traps in the northwesternSiberianplatform, Sovietgeol. and
Geophys.,32 (7), 70-77, 1991.
Renne,P. R. and A. R. Basu,Rapid eruptionof the SiberianTrap
Flood Basalts at the Permo-TriassicBoundary,Science,253,
176-179, 1991.
Richards, M. A., A. R. Duncan and V. E. Courtillot, Flood ba-
saltsand hot-spottracks:plume headsand tails, Science,246,
103-107, 1989.
Ripley, E. M., Sulphur isotopicabundancesof the Dunka Road
Cu-Ni deposit,Duluth Complex, Minnesota,Econ. Geol., 76,
619-620, 1981.
Shaw,D. M., J. Dostal, and R. R. Keays,Additional estimatesof
continentalsurfacePrecambrianshield compositionsin Can-
ada, Geochim. Cosmochim.Acta, 40, 73-83, 1976.
Simonov, O. N., V. A. Lul'ko, Yu. N. Amosov, and V. M. Salov,
Geologicalstructureof the Noril'sk District,in Proceedingsof
the Sudbury-Noril'skSymposium, Spec. IZol.5, editedby P.C.
Lightfoot and A. J. Naldrett, pp. 161-170, OntarioGeological
Survey,Toronto, 1994..
Smirnov, M. F., The Noril'sk nickeliferous intrusions and their
sulfideores,58 pp., Nedra Press,Moscow, 1966.
Taylor, S. R., and S. M. McLennan, The Continental Crust: its
Compositionand Evolution, 312 pp., Blackwell Scientific
Publications, 1985.
Thordarson,Th., Volatile releaseand atmosphericeffectsof ba-
salticfissureeruptions,Ph.D. thesis,580 pp., Univ. of Hawaii
at Manoa, 1995.
Torgashin,A. S., Geologyof the massiveand copperoresof the
westernpart of the Oktyabr'skyDeposit,in Proceedingsof the
Sudbury-Noril'sk Symposium,Spec. Fol. 5, edited by P.C.
Lightfootand A. J. Naldrett,pp. 231-242, OntarioGeological
Survey, Toronto, 1994.
Turner, S., and Hawkesworth, C. J., The nature of the continental
mantlelithosphere:constraintsfrom the major elementcompo-
sition of continental flood basalts, Chem. Geol., 120, 295-314,
380 FLOOD BASALTS AND MINERALIZATION
Turner, S., C. Hawkesworth,K. Gallagher,K. Stewart,D. Peate
and M. Mantovani,Mantle plumes,flood basalts,and thermal
modelsfor melt generationbeneathcontinents:Assessment of
a conductiveheating model and applicationto the Paranti,J.
Geophys.Res., 101, 11503-11518, 1996.
Ulff-Moller, F., Magmatic Pt-Ni mineralisationin the West
GreenlandBasalt Province: A compilationof the resultsof
prospectingby GreenexA/S in 1985-1988, Gronlandsgeolo-
giske Undersogelse,OpenFile, 91/1, 37 pp., 1991.
Wendlandt, R. F., Sulphide saturationof basaltsand andesite
melts at high pressuresand temperatures,Am. Mineral., 67,
877-885, 1982.
White, R., and D. McKenzie, Magmatismat rift zones:the gen-
eration of volcanic continentalmarginsand flood basalts,J.
Geophys.Res., 94, 7685-7729, 1989.
White, R., and D. McKenzie, Mantle plumesandflood basalts,J.
Geophys.Res., 100, 17,543-17,585, 1995.
Wooden, J. L., G. K. Czamanske, V. A. Fedorenko, N. T. Arndt,
C. Chauvel, R. M. Bouse,B. W. King, R. J. Knight, and D. F.
Siems, Isotopic and trace-elementconstraintson mantle and
crustal contributions to Siberian continental flood basalts,
Noril'sk Area, Siberia, Geochim. Cosmochim.Acta, 57, 3677-
3704, 1993.
Zientek, M. L., A. P. Likhachev, V. E. Kunilov, S.-J. Barnes,A.
L. Meier, R. R. Carlson,P. H. Briggs, T. L. Fries, and B. M.
Adrian, Cumulus processesand the compositionof magmatic
ore deposits:examplesfrom the Talnakh District, Russia,in
Proceedingsof the Sudbury-Noril'skSymposium,Spec. Vol. 5,
editedby P.C. Lightfoot and A. J. Naldrett, pp. 373-392, On-
tario GeologicalSurvey,Toronto, 1994.
C.J. Hawkesworth,Departmentof Earth Sciences,The Open
University, Milton Keynes,MK7 6AA, U.K.
P.C. Lightfoot, Departmentof Earth Sciences,The University
of Toronto, 22 Russell St, Toronto, Ontario, M5S 3B 1,
 Emplacementof ContinentalFlood BasaltLava Flows
I 2

StephenSelf, ThorvaldurThordarson,andLaszloKeszthelyi
Departmentof Geologyand Geophysics and Hawaii Centerfor Volcanology
Schoolof Oceanand Earth Scienceand Technology,Universityof Hawaii at Manoa, Honolulu,Hawaii

We proposethat continentalflood basalt (CFB) lavas were predominantly

emplacedas inflatedcompoundpahoehoeflow fields via prolonged,episodic
eruptions.Our most detailedobservationscome from the ---14.7Ma Roza flow
field of the Columbia River Basalt (CRB) Group. The Roza flow field seemsto
be typical of many flood basalt lavas.Individualflows show a wide range of
pahoehoesurfacefeaturesand a three-partinternalstructurein vesicularityand
othertexturalparameters.This three-folddivisioninto an uppercrust,core, and
basalcrustappearsto be diagnosticof the inflationprocessand is ubiquitousin
basaltic lava flows over a remarkablerange of sizes. The pahoehoesurface
featuresand indicationsof inflation are inconsistentwith rapid emplacementof
these lava flows. Instead,we interpretthe observationsto imply that the Roza,
and otherCFB flows, were emplacedover an extendedperiodof time. From the
thicknessof the upper crust, which we suggestformed while the flow was
actively inflating, and an empirical expressionfor the rate of crust growth of
Hawaiian inflated sheet flows, we estimate that individual Roza flows were
emplacedover 5 to 50 monthsandthat the Roza flow field was constructed over
a period of 6 to 14 years.However, even•withthis longereruptionduration,the
average lava effusion rate of-•4000 rn /s is similar to that of the highest-
effusion-rateeruptionin recordedhistory(the 1783-4 Laki eruptionin Iceland).
Our observations of lava characteristics in other CRB flows and in the Deccan
Traps suggestthat this emplacementstyle is typical of many, if not most,CFB
flows. Initial estimatesof the volatile releasefrom the Roza eruption indicate
that prodigiousamountsof S, C1,andF were injectedinto the uppertroposphere
and lowermoststratosphere; thusthis singleflood basalteruptioncouldhave had
a significanteffect on the global atmosphereIf other flood basalt eruptions
producedsimilar amountsof volatiles,volatile releasemight provide a link
betweenflood basalteruptionsand massextinctions.

1 Now at CSIRO,Exploration
andMining,PrivateBag,PO 1. INTRODUCTION
Wembley WA 6014, Australia.
2 Also at Hawaii Volcano Observatory,United States The physical volcanology of continental flood basalt
Geological Survey, P.O. Box 51, Hawaii Volcanoes National (CFB) lava flows has received relatively little attention
Park, HI 96718.
until recently. The emphasisof most previous research
on flood basalt provinces has been directed at defining
the gross chemical stratigraphy of the lava piles,
LargeIgneousProvinces:Continental,Oceanic,andPlanetary identifying different possible mantle sourcesand crustal
Flood Volcanism
or mantle contaminants for these huge volumes of
GeophysicalMonograph100
basalt, and investigating how these factors relate to the
Copyright1997 by the AmericanGeophysicalUnion
382 EMPLACEMENT OF FLOOD BASALT FLOWS

Lava units and

intercalated deposits

20mI
C
• Lava
Surface
breakouts
flow
lobes
Clastogenic
lava
45øN RBL
I Primarytephradeposit
[• Rootless
conetephra
120øW

Section locations

ß Locations of measured sections

I} Near vent and and dike locations RFS

o Sectionsby Martin (1991)

RSC
Lava flows

[•4
•3

RLG
100 km
River

1. Kupaianaha, 1986-1992 b
- 0.5km3
5.6 years
2-5 m3/s
2. Mauna Loa, 1859
,
0.27 km3
10 months

4 m3/s
3. Laki, 1783-4
14.7km3
,
8 months

I 150-4250
m3/s
4. Roza flow, CRB, c.a. 14.5 Ma o
1300km3
5-15 years ?
2600-8100
m3/s?
1O0
 SELF ET AL. 383

picture of global plate tectonic and mantle dynamics
[e.g., Wright et al., 1973, Macdougall, 1988; White and
McKenzie, 1989; Campbell and Griffiths, 1990; Kent et
al., 1992; Hooper and Hawkesworth, 1993; Arndt et al.,
1993; Anderson, 1994; Peng et al., 1994; Turner and
Hawkesworth, 1995; papers in this monograph].
Furthermore, earlier studies aimed specifically at the
physical emplacementof flood basalt lava flows [Shaw
and Swanson, 1970; Swanson et al., 1975; Long and
Wood, 1986; Reidel and Tolan, 1992] did not have the
benefit of recent advances in our understandingof lava
flow emplacementmechanics[e.g., Hon et al., 1994].
The emplacement of continental flood basalt lavas
also deserves attention in light of the apparent
correlation between the ages of flood basalt eruptions
and mass extinctions throughout the Phanerozoic
[Courtillot et al., 1986, 1988; Rampino and Stothers,
1988; Renne et al., 1992, 1995; Stothers,
Courtillot, 1994]. Although the apparent agreementin
ages is highly suggestive of a link, no causal
relationships can be established without first
understanding the eruptions that form flood basalt
provinces and examining their potential for releasing
volatiles into the atmosphere.
Here we describe the flow morphology and internal
structures of Columbia River Basalt lavas, indicate how
these observationsare inconsistentwith the previously
acceptedemplacementmodel and then introduce a new
model and describehow the flow features are explained
better by it. We also describe what is known about
vents for flood basalt lava flows and eruption rates. We
end by discussingsomeof the implicationsof our work
for the environmental impact of flood basalt volcanism.
For the purposes herein, we define a flood basalt
provinceas an area greaterthan 100,000 km2 covered
with at least 1 km thickness of basalt lavas
individual flows of lengths in excess of 100 km and
volumes of more than 100 km 3.
In exploring the physical volcanologyof flood basalts,
we have concentrated on the Columbia River
(CRB) Group as it is the youngestand best-studiedCFB
province [Hooper, 1982]. Maps of the areal extent of
individual flows and basic parameters such as flow
volumes [e.g., Tolan et al., 1989] are not available from
any other flood basalt province, making the CRB the
only province where one can quantitatively discuss the
emplacement of flood basalt lava flows. This fact
highlights the immense value of previous studies by
many workers in understandingthe CRB lava sequence.
The major lava production in the CRB occurred
between 16.5 and 14.5 Ma (the Grand Ronde Basalt and
Wanapum Formations), when -- 90% of the total volume
of the province was erupted [Tolan et al., 1989]. Even
within the CRB, the older units are exposedinfrequently
as they are covered by thick stacks of overlying lava
flows. Our information to date is largely from one of the
best-studiedvoluminous flows, the ---14.8Ma, 1300 km3,
Roza Member of the Wanapum Formation (Figure 1;
1993; Tables 1 and 2) [Swansonet al., 1975, 1979; Tolan et al.,
1989; Martin, 1989, 1991]. Observations made on other
CRB lava flows during this study and reportsfrom older
CFB provinces [e.g., Walker, 1971; Keszthelyi et al.,
1997] suggestthat the Roza is representativeof the type
of flood basalt flow that shows a simple internal
structure(Type I of Long and Wood [1986]).
2. PHYSICAL DESCRIPTION OF CONTINENTAL
FLOOD BASALT LAVA FLOWS
2.1. Terminology
Before describing CFB lava flows, we must define
our terminology. In choosingterminology we have used
two criteria: (1) simplicity and (2) ability to convey the
concepts relevant to the emplacement of CFB lava
and with flows. As such, the terms we use here are not
necessarilythe most useful for describingoutcropsor for
field mapping in ancient lava sequences. However, the
terminology serves well in describing active and young
Basalt flows in Hawaii and Iceland.
We divide the products of an eruption into three

Figure 1. (a) Map of known distributionof Roza member,Colombia River flood basalt province, showing (inset) extent of
Columbia River Basalt Group (dark shading) and Roza lava flow field (light shading). Dashed line delineates Roza eruptive
fissure [Swansonetal., 1975]. Dots mark locations of sections measured within the Roza lava; columns show divisions of lava
into flow lobes and the five lava flows that make up the flow field [after Martin, 1989; Thordarson, 1995]. Martin's Roza
chemicalsubtypesare shownhere as lava flows designatedas follows: SubtypeIA and B combined,IIA, IIB, III, IV are flows 1,
2, 3, 4, 5, respectively. Letter designationsare outcroplocation codesreferred to in text and Table 2; SG is Sentinel Gap
location. Maps of individual Roza lava flows are presentedon Figure 14. (b) Comparisonof the extent of Roza lava field with
thoseof three historicbasalticpahoehoeflow fields shownat samescale (see insets). For each eruption, volume, duration, and
averagetotal volumetric flux of lava is given. (Data: Kilauea [Mattox et al., 1993, and present authors]; Mauna Loa 1859
pahoehoeflow [Rowlandand Walker, 1990]; Laki [Thordarsonand Self, 1993]; Roza [ Tolan et al., 1989; Thordarson,
384 EMPLACEMENT OF FLOOD BASALT FLOWS

TABLE 1. WidespreadColumbiaRiver BasaltGroupLava UnitsEruptedBetween16.5 and 12 Ma with Occurrence

of Physical
Lava Features

Formation Agea Area Volume Pahoehoe Lobese Cone/ Lava rise Tumuli Elevated
Member/Unit/Flow (Ma) (kmb) (kmc) Aad B/U crust
f sutures treemolds
Saddle Mountain Basalt
Pomona 12 20,550 760 P X
Umatilla 15,110 720 --
Wanapum Basalt
Priest Rapids/
Rosalia 14.5 57,300 2,800 pa B X X X X
Roza 40,350 1,300 P B+U X X X X
FrenchmanSprings
Sentinel Gap 38,760 1,190 P B+U X X X
Sand Hollow 15.3 67,110 2,660 P B+U X X X X
Silver Falls 28,840 710 P B X
Gingko 37,170 1,570 P B+U X X
Grande Ronde Basalt2
N23 114,460 27,900
Sentinel Bluffs 15.6 pa X X X X
Slack Canyon P X X X X
Fields Springs
Winter Water P B X X X
Umtanum P

Ortley
ArmstrongCanyon
R2 117,730 53,100
N• 102,340 31,400
R1 16.5 96,650 36,200
aStratigraphy,isotopicage, and volume after Tolanet al. [1989] andReidelet al. [1989].
beachGrande
Rondepaleomagnetic
unitconsists
of manyflows.
CSubdivisions
of GrandeRondeBasaltGroupbasedon paleomagnetic
polarity: N = normal,R = reversed.
dDominant
upperandlowersurface
textural
characteristic
of flows;P=pahoehoe;
A=aa;padenotes
pahoehoe
withrubblyflowtop
material.
epresenceof pahoehoelobesand toes at base (B) or uppersurface(U) of flows.
fCleardivision
of flowintocoreanduppercrustzones
based
onvesicularity
andjointingcharacteristics
(seetext).
X Feature noted in flows; -- flow not examined.

"levels": flow field, lava flow, and flow lobe. In
simplest terms, a flow field is a field of lava flows and
eachlava flow is made up of a numberof lobes.
2.1.1 Flow fields and lava flows. A flow field is the
aggregate product of a single eruption or vent and is
built up of one or more lava flows [e.g., Pinkerton and
Sparks, 1976; Kilburn and Lopes, 1991; Mattox et al.,
1993]. For our purposes,a flow field is the productof a
single eruption within a flood basalt province and is
usually identified on the basis of the chemistryof the
constituent flows.
As per the definition in the Glossary of Geology
[Bates and Jackson, 1987] and usagein Hawaii, we use
the term "lava flow" to describethe productof a single
continuousoutpouringof lava. In principle, each flow
roughlycorresponds to one episodeof an eruption. For
example, most named flows within the current Kilauea
eruption formed after a short pause in the effusion of
lava [Mattoxet al., 1993]. Coolingand collapsesduring
the pauses can render the previous lava pathways
unusable before the eruption resumes, so that the new
lava is forced to flow over different areas. If a new lava
flow coversan olderlava flow while it is still hot, it may
be difficult later to distinguishthe two flows becausethe
lavas can weld together and cool as a single unit
[Walker, 1989]. It is also possiblefor an eruptionto
simultaneouslyform two or more separatelava flows.
Although defining the exact limits of an ancient lava
flow can be difficult to impossiblein the field, some
term is required to describethe units that build up a
large flow field. As such, in our usage,each lava flow
in a flood basaltprovinceis regionalin scaleand formed
by a single continuousoutpouringof lava. While these
definitionsseemsimple,it is not straightforward to
 SELF ET AL. 385

TABLE 2. Total Thickness,Core andUpperCrustThickness,andSurfaceandInternalFeaturesof Lava Flowsof the Roza

Member at Selected Locations on the Columbia River Plateau.
Location Name Location Lava Total Lava core Uppercrust Ratio core/ Surface Internal
Codea Flowb thickness thickness thickness
(m) totalth.c features
d features
e
(m) (m)
Asotin Creek RAC 1 11.5 7.5 4.0 0.65 s hz
Banks Lake RBL 2 15.3 8.4 6.9 0.55 s,l,t,it hz,hv,pv,t,vc
3 7.2 3.8 3.4 0.53
Black Butte RBB 1 17.3 12.3 5.0 0.71
2 31.0 20.0 11.0 0.65
3* 3.0 1.5 1.5 0.50
3* 3.0 1.7 1.3 0.57
DevilsCanyon RDC 4 28.8 21.3 7.5 0.74 s,l,it hz
5 6.5 4.0 2.5 0.62
Dry Falls RDF 2 11.0 6.5 5.5 0.59 t,l,s hz,hs,pv,vc
3 12.9 5.7 7.2 0.44
FrenchmanSprings RFS 4 42.5 29.0 13.5 0.68 1,t,s,it,r is
Horton Grade RHG 2 25.0 14.3 10.7 0.57 t,1
4 32.1 17.1 15.0 0.53
Moses Coulee RMC 3 20.6 12.4 8.2 0.60 s,p hz
Drumheller Channels RPR 2 52.0 37.7 14.3 0.73 t,s,it hs,hz,pv
4 9.0 4.5 4.5 0.50

Lyle RLG 2 10.0 5.5 4.5 0.55 1,t,s,it is,hz,tm

2* 4.0 2.5 1.5 0.63
Summer Falls RSF 1 12.0 6.5 5.5 0.54 s hz,hs,pv,vc
2 12.0 5.0 7.0 0.42
3 11.0 5.5 6.5 0.50
ValentineRidge Rd. RVL 2 3.5 2.2 1.3 0.63 1,t
WanapumVillage RWV 2 20.4 10.0 10.4 0.49 1,t,it hz,vc
Selah Creek, Yakima RSC 2 39.0 29.2 9.8 0.75 1,s,r,t hs,hz,pv,vc
Canyon

a Location code of Martin [1991] usedwhere applicable;Figure 1 showslocationfor most of theseexposures.Table 1 of

Thordarsonet al. [ 1996a] gives map referencesfor theselocations.
b Flowdesignation
1-5equivalent
to Martin's[1989,1991]
subtypes;
thus,IA andB = 1, IIA = 2, IIB=3, III=4, IV=5.
c Ratio of lava core thicknessto total flow thicknessfor eachflow; rangeis betwen0.4 and 0.75.
d Morphologic
features
of flowsurfaces
notedat eachexposure.
e Internal flow features noted at each exposure.
it = inflation tumulus

r = pahoehoeropes
1 = pahoehoelobe (1-100 m in long dimension)
s = pahoehoesheet lobe (> 100 m in long dimension)
p = pillow lava
t = pahoehoetoe
hs = horizontal vesicular sheets in lava core

pv = pipe vesiclesnear base of flow

hz = horizontal vesicle zones in upper lava crust
tm = tree molds
is = inflation sutureor pit
vc = vesicle cylinders in lava core
* Possiblylocal surfacebreakoutlobesfrom flow
386 EMPLACEMENT OF FLOOD BASALT FLOWS
them to the earlier mapping efforts in the various CFB
provinces.
In the CRB, a detailed stratigraphy based on the
superposition, chemistry, and paleomagnetic character
of the lavas has been built up successfullyover the years
[e.g., Mackin, 1961; Waters, 1961; Bingham and Walters,
1965; Schminke, 1967; Wright et al., 1973, 1989;
Swanson et al., 1979; Reidel, 1983; Hooper et al., 1984;
Beeson et al., 1985; Mangan et al., 1985, 1986; Reidel et
al., 1989; Landon and Long, 1989]. Similar work is
ongoing in other CFB provinces. Because the
stratigraphy is (necessarily) based largely on the
chemical composition of the lavas, there has been some
confusion about what constitutesthe product of a single
eruption. Flow contacts can often be seen within
individual chemically defined stratigraphic units.
Although a chemical stratigraphy is absolutely vital in
sorting out the history of a flood basalt province, even
more detail is needed to decipher the physical
emplacementprocessesthat formed each lava flow field.
It appears that each stratigraphic member in the
Wanapum and Saddle Mountain Formations of the CRB
Group (Table 1) is the product of a separate eruption.
Each member has a distinct chemical composition and
is usually separated from other members by a thin
weathering horizon [Tolan et al., 1989]. However,
current knowledge does not permit separation of
individual lava flow fields within the full extent
Grande Ronde Formation, which constitutes -85% of the
volume of the CRB province [Tolan et al., 1989]. The
Grande Ronde is presently divided by different schemes
into (a) paleomagnetically and chemically defined units
[e.g., Swansonet al., 1979; Mangan et al., 1986; Reidel et
al., 1989] and (b) stratigraphically defined units (flow
groups and flows) in the Grande Ronde N2 magneto-
stratigraphic unit [e.g., Landon and Long, 1989]. The
"flow" subdivision of Landon and Long probably
representsflow fields or, in some cases, individual lava
flows, but correlation of their units cannot be carried
widely acrossthe whole outcrop area of the CRB, due in
large part to a lack of exposure.
Within the Wanapum members, chemical subunits
with subtle differences are often recognized (e.g., the
chemical subtypes I-IV in the Roza Member [Martin,
1989, 1991] and six "basalts" of the FrenchmanSprings
Member [Tolan et al., 1989]). These may be the result of
slight compositional variations in the lava produced
along the length of a fissure systemduring the life of an
eruption. As such,they may generally correspondto our
usage of the term "lava flow."
2.1.2. Flow lobe. We use this term to describe
individual package of lava that is surrounded by a
chilled crust. Flow lobes in the CRB and other CFB
provinces can vary in size from tens of centimetersto
many kilometers in scale. Small, 10-50 cm thick, 30-
100 cm long lobes are usually called toes. Lobes only
rarely emanate directly from the vent. They are most
often fed from the interior of other, usually larger, lobes.
Many CRB flows are built up of lobes on the scale of
hundreds to thousands of meters across that are much
wider than they are thick and have relatively flat upper
surfaces. Such lobes have been referred to as sheet
flows [Hon et al., 1994]. Because these sheets are often
regional in scale and are the product of a single
continuousoutpouring of lava, large sheet-like lobes blur
the distinction between lobe and flow. When empha-
sizing the fact that the entire sheet is composed of a
single lobe we use the term "sheet lobe."
A flow composedof a single lobe has been called a
"simple lava flow" whereas a flow composedof two or
more lobes is called "compound" [Walker, 1971]. In
common usage, "compound lava flow" has come to
imply that the flow is built up of many, overlapping,
subequal-sized lobes. We do not make any such
implication when using the term "compound";the lobes
can be one or two ordersof magnitudedifferent in linear
dimension and differ more in terms of volume. The
designations "simple" and "compound" are often
applied on the basis of only a few outcropsof a lava
of the flow; in our experience, what is referred to as a simple
flow in the CRB is usually a large sheet lobe. If
followed for a great enoughdistance(in somecasesfor
tens of kilometers) these sheet lobes terminate against
other lobes of the same lava flow. Thus, a lava flow can
be compound, although in many localities only a single
large lobe is exposed,giving the impressionof a simple
lava flow. It should also be noted that "simple lava
flow" and "sheet lobe" are identical to the Type I lava
flow morphologydefined in the CRB by Long and Wood
[1986]. Type II and III flows, which are also commonin
the CRB, are variants having more complicated,
horizontal vesicular zones and/or jointing in their upper
crusts and interiors.
2.2. External Features of ContinentalFlood Basalt
Lava Flows
Most continental flood basalt provinces appear to be
built up of hundredsof thick (20-100 m) flow fields,
each consisting largely of sheet flows. On average,
these flows traversed very shallow slopes, (e.g.,-0.1%
in the CRB [Tolan et al., 1989]). The flows appearto be
an hundreds of kilometers long and some have volumes
reaching into the thousandsof cubic kilometers.

documented cases of the extent of individual flow fields,
e.g., the Roza Member of the CRB (Figure 1), show the
huge size of these lava bodies when compared with
historic lava flows. It is uncertain, however, whether a
flow field like the Roza covers the entire 40,300 km2
area within the boundary shown on Figure 1 or whether,
as is likely to be the case, there were areas that were
not covered by the lava (kipukas).
Upon closer examination, it can be seen that most
flood basalt lavas are compoundpahoehoeflows. The
pahoehoenature of many major flows in the Columbia
River Basalts can be seen in their smooth upper and
lower surfaces[e.g., Mackin, 1961;Swansonand Wright,
1980; Reidel and Tolan, 1992]. In many cases,ropes,30-
50 cm scale toes, and other features characteristic of
pahoehoe flows are preserved at the flow tops and
bottoms (Figure 2; Tables 1 and 2). Pillow lava
sequencesare often found at the base of CRB flows,
occasionally forming almost the whole flow thickness
[Swanson, 1967; Schminke, 1967; Swanson and Wright,
1980; Long and Wood, 1986]. These pillows are
essentially pahoehoe lobes that formed as the lava
invaded lacustrine and riverine environments across the
Columbia Plateau. Littoral examples of CRB pillow
sequencesare found along the Oregon coast [Snavely et
al., 1973].
Several workers have described aa flows in the CRB,
especially in near-vent areas of the Grande Ronde
Formation [e.g., Swanson and Wright, 1980; Reidel,
1983]. In our investigations,we have found only a few
cases of true aa in the CRB pile, even in the Grande
Ronde source area. To be considered true aa, lava flows
should possess (a) spinose rubble (clinker) at the flow
top and bottom, (b) elongate, ragged-shapedvesicles,
and (c) entrain upper and basal clinker into the flow
interior. Many CRB flows have a thin, rubbly flow top
[Reidel, 1983]. However, they are unlike true aa flows
in that the rubble is largely composed of disrupted
pahoehoecrust and in that the flows have smoothbases.
Such disrupted upper crusts are a common feature of
pahoehoe flows in Hawaii, Iceland, Australia, and
elsewhere, where they occur in patches on generally
undisruptedpahoehoeflow surfaces. These patchesare
often transitional forms of pahoehoe (e.g., slabby
pahoehoe) and are suggestive of either more viscous
lava or locally more rapid emplacement. In ancient
flows, if the only available outcrops of a lava have
surfacerubble the flows may have been describedas aa.
The compound nature of CRB lava flows can be
demonstrated by the fact that most of the chemically
defined lava flow packages consist of multiple,
physicallydistinct,lobes. The dominantlobe type is the
SELF ET AL. 387
kilometer-scale sheet lobe that is significantly larger
than anything seen in Hawaii or Iceland (Figure 3).
Where the contact between two lobes is visible, an
overlying lobe usually fills in the gap between them.
From a distance,this can give the impressionof a single
continuoussheet when in reality there are distinct lateral
discontinuities.
In most outcropsof the Roza Member and other CRB
flow fields, there are two to four distinct sheet lobes
[Martin, 1989, Thordarson, 1995], each probably a part
of a separatelava flow. Margins of sheetscan rarely be
observed because the sheets are usually larger in scale
than the outcrop. Smaller toes and lobes are commonly
associated with these large sheets [Finnemore et al.,
1993; Thordarson, 1995] (Figures 2, 4, and 5). In some
casesthe lobes are clearly associatedwith the overlying
unit, with small toes emplaced in front of a larger sheet
that eventually overran them (Figure 2c, d). In other
casesthey are associatedwith the underlyingunit, with
small lobes having oozed out from the larger sheet
(Figure 4 b).
Other lateral variations are identifiable in the CRB.
Most sheet lobes do not have truly flat tops, but instead
have hummocksand swales of 1-5 m amplitude and 10-
50 m wavelength (Figures 3 and 5). These undulations
have been overlooked because (1) they are relatively
small comparedto the dimensionsof the averageCRB
sheet lobe (typically 1-20% of the total thicknessand
0.1-10% of the lobe width) and (2) the contacts between
lobes are often very poorly exposed. This is becausethe
upper and lowermostparts of the lobes are vesicularand
erode more easily than the dense interiors. The 1-5 m
surfacetopographyis very often obscuredby talus and
vegetation.
When exposures are adequate, axial cracks can be
found at the crest of many hummocks which are
identical in size and shape to tumuli and other inflation
features found on pahoehoe flows across the globe.
Tumuli are broad and whaleback-shaped rises, and
usually have axial and medial cracks that form as the
brittle crust is uplifted [Walker, 1991; Hon et al., 1994].
On Hawaiian flows, tumuli are usually on the order of 5-
20 m in length, 5-10 m in width, and 2-5 m in height.
However, larger tumuli on the order of 50-1000 m in
length and width and 10-20 m in height are commonon
flows in the continental United States [e.g., Nichols,
1936; Theilig, 1986; Keszthelyi and Pieri, 1993;
Chitwood, 1994]. The Undara flow in Queensland,
Australia, has a 40-km-long tumulus (inflation-ridge)
[Atkinson, 1996]. Tumuli on all these scalesare found in
the tops of CRB sheet flows (Figure 5) and are often
associated with small breakouts. As noted above,
388 EMPLACEMENT OF FLOOD BASALT FLOWS

iB
 SELF ET AL. 389

flows have much broader surface undulations (Figure 3)
that appear analogous to inflation ridges and plateaus.
Other features in CRB flows (Figure 6) are analogousto
suture zones recognized on other pahoehoe flow fields
[Walker, 1991].
The abundanceof compoundpahoehoelava flows in
the CRB and other flood basalt provinces is significant
because pahoehoe and pillow lavas have only been
observed to form at low volumetric fluxes [Rowland and
Walker, 1990; Griffiths and Fink, 1992; Gregg and Fink,
1995]. The compound nature of the lava flow fields,
with small lobes and toes at the bases of the flows, also
indicates that they were eraplacedin a series of lobes
separatedin time and space, arguing for a drawn-out
eraplacement history. Compound pahoehoe lavas are
also common in the Deccan Traps [e.g., Agashe and
Gupte, 1971; Phadke and Sukhtankar, 1971; Walker,
1971; Keszthelyi et al., 1997] and other CFB provinces.
2.3. Internal Structure of ContinentalFlood Basalt
Lava Flows
The internal structureswithin sheet lobes also provide
clues to their style of emplacement. There are three key
types of internal structures: (a) vesicle patterns, (b)
jointing style, and (c) petrographic texture. Of these,
jointing is the least informative becausefractures form
only after the lava has solidified. These internal
structuresdivide each sheet lobe into (1) an upper crust,
(2) lava core, and (3) basal zone (Figures 7 and 8)
c
[Thordarson, 1995; Self et al., 1996].
This three-part division does not change from the
near-ventoutcropsto exposureshundredsof kilometers
from the source. The divisions can be recognized at
every one of the many outcropsof the Roza and other
CRB flows that we have examined. The same divisions
are also seen in other pahoehoelava flows of various
thicknesses in Hawaii (Figure 7d), Iceland (Figures
7b,c), the Deccan Traps [Keszthelyi et at., 1997], and
elsewhere [e.g., Aubele et at., 1988]. Contacts between
the three zones are not knife-sharp, but can usually be
located to within 10 cm, even in flows many tens of
meters thick.
2.3.1. Upper crust. The upper crust is defined by
relatively high vesicularity and small prismatic joints at
the very top with irregular jointing beneath and
hypohyaline (50-90% glass) to hypocrystalline (10-50%
glass) textures. The jointing in the upper crust (Figures
7 and 8) can be highly variable, including types
previously described as hackly, curvilinear, wine-glass,
etc. We generally avoid use of the terms entablature
and colonnade because in flood basalts they have
become associated with specific genetic models (e.g.,
water cooling for the entablature [Long and Wood,
1986]).
Vesicularity usually decreases while vesicle size
increases downward into the flow. Approximately
horizontal layers of increased vesicularity can often be
found in the upper crust (Figure 9a). The crystallinity
gradesfrom a cm-thick glassy rind at top to dominantly
microcrystalline downward over several meters (Figure
10a). In most cases in the CRB and elsewhere, the
upper crust constitutes 40-50% of the total flow thick-
ness (Table 2).
2.3.2. Lava core. The core of a sheet lobe is
characterized by very few primary vesicles, regular
jointing, and holocrystalline texture (90-100% crystals).
The crystals are fine to medium grained and most of the
porosity in the core comes from diktytaxitic voids
between the crystals (Figure 10b). Jointing tends to be
quite regular, and well-developed columnar jointing
(colonnade) is sometimes present. The core typically
makes up 40-60% of the flow thickness(Table 2).
Megascopic vesicles are usually confined to the late
stage residuum formed during in-situ crystallization.
Macroscopic bodies of this vesicular late-stageresiduum
are confined to the lava core and have been reported
from many thick basalt flows of all ages [e.g.,
Greenhoughand Dostal, 1992; Puffer and Horter, 1993].
As the lava crystallizes, volatiles are concentratedinto

Figure 2. Pahoehoelava lobes and associatedfeatures in CRB flow in Washingtonstate. (a) 60-m-high section through
FrenchmanSpringsmemberlava (aboveupperdashedline) and a flow lobe at the top of GrandeRondeBasaltN 2 SentinelBluffs
unit, probablyequivalentto Museumflow groupof Landonand Long [1989], betweenand below dashedlines. Near Lower
Monument Dam (area of RDC, Figure 1). (b) Boxed area to right of personin (a) exposesa section through a pahoehoe toe at
the base of the Museum flow lobe (center;scale is 10 cm across)with centimeter-thick chilled selvage. Note jointed vesicular
crustof underlying flow (below level of scale). (c) Basal 70 cm of lower Roza sheet lobe at Dry Falls, showing glassy selvage,
s, and thin vesicularcrust, c, pipe vesicles,p, and vesicle cylinders,vc. Photographlooks down onto fallen block lying on the
ground. Arrow showsoriginal up direction. (d) 40 cm long pahoehoetoe with ropy surface at base of Roza flow in Yakima
Canyon (RSC on Figure 1). (e) Upper 5 m of a flow consistingof shelly pahoehoelobes, seenin section,overlain by massive
base of another pahoehoeflow (above dashedline); both FrenchmanSpringsunits at Wallula Gap. Hammer (circled) is 35 cm
390 EMPLACEMENT OF FLOOD BASALT FLOWS

;.'3 "--"-'----'-•'•
-•"'•-
- - - - talus

ß *- Banks Lake
d/...""""
'""..'"/.
,'"",•m,,-
..i• O, ½.,,'h,,,.,,,•,.o,.
"% '"'••/,,,,,,,.,,•.o'• mm,•,
• •

Figure 3. Sheetflow lobesin the CRB. (a) 220-m-high section displaysthe geometry of large (>1 km scale) sheet lobes in
severallavasof GrandeRondeN 2 (probablySentinelBluffs) unit, west side of Banks Lake (Grande Coulee), Washington(near
RBL on Figure 1). At least four lava flows (1-4) can be seenbut it is uncertain how many flow fields are represented. Lobe 3c
fills the hollow betweentwo 50 m thick, previouslyemplacedsheetlobes, 3a and b; lobes of flow 3 cooled together as a single
cooling unit becausecoolingjoints passfrom one lobe into the adjacentlobes. Lobe 3a is constructedof two tumulus-like bodies
and thins in the middle where lessthickeningby inflation occurred. Sectionshownis about 3 km long and site is >200 km from
the suspectedvent area for theseflows. (b) Inflated sheet lobes in Roza and FrenchmanSprings (FS) membersat Frenchman
Spring Coulee (RFS on Figure 1). Note that Roza thickens into the hollow between two FS lobes that come together in the
center of the photo. On the right, a third Roza lobe (R3) fills in the gap between two earlier sheet lobes (R1, R2) at the
waterfall. Upper part of Roza has been scouredby catastrophicglacial floods that excavatedthe coulee; distinction between
crest and core is expressedby differentjointing (see text). Cliff is 120 m high and each sheetlobe varies from 20-50 m

the remaining melt, producing a vesicular siliceous
sludge that often concentratesinto segregationveins.
This segregatedvesicular material is less densethan the
still fluid surrounding lava and rises in cylindrical
conduits toward the upper crust, preserved as vesicle
cylinders [Goff, 1996] (Figure 9b, c). When these
conduits reach the solidifying roof of the flow, they are
deflected into horizontal sheets. This lateral spreading
of the highly vesicular residuum is a result of a
mechanical barrier and cannot be attributed
residuum reaching a neutral buoyancy level. In some
casesthe buoyant residuum attempts to form diapirs up
through the viscoelasticcrust. This can result in large,
bell-jar shaped, 5-30 cm diameter gas cavities that we
call "megavesicles." However, it must be noted that
large coalescedmegavesiclescan also occur within the
upper crust.
2.3.3. Basal zone. The basal zone is almost always
only 0.5-1 m thick, is hypohyaline (50-90% glass)
(Figure 10c), slightly vesicular, and sometimes has
poorly developed platy jointing (e.g., Figure 2c). In
thick flows, the basal zone usually forms much less than
10% of the flow thickness. A 1-3 cm thick, quenched
glassyselvagewith stretchedor round vesiclescan often
be found at the very base of the flow. Pipe vesicles
occur in the base of many CRB lava flows (Figure 2c),
consistent with emplacement on very low slopes
[Walker, 1987].
2.4. Ventsfor Continental Flood Basalt Lavas
Little has been written about the vents from
flood lavas are derived. In part this is simply because,
other than feeder dikes, vent structures or edifices
appear to be elusive features in CFB provinces.
However, the Roza linear vent system of the CRB,
identified as coeval with the Roza flow by field and
chemical characteristics [Swanson et al., 1975; Martin,
1989] is well documented. Identification of this vent
system strongly suggests that flood basalt lavas are
largely fissure-fed. Featuresrangingfrom dikes to small
shield-like edifices are exposed within a narrow zone
about 5 km wide along a 150-km-long swath of country
(Figure 11). An outlier to this zone, the vent-like
structuresat Pomeroy quarry, Washington (Figure 3 of
Swanson et al. [1975]), are possibly sections through
rootless cones (pseudocraters) in a lower Roza sheet
lobe that was buried beneath later Roza lobes (Th.
Thordarson,unpublisheddata).
Deposits of welded spatter and fountain-fed
(clastogenic) lava, identified by ghosts of agglutinated
clasts, along the trend of the Roza fissure provide most
SELF ET AL. 391
of the evidence for location of the vent system [Swanson
et al., 1975]. The central region has several small (50 m
long x 5 m high) outcrops of clastogeniclava that give
the impression of widespread fallout of material on
either side of a fissure and are convincing evidence of
proximity to the lava source. At some locations, e.g.,
Potter White Hill (site 15 of Swanson et al. [1975]), the
clasts are coarse (30-40 cm), suggestingthat a vent was
nearby. The best exposedaccumulationsof Roza spatter
to the and scoria are in two exposures forming an oblique,
1-km-long section across the strike of the fissure at
Winona, Washington [Thordarson and Self, 1996]. The
structures appear to be parts of 5-6 m high scoria
ramparts with associated welded spatter, and although
no convincing evidence exists of a section across an
actual vent, the structures are buried beneath >10 m of
fountain-fed lava, suggestiveof a nearby source.
The detailed geochemical study of Martin [1989]
showed that each chemical subtype of the Roza lava is
found in a restricted portion of the fissure system,
suggestingthat lava effusion migrated over time (Figure
11). We speculate, based on the behavior of historic
fissure eruptions, that only one or two segmentsof the
Roza fissure system, each several kilometers in length,
would have been active at any one time. Furthermore,
as in Hawaii and Iceland, each active segment should
have contracted to a few point-sources of lava in a
matter of days to weeks given the thermal instability
inherent in fissure systems [Wilson and Head, 1981;
Whitehead and Helfrich, 1991].
Reidel and Tolan [1992] describedanotherexample of
which a vent structurefrom the R1 magneto-stratigraphicunit of
the Grande Ronde Formation. A dike-fed, approximately
100-m-wide lava lake of unknown long dimension is
exposed in cross section, with associated shelly
pahoehoe overspills and scoria and Pele's tear deposits
extending to a few hundred meters on either side of the
lake. This structure is of average size by Hawaiian
standards;e.g., the Kupaianaha lava pond, active from
1986-1991, was <100 m in diameter. Yet this Grande
Ronde vent was part of a system that produced a lava
flow thought to extend for >300 km and have a volume
of-2,000 km3.
Having presented a compilation of the field data
available on CRB lavas, we move on to the (differing)
interpretations.
3. EARLY MODEL OF FLOOD BASALT
EMPLACEMENT
The pioneering, but highly exploratory,work by Shaw
and Swanson [1970] on the emplacement of the
392 EMPLACEMENT OF FLOOD BASALT FLOWS

, •f[ '•,:. •-'-.. • ,.. . . •/ /"-• •',•o : vesicular

flow
top
_,.-._1

b• •injec
ion
from
lobe
3 t• '• contact
w½½•

• ••. tree
molds':• •, Lobe2b
• core.• •
." • • contact
welded ••' ''
,,,,,,,,,,,,,,•
-"'"W,••
Lobe
1 ••
ush-',,.
..............................
....
Figure4. Flowlobes in CRBlavasdemonstrating locallycompoundnatureof flows.(a) Sketchof smallcompoundpahoehoe
lobes(1-3)in a Wanapum Basaltlavaflow0.5 kmnorthof SoapLake,Washington (RSLonFigure1). Notethatinflationof
lobe3 disruptedpartof thecrust
ofpreviouslyemplacedlobe2a. Individual
columnarjointspassing
through
theupperpackageof
lobesandweldingof lobecontacts indicate
thattheycooledtogetheras singleunit. (b) A 60-m-long
lobefed by breakout
throughcrustof underlyingsheet
lobein a GRBN2lavaflowoneastsideof BanksLake,Washington (nearRBLonFigure1).
Circularholes,h, aremoldsof treetrunksthatwerelyingonthecrustof lobeasit inflated.Notethattwo low-amplitudetumuli-
like bodiesformedin lobe aboveto eithersideof the thickestpart of break-outlobe. Pahoehoerubbleis disruptedglassy,
vesicular
flow-top
material.(c)Rozamember
at DryFalls(RDFonFigure1) showing
twoof thethreesheetflowlobesfoundin
thisarea. Note that the lower lava flow lobe graduallythinsaway from cameraand that the thinningis compensated
by
thickeningof upperlobe. Totalthicknessof Rozaat thisexposure
is -023
 SELF ET AL. 393

'- ' '":•"•.'•-'

-. ::'"'•:
'"'•. •...... •.•: $-•-::•'•'•.:•'E•'b.-•.-•:•:.'•:::-:.:'
:.•ri.:•....
•½.::
....... ;':• ........... - -

..•.-- .•..'.•..•'
..x•q
::•:.•,::-%•
::•.-•?•::2:*-:.•::•
:•:-e-•.•.,•:.......•
. • •:•, :,./--.:xr:
.:..•:.::•.•
..•
......•.• :• ...... • -• ß • '..................
-::•u . . e.•':..•-'/•
............ -a•.•:...•
..... '................ ..... : :: •-•:•
•.........,,•:.-•.
..... . '.: -
............................
.•:::'
.....
'-•,•,:-:•,•-•:•;.•.,.•.:•:g•:•:.•:•?•
'•..
•,..•;:•::
'":•-•::-
--:-.:;.•:::.....
.....
•:'"..:':•:-'•:.•-::-•:'•z•'•:--•-::--'.•;,.::•
.. ,---•:•'•':•.•
....•*•'-':•.-•-•...,
....
:.. :'..--.'
•.-:•?--.--.•::.•:
'" .::%.." :':'--.. :- • •..'•:-..,•:•;4•::"•' ..-•e•*'-•:: ....,' -.:-"....'
•-:...:•:•.•...•.•.•:.t.d•;
..........
.......
'-'.•,::-•:-•
•..'•:'.'• .:-:•.
'.'.....
:"::;'•....';
.- ":•:.-':..:--:
':: .....:.
2:'
......:.'•
...... ,........ '.Z.'-'ff•"•;:•'• -'-•...-- :'-•'""• --'---..
•,-'-:
.....
' t" ;-•'-':?•'•:
'•:-.•
.... • -• :::'•'::--'•'
'•' ' ' ': .............
"' ...... '"":*'
................
:•.'.".'..
' , "?.'.'•........ • .. e ,:,.:,.
........
:-----..---::•::•....'::•-:.'•"'
::-.%::;..
•:•d'•:...
-"--'
.....•.-..•'"'
:'.•----'--•
:•"::--.::-'•:'•:••••
.'•;f-':e;?'• .•:•,:'"•-•-•:.•e•:•-•:,
•.`•...•.f•:•Z•.;•x•:.•u:.:•:•..•x:•..•..:•:;..•:*:::.:%.?•:•x•:t;•..
............
,.-.-:,•.x'•:•'•
.........
, .....
---,•:•/•.,:-
•..-
.....•..::•.•-.-.-,•-.::•:.
............
::•:::•:
.... -:-...:
....-.......
-:•
•:-•::::::•:.•::.-
......
-..':'•...............
-............
-.•:::.:•:•:.•...
- '?'-
*'P--'---•':'-
......"--:-'
:-:•-•-•--:-
-:'- ' -•-'-•'' --:--:
---......................... * ..... :-::-•
.... :-.::'-:-::', '----d• *.... , '-. -'{•: :•'•- :--? ----•"
• - -:-• .'•''--:--:•-•::,.•:

:;•..e..L:•x:.-•?-:.•:..:•..::•.:•:•t•:,,•:•-•,:-•:"
-•::•-,:_c•::..:•:::•:•:•'
:..•-•--::•.---:-
• -:--•-•:::--:•.•.-:•-'-'-'•:-..•:•-.

•g•'•"::•-:•
•:.:-:d•:?'...;.•;•:•:•:%•4'"'
..•:::•::•
'-•-::½:::.:.:•:%:•h-
...........
.::"'::•:t7-"-"•:•:•½%•:•::'--':-•:•'•::•½
•::.:-:•;--•
......•'•'•' '••'-:•-•:....t..: •.:i-•?•;F•:•:•:•;
....;-':•::
';"•--':.,•
•{•::•.•":•-
..... •t:::•'--.
-•*... ":':•-:•::;•.:•
:?•::
:.'
.........
--:•-,
::-•.
.........................
.•:•:?.
::.-..
............
.........::•
.................
•--:•:• .......• -:•.%.•::--:;:•:.•
::...:.% -" ...•:•;.:;;:.•::•:....;•.
;•.::•-•:•:•:•.
?•:..:..•....?--
•-?•.--•:•••••••
:.••.••,..•-•:.:...:.•::•::•::•:.--
:-x:-•½ •':,.--
------.:.
'•::•...,.7
--..ex.-.•:•..
:•-•:---:
. •...::......
.-..•--:-'•/•.•:•
..............
.:.:::..:...
...............
-.

.....•...• ......'-.•.• ---,-•------••••••••

............ :•,:•:,•,..::•,•:•;g -:-•-• .•.• •-•'•:.•:•.•,:f-•..:::::.:•-:-.:.:•:"'•*
................

tumuli .' -

'• bsse
o•
ano•ner
C.RB
unit

..... l--- "• i / .I o-•,•1• • "/•_,.{,t'•

pahoehoe
rubble

Areaof½i
t I 11/ J' '½;
photograph_.
I ' l,! ;.... •, /:....
;0
..................
[,'.x
..•_
.........
L -J ; 5m ' road level

Roza 2

.:

Roza
Roza ß
ß
394 EMPLACEMENT OF FLOOD BASALT FLOWS

,,•i!!!
"• •'• -•7• • ....
•...'•!:•....i•-'.',.
ß
..... •.:•'?••'•'• .....
:..
....

.•-. • • •. .:,.•

..... . ,., •.... '•...ß - - .

.....

..... . .:......:•.•.:.:...•.:.,?•.
-.• .•.•• :::....
:: . :•:......

5`5`•`5`.5`;`5`5`5`5`5`5`5`•`•`•`5`5`5`•`;`5`5`•`;`5`5`5`5`5`5`5`5`5`5`•`5`5`5`5`5`5`;

'•...........
• •.....' ................ '........... !'
, ' . ,' ....---•
• '"
,

slabs
of
pahoehoe
crust •v
i
•
between
flows
' ,
I tumuli
attop
of -l 0

Figure 5. Tumuli in CRB and Kilauea pahoehoeflows.

(a) Section through a small tumulus (T) with associated
breakouttoesand lobes(L) in top of a sheetlobe, overlainby
massivebaseof anothersheetlobe (aboveupperdashedline)
thatfills in topography;exposedat FrenchmanSpringscoulee
(RFS on Figure 1). (b) Tumuluson active Kalapanaflow,
Kilauea,in 1990. Note breakouttoesand upturnedpahoehoe
crustalplates. In crosssectionthis would resemblethe tumulus
shownin (a). (c) Larger elongatetumulusof 5 m amplitude
and40 rn length on active Wahaula flow, Kilauea, 1991. (d)
Large tumuli of 10 rn amplitudeand 50 m length in upper
surfaceof a lava flow in the compoundRoza members,east
wall of Devil's Canyon(RDL on Figure 1). Anotherlava flow
is presentabovetumuli, with Rosaliaflow of Priest Rapids
memberoverlying; a FrenchmanSpringsunit underliesRoza
flow field. Sketchshowsinterpretation
of photo.

lavas has become the entrenched and (until recently) field data are inconsistent
with somepredictionsbased
unchallengeddogma that it was never intendedto be. In on this emplacementmodel.
this section,we describethe constraintsand assumptions Shaw and Swanson [1970] made one of the earliest
on which Shaw and Swanson [1970] based their work, the attempts to combine fluid mechanics with field
resulting emplacement model, and how new observationssuchas lava flow thicknesses,
topographic
observations over the past 25 years have forced re- slope, and an estimateof lava rheology in order to
evaluationof somekey assumptions.We also show that calculate velocity and style of flow. The

observationthat constrainedtheir modeling was that the
CRB lavas do not show evidence of measurable
crystallization during transport from their source vents.
Glassyrinds of flows 300-500 km from the ventsare just
as glassy as those within a few tens of kilometers of the
vents. This suggested to Shaw and Swanson that the
flows must have traveledvery rapidly, attainingtheir full
areal extent before any significantcooling took place.
Shaw and Swanson [1970] also assumed that the
glassy nature of the lava selvages indicated that the
lavas were erupted well above the liquidus temperature,
allowing the lava to cool substantiallyduring transport
without crystallization. The high assumed lava
temperatures (-1200øC) led to the use of very low
viscosities (50 Pa s) in their calculations, which further
corroborated their assumption of rapid emplacement.
Combined with the great observedflow thicknesses,this
resulted in calculated flow velocities of several km/hr,
flow in the turbulent regime, and eruption durationson
the order of weeks. Shaw and Swanson [1970] did
suggestlonger eruptive durationsfrom extensivefissures
as an alternative scenario, but this idea was not
generally adopted [e.g., Hooper, 1982].
Although a picture of cataclysmic floods of lava
charging across the Columbia Plateau is perhaps
appealing, good evidence exists that the CRB flood
basalt lava flows were not emplaced in this manner.
First, the phenocrystsseen in many CRB lava flows,
includingthe Roza flow [e.g., Swansonand Wright, 1980;
Marti•, 1989], are now known to be inconsistentwith
eruption temperatures well above the liquidus. The
crystallizationof metal oxides in the glassymarginsof
dikes and lobes suggestseruptiontemperaturesof 1090-
1070øC (Table 3) [Thordarson, 1995; Ho and Cashman,
1995]. The viscosity of the lava can be estimatedfrom
the liquid compositionand temperature[e.g., Bottinga
and Weill, 1972; Shaw, 1972] and then adjustedfor the
effect of entrainedcrystalsand bubbles[e.g., Pinkerton
and Stevenson, 1992]. For the range of glass (and
presumably liquid) compositionsfrom the CRB lavas
(49-56 wt.% SiO2 [Mangan et al., 1986]), estimated
viscositiesat the point of eruptionare one to two orders
of magnitudehigher than the 50 Pa s usedby Shaw and
Swanson [1970]. For the Roza lava, estimated viscosi-
ties are 500-700 Pa s (Table 3). Thesehigherviscosities
make it unlikely that even 20-m-thick CRB lava flows
would have been turbulent (Figure 12). However, lava
flows >30 m thick theoretically should have been
turbulent.
Turbulentlava flows shoulderodetheir basesby both
SELF ET AL. 395
thermal and mechanical erosion, resulting in distinct
channelswith bell-jar-shapedcrosssections[e.g., Jarvis,
1995]. CRB flows have no evidence for such
downcutting or channelization,being remarkably sheet-
like (see Figure 3). Delicate, centimeter-scale surface
textures are preserved between thick flows in the CRB,
indicating that no erosion took place at the contact (see
Figures 2 and 4). No field evidence collected to date
supports turbulent emplacement for any of the flows
examined.
Emplacement of CRB lava flows cannot be generally
explained by rapid laminar flow, either. Rapid flow
leads to high strain rates which, in the absence of
unusually low viscosities, lead to the production of aa
(or aa-like) flow surfaces [Peterson and Tilling, 1980].
The calculated viscosities of CRB lavas (Table 3) are in
fact even higher than those of typical Hawaiian lavas.
Furthermore, aa flows, with disrupted crusts, are
inherently thermally inefficient. Adopting the thermal
model of Crisp and Baloga [1994] for aa lava flows, it
can be shown that even if CRB flows were emplacedin
less than half a day, they would probably have cooled
and crystallized significantly. Only under very special
and unlikely conditions can aa lava flows 300-500 km
long be producedby laminar flow [Keszthelyiand Self,
1996].
The single greatest problem with the Shaw and
Swanson [1970] rapid emplacement model for flood
basalt lava flows was the assumption that the final
thicknessof the frozen lava flow was approximatelythe
thickness of the flow while it was moving. We now
present a new model based on the idea that most CRB
lava flows thickenedby one to two ordersof magnitude
during emplacementby the processof inflation.
4. A NEW MODEL FOR THE EMPLACEMENT OF
FLOOD BASALT LAVA FLOWS
The dominant flow morphology in the CRB and other
flood basalt provincesis the thick, compound,pahoehoe
sheetflow. Sheet lobes within the ongoingeruptionof
Kilauea Volcano, Hawaii, form exclusivelyby inflation
(endogenousgrowth), i.e., the injection of liquid lava
under a solidifying crust [Hon et al., 1994].
Comprehension of the inflation process and the
realization that inflation is the fundamental and
universal process through which pahoehoe lava flows
grow in thicknesshave led to a revolutionin our thinking
aboutthe formationof compoundpahoehoeflow fields in
generaland flood basalt sheetlobes in
396 EMPLACEMENT OF FLOOD BASALT
 SELF ET AL. 397

a
Roza member of the CRB
Flow 2 Banks Lake,
Washington

Key
JH
Partings and gas blisters

Megaves•cles(MV)
b Vesicle sheets (VS)
8500 BP Thj6rsg lava Vesicle cylinders(VC)
Flow unit at Thjbrs•rbrQ
South Iceland Irregular, closely-spaced joints
Master joints = columnar joints
Crust
,
Platy joints
lO

c
AD 1783-84 Laki lava d
Small flow unit at Kilauea lava
Crust Eldvatnsbr•)
Lobe emplaced in 1991
South Iceland

VF JH
100 -
Crust

Core
Crust

Core Core
I Core

ß- '• / Zone
Om ;i'"'::'"• 0m Basal Basal o cm Basal
Zone Zone Zone

Figure 7. Examples of measuredsectionsthrough sheet flows of various thicknessesin flow fields of different dimensions. (a)
16-m-thick sheet lobe within Roza member at Banks Lake, Washington(RBL on Figure 1). (b) 10-m-thick flow unit within the
21 km3 Thjorsalavain southIceland.(c) 3.5-m-thick
flowunitwithinthe 15km3 Lakilavaflow, southIceland. (d) 80-cm-thick
flow unit within the 1991-2 Wahaula flow, Kilauea. Note the overall structural symmetry of these flow units. VF indicates
vesiculationfeatures;JH indicatesjointing habits. Detailsof internalfeaturesare as in Figure 8b. Note presenceof pipe vesicles
at the base of each flow. Modified from Thordarson[1995].

4.1. The Inflation Processand Inflation Features contraction during cooling and has essentially no tensile
strength [Hon et al., 1994]. Thus the cooled, fractured,
The inflation processis involved in the formation of solid lava does not play an important mechanicalrole in
all pahoehoelava bodies from 20-cm-thick toes to 100- the emplacement of pahoehoe flows. Instead, it is the
m-thick sheets of lava. Pahoehoe toes and lobes resistance of the viscoelastic skin to continued

initially grow as inflating, liquid-filled balloons. The
liquid lava is held in by a partially cooled, viscoelastic
skin of lava. Over a period of a few minutes a cold
brittle crust will begin to form on top of the viscoelastic
skin [Hon et al., 1994; Keszthelyi and Denlinger, 1996].
The brittle crust is extensivelyfracthreddue to
stretchingthat constrainsinflation rates. The motion of
the viscoelastic skin is accommodatedin the overlying
brittle crust by the widening of the cooling-induced
fractures [e.g., Walker, 1991]. If the rate of inflation is
too high, the skin may burst and fluid lava is able to
break out from inside the inflating lobe.

Figure 6. Featuresin CRB lavas thoughtto be related to thickeningby inflation (lava rise). (a) Margins of small inflation
sutureor pit in a Roza sheetlobe at JasperCanyon, Blue Lake, Washington. Note curved foliation of stretchedvesicles;jagged
plates of lava fill the suture. (b) Margins of inflation suture in a FrenchmanSprings lava flow at Frenchman Springs Coulee.
Note foliation of stretchedvesiclesand small toesfilling upperpart of suture. (c) Large lava rise sutureor pit in Roza flow (note
personin center of feature) showing subhorizontalinterleavedplates of lava between two abutting parts of an inflated flow lobe.
Subhorizontalcracks formed by extensionduring inflation extend into massive lava cone on right and left. Both (b) and (c)
photographedat Frenchman Spring Coulee (RFS on Figure
398 EMPLACEMENT OF FLOOD BASALT FLOWS

Inflation in pahoehoe lavas forms a distinctive set of

external morphologic features, the simplest of which are
lobes, toes, and (in the presence of surface water)
pillows. When inflation is localized, tumuli and other
inflation features can form at the surface. Inflation also
forms pits (the "lava-rise pits" of Walker [1991]) where
a small section of the flow does not inflate and the
surroundinglava is raised up around it. The pits often
contain pahoehoe toes and small lobes where lava has
oozed from the marginal cracks. Such pits are typically
only 2-10 m acrossand 2-4 m deepin Hawaii, and the
CRB flows have similar size features, but larger
inflation pits can be found elsewhere, such as in the
Toomba and Kinrara flows, Queensland [Stephenson et
Structural Jointing Vesiculation Texture
component habits features al., 1996]. In the CRB, the enigmatic features in Figure
meters
ERODED
SURFACE b 6, previously thought to be spiracles caused by steam-
36 induced fracturing [Waters, 1961], can be explainedbest
LAVA CRUST
hypocrystailine as filled suturesor pits where two opposinglobesor parts
platy joints
of lobes inflated and thickened together (the lava rise
hoiocrystailine
LAVA CORE
dicktytaxitic sutures of Walker [ 1991 ]).
columnarjoints VC's
hoiocrystalline
dense The three-part internal division of the thick pahoehoe
BASALZONE hypocrystalline sheet lobes in the CRB is also consistentwith an origin
hypohyalinc
TVZ by inflation. Note that the internal structureof a true aa
irregularand hackly
LAVA CRUST joints
VZ
hypocrystalline flow (described in section 2.2) is very different from
platyjoints
VZ and MV's what we have commonly observed in the Roza and other
HVS's

LAVA CORE
CRB lava flows. Figure 13 shows how we believe the
holocrystailine
columnarjoints 16
VC's dense three-part division comes about. In an inflating lava
BASAL ZQNE
BVZ & PV hypocrystalline
flow, the upper and lower crustsform by freezing of the
TVZ hypohyaline lava while the flows are receiving an influx of fresh lava.
irregularandhackl
LAVA CRUST joints
The core of the flow cools and crystallizesonly after the
hypocrystailine
sheet has stagnated. The clearest evidence of this
VZ and MV's
process comes from the vesicle patterns, as previously
HVS's
LAVA CORE
suggestedfor the origin of vesicle zonation in thin lava
columnarjoints VC's hoiocrystalline
dense flows [Aubele et al., 1988].
BASAL ZONE
BVZ & PV
hypocrystailine
hvpohvaline The size distribution and the shape of the vesicles in
sediment

Sentinel Gap lava
Figure 8. (a) Outcropof Roza memberat SummerFalls (RSF
in Figure 1) showinglava flows 1, 2, and 3 (top eroded). Solid
line indicates boundary between individual flows and broken
line shows boundary between lava core (Co) and lava crust
(Cr), most easily discernible in photographby changesin
jointing. Flow below Roza is Sentinel Gap flow of the
Frenchman Springs Member. (b) Stratigraphicsection mea-
sured through the compound Roza member at Summer Falls
showingdivision of each of three lava flows (1, 2, 3) into crust,
core, and basal zone. Right side of the stratigraphiccolumn
showsvesiculationfeaturesand texturalpropertiesin eachflow
unit. Left side showsjointing patternin the lava. VZ denotesa
vesicularzone, where the prefix B = basal and T = top. MV
indicatesmegavesicles; PV and VC stand for pipe vesicles
and vesicle cylinders, respectively. HVS denoteshorizontal
vesicle sheet. Vertical scale is in meters. Modified
Thordarson [1995]. '
the upper crust suggest that they are primary, having
formed during the eruptionprocess[Mangan et al., 1993]
with modification due to bubble rise and coalescence.
Horizontal vesicular zones preserved in the crust can be
interpretedto form when the sheetis depressurizedby a
suddenmajor breakout and bubbles form inside the sheet
[cf., Hon et al., 1994]. An alternative possibility is that
the vesicular zones are the result of a more bubble-rich
batch of lava passing through the sheet lobe. In either
case, the bubbles migrate upward (and coalesce)only to
be trapped againstthe downward-growingupper crust of
the lava flow. Previous modeling of these horizontal
vesicular zones required episodic bubble formation at
the base of a stagnant lava flow [e.g., McMillan et al.,
1989], a situation difficult to explain physically, or
unrealisticallyhigh lava viscosities[Manga, 1996]. The
from
jointing in the upper crust may be more irregular, in
 SELF ET AL. 399

Figure9. Vesiculation
features
in CRBlavas.(a)Horizontal
vesicular
zones
(numbered
1-3)in thecrust
of a sheet
lobein a
Grande Ronde Basalt N 2 flow at Armour Draw, Moses Coulee, Washington. Tape is extended to 1 m. (b) Individual vesicle
cylinder in core of a Frenchman Springs sheet flow near Lyle, Washington (RLG on Figure 1). (c) Vesicle cylinders (above
hand) passinginto vesicular sheets(outlinesdashedfor clarity), upper part of lava core of Levering flow, Sentinel Bluffs unit of
GrandeRondeN 2 at Sentinel Gap, Washington(locationSG, Figure 1).

due to jostling of the brittle chilled lava that takes place
during inflation.
The features in the lava core are best explained by
slow, stagnant cooling. The bulk of the core is dense
becausethe bubbles have had time to migrate to the top
of the flow. Also, coherent cylinders and sheets of
vesicular residuum form only after the flow has
stagnatedbecause the flow of lava through the core
would disruptthe passiveconvectivepatternsthat lead
to the cylindrical diapirs. Furthermore, columnar
jointing is thought to require cooling under stagnant
conditions[Swansonand Wright, 1980].
The thin basal zone may appear to be inconsistent
with a protracted emplacement, but recent measure-
ments of cooling at the base of pahoehoeflows show
that the basecoolsmuch slowerthan was expectedfrom
any cooling model [Keszthelyi, 1995a].
4.2. EruptionDuration and EruptionRatesfor theRoza
Memberof the ColumbiaRiver BasaltGroup
The model for the formation of the internal divisions
within an inflated pahoehoe sheet lobe (Figure 13)
provides a means to estimate the duration of the effusive
activity that fed a lobe. Assumingthat the Roza flow
field was emplacedas an inflated pahoehoeflow field,
we now discussits eruption duration and the range of
volumetric effusion rates. We argue that many other
CRB (and other CFB) lava flows were emplaced in a
similar fashion, but we do not at this time have the field
data to quantify their eruptiondurationsor effusionrates.
Our model indicates that the boundary between the
vesicular upper crust and the dense lava core marks the
time when the flux of fresh lava into a lobe ended and
the fluid interior of the lobe became stagnant (Figure
13c). By calculating the time required for the upper
crust to form, it is possibleto estimatethe duration over
which the lobe was being fed fresh lava. In Hawaii, the
growth of thicknessof the upper crust conformsto the
empirical equation
Hc = 0.0779t1/2 ( 1)
where Hc is the thicknessof the upper crust in meters,
0.0779 is an empirically determined constant, and t is
time in hours [Hon et al., 1994]. We expect some
differences in the cooling rate of CRB and Hawaiian
lavas because of differences in rainfall and thermal
properties (heat capacity, diffusivity, and latent heat of
crystallization). We are currently investigating these
differences using a modified version of the thermal
model of Keszthelyi and Denlinger [1996] and
preliminary results suggest that the effects are small
(<25% errors in estimated
400 EMPLACEMENT OF FLOOD BASALT FLOWS

Figure 10. Photomicrographsof thin sectionsof lava samples

from the upper crust, core, and basal zone of the Roza member.
(a) Upper crust of lower Roza sheet flow at Summer Falls
(RSF on Figure 1). Upper photo plane polarized light (PPL),
lower photo reflected light (RL); field of view 1.3 mm across
(100 X). Note opaque tachylitic glass between plagioclase
pheno- and microcrysts in PPL. In RL the tachylite glass is
seento contain numerousdendritic opaque oxides, a reflection
of rapid quenching (but slower than basal selvage). (b) Core of
samelobe as in (a). Upper photo PPL; lower RL: field of view
5.2 mm across (25X). Note phenocrystsof plagioclase and
coarselycrystalline groundmass,including plagioclase,olivine,
pyroxene, and acicular opaque oxide microphenocrysts,
especiallywell seen in RL. (c) Basal glassy selvage (rapidly
cooled and quenched sideromelane) of the lower Roza sheet
lobe at Frenchman Springs Coulee (RFS on Figure. 1). Field
of view 5.2 mm across(25X); view changedslightly between
top (PPL) and lower (RL) photos. Note sparseopaque oxide
microcrysts. Phenocrystsand microphenocrysts are plagioclase
except for a pyroxene/plagioclaseglomerocryst at lower right
of
 SELF ET AL. 401

Assuming that equation (1) applies to CRB flows as

well as those in Hawaii, the thicknesses of the upper
! crusts of individual Roza sheet lobes (Table 2) indicate
4 that they were active for months. Most outcropsof the
Roza consist of two or more sheet lobes. Because it is
5
not likely that an underlying lobe would actively inflate
with an overlying active flow, it is possible to sum the
durations calculated for each lobe to arrive at a
minimum time required to form the flows at each
outcrop. For example, the outcrop at Summer Falls
3
shown in Figure 8 has three sheet lobes that we
calculate were active for 6.8, 11, and >9 months, based
on the thickness of their respective upper crusts
[Thordarson, 1995]. Thus this outcrop records over 2
years of effusive activity. Note that this result assumes
that the overlying flow arrived the moment the
underlying flow stopped inflating, which may be a
reasonableapproximation. The contactsbetween Roza
sheets are often welded together, suggesting that the
overlying unit arrived at least before the underlying unit
cooled completely.
While it is relatively straightforward to estimate the
duration of activity at each outcrop, it is not as clear
how to sum these durations. Unfortunately, it seemsthat
no location records the entire Roza eruption becauseno
PWH single outcrop contains all the chemical subtypesfound
by Martin [1989, 1991]. To estimatethe durationof the
entire Roza eruption, we sum the longest recorded
duration of each of the five lava flows identified within
the Roza flow field (see Table 2; Figure 1). This leads
to a total eruption duration of about 14 years (Table 4).
I I I I There are uncertainties in this calculation. We have
I
25 km already noted that estimated durations may be in error
by as much as 25% due to differences in the cooling
WA rates in the CRB versusHawaii. It is also possiblethat
more than one of the lava flows of the Roza flow field

OR 2 was active at some time in the eruption. Finally, the

chemical differences between the flows are subtle, so it
is possible that some flows are misidentified at some
Gran6e outcrops. However, becausethere is an outcrop (Horton
Grade) that appears to record 6.4 years of activity, we
feel that using the average durations of each lava flow
• Proximal spatter (for a total eruption duration of 5.9 years (Table 4)), is
E] Vent accumulation excessivelyconservative. In any case, we feel confident
1
A Dike

-'.•
ß Outcrop
ofRoza
lava Figure 11. Sketch map of featuresalong the Roza fissure and
fissure segmentsthought to be related to each Roza lava flow
2 (1-5) [after Swanson et al., 1975; Martin, 1989; and work
reported in this study]. PWH = Potter White Hill; WI =
Winona; BB = Big Butte; WA = Washington; OR =
 402 EMPLACEMENT OF FLOOD BASALT FLOWS

TABLE 3. Physicaland ChemicalPropertiesof the Roza and GingkoLava Flows, WanapumBasalts,CRB.

Physical
Property Dike selvage(MgO) Lava selvage (MgO) Dike selvage (CaO) Lava selvage
(CaO)
Roza flow

Temperaturea (øC) 1095 1084 1105 1100

Viscosity
b (Pa s): 1%H2¸ 200 380 175 190
0.5%H20 350 735 310 330

Gingko flow c
Temperature (øC) 1090 1070
Viscosity (Pa s) 630

Chemical

Oxide SiO2 TiO2 A1203 Fe203 MnO MgO CaO Na20 K20 P205
Roza d 50.23 3.12 13.40 15.25 0.22 4.47 8.61 2.75 1.29 0.68
Gingkoe 51.55 3.08 14.38 14.19' 0.23 4.16 8.03 2.34 1.23 0.58

a Temperaturecalculatedusing the empirical geothermometers of Helz and Thornber [1987] basedon compositions(MgO or CaO)
of Roza dike and lava flow glassyselvages[after Thordarson,1995].
b Viscosity
calculated
usingthemethodof BottingaandWeill[1972]for 1.0and0.5 wt%H20, estimated
reasonable
valuesfor
Roza magma and lava, respectively[ Thordarson,1995].
c Temperaturefrom Ha and Cashman [1995] basedon MgO glassgeothermometer.Viscosity from Ha and Cashman[1996].
dAverage(n=73)whole-rock
majorelement
analysis
of Rozachemical
subtype
IIA of Martin[1989],themostvoluminous
type,
equivalent to flow 3 in this paper.
e Average (n=38) whole-rock major element analysisof Gingko flow [Beesonet al., 1985].
* Total iron as FeO.

5O
in suggestingthat the emplacementof the 1300 km3
Roza flow field took on the order of a decade.
4O Even with the eruption continuing over a period on
the order of 10 years, the estimated volumetric flux of
lava during the Roza eruption is very large when
• 30
compared to historical eruptions. We calculate an
averagetotal eruptionrate of •-4000 m3/sfor the Roza.
•- 20 This is roughly equal to the peak eruption rate of the
o
1783-84 Laki eruption in Iceland, the largest historical
basaltic eruption [Thordarson and Self, 1993]. These
lO
high total eruption rates for the Roza are not
unreasonablegiven the potential lengths of the fissures
involved. A 4-km-long fissure segment active at one
0 500 1000 1500 2000 time would give averageeruptionrates on the order of 1
Lava Viscosity(Pa s) m3/sper meterlengthof fissure(or •-3000kg/sper meter
Figure 12. Plot of transition from laminar to turbulent flow as length of fissure), which is a typical value for many
a function of flow thickness and lava viscosity. Curves Hawaiian eruptions and only about half the peak
computed for a slope of 0.1% (0.05ø), appropriatefor most of eruptionrate of the Laki fissure. Fissuresegments4 km
the flows in the CRB [Shaw and Swanson, 1970] and a long need be active for only an averageof 3.2 monthsin
reasonable value for other CFB provinces. Estimated order to migrate across the 150-km-long Roza fissure
viscositiesfor CRB lavas, for realistic eruption temperatures,a system in 10 years. For comparison, the Laki fissure
reasonable range of volatile contents, and corrected for system was 27 km long and producedlava for 8 months,
entrained crystals and bubbles, range from 120 to 5500 Pa s.
though •-60% of the lava erupted from five fissures
This plot indicates that only flows with initial emplacement
thicknessesgreater than 20-25 m in the viscosity range of the
totaling 13.5 km in length in just 1.5 months[Thordarson
Roza lava could theoretically have flowed turbulently. and Self,
 SELF ET AL. 403

Breakouts(hours) a Stagnantfreezing C
(months-decades)
HVZ

:øsecond
::ol boiling
silicic in
differentiateøø
.....................................
•.0•.;o.•
.............................................................................................
•:?•.•...........
....... '• ...• .• ...... • • .... '•. .•, ß

Inflation(days-years) b Frozenlavaflow d
-Upper crust
cooling-induced
J• • •.. •.:½'.•";:•h' 'h -/ f t
.... , ½
'
brittle crust

visco-elastic
new
break
•.:..•.r'"'"'"'"''
'•"'7"•o øoøO
œ,2•...-:-••
øo:
ø?i'"'"'"--•:'•
o oo TM o o - 0 'o
_o o
layer
molten core

• • • • • • •. IBasalcrust
I

Figure 13. Schematic

crosssections
of emplacement
of a genericinflatingpahoehoe
sheetflow. Verticalscalevariesfrom 1-5
m for Hawaiian flows to 5-50 m for the CRB flows. (a) Flow arrivesas a small, slow-moving,lobe of molten lava held inside a
stretchable,chilled viscoelasticskin with brittle cruston top. Bubblesare initially trappedin both the upperand basalcrusts.(b)
Continuedinjectionof lava into the loberesultsin inflation(lifting of the uppercrust)and new breakouts.During inflation,
bubblesrisingfrom the fluid corebecometrappedin the viscoelastic mushat the baseof the uppercrust,forminghorizontal
vesicularzones.The growthof the lower crust,in whichpipe vesiclesdevelop,is muchslower. Relativelyrapidcoolingand
motionduringinflationresultsin irregularjointingin the uppercrust.(c) After stagnation,diapirsof vesicularresiduumform
verticalcylindersandhorizontalsheetswithinthe crystallizing lavacore. Slowcoolingof the stationaryliquidcoreformsmore
regularjoints.(d) Emplacementhistoryof flow is preservedin vesicledistribution
andjointingpatternof frozenlava.

Data from the vent areas of the Roza also support
analogiesto Laki and other historical fissure eruptions.
Significant amounts of volatiles were degassedat or
near the vent [Thordarson et al., 1996], so some
mechanismto promote volatile loss, such as high fire-
fountains, must have occurred. The dearth of vent
edifices supports the notion of high fire-fountaining.
Such fire-fountains generally do not producehigh cones;
insteadthey producefields of welded spatterfallout and
rootless lava flows, as in the 1986 Izu-Oshima eruption
[Sakaguchiet al., 1988].
More generally, the vent structuresin the CRB are
not extraordinarywhen comparedwith historicalbasaltic
eruptive vents elsewhere. Dike widths are similar and
spatter deposits are remarkably comparable in scale.
Thus, historicalfissureeruptionscan probablybe used as
analogs of the Roza and other CRB fissures, even
thoughthe total length of the Roza systemis larger than
that of any known historical eruption. No evidence
indicates that the activity producingthe immenseflows
of the CRB was significantly different than what has
been observed historically, except that high total
eruptionrates were maintainedfor much longerbecause
of the longer fissure system.
4.3. Flow Field Evolution and Thermal Efficiency
Working from the ideasproposedabove,we produce
a revisedpicturefor the emplacementof the Roza flow
field (Figure 14). We envision the advanceof each
Roza flow as being composedof a slowly advancing
(-1-10 cm/s), broad front (many kilometers wide) of
small (-20-50 cm tall) pahoehoelobes and toes. These
lobes inflate and coalesce over a distance of perhaps
several tens of meters and over a time scale of hours.
This part of the flow would be little differentfrom the
pahoehoesheet flows in Hawaii [Honet al., 1994] or
Iceland [Thordarson and Self, 1993], except for being
significantlywider. The flow lobeswouldcoalesceand
continueto inflate, producingbroad,flat sheet-like
404 EMPLACEMENT OF FLOOD BASALT FLOWS

TABLE4. Estimated
Duration
of theRozaEruption
UsingEquation
(1)Based
ontheAverage
andMaximum
Crustal
Thicknesses Measured in Roza Lava Flows.
Lavaflow Maximum
t (years) Average
t (years) Location of maximum t Other locations
1 0.57 0.44 RSF RAC, RBB
2 3.84 1.69 RPR RBB, RBL, RHG, RDF, RSF,
RVL, RWV, RSC
3 1.26 0.58 RMC RBB, RBL, RDF, RSF, RSL
4 4.22 1.83 RHG RDC, RFS, RPR
5 4.22 1.33 RC RDC, RLG

Total years 14.10 5.90

SeeTables1 and2 for the locationkey andmeasured
crustthicknesses.

1
• Active
fissure/vent
River
Flow
limit
ø• State
border

3 !lB

Figure14.Cartoon
map-view
of a possible
scenario
forthedevelopment
of theRozacompound
pahoehoe
flow-field
based
on
work
ofMartin[1989;
1991]andThordarson
[1995].Lavatravels
fromthevent
under
aninsulating
crust
ina preferred
pathway
(dashed
lines).
Thisfluxoflavaisused
toboth
inflate
thesheet
flowandtofeednewbreakouts,
whicharevirtually
theonly
placewhereincandescent
lavais exposed.Activeflowsarediverted
around
earlier-formed
flowsunless
the activeflowsare
inflated
toa thickness
greater
than
earlier
units.1-5= thefivemajor
flows
thatformtheRoza
flowfield;active
fissure
segment
at eachstageshownby thick

up to severalkilometerswide. These sheetlobes would
continue to inflate and feed the lava flow front for many
months,reachinga final thicknessof up to 50 m or more.
It is possible that several sheet flows could be fed
simultaneouslyfrom different fissure segments. As any
given lava flow becameinactive, as a result of either its
section of the fissure shuttingdown or having the lava
diverted to a new flow, it would become a temporary
barrier for new flows from either the same or subsequent
eruptions.
This picture must still answer the primary constraint
on the previous emplacement model of Shaw and
Swanson [1970]; the lava must not cool significantly
during its transport across hundredsof kilometers of
surfaceflowage. Ho and Cashman[1995] suggestedon
the basisof three data pointsthat the Gingko flow of the
Frenchman Springs Member of the CRB cooled an
averageof 0.06-0.11øC/km(Table 3). Keszthelyi[1995b]
showed that this kind of thermal efficiency would
theoretically be feasible for lava tubes in the CRB to
achieve. However, we have found no evidence so far for
cylindrical, drained lava tubes in the CRB.
We do not find the lack of evidence of lava tubes
problematic. On the shallow slopesat the time of CRB
flow emplacement, it is unlikely that lava tubes could
have drained. Furthermore, inflating sheet flows are
expectedto form broad preferredinternal pathwayswith
elliptical crosssections,not cylindrical conduits. This is
commonlythe case on the low slopesof the coastalflats
on Kilauea and was proposed much earlier for the
Buckboard Mesa flow, Nevada, by Lutton [1969].
Thermal modeling of such sheet-likepreferredpathways
suggests that lava can be transported hundreds of
kilometers with no detectable cooling [Keszthelyi and
Self, 1996].
5. POTENTIAL FOR ATMOSPHERIC EFFECTS
FROM FLOOD BASALT ERUPTIONS
For flood basalt eruptions to have had widespread
climatic impact, the gases emitted by the volcanic
activity must have reached high enough into the
atmosphereto be transportedwidely around the Earth.
The height to which volcanic gasescan be carried from
a fissure eruption dependscritically on the masseruption
rate per unit length of fissure [Stothers et al., 1986;
Stothers, 1989], the volatile content of the magma
[Wilson and Head, 1981], and the moisturecontent of the
atmosphereat the time [Woods, 1993]. Becausethere
are so many variables when modeling an ancient
eruption like that which formed the Roza flow, at
SELF ET AL. 405
presentwe can form only a generalidea of the heightsto
which plumes might reach above the fissures. At
eruptionrates of --102 kg/s per meter lengthof fissure,
estimatedplume heightswould be 3-6 km abovethe fire-
fountains;and at--3 x 103 kg/s per meter length of
fissure, 8-11 km [Woods, 1993]. Mass eruption rates
along the fissuresmay have waxed and waned over two
orders of magnitude, as at the 1783 Laki eruption
[Thorarinsson, 1968; Thordarson and Self, 1993], and
thus significantly higher plumes would be expected at
some periods in the eruption.
Clearly, from the first-order considerationspresented
above, it is plausible that the Roza, and presumably
other flood basalt eruptions, could inject volcanic gases
into the uppermost troposphere and even the lower
stratosphere (presently --12-13 km altitude at mid-
latitudes). Once aloft, the volatiles (the most important
of which is sulfur as SO2 or H2S) would behave much
like the products of any volcanic eruption and form
sulfate aerosols. Concentrations of S, F, and C1 were
measured in glassy samples and glass inclusions from
phenocrystscollected from dikes, near-vent spatter, lava
selvages, and lava cores (Figure 15) [Thordarson and
Self, 1996]. These analyses and a mass balance
calculation show that 66% of the total sulfur was lost
during the eruption processand that a significantfraction
was also degassedduring the flow and crystallizationof
the lava. The degassingduring crystallizationwould not
have formed an upper atmospheric aerosol perturbation
as it would have been confined to the boundary layer of
the lower troposphere,and it shouldnot have been trans-
ported globally. However, the local and regional impact
of such volcanic smog can be dramatic, as it was during
the Laki eruption [Thorarinsson, 1981; Thordarson and
Self, 1997].
The estimatedmass of volatiles degassedat the vents
during the Roza eruptionis staggering;9,000 Mt of SO2,
1300 Mt of HF, and 400 Mt of HC1. For comparison,the
1991 Mount Pinatubo eruption released an estimated 20
Mt of SO2 [e.g., McCormick et al., 1995]. Thus, if the
Roza flow field was eruptedover a period of 10 years, it
would be approximately equivalent to four times the
Pinatubo upper atmospheric SO2 perturbation every
month, maintained for a decade. Along with sulfuric
acid aerosols,the formation of acid droplets in the lower
atmosphere from the F and C1 would lead to extensive
acid rain. It should also be noted that considerablymore
voluminousflows (up to 4000 km3) are known from
regional mapping and correlation in the Grande Ronde
Basalt Formation [Reidel et al., 1989] and that initial
analyses on other Wanapum Formation basalts
406 EMPLACEMENT OF FLOOD BASALT
km Wind
18
12
/ Eruption
column Distal
haze
1370ppmS
9617MtSO2
-6 ,. , .( • •
,/Vents .... , .......
.....808.tO
,;•j•}&;,;,T;;,;,?55,5.55,5,5,57
,-•,.5,••5;, ',••,;,-;.--;-
.',','vC} la'b• '•Y •' •••k• •'-'-',' 520 ppmS •'-'-'-'-",
Figure 15. Schematic illustration of Roza eruption (not to
scale) and degassing budget, based on example of Laki
eruption[ Thordarsonet al., 1996]. The amountof S retainedin
samplesfrom various eruption stagesand the total amountof
SO2 dissolvedin the Roza magma prior to eruptionare given,
as well as the estimated SO2 yield at the vents (causing a
widespread(distal) haze or aerosol cloud) and from the lava
flow (causinga low altitude (local) haze or dry fog).
indicate the potential for large sulfur releases during
thoseeruptions[Ewart, 1987].
The climatic and environmental consequences of
such emissions maintained over periods of years to
decades cannot at present be modeled with any
confidence, but must have been severe.
atmospheric and environmental impact of flood basalt
volcanism alone could cause mass extinctions is not yet
clear. Certainly, individual flood lava events lasting
even decades are unlikely to have had such an extreme
effect, but, as typified by the CRB, flood basalt
provinces are composedof hundredsof such eruptions.
During the period of peak output, 50% of the Grande
Ronde lavas were emplaced in approximately 300,000
years from about 16.0 to 15.7 Ma [Baksi, 1989]. Even in
this time of peak activity, individual eruptions would
have average recurrenceintervals of 5,000-10,000 years,
which may have given the environment sufficient time
to recover between eruptions. The most reasonable
statement,given current knowledge, is that a continental
flood basalt eruption probably could not cause mass
extinctions, but a series of them during the growth of a
CFB province would have been able to stress the
environment to such an extent that any other major
perturbationwould have had a more extremeeffect.
FLOWS
6. CONCLUSIONS
We have shown what we believe to be
incontrovertible field evidence that the Roza Member
the CRB is a compoundpahoehoeflow field. We also
of
find pahoehoeflow fields to be common throughoutthe
Localhaze CRB and in other flood basalt provinces. This evidence
is in the form of surface textures, morphologicfeatures
at the flow tops, and the internal structureswithin the
lava flows. The three-part internal structurewe have
describedherein is ubiquitousin inflated pahoehoelava
flows of all scalesacrossthe globe and is different from
the internal structureof rapidly emplaced aa or aa-like
195ppmS '-"-',' lava flows. We have found no evidence for rapid,
turbulent emplacement as envisioned by Shaw and
Swanson [1970] in the CRB. Instead, the features we
have described in the Roza and other CFB lava flows
are easily explained by their emplacement as inflating
sheetflows. An emplacementmodel basedon pahoehoe
flow fields in Hawaii, Iceland, and elsewhere is able to
explain the gross geometry, internal structure, and
evidence for minimal cooling observedin the Roza and
other CFB lava flows.
Having said this, we caution the reader that our
simple descriptions and cartoon drawings are not
intended to explain, in detail, every feature seen in
every outcrop. For example, some flows in the CRB
have rubbly upper crusts that may have formed in a
manner transitional between pahoehoe and aa, (e.g.,
slabby pahoehoe). Instead, our model is intended to
show that the inflation processis generally applicableto
Whether the flood basalt lava flows. Detailed studies are needed to
decipher lava flow emplacement history at specific
locations.
Examination of the physical features of the Roza
Member has allowed us to quantitatively speculateon
its emplacement. An emplacementmodel, combined
with the cooling model of Hon et al. [1994], permits an
estimate of the duration of active flow recorded at each
outcrop of the Roza flow field. Translating theselocal
durations to the total eruption duration is not
straightforward. We expect that many of the lava flows
at various outcrops were active simultaneously,and we
have evidence that no single outcrop recordsthe entire
Roza eruption. However, because the examined
outcropsrecord activity for up to 6.4 years, we suggest
that the Roza eruptionlastedfor about a decade.
Using a 10-year eruption duration, we arrive at an
averageeffusionrate for the Roza of-4000 m3/s. This
average rate is as high as the peak rate of the largest
historical basaltic eruption, the 1783-1784 Laki

in Iceland. Because the Roza fissure system is 150 km
long, the high total effusion rate can be accommodated
by moderateactivity along a small fraction of the fissure
system. We therefore suggest that the Roza eruption
probably consisted of a number of shorter fissure
segments. Each segmentmay have fountainedfor just a
few monthsbefore activity migrated to a different part of
the fissure system.
The measured release of immense volumes of sulfur,
chlorine, and fluorine at flood basalt vents over periods
on the order of a decade may have had a strong,
detrimental effect on global climate. This provides a
plausible, though not yet proven, process to link
continental flood basalt eruptions to mass extinctions
[Courtillot, 1994].
Whereas the causes of flood basalt volcanism and the
sources of magmas that form flood basalt provinces
appear to be relatively well understood[e.g., Carlson,
1991], our knowledgeof the physicalprocessesoccurring
during flood basalt events, such as venting mechanisms,
lava flow emplacement, and degassingmechanisms,is
much poorer. Our initial studies on CRB lava flows
strongly suggest that the previous model of enormous
flow rates and catastrophiceruptionmechanismsof flood
basalt lavas is largely untenable and that CFB lavas in
general were emplaced more gradually as pahoehoe
sheet flows forming extensive lava flow fields, though
still at very high total effusion rates when comparedto
most basalticeruptionswitnessedby man.
We hope that this work will spur further, more
detailed studies on the physical volcanology of flood
basalt lava flows.
Acknowledgments.Supportfor this work was provided by
NSF grantsEAR-9118755 and 9316881, NASA grantsNAG5-
1839 and NAGW-3721, by a NASA Global Change Fellowship
to ThTh, and by an NSF post-doctoralfellowship to LPK. We
thank George Walker, Mark Murphy, Phil Long, Mike
Rampino, Steve Reidel, Terry Tolan, and Sara Finnemore for
various inputs to this study. Ken Hon and John Wolff are
thanked for critical and helpful reviews. This is SOEST
contribution no. 4169.
REFERENCES
Agashe, L. V., and R. B. Gupte, Mode of eruption of the
Deccan Traps basalts,Bull. Volcanol.,35,591-601, 1971.
Anderson, D. L., Sublithospheric mantle as the source of
continental flood basalts - the case against the continental
lithosphereand plume head reservoirs,Earth Planet. Sci. Lett.,
123, 269-280, 1994.
Arndt, N. T., G. K. Czamanske, J. L. Wooden, and V. A.
SELF ET AL. 407
Fedorenko, Mantle and crustal contributions to continental
flood volcanism, Tectonophysics,223, 39-52, 1993.
Atkinson, F. A., Some remarkable features of flows from
Undara, AGU Chapman Conference on Long Lava Flows,
ConferenceAbstract Volume, JamesCook University of North
Queensland,Townsville, pp. 4-5, 1996.
Aubele, J. C., L. S. Crumpier, and W. E. Elson, Vesicle
zonation and vertical structure of basalt flows, J. Volcanol.
Geotherm. Res., 35, 349-374, 1988.
Baksi, A. K., Reevaluation of the timing and duration of
extrusionof the Imnaha, Picture Gorge, and Grande Ronde
Basalts, Columbia River Basalt Group, in Volcanismand
Tectonismin the Columbia River Flood-Basalt Province, Spec.
Pap. 239, editedby S. P. Reidel, and P. R. Hooper, pp. 1-20,
Geological Society of America, Boulder, CO, 1989.
Bates, R. L., and J. A. Jackson (Eds), Glossary of Geology, 3rd
edition, 788 pp., American Geological Institute, Alexandria,
VA, 1987.
Beeson, M. H., K. R. Fecht, S. P. Reidel, and T. L. Tolan,
Regional correlations within the FrenchmanSprings Member
of the Columbia River Basalt Group: New insights into the
middle Miocene tectonics of northwestern Oregon, Oregon
Geol., 47, 87-96, 1985.
Bingham, J. W., and K. L. Walters, Stratigraphyof the upper
part of the Yakima Basalt in Whitman and easternFranklin
Counties, Washington, U.S. Geol. Surv.Prof. Pap. 525-C, 87-
90, 1965.
Bottinga, Y., and D. F. Weill, The viscosity of magmatic
silicate liquids: A model for calculation, Am. J. Sci., 272,
438-475, 1972.
Campbell, I. H., and R. W. Griffiths, Implications of mantle
plume structure for the evolution of flood basalts, Earth
Planet. Sci. Lett., 99, 79-93, 1990.
Carlson, R. W., Physical and chemical evidence for the cause
and source characteristics of flood basalt volcanism, Austr. J.
Earth Sci., 38, 525-544, 1991.
Chitwood, L. A., Inflated basaltic lava: examples of processes
and landforms from central and southeast Oregon, Oregon
Geol., 56, 11-21, 1994.
Courtillot, V. E., Mass extinctionsin the last 300 million years:
one impact and seven flood basalts, Israeli J. Earth Sci., 43,
259-266, 1994.
Courtillot, V. E., G. Fdraud, H. Maluski, D. Vandamme, M. G.
Moreau, and J. Besse, The Deccan flood basalts and the
Cretaceous/Tertiaryboundary, Nature, 333, 843-846, 1988.
Courtillot, V. E., J. Besse, D. Vandamme, R. Montigny, J. J.
Jaeger, and H. Capetta, Deccan flood basalts at the
Cretaceous-Tertiaryboundary,Earth Planet.Sci. Lett., 80, 361-
374, 1986.
Crisp, J., and S. Baloga, Influence of crystallization and
entrainment of cooler material on the emplacement of
basaltic aa lava flows, J. Geophys.Res., 99, 11,819-11,831,
1994.
Ewart, J. W., Sulfur in the FrenchmanSprings member of the
Wanapum Basalt in Washington and Oregon, Geol. Soc.
408 EMPLACEMENT OF FLOOD BASALT FLOWS
Abstracts with Programs, 19, Cordilleran Section, 376, 1987.
Finnemore, S. L., S. Self, and G. P. L. Walker, Inflation
features in lava flows of the Columbia River Basalts
(abstract),Eos Trans.AGU, 74(46), Fall Meeting Suppl., 555,
1993.
Goff, F., Vesicle cylinders in vapor-differentiatedbasalt flows,
J. Volcanol. Geotherm. Res., 71, 167-185, 1996.
Gregg T. K. P., and J. H. Fink, Quantification of submarine
lava-flow morphologythroughanalog experiments,Geology,
23, 73-76, 1995.
Greenhough J. D., and J. Dostal, Cooling history and
differentiation of a thick North Mountain Basalt flow (Nova
Scotia, Canada), Bull. Volcanol., 55, 63-73, 1992.
Griffiths, RoW., and J. H. Fink, Solidification and morphology
of submarine lavas: A dependence on extrusion rate, J.
Geophys.Res., 97, 19 729-19 737, 1992.
Helz, R. T., and C. R. Thornber, Geothermometryof Kilauea Iki
lava lake, Hawaii, Bull. Volcanol., 49,651-668, 1987.
Ho, A., and K. V. Cashman, Geothermometry of the Gingko
Flow, Columbia River Basalt Group (abstract), Eos Trans
AGU, 76(46), Fall Meeting Suppl., 679, 1995.
Ho, A., and K. V. Cashman,Temperatureconstraintson a flow
of the Columbia River Basalt Group (abstract), AGU
Chapman Conference on Long Lava Flows, Conference
Abstract Volume, James Cook University of North
Queensland,Townsville, p. 22-23, 1996.
Hon, K., J. Kauahikaua, R. Denlinger, and K. Mackay,
Emplacement and inflation of pahoehoe sheet flows: Obser-
vations and measurements of active lava flows on Kilauea
Volcano, Hawaii, Geol. Soc. Am. Bull., 106,351-370, 1994.
Hooper, P. R., The Columbia River Basalts, Science, 215,1463-
1468, 1982.
Hooper P. R., and C. J. Hawkesworth,Isotopicand geochemical
constraints on the origin and evolution of Columbia River
Basalts, J. Petrol., 34, 1203-1246, 1993.
Hooper, P. R., W. D. Kleck, C. R. Knowles, S. P. Reidel, and
R. L. Thiessen,Imnaha Basalt, Columbia River Basalt Group,
J. Petrol., 25,473-500, 1984.
Jarvis, R. A., On the cross-sectional geometry of thermal
erosion channels formed by turbulent lava flows, J. Geophys.
Res., 100, 10,127-10,140, 1995.
Kent, R. W., M. Storey, A.D. and Saunders, Large igneous
provinces--sites of plume impact or plume incubation,
Geology,20, 891-894, 1992.
Keszthelyi, L., Measurements of the cooling at the base of
pahoehoeflows, Geophys.Res.Lett., 22, 2195-2198, 1995a.
Keszthelyi, L., A preliminary thermal budget for lava tubes on
the Earth and planets,J. Geophys.Res. 100, 20 411-20 420,
1995b.
Keszthelyi, L., and R. Denlinger, The initial cooling of
pahoehoelava flows, Bull. Volcanol.,58, 5-18, 1996.
Keszthelyi, L. P., and D.C. Pieri, Emplacementof the 75-km-
long Carrizozo lava flow field, south-centralNew Mexico, J.
Volcanol. Geotherm. Res., 59, 59-75, 1993.
Keszthelyi, L., and S. Self, Some thermal and dynamical
considerationsfor the emplacementof long lava flows, AGU
Chapman Conference on Long Lava Flows, Conference
Abstract Volume, James Cook University of North
Queensland,Townsville, pp. 36-38, 1996.
Keszthelyi, L., S. Self, and Th. Thordarson, Application of
recent studies on the emplacementof basaltic lava flows to
the Deccan Traps, Mere. Geol.Soc.India, in press,1997.
Kilburn, C. R. J., Pahoehoe and aa lavas: A discussion and
continuationof the model of Petersonand Tilling, J. Volcanol.
Geotherm. Res., 11,373-382, 1981.
Kilburn, C. R. J., and R. M. C. Lopes, General patternsof flow
field growth: aa and blocky lavas, J. Geophys. Res., 96,
19,721-19,732, 1991.
Landon, R. D., and P. E. Long, Detailed stratigraphyof the N2
Grande Ronde Basalt, Columbia River Basalt Group, in the
central Columbia Plateau, in Volcanism and Tectonism in the
ColumbiaRiver Flood-Basalt Province, Spec. Pap. 239, edited
by S. P. Reidel, and P. R. Hooper, pp. 55-66, Geological
Society of America, 1989.
Long, P. E., and B. J. Wood, Structures,textures, and cooling
histories of Columbia River basalt flows, Geol. Soc. Am. Bull.,
97, 1144-1155, 1986.
Lutton, R. J., Internal structure of the Buckboard Mesa Basalt,
Bull. Volcanol., 33, 579-593, 1969.
Macdougall, J. D. (Ed.), ContinentalFlood Basalts, 341 pp.,
Kluwer Academic Publishers, Dordrecht, 1988.
Mackin, J. H., A stratigraphicsectionin the Yakima Basin and
the Ellensburg Formation in south-centralWashington, Wash.
Div. Mines Geol., Dept. Conserv.,Rept.Invest.19, 45 pp., 1961.
Manga, M., Waves of bubbles in basaltic magmasand lavas, J.
Geophys.Res., 101, 17,457-17,465, 1996.
Mangan, M. T., K. V. Cashman, and S. Newman, Vesiculation
of basaltic magma during eruption, Geology, 21, 157-160,
1993.
Mangan, M. T., T. L. Wright, D. A. Swanson,and G. R. Byerly,
Major oxide, trace element, and glass chemistry pertinent to
regional correlation of Grande Ronde Basalt flows, Columbia
River Basalt Group, Washington,U.S. Geol. Surv. Open-File
Rep. 85- 747, 74 pp., 1985.
Mangan, M. T., T. L. Wright, D. A. Swanson,and G. R. Byerly,
Regional correlation of Grande Ronde Basalt flows,
Columbia River Basalt Group, Washington, Oregon, and
Idaho, Geol. Soc. Am. Bull., 97, 1300-1318, 1986.
Martin, B. S., The Roza Member, Columbia River Basalt
Group: Chemical stratigraphy and flow distribution, in
Volcanism and Tectonism in the Columbia River Flood-Basalt
Province, Spec. Pap. 239, edited by S. P. Reidel, and P. R.
Hooper, pp. 85-104, Geological Societyof America, 1989.
Martin, B. S., Geochemical variations within the Roza
member, Wanapum basalt, CRBG: Implications for the
magmatic process affecting continental flood basalts,
unpublished Ph.D. thesis, University of Massachusetts,513
pp., 1991.
Mattox, T. N., C. Heliker, J. Kauahikaua, and K. Hon,
Development of the 1990 Kalapana flow field, Kilauea
Volcano, Hawaii, Bull. Volcanol., 55,407-413, 1993.
McCormick, M.P., L. W. Thompson, C. R.

Atmosphericeffects of the Mt Pinatuboeruption,Nature, 373,
399-404, 1995.
McMillan, K., P. E. Long, and R. W. Cross, Vesiculation in
Columbia River basalts, in Volcanism and Tectonism in the
ColumbiaRiver Flood-Basalt Province, Spec.Pap. 239, edited
by S. P. Reidel, and P. R. Hooper, pp. 157-167, Geological
Society of America, 1989.
Nichols, R. L., Flow-units in basalt, J. Geol., 44, 617-630, 1936.
Peng Z. X., J. J. Mahoney, P. Hooper, C. Harris, and J. Beane,
A role for lower continental crust in flood basalt genesis?
Isotopic and incompatible element study of the lower six
units of the western Deccan Traps, Geochim. Cosmochim.
Acta, 58,267-288, 1994.
Peterson,D. W., and R. I. Tilling, Transition of basaltic lava
from pahoehoe to aa, Kilauea Volcano, Hawaii: Field
observationsand key factors, J. Volcanol. Geotherm.Res., 7,
271-293, 1980.
Phadke, A. V., and Sukhtankar,R. K., Topographic studiesof
Deccan Trap hills around Poona, India, Bull. Volcanol., 35,
709-718, 1971.
Pinkerton,H., and R. S. J. Sparks,The 1975 sub-terminallavas,
Mount Etna: a case history of the formation of a compound
lava field, J. Volcanol. Geotherm. Res., 1, 167-182, 1976.
Pinkerton, H., and R. Stevenson,Methods of determining the
rheological
properties
ot•lava from their physico-chemical
properties,J. Volcanol.Geotherm.Res.,53, 47-66, 1992.
Puffer, J. H., and D. L. Horter, Origin of pegmatiticsegregation
veins within flood basalts, Geol. Soc. Am. Bull., 105,738-748,
1993.
Rampino, M. R., and R. B. Stothers, Flood basalt volcanism
during the past 250 million years, Science, 241, 663-668,
1988.
Reidel, S. P., Stratigraphy and petrogenesisof the Grande
Ronde Basalt from the deep canyon country of Washington,
Oregon,and Idaho, Geol.Soc.Am.Bull.,94, 519-542,1983.
Reidel, S. P., and T. L. Tolan, Eruption and emplacement of
flood basalt: An example from the large-volume Teepee
Butte Member, Columbia River Basalt Group, Geol. Soc. Am.
Bull., 104, 1650-1671, 1992.
Reidel, S. P., T. L. Tolan, P. R. Hooper, M. H. Beeson, K. R.
Fecht, R. D. Bentley, and J. L. Anderson,The Grande Ronde
Basalt, Columbia River Basalt Group: stratigraphic
descriptions and correlations in Washington, Oregon, and
Idaho, in Volcanism and Tectonism in the Columbia River
Flood-Basalt Province, Spec. Pap. 239, edited by S. P. Reidel,
and P. R. Hooper, pp. 21-53, GeologicalSociety of America,
1989.
Renne, P. R., M. Ernesto, I. G. Pacca, R. S. Coe, J. M. Glen, M
Pr6vot, and M. Perrin, The age of Parana flood volcanism,
rifting of Gondwanaland,and the Jurassic-Triassicboundary,
Science, 258,975-981, 1992.
Renne, P. R., Z. Zichao, M. A. Richards, M. T. Black, and, A.
R. Basu, Synchrony and Causal Relations between Permian-
Triassic Boundary Crises and Siberian Flood Volcanism,
Science, 269, 1413-1416, 1995.
Rowland, S. K., and G. P. L. Walker, Pahoehoe and aa in
SELF ET AL. 409
Hawaii: Volumetric flow rate controls the lava structure, Bull.
Volcanol., 52, 615-628, 1990.
Sakaguchi, K., A. Takada, K. Uto, and T. Soya, The 1986
eruption and products of Izu-Oshima Volcano, Japan, Bull.
Volcanol. Soc. Japan, 33, Special Issue, Prof. K. Nakamura
Memorial Volume, 20-32, 1988 (in Japanese).
Schminke, H. U., Stratigraphy and petrographyof four Upper
Yakima Basalt Flows in south-centralWashington, Geol. Soc.
Am. Bull., 78, 1385-1422, 1967.
Self, S., Th. Thordarson, L. Keszthelyi, G. P. L. Walker, K.
Hon, M. T. Murphy, P. Long, and S. Finnemore,A new model
for the emplacement of Columbia River Basalts as large
inflated pahoehoe lava flow fields, Geophys. Res. Lett., 23,
2,689-2,692, 1996.
Shaw, H. R., Visocosities of magmatic silicate liquids: an
empirical method of prediction, Am. J. Sci., 272, 870-893,
1972.
Shaw, H. R., and D. A. Swanson, Eruption and flow rates of
flood basalts, in Proc. Second Columbia River Basalt
Symposium,editedby E. H. Gilmour and D. Stradling,pp. 271-
299, Eastern Washington State College Press,Cheney, 1970.
Snavely, P. D., N. S. MacLeod, and H. C. Wagner, Miocene
tholeiitic basaltsof coastalOregon and Washington and their
relations to coeval basalts of the Columbia plateau, Geol.
Soc. Am. Bull., 84,387-424, 1973.
Stephenson,P. J., Burch-Johnson,D. Stanton, Long lava flows
in North Queensland--context, characteristics,emplacement
(abstract), AGU Chapman Conference on Long Lava Flows,
ConferenceAbstract Volume, JamesCook University of North
Queensland,Townsville, pp. 86-87, 1996.
Stothers, R. B., Turbulent atmospheric plumes above line
sourceswith an application to volcanic fissure eruptions on
the terrestrialplanets,J. Atmos. Sci., 46, 2662-2670, 1989.
Stothers,R. B., Flood basalts and extinction events, Geophys.
Res. Lett, 20, 1399-1402, 1993.
Stothers, R. B, J. A. Wolff, S. Self, and M. R. Rampino,
Basaltic fissure eruptions, plume heights, and atmospheric
aerosols,Geophys.Res. Lett, 13,725-728, 1986.
Swanson, D. A., Yakima Basalt of the Tieton River area, south-
central Washington, Geol. Soc.Am. Bull., 78, 1077-1110, 1967.
Swanson, D. A., and T. L. Wright, The regional approach to
studyingthe Columbia River Basalt Group, Memoir Geol. Soc.
India, 3, 58-80, 1980.
Swanson, D. A., T. L. Wright, and R. T. Helz, Linear vent
systems and estimated rates of magma production and
eruption for the Yakima Basalt on the Columbia Plateau, Am.
J. Sci., 275,877-905, 1975.
SwansonD. A., T. L. Wright, P. R. Hooper, and R. D. Bentley,
Revisions in stratigraphic nomenclature of the Columbia
River Basalt Group, U.S. Geol. Surv. Bull., 1457-G, 59 pp.,
1979.
Theilig, E., Formation of PressureRidges and Emplacementof
Compound Basaltic Lava Flows, Ph.D. thesis, Arizona State
University, Tempe. 212 pp., 1986.
Thorarinsson, S., On the rate of lava and tephra-productionand
upward migration of magma in four Icelandic eruptions,
410 EMPLACEMENT OF FLOOD BASALT FLOWS

Rundsch., 57,705-717, 1968. Walker, G. P. L., Spongy pahoehoe in Hawaii: a study of

Thorarinsson, S., Greetings from Iceland: ash-falls and
volcanic aerosols in Scandinavia, Geograf.Ann., 18, 63,109-
63,118, 1981.
Thordarson, Th., Volatile release and atmospheric effects of
basaltic fissure eruptions, Ph.D. thesis, University of Hawaii
at Manoa, Honolulu, 580 pp., 1995.
Thordarson, Th., and S. Self, The Laki (Skaftgr Fires) and
Grfmsv6tn eruptions in 1783-1785, Bull. Volcanol., 55, 233-
263, 1993.
Thordarson, Th., and S. Self, Sulfur, chlorine and fluorine
degassingand atmospheric loading by the Roza eruption,
Columbia River Basalt Group, Washington, J. Volcanol.
Geotherm. Res., 74, 49-73, 1996.
ThordarsonTh., and S. Self, Atmospheric and environmental
effects of the 1783-84 Laki eruption, Global and Planetary
Change, 1997, in revision.
Thordarson
)'h., S. Self,N. Oskarsson,
T. Hulsebosch,
Sulfur,
chlorine and fluorine degassingand atmosphericloading by
the 1783-84 Laki (Skaftar Fires) eruption in Iceland, Bull.
Volcanol., 58, 205-225, 1996.
Tolan, T. L., S. P. Reidel, M. H. Beeson, J. L. Anderson, K.R.
Fecht, and D.A. Swanson, Revisions to the estimates of the
areal extent and volume of the Columbia River Basalt Group,
in Volcanism and Tectonism in the Columbia River Flood-
vesicle distributionpatternsin basalt and their significance,
Bull. Volcanol., 51,199-209, 1989.
Walker, G. P. L., Structure, and origin by injection under
surface crust, of tumuli, "lava rises," "lava-rise pits," and
"lava inflation clefts" in Hawaii, Bull. Volcanol., 53,546-558,
1991o
Waters, A. C., Stratigraphic and lithologic variations in the
Columbia River Basalt, Am. J. Sci., 259,583-611, 1961.
White, R. S, and D. McKenzie, Magmatism at rift zones: the
generation of volcanic continental margins and flood basalts,
J. Geophys.Res.,94, 7685-7729, 1989.
Whitehead, J. A., and K. R. Helfrich, Instability of flow with
temperature-dependent viscosity: A model of magma
dynamics,J. Geophys.Res.,96, 4145-4155, 1991.
Wilson, L. and J. V. Head, Ascent and eruption of basaltic
magma on the Earth and Moon. J. Geophys.Res., 86, 2971-
3001, 1981.
Woods, A. W., A model of the plumes above basaltic fissure
eruptions,Geophys.Res.Lett., 20, 1115-1118, 1993.
Wright, T. L., M. J. Grolier, D. A. Swanson, Chemical variation
related to the stratigraphy of the Columbia River Basalt,
Geol. Soc. Am. Bull., 84, 371-386, 1973.
Wright, T. L., M. T. Mangan, and D. A. Swanson, Chemical
data for flows and feeder dikes of the Yakima Basalt

Basalt Province, Spec. Pap. 239, edited by S. P. Reidel, and P.
R. Hooper, pp. 1-20, Geological Societyof America, 1989.
Turner, S., and C. Hawkesworth, The nature of the sub-
continental mantle: Constraints from the major element
composition of Continental Flood Basalts, Geology, 120, 295-
314, 1995.
Walker, G. P. L., Compound and simple lava flows and flood
basalts, Bull. Volcanol., 35,579-590, 1971.
Walker, G. P. L., Pipe vesicles in Hawaiian basaltic lavas:
Their origin and potential as a paleoslope indicator, Geology,
15, 84-89, 1987.
Subgroup, Columbia River Group, Washington, Oregon,
Idaho, and their bearing on a petrogeneticmodel, U.S.Geol.
Surv. Bull., 1821, 71 pp, 1989.
Laszlo Keszthelyi, Stephen Self, Thorvaldur Thordarson,
Departmentof Geology and Geophysics,School of Ocean and
Earth Science and Technology, University of Hawaii, 2525
Correa Rd., Honolulu, HI 96822
 Large IgneousProvinces' A PlanetaryPerspective
JamesW. Head, III
Departmentof GeologicalSciences,
Brown University,Providence,RhodeIsland

Millard F. Coffin

Institutefor Geophysics,
The Universityof Texasat Austin,Austin,Texas

Large igneousprovinces(LIPs) are commonon the Moon, Mars and Venus,

and their presence,characteristics, and geologicand temporalsettingsoffer a
potentiallyimportantperspectivefor interpretingLIPs on Earth. On the Moon,
shallowmagmareservoirsand large shieldvolcanoesare unknown. The rela-
tively low-density,thick anorthositiccrustcreatesa densitytrap for risingbasal-
tic magmaswhich are thoughtto collect in reservoirsat the baseof the crust;
reservoiroverpressurization causesdikesto propagateto the surface. Dikes suf-
ficiently large to reach the surfaceare likely to result in large-volume,high-
effusion-rateeruptions;single eruptivephasesare predictedtheoreticallyand
observed
inthemariatobeseveral
hundred
toover103km3.OnMars,massive
shield volcanoeshave formed on the stablelithosphereover hot spotslasting
over a billion years;shieldheightsare up to 25 km abovethe adjacentplains.
Volumes
ofsingle
edifices
areoftheorder
of 1.5x 106km3 comparable
tothe
total volumesof many basaltprovinceson Earth. The impactcrateringrecordon
Venus suggeststhat Venus underwentrapid and massiveplanet-widevolcanic
resurfacingabout 300 m.y. ago, an event possiblyrelated to the overturn of a
depletedmantle layer resultingfrom the vertical accretionof a basalticcrust.
This hypothesizedevent could be the equivalentof a planet-wideLIP and un-
derlinesthe possibilityof episodicand catastrophicLIPs throughoutplanetary
history,resemblingmantle overturneventsproposedfor Earth. The planetary
record,in concertwith the detailedexaminationof exampleson Earth, can be of
use in developingand testingmodelsfor the emplacementof LIPs, and may help
to distinguishplate tectonic influencesfrom those linked to deeper interior
(mantle and core) processes.
INTRODUCTION AND BASIC CHARACTERISTICS minous emplacementof predominantlymafic extmsive
OF TERRESTRIAL LARGE IGNEOUS PROVINCES and intrusiverock whoseoriginslie in processesotherthan
"normal" seafloor spreading[e.g., Coffin and Eldholm,
Recently, attention has been drawn to large igneous 1992]. Large igneousprovincesare characterizedby tran-
provinceson Earth, which are definedas regionsof volu- sientlarge-scaleintrusiveand extmsiveactivity,including
continentalflood basalt(CFB) provinces(e.g., the Deccan
Large IgneousProvinces:Continental,Oceanic,andPlanetary Traps), volcanic passive margins (e.g., the Voring
Flood Volcanism Margin), oceanicplateaus(e.g., the OntongJavaPlateau),
GeophysicalMonograph100 ocean basin flood basalts (e.g., the Caribbean Flood
Copyright1997 by the AmericanGeophysicalUnion Basalts), and large seamount chains (e.g., Hawaiian-
412 LARGE IGNEOUS PROVINCES: A PLANETARY
Emperor) [Coffin and Eldholm, 1992]. Commonly
analyzedseparatelyin the past,recentstudies [e.g., Coffin
and Eldholm, 1992, 1993] have shown that there are
important temporal, spatial, and compositional
relationshipsamong terrestrial large igneousprovinces,
informally referredto as LIPs.
These studies,and numerousothersthat documentindi-
vidual occurrences(seereferencesin this volumeandthose
of Coffin and Eldholm [1994]), showthat the genesisand
evolution of LIPs are closely linked to mantle dynamics,
that some LIPs represent major global events (large
volumes of lava and associated intrusions are
producedin short episodes,which had potentiallymajor
effectson the global environment),and that emplacement
of some LIPs may be related to changesin rate and
direction of plate motion. Their formation may be
episodic, but modification and destruction of older
examples,and sedimentationand inaccessibilityof others,
makesthis difficult to determine. Althoughseveralmodels
have been proposed for the emplacement of LIPs
(primarily associated with mantleplumes)[e.g., Whiteand
McKenzie, 1989; 1995; Griffiths and Campbell, 1991;
Larson, 1991a, b], these models are not yet well
constrainedby observations.At presentonly a limited,but
growing (see articles in this volume), amount of
quantitative data is available to assessassociatedmantle
and crustalprocesses;to determineLIP dimensions,dura-
tions,ratesof emplacement,crustalstructure,andrelation-
ship to tectonism;and to predictenvironmentaleffectsof
LIP formation. For example, recent workers [Self et al.,
1996, and this volume] have presentedevidence that
Columbia River flood basalt lavas may have been
emplacedmore graduallyas inflatingpahoehoesheetflows
forming very extensiveflow fields ratherthan singlevery
high-effusion-rateeruptions.
THE PERSPECTIVE FROM THE PLANETARY
GEOLOGICAL RECORD
Large igneousprovincesare alsocommonon the terres-
trial planetarybodies(Figure 1) otherthan the Earth [e.g.,
Basaltic VolcanismStudy Project, 1981; Taylor, 1994],
and their presence, characteristics,and geologic and
temporalsettingsoffer a potentiallyimportantperspective
for understanding LIPs on Earth. For example,unlike the
Earth, the majority of which is coveredby water and thus
virtually unknown at high spatialresolution,global imag-
ing coverageexistsfor the solid surfacesof the Moon and
Mars, andthe Magellan projectimagedover 98% of Venus
at-200 m resolution. In addition,exposureand preserva-
tion are excellentdue primarily to fewer erosionalagents,
minimal erosionalrates, and relatively stablelithospheres.
PERSPECTIVE
Stablelithospheresalsomeanthat longertime intervalsare
availablefor study(Figure 1). The age of over one-halfof
the Earth'ssurface(the oceancrust)is lessthan 5% of the
age of the planet;the majority of the surfaceof the Moon
and Mars, however, datesto the first half of solar system
history. Terrestrial planetarybodies, by virtue of their
number, offer multiple examplesfor study. Thus, LIPs
might be studied in different places on one planet and
among several planets. Similarly, the terrestrialplanets
provide an opportunityto assesshow different environ-
mental conditions (e.g., different crustal and thermal
often structure)might influence the formation and effects of
LIPs. Furthermore,the segmented,laterally moving, and
constantlyrenewing terrestriallithosphereboth insulates
and obscuresthe view of many mantle convectionproc-
essesand, indeed,is an activeinfluenceon theseprocesses.
One-plate planets [Solomon, 1977] such as the Moon,
Mars, Mercury, and Venus can illustrate the long-term
influences of mantle plumes and their variations under
different thermal conditions in space and time. The
multiple, well-exposedLIPs on the planetscan alsohelp to
reveal the relation of plumes to tectonic structure. For
example, Venus has tens of thousandsof kilometers of
exposedrift zones [Senskeet al., 1992] which display a
wide variety of igneous centers [Senskeet al., 1992;
Magee Robertsand Head, 1993], manyof whichare LIPs.
The planetaryrecord can be instructivein terms of the
chronologyand episodicityof large igneousevents and
provinces. The extendedhistoricalrecord available for
study(e.g., the first half of solarsystemhistory;Figure 1)
permitsan assessment of changesin the styleof LIPs with
time (potentiallylinked to thermalevolution,for example),
andthe frequencyat any giventime. Althoughradiometric
datesfrom the planetaryrecordare sparse,cluesfrom well-
exposeddepositmorphologycan sometimeseven be used
to estimatesingle-eventduration[Head and Wilson,1980].
Finally, the planetary record can offer a temporally
completeperspectiveon many processesassociatedwith
LIPs. For example,lateralplate movementon Earth in the
caseof the Hawaiian-Emperorseamountchain (and other
hotspot-related chains)helpsto illustratemany stagesin
hotspotdevelopmentby spreadingthe signatureout into a
seriesof volcanic edifices;this same process,however,
destroysthe signatureof the initial plumewhichpresuma-
bly hasbeensubductedunderKamchatka.On the planets,
particularlyVenus, the start-to-finishprocessesof mantle
plumescan be studied(e.g., the relationof thermaluplift,
tectonics,and volcanism in a single example and from
examplesin different stagesof formation)[e.g., Stofanet
al., 1992; Keddie and Head, 1994a] and comparedto
Earth. In summary,the planetaryrecord,in concertwith
detailedexaminationof exampleson Earth, shouldhelp
 HEAD AND COFFIN 413

]a POSITION IN
SOLAR SYSTEM developand testmodelsfor the emplacementof LIPs, and
5000
(10• km) to distinguishplatetectonicinfluencesfrom thoselinkedto
SIZE
SUN deepermantle and coreprocesses.
(radius, km) Terrestrialplanetarybodiesshowa wide variety of char-
MERCURY
acteristics: e.g., size, density,gravity, presence/absence
0
and natureof atmospheres, thermalevolution,and starting
VENUS conditions(Figure 1). Obviously, all comparisonsto the
100-

5
terrestrial record must keep these variations and
EARTH/MOON differences in mind, as well as the positive aspectsof
DENSflY comparativeplanetologydescribedabove. The planetary
(g/cm3)
record is not, of course, a panacea. In many cases
200-
available information for specific aspectsof different
MARS
planets(e.g., the detailedcrustalthicknessand structureon
CO2 Mars and Venus) is limited, resultingin someuncertainties
'øø
1 involving correlations,relationships,and causal factors.
Nonetheless, the information provided by specific
ATMOSPHERIC
11
300 -
PRESSURE
(BARS) examples and the perspectiveprovided by considering
different conditionson differentplanetsshouldcontribute
C02
to our understandingof the formation and evolution of
lb LIPs. The purposeof this paper is to presenta range of
specificexamplesand to explore the potential application
of these examplesto current problems in understanding
....MOON
AND

MERCURYf LIPs on Earth.

MARS

..... •-•
PLANETARY EXAMPLES

We proceedin orderof increasingplanetarysize(Figure

1), first describingthe general crustal, lithospheric,tec-
tonic, and temporal setting of basaltic volcanism on the
I I I
Moon, Mercury, Mars, and Venus, and then discussing
ORIGIN OF

specificexamplesof large-volumebasalticmagmatismon
3 2 I PRESENT
PLAN ETS
TIME BEFORE PRESENT(BILLIONSOF YEARS)

eachplanetarybody and the potentialrelevanceto LIPs on

Figure 1. Characteristicsand history of terrestrial planetary
Earth.
bodies. (a) Basic characteristicsof the terrestrial(silicate-rich,
relatively high-density) planetary bodies of the inner solar The Moon
system. Phasesof planetaryevolutionand many of the present
geologic characteristicsof these bodies may derive from these The Moon's diameter is about one-quarterthat of the
factors [e.g., Head and Solomon,1981]. Although there are no Earth. The Moon is of lower density,has not retainedan
direct trendsin size or presenceand natureof an atmosphereas a
atmosphere,is characterizedby vertical tectonics of an
function of distancefrom the Sun, the general density charac-
teristicshave been interpretedin terms of temperature/pressure unsegmentedlithosphere(not lateral plate tectonics),and
gradientsin the collapsingsolarnebula.(b) Geologichistoriesof now has a very thick lithosphere.Most of its geological
the terrestrial planetary bodies. Generalized plot of the surfaceactivitytook placein the first half of solarsystem
approximatepercentageof presentlyexposedsurface area that history(Figure 1) [e.g., Head and Solomon,1981]. Infor-
formed at different times in the history of the planet. For mation about the Moon comes from remote observations
example, more than one-half of the Earth is seafloorformed in and surfaceexploration,includingreturnedsamplesand
the last 200 million years, whereasthe surfacearea represented scientificstations[seeHeiken et al., 1991].
by units formed in the first two-thirdsof Earth history is very
BasalticvolcanicdepositsontheMoon consistlargelyof
small. On the Moon, Mercury and Mars the majority of the
lunar maria which cover about 17% of the surface,
surfaces(e.g., the lunar highlandsand maria) formedduringthe
the first third of the historyof the solarsystem. Volcanic activity primarily on the near side (Figure 2a). The total area of
continuedbeyondthis periodon Mars. On Venus,like the Earth, thelunarmaria(6.3 x 106 km2 [Head,1975b;1976])is
the majority of the surface formed relatively recently, but considerablylarger than typical terrestrialLIPs but only
apparently due to processesdifferent than the plate recycling slightlylargerthanthe areaof the OntongJavaLIP (Figure
typical of the Earth. 3). The lunar maria were emplacedover about
414 LARGE IGNEOUS PROVINCES' A PLANETARY PERSPECTIVE

AREA VOLUME EMPLACEMENT RATE

1000..•Lunar
Sinuous
Rilles
7•Lunar
6

5 Maria 60
18
16
1

•
..
•.,
•12
E {Lunar
Average
-
PeakFlux

before 8
' 2( Mens 6

0
o,O,, : - o,O,, - o,O,,
uJ z • z • z

Figure2. TheMoon.(a) Globaldistribution ofthelunarmaria. Maredeposits youngerthanabout 3.9Gaareshown

in black;buriedmareplains(possible cryptomaria) inferredfromremotesensing dataandclustering of dark-halo
cratersarestippled.Selected multi-ringedbasinsareO = Orientale; H = Humorum; SZ = Schiller-Zucchius; I=
Imbrium; N = Nectaris; C = Crisium;B - Balmer; S= Smythii; LF = Lomonosov-Fleming; AK = A1-Khwarizmi-King;
TS-- Tsiolkovsky-Stark; M -- Milne;MS = Moscoviense [fromHeadandWilson, 1992a,modified fromSchultz and
Spudis, 1979,1983].(b) Blockdiagram illustrating
therelationship between risingmantlediapirs,anorthositic
crust,
theneutral buoyancy zone,andsurface emplacement of magma bydikes[afterHeadandWilson, 1992].Diapirs (1)
reaching thebaseof thickcrust(2) onthefarside andpartsofthenearside stallandpropagate dikesintothecrust, most
of whichsolidifyanddo notreachthesurface, exceptoccasionally in deeper craters
(3). Propagation of dikesfrom
diapirsstalled underthincrust(1) causes outflow of lavaandfillingofbasins. Withcontinuedthermal evolution,the
lithosphere thickens anda rheologicalbarrierto risingdiapirs forms(4);onlydikeswithextreme overpressurereach
thesurface (5, 6) forming veryhigh-effusion-rate
lavasandsinuous rilles.(c) Rimae Prinzsinuous rilles(27øN,317ø)
interpreted to havebeenformed byveryhigheffusion ratesandthermal erosion (LunarOrbiterLO V M191). Width
of imageis65km. (d) Oblique viewfromlunarorbitof extensive lavaflowsinMareImbrium.Lateinthevolcanic
fillingof theImbrium impactbasin,lavaflowserupted fromthesouthwestern edgeof thebasin(tothelowerleft,but
outside of thisimage)andflowedforupto 1200.In thisimage,flowsupto about20 km in widthandabout30 m in
heightcanbeseenextending frommiddleleft(thedirection of thesource) to upperright(themiddleof thebasin).
Ridgesextending fromupperleftto lowerrightaretectonic features formedlargelyafterlavaflowemplacement.
MonsLa Hire,themountain
in thecenter,isabout20 km in width. Apollomapping
camera
frameAS15-1555.

billionyears,largelyin the firsthalf of solarsystemhistory Volcanic features manifesting large-volumeeruptions

[Wilhelms, 1987], but the total volume was relatively include the individual maria themselves, extensive flow
small,about1 x 107 km3 [Head,1975b].Thisvaluefor fronts, some stretchingfor distancesof over 1200 km
the total planetis comparableto the volumeof the Deccan [Schaber,1973], volcaniccomplexesthat mightsignalthe
flood basaltdepositsalone,but considerably lessthanthe locationof hotspots,and sinuousrilles, which have been
total presentvolumeof the terrestrialoceaniccrust,about attributedto high-effusion-rate eruptionsinvolvingthermal
1.7x l09km3(Figure 3). Theaverage lunarglobal magma erosion of the substrate. Interestingly,no large shield
fluxwaslow,about10-2km3/a,evenat peakperiods of volcanoes,suchas thoseseenon the Earth (e.g., Hawaii),
mare emplacement (in the ImbrianPeriod,3.8-3.2 Ga). Mars (e.g., OlympusMons), or Venus(e.g., SapasMons),
Thisaverageglobalflux is comparable to the presentlocal are observedon the Moon; large caldera-likefeaturesare
outputrates for such individualterrestrialvolcanoesas alsoextremelyrare [Head and Wilson, 1991].
Kilaueaor Vesuvius.Outputratesfor individualeruptions The lunar maria are of diversesizesand shapes[Head,
on the Moon were occasionallyextremelyhigh; several 1975a], and individualmare occurrences might be thought
individualeruptionsassociatedwith sinuousrilles may of as equivalent to some terrestrialLIPs (Figure 3),
haveemplaced morethan103km3of lavainabout a year particularlythosemaria thattend to be concentratedwithin
[Hulme, 1973], a single event that would representthe largeimpactbasinsof variousstatesof preservation [Head,
equivalentof about70,000yearsof theaverageflux! 1975a, b; 1976]. Indeed,Altet al. [1988] proposed
 HEAD AND COFFIN 415

6O' 75' 75' 6O' FARSIDE NEARSIDE

90' • 180' 225'

. ,

-60' -75' -75' .60'

3a

3d
3c

Figure
3. Terrestrial
large
igneous
province
areas,
volumes,
and
crustal
emplacement
rates
(averaged
over
0.5-4.0
m.y.
emplacement
duration)
relative
tosome
planetary
examples.
Modified
from
Coffin
and Eldholm
[1994].
COLR
= Columbia
River
Basalts;
DECC = Deccan
Traps;
NAVP= North
Atlantic
volcanic
province;
KERG- Kerguelen
Plateau;
KERGLIP
=Kerguelen
Plateau
large
igneous
province;
ONTO =OntongJava
Plateau;
ONTOLIP= Ontong
Java
Plateau
large
igneous
province.
Forvolume
and
emplacement
rate,
black-only
oceanic
plateau
columns
indicate
off-ridge
emplacement
andblack
plus
white
columns
depict
on-ridge
emplacement.

flood basaltsthat form within plates,with no apparent mantle[Altet al., 1988]. Doesthelunarrecord support
tectoniccause,aretheterrestrialequivalentsof thelunar thismodel?Although earlytheoriessuggesteda causal
maria. In their model,an impactcrateron Earthlarge relationship between lunarimpactbasinformation and
enough tocause pressure-release
meltingwould bequickly basaltic
mare filling,the resultsof the Apolloand Luna
floodedto forma lavalake(equivalent to thelunarmaria) explorationprograms andmodels of basinformationand
andtheseevents,in turn,wouldinitiatehotspots, which evolution [Solomon etal., 1982;Brattetal., 1985]showed
woulddevelop
intopersistent cellswithinthe thatgeneration
low-pressure ofbasalts
viaimpacts
wasunlikdy
and
416 LARGE IGNEOUS PROVINCES: A PLANETARY
impactbasinformationand filling by marebasaltare sepa-
rated in time. In the case of the 900-km-diameter
Orientale impact basin, vast quantitiesof substratewere
impact-meltedby the basin-formingevent to produce a
sheet of high-albedoplains lining the basin interior and
floor,andareestimated
tohavea volume
of-200,000kms
[Head, 1974]. This unit has a compositionalaffinity to the
non-maretargetrocks [Head et al., 1993] and is distinctly
differentin compositionand age from the adjacentbasaltic
maria deposits,which spanan intervalof severalhundred
million years [Greeley et al., 1993]. In most other mare
basins,the vast majority of the exposedvolcanic plains
were emplaced over several hundred million years
following the impact event [Basaltic VolcanismStudy
Project, 1981]. There is no evidencefor the productionof
basin-sized lunar basaltic "lava lakes which crystallized
from the surfacedown" [Alt et al., 1988]. The stratigraphy
of lunar maria infilling documentsboth the long and
sequential development of extrusive events, and the
differencein age betweenthe basin-formingevent and its
basalticlava filling. Localizationof the maria in the basins
apparently was due to passive variations in crustal
thicknessand ponding in topographiclows [Head and
Wilson,1992a],processes discussed furtherbelow.
Althoughthe equivalenceof an impactorigin of a basin
and its fill on the Moon and LIPs on Earthproposedby Alt
et al. [1988] is not supportedby evidencefrom the Moon,
impactson Earth could potentiallyinitiatevolcanism. The
smallsize of the Moon (and correspondingly very different
pressure gradient), its thicker crust, and its variable
lithosphericthicknesscould all inhibit meltingrelativeto a
comparableevent on Earth. Convincingargumentshave
been put forth to indicate that impact-initiatedvolcanism
was not a factor in the large (-200 km diameter)Sudbury
basin formed in continentalcrust on Earth [Grieve et al.,
1991a]. Similar-size impactsinto thin crust and litho-
spheretypical of a youngoceanicfloor settingcouldcon-
ceivably producepressure-release melting and associated
volcanism[Rogers,1982]. Craterstypicallyformedduring
the time of emplacementof most well-documentedLIPs
(e.g., the last 250 m.y. [Co[fin and Eldholm, 1994]) are
characterizedby relatively small size, shallow depthsof
excavation,and lack of significantlava fill [Grieve et al.,
1991b; but see also Oberbecket al., 1993]. Large-scale
rifting and deep-sourceplume volcanismare more likely
candidatesfor LIP formation and evolution during this
time period. In early Earth history,however,very large
impactsinto ocean crust and thin lithospheremay have
been sitesof extensivevolcanismcausedby mantle uplift
and decompressionalmelting [e.g., Grieve, 1980; Frey,
1980; Grieve and Parmentier, 1984].
PERSPECTIVE
Otherlargevolcanicaccumulations
on the Moon include
the extensive lava flow fronts of Mare Imbrium which
were emplacedat least a billion yearsafter the formation
of the impact basin. These occur in three phaseswhich
extend1200, 600, and 400 km from the southwestern edge
of the basin into its interior. The three flow units have a
totalvolume
of>4 x 104klTl
3,andveryhigheffusion
rates
are implied by their lengthsand volumes;effusionrates
and flow volumes are comparable to some of those
reported for the Columbia River flood basalts[Schaber,
1973; Tolan et al., 1989], althoughfractal analysesraise
the possibility that the Imbrium flows could have been
emplacedas numerousthin pahoehoeflows [Bruno et al.,
1992]. The fact that theseunits are someof the youngest
on the Moon suggeststhat other more degradedflows
filling the earlierlunarmariamay alsohavebeenemplaced
similarly. Examinationof isolatedmare basaltponds in
the highlandsfringingthe continuous mariahasshownthat
typicalvolumes
rangefrom 100to 1200km3, values
similar to those of terrestrialflood basalt eruption units
[Yingstand Head, 1994; 1995; Tolan et al., 1989]. Thus,
many of the individual eruptionsthat make up the maria
may be equivalentto unitswithin flood basaltsand LIPs on
Earth,but the eruptionfrequencyseemsto havebeenmuch
less;the lunar maria were emplacedover many hundreds
of millions of years,ratherthan a few million yearsaswas
apparentlythe casein mostterrestrialexamples.
Another unusual characteristic of lunar maria relative to
LIPs on Earth is sinuousrilles (Figure 2c), which are
meanderingchannelspreferentiallylocatedalongthe edges
of the maria [Schubert et al., 1970]. They range up to
about 3 km wide and from a few kilometers to more than
300 km long. Sinuousrilles are generallyan order of
magnitudelargerand oftenmuchmoresinuousthanterres-
trial lava channels. Many characteristics
of lunar sinuous
rilles unexplainedby simple lava channel,tube or other
models[e.g., Oberbecket al., 1969, Greeley,1971;Spudis
et al., 1987] can be accountedfor by thermal erosion
[Hulme, 1973, 1982; Carr, 1974]. The length,width, and
depthof large sinuousrilles and the natureof their source
regionsprovide importantinformationon eruptioncondi-
tions. For a 50-km-long rille in the Marius Hills, Hulme
[1973]calculated
aneffusion
rateof4 x 104m3/s,
anerup-
tion durationof aboutone year, and a total magmavolume
of about1200km3. The sizesof source
depressions
of
sinuousrilles provide independentevidencefor extremely
high-effusion-rateeruptionsof long duration[Wilsonand
Head, 1980; Head and Wilson, 1980]. On the basis of
thesestudies,key factorsin the formationof sinuousrilles
by thermal erosionare (1) turbulentflow, requiringhigh
effusionratesandaidedby low yield strengthand(2)
 HEAD AND COFFIN 417

tained flow (implying very long-durationeruptionsand thositichighlandcrust,the latter derivedprimarily from

thus very high eruptionvolumes)to causethe continued global-scalemelting associatedwith planetaryaccretion.
downcuttingof the rille to the observeddepths. Thus, The low-densityhighlandcrustprovideda densitybarrier
eruptionsthat causedmany of the large sinuousrilles on [Solomon,1975] to ascendingmantle plumesand basaltic
the Moon were apparentlycharacterized by rapid effusion melts. In this view, rising diapirs and magma bodies
of low-yield-strengthlavasfor prolongedperiods,produc- tended to collect at the base of the 60-80 km thick crust
ing flows of extremelyhigh volumes(in the range 300- (Figure 2b). Following sufficient overpressurization of
1200km3),comparable to thosein terrestrialfloodbasalt sourceregionsby partial melting or arrival of additional
provinces (e.g.,the-1375 km3 RozaMemberof the material into the reservoir, individual dikes propagated
Columbia River Basalt [Martin, 1989]). In contrast, toward the surface. Thus, the thick highlandcrustcreated
typical eruptionvolumesfor shield-relatedflows on Earth a deep zone of neutral buoyancyfor rising magma that
are much lessthan a cubic kilometer [Petersonand Moore, could only be overcome by overpressurizationevents
1987], with the largest historic lava flow (Laki) being which causeddikesto propagateto the surface.
about15km3[Jonsson, 1983;Thordarson andSelf,1993 ]. In this model, whether intrusion or eruption occurred
Several mare-related areas show unusual concentrations was determined by variations in overpressurizationand
of volcanic featureson the Moon [Guest, 1971; Whitford- crustal thickness. Low levels of overpressurization
Stark and Head, 1977]. Two of the most significantof resulted in intrusion into the lower crust, forming dikes
these (Figure 3) aretheMarius Hillsarea(35,000km2), which cooledand solidified. Dikes characterizedby suffi-
which displays20 sinuousrilles and over 100 domesand cientoverpressurization to approachthe surfacecouldhave
cones, and the Aristarchus Plateau/Rima Prinz region severalfates. Overpressurization eventslarge enoughto
(40,000km2) whichis dominated by 36 sinuous rilles propagate dikes to the surface to cause eruptions are
(Figure 2c). The high concentration of sinuousrilles sug- predictedto involve very large volumesof magma [Head
geststhat thesecomplexesare the sitesof multiplehigh- and Wilson, 1992a], comparableto thoseassociatedwith
effusion-rate,high-volumeeruptionsand that thesecenters many observedlava flows, such as the flows extending
may be the surfacemanifestationof hotspots[Head and hundreds of kilometers into Mare Imbrium [Schaber,
Wilson, 1992a] and thus possibleanalogsto terrestrial 1973] and those associatedwith sinuousrilles. Intrusion
LIPs. The thick crust (about 60-80 km) and lithosphere close enoughto the surfaceto producea distinctivenear-
(in excessof the thicknessof the crust)characteristicof the surface stress field often resulted in the production of
Moon (and thusthe greaterdepthsof magmasources)may linear graben-likefeaturesalongthe strikeof the dike and
make thesecandidatehotspotslessrecognizableand more smallassociated effusionsanderuptions.In the caseof the
analogousto continentalvolcanicprovinces. In addition, linear graben Rima Parry V, small spatter cones are
lava flow depositson the Moon are much more widely aligned along the central part of the graben [Head and
dispersedfrom their sources. Wilson, 1994a]. Dikes propagatingto slightly deeper
In summary,the lunar maria are comparablein scaleto levelsmay not createnear-surfacestressfields sufficientto
some terrestrialLIPs (Figure 3) but on the basisof avail- form graben,but subsequent degassingmay form chainsof
able data appearto have beenemplacedover much longer pit cratersover the siteof the dike.
periods oftime(e.g.,108to 109years rather than106to 10? The model predictsthatthe relationshipbetweenthe size
years). Many individualeruptions,however,appearto be of the magma source and highland crustal thicknesswas
similar in volume and eruption rates to those in flood suchthat dikespropagatedto the near-surfaceand surface
basaltprovinces[Tolan et al., 1989]. Little evidenceexists relatively infrequently(Figure 2b). Thus, most dikes had
for shallow magma reservoirsand repeatedsmall-volume sufficienttime to cool beforethe next dike was emplaced.
eruptionsthat would build up large shieldvolcanoes.The Frequentemplacementof dikesto createa shallowmagma
observedcharacteristics seemto call for large batchesof reservoir was very difficult on the Moon. The lack of
magmaeruptedover shortperiodsof time from relatively Hawaii-like shield volcanoesand the paucity of caldera-
deepsourcesbut separatedin time by significantintervals. like features are thus attributed to the difficulty in
How can these characteristics be accounted for in terms of producing shallow magma reservoirs which result in
the nature of the source regions and the modes of emplacementof many individual flows, edifice-forming
emplacement? flows, and associatedcalderas[Head and Wilson, 1991].
One model [Head and Wilson, 1992a] beginswith the In addition, the same lack of multiple, continuousdike
observationsthat the basalticmaria are superposed on the eraplacementevents of sufficientmagnitudeto reach the
ancient,globally continuous,and thick low-densityanor- surfaceover short periods of time meant that the
418 LARGE IGNEOUS PROVINCES: A PLANETARY PERSPECTIVE

maria tendedto be producedfrom relativelylargeeruption conditionsin the innerpart of the condensing solarnebula,
events spaced over very long intervals, in contrastto which favored retention of refractory components[e.g.,
terrestrial LIPs. Goettel, 1988], and to the effectsof a giant impact event
The lunar situationdescribedin this model is analogous strippingoff a low densitycrustand uppermantleafter
in many ways to basalticmagmabodiesinteractingwith core formation [Cameronet al., 1988].
terrestrial continental crust. On Earth, zones of neutral Mercury is poorly exploredin terms of photographic
buoyancy[e.g., Glazner and Usslet, 1988] stallbuoyantly coverage and remote sensing data [Chapman, 1988].
rising basalticmagmabodieswithin the crust. Overpres- Knowledgeof internal structureis meager, althougha
surization events can cause the same features seen on the high-densitycorecomprisingwell over one-halfMercury's
Moon, as exemplified by many of the basaltic volcanic diameter(aboutthe size of the Earth'sMoon) is likely. In
fields in the westernUnited States[e.g., Crumpieret al., addition, prominent albedo variations such as those that
1994], and indeed large-scale flood basalts can be distinguish the lunar maria from the heavily cratered
emplacedthat are comparablein size to the large lunar highlandsare not apparenton Mercury. Smoothplainsare
flows [Tolan et al., 1989]. The low meltingtemperatureof present,but a possiblevolcanic origin cannot readily be
the continentalcrustrelativeto that of the more refractory distinguishedfrom plains producedby pondingof impact
lunar anorthositiccrustmeansthat stalledbasalticmagma ejecta, a process known to occur in the light plains
bodiesin continentalcrustmay causeassociated and large- surroundingimpactbasinson the Moon [Oberbeck,1975;
scale crustal melting, resulting in a geochemicaland Oberbecket al., 1975; Wilhelms,1976]. The stratigraphy
petrologic complexity unknown on the Moon. The and geologic history of Mercury suggest that major
continentalcrustand the lunar highlandsillustratethe role volcanicprovinceswere emplacedin the first third of solar
of large-scale density barriers impeding the creation of systemhistory [Spudisand Guest, 1988], but the details
significant shallow basaltic reservoirs, such as those are insufficientto providea basiccharacterization of such
observedat seafloorspreadingcentersand in largeedifices provinces or an understanding of their mode of
such as Hawaii. Complex shallow reservoirsdo exist in emplacement.If theseplainsare indeedof volcanicorigin,
continental crust, however, where local conditionsof melt their generallack of associatedvolcanic features[Stromet
generationand, unlike on the Moon, sustainedsupplyrates al., 1975; Trask and Strom, 1976] suggestspossibleflood
exist (as in continentmargin subductionzonesand hotspot basaltemplacement.
tracesor rifting environments). In thesecases,composite
volcanoes are common. No known analog of these Mars
features exists on the Moon and Venus, but several
examplesmay be presenton Mars (e.g., HecatesTholus Information about Mars [e.g., Kieffer et at, 1992a]
[Mouginis-Mark et al., 1982; Wilson and Head, 1994; comesfrom Earth-basedobservations,extensivespacecraft
Hodgesand Moore, 1994]). exploration(includingorbitersand landers[e.g., Kieffer et
al., 1992b; Snyder and Moroz, 1992]), and meteorites
Mercury believed to be ejected from Mars by impacts and
transportedto Earth [e.g., Longhi et al., 1992]. Mars is
Mercury remainsone of the most enigmaticand promis- about one-half the diameterand of much lower density
ing planets in the inner solar systemin terms of under- than the Earth, has a thin CO2 atmosphere,and is
standingthe relationshipof its unusualinteriorto its vol- characterizedby vertical tectonics of an unsegmented
canic and magmatic history [Chapman, 1988]. lithosphere(not lateral plate tectonics);most of its major
Information about Mercury comes from the Mariner 10 geological surface activity took place in the first half of
mission and Earth-based observations[see Vilas et al., solar systemhistory,with somevolcanismand significant
1988]. Mercury is about one-third the diameter but eolian activity continuingwell into the last half of solar
approximately the same density as the Earth, has not systemhistory (Figure 1, 4a) [e.g., Head and Solomon,
retained an atmosphere,and is characterizedby vertical 1981; Kieffer et al., 1992b]. The total area of Mars
and somelateraltectonicsof a largely unsegmented covered by volcanic material has been estimatedto be
litho-
sphere(not lateral plate tectonics). Most geologicalsur- about 58%ofthesurface (-0.84x 108km2)[Tanaka etal.,
face activity took place in the first third of solar system 1988], and the total volume of surfaceextrusionto be 2 x
history(Figure 1) [Head and Solomon,1981; Vilas et al., 108km3 [Greeley, 1987](Figure3). Thecorresponding
1988]. The very high density of Mercury relative to its intrusivevolume is not knownbut is likely to be largerby
sizehasbeen attributedboth to initial temperature-pressure at least a factor of 10, the ratio typical of the
 HEAD AND COFFIN 419

60 ø N

Arcadia
Acidalium
Planitia
Alba Patera % Planitia
Utopia Planitia

•l•ranius
Tholus

Isidis
Lunae •, j
Mons
Planum •,Planitia
/ ""
Pavonis ,.,',,
' Mons -•tJ - ' •30 3OO'•-•' 270•

("Hesperia
'• Planum
Planum
/,e

Argyre
Planitia
.Amphitrites

4a

Figure4. Mars.(a) Generalized topography andgeography.Thedashedlineextending fromupperleft (nearArcadia

Planitia)to lowerright (nearApellinarisPatera)acrossbothhemispheres separatesthe northernlowlandsfrom the
southern highlands;othercloseddashedlinesare impactbasindepressions (e.g.,ArgyrePlainita,HellasPlanitia)or
volcanicprovinces(e.g., HesperiaPlanum;LunaePlanum). Black spotsare shieldvolcanoesof varioussizes;the
concentration of shieldvolcanoesin the left hemisphereis the Tharsisregion(seeFigure4b for enlargement) and in
therighthemisphere is theElysiumregion. (b) Geologicsketchmapof theTharsisregion. Topography is indicated
by contourlines at 5 km intervalswith tick markspointingdownslope.Starsmark summitsof the major shield
volcanoes (Figure4a), OlympusMens andtheTharsisMontes(Arsia,Pavenis,andAscraeus), whichcommonlyreach
elevationsin exceessof 20 km; uppercontoursareomitted. Width of diagramis about4000 km at the equatorandthe
unitsare discussed in the text. Map is from Head and Solomon[ 1981] from dataof l/Viseet al. [ 1979] andScottand
Cart [1978]. (c) Obliqueview of OlympusMens volcano,oneof the largelavashieldsin the Tharsisregion(Figure
4a, b); summitis about25 km abovethebaseof thevolcanoandis characterized by a complexcalderaandtwo nearby
impactcraters. Flows emanatingfrom nearthe summitextenddownthe flanksand oftencascadeover the several-
kilometer-high scarpat the baseof thevolcano.Viking Orbiterphotograph VO 641A52. (d) Stratigraphic sequence
showingcontextandmaineventsin theevolutionof Tharsisin relationto globalprocesses [fromBanerdtet al., 1992].
Locationsof regionsare shownin Figure4a, b. Absoluteagesare from the time-scalemodelsof Hartmann-Tanaka
(H/T) [Hartmann,1978]andNeukum-Wise (N/W) [Neukum and 14/ise,
1976],assummarized by Tanakaet al. [1992].

regionson Earth [e.g.,Crisp, 1984;see Wilsonand Head, however(Figure 4b, c). Martian shieldspossess breadths
1994]. Volcanismhas decreasedover geologictime from of many hundredsof kilometers,and their heights are
broad regional resurfacing to local activity; areal commonlya factor of three greaterthan Hawaii (up to 25
resurfacing
rates
havesteadily
decreased
from-1 km2 a'• km!). Volumes of individualshieldsare gigantic(Figure
to-10'2kin
2 a'1[Tanaka
etal.,1992]. 3). OlympusMons (Figure4c) has a volumeof about2 x
On Mars, in contrastto the Moon, largeshieldvolcanoes 106km3(above
itsbase),
compared
to 1 x 10skm3(above
havebeenemplacedthat resemblethoseon Venusandthe its base)for the islandof Hawaii (which is composedof
Earth in morphology. They exhibit a wide range of rift several
different
shields)
and1.1x 106km3forthewhole
zone development, internal deformation related to Hawaiian-Emperorseamountchain [Barget and Jackson,
lithosphericloadingand flexure,flank and slopefailure, 1974]. Volumes of othersingleedificesare of the orderof
and summit caldera development [Cart, 1973, 1981; 1.5x 106km3,comparable
to extrusive
volumes
estimated
Hodges and Moore, 1994; Wilson and Head, 1994; for the Karoo, Paranti, Deccan and North Atlantic basalt
Crumpleret al., 1996]. Their scalesare quite different, provinceson Earth (Figure 3). Martian caldera
420 LARGE IGNEOUS PROVINCES: A PLANETARY PERSPECTIVE

1
in longer cooling-limitedflows and wider dikes character-
ized by higher effusion rates [Wilson and Head, 1994].
Becausethe lithospherehasbeen stableand hasnot moved
laterally over the majority of martian history, regions of
melting in the mantle (e.g., mantle plumes) concentrate
their effusiveproductsin a singlearea, ratherthan having
them spread out in conveyor-belt-likefashion, as in the
case of the Hawaiian-Emperor seamountchain on the
Pacific Oceanfloor. Thus,melt productsaccretevertically
into huge accumulations[Carr, 1973], loading the litho-
sphereand causingflexure, deformation,and edifice flank
failure.
The extremeheightof martianvolcanoesalso appearsto
be related to lithosphericstructure. Comer et al. [1985]
examined deformational structures surrounding several
Tharsis-regionvolcanoes(shown as large black spots in
-50 ø
70 ø 100 ø 40 ø

4b Figure 4a) to assesslithosphericflexure causedby volcano

loading and to estimatethe thicknessof the elasticlitho-
sphere. They foundthat elasticlithospherethicknesses are
in the range of 20-50 km for regions surroundingthe
majority of the Tharsisshields. The lithosphereappearsto
be at least 150 km thick in the region of Olympus Mons.
Thus, one factor contributingto the large height of the
martian volcanoesis the relativelythick elasticlithosphere
during their formation; the volcanic load and underlying
lithospheredid not subsideat a rate that would limit their
heights. In addition,variationsin lithosphericthicknessin

4d
GLOBAL REGIONAL ABSOLUTE
PROCESSES ACTIVITY AGES (Ga)

(o._• • .• o• .• E
(0 --

EPOCH •- ._ -- '• I-- LU H/T N/W

ß ß ß ß 0 ---- 0 -
Late . . ß
ß

Amazonian .' ß ß .' .

Middle . [ ' o 0.70
-- 0.25
-
Amazonian.' I i
I I 2.5o-- o.7o-
Early:
Amazonian
' I
I II I 3.55
--- 1.80
-
4c
Figure 4. (continue.
d) Late I I
Hesperian
Early
! ßß
3.70
--- 3.10
-
Hesperian ß
are alsomuchlargerthan thosetypical of Earth [Head and
Late i I 3.80
--- 3.50-
Wilson, 1994b; Crumpler et al., 1996; McGuire et al., Noachian I
1996]. How do we accountfor thesedifferences? Middle
4.30 ---- 3.85 --

On Mars, low gravity and low atmosphericpressureat Noachian

the surfaceresultin a crustalbulk densityprofile different Early • 4.50--3.92

-
Noachian
from otherplanets[Wilsonand Head, 1994], which means •! z• __ A •t• _

that magmareservoirsare predictedto be deeperthan on

Earthby a factorof aboutfour. The lower gravityresults Figure 4.
 HEAD AND COFFIN 421

spaceand time can be very importantin the construction shortening.Major volcanicedificesandstructures

and subsequent modificationof volcanicedifices. For the
volcanoesforming the GalfipagosArchipelago,Feighner
and Richards[ 1994] showedthat lithosphericthicknessis
relatedto volcanosize and structureacrossthe archipelago
betweenareasof effectiveelasticlithosphericthicknessof
6 and 12 km. McNutt et al. [1989] demonstratedthat the
thermal and mechanicalstateof the lithosphereapparently
controls the expressionof weak plumes such as the
Marquesas.In a studyrelevantto terrestrialshieldvolca-
noes and to the history of lava emplacementin LIPs,
McGovern and Solomon[ 1993] modeledlithosphericflex-
ure and time-dependentstressand faulting on the Tharsis
volcanoes and demonstratedsufficiently large flexural
stressesin severalexamplesto causefailure by faulting.
Suchstressesin turn couldhave influencedthe subsequent
path of magmaascentand emplacement.
One of the most impressiveglobal-scalefeatureson
dotthe
surfaceof Tharsisand four of these(indicatedby starson
their summitsin Figure4b) exceed25 km in elevation
above the surroundingterrain (Figure 4c). Younger
volcanicplainsunits(pt) surround themajorcentralshield
volcanoes,their most likely source. Tectonicfeaturesare
abundantand the most prominent of these, the Valles
Marineris rift system,extendsseveralthousandkilometers
from central Tharsis toward the east. Its floor is indicated
by unit cf, canyonfloor materials,in Figure4b.
Tharsis rise development involved complex episodic
tectonism and intimately associatedvolcanism on both
local andregionalscales.Early fracturedplains(Noachian
and Hesperian Epochs, Figure 4d) made up mostly of
volcanic rocks eruptedduring the early stagesof Tharsis
activity are cut by the most intensedeformationin Tharsis,
representedby fault systemsthat are radial and concentric
to volcanic centers such as Tharsis Montes and Alba

Mars is the Tharsisregion,a LIP comprising-20% of the Patera. These faults formed during the Noachian and
surfacearea of Mars that dwarfsthosepresentlyknown on Hesperianepochs;concentricridge systemsrepresenting
Earth in size, associatedfeatures,and duration(Figure 3). local shorteningwere formed mainly at distancesgreater
Tharsis, which forms a broad dome or rise about4000 km than 2000 km from the centerof the Tharsisrise, primarily
in diameterrising as much as 10 km above surrounding during the Late Noachian and Early Hesperian Epochs.
terrain, dominatesthe westernhemisphereof Mars (Figure The latest faulting occurred in the Amazonian Epoch
4b). Itsareaof >6.5x 106km2 is largerthanthelargest (Figure 4b, d) primarily in associationwith the active
known terrestrial LIP and totals over one-half the total area volcanic centers mentioned above [Tanaka et al., 1992].
of the lunar maria (Figure 3). The Tharsis rise is The Tharsis Montes (from southto north, Arsia, Pavonis
composedof areally extensivevolcanicplainsspanninga and AscraeusMontes; Figure 4a, b), composedof three
wide rangeof ages;massivesuperposed shieldvolcanoes massiveshieldvolcanoesalignedin a row along the crest
(e.g., >500 km wide and up to 25 km high) are associated of the Tharsisrise, are the primary sourcesfor the volcanic
with tectonic features that include radial fractures and TharsisMontes Formation(largely unit pt in Figure 4b),
grabenextendingbeyondthe rise and perhapsassociated which covers an area of almost 7 x 106 km2 and is
with uplift, and concentricwrinkleridgesindicatingcrustal composed of lobate sheet flows, some of which extend
shortening. The volcanicdepositsclearly associated with almost 1500 km from the source shields [Schaber et al.,
this province,the westernvolcanicassemblage [Tanakaet 1978; Plescia and Saunders, 1979]. Olympus Mons, a
al., 1992],cover
anareaof 1.4x 107km2(Figure
3);more- similar shield volcano to the west of Tharsis Montes, is the
degradeddepositsmay alsobe volcanic. sourcearea for Upper Amazonianlava flows, someof the
On the basisof geologicmappingat a variety of scales youngeston Mars (Figure 4b, c).
[Scott and Tanaka, 1986, and summarizedby Tanaka et Theories to account for the Tharsis rise abound [see
al., 1992] a general stratigraphyand chronology for discussionsby Schubertet al., 1992; Banerdt et al., 1992;
Tharsishas begunto emerge(Figure 4d). In contrastto and Tanaka et al., 1992]. Initial ideas centeredon an area
manyterrestrial LIPswhichformed over105-106 years, of convectiveupwelling producinga very large mantle
these data point to volcanic and tectonicactivity in the plume which generateduplift and volcanism[e.g., Carr,
Tharsis region
spanning 108-109 years.Ancient cratered 1974], an idea supported by calculations of mantle
terrain boundsTharsisto the southand is exposedat high convection under martian conditions in which a limited
elevations within Tharsis, suggestingextensive uplift. number of convectioncells are favored [e.g., Schubertet
Plains units interpretedas volcanic and major shield al., 1990]. The interpretedtopographicuplift, however,
volcanoesdominatethe rest of Tharsis. Undivided plains could not be explainedby thesedynamicprocessesalone.
(pu in Figure 4b) make up the vast majority of surface Isostaticuplift causedby lateralmigrationand intrusionof
units and extendnorth of Tharsis. Ridged plains (pr) are materialthermally erodedfrom the baseof the crustof the
characterizedby many mare-ridgetype featuresindicating northern lowlands was favored by Wise et al.
422 LARGE IGNEOUS PROVINCES: A PLANETARY PERSPECTIVE

whereas Schultz et al. [1982] suggestedpreferential A secondlarge domalareaon Mars, the Elysiumrise, is
concentrationof volcanismalong early impact-basinring about2000 km across(Figure4a). Althoughsmallerthan
structures. Debate also centers on the relative role of Tharsis,it also has high concentrations of volcanismand
volcanismand uplift, with someworkerspreferringmajor tectonism[Greeleyand Guest,1987;Mouginis-Market al.,
uplift and relatively minor volcanism[e.g., Plescia and 1984], including several volcanic formations,and three
Saunders, 1980] and others suggestingthat Tharsis major shieldvolcanoes(Albor Tholus,HecatesTholus,and
resultedfrom an extendedperiod of regionalvolcanismin ElysiumMons). Hall et al. [1986] arguedon the basisof
an areaof thin lithosphere[e.g.,Solomonand Head, 1982]. thermaland mechanicalargumentsthat flexuraluplift pre-
The inability of stressmodels to accountsimply for the ceded or was contemporaneous with the emplacementof
extensiveradial grabensystemshas led workersto accept the majority of the volcanicdeposits.
the idea that more than one mechanismof lithospheric Why does Mars have two prominent, long-lasting,
deformation is required;simpleisostatic
or flexuralloading extremelylarge igneousprovinces?Convectiveplanforms
models do not satisfy all observations. The present in the martianmantlewere modeledusingnumericalsimu-
conundrumis that stressmodels seem to require two lationsof fully three-dimensionalconvectionin a spherical
different events,but the geologicevidencesuggests that shell [e.g., Schubertet al., 1990, 1992]. Thesemodels
the radial grabenformedessentiallysimultaneously [e.g., suggestthat cylindricalplumesare the mostprobableform
Banerdt et al., 1992]. of upwellingin the mantleandthat downwellingoccursin
Finnerty et al. [1988] constructeda quantitativepetro- an interconnected networkof planarsheets;the numberof
logicmodelfor Tharsiswhichwasextendedto a moregen- upwellingplumesis a functionof the geometryof heating.
eral model for the evolutionof Tharsis [Phillips et al., Increasein bottomheatingcausesa decreasein the number
1990]. Using melt partitioningdataandmodelsfor likely of upwellingsand an increasein their intensity,with very
martian mantle compositions,they showedthat extraction substantialbottom heating producing only six plumes.
of basalt melt from the mantle and subsequent crustal Gradual cooling of the planet, and the core in particular,
intrusion and extrusion could have resulted in a net volume means that the planform and style of convectionlikely
increase in the crust-mantle column, producing a changedwith time, with fewer, more vigorousplumesear-
prominenttopographicrise with no net increasein mass. lier and more, smaller plumes later. In addition, the
Much of the supportfor the uplift would come from the temperature-dependence of mantle viscositywill have an
source-regionresiduum,and most of the magmaproduced influenceon plume structureandabundance.Althoughthe
by the requiredmeltingmust end up as intrusionsin the trend in early history might have been toward a small
crustanduppermantle. Althoughconsistent with manyof numberof vigorousplumes,a variable lithosphericthick-
the major characteristicsof Tharsis,thesemodels do not nessand a thickeninglithospherewith time [e.g., Comeret
easilysatisfythe gravity data. al., 1985; Solomonand Head, 1990] might hide the surface
What currentlysupportsthe Tharsisregion,someseveral effectsof all but the mostprominentplumes.
billion yearsafter its initial activity? Gravity datashowan What are some possiblelessonsfor those who study
extremely large free-air anomaly [e.g., Espositoet al., LIPs on Earth? First, it is clear that LIPs can achieve mas-
1992]; simple isostaticcompensationis essentiallyruled sive proportions and form over long periods. Tharsis
out and dynamicsupportby active mantle flow is very covers20% of the surfaceareaof the planetMars andwas
unlikely becauseof the difficulty of maintainingsuch activefor severalbillionsof years. In addition,the martian
large-scaleand consistentmantle flow for severalbillion LIPs confirm that scale and total duration of igneous
years. Many models have been proposed,and the most emplacementcan changeas a functionof time and thermal
likely have the Tharsisrise partially supportedby the evolution (large-scaleplanetary cooling). Early plumes
elasticstrengthof the lithosphere,with additionalsupport might have been lessnumerous,larger, and more vigorous
from the buoyancyof a crustalroot at depthsof about50- due to a larger role of bottomheating. Planetarythermal
100 km [e.g., Banerdtet al., 1992]. The early historyof evolutionwill also influencelithosphericthicknessand the
the Tharsisrise might have involved a transientmode of surface manifestationof plume impingement;thus we
support (e.g., a convective plume, or an upper mantle shouldanticipateconsiderablevariability in LIPs through
density-deficitinducedby thermalor chemicalfactors)and time. The abundanceof large shield volcanoeswithin
a regional crustal thicknessabout 25-30 km in excessof Tharsis,each of which would qualify as a LIP on Earth
that estimated in global-scale models. Subsequent (Figure 3), also suggeststhat individual plumesare likely
reduction and removal of the transient support were within a largerdiffuseupwellingsuchas may have formed
accompaniedby the generalcoolingof the planet,leaving Tharsis as a whole. Given the incompleteness of the
a superisostaticload on a cooling,thickeninglithosphere. terrestrial record, the martian record suggeststhat
 HEAD AND COFFIN 423

singleLIPs on Earth might be only one "tree" in a larger would indicate ongoing crustal spreadingand recycling
"forest"of a megaplume. Finally, the petrogeneticeffects [Solomonand Head, 1982; 1991; Solomonet al., 1992].
of shallowmelting and the resultingresiduummight leave The-80% of the surface area of Venus estimated to be
depletedmantle signaturesthat couldpersistfor hundreds covered
byvolcanic
plains
(-3.68x 108km2 [Headetal.,
of millions to billions of years, even on a planet as 1992]) can be combinedwith an estimateof the average
dynamicas Earth. plains thicknessof about 2.5 km based on stratigraphic
relationships[Head et al., 1996a] to predictthe total vol-
umeofsurface
extrusion
ofabout
9.2x 108km3(Figure
3).
On the basis of impact crater counts on volcanic units,
Venus is approximatelythe samediameterand density volcanism has apparently decreasedover geologic time
as the Earth and is Earth's closestplanetary neighbor from a period of global resurfacing to much less
(Figure 1). These similaritieshave led to frequentcom- voluminous local activity, with average effusion rates
parisonof Venus with the Earth and the idea that Venus changing fromabout5 km3/ato <1 km3/a[Headet al.,
might be a "sibling" or possiblyeven a "twin." Venus 1992].
offers an important test of major ideas about planetary On the Earth, typical basalticmelts are positivelybuoy-
evolutionin terms of the role of planetarysize and initial ant, but can stall at a neutralbuoyancyzone (NBZ) repre-
position in the solar system [Head and Solomon, 1981], senting a near-surface,low-density horizon related to
crustal formation and evolution [Head, 1990b], and weatheringand gas-exsolution porosity[e.g.,Ryan, 1988].
mechanismsof lithosphericheat transfer [Solomon and The resultsof theoreticalmodelingindicatethat very high
Head, 1982]. Although Venus has many similaritieswith atmosphericpressureon Venus reducesvolatile exsolution
the Earth, it also has importantdifferences.It has a thick, and magmafragmentation,servingto inhibit the formation
dense CO2 atmosphere, rotates very slowly, and has of NBZs and shallowmagmareservoirs[Head and Wilson,
essentiallyno magneticfield. The high surfacepressureis 1986]. For a rangeof commonterrestrialmagmavolatile
approximatelycomparableto that on the oceanfloor below contents (<0.5 wt% H20 , <0.35 wt% CO2), magma
abouta kilometer depth,and averagesurfacetemperatures ascending and eruptingnear or belowthe meanplanetary
are around475øC, precludingthe presenceof liquid water radius(MPR) on Venusshouldnot stallto produceshallow
andresultingin the preservationof landformsin their near- magma reservoirs. In this case, magma should ascend
pristine state. Information on the nature of Venus has directlyto the surface;sucheruptionsshouldbe character-
come from Earth-based observationsand a variety of ized by relatively high total volumesand effusion rates,
planetary probes, including flybys, orbiters, balloons, comparable to those observed in terrestrial flood-basalt
atmosphericprobesand landers[e.g., Hunten et al., 1983; provinces(Figure 3) [Head and Wilson,1992b].
Cruikshank, 1983]. Data from the recent Magellan Becauseatmosphericpressurechangesconsiderably with
missionprovidedhigh resolutionradarimagecoveragefor elevationon Venus,the samerangeof volatilecontentsre-
almost the whole planet and altimetry and gravity data sultsin the productionof NBZs and magma reservoirsat
[Saunderset al., 1992]; this information, together with elevationswell aboveMPR. For the samerangeof volatile
datafrom previousUS and Sovietmissions,hasresultedin contentsat higher elevations(about 2 km above MPR),
a more comprehensiveview of the geologyand geophysics abouthalf of the casestreatedby Head and Wilson[ 1992b]
of Venus (Figure 5a). resultin direct ascentof magmato the surfaceand half in
Approximately80% of the surfaceof Venus is madeup the productionof NBZs. In general,NBZs and shallow
of volcanicplainsand a wide varietyof volcaniclandforms magmareservoirson Venus are predictedto appearas gas
(Figure 5b-e), probably largely basaltic in composition content increasesand, becauseof the high atmospheric
[Head et al., 1992]. Unlike on the Earth, volcanic land- pressure,to be nominally shallower on Venus than on
forms are not distributedalong elongatedplate boundaries Earth. The shallowestdepthsfor NBZs are about1 km and
and hotspottraces;rather,they are broadly distributedover depthsincreaseslowly with increasingCO2 contentand
the whole planetand alsoclusteredin a largeregion(Beta- rapidly with increasingH20 content. For a fixed volatile
Atla-Themis) making up about 20% of the surface content,NBZs become deeperwith increasingelevation.
[Crumpier et al., 1993]. Although Venus has folded Over the range of elevations(-1 to +4.5 km) treatedby
mountainbelts [Crumpier et al., 1986], global rift zones Head and Wilson[ 1992b], depthsdiffer by a factorof 2-4,
[Senskeet al., 1992], and featuresthat resemble Earth's which is about the same factor as that induced by
convergentplate boundaries[Head, 1990a; McKenzie et variationsin CO2. NBZ reservoirscanbecomedeeperthan
al., 1992], Magellanrevealedno evidencefor the extensive reservoirson Earth producedwith similar volatile contents
globalplate-tectonicboundariesand crustalstructuresthat if common terrestrial volatile contents are exceeded. To
424 LARGE IGNEOUS PROVINCES: A PLANETARY PERSPECTIVE

;':
.... :'""'"
'..... -.........
'.,,;"--:
;?:*.---.
L NT.:'
'•'
..........
•--,.
.........
'•:'; ...................
"*'•!
x:• :"'.':':L..-::;.'*.*
I TI..'A
.: .::..•:.,i•½;?.•.,./:.....:
:
. ...:........::...:.•..
:•....
½:::--:
ß •;..-:..
..:-•:, .:,.:•

....

'"
.:.- .
.:-::.--.:'::::;•;
......
%..
........:.
:.
....%:;"gi..'i"
% .......
.....
'" . .:::....
. ½:,, •:..:

....
'..... .:. •. . ../: :•..::
".'EiS't':'-",..:,......
.....
::•.::..½•,.:..::..:........
,..•..,•
..... •;•:;•
- '"
........... •;•;?

•::i• '• s---c•,•..-½....::::..•:•::---::•

-•
.....•.... •. ,;..:•
,-*-••-- ..•.•'•:•:::•:.•
;:::'...•:.:'
•;• .:5:•::
......
:": ........
•..........
' '"",:•:
,•,•.:::":*•?';:•*
....':•X•,,:-
•:"'
......
'.:a•;•":•
•::"
,:•.,::::;-.....
?:..:...::;:..:-.
.....; ....
.,..
.... ........

,...:•: .'%$
......
........
•....::.:..••;:;:*:,::,
..•;*;:-...•,
..$:,.:::
':$:.½-:':•:•.•-•

1 80 240 300 0 6 0 1 20 1 80

Figure 5. Venus. (a) Magellan global synthetic-aperture radar-imagemosaic of Venus and selectedgeographic
features. The bright areashave generallyhigh radar backscatter(rough at centimeterscales;rift zones,deformed
highlands)and the dark areashave low backscatterand relativelysmooth(volcanicplainsand lowlands). Cylindrical
projection. One degreeof longitudeis aboutequivalentto 100 km at the equator. (b) A portionof Baltis Vallis, an
exampleof a sinuouschannelon Venus (width of imageis about 150 km). Baltis Vallis, about2-3 km wide, startsat
about 45øN, 186ø and winds for about 6800 km to the vicinity of 12øN, 168ø (Figure 5a). Dark areasare volcanic
plainsand brighterlinearpartsare scarpsand surfaceroughness at centimeter-meter scale. The network-likepatternof
ridgesis causedby post-emplacement subsidence anddeformationof lavas. Portionof C1-MIDR 45N159. North is at
the top. (c) A large lava flow field on the flanks of Sif Mons, locatedjust to the southof this image in Eistla Regio
(Figure 5a). Flows, appearingbright in the radar image,emanatefrom the summitand streamdown the flanks for
hundredsof kilometers,oftenbeingtrappedin narrowlineargrabenabout1-2 km wide. Width of imageis about 125
km. Portion of C1-MIDR 30N351. North is at the top. (d) SapasMons, a typical shieldvolcanoon Venus [see
Keddie and Head, 1994] is locatedin westernAtla Regio (Figure 5a). The 2.4-km-high edifice is comprisesradar-
bright (upper flanks of the volcano) and radar-dark(lower flanks) radial flows. The two dark summitfeaturesare
steep-sidedscalloped-margined domes. Image is about650 km wide. Portionof F-MIDRP 10N188. North is at the
top. (e) Linear graben related to radial fracturing around a central reservoirand interpretedto be the surface
manifestationof wide dikesat shallowdepth;notelargelava flows emanatingfrom severalof these. Locatedsoutheast
ofAtla Regio at about15øS,215ø. Width ofblack boxesis 20 km. Portionof C1-MIDR 15S215. North is at the top.

first order, the characteristicsand global distribution of
volcanic landformsthat are largely extrusive[Keddie and
Head, 1994a, b] and structures that reflect intrusive
activity [Grosfils and Head, 1995] supportthe idea that
neutral buoyancy contributesto major aspectsof volcano
growth and developmenton Venus.
How do these different conditions influence the forma-
tion of LIPs? Populatingthe >80% of the surfaceof Venus
comprisedof volcanicplains are more than 1500 edifices
or volcanic sourcesin excessof 20 km diameter [Head et
al., 1992]. Over 150 of theseare major shieldvolcanoes
(Figure 5d) in excessof 100 km in diameter. The lack of
 HEAD AND COFFIN 425

..

•:;'•:::':
•.:.•"' •:......r-:-•.::-'
.:.:?,<% ::':'!,;: •.':;:.':'• .:?.:.
;.: •• •:"•;:•":,•:
::.•:. ' ':• ,..2,:.
:: ['"::•*':'•" ,•.:',
.....
..... •.:::;: ..•:........ .......... .&•::...? :,:•
, •.•:.*•-••.} .:. .... .•';, ,%,:f ,:.:?...-
,,-•.... **;;5.•5'*:::;'a '": ..-i;•,-;a:• '?-'
;."':'• ,:::**:a
':•' :•:,....-'aS?• ;X"• ,-:-'.':• ;'
'•'.:•:.:;a:•;'-'";'• .*'5:::-
.:?-•.

../•}::::•-•*•:::;*-*:"*'
.?•:;-;-'.½,---½':•
%;4 .......
....•-..:.;..-•:a•½..-•.

'*
. :%:5 .....

Figure 5. (•ntinuexi)

hydrosphereandinsignificant
erosiononVenus meanthat thuskeeping pacewiththeincreasing elevation[Headand
theearlyrecordof thevolcano maybe exposed andthat Wilson, 1992b]. In contrast,
on Venus, because of the
differentphases
in its evolution
cansometimes be more major gradient in atmospheric
pressure with altitude,
the
readilyoutlinedthanis commonly the caseon Earth, chamber's centerbecomes deeper relative
to thesummit of
particularly
whereflowlengths havedecreasedwithtime, the growing edifice. Although the chamber's elevation
leavingexposed sequentialphases of volcanoevolution doesrisewith time,theriserateis low. Therefore,magma
[e.g.,KeddieandHead,1994a]. The heightsof these reservoirs on Venus will remain in the pre-volcano
shieldvolcanoesare considerablylessthanthoseon the substrate longer,andin manycases maynotemerge into
EarthandMars,typicallylessthanabout2 km abovethe the edifice at all. In addition,the lower rate of vertical
surrounding plains[KeddieandHead,1994b]. Several migration impliesthat,for a givenmagmasupplyrate,
factorshelpto account forthesedifferences
andillustrate magma reservoirs wouldtendto stabilize, undergo greater
how LIPs may be producedon Venus. One, the lateralgrowth, and become larger on Venus than Earth.
environment on Venus(surface temperature
andpressure) Thus, the proportion of the available magma going into
favorslargerprimarymagmareservoirs whichwill cause production of the edificerelativeto that intruded intothe
thewidedispersal of conduits
thatbuildedifices,
resulting substrateis smaller on Venus than Earth. The resulting
Two,modelsof shallowNBZ largereservoirs
in broader,flatterstructures. wouldencourage multipleand more
reservoirlocationsduringedificegrowthshowthat, for widely dispersed sourcevents and large volumes forindi-
Earth,the centerof the magmachamberremainsat a vidual eruptions. All of thesefactors resultin volcanic
constantdepthbelowthegrowingsummitof theedifice, edificesthat are low and broad,with reservoirs
426 LARGE IGNEOUS PROVINCES: A PLANETARY PERSPECTIVE

.,

:..'..$-:..

:.-.:•:'-•?..
ß
':?"'•:i:::
....
ß,.......:
..........
.'::.':•.:.::

, :::::::::::::::::::::::::::::::::::::
..•:::::::::::::::::::::::::::::::::::
ß..:•::•::•::::•?::.;:...,:,

".:::;?:::-•:X:

.:'..':.
,...:..::.:?:
:.?:
..... . ;:..:.. ..,,• :.,.•..::::".:•
',:-.-

:-:.<:;:.--,:.,.',:. .:::•,,::::';$•:.•:...:;:,:.,::.:•.-
• ,.-:..,; o.'?.:: .....

.:?. :...•

.
.• .

. ..:': .:,:...•:•.S--•
:'-••, ½.. -;

,. -½- '½.
:: ,, :..;,:.

:?--. , *' ' ' ..." .:..'. '::;•::.':a•:•:-':;:..,

->?:.
• .-<
-. •
.... : .. • :'•:: ::::::::::::::::::::::::::::::
.....

.,....:..-•..-<•: ,:' . : '.?•'•

?:' -..':-:
.:::
::"•....,.-,..._-:-.:,::
. , .. .........

, :::::::::::::::::::::
..... -.......:: , ,-.: ..:•**:½
.....
:,. --
-..* .
,.(...:.:..:.:.-.,

..,.. ., . .:f ,? :.:.,.?:•"."3-.:

..-.'-::'
",;

'"
........
;:':"*;
..'i<........::::::...':'-:'
' . ......,..:•:,..,
•,;• %,::,-:i...•
•,":'..}..:"".:::¾:
::.-.-::;::.•%:-.-;
.:

......

,½.:W"'* :::::::::::::::::::::::
....',:½;::%•W%•::::?<-s-•::X:'
.... .....

. .,: •:.::,...,.,.
:. ::...
,:.•:::<::
,½?':?
.....
..::.?..-? ........
Figure 5.
 HEAD AND COFFIN 427

Figure 5. (continued)

nantly in the substrate,ratherthan the edifice.Becauseof [Grosfils and Head, 1994]. These structuresare inter-
the inhibition of volatile exsolution in the terrestrial sub- preted to be the surface manifestationsof dike swarms
marine environment and its influence on volcanic radiating away from a central source, analogousto the
landforms[e.g., Head et al., 1996b], thesefactorscould giant radiating dike swarms on the Earth such as the
alsobe importantin the formationof LIPs on the seafloor Mackenzie dike swarm in Canada. On Venus, however,
and in the initial stages of Hawaiian-type edifice lack of erosionpermitsthe surfaceequivalentof the deeply
formation. erodedMackenzie-typeswarmsto be studied[e.g., Ernst et
On Venus, over 150 large radiatinglineamentsystems al., 1995]. Theoretical analysisand predictionsof the
with a radius in excessof 100 km have been mapped characteristicsof dikes emplacedin unbuffered(declining
428 LARGE IGNEOUS PROVINCES: A PLANETARY PERSPECTIVE

5e thermal structure of the overlying lithosphere [e.g.,

Erickson and Arkani-Hamed, 1992]. Ascendingmantle
diapirselevateand deformthe lithosphereasthey approach
the base of the lithosphere. There, they flatten, spread
laterally, and cool; on the surface the raised plateau
subsidesto producea centraldepression and a deformed
annulusand moat. One of the most impressiveaspectsof
the documentation of thesefeaturesis that they potentially
representhundredsof examplesof hotspotswhosehistory
canbe mappedbecauseof the lack of erosionand incom-
pletecoverageby later deposits.
A stepin this directionwas takenby MageeRobertsand
Head [1993], who studied the temporal and spatial
relationshipsof areally extensive volcanic flow fields
associatedwith coronae. They showedthat large-scale
flowfields(average
of 1.1x 105km2,butupto 1.5x 106
km2;Figure3) formeda significant
stage
in theevolution
of at least41% of all coronaeand that the timing and scale
Figure 5. (continued)
of many coronaeflow fields are consistent with the arrival
and pressure-release melting of materialin the head of a
driving pressure)and buffered(constantdriving pressure). mantleplume. They also showedthat thosecoronaewith
environments[Parfitt and Head, 1993] showthat buffered associated large-scalevolcanicactivitywere preferentially
driving pressurecan readily accountfor the very wide locatedin areasof rifting (Figure 5c) and interpretedthisto
(exceeding100 m) and very long (in excessof 2000 km) meanthat the intersectionof mantle upwellingswith zones
dikes observed on the Earth and Venus in some mafic dike of extensionis the most significantfactor in the volumeof
swarms. In addition, some dikes emplacedin buffered melt producedand eruptedfrom a corona.
conditions will grow vertically and laterally until they Among the featuresobservedon the volcanicplains of
reach the surface at some distance from the magma Venus are sinuouschannels[Baker et al., 1992], many of
reservoir. The high driving pressureand largedike widths which resemble sinuous rilles on the Moon. Lunar sinuous
typical of these conditionsmean that eruptionsproduced rilles, which range up to about 300 km in length, are
would be characterized by large volumes of lavas thought to have formed from thermal erosionassociated
emplacedat very high rates, potentiallyproducingflood with high-effusion-rateeruptions[e.g., Hulme, 1982]. In
basalts[Parfitt and Head, 1993] (Figure 5e). some caseson Venus the depositsfrom the lavasthat are
An interestingclass of featureswith associatedlarge- proposedto have eroded the sinuous channelscan be
scale effusive activity was discoveredin Earth-basedand identified and distinguishedfrom the surroundingplains.
Venera missionradar images. Named coronae,thesefea- Many of the channelsare much longer than 300 km and,
tures range in size from 60 to over 2000 km in diameter indeed,one of them, Baltis Vallis (Figure 5b), is over 6800
and are characterizedby circularto elongateoutlinesand km in length! These extreme lengthsimply very high
one or more discontinuousannuli comprisingconcentric effusionratesand large-volumeeruptionsandmay indicate
compressional and/or extensional troughs and ridges that the lavas associated with these features were of an
[Pronin and Stofan, 1990]. Magellanprovideddata for a unusualcomposition,temperature,and/or viscosity(e.g.,
global censusof coronae[Stofanet al., 1992; Head et al., possiblykomatiites[seeHead et al., 1994] or carbonatites
1992] that revealed over 300 of these features widely [Kargel et al., 1994]). Clearly, the eruption products
distributedover the surface [Magee Roberts and Head, formingthesechannelswould qualify asLIPs.
1993] and showedthe detailsof their structure[Stofanet Large lava flow fields approachingand exceedingthe
al., 1992; Squyreset al., 1992]. The prevailinginterpre- dimensions of many terrestrial LIPs are observed in
tation is that thesefeaturesrepresentthe surfacemanifes- numerousother settingson Venus. Mylitta Fluctusis one
tation of hotspots or plumes [e.g., $tofan et al., 1991, suchfield thatoriginatesalonga rift zoneandcoversabout
1992; $quyres et al., 1992; danes et al., 1992] and that 3 x 105km2, extending downintoLaviniaPlanitia, a
their diversity in structure and associatedvolcanism lowlandbasin. Maximum flow lengthsrangefrom 400 to
reflects differences in plume size and intensity and the 1000 km, flow widths from 30 to 100 km, and the
 HEAD AND COFFIN 429

volume of the flow field is of the order of 2 x 104 km3 alsocorresponds to the locationof threemajor rifled rises,
[Magee Roberts et al., 1992]. Magee and Head [1995] Beta Regio, Atla Regio, and Themis Regio (thusthe term
documentedthe morphology,morphometry,stratigraphy BAT region). This concentration suggests severalscalesof
and distribution of the global populationof large flow upwelling and instabilities (relatively small for the
fieldsonVenusin excess of 5 x 104km2. Thelargest of individualvolcanicsourceregions,a thousandkilometers
the208suchflowfieldsis 1.6x 106km2andtheaverage for the broad rises, and perhaps many thousandsof
areais 2.2 x 10s km2 (Figure3); collectively, theflow kilometersfor the BAT region). The BAT regioncoversa
fieldscoveran areaof 4.0 x 10?km2, about11%of the comparableplanetary surface area percentage(20%) to
plains regionsof Venus. The most commonsourcevents that of the Tharsisrise on Mars. The agesof its individual
for the largeflow fieldsare coronae,largevolcanicshields, components(rifling, rises,volcanoes)largely postdatethe
and fissures and fractures within rifts and fracture belts. earliest plains emplacement [Crumpier et al., 1994;
Most flow fields are associatedwith zones of extension, Basilevskyand Head, 1995a,b], and thus it may represent
such as major riff zones and fracture belts, and the mantle convectionpatternslinked to the aftermathof the
emplacementof the flow fieldstendedto postdatethe onset collapseof a negatively buoyant, depletedmantle layer
of extension. In reference to terrestrial flood basalts and remaining from the extractionof the basalticcrust [e.g.,
the discussionaboutthe relative importanceof large-scale Head, 1995].
mantle upwelling (e.g., plume heads) versus lithospheric Preliminaryanalysisof the global stratigraphyof Venus
extensioncausingenhanceddecompressional melting, the suggeststhat the dominantgeologicprocesses and stylesof
Venus data supportthe idea that lithosphericextensionand volcanismhave changedover time [Basilevskyand Head,
thinning accompany the formation of the majority of 1995a,b; 1996]. The oldestterrain exposedis known as
flood-basalt lavas there. In addition, examination of a tessera(tile in Greek, for the similarity of the terrain tex-
6800-km-longriff zone interpretedto have originatedfrom ture to parquet floor tiles). This terrain is high-standing
passiverifting in responseto stresseslinked to adjacent and very complexly deformed, somewhatcontinent-like,
downwellingshowsthat extensionoccurredgenerallyprior and comprisesabout 8% of the surfaceof Venus [Iranov
to the eruption of large-scale volcanic flow fields, and Head, 1996]. Tesserais embayedby two major plains
comparable to some terrestrial flood basalts [Magee units and is thus probably much more widespreadin the
Robertset al., 1992; Magee and Head, 1995]. This is in subsurfacethan its presentoutcrop would suggest. The
contrastto the ColumbiaRiver and DeccanBasaltGroups, oldestof the two major plains units (ridged plains) covers
where evidence has been presentedthat eruption of the most of the surface of Venus. Most ridged plains are
main tholeiitic phase precededsignificant extensionand homogeneousand flow-unit boundariesgenerally are not
crustalthinning[Hooper, 1990]. traceable; however, ridged plains are characterizedby
Many riff zoneson Venus are associatedwith broadrises numerous sinuous channels, suggesting large-volume,
resemblingthe Tharsisand Elysiumregionson Mars. The high-effusion-rate eruptions.The stratigraphically younger
Beta, Atla, and WesternEistlaregionsare eachup to 2000- regional plains unit is characterizedby smootherlobate
3000 km in diameter and rise up to several kilometers flows (suchas thoseat Mylitta Fluctus)usuallyemanating
above the surroundingplains (Figure 5a). They are from discrete sources. Although individual flows are
characterizedby riff systemswhich cross(WesternEistla) volumetrically significantand often akin to flood basalts,
or radiateaway from (Beta and Atla) the centralhigh, and they do not have the distinctive sinuouschannelsof the
large shield volcanoes are located on the summit and ridged plains and thus appearto have a differentmode of
flanksof the rise. Positivegravityanomaliesare consistent emplacement. Large volcanic edifices representing
with mantle upwelling [Senskeet al., 1992]. These rises individual volcanic sources, and the emplacementof
appear to representa scale of mantle upwelling much hundreds of flows from subjacent localized magma
larger than that related to individual volcanoes and reservoirs are the most recent features and these are
coronae[e.g., Head et al., 1992], althoughseveralcoronae superposed on most othertypesof plainsunits [Crumpier
reachtremendousdimensions(e.g., Heng-O, 1060 km, and et al., 1997]. Thus, following tesseraformation,two units
Artemis,-2500 km). The global distributionof volcanic that could be interpretedas LiP-related were formed: the
landforms revealed by the Magellan mission showed a ridged plains, where large-scale,sinuous channel-type
concentration of volcanic edifices and sources in an area emplacementoccurred, and the smooth/lobateplains,
comprising
about9.2 x 107 km2, or about20%of the whereflood-basalt-like
pr/vinceswereproduced,
oftenin
surfaceof Venus [Head et al., 1992; Crumpier et al., 1993, conjunctionwith riff zones. This major changeover time
1996b]. This areaof regionallyabundantvolcanicsources (togetherwith substantialchangesobservedin the
430 LARGE IGNEOUS PROVINCES: A PLANETARY PERSPECTIVE

evolutionof Mars and the Moon) suggests that the style, Althoughvolumesarenot well known,theymaybe of the
number, and size of LIPs may vary over the long-term orderof 9.2x 108km3 [Headetal., 1992].Effusion
rates
geologicrecordof the Earth. are likely to have beenvery high for many plainsunits
What causesthese changeson Venus? The size-fre- emplaced through sinuouschannels,but estimatesare
quencydistributionof exposedimpact cratersshowsthat difficult to make. Integratedfluxes were very high,
the averagesurfaceage(craterretentionage)is about300- however.If theridgedplainswereemplaced
in about108
500 Ma, much more similar to that of the Earth than the years,theaverage fluxwouldbe 5-7 km3/a(Figure3),
smallerterrestrialplanets(Figure 1). Even more surpris- which is aboutthree ordersof magnitudemorethan the
ingly, the areal distribution of craters cannot be distin- peak lunar mare flux, approximatelycomparableto the
guishedfrom a completelyspatiallyrandompopulation typical extrusivecomponentof the Earth at present
[Schaberet al., 1992; Phillips et al., 1992; Stromet al., (intraplateand plate boundary)and more than five times
1994] and nearly all of the cratershave not beenmodified greaterthan the flux typical of the last tens of millionsof
by post-emplacementvolcanism. On the basis of these yearsfor Venus. Two, the greatcontrastin magmaflux
results,it washypothesized that Venusunderwenta global betweenthe resurfacingevent and subsequent activity
tectonic and volcanic resurfacingevent about 300-500 showsthat majorchanges cantake placein the geologic
m.y. ago that eradicatedthe previouscrateringrecord. historyof a planet;differenttypesof LIPs can occurin
Subsequent
tothatevent
(thought
tohavelasted
about
10• relativelyrapid succession, and periodscan occurwhen
108years),
volcanism
wasrelatively
minorinvolume
and virtually none are emplaced. Three, if either the
areal distribution(on the basis of the small number of hypothesisconcerningthe buildup and collapseof the
craters modified by volcanic activity) [Schaberet al., depleted mantle layer or the episodic-plate-tectonic
1992; Strom et al., 1994]. Many mechanismshave been hypothesis is correct,thismeansthat large-scale planetary
proposedto explain this hypothesizedevent [e.g., see heatlosscanbe cataclysmic andepisodic,a phenomenon
review of Solomon, 1993], including episodic plate not considered in monotonic thermal evolution models.
tectonics[e.g., Turcotte,1993]. Amongthesehypotheses One implication of the tectonic and volcanic record of
[e.g., Schaber et al., 1992; Parmentier and Hess, 1992; Venusis that the crustformsand evolves[Head, 1990b]
Head et al., 1994] aremechanisms thatcall for near-global primarilyin a verticalsenseratherthan in a lateralsense,
volcanicresurfacing,in effect a planet-wideLIP! In the as is the case in terrestrialoceanicplate spreading,
scenarioproposedby Parmentierand Hess [ 1992], vertical althoughhypotheses for episodicplatetectonicson Venus
crustal accretion leads to formation of a thick, melt- havebeenproposed [e.g.,Turcotte,1993]. Thisconcept of
depleted mantle layer that evolves chemically and verticalcrustalaccretionhasimportantimplications for the
thermally over geologictime; the depletedmantle layer production of LIPs on Venusandthe generalevolutionof
ultimatelybecomesnegativelybuoyantandfounders.This secondarycrustover geologictime [Head et al., 1994], as
event is predictedto occur over a geologicallyshorttime discussed below.
[Parmentier and Hess, 1992] and the foundering,
downwellingdepletedmantlelayeris hypothesized to have Other Planetary Bodies
deformed much of the crust into tessera terrain, while the
complementary upwelling fertile mantle underwent Outsidethe orbits of the terrestrialplanetslie the aster-
massivepressure-release melting to producevoluminous oid belt and the outer gas giant planetsand their satellites.
sinuous-channel-relatedflood basalts over a relatively Some meteorites and asteroids show evidence for
shorttime(z 108yrs)[Headet al., 1994].Subsequently, differentiationand basalticvolcanism[e.g., Taylor et al.,
volcanismwanedbut was locally significantin riffs (where 1993], phasesof which may have been volumetrically
local flood basalt units were emplaced,such as Mylitta significant [e.g., Wilson and Keil, 1996]. Outer planet
Fluctus) and near hotspots, where magma reservoirs satellitesare predominantlylow-densitybodiescomposed
evolvedto producevolcanicedifices(e.g., SapasMons). primarily of water and related ices [e.g., Burns and
Three importantobservations canbe maderelativeto the Matthews, 1986]. One exceptionis the innermostof the
study of LIPs on Earth. One, the hypothesizeddepleted- Galilean satellitesof Jupiter,Io, which is approximately
mantle-layeroverturneventon Venus couldbe the equiva- the same size and densityas the Earth'smoon. In one of
lent of a planet-wide LIP. Approximately 80% of the the most spectacularpredictions[Peale et al., 1979] and
planet(3.68 x 108km2;Figure3) may havebeen discoveries [Morabito et al., 1979] of planetary
resurfacedover a very shorttime duringthe emplacement exploration,imagesreturnedby Voyager showednumer-
of the ridged plains [Basilevskyand Head, 1995a, b]. ous active volcanic eruptionson Io [Smith et al.,
 HEAD AND COFFIN 431

Througha combinationof pyroclasticeruptionsand lava influence on the occurrence,depth, and size of magma
flows, Io appearsto be resurfacedat the phenomenally reservoirsand thus on the possibility of flood basalts
highrateof 10'4to 1 cm/yr[Nashet al., 1986].Further [Headand Wilson,1992b]. Thisconsideration impliesthat
explorationby the Galileo missionwill provideevidence intraplate submarinereservoirsand extrusions(more
for the natureof changeson Io in the last 17 yearsandthe Venus-like) may be different from those in subaerial
relationof theseresurfacingratesandstylesto LIPs. environmentson Earth, as indeedmight many LIPs formed
earlier in Earth'shistory, when atmosphericpressuremay
SUMMARY, RELEVANCE TO TERRESTRIAL LIPS, have beenhigher.
AND OUTSTANDING QUESTIONS
Controls on Mode and Location

Environments,Associations,Settings of Formation and

Style of Emplacement The lunar record showshow a low-densitycrustallayer
analogous to continentalcruston Earthcanact asa filter to
The planetaryrecordprovidesa perspectiveon the three
plumesand associated volcanicactivity, obscuringtheir
main categoriesof terrestrialLIPs: oceanicplateaus,conti-
surfacemanifestationand evenprecludingthe construction
nental flood basalts,and volcanicpassivemargins. Multi-
of shallowmagmareservoirsand large shieldvolcanoes.
ple analogsto continentalflood basalts(e.g., Moon) andto
Variations in lithosphericthicknesson the terrestrial
oceanicplateaus(e.g., Mars, Venus)existon the terrestrial
planetsillustratehow thermalstructurecan influencethe
planets;the rift-related LIPs on Venus provide important
occurrenceand modeof large-volumeextrusionsand how
informationaboutsequenceand timing of emplacement in
changesin lithosphericthicknesswith time can alter the
relation to volcanic passivemargins. The role of large-
styleand abundanceof flood basalts.
volume, long-distance,lateral dike emplacementof flood
basaltsis illustratedby coronaeon Venus; theseprovide
Relation to Internal Structure
probableanalogsto the now-erodedgiant radiatingdike
swarmsof Earth and associated,but largely eroded,flood The wide range of plume-likefeatureswith associated
basalts. The greatrangeof scalesof upwellingson Venus large extrusive componentssuggeststhat plume sources
and Mars, the influence of crustalthicknessand composi- could possiblyextendfrom the uppermantleto the core-
tion, and lithosphericthicknessvariationsin spaceand mantle boundary. The extremely large and long-lasting
time also are significant for studies of terrestrial provinceson Mars (Tharsis)and Venus (BAT) strongly
environments. Venus showsthe potentialsignificanceof suggestthat even larger instabilitiesoccur than those
vertical crustal accretion and the influence of the
commonlyassociatedwith individualterrestrialplumes.
complementary depleted mantle layer on further Betterknowledgeof variationsin the volcanicflux associ-
petrogeneticevolution, as well as the possibility of ated with plume-like featureson Venus will help us to
episodic,cataclysmicplanetary-scale resurfacing. Theo- understandterrestrial plume and mantle structure[e.g.,
retical analysisof environmentson Venus also illustrates Bercovici and Mahoney, 1994]. Another potentially
that differences in thermal structure can cause fundamental
important perspectivecomes from the vertical crustal
differencesin the volume of melt produced in buoyant accretionhypothesisfor Venus; the formationand accu-
upwellingsin both rise [e.g.,Sotinet al., 1989] and plume mulation of a complementarymelt-depletedmantle layer
[e.g., Ericksonand Arkani-Hamed,1992] environments.It cansignificantlyalterthe natureof furthermelt production
hasbeenproposedthatmanyterrestrialLIPs areanalogous and can also influencecrustalbuoyancyand stability. The
to the lunar maria and resulted from impact-related lunarhighlandcrustdensitybarrieralso illustratesthat in
pressure-release melting[e,g.,Alt et al., 1988]. The lunar somecasesbasaltsmay localizeat the baseof the crust,as
geologicrecordshowsthat this is unlikelyat the present is believed to occur for terrestrial flood basalt magmas.
time because of the generally small size of impact Venus demonstrates that large-scalemelting in the mantle
projectiles,but that it may havebeenmore significantin (the hypothesizedlarge-scalemantleoverturn)may have
pastEarthhistorywhenlargerimpactsoccurred, especially hadvery long-lastingeffectson mantleconvection patterns
in oceanicsettings. andvolcanism[e.g.,Crumpleret al., 1993].

Influenceof SurfaceEnvironment Implicationsfor PlumeStructure

The exampleof Venus suggeststhat the externalenvi- Although the compositional,thermal and mechanical
ronmentinto which magmais extrudedcan have a major structureof the crustand lithosphereof the planetsis
432 LARGE IGNEOUS PROVINCES: A PLANETARY PERSPECTIVE

the same as on the Earth today, the planetaryrecordcan and degreesof melting in the earlier history of planets,
provide a frame of referencefor questionssuchas plume periodsof possiblecataclysmicresurfacing(e.g., on Venus
structure(plume headsand tails), plume incubationversus [Head et al., 1994]), and periods of anomalousrates of
plume impact [e.g., Kent et al., 1992], internal mantle mantle convection. The cataclysmic resurfacing
structure[e.g., Bercoviciand Mahoney, 1994], and plume hypothesisfor Venus also illustratesthe possibilitythat
duration(e.g., on Venusand Mars). thermal evolutionmay not be steady-stateand monotonic
but ratherepisodic[seealsoCondie, 1995].
Large IgneousProvinceSubstructure
PetrogeneticEvolution
Eroded, giant radiatingdike swarmson Earth (e.g., the
Mackenzie dike swarm) and their unerodedcounterparts Do extraterrestrial LIPs contain differentiates of basalt,
on Venus [e.g., Ernst et al., 1995] showthat flood basalts and if so, where do theseoccur in the sequence?In situ
neednot occuronly abovea plumehead;dike thicknesses, geochemicalanalyseson Venussuggestthat tholeiitesand
lengths,and flow ratesare suchthat flood basaltscan occur possiblymore alkaline basalts[Surkovet al., 1987] form
severalthousandkilometersaway from a plume, through the vast volcanic plains. The extensivesinuouschannels
lateral transportof magma in dikes and its eruptiondue to have been interpretedas evidencefor possiblekomatiites
buffered conditionsin the magma reservoir [Parfitt and [Head et al., 1994] and carbonatites[Kargel et al., 1994].
Head, 1993]. Steep-sideddomes [Pavri et al., 1992] and large deposits
of viscous-appearing deposits[Moore et al., 1992] have
Areas and Volumes been observed. Unfortunately,widespreadand detailed
chemicalanalyseshave not yet beenmadeon the planets.
PlanetaryLIPs have a wide range of areasand volumes
(Figure 3), showingthat terrestrialLIPs are not unique in Influenceon the Atmosphereand Environment
this respect. In addition, volumes range up to that of
Tharsison Mars, and scaleson Venus exceedthe present Voluminous and prolongedvolcanic outpouringscan
surfacearea of the oceaniccrust. Theseexamplessuggest make importantcontributionsto the atmosphereof planets
the possibilityof larger terrestrialLIPs than presentlyrec- throughouttheir evolution, as on Mars [e.g., Greeley,
ognized(e.g., terrestrialsuperplumes [Larson, 1991a,b]). 1987]. If large volumesof flood basaltsare extrudedover
The volumes of the larger planetary LIPs and the very short periods, outgassedvolatiles, heat flux, and
stationarylithosphereof most terrestrialplanetssuggest voluminous particulate matter can influence short-term
that voluminouspartial melting of mantle has occurred chemistryand circulationof the atmosphereand long-term
(e.g., seefigure 10 of Coffin and Eldholm[1994]) and that climate evolution. Potentiallythe most dramaticexample
residual, depletedmantle layers must play an important of this is the widespreadvolcanicresurfacinghypothesized
role in the continuedevolutionof planetaryuppermantles. for Venus. For example, Bullock and Grinspoon[1996]
showed that an increased flux of volcanism such as that
Duration and Ratesof Emplacement interpretedto be associatedwith the proposedglobalresur-
facing would precipitatea climaticcatastropheleadingto
Planetary LIPs are seen in which volumes were very muchhighertemperaturesandpressures.
high and eruptiondurationswere both short(the largeout-
flows and sinuousrilles on the Moon) and long (the shield Relation to Geologic History
volcanoeson Mars). In addition, the Tharsisrise on Mars
showsthat mantle melting anomaliescan last billions of A mostimportantperspectivefrom the planetsis that the
years, producing prodigiousLIPs, and the Venus global characteristicsand ratesof geologicprocesses as a function
resurfacing model suggests that large-scale mantle of time and thermal evolutionhave experiencedlarge-scale
overturnmay provideshort-term (•<108yr) pulsesof changes. The geological processesdominating the
global-scale igneous provinces. On the basis of the geologicrecordover the last severalhundredmillion years
planetaryperspective,high eruptionratesmay be due to a on the planets (i.e., the temporal equivalent of the
variety of conditions,includingtrappingof melt at density Phanerozoicon the Earth) are not the sameas thoseoperat-
barriers and subsequentoverpressurizationof reservoirs ing in the earlier historyof Mars, Venus,Mercury, and the
[Head and Wilson, 1992a], bufferedconditionsin magma Moon. Thus, we should anticipate potentially major
reservoirs[Parfitt and Head, 1993], higher temperatures changesin the style of volcanicextrusionas a function
 HEAD AND COFFIN 433

geologictime on Earth. It is clear from the planetaryrec- Jakosky,C. W. Snyder, and M. S. Matthews, pp. 249-297,
ord that coincidentwith the generalthermal evolutionof The University of Arizona Press,Tucson, 1992.
the planet, changescan occur in the mantle convection Barger, K. E., and E. D. Jackson,Calculatedvolumesof individ-
planform, the development and scale of mantle ual shield volcanoesalong the Hawaiian-EmperorChain, d.
Res., U.S. Geol. Surv., 2, 545-550, 1974.
instabilities,and the conditionsof melting in the Earth's
Basaltic Volcanism Study Project, Basaltic Volcanism on the
crust. Much of the evidencefrom the planetsand from Terrestrial Planets, 1286 pp., PergamonPress,New York,
thermal evolution modelssuggestsa more importantrole 1981.
for LIPs in the earlierhistoryof the Earth. Basilevsky,A. T., and J. W. Head, Global stratigraphyof Venus:
Analysis of a random sampleof thirty-six test areas,Earth,
Moon and Planets, 66, 285-336, 1995a.
Origin of LIPs.' A Planetary Perspective Basilevsky,A. T., and J. W. Head, Regionaland global strati-
graphyof Venus:A preliminaryassessment and implications
The planetary perspectiveprovidesmany examplesof for the geologic history of Venus, Planet. Space Sci., 43,
1523-1553, 1995b.
LIPs in a diverserangeof geologicalenvironments.This
Basilevsky,A. T., and J. W. Head, Evidencefor rapid and wide-
underlinesthe fact that there is no singleorigin for LIPs,
spread emplacementof volcanic plains on Venus: Strati-
but that, taken together,they can help to understandthe graphicstudiesin the BaltisVallis region,Geophys.Res.Lett.,
nature and significanceof large-scalemelting in the shal- 1497-1500, 1996.
low interiorsof planets[e.g., Coffin and Eldholm, 1994]. Bercovici,D., and J. Mahoney, Double flood basaltsand plume
For example,the broadrifted risesof Venus and Mars and head separationat the 660-kilometerdiscontinuity,Science,
their associatedLIPs serveas potentialanalogsfor early 266, 1367-1369, 1994.
continental breakup and the early stages of crustal Bratt, S. R., S.C. Solomon, and J. W. Head, The evolution of
spreadingon Earth. Continuedanalysisof data from the impact basins:Cooling, subsidence,and thermal stress,90,
12,415-12,433, 1985.
planetary record will help to provide perspectiveon
Bruno, B.C., G. J. Taylor, S. K. Rowland, P. G. Lucey, and S.
terrestrial LIP dimensions, durations, and rates of
Self, Lava flows are fractals,Geophys.Res.Lett., 19, 305-308,
emplacement,as well as mantle and crustal structureand 1992.
processes,
relationshipto tectonism,environmentaleffects Bullock, M. A., and D. H. Grinspoon,The stabilityof climateon
and petrological and geochemical characteristicsand Venus,d. Geophys.Res., I01, 7521-7529, 1996.
evolution. Burns, J. A., and M. S. Matthews, editors, Satellites, The
University of Arizona Press,Tucson, 1021 pp, 1986.
Cameron,A. G. W., B. Fegley Jr., W. Benz, and W. L. Slattery,
Acknowledgments.Researchfor this paperwas supportedby a The strangedensityof Mercury: Theoreticalconsiderations, in
grant from the National Aeronauticsand Space Administration Mercury, editedby F. Vilas, C. R. Chapman,M. S. Matthews,
(NAGW-2185) to JWH. MFC acknowledges the supportof the pp. 692-708, The University of Arizona Press,Tucson,1988.
Industrial Liaisons Programof the University of Oslo, Norway. Carr, M. H., Volcanism on Mars, o•. Geophys.Res., 78, 4049-
4062, 1973.
Thanksare extendedto Mary Ellen Murphy, Anne C. C6t6, Anne
Carr, M. H., The role of lava erosion in the formation of lunar
McKay and Peter Neivert for help in preparation of the
manuscript,and to G. J. Taylor and Laszlo Kesthelyi for very rilles and martian channels,Icarus, 22, 1-23, 1974.
helpful reviews. Specialthanksare extendedto JohnMahoney Carr, M. H., The Surface of Mars, Yale University Press,New
for his excellent scientificand editorial suggestions.University Haven and London, p. 232, 1981.
of Texas Institutefor Geophysicscontributionnumber1269. Chapman, C. R., Mercury: Introduction to an end-member
planet,in Mercury, editedby F. Vilas, C. R. Chapman,and M.
S. Matthews, pp. 1-23, The University of Arizona Press,
Tucson, 1988.
REFERENCES
Coffin, M. F., and O. Eldholm, Volcanism and continentalbreak-
up: A global compilation of large igneous provinces, in
Alt, D., J. M. Sears,and D. W. Hyndman,Terrestrialmaria:The Magmatism and the Causesof ContinentalBreak-Up, Spec.
originsof large basalticplateaus,hotspottracksand speading Publ. 68, edited by B.C. Storey, T. Alabaster, and R. J.
ridges,d. Geol., 96, 647-662, 1988. Pankhurst,pp. 21-34, The GeologicalSociety,London, 1992.
Baker, Vo R., G. Komatsu, T. J. Parker, V. C. Gulick, J. S. Coffin, M. F., and O. Eldholm, Scratchingthe surface:Estimat-
Kargel, and J. S. Lewis, Channels and valleys on Venus: ing dimensionsof large igneousprovinces,Geology,21, 515-
Preliminary analysisof Magellan data, d. Geophys.Res., 97, 518, 1993.
13,421-13,444, 1992o Coffin, M. F., and O. Eldholm, Large igneousprovinces:Crustal
Banerdt, W. B., M.P. Golombek, and K. L. Tanaka, Stressand structure, dimensions, and external consequences,Rev.
tectonicson Mars, in Mars, edited by H. H. Kieffer, B. M.
_
Geophysics,32, 1-36,
434 LARGE IGNEOUS PROVINCES: A PLANETARY PERSPECTIVE
Comer,R., S. Solomon,and J. W. Head, Mars: Thicknessof the
lithosphere
from the tectonicresponse
to volcanicloads,J.
Geophys.Res.,23, 61&92,1985.
Condie, K. C., Episodicages of greenstones:
A key to mantle
dynamics?,Geophys.Res.Lett., 22, 2215-2218, 1995.
Crisp,J. A., Ratesof magmaemplacement and volcanicoutput,
J. Volcanol. Geotherm. Res., 20, 177-211, 1984.
Cruikshank,D. P., The Developmentof Studiesof Venus, in
Venus,editedby D. M. Hunten,L. Colin, T. M. Donahue,and
V. I. Moroz, pp. 1-9, The University of Arizona Press,
Tucson, 1983.
Crumpier,L. S., J. W. Head, andD. B. Campbell,Orogenicbelts
on Venus, Geology,14, 1031-1034, 1986.
Crumpier,L. S., J. W. Head, and J. C. Aubele,Relationto major
volcanic center concentrationon Venus to global tectonic
patterns,Science,261, 591-595, 1993.
Crumpier,L. S., J. C. Aubele, and C. D. Condit,Volcanoesand
neotectoniccharacteristicsof the Springervillevolcanicfield,
Arizona, New Mexico Geological Society Guidebook,45th
Field Conference,Mogollon Slope,West-CentralNew Mexico
and East-CentralNew Mexico, 147-164, 1994.
Crumpier,L. S., J. W. Head, and J. C. Aubele,Calderason Mars:
Characteristics, structural evolution, and associated flank
structures,in Volcano Instability on the Earth and Other
Planets, Spec. Publ. 110, edited by W. J. McGuire, A. P.
Jones,and J. Neuberg, eds., pp. 307-348, The Geological
Society,London, 1996.
Crumpier,L. S., J. C. Aubele,D. A. Senske,S. T. Keddie,K. P.
Magee, and J. W. Head, Volcanoesand centersof volcanism
on Venus, VenusII, University of Arizona Press (in press),
1997.
Erickson, S. G., and J. Arkani-Hamed, Impingementof mantle
plumeson the lithosphere:ContrastbetweenEarthandVenus,
Geophys.Res.Lett., 19, 885-888, 1992.
Ernst, R. E., J. W. Head, E. Parfitt, E. Grofils, and L. Wilson,
Giant radiatingdyke swarmson Earth and Venus, Earth-Sci.
Rev., 39, 1-58, 1995.
Esposito,P. B., W. B. Banerdt,G. F. Lindal, W. L. Sjogren,M.
A. Slade, B. G. Bills, D. E. Smith, and G. Balmino, Gravity
and topography,in Mars, edited by H. H. Kieffer, B. M.
Jakosky,C.W. Snyder,and M. S. Matthews,pp. 209-248, The
University of Arizona Press,Tucson, 1992.
Feighner,M., and M. Richards,Lithosphericstructureand com-
pensation mechanisms of the Galapagos Archipelago,
Geophys.Res.,99, 6711-6729, 1994.
Finnerty, A. A., R. J. Phillips, and W. B. Banerdt, Igneous
processesand the closed system evolution of the Tharsis
regionof Mars, •/. Geophys.Res.,93, 10,225-10,235, 1988.
Frey, H., Crustalevolutionof the early Earth: The role of major
impacts,Precamb.Res., 10, 195-216, 1980.
Glazner, A., and W. Ussler,Trappingof magmaat midcrustal
density discontinuities,Geophys. Res. Lett., 15, 673-675,
1988.
Goettel, K. A., Present bounds on the bulk compositionof
Mercury: Implicationsfor the planetaryformationprocess,in
Mercury, editedby F• Vilas, C. R. Chapman,M. S. Matthews,
pp. 613-621, The Universityof ArizonaPress,Tucson,1988.
Greeley,R., Lunar Hadley Rille: Considerations
of its origin,
Science, 172, 722-725, 1971.
Greeley, R., Release of juvenile water on Mars: Estimated
amountsand timing associatedwith volcanism,Science,236,
1653-1654, 1987.
Greeley, R., and J. E. Guest,Geologicmap of the easternequa-
torial region of Mars, U.S. Geol. Surv. Misc. Inv. SeriesMap
I-1802-B, 1987.
Greeley, R., M. J. S. Belton, L. R. Gaddis, J. W. Head, S. D.
Kadel, A. S. McEwen, S. L. Murchie, G. Neukum, C. M.
Pieters,J. M. Sunshine,and D. A. Williams, Galileo imaging
observationsof lunar maria and relateddeposits,J. Geophys.
Res., 98, 17,183-17,205, 1993.
Grieve, R. A. F., Impact bombardmentand its role in proto-
continentalgrowthon the early Earth,Precamb.Res., 10, 217-
248, 1980.
Grieve, R. A. F., and E. M. Parmentier,Impact phenomenaas
factors in the evolution of the Earth, Proc. 27th Int. Geol.
Cong, 19, VNU SciencePress,99-114, 1984.
Grieve, R. A. F., P. B. Robertson,and M. R. Dence, Constraints
on the formationof ring impactstructures,basedon terrestrial
data, in Multi-Ring Basins,editedby P. H. Schultzand R. B.
Merrill, pp. 37-57, Pergamon,New York, 1991a.
Grieve, R. A. F., D. Stoffler, and A. Deutsch, The Sudbury
Structure: Controversialor misunderstood?, J. Geophys.Res.,
96, 22,753-22,764, 1991b.
Griffiths, R. W., and I. H. Campbell,Interactionof mantleplume
heads with the Earth's surface and onset of small-scale
convection,J. Geophys.Res.,96, 18,295-18,310, 1991.
Grosfils,E. B., and J. W. Head, The global distributionof giant
radiating dike swarmson Venus: Implicationsfor the global
stressstate,Geophys.Res.Lett., 21, 701-704, 1994.
Grosfils, E. B., and J. W. Head, Radiatingdike swarmson Venus:
Evidence for emplacementat zones of neutral buoyancy,
Planet. SpaceSci., 43, 1555-1560, 1995.
Guest,J. E., Centersof igneousactivity in the maria, in Geology
and Geophysicsof the Moon, editedby G. Fielder, pp. 41-53,
Elsevier, Amsterdam, 1971.
Hall, J. L., S.C. Solomon, and J. W. Head, Elysium Region,
Mars: Testsof lithosphericloadingmodelsfor the formation
of tectonic features, J. Geophys. Res., 91, 11,377-11,392,
1986.
Hartmann, W. K., Martian crateringV: Toward an empirical
Martian chronology,and its implications,Geophys.Res.Lett.,
5, 450-452, 1978.
Head, J. W., Orientalemulti-ringedbasin interior and implica-
tions for the petrogenesisof lunar highland samples,The
Moon, 11, 327-356, 1974.
Head, J. W., Mode of occurrenceand style of eraplacementof
lunar mare deposits,in Origin of Mare Basalts, Proc. Lunar
Sci. Institute Conf., 234, 61-65, 1975a.
Head,J. W., Lunarmaredeposits:Areas,volumes,sequence, and
implication for melting in sourceareas, in Origin of Mare
Basalts,Proc. Lunar Sci. InstituteConf., 234, 66-69, 1975b.
Head, J. W., Lunar volcanismin spaceand time, Rev. Geophys.
SpaceSci., 14, 265-300, 1976.
Head, J. W., The formation of mountain belts on Venus:

dencefor large-scaleconvergence, underthrusting and crustal
imbricationin Freyja Montes, Ishtar Terra, Geology,18, 99-
102, 1990a.
Head, J. W., Processesof crustal formation and evolution on
Venus: An analysis of topography and crustal thickness Hulme, G., A review of lava flow processesrelated to the
variations,Earth, Moon and Planets, 50/51, 25-55, 1990b.
Head, J. W., Processesof crustaland depletedmantle layer loss
on Venus: Evidence from basins in tesserae,uplands, and Hunten, D. M., L. Colin, T. M. Donahue,and V. I. Moroz (Eds.),
plains,Lunar Planet. Sci. Conf, 26, 577-578, 1995.
Head, J. W., and S.C. Solomon, Tectonic evolution of the
terrestrialplanets,Science,213, 62-76, 1981. Ivanov, M. A., and J. W. Head, Tessera terrain on Venus: A
Head, J. W., andL. Wilson, The formationof erodeddepressions
around the sources of lunar sinuous rilles: Observations,
Lunar and Planet. Sci. Conf., I I, 426-428, 1980.
Head, J. W., and L. Wilson, Volcanic processes and landformson Janes,D. M., S. W. Squyres,D. L. Bindschadler,G. Baer, G.
Venus: Theory, predictions,and observations,J. Geophys.
Res., 91, 9407-9446, 1986.
Head, J. W., and L. Wilson, Absence of large shield volcanoes
and calderason the Moon: Consequenceof magmatransport J6nsson,J., Vocanic eruptionin historicaltime on the Reykjanes
phenomena?,Geophys.Res.Lett., 18, 2121-2124, 1991.
Head, J. W., and L. Wilson, Lunar mare volcanism:Stratigraphy,
eruption conditions,and the evolution of secondarycrusts, Kargel, J. S., R. L. Kirk, B. Fegley Jr., and A. H. Treiman,
Geochim. Cosmochim. Acta, 55, 2155-2175, 1992a.
Head, J. W., and L. Wilson, Magma reservoirsand neutralbuoy-
ancy zones on Venus: Implications for the formation and
evolution of volcanic landforms,d. Geophys.Res., 97, 3877-
3903, 1992b.
Head, J. W., and L. Wilson, Lunar grabenformationdue to near-
surfacedeformationaccompanyingdike emplacement, Planet.
SpaceSci., 41, 719-727, 1994a.
Head, J. W., and L. Wilson, Mars: Formation and evolution of
magma reservoirs, Lunar Planet. Sci. Conf., 25, 527-528,
1994b.
Head, J. W., L. S. Crumpier,J. C. Aubele, J. E. Guest,and R. S.
Saunders, Venus volcanism: Classification of volcanic fea-
turesand structures,associations, and global distributionfrom
Magellan data,J. Geophys.Res.,97, 13,153-13,197, 1992.
Head, J. W., S. Murchie, J. F. Mustard, C. M. Pieters,J. Neukum,
A. McEwen, R. Greeley, E. Nagel, and M. J. S. Belton, Lunar
impact basins:New data for the western limb and far side
(Orientale and South Pole-Aitken Basins) from the first
Galileo flyby, J. Geophys.Res.,98, 17,149-17,181,1993.
Head, J. W., E. M. Parmentier, and P. C. Hess, Venus: Vertical
accretionof crust and depletedmantle and implicationsfor
geologicalhistory and processes,Planet. SpaceSci., 42, 803-
81 l, 1994.
Head, J. W., A. T. Basilevsky,L. Wilson, and P. C. Hess, Evo-
lution of volcanicstyleson Venus: Changebut not Noachian?,
Lunar Planet. Sci. Conf., 27, 525-526, 1996a.
Head, J. W., L. Wilson, and D. K. Smith, Mid-oceanridge erup-
tive vent morphologyand structure:Evidencefor dike widths,
eruptionrates, and evolution of eruptionsand axial volcanic
ridges,•/. Geophys.Res., I01, 28,265-28,280, 1996b.
Heiken, G. H., D. T. Vaniman, and B. M. French(Eds.), Lunar
Sourcebook: A User's Guide to the Moon, 736 pp.,
CambridgeUniversityPress,New York, 1991.
HodgesC. A., and H. J. Moore, Atlas of VolcanicLandformson
Mars, U.S. Geol. Surv. Prof. Pap. 1534, p. 194, 1994.
HEAD AND COFFIN 435
Hooper,P. R., The timingof crustalextensionandthe eruptionof
continentalflood basalts,Nature, 345, 246-249, 1990.
Hulme, G., Turbulent lava flow and the formationof lunar sinu-
ous rilles, Mod. Geol., 4, 107-117, 1973.
formation of lunar sinuousrilles, Geophys.Surveys,5, 245-
279, 1982.
Venus, 1143 pp., The University of Arizona Press,Tucson,
1983.
survey of the global distribution,characteristics,
and relation
to surroundingunits from Magellan data, d. Geophys.Res.,
I01, 14,861-14,908, 1996.
Schubert, V. L. Sharpton, and E. R. Stofan, Geophysical
models for the formation and evolution of coronae on Venus,
d. Geophys.Res., 97, 16,055-16,067, 1992.
Peninsula, southwest Iceland (in Icelandic with English
summary),N•ttt•rufraedingurlun,52, 127-139, 1983.
Carbonate-sulfatevolcanism on Venus?, Icarus, 112, 219-252,
1994.
Keddie, S. T., and J. W. Head, SapasMons, Venus:Evolutionof
a large shieldvolcano,Earth, Moon and Planets,65, 129-190,
1994a.
Keddie, S. T., and J. W. Head, Height and altitudedistributionof
large volcanoeson Venus, Planet. Space Sci., 42, 455-462,
1994b.
Kent, R. W., M. Storey, and A.D. Saunders,Large igneous
provinces: Site of plume impact or plume incubation?,
Geology,20, 891-894, 1992.
Kieffer, H. H., B. M. Jakosky, C. W. Snyder, and M. S.
Matthews (Eds.), Mars, 1498 pp., The University of Arizona
Press,Tucson, 1992a.
Kieffer, H. H., B. M. Jakosky, and C. W. Snyder, The planet
Mars: From antiquityto the present,in Mars, editedby H. H.
Kieffer, B. M. Jakosky, C. W. Snyder, and M. S. Matthews,
pp. 1-33, The University of Arizona Press,Tucson, 1992b.
Larson, R. L., Latest pulse of Earth: Evidence for a mid-
Cretaceoussuperplume,Geology,19, 547-550, 1991a.
Larson, R. L., Geological consequencesof superplumes,
Geology, 19, 963-966, 1991b.
Longhi, J., E. Knittle, J. R. Holloway, and H. W•inke, The Bulk
Composition,Mineralogy and Internal Structureof Mars, in
Mars, edited by H. H. Kieffer, B. M. Jakosky,C. W. Snyder,
and M. S. Matthews, pp. 184-208, The University of Arizona
Press, Tucson, 1992.
Magee Roberts, K., and J. W. Head, Large-scale volcanism
associatedwith coronaeon Venus: Implicationsfor formation
and evolution,Geophys.Res.Lett., 20, 1111-1114, 1993.
Magee, K., and J. W. Head, The role of rifting in the generation
of melt: Implicationsfor the origin and evolutionof the Lada
Terra-Lavinia Planitia region of Venus,d. Geophys.Res., 100,
1527-1552, 1995.
Magee Roberts,K., J. E. Guest,J. W. Head, and M. G. Lancaster,
Mylitta Fluctus, Venus: Rift-related, centralized
436 LARGE IGNEOUS PROVINCES: A PLANETARY PERSPECTIVE
and the emplacementof large-volumeflow units,d. Geophys.
Res., 97, 15,991-16,015, 1992.
Martin, B. S., The Roza Member, Columbia River Basalt Group;
Chemicalstratigraphyand flow distribution,in Volcanismand
Tectonismin the ColumbiaRiver Flood-basaltProvince,Spec.
Pap. 239, edited by S. Reidel and P. Hooper, Geological
Societyof America,, pp. 85-104, 1989.
McGovern, P., and S. Solomon, State of stress,faulting, and
eruption characteristicsof large volcanoes on Mars, d.
Geophys.Res., 98, 23,553-23,579, 1993.
McGuire, W. J., A. P. Jones,and J. Neuberg (Eds.), Volcano
Instability on the Earth and Other Planets, Spec.Publ. 110,
388 pp., The GeologicalSociety,London, 1996.
McKenzie, D., P. G. Ford, C. Johnson,B. Parsons,D. Sandwell,
S. Saunders,and S.C. Solomon,Featureson Venus generated
by plate boundaryprocesses, d. Geophys.Res., 97, 13,533-
13,544, 1992.
McNutt, M., K. Fischer, S. Kruse, and J. Natland, The origin of
the Marquesasfracturezoneridge and its implicationsfor the
nature of hot spots,Earth Planet. Sci. Lett., 91, 381-393,
1989.
Moore, H. J., J. J. Plaut, P.M. Schenk, and J. W. Head, An
unusual volcano on Venus, d. Geophys. Res., 97, 13,479-
13,493, 1992.
Morabito, L. A., S. P. Synnott,P. Kupfermanand S. A. Collins,
Discovery of currently active extraterrestrialvolcanism,
Science, 204, 972, 1979.
Mouginis-Mark, P. J., L. Wilson, and J. W. Head, Explosive
volcanismat HecatesTholus, Mars: Investigationof eruption
conditions,d. Geophys.Res.,87, 9890-9904, 1982.
Mouginis-Mark, P. J., L. Wilson, J. W. Head, S. H. Brown, J. L.
Hall, and K. Sullivan, Elysium Planitia, Mars: Regional
geology, volcanologyand evidencefor volcano/groundice
interactions,Earth, Moon, and Planets, 30, 149-173, 1984.
Nash, D. B., M. H. Carr, J. Gradie, D. M. Hunten and C. F.
Yoder, Io, in Satellites, J. A. Burns and M. S. Matthews,
editors, pp. 629-688, The University of Arizona Press,
Tucson, 1986.
Neukum, G., and D. U. Wise, Mars: A standardcrater curve and
possiblenew time scale,Science,194, 1381-1387,1976.
Oberbeck,V. R., The role of ballistic erosionand sedimentation
in lunar stratigraphy,Rev. Geophys.,13, 337-362, 1975.
Oberbeck,V. R., W. L. Quaideand R. Greeley,On the origin of
lunar sinuousrilles, Mod. Geol., 1, 75-80, 1969.
Oberbeck,V. R., F. H6rz, R. H. Morrison, W. L. Quaide, and D.
E. Gault, On the origin of the lunar smooth-plains,TheMoon,
12, 19-54, 1975.
Oberbeck, V. R., J. R. Marshall, and H. Aggarwal, Impacts,
tillites, andthe breakupof Gondwanaland,
d. Geology,101, 1-
19, 1993.
Parfitt, E. A., and J. W. Head, Buffered and unbuffered dike
emplacementon Earth and Venus: Implicationsfor magma
reservoirsize,depth,andrate of magmareplenishment,
Earth,
Moon, and Planets, 61, 249-281, 1993.
Parmentier,E. M., and P. C. Hess, Chemical differentiationof a
convectingplanetaryinterior: Consequences for a one plate
planet such as Venus, Geophys.Res. Lett., 19, 2015-2018,
1992.
Pavri, B., J. W. Head, K. B. Klose, and L. Wilson, Steep-sided
domes on Venus: Characteristics,geologic setting, and
eruptionconditionsfrom Magellan data,d. Geophys.Res., 97,
13,445-13,478, 1992.
Peale,S. J., P. Cassenand R. T. Reynolds,Melting of Io by tidal
dissipation,Science,203, 892-894, 1979.
Peterson,D. W., and R. B. Moore, Geologic history and evolu-
tion of geologic concepts,Island of Hawaii, in Volcanismin
Hawaii, Prof. Pap, 1350, edited by R. W. Decker, T. L.
Wright, and P. H. Stauffer, pp. 149-189, U.S. Geological
Survey,Washington,D.C., 1987.
Phillips,R. J., N.H. Sleep,andW. B. Banerdt,Permanentuplift
in magmaticsystemswith applicationto the Tharsisregionof
Mars, d. Geophys.Res.,95, 5089-5100, 1990.
Phillips, R. J., R. F. Raubertas,R. E. Arvidson, I. C. Sarkar,R. R.
Herrick, N. Izenberg, and R. E. Grimm, Impact cratersand
Venus resurfacing history, d. Geophys. Res., 97, 15,923-
15,984, 1992.
Plescia,J. B., and R. S. Saunders,The chronologyof the Martian
volcanoes,Proc. Lunar Planet. Sci. Conf., 10, 2841-2859,
1979.
Plescia, J. B., and R. S. Saunders,Estimation of the thicknessof
the Tharsislava flows and implicationsfor the natureof the
topographyof the Tharsis plateau,Proc. Lunar Planet. Sci.
Conf., 11, 2423-2426, 1980.
Pronin, A. A., and E. R. Stofan,Coronaeon Venus: Morphol-
ogy, classificationanddistribution,Icarus,87, 452-474, 1990.
Rogers,G. C., Oceanicplateausas meteoriteimpact signatures,
Nature, 299, 341-342, 1982.
Ryan, M.P., The mechanics and three-dimensionalinternal
structure of active magmatic systems: Kilauea volcano,
Hawaii, d. Geophys.Res., 93, 4213-4248, 1988.
Saunders,R. S., and 26 others,Magellan Mission Summary,d.
Geophys.Res., 97, 13,067-13,091, 1992.
Schaber,G. G., Lava flows in Mare Imbrium: Geologicevalua-
tion from Apollo orbital photography,Proc. Lunar Planet.
Sci. Conf. 4, 73-92, 1973.
Schaber, G. G., K. C. Horstman, and A.L. Dial Jr., Lava flow
materialsin the Tharsisregionsof Mars, Proc. Lunar Planet.
Sci. Conf. 9, 3433-3458, 1978.
Schaber,G. G., R. G. Strom, H. J. Moore, L. A. Soderblom,R. L.
Kirk, D. J. Chadwick, D. D. Dawson, L. R. Gaddis, J. M.
Boyce, and J. Russell, Geology and distribution of impact
craterson Venus: What are they telling us?,d. Geophys.Res.,
97, 13,257-13,301, 1992.
Schubert,G., R. E. Lingenfelter,and S. J. Peale, The morphol-
ogy, distrbution, and origin of lunar sinuous rilles, Rev.
Geophys.SpacePhys.,8, 199-224, 1970.
Schubert, G., D. Bercovici, and G. A. Glatzmaier, Mantle
dynamics in Mars and Venus: Influence of an immobile
lithosphere on three-dimensional mantle convection, d.
Geophys.Res., 95, 14,105-14,130, 1990.
Schubert,G. G., R. G. Strom, H. J. Moore, L. A. Soderblum,R.
L. Kirk, D. J. Chadwick, D. D. Dawson, L. R. Gaddis, J.

Boyce,
andJ. Rt•ssell,
Geology
anddistribution
of impact
craterson Venus:What are they telling us?,J. Geophys.Res.,
97, 13,257-13,302, 1992.
Schultz,P. H., and P. D. Spudis,Evidencefor ancientmare
volcanism,Proc. Lunar Planet. Sci. Conf., I0, 2899-2918,
1979.
Schultz,P. H., andP. D. Spudis,The beginning
andendof lunar
marevolcanism,Nature, 302, 233-236, 1983.
Schultz,P. H., R. A. Schultz,and J. R. Rogers,The structureand
evolutionof ancientimpactbasinson Mars, J. Geophys.Res.,
78, 9803-9820, 1982.
Scott, D. H., and M. H. Carr, Geologic Map of Mars, scale
1:25,000,000, Map 1-1083, U.S. Geol. Surv. Misc. Invest.
Ser., 1978.
Scott, D. H., and K. L. Tanaka, Geologic map of the western
equatorialregion of Mars, scale 1:15,000,000,Map I- 1802A,
U.S. Geol. Surv. Misc. Invest. Ser., 1986.
Self, S., T. Thordarson,L. Keszthelyi, G. P. L. Walker, K. Hon,
M. T. Murphy, P. Long, and S. Finnemore,A new model for
the emplacementof the Columbia River Basalt as large,
inflatedpahoehoesheetlava flow fields, Geophys.Res. Lett.,
23, 2689-2692, 1996.
Senske, D., G. G. Schaber, and E. R. Stofan, Regional topo-
graphicriseson Venus:Geologyof westernEistla regionand
comparisonto Beta Regio and Alta Regio, J. Geophys.Res.,
97, 13,395-13,420, 1992.
Smith,B. A., and 21 others,The Jupitersystemthroughthe eyes
of Voyager,Science,204, 951-972, 1979.
Snyder, C. W., and V. I. Moroz, TelescopicObservations:
Visual, Photographic,Polarimetric,in Mars, editedby H. H. Surkov, Yu. A., et al., Uranium, thorium, and potassiumin
Kieffer, B. M. Jakosky,C. W. Snyder,and M. S. Matthews,
pp. 71-119, The Universityof ArizonaPress,Tucson,1992.
Solomon, S.C., Mare volcanism and lunar crustal structure,
Proc. Lunar Sci. Conf., 6, 1021-1042, 1975.
Solomon,S.C., The relationshipbetweencrustaltectonicsand
internal evolution on the Moon and Mercury, Phys. Earth
Planet. Inter., 15, 135-145, 1977.
Solomon,S.C., The geophysicsof Venus,PhysicsToday,48-55,
1993.
Solomon, S.C., and J. W. Head, Mechanismsfor lithospheric
heat transporton Venus: Implicationsfor tectonicstyle and
volcanism,J. Geophys.Res.,87, 9236-9246, 1982.
Solomon,S.C., and J. W. Head, Heterogeneitiesin the thickness
of the elastic lithosphereof Mars: Constraintson heat flow
and internal dynamics,d. Geophys.Res., 95, 11,073-11,083,
1990.
Solomon, S.C., and J. W. Head, Fundamental issues in the
geology and geophysicsof Venus, Science,252, 252-260,
1991.
Solomon, S.C., R. P. Comer, and,J. W. Head, The evolution of
impact basins: Viscous relaxation of topographicrelief, J.
Geophys.Res., 87, 3975-3992, 1982.
Solomon, S.C., S. E. Smrekar, D. L. Bindschadler,R. E. Grimm,
W. M. Kaula, G. E. McGill, R. J. Phillips, R. S. Saunders,G.
Schubert,S. W. Squyres,and E. R. Stofan,Venus tectonics:
HEAD AND COFFIN 437
An overview of Magellan observations,
J. Geophys.Res., 97,
13,199-13,255, 1992.
Sotin, C., D. A. Senske, J. W. Head, and E. M. Parmentier,
Terrestrial spreading centers under Venus conditions:
Evaluationof a crustalspreadingmodel for westernAphrodite
Terra, Earth Planet. Sci. Lett., 95, 321-333, 1989.
Spudis,P. D., and J. E. Guest,in Mercury, editedby F. Vilas, C.
R. Chapman,M. S. Matthews,pp. 118-164, The University of
Arizona Press,Tucson, 1988.
Spudis,P. D., G. A. Swann and R. Greeley, The formation of
Hadley Rille and implicationsfor the geology of the Apollo
15 region,Proc. Lunar Planet. Sci. Conf, 18, 243-254, 1987.
Squyres, S. W., D. M James,G. Baer, D. L. Bindschadler,G.
Schubert,V. L. Sharpton,and E. R. Stofan,The morphology
and evolution of coronae on Venus, J. Geophys.Res., 97,
13,611-13,634, 1992.
Stofan E. R., D. L. Bindschadler, J. W. Head, and E. M.
Parmentier,Corona structureson Venus: Models of origin, J.
Geophys.Res., 96, 20,933-20,946, 1991.
Stofan E. R., V. L. Sharpton, G. Schubert, G. Baer, D. L.
Bindschadler, D. M. Janes, and S. W. Squyres, Global
distribution and characteristics of coronae and related features
on Venus: Implications for origin and relation to mantle
processes,
J. Geophys.Res.,97, 13,347-13,378, 1992.
Strom, R. G., N.J. Trask, and J. E. Guest, Tectonism and
volcanism on Mercury, J. Geophys.Res., 80, 2478-2507,
1975.
Strom, R. G., G. G. Schaber, and D. D. Dawson, The global
resurfacingof Venus, J. Geophys.Res., 99, 12,899-12,926,
1994.
Venusianrocksat the landingsiteof Vega 1 and2, J. Geophys
Res.,92, suppl.,E537-E540, 1987.
Tanaka, K. L., N. K. Isbell, D. H. Scott, R. Greeley, and J. E.
Guest,The resurfacinghistoryof Mars: A synthesisof digit-
ized, Viking-based geology. Proc. Lunar Planet. Sci. Conf.,
18, 665-678, 1988.
Tanaka, K. L., D. H. Scott, and R. Greeley, Global stratigraphy,
in Mars, edited by H. H. Kieffer, B. M. Jakosky, C. W.
Snyder,and M. S. Matthews,pp. 354-382, The Universityof
Arizona Press,Tucson, 1992.
Taylor, G. J., K. Keil, T. McCoy, H. Haack, and E. R. D. Scott,
Asteroid differentiation: Pyroclastic volcanism to magma
oceans,Meteoritics, 28, 34-52, 1993.
Taylor, S. R., Large-scalebasalticvolcanismon the Moon and
Venus, in Volcanism, edited by K. V. Subbarao,pp. 1-20,
1994.
Thordarson,Th., and S. Self, The Laki (Skaft/tr Fires) and
Grimsv6tn eruptionsin 1783-1785, Bull. Volcanol.,55, 233-
263, 1993.
Tolan, T., S. Reidel, M. Beeson, J. Anderson, K. Fecht, and D.
Swanson, Revisions to the estimates of the areal extent and
volume of the Columbia River Basalt Group, in Volcanism
and Tectonism in the Columbia River Flood-basalt Province,
Spec.Pap. 239, editedby S. ReidelandP. Hooper,Geological
Societyof America,, pp. 1-20,
438 LARGE IGNEOUS PROVINCES: A PLANETARY PERSPECTIVE
Trask, N.J., and R. G. Strom, Additional evidencefor mercurial
volcanism, Icarus, 28, 559-563, 1976.
Turcotte,D. L., An episodichypothesisfor venusiantectonics,d.
Geophys.Res.,98, 17,061-17,068, 1993.
Vilas, F., C. R. Chapman,M. S. Matthews(Eds.), Mercury, The
University of Arizona Press,Tucson,pp. 1-752, 1988.
White, R. S., and D. McKenzie, Magmatismat rift zones:The
generationof volcaniccontinentalmarginsand flood basalts,
d. Geophys.Res.,94, 7685-7729, 1989.
White, R. S., and D. McKenzie, Mantle plumesand flood basalts,
d. Geophys.Res., 100, 17,543-17,585, 1995.
Whitford-Stark, J. L., and J. W. Head, The Procellarum volcanic
complexes:Contrastingstylesof volcanism,Proc. Lunar Sci.
Conf, 8, 2705-2724, 1977.
Wilhelms, D. E., Mercurian volcanismquestioned,Icarus, 28,
551-558, 1976.
Wilhelms, D. E., The geologichistoryof the Moon, U.S. Geol.
Surv. Prof. Paper 1348, 302 pp., 1987.
Wilson, L., and J. W. Head, The formationof erodeddepressions
around the sourcesof lunar sinuousrilles: Theory, Lunar
Planet. $ci. Conf., 11, 1260-1262, 1980.
Wilson, L., and J. W. Head, Mars: Review and analysis of
volcanic eruption theory and relationshipsto observed
landforms,Rev. Geophys.,32, 221-263, 1994.
Wilson,L., andK. Keil, Volcaniceruptionsandintrusions on the
asteroid4 Vesta,or. Geophys.Res.,101, 18,927-18,940,1996.
Wise, D. U., M.P. Golombek,G. E. McGill, Tharsisprovinceof
Mars: Geologic sequence,geometry,and a deformation
mechanism,Icarus, 38, 456-472, 1979.
Yingst,R. A., and J. W. Head, Lunar mare depositvolumes,
composition,age, and location:Implicationsfor sourceareas
and modes of emplacement,Lunar Planet. Sci. Conf., 25,
1531-1532, 1994.
Yingst,R. A., and J. W. Head, Spatialandarealdistribution
of
lunarmaredepositsin Mare Orientaleand SouthPole/Aitken
Basin:Implicationsfor crustalthicknessrelationships,
Lunar
Planet. Sci. Conf., 26, 1539-1540, 1995.
M. F. Coffin, Institute for Geophysics,The University of
Texas at Austin, Austin, TX 78759-8397 USA.
(mikec•coffin.ig.utexas.edu)
J.W. Head, III, Department of Geological Sciences,Brown
University, Box 1846, Providence, RI 02912 USA.
(James