A NEW EVOLI.

NIONANY LAW Leigh Van Valen Department of Biology of Chicago The University 60537 Chicago, Illinois ASSTRACT: A11 groups for which data exist go extinct at a rate that is constant for group. given When this is recast in ecological for:n (the effective a group of organisms d.eteriorates at a stochastiany of homogeneous environment exist although a fev are possible. tlefinite exceptions rate), no constant ca11y very some broad categories ancl vary within are similar Extinction "ates A of new unit rates for discrete inhablted.. of area regularly with size Laws are appropriate in evolutionary is introduced.. phenomena, the macarthur, pred.ictions. The Law Extinction than correet of more Truth needs biology. A ncnand. of taxa. slgnifieance eomparability is evidence for eeological the law invokes mutually optima incompatible to explain Markovian hypothesis can fluctuation vhich A results self*perpetuating zone. an ad.aptive vithin gnrns fhe aspect of theory. zero-sun an unstudied of terms be stated. in is the e,mount hypothesis can be derived from a view that momentary fitness The hypothesis rernain eonstant in totalwhieh anount. of control of resources, of two and. eventg only eomponents that fitness has that long-term inplies pattern of largely hypothesis explains the observed rare. The mutualism are molecu]-ar evolution.

Introduction I During a study (Van Valen, submitted.) on the effects of extinction was l using oversimplified.. modeL was It assumed no that a to show vanted with age of the group, antl f thought of extinction of probability correlation groups should d.ie out first. A test using data general1y more vulnerable that (1953) that the to my astonishment assumption was showed. from Simpson generally I d"id. not it cases. believe could be in these reasonably correct in cases more detail-. The and other assr:nption these true and so tested Others (unpublished results) proved to be consistent with all available data. ind.ividual taxa. I will present a more fincting for this have now confir:ned paper present is contlensed.. the extend.ed"treatment efsewhere; The Evidence of the survivorship curve of population Ttre method is an application plot (includ.ing It a is simple of the proportion of the d.emography). ecology for various interval-s. In this case the sa,mple sanple that survive original is the set of all known subgroups of some larger group, no matter when in A logarithmic ord.inate, stand.ard. absolute time each subgroup originatecl. the property slope the that of gives curve at any age is the in eeology, extinction at probability of that age. Simpson (f9\\, proportional to the survivorship taxonomic curves but used an 1953) compiled two well-knovn ordinate (f-)+). arittmetic (fig;.-1-5) for over 25'OOO subtaxa show aL:nost unifotrn The results for effects attributabl-e except taxa to sa,npling error for extinct linearity (216). most at the is noticeabl-e bottom graphs, where error of the Sa.mpl-ing Evol. Theory 1:1-30 (.lulY 1973)

DINOFLAGELLATE CYSTS
(EXTINCT SPECIES)
I

100

CYSTS DINOFLAGELLATE
'.. _aaaa^ (ExrtNcr GENERA)
tta" a r aaaaa.

a a a

10

tttaaaa

o
50 M.Y.

l"ll

50

100

M.Y. 150

DIATOMS

1oo{i
t.

(GENERA)

r v ^ ^ x o o ou

o o oo

E X TrfrtC T, C EN TR A LE S . | 'VIi l G.C EN TR A LE S o PE t{i l ALE s x tvrno,

oa

COCCOLITHOPHYCEAE
(E X TIN C TGE N E R A )
a a O3 taa a oaaaaa otttaa

20

40

M.Y.

CENTRALES
(SPECIES)

L-qqL^
-o -

TorALEXTIt{crr
EXCEPT LATE PERMIANi EXTIIICTION$ ^

FORAMINIFERA
100 " " . ...
(E XT IN C TS PE C IE S)

X PLANKTONIC EXTINCT o BENTHONIC EXTINCT o LIVING

oa

to o o r

ol o20
Fig. 1. Taxonomic

M.Y. 30 survlvorshiP

200 curves f o r p ro t is t s . o

M.Y. 600

(EXTINCTFAMILIES)
tt, ta lX t',

BRACHIOPODA
a

BRACHIOPODA
(EXTIi.ICT GENERA) PERMIAN - JLATE IE X TIN C TION S .OTH E R A R TTC U LATA x OTH E R IN A R TICULATA
t8 lo t, xl x a ooo oo

oor

XCLUDING ATE PERMIAN XTINCTIONS

xx tx .r xxx ax a

?o

' ?oot
t

...

xx

ooo xX x ooo

O
""f

PELECYPODA
(GE N E R A )

M.Y. 100 '{'** \*,

\ e ** L.

200

100

r a a a a ra

.EX T IN C T . L IV IN G
O- t a a t- t

PELECYPODA
(FAMILIES)

BRYOZOA
(FAMILIES)

110

-'
x'o.aw

. . EXTINCT , LIVING
a

Ff-Frt!4,

-"7
200 M.Y. 400

NAUTILOIDEA
a

100
t\

RUDISTS ( HIPPURITACEA)
(GENERA)
".o \-

100

ttt

. EXTINCT GENERA XEXTINCT FAMIL]ES

10

"**-

^* q *_

?rf
e_
a

rALL 'rrw , EXTTNCT BEFORE

. ll@-i

'MAesrnrcxrtan
20

I
100

aI x

40

M.Y. 60

-]tO 200

M.Y. 300

AMMONOIDEA
t
{o)

ARCHAEOGASTROPODA
t
r ^?: . ^i9
o o'o ^o t .. Xr ^?. o o o o "i..

l ruoruoPLAcoPHoRA
(GEN ER A) . a L L E x rrrc r (ExrtNcrExcEPT
+ L AT E - JM | DDL E ^ IPERM IAN { Exr lNcr lo Ns o L IVING a xa o xo x
aa I-

. FAMILIES XGENERA

xx.
oo oaXr xl ooo"x aa ooo

x
I

a
I.

"::.

o
F iC. 2.

200

M.Y. 300 curves

100

M.Y, r50

Taxonomic survivorshiP

for Mollueca and BrachloPoda.

too
ro
a

PTERIDOPHYTA (SENSU LATO)

(FAMILIES)

GRAPTOLITHINA
. (GENERA)

100

xxrr:x'xxx xxxxxxx
aa aa a . aa aa aaaa
t'l

.A LL

xx

: E?llllft
xxxxxxr xx x x r x

r I ^I
'

a x ol o xxxo

xIPENNSYLVANTAN lEXTlNCTlot.l
r " " aaaaa oa

Iexcepr(FTNAL)

roo

200 M.Y. 300

100 M.Y.

ECHINOIDEA
(FAMILIES)
XXX.
a-

-t...5xxxrl <
.X a

xx
tt.

\ M.Y.

(GENERA) . TOTAL EXTINCT EXCEPTI r EXTINCT ^'^ PERMIAN IX OOI LATE ) EXTINCTIONS -l?. o LtvtNG o6t3
^ lo

ZOANTHARIA

MALACOSTRACA
(FA MIL I ES) S Itt* *
ta a-o o a taa aa XXx x x x x XXX X x xx xx a

xx

o' @O0(...o

. EXTINCT XLIVING

..

xxx I

200 M.Y. 400 100 M.Y.

. . x x

MALACOSTRACA
(GE N E R A )
a oo Xao Xa x a ^^ Xl o taoo

""!

10

o LIV IN G . . .TOTA LE X TIN C T xx. xEXTINCT DECAPODA

oo

o
Flg. 3. Taxonomic survivorshiP

o
curves for
pl " ants

50

r'ti '

'r@

M.Y. r50

and i nvertebrates.

OSTEICHTHYES
(F AMIL IE S}

100 M.Y. o 200 M.Y. 300

OSTEICHTHYES
(exltNcr GENERA)

OSTEICHTHYES
(EX T IN C TGE N E RA )

100

1-r-r-+-e--J

M.Y. 150

REPTILIA
(FAMILIES) TOTALEXTINCTo EXTINCT EXCEPTT x LATE CRETACEOUSI EXTINCTIONS) LIV IN Go

. E X T I NCT XL I V I NG
20 M.Y. 30

loo
aaaa

(E XT IN C T GEN ER A )

PROBOSCIDEA

EDENTATA (XENARTHRA)
(EXTINCT GENERA)

10

aaa

aaaaaaa tttttttt aa aq
I

I
o Flg. 4. 10 M.Y. 20

o
curves for vertebrates. plots. lndlvidual

10

M.Y. 20

TaxonomLc survivorship ordersrr are those nith

rrMajor therian

1000

+ HYAENODONTA CARNIVORA
(GE NE RA )
aaaaa

r. -".xxxxx xxx - o.
oa a

'x EXTINCT LIVING

10

xxxxx

t.

*rrrri.

xx.xx
aa
I

1000

o
....
a a a

10

20

M.Y. 30

ARTIODACTYLA
(GE N E R A )

100

xxxx

"..

-...
tta

'E XT I NCT
xLlVlNG

100 .... ..
aa

POLYPROTODONTA
(EXTINCT GENERA)

ttaaaataaa XXXxxxX r aaa aa

I' I
tta

1-f-_

10

20

'l

M.Y. 50

o
100

10

20

M.Y. 30

1000

PERISSODACTYLA
(E XT IN C T GE N E R A )
aaaaaa

....
...

(EXTTNCTGENERA)

P RI M A T E S

100
tOr -a aa a a a a

10
a aa aa aaaaa aaa

10

'i
M.Y. '15 M.Y. 30

to

20

rl

loo

...

+ NOTOUNGULATA LITOPTERNA
o
oa

(GENERA)

O3

ttaa

10
aa

1 F ig.

o
5. cr.lrves for manunals. For Primates, Taxonomic survivorship Madagascar genera are omitted because the island lacks prePolyprotodonta lncludes caenotestoldea. Pleistocene fosslls'

l"

A NEI^IEVOLUTIONARY LAI^I weight, and at the left a few taxa have disproportionate sid.e, where inaccuracies (T). Usually, the more taxa included the cl-oser in dating are most ircportant is the approach to linearity. of the d-istribution taxa linearity For living requires both constant It further requires that both be more and constant origination. extinetion nearly constant over absol-ute time than does a d.istribution for extinct taxa, which nonnally spans many overlapping half-l-ives of taxa, and that probability of discovery not d.ecrease appreciably vith age of strata. It is therefore (8). that many even of these d.istributions are linear surprising of the method.. Most survivorship is not an artifact Linearity curves for Most smal-l ind"ividuals are either markedly concave or markedly convex (9). passerine bird.s are exceptional in having l-inear survivorship curves. The (biotogically and stratigraphica[y) very wide d.iversity of groups plotted. here argues against any kind of special artifact. The sources of data usually d.o not distinguish between real extinction by evolution of one taxon into a successor of a lineage and pseudo-extinction The latter proves to be negligible. Most taxa which taxon (Sinpson, 1953). give rise to suecessor taxa continue in their original form for an appreciable For farnilies of mannals, for which I am familiar period. after the branching. estinate is 20 per cent with the phylogeny, a maximum (and surely inflated.) estimate is ! per cent. Anmonites (!) give a more likely pseudo-extinction; for fa.rnilies). A1l- other available 5 per cent (I2 of l8B taxonomic extinctions The pattern of constant extinction remains phylogenies give similar results. are included or excluded.. AdditionalJ-y, r:nchanged whether pseudo-extinctions usually implies the end of an that even pseudo-extinction it is plausible Pseudointo the hypothesis given below. ad.aptive mod.e and so woufd fit are probably more eomnon at lower taxonomic l-evel-s, d.espite the exbinctions counter-claim made or inplled by Ruzhentsov (rg5S), MacGillavry (1958), and Eld"redge and Goul-d.,1972) that they do not occur for species. Eldrefue (lgtf; with a descendant taxon viol-ates Any example of an ancestral taxon co-existing systematics (Hennig, 1966). As noted, such a basic premise of cladistic exa,mples not merely exist but greatly predominate. Apparent Exceptions

There are some real and some sptrrj-ous exceptions to constant extinction taxa are not directly cornparable As noted above, the ages of living rates. Sa.npling error al-so causes d.eviations from taxa. to those of extinct linearitY. of some of geologic time have had massive extinctions Some short intervals became extinct kinds of organisms (Newell-, 1963, 1967, 1971). A11 graptolites in the Pennsylvanian and al-rnost all stony corals (Zoantharia) aia so in the sort of event from the usual- process a different This is clearly l-ate Permian. In the and does not fit the general explanation given below. of extinction in the ad.aptive zone of stony corals was aL:nost everything late pernian, If was demolished for a while. The ad.aptive zone itself el_iminated (10). from events within an as being different ve eliminate such extinctions in parts of Fi5.. 1-l+)rlinearity is never adaptive zone (d"one supplementarily (ff). increasea When, as with corals, with as ancl sometimes, reduced numerous to at such a time are suffieiently brachiopod. genera, extinctions than the ages of less for: others, as for is slope the separately, plotted be reason. today and for the same trivial genera living Two of coral-s at the famlly level-. There is no accepted cfassification plotted One is are in 6. Figure classifications orthogonal fess or more the other convex. This gives some evidence that the Treatise linear,

1000

x x a a

TRILOBITA
(GE N E R A ) ORIGIN CAMBRIAN ORI EARLYORDOVICIAN I^ ORIGIN LATER.
a

loo X
-ool-

- "

, tlr

TRILOBITA
(FAMtLlEs)

100

l tt.

10

ol

o
aa

x'
Xa a

10

x o ORIGIN LATER x x x XORIGIN CAMBRIAN t

too
100

t""

M.Y.

o
100

50

r "l

M.Y.

100

(THERIA) MAMMALIA
(EXTINCT FAMILIES)

( T HE RI A ) MAMMALIA
(FAMILIES ORIGINATING BEFORE LATE EOCENE)

10
rl

o
-

20

40 M.Y.

a?
I

60 100
( f ) oooooooof , 6 t x
ll^

'.
taa a ooooo^^'aa

(oRtGlNATtNG tN PALEOZOC)

GASTROPODA

xxxxxi l ooooooo

^ ' , , t llriiilo o o o o x x x rx P l? g e
t... ^tt""aaa t..qD

--rttdx

10
. GEN E R A

a xxx EXTINCT(TREATISE) " O'! O ( TREATI SEI LI VI NG ooxx (FOSSIL xEXTINCT RECORD)

100 M.Y. 200

O ALL FAMILIES .....o.. EXCEPT ^IP EMILIE R MIA N ' JFA SX TIN C TION S E

200 M.Y 300

'

J-

100

(GENERA ORIGINATING IN PALEOZOIC) x .x x x

a

FORAMINIFERA

txxx*t
ol

^xXxxxxxxxxxxr ^a a ttaa

X ALL . EXTINCT
aaaaa aaaaaaaa

o
F ig. 6.

Taxonomic survivorship curves for apparent exceptlons See text. Zoantharia families are from very different

to linearity. c lass i ficat ions .

A NEI^IEVOLUTTONARY LAW

is ecoJ-ogically more realistico but a decision must ultirnately classification come from phylogeny. 6). As shorm by plotting Manmalian fa.mil-ies also give a convex curve (fig. (extinct origlnating before the late Eocenen hovever, and. living) only fanilies of the short duration of the Cenozoic in relation to the this is an artifact Living fa^noilies seem similar to extinct oneq because durations of many fanilies. This is al-so true for gastropod.s d.eter':nine a single linear curve. they jointly but not for foraniniferans. ta:ra from most extinct This d.ifference in fora,minj-ferans of some living ft is ind.ete:minabl-e ones also occurs elsewhere, notably in the Pterid.opffia. from these d.ata whether reaf exceptions are involved or whether the deviant ad.aptive zones as not to interact appredifferent taxa occupy sufficiently Simpson (fg\t+) noted this phenociably with the maJority of related taxa. evoluti.on. menon for pelecypod.s and mad.e it the basis of his bradytelic Deviant adaptive zones are obvious for genera of most of the d.ifferent peleeypod.s mammal-ianord.ers (Van Valen, I97I) and. al-so for rud.ists, coral-like which forrned reefs i-n the late Cretaeeous and seem to have caused the extinction Their extinction rate is greater than of many coral-s while d.oing so (fe). that of no:maI pelecytrlods o althor:gh also greater than that of genera of corals, (f:). similar which may not be taxonomically give ostensibly generao but not fanilies, concave Anmonite and nautiloid (fig. approximate classes grouped more homogeneous into When 7)' crlrves. branches This also applles to the lengths of the tentinal results. linearity of the phylogeny of e.nmonite fanil-ies (ll+). (after each final branch-point) has a similar effect for genera as Removal of the Late Devonian extinctions plotted. I d.id. not try to subd.ivicle the not is but branches tenninal for pseudo-extinction is more cotlmon for that further suspect I nautiloids. genera than is true elsevhereo which would nmmonite and perhaps nautiloid increase eoncavity by increasing short-lived. ta:ra at the expense of longer-lived lineages. The need to separate Pafeozoic and Mesozoic ammonites shows that there is The extinction rate is definitely real exception here. at least a d.escriptively group. The sa,me is true for of the existence the not constant throrrghout Linearity in the Ordovician. being (fig. here the separation 6), trilobites caused either change' but it what not know I d.o again hold.s for each segment. from one d.escended Mesozoic ammonites alf may be refevant that effeetively In declined greatly in the Ord.ovician. paleozoic lineage and that trilobites the to corresponds extinction rates for time found. each case the d.ivision (U.)' greate* separation in the group on other criteria and incomplete collecting bias and preservational Moreover, because of taxa. However, than long-1lved l-ess often be found. taxa will stuctyn short-1ived for equally frequent groups the effect of this bias will be a red.uction in observed longevity by a constant absofute ainount. This will leave linearity have a greater and even the slope of the curve unchanged.. Rarer groups will groups. Any effect of than cotmoner longevity expected. reduction in observed and I longevity, with is correfated rarity whether this property clepends on clata. know of no relevant rates produces Any combination of subgroups with unequal constant extinction by thls can be estimated of inequality a.mount The a concave resul-tant curve. regions in the manifests most itself concavity any concavity, but unfortunately exceptions that sufficient is nevertheless Linearity sanpling enor. of greatlst as with echinoid families' Abrupt end.s to tiistributions, must be rare or slight. range of tne stratigraphic fa,nilies) ma:nmal when (as we have seen with ;;;.* become already taxa long-1ived having of possibility the group inplnges on the logaritha with ad.dition or muttiplication is unaffeeted. by Linearity extinct. (abscissa) constant' mic (ordinate) or Lrithmetic

10

AMMONOIDEA
(GENERA) . MESOZOTC (MESOZOIC EXCEPT ' IMA E S TR IC H TIA N (EXTINCTIONS O PALEOZOIC

t*-'a8
uxegsroa

o a
-E

M.Y,

curves for annonites. See text. Fig. 7. Taxonomic survivorship the lower left is for ter-minal twigs of a phylogeny of fa,nilies.
tt**

The graph on

We see that the exceptions are either spurious, rare, slight, doubtful, general The pattern is therefore sufficiently or exceptions only in part. pecuthat the minor exceptions are best explained by unusual circurstances liar to each case. There is a strong first-order effect of linearity, and it is this, rather than its perturbations by special and diverse circrrnstances, that deserves prlmary attention. Related Phenomena The constancy of extinction in terms of survivorship does not imply Any form of survivorship constancy over geological time, and vice versa. curve has a mean, as d.oes any pattern of extinction rates over absolute time, and this is the only formal connection between the two phenomena. fhere may nevertheless be a d"eeper causal connection. I give some extinction and origination curves over geological time to illustrate their variability in the measurement framework proposed. here (Fig. B). rate of mnmmalian There is an extraord.i.nary exponential d.ecrease in origination femilies, by two orders of magnitud.e, from the beginning of the Cenozoic. An inverse phenomenon occurs in d.iatomso where the species of Pennales (a largely as in benthonic groups, unlike the Centrales) have increased exponentially, (f5; Fie. 1). We can the 1og phase of bacterial culture, in the sane intervaf l-ook at Figure 8F inversely: there is a nearly linear increase in the proportion of fanilies that had. originatetl by a given time a^ndthat are now extinct. (f1) trave originated Large foraniniferans about hl tines independently from smaller fora.miniferans, almost always from much smaller ones. About 33 of these clades have existed. since the end of the early Cretaceous, when an approximate equilibrir:m Figure p shows that the numbers of was establ-ished. clad,es and genera present simr:ltaneously have followed nore or less Iog-no:mal

1200

r;
I

1100 I 1000

t;

CENTRALES PE NNALES ORIGINATION EXTI NCTI O N

11

BENTHONIC. FORAMINIFERA
(EXTINCT GENERA)

,t

lco 1l o{
roo ]
100 300 M.Y .8.C .200

or

50

zoo--l . exirrucrroru I
o ORIGINATION

--

I il, . "li/
400

= 20 {.

40 30 M.YB .C . 3 0 0 M .Y.B .C2 . 00

20

10

MA MMA LIA
( FAM I LI ES)

60 40 20 M.YB.C

50

40 30 M.Y .B .C .

20

10

Fig.

8.

rates for species of diatoms. and extlnction A: Orlgination B: Extinctlon rates for genera of benthonic foraminiferans. rates for famllies of anrnonites. and extlnction C: Orlgination rates except for rates precede hlgh extinction High originatlon the Permlan and latest Cretaceous. D3 Extinction rates of rates for mammalian mammallan famiLy lineages, and orlgination E: Origination rates for all mammaLian famlLles now surviving. famllles (new famllles per famtly ln previous age per unit time) . (cumul-ative recent families originated)/ F: For manrnalian families, (cumuLative total families originated).

L2

LARGE FORAM INIFERA

. EXTINCT CLADES XEXTINCT GENERA
a

10 15 20 25 30 NO.SIMULTANEOUSLY PRESENT 100

?-?-?-?-?-

M.Y. 150

Survivorship curves for genera and clades of large foraminifers,ns, Fig. 9. and present after the of m:mber of genera and clades simultaneously d.istribution early Cretaceous (Rrlian).
**tt

Survivorship since that time. d.istributions very heterogeneous, but this taxonomically group are linear within sa,mpling error and of all benthonic foraniniferans. It woul-d be useful- to knov vhether the al-so occurs at a constant rate, or if some fhis appears to represent suecessor taxa. (18).

curves for clades and genera of ecologically rather homogeneous, d.iffer from the curve for genera extinction of entire lineages taxa can avoid it by evolving an unsol-ved problem in graph theory

Measurement of Rates I propose a general unit of rates for phenomena which can be treated as d"iscrete. of an event per A macarthur (na) is the rate at which the probability MacArthur the of extinction in H. showed. importance years Robert is 0.5. !00 ecolory (U.). per t thousand. years. Let P be the probability

ma = -logr(r-gegy

(r)

(fl) one macarthur is the rate of extinctlon With respect to extinctlon, one macarthur With respect to origination, of 500 years. giving a half-life ancestor. is the oceuruence of one origin per thousand. years per potential (nna) is the rate one millimacarthur With respeet to molecular evolution, per million years (eO). giving one substitution rates of bird. species on islands that have extinction Apparent equilibriun been stud1ed (21) are 0.5 to 10 ma. Ma.rnnalspecies in the late Pleistocene of Florid.a (Martii-and. Webb, i.n press) naa a rate of regional turnover (time frori (r'is. 10). of about ? to locaI extinction) irnnigration 'ma rates for the taxa stud.ied.. The Table 1 gives estimated extinction marine organlsms sedentary marine benthos is remarkably homogeneous, and notile Me.rnnalian Senera among themselves. have somewhat higher rates less similar fa,nllles (and. Mesozoic annonites) have the highest rates, but rateg for reptlllan For everlrthlng except manmals are as high as those for manrmalian families. Groupe ln rate for fa.rnilies is about haLf that for genera. the extinctlon new adaptive zones (at least to them) usually have hlgher ratee than relatively groups. ecology ls a better predlctor of Not surprlsingly, long-established DNA (eZ), rate than is a^nor:nt of inforrnation-bearlng evolutionary ind.epend.entl-y, the extlnctlon rate of ff genera of a fa.nily go extinct on the number of contemporaneous genera per fanlLy d.epend.directly will fanilies

LAW A NEW EVOLUTIONARY TABLE l: Extinction Rates (in Ana). Abbreviations: fa.m.o fa.uaily; gen., genus; sp., GROUP fam. Fterid.ophyta d.iatoms Coccolithophyceae DinoflagelJ-ata Foraninifera benthonic planktonic large 0stracoda GraptoJ.ithina Bryozoa Brachiopoda Articulata fnarticulata Malacostraca Trilobita early ].ate Echinoidea Zoantharia Anmonoiclea Paleozoic Mesozoic Nautiloidea Arehaeogastropoda + Monoplacophora Gastropod.a Paleozoie Pelecypod.a rud.ists 0steichthyes Teleostei Holostei + Chond.rostei Sarcopterygii Acanthodii ReptiJ-ia Ma^umalia (Tueria) Polyprotod.onta Primates Rodentia Carnivora + Hyaenodonta Edentata Proboscidea Notoungulata * LitoPterna Perissodactyla ArtiocLactyla Cetacea 20

13

speci.es

RATE gen.

sp. 90 55 100

50 25 20
10

10 T L5

20 30 50 25 30
t+j

12 20 10 10 10

25 30 80 35 2' 25 35 150 30 20 20 20 50 30 35 25 30 30 150 150 220 150 120 180

1?o 120 120 200

20 75 15
h
R 'l q

20 t2

30 30

5o

L4

100
'o o o o o l ojo

10

6'
pig. lO. Regional survivorship the late Pleistocene.

'
curve for

5b

T.Y.

100
in Fl-orid.a d.uring

ma.wtalian species

* * genera in the fanily. Rough estimates new of if there is no origination ' give interval, contemporaneous each 2.9 taxa in of number weighted by the genera per fanily of Ostracod.ar 2.S for Echinoid.ea, and5.l- and5.8 species With 2.9 genera per genus respectively for pennate and centric diatoms. give rate per genus 2.6 w:ill an extinction pu" irroify, extinction independent and.3.5 for for 2.5 echinoids This ratio would" be per fa:nily. tlmes that to be time of farnily extinction I the expected. take d.iatom speeies. average With branching when the expected. number of genera per fa,nlly reaches 0.!. (origination of course have longer expected of new genera), the fa,milies will If the rate of branching increase. and so the expected ratio will longevity roughly simrrlation in the above range indicates equals that of extinction, in although a high variance aJnong fa.nilies a doubling of the expected ratio, genera per fa.nily probably increases it more. Table 1 shows that genera of echinoids and ostracod.es have about 2.5 from that a result ind.istinguishable times the extinction rate of fanilies, Because branching exists, on the assumptions of ind.epencience and no branching. of genera within fanilies seems probable. some degree of correlated. extinction Ttre observed rate for species of diatoms is only twice that for genera, much in less than the expected value of 3.5, so there i.s a strong correlation extinction here even without consid.ering branching. Contemporaneous Subgroups Extinction rate obviously depends on the area consid.ered as velL as on rate, between area and. extinction Fig. 11 gives a relation the inhabitants. Approximate linearity d.ata (Zf). holds over 11 orders of using all available the extinction rate in 1oglo macarthurs and With magnitud.e on a 1og seaIe. A the area in logln sq. km, the regression is J) = -Q.669+ 1.53 when calculated from A regression from the birds and-manmals and excluding the worl-d. fauna. somewhat. As expected, the arthropods complete data would presumably differ on Simberloffia seem to require less area for the sa^ne exti.nction rate than (ef). d.o vertebrates A continent seems to be the largest area in which organisms can interact more or J-ess as they do in any smaller area that is large enough for a population. Therefore species and higher taxa occupying smaller areas, as will often greater have an expectecl rate of extinction be the case when they originate, (fg)+5, Sinilarly, Smalf as 19h8b) noted. in than that of more wid.espreacl taxa.

15

co
-2
-4
A

ARTHROPODS B IR DS o MA M M ALS

\"

-2

o24
LOG rc AREA

Fie. 11.. Extinction (sq. lon. ).

rate

(f[nin

loglo

ma) of species as a function

of area

context for tliatons, genera with fewer species have a higher a different This is of than do genera with more species. probability of extinction of course because branching at the specific l-evel inereases the probability survival of the entire sub-tree n since the species will usually have different ecologies. For these and other, less well-documenteti, reasonso there are definabfe rate than others, and and large subgroups that have a higher extinction most frequent among younger taxa. moreover these subgroups should be relatively How can it be that the curve should. result. Therefore a concave survivorship curves are nevertheless linear? assr:mptions and. eompensatory incorrect There are two possibilities: ft nay be that, by the time taxa have an appreciable probability feedback. record, they have the sa.memean area oceupied as of appearing in the fossil fhis sort of rapid increase may also be true, although it doesnrt ol-der taxa. seem as 1ikeIy, for nr:mber of speeies in a genus. Enough data.are probably point. Conpilations by Sloss (f950) ana availabfe to resolve the latter (e.e. (fg:S)n 19h6) on d.iatomso and" d.ata by Berggren Smallf s studies Sinpson suggest that there is indeed (tg6g) and others on planktonic foraminiferans, taxa approaches its of subordinate usually an appreeiable lag before the number papers by no means conclusive are The kind.s of evidence in these maximum value. a threshol-d. may well be there Fr.rrthermore, on the present point, however. population size in (tite others for and those folnd by MacArthur lryfZl assured is virtually which survival above models of colonization) simplified of possible inad.equacies Other likely. is and below which rapid extinction they that deteeted or to be too smaLl the assr:mptions, that the effects are of taxa, are clearly unrealistic. affect only a small proportion curves is avoided by the concavi.ty in survivorship Even if an initial anong major heterogeneity remains a still rapid. ehanges invoked above, there genera, and nr:mber of or species to area, the surviving taxa with respect There must then be some kind. of compensaother faetors relevant to survival. Two kind.s of feedback of linearity. degree give the observed. tory feedback to age and area (or nu:nber interaction between an be possible. There might seem effects of area of species) such that older and younger toca have d.ifferent inplauNo simple mod.el fits this biol.ogically of extinction. on probability However, if for new ta:ca small area and low nr:mber of sible alternative. concavity the expected initial species were not ha:mful , or as har"nful as later, would be reduced. or eliminated.

L6

L. VAN VALEN The other kind. of feedback is by the continual appearanee at a1l- a,ges of taxa of low area and. other characters giving greater susceptibility to extincfn fact this could. be regard.ed.as an aspect of the extinction tion. process: taxa would. become more susceptible at a constant average rate per taxon per This transforuation time. need. not be by feedback per se, but it d.oes require that the production of greater susceptibility be part of the sarne process as the extinction. There j.s no requirement that eaeh subord.inate taxon be equally likely at any given time to beeome more susceptible, but the overall controlmust eompensate for survival by greater susceptibility at some other time or (given linear for some other species. fhis second. aLternative is plausible ft pred.icts that, at least after a short sr:rvivorship) but largely untested. period o criteria of susceptibility initial will have a steady-state distribution time ind.ependent of the age of the taxon. i-n survivorship Smal-l (fgl+6, l9\7, 191+8arc) found. such a steady-state distribution in real- time for generic sizes of d.iatons through the Cenozoic. An Evolutionary Law

The effective environment (el+) of the menbers of any homogeneous group of constant rate. organisms d.eteriorates at a stochastically This law goes a bit beyond the observations in postulating the cause to like nucl-ear rather than intrinsic d.ecay. There be extrinsic is of course n much other evid.ence (Sinpson, 1953) for such a step. The law is also stated timeo although lt appties to in terms of reaf time rather than survivorship A more neutral statement is that extinction both and. ties them together. in constant rate. any adapti-ve zone occurs at a stochastically ttHomogeneousttis a necessarily arnbiguous word., and its meaning in a partlcj.rcunstances and on the d"egree of cular case must depend. on the particular precision desired. Pal-eozoic and Mesozoic amrtonites nay d.iffer sufficiently to be nonhomogeneous, but the entire marine benthos could be treated. together This d.oes not lead to circularity; aLmost as well- as subdivid.ing it. no subdivision of a group with racial senescence (25) would give a set of linear The homogeneity required. (and so its verification) curves. survivorship is factors (Van Va1en, entireJ-y ecologicaf and is in terms of ultimate regulatory Mice and fruit-eating flies night be honogeneous 19?3) of population d.ensity. flies, although but not with blood"-sucking a scaLing problem other with each there always could exists a degree We say that remain. of homogeneity vould measure a4d try to it true, ind.ependently. the law is Homogeneity at which respond to to the of abil-ity d.eterioration; equality d.oes not inply but of such abilities would. disprove the counterexamples to a eonstant d.istribution form as a universal phenomenon (e5) 1aw in its first a,re observable Like a^ny 1av (ZT), the effects of the lav of extinction (here, persistence of the wid.th of the ad.aptive only under eppropriate cond.itions If it ls not universal, zone). Un1ike many laws (e8), it nay not be universal. its domain in an obJeetive manner. Such limitait should be possible to limit tion should. d.erive from an understand.ing of the causal basis for the law. biology or for ft is not fashionabfe to speak of l-aws in evolutionary I think this is baged on both a misunderstand.lng proeesses generally. historical to poorly of actual processes (A) and on an over-reaction of the regularity Laws are propositions that specify fonrulated. laws of earlier workers (30). cond.itions for a result; Sven the condiiions, the result willsufficient occur, although some of the eonditions (ttre bounds of the d.onain) may be of a law is of course a different matter Ttre clegree of confirnation inplicit. (or represents) a law in this sense. Any general from whether a proposition is statement of the nature of a causal process states a faw (ff1.

A NEW EVOLUTIONARY I.AW The Red Queenrs Hypothesis (3e)

T7

fhe probability of extinction of a taxon is then effectively ind.epend.ent of its age. This suggests a randomly acting process. But the probatifity ts strongly related. to ad"aptive zones. This shows that.a randomly acting process eannot be operating unifor:nly. How can it be that extinction occurs rand.omly v"ith respect to age but nonrand.omly with respect to ecology? We can consi-der the situation in te:rns.of an ensemble of mutually ineompaIt is selectively tible optima. advantageous for a prey or host (includ.ing plants) to d.ecrease its probability of being eaten or parasitized.. ft is often selectively ad.vantegeous for a predator or parasite species, an4 much more often for a predator or parasite ind.ividual, to increase its expected rate of capture of footi. It is selectively advantageous for a competitor for resources in short supply (fooct and space in the broadest senses, and sometimes also externally supplied. ad.juncts'bo reproduction or d.ispersat) both to i-nerease its own effect on its competitors and to d.ecrease the effect of its competitors on itset-f (Sf ). Rvery species d.oes the best it can in the faee of these pressures. Probably all species are affected. importantly by them at least over intervals of a few generations. Response to one kind of pressure may well deerease resistance to some other, at that time weaker, kind.. The various species in an ad.aptive zone (Van Va1en, J.97l-), whether or not this zone is sharply d.elimited from others, can be considered together. We can assume as an approximation that a proportional a^urount w of suecessfulresponse by one species prod.uces a total negative effect oT v = w on other usually less than v for any one of these spEcieF. species Jointlyo For n speeies, the mean d.ecrement of fitness per species is v/n. For m suc"esiful (in an intervalt), responses simultaneously the *ean ToTal decr6nent per Over some interval t (in units of t), species i-s sg/g. the d.eerenent is then gob-/". To maintain itself as before, the species must increase its fitness by gnnb-/g. Since each d.ecrement generates a responser g =.g.. Most species Fa.ilount will be abl-e to recoup their loss o more or less while it occurs, but in doing prod.uee another d.isadvantage of v for the avera.ge species. so they Jointly fhis process of successive o'rerlapping d.ecrements of v may continue indefinitely. Species at feast 1oca1ly new may replace those for which the rate of environhas been too great. mental deterioration For the momentary fitness F of a rand.om species o

$ =

d q = m(w-v)
dt

For a given species, { will of course vary from time to time as its specific However, since w = I on the average, decrements and possible responses covary. = fitness in the adaptive the mean i.e. zone is constant in the long O, E(0 ) fitness average a standard the we take real average fitness will If run. f1, = of any response one speeies because above $ will bring a b. & If - fr cortnter-response from the of other species. deerenent v is the corresponding Thus, for a single even speeies, the total loacl. selecTion environmental that are long enor:gh to average out irregupressure is constant over intervals larities. anong the species at any time, and its { is of course itsel-f a variable (or the extinction determine w111 rate in the ad.aptive of that variation E) (anT an unknolrn to extent) aiffer d.o, but may in the epectral zone. Species

18

L. vAN vALEN threshold T (Reyrnent and Van Valen, 1969) of S or lts components to which they can no longer respond successfully and. at vhich they therefore become extinct. Which speci-es have which values of + and T wiJ-l change markedly with the nature of the stresses at a given time. The observed. pattern of extinction seems to inply that the stresses are sufficiently diverse that they affect most species simil-ar1y over a very long interval. Assumeo as Justified. in part by the Central- Linit Theorem, that both $ and T are normally d.istributed., r+-ith neans y',1 and.y, respectively. Then, since-toth distributions range over the sa.meset-of spedies and so have the s,tnearea, the rate of extinction is the overl-ap of the d.istributions. In the case of equal varianceg for and !o f

-r

(^) = J'9 4,
Lt

( 3)

Uncompensated. d.epartures from no:mality and from equal variances in the d.istribution will of course affect the accuracy of this expression but not the dependence ofil on the real area of overlap. On this self-contained system we must now impose the physical environ:nent and irregular biotie perturbations. Almost all changes in the physical environment of an ad.aptive zone will- be deleterious to its inhabitants, either directly the establ-isiment of other species which can then survive or by pe:mitting there. Therefore regular changes in the physical environment ean be treated. as constant factors of p . The observed. constancy of fl is evid.ence that such a treatment is pernissible. clo, however, occrr, from both biotic Important perturbations anil physical They can be treated together. causes. The probability of extinction is not constant over geological time, as Fig. 8 anct the extinctions at the end. of the Cretaceous suffice to remind us. When maJor perturbations occur wel-I within the geological range of a group, as for brachiopod.s in the late Permiann Paleozoic ammonites at the end of the Devonian, or coccoliths at the end of the Cretaeeous, it happens that if we ignore the subgroups that bece.me extinct (34). then, the remainder show a constant rate of extinction There is no irnportant lasting effect of the unique event, unless total extinction occurred., nor d.o its effects extend d"etectably beyond. the subgroups prematr.rely eliminated. Newly appearing subgroups after the event continue in the sr.me way as those Smal-ler perturbations that l-ived through it. occur much more frequentlyn however (fig. 8), and. it is the very resultant of these perturbations that d.etermines the long-term cgnstancy. Therefore the effects of these minor perturbations are not ind.epend.ent If they wereo they should not be distributed. of each other over time. about a constant mean. The rate of d.eterioration of the effective envirorunent has call the property of homeostasis. what we can d.escriptively A large change in one i.nterval is compensated for by a smaIl one latero on the average, and. vice d.oes not undergo a rand.omwalk. versa; the mean itself ft is the organisms in the adaptive zone that can l-ink the perturbations across tine in that the may d.epend on the effects of those before it. effects of one perturbation are not there if it comes Species easily removed by one kind. of perturbation again soon, and may accumulate if it d.oes not come for a long time. Meanwhile are taking their own toll-, more or less indepenother kinds of perturbations dently of the first kind (35). We see here a maJor d.ifference from the usual theory of genic selection. The l-atter depends only on the current distributlon of frequencies of alleles with each other and the environment. It d.oes not depend and their interactlons

A NEW EVOLUTIONARY LAW at all on the process by whieh the current d.istribution was obtained.. fn forual languager it is a Markov process. But any process can be made Markovian by choice of a suitable leveI of analysis. With extinction we can see that a selection is appropriate. non-Markovian analysis,of This is probably true for the general case arso (35). The Red Queen does not need. changes in the physical environment, although she ean accommodate them. Biotic forces provid.e the basis for a self-d.riving (at this leveI) perpetual motion of the effective envlrorulent and so of the evolution of the species affected by it (SZ). fne Red. Queenrs Ilypothesis is a sufficient explanation of the law of but not one that is yet d.erivable with confidence frcm lower-l-evel extinction knowled.ge of the causes of ind.ivid.ual extinctions anci the nature of species There may be other sufficient interactions. explanations that I have not been enough to see, anil pred.ictions by the Red. Queen may be duplicated. inaginative Disproof of the Red Queenfs Hypothesis is by such und.iscovered explanations. possible in several ways, but f'ully adequate confirmation must await d.erivation of it from what we can reasonably regard as facts. We can go a step further by thinking of an ad.aptive landscape in a resource The a,moun+. of resources is fixed and. ean be thor:ght space (Van Valen, l-971). ge1 neutrally stabl-e in configuration, of as an incompressible supporting the ff one peak is diminished. there must be an equal total peaks and ridges. increase elsewhere, in one related peak or more unifonoly. Similarly, increase in an equal d.ecrease elsewhere. in a peak results Species occupy thi-s land.scaFe and can be thought of as trying to maxito its use and. mize their share of whatever resource is scarcest relative fhis resource will- take the role of the geI, and the momentary availability. to ihe a^mountof ge1 r.md"erits of a species wil-l be proportional fitness resource it controls). To a sufficiently areaTttre smount of the liniting momentary fitness seems this to be what natural sel-ection approximation close maximi:zes. at three levels. Species d.isplece fhe lanilscape is changing continuously, the aclaptive surface. This areas of can from be by resistance to each other predation as wel-l as other means, and some parts of the land.seape may in this Incorrpatible optirna way or others (e.S. severe weather) be under-occupied.. gel the d.istribution of Secondly, within the land"scape d.iversity. maintain (for change or herbj"vores) when the cLimatic flora changes. as with changes, (or ge1 amount the of in fact useci.) can change. This a.mount Fi.nally, the total beconing limiting, resou.rce necessary bJr constraints on can occur by another occupation of the ad.aptive zone, or by a change ln the arnowt of the sa,me fhe entire lanclscape can in this way d.isappear. resource. liniting two kinds of change ln the land.scape lead" to the the Red Qrreents The first with Hypothesis, while the third. is, without compensatory changes, inconsistent is less independent of age when extinctions caused extinction Enpirically it. The observed. d.egree of of adaptive zones are includ.ed. by maJor constrictions the third. kind of change is rel-atively unimportant independence suggests that time scal-e. We can hard.ly regard such a conclusion as on the evolutionary in the size of but it d"oes show that the question of variation established, maJor ailaptive zones is firnd'amental. Molecul-ar EVolution approxiThe Red Queenfs Hypothesis provid.es reason to expeet a long-tera proteins within ar5r of ind.ividual mate constancy in the rate of evolution It does so without ad hoc assumptions, although single adaptive zone (38).

19

20

L. VAN VALEN such assunptions could be invoked. to explain away exceptions Just as they have been invoked (Clarke, 1970) to explain away constancy by the usual selective frarroework. Constancy of the rates of protein evolution is often regarded as the most important evidenee for what Kirrg and. Jr:kes OgAg) called. non-Darwinian (-:g). despite serious misinterpretations evolution, A further pred.iction is that rates of protein evofution will be related monotonically to rates of taxonomic evotrution, to the extent that the subtaxa Some eviaence ()+1) is available in d.ifferent groups are comparable (r3rl+O). f obtained long results that rates are not always constanto and. preliminary suggest that the mean rate of protein before d.iscovering constant extinction short time while the orders of placental evolution during the relatively ()+2). Lingufa, me.mmalsd.iverged. from each other was greater than that later anil Triops a brachiopod genus present since the Ord.ovician or Silurian, species not detectably changed since t"""hiopod.) a branchiopod. (q[ cancriformiso (1), would. be expected. to have proteins more similar at least the early Triassic The to those of the eonmon ancestor than would- their more d.ivergent relatives. constant evoLutionary rates of proteins comes from evid.ence for approxinately organisms (nainly vertebratels) whieh have all evolved. appreciably since their separati-on from eaeh otherr fhe Red. Queen in her simplest gown also predicts a perhaps real phenothat iregularities menon, which Ohta and Kimura (fgff ) noted as mysterious: less caricelled. out over to be more or seem evol-ution rate molecular the of in autocorrelation. negative a seemingly long intervals by Implications of the Red. Queen

of survival Thod.ay (fg:S) propcrsed. a concept of fitness as the probability generally, we can More in future. tine the dlstant some very of a lineege to take
F=

z*

r * (t )r ( t ) a t I ---J o

()+)

(h:) at time O of survival to time !r and w(t) where p(t) is the probability (the sarne for all lineages and perhaps integrating to is a weightlng functlon l for scale) for whlch I vould choose exponential decay a*, a 1ow rate (lrh). Tlme 0 :ls the varlable present. The Red Queen propoees that thle fltnesn has only two components for whlch adaptlve zone J.s occupled. and what the probabillty a"lmoet any real lineage: have Artiodactyls cllstrlbutlon ls of new sublineages occurring by branching. largely replaced perissod.actyls in the sa.mead.aptive zone, yet their extinction It is the rates of branching that are rates are id.entical (Tables 1,2). d.ecisive. Monotreures sti11 llnger on in two isolated sutrzones, one of vhich The gradual extinction of nultihas already been invaded. by marsupials. tubercul-ates (Van Valen and Sloan, 1966) occurred. by herbivorous placentals d.iversifling lnto parts of the Joint ad.aptive zone foranerly held by multiyears after the survived. 1l or 20 million One multituberculate tubercufates. had. taken only about was eompleted (h5); the fatter rest of the extinction years. of fhe Red Queents proposal implies that extinction 10 nillion lineages (subtrees) will prove not to be constant when j.t can be measured. By does not d.eny progress in evolution. For the Red Queen, curiously, properties of group seLectlon such as thls, as weJ.J.as by ind.ividual selection, manner. It may well be that the conmunities can change in a clirectional everage Cenozoie species could. outcompete the average Ca^:nbrianone; some It is analysis. on this can probably be derived from ftrnctional informatlon

A NEI{ EVOLUTIONARY LAW TABLE 2: Extinetion and. origination rates (in orders of ma:nmals. E: early; M: middle; E}MTNCTION Epoch E. M, L. E. M. L. E. M. L. E. L. Eocene Eocene Eocene Oligoeene Oligocene Oligocene Miocene Miocene Miocene Plioeene Pliocene Perissodactyla Artiod.actyla ma) for L: late ORTGTNATION Perissodactyla Artioclactyla genera of two competing

2T

120 80 120 r20 100 170

150

1300 290
An

9o
9O 100 150 130 130 9O

3+0
l+o

9o 30 l+o 90 90
l8

6o

130 BO
t3

130 90 60 90 20

500 920 260 150 130 BO 260 90 100 320

*

7o

well known (cf. Roner ftgee]) ttrat such fi.mctional progress has occurred in vertebrate The Red Queen measures environmental d.eterioration evoluti"on. on a scale that is deter:nined. by the resistance of eontempory species to it, so the scale and real d.eterioration themselves may weJ.l change without a change in the measured cleterioration. Darrinrs example of wolf and deer exemplifies this. The RecI Qu.een proposes that events of mutualism, at least on the sa.rne trophic leveL, are of l"ittle importance in evolution in comparison to negative lnteractions, although she does not consid.er other cases where mutualism is so great that the mutuallsts function as an evolutionary unit, as with fichens. and perhaps chloroplasts. She consitiers the usual contrary view to be a result of wlshful thlnklng, the imposition of hr:man values on nonhrrnan processes. The existenee of the law of extinetion signiis evj.d.ence for ecological ficance and. ecological of tarca from species to fa,mily, within comparability any adaptive zone. We can think of the Red Queenrs Hypothesis in terms of an unorthodox game theory (l+5). To a good" approximati.on, each species is part of a zero-sum ga,ne against other species. Which ad.versary is most important for a speei.es may vary from time to timen and for some or even most species no one adversary may ever be para,nowrt. Furthermore, no species can ever win, and new ad.versaries grinningly of ga,me replaee the fosers. This is a d.irection of generalization theory which I think has not been explored. From this overlook we see dyna.nic equilibria on an immense scale, deternining much of the course of evolution by their self-perpetuating fluctuations. This is a novel way of looking at the world, one with which f a:n not yet comfortable. But f have not yet found. evid.ence against it, and it does make visible new paths and. it may even approach reality. Aeknowled.gments I thank the National Science Foundation for regularly reJecting my (honest) grant applications L972), thus for work on real organisms (cf. Szent-Gy8rgyi, forcing me Lnto theoretical Thls paper has been circulating in glzdat work. since December, 1972, and. I have given talks based on it before and after then. I thank Dr. R. A. Martin for unpublishetl d.ata and Drs. P. BillingsJ-e;rr J. Cracraft,

22

L. VAN VALEN

J. F. Crow, D. H. Janzen, T. H. Jukes, S. A. Kauffinano H. W. Kerster, M. Kimr:ra, E. G. Leigh, J. S. Levintotro R. C. Lewontin, J. F. Lynch, V. C. Maiorane, J. Maynard Snith, P. Meier, D. M. Raupo G. A. Sachero T. J. M. Sehopfo and. E. O. The Louis Bloek Fund. of the University Wilson for d.iseussion. of Chicago paid. for the preparation of the figures. Notes (f). sources: fhe taxonomic d.ata I used are from the following For vertebrates, (tg66), a few modifications from vari.ous with later work. Romer and genera and fanil-j.es of Foraninifera, For invertebrates R.C. Moore (ed.), Treatise on Inverlebrd.te Paleontology (goutaer, Colorado: Geologi.TAGA;TrAs), 1953-present, ineluding cal- soe ed..), with supplementation or (for Bryozoa) revisions (C. feictrert, Rdcord (W. S. Harland et a1., ed.s.; London: from fhe Fossil substitution Geological societyaT ffi r deliberately lgnored. later TltdTI). pertinent work (e.g. Yochel-son ltg6gl) on invertebrates. Pberidophytes The from Fossil are also Record. For and coccoliths species of Forarnini-

rers.- Bersgren (tg6g).

vvro - v r er

ror n-norlffialil-s-irJeant

(ig6l).

For diatoms,

Q) .

(:).

(!.).

(:.).

(5.).

Snall- (rg\:,191+5). sources: Berggren (l-gtZ); irhe time scal-e I used is from the following (lgeb); Everndon (rg5l+); Curtis, Savage, Jemes Everndon, and Anonynous (1970); (1966). (t955); and Kauffman Van Sloan Valen and and Curtis are the results robust to reasonable that changes in the show Experiments time scafe. Some general Various conventions are necessa.ry in any such compilation. I took d.uration of a as from are following: the tarcon the ones I used (or shortest other interval) before the first epoch the nid.dle of the quesof epoch of the last record. I the ignored. recorcl, to the mid.trLe period. taxa. accurate unrecognizable For d.ata only to tioned. record.s and the niddle period,. in unit the f of used I plotted the range as ending (unless imprecise beyond a all d.ata regardless of degree of precision A1so, summaries of period.) but used then to the precision allowable. (even too often inaccurate, as shovn by of ind.ividual taxa) are ranges rmquestioned record.s of subordinate tarca beyond the stated limits o so whenever possible I compiled d.ata at the level of genus. f knew of this deficiency for manyyears but saw no reason to pursue it, In hind.sight as I expected" the shape of the culn'es to renain concave. group progressively as such might a have one could. also expect that the as the biota around j.t evolves vhile it of extinction higher probability This woufd. give convex survi"vorship curves. d.oes not. I mad.e several- d.ozen plots of subsets In ad.d.ition to the d.ata plotted, as exclusion of a maJor extinetion of the sa,medatao using such criteria the time used. A11 important of interval or subtaxon, or restriction includ.ed in the figures given are deviations from the total d.istribution group in subta:ca of a living There is a bias in using all extinct here. reeently are stillal-ive. of subtaxa originating that the longer-lived (not, groups for long-ranging Tests show that this effect is negligible anci and obviously al-l living e.g. r for fa,milies of ma.wtals or echinoids) (Sinpson, into one useful cr:rve 1953). erbinct subtaxa cannot be eombined. Omissiona are I used. all gror:trrs for wtrich ad.equate data were available. (e.g. Insectivora), or its study due to poverty of the fossil record part small number of extinct taxa, error of d.ating being a substantial (e.g. lack of adequate compilaArchaeocyatha) or o of estimatecl tlurations tion (e.g. Gtrrnnospe:ruae).

A NEWEVOLUTIONARY LAW (1). (g). This is the reason for the apparently flat tops of some distributions. The initial figure in parentheses in distributions of most livlng taxa represents the total now aliven includ.ing those not known as fossil-s. Many living tarca in some groups are not easily fossil-izable, and. our knowled.ge of the present fauna is better than that at any other time even aside fron this. UxelnFJ.escan be found in the following: Pearl (fgZ8); A11ee, Thtersonr (191+9); (fg:S); Park, Park, and Schmidt (fgff ). Kurten Od"rrn Coral-like brachiopod.s of the Permian also becane extinet, as d.id. many other organisms superfieially less simil_ar to corals. Tests are necessarily weak, but data from several groups give no indication that a major extinction affects taxa of d.ifferent ages to a d.ifferent extent. A third (fT or 51) of the fainilies of stony corals present in the late Cretaceous beca,me extinct then, but only one of these d.id so at the no'lnr anigis at the end of the Cretaceous. The Fossil Record. famiLies are the only ones with aciequate d.ata on this point. Rud.ists diversified. throughout the late Cretaceous and abruptly d.isappeared at its close. f give several ways of evaluating comparability of taxa elsewhere (Van Valen, in press). The plrylogeny is from the Treati?g (1) and excludes ! fe.nrilles of unknown ancestry. Schinctewolf (1961-1968) has given a rather different phylogeny, but parts of it are not d.etaiLed enough for this application. A brief d.iscussi.on of possible factors in the crisis for trilobites can p. 51+. Sone other curves also probably be for:nd. in The FoSSil Record (1) ' exhibit real deviations from linearity, but the ones discussed. are the largest. this exponentiaL increase for some individual Small (fghg,l?r2).noticed genera, as Sma]l (1950) and Tappan and Loeblich (fgff) d"id for the entire group. d.efine these as having a maxj.mu:nd.ianeter of at l-east 5 m. I arbitrarily and most are discoid.aL, fusifo:m, Most are highly complex internally, or low conical. of the d.urations of all llhat is necessary is the expected. d.istribution incJ-uding parts of larger sub-trees o given a constant extincsub-trees, Harris tion rate and a branching rate that varies from time to tine. (f953, p. 32) considereti the problem for a constant branching rate and It would further be useful if some sub-trees for:nd it intractable. (extending an unknown arnowrt infinite could be ignored as effectively The point is directly beyond. some absolute d.ate such as the present). but this requires more money for computer use resolvable by simulation, than f have. MacArthur died. (at the age of \e) two weeks after the discovery of constant extinction. of 5OOyears; in generaln ma = I ma is the reciprocal of a half-life (naff-fife Macarthurs apply also to phenomena in units of 5OOy"ars)-r. (such as origination are inappropriate. rates) for which half-lives with and without in the text are to phenomenavariously Applications in replacement of the iten sampled.; the d.erivations therefore differ Some conputational aids: . ma. = (1og1921-1(-Ioelg(f aetait. -p2!.1 ;, Pn-s(l ctay) = 183 kilomacarthurs (t<na); Po.5(1 minute) =-263 meganX66rtnur" (rima). Given p(t) of an event;[5r interval !, P(kt) = [p(t)]E. given the sarne probability kt is na(trt) = k-rma(t), The rate for interval Haldan" (f9[g) pioposeE'an analogous mee,sure for for both intervals. rates of continuous phenomena: a d.arwin is the rate giving a change by a

23

(g). (fO). (ff1.

(re).

(fS). (rt+).

(f:).

(f6).

(U.).

(rB ).

(U.). (eO).

24

L. vAN vALEN years. Kimura (tg6g) d.efined. a unit of rates for factor of e per million pauling a is, in this context, the seme as a molecular evolution: mil]-inacarthur. (2f1. For bird.s, the d.gta are for islands and. of variable quality: MacArthur e"ndWilson (t967); Lack (rghe); Dianond, (t969,1971). The island.s are, in ord.er of size, Los Coronados, Santa Barbarao Anacapa, St. Kild.an the Krakatoa, San Miguel, San Nicolas, San Cl-ementenSanta Catalina, Scillies, Santa Rosa, Santa Cruz, Karkar, Mann and. the Orkneys. For arthropods, Simberloff and Wilson (tg6g) have approximate d.ata from mangrove islets. fhe inaccuracy of the estimated extinction rates for bird.s and arthropod.s, in the present contexto is about an orcler of magnitucle. Estimates for Manmal both groups are probably too high, but I use available val-ues. data are from Kurten (f968) for Europen Martin and Webb (in press) for Webb (tg6g) for North America, and the data of Table l for the Florida, There are nolr about I+ species per genus of me.mnal-sin both North world. lnd.epend.ent1y, the extinction Ameriea and" the world; if they go extinct This inaccurate assunprate of species is three times that of genera. between number of species present tion, which ignores the correlation and number of new species produced"o is reasonable at the seale used.. The into one equation is less defensible but grouping of bird"s and ma,:rm.als d.ata. The sa,me is true for grouping d.ata is supported. by the Florida a peninsula, and two partly isolated continents. from islands, (ZZ). rates in paleontology not in other notes can Treatments of evolutionary and be found in, e.g., the books by SchindewoLf (fgfO), Zeuner (fg:8), 297-39\ l].9fZl); anct papers Kurten (rg58); a syrnposir.un(.1. Paleont. 26: by willia.ns (195?), Kurten (rg5o), Bone (rg5f), Lerman (tg6S), House (:?6t), Newel1 (ry6f), valentine (1969), Lipps (rgfo), Kurten (r9?r), DNA estlmates are readily Olsson (tglZ) o and Cooke and Maglio (tgTZ). accessibl-e in the At1as (Dayhoff, L972); various guesses exist on the proportions that are info::national. (eg). be expected for smaller A more extreme divergence from the regression wou-l-d. organisms. Cairns, Dahlbergo Dickson, Smith, and Wal1er (f959) in fact substrate in a l-ake. give d.ata for protozoans on bfocks of an artificial = O.Oh3 per species rate of about 12 kma [P(extinction) An extinction However, the substrate was a per d.ay] can be derived from their data. area is unknovn; the area of the top was 5 x 10-9 foa,n and so the effective rate may be higher than that in sq.. 1gn. Furthe:more, the extinction areao and occurring isolated. substrates of the se.meeffective naturally the experiment tasted. only a,bout l+O d.ays. The glass slides that Patrick (lg6l) used as isfand.s for diatoms would seem an excel-lent model system because the effect of area itself can especially for such estimations, and. both stuilied. together. be isolated from that of spatial heterogeneity nay also be higher than a rate on Simberloffia 1'he estimated. extinction and Wilson (1959) Sinberloff rate comparable to that for vertebrates; trextinctionsrt seem to have been of species that coulclntt say that most und.er any circr:mstances, so no real populations of colonize the islets This is a serious bias even if one them existeti to become extinct" what we the d.anger of letting to overcome, and ilLustrates d.ifficult can easily measure d.etermine what we think we want to measuret the tyranny of epistemology on ontology. (Zl1). environment- of any organism is its ad.aptive zone (Van Valen, t{:e efiectilq .-.rgrr@arryotherorganismswithinthatad.aptivezone. (Z>). (26). A hypothesis recently revived in terns of DNA by Bachmann' Goin and Goin ( L972). ff we look at too nanow a part of the zone' with only a fev tanca, discrete

A NEIII EVOLUTIQNARY LAW

25

(Zl).

(28). (29). (30).

(Sf).

(:Z). (:f).

(3I). (::.).

(36.).

(:f).

(SA).

(-fg.) .

noticeable, as with argr rand.om process single events will be inclividually In a causal wtiverse a claim of randomness is a bad.ge in the real vorld.. d.iversification With evolutionary the causes of seemingly of ignorance. (Van Valen and. random patterns nay well be important and. cliscoverable give an exanple). The Iaw of extinction Sloan ltgtZl is on the next from such causes. 1evel of abstraction does not cause an obJect resting on the floor to fa1l. Gravitation Lakatost (tg6l-tg6h) of mathematieal proof is based on the ilifficulty critique of d.omains obJ eetively. d.elimiting But liker e.8., Mendelrs Lavs or the gas 1aws. I have treated this subJect elsewhere (Van Valen , 1972). For instance, E.D. Cope proposed a farnous law in the nineteenth century than their d.escentaxa have a greater expected longevity that prinitive This has never been aciequately tested. and should be re-formul-ated. dants. (assuming a threshold. is absent) and in terms of d.egree of primitiveness by entrance to en adaptlve zone. of prirnltiveness a d.efinition (although the 1aw of extinctlon is). The A lar.r need not be quantltative ls a rlyth derived, as Egbert Lelgh has eaid, from contrary tradition physics enry. ttNowherer you.see, it tekes all the running you can do, to keep ln the p].ace.tr (t. Carroll-, lhroueL the Looking Glaes.) sanne Fisher (r9:o) and others, including Darwin ancl especially Lyell (:-B:e), foreshadoved. the Red Queenrs Hytrlothesis but had no reason to impose the I regartl interference of constancy, and d.id. not do so. crucial- constraint a proximal competition as causally a mechanism of resource competition, regulator. rather than ultimate of longevities Whether the total- group does also depentls on the d.istribution of the subgrouPs omitted-. of extinction of a group is related to its On one leveI, the probability for different reasons beeause d.ifferent groups go extinct own properties and so at d.ifferent times. But on the next level, the Red Queen says that is not appreciably better the.n having another having one set of properties is the se.me. because the expected. time to extinction tor Drosgphilp Levine and Van Valen (rg5t+) showed experimentally !!:t Lewontin (1955) natural selection has rather non-Markovian aspects. results. but without specific later elaborated. the point theoretically are implicit in $frnFsonrs work equilibria Origination-extinction I realize that the Red Queenrs (fghl+,1953), ancl in Lyel.lrs (fBfZ). ft is d.irectly of reality. Hypothesis is at feast a sirnplification analogous to Newtonrs third lalr of motion. That the Recl Queen in her simplest gown implies long-term constancy in For any slngle protein we mugt invoke rate is obvious. total evolutionary makes the an analogue of the Central- Limit Theorern: pervasive pleiotropy to that for a1lo or linkage rate for one protein roughty proportional For i.nstancer many proteins are to some effects have a similar resul-t. ertent attachedo ancl the other components of the attachment may change for extraneous reasons, making the previous structure nonoptimal. Dickerson (fg?f) antt others have made a similar point from the other end of the microscoPe. argunent think that I'the entire Stebbins and Lewontin ( L972) actually What is a and constant.tt an average between confilsion a upon is based (over rate shorter avergge precisety that the however, is, remarkable segnents of a ptrylogeny) is so nearly constant (anong these segments) a branching rand.om walk or for a given proteino rather than reflecting some other Process'

26 (!9.).

L. VAN VALEN the precliction is of a monotone relationship of the More precisely, with the average rate of change arnong evolution average rate of protein the origin of netr eharacters). phenotypic characters (including Horne (rg6f)n Kohne (rgfo), ohta and Kirnr:ra (r9?r), uzze11 and Pilbeam, (fgtf), Also, constancy predicts the salne Jukes and Hol-mquist (L972). conmon e:cpected. nr-mber of changes in eaeh lineage after the latest The data for hemoglobins in the Atlas (Dayhoff, 1]72) seem ancestor. The assumption of total constancy with this expectation. inconsistent leads to the expecta'tion, presented seriously by D. Boulter (tglZ) and. Ranshav ei al-. (1972) that angiosperms originated. in the early or mid.d.le Paleozoic. The approach used the protein sequenee data of the &fes-'195p edition, placental phylogenies estimates of various alternative a^ndprobability as determined. bY mYself. a property of any single in the sense of a propensity, Probability lineage. the rank-order of d.ifferent groupe wlth respect io & can depend. on the It is an almoet univereal raisteJre tU think that evolutlon choice of w(t). Evolutionary fitness le ft except to some 1oca]Iy marcimlzes fitness. Selection at but evolutlon doesnrt marclmizE it. populatlon geneticists, for that level' but the any 1evel local1y maxinizes momentary. fltness Thrs is obvlous betneen levels need not eoincide. optima of different selection trut is equally true for higher levels. prezygotic and inclividual force' can decrease the most important evolutionary selection, Indivictual by, e.B. r foz'cing the occupation of only a temporary extinctlon F1 until rrfche. University of l(ansas been for,md.by J.F. The tatest record-h3l_lo* futtgn* (tatts at l-971 and. L972 meetings of the Society of Vertebrate Paleontology). Warburton (f96?), ancl Maynaril Snith (fgre) harre mad.e Lewontfn (f96f), within the usual evolutionary of gane theory to evolution applications r'ranework. Literature Allee, Cited.

(hr).

(l+2).

(hf). (!.L).

(!t). . 46)

W.C. o A.E. Enerson, O. Park, T. Park, and. K.P. Schmictt. 19h9. Prlnclples Saunders. 837 pp. Philadelphia: of Aninal Ecology. 195\. Phanerozoic time-scale Society Geological l.96)+. Quart. Jour. Anon;mous. Z6O-262. S: 120 London Geol. Soc. L972. Nuclear DNA e.mounts in vertebrates. Bachnann, K. o 0.B. Goin, and. C.J. Goin. (tt.H. Smitho ed.), pp. l+r9-\50. New York: Systems of Genetic In evolution Breach. 6rd.on and fora.niniin some Cenozoic planktonic Berggren, W.A. L969. Rates of evoluiion 1): 3rI-355. Micropaleontology fera. for regional geology and . L972. A Cenozoic time-sca1e--some implications 195-215. Lethaia paleobiogeography. 5: South African species and. human speciation. 1963. Paleontological Bone, E.L. 273-277. Jour. Sci. 59: to the evolution of higher in relationship D. 1972. Protein structure Boulter, 217-229. 5o: Sci. Prog. plants. 1959. The Cairns, J., M.L. Dahlberg, K.L. Dickson, N. Snith, and W.T. Wa]ler. protozoan to the MacArthurconmunities fresh-water of relationship modeI. Arier. Nat. 103: l+19-\:h. Wilson equititrium Science 158: 1009-1011. Clarke, B. 19?0. Darwinian evolution of proteins. in East \972, Plio-Pl-eistocene stratigraphy Cooke, H.B.S., and. V.J. Maglio. fn Calibration anrl suid evolution. probosciclean to Afriea in relation ed.s.f pp. 303-329. of HominoiC Evolution ill.W. Bishop and. J.A. Miller, Scottish Acad.emic Press.

A NEW EVOLUTIONARY LAhI

27

Dayhoff, M.O. National Dianond", E.M. Channel . tropicaf

Atlas of Protein Sequence and Structure 1972. Washington: ])72. 121++382pp. Biometlical Research Foundation. and tr:rnover rates on the species equilibria 1969. Avifaunal Sci. U.S.A. 6l+. ,T-63. Acad. Nat1. Proc. Islands of Cal-ifornia. rates on a turnover equilibrium faunal Comparison of 19?1. Sci. U.S.A. Natl. Acatl. island.. Froc. island ancl a temperate

The strueture of cytochrome c and the rates of molecular 1]7l-. Dickerson, R.E. Jor:r. Molec. Evol. ,"26-\5 evolution' model and. phylogeny in Paleozoic invertebrates. The allopatric Eldredge, N. l,7t. E'volution 25: L56-167. to an alternative an6 S.J. Goul-d. 1972. Punctuateit equillbria: -.l1yf-etic (t.l.l't. ed. Schopf )n gradualism. In Models in Paleobiology , Freemano Cooper. pp. 8Z-1.l-5. San Francisco: 1965. The potassium-8rgon d'ating of late Rvernclon, J.F., and G.H. Curtis. Cr:r. Anth. 5: 3\3-385. Cenozoic rocks in East Africa and lta1y. Potassium-argon Everndon, J.F., D.E. Savage, G.H. Curtis, and. G.T. James. t96\. Amer. North America. dates and. the Cenozoic manmalian chronology of Jour. Sci. 26zz 1l+5-198. Oxford.: 19S0. The Genetical Theory of Natr:ral Selection. Fisher, R.A. 272 PP. Clarend.on Press. measurenent of rates 1.9L+9. Suggestions as to the quantitative Haldane, J.B.S. Evolution 3: 5I-55. of evolution. SpringerBerlin: 1963. The Theory of Branching Processes. Harris, T.E. 230 PP. Verlag. Prees. Urbana: Univ. Illinois Ilennig, W. 1;966, Ptrylogenetic Systenatics.

6*

2T\2-2745

e63 'i.967. peptides and iT!11Comparisons of primate catalase tryptic Horne, S.L. 21: 7TL'785. Evolution of molecular evolution. cations for the "trray invertebrates. in the evolution of Paleozoic Fluctuations House, M.R. fg67. London' In frhe Fossil Record (W.g. Harland et al. ' eds.), pp. h-5h. SocietY of London. 6o1ogica1 nonconstancy of L972, Evolutione.:qr clock: R. Holmquist. and T.H., Jukeso science r.TTt 530-532' rate in d'ifferent species' new and zonation--a radiometrics systematics, Population Ig7O. E.G. Kauff)man, ( r ) : 6 t z 6 6 6. Proc. North American Paleont. conv. biostratigraphy. Larrrence: Allen Press. consideretl from the standKimura, M. ]:969. fhe rate of molecular evolution 11 81Proc' Natl' Acad'' Sci' u.s.A.63t point of population genetics.
1188. Science 15\: evolution' King, J.L., and T. H. Jukes. f959. Non-Darwinj.an 798. Biophys' Kohne, D.E. L9TO. Evolution of higher-organisn DNA. Qgart. Rev'
Kr:rtdn, and population tlyna.mics of fossll On the variation Ig53. B. ma.rnmal-populations.ActaZool'trennica?5:I-];22' --. mqffmals. Cold' $pring in fossil igeO-. Rates of evolution

pp.

?88-

3:

327-375.

and recent Harbor

frp.

Quant. Bio1. 2\ (ror L959)z 205-2:--5'

.' fgeg. Nicholson. IITL, .

London: Weidenfeld and Pleistocene Mama1s of Europe. 31? pp. Connent Biol. Soc. Sci. Fennica Tine-and. honinid brain size.

of British small islands' Lack, pl,- WiZ-. Ecological features of the bircl faunas Jor:r. Anin. Eco1. 11: 946' -: Brit. Jouri Philos. Sci. Ih: L963-].)61+. Proofs and refutations' Lakatos, I. r-25, 120-139, 22t-2\5, 296-3\2'

r-8.

28
Lerran,

L. VAN VALEN A. L965. 0n rates of evolution of unit characters and character complexes. Evolution 19: L6-25. I95l+. Genetic response to the sequence of two Levlne , L., and L. Van Valen. Heredlty 19: environnents. 73\-735. 795lEvolution and the theory of ge.mes. Jour. Theor. Biol. Lewontln, R.C. 1: 382-h03. 25-27. BioScience 16: Is nature probable or capricious? . ry66. 1-21. Evolution 2\: Plankton evolution. J.H. lgfO. Ilpps' Lyell, C. 1832. Principles of Geology. 1st ed., vol. 2. London: J. Murray. 330 PP New York: Harper and Row. 269 pp, MacArthur, R.H. 1972, Geographical Ecology. Princeton: A967. The Theory of Is1and Biogeography. ancl E.O. Wilson. ---ffion 203 pp. univ. Press. H.J. 1968. Modes of evolution mainly anong marine invertebrates. MacGillavry, BiJdr. Dierk. 38: 69-T\. Late Pleistocene mannals from Devilrs Martin, R.A., and. S.D. Webb. fn press. In Pleistocene Manmals of Florida (S.O. Webb, ed.. ) Deno Levy County. Univ. Florida Press. Gainesville: Ed.inburgh: Eclinburgh Univ. Press. L972. 0n Errol-ution. Maynard Smith, J. Lzj pp. Sci. Amer. 28(2)z 7Ca2. Newe11, N.D. 1953. Crises in the history of life. Geo1. Soc. Amer. Spec. . 1967. Revolutions in the history of life.

Fapl 8g: 6s-91.

I97I. An outline history of tropical organic reefs. Amer. Mus.

N ovit. 2\6 5 :
--l

1 -37.

Saunders. L97:--, Fundanentals of Ecology. Odr:m, E.P. 3rd. Ed. Philadelphia: > 14 pp. rate of 0n the constaney of the evolutionary Ohta, T., and Ki.mura, M. I97l-. 18-25. Jour. Molec. Evol. 1: eistrons. fohsi lineage. Eclogae Growbh changes in the Globorotalia 1)72. OLsson, R.K. Geol. Helvetiae 65: 165-18b. Patrick, R. L967. The effect of invasion rate, species pooI, and size of area on the stnrcture of the d.laton eornmunlty. Proc. Natl. Acad.. Sci. U.S.A. 5B: 1335-131+2. New York: Knopf. 185 pp. Pearl, R. 1928. The Rate of Llvlng. Ra:nshaw,J.A.M., D.L. Rlchardeon, B.T. Meatyard., R.H. Brown' M. Rlchard.son, 1972. The tlne of orlgln of the E.W. Thompeon, and. D. Boulter. plants determlned by ueing anlno acld sequence of cybochrome fl-owerlng c. New Phytol. 71: T'3-779. 1969. Buntonia olokunduclui sp. nov. (Ostracoita, Reynentl R., and L. Van Valen. mertstie variation in Paleocene and. Recent a study of Crustacea): ostracod.s. BuII. Geol. Inst. Univ. Uppsala (tt.S. ) f : 83-9\. Univ. Chicago 1956. Vertebrate Paleontology. Romer, A.S. 3rd ed.. Chicago: pp. Press. \58 Paleont. in paleontology. Ruzhentsov, V. Ye. 1963. The problen of transition (Translated 1951+, Rev. Int. Geol. 5: 22O\-22L3). Zhur. tg6SQ)z 3-16. of fossil dinoflagellates. The stratigraphical clistribution SarJeant, W.A.S. f967. 1: Palyno1. Rev. Palaeobot. 323-343. in Geologie und PalH,ontologie. Stuttgart: Sehindewol-f , O.H. l-]50, Dez, Zeitfektor pp. 114 Sehweizerbart. Akad.. Studien zur Stamnesgeschichte der Anmoniten. . f96I-I968.

Wiss. t1tt. Mainz, Abhan&l. Math.-Naturw. K]. 1960:

635-TU+;a95Z: )+25-

I3T-238; 19662 323-\51+, 7r9-BO8; 1968: 39-209. 572; L963: ZB5-l+:2; I965t ]-969. Experimental zoogeography of islands: anil E.O. Wilson. Simberloff , D., Ecology 50: 2TB-295. of enpty islands. the colonization EVolution, Mode in Tenpo and Col:mbia Univ. G.G. 19hh. Simpson, $ew York: press. 23T pp.

A NEWEVOLUTIONARY LAW

29

S'l

nqq

Small,

New York: Columbia . 1953. The MaJor Features of Evolution. Univ. Press . l+ll+ pp. 131-139. L.L. Rates of evolution. Jor:r. Pal-eont. 2\: the gological history of species-nr.:mber Tables to illustrate 79\5. J. 295-309. Proc. Roy. Irish Acad. (s):Oz in diatoms. Nr:merical analysis of tables evolution--Vlf]. 19)16. Quentitative Proc. the geological history of species number in diatoms. illustrate Roy. rrish A cad. (s)51: 5 3 -8 0 .

19hT. SomeLaws of Organic EvoLution. Bel-fast:

-L/P '

privately

printed..

-..',

L fo I

19LB a. Quantitative
with three laws of organic

evolut ion--IX. evolutj.on.

of species-durations' Distribution 26tProc. Roy. Irish Acad. (s)tr:

Generic sizes in relation to time evolution--X. 191+Bb. Quantitative (n):fz 279-295. Proc. Roy. Irish Acad. and type. of Frequency-d.i stributions evolution--XTl: . 191+Bc. Quantitative Proc. Roy. rrish Aead. (e)51: 3I1generic sizes in relation to tine.

2 7 8.

321+.
1950.

Eiatons.
1952.

Ann. Bot. (ttt.s.)rl+:

Quantitative

Quantitative

evolution--XVl.

Increase

of species-nrrmber in in rates of diversi-

91-113.
Correlations

evofution=-XX.

Fication.

Proe. Roy. Soc. Ed.inbureh(S)5)+: 277'29L.

1972. Comparative evolution at the level-s Stebbins, G.L., ancl R.C. Lewontin. Proc. Sixth Berkeley Synp. of molecules, organisms, ancl populations. (1. LeCam, J. Neymanoand. E.L. Scott, eds.), vol. 5, Math. Stat. Probab. Press. Univ. California Berkeley: pp. 23-\2. Generation and maintenance Stehli, F.G., R.G. Douglas, anil N.D. NewelL. f969, Science 15)+t 9\7-9\9. of gradients in taxonomic d.iversity. Science 176: 956. R. Ig72. Dionysians and Appolonians. Szent-Gy8tlyi, of fossil Geobiologie implications L}TI. Tappan, H., and A.R. Loeblich o Jr. Amer. Soc. Geol. phytoplankton evolution and time-space d.istribution. Spec. Pap. l.27 (for 19TO): 2\T-3b0. Symp. Soc. Exper. Bio1. 7: 96'I]-'3. Lgr3. Components of fitness. Thod"ay, ].1t. I97:-. Ptryletic divergence dates of hominoid. Uzzello T., and D. Pilbea.n. and molecular d.ata. Evolution 252 primates: a comparison of fossil of the J.W. 1969. Patterns of taxonomie and. ecologi.eal structure Valentine, 581+-709. 12: Palaeontology shelf benthos during Phanerozoic time. Seience l-55: I656-L658. 1;969. Criurate and evolutionary rate. Van Va1en, L. Adaptive zones and. the orders of marnmals. Evolution 25: . I9?1. 420-42U. of similarities structural Laws in blology and history: IgT2. New Literary Hist. 3: \09-)+19. academic disciplines. Evol. Theory' this 1973. Pattern and. the balance of nature.

6t>-6s>.

r ssue.

phyla comparable? Taxon. Are categories in different . In press. Group selectionr sex, and fossils. . Subnitted.. of the nultituberculates. ,and. R.E. Sloan. A966. The extinction Syst. Zoo1. 15: 25L-278. of the d.inosaurs extinction 1972. EcologY and the exti and

rcract)ffizl+ttr

tnt. Geol. cong., p. 2\7.

on a theory of guessing Warburton, F.G. 1967. Anodel of natural selectionbased games. Jor:r. Theor. Biol. 15: 78-96. land mnmmals 1.969. Extinction-origination Webb, S.D. _equil-ibria in J-ate Cenozoic Bvo]-ution 23: 588-TOe. of North Anerica.

30

L. VAN VALEN

Geol. Mag. 9\: 201-211. J)57. Evolutionary rates of brachiopods. Williarns, A. yochelsono E.L. Stenothecoidao a proposed. new class of Cambrian Mollusca. f969. Lethaia 2: )+g-62. 1958. Dating the Past. \tn ed.. Lond.on: Methuen. 515 pp. Zeuner, F.E.