Paleontological Society

Exaptation-A Missing Term in the Science of Form Author(s): Stephen Jay Gould and Elisabeth S. Vrba Reviewed work(s): Source: Paleobiology, Vol. 8, No. 1 (Winter, 1982), pp. 4-15 Published by: Paleontological Society Stable URL: http://www.jstor.org/stable/2400563 . Accessed: 06/11/2011 20:37
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Paleontological Society is collaborating with JSTOR to digitize, preserve and extend access to Paleobiology.

http://www.jstor.org

Paleobiology,8(1), 1982, pp. 4-15

Exaptation-a missing term in the science of form

StephenJayGould and ElisabethS. Vrba*

Abstract.-Adaptationhas been defined and recognized two different by criteria: historical genesis(featuresbuiltby naturalselection theirpresentrole) and current for utility (features now enhancing fitness no matter how theyarose). Biologists have often failedto recognize potential the confusion betweenthese different definitions because we have tendedto view naturalselection so dominant as amongevolutionary mechanisms thathistorical processand current productbecomeone. Yet if manyfeatures organisms of are non-adapted, but available forusefulcooptationin descendants, thenan important concepthas no name in our lexicon (and unnamed ideas generallyremainunconsidered): featuresthat now enhance but were not built by naturalselectionfortheircurrent fitness role. We proposethat such features be called exaptations and that adaptationbe restricted, Darwin suggested, features as to builtby selection fortheircurrent role. We presentseveralexamplesof exaptation, wherea failureto concepindicating tualizesuchan idea limited rangeofhypotheses the available. We exploreseveralconsequences previously of exaptation and proposea terminological solutionto the problemof preadaptation. StephenJ. Gould. Museum of Comparative Zoology,Harvard University, Massachusetts Cambridge, 02138 Elisabeth S. Vrba. TransvaalMuseum,Paul KrugerStreet,P.O. Box 413, Pretoria,South Africa Accepted: October 15, 1981

I. Introduction

We wishto proposea term a missing for item in the taxonomyof evolutionary morphology. Termsin themselves trivial, taxonomies are but revisedfor a different ordering thoughtare of not withoutinterest. Taxonomies are not neutralor arbitrary hat-racks a setofunvarying for concepts;theyreflect even create)different (or II. Two Meaningsof Adaptation theoriesabout the structure the world. As of In thevernacular, and in sciencesotherthan Michel Foucault has shown in several elegant evolutionary biology,the word adaptation has books (1965 and 1970, forexample),when you in knowwhypeople classify a certainway, you several meanings all consistentwith the etymology ad + aptus,ortowardsa fit a parof (for understand how theythink. ticularrole). When we adapt a tool for a new Successivetaxonomies thefossiltracesof are so substantialchanges in human culture.In the role,we changeits designconsciously thatit will workwell in its appointedtask. When cremid 17thcentury, madmenwereconfined inin stitutions along with the indigentand unem- ationistsbeforeDarwin spoke of adaptationployed,thusendinga long tradition exile or for the term long precedes evolutionary of to actoleration the insane. But what is the com- thought-theyreferred God's intelligent for for definite roles. mongroundfora taxonomy thatmixesthe mad tion in designingorganisms claim that largerlungs of with the unemployed-an arrangementthat When physiologists strikesus as absurd. The "key character"for Andean mountainpeoples are adapted to local directed changeforbetter the "highertaxon," Foucault argues, was idle- climates,theyspecify In ness, the cardinalsin and dangerin an age on function. short,all these meaningsreferto the brinkof universalcommerce and industry historicalprocesses of change or creation for definite functions. The "adaptation"is designed for specifically the task it performs. * An equal timeproduction; orderof authorship was determined a transoceanic by coin flip. In evolutionarybiology, however, we enreserved. (?) 1982 The Paleontological Society.All rights
0094-83 73/82/0801-0002/$1.00

(Foucault's interpretation been challenged has by British historian scienceRoy Porter, of MS). In other systemsof thought,what seems peripheral us becomescentral, to and distinctions essentialto us do not matter(whether idleness is internally inevitable,as in insanity, exteror nallyimposed,as in unemployment).

EXAPTATION

TABLE 1. A taxonomy fitness. of

Process Character Usage

Natural selection shapes the character fora current use-adaptation A character, previously shaped by naturalselection fora particular function adaptation),is coopted (an fora new use-cooptation A character whose origincannotbe ascribedto the directaction of naturalselection nonaptation), coopted (a is fora current use-cooptation

adaptation

function

aptation
exaptation effect

counter two different meanings-and a possible conflation concepts-for featurescalled adof is aptations.The first consistent with the vernacular usages cited above: a feature an adis aptationonlyifit was builtby naturalselection it The second for the function now performs. way defines adaptationin a static,or immediate as any feature thatenhancescurrent fitness, regardless of its historicalorigin.(As a further both to a process confusion, adaptation refers and a state of being. We are only discussing contributstateof beinghere-that is, features about ingto fitness. includesomecomments We thisfurther problemin sectionVIE.) Williams,in his classic book on adaptation, the recognized thisdilemmaand restricted term to its first, narrower, or meaning.We should speak of adaptation,he argues,only when we the of can "attribute originand perfection this design to a long period of selectionfor effectivenessin thisparticular role" (1966, p. 6). In his terminology, "function" refersonly to the aroperationof adaptations.Williamsfurther gues that we mustdistinguish adaptationsand theirfunctions He fromfortuitous effects. uses "effect" in its vernacular sense-something caused or produced, a resultor consequence. Williams'conceptof "effect" may be applied to or a character, to its usage, or to a potential (or process),arisingas a consequenceof true adaptation. Fortuitouseffect always connotesa and not consequencefollowing "accidentally," construction naturalsefrom arising directly by lection.Othershave adoptedvariousaspectsof this terminology "effects" for sensu Williams (Paterson 1981; Vrba 1980; Lambert, MS). However, Williams and othersusually invoke the term'effect' designatethe operationof a to

useful characternot built by selectionfor its current role-and we shall followthis restriction here (Table 1). Williams also recognizes that much haggling about adaptationhas been "encouragedby imperfections terminology" of (1966, p. 8), a situation thatwe hopeto alleviate slightly. Bock, on the otherhand, champions secthe ond, or broader, meaning in the other most widely-cited analysis of adaptation from the 1960's(Bock and von Wahlert1965;Bock 1967, 1979, 1980). "An adaptationis, thus,a feature of the organism,which interacts operationally withsome factor its environment thatthe of so individualsurvivesand reproduces"(1979, p. 39). The dilemmaof subsuming different criteria of historical genesisand current undera utility singletermmay be illustrated witha neglected example froma famoussource. In his chapter devotedto "difficulties theory," on Darwin wrote (1859, p. 197): The suturesin the skullsof youngmammals have been advanced as a beautiful adaptation foraiding parturition, no doubt theyfaand cilitate,or may be indispensable thisact; for but as suturesoccur in the skulls of young birdsand reptiles, whichhave onlyto escape froma brokenegg, we may inferthat this structure arisenfrom laws of growth, has the and has been taken advantageof in the parturition the higheranimals. of Darwin assertstheutility, indeedthe necessity, of unfusedsuturesbut explicitly declinesto label theman adaptationbecause theywere not builtby selection function theynow do in to as mammals.Williamsfollows Darwin and would

STEPHEN JAY GOULD & ELISABETH S. VRBA

decline to call this featurean adaptation; he towardsfitness. They owe theirfitness feato would designateits role in aiding the survival turespresentforotherreasons,and are thereof mammals as a fortuitous effect.But Bock fore (aptus) by reason of (ex) theirform,or fit would call the suturesand the timingof their ex aptus. Mammaliansutures an exaptation are fusionan adaptation,and a vital one at that. forparturition. Adaptations have functions; exAs an example of unrecognized confusion, aptationshave effects. The general,staticpheconsiderthisdefinition adaptationfroma bi- nomenonof beingfitshouldbe called aptation, of ologicaldictionary (Abercrombie al. 1951, p. notadaptation.(The set of aptations et existing at 10): "Any characteristic living organisms any one timeconsistsof two partially of overlapwhich, in the environment they inhabit, im- ping subsets:the subsetof adaptationsand the provestheirchancesof survivaland ultimately subset of exaptations.This also applies to the leaving descendants,in comparisonwith the moreinclusiveset of aptationsexisting through chances of similarorganisms without char- time;see Table 1.) the tends to acteristic;natural selectiontherefore Need fora in " establish adaptations a population. This def- IV. The Current initionconflates currentutility with historical Conceptof Exaptation genesis.What is to be done with usefulstrucof Why has this conflation historical genesis turesnotbuiltby naturalselection theircur- with current for attracted littleattention so utility rentrole? heretofore? Every biologist surely recognizes thatsome usefulcharacters not arise by sedid III. A Definition Exaptation of lectionfortheircurrent roles;whyhave we not We have identifiedconfusionsurrounding honored thatknowledge witha name?Does our one of the centralconceptsin evolutionary the- failureto do so simplyunderscorethe unimory. This confusion arises,in part,because the portanceof the subject? Or mightthis absent taxonomy formin relationto fitness of lacks a term,in Foucault's sense, reflect conceptual a term.FollowingWilliams(see Table 1), we may structure excludedit?And,finally, that does the as thatpro- potentialneed forsuch a termat this time indesignate an adaptationanyfeature and was built by selectionforits dicate that the conceptualstructure motesfitness itselfmay role of current (criterion historical genesis).The be altering? of operation an adaptationis its function.(Bock Why did Williamsnot suggesta term,since uses the term functionsomewhatdifferently,he clearlyrecognized problemand did septhe but we believe we are following biological arate usages into functions the and effects (correvernacularhere.)We may also followWilliams spondingrespectively adaptationsand to the to in labellingthe operationof a usefulcharacter unnamed featuresthat we call exaptations)? role as an Whydid Bock failto specify problem all? not built by selectionforits current at the as effect. (We designate an effect onlythe usage We suspect that the conceptualframework of not see modern evolutionary of such a character, thecharacter itself, thought,by continually p. 5.) But what is the unselected,but useful emphasizing supremeimportance and conthe characteritselfto be called? Indeed it has no tinuity adaptationand naturalselection all at of recognized name(unlesswe acceptBock's broad levels, subtlyrelegatedthe issue of exaptation of definition adaptation-the criterion cur- to a periphery unimportance. of of How could rentutility alone and rejectbothDarwin and nonadaptive aspectsofform gaina properhearWilliams).Its space on the logical chartis cur- ing under Bock's definition (1967, p. 63): "On blank. rently features anof theoretical grounds,all existing We suggest thatsuch characters, evolved for imals are adaptive. If theywere not adaptive, otherusages(or forno function all), and later thentheywould be eliminated selectionand at by "coopted" for theircurrent role, be called ex- would disappear."Williamsrecognized phethe aptations. (See VIA on the relatedconcept of nomenon of exaptation and even granted it "preadaptation.") They are fitfortheircurrent some importance assessingthe capacitiesof (in role, hence aptus, but theywere not designed the humanmind,forexample),but he retained forit, and are therefore ad aptus,or pushed a preeminent foradaptationand oftendesnot role

EXAPTATION

ignated effectsas fortuitous peripheral- V. Examples of Exaptation or "merely incidental an consequence"he statesin A) Feathers and flight-sequential exaptation one passage (p. 8). in the evolution birds.-Consider a common of We believethatthe adaptationist of scenariofrom evolution birds.(We do not program the of modernevolutionary thought(Gould and Le- assertits correctness, onlywish to examine but wontin1979) has been weakening a resultof appropriateterminology a common set of as for challengesfromall levels, moleculesto macro- hypotheses.)Skeletal features,including the evolution.At the biochemicallevel, we have sternum, basket and shoulderjoint, in late rib theories neutralism suggestions of and thatsub- Jurassicfossilsof Archaeopteryx indicate that stantialamountsof DNA may be nonadaptive this earliestknownbird was probablycapable at the level of the phenotype (Orgel and Crick of only the simplestfeatsof flight. Yet it was 1980; Doolittleand Sapienza 1980). Studentsof quite thoroughly feathered. This has suggested macroevolution have arguedthatadaptationsin to many authorsthat selectionfor the initial translate effects yieldthe pat- development feathers an ancestorwas for as populations to of in ternsof differential species diversification that the functionof insulationand not for flight may resultin evolutionary trends (Vrba's effect (Ostrom1974, 1979; Bakker 1975). Such a funhypothesis,1980). If nonadaptation(or what damental innovationwould, of course, have shouldbe called nonaptation) aboutto assume many small as well as far-reaching, is incidental an important in a revisedevolutionary role the- consequences.For example, along no descenory,thenour terminology formmust recog- dant lineage of this first of feathered species did nize its cardinal evolutionarysignificance- (so faras we know)a furry covering thebody of cooptability fitness for (see Seilacher 1972, on evolve. The fixation earlyin thelifeof the emimportant effects a nonaptivepattern the bryo,of cellular changes that lead on the one of in structure coloration molluscs). and of hand to hair,and on the otherto feathers, conSome colleagues have said that they prefer strainedthe subsequentcourse of evolutionin Bock's broad definition because it is moreeasily body covering (Oster 1980). operational.We can observeand experiment to Archaeopteryx alreadyhad largecontour-type determine what good a feature does foran or- feathers, arrangedalong its arms in a pattern the ganismnow. To reconstruct historical path- very much as in the wings of modernbirds. way of its originis always more difficult and Ostrom(1979, p. 55) asks: "Is it possible that often (whencrucialevidenceis missing) intract- the initial (pre-Archaeopteryx) enlargement of able. featherson those narrow hands might have To this we replythat we are not trying to been to increasethe hand surface area, thereby dismantle Bock's concept.We merely arguethat makingit more effective catchinginsects?" in it should be called aptation (with adaptation He concludes(1979, p. 56): "I do believe that and exaptationas its modes). As aptation, it the predatory design of the wing skeletonin retainsall the favorableproperties testing Archaeopteryx strong evidence of a prior for is enumerated above. predatory function the proto-wing a curof in Historical genesis is, undoubtedly,a. more sorial proto-Archaeopteryx. " Later selectionfor but difficult problem we cannottherefore ignore changes in skeletalfeaturesand feathers, and it. As evolutionists, are charged,almostby forspecific we neuromotor in resulted the patterns, to definition, regardhistorical pathwaysas the evolutionof flight. essenceofour subject.We cannotbe indifferent The Black Heron (or Black Egret, Egretta to the factthatsimilarresults can arise by difardesiaca) of Africa,like most modernbirds, ferent historical routes.Moreover,the distincBut it also uses themin uses its wingsin flight. tion betweenad- and exaptation, howeverdifan interesting way to preyon small fish:"Its ficult,is not unresolvable.If we ever find a is standingin shallow water small running dinosaur,ancestralto birds and fishing performed withwingsstretched and forward, out forming clothedwithfeathers, will know that early we canopywhichcasts a shadow feathers were exaptations, not adaptations,for an umbrella-like flight. on the water.In thisway its foodcan be seen"

STEPHEN JAY GOULD & ELISABETH S. VRBA

(McLachlan and Liversidge1978, p. 39, Plate 6). This "mantling" the wingsappears to be of a characteristic behaviorpattern, witha genetic basis. The wing and featherstructures themin selvesdo notseemto be modified comparison with those of closelyrelatedspecies, the individuals of whichdo not huntin thisway (A. C. Kemp, pers. comm.). We see, in this scenario,a sequentialset of adaptations, each converted an exaptation to of different effect that sets the basis fora subsequent adaptation. By this interplay, major a evolutionary transformation occursthat probably could not have arisenby purelyincreasing adaptation. Thus, the basic design of feathers is an adaptationforthermoregulation latand, er, an exaptationforcatchinginsects.The developmentof large contourfeathers and their arrangement thearmariseas adaptations on for insect catching and become exaptations for flight. Mantlingbehavioruses wingsthatarose as an adaptation for flight.The neuromotor modifications governing mantling behavior,and therefore mantling the are posture, adaptations in forfishing. The wing per se is an exaptation its current effect shading, of just as thefeathers it covering also arose in different adaptive conbut texts have providedmuchevolutionary flexibilityfor other uses duringthe evolution of birds. B) Bone as storageand support.-The developmentof bone was an eventof major significance in the evolutionof vertebrates. Without could nothave latertakenup bone, vertebrates life on land. Halstead (1969) has investigated the question: grantingits subsequent importance as body supportin the laterevolutionof vertebrates, why did bone evolve at such an Some authors earlystage in vertebrate history? have hypothesized that bone initially arose as an osmoregulatory responseto life in freshwater. Others,like Romer(1963), postulateinitial adaptation of bony "armor" for a protective function. Pautard (1961, 1962)pointedout that withmuchmuscular needs anyorganism activity a conveniently accessible store of phosphate. Following Pautard, and noting the seasonal cycleof phosphateavailabilityin the sea, Halstead (1969) suggestedthe following scenario: Calcium phosphates,laid down in the skin of as the earliestvertebrates, evolved initially an

adaptation for storingphosphatesneeded for metabolicactivity.Only considerably later in evolution bone replacethecartilaginous did endoskeletonand adopt the function support of forwhichit is now mostnoted. Thus, bone has two majoruses in extantvertebrates: support/protection storage/homeoand stasis (as a storehouse certainmineralions, for includingphosphateions). The ions in vertebratebone are in equilibrium withthosein tissue fluidsand blood, and function certain in metabolicactivities(Scott and Symons 1977). For instance,in humans, 90% of body phosphorusis present theinorganic in phase of bone (Duthie and Ferguson1973). FollowingHalstead's analysis,thedeposition of phosphatein body tissuesoriginally evolved as an adaptationfora storage/metabolic funcfor tion. The metabolicmechanism producing bone per se can thus be interpreted an exas aptation for support. The metabolic mechanisms for depositingan increasedquantityof phosphatesand for mineralization, well as as in thearrangement bonyelements an internal of are skeleton, thenadaptationsforsupport. C) The evolutionof mammalianlactation.Dickerson and Geis (1969) recounthow Alexander Fleming,in 1922,discovered enzyme the lysozyme. had a cold and, forinterest's He sake, added a fewdropsof nasal mucusto a bacterial culture.To his surprisehe found,aftera few in days, that something the mucus was killing the bacteria:the enzymelysozyme, since found in mostbodilysecretions in largequantities and in the whitesof eggs. Lysozyme destroys many the or bacteriaby lysing, dissolving, mucopolyof saccharidestructure thecell wall. The amino acid sequence of a-lactalbumin, milk protein a ofpreviously unknown was function, thenfound that some to be so close to that of lysozyme, relationshipof close homology must be involved. Dickersonand Geis (1969, pp. 77-78) write: a-Lactalbuminby itself notan enzymebut is was foundto be one component a two-proof tein lactose synthetase system,presentonly in mammaryglands duringlactation . . .. The othercomponent(the "A" protein)had in been discovered theliverand otherorgans as an enzyme thesynthesis N-acetyllacfor of

EXAPTATION

forcedpreviousauthorsinto this tosaminefromgalactose and NAG. But the unconsciously conceptualbind. of the A proteinand a-lactal- erroneous combination of Kruuk (1972), the leading student spotted the buminsynthesizes milksugarlactosefrom galactose and glucose instead. The non-cat- hyenas, for example, notes that the enlarged acts as a con- sexual organsof femalesare used in an imporalytica-lactalbuminevidently one po- tant behaviorknow as the meetingceremony. from troldevice to switchits partner wanderIt appears Hyenas spend long periodsas solitary to tentialsynthesis another.... for was thatwhena milk-producing-system being ers searching carrion,but theyalso live in and clans that defendterritory of developedduringthe evolution mammals, well integrated A for and when a need for a polysaccharide-syn- engagein communalhunting. mechanism propintotheir wanderers solitary thesizingenzymearose, a suitable one was reintegrating cera found in part by modifying pre-existing er clan mustbe developed. In the meeting emony,two hyenasstand side to side, facingin enzyme. polysaccaride-cutting opposite directions.Each liftsthe inside hind to an in in Thus, lysozyme, all vertebrates which leg, exposing erectpenisor clitoris itspartgenand for an it occurs,is probably adaptation thefunc- ner'steeth.They sniff lickeach other's at tion of killingbacteria. Furtherevolution in itals for10 to 15 seconds,largely the base of of and in front the scrotum mammals (alterationof a duplicated gene ac- the penisor clitoris cordingto Dickersonand Geis, 1969) resulted or falsescrotum. utilityfor the Having discovereda current with an in a-lactalbumin, adaptation(together genitaliaof females,Kruuk external of the A protein)for the function lactose syn- prominent thattheymust have in thesisand lactation.Human lysozyme, this (1972, pp. 229-230) infers scenario, is an adaptation for lysingthe cell evolved forthispurpose: walls of bacteria, and an exaptationwith reto It is impossible thinkof any otherpurpose system. spectto the lactose synthetase thanforuse in forthisspecial femalefeature in D) Sexual "mimicry" hyenas.-Females of . ceremony . . It may also be, the meeting the spottedhyena,Crocutacrocuta,are larger then,that an individualwith a familiarbut overthem.Pliny,and thanmalesand dominant complexand conspicuousstructure relatively a had alreadyrecognized otherancientwriters, has an advansniffed duringthe meeting at of relatedand unusual feature theirbiologyin would often tage over others;the structure (falsely,as Ariscalling them hermaphrodites thisreestablishment social bonds of facilitate genitaliaof females totleshowed).The external over a longer by keepingpartnerstogether fromthe sexual indistinguishable are virtually meetingperiod. This could be the selective is organsofmalesbysight.The clitoris enlarged advantage that has caused the evolutionof structure and extended to form a cylindrical the females'and cubs' genitalstructure. slit witha narrow at itsdistalend; it is no smallYet another hypothesis,based upon facts er thanthemale's penisand can also be erected. The labia majora are folded over and fused known to every Biology I student,virtually The penis and clitoris along the midlineto forma false scrotal sac criesout forrecognition. and organs,as are the scrotum (thoughwithouttesticlesof course), virtually are homologous identicalin formand positionwith the male's labia majora. We know thathighlevels of anof drogen induce the enlargement the clitoris scrotum (HarrisonMatthews1939). of is overand fusion thelabia until "mimicry" full and thefolding on The literature thissexual sac penisand scrotal respectively. of speculationsabout adaptive meaning.Most theyresemble util- (In fact,in an important sense,theyare thena current have conflated of thesearguments genesisin assumingthat the penis and scrotal sac, given the homologies.) ity and historical of demonstration modernuse (Bockian adap- Human baby girlswithunusuallyenlargedadthe path of origin(adaptation renalssecretehighlevels of androgen,and are tation)specifies scroclitoris an empty and as used by Williams and Darwin, and as ad- bornwitha peniform of thatthe ab- tal sac formed the fusedlabia. in thispaper). We suggest vocated has of Female hyenas are larger than males and concept exaptation senceofan articulated

10

STEPHEN JAY GOULD & ELISABETH S. VRBA

dominantover them. Since these featuresare that the architecture geneticmaterialmight of often hormonally mediatedin mammals,should fitall theirpresuppositions about evolutionary we not conjecture thatfemalesattaintheirsta- processes.The linearorderofnucleotides might tus by secreting androgensand that the peni- be the beads on a stringof classical genetics: form clitoris and falsescrotalsac are automatic, one gene, one enzyme;one nucleotide substitusecondaryby-products. Since theyare formed tion,one minutealteration naturalselection for anyway,a laterand secondary utility might en- to scrutinize.We are now, not even 20 years sue; theymay be cooptedto enhancefitness in later,faced withgenesin pieces,complexhierthe meeting ceremonyand then secondarily archies of regulation and, above all, vast modified thisnew role. We suggest for thatthe amountsof repetitive DNA. Highlyrepetitive, peniform clitoris and false scrotalsac arose as or satellite, DNA can existin millions copies; of nonaptive consequences highandrogen of levels middle-repetitive DNA, withitstensto hundreds (a primaryadaptation related to the unusual of copies, formsabout one quarterof the gebehavioralroleoffemales).They are, therefore, nome in both Drosophila and Homo. What is exaptations themeeting for and ceremony, their all the repetitive DNA for(if anything)? How effectin enhancingfitness throughthat cere- did it getthere? monydoes notspecify historical the pathwayof A survey previousliterature of (Doolittleand theirorigin. Sapienza 1980; Gould 1981) revealstwo emergYet this obvious hypothesis, with its easily ing traditions argument, of both based on the testablecardinalpremise, was notexplicitly ex- selectionistassumption that repetitiveDNA amined until 1979 after,literally, more than must be good for something so much of it if in 2000 yearsof speculation the adaptive mode exists.One tradition (see Britten and Davidson (both ancient authorsand medieval bestiaries 1971 forits locus classicus) holds thatrepeated triedto inferGod's intentin creatingsuch an copies are conventionaladaptations, selected odd beast). Racey and Skinner (1979) foundno foran immediate rolein regulation bringing (by in differences levels of androgenin blood plas- previously isolatedpartsofthegenome intonew ma of male and femalespottedhyenas.Female and favorable for combinations, example,when fetuses containedthe same highlevel of testos- repeated copies disperse among several chroterone adultfemales.In theother as two species mosomes).We do not doubt thatconventional of the familyHyaenidae, however, androgen adaptation explains the preservation much of levels in blood plasma are much lower forfe- repeatedDNA in thismanner. males than formales. Females of thesespecies But many molecular evolutionists now are notdominant overmalesand do notdevelop strongly suspect that directadaptation cannot clitorises falsescrotalsacs. or peniform explain the existence of all repetitiveDNA: We do notassertthatour alternative hypoth- thereis simplytoo much of it. The second traesis of exaptationmust be correct.One could ditiontherefore holdsthatrepetitive DNA must runthescenarioin reverse (witha bit offorcing exist because evolutionneeds it so badly fora in our judgment): females "need" prominent flexible future-as in thefavored that argument genitaliaforthe meetingceremony; theybuild "unemployed," redundant copies are freeto althembyselection highandrogen for levels;large terbecause their is necessary product stillbeing size and dominanceare a secondary by-product generated theoriginal by copy(see Cohen 1976; of the androgen.We raise, rathera different Lewin 1975; and Kleckner 1977, all of whom issue: whywas thisevidentalternative con- also follow the firsttraditionand argue both not sidered,especially Kruukin his excellent by ex- sides). While we do not doubt thatsuch future haustivebook on the species?We suggestthat uses are vitallyimportant consequencesof rethe absence of an explicitly articulated concept peated DNA, theysimplycannotbe the cause of exaptationhas constrained range of our ofitsexistence, the unlesswe return certain to theisin hypotheses subtleand unexamined ways. tic views that permitthe controlof present E) The uses of repetitive DNA.-For a few eventsby future needs. years afterWatson and Crick elucidated the This second tradition expressesa correctinstructure DNA, many evolutionists of hoped tuitionin a patently nonsensical(in its nonpe-

EXAPTATION

11

thought formed a different (thermoregulation in manner.The missing jorativemeaning) way for concept feathers,for example). Yet we traditionally thatsuppliessense is a well articulated of in Defenders thesecondtradition apologizefor"preadaptation" our textbooks, of exaptation. DNA is to and laboriously repetitive how important understand pointout to students thatwe do lan- not mean to implyforeordination, that the but onlyknow theconventional evolution, and this convic- word is somehowwrong(though conceptis guage of adaptationforexpressing the condition (when secure). Frazzetta (1975, p. 212), forexample, is tion.But sinceutility a future function writes: "The associationbetweentheword'precopyassumesa different theredundant stilllingers, or undergoessecondaryadaptationfor a new adaptation'and dubious teleology role), an impasse in expressiondevelops. To and I can oftenproduce a wave of nausea in break this impasse, we mightsuggestthat re- some evolutionary biologistswhen I use the available word unlessI am quick to say what I mean by peated copies are nonaptedfeatures, forcooptation later,but not servingany direct it." Indeed, thewordis wrongand ourlongstandat function the moment.When coopted, they discomfort justified(see Lamis in will be exaptations theirnew role (withsec- ing intuitive bert,MS). For ifwe dividetheclass of features if ondaryadaptive modifications altered). to What thenis the sourceof theseexaptations? contributing fitness intoadaptationsand exthey arise as aptations,and if adaptationswere constructed Accordingto the firsttradition, different (and exaptations and laterassumetheir trueadaptations coopted)fortheircurrent use, function.The second tradition,we have ar- then featuresworkingin one way cannot be and subsequent gued, must be abandoned. A thirdpossibility preadaptationsto a different been proposed (or, rather,better usage: the termmakes no sense at all. has recently The recognition exaptationsolves the diof afterprevioushints):perhapsrepeated codified for call for copies can originate no adaptivereasonthat lemmaneatly, what we now incorrectly the Darwinianlevel ofphe- "preadaptation"is merelya categoryof exapconcerns traditional advantage(Orgel and Crick 1980; Doo- tation consideredbefore the fact. If feathers notypic littleand Sapienza 1980). Some DNA elements evolved forthermoregulation, become exthey once birdstake off.If, howare transposable;if these can duplicate and aptationsforflight of as we accumulation long ever, with the hindsight history, choose move,whatis to stoptheir while theystill encase the as they remain invisibleto the phenotype(if to look at feathers thattheybeginto ex- running,dinosaurianancestorsof birds, then theybecomeso numerous ertan energetic constraint exaptations flight, for or upon thephenotype, theyare onlypotential (thatis, aptus-or fit-beforetheir Such preaptations them)? will thennaturalselection eliminate DNA" may be playingits own Darwin- actual cooptation).The term "preadaptation" "selfish a ian game at a genic level, but it represents should be dropped in favor of "preaptation." are potential,but unrealized,exat truenonaptation the level of the phenotype. Preaptations arise as a non- aptations; theyresolveMivart'smajorchallenge Thus, repeatedDNA may often in aptation. Such a statement no way argues to Darwin. fuB) Primaryexaptationsand secondaryadits vital importance evolutionary for against tures. When used to great advantage in that aptations.-Feathers, in theirbasic design,are theserepeatedcopies are exaptations. for but future, exaptations flight, once this new effect was added to the function thermoregulation of VI. Significance Exaptation of as an importantsource of fitness,feathers A) A solution to the problemof preadapta- underwenta suite of secondary adaptations has called post-adaptations) enhance tion.-The conceptofpreadaptation always (sometimes to in to We acknowl- theirutility flight.The orderand arrangebeen troubling evolutionists. limbbonesis an exaptation as for edge itsnecessity theonlyDarwiniansolution mentoftetrapod stages walkingon land; many modifications shape of to Mivart's(1871)old tauntthat"incipient as are could not function the and musculature secondary of usefulstructures" adaptationsfor life. forms (whrat do good is 5% ofa wing). terrestrial perfected of The evolutionary The incipient stages,we argue,musthave perhistory any complexfea-

12

STEPHEN JAY GOULD & ELISABETH

S. VRBA

turewill probablyincludea sequentialmixture of adaptations, primaryexaptationsand secondaryadaptations.Justas any featureis plesiomorphicat one taxonomiclevel and apomorphic at another (torsion in the class Gastropoda and in the phylumMollusca), we are not disturbedthat complexfeaturesare a mixtureof exaptationsand adaptations. Any cooptedstructure exaptation) (an will probably not arise perfected its new effect.It will for therefore develop secondary adaptations the for new role. The primary exaptations and secondary adaptations can, in principle,be distinguished. C) The sources of exaptation.-Features coopted as exaptationshave two possible previous statuses.Theymayhave beenadaptations for anotherfunction, theymay have been or non-aptivestructures. The first has long been recognized as important,the second underplayed. Yet the enormous pool of nonaptations of must be the wellspring and reservoir most We evolutionary flexibility. need to recognize the centralrole of "cooptability fitness" for as the primary evolutionary significance ubiqof in In uitous nonaptation organisms. thissense, this and at itslevel ofthephenotype, nonaptive pool is an analog of mutation-a sourceof raw selection. materialforfurther Both adaptations and nonaptations,while theymay have non-random proximate causes, as can be regarded randomly producedwithrerespect to any potential cooptation further by gimes of selection.Simplyput: all exaptations to withrespect their effects. originate randomly Together,these two classes of characters,adaptations and nonaptations, provide an enorat mous pool of variability, a level higherthan for as mutations, cooptation exaptations. (Lamto bert,MS, has discussedthiswithrespect preadaptations only-preaptations in our terminology. He explored the evolutionary implications thenotionthatforany function, of fromnaturalselectionat any resulting directly one time,theremay be multiple effects.) If all exaptationsbegan as adaptations for in anotherfunction ancestors,we would not this have written paper. For theconceptwould of be coveredby theprinciple "preadaptation"and we would only need to point out that would be a better "preaptation" term,and that

etymology requiresa different name forpreaptationsaftertheyare established.Exaptations that began as nonaptations represent missthe ing concept.They are not coveredby the principle of preaptation, theywere not adaptafor tions in ancestors.They trulyhave no name, and concepts without namescannotbe properly in incorporated thought.The greatconfusions ofhistorical and current genesis utility primarily involve usefulfeaturesthat were not adaptationsin ancestors-as in our examplesof sexual "mimicry" hyenasand theuses of middle-rein petitive DNA. D) The ironyof our terminology nonapfor tation.-It seems odd to definean important thing whatit is not. Students earlygeology by of are rightly offended thatwe refer 5/6 of earth to history Precambrian. as Featuresnot now contributing fitness usuallycalled nonadapto are tations.(In our terminology theyare nonaptations.)This curiousnegative definition only can recorda feeling thatthesubjectis "lesser"than the thingit is not. We believe that thisfeeling is wrong,and that the size of the pool of nonaptationsis a centralphenomenon evolution. in The term"nonadaptive"is but anotherindication of previous-and in our view false-convictions about the supremacyof adaptation. The burden of nomenclature already great is enoughin thispaper and we do not propose a new termforfeatureswithoutcurrent fitness. But we do wish to recordthe irony. E) Process and state-of-being. -Evolutionary biologistsuse the termadaptationto describe both a currentstate-of-being discussed in (as this paper) and the processleading to it. This no dualitypresents problemin cases of trueadaptation,where a process of selectiondirectly produces the state of fitness.Exaptations, on the otherhand, are not fashioned theircurfor rent role and reflect attendantprocess beno yond cooptation(Table 1); theywere built in the past eitheras nonaptiveby-products as or roles. adaptationsfordifferent Perhapswe shouldbeginour analysisof process witha descriptive approachand simplyfocus upon the set of features thatincreasetheir relativeor absolute abundance withinpopulations,speciesor clades by theonlygeneralproor cesses that can yield such "plurifaction," "more making":differential branchingor per-

EXAPTATION

13

and Fristrup,MS). This the flexibility phenotypic of sistence(see Arnoldcharacters a prias has processofplurifaction two basic maryenhancerof or damperupon futureevodescriptive causes. First, featuresmay increasetheirrep- lutionary change. Flexibility in the pool of lies to activelyby contributing branch- features resentation available forcooptation(eitheras adeitheras adaptationsevolved aptationsto something ing or persistence else that has ceased to or function, ex- be important new selectiveregimes,as adby selectionfor theircurrent in routeand coopted aptationswhose original aptationsevolvedby another function continues but and particular- whichmaybe cooptedforan additionalrole,or Secondly, fortheir usefuleffect. ly at the higherlevel of species withinclades, as nonaptationsalways potentially available). may increasetheirown representation The paths of evolution-both the constraints features causal and the opportunities-must largelyset by reasons,including fora hostofnonaptive be to contributing fitness, the size and natureof thispool of potentialexwithfeatures correlation found at such sur- aptations.Exaptive possibilities correlation and fortuitous definethe "inin highfrequency randomsimulations ternal" contribution prisingly that organismsmake to by Raup and Gould (1974). These nonaptive theirown evolutionary future. A. R. Wallace, a strictadaptationist ever for establishan enormous-pool potenif features tial exaptation. therewas one, nonetheless denied that natural selection had builtthe humanbrain. "Savages" VII. Conclusion (living primitives),he argued, have mental we decisionaboutwhether have equipmentequal to ours, but maintainonly a The ultimate or a written trivialessay on terminology made rude and primitive culture-thatis, theydo not about evo- use mostof theirmentalcapacitiesand natural statement a potentially interesting of lution must hinge upon the importance ex- selection onlybuildforimmediate can use. Darand aptation,bothin frequency in role. We be- win, who was not a strictadaptationist, was to lieve that the failureof evolutionists codify both bemusedand angered.He recognizedthe be- hiddenfallacyin Wallace's argument: such a conceptmustrecordan inarticulated that the liefin its relativeinsignificance. brain,though builtbyselection for undoubtedly We suspect, however, that the subjects of some complexset of functions, can, as a result are and cooptability of paramount of its intricate nonaptation workin an unlimited structure, importancein evolution. (When cooptability numberof ways quite unrelated the selective to has been recognized-in the principleof "pre- pressure thatconstructed Manyoftheseways it. adaptation"-we have focussedupon shiftin mightbecome important, not indispensable, if adapted for some- forfuture role forfeaturespreviously survivalin later social contexts (like in afternoontea for Wallace's contemporaries). for else, noton thepotential exaptation thing of The flexibility evolu- But current nonapted structures.) utility carriesno automaticimplitionlies in the rangeof raw materialpresented cationabout historical origin.Most of what the to processesof selection.We all recognizethis brain now does to enhanceour survivallies in sourcesofgenetic the domain of exaptation-and does not allow in discussing conventional the about theselectivepaths variation-mutation, recombination,and so us to make hypotheses How muchof the evolutionto forth-presented naturalselectionfromthe of humanhistory. on levelbelow. But we have notadequately ary literature human behavior would colgenetic the appreciated that features of the phenotype lapse if we incorporated principleof exapthemselves intothecoreofourevolutionary (with theirusuallycomplexgenetic tation thinking? because it bases) can also act as variantsto enhance and This collapse would be constructive change. Thus the would vastlybroadenour rangeof hypotheses, futureevolutionary restrict on function and deimportantstatementof Fisher's fundamental and focusattention current instead of (all testablepropositions) considersonly geneticvariance in re- velopment theorem about primal "The rate of increasein fitness leadingus to unprovablereveries lationto fitness: on at of any organism any timeis equal to its ge- fratricide the Africansavanna or dispatchat at netic variance in fitness that time" (Fisher ing mammoths the edge of greatice sheetsconsider a valid subject,but one better in treated novels 1958). In an analogous way, we might

14

STEPHEN JAY GOULD & ELISABETH

S. VRBA

that can be quite enlightening scientifically is not anti-selectionist, we view this paper and (Kurten1980). as a contribution Darwinism,not as a skirto Consideralso theapparently crucialrole that mish in a nihilistic vendetta.The main theme repeatedDNA has played in the evolutionof is, after cooptability all, forfitness. Exaptations phenotypic complexity organisms.If each are vital components any organism's in of success. gene codes foran indispensable enzyme per(or forms necessary any function), asks Ohno (1970) Acknowledgments in his seminalbook, how does evolutiontranThe following have commented themanuon scendmeretinkering alongestablished linesand script:C. K. Brain, C. A. Green,A. C. Kemp, achieve the flexibility build new typesof or- H. E. H. Paterson.One of us (E.S.V.) owes a to ganization. Ohno arguesthatthisflexibility must debtto Hugh Patersonforan introduction, durariseas theincidental result geneduplication, ing extensive of discussions, the terminology to of withits production redundant of geneticmate- effects (sensuWilliams).We boththankhimfor rial: "Had evolutionbeen entirely dependent referring to theexamplesofmantling us behavupon naturalselection,froma bacteriumonly ior in the Black Heron and lysozyme/a-lactalnumerous forms of bacteria would have buminevolution.D. M. Lamberthas given us emerged.. . . Only the cistronwhich became access to an unpublishedmanuscript, and has redundant was able to escape from relentless discussedwithus the ubiquitouspresence,and the pressureof naturalselection,and by escaping, enormousimportance, evolutionof what he in it accumulatedformerly forbidden mutations to and otherscall preadaptation. the emergeas a new gene locus" (from preface to Ohno 1970). Cited We argued in sectionVE that much of this Literature repetitive DNA mayarisefornonaptive reasons ABERCROMBIE, M., C. H. HICKMAN, AND M. L. JOHNSON. 1951. A Dictionary of Biology. 5th edition, 1966. Hunt Bernard and at the level of the individualphenotype in (as Co. Ltd., Aylesbury, Great Britain. the "selfishDNA" hypothesis). The repeated ARNOLD, A. J. AND K. FRISTRUP. 1982. The hierarchical basis for copies are thenexaptations, cooptedforfitness a unified theory of evolution. Paleobiology, in press. and secondarilyadapted for new roles. And BAKKER, R. T. 1975. Dinosaur renaissance. Sci. Am. 232(4):58-78. BOCK, W. 1967. The use of adaptive characters in avian classifiin theyare exaptations the interesting category cation. Proc. XIV Int. Ornith.Cong., pp. 66-74. of structures that arose as nonaptations, when BOCK, W. 1979. A syntheticexplanationof macroevolutionary change-a reductionistic approach. Bull. Carnegie Mus. Nat. DNA" hypothesis the "selfish applies. Hist. No. 13:20-69. Thus, the two evolutionary phenomenathat BOCK, W. J. 1980. The definition recognition biologicaladand of may have been mostcrucialto thedevelopment aptation.Am. Zool. 20:217-227. of complexity withconsciousness our planet BOCK. W. J. AND G. VON WAHLERT. 1965. Adaptationand the on form-function complex.Evolution. 10:269-299. (if readerswill pardon some drippinganthro- BRITTEN,R. J. AND E. H. DAVIDSON. 1971. Repetitiveand nonfor pocentrism themoment)-theprocessofcre- repetitiveDNA sequences and a speculationon the originsof evolutionary in atinggeneticredundancy thefirst place, and COHEN, S. N. novelty.Q. Rev. Biol. 46:111-131. and plasmid 1976. Transposable geneticelements the myriad and inescapable consequences of evolution.Nature. 263:731-738. device as complex as DARWIN,C. 1859. On the Originof Species. J. Murray:London. buildingany computing and the human brain-may both represent exapta- DICKERSON,R. E. AND I. GEIS. 1969. The Structure Action of Proteins.Harper and Row; New York. tions that began as nonaptations, concept DOOLITTLE,W. F. AND C. SAPIENZA. 1980. Selfishgenes,thephethe previouslymissingin our evolutionary termi- notypeparadigm,and genomeevolution.Nature. 284:601-603. nology.With examplessuch as these,the sub- DUTHIE, R. B. AND A. B. FERGUSON. 1973. Mercer'sOrthopaedic Surgery.7thedition.Edward Arnold;London. ject cannotbe deemed unimportant! FISHER, R. A. 1958. GeneticalTheoryof Natural Selection. (2nd In short,the codification exaptationnot revisededition).Dover; New York. of M. a only identifies commonflaw in much evolu- FOUCAULT, 1965. Madness and Civilization.Random House; New York. of tionaryreasoning-the inference historical FOUCAULT, 1970. The Orderof Things. Random House; New M. current It York. genesisfrom utility. also focussesatT. tention but paramountrole FRAZZETTA, H. 1975. ComplexAdaptationsin Evolving Popuupon the neglected lations. 267 pp. SinauerAssociates;Sunderland, Massachusetts. in of nonaptivefeatures both constraining and GOULD,S. J. 1981. What happensto bodies if genesact forthemthe The argument selves?Nat. Hist. November. facilitating pathofevolution.

EXAPTATION

15
1979. Bird flight:how did it begin? Am. Sci.

GOULD,S. J. ANDR. C. LEWONTIN. 1979. The spandrelsof San Marco and the PanglossianParadigm: a critiqueof the adaptationistprogramme. Pp. 147-164. In: MaynardSmith,J. and R. Holliday,eds. The EvolutionofAdaptation Natural Selection. by R. Soc. London. HALSTEAD, L. B. 1969. The Pattern Vertebrate of Evolution.Oliver and Boyd; Edinburgh. HARRISON MATTHEWS, L. 1939. Reproduction thespotted in hyena Crocutacrocuta(Erxleben).Phil. Trans. R. Soc. (B) 230:1-78. in Cell. KLECKNER, N. 1977. Translocatable elements procaryotes.
11:11-23. KRUUK, H. 1972. The Spotted Hyena, a Study of Predationand

OSTROM, J. H.

67:46-56.

PATERSON, H. E. H. 1982. Species as a consequence of sex, in press. Am. Sci. PAUTARD, F. G. E. 1961. Calcium, phosphorus, and the originof PAUTARD, F. G. E. 1962. The molecular-biologic backgroundto PORTER, R. MS. Problems thetreatment 'madness'in English in of RACEY, P. A. AND J. C. SKINNER. 1979. Endocrineaspectsof sex-

backbones.New Sci. 12: 364-366.

the evolution bone. Clin. Orthopaed.24:230-244. of

science,medicineand literature the eighteenth in century.

Social Behavior. Univ. Chicago Press; Chicago, Illinois. KURTEN, B. 1980. Dance of the Tiger. Pantheon;New York. LEWIN, B. 1975. Units of transcription and translation.Cell. 4:7 7-93. McLACHLAN, R. AND R. LIVERSIDGE.1978. Roberts'Birds of G. SouthAfrica.4th edition(first publ. in 1940).JohnVoelckerBird Book Fund; Cape Town. MIVART, ST. G. 1871. On the Genesis of Species. MacMillan; London. OHNO, S. 1970. Evolutionby Gene Duplication. 160 pp. Springer; New York. ORGEL, L. E. AND F. H. C. CRICK. 1980. Selfish DNA: theultimate parasite.Nature. 284:604-607. OSTER, G. 1980. Mechanics,morphogenesis evolution. and Address to Conference Macroevolution, on October1980, Chicago. and the originof flight.Q. OSTROM, J. H. 1974. Archaeopteryx Rev. Biol. 49:27-47.

in ual mimicry spottedhyenasCrocutacrocuta.J. Zool. London. 187:315-326. RAuP, D. M. AND S. J. GOULD. 1974. Stochasticsimulationand evolution of morphology-towardsa nomothetic paleontology. Syst. Zool. 23:305-322. ROMER, A. S. 1963. The 'ancienthistory' bone. Ann. N.Y. Acad. of Sci. 109:168-176. SCOTT, J. D. AND N. B. B. SYMONS. 1977. Introduction Dental to Anatomy.Churchill Livingstone; London. SEILACHER, A. 1970. Arbeitskonzept Konstruktionsmorpholozur gie. Lethaia. 3:393-396. in SEILACHER, A. 1972. Divariate patterns pelecypodshells. Lethaia. 5:325-343. VRBA, E. S. 1980. Evolution, species and fossils:how does life evolve? S. Afr.J. Sci. 76:61-84. WILLIAMS, G. C. 1966. Adaptationand Natural Selection.Princeton University Press; Princeton, New Jersey.