invention by evolution: functional analysis in paleobiology

BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors nonprofit publishers academic institutions researchlibraries and research funders in the common goal of maximizing access to critical research

Invention by evolution functional analysis in paleobiologyAuthor(s) Roy E Plotnick and Tomasz K BaumillerSource Paleobiology 26(sp4)305-323 2000Published By The Paleontological SocietyDOI httpdxdoiorg1016660094-8373(2000)26[305IBEFAI]20CO2URL httpwwwbiooneorgdoifull1016660094-837328200029265B3053AIBEFAI5D20CO3B2

BioOne (wwwbiooneorg) is a nonprofit online aggregation of core research in the biological ecological andenvironmental sciences BioOne provides a sustainable online platform for over 170 journals and books publishedby nonprofit societies associations museums institutions and presses

Your use of this PDF the BioOne Web site and all posted and associated content indicates your acceptance ofBioOnersquos Terms of Use available at wwwbiooneorgpageterms_of_use

Usage of BioOne content is strictly limited to personal educational and non-commercial use Commercial inquiriesor rights and permissions requests should be directed to the individual publisher as copyright holder

q 2000 The Paleontological Society All rights reserved 0094-8373002604-0013$100

Copyright ( 2000 The Paleontological Society

Invention by evolution functional analysis in paleobiology

Roy E Plotnick and Tomasz K Baumiller

AbstractmdashFunctional analysis of fossils is and should remain a key component of paleobiologicalresearch Despite recently expressed doubts conceptual and methodological developments overthe past 25 years indicate that robust and testable claims about function can be produced Func-tional statements can be made in at least three different hierarchical contexts corresponding to thedegree of structural information available the position in the phylogenetic hierarchy and the de-gree of anatomical specificity The paradigm approach which dominated thinking about functionin the 1960s and 1970s has been supplanted with a methodology based on biomechanics Paleo-biomechanics does not assume optimality in organismal design but determines whether structureswere capable of carrying out a given function The paradigm approach can best be viewed as a wayof generating rather than testing functional hypotheses Hypotheses about function can also bedeveloped and supported by well-corroborated phylogenetic arguments Additional functional ev-idence can be derived from studies of trace fossils and of taphonomy New computer techniquesincluding lsquolsquoArtificial Lifersquorsquo studies have the potential for producing far more detailed ideas aboutfunction and mode of life than have been previously possible Functional analysis remains the basisfor studies of the history of adaptation It is also an essential component of many paleoecologicaland paleoenvironmental studies

Roy E Plotnick Department of Earth and Environmental Sciences University of Illinois at Chicago 845West Taylor Chicago Illinois 60607 E-mail plotnickuicedu

Tomasz K Baumiller Department of Geological Sciences and Museum of Paleontology University of Mich-igan 1109 Geddes Road Ann Arbor Michigan 48109-1079 E-mail tomaszBumichedu

Accepted 1 May 2000

Introduction

lsquolsquoA science of form is now being forged withinevolutionary theory It studies adaptation byquantitative methods using the organism-ma-chine analogy as a guide it seeks to reduce com-plex form to fewer generating factors and causalinfluencesrsquorsquo (Gould 1970 p 77)

lsquolsquoThe flowering of functional morphology hasyielded a panoply of elegant individual examplesand few principles beyond the unenlighteningconclusion that animals work well I at leastonce harbored the naive belief that a simple enu-meration of more and more cases would yield newprinciples for the study of form But Newtonianprocedures yield Newtonian answers and whodoubts that animals tend to be well designedrsquorsquo(Gould 1980 p 101)

lsquolsquo[W]e have placed unwarranted faith in ourunderstanding of the relationship between formand function Of particular concern arethe nature and accuracy of predictions of func-tion from morphology in fossil taxarsquorsquo (Lauder1995 p2)

Thirty years have passed since Gould ex-pressed his enthusiasm for human engineereddevices as guides to the study of fossil forms

and 20 years since his more pessimistic ap-praisal of functional morphology His laterskepticism apparently stemmed from a con-cern that although functional morphology isgood at determining functional design of par-ticular organisms the significance of thesestudies to understanding evolution is obscureLauder (1995) went further asserting that thestructure-function linkage is so weak that in-ferences of function from morphology in fos-sils are themselves questionable

These concerns make clear that we mustdistinguish between the methodology of func-tional morphology and its goals in order toproperly assess its importance in paleontolo-gy In this we follow Fisher (1985) who dis-tinguished functional analysis from functionalmorphology Functional morphology whichfocuses on the lsquolsquonature evolution and histor-ical consequences of adaptationrsquorsquo (Fisher 1985p 121) provides the historical context for theessentially ahistorical results of functionalanalysis

This paper will focus on methods and ap-proaches to functional analysis We will dis-cuss some recent methodological develop-ments provide an example of a taphonomic

306 ROY E PLOTNICK AND TOMASZ K BAUMILLER

approach to function critique the paradigmmethod and argue that although biomechan-ics (or paleobiomechanics) has proven to bethe best approach to the testing of functionalhypotheses it can be usefully supplementedby other methods We will then briefly discussthe broader implications and uses of function-al analysis including the insights it can pro-vide into the dynamics of the evolutionaryprocess

We will not be concerned here with recentdevelopments in other aspects of the lsquolsquoscienceof formrsquorsquo (Gould 1970) such as allometricstudies or theoretical morphology (McGhee1998) Instead we will focus on approaches tothe determination of both the lsquolsquofunctionrsquorsquo andlsquolsquobiological rolersquorsquo (sensu Bock and von Wahlert1965) of the preserved structures of fossil or-ganisms and on the evaluation of their evolu-tionary and ecological significance

The Form-Function Relationship in Fossils

Form and Function In a highly influentialpaper Bock and von Wahlert (1965) attemptedto clarify the terminology surrounding func-tional morphology Their key definitions are

1 Feature any part of an organism includ-ing morphological behavioral and physiolog-ical the structures of the organism are itsmorphological features

2 Form the appearance configuration com-position shape etc of a feature

3 Function what a feature does or how itworks includes chemical and physical prop-erties arising from its form A given featurecan have multiple functions

4 Faculty the combination of a given formand a particular function this is the lsquolsquoform-function complexrsquorsquo Faculty is defined aslsquolsquowhat the feature is capable of doing in thelife of the organismrsquorsquo(p 277)

5 Biological role how the organism uses thefaculty during its lifetime in the context of itsenvironment The same faculty can have mul-tiple biological roles Bock and von Wahlertstressed that the biological role cannot be pre-dicted with certainty from the study of formand function and must be directly observedThey considered this a particular problem for

fossil organisms Biological role generally cor-responds to the concept of lsquolsquolife habitrsquorsquo

In this context the shape of a birdrsquos wing ispart of its form the production of lift is one ofits functions the use of the wing for flight isa faculty and the use of flight to escape pred-ators is a biological role Similarly the ar-rangement of bones in a skull is a form theforces the skull can exert are a function theuse of these forces to bite is a faculty and thebiting of a prey animal is a biological role Thissequence also corresponds to the degree ofcertainty available to a paleontologist in afunctional analysis ie we can be quite cer-tain about the form model or test the functionmake reasonable hypotheses about facultiesand speculate about biological roles (eg Wit-mer and Rose 1991)

These definitions can provide a very usefulframework for functional interpretation of fos-sil organisms Nevertheless the Bock and vonWahlert definition of function referring to lit-tle more than the physical and chemical prop-erties of structure is perhaps unnecessarilynarrow In common usage the term functionencompasses their concepts of function and offaculty This paper will generally follow theusual practice and use the term in this broadersense In specific cases the distinction be-tween the two concepts will be made explicit

Resolving Lauderrsquos Dilemma Is Function Pre-dictable from Structure In two recent articlesLauder (1995 1996) expressed marked skep-ticism about the ability to decipher functionfrom structure These doubts are not based onan assumption that a relationship betweenform and function does not exist or cannot bedeciphered but that the required structuraldata to do so are rarely if ever available forfossil taxa In particular this implies that themost widely used fossil data in functionalanalysis skeletal morphology is of little directuse in interpreting fossil function We wouldargue that the situation is not quite so dire asLauder suggests mostly because his concernsare relevant only to a very specific subset ofstructurendashfunction problems

First Lauder (1995) uses a concept of func-tion different from that of Bock and von Wah-lert (1965) Lauder defines function as the me-

307FUNCTIONAL ANALYSIS IN PALEOBIOLOGY

chanical or physical role that a structure playsin the organism that is how a phenotypic fea-ture is used This definition seems closer toBock and von Wahlertrsquos concept of a facultyImplicitly Lauderrsquos view of function appearsmuch narrower than this referring to species-(or even population-) level statements aboutprecise patterns of structural kinematicsSince these kinematics are dependent on neu-romuscular features that cannot be observedin fossils Lauder inescapably rejects the pos-sibility of unequivocally predicting functionfrom form in fossils Functional studies wouldthus be restricted to direct observations in liv-ing organisms (Savazzi 1999)

Second Lauderrsquos admonition about theweakness of the link between morphologyand function is scale-dependent At the lowerhistological level and an upper lsquolsquogeneral levelof behavior and ecologyrsquorsquo he accepts that amuch tighter correlation between structureand function can be demonstrated For ex-ample at the histological level cross-sectionalarea of muscle is a good predictor of muscleforce and at the more general level accuratepredictions of habitat or diet can be madefrom analyses of structure His caution is re-stricted to what he refers to as the intermedi-ate level of generality that is where the neu-romuscular system interacts with the skeletalsystem to generate patterns of movement Atthat level predictions about patterns of move-ment require understanding the interactionsbetween the skeleton muscles and the ner-vous system For example a number of osteo-glossomorph fishes have the same evolution-ary novelty the tongue-bite Although thesetaxa have basically the same musculoskeletalsystems they have markedly different feedingkinematics due to differences in their nervoussystems Differences in kinematics of feedingin these fishes cannot be predicted from theirmorphology alone information on their ner-vous system is required More generally giventhat a particular morphology is consistentwith a wide repertoire of possible movements(ie functions in Lauderrsquos view) and that thestructure of the nervous system is unknown infossils function at this level is unrecoverablefrom fossil organisms

This is undeniably true but whereas Lauder

sees this as a lack of fit between structure andfunction we view it somewhat differently Thecritical issue is the level of desired precision ofthe functional analysis In the case of osteo-glossomorph fish the significant question isWhat is the functional significance of the pres-enceabsence of a tongue-bite As pointed outby Lauder (1995 p 7) the lsquolsquotongue-bite is asignificant evolutionary novelty that if pre-sent in an extinct taxon would provoke func-tional speculation and hypotheses as to its rolein the feeding mechanismrsquorsquo For a paleontolo-gist therefore the problem would be identi-fying functions that correlate with the pres-ence of a tongue-bite If it could be shown thatfishes with the tongue-bite exhibit character-istic feeding behaviors then regardless of therange of those behaviors a fit between struc-ture and function would have been estab-lished at this level of analysis

Has such a fit been established for Lauderrsquosexample Since lsquolsquoone could be reasonably con-fident given these results that a fossil taxonpossessing a tongue-bite morphology usedthese teeth to manipulate and puncture preyrsquorsquo(Lauder 1995 p 8) the presence of a tongue-bite indicates a limited and characteristic rep-ertoire of functions Lauder makes the glassappear half empty by stressing the impreci-sion inherent in inferring function from struc-ture because a structure can have a wide rep-ertoire of functions We see the glass as halffull in that when a unique morphology is ob-served it is predictably associated with sucha repertoire We would thus reinterpret Lau-derrsquos claims as a lesson about the precision ofthe link between structure and function Theprecision with which function can be inferredwill depend upon the amount of structural in-formation available with the degree of preci-sion of the functional statements increasing aswe progressively add more information aboutmuscles the nervous system etc Functionalstatements are thus hierarchical eg a generalstatement based on the skeleton alone includesa variety of more precise statements possibleif direct information on musculature was alsoavailable

Hierarchical approaches to biological sys-tems have received a great deal of attentionover the past several decades (eg Jacob 1977


Eldredge 1985 Allen and Hoekstra 1992 Val-entine and May 1996) As pointed out by Med-awar (1974 cited in Valentine and May 1996see also Jacob 1977) as one descends the ranksof a hierarchy the smaller becomes the scopebut the greater the complexity of the possiblephenomena For example lsquolsquobonersquorsquo can be con-sidered one rank of a form-function hierarchyAlthough bone is a complex tissue the varietyof bone morphology pales in comparison tothe diversity of structural elements that can bemade from it Functional statements aboutbone such as its compressive and tensilestrength are of much more general naturethan those about particular bones A hierarchyof anatomical specificity and correspondingfunctional statements can be identified egbonemdashvertebral bonemdashthoracic vertebraemdashfirst thoracic vertebra In the same way a hi-erarchy of functional statements can be basedon the amount of information available astatement based only on hard-part anatomy ismore inclusive of possible faculties than onethat includes additional anatomical informa-tion

Another form-function hierarchy parallelsthe phylogenetic one functional statementsbecome more specific as one descends the Lin-nean hierarchy In the same sense that a pa-leontologist might look at the diversity of fam-ilies or genera rather than species functionalstudies tend to focus on general attributescharacteristic of large taxonomic groups egpterosaurs (Padian 1991) stromatoporoids(LaBarbera and Boyajian 1991) or eurypterids(Plotnick 1985) A notable exception is Fisher(1977) who elegantly examined function in asingle species of horseshoe crab In additionespecially among invertebrate paleontolo-gists these functional analyses are generalstatements of life habits or function ratherthan specific statements of kinematics or be-havior (eg Baumiller 1990 Labandeira 1997)ie they are at Lauderrsquos upper hierarchical lev-el At these levels of analysis we believe theform-function relationship to be reliable

In sum functional statements can be madein at least three different hierarchical contextscorresponding to the degree of structural in-formation available the position in the phy-logenetic hierarchy and the degree of anatom-

ical specificity The situation thus is not quiteso dire as Lauder suggests mostly because hisconcerns are relevant only to a restricted cat-egory of structure-function problems Theseconcerns are however valid cautions againstoverestimating the precision of functional in-terpretations

Functional Analysis of Fossil Organisms

Phylogenetic Approach The interpretationof function in fossils has primarily followed avariety of inductive comparative approaches(Savazzi 1999) Principal among these hasbeen the comparison of homologous struc-tures in fossil and living organisms ie thefunction of a structure in an extinct organismis inferred to be similar to that of the homol-ogous structure in a living relative For ex-ample on the basis of their phylogenetic re-lationship with modern Limulus it has beensuggested that eurypterids swam on theirbacks an idea rejected by Plotnick (1985) onboth anatomical and hydrodynamic groundsCowen (1979) considered this approach themost reliable of those available

In recent years more explicit methods forinferring function in fossils using the homol-ogy approach have been formulated (eg Lau-der 1990 Weishampel 1995 Witmer 1995)These methods rely on the use of phylogeniesand they treat functions as traits and characteroptimization as a criterion for assessing thedistribution of these traits among taxa Thebasic premise is that genealogy can serve as aguide for reconstructing the unknown traits oforganisms Since functional characters can betreated as any other organismal attribute theyjust like structures may have synapomorphicapomorphic or plesiomorphic distributions

In principle the phylogenetic approach al-lows functional inferences to be made purelyby optimizing the functional characters on thecladogram no knowledge of the distributionof the functionally relevant morphologicaltraits is necessary By keeping function andform separate one may then use correlates ofform and function or other relevant informa-tion as independent tests for reconstructingfunction In the simplest case given an inde-pendently corroborated phylogenetic hypoth-esis for three taxa if taxa A and C share a


known function and phylogenetically brackettaxon B such that A represents the outgroupand C the sister taxon to B we may infer thatB shares the function of A and C This infer-ence can be further corroborated if a tightlinkage exists between a given structure andthe function in taxa A and C and if the struc-ture is also found in taxon B

It is clear that the phylogenetic methods of-fer little for structures and functions that areuniquely derived in fossils since they basetheir inferences on homologous structures andfunctions The chief danger of the argumentfrom homology is that a highly specific func-tion is assigned on the basis of a very generalhomology ie homologous structures oftenperform very different functions in even close-ly related organisms The more detailed thedescribed function and the more distantly re-lated the taxa are the more likely that the ho-mologous structure-function relationship willbreak down (Lauder 1995) Also for many pa-leontologically interesting questions such asabout flight in Archaeopteryx or pterosaursfunctionally bracketing the fossil taxa is notpossible

Another danger is that the wrong homologymay be used For example Jacobs and Land-man (1993) strongly questioned the commonuse of Nautilus as a model for the life habits ofammonoids Instead they pointed to phylo-genetic evidence for a coleoid-ammonoid re-lationship and suggested that the biology ofammonoids be interpreted on that basis Pur-nell (1999) described similar problems withthe interpretation of conodont elements

Analogy and the Paradigm Approach If suit-able homologies are not available the tenden-cy has been to argue for function based onanalogy usually biological Radinsky (1987)termed this the lsquolsquoform-function correlation ap-proachrsquorsquo It assumes that a close relation existsbetween form and function so that the lattercan be predicted from the former For exam-ple as discussed by Radinsky (1987) since ex-tant animals with long legs are usually fastrunners it is reasonable to assume that extinctanimals with long legs whether or not theyare related to modern forms also ran fast(note that lsquolsquorunning fastrsquorsquo is a faculty) This ap-proach also underlies Stanleyrsquos (1970) classic

analysis of the relationship between bivalveshell form and life habit and Labandeirarsquos(1997) interpretation of insect feeding mech-anisms based on mouthpart morphology

When biological analogues are not avail-able mechanical ones have often been usedFor example Cowen (1975) argued for a lsquolsquoflap-ping valversquorsquo in richtofeniacean brachiopodsbased on an analogy with a single-valvedpump (cf Grant 1975) Similarly Myhrvoldand Currie (1997) using analogy with whipssuggested that sauropod tails cracked andwere used in communication

The identification of a functional analoguefor a structure in a fossil is a hypothesis thatmust be tested By far the most influential con-ceptual approach to the functional morphol-ogy of extinct invertebrates the lsquolsquoparadigmrsquorsquoapproach of Rudwick (1964) has been sug-gested as a way to carry out such a test Theparadigm method was extensively describedin older reviews of the field (eg Raup 1972Gould and Lewontin 1979) and remains prom-inent in more recent articles and textbooks(Hickman 1988 Lauder 1995 Prothero 1998Moon 1999 Paul 1999)

Rudwickrsquos (1964) original statement of theconcept was that a paradigm is lsquolsquothe structurethat would be capable of fulfilling the functionwith the maximum efficiency attainable underthe limitations imposed by the nature of thematerialsrsquorsquo (p 36) Function is used here in thebroader definition ie as essentially synony-mous with faculty This approach involvesseveral steps

1 A function is suggested for a morphologicfeature perhaps based on analogy with a liv-ing organism or with a mechanical device

2 From a knowledge of engineering and ofthe nature of the biological materials involveda paradigm is developed for the performance ofthis function The paradigm is thus a model(in Rudwickrsquos term a lsquolsquostructural predictionrsquorsquo)of the optimum structure for the performanceof the function

3 The paradigm is compared with the ob-served structure The degree of correspon-dence between the two acts as a test of the par-adigm as a functional hypothesis The expec-tation is that if the paradigm is valid and no


other constraints hold the paradigm and thestructure will closely agree

4 Each alternative function for a given fea-ture generates its own paradigm The functionwhose paradigm most closely matches the ob-served structure would have been lsquolsquofulfilledmost effectivelyrsquorsquo by the structure

An often overlooked point is that Rudwick(1964) in his discussion of the paradigm ap-proach indicated that the comparison be-tween paradigm and structure shows whetherthe structure would be capable of performingthe function but lsquolsquocannot however establish infact that it did fulfil that functionrsquorsquo (p 38)

The paradigm approach has been bothstrongly criticized (Grant 1972 1975 Lauder1995) and defended (Cowen 1975 Paul 19751999 DeMar 1976 Fisher 1985) The key ob-jections can be summarized as follows

1 The paradigm method assumes that nat-ural selection produces an optimal structurefor a particular function This assumption issuggested to be invalid since other factors in-cluding developmental constraints and phy-logenetic history can exert comparable con-trol over morphology (Seilacher 1970 Grant1972 Signor 1982 Seilacher and LaBarbera1995) Implicitly this concept is contained inthe lsquolsquolimitations imposed by the nature of thematerialsrsquorsquo of the original formulation of theparadigm concept

2 There may be multiple possible structuraloptima ie alternative equally (or nearlyequally) valid paradigms could exist for a sin-gle function (Signor 1982)

3 Competing functional requirements(lsquolsquotrade-offsrsquorsquo) may produce suboptimal struc-tures for the functions considered indepen-dently (a point conceded by Rudwick [1964])

4 The comparison between the paradigmand the observed structure is essentially vi-sual and qualitative the lsquolsquotestrsquorsquo is thereforesubjective (Signor 1982)

5 Paradigms rely too heavily on analogiesto mechanical devices they thus tend to over-look important biological factors such asphysiology (Grant 1972 1975)

Grant concluded that the paradigm meth-od as stated by Rudwick is a lsquolsquopoint of view

an approach it is not a complete meth-odology rsquorsquo Signor (1982) was far more crit-ical suggesting that it should be used only ifother methods are not available

Despite these criticisms the paradigmmethod still has proponents Even Grant(1972) in a generally critical article labeled itlsquolsquoa watershed in the conceptual methodologyof invertebrate paleontologyrsquorsquo (p 236) Paul(1999) asserted that it was a simple step-by-step approach that allows the rejection of in-appropriate hypotheses and makes it possibleto compare competing hypotheses We will ar-gue below that the value of the paradigmmethod is as a source of testable hypothesesfor function not as the test itself Before we doso we want to correct two claims about themethod that we believe to be misconceptionsWe call these the Fallacy of the Perfect Engineerand the Fallacy of the Mechanical Analogy

The Fallacy of the Perfect Engineer FrancoisJacob in his generally overlooked essay lsquolsquoEvo-lution and Tinkeringrsquorsquo (Jacob 1977) pointedout several differences between the process ofnatural selection and actions of an engineerOne of these is that lsquolsquothe objects produced bythe engineer at least by a good engineer ap-proach the level of perfection made possibleby the technology of the time In contrast evo-lution is far from perfectionrsquorsquo (p 1161) As dis-cussed above the lsquolsquoimperfectionrsquorsquo of morphol-ogy produced by biological evolution is a keypart of much of the criticism of the paradigmmethod (eg Gould and Lewontin 1979) Lat-er on in comparing the action of evolution tothat of a lsquolsquotinkererrsquorsquo rather than an engineerJacob stated lsquolsquoUnlike engineers tinkerers whotackle the same problem are likely to end upwith different solutions This also applies toevolution rsquorsquo (p 1164) On this premise Ja-cob made a strong case for the importance ofcontingency in evolution Again there is theexplicit assumption that a trained engineerwill wind up with a perfect optimal productThis idea that an engineering approach willunerringly produce the optimal form to solvea given functional problem has been funda-mental to the paradigm method We arguethat this concept which we term lsquolsquothe fallacyof the perfect engineerrsquorsquo is both false and mis-leading


We base our argument on a reading of thenontechnical literature of engineering espe-cially the popular works of civil engineer Hen-ry Petroski (1985 1993 1996) Petroski has ex-tensively analyzed the engineering designprocess used in producing such mundaneitems as paper clips and can openers and suchspectacular items as bridges One of his keyobservations (Petroski 1993) is that even forengineers form does not follow function In-stead form follows failure ie engineering de-sign advances by recognizing the limitationsof existing products New forms develop as anattempt to overcome these failures Implicit inthis is the concept that very few if any hu-man-designed objects are optimal for theirtask There is always room for improvement

In addition human-engineered objects aresubject to many of the same kinds of con-straints and influences that Seilacher (1970)recognized for biologically evolved forms Forinstance there is clear evidence for a form ofphylogenetic constraint in engineering de-sign The design maxim known as MAYAlsquolsquomost advanced yet acceptablersquorsquo (Petroski1996) indicates that new designs cannot betoo radically different from existing forms orthey wonrsquot be adopted Many details of thefirst iron bridges closely resembled those oftheir wooden predecessors even though thiswas not required by the nature of the materi-als (Petroski 1996) In addition even lsquolsquoidealrsquorsquoengineering objects such as the standard pa-per clip (Petroski 1993) have identifiableshortcomings These shortcomings resultfrom such factors as limitations imposed bythe nature of the materials competing func-tional requirements or simply design mis-takes (Dennett 1998) And of course as arguedby Gould and Lewontin (1979) many featuresof human-made structures are not lsquolsquoadaptiversquorsquobut inescapable side effects of how the struc-ture must be built The products of engineer-ing design cannot be considered as unerringlyoptimal for the same reasons that organic de-sign cannot They thus should not be used astests of functional hypotheses for extinct orliving organisms

Interestingly as pointed out by Vogel(1998) many of those who have previouslyrecognized the failings of human design have

pointed to nature as providing examples ofdesign excellence Manned flight provides anexcellent example of where a too slavish at-tempt to copy nature by the construction ofornithopters led to a technological dead end

In summary the concept that the engineer-ing design process leads to optimality where-as evolution does not is incorrect In fact theremay be more similarities between the twothan has been generally accepted In both his-torical legacies material constraints costs inproduction of different structures under dif-ferent conditions and competing functionalrequirements mean that the concept of globaloptima is less useful than that of optimiza-tionmdashthe climb to local peaks on an everchanging landscape Nevertheless the simi-larities between the engineered and theevolved cannot be carried too far althoughthe processes may be similar we will argue inthe next section that the failures of organism-machine comparisons stem from the use of toostrict an analogy between the results of naturaland human design

The Engineered and the Evolved The Fallacy ofthe Mechanical Analogy Implicit in the para-digm method and in other discussions offunctional interpretation (eg Gould 1970Cowen 1975 Frazetta 1975 Hickman 1988) isthe use of analogy ie the comparison of theobserved structure with lsquolsquosimple machinesarchitecture industrial design and otherman-made systems designed for efficient andcost-effective functionrsquorsquo (Hickman 1988 p782) Recent examples include the comparisonof sauropod tails with bullwhips by Myhrvoldand Currie (1997) and the lsquolsquoammonites as Car-tesian diversrsquorsquo hypothesis of Seilacher andLaBarbera (1995 cf Jacobs 1996)

The difficulty with this approach was co-gently stated by Wainwright (1988 p 8)lsquolsquoMan-made buildings are large dry rectan-gular rigid and static In comparison plantsand animals are small damp cylindrical flex-ible and dynamicrsquorsquo As discussed by Vogel(1998) and Dennett (1998) the technology ofnature and human technology have far moredifferences than similarities these differencesspring from both the nature of the materialsand the design process They include the fol-lowing


1 Unlike many manufactured artifactsthere are very few corners or right angles innature organisms tend to favor round surfac-es and cylindrical shapes

2 Units of engineered structures tend be ho-mogeneous whereas biological units are in-ternally variable (ie individual steel beamshave the same physical properties throughoutbut individual bones or crab sclerites have re-gions with different composition and organi-zation)

3 Metallic materials are absent in organ-isms

4 Very few organisms roll and the wheeland axle are essentially absent in the livingworld (LaBarbera 1983)

5 Human artifacts are designed to be stiffand are consequently often brittle organismaldesign favors strength over stiffness and thusproduces toughness

6 In most complicated mechanical deviceseach separate part usually performs one ortwo discrete functions multiple functions foreach part are rare For example in a computerprinter the paper feeder the drum unit andthe output tray each perform a separate andsingle role In contrast in biological systemsthe same feature can perform multiple func-tions (eg the jaw) and many functions areperformed by the joint action of many struc-tures

Obviously the list can go on the reader isreferred to Vogel (1998) for a far more com-plete rendition The essential point is simplythat most machines make poor analogues toliving organisms The use of engineeringstructures as analogues to biological systemsis fraught with difficulties and must be usedwith extreme caution

Paleobiomechanical Approach Our discus-sion of the machine-organism analogy maysound pessimistic but by becoming cognizantof the very real differences between machinesand organisms we can focus on their similar-ities These similarities as pointed out by Vo-gel come from lsquolsquoinescapable physical rulesand environmental circumstancesrsquorsquo (p 292) Itis the relationship between these physicalrules and organisms that is the foundation of

the paleobiomechanical approach to function-al analysis

At its most fundamental biomechanics ex-amines the interrelationships between biolog-ical structures and physical processes (cf def-inition in Rayner and Wootton 1991) The as-sumption is that such factors as the strengthof biological materials the kinetics of linkedmechanisms fluid drag and lift and diffusionall have directly observable and measurableconsequences on both the possible facultiesand the biological roles of morphological fea-tures (Wainwright et al 1976 LaBarbera 1990Vogel 1994) Biomechanics thus allows us toquantify the functional properties of biologi-cal structures and thus test their effects on fac-ulties and biological roles

Paleobiomechanics therefore is simply theuniformitarian extension of this the conse-quences of physical processes existed to thesame extent in the past as they do today (Al-exander 1989) As a result the principles ofphysics that describe bird bat and airplaneflight can be used to understand pterosaurflight (Padian 1991) The dynamics of wavesalong rocky coasts were the same in the De-vonian as today and thus had the same influ-ence on morphology (Denny 1995) The bio-mechanics of extinct organisms is thus one ofthe only areas within paleontology amenableto direct experimental investigation (taphon-omy is another)

Paleobiomechanics also does not require theexistence of a living homologue or living ormachine analogue (Radinsky 1987 Witmerand Rose 1991) although one can be suggest-ed as a starting point of the analysis Insteadprinciples of physics and engineering are di-rectly applied to the observed structure to in-fer its function and faculty as we will discussbelow this is directly comparable to the prac-tice of reverse engineering

We can summarize the paleobiomechanicalapproach as follows it is clearly derived fromthe paradigm method but does not rely on theflawed assumptions of that approach

1 A possible faculty (not a biological role)for a structure is proposed This proposal is ahypothesis that could be derived from ho-mology or analogy For example it is straight-


forward to hypothesize that the forelimbs ofpterosaurs and Archaeopteryx were used forflight (Padian 1991 Rayner 1991)

2 The hypothesized faculty is then used tomake a prediction of function (sensu Bock andvon Wahlert 1965) and of form If for examplethe wing of Archaeopteryx was used for flightthen it should have been capable of generatingsufficient lift to support the weight of the an-imal It should also have a form consistentwith the production of that lift (Rayner 1991)eg a cambered wing produces more lift thanone with a symmetrical cross-section (Vogel1998) Similarly the aerodynamics of flappingflight predict a large discrepancy in musclemass between downstroke and upstroke mus-cles (Greenewalt 1975)

3 A model either physical or computer-based (see below) is produced that allows theexperimental determination of the structurersquosfunction (sensu Bock and von Wahlert 1965)and a test of whether the observed structureis capable of carrying out the hypothesized fac-ulty In the case of Archaeopteryx one couldplace a model of the wing in a wind tunneland measure the amount of lift produced Ifthe measured lift proves sufficient to supportthe weight of the animal in air under a rea-sonable set of conditions the wingrsquos hypoth-esized faculty has not been rejected Note thatthe question is not whether the wing of Ar-chaeopteryx is optimally designed but whetherthe actual wing of the animal could producesufficient lift to overcome its weight and bodydrag

4 If direct experimental tests are not pos-sible or in addition to experiments predictedaspects of form are compared with the ob-served form Is the wing of Archaeopteryx cam-bered Does the skeleton reflect greater down-stroke muscle mass Again the goal is not see-ing whether the form is the optimal for a par-ticular function (lift generation) but insteadwhether the form has characteristics that areassociated with performance of the functionA structure may indeed be optimal but this isa hypothesis to be tested rather than an as-sumption of the approach

A superb example of the paleobiomechani-cal approach to function and faculty is the

study of the skull and jaw of Diatryma by Wit-mer and Rose (1991) They began by pointingout that there are no modern avian analoguesto the Diatryma so that the form-function cor-relation approach (Radinsky 1987) could notbe used Instead they utilized beam theory topredict what design features a bird skullshould have to maximize biting forces and de-cided that lsquolsquoDiatryma exhibits virtually all ofthe predicted featuresrsquorsquo (p 103) They conclud-ed that the jaw apparatus of Diatryma was ca-pable of exerting tremendous bite forcesCommendably they clearly distinguish the bi-ological role of the jaw (eg herbivory or car-nivory) from its function and faculty (lsquolsquoWhat-ever Diatryma ate it could bite hardrsquorsquo [p 117])After analyzing food availability and consid-ering the requirements of jaw form imposedby eating vegetation versus meat and bonesWitmer and Rose then interpreted the bird asa carnivore The forces generated by the mech-anism are functions these functions are ap-propriate for crushing certain objects in biting(faculty) this allowed Witmer and Rose toevaluate the biological role of Diatryma jaws byconsidering objects that could actually becrushed

In another example Plotnick and Baumiller(1988) examined two alternative hypothesesfor the faculty of the wide flat telson of pter-ygotid eurypterids Each of these hypothesesmade testable predictions about the morphol-ogy and function of the telson and of the restof the animal The first hypothesis was that thetelson actively flapped and acted to propel theanimal similar to the caudal fins of cetaceansThis hypothesis of faculty predicts morpho-logic features such as large condyles for flex-ibility large muscle insertions and a high as-pect ratio (width to length) for the telsonNone of these features are found in pterygo-tids The second hypothesis of faculty was thatthe telson was used to steer the animal that itacted as a rudder The functional properties ofa rudder require numerous characteristics andthese can be used to make morphologic andallometric predictions All of these predictionsare consistent with the observed features ofpterygotids and their telsons Further a com-parison of alternative telson designs showedthat the observed morphology produced


greater steering forces than the alternativessuggesting that it approaches an optimal de-sign

In a study of crinoid functional morpholo-gy Baumiller (1992) examined the hypothesisthat lift on the crinoid crown was sufficient tomaintain the position of the crown above thesubstrate ie that the faculty of the crinoidcrown was to act as a kite with the stem actingas the string that tethered it to the substrateHis experiments allowed him to estimate thelift that the crowns of two representative cri-noid genera may have experienced These re-sults combined with analyses of the otherforces (weight drag) acting on the crinoid ledBaumiller to conclude that the kite hypothesiswas untenable at the level of function andtherefore of faculty and role

As a final example in a pair of studies Bo-yajian and LaBarbera (Boyajian and La-Barbera 1987 LaBarbera and Boyajian 1991)explored alternative hypotheses for the sys-tematics and paleoecology of stromatopo-roids specifically the role of the astrorhizaeBoyajian and LaBarbera (1987) examinedwhether the astrorhizae represented an excur-rent canal system for a filter feeder similar tothose in living sclerosponges Using scalemodels they studied flow patterns throughastrorhizae and concluded they were indeedable to function as excurrent canals Theycompared alternative hypotheses for astror-hizae based on alternative predictions madefor the distributions of canal diameters (La-Barbera and Boyajian 1991) Their results al-lowed them to reject several hypotheses butwere consistent with the interpretation of theastrorhizae as sponge-like mass transport sys-tems Their conclusion of a close relationshipbetween stromatoporoids and sponges sug-gests how functional analysis could be used asa tool for phylogeny reconstruction

Functional Analysis As Reverse EngineeringOne noticeable similarity between functionalanalysis particularly paleobiomechanics anda human design process is to reverse engineer-ing (Petroski 1996 Dennett 1998) In reverseengineering a product is disassembled fre-quently by a business competitor to deter-mine how it works and how it might be du-plicated This is perhaps most common today

for software where executable code is reverseengineered to obtain the original program-ming Reverse engineering can be distin-guished from lsquolsquoforwardrsquorsquo engineering whichcreates the object

Cognitive scientist D C Dennett (1999 p256) pointed out that lsquolsquoin spite of the differ-ence in the design processes reverse engi-neering is just as applicable a methodology tosystems designed by Nature as to systems de-signed by engineersrsquorsquo If reasons for subopti-mality and historical contingency are recog-nized the techniques of reverse engineeringshould lead to a sound understanding of thedesign of organisms In fact Dennett (1999 p256) even went on to redefine biology as thelsquolsquoreverse engineering of natural systemsrsquorsquo Forliving things evolution is the forward designprocess

A key similarity between the reverse engi-neering of extant organisms and that of engi-neered systems is that the lsquolsquopurposersquorsquo of the re-verse-engineered entity is already known thegoal is to uncover the specific way it performsthis purpose We already know that a com-puter printer produces printed output andthat a bird flies in order to find food On theother hand implicit in the methodology of re-verse engineering is the determination of howa particular part of the device performs itsrole eg How much current does a particularcircuit carry How much lift does a particularairplane wing design generate How muchbending can a particular strut withstand Thepurpose of a reverse-engineered object thuscorresponds to the biological role of Bock andvon Wahlert (1965) whereas the propertiesuncovered by reverse engineering correspondto their definitions of function and faculty

Paleontologists in contrast are in a positionsimilar to those who try to uncover the oftenforgotten uses of obsolete tools and utensils(Petroski 1993) From familiarity with similarobjects (ie either through homology or anal-ogy) a purpose can be surmised We then lsquolsquore-verse engineerrsquorsquo the fossil to see if it could car-ry out the surmised purpose The critical pointhere is that we do not ask Was it the best struc-ture to do the assumed task but Could it havebeen at least minimally capable of carryingout this task Physical rules may be inescap-


able but they are not dictatorial multiple al-ternatives could exist to perform the samefunction

Real Animals in a Virtual World Computersand A-life No survey of any recent develop-ment in science is complete without a discus-sion of the role played by computers Com-puters allow the detailed examination of farmore complex systems than are generallyamenable to standard experimental methodsFour areas can be identified in which func-tional analysis has benefited or could benefitfrom the new technology kinematics of com-plex skeletal systems finite element analysisof stress and strain computational fluid me-chanics and artificial life

Vertebrate kinematics are usually studiedby manipulation of bones or models of bonesThis is often impractical because there may bemany separate elements their pattern of con-nections could be complex or the bones them-selves might be very large (Stevens and Par-rish 1999) As a result investigators have be-gun to use software similar to that used by en-gineers to model these systems One exampleis the previously cited work of Myhrvold andCurrie (1997) on sauropod tails Morphologicinformation on sauropod tail vertebrae wasinput into a physics-based simulation pro-gram The flexibility and possible velocity ofmotion of the tail along its length were mod-eled leading to the conclusion that the tip mayhave moved at supersonic speeds Another ex-ample is the work of Stevens and Parrish(1999) who examined the flexibility and pos-ture of the sauropod neck They decided thatthe neck was markedly less flexible than pre-viously suggested

Finite element analysis is an important andwidely used technique among engineers andphysicists (Huebner et al 1995 Gershenfeld1999) It is a method for finding approximatesolutions for the values of variables such asstress within a complexly shaped surface orvolume It does this by dividing (discretizing)the region into contiguous pieces or elementsand solving the relevant equations (usually apolynomial) within each element separatelyThe results from all elements are then assem-bled to produce a set of simultaneous equa-tions that describe the behavior of the desired

property for the entire region The equationsare then solved for a specific set of boundaryconditions (Huebner et al 1995) Finite ele-ment methods have the potential for studyingfar more complex structural situations thanare usually amenable to direct experimentalanalysis

Recent applications of finite element meth-ods in functional analysis include Philippiand Nachtigall (1996) Daniel et al (1997) andKesel et al (1998) Kesel et al analyzed thedistribution of material in the wings of drag-onflies and flies and examined the function ofwing veins for stiffening Philippi and Nach-tigall investigated the distribution of forces inthe test of regular echinoids under differentloadings and rejected the pneu hypothesis fortest shape

A specifically paleontological applicationwas Daniel et al (1997) They used finite ele-ment analysis to examine the distribution ofpressure stresses in ammonoid septa Theytested the idea that greater septal complexityallowed greater resistance to hydrostatic pres-sure so that sutural complexity correlatedpositively with greater depth during lifeTheir results suggested that highly complexsutures actually lead to diminished resistanceto hydrostatic pressures

The use of computer models that incorpo-rate the basic principles of fluid dynamicssuch as the Navier-Stokes equations (Vogel1994) is also possible These computationalfluid dynamics (CFD) models often rely on thefinite element approach discussed above(Huebner et al 1995) These models make itpossible to analyze the fluid flow around com-plex objects and have become extensivelyused in the aircraft industry replacing moretraditional physical modeling approaches (Pe-troski 1996) We are aware of no studies ap-plying these models to functional analysisbut they clearly have great promise

One example of a computer study of fossilfluid mechanics is that of Knight (1996) Hedeveloped a computer model that incorporat-ed the equations for lift and drag These werethen used to study how eurypterids may haveswum He suggested that lift was the primarymechanism for eurypterid swimming

An especially exciting recent development


in computer science with direct implicationsfor paleontology and the functional analysisof fossils is lsquolsquoartificial lifersquorsquo often called AL orA-life (Plotnick 1997 Dennett 1998) Ray(1994 p 179) one of its leading developersstated that lsquolsquoArtificial Life (AL) is the enter-prise of understanding biology by construct-ing biological phenomena out of artificialcomponents rather than breaking natural lifeforms down into their component partsrsquorsquo Thebasic approach of A-life is to create entitiesthat possess properties and operate underrules similar to those of biological entities andsystems For example an lsquolsquoindividualrsquorsquo in anA-life system which is in reality a string ofcomputer code can reproduce (the code du-plicates) and mutate (the code can change dur-ing replication) Individuals may also be al-lowed to mate (code pieces are exchanged) tofind resources and to die A group of similarindividuals (ie a lsquolsquospeciesrsquorsquo) can thus be sub-ject to natural selection Even given a simpleset of rules and properties highly complex be-haviors can emerge It is this development ofemergent system properties that is character-istic of A-life Dennett (1998 p 256) describedit as a form of lsquolsquobottom up reverse engineer-ingrsquorsquo

Two A-life projects particularly relevant tofunctional analysis are Karl Simsrsquos lsquolsquocrea-turesrsquorsquo (Sims 1994) and the lsquolsquoartificial fishrsquorsquo ofTerzopoulos and colleagues (Terzopoulos etal 1994 1996) Simsrsquos creatures are morpho-logically simple virtual organisms made upof rectangular blocks of various sizes Theyare supplied with basic control systems andoccupy a virtual environment with realisticphysical laws (eg gravity fluid mechanics)The codes for these organisms which describetheir form and their behavior can reproduceand mutate Natural selection is introduced byallowing only those forms that are best able toperform a task in the environment such asswimming to survive Although morpholog-ically crude the resulting creatures exhibit awide range of plausible behaviors includingundulatory swimming and sculling These be-haviors were not introduced but arise as aconsequence of the interaction of the evolvingvirtual life forms with their environment

The goal of Terzopoulos and his colleagues

is to produce visually realistic virtual organ-isms that are self-animating that is they aregiven a set of rules and behaviors and then actautonomously rather than being directed by aprogrammer To this end they designed arti-ficial fishes whose external morphology isbased on living examples and whose bodymovements are controlled by relatively real-istic representation of the skeletal and neuro-muscular systems These virtual fish combinesimple yet realistic algorithms for biomechan-ics (including fluid mechanics) perceptionand learning (Grzeszczuk and Terzopoulos1995) Movements of their bodies allow thefish to move in their environment thosemovement patterns that lead to faster move-ments are kept whereas other movement pat-terns are discarded As a result the artificialfish lsquolsquolearnrsquorsquo how to swim The resulting move-ment patterns closely resemble those seen inbiological fish A similar learning pattern re-sults in fish that pursue lsquolsquopreyrsquorsquo

The methods developed by Terzopoulosand Sims have tremendous potential forstudying function in fossils In particular theymay lead to at least a partial solution to Lau-derrsquos (1995) concerns about kinematics of fos-sil forms We envision for example a virtualfish based on the anatomy of such organismsas Silurian ostracoderms A virtual experi-ment is then conducted in which the artificialostracoderm fish learns to swim If properlydesigned a set of virtual experiments coulddetermine if there are uniquely predicted ki-nematics for ostracoderm swimming or ifthere is a range of equally likely alternatives

Ichnological and Taphonomic ApproachesOther types of paleontological data besidesmorphology are relevant to reconstructingfossil function and life habits Although theinability to test function in extinct organismsdirectly is sometimes viewed as a limitation itprovides for paleontologists an opportunityto find different means of extracting function-al information from the fossil record (Savazzi1999) In particular data from trace fossils andtaphonomy have great potential for develop-ing and testing functional and life-habit hy-potheses

An extensive body of literature on ichnofos-sil characterization and interpretation exists


FIGURE 1 A Distribution of the stalk-shedding function and life habit among extant crinoids B Stratigraphicranges and inferred relationships of the five extant crinoid groups and the Triassic holocrinids (modified fromSimms 1999) The position of the bourgueticrinids is controversial they are thought to be either a neotenous off-shoot of the comatulids (Simms 1988) or a subgroup of the millecrinids (Gislen 1938 Roux 1978)

(eg Bromley 1990) and we will not review itin detail here Ichnofossils have provided in-formation on modes and speeds of locomotionin groups such as arthropods (Briggs et al1991) and vertebrates (Alexander 1989) Ich-nofossils have also been used for the interpre-tation of life habit (eg suspension versus de-posit feeding) or behavior of extinct organ-isms (Seilacher 1964 Bromley 1990) Obvious-ly the use of trace fossils in functionalinterpretation requires that the trace maker becorrectly identified

Ichnofossils often act as tests of hypothesesof kinematics based on limb morphology One

especially exciting study is that of Gatesy et al(1999) who analyzed a suite of footprints ofTriassic theropods Combining detailed ex-amination of the tracks studies of modernbird locomotion and computer graphics theyproduced a detailed reconstruction of the footmovements of these dinosaurs

Taphonomy is another source of paleonto-logical data that can provide critical insightsinto function either directly or by providingcritical information on soft-tissue morpholo-gy Although biostratinomic processes are of-ten viewed as filters that remove information(Behrensmeyer and Kidwell 1985) decay and


fossilization can also leave signatures of lifehabit or function that would otherwise be un-available For example the preserved post-mortem gape of extinct lamellibranchs pro-vides information relevant to reconstructingsoft-tissue distribution (muscles and liga-ment) in these organisms This information inturn constrains hypotheses of function

Taphonomic information used in recon-structing soft tissues and function in extincttaxa may even be used to overturn skeletalproxies One example is the analysis of stalkfunction in fossil crinoids Two of the five ex-tant groups of crinoids comatulids and iso-crinids are capable of freeing themselvesfrom the substrate and crawling or swimming(Fig 1) (Messing et al 1988) The functionalconsequences of a free-living versus fully ses-sile life habit may have profound ecologicaland evolutionary implications (Meyer and Ma-curda 1977) so it is important to determinethe distribution and origin of the free-livinghabit within the post-Paleozoic clade

In all extant crinoids the juveniles are ce-mented by a holdfast to the substrate Thefree-living habit thus requires that the animalbe capable of shedding its stalk at some stageof life In comatulids the stalk is shed in earlyontogeny and is never regenerated In isocrin-ids the stalk continues to grow throughoutlife but as new elements are added in theproximal portion the older distal end of thestalk is shed Stalk shedding in isocrinids oc-curs at specialized rupture points that arespaced at regular intervals along the length ofthe stalk These rupture points possess a char-acteristic skeletal morphology and specializedligamentary organization that are not found atthe non-shedding articulations

Holocrinus is the first crinoid to appear fol-lowing the Permo-Triassic extinction and thesister taxon of the entire isocrinid-comatulidclade (Simms and Sevastopulo 1993 Hagdorn1995) This genus lacks skeletally differenti-ated articulations Since ligament organiza-tion cannot be studied directly in fossils theskeletal data suggest that Holocrinus was in-capable of shedding its stalk and that thisfunction is derived in the comatulid-isocrinidclade

However stalk shedding also produces a

characteristic non-random pattern in the shedstalk segments (Baumiller and Ausich 1992Baumiller et al 1995) This pattern can be usedas an independent taphonomic test of theshedding function An analysis of Holocrinusstalk segments revealed just such a pattern(Baumiller and Hagdorn 1995) Thus Holocri-nus was capable of stalk shedding despite thefailure of skeletal morphology to reflect thisspecialization

The original functional hypothesis for Hol-ocrinus was based on a correlation betweenmorphology (articulation type) and function(shedding ability) in extant taxa and was over-turned using taphonomic criteria This ex-ample thus appears to support Lauderrsquosclaims about a lack of fit between structureand function In contrast we believe it conveysa different and instructive message that bas-ing claims of a linkage between structure andfunction solely on correlation can lead tofaulty conclusions not only about a specific sit-uation but also about the general fit betweenfunction and structure For example since al-most all airplanes have wheels using onlycorrelation we might construe a link betweenwheels and flight Seeing a wheel-less sea-plane could then lead to the conclusion thatthe link between structure and function wasweak Obviously this conclusion would be in-correct because the wrong structure-functioncouplet was initially selected That is also whywe are skeptical of studies concluding a gen-eral lack of structure-function fit from exam-inations only of congruence between structureand function without exploring the biome-chanical linkage between function and struc-ture (see Lauder 1995 his Case Study 2) Inlinking function and structure it is critical todevelop and test biomechanically how a mor-phological feature affects function (Lauder1991) The goal of functional morphologyshould be not simply to find correlations be-tween structure and function but to find caus-al explanations for them

The Roles of Functional Morphology inPaleobiology

In the previous sections we have discussedsome of the available methods for reconstruct-ing function in fossils and for getting at the


link between structure and function In thissection we will show that functional interpre-tations remain at the core of many areas of pa-leobiological research and that their uses gobeyond demonstrating good design and cur-rent utility Paleobiologists attempt to inter-pret patterns in the history of life in functionaland ecological terms Implicitly function iscausally connected with the origin of pheno-types evolutionary trends evolutionary prop-erties of taxa and long-term changes in thestructure and dynamics of the biosphere Inaddition the association of functions and lifehabits with morphology at whatever hierar-chical level is an essential part of paleoecol-ogy including the distribution of organismsamong environments and the distribution ofpaleocommunities As pointed out by Boucot(1990) in his extensive review functional anal-ysis is also the best available method for thestudy of the evolution of behavior in the fossilrecord The underlying assumption of theseapproaches is that functional attributes of in-dividuals populations species and cladescan have ecological and evolutionary conse-quences A few examples will suffice to illus-trate the role that functional interpretationsplay in paleobiological research

Jacobs et al (1994) used a well-known bio-mechanical principle and experiments to ex-plain the pattern of distribution of differentammonite morphs among facies They showedthat more spherical less compressed morphsexperience a lower total drag under condi-tions when drag is dominated by frictionalforces (small Reynolds number) than do com-pressed streamlined morphs The opposite istrue under conditions when drag is dominat-ed by dynamic forces (large Reynolds num-ber) Since overcoming drag is energeticallyexpensive for active swimmers it was hypoth-esized that environments with different cur-rent energies should be characterized by thepresence of the least drag-inducing ammonitemorphologies The patterns observed in dif-ferent facies of the Western Interior Creta-ceous Seaway were consistent with this pre-diction

In a seminal paper Bambach (1983) intro-duced the concept of the lsquolsquoguildrsquorsquo to paleon-tology where it has become commonly used

in paleocommunity studies (eg Watkins1994) Species that belong to the same guildexploit environmental resources in a similarway Species in a paleocommunity are as-signed to a guild on the basis of their phylo-genetic class (eg Trilobita) their feedingtype and food source and their life habit orlife position As indicated by Bambach rec-ognition of food source feeding type life po-sitions and life habits for a particular speciesrelies heavily on functional analysis (see alsoBambach 1994)

Paleobiologists also continue to offer eco-logical descriptions of evolutionary trendsbased on functional claims Vermeij (1977) ex-amined the interactions between predatorsand prey in what he termed the Mesozoic ma-rine revolution He documented in detail anincrease over time in the frequency of gastro-pod shell designs resistant to crushing in-cluding changes in shell coiling and the abun-dance and elaboration of ornamentation con-current with an increase in diversity of du-rophagous predators The identification ofshell crushers and of designs resistant to shellcrushing is based on functional arguments

Thayer (1979) attributed the changes in thestructure of benthic communities during thePhanerozoic to the diversification of depositfeeders The diversification led to increaseddisturbance of sediment and a consequent re-placement of immobile suspension feeders liv-ing on soft substrates by mobile taxa and im-mobile hard-surface dwellers

Labandeira (1997) examined the evolution-ary history of insect mouthparts Using clusteranalysis he identified 34 distinctive mouth-part classes among modern insects Many ofthese classes are polyphyletic Each mouth-part class is associated with a characteristicfeeding strategy such as piercing and suckingBy examining the history of insect mouth-parts Labandeira was able to identify fivephases in hexapod evolution which he asso-ciated with increased partitioning of food re-sources The correlation between preservedmorphology and function is essential to hisanalyses

Taxon longevities have also been exploredusing functional arguments Baumiller (1992)used filtration theory experiments and ener-


gy budgets to claim that filter morphologyplaced constraints on distributions of passivefilter feeders among environments of differentcurrent energies This claim was confirmed bydocumenting patterns of distribution of fossilcrinoids with different filter morphotypesamong facies More widely distributed andthus lsquolsquoeurytopicrsquorsquo morphotypes were predict-ed to be less prone to extinction than narrowlydistributed and thus lsquolsquostenotopicrsquorsquo morpho-types Stratigraphic ranges of crinoids cate-gorized by filter morphotype confirmed thisprediction (Baumiller 1993 Kammer et al1998)

What the above examples illustrate andwhat we would like to emphasize is that thequestion of function permeates paleobiologi-cal research However functional argumentsare not always well constrained by rigorousfunctional analyses nor is the link betweenfunctional attributes and their evolutionaryconsequences always made clear Often im-plicit is the notion that functional traits confersome performance advantage and that natu-ral selection is the mechanism for the originand proliferation of particular phenotypesBut because natural selection is not the onlymechanism of evolutionary change and be-cause a variety of processes of sorting and se-lection can affect the distribution of traits atdifferent levels of the biological hierarchy anassumption of natural selection as the causerequires further testing (Lewontin 1978Gould and Lewontin 1979 also see Rose andLauder 1996a and Vermeij 1996) When infer-ences on function are well constrained adap-tive scenarios are plausible and become goodstarting points for deeper evolutionary anal-yses into the origin maintenance or evolu-tionary consequences of traits Tests withvarying degrees of rigor can be applied tosuch scenarios Tests might include data fromphylogenetic analyses to supply informationon the history of transformation of traits andtheir independent or nonindependent origins(convergence vs homology) as well as datafrom paleoecology to provide relevant infor-mation on selective regimes Alternate scenar-ios deemed more plausible can replace themand likewise be tested

Conclusions

Despite the concerns expressed over thepast two decades we are convinced that func-tional analysis is capable of producing robustand testable statements about function and lifehabits in fossil organisms These functionalstatements should not be more detailed thanis allowed by the amount of preserved infor-mation and should be made in their appro-priate hierarchical contexts These hierarchiesare based on the anatomical detail of the anal-ysis the amount of structural informationavailable and the phylogenetic level of thegroup studied A study of the function of thesepta of nautiloids has quite different data re-quirements and range of generality than oneof forelimb movements in Anomalocaris

There is no single source of information ormethodology sufficient on its own for recon-structing function in fossils Data and meth-ods derived from biomechanics phylogenet-ics ichnology and taphonomy all can play im-portant roles We do not believe that function-al morphology would benefit at this time fromrigid methodological standardization and weencourage the use of new nonstandard meth-ods and data We are especially excited aboutthe potential for new methods derived fromstudies of artificial life and artificial intelli-gence to produce far more detailed functionalreconstructions than have been previouslyavailable

For paleobiologists functional attributes oftaxa continue to be a source of explanatory hy-potheses about ecological and evolutionarypatterns and trends Their appeal is in partbased on the theoretical underpinning thatDarwin provided But in using functional ex-planations one is not constrained solely to in-voking natural selection functional explana-tions may play a role in scenarios invokingsorting or selection operating at a variety oflevels (Vermeij 1996) or even those that do notinvoke selection at all

The focus of this journal as given by its titleis paleobiology To paleobiologists fossils arenot simply organic constituents of rocks theyare remains of once living organisms We aredenied our neontological colleaguesrsquo ability tomake the direct observations that are some of


the chief joys and interests of natural historyto see our organisms swim fly walk mateand eat

We strongly believe that the attempt tobreathe life back into extinct animals to at-tempt to visualize a once living world is lsquolsquosci-entifically as well as spiritually uplifting andrewardingrsquorsquo (Eldredge 1979 p 195)

Acknowledgments

G Lauder and an anonymous reviewer arethanked for their insightful comments whichgreatly improved the manuscript L Ivanyand P Kaplan kindly read over the manu-script and made many useful comments Wealso thank the editors S Wing and D Erwinfor their incredible patience and forbearanceFinally we would like to gratefully acknowl-edge three individuals who shaped and in-spired our research into the functional mor-phology of living and extinct animals S Vo-gel M LaBarbera and D C Fisher Partialsupport was provided by the National ScienceFoundation (grant EAR-97601 to T K B)

Literature Cited

Alexander R M 1989 Mechanics of fossil vertebrates Journalof the Geological Society London 14641ndash52

Allen T F H and T W Hoekstra 1992 Toward a unified ecol-ogy Columbia University Press New York

Bambach R K 1983 Ecospace utilization and guilds in marinecommunities through the Phanerozoic Pp 719ndash746 in M J STevesz and P L McCall eds Biotic interactions in fossil andRecent benthic communities Plenum New York

mdashmdashmdash 1994 Seafood through time changes in biomass ener-getics and productivity in the marine ecosystem Paleobiol-ogy 19372ndash397

Baumiller T K 1990 Physical modeling of the batocrinid analtube functional analysis and multiple hypothesis-testing Le-thaia 23399ndash408

mdashmdashmdash 1992 Importance of hydrodynamic lift to crinoid aut-ecology or could crinoids function as kites Journal of Pale-ontology 66658ndash665

mdashmdashmdash 1993 Survivorship analysis of Paleozoic Crinoidea ef-fect of filter morphology on evolutionary rates Paleobiology19304ndash321

Baumiller T K and W I Ausich 1992 The broken-stick modelas a null hypothesis for crinoid stalk taphonomy and as aguide to the distribution of connective tissue in fossils Paleo-biology 18288ndash298

Baumiller T K and H Hagdorn 1995 Taphonomy as a guideto functional morphology of Holocrinus the first post-Paleo-zoic crinoid Lethaia 28221ndash228

Baumiller T K G Llewellyn C G Messing and W I Ausich1995 Taphonomy of isocrinid stalks influence of decay andautotomy Palaios 1087ndash95

Behrensmeyer A K and S M Kidwell 1985 Taphonomyrsquoscontributions to paleobiology Paleobiology 11105ndash119

Bock W J and G von Wahlert 1965 Adaptation and the form-function complex Evolution 19269ndash299

Boucot A J 1990 Evolutionary paleobiology of behavior andcoevolution Elsevier Amsterdam

Boyajian G E and M LaBarbera 1987 Biomechanical analysisof passive flow of stromatoporoidsmdashmorphological paleo-ecological and systematic implications Lethaia 20223ndash229

Briggs D E G J E Dalingwater and P A Selden 1991 Bio-mechanics of locomotion in fossil arthropods Pp 37ndash56 inRayner and Wootton 1991

Bromley R G 1990 Trace fossils Unwin Hyman LondonCoddington J A 1988 Cladistic tests of adaptational hypoth-

eses Cladistics 253ndash67Cowen R 1975 lsquoFlapping valvesrsquo in brachiopods Lethaia 823ndash

29mdashmdashmdash 1979 Functional morphology Pp 487ndash489 in R Fair-

bridge and D Jablonski eds Encyclopedia of paleontologyDowden Hutchinson and Ross Stroudsburg Penn

Daniel T L B S Helmuth W B Saunders and P D Ward1997 Septal complexity in ammonoid cephalopods increasedmechanical risk and limited depth Paleobiology 23470ndash481

DeMar R 1976 Functional morphological models evolutionaryand non-evolutionary Fieldiana (Geology) 33333ndash354

Dennett D C 1998 Brainchildren essays on designing mindsMIT Press Cambridge

Denny M 1995 Predicting physical disturbancemdashmechanisticapproaches to the study of survivorship on wave-sweptshores Ecological Monographs 65371ndash418

Eldredge N 1979 Cladism and common sense Pp 165ndash198 inJ Cracraft and N Eldredge eds Phylogenetic analysis andpaleontology Columbia University Press New York

mdashmdashmdash 1985 Unfinished synthesis Oxford University PressNew York

Fisher D C 1977 Functional morphology of spines in the Penn-sylvanian horseshoe crab Euproops danae Paleobiology 3175ndash195

mdashmdashmdash 1985 Evolutionary morphology beyond the analogousthe anecdotal and the ad hoc Paleobiology 11120ndash138

Frazetta T H 1975 Complex adaptations in evolving popula-tions Sinauer Sunderland Mass

Gatesy S M K M Middleton F A Jenkins Jr and N H Shu-bin 1999 Three-dimensional preservation of foot movementsin Triassic theropod dinosaurs Nature 399141ndash144

Gershenfeld N 1999 The nature of mathematical modelingCambridge University Press Cambridge

Gislen T 1938 A revision of the recent Bathycrinidae ActaUniversitatis Lundensis 341ndash30

Gould S J 1970 Evolutionary paleontology and science ofform Earth Science Reviews 677ndash119

mdashmdashmdash 1980 The promise of paleobiology as a nomothetic evo-lutionary discipline Paleobiology 696ndash118

Gould S J and R C Lewontin 1979 The spandrels of San Mar-co and the Panglossian paradigm a critique of the adapta-tionist programme Proceedings of the Royal Society of Lon-don B 205581ndash598

Grant R E 1972 The lophophore and feeding mechanism ofthe Productidina (Brachiopoda) Journal of Paleontology 46213ndash249

mdashmdashmdash 1975 Methods and conclusions in functional analysis areply Lethaia 831ndash34

Greenewalt C J 1975 The flight of birds Transactions of theAmerican Philosophical Society new series 65(4)1ndash67

Grzeszczuk R and D Terzopoulos 1995 Automated learningof muscle-actuated locomotion through control abstractionSIGGRAPH (Conference 1995) Computer graphics proceed-ings annual conference series pp 63ndash70 Special InterestGroup on Computer Graphics Association for ComputingMachinery New York


Hagdorn H 1995 Triassic crinoids Zentralblatt fur Geologieund Palaontologie Teil II1ndash22

Hickman C 1988 Analysis of form and function in fossilsAmerican Zoologist 28775ndash783

Holland N D J C Grimmer and K Wiegmann 1991 Thestructure of the sea lily Calamocrinus diomedae with specialreference to the articulations skeletal microstructure sym-biotic bacteria axial organs and stalk tissues (Crinoidea Mil-lericrinida) Zoomorphology 110115ndash132

Huebner K H E A Thornton and T G Byrom 1995 The fi-nite element method for engineers Wiley-Interscience NewWork

Jacob F 1977 Evolution and tinkering Science 1961161ndash1167Jacobs D K 1996 Chambered cephalopod shells buoyancy

structure and decoupling history and red herrings Palaios11610ndash614

Jacobs D K and N H Landman 1993 Nautilusmdasha poor modelfor the function and behavior of ammonoids Lethaia 26101ndash111

Jacobs D K N H Landman and J A Chamberlain Jr 1994Ammonite shell shape covaries with facies and hydrodynam-ics iterative evolution as a response to changes in basinal en-vironment Geology 22905ndash908

Kammer T W T K Baumiller and W I Ausich 1998 Evolu-tionary significance of differential species longevity in Osa-geanndashMeramecian (Mississippian) crinoid clades Paleobiol-ogy 24155ndash176

Kesel A B U Philippi and W Nachtigall 1998 Biomechanicalaspects of the insect wing an analysis using the finite elementmethod Computers in Biology and Medicine 28423ndash437

Knight G J 1996 Making rocks swim In J E Repetski edSixth North American paleontological convention Abstractsof papers Paleontological Society Special Publication 8214

Labandeira C C 1997 Insect mouthparts ascertaining the pa-leobiology of insect feeding strategies Annual Review ofEcology and Systematics 28153ndash193

LaBarbera M 1983 Why the wheels wonrsquot go American Nat-uralist 121395ndash408

mdashmdashmdash 1990 Principles of design of fluid transport systems inzoology Science 249992ndash1000

LaBarbera M and G E Boyajian 1991 The function of astror-hizae in stromatoporoidsmdashquantitative tests Paleobiology17121ndash132

Lauder G V 1990 Functional morphology and systematicsstudying functional patterns in an historical context AnnualReview of Ecology and Systematics 21317ndash340

mdashmdashmdash 1991 Biomechanics and evolution integrating physicaland historical biology in the study of complex systems Pp 1ndash19 in Rayner and Wootton 1991

mdashmdashmdash 1995 On the inference of function from structure Pp 1ndash18 in Thomason 1995

mdashmdashmdash 1996 The argument from design Pp 55ndash91 in Rose andLauder 1996b

Lewontin R C 1978 Adaptation Scientific American 239156ndash169

McGhee G 1998 Theoretical morphology the concept and itsapplications Columbia University Press New York

Medawar P 1974 A geometric model of reduction and emer-gence Pp 57ndash63 in F J Ayala and T Dobzhansky eds Studiesin the philosophy of biology University of California PressBerkeley and Los Angeles

Messing C G M C RoseSmyth S R Mailer and J E Miller1988 Relocation movement in a stalked crinoid (Echinoder-mata) Bulletin of Marine Science 42480ndash487

Meyer D L and D B Macurda 1977 Adaptive radiation of co-matulid crinoids Paleobiology 374ndash82

Moon B R 1999 Testing an inference of function from struc-

ture snake vertebrae do the twist Journal of Morphology 241217ndash225

Myhrvold N P and P J Currie 1997 Supersonic sauropodsTail dynamics in the diplodocids Paleobiology 23393ndash409

Padian K 1991 Pterosaurs were they functional birds or func-tional bats Pp 145ndash160 in Rayner and Wootton 1991

Paul C R C 1975 A reappraisal of the paradigm method offunctional analysis in fossils Lethaia 815ndash21

mdashmdashmdash 1999 The paradigm method Pp 25ndash28 in E Savazzi edFunctional morphology of the invertebrate skeleton WileyChichester England

Petroski H 1985 To engineer is human St Martins New Yorkmdashmdashmdash 1993 The evolution of useful things Knopf New Yorkmdashmdashmdash 1996 Invention by design Harvard University Press

CambridgePhilippi U and W Nachtigall 1996 Functional morphology of

regular echinoid tests (Echinodermata Echinoida) a finite el-ement study Zoomorphology 11635ndash50

Plotnick R 1985 Lift-based mechanisms for swimming in eu-rypterids and portunid crabs Transactions of the Royal So-ciety of Edinburgh 76325ndash337

mdashmdashmdash 1997 Wonderful interactions the Digital Burgess con-ference American Paleontologist 52ndash4

Plotnick R and T Baumiller 1988 The pterygotid telson as abiological rudder Lethaia 2113ndash27

Prothero D 1998 Bringing fossils to life WCBMcGraw HillBoston

Radinsky L B 1987 The evolution of vertebrate design Uni-versity of Chicago Press Chicago

Raup D 1972 Approaches to morphologic analysis Pp 28ndash45in T J M Schopf ed Models in paleobiology Freeman Coo-per San Francisco

Ray T 1994 An evolutionary approach to synthetic biologyZen and the art of creating life Artificial Life Journal 1179ndash209

Rayner J M V 1991 Avian flight evolution and the problem ofArchaeopteryx Pp 183ndash212 in Rayner and Wootton 1991

Rayner J M V and R J Wootton eds 1991 Biomechanics inevolution Society for Experimental Biology Seminar Series36 Cambridge University Press Cambridge

Rose M R and G V Lauder 1996a Post-spandrel adaptation-ism Pp 1ndash8 in Rose and Lauder 1996b

mdashmdashmdash eds 1996b Adaptation Academic Press San DiegoRoux M 1978 Ontogenese variabilite et evolution morpho-

fonctionnelle du pedoncule et du calice chez les Millericrinida(Echinodermes Crinoıdes) Geobios 11213ndash241

Rudwick M J S 1964 The inference of function from structurein fossils British Journal for the Philosophy of Science 1527ndash40

Savazzi E 1999 Introduction to functional morphology Pp 3ndash14 in E Savazzi ed Functional morphology of the inverte-brate skeleton Wiley Chichester England

Seilacher A 1964 Biogenic sedimentary structures Pp 293ndash316in J Imbrie and N D Newell eds Approaches to paleoecol-ogy Wiley New York

mdashmdashmdash 1970 Arbeitskonzept zur Konstruktions-MorphologieLethaia 3393ndash396

Seilacher A and M LaBarbera 1995 Ammonites as Cartesiandivers Palaios 10493ndash506

Signor P 1982 A critical re-evaluation of the paradigm methodof constructional inference Neues Jahrbuch fur Geologie undPalaontologie Abhandlungen 16459ndash63

Simms M J 1988 The phylogeny of post-Paleozoic crinoidsPp 269ndash284 in C R C Paul and A B Smith eds Echinodermphylogeny and evolutionary biology Clarendon Oxford

mdashmdashmdash 1999 Systematics phylogeny and evolutionary historyPp 31ndash40 in H Hess W I Ausich C E Brett and M J Simmseds Fossil crinoids Cambridge University Press Cambridge


Simms M J and G D Sevastopulo 1993 The origin of artic-ulate crinoids Palaeontology 3691ndash109

Sims K 1994 Evolving virtual creatures SIGGRAPH (Confer-ence 1994) Computer graphics proceedings annual confer-ence series pp 15ndash22 Special Interest Group on ComputerGraphics Association for Computing Machinery New York

Stanley S M 1970 Relation of shell form to life habits in theBivalvia (Mollusca) Geological Society of America Memoir125

Stevens K A and J M Parrish 1999 Neck posture and feedinghabits of two Jurassic sauropod dinosaurs Science 284798ndash800

Terzopoulos D X Tu and R Grzeszczuk 1994 Artificial fish-es autonomous locomotion perception behavior and learn-ing in a simulated physical world Artificial Life 1327ndash351

Terzopoulos D T Rabie and R Grzeszczuk 1997 Perceptionand learning in artificial animals Pp 1ndash8 in C G Langton andK Shimohara eds Artificial life V proceedings of the fifthinternational workshop on the synthesis and simulation ofliving systems Nara-shi Japan 1996 MIT Press Cambridge

Thayer C W 1979 Biological bulldozers and the evolution ofmarine benthic communities Science 203458ndash461

Thomason J ed 1995 Functional morphology in vertebrate pa-leontology Cambridge University Press Cambridge

Valentine J M and C M May 1996 Hierarchies in biology andpaleontology Paleobiology 2223ndash33

Vermeij G J 1977 The Mesozoic marine revolution evidencefrom snails predators and grazers Paleobiology 3245ndash258

mdashmdashmdash 1996 Adaptations of clades resistance and responsePp 363ndash380 in Rose and Lauder 1996b

Vogel S 1994 Life in moving fluids the physical biology offlow Princeton University Press Princeton NJ

mdashmdashmdash 1998 Catsrsquo paws and catapults Norton New YorkWainwright S 1988 Axis and circumference the cylindrical

shape of plants and animals Harvard University Press Cam-bridge

Wainwright S W Biggs J Currey and M Gosline 1976 Me-chanical design in organisms Edward Arnold London

Watkins R 1994 Evolution of Silurian pentamerid communitiesin Wisconsin Palaios 9488ndash499

Weishampel D B 1995 Fossils function and phylogeny Pp34ndash54 in Thomason 1995

Witmer L M 1995 The extant phylogenetic bracket and the im-portance of reconstructing soft tissues in fossils Pp 19ndash33 inThomason 1995

Witmer L M and K D Rose 1991 Biomechanics of the jawapparatus of the gigantic Eocene bird Diatryma implicationsfor diet and mode of life Paleobiology 1795ndash120

q 2000 The Paleontological Society All rights reserved 0094-8373002604-0013$100

Copyright ( 2000 The Paleontological Society

Invention by evolution functional analysis in paleobiology

Roy E Plotnick and Tomasz K Baumiller

AbstractmdashFunctional analysis of fossils is and should remain a key component of paleobiologicalresearch Despite recently expressed doubts conceptual and methodological developments overthe past 25 years indicate that robust and testable claims about function can be produced Func-tional statements can be made in at least three different hierarchical contexts corresponding to thedegree of structural information available the position in the phylogenetic hierarchy and the de-gree of anatomical specificity The paradigm approach which dominated thinking about functionin the 1960s and 1970s has been supplanted with a methodology based on biomechanics Paleo-biomechanics does not assume optimality in organismal design but determines whether structureswere capable of carrying out a given function The paradigm approach can best be viewed as a wayof generating rather than testing functional hypotheses Hypotheses about function can also bedeveloped and supported by well-corroborated phylogenetic arguments Additional functional ev-idence can be derived from studies of trace fossils and of taphonomy New computer techniquesincluding lsquolsquoArtificial Lifersquorsquo studies have the potential for producing far more detailed ideas aboutfunction and mode of life than have been previously possible Functional analysis remains the basisfor studies of the history of adaptation It is also an essential component of many paleoecologicaland paleoenvironmental studies

Roy E Plotnick Department of Earth and Environmental Sciences University of Illinois at Chicago 845West Taylor Chicago Illinois 60607 E-mail plotnickuicedu

Tomasz K Baumiller Department of Geological Sciences and Museum of Paleontology University of Mich-igan 1109 Geddes Road Ann Arbor Michigan 48109-1079 E-mail tomaszBumichedu

Accepted 1 May 2000

Introduction

lsquolsquoA science of form is now being forged withinevolutionary theory It studies adaptation byquantitative methods using the organism-ma-chine analogy as a guide it seeks to reduce com-plex form to fewer generating factors and causalinfluencesrsquorsquo (Gould 1970 p 77)

lsquolsquoThe flowering of functional morphology hasyielded a panoply of elegant individual examplesand few principles beyond the unenlighteningconclusion that animals work well I at leastonce harbored the naive belief that a simple enu-meration of more and more cases would yield newprinciples for the study of form But Newtonianprocedures yield Newtonian answers and whodoubts that animals tend to be well designedrsquorsquo(Gould 1980 p 101)

lsquolsquo[W]e have placed unwarranted faith in ourunderstanding of the relationship between formand function Of particular concern arethe nature and accuracy of predictions of func-tion from morphology in fossil taxarsquorsquo (Lauder1995 p2)

Thirty years have passed since Gould ex-pressed his enthusiasm for human engineereddevices as guides to the study of fossil forms

and 20 years since his more pessimistic ap-praisal of functional morphology His laterskepticism apparently stemmed from a con-cern that although functional morphology isgood at determining functional design of par-ticular organisms the significance of thesestudies to understanding evolution is obscureLauder (1995) went further asserting that thestructure-function linkage is so weak that in-ferences of function from morphology in fos-sils are themselves questionable

These concerns make clear that we mustdistinguish between the methodology of func-tional morphology and its goals in order toproperly assess its importance in paleontolo-gy In this we follow Fisher (1985) who dis-tinguished functional analysis from functionalmorphology Functional morphology whichfocuses on the lsquolsquonature evolution and histor-ical consequences of adaptationrsquorsquo (Fisher 1985p 121) provides the historical context for theessentially ahistorical results of functionalanalysis

This paper will focus on methods and ap-proaches to functional analysis We will dis-cuss some recent methodological develop-ments provide an example of a taphonomic

















































































































































Conclusions








Acknowledgments


Literature Cited

































































































































































































































































Conclusions








Acknowledgments


Literature Cited




















































































































Acknowledgments


Literature Cited

















































































































invention by evolution: functional analysis in paleobiology

Documents