1BEYOND SELECTION Revista Chilena de Historia Natural78: ¿¿¿-???, 2005

PRUEBA DE PÁGINA

COMMENTARY

Beyond selection

Más allá de la selección

ALEXANDER O. VARGAS

Departamento de Anatomía y Biología del Desarrollo, Facultad de Medicina, Universidad de Chile, Independencia 1027,Casilla 70.079, Santiago 7, Santiago, Chile

Present address: Department of Ecology and Evolutionary Biology, Yale University, New Haven, Connecticut 06520,USA; e-mail: alexander.vargas@yale.edu

ABSTRACT

It has been argued that the study of natural selection and quantitative genetics should have a central role inevolutionary thinking and undergraduate teaching in Chile. Extensive operational use of the concept of naturalselection may seem consistent with this argument. However, advances of evolutionary knowledge inindependent fields such as phylogenetic analysis, developmental evolution, and paleontology cannot beignored. I argue here that the role of natural selection in contemporary evolutionary biology can be comparedto that of Newtonian mechanics in contemporary physics: it can describe a given domain of observations, butit is insufficient to handle the different sources of evolutionary knowledge. Overemphasis on natural selectionas the immediate mechanism of evolution may lead to disregard phylogenetic-historical evidence, and toignore the important evolutionary role of non-adaptive change and epigenetic phenotypic plasticity. Naturalselection deals with populations and leads to conceive the environment as a “sieve” of genetic variation,bypassing the role of the environment as a trigger of phenotypic and behavioral diversification. Alternatively,it is possible to conceive how part of the medium participates as an ontogenic niche in the trans-reproductivechange or conservation of an ontogenic phenotype. The concept of drift, currently accepted for molecular anddevelopmental change, can be applied to the level of the phenotype as an alternative to the concept ofevolution as adaptation by natural selection.

Key words: evolution, history, development, epigenesis, drift.

RESUMEN

Se ha argumentado que el estudio de la selección natural y la genética cuantitativa debiera ocupar un rolcentral en el pensamiento evolutivo y la enseñanza de pregrado de la evolución en Chile. El extenso usooperacional del concepto de selección puede parecer consistente con este argumento. No obstante, los avancesen el conocimiento evolutivo en áreas independientes, como los estudios de análisis filogenéticos, laevolución del desarrollo, y la paleontología, no pueden ser ignorados. Es posible argumentar que el rol de laselección natural en la biología evolutiva puede compararse al de la mecánica newtoniana en la físicacontemporánea. Puede describir un cierto dominio de observaciones, pero es insuficiente para manejar lasdiferentes fuentes de conocimiento evolutivo. Un énfasis excesivo en la selección natural como mecanismoinmediato de la evolución puede conducir a desconocer la evidencia filogenético-histórica, y a ignorar el rolevolutivo del cambio no adaptativo y de la plasticidad fenotípica epigenética. La selección natural trata depoblaciones y conduce a concebir al medio como un “cedazo” de la variación genética, evadiendo el rol delambiente como un gatillador de diversificación conductual y fenotípica. Alternativamente, es posible concebircómo un parte del medio participa como un nicho ontogénico en la conservación o cambio trans-reproductivode un fenotipo ontogénico. El concepto de deriva, actualmente aceptado para cambios moleculares y deldesarrollo, puede ser aplicado al nivel del fenotipo como una alternativa al concepto de evolución comoadaptación por selección natural.

Palabras clave: evolución, historia, desarrollo, epigénesis, deriva.

2 VARGAS

INTRODUCTION

It is frequently conceived that evolution is aprocess of adaptive change through naturalselection. This can easily lead to consider thatevolutionary thinking and undergraduateteaching of evolution should be centered onnatural selection. It has been argued that asystematic-taxonomic and historical view ofevolution currently prevails in undergraduatecourses in Chile, that does not provide arealistic picture of current research onevolution by natural selection and its tools ofquantitative genetics (Nespolo 2003).Insufficent teaching of natural selection andquantitative genetics in Chile has been arguedto have “devastating consequences for theformation of scientists”. The existence in Chileof evolutionary thinking based on structuraldeterminism and autopoiesis rather than naturalselection (Maturana & Varela 1973, 1980,1984, Maturana & Mpodozis 1992, 2000), hasfurther been commented as an outbreak ofunscientific and dogmatic thinking (Nespolo2003). From an alternative point of view, thiscommentary will discuss the insufficiencies ofconsidering natural selection as a generalframework for understanding evolution. I willalso argue that in Chile the presence of anapproach that is not based on selection is farfrom being unscientific. Rather, by questioningthe concept of evolution as adaptive change, itencourages a discussion that is necessary tograsp the real phenomenological scope ofevolution.

CURRENT EVOLUTIONARY RESEARCH:

A CORRECTED PICTURE

Advances in the phylogenetic-historicalapproach

In order to adequately describe currentresearch in evolut ionary biology, i t isnecessary to understand that the logic ofnatural selection and quantitative genetics isirrelevant for the treatment of extensiveempirical sources of evidence on evolution.The collection of this information can bedescribed as the phylogenetic-historicalperspective. The commentary by Nespolo(2003) may provoke the wrong impression that

the phylogenetic-historical approach has notundergone significant reinterpretation orexpansion of its knowledge. In fact, researchand thinking in this independent field hasprogressed dramatically in recent years,leading to new common methods of researchand consensus on several facts. For example,only in the last decade fossil taxa have beenfound that are relevant to understand theorigin of vertebrates (Shu et al. 1999), birds(Ji et al. 1998, Xu et al. 2003), snakes(Caldwell & Lee 1997), and sirenians(Domning 2001). Only in the 1990’s, thenumber of species of mesozoic birds tripledthose discovered in al l previous years(Chiappe 1997). Any serious evolutionarystudy involving the comparison of taxa willrequire phylogenetic analysis to identifywhich taxa share a more recent commonancestor. A logic of phylogenetic analysisbased on parsimony has been treatedmathematically for decades (Hennig 1965) andcurrently is also aided by molecular evidence,shedding new light on several importantevolutionary subjects. For example, newagreements have been arrived according bothto morphological and molecular evidence onthe affinities of animal phyla (Peterson &Eernisse 2001), and the relationships amongthe main divisions within amphibians, birds,and mammals (Meyer & Zardoya 2003).Evolutionary biology will always benefit froma historical-systematic perspective, whichal lows hypotheses on evolut ionarymechanisms to be contrasted with availableinformation on how evolution has actuallyoccurred. Examples of historically inspiredinsight of great importance to the study ofevolution are the concepts of phylogeneticconstraints , the role of chance andopportunism in events of mass extinction, andthe origin of adaptation by exaptation as analternative to directional selection (Gould &Vrba 1982, Gould 2002). However, a l lknowledge from the phylogenetic-historicapproach is stomped out by the commonargument that only the study of naturalselection deals with the immediate mechanismof evolution. Evolutionary biologists couldtherefore ignore the transitions documented bythe fossil record, and be generally unaware ofthe new agreements reached on thephylogenetic relationships among organisms.

3BEYOND SELECTION

In fact, the logic of natural selection andquantitative genetics is insufficient to claim anadequate knowledge of evolution, which mustalways update the unambiguous informationavailable from the phylogenetic-historicalperspective.

For example, it is possible to ask thefollowing question: can population geneticscontribute to understand how did the bonesarticulating the skull and jaw in reptiles endup in the middle ear of mammals? Not much:important evolutionary topics like this must beanswered in structural terms. The fossil recordshows how previous participation of otherbones in a “double” jaw articulation oftherapsids (Fig.1) allowed the quadrate andarticular bones to change position, while theother bones (dentary and squamosal) wereable to keep a functional articulation betweenthe skull and mandible (Caroll 1988, Benton2002).

Evolutionary developmental biology

Natural selection is also clearly insufficient forthe study of evolutionary developmentalbiology (evo-devo), a growing field of researchthat has delivered remarkable new insight intoevolution. The logic of selection can be appliedto differential survival within a population, butit is widely acknowledged that selection doesnot provide a mechanism for the origin ofvariation. This relates directly to the questionon the relationship between development andevolution: how does development“mechanistically” produce and restrictphenotypic variation? (Wagner et al. 2000).Current research within molecular-developmental biology has also revealed theactual complexity of the genotype-phenotyperelationship (phenogenetics, Weiss 2005),which goes far beyond the concept ofquantitative genetics, that small andaccumulative effects of several genes determinecontinuous variation of metric traits.Evolutionary developmental biology has alsointroduced the comparison among taxa of theembryonic expression patterns of orthologousgenes, providing an important new source ofevidence for the discussion of homology orconvergence of similar traits. For example,several resemblances between protostomes anddeuterostomes, previously assumed to beconvergences from an adaptationist perspectiveof evolution, have been found to share a similarmolecular-developmental basis. Processes ofanterior differentiation, dorsoventral patterning,eye development, peripheral and centralnervous system development, cardiacdevelopment, gut regionalization, segmentationand appendage formation and patterning haveall been proposed as conserved developmentalmechanisms that were already being utilized inthe most recent common ancestor ofprotostomes and deuterostomes (reviewed byKnoll & Carroll 1999).

Research on developmental evolution isfrequently published in several specificjournals dedicated to the subject, as well ashigh-impact developmental, evolutionary, andinterdisciplinary journals. According to areview on evolutionary developmental biology,“the evidence for evolution is better than ever.Natural selection in evolution, however, is seento play less an important role. It is merely a

Fig. 1: The double articulation in the jaw jointof the therapsid cynodont Probainognathus,which explains how the articular (art) and qua-drate (q) of mammals were able to move intothe middle ear, while the dentary (dent) andsquamosal (sq) maintained a functional jaw jo-int. Image taken from Carroll (1988).

La doble articulación en la articulación mandibular de elterápsido cynodonto ‘robainognathus, que explica cómo elarticular (art) y el cuadrado (q) de los mañíferos fueroncapaces de moverse hacia el oído medio, mientras que eldentario y el escamoso retuvieron una articulación mandi-bular funcional. Imagen tomada de Carroll (1988).

4 VARGAS

filter for unsuccessful morphologies generatedby development. Population genetics isdestined to change if it is not to become asirrelevant to evolution as Newtonian mechanicsis to contemporary physics” (Gilbert et al.1996).

NATURAL SELECTION VERSUS NATURAL DRIFT

The evolution of developmental pathways

Although it is possible to state that theevolution of phenotypic traits is caused by theevolutionary modification of theirdevelopmental pathways (Hall 1998, Raff1996), it is important to point out that theevolutionary conservation of a phenotypic traitdoes not imply the conservation of i tsdevelopmental pathway. Countless cases ofvariation in developmental pathways are know(Wagner & Misof 1993), such as variation ofthe germ layer of origin (endoderm , ectodermor mesoderm), variation in the early inductiverole of tissues, and homeotic transformations(the same structure derived from a differentbody region). The fact that the developmentalpathway of a trait can vary is also evidenced inthat genetic and environmental factors cansubstitute for each other in triggering a certainphenotype, a fact recognized withindevelopmental biology by the use of the terms“phenocopy” and “genocopy”. Since in thesecases, phenotypes are identical to selectionregardless of their genetic or epigenetictriggering, the exchange of genetic orepigenetic triggering of a trait along evolutionhas been described as an adaptively neutralprocess of phenogenetic drift (Weiss &Fullerton 2000). Cases where an epigeneticallyinduced phenotype thereafter becomesintrinsically inheritable are discussed under theconcepts of genetic assimilation (Waddington1953, Palmer 2004) and stabilizing selection(Schmalhausen 1949, cited in West-Eberhard2003). The “Baldwin effect” in turn candescribe how traits also may evolve towardsincreased condition sensitivity, rather thangreater genotypic influence (West-Eberhard2003). Several conceptual and empiricalreasons have been exposed to argue thatepigenetic plasticity, rather than geneticvariation, is frequently involved in the

initiation of evolutionary novelty (Newman &Müller 2000, Müller 2003, Palmer 2004), witha role of genes as followers of the phenotype(Maturana & Mpodozis 2000, West-Eberhard2003).

A renewed concern currently exists amongevolutionary biologists over the relevance ofepigenetic plasticity in evolution.Unfortunately, epigenetic plasticity is seldommentioned when natural selection is consideredto be the immediate mechanism of evolution.This is because in evolution by naturalselection, the effect of the environment isunderstood to act mainly as a selective agent or“sieve” of genetically determined variation,missing its role as a trigger of phenotypic andbehavioral diversification, and starting point ofevolutionary changes. Traits withoutheritability are incorrectly considered to haveno evolutionary potential because they do notmeet the requisites for evolution “by naturalselection” and are frequently considered to be“noise”. Therefore, the actual evolutionaryrelevance of epigenetic plasticity providesanother reason why natural selection should notbe ideologically assumed to be the mainmechanism of evolution. The theoreticalframeworks of genetic assimilation, stabilizingselection, and the “Baldwin effect” have beenunable to effectively introduce the evolutionaryimportance of epigenetic plasticity, and severalarguments have been introduced to dismissthem (West-Eberhard 2003). This may bebecause they are ultimately framed in selectiveterms, regarding the effects of phenotypicplasticity on fitness and gene frequencies inpopulations. To study evolution, adequateunderstanding of the relationship betweenorganism and environment is of evidentimportance. In fact, full acknowledgement ofthe developmental and evolutionary importanceof epigenetic plasticity is at odds withinterpretation of the environment to beultimately a “sieve” or effector of selection.Although the concept of selection can be usedto deal with the relationship to the environmentat a population level, it does not deal with themore basic level of the individual. At this level,understanding of the relationship to theenvironment can be discussed in structuralterms. In contrast to conceiving theenvironment as the effector of selection,focusing on the individual level allows to

5BEYOND SELECTION

describe the participation of the environment asan ontogenic niche in the conservation orshifting of an ontogenic phenotype (Maturana& Mpodozis 1992, 2000). The organism is thestarting point for the definition of its niche, andthe niche itself can be distinguished as part of agreater medium that contains it (Fig 2). Themedium, therefore, participates not only as ageneral container, but it operates fundamentallyas the domain of the realization of theontogenic niche of the living systems that itcontains. It provides an independent source ofopportunities for the shifting of ontogenicphenotypes, and for the realization of variationsin epigenesis along the history of conservationand diversification of lineages (Maturana &Mpodozis 1992, 2000).

The adaptively neutral exchange in theenvironmental or genetic triggering of thedevelopment of a trait described by Weiss &Füllerton (2000) can be described as fitting theproposal that the epigenetic field of possibledevelopmental pathways changes trans-reproductively, even when the ontogenicphenotype-ontogenic niche relationshipremains unaltered (Fig 3; Maturana &Mpodozis 1992, 2000). Although the change inthe epigenetic field of possible developmentalpathways is non-adaptive, note that it definesthe possibilities of future shifts in the ontogenicphenotype-ontogenic niche relation. Theadaptively neutral exchange betweenenvironmental and genetic triggering of thedevelopment of traits that are required forsurvival has the non-trivial consequence ofdefining which environmental or geneticchanges can or cannot occur without losingadaptation.

Adaptationism: the overestimation of selection

A recent commentary argued that evidence forevolution by natural selection continues toaccumulate. To prove this point, 94 recent casestudies were listed where estimates ofheritability or selection differentials wereprovided for traits, as well as cases of artificialselection and experimental evolution (Nespolo2003). However, the basic logic of selectiondoes not require for increasing accumulation ofevidence to be accepted, and can be understoodthrough simple qualitative examples, whichmay or may not belong within the field of

biology. A comparison of the role of selectionto that of Newtonian mechanics in physics isinteresting, since selection can only be used toexplain a given domain of observations. Theuse of “accumulation of replicas” as“accumulation of evidence” ignores theimportance of how theories specify the domainof observations they are capable of explaining.Alternatively, an extensive list of case studiesof natural selection may be considered to

Fig. 2: This figure attempts to evoke the diffe-rent views that an observer can have of a livingsystem as he or she beholds it and reflectsabout its existence. As the observer beholds theliving system from a distance: (a) the mediumappears to him or her as all that he or she mayimagine as the great container in which itexists; (b) the niche appears to him or her asthat part of the medium with which the livingsystem interacts and which it obscures, so thatit can only be shown by the operation of theliving system itself: and (c) the ambient or thatwhich surrounds the living system, appears tohim or her as that which he or she sees aroundit but which being part of the medium is notpart of its niche. Conceptually, the niche andthe ambient together constitute the medium. (fi-gure and legend taken from Maturana & Mpo-dozis 2000).

Esta figura ilustra las diferentes distinciones que un obser-vador puede realizar respecto del dominio de relaciones deun organismo con su entorno. El medio es distinguidocomo todo aquello que, a juicio del observador, forma par-te del gran continente o ámbito relacional y de interaccio-nes en que el ser vivo existe. El nicho es distinguido comoaquella parte del medio con la cual el ser vivo está interac-tuando, y que por tanto queda oculta al observador, el quepuede inducirla solo a partir de las operaciones del propioser vivo. El ambiente queda distinguido como toda aquellaparte del medio que no forma parte del nicho. El nicho y elambiente juntos constituyen el medio.

6 VARGAS

support the frequent occurrence of thisevolutionary mechanism in nature. However,most cases listed by Nespolo (2003) do notprovide actual evidence of evolution by naturalselection. The conceptual framework forevolution by natural selection requires traits toshow heritability “and” to be under selection.Remarkably, only three cases cited in thatcommentary present estimates of “both”heritabili ty and a directional selectiondifferential for a trait. Moreover, it is wellknown that the heritability of traits can changedepending on environmental conditions (West-Eberhard 2003). Therefore, it is necessary toprecise that in such cases heritability waspresent “under the conditions observed” andthat this requisite for evolution by naturalselection could disappear under anotherenvironmental circumstance. Furthermore, toaccept that most evolutionary change occurs bynatural selection, this would require a rigorouscomparative method to contrast with thefrequency of cases in which evolution does notproceed by natural selection (for example, byadaptively neutral drift, or exapted traits thatdid not originate by selection for currentfunction), specially since Nespolo (2003)admits in the same commentary that non-adaptive processes are commonplace in nature,

and that “nothing could be more misleadingthan to believe in natural selection as theunique cause of evolution”. Finally, cases ofevolution by artificial selection andexperimental evolution certainly do notdemonstrate the predominance of naturalselection. Evolution by artificial selection isachieved in a homogeneous environment,where environmentally triggered variation isdiminished, and genetic differences affectingvariation of a trait are made by the researcherto reflect differences in offspring survival.Similar conditions can be rare in nature,especially regarding the directionality ofselection as imposed by humans with a well-defined objective. Evolution also requires to beaddressed in terms of structural, developmentaland phylogenetic constraints, which set limitsto the variation available to both artificial andnatural selection (Thompson 1917, Whitman1919, Gould 2002).

Natural selection has been argued to have apredictive power that is in contrast with morerecent evolutionary frameworks (Nespolo2003). Evolutionary biology certainly possessesgood predictive power that is completelyindependent of the logic of natural selection.Specially evident examples are the frequentprediction by phylogenetic inference of the

Fig. 3: This figure attempts to illustrate the change or shift of the epigenetic field as well as thechange of the genetic constitution that takes place in the course of the generations while theparticular ontogenic phenotype/ontogenic niche relation that defines a lineage (and that is epigene-tically realized as the arrow and the r indicate), is conserved from generation to generation throughsystemic reproduction (figure and legend taken from Maturana & Mpodozis 2000).

Esta figura ilustra el corrimiento del campo epigenético, y también de la constitución genética inicial, que ocurren transge-neracionalmente en un linaje, mientras la relación fenotipo ontogénico/nicho ontogénico que define a ese linaje (y que esepigenéticamente realizada, como lo indican la flecha y la r), es conservada generación tras generación a través de lareproducción sistémica.

7BEYOND SELECTION

discovery of representatives of clades in theappropiate geological time (Norell & Novacek1992), or the extrapolation of previouslydescribed developmental trends such asallometric growth (Gould 1977). The closephylogenetic affinity between Archaeopteryxand dinosaurs allowed the prediction that somedinosaurs must have presented feathers (Bakker1986, Paul 1988) before feathereddromaeosaurid and oviraptorid dinosaurs wereactually discovered (Ji et al. 1998, Xu et al.2001, Xu et al. 2003). In the case of naturalselection, despite the mathematicalformalization of quantitative genetics, it wouldcertainly be naïve to think this has led to apredictive power comparable to that ofNewtonian mechanics. In fact, predictionsmade in selective frameworks can frequentlyturn out to be incorrect. Moreover, predictivepower achieved within selective frameworkscan be argued to always rely heavily on anadequate description and understanding ofunderlying biological mechanisms and otherrelevant factors, which in ecology arefrequently designated as a whole through awastepaper-basket use of the term “naturalhistory”.

One of the main legacies of Darwinianthinking is the uniformitarian concept that“each slight variation, if useful, is preserved,by ... natural selection” (Darwin 1859). It isfrequently argued that selection has the powerto detect even slight adaptive advantages ordisadvantages which become reflected inoffspring survival, accumulating slightlybeneficial genes and eliminating the slightlynegative. However, this way of thinkingunderestimates the actual relevance of non-adaptive drift and epigenetic plasticity onoffspring survival, which can frequentlyoverride any small effects of genes (Weiss &Buchanan 2003). The continuous variation ofmetric traits in quantitative genetics and theassumed accumulation under selection ofseveral genes with small effects is consistentwith the Darwinian concept ofuniformitarianism. However, one-step geneticor environmental changes can trigger a largespectrum of plasticity, from slight to drasticphenotypic changes, as well as the emergenceof traits that are qualitatively different. Thelimitations of a gradualist, microevolutionaryapproach as a general framework for evolution

have been extensively discussed elsewhere(Gould 2002, West-Eberhard 2003).

Biases of selection: cases from tetrapod limbs

The concept that evolution is a process ofadaptation by natural selection has frequentlyshown inconsistencies with empiricalobservations of non adaptive evolutionarychange, and can be found to mislead fromadequate recognition of unambiguousphylogenetic-historic evidence.

One interesting case is provided by theevolution of the development of the bird wing.In the embryos of crocodiles (the closest livingrelatives of birds), as well as in any pentadactylamniote, the first digit to initiate cartilageformation is digit 4 (Burke & Feduccia 1997,counting 1-5 from the thumb to the pinky). Thefossil record documenting the evolutionarytransition from theropod dinosaurs to birdsindicates unambiguosly that digits 4 and 5 werelost, retaining digits 1, 2 and 3 in their tridactylwing (Fig. 4; Padian & Chiappe 1998, Wagner& Gauthier 1999). The development of thewing is unique in that the first digit to initiatecartilage formation is digit 3. It is thereforeconcluded that the developmental pathway ofthe forelimb has changed along the theropod-bird transition (Chatterjee 1998), and proposedto be a case of a homeotic frameshift of digitalidentity (Wagner & Gauthier 1999). Despite thefact that the evolutionary variation ofdevelopmental pathways is well-documented,and that the origin of birds from dinosaurs hasbeen confirmed by several independent sourcesof evidence (skeletal, tegumental, oological,ethological, and molecular; Padian & Chiappe1998, Sereno 1999, Prum 2002, 2003, Witmer2002, Chiappe & Vargas 2003, Zhou 2004,Vargas & Fallon 2005), the assumedpredominance of natural selection has allowedfor the argument that a change in theembryology of the wing would have noadaptive value, and therefore, that i tsevolutionary plausibility can be questioned,basically because no reason can be found whythis change would be selected for (Feduccia2002, Galis et al. 2003). Ultimately, it is evenargued that birds did not descend fromtheropod dinosaurs, but rather from ancestorsthat had lost digits 1 and 5 (Feduccia 2002,2003).

8 VARGAS

Fig. 4: Phylogenetic-historical evidence unambiguously indicates that birds lost digits 4 and 5,retaining digits 1, 2 and3. Succesive taxa share a more recent common ancestor with modern birds(Neornithes). (A-C) Triassic dinosaurs: (A) the ornithischian Heterodontosaurus, (B) the earlytheropod Herrerasaurus, (C) the neotheropod Coelophysis. (D-E) Jurassic theropods: (D) the teta-nuran Allosaurus, (E) the early maniraptoran Ornitholestes. (F-G) Mesosozoic birds: (F) the Juras-sic avialae Archaeopteryx, (G) the cretaceous enantiornithe Sinornis. (H-I) Modern birds: (H) thewing of an Opisthocomus (hoatzin) hatchling, (I) the wing of the adult chicken Gallus. Modifiedfrom Vargas & Fallon (2005).

La evidencia filogenético-histórica indica sin ambigüedades que las aves perdieron los dígitos 4 y 5, reteniendo los dígitos1, 2 y 3. Los taxa sucesivos comparten un ancestro en común más reciente con las aves actuales (Neornithes). (A-C)Dinosaurios triásicos: (A) ornitsquio Heterodontosaurus, (B) terópdo temprano Herrerasaurus, (C) neoterópodo Coelophy-sis. (D-E) Terópodos jurásicos: (D) tetanuro Allosaurus, (E) maniraptor temprano Ornitholestes. (F-G) Aves mesosozoicas:(F) Avialae jurásico Archaeopteryx, (G) enantiornite Sinornis. (H-I) Aves actuales: (H) ala de polluelo de Opisthocomus,(I) ala de adulto del pollo Gallus. Modificado de Vargas & Fallon (2005).

9BEYOND SELECTION

Adaptationism is evident in this argumentagainst the theropod-bird link. Phylogenetic-historical evidence is discarded beforeaccepting the possibility that even a “rare”event of non-adaptive evolutionary changecould have occurred. Further molecular-developmental evidence, however,demonstrates that the development of the wingprovides support for the theropod-bird link: theexpression pattern of the Hoxd12 and Hoxd13genes in the chicken wing has been shown to beas expected for a limb with digits 1, 2 and 3,rather than digits 2, 3 and 4 (Pennisi 2005,Vargas & Fallon 2005). In contrast, noexpression pattern or molecular marker hasbeen described to support the 2, 3, 4identification of the wing digits. The bestapproach is to accept the abundant fossilevidence for the theropod-bird transition, andassume that the wing of birds represents yetanother case of adaptively neutral variation in adevelopmental pathway.

Unfortunately, under an adaptationist pointof view, the similarities of theropod dinosaursand birds can always be interpreted to beconvergent solutions to similar adaptiveproblems. It is considered that phylogeneticanalysis based on parsimony cannot deal with“massive convergence” (Feduccia 1999), leadingto an ultimately mysterianist attitude towards thesubject, which considers (to the outrage ofpaleontologists) that the origin of birds may be ascientifically unsolvable problem (Feduccia2002), as in the comment that “the debate overthe phylogenetic position of birds seems farfrom any conclusion (...). Why? Perhaps becausewhere natural selection meets the strictconstraints of biomechanics, convergence isinevitable, and separating common inheritancefrom common function may be near-impossiblein a system so highly derived” (Thomas &Garner 1998). Within the subject of bird origins,the concept that natural selection is the mainevolutionary mechanism does not pass the test ofinterdisciplinary integration, failing to agreewith what paleontologists consider to be therelevant evidence.

Another case where selective thinking maybe misleading regards the evolvability ofpolydactyly. It is well-known that polydactylycan be observed within intraspecific variation,and that it is inheritable. However, it has beenargued that a true case of polydactyly has never

occurred at the evolutionary level ofinterspecific variation (Galis et al. 2001). Insome apparent cases, such as the panda’sthumb, a sesamoid or pisiform bone is derivedinto a functional digit-like structure. Given thatheritability for polydactly is high withinintraspecific variation, and that selection for anextra digit is conceivable, it has been arguedthat polydactyly may imply some negative side-effect on fitness that constrains its evolutionaryoccurrence. It has been proposed that selectionagainst polydactyly could be explained becausegenetic changes required for polydactyly couldhave pleiotropic effects such as a greatersusceptibility to cancer (Galis et al. 2001). Thishypothesis therefore provides an adaptivecontext to understand the apparently non-adaptive constraint on polydactyly. However,the recent discovery of an unambiguouslypolydactyl aquatic amniote from the Triassic ofChina (Wu et al. 2003) indicates that theevolutionary occcurence of polydactyly inamniotes is certainly not impossible, a fact thatshould caution against assuming the generalityand pervasiveness of hypothesized adaptiveexplanations.

The assumption that evolution mostlyproceeds by natural selection can lead to“Panglossian” adaptationism and ad-hochypotheses of trade-offs as ultimately adaptiveexplanations for the presence of any apparentlynon-adaptive or “negative” trait (Gould &Lewontin 1979). Exceptions to hypothesizedselective factors can always be explained bypreviously “unrecognized” advantages ordisadvantages which can lead to an ultimatelyuntestable philosophy, where all evolutionarychange is allowed or restricted for purelyadaptive reasons Although the operationalcommitment and specialization of some traits isobvious (specially in highly specializedlifestyles), hypotheses of adaptive value oftraits can frequently turn out to beinsufficiently supported by evidence.Unfortunately, purely hypothetical adaptiveconsiderations are sometimes taken to be morerelevant than well-documented phylogenetic-historic facts.

Adaptation and fitness: terms to be revised

Within the study of natural selection, estimatesof fitness of individuals cannot rely on the

10 VARGAS

description of traits or mechanisms, but areultimately judged by offspring survival. Thisdefinition of fitness can be controversial,because offspring survival should beconsidered the result of fitness, rather than itsequivalent or cause. As noticed by earlyopposition to natural selection, the concept of“survival of the fittest” can be argued to benothing but “survival of the survivors” (Milner1990). Rather than describe greater offspringsurvival as a progressive increase in fitness, itcan be understood in a more contingential andneutral sense, by full acknowledgement of hownon-genetic factors and other circumstancescan affect offspring survival. Offspring survivalcertainly can vary when compared amongdifferent individuals. However, differences inoffspring survival are not implied to be a directresult of genetic differences among individuals,which can be easily overcome byenvironmental-circumstantial factors.Unfortunately the concept of fitness as asubstitute for offspring survival will continueto mislead towards interpretation of individualswith a greater offspring survival as somehowhaving a greater “genetic quality” than others.As an alternative, it has been proposed that in astrict sense it is not possible to defineadaptation as a variable, such that no individualcan be considered intrinsically better adaptedthan another (Maturana & Mpodozis 2000).Adaptation, considered as the operationalcongruence and structural coupling of theorganism to the environment, is a condition ofexistence of l iving systems, and eitheradaptation is present or the systemdisintegrates. From this perspective, adaptationdoes not arise or increase through greateroffspring survival, but rather it is a pre-requisite even for reproduction. Evolutionarychange does not require for an increase inoffspring survival. Conversely, an increase inoffspring survival can occur without anyevolutionary change. The fact that certain traitscannot change without leading to thedisintegration of the organism, does not implythat these traits originated through acompetitive process of natural selection fortheir current function, as exemplified by theoccurrence of exaptation (Gould & Vrba 1982).

It is conceivable that a non-adaptivephenotypic trait can become fixed in apopulation without natural selection, but as a

result of a bottleneck or foundation effect.However, even in cases in which differentialsurvival can be found to relate directly to traitsof organisms (for example, when only immuneindividuals survive a disease, or in the classicexample of Kettelwell’s industrial melanism) ,it is legitimate to ask whether this process canonly be described as a process of adaptation bynatural selection, or if it can in fact be arguedto be more contingential and comparable to abottleneck, specially taking into account thatunder another circumstance (e.g., another typeof sickness), other traits would be selected for.When selection is described to have occurredfor a trait, it is possible to consider it as apossible pathway under a specific circumstance(drift), rather than describing it as a directional,adaptive increase of fitness leading to the finalresult.

Although it is frequently conceived thatevolutionary change is a process of progressiveincrease in fitness or adaptation through naturalselection, it is possible to apply the concept ofnon-adaptive drift, accepted for the evolutionof developmental pathways and evolution at amolecular level (Kimura 1968, Weiss &Fullerton 2000), to the level of the phenotype(Maturana & Varela 1984, Maturana &Mpodozis 1992, 2000, Weiss & Buchanan2003), conceiving the change and conservationof phenotypes as a neutral process of naturaldrift (Maturana & Mpodozis 1992, 2000),rather than a process of adaptation throughnatural selection. Conservation and change ofontogenic (life cycle) phenotypes is establishedthrough the trans-reproductive repetition orshifting in the epigenetic pathway that emergesalong the organism-environment interaction(Maturana & Mpodozis 1992, 2000). Bothgenetic and environmental changes in evolutioncan be described as co-opted to the epigeneticpathway in which adaptation is conserved(Maturana & Mpodozis 1992, 2000).

The epistemological status of natural drift

Epistemologically, the systemic conceptualframework of natural drift (Maturana &Mpodozis 1992, 2000) is not teleological(“finalist”), it does not imply any sort of vitalistor metaphysical reasoning, and does not makeany reductionist assumptions. No force isinvoked that cannot be observed to operate in

11BEYOND SELECTION

the natural world. It is therefore mostremarkable that this conceptual framework hasbeen argued to be unscientific and dogmatic,and even framed as holding a potential to fuelcreationism (Nespolo 2003). The mainargument for this overreaction is that naturaldrift cannot be falsified. Ironically, it is well-known that the same argument has been used toseriously question the scientific status ofnatural selection, because no crucialexperiment or case can be conceived to prove itwrong (Popper 1976). As discussed by Mayr(1998), when dealing with a general conceptualframework for a complex historicalphenomenon such as the evolution of life, anextreme version of falsation is inadequate.Rather, within evolutionary biology, generaltheories are assessed by contrasting theconsistency of their logical predictions withseveral independent sources of information (seeAhumada et al. 2005). As discussed above,evolutionary biology can have good predictivepower, but it cannot be attributed to the soleapplication of the concept of selection.

Creationists or advocates of “intelligentdesign” frequently refer to the origin of life asa roadblock to scientific explanation. Naturaldrift is an explanation of evolution that uses theconcept of autopoiesis as a definition of life.The concept of autopoiesis does not recur toany reductionist or vitalist assumptions toexplain the origin of life, and as such hasreceived considerable attention in this field(Maturana & Varela 1973, 1980, Bachmann etal. 1992, McMullin 2000, Luisi 2003).Autopoiesis conceives organisms as self-producing units. Reproduction and therefore,evolution, are consequences of autopoiesis(Maturana & Mpodozis 1992, 2000). It is alsointeresting to point out that, although themetaphysical construct of modern catholic faithcurrently considers itself compatible with thescientific fact of evolution, it continues to setthe limit as to a non-evolutionary origin of ahuman soul (Vicuña 2004), and further rejectsany biological origin of the human mind.Alternatively, all human cognitive experiencescan be approached in biological terms ofstructural determinism (Maturana & Pörksen2004).

The bold epistemological approach behindnatural drift dispels any attempted linking ofnatural drift to creationism. Similarly, the

proposal that natural drift is an actualequivalent to the “daisyworld parable” (asimulation on how a homeostatic, regulativeglobal environment can evolve without naturalselection, Watson & Lovelock 1983) is clearlyinappropriate. The conceptual framework ofnatural drift provides a battery of concepts todeal with several important subjects thatrequire attention in evolutionary biology. Theonly resemblance of natural drift to the“daisyworld parable” is that it does notconsider a central role for natural selection.

Creationism is fueled by the mistake ofconsidering “natural selection” and “evolution”to be equivalent, as in the popularmisconception of “Darwin’s theory ofevolution”. This leads to confuse the debate onthe evolutionary role of selection (which isnecessary and appropriate), with the existenceof doubt on the phenomenon of evolution initself. Dobzhansky probably contributed to thisconfusion with his famous reductionist dictumthat “evolution is a change in the geneticcomposition of populations”. The frequentepistemological mistake of reducing aphenomenological domain to another that isassumed to be more fundamental can berecognized and avoided (Lewontin 1974,Maturana & Mpodozis 2000). In fact, a centralrole for selection and quantitative genetics isnot required to acknowledge the phenomenonof evolution in itself. Common ancestry anddescent with modification are evident fromchange and conservation throughout thecomparison of the diversity of living and fossiltaxa. If this is understood, controversy on therole of selection may cease to be confused withuncertainty on the fact of evolution itself.

CONCLUDING REMARKS

The framework of selection has been applied toseveral biological levels (genes, cells,organisms, populations, species) and even tonon-biological cases (Nespolo 2003). Althoughthis may be considered to reflect the greatoperational utility of the concept of selection, ithas also lead to remarkable controversies: dogenes confer fitness to individuals which actmainly as transient vessels, or is the frequencyof genes a result of the fitness of individuals?145 years after publication of “the origin of

12 VARGAS

species”, the actual agents of selection are stillunder discussion, with paradoxicaldisagreement on the hierarchical priority ofselection for one level or another (Gould 2002,Dawkins 1976, West-Eberhard 2003). Even ifselection is a highly operational concept, thisdoes not justify conceiving natural selectionideologically as a pre-eminent, fundamentaldeterminant of evolutionary change, that shouldbe central in the study of evolutionary biology(Camus 2000). An ideological approach toselection leads to disregard phylogenetic-historical evidence, dismiss non-adaptiveevolutionary change, and underestimate therole of epigenetic phenotypic plasticity. Thepresence in Chile of alternative conceptualframeworks is in part a response to theinsufficiencies of selection for theunderstanding of evolution (Maturana &Mpodozis 2000).

A historical description of the acceptance ofnatural selection in the scientific communityshows an initial enthusiasm in the late 19th-early 20th century, followed by an importanteclipse towards the 1930’s when it was almostabandoned. Towards the mid 20th century, themodern synthesis had resurrected Darwinism,and attention focused on the theoreticalconstruct of population genetics, with adogmatic hardening of the concept of naturalselection as the main mechanism ofevolutionary change. However, since the1960´s, research on evolutionary biology hasgone far beyond the scope of natural selectionand population genetics, with great advances inphylogenetic, historical, molecular, anddevelopmental research. The concepts ofspandrels, exaptation, and non-adaptive changehave become well-accepted. Within the actualstructure of current evolutionary theory, thestudy of natural selection and populationgenetics can be described to coexist with othersubstantially different branches of evolutionarytheory (Gould 2002).

Several researchers have defended thatnatural selection should have a central role inevolutionary thinking, as it effectively did inthe mid 20th century. Some may even considertheir own line of evolutionary thinking has“attained maturity” (Nespolo 2003). However,it is evident that a great deal of progress inevolutionary research and thinking is still due,and is not restricted to the study of natural

selection. Current evolutionary biology dealswith several independent research fields onevolution that are almost absent in Chile, suchas paleontology and developmental evolution.We can not expect to develop a truly scholarlyvision of evolution if an hegemonic role forselection is defended at undergraduate coursesof evolution. Greater interdisciplinaryknowledge will be achieved by accepting thegrowing phenomenological scope ofevolutionary biology, and the need for adiversity of theoretical and empiricalapproaches. New conceptual frameworks arewelcome that may help remove the biasesfacing the recognition of actual evolutionaryfacts and mechanisms.

ACKNOWLEDGEMENTS

Thanks to Jorge Mpodozis and Carlos Medinaat the Uiversidad de Chile for their excellentcourses of evolution and for enlighteningdiscussions. Thanks to the Decenio biologydiscussion group and to everyone at the lab ofcognitive biology. This article is dedicated tothe memory of Gonzalo Farfán.

LITERATURE CITED

AHUMADA P, G ASTORGA, K BULL-HEREÑU, CCABRERA, L CAPUTO, C CARMONA, MCCECCHI, N CORTÉS, S CORTEZ, C DORADOR,F FREDES, C GUERRERO-BOSAGNA, PHENNY, C LIBEDINSKY, W MARCELO, NMARQUEZ, J MARTÍNEZ, M MEDINA, P OSSA,D PARRA, N POHL, H POTTSTOCK, TOSSANDON, P RAZETO, D RUBILAR, CSALAS, F SALINAS, E SEPÚLVEDA, C SOTO, RSUÁREZ, A VARGAS, T VEGA-ZÚNIGA, NVELÁSQUEZ, J VIDAL, C VILLAGRA & CPVILLAVICENCIO (2005) Undergraduate teachingof evolution in Chile: more than natural selection.Revista Chilena de Historia Natural 78: 169-170.

BACHMANN PA, PL LUISI & J LANG (1992)Autocatalytic self-replicating micelles as modelsfor prebiotic structures. Nature 357: 57-59.

BAKKER, RT (1986) The dinosaur heresies. Zebra books,New York, New York, USA. 481 pp.

BENTON MJ (2000) Vertebrate paleontology. Secondedit ion. Blackwell Science, Oxford, UnitedKingdom. 452 pp.

BURKE AC & A FEDUCCIA (1997) Developmentalpatterns and the identification of homologies in theavian hand. Science 278: 666-668.

CAMUS PA (2000) Evolution in Chile: natural drift versusnatural selection, or the preservation of favouredtheories in the struggle for knowledge. RevistaChilena de Historia Natural 73: 215-219.

13BEYOND SELECTION

CALDWELL MW & MSY LEE (1997) A snake with legsfrom the marine cretaceous of the middle east.Nature 386: 705-709.

CARROL R (1988) Vertebrate paleontology and evolution.W.H. Freeman & Co., New York, New York, USA.698 pp.

CHATTERJEE S (1998) Counting the fingers of birds anddinosaurs. Science 280: 355.

CHIAPPE LM (1997) Aves. In: Currie P & K Padian (eds)The encyclopedia of dinosaurs: 32-38. AcademicPress, New York, New York, USA.

CHIAPPE LM & A VARGAS (2003) Emplumandodinosaurios: la transición evolutiva de terópodos aaves. Hornero 18: 1-11.

DARWIN C (1859) [1977] El origen de las especies pormedio de la selección natural. Editorial Edaf,Madrid, España. 533 pp.

DAWKINS R (1976) [1993] El gen egoísta. SalvatEdiciones, Barcelona, España. 407 pp.

DOMNING DP (2001) The earl iest known fullyquadrupedal sirenian. Nature 413: 625-627.

FEDUCCIA A (1999) 1,2,3 = 2,3,4: accomodating thecladogram. Proceedings of the National Academyof Sciences USA 96: 4740-4742.

FEDUCCIA A (2002) Birds are dinosaurs: simple answerto a complex problem. Auk 119: 1187-1201.

FEDUCCIA A (2003) Bird origins: problem solved, butthe debate continues. Trends in Ecology andEvolution 18: 9-10.

GALIS F, J VAN ALPHEN & J METZ (2001) Why fivefingers? Evolutionary constraints on digit numbers.Trends in Ecology and Evolution 16: 637-646.

GALIS F, M KUNDRÁT & B SINERVO (2003) An oldcontroversy solved: bird embryos have five fingers.Trends in Ecology and Evolution 18: 7-9.

GILBERT SF, JM OPITZ & RA RAFF (1996)Resynthesizing evolutionary and developmentalbiology. Developmental Biology 173: 357-372.

GOULD SJ (1977) Ontogeny and phylogeny. The BelknapPress of Harvard University Press, Cambridge,Massachusetts, USA. 501 pp.

GOULD SJ (2002) The structure of evolutionary theory.The Belknap Press of Harvard University Press,Cambridge, Massachusetts, USA. 1,433 pp.

GOULD SJ & RC LEWONTIN (1979) The spandrels ofSan Marco and the Panglossian paradigm: a critiqueof the adaptationist program. Proceedings of theRoyal Society of London B 205: 581-598.

GOULD SJ & ES VRBA (1982) Exaptation: a missingterm in the science of form. Paleobiology 8: 4-15.

HALL BK (1998) Evolutionary developmental biology.Chapman and Hall, London, United Kingdom. 516 pp.

HENNIG W (1965) Phylogenetic systematics. AnnualReview of Entomology 10: 97-116.

JI Q, P CURRIE, MA NORELL & S-A JI (1998) Twofeathered dinosaurs from northeastern China.Nature 393: 753-761.

KIMURA M (1968) Evolutionary rate at the molecularlevel. Nature 217: 624-626.

KNOLL AH & SB CARROLL (1999) Early evolution:emerging views from comparative biology andgeology. Science 284: 2129-2137.

LEWONTIN RC (1974) The genetic basis of evolutionarychange. Columbia University Press, New York,New York, USA. 346 pp.

LUISI PL (2003) Autopoiesis: a review and a reappraisal.Naturwissenschaften 90: 49-59.

MATURANA H & B PÖRKSEN (2004) Del ser al hacer.Los orígenes de la biología del conocer. JC Sáez,Santiago, Chile. 239 pp.

MATURANA H & J MPODOZIS (2000) The origin ofspecies by means of natural drift. Revista Chilenade Historia Natural 73: 261-310.

MATURANA H & J MPODOZIS (1992) Origen de lasespecies por medio de la deriva natural o ladiversif icaión de los l inajes a través de laconservación y cambio de los fenotiposontogénicos. Museo Nacional de Historia Natural,Publicación Ocasional (Chile) 46: 1-48.

MATURANA H & F VARELA (1973) De máquinas y seresvivos. Editorial Universitaria, Santiago, Chile. 122 pp.

MATURANA H & F VARELA (1980) Autopoiesis andcognition: the realization of the living. D. ReídlePublishing Company, Dordrecht, The Netherlands.140 pp.

MATURANA H & F VARELA (1984) El árbol delconocimiento. Editorial Universitaria, Santiago,Chile. 172 pp.

MAYR E (1998) Así es la biología. Editorial Debate,Madrid, España. 326 pp.

MCMULLIN B (2000) Remarks on autocatalysis andautopoiesis. Annals of the New York Academy ofScience 901:163-74.

MEYER A & R ZARDOYA (2003) Recent advances in the(molecular) phylogeny of vertebrates. AnnualReviews in Ecology and Evolutionary Systematics34: 311-348.

MILNER R (1990) The encyclopedia of evolution. Factson File Press, New York, New York, USA. 481 pp.

MÜLLER GB (2003) Embryonic motility: environmentalinfluences and evolutionary innovation. Evolution& Development 5: 56-60.

NESPOLO RF (2003) Evolution by natural selection: moreevidence than ever before. Revista Chilena deHistoria Natural 76: 699-716.

NEWMAN SA & GB MÜLLER (2000) Epigeneticmechanisms of character origination. Journal ofExperimental Zoology B 288: 304-317.

NORELL MA & MJ NOVACEK (1992) The fossil recordand evolution. Comparing cladist ics andpaleontologic evidence for vertebrate history.Science 255: 1690-1693.

PADIAN K & L CHIAPPE (1998) The origin and earlyevolution of birds. Biological Reviews 73: 1-42.

PALMER AR (2004) Symmetry breaking and theevolution of development. Science 306: 828-833.

PAUL G (1988) Predatory dinosaurs of the world. Simon& Schuster, New York, New York, USA. 464 pp.

PENNISI E (2005) Bird wings really are like dinosaurs’hands. Science 307: 194-195.

PETERSON KJ & DJ EERNISSE (2001) Animalphylogeny and the ancestry of bilaterians:inferences from morphology and 18S r DNAsequences. Evolution & Development 3: 170-205.

POPPER K (1976) El darwinismo como programametafísico de investigación. Third edition. Tecnos,Madrid, España. 320 pp.

PRUM RO (2002) Why ornithologists should care aboutthe theropod origin of birds. Auk 119: 1-17.

PRUM RO (2003) Are current critiques of the theropodorigin of birds science? Rebuttal to Feduccia(2002). Auk 120: 550 -561.

RAFF R (1996) The shape of life. University of ChicagoPress, Chicago, Illinois, USA. 520 pp.

SERENO PC (1999) The evolution of dinosaurs. Science284: 2137-2147.

SHU D-G, H-L LUO, S CONWAY MORRIS, X-LZHANG, S-X HU, L CHEN, J HAN, M ZHU, Y LI& L-Z CHEN (1999) Lower Cambrian vertebratesfrom south China. Nature 402: 42-46.

14 VARGAS

THOMAS ALR & JP GARNER (1998) Are birdsdinosaurs? Trends in Ecology and Evolution 13:129-130.

THOMPSON DAW (1917) [1977] On growth and form(abridged edition). Cambridge University Press,Cambridge, United Kingdom. 346 pp.

VARGAS AO & JF FALLON (2005) Birds have dinosaurwings: the molecular evidence. Journal ofExperimental Zoology B 304: 85-89.

VICUÑA R (2004) Alma humana y evolución. Revista deEstudios Médico Humanísticos (Chile) 1: 1.

WADDINGTON CH (1953) Genetic assimilation of anacquired character. Evolution 7: 118-126.

WAGNER GP & BY MISOF (1993) How can a characterbe developmentally constrained despite variation indevelopmental pathways? Journal of EvolutionaryBiology 6: 449-455.

WAGNER GP & JA GAUTHIER (1999) 1, 2, 3 = 2, 3, 4: asolution to the problem of homology of the digits ofthe avian hand. Proceedings of the NationalAcademy of Sciences USA 96: 5111-5116.

WAGNER GP, C-H CHIU & M LAUBICHLER (2000)Developmental evolution as a mechanistic science:the inference from developmental mechanisms toevolutionary processes. American Zoologist 40:819-831.

WATSON AJ & JE LOVELOCK (1983) Biologicalhomeostasis of the global environment: the parableof Daisyworld. Tellus 35B: 284-289.

WEISS KM (2005) The phenogenetic logic of life. NatureReviews in Genetics 6: 36-46.

WEISS KM & SM FULLERTON (2000) Phenogeneticdrift and the evolution of genotype-phenotyperelationships. Theoretical Population Biology 57:187-95.

WEISS KM & AV BUCHANAN (2003) Evolution byphenotype. A biomedical perspective. Perspectivesin Biology and Medicine 46: 159-82.

WEST-EBERHARD MJ (2003) Developmental plasticityand evolution. Oxford University Press, New York,New York, USA. 794 pp.

WITMER LM (2002) The debate on avian ancestry:phylogeny, function, and fossils. In: Chiappe LM &LM Witmer (eds) Mesozoic birds: above the headsof dinosaurs: 3-30. University of California Press,Berkeley, California, USA.

WHITMAN CO (1919) Orthogenetic evolution in pigeons.Carnegie Institute of Washington Publications 257.

XU X, Z-H ZHOU & RO PRUM (2001) Branchedintegumental structures in Sinornithosaurus and theorigin of feathers. Nature 410: 200-204.

WU XC, Z LI, BC ZHOU & ZM DONG (2003)Palaeontology: a polydactylous amniote from theTriassic period. Nature 426: 516.

XU X, Z ZHOU, X WANG, X KUANG, F ZHANG & YXDU (2003) Four-winged dinosaurs from China.Nature 421: 335-340.

ZHOU Z (2004) The origin and early evolution of birds:discoveries, disputes, and perspectives from fossilevidence. Naturwissenschaften 91: 455-471.

Associate Editor: Patricio CamusReceived August 30, 2004; accepted March 7, 2005