ARTICLE IN PRESS TREE 464 Opinion TRENDS in Ecology and Evolution Vol.xx No.xx Monthxxxx Full text provided by www.sciencedirect.com -ů° Genome duplication, extinction and vertebrate evolution Philip C.J. Donoghue1 and Mark A. Purnell2 department of Earth Sciences, University of Bristol, Wills Memorial Building, Queen's Road, Bristol, UK, BS8 1RJ department of Geology, University of Leicester, University Road, Leicester, UK, LE1 7RH Vertebrate evolution has been punctuated by three episodes of widespread gene or genome duplication, which have been linked with the origin of vertebrates, gnathostomes and teleosts, respectively. These three events coincide with bursts of character acquisition and increases in phenotypic complexity, and many researchers have suggested a causal relationship between the two. However, this pattern is derived from data for living taxa only; we argue here that, when fossils are taken into account, bursts of character acquisition disappear and gen(om)e duplication in vertebrate phylogeny can no longer be correlated with the origin of body plans. If patterns of character acquisition or morphological gaps between higher taxa are a reflection of phenotypic complexity, then more inclusive data sets incorporating fossil taxa provide no support for hypotheses linking gen(om)e duplications and the evolution of complexity in vertebrates. Duplication, duplication, duplication The distinction between invertebrates and vertebrates has long been considered fundamental, and it pervades all discussions of animal biology. With the recognition that invertebrates are paraphyletic (see Glossary) and do not constitute a natural group, the distinction became little more than a convenient shorthand in communication, but discoveries in molecular biology over the past decade have revealed that the division has real genetic significance. Comparative analysis of invertebrate and vertebrate genes has borne out earlier suggestions [1] that vertebrate genomes are more complex than are those of their invertebrate relatives. Hox genes, a family of homeobox-containing transcription factors that lie in adjacent positions along a chromosome, have been particularly significant in these explorations of molecular complexity. Adjacent Hox genes form a 'cluster', and all invertebrates, including amphioxus (the closest living invertebrate relative of vertebrates), have only one Hox cluster. All vertebrates, however, have more than one [2]. A similar pattern has been found in many other gene families [3-5], and it now appears that the origin of vertebrates coincided with a widespread (possibly even genome-wide) gene duplication event [3]. Indeed, continued surveying has revealed that vertebrate evolution has been punctuated by at least two other duplication events, one at the origin of the gnathostomes [6] and the other somewhere along the lineage leading to teleost fishes, after their divergence from sturgeons [7] or gars [8]. All three duplication events are associated either with dramatic jumps in morphological complexity or with adaptive radiations and innovations of body design. This has led several authors to suggest a relationship between gene duplication and complexity [5-7,9-23], with some arguing that increases in complexity were caused by gene duplications [11,13,24]. We contend, however, that the apparent correlation between the phylogenetic timing of duplication events and increases in phenotypic complexity and diversity is an artefact of incomplete taxonomie sampling. Evolutionary jumps, fossils and extinction A fundamental problem with the hypothesis linking gen(om)e duplication events in vertebrates with evolutionary jumps is that extinct lineages are ignored. This is unfortunate because, in all three instances, the gulf between the living branches of the vertebrate tree is bridged by a series of extinct clades that are taxonomically and anatomically intermediate. These fossils can provide insight into the nature of morphological evolution through the period in which gen(om)e duplication is implicated, Glossary Corresponding author: Donoghue, P.C.J, (phil.donoghue@bristol.ac.uk). Adaptive radiation: rapid diversification of a lineage, linked with adaptation to a range of different ecological niches. Apomorphy: an advanced group-diagnostic character. Basal clade: a monophyletic group that occurs low within the topology of a phylogenetic tree. Clade: monophyletic group. Crown group: least inclusive clade encompassing the living members of a group and the extinct taxa that nest among them. Monophyly: the condition where a group contains the common ancestor of all its members and all the descendents of that common ancestor. Paralogue: used to describe two or more genes that occur within a genome that are not homologous but are derived through duplication from a single ancestral gene. Paraphyly: the condition where a group contains the common ancestor of all its members, but not all of the descendents of that common ancestor. Plesiomorphy: a primitive trait, not diagnostic within the phylogenetic context considered; the quality of primitiveness. Symplesiomorphy: a shared, primitive trait that diagnoses a more inclusive clade. Stem lineage: a paraphyletic assemblage of extinct taxa that are not members of a given crown group, but are more closely related to it than they are to any other crown group. For example, stem vertebrates lie outside the vertebrate crown group, but are more closely related to crown vertebrates than they are to their nearest living relative, amphioxus (a crown cephalochordate). www.sciencedirect.com 0169-5347/$ - see front matter © 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.tree.2005.04.008 ARTICLE IN PRESS TREE 464 Opinion TRENDS in Ecology and Evolution Vol.xx No.xx Monthxxxx constraining models of developmental evolution, rates of character acquisition and hypotheses of adaptive radiation. In the case of comparative genomics, omission of fossils is a matter of necessity, given that no genes are preserved. However, when attempts are made to compare or integrate analyses of genomic and morphological evolution, there is no logical reason for excluding extinct taxa. One consequence of their exclusion is that attempts at integrated analysis are based on an incomplete morphological data set. More significantly, analyses based on extant diversity alone reveal an apparent pattern of bursts in character evolution and morphological complexity that is misleading. We argue here that this is not merely a scientific inconvenience: the pattern is a consequence of the incomplete sampling of taxonomie and morphological diversity. The best understood of these three important episodes of vertebrate phylogeny is the origin of Gnathostomata, where accelerated rates of increasing morphological complexity have been identified as coincident with an inferred genomic doubling [4-7,11,17,22,25]. All living vertebrates, except the hagfishes and lampreys, are members of Gnathostomata, but what are gnathostomes? Etymologically, and in common parlance, gnathostomes are widely considered to be vertebrates with jaws. Curiously, however, they are both more and less than this. Gnathostomes: more than just a pretty smile The jaw is just one of a long inventory of characters that separate extant jawed vertebrates from lampreys, their nearest living relatives (Mazan et al. [26], for example, record 56 characters, and this is a considerable underestimate). Indeed, most of the typical features of what is generally thought of as the vertebrate body plan (e.g. a phosphatic internal skeleton, vertebrae and paired appendages) are gnathostome characters, and some authors recognize a distinct gnathostome body plan [27]. By any measure, there is an abrupt step-change in the rate of phenotypic character acquisition in the portion of the phylogenetic tree between living jawless and jawed vertebrates and this is congruent with patterns of genetic complexity. Does this abrupt change constitute evidence of correlated increases in molecular and morphological complexity? Possibly, but incorporation of palaeontological data reveals a much less dramatic pattern of morphological and physiological change. This is because a spectrum of extinct lineages lies between the extant jawless and jawed vertebrates (Figure 1). We can resolve their intermediate position, and their degrees of relationship to living jawed vertebrates because, in addition to the plesiomorphic vertebrate characters that they all share, they also have subsets of that inventory of characters that distinguishes living jawed vertebrates from lampreys. For instance, conodonts, which were otherwise anatomically comparable to lampreys, have a single gnathostome apomorphy (a mineralized skeleton), indicating that they are more closely related to living jawed vertebrates than they are to lampreys [28]. Pteraspidomorphs have a mineralized dermoskeleton, as do anaspids, which also have an anal fin and a distinct stomach, betraying still closer affinity to sharks and bony fishes. Galeaspids have these characters and more, including a mineralized endoskeleton, which indicates that they are even more closely related to living jawed vertebrates [29,30]. And osteostracans are more closely related still, having paired pectoral appendages, slit-shaped gill openings, a dorsally elongated tail fin, cellular bone, sclerotic ring and ossified sclera; all characters that are absent from lampreys but present in living jawed vertebrates [29,31]. Finally, placoderms, the extinct group that is most closely related to living jawed vertebrates, diminish the inventory of characters that distinguish extant jawed vertebrates still further, having mineralized neural and haemal vertebral elements, a horizontal semi-circular canal, paired pelvic fins and, among other characters, jaws. Few characters remain that are exclusive to living jawed vertebrates, of which the anterodorsal attachment of the superior oblique eye muscle [31] is perhaps the most convincing. Is this sufficient justification for recognizing a distinct gnathostome body plan? One might argue that placoderms are more appropriately considered members of Gnathostomata because, after all, they have jaws. But why should this single character be considered more 'significant' or 'essential' than any of the many other characters that distinguish sharks and bony fishes from lampreys? As with the other extinct lineages considered above, placoderms do not exhibit the full complement of characters shared by living jawed vertebrates. If placoderms are to be incorporated into Gnathostomata, then so should all the others, from osteostracans to conodonts, which on the basis of gnathostome synapomorphies are demonstrably more closely related to sharks and bony fishes than they are to lampreys. Regardless of whether these extinct clades are formally considered members of Gnathostomata, their affinity to extant jawed vertebrates is conventionally recognized through their inclusion in what is known as the gnathostome total group, within which the extant lineages are distinguished as crown Gnathostomata, and the remaining paraphyletic ensemble of extinct lineages as stem Gnathostomata [28]. Stem lineages and evolutionary patterns Whichever way one classifies this parade of dead fishes, stem gnathostomes are fundamental to our discussion because their subsets of gnathostome characters reveal that the construction of the gnathostome body plan occurred piecemeal and, if the dates of appearance of these groups in the fossil record are anything to go by [32], this assembly of characters occurred over a protracted period of time between the Middle Cambrian and the Late Ordovician, or possibly even later, a matter of 70 million years (myr) or more (molecular clock dates suggest an even more protracted interval [32]). Stem gnathostomes also demonstrate that the pattern of character acquisition was not smooth, with more characters acquired at some nodes than at others (Figure 1). This apparent variation in rate might be real, but, given the nature of the fossil record, it is possible that it is itself an artefact of incomplete taxon sampling in that we might have omitted some extinct lineages that are as yet unknown or incompletely understood from the fossil record [32]. After www.sciencedirect.com ARTICLE IN PRESS TREE 464 Opinion TRENDS in Ecology and Evolution Vol.xx No.xx Monthxxxx 100 90 80 >,70 ľ 60 Š 50 5 40 30 20 10 1 1 SI em G TfTlifKstomes Crown Q. i« aj O >. — E O) O) ro o T3 ra c ■a o ro E o T3 CT (1) £• o T3 o c o Q. ro (Ď ■a Q. ro c o ■a o aj .c ro . U) Ü. B 9- tS -y -2 ! e «k ~ Q- « ■g III i L¥ mjj u Duplication TRENDS in Ecology & Evolution Figure I. Hypothesis of relationships among living (black lines) and extinct (white lines) clades of ray-finned fishes. The vertical bars and scale above the cladogram show diversity (number of families, total for each clade; values for most derived teleosts not shown). The grey box shows the degree of uncertainty in placing the fish-specific duplication event within the teleost stem. The relationships among the most derived non-teleost fishes are shown as unresolved because of uncertainties concerning whether gars (Ginglymodi) or bowfin (Halecomorphi) are more closely related toteleosts [40], and because if they form a cladewith sturgeons (Acispenseroidei) [51] the positions of the intervening extinct clades are currently unknown. This lack of resolution in this part of the tree does not affect the fact that an extensive teleost stem exists, and that there is no correlation between taxon diversity and the phylogenetic timing of the duplication. The position of Osteoglossomorpha as the most basal crown teleosts follows [37,51,56,57]. Relationships and taxon counts based on [37,52-55,58], and the summary cladograms from [42]. The situation regarding the gen(om)e duplication at the origin of vertebrates is more problematic. It appears to be associated with an evolutionary jump from amphioxus to hagfish but determining whether this is real or an artefact of extinction is difficult. Only the pattern of character evolution through the vertebrate stem can reveal the answer. However, this is confused by preservational biases and difficulties inherent in assigning fossils to the stems of clades with deep divergence times (Box 2). Even among crown vertebrates, the most basal clades, hagfish and lampreys, lack mineralized tissues, and the preservation potential of chordates with no hard parts is low, even in instances of exceptional preservation. This is further compounded by the fact that most crown-vertebrate apomorphies are soft tissue, ultrastructural and especially embryological characters. Therefore, the current inventory of exceptionally preserved soft-bodied animals could include the remains of stem vertebrates that have not www.sciencedirect.com ARTICLE IN PRESS TREE 464 Opinion TRENDS in Ecology and Evolution Vol.xx No.xx Monthxxxx Box 2. Recognizing stem vertebrates Unravelling the pattern of character acquisition through the origin of vertebrates is complicated by the low preservation potential of chordates lacking mineralized skeletal components, and the absence of any unequivocal stem vertebrates in the known fossil record. Conodonts provide an interesting illustration of this preservational bias. Their fossil record extends from the late Cambrian to the end of the Triassic (some 300 myr), and comprises millions of specimens of thousands of species. However, if it were not for the phosphatic elements of the conodont feeding apparatus (their sole gnathostome apomorphy), the known fossil record would comprise 12 specimens assigned to three species, the number t hat preserve t races of t he body. Furthermore, some of these specimens preserve few vertebrate characters, and it is doubtful whether one would be recognized as a chordate if it were not clear from its phosphatic skeletal apparatus that it is a conodont. Most crown-vertebrate apomorphies are soft tissue, ultrastructural and embryological characters and, for this reason, phylogenetic placement of the few fossils currently identified as hagfishes and lampreys is open to question (Figure I). Although they have features that are interpreted as hagfish and lamprey apomorphies, they lack key vertebrate characters [59-62], and some could be stem vertebrates. Similarly, Pikaia, Yunnanozoon, Haikouella and Myllo-kunmingia from the Lower Cambrian Chengjiang Fauna, as well as a spectrum of mitrate carpoids, are candidate stem vertebrates. With the recognition that echinoderms and hemichordates are each other's nearest relatives, however, the many putative homologies uniting carpoids and vertebrates have been resolved as deuterostome symplesiomorphies, and carpoids are most readily interpreted as stem echinoderms [63]. Of the remaining taxa, it is currently unclear whether the chordate characters that they have are unique to the crown group, or appeared somewhere along the chordate stem. Some, such as a perforated pharynx and pharyngeal arches, are evidently deuterostome symplesiomorphies [63]. If these fossils do not unequivocally qualify as members of the chordate crown group, interpretations placing them as stem cephalochordates or stem vertebrates are even more doubtful. From what we know of other major clades, it seems most unlikely that there was not an extensive vertebrate stem. Unfortunately, however, we lack the evidence needed even to begin an assessment of the taxon richness or character evolution, and their possible congruence with gen(om)e duplication, through this crucial phase of vertebrate evolution. i c to c I LJ X LU O Q. • g to E o) o O -L £ D- ° O TRENDS in Ecology & Evolution Figure I. Hypothesis of relationships among living (black lines) and extinct (white lines) clades of deuterostomes. Relationships of extant groups are based on [63] and references therein. The affinities of Vetulicolia are extremely uncertain. Yunnanozoon and the suspiciously similar Haikouella have been variously interpreted as stem deuterostomes, hemichordates, cephalochordates, craniates or vertebrates owing to equivocation over interpretation and character polarity. Pikaia and Cathaymyrus have hitherto been recognized as fossil cephalochordates but they fail to exhibit any unequivocal cephalochordate apomorphies. been recognized as such. Candidates include Pikaia from the Middle Cambrian Burgess Shale, and Yunnanozoon, Haikouella and Myllokunmingia from the Early Cambrian Chengjiang Fauna, but none of these can be placed unequivocally within the vertebrate stem (Box 2). What a more inclusive and integrated perspective on vertebrate phylogeny reveals, therefore, is that the purported jumps in character acquisition and diversity in vertebrates coincident with the gen(om)e duplication events are artefacts of higher than average pruning of lineages from the Tree of Life by extinction. Neither is there anything particularly special about the extent of these stem lineages that could be linked to inferred gen(om)e duplication events. Tetrapoda, Mammalia and Aves also have extensive stem lineages [42], and we should not be seduced into thinking that there is anything special about the nature of the origin of vertebrate higher taxa just because they have long stem lineages. Rather, the anatomical distinctiveness of these groups of organisms is an artefact of extinction of phylogenetic intermediates. Constraining the timing of gen(om)e duplication events Although the elucidation of stem vertebrates, stem gnathostomes and stem teleosts refutes the hypothesis that dramatic jumps in character acquisition are associated with the origin of their respective crown groups, some authors have been careful to couch their hypotheses of causality between gen(om)e duplication, molecular and morphological evolution in terms that allow for these events to have occurred somewhere between extant lineages [6,18,43]. Yet the precise timing of the inferred www.sciencedirect.com ARTICLE IN PRESS TREE 464 Opinion TRENDS in Ecology and Evolution Vol.xx No.xx Monthxxxx gen(om)e duplications relative to morphological character evolution is crucial to the entire thesis of causality (or permissiveness [6]). If the gen(om)e doubling were found to postdate the acquisition of most of the characters within the stem, hypotheses of causality and permissiveness would be rejected; gen(om)e duplication would be consigned to an interesting but insignificant phenomenon of vertebrate evolution. Clearly, only data from stem taxa can resolve this issue. We are never going to be able to sequence any of their genes but it might be possible to constrain the timing of gen(om)e duplication by inferring the existence of new paralogues from evidence of their first phenotypic expression in fossils from within the stem (phenotypic expression being understood from studies of knockout, mis-expression and in situ hybridization in extant crown group representatives). Such fossil evidence would postdate true paralogue origin (and even origin of function) but might provide some insight into the relative timing of gen(om)e duplication events (i.e. early or late within the stem). For example, Dlx genes occur as six paralogues in gnathostomes (Dlxl-6), and four in lampreys (DlxA-D). Two of the lamprey paralogues (DlxA and DlxC) appear to be unique to lampreys, whereas DlxD and DlxB appear to be orthologous to gnathostome Dlxl, Dlx6, Dlx7, and Dlx2, Dlx3, Dlx5, respectively [4]. In gnathostomes, all six genes coordinately regulate jaw [17] and tooth development [44]; Dlx3, Dlx5 and Dlx6 are integral to cartilage, bone and tooth development in the pharyngeal, axial and appendicular skeletons [45]. Given that placoderms are the most plesiomorphic vertebrates in which these skeletal systems occur in mineralized form [46], it would not be unreasonable to conclude that manifest function of gnathostome Dlx paralogues was late in the gnathostome stem, coincident with the acquisition of jaws. However, all six paralogues are also expressed in fin and/or limb development [45], and the presence of pectoral fins in osteostra-cans and fin-like structures in even more plesiomorphic members of the gnathostome stem (e.g. anaspids and thelodonts) could, therefore, indicate much earlier manifestation oiDlx function. However, deletion experiments show that, with respect to the skeleton, there is considerable redundancy between paralogues [45], perhaps because a similar range of functions was once performed by a smaller repertoire of ancestral genes. In other words, function and phenotype were acquired before the origin of gnathostome paralogues, among which function was subsequently partitioned. This lies at the heart of the duplication-degeneration-complementation model of paralogue fate following gene duplication [47]. Under this model, paralogues, which initially are functionally redundant, can be either silenced or retained, with retention favoured where paralogues acquire novel functions (neofunctionalization), or because they accumulate complementary degenerative mutations such that both are required for combined maintenance of ancestral functions (subfunctionaliza-tion). Subfunctionalization in particular provides an explanation for the prevalence of paralogues with a partitioning of ancestral expression patterns and protein functional domains. One corollary of the subfunctionalization model is that complexity, in terms of the body plan, regulatory gene networks and/or gene expression patterns, is established before gen(om)e duplication, and genomic complexity is thus an effect, rather than the cause of developmental and/or phenotypic evolution (although neither model precludes further increases in developmental complexity and, potentially, phenotypic complexity, after duplication). With respect to Dlx and gnathostomes, however, under either model of paralogue retention, the phenotypic effect of Dlx duplication occurred late within the gnathostome stem, after many gnathostome characters had already been acquired. Testing the congruence of genetic and phenotypic complexity The hypothesis that gen(om)e duplications have been significant in facilitating or driving vertebrate evolution requires duplications to be congruent with increases in phenotypic complexity and/or morphological innovation and/or taxon richness. If only extant taxa are considered, the coincidence of character acquisition and taxon richness with duplication events in the lineages leading to living jawed vertebrates and to teleosts, for example, seems obvious, but it is an artefact of incomplete taxon sampling, resulting entirely from the unjustified exclusion of extinct lineages. The pattern disappears when the fossil record of extinct members of those lineages is taken into account. However, it could be argued that the hypothesis is not falsified by this result because phenotypic complexity is something more than character acquisition or taxon richness. So is there a correlation between genomic and phenotypic complexity? Unfortunately, finding an answer is problematic. First, research effort is strongly biased towards the molecular side of the hypothesis and, in contrast to the hundreds of papers investigating genome complexity, very few have attempted to analyse phenotypic complexity, innovation or taxon richness through the phylogenetic intervals in question (see Box 3 for a discussion of the interrelationships between complexity, innovations, taxon richness and body plans). This might be because, when extant taxa only are considered, the apparent evolutionary bursts within vertebrate phylogeny appear too obvious to be worth further enquiry, but extinct taxa cannot simply be ignored. A further problem arises from the meaning of complexity. In the context of duplication events, 'genomic complexity' is unambiguous. However, with very few exceptions [48], the meaning of 'phenotypic complexity' or how it might be measured has not been addressed (Box 3). We are aware of only two attempts to measure the evolution of complexity that have included the interval of vertebrate evolution of interest here: one counted cell morphotypes [48,49], the other used an index based on character counts [5]. Unfortunately, these analyses were based on extant vertebrates only (cell morphotype number cannot be determined directly in extinct vertebrates); thus, in both cases, the magnitude of the increase in complexity measured between living jawless and jawed vertebrates is, similar to the pattern of character acquisition, a www.sciencedirect.com ARTICLE IN PRESS TREE 464 Opinion TRENDS in Ecology and Evolution Vol.xx No.xx Monthxxxx Box 3. Complexity, characters and body plans Suggested correlates of vertebrate genome duplications include increases in phenotypic complexity, morphological innovation, taxon richness (i.e. diversification) and the origin of body plans. All these are linked. A detailed discussion of complexity in this context is beyond the scope of this article. 'Complexity' is considered by some authoritiesto have lost any precision or general meaning [64,65], but there is a degree of general agreement that the morphological complexity of an organism is some function of its component parts, with more complex organisms have more parts or more interactions between parts. The difficulties of actually measuring organismal complexity, however, are not insignificant. Some progress has been made in formulating narrower definitions of complexity [65,66] and these at least provide a context within which comparisons can be made and scientific questions framed, but research into patterns of phenotypic complexity has not progressed much beyond this. With the single exception noted in the main text [48], these problems remain unacknowledged in hypotheses of genome duplication and complexity increase in vertebrates. Used without qualification or explicit explanation, 'phenotypic complexity' has no meaning or value in scientific investigation. Body plans are linked to complexity through the hypothesis that the origin of body plans and establishment of body-plan characters are emergent properties of increasing phenotypic complexity [67]. Characters that define the body plan are those that attained greater importance and became evolutionary crystallized because other characters are functionally or developmentally predicated on them. This idea applies to both phenotypic and genomic complexity, and has some theoretical support [68], but one of the difficulties in this area of research is determining whether body plans are real or merely perceptual artefact [67]. Consideration of the stems of extant clades points strongly toward perceptual artefact. Some authors have linked genomic duplication to innovation, rather than phenotypic complexity per se, but this is really just character acquisition by another name in that: • Innovations (or 'key innovations') are causally linked to intervals of accelerated phenotypic evolution (reviewed in [69]); • Such intervals are usually identified by increases in taxon richness; • From a morphological perspective, an interval of increased taxon richness is the same thing as an increase in rate of character acquisition. A further difficulty with key innovations comes in establishing the causal relationship with accelerated evolution. However, multiple comparisons of the evolutionary history of clades with and without a putative innovation, and some level of correspondence between the appearance and diversification of the clade are generally taken to be valid tests [70]. It is difficult to see how this could be applied in the context of vertebrate and gnathostome origin, or teleost diversification, however, as they are, by their nature, one-off events. The relationship between key innovations and body-plan characters is more subtle, but becomes clearer when extinction is taken into account. Both refer to characters shared by a clade that are interpreted as having special significance and are correlated with diversification. The only difference is that, in the case of body-plan characters, diversification occurred deep within phylogeny and the subsequent extinctions of basal taxa have left the clade in question as a terminal survivor, morphologically isolated from its nearest living relative. consequence of missing all the taxa in intermediate phylogenetic positions. The upshot is that we cannot currently test for congruence between genomic duplications and increases in phenotypic complexity (or morphological innovation; Box 3) because these concepts are defined too loosely for falsification of the hypothesis. If we are to incorporate the fossil record into the hypothesis (and to do otherwise renders analysis futile) then the taxon richness of clades is directly correlated with rates of character acquisition and this, as we have shown, provides no support for congruence. This discussion also serves to emphasize a broader point: to understand fully the nature of developmental evolution it is necessary to include extinct, not just extant, organisms. The fossil record might be famed for its incompleteness but it is far more complete than a phylogeny that includes only those organisms that, by chance, are alive today. Acknowledgements We thank the following for discussion, comments and reviews of the article: Mike Bell, Mike Benton, Peter Forey, Anthony Graham, Philippe Janvier, Gunter Wagner and three anonymous reviewers. Funding from NERC (GT5/99/ES2 to P.C.J.D.; GT5/98/4/ES and NER/J/S/2002/00673 to M.A.P.) and NESTA (P.C.J.D) is gratefully acknowledged. References 1 Ohno, M. (1970) Evolution by Gene Duplication, Springer-Verlag 2 Garcia-Fernändez, J. and Holland, P.W.H. (1994) Archetypal organisation of the amphioxus Hox gene cluster. Nature 370, 563-566 3 Furlong, R.F and Holland, P.W.H. (2002) Were vertebrates octoploid? Philos. Trans. R. Soc. Land. Ser B Biol. Sei. 357, 531-544 4 Neidert, A.H. et al. (2001) Lamprey Dlx genes and early vertebrate evolution. Proc. Natl. Acad. Sei. U. S. A. 98, 1665-1670 5 Aburomia, R. et al. (2003) Functional evolution in the ancestral lineage of vertebrates or when genomic complexity was wagging its morphological tail. J. Struct. Fund. Genom. 3, 45-52 6 Holland, P.W.H. et al. (1994) Gene duplications and the origins of vertebrate development. Development (Suppl.), 125-133 7 Wagner, G.P et al. (2003) Hox cluster duplications and the opportunity for evolutionary novelties. Proc. Natl. Acad. Sei. U. S. A. 100, 146043-114606 8 Hoegg, S.etal. (2004) Phylogenetic timing of the fish-specific genome duplication correlates with the diversification of teleost fish. J. Mol. Evol. 59, 190-203 9 Ruddle, FH. et al. (1994) Gene loss and gain in the evolution of the vertebrates. Development (Suppl.), 155-161 10 Holland, P.W.H. (1992) Homeobox genes in vertebrate evolution. BioEssays 14, 267-273 11 Sidow, A. (1996) Gen(om)e duplications in the evolution of early vertebrates. Curr Opin. Genet. Dev. 6, 715-722 12 Carroll, S.B. (1995) Homeotic genes and the evolution of arthropods and chordates. Nature 376, 479-485 13 Ruddle, FH. et al. (1994) Evolution oiHox genes. Ann. Rev. Genet. 28, 423-424 14 Holland, P.W.H. and Garcia-Fernändez, J. (1996) HOX genes and chordate evolution. Dev. Biol. 173, 382-395 15 Amores, A. et al. (1998) Zebrafish hox clusters and vertebrate genome evolution. Science 282, 1711-1714 16 Holland, P.W.H. (1999) Introduction: gene duplication in development and evolution. Sem. Cell Dev. Biol. 10, 515-516 17 Depew, M.J. et al. (2002) Specification of jaw subdivisions by Dlx genes. Science 298, 381-385 18 Shimeld, S.M. and Holland, P.W.H. (2000) Vertebrate innovations. Proc. Natl. Acad. Sei. V. S. A. 97, 4449-4452 19 Meyer, A. and Schartl, M. (1999) Gene and genome duplications in vertebrates: the one-to-four (-to-eight in fish) rule and the evolution of novel gene functions. Curr. Opin. Cell Biol. 11, 699-704 20 Taylor, J.S. et al. (2003) Genome duplication, a trait shared by 22,000 species of ray-finned fish. Genome Res. 13, 382-390 21 Holland, P.W.H. (2003) More genes in vertebrates? J. Struct. Fund. Genom. 3, 75-84 22 Stadler, PF et al. (2004) Evidence for independent Hox gene duplications in the hagfish lineage: a PCR-based gene inventory of Eptatretus stoutii. Mol. Phylog. Evol. 32, 686-694 23 Amores, A. et al. (2004) Developmental roles of Pufferfish Hox clusters and genome evolution in ray-fin fish. Genome Res. 14, 1-10 www.sciencedirect.com ARTICLE IN PRESS TREE 464 Opinion TRENDS in Ecology and Evolution Vol.xx No.xx Monthxxxx 24 Stellwag, E.J. (1999) Hox gene duplication in fish. Sem. Cell Dev. Biol. 10, 531-540 25 Holland, P.W.H. (1999) Gene duplication: past, present and future. Sem. Cell Dev. Biol. 10, 541-547 26 Mazan, S. et al. (2000) Otxl gene-controlled morphogenesis of the horizontal semicircular canal and the origin of the gnathostome characteristics. Evol. Dev. 2, 186-193 27 Woodger, J.H. (1945) On biological transformations. In Essays on Growth and Form presented to D'Arcy Wentworth Thompson (Le Gros Clark, W.E. and Medawar, P.B., eds), pp. 95-120, Cambridge University Press 28 Donoghue, P.C.J, et al. (2000) Conodont affinity and chordate phylogeny Biol. Rev. 75, 191-251 29 Janvier, P. (1981) The phylogeny of the Craniata, with particular reference to the significance of fossil 'agnathans'. J. Vert. Paleontol. 1, 121-159 30 Janvier, P. (1984) The relationships of the Osteostraci and the Galeaspida. J. Vert. Paleontol. 4, 344-358 31 Janvier, P. (2001) Ostracoderms and the shaping of the gnathostome characters. In Major Events in Early Vertebrate Evolution: Palaeontology, Phylogeny, Genetics and Development (Ahlberg, P.E., ed.), pp. 172-186, Taylor & Francis 32 Donoghue, P.C.J, et al. (2003) The origin and early evolution of chordates: molecular clocks and the fossil record. In Telling the Evolutionary Time: Molecular Clocks and the Fossil Record (Donoghue, P.C.J, and Smith, M.P, eds), pp. 190-223, CRC Press 33 Agassiz, L. (1833-43)Recherches sur les Poissons Fossiles, Imprimerie de Petitpierre 34 Liu, Y. (1965) New Devonian agnathans of Yunnan. Vert. Paleontol. Asiat. 9, 125-134 35 Young, GC. (1991) The first armoured agnathan vertebrates from the Devonian of Australia. In Early Vertebrates and Related Problems in Evolutionary Biology (Chang, M.M. et al., eds), pp. 67-85, Science Press 36 Aldridge, R.J. et al. (1993) The anatomy of conodonts. Philos. Trans. R. Soc. Land. Ser B Biol. Sei. 340, 405-421 37 Nelson, J.S. (1994) Fishes of the World, John Wiley & Sons 38 Cresko, W.A. et al. (2003) Genome duplication, subfunction partitioning, and lineage divergence: Sox9 in stickleback and zebrafish. Dev. Dynam. 228, 480-489 39 de Pinna, M.C.C. (1996) Teleostean monophyly In Interrelationships of Fishes (Stiassny, M.L.J, et al., eds), pp. 147-162, Academic Press 40 Arratia, G (2001) The sister-group of Teleostei: consensus and disagreements. J. Vert. Paleontol. 21, 767-773 41 Purneil, M.A. (2001) Scenarios, selection and the ecology of early vertebrates. In Major Events in Early Vertebrate Evolution: Palaeontology, Phylogeny, Genetics and Development (Ahlberg, P.E., ed.), pp. 187-208, Taylor & Francis 42 Benton, M.J. (2005) Vertebrate Palaeontology, Blackwell 43 Furlong, R.F and Holland, P.W.H. (2004) Polyploidy in vertebrate ancestry: Ohno and beyond. Biol. J. Linn. Soc. 82, 425-430 44 Weiss, K.M. et al. (1998) Dynamic interactions and the evolutionary genetics of dental patterning. Crit. Rev. Oral Biol. Med. 9, 369-398 45 Panganiban, G and Rubenstein, J.L.R. (2002) Developmental functions of the Distal-less/Dlx homeobox genes. Development 129, 4371-4386 46 Donoghue, P.C.J, and Sansom, I.J. (2002) Origin and early evolution of vertebrate skeletonization. Microsc. Res. Tech. 59, 352-372 47 Force, A.G. et al. (2004) Informational accretion, gene duplication, and the mechanisms of genetic module parcellation. In Modularity in Development and Evolution (Schlosser, G and Wagner, GP, eds), pp. 315-337, Univerity of Chicago Press 48 Valentine, J.W (2000) Two genomic pathways to the evolution of complexity in bodyplans. Paleobiology 26, 513-519 49 Valentine, J.W. et al. (1994) Morphological complexity increase in metazoans. Paleobiology 20, 131-142 50 Halstead, L.B. (1993) Agnatha. In The fossil Record 2 (Benton, M.J., ed.), pp. 573-581, Chapman & Hall 51 Inoue, J.G et al. (2003) Basal actinopterygian relationships: a mitogenomic perspective on the phylogeny of the 'ancient fish'. Mól. Phylog. Evol. 26, 110-120 52 Grande, L. and Bemis, W.E. (1996) Interrelationships of Acipenser-iformes, with comments on 'Chondrostei'. In Interrelationships of Fishes (Stiassny, M.L.J, et al., eds), pp. 85-115, Academic Press 53 Gardiner, B.G et al. (1996) Interrelationships of basal neopterygians. In Interrelationships of Fishes (Stiassny, M.L.J, et al., eds), pp. 117-146, Academic Press 54 Patterson, C. (1993) Osteicthyes: Teleostei. In The fossil Record 2 (Benton, M.J., ed.), pp. 621-656, Chapman & Hall 55 Li, G-Q. and Wilson, M.V.H. (1996) Phylogeny of Osteoglossopmorpha. In Interrelationships of Fishes (Stiassny, M.L.J, et al., eds), pp. 163-174, Academic Press 56 Lauder, GV and Liem, K.F (1983) The evolution and interrelationships of the actinopterygian fishes. Bull. Mus. Comp. Zool. 150, 95-197 57 Patterson, C. and Rosen, D.E. (1977) Review of ichthyodectiform and other Mesozoic Teleost fishes and the theory and practice of classifying fossils. Bull. Am. Mus. Nat. Hist. 158, 83-172 58 Forey PL. et al. (1996) Interrelationships of elopomorph fishes. In Interrelationshipps of Fishes (Stiassny, M.L.J, et al., eds), pp. 175-191, Academic Press 59 Bardack, D. (1998) Relationship of living and fossil hagfishes. In The Biology of Hagfishes (J0rgensen, J.M. et al., eds), pp. 3-14, Chapman &Hall 60 Hou, X.G et al. (2002) New evidence on the anatomy and phylogeny of the earliest vertebrates. Proc. R. Soc. Land. Ser B Biol. Sei. 269, 1865-1869 61 Janvier, P. and Lund, R. (1983) Hardistiella montanensis n.gen. et sp. (Petromyzontida) from the Lower Carboniferous of Montana, with remarks on the affinities of lampreys. J. Vert. Paleontol. 2, 407-413 62 Poplin, C. et al. (2001) Un Myxinoide (Craniata, Hypotreti) dans le Konservat-Lagerstätte Carbonifere supérieur de Montceau-les-Mines (Allier, France). Comp. Rendu Acad. Sei. Paris, Sciences de la Terre et des planětes 332, 345-350 63 Smith, A.B. The preradial history of echinoderms. Geol. Mag. (in press) 64 Carroll, S.B. (2001) Chance and necessity: the evolution of morphological complexity and diversity. Nature 409, 1102-1109 65 McShea, D.W. (1996) Metazoan complexity and evolution: is there a trend? Evolution 50, 477-492 66 Bonner, J. T (1988) The Evolution of Complexity, Princeton University Press 67 Wagner, G (1998) Complexity matters. Science 279, 1158-1159 68 Waxman, D. and Peck, J.R. (1998) Pleiotropy and the preservation of perfection. Science 279, 1210-1213 69 Galis, F (2001) Key innovations and radiations. In The Character Concept in Evolutionary Biology (Wagner, GP, ed.), pp. 581-605, Academic Press 70 Hunter, J.P (1998) Key innovations and the ecology of macroevolu-tion. Trends Ecol. Evol. 13, 31-36 www.sciencedirect.com