* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download as a PDF
The Selfish Gene wikipedia , lookup
Evolving digital ecological networks wikipedia , lookup
Sexual selection wikipedia , lookup
Catholic Church and evolution wikipedia , lookup
Plant evolutionary developmental biology wikipedia , lookup
Sociobiology wikipedia , lookup
Dawkins vs. Gould wikipedia , lookup
Natural selection wikipedia , lookup
Darwinian literary studies wikipedia , lookup
Hologenome theory of evolution wikipedia , lookup
Evolutionary mismatch wikipedia , lookup
Evolutionary developmental biology wikipedia , lookup
Population genetics wikipedia , lookup
Punctuated equilibrium wikipedia , lookup
Theistic evolution wikipedia , lookup
State switching wikipedia , lookup
Zoology 106 (2003): 283–290 © by Urban & Fischer Verlag http://www.urbanfischer.de/journals/zoology REVIEW The power of evo-devo to explore evolutionary constraints: experiments with butterfly eyespots** Paul M. Brakefield* Institute of Biology, Leiden University, Leiden, The Netherlands Summary Examples of adaptive radiation by Darwinian evolution will always enthral, and the theory of natural selection binds the whole of biology. But is selection all-powerful? Evolutionary developmental biology is beginning to provide the background to understanding how the internal organisation of organisms can influence the tempo and direction of evolutionary change. Do the mechanisms that generate the phenotype channel, bias or limit morphological evolution? The formation of eyespots on the wings of the butterfly Bicyclus anynana is a process which is becoming comparatively well understood from the genetical, developmental and ecological perspectives. We have begun to combine this knowledge with artificial selection experiments to examine the rates at which morphological changes of a particular eyespot pattern in different directions can be achieved. Our initial selection experiment performed over twenty-five generations has shown that in contrast to predictions based on shared genetical and developmental properties, two butterfly eyespots on the same wing surface can be freely uncoupled from each other with respect to their pattern of relative sizes. This flexibility in development and in response to selection may occur because our base population is part of a lineage with a long legacy of natural selection building up genetic variation that enables independent behaviour for eyespot size, and thus it no longer reflects a tightly modular organisation of eyespots as may well have existed close to their evolutionary origin in basal Lepidoptera. Comparisons can be made between such descriptions of the potential for short-term changes in morphology in B. anynana and observed patterns of divergence among all extant species in this species-rich genus. Being able to map phenotypes onto genotypes via developmental processes provides a powerful basis for exploring genetical and developmental constraints. This will eventually lead to examples of evolutionary constraints that represent more than ‘just-so’ stories, and thus to a more balanced view of Darwinian evolution. Key words: evolution, development, generative constraints, artificial selection, morphological stasis Introduction Evolutionary biologists from a variety of different backgrounds have displayed intense interest in the concept of evolutionary constraints. Although their thoughts and conjectures have led to a rich literature (e.g. Maynard Smith et al., 1985; Schwenk, 1995; Schluter, 1996; Schlichting and Pigliucci, 1998; Gould, 2002), there are few areas of agreement and much controversy (see Antonovics and van Tienderen, 1991; Stearns, 2002). At least some of the problem of agreeing about constraints stems from the difficulty in providing convincing tests of proposed examples (Beldade and Brakefield, 2003a). We believe that an approach of integrating evolutionary genetics and developmental studies to analysing the origin of evolutionarily relevant phenotypes can yield insights about potential constraints in terms of the processes by which such phenotypes are generated (Brakefield et al., 2003). This endeavour forms part of the expanding field known as evolutionary developmental bi- * Corresponding author: Paul M. Brakefield, Institute of Biology, Leiden University, P.O. Box 9516, 2300 RA Leiden, The Netherlands; phone: ++31-71-527-4880; fax: ++31-71-527-4900; e-mail: [email protected] **Presented at the 96th Annual Meeting of the Deutsche Zoologische Gesellschaft in Berlin, in association with the Deutsche Gesellschaft für Parasitologie, Berlin, June 9–13, 2003 0944-2006/03/106/04-283 $ 15.00/0 P. M. Brakefield ology or ‘evo-devo’. Development is of central importance to evolution because the processes of development translate genotypes into phenotypes, and thus changes in development generate the variation which natural selection can sort through. Here I will describe the background to, and the results of our recent experiment which has begun to build on ‘evodevo’ studies of eyespot formation on butterfly wings to explore a proposed constraint involving the mechanisms of generating the phenotype (Beldade et al., 2002a). Our perspective is primarily one of evolutionary genetics, but in general we believe that it will only be through the application of such tests that claims for constraints will become more than biological ‘just-so’ stories (Beldade and Brakefield, 2003a). This point of view parallels in some ways the highly influential initiative of Gould and Lewontin (1979) that led to more rigorous and critical thinking about the evolutionary mechanisms that could explain what appear to be adaptive traits. What are evolutionary constraints? In theory, evolutionary constraints can be of many different types. However, historical, phylogenetic and selective constraints are not the topic of this paper (see Schlichting and Pigliucci, 1998). Rather, I am concerned here with types of internal constraints that involve how phenotypes are generated by genetical and developmental mechanisms. Nearly twenty years ago, a diverse group of evolutionary and developmental biologists representing the ‘state of the art’ proposed the following definition of a developmental constraint: “A developmental constraint is a bias on the production of variant phenotypes or a limitation on phenotypic variability caused by the structure, character, composition, or dynamics of the developmental system” (Maynard Smith et al., 1985). Schlichting and Pigliucci (1998) have pointed out that substituting the phrase ‘genetic architecture’ for ‘developmental system’ provides a good definition for genetic constraints. Indeed this lack of specificity illustrates one of the problems about evolutionary constraints, namely that we just do not have hard data for sufficient examples to be able to apply appropriate adjectives with any real conviction. Here, I use the phrase ‘generative constraints’ to refer to the class of constraints which potentially can have both genetical and developmental explanations. In this context it is worth emphasizing that for complex traits the genetic and developmental organisations may be very different. Thus, concerning eyespot formation on butterfly wings a gene network has been co-opted and elaborated in the evolution of the Lepidoptera, and during wing growth the genes of this network are involved in the establishment of organizers, in signal generation, in responses to the signals and in pigment synthesis at different stages of development. 284 The time frame An important factor in considering potential constraints on evolution is the time horizon (Table 1). Developmental biologists and analysers of the fossil record tend to think in terms of macroevolutionary change and divergence across widely disparate taxa; in other words, in terms of what could or could not be ‘built’ given sufficient time and mutational input. This leads to a more absolute view of constraints. At the other end of the spectrum, evolutionary geneticists are concerned primarily with the processes of adaptive evolution within populations that experience environmental change (Brakefield, 2003). The pattern of evolution will then depend mainly on the standing genetic variation, and on whether populations adapt and survive, or alternatively go extinct because of a lack of time. This is a more relative view of constraints, the questions being whether some changes in phenotype are in some sense easier or more likely to occur than others, and how such potential differences in tempo and direction influence the patterns of evolutionary change. In our view it is important to consider the whole continuum from absolute through to relative constraints (Beldade and Brakefield, 2003a). However, the approach to be considered here is concerned with the analysis of relative constraints which could in theory bias or channel evolutionary change. Why are constraints important to evolutionary theory? Text book accounts of evolution include wonderful descriptions of so-called patterns of adaptive radiation that involve groups of related species showing divergence in key phenotypic traits. Explanations focus on Table 1. Some descriptive terms illustrating contrasting perspectives characterising the two ends of the continuum for potential generative constraints. Absolute Constraint Relative Constraint long-term short(er)-term phylogenetic relationships local adaptation within lineages, or groups of lineages among populations; across closely related species historical view extant species in spite of mutational input standing genetic variances; responses to selection Zoology 106 (2003) 4 Exploring evolutionary constraints the role of natural selection in different ecological environments leading to divergent designs in terms of form and function. Such accounts are indeed impressive and it is perhaps all too easy to infer that natural selection is all-powerful. However, different types of generative constraint could also play a role in shaping, biasing and channeling the observed patterns of phenotypic divergence, even in such dramatic examples as the Galapagos finches (Grant, 2003), cichlid fishes in African lakes (Fryer and Iles, 1972), and the hummingbird-Heliconia system (Temeles and Kress, 2003). Perhaps some potential phenotypes could have been even more successful (i.e. climbing higher adaptive peaks) in terms of matching functional design to a specific environment, but have not become established over time because of some absolute or relative constraint associated with the process of generating them. In genetic terms, lineages may tend to follow genetic lines of least resistance when responding to environmental change (Schulter, 1996). Again, we believe that insights about such hypotheses and the role of generative constraints will ultimately only be gained through properly integrated studies using certain systems and potential examples that lend themselves to the whole range of field, laboratory and molecular analyses. In terms of rates of change and the dynamics of evolution, the opposite end of the spectrum in relation to patterns of continuing evolutionary divergence and adaptive radiation is morphological stasis (Jackson and Cheetham, 1999). Evolutionary geneticists and biologists, especially those interested in natural selection and the process of adaptation, tend to focus on patterns of phenotypic divergence within groups of related species. However, in many respects lineages that show little or no change over long periods of time in the fossil record are of equal fascination. There are now several case studies where thorough and careful sampling, sometimes in combination with ecological surveys of extant taxa, provide more rigorous descriptions of patterns of stasis occurring in many (but not all) lineages. In the case of marine bryozoans (reviewed by Jackson and Cheetham, 1999), studies of extant species have revealed the life history consequences of changes in morphology, and also demonstrate the presence of heritable variation for the morphological traits within species. Why do such taxa show patterns of stasis? Generative constraints could play a role, although it seems more likely that explanations are to be found in the processes of ecology and population structure that determine the chance of novel phenotypes which have become established in local populations to spread throughout the species range, and thus eventually to become detectable in the fossil record (Eldredge et al., 2003). Zoology 106 (2003) 4 Eyespot formation and ‘evo-devo’ Some years ago, we began to integrate studies on evolutionary genetics and development using a system of eyespot formation on the wings of butterflies (reviewed in Brakefield and French 1999; Beldade and Brakefield 2002). The work was built on the classic experiments by Fred Nijhout on eyespot development in Precis (= Junonia) coenia (Nijhout 1980). Our research used an African butterfly, Bicyclus anynana, which can be readily reared in the laboratory with a short generation time. Butterfly wing colour patterns are attractive from a developmental perspective because, compared to many more complex morphological features, they are structurally simple and amenable to developmental characterisation. Wing patterns are created by spatially ordered arrangements of monochromatic scale cells in a single epithelial cell layer. Studying development is about understanding the way in which these scale cells gain the necessary information during growth to make the appropriate decisions that determine which one of a limited set of colour pigments will be synthesized by each cell just before adult eclosion from the pupa. An eyespot will only appear as a functional morphological unit if the many individual scale cells within each of its concentric rings of colour all contain the same pigment having all made the same decision in the specification of cell fate. These decisions are made in the growing wing disc at around the time of pupation, and sometimes many months before pigment synthesis and adult eclosion. Nijhout (1980) used surgical manipulations on the growing wing in the early pupa to show that an eyespot is formed around a central group of cells that acts as a developmental organizer. He postulated a model for eyespot formation based on this organizer, or eyespot focus, establishing a cone-shaped information gradient in the surrounding cell layer, perhaps through diffusion of a morphogen (Nijhout 1980, 1991). Following focal signalling in early pupae the surrounding cells interpret the gradient and, depending on the concentration of the morphogen, become fated to synthesize a particular colour pigment. The more recent results remain broadly consistent with this type of model although the morphogen itself remains elusive (see reviews: Beldade and Brakefield, 2002; McMillan et al., 2002). Application of molecular tools, and in particular examination of gene expression patterns, has identified a number of genes involved in eyespot formation both at the early stages of focal determination and later, at the end of focal signalling (Brunetti et al., 2001). A recent association study has also shown that one of these genes, Distal-less, contributes to inter-individual variation in eyespot size in B. anynana (Beldade et al., 2002b). Surveys of wing patterns across many species of butterflies inspired the concept of the Nymphalid groundplan 285 P. M. Brakefield (Nijhout, 1991). This is composed of different types of pattern elements: chevrons and bands as well as eyespots. The eyespots are present as a repeated series, with one element along the midline of each marginal wing cell (a wing cell is a compartment bounded by veins or trachea). Together, the eyespots can be viewed as a single module that has some independence from other modules comprised of band or chevron elements. We have found genetic variation in a number of the features of the eyespots in B. anynana, namely in their size, colour composition, shape and position (Brakefield, 1998). Each of these traits responds to artificial selection targetted at a single eyespot; only eyespot shape shows a low heritability with slow and comparatively subtle changes in phenotypes (Monterio et al., 1997a). Eyespots and exploring constraints Why use butterfly eyespots to explore the potential of generative constraints in the evolution of wing patterns? The answer lies in the notion of a nymphalid groundplan in combination with several developmental and genetical properties of groups of eyespots. The groundplan shows a series of marginal eyespots each of which is more or less of similar size and colour pattern. How do differences evolve across repeated eyespots, or in other words, how can the different eyespots within specific wing cells gain individuality during the course of evolution and thus diverge in phenotype? A very useful framework for examining how such patterns may evolve is that of modularity. The potential to produce a simple, undifferentiated pattern element or spot along the midline of each of the marginal wing cells is considered the evolutionary origin of butterfly eyespots (see Brunetti et al., 2001). Further evolution involving genetical and developmental elaboration of eyespot formation has led to the more complex compositions of individual eyespots with rings or patterns of different colours as found in many extant species. An initial, tightly coupled or integrated module may thus have changed over evolutionary time with respect to the complexity of the underlying genetical and developmental organisation, perhaps to the extent of interacting levels of modularity or different sub-modules (Beldade et al., 2002c; Beldade and Brakefield, 2003b; Monteiro et al., 2003). Such questions concerning the different individual eyespots are comparable to considering how different teeth on a jaw bone, or segments along a body, may become differentiated (Beldade et al., 2002c; Beldade and Brakefield, 2003b). In the case of butterfly eyespots, we know that all eyespots on a wing surface share a common sequence regarding the expression of developmental genes during their formation (Brake286 field et al., 1996; Brunetti et al., 2001). Furthermore, nearly all single gene mutants established in laboratory stocks influence all eyespots in a similar way (see Fig. 1). Finally, in artificial selection experiments targetted at either the size or the colour composition of a single eyespot, other eyespots, especially those on the same wing surface, respond to the selection in a highly concerted way (Monteiro et al., 1994, 1997b). These observations indicate both a genetical and a developmental coupling of the eyespots; they indeed form serial repeated units within a single module. These various properties of the eyespot pattern in B. anynana led to the prediction that directions of evolutionary change that involved uncoupling of individual eyespots might be more ‘difficult’ to achieve than others involving parallel changes, and thus, that the processes of pattern differentiation might have biased the evolutionary trajectories in this speciose genus (Brakefield, 1998). The forewings of B. anynana have two eyespots, a small anterior eyespot and a large posterior one. An artificial selection experiment (Fig. 2) was designed to test the predictions about evolutionary change in the pattern of eyespot changes, and more specifically to explore the tempo of evolutionary change in different directions of the potential morphological space regarding the pattern of relative eyespot size on the forewing. The results of the experiment have been published in detail elsewhere (Beldade et al., 2002a) so that here I Fig. 1. The ventral wing pattern of a gynandromorph individual of the butterfly Bicyclus anynana reared at 27 °C. The left side is male, and the right side, female. The left forewing is about 20 mm long. The female wings show the wild-type pattern while the male wings have larger marginal eyespots due to expression of the Bigeye allele. Note the pattern of fusion that occurs in some adjacent eyespots on the left hindwing due to Bigeye. Zoology 106 (2003) 4 Exploring evolutionary constraints Fig. 2. Design of the artificial selection experiment using Bicyclus anynana to explore the rate of response to selection in different directions of morphological space regarding the forewing pattern of eyespot size (from Brakefield, 1998). The centrally positioned forewing shows the wild-type pattern of this species with a small anterior eyespot and a large posterior eyespot. Selection attempted to change this pattern of eyespot sizes along both an axis of coupled change (i.e., both smaller or both larger) and of uncoupled change (one smaller and the other larger). Selection experiments targetted at the larger, posterior eyespot had already demonstrated positive genetic correlations and thus we predicted that movement along the ‘coupled’ axis (solid lines) with concerted changes in both eyespots would be rapid and unconstrained (√ ). In contrast, selection along the ‘uncoupled’ axis (dotted lines) was expected to be more limited (?). Fig. 3. Representative female forewings from the populations of Bicyclus anynana selected in the direction of each corner of morphological space for the two eyespots: anterior (A, a) and posterior (P, p) (see Fig. 2). The butterflies of each generation are arranged in the same pattern as in Fig. 2, without the wild-type pattern in the centre. Butterfly wings are shown from generations 11 and 25 (from Beldade et al., 2002a). Note the more extreme phenotypes of the later generation, and that highly divergent phenotypes were obtained along both the leading axis of concerted change in eyespot size (both smaller or both larger), and also along the uncoupled directions in which one eyespot became smaller and the other larger. will only summarize them. Figure 3 shows representative butterflies in the middle of the experiment (generation 11) and at the end of the experiment (generation 25) placed in the same arrangement regarding the directions of selection as shown in Fig. 2. They illustrate Zoology 106 (2003) 4 clearly that the standing genetic variation within our laboratory stock, in combination with a flexibility in the developmental mechanisms underlying specification of eyespot size, was sufficient to produce responses to selection that were effectively uncon287 P. M. Brakefield strained. This extreme flexibility in the response to artificial selection occurred in spite of the prior predictions about the likelihood of more limited responses in those directions involving the necessity of uncoupling the shared genetical and developmental properties of the individual eyespots. In terms of size, the anterior and posterior forewing eyespots did not behave as members of a tightly integrated single module. Final phenotypes after twenty-five generations of selection were widely divergent in all directions of selection whether involving parallel (‘coupled’) or opposing (‘uncoupled’) changes in the two eyespots. This was especially impressive in one of the uncoupled directions (small posterior, large anterior eyespot), since none of the eighty or so species of Bicyclus butterflies show this phenotype (Fig. 4). Thus, although this eyespot pattern might have been suspected to represent a ‘forbidden morphology’, the experimental result strongly suggests that its absence in nature is due to a lack of appropriate natural selection in any lineage. Perhaps this pattern has not proved advantageous in any natural environment either in terms of potential natural selection by visual predators (see Lyytinen et al., 2003a, b) or of sexual selection involving mate choice (Breuker and Brakefield, 2002). Fig. 4. Distribution of species of Bicyclus in the morphological space for the two forewing eyespots (see also Fig 2). Representative species names are given. Note the absence of any extant species that shows a pattern of a large posterior eyespot in combination with no anterior eyespot. This pattern was, however, obtained in B. anynana following 25 generations of artificial selection in that direction (see Fig. 3) (Fig. redrawn by Steve Thurston from Beldade et al., 2002a). 288 Discussion This artificial selection experiment shows that although eyespots share a common developmental mechanism, the relative size of the dorsal forewing eyespots can readily move through morphospace under appropriate selection regimes. This result does not mean that there are no generative constraints on evolutionary change in the eyespot patterns of B. anynana and other butterflies. Indeed, we have recently analysed the responses recorded in our selection experiment with regard to the full module of marginal eyespots on each wing, and on both dorsal and ventral wing surfaces (Beldade et al., 2002c; Beldade and Brakefield, 2003b). The response on the ventral surface of the forewing was straightforward: the pattern of change in the anterior and posterior eyespots in each direction of selection closely paralleled the response on the dorsal surface, although it was less extreme. For the ventral hindwing, the pattern of relative sizes within the full series of seven eyespots (one within each marginal wing cell) was, however, more illuminating (Fig. 5). The eyespots could be classified into two groups: the four more anterior eyespots tended to respond in parallel with the anterior forewing eyespot, while the three more posterior eyespots (especially numbers 6 and 7) changed in accordance with the posterior forewing eyespot. Thus, there may be two domains or subcompartments of the wing leading to some subdivision of the eyespot module into submodules (see also Monteiro et al., 2003). Perhaps the eyespots within each of these groups are more tightly coupled than are the two forewing eyespots (in wing cells 2 and 5) or the eyespots in different submodules of the hindwing. We will test this hypothesis and study the consequences of the organisation of the hindwing eyespots in a follow-up artificial selection experiment (see Fig. 5). Our selection experiment detected no important developmental constraints on the evolution of the pattern of relative sizes of the two forewing eyespots in B. anynana. This result demonstrated sufficient variation in genes that differentially modify the developmental process that underlies eyespot formation to give the anterior and posterior eyespots individuality in size within the overall wing pattern. We now believe that our selection experiment was actually based on a stock of a species from a lineage that had already experienced a long period of natural selection involving an accumulation of individuality in the specification of size of the anterior and posterior eyespots. Thus the notion of a groundplan with a module of tightly coupled, serial eyespots is probably more relevant to the genetic and developmental architecture nearer the time of the evolutionary origin of eyespots in basal Lepidoptera (Beldade et al., 2002c; Beldade and Brakefield, 2003b; Monteiro et al., 2003). Zoology 106 (2003) 4 Exploring evolutionary constraints Fig. 5. Analysis of the modularity of eyespots in the butterfly Bicyclus anynana. The specimen to the left shows the ventral forewing and hindwing patterns for the wild type (wet season phenotype) with small anterior (A) and large posterior (P) forewing eyespots, and a series of seven hindwing eyespots, one in each wing cell (1–7) that show a characteristic pattern of relative sizes. Analysis of butterflies from the selection experiment of Beldade et al. (2002a) as illustrated in Figs. 2 and 3 showed that hindwing eyespots1–4 responded in concert with the direction of selection of the anterior eyespot of the forewing (A), while 5–7 responded in line with the posterior eyespot (P) (Beldade et al., 2002c; Beldade and Brakefield, 2003). Thus they tended to shift in two blocks, and in opposing directions if selection for size on the two dorsal forewing eyespots was in opposite directions. This suggests a substructure of two domains or submodules within the hindwing (note that these do not reflect the anterior-posterior compartments known from the Drosophila wing). The righthand side of the diagram illustrates a protocol to be used in a follow-up selection experiment on pairs of hindwing eyespots. It is designed to test the consequences of such a substructuring of modularity and to compare patterns of responses to selection regarding two different features of the eyespots, namely their size and colour composition. Different predictions can be made depending on the developmental basis of variation of each feature and whether the pairs of targetted eyespots are within the same submodule or not (for further information see the text). A future experiment (see Fig. 5) will explore whether the same degree of flexibility as we have found for the forewing eyespots is also evident within each of the apparent sub-compartments on the hindwing, and also regarding a different feature of eyespots, their colour composition (i.e. the relative dimensions of the different concentric colour rings). The latter trait appears to show less variation among eyespots within a species than does size; all eyespots on a wing tend to show a similar colour composition, although the overall pattern can differ across species (see Condamin, 1973). This might have an explanation in terms of natural selection but could also reflect the existence of stronger constraints on colour composition than on size. The latter hypothesis may be consistent with the results from an earlier comparison (Monteiro et al., 1994, 1997b) of the responses of the posterior forewing eyespot to artificial selection regarding size or colour composition. In terms of realized heritability and rate of responses these two features of the eyespot behaved in a closely similar manner. However, the underlying development was found to be very different. While divergence in eyespot size was largely accounted for by the strength of the focal signal, differences in colour were explained solely through the inZoology 106 (2003) 4 terpretation of the focal signal by surrounding cells, and thus presumably by variation in threshold responses (see Beldade and Brakefield, 2002). The latter may tend to be a more or less uniform property across the whole distal portion of the wing, while the former may be more tightly coupled to establishment of the individual focal signals in each marginal wing cell. This sort of reasoning underlies the qualitative predictions regarding the responses in a new selection experiment targetted at both these features in the hindwing eyespots (see Fig. 5). We believe that our experiments to date already illustrate the potential power of an evolutionary genetics approach in combination with ‘evo-devo’ insights about developmental processes to explore the likelihood of generative constraints, and to understand the underlying mechanisms (see also Brakefield et al., 2003). A new selection experiment should reveal more about just how successful this approach is likely to be in helping to unravel the role of constraints in shaping patterns of evolution. A more balanced view of Darwinian evolution will result from a better understanding of how the processes of development and physiology that generate variation in the phenotype influence the patterns of evolution. 289 P. M. Brakefield References Antonovics, J., and P.H. van Tienderen. 1991. Ontoecogenophyloconstraints? The chaos of constraint terminology. Trends Ecol. Evol. 6: 166 –168. Beldade, P., and P.M. Brakefield. 2002. The genetics and evodevo of butterfly wing patterns. Nat. Rev. Genet. 3: 442–452. Beldade, P., K. Koops and P.M. Brakefield. 2002a. Developmental constraints versus flexibility in morphological evolution. Nature 416: 844 –847. Beldade, P., P.M. Brakefield and A.D. Long. 2002b. Contribution of Distal-less to quantitative variation in butterfly eyespots. Nature 415: 315 –318. Beldade, P., K. Koops and P.M. Brakefield. 2002c. Modularity, individuality, and evo-devo in butterfly wings. Proc. Natl. Acad. Sci. USA 99: 14262–14267. Beldade, P. and P.M. Brakefield. 2003a. The difficulty of agreeing about constraints. Evol. Dev. 5: 119–120. Beldade, P. and P.M. Brakefield. 2003b. Concerted evolution and developmental integration in modular butterfly wing patterns. Evol. Dev. 5: 169–179. Brakefield, P.M. 1998. The evolution-development interface and advances with the eyespot patterns of Bicyclus butterflies. Heredity 80: 265–272. Brakefield, P.M. 2003. Artificial selection and the development of ecologically relevant phenotypes. Ecology 84: 1661–1671. Brakefield, P.M. and V. French. 1999. Butterfly wings: the evolution of development of colour patterns. BioEssays 21: 391–401. Brakefield, P.M., V. French and B.J. Zwaan. 2003. Development and the genetics of evolutionary change within species. Annu. Rev. Ecol. Evol. Syst. 34: 633– 660. Brakefield, P.M., J. Gates, D. Keys, F. Kesbeke, P.J. Wijngaarden, A. Monteiro, V. French and S.B. Carroll. 1996. Development, plasticity and evolution of butterfly eyespot patterns. Nature 384: 236 –242. Breuker, C.J. and P.M. Brakefield. 2002. Female choice depends on size but not symmetry of dorsal eyespots in the butterfly Bicyclus anynana. Proc. R. Soc. Lond. B 269: 1233–1239. Brunetti, C.R., J.E. Selegue, A. Monteiro, V. French, P.M. Brakefield and S.B. Carroll. 2001. The generation and diversification of butterfly eyespot color patterns. Curr. Biol. 11: 1578 –1585. Condamin, M. 1973. Monographie du Genre Bicyclus (Lepidoptera, Satyridae). Memoires de l’Institut Fondamental d’Afrique Noire 88. Ifan-Dakar. Eldredge, N., J.N. Thompson, P.M. Brakefield, S. Gavrilets, D. Jablonski, J.B.C. Jackson, R.E. Lenski, B.S. Lieberman, M.A. McPeek and W. Miller III. 2003. The dynamics of evolutionary stasis. Paleobiology: in press. Fryer, G. and T.D. Iles. 1972. The Cichlid Fishes of the Great Lakes of Africa: Their Biology and Evolution. Oliver and Boyd, London. 290 Gould, S.J. 2002. The Structure of Evolutionary Theory. Harvard University Press, Cambridge, Massachusetts. Gould, S.J. and R.C. Lewontin. 1979. The spandrels of San Marco and the Panglossian paradigm: a critique of the adaptationist programme. Proc. R. Soc. Lond. B. 205: 581–598. Grant, R. 2003. Evolution in Darwin’s finches: a review of a study on Isla Daphne Major in the Galápagos Archipelago. Zoology 106: Jackson, J.B.C. and A.H. Cheetham. 1999. Tempo and mode of speciation in the sea. TrendsEcol. Evol. 14: 72–77. Lyytinen, A., P.M. Brakefield and J. Mappes. 2003. Significance of butterfly eyespots as an anti-predator device in groundbased and aerial attacks. Oikos 100: 373–379. Lyytinen, A., P.M. Brakefield, L. Lindström and J. Mappes. 2003. Does predation maintain eyespot plasticity in Bicyclus anynana? Proc. R. Soc. Lond. B.: in press. Maynard Smith, J., R. Burian, S. Kaufman, P. Alberch, J. Campbell, B. Goodwin, R. Lande, D. Raup and L. Wolpert. 1985. Developmental constraints and evolution. Quart. Rev. Biol. 60: 265–287. McMillan, W.O., A. Monteiro and D.D. Kapan. 2002. Development and evolution on the wing. Trends Ecol. Evol. 17: 125–133. Monteiro, A.F., P.M. Brakefield and V. French. 1994. The evolutionary genetics and developmental basis of wing pattern variation in the butterfly Bicyclus anynana. Evolution 48: 1147–1157. Monteiro, A., P.M. Brakefield and V. French. 1997a. The genetics and development of an eyespot pattern in the butterfly Bicyclus anynana: response to selection for eyespot shape. Genetics 146: 287–294. Monteiro. A., P.M. Brakefield and V. French. 1997b. Butterfly eyespots: the genetics and development of the color rings. Evolution 51: 1207–1216. Monteiro, A., J. Prijs, M. Bax, T. Hakkart and P.M. Brakefield. 2003. Mutants highlight the modular control of butterfly eyespot patterns. Evol. Dev. 5: 180–187. Nijhout, H.F. 1980. Pattern formation on lepidopteran wings: determination of an eyespot. Developmental Biology 80: 267–274. Nijhout, H.F. 1991. The Development and Evolution of Butterfly Wing Patterns. Smithsonian Institute Press, Washington. Schlichting, C.D. and M. Pigliucci. 1998. Phenotypic Evolution, a Reaction Norm Perspective. Sinauer, Sunderland, Massachusetts. Schluter, D. 1996. Adaptive radiation along genetic lines of least resistance. Evolution 50: 1766–1774. Schwenk, K. 1995. A utilitarian approach to evolutionary constraints. Zoology98: 251–262. Stearns, S. C. 2002. Less would have been more, Evolution 56: 2339–2345. Temeles, E.J. and W.J. Kress. 2003. Adaptation in a plant-hummingbird association. Science 300: 630–633. Zoology 106 (2003) 4