EVOLUTION & DEVELOPMENT
4:1, 1–6 (2002)
Evolutionary innovations overcome ancestral constraints: a re-examination of character evolution in male sepsid flies (Diptera: Sepsidae) Günter P. Wagnera,* and Gerd B. Müllerb a
Yale University, New Haven, CT 06520-8106, USA; and bUniversity of Vienna, A-1090 Vienna, Austria
*Correspondence (email:
[email protected])
In a recent article, Eberhard discussed whether the origin of morphological innovations is rare because of developmental constraints or whether their appearance is primarily driven by natural selection, and he argued in favor of the latter (Eberhard 2001). Eberhard’s reasoning is based on a study that traces the evolutionary origination of a major morphological innovation in a group of flies (Diptera: Sepsidae) using Meier’s (1995) phylogenetic hypothesis. The innovation consists of movable abdominal lobes in the males of this group, a character that is extremely rare in other flies. Eberhard also presented evidence that the appendages are used to stimulate the female during copulation. According to his preliminary character state reconstruction, this character originated two or three times in the clade of 240 species of flies. Eberhard concluded that abdominal lobes must evolve easily because there are multiple originations of this character within the sepsids. Thus, the ancestral lineage is inferred to be not constrained to produce the derived character. At issue in this study is a suggestion that we proposed in a 1991 review of the novelty concept (Müller and Wagner 1991). From a survey of examples from vertebrate evolution, we concluded that the lineage ancestral to a major innovation is developmentally constrained to acquire the derived character and that the rareness of innovations is explained by the need to overcome this ancestral constraint. If this is true, the study of innovations would have a different research agenda than the study of adaptation. This would justify recognition of innovation as a concept distinct from adaptation. A major research objective in the study of innovations would be to identify the developmental apomorphies that allowed the derived lineage to overcome the ancestral constraint (Prum 1999). Eberhard, however, concluded that his data refuted our suggestion and by implication would render the concept of innovations superfluous. As an alternative, he proposed that sexual selection is sufficient to explain the origin of this © BLACKWELL SCIENCE, INC.
character in sepsids and the almost uniform absence of this character in other dipterans. A close examination of the facts presented in the article led us to conclude that the data do not support his conclusion. In contrast, we think that the data actually strongly support our suggestion that natural selection alone is not sufficient to explain the origination of morphological novelties. Later here, we outline the argument supporting this conclusion.
CHARACTER EVOLUTION IN MALE SEPSID FLIES Eberhard reports four character states for the fourth abdominal sternite of male sepsids. Character state “0” represents a case in which no sexual dimorphism exists in this character. This character state is found in the most basal lineage of sepsids, Orygma luctuosum, and in one derived species of uncertain phylogenetic position, Sepsis tuberculata. In the latter species, however, the tergite is sexually dimorphic. All of the other character states of the fourth sternite are sexually dimorphic. Character state “A” represents cases in which the male has more and longer bristles on the fourth abdominal sternite than the female but no further modifications. The third character state “B” is one in which the fourth abdominal sternite has acquired, in addition to the long bristles of character state A, a modified shape but still consists of only one sclerite. In the most derived character state “C,” the sternite consists of several individualized skeletal elements, some of which are independently movable. According to our reading, it is this character state that Eberhard identified as a novelty in the sense of “a pronounced and unusual departure in morphology” (Arnold et al. 1989, p. 410) or as a new homologue (Müller and Wagner 1991). From Eberhard (2001), we mapped, using MacClade 4.0 (Maddison and Maddison 2000) (see the Methods section for details), the character states of those species that are included 1
2
EVOLUTION & DEVELOPMENT Vol. 4, No. 1, January-February 2002
in the phylogenetic analysis of Meier (1995). For these taxa, the “preferred tree” of Meier (1995, Figures 76 and 77) contains several polytomies, “Tree 1” (Fig. 1; nine steps, CI 0.33). The unresolved relationships among the included Sepsis species are irrelevant because they all represent the same character state “A” and are thus represented as a unitary taxon “Sepsis” in our analysis. Assuming ordered character state transformations, the analysis suggests at least two independent derivations of the novelty, character state C, in accordance with Eberhard (2001). However, ancestral states could not be reconstructed in many places because of the limited resolution of the tree. Therefore, we created a resolved tree by excluding two Themira species with a highly uncertain phylogenetic position (T. minor and T. putris) based on the alternative, most parsimonious trees in Figure 78 of Meier (1995). In addition, we represent the three Nemopoda species by only one species, as they all have the same
character state. The polytomy between two Themira clades (Methods section) is arbitrarily resolved because the alternative resolutions do no affect the character state reconstructions (results not shown). These modifications lead to a completely resolved tree, “Tree 2” (Figs. 2 and 3), which allows the application of ordered as well as irreversible character state transformation schemes. Character state reconstruction in Tree 2, assuming irreversible evolution, suggests that the novelty C evolved three times independently among the species included in this analysis (Fig. 2; 10 steps, CI 0.30). These originations occur in three distantly related lineages, within a clade that contains Themira leachi and T. superba, in the Nemopoda clade, and in the Meroplius minutus lineage. Most of the ancestral lineages, however, are reconstructed to have character state A, that is, only modified bristles in the males. This character state is reconstructed to be ancestral for all lineages except
Fig. 1. Tree 1 contains most of the species that overlap between the data sets of Eberhard (2001) and Meier (1995). The character states are reconstructed using an ordered character transformation scheme. Note that there are at least two independent originations of movable abdominal lobes, that is, character state C.
Wagner and Müller
Evolutionary innovations overcome ancestral constraints
3
Fig. 2. Tree 2 has been obtained from Tree 1 by pruning species with uncertain phylogenetic position and by arbitrarily resolving polytomies that do not affect character reconstruction (see the Methods section for details). The character evolution reconstruction in this figure assumes that the evolution of the character is irreversible (10 steps, CI 0.3). Under this assumption, it is inferred that the novelty evolved three times. The ancestral lineage is reconstructed to have character state A, which is sexually dimorphic. From this, we conclude that sexual selection for modified male abdominal morphology was acting in all sepsid lineages, at least since the most recent common ancestor of Paradoxopoda and the rest of the species in our sample.
the most basal one, Orygma luctuosum. From this, we conclude that sexual selection for a modified male abdominal morphology has been acting for most of sepsid evolution. Similar results are obtained if we assume ordered but reversible character transformations (Fig. 3; eight steps, CI 0.38). The only differences are that only two derivations of the novelty C can be inferred unambiguously, and most of the ancestral states are reconstructed as B rather than A. Un-
der this assumption, character state B would have originated only once, above the node of Paradoxopoda and the rest of the sexually dimorphic sepsids. The reconstruction implies at least two reversals to character state A. Whether this is functionally likely given the distribution of sexual selection in this group needs to be determined by experimental work. The ancestral state for Nemopoda and Meropilus is either C or B. However, the basic conclusion that sexual selection
4
EVOLUTION & DEVELOPMENT Vol. 4, No. 1, January-February 2002
Fig. 3. Tree 2 is as in Figure 2, but assuming an ordered but reversible character transformation scheme (eight steps, CI 0.38). Character state C is reconstructed to have originated two or three times. Character state B only evolved once with two reversions to character state A. Whether a reversion of character state B to A is functionally likely cannot be decided from these data.
was acting for most of sepsid phylogeny remains intact. Nevertheless, many lineages linger at character states A or B, with only two or three independent transitions to the novel character state C. Based on this phylogenetic scenario, we conclude that sexual selection for modified male abdominal morphology was acting in all sepsid lineages except the most basal one. We think that this conclusion is inevitable given the current phylogenetic information. Given the prevalence of sexual selection in this clade, it is indeed surprising that the innova-
tion C only evolved twice (or three times). Many species in this clade remain in the less derived but still sexually dimorphic character states A or B. From this comparative evidence, one has to infer that a constraint is preventing the evolution of character state C in all of these species. Of course, we cannot be sure whether the constraint is developmental or functional. However, the data presented certainly cannot refute the hypothesis that developmental constraints may be acting in the species with character states A and B, preventing the ready realization of character state C.
Wagner and Müller
CONCLUSIONS We agree with Eberhard that a careful analysis of character state variation and evolution is essential for a deeper understanding of evolutionary novelties. His work provides a valuable service to Evo-Devo research by making the facts of sepsid morphology accessible to nonspecialists. However, we respectfully disagree with the evolutionary interpretation of the data that he presented. In fact, we think that this example is particularly important for the theory of evolutionary innovations because it is rarely possible to reconstruct the selection pressures acting in the ancestral lineage. The example shows that selection acted in favor of the derived character long before the innovation itself occurred in this clade. Thus, we think that this data set represents a strong case for the presence of a constraint preventing the novelty in most of the lineages in this clade. Eberhard successfully demonstrated that one factor that explains the absence of movable abdominal lobes in most flies other than sepsids is the absence of sexual selection for this male copulatory function. From this, it is possible to conclude that sexual selection is a necessary factor in the origin of the abdominal lobes in flies. At the same time, the analysis of character variation within the sepsid demonstrates that sexual selection is not sufficient to explain the origin of this morphological innovation in flies. Some factor seems to prevent the realization of this character for most of sepsid phylogeny. One possibility is the presence of developmental constraints that can be overcome by rare mutational events (Carroll et al. 2001), pleiotropic effects caused by selection or other characters (Müller 1990), developmental threshold effects (Müller 1990; Raff 1999), or neutral phenotypic drift (Mayr 1960). The reason that Eberhard reached a different conclusion from ours could be that he has a different understanding of what constitutes a developmental constraint in evolution. Eberhard’s introduction documents that he sees constraints as a prevention of alternative designs to arise, inferred from phylogenetic uniformity: “Whatever has not changed must be constrained.” The definition of constraint on which our argument was based differs substantially. We followed Maynard-Smith et al. (1985), who define developmental constraint as “a limitation or 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.” Hence, a developmental constraint is not merely defined as a lack of evolutionary change, but rather by what is likely to arise by natural variation (Schwenk 1995). The effect of developmental constraints on phenotypic evolution depends on the circumstances. Of course, developmental constraints can inhibit certain phenotypic transformations by limiting the amount of heritable phenotypic variation. However, constraints can also enhance the proba-
Evolutionary innovations overcome ancestral constraints
5
bility of other evolutionary transformations by increasing the probability of a particular outcome through eliminating alternatives (Gould 1989; Müller and Newman 1999; Wagner 1988; Wake 1991). Although a constraint needs to be overcome for an innovation to arise, it does not imply that something is developmentally difficult to realize under the right circumstances (Blanco et al. 1998). It implies only that genetic variation for the derived character is not likely to segregate in natural populations and that selection is therefore not a sufficient cause for the evolutionary realization of the trait. In the case of the sepsid flies, the hypothesis of a developmental constraint means that a population with the character states A or B does not contain a heritable phenotypic variation that would allow the ready realization of the B→C transition by natural selection. Does this imply that this transition is impossible? Of course not. However, the rate at which this transition occurs and the specific form it takes are not determined by selection alone but by the occurrence of developmental changes that are necessary to overcome the ancestral constraint(s). In this example, we have no information as to the nature of the constraint and how it was overcome. In other examples, however, we do have a better idea about the developmental changes that open up new dimensions of character space, such as in the case of feather development and evolution (Prum 1999), the fibular crest of birds and theropod dinosaurs (Müller and Streicher 1989), or the eyespot patterns on butterfly wings (Keys et al. 1999). With sepsid flies, the next and most important step would be to obtain a better resolved and more complete phylogeny of this group to allow a detailed reconstruction of character evolution.
METHODS Character evolution is reconstructed from the set of species that overlap between the data sets of Meier (1995) and Eberhard (2001) with a few exceptions. Dicranosepsis transita (Ozerov) of Eberhard (2001) was placed into the phylogenetic position of D. bicolor. All of the species of Sepsis in Eberhard (2001) that are included in Meier (1995) have the same character state A and the same relative phylogenetic position and are represented in the analysis by a single inclusive taxon Sepsis. We fixed the character state of the root at 0. Tree 1 (Fig. 1) consists of all previously mentioned species placed on the phylogenetic tree represented in Figures 76 and 77 of Meier (1995). This tree contains several unresolved nodes, particularly among the species of the genus Themira and Nemopoda. Therefore, this tree is of limited value for the reconstruction of character state evolution. In Tree 2 (Fig. 2), to allow for the application of more character reconstruction methods, we obtained a resolved tree by eliminating certain species from Tree 1. Inspection of the equally parsimonious trees of Themira species in Figure 78 of Meier (1995) reveals that two of the species included in our analysis are particularly ambigu-
6
EVOLUTION & DEVELOPMENT Vol. 4, No. 1, January-February 2002
ous in their phylogenetic affiliation (T. minor and T. putris). We eliminated these species from the tree and arbitrarily resolved the polytomy at the base of the clades T. leachi, T. superba, and T. lucida and T. nigricornis and T. lutulenta. The three different arrangements of these two clades lead to the same conclusion about ancestral character reconstruction. The three species of Nemopoda in this data set all have the same character state and are thus represented by one of them in this tree. Based on the morphological data that Eberhard (2001) presented, we use two character state options. Ordered means that characters are transformed only according to the scheme O↔A↔ B↔C. The irreversible transformation scheme assumes O→A→ B→C. Acknowledgments The authors thank Drs. Chi-hua Chiu, Joachim Hermisson, Hans Larsson, and Terri Williams for reading a previous version of this manuscripts and for helpful comments. We gratefully acknowledge NSF’s financial support for the research in the GPW lab.
REFERENCES Arnold, S. J., Alberch, P., Csányi, V., Dawkins, R. C., and Emerson, S. B., Fritzsch, B., Horder, T. J., Maynard-Smith, J., Starck, M. J., Vrba, E. S., Wagner, G. P., and Wake, D. B. 1989. How do complex organisms evolve? In D. B. Wake and G. Roth (eds.). Complex Organismal Functions: Integration and Evolution in Vertebrates. John Wiley & Sons, New York, pp. 403–433. Blanco, M. J., Misof, B. Y., and Wagner, G. P. 1998. Heterochronic differences of Hoxa-11 expression in Xenopus fore- and hind limb development: evidence for a lower limb identity of the anuran ankle bones. Dev. Genes Evol. 208: 175–187. Carroll, S. B., Grenier, J. K., and Weatherbee, S. D. 2001. From DNA to Diversity. Blackwell Science, Malden, MA. Eberhard, W. G. 2001. Multiple origins of a major novelty: moveable abdominal lobes in male sepsid flied (Diptera: [S]epsidae), and the question of developmental constraints. Evol. Dev. 3: 206–222.
Gould, S. J. 1989. A developmental constraint in Cerion, with comments on the definition and interpretation of constraint in evolution. Evolution 43: 516–539. Keys, D. N., Lewis, D. L., Selegue, J. E., Pearson, B. J., and Goodrich, L. V., Johnson, R. L., Gates, J., Scott, M. P., and Carroll, S. B. 1999. Recruitment of a hedgehog regulatory circuit in butterfly eyespot evolution. Science 283: 532–534. Maddison, W. P., and Maddison, D. R. 2000. MacClade Version 4: Analysis of Phylogeny and Character Evolution. Sinauer, Sunderland, MA. Maynard-Smith, J., Burian, R., Kauffman, S., Alberch, P., and Campbell, J., Goodwin, B., Lande, R., Raup, D., and Wolpert, L. 1985. Developmental constraints and evolution. Q. Rev. Biol. 60: 265–287. Mayr, E. 1960. The emergence of evolutionary novelties. In S. Tax (ed.). Evolution after Darwin. Harvard University Press, Cambridge, MA, pp. 349–380. Meier, R. 1995. Cladistic analysis of the Sepsidae (Cyclorrhapha: Diptera) based on a comparative scanning electron microscopic study of larvae. Syst. Entomol. 20: 99–128. Müller, G. B. 1990. Developmental mechanisms at the origin on morphological novelty: a side-effect hypothesis. In M. H. Nitecki (ed.). Evolutionary Innovations. The University of Chicago Press, Chicago, pp. 99– 130. Müller, G. B., and Newman, S. A. 1999. Generation, integration, autonomy: three steps in the evolution of homology. In G. R. Bock and G. Cardew (eds.). Homology. John Wiley & Sons, New York, pp. 65–79. Müller, G. B., and Streicher, J. 1989. Ontogeny of the syndesmosis tibiofibularis and the evolution of the bird hindlimb: a caenogenetic feature triggers phenotypc novelty. Anat. Embryol. 179: 327–339. Müller, G. B., and Wagner, G. P. 1991. Novelty in evolution: restructuring the concept. Ann. Rev. Ecol. Syst. 22: 229–256. Prum, R. O. 1999. Development and evolutionary origin of feathers. J. Exp. Zool. (Mol. Dev. Evol.) 285: 291–306. Raff, R. A. 1999. Larval homologies and radical evolutionary changes in early development. In G. R. Bock and G. Cardew (eds.). Homology. Wiley & Sons, Chichester, MA, pp. 110–121. Schwenk, K. 1995. A utilitarian approach to evolutionary constraint. Zoology 98: 251–262. Wagner, G. P. 1988. The influence of variation and of developmental constraints on the rate of multivariate phenotypic evolution. J. Evol. Biol. 1: 45–66. Wake, D. B. 1991. Homoplasy: the result of natural selection, or evidence of design limitations? Am. Nat. 138: 543–567.
