3; Tables 3, S2). Am. Table S1. The role of natural selection in the repeated evolution of ecomorph‐specific traits, which is supported by a large body of evidence (Losos 2009), suggests that phenotypic divergence in anoles is unlikely to have been limited by genetic architecture. On each island, very similar patterns of … Parameters from our animal models were used to quantify species divergence in morphology. The major axis of morphological divergence (d1, explaining 73.5% of divergence) separated species with long limbs and wide heads from those with shorter limbs and narrow heads (Fig. This alignment echoes results from a recent study of a vastly different group of traits in flies (Houle et al. The relationship between sexual size dimorphism and habitat use in Greater Antillean Anolis lizards. drafted the manuscript and all authors contributed to the final version of the manuscript. Starting with a single ancestor, this process results in the speciation and phenotypic … Particular ecomorphs may evolve repeatedly, that is, convergently, as separate lineages arrive at similar solutions. Although these plots were not used for any formal analyses, they facilitate visual comparison of G‐matrix size, shape, and orientation (see Figs. This process may be driven both by directional selection (i.e., movement of a population toward a new fitness peak on the adaptive landscape) and multivariate stabilizing selection (i.e., selection that stabilizes a population's occupancy of its current fitness peak) (Lande 1980; Cheverud 1982; Jones et al. It is surprising, then, that such a pervasive source of biological diversity has not been integrated into studies of adaptive radiation, despite extensive and growing attention to both phenomena1,3,4,5,6,7. Traits were measured from X‐ray images of juveniles taken at four points during development, and G matrices were estimated ASReml (Gilmour et al. Syst. 60, 369–388 (1990), Selander, R. K. in Sexual Selection and the Descent of Man, 1871–1971 (ed. and J.J.K. Thank you for visiting nature.com. 2003; Arnold et al. VSN International Ltd, Out of Cuba: overwater dispersal and speciation among lizards in the, Predicting microevolutionary responses to directional selection on heritable variation, Measuring and comparing evolvability and constraint in multivariate characters, Characterizing the evolution of genetic variance using genetic covariance tensors, Mutation predicts 40 million years of fly wing evolution, A comparative analysis of the ecological significance of maximal locomotor performance in Caribbean, The mutation matrix and the evolution of evolvability, Epistasis and natural selection shape the mutational architecture of complex traits, The effects of stochastic and episodic movement of the optimum on the evolution of the, Genetic variation increases during biological invasion by a Cuban lizard, Between‐groups comparison of principal components, Natural selection and random genetic drift in phenotypic evolution, Quantitative genetic analysis of multivariate evolution, applied to brain:body size allometry, The genetic covariance between characters maintained by pleiotropic mutations, Pleiotropy of quantitative trait loci for organ weights and limb bone lengths in mice, Ecomorphology, performance capability, and scaling of West Indian, A phylogenetic analysis of character displacement in Caribbean, Lizards in an evolutionary tree: Ecology and adaptive radiation of anoles, The effect of perch diameter on escape behaviour of, Adaptation and constraint in the evolution of specialization of Bahamian, Contingency and determinism in replicated adaptive radiations of island lizards, Exceptional convergence on the macroevolutionary landscape in island lizard radiations, Studying phenotypic evolution using multivariate quantitative genetics, Phenotypic divergence along lines of genetic variance, EvolQG–an R package for evolutionary quantitative genetics, APE: analyses of phylogenetics and evolution in R language, Constraints evolve: context dependency of gene effects allows evolution of pleiotropy, A phylogenetic, biogeographic, and taxonomic study of all extant species of, Integration of the transcriptional networks regulating limb morphogenesis, A test of the hypothesis that correlational selection generates genetic correlations, Adaptive radiation along genetic lines of least resistance, Comparative quantitative genetics: evolution of the, Rapid evolution of a native species following invasion by a congener. 2004, 2012). Although the pattern demonstrated here is clear, the mechanisms underlying it are not. 2009; Aguirre et al. A software package written in the R statistical language for kernal smoothing, discriminant function analysis, and associated 2- and 3D plotting. Finally, to assess the degree to which the eigenvectors of H were similar to those of G for each species, we calculated the angle between each h vector and the corresponding eigenvector of G (i.e., h1 vs gmax, h2 vs g2, etc.). Analogously, to ask how change in G for a particular trait combination scaled with available genetic variation, we performed a similar regression of log10‐transformed among‐species divergence in genetic variance (log DG) on log G. For the latter analysis, we used species‐specific genetic variances in the eigenvectors of Ganc to calculate log E and the eigenvalues of Ganc to calculate log G. Of the three metrics we used to characterize G matrices, size (total genetic variance) varied the most across species (coefficient of variation = 27%), followed by orientation (angle of gmax, 21%), and shape (% variance explained by gmax, 7.5%; Fig. If a particular h vector is aligned with all G matrices, this would further indicate that it describes an axis of genetic variance that is conserved across species. 2013; Careau et al. The observed changes in G across species may reflect a deeper constraint on G imposed by the rate and phenotypic effect of mutations. 2014; but see Houle et al. We constructed each random vector by drawing its eight elements from a uniform distribution bounded by –1 and 1 and then standardizing the vector to unit length. In a common laboratory environment, we estimated G matrices for seven species of West Indian Anolis lizards representing three different ecomorphs (trunk‐crown, trunk‐ground, and grass‐bush) that originated independently on three different islands of the Greater Antilles: A. cristatellus (trunk‐ground), A. evermanni (trunk‐crown), and A. pulchellus (grass‐bush) from Puerto Rico; A. grahami (trunk‐crown) and A. lineatopus (trunk‐ground) from Jamaica; and A. sagrei (trunk‐ground) and A. smaragdinus (trunk‐crown) from South Bimini, the Bahamas (Table 1). 2013). As we discuss below, our results contain a signature of G‐matrix evolution consistent with this scenario, suggesting that selection is a more plausible source of the alignment of G and D than constraint. 1994; Stuart and Losos 2013; Stuart et al. We use animal models to estimate both species‐specific genetic architecture (G matrices) for a suite of skeletal traits and the direction and magnitude of evolutionary divergence among Anolis species in size‐corrected morphological space. Press, Oxford, 2000), Gillespie, R. Community assembly through adaptive radiation in Hawaiian spiders. The percent variance explained by gmax (the axis of greatest additive genetic variance) was used as an index of G matrix shape, which indicates a population's potential to respond to selection aligned with gmax relative to other directions. Size‐corrected species means and divergence matrix. Most tests of evolution along genetic lines of least resistance follow Schluter (1996), comparing divergence of species means to a single estimate of G. We used a different approach that allows us to capture information from all of estimates of G within the radiation. To ask whether G itself evolved during the adaptive radiation, we analyze covariance tensors and find that most differentiation of genetic architecture occurs in subspaces that include limb traits. Nat. A second axis (h2) explained between 13 and 30% of genetic variance within species (Tables 3, S4). Stanley A Sawyer 7 (eds Gans, C. & Tinkle, D. W.) 35–136 (Academic, New York, 1977), Slatkin, M. Ecological causes of sexual dimorphism. The classification of Anolis: conflict between genetic and osteological interpretation as exemplified by Anolis cybotes. 2013). III and DEB 0519777 and 0722475 to J.B.L. This approach allowed us to determine how species had diverged in shape while controlling for species differences in overall size. We find evidence for island (i.e., opportunity) effects and no evidence for trait (i.e., key innovation) effects causing accelerated body size evolution within Anolis. We thank Simon Pearish and Michelle Sivilich for managing the lizard colony. The authors declare no competing financial interests. We also calculated the amount of species‐specific genetic variance explained by each h vector by projecting it through each G matrix using the equation hTGh, where T denotes transposition (Aguirre et al. Radiation of α vs. β traits: evolutionary sequences in speciation vs. radiation Ceanothus in California (Ackerly et al.) Then we calculated the average of the rescaled Ganc and D and calculated the eigenvectors of the resultant matrix, providing a set of orthogonal vectors representing a subspace common to Ganc and D. Next, these eigenvectors were projected through both of the original matrices to determine the amount of within‐species genetic variance and among‐species variance, respectively, for each trait combination. Third, we show that this pattern likely persisted because the evolution of G was proportional to both within‐species genetic variance and divergence in species means. Phenotypic evolution, constant covariances, and the maintenance of additive variance, Abundant genetic variation plus strong selection = multivariate genetic constraints: a geometric view of adaptation, Evolution of genetic variance during adaptive radiation, The origin of faunas. 2010; Bolstad et al. Sexual dimorphism is widespread and substantial throughout the animal world1,2. Table S6. The major axis of divergence (d1) was closely aligned with the major axis of conserved genetic variance (h1; Table 4), suggesting that a majority of phenotypic divergence has occurred along the genetic line of least resistance. To visualize G matrices in two dimensions, we estimated best‐linear unbiased predictors of breeding values for each trait in ASReml and transformed them using the coefficients of d1 and d2, the axes of greatest morphological divergence. Nature 2 Answers. Biol. However, the extent to which genetic constraints influence larger scale evolutionary change, such as phenotypic divergence in species radiations, remains a major unresolved question in biology (Schluter 2000; Gould 2002). The importance of selection has been convincingly demonstrated many times, but the extent to which genetic architecture might constrain the long‐term outcomes of selection is poorly understood. 2009) using multivariate repeated‐measures animal models of natural‐log transformed traits with natural‐log snout‐vent length (SVL) as a covariate. 2005; Hansen and Houle 2008; Chenoweth et al. Shared and unique features of evolutionary diversification. 166, S1–S4 (2005), Article  Additive genetic (co)variance matrices (G) for seven Anolis species. Each species displays distinctive adaptations in body form, coloration and behavior suited to particular aspects of the microhabitat they occupy. These patterns are consistent with previous analyses of divergence in the West Indian Anolis radiation and reflect both divergence among islands and habitat specialization within lineages (Losos et al. 1998; Mahler et al. Immigration history controls diversification in experimental adaptive radiation. Educ. Nature 446, 436–439 (2007), Ricklefs, R. E. Cladogenesis and morphological diversification in passerine birds. Here, the axis capturing the most genetic variance in this subspace—the major axis of each ellipse—tends to be biased toward d1. Taken together, these results suggest that the alignment of phenotypic divergence and G‐matrix evolution may be a general phenomenon. Anolis lizards of the Greater Antilles represent one of the classic examples of vertebrate adaptive radiation. Psychol. I. This trait combination (e11) was highly aligned with both the first axis of divergence (d1) and the first axis of genetic variation (h1). 72, 541–559 (2002), Losos, J. In other words, evolution of G occurred in such a way as to preserve the relationship between axes of genetic variation and morphological divergence. 2005; Hansen and Houle 2008; Chenoweth et al. Unlike other Caribbean islands where different species have adapted to occupy different habitats, the island of … Such a process should not only stabilize the orientation of gmax but also cause changes in the magnitude of genetic variance explained by gmax. J.B.L. Working off-campus? Science 303, 356–359 (2004), ADS  Adaptive radiation, which results when a single ancestral species gives rise to many descendants, each adapted to a different part of the environment, is possibly the single most important source of biological diversity in the living world. 2003). The second axis of divergence (d2) was nearly orthogonal to h1 (Table 4) and was significantly but weakly aligned with the next axis of available genetic variation (h2; Table 4). 2008). Both Bahamian species are from lineages that originated in Cuba and colonized the Bahamas naturally (Kolbe et al. 2015) indicate that selection and drift can alter the characteristics of G, but it is unknown whether such changes tend to preserve or alter genetic lines of least resistance (but see Walter et al. Abstract Sexual dimorphism is widespread and substantial throughout the animal world. Sexual dimorphism and adaptive radiation in Anolis lizards. Such similarities can be characterized using Krzanowski's common subspace analysis (Krzanowski 1979; Aguirre et al. 2017). (2017) was able to estimate M in addition to G and D, demonstrating that both divergence and genetic variation could be predicted by mutation and suggesting a role for deep constraints in phenotypic evolution. Substituting these eigenvectors for hi in our subsequent analyses did not change our results. Biol. 425 spp., see text Anolis is a genus of anoles, iguanian lizards in the family Dactyloidae, native to the Americas. behavioral convergence and adaptive radiation: effects of habitat use on territorial behavior in anolis lizards Michele A. Johnson, 1 , 2 , 3 Liam J. Revell, 4 , 5 and Jonathan B. Losos 4 2010; Bolstad et al. Science 279, 2115–2118 (1998), Stamps, J. However, the observed changes in the G matrix across the seven Anolis species in this study occurred in a way that preserved the major axes of genetic variance. Whether such genetic constraints shape phenotypic diversity over macroevolutionary timescales is more controversial, however. One reason constraint might be less of a factor on macroevolutionary timescales is that G itself can evolve (Turelli 1988; Steppan et al. Symbols used in this article are listed in Table 2. Google Scholar, Hendry, A. P., Kelly, M. L., Kinnison, M. T. & Reznick, D. N. Parallel evolution of the sexes? Below, we discuss potential mechanisms that may maintain a relationship between genetic variation and adaptive divergence across an ancient radiation. commun. Monogr. We seek … By submitting a comment you agree to abide by our Terms and Community Guidelines. Press, Cambridge, 2004), Losos, J. 2018). Anoles comprise one of the most diverse vertebrate genera, with nearly 400 species known and more being discovered every year. Enter your email address below and we will send you your username, If the address matches an existing account you will receive an email with instructions to retrieve your username, Divergence of genetic architecture across the, Relationship between species divergence and genetic architecture. 2014) to characterize the directions in which G diverged across species (Table S2). Lizards in an Evolutionary Tree: Ecology and Adaptive Radiation of Anolesis the tenth volume in the University of California Press’s series on organisms and environments, whose unifying themes are the diversity of plants and animals, the ways they interact with each other and with their surroundings, and the implications of those relationships for science and society. The evolution of sexual dimorphism in animals: Hypotheses and tests. We used genetic covariance tensor analysis (Hine et al. Second, we show that despite this divergence, all G matrices retain at least two similar axes of genetic variation and that the divergence of morphological trait means is biased toward the greatest of these (h1 or gmax). out of the University of Liverpool. The closer these angles are to 0°, the better a particular h vector describes genetic variance within a particular species. & Doebeli, M. Sexual dimorphism and adaptive speciation: Two sides of the same ecological coin. A number of studies have provided empirical support for this prediction, but most work has been conducted on relatively short evolutionary timescales (1–2 million years, Schluter 1996; Blows and Higgie 2003; Bégin and Roff 2004; McGuigan et al. 1). Because these vectors describe axes of genetic variance across all species, they represent putative genetic lines of least resistance that may have influenced divergence. We acknowledge the National Science Foundation for financial support. (2021), Proceedings of the Royal Society B: Biological Sciences One of the best-studied examples involves Caribbean Anolis… Nature An eigenvalue of 7 would indicate that a particular h vector can be reconstructed exactly for all seven species using the eigenvectors of its G matrix, and would suggest that a given eigenvector represented a conserved axis of genetic variance. 2009). But few studies have analyzed the equally species-diverse mainland Anolis. Journal of Herpetology 14 (2): 149-153 - get paper here; Yuan, M. L., M. H. Wake, and I. J. Wang. Using estimated intercepts, slopes from the regression of ln‐transformed trait values on ln‐transformed SVL, and the grand mean SVL across all seven species (34.81 mm; Table S2), we calculated size‐corrected species means for each trait. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Nat. Evidence for speciation The extent of speciation in Caribbean anoles can be viewed at severallevels. We then regressed log10‐transformed among‐species variances (log D) on log10‐transformed genetic variances (log G). The similarity of h1 to each species‐specific gmax and the alignment of e11 with both h1 and d1 are all consistent with this scenario. Use evidence from the trees to explain how the Anolis lizards are an example of this concept. 2014). Like gmax, this axis was most strongly loaded with limb traits. 2001). 1998; Losos 2009; Mahler et al. A. in Lizard Ecology: Studies Of A Model Organism (eds Huey, R. B., Pianka, E. R. & Schoener, T. W.) 169–204 (Harvard Univ. This test was repeated using two other measures of species divergence as well as the evolutionary rate matrix (see Additional Methods). Across species, all gmax vectors were significantly aligned with one another (θ = 11.8–46.7°, P < 0.05), but none were collinear (Fig. Caribbean Anolis lizards are a classic case of adaptive radiation, repeated four times across islands of the Greater Antilles. In addition, the age of the Anolis radiation (46.3–64.4 million years, Poe et al. Much of the divergence among species in this study (as well as across all species of West Indian anoles, Beuttell and Losos 1999; Mahler et al. 2017). For West Indian anoles, it is reasonable to expect that the adaptive landscape resembles a surface with multiple fitness peaks representing the ecomorphs that we see today (Mahler et al. Such adaptations are 2014). Drift should also cause population means to diverge in directions with more genetic variance, resulting in divergence along gmax and thus alignment between d1 and the direction of most change in G. Although this scenario is theoretically plausible, the well‐established role of selection in the evolution of Anolis ecomorphs (Losos 2009) suggests that neutral processes are highly unlikely to be the only factor explaining such alignment. Together, these findings suggest that groups of species may diverge along lines of genetic least resistance for millions of years and that this pattern is unlikely to be disrupted by concomitant changes in underlying genetic architecture. contributed to study design; J.W.M., J.J.K., D.L.M., J.B.L., and E.D.B. To determine whether vectors were significantly aligned, we compared this angle to a null distribution generated from a simulation of 100,000 pairs of randomly generated vectors. 2004; Glor et al. Google Scholar, Temeles, E. J., Pan, I. L., Brennan, J. L. & Horwitt, J. N. Evidence for ecological causation of sexual dimorphism in a hummingbird. The trace, or the sum of the eigenvalues of each G matrix (which is equivalent to the sum of the genetic variances), was used as an index of its overall size, which should predict the potential magnitude of a population's overall response to selection. lizards. 5. 3, Table S5; r = 0.95, P = 0.001). King for advice on analyses; and L. Harmon, J. Kolbe, B. Langerhans, S. Gavrilets and D. Simberloff for critical review of the manuscript. Here, we use a comparative study of Anolis lizards to assess the relationship between genetic constraints and phenotypic divergence in adaptive radiation. S1). Examination of first ten e vectors showed that divergence in G was more closely aligned with axes of conserved genetic variance (h1 and h2) than with axes of morphological divergence (d1 and d2; two‐tailed sign test, P = 0.04; Table S5). 164, 335–349 (2004), Olejnik, S. & Algina, J. On the other hand, genetic constraints may respond to natural selection in such a way as to facilitate further adaptive evolution. 2014; Walter et al. 2017). In other words, trait combinations with more genetic variance showed greater divergence. In simulation studies, this evolutionary scenario leads to an elongation of G in the direction of the moving optimum (Jones et al. Nature 447(7141): 202-205. Further, this signal persisted even as G itself evolved, apparently because the largest evolutionary changes in G were themselves aligned with the line of genetic least resistance. & Tautz, D.) 335–343 (Cambridge Univ. Next, we calculated the angles between each h vector and the subspace defined by the first four eigenvectors of G within each species (Aguirre et al. & Rainey, P. B. 4, 136–138 (1989), CAS  Eigenanalysis of this subspace provides a set of orthogonal vectors (hi) that represent axes of genetic variance that are shared to some extent across species, and its eigenvalues (p) indicate the extent to which those axes are shared. Analogous to a first principal component, the first eigentensor describes the subspace in which G varies the most across species. 25, 241–286 (2000), Duong, T. ks: Kernal Smoothing, version 1.3.4. The common subspace of genetic variation for all seven species; describes the orientations of trait space that share the most genetic variation and is defined using the first four eigenvectors of each, The eigentensors describing subspaces in which. A parallel exists between the divergence of experimental populations of A. sagrei and the adaptive radiation of Anolis lizards in the Greater Antilles; in both cases, relative hindlimb length and perch diameter are strongly correlated. Table S2. 2013). Science 289, 441–443 (2000), Fukami, T., Beaumont, H. J. E., Zhang, X.-X. However, the well‐established adaptive basis of morphological divergence in the Anolis radiation suggests that selection likely plays a role in generating this pattern. Although we cannot definitively distinguish between genetic constraint, drift, and selection as the cause of this pattern, the alternatives lead to equally compelling conclusions about the evolutionary process. 3. Alignment of d and h vectors would indicate that evolutionary change was biased toward such lines of least resistance. In the study of adaptive radiation, few organisms serve as a better model than Caribbean . 19, 741–754 (2006), Jarman, P. J. Social-organization of antelope in relation to their ecology. (2017). & Losos, J. The relevant genetic constraint on the evolution of G is the mutational (co)variance matrix M, which describes the per‐generation input of new genetic variation in a population. Conversely, if sexual dimorphism and interspecific divergence are alternative means of ecological diversification, then the degree of sexual dimorphism may be negatively related to the extent of adaptive radiation. Examination of the loadings of e11 indicates that it almost entirely represents divergence in the components of G involving limb length. Effects of predation and habitat features on the size and shape of wild guppies. We then plotted the 95% confidence ellipse centered at the species mean using JMP Pro 13.0. Eigenanalysis of this tensor provides genetic covariance eigentensors (Ei), which are square matrices describing independent subspaces in which in G varies across species. All results were similar when using alternative measures of species divergence (Table S6, Fig. Species coordinates within the second eigentensor (E2) were marginally correlated with the orientation of G (Table S5; r = 0.73, P = 0.06). 4; log‐log slope = 1.40 ± 0.208; P = 0.0005, R2 = 0.88). Genetic constraints are often considered to be less important over the longer evolutionary spans that generate species differences, but there are few empirical tests of this prediction (but see Houle et al. Oecologia 95, 525–532 (1993), ADS  Led by Tobias Uller & Nathalie Feiner . In all species, the axis of greatest genetic variance (gmax) was strongly associated with genetic variance in limb traits, which consistently showed strong positive loadings (Table S1). 2, where G matrices are plotted as ellipses in the subspace defined by d1 and d2 and centered on species means. This pattern may have arisen either because pervasive constraints have biased the course of adaptive evolution or because the G matrix itself has been shaped by selection to conform to the adaptive landscape. 2008). Second, multivariate nonlinear selection may further contribute to the stability of G by conforming its orientation to the adaptive landscape (Cheverud 1982; Jones et al. Such an adaptive landscape could stabilize certain aspects of G (such as gmax) and lead to the alignment between G and D. The repeated evolution of ecomorphs may resemble the repeated movement of fitness peaks along the same trait axis in response to interspecific competition (Schoener 1968; Williams 1972; Losos 1990b; Losos et al. After estimating G matrices, we conducted analyses allowing us to characterize both similarities and differences across species.