SUPPLEMENTARY INFORMATION

Transcription

1 doi: /nature05774 SUPPLEMENTARY INFORMATION Sexual Dimorphism is Greater on Jamaica than on Puerto Rico. Analyses. We used Mahalanobis distances to compare the degree of multivariate shape dimorphism across islands. Using the shape canonical variate scores used in the Manova analysis, we calculated mean coordinates for each species-sex class in shape morphology space, and used these to compute distances between sexes for each species (a summary variable for multivariate sexual shape dimorphism). We then compared mean shape SD across islands. We used ANOVA to test for significance. From the full factorial ANOVA model (island, ecomorph and their interaction) on shape variation, we examined the island effect for significant difference between Jamaica and Puerto Rico in levels of shape SD. Incorporation of phylogenetic information into this analysis is not possible because phylogenetic relationships and island are for the most part confounded. The ANOVA/MANOVA analyses for this paper were generated using SAS/STAT software 1. Supplementary Table 1. ANOVA results comparing shape dimorphism across islands. Effect Type III Sums of Squares Mean Square F-Value Degrees of Freedom Pr>F ISLAND ECOMORPH ECOMORPH*ISLAND

2 Supplementary Table 2. Mean values of morphological variables and sample sizes used (N). ECOMORPH SPECIES SEX N SVL MASS FOREL HINDL LAMN SIZE SSVL SMASS SFOREL SHINDL STAIL SLAMN Crown-Giant A. cuvieri F M A. garmani F M Grass-Bush A. krugi F M A. poncensis F M A. pulchellus F M Trunk-Crown A. evermanni F M A. grahami F M A. opalinus F M A. stratulus F M Trunk-Ground A. cristatellus F M A. gundlachi F M A. lineatopus F M A. sagrei F M Twig A. occultus F M A. valencienni F M SVL = snout-to-vent length (mm) MASS (g) FOREL = forelimb length (mm) 2

3 HINDL = hindlimb length(mm) LAMN = lamella number SIZE = natural log of the geometric mean size SSVL = size-adjusted SVL = log(svl) SIZE All species are of the genus Anolis. The size-adjusted variables are calculated in the same fashion as SSVL (see the equation above). Some of these data were reported previously

4 Supplementary Table 3. Results of Morphological Canonical Variate Analysis. (A) Significance tests of Canonical Variate Axes Eigen- Cumulative Likelihood Approx value Variance Ratio F-Value NumDF DenDF P-value CAN < CAN CAN < < CAN < (B) Canonical Structure Total Variable Can1 Can2 Can3 Can4 SSVL SMASS SFOREL SHINDL SLAMN Between Species-Sex Classes Variable Can1 Can2 Can3 Can4 SSVL SMASS SFOREL SHINDL SLAMN Pooled Within Species-Sex Classes Variable Can1 Can2 Can3 Can4 SSVL SMASS SFOREL SHINDL SLAMN NumDF = numerator degrees of freedom 4

5 DenDF = demoninator degrees of freedom Can1-4 = Canonical variate axes 1 through 4 Associated eigenvalues and significance levels of canonical variate axes (A) followed by canonical structure (B: total, between group, and pooled within group loadings of shape morphology variables on each canonical axis). Groups in this analysis were the species-sex classes. Variable abbreviations are given in supplementary table

6 Supplementary Appendix. Comparison of Canonical Variates Analysis and Principal Components Analysis Canonical Variates Analysis (CVA) was used to reduce dimensionality in the analysis presented in the main text. CVA accounts for group-structure, in this case, variation within species-sex groups, which is the smallest unit of biological organization in our data. Prinicpal Components Analysis (PCA) treats all data points equally. Because PCA assumes that there is no sub-structure within the data, it blends within and between-group variation. Here we present the results of the niche-filling analysis repeated using PCA. The result remains significant (females increase morphospace occupation; P=0.017), and large portions of female-ecomorph and male-ecomorph space do not overlap (Supplementary Table 4). Supplementary Table 4. Morphological niche-packing analysis using Principal Components Analysis Ecomorph N ind Tot cub es Unique c ubes Both Sexes * Males Females Unique% N ind Tot cubesuni que cubes Unique % N ind Tot cubesuni que cubes Unique % Trunk-Ground Trunk-Crown Crown-Giant Grass-Bush Twig All Total cubes We compared the overlapping versus unique space occupied by ecomorphs and ecomorph-sex classes to assess the relative contributions to morphospace volume. 6

7 * 'Both sexes' refers to ecomorph analyses with sexes combined, whereas ecomorph-females and ecomorph-males separates each ecomorph class by sex. N ind, number of individuals. Tot cubes, the number of (morphospace volume) cubes filled by each sex, ecomorph or ecomorph-sex class; these cubes may contain individuals of more than one class. See Methods for details. We measured 'Unique cubes ', the volume occupied by each class. It is the volume of morphospace that is occupied solely by the given class relative to the entire volume occupied all classes. Unique cubes excludes those cubes which contain more than one class. Unique% refers to the percentage of cubes occupied by only the given class relative to the total number of cubes. Forty-four cubes are occupied by both sexes. 7

8 Morphological separation between sexes and ecomorphs in canonical variate space. Twig anoles have high values on CVA1 with long bodies, many lamellae, and short hindlimbs (Figure 2, supplementary movie, Supplementary Figure 2, and Supplementary Table 2). In contrast, trunk-ground anoles and crown-giant anoles have low values of CVA1, short bodies, few lamellae, and long limbs. CVA1 is not particularly effective in separating sexes, except perhaps for trunk-crown anoles, whose sexes are well separated by a combination of CVA1 and 2. High CVA2 contrasts greater lamella number with shorter bodies. Both grass-bush and trunk-crown anoles have high CVA2, whereas crown-giant and twig anoles have low values. As a whole, trunk-ground anoles are intermediate in CVA2, but the sexes of these anoles separate well along CVA2 with male trunk-ground anoles having lower CVA2 than females. CVA3 describes a contrast between long forelimbs and to a lesser extent, shorter hindlimbs. CVA3 separates trunk-crown (high) and grass-bush anoles well. In general, the sexes of trunk-ground and trunk-crown anoles are well-separated, whereas the sexes of twig anoles appear to form a more mixed cluster (Figure 2). Sex differences in Ecology Sex differences in resource use have been reported for many species of Anolis lizards. Important niche parameters include differences in prey characteristics 3-11 and micro-habitat use 2,7,8, Analysis of ecological data for the species in this study demonstrate that sexual differences in habitat use are extensive and that degree of intersexual habitat differentiation is 8

9 correlated with degree of sexual dimorphism in morphology. Furthermore, the ecologicallybased patterns are parallel to the morphological pattern described in the text. In particular, for the species in this study, analyses of niche packing show that the sexes of each ecomorph tend to occupy non-overlapping regions in ecospace, in similar fashion to the morphological data (supplementary table 5). Only 25% of occupied cells contain both male and female indiviuduals. By contrast, females uniquely occupy 34% of ecospace and males uniquely occupy 31% of ecospace. Thus, sexual differences significantly increase the density of habitat-use space occupation in the anole radiation (P<0.0042). This result remains significant if closely-related species are excluded (P<0.0014). Additionally, MANOVA finds that ecological variation is ordered by ecomorph, sex, and by the interaction of ecomorph and sex (Supplementary Table 6). The link between dimorphism is morphology and niche use is revealed by canonical correlation analysis, which shows a strong relationship between degree of sexual shape dimorphism and extent of intersexual difference in habitat use: species with great intersexual differences in habitat use also exhibit high levels of shape dimorphism. Two canonical correlation axes explain 93% of the variation (r 2 CC1 = 0.82, r 2 CC2 = 0.50; Roy s Greatest Root = 2.080, P< ). The variables which correlate most strongly with the first canonical axes are dimorphism in snout-to-vent length and forelimb lengh with sex differences in perch diameter and distance to medium-diameter perches (Supplementary Table 7). The second canonical axis describes multivariate correlation between sexual differences in lamella number and mass with distance to medium and large-diameter perches. 9

10 METHODS (ecological analyses) Data collection. We measured five habitat use variables following the methodology of previous studies 2,28. Briefly, the observer walked slowly through the habitat and for each lizard sighted, recorded its species, sex, height above the ground (perch height; PHT), perch diameter (PDIA), and distance to the closest perch in three diameter categories (DIST1 DIST3): (1) cm; (2) cm; and (3) 1.5 cm. Species mean values for habitat-use based on 1024 individuals are presented in Supplementary Table 8 for both sexes of 14 species. Some of these data have been published previously 2. All analyses were based on log-transformed data. Multivariate analyses of habitat use. We conduced niche-filling and MANOVA analyses on habitat-use data following the same methodology used for morphological data (see Methods). The relationship between sexual dimorphism and sex differences in habitat use. We used canonical correlation analysis to test for a multivariate association between sexual shape dimorphism and sex differences in habitat use. Sexual difference in habitat use was computed as the difference between mean female and male values for log-transformed habitat use variables. Sexual dimorphism values were computed in the same way using mean shape variables. We conducted a phylogenetic version of the canonical correlation analysis using the phylogenetic GLS method described previously 29 on species mean values. 10

11 Supplementary Table 5. Habitat-use niche-packing analysis Ecomorph N ind Tot cubes Both Sexes * Males Females Unique cubes Unique % N ind Tot cubes Unique cubes Unique % N ind Tot cubes Unique cubes Unique % Trunk-Ground Trunk-Crown Crown-Giant Grass-Bush Twig All Refer to Supplementary Table 4 legend for explanation of terms. 11

12 Supplementary Table 6. MANOVA results for sex and ecomorphological differences in habitat use. Effect Wilks' λ F-Value P-value p q r η 2 SEX < % ECOMORPH < % SEX*ECOMORPH % SPECIES(ECOMORPH) < % Wilks λ (lambda) F-value = value from F distribution p = number of dependent variables q = number of independent degrees of freedom r = minimum of p or q η 2 (eta-squared) = multivariate partial variance 30,31 = 1 - λ (1/r) SPECIES(ECOMORPH) = species nested within ecomorph All habitat use variables entered into the model as dependent variables. Independent variables included in the model are listed under Effect. 12

13 Supplementary Table 7. Correlation (r 2 ) between canonical correlation axes and their respective variables: sexual dimorphism (size-adjusted) and sex differences in ecology. Delta (Δ) indicates difference between sexes. Morphology Ecology CC1 CC2 CC1 CC2 SSVLΔ PDIAΔ SMASSΔ PHTΔ SFORELΔ DIST1Δ SHINDLΔ DIST2Δ SLAMNΔ DIST3Δ Supplementary Table 8. Mean values of habitat use variables and sample sizes used (N). ECOMORPH SPECIES SEX N PHT PDIA DIST1 DIST2 DIST3 Crown-Giant A. cuvieri F M A. garmani F M Grass-Bush A. krugi F M A. poncensis F M A. pulchellus F M Trunk-Crown A. evermanni F M A. grahami F M A. opalinus F M A. stratulus F M Trunk-Ground A. cristatellus F M A. gundlachi F M A. lineatopus F M A. sagrei F M Twig A. valencienni F M PHT = perch height (m) 13

14 PDIA = perch diameter (cm) DIST1 = distance (cm) to nearest available thin diameter ( cm) perch DIST2 = distance (cm) to nearest available medium diameter ( cm) perch DIST3 = distance (cm) to nearest available wide diameter ( 1.5 cm) perch All species are of the genus Anolis. Some of these data were reported previously

15 cuvieri lineatopus valencienni grahami garmani opalinus sagrei evermanni stratulus krugi pulchellus cristatellus gundchi poncensis occultus

16 a b can can can 3

17 Supplementary Figure Legends Supplementary Movie. A rotating view of the 3D positions of male and female ecomorph densities shown in figure 2. The 3D object was plotted using ks 32 and misc3d 33 in the R statistical language 34. We created a movie of the 3D object using the rgl package 35 to create a stack of pictures, and imagej 36 to produce a movie from the stacked images. Supplementary figure 1. The phylogenetic hypotheses used in the study. The species included in the study were a subset of the larger Anolis phylogeny 37. The tree is ultrametric and branches are drawn proportionally to time, with the time depth from tip species to root equal to time=1.0. Supplementary figure 2. Two dimensional graphs of morphospace overlap between ecomorphsex classes. These graphs were constructed using 2D KDA 32 on can1, can2 (A) and separately on can2, can3 (B; see description for 3D analysis). Plotted are the inner 20% contours for the ecomorph-sex classes. Individual datapoints are overlaid on the color regions. crown-giant = green diamonds, grass-bush = yellow triangles, trunk-crown = red circles, trunk-ground= blue squares, twig = purple crossed boxes and asterisks. Females are depicted in lighter colors and open symbols, males are darker colors with filled symbols. 17

18 Supplementary References: 1. SAS Institute Inc. Version 9 of the SAS system for Unix. ( ). SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. indicates USA registration. 2. Butler, M. A. & Losos, J. B. Multivariate sexual dimorphism, sexual selection, and adaptation in Greater Antillean Anolis lizards. Ecological Monographs 72, (2002). 3. Floyd, H. B. & Jenssen, T. A. Food habits of the Jamaican lizard Anolis opalinus: Resource partitioning and seasonal effects examined. Copeia 1983, (1983). 4. Preest, M. R. Sexual size dimorphism and feeding energetics in Anolis carolinensis: Why do females take smaller prey than males Journal of Herpetology 28, (1994). 5. Perry, G. The evolution of sexual dimorphism in the lizard Anolis polylepis (Iguania): Evidence from intraspecific variation in foraging behavior and diet. Canadian Journal of Zoology 74, (1996). 6. Parmelee, J. R. & Guyer, C. Sexual differences in foraging behavior of an anoline lizard, Norops humilis. Journal of Herpetology 29, (1995). 7. Lister, B. C. Seasonal niche relationships of rain forest anoles. Ecology 62, (1981). 8. Schoener, T. W. Ecological significance of sexual dimorphism in size in the lizard Anolis conspersus. Science 155, (1967). 9. Schoener, T. W. Anolis lizards of Bimini: resource partitioning in a complex fauna. Ecology 49, (1968). 10. Schoener, T. W. & Gorman, G. C. Some niche differences in three Lesser Antillean lizards of the genus Anolis. Ecology 49, (1968). 11. Talbot, J. J. Time budget, niche overlap, inter- and intraspecific aggression in Anolis humlis and Anolis limifrons from Costa Rica. Copeia 1979, (1979). 12. Collette, B. B. Correlations between ecology and morphology in anoline lizards from Havana, Cuba and southern Florida. Bulletin of the Museum of Comparative Zoology 125, (1961). 13. Andrews, R. M. Structural habitat and time budget of a tropical Anolis lizard. Ecology 52, (1971). 14. Lister, B. C. Nature of niche expansion in West-Indian Anolis lizards.1. Ecological consequences of reduced competition. Evolution 30, (1976). 15. Lister, B. C. & Aguayo, A. G. Seasonality, predation, and the behavior of a tropical mainland anole. Journal of Animal Ecology 61, (1992). 16. Pounds, J. A. Ecomorphology, locomotion, and microhabitat structure: patterns in a tropical mainland Anolis community. Ecological Monographs 58, (1988). 17. Powell, G. L. & Russell, A. P. Locomotor correlates of ecomorph designation in Anolis: an examination of three sympatric species from Jamaica. Canadian Journal of Zoology 70, (1992). 18. Rand, A. S. & Rand, P. J. Field notes on Anolis lineatus in Curacao. Studies on the Fauna of Curaçao and other Caribbean Islands 24, (1967). 19. Rand, A. S. Ecology and social organization in the iguanid lizard Anolis lineatopus. Proceedings of the United States National Museum 122, 1-79 (1967). 20. Rand, A. S. The ecological distribution of the anoline lizards around Kingston, Jamaica. 18

19 Breviora 272, 1-18 (1967). 21. Rodríguez Schettino, L. & Martínez Reyes, M. Algunos aspectos de la ecología trófica de Anolis argenteolus (Sauria: Polychridae) en una localidad de la Costa Suroriental de Cuba. Biotropica 28, (1996). 22. Schoener, T. W. & Schoener, A. Structural habitats of West Indian Anolis lizards. I. Lowland Jamaica. Breviora 368, 1-53 (1971). 23. Schoener, T. W. & Schoener, A. Structural habitats of West Indian Anolis lizards. II. Puerto Rican uplands. Breviora 375, 1-39 (1971). 24. Scott, N. J., Wilson, D. E., Jones, C. & Andrews, R. M. The choice of perch diameters by lizards of the genus Anolis (Reptilia, Lacertilia, Iguanidae). Journal of Herpetology 10, (1976). 25. Jenssen, T. A. The ethoecology of Anolis nebulosus (Sauria, Iguanidae). Journal of Herpetology 4, 1-38 (1970). 26. Irschick, D. J., Vanhooydonck, B., Herrel, A. & Meyers, J. Intraspecific correlations among morphology, performance and habitat use within a green anole lizard (Anolis carolinensis) population. Biological Journal of the Linnean Society 85, (2005). 27. Herrel, A., Joachim, R., Vanhooydonck, B. & Irschick D. Ecological consequences of ontogenetic changes in head shape and bite performance in the Jamaican lizard Anolis lineatopus. Biological Journal of the Linnean Society 89, (2006). 28. Losos, J. B. Ecomorphology, performance capability, and scaling of West-Indian Anolis lizards: an evolutionary analysis. Ecological Monographs 60, (1990). 29. Butler, M. A., Schoener, T. W. & Losos, J. B. The relationship between sexual size dimorphism and habitat use in Greater Antillean Anolis Lizards. Evolution 54, (2000). 30. Langerhans, R. B. & DeWitt, T. J. Shared and unique features of evolutionary diversification. American Naturalist 164, (2004). 31. Olejnik, S. & Algina, J. Measures of effect size for comparative studies: Applications, interpretations, and limitations. Contemporary Educational Psychology 25, (2000). 32. Duong, T. ks: Kernel smoothing, R package version (2005). 33. Feng, D. & Tierney, L. misc3d: Miscellaneous 3D Plots, R package version (2005). 34. R Development Core Team. A language and environment for statistical computing, version (R Foundation for Statistical Computing, Vienna, 2005). 35. Adler, D. rgl: 3D visualization device system (OpenGL), R package version (2005). 36. Rasband, W. S. ImageJ, (U.S. National Institutes of Health, Bethesda, ). 37. Nicholson, K. E., Glor R. E., Kolbe, J. J., Larson A., Hedges, S. B. and Losos, J. B. Mainland colonization by island lizards. Journal of Biogeography 32, (2005). 19