Which trait is advantageous for anoles that live in trees and cling to leaves?

The green anole is five to eight inches in length. It has a sharp nose, a narrow head, a slender body and a long, skinny tail. It has pads on its feet that help it climb, run on and cling to lots of different surfaces. It has a white belly and lips, very long hind legs, and moveable eyelids. Its eyes can move independent of each other.

Nội dung chính Show

INTRODUCTION
MATERIAL AND METHODS
Specimen sampling and morphological measurements
Assessing support for the ecomorphs
Classification of mainland anoles to the Caribbean ecomorphs based on morphology
Assessing a novel ground ecomorph
Reconstructing the evolution of mainland ecomorphs and their convergence
Investigating faunal convergence between island and mainland radiations
Data availability
Mainland and Caribbean anoles in the morphospace
Support for the Caribbean ecomorphs
Classification of mainland anoles into the six Caribbean ecomorphs
Evaluation of a new ground ecomorph
Repeated ecomorph evolution on the mainland
Faunal convergence between island and mainland anole radiations
Testable ecomorph classifications based on morphology
An ecomorph of ground anoles
Replicate adaptive radiations in island and mainland settings
SUPPORTING INFORMATION
ACKNOWLEDGEMENTS

On most green anole, the eyes are surrounded by a thin turquoise border. The green anole is normally bright green, but it can change its color to brownish-green or dark brown in seconds.

It is sometimes call the American chameleon because it can change color, but it is not a true chameleon. The green anole changes color because of changes in temperature, humidity, health and mood, not to blend into its background. Males have a pink or red throat fan or dewlap.

Range

The green anole can be found in the Southeastern United States from southern Virginia to the Florida Keys and west to Central Texas and Oklahoma.

It has been introduced to Hawaii.

HabitatThe green anole lives in habitats with very high humidity. It can be found in swamps, forests, wooded beaches and other areas with trees. It may also be found in parks and yards. The green anole is often found basking in the sun as it clings to trees, shrubs, vines, palm fronds, fence posts and walls.

DietThe green anole eats spiders, flies, crickets, small beetles, moths, butterflies, small slugs, worms, ants and termites. It only notices prey that is moving. It gets most of its water from the dew on plants. Life CycleThe green anole breeds from March to September. The male will establish a territory and patrol it. He will attract females by puffing out his dewlap. He mates with females in his area and aggressively defends his territory from other males.

When a female and male mate, the female stores the sperm. If she doesn't mate with another male, the stored sperm will fertilize her eggs. The female lays a single egg and buries it in moist leaf litter, hollow logs or the soil. She will lay one egg every two weeks during breeding season. She may lay up to 15-18 eggs during the summer. The female does not stay with the egg or care for the young that will hatch in five to seven weeks. Young green anole eat small insects like mealworms, fruit and house flies and termites.

Uncovering convergent and divergent patterns of diversification is a major goal of evolutionary biology. On four Greater Antillean islands, Anolis lizards have convergently evolved sets of species with similar ecologies and morphologies (ecomorphs). However, it is unclear whether closely related anoles from Central and South America exhibit similar patterns of diversification. We generated an extensive morphological data set to test whether mainland Draconura-clade anoles are assignable to the Caribbean ecomorphs. Based on a new classification framework that accounts for different degrees of morphological support, we found morphological evidence for mainland representatives of all six Caribbean ecomorphs and evidence that many ecomorphs have also evolved repeatedly on the mainland. We also found strong evidence that ground-dwelling anoles from both the Caribbean and the mainland constitute a new and distinct ecomorph class. Beyond the ecomorph concept, we show that the island and mainland anole faunas exhibit exceptional morphological convergence, suggesting that they are more similar than previously understood. However, the island and mainland radiations are not identical, indicating that regional differences and historical contingencies can lead to replicate yet variable radiations. More broadly, our findings suggest that replicated radiations occur beyond island settings more often than previously recognized.

INTRODUCTION

A primary goal of evolutionary biology is to understand the causes of similar and dissimilar evolutionary outcomes. Adaptive radiation is the rise of diverse ecological and morphological forms, descended from a single common ancestor (Simpson, 1953; Schluter, 2000; Losos, 2011; Gillespie et al., 2020). This hallmark evolutionary process often works in tandem with convergence to produce repeated evolutions of similar form-function relationships. When different species occupy similar ecological niches, selection for similar functional attributes often leads to morphological convergence (Schluter & Nagel, 1995; Losos, 2011; Santana & Cheung, 2016; Burress et al., 2018; Moen, 2019). In some cases, repeated ecological diversification leads to exceptional morphological convergence between entire radiations in distinct regions, as seen in African rift lake cichlids, Hawaiian spiders, stickleback fishes, Bonin Islands snails, Caribbean rain frogs and Caribbean anole lizards (Kocher et al., 1993; Losos et al., 1998; Rundle et al., 2000; Chiba, 2004; Gillespie et al., 2018; Dugo-Cota et al., 2019). These replicated adaptive radiations illustrate the deterministic aspects of evolution. However, the generality of replicated radiations is not clear, as most recognized cases occur in insular and lacustrine systems, where neighbouring islands and lakes present similar environments (Schluter, 2000; Losos, 2009; Prates & Singhal, 2020).

Comparisons of closely related taxa in island and mainland settings are useful for testing the pervasiveness of replicated adaptive radiations. Many factors differ between island and mainland environments that might lead to disparate evolutionary outcomes. In general, islands present fewer competitor and predator species and more “unoccupied” niches compared to mainland environments, which tend to be older and often have more diverse and established communities (MacArthur & Wilson, 1967; Carlquist, 1974; Schluter, 1988). However, recent studies suggest that some mainland environments provide ample opportunities, if not more, for physiological diversification than insular habitats (Velasco et al., 2016, 2018; Salazar et al., 2019). To test for convergent patterns of adaptive radiation, we can characterize diversification in closely related groups of species with similar ecologies across regions (Schluter, 1988; Irschick et al., 1997; Bossuyt & Milinkovich, 2000; Schaad & Poe, 2010). The independent evolution of similar morphologies with strong ties to ecology would suggest that island and mainland lineages are evolving under similar selective regimes and support the hypothesis of convergent adaptive radiations. By contrast, the absence of morphological convergence in mainland lineages would suggest that replicated radiations are primarily restricted to island settings.

Anole lizards (Anolis) provide an exemplary system for investigating convergent evolution and adaptive radiation (Losos, 2009). In the Greater Antilles, anoles have radiated four times on four different islands, where they repeatedly evolved habitat specialists with similar morphological adaptations (Losos et al., 1998). Based on their shared ecological and morphological traits, most Greater Antillean anole species have been assigned to one of six classes termed “ecomorphs”, named (mostly) after the structural microhabitats characteristically used by their members: crown-giant, grass-bush, trunk, trunk-crown, trunk-ground and twig (Rand & Williams, 1969; Williams, 1972, 1983; Losos, 2009; Fig. 1). Although many investigations have studied the evolution and pervasiveness of the Caribbean ecomorphs (Irschick et al., 1997; Losos & de Queiroz, 1997; Losos et al., 1998; Velasco & Herrel, 2007; Schaad & Poe, 2010), the six ecomorphs do not describe the full range of ecological diversity exhibited by anoles, even in the West Indies (Mahler et al., 2013). For instance, there are ground-dwelling, rock-dwelling and semi-aquatic anoles in both island and mainland settings (Losos, 2009). The semi-aquatic species have not converged on similar morphologies (Leal et al., 2002; Muñoz et al., 2015); however, other habitat specialists might constitute undescribed ecomorphs that have been overlooked by studies that used only the six Caribbean ecomorphs to investigate convergent patterns of adaptive radiation (Mahler et al., 2013; Moreno-Arias & Calderón-Espinosa, 2016; Poe & Anderson, 2019; Moreno-Arias et al., 2020).

Which trait is advantageous for anoles that live in trees and cling to leaves?

West Indian Anolis lizards representing the six Caribbean ecomorphs (A-F) and the morphologically similar mainland Draconura-clade species (a-f). (A) Anolis cuvieri, photo courtesy of Matt McElroy; (B) Anolispulchellus, photo courtesy of Kristiina Ovaska; (C) Anolis distichus, photo courtesy of Pierson Hill; (D) Anolis stratulus, photo courtesy of Alberto López; (E) Anolis lineatopus, photo courtesy of D. Luke Mahler; (F) Anolis occultus, photo courtesy of D. Luke Mahler; (a) Anolis petersii, photo courtesy of Carlos R. Beutelspacher; (b) Anolis auratus, photo courtesy of Kenro Kusumi; (c) Anolis ortonii, photo by Ivan Prates; (d) Anolis omiltemanus, photo courtesy of Gunther Köhler; (e) Anolis bicaorum, photo courtesy of Sofia Prado-Irwin; (f) Anolis salvini, photo courtesy of Sebastian Lotzkat.

Of the 430 described species of anoles (Uetz et al., 2020), the majority occur in Central and South America, yet the ecological and morphological evolution of mainland anoles remains understudied (Nicholson et al., 2005; Losos, 2009). There are two clades of mainland anoles; one of which, the Draconura clade, is derived from a Caribbean ancestor that recolonized the mainland (Nicholson et al., 2005; Poe, 2017). The Draconura clade exhibits comparable species richness, rates of morphological evolution and physiological diversity to the Caribbean anoles, suggesting that this clade underwent adaptive radiation on the mainland (Pinto et al., 2008; Nicholson et al., 2012; Poe et al., 2018; Salazar et al., 2019; Poe & Anderson, 2019). Relatively little is known about their ecology, but natural history data suggest that many Draconura species are associated with similar structural habitats to those of the Caribbean ecomorphs (Savage, 2002; Köhler et al., 2014; McCranie & Köhler, 2015). Additionally, several Draconura species are most often found on the ground, a habitat preference found in only a few Caribbean anoles (Vitt et al., 2001, 2002, 2003; Henderson & Powell, 2009; Fig. 2). These ecological similarities and differences set the stage to investigate whether mainland Draconura species have converged on similar morphological adaptations as their island counterparts.

Four species of ground-dwelling Anolis lizards. (A) A. humilis (mainland), photo courtesy of Sebastian Lotzkat; (B) A. trachyderma (mainland), photo by Ivan Prates; (C) A. tandai (mainland), photo by Ivan Prates; (D) A. barbouri (Caribbean), photo by Kevin de Queiroz.

Previous attempts to classify mainland anoles into the Caribbean ecomorphs have found a low number of mainland ecomorph species, suggesting that island and mainland anoles are on separate evolutionary trajectories (Irschick et al., 1997; Velasco & Herrell, 2007; Schaad & Poe, 2010). These studies relied on the well-established links between ecology and morphology in Caribbean anoles to assign mainland species to the same ecomorphs based on morphological similarities to the Caribbean anoles (Irschick et al., 1997; Velasco & Herrell, 2007; Schaad & Poe, 2010). This approach may be too restrictive if the assignment criteria are unrealistic or the Caribbean species used to delimit the ecomorphs do not encompass the full extent of morphological diversity associated with a given ecomorph. As a result, mainland species may have remained unclassified despite potentially being ecologically and morphologically similar to one of the ecomorphs. However, recent studies using cluster methods have objectively detected similar island and mainland anole morphotypes (Moreno-Arias et al., 2016; Poe & Anderson, 2019). Recognizing species with less extreme ecomorphological adaptations and distinguishing them from those that do not conform to any ecomorph (ecologically or morphologically) may provide more insights into whether species in distinct geographic regions have evolved similar traits in response to similar selective pressures.

Here, we reassess patterns of ecomorph convergence between Caribbean anoles and the mainland Draconura clade of anoles using an extensive morphological data set. We propose a framework to assign mainland species to the Caribbean ecomorphs with different levels of morphological support. We also investigate whether island and mainland ground-dwelling anoles share morphological traits that are distinct from those of the six Caribbean ecomorphs. We regard our ecomorph assignments as hypotheses of habitat use and evaluate our assignments based on currently available ecological data. We then use phylogenetic comparative methods to investigate whether the ecomorphs have evolved repeatedly on the mainland and test the strength of their convergence. Lastly, we test whether the mainland Draconura clade and Caribbean radiations are convergent without a priori ecomorph assignments. To assess convergent evolutionary outcomes within and between island and mainland anoles, we specifically ask: (1) Are mainland Draconura species assignable to the Caribbean ecomorphs based on morphology? (2) Do island and mainland ground-dwelling anoles constitute a previously unrecognized ecomorph class? (3) Have the six Caribbean ecomorphs and potential ground ecomorph evolved on the mainland independently from the Caribbean, and if so, did they evolve repeatedly? (4) To what extent do the Greater Antillean and mainland Draconura radiations exhibit morphological similarity?

MATERIAL AND METHODS

We approached our investigation of morphological convergence between Caribbean and mainland Draconura-clade anoles with two complementary approaches. We first assessed whether mainland Draconura and previously unclassified Caribbean species can be assigned to ecomorph classes based on their morphological similarity to Caribbean ecomorph species. We also assessed their degree of convergence. However, these results may be biased by subjective a priori assignments of reference ecomorph species, so we also used more objective methods (e.g. cluster analyses, tests for species-for-species matching and the detection of morphological shifts across the adaptive evolutionary landscape) to identify instances of convergence without a priori ecomorph assignments. Congruent results across approaches would support the robustness of our conclusions.

Specimen sampling and morphological measurements

We collected morphological data for 205 Anolis species (including 99 Draconura-clade species) using preserved specimens deposited in natural history collections. We examined 347 adult male lizards, with an average of 1.69 (range 1–4) specimens per species. Armstead & Poe (2015) demonstrated that a single specimen per species is sufficient to capture ecologically relevant interspecific morphological variation in anoles. Specimens were identified as male based on the presence of hemipenes or enlarged post-anal scales. Of the 99 Draconura-clade species, 91 were found in mainland Central and South America, and eight species were endemic to Caribbean or Pacific islands. Together they represent roughly 67% of the total known Draconura species (Nicholson, 2012; Poe et al., 2017). Of the remaining 106 species that we sampled, 71 were Caribbean species that have sufficient ecological and morphological data to support previous ecomorph assignments (hereafter referred to as “Caribbean ecomorph” species), and 35 were Greater and Lesser Antillean species not assigned to an ecomorph (hereafter referred to as “previously unclassified Caribbean” species). The Supporting Information (Table S1) contains a full species list with ecomorph assignments.

To compare the ecomorphology of mainland Draconura and Caribbean species, we measured 13 external morphological traits previously associated with ecology and locomotor performance (Losos, 1990; Beuttell & Losos, 1999; Macrini et al., 2003; Elstrott & Irschick, 2004). Measurements were taken as defined by Köhler (2014) and included: snout-vent length (SVL), tail length, head length, snout length, head width, humerus length, radius length, hand length, femur length, tibia length, foot length and the width of the subdigital pads on the fourth finger and fourth toe. For tail length, we only measured individuals that possessed a complete, non-regenerated tail, and obtained at least one tail measurement per species. We were unable to obtain any specimens with complete tails for 11 species, so we used the body proportions from Poe & Anderson (2019) to estimate tail length. In the single species where subdigital pads are absent (Anolis onca), we measured the width of the digit in the corresponding region (antepenultimate phalanx). All measurements were taken to the nearest 0.1 mm with digital calipers or to the nearest 0.5 mm with a ruler by a single author (J.M.H.).

Prior to all analyses, the raw data were natural log transformed and averaged by species to produce mean trait values. To remove the effects of size and account for phylogenetic non-independence among species, the “phyl.resid” function from the phytools R package was used to phylogenetically regress each trait against SVL (Revell, 2012). When size-correcting tail length, we regressed the data separately using the average SVL values calculated from only the specimens with tail data, including the specimens for which we estimated tail data. The residuals from each regression were treated as our size-corrected data. For this and subsequent phylogenetic analyses, we used the maximum clade credibility (MCC) tree from Poe et al. (2017), which was estimated from both morphological and molecular data, and pruned it to include only our sampled taxa (Supporting Information, Materials and methods S1). However, we also repeated all of our analyses using a time-calibrated phylogeny from Poe et al. (2017), which was estimated from genetic data only (referred to as the maximum clade credibility time, or MRCT, by the original authors). We qualitatively compared the results inferred using the MCC and time-calibrated phylogenies and report notable deviations below.

Assessing support for the ecomorphs

To investigate morphological variation of the mainland Draconura and Caribbean anoles, we plotted their positions in a multidimensional trait space. We used the “phyl.pca” function in the phytools package (Revell, 2012) to perform a phylogenetically informed principal component analysis (pPCA) with a covariance matrix that incorporated the phylogenetically size-corrected traits and SVL as a proxy for body size. Based on a visual scree test, we retained the first five pPC axes for subsequent analyses. These axes accounted for 85.9% of the total trait variation and exhibited clear correlations with specific body traits.

To test whether the Caribbean ecomorphs occupied distinct areas of the morphospace, we performed a series of randomization tests to assess the significance of the Euclidean distances between their centroids. Using the pPC scores for the a priori Caribbean ecomorph species (from the retained pPC axes), we calculated the Euclidean distances between the centroids of each pairwise ecomorph combination. To determine whether the distance between two ecomorph centroids was greater than expected under a null model, we combined the members of those ecomorphs into a single set and then randomly assigned them to two groups, each with the same number of species originally assigned to each ecomorph. Then, using the pPC scores for the randomly assigned species, we calculated the centroids for each group and the Euclidean distance between them. This process was repeated 1000 times per ecomorph pair and the proportion of randomly sampled values less than the observed value was used as a p-value (alpha level = 0.05 with a Bonferroni correction for multiple comparisons).

We also assessed support for the ecomorphs using a hierarchical cluster analysis to group species based on morphological similarity. First, we calculated a Euclidean distance matrix using the size-corrected trait values and SVL for all a priori Caribbean ecomorph species. This matrix was the input for a hierarchical cluster analysis using the Ward method (Murtagh & Legendre, 2014), implemented with the “hclust” R function.

As a third method for assessing support for the ecomorphs, we used a discriminant function analysis (DFA). We performed a DFA and a MANOVA to test for significant differences among the Caribbean ecomorph classes with the size-corrected trait values and SVL using the MASS package in R (Venables & Ripley, 2002). We then assessed the DFA’s ability to correctly assign species to their respective ecomorphs using re-substitution and cross-validation.

Classification of mainland anoles to the Caribbean ecomorphs based on morphology

To assess whether any mainland Draconura species can be assigned to a Caribbean ecomorph based on morphology, we used the DFA trained with Caribbean ecomorph species data to classify Draconura species and the previously unclassified Caribbean species into ecomorph classes. A caveat with the DFA method is that it classifies all species, even those that are a poor fit to any of the possible classifications. Therefore, we also developed a set of criteria to assign species to ecomorphs based on their relative positions in the morphospace. Using the retained pPC scores for the Caribbean ecomorph species, we calculated three measures to capture the morphospatial relationships between species in the same ecomorph (Supporting Information, Table S3). First, we calculated the Euclidean distance between each Caribbean ecomorph species and the centroid (centroid distance, CD) of its respective ecomorph. Then we calculated the mean pairwise distance (MPD) between each ecomorph species and all of the members of its ecomorph. Lastly, we calculated the distance between each ecomorph species and its nearest neighbour (NND) from the same ecomorph. To classify the Draconura and previously unclassified Caribbean species based on whether they exhibit similar morphospatial relationships as the Caribbean ecomorph species, we calculated CD, MPD and NND for each Draconura and unclassified Caribbean species, relative to each ecomorph.

Based on the Euclidean distances, a species was assigned to an ecomorph if: (1) its distance to that ecomorph’s centroid was ≤ the CD of the furthest member of that class; (2) its average distance to all the members of that ecomorph was ≤ the largest MPD among members of that ecomorph; and (3) its distance to the nearest member of that ecomorph was ≤ the largest NND among the members of that ecomorph. We assessed every Draconura and previously unclassified Caribbean species for each criterion independently so that each species could satisfy one or more of the three criteria. We also allowed species to be classified into multiple ecomorphs using the same criterion to assess uncertainty in ecomorph assignments and support for alternative assignments.

We consolidated the results of the DFA and Euclidean distance classifications into a final ecomorph assignment using two composite criteria. A species was confidently assigned to a single ecomorph if the DFA assigned it with a posterior probability ≥ 0.90, and it also satisfied any combination of at least two Euclidean distance criteria for the same ecomorph as the DFA assignment. Alternatively, a species was confidently assigned to an ecomorph if the DFA assigned it with a posterior probability ≥ 0.95, and it also satisfied any one of the Euclidean distance criteria for the same ecomorph as the DFA assignment. With this approach, species that did not meet both the DFA and Euclidean distance criteria were considered unclassified even if they were assigned to an ecomorph with a high DFA posterior probability. However, several species with a DFA posterior probability < 0.9 for a single ecomorph exhibited strong DFA and Euclidean distance support for two ecomorphs and likely represent intermediate or less specialized ecomorph species. These species were given two potential ecomorph assignments if the DFA assigned them to two ecomorphs with a total posterior probability ≥ 0.90, they were closest to the centroids of those ecomorphs, and their nearest ecomorph neighbour was a member of one of those ecomorphs. The final ecomorph assignments inferred from morphology were evaluated using available ecological and natural history data in the literature (Supporting Information, Table S9).

Assessing a novel ground ecomorph

To assess the existence of a previously unrecognized ground ecomorph, we conducted a literature review to identify Caribbean and mainland Draconura species that are predominantly ground-dwelling. Based on observations of perch use and other natural history data, we designated ten mainland Draconura species (Anolis bombiceps, Anolis brasiliensis, Anolis chrysolepis, Anolis humilis, Anolis planiceps, Anolis quaggulus, Anolis scypheus, Anolis tandai, Anolis trachyderma, Anolis uniformis), and one unclassified Caribbean species (Anolis barbouri) as exemplars of a hypothesized ground ecomorph (Avila-Pires, 1995; Howard et al., 1999, Vitt et al., 2001, 2002, 2003; Savage, 2002; McCranie & Köhler, 2015; Fig. 2). We then performed randomization tests and a cluster analysis as described above to assess whether the putative ground ecomorph occupied an area of the morphospace and formed a morphological cluster distinct from the six Caribbean ecomorphs, respectively.

To determine whether other mainland Draconura or unclassified Caribbean anoles might belong to the putative ground ecomorph, we repeated our ecomorph classification analyses with the addition of the ground ecomorph. In cases where previously classified mainland species were assigned to different ecomorphs in the analyses with and without the ground ecomorph, we incorporated their latter assignment into the final classification. However, some species were classified into a Caribbean ecomorph without the ground ecomorph but were not classified when it was included. This generally occurred because the DFA support for those species’ original assignments fell below the 0.90 threshold, causing them to be unclassified. Because these species were assigned to a Caribbean ecomorph before the addition of the ground ecomorph class, it is likely they still belong to an ecomorph (either their original ecomorph or the ground ecomorph). Therefore, we assessed the newly unclassified species based on their DFA posterior probabilities and their ecomorph centroid distances. If a newly unclassified species had a DFA posterior probability ≥ 0.75 for its original ecomorph assignment and was closest to the centroid of that ecomorph, it retained its original assignment. Conversely, if a species had a DFA posterior probability ≥ 0.75 for the ground ecomorph and was closest to the centroid of the ground ecomorph, it was reassigned to it.

Reconstructing the evolution of mainland ecomorphs and their convergence

We reconstructed ecomorph evolution across the Caribbean and mainland Draconura-clade anoles to investigate how many times each ecomorph has evolved on the mainland. To estimate the frequency of ecomorph transitions, we used discrete stochastic character mapping via SIMMAP (Bollback, 2006). Two separate SIMMAP analyses were performed: one with just the six Caribbean ecomorphs, and another with the addition of the ground ecomorph. Both analyses included all species in our morphometric data set. We scored the Caribbean ecomorph species with their respective assignments in both analyses and scored the a priori ground ecomorph species as such in the second SIMMAP analysis. The Draconura species and the previously unclassified Caribbean species were scored with their final morphology-based ecomorph classifications (when applicable). Species we considered to be intermediate between ecomorphs were assigned prior probabilities equivalent to their DFA posterior probabilities for each ecomorph to which they were assigned. All species not assigned to an ecomorph were treated as non-ecomorph species.

We conducted our SIMMAP analyses using the phytools package (Revell, 2012), with the symmetrical-rates model of character state transitions. We determined that this model fitted our data better than the equal-rates and all-rates-different models using likelihood-ratio tests. To account for uncertainty in the estimation of branch lengths and topology, we performed the SIMMAP analyses on a set of 100 randomly sampled post-burn-in trees from Poe et al. (2017), with 1000 simulations per tree. To estimate the number of independent ecomorph transitions among only the mainland species, we pruned the SIMMAP results to include only the mainland members of the Draconura clade using phytools (Revell, 2012). We considered a posterior probability > 0.95 for at least one transition and for more than one transition as strong support for a separate mainland origin and multiple independent mainland origins of a given ecomorph, respectively.

To evaluate the degree of morphological convergence among species belonging to the same ecomorph, we used a distance-based measure, C1, described by Stayton (2015). The C1 metric measures the degree of convergence between two taxa as how similar they are in the morphospace relative to the largest reconstructed distance between their ancestors. C1 values range from zero to one, where values closer to one indicate stronger convergence. We calculated C1 values using the scores from the five retained pPC axes for multiple sets of putatively convergent taxa, each representing a different ecomorph. To assess the significance of the detected convergence, the calculated C1 values were compared against a null distribution of C1 values calculated from 1000 sets of trait values simulated across the tree under Brownian Motion (BM).

First, we assessed the strength of convergence among just the a priori Caribbean and ground ecomorph members. Then we evaluated the strength of convergence among the a priori members and all species assigned to each ecomorph in our final classifications. For this second set of analyses, C1 was calculated once including species classified with only the six Caribbean ecomorphs and once with the species classified with the inclusion of the ground ecomorph. All species with intermediate classifications were treated as non-convergent taxa given the uncertainty in their classifications. Finally, we evaluated the strength of convergence among mainland Draconura-clade species assigned to the same ecomorph by calculating C1 for all the ecomorphs for which the SIMMAP analyses indicated strong support for more than one independent transition. Again, we calculated C1 once for the classifications made with and without the addition of the ground ecomorph and did not include species with intermediate classifications.

Investigating faunal convergence between island and mainland radiations

We visualized morphological convergence across the phylogeny between Caribbean and mainland Draconura-clade anoles using a tanglegram. First, we performed another hierarchical cluster analysis with all species in our morphological data set. Then we compared the topologies of the MCC phylogeny and a dendrogram showing the morphological clusters using the tanglegram, produced with the “cophylo” function in phytools (Revell, 2012). Lines were drawn to show the location of species in both diagrams and indicate cases of convergence across regions.

To investigate whether the radiation of mainland Draconura anoles is convergent with the Greater Antillean radiations, we tested for species-for-species matching between the different regions by adopting the approach by Mahler et al. (2013). For each Greater Antillean and mainland Draconura species (excluding any island endemics), we calculated the Euclidean distance between it and its nearest neighbours (NND) in the morphospace from each of the other four regions (using pPC scores from the five retained axes). To quantify the average faunal similarity among regions with a single metric, we first averaged the calculated NND values for all species occurring in the same region, and then averaged those five values together. We compared the resulting mean against a null distribution of mean distances calculated from 1000 simulated morphospaces. To simulate morphospaces, we fitted the first five pPC axes to a BM model of evolution and simulated trait data using empirically-estimated rate parameters (σ 2). For each morphospace, the pPC axes were simulated independently and then combined. We then calculated the mean NND values using the same methods for each morphospace. A significantly lower empirical mean NND value (alpha level = 0.05) would provide evidence that the Greater Antillean and mainland Draconura fauna are more similar than expected by chance.

However, strong similarities among the Greater Antillean radiations may confound our results. Thus, we also isolated the average similarity between the mainland Draconura clade and the four different Caribbean radiations (by averaging all of the calculated NND values between each mainland species and its Greater Antillean neighbours from each island), as well as the average similarity between the Greater Antillean radiations and the mainland Draconura radiation (by averaging all of the calculated NND values between Greater Antillean species and their nearest mainland neighbours). We compared these additional metrics against two null distributions of mean distances calculated from the simulated morphospaces.

We performed a final test of morphological convergence between island and mainland anoles without a priori ecomorph assignments. For this, we used the l1ou method and corresponding R package (Khabbazian et al., 2016) to detect the number and location of adaptive evolutionary shifts in morphology. This method uses the Ornstein–Uhlenbeck (OU) process to model changes in multivariate phenotypic data over time and over lineages and selects a shift configuration using the LASSO (least absolute shrinkage and selection operator) method (Khabbazian et al., 2016). The l1ou method also uses a non-parametric bootstrap procedure to quantify the support for a given shift as well as determine which regimes are converging towards the same adaptive optima (Khabbazian et al., 2016). We used the Akaike information criterion (AIC) to evaluate models because AIC performed best in identifying regimes associated with the Caribbean ecomorphs in an empirical study (Khabbazian et al., 2016). Our analyses were performed using the retained pPC scores for all species in our study. If island and mainland anoles are evolving under similar selective pressures, we would expect to find convergent regimes.

Data availability

The full data set used in this study and the R scripts used to analyse it have been archived in GitHub: https://github.com/jmhuie/Mainland-Anole-Ecomorphology.

RESULTS

Mainland and Caribbean anoles in the morphospace

A large part of the morphological space occupied by the mainland Draconura clade overlapped with that occupied by the Caribbean ecomorphs (Fig. 3; Supporting Information, Table S2). However, while some Draconura species resembled the large crown-giants and others the short-legged twig anoles, they did not achieve the same extremes in the morphospace as the Caribbean exemplars of those ecomorphs (Fig. 3A). Likewise, mainland species did not have tails as long as those of the most extreme Caribbean grass-bush anoles (Fig. 3C). Conversely, several mainland anoles occupied regions of the morphospace not occupied by any Caribbean ecomorph species. For example, many Draconura species had longer limbs than those of the Caribbean ecomorph species but resembled several of the previously unclassified Caribbean anoles. Mainland anoles also exhibited a wider range of finger- and toe-pad sizes than Caribbean ecomorph species (Fig. 3B). They exhibited similar maximum pad widths; however, the pad-less A. onca and a few other species had narrower digits/pads than any of the Caribbean ecomorph species. Draconura species also did not achieve head lengths as great as those of some Caribbean species, particularly ones representing the crown-giant, grass-bush, trunk-crown and twig ecomorphs (Fig. 3D).

Relative positions of 205 species of Caribbean and Draconura anoles in a morphological space. A, pPC2 vs. pPC1; (B) pPC3 vs. pPC1; (C) pPC3 vs. pPC4; (D) pPC5 vs. pPC4. Silhouettes of lizards or their parts represent the predominant trait variation on each pPC axis (body size, limb length, finger- and toe-pad width, tail length and head shape). See Supporting Information (Table S2) for the detailed trait loadings for each pPC axis. Convex hulls were drawn around the Caribbean ecomorphs. Grey triangles indicate mainland Draconura species, grey squares indicate island Draconura species and coloured circles indicate Caribbean species. CG (blue) = Crown-Giant; GB (yellow) = Grass-Bush; T (red) = Trunk; TC (green) = Trunk-Crown; Tw (purple) = Twig; U (orange) = previously unclassified Caribbean species.

Support for the Caribbean ecomorphs

The randomization tests indicated that the centroid distances between all Caribbean ecomorph pairs were significant at the P < 0.001 level. The hierarchical cluster analysis indicated that most of the ecomorphs formed single clusters, although the trunk-crown species formed two (Supporting Information, Fig. S2). The discriminant function analysis revealed significant differences among ecomorphs (MANOVA: Wilks’ λ < 0.001, F = 17.160, P < 0.001) and successfully classified 98.6% and 94.4% of the Caribbean ecomorph species to their respective classes when using re-substitution and cross-validation, respectively (Supporting Information, Result S1).

Classification of mainland anoles into the six Caribbean ecomorphs

Out of the 99 Draconura species, the DFA classified 70 mainland and eight island species into ecomorphs with a posterior probability of 0.90 or greater based on morphological characters (Supporting Information, Table S4). The DFA also classified 28 out of 35 previously unclassified Caribbean species (Supporting Information, Table S4). Ecomorph assignments for Draconura species were as follows: eight grass-bush, five trunk-crown, 63 trunk-ground and two twig anoles (Supporting Information, Table S4). The ecomorph assignments for the previously unclassified Caribbean species were as follows: three crown-giant, four grass-bush, two trunk, four trunk-crown, 13 trunk-ground and two twig anoles (Supporting Information, Table S4).

The Euclidean distance criteria varied in the number of species they classified, but all assigned at least one mainland Draconura species to each Caribbean ecomorph (Supporting Information, Tables S5–S7). Seventy-seven species (55 mainland Draconura, five island Draconura and 17 unclassified Caribbean species) were as close or closer to the centroid of at least one ecomorph than was the furthest member of that ecomorph (Supporting Information, Table S5). Eighty-five species (58 mainland Draconura, five island Draconura and 22 unclassified Caribbean species) were assigned to at least one ecomorph based on having a mean pairwise distance to the members of an ecomorph that was less than or equal to the largest MPD among members of that ecomorph (Supporting Information, Table S6). Eighty-two species (56 mainland Draconura, five island Draconura and 21 unclassified Caribbean species) were assigned to at least one ecomorph based on having a distance to a neighbouring ecomorph species that was less than the largest NND among the members of that ecomorph (Supporting Information, Table S7).

After evaluating the DFA and Euclidean distance classifications to produce final ecomorph assignments based on morphology, 61 species (39 mainland Draconura, five island Draconura and 17 unclassified Caribbean species) showed strong morphological support for belonging to a Caribbean ecomorph (Table 1). An additional 14 mainland Draconura species and five previously unclassified Caribbean species were considered to be intermediate between two ecomorphs. Most of these intermediate species were between the grass-bush and trunk-ground ecomorphs, with some between the crown-giant and three other ecomorphs. Among the Draconura anoles assigned to a single ecomorph, there were eight grass-bush, three trunk-crown, 31 trunk-ground and two twig anoles. The previously unclassified Caribbean species were assigned as one crown-giant, three grass-bush, one trunk, 11 trunk-ground and one twig anole. Except for the Caribbean species A. barbouri, none of our a priori ground anoles was assigned to an ecomorph in this analysis.

Table 1.

List of the 80 Draconura and previously unclassified Caribbean anole species that were assigned to one or two different ecomorph classes when only using the Caribbean ecomorphs. Also shown are the DFA posterior probabilities and the Euclidean distances for each of the classification criterion that each species satisfied for its predicted ecomorph. For intermediate species, where we adopted different CD and NND (and no MPD) criteria, all relevant Euclidean distances are shown for their predicted ecomorphs. Although intermediate species did not need to satisfy the same CD and NND criteria as those assigned to a single ecomorph, asterisks (*) indicate instances where they did.

Species . Region§ . Predicted ecomorph† . DFA . CD . MPD . NND . A. altavelensis C T 1.00 0.25 0.32 0.19 A. amplisquamosus M GB 0.94 0.45 0.54 - A. apletophallus M TG 1.00 0.37 0.43 - A. argenteolus C T / TG 0.78 / 0.22 0.37* / 0.36* NA - / 0.25* A. armouri C TG 0.96 - - 0.24 A. auratus M GB 1.00 0.39 0.50 0.23 A. barbatus C Tw / CG 0.56 / 0.44 1.19 / 0.96 NA 0.74 / - A. barbouri C GB 1.00 0.33 0.45 0.19 A. barkeri M TG 1.00 0.35 0.41 - A. benedikti M TG 1.00 0.28 0.35 0.19 A. bicaorum I TG 1.00 0.31 0.38 0.26 A. biporcatus M CG / TG 0.63 / 0.37 0.57 / 0.60 NA 0.44 / - A. boulengerianus M TG 0.98 0.26 0.34 0.19 A. charlesmyersi M Tw 0.97 0.50 0.62 0.28 A. concolor I TG 1.00 0.24 0.33 0.22 A. conspersus C TG 0.99 0.19 0.30 0.20 A. cristifer M Tw 1.00 0.65 0.75 0.28 A. cusuco M TC 0.99 - 0.49 0.16 A. desechensis C TG 0.92 0.36 0.42 0.20 A. dollfusianus M TG / GB 0.80 / 0.20 0.51 / 0.47* NA - / 0.30* A. dunni M TG 1.00 0.23 0.31 0.10 A. ernestwilliamsi C TG 1.00 0.35 0.42 0.22 A. fuscoauratus M GB 0.99 0.37 0.48 0.25 A. gadovii M TG 1.00 0.30 0.37 0.26 A. gaigei M TG / GB 0.72 / 0.28 0.48 / 0.48* NA - / 0.37 A. imias C TG 0.98 0.36 0.42 0.25 A. isolepis C Tw 1.00 0.49 0.60 0.18 A. johnmeyeri M TG 0.98 0.16 0.28 0.17 A. kemptoni M TC 1.00 0.35 0.43 0.21 A. lemurinus M TG 1.00 - 0.43 - A. liogaster M TG 0.99 - 0.43 0.18 A. longitibialis C TG 1.00 - 0.43 - A. loveridgei M CG / TG 0.78 / 0.22 0.35* / 0.70 NA 0.23* / - A. lucius C TG / T 0.79 / 0.15 0.48 / 0.49 NA 0.32* / - A. lyra M TG 1.00 0.36 0.42 0.26 A. macrolepis M TG 1.00 0.36 0.42 0.23 A. magnaphallus M TG 1.00 0.27 0.35 0.20 A. mariarum M TG 0.96 - - 0.26 A. matudai M TG 1.00 0.17 0.28 0.06 A. mccraniei M TG 1.00 0.33 0.40 - A. meridionalis M GB 1.00 0.50 0.59 - A. milleri M TG 1.00 - - 0.26 A. monensis C TG / T 0.78 / 0.20 0.24* / 0.33* NA 0.18* / - A. morazani M GB 1.00 0.47 0.55 0.21 A. muralla M GB 0.99 0.44 0.53 0.23 A. nebulosus M TG 0.95 0.26 0.34 0.17 A. notopholis M TG / GB 0.83 / 0.17 0.49 / 0.69 NA 0.41 / - A. ortonii M T / TC 0.52 / 0.48 0.26* / 0.46 NA - / 0.19* A. osa M TG 1.00 0.34 0.41 0.27 A. petersii M TC / CG 0.80 / 0.17 0.61 / 0.55 NA - / 0.34 A. pijolense M TG 1.00 0.23 0.32 0.21 A. pinchoti I TG 1.00 0.27 0.35 0.24 A. poecilopus M TG 1.00 0.21 0.31 0.17 A. pogus C GB 1.00 0.46 0.54 0.29 A. purpurgularis M TG 0.99 0.25 0.33 0.23 A. reconditus C TG 1.00 - - 0.21 A. rivalis M TG 1.00 0.29 0.36 0.17 A. roatanensis I TG 1.00 0.24 0.33 0.16 A. rodriguezii M GB / TG 0.78 / 0.22 0.43* / 0.49 NA 0.28* / - A. rubribarbaris M TC 0.98 0.40 0.47 0.25 A. rupinae C GB / TG 0.75 / 0.25 0.79 / 0.59 NA - / 0.46 A. salvini M Tw / TC 0.65 / 0.35 0.54* / 0.54 NA 0.34* / - A. schwartzi C GB 0.96 - 0.64 - A. scriptus C TG 1.00 - - 0.25 A. serranoi M TG / GB 0.78 / 0.22 0.31* / 0.63 NA 0.23* / - A. shrevei C TG 1.00 0.21 0.31 0.15 A. sminthus M GB / TG 0.89 / 0.09 0.46* / 0.32* NA 0.13* / - A. strahmi C TG 1.00 - 0.43 0.24 A. subocularis M TG 1.00 0.37 0.43 - A. taylori M TG 1.00 0.25 0.33 0.19 A. tolimensis M TG 0.92 0.31 0.38 0.20 A. townsendi I TG 1.00 0.27 0.35 0.23 A. tropidolepis M TG 0.98 0.36 0.42 - A. unilobatus M GB 0.90 0.33 0.45 0.31 A. vermiculatus C CG 1.00 0.38 0.45 0.12 A. vittigerus M GB / TG 0.50 / 0.50 0.58 / 0.42 NA - / 0.31* A. wattsi C TG 1.00 0.26 0.34 0.17 A. wellbornae M TG / GB 0.75 / 0.22 0.58 / 0.36* NA - / 0.31* A. wermuthi M GB 0.99 0.53 0.60 0.16 A. zeus M GB / TG 0.66 / 0.34 0.47* / 0.55 NA 0.34 / -

Table 1.

Evaluation of a new ground ecomorph

The ground anoles exhibited morphologies that distinguished them from each of the Caribbean ecomorphs (Fig. 4). In general, the ground anoles had longer forelimbs and hindlimbs than the Caribbean ecomorphs, but their limbs were only slightly longer than those of the grass-bush, trunk and trunk-ground anoles (Fig. 4A; Supporting Information, Fig. S1). They also had narrower subdigital pads than the Caribbean ecomorphs but were only slightly narrower than those of the grass-bush anoles (Fig. 4B; Supporting Information, Fig. S1). The ground anoles had comparable body sizes (Fig. 4A; Supporting Information, Fig. S1) and tail lengths to most Caribbean ecomorph species (Fig. 4C; Supporting Information, Fig. S1) but were smaller than the crown-giant species and had shorter tails than the grass-bush species. The ground anoles also generally had shorter heads and snouts than most of the Caribbean ecomorph species, although these features were similar to those of the trunk and some of the trunk-ground species (Fig. 4D; Supporting Information, Fig. S1). The randomization tests indicated that the ground ecomorph was morphologically different from each Caribbean ecomorph (P < 0.001 in all cases). However, hierarchical cluster analysis indicated that the Draconura ground species formed two clusters and A. barbouri clustered with the grass-bush ecomorph (Supporting Information, Fig. S2).

Relative positions of the six Caribbean ecomorphs and the putative ground ecomorph in a morphological space. A, pPC2 vs. pPC1; (B) pPC3 vs. pPC1; (C) pPC3 vs. pPC4; (D) pPC5 vs. pPC4. Silhouettes of lizards or their parts represent the predominant trait variation on each pPC axis (body size, limb length, finger- and toe-pad width, tail length and head shape). See Supporting Information (Table S2) for the detailed trait loadings for each pPC axis. Convex hulls were drawn around each ecomorph. Triangles indicate mainland Draconura species and circles indicate Caribbean species. Abbreviations and colours as in Figure 3 (when applicable), but with the addition of G (cyan) = Ground.

A DFA that included the putative ground ecomorph class revealed significant differences among ecomorphs (MANOVA: Wilks’ λ < 0.001, F = 14.84, P < 0.001) and successfully classified 98.8% and 93.9% of the Caribbean and ground ecomorph species to their respective ecomorphs when using re-substitution and cross-validation, respectively (Supporting Information, Result S1). The DFA classified 78 species (44 mainland Draconura, six island Draconura and 28 unclassified Caribbean anoles) into ecomorphs with a posterior probability of 0.90 or greater (Supporting Information, Table S8). The classifications for Draconura species were as follows: 22 ground, two grass-bush, one trunk, four trunk-crown, 18 trunk-ground and three twig anoles (Supporting Information, Table S8). The classifications for previously unclassified Caribbean species were as follows: two ground, three crown-giants, two grass-bush, three trunk, three trunk-crown, 12 trunk-ground and three twig anoles (Supporting Information, Table S8).

Considering the Draconura and unclassified Caribbean species that satisfied Euclidean distance criteria for the ground ecomorph (Supporting Information, Tables S5–S7), 29 species (25 mainland Draconura, two island Draconura and two unclassified Caribbean species) were closer to the ground ecomorph centroid than was the furthest member of that ecomorph (Supporting Information, Table S5). Of the Draconura species, 13 satisfied this criterion for only the ground ecomorph; the remaining 14 also satisfied the CD criterion for the grass-bush and/or trunk-ground ecomorphs (Supporting Information, Table S5). Twenty-four species (20 mainland Draconura, two island Draconura and two unclassified Caribbean species) had a mean pairwise distance to all of the ground ecomorph species less than the largest MPD among the members of that ecomorph (Supporting Information, Table S6). Of the Draconura species, 12 satisfied the MPD criterion for only the ground ecomorph, while the remaining 10 also satisfied it for the grass-bush and/or trunk-ground ecomorph (Supporting Information, Table S6). Twenty-eight species (22 mainland Draconura, three island Draconura and three unclassified Caribbean species) were closer to a ground ecomorph species than the largest nearest-neighbour distance among the members of that ecomorph (Supporting Information, Table S7). Of the Draconura species, 15 satisfied the NND criterion for only the ground ecomorph, while the remaining 10 also satisfied this criterion for the grass-bush and/or trunk-ground ecomorphs (Supporting Information, Table S7).

After evaluating the morphology-based DFA and Euclidean distance ecomorph assignments with the inclusion of the ground ecomorph, we classified 59 species (36 mainland Draconura, five island Draconura and 18 unclassified Caribbean anoles) with strong morphological support (Table 2). An additional 13 Draconura species (seven of which were previously assigned to an ecomorph before the addition of the ground ecomorph) and two previously unclassified Caribbean species were assigned to two ecomorphs as potential intermediate species. The majority of the intermediate species were between the ground and trunk-ground ecomorphs. Among the mainland species assigned to a single ecomorph, there were 22 ground, two grass-bush, one trunk, four trunk-crown, 18 trunk-ground and three twig anoles. Of the 22 species of Draconura ground anoles, 16 were not previously classified into a Caribbean ecomorph, while six species were previously classified as trunk-ground anoles and reassigned to the ground ecomorph (A. lemurinus, A. meridionalis, A. milleri, A. subocularis and A. tropidolepis). Most other ecomorph assignments remained consistent with the previous assignments made without the ground ecomorph, except for the addition of one mainland trunk anole (Anolis ortonii) and one mainland twig anole (Anolis salvini). Meanwhile, the previously unclassified Caribbean species were classified as two ground (Anolis rupinae and Anolis etheridgei), one crown-giant, one grass-bush, two trunk, 11 trunk-ground and one twig anole.

Table 2.

List of the 76 Draconura and previously unclassified Caribbean anole species that were assigned to one or two different ecomorph classes with the inclusion of the ground ecomorph. Also shown are the DFA posterior probabilities and the Euclidean distances for each of the classification criterion that each species satisfied for its predicted ecomorph. For intermediate species, where we adopted different CD and NND (and no MPD) criteria, all relevant Euclidean distances are shown for their predicted ecomorphs. Although intermediate species did not need to satisfy the same CD and NND criteria as those assigned to a single ecomorph, asterisks (*) indicate instances where they did.

Species . Region . Predicted ecomorph† . DFA . CD . MPD . NND . A. altavelensis C T 1.00 0.25 0.32 0.19 A. aquaticus M G / TG 0.63 / 0.37 0.36* / 0.41 NA - / 0.29* A. argenteolus C T 0.91 0.37 0.40 0.27 A. armouri C TG 0.99 - - 0.24 A. auratus M GB 0.99 0.39 0.50 0.23 A. barkeri M TG / G 0.60 / 0.40 0.35* / 0.43* NA 0.28* / - A. benedikti M TG 0.92 0.28 0.35 0.19 A. bicaorum M G / TG 0.73 / 0.27 0.36* / 0.31* NA 0.24* / - A. biporcatus M TG / CG 0.84 / 0.16 0.60 / 0.57 NA - / 0.44 A. boulengerianus M G / TG 0.87 / 0.12 0.33* / 0.26* NA - / 0.19* A. charlesmyersi M Tw 1.00 0.50 0.62 0.28 A. christophei C G / TG 0.56 / 0.38 0.27* / 0.26* NA - / 0.21* A. cobanensis M G / TG 0.82 / 0.10 0.43* / 0.41 NA - / 0.32* A. concolor I TG 0.99 0.24 0.33 0.22 A. conspersus C TG 0.97 0.19 0.30 0.20 A. cristifer M Tw 1.00 0.65 0.75 0.28 A. cryptolimifrons M G 0.99 0.31 0.45 0.18 A. cupreus M G 0.98 0.39 0.50 0.13 A. cusuco M TC / T 0.57 / 0.43 0.44 / 0.40 NA 0.16* / - A. desechensis C TG 0.94 0.36 0.42 0.20 A. dollfusianus M G 0.99 0.40 0.50 0.15 A. dunni M TG 0.99 0.23 0.31 0.10 A. ernestwilliamsi C TG 0.99 0.35 0.42 0.22 A. etheridgei C G 1.00 0.42 0.54 0.33 A. fuscoauratus M GB / G 0.57 / 0.42 0.37* / 0.57 NA 0.25* / - A. gaigei M G 1.00 0.23 0.41 0.27 A. gracilipes M G 1.00 0.28 0.42 0.16 A. granuliceps M G 1.00 0.29 0.45 0.30 A. imias C TG 0.98 0.36 0.42 0.25 A. isolepis C Tw 1.00 0.49 0.60 0.18 A. johnmeyeri M TG 0.97 0.16 0.28 0.17 A. kemptoni M TC 0.99 0.35 0.43 0.21 A. lemurinus M G 0.99 0.22 0.41 0.22 A. liogaster M TG 1.00 - 0.43 0.18 A. longitibialis C TG 1.00 - 0.43 - A. loveridgei M CG / TG 0.51 / 0.49 0.35* / 0.70 NA 0.23* / - A. lynchi M G 0.93 0.38 0.52 - A. lyra M TG 0.98 0.36 0.42 0.26 A. maculiventris M G / TG 0.64 / 0.36 0.42* / 0.57 NA 0.25* / - A. mariarum M TG 0.97 - - 0.26 A. matudai M TG 0.99 0.17 0.28 0.06 A. medemi I G 1.00 0.25 0.43 0.18 A. milleri M G 0.97 0.35 0.49 - A. monensis C TG / T 0.88 / 0.10 0.24* / 0.33* NA 0.18* / - A. nebulosus M TG 0.94 0.26 0.34 0.17 A. notopholis M G 1.00 0.22 0.41 0.27 A. ortonii M T 0.98 0.26 0.33 0.24 A. osa M G 0.97 0.18 0.39 0.23 A. oxylophus M TG / G 0.90 / 0.10 0.45 / 0.47 NA 0.34* / - A. pijolense M TG 0.99 0.23 0.32 0.21 A. pinchoti I TG 0.99 0.27 0.35 0.24 A. poecilopus M TG 0.96 0.21 0.31 0.17 A. pogus C GB 1.00 0.46 0.54 0.29 A. reconditus C TG 1.00 - - 0.21 A. roatanensis M TG / G 0.64 / 0.36 0.24* / 0.34* NA 0.16* / - A. rodriguezii M G 1.00 0.36 0.47 0.19 A. rubribarbaris M TC 1.00 0.40 0.47 0.25 A. rupinae C G 0.98 0.36 0.48 0.20 A. salvini M Tw 0.91 0.54 0.64 0.34 A. scriptus C TG 1.00 - - 0.25 A. shrevei C TG 0.98 0.21 0.31 0.15 A. strahmi C TG 1.00 - 0.43 0.24 A. subocularis M G 0.96 0.29 0.44 0.24 A. taylori M TG 0.99 0.25 0.33 0.19 A. tolimensis M G / TG 0.51 / 0.48 0.45 / 0.31* NA - / 0.20* A. townsendi I TG 0.93 0.27 0.35 0.23 A. tropidolepis M G 0.99 0.19 0.40 0.31 A. vermiculatus C CG 0.99 0.38 0.45 0.12 A. villai I G 1.00 0.28 0.44 0.23 A. vittigerus M G / TG 0.68 / 0.23 0.49 / 0.42 NA - / 0.31* A. wampuensis M G 0.99 0.15 0.38 0.27 A. wattsi C TG 0.97 0.26 0.34 0.17 A. wermuthi M GB 0.91 0.53 0.60 0.16 A. woodi M G / TG 0.56 / 0.44 0.49 / 0.41 NA - / 0.29* A. yoroensis M G 0.94 0.39 0.50 0.24 A. zeus M G 0.95 0.37 0.49 0.26

Table 2.

Repeated ecomorph evolution on the mainland

The stochastic character mapping analyses indicated that most of the ecomorphs arose within the Draconura clade in Central and South America independently from the Caribbean (Greater Antillean) anoles with a posterior probability greater than 0.98 (Table 3). The exceptions were the crown-giant and trunk ecomorphs, which showed weak support for a mainland origin because no mainland Draconura species were classified to those ecomorphs unambiguously (intermediate species only). However, the trunk ecomorph showed strong support for a mainland origin with the inclusion of the ground ecomorph, when A. ortonii was confidently assigned to the trunk ecomorph. For all other ecomorphs that did evolve independently on the mainland, both SIMMAP analyses performed both with and without the ground ecomorph found strong support for multiple origins of those ecomorphs on the mainland (Fig. 5; Table 3).

Table 3.

Posterior probabilities of a mainland origin and multiple independent transitions for each ecomorph within the Draconura clade based on two separate stochastic character mapping analyses performed on 100 post-burn-in trees with 1000 simulations per tree. Bold values indicate posterior probabilities that show strong support (≥ 0.95).

. Caribbean ecomorphs only . With ground ecomorph . Ecomorph . Probability of a mainland origin . Probability of > 1 independent transitions . Best-supported number of transitions (probability) . Probability of a mainland origin . Probability of > 1 independent transitions . Best-supported number of transitions (probability) . Crown-Giant 0.463 0.119 0 (0.537) 0.173 0.016 0 (0.827) Grass-Bush 1.000 1.000 8 (0.204) 1.000 1.000 4 (0.295) Trunk 0.152 0.003 0 (0.848) 1.000 0.376 1 (0.623) Trunk-Crown 1.000 1.000 6 (0.345) 1.000 1.000 3 (0.509) Trunk-Ground 0.986 0.939 4 (0.113) 1.000 1.000 13 (0.178) Twig 1.000 0.810 2 (0.378) 1.000 0.993 3 (0.396) Ground - - - 1.000 1.000 7 (0.170)

Table 3.

The MCC phylogeny showing the ancestral state reconstructions of ecomorph evolution among mainland Draconura-clade anoles based on the final classifications (including intermediates) made with only the six Caribbean ecomorphs (left) and with the addition of the ground ecomorph (right). Tip circles show the final ecomorph classifications or (as pies) the DFA probabilities for the intermediate ecomorph species. Node pies reflect the posterior probabilities of each ecomorph summarized from the two stochastic character mapping analyses performed on 100 post-burn-in trees. Terminal branches are coloured by the ecomorph with the highest tip probability.

The C1 convergence analyses indicated that the Caribbean and ground ecomorphs varied between 0.139 and 0.09 in their degree of morphological convergence (Table 4). All of the a priori ecomorphs exhibited significantly stronger morphological convergence than expected compared to a null distribution of C1 values (Table 4). Convergence was also found among the Draconura and previously unclassified Caribbean species that were assigned to an ecomorph in our final morphology-based classifications, both with and without the ground ecomorph (Table 4). Within the mainland Draconura clade, species classified as grass-bush and trunk-ground before the inclusion of the ground ecomorph exhibited significant morphological convergence (Table 4). With the addition of the ground ecomorph, mainland species assigned to the ground, trunk-ground and twig ecomorphs exhibited significant morphological convergence (Table 4).

Table 4.

C1 values showing the estimated degrees of convergence for each ecomorph with different sets of putatively convergent taxa. There were no mainland Draconura species assigned unambiguously to the crown-giant ecomorph (either with or without the inclusion of the ground ecomorph) and only one mainland species assigned to the trunk ecomorph, thus C1 was not calculated for these ecomorphs when assessing convergence among only the classified mainland Draconura species. Bold values represent statistically significant amounts of morphological convergence (P < 0.05).

. . A priori species and final classifications . Final mainland classifications only . Ecomorph . A priori ecomorphs species . Caribbean ecomorphs only . With ground ecomorph . Caribbean ecomorphs only . With ground ecomorph . Crown-Giant 0.409 0.425 0.425 - - Grass-Bush 0.334 0.272 0.307 0.164 0.072 Trunk 0.156 0.140 0.215 - - Trunk-Crown 0.248 0.256 0.259 0.222 0.213 Trunk-Ground 0.180 0.173 0.191 0.153 0.200 Twig 0.381 0.389 0.395 0.043 0.437 Ground 0.139 - 0.205 - 0.160

Table 4.

Faunal convergence between island and mainland anole radiations

The tanglegram indicated several instances of morphological convergence between Caribbean and mainland Draconura-clade anoles (Fig. 6). Morphological clusters generally corresponded to the different Caribbean ecomorphs and several mainland Draconura species clustered with each of them. However, the tanglegram also revealed a large cluster of mostly Draconura species with only a few instances of previously unclassified Caribbean species converging on those morphologies. Many Draconura species within this cluster also exhibited a mismatch between phylogeny and morphology, suggesting convergence within the Draconura clade. All a priori ground ecomorph species, except for A. barbouri, fell within this predominantly mainland Draconura cluster.

A tanglegram with the MCC phylogeny (left) and a dendrogram showing morphological clusters inferred from a hierarchical cluster analysis (right). Lines are drawn to indicate the location of each species in both diagrams. Lines that originate from areas of the phylogeny corresponding to Caribbean and mainland anoles but terminate in the same cluster in the dendrogram indicate convergence between the two geographic regions. The branches and tips of the phylogeny are coloured to indicate whether a species is from the Caribbean (blue) or the mainland Draconura clade (red). Some dendrogram branches are coloured to indicate which clusters generally correspond to clusters of a priori ecomorph species (black represents a neutral branch colour) and the tips are coloured to indicate our final classifications made with the inclusion of the ground ecomorph. The arrow indicates a morphological cluster composed of mostly Draconura species.

The tests for morphological similarity among the mainland Draconura radiation and each of the Greater Antillean radiations found moderate to strong support for species-for-species matching across regions. Analyses performed with the MCC phylogeny did not find significant faunal similarity across regions (mean NND = 0.37, p = 0.09). However, analyses performed with the time-calibrated tree did (mean NND = 0.37, p = 0.014). Moreover, the isolated island vs. mainland comparisons showed that mainland Draconura species were, on average, more similar to species on each of the Greater Antillean islands than expected by chance (mean NND = 0.376; p = 0.044). However, the Greater Antillean fauna was not more similar to the mainland Draconura species than expected by chance (mean NND = 0.312, p = 0.669).

Adaptive shifts were common across the Caribbean and within the Draconura clade (Supporting Information, Fig. S3). The l1ou analyses using the MCC phylogeny detected 31 independent shifts, all with moderate to strong bootstrap support (> 55%). Of the 16 adaptive peaks, 11 were occupied by convergent lineages, of which three were shared between the Caribbean and the mainland. One of the convergent peaks corresponded to Caribbean twig anoles of the Anolis angusticeps group and the mainland Anolispentaprion group (Supporting Information, Fig. S3, Regime 7). Another peak was occupied by Caribbean grass-bush anoles from the Anolis olssoni clade and the mainland species Anolis auratus and Anolis meridionalis (Supporting Information, Fig. S3, Regime 11). The third shared peak between the Caribbean and the mainland was occupied by the previously unclassified Caribbean anole Anolis etheridgei and several Draconura-clade anoles that we assigned to the ground or trunk-ground ecomorphs (Supporting Information, Fig. S3, Regime 15). The remaining peaks in the Caribbean broadly corresponded to the Caribbean ecomorphs, and one exclusively mainland peak (Regime 14) was occupied by species assigned to the ground ecomorph both a priori and a posteriori. The analyses using the time-calibrated tree found fewer shifts and only some of the same regimes detected using the MCC tree. However, the time-calibrated tree analyses still found evidence for convergent regimes between the Caribbean and the mainland (Supporting Information, Fig. S4).

DISCUSSION

We found multiple lines of evidence suggesting that the mainland Draconura and Caribbean anole radiations exhibit convergent patterns of phenotypic diversification. Our morphological classifications provided strong evidence for several representatives of all six Caribbean ecomorphs on the mainland (Tables 1, 2; Fig. 1). Many of these species also occupy similar structural microhabitats as the Caribbean members of their corresponding ecomorphs (Supporting Information, Table S9; see below). These findings differ from previous studies that suggested the Caribbean ecomorphs are either absent or uncommon in Central and South America (Irschick et al., 1997; Schaad & Poe, 2010), which led to the idea that different environmental factors are driving distinct evolutionary outcomes in island and mainland anoles (Irschick et al., 1997; Velasco & Herrell, 2007; Pinto et al., 2008; Schaad & Poe, 2010). By contrast, our results indicate that the mainland Draconura radiation generally resembles the Greater Antillean fauna, with species from both regions converging on similar morphologies and adaptive peaks (Fig. 6; Supporting Information, Fig. S3). Furthermore, many of the ecomorphs evolved more than once within the Draconura clade (also see Poe & Anderson, 2019), signifying that selection related to microhabitat specialization has led to convergent morphologies in separate mainland anole lineages.

Many mainland Draconura species resembled one or two ecomorphs but not closely enough to be classified by our methods, suggesting that they may be less specialized or intermediate ecomorph species. Likewise, Caribbean ecomorph species also vary in their degree of ecological and morphological specializations. Because ecomorphological specialization is generally continuous, it is difficult to categorize intermediate species (Grizante et al., 2010; Arbuckle et al., 2014; Huie et al., 2020), as the misclassifications of a few Caribbean ecomorph species by the DFA and cluster analysis demonstrate (Fig. 6; Supporting Information, Fig. S2). Recent studies, as well as this one, have suggested there is a mix of similar and distinct ecotypes and morphotypes on the mainland relative to the Caribbean (Poe & Anderson, 2019; Moreno-Arias et al., 2020). Although some intermediate species might represent less specialized members of the currently recognized ecomorphs, or transitional species between ecomorphs, others may have morphologies that reflect distinct patterns of habitat specialization relative to the Caribbean ecomorphs. Alternatively, these species might belong to a more inclusive ecomorph, such as a bush-trunk-ground ecomorph. Future studies that examine the relationships between habitat use and morphology in these intermediate species will help refine our understanding of anole ecomorphology.

Testable ecomorph classifications based on morphology

Our analytical framework, which provides an explicit test for morphological differences between ecomorphs and predicts ecomorph classifications based on morphology assigned a greater relative proportion of mainland Draconura species to ecomorph classes compared to the morphology-based methods employed by previous studies (Irschick et al., 1997; Schaad & Poe, 2010). Some differences are likely explained by our more permissive, yet still conservative criteria, which assign species based on the observed variation among a priori members of a given ecomorph. However, we also examined more and slightly different traits compared to previous studies (e.g. head dimensions and toepad width rather than lamella count). Schaad & Poe (2010) used scale counts and sexual size dimorphism (among other traits similar to ours) to classify mainland species into ecomorphs; however, these traits are not directly tied to structural habitat use. As such, they may be poor predictors of habitat use on the mainland and could explain differences in our results. Furthermore, our classification approach quantifies the similarities between less specialized species and the different ecomorphs to propose alternative ecomorph assignments. That said, our morphological classifications are presented as hypotheses to be tested as additional microhabitat use data become available.

For several better-known Draconura species, the available natural history data support our hypothesized ecomorph assignments. For example, some mainland species from the A. pentaprion clade plus A. salvini were classified as twig anoles (Fig. 1). These species are all highly arboreal relative to other Draconura species (Savage, 2002; McCranie & Köhler, 2015), and at least A. pentaprion is often found on narrow perches, similar to Caribbean twig anoles (J. Losos, pers. comm.). We also classified the mainland Draconura species A. auratus as a grass-bush anole, a long-recognized assignment based on both ecology and morphology (Avila-Pires, 1995; Irschick et al., 1997; Schaad & Poe, 2010; Fig. 1). Lastly, several Draconura species assigned to the trunk-ground ecomorph (Anolis cryptolimifrons, Anolis mccraniei, Anolis osa, Anolis serranoi and Anolis subocularis) often perch head-down low on tree trunks and regularly descend to the ground (Andrews, 1971; Jackson, 1973; Savage, 2002; Köhler & Acevado, 2004; Köhler et al., 2014; Fig. 1), behaviours that are characteristic of Caribbean trunk-ground anoles. Other well-supported examples of ecomorph assignments are outlined in Supporting Information (Table S9). These corroborated ecomorph classifications highlight the usefulness and potential of our method to generate testable hypotheses based on the relationships between form and function as seen in, but not limited to, fish pharyngeal jaws, bat cranial morphologies, grass leaf shapes and the limb proportions of several frog clades (Santana & Cheung, 2016; Burress et al., 2018; Gallaher et al., 2019; Moen, 2019).

Although some of our assignment criteria are less strict than those applied in previous investigations, our final classifications may still fail to detect species that belong to an ecomorph. For example, A. biporcatus and A. petersii are two large, green and highly arboreal species that might represent mainland crown-giants but were not assigned as such (Irschick et al., 1997; Losos, 2009; McCranie & Köhler, 2015), potentially because these “mainland crown-giants” do not attain body sizes as large as their Caribbean counterparts (McCranie & Köhler, 2015; Armstead et al., 2017). Both A. biporcatus and A. petersii were identified as intermediates between crown-giant and another ecomorph, clustered morphologically with the Caribbean crown-giants, and occupied adaptive peaks close to those occupied by the Caribbean crown-giant species, supporting that they are less specialized (i.e. smaller) crown-giant anoles. These results also highlight how considering only the Caribbean ecomorph species may result in overly narrow morphological (and likely ecological) characterizations of the ecomorphs. Conversely, our methods may have incorrectly assigned some Draconura species to ecomorphs. For instance, A. kemptoni was classified as a trunk-crown anole, but it appears to be ecologically more similar to trunk-ground anoles (Savage, 2002), and yet this species showed only weak support for an alternative trunk-ground classification (Supporting Information, Tables S4–S8).

What factors explain these instances where the relationship between morphology and ecology do not correspond to those observed in the ecomorph species? One possibility is that higher or lower diversity in local anole assemblages leads to variable ecomorphological relationships. At some localities on the mainland, as well as in the Caribbean, anole communities are composed of a few or no sympatric anole species (Poe & Anderson, 2019). For example, some Draconura species are endemic to islands (Calderón-Espinosa & Barragán Forero, 2011; McCranie & Köhler, 2015; Phillips et al., 2019) or occur in high elevation environments (as is the case for A. kemptoni), where they co-exist with few anole species and fewer competitors in general (Savage, 2002; McDiarmid & Donnelly, 2005; Leenders, 2019). Likewise, communities of anoles in the Lesser Antilles, which include many of our previously unclassified Caribbean species, typically include just one or two species per island (Henderson & Powell, 2009; Yuan et al., 2020). There is strong evidence that niche partitioning and morphological specialization are tied to interspecific competition (Losos, 1992; Schluter, 1998; Pfennig & Pfennig, 2009; Adams, 2010; Yuan et al., 2020). Therefore, the presence of fewer anole competitors in some communities might lead to ecological release or natural selection favouring different patterns of microhabitat use and associated morphological adaptations compared to species-rich assemblages (Carpenter, 1965; Losos & de Queiroz, 1997; Calderón-Espinosa & Barragán Forero, 2011; McCranie & Köhler, 2015). More broadly, community composition is an important factor to consider when assessing the presence (and absence) of convergent evolution.

An ecomorph of ground anoles

We found strong evidence for a previously undescribed ecomorph consisting of predominately ground-dwelling mainland and Caribbean anoles. These species exhibit strong morphological convergence (Table 4) and occupy an area of the morphospace that is statistically distinct from all of the other anole ecomorphs. Ground anoles share longer limbs, reduced finger- and toe-pads and shorter heads (Figs 2, 4), traits that are associated with life on the ground in other lizards. For instance, longer limbs allow anoles and other ground-dwelling lizards to sprint more efficiently on broader surfaces (Losos, 1990; Melville & Swain, 2000; Herrel et al., 2002). Similarly, narrower toepads in ground anoles are consistent with the positive correlation between perch height and toepad size in anoles, and the understanding that toepads are an adaptation for clinging to vertical surfaces (Macrini et al., 2003; Elstrott & Irschick, 2004; Crandell et al., 2014; Collins et al., 2015). A short and stout head, presumably associated with diet, has also evolved more than once in ground-dwelling Australian agamid lizards (Gray et al., 2019). Finally, many of our putative ground ecomorph species have flatter claws than the more arboreal Caribbean ecomorph species, consistent with the inference that flatter claws improve performance while running, whereas highly curved claws are better suited to climbing [Yuan et al. (2019) and references therein].

Additional support for the ground ecomorph stems from our ability to classify ecologically similar species to this ecomorph based on morphology. For example, Anolis cupreus, Anolis granuliceps, Anolis notopholis and Anolis wampuensis are some of the species commonly found on the ground that remained unclassified both in our study and in past studies prior to the addition of the ground ecomorph (Savage, 2002; Castro-Herrera, 1988; McCranie & Köhler, 2015; Moreno-Arias et al., 2020). A. notopholis and three other species classified as either ground or intermediates between ground and trunk-ground also appear to be converging towards the same adaptive peak as the ground anoles from the A. chrysolepis complex, assigned a priori to this ecomorph (Supporting Information, Fig. S3, Regime 14). However, ground-dwelling is unlikely to be an ecological monolith. Although the ground species we assigned a priori predominately live among leaf-litter, other terrestrial anoles perch on ground-level structures such as rocks, logs, tree roots and other low vegetation. Moreover, several species we classified as ground anoles may perch on a combination of tree trunks and shrubs at low heights, but potentially less so than typical grass-bush or trunk-ground anoles, in addition to being active on the ground (Savage, 2002; Köhler et al., 2014, McCranie & Köhler, 2015; Muñoz et al., 2015; Supporting Information, Table S9). This variation introduces the possibility of finer partitioning of the broad “ground” category; however, the ecological data needed to tease apart microhabitat preferences for these species are not currently available. Correspondingly, ecomorph classifications near the intersection of the ground, trunk-ground and grass-bush ecomorphs in the morphospace are not always straightforward given the ecological and morphological similarities between those ecomorphs (this study).

The recognition of the ground ecomorph further supports our conclusion that the island and mainland anole radiations are more similar than previously understood. Under the original anole ecomorph concept, the Caribbean ground anole A. barbouri and potential ground species A. rupinae (Williams & Webster, 1974) have been regarded as ecological “singletons” that are dissimilar to any of the ecomorph species (Losos, 2009). However, our data reveal that these Caribbean “singletons” resemble some mainland Draconura species, specifically the ground-dwelling anoles, suggesting that at least some previously unclassified Caribbean species are not unique when considering the broader radiation of anoles. This finding raises the question of whether there are other undescribed ecomorph classes consisting of both mainland and Caribbean species. For example, there are several island and mainland species that perch on vertical rock walls including those of caves (Glor et al., 2003; Henderson & Powell, 2009; Nicholson et al., 2012; Scarpetta et al., 2015; Muñoz & Losos, 2018), as well as species that commonly perch on top of rock-piles (Birt et al., 2001; Losos, 2009; Muñoz et al., 2015). Most rock-wall species were assigned to the trunk-ground ecomorph in our study, and a few species that perch on top of rocks were intermediate between the ground and trunk-ground ecomorphs; thus, it seems unlikely that rock-dwelling anoles represent a single ecomorph class (Supporting Information, Table S9; Fig. S5). Our classifications are consistent with a difference between vertical and horizontal rock-perching that aligns well with the functional challenges faced by trunk-ground and ground anoles, respectively. This finding illustrates that not all species with distinctive ecologies also have distinct morphologies that warrant a separate ecomorph class, suggesting that some ecomorphs are more inclusive and/or contain subclasses. Along the same vein, it is equally important to recognize that distinct morphologies do not necessarily equate to distinct ecologies as different trait combinations may result in similar performance outcomes (Wainwright et al., 2005). Addressing the possibility of currently unrecognized ecomorphs will require more in-depth ecological and morphological investigations, particularly for mainland taxa.

Replicate adaptive radiations in island and mainland settings

Our data provide compelling morphological evidence that the evolution of anole ecomorphs, which has occurred independently multiple times in the Greater Antilles, occurred an additional time when anoles recolonized the mainland. Despite the general similarities, there are subtle differences between the Greater Antillean and mainland Draconura radiations. Species representing some of the more arboreal ecomorphs do not appear to be as ubiquitous in the mainland Draconura clade as they are in the Caribbean. In particular, relatively few Draconura anoles were assigned to the crown-giant, trunk-crown, trunk and twig ecomorphs. Conversely, there is an abundance of ground-dwelling anole species on the mainland but few in the Caribbean. These patterns may reflect historical contingencies in the evolution of the Draconura clade. By the time the ancestor of the Draconura clade recolonized the mainland from the Caribbean, Central and South America were already populated by a diverse assemblage of arboreal lizards, including members of the Dactyloa clade of anoles (Nicholson et al., 2005, 2012; Lotzkat et al., 2013; Prates et al., 2020). We propose that the limited availability of arboreal niches both restricted the evolution of Draconura species into the highly arboreal ecomorphs and facilitated the diversification of ground anoles. In sum, we posit that occupying similar microhabitats in island and mainland settings led to broadscale convergence in the adaptive radiation of anoles; however, regional differences in selective landscapes led to variation in how the details of those radiations unfolded.

The anole system is one of the few recognized examples of replicated radiations that occur outside of just insular and lacustrine environments (Bossuyt & Milinkovich, 2000; Ruedi & Mayer, 2001, Burress et al., 2018; Rincon-Sandoval et al., 2020). Our results are congruent with the idea that continental settings can provide equal or even greater ecological opportunities than island settings (Velasco et al., 2016, 2018; Salazar et al., 2019). This study also suggests that the favourable conditions provided by islands and lakes are not absolutely crucial for replicated radiations to occur. With this in mind, we invite future studies to investigate other lineages that span distinct geographic regions to better understand the factors that facilitate and impede replicated radiations.

SUPPORTING INFORMATION

Additional Supporting Information may be found in the online version of this article at the publisher’s web-site:

Material and Methods S1. Species substitutions in the phylogeny.

Results S1. DFA misclassifications.

Table S1. List of the 205 anole species used in this study and their ecomorph assignments.

Table S2. Loadings for the first five axes of the phylogenetic principal component analysis.

Table S3. Summary of the morphospatial Euclidean distances for a priori ecomorph species.

Table S4. DFA posterior probabilities for the ecomorph classifications made with only the six Caribbean ecomorphs.

Table S5. Centroid distances for the Draconura anoles and previously unclassified Caribbean species.

Table S6. Mean pairwise distances for the Draconura anoles and previously unclassified Caribbean species.

Table S7. Nearest neighbour distances for the Draconura anoles and previously unclassified Caribbean species.

Table S8. DFA posterior probabilities for the ecomorph classifications made with the inclusion of the ground ecomorph.

Table S9. Assessments of the final ecomorph assignments inferred from morphology with ecological data.

Figure S1. Individual trait variation among the a priori Caribbean and ground ecomorph species.

Figure S2. Results from the cluster analyses using the a priori ecomorph species.

Figure S3.l1ou results inferred with the MCC phylogeny.

Figure S4.l1ou results inferred with the time-calibrated phylogeny.

Figure S5. Relative positions of the Caribbean ecomorphs, ground ecomorph and rock anoles in the morphospace.

ACKNOWLEDGEMENTS

For facilitating access to specimens, we thank: Addison Wynn, Esther Langan and Kenneth Tighe of the National Museum of Natural History, Smithsonian Institution; Erica Ely and Lauren Scheinberg of the California Academy of Sciences; Carol Spencer of the Museum of Vertebrate Zoology; and José Rosado of the Museum of Comparative Zoology. For giving important feedback at different stages of this study, we thank Mike Yuan, Kyle O’Connell, Ryan Schott, Molly Womack, Kevin Mulder and Anna Penna. In addition, we thank three anonymous reviewers for their thoughtful feedback that substantially improved the manuscript. We also thank Luke Tornabene for hosting a specimen loan. J.M.H. would also like to thank Elizabeth Cottrell, Eugene Hunt, Virginia Powers and the National History Research Experience program National Science Foundation (REU Site OCE-1560088) for providing the opportunity to conduct this research. I.P. was supported by a Smithsonian Peter Buck post-doctoral fellowship. Additional support was provided by the Gaver Fund of the Division of Amphibians and Reptiles, National Museum of Natural History. Author contributions: I.P., J.M.H. and K.d.Q. conceptualized the study. K.d.Q. and R.C.B. provided financial support. J.M.H. collected and analysed the data with input from I.P., K.d.Q. and R.C.B. J.M.H. wrote the manuscript with contributions from I.P., K.d.Q. and R.C.B. The authors have no conflict of interests.

REFERENCES

Parallel evolution of character displacement driven by competitive selection in terrestrial salamanders

. : .

Structural habitat and time budget of a tropical Anolis lizard

. : –.

, , .

A simple measure of the strength of convergent evolution

Methods in Ecology and Evolution

: –.

, .

Windex: analyzing convergent evolution using the Wheatsheaf index in R

Evolutionary Bioinformatics

: .

, , , , .

Systematics and ecology of Anolis biporcatus (Squamata: Iguanidae)

. : –.

, .

Use of an exemplar versus use of a sample for calculating summary metrics of morphological traits in comparative studies of Anolis lizards

. : –.

Lizards of Brazilian Amazonia (Reptilia: Squamata)

Zoologische Verhandelingen

: –.

, .

Ecological morphology of Caribbean anoles

Herpetological Monographs

: –.

, , .

Natural history of Anolis barkeri: a semiaquatic lizard from southern Mexico

. : –.

SIMMAP: stochastic character mapping of discrete traits on phylogenies

. : .

, .

Convergent adaptive radiations in Madagascan and Asian ranid frogs reveal covariation between larval and adult traits

Proceedings of the National Academy of Sciences

: –.

, , , , , , .

Island- and lake-like parallel adaptive radiations replicated in rivers

Proceedings of the Royal Society B: Biological Sciences

: .

, .

Morphological diversification in solitary endemic anoles: Anolis concolor and Anolis pinchoti from San Andres and Providence islands, Colombia

South American Journal of Herpetology

: –.

. :

Columbia University Press

The display of the Cocos Island anole

. : –.

Niche structure of an anole community in a tropical rain forest within the Choco region of Colombia.

Unpublished D. Phil. Thesis,

North Texas State University

Ecological and morphological patterns in communities of land snails of the genus Mandarina from the Bonin Islands

Journal of Evolutionary Biology

: –.

, , .

Subdigital adhesive pad morphology varies in relation to structural habitat use in the Namib day gecko

. : –.

, , , , .

Stick or grip? Co-evolution of adhesive toepads and claws in Anolis lizards

. : –.

, , , .

Ecomorphological convergence in Eleutherodactylus frogs: a case of replicate radiations in the Caribbean

. : –.

, .

Evolutionary correlations among morphology, habitat use and clinging performance in Caribbean Anolis lizards

Biological Journal of the Linnean Society

: –.

, , , , , , , , , , .

Leaf shape and size track habitat transitions across forest-grassland boundaries in the grass family (Poaceae)

. : –.

, , , , .

Repeated diversification of ecomorphs in Hawaiian stick spiders

. : –.

, , , , , , , , , , , , , , , , , , , , , , , .

Comparing adaptive radiations across space, time, and taxa

. : –.

, , , , .

Phylogenetic analysis of ecological and morphological diversification in Hispaniolan trunk-ground anoles (Anolis cybotes group)

. : –.

, , , .

Evolution of cranial shape in a continental-scale evolutionary radiation of Australian lizards

. : –.

, , Jr, .

Morphological evolution in Tropidurinae squamates: an integrated view along a continuum of ecological settings

Journal of Evolutionary Biology

: –.

, .

Natural history of West Indian reptiles and amphibians.

University Press of Florida

, , .

Relations between microhabitat use and limb shape in phrynosomatid lizards

Biological Journal of the Linnean Society

: –.

, , , , .

Natural history of a terrestrial Hispaniolan anole: Anolis barbouri

. : –.

, , .

Co-evolution of cleaning and feeding morphology in western Atlantic and eastern Pacific gobies

. : –.

, , , .

A comparison of evolutionary radiations in mainland and Caribbean Anolis lizards

. : –.

Notes on the population biology of Anolis tropidonotus in a Honduran highland pine forest

. : –.

, , , .

Fast and accurate detection of evolutionary shifts in Ornstein–Uhlenbeck models

Methods in Ecology and Evolution

: –.

, , , .

Similar morphologies of cichlid fish in Lakes Tanganyika and Malawi are due to convergence

Molecular Phylogenetics and Evolution

: –.

Characters of external morphology used in Anolis taxonomy—definition of terms, advice on usage, and illustrated examples

. : –.

, .

The anoles (genus Norops) of Guatemala. I. The species of the Pacific versant below 1500 m elevation

. : –.

, , , .

A revision of the Mexican Anolis (Reptilia, Squamata, Dactyloidae) from the Pacific versant west of the Isthmus de Tehuantepec in the states of Oaxaca, Guerrero, and Puebla, with the description of six new species

. : –.

, , .

Lack of convergence in aquatic Anolis lizards

. : –.

Reptiles of Costa Rica: a field guide.

: .

Ecomorphology, performance capability, and scaling of West Indian Anolis lizards: an evolutionary analysis

. : –.

The evolution of convergent structure in Caribbean Anolis communities

. : –.

Lizards in an evolutionary tree: ecology and adaptive radiation of anoles.

University of California Press

Convergence, adaptation, and constraint

. : –.

, .

Evolutionary consequences of ecological release in Caribbean Anolis lizards

Biological Journal of the Linnean Society

: –.

, , , , .

Contingency and determinism in replicated adaptive radiations of island lizards

. : –.

, , , .

Distribution and variation of the giant alpha anoles (Squamata: Dactyloidae) of the genus Dactyloa in the highlands of western Panama, with the description of a new species formerly referred to as D. microtus

. : –.

, .

The theory of island biogeography.

Princeton University Press

, , .

Ecomorphological differences in toepad characteristics between mainland and island anoles

. : –.

, , , .

Exceptional convergence on the macroevolutionary landscape in island lizard radiations

. : –.

, .

The anoles (Reptilia: Squamata: Dactyloidae: Anolis: Norops) of Honduras. Systematics, distribution, and conservation

Bulletin of the Museum of Comparative Zoology

: –.

, .

The herpetofauna of the Guayana highlands: amphibians and reptiles of the lost world.

Ecology and evolution in the Tropics: a herpetological perspective.

University of Chicago Press

, .

Evolutionary relationships between morphology, performance and habitat openness in the lizard genus Niveoscincus (Scincidae: Lygosominae)

Biological Journal of the Linnean Society

: –.

What determines the distinct morphology of species with a particular ecology? The roles of many-to-one mapping and trade-offs in the evolution of frog ecomorphology and performance

. : –.

, , .

Evolution of ecological structure of anole communities in tropical rain forests from north-western South America

Zoological Journal of the Linnean Society

: –.

, .

Patterns of morphological diversification of mainland Anolis lizards from northwestern South America

Zoological Journal of the Linnean Society

: –.

, , , , , , , , .

Multiple paths to aquatic specialisation in four species of Central American Anolis lizards

Journal of Natural History

: –.

, .

Thermoregulatory behavior simultaneously promotes and forestalls evolution in a tropical lizard

. : –.

, .

Ward’s hierarchical agglomerative clustering method: which algorithms implement Ward’s criterion?

Journal of classification

: –.

, , , .

It is time for a new classification of anoles (Squamata: Dactyloidae)

. : –.

, , , , , .

Mainland colonization by island lizards

. : –.

, .

Character displacement: ecological and reproductive responses to a common evolutionary problem

The Quarterly Review of Biology

: –.

, , , , .

Biogeography, systematics, and ecomorphology of Pacific island anoles

. : .

, , , .

Testing the island effect in adaptive radiation: rates and patterns of morphological diversification in Caribbean and mainland Anolis lizards

Proceedings of the Royal Society B: Biological Sciences

: –.

, .

The existence and evolution of morphotypes in Anolis lizards: coexistence patterns, not adaptive radiations, distinguish mainland and island faunas

. : .

, , , , , , , , , , .

Comparative evolution of an archetypal adaptive radiation: innovation and opportunity in Anolis lizards

. : –.

, , , , , , , , , , .

A phylogenetic, biogeographic, and taxonomic study of all extant species of Anolis (Squamata; Iguanidae)

. : –.

, , , , , .

Discovery of a new species of Anolis lizards from Brazil and its implications for the historical biogeography of montane Atlantic Forest endemics

. : –.

, .

Predicting speciation probability from replicated population histories

. : –.

, .

The anoles of La Palma: aspects of their ecological relationships

. : –.

phytools: an R package for phylogenetic comparative biology (and other things)

Methods in Ecology and Evolution

: –.

, , , , , , , , , , .

Evolutionary determinism and convergence associated with water-column transitions in marine fishes

Proceedings of the National Academy of Sciences

: –.

, .

Molecular systematics of bats of the genus Myotis (Vespertilionidae) suggests deterministic ecomorphological convergences

Molecular Phylogenetics and Evolution

: –.

, , , .

Natural selection and parallel speciation in sympatric sticklebacks

. : –.

, , , , .

Physiological evolution during adaptive radiation: a test of the island effect in Anolis lizards

. : –.

, .

Go big or go fish: morphological specializations in carnivorous bats

Proceedings of the Royal Society B: Biological Sciences

: .

The amphibians and reptiles of Costa Rica: a herpetofauna between two continents, between two seas

. :

University of Chicago Press

, , , , , , , , .

Morphology and ecology of the Mexican cave anole Anolis alvarezdeltoroi

. : –.

, .

Patterns of ecomorphological convergence among mainland and island Anolis lizards

Biological Journal of the Linnean Society

: –.

Character displacement and the adaptive divergence of finches on islands and continents

. : –.

The ecology of adaptive radiation.

: .

, .

Parallel speciation by natural selection

. : –.

The major features of evolution

. :

Columbia University Press

The definition, recognition, and interpretation of convergent evolution, and two new measures for quantifying and assessing the significance of convergence

. : –.

, .

Ecomorphology of Anolis lizards of the Choco′ region in Colombia and comparisons with Greater Antillean ecomorphs

Biological Journal of the Linnean Society

: –.

, , , , , , .

Climatic niche attributes and diversification in Anolis lizards

. : –.

, , , , , , , .

Climatic and evolutionary factors shaping geographical gradients of species richness in Anolis lizards

Biological Journal of the Linnean Society

: –.

, .

Modern applied statistics with S,

4th edn. : .

, , , , .

Life above ground: ecology of Anolis fuscoauratus in the Amazon rain forest, and comparisons with its nearest relatives

Canadian Journal of Zoology

: –.

, , , , .

Life in shade: the ecology of Anolis trachyderma (Squamata: Polychrotidae) in Amazonian Ecuador and Brazil, with comparisons to ecologically similar anoles

. : –.

, , , .

Life on the leaf litter: the ecology of Anolis nitens tandai in the Brazilian Amazon

. : –.

, , , .

Many-to-one mapping of form to function: a general principle in organismal design?

Integrative and Comparative Biology

: –.

The origin of faunas. Evolution of lizard congeners in a complex island fauna: a trial analysis

. : –.

Ecomorphs, faunas, island size, and diverse end points in island radiations of Anolis

. In: , , , eds.

Lizard ecology: studies of a model organism.

: , –.

, .

Anolis rupinae new species: a syntopic sibling of A. monticola Shreve

. : –.

, , , .

Habitat use, interspecific competition and phylogenetic history shape the evolution of claw and toepad morphology in Lesser Antillean anoles

Biological Journal of the Linnean Society

: –.

, , .

Phenotypic integration between claw and toepad traits promotes microhabitat specialization in the Anolis adaptive radiation

. : –.

Published by Oxford University Press on behalf of The Biological Journal of the Linnean Society 2021.

Giới Tính Trai

Which trait is advantageous for anoles that live in trees and cling to leaves?

INTRODUCTION

MATERIAL AND METHODS

Specimen sampling and morphological measurements

Assessing support for the ecomorphs

Classification of mainland anoles to the Caribbean ecomorphs based on morphology

Assessing a novel ground ecomorph

Reconstructing the evolution of mainland ecomorphs and their convergence

Investigating faunal convergence between island and mainland radiations

Data availability

RESULTS

Mainland and Caribbean anoles in the morphospace

Support for the Caribbean ecomorphs

Classification of mainland anoles into the six Caribbean ecomorphs

Evaluation of a new ground ecomorph

Repeated ecomorph evolution on the mainland

Faunal convergence between island and mainland anole radiations

DISCUSSION

Testable ecomorph classifications based on morphology

An ecomorph of ground anoles

Replicate adaptive radiations in island and mainland settings

SUPPORTING INFORMATION

ACKNOWLEDGEMENTS

REFERENCES

Bài Viết Liên Quan

Quảng Cáo

Có thể bạn quan tâm

Toplist được quan tâm

Quảng cáo

Xem Nhiều

Quảng cáo

Chúng tôi

Điều khoản

Trợ giúp

Mạng xã hội