Photo by Richard Glor

Anolis distichus, the North Caribbean bark anole, is probably best known by readers of Anole Annals for its striking variability in dewlap color among populations. Primarily based on these differences in dewlap color, Albert Schwartz published a monograph with descriptions for 18 subspecies distributed across Hispaniola and the Bahamas (1968). The large number of subspecies and the question of their origin has helped establish Anolis distichus as one of the most intriguing cases for the study of speciation in anoles. Do the subspecies of Anolis distichus represent geographic patterns in dewlap color variation? Or, are the subspecies evolutionarily isolated lineages worthy of species status?

Molecular genetic data have revealed several genetically distinct populations of Anolis distichus that appear to be at varying stages of the speciation process. However, with the exception of A. d. ignigularis and A. d. favillarum, these genetically distinct groups did not align with the subspecies described by Schwartz, corresponding better to geography than patterns of dewlap color variation (Geneva et al. 2015). Most recently, using genome wide markers and multicoalescent species delimitation methods, MacGuigan et al. (2017) identified six candidate species within A. distichus. The authors, however, did not update the taxonomy because the boundaries among the candidate species remained unclear.

To clear up those boundaries, we tested if the candidate species identified by MacGuigan could be distinguished by morphological characters. Because Schwartz considered scale counts along with dewlap and body color pattern in his monograph and was not able to recover any diagnostic differences, we focused on morphometric characters and measured 13 traits from more than 500 animals (available on Dryad). We conducted univariate and multivariate analyses to test if (a) any of the individual characters distinguished candidate species; and (b) if characters considered in aggregate could distinguish the candidate species. Because the candidate species are parapatrically distributed across Hispaniola, locality information has the potential to aid diagnosis. To account for this, we carried out comparisons of all candidate species together and only the pairs of candidate species that potentially come into contact. 

Although ANOVAs recovered significant differences in mean character values, visual examination of violin plots showed that none of them were diagnostic for any of the candidate species.  Discriminant Function Analysis (DFA) revealed clustering of some of the candidate species, but there was still substantial overlap in multivariate space and candidate species were diagnosed with poor accuracy. DFA did prove to be more accurate when asked to classify individuals only between the pairs of potentially co-occurring candidate species instead of all of the candidate species together. Ultimately, we still rejected the hypothesis that candidate species could be distinguished on the basis of our morphometric dataset due to gaps in our sampling and overall similarity among the candidate species.

Univariate and multivariate results from Myers et al. (2020). (A) Example violin plots for two characters, snout-vent length and head width. (B) Multivariate plots of DFA results.

Because both univariate and multivariate analyses did not recover support for the hypothesis that the candidate species could be distinguished by morphometric characters, we decided to test how many species were supported by the data.  We tested alternative species delimitation scenarios with a model fitting approach that uses normal mixture models (Cadena et al. 2018). Normal mixture models consider morphological variation to consist of a mix of different normal distributions and, unlike DFA, does not require individuals to be assigned to different groups a priori. We tested the support for models specifying up to as many as 12 species and compared the support for these generic models to models specifying MacGuigan’s candidate species and Schwartz’s subspecies. The best supported normal mixture model specified two groups; however, the model specified a group containing 489 individuals and another with 24 individuals and followed no clear geographic trend. We scrutinized the principal components used to estimate these models and determined that longitudinal and vertical ear opening diameter were driving this result. We removed them and conducted the normal mixture model analysis with the reduced dataset and recovered a single morphological group.

Is Anolis distichus only one species? Do the candidate species lack distinguishing features? We weren’t comfortable making either conclusion. Our dataset of linear morphometric characters was not capable of distinguishing candidate species, but future datasets featuring other aspects of phenotypic variation e.g., geometric morphometrics, might. Larger, genome-scale datasets with more comprehensive geographic sampling than previous molecular genetic studies will also help address the question of species boundaries in A. distichus. We also discuss some possibilities for why we were unable to recover distinguishing morphological differences, including that local adaptation across Hispaniola’s environmentally heterogeneous landscapes has resulted in morphometric variation that does not align with candidate species boundaries. Ng et al. (2013) found that dewlap color in A. distichus is correlated with local environmental signaling conditions, which would explain why dewlap color does not correspond with putative evolutionary lineages in this group of lizards. Many of the morphometric traits we considered (e.g., limb length) affect ecological performance and may have responded similarly to selective pressures. 

We were not able to resolve the confounding taxonomy of Anolis distichus in this paper, but I enjoyed the project and found it to be a very rewarding first publication. I was able to travel to the Dominican Republic to catch lizards twice as an undergraduate for this project and we were able to amass a large and geographically comprehensive dataset. Working on the taxonomy of this confounding group of lizards helped me realize my interest in speciation, which I plan to pursue further in my PhD. A lot of friends and collaborators helped with this project by assisting with fieldwork and providing input on manuscript drafts. And, of course, this work wouldn’t have been possible without my co-authors and mentors, Pietro de Mello and Rich Glor. 

References

Cadena, C. D., F. Zapata, and I. Jiménez. 2018. Issues and perspectives in species delimitation using phenotypic data: Atlantean evolution in Darwin’s finches. Syst. Biol. 67:181–194.

Geneva, A. J., J. Hilton, S. Noll, and R. E. Glor. 2015. Multilocus phylogenetic analyses of Hispaniolan and Bahamian trunk anoles (distichus species group). Mol. Phylogenet. Evol. 87:105–117.

MacGuigan, D. J., A. J. Geneva, and R. E. Glor. 2017. A genomic assessment of species boundaries and hybridization in a group of highly polymorphic anoles (distichus species complex). Ecol. Evol. 7:3657–3671.

Myers, T. C., P. L. H. de Mello, and R. E. Glor. 2020. A morphometric assessment of species boundaries in a widespread anole lizard (Squamata: Dactyloidae). Biol. J. Linn. Soc. 130:813-825.

Ng, J., E. L. Landeen, R. M. Logsdon, and R. E. Glor. 2013. Correlation between Anolis lizard dewlap phenotype and environmental variation indicates adaptive divergence of a signal important to sexual selection and species recognition. Evolution 67:573–582.

Schwartz, A. 1968. Geographic variation in Anolis distichus Cope (Lacertilia, Iguanidae) in the Bahama Islands and Hispaniola. Bull. Mus. Comp. Zool. 137:255–309.