Dewlap colour variation in Anolis sagrei is maintained among habitats within islands of the West Indies

Abstract

Animal signals evolve in an ecological context. Locally adapting animal sexual signals can be especially important for initiating or reinforcing reproductive isolation during the early stages of speciation. Previous studies have demonstrated that dewlap colour in Anolis lizards can be highly variable between populations in relation to both biotic and abiotic adaptive drivers at relatively large geographical scales. Here, we investigated differentiation of dewlap coloration among habitat types at a small spatial scale, within multiple islands of the West Indies, to test the hypothesis that similar local adaptive processes occur over smaller spatial scales. We explored variation in dewlap coloration in the most widespread species of anole, Anolis sagrei, across three characteristic habitats spanning the Bahamas and the Cayman Islands, namely beach scrub, primary coppice forest and mangrove forest. Using reflectance spectrometry paired with supervised machine learning, we found significant differences in spectral properties of the dewlap between habitats within small islands, sometimes over very short distances. Passive divergence in dewlap phenotype associated with isolation-by-distance did not seem to explain our results. On the other hand, these habitat-specific dewlap differences varied in magnitude and direction across islands, and thus, our primary test for adaptation-parallel responses across islands-was not supported. We suggest that neutral processes or selection could be involved in several ways, including sexual selection. Our results shed new light on the scale at which signal colour polymorphism can be maintained in the presence of gene flow, and the relative role of local adaptation and other processes in driving these patterns of dewlap colour variation across islands.

Keywords: adaptation; machine learning; polymorphism; reflectance; sexual signal.

FIGURE 1

Overview of our study design, including a map of the Bahamas and the Cayman Islands, on which are indicated the nine islands we sampled, two representatives of our study species Anolis sagrei with their dewlaps deployed, and the three types of habitats we considered on each island

FIGURE 2

Comparison of dewlap colouration across habitats on Abaco. (a) Map of the island with the sampling sites coloured by habitat. (b) Reflectance profiles of all the dewlaps on the island. (c) How reflectance profiles map onto the within‐island principal components. (d) Confusion matrix showing the proportion of lizards from each (true) habitat reassigned to each (predicted) habitat by the random forests, based on the first four within‐island principal components and averaged across replicates. Each column sums to one. (e) Within‐island principal component scores across habitats. Bars indicate significant contrasts. *$p<0.05$; **$p<0.01$; ***$p<0.001$