Community partitioning into sub-faunas
As predicted by the availability of greater highland evolutionary opportunity in Hispaniola, the two faunas diverge in community turnover across elevations. As such, total beta-diversity is greater on Hispaniola than Jamaica, and Hispaniola’s beta-diversity derives almost entirely from among-site species turnover (Table 1). Jamaica, in contrast, shows markedly less turnover, with a substantial portion of its compositional dissimilarity stemming from gradients in richness and abundance. This difference is a community-level signature from the evolution of high-elevation specialists on Hispaniola. Specifically, the Hispaniolan fauna contains 5 species that are broadly distributed across the lowlands, 1 that is a specialist on dry lowland regions, and 5 that maximize their abundance above 700m and do not occur at sea level. By comparison, Jamaica has 5 species that are distributed across the lowlands, and only 1 high-elevation specialist.
Hispaniola’s species are partitioned into two entirely distinct sub-faunas: the lowest and highest communities share no species in common. In contrast, some of the lowland species in Jamaica are among those that persist at the highest sites (Fig 3a-c). Interestingly the elevated beta-diversity of Hispaniola contains a strong phylogenetic signal linked to historical elevational specialization. The Hispaniolan highland fauna appears to be derived from a combination of in situ radiation and evolutionary colonization from lowland derived species. While two highland species each have a lowland species as their closest relative, the most abundant highland forest understory species primarily derive from a single clade, which colonized the highlands reasonably early in the history of the island, seeming to have initiated an adaptive radiation in miniature (Fig 3d,e). The presence of this clade drives the formation of the sub-regional fauna on Hispaniola, resulting in substantial elevational turnover that is absent from Jamaica.
Evolution of this highland sub-fauna on Hispaniola also results in community-level differences in functional trait distribution. Average morphology on Hispaniola diverges between lowland and highland communities but remains steady on Jamaica across elevations (Fig 3f). Different environments in the highlands and lowlands should require different morphological solutions for optimal fitness. Given ample evolutionary opportunity, communities in similar environments should reach the same solution. Indeed, the community abundance-weighted mean morphology in the lowlands of Jamaica and Hispaniola is extremely similar (Fig. 3g)—more similar than expected if species morphologies are randomly shuffled (Fig. 4, P = 0.033). In contrast the highlands are morphologically distinct, without evidence for convergence between islands (P = 0.44), suggesting that evolutionary opportunity on Hispaniola has created a community unique in both diversity and morphology, that has not been reproduced on Jamaica. As a result, islandwide average morphologies are not more similar across islands than expected by chance (Fig. S2, P = 0.076)
Finally, the development of the high-elevation fauna on Hispaniola appears to have granted access to niche space that would otherwise go unoccupied by elevational generalists. For example, on Hispaniola above 1800m three species were common and species continued to occupy montane forests up until 2,300m (Fig 2b,c). But while Jamaica reaches 2,250m at its peak, we detected no anoles in the forests at or above 1800m (however, even with no detections at the 4 highest plots over six two-hour surveys, resight models cannot rule out a small number of individuals, and so model-predicted species richnesses are non-zero; as in Fig. 1b).