Huntsmen of the Caribbean: Multiple tests of the GAARlandia hypothesis

The origin of the Caribbean biota remains debated, but amassing evidence suggests important roles of both dispersal and vicariance events in the colonization the archipelago. The most prominent vicariance hypothesis is colonization over the GAARlandia land bridge that putatively connected the Greater Antilles to South America around 33 mya. This hypothesis has received support from studies of individual lineages, but its main prediction—the simultaneous colonization of multiple lineages during that time window—requires further unambiguous corroboration. Here, we examine the phylogenetic structure of huntsman spiders (Sparassidae) of the Caribbean. Huntsman spiders are appropriate models for this question, as they are expected to be dispersal limited as substrate and foliage dwelling spiders that rarely balloon, yet are found on some volcanic islands, and thus at least some overwater dispersal must have occurred. We focus on the Caribbean endemic Neostasina, but also include Caribbean Olios, for a deeper biogeographical understanding. We use two mitochondrial and four nuclear markers to reconstruct dated phylogenetic trees and to test taxonomic and biogeographic hypotheses. Our analyses strongly support the monophyly of Neostasina and the polyphyly of Olios, with a new clade endemic to the Caribbean. Both Neostasina and Caribbean Olios occur on the Greater and Lesser Antilles and independently colonized the Caribbean around 36–28 mya. Hypothesis testing in BioGeoBEARS suggests a role of the GAARlandia landbridge in the colonization of both clades. The ‘Olios-like’ clade, in contrast, is restricted to the southern Lesser Antilles and shows a biogeographic history consistent with colonization from S. America, probably within the last 10 my. Thus, many spider lineages on the Greater Antilles seem to have colonized the Caribbean during a relatively short time span approximately coinciding with the proposed timing of GAARlandia. The synchronous colonization of multiple lineages suggests a temporary land connection. However, the main problem in concluding synchronous events across lineages in this study, as in most others, is the ambiguity in chronogram analyses meaning that many different patterns can be ‘consistent’ with GAARlandia, thus potentially providing a false positive result. Broad comparative biogeographical studies such as the CarBio project will offer the best opportunity to multiply test shared biogeographic patterns among independent lineages. The current paper contributes evidence from multiple lineages that will contribute to this synthesis.