Abstract
To determine the evolutionary relationships within the Anolis cristatellus series, we employed phylogenetic analyses of previously published karyotype and allozyme data as well as newly collected morphological data and mitochondrial DNA sequences (fragments of the 12S RNA and cytochrome b genes). The relationships inferred from continuous maximum likelihood reanalyses of allozyme data were largely poorly supported. A similar analysis of the morphological data gave strong to moderate support for sister relationships of the two included distichoid species, the two trunk-crown species, the grass-bush species A. poncensis and A. pulchellus, and a clade of trunk-ground and grass-bush species. The results of maximum likelihood and Bayesian analyses of the 12S, cyt b, and combined mtDNA data sets were largely congruent, but nonetheless exhibit some differences both with one another and with those based on the morphological data. We therefore took advantage of the additive properties of likelihoods to compare alternative phylogenetic trees and determined that the tree inferred from the combined 12S and cyt b data is also the best estimate of the phylogeny for the morphological and mtDNA data sets considered together. We also performed mixed-model Bayesian analyses of the combined morphology and mtDNA data; the resultant tree was topologically identical to the combined mtDNA tree with generally high nodal support. This phylogenetic hypothesis has a basal dichotomy between the Hispañolan distichoids and the bimaculatus series, on the one hand, and the cristatellus series inhabiting the Puerto Rican Island Bank, its satellite islands, the Bahamas, and St. Croix, on the other. The trunk-crown species form a clade, while the trunk-ground and grass-bush species do not as A. gundlachi, a trunk-ground species, is nested within a clade of grass-bush species. The patterns of relationships among the trunk-ground and grass-bush species suggest that one of these ecomorphs may have been ancestral to the other and that one or both evolved convergently. In the context of our preferred phylogeny and divergence dates estimated by NPRS analyses, we propose several biogeographical hypotheses that explain the current distribution of the cristatellus series. The presence of endemic species on the islands of the Bahamas, Desecheo, Mona, Monita, and St. Croix are likely due to over-water dispersal. Vicariance resulting from Pliocene or Pleistocene changes in sea levels likely explains the occurrence of A. cristatellus (including A. ernestwilliamsi), A. pulchellus, and A. stratulus on different islands of the Puerto Rican Bank.