When did anoles diverge? An analysis of multiple dating strategies.

Whereas most of the studies that discuss the evolutionary divergence of Anolis lizards have dated the clade's crown group in between 31 and 64 Ma, a single study has recovered a significantly older age for the same node (87 Ma). These differences also entail notable consequences on the preferred biogeographical hypothesis for the whole clade. Here we analyze a total of seven dating strategies by combining three calibration sources in independent BEAST runs to infer the most probable divergence timing for anole lizards (a mitochondrial rate for ND2 gene, the Anolis dominicanus fossil, and a group of fossils assigned to the Priscagamines, Iguanines, and Idontosaurus clades). Based on the estimated timing, we also addressed whether chronograms differ the most in deeper or shallower nodes by exploring the trend in the standard deviation of mean ages between chronograms across time. Next, we focus on the pattern for a single shallow node by hypothesizing the biogeography of the island-endemic Malpelo anole (Anolis agassizi), and evaluating the temporal congruence between the species' divergence and the island geology. The estimated set of ages suggests that anoles most likely diverged 72 Ma (71-73 Ma), with the crown group established around 58 Ma (51-65 Ma). Dispersal is therefore supported as the major driver in the biogeography of the group (and in Caribbean lineages in particular). Our analyses also indicated that (1) rate-based analyses pulled dates toward younger ages, (2) the differences in node ages between chronograms decrease towards the tips regardless of the position of the constrained node, and that (3) the estimated age for deep nodes (e.g. Anolis stem) is highly influenced when deep nodes are also constrained. The latter two results imply that the estimated age for shallower nodes is largely unaffected by the used temporal constraint. The congruence of all chronograms for the Malpelo anole also supports this finding. Anolis agassizi was found to have diverged before the emergence of Malpelo island in each analysis (anole: 19-31 Ma vs. Malpelo island: 16-17 Ma). Finally, we recommend when performing absolute dating analyses to first test for sequence saturation in the analyzed dataset (especially when calibrations are based on molecular rates). Our study also points out the importance of using multiple node constraints, especially when placed deeply in the tree, for fossil-based divergence dating analyses.