Combining molecular and morphological information in a new molecular clock for tortoises (Testudines: Pan-Testudinidae)

JUAN ALFREDO HOLLEY; VLACHOS EVANGELOS

Congreso; Congreso Argentino-Paraguayo de Herpetologia; 2016

Asociación Herpetológica Argentina - Paraguaya

The use of fossils in tree calibrations has been rigorously debated, mainly as the result of the uncertainty of the phylogenetic position and age of extinct taxa. Also, the addition of fossil information can be done with various methods, reaching sometimes significantly different results. Here we present our methodological comparative study in which we estimate the divergence times of a new tortoise phylogeny (Testudines: Pan-Testudinidae) based on a broad sampling of extant and extinct taxa. Our comparison includes the exploration of different molecular clock approaches (e.g. node- and tip-dating) as well as the performance of different methods (e.g. maximum parsimony and Bayesian inference) in the phylogenetic resolution when facing combined data. Our total-evidence matrix includes 84 taxa, 170 morphological and 5,735 molecular characters (including 4 mitochondrial and 3 nuclear genes). The divergence estimations over the Bayesian phylogeny were performed with BEAST2 under the relaxed uncorrelated lognormal clock model and the fossilized birth-death (FBD) tree model, and the parsimony analysis was performed using TNT. Our results show that the early evolution of testudinids occurred in the early Cenozoic, whereas the split and early diversification of the main clades of crown Testudinidae in the Eocene. The inclusion of extinct taxa allows us to estimate the timing of diversification of the stem Testudinidae to the Late Cretaceous. The comparison of the phylogenetic results suggests that the lack of molecular information in some taxa has a significant impact in the Bayesian inference, making difficult the convergence of some parameters and producing unlikely solutions.