Abstract

Phylogenetic relationships among sloths (Folivora) have been extensively studied in the past few decades using maximum parsimony approaches. Recently, Bayesian phylogenetic methods also began to be employed for this task, with advances in methods for data partitioning and tip-dating analyses leading to exciting new possibilities in morphological phylogenetics. In this context, we assembled the largest morphological data set ever applied to sloths and reassessed their phylogeny and divergence times, evaluating alternative models of partitioning and dating in a Bayesian framework. The updated phylogeny of sloths is largely in agreement with previous morphological studies, with Bradypus recovered as sister to Eufolivora, the presence of two major sloth clades (Mylodontoidea and Megatherioidea) and Choloepus among Megalonychidae. However, the present study yields some important advances in understanding the relationships of genera with historically unresolved or controversial allocations. The major sloth clades diversified from the Late Eocene to the Early Miocene. Homoplasy-based partition models outperformed anatomical partitioning and unpartitioned analyses, with considerable impacts on topology and posterior probabilities. Estimates obtained using homoplasy-partitioned models with Bayesian analyses were in strong agreement with those of maximum parsimony. We emphasize the importance of model comparison with Bayes factors and the assessment of synapomorphies in Bayesian morphological phylogenetics.