Exploring morphological data partitioning and reassessing the phylogeny, divergence times and morphological evolution in Pilosa (Mammalia: Xenarthra)

Abstract

Morphological evolution is heterogeneous and analytical methods should take this into consideration in order to produce reliable inferences, both for phylogenetic analyses and for studies of morphological evolution based on discrete characters. Data partitioning is well understood for molecular datasets, but only recently has begun to be applied to morphology, and its impacts on phylogenetic and evolutionary inferences for this kind of data remain underinvestigated. This thesis evaluates the performance of morphological data partitioning with different approaches and goals, including simulations and systematic empirical explorations. This thesis also aimed to reassess the phylogeny, divergence times and morphological evolution of Vermilingua and Folivora, and these empirical investigations also work as case studies for evaluating the performance of morphological data partitioning and the use of morphology in dating analyses. Evidence from simulations and empirical data suggests that anatomical partitioning is not an efficient way to segregate characters according to their evolutionary rates, what can be achieved using homoplasy-based partitions in phylogenetic analyses. Nevertheless, anatomically-oriented partitions can be valuable when studying morphological disparity and evolutionary rates, allowing to investigate these patterns for different regions of interest, which could be obscured if unpartitioned datasets are considered. Phylogenetic investigations of Vermilingua and Folivora returned overall similar results with those previously obtained, but provided some novel evidences related to the position of contentious taxa, divergence time estimates and new synapomorphies for previously recognized clades. Some methodological insights were also obtained. For the study of Vermilingua, the importance of sampling additional taxa and characters and the presence of hidden phylogenetic signals in separate datasets only became evident when they were combined. For the phylogenetic study of Folivora, the consistency with previously published results was affected by the partitioning model applied, with homoplasy-based partitioning being preferred and leading to similar topologies to those obtained by maximum parsimony inferences. Both studies reafirmed the importance of model selection in Bayesian phylogenetics of morphological data. While investigating the patterns of morphological evolution in Folivora, partitioning of cranial and postcranial data was fundamental to shed light in the distinct evolutionary patterns in these subsets of characters, with cranial morphology mostly reflecting phylogenetic inertia, whereas for postcranium, ecological adaptations seem to have also played an important role, leading to convergences related to the functional morphology.