Bayesian total-evidence dating revisits sloth phylogeny and biogeography: a cautionary tale on morphological clock analyses.

Combining morphological and molecular characters through Bayesian total-evidence dating allows inferring the phylogenetic and timescale framework of both extant and fossil taxa, while accounting for the stochasticity and incompleteness of the fossil record. Such an integrative approach is particularly needed when dealing with clades such as sloths (Mammalia: Folivora), for which developmental and biomechanical studies have shown high levels of morphological convergence whereas molecular data can only account for a limited percentage of their total species richness. Here, we propose an alternative hypothesis of sloth evolution that emphasizes the pervasiveness of morphological convergence and the importance of considering the fossil record and an adequate taxon sampling in both phylogenetic and biogeographic inferences. Regardless of different clock models and morphological datasets, the extant sloth Bradypus is consistently recovered as a megatherioid, and Choloepus as a mylodontoid, in agreement with molecular-only analyses. The recently extinct Caribbean sloths (Megalocnoidea) are found to be a monophyletic sister-clade of Megatherioidea, in contrast to previous phylogenetic hypotheses. Our results contradict previous morphological analyses and further support the polyphyly of "Megalonychidae", whose members were found in five different clades. Regardless of taxon sampling and clock models, the Caribbean colonization of sloths is compatible with the exhumation of islands along Aves Ridge and its geological time frame. Overall, our total-evidence analysis illustrates the difficulty of positioning highly incomplete fossils, although a robust phylogenetic framework was recovered by an a posteriori removal of taxa with high percentages of missing characters. Elimination of these taxa improved topological resolution by reducing polytomies and increasing node support. However, it introduced a systematic and geographic bias because most of these incomplete specimens are from northern South America. This is evident in biogeographic reconstructions, which suggest Patagonia as the area of origin of many clades when taxa are underrepresented, but Amazonia and/or Central and Southern Andes when all taxa are included. More generally, our analyses demonstrate the instability of topology and divergence time estimates when using different morphological datasets and clock models, and thus caution against making macroevolutionary inferences when node support is weak or when uncertainties in the fossil record are not considered.

Correction: Eocene intra-plate shortening responsible for the rise of a faunal pathway in the northeastern Caribbean realm.

[This corrects the article DOI: 10.1371/journal.pone.0241000.].

Eocene intra-plate shortening responsible for the rise of a faunal pathway in the northeastern Caribbean realm.

Intriguing latest Eocene land-faunal dispersals between South America and the Greater Antilles (northern Caribbean) has inspired the hypothesis of the GAARlandia (Greater Antilles Aves Ridge) land bridge. This landbridge, however, should have crossed the Caribbean oceanic plate, and the geological evolution of its rise and demise, or its geodynamic forcing, remain unknown. Here we present the results of a land-sea survey from the northeast Caribbean plate, combined with chronostratigraphic data, revealing a regional episode of mid to late Eocene, trench-normal, E-W shortening and crustal thickening by ∼25%. This shortening led to a regional late Eocene-early Oligocene hiatus in the sedimentary record revealing the location of an emerged land (the Greater Antilles-Northern Lesser Antilles, or GrANoLA, landmass), consistent with the GAARlandia hypothesis. Subsequent submergence is explained by combined trench-parallel extension and thermal relaxation following a shift of arc magmatism, expressed by a regional early Miocene transgression. We tentatively link the NE Caribbean intra-plate shortening to a well-known absolute and relative North American and Caribbean plate motion change, which may provide focus for the search of the remaining connection between 'GrANoLA' land and South America, through the Aves Ridge or Lesser Antilles island arc. Our study highlights the how regional geodynamic evolution may have driven paleogeographic change that is still reflected in current biology.

Early Oligocene chinchilloid caviomorphs from Puerto Rico and the initial rodent colonization of the West Indies.

By their past and present diversity, rodents are among the richest components of Caribbean land mammals. Many of these became extinct recently. Causes of their extirpation, their phylogenetic affinities, the timing of their arrival in the West Indies and their biogeographic history are all ongoing debated issues. Here, we report the discovery of dental remains from Lower Oligocene deposits (ca 29.5 Ma) of Puerto Rico. Their morphology attests to the presence of two distinct species of chinchilloid caviomorphs, closely related to dinomyids in a phylogenetic analysis, and thus of undisputable South American origin. These fossils represent the earliest Caribbean rodents known thus far. They could extend back to 30 Ma the lineages of some recently extinct Caribbean giant rodents (Elasmodontomys and Amblyrhiza), which are also retrieved here as chinchilloids. This new find has substantial biogeographic implications because it demonstrates an early dispersal of land mammals from South America to the West Indies, perhaps via the emergence of the Aves Ridge that occurred ca 35-33 Ma (GAARlandia hypothesis). Considering both this new palaeontological evidence and recent molecular divergence estimates, the natural colonization of the West Indies by rodents probably occurred through multiple and time-staggered dispersal events (chinchilloids, then echimyid octodontoids (spiny rats/hutias), caviids and lastly oryzomyin muroids (rice rats)).