| dc.description.abstract | The present study provides a new hypothesis of the evolutionary relationships of Glassfrogs (Centrolenidae) inferred from mitochondrial and nuclear loci, and addresses questions on the evolution and speciation of the group. Ingroup sampling includes 55.5% of the named taxa in the family, representing most of the phenotypic diversity described for the group. As outgroups, I included 35 species, most of which represent families traditionally associated with Glassfrogs. Gene sampling consisted of complete or partial sequences of three mitochondrial (12S, 16S, ND1) and three nuclear markers (c-myc exon 2, RAG1, POMC) for a total of ~4362 bp. Phylogenies were estimated using maximum parsimony, maximum likelihood, and Bayesian analyses for individual genes and combined datasets. The importance of analyzing mitochondrial and molecular datasets separately is discussed, with particular emphasis on the ways in which this approach clarifies interpretations of relationships within Glassfrogs and other Neotropical anurans. Based on the phylogeny obtained, I propose a new taxonomy of Glassfrogs and its sister taxon Allophryne ruthveni. This arrangement formalizes clades that have significant statistical support under Parsimony, Maximum Likelihood, and Bayesian inference criteria, and that, in most cases, are phenotipically diagnosable. Centrolenid diversity is arranged into two subfamilies and a total of 12 genera, seven of which are new. Using Maximum Likelihood and Parsimony, I explore the evolution of morphological and behavioral features that characterize this unique group of Neotropical anurans. Each trait that has been postulated to trace relationships unambiguously in this group turns out to have had a complex evolutionary history with multiple origins and losses. Complete ventral transparency has evolved multiple independent times under all models of reconstruction, even those biased toward single origins. I demonstrate that repeated evolution of transparency has a significant inverse correlation across phylogeny with the presence of iridophores, which are hypothesized to protect internal organs from detrimental effects of solar radiation and heat. Complex derived behaviors, such as deposition of eggs on undersides of leaves, have evolved at least four times, and may be correlated with evolution of parental care. The evolution of sexually dimorphic traits (i.e., humeral spines) is ambiguous and depends on the method of inference; however, a scenario with recurrent origins of spines, probably selected in response to male-male intraspecific competition, is favored. The effect of incomplete sampling on reconstruction of ancestral character states is considered by comparing topologies and patterns of character evolution derived from analyses of complete and pruned datasets. Given the results, I suggest that when working with relatively large groups (i.e., ? 90 species), complete taxon sampling may not be so critical for an accurate reconstruction of character evolution, as long as morphological/behavioral diversity of all major clades is represented. The biogeography of centrolenid frogs is partially resolved. Glassfrogs originated in South America and dispersed multiple times into Central America. The most likely scenario for the current distribution of the Centrolenidae suggests a Guianan origin with subsequent dispersal into the Amazonia and the Chocó. Once the land connection between South and Central America was complete, and before the uplift of the Eastern Andean Cordillera, Glassfrogs dispersed to Central America at least four times. The uplift of the Eastern Cordillera and the Mérida Andes, coupled with drops in temperature during Pleistocene climatic fluctuations, could have facilitated dispersal of cold-adapted species from the Guianas to the Cordillera de la Costa and then into the Northern Andes. Glassfrogs reached the Southern Andes via repeated dispersal from the Northern Andes. Finally, comparisons of phylogeny and species distributions strongly supports vicariance as the main speciation mechanism in centrolenid frogs. | |
