Date: April 26, 2019, 14:00
Candidate: João Claudio de Sousa Nascimento
Advisor: Vera Nisaka Solferini
Pós-Graduação do Instituto de Biologia
Identifying the processes that were responsible for the diversification of Neotropical groups was traditionally mainly associated with identifying the period in which said diversification occurred. Groups that presented younger datings were supposed to have been influenced by the Quaternary climatic oscillations, while older dating were supposed to be associated with orogenic events that occurred in the Tertiary. Nevertheless, it was not well explored which specific traits of the organisms are associated with a younger or older dating. In this work, this question was approached by using a meta-analysis of the published works that presented an estimate of the oldest intraspecific lineages divergence for a number of species. The random forest algorithm, a machine-learning approach, was then used to perform the analysis and classification of these observations. This algorithm uses a set of predictor variables to construct a classification rule for the observations. In this process, it also allows the estimation of the importance of each predictor variable for the correct classification of the groups in these two periods. For the analysis, a dataset comprising of species with known dating and sampling localities was constructed from a literature survey. From this, the analysis was performed for all localities obtained, for the full dataset of organisms as well as for plants and animals separately, using a number of climatic, ecological and taxonomic characteristics as predictor variables in the algorithm and the period of diversification as the response variable. The results showed that the taxonomic rankings were the most important variables for the classification, which suggest an important role for natural history, morphological or physiological characteristics of the groups in shaping the way the diversification process occurs. Latitude and longitude were also among the most important variables, adding a geographical context to the diversification processes. The analysis that did not included taxonomic identified the form of locomotion in animals as highly important, probably due to its influence in the dispersal capability of organisms. For plants, the kind of gene used in the analysis was the most important variable, highlighting the carefulness that is necessary in making inferences from results from one of few genetic markers. The results here presented are a first attempt in trying to uncover how the traits of organisms influence the diversification patterns and genetic diversity distribution in groups of the Neotropical region.