Duas Abordagens na Modelagem da Distribuição de Aves na Amazônia: Áreas de Endemismo versus Variáveis Abióticas

Abstract

The birds besides being one of the most representative taxonomic groups of the Amazon, are not yet known aspects related with its the distribution and ecological and historical factors that can determine it. In general, to model their distribution are used climatic and environmental variables depending on the scale, that can not clearly explain its distribution. Thus, the aim of this study was to determine how models based on three different approaches, predict the spatial distribution of species of Amazonian birds in areas with different biogeographic characteristics. First was made species distribution models using layers of polygons of Amazon Birds endemism areas and distribution of flooded environments. Later, were built models with climate and topography variables, using the mathematical algorithm of maximum entropy (MAXENT). There were no significant differences between the two approaches to Topaza pyra, Rhegmatorhina gymnops, Touit huetii, Lophotriccus galeatus and Knipoleugus orenocensis and could not identify a pattern to identify in general which the predictions were the most likely. It was found that for T. pyra, R. gymnops and T. huetii models made with areas of endemism (MAE) were more likely than models with abiotic variables (MVA). In turn, the abiotic variables models (MVA) were more likely than endemism areas models (MAE) for L. galeatus, T. pyra, R. melanosticta and Knipolegus orenocensis. In general all models showed high performance values, and only for Heliodoxa schreibersii and Synallaxis propinqua the failure rate was higher than the sensitivity. We conclude that in general none of the two approaches were more likely than the other. Still you need to understand how they interact the two approaches in traditional sets models, extending the analysis to other species and achieve greater quantity and quality of data and environmental information to make more accurate and reliable inferences about the predictions of distribution and draw definitive conclusions.