Modelos nulos, efeitos estocásticos e padrões de distribuição e abundância de espécies em paisagens fragmentadas

Abstract

The loss and fragmentation of natural habitats reduce large, continuous areas and isolate them into smaller patches of habitat, creating fragmented landscapes. This land-use conversion affects species survival and alters the pattern of species distribution and abundance in the landscape. Some hypotheses based on ecological effects have been suggested to understand the processes behind these patterns, such as immigration and extinction, resource availability, among others. However, hypotheses based on simple random processes are alternatives already recognized in some studies. A major challenge for Landscape Ecology is to assess the relative contribution of ecological and probabilistic processes as determinants of these patterns of species richness and abundance in fragmented landscapes. As a way to meet these demands and advance in filling these knowledge gaps, this study aims to quantify the importance of random effects on diversity patterns in fragmented landscapes. To this end, this study was structured in three chapters where: (1) the contribution and efficiency of the area-based (or Coleman´s) random placement model was investigated, through an investigation of its practical application in published studies on the species-area relationship; (2) a new approach was taken by adapting the null model discussed above and adding spatial parameters from fragmented landscapes to predict patterns of species richness and composition using empirical data from small non-flying mammals; and, finally,(3) this same database was used to test another approach already known in macroecology, but relatively new at the landscape scale, through a model called Landscape Model. In the first chapter, it was detected that Coleman's random placement model was little considered in the literature, although efficient in predicting the richness pattern in almost half of the case studies evaluated. A clear relationship was evidenced between the degree of explanation of the model with taxon (favoring plants more than animals) and latitude. In the second chapter, it was possible to partially explain the richness and composition of small mammal species through null models based on individuals or spatial parameters of the landscape. These models also showed how patterns may simply reflect the spatial configuration of the landscape. In the third chapter, models based on the individual home range generated similar response patterns based on both a discrete landscape (MPD) and continuous (MPC) approach that considers habitat quality. Both models were able to partially explain small mammal abundance and richness, however only the MPC had no statistical bias. The models developed in this thesis represent new theoretical and methodological advances on the causal processes of the pattern the distribution and abundance of species at the landscape scale, as they offer new models that allow exploring and estimating the relative contribution of these probabilistic effects. Understanding the processes behind these patterns will help in future strategies of species conservation in fragmented landscapes