Pedro R. Peres-Neto, University of Quebec at Montreal
Ecologists increasingly consider how local and regional factors affect patterns of movement and extinction-colonization across the landscape, which in turn shape the structure of local metacommunities. A variety of statistical models and a number of stochastic patch occupancy models can be used to models dispersing populations. However, these frameworks are designed to model single species in the landscape, precluding the analysis of entire communities. Analytical frameworks for environmental gradient analysis such as redundancy analysis and canonical correspondence analysis can handle multiple species, but they are also limited as they cannot consider dispersal predictors and site isolation. The goal of this study is to develop a quantitative framework for estimating the relative importance of local habitat suitability, species spatial distribution and dispersal predictors in dictating metacommunity patterns in fragmented landscapes. The framework also controls for the effects of spatial autocorrelation while estimating the importance of these ecological factors. The proposed approach will provide output results similar to those of canonical analysis where species can be contrasted according to both their habitat affinities and dispersal capabilities.