Modern applied statistics and animal populations in geographically complex landscapes: Data visualization and modeling

This dissertation includes three papers. The first paper brings a methodology to apply exploratory tools of modern statistics such as interactive data visualization (brushing, identification), or bivariate scatterplots, and statistical simulation to extract populational characteristics of three species of small mammals living in the geographically complex landscape of the Savannah River Site (SRS). This first paper shows the importance of looking into statistical properties of the data analyzed, such as skewness of the empirical distribution of response variables to assess patterns observed through exploratory graphs. The second paper uses Monte Carlo simulation to investigate the behavior of metapopulations of two competing species in a temporarily dynamical environment through a spatially explicit model. In particular, this paper investigates the role of local ecological processes (within-patch populations) in the dynamics of a metapopulation in space and time. Competition between species proved to be sufficient to cause spatial segregation of species. The additional competition between immigrants and inhabitants of a patch enhances spatial segregation of species. A sensitivity analysis of the simulation model developed in the second paper follows in the third paper. The spatial parameter of the model showed to be the most influential one in simulation results. Further investigation of the role of the spatial parameter (average species migration distance) in species' metapopulation probability of extinction shows results where small changes in the spatial parameter produces abrupt shifts in the metapopulation persistence. This suggests that the problem of persistence of a metapopulation is analogous to a theoretical percolation problem in three dimensions. In addition, species ability to disperse shows to be the connectivity of a patchy landscape.