Linking local population structure to regional trends in community structure of Wisconsin upland forest plant communities

The decline of biodiversity through habitat loss and exotic invasions as a result of human activity is a global phenomenon. This research explores several aspects of how regional species invasions and declines recently observed in Wisconsin upland forests understories relate to multi-scale spatial distributions of individual species and uses local and regional abundance and site occupancy to test theories of predictability of metacommunity species dynamics.;I conduct a fine scale, spatially explicit study of the nature and strength of species-environment linkages and spatial patterns in populations of spring ephemeral and early summer herbs in a maple-basswood forest in southern Wisconsin. Species declining in abundance tended to be more consistently clustered at smaller scales and highly influenced by soil nutrient variables, while increasing persistent species were less consistent among populations both in environmental dependencies and spatial structure. Response to early spring microenvironments varies greatly by species, giving evidence of micro-scale spatial niche partitioning.;I revisit a technique to visualize the variation of species abundance over sites and expand upon it by quantifying these patterns. Frequency data for 89 species across 62 sites best fits a beta-binomial distribution, as predicted by theory regarding spatially aggregated populations at smaller scales. A spatial Heterogeneity Index calculated directly from the beta-binomial parameters for each species was highly correlated with Morisita's empirical index of dispersion, and may allow us to describe patterns in variation in abundance from several different types of data.;I expand upon nested subset theory to include frequency data, and show that this framework has some power to predict local changes in incidence and abundance. Models based simply on matrices of species incidence and frequency across sites serve to provide predictions of regional dynamics even in the absence of more detailed information on species traits or site characteristics.;I also begin to explore ways to use join-count statistics to quantify spatial distribution of species sampled with a bent-transect of contiguous quadrats. These spatially explicit data replicated over dozens of sites can provide us with an excellent opportunity to assess how local species patterns vary over the larger region.