Spatial structure and invasion success of populations undergoing range expansion: Effects of dispersal strategy and landscape heterogeneity

This dissertation explores the effects of dispersal strategy and landscape heterogeneity on the spatial structure and invasion success of populations undergoing range expansion. First, the spatial structure of an invasive population was evaluated using semivariogram analysis (a characterization of the dependence of data on spatial location). Twelve years (1985--1996) of historic gypsy moth (Lymantria dispar) monitoring records from the Lower Peninsula of Michigan were analyzed. Seven models were statistically fit to annual plots of semivariance against distance at two different spatial extents. The model with the lowest Akaike's Information Criterion value was chosen to characterize overall semivariogram behavior for each year and spatial extent. At large spatial extents, three distinct patterns of semivariogram behavior were observed as the invasion progressed. This three-stage semivariance progression pattern likely represents the invasion stages of establishment, expansion, and stabilization and may be indicators of important mechanistic changes in invasion dynamics.; Second, the relationship between landscape heterogeneity and invasion success of the gypsy moth in Michigan was analyzed by regressing invasion success on seven measures of landscape fragmentation and by identifying potential threshold responses (sudden, nonlinear changes) in invasion success with habitat loss. Invasion success thresholds were then compared with that of seven landscape structure metrics. Both patch- and gap-based landscape metrics exhibited threshold responses. However, invasion success increased linearly with increasing proportion of habitat, indicating that the negative effects of habitat fragmentation do not compound that of habitat loss. The absence of detectable thresholds in invasion success suggests that aerial dispersers like the invasive gypsy moth may not be as strongly affected by habitat fragmentation as previously thought.; Lastly, semivariogram analysis was used to evaluate patterns of spatial autocorrelation generated by simple and stratified dispersal models. Dispersal was simulated across a homogeneous landscape composed of 100% habitat and a real, heterogeneous landscape composed of 54% habitat. Population growth rate and stratified dispersal distributions were varied as well. Semivariograms were calculated from model simulations at every other time step and fit with four different models to characterize changes in spatial structure of the population over time. Both simple and stratified dispersal modeled across real, heterogeneous landscapes generated semivariogram progression patterns similar to that of gypsy moths. Models of dispersal across 100% habitat landscapes and models of stratified dispersal with a long-tailed dispersal distribution across real landscapes generated semivariance progression patterns dissimilar to that of gypsy moths. Thus, landscape heterogeneity (and not dispersal strategy) was a major driving force behind semivariance patterns observed during the gypsy moth invasion of Michigan.