An important question in biotic invasions is why invaders spread more rapidly through particular environments than others. A number of challenges make this question difficult to answer. Monitoring spread over time and space is resource-intensive and may be unfeasible, and conventional methods of calculating spread rate from available data are not suitable for representing variability in spread rate at local scales. This presentation introduces local polynomial fitting as a method for deriving local estimates of invasion speed and direction. Using the gypsy moth as a model system, we demonstrate the utility of this method for detecting geographic variability in invasion rate and use geographically weighted regression (GWR) to explore how invasion rate is influenced by forest fragmentation (the cohesion index), host tree basal area, and human population density.
Results/Conclusions
Using local polynomial fitting, the invasion rate of the gypsy moth is shown to vary between 13 and 55 km yr-1. The mean invasion rate of 22.4 km yr-1 is consistent with previous region-wide estimates. GWR models showed that forest fragmentation, human population density, and host tree basal area all are good predictors of invasion rate; however, measures of model fit indicate that the combination of human population density and forest fragmentation provide the best predictions of gypsy moth invasion rate, explaining 89.7% of the variability. GWR also reveals that the relationship between invasion rate and predictor variables varies considerably over space, including a region where spread is faster in more fragmented areas (Virginia and West Virginia) and a region where spread is slower in more fragmented areas (the upper Midwest).