Conventionally, species dispersal has been conceptualized as the result of natural processes based on life history traits and habitat suitability as it relates to individual species’ constraints. However, if human transport mechanisms dominate over individual species’ processes, then human agency may be overshadowing traditional ecological mechanisms of dispersal, such as natural flight capacity, wind-driven dispersal, and community assembly mechanisms. We hypothesized that across invasions, natural ecological processes are essentially being overridden by anthropogenic ones, and that predictable generalities that operate across entire suites of species arise as a consequence of these processes’ broad effects. We previously demonstrated support for this hypothesis by creating a general dispersal model for all United States invasive forest pests (generalized dispersal kernel, (GDK)), which explained 76% of the variation in their range extent, with 74% locational accuracy. However, it was not clear how much predictive power was lost by building this general model, compared to building customized models fit to individual focal species. To examine this question, we built customized spread models for three invasive forest pests (beech bark disease (Cryptococcus fagisuga), gypsy moth (Lymantria dispar), and hemlock woolly adelgid (Adelges tsugae)), and compared the predictive power of these more complex models to the GDK.
Results/Conclusions
By adding a simple correction factor to the intercept of the GDK based on the residuals of each species' predicted vs. observed spread radius in our original model, we were able to improve the GDK's forecasting ability such that it roughly matched that of the single-species models (within ±9% of the spatial variation explained, relative to a null model of spatial dispersal predictions), and outperformed the complex model in one case. The mean strength of spatial predictions across species for a 5-year forecasting horizon was 69% of the spatial variation explained. Forecasting ability remained across a 30-year horizon (62% spatial variation explained), though it dropped disproportionally for hemlock woolly adelgid (-22% spatial variation explained). Our results suggest that general models can be adjusted based on simple one-dimensional theoretical predictions in order to respond quickly to invasions, and they provide additional evidence for a macroecological theory of invasive species dispersal.