Mathematical models of short-range dispersal in defoliating lepidopterans

The gypsy moth (Lymantria dispar) is among the most destructive invasive species in North America, responsible for defoliating millions of hectares of oak forest. The spatial dynamics of defoliating lepidopteran populations, such as those of the gypsy moth, are thus of great interest to forestry and conservation efforts. Despite numerous studies on the long-range dispersal patterns of defoliators, there is comparatively little theoretical understanding or field research concerning short-range dispersal via ballooning. Previous studies of ballooning have assumed random diffusion, but these models cannot account for non-random biases, such as the effect of wind on the angle of dispersal.Here, I develop models of short-range dispersal in larvae via ballooning, informed by methods from the seed dispersal kernel literature. I then fit models to field data of gypsy moth larvae dispersal using MCMC to perform Bayesian inference, and PSIS-LOO to perform model selection. I found that dispersal kernel models are able to reliably detect biases in angle of dispersal due to wind direction, and allow for testing of correlations between experimental variables and measures of dispersal. These modeling methods can help inform future studies into short-range larval dispersal and provide a novel framework with which to analyze dispersal data.