Challenges of modeling invasive species spread

The idiosyncratic nature of the invasion process, where both historical and local conditions affect the likelihood of establishment, makes generalization of the invasion process very difficult and hampers our ability to generate guidelines for the control of potentially invasive species. To develop reliable models of future plant invasions in New England we studied the past patterns of plant spread into the region. Information on invasive plant species presence and abundance, were gathered for more than 4200 locations in New England by volunteers working with the Invasive Plant Atlas of New England (IPANE). This data set was combined with climate, land-use and other environmental variables to construct a hierarchical Bayesian (HB) model describing the invasion process. HB modeling allowed us to incorporate data collected at different spatial and temporal scales, and more importantly, to deal with the fact that the spread of recently introduced species is an ongoing process. Results for three focal invasive species showed that Celastrus orbiculatus (vine) seems to have invaded most of the suitable area, while Berberis thunbergii (shrub) and Elaeagnus umbellata (small tree) could still expand their current distributions. Here we were able to identify areas that are prone to invasions in the region, predict future spread within the region, and single out the set of environmental conditions that make an exotic species able to become an invader.