Predicting establishment and spread of invasive species is critical to informing management guidelines for early detection and effective eradication strategies. Predicting the potential range of an introduced species is difficult since, for most species, invaded ranges are dynamic and are a function of both species tolerance limits and attributes of the recipient environment. To overcome these challenges, we examine the spread of invasive species using spatially explicit hierarchical Bayesian models and include data from both the invasive and native species’ ranges. Key environmental variables, including climate data and land-use-land-cover (LULC) from both the native and introduced ranges along with local habitat type, and canopy closure in the invasive range are contained in the models. These variables are coupled to a neighborhood-based site colonization probability to predict potential species distributions in New England.
Results/Conclusions Widespread New England invasive species, including Fallopia japonica, Lonicera japonica, and Rosa multiflora, demonstrate that when abundant presence-absence data are available in both native and introduced ranges, quality predictions can be made at the species-level. Newly introduced or sparsely distributed species represent a unique challenge in prediction quality and model validation due to limited information from the introduced ranges. However, given climate, LULC, and abundant presence/absence data for multiple species in their native ranges (specifically Japan), adequate prediction of potential distributions in northeastern
