Frontiers in invasive species distribution modeling (iSDM): Assessing effects of absence data, dispersal constraints, stage of invasion and spatial dependence

Successful management of biological invasions depends heavily on our ability to predict their geographic ranges and potential habitats. Species distribution modeling (SDM) provides a methodological framework to predict spatial distributions of organisms but the unique aspects of modeling invasive species have been largely ignored in previous applications. Here, three unresolved challenges facing invasive species distribution modeling (iSDM) were examined in an effort to increase prediction accuracy and improve ecological understanding of actual and potential distributions of biological invasions. The effects of absence data and dispersal constraints, stage of invasion, and spatial dependence were assessed, using an extensive collection of field-based data on the invasive forest pathogen Phytophthora ramorum. Spatial analyses were based on a range of statistical techniques (generalized linear models, classification trees, maximum entropy, ecological niche factor analysis, multicriteria evaluation) and four groups of environmental parameters that varied in space and time: atmospheric moisture and temperature, topographic variability, abundance and susceptibility of host vegetation, and dispersal pressure. Results show that incorporating data on species absence and dispersal limitations is crucial not only to avoid overpredictions of the actual invaded range in a specific period of time but also for ecologically meaningful evaluation of iSDMs. When dispersal and colonization cannot be estimated explicitly, e.g. via dispersal kernels of propagule pressure, spatial dependence measured as spatial autocorrelation at multiple scales can serve as an important surrogate for dynamic processes that explain ecological mechanisms of invasion. If the goal is to identify habitats at potential risk of future spread, the stage of invasion should be considered because it represents the degree to which an organism is at equilibrium with its environment and limits the extent to which occurrence observations provide a sample of the species ecological niche. This research provides insight into several key principles of the SDM discipline, with implications for practical management of biological invasions.