Refining species distribution model outputs using landscape-scale habitat data: Forecasting grass carp and Hydrilla establishment in the Great Lakes region

Abstract

Forecasts of the locations of species invasions can improve by integrating species-specific climate and habitat variables and the effects of other invaders into predictive models of species distribution. We developed two species distribution models (SDMs) using a new algorithm to predict the global distributions of two nonindigenous species, grass carp (Ctenopharyngodon idella) and Hydrilla (Hydrilla verticillata), with special attention to the North American Great Lakes. We restricted the projected suitable habitat for these species using relevant habitat data layers including accumulated Growing Degree Days (GDD), submersed aquatic vegetation (SAV), wetlands, and photic zone. In addition, we restricted the grass carp niche by the projected Hydrilla niche to explore the potential spatial extent for grass carp given a joint invasion scenario. SDMs showed that climate conditions in the Great Lakes basin were often suitable for both species, with a high overlap between the areas predicted to be climatologically suitable to both species. Restricting Hydrilla regions by GDD and photic zone depth showed that the nearshore zones are primary regions for its establishment. The area of predicted habitat for grass carp increased greatly when including Hydrilla niche as a potential habitat for this species. Integrated risk maps can provide a means for the scientifically informed prioritization of management resources toward particular species and geographic regions.