Dynamic distribution modelling using a native invasive species, the mountain pine beetle

Abstract

Efforts to minimize impacts of highly mobile insect pests in a warming environment are complicated by the dynamicity and uncertainty of their distributions. Tools that aid such management efforts are urgently required for successful outcomes. Dynamic species distribution models (DSDM) represent temporal variations in the environment that match the timing and location of species occurrence leading to improved predictions of species distributions while reducing over prediction. In contrast, static SDMs treat the environment and ecological niche of the species as essentially invariant, such as climatic data averaged over long reference periods, which can result in inaccurate assessments of a species' ecological niche. We modified the traditional static SDM framework such that it incorporates both dynamic biotic and abiotic predictors to account for temporal variability in the environment. We used the native-invasive mountain pine beetle (MPB; Dendroctonus ponderosae, Hopkins) as a model species. MPB is an irruptive forest insect native to western North America; however, epidemic populations have recently crossed the Rocky Mountains and invaded the province of Alberta, Canada, giving rise to concern that the beetle will spread into the transcontinental boreal forest. We compared the ability of static and dynamic SDMs to predict the potential distribution of MPB, including the impact of host tree availability, degree of native-novel host tree introgression and interannual variation in climatic variables. Evaluation results using independent test data revealed higher predictive capacity by DSDMs when compared with SDMs. DSDMs provided robust temporal trends of MPB potential distributions while generating predictions that support the known MPB distributional shifts. The DSDM framework will help species distribution modellers in understanding the distributional dynamics of mobile species in climatically variable regions.