Predicting invasive plant species range expansion in Alaska

Alaska is vulnerable to a variety of changes causing by a warming climate. Alaska is no longer thought to be immune to wide scale invasive plant species infestations. Planning tools are needed to anticipate area of potential change and to identify invasive species of concern. I conducted a field study to determine presence or absence of any non-native vascular plant species per 100 m of transect keyed to vegetation type, canopy cover class, aspect, visitor use level, and use intensity on all major trails on and near Forest Service lands on the Kenai Peninsula, Alaska. Trailheads are thought to be sites of non-native species introduction to trail systems. Little is known about the number of introductions needed to establish a non-native species in northern climates, or the expected distances a particular species can be expected to spread up a trail. Prediction of non-native species spread along Alaskan trails is possible by examining vegetation type and intensity of use, with greatest impacts in open types with high use. I used biogeoclimatic models to forecast potential vulnerabilities with respect to invasive species distribution in Alaska. I selected three invasive plant species of interest in Alaska with different current distribution (reed canarygrass, present and widespread; purple loosestrife, present and limited; and leafy spurge, not yet present but considered potentially invasive). Species were modeling using two different predictive models (BIOCLIM in the DIVA-GIS platform and MaxEnt), two different future climates (Hadley and CCC), two emissions scenarios (A2, high and B2, low), for current climate plus three time steps (2020, 2050, 2080). Models were assessed with 25% test data, and then trained with 75% of the data. MaxEnt models performed better than DIVA-GIS models. All models showed current potential range that exceeds their known occurrence. For each species, we compared area difference in predicted habitat suitability between scenarios and between time steps in both models as a quantified measure of potential habitat change. I applied the modeling procedure to an additional twenty-four species to create an atlas of scenario maps for Alaska. All models showed current potential range that exceeds their known occurrence.