Predictive habitat selection modeling for an endangered insect species across a Wisconsin landscape

Selecting appropriates sites to re-establish or promote an endangered insect species is a critical part of conservation and restoration. However, this is difficult when considering the sheer scale of the landscape and the complexities of habitat selection dynamics. Predicting habitat selection requires an understanding of the species in question, the area being considered, and isolation of focal environmental characteristics such as geographic barriers and vegetation preferences that primarily determine how a species selects habitat patches. Habitat selection at the landscape-scale was predicted  for an endangered lepidopteran,  Lycaeides melissa samuelis Nabokov,  using multi-criteria risk models. These models were developed by assigning ‘risks’ to focal environmental characteristics such as soils, canopy cover, hydrology and corridors. Areas with known L. melissa populations were used to both train and test the models across the range in Wisconsin. Using risk models to predict habitat selection at the landscape-scale could act as a first step in restoration and conservation efforts, focusing management efforts on areas with the greatest chance of successful endangered species establishment.