Spatial Transferability Of Expert Opinion Models For American Beaver Habitat

Species distribution models and habitat suitability models (HSMs) have become a popular tool in the conser-vation of biodiversity. However, the ability to predict species spatial distributions at sites beyond the data source sites (i.e., spatial transferability) is critical for the applications of HSMs in the management and conservation of rare or endangered species. The main objective of our study was to assess the predictive performance and spatial transferability of expert opinion models (EOMs). To build EOMs, we identified through extensive literature re-views 17 key landscape variables to characterize habitat use by American beaver (Castor canadensis). We developed 31 pairwise opinion questions on the relative importance of the 17 selected habitat variables for an online survey in Qualtrics?. We used Saaty?s analytical hierarchy process (AHP) and geospatial analysis to build EOMs for beaver. We tested the transferability of EOMs by assessing model predictive performance using the area under the curve (AUC > 0.7) in northcentral Mississippi and northern Alabama, USA. Thirty-five of 63 survey participants submitted complete, consistent surveys. Expert opinion models had fair predictive performance for beaver at the two study sites (AUC = 0.70?0.76). The fair predictive performance of EOM for the two sites, from which no opinion survey data were collected, indicated acceptable spatial transferability. The American beaver exhibits stable realized niche space throughout its geographic range, restricting habitat selection to open water bodies and associated wetlands, which may subsequently result in high transferability of HSMs.