GIS-Generated, Expert-Based Models for Identifying Wildlife Habitat Linkages and Planning Mitigation Passages

We developed three black bear (Ursus americanus) habitat models in the context of a geographic information system to identify linkage areas across a major transportation corridor. One model was based on empirical habitat data, and the other two (opinion- and literature-based) were based on expert information developed in a multicriteria decision-making process. We validated the performance of the models with an independent data set. Four classes of highway linkage zones were generated. Class 3 linkages were the most accurate for mapping cross-highway movement. Our tests showed that the model based on expert literature most closely approximated the empirical model, both in the results of statistical tests and the description of the class 3 linkages. In addition, the expert literature-based model was consistently more similar to the empirical model than either of two seasonal, expert opinion-based models. Among the expert models, the literature-based model had the strongest correlation with the empirical model. Expert-opinion models were less in agreement with the empirical model. The poor performance of the expert-opinion model may be explained by an overestimation of the importance of riparian habitat by experts compared with the literature. A small portion of the empirical data to test the models was from the pre-berry season and may have affected how well the model predicted linkage areas. Our empirical and expert models represent useful tools for resource and transportation planners charged with determining the location of mitigation passages for wildlife when baseline information is lacking and when time constraints do not allow for data collection before construction.