Gene Flow in Complex Landscapes: Testing Multiple Hypotheses with Causal Modeling

Predicting population-level effects of landscape change depends on identifying factors that influence population connectivity in complex landscapes. However, most putative movement corridors and barriers have not been based on empirical data. In this study, we identify factors that influence connectivity by comparing patterns of genetic similarity among 146 black bears (Ursus americanus), sampled across a 3,000-km(2) study area in northern Idaho, with 110 landscape-resistance hypotheses. Genetic similarities were based on the pairwise percentage dissimilarity among all individuals based on nine microsatellite loci (average expected heterozygosity=0.79). Landscape-resistance hypotheses describe a range of potential relationships between movement cost and land cover, slope, elevation, roads, Euclidean distance, and a putative movement barrier. These hypotheses were divided into seven organizational models in which the influences of barriers, distance, and landscape features were statistically separated using partial Mantel tests. Only one of the competing organizational models was fully supported: patterns of genetic structure are primarily related to landscape gradients of land cover and elevation. The alternative landscape models, isolation by barriers and isolation by distance, are not supported. In this black bear population, gene flow is facilitated by contiguous forest cover at middle elevations.