Modeling landscape connectivity for bobcats using expert-opinion and empirically derived models: how well do they work?

Efforts to retain ecological connectivity have become a conservation priority to permit animal movements within home ranges, allow dispersal between populations and provide opportunities for animals to respond to climate change. We used expert-opinion and empirically derived models to investigate landscape connectivity at two spatial scales among bobcats Lynx rufus in New Hampshire, USA. Paths of marked bobcats were compared to random movements in the context of program CIRCUITSCAPE. At the local scale (within home ranges), the empirical model (based on observations and telemetry locations) performed better than the expert-opinion model. At the regional scale (state of New Hampshire), both models identified urban development as a potential barrier; however, the models differed in predicting how specific natural features (e.g. mountains and large water bodies) and some roads affected bobcat movements. When compared with bobcat population structure based on genetic information, the expert-opinion model overestimated the influence of roads. Alternatively, the empirical model overestimated the influence of snow. Our findings indicate that the empirically based resistance model was better at describing landscape-scale effects, whereas the expert-opinion model provided a good understanding of gene flow at a regional scale. As such, both models may be considered complementary. Bobcats were sensitive to disruptions imposed by habitat fragmentation and thus may be a suitable focal species for evaluating the consequences of land-use changes on the regional suite of mesocarnivores.