Print

Williams, Sara H.

Natural Resources

2013

MS

Habitat fragmentation is one of the most challenging issues facing terrestrial mammal conservation. I used detection-non detection data collected at forest fragments (n=46) in the Champlain Valley, Vermont during 3 surveys from August 2011 to January 2012 in an occupancy modeling framework to examine the effects of forest fragment and surrounding landscape characteristics on the occurrence of fisher (Martes pennanti). I developed 26 potential hypotheses to explain occupancy and model selection techniques to assess which best represented the data. I detected fishers during 19% of surveys (26/136) and at 35% of sites (17/48). Among the variables I measured, the most important included fragment area and the amount offorest cover surrounding a fragment based on variable importance weights. Five models showed strong empirical support, including fragment area*surrounding pasture, patch ruggedness+surrounding forest, patch use by humans*surrounding forest, surrounding forest, and patch ruggedness*surrounding forest. Estimates of detection probability indicated that periods of 80-days would be required to achieve a 50% probability of detection. The results suggest that area and isolation are important factors affecting fisher occupancy, but that landscape context should also be considered from a management perspective. The top models indicated that interactions between fragment characteristics and the landscape were meaningful and provide tools for landscape conservation planning.