MULTISCALE PERSPECTIVES ON OCCUPANCY OF MEADOW JUMPING MICE IN LANDSCAPES DOMINATED BY AGRICULTURE

Previous studies of distribution and habitat use by meadow jumping mice (Zapus hudsonius) have focused on local habitat associations. Our objectives were to predict occupancy of meadow jumping mice and identify local-, patch-, and landscape-specific factors correlated with the species' occurrence. We livetrapped for 5 consecutive days at each of 922 sites during summers of 2001-2003 in thirty-five 23-km(2) landscapes located across the upper Wabash River basin, Indiana. At 474 of these sites (300 forest and 174 grassland), we measured local habitat variables and computed metrics of patch size, shape, and isolation. Meadow jumping mice were captured 242 times at 69 of the 474 sites. The species did not associate with grassland habitat more frequently than forest habitat and was slightly more likely to be captured in forested corridors than in forest patches. Occupancy models fitted separately for grassland and forest sites revealed positive effects of stern density <1 m (forest) and 1-2 m (grassland) aboveground, but no effects of patch attributes. Probability of detection increased with lower temperatures at forest sites. Surprisingly, the extent of grasslands in landscapes had no effect on overall occupancy rates. Instead, variation in mean site-specific occupancy estimates among landscapes was related to the amount of forest cover; more heavily forested landscapes exhibited greater overall occupancy probabilities. Tests failed to identify a critical threshold of forest cover below which occupancy declined at an accelerated rate. More powerful tests are needed, because the existence of a threshold could have important implications for long-term viability of meadow jumping mice across the basin.