218 Evaluating the scale of resource use by beef cows on winter range.

Abstract Foraging animals select for resources across spatial scales, however, information quantifying this is limited for livestock species. Therefore, this study evaluated the relative selection for resources across multiple spatial scales. Over the course of two winters, a herd of 140 mature, Angus-based cows grazed a 645-ha native rangeland pasture in southwestern Montana. Each winter, 17 cows were randomly selected and equipped with global positioning system (GPS) collars and monitored between December and February. For each animal, we generated a utilization distribution using GPS locations identified as occurring during grazing activity. We obtained gridded, 30-m resolution data on pasture attributes, including topography (elevation, slope, terrain roughness, aspect), vegetation (cover and biomass by plant functional group), and distances to water and supplement sites. To create coarser-scale data for each attribute, we used a moving window to calculate the mean value up to a 1000 m neighborhood size for each cell of the grid. Then, for each metric, we built a multi-level resource utilization function that calculated the selection coefficient for each animal at each scale. To evaluate how population-level selection varies across scales, we analyzed selection coefficients with a set of candidate linear mixed-effect models. Of the 18 attributes evaluated, 17 showed a significant relationship between scale and resource selection (P ≤ 0.03). Fourteen of the attributes indicated strongest support for a second- or third-order quadratic relationship. The remaining four showed strongest support for a linear relationship. Cows selected against the topographical attributes of elevation, slope and terrain ruggedness across all scales (P < 0.01) and most strongly at coarser scales (620-1000 m resolution). Additionally, cows selected for directional components of aspect at coarser scales (780-1000 m resolution; P < 0.01). Cows selected for annual and perennial herbaceous cover at fine scales (30-110 m resolution; P < 0.01). Cows selected against bare ground, litter, and tree cover across all scales and (P < 0.01) most strongly at the coarsest scale considered (1000 m). Cows most strongly selected for perennial herbaceous biomass at the finest scale considered (30 m; P < 0.01) and for annual herbaceous biomass at a moderate scale (620 m; P < 0.01). Cows selected against distance from water and supplement across all scales (P < 0.01). The intraclass correlation coefficient (ICC) was highest for distance attributes (ICC = 0.99), followed by topographical attributes (ICC = 0.45), and lowest for vegetation attributes (ICC = 0.23), indicating that selection for vegetation attributes was more consistent among individuals, whereas selection for distance attributes was least consistent. This information provides insight for rangeland scientists and livestock managers to inform the development of resource use models and evaluate the effective scale for various rangeland improvements.