The traditional models used to characterize animal home ranges have no mechanistic basis underlying their descriptions of space-use, and as a result, the analysis of animal home ranges has primarily been a descriptive endeavor. In this paper, we characterize coyote (Canis latrans) home range patterns, using partial differential equations for expected space use that are formally derived from underlying descriptions of individual movement behavior. To our knowledge, this is the first time mechanistic models have been used to characterize animal home ranges. The results provide empirical support for a model formulation of movement responses to foreign scent-marks, and suggest that having relocation data for individuals in adjacent groups is necessary to capture the spatial arrangement of home range boundaries. We then show how the model fits can be used to obtain predictions for individual movement and scent-marking behavior and to predict changes in home range patterns. More generally, our findings illustrate how mechanistic models permit the development of a predictive theory for the relationship between movement behavior and animal spatial distribution.