• cervus elaphus;
  • diffusion process;
  • potential functions;
  • random vector field;
  • splines;
  • telemetry data


We describe the use of bivariate stochastic differential equations (SDE) for modeling movements of 216 radio-collared female Rocky Mountain elk at the Starkey Experimental Forest and Range in northeastern Oregon. Spatially and temporally explicit vector fields were estimated using approximating difference equations and nonparametric regression techniques. Estimated vector fields of movement were mapped onto the project area at selected times of the day to examine spatial patterns of movement in relation to topography. Using the concept of a potential function, we were able to study the influence of roads and grassland foraging areas on elk movements. Doing so we identified broad spatial patterns of elk movements and showed the time dependent effects of habitat features within the habitat mosaic at Starkey. Our analyses quantify the cycles of movements in spring and summer in terms of attraction or repulsion to specific habitat features, and illustrate the magnitude, timing and direction of these movements. An extensive list of references is included. Published in 2004 by John Wiley & Sons, Ltd.