A LANDSCAPE SIMULATION-MODEL OF WINTER FORAGING BY LARGE UNGULATES

Ungulate winter grazing was simulated on simple random and actual landscape patterns using an individual-based modeling approach to explore the effect of landscape heterogeneity on foraging dynamics. The landscape was represented as a 100 x 100 grid with each cell considered to be either a resource or nonresource site. Random maps were generated by specifying the proportion, p, of the landscape occupied by resource sites. Actual landscape maps were obtained from the spatial arrangement of sagebrush-grassland habitats in subsections of northern Yellowstone National Park, Wyoming. Each resource site was assigned an initial forage abundance, and a specified number of ungulates were distributed randomly across the landscape on resource sites. Three alternative search-and-movement rules, which incorporated different movement scales and assumptions about ungulates' knowledge of the landscape, were compared. Grazing was simulated as a recipient-determined, donor-controlled flow with a nonlinear feedback. Daily energy balances were computed for each ungulate by subtracting energy cost from energy gain, and ungulates were assumed to die when they reached 70% of their lean body weight. Simulation results suggested that when resources were abundant across the landscape (i.e., high p), the search-and-movement rule selected to simulate foraging was not important. That is, a variety of strategies should suffice under high-resource conditions, and there was no benefit to having a more efficient rule. However, when resources were scarce (e.g., low p or high ungulate densities), then the ability to discern resource abundances and to move over greater distances resulted in lower mortality. For a given p, the difference between a fragmented (i.e., random) and aggregated (i.e., real) arrangement of resource sites was only pronounced when the ability of the animals to move was very limited. In these situations, survival was always greater in the real landscapes. Individual-based spatial models developed for specific landscapes and species could be quite valuable in enhancing our understanding of landscape dynamics and large herbivores.