Many vertebrate species can use their cognitive abilities to make rewarding decisions when moving across the landscape. Because of uncertainty about the spatial distribution of high-quality patches, animals could benefit from returning to previously visited patches (i.e., known high-quality patches) after visiting a sequence of poor-quality patches. Such use of spatial memory could be a key determinant of animal spatial dynamics and home range formation, where site fidelity emerges as a product of memory-based patch selection under uncertainty. Yet, field evidence of how animals use spatial memory remains limited, especially for free-ranging animals in natural settings. Our objective was to test if past experience influenced selection of resource patches (i.e., meadows) by free-ranging bison (Bison bison), including patches previously visited. We developed patch selection models accounting for different discounting functions of past experience using GPS location data from 22 female bison living in the meadow-forest matrix of Prince Albert National park, Saskatchewan.
Results/Conclusions
We found that bison selected relatively large meadows of intermediate mean plant biomass, and this selection varied with short-term (1-2 days) past experience. Selection strength for high-quality meadows (i.e., meadows of intermediate plant biomass) decreased when bison had recently visited high-quality meadows. Bison also selected meadows they had previously visited, and this selection was influenced by long-term previous experience. Indeed, as expected, bison were more likely to revisit a meadow (particularly meadows of high-quality) if they had experienced a run of poor-quality meadows. Our findings suggest that free-ranging animals rely on past experience to adjust patch selection decisions, and they use both short-term and long-term memory to select forage patches. Furthermore, incomplete information about the distribution and quality of food leads to a higher probability of return to previously visited patches, resulting in area-restricted space use, thereby revealing a possible mechanism of home range formation. These principles can be applied to individual-based models of animal movement to better simulate and test hypotheses about the spatial distribution of phylopatric species at broad scales.