Decision-making in group foragers with incomplete information: Test of individual-based model in Geese

One important challenge of spatial ecology is to generate models linking individual behavior to population-level phenomena. Although animals often face great uncertainty regarding foraging patch quality, earlier models explaining the aggregation of animals have rarely specified how stable outcomes are achieved through individual decisions, especially under realistic assumptions for incompletely informed foragers. We developed a new foraging model that assumed a realistic decision-making rule for incompletely informed group foragers, and we tested its performance against existing models with different assumptions by comparing how well they reproduce the patterns observed in foraging White-fronted Geese (Anser albifrons). The assumptions in each of the four compared models were: (1) incompletely informed foraging with benefits of group foraging, which uses the expected gain rates for making decisions on diet choice, patch departure, and flock joining; (2) incompletely informed foraging without benefits of group foraging, which uses the expected gain rates to determine the timing of patch departure but selects a new patch randomly; (3) completely informed foraging without benefits of group foraging, which simply selects the most pro. table patches; and (4) completely informed foraging with benefits of group foraging, which selects the most pro. table patches, considering benefits from the presence of conspecifics. The model that assumed incompletely informed foragers with benefits of group foraging was best in agreement with the observed patterns in all of the five spatial distribution and fat deposition parameters. The models that assumed no benefits of group foraging could not reproduce the observed seasonal variation in flock sizes, whereas the models with completely informed foragers overestimated the flock size as well as usage by the geese of alternative food and the fields near the roost. These results supported the idea that the geese can be assumed to use the expected gain rates for decision-making on diet choice, patch departure, and flock joining. Further, the incompletely informed foragers showed greater disparity in foraging performance among individuals. We discuss the necessity of assuming, when appropriate, that foragers have incomplete information on patch quality in models explaining spatial distribution and foraging success. We also make some references to the applicability of the presented model in other studies.