SHOULD PREY OVERESTIMATE THE RISK OF PREDATION

Mathematical models are used to determine the optimal foraging effort of individuals that face increased risk of predation when they exert g reater foraging effort but have imperfect information about the degree of risk. If the fitness cost of underestimating predation risk is les s than that of overestimating risk, imperfect information should lead to behavior that is appropriate for a lesser risk than is actually pre sent. Overestimation is favored under the opposite condition. If there is a trade-off between starvation and predation, an animal will usual ly underestimate (overestimate) risk if the third derivative of the st arvation-versus-risk relation is positive (negative), provided uncerta inty is not too large. Different, plausible starvation functions can f avor either under- or overestimates of risk. If there is a trade-off b etween reproduction and predation, a more complex condition determines which type of bias is adaptive; this condition involves the reproduct ion-versus-risk function and its first three derivatives, and again, o ver- or underestimation of risk can be advantageous. In almost all mod els, increased accuracy of estimation is favored when costs of increas ed accuracy are sufficiently small. These results differ from those of previous analyses, and reasons for these differences are discussed.