Cloutier, Z.; Manser, M. B.; Thorley, J.; Behr, D. M.; Clutton-Brock, T. H.; Ozgul, A.

Prey can adjust their behaviour to perceived predation risk, which may induce non-consumptive fitness effects. However, quantifying perceived risk in field studies remains challenging as anti-predator behaviours often integrate multiple processes such as encounter probability and detection capacity. Partitioning prey responses into sub-processes may therefore provide novel insights on predation risk perception. To this end, we decomposed meerkat alarm-calling behaviour into threat discrimination, and response escalation. Using 23 years of data from a wild population in the Kalahari, we evaluated the context-dependence of both components in relation to landscape features, and group composition. We defined threat discrimination as the probability to emit an alarm call at predators (eagles) relative to non-predators (vultures) and response escalation as the strength and intensity of the ensuing collective reaction. We found that discrimination varied with habitat and vegetation productivity. Furthermore, larger groups and the presence of pups increased meerkat threat discrimination, but only during the dry season. Response escalation varied with habitat and group size, but pup presence only increased reaction strength. The contrasting context-dependence among the two alarm-calling components support decomposing compound behaviours rather than quantifying perceived risk from a single behaviour. Thus, we recommend future studies to use this approach, when possible, to improve inference on prey perception and responses to risk.