Event-level mobbing responses of black-tailed gulls (Larus crassirostris) to a resident and a transient predator, Hongdo Island, Korea, 2003

Collective mobbing of predators is a public good: it lowers risk for the whole colony but is costly to each participant, favouring conditional rather than unconditional investment. Allocating that investment to where risk is highest should require prey to anticipate the predator, and hence to have information available from repeated encounters. Using a natural contrast at a Black-tailed gull (Larus crassirostris) colony between a resident predator (Peregrine falcon, Falco peregrinus) and a transient one (Crested honey buzzard, Pernis ptilorhynchus), operationally familiar/predictable versus unfamiliar/unpredictable, we asked whether directional selectivity differed between them. For each attack we recorded mobbing duration and participation in the two colony sectors relative to the predator's approach direction. Against the falcon, the approached sector mobbed about 3.5 times longer than the opposite sector, whereas against the buzzard both sectors mobbed intensely irrespective of direction. The predator × direction interaction was strong and robust to predator exposure time. Gulls thus targeted defence spatially against the resident but not the transient predator, consistent with an information-dependent predictability hypothesis. Because familiarity co-varies with predator identity, threat type and season, this natural contrast identifies a strong association rather than a causal mechanism; we outline the model-predator experiment needed to test it.