Prey responses to direct and indirect predation risk cues reveal the importance of multiple information sources.

Prey can use several information sources (cues) to assess predation risk and avoid predation with a variety of behavioural responses (e.g., changes in activity, foraging, vigilance, social behaviour, space use, and reproductive behaviour). Direct cues produced by predators and indirect cues from environmental features or conspecific and heterospecific prey generally provide different types of information about predation risk. Despite widespread interest in understanding behavioural antipredator responses to direct and indirect cues, a clear general pattern of relative response strength across taxa and environments has yet to emerge. We conducted a meta-analysis of studies (N = 113 articles and 999 effect sizes taken from a search of over 7500 articles) testing behavioural responses to direct and indirect cues of predation risk, and their combination, across terrestrial and aquatic ecosystems. We further contrasted if effects were moderated by ecosystem type (terrestrial, marine, or freshwater), cue source (predator, conspecific, heterospecific, or environmental feature), or sensory modality (visual, auditory, or chemosensory). Overall, there were strong effects of risk cues on prey behaviour. We found that prey responded more strongly when both types of cues were presented together compared with either cue in isolation, which was driven by changes in prey activity levels but not other behaviours. There was no general pattern in response strength to direct compared with indirect cues. Responses to these cues were moderated by interactions between environment, cue source, and cue sensory modality (e.g., visual cues elicited stronger responses than other modalities, and responses to conspecific chemosensory cues were stronger than those to predator chemosensory cues in aquatic systems). These results suggest that rather than a broad framework of direct and indirect cues, the specific context of the system should be considered in tests and predictions of how prey respond to risk to elucidate general patterns of antipredator responses.