Deconstructing the landscape of fear in stable multi-species societies.

Abstract

Animal distributions are influenced by variation in predation risk in space, which has been described as the "landscape of fear." Many studies suggest animals also reduce predation risk by eavesdropping on heterospecific alarm calls, allowing them to occupy otherwise risky habitats. One unexplored area of study is understanding how different species' alarms vary in quality, and how this variation is distributed in the landscape. We tested this phenomenon in a unique system of avian mixed species flocks in Amazonian rainforests: flock mates (eavesdropping species) strongly associate with alarm-calling antshrikes (genus Thamnomanes), which act as sentinel species. Up to 70 species join these flocks, presumably following antshrike behavioral cues. Since flocks in this region of the Amazon are exclusively led by a single antshrike species, this provides a unique natural system to compare differences in sentinel quality between flocks. We simulated predation threat by flying three species of live trained raptors (predators) towards flocks to compare sentinel probability to (1) produce alarm calls, and (2) encode information about magnitude and type of threat within such alarm calls. Our field experiments show significant differences in the probability of different sentinel species to produce alarm calls and distinguish predators. This variation may have important fitness consequences and shape the "landscape of fear" for eavesdropping species.