Aerial alarm calls and fitness in male fowl (Gallus gallus)

Few previous studies have examined fine-grained variation in the structure of alarm calls, as distinct from differences between call types, and most of these have focused on information content (eg level of predation threat). We focused on the design of aerial alarm calls in male fowl, Gallus gallus, with the goal of explaining observed changes in call structure during encounters with simulated predators. These calls typically begin with one or more shot pulsatile elements, which are followed by a sustained element consisting of one or more units. We analysed bouts of aerial alarm calling from 30 individual roosters, measuring duration, dominant frequency and frequency bandwidth (-10dB), for all pulse and sustained elements. Pulse duration was significantly reduced between first, second and third calls in a bout. This change was highly selective. Both mulivariate analyses of other features and spectrogram cross-correlation reveal that pulse reduction was not accompanied by other variation in structure. Recent studies have shown that raptors can detect the aerial alarm calls of G. gallus, but that they cannot easily locate the source of the call. Nevertheless, the pulse element of the aerial alarm call has spectral and temporal characteristics known to be readily localisable by raptors, and hence costly. By reducing the duration of the pulse, the caller likely reduces his conspicuousness and thus manages predation risk. We suggest a possible functional benefit to account for the inclusion of a high-risk element in aerial alarm calls.