Bird song as a signal of aggressive intent

A central question in animal communication research concerns the reliability of animal signals. The question is particularly relevant to aggressive communication, where there often may be advantages to signaling an exaggerated likelihood of attack. We tested whether aggressive signals are indeed reliable signals of attack in song sparrows (Melospiza melodia). We elicited aggressive signaling using a 1-min playback on a male's territory, recorded the behavior of the male for 5 min, and then gave him the opportunity to attack a taxidermic mount of a song sparrow associated with further playback. Twenty subjects attacked the mount and 75 did not. Distance to the speaker was a significant predictor of attack for both the initial recording period and the I min before attack. For the initial recording period, none of the measures of singing behavior that we made was a significant predictor of attack, including song-type matching, type-switching frequency, and song rate. For the 1-min period immediately before attack, only the number of low amplitude soft songs was a significant predictor of attack. Although most aggressive signals contained little information on attack likelihood, as some models suggest should be the case, the unreliability of these signals was not caused by convergence of individuals on a single signaling strategy, as those models argue should occur.