Song as an aggressive signal in songbirds

Birdsong is often regarded as an aggressive signal. More specifically, particular singing behaviours are hypothesized to be threatening, including song type matching, frequency matching, song overlapping, song type switching and low-amplitude song. The term aggressive signal should be reserved for behaviours that are associated with, and, in that sense, signal aggressive escalation. Three criteria are relevant to whether a signal should be classified as aggressive: (1) whether the signal increases in aggressive contexts (the context criterion); (2) whether the signal predicts aggressive escalation by the signaller (the predictive criterion); and (3) whether receivers respond to the signal (the response criterion). Adequate evidence on all three criteria is not available for most putative aggressive signals. The evidence that is available suggests that low-amplitude song and perhaps frequency matching are strongly aggressive signals in some species; that type matching is not a signal of attack but may be a signal of lower levels of aggressive escalation; that type switching can be either a signal of escalation or of de-escalation; and that overlapping may not be a signal at all. Song performance is probably mainly a signal of male quality, but may additionally signal aggressive escalation in some cases. We propose that more attention be devoted to testing the predictive criterion: an aggressive signal should either predict attack, or if not, it should predict escalation to the next higher level of aggressive signalling, and each higher level of signalling should more reliably predict the ultimate response of actual attack.