Reliable aggressive signalling in swamp sparrows

Abstract

Whether aggressive displays are reliable predictors of attack is an important, unresolved issue in animal communication research. Here we test the extent to which vocal and visual displays predict subsequent attack in territorial male swamp sparrows, Melospiza georgiana. A brief playback of swamp sparrow song was used to provoke aggressive signalling from a territorial male, and the subject's displays were recorded for 5 min. A taxidermic mount of a male swamp sparrow was then revealed, coupled with additional playback, and the subject was given 14 min to attack while we continued to record its displays. Of 40 subjects, nine attacked the mount and 31 did not. For both the initial recording period and the 1 min before attack, attackers produced significantly more low-amplitude ‘soft songs’ and more bouts of wing waving than did nonattackers. Attackers and nonattackers did not differ significantly in song type-switching frequency or in numbers of broadcast songs, matching songs, wheezes or rasps. In discriminant function analyses, soft song was consistently the display that best predicted subsequent attack. Different acoustic forms of soft song were found, all of which appeared to be equally aggressive. Combinations of displays predicted attack better than did single displays. The overall reliability of swamp sparrow displays as predictors of aggression was impressively high.