Abstract

MS. number: A13-00177 ay b h a phy arro ck. song, but only a weak predictor of attack. We used a two-part playback design to test these predictions, with an initial edge playback from just off the subject’s territory using a song type that the subject matching. Matching the edge playback also was not associated with physical measures of aggression ggress Waas anation at disp essive progression of discretely different signals. Here we test a specific model of hierarchical signalling proposed for song sparrows, Melospizamelodia, by Beecher and colleagues (Beecher & Campbell 2005; Searcy & Beecher 2009; Akçay et al. 2013). s (Clutton-Brock & ed for fallow deer, d here it has been ression, with one other (Bartoset al. e blue penguins, c displays given by these birds (Waas 1990,1991b), two vocalizations, growls andhisses, appear to form a hierarchy of threat. When confronted with a model penguin at their burrows, lonemales that give growls aremore likely to attack thanmales that remain silent, andmales that give hisses are more likely to attack than those that growl (Waas 1991a). Thus hisses are a more reliable threat than growls. Themodel of hierarchical aggressive signalling in song sparrows emerged from work by Beecher and colleagues on matching be-haviours. In song type matching, one individual replies to another