Song overlapping, a behaviour in which an individual begins singing before its counterpart has completed its song, has been the subject of recent debate. Although many studies have suggested that song overlapping functions as a signal, the majority of these studies fail to address the possibility that overlapping is a chance occurrence. Part of the difficulty in determining whether overlap is intentional or accidental lies in the lack of compelling null models for estimating chance levels of song overlap. We have developed the Song Overlap Null model Generator (SONG), a software package for R. SONG uses resampling randomization to predict the expected amount of overlap due to chance, and is applicable to any system in which individuals engage in signalling interactions. To evaluate the effectiveness of SONG, we examined the overlapping behaviour of three avian species: black-capped chickadees (Poecile atricapillus), rufous-and-white wrens (Thryophilus rufalbus) and long-tailed manakins (Chiroxiphia linearis). Our analyses revealed that black-capped chickadees avoided overlapping the songs of playback-simulated intruders, duetting wrens overlapped the songs of their mates and manakins avoided overlapping the duets of their neighbours. We believe that SONG will prove to be a valuable tool for understanding signal timing in songbirds as well as other taxa.
Animal communication, birdsong, signal timing, song overlapping, vocal interaction