Familiarity, homogeneity, and discrimination of song dialects: Data and playback study stimuli

Male songbirds of many species sing local song dialects that are restricted to defined geographical areas. In most tests of responses to local versus foreign dialects, males respond more aggressively to songs from their own dialect, presumably because local males represent more of a threat to their success. We asked how hearing foreign songs during development and territory establishment affects discrimination of the local dialect in wild Savannah sparrows, Passerculus sandwichensis. After foreign songs had been heard from loudspeakers in the study area in at least two consecutive breeding seasons, males reduced the intensity of their responses to the local version of population-specific buzz segment of the song. Four years after the foreign songs were last broadcast on the study area, males again responded more aggressively to the local version of the buzz. As for the basis of these responses, we found no evidence that birds discriminated among dialects by comparing them to their own songs. However, auditory experience with a foreign song, whether during song development (from speaker-simulated song tutors) or during the current breeding season (from neighbours’ songs), reduced the intensity of birds’ responses to the local buzz type. Both familiarity, in the form of auditory experience with a song type, and homogeneity, when a song type is sung by all or nearly all of the population, appear to contribute to heightened aggressive responses to a local song dialect.

Methods
The songs of Savannah sparrows (Passerculus sandwichensis) in the Grand Manan archipelago (New Brunswick, Canada) were recorded using a Marantz digital recorder (PMD670 or PMD660) and a Sennheiser ME-66 directional microphone; our digital recordings used a 44 kHz rate and a 16-bit depth. We also used some songs recorded between 1993 and 2011 to characterize Kent Island variant buzzes and generate stimuli for buzz playbacks (see Wheelwright et al., 2008 for details on methods for those recordings). We digitized songs originally recorded on tape using SoundEdit Pro (Macromedia; 44 kHz and 16 bits). 

The buzz segments of all songs and stimuli used in this study were characterized by measuring their mean frequency (using SoundAnalysisPro; Tchernichovski et al., 1999; http://soundanalysispro.com) and pulse duration (buzz duration divided by the number of repeated sound pulses). These two parameters are sufficient for distinguishing the different dialects of Savannah sparrow buzzes (see Williams et al, 2019).

Stimuli used in playback studies were buzz segments drawn from four different Savannah sparrow song types: (1) the local Kent Island dialect, (2) a variant that was no longer sung on Kent Island, (3) a dialect from Williamstown, MA, USA, 500 km distant, and (4) songs from western North America that were broadcast through speakers on Kent Island from 2013-2018 and were learned by some local birds (see Mennill et al., 2018). Each set of four stimuli (one of each type) was matched for duration and amplitude.

Subjects of playback studies included (1) birds breeding on Kent Island in 2012; (2) birds breeding on Kent Island in 2016 and 2017 (3) birds breeding on nearby islands in 2016, and (4) birds breeding on Kent Island in 2022. The second category, birds breeding on Kent Island in 2016 and 2017, heard foreign songs broadcast from speakers on their breeding site, and some of them had copied those songs. All subjects were mated males.

A playback session consisted of a primer stimulus, which was a song in which the buzz had been replaced with silence, which was used to try to minimize first-stimulus effects, followed by the four stimulus types deiivered in a varied, balanced order. Each stimulus was played for two minutes (5 buzzes per minute). Response duration was measured as the time between the end of the stimulus playback and the time when the subject ceased to respond aggressively (either  and the time when the male ceased to respond was noted. Responses to the playback were scored as ending when a bird either sang a full, loud song; began foraging; started preening; or flew more than 30 meters away from the speaker. We used this measure because individual birds used different aggressive behaviors when responding to a playback (see Williams et al., 2019).

These responses were evaluated by GLMM models that included stimulus order, stimulus type, subject identity, acoustic similarity of the stimulus to the subject's own buzz, and the subjects' experience with foreign songs (during development and during the season of the playback study).

Data files include the acoustic measurements of different buzz types; responses to playback stimuli; and an R script recapitulating the analyses reported in the paper. Also included are four sound files that include eight versions of each of the different buzz types used as playback stimuli.