In many songbirds, individuals have repertoires of multiple song types, some of which may be shared with others in the local area. Hypotheses about the evolution of song repertoires differ as to whether selection acts primarily on repertoire size itself or the ability to match songs of neighbours. We used a 16-channel acoustic location system to record neighbourhoods of song sparrows (Melospiza melodia melodia) during the dawn chorus. We asked whether males sing all songs with similar frequency as predicted by the Repertoire Size Hypothesis, whether males preferentially sing highly shared songs as predicted by the General Sharing Hypothesis, or whether use of highly shared songs is associated with phenotype as predicted by the Conditional Sharing Hypothesis. Contrary to the Repertoire Size Hypothesis, most males did not sing all songs equally often. Contrary to the General Sharing Hypothesis, we found no general tendency to overproduce highly shared songs. The degree to which males overproduced highly shared songs was repeatable across days, indicating consistent individual differences, and varied across neighbourhoods. Moreover, and consistent with the Conditional Sharing Hypothesis, older males were more likely to overproduce highly shared songs. If highly shared song is a conventional signal of aggression, with the threat of receiver retaliation maintaining honesty, older males may be more willing or able to risk conflict. Alternatively, males may learn which songs are effective signals for an area. Finally, age-related variation in vocal performance may shape the adaptive value of highly shared song.
Study site and population
Field work was conducted during spring 2008 and 2009, on the Bracken property owned by the Queen’s University Biological Station, near Newboro, Ontario, Canada (44° 38.6′ N, 76° 19.0′ W). This site supports approximately 35–40 breeding pairs of song sparrows (M. melodia melodia) that have been studied by our research group since 2002. Song sparrows in this area are migratory, but philopatric as adults (40–60% return rates, <75 m territory movement between years).
Shortly after birds returned from spring migration in April, we captured adult males in seed-baited Potter traps or mist nets and provided each with a unique combination of coloured leg bands. We measured mass to the nearest 0.2 g using a spring-loaded scale, and measured right tarsus length to the nearest 0.1 mm using dial callipers. Body condition was estimated as the ratio of mass to tarsus length. We determined age (range = 1–5 years) based on banding records dating back to 2002. Two males had been banded as nestlings in previous years, so their ages were known with certainty. Others were first captured and banded as adults; we considered these adult recruits to be yearlings. In support of this assumption, adult recruits have shorter wings than birds known to be 2 years or older (unpublished data). Moreover, each spring we exhaustively search the study area to capture and band all breeding adults. Given the short distance of territory movement by banded birds between years, we assume that adult recruits are breeding for the first time. We determined pairing status through field observations, and monitored nests to confirm the breeding status of all males in the study.
ALS recording and analysis
The ALS consisted of an array of 16 omni-directional microphones, similar to the eight-channel ALS described by Mennill et al. (2006), distributed over five to 11 song sparrow territories. Microphones were protected by rain guards made from polyvinyl chloride tubing and plastic mesh, and mounted with shelf brackets onto 3-m wooden poles that had been painted in a camouflage pattern and attached to trees or stakes using elastic cords. Microphones were connected by 2,200 m of cable to a central laptop computer, where data from all 16 simultaneously recording channels were digitized using a multichannel data acquisition card (National Instruments DAQ-6260) and stored as a 16-channel AIFF sound file using Chickadee recording software (version 1.8; J. Burt, Seattle, WA). ALS recording began at sunrise; we annotated the first 30 min of song recorded each day. We have previously shown that 30 min is sufficient to achieve repeatable measures of song output in this population (MacDougall-Shackleton et al. 2009). This interval encompassed most of the dawn chorus, as song rates declined substantially by 30 min after sunrise, and was long enough for most singers to use most or all of the songs within their repertoire (Table 1). To minimize disturbance, we did not intrude onto territories while the ALS was operating.