Sexual selection for both diversity and repetition in birdsong

Abstract

From fiddler crabs to humans, animals perform repetitive displays showing neuromotor skill and vigour. Consistent repetition of identical notes (vocal consistency) facilitates the assessment of neuromotor skills and is important in communication in birds. Most birdsong research has focused on song diversity as a signal of individual quality, which seems contradictory as repetition is extremely common in most species. Here we show that consistent repetition within songs is positively correlated with reproductive success in male blue tits (Cyanistes caeruleus). A playback experiment shows that females are sexually aroused by male songs with high levels of vocal consistency, which also peaks seasonally during the fertile period of the female, supporting the role of vocal consistency in mate choice. Male vocal consistency also increases with subsequent repetitions of the same song type (a warm-up effect) which conflicts with the fact that females habituate to repeated song, showing decreased arousal. Importantly, we find that switching song types elicits significant dishabituation within the playback, supporting the habituation hypothesis as an evolutionary mechanism driving song diversity in birds. An optimal balance between repetition and diversity may explain the singing style of many bird species and displays of other animals.

Fig. 1. Spectrograms of typical songs of blue tits.

Each of the three pairs of spectrograms (a–c) shows two songs recorded from the same individual male, one before the breeding period (left) and the second during the females’ fertile period (right), with the frequency (kHz) in the Y-axis and time in seconds in the X-axis. Spectrograms (a, b) show annotations describing the basic structure of song in this species. The vocal consistency measured for each song is shown in the upper right corner, with 1 as the maximum consistency possible. The time when it was recorded during the season is also indicated as weeks in relation to the first egg date.

Fig. 2. Correlation of vocal consistency with reproductive success, season and context in blue tits.

Points in (a) show mean vocal consistency of each male per breeding season with the respective clutch size as number of eggs, corrected for the seasonal effect of date of first egg using the estimated coefficient from the model. Points in (b) show the male vocal consistency ± one standard error (SE) per clutch size category (number of males in each category shown underneath error bars). c The variation of male vocal consistency throughout the breeding period, with a temporal resolution of 1 day. Points indicate the mean ± SE of vocal consistency of all males recorded on each day, the sample size of number of males recorded each day shown underneath error bars. In all plots, red lines trace the model predicted values, with the associated 95% confidence interval (CI) in pink shaded area. The grey area in the centre of (c) indicates the female receptive period. In (d), we see that male vocal consistency is significantly higher during dawn chorus (N = 90) than during day-time singing (N = 74), measured in a total of 95 birds recorded during the breeding period. Box and whisker plots showing median, upper and lower quartiles, and 1.5 interquartile range and outliers as points. Source data are provided in Supplementary Data 2.

Fig. 3. Increase in vocal consistency within song in the first 15 repetitions of the whole song in a song-type bout.

Vocal consistency is standardized by subtracting the consistency measured in the first song per song-type bout (consistency of 1st repetition = 0). Points indicate mean values per song position across all song bouts for 18 individuals, with the associated SE. The red line indicates the predicted values of the model, with the associated 95% confidence interval in the shaded area. Source data are provided in Supplementary Data 2.

Fig. 4. Female vocal response during the playback experiment in the nest box.

Female response is shown as the proportion of song bouts where females produced at least one copulation solicitation call, in relation to the total number of bouts played while the female was inside the nest box. For the same 13 female subjects and for 442 bouts, a shows that female vocal response was significantly higher during the Song treatment than during the Silence treatment. Box and whisker plots showing median, upper and lower quartiles and 1.5 interquartile range. Points overlaid represent individual females, as the proportion of bouts with vocal response out of all bouts presented. b The significant, positive correlation between female vocal response and vocal consistency of playback song. In (b), points represent the proportion of bouts with vocal response out of all song bouts played for each song type during each trial. Black lines connect the response of the same female to both song types presented within the same trial. The red line represents the predicted values derived from the model, with the 95% CI interval in the pink shaded area. Source data are provided in Supplementary Data 2.

Fig. 5. Female vocal response per song position within playback song-type bouts.

The X-axis shows the sequential position of each song within a song-type bout. In this case, the first 15 repetitions are shown, matching the analysis of change in vocal consistency within song-type bout in male dawn song. The Y-axis shows the proportion of playback songs with at least one female vocalisation, out of all songs played per trial in that position. Each point represents the mean ± SE female vocal response per song position across all trials, summarizing the response of 13 females to 34 song-type bouts preceded by silence (in red) and 60 song-type bouts preceded by a song type switch (in blue). The lines show the predicted values for female vocal response derived from the model, with the associated 95% confidence intervals in shaded area. The decrease in response as the song is repeated indicates behavioural habituation of the females while the switch in song type elicited dishabituation. The data included song type switches between ten different song types across trials and in all possible directions, i.e., from song type A to song type B but also from song type B to song type A. Source data are provided in Supplementary Data 2.

Fig. 6. Diagram of methods used in the female choice playback experiment.

In (a), a schematic overview of the equipment set up at the nest box, see also Fig. S5. In (b), a schematic timeline indicating the structure of a playback stimulus, alternating periods of song with silence. Spectrogram (c) shows an audio snapshot recorded inside the nest box during a playback trial, showing the vocal interaction of a subject female with the playback song, with frequency (kHz) in the Y-axis and time in seconds in the X-axis. Spectrograms (d) and (e) show the repetition at different time scales in a natural recording of a male singing. In (d) one note is repeated within song (this is where we measure vocal consistency), and in (e) the entire song is repeated within a song-type bout.