Cultural conformity generates extremely stable traditions in bird song

Abstract

Cultural traditions have been observed in a wide variety of animal species. It remains unclear, however, what is required for social learning to give rise to stable traditions: what level of precision and what learning strategies are required. We address these questions by fitting models of cultural evolution to learned bird song. We recorded 615 swamp sparrow (Melospiza georgiana) song repertoires, and compared syllable frequency distributions to the output of individual-based simulations. We find that syllables are learned with an estimated error rate of 1.85% and with a conformist bias in learning. This bias is consistent with a simple mechanism of overproduction and selective attrition. Finally, we estimate that syllable types could frequently persist for more than 500 years. Our results demonstrate conformist bias in natural animal behaviour and show that this, along with moderately precise learning, may support traditions whose stability rivals those of humans.

Fig. 1

Swamp sparrow song structure. a Spectrogram of a swamp sparrow song sung by an individual from the Hudson Valley population. Swamp sparrow songs consist of one syllable (red) repeated 10 or more times. Each syllable consists of 2-5 elements or notes. b–g Examples of different syllable types. b–d Show the syllables that make up the repertoire of the same individual whose song is shown in a. e–g Show three syllable types sung by other males in the population, illustrating both the considerable diversity in syllable structure found within a population, but also how, as a consequence of vocal learning, different individuals also sometimes share the same syllable-type (b, f)

Fig. 2

Simulated and empirical syllable type frequencies. a Frequency distributions of syllable types found in simulations with different types of bias, illustrating how different aspects of individual learning influence population-levels of diversity. A total of 100 simulations were carried out for each of 6 conditions. In the Control case, individuals had a mutation rate of 0.01 and no bias to learning. In the Low-mutation rate condition, μ = 0.001. In the Demonstrator Bias simulations, v = 2. In the Content Bias simulations, p_att = 0.1. In the Conformist Bias simulations, α = 1.2. b Frequency distributions of syllable types found in the empirical data set, from the 6 populations sampled (Conneaut, n = 208, Montezuma, n = 71, Adirondacks, n = 70, Waterloo, n = 74, Horicon, n = 91, Hudson Valley, n = 101), illustrating the consistency found between populations and suggesting a match between the empirical data and the simulations with Conformist Bias

Fig. 3

Geographic variation in syllable sharing. a, b Show the amount of syllable sharing between individuals’ (n = 615) repertoires, measured using the Jaccard Index (proportion of shared syllable types out of all types found in a pair of individuals), against the geographical distance between them. a Syllable sharing was high within populations, with no significant within-population geographic structure, but much lower between populations. b Syllable sharing was similar between the six populations sampled

Fig. 4

Age of syllable types in fitted simulations. The age of syllable types in the 1000 simulations that best fit the empirical data. Each point represents a combination of age in years since the syllable type was innovated and frequency within the population (point size represents the number of times that combination arose). More common syllable types tended to be older, and many syllable types are estimated to be older than 500 years