Listening to the birds: how does song culture evolve?

The genesis of a long-term field data set can be prosaic: a researcher measures something, then comes back and repeats that measurement again, and again. It is not always clear what might emerge from the accretion of data over the long term; sometimes the focus evolves as the study progresses.

Kent Island, in the Bay of Fundy, the home of the Bowdoin College Scientific Station, has produced more than its fair share of long-term data sets. Weather data gathered beginning in 1937 was instrumental in finding the source of acid precipitation and its effects on forests1. Individual Leach’s storm petrels (Hydrobates leucorhoa) have been banded and their reproductive success tracked since 19532. Savannah sparrow (Passerculus sandwichensis) data collection began in the 1960s3; birds have been colour-banded since 1987, demographic and reproductive success has been tracked in most years4,5, and songs have been recorded, sporadically at first and then each year since 2003 (except for 2020, because of the pandemic)6.

Undergraduates, graduate students and postdoctoral fellows do much of the day-to-day work of mist-netting adults, measuring and colour-banding them, searching for nests, banding nestlings, and recording songs. Undergraduates, graduate students and postdoctoral fellows do much of the day-to-day work of mist-netting adults, measuring and colour-banding them, searching for nests, banding nestlings, and recording songs. Four labs in Canada (those of Ryan Norris and Amy Newman at the University of Guelph University and Dan Mennill and Stéphanie Doucet at the University of Windsor) and one in the U.S. (Heather Williams at Williams College) currently participate in the project and share long-term data. 

Just as a genetically encoded trait spreads through a population to result in organic evolution, a socially learned behavior can spread through a population to result in cultural evolution. Bird song is a socially learned behavior: young birds copy their songs from adults – sometimes from a parent, sometimes from other birds they hear during development6. This pattern of transmission makes birdsong ideal for studying cultural evolution. In 2004, we started looking at one specific learned song trait: the soft “interstitial” notes sung between the loud introductory notes. “Looking” is the right word; the notes are fast and short, and hard to distinguish by ear; in a sound spectrogram, which plots frequency against time, they can be visually examined and compared to other songs. We found two types of interstitial note groups: “high note clusters”, which usually had three distinct note types including a very high whistle (a horizontal line in the sound spectrogram), and “click trains”, which look like 2-4 vertical strokes in the sound spectrogram. Some birds sang both of these interstitial note types. 

Looking back through the older recordings, we saw that the click trains had not been present in 1980 and 1982, but first appeared in 1988 (songs were not recorded before 1980 or from 1983-7). Once they appeared, the click trains were copied by young birds learning their songs and spread through the population, first within songs that also had high note clusters. The high note clusters were copied less and less often, and finally disappeared altogether from the population.

In a graph, the increase in click trains follows an S-shaped curve, similar both to the spread of a novel gene under positive selection and to the “battleship curve” that anthropologists observe when one cultural form, such as a language variant, replaces another. The click trains also seemed to have a functional correlate: birds singing this form of interstitial note raised more babies to independence. The replacement of high note clusters by click trains is thus a good example of cultural evolution in an animal vocal communication signal7.

But the birds didn’t stop there. Instead they began to elaborate their click trains – young birds added more clicks to the trains they copied. The average number of clicks in a train had been three from 1983 to 2003; then, after a supermajority of birds in the population sang click trains, the maximum and average number of clicks in a train began to increase, with the average approaching 5 in 2013, and the “virtuosos” singing as many as 8 clicks in a row.  

When we tested the effects of these added clicks by playing click trains of different lengths to birds in the field, we found that songs with 7 clicks caused more aggression in males and more approach by females than did songs with fewer clicks. This meant that we had observed the birds making two rounds of changes in the interstitial notes: first, replacing high note clusters with click trains, and later, adding more clicks to those trains. Each of those changes had spread through the population via social learning, and each of them had also made their singers’ vocal communication more effective. These two successive accretions of change in a socially learned behavior, with improved effectiveness after each round, represent cumulative cultural evolution. Naturally occurring cumulative cultural evolution is difficult to document in wild animal populations; so far, only Savannah sparrows (and potentially also humpback whales8) appear to have satisfied the core criteria9 for demonstrating cumulative cultural evolution.

We were also curious about whether the changes we saw were driven by some form of selection or were the result of random processes. We know that Savannah sparrows listen to their social fathers and neighbors in their hatching year, and then settle on their final song when they return to breed early in the following spring10,11. Working with Julie Blackwood, two talented Williams College undergraduates – Andrew Scharf and Anna Ryba - developed a mathematical model of how Savannah sparrows learn their song. Different “learning curves” corresponded to a) random copying (cultural drift), b) preferentially learning either a rare or a common song type (frequency-dependent bias), or c) a preference for learning one particular song type regardless of other factors (selection). The historical data best match the selection model: birds preferred to learn click trains instead of high note clusters.

Looking at the survival of males singing high note clusters, click trains, and different numbers of clicks, we saw that what a male sang was not correlated to his survival and hence to his genetic fitness – even though his ability to raise offspring and attract females increased when a male sang click trains, and later, longer click trains. So it was the song, independent of the singer’s quality, that was more attractive to females and was more likely to be copied. Copying attractive song features made a male more attractive, independent of his intrinsic genetic “quality”, and it may be this shortcut to reproductive success that drives rapid cumulative cultural evolution in these birds.


  • Jagels, R., Carlisle, J., Cunningham, R., Serreze, S. & Tsai, P. Impact of acid fog and ozone on coastal red spruce. Water Air Soil Pollut. 48, 193–208 (1989).
  • Mauck, R. A., Huntington, C. E. & Grubb, T. C. J. Age-specific reproductive success: evidence for the selection hypothesis. Evolution (N. Y). 58, 880–885 (2007).
  • Dixon, C. L. Breeding biology of the Savannah sparrow on Kent Island. Auk 95, 235–246 (1978).
  • Wheelwright, N. T. & Mauck, R. A. Philopatry, natal dispersal, and Inbreeding avoidance in an island population of Savannah sparrows. Ecology 79, 755–767 (1998).
  • Woodworth, B. K., Wheelwright, N. T., Newman, A. E. M. & Norris, D. R. Local density regulates migratory songbird reproductive success through effects on double-brooding and nest predation. Ecology 98, 2039–2048 (2017).
  • Wheelwright, N. T. et al. The influence of different tutor types on song learning in a natural bird population. Anim. Behav. 75, 1479–1493 (2008).
  • Williams, H., Levin, I. I., Norris, D. R., Newman, A. E. M. & Wheelwright, N. T. Three decades of cultural evolution in Savannah sparrow songs. Anim. Behav. 85, 213–223 (2013).
  • Garland, E. C., Garrigue, C. & Noad, M. J. When does cultural evolution become cumulative culture? A case study of humpback whale song. Phil. Trans. R. Soc. B. 20200313 (2021). doi:10.1098/rstb.2020.0313
  • Mesoudi, A. & Thornton, A. What is cumulative cultural evolution? Proc. R. Soc. B-Biological Sci. 285, 20180712 (2018).
  • Mennill, D. J. et al. Wild birds learn songs from experimental vocal tutors. Curr. Biol. 28, 3273-3278.e4 (2018).
  • Thomas, I. P. et al. Vocal learning in Savannah sparrows: acoustic similarity to neighbours shapes song development and territorial aggression. Anim. Behav. 176, 77–86 (2021).