skip to main content

Title: Machine learning and statistical classification of birdsong link vocal acoustic features with phylogeny

Birdsong is a longstanding model system for studying evolution and biodiversity. Here, we collected and analyzed high quality song recordings from seven species in the familyEstrildidae. We measured the acoustic features of syllables and then used dimensionality reduction and machine learning classifiers to identify features that accurately assigned syllables to species. Species differences were captured by the first 3 principal components, corresponding to basic frequency, power distribution, and spectrotemporal features. We then identified the measured features underlying classification accuracy. We found that fundamental frequency, mean frequency, spectral flatness, and syllable duration were the most informative features for species identification. Next, we tested whether specific acoustic features of species’ songs predicted phylogenetic distance. We found significant phylogenetic signal in syllable frequency features, but not in power distribution or spectrotemporal features. Results suggest that frequency features are more constrained by species’ genetics than are other features, and are the best signal features for identifying species from song recordings. The absence of phylogenetic signal in power distribution and spectrotemporal features suggests that these song features are labile, reflecting learning processes and individual recognition.

more » « less
Author(s) / Creator(s):
; ; ;
Publisher / Repository:
Nature Publishing Group
Date Published:
Journal Name:
Scientific Reports
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
    more » « less
  2. 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.

    more » « less
  3. Abstract

    Previous work has demonstrated that there is extensive variation in the songs of White-crowned Sparrow (Zonotrichia leucophrys) throughout the species range, including between neighboring (and genetically distinct) subspecies Z. l. nuttalli and Z. l. pugetensis. Using a machine learning approach to bioacoustic analysis, we demonstrate that variation in song is correlated with year of recording (representing cultural drift), geographic distance, and climatic differences, but the response is subspecies- and season-specific. Automated machine learning methods of bird song annotation can process large datasets more efficiently, allowing us to examine 1,913 recordings across ~60 years. We utilize a recently published artificial neural network to automatically annotate White-crowned Sparrow vocalizations. By analyzing differences in syllable usage and composition, we recapitulate the known pattern where Z. l. nuttalli and Z. l. pugetensis have significantly different songs. Our results are consistent with the interpretation that these differences are caused by the changes in characteristics of syllables in the White-crowned Sparrow repertoire. This supports the hypothesis that the evolution of vocalization behavior is affected by the environment, in addition to population structure.

    more » « less
  4. Candolin, Ulrika (Ed.)
    Abstract Learned traits, such as foraging strategies and communication signals, can change over time via cultural evolution. Using historical recordings, we investigate the cultural evolution of birdsong over nearly a 50-year period. Specifically, we examine the parts of white-crowned sparrow (Zonotrichia leucophrys nuttalli) songs used for mate attraction and territorial defense. We compared historical (early 1970s) recordings with contemporary (mid-2010s) recordings from populations within and near San Francisco, CA and assessed the vocal performance of these songs. Because birds exposed to anthropogenic noise tend to sing at higher minimum frequencies with narrower frequency bandwidths, potentially reducing one measure of song performance, we hypothesized that other song features, such as syllable complexity, might be exaggerated, as an alternative means to display performance capabilities. We found that vocal performance increased between historical and contemporary songs, with a larger effect size for urban songs, and that syllable complexity, measured as the number of frequency modulations per syllable, was historically low for urban males but increased significantly in urban songs. We interpret these results as evidence for males increasing song complexity and trilled performance over time in urban habitats, despite performance constraints from urban noise, and suggest a new line of inquiry into how environments alter vocal performance over time. 
    more » « less
  5. Abstract

    Audio recording devices have changed significantly over the last 50 years, making large datasets of recordings of natural sounds, such as birdsong, easier to obtain. This increase in digital recordings necessitates an increase in high‐throughput methods of analysis for researchers. Specifically, there is a need in the community for open‐source methods that are tailored to recordings of varying qualities and from multiple species collected in nature.

    We developed Chipper, a Python‐based software to semi‐automate both the segmentation of acoustic signals and the subsequent analysis of their frequencies and durations. For avian recordings, we provide widgets to best determine appropriate thresholds for noise and syllable similarity, which aid in calculating note measurements and determining song syntax. In addition, we generated a set of synthetic songs with various levels of background noise to test Chipper's accuracy, repeatability and reproducibility.

    Chipper provides an effective way to quickly generate quantitative, reproducible measures of birdsong. The cross‐platform graphical user interface allows the user to adjust parameters and visualize the resulting spectrogram and signal segmentation, providing a simplified method for analysing field recordings.

    Chipper streamlines the processing of audio recordings with multiple user‐friendly tools and is optimized for multiple species and varying recording qualities. Ultimately, Chipper supports the use of citizen‐science data and increases the feasibility of large‐scale multi‐species birdsong studies.

    more » « less