A collaborative research project funded by
How do animals coordinate with one another across varying contexts and spatial scales?
We are an interdisciplinary and international team of researchers investigating how communication and coordination interact in animal societies.
How do vocal communication and social relationships shape collective behaviors?
We combine traditional behavioral field biology with modern tracking technologies and computational approaches to address these questions
We focus on three social carnivores spanning a range of group cohesion levels, from cohesive meerkat groups to fission-fusion hyena clans
We use tracking collars to simultaneously record the movements, behaviors, and vocalizations of entire animal groups
We develop machine learning approaches to detect and classify calls, and to identify who is calling
From raw audio and accelerometer data, we use machine learning to extract and identify calls and behaviors of interest.
Above is an illustration of our pipeline showing data collection (A), processing (B), and sensor integration (C), using data from a hyena pilot study. (B) Example time series of sound intensity, time x frequency spectrogram, and accelerometer (ACC) signal from one hyena. ACC can help separate calls of a tagged focal hyena (dashed black boxes) from calls of other hyenas (dashed red boxes), even if they overlap. (C) 15-minute group foraging example involving 4 hyenas. Top panel shows ODBA (energy expenditure proxy) of two individuals, with calls overlaid (circles). Bottom panels show 1 Hz GPS positions of four hyenas, with focal calls overlaid (circles).
We are using these data to understand
--> drivers of individual decision-making
--> patterns of information flow within groups
--> collective outcomes such as group movement and collective action
Movement and signaling data from a single meerkat group over 20 min, collected during a pilot study. Lines show trajectories of each meerkat (individuals denoted by different shades of gray). Each column represents a single meerkat. Colored points on lines indicate positions, with color showing time elapsed. Inset shows the time series of vocalizations from all individuals corresponding to the same 20 minutes. Horizontal colored lines indicate calls of different types, with examples shown as spectrograms in insets (red: cohesion call, blue: move call, green: short note call).