Using Hidden Markov Models to characterise intermittent social behaviour in fish shoals

Area of Science:

• Ecology
• Animal Behavior
• Statistical Modeling

Background:

• Group movement is common in nature, crucial for ecology, conservation, and robotics.
• Understanding social interactions is key to explaining collective animal movement.
• Statistical methods for inferring social behavior from movement data are limited.

Purpose of the Study:

• To apply Hidden Markov Models (HMMs) to characterize individual social movement in fish shoals.
• To formally test for and quantify speed-mediated social interactions.
• To compare social behavior dynamics between three-spined stickleback and guppy shoals.

Main Methods:

• Utilized HMMs to analyze trajectory data from three-spined stickleback (Gasterosteus aculeatus) and guppy (Poecilia reticulata) shoals.
• Developed models to infer social interactions based on individual movement patterns.
• Formally tested hypotheses regarding speed-mediated interactions.

Main Results:

• Confirmed the presence of speed-mediated social interactions in both species.
• Social behavior is intermittent and involves varying numbers of individuals.
• Interacting fish speeds exhibit a bimodal distribution (stationary or preferred speed).
• Fish with more social neighbors tend to move at higher speeds.

Conclusions:

• HMMs are effective for characterizing individual social movement and interactions in animal groups.
• Despite species-specific differences in overall movement, the nature of social interactions is qualitatively similar in sticklebacks and guppies.
• Findings provide a framework for understanding and quantifying social behavior in collective animal movement.