TITLE: “Decoding alarm signal propagation of seed-harvester ants, Pogonomyrmex californicus.”
ABSTRACT: Alarm signal propagation through social-insect colonies provides an empirically tractable context for analyzing information flow through a natural system, with useful insights for network dynamics in other social groups, including human social networks. Here, I develop a methodological approach to track alarm spread within the group of harvester ants, Pogonomyrmex californicus. I alarmed initial 3 individuals and tracked subsequent signal transmission through the group. Because there was no actual threat, the alarm was false, allowing us to assess amplification and adaptive damping of collective alarm response. I trained a Random-Forest machine learning regression model to quantify alarm behavior of individual workers from multiple movement features associated with alarm behavior in this species. This approach provides reliable continuous assessments of an individual’s behavioral state at much finer temporal scales and more consistently than can be achieved visually. I combined the ML alarm state assessments with proximity data from tracking software (ABCTracker) to construct a propagation network of alarm spread. Using this system, alarm propagation can be manipulated and assessed to ask and answer a wide range of questions on information and misinformation flow in social networks.
For lunch at 12:30pm with the speaker, please RSVP at firstname.lastname@example.org