Model-Based Analysis of Honeybee Communication during Aggregation and Food Distribution

Honeybees exhibit remarkable distributed behaviors due to their unique structure, one being how they distribute food in a hive. A few forager bees act as mobile food stores from which others obtain food through trophallaxis, the direct exchange of food. Newly fed bees then continue the food-exchange chain until all bees are sufficiently fed. However, how fed bees are located by the rest of the group, and the communication mechanisms between them remains largely unexplored. We leverage computer-vision and machine learning to detect food exchange and scenting behaviors (vigorous wing-flapping to propagate pheromones directionally) in video data from laboratory experiments. We demonstrate that honeybees communicate through scenting, aggregate around mobile food sources, and efficiently distribute food. Finally, we incorporate scenting into an agent-based model to study its impact on food distribution times. We find that directional scenting increases the efficiency of food distribution and makes model dynamics more consistent with experimental observations, demonstrating that scenting is an effective communication mechanism for communicating the location of moving food sources in honeybee groups.