AbstractHoney bees (Apis mellifera L.) are social insects that makes frequent use of volatile pheromone signals to collectively navigate unpredictable and unknown environments. Ants have been shown to effectively use pheromone trails to find the shortest path between two points, the nest and the food source. The ant pheromone trails are accomplished by depositing of pheromones which are then diffused passively, creating isotropic (i.e., non-directional and axi-symmetric) signals. In this study, we report the first instance of the honey bees’ ability to solve the shortest path problem to localize the queen and aggregate around her by using a collective flow-mediated scenting strategy. In this strategy, individual bees not only emit pheromones, but also fan their wings to actively direct the flow of the signals, providing colony members with directional messages to the queen’s location. We use computer vision and deep learning approaches to perform automatic and accurate image analysis. As a result, we quantify the number of bees in the short and long paths, and show that the short path is frequented by significantly more bees over time. We also reconstruct attractive surfaces using the positions and directions of scenting bees, and show that this surface is more “attractive” along the short path and around the queen as scenting bees send out directional messages and the swarm makes their way to the queen. Overall, we show that the honey bees can effectively use the collective scenting behavior to overcome local and volatile pheromone communication and find the shortest path to the queen.
