Neural mechanisms of dance communication in honeybees

Abstract

This thesis brings a new approach to the study of honeybee dance communication: evaluating computational models against observed behaviour to understand how dance communication could be implemented by the insect brain. After a foraging trip, a bee has internal knowledge of the flight vector to the food which it communicates by producing a stereotyped motor pattern known as the waggle dance. This creates mechanical cues that surrounding nestmates can assimilate to obtain their own flight vector. Independently of communication, the acquisition and utilisation of vectors for navigation is believed to occur in the central complex in the insect brain. Our key hypothesis is that this circuit is sufficient, with minor adaptations, to explain both the production and assimilation of dance information.

First, we propose how an anatomically grounded model of path integration for foraging, based on the central complex, could be adapted to produce the dance in the hive. We assume the existence of vector memory, where a snapshot of the bee's path integrator state at the food source can later be utilised by the steering circuitry to guide a return journey. We also impose a parameter to limit the angular velocity of the bee when waggling based on properties of natural dances. By simulating neural activity for foraging routes, we demonstrate that natural features emerge from the subsequent dancing behaviour produced by the circuit, including waggle and return phases. The simulated dances also exhibit patterns of angular scatter that align with those observed in real bees for different feeder distances. Our results suggest that performance of the dance could arise from a pre-existing neural circuit that underlies large-scale navigation and supports the idea that the dance is a miniature re-enactment of the foraging flight.

We then explore how this circuit could be used by the receivers of information: nestmates following the dance. We present a new dataset of nestmates as they follow a dance and uncover a previously unreported feature of their antennal positioning that correlates to their relative angle to the dancer. Knowing its own orientation to gravity on the vertical comb, this could allow followers to deduce the dancer's orientation, which indicates the direction of the food relative to the sun. Integrating the estimates of this direction during the waggle phase could then enable the follower to obtain a vector to the food. Based on recent evidence of antennal inputs and spatial encoding in the central complex of the fruit fly, we propose how the circuit could be adapted to use the antennal information to perform these computations and recover the signalled dance vector. Using the real positional data of the followers as input, the simulated circuit predicts that their recovered vectors would be appropriately centred on the signalled direction. It also predicts the distribution to be up to ±90° wide from this direction.

To follow up this result, we devise an experiment to compare these predictions with the real flight vectors expressed by bees. Inspired by the forced-detour paradigm in ants, we track the correction angle of bees navigating to a feeder after an imposed detour, as a measure of the estimated food location. Similar to the circuit's predictions, we observe a characteristic spread of vectors centred on the feeder. Our work thus indicates that the central complex could underlie the encoding, decoding and expression of spatial information in dance communication.