Evaluating the feasibility of using sound to track automatically honey bee recruitment dances
Honey bee foragers communicate to nestmates where they have found food, and if we eavesdrop on their conversations honey bees can be powerful bio-indicators of information on food availability for declining insect pollinators in space and time. Determining when and where food is available for pollinators is a critical first step in conserving and enhancing their dwindling diversity and abundance. Waggle dances that are used by the honey bees to communicate contain information on honey bee foraging dynamics, and the dances can be manually decoded from video, but only if the bees are housed indoor, in glass-walled observation hives as opposed to regular, outdoor hives. For a number of reasons, doing large-scale monitoring of bee dances is almost impossible. First, the manpower needed for manual decoding of waggle dances from video recordings of observation hives is immense. Second, observation hives are difficult to maintain and are best suited for short-term experimental setups. Last, video equipment is expensive and not suited for everyday unsupervised field conditions. These constraints hinder us to achieve meaningful large-scale spatio-temporal data to support the conservation of pollinators. To overcome manpower limitations, researchers have improved the automation of dance decoding from video. Still, the other prohibitive factor around using bees as bio-indicators is the expensive specialist equipment needed for automated decoding of bee dances. This limitation can potentially be overcome by decoding dances not from the visual signal on video recordings, but from the sound that dancing bees emit. Using acoustic near-field holography, we may be able to locate dancing bees on the dancefloor and track them over time to decode the food source locations they had been visiting.
The goal of this study was to establish a collaboration between the Departments of Entomology and Mechanical Engineering, to study the acoustic properties of the waggle dance, and to compare the acoustic source localization method to the gold standard of decoding dances from video. Over the past year, we have successfully established the collaboration, and we have undertaken concurrent acoustic and video recordings on multiple occasions. Our preliminary data suggest that acoustic source localization may be feasible. Recordings undertaken last summer and fall suggest that dances may be isolated from the noise in the bee hive. However, our recordings also demonstrated that the bees will readily deposit wax on the microphones, and that we need to protect microphones against this. Over the winter months, we worked to improve the microphone array design, and we had hoped to pick up measurements this spring. Unfortunately, high winter mortality in the apiary made that impossible, as we had to order bee colonies from further south to install observation hives. We have now installed these new colonies in observation hives at the Prices Fork Research Station, and they have reached sufficient strength to test further the acoustic properties of the dance. In the coming summer months, we will continue our measurements in the observation hives to firmly establish whether acoustic source localization is possible. If we succeed in our proposed methodology, using honey bees as bio-indicators will become possible for wide-scale adoption with immense applicability for agricultural settings.