Foraging for food: multimodel sensory fusion in freely flying fruit flies

Abstract

The ability to find food by tracking wind−borne odor plumes to their source is one of the most critical yet difficult tasks an insect performs. In a natural environment, turbulent air breaks apart the odor distribution in a plume, resulting in packets of high concentration interspersed with clean air. The visual sense, however, provides continuous information about where objects are, but very little about what they are. Thus, it would seem prudent for an animal to integrate the two sensory cues to maximize their ability to localize food sources. In this study we focus on the fruit fly, and how they are able to track a time varying plume of an attractive odor to its physical source, and whether or not they decide to land on it. To answer these questions we built an experimental rig capable of delivering predictable pulses of odor into a windtunnel with minimal turbulence. We used a mini PID to characterize the odor pulses and build an accurate model, allowing us to predict the time varying odor landscape in the wind tunnel. To study how the flies integrate this olfactory cue with their visual sense we added a vertical black post near the plume. Using a 9−camera tracking system we were able to track the flies in 3D as they flew through the wind tunnel with different olfactory and visual scenarios. Preliminary results suggest that flies that recently passed through an odor plume are 3 times more likely to land on a nearby object (N=699), compared to flies who have not experienced any odor, yet flew within the same general area (N=879). Furthermore, the effect of the odor stimulus appears to persist − flies that have experienced odor, but less recently, are 7 times more likely to land than in the control case (N=679, 686, resp.). In summary, our unique experimental paradigm has allowed us to begin probing the roles of olfaction, vision, and memory, in food finding behavior in freely flying fruit flies.