Visuo-olfactory integration in Drosophila flight control: Neural circuits, behavior, and ecological implications

Most human beings on the planet perceive flies as nuisances that are maddeningly difficult to swat. However, if we consider the sophistication of flight maneuvers in these animals, along with the limited computational and sensory capacity available to them, one realizes that we might learn a lot from studying exactly how the fly flies. The fruit fly Drosophila melanogaster has a brain roughly the size of a sesame seed, consisting of 100,000 neurons, about 10,000 times fewer than that of a human. The fly visual system has a resolution 1,000 times worse spatially than ours, but perhaps makes up for this by having 3-4 times better temporal resolution. It is the fly's sophisticated ability flight ability that has made Diptera one of the most successful genera on the planet, and as of yet, no human engineer has come even close to replicating its capabilities. Throughout my graduate career, I have sought to understand how the fruit fly integrates visual and olfactory information to control its flight. By understanding how such sophisticated control architecture can function with limited sensory information and be packaged into such a small space, we may be better able to gain insight into how nature has solved a difficult engineering problem, as well as gain basic knowledge on how animals transform multi-modal sensory information into behavior.;Because of poor visual resolution and limited neural resources, flies cannot rely on visual object identification to locate food resources. Instead, they must integrate information from visual motion and olfactory cues. Like all animals possessing image-forming eyes, fruit flies possess strong visually-mediated reflexes in response to motion. This reflex, the optomotor response, analogous to the optokinetic response in human vision, is highly robust and results in a turn in the same direction as visual motion. Although the optomotor response is critical for flight stability, it provides no information about food resources in the environment. Nor is olfactory information alone enough for fruit flies to locate food resources. Instead, flies require both visual and olfactory information to successfully track odors to the source.;The fly's visual reflexes are so strong that same animal will respond stereotypically to the same stimulus hundreds of times. However, this poses a dilemma: robust reflexes must produce adaptive behavior in a number of different contexts. In the presence of an odor plume, the fly must reliably track the plume to its source, and in the absence of odor, the fly requires greater variability in its flight control to locate spatially unpredictable resources. One way in which the fly might resolve this dilemma is to modify visual reflexes in a context-dependent manner depending on the presence or absence of odor. This has the advantage of leaving the fly's basic flight control algorithm intact, while inserting enough flexibility to accomplish differing goals.;In this dissertation, I demonstrate that olfactory cues modify the basic flight algorithm in two ways: 1) An odor gradient across the antennae creates a tonic steering bias towards the more heavily stimulated side and 2) olfactory cues modify the gain and reliability of optomotor reflexes (the animal equivalent to human optokinetic eye movements). In the presence of an odor plume, such reflexes ultimately result in successful localization of the odor source and constitute a robust algorithm for flight control. However, despite explaining a large amount of the variation of the fly's response to external motion perturbations, optomotor responses cannot fully account for the self-generated distribution of turns by flies starved for differing periods of time. Lastly, via chemical and genetic manipulation, I show that olfactory-mediated increase in optomotor reliability depends specifically on the mushroom bodies, the fly's second order olfactory center. The mushroom bodies are a highly conserved and ancient brain structure in insects, previously supposed to be involved primarily in olfactory learning and memory. I posit that the mushroom bodies function as a sensory weighting mechanism to switch between two states; if an odor is present, it increases the influence of visual information on behavioral output to increase the fidelity of odor plume tracking, and in the absence of odor, decreases the importance of visual information to increase behavioral variability. These results add to accumulating evidence that the mushroom bodies subserve behavioral flexibility in insects and plays a critical role in their ecology...