Long gone are the days in which behavior analysis of animal models relied on labor-intensive, frame-by-frame manual movie analysis. Newer, sophisticated techniques are becoming available that can automatically annotate complex and detailed behaviors even in tiny animals.

Fly on a floating ball. Credit: Image courtesy of B. de Bivort.

The fruit fly Drosophila melanogaster has been an all-time favorite for studying the genes and neural circuits underlying behavior. Despite its name, the fly is most interesting to researchers when it walks. “To generalize the behavior very broadly, what flies are doing while they are flying is smelling fruit and going towards the fruit; but once they find it, all the interesting behaviors that flies do—like battling for dominance or mating—are done while they are walking,” explains Benjamin de Bivort, who runs a lab as a Junior Fellow at the Rowland Institute at Harvard University.

However, tracking walking behavior in flies has up to now been relegated to relatively low-resolution analyses such as monitoring walking trajectories, the average speed of a group of flies or the ability of a fly to fulfill a simple motor task such as climbing a wall. de Bivort and others soon realized that to understand the mechanisms behind locomotion in this organism, they needed to be able to record the position and motion of each leg. Two methods now provide tools to track and quantify fly walking behavior at this level of detail, expanding the potential of the fruit fly as a model organism for the study of locomotion and the neural circuits that control it.

The first of these methods, developed by Richard Mann and his colleagues at Columbia University, uses an optical technique called frustrated total-internal reflection to track the legs of individual flies as they walk (Mendes et al., 20132). In this setup, the flies walk freely in a small arena with a glass surface. The researchers illuminate the glass, and a camera positioned under the surface detects the interference in the light reflection patterns produced by the insect's legs as it walks. Tracking the fly's footprints is automated thanks to software that the group also developed.

de Bivort, on the other hand, was interested in tracking walking flies over long time periods, and he developed a system in which flies are tethered by their thorax and walk on a treadmill of sorts (Kain et al., 20131). The fly is mounted on top of a clear plastic ball that floats on a cushion of air, and the fly turns the ball as it walks. By making the ball transparent, the researchers managed to image the legs from below using a custom-built imaging system. They glued small pieces of film colored with infrared dye onto each of the fly's legs and illuminated the legs with a laser while recording the fluorescence from below using a couple of cameras. They used another sphere, positioned under the first one, to collimate the image coming from the fly's legs as they moved. de Bivort and colleagues also developed custom software that tracks and quantifies walking behaviors and that uses machine learning approaches to identify instances in which the fly is engaged in specific behaviors such as turning or grooming.

de Bivort now wants to use two-photon microscopy to visualize the activity of specific neurons in the fly's nervous system while tracking walking behavior at high resolution. Pairing detailed behavioral analysis with cellular-level physiological measurements seems a promising path to take.