Max-Planck-Institut für biologische Kybernetik, Tübingen, F.R.G.
Our goal is to analyze the neuronal circuitry underlying the turning behavior of walking flies (Drosophila melanogaster). We modified a video-based automated leg-motion detector designed for straight walking so that we could collect a large body of data for the statistical analysis of wild-type turning behavior, a prerequisite for further analyses in mutant flies.
Visually induced turns of wild type flies (Berlin) were categorized as "turning on the spot" when the translational progress during turning was smaller than half a fly's body length; otherwise they were "curved paths".
Velocity in curved paths is variable (10..35 mm/s). As in straight walking, variation is achieved by concomitant changes in step length and stepping frequency (6..14 steps/s). (i) For legs on the outside of a curve, step length depends on stepping frequency, just like during straight walking. (ii) Turning is achieved exclusively by diminishing the step lengths on the inner side. (iii) This adaptation is at least partially achieved by shortening the duration of swing phases. (iv) The number of steps per leg on the inner and the outer body side were always identical, even during rarely observed transitions from the standard alternating tripod gait into a tetrapod coordination of legs.
The orientation of the body relative to the direction of progression is astonishingly variable and may differ by up to 40°. In curves, the direction of the body may lead or trail. All recovery strokes then point in the intended direction of progression and, hence, are oriented obliquely to the body axis.
Turning on the spot may consist of two or three consecutive phases distinguished by different translational components. Only in 3-phase turns is the first phase directed backwards. In the next phase the fly turns around its abdominal tip and finally it increases its forward velocity. The angular velocity during the latter two phases remains about constant until the fly is headed in the desired direction. During several consecutive steps of a final phase all the legs already recover in the desired direction. A representation of this direction is likely to exist.
We are currently screening 230 mutant lines that were behaviorally isolated for their walking impairments (Strauss, J Neurogenet, in press). Of special interest is the role of the central complex in turning. Structural central complex mutants show altered step length characteristics (e.g. Strauss et al. J Neurogenet 8:125-55).
The method in brief: The detector system was initially developed to analyze straight walking (Strauss, J Neurogenet 8:250). Flies walk freely on a glass surface covered with a thin layer of red laser light. CCD cameras (overall time resolution: 100 Hz) image the laser light reflected only by the tarsi of legs in contact to the ground. For the present analysis we added a third camera that images the fly from above. Orientation and position of the longitudinal body axis is extracted on-line and automatically combined with the spatial and temporal reconstruction of the leg actions. A concentric cylinder of 4 x 180 LED's around the detector allows for the display of rotating visual stimuli.