An investigation was made to find which simple parameters of a visual stimulus are critical to induce a landing response in tethered flying Drosophila melanogaster.
The stimulus consisted of a circular black disk, against a white background, moving towards the fly. A glass plate was placed between fly and disk to ensure that only a visual stimulation was presented. The landing response involves an extension of the forelegs in front of and above the head. To monitor this movement, the shadow of the fly was projected by red light through a microscope onto a small screen (Erber & Schildberger, 1980) and the motion of the forelegs was recorded by two light gates at different positions with respect to the fly’s shadow (Fig. 1 inset). The disk approached the fly at a given velocity and its movement was timed, from a marker at a fixed distance from the fly, up to the moment when the first light gate detected the start of the landing response. The expansion of the stimulus covered an object angle subtended on the fly’s retina (a) of 10—95°. The diameter of the disk was 35 or 70 mm. The range of disk velocities was 5-40 cm s-1.
In a first experiment, short (20 ms) expansion stimuli were used which elicited a landing response with a latency of 50 ± 15 ms, similar to values obtained for Musca domestica (Wagner, 1982; Borst, 1986).
The object angle subtended on the fly’s retina 50 ms before the onset of the landing response dropped from 80° at small retinal expansion velocities to a nearly constant value, about 50°, above retinal expansion velocities dα/dt of 300°s−1 (Fig. 1). This angle was independent of disk diameter. So the absolute disk angle on the fly’s retina was the critical parameter and not the absolute distance between stimulus and fly. From measurements of landing activity an analogous result for larger flies was obtained when they were tested with variable disk diameters and a constant disk velocity (Goodman, 1960; Eckert & Hamdorf, 1980).
The angle subtended on the retina has also been found to be critical for male Syritta pipiens to compute their distance from a female (Collet & Land, 1975). In contrast, the velocity of the landing object has been found to be critical for the landing response in Lucilia sericata (Goodman, 1960) and for the onset of final deceleration before landing of free-flying Musca domestica (Wagner, 1982).
Since larger flies fly faster and have a greater momentum to drag ratio, it might be necessary for them to shift the landing manoeuvre to greater distances at higher flying speed which would require a longer stopping distance. Since the landing response of Drosophila is elicited at a constant object angle, there will be insufficient time to exhibit the landing response at high disk velocities.
The performance time of the landing response was measured as the time taken for the flies to extend their forelegs from light gate 1 to light gate 2. The performance time was 35 ms and was independent of the disk velocity. Together with the latency of 50 ms this gives Drosophila 85 ms to land before the object expands to 180°. This indicates that at an expansion velocity above 1600°s-1 (at α=95°) it is impossible for Drosophila to land in time when the landing response is induced at an object angle of 50°. When the landing response was carried out, all the experimental flies landed in time, but at high expansion velocities, above 800°s-1, the probability of landing decreased (Fig. 2). Thus the experiments were carried out over the whole range of expansion velocities in which the landing responses could be elicited with high probability. Furthermore, this shows that the landing mechanism of Drosophila is optimally triggered only when the landing response can be carried out in time.
I thank Professor H. C. Spatz, Dr J. E. Treherne, Dr R. Willmund and Professor K. F. Fischbach for careful reading and for useful comments on the manuscript.