Role of wing pronation in evasive steering of locusts

Abstract

Evasive steering is crucial for flying in a crowded environment such as a locust swarm. We investigated how flying locusts alter wing-flapping symmetry in response to a looming object approaching from the side. Desert locusts (Schistocerca gregaria) were tethered to a rotatable shaft that allowed them to initiate a banked turn. A visual stimulus of an expending disk on one side of the locust was used to evoke steering while recording the change in wingbeat kinematics and electromyography (EMG) of metathoracic wing depressors. Locusts responded to the looming object by rolling to the contralateral direction. During turning, EMG of hindwing depressors showed an omission of one action potential in the subalar depressor (M129) of the hindwing inside the turn. This omission was associated with increased pronation of the same wing, reducing its angle-of-attack during the downstroke. The link between spike-omission in M129 and wing pronation was verified by stimulating the hindwing depressor muscles with an artificial motor pattern that included the misfire of M129. These results suggest that hindwing pronation is instrumental in rotating the body to the side opposite of the approaching threat. Turning away from the threat would be highly adaptive for collision avoidance when flying in dense swarms.

Appendix: Extraction of wingbeat kinematics from movies

Data from three spatially calibrated high-speed video cameras were used to extract the motion of the four wings in 3D space. In the films, we digitized 14 points on the body and wings. Points 1–4 were the bases of the four wings, and their mean change in position with time (frame rate) was used to monitor the rotation of the thorax. Points 5–8, the wing tips, were used to extract the wing tip path. On each of the forewings we also digitized two points (points 9–12) one at the leading edge and the other at the trailing edge of the wings at 0.6 of the wing length (wing length was defined as the distance between the base and tip of the same wing). The last two points (points 13, 14) were at 0.65 of the wing length on the trailing edge of the hindwings. To locate points 9–14 in the films, we used the natural pigmentation of the forewings and the third annal vein of the hind wing (Wootton et al. 2000). Points 9–14 were used to determine the orientation of the local chord of the wing, i.e. the angle of incidence and the angle-of-attack (AoA). The digitized data from the three camera views were converted to 3D coordinates using a DLT coefficient matrix derived during camera calibration (Hedrick 2008). To describe the wingbeat kinematics from the time series of the coordinates, we used conventions similar to Walker et al. (2009). The trailing edge of the hindwings is highly deformable and the wing camber undergoes change during flapping (Wootton et al. 2000; Walker et al. 2009). It is thus impossible to represent accurately the kinematics of this wing using three points on the wing edges. Nevertheless, our objective here was not to extract the detailed deformability of the wing but rather to arrive at a simple quantitative estimate of the flapping asymmetry. For this purpose, we extracted the kinematics as if all wings were rigid and flat. Thus, the terms ‘angle of incidence’ and AoA are used loosely here to refer to the angles of the straight line connecting the leading and trailing edge in the sagittal plane. The above description refers to the kinematics of one wing as the locust flies horizontally in a straight line. As will be shown, the locust responded to the looming object by rolling its body. When extracting the kinematics of all four wings we corrected for rotation of the body and extracted the kinematics in a frame of references where X is the horizontal upwind (flight) direction, Y is the lateral axis of the insect (left side positive) and Z is orthogonal (positive toward the dorsal side of the insect). The axes reflect an insect frame of reference, correcting for the rotation of the locust relative to the cameras. In this frame of reference, the wing movements are relative to the thorax, providing a valid comparison with EMG patterns. To indicate graphically flapping asymmetry between the left and right wings (e.g. Fig. 3 of the main text), we repeated the analysis with the left and right forewings having the same stroke plane (averaging the left and right stroke plane angles). The same was performed for the hindwings. The data points at the trailing edges were also used to determine numerically the wing velocity vector from the change in position with time (Rayner and Aldridge 1985) in the fixed lab-based coordinate system, where X and Y were horizontal with the positive side of X pointing upwind. The wind speed was added along the X component of the wing velocity vector and the new vector for motion relative to air was used to find the local AoA of the wing.