1. Obstacles approaching in the flight path trigger postural and wing kinematic adjustments in tethered flying locusts. We sought to confirm that these behaviours were steering behaviours by measuring the changes in the flight forces associated with their execution. We also investigated the coordination of these behaviours in the execution of collision avoidance manoeuvres and the effect of speed or size of the obstacle on the timing and magnitude of the response. 2. Locusts were tethered and suspended in a wind tunnel from orthogonally arranged leaf springs mounted with strain gauges. Lift and yaw torque could be monitored unambiguously. We also monitored a forward translation force which combined pitch and thrust. During flight, the locusts were videotaped from behind while targets of different sizes (5 cmx5 cm, 7 cmx7 cm, 9 cmx9cm, 11 cmx11 cm) were transported towards the head at different speeds (1, 2, 3 or 4 m s-1). 3. Angular asymmetry of the forewings during the downstroke with the right forewing high, and abdomen and hindleg movement to the left, were temporally associated with an increase in yaw torque to the left. With the left forewing high, abdomen and hindleg movement to the right were temporally associated with a decrease in yaw torque to the left. Obstacle avoidance behaviours could be associated with either an increase or a decrease in the pitch/thrust component. 4. Leg, abdomen and wingbeat alterations in response to the approach of an obstacle were independent but tightly coordinated. Slower approaches increased the magnitude of the responses. However, the size of the obstacle had no effect on the magnitude of the response. Slower and larger targets generated earlier reactions (i.e. locusts reacted when the targets were further from the head). 5. We conclude that the behaviours we have described were steering behaviours which would have directed the animal around an obstacle in its flight path, and that there were at least two strategies for collision avoidance associated with slowing or speeding flight. Leg, abdomen and wingbeat alterations formed a coherent avoidance response, the magnitude of which was dependent upon the time available for it to develop. We further conclude that the manoeuvre was not initiated at a constant time to collision and we propose that the avoidance strategy was to initiate the manoeuvre when the targets subtended more than 10 s in the insect's field of view.

This content is only available via PDF.