Locust Wind Receptors: III. Contribution to Flight Initiation and Lift Control^*

The previous paper showed that interneurone responses of four different types can be recorded from axons in the locust cervical connective upon wind stimulation of the facial setae (Camhi, 1969 b). Of these four interneurone categories, the ‘wind-indicator’ and the ‘acceleration’ cells showed bilaterally symmetrical responses, producing about equal numbers of impulses for wind directed toward the head from either side. Since these two types of interneurone are unable to discriminate between wind from the left and wind from the right, it seems reasonable that any flight manoeuvres resulting from their activity would occur about the pitch axis; therefore, turns, yaws and rolls presumably are not controlled by these symmetrically responding cells.

Weis-Fogh (1949, 1950) demonstrated two bilaterally symmetrical flight reflexes of tethered locusts: (1) initiation, and (2) maintenance of wing beating, each resulting from wind stimulation of the facial setae. The previous paper showed that the flightmaintenance reflex probably results from the activity of wind-indicator interneurones. Flight initiation, consisting of the opening and incipient beating of the wings at wind onset, has not been studied in detail. Evidence presented here will show that the flight initiation following normal free jumps probably does not result from wind-receptor stimulation, or from loss of tarsal contact.

A third vertical flight control, lift regulation, is shown to be controlled in part by facial wind-receptor stimulation.

As in the previous papers all experiments were performed on male Schistocerca gregaria Forsk., phasis gregaria, between 2 and 6 weeks following final moult. The wind-tunnel construction was described earlier (Camhi, 1969 a).

For studying the start of a locust’s flight I photographed the insect with a Bolex H 16 ciné camera at 64 frames/sec. as he leaped from a rough-surfaced platform. Photographing a rapid-sweep stop-watch served to check the camera’s frame rate.

For further observations of flight initiation I arranged a tethered locust, having all legs and antennae amputated, in front of a black card mounted just above the screen of a dual-beam oscilloscope (Tektronix 502A). The card was visible to the oscilloscope camera, but the locust was not. A Leitz slide projector focused a 1 mm. diameter spot of light onto the extreme posterior tip of the closed forewings. At the instant the wings made the slightest move in the direction of opening, they ceased interrupting the light beam which therefore flashed a spot instantaneously on to the black card. The geometry was such that the spot flashed on to the card was in perfect vertical alignment with the two oscilloscope traces, which were spots, the sweep trigger having been turned off.

The oscilloscope displayed the instantaneous wind velocity, registered by a Flow Corporation 55A1 hot-wire anemometer. The anemometer’s active wire (200 μ diameter) remained fixed about 3 mm. above the facial setae. A Grass C4 kymograph camera recorded on continuously moving film at 1 m./sec. the oscilloscope wind trace and the projected spot. The distance on the film between the signal marking wind onset and the first appearance of the projected spot gave the latency of the wingopening response. I shall describe later a modification of this method which was also useful.

Slight modification of a technique used by Gettrup & Wilson (1964) allowed determination of forewing twist as a function of wind velocity flowing past the head. I positioned in front of the oscilloscope screen a tethered locust, having all legs and antennae amputated and the wounds sealed with wax to prevent bleeding. Each forewing bore a narrow marker of light-weight paper glued across the wing chord. The locust wore on the anterior and lateral margins of the prothoracic cuticle a shield of aluminium foil, firmly attached with wax. Probing measurements with the anemometer showed that the shield, if properly shaped, effectively prevented the wind from flowing over any part of the body other than the head. It also prevented the wind from coursing into the wing path. Certain of these shapes effectively allow flow over the facial setae to be laminar.

With the room darkened, a 35 mm. Nikon camera viewed the oscilloscope screen and the locust. Single or repetitive flashes of a strobe light (General Radio Corporation Strobotac) mounted above the locust allowed unblurred images of the moving wing and its marker to be photographed without illuminating the oscilloscope screen.

It was difficult to make a locust fly for long periods at low wind speeds. Occasionally, however, certain individuals would perform well at low air speeds or in still air when I gradually reduced the velocity from the normal flight speed. After reducing the wind to a particular velocity, I waited at least 1 min. before taking the photograph. Opening the camera shutter, set for a 1 sec. exposure, triggered a 1 sec. sweep of the oscilloscope beams. One beam displayed the instantaneous wind velocity recorded by the anemometer probe just above the setae. The other beam, connected to the output of the strobe light, marked the instant of each strobe flash.

To measure the effect of a given constant wind velocity on forewing twisting, I delivered either single or repetitive strobe flashes during 1 sec. exposures at different wind speeds. To observe any effect of wind acceleration on forewing twist, I opened the shutter, then rapidly accelerated the wind by a predetermined amount and immediately delivered a single strobe flash.

Using this technique different photographs showed the wing in different positions within the stroke cycle. Only those records showing the wing in the mid-downstroke were of interest, since it is at this instant that the wing generates most power (Jensen, 1956; Gettrup & Wilson, 1964). As the results will indicate, differentiating between mid-upstroke and mid-downstroke wing images was accomplished easily on the basis of the wing-twist angle.

The first series of experiments compared the latency from wind onset to wing openings for locusts under different conditions: first, with the insect creating its relative wind by jumping freely into flight, and secondly, with the insect tethered and exposed to wind flow from the tunnel. A similarity between the two latencies would lend support to the view that in natural flight the onset or acceleration of relative wind created by the take-off leap promotes reflex wing opening.

A locust, when placed on the launching platform, would usually explore the area for several minutes and then remain quiescent for a long while. Occasionally he would jump, usually several seconds after bringing the hindlegs into kicking position, allowing the photographer to capture the jump sequence on film. Reluctant individuals would usually jump in response to slight hand movements or gentle prodding. I photographed over 40 take-off sequences of six locusts.

Plate 1 shows the frame sequences of five typical take-offs. Sequences a and b show spontaneous jumps, while those of c–e were induced by hand motions or prodding. Approximately 15 m./sec. separates the action recorded in each frame. In each of the jump sequences, some degree of wing opening has ensued within 15 msec, of the onset of the take-off.

In the first paper of this series (Camhi, 1969 a) I showed that the conduction time of the wind-receptor neurones to the suboeso-phageal ganglion ranges between 7 and 14 msec. It seemed unlikely, therefore, that these receptors could mediate wing opening within 15 msec, of the onset of the jump wind. The point was examined, however, in the following series of experiments.

I arranged a tethered locust as described in Part 1 of the Methods section. I then accelerated the wind at different rates ranging from 100 to 600 m./sec.², terminating at velocities between 2·5 and 4·5 m./sec. These ranges encompassed all the values observed in records of free jumps, such as those shown in Pl. 1. I recorded over 150 such flight initiations of six animals.

Only velocities greater than 2·5 m./sec. provoked wing opening. The latency was shorter for faster accelerations within the range studied.

Text-fig. 1 shows the records on continuously moving film of two typical flightinitiation responses to wind onset. The latency to the beginning of wing opening is about 45 msec. The fastest response ever recorded was 35 msec. Thus the latency is two to three times as great as that observed in flight initiated by a free jump.

Controls performed on four of the locusts showed that covering the facial aerodynamic setae with petroleum jelly or low melting-point paraffin prevented wing opening in response to wind stimulation. The response returned upon peeling away the paraflin.

A second way of stimulating tethered locusts to fly is by removing contact between the tarsi and the substratum, thereby discontinuing a flight-inhibition reflex (Fraenkel, 1932; Weis-Fogh, 1956b). It seemed possible, therefore, that on a take-off jump the wings might open reflexively in response to loss of contact inhibition. Careful scrutiny of the jump sequences of Pl. 1 reveals that normally the wings begin to open before the tarsi appear to have left the ground. However, since it is difficult to determine from such photographs the exact instant a leg loses contact, I determined experimentally the latency between tarsal release and wing opening for a tethered locust. For tarsal release to control wing opening in free flight, this latency would have to be no more than about 15 msec.

I arranged a tethered locust in front of the black card as before, the closed forewing tips interrupting the projected 1 mm. light beam. The insect stood on a roughened rubber rod oriented in the plane of the body’s long axis. The rod, loosely hinged to a separate ring stand in front of the animal, formed an angle of about 30°, front end uppermost, with the body’s long axis. The locust usually extended his legs until contact had been made with the rod and then remained quiescent. Because the metathoracic legs are longest and the prothoracic shortest, the rod’s angle enabled the locust to stand with all legs almost fully extended. Any downward pivoting of the rod on its hinge attachment would therefore release contact of all tarsi almost simultaneously. All tarsi were released well before the rod’s tip had moved 1 cm.

The oscilloscope traces were two spots whose vertical positions were independently adjustable. Looking through the lens of the kymograph camera, I positioned the upper spot just below the rod, so that further lowering the rod by as little as 1 mm. would cut off the camera’s view of the spot. I then brought the lower oscilloscope spot exactly 1 cm. below the first. The room was kept dark throughout the preparative and experimental procedures.

After the locust had made contact and remained motionless for at least 30 sec., I turned on the camera to continuous moving film at 1 m./sec. and then released all tarsal contact by giving a gentle downward tap to the rubber rod. Flight ensued. The photographic record indicated the onset of tarsal release by the rod’s interruption of the upper oscilloscope spot. The interruption of the lower spot, 1 cm. below, indicated the instant before which all tarsal release must have been achieved. The first appearance on the film of the projected light beam indicated the first instant of wing opening.

Because the rod was mounted independently of the locust’s tether, virtually no mechanical shock reached the insect. Hand motions similar to those used in depressing the rod provoked no motion by the locust. I recorded more than 100 flight initiations using six animals, on three of which I amputated varying numbers of legs successively during the experiment.

Text-fig. 2 shows a series of typical records from one locust. The upper two traces are from the intact animal. The interval between lower beam interruption and wing opening is about 30 msec. In these two cases, then, the latency from the last instant of tarsal release to the first instant of wing opening was at least twice that required for a locust to open his wings following a free jump (Pl. 1).

The remaining three traces in Text-fig. 2 show the results of successive leg amputation. Reducing the number of legs to one (any leg) does not alter appreciably the latency from tarsal release to wing opening. This result gives confidence that the latencies reported here are not unnaturally long owing to insufficient tarsal contact with the rod prior to release—reducing over-all tarsal contact to one-sixth (one remaining leg) left the latency unaltered.

In summary, the results of this section suggest that in locusts, neither release from tarsal contact inhibition nor onset of the jump wind detected by the facial setae initiates reflexive wing opening in free flight. Either factor, however, is capable of producing wing opening under the artificial conditions of tethered flight.

To further this study of wind-stimulated symmetrical flight responses, it seemed of interest to determine whether forewing twisting—a critical factor determining the locust’s flight path (Weis-Fogh, 1956a; Gettrup & Wilson, 1964)—varies with the speed of a head-on wind directed at the facial wind receptors. Tethered locusts flew with prothoracic wind shields and forewing markers as described in part 2 of the Methods section. I made a total of more than 2400 separate exposures using six different locusts flying in response to winds from o to 4·0 m./sec. For about two-thirds of these photographs the wind tunnel maintained a constant flow rate. For most of the remaining photographs I accelerated the wind and then immediately flashed the strobe light. Finally, I performed several control measurements by moving the wind jet away from the head and directing the air stream directly over the wing. Frequent checks indicated that both forewings acted symmetrically, although only one was consistently photographed.

Plate 2 shows some typical photographic records. Plate 2 a is an example of a multiple exposure at constant flow rate. Seven strobe flashes captured the wing in about the mid-positions of seven different strokes. Of the seven marker images produced, four are almost parallel and show the wing to be supinated. This indicates that in these four cases the wing was captured in the upstroke (Jensen, 1956). The other three parallel marker images show the wing to be pronated, and therefore in the downstroke. The angle between a wing marker and a horizontal oscilloscope grid line provided a convenient measure of relative twist angle.

Plate 2 b and c shows typical single-exposure photographs of wing angle following an abrupt wind acceleration. The upper beam marks the instant of the single strobe flash.

The histogram of Text-fig. 3 plots all the mid-cycle forewing twist angles recorded for a typical locust in response to maintained and accelerated wind. The twist angles labelled along and abscissa, begin arbitrarily with o°, the angle of the most pronated wing image photographed. The left-hand groups of bars represent mid-downstroke and the right-hand groups mid-upstroke twist angles. Stippled bars represent the twist angles of wing images produced between 80 and 200 msec, after wind acceleration from rest. Bars marked by an ‘×’ show twist angles less than 80 msec, after acceleration. Empty bars show twist angles for locusts in wind of maintained velocity.

In agreement with the view that wing twisting is passive on the upstroke (Weis-Fogh, 1956a;,Gettrup & Wilson, 1964), the upstroke twist angle remained essentially unchanged for all wind speeds and accelerations. The downstroke twist angle of the forewing, however, depended upon the velocity of a constant wind. Higher velocities resulted in less than normal downstroke pronation. The effect continued for at least 10 min., the longest interval studied. Text-fig. 4 plots the relationship between maintained wind speed and downstroke twist angle, and shows it to be non-linear; small speed differences have greater effect at low than at high wind speeds.

The histogram of Text-fig. 3 further shows that between 80 and 200 msec, after an acceleration from rest to a given final speed, the decrease of downstroke pronation was exaggerated as compared with values for constant wind at that same speed (see also Text-fig. 4). However, photographs taken less than 80 msec, after such acceleration displayed twist angles similar to those of a locust flying in still air. Control measurements, with wind directed at the wing, likewise revealed twist angles like those in still air.

In summary, then, this section shows that for higher velocities of wind over the head, forewing pronation decreases on the downstroke. Wind acceleration from rest to some maintained speed results, after a latency of about 80 msec., in an exaggerated decrease of forewing pronation. This extreme initial decrease levels off over the next few hundred milliseconds, to the value characteristic of the given maintained wind speed.

Experiments reported in this paper show that the wind-stimulated flight-initiation response, observed in tethered locusts, proceeds too slowly to be the mechanism inducing wing opening following a free jump. The flight initiation resulting from loss of tarsal contact is likewise too slow. It is of interest, then, to inquire how the locust in nature opens his wings and begins to fly.

A fairly simple explanation suggests itself. Locusts, like most other insects, do not possess any flight muscles whose sole function is opening the wings. Rather this task devolves largely upon the wing elevator muscles, which also are crucial to the normal, patterned flapping of the wings (Wilson & Weis-Fogh, 1962). Waldron (1968) has shown that these muscles are the first to contract at the onset of flight. It is reasonable to suggest, then, that to open its wings at the beginning of a flight, a locust need only turn on its central flight motor. One can induce this process in a tethered locust by stimulating the facial wind receptor or by releasing the tarsi. However, when a locust jumps freely into the air, presumably a separate pathway, probably within the central nervous system, turns on the flight motor. Waldron (1968) has recently made a similar suggestion.

A second finding reported in this paper is that increased wind flow over the facial setae of a tethered locust results in decreased forewing pronation on the downstroke. The extent of the decrease is a function of wind velocity and is exaggerated just following acceleration.

The effects these wing-stroke changes impose on an insect flight are well established (Weis-Fogh, 1956a; Jensen, 1956). Decreased downstroke pronation in both forewings increases the ratio lift/drag, resulting in ascending flight.

It was usually not possible to perform the control experiment of covering the facial setae in order to prove definitively the involvement of the aerodynamic sensory structures. Locusts rarely would maintain flight with the setae covered. Nevertheless, with the antennae removed and the shield restricting wind to a course over the head, no other receptors appear capable of subserving this function.

The drastic leg amputation, without which the locusts would break the anemometer wire, was a potential source of error. However, Wilson (1961) showed that apparently normal flight occurs with the legs removed. Moreover, the reproducible flight response to different air speeds could hardly have resulted from amputation shock.

The previous paper (Camhi, 1969 b) reported neuronal responses of four different types recorded in wind-sensitive interneurones of the cervical connective. Two of these, the ‘wind-indicator’ and the ‘acceleration’ cells, gave symmetrical responses to wind from either side of the head, suggesting that any wind-stimulated flight manoeuvres which they mediate would be restricted to movements about the pitch axis. The major difference between these two cell groups was that wind-indicator cells adapted very slowly, while acceleration cells were highly phasic. Also, the windindicator cells responded to threshold air speeds as low as 0·1 m./sec., whereas threshold for acceleration cells was about 1·0–1·5 m./sec.

The lift-control reflex reported in this chapter cannot result from the activity of acceleration interneurones. The behavioural response is greatest for wind velocity changes below 1 o m./sec., speeds at which acceleration cells remain silent (Camhi, 1969b). Moreover, the acceleration cells’ phasic response expires after a few hundred milliseconds, whereas the lift-control reflex continues for at least 10 min. (Camhi, 1969b). The responses of the other two interneurone types found, the 𠆘wind-direction’ and the ‘recentre’ cells, were also too phasic to subserve the prolonged lift regulation reported here.

It therefore seems reasonable to suggest that wind-indicator interneurones mediate the lift-control reflex. This suggestion implies that wind-indicator cells connect possibly through other interneurones, with mesothoracic motorneurones controlling the forewing twist muscles. Earlier findings (Camhi, 1969b) suggested that these same interneurones also connect to the pterothoracic flight motor.

The histogram of Text-fig. 3 indicates that the lift-control reflex requires some 80 msec., or between one and two wingbeats, to affect forewing twisting. That the locust responds thus rapidly to sensory input contrasts with the view of Waldron (1968) that sensory control of flight in this insect generally requires tens of wingbeats to produce any observable effect.

The role that the lift-control reflex might play in natural flight is a matter of speculation. However, the physiological data correlate closely with at least one aspect of flight behaviour. As Pl. 1 shows, a take-off jump is generally followed by ascending flight, which implies less than normal downstroke pronation of the forewings. The take-off jump produces a wind acceleration of magnitudes which, as this paper shows, result in a temporarily exaggerated decrease in forewing pronation within about 80 msec. At this time after the onset of a jump a locust would be in his first or second complete wing cycle. It is possible, then, that one function of this wind-stimulated lift-control reflex is to promote rapid ascent just following wing opening on the take-off jump.

I have also shown (Camhi, in preparation) that the reflex is probably used, in conjunction with changes in the abdomen posture, to prevent stalls at low flight speed by inducing a dive when air speed becomes critically low.

I wish to thank Professors Ian Cooke and Kenneth Roeder for their most helpful critical reading of this manuscript.