The Role of Certain Optomotor Reactions in Regulating Stability in the Rolling Plane During Flight in the Desert Locust, Schistocerca Gregaria

Rainey & Ashall (1953) reported field observations on the desert locust, Schisto-cerca gregaria, which suggested that a dorsal light reaction was operating. Gettrup (personal communication), in experiments on the control of forewing twisting in S. gregaria with insects previously selected for straight, forward flight, found that blind locusts flew stably over a distance of 10 m. Haskell (1960) in field experiments observed that although complete blinding did not interfere with the initiation and maintenance of flight in S. gregaria, in the majority of cases instability eventually became apparent and the insects went into a spin and crashed, although some managed to fly as far as 100 m. In view of these results the effect of light on stability in the rolling plane was examined first when surveying the part played by optomotor reactions in controlling locust flight, and in particular tests were made for a dorsal light reaction (Goodman, 1959). Although such a reaction undoubtedly does play a large part in controlling stability in this plane in some insects it cannot be the only mechanism operating since it would presumably produce curious effects as the sun changes its position in the sky. Another possible source of visual stimulation which does not suffer from this disadvantage is the line of the horizon, and an attempt has been made to determine whether the flying locust responds in any way to rotation of the horizon.

Mature male and female specimens of S. gregaria were used in the experiments described in this paper. They were mounted in a support which permitted them to fly but gave them freedom of movement in one plane only, i.e. freedom to rotate about their longitudinal axis in the rolling plane. Each locust was suspended between two metal rods attached at each end to a wheel which could pivot freely about a horizontal bar projecting from a vertical supporting pillar (see Fig. 1 A). It was necessary to mount the locust so that its centre of gravity was in such a position that it could turn freely and remain stable in any position about its longitudinal axis. The ventral surface of the thorax and abdomen was waxed to a metal plate borne on one of the rods, the tarsi being removed to prevent them from gripping the framework. When the locust was in position a screw was lowered through the upper metal rod until it just touched the dorsal surface of the thorax. The centre of gravity of the locust and its support could be adjusted if necessary by adjusting the position of nuts borne on horizontal and ventral arms of the supporting framework. Rotation of the locust was measured with the aid of a circular Perspex disk, graduated into 360°, attached to the rear vertical pillar with its centre in line with the longitudinal axis of the locust. When the locust was in a normal upright position its median sagittal plane was aligned exactly with the vertical (0−180°) axis of the disk. Any departure from this position was measured by recording the angle between 0° and the median sagittal plane (see Fig. 1B).

Each piece of apparatus used in the experiments was fitted to the end of a wind tunnel having speeds continuously variable up to 6 m/sec. The locust was placed with its head in line with the axis of the wind tunnel and the wind speed was adjusted until it was flying steadily. Locusts were first flown in diffuse illumination in a square metal box, apparatus A. The wind tunnel fitted into the front wall ; the rear wall was completely removed, three of the side walls were matt black metal and the fourth consisted of a sheet of opal glass (see Fig. 2 A). Depending upon the orientation of the box the locust could be flown with diffuse illumination overhead, beneath or on either side. In a number of cases the fourth wall was not illuminated and the orientation of flying locusts in darkness was examined. A microphone placed just beneath the locust made it possible to detect whether it was flying during the test period. Its orientation at a given moment was examined by switching on a dim red light outside the box for the shortest possible time necessary to note the angle between the median sagittal plane and o^°.

A second piece of apparatus (apparatus B, Fig. 2B) was used to stimulate the locust with parallel beams of light of variable diameter and intensity whose direction, in relation to the position of the locust’s head, could be known accurately. As before, it consisted basically of a square, matt black box, with the wind tunnel projecting into the front wall and the rear wall completely removed. On the outside of each of the four remaining walls was a box in which a parallel beam of light was produced by a suitable arrangement of convex lenses and slits in front of a 150 W. tungsten filament lamp. After passing through a heat filter the beam was projected into the central box. The four beams met in the exact centre of the box at 90° to one another, a position occupied by the locust’s head under experimental conditions. The intensity of each beam could be varied by means of a rheostat in series with the lamp and the diameter by inserting slits of different sizes in the path of the beam. The intensity of the beams was measured by placing a black pyramid mounted horizontally in the position normally occupied by the locust’s head so that one beam fell on each face. Intensities could be read directly in log. ft.-lambert with an S.E.I. photometer.

Each locust was flown in front of the wind tunnel for half an hour in a dim light, and if flying strongly at the end of this period was used for experimental purposes. After i min. in darkness one of the beams of light was switched on and the response of the locust was recorded. Its position when the light was switched on, the direction anddegree of rotation, if any, and the time taken to reach a stable position were each noted. The settled orientation of the locust was measured in relation to the direction of the beam of light by measuring the angle between the axis of the beam and the median sagittal plane of the locust to the nearest 5⁰ (see Fig. 1B). This angle is hereafter referred to as the angle of orientation. The beam was switched off, and, after a suitable interval which preliminary tests established could be as short as 1 min., another beam was switched on and the recordings were repeated. Beams were switched on in a random sequence for the required number of tests.

Locusts were also stimulated by rotation of an artificial horizon in apparatus C (see Fig. 3), by projecting the framework in which they were mounted into the centre of a hemisphere (radius 1 ft.) constructed of stiff cardboard. A black line, 1 in. in width, forming the horizon, ran round the horizontal circumference of the hemisphere, separating two equally bright, white sectors. In order to produce differences of brightness between the two sectors for some experiments one sector was painted successively darker shades of grey. A plain white hemisphere, with no horizon, was used in some cases as a control. The wind tunnel was admitted through an aperture in the hemisphere at horizon level. Diffuse illumination was provided by a 150 W. lamp placed behind two sheets of opal glass at the rear of the locust. The intensity could be varied by means of a rheostat. The hemisphere was rotated by hand, the degree of rotation being controlled by placing stops at suitable intervals around the circumference.

Electrical recordings were made from the ventral nerve cord of some locusts whilst they were stimulated by rotation of an artificial horizon. A second piece of apparatus (apparatus D, Fig. 4) was used here so that the degree and speed of rotation might be more accurately controlled. The locusts were placed in a matt black metal box used for screening purposes, facing an aperture into which the end of a black cylinder projected very slightly. The end of the cylinder was closed by a sheet of opal glass half of which was painted black. It was illuminated from inside by a 100 W. bulb whose intensity could be varied by means of a rheostat. The cylinder was rotated through known angles by means of a $16$ h.p. motor in conjunction with a variable speed gear.

Wingless and legless locusts were waxed by their ventral surface to the edge of an adjustable Perspex platform so that their heads projected over the edge and were free to move. The platform was adjusted so that the centre of the locust’s head was in line with the centre of the artificial horizon. A rectangular window was cut in the dorsal surface of the thorax, taking care to leave the prothorax intact. The gut was drawn through this opening and removed and two silver wire electrodes, leading to a d.c. amplifier and oscilloscope, were hooked separately over each connective between the first and the second thoracic ganglia, so that simultaneous recordings could be made from the two ventral connectives.

The eleven locusts tested were mounted and placed in front of the wind tunnel in apparatus A in complete darkness with the dorsal surface uppermost. Their flight for the next half an hour was monitored by listening to their wing beat and by occasional brief examinations in red light. If they flew steadily for this period with a normal wing movement they were accepted for the experiment. The position of each locust was examined and recorded at frequent intervals for as long as it appeared to be flying well, periods which ranged from 1 ·5 to 5 hr. As soon as the locust appeared fatigued the experiment was discontinued. None of the locusts tested remained in their initial position but rotated about their longitudinal axis, generally fairly slowly, sometimes coming to rest in one position for some minutes and then continuing to rotate in the same or in the opposite direction. Occasionally they began to spin rapidly and this would be maintained for periods of up to 5 min. At each examination the angle which the median sagittal plane of the locust made with the vertical at that moment was recorded. When the locust was spinning too rapidly for any reading to be taken the direction of the rotation was recorded. A typical response given by one locust is illustrated in Fig. 5. The locust’s position was noted 250 times over a period of 4·5 hr. On 180 occasions it was possible to record its position with respect to the vertical axis, on 42 occasions it was spinning rapidly in an anticlockwise direction and on 28 occasions it was spinning clockwise. It can be seen that the locust had no preferred orientation about its longitudinal axis and in fact it was continually rotating at varying speeds during the whole of this period. All the other locusts tested behaved in exactly the same way, none of them showing any inherent stability in this plane during complete darkness.

Locusts normally fly under conditions where the diffuse overhead illumination is greater than that from other sources. To reproduce these conditions experimentally locusts were placed in front of the wind tunnel in apparatus A (Fig. 2 A) with the median sagittal plane initially at various angles to the vertical. The light intensity of the illuminated roof of the box was 1·5 log. ft.-lamberts. Ten locusts were placed in the apparatus in turn in complete darkness and, if flying well after half an hour, the roof light was switched on and they were kept under observation until they were fatigued. At intervals of 1 min. their positions were recorded. As soon as the light was switched on each of the locusts rotated about its longitudinal axis until the dorsal surface was uppermost with the dorsal ommatidia receiving maximal illumination, and continued to fly steadily in this position with very little movement in the rolling plane. Fig. 6 A shows the orientation of one locust recorded 74 times over a period of 4.5 hr.

The box was then turned upside down so that the floor now provided the diffuse source of illumination and the experiment was repeated. In this case each of the locusts rotated about its longitudinal axis until the dorsal region of the eye was receiving maximal illumination and then remained flying steadily upside down (see Fig. 6B). The same effect was noted if the illumination came from either of the side walls, the locusts turned and flew with the dorsal ommatidia facing that wall. All the locusts flown in the following experiments showed this dorsal light reaction, turning so that the dorsal region of the compound eyes was maximally illuminated.

The reaction was examined in greater detail by using as a stimulus narrow beams of light whose diameter and intensity could be controlled. The diameter of the four parallel beams of apparatus B was adjusted to 3 mm. and their light intensity to 1·5 log. ft.-lamberts. The locust was inserted and, if flying well after the trial period, was stimulated by switching on one of the beams. As soon as the light came on the locust began to rotate about its longitudinal axis, continuing until the dorsal surface of the head was perpendicular to the beam when it ceased rotating and flew steadily. Once orientated in this manner the locust continued to fly in this position with little or no movement as long as the light was switched on. When the light was switched off the locust resumed the slow rolling characteristic of darkness, either immediately or after a few seconds. When another beam was switched on the locust immediately rotated about its axis from whatever position it had reached until the light from this beam fell on to the dorsal surface of the head, and remained flying in this position. A similar orientation was shown with respect to each of the four beams, although in two cases it meant that the locust was flying on its side and in one case that it was flying upside down.

Nine locusts were tested in turn, being stimulated by single beams in a random sequence at 1 min. intervals until they were fatigued. In all the tests made the locusts turned so that the angle of orientation was within ± 35° of the axis of the stimulating beam and in 64% of the tests it was within + 10° (see Fig. 7). The field of view of S. gregaria in the horizontal and transverse vertical planes through the centres of the eyes has been measured by Whittington (unpublished observations) who gives figures of 220° and 172° respectively. The field of binocular vision is 42° in the vertical plane and 64° in the horizontal. This means that in all cases the locust turned so that the beam lay in the uppermost quarter of the field of view of one eye and in approximately 85 % of the tests the locusts turned so that the beam fell on the dorsal ommatidia of both eyes. The locusts, with very few exceptions, always turned through the shortest distance necessary to achieve this orientation. Since the position of the locust relative to the beam when it was switched on differed in each case the angles through which the locust turned varied considerably. The average time taken to become orientated was 12 sec.

(a) The intensity of the four light beams could be varied by means of a rheostat between 2·0 and log. ft.-lamberts. Nine locusts were given 25 stimuli from each direction at intensities of 2·0, 1·5, 1·0, 0·70, 0·0, ⁠, and log. ft.-lamberts. Table 1 shows the range of orientations to a beam from the floor of the box at each intensity. The dorsal light reaction is shown in an equally marked manner at each of the intensities used. There seems to be no reduction in the accuracy of orientation or in the time taken to orientate as the intensity of the beam is reduced down to an intensity of log. ft.-lamberts which is of the same order of intensity as that of the night sky at Buraiman, Saudi Arabia, at the end of twilight (Roffey, 1963). If two beams having differing intensities are switched on simultaneously the locusts orientate to the beam of higher intensity. Intensity differences between two test beams of 1·0 and 0·5 log. ft.-lamberts were reduced gradually and locusts were found to be able to discriminate between an intensity difference of the order of 0·3 log. ft.-lamberts which was the smallest intensity difference it was practicable to obtain, (b) Beams of diameter 7·5, 1·5, and 0·5 mm. were used to stimulate five locusts. Locusts stimulated by a beam of diameter 7·5 mm. rotated until the dorsal ommatidia were receiving maximal illumination. The range of recorded angles of orientation was slightly greater (±40°) than that recorded with a beam of diameter 3 mm., the diameter used in the events described àbove, but not so great as that recorded with diffuse overhead illumination (see Fig. 8 A). Fig. 8 B shows the response of a locust to 29 stimuli from the right-hand side by a beam 1·5 mm. in diameter. The locust always turns so that the dorsal ommatidia are maximally illuminated but does not appear able to hold itself steadily in this position, rolling a short way to one side through approx. 45°, then re-orientating and then rolling again. When the smallest beam was used, 0·5 mm. in diameter, the locusts always turned so that the dorsal ommatidia were maximally stimulated but again did not seem able to remain in this position and there was constant rotation both clockwise and anticlockwise followed by recovery and further rotation. To a certain extent increasing the light intensity improved orientation to the narrower beams. At an intensity of 2·0 log. ft.-lamberts, the maximum obtainable in this apparatus, the locusts’ performance when stimulated with a 1·5 mm. beam was improved (see Fig. 8C), but locusts were still not able to fly steadily beneath a 0·5 mm. beam. It would seem that a certain number of ommatidia must be stimulated for the response to operate efficiently.

Nine locusts were each tested with the compound eyes and the ocelli intact, with the ocelli occluded by a mixture of blackboard paint, wax and varnish, and finally with the ocelli thoroughly cleaned and the compound eyes occluded. Each locust was given 30 stimuli from one direction at min. intervals in each of the three tests. Intact locusts orientated themselves so that the light fell on the dorsal surface of the head as described earlier. Locusts with the ocelli covered responded in a like manner when the light was switched on but did not on the whole align the median sagittal plane so closely with the long axis of the beam, there being a greater range in the angles of orientation recorded (± 60° as compared to + 35°). In addition the average time taken for orientation was much longer, being 1 min. 40 sec. Locusts with the compound eyes occluded did not orientate themselves in any way to the light but continued the slow interrupted rolling characteristic of complete darkness. Fig. 9 A-C show typical responses made by one of the locusts in the three test situado ns. It appears that the compound eyes are the sense organs controlling the dorsal light reaction and that the ocelli, whilst themselves unable to control the reaction, in some way enhance the performance of the compound eyes.

The left compound eye of ten locusts was occluded and each locust was stimulated 30 times by each beam in a random sequence. When the light was switched on the locusts rotated about the longitudinal axis until the dorsal ommatidia of the right, intact eye were maximally stimulated and then flew in this position. The range of angles of orientation recorded was +25° to — 80° about the axis of the beam. A further nine locusts were stimulated with the right compound eye occluded and the left one intact, and these locusts rotated until the dorsal ommatidia of the left eye were maximally stimulated and giving a range of angles of orientation — 25° to + 80°. The time taken for orientation was approximately the same as that taken when both eyes were intact but the alignment with the axis of the beam was not quite so precise, the range of orientation positions adopted being greater with one eye occluded. In addition, the results (see Fig. 10A and B ) show a heavy bias in the increase of spread to the side of the occluded eye which suggests that the normal mechanism involves correction initiated by both eyes. The effect of partial occlusion of both eyes was examined by blacking out the upper third of each compound eye of five locusts, obscuring approximately 55° of the field of view of each eye in the vertical plane. Each locust was stimulated 25 times by each beam at intervals of 1 min. in apparatus B. When a beam was switched on the locust turned so that the uppermost intact region of one of the two eyes was maximally illuminated and remained flying in this position. The direction in which the locust turned depended upon the position it occupied when the light was switched on. If the light fell within the field of vision of one of the two eyes the locust rotated so that the light fell on the uppermost region of that eye unless the uppermost region happened to be facing the light in which case no movement occurred. If the blacked-out region faced the light the locust continued the slow rolling characteristic of darkness until the light fell on the uppermost intact region of one of the two eyes. Fig. 11A and B show the responses of one of the five locusts to 25 stimuli from the left (— 90°) and from below (180°). The locust turned so that the range of angles of orientation lay between — 30° and — 75° or between +30° and + 75° with respect to the axis of each beam, positions in which the beam falls on the uppermost ommatidia of the intact area of one or the other of the two eyes.

The experiment described above was repeated with five locusts having the upper two-thirds of both compound eyes occluded. When the beam was switched on these locusts continued the slow rolling characteristic of darkness until the beam came into the field of vision of the remaining part of one of the eyes. The locusts then turned and flew so that the light fell on the uppermost ommatidia of this remaining region.

The exposed region of both eyes was reduced still further in some cases so that only a small ventral area of each eye remained uncovered. Under these conditions the locusts rotated slowly as if in darkness until the beam came into the field of vision of one of the remaining regions, when they stopped turning and remained flying fairly steadily with some occasional rotations in either direction.

If the behaviour of the locust is observed closely it can be seen that when the light is switched on the locust first begins to rotate the head and then the wings are moved in such a way that the body is turned after the head, lagging a little way behind it. The head ceased to rotate when the dorsal ommatidia were maximally stimulated and the body ceased to move when it was in line with the head. As mentioned earlier, Mittlestaedt (1950) showed the presence of proprioceptive organs in the neck region of the dragonfly, namely, four groups of prothoracic hair plates which registered relative head and body movement. In view of this work the neck region of S. gregaria was examined and two possible proprioceptor mechanisms which could be involved in maintaining relative positions of head and body were found (Goodman, 1959). A hair plate was found on the first cervical sclerite which, with the second cervical sclerite, forms the articulation of the head with the body. The tip of the sclerite bears an oval swelling at the point where it forms a ball-and-socket joint with the post occipital condyle of the head (see Fig. 12). The swelling bears numerous fine hair receptors; interspersed between them are circular, non-orientated campaniform sensilla and a few small tactile hair sensilla. In addition there are a few long-haired sensilla (Haskell, 1960). Several types of head movement, in particular rotation of the head relative to the long axis of the body, deflect the hair receptors on the cervical hair plates, and in addition, some of the hairs that underhang the anterior edge of the prothorax (see Fig. 12). The hair plates are innervated by a branch of the second pair of nerves from the first thoracic ganglion and a second branch of this nerve runs up to the anterior edge of the prothorax. From their position it would seem that the hair plates should be equally stimulated when the head is at rest and differentially stimulated when it is moving, whilst the prothoracic hairs will be stimulated only whilst the head is moving. Thus the locust should have a dynamic sense for head rotations similar to that of the dragonfly.

Eleven intact locusts were given 25 stimuli with a beam of light from each of the four directions at 1 min. intervals. Their cervical hair plates were then destroyed either by cautery or by a needle and the experiments repeated. The prothoracic hair fringe was then destroyed by cautery or by cutting off the rim of the prothorax and the locusts were stimulated again.

When the cervical hair plates are removed the locusts still turn so that the dorsal ommatidia are maximally illuminated but they are no longer able easily to remain in this position. The head turns as before but the body is slow in turning after it and when the head remains still in the normal position of orientation the locust does not appear to be able to hold the body in line with it. The body rolls from side to side dragging the head after it if the excursion is great enough. The head then starts to turn again to the position of orientation, followed again by the body and the whole process is repeated. There is constant movement about the beam so that the angle of orientation is continuously varying between approximately +60° and — 60° (see Fig. 13 B). Occasionally the body rolls so far that a complete revolution is made.

When the prothoracic hair plates were destroyed as well the locusts were completely unable to hold the body in line with the head. Although they responded in the usual manner to the onset of illumination they were unable to remain orientated but rolled continuously in varying degrees about the longitudinal axis as the head turned towards the normal position of orientation and the body stopped short of the head or went beyond it (see Fig. 13C ).

Ten mounted locusts were placed with the centre of the head in line with the centre of the horizontally placed artificial horizon described earlier (apparatus C, Fig. 3), and with the median sagittal plane at varying angles to the horizon. Their positions were noted after 1 min. Eight of the locusts turned from their initial position so that the median sagittal plane was at approximately 90° to the horizon, i.e. so that the horizon was horizontal in their field of view. Table 2 shows a typical response given by one of the locusts. They showed no preferential orientation to either the upper or lower half of the visual field when these were equally bright but occasionally rotated through 180°. If, however, one of the two was made less bright with grey paint then the locusts turned so that the horizon was horizontal and the brighter half uppermost in their visual field and no rotation occurred. This behaviour persisted even when the overall illumination of the hemisphere was low, of the order of log. ft.-lamberts, and when the difference of intensity between the two halves of the visual field was small (⁠ log. ft.-lamberts). As a control the experiment was repeated with a plain white hemisphere lacking a horizon. In this case the locusts remained approximately in the positions in which they had been inserted. Occasionally some of them would roll slowly for a while.

The horizon was then set horizontally with a white upper sector and a grey lower sector and the locusts were inserted with the median sagittal plane at 90° to the horizon. It was slowly tilted through 10° in a clockwise direction by hand and after J hr. returned to its former position. In successive tests it was turned through a total of 180° at 10° intervals and the test was then repeated with the horizon moving in an anticlockwise direction. The locusts began to move their heads in the same direction as the horizon as soon as it moved and then turned the body after the head. The angle through which the locust rotated was approximately equal to the angle of rotation of the horizon (see Table 3). Rotation ceased when the horizon once again lay approximately horizontally across the locust’s visual field and the locust remained flying in this position. When the horizon was returned to the horizontal the locust followed.

A number of locusts were stimulated by simultaneously tilting the artificial horizon and switching on a small light in order to examine the relationship between the horizon reaction and the dorsal light reaction. The horizon was tilted in an anticlockwise direction whilst a small light source (a 2·5 W. torch bulb) was switched on 6 cm. away from the right-hand side of the locust’s head. The test was repeated several times with different speeds of rotation of the horizon, with the intensity of the small light source varied and with the overall illumination of the white hemisphere of apparatus C varied. It was not possible to compare directly the strength of the stimulus given by a dark object rotating in the visual field against a white background with that given by switching on a stationary light source so that detailed quantitative results were not obtained from these tests. Observations of the locusts’ behaviour suggest that the response given depends on the relative strength of the two stimuli. When the intensity of the small light source was markedly above that of the white hemisphere (light source 2·0, hemisphere 1·25 log. ft.-lamberts) the locusts rotated until the light from this source fell on to the dorsal ommatidia and remained flying in this position, appearing to ignore the movement and subsequent position of the artificial horizon. When the intensity of the light source was low compared with that of the hemisphere (fight source 0·5, hemisphere 1·5 log. ft.-lamberts) the locusts rotated in an anticlockwise direction with the artificial horizon and remained flying with it aligned horizontally in their visual field. As the intensity of the light source approached that of the hemisphere the locusts’ behaviour became less precise. They showed a tendency to turn with the horizon whilst it was moving, to remain aligned with it for some moments when it stopped and then to rotate clockwise until the dorsal ommatidia were maximally stimulated by the light source. In some cases the locusts alternated between the two positions. Increasing the intensity of the light source increased the likelihood of the locust turning first towards the light source but this could be counteracted to a certain extent by increasing the rate of rotation of the horizon. The relationship of these two reactions requires further study under conditions where quantitative results can usefully be obtained.

In order to study the horizon response further electrical recordings from the ventral nerve cord of locusts were made when the latter were stimulated by rotation of the horizon. Locusts were mounted as described earlier in apparatus D, Fig. 4. The horizon was rotated from a horizontal position through 90° at a speed of 25 °/sec. in a clockwise direction. It was returned to the horizontal after an interval of 1 o min. The stimulus was then repeated reducing the angle through which the horizon was rotated by 5° each time until finally it only moved through 5°. The complete test was repeated with other locusts with the horizon rotating anticlockwise, with the illuminated sector set at different intensities and with various speeds of rotation.

A marked discharge was recorded from several fibres in both the ventral connectives as soon as the horizon started to rotate. This discharge persisted as long as the horizon was moving. Some fibres showed signs of adaptation whilst others did not (see Figs. 14 and 15). Discharge ceased in the majority of fibres as soon as the horizon stopped moving. One or two fibres continued to discharge at a low rate, 15−20 impulses/sec. as long as the horizon was tilted, with little or no signs of adaptation. A similar marked discharge occurred when the horizon was restored to the horizontal, ceasing when movement stopped except for the low-frequency discharge which, however, only persisted for 1 or 2 sec. after movement ceased (see Fig. 15). The locusts were extremely sensitive to movement of the horizon ; tilting through only 5°at low speeds produced a marked response (see Fig. 14).

A comparison of the discharge in the two connectives, whilst it cannot of course be quantitative, showed a consistent difference between them, there being greater activity in the left-hand connective when the horizon was rotated clockwise and in the right-hand connective when it was rotated anticlockwise (see Fig. 16). This means that there is a larger discharge on the side on which the larger amplitude of wing beat would be necessary to turn the locust after the horizon.

In a few cases recordings were made after the locust’s head had been fastened to the prothorax by wax so that it could no longer turn after the horizon, under these conditions, much smaller, approximately similar sized responses were recorded from both connectives whilst the horizon was moving, but no discharge was recorded after the horizon stopped moving (see Fig. 17). In one instance recordings were made from a locust whose head was free to move but whose cervical hair plates were destroyed; an unequal discharge was recorded from the two connectives whilst the horizon was moving which ceased completely when rotation stopped (see Fig. 18).

At least two optomotor reactions, mediated initially by the compound eyes, appear to play some part in regulating stability in the rolling plane in the desert locust. Locusts flying in complete darkness or locusts with only the ocelli intact lack stability in this plane under experimental conditions and when tested over fairly long distances (between 50 and 100 m.) in the field. With the compound eyes intact, however, a marked dorsal light reaction and a reaction to the position of the horizon in the visual field can operate to promote stability.

In the absence of a well-defined horizon the locust rotates about its longitudinal axis until the dorsal ommatidia are maximally illuminated and remains flying steadily in this position. The locust orientates in the same way if one compound eye is covered so that the response cannot be a phototropic one involving a balance of stimulation between the two eyes. It does not depend upon fixing the light source on one particular part of the eye since the locust always turns so that the uppermost region of the field is maximally stimulated however restricted the visual field may be. It would appear rather that there is a spatial representation of the visual area in the central nervous system and that the locust always turns so that the maximal illumination falls on the uppermost region of this area.

It is clear from the observed behaviour of locusts that a simple dorsal light reaction cannot be the sole factor controlling stability in the rolling plane since, for example, orientation is not regulated by the sun’s position in the sky. As blinding removes stability in this plane some other visual stimulus must be operating. Locusts have been shown to be very sensitive to the rotation of an artificial horizon, aligning themselves so that the horizon lies horizontally across their visual field. If the visual field presented consists of two equally illuminated areas separated by a black boundary the locust will align itself so that the boundary is horizontal in its visual field, showing no preferred orientation to either of the illuminated areas. If one half of the visual field is brighter than the other the locust orientates itself with the boundary horizontal and the brighter area uppermost in its visual field. These two optomotor reactions may very well reinforce each other in nature in the same way, the more general dorsal light reaction ensuring that the brighter half of the visual field, which under normal circumstances will be the sky, is uppermost, and acting as a coarse control, whilst the more specific reaction to the horizon serves as a means of fine adjustment of position.

Swarms of S. gregaria generally fly by day and settle in the late afternoon (Waloff & Rainey, 1951), but locusts of the phase solitaria appear to have a different behaviour pattern and to fly chiefly at night (Waloff, 1963 ; Roffey, 1963). There is some evidence that sustained night flying is associated particularly with moonlight nights or nights when there is some light in the sky (Roffey, 1963). Is it possible for night-flying locusts to use optomotor reactions to control stability or have they some other means of control? Field experiments in which the beam of an Aldis lamp has been shone upwards on moonless nights have shown that solitaria individuals passing near the beam alter their orientation about the rolling axis and show signs of a dorsal light reaction, in some cases diving down the beam of light (Rainey & Ashall, 1953 ; Roffey, 1963). Their stability at night therefore can be affected by a visual stimulus. Gregaria individuals used in the present experiments were able to orientate precisely in experiments on the dorsal light reaction down to test intensities comparable to those measured in the night sky at Buraiman, Saudi Arabia, at the end of twilight, of the order of log. ft.-lamberts (Roffey, 1963) and to detect the differences of intensity between ‘sky ‘and ‘land ‘of a similar order in the artificial horizon apparatus. It has been suggested that the difference in the flight-behaviour pattern between individuals of the gregaria and solitaria phases is due to a difference in the sensitivity of their visual systems, solitaria individuals being adapted for vision in low intensities (Roffey, 1963). If this is so then they should be even more sensitive to the difference of intensity between sky and land than gregaria individuals and there seems no reason why stability in the rolling plane should not be controlled by these two optomotor reactions even in the night sky.

Both the responses are mediated by the compound eyes alone but, although the ocelli are not capable of mediating the response themselves, the performance of the compound eyes appears to be impaired when the ocelli are occluded; the locusts do not orientate so quickly or so accurately. If occlusion of the ocelli resulted in an overall decrease in sensitivity of the visual system this should not necessarily interfere with the locust’s ability to orientate accurately. Possibly the effects of occlusion of the ocelli are unequally distributed in the central representation of the visual area so that the spatial distribution of intensity appears to be modified. The orientation obtained with the ocelli occluded would thus represent the locust’s attempt to orientate to this modified representation. Cornwell (1955) has noted a somewhat similar effect in Locusta nymphs where the efficiency of certain optomotor reactions mediated by the compound eyes was greater when the ocelli were intact than when they were occluded. The nymphs behaved as if ocellar occlusion caused a decrease in sensitivity of the anterior part of the compound eye, an effect which might be produced if occlusion of the ocelli caused distortion of central visual representation as suggested above.

The visual stimuli control rolling indirectly through head movements, the head always being positioned first and then the body aligned with it by differential wing movements. Unlike the system in the dragonfly however, in the locust the head does not act as an inertial stabilizer in darkness since it is tightly coupled to the body. Relative head and body movements appear to be registered by the cervical hair plates and certain of the prothoracic hairs, since ability of the locust to remain orientated to the light is considerably reduced if the hair plates are destroyed, and made impossible if both hair plates and prothoracic hairs are missing. Haskell (1959, 1960) has recorded the electrical responses of the prothoracic hairs and the receptors of the cervical hair plates in response to bending and shown that they would be capable of fulfilling the proprioceptive role assigned to them. A sustained deflexion of the relevant prothoracic hairs in any direction produced a high initial discharge rate, adapting rapidly to a much lower steady level. The initial frequency was related to both the speed and the magnitude of the deflexion. Recordings from the cervical hair plate nerve showed responses from two of the several receptors found there; the main hair sensilla on bending showed fairly slow, practically non-adapting discharges, with frequency generally related to the degree of bending, while pressure on the hair plate resulted in responses with a high initial discharge rate showing incomplete adaptation which Haskell believes originates from the campaniform sensilla. The hair plates are partially stimulated in the resting position and movement will change their relative stimulation. They will thus be able to register the equilibrium position, and the prothoracic hairs, which are only stimulated on movement, will act as motion proprioceptors.

Electrical recordings taken from the ventral nerve cord as the horizon is rotated suggest that the proprioceptors are in fact functioning in this manner. If head movement is prevented only a small discharge of approximately equal size in both connectives is recorded during the actual movement of the horizon; this discharge is presumably the result of direct visual stimulation. When the head is free to turn, however, there is a much more marked, though unequal, discharge in the connectives. It seems likely that this difference in discharge in the two connectives is due to the differential stimulus of the proprioceptors and will presumably be reflected in differences in the motor nerve input to the relevant wing muscles, although this has not yet been demonstrated.

Normally, a few fibres continue to discharge in both connectives as long as the horizon remains tilted. In the experiment described above where the head was free to follow the horizon but the body was immobilized preventing alignment with the head, when the horizon was restored to the horizontal, followed by the head, this residual discharge very quickly died away. If the prothoracic hairs are only stimulated when the head is moving, this discharge, which persists when head and body are not aligned, must result from unequal stimulation of the cervical hair plates and indeed this is confirmed, since, if they are destroyed, the discharge in the connectives does not persist after the horizon stops moving.

1. Locusts given freedom of movement in the rolling plane show complete lack of stability when flown in darkness, continually rotating about their longitudinal axis.

2. Stability in this plane appears to be controlled by two optomotor reactions, a dorsal light reaction and a reaction to the position of the horizon in the visual field.

3. Locusts behave as if there is a spatial representation of the visual area in the central nervous system and always turn so that the horizon is horizontal and the brighter half of the visual field uppermost in this representation.

4. The optomotor reactions control stability in the rolling plane indirectly through head movements, the head being orientated first and the body aligned with it by differential wing movements.

5. Relative head and body movements appear to be registered by two sets of proprioceptors, hair plates borne on the first cervical sclerites where they articulate with the head and a row of tactile hairs on the under edge of the prothorax.

6. The reactions described are operative in fight intensities down to o-1 ft.-lamberts, approximately comparable to tropical twilight conditions.

This work was carried out while I held a Senior Research Award from the AntiLocust Centre, London. I am indebted to the Director for his advice and encouragement.