ABSTRACT
Although many of the structural adaptations associated with the locomotion of aquatic insects are well known, there are few observations on the details of their propulsive mechanisms. In general, these may be subdivided into those which effect swimming by undulatory movements of the body (Kalmus, 1936; Gray, 1953), by oar-like movements of the legs or wings (e.g. Polynema, Hymenoptera), or by jet propulsion methods involving either the use of a surface active secretion (Stenus, Staphylinidae) or the active expulsion of water from the insect. Of these the use of legs as oars is the most common among adults. In this paper cinemato-graphic analyses are given of two beetles which use this method and of the abdominal and leg movements involved in the jet propulsion of a dragonfly nymph. In addition, some observations are described on the effect of amputating legs on the swimming of Dytiscus. These differ significantly from the observations of Bethe & Woitas (1930) and support conclusions based on a study of cockroach movements (Hughes, 1957).
MATERIALS AND METHODS
The species used were the giant water-beetles, Dytiscus marginalis and Hydrophilus pic eus, and late instar nymphs of Anax imperator. Swimming of the beetles was observed in a large sink containing water about 6 in. deep. Films were taken while the insects swam freely above a glass plate marked with a grid and placed in the field of the camera. Similar films were taken of the dragonfly nymphs but the results were not satisfactory for all purposes. Close-up films, either in side or in dorsal view, of nymphs freely suspended from the wing pads proved of much greater value in revealing the sequence of events in the swimming cycle. Spots of enamel paint placed on the abdominal segments and legs helped in plotting their position on successive frames. Sinclair 35 mm. and Zeiss Movikon 16 mm. cameras were used at speeds of 24-30 frames/sec. for the beetles, but much faster speeds up to 60 frames/sec.) and briefer exposures were necessary to‘fix’ the dragonfly movements. General lighting was provided by two Photofloods and in addition carefully placed Point-o-Lites were used for the close-up shots.
Additional information about the events during swimming of Anax was obtained from films and simultaneous recordings on an oscilloscope of the time course of pressure changes within the branchial chamber and/or the forces on the animal resulting from the jet. As these events are extremely rapid great care was taken in the selection of recording apparatus and in the method of synchronizing films and oscillograph records. The wing pads of the insect were attached by a small paper clip to a short arm soldered to the anode of an RCA 5734 mechanotransducer valve. Although the natural frequency of the latter is 12 kcyc./sec. it was reduced considerably by these attachments. After suitable reductions in the length of the arm it was raised to 120 cyc./sec. which is sufficiently high to reproduce faithfully the true time course of the forces generated by the jet. Similarly, by using a Hansen condenser manometer (Hansen, 1949) with a natural frequency of nearly 100 cyc./sec., the time course of the pressure changes was accurately recorded even when fine (0·5 mm. outside diameter) hypodermic tubing was used. The following method of synchronization was suggested by Dr R. H. J. Brown, to whom I am grateful for advice on this and on other matters of technique. The point where a wire on a time marker, rotating times/sec., touched a mercury drop was brought into focus in the camera field by a lens or mirror; this contact completed a circuit which modulated the intensity of the beams on the oscilloscope. It was thus possible not only to see which frame was related to the signal on the oscilloscope trace, but also to decide whether the frame was a fraction of a second before or after the signal by observing the actual position of the rotating contact on that frame.
RESULTS
(A) The swimming of Dytiscus and Hydrophilus
(1) Morphological
Dytiscus and Hydrophilus show many structural adaptations to their aquatic habitat some of which display striking convergences. On the whole, Dytiscus is better adapted to aquatic conditions than Hydrophilus which is correspondingly more efficient on land. The body is keel-shaped in both genera and offers little resistance to water flow, but it is in the legs that the most striking adaptations are found. The femur, tibia and tarsus of the meso- and metathoracic legs are all flattened. The hindlegs of swimming beetles are shorter relative to the body length than in terrestrial forms and the proportions of the segments are different. In Dytiscidae the tarsus is about twice the length of the tibia, whereas in terrestrial beetles they are nearly equal (Roth, 1909). Hairs are found on the tarsal joints of Hydrophilus and on both the tibia and tarsus of Dytiscus. In both genera these tarsal joints overlap one another on the surface which is anterior during retraction and thus increase the rigidity of the structure. They have a mechanism whereby the tarsi of the middle and hind pairs of legs can rotate, enabling them to ‘feather’ during the forward stroke. In Dytiscus, the articulation of the proximal segment with the tibia is similar to a ball-and-socket and facilitates a 100° rotation of the tarsus. This joint is fairly normal in Hydrophilus, but the articulation between the two proximal joints of the tarsus is very oblique and the rotation principally takes place about its axis. The other segments of the leg are also modified in both genera. Promotor-remotor movements of the coxae are completely excluded in Dytiscus as the coxae are fused to the sternum, while in Hydrophilus they are restricted by the coxae being sunk into spaces in the sternum. This adaptation probably serves to lessen the drag of the insects in the water.
The coxo-trochanteral joints are dicondylic and determine the movement of the femur very rigidly. This movement is not in one plane, because of the shape of the trochanter, and has the result that the tip of the femur moves parallel to the ventral body surface throughout its stroke. The femur is flattened and curved so that it is almost flush with this surface throughout the arc of its movement, adaptations which will again decrease the resistance to forward movement.
(2) Cycle of hindleg movements
A series of diagrams (Fig. 1) showing the position of the hindlegs throughout the cycle has been constructed from the photographs of the swimming insect. At the completion of the effective stroke (ii), the hindleg of Dytiscus appears narrow in the photographs because the flattened surfaces of the tibia (a) and tarsus (b) are pressing against the water and the hairs are spread out dorsally and ventrally to offer their maximum resistance. Protraction commences with flexion of the coxotrochanteral and femoro-tibial joints (iii). The tarsus rotates 100° (b to d) so that the edge which was anterior during retraction now becomes dorsal and the outline becomes blurred as the hairs are visible (iv). Roth (1909) described this tarsal rotation but in the opposite direction. His account was shown to be wrong by filming insects with the tarsi painted white on one surface only. The anterior side during retraction becomes dorsal as the leg is drawn forwards and paint on its surface is only visible during protraction. The femur continues to protract and it forms the leading edge of the limb behind which trail the flexed tibia and tarsus (v). The tibia now begins to extend and takes up its fully protracted position (vi, vii). As Amans (1888) described, the rotation of the tibia (a to c) which occurs during the cycle is about 45°, but in the opposite direction to that of the tarsus. The femur is beginning to retract at this stage and as the tarsus rotates its tip passes dorsally to the rest of the limb. As the femur has already retracted by the time the tarsus is presenting its maximum surface to the water, it is probable that its extension and rotation are entirely passive. This may be partly true of the femoro-tibial joint also. The musculature agrees with this suggestion for, as Bauer (1910) pointed out, the extensor tibiae is a much smaller muscle than its antagonist and there are no extensors of the tarsus. During retraction the tibia rotates progressively until it is inclined as shown in (a) at the end of the active stroke. The whole limb acts as an almost straight oar (i) and the tarsi are curved by the water pressure. The films of insects with painted tarsi showed that during retraction the posterior surface of the tarsus is inclined slightly upwards which will produce a force tending to move the insect downwards. Towards the end of the stroke, the tibia flexes while the tarsus is still exerting pressure (iii) and protraction begins once more.
This description indicates the principle underlying the ‘rowing’ movements of aquatic insects. It is highly developed in Dytiscus and the individual movements of the swimming legs of Hydrophilus conform to the same general plan. A comparison of the path followed by these legs illustrates the greater efficiency of Dytiscus in its swimming movements (Fig. 2 A). During protraction of the legs in Hydrophilus (Fig. 2B) there is relatively little flexion of the femoro-tibial joint and the course followed by the limb tip during protraction is consequently further away from the body and offers a greater resistance to forward movement of the body. On the other hand, the middle legs of Hydrophilus exert a more powerful thrust against the water than does this pair of legs in Dytiscus. In both insects the highly-developed extensor trochanteris is the most important muscle in propelling the body forwards.
These movements of the legs during the swimming of Dytiscus are obviously quite different from those employed by the insect when walking, but in Hydrophilus the differences are not so well marked. The greater activity of the flexor tibiae muscles and the increased promotor-remotor movements of the coxae, result in the path followed by the hindlegs being much closer to the body when the insect walks.
(3) The rhythm of the swimming movements
In Dytiscus, the rhythm in which the legs are retracted is completely different from that described for the walking insect (Hughes, 1952). The two legs of a limb pair retract simultaneously, ensuring that the body is subjected to only very slight turning moments (Fig. 3 A). When swimming along a path which is not perfectly straight the timing of a pair of legs may be altered. Fig. 2 A was plotted from such an insect and it will be noticed that the right legs are retracted a little in advance of the left legs and also with a greater amplitude. Both these features will tend to correct the slight turn to the right which was made in the previous stroke. More rapid turning takes place when one of the hindlegs is held outstretched in the protracted position, while the contralateral leg retracts actively (Fig. 4B). The extended hindleg and also the middle leg on that side may make slight ‘back-paddling’ movements. The duration of protraction and retraction are usually equal in this insect, but sometimes protraction is shorter as in Fig. 4A. The difference is slightly exaggerated here, however, because the legs appear to be in a protracted position for a longer period on account of the inclination of the body. Previous accounts of the rhythm hardly mention the middle pair of legs and suggest that they perform only occasional movements during turning and to preserve the balance. This is true when the insect is undisturbed and swims slowly but in more rapid swimming, these legs play a regular part and retract together but in opposite phase to the hind pair of legs (Figs. 3 A, 4). One function of the middle pair is to stabilize the body during the recovery stroke of the hindlegs. Their retractor movements are directed more ventrally than those of the hind pair, and this tends to raise slightly the front part of the body. For instance, in Fig. 4A the apparent changes in length of the body are plotted beneath the leg movements. A decrease in length is due to the abdomen being raised relative to the head by the buoyancy of the air store. As the middle legs retract, the body becomes more level in the water and its apparent length increases. It is noticeable that the length starts to decrease before the hindlegs are completely retracted, suggesting that perhaps they do not play such a great part in regulating the horizontal position of the body. The middle legs also function to maintain a uniform speed of forward movement, since when they are amputated progression is somewhat jerky, especially when the insect is excited and the body does not keep on too even a keel.
The rhythm of leg movements in Hydrophilus differs from that of the Dytiscidae in that the two legs of a segment are in opposite phase. The middle and hind pairs are both employed and diagonal legs retract at exactly the same instant (Fig. 3 B). The fully retracted middle legs lie beneath the body so that it is difficult to determine whether this is also true at the beginning of protraction. The duration of protraction of the hindlegs tends to be longer than retraction by about a fifth of a cycle. The position of the head of this insect has also been plotted in Fig. 3 C, D and demonstrates that the path followed by Hydrophilus is much straighter than has been generally supposed. Deviations of the head to the right and left are associated with retraction of the hindleg on the opposite side. One of the reasons why these deviations are not so great is because the legs move simultaneously on the two sides. Synchronous retraction of diagonal legs is of further significance in that it distinguishes the swimming from the walking rhythm where there is always a slight delay between the instant at which R2 and L3 or L2 and R3 are retracted (Hughes, 1952).
(4) The effects of limb removal in Dytiscus
Bethe & Woitas (1930) described changes in the swimming movements which occur when one or more legs are amputated. The most striking changes take place when a single hindleg is removed. In such an insect they describe how the middle leg on the same side, previously inactive, became important in swimming and retracted simultaneously with the intact hindleg. However, analysis of films (Figs. 5, 6) taken in the present investigations has failed to confirm this and other details of their description. It can be readily observed that when the left hindleg (L3) is amputated, the suddenly alarmed insect tends to circle to the left owing to the action of the right hindleg, but it soon swims along a straighter path because of the following modifications in leg movements:
The intact hindleg (R3) does not retract so far back as in the normal insect and during protraction it is not feathered so closely to the body nor does it move so far forwards. The movements of this leg do not have such a regular rhythm and give the impression of being constantly regulated. All these modifications tend to decrease the tendency for this leg to rotate the insect anticlockwise.
The left middle leg (L2) continues in the normal rhythm so that it retracts as soon as R3 has completed its effective stroke. The amplitude of its movements is increased both anteriorly and posteriorly.
Retraction is delayed in which R2 therefore remains outstretched while L2 retracts and, together with the protracting hindleg, acts as a fulcrum about which the body pivots.
Modifications (ii) and (iii) both increase the turning effect of Lt and it has been noted already that similar changes occur in the normal insect when it turns (Figs. 2 A, 4B). The net effect of the changes in rhythm and posture of the legs is that progression towards a given point is accompanied by oscillations which are due to an alternation of the turning moments produced by R3 and L2. This is clearly shown by tracing the path of the head or centre of gravity (Fig. 5). The overall path is fairly straight, which would not be the case if these modifications had not taken place. Bethe & Woitas (1930) did not plot the movements of the legs from their photographs, which in fact clearly show that the intact hindleg and contralateral middle leg do not retract simultaneously. Most later illustrations of this phenomenon (e.g. Wigglesworth, 1953) reproduce Bethe’s diagram and not the actual photographs.
Observations on the effects of amputating other legs differed slightly from those of Bethe & Woitas but will not be described here. Changes in posture of the remaining legs occur in addition to their taking over of the swimming function. One very striking example of this occurs if both the hind- and middle legs of one side (L2 and L3) are removed. As was to be expected the insect tended to move to the left, again especially when it was startled. But it was able to swim along a relatively straight path (Fig. 6 C) surprisingly well when left undisturbed. In doing this the intact middle and hindleg both altered their normal movements a great deal although the rhythm remained relatively unaltered. The hindleg was held outstretched and scarcely feathered during protraction, which was longer than retraction. The middle leg, however, protracted very much farther forwards and during this phase was held away from the body. During retraction it moved ventrally and its active stroke was directed under the body and towards the left-hand side. This leg, therefore, moved in an anticlockwise direction (Fig. 6B) and tended to rotate the head towards the right. Progression was inevitably slow by this method which reminds one of the manoeuvres which can be made with a canoe paddle.
Several neurological mechanisms are involved in the production of these plasticity changes and while a complete analysis cannot be given some responses of intact Dytiscus are suggestive. It has already been noted how during a slightly ‘wavy’ course and when turning (Figs. 2A, 4B) the posture and rhythm of the legs are modified which is accentuated following amputation. Even more striking is the observation that when Dytiscus is held and moved through water in a circle of about 6 in. diameter, a characteristic compensatory response is invariably observed (Fig. 7). The outer hindleg is outstretched and shows little active retraction, but occasionally moves forwards in this stiff condition in a way very reminiscent of the single hind-leg of the insects described above. At the same time, the other hindleg retracts strongly and frequently, while the two middle legs show movements which would also produce rotation in the opposite direction to the imposed movement. The most important sensory structures involved in this response are the eyes, antennae, and receptors on the legs themselves. The response persists in red light, but is not so marked after covering the eyes with wax containing lampblack, or when a striped environment is rotated at the same speed as the insect. If the antennae are also removed the response is sometimes absent and if it occurs is not very definite. During forced rotation of an insect there will, of course, be an asymmetric stimulation of the legs on the two sides because of the centrifugal force as is shown by the positions assumed by the legs when a dead insect is rotated.
It has not been possible to prove conclusively that any one of these receptors is solely responsible for this response. It seems likely that all three are able to produce it by themselves. The eyes are certainly involved a great deal in the orientation of this insect (Zeiser, 1934). They are probably the most important receptors as was concluded by Tonner (1938) in his study of similar compensatory responses of dragonfly nymphs. The importance of differential stimulation of antennae has not been discussed previously in Dytiscus but it must certainly occur even in normal swimming. The antennae are held forwards at an angle to the axis of progression and when the path deviates slightly a difference in their degree of bending is apparent on the films. The basal joints are profusely supplied with sensory structures which include Johnston’s organ, a row of campaniform sensilla and several hair plates at the articulation. Both slowly and rapidly adapting responses to deflections of the antennae have been recorded from the antennal nerve. If a fine jet of water is directed at the antennae of a fixed Dytiscus from in front and to one side the contralateral hindleg usually kicks.
The above observations suggest that if the insect departs from a straight course the central nervous system will receive a changed pattern of sensory impulses from these sources and that the normal responses tend to restore the original orientation. As these responses have much in common with the alterations in motor activity which follow limb amputation, it is probable that these too are responses to the inevitable changes in peripheral inflow resulting from the tendency to circle. If insects with a single hindleg amputated are moved through the water in a straight path the remaining leg movements are almost normal. Correspondingly, the alterations in movements do not occur if the hindleg is removed from a blinded insect so that the insect circles for some time, particularly if the antennae are also removed. Insects which show almost rectilinear swimming after a hindleg has been amputated immediately circle when blinded. Some compensation does eventually take place in these insects, however, because of the proprioceptive inflow from the legs, but the majority of their movements are normal. Unilaterally blinded insects with a hindleg removed tend to circle, but may move fairly straight for short distances (Fig. 5 D). In this insect the intact hindleg moves normally and the two middle legs retract almost simultaneously (Fig. 5 B).
The supra- and suboesophageal ganglia must play a large part in the integration of the inputs from these different sources. Several lines of evidence suggest that they probably contain a mechanism which compares the inputs from the two sides of the body (cf. Kalmus, 1949) and controls the responses which occur when the insect departs from a given path. The latter need not necessarily be directly to or away from a given stimulus but can vary in its orientation. The factors controlling the ‘set’ of this mechanism may involve central as well as peripheral patterns of stimulation, but once the ‘set’ has been established changes in peripheral inflow will be analysed with respect to it. Such a mechanism is essential otherwise the compensatory responses would prevent the insect from turning. Unilateral blinding or extirpation of a supraoesophageal ganglion so alters the balance of the coordinating mechanism that persistent circling results.
Faivre’s (1857) observation that the direction of circling is away from the injured side has been substantiated by several workers and Baldus (1927) has established that alterations in the movements of the legs on both sides of the body are involved. His photographs show that they are strikingly similar to those reported above following unilateral amputation of legs or rotation of the insect. Circling away from the blinded or decerebrate side continued even when the middle and hindlegs were amputated on one side. If the legs removed were from the side opposite to that from which the supraoesophageal ganglion was extirpated, circling resulted from only slightly exaggerated movements of the remaining middle and hindlegs. If, however, the amputation and brain injury were on the same side the leg movements were very different. His description is almost identical to that given above when the two legs were amputated but there was no brain injury. Baldus mentions that he observed this activity but he holds quite the opposite view to Bethe & Woitas, i.e. that following limb amputation the remaining legs move quite normally and that these asymmetries occur only after unilateral brain injuries have also been performed.
It is evident that the brain has a large number of both homo-lateral and contralateral connexions with the lower motor centres, but their complexity makes any description at this stage not only inadequate but probably misleading. Nevertheless, the very striking similarities between the leg movements following the amputation of a single hindleg and those involved in the compensatory and circus movements strongly suggests the operation of a single co-ordination mechanism influenced by peripheral changes and is quite contrary to the plasticity theory of Bethe.
(B) The swimming of the dragonfly nymph, Anax imperator
Normally these predators creep extremely slowly among weeds and await their prey, but they can swim quite rapidly especially when disturbed. They usually swim when placed in a large volume of water particularly if there are few objects to which they can attach themselves. Nymphs also swim if the legs are free when they are suspended from their wing pads by a small paper clip fixed to a steel rod. Periods of swimming are often produced when the rod is struck or the insect touched with a brush.
Jet propulsion is achieved by an extremely rapid contraction of the abdominal musculature (Whedon, 1918; Tonner, 1936; Snodgrass, 1954) producing a pressure head within the enlarged hindgut which forms the branchial chamber. Water escapes as a fine jet at high velocity from the anal opening which is guarded by valves. The reaction from this jet propels the insect forwards at speeds which reach 30–50 cm./sec. during the first centimetre of propulsion. The frequency of the jets may be up to 3/sec. and they may continue for some time but in a fixed insect a pause usually occurs after ten to twenty of them. These features were conveniently recorded when the insect was fixed to the RCA 5734 transducer valve. The time course of the propulsive forces generated by the jet was also registered and as Figs. 9 and 11 show, a maximum value of 0·6 g. is reached within 0·03 sec. and the total duration is a little under 0·2 sec. The area beneath the curve is a measure of the impulse acting on the insect and is equal to the change of momentum which was about 52 dyne-sec. in an actual experiment. As the insect weighed 1 g. the expected velocity had the insect been free was 52 cm./sec. This figure, however, neglects the drag forces to which the free insect would be subjected, but, even so, it is in good agreement with speeds measured from films at the onset of swimming.
The precise sequence of muscular movements is difficult to establish as the total duration of the contraction is only about 0·1 sec. The most obvious movements are the longitudinal and dorso-ventral contractions of the abdomen, and the retraction of all six legs so that they lie alongside the body. When the insect is held stationary the legs come forwards again before the next jet (Figs. 8-10), but this does not occur normally because of an ‘Angelegtbleiben’ reflex (Tonner, 1935) elicited by water stimulating the antennae during motion. The two legs of a segment usually retract simultaneously, though cases have been observed where only the three legs on one side retracted. Some shots suggest that the order of retraction is hind, middle, fore, but the intervals separating these movements are extremely small, being at most 1/50 sec. (Fig. 8). One of the difficulties is to decide the exact phase at which retraction of a leg is associated with a given jet, but usually this is recognizable when a leg moves more rapidly backwards, regardless of its other activity.
The interval between the abdominal and leg movements is brief and in many cases they appear to be simultaneous. Most analyses suggest, however, that the hindlegs, at least, retract slightly before the abdomen shortens. Transducer records taken simultaneously tend to confirm this as there is a slight forward force synchronous with the leg movements and slightly preceding a rapid rise which coincides with shortening of the abdomen as the jet efflux occurs. The resulting propulsive forces on the insect cease when the abdomen is maximally shortened (Fig. 9). Flattening of the sternum seems to occur a little before the abdomen shortens, but it does not reach a maximum until after the abdomen has partly extended again (Figs. 9, 10). Water probably begins to be drawn in when the abdomen lengthens and continues to do so as the sternum is depressed. The most striking feature of the whole response is the longitudinal contraction, however, which may cause the abdomen to shorten by 7-10% of its resting length. Most of this appears to take place in the posterior segments but plots of the positions of individual segments (Fig. 10) shows that all segments are involved to some extent. The largest proportion (40 %) occurs in segments 6-8 and this is also the region where flattening of the sternum is greatest. These observations correlate with the internal anatomy as the branchial chamber is situated in these segments which are therefore the region where the maximum decrease of volume would be expected when water is expelled. The segments do not all contract at exactly the same instant as can be seen particularly during some of the slower contractions. The hind segments contract first and the telescoping spreads forwards very rapidly. The very first part of the whole response appears to be the closing together of the three valves which guard the external opening. It has not been possible to obtain good films of the valves, but direct observation certainly suggests that the aperture is completely closed just before the longitudinal contraction commences. It then appears to open slightly as water is expelled and later suddenly opens very widely and water is drawn in. The phase of valve closure is often associated with the slight apposition of the epiproct and two paraprocts. Opposite each of these terminal appendages is situated one of the three anal valves.
Analyses made so far of pressure changes in the chamber are in agreement with the account given above. Recordings taken when a fine needle was inserted through the anus into the branchial chamber show a rapid rise in pressure at the onset of the longitudinal contraction and may reach pressures of about 30 cm. water (Fig. 11). The pressure falls to that of the surrounding water (taken as zero) as the abdomen begins to elongate. This fall may be even more rapid than the rise in pressure especially during contraction at the end of a series. The opening wide of the valve is presumably associated with this rapid fall and is succeeded by a very small negative pressure as water is drawn in. The positive pressure occurs simultaneously with the impulse on the insect and has a similar time course. Some preliminary attempts to record pressures in the body cavity suggest that the time course of these follows that of the longitudinal movements of the abdomen and not of the jet pressure or impulse records. Several other features of the pressure changes recorded during swimming as well as those occurring in respiration are of interest and will be considered in a later paper.
DISCUSSION
The importance of swimming in the life of these three insects varies. It is probable that Dytiscus swims the most and dragonfly nymphs the least. Both are carnivores but catch their prey in different ways. Dytiscus hunts actively in search of food (Tinbergen, 1951) but dragonfly nymphs lurk among weeds and await their prey. Swimming in the latter is mainly an escape mechanism for which rapid acceleration is of great importance. This is achieved by the expulsion of water at high speeds through a small aperture. The pressure and impulse records suggest that the jet velocity is about 250 cm./sec. and the aperture less than 0·01 mm.2 in cross-section. The same volume of water is drawn in through the anal opening, but by increasing the cross-section the velocity of the water relative to the animal is low and the consequent change of momentum retarding progression will be considerably reduced. The use of the same aperture for both the expulsion and entry of water is probably unique among animals which use this method of propulsion. In Cephalopods, for instance, water is drawn in through the lateral openings of the mantle cavity. The extreme rapidity and synchronization of activities at the beginning of dragonfly flight was described as ‘die Gesamtreflex’ by von Uexkiill (1908) and the same term might well be applied to the swimming insect. Its co-ordination probably involves nerve fibres which run the whole length of at least the abdominal nerve cord. Some of these are relatively large (Hughes, 1953) and provide rapidly conducting pathways such as are frequently found in escape mechanisms.
Hydrophilus is a vegetarian and is the least modified both structurally and functionally. This is apparent in the movements of pairs of limbs which alternate in Hydrophilus but retract together in Dytiscus and Anax. It has been shown, however, that these alternating movements are not exactly the same as in walking because of differences in their nature and of the timing between the middle and hind pairs. It is of interest to note that in the larval stages of the beetles the opposite is true, for Dytiscus larvae ‘trample’ along with the two legs of a segment alternating whereas it is reported that those of Hydrophilus move together as the body performs dorsoventral undulatory movements.
All three insects are adapted to swim along a straight path and their resistance to progression is reduced. Deviations from this path are corrected by a number of compensatory responses of the animal. The observations reported on the effects of limb amputation on Dytiscus indicate the extent of these in the normal animal, how they operate under modified circumstances and within limits produce the same end result. Thus although a precise analysis of any so-called plasticity change is not possible at a neurophysiological level, sufficient is now known of the extensive proprioceptive mechanisms of terrestrial insects to support the view that it is due to the altered stimulation of the proprioceptive mechanisms which occur when the mechanical conditions change (Hughes, 1957). The present work, while not ignoring such mechanisms, emphasizes the importance of extero-receptors in the orientation of aquatic animals where gravity does not have such a controlling influence. Of these the eyes are the most important, but there is some evidence that the antennae are important in detecting water currents as they are known to do in dragonfly nymphs (Tonner, 1935). The considerable importance of impulses reaching the central nervous system is shown in all these observations. The fineness of control which they exercise on motor responses probably involving extensive feed-back mechanisms, results in them often being taken for granted. The striking changes which occur following amputation, etc., show up their existence and give us some insight into their complexity. However, it must not be assumed that the original patterns of motor activity are completely discarded either in Dytiscus or in Blatta, for in both insects there remains a tendency for these to show up under ‘stress’. Thus, if Dytiscus is dropped on water immediately after the amputation of a single hindleg, the remaining legs move rapidly and as a result the insect swims in small circles. But after a short time it slows up and makes a straighter path as described earlier. The same sequence is observed when the experiment is repeated on subsequent occasions even after a period of weeks.
SUMMARY
Movements of the legs during the swimming of adult Dytiscus marginalis and Hydrophilus piceus have been analysed from films. In both beetles the middle and hind pairs of legs play an active part and the front pair is held stationary beneath the thorax.
The two legs of a segment are retracted simultaneously in Dytiscus, the two limb pairs acting alternately. In Hydrophilus retraction of a hindleg is simultaneous with that of the contralateral middle leg. Movement is rectilinear in both insects, but the head of Hydrophilus rotates a little to the side opposite that on which the hindleg is retracting.
Following amputation of a single hindleg, Dytiscus swims along a relatively straight path as a result of modifications in the movements of the two middle legs and the remaining hindleg. Contrary to the description of Bethe & Woitas, the hindleg continues to alternate with the contralateral middle leg, the only change in rhythm being a delay in the retraction of the other middle leg. Some of the changes in action of these legs are similar to those used in turning of the normal insect and those which produce circling in unilaterally blinded insects.
Forced rotation of Dytiscus elicits compensatory movements which tend to produce circling in the opposite direction. The eyes, antennae and legs are the sites of receptors whose asymmetric stimulation produces this response. It is suggested that a similar asymmetry in the inflow from these sense organs is responsible for the modified movements following limb amputations.
In the swimming of nymphs of Anax the legs retract slightly before shortening of the abdomen takes place. Although all the segments take part, a large proportion of this shortening occurs in segments 6–8 which contain the branchial chamber. Water is ejected through a small aperture during this contraction but when water is drawn in the anal valves are open wide.
The pressure in the branchial chamber rises to about 30 cm. water within 0·03 sec. and the reaction from the jet enables the animal to attain speeds of 30-50 cm./sec. within the first centimetre of propulsion. The duration of this pressure and of the impulse on the animal correspond with the time during which the abdomen contracts longitudinally.
ACKNOWLEDGEMENTS
I wish to thank Dr Tybjaerg Hansen of the Rigshospitalet, Copenhagen, for allowing me the use of his electric manometers to measure pressure changes in the dragonfly nymph.