Insects are Both Impeded and Propelled by Their Legs During Walking

It has been known for several years (Graham, 1972) that some insects do not walk with a constant body velocity. Rapid variations in forward velocity during each leg cycle have been found in locust (Burns, 1973), grasshopper (D. Graham, unpublished) and cricket (Weber, Thorson & Huber, 1981). The velocity variations are complex at slow speeds but in the tripod step pattern, used at the highest speeds, two clear maxima and minima appear in each leg cycle. In the stick insect the velocity minimum is usually close to zero and the maximum velocity of the body occurs near the middle of each retraction stroke. The behaviour is markedly different from that reported for walking in man, monkey and the dog where the maximum speed of the body is achieved at the end of the retraction stroke (Alexander, 1976; Cavagna, Heglund & Taylor, 1977) and the momentum of the body is maintained while recovering the supporting legs.

The present paper examines the velocity developed by the legs on each side of the body when walking on a pair of independent treadwheels and compares these results with the force platform measurements of Cruse (1976) and the motor output to the middle leg during walking on a self-propelled tread-wheel (Graham & Wendler, 1981). The results indicate that walking locomotion in the stick insect, and probably in many other insects, has special characteristics quite different from those of vertebrate locomotion, resulting in a rather bizarre walking behaviour which could be described as ‘lurching’ locomotion. It appears to be essential that the animal achieve a static support condition each time the recovered legs touch the walking surface. This ‘braking’ of forward movement is clearly demonstrated in slow motion films of free-walking animals. A physiological correlate of this effect is observed under conditions where the apparent inertia of the body can be increased. For example, walks on heavy wheels show that the promotor musculature is strongly active during the first part each stance phase. The experimental observations are discussed and it is concluded that such a characteristic is most probably a feature of a climbing rather than a flat surface walking system and may be of considerable importance in controlling walks down an inclined plane and on surfaces at arbitrary angles to the gravity vector.

The experimental arrangement for wheel walking is shown in Fig. 1 and has been described in detail previously (Graham, 1981). The animal ran on a wheel system consisting of two light independent wheels rotating on a common axle. The wheel axle was counter-balanced to give an upthrust of half the animal’s body weight (Carausius adult body weight, 0·8g). The walking behaviour was filmed at 18 frames/s using a upper 8 Cine Beaulieu camera. The distance moved between each film frame was measured by the movement of wheel markings or the displacement of the leg tarsus.

An example of the step pattern and the velocity recorded for each wheel is shown in Fig. 2A for an intact animal walking on the wheels and in Fig. 2B for a right, middle leg, amputee. These records can be compared with the free-walking data at two speeds shown in Fig. 2C and D. Fig. 3 shows the average velocity for 6–12 steps with the same step period to illustrate the phase constancy of the velocity profiles for each independent wheel, at different speeds. At a high step frequency the legs of the same segment are coordinated with a phase of 0·5 but because force maxima appear twice during each leg cycle approximately equal accelerations are produced on both side of the body at the same moment (Figs 2A, 3A). This minimizes rotation about a vertical axis through the centre of gravity and under this condition the middle leg generates approximately the same acceleration and maximum velocity as the combined contributions of the front and hind legs on the other side.

Removal of a middle leg on one side causes the animal to walk more slowly and the intact side only shows the slow walk pattern observed in free-walking stick insects (Fig. 2D). In this step pattern the sum of the intervals between the protractions of adjacent legs on the same side is shorter than the step period. This consequence of middle leg amputation in the adult is similar to that reported for free-walking 1st instar nymphs (Graham, 1977). This change in the relative timing of the front and hind leg protractions is accompanied by a separation of the velocity distribution into three separate peaks (Fig. 3B). Surprisingly, the amputated right hand side, which now possesses only two legs, also shows three peaks in velocity of the wheel which are coordinated with the intact left side. However, the first peak (upper trace, Fig. 3B) is considerably reduced in height compared with the corresponding peak on the unoperated side. The intact side of the amputee (lower trace, Fig. 3B) shows a similar velocity profile to the body velocity in the free, slow-walking adult (right trace, Fig. 3C) but in the free animal all six legs contribute to the velocity record of me body.

A close examination of the velocity curve for the intact side of the amputee (lower trace, Fig. 3B) suggests that each peak corresponds to the activity of a different leg. The largest peak approximately equals the sum of the two smaller ones and this largest peak occurs during the last half of the middle leg retraction stroke. The smallest peak on the intact side of the amputee appears when the front leg is lifting from the ground and therefore corresponds to the action of the hind leg which is in the last half of its retraction stroke. The middle sized peak occurs when leg 2 is off the ground (Fig. 3B) and in this case legs 1 and/or 3 must be responsible for this acceleration. This velocity maximum could be associated with either the front or hind leg, but it seems most likely that it is produced by the front leg during the time at which the large tibial flexor is expected to provide the main contribution to the power stroke. Including the observations on the intact side of the amputee and the intact animal the legs appear to provide a relative contribution of 100% (leg 2), 70% (leg 1) and 30% (leg 3) to the velocity peak on each side. When the animal walks at high speed (Fig. 3A and fast walk Fig. 3C) legs 1 and 3 overlap and combine to give a single peak similar in amplitude to that for the middle leg.

These assumptions allow a consistent explanation of the velocity peaks for the intact side of the amputee and the intact animal walking on the wheel, and they also provide a partial explanation of the data for the free-walking animal. Clearly this simplistic explanation of force production is inadequate for the operated side of the amputee for this side consistently shows three velocity peaks in the wheel for only two legs acting upon it (Fig. 3B). The only explanation of this velocity profile is that either the front or hind leg is required to provide two bursts of propulsion per stroke in this amputated configuration.

The simple model for force development described above does not explain the remarkable decelerations which occur twice in each cycle of leg movement. In the wheel-walking behaviour of adults and at high speeds of walking in free adults and at all speeds of walking in the 1st instar nymph the velocity is reduced almost to zero between each velocity peak.

When the animals walk on the treadmill, the legs on a side propel each wheel independently. Referring to Fig. 3A and C at the beginning of the stance phase in a fast walk, the velocity of each wheel drops rapidly from the maximum value to 10% of maximum (or zero in some instances) within 100 ms. This sharp fall in velocity occurs twice in each leg cycle for each alternating tripod of legs at the onset of the retraction stroke. This deceleration could arise from several possible sources: (1) the retracting legs might apply forces opposing the normal rearward wheel movement at the end of their stroke; (2) the protracting legs when they contact the wheel might decrease its velocity by (a) passive attachment to the wheel, (b) a strut effect moving the centre of gravity of the wheels downwards or (c) applying a force directly opposing the wheel rotation.

Let us consider the first deceleration interval of the rear leg cycle in Fig. 3A (right leg, step period 555 ms). A study of motor output during wheel walking for the middle leg of the stick insect (Graham & Wendler, 1981) shows that the tergo-coxal remotor museles of the middle leg have very stereotyped activity patterns. They are strongly active towards the end of retraction until the leg leaves the walking surface (with ± 30ms, see Fig. 8, Graham & Wendler, 1981). The activity of the excitatory axons of the coxal retractor muscles occurs throughout the wheel deceleration and there is no protractor activity. Therefore, the retracting middle leg cannot be decelerating the wheel. The second alternative is that the front and hind legs contact the wheel at the beginning of the deceleration and produce the observed fall in velocity. Let us consider the three possibilities (a), (b) and (c) in order. It can easily be shown that the addition of a load of 50 mg to the wheel rim, corresponding to attaching two legs to the wheel, will have a negligible influence on the wheel velocity (< 1 % change in 100 ms). In case (b) a calculation of the energy required to displace the centre of gravity of the wheel system downwards through a few millimetres by the three legs of a tripod acting as rigid struts shows that the wheel velocity would not be reduced by more than 30–40% within the required time interval, unless the retracting legs also apply forces pulling the wheel towards the body. There is no evidence of this in either wheel-walking animals (Graham & Baessler, 1981), or free-walking animals (see Fig. 7 of Cruse, 1976). In both studies the legs always provide downward acting motor output or forces when in contact with the walking surface. Thus one must conclude that possibility (c) is the correct interpretation of this behaviour. The legs are capable of producing sufficiently large forces to oppose the normal direction of movement for short time intervals using the protractor muscles and these forces may be even greater than those available during the retraction stroke. The data of Fig. 3A (right side) and 3C at 555 ms for the 1st peak show a deceleration which is greater than the acceleration of the middle leg which precedes it or the combined action of the front and hind legs which follows it. One must conclude that the front and/or hind leg actively opposes the middle leg retraction stroke during walking on the wheel, and a similar effect is observed when the middle leg first contacts the ground while the front and hind legs are retracting.

Is there any independent evidence for this unusual behaviour in which certain legs momentarily try to resist rearward movement while other legs are attempting to continue their retraction? Firstly, it should be remembered that these body velocity changes are found in several free-walking insects. Similar, profiles have been observed for locust (Burns, 1973), cricket (Weber et al. 1981) and grasshopper (D. Graham, unpublished data). No direct evidence is available for the cockroach but the data of Hughes (1952) suggest that leg displacement as a function of time is similar to that found here with a slow rate of change of position as the protraction phase changes in retraction even at step frequencies as high as 4 Hz.

Direct evidence that the legs in stance phase apply a force against the direction of body movement comes from the experiments of Cruse (1976) in which adult stick insects were enticed to step onto a force transducer in the walking substrate. These experiments show that on a horizontal plane the forces applied by a leg act against the direction of motion for the first part of the retraction stroke in both front and middle legs and have an approximately equal amplitude for both forward and rearward forces (see Fig. 4, Cruse, 1976). This behaviour has recently been confirmed at the motor output level for walking on a wheel while recording from the nerves innervating the protractor and retractor muscles driving the mesothoracic coxa (Graham & Wendler, 1981). These results show that for the first 10—50 % of the stance phase the protractor muscles are receiving a burst of excitatory input while the retractor muscles remain relatively silent. During the second half of the stance phase strong retractor output appears and the protractor muscles are inactive at this time (see Fig. 4, Graham & Wendler, 1981).

Forward directed forces during stance phase appear to be present primarily in the front and middle legs and so far there is no evidence for strong body decelerating forces in the hind legs. This is consistent with the velocity profiles of the slow, free-walking animal of Figs 2D and 3C, which shows that the minimum velocity occurs when leg pairs which do not include a hind leg begin their stance phase. This is also shown in the measurements of Cruse (1976) which indicate that hind legs only produce forces towards the rear.

Each front or middle leg produces three separate actions:

1. The leg is placed on the surface and actively resists forward movement of the body by other legs.

2. After being moved some 10–30 % towards the rear by the other legs it retracts strongly propelling the body forward. During retraction the other leg tripod is lifted from the ground.

3. The supporting leg then lifts from the surface and moves forward, in protraction, to the anterior extreme position.

Thus a leg in the retraction phase may or may not be active in retraction and normally resists forward movement of the body during the early part of what should be more accurately referred to as the stance or support phase of the walking cycle.

Recent experimental work on adult stick insects supported above a mercury surface (Graham & Cruse, 1981) demonstrates that the animals use coordinated movements when walking on mercury and have a stride length and anterior extreme position very similar to that seen in wheel walking. There is no indication of any forward movement of the leg after touch down on the surface. Thus in this situation where mechanical coupling of the legs cannot occur through the substrate (a liquid surface) the legs rarely show forward movements which would indicate protractor activity at the beginning of the stance phase. The protraction burst at the beginning of the stance phase in free and wheel-walking animals does not appear to be part of the motor output when legs are allowed to function independently. Most probably, it is a reflex response to displacement caused by the other legs acting directly through the substrate at a time when the leg is attempting to stand still. One must conclude that in the stick insect the legs are used actively to decelerate the body at regular intervals during walking as strongly as they are used to make the active propulsive movements which carry the animal forward over the ground. The forward directed forces increase the energy cost of locomotion for walking above a horizontal surface, for if the legs were used as passive articulated struts or even passive rigid supporting struts the average speeds achieved would be much higher and the energy cost would be greatly reduced, although the risk of falling might be increased.

The stand-push-recover system described here may be particularly useful for climbing animals where the advantages of momentum transfer cannot be used at the low speeds demanded by the requirement to find suitable tarsal support. This a relevant to adult stick insect locomotion, but it seems rather inappropriate for 1st instar stick insects since they spend much of their time walking rapidly over the relatively large and smooth surfaces presented by small leaves and twigs. Perhaps this rigid standing phase is necessary to hold the body in a suitable orientation to an inclined or vertical substrate and is important in maintaining contact with the surface?

If we imagine walking across a surface which slowly tilts until one is walking upside down then it is clear that maintenance of contact and orientation become vital considerations. Under such awesome conditions a rigid standing phase interposed between each cycle of forward propulsion could be essential in maintaining the body in a suitable position relative to the substrate. This orientation maintenance could be provided most simply by the standing reflexes which hold the legs in the required orientation relative to the body and automatically compensate for the externally imposed gravitational forces. In this situation presumably the maintenance of contact and/or orientation relative to the walking surface is more important for the insect than a reduction in energy cost on those rare occasions when the animal is walking in the upright position.

Finally, some mention should be made of the unusual ‘conflict situation’ created between those legs which are making contact with the substrate and are braking forward movement and the supporting legs which are pushing vigorously towards the end of their retraction stroke. There is no indication that motor activity is decreased in middle legs when front and hind legs contact the ground. On the contrary, it is at this time that the fast motor axons are recruited. At this point in the step cycle the front and hind legs succeed in decelerating the wheel while the middle leg on the same side is increasing the force it develops, perhaps to some extent reacting to the ‘braking’ activities of the front and hind legs.

Work by Wendler (1964), Cruse (1981) and Cruse & Pflueger (1981) has established that resistance reflexes are active during walking movements. They are activated whenever the insect leg has external forces applied to it which modify the intended movement. Essentially the leg acts as a set point servo system and can remain stationary or move at a rate determined by the set point but reacts to any externally imposed deviation from the set point.

The interesting question in the present context is whether an intended movement is a completely cooperative phenomenon involving all the legs, or can individual legs display their own intentions related to the particular part of the locomotor cycle they are performing.

In some respects this ‘conflict situation’ resembles the ‘collection’ of a horse by its rider in preparation for a jump or other difficult manoeuvre. Thus it may represent an increase in upthrust on the body produced by pushing the body forward with the middle leg at the same time as the front leg slows the body down.

Experiments are now in progress to examine motor output simultaneously in middle and hind legs whilst changing the inertia and propulsive loading of the system to investigate this ‘conflict situation’ in greater detail. The braking behaviour described should be of particular importance in slowing the animal when walking down a gradient where the body tends to ‘run away’ with the legs.

This work was carried out with DFG support from Ba 578 and I would like to thank the staff and students of the Biology Department, Kaiserslautern University for their help and advice, and the exciting research environment that they provide. I would also like to thank Dr F. Delcomyn for suggesting the title of this article and his assistance with the preparation of the manuscript.