Locusts use a composite of resilin and hard cuticle as an energy store for jumping and kicking

To jump, insects have to use mechanisms that generate rapid movements, usually of their hind legs, while delivering sufficient power to raise the weight of their body from the ground. To overcome the limitations of direct muscular contractions in propelling such fast but powerful movements, insects often use a catapult mechanism (Gronenberg, 1996; Patek et al., 2011). This mechanism is used by insects as diverse as fleas (Siphonaptera) (Bennet-Clark and Lucey, 1967; Rothschild et al., 1972; Rothschild et al., 1975), froghoppers (Hemiptera, Cercopidae) (Burrows, 2003; Burrows, 2006), planthoppers (Hemiptera, Fulgoridae) (Burrows, 2009), flea beetles (Coleoptera, Chrysomelidae) (Furth et al., 1983), and grasshoppers and locusts (Orthoptera) (Bennet-Clark, 1975; Brown, 1967; Burrows, 1995; Godden, 1975; Heitler and Burrows, 1977). The jumping performance achieved by using a catapult is impressive. Fleas accelerate their bodies in ~1 ms to take-off velocities of 1 m s⁻¹ (Bennet-Clark and Lucey, 1967; Rothschild et al., 1975; Rothschild et al., 1972; Sutton and Burrows, 2011). Froghoppers, with a body mass some 27 times that of a flea, accelerate in less than 1 ms to take-off velocities up to 4.7 m s⁻¹ (Burrows, 2003; Burrows, 2006), and some planthoppers with even heavier bodies accelerate in 0.8 ms to a take-off velocity of 5.5 m s⁻¹ (Burrows, 2009). Gregarious-phase locusts with a body mass of 1.5 to 2 g can accelerate their bodies in 20–30 ms to a take-off velocity of 3 m s⁻¹ by extension of their large hind legs (Bennet-Clark, 1975; Brown, 1967).

A catapult mechanism places two crucial requirements on the nervous system and on the body. First, the muscles must load the catapult by contracting slowly and without moving the legs, before the stored energy is released to extend the hind legs rapidly in the jump. This implies the generation of a complex and repeatable motor pattern. The motor pattern for such a catapult mechanism in locusts (Burrows, 1995; Godden, 1975; Heitler and Burrows, 1977) consists of three phases. First, a cocking phase in which the tibia is flexed fully about the femur of a hind leg. Second, a co-contraction phase in which the flexor and extensor tibiae muscles contract but the tibia remains fully flexed about the femur. A key action here is the spike pattern of the single fast extensor motor neuron to the large extensor tibiae muscle. Mechanical specialisation of the femoro-tibial joint gives the smaller flexor muscle a mechanical advantage over this large muscle when the tibia is fully flexed, thus preventing the tibia from extending (Heitler, 1974; Heitler, 1977). Third, a rapid extension or jump phase in which the flexor motor neurons are inhibited and the energy stored by the prolonged contraction of the extensor is suddenly released to power the rapid extension of the hind tibia. A similar sequence of muscle actions also occurs in froghoppers (Burrows, 2007) and planthoppers (Burrows and Bräunig, 2010), but with a more prolonged energy storage phase.

Second, the energy generated by the prolonged contractions of the muscles must be stored efficiently and then delivered rapidly to power the jump. There are two competing demands: (1) stiffness to store energy and (2) elasticity to restore shape. For 50 years, it has been thought that the energy storage devices of locusts were made only of hard material in the form of black cuticle. Hard materials can store much energy with little bending but are liable to fracture. Soft elastic materials cannot store much energy but can restore shape quickly. Froghoppers (Burrows et al., 2008) and planthoppers (Burrows, 2010) store energy by bending bow-shaped pleural arches that are a composite of hard cuticle and soft resilin. The design of these storage devices is similar to that of an archery bow, which depends heavily for its performance on the use of composite materials (Miller et al., 1986). Fleas are suggested to store energy in a pad of resilin located in the internal skeleton of the metathorax at the articulation between the notum and pleuron (Bennet-Clark and Lucey, 1967; Rothschild and Schlein, 1975; Rothschild et al., 1975). By itself, however, calculations suggest that resilin can store very little energy (Burrows et al., 2008). Instead, one role may be to prevent fracture during bending or distortion of hard, black cuticle, which can store the required energy.

A key element in many of these energy storage structures is the rubber-like protein resilin (Weis-Fogh, 1960). Resilin consists of coiled peptide chains linked together in a stable, isotropic, three-dimensional network by the fluorescent amino acids dityrosine and trityrosine (Andersen, 1963; Andersen, 1964; Malencik et al., 1996). It is mechanically highly deformable and shows almost perfect elastic recovery (Andersen and Weis-Fogh, 1964). Energy loss from resilin during movements at 200 Hz is less than 5% (Jensen and Weis-Fogh, 1962), suggesting that it can act as a useful spring over a wide range of speeds (Bennet-Clark, 1997; Fonseca and Bennet-Clark, 1998). Recently, the gene CG15920 in Drosophila encoding a resilin-like protein (pro-resilin) has been identified (Ardell and Andersen, 2001) and has been inserted into Escherichia coli to obtain the gene product, which has then been cast into a rubber-like material by photochemical cross linking; the first exon of this gene product was used to raise an antibody (Elvin et al., 2005). This antibody has recently been shown to stain resilin in the energy storage devices for jumping in froghoppers and planthoppers (Burrows et al., 2011), and a flea (Siphonaptera), a fly (Diptera) and a dragonfly (Odonata) (Lyons et al., 2011).

Adult locusts and all of the larval instars store half the energy generated by the large extensor tibiae muscles by bending two bow-shaped semi-lunar processes at the femoro-tibial joint of each hind leg. The other half is stored in the elasticity of the muscles, their tendons, and in distortions of the femoral cuticle from which the muscle fibres arise (Bennet-Clark, 1975; Burrows and Morris, 2001; Cofer et al., 2010; Cofer et al., 2007). Only the hind legs generate the forces necessary to propel jumping, and semi-lunar processes are not present in the front and middle pairs of legs. These semi-lunar processes have been considered to be made only of hard cuticle and might therefore be supposed to be brittle and subject to fracture when subjected to high stresses. Bennet-Clark (Bennet-Clark, 1975) reports that if the external surface cuticle of a semi-lunar process is scratched with a scalpel, a transverse fracture occurs when the locust next attempts to jump. When, however, a hind leg is externally loaded, the head of the tibia, the suspensory ligaments of the tibia and the femur proximal to the femoro-tibial joint break before the semi-lunar processes. The presence of resilin in the semi-lunar processes has not been reported.

In this study we have sought to determine whether resilin is involved in energy storage by locusts for jumping and kicking. We show that resilin, as detected by its fluorescent properties, forms an internal layer inside the hard, black cuticle of each semi-lunar process and externally forms a triangular-shaped area ventral to a semi-lunar process. These storage devices are thus composite structures, a conclusion supported by calculations of the Young's modulus (which is a measure of stiffness) for a semi-lunar process. We then analyse the distortions of the resilin during natural kicking movements and during contractions of the extensor tibiae muscle that simulate natural kicking. We show that the resilin and hard cuticle of the semi-lunar processes are progressively bent during co-contraction of the muscles, and that the externally visible resilin is compressed. The resilin quickly resumes its natural shape at the end of a jump or kick, returning all the stored energy and ensuring that the hind leg is also restored to its original shape, ready for the next movement.

Forty-two adult male and female gregarious-phase locusts [Schistocerca gregaria (Forskål 1775)] were analysed from our crowded culture. Five juvenile locusts from each of the five, crowded larval instars were also used. The ages of these instars ranged from 3 to 5 days from their last moult. Cast skins of fourth instar locusts were examined within 2 days of the moult.

The anatomical features of the medial, lateral and dorsal surfaces of the distal femur of the hind leg were drawn with the aid of a drawing tube attached to a Leica MZ16 stereo microscope (Wetzlar, Germany) and photographed on the same microscope with a Nikon DXM 1200 digital camera (Kingston upon Thames, Surrey, UK). Longitudinal incisions along the dorsal midline were made with a thin razor blade through the distal part of the hind femur to reveal the structure of the inside, lateral and medial surfaces. A series of transverse sections were also cut by hand through the same region of freshly moulted, and thus soft, adults and larvae.

To reveal the possible presence of the elastic protein resilin, hind femoro-tibial joints were placed in a Petri dish with a floor of Sylgard on the stage of an Olympus BX51WI compound microscope (Olympus UK, London, UK). The external surfaces of the distal end of the femur were examined in air, and longitudinal or transverse sections of this region were examined in locust saline (Usherwood and Grundfest, 1965). Both surfaces were viewed through Olympus MPlan 5×/0.1 NA, MPlan 10×/0.25 NA and LUCPlanFLN 20×/0.45×NA objective lenses under ultraviolet (UV) or white epi-illumination. Images were captured with a Micropublisher 5.0 digital camera (QImaging, Marlow, Bucks, UK). The UV light was provided by an X-Cite series 120 metal halide light source (EXFO, Chandlers Ford, Hants, UK), conditioned by a Semrock DAPI-5060B Brightline series UV filter set (Semrock, Rochester, NY, USA) with a sharp-edged (1% transmission limits) band from 350 to 407 nm. The resulting blue fluorescence emission was collected in a similarly sharp-edged band at wavelengths from 413 to 483 nm through a dichroic beam splitter. Images captured at the same focal planes under UV and white light were superimposed in Canvas 12 (ACD Systems of America, Miami, FL, USA). To establish whether the fluorescence was sensitive to pH, the bathing saline was changed from its normal value of 7.2 to pH 2 with 2 mol l⁻¹ hydrochloric acid, and to pH 12 with 2 mol l⁻¹ sodium hydroxide. During each test sequence, the camera gain and exposure time were constant.

Images of the distortions that occur in the distal femur during natural kicking were captured at a rate of 1000 images s⁻¹ and an exposure time of 0.2 ms (supplementary material Movie 1) with a single Photron Fastcam 1024 PCI camera (Photron Europe, High Wycombe, Bucks, UK). The body of the locust was restrained in Plasticene but with the hind tibiae free to move. In different kicks, the orientation of the locust was changed to allow the distal femur to be viewed laterally, dorsally and from its distal end. To simulate the distortions of the femur that occur during natural jumping and kicking, a pair of 50 μm silver wires, insulated but for their tips, was pushed through small holes in the proximal femoral cuticle and into the extensor tibiae muscle of an isolated hind leg. A sequence of stimuli, 0.5 ms in duration and at a frequency of 50 Hz, generated by a Master8 stimulator (AMPI, Jerusalem, Israel), was delivered to the muscle for a period of 0.2 to 1 s, reflecting the pattern of spikes in the fast extensor tibiae motor neuron that is recorded during natural kicking movements (Burrows, 1995; Heitler and Burrows, 1977). The amplitude of the stimuli was adjusted so that a single stimulus elicited a twitch movement of the tibia, a characteristic outcome of a spike in the large, fast extensor tibiae motor neuron. The tibia was restrained fully flexed about the femur, a position it adopts under the control of the flexor tibiae muscle in preparation for a natural kick or jump. The femoro-tibial joint was illuminated either with white or UV light. The resulting distortions of the distal femur were captured in two ways: first, as a series of 50 colour images with the Micropublisher colour camera; second, as black and white images at rates of 250 images s⁻¹ and exposure times of 0.3 ms (supplementary material Movies 2–4) with a monochrome Photron Fastcam 1024 PCI. Both cameras were mounted on the Olympus BX51WI microscope. Each camera was triggered to record by the same electrical pulse that initiated the sequence of stimuli. The images captured were then fed directly to a computer.

Lengths of the legs and of the semi-lunar processes of fixed specimens were measured to an accuracy of 0.1 mm from images captured with a digital camera attached to a Leica MZ16 microscope and projected onto a 24 inch monitor. Body masses were determined to an accuracy of 0.1 mg with a Mettler Toledo AB104 balance (Beaumont Leys, Leicester, UK).

Especially prominent on the distal hind femora are black semi-lunar processes, one on the lateral (anterior) and one on the medial (posterior) surface (Figs 1, 2). Each weighs approximately 1.3 mg (Bennet-Clark, 1975) and in adult females was approximately 2.4 mm long and 0.7 mm from dorsal to ventral at its widest point. Each semi-lunar process merges dorsally into the cuticle of the surrounding femur and proximally joins a dorso-ventral band of dark cuticle. At its ventral edge, however, there is an abrupt transition distally to a triangular-shaped region of translucent material that tapers more proximally to a narrow finger of apparently similar material (Fig. 1, Fig. 2, blue colouration). This translucent material is fused with the black semi-lunar process dorsally and with the lighter coloured cover plate ventrally and occurs on both the lateral and medial surfaces. Distally, each semi-lunar process tapers and curves ventrally to form the articulations with the paired horns of the tibia, which curve dorsally (Fig. 1).

When both the medial and lateral external surfaces of the distal hind femur were illuminated with UV light of a wavelength between 350 and 407 nm, blue fluorescence was seen that exactly matched the distribution of translucent material seen under white light (Fig. 1, Fig. 2A,B, Fig. 3A,B). A distal triangular area delineated the semi-lunar process dorsally from the cover plate ventrally, and a narrow finger of fluorescence extended to the proximal third of each cover plate. When the internal surfaces were revealed by splitting the distal femur along its longitudinal midline and removing the tracheae and the hypodermis, the entire internal surface of each semi-lunar process showed intense blue fluorescence (Fig. 3C,D). On each internal face, the distal triangular region and its tapering proximal tail (black arrows in Fig. 3), which were also visible externally, also fluoresced, and were separated distally from the semi-lunar process by a region of non-fluorescent cuticle. The ventral and distally facing surface of Heitler's lump (Heitler, 1974), an inward protrusion of the ventral surface of the femur, also showed some weaker blue fluorescence (Fig. 3C,D yellow arrows).

To follow the shape of the semi-lunar processes and the relative distribution of black cuticle and fluorescent material, a series of thick, transverse sections were cut by hand through the distal femur (Fig. 4). A sawing motion of the razor blade generally avoided fracturing the structures, particularly in young adults. Three key features became apparent. First, the two semi-lunar processes were bilaterally symmetrical (Fig. 4A). Second, their three-dimensional shape was complex, with curves in all planes and an internally pointing ridge running from proximal to distal (Fig. 4). No muscle fibres or other structures attached to these ridges. Third, an entire semi-lunar process is a composite structure with an outer surface of black cuticle and an inner surface of blue fluorescent material (Fig. 4B–D,Bi–Di). The black and fluorescent materials did not separate during sectioning.

Over most of its length, a semi-lunar process was 80 μm thick, with the two material components contributing differently. At the internal protruding ridge, the fluorescent material was much thicker (165 μm) than the black cuticle. In other places, the black cuticle was as thin as 30 μm, with the fluorescent material correspondingly thicker.

The specificity of the wavelength of the UV light needed to elicit the blue fluorescence from the distal femur and its semi-lunar processes is one of two key signatures of resilin (Andersen and Weis-Fogh, 1964; Burrows et al., 2008; Neff et al., 2001). The second key characteristic of resilin is the reversible, pH dependence of the blue fluorescence. To test whether the fluorescence met this key signature, the distal femur was immersed in locust saline and the pH was changed. In saline with a pH of 2.0, the fluorescence gradually declined after 30 min, but recovered to its former full intensity after a similar period when the pH was returned to its normal value of 7.4. If the pH was changed to 12, then the intensity of the fluorescence increased, but again returned to its former level when the pH was restored to normal.

During natural kicking, the semi lunar processes are known to bend in the dorso-ventral plane as viewed from the lateral surface (Burrows and Morris, 2001; Bennet-Clark, 1975). In the present study, the tips of each semi-lunar process moved ventrally and posteriorly so that they disappeared behind the ventral coverplate. Images captured here showed that bending also occurred in the other two dimensions, as seen in dorsal and end-on views (Fig. 5A–C). Both semi-lunar processes bowed outwards and twisted so that the width of the distal femur increased by 15±4% (grand mean ± s.d., for a minimum of four kicks by each of seven adult gregarious locusts). The complex distortions are more readily seen in movies of natural kicks captured at 1000 images s⁻¹ with the Fastcam high-speed camera (supplementary material Movie 1). Alternatively, they can be seen in tracings of individual frames of such movies before a kick, at the end of the contractions of the tibial muscles and before the tibia is extended rapidly in a kick (Fig. 5D). After the initial rapid extension of the tibia, the semi-lunar processes rapidly resumed their normal shape (compare the shapes in Fig. 5A,C).

To determine how the fluorescent material in the distal femur might be distorted and therefore store energy, kicking movements were simulated by electrical stimulation of the extensor tibiae muscle of an isolated hind leg. The motor pattern for kicking and, in particular, the sequence of spikes in the fast extensor tibiae motor neuron is known in detail (Burrows, 1995; Heitler and Burrows, 1977). A sequence of stimuli that simulated the natural pattern of motor spikes of the single fast extensor tibiae motor neuron during kicking was therefore used to activate the extensor tibiae muscle. The tibia was held fully flexed about the femur and images were captured at a rate of 250 s⁻¹. Stimulation caused the semi-lunar process to bend so that its distal tip moved progressively ventrally, eventually disappearing inside the cover plate (Fig. 6A, supplementary material Movie 2). The fluorescent material that is bonded to the inside surface of the semi-lunar process is presumed to bend in the same way, but cannot be imaged. The bending movement of the semi-lunar process also compressed the externally visible translucent material. When the stimulation stopped, the semi-lunar process unfurled and the whole distal femur rapidly resumed its former shape (Fig. 6B). Plotting the movement of the distal tip of the semi-lunar process against time showed that the tip moved by 400 μm and that the recovery was consistently faster that the bending (Fig. 6C).

The simulation was then repeated under UV illumination and 50 sequential images were captured by a Micropublisher colour camera to determine in detail the effects on the fluorescent material under such conditions (Fig. 7, supplementary material Movie 3). Before stimulation, colour images showed the typical triangular area of blue fluorescence tapering to a narrow finger between the semi-lunar process dorsally and the cover plate ventrally (Fig. 7A). During stimulation, the area of blue fluorescence was progressively compressed to form a narrow slit when the semi-lunar process was maximally bent. When the stimulation stopped, the fluorescence rapidly resumed its original shape (Fig. 7B). At higher magnification and with black and white images captured at a rate of 250 s⁻¹ with the Fastcam high-speed camera, the material that fluoresces blue under UV light could be seen to wrinkle and compress during the stimulation period, and then to unwrinkle and spring back to its natural shape when the stimulation stopped (Fig. 8, supplementary material Movie 4).

Locusts typically develop through a series of five nymphal instars, which resemble miniature adults (hemimetabolous development) and, like the adults, jump and kick, except for brief periods before and after each moult (Norman, 1995). We therefore sought to determine whether the energy storage mechanisms in the distal femur contained blue fluorescent material indicating the presence of resilin (Figs 9, 10).

In first instar nymphs viewed externally, the distal femur was not black except for the semi-lunar processes and thus allowed light to penetrate the surrounding cuticle (Fig. 9A). A crescent-shaped area of blue fluorescence was visible both dorsal and ventral to a semi-lunar process. A paler triangular-shaped area of fluorescence between the semi-lunar process and the cover plate was also present. Some fluorescence was also apparent at the insertion of the flexor tibiae tendon on the ventral tibia. In second instar nymphs, the external, distal femur and the proximal tibia were darker, but a blue crescent of fluorescence dorsal to the semi-lunar process and a triangular area ventral to it were again present (Fig. 9B). In third and fourth instar nymphs (Fig. 9C,D), this pattern was repeated with a dorsal crescent and a ventral triangular area visible in each. In fifth instar nymphs, the dorsal crescent was narrow and was accompanied by two thin and more ventral crescents (Fig. 9E). A triangular area of fluorescence was also present in each of these instars.

When viewed internally, both the medial and lateral faces of the semi-lunar processes of all five instars showed bright blue fluorescence (Fig. 10, second instar not shown). As in the adults, a layer of blue fluorescent material covers the entire internal face of each semi-lunar process in a femur.

At each moult the fluorescent material was shed with the exoskeleton. The dry exoskeleton of a fourth instar nymph that had moulted into a fifth instar still showed the same pattern of fluorescence as in the live locust (Fig. 11). The lateral, external surface of a hind femur had a triangular area of bright blue fluorescence (Fig. 11A). The internal surface was dominated by the blue fluorescence from the entire face of the semi-lunar process and with the triangular area visible ventral to it (Fig. 11B).

This paper has shown that key energy storage devices for jumping and kicking in locusts are made of a composite material. This material contains hard black cuticle on its external surfaces and soft resilin forming a lining along the whole internal surfaces, with some also visible from the outside. These paired semi-lunar processes are only present at the femoro-tibial joints of each hind leg that propel jumping. These composite structures are bent in three dimensions when the flexor and extensor tibial muscles co-contract in advance of a natural jump or a kick and energy is stored. In the simulation experiments reported here, the extensor tibiae muscle was stimulated with electrical pulses reproducing the pattern of spikes recorded from the fast extensor motor neuron during a natural kick, and with the tibia restrained mechanically in its fully flexed position. These experiments therefore did not simulate the co-contraction that occurs in natural kicking and may not have revealed the full extent of the distortion of the resilin. Nevertheless, the semi-lunar processes were bent by an amount similar to that in natural movements, and an externally visible area of resilin was compressed and wrinkled. When the catapult mechanism was released, the semi-lunar processes unfurled and, together with the compressed areas of resilin, rapidly resumed their natural profile. The result was that all the stored energy was returned and the distal end of the femur was quickly restored to its natural shape, ready for another kick or jump. The same composite of materials was present in the semi-lunar processes of each larval stage, which, like the adults, jump and kick, and was shed with the external skeleton at each moult.

The fluorescent material associated with the semi-lunar processes had the same key signatures as resilin (Andersen and Weis-Fogh, 1964; Burrows et al., 2008; Neff et al., 2001; Weis-Fogh, 1960). First, the intense blue fluorescence had the general excitation and emission characteristics expected for resilin. Excitation was elicited by a sharp-edged band of wavelengths from 350 to 407 nm and the resulting blue fluorescence emission was collected in a similarly sharp-edged band at wavelengths from 413 to 483 nm. The fluorescence was strongly delineated to the structures that we describe and none was visible as a background of fluorescence in surrounding tissue. Less intense fluorescence was seen at Heitler's lump (Heitler, 1974), a ventral invagination into the ventral wall of the distal tibia, and more intense fluorescence at a clearly delineated band at the buckling region of the proximal tibia (Bayley et al., 2012). Second, the fluorescence was sensitive to the pH of the locust saline: when made acidic, the fluorescence faded but did not completely disappear, and when made alkaline, the fluorescence increased in intensity. Both effects were completely reversible. These two tests are currently the only reliable methods to test for the presence of resilin, and therefore on this basis we conclude that the fluorescent material is resilin. A recently introduced antibody that stains resilin in several orders of insects should add a further means of identification in future studies (Burrows et al., 2011; Lyons et al., 2011).

Resilin was associated with the semi-lunar processes in all larval instars, a distribution that correlates well with the fact that they all jump. We only used larvae 3 days after a moult and thus did not attempt to follow changes that might accompany maturation. The resilin was shed with the exoskeleton at each moult so that the composite structure of the semi-lunar processes must be reformed at each moult. At the final moult to adulthood, locusts start making resilin 3 days before the moult and continue to do so for 15–20 days after this final moult (Neville, 1963). The resilin is secreted and probably added in layers to previous resilin, perhaps explaining the non-entanglement of chains of its molecules (Neville, 1963). In fleas, which have three larval stages, resilin appears to be laid down late in development (Rothschild and Schlein, 1975). In froghoppers, larval stages do not jump as they live in a protective froth from which they emerge only at the final moult to adulthood when the energy stores are completely formed (Burrows et al., 2008). By contrast, the larval stages of planthoppers live a free life in the same habitat as the adults and, like the adults, jump. As in locusts, each larval instar and the adults have resilin in their energy storage devices, the pleural arches. These observations indicate a clear correlation between the presence of resilin and the need for energy storage in jumping.

The calculated value of 35 MPa for the Young's modulus of a semilunar process is within the range of values observed for insect cuticle (Vincent and Wegst, 2004). A semi-lunar process also contains resilin, which has a maximum Young's modulus of only 1–2 MPa, making the semi-lunar process an order of magnitude stiffer than resilin alone (Vincent and Wegst, 2004). If we therefore consider the two constituent materials of a semi-lunar process separately, applying Eqn 3 to the black cuticle only, this requires reducing the value for area by one-half, while all the other terms remain the same. The result is that to store the required energy, the Young's modulus of the black cuticle would have to be much higher at 80 MPa. The conclusion is that, like the froghopper (Burrows et al., 2008), the energy within the semi-lunar process is mostly stored in the hard cuticle, with resilin contributing some function other than purely energy storage. What could that function be?

If the surface cuticle of a semi-lunar process is scratched with a blade, a transverse fracture can occur when the locust next tries to jump (Bennet-Clark, 1975). When, however, the extensor tibiae muscle of a fully flexed hind leg is stimulated experimentally, any fracture normally occurs at the extensor apodeme, although fracture at a semi-lunar process has been reported. External loading to destruction usually breaks the head of the tibia, the suspensory ligaments of the tibia, or the femur proximal to the semi-lunar processes; such fractures always occur if the load exceeds 17 N, but loads of 14 N can be sustained for several seconds. These observations are explained by the demonstration that the semi-lunar processes are a composite of hard cuticle and soft resilin. Resilin is known to be incorporated into such composites or metamaterials through chitin-binding domains (Neville, 1963), now identified as a consensus sequence (Andersen, 2010; Rebers and Riddiford, 1988). The placement of the resilin reflects the likely bending forces to which the semilunar processes will be subjected during muscular contractions that precede a kick or a jump. The triangular area of resilin visible from outside is placed where the compression is greatest when a semi-lunar process is bent in the dorso-ventral plane (Fig. 7). The internal placement of the resilin as a layer along the inside of a semi-lunar process is where the greatest compression will occur when the whole structure bends outwards during the same period. The resilin in the semi-lunar process may be there as a protection against fracture when the semi-lunar process is excessively loaded. Analogies here would be with plywood in which the laminations prevent fracture and give added strength, and with bricks linked by mortar, which prevents cracks from spreading. It may also be there to restore a recoiling semi-lunar process rapidly to its original shape, and thus return all the stored energy, and allow a jump or kick to be repeated.

This paper adds further evidence to an emerging and unifying picture of the construction and action of energy storage devices across different insects in which composite materials play a crucial role. Until recently, these devices were thought to work in different ways in different insects. In fleas, for example, resilin was proposed as the only component needed for energy storage (Bennet-Clark and Lucey, 1967). Resilin is present in a pad at the articulation of the pleuron with the notum in the metathorax; this pad is compressed during the build-up to a jump and then recoils like a spring to power the rapid movements of the hind trochantera. Based on the volume and properties (Weis-Fogh, 1960) of resilin present, and on the assumption that it is stretched by 100% during jumping, calculations showed that it could store sufficient energy for jumping (Bennet-Clark and Lucey, 1967). In fleas, therefore, energy is supposed to be stored by a soft material. By contrast, the spring in the hind femur of flea beetles is made of hard cuticle and resilin has not been found to be associated with it (Furth, 1988; Furth et al., 1983; Ker, 1977). Similarly, in locusts, resilin had not previously been found in any of the places where energy for jumping and kicking was proposed to be stored. In particular, the semi-lunar processes in locusts, which store approximately half of the required energy (Bennet-Clark, 1975), were thought to be made only of hard cuticle. This paper has shown that they are composites of black cuticle and resilin.

In froghoppers and planthoppers, a merging of these two divergent views has been proposed (Burrows et al., 2008). The energy stores in these insects, the metathoracic pleural arches, are a composite of hard cuticle and resilin. They bend in preparation for jumping and return to their precise former shape once the jump is completed. Calculations based on the stiffness of resilin (Vincent and Wegst, 2004) undergoing the observed distortions during preparation for a jump indicated that resilin alone could store only a small proportion (1–2%) of the required energy, with the rest stored in the hard cuticle. This arrangement means that the whole structure needs only to be bent by a small amount and hence the muscles need only to shorten by a little. It has been suggested (Burrows et al., 2008) that the resilin makes it possible to do this by preventing fracture of the hard cuticle. By virtue of its highly resilient properties, resilin also enables all of the stored energy to be returned for jumping, and may also return the distorted structure of the legs or body quickly to their original shape, thus allowing further jumps to be performed. This would be an important survival mechanism when rapid escape movements from a predator are needed. The use of a composite material in this way provides a much greater impact than either material on its own.

The demonstration in this paper that composite materials are likewise used for energy storage in locusts raises three questions for insect jumping more broadly. First, is a similar arrangement used by fleas, where so far only resilin has been implicated (Bennet-Clark and Lucey, 1967)? For example, is there enough resilin in fleas and can it be compressed sufficiently to be able to store the energy needed for jumping, or must bending of some of the cuticular struts in the thorax also be implicated? Second, in the femoral springs of flea beetles, in which hard cuticle alone is implicated in energy storage (Furth, 1988; Ker, 1977), could resilin also be present? Third, how is resilin distributed in other orthopterans, which include a large number of powerful jumpers, but which use two distinct mechanisms? A catapult mechanism is used by locusts, pygmy mole crickets (Tridactylidae) (Burrows and Picker, 2010), horse head grasshoppers or false stick insects (Proscopidae) (Burrows and Wolf, 2002) and crickets (Gryllidae) (Hustert and Baldus, 2010; Hustert and Gnatzy, 1995). Leverage provided by their very long hind legs is used by bush crickets (Tettigonidae) (Burrows and Morris, 2003). Does the construction of the semi-lunar processes in these insects reflect the energy storage requirements of their jumping and is this therefore expressed by differences in the contribution from resilin in the two groups? The results of this paper suggest that layered resilin/cuticle composites may be ubiquitous in jumping insects.

We thank Jo Riley and Marina Dorosenko for their invaluable help and Cambridge colleagues for their many constructive suggestions during the course of this work and on the manuscript.