Jumping and take-off in a winged scorpion fly (Mecoptera, Panorpa communis)

To jump from the ground or a plant, many insects, with a few notable exceptions, rely on propulsion by the legs, sometimes with the assistance of wing movements. Two basic strategies provide leg propulsion. The first is a catapult method that generates some of the fastest leg movements in any animal and some of the fastest take-off velocities. This requires mechanical specialisations to enable the energy that is built up by slow muscle contractions to be first stored without moving the legs, usually in distortions of the skeleton, and then released suddenly to power the rapid movements of the legs. The power (energy/time) of the muscles is thereby amplified greatly. In planthoppers (Hemiptera, Fulgoridae), some of the best exponents of this strategy, the slow preparatory phase of a jump, is followed by a short acceleration time, often less than 1 ms, in which rapid movements of the hind legs generate propulsion. The accelerations can exceed 500 g and the final take-off velocity can reach 5 m s⁻¹ (Burrows, 2009). Other insects that use this mechanism are grasshoppers (Orthoptera, Acrididae) (Bennet-Clark, 1975), flea beetles (Coleoptera, Chrysomelidae) (Brackenbury and Wang, 1995; Nadein and Betz, 2016) and many other hemipteran plant-sucking bugs such as froghoppers (Hemiptera, Cercopidae) (Burrows, 2006). Such extreme movements require specialisations of the body. For example, mechanical, interacting gears may synchronise movements of the propulsive legs to within 30 µs of each other so that thrust can be generated in a particular direction and energy is not lost to rotation of the body (Burrows and Sutton, 2013).

The second method uses direct contractions of muscles to power the propulsive leg movements. The acceleration times of these jumps are usually longer and the final take-off velocities are generally lower. Many insects using this mechanism propel themselves with elongated hind legs, as exemplified by bush crickets (Orthoptera, Tettigoniidae), which can have propulsive hind legs that are four times longer than the front legs (Burrows and Morris, 2003). The leverage provided by these long legs is offset by the longer time (30 ms) it takes to extend them fully. Many true flies (Diptera), however, use their middle legs (Card, 2012), which are no longer than the other two pairs of legs. For some insects, a further variation to this mechanism is to use the middle and hind pairs of legs together; examples are lacewings (Neuroptera, Chrysopidae) (Burrows and Dorosenko, 2014), caddis flies (Trichoptera) (Burrows and Dorosenko, 2015b), moths (Lepidoptera) (Burrows and Dorosenko, 2015a), praying mantises (Mantodea, Mantidae) (Sutton et al., 2016) and ants (Hymenoptera) (Baroni Urbani et al., 1994; Tautz et al., 1994). Effects of this are to distribute the forces applied to the substrate over a larger area and to increase muscle mass to reduce the power requirements per unit of muscle. The synchronisation between the leg movements does not have to be closely controlled, so that in different jumps one pair of legs may move first and one pair may leave the ground before the other. In some winged insects, propulsive leg movements can also be accompanied by the start of flapping movements of the wings, for example, in moths (Burrows and Dorosenko, 2015a), and even if the wings are not moved, their shape can influence stability of the trajectory in whiteflies (Ribak et al., 2016).

Flightless scorpion snow fleas (order Mecoptera, family Boreidae) use strategies that incorporate elements of both catapult and direct contraction mechanisms to produce a jumping performance that lies at the boundary between the two; acceleration times are approximately 7 ms and take-off velocities are 0.7 to 0.8 m s⁻¹. Jumps are propelled by the combined movements of the middle and hind pairs of legs with three factors indicating that there must also be some power amplification by storage of muscle energy (Burrows, 2011). First, the power required for the fastest jumps reaches and may exceed the maximal contractile limits of muscle (Askew and Marsh, 2002; Ellington, 1985; Josephson, 1993; Weis-Fogh and Alexander, 1977). Second, performance is not affected by temperatures of −3°C to +3°C in their natural habitat when jumping on snow, as would be expected were direct contractions of the muscles alone involved (Burrows, 2011). Third, a thoracic region containing the elastic protein resilin and where energy could be stored is associated with each of the four propulsive legs (Burrows, 2011). This small family (Boreidae) contains only about 30 of the 600 total species in the order Mecoptera (Whiting, 2002), and Whiting indicates that ‘boreids are unique among Mecoptera in their ability to jump up to 30 cm’.

To test the hypothesis that jumping may be present more widely in this order of insects and that the same mechanism is used, analysis was undertaken of a larger, winged species, the scorpion fly Panorpa communis, from a genus that itself contains almost half of mecopteran species. Hasken (1939), in his detailed anatomical studies of the skeleton and musculature of this scorpion fly, gained the impression that it could move from the ground in what he described as a clumsy flutter involving the participation of the hind and middle legs (Hasken, 1939). When he removed the wings, a clearer propulsive contribution from the legs became apparent. Can these scorpion flies jump, and if so, are there any common features with scorpion snow fleas in the mechanism that they use and the performance that they generate?

Adult, winged scorpion flies Panorpis communis Linnaeus 1758 were collected during the summers of 2013–2019 from hedgerows and from patches of stinging nettles (Urtica dioica) or bramble (Rubus fruticosus) in the countryside around Cambridge, UK. They are omnivorous, feeding on pollen and fruit, and are scavengers of dead insects and rotting fruit. They belong to the order Mecoptera and the family Panorpidae. Fossil representatives of these groups date to the Permian, 250 million years ago.

Photographs of live P. communis (Fig. 1A,B) were taken with a Nikon D7200 camera fitted with a 105 mm Nikon macro lens. The lengths of the legs and of the body were examined in live insects, and in those fixed in 70% alcohol. Images were captured with a GXCAM-5C digital camera (GT Vision Ltd, Stansfield, Suffolk, UK) attached to a Leica MZ16 microscope (Wetzlar, Germany) and projected onto a large monitor. Leg and body lengths were then measured from these images to an accuracy of 0.1 mm with a ruler (Table 1). Body masses were determined to within 0.1 mg with a Mettler Toledo AB104 balance (Beaumont Leys, Leicester, UK). The body shape of male and female P. communis is characterised by a head with a large ventrally pointing beak-like rostrum with biting mouthparts at the distal end and two pairs of similarly-sized wings that project beyond the tip of the abdomen (Fig. 1A,B). Males also have an upwardly curved posterior region of the abdomen that ends in a prominently enlarged distal part that is red in colour (Fig. 1A) and contains a curved, terminal pair of claspers pointing anteriorly that are used to hold females during copulation, which can be prolonged. This structure resembles in appearance, but not in action, the stinging apparatus of a scorpion, hence the common name given to members of the family Panorpidae.

Sequential images of jumps were captured at a rate of 1000 s⁻¹ and an exposure time of 0.2 ms, with a Photron Fastcam SA3 high-speed camera [Photron (Europe) Ltd, West Wycombe, Buckinghamshire, UK] fitted with a 100 mm macro f/2.8 Tokina lens. Images were fed directly to a computer. The insects were free to jump in a chamber with an internal width of 80 mm, height of 80 mm and depth of 25 mm. The front wall was made of optical quality glass, the floor, side walls and ceiling of 12 mm thick, closed-cell foam (Plastazote, Watkins and Doncaster, Leominster, UK). If an insect bumped into the side of the chamber during take-off, then that jump was excluded from the analysis of performance. If contact occurred after take-off, then it fell outside the period of analysis and was thus not part of measured performance. Jumps could occur from any of these surfaces, but only those from the floor that occurred without any apparent stimulus or followed touching with a 100 µm diameter silver wire were used to determine jumping performance. Some jumps occurred from the front glass wall and were thus seen from underneath, allowing good views of angular changes of some key leg joints. Such jumps were excluded from analyses of jumping performance, either because the propulsive legs slipped on the glass surface or because the forward movement of the insect could not be measured. All other jumps in which there was any indication that the propulsive legs slipped were also excluded.

The camera pointed at the centre of a chamber. Most jumps were constrained by the shape of the chamber and were in the image plane of the camera. If the direction of the jump deviated from the image plane by more than 30 deg, then the error was obvious as changes in focus of the insect during take-off. Such videos were excluded from analysis. If the deviation was less than 30 deg, then calculations from trigonometry indicated that measures of angle or length would contain a maximum error of 10%. A further check by comparing real lengths against those recorded by the camera showed that distortions did not exceed this level in videos that were analysed. Tracks of the movements of specific body parts were made manually frame by frame with Tracker software (http://physlets.org/tracker/); auto-tracking was not used. The frame at which a particular leg lost contact with the ground was determined by playing the video backwards and forwards frame by frame. A shift in position of a tarsus indicated that the leg was no longer load-bearing and had therefore lost contact with the ground. Take-off occurred when the last leg lost contact with the ground and was designated as time (t)=0 ms. The acceleration time was defined as the period from the first detectable movement of the propulsive legs until take-off. Peak velocity was calculated from the distance moved in a rolling three-point average of successive images before take-off. A point on the body that could be recognized in successive frames and was close to the centre of mass (estimated by balancing the insect on a pin) was used for measurements of the trajectory. The angle subtended by a line joining these positions after take-off, relative to the natural horizontal, gave the trajectory angle. The body angle at take-off was defined as the angle subtended by the longitudinal axis of the body relative to the natural horizontal. The results are based on high-speed videos of 205 jumps by 22 P. communis (10 males and 12 females) at temperatures of 25–28°C. At least three jumps were analysed for each individual. Measurements are given as means±s.e.m. unless otherwise stated for an individual and as mean of means (grand means) for a particular sex.

Females had a mean body mass of 39.7±11.3 mg (N=10) and thus were significantly heavier (t-test: t₁₈=3.253, P=0.004) than males, which had a mean mass of 26.7±1.8 mg (N=10). Females had a mean body length of 15.1±1.1 mm (N=10) that was not significantly different (t₁₈=1.242, P=0.238) from that of males, which had a mean body length to the tip of the abdomen of 13.2±1.1 mm (N=10).

The mesothorax and metathorax formed a rigid structure supporting the middle and hind legs; the prothorax supporting the front legs was separately articulated (Hasken, 1939) (Fig. 1C). The three pairs of legs moved in two parallel planes on either side of the body. A front leg of females, from the trochanter to the tarsus, was the shortest at 9.1±0.5 mm, a middle leg was 11.5±0.9 mm and a hind leg was the longest at 14.1±1.1 mm (N=7) (Table 1). Comparable values in males were: front leg 7.3±0.5 mm, middle leg 8.5±0.7 mm and hind leg 11.6±0.8 mm (N=7). Relative to the front legs, this gave a ratio of mean leg lengths in females of 1:1.3:1.6 (front:middle:hind) and in males of 1:1.2:1.6. In females, the length of the front legs represented 61% of the body length, the middle legs 77% and the hind legs 98% (N=7). In males (N=7), the comparable values were 56, 65 and 89% of body length. Front, middle and hind legs differed significantly in length (ANOVA, F_2,36=17.05, P=6×10⁻⁶), and there was also a strong difference between males and females (F_1,36=13.80, P=6.9×10⁻⁴), with the longer and heavier females having longer legs than males. The lack of a significant sex×leg interaction (F_{2, 36}=0.32, P=0.727) indicates that the relative proportions of the legs remain the same in both sexes. The hind and middle coxae of both sexes were broader than the more distal segments of the legs and their articulation with the thorax allowed only small forward and backward rotations. The increased length of a hind leg was due to the femur, tibia and tarsus that were, respectively, 11, 40 and 35% longer than the equivalent parts of a middle leg. The femora and tibia of all the legs were tubes of similar diameter along their length and none were swollen to accommodate a greater volume of muscle that could make an increased contribution to jumping.

Both male and female P. communis used the same, single strategy for propelling take-off that involved fast movements of the middle and hind pairs of legs, accompanied by depression of both pairs wings. The following initial analysis is of jumping in females (Figs 2–5, Movies 1 and 2) and the similarity of the mechanism used by males is then analysed and illustrated (Figs 6 and 7, Movie 3). Of the 205 jumps analysed, 89.8% (184 jumps) occurred without any observable stimulus in the experimental chamber while just 10.2% (21 jumps) followed a touch to an antenna, a wing tip or a hind leg with a 100 µm diameter silver wire. The place on the insect that was touched resulted in a forward jump and did not influence the direction of a jump.

In preparation for a jump, the wings started to open 20–25 ms before take-off, followed by slow upwards and forwards movements of the middle and hind pairs of legs generated by levation of their coxo-trochanteral joints, and flexion of their femoro-tibial joints (Figs 2 and 3). In different jumps by the same or a different scorpion fly, these legs did not move to a constant position in advance of their propulsive movements. For example, a trochanter was not necessarily fully levated about its coxa and a tibia was not always fully flexed about its femur as observed either from the side (Fig. 2, Movie 1) or from underneath (Fig. 3, Movie 2). To quantify this variability, the angles adopted by a hind femur about the body and the tibia about the femur were measured just before these joints began to move in the propulsive movements of 109 jumps by 10 males and 10 females. The mean±s.d. coxo-trochanteral (body/femur) angle was 34.1±20.1 deg (range 4–90 deg) and the mean femoro-tibial angle was 68.4±25.9 deg (range 12−131 deg). From these variable initial positions, the start of the acceleration phase of a jump was marked by increasingly rapid movements of the hind femora and tibiae and by movements of the same joints of the middle legs, accompanied by a depression of the wings. In different jumps by the same or other individuals, the start of the wing movements was not tightly coupled to the movements of the legs; the wings could begin to move a few milliseconds before the legs, or at the same time (Figs 2 and 4). The leg movements started at a mean of 16.1±0.9 ms (N=10 females) and 19.0±1.0 ms (N=10 males) before take-off (Table 2). The time between these first propulsive leg movements and take-off defined the acceleration phase (take-off time) of a jump. There was no significant difference in these times between males and females (t-test: t₁₉=−1.602, P=0.126). The rapid depression and extension of these joints resulted in a progressive straightening of the two pairs of legs that raised the body from the ground and propelled it forwards (Fig. 4). In different jumps, the peak angular rotation of these joints ranged from 13,000 to 30,000 deg s⁻¹ (Fig. 5). In the majority of jumps, the movements of the middle and hind legs tracked each other closely (Fig. 5A), but for jumps in which the angular velocities were lower, the movements of the middle legs were of lower amplitude and were less closely coupled to those of the hind legs (Fig. 5B). The front legs showed no movements that were consistent with a contribution to the forward and upward propulsion of the body, and could leave the ground as early as 7 ms before take-off, and thus could not make any contribution to propulsion during the latter part of the acceleration phase. The middle legs were the next to lose contact with the ground some 4 ms before take-off (Figs 2 and 4B). The final propulsion was thus provided by the hind legs and by the wings, which completed their first cycle of depression before the insect became airborne. In some jumps, the wings had begun the first elevation phase of the wing beat cycle when take-off occurred (Figs 2 and 4).

Males used essentially the same pattern of leg and wing movements to propel take-off (Figs 6 and 7, Movie 3). In preparation for a jump, the wings were first opened and the middle and hind legs were levated at the coxo-trochanteral joints and flexed at the femoro-tibial joints. These movements again did not bring the middle and hind legs into a constant starting position for each jump. Depression of the wings and propulsive depression and extension movements of the middle and hind legs then began within a few milliseconds of each other (Fig. 6, Movie 3). The joints of the hind and middle legs progressively straightened and the wings depressed to power the forward and upward movement of the body to take off. The straightening of the hind legs was, as in females, more clearly seen when viewed from underneath (Fig. 7). The middle legs did not straighten as much. The front legs were the first to lose contact with the ground, followed by the middle legs, and finally the hind legs provided the last contribution to propulsion.

In neither females nor males were any take-offs observed in which leg movements alone, or wing movements alone, were the sole source of propulsion; all take-offs involved the combined actions of legs and wings.

Measurements taken from the high-speed videos and calculations based on these data enabled the jumping performance of the scorpion flies to be determined (Table 2).

The mean take-off velocity of females was 0.82±0.04 m s⁻¹ (best 1.05 m s⁻¹; N=10 insects) and of males was 0.73±0.03 m s⁻¹ (best 1.05 m s⁻¹; N=10). There was no significant difference in these mean take-off velocities between males and females (t-test: t₁₉=1.782, P=0.091).

To determine whether there was a difference in either acceleration time or take-off velocity between jumps that were evoked by a mechanical stimulus and those which occurred without any observable stimulus, the following test was performed. Twelve jumps in 12 animals (six males and six females) evoked by a mechanical stimulus were individually compared with 12 jumps by the same insects that occurred without any detectable stimulus (‘spontaneous’). Neither the acceleration time to take-off (touch-evoked 18.1±1.2 ms, ‘spontaneous’ 18.9±1.4 ms, paired t-test, t₁₁=0.44, P=0.670) nor the take-off velocity (mean touch-evoked 0.80±0.04 m s⁻¹, mean ‘spontaneous’ 0.78±0.04 m s⁻¹, paired t-test, t₁₁=2.2, P=0.44) were significantly different.

The mean body angle (measured as the angle of the longitudinal axis of the thorax relative to the natural horizontal) of females at take-off was 30.8±4.8 deg (N=10 insects) and of males was 35.4±4.1 deg (N=10). There was no significant difference in this angle between males and females (t-test: t₁₉=−0.818, P=0.423).

The mean trajectory angle during the first 5 ms when females were airborne was 51.5±5.6 deg (N=10 insects) and in males was 57.6±4.2 deg (N=10). There was no significant difference in the trajectories of males and females (t-test: t₁₉=−1.017, P=0.322). To determine the stability of the insect once airborne, changes in the longitudinal axis of the body in the pitch plane were measured in 63 jumps by 10 females and 10 males. Five milliseconds after take-off, the mean change of the pitch angle since take-off was 2.1±0.39 deg with no change detectable in 41 of those jumps and the maximum change being 10 deg. If this mean angular change were to continue longer into the trajectory it would lead to a spin rotation rate of 1.2 Hz, but this was not observed because during this time there was a smooth transition to flapping flight. The action of the four legs and the first downward movement of the wings thus provided the initial propulsion with a small pitch rotation of the body, but once the legs were clear of the ground, the subsequent flapping movements of the wings provided stability so that little energy was dissipated in rotating the body.

The potential contribution of wing movement to take-off was analysed by measuring the times to take-off and the take-off velocities in four jumps by each of five individual scorpion flies (three females and two males) both before and after the wings had been reduced in area by 25 to 50%. This reduced the mass of the insect by a mean of 5.8±0.8 mg (range 3.1 to 8.4 mg). The mean time to take-off was only slightly longer when the wings were clipped (mean 19.8±0.8 ms) compared with when the wings were intact (19.5±0.8 ms), and this difference was not significant (repeated-measures ANOVA, F_1,20=0.08, P=0.792). The mean take-off velocity was slightly lower in scorpion flies with clipped wings (0.69±0.06 m s⁻¹) compared with when they were intact (0.73±0.06 m s⁻¹), but this difference was again not significantly different (repeated-measures ANOVA, F_1,20=0.21, P=0.679).

The energy and power of the jumps was calculated from the means of the measured data. The energy required for a mean jump was 13.2 µJ in females and 7.1 µJ in males. The mean power was 0.8 mW in females and 0.4 mW in males (Table 2).

The mass of leg muscle that is used in jumping is assumed to be 20% of total body mass because two pairs of legs are used, and thus it is twice the value of the measured mass of muscle used by insects such as froghoppers (Burrows, 2007) or planthoppers (Burrows et al., 2014), which are propelled by just the hind legs. The power per kilogram of muscle needed to generate the mean take-off velocity in females was 103 W kg⁻¹ of muscle and in males was 70 W kg⁻¹ of muscle. In the jumps with the fastest take-off velocities, labelled as best jumps by an individual (Table 2), these values rose to 250 W kg⁻¹ of muscle in females and 170 W kg⁻¹ of muscle in males. The mass of muscles involved in take-off will also be increased by any participation of the wings, so further distributing the power demands and lowering the figures given above.

The high-speed videos show that scorpion flies are able jumpers that use a repeatable pattern of propulsive middle and hind leg movements and wing movements in individuals of both sexes. In preparation for a jump, the middle and hind legs were moved forwards and upwards by levation at their coxo-trochanteral and extension at their femoro-tibial joints. The starting angles of these leg joints were not always the same in each jump. In the propulsive acceleration phase of all jumps, the middle and hind pairs of legs were rapidly depressed and extended whilst the wings were depressing at the point of take-off. These movements propelled an insect in its best jumps to take-off velocities of just over 1 m s⁻¹. The jumps suggest that they might serve to move the insect quickly away from danger posed by a potential predator such as a bird or wasp, or to launch into stable flight to move to another site.

Three lines of evidence indicate that the jumping method used by scorpion flies relies on direct contraction of muscles rather than a catapult mechanism. First, in preparation for a jump, the joints of the propulsive legs do not adopt a constant starting position. This indicates that they are not moved to a position that favours a particular ratio of lever arms of the muscles that are moving a particular joint. Similarly, the legs are apparently not engaging a mechanical locking mechanism that would restrain a joint and allow energy to be stored; there is no evidence of a ‘latch’ as discussed by Ilton et al. (2018). At least one of these requirements would need to be met if a catapult mechanism was to engage. The contrast is therefore strong between the variable starting positions of these leg joints of the scorpion fly, using direct muscle contractions, compared with insects such as grasshoppers (Bennet-Clark, 1975) or froghoppers (Burrows, 2006), which use a catapult mechanism where the starting positions have always to be the same for loading the catapult. Second, the long duration of the acceleration phase suggests that the propulsive leg movements are not the result of the recoil of a spring loaded by the preceding storage of energy in a catapult mechanism. Third, calculation of the power requirements for a jump assumed that the muscles of the two pairs of legs used in propelling the jump comprised 20% of body mass, extrapolating from measurements (Burrows, 2007; Burrows et al., 2014) made in insects that use just one pair of legs for propulsion. The calculated power requirements fall well within the known maximal contractile limits of muscle (Askew and Marsh, 2002; Ellington, 1985; Josephson, 1993; Weis-Fogh and Alexander, 1977). The conclusion is that an explanation of jumping performance does not need to invoke power amplification through a mechanical energy storage mechanism. This does not, however, rule out the possibility that some energy generated by the muscles may be stored and then contributes to the jump. The power requirements for jumping in scorpion flies are much lower than those of insects that use a catapult; for example, in a siphonapteran insect, the hedgehog flea, the power requirements can reach 14,000 W kg⁻¹ of muscle, almost 28 times larger (Table 2) (Sutton and Burrows, 2011).

The kinematics indicate that take-off is associated with rapid propulsive movements of the middle and hind pairs of legs and the first cycle of depression by the four wings. Can the relative contribution of the legs and the wings be assessed? A resolution of this was attempted by clipping the wings of a group of scorpion flies and measuring jump performance before and after this procedure. No significant difference was found in the acceleration time or the take-off velocity of jumps before or after this procedure, implying that the wings add little to the thrust generated by the two pairs of legs. Reducing the area of the wing in this way will also reduce the volume of air that can be moved by the wings, an effect that may be offset by an approximately 9% reduction of body mass in these particular experiments. Overall, the conclusion that can be drawn is that the movements of the middle and hind legs are the main contributors to propulsion of take-off while the accompanying depression of the wings contributes to the smooth transition to flapping flight and thus stability to the take-off trajectory.

How does the mechanism used by scorpion flies compare with that used by the only other mecopteran known to jump, the snow flea, Boreus hyemalis (Burrows, 2011)? Both propel their jumps by rapid trochanteral depression and tibial extension movements of their middle and hind pairs of legs. In P. communis, the hind legs often moved before the middle legs whereas in snow fleas there was no discernible pattern as to which pair of legs moved first. The power for the depression of the middle and hind trochantera is provided in both species by large trochanteral depressor muscles in the mesothorax and metathorax so that the more distal segments of the legs can be thin and light with small muscles, allowing their more rapid acceleration. Boreus hyemalis does not have wings that can be moved so that they cannot provide further thrust to aid take-off. Both scorpion flies and snow fleas achieve best take-off velocities of 1 m s⁻¹ and the same mean velocities for all jumps of 0.8 m s⁻¹ in males and 0.7 m s⁻¹ in females. The initial trajectory of scorpion flies in the air propelled from a four-legged base imparted a small pitch rotation of 2 deg in the first 5 ms (400 deg s⁻¹) after take-off that was then stabilized by the flapping movements of the wings when airborne. In B. hyemalis without the stabilising addition of wing movements, the pitch rotation increases to 10−15 deg during the initial 6–7 ms (1900 deg s⁻¹) once airborne (Burrows, 2011). Both mecopteran examples, however, contrast with the jump of the wingless hedgehog flea (Siphonaptera), which is propelled only by the two hind legs and in which rotation of the body occurs at 10,000 deg s⁻¹ (Sutton and Burrows, 2011). Taken together, this is further evidence that the use of four legs for jump propulsion provides a more stable base for take-off.

Although jump performance of both insects is similar in terms of the take-off velocities, the acceleration time of P. communis is 16–19 ms and is thus almost three times longer than the 6.6 ms in B. hyemalis (Table 2). This may reflect the need to lift and accelerate their 10-times-greater mass, and to extend and accelerate their legs, which are three times longer than those of the snow flea. The biggest difference between these two insects is in the calculated power requirements for jumping and the implications that follow. In snow fleas, the fastest jumps require more power than could be delivered by direct muscle contractions. This has led to a suggested power-amplification mechanism in each of the middle and hind legs that could store energy generated by the contractions of their respective trochanteral depressor muscles. Supporting this suggestion is that in each of these four legs is a thoracic region containing the elastic protein resilin (Burrows, 2011). Resilin has been associated with energy storage mechanisms for jumping in diverse insect such as fleas (Siphonaptera) (Bennet-Clark and Lucey, 1967; Lyons et al., 2011), grasshoppers (Orthoptera) (Burrows and Sutton, 2012) and froghoppers (Hemiptera) (Burrows et al., 2008). The problem in snow fleas remains of how four catapult devices could be co-ordinated to release their stored energy at the same time. In P. communis, the power requirements for the fastest jump fall within the capability of normal muscle contractions so that no storage mechanisms are needed to explain the observed jumping performance. In keeping with this is that the search for resilin using specific wavelengths of ultraviolet light to illuminate each of the propulsive legs was not successful.

The scorpion fly analysed here joins a number of other arthropods that use two pairs of legs to propel jumping. Among insects are lacewings (Neuroptera, Chrysopidae) (Burrows and Dorosenko, 2014), praying mantises (Mantodea, Mantidae) (Sutton et al., 2016), mirid bugs (Hemiptera, Miridae) (Burrows and Dorosenko, 2017), caddis flies (Trichoptera) (Burrows and Dorosenko, 2015b), moths (Lepidoptera) (Burrows and Dorosenko, 2015a), true flies (Diptera) (Card, 2012), a few ants (Hymenoptera) (Baroni Urbani et al., 1994; Tautz et al., 1994) and some stick insects (Phasmida, Heteronemiidae) (Burrows and Morris, 2002), and among arachnids, a few spiders (Parry and Brown, 1959; Weihmann et al., 2010). The propulsive mechanisms in the insects above generate similar jumping performances to those of the scorpion fly; the take-off velocities all fall below 1 m s⁻¹ while acceleration times range from 6 to 100 ms. In comparison with insects using a catapult mechanism, the take-off velocities are lower and the acceleration times are longer. Of the insects that use four legs to jump, the only catapult mechanism suggested is in the snow flea (B. hyemalis), which has a maximum take-off velocity of 1.3 m s⁻¹ (Burrows, 2011) notably slower than most other catapult jumping insects. The fact that the mecopterans P. communis and B. hyemalis take off with relatively slow pitch rates of their body suggests the testable hypothesis that jumps by other insects propelled by four legs will also be more stable.

These data indicate two distinct categories of jumping insects with minimal overlap and suggest responses to different evolutionary pressures. A faster catapult mechanism is offset by the expense of building, maintaining and operating more specialized mechanisms that may also impinge on the many other functions that legs have to perform. A slower jumping mechanism involving the use of two pairs of propulsive legs runs the higher risk of predation, but this is offset by a more stable launch into flapping flight. The similarities revealed here between the jumping mechanism in a representative of the large family Panorpidae and the smaller family Boreidae adds a further characteristic that can be used in the continuing discussion about the phylogeny of the Mecoptera.

I thank Roger Northfield. Gabriel Jamie, Peter Lawrence and Pat Simpson for help in collecting the insects used in this study. Steve Rogers helped with the statistics. I am also grateful to University of Cambridge colleagues for their encouragement during the experimental work and for their comments on earlier drafts of the manuscript.