4607

The Journal of Experimental Biology 209, 4607-4621 Published by The Company of Biologists 2006 doi:10.1242/jeb.02539

Jumping performance of froghopper insects Malcolm Burrows Department of Zoology, University of Cambridge, Cambridge CB2 3EJ, UK e-mail: [email protected]

Accepted 6 September 2006

Summary The kinematics of jumping in froghopper insects were are in contact with the ground. In this position, the hind analysed from high speed sequences of images captured at legs stay motionless for 1–2·s. Both trochantera are then rates up to 8000·s–1. In a jump, the attitude of the body is synchronously and rapidly depressed about the coxae at set by the front and middle legs, and the propulsion is rotational velocities of 75·500·deg.·s–1 and the tibiae delivered by rapid and synchronous movements of the extended, to launch a jump that in Philaenus reaches a hind legs that are 1.5 times longer than the other legs, but height of 700·mm, or 115 body lengths. are only about half the length of the body and represent In the best jumps by Philaenus, take-off occurs within just 2% of the body mass. The wings are not moved and 0.875·ms of the start of movements of the hind legs at a the front and middle legs may be raised off the ground peak velocity of 4.7·m·s–1 and involves an acceleration of before take-off. The hind legs are first cocked by a slow 5400·m·s–2, equivalent to 550 times gravity. This jumping levation of the trochantera about the coxae so that the performance requires an energy output of 136·mJ, a power femora are pressed against the ventral, indented wall of output of 155·mW and exerts a force of 66·mN. the thorax, with the femoro-tibial joints tucked between the middle legs and body. Only the tips of the hind tarsi Key words: locomotion, kinematics, Hemiptera, Auchenorrhyncha.

Introduction allowing force to be delivered over a long period and over a Insects have evolved many different mechanisms for long distance. Bush crickets with very long hind legs (Burrows jumping so that they may increase the speed of their and Morris, 2003) have therefore adopted the same strategy as locomotion, launch themselves into flight, or escape rapidly frogs, kangaroos and bush babies in using direct muscle from a potential predator. The repeatable nature of these contractions to move long levers. Second, insects with short movements has enabled detailed analyses of the underlying legs such as fleas use a catapult mechanism in which muscles neuronal mechanisms (Burrows, 1996) and determination of are assumed to contract slowly and the force they generate is the mechanical and muscular solutions to these extreme stored in elastic elements of the skeleton and then suddenly locomotory demands. Click beetles (Elateridae) jack-knife a released (Bennet-Clark and Lucey, 1967; Rothschild and joint in their thorax (Evans, 1972; Evans, 1973), bristletails Schlein, 1975). Some insects, such as locusts, combine energy (Archaeognatha) (Evans, 1975), springtails (Collembola) storage and long legs (Bennet-Clark, 1975). (Brackenbury and Hunt, 1993) and the larvae of some flies A largely unexplored group of jumping insects are the (Maitland, 1992) use movements of their abdomens. Particular abundant and diverse plant sucking Hemipteran bugs, ants (Baroni et al., 1994; Tautz et al., 1994) and the stick insect belonging to the sub-order Auchenorrhyncha. One family of Sipyloidea sp. (Burrows and Morris, 2002) combine forward this group are the froghoppers (Cercopidae), which are movements of their abdomens with movements of their legs. common insects in many parts of the world. Their larvae In many other insects simultaneous movements of develop on plants, some species below and some above ground. specialised hind legs power jumping movements, though The latter secrete a froth, known colloquially in different muscles of different leg segments may be used in different countries as ‘cuckoo-’, ‘witches-’ or ‘frog-spit’, which may species. Fleas, for example, use the trochanteral depressor afford protection from desiccation and predation. Both larvae muscles to generate the necessary force whereas locusts and and adults feed on xylem sap, as a result often transmitting bush crickets use the tibial extensor muscles. Two design viruses between crop plants, but only the adults jump from principles emerge as different ways to overcome the need to plant to plant. Their name derives from the resemblance of their produce leg movements that are both rapid and powerful body shape to that of a frog and their prodigious jumping ability (Alexander, 1995). First, some insects have long hind legs that is the focus of this paper.

THE JOURNAL OF EXPERIMENTAL BIOLOGY 4608 M. Burrows

Table·1. Body form in froghoppers Length (mm) Ratio of leg lengths Length (% of body length) Body mass Hind leg, Hind leg, Insect N (mg) Body tibia femur Front Middle Hind Front Middle Hind Neophilaenus 7 3.2±0.08 4.0±0.03 1.2±0.01 0.8±0.02 1 1 1.5 43 42 66 Philaenus 34 12.3±0.74 6.1±0.08 1.8±0.07 1.1±0.03 1 1 1.5 46 46 66 Lepyronia 7 17.6±0.18 7.2±0.18 2.0±0.06 1.3±0.03 1 1 1.4 44 46 61 Aphrophora 23 28.3±1.1 9.8±0.24 2.5±0.09 1.3±0.04 1 1 1.5 36 36 52 Cercopis 16 32.9±1.0 9.5±0.13 2.9±0.04 1.6±0.03 1 1 1.6 43 44 63

A brief report on the kinematics of the jumping movements and near Ljubljana. The other species were collected near of a froghopper, Philaenus spumarius (Burrows, 2003), has Wells-next-the-sea and around Cambridge, UK. Observations demonstrated its jumping prowess, and a mechanism for on live insects were made on the same day of collection, or jumping has been proposed for Cercopis vulnerata (Gorb, after they had been in the laboratory for no more than a few 2004). This paper analyses the detailed jumping performance days feeding on live Chrysanthemum plants. of froghoppers and shows that in a jump they are airborne in Sequential images of jumps were captured at rates of 1000 less than 1·ms from the first propulsive movement of the hind or 2000·s–1 with a high speed camera (Redlake Imaging, legs. The enormous acceleration needed to achieve take-off San Diego, CA, USA) and associated computer, or at velocities of over 4.7·m·s–1 in this short time is equivalent to 4000–8000·s–1 with a Photron Fastcam 512 or 1024PCI camera 550·g. [Photron (Europe) Ltd, Marlow, Bucks., UK] and with exposure times of 0.05–0.25·ms. Spontaneous jumps, or jumps encouraged by delicate mechanical stimulation with a fine Materials and methods paintbrush or a 100·mm silver wire, were performed in a Five species of froghoppers were analysed: Aphrophora alni chamber of optical quality glass 80·mm wide, 80·mm tall (Fallén 1805), Cercopis vulnerata (Rossi 1807), Lepyronia and 25·mm deep with a floor of high density foam. Selected coleoptrata (Linnaeus 1758), Philaenus spumarius (Linnaeus image files were analysed with Motionscope camera software 1758), and Neophilaenus exclamationis (Thunberg 1784). (Redlake Imaging) or with Canvas (ACD Systems of America). Neophilaenus were collected near Wells-next-the-sea in Jumps were aligned by designating the point of take-off as time Norfolk, UK and Lepyronia from the Nanus region of Slovenia t=0·ms.

Transverse axis Longitudinal axis

80° 90° 155° 20° 40° Coxa Coxa 40° Coxa

Trochanter Trochanter Trochanter Femur Femur Tibia Femur Hind Middle Front leg leg Tibia leg 1 mm

Fig.·1. Drawing of the anterior part of Philaenus, viewed from the side, to show the orientation of the proximal parts of its three pairs of legs. Each leg is shown in its most anterior position (black) and in its most posterior position (grey). The pivots of the coxae with the thorax are indicated by black dots and vertical arrows. The plane of movement is not orthogonal to the plane of the drawing.

THE JOURNAL OF EXPERIMENTAL BIOLOGY Jumping performance of froghoppers 4609

Higher temporal resolution of the movements of a hind leg many small indentations and its ventral cuticle is ribbed. The of a restrained Aphrophora was obtained by gluing a 0.2·mm mouthparts point backwards and in Aphrophora extend to the disc of reflective tape to a hind femur close to the femoro-tibial coxae of the hind legs. The folded fore wings cover the body, joint. A modified single lens reflex camera with a concentric extend beyond the abdomen posteriorly and cover most of the light around the lens was focussed on the disc and the light it hind legs when viewed from the side. The five species of reflected was captured by a photocell in the film plane of the froghopper analysed have a tenfold range of body mass, from camera (Hedwig, 2000). This method recorded the movement 3.2±0.08·mg (N=7) in Neophilaenus to 32.9±1.0·mg (N=16) in of the femur and was combined with sequential images of the Cercopis (Table·1). Their body lengths have a 2.5 fold range hind legs captured by a high speed camera. from 4.0±0.03·mm (N=7) in Neophilaenus to 9.8±0.24·mm Seventy nine jumps by 19 Aphrophora, 92 jumps by 19 (N=23) in Aphrophora. Philaenus is toward the middle of this Philaenus, 47 jumps by 13 Cercopis, 8 jumps by 5 range with a body mass of 12.3±0.41·mg (N=34) and a body Neophilaenus and 16 jumps by 4 Lepyronia were captured and length of 6.1±0.08·mm. analysed. Data are given as means ± standard error of the mean The hind legs are only just over half the length of the body, (s.e.m.). Temperatures ranged from 24–30°C. ranging from 52.3±1.24% (N=23) of body length in Aphrophora to 66% in Philaenus and Neophilaenus (Table·1). In all species the front and middle legs are of similar length, Results but the hind legs are about one and half times longer so that Body form the ratio of leg lengths ranges from 1 (front):1 (middle):1.4–1.6 The head of froghoppers is flattened dorso-ventrally and has (hind legs) in the different species. The increased length of a short antennae. Its dorsal cuticle, and that of the prothorax, has hind leg is due to its longer tibia. By contrast, the femur of a

Fig.·2. Sequential images of a jump by Philaenus captured at 7500·s–1 and with an exposure time for each of 0.05·ms. The images are arranged vertically in two columns and are timed from the image at take-off (0·ms). The first movements of the right hind leg (white arrow) occurred 1.04·ms before take-off.

THE JOURNAL OF EXPERIMENTAL BIOLOGY 4610 M. Burrows hind leg is the same length and shape as those of the other legs. The coxae of the three pairs of legs articulate with the The mass of the two hind legs of Aphrophora, including the thorax at different angles (Fig.·1). In its most forward position trochanter and all the more distal segments, represents only the coxa of a hind leg subtends an angle of 155° to the 2.0±0.11% (N=7) of the total body mass. longitudinal axis of the body and rotates, as determined by imposed movements, backwards and upwards in one plane about its paired pivots by only a further 20–25°. Movements Philaenus of segments distal to the coxa are in this plane. By contrast, Take-off 6 in their most forward positions the coxae of the front and A Front and middle middle legs subtend angles of 80–90° and can rotate legs off ground backwards through an angle of about 40°, or almost twice the range of a hind leg. 4 4 Hind leg )

–1 Kinematics of the jump moves The same rapid movements of the hind legs propelled jumping by all species but the following analysis focuses on 2 2 Philaenus, with information from other species illustrating Hind wing (m s Velocity particular features. In preparation for a jump from a horizontal surface, the front of the body was raised or lowered by movements of the front 0 and middle legs to give a mean attitude of the body relative to 0 Front tarsus Hind tarsus Middle tarsus the ground at take-off of 28±1.9° (N=20). After adjustment of Hind femoro-tibial joint the body attitude was complete, the hind legs then remained –1.5 –1 –0.5 0 0.5 1 still for 1–2·s with only the distal tips of their tarsi in contact Time (ms) with the ground. A rapid and simultaneous depression of both hind legs then powered an explosive take-off. No differences could be detected in the timing of the movements of the two B hind legs and both left the ground at the same time. The first movement of a hind leg in a jump was a downward and backward thrust of the trochanter and femur (their 4 individual movements cannot be distinguished in these images viewed from the side) which, as transmitted through the tibia, forced the whole ventral surface of the tarsus to the ground (Fig.·2, Fig.·3A). Images captured at 8000·s–1 showed 2 that this first movement of a hind leg occurred only 0.875·ms

Vertical distance (mm)Vertical distance (mm) Vertical (7 frames) before the insect became airborne. The force from the continuing backwards movement of the hind legs began to lift the body because their tarsi were now directly applied 0 to the ground (Fig.·2, Fig.·3A,B). The body continued to be raised as the hind femora were further depressed and the hind 0246 tibiae were extended so that the tarsi of the front and middle Horizontal distance (mm) legs were raised from the ground before take-off. The velocity Fig.·3. Graphs of leg and body movements during a jump by Philaenus of the insect followed these movements of the legs. The first captured at 8000·s–1. (A) Five points on the body (see cartoon inset) surge in velocity corresponded to the initial movement of the are plotted against time. Take-off is indicated by the right arrow and femur (Fig.·3A) and after a short pause of 0.25·ms was vertical yellow bar. The first movement of a hind leg occurred followed by a rapid acceleration to a peak velocity of 0.875·ms before take-off (left arrow and yellow bar). The tarsi of the 4.7·m·s–1 at take-off. front and middle legs left the ground 0.5·ms before take-off (middle To resolve the movements of particular joints of the hind arrow and yellow bar). Velocity, measured as the movement of the legs during jumping, images of Aphrophora jumping were centre of an ellipse representing the overall shape of the body, is captured from a ventral view (Fig.·4). Inevitably this meant plotted as a two-point average of successive frames (blue line). (B) allowing it to jump from a transparent glass surface, with the Sequential movements of the five points on the body as the insect result that the hind legs gained little purchase and the whole moved through the field of view of the stationary camera, movement was completed in 0.4·ms or 2 frames at 5000·s–1. superimposed on an image of the froghopper in its starting position. The black arrowheads and the linking black lines show the position In preparation for a jump, the hind legs were levated at the of these five points at take-off. The corresponding positions of these coxo-trochanteral joints so that they were tucked between the points at different times during the jump can be read frame-by-frame femora of the middle legs and the thorax (Fig.·4A,B). The from these positions at take-off. tibiae were also flexed about the femora so that they lay

THE JOURNAL OF EXPERIMENTAL BIOLOGY Jumping performance of froghoppers 4611

Fig.·4. Jump by Aphrophora viewed ventrally and captured at 5000·s–1 with an exposure time of 0.05·ms. (A) Sequence of four images from the jump. (B) Drawings to show the joint positions of the hind legs before (–0.4·ms, fully levated) and after (0·ms, fully depressed) their rapid movements.

ventral to the abdomen along its lateral edges. The legs then levated and the tibiae flexed again to move the hind legs back remained stationary in this cocked position for a few seconds. into their cocked position. The first movements of the hind legs in a jump were the The same sequence of movements of the joints of a hind sudden depression of the trochantera about the coxae, most leg were also seen in Philaenus jumping from a horizontal notable as a closure of the gap between the trochantera at the position toward the camera and therefore moving out of its midline (Fig.·4A). This resulted in the femora moving focal plane. (Fig.·5A–C). The first movement of a hind leg posteriorly, without an apparent change in the angle of was a downward movement of the femur resulting from a trochantero-femoral joints and the tibiae extended about the depression of the trochanter about the coxa, accompanied by femora. This combination of joint movements continued and an extension of the tibia. With the tarsus pushed fully to the resulted in a full depression of the trochantera about the coxae ground, further depression of the femur and extension of the and a full extension of the tibiae about the femora. After these tibia resulted in an upward movement of the body. At take- movements were completed at take-off, the trochantera then off the coxo-trochanteral joint had been depressed through its

THE JOURNAL OF EXPERIMENTAL BIOLOGY 4612 M. Burrows

Fig.·5. (A) Sequential images captured at 5000·s–1 and with an exposure of 0.05·ms of Philaenus jumping toward and to the right of the camera to show the leg movements. Movements of the right hind femur are indicated by the arrows. (B) Graphs of changes in the angle between the femur and the longitudinal axis of the body and of the femoro-tibial joint (see inset drawings) in a jump by another Philaenus captured at 4000·s–1. (C) Movements of four points on the body (see cartoon) and of the angular changes of the femur and tibia (coloured lines). The changes in the femoro-tibial angle at the times indicated (ms) are shown in detail on the right.

THE JOURNAL OF EXPERIMENTAL BIOLOGY Jumping performance of froghoppers 4613

Fig.·6. Hind leg movements of a restrained Aphrophora recorded simultaneously by a camera capturing images at 1000·s–1 and with an exposure of 0.25·ms and by a photoelectric device (see Materials and methods). (A) Two sequential images, the first showing the hind legs fully levated before the attempted jump and the second after the rapid movement. The inset drawings show the position of the hind legs in these two frames. Note the small piece of reflective tape on the hind femur. The outputs of the photoelectric device during six jumps were captured at a sampling rate of 45·kHz and low-pass filtered at 2·kHz. One trace in blue shows the unfiltered recording. The leg movements were complete in 0.3·ms. (B) Seven attempted jumps by a second Aphrophora show that the movement was again complete in 0.3·ms.

full range at angular velocities of 75·500·deg.·s–1 and the was also achieved within 1·ms of the first movements of the femoro-tibial joint extended at an angular velocity of hind legs (Fig.·7A,B). The first and key movements of the hind 105·000·deg.·s–1. legs were again a rapid depression of the trochanter, with an Further detail of the joint movements was obtained by fixing accompanying extension of the tibia. Before take-off in some Aphrophora ventral surface uppermost in PlasticeneTM with the jumps, the tarsi of the front and middle legs had already lost hind legs free to move (Fig.·6). Rapid and simultaneous contact with the ground (Fig.·7B). movements of both hind legs occasionally occurred In the heaviest of the froghoppers, Cercopis, the body was spontaneously or could be evoked by gently tickling hairs on accelerated for a longer period to achieve take-off, with the the abdomen with a fine paintbrush. No differences in the form movements of the hind legs beginning 1.5·ms before take-off of these attempted jump movements could be discerned (Fig.·8A,B). The whole jumping sequence began with the compared with those in free jumping. They were, however, front and middle legs adjusting the attitude of the body, much faster and were completed in 0.3·ms because they did not which, in this example was only 16°. Both front and middle lift the mass of the body. The key movement was again a pairs of legs were again off the ground before take-off simultaneous depression of both trochantera about the coxae (Fig.·8B,C). Movements of the hind legs led to the head which occurred at 267·000·deg.·s-1, almost three times faster being raised while the posterior of the body was lowered, than in a real jump. The speed of these movements was giving a take-off angle of 45° despite the initial shallow body consistent in 6 attempted jumps by one Aphrophora and in 7 attitude. by a second (Fig.·6A,B). In some jumps when Lepyronia took off almost vertically the middle legs were already off the ground and the front legs Kinematics of the jump in other froghoppers were fully depressed and extended even before the first In the smallest of the froghoppers, Neophilaenus, take-off movements of the hind legs began (Fig.·9). At 1·ms before take-

THE JOURNAL OF EXPERIMENTAL BIOLOGY 4614 M. Burrows

Fig.·7. Images from two jumps by Neophilaenus captured at 2000·s–1 with an exposure of 0.25·s. (A) A jump in which the take-off occurred within 1·ms of the first movements of the hind legs: body angle, 36°; take-off angle, 55°; take-off velocity, 4.2m·s–1. (B) A jump toward and to the right of the camera in which the body was raised by the front and middle legs to assume a higher take-off angle: body angle, 50°; take-off angle, 72°; take-off velocity, 4.0·m·s–1; body mass, 3.2·mg. off, the front and middle legs were clear of the ground but the abdomen started to rotate forward about the transverse axis take-off velocity of 4.0·m·s–1 was, nevertheless, as great as that 10·ms after take-off so that it rather than the head pointed achieved at take-off angles closer to the mean of 45° when the forwards. In a third jump, the body began to rotate about its front and middle legs remained in contact with the ground for long axis after 10·ms and after 19·ms had rotated by 180° so a longer period. that the legs were pointed upwards. The rotation was completed 28·ms after take-off and then the next cycle of rotation began. Trajectories In a fourth jump, the body first started to rotate about its long Philaenus had a mean take-off angle of 46.8±2.0° (range 18° axis and then some 5·ms later also began to rotate about its to 90°, N=50) and a mode of 45° (Fig.·10A). In the first few transverse axis. milliseconds after take-off the insect typically maintained a In all species the wings remained folded during take-off, so stable orientation, and in many jumps this continued until a that the movement was a pure jump powered by the hind legs landing feet-first on a vertical or horizontal surface. In other and not assisted by active wing movements. Occasionally, jumps, however, the body rotated about its long or transverse however, the wings were opened after take-off, and flapping axes and occasionally about both axes (Fig.·10B). In the flight was assumed though this did not always lead to sustained example shown, Philaenus spun through four complete cycles flight or even to maintaining the height attained by the initial during the first 50·ms after take-off. In a second jump, the jump (Fig.·11).

THE JOURNAL OF EXPERIMENTAL BIOLOGY Jumping performance of froghoppers 4615

Fig.·8. Jump by Cercopis. (A) Sequential frames from the jump viewed from the side and captured at 2000·s–1 with an exposure of 0.25·ms. (B) Graph of the movements of six points on the body (see cartoon inset) during this jump. (C) Sequential movements of the same six points superimposed on an image of Cercopis in its starting position to show their vertical and horizontal displacement. The yellow arrows show the direction of the initial movements of the femoro-tibial joint. Body angle at take-off, 16°; take-off angle, 45°; take-off velocity, 3.8m·s–1; body mass, 41.9·mg.

Jumping performance achieving a take-off velocity of 4.7·m·s–1. In the other species, The jumping performance of these insects was calculated the take-off velocity in the best jumps when measured with the from the kinematic analysis of the jumping movements same image capture rates ranged from 4.6·m·s–1 in Lepyronia (Table·2). The take-off velocity was calculated as the mean to 3.4·m·s–1 in Aphrophora. The heavier insects generated the velocity over the first 3·ms when airborne. This will lower take-off velocities. The time over which the body was underestimate performance as peak velocity occurs just before accelerated was measured from the first visible movement of take-off and gradually declines thereafter (Fig.·3A). Such a the hind legs until the insect became airborne. In Philaenus calculation was necessary because images of most jumps were this acceleration period was only 0.875·ms and in captured at either 1000 or 2000·s–1, which did not record the Neophilaenus no more than 1·ms, but in Lepyronia, accelerations before take-off with sufficient resolution. In ten Aphrophora and Cercopis it was 1.5·ms. For the best jumps jumps by Philaenus the mean take-off velocity was this meant that the applied acceleration ranged from 2.8±0.1·m·s–1 with the best jumps captured at 8000·s–1, 2267–5400·m·s–2.

THE JOURNAL OF EXPERIMENTAL BIOLOGY 4616 M. Burrows

Fig.·9. Jump by Lepyronia. (A) Two columns of sequential images from a jump captured at 4000·s–1 and with an exposure of 0.25·ms. At the start, the body was raised at an angle of 61° to the ground so that only the tips of the tarsi of the front and middle legs were in contact with the ground. (B) Movements of five points on the body (see image in C) against time. (C) Movements of the same body points to show their vertical and horizontal displacement. The black curved arrows show the direction of the initial movements of the femoro-tibial joint. Body angle at take- off, 61°; take-off angle, 90°; take-off velocity, 4.0·m·s–1; body mass, 17·mg; temperature, 36°C.

The energy required to achieve this performance depends on jumping performance of these insects. The best indication of body mass. In the heavier species such as Cercopis the best ability came by taking the maximal performance of particular jumps required 238·mJ but in the much lighter Neophilaenus individuals, which under laboratory conditions and temperatures this fell to 28·mJ; Philaenus required 136·mJ. The power output may still be an underestimate. in a jump depends on the time during which the energy is expended. In the 0.875·ms that Philaenus took to accelerate its Walking body the power output was thus 155·mW. The force exerted The orientation of the hind legs, and their key role in powering during the best jumps by Philaenus was 66·mN. For the heavier jumping, raised the question of whether this compromised their Cercopis the force was highest at 83·mN and for the lighter ability to contribute to walking. The striking feature of horizontal Neophilaenus was lowest at 13·mN. walking was that the hind legs did not show rhythmic movements In a laboratory chamber at a temperature of 25°C and in still in the walking pattern and were not sequentially placed on the air, the average height jumped by Philaenus was 428±26·mm ground and then lifted (Fig.·12A). Instead they were held in the (N=17 insects) with the highest jumps reaching 700·mm, or 115 cocked position with the trochantera fully levated about the times its body length. None of the other species bettered these coxae so that the tarsi did not contact the ground. The hind legs performances; for example, in the same conditions Aphrophora were, however, used when climbing on a vertical surface on reached an average height of 263±20·mm (N=13). In a particular which there was limited traction (Fig.·12B). They moved individual, jumping performance declined with increasing rhythmically and were placed on the ground in time with the attempts to encourage jumping with the consequence that walking pattern so that they might therefore be expected to averaged values are likely to have underestimated the true contribute thrust to the movement.

THE JOURNAL OF EXPERIMENTAL BIOLOGY Jumping performance of froghoppers 4617

Philaenus 60 43° A 56° 35° 49°

40

Height (mm) 20 Take-off

0 –3 0 +10 20 30 Time (ms) 54 62 58 22 51 10 B 15 +10 10

5 0 Take-off axis (mm)

5 axis (mm) x y 0

0 –5 0204060Fig.·11. A jump in which Aphrophora flapped its wings after take-off. Time (ms) The selected images captured at 1000·s–1 and with an exposure of Fig.·10. (A) Trajectories of five jumps by the same Philaenus. Take- 0.25·ms are arranged in two columns. At take-off (time 0·ms) the off angles are the average over the first 10·ms after take-off. (B) wings were not opened. 5·ms later it started to lose height and 8·ms Rotation of the body during a single jump by Philaenus. A fixed point after take-off opened its hind wings and flapped them. on the head in the vertical or in the horizontal plane was plotted against time. The body spins about its longitudinal (y axis) and transverse (x) axis undergoing five cycles of rotation in the first 70·ms that it is airborne. Images were captured at 1000·s–1; body mass, 12·mg. muscle as in jumping insects such as grasshoppers, bush crickets (Tettigonidae) and flea beetles (Chrysomelidae, Halticinae). The overall design is therefore of a body that can be accelerated rapidly and which is able to transmit power to Discussion the ground through short but light hind legs. Body structure The hind legs of froghoppers are short relative to the body, Jumping performance ranging from only 52–66% of body length, and are only 1.4–1.6 The main thrust for jumping is provided by the hind legs. times longer, by virtue of longer tibiae, than the front or middle Any contribution from the front and middle legs is limited as legs in the different species. This contrasts with other they are often lifted from the ground before movements of the prodigious jumping insects such as locusts, in which the hind hind legs begin. Their primary responsibilities are therefore to legs are equal to body length, and most strongly with bush provide a stable platform and to set the angle of the body and crickets, in which the hind legs are longer than the body and the take-off trajectory, by raising or lowering the anterior part several times the length of the front legs (Burrows and Morris, of the body. The critical movement of the hind legs in 2003). This implies that the hind legs of froghoppers can generating the thrust for a jump is the depression of the provide much less leverage, a design that they share with fleas trochantera about the coxae. Before take-off, the hind legs are (Bennet-Clark and Lucey, 1967). The mass of the hind legs is levated forwards at the coxo-trochanteral joints so that that the also only a small (2%) proportion of body mass, contrasting femora are tucked between the thorax and the middle legs and again with locusts where it is 14% (Bennet-Clark, 1975). The are apposed to the lateral and ventral surfaces of the coxae. The hind femur is not enlarged to contain a powerful extensor tibiae tibiae are also flexed about the femora. By contrast, the front

THE JOURNAL OF EXPERIMENTAL BIOLOGY 4618 M. Burrows

Table·2. Jumping performance of froghoppers Take-off Body mass, Body length Time to velocity, Acceleration Energy Power Force –1 –2 Mb (mg) (mm) take-off (ms) v (m s )(m·s ) g force (mJ) (mW) (mN) Neophilaenus Best 3.2 4.0 1 4.2 4200 428 28 28 13 Philaenus Averagea 12.3 6.1 1 2.8 2800 286 48 48 34 Best 0. 875 4.7 5400 550 136 155 66 Lepyronia Averageb 17.6 7.2 1.5 4 2667 272 141 94 ·47 Best 4.6 3067 313 190 127 54 Aphrophora Averageb 28.3 9.6 1.5 2.5 1667 170 88 59 47 Best 3.4 2267 231 164 109 64 Cercopis Best 32.9 9.5 1.5 3.8 2533 258 238 158 83

a(N=10 jumps by 5 insects); b(N=10 jumps by 4 insects). N, number of jumps used to determine time to take-off and take-off velocity. 2 Formulae: acceleration f=v/t; g=f/9.86; energy e=0.5Mbv ; power=e/t; force=Mbf.

and middle tibiae are held in their most forward positions at jump suggests that its overriding objective is to move a angles of 80–90° to the longitudinal axis of the body. These froghopper from one position to another as rapidly as possible critical actions of the hind legs in jumping are at the expense at the expense of a controlled path through the air, or a of their ability to contribute to the propulsion of the body in controlled feet-first landing. horizontal walking. Which of the five species of froghoppers examined is the best From the start of the first visible movements of the hind legs jumper? The answer lies in which aspects of jumping to a froghopper becoming airborne takes no more than 0.875·ms performance are considered and how they are related to body in Philaenus and a maximum of 1.5·ms in the heavier Cercopis mass and volume. The five species of froghopper analysed have or Aphrophora. In this short time, the body is accelerated to a a tenfold range of body masses (3.2·mg in Neophilaenus to take-off velocity of 4.7·m·s–1 in the best jumps by Philaenus. In 32.9·mg in Cercopis), and vary in length from 4.0·mm in the best jumps by the different species, the applied acceleration Neophilaenus to 9.8·mm in Aphrophora. ranged from 2267–5400·m·s–2. Philaenus experiences the In terms of the height jumped then Philaenus comes out on equivalent of 550·g at take-off and the others from 231–428·g. top. Its average height jumped was 428·mm with the best jumps The best jumps by Philaenus require an energy output of 136·mJ, attaining heights of 700·mm, or 115 times its body length. By a power output of 155·mW and exert a force of 66·mN. These contrast, in the same conditions Aphrophora reached an forces and accelerations generated in jumping could not be average height of only 263·mm, or 27 times its body length. produced by direct contractions of the muscles and indicate that Distance and height achieved will be determined by take-off muscular force must be generated in advance of the movement, velocity, take-off angle and by the drag. All the froghoppers energy stored and then released rapidly. achieve high take-off velocities ranging from 3.4 to 4.7·m·s–1 None of the five species of froghoppers were captured and average take-off angles are close to 45°. Drag will, opening their wings before take-off or flapping them to assist however, be different because of the different body sizes and take-off. Indeed, the high accelerations and velocities at take- masses (Bennet-Clark and Alder, 1979). The distance lost due off may preclude opening the wings. Occasionally Aphrophora to drag by Philaenus is estimated to be about 25% (Vogel, and more frequently Cercopis opened their wings when they 2005) based on my data. The smaller Neophilaenus would were airborne and then flapped them in a flight pattern. be expected to experience greater drag while the larger Flapping the wings after take-off could presumably generate froghoppers should experience less. further lift or forward momentum, but could also act as an air In terms of velocity, acceleration and force relative to body brake to stabilise the movements and increase the likelihood of mass generated at take-off, then Philaenus again comes out on a soft landing. The hind legs were held fully extended after top. It accelerates its body in less than 1·ms to achieve an take-off so that adjustments of their posture could provide some average velocity over the first 3·ms of the jump of 4.7·m·s–1. ruddering control. The body may nevertheless still rotate about Both Neophilaenus and Lepyronia approach but never better its longitudinal and transverse axes. These characteristics of a these velocities at take-off, but the heavier Aphrophora and

THE JOURNAL OF EXPERIMENTAL BIOLOGY Jumping performance of froghoppers 4619

A Horizontal walk Philaenus

131 ms Right 190 ms 256 ms Hind

Middle

Front Stance Swing Left Hind Middle Front

0 200 400 600 800 1000 Time (ms) B Vertical walk

557 ms 586 ms 684 ms Right Swing Stance Hind

Middle

Front

Left Hind

Middle Front

0 200 400 600 800 1000 Time (ms)

Fig.·12. Contribution of the hind legs of Philaenus in walking. Images were captured at 1000·s–1 with an exposure of 0.5·ms. Selected images, viewed ventrally, are shown at the times indicated. (A) Horizontal walking. The middle and front legs were lifted in a sequence to perform a swing phase of similar duration (thick black bars) and contact the ground during a stance phase of variable duration (thin blue bars). The hind legs did not move. (B) Vertical walking in which the hind legs did contribute.

Cercopis both take longer (1.5·ms) to accelerate their bodies experienced by Philaenus (Vogel, 2005). Heavier Orthopteran and achieve lower velocities. insects such as locusts (Schistocerca gregaria) with a mass of 1–2·g take 20–30·ms to extend their hind legs and accelerate Jumping performance relative to other animals their body (Brown, 1967) to a take-off velocity of 3·m·s–1 Fleas have been considered the best jumpers amongst the (Bennet-Clark, 1975) while Prosarthria teretrirostris with a insects, accelerating their body within 1·ms to a take-off mass of 280·mg takes 30·ms of acceleration to achieve a take- velocity of 1·m·s–1 (Bennet-Clark and Lucey, 1967; Rothschild off velocity of 2.5·m·s–1 (Burrows and Wolf, 2002). The et al., 1975; Rothschild et al., 1972). Froghoppers produce jumping distance of these larger insects is likely to be curtailed a substantially better jumping performance. Philaenus by only some 6% due to drag (Vogel, 2005). If jumping accelerates its body to a take-off velocity that is more than 4.7 performance is expressed as the force exerted relative to body times faster than a flea despite having a body mass that is 27 mass, then froghoppers again outperform other insects and times greater and a body that is four times longer. Once other jumping animals. The force that froghoppers exert at airborne, however, the flea is likely to have its jumping distance take-off is more than 400 times their body weight and is, reduced by 80% due to drag compared to the 25% reduction therefore, much higher than in other jumpers such as the flea

THE JOURNAL OF EXPERIMENTAL BIOLOGY 4620 M. Burrows (~135 times) (Bennet-Clark and Lucey, 1967), locust (~8) performance is analysed in two subsequent papers (Burrows, (Bennet-Clark, 1975) and humans (~2–3) (Dowling and 2006; Burrows, 2007) and further papers will explore the Vamos, 1993). evolution of the particular jumping mechanisms in other families of these plant sucking bugs. Biology of the jump What do froghoppers gain by investing so much in their I thank my Cambridge colleagues for their help in prodigious jumping performance? Is it simply a way of collecting these bugs, for their constructive suggestions during improving locomotion so that a froghopper can move quickly the course of this work, and for their comments on the from one food plant to another without being spotted? Is it to manuscript. This work was initiated at the Wells Field Study avoid predation or being parasitized? Only the adults jump. The Centre, Wells-next-the-sea, Norfolk, England during an nymphs of Philaenus surround themselves with froth that is undergraduate field course, and I am most grateful to the generated by blowing air into secretions from their Malpighian warden Christine Marshall for the use of these facilities. Some tubules. They do not jump if their frothy surroundings are experiments were also carried out at the Department of removed. Jumping is thus associated with the free-living life Entomology, National Institute of Biology, Ljubljana, style of adults, but very little is known of what determines Slovenia and I am most grateful to Dr Meta Virant and her movements from one host plant to another, and what dangers colleagues for their support. might be posed by other animals. The adults generally feed on the underside of leaves but the aposematically coloured Cercopis feeds more frequently on the more exposed stems of References plants. Alexander, R. M. (1995). Leg design and jumping technique for humans, other Potential predators of froghoppers are many and include vertebrates and insects. Philos. Trans. R. Soc. Lond. B Biol. Sci. 347, 235- 248. birds, solitary wasps that provision their nests with Baroni, U. C., Boyan, G. S., Blarer, A., Billen, J. and Musthak, A. T. M. froghoppers, and predatory social wasps or flies. Parasitoids (1994). A novel mechanism for jumping in the Indian ant Harpegnathos such as Pipunculidae attack the pre-adult stages which are saltator (Jerdon) (Formicidae, Ponerinae). Experientia 50, 63-71. Bennet-Clark, H. C. (1975). The energetics of the jump of the locust unable to jump. A further major danger may be unwitting Schistocerca gregaria. J. Exp. Biol. 63, 53-83. predation by grazing mammals. They share this danger with all Bennet-Clark, H. C. and Alder, G. M. (1979). The effect of air resistance on the other insects that live or feed on vegetation, but a rapid and the jumping performance of insects. J. Exp. Biol. 82, 105-121. Bennet-Clark, H. C. and Lucey, E. C. A. (1967). The jump of the flea: a long jump out of harm’s way becomes advantageous. It may study of the energetics and a model of the mechanism. J. Exp. Biol. 47, 59- be that froghoppers are vulnerable while airborne and thus need 76. to minimise exposure time in the air by jumping rapidly. This Brackenbury, J. and Hunt, H. (1993). Jumping in springtails: mechanism and dynamics. J. Zool. Lond. 229, 217-236. assessment of the value of jumping, however, poses further Brown, R. H. J. (1967). The mechanism of locust jumping. Nature 214, 939. questions. Burrows, M. (1996). The Neurobiology of an Insect Brain. Oxford: Oxford First, at what distance and with what sense does a froghopper University Press. Burrows, M. (2003). Froghopper insects leap to new heights. Nature 424, 509. detect an approaching predator? A vibratory sense could give Burrows, M. (2006). Morphology and action of the hind leg joints controlling advanced warning of an approaching danger by detecting jumping in froghopper insects. J. Exp. Biol. 209, 4622-4637. footfalls or movements of the plant on which it is feeding. This Burrows, M. (2007). Neural control and co-ordination of jumping in froghopper insects. J. Neurophysiol. In press. would allow the necessary time for developing the forces Burrows, M. and Morris, O. (2002). Jumping in a winged stick insect. J. Exp. needed to jump (Burrows, 2007). Related families of Biol. 205, 2399-2412. Auchenorrhyncha use this sense to communicate with each Burrows, M. and Morris, O. (2003). Jumping and kicking in bush crickets. J. Exp. Biol. 206, 1035-1049. other on the same plant (Claridge, 1985; Cocroft et al., 2000; Burrows, M. and Wolf, H. (2002). Jumping and kicking in the false stick Cokl and Virant-Doberlet, 2003) and Cercopis appears to insect Prosarthria: kinematics and neural control. J. Exp. Biol. 205, 1519- signal by vibrating its wings while remaining stationary on a 1530. Claridge, M. F. (1985). Acoustic signals in the Homoptera: behavior, plant (Kehlmaier, 2000). taxonomy, and evolution. Annu. Rev. Entomol. 30, 297-317. Second, does a froghopper have to anticipate the possible Cocroft, R. B., Tieu, T. D., Hoy, R. R. and Miles, R. N. (2000). need to jump and thus hold its hind legs in readiness? This Directionality in the mechanical response to substrate vibration in a treehopper (Hemiptera: Membracidae: Umbonia crassicornis). J. Comp. would explain the cocked position adopted by the hind legs Physiol. A 186, 695-705. during walking. Cokl, A. and Virant-Doberlet, M. (2003). Communication with substrate- Third, how quickly can a froghopper unplug its stylets? Are borne signals in small plant-dwelling insects. Annu. Rev. Entomol. 48, 29- 50. the stylets withdrawn before take-off or does the act of jumping Dowling, J. J. and Vamos, L. (1993). Identification of kinetic and temporal merely rip them from the plant? factors related to vertical jump performance. J. Appl. Biomech. 9, 95-110. Fourth, does a froghopper orient itself out of the line of Evans, M. E. G. (1972). The jump of the click beetle (Coleoptera: Elateridae) – a preliminary study. J. Zool. Lond. 167, 319-336. approach of a predator before it takes off, or is the direction of Evans, M. E. G. (1973). The jump of the click beetle (Coleoptera, Elateridae) its jump determined by the way it was facing when feeding and – energetics and mechanics. J. Zool. Lond. 169, 181-194. had to unplug its stylets? Evans, M. E. G. (1975). The jump of Petrobius (Thysanura, Machilidae). J. Zool. Lond. 176, 49-65. The structural specialisations of the joints and the sequence Gorb, S. N. (2004). The jumping mechanism of the cicada Cercopis vulnerata of muscle actions that enable this remarkable jumping (Auchenorrhyncha, Cercopidae): skeleton-muscle organisation, frictional

THE JOURNAL OF EXPERIMENTAL BIOLOGY Jumping performance of froghoppers 4621

surfaces, and inverse-kinematic model of leg movements. Arthropod Struct. cheopis. Exoskeletal structures and musculature. Philos. Trans. R. Soc. Dev. 33, 210-220. Lond. B Biol. Sci. 271, 457-490. Hedwig, B. (2000). A highly sensitive opto-electronic system for the Rothschild, M., Schlein, Y., Parker, K. and Sternberg, S. (1972). Jump of measurement of movements. J. Neurosci. Methods 100, 165-171. the Oriental Rat Flea Xenopsylla cheopis (Roths.). Nature 239, 45-47. Kehlmaier, C. (2000). Zu Verbreitung, Lebenszyklus und Gesang der Rothschild, M., Schlein, J., Parker, K., Neville, C. and Sternberg, S. (1975). Blutzikade Cercopis vulnerata ROSSI, 1807 (Auchenorrhyncha: The jumping mechanism of Xenopsylla cheopis. III. Execution of the jump Cercopidae) am Lüneburger Schildstein (Niedersachsen). Braunschw. and activity. Philos. Trans. R. Soc. Lond. B Biol. Sci. 271, 499-515. Naturkdl. Schr. 6, 69-84. Tautz, J., Holldobler, B. and Danker, T. (1994). The ants that jump: different Maitland, D. P. (1992). Locomotion by jumping in the Mediterranean fruit- techniques to take off. Zoology 98, 1-6. fly larva Ceratitis capitata. Nature 355, 159-161. Vogel, S. (2005). Living in a physical world. II. The bio-ballistics of small Rothschild, M. and Schlein, J. (1975). The jumping mechanism of Xenopsylla projectiles. J. Biosci. 30, 167-175.

THE JOURNAL OF EXPERIMENTAL BIOLOGY