Sexual dimorphism in jump kinematics and choreography in peacock spiders

To execute a jump, animals require a quick burst of energy to launch themselves from a substrate, during which they coordinate various body parts to move synchronously within a brief period. Jumping is an energetically expensive form of locomotion (Burrows, 2006; Krasnov et al., 2004) that rapidly moves an animal either towards or away from a target. Many taxa have developed distinct strategies for jumping or leaping, including mammals, marsupials, amphibians and arthropods (Burrows et al., 2021; Delciellos and Vieira, 2009; Juarez et al., 2023; Kiørboe et al., 2010; Moore and Clifton, 2023). Animal jumps are classified as muscle-actuated or spring-actuated systems based on how they are powered. In arthropods, muscle-actuated jumps are seen in ants (Aibekova et al., 2023), lacewings (Burrows and Dorosenko, 2014) and bush crickets (Burrows and Morris, 2003), where animals rely on the power generated by the contraction of leg muscles (e.g. muscles in the coxae, trochanter and tibia). The volume of the muscle and its physiological properties limit the jump performance (Aibekova et al., 2023; Sutton et al., 2016). Spring-actuated systems, also referred to as catapult systems or power-amplification systems, as seen in grasshoppers, click beetles, true bugs and fleas, bypass muscular limitations and output more power by relying on internal structures to store energy, which is then rapidly released to trigger the jump (Bolmin et al., 2022; Patek et al., 2011).

Jumping spiders of the family Salticidae are exceptional jumpers, as their name suggests. They leap to cross gaps, navigate detours, capture prey and evade predators (Aguilar-Arguello et al., 2020; Bolmin et al., 2022; Brandt et al., 2021; Nabawy et al., 2018; Nelson, 2024; Patek et al., 2011). Salticids jump to targets at different elevations and often cross gaps 10–15 times their body length (Nelson, 2024). Spider jumps are neither muscle nor spring actuated; instead, they use an unusual strategy referred to as the semi-hydraulic system (Brandt et al., 2021; Nentwig et al., 2022; Parry and Brown, 1959a; Weihmann et al., 2012). Spider legs have flexor muscles for postural movements but lack the extensor muscles required to extend their legs and power jumps. The legs are extended by increasing the haemolymph pressure within them. This pressure is generated by the contraction of muscles in the cephalothorax, which pumps haemolymph into the leg. Currently, though jumps have been described in five genera of Salticidae – Attulus, Habronattus, Phidippus, Portia and Trite (Aguilar-Arguello et al., 2021; Brandt et al., 2021; Hill, 2006; Nabawy et al., 2018; Parry and Brown, 1959b) – sex-specific differences in jump kinematics have not been investigated. Male and female jumping spiders experience different selective pressures. Similar to many other spiders, females tend to have larger abdomens as they must accumulate resources for egg production. More uniquely, male salticids use leg waving as part of a courtship display and also to rapidly escape from females and male competitors. These differences are likely to affect locomotion in distinct ways in male and female spiders.

Here, we investigated the jump kinematics and choreography of males and females of the Australian splendid peacock spider, Maratus splendens (Fig. 1A,B). Peacock spiders are native to Australia, with males renowned for their vibrant iridescent colours and elaborate courtship displays (Girard and Endler, 2014; Stavenga et al., 2016). During courtship, males of Maratus species spread open their brightly coloured opisthosomal flaps, which are usually folded around the abdomen, and oscillate around a female while waving the ornamented third pair of legs (Girard and Endler, 2014; Girard et al., 2021; Hill and Otto, 2011). Using high-speed videography, we studied locomotory jumps and determined the jump kinematics and choreography in male and female spiders.

We collected adult males and females of the Splendid Peacock spider, Maratus splendens (Rainbow, 1896) at the Lake Paramatta Reserve, Sydney, NSW, Australia. Spiders were individually housed in containers and provided with water and food (Drosophila melanogaster). Experiments were carried out between October and December 2023 at the Wallumattagal campus, Macquarie University.

Experiments were carried out in the laboratory with flicker-free LED lighting, and constant temperature conditions of 24°C. We fixed a landing and take-off platform to an optical bench which was placed on a vibration isolation table. The platforms were 3D printed with black PLA using Ultimaker S3. The take-off platform (4.0×0.5×0.1 cm, L×W×H) was horizontal to the ground on which a spider walked on their own from a collection jar. We placed a vertical landing platform (0.1×0.5×4.0 cm, L×W×H) 4 cm away from the take-off platform. Individual spiders jumped from the take-off to the landing platform of their own accord (Fig. 2). Thus, jumps were not elicited but allowed to happen naturally. We removed all other visual targets in the vicinity, which ensured that spiders carried out locomotory jumps to the landing platform. Most spiders jumped immediately, whereas some took a few minutes to jump towards the landing platform. For each individual, we recorded three consecutive jumps. The one exception was a female spider for which we recorded only two jumps, as this animal was reluctant to jump to the target on its third attempt. Animals were weighed immediately after they carried out the jumps (Sartorius Entris ii Advanced line Balance, Göttingen, Germany, 0.1 mg repeatability) and subsequently released back into their natural habitat.

We filmed the jumps with a 105 mm Sigma lens attached to a high-speed camera (T1340, Phantom, Adept Turnkey Pty Ltd, Osborne Park, WA, Australia) at 5000 frames s⁻¹ (2048×1024 pixels). The spider and filming area were illuminated with Godox SL 300III light source. We filmed either the entire jump trajectory (male: n=4 spiders, 12 jumps; female: n=7 spiders, 20 jumps), or magnified views of take-off jumps to track leg joints accurately (male: n=6 spiders, 18 jumps; female: n=5 spiders, 15 jumps).

Raw footage was converted from cine to uncompressed mp4 files and trimmed in Videoloupe (Corduroy Code Inc., Vancouver, BC, Canada). Trimmed files were imported to DLTdv8 (Hedrick, 2008) in Matlab 2023b (Mathworks, Natick, MA, USA) where we carried out a manual frame-by-frame analysis to track different body structures and extract x,y coordinates. In both full-jumps and take-off-only jumps, we tracked the centre of mass (CoM; see below). From these coordinates, we reconstructed the jump trajectory. In full jumps, we identified the maximum height attained during a jump and the height above ground when animals reached the landing platform. We used the displacement of the CoM to determine jump kinematics, which included: (i) take-off duration (ms) – the duration between the first instance when acceleration increased from zero (i.e. the first frame where there was movement) to the first instance when all legs were off the ground. All kinematic measurements below were calculated for the same period as the take-off duration and maximum kinematic values for each jump was identified: (ii) take-off velocity (m s⁻¹), v=d/t, where d is distance and t is time, determined for every frame; maximum take-off velocity was identified and used to calculate the following variables; (iii) acceleration (m s⁻²), a=Δv/Δt, where Δv is change in velocity and Δt is change in time, determined for every frame; peak acceleration was identified and used to calculate the following variables; (iv) kinetic energy (µJ), ke=½mv², where m is body mass and v is velocity; jump force (mN), F=m×a, where m is body mass and a is acceleration; (v) jump power (mW), P=ke/take-off duration; (vi) g-force, g=a/9.81; and (vii) take-off angle (deg): the angle of the CoM between the first frame prior to when the abdomen moves and the first frame when all legs are off the ground, relative to the plane of the substrate.

To identify the contribution of different legs in a jump, we tracked four leg joints in each leg in one randomly selected jump of all male (n=6) and female spiders (n=5) from the high-magnification footage. We tracked the following joints: trochanto-femoral, femoro-patellar, tibio-metatarsal and tarsal claw (Fig. 1E). Over the course of the jump, we determined leg extension using two methods, both adapted from Brandt et al. (2021). In the first, we calculated the effective leg length (ELL), which was the absolute distance between the trochanter and tarsal claw divided by the sum of length of all segments. Values range between 0 and 1, where values close to 1 indicate a fully extended leg. In the second, we determined the flex or the leg joint angle, i.e. the angle formed by the trochanto-femoral, femoro-patellar and patellar–tibial joints (Fig. 1E). Lastly, we identified the time difference between the moment the animals reached peak acceleration and the instant all their legs completely left the ground.

Specimens were fixed in 4% paraformaldehyde for 4 h and then rinsed in 0.1 mol l⁻¹ phosphate buffer (3×10 min). Specimens were stained in 2% Lugol's solution for 48 h followed by washes in phosphate buffer (3×10 min). Samples were individually placed in small Eppendorf tubes filled with 70% ethanol for imaging for micro-computed tomography (µCT). We imaged the samples with a Bruker SkyScan1272 µCT scanner using a voltage of 45 kV, a current of 200 μA and a 0.5 mm aluminium filter. Projection images were reconstructed using SkyScan's NRecon v.2.2.0.6 software (Bruker, Kontich, Belgium) and data were output as a bitmap stack. Scans had a voxel resolution of between 1.5 and 5.0 µm.

The µCT image stacks were imported into Fiji (Schindelin et al., 2012). Scale information was added and a binning transformation (averaging) was applied to the x,y,z axes to halve the resolution in each axis to reduce the file size. Stacks were then exported as a NIfTI file. This file was opened in ITK-snap v.4.0.0 (Yushkevich et al., 2006) where we used the thresholding user guided auto-segmentation tool to segment the images. The resulting segmentation was manually refined to ensure there were no interior gaps and we assumed that the cephalothorax and abdomen were homogeneous and solid for CoM calculations. The legs were additionally separated into a different label as they were not included in the CoM calculation. Finally, a volume mesh was generated and exported as an STL file. The mesh was imported into MeshLab v.2022.02 (Cignoni et al., 2008), where the mesh was simplified using quadratic edge collapse decimation, isolated faces were removed and non-manifold edges were repaired before re-exporting as an STL file again. The cleaned mesh was loaded into AutoDesk Fusion 360 v.2.0.17721 where the CoM was identified.

We used circular statistics, the Watson–Williams test in Oriana (Kovach Computing Services), to determine whether the mean take-off angles differed between male and female spiders.

We used linear mixed-effects models to examine the relationship between the sex of the spider, body mass and kinematic variables. The modelling was carried out in R using the ‘lme4’ and ‘lmerTest’ packages. The initial analyses involved fitting a full model with the sex of the spider and body mass as fixed effects. In addition, the model included an interaction term (body mass×sex) to evaluate the combined effect of these two factors. Individual spider IDs were used as random effects. The second model was similar but without the interaction term, assuming the relationship between the body mass and the kinematic variable does not differ by sex. The final model was determined based on Akaike's information criterion (AIC) by comparing the fits of the two models (Table S1). After selecting the optimal random effects' structure, the significance of the fixed effects was assessed using the F-test with Satterthwaite correction as implemented in the ‘lmerTest’ package.

Males of M. splendens were distinctly lighter compared with their female counterparts (male: 4.94±1.86 mg, n=10; female: 10.40±2.05 mg, n=12; means±s.d.; Mann–Whitney, U=3.0, P<0.001; Table 1). The CoM in the male was located on the cephalothorax, just above the fourth coxae (Fig. 1C). The CoM in the female was located on the anterior-most region of the abdomen (Fig. 1D).

We studied the jump kinematics of 10 male (30 jumps) and 12 female spiders (35 jumps) and on every occasion, all animals jumped across the 4 cm gap to successfully reach the landing platform (Fig. 3).

The smallest individual filmed was a 2.0 mg male that had a take-off duration of 22 ms, peak acceleration of 141.80 m s⁻² and maximum take-off velocity of 0.89 m s⁻¹. In comparison, the heaviest individual filmed was a 13.2 mg female that had a take-off duration of 14 ms, peak acceleration of 122.33 m s⁻² and a maximum take-off velocity of 0.76 m s⁻¹. Between the two sexes, the highest and lowest acceleration was seen in female spiders (155.22 and 90.06 m s⁻²). Between the sexes, the highest take-off velocity was in males (0.98 m s⁻¹) and the lowest was in females (0.61 m s⁻¹). There was a noticeable overlap in body mass between the sexes, where the heaviest males were heavier than the lightest females.

Body mass had a significant effect on all the kinematic variables we measured (Fig. 4, Table 2). Sex of the spider had a significant effect on all kinematic variables except take-off velocity (Fig. 4, Table 2). After controlling for body mass, we found that all kinematic variables, with the exception of jump power, were lower in males compared with females (Table 2). We found a significant interaction between sex and body mass for all kinematic variables except take-off velocity and kinetic energy (Fig. 4, Table 2).

Males had a significantly higher take-off angle compared with females (males: 25.08±8.55 deg, r=0.98, n=30; females: 20.89±6.25 deg, r=0.99, n=35; means±s.d.; length of mean vector; Watson–William test F=5.011, P=0.029; Figs 3B,C, 4, Table 1). From full jump trajectories, we measured the maximum height of the trajectory and trajectory height when spiders reached the landing platform (Fig. 2) and found no differences between sexes (Mann–Whitney: maximum trajectory height: U=98.0, P=0.408; trajectory end height: U=98.0, P=0.146; Fig. 3D,E).

The sequence of leg and body movements between the male and female spiders was similar. Prior to jumping, spiders of both sexes raised and extended their first two pairs of legs (Figs 2 and 5; Movie 1). Spiders waved their first two pairs of legs for 0.4–60 ms, and gradually lowered their abdomen. The tip of the abdomen touched the take-off platform briefly (∼0.2 ms; equivalent to our interframe interval), presumably to attach a drag line thread. The fourth pair of legs left the take-off platform first, followed by the third pair (Fig. 5; Movie 1). In females, the third pair of legs left the platform 3.5±0.4 ms (means±s.d., n=35) after the fourth pair, whereas in males it was slightly delayed (3.8±1.8 ms, n=30; Mann–Whitney U=216, P<0.001; Fig. 5; Movie 1).

We identified the duration between the moment the animals reached peak acceleration and the instant all their legs completely left the take-off platform. In both sexes, the fourth pair of legs left the platform close to when peak acceleration occurred (females: 0.4±0.2 ms after peak acceleration, means±s.d., n=35; males: 0.3±0.6 ms before peak acceleration, n=30). This was true for almost all jumps except one jump in a male (out of 30) and a female (out of 35). In both sexes, the third pair of legs left the platform after peak acceleration (Fig. 5; Movie 1; males: 3.4±1.4 ms after peak acceleration, n=30; females: 4.0±1.5 ms after peak acceleration, n=35).

In a smaller sample size (6 males, 5 females, 1 jump each), we determined the extension and the timing of leg movement by calculating the ELL and angle of the leg joint relative to acceleration (Figs. 1E and 6). The ELL of leg IV in males and females gradually increased and was fully extended before animals reached peak acceleration (with one exception in a male and a female; Figs 1E and 6). In both sexes, maximum ELL and the least acute angle in leg III occurred after peak acceleration (Fig. 6).

Male jumping spiders must execute rapid jumps to escape attacks from female and male competitors during courtship and to avoid being preyed upon. Hence, we focused on characterising sex-specific differences in the kinematics and choreography of jumps in the Australian splendid peacock spider M. splendens. Though we studied only mature adults, the body mass varied between individuals within both males and females. The variation in body mass in males (2.0–8.3 mg) and females (6.4–13.2 mg) could result from slightly different feeding habits, as well as the additional energy reserves females require for egg production. On average, females weighed more than twice as much as males, and the heaviest female was 6.6 times heavier than the lightest male.

The acceleration of the peacock spider, M. splendens, was the fastest among all known jumping spiders: more than twice as fast as the previous highest acceleration in Phidippus princeps and more than 4 times faster than that of Habronattus conjunctus, which had the slowest acceleration among all Salticids (Table 2; Brandt et al., 2021; Hill, 2006). The fast acceleration in peacock spiders may be an indication of the predation pressure or the speed of the prey they need to track and capture in natural environments. Although the lighter males had higher acceleration compared with the heavier females, after controlling for body mass, we found that acceleration in males was slower compared with that in females (Fig. 4, Table 2).

The lighter males had significantly shorter take-off times compared with those of females (Table 1). On average, males and females experienced the equivalent of 13.03 g-force and 12.5 g-force, respectively, during take-off, which is higher than any other known jumping spider (Table 1). In comparison, froghoppers (Philaenus sp.) that use stored energy for their jump, experience 550 g-force (Burrows, 2006). As the g-force experienced by M. splendens and other Salticids is several magnitudes lower than that of animals that rely on stored energy, the semi-hydraulic system that drives jumping in spiders is unlikely to have any elastic or stored energy component. Although the lighter males had a steeper jump take-off angle compared with the females (Fig. 3B,C), there was a large spread in both sexes for this measure. Surprisingly, the difference in the take-off angle between sexes was not reflected in differences in the maximum height of the jump trajectory (Fig. 3A,D,E). A potential reason for steeper take-off angles in males compared with females is their small body mass and reduced muscle power output. Male spiders, being lighter than females, take off at steeper angles to achieve the same height with less energy. Females, being heavier, may not need as steep an angle to reach the same height because of their heavier body, which provides more inertia. Because the target, the landing platform, was at the same distance for all spiders, males and females appear to have optimised their take-off angle to match their differing morphologies to achieve similar trajectory heights. Sexual dimorphism in locomotory strategies is known in spiders and insects. In spiders, small-bodied males outperformed heavier females in climbing vertical features or walking faster (Moya-Laraño et al., 2002; Prenter et al., 2010). In the oriental fruit moth, Cydia molesta, females outperformed males in velocity and flight distance (Hughes and Dorn, 2002), and in the butterfly Euphydryas phaseton, female flights are sluggish compared with those of their male counterparts, which is likely to be due to their greater wing loading, which requires a higher wing-beat frequency (Gilchrist, 1990). In ants, males lead a life on the wing, female workers are exclusively pedestrian, and reproductive females initially lead a life on the wing and subsequently become pedestrian (Narendra et al., 2011). While sexual dimorphism is prevalent in jumping spiders including M. splendens, they have been studied mostly in the context of courtship displays (Girard and Endler, 2014; Hill and Otto, 2011), and investigation into their foraging habits, spatial memory and biomechanics is lacking.

Peacock spiders are best known for the elaborate courtship display that males carry out to entice the female (Girard et al., 2011, 2021; Hill and Otto, 2011). As part of this ritual, males of M. splendens extend, wave and lower the third pair of legs in front of the female. In contrast to the other legs, the third pair tend to be long and dark with bright white tarsal setae, indicating leg III is under selection pressure. While the third pair of legs are considered long, measurements of the leg have not been reported. Hence, we measured the length and 2D area of the leg segments by dissecting legs III and IV in one male and one female of M. splendens. Leg III was 1.23 times longer in the male than in the female; leg IV was 1.04 times longer in the male than in the female. In the male, leg III was the longest, with the femur being the longest segment. In the female, leg IV was the longest, with the metatarsus+tarsus section being the longest segment. The 2D area of the femur was largest in the male (leg III: 0.386 mm²; leg IV: 0.348 mm²) versus the female (leg III: 0.303 mm²; leg IV: 0.308 mm²). In comparison to female spiders, male spiders invested more in their femurs, which are larger in leg III than in leg IV, at least in two-dimensional space.

The third and fourth pair of legs played a crucial role in the jumps of both males and females. Well before the start of the jump or displacement of the CoM, in both sexes, spiders raised the first two pairs of legs and extended them in front of their cephalothorax (Fig. 5; Movie 1). The fourth pair of legs left the platform next, followed a few milliseconds later by the third pair. In most jumping spiders, similar to M. splendens, the third pair of legs are the last to leave the ground. The only exception is Attulus pubescens, in which the fourth pair left the ground last (Parry and Brown, 1959b). We followed the elegant strategy that Brandt et al. (2021) developed to identify the propulsive leg, by calculating the ELL and the angle between leg joints relative to peak acceleration. It was evident that spiders extended and held the first and second pair of legs straight, well before peak acceleration occurred. Leg IV attained the highest ELL and its leg joint was almost straight before the animals attained peak acceleration, with one exception, in all male and female spiders (Fig. 6). This meant that spiders continued to accelerate after leg IV was fully extended and had left the ground. Leg III reached the highest ELL after peak acceleration had occurred. These two findings suggest that leg III is the propulsive leg in both males and females of M. splendens, similar to findings in H. conjunctus (Brandt et al., 2021). Thus, leg III is under extreme selection pressure as it drives two significant functions: locomotory jumps and courtship behaviours. It remains to be seen whether males are more vulnerable to predation during courtship, as leg III is extended and waved as part of a visual signal. Equally, it is essential to characterise the 3D volume of the femur and investigate the internal structure and muscles involved in the jumps.

Jumping spiders occupy 3D-cluttered environments where they jump under various contexts. Here, we studied jumps exclusively in the context of locomotion that was planned and well executed. Escape or startle jumps provide a strikingly different context from locomotory jumps, with the former usually unsteady, non-directed and tending to have high accelerations. Identifying the kinematics of escape or startle jumps will pinpoint the limits of the system, which locomotory jumps do not capture (Kiørboe et al., 2010). Thus, the currently known acceleration and take-off velocity in M. splendens and H. conjunctus are likely to be lower than the maximum kinematics these species can achieve (Brandt et al., 2021). A second and unusual context is multiple jumping or continuous jumping, which is a form of endurance locomotion that jumping spiders and some insects exhibit (Ali et al., 1992). Here, we studied only a single jump, and it remains unknown how acceleration and jump forces change when animals perform consecutive jumps. A third context is the plane of jumping, i.e. whether an animal carries out horizontal, ascending or descending jumps at different elevations. Here, we studied jumps towards a 4 cm vertical target where animals could land anywhere on the target. Jumping to smaller targets at higher planes will probably require more energy to take-off. Ascending and descending jumps have been described in jumping spiders (Phidippus regius and P. princeps), but outcomes remain unclear, as only one individual was studied (Hill, 2006; Nabawy et al., 2018). In pigeons, potential energy and the power required to fly increases with flight angle (Berg and Biewener, 2008). Thus, systematically probing jump kinematics to targets at different elevations would provide valuable insights.

Salticids vary greatly in body mass and inhabit a diverse range of microhabitats. After controlling for body mass, we found that nearly all kinematic variables, except jump power, were lower in males compared with females. Jump kinematics have been studied in only a few spider species, with the smallest being males of M. splendens. Spiders rely on a combination of hydraulic pressure and muscular action to propel their jump. The jump performance of peacock spiders and other jumping spiders suggests that their jump strategy is more akin to a muscle-actuated system, which operates at several magnitudes slower than spring-actuated systems. A comparative approach to analyse both the jump kinematics and hydraulic mechanism driving the jump is very much needed. Furthermore, analysing the geometry and anatomy of the legs would be fascinating, as they play a crucial role in shaping the fluid dynamics of leg extension.

We acknowledge technical and scientific assistance with µCT scanning from Matthew Foley at the Sydney Microscopy & Microanalysis, the University of Sydney. We thank Drew Allen for statistical advice and two anonymous reviewers for feedback.