Effect of shear forces and ageing on the compliance of adhesive pads in adult cockroaches

The locomotion of insects is often constrained by their ability to attach to surfaces in their environment. In order to climb successfully on a wide range of rough surfaces found in nature, insects have evolved attachment organs, which include claws and soft attachment pads (Beutel and Gorb, 2001). Claws can interlock with most surface asperities if larger than the diameter of the claw tips, while soft pads are most useful when surface protrusions are too small for the claws to grip (Dai et al., 2002; Bullock and Federle, 2011).

The mechanisms by which insect adhesive pads cope with surface roughness are still not well understood. The pads need to be compliant to be able to deform and achieve sufficient contact (Gorb et al., 2000; Clemente et al., 2009). In smooth adhesive pads of insects, compliance at different length scales is achieved by a hierarchically organised cuticular structure. The specialised cuticle of smooth pads consists of large inner rods oriented at an angle almost perpendicular to the surface. They branch out into finer fibrils near the surface, conveying smaller-scale compliance. These fibres are covered by a thin epicuticle, which in stick insects is folded longitudinally (Scholz et al., 2008; Bennemann et al., 2014).

A further key feature of animal attachment organs is their direction dependence. Forces of adhesive pads are usually maximised when insect legs are pulled toward the body and they detach easily when this force is released or when they are pushed (e.g. Federle et al., 2001; Federle and Endlein, 2004; Autumn et al., 2006; Clemente and Federle, 2008). Although this direction dependence is wide spread among animals with adhesive footpads, the mechanical systems underlying it can be diverse. In hairy adhesive systems, direction dependence is caused by the default position of seta tips, which need to be bent by a pull in order to make full contact to the substrate. In smooth adhesive pads, the direction dependence can be based on the unfolding of the adhesive pad, or on a hydraulically mediated contact area increase.

It is still unclear, however, whether the ability of the pad to deform around surface asperities is also dependent on the shear force acting on the pad. The internal fibrous cuticle structure of smooth pads would allow a shear-dependent change in compliance. It has been shown that the internal fibres assume a more oblique orientation when the pad is pulled. This may not only result in a more compact packing of the cuticular rods (thereby potentially increasing the cuticle's modulus) but also reduce the thickness of the cuticle (thereby reducing its bending stiffness). It is unclear which of the two effects dominates, or whether they will cancel each other out. If there is a change in pad stiffness mediated by a pull, it could have different effects, depending on the substrate. On a smooth substrate, peeling or fracture mechanics models would predict a stiffer pad to achieve higher adhesion (Maugis and Barquins, 1978; Bartlett et al., 2012). On a rough substrate, compliant adhesives will adhere better as they will achieve a larger contact area. Here, we studied how shear forces influence the ability of smooth adhesive pads to make contact to a transparent microstructured substrate.

The stiffness of tarsal pads has been found to be affected by the insect’s age. The pads of aged cockroaches were found to be less flexible than those of younger ones, concurrent with a reduced climbing performance (Ridgel et al., 2003; Ridgel and Ritzmann, 2005), but it is still unclear to what extent these changes are the result of wear and damage to the cuticle. Insect pads are often found with brown ‘scars’, which are much stiffer than the rest of pad. Lai-Fook (1966) demonstrated that, because of the action of phenolases, a permanent localised darkening and hardening of the cuticle (sclerotisation) occurred in insects after superficial abrasions. Insect pads are affected by this ‘scarring’ process with age, which affects their flexibility.

Age may also provide a possible explanation for the observed, high intraspecific variation of adhesion in insects. Even on the same substrates, insects from the same colony can show very different shear and adhesive stresses (e.g. Clemente et al., 2009; Zhou et al., 2014).

To quantify the stiffness of smooth insect pads, previous studies measured load–displacement curves of whole pads (Gorb et al., 2000; Jiao et al., 2000) or indented pads with spherical tips, or used atomic force microscopy (AFM) to characterise the micromechanics with high spatial resolution (Scholz et al., 2008). In these studies, the material properties were measured at a single point under approximately static conditions.

To assess the effective stiffness of the whole adhesive pad contact zone, we used a transparent micropillar substrate with a gradient of pillar spacings. Microstructured substrates are powerful tools to study insect adhesion (Clemente et al., 2009; Zhou et al., 2014). Smooth adhesive pads of insects make full contact (touching the substrate between the pillars) where the pillars are spaced widely but only partial contact (touching the pillar tops only) where the pillar spacing is narrow (Zhou et al., 2014). The effective elastic modulus of pads can be estimated from the transition point from full to partial contact (Zhou et al., 2014).

When the surface of the cuticle has been damaged, the water-proofing wax layer of the epicuticle may be destroyed, making the cuticle permeable to water-soluble dyes such as Methylene Blue (Slifer, 1950). Following damage, the lipid layer can be repaired and become again impermeable (Wigglesworth, 1945; Slifer, 1950). For soft insect cuticle, more than superficial damage can result in sclerotisation, leading to permanent brown scars (Lai-Fook, 1966). We therefore studied the cuticular damage to the arolium by examining blue (stain) and brown (permanent scar) coloration.

In this study on adult speckled cockroaches [Nauphoeta cinerea (Olivier 1789)], we used the above techniques to address the following questions: (1) how do pulling forces affect the stiffness of adhesive pads?; (2) does the compliance of pads change with age?; and (3) how does pad damage accumulate over time in cockroaches?

When cockroach adhesive pads were brought into contact with the gradient micropillar substrate (Fig. 1), they made full contact on the side with large pillar spacings, and partial contact on the side with dense pillar arrays. The transition from full to partial contact mostly coincided with a change in pillar spacing on the substrate (Fig. 2). When pads were pulled across the substrate, the transition changed significantly from larger to smaller pillar spacings (supplementary material Movie 1). The values of λ (indicating compliance) for cockroach pads increased significantly with sliding distance (Page's L-test, L_21,60=1.73×10⁵, P<0.001; Fig. 3). The position of the contact transition approached a steady state during the pulling movement. The change was fastest when the pulling movement started. In the course of sliding, the pads' total projected contact area increased, compared with the initial contact (Wilcoxon signed rank test, V=272, N=46, P=0.003).

The changes in pad compliance (λ) cannot be explained by simple creep, as λ did not change significantly from 5 s before to the start of sliding (Wilcoxon signed rank test, V=57, N=34, P=0.445). However, λ then increased significantly with sliding (Wilcoxon signed rank test, start of sliding versus 5 s thereafter, V=52, N=34, P=0.008; P-values corrected for multiple testing using the Holm–Bonferroni method; Holm, 1979).

In order to investigate whether the observed change in compliance is based on a change in material properties or caused by different areas of the pad cuticle coming into contact with the substrate, we made an attempt to track ‘landmark’ points in the adhesive contact zone. Unfortunately, only one adhesive pad had a visible air bubble that could be tracked as a landmark (see Fig. 2). In this cockroach, λ increased within the first second from 0.087 to 0.096 µm. The distance of the air bubble from the proximal edge of the contact zone increased by 19.6 µm within 3 s of the start of the pulling movement. This suggests that the pad underwent a small amount of rolling, bringing new areas of adhesive pad cuticle into contact on the proximal side. However, as the movement of the air bubble was considerably shorter than the length of the adhesive pad (97.1 µm, see Fig. 2), the bulk of the adhesive contact area during initial contact was still in contact during the steady-state sliding phase.

To determine whether the ‘new’ areas on the proximal side of the pad could be responsible for the measured change in pad stiffness, we tested whether pads are more compliant on the proximal side than on the distal one. We measured the transition from partial to full contact separately for the distal and the proximal half of the pad contact zone for pads in initial contact and after sliding. We did not find any evidence for a higher compliance on the proximal side, either at initial contact or after sliding; there was even a slight trend towards higher compliance on the distal side (Fig. 4; Wilcoxon signed rank tests for initial contact, V=572.5, N=60, P<0.001; after sliding, V=32, N=60, P=0.114). This indicates that the observed decrease in stiffness after shear is not caused by simple rolling of the pads.

The transition from full to partial contact changed and hence the value of λ increased gradually to a steady state during the pulling movement (Fig. 3). To compare the pad compliance for cockroaches of different ages, we used the steady-state value of λ (measured at t=50 s). Pad compliance significantly decreased with age (Spearman's rank correlation, r_s=−0.588, N=60, P<0.001; Fig. 5). This decrease appeared to be relatively abrupt. Specifically, the difference in compliance was significant between consecutive age groups only for cockroaches that were 2.5 and 4 months old (Mann–Whitney U-test, W=100, N=20, P<0.001, all other comparisons P>0.05; P-values corrected for multiple testing) (Holm, 1979).

In order to verify that the observed ‘age’ effect represented a real change and not a difference between separate groups of cockroaches kept in different boxes, we collected a smaller series of data using untested cockroaches from the 2.5 month group, tested 3 months later, i.e. after 5.5 months (N=10). Again, pad compliance was significantly lower for the 5.5 month old cockroaches compared with the 2.5 month old cockroaches (Mann–Whitney U-test, W=84, N=20, P=0.017; Fig. 6). The two groups tested at 5.5 months of age were not significantly different from one another (Mann–Whitney U-test, W=61, N=20, P=0.418; P-values corrected for multiple testing) (Holm, 1979).

We compared the adhesion forces of cockroaches aged 0.5 and 4.5 months on a microrough test substrate. The adhesion forces of the younger cockroaches were significantly higher than those of the aged ones (Mann–Whitney U-test, W=82, N=20, P=0.014; Fig. 7). This suggests that the softer pads in younger cockroaches help them to deform and achieve sufficient contact on rough surfaces.

We assessed both Methylene Blue staining (indicating recent wear) and brown cuticle coloration (indicating cuticle repair and sclerotisation) of the arolium in newly hatched, and 3 and 7 month old cockroaches. There was no correlation between the amount of blue arolium staining and the age of the cockroaches (Spearman's rank correlation, r_s=0.081, N=26, P=0.694). However, brown arolium coloration increased significantly with age (Spearman's rank correlation, r_s=0.672, N=26, P<0.001; Fig. 8). A repeated analysis of the same pads of the newly hatched cockroaches showed that there was a strong correlation between the amount of blue staining on day 3 and the brown coloration on day 10 (Spearman's Rank correlation, r_s=0.628, N=18, P=0.005).

The brown scars appeared to reduce the pads' flexibility. We visualised this effect in a 7 day old adult cockroach with a single brown scar on one of its arolia. This arolium was brought into contact with a pillar substrate with 1.4 µm pillar height and diameter, and 4 µm spacing (Fig. 9). In the scarred area, the arolium cuticle was unable to deform around the pillars to make full contact on this substrate, whereas all other areas were soft enough to achieve full contact.

Our results show that the ability of cockroach adhesive pads to compensate for surface roughness increases when the pads are pulled toward the body. Following shear, the pads made full contact on pillar arrays with a smaller spacing than before the shear movement. The median effective compliance parameter (estimated from the topography) increased by 22% within the first 9 s of the start of the pull, and stayed approximately constant thereafter (Fig. 3). What is the mechanism underlying this increase in compliance?

First, the change in stiffness could be explained by time-dependent viscoelastic properties of the pad cuticle. Viscoelasticity has been demonstrated for smooth adhesive pads of other insects at the level of whole pads (Gorb et al., 2000), but its importance for the pad cuticle itself is still unclear. After bringing the pads into contact with the gradient pillar substrate, the pad cuticle did not show any significant creep, i.e. the compliance parameter λ remained approximately constant over the 5 s of contact before the sliding movement started. Only with the start of sliding did the pad's compliance increase. Thus, the observed changes in stiffness cannot be explained by viscoelasticity. Strong viscoelasticity may generally be undesirable for adhesive pads of climbing animals, as they have to attach and detach rapidly during locomotion.

Second, shearing results in some rolling of the pad, bringing new areas of cuticle into contact at the proximal side. Thus, it is possible that different areas of pad cuticle are in contact during initial contact and after shear. However, our findings indicate that not only was the amount of rolling relatively small (so that only about 20% of the pad's contact area was new) but also there was no evidence that regions on the proximal side of the pad were any softer than those on the distal side. Thus, although some rolling may have occurred, it cannot explain the observed shear-induced increase in pad compliance.

Third, the transition from partial to full contact of the pad could be influenced by the normal pressure in the pad contact zone. Although a constant normal force was maintained during our experiments, it is possible that a shear-induced reduction in contact area (as observed for rubber hemispheres pressed against glass; Savkoor and Briggs, 1977) increases the pressure within the contact zone, leading to a shift of the compliance parameter λ. However, our results showed that the overall projected contact area increased when the pads were sheared, so there is no evidence for a pressure-induced transition from partial to full contact.

In the absence of evidence in favour of the previous three explanations, the observed increase in pad compliance may indicate a change in cuticle material properties in response to a pull. A change in cuticle material properties could be based on the pad's internal fibrillar ultrastructure. In stick insects, the default angle of the larger cuticular rods was measured to be 57 or 71 deg to the pad surface in the proximal-to-distal orientation (Dirks et al., 2012; Bennemann et al., 2014). A pull reduces this angle (Dirks et al., 2012). The consequences of this angle change for effective cuticle stiffness are non-trivial. Because of the smaller angle, both the main rods and the finer cuticular fibrils of the outer ‘branching’ zone may bend more in the tangential direction, thereby increasing compliance. However, the distance between adjacent rods will decrease, making the pad cuticle more compact, potentially increasing the coupling between rods and reducing pad compliance.

It is likely that the observed stiffness change is an adaptation to enhance adhesion on rough surfaces when the foot is pulled towards the body. So far, however, it is unclear whether the change in compliance is reversible and whether it occurs only when the pad is sheared in the pulling direction.

The compliance of cockroach arolia (values of λ) decreased significantly with age. This decrease appeared to be relatively abrupt, with the greatest change occurring in cockroaches between 2.5 and 4 months of age. As the pads' compliance parameter λ decreased from around 0.14 to 0.10 μm, their elastic modulus may have increased by ca. 40% (assuming no change in the work of adhesion with age). This stiffening occurred in pads without any visible damage or brown coloration. Thus, if pads with visibly damaged areas (and sclerotised scars) had been included, an even greater effect would have been recorded.

In a previous study, we quantified pad stiffness in N. cinerea cockroaches without considering their age, and found a mean compliance parameter λ of 0.15 µm (Zhou et al., 2014). This value corresponds well to the pad compliance of young cockroaches found here, suggesting that most insects used in our previous study were less than 4 months old. This age bias may be based on our selection of insects and pads without any visible damage (Zhou et al., 2014).

Adhesive pads are subject to damage and wear throughout an insect's lifetime. We monitored wear and damage of the cockroach arolia by observing cuticle coloration and its stainability with Methylene Blue.

Pad staining by Methylene Blue indicates recent damage by abrasion of the impermeable wax layer on the surface of the epicuticle (Wigglesworth, 1945; Slifer, 1950). The wax layer can be repaired, thereby restoring its water-proofing function (Slifer, 1950; Lai-Fook, 1966). Consistently, the level of Methylene Blue staining in our study did not increase with age but remained approximately constant. It was found for Calpodes fly larvae that wounded cuticle regions started to darken (indicating sclerotisation by phenolic tanning) before they lost permeability to Methylene Blue (Lai-Fook, 1966). When larvae were prevented from sclerotisation by dipping them in a suspension of phenylthiourea, their permeability to Methylene Blue persisted longer (Lai-Fook, 1966). These findings suggest that repair of the wax layer and cuticle sclerotisation are correlated. Therefore, it is likely that the increased pad stiffness observed in our study was also based on a weak sclerotisation of the cuticle by phenolic tanning (Vincent and Wegst, 2004; Andersen, 2010), even though no brown scars were visible. The detailed mechanism underlying the invisible stiffening is still unclear; it could be simply based on smaller amounts of melanin produced during the repair of small and superficial cuticle wounds (Slifer, 1950; Lai-Fook, 1966), or by a different mechanism of cuticle sclerotisation (Andersen, 2010).

Brown scars were more frequent and occupied larger areas on the pads of older insects. The stiffening of adhesive pads reduces the insects' adhesive performance on rough surfaces, and it may contribute to the age-dependent decline in locomotor activity of cockroaches (Ridgel et al., 2003). Stiffening may represent an unwanted by-product of cuticle repair mechanisms, and selection should favour physiological mechanisms that allow the repair of the wax layer in adhesive pads with a minimal stiffness increase.

In many insect species, individuals of different ages vary in their activity, leading to a simple decline in spontaneous locomotion with age as reported in flies (Le Bourg, 1987), or a complex division of labour between age groups in some social insects (Seeley, 1995). Age-dependent activity could result in a different frequency of adhesive pad surface contacts that contribute to wear (Ridgel and Ritzmann, 2005). The fast decrease in pad compliance between 2.5 and 4 months of age may be linked with a higher locomotory activity for cockroaches of this age, although no higher Methylene Blue stainability was observed.

The microstructured, transparent substrates with standardised topographies are a powerful tool to study the performance of natural adhesives on rough surfaces. The gradient pillar substrates used in this study allow an instantaneous assessment of pad stiffness over the whole contact zone and under different experimental conditions.

Adult cockroaches (N. cinerea; body mass 549.3±14 mg, mean±s.e.m., N=60) were taken from laboratory colonies. In order to group the cockroaches by age, newly hatched adult cockroaches were collected and kept in separate plastic boxes together with those hatched in the same month. Newly hatched adult cockroaches were easily recognised by their completely white cuticle; most parts of the exoskeleton turn brown in the course of cuticle sclerotisation in less than 1 day. We are therefore confident that the collected cockroaches were less than 1 day old. We separated six groups of cockroaches; within each group all insects had hatched within the same month. The cockroaches were kept at 24°C on a 12 h:12 h light:dark cycle and fed on dog food. There was a significant increase of body mass with age (Spearman's rank correlation, r_s=0.490, N=60, P<0.001; body mass, newly hatched: 419±10.2 mg, N=10; 7 months old: 634±28.2 mg, N=10, means±s.e.m.). For the experiments with microstructured substrates, cockroaches were anaesthetised with CO₂ and immobilised by tying them on their back using Parafilm to a microscope slide glued on a glass tube. One of the hindlegs was fixed with vinyl polysiloxane impression material (Elite HD+ light body, Zhermack, Badia Polesine, Italy) to a piece of soldering wire attached to the microscope slide, so that the whole tarsus was immobilised and the adhesive pad stood out as the highest point. To ensure that only the adhesive pad came into contact with the test substrate, we trimmed the tips of the claws. We only tested arolia without any visible damage.

Microstructured transparent substrates were fabricated using photolithography and nanoimprinting. The fabrication method followed a previous study (Zhou et al., 2014). In brief, a lithography shadow mask was designed and produced (Compugraphics, Glenrothes, Fife, UK). SU-8 2002 photoresist (viscosity 7.5 cSt; MicroChem, Newton, MA, USA) was spin-coated onto a silicon wafer for 30 s at 2000 rpm, resulting in a feature height of 1.4 µm. The features were produced on the silicon wafer by exposing the photoresist to UV light through the mask, followed by developer treatment. A soft polydimethylsiloxane (PDMS; Sylgard 184, Dow Corning, Midland, MI, USA) mould was then used to transfer the features from the silicon wafer to transparent epoxy (PX672H/NC; Robnor Resins, Swindon, Wiltshire, UK) on glass coverslips (18×18×0.1 mm).

Using this method, transparent substrates were produced with a 525 µm×2 mm ‘gradient’ pattern consisting of cylindrical pillars arranged in a series of 11 square arrays increasing in pillar spacing (Fig. 1). The pillars were 2 µm in diameter and 1.4 µm in height. The centre-to-centre pillar spacing increased from 3 to 8 µm in steps of 0.5 µm (Fig. 1).

For scanning electron microscopy (SEM) of microstructured substrates, the surfaces were mounted on SEM stubs and sputter-coated with gold to prevent charging. Samples were viewed using a Philips XL 30 FEG microscope (Amsterdam, The Netherlands) with a beam voltage of 5 kV.

Following previous studies (Drechsler and Federle, 2006; Bullock et al., 2008), a custom-made setup was used to perform pulling movements of single adhesive pads of live cockroaches on the gradient micropillar substrate. The spring constant of the 2D force transducer used was 31.8 N m⁻¹ in the normal direction and 41.9 N m⁻¹ in the lateral direction. The force transducer was mounted on a 3D motor positioning stage (M-126PD, C-843, Physik Instrumente, Karisruhe, Germany), controlled by a custom-made LABVIEW program (National Instruments, Austin, TX, USA). The contact of insect pad through the transparent substrate was recorded at 10 Hz using an externally triggered Redlake PCI 1000 B/W video camera (Cheshire, CT, USA).

A coverslip with the gradient micropillar substrate was mounted on the force transducer so that the axis of the gradient was aligned to one of the axes of the motor. The hindleg adhesive pad of a cockroach was brought into contact with the centre of the micropillar gradient with a normal force of 1 mN. The pad was brought close to the surface manually, followed by 10 s approach using the force feedback system. The axis of the leg was perpendicular to the gradient, and the pad touched each of the zones with different spacings. The substrate was then moved for 20 s across the insect pad, away from the insect (corresponding to a horizontal pull of the leg toward the body), while maintaining a constant normal force of 1 mN (feedback control frequency 20 Hz) and a sliding velocity of 0.05 mm s⁻¹. Because of the alignment of the substrate, the pad moved perpendicular to the gradient, and therefore stayed in the centre of the pattern throughout the movement. At the end of the sliding movement, the pad was left in contact with the same normal force for another 30 s before the substrate was pulled off.

The image of the contact area was used to estimate the compliance of the arolium. As the pattern was designed to be slightly narrower than the width of the cockroach pads, a transition from partial contact (pad only in contact with the top of the pillars) to full contact (pad also in contact with the area in between the pillars) was always observed. We measured the pad's total contact area during initial contact and at the end of sliding as twice the (fully visible) pad half-area on the side with larger pillar spacing; the pad mid-line was defined relative to the visible left and right pad edges.

From the contact images, we measured the pillar spacing where the pad changed from full to partial contact (Fig. 2). The position of this transition was measured relative to the visible left and right margins of the gradient to find the smallest pillar spacing on which the pad could still make full contact. This spacing was used to estimate the pad's flexibility. One image per second was analysed for the first 30 s from when the pulling movement started, while one image per 5 s was analysed for the last 20 s as the contact hardly changed during this period.

To evaluate whether ‘new’ areas on the proximal side of the pad could mediate a shear-induced change in pad stiffness, we measured λ separately for the distal and the proximal half of the pad contact zone (see Fig. 2).

In order to evaluate how the variation in pad compliance affects adhesion, we measured pull-off forces of single cockroach pads on a microrough substrate. A piece of aluminium oxide polishing paper (asperity size 50 nm; Ultra Tec, Santa Ana, CA, USA) glued to a glass coverslip (18×18×0.1 mm) was mounted at the end of a metal bending beam with a spring constant of 14.3 N m⁻¹. The bending beam was mounted on a three-dimensional motor positioning stage (see above). A fibre optic sensor (D12, PHILTEC, Annapolis, MD, USA) measured the beam's deflection as the distance to a smooth reflective metal foil target glued onto the beam just before the substrate. The fibre optic sensor was used in its near field (ca. 30–130 µm distance to reflective target), and was calibrated to obtain force. The substrate was moved into contact with the insect pad with a normal force of 1 mN, kept in contact for 30 s (feedback frequency 20 Hz), and then pulled off perpendicularly at a velocity of 0.1 mm s⁻¹. The peak forces were used for further analysis.

The cockroaches used in this experiment were 0.5 and 4.5 months old (body mass: 480±31.1 mg and 645±20.3 mg, means±s.e.m., N=10 each).

We visualised damage and wear of the adhesive pad cuticle with a method similar to that used by Slifer (1950). A 0.1% solution of Methylene Blue was obtained by dissolving Methylene Blue powder (general purpose grade, Fisher Scientific, Hampton, NH, USA) in deionised water. A piece of tissue paper was folded, fully soaked with Methylene Blue solution, and laid out in a Petri dish. A cockroach was then placed on the tissue paper with a Petri dish lid over it. This ensured that the cockroach could not escape and its tarsi were in contact with the dye. Tissue paper was used rather than liquid Methylene Blue solution to prevent the cockroach from drinking the dye. The insect was kept in the Petri dish for 1 h and was then transferred to a clean Petri dish, which was rinsed four times (with the cockroach inside) to remove any surplus dye. The cockroach was then anaesthetised by cooling at −10°C for approximately 2 min until it stopped moving so that it could be observed under a stereomicroscope.

In order to study the age dependence of wear, cockroaches were randomly selected from newly hatched (N=6), 3 month old (N=10) and 7 month old (N=10) groups. In order to score the arolia for blue (Methylene Blue stain) and brown (sclerotised cuticle) coloration, each pad contact area was evenly divided into six segments of equal size (three distal and three proximal). Each pad segment was given a score from 0 to 3 for the intensity of blue and brown coloration. We calculated the average score of blue and brown coloration for each cockroach (pooling across all measurements and arolia).

We thank Phil Taylor (AkzoNobel) and members of the Insect Biomechanics Group (University of Cambridge) for helpful discussions.