**Resilin Distribution and Sexual Dimorphism in the Midge Antenna and Their Influence on Frequency Sensitivity**

**Brian D. Saltin 1,2,\* , Yoko Matsumura <sup>3</sup> , Andrew Reid <sup>1</sup> , James F. Windmill <sup>1</sup> , Stanislav N. Gorb <sup>3</sup> and Joseph C. Jackson <sup>1</sup>**


Received: 10 July 2020; Accepted: 3 August 2020; Published: 11 August 2020

**Simple Summary:** The antennae of insects are multipurpose sensory organs that can detect chemicals, gravity, vibrations, and sound, among others. While such sensors are very specialized and adapted to their specific needs, the way the antenna itself is built has often been considered either uninteresting or unimportant. We used a laser to scan the antenna of the midge *Chironomus riparius*. Insect cuticle, if illuminated with laser light, reflects autofluorescent light, an emission that has long been known to indicate the material properties of the scanned cuticle sample. Rather than a simple beam-like structure of constant material stiffness, we saw bands of hard and soft material, distributed along the length of the antenna. We were able to computer-simulate the effect of this banded structure on the antenna's resonant frequency and showed that it allows the beam to vibrate at different frequencies than would be expected only by its shape. This discovery will help us to better understand these animals' biology and can inspire future biomimetic sensors for detecting sound or vibration.

**Abstract:** Small-scale bioacoustic sensors, such as antennae in insects, are often considered, biomechanically, to be not much more than the sum of their basic geometric features. Therefore, little is known about the fine structure and material properties of these sensors—even less so about the degree to which the well-known sexual dimorphism of the insect antenna structure affects those properties. By using confocal laser scanning microscopy (CLSM), we determined material composition patterns and estimated distribution of stiffer and softer materials in the antennae of males and females of the non-biting midge *Chironomus riparius*. Using finite element modelling (FEM), we also have evidence that the differences in composition of these antennae can influence their mechanical responses. This study points to the possibility that modulating the elastic and viscoelastic properties along the length of the antennae can affect resonant characteristics beyond those expected of simple mass-on-a-spring systems—in this case, a simple banded structure can change the antennal frequency sensitivity. This constitutes a simple principle that, now demonstrated in another Dipteran group, could be widespread in insects to improve various passive and active sensory performances.

**Keywords:** *Chironomus riparius*; Diptera; insects; confocal laser scanning microscopy; finite element modelling; antennal hearing; biomechanics; multimodal sensor

#### **1. Introduction**

Contrary to mosquitoes, whose bite is not only a nuisance but also a pathway for the transmission of disease, midges receive limited scientific attention. However, midges are numerous in both numbers of individuals and number of species [1] and have been shown to be ecologically important for aquatic and lotic systems [2,3], in terms of biomass and production [4]. The present study on the intricate antennal structure, especially of the male non-biting midge *Chironomus riparius*, aims to reveal some adaptations of these animals' biology.

*Chironomus riparius* is a non-biting midge that, like many mosquitoes, displays swarming behaviour [5–7]. Since acoustic communications play an essential role in finding mating partners [5,7–9], it is reasonable to expect that there are similarities in the antennal form and hence properties in species of midges and mosquitoes whose mating behaviour includes swarming. Antennae are remarkable sense organs capable of responding to a variety of sense modalities all at once [10,11]. Known functions include senses of smell and gravity, windspeed detection and, in many species, acoustic perception, the latter postulated as long ago as the 19th century [12]. The flagellar nematoceran antenna is built by three elements: most proximally-the scapus, which is partially responsible for orienting the rest of the antenna, followed by the spherical pedicel, housing a Johnston's organ, and most distally the flagellum, which in both sexes appears sub-divided. The number of sensory neurons in the pedicel of mosquitoes has been estimated to be around 16,000 [13]. Most neurons in the Johnston's organ are thought to be involved with acoustic perception, although which ones remains a matter for debate [14].

In males the flagellum is densely covered by fibrillae (also known as setae). These are hair-like structures which are thought to improve sensory performance by increasing the drag of the antenna [8]. Antennae exhibit strong sexual dimorphism, and the female antennae have shorter and fewer fibrillae than the male antennae, which are often referred to as plumose. Despite the known complexity of these auditory systems, mechanical properties of insect sensory organs are often overlooked [15], with just one recent study on the antenna of swarming and non-swarming mosquitoes [16].

To provide another mechanical case study on dipteran antennae, we chose the swarming midge, *Chironomus riparius*. As in the previous study [16], which deployed state-of-the-art confocal laser scanning microscopy (CLSM), the present study presents morphology of the male and female antenna of *C. riparius* through observation of different autofluorescences of varying cuticle configurations. In turn, this study hints at a potential functional influence of the distribution of material composition on resonant tuning of the flagellum. During the last decade, inferring material properties in this way has become an established method [17–23]. CLSM furthermore has the advantage of allowing the imaging of whole structures with no loss of depth resolution, at higher resolutions than conventional light microscopy. In addition to this structural observation, finite element modelling (FEM) of the mechanical behaviour of the antennae with an elasticity distribution in accordance to the observed CLSM data (following the method of [24], see also [16]) shows the potential effect of element position on the mechanical sensitivity. Finally, we discuss the impact of sexual dimorphism of structures and material composition patterns on resonant tuning and its diversity among species in relation to their mating biology.

#### **2. Material and Methods**

#### *2.1. Specimen Preparation*

Prior to dissection, the animals were anaesthetised with CO2. Dissection was performed in phosphate buffer solution (PBS) (Carl Roth GmbH & Co KG, Karlsruhe, Germany). The specimens were briefly subjected to small amounts of Triton X-100 (Sigma-Aldrich Chemie GmbH, Steinheim, Germany), to remove air bubbles trapped on the surface by decreasing water surface tension. Triton X-100 then was washed repeatedly with the PBS to fully remove traces of Triton X-100. Microscopical observations were made after transfer of antennae or antennal fragments to glycerine (Carl Roth GmbH & Co. KG, Karlsruhe, Germany).

#### *2.2. Confocal Laser Scanning Microscopy (CLSM)*

To analyse local distributing patterns of material compositions within the antenna, we applied CLSM for insect cuticles according to the method established by Michels and Gorb [17]. This technique is successfully used in studies of a wide range of insect exoskeletons [17,23,25,26] including the antennae of mosquitos [16]. The method was applied here as described by Michels and Gorb [17] using a confocal laser scanning microscope, CLSM Zeiss LSM 700 (Carl Zeiss Microscopy GmbH, Jena, Germany). Samples were sequentially exposed to four stable solid-state lasers with wavelengths of 405, 488, 555, and 639 nm, and the excited autofluorescences were filtered with 420–480 nm band-pass and long-pass emission filters transmitting light with wavelengths ≥490, ≥560, and ≥640 nm, respectively. Then, we assigned blue, green, red, and (again) red to the micrographs captured using the filters, respectively, and superimposed them into a final image. To avoid oversaturation, the last two laser lines were combined into one "red" channel, each on 50% intensity. It has to be noted that colours are a product of the colour code applied to the material autofluorescence, and it does not reflect the natural appearance of the antennae. In superimposed images of insect exoskeleton parts, the colour code is as follows: (1) well-sclerotized structures are shown in red, (2) tough flexible cuticular structures are indicated in yellow-green, (3) relatively flexible parts containing a relatively high proportion of resilin appear light-blue and (4) resilin-dominated regions are visualized as deep-blue.

#### *2.3. Finite Element Modelling (FEM)*

Finite Element Modelling (FEM) with COMSOL 5.3a (Comsol Inc., Stockholm, Sweden) was conducted to determine the effects of the CLSM results on the mechanical behaviour of the antenna. As with the previous study [16]—where details of the modelling method were already described—the sole purpose of the present simulations is to show that banding and the location of said bands have the potential to influence the beam mechanics. Hence a simplification to a cylinder (10% shell volume) was deemed justifiable to limit computation time, while still encompassing all relevant features of the system.

Similarly, as opposed to stiffness, mass does not tend to be dramatically different between different types of specialised cuticle [27], and therefore it is assumed to be constant in the present simulations. The parts with higher stiffness were simulated with 5 GPa, medium-hard stiffness elements with about 0.5 GPa, and soft material is around 1 MPa (mimicking a typical value for resilin) [27–29].

While the effect of the articulation in the pedicel was not the subject of our study, an approximation was needed and this was achieved by modelling the entire articulations as a round disc at the base of the flagellum, whose flexibility was fixed [16]. For illustration of the basic cylindrical model, please refer to the inset in the FEM simulation figure.

#### **3. Results**

In the male *Chironomus riparius*, the pedicel is spherical and exhibits weak autofluorescence in comparison to the rest of the antenna (Figure 1a). The flagellum is composed of 11 units, called flagellomeres. With the exception of the most proximal flagellomere—whose flexible part might be hidden by the pedicel or be part of the articulation—the following ten flagellomeres consist of a basal flexible ring (blue) followed by a sclerotized (red) part, where, except for the 11th flagellomere (Figure 1a \*), a circular crest of fibrillae emerges. In the most proximal 11th flagellomere (Figure 1b), fibrillae emerge in an apparently arbitrary pattern. The length of the flagellomeres decreases from the 2nd to 9th, and the lengths are approximately 20–30 µm. The flexible part of the proximal flagellomeres is similar in length to the sclerotized part. In more distal flagellomeres approaching the 10th flagellomere, the flexible part decreases in length to about half of the length of the sclerotized part. The sclerotized part, which is approximately similar in length, gradually loses the dominance of red autofluorescence. From the 5th or 6th flagellomeres onwards, their autofluorescence becomes entirely green (i.e., tough and flexible). The whole structure tapers continuously from the base to the 10th

flagellomere—the diameter of the flagellomeres decreases from around 65 µm to 40 µm and continues to taper towards a pointed tip. After the 10th flexible ring (showing strongly blue autofluorescence), the antenna shows less intense autofluorescence until the tip. The fibrillae continuously become shorter along the flagellum up to the very short and irregular fibrillae at the tip. Along the flagellum, none of the fibrillae exhibits any strong autofluorescence (Figure 1a).

As in the other species previously investigated [16], all prongs are of the same diameter and show homogeneous green autofluorescence. The red-orange autofluorescence of the rim of the pedicel, already visible from the outside (Figure 1a), is also visible from the inside (Figure 1c). This indicates that the rim is relatively well sclerotized. The ridge, where the prongs attach, is deep-blue autofluorescent and possibly resilin-enriched, which is not encountered in any other species studied, while the prongs between attachment and flagellum appear to be of stiffer material indicated by reddish autofluorescence (Figure 1c).

In the female *C. riparius*, the pedicel is slightly rectangular in shape and exhibits comparatively strong green fluorescence (Figure 1d). The flagellum is composed of five flagellomeres, which are cylindrical but not constant in diameter within a flagellomere. All flagellomere have 6–8 separate long fibrillae emerging in a crest. There are rings of blue fluorescence (Figure 1d), which are likely resilin-enriched for flexibility. There is also another crest of shorter fibrillae present on each flagellomere. The long fibrillae emerge in one crest at the widest part of each flagellomere as it broadens, before the flagellomere tapers again. The bottom of each flagellomere, with diameters of 40 to 50 µm, tapers to about half this width (Figure 1c,d).

In the first and second flagellomere of *C. riparius*, fibrillae sockets of the fibrillae crest are apparent and are distinctly more orange/red than its remainder. To the right in Figure 1d, the optical section shows the articulation of the flagellum: no further internal details are visible. In comparison to the male and to the other species, the articulation is more flattened than domed (Figure 1d). An optical section (Figure 1e) of the pedicel shows a rather flexible soft articulation with a central blue area (Figure 1e). The pedicel as such is more fluorescent than the flagellum.

The male pedicel (Figure 1a,c), described in detail above, has a more detailed substructure than the female pedicel (Figure 1d,e). The female pedicel, described in detail above (Figure 1d,e), is in general more angular and less spherical, and the articulation of the flagellum is rather flat. There are clear differences in the flagellum's subdivision in to flagellomeres and material distribution along the flagellum between sexes. Female antennae have five flagellomere, in contrast to the eleven flagellomere of the males. In both sexes, the hard parts of the flagellomeres are separated by blue-fluorescent joints. Furthermore, the female antenna is less covered by fibrillae (Figure 1a,d), which in both sexes similarly show relatively weak autofluorescences. The female does not exhibit the characteristic short intervals among green, red-orange, and light blue bands observed in the first ten antennal flagellomeres in males. Instead, the fewer flagellomeres are more evenly spaced out along the whole length of the female antennae.

Based on the observed flagellomere distribution and material distributions, FEM simulations of the mechanical response of a beam structure with and without the revealed substructure were performed (Figure 2). For the shorter female structure (green lines in Figure 2), no effect of the banded structure on the resonant frequency is seen. For the male antenna (blue lines in Figure 2) a small downwards shift of 4 Hz between the uniform structure (continuous light-blue line) and the more realistic substructured beam (dashed, dark blue line) can be observed. In the lower right corner, a more detailed 1 Hz step simulation of the frequency range around the strongest response 445–475 Hz in males is shown.

**Figure 1.** CLSM observations of the antenna of *C. riparius*. The colour code runs from blue colours for comparatively soft structures to increasingly stiff structures in red. (**a**) Maximum intensity projection of the male *C. riparius* antenna, pedicel and basal antenna part with magnified inset. (**b**) Maximum intensity projection of the male *C. riparius* antenna, tip region. (**c**) Maximum intensity projection of the male *C. riparius* pedicel, seen from inside. (**d**) Left: Maximum intensity projection of the female *C. riparius* antenna, right: cross-section. (**e**) Optical cross-section of the female *C. riparius* pedicel. Abbreviations: bp: basal plate, fl: flagellum, fm: flagellomere, fb: fibrillae, j: joint, pd: pedicel, pdw: pedicel wall, pr: prongs, rg: ridge.

**Figure 2.** Simulation results for models of uniform and structured female and male *C. riparius* antenna in 5 Hz steps between 20 and 620 Hz. The figure includes for illustrative purposes the simulated female and male model (left and right, respectively). In each model the point (node), whose displacement is shown in the figure, is marked with a black triangle. The displacement is codified as gradient from low (dark blue) to high (red). Indicated in green (female) and blue (male), triangles point to the frequency of the strongest mechanical response. The figures underneath show the zoomed-in response for the female (left, green) and the male (right, blue). In each panel the comparison between a uniform beam of the sex-specific dimension depicted as solid line and the more natural situation of a substructured flagellum-beam depicted as dashed line, also in the sex-specific dimension. The used distribution of substructure is deduced from the antenna CLSM images Figure 1a,b (male) and Figure 1d (female). Simulating an impinging sound field, load was applied perpendicular to the beam axis in the +X direction on all but the lowest element.

#### **4. Discussion**

Our results show that the well-known sexual dimorphism of dipteran antennae goes further than morphological structure alone, and in midge antennae also includes differences in material elasticity. Like in mosquitoes [16], material composition of the antennae is not homogeneous along the flagellum, but instead comprises hard and soft elements. Taken together with the structural complexity of the antenna in mosquitoes [16] and stick insects [30], it is becoming more evident that the structure and especially material and composition of insect antennae is much more complex as previously thought. Despite the statement that material properties of insect sensors are largely overlooked made as early as 2009 by Sane and McHenry [15], only limited research has been conducted to amend this lack of understanding. A lot of questions remain open and there is much potential for future research given the vast diversity of insect antennae not yet sampled. To our knowledge, none of the hitherto investigated species here and in our previous study [16] closely resemble each other regarding material distribution irrespective of their mating ecology. This indicates that further research will be necessary to better understand the various factors influencing antenna morphology. Given the different sensory functions of insect antennae that include, but are by far not limited to, olfaction, tactile sensation, and hearing, it is clear that the structure balances various trade-offs and functional constraints. One example of intricate structures of unknown function is the rapid sequence of flexible and sclerotized material at the base of the male flagellum of *C. riparius*.

Pedicels have a large variation of autofluorescence intensity and are largest in *C. riparius* females. In male *C. riparius,* a hard area on the distal ridge, where the flagellum emerges from the pedicel, is most prominently visible. The two important messages regarding the prongs are as follows. First, the prongs are neither particularly flexible nor stiff and are all amongst each other consistent in their autofluorescence within an individual animal. Secondly, judging from our CLSM images, they seem rather similar in dimensions. The uniformity of the prongs in their stiffness and dimension underpins previous assumptions by Avitabile et al. [31] that the prongs act more or less as rigid-body extensions of the flagellum. The flagellum, however, is by virtue of stiffness variation, shown by our study, potentially acts in a more complex manner than simply rigid beam of uniform stiffness. Similar to our study on mosquito antennae [16], we confirmed here the presence of variation and increased small-scale complexity of the dipteran antenna.

While the degree of effect remains under dispute, the direct fitness improvement of traits involved in sexual selection is not [32–34]. An impact of these differences on mating behaviour seems likely given the combination of the following three points: (1) certain mosquitoes (7–9) and at least some midges [5] respond to acoustic stimuli; (2) their antennae clearly show a well-known structural sexual dimorphism (e.g., [35]), and as demonstrated here also a dimorphism in material composition; (3) considerations by Loudon [36] heavily imply the importance of getting the flexural stiffness of antenna right for any given insect. This means that while the function of the different banded structure between species [16] and sexes reported remains unclear for now, they will be meaningful for the behaviour and biology of those animals.

Possible reasons for these antennal observations are that a different stiffness will inadvertently correspond to a different resonant tuning for acoustic perception, or for reasons of static integrity of the antenna, or perhaps another behavioural or ecological aspect of these animals' biology. Compared to results in mosquitoes [16], the effect might be smaller in male Chironomidae or different in principle—both hypotheses require further investigation. Whatever the ultimate reasons for the observed specialisation are, it is fairly clear that different specialisations of males and females might require strong tuning of their acoustic sensors (antennae), which is not understood yet, but this study shows further evidence for the presence of such a specialisation. A limitation of both these studies is the lack of direct correlation of CLSM-based autofluorescence analysis with mechanical measurements, which should be tackled in follow-up investigations.

#### **5. Conclusions**

We have demonstrated that the sexual dimorphism in the antenna of *Chironomus riparius* pertains beyond geometry to material composition. The antennae of both sexes balance a variety of functions. Hence it is difficult to decide—without further research—how much of the newly found complexity actually is adaptive to a given sensory function. While effects on resonant tuning in male midges are small compared to the hundreds of Hz shifts observed in mosquitoes [16], variation in stiffness can alter the antenna's vibrational characteristics in different species.

This result and other studies on the mechanics of antennae [16,30] as well as other appendages [26], underlines the necessity of a more holistic and realistic future approach not only but especially for modelling. That includes the hitherto unknown material complexity in these structures.

Future studies of insect antenna could include investigations on other species or be combined with direct mechanical measurements, such as bending and indentation tests, which would provide better understanding of their structure-function relationships. Such outcomes will improve the quality of simulation results, as we clearly see how the mechanical responses can deviate due to structural and material complexities so far observed. The importance of knowledge about material properties of insect cuticle for understanding functional mechanisms of different organs is huge [21,26,30,37–41] e.g., for robotics [40,41], and can be extended to the sensory structures [16,41–45]. This is not only a matter of academic interest but could also feature in the improvement of biomimetic sensory systems with wide applications. Rather than trying to find materials with a given Young's modulus to satisfy a design constraint, stiffness can be altered through careful design of banding with standard materials.

**Author Contributions:** B.D.S. and Y.M. carried out preparations and CLSM imaging. A.R. and B.D.S. conducted FEM simulations. B.D.S., J.F.W., S.N.G. and J.C.J. designed the study. B.D.S., A.R., S.N.G., Y.M., and J.C.J. wrote the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported in part by the European Research Council under the European Union's Seventh Framework Programme FP/2007-2013/ERC under Grant Agreement n. 615030 to J.F.C.W. This work was partially supported by the European Research Council under the European Union's Seventh Framework Programme FP/2007-2013/ERC under grant agreement no. 615030 to J.F.C.W. by the EPSRC (J.C.J., EP/H02848X/1) and by the German Research Foundation (Y.M., DFG grant no. MA 7400/1-1). B.D.S. is funded by the HSB Research Fellowship. The APC were funded by UKRI.

**Acknowledgments:** We thank Jan Michels (University of Kiel) for theoretical and practical CLSM training, as well as members of staff of the Centre for Ultrasonic Engineering at the University of Strathclyde for their support, especially Jeremy Gibson. Thanks also go to Carla Lorenz and Heinz-R. Köhler and the Animal Physiological Ecology group (University of Tübingen) for provision of *Chironomus* eggs to start a local culture.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

## **Adhesion Performance in the Eggs of the Philippine Leaf Insect** *Phyllium philippinicum* **(Phasmatodea: Phylliidae)**

#### **Thies H. Büscher \* , Elise Quigley and Stanislav N. Gorb**

Department of Functional Morphology and Biomechanics, Institute of Zoology, Kiel University, Am Botanischen Garten 9, 24118 Kiel, Germany; eliseq2@scarletmail.rutgers.edu (E.Q.); sgorb@zoologie.uni-kiel.de (S.N.G.) **\*** Correspondence: tbuescher@zoologie.uni-kiel.de

Received: 12 June 2020; Accepted: 25 June 2020; Published: 28 June 2020

**Abstract:** Leaf insects (Phasmatodea: Phylliidae) exhibit perfect crypsis imitating leaves. Although the special appearance of the eggs of the species *Phyllium philippinicum*, which imitate plant seeds, has received attention in different taxonomic studies, the attachment capability of the eggs remains rather anecdotical. We herein elucidate the specialized attachment mechanism of the eggs of this species and provide the first experimental approach to systematically characterize the functional properties of their adhesion by using different microscopy techniques and attachment force measurements on substrates with differing degrees of roughness and surface chemistry, as well as repetitive attachment/detachment cycles while under the influence of water contact. We found that a combination of folded exochorionic structures (pinnae) and a film of adhesive secretion contribute to attachment, which both respond to water. Adhesion is initiated by the glue, which becomes fluid through hydration, enabling adaption to the surface profile. Hierarchically structured pinnae support the spreading of the glue and reinforcement of the film. This combination aids the egg's surface in adapting to the surface roughness, yet the attachment strength is additionally influenced by the egg's surface chemistry, favoring hydrophilic substrates. Repetitive detachment and water-mediated adhesion can optimize the location of the egg to ensure suitable environmental conditions for embryonic development. Furthermore, this repeatable and water-controlled adhesion mechanism can stimulate further research for biomimeticists, ecologists and conservationalists.

**Keywords:** attachment; glue; oviposition; biomechanics; walking leaf; morphology; plant surface interactions; insect–plant relations; egg dispersal

#### **1. Introduction**

Stick insects (Phasmatodea) are rather large terrestrial herbivores and well known for their remarkable camouflage [1,2]. This masquerade, imitating parts of their environment, is particularly striking in the lineage Phylliidae (leaf insects). Consequently, these insects are commonly called "walking leaves" [3–5]. Leaf insects extraordinarily imitate the leaves of plants and visually merge with their environment. The first fossil records of Phylliidae date back 47 mya with *Eophyllium messelensis* Wedmann, Bradler and Rust 2006 as the oldest known representative of this lineage [6]. Visual camouflage in stick insects had already evolved during the Cretaceous period (approximately 125 mya), to avoid predators at a time when gymnosperm plants represented the majority of plant diversity [6–8]. During the emergence of angiosperms and their major radiation [9,10], stick insects evolved in a similar rapid fashion, possibly as a response to the burgeoning diversity of plants [8,11–14]. Camouflage is not only used by the insects to deceive predators, but also exhibited by the eggs in their resemblance to plant seeds [1,13,15]. Beyond visual aspects, ranging from the imitation of

twigs, bark, moss and other environmental elements, along with the convergent evolution of leaf mimicry in Phyliidae and several other groups of stick insects [16], other characteristics diversified as well. The attachment systems of phasmids, for example, adapted to the abundance of different plant surfaces [5,17–21]. Females also made use of a remarkably broad range of oviposition techniques, which differ between species depending on their ecological niche [2,5]. As a result, the egg morphology reflects an oviposition technique and ecological niche [15,22,23]. Some species simply drop their eggs passively while others catapult them actively. Passively dropping the eggs is considered an extension of phasmids' notorious masquerade crypsis [2] and probably an ancestral technique [2,13,24]. Another widespread principle is fixation of one egg or groups of eggs to specific spots, e.g., their host plants [13]. While some species mechanically drill their eggs into the soil, into crevices or even leaves and bark, other species secrete a glue during oviposition to permanently fix the eggs to the substrate [2,5,13,20]. The latter is either used to attach single eggs or batches to a certain place, and in one striking case, the egg batches are deposited in the form of an ootheca with a protective case [13]. The different strategies for egg deposition are a result of the low spatial distribution and extensive radiation of phasmids, which presumably led to the co-evolution with angiosperms.

Interestingly, not only has the outer appearance of phasmids been shaped by their co-existence with plants, but also the eggs of phasmids mimic the appearance of seeds and even copy functional principles of seeds. Phasmids are not only the sole insect lineage with species-specific egg appearances [2], but also the only lineage with eggs adapted to different oviposition techniques. Some taxa, in which the eggs are dropped passively, produce eggs which bear a capitulum. This extension of the egg's operculum is not only a signal adaptation for zoochory by ants, but also a result of co-evolution with plant seeds and ants [25,26]. In both capitulate phasmid eggs and elaiosome-bearing seeds, such a lipid-rich extension mimics ant-specific signaling and convinces ants to carry the egg or seed and thereby mediate dispersal [27–29]. Besides ant-mediated zoochory, the eggs of several species of phasmids follow the same principles that plant seeds deploy for dispersion, or aggregation respectively. Many plant seeds disperse via endozoochory, especially via birds [30,31]. Although an initial study has shown that phasmid eggs (directly fed to birds) of a few species do not survive the digestion by quills and ducks [32], a subsequent study found the eggs of several other phasmid species remain viable inside a gravid female phasmid that has been consumed by a bird [33]. Other phasmid eggs, especially *Megacrania* species, are experimentally shown to float in sea water, and disperse via the ocean [34–37], like the seeds of *Cycas* spp. (Cycadaceae) or screw pines do [38,39].

*Phyllium philippinicum* Hennemann, Conle, Gottardo and Bresseel, 2009 (Phylliidae) is a species of leaf insect commonly bred in labs and private cultures (Figure 1A). However, most of the literature on the species revolves around taxonomic and phylogenetic classification and is mainly based on adult morphology [3,40–42]. Leaf insects in general are reported to drop or catapult their egg for deposition from the canopy tops of their host tree [2,13]. Basically, the eggs of this species, as well as those of closely related species, employ a more specialized mechanism for host plant association than previously reported. The specialized exochorionic morphology of leaf insect eggs is predominantly accounted in descriptive morphological studies and taxonomic descriptions [4,15,23,42] and, hence, functional aspects have been widely undocumented. The eggs of several *Phyllium* species, including *P. philippinicum*, resemble plant seeds and bear protruding exochorionic structures (pinnae, according to Clark [43]). The morphology of these pinnae is suggested to be species-specific and their taxonomic use has been previously well demonstrated [4,42]. The functionality of these structures is thus far largely unknown, but the unfolding behavior of the pinnae is often observed in captive breeding. The fact that the pinnae morphologically respond to water has anecdotally raised questions amongst the phasmid breeding community on what purpose this mechanism might serve. Only very few taxonomic studies hypothesized the function of these pinnae. Hennemann et al. [3] described the unfolding of the pinnae after their contact with water and suggested an adhesive function of this system, however did not further elucidate this idea. Additionally curious, the oviposition technique employed by the females, does not involve active gluing of the eggs, which begs the question of whether there is

a presence or absence of accessory reproductive glands in *P. philippinicum*. Unfortunately, most studies on leaf insects solely focus on external morphological features, leaving this question unanswered.

**Figure 1.** Examined species and experimental setup. (**A**) Female of *Phyllium philippinicum* (image is provided by Daniel Dittmar). (**B**) Experimental set-up for detachment force measurements. The egg, which was glued onto the particular substrate fixed on a lab boy using double-sided sticky tape. A hair was glued onto the egg and connected to a force sensor. To detach the egg from the substrate, the force sensor was moved away from the egg in perpendicular direction. The time force signal was amplified and finally processed in a computer.

Overall, strong egg attachment has been reported in a number of other insect species on natural substrates [44–47] and even stronger adhesion was measured from extracted egg glue on various artificial substrates [48–50]. The specific properties of the egg glue seemingly depend on the level of specialization in the attachment system and the habitat/substrate it is specialized for, therefore resulting in different strategies [51]. One important component influencing attachment efficiency is the roughness of the substrate: rougher surfaces create a greater contact area for glue and stronger adhesion after glue solidification [52]. The high complexity in the structural features of plant leaves (trichomes, wax crystals, stomata and cuticle foldings) and fruits (microcracks and epicuticular wax crystals) of various plant cultivars leads to rougher surfaces and increases the adhesion of the eggs of the codling moth *Cydia pomonella* (Linnaeus, 1758) (Lepidoptera, Tortricidae), as experimentally shown [44,45]. The egg attachment strength of the parasitic warble fly (Diptera, Hypoderminae) positively correlates to the roughness of the hairs on its host species [53]. Insect vectors of the human bot fly *Dermatobia hominis* (Linnaeus Jr., 1781) (Diptera, Oestridae) are covered with setae, which enhances egg adhesion for the human bot fly [54]. Another important factor influencing the attachment of eggs is surface chemistry. Eggs of the asparagus beetle *Crioceris asparagi* (Linnaeus, 1758) (Coleoptera, Chrysomelidae) adhere well to the surfaces of the plant *Asparagus o*ffi*cinalis* L. (Asparagaceae), which have superhydrophobic and microstructured surfaces due to the coating by wax crystals [46].

Eggs with adhesive responses in contact with water are only reported for a few insect species. The dragonfly *Libellula depressa* Linnaeus, 1758 (Odonata, Libellulidae), and other Anisoptera [55–64] lay eggs which possess an adhesive coating that swells and generates adhesive properties after the female deposits them in water [65]. The eggs of Ephemeroptera are covered with a thick layer composed of tightly entwined filaments, causing cohesion of the eggs and adhesion to a substrate after deposition into water [66]. The exochorionic structures of these species undergo modifications upon interaction with water, in turn generating adhesion [65,66]. It is assumed that in lieu of colleterial glands [55,56,67], these adhesive coatings are synthesized by follicle cells [65,68] which are involved in eggshell deposition [66,69–71].

On one hand, exploring the adhesive properties and response to water contact of the eggs of *P. philippinicum* can enhance our knowledge of multifunctional bioadhesives. On the other hand, this functional system can provide insights into the life history of this species and shed light on the ecological environments this species inhabits, as this knowledge is usually missing in taxonomic descriptions of museum specimens. This could assist future studies in obtaining broader ecological knowledge of this species, contributing to conservational aspects for both phasmids and plants that can be subject to damage by insects, and also give input on evolutionary studies, as the highly specialized attachment mechanism of *P. philippinicum* is highly derived. In this paper, we asked the following specific questions. (i) How do the eggs of *P. philippinicum* adhere? (ii) How do water contact, surface topography and surface chemistry influence egg adhesion in this species? (iii) Is attachment in *P. philippinicum* eggs reversible and repeatable?

#### **2. Materials and Methods**

#### *2.1. Specimens*

The eggs of *Phyllium philippinicum* Hennemann, Conle, Gottardo and Bresseel, 2009 were obtained shortly after being laid by female insects from the culture of Kirsten Weibert (Jena, Germany). The animals were fed with blackberry leaves ad libitum and kept in a natural day/night cycle. The weight of freshly laid eggs (*N* = 20) was measured using an analytical balance AG204 Delta Range microbalance (Mettler Toledo, Greifensee, Switzerland; d = 0.1 mg).

#### *2.2. Morphology*

Eggs attached to microscopy glass slides were observed with the Leica Microscope M205 (Leica Microsystems Ltd., Wetzlar, Germany). Images were captured from both sides, overview of the egg and view of the contact through the glass slide, using the microscope camera Leica DFC420 (Leica Microsystems Ltd., Wetzlar, Germany). Multifocus stacked images were postprocessed using the software Leica Application Suite (LAS) version 3.8.0 (Leica Microsystems Ltd., Wetzlar, Germany) and Affinity Photo (Apple Inc., Cupertino, CA, USA).

For higher magnification, eggs in contact with different substrates, as well as detached and untreated eggs, were air-dried and sputter-coated with gold-palladium of 10 nm thickness. The substrates corresponding to the detached eggs were sputter-coated as well. Additionally, some untreated eggs were dehydrated using an ascending alcohol series, critical point-dried and sputter-coated as well. These samples were observed in the SEM Hitachi S4800 (Hitachi High-technologies Corp., Tokio, Japan) at an acceleration voltage of 5 kV. Subsequently, the images were processed with Affinity Photo (Apple Inc., Cupertino, CA, USA).

The nomenclature of the egg morphology follows Sellick [23].

#### *2.3. Detachment Force Measurements*

The detachment force of individual eggs was measured in four different experiments. In all experiments, the eggs were mounted on standardized surfaces, as described below, and individually attached to a force transducer (100 g capacity; FORT100, World Precision Instruments Inc., Sarasota, FL, USA) by gluing a horsehair with bees wax onto the lateral side of the egg (Figure 2B) and attaching the hair to the sensor (Figure 1B). The force transducer was connected to a BIOPAC Model MP100 and a BIOPAC TCI-102 system (BIOPAC Systems, Inc., Goleta, CA, USA). Force–time curves were recorded by pulling the eggs off the surfaces using the software Acqknowledge 3.7.0 (BIOPAC Systems Inc., Goleta, CA, USA). The test surfaces were lowered away from the sensor with a speed of approximately 2–3 cm/s using a laboratory lifting platform. In all four experiments, the detachment force was measured by pulling the egg off of a surface at an angle of 90◦ , with the same setup, as described by Wohlfart et al. [72] for spiders and later used for adult stick insects [19]. The highest peak of the visualized graph was interpreted as the maximum detachment force. All surfaces were carefully cleaned with 70% isopropylic alcohol prior to each experiment. Detachment forces were measured in the following four different experiments:

̈

**Figure 2.** Morphology of the eggs of *Phyllium philippinicum*. (**A**) Dorsal view. (**B**) Lateral view. (**C**) Ventral view. **mp**, micropyle; **op**, operculum; **or**, opercular rim; **pi**, pinnae; **ri**, ribs; **sc**, serosal cuticle. Scale bars: 1 mm.

(1) Freshly laid eggs (*N* = 32 per substrate) were mounted on four test substrates with different roughness (0, 1 and 12 µm, and standardized p40 polishing paper) made of epoxy resin (as described below). Eggs were prepared on the test substrates by placing individual droplets of distilled water (~100 µL) on the epoxide plates and then placing one egg in a single droplet, to trigger the unfolding of the pinnae. Subsequently, the eggs were allowed to dry completely (~24 h) and then attached to the sensor.

(2) Eggs (*N* = 20 per substrate) were mounted on three surfaces with different chemical surface properties with the same procedure as described above. The surfaces used differed in the wettability, indicated by the contact angle of the water, which was 36.25 ± 1.15◦ (mean ± SD, *n* = 10) (hydrophilic), 83.38 ± 0.89◦ (the same epoxy resin as used for experiment 1) and 98.9 ± 0.47◦ (hydrophobic).

(3) Additionally, eggs were placed on the hydrophobic and the hydrophilic substrates in wet condition (*N* = 20 per substrate) and the detachment force was measured. The eggs were individually fastened with a horsehair as described above and fully submerged in distilled water for 20 min; afterwards, they were attached to the force transducer and then placed on the test substrate. After letting the eggs sit on the substrate for 1 min, the detachment force from the substrate was measured in the same manner as in the other experiments.

(4) The reproducibility of egg attachment was tested by subsequent pull-off measurements of the same egg. Individual eggs (*N* = 8) were prepared as described in the first experiment and attached to the smooth epoxy resin substrate (0 µm roughness). Then, the detachment force was measured by pulling off the egg. Afterwards, the same egg was then reattached once again using a droplet of water and left to dry for another 24 h. This procedure was repeated for each of the eight eggs six different times, until the measured detachment forces were similar in comparison to the previous day (i.e., revealed no significant difference).

All experiments were performed at 19–21 ◦C temperature and 45–55% relative humidity.

#### *2.4. Surface Preparation*

Two different types of surfaces were used in the experiments. Epoxy resin with a different surface roughness for the first and the fourth experiment and glass with different wettability, as well as epoxy resin, for the second and third experiments.

#### 2.4.1. Glass

Clean microscope glass slides (Carl Roth GmbH & Co. KG, Karlsruhe, Germany) were used as the hydrophilic substrate and silanized, as described by Voigt and Gorb [46], to obtain a hydrophobic substrate. The surface chemistry was characterized by measuring the contact angle of the water on the substrate (aqua Millipore, droplet size = 1 µL, sessile drop method; *n* = 10 per substrate) using the contact angle measurement instrument OCAH 200 (Dataphysics Instruments GmbH, Filderstadt, Germany). The contact angle of the water was 36.25 ± 1.15◦ for the hydrophilic glass substrate and 98.9 ± 0.47◦ for the hydrophobic one.

#### 2.4.2. Epoxy Resin

Substrates with different roughness were produced using epoxy resin [73] following the protocol of Salerno et al. [74]. Negative replicas were cast using polyvinylsiloxane (PVS)-based two-component dental wax (Colthéne/Whaledent AG, Altstatten, Switzerland). Negatives were then filled with epoxy resin and cured at 70 ◦C for 24 h. Glass (0 µm roughness) and polishing papers with the roughness of 1 µm, 12 µm (Buehler, Lake Bluff, IL, USA) and industrially standardized p40 polishing paper (particle size ~440 µm) were used as templates for the resin replicas. The contact angle of the water on the smooth epoxy resin was 83.38 ± 0.89◦ (mean ± SD, *n* = 10).

#### *2.5. Statistical Analysis*

Statistical analyses were performed with SigmaPlot 12.0 (Systat Software Inc., San José, CA, USA). Normal distribution and homoscedasticity were tested using the Shapiro–Wilk test and Levene's test, respectively, prior to other tests. As the respective data were neither parametric nor showed homoscedasticity, detachment forces of eggs on substrates with different surface roughness, as well as on surfaces with different chemical properties represented by corresponding contact angles, were compared using Kruskal–Wallis one-way analyses of variance (ANOVA) on ranks followed by Tukey's post hoc test. Detachment forces of wet and dry eggs on surfaces with different contact angles were compared using Kruskal–Wallis one-way ANOVA and Tukey's test as well. The Mann–Whitney rank sum test was used to compare the detachment forces of eggs in the wet condition on hydrophilic and hydrophobic surfaces. For a comparison of the detachment forces over a six-day period of repeated measurements, a Friedman repeated measures ANOVA was performed along with a Tukey's post hoc test.

#### **3. Results**

#### *3.1. Egg Morphology*

The eggs of *Phyllium philippinicum* are laterally compressed and densely covered with small exochorionic appendages (pinnae, sensu Clark [43]). These pinnae cover most of the egg's surface, except for some circular pits and the center of the micropylar plate (Figure 2A). A corona of shorter expansion surrounds the micropylar plate, oriented away from it. The anterior pole of the egg is covered by an operculum, the lid of the egg, which is released during the hatching of the nymph (Figure 2B). A formation of larger pinnae surrounds the outer rim of the operculum anteriorly. Two ribs along the lateral ridges of the egg are covered with long pinnae as well, expanding the lateral dimensions of the egg (Figure 2). The ribs meet on the ventral side of the egg (Figure 2C). Pinnae of freshly laid eggs lie flat on the surface of the egg, but unfold after contact with water, as described below (Figure 3). Dimensions of the eggs are measured according to Sellick [23]. They measure 4.39 ± 0.36 mm (mean ± SD, *N* = 7) in length, with a height of 2.77 ± 0.25 mm, and width of 2.16 ± 0.14 mm. The mean weight was 15.9 ± 1.3 mg (*N* = 20).

**Figure 3.** Unfolding behavior of *Phyllium* egg pinnae. (**A**) Succession of pinnae unfolding in *Phyllium rubrum* Cumming, Le Tirant and Teemsma, 2018, after exposure to water (images are provided by Bruno Kneubühler), lateral views. *B,C.* Lateral view of untreated *Phyllium philippinicum* egg. (**B**) Overview. (**C**) Detail of folded pinnae. (**D**) Detail of unfolded pinnae of a *Phyllium philippinicum* egg after water exposure. **gl**, glue; **pi**, unfolded pinna; **pn**, folded pinna. Scale bars: 1 mm (**A**,**B**), 500 µm (**C**,**D**).

#### *3.2. Pinnae Behavior and Adhesive Secretion*

The eggs are deposited by the female with the pinnae folded on the surface of the egg. A single pinna consists of a central shaft that is hierarchically split several times towards the tip (Figure 3D, 4). After oviposition, before initial contact with water, the folded pinnae are covered with an iridescent layer of a solidified secretion deposited by the female (Figure 3B,C). The pinnae unfold after contact with water and the secretion liquefies (Figure 3A). The larger pinnae on the operculum and the lateral ribs of the egg unroll and expand the dimension of the projected lateral area of the egg. Smaller pinnae, as well as hierarchical expansions of the main fringes of the pinnae, expand and increase the egg surface as well. The liquefied secretion on the surface of the eggs, after expansion of the pinnae, concentrates on the tips of the expansions (Figure 3D). Along the length of larger pinnae, a reservoir of the secretion forms a bridging film between the shafts of the pinnae. During contact with a substrate, the pinnae deform and spread the viscous secretion, in which they are imbedded, onto the substrate. After some time without contact to water (5–6 h), the secretion dehydrates and solidifies again (Figure 4A,B). After the curing off the secretion, the egg remains attached to the substrate. The adhesive function of the glue is characterized below.

**Figure 4.** Glue associated with *Phyllium philippinicum* pinnae. (**A**,**B**) Stereomicroscopic images of pinnae attached to a glass surface, view through the glass slide. (**A**) Glue deposition on a glass surface and pinnae interaction with the substrate (arrowheads). (**B**). Reinforcement and distribution of the glue by the pinnae. (**C**,**D**) Scanning electron microscopy images of glue–pinnae interactions. *C.* Glue film adhering to pinnae. (**D**) Dried glue residuals on a pinna after detachment from a smooth glass surface. **gl**, glue; **pi**, pinnae. Scale bars: 500 µm (**A**), 300 µm (**B**), 100 µm (**C**), 10 µm (**D**).

#### *3.3. Egg Attachment*

The attachment performance of *P. philippinicum* eggs on different surface roughnesses is illustrated in Figure 5. The maximum pull-off force measured before the egg detached from the respective substrate (maximum detachment force, Figure 5A) is considered a measure for the attachment capability of the egg to the substrate. The maximum detachment force values were highest on the intermediate roughnesses, 12 µm with 144.65 ± 133.38 mN (median ± SD) and 1 µm with 144.23 ± 137.18 mN. The lowest detachment forces were recorded on the roughest (p40; 81.71 ± 104.11 mN) and the smoothest (0 µm; 122.94 ± 95.28 mN) surfaces. However, the differences in median detachment force values between the four surface roughnesses were not significant (Kruskal–Wallis one-way analysis of variance (ANOVA), *H* = 7.278, d.f. = 3, *p* = 0.064, *N* = 32 per roughness).

≤

≤

**Figure 5.** Influence of roughness on egg adhesion. (**A**). Exemplary force-time curve from measurements of the detachment force. (**B**) Detachment forces from substrates with different surface asperity (*N* = 32 for each roughness). Boxes are 25th and 75th percentiles, the line within the boxes defines the median, and whiskers represent the 10th and 90th percentiles. **n. s.** = no statistical difference (*p* > 0.05, Kruskal–Wallis ANOVA on ranks). (**C**) Scanning electron microscopy image of pinnae deformation showing the adaptation of pinna extensions to surface corrugations. (**D**) Schematic interpretation of the eggs' glue with differing degrees of surface roughness. Roughness parameters are given in detail by Salerno et al. [74]. Scale bar: 60 µm.

The attachment performance of eggs on surfaces of differing surface chemistry is displayed in Figure 6A. The detachment force from pulling the eggs off of the hydrophilic surface (water contact angle 36.25◦ ) was very high (792.37 ± 293.94 mN) and significantly higher than the force measured on surfaces with a higher water contact angle (Kruskal–Wallis one-way ANOVA, *H* = 38.543, d.f. = 2, *p* ≤ 0.001, *N* = 20 per surface; Tukey's test, *p* < 0.05). The adhesion to epoxy resin (water contact angle 83.38◦ ) was significantly lower than that of the hydrophilic glass with 159.03 ± 117.31 mN (Tukey's test, *p* < 0.05), but higher than the adhesion to the hydrophobic glass (water contact angle 98.9 ◦ ) with 88.03 ± 114.81 mN. The latter difference, between the epoxy resin and hydrophobic silanized glass, was not found to be statistically significant according to Tukey's post hoc test (Tukey's test, *p* > 0.05).

*χ* ≤

≤ **Figure 6.** Influence of surface chemistry and repetitive detachment on *Phyllium philippinicum* eggs. (**A**) Detachment forces from surfaces with different water contact angles (*N* = 20 for each contact angle). (**B**) Detachment forces from wet and dry surfaces with different chemical properties (*N* = 20 for each treatment). "Hydrophilic" corresponds to a contact angle of 36◦ and "hydrophobic" corresponds to a 99◦ water contact angle. Boxes are 25th and 75th percentiles, the line within the boxes defines the median, and whiskers represent the 10th and 90th percentiles. \* *p* ≤ 0.001, only significant comparisons are highlighted (Kruskal–Wallis one-way ANOVA on ranks followed by Tukey's post hoc test). (**C**) Detachment forces during sequential detachment events (*N* = 8 eggs). Dots indicate the median, whiskers represent the standard deviation. Lowercase letters indicate statistically similarity. Groups with the same letter are statistically equal (Friedman repeated measurements ANOVA, followed by Tukey's test). (**D**) Scanning electron microscopy image of glue residuals on a smooth hydrophobic glass surface. Scale bar: 300 µm.

Adhesion to both hydrophobic and hydrophilic surfaces was very low and practically negligible in the presence of water on the surface (Figure 6B). Wet eggs on the hydrophobic surface (2.74 ± 0.34 mN) showed no significant difference in detachment values compared with wet eggs on the hydrophilic surface (3.17 ± 2.30 mN; Kruskal–Wallis one-way ANOVA, *H* = 66.77, d.f. = 3, *p* ≤ 0.001, *N* = 20 per treatment; Tukey's post hoc test, *p* > 0.05). The comparison of the egg adhesion performance between hydrophobic and hydrophilic surfaces in both wet and dry conditions yielded significant differences between all comparisons, except for the comparison between the two wet surfaces (*p* < 0.05, Tukey's test). Eggs dried in adherence to hydrophilic surfaces showed significantly higher detachment forces than eggs in contact with wet surfaces, as well as higher adhesion than eggs dried on hydrophobic surfaces (all *p* < 0.05, Tukey's test). The detachment force of dried eggs from the hydrophobic glass was lower than from the hydrophilic substrates, but higher than wet eggs from both substrates (all *p* < 0.05, Tukey's test).

Figure 6C illustrates the attachment performance of eggs over a sequence of six repeated detachment events. The median detachment force initially increased from day 1 (94.8 ± 38.0 mN) to day 2 (188.68 ± 66.75 mN). Subsequently, the detachment force consistently decreased from day 2 until day 6 (9.79 ± 4.56 mN). The detachment force was statistically different (Friedman repeated measures ANOVA on ranks, χ <sup>2</sup> = 35.179, d.f. = 5, *p* ≤ 0.001, *N* = 8 per day) and decreased between day 2 and day 6. However, the first three days were statistically similar, but each of the days 1–3 were significantly higher than days 5 and 6 (all *p* < 0.05, Tukey's test). Although the overall decrease in attachment performance is significant, the initially higher median detachment force on day 2 is not significantly different from day 1 and day 3 (*p* > 0.05, Tukey's test).

#### **4. Discussion**

#### *4.1. Attachment Mechanism*

The attachment capabilities of the eggs of *Phyllium philippinicum* were not readily paid attention to in the past and recognized only anecdotally in the literature [3]. However, the combination of an adhesive secretion and reinforcing microstructured exochorionic structures has proven to provide excellent attachment. The mean safety factor (*Fa*/*Fm*; mean detachment force per weight force) of eggs on a smooth epoxy resin substrate ranges around 924, i.e., the adhesion of one egg sufficiently attaches 924 times its own weight. On hydrophilic substrates, the average *Fa*/*F<sup>m</sup>* is 4825.

Water exposure has two main effects on the egg: (1) unfolding of the pinnae and (2) liquefaction of the glue (Figure 3). Both effects contribute to the enhancement of the adhesive properties of the eggs. Like a solvent-based adhesive, the egg adhesive dissolves partially in water and once the water evaporates, the adhesive dries and hardens on the substrate. When introduced to water, the pinnae extend and fan out, adapting to the texture of the substrate. The liquid glue covers the pinnae, which transmit and spread it out onto the substrate. Such horizontally oriented fibrillary structures, that lay parallel to a surface, facilitate the spreading of a fluid, hence enhancing the surface contact of the adhesive fluid [75]. Therefore, bridges of dried adhesive material between adjacent pinnae are visible, when re-solidified (Figure 3D, 4C). To achieve proper attachment, the glue becomes fluid to interact with an adjoining surface, then the adhesive fluid dries to either create a sufficient contact area [76,77] at the interface or mechanically interlock with the surface irregularities [77,78]. Whether high humidity in the surroundings or solely the contact to water droplets cause the glue to liquefy remains untested. To evaluate the effect of ambient humidity, further experiments with exposure to differing humidity are necessary. While the pinnae facilitate adhesion through an increased contact area with the surface, the fluid adhesive makes large real contact with the surface. The exochorionic extensions may also be able to extend into and interdigitate with surface asperities and further spread the adhesive fluid [79–81], depending on the roughness profile of the surface (Figure 5C). Their hierarchical structure offers finer subcontacts with the substrate [82,83], and, hence, optimizes contact formation on natural surfaces of fractal roughness with overlapping wavelengths (e.g., tree bark) [84]. Overall, the pinnae reinforce the film of the glue, thus achieving a viable adhesive system: soft enough to form intimate contact, yet stiff enough after solidification, to decrease elastic deformation and hold a strong bond [84]. This pinnae-based reinforcement offers structural integrity to the adhesive system of the egg.

Besides the mechanical interlocking of the solidified glue with surface corrugations, the glue adheres by physiochemical interactions, presumably van der Waals that can prove very strong with sufficient interfacial contact [77].

#### *4.2. Influence of Substrate Roughness*

Attachment on substrates with different surface roughness revealed no significant differences among all tested surfaces (Figure 5B). Other biological attachment systems are found to be significantly affected by surface roughness. The tarsal attachment systems of some flies and beetles consist of tenent setae that, similar to the pinnae of *P. philippinicum* eggs, adapt to the surface profile [85–91]. However, the performance of these attachment systems for the purpose of locomotion fare better on

smooth surfaces or rougher surfaces exceeding asperity sizes of 3 µm, with the worst performance on micro-roughnesses ranging from 0.1–0.3 µm. This is explained by the spatula-like terminal elements of insects' tenent setae interaction with the surface [85,89,90]. These setae tips are able to make sufficient contact with large surface asperities but are confounded by micro-rough surfaces that inhibit real contact of the setae to the surface. The eggs of *P. philippicum,* in contrast, performed well on all surface roughnesses tested.

This ability of *P. philippicum* eggs is presumably based on the action of initially fluid and later solidified glue. For glues, like that of the adhesive material of the eggs examined herein, rougher surfaces create a larger contact area and stronger adhesion [52]. This applies to the performance of *P. philippinicum* eggs, with an increasing trend in attachment strength from 0–12 µm roughness. Adhesion relies on the area of actual contact made with a surface [79–81]. Although surfaces with micro-roughness are not generally favourable for many insect attachment systems associated with locomotion, the egg's adhesive fluid is able to conform around small surface irregularities and hence increase the actual contact area (Figure 5D). Rough surfaces are beneficial for egg adhesion in different insect species. For the codling moth, it has been previously shown that smoother surfaces with fewer trichomes and rather low free surface energy deter their eggs' attachment [44], while structural features creating a rougher surface on leaves or fruits (e.g., trichomes, microcracks or epicutilar wax crystals) lead to stronger attachment of codling moth eggs due to an increase in the contact area with the egg's glue [45]. Rough surfaces on plants are known to be favorable choices for oviposition sites in other lepidopterans [92,93] and the willow leaf beetle *Phratora vulgatissima* (Linnaeus, 1758) (Coleoptera, Chrysomelidae) [94], leading to an enhanced attachment strength. The same applies for the surface texture of the oviposition substrates of parasitic flies [53,54]. In contrast, the eggs of *P. philippinicum* adhere similarly strong to surfaces of all tested roughnesses. As the eggs are dropped without direct oviposition onto specific substrates, this case of universal adhesion is most likely adapted to a broad spectrum of surface roughness.

However, p40, the roughest surface tested, revealed the weakest overall attachment to the eggs. The adhesive fluid probably generates a larger contact area on substrates with micro-rough surface corrugations. The large surface asperities of the p40 substrate (~440 µm asperity size) principally offer a larger surface for contact formation, but the glue of the egg presumably does not fill the deeper asperities, creating only partial interaction with the walls of the surface asperities (Figure 5D). The viscosity of the glue presumably prevents the glue from properly filling the surface texture.

#### *4.3. Influence of Surface Chemistry*

The adhesive strengths on surfaces with different water contact angles revealed significant differences between the attachment strength of eggs on the hydrophilic substrate compared with the two substrates with higher water contact angles. The contact formation and generation of attachment by the adhesive fluid depends on the surfaces' chemical properties. Higher free surface energy and lower contact angles, which essentially vary inversely to one another [95], are characteristics of surface hydrophilicity that invite greater wetting of liquids on such a surface, in turn forming greater contact at the liquid–surface interface, which creates stronger adhesion [82,96,97]. Lower surface energy reduces the overall attachment ability of a system [84] and therefore a lower surface energy results in a lower detachment force of the eggs tested on the hydrophobic substrate. The same correlation between surface chemistry and attachment is reported for the tarsal attachment system of several groups of insects [87,88,98,99]. The water-mobilized adhesive presumably does not wet hydrophobic surfaces properly and therefore attachment is reduced, as wetting is an important prerequisite for this type of adhesion [78,100]. The adhesive fluid and its composition presumably best perform on hydrophilic surfaces, most likely due to polarity within the adhesive fluid and water that is attracted to polar/hydrophilic surfaces [78].

The range of suitable surface chemistry regimes in insect attachment is presumably a result of co-evolution of the insects and their corresponding plants [95]. Adaptation to substrate chemistry is species-specific and depends on the degrees of specialization to various natural substrates. Unfortunately, not many aspects of the ecology and natural habitat of *P. philippinicum* are known. Therefore, assumptions about their host trees are based on diet compatibility with certain leaves from their endemic region [3] and known host species of closely related *Phyllium* [1,101]. As *P. philippinicum* are generalist phytophages, there are several potential host plants. Some of the supposed host species of *P. philippinicum* are *Psidium guajava* L. (Myrtaceae), *Mangifera indica* L. (Anacardiaceae) and *Nephelium lappaceum* L. (Sapindaceae) [1,3,101]. The higher attachment performance on hydrophilic surfaces does not allow to specifically determine the actual host plants but enables an approximation of natural substrates the eggs are adapted to. All three putative food plants are evergreen tree species which have leaves that are generally hydrophobic with water contact angles around 100◦ [102]. In contrast, the bark of guava and mango was estimated to have contact angles around 52◦ and 50◦ [103]. The eggs more likely adhere to the bark than to the leaves. This is also reflected in the coloration of the eggs and freshly hatched nymphs. Nymphs hatch a dark brown hue and change to green after feeding on the foliage of the tree. Further tests are certainly necessary to test this assumption, but based on the findings herein, *P. philippinicum*'s eggs have a better chance at attaching to wood surfaces of its host tree species compared with leaf surfaces. Additionally, anchorage of the egg to a substrate could help the new-born nymph successfully hatch from the inside of the egg while remaining arboreal.

Apparently, for the eggs of *P. philippinicum*, the surface chemistry influences egg adhesion more than surface roughness. In experiments with beetles, their tarsal attachment systems worked the other way around. The hairy, wet adhesive pads of *Coccinella septempunctata* Linnaeus, 1758 (Coleoptera: Coccinellidae) and *Leptinotarsa decemlineata* Say, 1824 (Coleoptera: Chrysomelidae) responded more to surface roughness than to surface chemistry [51,104]. However, *Gastrophysa. viridula* (De Geer, 1775) (Coleoptera: Chrysomelidae) exhibited a decreasing attachment performance with hydrophobicity and stronger performance on smooth surfaces [87]. On the other hand, the eggs of *P. philippinicum* show decreased attachment on hydrophobic surfaces, but attachment values were high for all surface roughnesses. However, for attachment of the eggs, in contrast to moving insects, the preconditions are different.

Two studies on egg adhesion of the codling moth used various natural surfaces with specific topographies and physiochemistries [44,45], however, the glue of the codling moth is permanent and most likely water-insoluble, and the adhesive of *P. philippinicum* is reversible and mostly water-soluble. Furthermore, these studies also found an effect of surface topography, which is probably due to the specific topography of the natural surfaces. Regardless, further investigation into the combined effect of surface chemistry and topography using a wider scope for both would be advantageous for the comparative effect on the attachment of *P. philippinicum* eggs.

#### *4.4. Glue Properties*

Detachment force values from *P. philippinicum* eggs in the wet condition were extremely low and practically negligible in comparison with other force measurements. It is assumed that the low force values in the wet condition reflects surface tension and capillary forces exerted by the water droplet that the egg was submerged in upon detachment, not the eggs' actual attachment to the surface.

Many insect eggs require hydration to become adhesive [55,57–66]. For example, eggs of the mayfly *Siphlonurus lacustris* (Eaton, 1870) (Ephemeroptera: Siphlonuridae) have a thick fibrous coat surrounding their eggs that undergo exochorionic changes once deposited in water, creating cohesion between egg masses and adhesion to a substrate [66]. However, these eggs attach in the wet condition and remain underwater. Contrastingly, the eggs examined herein need to dry after hydration to generate adhesion. The egg of *P. philippinicum* does not achieve attachment in water and necessitates a phase change from liquid to solid for the adhesive material to adhere. This could serve as a mechanism for the optimum site selection for the incubation of the egg to avoid adhesion under water, as attachment under water would be lethal to the hatching nymph. The eggs of *P. philippinicum* may be almost immediately ready to attach to a substrate once produced from the mother due to high humidity and the prevalence of water in a tropical rainforest [105]. This makes sufficient hydration of the egg highly probable and triggers its adhesive capabilities. However, if suitable conditions are not found, the egg retains the potential to adhere in a suitable environment in the next attachment event.

The glue mediated attachment of *P. philippinicum* eggs is reversible and reproducible over several cycles of attachment, detachment and reattachment (Figure 6C). Apparently, no glue is secreted by the egg itself; furthermore, colleterial glands for glue production are probably absent in the females as well [106], similar to some odonates and mayflies [65,66]. The fact that the attachment strength decreases after a few cycles of reattachment is probably due to the depletion in the supply of the glue covering the egg. The test surfaces revealed residuals of the glue after detachment (Figure 6D). However, the repeated ability to attach after submersion in water shows that the adhesive material is not entirely water-soluble and most of the secretion remains on the egg. In all likelihood, a hydrophilic polar portion of the material allows diffusion in water, facilitating adsorption at the glue interface [107] and consequently facilitates contact adaptation to the substrate [78]. A hydrophobic nonpolar portion most likely remains on the egg, preventing full dissolution [108]. The oviposition method that *P. philippinicum* females employ does not give them as much control over the oviposition site compared with other species that directly deposit eggs. The reversibility of adhesion may be a technique to correct maladaptive attachment sites or to adapt to seasonal changes in the environment, as reported for the egg glue of some alpine butterflies [47].

The chemical composition of the glue remains ambiguous and is not the subject of this study. However, some assumptions can be drawn based on the experimental results. Most permanent bioadhesives involved in egg attachment are largely proteinaceous [48,109–115]. The amphiphilic nature of the glue could be achieved by glycoproteins, as in many other insect glues [116]. The highly soluble glycan would serve as the polar portion [48,114,117], facilitating non-covalent bonding with hydrophilic substrates. The protein serves as the hydrophobic portion [108,118], providing adherence of the glue to the surface of the egg and its appendages.

#### **5. Conclusions**

Although the special appearance of the eggs of *Phyllium* species, including *P. philippinicum*, received attention in different taxonomic and evolutionary studies [15,23,42], only a few hypotheses on the function of the special morphological features were presented [3]. We herein elucidate the specialized attachment mechanism of the eggs of this species and provide the first experimental approach to systematically characterize the functional properties of their adhesion. The adhesive mechanism of the egg exploits a combination of folded exochorionic structures (pinnae) and a film of adhesive secretion. Both components respond to contact with water. The glue becomes fluid through hydration, adapts to the substrate profile and adheres after solidification. The pinnae facilitate the spreading of the glue, support adaptability using hierarchically splitting filaments and reinforce the hardened film. This mechanism copes with surface roughness using this combination but is affected by surface chemistry. The glue adheres very well to hydrophilic surfaces, but the attachment force decreases with an increasing water contact angle. Although the egg cannot achieve attachment while submerged in water, it can reattach itself after dislodgement from a surface, making its adhesive mechanism temporary, and arguably long-term [77], depending on the conditions. This replicability of attachment can accomplish attachment site optimization to ensure suitable environmental conditions for embryonic development. This includes fixation in preferable environmental conditions, but also adjustment in case of environmental changes. The mechanism described herein copes with different degrees of surface roughness but is affected by the surface chemistry of the substrate. Other adhesive secretions in insects interestingly perform differently, although they serve a very similar function: the larval glue of the fly *Drosophila melanogaster* consists of glycosylated proteins and is used to anchor the pupa to different substrates [119]. In contrast to the egg glue of *P. philippinicum*, the glue of *D. melanogaster* larvae adheres well to various substrates independent of their surface chemistry or roughness [120]. This leads to the assumption that the egg deposition in *P. philippinicum* favors hydrophilic substrates and suggests preferable deposition site selection for the eggs in the natural habitat.

Knowledge about this mechanism can support ecologists and conservationists. Elucidating the nature of the attachment mechanism helps in understanding the dispersal as well as the life history of the species. This can help in quantifying fecundity for conservation purposes of the insect species [120]. Information on the attachment sites can help the conservation of plants and gauging the population density [1,121]. The details of this potential transitory state between non-adhesive and permanently attached eggs can be useful for evolutionary biologists.

Furthermore, this repeatable and water-controlled mechanism can stimulate biomimetic research in the field of bioadhesives [48,122–124]. The origin and biochemical nature of the glue, however, remain elusive and should be subject to future studies.

**Author Contributions:** Conceptualization, T.H.B. and S.N.G.; methodology, T.H.B. and S.N.G.; validation, T.H.B. and S.N.G., formal analysis, T.H.B. and E.Q.; investigation, T.H.B. and E.Q.; resources, S.N.G.; data curation, T.H.B.; writing—original draft preparation, T.H.B.; writing—review and editing, S.N.G., E.Q. and T.H.B.; visualization, T.H.B. and E.Q.; supervision, T.H.B. and S.N.G.; project administration, S.N.G.; funding acquisition, S.N.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was partially funded by the German Science Foundation, DFG grant GO 995/34-1.

**Acknowledgments:** We thank Jan Michels (Department of Functional Morphology and Biomechanics, Kiel University, Germany) for support in microscopy techniques. Kirsten Weibert (Jena, Germany) is thanked for providing specimens for this study, Bruno Kneubühler (Zurich, Switzerland) and Daniel Dittmar (Berlin, Germany) for providing images. We acknowledge financial support by DFG within the funding programme Open Access Publizieren. E.Q. was financially supported by the European Commission through the program Erasmus Mundus Master Course—International Master in Applied Ecology" (EMMC-IMAE) (FPA 2023-0224/532524-1-FR-2012-1-ERA MUNDUS-EMMC). Coordination F.-J. Richard, Université de Poitiers, France.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **The E**ff**ect of Ground Type on the Jump Performance of Adults of the Locust** *Locusta migratoria manilensis***: A Preliminary Study**

#### **Chao Wan 1,2,\* , Rentian Cao <sup>1</sup> and Zhixiu Hao 1,\***


Received: 18 March 2020; Accepted: 12 April 2020; Published: 23 April 2020

**Abstract:** The jump performance of locusts depends on several physiological and environmental factors. Few studies have examined the effects of different ground types on the jump performance of locusts. Here, mature adult locusts (*Locusta migratoria manilensis*) were examined using a custom-developed measuring system to test their jump performance (including postural features, kinematics, and reaction forces) on three types of ground (sand, soil, and wood). Significant differences were primarily observed in the elevation angle at take-off, the tibial angle at take-off, and the component of the mass-specific reaction force along the aft direction of the insect body between wood and the other two ground types (sand and soil). Slippage of the tarsus and insertion of the tibia were often observed when the locusts jumped on sand and soil, respectively. Nevertheless, comparisons of the different parameters of jump initiation (i.e., take-off speed and mass-specific kinetic energy) did not reveal any differences among the three types of ground, indicating that locusts were able to achieve robust jump performance on various substrates. This study provides insights into the biomechanical basis of the locust jump on different types of ground and enhances our understanding of the mechanism underlying the locust jump.

**Keywords:** ground type; jump; locust; reaction force; kinematics; elevation angle

#### **1. Introduction**

Locusts are some of the most famous insects for their jumping ability, as they can achieve velocities as high as 2.6 m/s, accelerations as high as 75 m/s 2 , and can cover dozens of times their body length in a single jump [1]. Their jumps serve several critical functions: to escape from predators, to achieve an initial velocity for flight, and provide a more rapid alternative to travel than crawling. The locust jump has been extensively studied, especially its postural control, the mechanics of the hind leg, patterns of muscle and motoneuron activity, and mechanisms of energy storage and release [2–11]. The action of jumping in locusts is fueled by their hind legs in the following steps: initial flexion of the tibiae, co-contraction of the flexor and extensor muscles, and rapid tibial extension after trigger activity [2]. A large amount of strain energy is stored by the deformed exoskeleton during muscle co-contraction (especially by the semi-lunar process (SLP) cuticle on the distal end of the metathoracic femur) and is released during tibial extension to overcome the weaknesses of insect muscles [3,4].

Various physiological factors have been shown to have different effects on the jumping performance of locusts. Compared with immature juvenile hoppers, adult locusts have three times the range of escape jumps and at least twice the specific energy output [12,13]. Older juvenile insects hop less frequently than younger ones within the same instar because of increased anaerobic metabolism and locomotory fatigue [14]. Adaptive change in muscle contraction has also been observed in newly molted locusts to avoid cuticle damage during jumping [5]. In contrast, gravid females (20% heavier) have the same jump distance and significantly lower endurance compared with non-gravid females, as non-gravid females show a 20% increase in the duration of muscle contraction relative to gravid females [15,16]. Furthermore, Katz and Gosline [17] found that the take-off speed of the locust jump is relatively scale-independent (0.9–1.2 m/s for juveniles and 2.5 m/s for adults), showing that juvenile insects often jump to maximize the distance traveled while the purpose of jumping in adults more often serves to achieve a velocity necessary for initiating flight.

In addition, the effects of a few natural environmental factors have also been investigated. Hawlena et al. [18] reported that the chronic risk of predation can increase both the take-off speed and the jump distance of grasshoppers and that this pattern cannot be explained by morphological variation. Air resistance reduces the kinetic energy of locust jumps by less than 10% at lower initial speeds [19], and environmental temperatures ranging from 15 ◦C to 35 ◦C only weakly affect jump energy [20]. The properties of the ground are another set of environmental factors that can potentially affect the locust jump, especially during the take-off stage. For example, surface roughness with an Ra value of 1–2 µm can reduce the ability of locust legs to attach to the substrate, resulting in considerable slippage of the hind legs on the ground and thus take-off failure [21,22]. Similar effects of surface roughness have also been documented in females of the Mediterranean field cricket (*Gryllus bimaculatus*) crawling on smooth surfaces (R<sup>q</sup> = 7.3 µm), which resulted in significantly lower phonotactic responses compared with rougher surfaces (R<sup>q</sup> = 16 or 180 µm) [23]. Several natural types of ground surfaces have other key physical properties as well as surface roughness, such as normal stiffness, hardness, and tangential friction/shear stress strength. However, whether natural ground types can affect the jump performance of locusts has not been explored.

Here, we compared the kinematics, insect posture, and reaction force of the jump performance of adult locusts on three natural ground types (sand, soil, and wood). We hypothesized that the jump performance of locusts would differ among the three types of ground. Our results provide insight into the biomechanical basis of the locust jump on different types of ground and enhance our understanding of the mechanisms underlying the locust jump.

#### **2. Materials and Methods**

Mature Oriental migratory locusts (*Locusta migratoria manilensis*) were purchased from the Jiyuan locust-breeding facility in Anhui Province of China. Before tests, all locusts were fed with wheat leaves under natural light and room temperature (23–30 ◦C) for at least 2 weeks to ensure fully sclerotized cuticles after their final molt.

A custom-developed test system was built to simultaneously measure both the kinematics and reaction force of the locust jumps. Figure 1 shows the components of the test system, including a coordinate background plate (indicated by the letter G), two high-speed cameras (CR600 × 2, Optronis GmbH, Kehl, Germany; Hero4, GoPro Inc., San Mateo, CA, USA; indicated by letters D and E), a video camera (IXUS 210, Canon Inc., Tokyo, Japan; indicated by the letter F), a desktop for data acquisition (indicated by the letter H), and two custom-designed platforms (indicated by the letter B) supported by a stand column and a three-dimensional high-precision force sensor (S3-001NTO-003, Bio-inspired Technology Corp., Nanjing, China; indicated by the letter A). The two platforms were set in the same horizontal plane with 1 mm of space. The two high-speed cameras were placed on the side of the platforms. The GoPro camera (120 fps) was used to obtain the jump trajectory of the animals, and the Optronis camera (1000 fps) was used to record the rapid movement of their hind legs before take-off. The Canon camera was placed above the system to take images of insect posture and the jump azimuth. The force sensor was used to measure the three-dimensional reaction force from one hind leg of the locust during jumping (full-scaled range of 1 N, resolution of 1 mN, and sample frequency of 10 kHz).

**Figure 1.** A custom-developed test system for measuring the kinematics and reaction force of adult locusts while jumping. (**A**) Three-dimensional high-precision force sensor; (**B**) two platforms, one of which is fixed to the force sensor and the other fixed to a stand column; (**C**) adult locust whose left and right hind legs each stood on one platform; (**D**) a high-speed camera for obtaining the rapid movement of the hind leg during the jump; (**E**) a high-speed camera for obtaining the trajectory of the locust jump; (**F**) a video camera for imaging the posture of the locust from above; (**G**) a coordinate background plate; and (**H**) desktop computer for data acquisition.

Three ground types (sand, soil, and wood) were studied. First, balsa wood was fixed on the surface of the platforms to simulate wooden ground. The balsa wood was cut from some timbers and polished using moderate sandpaper (#: P400). All obvious visual burrs were removed from the wood surface. The locust was prepared by attaching its wings together using adhesive tape and placing it on the platforms such that its left and right hind legs each stood on one platform. Either a sudden sound or the touch of a brush was used to trigger the locust's escaping jump. Data were excluded for individuals that had two hind legs on the platform of the force sensor. Next, the balsa wood was replaced by two boxes (width × length × depth = 15 mm × 40 mm × 10 mm) that were filled with local sand (grain size of approximately 0.4 mm, density = 1.27 g/cm<sup>3</sup> ) to create a sandy substrate or commercial potting soil (soft and rich in organic matter, density = 0.317 g/cm<sup>3</sup> ) to create a soil substrate. Both of the fillings were lightly compacted, and the surface was carefully flattened before tests. Jumps of the locusts on the sand and the soil were tested following the aforementioned procedures. In total, nine adult locusts (4 females and 5 males) were included (body mass = 1.8 ± 0.51 g (mean ± SD)) and used for all the three ground types in this study. For avoiding the animal fatigue, only one ground type was tested in one day. For each ground type, each animal was tested less than seven trials and its first successful jump was selected to represent the jump performance of this individual. In brief, nine jumps were included for each ground type (one jump from each animal for each ground type). The sand, soil, and wood ground types corresponded to a granular substrate with weak cohesion, a granular substrate with strong cohesion, and a solid substrate, respectively.

*φ*

d

x y z

e

*φ* <sup>e</sup> arctan(cos( )tan( )) *φ β*

The yaw angle of a jump (ϕ) was obtained from the top camera image as the angle between the body axis of the locust and the direction *X* of the force sensor (Figure 2). The real distance of each locust jump (*S*d) was then obtained by correcting the camera recordings using the following equation:

$$S\_{\mathbf{d}} = \frac{\overline{S}\_{\mathbf{d}}}{\cos(\varphi)}\tag{1}$$

where *S*<sup>d</sup> is the distance recordings from the lateral images. The physiological components of the reaction force for the locusts were calculated from the force measurements of the sensor using the following equation: d

$$\overline{F}\_\mathbf{a} = F\_\mathbf{x} \cos(\boldsymbol{\varphi}) - F\_\mathbf{y} \sin(\boldsymbol{\varphi}),\\\overline{F}\_\mathbf{l} = F\_\mathbf{x} \sin(\boldsymbol{\varphi}) + F\_\mathbf{y} \cos(\boldsymbol{\varphi}),\\\overline{F}\_\mathbf{n} = F\_\mathbf{z} \tag{2}$$

where *F*x, *F*y, and *F*<sup>z</sup> are the measurements from the force sensor along its axes X, Y, and Z, respectively. *F*a, *F*<sup>l</sup> , and *F*<sup>n</sup> are the physiological components of the reaction force along the aft, lateral, and normal directions of the locust, respectively (Figure 2b). To eliminate the effect of body mass, the mass-specific reaction force was calculated by normalizing the physiological components with body mass as follows: a l n

$$F\_{\mathbf{a}} = \frac{\overline{F}\_{\mathbf{a}}}{M}, F\_{\mathbf{l}} = \frac{\left| \overline{F}\_{\mathbf{l}} \right|}{M}, F\_{\mathbf{n}} = \frac{\overline{F}\_{\mathbf{n}}}{M}, F\_{\mathbf{l}} = \sqrt{F\_{\mathbf{a}}^2 + F\_{\mathbf{l}}^2 + F\_{\mathbf{n}}^2} \tag{3}$$

where *M* is body mass. *F*a, *F*<sup>l</sup> , and *F*<sup>n</sup> are the aft, lateral, and normal components of the mass-specific reaction force, respectively. *F*<sup>t</sup> is the total magnitude of the mass-specific reaction force. The elevation angle of the locust at take-off (βt) was determined based on the real trajectory of the jump from the images of the lateral GoPro camera after take-off (Figure 2). a l n t t *β*

*φ* t *β* **Figure 2.** Schematic diagram illustrating how the kinematics and reaction force of the locust jumps were calculated. Because of the yaw angle (ϕ), the real trajectory of the jump was corrected from the image recordings of the sideway high-speed camera. The jump direction and real trajectory of the locust are indicated by the blue and red dashed lines, respectively. The black dashed line is parallel to both the lateral camera and the direction *X* of the force sensor. A<sup>0</sup> : the initial locust position; A<sup>1</sup> : the real locust position after jumping; B: the locust position in the camera images after jumping; and βt : the elevation angle of the locust at take-off. Similarly, the physiological components of the reaction force were transformed from the sensor measurements using the yaw angle. The coordinate system of the force sensor is shown by black axes, and the physiological coordinate system of the locust is shown by red axes.

*θ* e *β β* <sup>e</sup> *β* Three postural features of the locust during jumps were defined based on the camera images as follows (Figure 3). First, the opening angle between the hind femur and the central line of the body (θ) was measured from the image of the top camera just before tibial extension. Following a previous study [6], a line was drawn through the proximal femoral and the distal tibial ends of the locust just before tibial extension, and its tilt angle relative to the ground surface was defined as βe. Here, the angle β<sup>e</sup> was determined by both the projected tilt angle in the Optronis camera image (β<sup>e</sup> ) and the yaw angle ϕ as arctan(cos(ϕ)tan(β<sup>e</sup> )). Similarly, the angle between the tibial axis and the ground surface at take-off (γ) was calculated as arctan(cos(ϕ)tan(γ)), where γ is the projected tibia–ground angle in the Optronis camera image. Given the small waste in energy because of air resistance [19], the take-off speed (*V*t) was calculated based on the real jump distance *S*<sup>d</sup> and the angle β<sup>t</sup> as

t d t

$$V\_{\rm t} = \sqrt{\frac{gS\_{\rm d} \tan(\beta\_{\rm t})}{2} + \frac{gS\_{\rm d}}{2 \tan(\beta\_{\rm t})}}\tag{4}$$

*β*

where *g* is the acceleration of gravity (i.e., 9.8 m s−<sup>2</sup> ). The mass-specific kinetic energy (*E*m) for each jump was determined as 0.5*V* 2 t . m 2 0.5

t

*θ* e *β γ* **Figure 3.** Schematic diagram for three postural features of adult locusts while jumping. (**a**) The opening angle between the hind femur and central line of the body just before tibial extension (θ); (**b**) the tilt angle of the line through the proximal femur and distal tibia just before tibial extension (βe); and (**c**) the tibial angle relative to the ground at take-off (γ).

t *β* <sup>e</sup> *β* Statistical analyses were performed to clarify the effect of ground type on the locust jump. First, the normal distribution and homoscedasticity of the results were checked using Shapiro–Wilk tests and F-tests/Bartlett's tests, respectively. If the data met these requirements, *t*-tests were performed for all of the data between female and male animals and between the angles β<sup>t</sup> and β<sup>e</sup> for each ground type. Repeated one-way analyses of variance (ANOVA) with a Bonferroni post-hoc correction were also used to compare results between the three different ground types. Mann–Whitney tests were used instead of *t*-tests if the data were not normally distributed, and *t*-tests with Welch's correction were used if the data had unequal variances. For repeated one-way ANOVA, Friedman tests with Dunn's multiple comparisons were used if the data were either not normally distributed or variances were not homogeneous. Results were reported as mean ± SD. Significance was defined as *p* < 0.05.

#### **3. Results**

t

*θ*

*θ γ* <sup>t</sup> *β* <sup>e</sup> *β* <sup>t</sup> <sup>m</sup> *F*<sup>a</sup> *F*<sup>l</sup> *F*<sup>n</sup> *F*<sup>t</sup> Both the postural features (θ, γ, β<sup>t</sup> , and βe) and jump performance (*V*<sup>t</sup> and *E*m), including the reaction force (the maximal values of *F*a, *F*<sup>l</sup> , *F*n, and *F*t), of adult female locusts were not significantly (*p* > 0.05) different from those of adult male locusts regardless of ground type. As a result, data for both female and male locusts were pooled and analyzed together in subsequent statistical analyses. All of the experimental data are shown in Tables S1–S6 of the Supplementary Materials.

e *β* <sup>t</sup> *β* e *β* The four postural angles of the locust jumps on the three ground types are shown in Figure 4. The angles β<sup>e</sup> and β<sup>t</sup> were 57 ± 7.3 degrees and 36 ± 9.5 degrees, 52 ± 9.2 degrees and 43 ± 6.2 degrees, and 64 ± 12 degrees and 63 ± 12 degrees for the sand, soil, and wood substrates, respectively. Significant differences were detected in the angle β<sup>e</sup> between soil and wood (*p* < 0.05), in the angle β<sup>t</sup> between sand and wood (*p* < 0.001), and in the angle β<sup>t</sup> between soil and wood (*p* < 0.01). Comparisons between these two angles indicated that differences were only significant between sand and soil (*p* < 0.001). Moreover, the angle γ for the wood (68 ± 10 degrees) was significantly (*p* < 0.001) higher than that for

*γ*

*β*

*β* <sup>t</sup>

the sand (44 ± 8.6 degrees) as well as that for the soil (51 ± 6.5 degrees). By contrast, the angle θ was not significantly affected by changes in ground type.

t *β* e *β θ γ* **Figure 4.** Comparisons of the postural features of adult locust jumps on the three ground types: (**a**) the elevation angle of locust jumps at take-off (βt) and the tilt angle of the line through femoral proximal and tibial distal ends just before tibial extension (βe); (**b**) the open angle of the femur to the central line of the body (θ); and (**c**) the angle between the tibial axis and the ground at take-off (γ). Results were reported as mean ± SD. \* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001.

t m a a Figure 5 shows comparisons of the kinematics and reaction force among the three ground types. Neither the take-off speed (*V*t) nor the mass-specific kinetic energy (*E*m) was significantly different among the three ground types (2.3 ± 0.33 m/s and 2.7 ± 0.83 mJ/g for sand, 2.3 ± 0.13 m/s and 2.7 ± 0.31 mJ/g for soil, and 2.6 ± 0.17 m/s and 3.3 ± 0.42 mJ/g for wood). There were no significant differences in the mass-specific reaction force among the three ground types except for *F*a. The *F*<sup>a</sup> for wood (20 ± 6.3 mN/g) was significantly (*p* < 0.05) lower than that for sand (38 ± 5.6 mN/g) and soil (31 ± 10 mN/g). In addition, the hind legs of the locusts during the jumps firmly adhered to the wood substrate, often inserted into the sandy substrate (six of the nine jumps), or noticeably slid on the soil substrate (seven of the nine jumps). To illustrate the different interactions between the hind legs of locusts and the three ground types, some typical high-speed camera images and videos of the locust hind legs during jumping are provided in Figure 6 and the Supplementary Videos S1–S3.

t a

l

t m

n

t *β*

t m

e *β*

*θ γ*

a a

t m t a l **Figure 5.** Comparisons of the kinematics and reaction force of locust jumps on three ground types: (**a**) the take-off speed (*V*t); (**b**) the mass-specific kinetic energy (*E*m); (**c**) the total magnitude of the mass-specific reaction force (*F*t); (**d**) the aft component of the mass-specific reaction force (*F*a); (**e**) the absolute lateral component of the mass-specific reaction force (*F*<sup>l</sup> ); and (**f**) the normal component of the mass-specific reaction force (*F*n). Results were reported as mean ± SD. \* *p* < 0.05; \*\*\* *p* < 0.001.

**Figure 6.** High-speed camera images for the rapid movement of the hind legs of locusts while jumpin **Figure 6.** High-speed camera images for the rapid movement of the hind legs of locusts while jumping. Locusts jumped on sand (**a**), soil (**b**), and wood (**c**) ground types. The frame at 0 ms shows the configuration of the hind leg at take-off. To analyze the jump process, white dashed lines were marked for each panel to represent the location of the tibial distal end at 0 ms in the view of the camera.

*γ* <sup>t</sup>

*β*

t *β* <sup>e</sup> *β*

*β* <sup>e</sup>

t

t *β*

> e *β*

*β*

#### **4. Discussion**

We suggest that locusts have a robust jump sequence on various substrates. Both the take-off speed and the mass-specific kinetic energy of the locust jump were not significantly affected by the change in ground type even though tibial slip or insertion occurred on sand and soil substrates during the initial stage of jumping (Figure 6b,c). Given that the main purpose of jumping for adult locusts is to achieve a sufficient initial speed for initiating flight [17], the fact that the take-off speed did not change among the different ground types revealed that the jump behaviors of the locust were robust on various substrates. The kinetic energy of the locust jump was primarily released from the elastic strain energy stored by the SLP cuticle during the extension of the hind leg [3,4]. The similar mass-specific kinetic energy for the jumps on the three ground types meant that the stored strain energy was not wasted by the somewhat useless extension arising from the slip or insertion of tibia. This finding coincides with the previous observation that the semi-lunar processes of locusts do not start to unfurl (i.e., release energy) until the tibia extends by 55 degrees [24]. In other words, the specific mechanism underlying the storage and release of strain energy helps locusts achieve high jumping performance on different ground types because the energy does not begin to release during the initial extension, regardless as to whether the tibia slips on or inserts into the substrate. This property might enhance the ability of locusts to evade predators and thus their survival in different environments.

By contrast, the postural features of locusts among the three types of ground indicated that significant differences were primarily observed in the elevation angle at take-off (βt) and the tibial angle at take-off (γ) (Figure 4). Sutton and Burrows [6] found that the angle β<sup>t</sup> was almost the same as the tilt angle of the femoral proximal femur–distal tibia line before tibial extension (βe) when locusts jumped on wood. Similar results were obtained by our measurements in that the difference between the two angles β<sup>t</sup> and β<sup>e</sup> was only 1.5 ± 0.83 degrees for jumps on wood. By contrast, the angle β<sup>t</sup> was significantly higher than the angle β<sup>e</sup> for jumps on the other two ground types (sand and soil). The differences between the three ground types stem from the specific conditions of the tibia on the ground (i.e., sliding or insertion). It was clearly illustrated in Figure 6 that the locust tibia firmly adhered to wood but often inserted into sand or slid on soil during the take-off process. The sliding or insertion altered the swinging angle of the hind leg, resulting in an altered take-off elevation angle. Although locusts can use their claws and tarsal adhesive pads to firmly stick to a few types of substrates [25,26], these strategies appear to be useless for successfully attaching to natural ground types, such as sand and soil. In addition, the significant difference in the angle γ resulted from the different β<sup>t</sup> among the three types of ground following the quantitative relationship between these two angles established by Sutton and Burrows [6].

Another interesting finding is that only the aft component of the mass-specific reaction force (*F*a) was significantly different among the three ground types while its total magnitude and other components were not (Figure 5). This finding might stem from the differing effect of the tibial angle on the three physiological components. Given that the total reaction force (*F*t) is along the tibial axis, its aft and normal components were calculated by the decomposition theory of force as *F*<sup>t</sup> cos(γ) and *F*<sup>t</sup> sin(γ), respectively. Thus, a higher tibial angle leads to a reduced aft component and an elevated normal component. Although the *F*<sup>t</sup> values did not significantly change among the three ground types, the aft component was increased by 0.35*F*<sup>t</sup> (from 0.37*F*<sup>t</sup> to 0.72*F*t) when the angle γ changed from 68 degrees (on wood) to 44 degrees (on sand) and by 0.26*F*<sup>t</sup> (from 0.37*F*<sup>t</sup> to 0.63*F*t) when the angle γ changed from 68 degrees (on wood) to 51 degrees (on soil). Similarly, the normal component decreased by 0.24*F*<sup>t</sup> (from 0.93*F*<sup>t</sup> to 0.69*F*t) when the angle γ changed from 68 degrees (on wood) to 44 degrees (on sand) and by 0.15*F*<sup>t</sup> (from 0.93*F*<sup>t</sup> to 0.78*F*t) when the angle γ changed from 68 degrees (on wood) to 51 degrees (on soil). Briefly, the influence of the angle γ on *F*<sup>a</sup> was 1.5–1.7 times that of *F*n, which might explain why significant differences only existed in *F*<sup>a</sup> among the different ground types.

However, some limitations of this study require consideration. First, the sample size of jumps for each ground type was small, although paired statistical analyses were used in the comparisons. It should be emphasized that the results of this study are preliminary and should be used as a basis

from which future studies examining the effect of ground type on locust jumps could be conducted. Second, the properties of the three ground types need to be quantitatively measured. Based on these measurements, the quantitative relationship between ground properties and locust jump behaviors could be characterized and provide some fundamental mechanical data for optimizing the jumping tactics of bioinspired robots on different ground types.

#### **5. Conclusions**

In this study, jumps of nine adult *L. m. manilensis* locusts on three different types of ground (sand, soil, and wood) were measured using a custom-developed test system. Specifically, measurements were made of the postural features, kinematics, and reaction force. Both the elevation angle β<sup>t</sup> and tibial angle γ at take-off were significantly different among the three types of ground, which might have been caused by the hind legs slipping on or inserting into the ground. Nevertheless, the jumping kinematics (including the take-off speed and the mass-specific kinetic energy) were not significantly different among the different ground types, indicating that locusts were able to achieve robust jumping performance on the various substrates. This study provides preliminary data that contribute to enhancing our understanding of the jumping mechanisms in locusts, especially how jumping behaviors are adapted to different types of ground.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2075-4450/11/4/259/s1, Table S1: Experimental data of the kinematics for the locusts jumping on sand, Table S2: Experimental data of the kinematics for the locusts jumping on soil, Table S3. Experimental data of the kinematics for the locusts jumping on wood, Table S4: Experimental data of the reaction force for the locusts jumping on sand, Table S5: Experimental data of the reaction force for the locusts jumping on soil, Table S6: Experimental data of the reaction force for the locusts jumping on wood, Video S1: A typical video of the locust jumping on sand, Video S2: A typical video of the locust jumping on soil, Video S3: A typical video of the locust jumping on wood.

**Author Contributions:** Conceptualization, C.W. and Z.H.; experiment and formal analysis, R.C. and C.W.; writing—original draft preparation, C.W.; writing—review and editing, Z.H.; funding acquisition, C.W. and Z.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by grants from the State Key Laboratory of Tribology (grant number: SKLT2015B06, SKLT2018B07) and the Beijing Institute of Technology Research Fund Program for Young Scholars (grant number: 3052019066).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Insects* Editorial Office E-mail: insects@mdpi.com www.mdpi.com/journal/insects

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18

www.mdpi.com ISBN 978-3-0365-2892-2