An insect trap as habitat: cohesion-failure mechanism prevents adhesion of Pameridea roridulae bugs to the sticky surface of the plant Roridula gorgonias

Glandular trichomes that release sticky fluids are common in a variety of plant species. They can have antixenotic effects on insects and mites,preventing them from settling and damaging plant surfaces (e.g. Levin, 1973). Several carnivorous and protocarnivorous plants use `fly-paper' traps with tentacles and/or capitate trichomes to capture their prey(Lloyd, 1942; Juniper, 1986). The released glandular fluids make the surface shiny, and, by means of these reflective surface properties, a large number of insects are attracted and can be suddenly entangled when coming in contact with the sticky surface of the plant.

Viscid secretions at trichome tips can be, in general, of three kinds: oily(often aromatic), mucilaginous and resinous(Lloyd, 1942). In the genera Byblis, Drosera, Drosophyllum and Pinguicula, a hydrous mucilage is found, whereas, in the South African perennial shrubs Roridula gorgonias Planch. and R. dentata L. (Roridulaceae), the transparent exudates of their glands are very different from those of Drosera (Lloyd,1934). The water-insoluble pellucid droplets of Roridulaplants have been observed to persist indefinitely on dead and dried-up leaves(Bruce, 1907) or leaves preserved in formalin (Lloyd,1934). Marloth (Marloth,1925) washed dried leaves with chloroform and found a very viscid resinous residue containing ∼10% of caoutchouc. These results were confirmed by Lloyd (Lloyd,1934), who extracted fresh leaves with acetone, yielding a resin or a mixture of resins. Following this by extraction with petroleum ether yielded an acetone-insoluble material having properties of caoutchouc. However, no distinct chemical analyses on the composition of the Roridula secretion have been performed.

The presumably resinous nature led to the conclusion that proteolytic enzymes can be neither transported nor dissolved in the adhesive secretion(Marloth, 1925). That is why the `carnivorous syndrome' of Roridula plants is considered controversial. Primarily, these plants were supposed to be carnivorous because of the similarity of the tentacle-shaped trichomes to those in representatives of the genera Drosera and Drosophyllum(Darwin, 1875; Marloth, 1903; Fenner, 1904; Bruce, 1907). As no digestive organs in the trichomes and leaves were found, Roridula was placed in the group of protocarnivorous plants – those that trap insects without the ability to digest them (Marloth,1910; Marloth,1925; Lloyd, 1934; Lloyd, 1942; Juniper et al., 1989). However, Midgley and Stock (Midgley and Stock, 1998) found higher levels of nitrogen and stronger ultraviolet reflectivity in the mature leaves of plants that captured insects,compared with control plants. These data confirm the carnivorous nature of R. gorgonias, despite the apparent lack of proteolytic enzymes. However, Plachno and colleagues (Plachno et al., 2006) observed phosphatase activity in the leaf epidermis using enzyme-labeled fluorescence microscopy.

The leaf surface of Roridula is covered with numerous tentacle-shaped and capitate trichomes of different size but similar structure, concentrated along leaf margins and the main vein(Fenner, 1904; Bruce, 1907; Uphof, 1962). The adhesive secretion of the trichomes is extremely effective, as shown by the conspicuous number of trapped insects of considerable size and mass, particularly flying ones (Marloth, 1910; Barthlott et al., 2004). Hartmeyer (Hartmeyer, 1998)suggested that the secretion of Roridula is the strongest glue of all insect-trapping plants. Stuck insect parts and entire corpses of occasionally considerable size and body mass have been found on the plant surface(Marloth, 1903; Marloth, 1910). Mean prey length averaged 3.55±0.57 mm (±s.d.; N=109)(Ellis and Midgley, 1996). However, no representative systematic study of the insect species trapped by Roridula plants has been performed to date. From a small sample of R. dentata, representatives of Hymenoptera (25 individuals from subfamilies Sphecinae and Apinae), Diptera (20 individuals from the family Muscidae), Coleoptera (a few individuals from the families Coccinellidae and Scarabaeidae), Hemiptera (single individuals from the families Lygaeidae,Reduviidae and Membracidae) and Lepidoptera have been reported previously,whereas bees, wasps and flies were the most abundant specimens found(Marloth, 1903; Marloth, 1910). Samplings of individual leaf rosettes of 15 R. gorgonias plants after an eight-week period resulted in a total of 109 trapped individuals belonging to 32 macro-invertebrate species (>2mm in length) consisting mainly of dipterans and coleopterans (Ellis and Midgley, 1996). Additionally, 122 micro-invertebrates (<2mm in length) mainly from the orders Thysanoptera and Diptera have been counted. Illustrating their potency, South African farmers call the plant `Vliegebos'and suspend the plants in their houses as flytraps(Marloth, 1925).

However, mirid bugs of the genus Pameridea (Heteroptera, Miridae,Bryocorinae, Dicyphini) are obligately associated with Roridulaplants (Reuter, 1907; Dolling and Palmer, 1991; Picker et al., 2004) in a form of digestive and pollinating mutualism(Ellis and Midgley, 1996; Reiner, 2003; Anderson, 2005; Anderson, 2006; Anderson and Midgley, 2002; Anderson and Midgley, 2003; Anderson and Midgley, 2007; Anderson et al., 2003). They live omnivorously and walk confidently and quickly on the sticky plant surface without becoming entangled and without hindrance. The bugs feed on the glued insects and defecate on the leaves(Marloth, 1903; Lloyd, 1934). The nitrogen in the faeces of the bugs is absorbed through the thin leaf cuticle of the plant,thus providing up to 70% of the total plant nitrogen uptake(Ellis and Midgley, 1996; Anderson and Midgley, 2002; Anderson and Midgley, 2003). Similar relationships between mirid bugs and carnivorous plants are known from representatives of the genera Byblis and Drosera in Australia (Schuh, 1995). Moreover, mirid bugs from the subfamilies Orthotylinae and Bryocorinae seem to be specialized for living on glandular hairy plants(Reuter, 1913; Dolling and Palmer, 1991; Falkingham, 1995; Schuh, 1995; Wheeler, 2001; Sugiura and Yamazaki, 2006). For example, the mirid species Dicyphus errans Wolff (Bryocorinae,Dicyphini) avoids contact with sticky glandular secretions by means of morphological (slim body, long and slender legs, elongated curved claws) and behavioral (mode of locomotion, grooming) adaptations to hairy plant substrates (Southwood, 1986; Voigt et al., 2006a; Voigt et al., 2007). D. errans is often observed grooming various body parts(Voigt, 2005). By contrast, Pameridea bugs frequently touch the released viscid plant secretion,which is spread evenly over ovoid glands and often over the trichome multiseriate stalks and the plant stem(Marloth, 1910).

Previous authors (Lloyd,1934; Hartmeyer,1996; Hartmeyer,1998) have suggested that bugs must have some kind of defense and resistance mechanisms to adhesive substances such as a sophisticated arrangement of short bristles capable of protecting the body against adhesives, specialized body cleaning adaptations and/or a locomotion mode in which the body is kept elevated above sticky trichomes. Intensive body grooming has been reported previously to be an essential adaptation of mirid bugs to living permanently, and walking, on hairy and glandular hairy plant substrates (Kullenberg, 1946; Voigt et al., 2006b). Our preliminary observations showed that Pameridea roridulae Reuter,after being wrapped in a sticky leaf of R. gorgonias, continued normal walking without body grooming.

Why does P. roridulae not stick to the adhesive secretion, whereas numerous other insect species do? Is such an anti-adhesive property of the mirid bug surface related to some specialized microstructure(Fig. 1C), similar to that previously described on unwettable surfaces of aquatic bugs (Anderson, 1976;Anderson, 1977; Anderson, 1982; Perez Goodwyn, in press; Perez Goodwyn et al., 2008)? Does the solid epicuticle have a strong repelling ability to nonpolar fluids (Fig. 1B)? Are there some fluids or easy-to-break solid layers on the epicuticle that function to prevent direct contact between plant secretions and the insect surface (Fig. 1D,E), as suggested by Lloyd(Lloyd, 1934)?

This study was performed to analyze the surface structure and adhesive characteristics of insect cuticle in contact with glandular droplets of trichomes of R. gorgonias. Using light and cryo-scanning electron microscopy (cryo-SEM), adhesive droplets of the plant, the surface of the insect cuticle and its prints on a glass surface were visualized in the intact state and after treatment with different solvents. The adhesion forces of adhesive droplets of R. gorgonias on the mutualistic mirid bug P. roridulae, the representative prey insect Calliphora vicinaRob.-Des. (Diptera, Calliphoridae) and a glass surface (control) were estimated through experimentation.

Seeded R. gorgonias plants, of age 1–3 years, occupied by P. roridulae, were obtained from a private greenhouse culture (Klaus Keller, Augsburg, Germany), kept under laboratory conditions during experiments (23.7±1.7°C, 47.3±10.0% relative humidity, 16h photoperiod) and fed with wingless, adult Drosophila melanogasterMeigen fruit flies (Diptera, Drosophilidae; Zoo-Schöniger, Stuttgart,Germany).

Adult C. vicina blow-flies were reared from commercially offered larvae (Angelmarkt Stephan, Stuttgart, Germany).

A stereomicroscope Olympus SZX 12 with a DF PLAPO 1×PF objective(Olympus Corporation, Tokyo, Japan) was used to observe P. roridulaemirid bugs and captured insects on R. gorgonias. Images were taken using a Nikon Coolpix E995 digital camera adapted to the stereomicroscope with a C-Mount adapter and a MDC 2 relay lens MXA 29005 (Nikon Corporation, Tokyo,Japan).

Cuticle prints of living and dead P. roridulae and C. vicina, pressed against a glass slide, were visualized using an upright AXIOPLAN microscope with an AxioCam MRc digital camera (Carl Zeiss MicroImaging GmbH, Jena, Germany) and AxioVision 3.1 software (AxioVision GmbH, München-Hallbergmoos, Germany).

To visualize the details of interactions between plant and insect surfaces in the freshly frozen condition, microscopy studies were carried out using a Hitachi S-4800 cryo-SEM (Hitachi High-Technologies Corp., Tokyo, Japan)equipped with a Gatan ALTO 2500 cryo-preparation system (Gatan, Inc.,Abingdon, UK). Samples were mounted on metal holders, frozen in the preparation chamber at –140°C, sputter-coated with gold–palladium (3nm) and examined in a frozen state in the cryo-SEM at 3 kV accelerating voltage while at –120°C.

Fresh samples of R. gorgonias, P. roridulae and C. vicinawere prepared and studied as described above.

Cryo-SEM has been reported previously to be a successful method for visualizing droplets of glycerine, water, lipids and other biological fluids and their mixtures (Gorb et al.,2007). Using tweezers, droplets of single tentacle-shaped and capitate trichomes were manually brought into contact with surfaces of freshly killed insects (C. vicina, P. roridulae), and small hemispherical pieces of smooth Au–Pa-metallized epoxy resin Spurr(Spurr, 1969; Gorb, 2006), and examined according to the method described above. The solid metallized Spurr surface is known to increase the material contrast between the substrate and fluid droplets as well as, owing to the substrate profile, allowing the observation of droplets at an appreciable angle (Gorb 2006, Gorb et al.,2007).

Using tweezers, single tentacle-shaped trichomes with glandular droplets were removed from the leaf margin, washed for 5 min in ethanol (Rotipuran®≥99.8%, p.a., Carl Roth GmbH & Co. KG, Karlsruhe, Germany), acetone(Rotipuran® ≥99.8%, p.a., ACS, ISO, Carl Roth) or cold chloroform(≥99.8%, p.a., Merck KgaA, Darmstadt, Germany), placed on sample holders and observed as described above. The results were compared with data on untreated trichomes and those washed with aqua millipore.

Samples of P. roridulae and C. vicina – living,dead, and dead washed with cold chloroform (5 min) – were mounted on holders. Using a cold scalpel, fractures of frozen legs were performed in the preparation chamber of the cryo-SEM at –140°C. Next, samples were sputter-coated in the frozen condition (3 nm thickness of Au–Pa) and examined in the cryo-SEM at –120°C and an accelerating voltage of 3 kV.

From digital images of 10 randomly selected points, the thickness of the epicuticular grease layer was estimated, using Sigma Scan Pro 5 (SPSS, Inc.,Chicago, IL, USA) software. The data obtained were statistically processed using Kruskal–Wallis one-way ANOVA on ranks and an all pairwise multiple comparison Tukey test (SigmaStat 3.1.1® software, Systat Software, Inc.,Richmond, CA, USA).

The measurements were carried out using a force transducer (10g capacity,Biopac Systems, Santa Barbara, CA, USA), attached to a motorized DC3314R micromanipulator with MS314 controller (World Precision Instruments, Sarasota,FL, USA). A piece of double-sided carbon tape (5×5mm) was firmly attached to the force transducer. Using tweezers, a tentacle-like glandular trichome was removed from the leaf margin and attached by its base to the tape, perpendicular to the force transducer(Fig. 2). The trichome could be moved up and down with a velocity of 100μm s^–1. The adhesive droplet on the trichome tip was brought into contact with the insect cuticle of the ventral side of the abdomen (living and dead bugs and flies, as well as chloroform-washed bugs) or with the glass surface. The droplet was preloaded to a maximum of 50μN force and then withdrawn. Force–time curves were used to estimate the maximum pull-off (adhesion) force. For each test surface, 20 measurements on different sites (N=5 insects, n=4 measurements per insect), and 120 single measurements in total were carried out (N=6 surfaces, n=20 measurements per surface). Kruskal–Wallis one-way ANOVA on ranks followed by an all pairwise multiple comparison procedure (Tukey test) was used to evaluate differences in the adhesion force values between the substrates (SigmaStat 3.1.1® software, Systat Software, Inc.). Laboratory conditions were similar to those mentioned previously.

Using a few representative force–time curves, measured on both living P. roridulae (N=3) and C. vicina (N=3),force–distance curves were calculated to estimate the work necessary to retract the adhering plant trichome from an insect cuticle at a distance of 1.5 mm.

Leaves of R. gorgonias are densely covered with capitate trichomes of varying sizes, mainly situated on leaf margins as well as on adaxial and abaxial mid veins. All instars of P. roridulae bugs moved easily on the adhesive glandular plant surface (Fig. 3A–D). To attach to the plant during locomotion, mirid bugs either hooked themselves to single trichomes using claws or attached to the flat leaf surface between the trichomes apparently by using their adhesive pads, called pseudopulvilli. The body and especially the legs were held close to adhesive secretion droplets and contacted them frequently without being trapped. However, representatives of Homoptera, Lepidoptera, Coleoptera and Diptera were observed to be caught immediately after touching the plant surface, always with either the ventral or lateral body surface. An individual of D. melanogaster, for example, can be trapped by a single capitate trichome (Fig. 3E). Prey insects that struggled intensively to free themselves after capture were held by sticky secretion droplets of two or numerous trichomes. Insect activity apparently effected an increase in the number of droplets in contact with the body of the insect (Fig. 3F,G).

The capitate and tentacle-shaped trichomes of R. gorgonias consist of multicellular stalks and multicellular glandular heads releasing the adhesive secretion (Fig. 4). The width in the middle of secretion droplets ranged from 129 to 405 μm(226.4±54.8 μm, mean ± s.d., N=120). The consistency of the secretion seems to vary slightly between the very long, tentacle-shaped and shorter capitate trichomes. The latter possess round, apparently more viscous, stable droplets (Fig. 4B,C) that appear rough in the cryo-SEM preparations(Fig. 4D). The tips of the longer tentacle-shaped trichomes are equipped with ovoid, apparently less viscous, secretion droplets, having a smooth surface at high magnification(×1000) in the cryo-SEM (Fig. 4E). The adhesive fluid of different trichomes can spread over the trichome stalk (Fig. 4E,F) and even the leaf lamina (Fig. 4B). Furthermore, we observed fluid filaments of up to 5 cm in length into which the secretion, in particular those produced by the tentacle-shaped trichomes,can be pulled after contacting a surface(Fig. 3F,H, Fig. 4A).

Treatment of the secretion droplets of the tentacle-shaped trichomes with different fluids demonstrated differences in droplet solubility. The droplets kept their volume and ovoid shape after treatment with aqua millipore(Fig. 5A). After trichome treatment with absolute ethanol, they were deformed, and the secretion was partially removed such that some intact, convex glandular cells could be seen(Fig. 5B,C).

After treatment with chloroform, fibrous residues of the glandular secretion covering the trichome tips were found(Fig. 5E). The stalk and gland of the trichome collapsed (Fig. 5D). Treatment of the trichome with acetone resulted in total removal of the glandular secretion from the trichome tip. The terminal gland with its opening was then clearly visible.

Intact droplets were manually attached to insect cuticles or metallized epoxy-resin surfaces. Contact formation between the glandular tip of tentacle-shaped trichomes and P. roridulae was hardly detectable. The secretion spreads on the surface of the mirid bug(Fig. 6A,B). However, on single bug setae, residual round droplets of the secretion were observed(Fig. 6C). In contrast to the mirid bug, the surface of the fly C. vicina induced formation of round, droplet-shaped patches with a distinct boundary(Fig. 6D–F). Similar droplets formed on the surface of epoxy resin coated with Au–Pa(Fig. 6G,H). Furthermore,numerous flat micro-droplets were observed when the tip of a tentacle-shaped trichome briefly contacted the metallized surface(Fig. 6I).

The measured adhesive force depended on the surface to which the tentacle-shaped trichomes were brought into contact(Fig. 7). The lowest values were obtained on living mirid bug cuticle, and the highest ones on the cuticle of living and dry flies, as well as on the cuticle of chloroform-washed mirid bugs. These differences were statistically highly significant. During the experiment, secretion frequently formed filaments with lengths ranging from 2 to 51 mm in 48% of the individual tests (for details, see Materials and methods). The length of filaments (l) did not influence the measured pull-off force (F_A)(F_A=158.20–0.68l, r²=0.01, F_1,58=0.41, P=0.52, linear-regression analysis). The work required to retract the trichome from the insect cuticle was estimated from force–distance curves(Fig. 8). It was much lower(0.07±0.026 J) in living P. roridulae than in living C. vicina (0.18±0.093 J) (for each species, N=3).

Both insect species studied have an integument that bears large setae(Fig. 6A,D) and a layer of much smaller, procumbent microtrichia. When light pressure was applied by a clean glass slide to the insect cuticle, prints of epicuticle grease became visible in the phase-contrast mode of a light microscope(Fig. 9). The cuticle of living and dried P. roridulae mirid bugs left a distinct amount of fluid residues (Fig. 9A–D), in most cases forming a continuous layer on the glass surface. In contrast to the mirid bug cuticle, cuticle prints of living C. vicina flies were seen as discrete micro-droplets on the glass slide(Fig. 9E,F).

Cryo-SEM fractures clearly revealed an amorphous epicuticular grease layer on the surface of the cuticle in living and dry P. roridulae. This layer covers the entire insect surface, including procumbent microtrichia(Fig. 10A,B,D,E). In living and dry C. vicina, such a clear grease layer was not observed. The thickness values of the layer were significantly higher (∼30-fold) in living and dry mirid bugs compared with those of living and dry flies(Table 1). In both insects treated with chloroform, the grease layers seemed to be nearly removed(significantly lower thickness values were measured; see Table 1). Furthermore, in chloroform-rinsed mirid bugs, a clean cuticle with clear uncovered microtrichia was found (Fig. 10F, compare with Fig. 10D,E).

It has been reported previously that the surface of R. gorgoniasis very sticky (Marloth, 1910; Barthlott et al., 2004). We showed here that not only the tips of tentacle-shaped trichomes, but also their stalks and the leaf lamina, are mostly coated with an adhesive secretion. This coverage is responsible for the adhesive property of almost the entire plant. The adhesive secretion is released in the form of shiny droplets on the ovoid glandular heads of numerous capitate trichomes of varying lengths. Very long tentacle-shaped trichomes, shorter trichomes and very short capitate trichomes together comprise a three-dimensional trap. Once stuck to a fluid glandular droplet of a long, flexible, tentacle-shaped trichome, an insect will struggle and move, pulling the droplet into a long fluid filament. A moving insect usually remains trapped and contacts further droplets from the long tentacle-shaped trichomes. The arrangement of trichomes in combination with the insect behavior results in an entanglement of the insect prey between trichomes and pulled filaments. Finally, the insects,caught by the presumably more viscous secretion of the stiffer short capitate trichomes, adhere firmly to the plant surface. Our results show that the work required to break contacts between the fluid filaments of the plant and the insect surface is higher in flies than in mirid bugs. This result suggests that struggling flies, compared with P. roridulae mirid bugs, must apply considerably more energy to free themselves from the plant.

Captured insects were usually observed sticking ventrally or laterally to the plant surface. Insects initially contact the plant surface with their feet but finally become trapped at their ventral and lateral surfaces, which seem to be more sensitive to the sticky traps.

The presence of long, thin, extensible fluid filaments, caused by the pulling of adhesive droplets, and their ability to recover their shape afterwards, supports previous assumptions about the existence of a caoutchouc-like fluid in this plant(Marloth, 1925; Lloyd, 1934). This behavior indicates that the secretion possesses a viscoelastic property. The trichomes seem to release substances of varying viscosity. The fibrous residues we found on trichome tips after washing them with cold chloroform might indicate a composite nature of the secretion consisting of a fibrous network embedded in a fluid matrix. Such a composition of the secretion could well explain its behavior as being similar to that of an elastomere with fluid-like behavior. Our efforts in visualizing solvent-treated droplets indicate the presence of a multicomponent substance, partly soluble in ethanol and chloroform and entirely soluble in acetone. These solubility characteristics suggest that the adhesive fluid might comprise various lipids and terpenes (polar resins). Thus, our results support previous observations that the adhesive secretion has a resinous character (Marloth,1925; Lloyd,1934). However, we did not observe an acetone-insoluble material,as reported by Marloth (Marloth,1925). Terpenes in plant secretions have been reported to be released in great diversity and, normally, as a mixture of several terpenes(Michie and Reid, 1968; Schnepf, 1969; Dell and McComb, 1978). Fatty or oily substances have been found as concomitant fluids of essential oils. In the sticky sage Salvia glutinosa L. (Lamiaceae), for example, sticky secretions of glandular trichomes contain a large amount of lipids(Schnepf, 1969). In addition to the solvent-treatment experiment, the probable presence of lipid-like components was supported by the particular pattern of droplet prints spreading into numerous micro-droplets on a metallized epoxy-resin surface. The formation of such extremely small fluid droplets is hardly possible for a highly viscous material. That is why we assume that the margins of the large droplets contain oily substances. Also, the affinity of the adhesive secretion to the grease-covered, hydrophobic insect cuticle is an additional indicator of the possible presence of oily substances in the plant adhesive fluid.

The adhesive secretion of R. gorgonias has been reported previously to be very viscous and to be the strongest glue of all plant flypapers (Hartmeyer, 1998). In our adhesion measurements, we obtained pull-off forces of the secretion in tentacle-shaped trichomes at the micro-Newton scale. In the protocarnivorous tar flower Befaria racemosa Venten (Ericaceae), the stickiness of the viscid secretion corresponded to that of commercial flypapers and ranged between 40 and 50 kPa (Eisner and Aneshansley, 1983). If we assume the width of secretion droplets(∼100 μm; see Fig. 6I)to be the contact area, we can estimate the adhesive strength in R. gorgonias as being ∼13 kPa. Such a discrepancy between our results and those in the literature might be explained by the fact that our experimental design differed drastically from that used in experiments on B. racemosa. Eisner and Aneshansley(Eisner and Aneshansley, 1983)measured the pressure-dependent adhesive strength on the fluid trapped between two flat surfaces, whereas we estimated the adhesion strength of a single ovoid droplet slightly brought into contact with the insect or glass surface and then retracted from it. Direct comparison of these two approaches is not possible because of the strong differences in the fluid thickness (a thinner adhesive layer attaches two bodies much more strongly), which was not measured in both studies. Our experiment was designed to simulate the natural situation in which insects contact the surface of R. gorgonias (usually through long tentacle-shaped trichomes) and then pull the adhesive secretion into the form of filaments. This is also the reason why we used long tentacle-shaped trichomes in our experiments.

P. roridulae bugs move easily and quickly on R. gorgonias, adhering neither to trichomes nor to the epidermis. Related mirid species avoid contact with sticky plant fluids by means of a combination of morphological and behavioral adaptations(Southwood, 1986; Voigt et al., 2006a; Voigt et al., 2007), whereas P. roridulae bugs seem not to avoid particular plant parts, even regions between the trichomes, and frequently touch the glandular fluid. We observed that they also seldom groom their body, contrary to previous suggestions (Lloyd, 1934; Hartmeyer, 1996; Hartmeyer, 1998). These authors discussed the arrangement of short setae as being the mechanism protecting the mirid bug against adhesives. Our comparative SEM study has revealed that the surfaces of both P. roridulae bugs and C. vicina flies are covered with cuticle outgrowths of different origins. Adhesive droplets adhere to cuticle protuberances of both mirid bugs and flies. However, flies are caught by R. gorgonias, whereas mirid bugs are not caught. Thus, we presume that the hairy covering of the surface is not the main reason for the resistance of the bug to the sticky secretion. The present study shows that P. roridulae bugs employ a different mechanism to resist plant adhesion.

Our experiments show that the adhesion force was significantly lower on surfaces of live and dead P. roridulae bugs compared with that of live and dead C. vicina as well as compared with that of chloroform-washed mirid bugs (the latter still bear a hairy covering). That is why our results support the assumption of Lloyd(Lloyd, 1934) that mirid bugs release some substances repelling the adhesive secretion of the plant. Even if the greasy layer does not really repel the adhesive secretion but rather seems to have a strong affinity for it (the adhesive wets the surface of the bug),this layer sloughs off easily (cohesion failure), preventing trapping of the bug. According to both light-microscopic data on cuticle prints of P. roridulae and to the cryo-SEM results showing the distinct amorphous layer on the surface of the epicuticle of P. roridulae, we hypothesize that an epicuticular, greasy substance, removable with chloroform,is responsible for the anti-adhesive property of the mirid bug surface. Lipid-containing epicuticular grease has been reported previously from other arthropods: Calliphora erythrocephala Meigen (Diptera, Calliphoridae)(Wolfe, 1954), Rhodnius prolixus Stål (Wigglesworth,1933) (Heteroptera, Reduviidae), Periplaneta sp.(Blattaria, Blattidae) (Beament,1945; Beament,1958; Gilby and Cox,1963), Acheta domesticus L. (Orthoptera, Gryllidae)(Hendricks and Hadley, 1983), Locusta migratoria L. (Orthoptera, Acrididae)(Vötsch et al., 2002), Leptinotarsa decemlineata Say (Coleoptera, Chrysomelidae)(Tower, 1906), Cicindelinae(Carabidae, Coleoptera) (Stegemann,1929), Boophilus microplus Canestrini (Acarina, Ixodidae)(Gilby, 1957) and Cupiennius salei Keys (Araneae, Ctenidae)(McConney et al., 2007). Such a covering has been extracted from the fresh cuticle of the cockroach either as hard wax, mobile grease (Beament,1945; Beament,1958) or as a relatively fluid lipid layer hardening through chemical reactions (Gilby,1962; Gilby and Cox,1963). In Periplaneta sp., the polar, reducing-agent material spreads over the entire surface(Lees and Beament, 1948). Epicuticular grease has been considered to play a fundamental role in limiting water loss (e.g. Ramsay, 1935; Neville, 1975) and serving as a behavioral cue for insects (Dubis et al., 1987; Espelie et al.,1991). Its thickness previously has been suggested to be at a molecular scale (Beament,1955), varying from less than 1 nm to several microns(Hendricks and Hadley, 1983),or ranging from 0.1 μm to 1.0 μm(Locke, 1964). In the hunting spider Cupiennius salei, the thickness of the surface viscous layer has been quantitatively estimated to be 20–40 nm(McConney et al., 2007). Atomic force microscopy of elytra of the Colorado potato beetle L. decemlineata revealed a grease layer with a thickness of ∼8 nm(Gorb et al., 2008; Voigt et al., in press). Our measurements of grease thickness on leg cuticle in flies and mirid bugs confirm the previously suggested nanoscale dimensions. In C. vicina,a thin layer ranging from 19.8 to 29.7 nm was found. Remarkably, the grease thickness in P. roridulae is much more prominent, ranging from 516 nm to 713 nm. This layer covers the surface of the mirid bugs, embedding also their microtrichia.

Assuming that the epicuticular grease appears lipid like, we suggest that it functions as a layer causing cohesion failure when the plant adhesive contacts the surface of the mirid bug(Fig. 11). This means that plant adhesive secretions can adhere to the layer of grease, but, under pulling force, the grease layer breaks apart, preventing adherence to the underlying solid cuticle. On the surface of the fly, the very thin, greasy coverage presumably consists of fragmentary patches where the sticky plant secretion can get into contact with the solid epicuticle by filling the gaps of the discontinuous greasy layer (Fig. 11A). The 30-fold thicker epicuticular layer of the surface of the mirid bug prevents contact formation between the plant adhesive and the bug cuticle (Fig. 11B). The microtrichia, submersed in the grease layer, might be interpreted as a microstructure maintaining the thickness of the layer. The sticky droplets cannot penetrate the epicuticular grease of the mirid bug. As a result, these mutualistic mirid bugs cannot be trapped by the plant.

An adhesive secretion enables the R. gorgonias plant to capture effectively a large number of prey insects. The secretion exhibits adhesive properties, presumably differing depending on the type of trichome. Further mechanical characterization of the adhesive secretion combined with the analysis of the chemical composition of secretion from different trichomes will aid in further understanding the trapping strategy of protocarnivorous R. gorgonias.

The mutualistic mirid bug P. roridulae bears a thick,anti-adhesive greasy layer, which is a `smart' surface adaptation for living on the strongly adhesive surfaces of the plant. The anti-adhesion mechanism of the mirid bug is based on the cohesion failure caused by grease. The non-continuous, patch-like pattern of grease found in potential dipteran prey does not provide sufficient protection against the plant adhesive. Additionally, the presence of specific substances in the bug grease targeted to prevent adhesion of Roridula secretion cannot be entirely excluded. In future studies, this mechanism should be compared with that of the cuticle structures of various Heteroptera. Possibly, similar adaptations might be found in other insect–plant associations. Finally, the cohesion failure anti-adhesive mechanism might be potentially interesting for technical developments of novel biologically inspired surfaces.

K. Keller (Augsburg, Germany) kindly provided plants, bugs and valuable information. Dr E. Gorb, Dr M. Varenberg and Dr E. De Souza (Max-Planck Institute for Metals Research, Stuttgart, Germany) as well as Mag. C. Schaber(Dept of Neurobiology and Cognition Research, University of Vienna, Austria)are acknowledged for motivating discussions. Vicky Kastner (Max-Planck Institute of Developmental Biology, Tübingen, Germany) helped with linguistic corrections of the manuscript. This study was partly supported by the Federal Ministry of Education and Research, Germany (project InspiRat 01RI0633D).