Slippery surfaces of pitcher plants: Nepenthes wax crystals minimize insect attachment via microscopic surface roughness

Plants of the carnivorous genus Nepenthes (Nepenthaceae) efficiently attract, trap, retain and finally digest animal prey in highly modified leaves (Juniper et al., 1989). Within these pitcher-shaped leaves, different functional zones can be distinguished by their morphological and physicochemical surface characteristics, all of which contribute to the carnivorous syndrome (Knoll, 1914; Juniper et al., 1989; Gaume et al., 2002; Gaume et al., 2004; Bohn and Federle, 2004). The opening of these pitchers is lined by the peristome, which plays an important role in prey attraction by visual and olfactory cues and by nectar secreted at its inner margin (Moran, 1996; Owen and Lennon, 1999; Lloyd, 1942). It is also important for the trapping of insects, which slide on its fully wettable anisotropic surface (Bohn and Federle, 2004; Bauer et al., 2008). The surface of the conductive zone immediately below the peristome has also been recognized as essential for trapping and retaining insect prey, but in this case due to its special downward-directed lunate cells and its coverage with a thick layer of epicuticular wax crystals (McFarlane, 1893; Knoll, 1914). In spite of these very different mechanisms, both surfaces are extremely slippery for insects and cause the prey to fall into the lower part of the pitcher. In this part, specialized glands secrete a digestive solution in which the insects get trapped (Gaume and Forterre, 2007) and which finally absorbs the insect-derived nutrients (An et al., 2002a; An et al., 2002b; Schulze et al., 1997; Owen and Lennon, 1999; Owen et al., 1999).

Surface attachment of insects is principally enabled by the interlocking of hard claws and/or adhesion of flexible attachment devices to a wide variety of substrates (Beutel and Gorb, 2001). At the cellular level of both Nepenthes pitcher surfaces, trapping efficiency is enhanced by the anisotropic arrangement of smooth overlapping epidermal cells (Bohn and Federle, 2004; Gaume et al., 2004). This arrangement only allows the interlocking of insect tarsal claws during inward locomotion of the prey. Escape from the pitcher by using adhesive pads is also hampered by the physicochemical properties of the cuticular surfaces of the different pitcher zones. The peristome was shown to be completely wettable by nectar or rain water, leading to coverage with homogeneous liquid films that impede close contact between the plant and tarsal pad surfaces (Bohn and Federle, 2004). For the conductive zone, several mechanisms for its slipperiness have been discussed. As early as 1914, Knoll noted that the epicuticular wax crystals on the surface are easily detachable under the weight load of an insect foot, leading to contamination and hence inactivation of the adhesive device (Knoll, 1914). This interpretation was supported by the observation that crystal platelets are fastened to lower epicuticular layers by thin fragile ‘stalks’ (Juniper and Burras, 1962; Gorb et al., 2005). Experiments on flies (Calliphora vomitoria) walking on the waxy surface of Nepenthes ventrata supported this theory: after 10 min of walking little contamination of the adhesive setae was found but after 2 h the tarsus was completely covered (Gaume et al., 2004). However, investigations that used mechanical sampling strategies to prepare the crystals selectively (Riedel et al., 2003; Riedel et al., 2007) showed that the outermost crystals visible in the SEM do not consist of independent platelets but form a continuous 3D network of interconnected structures that keeps its structural integrity even after mechanical removal from the plant surface (Riedel et al., 2003; Riedel et al., 2007). These observations suggest that the extreme slipperiness of the waxy pitcher walls for insects cannot simply be explained by the fracture and detachment of wax crystals. In this study we investigated the mechanism of slipperiness by combining scanning electron microscopy (SEM), focused ion beam (FIB) treatment, atomic force microscopy (AFM) and attachment tests on live insects. We addressed the following questions. (1) How is the wax structure arranged beneath the outermost crystal parts? (2) What force values are necessary to break off the outermost crystal parts? (3) Do insect adhesive pads get contaminated by detached wax crystals? (4) Is the breaking of wax crystals necessary to capture prey?

Specimens of Nepenthes alata Blanco (Nepenthaceae) were taken from a continuous greenhouse culture in the Botanical Garden of the University of Würzburg. Pitchers open for approximately 7 days were freshly harvested for preparations. Rectangular leaf sections of 10–15 mm side length were cut from the conductive zone and either air dried for FIB imaging or freshly used without further preparation for AFM measurements (see below).

Female stick insects (Carausius morosus Sinety, Phasmatidae) were taken from a laboratory colony. Juvenile specimens (larval state L1–L3; length 20–25 mm; mass 19–88 mg) were used for centrifugation experiments. For slip-off experiments juveniles of larval state L3–L5 (length 40–65 mm, mass 144–539 mg) were used. Additionally, living adult black garden ants (Lasius niger L., Formicidae) were captured near the research department in Aachen (Germany).

Ants are the native prey of Nepenthes, thus these insects were of greatest interest; however, experiments were also performed with stick insects, because these have similar adhesive pads which are, in contrast to those of ants, very well characterized with respect to mechanical properties. This allows theoretical modelling (see below).

For friction force measurements, comparison to the glaucous Nepenthes surface, and calculation of a model for the interaction of insect tarsi with structured surfaces we used artificial surfaces with different surface parameters. As a flat and relatively non-structured surface we used polycarbonate. Surfaces with different profile parameters were represented by different types of polishing paper. The grain sizes of the so-called microgrits abrasive are defined by the sedimentation time of the particles. The particles are made of silicon carbide. Further information on particle size and size distribution is given by the FEPA-standards 42-2:2006, 43-2: 2006 and ISO 6344 and 8486 (Federation of European Producers of Abrasives, France).

The polishing paper (Mirka, Karlsruhe, Germany) and its mean grain sizes (in brakes) used in our experiments were: P320 (46.2±1.5 μm), P600 (25.8±1 μm), P2000 (10.3±0.8 μm) and P4000 (3.0±0.5 μm).

For SEM imaging, air-dried pitcher material from the conductive zone was sputter-coated with gold for 8 min with a sputter current of 10 mA, resulting in a gold layer of less than 2 nm. Observations were made using a Stereoscan S604 SEM (Cambridge Scientific Instruments Ltd, Cambridge, UK). Images were digitally recorded using an attached i-scan digitizer (ISS Group Services Ltd, Manchester, UK) with an image acquisition time of 50 s.

For determining 3D structure, FIB cutting was performed using a Strata FIB 205 (FEI, Hillsborough, OR, USA) single beam workstation. A focused beam of Ga ions from a liquid metal source was projected onto the gold-coated specimen. The gold coating was approximately 100 nm thick in order to prevent the surface from immediate damaging by the Ga ions. An acceleration voltage of 30 kV was used and the beam current was adjusted to 30 pA for imaging of the surface, which was observed without further treatment. Typically for cutting, a beam current of 300 pA was used and the cutting edge was finally polished using a beam current of 50 pA. However, the beam current for cutting was varied for control purposes from 10 to 1000 pA in order to exclude artefacts due to local heating.

AFM measurements were performed using a commercially available AFM (Explorer, Veeco Instruments, Woodbury, NY, USA) equipped with a 100 μm scanner. For imaging the scale topography the AFM was equipped with high reflection coated triangular silicon nitride cantilevers (Type MLCT, Veeco Instruments) with a nominal spring constant of 0.5 N m⁻¹, a four-sided pyramidal tip with a tip angle of 35 deg. (centreline-to-face) and a tip radius of 20 nm. Scanning was performed in contact mode to obtain information about the stability of the glaucous surface on N. alata. The scanning force was adjusted from 20 nN to 250 nN and images were obtained at various scanning velocities ranging from 1 μm s⁻¹ to 50 μm s⁻¹. Scans were performed at room temperature in an aqueous solution containing 5 g NaCl, 0.15 g KCl, 0.25 g CaCl, 0.15 g NaHCO₃ in 1 l pure water, pH 7.4. The freshly cut plant material (approximately 15 mm×15 mm) was glued using hot glue onto the bottom of a Petri dish. The glue was only applied to the corners of the square in order to avoid structural changes of the wax. The Petri dish was immediately filled with the aqueous solution and degassed in an evacuated desiccator. Immersion was found to be vital in order to prevent artefacts due to shrinkage of the plant material. Furthermore, fresh material had to be used as dried plant material sometimes exhibited reduced stability of the wax crystals (see below).

Three kinds of experiments were carried out to find out whether loose wax crystals from the conductive zone of N. alata would adhere to tarsi of insects and whether the intact wax layer of that region stays stable under the mechanical load of a slipping insect.

In the first of the experiments, wax from the conductive zone was separated from two pitchers and applied to a glass plate using a stiff paint brush. Carausius morosus and L. niger individuals were put onto this glass plate to run on the loose pieces of wax for a few seconds.

In the second of the experiments, six pitchers open for about 12 h per day for up to 16 days were taken from a pool of N. alata grown under greenhouse conditions. The digestive zone and the lid were cut and the pitchers clamped in the horizontal direction approximately 10 cm above a beaker, filled with a gentle flow of CO₂. Specimens of L. niger and C. morosus were put onto the waxy pitcher surface, directly underneath the peristome. The pitchers were then rotated to the vertical direction to induce slip-off into the beaker. On each pitcher surface this was carried out with three L. niger and one C. morosus individual. Afterwards, the insects from both the first and second experiments were anaesthetized with CO₂ and decapitated. The legs were cut off, air dried in a Petri dish on silica gel at a temperature of 28°C for 24 h and gold coated for 5 min (10 mA sputter current, gold layer less than 2 nm). Observations were made using the Stereoscan S604 SEM. Possible contamination of the adhesive organs, especially the arolii and tarsi, as well as the tarsal hairs by wax crystals or other wax residuals was documented. Pieces of the pitcher surfaces which the ants had slipped off were removed, air dried and gold sputtered. The waxy surfaces were observed using the SEM, to obtain information about the crystals after the insects had slipped off.

In the third of the slip-off experiments the ability of ants to cling to a smooth glass surface was tested: first, directly after walking on loose wax crystals; and second, after walking on an intact Nepenthes wax surface. Two watch glasses were prepared: the first with loose wax brushed from dry pitcher surfaces and the second with pieces of an intact wax layer on the bottom. After the ants were put onto the watch glass it was covered with a cleaned glass tube. The movement of the ants was filmed using a standard digital photo camera to get information on the ants' ability to cling to the smooth glass after walking on wax crystals and the intact wax layer.

Surface profile parameters of the polishing paper P4000 and the glaucous Nepenthes surface were obtained using a μsurf customized (NanoFocus, Oberhausen, Germany) Confocal-Multi-Pinhole microscope. Employing an ocular lens with ×100 magnification, the parameters R_a (roughness average), R_q (root mean square roughness), R_Pc (roughness peak count) and R_Sm (mean spacing of profile irregularities) were determined with regard to DIN EN ISO standards.

To measure attachment forces of C. morosus and L. niger, we used a centrifuge technique in order to determine the maximal frictional force, similar to the method described by Federle and colleagues (Federle et al., 2000; Federle et al., 2004). Insects were placed on a turntable coated with either a flat polycarbonate plate or polishing paper (see above). A strobe light synchronized to the revolutions of the centrifuge through a photoelectric barrier was used for illumination so that a standing image of the insect on the rotating surface could be seen. The centrifuge was filmed from above (distance 0.9 m) with a standard 25 Hz CCD video camera (Panasonic F15).

Revolutions per minute of the centrifuge were recorded with the stroboscope and the radius at which the insect detached was measured by marks placed at 10 mm radial intervals on the surface under investigation. Carausius morosus and L. niger individuals were placed on the turntable and the rotation was slowly accelerated until detachment occurred. The centrifugal force at the time of detachment was taken as the friction force reached by the insect under investigation. As shown previously (Federle et al., 2004), other forces are small in comparison to the centrifugal force. Additionally, in the present study only relative changes and not the exact absolute values of the forces are important, yielding the described method sufficient and adequate. The adhesion forces were measured on polishing papers with grain sizes P320, P600, P2000 and P4000 and a flat polycarbonate surface for both species (C. morosus and L. niger; N=10).

Scanning electron microscopic investigation of N. alata pitchers (Fig. 1) revealed exactly the morphology previously described in the literature (Riedel et al., 2003; Riedel et al., 2007; Gaume et al., 2004). Because of a dense wax bloom immediately below the rim of the pitcher, the inner wall surface of the conductive zone appears glaucous (Fig. 1A). In SEM this wax bloom was shown to cover all epidermal cells, including the lunate cells, as a continuous layer (Fig. 1B). Platelet-shaped wax crystals with entire margins protruded from the underlying structures, standing almost upright and therefore exposing their narrow sides to the lumen of the pitcher (Fig. 1C). Angles between the outermost crystal parts showed no preference, creating a random pattern of small air spaces in between (Riedel et al., 2003). In order to clarify the 3D structure of the epicuticular wax layer and thereby the organization of the wax crystals, FIB cutting was performed. This technique was found to be necessary because freeze fracture of the wax crystal zone yielded samples with damaged breaking edges. For FIB cutting, gold-coated samples of the conductive zone were used with an apparently intact wax layer in the proximity of the cutting edge (Fig. 2A,C). A rectangular area of about 20 μm×5 μm was cut from the sample (Fig. 2B,D) and the FIB-cut edge was finally ion-polished. Scanning electron micrographs of the FIB-cut edge at higher magnifications (Fig. 3) clearly showed the 3D structure of the epicuticular wax. Instead of a single layer of isomorphic platelet-shaped crystals, a sponge-like or trabecular structure became evident. This sponge-like layer showed a fairly constant thickness of about 3 μm and the mesh size of the sponge decreased towards the compact lower epicuticular wax layer. Apparently, the platelet-shaped crystals described so far are only the outermost walls of this porous structure. To rule out artefacts due to local heating by the FIB, the experiments were repeated with different ion currents. As the structures observed were approximately identical in all samples (not shown), we were able to conclude that no local heating took place which would have led to structural changes dependent on the ion current.

It appears unlikely that these trabecular cross-linked structures will be detached under the weight of insects leading to contamination and hence inactivation of the adhesive devices as suggested by Knoll and others (Knoll, 1914; Juniper and Burras, 1962; Gorb et al., 2005). However, in order to understand the mechanical properties and to find out whether detachment of crystal parts of the sponge-like wax layer alone can be responsible for the anti-adhesive properties of the conductive zone, AFM was performed. This was done using freshly prepared 1 cm² samples from the conductive zone, immersed in a slightly hypotonic salt solution. Using fresh material was necessary because the wax layer on dry material was found to be mechanically unstable in comparison with fresh material (data not shown). Immersion was found to be necessary because otherwise substantial shrinkage of the tissue during the AFM investigation occurred, disturbing the imaging process. Degassing of the immersed plant material was also essential as the wax layer is highly hydrophobic yielding a thin air film on the surface which would disturb the AFM measurements. This film, however, can be almost entirely removed by short (5 min) degassing in an evacuated desiccator.

Typical AFM images, obtained in the contact mode, are shown in Fig. 4. Because of the AFM tip geometry the thin and steep walls of the sponge-like wax layer could not be resolved in greater detail. The tip yields a wall perpendicular to the scanning direction appearing under an angle identical to the tip angle α, which is 35 deg. centreline-to-face in our case. This is depicted in principle in Fig. 5. Taking this fact into account, the AFM images very much resemble the structures observed under the SEM, showing that an intact wax layer was indeed scanned. The images obtained were not altered when scanning force, direction and/or speed were changed. Scanning forces (i.e. forces perpendicular to the surface) up to about F_N=250 nN could not destroy the wax structures (Fig. 4). Higher forces were not tested because of technical limitations of the apparatus and cantilevers used.

Considering the geometry shown in Fig. 5, we obtained the total force (F_tot) acting between the tip and a steep wall (wax crystal) to be F_tot=F_N/sin(α). This results in a bending force F_H=F_N/tan(α) of approximately 400 nN acting locally on the wax crystal. Thus, a single crystal, i.e. a single wall of the sponge-like structure, can at least withstand forces which correspond to a mass of 40 μg attached at the outermost edge of a single crystal. As this represents the maximum bending momentum we can assume this arrangement to be the worst case. A single adhesive organ of a typical prey insect has a contact area of more than 10,000 μm². According to the SEM and FIB images, the density of individual walls (crystals) is about 1 per μm². Thus, the force acting locally on a single crystal caused e.g. by a large ant of 5 mg is of the order of 5 nN, corresponding to 0.5 μg of attached mass per crystal if only one small arolium of a heavy ant is in contact (worst case), which is two orders of magnitude lower than the force a single crystal can withstand in the experiment. Thus, even if only 1% of the area of the arolium is in contact, the crystals would still withstand the force acting upon them. For control experiments we repeated the AFM measurements on detached wax layers (removed by drops of frozen water) (Riedel et al., 2003) having an identical geometry but being less stable as only the outermost part of the wax is removed. As clearly demonstrated in Fig. 6, the AFM tip can break and/or remove these wax crystals. Thus, if considerable breaking of wax crystals took place on fresh, intact pitchers, the AFM technique in combination with the accompanying SEM investigation would detect this. However, we never found any evidence of breaking due to AFM scanning on intact pitchers as we always got stable AFM images and non-contaminated tips after scanning, and we never found areas of broken wax crystals.

In addition to the atomic force microscopy, slip-off experiments on intact pitchers as well as detached wax layers were performed. The results of the first of these experiments on detached wax layers clearly show that loose pieces of wax from the pitcher plants adhere to the arolia, hairs and claws of the tarsi of both insect species, C. morosus and L. niger. Fig. 7D,E clearly shows huge amounts of wax, agglutinated to a certain degree and covering nearly the whole area of the last tarsal segments of C. morosus. The hairs on the tarsi are also covered with wax and are stuck together (Fig. 7F). Comparable results were found in L. niger (Fig. 8D–F); the hairs of all tarsal segments are stuck together by agglutinated waxes, and the claws and the arolia are also covered with wax crystals.

The second of these experiments was performed to test whether breakage of the waxy surface affects the arolii of the insects and consequently minimizes adhesion. The SEM images of C. morosus (Fig. 7A–C) and L. niger (Fig. 8A–C) after the experiments clearly show just a little wax residue on parts of the tarsi like the hairs and parts of the arolium. Even in higher magnification images only a small number of wax crystals and other indicators of wax residue from the waxy surface of the Nepenthes plant could be found on the arolium of C. morosus (Fig. 7B) and L. niger (Fig. 8B) or on the hairs and claws of the insects (Fig. 7C, Fig. 8C).

The third of the experiments was performed to test whether walking on loose wax crystals or on the intact wax layer of the conductive zone had any influence on the ants' ability to cling to smooth surfaces afterwards. The results are given as videos in the supplementary section (see supplementary material Movies 1–3). Those ants that walked on artificially detached wax clearly showed a reduced adhesion while walking on a vertical, smooth glass surface afterwards; thus, the friction force was below 1 g (where g is the gravitational acceleration). Contrary to those findings, ants walking on an intact pitcher surface showed no reduction of adhesion.

The above-mentioned findings indicate that detachment of individual platelets cannot be the main cause of the inability of insects to climb on the surface of the neck region of N. alata. In order to find out whether the structure itself or the mechanics are responsible for the described effect, we performed measurements of the adhesive forces of juvenile stick insects (C. morosus) on different artificial surfaces. In order to simulate the structure of the N. alata wax crystals, polishing paper of defined grain size was used. The silicon carbide crystals on polishing paper are platelet shaped and very thin and, thus, resemble the principal structural feature of the N. alata pitcher wax under investigation (Fig. 9). Because of their mechanical robustness, breakage of the silicon carbide crystals under the weight of a stick insect can be ruled out. For distinguishing between adhesion due to the adhesive pad (the arolium) and the claws of the insects, experiments were done with untreated animals and with insects where the claws had been removed. The frictional forces were measured using the centrifugation assay described above. The frictional force was defined as the centrifugal force at which detachment occurred. The results, normalized to the weight of the insects in units of the gravitational acceleration g are summarized in Fig. 10. Neat polycarbonate surfaces and polishing paper of different grain sizes (P4000–P320) were compared for untreated insects and insects with claws removed. It is evident that for untreated insects adhesion on the flat polycarbonate plates and on the rough polishing papers (P2000–P320) was fairly constant. However, the polishing paper P4000 was an exception as the insects could hardly adhere. They could not withstand a force that corresponds to an acceleration of 1 g. Thus they were not able to walk or even stay on a vertical wall. The structure of P4000 shows a striking similarity of the grain dimensions with N. alata wax crystals (Fig. 9B). This finding was strongly supported by comparing the surface profile parameters of the two surfaces (Table 1). Both resemble thin and steep walls spaced by values (R_Sm) from 1.33 to 1.83 μm in the case of the polish paper and from 1.15 to 1.53 μm in the case of the Nepenthes surface. The R_a values of the polishing paper range from 0.241 to 0.448 μm and those of the Nepenthes surface from 0.17 to 0.369 μm. Insects with claws removed showed a different behaviour. While the adhesive force on the flat polycarbonate was similar to that observed for untreated insects, hardly any adhesion could be found on rough surfaces. Similar experiments performed with L. niger produced very similar results: while L. niger could withstand forces severalfold their own weight on the flat polycarbonate surface (acceleration of 26.6±14.5 g) and on the polishing papers P320 (38.4±14.3 g), P600 (23.9±7.3 g) and P2000 (11.9±3.5 g), they could hardly adhere on the P4000 surface (0.73±0.42 g).

In the present study we could elucidate the 3D structure and investigate the mechanical properties of the epicuticular wax layer of the inner pitcher wall of N. alata. In the plant specimens in our hands we found the wax structures to be sponge-like (porous) with an outer pore size of about 1.3–1.5 μm, and to be mechanically very stable. There are some indications that the chemical composition and therefore probably the mechanical properties can change during maturation of the pitcher and according to environmental conditions (M.R. and M.R., unpublished data). However, until now we have failed to clarify under which conditions the wax layers become stable. Nevertheless, in the greenhouse-grown plant material we never found pitchers with crystals that broke under the weight of stick insects or ants or which could be broken by AFM scans. We cannot rule out the possibility that breaking of wax crystals may occur but what we clearly demonstrate is that the inner walls of the pitchers we investigated have stable crystals that are capable of catching prey. In this study we focused on insects having smooth pad-like adhesive organs. Ants were chosen as they are typical prey insects for Nepenthes. Additionally, stick insects were investigated because they are excellent model organisms which can be easily handled (much easier than ants) and for which mechanical parameters of the arolium are available (Scholz et al., 2008).

Stick insects were found to adhere well on artificial surfaces with either lower roughness like polycarbonate or higher roughness like polishing paper P2000 or lower. However, polishing paper with grain size P4000, which closely resembles the structure of N. alata wax crystals, allowed hardly any adhesion by the stick insects. When the claws of the stick insects were removed prior to the adhesion force measurements, the animals could adhere well on smoother surfaces like polycarbonate but failed to adhere to surfaces with greater roughness. This shows that the insect arolium is responsible for adhesion on flat surfaces whereas the claws perform well on rough surfaces. We propose that the surface of N. alata can exhibit a mechanically stable surface with a defined roughness which is too high for the arolium and too low for the claws to be effective.

The force which can be transmitted by the claws is dependent on the contact angle between the claw and the wall tip (Dai et al., 2002). This angle depends on the tip radius r of the claw tip and the pore size λ of the surface. In our SEM studies the tip radius of C. morosus was found to be about 7 μm and that of L. niger about 3 μm. If the pore size is larger than the diameter of the tip, the force, i.e. the adhesive energy, stays more or less constant. If 2r<λ, the contact angle follows as arctan(λ/r). The ratio of the adhesive energy to the repulsive energy for the claw is given by the ratio of the contact angle and the friction angle, which is dependent on the materials in contact and which is equal to the arctan of the friction coefficient μ. Assuming a high friction angle of 30 deg., according to Dai and colleagues (Dai et al., 2002), we obtain the ratio of adhesive and repulsive energies for the claw as depicted in Fig. 11B. Obviously values above 1 require a pore size λ to be larger than about 2 μm. This behaviour is in line with recent observations (Dai et al., 2002).

Of course, the described theoretical model is only a first, rough estimation. However, it is based on experimentally obtained parameters and some conservative assumptions clearly showing that a small range of pore sizes (λ; the spacing between profile peaks in the case of polishing paper) exists, where neither the arolium nor the claws can adhere sufficiently to the surface. This range is between 0.5 and 2 μm, a result matching the pore sizes found for N. alata and the artificial surface P4000 having a pore size of approximately 1–1.5 μm.

The findings of this study show that the solid structure of the wax layer of the inner pitcher wall of N. alata together with the high hydrophobicity, which prevents wet adhesion, can explain the anti-adhesive properties of the conductive zone with respect to insect adhesive organs. Adhesion on the wax surface mediated by insect arolia can be excluded. The area of real contact between the arolium and the thin upwards-directed walls of the wax layer on the conductive pitcher surface is insufficient, because the thickness of the walls (clearly below 100 nm) and the given pore size of about 1 μm results in less than 10% effective contact area.

Thus, detachment of single wax crystals and contamination of the adhesive organs of insects is not necessary to explain the phenomena observed. Furthermore, the often-suggested detachment, which appears to play a major role for other plants [e.g. Catopsis berteroniana or Brocchinia reducta (Gaume et al., 2004); Macaranga spp. (I.S., M.B., L.D., T.E., A.W., F.H., J.M., S.H., M.R., M.R. and W.B., unpublished observations)] appears unlikely for the Nepenthes species investigated because of the mechanical properties of the wax layer measured experimentally. Indirect hints on the stability of Nepenthes wax crystals have already been given by Gaume and coworkers (Gaume et al., 2004). In their work they found contamination of the hairy adhesive organs of C. vomitoria after running for 10 min on intact plant surfaces of C. berteroniana and B. reducta, but only a small residue after running on N. alata. Massive contamination due to walking on Nepenthes ventrata surface could only be documented after 180 min, indicating the N. ventrata wax layer to be much more stabile. In a recent work, Gorb and coworkers (Gorb et al., 2005) found contamination of the hairy adhesive organs of Adalia bipunctata after 1 min of walking on N. alata surface which appears to be in contrast to the recent findings of Gaume and colleagues (Gaume et al., 2004) and the present study. This, once again, was found for hairy adhesive organs. For these organs breaking off material from the pitcher surface might be effective. However, it should be emphasized that the structure of the wax surface of Nepenthes is fully sufficient for catching prey with smooth adhesive organs like ants, which are the typical prey insects (Kato et al., 1993; Adam, 1997; Moran et al., 1999) under natural conditions. Furthermore, contamination of adhesive organs after minutes of walking on plant surfaces does not explain the immediate effect on insects trying to walk on the conductive zone of the pitcher, which results in instantaneous falling of the animal.

Lasius niger and C. morosus were chosen as experimental animals. The former were used because ants are the prey most frequently captured by Nepenthes plants, but they are not easy to handle in experiments. With C. morosus we used an insect which might not be a typical prey, but which has several advantages: first, the structure and the mechanical properties (elasticity module) of the adhesive organ are well documented; second, the insect has a much greater weight than the ants, which gives higher impact onto the wax layer of Nepenthes; third, stick insects are easy to handle.

Under these close-to-nature conditions the insects showed only a few wax crystals or residues on their tarsi, but nevertheless failed to adhere to the wax surface and slipped off. Comparing images of tarsi of C. morosus with those of L. niger after slipping showed only negligible contamination on the stick insect tarsi. This might be explained by the higher load per area of the animals. Based on measurements of the contact area of adult specimens of C. morosus (Drechsel and Federle, 2006) and ants of the genus Oecophylla smaragdina (Endlein and Federle, 2008), the load of a juvenile stick insect is about 20–30 times higher (about 9–12 mN mm⁻² in comparison to about 0.42 mN mm⁻² load for L. niger). Additionally, in the third part of our slip-off experiments we clearly showed that only artificially detached wax from the conductive zone of the pitcher leads to a reduced adhesion of insects on a smooth surface. Walking on the intact wax layer had no such effect.

These findings strongly support the theory of a stable wax-foam surface that minimizes adhesion by its surface parameters even if bigger insects try to adhere.

In conclusion we cannot rule out the possibility that to a certain degree breaking of extensively protruding single wax crystals may occur and consequently be found as residues attached to the tarsi. Presumably this would increase the anti-adhesive effect of the surface of N. alata. Furthermore, breaking of crystals for catching prey may occur in different species of pitcher plants or even individuals of N. alata under certain circumstances or environmental conditions. However, the results presented clearly indicate that the structural properties of the mechanically very stabile epicuticular wax alone are sufficient to prevent insect adhesion to the conductive zone of N. alata.

The authors thank Dr Walter Federle (Cambridge, UK) for helpful discussion and comments on the manuscript. The work was financially supported by grants from the Deutsche Forschungsgemeinschaft (SFB 567 ‘Mechanisms of interspecific interactions of organisms’ and SPP 1207 ‘Strömungsbeeinflussung in der Natur und Technik’). We are also indebted to the staff of the Botanical Garden of the University of Würzburg for cultivating the plants used. Deposited in PMC for immediate release.