Mechanical Properties and Morphological Correlates of Fragile Skin in Gekkonid Lizards

Autotomy, or the ability to ‘shed’ or ‘drop’ portions of the body, is a widely employed antipredator strategy among vertebrates. Caudal autotomy is the most common manifestation of this phenomenon among tetrapods. It occurs in certain salamanders (Wake & Dresner, 1967), rodents (Mohr, 1941; Dubost & Gasc, 1987) and a great many lepidosaurian reptiles (Arnold, 1984, 1988; Bellairs & Bryant, 1985). In those forms in which the tail is autotomized intravertebrally (except amphisbaenians, see Gans, 1978), loss is followed by regeneration. Other forms of autotomy are necessarily rare as animals cannot dispense with body parts that are absolutely essential to survival (Goss, 1969; Reichman, 1984).

Autotomy, in the literal sense, pertains to self-induced loss. However, the concept as currently employed has been extended (e.g. Dubost & Gasc, 1987) to include losses induced by an external agency but facilitated by the structural properties of the organism in question. Within this broader context, the ability of certain lizards to exfoliate large portions of their integument in response to physical contact by a putative predator may be construed as autotomy. We prefer, however, to restrict the term autotomy to its more literal meaning with its implications of neurological control. The expression ‘regional integumentary loss’ is used instead for the phenomenon described herein.

Regional integumentary loss is found only in certain species of the scincid lizard genera Ctenotus and Lerista (Greer, 1986), and a numer of taxa of gekkonid lizards in the subfamily Gekkoninae. Such an antipredator mechanism has been reported in all or most members of the genera Aristelliger (Grant, 1941; Greene, 1988; A. M. Bauer, personal observations), Geckolepis (Schmidt, 1911; Mertens, 1955, 1960; Bellairs, 1969; Anderson, 1972; Meier, 1980; Schubert & Christophers, 1985), Gehyra (Bustard, 1970; Anderson, 1972; McKeown, 1978; A. M. Bauer, personal observation), Perochirus (J. V. Vindum, personal communication), Teratoscincus (Minton, 1966; Anderson, 1972), and insular populations of Thecadactylus (T. J. Papenfuss, personal communication). It is also typical of some members of the genus Phelsuma (McKeown & Miller, 1985; Greene, 1988) and of Pachydactylus namaquensis (Greene, 1988). Such regional integumentary loss also occurs in the large Seychelles gecko Ailuronyx seychellensis (Honegger, 1966; Evans & Evans, 1980; McKeown & Miller, 1985), and it is in this taxon, along with Geckolepis, another Indian Ocean island endemic, that this ability is most clearly manifested.

The anatomy of the integument of Geckolepis typica has been investigated by Schubert & Christophers (1985) who found a preformed zone of weakness lying in the fatty subcutis. On the basis of transmission electron microscopy, they identified a network of myofibroblasts which they believed actively initiated integumentary autotomy. Their work attempted to provide a precise mechanistic explanation for the fragility of the dermis of Geckolepis. While convincing at one level, however, their concentration on ultrastructural details is unsatisfying from the perspective of whole-organism biology as it fails to address the properties of the integument as an organ system with inherent mechanical properties.

In this paper we examine the skin of geckos in general, and Ailuronyx seychellensis in particular, in an attempt to characterize the properties of fragile skin and delineate their morphological correlates.

In this regard we distinguish fragile skin (which may be easily torn and the outer layers released from the body as a result of specific morphological modifications) from weak skin (which is more uniform in its structural properties and only fails under relatively low induced loading when removed from contact with underlying tissues). Little is known about the biomechanical parameters of tetrapod skin except in relation to a limited number of laboratory mammals (e.g. Fung, 1981; Purslow, 1983). This general paucity of information makes it difficult to place the details of particular observed phenomena into a more general mechanical and morphological context. To attempt to overcome this we have examined the pertinent properties of a variety of geckos that do not have fragile skin. This has enabled us to outline the basic mechanical and morphological parameters of gecko skin.

Specimens of Ailuronyx seychellensis were collected on Aride Island, Republic of Seychelles, in August 1987, killed with T-61 solution (Hoechst Pharmaceuticals, Montreal), fixed in 10% neutral buffered formalin, and stored in 70% ethanol. Gekko gecko were obtained from a commercial dealer and prepared as above. Skin strips were also obtained from similarly fixed and preserved specimens of 17 other gekkonid taxa (see Table 2 for listing). Skin samples for mechanical testing and light and electron microscopy were excised in strips of approximately 20 mm ×70mm and subsequently trimmed for the required purposes. Strips of fresh Gekko gecko skin were also taken from freshly killed specimens, and either tested immediately or after formalin fixation.

Strips for mechanical testing were held in metal vice-type clamps with hard rubber faces which prevented slippage, and stretched at a rate of 20 mm min⁻¹ on a Monsanto T10 tensometer. This gave strain rates in the range 0·02–0·04 s⁻¹. Fixed samples were rehydrated in distilled water before use and kept moist during the tests. For each sample the width and length between clamps were measured with vernier calipers before the test, while the thickness was determined afterwards from frozen transverse sections using a light microscope with a digital micrometer eyepiece. Signals for force and cross-head movement from the tensometer were recorded as force vs displacement on an x,y plotter. These data were normalized to stress (σ= force per cross-sectional area), measured in pascals (1Pa = INm⁻²), and strain (ϵ = the increment in length divided by the initial length).

Loading was applied to skin strips taken either parallel with or at 90° to the body long axis. The skin of Ailuronyx seychellensis was tested intact and also after separating the outer from the inner layer of the dermis (see Results). Most commonly, the sample was stretched to the break point and from the data the failure strain (ϵ_f) and tensile strength (i.e. the maximum stress achieved prior to failure, σ_m) were determined. The modulus of elasticity (E), a measure of the material stiffness, was calculated from the slope of the linear portion of the stress-strain curve. The total area under the stress-strain curve is the energy required to break a unit volume of material (W), expressed in Jm⁻³, and is a measure of toughness (Gordon, 1978). This was calculated by cutting out photocopies of the stress-strain curves and comparing their weights with the weight of a calibrated area of paper.

For the specimens of G. gecko and A. seychellensis the work of fracture (F) was also determined by using a ‘trouser tear’ technique (Purslow, 1983; Biedka et al. 1987). A short cut was made along the longitudinal axis of a skin specimen to initiate the fracture. The specimen was torn by pulling the ends of the two ‘legs’ formed by the cut with the tensometer so that a crack was propagated in a controlled fashion at a relatively constant force (see Fig. 7). The work done was calculated as the mean plateau force times the tear length (L). This was divided by the area of the two new surfaces formed (i.e. 2Lt, where t is the sample thickness) to yield the work of fracture in J m⁻². Owing to limited availability of Ailuronyx skin, trouser tear tests were conducted only longitudinally as this orientation accords with the direction of greatest strength predicted from collagen fibre orientation and thickness, and from tensile tests. All mechanical tests were recorded on a VHS video system and still photographs were taken from stopframe video images with a 35 mm format camera.

Material for light microscopy was embedded in Paraplast and sectioned on a rotary microtome at a thickness of 8–10 μm. Stains used were haematoxylin and eosin, Masson’s trichrome and Mallory’s azan trichrome (Humason, 1979).

Specimens for scanning electron microscopy (SEM) were critical-point dried, mounted so as to expose the cut surface of the skin, sputter coated with gold-palladium alloy, and examined using a Hitachi S-450 SEM. Skin samples fractured as a result of mechanical testing (see above) were similarly prepared for SEM examination.

Collagen fibre angles in the skins were measured by viewing wet-mounted samples under a Wild M5 polarizing microscope fitted with a graduated rotating stage.

As reported by McKeown & Miller (1985), wild-captured specimens of Ailuronyx seychellensis readily part with extensive pieces of skin. Even gentle handling of these animals frequently resulted in extensive cutaneous trauma, as Ailuronyx tends to twist its body when restrained, inducing torsional stress on the integument.

The integumentary fragility of Ailuronyx was verified in the course of procurement of specimens for this project. Initial attempts to capture specimens by hand were unsuccessful as skin loss, although expected, proved distracting enough to collectors to allow the escape of the wounded animal. Ultimately, undamaged specimens were obtained with a butterfly net.

The common invocation of regional integumentary loss as a defence in Ailuronyx seychellensis is attested to by the prevalence in museum collections of specimens with fresh or healed wounds. Twenty-four of 30 specimens examined [British Museum (Natural History), California Academy of Sciences, Zoolo-gisches Forschungsinstitut und Museum Alexander Koenig] exhibited such wounds, all of which were confined to the dorsal and lateral surfaces of the trunk and head.

Mechanical tests revealed major differences between Ailuronyx seychellensis and Gekko gecko in the strength, stiffness and toughness of their skins (Fig. 1). In both species the load curves are J-shaped, consisting of an initial ‘toe’ region followed by a linear portion, as is typical of soft biological materials. The peak in stress (σ_m) occurs at the limit of elastic deformation, where the skin structure begins to fail (ϵ_f). Fig. 1A indicates that the dorsal skin of A. seychellensis is relatively weak, with the longitudinal elastic modulus (E) and tensile strength (σ_m) being about an order of magnitude lower than in G. gecko (Fig. IB). Differences are also seen in the way the skin from these species breaks. Tensile failure in A. seychellensis skin begins as a tear at a strain of about 0-28, which propagates further with continued extension. This is shown by the gradual decrease in stress to zero beyond e_f in Fig. 1A. In contrast, the skin of G. gecko breaks catastrophically, as indicated by the sharp drop in stress at ϵ_f, and at a failure strain which is twice as high as in A. seychellensis (Fig. IB). The breaking energy (W), determined from these specimens, is only 0·25 MJ m⁻³ for A. seychellensis compared with 2·6MJm⁻³ for G. gecko. This shows that the dorsal skin of the former species is only one-tenth as tough as that of the latter. Mean values of the tensile mechanical properties of the skins of these two species are given in Table 1.

Fig. 2 compares the mechanical properties of skin samples taken longitudinally and transversely from the dorsal and ventral surfaces of Ailuronyx seychellensis. These data show that the failure strains are virtually the same in all cases, but that tensile strength and elastic moduli are only half as great in the longitudinally oriented dorsal skin samples as in the other samples. That the skin is preferentially weakest dorsally in the direction of the body long axis is consistent with the observation that the occurrence of natural skin wounds in A. seychellensis is most prevalent on the dorsum (see above). In contrast, similar tests on the skin of Gekko gecko indicate that there are no major differences in mechanical properties between dorsal and ventral locations or transverse and longitudinal orientations.

The difficulty in obtaining the exotic species for this study necessitated the use of preserved material. To determine what effect formalin fixation might have on the mechanical properties of lizard skin, fresh and preserved skin samples from an individual specimen of the commercially available G. gecko were compared. Examples of stress-strain plots for dorsal longitudinal skin samples shown in Fig. 3 reveal no major differences in mechanical behaviour attributable to fixation. This is surprising since the effect of formalin fixation on mammalian skin is significant. Viidik (1980) found that rat skin σ_m and E were increased by about 50% following formalin fixation. It was not possible to assess the preservation effects on the skin of Ailuronyx seychellensis and the other species tested (see Table 2) but, based on the results from G. gecko, we have operated under the assumption that the mechanical data obtained are fairly representative of the properties of fresh skin, and that comparisons within a group of preserved specimens should still yield qualitatively, if not quantitatively, valid conclusions.

Two distinct dermal layers are present in Ailuronyx seychellensis dorsal skin, a thick outer layer and a much thinner underlying layer (see below). These are easily separable upon dissection and were tested individually. It was found that these two layers were mechanically quite different. The thin inner layer, which is not ruptured during regional integumentary loss, is much stronger, stiffer and tougher than the thick outer layer. Fig. 4A shows three longitudinal force-extension curves: one for a test on the outer skin layer obtained from the dorsum of A. seychellensis, one for the inner layer of the same skin strip, and one for an equivalent sized strip of whole skin from the same animal. Up to an extension of about 4 mm, at which point the outer layer broke, the force-extension curves for both layers are almost identical. This indicates that the total force to stretch whole skin up to 30 %, where it began to fail, was shared equally by these two layers, even though the inner layer was only l/15th as thick as the outer layer (see Table 1). When the forces are normalized (Fig. 4B) it can be seen that the inner layer sustains much higher stresses at all strains than does the outer layer. The tensile strength, yield strain, breaking energy and elastic modulus are all much greater for the inner layer than for the outer one (see Table 1 for summaries). In fact, the inner layer of A. seychellensis resembles the whole skin of G. gecko in its mechanical integrity, with no apparent modifications for regional weakness. While the outer layer broke at a strain of 0-3, the inner layer broke only after being strained to about 0·7 (Fig. 4B). Failure of the whole skin sample reflected the combined properties of both inner and outer layers acting in parallel. The peak stress coincided with fracture of the outer layer and the transfer of the entire stress to the remaining inner layer, which subsequently broke under further extension. This can be seen in the visual records of extension of a longitudinal strip of whole A. seychellensis dorsal skin (Fig. 5).

The mechanical differences between the inner and outer layers of Ailuronyx seychellensis dorsal skin are illustrated qualitatively in Fig. 5C–G. While the outer layer tears at low strain, the inner layer is highly extensible. The decrease in width seen in the inner layer as it is stretched (Fig. 5G) indicates the reorientation of its constituent collagen fibres. These fibres are 15–20μm in diameter and occur as criss-cross arrays which make angles of about 55° with the body long axis when the skin is unstrained (Fig. 6). From this angle it can be predicted that a strain of 0·74 would occur before the collagen fibres were reoriented directly along the axis of tension. Fig. 5G shows a strip of the inner layer that was strained up to 0·9 just before breaking. The average value of e_f for the inner layer was 0-85 (Table 1). The outer layer has thick collagen fibres (40–60,μm in diameter) oriented at about 90° to the body long axis, and much thinner fibres at 0° (Fig. 6). This fibre arrangement should reinforce the outer layer transversely and provide lower strength longitudinally, as is indeed the case (Fig. 2). Fig. 5C,D,E shows one example of an outer layer strip that fractured at a tensile strain of 0·20. The average value of ϵ_f for the outer layer was 0·29 (Table 1). Fig. 5H-J shows a strip of G. gecko skin which was extended and broke at a strain of 0·6. The lateral ‘necking’ of the sample at high strains suggests, as with the inner layer of A. seychellensis skin, that the fibre component has reoriented towards the axis of stress. Collagen fibres in G. gecko skin were in crossed arrays with an average fibre angle of 50° to the body long axis. From this angle it may be predicted that a strain of 0·56 would be possible before the fibres were loaded directly along their axis. The average value of ϵ_f for G. gecko skin is 0·57 (Table 1). The mechanical properties of these tissues are highly dependent on the fibre morphology and are consistent with the behaviour of fibre-reinforced composite materials.

From tear tests on skin samples, the work of fracture (г), an indication of the fracture toughness of a material, was determined. Typical force-extension records are shown in Fig. 7. In each case the initial region of the curve represents elastic deformation of the two arms of the test piece. When the force is high enough to propagate the tear, it then proceeds in a stable fashion with no additional force required. The plateau region is not smooth but oscillates about a mean value in a ‘stick-slip’ behaviour typical of fibrous biological materials (Purslow, 1983). In this type of controlled fracture, the strain energy input within the plateau region is approximately equal to the work done to form the new surfaces. Values of T for the skin of Ailuronyx seychellensis and Gekko gecko are provided in Table 1. Not surprisingly, these results indicate that the outer layer of A. seychellensis skin could be torn longitudinally with very little energy input. The cracks so propagated are not straight but follow the fibre pattern. We would not expect any major differences if the skin was torn transversely, although, if anything, it may be slightly less tough. However, the inner layer of the skin of this species had a high work of fracture, indicating that it is a very tough material. The skin of G. gecko is also much tougher than the outer layer of A. seychellensis, although not as tough as the inner layer of this species.

Finally, we compared the mechanical properties of the skin of 17 other species of gekkonid lizards, some of which have been described as having fragile skin or exhibiting regional integumentary loss. The results are summarized in Table 2. Based on the mechanical properties of the skin, we have arbitrarily divided these species into two groups. Group A represents species having skins of relatively low strength and toughness (σ_m<10MPa and W<2MJm⁻³). Group B contains species having skins of relatively high strength and toughness (σ_m> 10 MPa and W > 2MJ m⁻³). Ailuronyx seychellensis is included in group A, and Gekko gecko in group B. Only two species, Cosymbotus platyurus and Tarentola mauritanien, exhibit the toughness characteristic of one group and the strength characteristic of the other. In general, skin thickness and failure strains do not appear to be correlated with strength and toughness. The species previously described as exhibiting fragile integuments (see Introduction) all fit into group A, thus establishing a correlation between this characteristic and the mechanical properties of the skin. Interestingly, it may also be noted that the tensile failure behaviour of skin samples from most of the species suspected of having fragile skins (marked by asterisks in Table 2) resembles that of A. seychellensis (see Fig. 1A), while failure in skin from all other species resembles the catastrophic type of rupture seen in G. gecko (see Fig. IB).

The integument of Gekko gecko is typical in structure of that of gekkonid lizards in general. Dorsal scalation is heterogeneous, consisting of juxtaposing granules and tubercles of varying size. Weakly keratinized hinge regions between the scales provide flexibility to the skin as a unit (see Maderson, 1965). The granules themselves are generally flattened and polygonal in outline, and form rosettes around the caudally directed, regularly arranged, subconical tubercles. Certain of the more anterior mid-dorsal tubercles are underlain by osteoderms (Bauer & Russell, 1989). Ventral scales are somewhat more imbricate and less heterogeneous in both shape and surface relief (see Maderson et al. 1978a).

The epidermis of the dorsum is noticeably thicker on the exposed scale surfaces than it is in the hinge regions, and is underlain by pigment cells in the outer region of the dermis. On average, the epidermis accounts for approximately 4 % of total integumentary thickness. The underlying dermis may be divided into a thicker (approx. 90% of total skin thickness) outer region and a thinner (approx. 6% of total skin thickness), but distinct, inner region (Fig. 8A), with little loose connective tissue intervening between the two. The outer region is a composite of the thin stratum laxum and the majority of the collagenous matrix of the stratum compactum. The inner region consists of the deepest layers of the stratum compactum and possesses prominent pigment cells associated with its outer periphery (Fig. 8A). A thin layer of loose connective tissue separates the base of the dermis from the body wall musculature.

The ventral skin (Fig. 8B) differs from that of the dorsum in terms of the heterogeneity of the scales outlined above. Further, the stratum laxum is somewhat more extensive and the inner layer of the stratum compactum is more closely adpressed to the outer layer.

The dorsal scalation of original skin consists of pronounced conical tubercles arranged in an alternating pattern of large and smaller units (Figs 9, 10A,B). Narrow hinge regions appear as grooves within the high relief of the conical scales (Fig. 9). The larger tubercles are characterized by their generally circular bases, acute apices and complex surface microarchitecture (Fig. 10A,B), while the smaller ones have a more rounded apex and lack the well-defined pattern of radiating ridges.

The ventral scales of Ailuronyx seychellensis are, on the whole, more similar in appearance to those of Gekko gecko.

Wounded areas of the dorsum appear smooth and shiny, and the underlying muscle is clearly visible through windows of the deeper layer of the pigmented stratum compactum (Fig. 9, arrowheads). Fresh wound edges appear to coincide with the grooved hinge regions mentioned above (Figs 9, 10B) and are somewhat irregular. Healed wounds are generally similar in appearance to original skin but may be distinguished by their smaller, more irregular tubercles, as well as on histological grounds (see below).

The epidermis of the dorsum is noticeably thinned in the hinge regions between tubercles (Fig. 11A), more so than in Gekko gecko. The stratum compactum consists of a thick outer zone and a much thinner, distinct inner zone (Figs 10C, 11A). The outer layer is much more heterogeneous than that of G. gecko. The outer region of the core of the conical tubercles and the majority of the content of the smaller, dome-like tubercles consists of expanded parenchymatous cells of the stratum laxum, interrupted by thin and irregularly disposed collagen fibres (Figs 10C, 11A). The stratum compactum of the outer zone lacks the continuity of that of G. gecko and there is less crimping of the collagen fibres. Vertical fibres are not prominent in this zone. The collagen fibres of this zone are continuous or only partially interrupted beneath the smaller, dome-like tubercles and are continuous beneath the hinge regions (where a distinct stratum laxum is absent), but are significantly interrupted within the larger, conical tubercles. In the latter, collagen fibres are poorly represented and are divided around a central, collagen-free core of expanded parenchymatous cells that are encompassed entirely within the stratum compactum. Pigment cells are evident but scattered in all components of the outer layer.

The thin inner zone (6 % of integumentary thickness measured relative to the mean hinge region thickness) consists of collagen fibres and contains conspicuous pigment cells. The gap between the inner and outer dermal layers is more pronounced than in G. gecko and is occupied by loose connective tissue. A similar gap separates the inner layer from the body wall.

The ventral skin (Fig. 11B) is generally similar to that of the dorsum (see Fig. 11 A). The core of the subimbricate scales is filled with expanded parenchymatous cells, but these are all derived from the stratum laxum. Although fibres of the stratum compactum intrude into the stratum laxum, there is no parenchymatous interruption within the stratum compactum itself. The collagen fibres of the inner layer of the stratum compactum are straighter than those of the outer layer and, unlike the inner layer of the dorsum, appear to contain no pigment cells.

At the site of the wound only the inner layer of the stratum compactum remains intact (Fig. 11C). The wound is bounded by tubercles and rupture of the integument appears to have taken place at the junction of these scales with the thinner intervening hinge region, or perhaps at the regions underlying the larger tubercles. As a result of this, the collagen fibres of the outer zone of the stratum compactum have exposed free ends (Figs 10B,D, 11C).

In healed skin the inner zone of the stratum compactum retains its original appearance. The outer layer, however, regenerates without or almost without the conspicuous interruptions that characterize the subtubercular region of the undamaged integument (Fig. 11D). The tubercles themselves are lower in relief than those of intact skin and only weakly endowed with enlarged parenchymatous cells. In many respects these regenerated scales resemble more closely those of original ventral skin than they do original dorsal skin (Fig. 11A,B,D).

Vertebrate skin is typically involved in many functional roles, among them support, and protection of the soft tissues, transport and storage of gases, metabolites and nutrients, locomotion, behavioural display, and sensory perception (Kresja, 1979). The observation that Ailuronyx and certain other geckos so readily part with the majority of the integument over large areas of the body begs the question of how and why such loss occurs without obvious detriment to the animal. Clearly, the independent aquisition of fragile skin by several gekkonid lineages suggests that under certain ecological conditions the ability to part with skin may be more valuable than keeping it. Indeed, observations of animals both in the wild and in captivity demonstrate that Ailuronyx does employ the facility of skin loss as an antipredator strategy against potential aggressors.

Understanding the evolution of regional integumentary loss requires an appreciation of the cutaneous mechanical properties that facilitate loss and their underlying morphological bases. The morphological specializations that appear to be correlated with skin loss in Ailuronyx seychellensis are limited to a number of features that are absent from, or only incipiently developed in, Gekko gecko.

One major feature which is present in G. gecko, and indeed most gekkonids (Schmidt, 1911, 1912, 1913; Lange, 1931), is the segregation of the stratum compactum of the dermis into physically distinct inner and outer layers. In most geckos these are relatively tightly bound together, while in Ailuronyx (see above) and Geckolepis (Schmidt, 1911, plate 25, fig. 18; Schubert & Christophers, 1985, figs 2, 3) these two layers are much less intimately interconnected. This loose connection provides a zone for slippage between the two layers of the dermis and, as a result, the outermost layers of the skin are mechanically independent of the deep layer of the dermis. In Gekko gecko the deep layer of the stratum compactum is only loosely bound to the underlying body wall muscles (Maderson et al. 1978a), while in Ailuronyx this association is much firmer. The more definitive subdivision of the integument into two units in Ailuronyx is reflected in the differing mechanical properties of each considered separately (see above) and in the emergent properties (O’Grady, 1984) of the whole skin (Fig. 4).

The inherent fragility of the outer integumentary complex is a result of interruption of the collagenous network of the stratum compactum by intervening cones of parenchymatous cells (Fig. 11A). The reduced number of collagenous fibres per unit area in these zones of weakness results in stress concentrations in the region of the large tubercles and failure occurs under relatively low tensile loading. Such zones of weakness are not uncommon in reptilian integuments, but are generally more restricted in their distribution. In particular, they are found in caudal autotomy planes (Schmidt, 1912; Woodland, 1920; Ali, 1950; Werner, 1967; Sheppard & Bellairs, 1972; Bellairs & Bryant, 1985) and in association with the exit points for defensive secretion in the tail-squirting geckos of the genus Diplodactylus (Rosenberg & Russell, 1980).

The relative strengths of the inner and outer integumentary units are a function not only of their gross morphological architecture, however, but also of the collagen fibre orientation and continuity. The thicker outer collagen fibres (40–60 μm in diameter) are oriented at about 90° to the body long axis, with thinner fibres being oriented at about 0°. If the thicker fibres were continuous they would reinforce the integument transversely. They are, however, interrupted within each of the large tubercles, thus providing only localized areas of significant bracing. Imposed loads result in failure at the zones of weakness and give rise to an integument that is both weak and fragile.

The criss-cross arrays exhibited by the inner layer (mean angle of 55° to the body long axis, Fig. 6) indicate that, as tested, considerable reorientation of fibres took place before they became aligned along the direction of applied loading, although this was not measured. The observed fibre angle of 55° in Ailuronyx approximates but does not match exactly the idealized fibre angle of 60° predicted by Alexander (1987), probably because the body cross-section is not circular. Fibre angles of 45°, 55° and 60° occur frequently in animal integumentary collagens and the significance of these angles has been interpreted through a number of functional models. These, however, are primarily locomotory models (Clark & Cowey, 1958; Wainwright et al. 1976; Alexander, 1987) and are most applicable to aquatic animals that utilize the skin as a primary or secondary locomotory aid. The significance of specific fibre angles to function in Ailuronyx or other terrestrial vertebrates is probably more directly associated with skin extensibility or other non-locomotor functions.

In Ailuronyx the general properties of the integument associated with mechanical protection have been forsaken in favour of escape from potential predators. This has implications for the survival of the individual; while mechanical strength of the outer component of the skin is permanently reduced, upon loss the osmoregulatory properties of the integument (Maderson et al. 1978b;,Landmann, 1985; Lillywhite & Maderson, 1982) are also temporarily compromised, at least until regeneration is complete. This impairment of function is due to the loss of the keratinized epidermis (Maderson et al. 1978b) and not to the reduction of the outer layer of the dermis (Licht & Bennett, 1972).

The mechanical properties of Ailuronyx skin represent one end of a continuum which also includes many species characterized by more ‘typical’ cutaneous attributes (Table 2). The 19 taxa examined in this study may be roughly segregated into two clusters, although absolute distinction between them is not possible. These clusters are demarcated on the basis of the mechanical properties of their skin, with group A representing forms with weak skin and group B forms with tougher skin (Table 2). Group A includes several species for which regional integumentary loss is well documented - Ailuronyx seychellensis, Geckolepis typica and Teratoscincus scincus. We believe that the mechanism of skin fracture is similar in all three taxa. The latter two species, however, differ from Ailuronyx in possessing large, flattened, imbricating scales. Anderson (1972) suggested that such scutellation might be easily cast off while retaining some underlying cutaneous tissue as a protective barrier. Schubert & Christophers (1985), however, claimed that all components of the integument were lost in the affected areas in Geckolepis. This interpretation is not consistent with the impheations of the morphological (Schmidt, 1911) or behavioural (Meier, 1980) investigations of earlier workers. Further, the loss of the entire integument over a broad area with no negative consequences is seemingly incompatible with the normal operation of the immune system, given the reasonably long period (2 months or more, McKeown & Miller, 1985) required for regeneration of the skin from the wound edges. Indeed, our interpretation of the light micrographs of Schubert & Christophers (1985) is that separation of the outer layers of skin occurs at the border of the lower layer of the stratum compactum with the overlying loose connective tissue, or perhaps within the lower zone itself.

Also included in group A are species such as Aristelliger praesignis and Phelsuma madagascariensis for which skin loss has been reported, and Pachy dactylus bibronii, a closely related congener of a known fragile-skinned form. The remainder of the species in this group do not exhibit this antipredator defence (A. M. Bauer, personal observation), but in these forms the weakness of the skin is largely irrelevant in a functional context if the integument is a unitary structure, itself bound tightly to the underlying body wall.

For the most part, the geckos in group B, as would be predicted by the high tensile strength of their skin, do not exhibit fragile skin in life. Among the taxa included, only Thecadactylus rapicauda has been reported to display regional integumentary loss. However, this property is confined to certain insular forms (R. I. Crombie & T. J. Papenfuss, personal communication), and our specimen is from a mainland population (Rio Santiago, Amazonas, Peru). Two of the 19 taxa show strength and toughness values that are marginally incongruent, but these do not greatly disrupt the overall emergent pattern.

Relatively little is known about the biomechanical properties of reptilian skin, but what is known does not differ in any major way from the basic values obtained from the whole skin of Gekko or from the inner layer of Ailuronyx skin. In a comparative overview of the mechanics of snake skin, Jayne (1988) found that the typical stress-strain curve was J-shaped, the elastic modulus generally varied from 15 to 250 MPa and the tensile strength ranged from 5 to 80 MPa. Comparative data for gekkonid species that do not have fragile skins accord well with this general picture (Table 2).

The known details of the structure of reptile skin differ significantly in some respects from those of mammals. For example, elastin fibres play an important role in mammalian skin mechanics, but their distribution, relationships and functions are only poorly understood in reptiles (Lange, 1931; Pockrandt, 1937). In mammals, collagen forms a two-dimensional network of coiled fibres, often associated with elastin cores (Serafini-Fracassini et al. 1975), with elastin being the primary agent of recoil at low loads, while collagen provides high stiffness at larger extensions (Oxlund et al. 1988). The composite structure yields J-shaped stress-strain curves and tensile properties similar to those of reptiles. In the rat, the skin has a toughness of approximately 15 kJ m⁻², and an elastic modulus of about 10 MPa (Purslow, 1983). The tensile strength of cat skin is approximately 12 MPa (Wainwright et al. 1976).

Comparative study of gekkonid skin reveals two key innovations (the mechanical independence of the outer region of the integument, and inclusion in that area of preformed zones of weakness) that are co-opted into exaptive systems (Gould & Vrba, 1982) and permit regional integumentary loss in Ailuronyx and perhaps other forms with fragile integuments. Thus, in Ailuronyx only the mechanical behaviour of outer components of the skin (epidermis, stratum laxum and the outer layer of the stratum compactum) is of significance in the context of naturally occurring predator-prey interactions. Likewise, measurements of whole skin mechanics of other species, perhaps including weak-skinned taxa from group A (Table 2) that do not lose the skin in nature, may be irrelevant in the context of the skin-loss defence strategy. Weak skin bound firmly to a strong and stiff backing, such as the body wall musculature, will not facilitate regional integumentary loss. Our data on gecko skin mechanics serve to highlight Wainwright’s warning (1988) that mechanical ‘properties must be seen in the context of natural history of the species being studied’ if ‘the properties are to relate structure and function to larger issues of ecology, behavior, and evolution’.

We thank Lindsay Chong Seng for permission to collect in the Seychelles and Ian Bullock (Royal Society for Nature Conservation) for allowing us to work on Aride Island. Field assistance was provided by Blake Edgar. Randolf Lauff and Jacques Rantallion assisted in the laboratory studies. Specimens used for comparative testing were generously provided by Robert Drewes (California Academy of Sciences) and Harry Greene (Museum of Vertebrate Zoology, Berkeley). The manuscript was improved by the suggestions of two anonymous reviewers. Jim Alston kindly prepared the scanning electron micrographs. The manuscript was typed by Susan Stauffer. This research was supported by funding from the Herpetology Department, California Academy of Sciences, a University of Calgary Postdoctoral Fellowship to AMB and NSERC operating grants to APR and RES.