The Ultrastructure Of The Epicuticular Interference Reflectors Of Tiger Beetles (Cicindela)

The widespread occurrence of structural coloration in insect cuticle received vigorous attention in the early part of this century (Biederman, 1914; Onslow, 1920; Mason, 1927). These researchers ascribed the physical colours of the integument to the constructive interference of light reflected from thin layers within the exoskeleton. However, the exact anatorny of the reflector remained obscure, and proposals of morphologies varied widely. The advent of the electron microscope finally allowed ultrastructural studies of cuticular reflectors. The bulk of this work focused on inter-ference reflectance by specialized structures such as insect cornea and lepidopteran wing scales (Bernard & Miller, 1968; Ghiradella et al. 1972). However, the most complete study of interference reflectance by whole cuticle was Neville & Caveney’s (1969) work on scarab beetles. While that work elegantly described the exocuticular reflectors within that family, the morphologies of cuticular reflectors in other metallic-coloured insects have remained undescribed.

Coincidental with the early work on insect coloration, several authors described the elytral structure of several groups of beetles which exhibit interference reflectance (Buprestidae; Hass, 1916; Cicindelidae: Stegemann, 1930; Carabidae: Sprung, 1931). Stegemann (1930) described a superficial layer, the ‘Sekretschicht’, in tiger beetles (Cicindelidae) which dissolves in dilute KOH at 60 °C and pours from secret-ory glands after ecdysis. Mandl (1931) later ascribed the iridescence of Cicindela to this ‘Sekretschicht’. Shelford (1917) independently suggested that a surface film reflects interference colours in Cicindela.

Many of the elytral structures described by early histological and morphological studies have not been re-examined rigorously under current concepts of cuticle struc-ture. However, homologies have been proposed for these structures based upon more recent research involving other cuticles. Wigglesworth (1972) suggested that the ‘Sekretschicht’ may be analogous to the cement layer of other insects. Richards (1951) preferred the term ‘tectocuticle’ for this outermost layer presumed to be secreted by the dermal glands. A recent electron microscope study by Mossakowski (1980) iden-tified an interference reflector in the outermost layers of cuticle in Cicindela campestris. He identified the reflective cuticle as exocuticle.

The following study attempts to locate the multilayer reflective system of cicindelid cuticle, to describe its ultrastructure, and to determine to what extent this cuticle conforms to the current models of insect morphology.

Reflection from elytral surfaces was measured using a Cary 17D spectro-photorneter. Elytra for reflectance measurements were taken from four species of tiger beetles, Cicindela formosa Say, C. splendida Hentz, C. scutellaris rugata Dejean and C. repanda Dejean. Adults of all four species were collected at Bastrop State Park, Bastrop County, Texas.

Samples of adult cuticle were cut or torn from the elytra and thoraces, and subjected to a variety of chemical treatments. They were then prepared for either scanning or transmission electron microscopy.

In order to differentiate the epicuticle from the exocuticle, fragments of cuticle were immersed in 8 % KOH at 60 °C for periods of 4, 9, 12, 24, 48 h and 6 days. One set of samples was treated with concentrated KOH for 15 min at 160 °C. A second set was treated similarly, but followed with a 17-h immersion in 8 % KOH. Treatment with hot concentrated alkali is the initial step in the chitosan method of Campbell (1929) for detection of chitinous cuticle. All samples were rinsed with distilled water, dried, and any colour or gross textural change observed with a dissecting microscope. The fragments were cut to smaller size and prepared for either scanning or trans-mission electron microscopy.

In order to remove procuticular layers, elytra were immersed in concentrated H₂SO₄ for 1,12, 24, 48 and 72 h. The elytra were rinsed with distilled water, dried, and any colour changes noted. The elytral fragments were then prepared for scanning electron microscopy (SEM).

Since H₂O₂ will bleach melanin and alter the interference colours if melanin is one component of the interference reflector, the role of melanin in cicindelid coloration was investigated by treating C. formosa and C. scutellaris elytral fragments for 6 days with 10% H₂O₂ at 26 °C. When no colour change was observed, the samples were treated for a seventh day at 60 °C. Samples were dried, observed and prepared for transmission electron microscopy.

Dried strips of treated or untreated cuticle were mounted on edge, sputter-coated with 2·5 nm of gold-palladium, and examined with a Hitachi scanning electron microscope. This orientation provided a view of the torn edge perpendicular to the elytral surface with each cuticle layer exposed in relief.

For transmission electron microscopy, treated and untreated samples of cuticle were cut into extremely fine slivers (0·3 × 1 mm) with a tapering edge. Cuts were positioned between elytral tracheae to avoid trapped air bubbles. The cuticle was fixed in 2·5 % gluteraldehyde and post-fixed in 2% O_sO₄ for 2h, dehydrated in a standard ethanol and propylene oxide series, and infiltrated in a long series of propylene oxide/Epon-Araldite mixes. Each infiltration step required agitation, and the final two immersions in pure Epon-Araldite were evacuated completely.

Thin sections were cut perpendicular to the elytral surface. Some sections from each treatment group were post-stained for 5 min in 1 % uranyl acetate in 50 % ethanol, followed by 5 min in lead citrate. Portions of the dorsal surface of C. scutellaris were also sectioned parallel to the plane of the elytral surface. All sections were examined under a Zeiss 10CA transmission electron miscroscope.

Thin sections of samples from each treatment group were mounted and reserved for interference and polarized light microscopy.

The elytra and thorax of C. formosa are deep metallic red with white maculations-peak reflectance occurring at 655 nm, with a secondary peak near 360nm (Fig. 1A). The Bastrop population of C. formosa exhibits the elytral coloration of the race C. formosa pigmentosignata Horn. While the maculae of Bastrop individuals are rarely reduced, many individuals exhibit a distal deep purple coloration on the elytron, which is characteristic of C. formosa pigmentosignata (Gaumer, 1977). For reflec-tance measurements and electron microscope studies, elytral fragments were selected from elytral areas which were homogeneously red.

The elytra of C. splendida are metallic brick red with a metallic green border. An elytron with the border removed shows maximal reflectance near 635 nm, with a very small secondary peak at 410nm (Fig. 1B).

C. scutellaris rugata elytra reflect maximally at 499 nm in a relatively narrow band (Fig. 1C). C. scutellaris varies greatly in colour among its races, with rugata display-ing an iridescent blue-green coloration. Both the intensity and the purity of this reflected colour exceed those of C. formosa and C. splendida.

The brown C. repanda elytron has a very different reflectance curve (Fig. 1D). Instead of a sharp peak in the visible wavelengths, the reflectance increases steadily towards the near-infrared.

The colour of tiger beetle elytra, like all interference colours, changes with the angle of viewing. As the angle of reflectance departs from the angle of incidence, the reflected wavelengths become shorter. The elytra of Cicindela are alveolate (Fig. 2) with small (10 × 10μm) hexagonal depressions. The alveoli are superimposed upon larger irregularities in the upper portion of the elytron. Light reflected from this surface is a composite of wavelengths reflected from a variety of angles. Not only would the total intensity of reflection be diminished, but the peak reflectance would be less sharply defined. Of the four species, C. scutellaris displays the most regular elytral surface with smooth, shallow alveoli. C. repanda and C. splendida possess the deepest alveoli and most acute ridges.

The general features of elytral structure observed under SEM were consistent in the four species. Above the tracheae, the procuticular layers (region D in Fig. 2) are organized in the pseudo-orthogonal ‘plywood’ fashion that has been described for other Coleoptera (Hepburn, 1972). This region has been defined as endocuticle (Zelazny & Neville, 1972) or as mesocuticle (Hepburn, 1972), and is considered to be untanned or non-sclerotized.

Observed under the SEM, the outermost layer of Cicindela cuticle appears as a dense, solid material which bears the alveolate microsculpture of the integument (region AB in Fig. 2). This layer ranges in thickness from 3·5 μm at the bottorn of a cell, to 7·0 μm at the top of an adjacent ridge. The torn vertical face of this layer appears less fibrous and more granular than the underlying cuticle (region D in Fig. 2). Occasionally, samples will tear obliquely and reveal two distinct portions of the outer layer in relief (Fig. 3). In these torn samples, a regular banding or lamination of 6–8 extremely even bands (150–200 nm thick) is visible in the outer 1-2 μm of the layer (regionA in Fig. 3B). These laminations appear in contrast to the rougher, non-laminated face of the inner portion (region B in Figs 3B and 4). Based upon their ultrastructure, location, development and reaction to solvents, the outer and inner portions of the outer cuticle (layers A and B in Fig. 4) are here designated the epi-cuticle and outer exocuticle, respectively, and will be referred to as such throughout this report.

Dermal gland canals measure 0·8 μm in diameter and their orifices are distributed regularly over the surface of the elytron (Fig. 4, dc). The pattern of lamination is not disrupted around the canal opening. Epicuticular thickness is enhanced on the ridges, compared to the basins of the alveoli, but is not greater at the dermal gland opening than on adjacent ridges.

Transmission electron microscopy (TEM) reveals that the alveolate superficial layer of the scanning electron micrograph is indeed a composite of two complex layers (layersA andB in Fig. 5). Immediately below this composite, lie 2-3 thin exocuticular layers (layer C in Figs. 5, 6A) which exhibit the distinct arciform appearance of helocoidal cuticle as described by Bouligand (1972).

The outer exocuticle (layer B in Fig. 5) appears more electron-dense than the helicoidal or preferred layers below. This layer consists of 25-35 densely packed lamellae ranging in width from 0·055 to 0·1 μm. In some areas, electron-dense granules lie between or within lamellae. In post-stained sections, the horizontal lamellae are resolved into alternate patterns of microfibrils viewed in perpendicular and transverse orientations (Fig. 6B). This pattern closely resembles the description of the non-perfect helicoidal outer exocuticle of Tenebrio given by Filshie (1982).

Pore canals, 50—70nm in diameter, traverse the outer exocuticle (Fig. 6C, pc). Within the exocuticle, they appear as segments of tubes containing material much denser than the surrounding fibrils. The pore canals extend into the epicuticle but appear much narrower in diameter (Fig. 6C, arrow). This pattern is similar to the branching of epicuticular filaments described in other insect cuticles (Filshie, 1982).

The epicuticle (Fig. 6D) is composed of a series of extremely dense bands (D) be-tween 0·03 and 0·1 μm thick, regularly spaced by electron-lucent bands (L) of 0·06-0·125 gm. Each pair of electron-dense and electron-lucent layers corresponds to a single band observed in relief in the scanning electron micrographs. The number of bands varies between four dense layers at the base of a depression to nine layers within an adjacent ridge. Under high magnification, electron-dense layers appear to be composed of vertically aligned granules (Fig. 6D). Post-staining enhances this appearance. The electron-lucent bands exhibit no apparent ultrastructure. The outer-most layer is always electron-lucent and 20-30 nm thicker than the other light bands. The outer 20-30 nm appears as a faint shadow of higher density, possibly representing a cuticulin layer (Fig. 6E, cu). Mossakowski (1980) identified the outer series of electron-dense and electron-lucent bands as exocuticle. He concluded that the banding was a diffraction pattern resulting from the structural organization of exocuticular fibrils.

Although there are occasional distortions, the organization of the epicuticular layers is exceedingly regular. One individual L or D layer may vary in thickness horizontally within a section. This is usually an artifact resulting from the section passing obliquely through the layer as it conforms to the alveolate surface. Each L or D layer may differ slightly in thickness from other L/D layers in that region of the stack. In C. scutellaris, the peripheral dense band is typically fragmented and particulate. Partially formed bands occur regularly at the outer exocuticular border.

When the honeycombed surface is sectioned horizontally, the section passes through the hexagonal ridges and across the intervening alveoli. Initial sections thus possess holes, surrounded by material from adjacent ridges (section x in Fig. 7). Within the ridge itself, the section passes obliquely across the outermost epicuticular layers, resulting in concentric rings around the hole. As the section intersects the ridge axis, it passes through the portion of an epicuticular layer that is oriented parallel to the section; this therefore appears as a continuous patch of either electron-dense or electron-lucent material. Deeper sections include the elevated portions of outer exocuticle within the base of a ridge (sectiony through region B in Fig. 7). Sections taken across the alveolar basin (section z in Fig. 7) result in a central spot of light or dense epicuticular material ringed by obliquely sectioned layers. These islands of epicuticle are surrounded by exocuticle.

The horizontal sections demonstrate that electron-dense and electron-lucent layers have the same individual density and location when observed horizontally or vertically (Fig. 8A). The banding pattern is not, therefore, an artifact of diffraction. The layer that borders the holes in horizontal sections is the outermost layer of the epicuticle, and is electron-lucent in both orientations. Furthermore, horizontal sections within a single epicuticular layer show continuous patches of electron-dense material. This material is very finely particulate when viewed perpendicular to the layer (Fig. 8B).

When thin sections are viewed through an optical microscope, the outer exocuticle and the epicuticle appear dark brown (Fig. 9B). This apparent pigmentation is con-fined to the outer 5 μm of the elytral cuticle. It is assumed that the dark brown appearance is due to the presence of melanin in this region, and that it is responsible for the dark opaque appearance of the elytron in transmitted light. The epicuticle and outer exocuticle display a greater retardation of light than the inner layers of procuticle under interference microscopy.

Sections of elytral cuticle were also cut from the white maculations of C. formosa. The white areas show no trace of either iridescent colour or brown or black pigmenta-tion (Fig. 10B). Under interference microscopy, the sections of white cuticle show a uniformity in refractive index throughout the elytron. The epicuticle shows a refrac-tive index of approximately 1·5 by the Becke line test (Neville, 1980).

Electron micrographs of this non-pigmented cuticle show an epicuticle which lacks the electron-dense bands (layer A in Fig. 10A). In general, this homogeneous epi-cuticle is thinner and lower in contrast than the laminated epicuticle of pigmented areas (layer A in Fig. 9A). SEM confirmed the absence of the epicuticular laminations in the region of the elytral maculations. The outer exocuticle of these sections lacks the electron-dense granules that were evident in the exocuticle of pigmented areas (layers B in Figs 9A and 10A).

Tiger beetle cuticle treated with dilute KOH displayed a gradual loss of structural coloration, as well as pigmentation. Immersion in ethanol, xylene, acetone, sulphuric or nitric acid failed to produce any change in structural colour. Initially, structural colours became longer in wavelength as the solvent temporarily swelled the reflec-tance layers. After 4h, however, the maroon portion of the elytron of C. formosa appeared blue-green. The anterior portion, originally red, became ochreous with slight orange-green reflections. The overall iridescence was reduced to a dull sheen. Scanning electron micrographs of the treated elytra revealed a distorted and heavily eroded surface microsculpture. Under transmission electron microscopy, the regular laminations at the periphery of the cuticle were disrupted and often stripped from the sample (Fig. 11).

With the exception of a few blue-green or violet reflections in the basins of punctures, elytra treated with KOH for 12-15 h were virtually devoid of structural colours. The surface had a flat, brown, darker appearance at the elytral base, which is assumed to be due to the residual presence of melanin in the exocuticle. The deep hexagonal depressions of the surface microsculpture were eroded to a shallow outline of the surface pattern. TEM micrographs revealed that the laminated epicuticle was completely removed from the surface of the outer exocuticle (Fig. 12). A light band, 0·29 μm thick, remained at the border of the exocuticle. The exocuticular lamination also became more obscure and fragmented.

Longer KOH treatments resulted in a progressive loss of the remaining brown colour. By 48 h, only a slight trace of the original maculation pattern was evident, and the elytron was a light straw colour. During this treatment, the outer exocuticle remained -the alveolate pattern was still visible in the surface. However, the thick-ness of this layer was reduced by 40%, suggesting a leaching of exocuticular com-ponents.

The chitosan method of Campbell (1929) and van Wisselingh (1914) has long been used to distinguish chitinous from non-chitinous cuticle, although this technique is not conclusive (Richards, 1951). It requires the conversion of chitin to chitosan by treatment with concentrated KOH at 160 °C for 15 min. It is assumed that the non-chitinous elements of the exoskeleton (epicuticular components) are dissolved under this alkali treatment. Samples of tiger beetle cuticle subjected to the chitosan test showed a severe break-up of the presumptive epicuticle. In C. scutellaris, an eroded epicuticle (areaA in Fig. 13) was evident in some areas where a feebly iridescent sheen still appeared. If these treated samples were immersed in 8 % KOH for 18 h at room temperature, all traces of the presumptive epicuticle were removed, as well as the remaining pigmentation. In C. formosa, the epicuticle and iridescence were complete-ly removed by the chitosan treatment (Fig. 14).

In contrast to the results of the alkali treatments, endocuticle is dissolved rapidly in concentrated sulphuric acid (Sprung, 1931 ; Wigglesworth, 1933; Hackman, 1974), whereas sclerotized exocuticle and epicuticle are said to remain intact. Tiger beetle elytra immersed in sulphuric acid showed a rapid dissolution of inner ‘plywood’ mesocuticle, leaving a thin, flexible sleeve of putative epicuticle and outer exocuticle, while the structural coloration of the elytron remained unaltered (Fig. 15A). SEM examinations of the treated cuticle revealed that the outer laminated layer (layer A) retained its ultrastructure intact (Fig. 15B). The outer exocuticle was also intact and pigmented through the first 48 h of treatment. The residual cuticle still appeared dark brown in transmitted light. Portions of the helicoidal exocuticle below could be seen adhering to the lower side of the outer exocuticle. Frequently, areas were observed in which the epicuticle had separated from the outer exocuticle, and these fragments were found floating on the surface of the acid. Even these isolated fragments main-tained a laminated appearance under SEM, as well as an iridescence.

Treatments of elytral cuticle with H2O2 significantly altered the cuticle coloration. When observed in transmitted light, the brown coloration became more translucent. The outer exocuticle and epicuticle appeared either light brown or clear of pigment in microscopic sections (Fig. 16B). The structural colour of the cuticle surface became duller and shorter in principal wavelength. Similar results were obtained when samples were treated with 20 % H₂O₂ at 26 °C for 24-48 h.

TEM micrographs revealed a 40 % reduction in thickness of dense epicuticular layers in the H₂O₂-treated samples (Fig. 16A). The bands were progressively thinner towards the outer surface. The overall thickness of the epicuticle was not diminished significantly.

Probably because it is extremely thin, the epicuticle is the least understood portion of the insect integument. Its chemistry, structure and the terminology of epicuticular components have been controversial. Only a few, diverse arthropods have been studied intensively, and very little work has focused on the epicuticle of hard, adult exoskeletons (Delachambre, 1970). This is not surprising since hard cuticle is dif-ficult to prepare and section for transmission electron microscopy.

The epicuticle, by definition, is non-chitinous or reacts negatively to tests for chitin (chitosan method). It is the first layer formed during the moult sequence and assumes the outermost position in the strata of the exoskeleton, bearing the surface micro-sculpture. The general model of epicuticle structure recognizes five components. These are the cuticulin layer, the inner epicuticle, the outer epicuticle, the wax layer and the cement layer, listed in order of deposition (Filshie, 1982). The wax and cement layers are secreted upon the epicuticular surface by the epicuticular filaments and dermal glands, respectively. Filshie (1982) notes that the saturated lipids of wax layers do not survive conventional electron microscope preparations, and their sup-posed appearance in electron micrographs is probably an artifact or misidentification of other layers. Identifications of cement layers have been made by location only and remain unsupported by developmental studies.

The reflective layers of cicindelid cuticle are not a form of cement or tectocuticle. The so-called ‘Sekretschicht’-the source of the structural colours-is formed pre-ecdysially, not secreted from dermal glands (Schultz & Rankin, 1985). Furthermore, the very even pattern of layering over a contoured surface would be unlikely if it were formed from secretions of relatively dispersed glands. The layers remain evenly spaced around and even within the mouth of dermal glands (Fig. 4). Extraordinary properties of the two components and the dermal glands themselves would be required to produce such regular secretions over the entire cuticular surface.

The cuticulin layer (or outer epicuticle of Neville, 1975) is the first epicuticlar layer secreted by the epidermis during the moult sequence. It appears as a thin ( 10-20 nm), dense membrane at the outer edge of the cuticle in electron micrographs. In some insects (Locke, 1966; Delachambre, 1970), it initially appears laminated, but con-denses to a uniform appearance. Micrographs of C. formosa show a dense region, 30-40 nm thick, at the edge of the outermost electron-lucent band (cu in Fig. 6E). The inner side of this region appears finely laminated, becoming more uniformly dense towards the surface of the cuticle. This region appears in all the Cicindela species and samples studied, including the pharate cuticle (see Schultz & Rankin, 1985), and is interpreted as the cuticulin layer. In mature cicindelid cuticle only, a very thin (10 nm), electron-lucent membrane appears above the cuticulin. This struc-ture may represent the outer epicuticle that is thought to arise by separation from the cuticulin layer prior to ecdysis (Filshie, 1982).

The bulk of the epicuticle is composed of the inner epicuticle, 0·5-1 μm thick. Its appearance varies among species, and consequently, a wide range of chemical con-stituents have been proposed. These include protein, lipids, lipoprotein and dihydroxy phenols. This layer borders directly on the helicoidal exocuticle beneath it (Neville, 1975). Laminated inner epicuticles have been described, but these lamina-tions have been transitory during the moult sequence (Delachambre, 1970), or have extremely short periods of less than 40 nm (Gupta & Smith, 1969; Gluud, 1968; Zacharuk, 1972).

The multilayered outer region of cicindelid cuticle is best described as inner epicut-icle, due to its location, ultrastructure and solubility in KOH, in combination with a resistance to strong acid (Region A in Figs 2–7). Treatments with solvents demon-strate that this region is chemically distinct from the chitinous layers below. If it were exocuticle, the ultrastructure of the layers would survive the alkali treatment (Hack-man, 1974). Furthermore, it lacks any indication of fibrillar architecture that is evident in the alkali-resistant layers or in the exocuticles of other insects.

The outermost layer of inner epicuticle, presumably secreted first, is always electron-lucent, while the innermost layer may be electron-lucent or electron-dense. More layers are produced at the borders of the epidermal cells than at their centres. These borders are indicated by the hexagonal ridges in the surface microsculpture (Neville, 1975). Areas of elytral cuticle which lack pigmentation (i.e, the macula-tions), also lack electron-dense layers in the epicuticle. Perhaps the cicindelid inner epicuticle is described best as a homogeneous electron-lucent material, impregnated by electron-dense material at discrete intervals in those cuticles that are pigmented.

It appears that the epicuticle is the source of structural colours in Cicindela. Maceration or removal of the epicuticle with KOH results in the deterioration or loss of iridescent coloration. Conversely, removal of the other components of the integu-ment produces no loss of colour. The change of hue under oblique incidence demon-strates that the structural colours result from constructive interference of reflected light.

The repetitive layering of two materials in the epicuticle, whose thicknesses are less than half the wavelength of visible light, is characteristic of a multilayer interference reflector. The two layers are distinct in their electron densities, and are not an artifact of the helicoidal arrangement of cuticle fibrils, as in the reflective exocuticle of optic-ally active scarab beetles (Neville & Caveney, 1969). Multilayer reflectors have been described in a wide diversity of animal tissues, and the basic principles of interference in animal reflectors are reviewed by Land (1972).

A quarter-wave interference reflector has been described by Durrer & Villiger (1972) in the exocuticle of Euchromagigantea, which consists of a series of electron-lucent and electron-dense bands similar to the epicuticle of Cicindela. The dimen-sions of the layers imply an average refractive index of 1·75 for the reflector. The authors propose that the multilayer system contains electron-dense melanin layers (RI = 2·0) separated by chitin (RI = 1·5), similar to the keratin-melanin interference reflectors of bird feathers (Durrer & Villiger, 1970). In their approach, the authors calculated the wavelength of maximum reflectance for each layer separately, in addition to determining the reflectance from the series of bilayers (equation 2). They concluded that a mixture of these reflected components results, producing the total colour.

Mossakowski (1980) examined interference reflectors in the cuticles of the buprestid, Chrysocroa vittata, and the tiger beetle, Cicindela campestris. Mossakowski calculated reflectances for a number of possible refractive indices using the theoretical treatment of Huxley (1968). He proposed that an average refractive index of 1·75 for the bilayers of C. camprestis would cause a total reflectance value far exceeding those determined experimentally. Mossakowski’s measurements of total reflectance predicted refractive indices of 1·5 and 1·6 for the light and dense layers respectively. However, these values and the dimensions of the layers observed under electron microscopy predicted a wavelength of peak reflectance substantially below those recorded spectrophotornetrically. Mossakowski assumed that the reflecting layers had experienced a 10—15 % shrinkage due to specimen preparation or exposure to the electron beam.

The dimensions of the interference layers of cicindelids presented in this paper suggest a relatively high average refractive index near 1·75 for the pairs of layers. Values of 1·5 and 2·0 for the light and dense layers predict peak wavelengths which approximate to the spectrophotornetric results for C. scutellaris and C. formosa (Table 1). Refractive indices of 1·5 and 1·6 would predict substantially shorter wavelengths of peak reflectance. The epicuticle of C. repanda varied greatly in layer thickness and its reflectance exhibited no single peak within the visible wavelengths.

Even assuming an average refractive index of 1·75, calculations using the average thicknesses of epicuticular layers predict values of  max well below the reflectance peak of C. splendida. Unlike the other specimens, the outermost electron-lucent layer of C. splendida is 40–50 nm thicker than all the underlying electron-lucent layers.

This layer makes a significant contribution to the large standard deviation in the average thickness of these layers. Variation in the thickness of the inner layers is less than that of C. repanda but greater than the variation in C. scutellaris. A λ max of 638 nm is predicted if calculations are based upon the superficial electron-lucent layer (d_L — 130 n_L, = 1·5) and the underlying dense layer (d_L = 62n_L = 2·0). This value is close to the experimental reflectance peak of 635 nm. Since the surface of C. splendida is so deeply sculptured with the alveolate pattern, the superficial epicuticular layers may contribute more to the peak reflectance than the underlying layers. Peak reflectances from deeper layers may occur at angles which depart from the angle of reflection that is sampled by the spectrophotorneter.

No evidence of shrinkage appeared in the samples of cicindelid cuticle. The struc-tural colours of specimens were observed unchanged within the embedding medium, indicating that shrinkage did not result from specimen preparation. SEM measure-ments of epicuticular bilayers corresponded well with bilayer measurements made from transmission electron micrographs. If bilayers measured from SEM micrographs were assumed to shrink 10% under the electron beam, they would appear 10-20 nm thinner than the average thickness of bilayers measured from TEM micrographs. If shrinkage had occurred under the electron beam, TEM measure-ments should fall below the thicknesses measured under SEM.

The peak reflectance is also dependent upon the constancy of layer thickness throughout the epicuticle. In Cicindela, the thicknesses of the bilayers vary both horizontally and vertically within the epicuticle (Table 1). The theoretical treatment of non-ideal, multilayer reflectors assumes that these dimensions remain constant. This non-uniformity in layer thickness and angle, with respect to incident light, clearly places the cicindelid epicuticle as an extremely complex example of a biological interference reflector. The epicuticle represents a non-ideal, multilayer system moulded into a shallow honeycomb form, and is not adequately described by theoreti-cal treatments for regularly spaced, planar systems.

Correlations between electron microscope measurements and reflectance data should not be over-emphasized. The amount of cuticle sectioned is not comparable to the area of elytral material sampled for reflectance. The variability within the interference reflector has already been described. In addition, there is always un-certainty as to whether the section is precisely perpendicular to the interference layers.

With rare exceptions, the brown or black colours of insect cuticles are attributed to sclerotization and/or melanization of the cuticle (Richards, 1967). Eumelanins predominate as the brown or black pigments of insect integument (Hackman, 1974), and the pigmentation of Cicindela has been ascribed to melanin since the work of Shelford (1917). Unfortunately, it is exceedingly difficult to identify small amounts of melanin by staining or chemical analysis. Since the individual layers of dense material in the epicuticle are not resolvable under light microscopy, their optical properties cannot be examined separately. The treatment with KOH removes the epicuticle from the pigmented exocuticle, and produces a solution that is stained brown with dissolved epicuticle. But it is also possible that melanin may be leached from the exocuticle by the treatment (Richards, 1967).

Richards (1967) found that treatment with H₂O₂ will bleach melanin particles without removing them from the cuticle. Tiger beetle elytra were treated with H₂O₂ to determine the role of melanin in the elytral coloration. If melanin is involved simply as an absorbent background to the interference reflector, the iridescent colour should be reduced in intensity, but not altered in hue when the pigment is bleached (Mason, 1927). In treatments of C. formosa, a reduction in intensity and predominant wavelength resulted, as well as a thinning of pigmented colour in the cuticle. The change in reflected colour may be due to the bleaching of melanin within the structure of the reflecting epicuticle itself.

The structural dimensions of the two epicuticular layers, in conjunction with the experimental reflectance results, suggest that the epicuticular reflector is composed of an electron-lucent layer with a refractive index of 1·5 and a dense layer of 2·0. A refrac-tive index of 1·5 is acceptable for insect cuticular proteins (Neville, 1975). This study proposes that the unusually high refractive index of 2·0 is provided by layers of melanin or melanoproteins. Precedents for such a system can be found in iridescent bird feather colouration (Greenwalt, Brandt & Friel, 1960; Rutschke, 1966; Durrer & Villiger, 1966, 1970). Additional melanin in the outer exocuticle provides an absorbent back-ground for transmitted and extraneous wavelengths, hence purifying the reflected colour. Circumstantial evidence for the role of melanins is found in the restrictive co-occurence of pigmentation, iridescence and epicuticular lamination. Specific chemical analysis is still required to verify the identity of the electron-dense layer as melanin.

The specific coloration of structurally-coloured tiger beetles depends upon the relative thicknesses of the epicuticular layers. In C.formosa, C. scutellaris and C. splen-dida, the dimensions of these layers are sufficient to cause the constructive interference of visible wavelengths of light. In epicuticles with regular laminations and a relatively smooth microsculpture, as in C. scutellaris, the reflectance colours are purer.

In some areas of C. repanda elytra, epicuticular laminations are too wide to reflect light at visible wavelengths (Fig. 17). However, visible wavelengths are reflected from areas where the layers are sufficiently thin in the highly variable epicuticle, e.g. within the alveoli. Furthermore, visible wavelengths may be reflected from thicker layers at oblique incidence due to the effect of angle of incidence on the wavelength of maximum reflectance. Thus, while the layers of C. repanda may reflect maximally above 760 nm, shorter red wavelengths will be reflected by the highly angled pattern of ridges in the microsculpture. The results of these factors is a relatively poor reflec-tance of any one visible wavelength, producing an overall flat brown appearance upon casual observation. Under the dissecting microscope, a wide range of reflected colours (predominantly red) can be observed from various points on the surface micro-sculpture. Although at first glance the elytra of C. repanda do not appear to be structurally coloured, the epicuticular structure is consistent with the epicuticular reflectors of other iridescent species.

Thus, tiger beetle structural coloration depends upon three variables: the thick-ness, uniformity and microsculpture of the reflecting layers. In species which reflect a narrow band of short wavelengths, the epicuticular bilayers are relatively thin and the intensity and purity of the colour depend upon the severity of the alveolate microsculpture. Several species which appear brown or coppery bear relatively thick and variable epicuticular bilayers. The deeply pitted microsculpture of the reflector further diminishes the intensity and purity of the reflected colour, resulting in a broad range of reflected wavelengths of low intensity. Small variations in the basic multi-layer structure of the epicuticle are responsible for the wide variation in colour obser-ved in the genus Cicindela.

The authors thank Dr S. Meier, University of Texas, for the use of the SEM, Dr Bird and K. Kotora of Rutgers University, for their assistance in the reflectance spectrophotornetry, and the Texas Department of Parks and Wildlife for permission to collect specimens. We especially appreciate the assistance of Susan Houghton, MBL, Woods Hole, MA, who sectioned and photographed the TEM material. This study was supported partially by NSF Grant 2610711250.