The pronotum and elytra of the jewel beetle Chrysochroa fulgidissima display a prominent green colour, which is interrupted by purplish-reddish stripes (Fig. 4A). The ventral side of the exoskeleton of the abdomen has a rather green-goldish colour (Fig. 4B). Transverse transmission electron microscopy sections demonstrate in the distal 2–3 μm of the cuticle a number of differently stained layers (Fig. 4C–H). The cuticular material is mainly chitin, but as the elytra observed in transmitted light show a characteristic brown melanin colour, the darker layers must be due to infused melanin (Hariyama et al. 2005; Stavenga et al. 2011; Yoshioka and Kinoshita 2011; Onelli et al. 2017).

Colouration of the jewel beetle Chrysochroa fulgidissima. A Dorsal view, with green and red-purplish areas of the pronotum marked by 1 and 2, and green and red-purplish areas of the elytra marked by 3 and 4. B Ventral view, with green-goldish areas of the exoskeleton marked by 5 and 6. C-H TEM images of areas 1–6. I-N Real part of the refractive index profiles of the central 0.5 μm wide cuticular lane at 550 nm. O-T Reflectance spectra calculated with the estimated refractive index profiles for the areas 1–6. Scale bars: A and B 1 cm, C-H 2 μm

The TEM images allow estimation of the refractive index gradients (see Methods). The refractive index profiles of 0.5 μm wide cuticular lanes, implemented in a matrix transfer calculation procedure, yielded the reflectance spectrum of each lane, as shown in detail in Fig. S2. To avoid cluttering, Fig. 4I–N show the refractive index profiles of the central lane of Fig. 4C–H, O-T present the calculated reflectance spectra. The reflectance peak wavelengths of the green and red areas 1 and 2 of the pronotum are 516 and 725 nm (Fig. 4A, O,P), those of areas 3 and 4 of the elytra are 526 and 705 nm (Fig. 4A, Q, R), while the ventral golden areas 5 and 6 yielded peak wavelengths 580 and 557 nm (Fig. 4B, S, T). The half-bandwidths are 50–70 nm.

These results can be compared with reflectance spectra measured with a bifurcated probe spectrophotometer. The spectra vary with the local areas. Figure 5 shows a few examples. The shape of the calculated spectra corresponds well with the measured spectra but the peak wavelengths of the latter spectra are generally slightly shifted to longer wavelengths.

Reflectance spectra of C. fulgidissima measured with a bifurcated probe. A Green and red areas of the pronotum. B Green and red areas of the elytra. C Ventral exoskeleton

This impressive buprestid displays an overall bright-green colour (Fig. 6A), which is created by a large number of chitin-melanin layers (Fig. 6B). Similar as in the TEM images of C. fulgidissima, the density contrast in the layers initially increases with depth, but it rapidly decreases in the more proximal layers. The density converted to refractive index (Fig. 6C) yields reflectance spectra, peaking around 535 nm, with bandwidth ~ 50 nm (Fig. 6E), slightly narrower than those of experimentally obtained spectra, ~ 70 nm, which have a peak wavelength around 560 nm (Fig. 6D).

Colouration of Eucallopistus castelnaudii. A Dorsal view, showing an overall bright-green colour of both pronotum and elytra; scale bar: 1 cm. B TEM section, showing a regular stack of layers; scale bar: 1 μm (derived from Fig. A.2 of (Chow 2022). C Profiles of the real part of the refractive index at 550 nm of 5 adjacent cuticular lanes of panel B, each 0.5 μm wide, and their average (bold green curve). D Reflectance spectra measured from locations 1 and 2 of panel A. E Reflectance spectra calculated for the 5 cuticular lanes of panel C and their average (bold green curve)

Colouration of Eurema gigantea. A Dorsal view, showing a dull-coloured pronotum and overall red-coloured elytra, with greenish rims; scale bar: 1 cm. B TEM section, showing a regular stack with 5 melanised layers; scale bar: 1 μm (derived from Fig. 8 of (Durrer and Villiger 1972). C Profiles of the real part of the refractive index at 550 nm of 5 adjacent cuticular lanes of panel B, each 0.5 μm wide, and their average (bold green curve). D Reflectance spectra measured from locations 1–3 of panel A. E Reflectance spectra calculated for the 5 cuticular lanes of panel C and their average (bold green curve)

Durrer and Villiger published the first demonstration of melanised multilayers in a buprestid, Euchroma gigantea (Durrer and Villiger 1972). This indeed very large beetle has prominently red-coloured elytra with greenish border areas (Fig. 7A). The reported TEM micrographs clearly show the presence of some five melanin layers, with spacing depending on the area’s colour (unfortunately the magnification factors given with the different TEM images are not fully consistent). Figure 7B shows the section of a greenish area, which yields the refractive index profiles of Fig. 7C and the reflectance spectra of Fig. 7E. Figure 7D presents reflectance spectra measured from the pronotum and elytral areas.

The mostly green elytra of Chrysochra saundersii are marked by yellow-whitish midbands (Fig. 8A). In TEM micrographs of a green area, the density of the second melanin layer is remarkably reduced (Fig. 8B), yielding a refractive index profile that distinctly differs from those of the other species (Fig. 8C). The regular spacing of the layers nevertheless results in a reflectance spectrum with a prominent peak at ~ 540 nm (Fig. 7E). Reflectance spectra measured from the green elytra areas peak at ~ 570 nm (Fig. 8D). The reflectance spectrum of the midband indicates the presence of a UV-absorbing pigment (Fig. 8D).

Colouration of Chrysochra saundersii. A Dorsal view, showing a greenish pronotum and bright-green elytra with yellow-whitish midbands; scale bar: 1 cm. B TEM section, showing a regular stack of layers, with a strongly reduced density of the second melanin layer; scale bar: 1 μm (derived from Fig. A.64 of (Chow 2022). C Profiles of the real part of the refractive index at 550 nm of 5 adjacent cuticular lanes of panel B, each 0.5 μm wide, and their average (bold red curve). D Reflectance spectra measured from locations 1–3 of panel A. E Reflectance spectra calculated for the 5 cuticular lanes of panel C and their average (bold red curve)

Similar as C. saundersii, the elytra of C. fulgens display a yellow-whitish midband, but the other areas are markedly multicoloured (Fig. 9A). Whereas the midbands lack a structural colour, the multicoloured areas have clearly a structural basis. The reflectance spectra calculated with the refractive index profiles derived from a TEM micrograph peak at 670–680 nm (Fig. 9B, C, E). Reflectance spectra measured from various locations in the multicoloured areas widely vary. Figure 9D presents a few examples.

Colouration of Chrysochroa fulgens. A Dorsal view, showing the multicoloured pronotum and elytra, together with yellow-whitish midbands; scale bar: 1 cm. B TEM section; scale bar: 1 μm (derived from Fig. A.108 of (Chow 2022). C Profiles of the real part of the refractive index at 550 nm of 5 adjacent cuticular lanes of panel B, each 0.5 μm wide, and their average (bold red curve). D Reflectance spectra measured from locations 1–3 of panel A. E Reflectance spectra calculated for the 5 cuticular lanes of panel C and their average (bold red curve)

The elytra of Chrysochroa gratiosa feature a pale midband, similar as C. saundersii and C. fulgens (Fig. 10A). The elytra rims are green, similar as in E. gigantea, but the coloured areas of the elytra are extraordinary, as they display a rather dull reddish-brown colour (Fig. 10A). TEM images of the latter areas nevertheless show a common multilayer structure (Fig. 10B), which transferred into refractive index profiles yield narrowband reflectance spectra (Fig. 10E). The calculated spectra do not closely correspond to measured reflectance spectra (Fig. 10D, E). However, this is probably due to strongly varying elytral structures, suggested by largely varying experimental reflectance spectra obtained with the bifurcated probe (Fig. 10D). Measurements made with a microspectrophotometer (Fig. S3), yielded reflectance spectra of the green rim similar to that of Fig. 10D, #4, and spectra similar to those of the main elytra area peaking in the far-red with a sideband in the green (Fig. 10D, #1,3) were obtained from the areas in between the red valleys (Fig. S3, #3). The reflectance spectra of the walls of the red-valleys showed a single peak in the red (Fig. S3, #2), similar to the calculated spectrum of Fig. 10E).

Colouration of Chrysochroa gratiosa. A Dorsal view, showing a multicoloured pronotum and also elytra, with yellow-whitish midbands; scale bar: 1 cm. B TEM section (derived from Fig. A.87 of (Chow 2022); scale bar: 1 μm. C Profiles of the real part of the refractive index at 550 nm of 5 adjacent cuticular lanes, 0.5 μm wide, and their average (bold red curve). D Reflectance spectra measured from locations 1–3 of panel A. E Reflectance spectra calculated for the 5 cuticular lanes of panel C and their average (bold red curve)

Figure 11 shows two additional cases with area-dependent colours. The area-dependent reflectance spectra result from the extreme flexibility as well as variability that is possible by slight changes in the multilayer reflector’s dimensions. Note that a change in peak wavelength from 600 to 700 nm only needs a layer thickness increase of no more than ~ 10%.

Variously coloured jewel beetles. A Reflectance spectra of locations 1–5 of Chrysochroa fulminans. B Reflectance spectra of locations 1–4 of Cyphogastra javanica

The cuticular surface of various jewel beetles appears to vary widely (Fig. 12). The multilayer reflector calculations assume a perfectly flat surface, but this is only locally reasonably well approximated. Light-microscopic observations readily reveal that the surface is modulated by 5–10 μm-sized pits (e.g., Fig. 12G–I) and/or can even contain up to 100 μm-sized valleys (e.g., Fig. 12A-C, E, J, M). The valleys contain a hair or bristle, which is usually rather short (Chow 2022) but occasionally quite pronounced, as for instance in E. castelnaudii (Fig. 12D–F). The valleys traverse the endocuticle, which is clearly seen in images of the pale midband areas (Fig. 12N, Q).

Surface structure of jewel beetles. A-C Chrysochroa fulgidissima; scale bar: 200 μm. D-F Eucallopistus castelnaudii. G-I Euchroma gigantea. J-L Chrysochroa saundersii. M-O Chrysochroa fulgens. P-R Chrysochroa gratiosa. el – elytra, pr – pronotum, mb – midband, ea – elytra apex, em – elytra margin

The surface of C. gratiosa elytra is excessively sculpted, with valley walls that are distinctly red. Similar optical effects, but less expressive, are observable in other species, as the colour of the (often somewhat faint) light reflected by the valley walls differs from surface reflections in between the valleys (Fig. 12A, B, E, L). As a consequence of the angle-dependent reflections by the sculpted surface elements, the beetle’s appearance thus can strongly depend on the angle of illumination and observation. Although this phenomenon satisfies the definition of iridescence, the beetle optics essentially differs from that of a smooth multilayer reflector. The latter’s reflectance spectrum is strongly polarization dependent and shifts to shorter wavelengths with increasing angle of light incidence, which occurs with e.g. C. fulgidissima (Stavenga et al. 2011; Schenk et al. 2013; Chow 2022) and also E. castelnaudii. Very different angle-dependent colouration occurs when observing a variously-coloured jewel beetle from different directions.