Experimental degradation of helicoidal photonic nanostructures in scarab beetles (Coleoptera: Scarabaeidae): implications for the identification of circularly polarizing cuticle in the fossil record

Abstract

Scarab beetles (Coleoptera: Scarabaeidae) can exhibit striking colours produced by pigments and/or nanostructures. The latter include helicoidal (Bouligand) structures that can generate circularly polarized light. These have a cryptic evolutionary history in part because fossil examples are unknown. This suggests either a real biological signal, i.e. that Bouligand structures did not evolve until recently, or a taphonomic signal, i.e. that conditions during the fossilization process were not conducive to their preservation. We address this issue by experimentally degrading circularly polarizing cuticle of modern scarab beetles to test the relative roles of decay, maturation and taxonomy in controlling preservation. The results reveal that Bouligand structures have the potential to survive fossilization, but preservation is controlled by taxonomy and the diagenetic history of specimens. Further, cuticle of specific genus (Chrysina) is particularly decay-prone in alkaline conditions; this may relate to the presence of certain compounds, e.g. uric acid, in the cuticle of these taxa.

1. Introduction

The cuticle of insects is composed primarily of chitin microfibrils embedded in a protein- and lipid-rich matrix, forming a multilaminate system [1]. These microfibrils constitute the bulk of the endo- and exocuticle layers that provide rigidity to the exoskeleton and are covered by a lipid-rich epicuticle that forms a protective barrier and a sensory interface [2]. In the exocuticle, chitin microfibrils are usually mutually parallel in each layer [2,3]. In some beetles, however, microfibrils of the outer exocuticle are slightly offset in successive layers, giving rise to helicoidal ‘Bouligand' structures [4] that generate circularly polarized light (CPL; herein this term includes near-circular polarization) [2,3].

Bouligand structures have been reported in plants [5], crustaceans [6] and beetles; the last of these includes eight genera of Scarabaeidae and one of Hybosoridae [7,8]. The biological significance of CPL is controversial, with much debate focusing on whether CPL functions in signalling (compare [9,10] with [11] for behavioural studies) or is merely a by-product of the fibrillar and lamellar nature of the cuticle [12,13]. Recent discoveries of dedicated physiological characters for CPL detection in the eyes of stomatopod crustaceans [14–16] support the hypothesis that CPL has important, albeit incompletely understood, biological functions.

Fossils have the potential to shed light on the functions and evolutionary history of CPL use in insects, providing Bouligand structures can be preserved. Fossil examples, however, have not been reported. This may reflect a real biological signal or an artefact of the fossilization process, i.e. reflecting poor preservation potential. This is a common issue in palaeobiological studies due to the incompleteness of the fossil record [17]. The two scenarios can, however, be discriminated with the aid of controlled laboratory experiments [18] designed to investigate the taphonomy of specific taxa, tissue structures and/or biomolecules, thus informing on their preservation potential [19]. Previous taphonomic experiments investigating aspects of the fossilization process in various organisms, including arthropods, have identified a suite of factors which can influence the preservation potential of organisms (reviewed in [19,20]). Known environmental factors include the extent of transport, relative timing of transport and decay, water temperature, salinity, pH, the presence of microbes, sediment type, sediment consistency, the availability of O₂ and mineral ions, and burial temperature and pressure. Biological factors include the nature and extent of decay, body geometry and tissue composition. Clearly, not all potential factors can be investigated in a single experiment. Instead, rigorous taphonomic experiments typically investigate the impact of a small number of specific variables in a controlled manner (the ‘reductionist' approach [18]). In particular, previous taphonomic experiments have shed light on the preservational conditions and processes involved in the fossilization of multilayer reflectors and three-dimensional photonic crystals in insects [21]. Here, we experimentally degraded extant scarab beetles with and without Bouligand structures in controlled laboratory decay and maturation experiments. Untreated and degraded cuticles were analysed using scanning electron microscopy, reflectance microspectrophotometry and Mueller matrix spectroscopic ellipsometry.

2. Material and methods

2.1. Material

Dried specimens of five green beetle taxa representing three sub-families of Scarabaeidae were procured from commercial suppliers: Chrysina boucardi (n = 3) and Chrysina gloriosa (n = 3) (both Rutelinae), Ischiopsopha jamesi (n = 3) and Torynorrhina flammea (n = 3) (both Cetoniinae) and Gymnopleurus virens (n = 5) (Scarabaeinae). The cuticle of T. flammea does not generate CPL and served as a control; the cuticle of all other taxa used exhibits CPL [3,22–24].

2.2. Light microscopy

Beetle elytra were imaged using a Leica DM2700M stereomicroscope coupled to a Leica MC190HD camera using unpolarized light from a LED lamp. Elytra were tested for circular polarization using a Nikon D3100 camera and a Leica EZ4 W stereomicroscope equipped with left-handed and right-handed circular polarizers (Edmund Optics).

2.3. Scanning electron microscopy

Samples of elytra were mounted on aluminium stubs using carbon tape, sputter-coated with Au/Pd and examined using an FEI Inspect F50FE-SEM and an FEI Quanta 650 FE-SEM at 5 kV in secondary electron mode.

The period of the helicoidal structure, called pitch (p), corresponds to a full turn (360°) of the chitin fibrils and consists of two successive lamellae (thickness α), i.e. p = 2α. The lamellar thicknesses were obtained from scanning electron microscopy (SEM) images (see below) using the fast Fourier transform (FFT) function in ImageJ [25].

2.4. Microspectrophotometry

Reflectance spectra were collected on elytra with a Zeiss Axioskop 2 microscope using the x50 objective and equipped with a halogen lamp (HAL 100) set to 6 V and an Ocean Optics USB4000 spectrometer. Specimens were manually positioned so that the region to be analysed was approximately at normal incidence to the light path. Acquisition time varied from 5 to 10 s. Between three and five replicates were used for each sample. Data were normalized relative to a white standard (Certified Reflectance Standard, Labsphere) and analysed using Ocean Optics SpectraSuite software. Models of the theoretical wavelengths produced from the exocuticular lamination were based on the circular Bragg equation for normally incident light λ_max = pn_av = 2αn_av, where λ_max is the wavelength at the circular Bragg peak and n_av is the average of the two refractive indices in the anisotropic lamellae plane. In this study, we do not consider dispersion and use n_av = 1.56 [26].

2.5. Mueller matrix spectrometric ellipsometry

Normalized Mueller matrices of cuticle samples were measured with a dual rotating compensator ellipsometer (RC2, J. A. Woollam Co., Inc.) at an angle of incidence of 20°. Focusing probes were used to obtain a spot size of less than 100 µm. The Mueller matrices presented here are primary data without any filtering or data processing. Further technical details and limitations are reported in [27].

2.6. Decay experiments

Experiments were designed to approximate decay of insect cuticles in still aqueous environments as most fossil insects are from ancient lacustrine sediments [28]. We considered the following variables: temperature, pH, microbes and sediment. Water temperature is a major control on preservation as it controls the rate of decay [29]. We conducted our experiments at 30°C to promote decay and delivery of useful results within the project timeframe (even though benthic waters can be much cooler than this in many extant lakes). Our experiments used an alkaline pH as insect cuticles decay slightly faster in such media than in acidic media (M.E.M. 2018, unpublished data). Although previous experiments have demonstrated that decay is enhanced in the presence of sediment [30], even particular clay minerals [31], and in the presence of an external source of microbes [30] (i.e. not those in/on the carcass itself), we conducted our experiments using distilled water without inocula of sediment and bacteria in order to mitigate against occlusion of the cuticle surface by sedimentary particles and/or a microbial biofilm.

Small fragments of elytra (ca 5 × 5 mm²; n = 3 for each taxon) were immersed in individual glass vials each filled with 10 ml of water at pH 11.5 ± 0.1 and degraded for three months in a controlled temperature chamber at 30°C (±1°C) in the dark. The pH of the decay media was measured at the start and termination of the experiments. In order to track short-term changes in visible colour, additional specimens of C. gloriosa and C. boucardi were degraded in identical conditions to those described above and photographed at two-week intervals for three months.

2.7. High-temperature maturation experiments

Small fragments of untreated elytron (ca 5 × 5 mm²) were wrapped in aluminium foil and matured at one of each of the following conditions for 24 h at 1 bar: 95°C, 100°C, 125°C, 150°C, 175°C, 200°C.

2.8. Chemical tests

A series of supplementary experiments investigated the impact of other factors on the preservation of Bouligand structures, using techniques that have been applied to studies of the origins of structural colour in other taxa [32]. In particular, we carried out tests designed to reveal which cuticle layer(s) are involved in production of the visible hue, and whether the observed hue is entirely structural or includes a pigmentary component. Small fragments of untreated elytra from C. boucardi and T. flammea were treated in the following ways: immersion in (i) 8% KOH (60°C, 1 bar, 24 h; to remove the epicuticle [32]), (ii) 10% H₂O₂ (60°C, 1 bar, 24 h; to remove the procuticular layers [32]), (iii) 20% H₂SO₄ (20°C, 1 bar, 24 h; to bleach melanin [32]) and (iv) wrapped in polytetrafluoroethylene (200°C, 1 bar, 24 h) to simulate maturation in anoxic conditions. To test the influence of the aluminium foil upon maturation, some cuticle samples were also matured without being wrapped and placed in a glass Petri dish. The results were assessed qualitatively via visual inspection of cuticle after treatment.

Untreated cuticle samples were also immersed in methyl salicylate (refractive index 1.54) to test whether the observed colour is structural [33].

3. Results

3.1. Visible hue

The visible hue was investigated by eye and light microscopy. Untreated cuticle in all taxa is highly reflective and exhibits a bright green colour (figure 1a–e) that in C. gloriosa is combined with golden stripes (figure 1b) and in I. jamesi with red regions (figure 1d).

Figure 1.

Photographs of the five scarab taxa studied, with and without polarizing filters, on untreated, decayed and matured cuticle. Scale in the left-hand column: 20 mm (except G. virens: 10 mm); scale in the right-hand column: 5 mm.

Colour did not change during decay of G. virens, I. jamesi and T. flammea (figure 1c–e). During decay of C. gloriosa and C. boucardi, however, colour altered from green to red-orange (figure 2a), and ultimately to a matte dark brown (figure 1a,b); the region of C. boucardi cuticle that remained green after decay was not in contact with the decay medium (figure 1a). Alteration initiated on the golden stripes of C. gloriosa and extended progressively to the green cuticle regions. In contrast, alteration of hue in C. boucardi initiated at cut edges and cuticular pores (see electronic supplementary material, figure S1).

Figure 2.

Light micrographs and scanning electron micrographs of the surface of untreated, decayed and matured elytra. Scale: 50 µm for light micrographs (LM) and 10 µm for scanning electron micrographs (SEM).

All cuticle samples were altered to black during maturation, both in the presence (figure 1a–e) and absence of oxygen (electronic supplementary material, figure S2a) except for C. boucardi which altered to a dark green colour (figure 1a). Wrapping cuticle in aluminium foil does not influence the results (electronic supplementary material, figure S2b).

The supplementary experiments reveal that the green hue of the cuticle in all taxa is less intense after H₂SO₄ treatment and is lost upon treatment with KOH (electronic supplementary material, figure S3). Treatment with H₂O₂ resulted in loss of visible colour in C. boucardi and reduction in intensity plus a blue shift in hue in T. flammea.

3.2. Surface texture

The surface texture was investigated using both light and electron microscopies. Green cuticle regions in C. boucardi, C. gloriosa, I. jamesi and T. flammea exhibit a tessellated surface pattern defined by a combination of pentagonal, hexagonal and heptagonal units 7–10 µm wide (figure 2a,b,d,e). In C. boucardi and C. gloriosa, the units are green with yellow cores (figure 2a,b), those in I. jamesi are green throughout (figure 2d) and those of T. flammea are green-blue throughout (figure 2e). These tessellated surface patterns are superimposed onto taxon-specific microtextures, as evidenced by SEM micrographs. In C. gloriosa, green (but not gold-coloured) regions exhibit a stellate pattern comprising a series of star-shaped units each ca 5 µm wide (figure 2b). In I. jamesi, the elytral surface texture (both green and red regions) is dominated by near-parallel rows of elongate projections, each ca 5 µm long (figure 2d). Torynorrhina flammea exhibits a reticulate structure, with alveoli ca 10 µm wide, sometimes with a dark central spot (figure 2e). A similar, but subtler, structure is present in C. boucardi (figure 2a). Gymnopleurus virens elytra do not exhibit surface patterns or microtextures (figure 2c).

After decay, brown regions of C. boucardi and C. gloriosa cuticle retain a tessellated surface texture but the polygonal units no longer differ in hue between core and margins (figure 2a,b). SEM analysis reveals that the green regions of degraded C. boucardi cuticle exhibit surface microtextures that are identical to those in untreated cuticle (figure 2a). Brown cuticle regions show surface pitting (figure 2a). The surface microtextures of C. gloriosa are clearly altered during decay, via expansion of the star-shaped units and loss of surface cohesion (figure 2b). Finally, decay does not affect the visible colour and surface microtexture of G. virens, I. jamesi and T. flammea (figure 2c–e).

Following maturation, all taxa retain surface microtextures despite loss of visible colour (figure 2a–e).

3.3. Reflectance microspectrophotometry

Reflectance spectra (figure 3; electronic supplementary material, figure S4) are consistent with the observations on visible hue described above. In C. boucardi (figure 3a), untreated cuticle samples display two primary peaks at ca 545 and 585 nm, plus secondary peaks (up to three) between 605 and 675 nm; after decay, spectra of green cuticle regions change slightly (the primary peak at 545 nm shifts to 555 nm; not shown) while spectra from brown cuticle regions lack peaks (figure 3a). After maturation, cuticle exhibits a single broad peak centred on 530 nm and a secondary peak at 585 nm.

Figure 3.

(a–e) Reflectance spectra from untreated (solid black), decayed (red dash) and matured (blue dot) samples of beetle elytra. The spectra were chosen as mostly representative as possible of the global results.

In C. gloriosa (figure 3b), spectra of untreated cuticle samples exhibit a primary peak at ca 530 nm with secondary peaks at 580 nm, 605 nm and 630 nm. Spectra of decayed and matured cuticle lack peaks.

Spectral characteristics of G. virens vary markedly among specimens. In four specimens, spectra of untreated cuticle usually exhibit two primary reflectance peaks at 485–520 nm and 530–570 nm. The remaining two specimens, however, show three peaks at ca 500 nm, 530 nm and 555 nm (figure 3c). In all specimens, decay resulted in a shift of these peaks towards longer wavelengths; spectra of matured cuticle lack peaks.

Spectra of untreated I. jamesi cuticle show one primary peak at ca 550 nm and, sometimes, a secondary peak at ca 635 nm for male specimens or 605 nm for females (figure 3d). Decay did not result in major spectral changes (peak position shifted by ±5 nm); spectra of matured cuticle lack peaks.

Finally, spectra from T. flammea cuticle can exhibit one, two or three peaks at 535 nm, 555 nm and 580 nm (figure 3e). Decay results in minimal shift of these peaks (±5 nm) and the appearance of additional major peaks at 510 nm and 615 nm. Spectra of matured cuticle lack peaks.

3.4. Polarization analysis

Polarization was analysed qualitatively through circular polarizers attached to visualization systems (camera and stereomicroscope), and quantitatively by Mueller matrix spectrometric ellipsometry. All CPL-producing taxa studied exhibit a green colour when observed using the left-handed circular polarizer but a brown colour when observed using the right-handed circular polarizer, confirming reflection of left-handed polarized light (figure 1a–d). The brown (not black) colour indicates some reflected broadband CPL. These effects are apparent in cuticle regions of all colours and from all body regions; the structures responsible for circular polarization are thus ubiquitous and independent of visible hue. Torynorrhina flammea appears green when observed using each polarizer, confirming its optical inactivity (figure 1e).

Circular polarization is retained where original cuticle colour persists during decay (observed in G. virens, I. jamesi and T. flammea; figure 1c–e). Visual assessment of polarization is not possible where visible hue altered from green to matte brown (as for decayed cuticle; figure 1a,b) or black (as for matured cuticle; figure 1a–e).

These results are confirmed and explored further by Mueller matrix analysis on cuticle samples from C. boucardi and I. jamesi. Figure 4 presents the Mueller matrix element m₄₁, which is non-zero if the light has a circularly polarized component (the full Mueller matrices are presented in electronic supplementary material, figure S5). The negative values obtained on m₄₁ verify that left-handed CPL is reflected. The m₄₁ obtained on untreated cuticle of C. boucardi exhibits two major spectral bands centred around 550 nm and 800 nm corresponding to reflection of green and red/infrared CPL, respectively (figure 4a). For I. jamesi, a single narrow spectral band is obtained around 540 nm (figure 4b). The effect of decay and maturation is detailed in figure 4a,b. On the brown-coloured decayed cuticle of C. boucardi, the Mueller matrices lack features in all the off-diagonal block elements (electronic supplementary material, figure S5a), including the m₄₁ where the signal disappears (figure 4a). By contrast, left-handed polarized light is reflected from decayed cuticle that retain a green hue (figure 4b; I. jamesi). A blue-shift of the m₄₁ feature from 540 nm to 529 nm is, however, visible.

Figure 4.

Mueller matrix 41 obtained on degraded cuticle: (a) from untreated (solid black), decayed (red dash) and matured (blue dot) samples of C. boucardi, (b) from untreated (solid black), decayed (red dash) and matured (blue dot) samples of I. jamesi, (c) from temperature-series maturation of C. boucardi and (d) from temperature-series maturation of I. jamesi.

The maturation process has similar effects on I. jamesi and C. boucardi whereby loss of colour is accompanied by a weakening of the circular polarization, and ultimately its loss in I. jamesi (figure 4a,b).

Figure 4c,d shows the effect of thermal treatment during maturation: the m₄₁ signal becomes less negative with increasing temperature, approaching zero at 200°C for C. boucardi and 175°C for I. jamesi, whereby the circular polarization is lost. For C. boucardi, the green reflection band is more susceptible to alteration than the red/infrared: it is altered at lower temperature relative to the red/infrared band and indicates a colour shift from green to red/brown. At 200°C, while the visible contribution has disappeared, the red/infrared band is still present. For I. jamesi, a small blue shift occurs along with increasing temperature before the production of a flat signal.

3.5. Scanning electron micrographs of vertical cross sections

All CPL-producing beetle cuticle samples exhibit a thin homogeneous superficial epicuticle underlain by a thinly laminated exocuticle and, in turn, by a thicker endocuticle comprising a cross-ply arrangement of bundles of chitin microfibrils (figure 5; electronic supplementary material, figure S6). Except for G. virens, the exocuticle is subdivided into outer and inner regions with thin and thick lamellae, respectively. This subdivision is not apparent in the non-CPL-producing cuticle of T. flammea, but striking vertical structures are present (figure 5e; electronic supplementary material, figure S6e). The dimensions of various cuticular components are summarized in table 1.

Figure 5.

SEM micrographs of longitudinal cross sections through untreated, decayed and matured elytra, showing the epicuticle and part of the outer exocuticle. Scale: 2 µm.

Table 1.

Thicknesses of cuticular components measured manually from scanning electron micrographs of green cuticle. Data are based on at least 10 measurements from two different images.

		thickness			decrease (%)
taxon	layer	untreated (u)	decayed (d)	matured (m)	u/d	u/m
C. boucardi	epicuticle (µm)	0.2–1.0	0.2–0.6	0.3–0.9	33	0
	outer exocuticle (µm)	11.4–14.1	5.3–6.4	10.2–12.1	54	13
	inner exocuticle (µm)	3.8–6.4	5.0–7.0	2.0–4.0	18	41
	lamellae (nm)	100–270	70–120	170–230	49	8
C. gloriosa	epicuticle (µm)	0.3–2.3	0.2–2.6	0.4–2.0	8	8
	outer exocuticle (µm)	5.9–10.5	4.0–7.7	7.2–8.7	29	3
	inner exocuticle (µm)	3.0–6.4	12.9–16.4	5.2–6.3	212	22
	lamellae (nm)	130–240	60–160	150–190	48	8
I. jamesi	epicuticle (µm)	0.25–1.0	0.5–0.7	0.3–0.5	4	36
	outer exocuticle (µm)	13.0–16.4	8.2–10.5	9.4–11.3	36	30
	inner exocuticle (µm)	7.2–10.0	18.7–20.3	7.0–8.7	127	9
	lamellae (nm)	110–200	130–190	120–170	3	6
G. virens	epicuticle (µm)	0.2–0.7	0.2–0.5	0.1–0.3	22	56
	exocuticle (µm)	5.4–13.7	5.4–8.8	8.6– 13.9	26	18
	lamellae (nm)	90–200	90–160	100– 230	14	14
T. flammea	epicuticle (µm)	0.4–0.6	0.6–0.9	0.3–0.4	50	30
	exocuticle (µm)	18.6–24.4	27.5–27.8	17.8 –20.5	29	11
	lamellae (nm)	130–210	140–200	130–240	0	9

The morphology of the lamellae in the outer exocuticule varies markedly among specimens. Lamellae form distinctive concentric cusps 7–10 µm wide (each centred on the surficial ‘star' of the texture) in the green cuticular regions of both Chrysina species (figure 5a,b; electronic supplementary material, figure S6a,b). Lamellae are flat and parallel to the cuticle surface in G. virens (figure 5c; electronic supplementary material, figure S6c), T. flammea (figure 5e; electronic supplementary material, figure S6e) and gold regions of C. gloriosa, and undulose in I. jamesi (figure 5d; electronic supplementary material, figure S6d). Lamellae are flat and parallel to the surface in the inner exocuticle of all taxa (electronic supplementary material, figure S6).

Decay does not alter the morphology of the exocuticular lamellae but reduces the thickness of the latter in C. boucardi and C. gloriosa by 48% and 49%, respectively (table 1). By contrast, maturation results in a less marked reduction in lamella thickness in all taxa (6–14%).

4. Discussion

4.1. Source of visible colour

We examined multiple specimens of each of the extant taxa studied in order to assess the extent of any intraspecific variations in colour (common in extant beetles, e.g. T. flammea [24], Cetonia aurata [34] and Plateumaris sericea [35]). Our reflectance data on the green cuticle regions are consistent with previously reported λ_max values for C. gloriosa (530 and 580 nm [36]), C. boucardi (519, 588 and 620 nm [22]) and T. flammea (490, 530, 568 and 584 nm [24]). These spectroscopic data, plus those for I. jamesi and G. virens, are consistent with our models of the theoretical wavelengths (table 2): observed and predicted λ_max values differ by only 0.3–6.8% across taxa, supporting that the visible colour is primarily structural; the slight discrepancy between spectroscopic data and theoretical data may originate from a slight deviation of the cuticle surface from normal during spectrophotometric analysis, or from the value of the refractive index used.

Table 2.

Experimental and theoretical maximum reflection wavelengths of green and red regions of untreated elytra. Lamellar thicknesses are from FFT of scanning electron micrographs; theoretical maximum reflection wavelengths are from Bragg's equation, using a refractive index for chitin of 1.56 and the lamellar thicknesses from FFT.

taxon [region]	lamellar thickness (nm)	λtheoretical (nm)	λexperimental (nm)	difference (%)
C. boucardi [green]	175 (+232)	546	584	+6.7
C. gloriosa [green]	172	537	530	−1.3
I. jamesi [green]	185	577	550	−4.9
I. jamesi [red]	203	633	636	+0.3
G. virens [green]	152	474	507	+6.5
T. flammea [green]	173	540	533	−1.5

Immersion of structurally coloured materials in a medium of similar refractive index as the chitin matrix results in loss of iridescence; this has been considered a proxy for structural colour [33]. Immersion of cuticle from C. boucardi in methyl salicycliate (with a refractive index of 1.54, close to chitin of 1.56) does not affect the observed hue, suggesting that the visible colour results from a combination of structural colour (i.e. generated by the Bouligand structure) plus pigment, i.e. melanin. This is consistent with our experimental evidence for melanin in the cuticle (i.e. a reduction in the intensity of the observed colour upon treatment with H₂O₂) and with studies of C. boucardi where a contribution from melanin is required for consistency between experimental and theoretical models of visible colour [22].

Finally, the supplementary experiments demonstrated that the epicuticle, but not the endocuticle, is also involved in the production of visible colour. The role of the former may relate to scattering of light from epicuticular lipids [37].

4.2. Origins of colour changes during taphonomic experiments

Structural colours in scarabs have been attributed to constructive interference produced by highly regular spacing of lamellae in the exocuticle [3]. The colour reflected depends on the periodicity of the structure and its refractive index [3,26] while its brightness and purity relate (among other factors) to the order of the structures and the thickness of the colour and CPL-generating stack [26]. Based on the lamellar thickness in decayed cuticle and using a refractive index of 1.56 for chitin, C. boucardi and C. gloriosa should exhibit λ_max in the UV part of the spectrum (λ_max = 296 nm for C. bourcardi, λ_max = 343 nm for C. gloriosa). Similarly, after maturation, lamellar thickness predicts that samples should exhibit λ_max in the blue-green to green-yellow regions of the visible spectrum (λ_max = 544 nm for C. boucardi, 561 nm for C. gloriosa, 484 nm for I. jamesi, 515 nm for G. virens and 577 nm for T. flammea). Instead, in all cases, the cuticle is brown and their reflectance spectra lack peaks (in the visible range).

In brown decayed cuticle, the thickness of exocuticular lamellae is clearly altered and the structural colour contribution shifts toward shorter wavelengths in the UV part of the spectrum. The brown colour displayed by decayed cuticle must, therefore, relate to the presence of melanin (see above), which commonly functions as a light-absorbing layer in photonic nanostructures [37].

For matured cuticle, there is no evidence for striking changes in the thickness or regularity of exocuticular laminae. This suggests that the colour change (to black) in the experimentally matured beetles is chemical rather than structural in origin. The supplementary experiments demonstrate that the preservation potential of the cuticle is independent of ambient oxygen levels. Critically, our experiments demonstrate that preservation of colour is subject to a strong temperature control: colour survives below, but not above, 150°C. Colour change is thus unlikely to indicate marked dehydration (as in the scarab Dynastes hercules [38]) since cuticle does not change in visible colour after experimental maturation for 24 h at 95°C, 105°C or 125°C. Rather, these data suggest that colour changes during maturation most probably relate to polymerization, which is typical of maturation (but initiates during decay) [39]. Notably, polymerization alters the refractive index of the cuticle, which affects the absorbance windows, thus controlling the visible colour produced. This process has been reported from other experimentally degraded structurally coloured insect cuticle [21].

Combined, these results suggest that preservation of Bouligand structures in fossils is controlled by diagenetic history, in particular, the maximum burial temperatures of the host sediment.

4.3. Changes in polarization during experiments

All CPL-producing taxa studied reflect left-handed polarized light, which is consistent with previous studies [3,8,22,36,40]. In most elements, oscillations are observed. These are due to interference effects in various parts of the cuticle and can in some cases be used to determine pitch profiles and cuticle internal dimensions, as shown in [41].

Our data demonstrate a strong relationship between either loss or survival of visual colour and polarization upon degradation. Typically, the features observed on the decayed cuticle of C. boucardi indicate that the cuticle is a dielectric material, e.g. like a plastic surface, reflecting light as an isotropic material without any active Bouligand structure or other anisotropic features. Moreover, the ellipsometry data recorded on matured cuticle confirm that the optical effect of Bouligand structures is indeed destroyed during maturation, and not shifted toward wavelengths outside the visible spectral range, despite survival of the Bouligand structure itself (figure 5). The loss of the CPL effect may rather be due to chemical alteration causing a major change in refractive index or increased near-surface absorption preventing the light from interacting with the Bouligand structures.

The shift in the circular polarization signal differs between taxa. Since the structures responsible for CPL are indeed preserved (figure 5; electronic supplementary material, figure S6), these differences likely reflect a difference in chemistry between C. boucardi and I. jamesi.

4.4. Taxonomic controls on fossilization

The preservation potential of insects in the fossil record is influenced by body size, morphology and the hardness of the cuticle [42,43,44]. Small and robust beetles, and their more sclerotized body parts, e.g. elytra, are more likely to be preserved [43,45,46,47]. The cuticle of scarabs is one of the most robust among insects [43]; preservation of Bouligand structures in fossil scarabs does not, therefore, appear unlikely.

Our study reveals a strong taxonomic control on the preservation of visual colour and polarization properties of Bouligand structures during experimental decay, whereby the response of Chrysina cuticle differs from that of all other taxa studied. This contrasts with the taphonomy of multilayer reflectors, which is independent of taxonomy [33]. Chrysina boucardi and C. gloriosa differ in their ecology from the other taxa studied here: Chrysina species are nocturnal and feed on foliage whereas the other taxa are diurnal and feed on dung (G. virens) or on flowers and fruits (I. jamesi and T. flammea). These parameters are unlikely to relate to the preservation of Bouligand structures. The exocuticular laminae in green cuticle regions of the studied Chrysina show a distinctive cuspate arrangement (see [48]), but it is unclear how this physical structure could relate to preservation of Bouligand structures. Finally, experimental degradation does not alter the structure of the exocuticular laminae in any taxon studied.

Instead, the taxonomic signal most probably reflects chemical alteration of the cuticle. Variations in chitin chemistry are unlikely to be a factor as the cuticle of all arthropods comprises α-chitin [49] and degrades thermally only above 300°C [50]. Given the contribution of the lipid-rich epicuticle and of melanin to visible colour in the taxa studied (see above), it is possible that such cuticular components could chemically alter during the experiments, thus changing the refractive index of the cuticle. Indeed, epicuticular lipids are sensitive to basic treatment (similar to our decay experiments [32]) and are known to vary in composition among taxa [51]; the cuticular lipid inventory (see review in [52]) and pigment chemistry (see review in [53]) of the taxa studied here have not, however, been characterized, so we cannot assess whether these indices contribute to the observed taxonomic signal.

The studied taxa do, however, vary in other aspects of cuticle chemistry, notably the presence/absence of uric acid, which increases refractive index and enhances the birefringence of the chitin matrix [54], thus resulting in more efficient reflection of incoming light. The taxonomic distribution of uric acid in insect cuticle has been investigated only in a single study in which uric acid was reported from three Chrysina species but not in the only non-Chrysina genus (Potosia speciossima) examined [54]. In living insects, the acidic physiological pH [55] implies that uric acid is present in the acid form (C₅H₂N₄O₃H₂). At high pH (as in our decay experiments), however, uric acid deprotonates into urate (C₅H₂N₄O₃²⁻). The impact of this chemical change on the optical properties of cuticle is unknown. It is, therefore, not implausible that the loss of visible colour and polarization in Chrysina cuticle during decay may reflect the degradation of uric acid to urate or other chemical compound.

These results may have broader applications in the fossil record, suggesting the presence of a taphonomic bias related to taxon-specific cuticle chemistry. While acidic environments have been hypothesized to promote chitin degradation through hydrolysis [56], we hypothesize that cuticle rich in uric acid will be relatively susceptible to degradation in alkaline conditions. Fossil insect biotas preserved in sediments that have experienced relatively alkaline conditions during decay may thus be biased towards taxa without cuticular uric acid; this bias may be less pronounced in biotas that experienced relatively milder conditions. Future experimental studies will characterize more fully the impact of uric acid on the taphonomy of cuticle in diverse insect taxa.

5. Conclusion

This work is the first study investigating the preservation of Bouligand structures in insects. Our results demonstrate that Bouligand structures can survive decay and maturation and thus have the potential to be fossilized. Future analyses of fossil insects preserving structural colours should, therefore, test for ultrastructural evidence for Bouligand structures. The presence of strong temperature-related control on preservation suggests that the fossil record of Bouligand structures may be biased, whereby preservation is more likely for cuticle subjected to lower temperatures during diagenesis. This may belie a broader taphonomic bias relating to the nature of epicuticular lipids and/or pigments (e.g. melanin) and/or the concentration of uric acid, whereby insects with cuticle that contains uric acid may be less likely to survive decay in alkaline conditions than cuticle lacking uric acid. These hypotheses can be tested by future experimental investigations and analyses of fossil insects preserving structural colours should test for ultrastructural evidence of Bouligand structures. Additionally, geochemical studies should incorporate data on the diagenetic history of the host sediment. This will help to constrain fossil localities where Bouligand structures are more likely to be preserved. Finally, such results have broader implications in the field of bioinspired photonic materials, whereby the stability of Bouligand structures during thermal treatment may be of interest for banknote sensors (anti-counterfeiting) or light-emitting sources (screen technology).

Data accessibility

The data are publically available on CORA (https://cora.ucc.ie/handle/10468/6491).

Authors' contributions

M.E.M. and G.P.O. designed the research. H.A. and K.J. performed Mueller matrix analysis and G.P.O. conducted all other analyses. G.P.O. and M.E.M. interpreted results and wrote the manuscript with contributions from all other authors.

Competing interests

The authors declare no competing interests.

Funding

The research is funded by Science Foundation Ireland ERC Support grant no. SFI 16/ERCS/3904 and European Research Council Starting grant no. H2020-2014-StG-637691-ANICOLEVO awarded to M.E.M.