Pterin-pigmented nanospheres create the colours of the polymorphic damselfly Ischnura elegans

Abstract

Animal colours commonly act as signals for mates or predators. In many damselfly species, both sexes go through a developmental colour change as adults, and females often show colour polymorphism, which may have a function in mate choice, avoidance of mating harassment and camouflage. In the blue-tailed damselfly, Ischnura elegans, young males are bright green and turn blue as they reach maturity. Females are red (rufescens) or violet (violacea) as immatures and, when mature, either mimic the blue colour of the males (androchrome), or acquire an inconspicuous olive-green (infuscans) or olive-brown (obsoleta). The genetic basis of these differences is still unknown. Here, we quantify the colour development of all morphs of I. elegans and investigate colour formation by combining anatomical data and reflectance spectra with optical finite-difference time-domain simulations. While the coloration primarily arises from a disordered assembly of nanospheres in the epidermis, morph-dependent changes result from adjustments in the composition of pterin pigments within the nanospheres, and from associated shifts in optical density. Other pigments fine-tune hue and brilliance by absorbing stray light. These mechanisms produce an impressive palette of colours and offer guidance for genetic studies on the evolution of colour polymorphism and visual communication.

1. Background

Coloration plays a crucial role in communication among animals, in particular, in predator deterrence and deception, mate choice and territorial displays [1,2]. To decode the signalling language of animals, we have to understand both their visual systems and their colour signals [3,4]. Odonates (dragonflies and damselflies) are an especially interesting group in this respect as they are one of the earliest branching extant insect taxa [5]. They have evolved sophisticated eyes expressing an unusually large variety of visual opsins [6] with the potential for well-developed colour vision [7–11]. Odonate coloration is complex and dynamic even after the final moult. As adults, males and females often differ in colour (sexual dimorphism), and one sex, generally the female, may occur in two or three colour variants (sex-limited colour polymorphism) [11–13]. In addition, developmental colour transitions during sexual maturation are widespread [11,12], and various species show reversible, temperature-dependent changes in coloration [14], sometimes including the eyes [15].

Our knowledge about the mechanisms that produce odonate body colours and colour transitions is scarce. The pigments that contribute to the coloration are embedded either in the cuticle or in living epidermal cells below the cuticle [16]. A redox reaction of ommochrome pigments in the epidermis causes a developmental colour change from yellow to red in several dragonflies [17]. The distinct blue colour of other odonates has been attributed to a layer of small spherical granules called nanospheres, which develop in the epidermis within a few hours after the final moult [16,18,19]. Dark pigment, normally located proximal to this layer, migrates distally and destroys the dense packing of the nanospheres in the dull phase of temperature-related colour changes [18]. While the wings of many odonates are transparent, those of some species are conspicuously coloured. Recent studies have shown that colour formation on the wing membranes relies on pigmentation, optical interference by multilayers in the cuticle, light scattering by waxy structures, or a combination of these mechanisms [20–26].

Here, we investigate the body colours in the adult stages of the blue-tailed damselfly, Ischnura elegans, a model species in research on female colour polymorphism that also goes through a developmental colour change [27,28]. Females have a red (rufescens) or violet (violacea) thorax when immature, and turn into three distinct sexually mature morphs, coloured blue (androchrome), olive-green (infuscans) or olive-brown (obsoleta). The thorax of the mature male is invariably blue, while younger males show a bright green coloration. A well-defined black stripe on the thorax of all other morphs is replaced by a fuzzy reddish or brownish line in rufescens and obsoleta females. Developmental changes also affect the blue eighth abdominal segment on the tail, after which the species is named. This segment becomes brownish to dark-brown in infuscans and obsoleta females.

Breeding and hybridization experiments indicate that the colour morphs are genetically determined by three alleles on an autosomal locus with female-limited phenotypic expression following a dominance hierarchy [29]. Some genes that are differentially expressed in the sexes and colour morphs have been identified in I. elegans [30] and a closely related species (I. senegalensis, [31]), but the regulatory network controlling the colour polymorphism and developmental colour transitions is still unknown. We describe the colour development of the male and all three female morphs of the blue-tailed damselfly in detail, present an optical explanation of their colours and identify the major processes underlying the colour changes. Our study is the most comprehensive analysis of body coloration and colour development in an odonate species so far and it provides an informative basis for genetic investigations on the evolution of colour polymorphism and visual communication.

2. Methods

2.1. Animals

We caught adult blue-tailed damselflies (I. elegans) at two locations in southern Sweden (N 55°41′ 10″, E 13°5′ 10″ and N 55°41′ 24″, E 13°9′ 57″) from May through August between 2009 and 2013. Previously reported catch rates of androchrome : infuscans : obsoleta morphs at these sites were approximately 7 : 2 : 1 and 8 : 1 : 1, respectively [32]. For developmental time series, we marked individuals on the abdomen with a colour code using acrylic paint (Tamiya colour; Tamiya Europe GmbH, Fürth, Germany) and released them into a large outdoor flight tent (Solig; Ikea, Älmhult, Sweden). The tent was placed next to a pond in half-shade and enclosed natural vegetation, partly dangling into a little water basin. Temperatures in the tent ranged from a minimum of 9°C at night to a maximum of 33°C during the day. Small insects caught in the habitat of the damselflies were provided as live food.

2.2. Spectral reflectance of body parts

We measured reflectance spectra of coloured parts of the thorax and the compound eyes using a diode-array spectrophotometer (S2000; Ocean Optics, Dunedin, FL, USA) with a fibre-optic light guide (200 µm diameter, FC-UV200-2-ME; Avantes, Apeldoorn, Netherlands) positioned at a distance of approximately 5 mm at an angle of 45° from the surface of the sample. A white standard (WS-2; Avantes) served as a reference. The spectral reflectance of animals used for histological sections was measured before they were sacrificed on the day they were caught or after they had been kept at 6–10°C overnight. In developmental time series, we recorded the spectral reflectance of individuals for up to 23 days on a daily basis. In addition, we documented their colour change on photographs taken by a Coolpix 8800 VR camera (Nikon, Tokyo, Japan).

2.3. Histology

The thorax and the compound eyes of decapitated damselflies were isolated and the thorax was divided into right and left halves. In some specimens, we removed the flight muscles to expose the epidermis and achieve better penetration. All samples were fixed in 2% paraformaldehyde and 2.5% glutaraldehyde in 0.1 or 0.15 M sodium cacodylate buffer (pH 7.2) for 30–100 min at room temperature (short fixation) or for 24 h to two weeks at 4°C (long fixation), and rinsed in buffer. We varied the duration of the fixation step, because certain structures require longer fixation, while some pigments dissolve with extended incubation time [16]. For electron microscopy, the tissue was post-fixed in 1% osmium tetroxide in buffer for 1–2 h and rinsed. All samples were dehydrated and embedded in Epon following standard procedures. Sections of 3–5 µm thickness were mounted on glass slides and photographed, unstained or stained with toluidine blue, under an Axiophot light microscope (Zeiss, Jena, Germany) with a DS-Fi2-U3 CCD camera (Nikon, Tokyo, Japan). Ultrathin sections were mounted on copper grids, stained with 2% uranyl acetate and lead citrate in distilled water, and investigated with a transmission electron microscope (JEOL JEM-1230; JEOL, Tokyo, Japan).

2.4. Absorbance and reflectance of epidermal layers

We imaged cross-sections of the integument, obtained as described above, under a Zeiss Axio.Scope A1 light microscope (Zeiss, Oberkochen, Germany) with reflected and transmitted light using a Point Grey Grasshopper 3 camera (FLIR, Richmond, Canada). Additionally, thorax halves of one or two damselflies of each morph were embedded in Richard-Allan Scientific Neg-50™ freezing medium (Thermo Scientific, Kalamazoo, MI, USA) and sectioned at −22°C in a Microm HM 560 cryostat (Thermo Scientific). We collected approximately 10 µm thick cross-sections of the integument on slides coated with chrome alum gelatin and stored them at a temperature below −20°C until further processing. Absorbance spectra of the different layers within the epidermis of frozen or fixed sections were recorded using a microspectrophotometer, which consisted of an adapted Axio.Scope A1 light microscope (Zeiss, Oberkochen, Germany), a 100 W halogen lamp as light source and a spectrometer (QE Pro; Ocean Optics, Dunedin, FL, USA). To prevent degradation effects, the frozen sections were removed from the dry ice container, placed on the microscope stage and measured within approximately 20 s.

2.5. Pigment identification

The eighth abdominal segment and thorax halves of freshly killed or frozen animals were placed in aqueous ammonia (33%), which extracts pterins [33–35], or in hydrogen peroxide solution (30%), which bleaches melanin and ommochromes [36,37]. Controls were monitored in distilled water for more than 4 h to exclude any post-mortem colour change. After up to 30 min in contact with the solution, the tissue was retrieved and the reflectance spectra before and after treatment were compared.

2.6. Imaging scatterometry

The spatial reflection characteristics of the integument were measured by an imaging scatterometer [38,39]. We attached a coloured piece of the thorax of a freshly killed animal to a glass micropipette and positioned it at the first focal point of the ellipsoidal mirror of the scatterometer. A white light beam with a narrow aperture (less than 5°) was focused onto a small circular area (diameter approx. 60 µm) of the sample, and the spatial distribution of the far-field scattered light (the scatterogram) was monitored by a digital camera (Olympus DP-70). A flake of magnesium oxide, which acted as a white diffuser, served as a reference object.

2.7. Finite-difference time-domain modelling

Light scattering by the internal structures of the epidermis was simulated with the three-dimensional finite-difference time-domain (FDTD) method, using Lumerical 8.16 (Lumerical Solutions, Vancouver, Canada), a commercial-grade Maxwell equation solver. The photonic structure inside the epidermal cells was approximated by a randomly-arranged assembly of nanospheres with diameters varying randomly between 0.28 and 0.40 µm, as measured in electron microscopic sections (electronic supplementary material, table T1). We assumed that the nanospheres were arranged at random in a watery medium with refractive index (RI) n_w = 1.33, and that they had a base RI of 1.57 at 600 nm (close to that of transparent chitinous cuticle, [40]) and contained various pterin pigments [35,41]. The (real part of the) complex RI was calculated based on pterin spectra by applying Kramers–Kronig theory for absorbing media as in [41,42].

2.8. Model of predator colour vision

We estimated the visual contrast of the colour morphs against a background of green reed for a predator with tetrachromatic colour vision, the Eurasian blue tit (Cyanistes caeruleus), using the receptor noise-limited model of colour discrimination [43]. For details on the method, see [44].

3. Results

3.1. Colour morphs and developmental colour transitions

Male and female morphs of I. elegans differ in colour in a number of characteristic ways. Here, we show the appearance and spectral reflectance of all morphs during their adult development, from the newly enclosed teneral (electronic supplementary material, figure S1) via immature stages to the mature state (figure 1; for a more detailed developmental series and values, see electronic supplementary material, movie M1 and file D1, respectively). In our survey, maturity was typically reached after one to two weeks depending on temperature, although one individual had not completed adult colour development after three weeks. The thorax of tenerals is pale and grey-brown except for the black or brownish markings embedded in the cuticle. In males, the development proceeds via a bright green immature state to the blue-coloured mature morph. A secondary reflectance peak in the ultraviolet initially exists in the green phase. The ultraviolet peak disappears before the main peak above 500 nm shifts towards shorter wavelengths by another peak rising just above 400 nm, which gradually takes over when the animal turns blue. Androchrome females arrive at the same blue colour as mature males, yet take a distinct developmental path. They first pass through a violet state (violacea) with a broad reflectance peak around 400 nm and then obtain the reflectance peaks above 500 and 400 nm, similar to the males. Females of the infuscans morph also undergo the violacea stage, but only develop the green peak above 500 nm with little reflectance at lower wavelengths. Their colour finally shifts to a duller, olive-green hue. The female morph that differs the most from the others in its developmental path is obsoleta, which goes through a red state (rufescens) before turning olive-brown when becoming mature. In the reflectance curves, a shoulder above 400 nm and a later emerging one above 500 nm occur at similar wavelengths as the peaks of the other morphs. However, the reflectance is generally high above 400 nm, particularly at longer wavelengths. In parallel to the thorax, the lower two-thirds of the eye change colour in a defined sequence in all morphs, but are never violet or red (figures 1, 5, 6; electronic supplementary material, figure S1 and movie M1).

Figure 1.

Colours of the blue-tailed damselfly Ischnura elegans. Reflectance spectra of the flanks of the thorax at selected stages during male (a) and female maturation (b–d). The colours of the thorax and compound eye, documented on RGB photographs under standard illumination, are illustrated on the sketches to the right. For a more detailed compilation of developmental stages, see electronic supplementary material, figure S1 and movie M1.

Figure 5.

Sketch of the proposed changes during the male (a) and the three female colour developments (b–d). The thoracic integument is divided into the cuticle and the distal and proximal layers of the underlying epidermis. All three layers contribute to creating colour and contrast, while major modifications take place in the distal epidermal layer. Text in the arrows lists the order of events. Timeline (horizontal) not to scale. Colours indicate the reflected spectrum of the respective layer. NS, nanospheres; X, xanthopterin and E, erythopterin in nanospheres. Note that the second and third stage shown for the male differ in UV reflectance (see also electronic supplementary material, movie M1).

Figure 6.

Colour formation in the compound eyes of Ischnura elegans. (a) Light microscopic longitudinal section through the central part of the eye stained with toluidine blue. (b–d) TEM cross-sections through ommatidia at the first (b) and second (c) level of pigment cell nuclei and at the level of the pigment cap at the distal tip of the rhabdom (d) as indicated by the arrowheads in (a). (e–g) Magnification of the pigment granules visible in (b–d): nanospheres (b,e), big granules of lower electron density (c,f) and smaller granules of high electron density (d,g). (h) The spectral reflectance of the eye of the male shown in the inset (for values, see electronic supplementary material, file D1) has the same shape as that of the blue-green thorax of males and androchrome females (figure 1; electronic supplementary material, movie M1). Light microscopic and TEM sections are from two males with blue-green eyes. BM, basement membrane; C, cornea; CC, crystalline cone; N, nucleus; NS, nanospheres; PC, pigment cap; R, rhabdom (photosensitive part of the cells); s.d., standard deviation. Scale bars: (a) 50 µm, (b–d) 5 µm, (e–g) 1 µm.

3.2. Structure of the epidermis

The developmental colour transitions must be due to pigmentary and/or structural changes in the upper body layers. We therefore investigated the anatomical basis of the coloration in cross-sections through the integument of the thorax under a microscope with transmitted light (figures 2 and 3a–d). In all morphs, a transparent cuticle of 5–10 µm thickness covers the epidermal cells. Only in body regions that appear dark, the cuticle is opaque and black or brownish. The epidermal cells contain pigments that are arranged in a specific manner. In rufescens, red granules are visible directly below the cuticle (figure 3d). Pigment distribution in obsoleta is more variable, with orange or light-brown granules appearing directly below the cuticle and/or in a proximal layer (figure 3c). All other colour morphs of I. elegans except the very young adult stages possess dark-brown pigment and a smaller amount of light-brown pigment that together form a clearly defined proximal layer in the epidermis (figures 2 and 3a,b).

Figure 2.

A characteristic layer in the epidermis of Ischnura elegans. Cross-section through the thoracic integument of a blue male, imaged under a microscope with reflected and transmitted light. The yellowish layer (arrowheads) directly below the transparent cuticle produces a blue reflectance. Absorbance spectra of this layer in the frozen sections of different morphs resemble pterin spectra (compare figure 4 and [33]).

Figure 3.

The integument of Ischnura elegans. (a–d) Cross-sections through the thoracic integument of the female colour morphs androchrome (a), infuscans (b), obsoleta (c) and rufescens (d), viewed under a microscope with transmitted light. The cuticle is transparent or half-transparent with black or ginger stripes. Pigments in the underlying epidermis are distributed such that they produce a distal and a proximal layer. (e–h) TEM sections of the thoracic epidermis. The distal epidermis contains nanospheres. Bigger pigment granules build the proximal layer (g,h) or intermingle with the nanospheres in the distal layer as in rufescens (f). Scale bars: (a) 40 µm, (b–d) 10 µm, (e) 3 µm, (f–h) 1 µm. For sections of males, see figure 2 and electronic supplementary material, figure S1.

We used transmission electron microscopy (TEM) to investigate the fine structure of the epidermal cells and found that their distal layer always contained nanospheres with a diameter of approximately 0.3 µm (figure 3e–h; electronic supplementary material, figure S1 and table T1). While the nanospheres are loosely arranged and surrounded by cell organelles indicating high cell metabolism and synthetic activity in early developmental stages (electronic supplementary material, figure S1b–g), they become more tightly packed over time (figure 3e). In rufescens and sometimes also in obsoleta, larger, electron-dense pigment granules interrupt the packing of the nanosphere layer (figure 3e,f). The proximal layer of the epidermis is filled with pigment granules of different sizes, bigger granules with a diameter of about 1 µm and smaller ones of about half this size (figure 3e,g,h).

Based on previous studies [15,16,18,19], the observed structures suggest that the coloration of the damselflies mainly originates from the proximal nanosphere-containing layer, while the pigmented proximal layer serves as a contrast-enhancing, stray light-absorbing backing. Light micrographs confirm that light reflected from the nanosphere layer of epidermal thorax sections of a mature male appears blue (figure 2, see also [18]), while light transmitted through the same region of the epidermis appears yellowish (compare the upper and lower light micrographs in figure 2).

3.3. Pigmentation and optics

Previous studies on odonates (cited in [19]) found that the proximal layer of the epidermis contains ommochromes. However, the dark-brown colour of the majority of the pigment in the proximal epidermis of all morphs of I. elegans, except obsoleta and its younger developmental stages, suggested the presence of melanin, a pigment also found in Calopteryx, another damselfly genus [20]. We therefore investigated the proximal epidermal layer of the thorax by microspectrophotometry (electronic supplementary material, figure S2). Absorbance spectra obtained from sections of a mature male, a violacea and an infuscans were indeed very similar to known melanin spectra [45,46], while the absorbance spectrum of an obsoleta closely resembled the absorbance of ommochromes [47].

We also measured the absorbance of the nanosphere layer of the different morphs, yielding spectra reminiscent of pterin pigments [20,33,41,48] (figure 2). Because pterins (and not ommochromes or melanins) are dissolved in aqueous ammonia [33,35], we tested the possible presence of pterins by placing thorax halves in aqueous ammonia. This caused the rapid disappearance of the typical colours of the morphs and resulted in a dull ginger, reddish-brown or dark violet hue depending on the amount of melanin and ommochrome in the cuticle and epidermis (electronic supplementary material, figure S3).

Comparing the measured absorbance spectra with the absorbance spectra of pterins (figure 4a) strongly suggested the presence of xanthopterin and erythopterin in the distal epidermal layer (figure 2). We hence concluded that the nanospheres in the distal epidermal layer contain pterin pigments, which play a role in the coloration of the different damselfly morphs. Yet, with a nanosphere diameter of approximately 0.3 µm, light scattering and/or interference by the spheres must also contribute to colour formation. To estimate this contribution, knowledge of the RI of the nanospheres is crucial.

Figure 4.

Modelling the optics of the nanosphere layer. (a) Normalized absorbance spectra of pterins identified in pierid butterflies [33]. (b,c) Refractive index spectra calculated for a medium containing pure xanthopterin or erythropterin (b) or mixtures of both (c) with a peak absorption coefficient of 1 µm⁻¹. (d) Simulation volume used for the FDTD modelling. (e,f) Reflectance spectra obtained by FDTD modelling of a watery medium with randomly arranged nanospheres that contain either (e) pure xanthopterin or erythopterin or (f) mixtures of both pigments. Colour coding in (e) as in (b), and in (f) as in (c).

In an absorbing medium, the absorption coefficient is intimately connected to the imaginary part of the RI of the material, and the latter is directly related to the real part of the RI via the Kramers–Kronig relations (e.g. [42,49]). We thus calculated the RI dispersion curves for xanthopterin, erythopterin and various mixtures of these two pigments using a peak absorption coefficient of 1 µm⁻¹. As figure 4b,c and electronic supplementary material, figure S4 show, the presence of strongly absorbing pigments causes distinct anomalous dispersion, i.e. a non-monotonic behaviour of the RI spectra.

In order to understand the optical mechanisms that produce the colours of I. elegans, we used the calculated RI spectra in FDTD simulations on idealized photonic structures that mimic the anatomical conditions observed in the damselfly morphs in different developmental stages (figure 4d). We assumed that the nanospheres of the distal epidermal layer (i) are filled with (a mixture of) pterin pigments, (ii) are randomly packed, (iii) have varying diameters around a mean value of approximately 0.3 µm and (iv) are surrounded by a watery medium (see Methods).

The reflectance spectrum calculated for the medium with only xanthopterin-containing nanospheres has a prominent band in the blue wavelength range, while purely erythropterin-containing nanospheres produce a reflectance peak in the green and a secondary peak in the UV (figure 4e). Mixtures of both pterins (as indicated in figure 2) show reflectance spectra with pronounced shoulders in the blue wavelength range (figure 4f). A simulation of empty, i.e. non-pigmented (transparent), nanospheres arranged in a random fashion yielded a low reflectance with a broad peak in the UV and blue wavelength range, similar to the reflectance of violacea females (compare grey curve in figure 4e with the earliest curves in figure 1b,c). The thickness and filling fraction of the nanosphere layer have an influence on the strength of the reflectance, but not on the shape of the reflectance spectrum (electronic supplementary material, figure S6).

The measured reflectance spectra of the animals (figure 1; electronic supplementary material, movie M1 and file D1) are strikingly similar in shape to the modelled spectra (figure 4e,f). We hence concluded that the colours of the different developmental stages of the damselflies result from the generation of nanospheres in the distal epidermal layer and the synthesis (or incorporation) and gradual change of the different pterin pigments in the nanospheres (figure 5).

We furthermore investigated the directionality of the light scattering by the green thorax of a young male using an imaging scatterometer. Illumination of the thorax with a narrow-aperture beam resulted in a diffuse, green reflection pattern with a directional white area at scattering angles of approximately 30° (electronic supplementary material, figure S5a). The whitish reflection moves in a very mirror-like fashion when rotating the sample (not shown) and is reminiscent of a surface reflection from the cuticle. By contrast, the light reflection from the epidermal nanosphere layer is diffuse and fills the entire hemisphere above the object. The experimentally obtained scatterogram corresponds well with the spatial distribution of light scattered from the nanosphere layer in the FDTD model (electronic supplementary material, figure S5b). The far-field projection of the simulated reflectance confirms that the diffuse reflectance originates from the nanosphere layer.

4. Discussion

4.1. The coloration of Ischnura elegans

Our study of the damselfly I. elegans has revealed the mechanisms behind an unusual case of animal coloration. In most insects, colours originate either from pigmentation or from photonic structures in the cuticle, which are generally static, i.e. the colours do not change after the final moult. However, our analyses suggest that in I. elegans—and likely in many other odonate species—several processes in the epidermal cells are responsible for a dynamic colour display, resulting in time- and morph-dependent colour changes. These processes include the development of a distal layer of nanospheres in each epidermal cell [12,15,16], which creates the broad UV-blue reflectance of violacea. When the nanospheres are filled with morph-specific mixtures of strongly absorbing pterins, their RI changes, producing the green and blue colours of males, and androchrome and infuscans female morphs. A disordered array of nanospheres, together with red or orange-brownish ommochrome pigment, leads to the red and dull olive-brown coloration of rufescens and obsoleta (figures 3 and 5).

A comparison of the modelled spectra for various pterin mixtures with the reflectance spectra of the morphs during their developmental stages allows us to infer a schematic model of the major processes underlying thorax coloration of I. elegans (figure 5). All variants start with a random arrangement of few granules, resulting in pale colours. In males, erythopterin is expressed, while the nanosphere layer develops, producing a strong green- and UV-reflection. Subsequently, xanthopterin is added, reducing the UV peak and generating a blue peak, and finally erythopterin is degraded, leading to the blue colour of the mature male. The same final blue is reached in a different way in androchrome females. The purple colour of their immature violacea stage likely results from the nanosphere layer initially developing without pigmentation. Both pterin pigments are then expressed with a slight delay between them, and eventually erythopterin is degraded. Infuscans females also go through the violacea stage, but only express small amounts of xanthopterin and do not degrade erythopterin in a later step, thus conserving a green coloration. In rufescens and obsoleta, the red, orange or light-brown ommochrome pigment granules in the epidermis create local disorder. This reduces the coherent optical effects of the pterin-containing nanospheres and results in the red or dull olive-brown reflectance shape that is typical of pigmentary colours (figure 5).

As far as we have seen (M.J.H. and A.K. 2009–2014, unpublished data), no morph of I. elegans displays a dull phase in response to low temperatures as described in other odonate species [14]. Interestingly, temperature-dependent colour changes can include the eyes of damselflies [15], and our anatomical investigations show that the structures responsible for the colour of the compound eyes are very similar to those of the thorax (figure 6).

4.2. Colour from a combination of coherent scattering and pigmentary absorption

The coloration of I. elegans constitutes a remarkable optical system in its own right. The densely packed assembly of nanospheres in a watery matrix in the distal layer of the epidermis presents an efficient way to create bright, angle-independent colours (figure 4; electronic supplementary material, figure S5), where colour tuning can be achieved with different pigments. We note here that non-pterin, short-wavelength absorbing material (presumably uric acid) has been found in other insects with similar nanosphere structures [50]. We therefore assumed that the base RI of the nanospheres was approximately 1.57, slightly higher than 1.5, the typical RI of normal protein, and close to that of unpigmented chitinous cuticle [40]. A higher base RI does not alter the conclusions reached here, since it increases the RI contrast in a constant manner in the wavelength range of interest and therefore only quantitatively shifts the reflectance spectra (see electronic supplementary material, figure S6). Clearly, more studies are needed to unravel the fine details of these colours, including the precise chemical composition of the base material and the exact size distribution of the nanospheres in the distal layer.

Colours that are produced by structures of random or correlated disorder and tuned or adjusted by pigments have been found in butterflies [41] and in feather barbs of parrots, budgerigars and kingfishers [51–53]. However, in all of these cases, ‘dead’ solid materials are responsible for the colour, prohibiting developmental changes like those reported here. In fact, a bird has to moult in order to change feather colour, and butterflies will not change wing colours after emergence from the chrysalis.

Melanin or other broadband absorbing pigments like ommochromes are localized in the proximal layer backing the photonic structures (see figures 2 and 3; electronic supplementary material, figure S2). This layer aids the coloration in absorbing stray light that is not reflected by the photonic structure, and thereby increases the saturation and brilliance of the colour. A similar mechanism has been found in butterflies [54], cephalopods [55] and chameleons [56].

4.3. The biological significance of damselfly colours

As odonate colour vision extends from the UV into the red wavelength range [7–9,57], the different colours of the morphs may serve as visual signals in mate choice and/or as camouflage. Male I. elegans use colour as one of several cues for choosing a mate [13,27,28,58,59] and rarely approach immature females [59]. The male-like blue colour might protect androchrome females from constant harassment by males [60–62], but, because of their conspicuous body coloration, mature males and androchrome females could be more vulnerable to predation than green or brownish morphs. Model calculations estimating the colour contrast between the different morphs and their natural background confirm that a common predator, such as the Eurasian blue tit, can detect the blue morphs more easily than the green or brown morphs (electronic supplementary material, figure S7). The development of the nanospheres and the pigmentation in the epidermal cells may thus play a crucial role in the dual function of damselfly body coloration for camouflage and sexual signalling.

4.4. Understanding the genetic basis of colour polymorphism

Ischnura elegans is a model organism for the biological functions of colour polymorphism [63] and has been one of the first odonate species, for which transcriptomic data became available, allowing studies on gene expression [30,64]. Our findings indicate that the distinct colours of the morphs result from differences in the onset and time course of the synthesis or transport of pterin pigments, ommochromes, melanin and epidermal nanospheres. While some genes controlling pigment syntheses in I. elegans have been identified and their sex- and morph-specific expression has been investigated [30,64], gene regulation remains unclear. We highlight the need for studying the genetic basis of colour polymorphism in I. elegans, particularly the upstream control elements of the synthesis and transport pathways of the components listed above.

5. Conclusion

We have described the physico-chemical basis of the striking colours of I. elegans and shown that their coloration is mainly due to a combination of pigments and nanospheres acting as photonic structures in the distal layer of the epidermal cells. Changes in the pigment composition and the packing of the nanospheres during maturation modify the reflectance spectra of the morphs. This dynamic colour display seems to be important for visual signalling and camouflage. Our results should provide ample inspiration for further investigations on the genetic basis of colour polymorphism.

Data accessibility

All data used for our analyses can be found in this article and its electronic supplementary material.

Authors' contributions

M.J.H., B.D.W. and A.K. designed the study; M.J.H. and B.D.W. performed and analysed experiments; B.D.W. carried out FDTD simulations; O.L. calculated colour contrasts; A.K., B.D.W. and M.J.H. drafted the text and the figures; all authors edited and approved the final version of the manuscript.

Competing interests

We declare no competing interests.

Funding

This research was supported by the Swedish Research Council (2009-5683 to A.K.), the Carl Tryggers Foundation (to A.K.), the Swiss National Science Foundation (PBZHP3-131029 to M.J.H.), the National Centre of Competence in Research ‘Bio-Inspired Materials' and the Ambizione program of the Swiss National Science Foundation (168223 to B.D.W.).