Visual and optical properties of the blue spots of Taeniura lymma
Natural color is a product of context: of incident light wavelengths, the properties of the color-producing tissue (which dictate the range of wavelengths reflected), and the position and sensory capabilities of the observer, whether predator, prey or potential mate. Given the use of multiple reef microhabitats by the ribbontail ray, one can assume variations in the spectral habitat that the species may occupy (19) (Fig. 1e). For instance, in-water light transmission and attenuation typically increase from backreef to forereef waters, with blue and green wavelengths dominating in eutrophic coastal waters and lagoons, and a greater availability of blue wavelengths when moving to the forereef and offshore waters (19, 20). These differences in light availability can also be affected by natural variations in suspended particulates and dissolved organic matter and may lead to variations in the color perception of the ribbontail ray across local and regional scales (20, 21). Understanding how natural color varies with context can therefore provide critical baseline information, not only on the properties and basis of color production, but also on the role that skin color plays in an animal’s ecology and its structural and/or compositional stability.
In this study, all sampled specimens, including both juvenile and adult stingrays, exhibited the bright electric blue coloration distinctive for this species, displaying spots on their dorsal body, lateral stripes on their tails, and an ovular swath of color surrounding the lower caudal finfold to the tip of the tail and the stinging spine (Fig. 1b). Color profile measurements on ribbontail ray skin in water showed that, in sunlight (Fig. 2a), the juvenile animal’s blue spots showed peaks in the blue region (~ 487 nm), while the non-blue portions of skin strongly reflected in the red (~ 674 nm) (Fig. 2c). In lab conditions (i.e., under a controlled range of incident wavelengths, Fig. 2b), the blue spots predominantly reflected short wavelengths in the blue-cyan region of the visible spectrum (juvenile: ~452 nm; adult: 447 nm), whereas the non-blue regions reflected longer wavelengths. Surprisingly, a strong UV peak (juvenile: ~344 nm and adult: ~335 nm) was common to both blue and non-blue regions of the skin (Fig. 2d,e). This variation between sunlight and lab measurements is likely due to the nature of the incident light in the two conditions, being directional and broad-spectrum in the lab, but diffuse with more variable wavelengths in ambient conditions. Measurements at various angles of light incidence indicated that the blue spots are non-iridescent, confirming our visual observations from intact specimens that the color is unchanging with viewing angle (Fig. 2f). Although non-iridescence is often taken to indicate a lack of structural color, it can also be due to an irregularity of the structure (22) that results in scattering of light, regardless of the angle of incidence; we explore these options at a tissue level below.
The ecology and visual capabilities of this species provide some suggestions as to the roles its electric blue coloration may play for this species. Ribbontail ray can perceive blue wavelengths with absorbance maxima at wavelengths ~ 479 nm (S-cone photoreceptor cells), 500 nm (rod cells) and 557 nm (L-cone cells) (23). Moreover, the ribbontail ray has a specialized visual apparatus for low-light scotopic vision and a greater spatial resolving power compared to other benthic batoids (23, 24), which may facilitate improved transmission of visual cues between conspecifics, especially during crepuscular periods where rays are often seen gathered in shallow waters (17). In these regions, short-wavelength photons dominate the underwater spectrum (25) and the resulting blue illumination may emphasize and further saturate the blue spot pattern. The mixture of yellow/green and blue coloration in the ribbon tail ray is also an effective color combination that transmits well in reef waters, providing the most chromatic contrast for a range of animal visual systems (26) and possibly serving as an aposematic warning system to nearby predators. However, given the reduced color-vision range of other reef fish, the once conspicuous yellow/green and blue colors of the ribbontail ray would quickly merge and fade into the background over the span of a few meters towards the achromatic point (27), suggesting that the blue spot coloration of the ribbontail ray acts primarily as a near-field signal for inter- or intra-specific communication.
Our observations also demonstrate a uniform reflectance of soft UV (UV-A), suggesting that the components responsible for UV reflection should be identical in both blue and non-blue regions of the skin. This was unexpected and suggests the two dominant reflectance peaks (blue and UV) serve different functions. Ribbontail ray visual pigments block UV radiation (28) and therefore UV-reflections from their skin would not be perceived by conspecifics, or even by potential predators (e.g. sharks), since many of those lack UV perception (28, 29). Instead, since behavioral observations show that the ribbontail ray rarely covers itself in the sand like other stingray species (14), UV-A is not especially damaging to skin and coastal sand is known to reflect UV light (Baltic seaside at Juodkrantė, 30), we hypothesize that the UV-reflectivity of ribbontail ray skin provides camouflage in sandy, intertidal habitats, perhaps a useful deterrent against predation by birds (which can perceive UV; 31).
On close examination of skin regions with a stereomicroscope, the blue color, especially in the larger body spots, had the quality of a milky paste, dabbed onto the skin surface, with black dendritic splotches coalescing in a dense layer beneath it; this suggested that the regions responsible for the blue color may be quite superficial (Fig. 2g,h). The microscale splotches also extended into non-blue regions, albeit only sparsely (Fig. 2h). As our specimen lacked skin denticles, the soft nature of its skin allowed us to gently peel away the outer skin layers (~ 0.4 mm, with a black spot left behind on the specimen after peeling) and transfer them onto synthetic substrates to explore the effect of the black dendritic layer on the overall color appearance. The isolated tissue, when affixed to black carbon tape, exhibited a similar blue to that of the intact animal, whereas on a white substrate, the color was much paler and the tissue was almost transparent (Fig. 2i). Previous studies on biological and artificial optical materials have shown that black backgrounds can serve to absorb longer and incoherently scattered wavelengths and help in the saturation of blue color (32, 33). In our study, while the paler color of ribbontail ray tissue on white backgrounds suggests a superimposition of all colors in the spectrum, the brighter blue color on carbon tape indicates that the tissue’s natural black backing acts as a broadband absorber of longer wavelengths and is therefore an important component of the blue coloration.
In contrast to our fresh specimens, in one specimen that was frozen immediately after death for ~ 4 years, all characteristic blue regions were dark blue or nearly black in color (Fig. S1). Strikingly, however, spots cut from this hugely shriveled specimen returned to their natural electric blue immediately when placed in water; moreover, this color switching was reversible, with the blue being recoverable even after multiple freeze-thaw cycles. Likewise, in the tissue affixed to black carbon tape, the color reversal process was repeatedly achieved following de-/rehydration cycles, with the drying of the tissue producing black and re-wetting recovering blue hues, respectively. In fact, even after being left dry at room temperature for six months, the color of the excised tissue returned to blue soon after rehydration. Tissue colors that vary with hydration can be the result of changes in structure and/or refractive index when water leaves the tissue (34, 35), or pigmentary changes arising from pH differences (36). Our observations that the blue skin color of ribbontail ray is highly robust and reversible with de- and re-hydration are not enough to prove a structural origin of color —thus, tissue-level characterizations were necessary.
Unique ultrastructure of blue spots of ribbontail ray skin
The skin of all vertebrates contains specialized color-producing cells called chromatophores. Chromatophores exist in various forms: light-absorbing pigment-containing melanophores (black or brown pigments), erythrophores (red or yellow pigments), cyanophores (blue pigments), and colorless reflective purine-containing leucophores and iridophores (7, 8, 9). The distribution of these cells in skin layers (epidermis and dermis) and their interactions with other tissue components are responsible for the diverse and colorful patterns of fish skin (2, 37, 38).
Despite the deep interest in the denticle morphology and hydrodynamics of shark skin (39, 40), the histology of shark and ray skin is surprisingly poorly described. Our light and transmission electron microscopy investigations showed that, as in other fishes, the skin of ribbontail ray (Fig. 3a) consists of an epidermis (E: 100–250 µm thick), separated from a thicker underlying dermis (D: 350–500 µm) by a thin basement membrane (B: 0.3 µm). In both blue and non-blue skin regions, the dermis was visibly stratified. A horizontal layer (H) with a unique ultrastructure (see below) in the upper dermis was effectively sandwiched between two densely collagenous layers: a thin upper layer of collagen fibers (UC, 1–5 µm) below the basement membrane and a thick (LC, 100–150 µm) layer of interwoven collagen fiber bundles in the lower dermis.
We found three distinct cell types in the skin layers: mucous cells, melanophores (with either black or brown pigments), and specialized cells we refer to here as ‘pale cells’. These three cell types varied in their arrangement, preponderance, and density in the epidermis and dermis and, more importantly, in ways that distinguished blue from non-blue regions. The epidermis of both blue and non-blue areas were histologically relatively similar, dominated by large mucous cells superficially, but containing a scattered mix of mucous cells, melanophores, and pale cells closer to the basement membrane. We verified that the dark dendritic splotches observed by stereomicroscope (see above) were melanophores, exhibiting stellate morphologies and squeezed between the other cell types. The pale cells were so named because of their nearly featureless appearance in histology/light microscopy, such that they initially looked to be vacuolated (our electron microscopy observations revealed this not to be the case; see below). Although blue and non-blue regions were compositionally similar, the blue epidermis tended to contain fewer mucus-producing cells in some regions/specimens, more pale cells, and “black melanophores” (sensu 41, 42) that were more dendritic and darker (in contrast to the non-blue region, which contained less-electron-dense “brown melanophores” and more sparsely distributed “black melanophores”).
Unlike the epidermis, the upper dermis exhibited key ultrastructural differences between blue and non-blue regions of the skin. In blue regions only, the upper dermis (~ 35 µm thick) was densely packed with pale cells, amalgamated into large multi-cell complexes which often enclosed branched black melanophores. The tight association of pale cells with melanophores and their extended dendritic processes resulted in intracellular melanosomes being distributed throughout the pale cell layer, with the local concentration of melanosomes higher deeper in the tissue (corresponding to the “black dendritic backing layer” we observed in the previous section). In contrast, in non-blue regions, the horizontal layer (H) was only one-third as thick (up to 15 µm wide), contained more sparsely distributed black and brown melanophores, and a continuous layer of pale cells. The collagen layers (UC and LC) that surround the horizontal layer (H) were present in both blue and non-blue regions of the skin with no obvious differences, although the lower collagen layer was slightly thicker in the non-blue region. Our electron microscopy investigations revealed that pale cells do in fact contain a highly organized internal structure, which only became visible when samples were carefully and immediately fixed and then contrast-stained. In such samples, the lumen of pale cells was observed to be entirely full of nano-vesicles (~ 126.79 ± 13.08 nm in diameter), the morphology and arrangement of which varied with location in the tissue: whereas pale cells in the non-blue region contained vesicles that were more elongate and random in their arrangement, those in blue skin regions were spherical and mono-disperse with a strikingly uniform arrangement and packing density (~ 28 vesicles/µm). In the upper dermis of the blue skin regions, intracellular melanosomes (M in Fig. 3c,d) were jet black and of variable cross-section, ranging from ellipsoidal to spherical shape and roughly similar in size (500–600 nm), several times larger than pale cell vesicles. In well-fixed specimens, the extremely tight encasement of pale cells around melanophores was particularly impressive — with the many surrounding pale cells creating a nearly uniform, nano-vesicular corona around each pigment cell. In contrast to the non-blue regions, epi-illuminated light microscopy observations of skin cross-sections showed a conspicuous blue hue in blue spot regions, just above the dense black melanophore layer, suggesting that the blue color originates in the overlying pale cell layer (Fig. 3b).
TEM observations supported by SEM imaging of the isolated skin revealed that the spherical nano-vesicles contained crystalline structures (85.01 ± 14.58 nm wide, Fig. 3e,f), often rhomboidal in cross-section, occurring mostly in pairs within the matrix of each vesicle and with no defined orientation. However, in many vesicles, the crystals are extracted during sample preparation, leaving an empty space inside vesicles in TEM images. EDX and Raman spectroscopic measurements (Fig. S2) confirmed that these nano-crystals were nitrogen-rich and made of anhydrous beta-guanine (43, 44). Beta-guanine is a common structural-color-producing purine in bony fishes, typically manifesting as long guanine platelets (each tens of µm long and tens of nm thick), conspicuously stacked inside of iridophore cells. During development, the guanine platelets nucleate from an amorphous precursor inside small vesicles in iridophores, grow, and then coalesce to form single crystal platelets. Crystal growth propagates, partitioned by intravesicular bands, resulting in a layered arrangement of long guanine crystals (45–49). These large crystalline stacks act as multilayer reflectors to produce color and/or a silvery appearance in the skin (50, 51).
Pale cells are clearly a widespread and characteristic cell type in ribbontail ray skin and, we propose, at least partly responsible for this species’ blue coloration. With the exception of black and brown melanophores, we have not observed any pigmentary cells that might contribute to the blue color in ribbontail ray nor found any previous description of pale cells in elasmobranch or bony fishes in the literature; the careful preparations required to visualize pale cell contents, however, could mean these cells are more prevalent than realized, at least in elasmobranchs. Although the distribution and composition of pale cells indicates they are likely involved in a guanine-based blue structural color, the morphology and arrangement of guanine crystals in the cells of ribbontail ray is strikingly different from that of bony fishes. To the best of our knowledge, such a uniform and stable colloidal arrangement of guanine nano-crystals embedded in nano-vesicles has never been reported in fishes. Oddly, the most similar example we find in the literature is from an invertebrate, the Spanish shawl nudibranch mollusk (Flabellina iodinea), where short stacks of guanine nanocrystals, bound by membranes, form ‘punctate reflectors’ (52). Although these nanoreflectors are also localized in specialized cells near the epithelial basal membrane, as in ribbontail ray, their vesicles are less organized in their arrangement, instead packed in high densities within their specialized cells, but also scattered and mobile throughout the epithelium. The distinct morphology of ribbontail ray pale cells argues that their mechanism of guanine-based color production likely differs from that of other fishes, as well as the nudibranch example with which they are most similar; we examine possible mechanisms more deeply in the next section.
In addition to our dehydration-rehydration experiments above, the response of our specimens to fixation support our assertion that pale cells are involved in producing a structural blue in ribbontail ray. In our study, blue spots from adult specimens fixed and stored for ~ 4 years in a buffer solution lost their blue color with time. Such color loss is common in fixed specimens with structural color, attributed to architectural or refractive index changes that occur as a result of fixation (53–56). In our ribbontail ray samples, ultrastructural characterisation showed that pale cell nano-vesicles in blue regions were less densely arranged than in more freshly-fixed specimens and intravesicular guanine crystals were rare. In contrast, tissue from samples frozen for the same amount of time remained blue (albeit comparatively dark), with SEM of isolated skin verifying the presence of guanine crystals. Furthermore, the freshly-fixed spots from the juvenile specimen remained electric blue for several weeks after fixation with glutaraldehyde and therefore we suppose their ultrastructures are close to natural tissue architectures. The exact cause of color loss in long-fixed tissue remains to be identified, but these observations argue that the spacing of nano-vesicles and/or the presence of guanine nano-crystals (or the composite of crystals and vesicles, see below) are vital for color-production in ribbontail ray, a hypothesis we test below.
Coherent scattering from core-shell photonic glass ultrastructure of blue spots
In natural structural colors, the color produced (i.e., wavelengths scattered by the nanostructures) and the color saturation (i.e., the intensity of color) depend on the morphology and arrangement of the scatterers and their refractive index relative to their surrounding medium (57, 58). When nanostructures have a long-range order, as in a photonic crystal, the light waves interfere coherently to produce angle-dependent color or iridescence. In contrast, when nanostructural arrangements are completely random, incident light waves scatter incoherently (e.g., Rayleigh or Tyndall scattering) and the reflected color is non-iridescent. Accordingly, it was long believed that non-iridescent natural colors are caused only by incoherent scattering from zero-order architectures (colloidal nano-spheres in damselfly integuments: 59; spongy keratin-air matrix of feather barbs: 45, 60). This hypothesis was recently disproved by Prum et al. (61–63), who performed 2D Fast Fourier Transform analyses on transmission electron micrographs of structural colored tissues to calculate dominant frequency components in the arrangement of nanostructural arrays, revealing aspects of order and disorder otherwise imperceptible in direct observations of electron microscope images. Since electron density (gray-scale) variation in properly-stained electron micrographs is indicative of differences in refractive index (RI), the patterns observed in resultant 2D FFT power spectra can be used to distinguish between completely ordered, quasi-ordered and disordered structures, with peak widths corresponding to the spatial frequency of refractive index differences. In this way, Prum et al. (54, 56, 61) were able to confirm that the keratin-air structures in feather barbs and nano-spheres in damselfly integuments exhibit a quasi- or short-range order in spatial scales comparable to visible wavelengths to produce a coherently-scattered, but viewing angle-independent color.
We apply a similar approach to Prum et al. (54) in our examination of ribbontail ray micrographs, using 2D FFT analysis to explore the degree of order and predicted optical appearance of tissues we observed in our anatomical characterizations. Nano-vesicle arrays in pale cells showed a ring-like power spectrum (Fig. 4a), confirming they are mono-disperse, quasi-ordered and with a spatial periodicity that the Fourier power spectra prediction produces a mean reflectance peak at a wavelength of 468 nm, closely corresponding to an electric blue color (Fig. 4b). This supports our hypothesis that the spatial correlation between vesicles leads to constructive interference of blue light reflected from the nanostructural arrays in pale cells. Furthermore, the consistent FFT pattern observed from various TEM images —involving a nearly-circular single Fourier ring— indicates that the spatial arrangement of vesicles in pale cells is isotropic, resembling a ‘direct’ photonic glass ultrastructure (64–66). This explains the origin of the distinct and angle-independent color of blue spots in ribbontail ray. A direct photonic glass consists of mono-disperse colloidal structures in an isotropic disordered arrangement, but with a short-range order and higher refractive index compared to the surrounding matrix. It is this material constellation that produces the interesting combination of color characteristics in ribbontail ray, where the structural correlations from the quasi-order of vesicles with guanine crystals in the pale cell lead to coherent scattering of light, but the isotropic nature of the nanostructures results in angle-independence of the structural color (67). As a counterpoint, in a poorly-preserved adult specimen where tissues had lost their blue hue, guanine crystals were lacking, and vesicles were less densely packed, the Fourier spectra from pale cell micrographs were disk-like (Fig. S3), further illustrating that the quasi-order of guanine-containing vesicles is essential for the blue color production.
Recently, it was suggested that guanine crystals in the scales of Koi fish Cyprinus rubrofuscus and the skin of white widow spiders Latrodectus pallidus crystallize from an amorphous precursor distributed within vesicles (47, 48). If guanine crystal maturation also takes place in the same way in ribbontail ray pale cells, we suppose that the refractive index of intravesicular space and guanine depends upon the stage of crystal development. The presence of guanine crystals inside vesicles thus results in a non-monotonous variation in refractive index, with most intravesicular space (i.e., the region between guanine and outer vesicular membrane) having an intermediate refractive index when compared with their guanine ‘cores’ and the surrounding cytoplasm, and the guanine cores may act as the dominant scattering sites. As such, in matured vesicles, the guanine nanocrystals’ morphology, high refractive index, and lack of defined orientation may help to significantly enhance scattering in all directions, while the quasi-order of the vesicles provides the resonance that is necessary for angle-independent blue structural color (Fig. 4c). Also, the fact that vesicles do not appear to coalesce suggests that an unknown mechanism maintains the stability of the colloidal system, perhaps the structure and material properties of the cytoplasm and/or electrostatic repulsion between the vesicles. The mechanism determining crystal shape/growth is still unclear, but our observations of a consistent vesicle morphology in juvenile and adult individuals indicate that the stability of the vesicular core-shell ultrastructure is preserved throughout the ontogeny of the animal. Thus, it appears that the vesicles act as stable containers for the scattering guanine cores, while the dense cytoplasm controls the inter-vesicular spacing, critical for controlling the wavelengths scattered (67–69).
It is conceivable that the thick collagen fiber bundles in the lower dermis of the skin also contribute to the blue color of ribbontail ray. In some bird and mammalian skins, a thick layer of quasi-ordered collagen fibers scatter light coherently to produce structural blue colors (55, 70). Similarly, in electric rays (Torpedo ocellata, a batoid relative of ribbontail ray), it was suggested that collagen fibers are responsible for the bright blue spots on the skin (2). In the ribbontail ray, 2D FFT spectra showed that the arrangement of collagen fibers in the lower dermis is also quasi-ordered, however, the mean predicted reflectance from the Fourier power spectra showed peaks in the UV region (~ 344 nm) and in the violet region (~ 400 nm) of the visible spectrum (Fig. 4d,e). Moreover, the architecture of lower dermis collagen fiber bundles is almost identical in both blue and non-blue regions of the skin, only slightly thicker in the non-blue region, indicating that the blue color cannot originate from the collagen layers. In addition, the pervasiveness of melanosomes — in a dense layer of black melanosomes at the bottom of the upper dermis in blue skin regions, and brown melanosomes in the upper dermis in the non-blue regions should restrict most incident wavelengths (including blues) from even reaching the collagen bundles in the lower dermis, with the melanosomes also probably absorbing any light that would be back-scattered from the collagen bundles (Fig. 4c). Rather, our FFT predictions and skin reflectance measurements, argue that the extensive lower dermis collagen layer is responsible for the distinct UV reflectance we measured from all skin regions, and which proved extremely robust to different specimen storage and processing conditions (e.g. present even in spots that had been dried or had their top layer removed).
Although photonic glass ultrastructures (e.g. the quasi-order of pale cell vesicles) result in strong scattering, their comparative disorder also translates to a poor spectral selectivity or less-saturated structural color (71). In ribbontail ray, this was evidenced by the pale blue of spots when viewed on a white background. Our tissue isolation experiments and characterizations indicate that spectral selectivity is improved by adding a black material in close association, namely the ubiquitous black melanophores within the pale cell layer in the upper dermis of blue tissue regions (Fig. 4c). While the melanophores with their horizontal processes deeper in the tissue absorb wavelengths transmitted through the pale cells and scattered from the lower collagen layer, the intracellular melanosomes distributed throughout the pale cell layer should absorb incoherently- or multiply-scattered long wavelengths, enhancing the saturation of the blue color (33, 71, 72). We expect that the high aspect ratio of ellipsoidal melanosomes in black melanophores also helps to further enhance broadband absorption (73). Therefore, we argue that a combination of structural scattering from vesicles with randomly oriented guanine cores and broadband absorption from melanin pigments is crucial for the strong non-iridescent electric blue color of ribbontail ray skin. This unique nano-structural skin architecture represents an interestingly disordered arrangement for color production, with the mechanism not only demanding the evolution of the unusual pale cell type but also its intimate association with melanophores. Our demonstration that certain tissue components and structural arrangements are necessary to produce the ribbontail ray's blue color also argues that subtle structural modifications to the observed tissue nano-architectures may be the root of the natural color variation we observed in the field (Fig. 1a). The presence of pale cells, even in non-blue areas, suggests that they may have (or have had) some other tissue function, but it was their evolutionary co-optation into a structural partnership with melanophores that gave rise to the novel and ecologically important tissue color. Understanding pale cell origins and their prevalence in other species and tissues will therefore be key for understanding the evolution of structural color in elasmobranch skin, but also for designing biomimetic structural blues.