Expanding the eggshell colour gamut: uroerythrin and bilirubin from tinamou (Tinamidae) eggshells

To date, only two pigments have been identified in avian eggshells: rusty-brown protoporphyrin IX and blue-green biliverdin IXα. Most avian eggshell colours can be produced by a mixture of these two tetrapyrrolic pigments. However, tinamou (Tinamidae) eggshells display colours not easily rationalised by combination of these two pigments alone, suggesting the presence of other pigments. Here, through extraction, derivatization, spectroscopy, chromatography, and mass spectrometry, we identify two novel eggshell pigments: yellow–brown tetrapyrrolic bilirubin from the guacamole-green eggshells of Eudromia elegans, and red–orange tripyrrolic uroerythrin from the purplish-brown eggshells of Nothura maculosa. Both pigments are known porphyrin catabolites and are found in the eggshells in conjunction with biliverdin IXα. A colour mixing model using the new pigments and biliverdin reproduces the respective eggshell colours. These discoveries expand our understanding of how eggshell colour diversity is achieved. We suggest that the ability of these pigments to photo-degrade may have an adaptive value for the tinamous.


H 2 SO 4 /MeOH-based Pigment Extraction Protocol with Concomitant Esterification:
A slightly modified version of the classic eggshell pigments extraction protocol using methanolic H 2 SO 4 solution was used: 1-3 Eggshells (100-1000 mg) were weighted and ground into a fine powder in a porcelain mortar. The powder was placed in a 25 ml round-bottom flask equipped with a magnetic stir bar and reflux condenser. To this was added a methanolic H 2 SO 4 solution (1.8 M, 5%; 5 mL). The mixture was heated on a heating mantle to a gentle reflux for 45 min.
The pH of the cooled solution was adjusted to between 6-8 by addition of sat. aq. NaHCO 3 (~3.5 mL) until the effervescence stopped. The suspension was then filtered through a small (8 cm diameter) Büchner funnel and the filter cake was washed with ethyl acetate (EtOAc) (~10 mL).
The filtrate was transferred to a 50 mL separatory funnel containing water (10 mL). The aqueous layer was extracted with EtOAc (2 × 5 mL). The combined organic layers were dried over anhydrous Na 2 CO 3 (~ 1 g), the drying agent was removed by gravity filtration, and the clear but coloured solution was concentrated by rotary evaporation. The volume of the organic fraction was determined at this point for the determination of the optical pigment concentration.
The E. elegans extract was passed through a short silica gel column (0.5 × 3 cm). The nonpolar main blue-green band was collected (EtOAc) and the eluent changed (to 4:1 EtOAc:MeOH) to elute a yellow-brown fraction. The fractions were evaporated to dryness under a stream of N 2 and stored at -20˚C.
We verified that the dye extraction and esterification protocol did not trigger the degradation of protoporphyrin 1 H or biliverdin 2 H , thereby generating artefacts by applying independently sourced protoporphyrin 1 H or biliverdin 2 H , dissolved in MeOH, onto commercial (non-waxed) white hen eggshells. We then submitted the dyed eggshells to the same extraction protocol as the tinamou eggshells described above and verified the presence of the pigments, as their dimethyl esters, 1 Me or 2 Me , and the absence of any notable degradation products by HPLC.
The extinction coefficients for 3 H or 3 Me have not been reported but can reasonably be assumed to be essentially identical since the esterification does not affect the chromophore; furthermore, S5 we assumed the extinction coefficients for 3 H and 3 MeOMe at 271 nm (the wavelength least variant upon chromophore etherification) to be identical. Solvatochromic effects were ignored.

Colour Mixing Modelling
To determine what combination of biliverdin and novel pigment (bilirubin for E. elegans and uroerythrin for N. maculosa) could produce these novel colours, we simulated the variable admixture of the two main eggshell pigments for each species using measured transmission spectra (see above), and compared these predicted spectra to measured eggshells reflectance.
We measured the spectral reflectance of preserved E. elegans (N = 1) and N. maculosa (N = 1) eggshells, and also fresh gray catbird Dumetella carolinensis (N = 1) eggshells, using a field portable spectrometer (Jaz, Ocean Optics) relative to a diffuse white standard (WS-1, Ocean Optics). Each eggshell was measured six times: twice at the blunt pole, equator, and sharp pole. For each species, we normalized the transmission spectra of all of their extracted pigments by dividing by the maximum value for these pigments (e.g., bilirubin had a transmission of 106.85% for E. elegans, while an unidentified pigment had a transmission of 98.44% in N. maculosa). The solvents shifted the peak transmission for biliverdin relative to the peak reflectance of the pure blue-green eggshell of D. carolinensis (peak at 501 nm) measured in the field by 14 nm and 21 nm (E. elegans and N. maculosa, respectively); therefore, we shifted each spectrum so that the peak transmission of biliverdin matched the field measured peak reflectance. We then mixed the two major pigments, using a simple subtractive colour mixing model: where R represents reflectance and c represents the relative concentration of each colourant across all wavelengths. In these calculation we varied the concentration of each colourant to generate a range of transmission spectra representing the colours of variable mixtures (N = 1001) of the two pigments (Fig 4a, b). Because the transformation resulted in missing values in the short wavelength range (< 314 nm and 321 nm respectively), we extrapolated these data using a natural spline. This resulted in reflectance values ranging from 300 to 700 nm.
Measured and predicted spectra were then converted to coordinates within the CIE colour space (Fig 4c, d) 12 . To determine the admixture ratio (novel pigment: biliverdin) we then calculated the Euclidean distance between the coordinates within this colour space for each simulated intermediate spectrum and each measured reflectance spectrum for each species (N = 6). The admixture that resulted in the smallest Euclidean distance against any measured egg spectrum was recorded. The coordinates of these 'best-matched' predictions were used to S23 determine the likely variable contribution of these two pigments and are depicted on the CIE plots (on Fig. 4c and d).
We then plotted the spectral reflectance of each eggshell alongside the simulated spectra; however, since these simulated values were based on normalized data we needed to multiply measured spectral reflectance values by a constant. We used the constant that resulted in the smallest just noticeable difference (JND) between the constant-modified-reflectance and any simulated spectrum (the smaller the JND the less noticeable the colour difference). The JND values were calculated following previous research in this group, 13 using a receptor noise limited visual model 14 to compare the perceived achromatic difference between each simulated colour and the measured spectral reflectance modified by a constant (across a range of potential constants). For these visual models, we used photoreceptor sensitivities for the average ultraviolet sensitive viewer 15 and the double cone sensitivity estimates for the blue-tit (Cyanistes caeruleus), because tinamous are believed to be sensitive to ultraviolet light. 16,17 We assumed ideal (constant at 100%) illumination and transmission for these calculations.