The taphonomy of colour in fossil insects and feathers

Authors

  • Maria E. McNamara

    1. Department of Geology & Geophysics, Yale University, New Haven, CT, USA
    2. UCD School of Geological Sciences, University College Dublin, Belfield, Dublin 4, Ireland
    Current affiliation:
    1. School of Earth Sciences, University of Bristol, Bristol, UK
    Search for more papers by this author

Abstract

Colouration is an important multifunctional attribute of modern animals, but its evolutionary history is poorly resolved, in part because of our limited ability to recognize and interpret fossil evidence of colour. Recent studies on structural and pigmentary colours in fossil insects and feathers have illuminated important aspects of the anatomy, taphonomy, evolution and function of colour in these fossils. An understanding of the taphonomic factors that control the preservation of colour is key to assessing the fidelity with which original colours are preserved and can constrain interpretations of the visual appearance of fossil insects and theropods. Various analytical approaches can identify anatomical and chemical evidence of colour in fossils; experimental taphonomic studies inform on how colour alters during diagenesis. Preservation of colour is controlled by a suite of factors, the most important of which relate to the diagenetic history of the host sediment, that is, maximum burial temperatures and fluid flow, and subsurface weathering. Future studies focussing on key morphological and chemical aspects of colour preservation relating to cuticular pigments in insects and keratinous structures and nonmelanin pigments in feathers, for example, will resolve outstanding questions regarding the taphonomy of colour and will enhance our ability to infer original colouration and its functions in fossil insects and theropods.

Colour is a phylogenetically important and multifunctional attribute of extant animals (Hill and McGraw 2006), with important roles in inter- and intraspecific signalling (e.g. sexual and social display (Parker 2000; Bokony et al. 2003; Seago et al. 2009), camouflage (Riley 1997), UV protection (Butler et al. 2005), thermoregulation (Riley 1997), sequestration of toxic metal ions (McGraw 2003) and as a sink for free radicals (Césarini and INSERM 1996)). Visual evidence of colour associated with fossils can be conspicuous, for example metallic colours in fossil insects (Fig. 1; Lutz 1990; Parker and McKenzie 2003; Tanaka et al. 2010; Wedmann et al. 2010) and colour patterning in fossil molluscs (Hagdorn and Sandy 1998), insects (Fig. 2; Wang et al. 2010) and some feathers (Fig. 3; Vinther et al. 2008; Li et al. 2010; Wogelius et al. 2011), and can yield taxonomic and palaeoecological information (Blodgett et al. 1983; Turek 2009). Despite their palaeobiological potential, such ‘coloured’ fossils have, until recently, received little attention. The original colouration of most fossil organisms thus remains speculative (Labandeira 2005; Li et al. 2010; Carney et al. 2012). A profusion of recent studies on the colour of ancient insects and feathers, however, has resulted in major advances in our understanding of fossil colouration and its functional evolution. Key discoveries include fossilized melanin-bearing organelles (melanosomes) in feathers (Fig. 4; Vinther et al. 2008) and structural colours in fossil insects from many localities (Figs 1, 5; McNamara et al. 2012b). Recent studies have also contributed novel methods for analysing fossil colour (Figs 6, 7; Li et al. 2010; Wogelius et al. 2011; McNamara et al. 2011, 2012a, 2012b), provided insights into the taphonomy of colour (McNamara et al. 2011, 2012a, b, 2013a, b), constrained interpretations of enigmatic anatomical features (Zhang et al. 2010) and allowed inferences on behaviour and evolution (Clarke et al. 2010; Li et al. 2010, 2012; Zhang et al. 2010; McNamara et al. 2011). Insects and feathers each have an important fossil record: insects have been important members of continental ecosystems since at least the Early Devonian (Grimaldi and Engel 2005); feathers, a key derived avian character (Bergmann et al. 2010), were diverse by the late Cretaceous and may extend to the base of the archosaur tree (Norell 2011). Evidence of original colour in fossil examples of these taxa thus has the potential to inform on the evolution of colour and its functions in important fossil groups, and the role of colour, in particular visual signalling, in ancient ecosystems. Even where evidence of colour is preserved in fossils, however, it may be the result of diagenetic alteration (Klug et al. 2009; Turek 2009; McNamara et al. 2011, 2012a, b). Interpretations of the colours and colour patterns of fossil organisms thus require an understanding of the processes leading to their preservation. The purpose of this paper is to review recent developments in the study of colour in fossil insects and feathers and summarize how they have transformed our understanding of how colour-generating mechanisms can be recognized in the fossil record, how colours are modified by diagenetic processes and how ‘coloured’ fossils can shed light on the ecology and evolution of ancient insects and theropods.

Figure 1.

Metallic colours in fossil insects. A, leaf beetle (Coleoptera: Chrysomelidae) from the middle Eocene of Eckfeld, Germany. NHMM PE1997 003a. B, weevil (Coleoptera: Curculionidae) from the middle Eocene of Messel, Germany. SMF MeI13011. C, jewel beetle (Coleoptera: Buprestidae) from Messel showing patterning on thorax and elytra. SMF MeI14586. D, partial specimen of a jewel beetle from the late Oligocene of Enspel, Germany. GDKE 1996 PE 1592. E, moth from Messel with, inset, reconstruction of original wing colours (modified from McNamara et al. 2011). SMF MeI12269. F, detail of coprolite from Messel showing well-defined moth scales. SMF MeI11808. Scale bars represent: A, 2 mm; B–E, 5 mm; F, 5 mm.

Figure 2.

Monotonal colour patterning in fossil insects. A, leaf beetle from the late Eocene of Florissant, USA, showing symmetrical spotted patterning on elytra. UCM 51311a. B, unidentified bug (Hemiptera) from the late Miocene of Oeningen, Switzerland with, inset, detail of area indicated showing submillimetre-scale spotted pattern. MCZ 14431. C, leaf beetle from the late Eocene of Florissant, Colorado, USA, showing banding on elytra and abdomen. UCM 51324. D, hindwing from the Neuropteran Limnogramma mongolicum (Kalligrammatidae) from the Jurassic of Daohuguo, China, showing prominent eyespot. NIGPAS NND11021. Scale bars represent: A, C, 2 mm; B, 5 mm; D, 10 mm.

Figure 3.

Fossil feathers showing monotonal colour banding. A, feather from the Oliogocene Creede Formation, USA, showing proximal to distal gradation in tone. UCMP 169013. B, feather from Archaeopteryx lithographica (from Carney et al. 2012) showing proximal to distal gradation in tone. MfN MB.Av.100. C, feather from the Miocene Alvord Creek Formation, Oregon, USA, showing distal to proximal gradation in tone. UCMP 391006. D, left forelimb of the troodontid Anchiornis huxleyi from the late Jurassic Taojishan Formation, China, showing variation in tone across the wing. BMNHC PH828 (from Li et al. 2010). Scale bars represent: A, 2 mm; B, 5 mm; C, D, 10 mm.

Figure 4.

Scanning electron micrographs of modern (A–B) and fossil (C–H) feathers. A, eumelanosomes within a barbule of a satin bowerbird (Ptilonorhynchus violaceus) feather. B, phaeomelanosomes within a barbule of a pigeon (Columba livia) feather. C, three-dimensionally preserved eumelanosomes associated with feathers in a grebe (Podicipedidae) from the early Miocene of Libros, Spain. MNCN-63817. D–F, feathers from the bird Confuciusornis from the Early Jurassic Jehol Group, China (modified from Zhang et al. 2010). IVPP-V13171. D, eumelanosomes preserved as external moulds. E, F, phaeomelanosomes preserved in three dimensions (E) and as external moulds (F). G, rod-shaped feather degrading bacteria on a decaying contour feather from an extant finch (Taeniopygia guttata). H, fossil feather from Paraprejica kelleri (Aves: Caprimulgiformes) from the middle Eocene of Messel, Germany, showing incompletely degraded keratinous feather matrix. SMF MeV 1635a. All scale bars represent 2 μm.

Figure 5.

Biophotonic nanostructures in fossil insects. A–C, scanning (A) and transmission (B–C) electron micrographs of multilayer reflectors in fossil beetles. A, fractured vertical section of cuticle from a jewel beetle (Coleoptera: Buprestidae) from the late Oligocene of Enspel, Germany, showing laminated exocuticle (ex), epicuticular multilayer reflector (m) and vertical pore canals (arrow) traversing exocuticle. GDKE ENS 2009 PE 5889. B, vertical section of cuticle from a leaf beetle from the middle Eocene of Messel, Germany, showing laminated exocuticle (ex) and epicuticular multilayer reflector (m) characterized by alternating layers of high- and low-electron contrast. r, resin; s, sediment; t, trabecula. SMF MeI 15553. C, detail of epicuticular multilayer reflector (m) preserved in an unidentified beetle from the early Miocene of Clarkia, USA. YPM 2010 P37b 005. D–G, scanning (D–E) and transmission (F–G) electron micrographs of multilayer reflectors from metallic moths from Messel; each of the nanostructures described below contribute to the observed hue of the fossils. D, dorsal surface of cover scale showing longitudinal ridges flanked by short microribs and connected by transverse cross-ribs. Internal laminae of the multilayer reflector are visible top left. E, dorsal surface of cover scale showing detail of cross-ribs, microribs and perforations in the scale surface. F, vertical transverse section of two cover scales showing concave geometry of the multilayer reflector in between ridges (arrows). SMF MeI 11808 and MeI 12269. Scale bars represent: A–C, E–F, 1 μm; D, 5 μm, G, 500 nm.

Figure 6.

Quadratic discriminant analysis of plumage colour based on the morphology of feather melanosomes (from Li et al. 2010). Black, brown and grey dots represent data from extant birds and numbers represent samples from the troodontid Anchiornis huxleyi from the late Jurassic Taojishan Formation, China. The geometry of melanosomes from the fossil feathers clearly lies within the range exhibited in modern birds and, in many samples, corresponds to distinct feather colours.

Figure 7.

Synchrotron rapid scanning X-ray fluorescence (SRS-XRF) mapping of the plumage of Confuciusornis sanctus from the Lower Cretaceous Jehol Group, China (from Wogelius et al. 2011). A, optical image. IVPP MGSF 315B. B, SRS-XRF map of (A) showing concentrations of Cu (red) in the plumage, Ca (blue) in the bones, and Zn (green) in the sedimentary matrix. C–E, SRS-XRF maps of the region indicated in (A) for S (C), Ca (D) and Zn (E). Scale bar represents 20 mm.

Light visible to humans is represented by a narrow region of the electromagnetic spectrum characterized by wavelengths between c. 380 and 750 nm (Farrant 1997). Certain animals, including many insects and birds, are sensitive to colours in the ultraviolet (10–400 nm) regions of the spectrum (Shi and Yokoyama 2003); rare insects (e.g. Melanophila (fire bugs; Vondran et al. 1995)) are sensitive to infrared radiation (750 nm–300 μm). In addition, many insects are sensitive to polarization (Land 1997). Colour-generating mechanisms in the cuticle of extant insects and feathers fall into two broad classes (with some overlap, see below) (Farrant 1997). Pigments are chemical compounds that are efficient absorbers of specific wavelengths of light; light of reflected wavelengths produces visible colour (Cromartie 1959). Structural colours are generated by constructive inference when light is scattered in the visible part of the electromagnetic spectrum by variations in tissue nanostructure; such biophotonic nanostructures comprise materials of alternating high- and low-refractive index and can be organized into one-, two- or three-dimensional arrays (Prum and Torres 2003). Each of these colour-generating mechanisms is discussed in detail below.

Institutional abbreviations

BMNHC, Beijing Museum of Natural History, China; GDKE, Generaldirecktion Kulturelles Erbe, Mainz, Germany; IVPP, Institute of Vertebrate Paleontology and Paleoanthropology, Beijing; MCZ, Museum of Comparative Zoology, Cambridge, MA, USA; MfN, Humboldt Museum für Naturkunde, Berlin, Germany; NHMM, Naturhistorische Museum Mainz, Germany; MNCN, Museo Nacional de Ciencias Naturales, Madrid, Spain; NIGPAS, Nanjing Institute of Geology and Paleontology, China; SMF, Forschungsinstitut und Naturmuseum Senckenberg, Frankfurt, Germany; UCM, University of Colorado Museum of Natural History, Boulder, CO, USA; UCMP, University College Berkeley Museum of Paleontology, USA; YPM, Yale Peabody Museum of Natural History, New Haven, CT, USA.

Generation of colour in insect cuticle and feathers

Pigmentary colours

The mechanism by which pigments produce colour is well understood. In most pigments, absorption of light occurs in specific chemical regions (chromophores) that comprise either conjugated π-systems (i.e. alternating single and double bonds) or metal complexes (Shawkey and Hill 2006). In each type of chromophore, empty electron orbitals are available for electron excitation (due to sharing of electrons between adjacent atoms); the energy difference between the ground- and excited state for a given electron is equal to the frequency of light that is absorbed.

Diverse pigments occur in insect cuticle and feathers. Melanins, carotenoids and flavonoids are common in both (Hill and McGraw 2006; Ghiradella 2009); pterins, ommochromes, tetrapyrroles, bilins and quinones can also occur in insect cuticle (Ghiradella 2009), and psittacofulvins and porphyrins, in feathers (Hill and McGraw 2006). Melanins are the most abundant and phylogenetically broadly distributed pigments in extant insects (Liu and Simon 2003) and feathers (McGraw 2006; Ghiradella 2009; Stoddard and Prum 2011), and they are the only class of pigments in these taxa for which there is fossil evidence. Chemically, melanins are large, inert polymers of dihydroxyindole and dihydroxyindole carboxylic acid that crosslink strongly with proteins (McGraw 2006); the precise structure is not well understood (Li et al. 2012). In addition to their contribution to inter- and intraspecific signalling, melanins fulfil essential biological functions in metal scavenging, radioprotection and photoprotection (Liu and Simon 2003; Liu et al. 2005); they also confer resistance to abrasion and immune attack, including bacterial degradation (Goldstein et al. 2004; Nappi and Christensen 2005; Gunderson et al. 2008). In most insects, melanins are disseminated throughout the epi- and exocuticle and are implicated in cuticle sclerotization (Andersen 2010). In feathers, melanins occur within discrete membrane-bound lysosome-related organelles termed melanosomes that are embedded in the feather β-keratin matrix (Marks and Seabra 2001; Fig. 4). Melanosomes are typically 470–2000 nm long and vary in morphology according to the chemical variant of melanin they contain: eumelanin occurs in elongate, rod-shaped eumelanosomes (usually 900–1100 nm long; Fig. 4A), and phaeomelanin, in oblate to spheroidal phaeomelanosomes (c. 470 nm long; Fig. 4B). Eu- and phaeomelanosomes impart black, and brown to rufous, tones in feathers, respectively; melanosomes that impart grey tones are intermediate in morphology (Li et al. 2010). Melanosomes of different morphology can occur within individual feathers (Li et al. 2010) and can co-occur with other pigments (e.g. carotenoids; Lucas and Stettenheim 1972; Hofmann et al. 2007). In Aves, melanin-based plumage pigmentation (and its absence) is evolutionarily primitive; carotenoids are considered to have several independent evolutionary origins, and other key pigments, for example psittacofulvins, unique origins (Stoddard and Prum 2011).

Structural colours

Structural colours are the brightest and most intense colours in nature (Vukusic and Sambles 2003; Parker and Townley 2007) and are widespread in extant insects and birds (Kinoshita and Yoshioka 2005). Such colours function primarily in inter- and intraspecific visual signalling (Parker 2000; Seago et al. 2009) and are usually associated with one or more striking optical effects, for example iridescence (change in hue with observation angle), opalescence and reflection that is metallic (highly directional, near-100 per cent reflectance), ultraviolet or polarized (Vukusic and Sambles 2003; Doucet and Meadows 2009). Rare matte structural colours (with probable functions in crypsis) are also known (Parker et al. 1998a; Wilts et al. 2009). Biophotonic nanostructures in extant insects comprise three main classes: (1) diffraction gratings, that is, arrays of parallel ridges or slits, which usually generate weak spectral iridescence; (2) two- or three-dimensional photonic crystals with hexagonal, cubic, diamond or gyroid lattices that generate a spectrum of visual effects from dull matte colours to bright opalescence; and (3) multilayer reflectors, that is, alternating layers of high- and low-refractive index that usually generate bright metallic iridescence (Parker et al. 2001; Vukusic 2003; Seago et al. 2009). Multilayer reflectors are the most common biophotonic nanostructure in insects, where they can occur within the epicuticle (Schultz and Rankin 1985; Kurachi et al. 2002), exocuticle (Neville and Caveney 1969) or endocuticle (Hinton 1973). Brighter colours are generated in multilayer reflectors with more layers; mirror-like (near-100 per cent) reflection is achieved where there are 10 or more high refractive index layers (Land 1972). In lepidopteran scales, colour-producing nanostructures in cover scales are typically underlain by melanin-rich basal scales; the melanin absorbs light transmitted through the biophotonic array, thus preventing incoherent scattering and enhancing the spectral purity of the observed colour (Ghiradella 1998). The role of melanin in enhancing structural colours in the cuticle of other insect groups is unclear; in some beetles, melanin may be associated with the high-index layers in multilayer reflectors (Schultz and Rankin 1985) or may occur in a discrete layer underlying multilayer reflectors (Parker et al. 1998a). The evolutionary origins of the various biophotonic nanostructures in insects have been considered only for beetles (Seago et al. 2009) and some Lepidoptera (Tilley and Eliot 2002; Wilts et al. 2009). Diffraction gratings and multilayer reflectors are considered to have multiple evolutionary origins (Seago et al. 2009); there is limited evidence that multilayer reflectors are evolutionarily primitive (Tilley and Eliot 2002; Wilts et al. 2009), although this hypothesis is controversial (Ingram and Parker 2008).

In feathers, structural colours originate via two distinct mechanisms that are each localized to specific regions of the feather. Quasi-ordered three-dimensional nanostructures (with either a channel-type or microsphere architecture) occur within the keratinous medulla of barb rami (Shawkey et al. 2003; Shawkey and Hill 2006; Noh et al. 2010). Such keratin-air nanostructures generate bright matte colours and are underlain by a thin layer of melanosomes; as in lepidopteran scales (see above), the melanin absorbs incoherently scattered light within the structure and does not serve in colour production per se (Shawkey and Hill 2006). Sheet-like arrays of melanosomes within barbules act as thin-film reflectors and generate glossy black to highly iridescent colours (Doucet et al. 2006; Yin et al. 2006; Yoshioka et al. 2007; Shawkey et al. 2011); even slight organization can produce weakly iridescent, glossy visual effects (Li et al. 2012). The precise hue produced is a function of the thickness of the thin (100–300 nm) keratin cortex that envelops the melanosome layer(s) (Prum 2006). Quasi-ordered and thin-film biophotonic nanostructures in birds are considered to have multiple independent origins (Stoddard and Prum 2011).

Fossil evidence for colour-generating mechanisms

Hand specimens and light microscopy

Fossil insects and feathers can manifest evidence for colour in hand specimen, for example metallic colours and/or tonal (monochromatic) patterning. Metallic reflection is evidence of structural colour (Parker et al. 1998a) and is known in fossil insects from many Cenozoic localities (Fig. 1; McNamara et al. 2012b and references therein). Analysis of samples of cuticle from metallic fossil insects using electron microscopy and mathematical modelling confirms that the preserved colours in these specimens are structural in origin (Parker and McKenzie 2003; Tanaka et al. 2010; McNamara et al. 2012b). Barbules in a single fossil feather from Messel (middle Eocene, Germany) exhibit a silvery metallic sheen (Vinther et al. 2010); preserved ultrastructural evidence (see below) strongly suggests that the feather is structurally coloured (Vinther et al. 2010). The striking red-brown and violet hues of the feather barbs and rami may reflect diagenetic incorporation of metals, especially Fe, into feather melanosomes (Vinther et al. 2010). Metallic hues may also form during diagenesis of other organic tissues: fossil graptolites associated with clay minerals can exhibit blue colours (U. Farrell, pers. comm. 2012), possibly due to light scattering from tectonically aligned clay minerals. Whether or not preserved metallic colours are structural in origin can be tested using electron microscopy and computer modelling (see below). Where the colour-producing array is suspected to comprise a permeable array of air plus another material, however, an expedient alternative is to immerse the fossil in media of different refractive index; this alters the average refractive index of the colour-producing array and thus the visible hue (Mason 1927). This technique has been successfully applied to metallic scales in fossil lepidopterans from Messel, in which the nanostructure in the scales comprises a chitin-air nanostructure. The technique could also be applied to metallic scales in other fossil insect taxa (e.g. weevils) and in fossil feathers where colour is generated by quasi-ordered nanostructures in the barbs.

Many fossil insects exhibit monochromatic patterning expressed as varying tones of brown (Fig. 2) but the origins of this phenomenon have not been investigated. It is conceivable that such patterning reflects original variations in cuticle thickness (Rasnitsyn and Quicke, 2002) and/or pigment concentrations, in particular, eumelanin (Vinther et al. 2008). Colour patterning can also occur in fossil insects with metallic colours (e.g. Fig. 1C), but has not been studied in detail. Chromatic variations in metallic hue across a specimen presumably reflect variations in the thickness of the multilayer reflector; patterning consisting of alternating metallic and black regions may reflect presence or absence of a multilayer reflector. Interpretations of patterning in fossil insects are most robust where specimens exhibit features that occur (and have known visual functions) in extant insects, for example symmetry systems (i.e. mirror images in paired tissues such as insect wings), disruptive markings or eyespots (Heads et al. 2005; Wang et al. 2010). Rare subfossil and Miocene insect specimens exhibit patches of dull red to yellow colouration when freshly exposed, but these colours fade rapidly upon exposure to sunlight and air; this process may reflect oxidation of partially degraded carotenoid-based pigments (Rasnitsyn and Quicke 2002).

Tonal variation is also known in fossil feathers, both within individual feathers (Fig. 3A–C; Vinther et al. 2008; Li et al. 2010; Barden et al. 2011; Wogelius et al. 2011; Carney et al. 2012) and among feathers from an individual specimen (Fig. 3D; Li et al. 2010); this patterning can relate, at least in part, to variation in melanin-based hues (Vinther et al. 2008; Li et al. 2010, 2012; Barden et al. 2011; Wogelius et al. 2011; Carney et al. 2012). Patterning within individual barbules of fossil feathers can closely resemble melanin-based patterns in modern feathers and can aid interpretation of feather microstructures (McKellar et al. 2011). Survival of biological patterning can be reasonably inferred in fossil feathers where the margins of different colour bands match isochronic sections in pigmentary patterning in modern feathers (Vinther et al. 2008).

Fossils with metallic colours and colour patterns in hand specimens are obvious targets for studies of fossil colours, but monotonal fossils may also yield evidence of colour. Indeed, many fossil feathers lack obvious variations in tone but exhibit ultrastructural or chemical evidence of colour (see below). Further, recent taphonomic experiments using extant structurally coloured insects demonstrate that ultrastructural evidence of biophotonic nanostructures can endure the effects of burial even where visible evidence of structural colour is degraded (Fig. 8; McNamara et al. 2013a; see ‘The fate of original colours’, below). Fossils that lack obvious colour thus have the potential to provide key insights into the evolution of structural colour and behaviour.

Figure 8.

Maturation experiments using the jewel beetle Chrysochroa raja (Coleoptera: Buprestidae; modified from McNamara et al. 2013a). A, untreated specimen. B–E, light micrographs of cuticle from untreated (B) and experimentally decayed (C) specimens, and specimens experimentally matured at 200°C, 500 bar (D) and 270°C, 500 bar (E). F–G, transmission electron micrographs of untreated (F) cuticles and cuticles matured at 270°C, 500 bar (G). H, reflectance spectra for cuticles in (A), (D) and (E). Scale bars represent: A, 10 mm; B–E, 1 mm; F–G, 1 μm.

Electron microscopy

Electron microscopic imaging of preserved anatomical detail is critical to identification of colour-producing structures in fossil feathers (Fig. 4) and insects (Fig. 5). Scanning electron microscopy (SEM) cannot distinguish between multilayer reflectors and other laminated regions of cuticle in extant and fossil insects (Fig. 5A). Transmission electron microscopy (TEM) of multilayer reflectors in extant insects, however, reveals a characteristic ‘barcode’ pattern of alternating layers (each c. 100–150 nm thick) of high- and low-electron contrast (Fig. 5B–C, F–G). The high- and low-contrast layers are assumed to represent layers of high- and low-refractive index, but their precise chemical composition and refractive index is notoriously difficult to assess (Noyes et al. 2007). Identification of a ‘barcode’ pattern in TEM images of a laminated structure in fossil insect cuticle is a strong indicator of a fossil multilayer reflector (Parker and McKenzie 2003; Tanaka et al. 2010; McNamara et al. 2011, 2012b), but unequivocal determination of this requires mathematical modelling (see below). SEM is also useful for determining the presence of other ultrastructures capable of producing or modifying colour. SEM imaging of structurally coloured scales in fossil lepidopterans from Messel revealed evidence of a fossil multilayer reflector and various colour-modifying ultrastructures (Fig. 5D–G); the latter produce visual effects that are ecologically significant (McNamara et al. 2011). SEM has the potential to reveal preserved remains of other biophotonic nanostructures in fossil insects, for example diffraction gratings and three-dimensional photonic crystals. Diffraction gratings with a postulated anti-reflective function are known from fossil flies hosted in amber (Parker et al. 1998b; Tanaka et al. 2009) and have been reported in arthropods from the Cambrian (Parker 1998), but examples of three-dimensional photonic crystals have not yet been reported in the fossil record.

Scanning electron microscopy imaging of many fossil feathers reveals ovoid to rod-shaped microstructures that can be closely spaced and aligned parallel to the barb or barbule long axis (Fig. 4B–F; Vinther et al. 2008). Such microstructures were originally interpreted as the fossilized remains of keratinophilic decay bacteria preserved in three dimensions via rapid replacement by authigenic minerals (Wuttke 1983), but were reinterpreted recently as the decay-resistant remains of feather melanosomes (Vinther et al. 2008). Melanosomes and many bacteria (e.g. Fig. 4G) are similar in gross morphology and size (Davis and Briggs 1995; Vinther et al. 2008; Zhang et al. 2010; Knight et al. 2011); reinterpretation of the fossil microstructures as preserved melanosomes is therefore based on several arguments relating to their spatial organization and location within the feather. A melanosome interpretation is supported where the fossil microstructures exhibit most or all of the following features: (1) location within, or envelopment by, an organic matrix (presumably the degraded remains of the feather keratin), that is, the structures are internal to the feather and are not external films of decay bacteria that grew on the feather tissue during diagenesis (Zhang et al. 2010); (2) preferential location, or at least high abundance, within dark regions of fossil feathers (Vinther et al. 2008; Barden et al. 2011); (3) dense packing, forming a discrete uniform surficial layer (e.g. Vinther et al. 2010) as in the barbules of extant birds (Shawkey et al. 2006); and (4) diagnostic nanoscale organizations, for example alignment parallel to the barb long axis (Knight et al. 2011) that cannot be generated by bacteria (Vinther et al. 2010; Li et al. 2012). Despite increasing evidence for the preservation of melanosomes and melanin within theropods and other fossil groups (Glass et al. 2012; Lindgren et al. 2012), interpretation of the fossil microstructures as preserved feather melanosomes (and thus survival of the pigment melanin on geological timescales) is not universally accepted on the basis of microstructure alone (Schweitzer 2011; Glass et al. 2012). Some authors have proposed that the fossil microstructures represent melanosomes released from the skin of the animals during decay (Lingham-Soliar and Plodowski 2010) or the degraded remains of structural collagen or keratin (Lingham-Soliar 2011). Interpretations of fossil melanosomes may be more reliable where they are supported by chemical evidence of melanin survival (Schweitzer 2011; Wogelius et al. 2011; see below). Some fossil feathers retain three-dimensional details of keratinous feather structures (e.g. Fig. 4H), which are obvious targets for the recovery of nonmelanin feather pigments and biophotonic nanostructures.

Modelling and statistics

Identification of biophotonic nanostructures in fossil insects is contingent upon optical modelling of nanostructures within the cuticle. In extant insects, coherent scattering can be analysed using various techniques, in particular, the matrix method of Macleod (1969) and applications of the discrete 2D Fourier transform (Prum and Torres 2003). The matrix method is a powerful technique that calculates the optical properties of laminar nanostructures; average values for the thickness of each layer (measured from TEM images), and estimates of each layer's refractive index, are used to generate a characteristic matrix for each lamina (Macleod 1969). Matrices are then analysed using purpose-built software, for example TFCalc (Software Spectra, Inc., Portland, OR, USA). This approach is routinely used in thin-film optics and has been applied to beetles from Messel (Parker and McKenzie 2003) and from the Pleistocene of Japan (Tanaka et al. 2010). Predicted reflectance spectra generated using this technique are compared with observed reflectance spectra measured from the surface of the fossil to assess whether the putative biophotonic nanostructure in the fossil can produce the observed hue.

The 2D Fourier transform uses direct observations of spatial variation in refractive index rather than idealized or average values and can be applied to all types of biophotonic nanostructure, not just laminar arrays (Prum and Torres 2003); it has been used to analyse colour-producing structures in extant insects, birds and mammals (Prum and Torres 2003; Shawkey et al. 2009; Noh et al. 2010). Digital TEM micrographs of cuticle are analysed using a 2D Fourier tool that is freely available (http://www.yale.edu/eeb/prum/fourier.htm) and implemented in the matrix algebra program MATLAB. The tool uses the distribution of lighter and darker areas in the TEM image (and therefore the distribution of materials of different refractive index) to estimate the average refractive index of the structure. 2D fast Fourier transform analyses of spatial variation in refractive index generate radial averages of Fourier power spectra (useful for assessing whether a particular structure can produce visible wavelengths) and predicted reflectance spectra. This technique has been applied successfully to cuticular nanostructures in fossil insects from several localities (McNamara et al. 2011, 2012b).

Studies of the colouration of fossil feathers have used statistical analysis of the morphology of fossil melanosomes to predict colour within certain confidence intervals (Clarke et al. 2010; Li et al. 2010, 2012; Carney et al. 2012). These analyses were based on a data set of melanosomes from a phylogenetically diverse sample of extant bird feathers with melanin pigmentation (Li et al. 2010, 2012; Clarke et al. 2010). The data set included parameters such as long-axis, short-axis, long- and short-axis skew, long- and short-axis variation, aspect ratio and ‘density’ (i.e. number of melanosomes per unit area). Data on these aspects of the morphology and packing of fossil melanosomes were compared with the data set of modern samples using quadratic discriminant analysis, a statistical technique that estimates the probability with which unknown samples can be classified correctly using data on known samples (Li et al. 2012). In each study, forward stepwise analysis was used to determine which melanosome parameters contributed significantly to the analysis. Different combinations of parameters were significant in different studies (aspect ratio and density in the troodontid paravian Anchiornis (Fig. 6; Li et al. 2010); long-axis variation, short-axis skew, aspect ratio and density in the fossil penguin Inkayacu (Clarke et al. 2010); and aspect ratio, long-axis, short-axis, long- and short-axis variation, and aspect ratio skew in the paravian Microraptor (Li et al. 2012)), but the importance of these differences is unclear. The data from modern feathers treated melanosomes from barb rami and barbules separately (presumably to account for known intrafeather variation in melanosome geometry; Prum 2006), but not all analyses of fossil feathers made this distinction (e.g. Li et al. 2010). The results of the statistical analyses were used to predict the colours of fossil feathers from different plumage regions with probabilities ranging from 56 to 100 per cent.

Chemistry

The chemistry of structurally coloured fossil insects is incompletely resolved. Electron dispersive spectroscopy (EDS) of structurally coloured cuticle in specimens from Messel, Enspel (late Oligocene, Germany), Eckfeld (middle Eocene, Germany), Clarkia (middle Miocene, USA) and Randecker Maar (early Miocene, Germany) confirm that the cuticle is invariably organically preserved (McNamara et al. 2011, 2012b). The extent to which original biochemical components of the lipid-rich cuticle are preserved could be determined using techniques such as pyrolysis–gas chromatography–mass spectrometry (py-GC-MS) and nuclear magnetic resonance (NMR), which inform on macromolecular complexes and proteinaceous moieties, respectively. In contrast, several recent studies have investigated the chemistry of fossil feathers (most of which are organically preserved; Davis and Briggs 1995) using techniques that provide insights into their elemental composition (EDS), functional groups (Fourier transform infrared spectroscopy (FTIR)), involatile macromolecular complexes (py-GC-MS), organic free radicals (electron paramagnetic resonance (EPR)), spatial distribution of elements (synchrotron rapid scanning X-ray fluorescence (SRS-XRF)) or local structure of specific metallic elements (extended X-ray absorption fine structure (EXAFS) and X-ray absorption near-edge structure (XANES)) (Barden et al. 2011; Wogelius et al. 2011; Carney et al. 2012). Combination of several such techniques is a powerful approach to investigating the structural properties of organic constituents, especially melanin and its degradation products, within fossils (Glass et al. 2012), and can test interpretations of melanin fossilization based on morphological evidence for preserved melanosomes (Schweitzer 2011; Wogelius et al. 2011). EPR is a useful technique for studies of eumelanin-based colouration in fossils as eumelanin possess a unique free radical signature (Glass et al. 2012). However, despite successful identification of eumelanin and its derivatives in the ink sac of fossil squid using this technique (Glass et al. 2012), application of EPR to fossil feathers has met with only limited success; analyses have demonstrated different free radical signatures in fossil feathers and the surrounding sedimentary matrix, but have not yielded diagnostic spectra for melanin (Barden et al. 2011).

Recent SRS-XRF analyses of fossil feathers indicate that certain trace elements (especially organic-bound Cu) in fossil tissues rich in spheroidal and rod-shaped microstructures have melanin affinities and may act as biomarkers for melanin-derived compounds in fossils (Fig. 7; Wogelius et al. 2011). This technique allows rapid, nondestructive chemical mapping of entire fossil specimens at concentrations in the ppm range (Wogelius et al. 2011). Some authors consider the resulting chemical data superior to morphological data in studies of melanin-based colour mechanisms in fossils as the effects of diagenesis on the geometry of melanosomes is poorly resolved (Norell 2011; Wogelius et al. 2011; but see McNamara et al. 2013b, and ‘The fate of original colours’, below). Other analytical techniques that could be applied to studies of fossil melanin in theropods include ToF-SIMS, which allows simultaneous identification and mapping of molecules and their structures at high spatial resolution (Lindgren et al. 2012). Recent studies of fossil squid and fish using ToF-SIMS confirm that chemical evidence of melanin is restricted to micron-sized melanosome-like bodies in tissues where melanin was probably a major constituent in life (Glass et al. 2012; Lindgren et al. 2012).

The fate of original colours

Morphological evidence of colour

Insects

The taphonomy of colour-producing nanostructures in insects is reasonably well understood. Multilayer reflectors in Cenozoic insects have a similar preservation potential to other cuticular nanostructures (McNamara et al. 2012b). The suite of ultrastructural features preserved in fossil cuticles is therefore key to assessing whether black colours are a taphonomic artefact; the preservation of diverse cuticular ultrastructures, but not biophotonic structures, indicates that biophotonic nanostructures were originally absent. This does not necessarily imply, however, that the cuticle was black in vivo; nonmelanin pigments may have been present originally, but visual evidence thereof is not preserved.

Despite widespread preservation of multilayer reflectors in metallic fossil beetles, original hues are not preserved (McNamara et al. 2011, 2012b); observed reflectance spectra of the fossils are redshifted from spectra predicted using the preserved biophotonic nanostructure. This phenomenon was attributed to alteration of the refractive index of the cuticle; changes in periodicity can also effect colour change in multilayer reflectors (Adachi 2007), but cuticular features in the fossils lack clear evidence of volume change, for example buckling, pull-apart structures and desiccation cracks (McNamara et al. 2012b). In contrast to these results, high-pressure/high-temperature maturation experiments using extant structurally coloured beetles resulted in a blueshift in observed hue (Fig. 8; McNamara et al. 2013a). The experiments used the extant jewel beetle Chrysochroa raja, which generates metallic colour using an epicuticular multilayer reflector; specimens were matured for 24 hours using various pressure–temperature regimes (117 bar, 200°C; 250 bar, 200°C; 500 bar, 200°C; and 500 bar, 270°C). The hue of the beetles changed progressively (decreasing wavelength) with increasing pressure. This change resulted from alteration of both the refractive index and periodicity of the multilayer reflector; the dimensions of the reflector and of various other cuticular structures were altered without obvious distortion. The colour change had two discrete components: a large blueshift caused by a decrease in periodicity of the multilayer reflector, partly offset by a smaller redshift relating to considerable alteration of the chemistry of the epicuticle and, in turn, an increase in its refractive index. The redshift is identical in magnitude and direction to the discrepancy in wavelength between observed and predicted data for the fossil beetles, supporting the hypothesis that the fossil redshift results from a change in refractive index. The chemistry of structurally coloured fossil insects has yet to be investigated comprehensively (but see Parker and McKenzie 2003; Tanaka et al. 2010), but it is clear that both chemical and morphological data are critical to future attempts to reconstruct original structural colours in fossil insects.

As with fossil beetle colours, the hues produced by multilayer reflectors in fossil lepidopterans also alter during diagenesis (McNamara et al. 2011); unlike the fossil beetles, however, predicted colours for the lepidopterans are blueshifted from preserved hues. This difference could plausibly relate to differences in the chemistry of the biophotonic tissues in each taxon: in extant insects, beetle epicuticle comprises lipid and protein (i.e. chitin is absent; Neville 1975), whereas lepidopteran scales comprise predominantly chitin (Powell 2003). Regardless of the extent to which original colours have been altered, however, certain aspects of the visual ecology of structurally coloured insects may be inferred using anatomical evidence preserved in fossils. For instance, the colour-producing nanostructures in the fossil moths described above are associated with other ultrastructural features in the scales that modify the visual signal (the inherent iridescence and specular reflection of the multilayer reflector are suppressed), implying a defensive function for the colour (McNamara et al. 2011).

All fossil examples of structurally coloured insects contain multilayer reflectors. This may be a function of the relative abundance of different colour-producing mechanisms in extant insects: multilayer reflectors are the most common biophotonic nanostructure in animals (Parker 2002), including beetles (Seago et al. 2009). Alternatively, the high abundance of fossil multilayer reflectors may be taphonomic in origin. All published examples of fossil multilayer reflectors are hosted within organic-rich lacustrine sediments with significant volcaniclastic input (McNamara et al. 2012a); pore waters from such sediments would be expected to have a slightly acidic pH. Epicuticular lipids are insoluble in acidic media, and thus, the composition and chemistry of host sediments may influence the preservation potential of multilayer reflectors. Maturation experiments on 3D photonic crystals show that they have similar preservation potential to multilayer reflectors. The absence of 3D photonic crystals in the fossil record (at least in biotas of Miocene age and older) is thus considered to represent a real, evolutionary absence (McNamara et al. 2013a). Critically, maturation experiments also demonstrate that physical evidence of colour-producing nanostructures survives in insects even where visual evidence of colour is lost (McNamara et al. 2013a). Structural colour may thus have an extensive cryptic fossil record in insect specimens that lack obvious metallic colouration.

Feathers

Despite intense interest in the colour of fossil feathers, the taphonomy of colour-producing mechanisms in feathers has not been a focus of investigation. Nonetheless, it is clear that the fidelity of preservation of fossil feathers, and their melanosomes, varies considerably. Fossil melanosomes can be preserved as three-dimensional bodies (Fig. 4C, E) or as external moulds embedded in amorphous organic material or diagenetic minerals (Fig. 4D, F; Clarke et al. 2010; Li et al. 2010, 2012; Zhang et al. 2010); both preservational modes can occur within a single feather (Zhang et al. 2010). The nature of the organic matrix surrounding some mouldic melanosomes may represent degraded feather keratin (Zhang et al. 2010) or melanin (Clarke et al. 2010; Li et al. 2010). Further, dark visual tones in fossil feathers can (Li et al. 2010), but do not always (Li et al. 2012), correspond to a high abundance of melanosomes and do not correlate with the mode of melanosome preservation (Li et al. 2010). Dark tones (i.e. a high absorbance of visible light) commonly originate in organometal- or conjugated bonds in modern pigments (Farrant 1997). The precise chemical structure, and taphonomy, of the chromophore in fossil feathers is, as yet, uncertain.

In structurally coloured fossil feathers, reconstructions of original hue are precluded by degradation of the keratin cortex that envelops the melanosomes in vivo; this cortex is responsible for the exact hue produced by the highly ordered melanosome array (Vinther et al. 2010; Li et al. 2012). In other fossil feather examples, however (i.e. those lacking structural colour), accurate predictions of precise hue are contingent upon the geometry of melansomes (Li et al. 2010). Reconstructions of original plumage colouration in fossil theropods have assumed that the original geometry of melanosomes is preserved in the fossils (Clarke et al. 2010; Li et al. 2010, 2012; Knight et al. 2011). Fossil melanosomes, however, vary in the mode of preservation: melanosomes preserved as moulds and three-dimensional bodies from the same feather region differ in size (Clarke et al. 2010) and yield differing colour predictions (Li et al. 2010, 2012). Maturation experiments on feathers from extant birds reveal that melanosome geometry is altered by the effects of elevated pressure and temperature (McNamara et al. 2013b). These experiments used melanosome-bearing feathers from 12 extant taxa; feathers encompassed diverse hues and melanosome types (eu- and phaeomelanosomes, solid and hollow melanosomes). The experiments used two different pressure–temperature regimes (200°C, 250 bar; 250 bar, 250°C) and lasted 24 hours; melanosomes in all feathers altered progressively in geometry (both long and short axes reduced in length) between the 200°C, 250 bar experiment and that using 250°C, 250 bar. Survival of original melanosome geometries in fossils is thus most likely where the host sediments experienced limited burial. Not all feather-containing fossil deposits, however, meet this criterion (McNamara et al. 2013b). Some studies of melanin-based colouration in fossil feathers have considered that contributions by other pigmentary and structural colouration mechanisms to the visible hue in vivo would have been masked by melanin (Carney et al. 2012; Li et al. 2012), but this is likely only in feather regions with very abundant melanosomes. Feathers in many extant birds contain melanosomes but the visible hue derives from nonmelanin pigments or biophotonic architectures (McNamara et al. 2013b and references therein). Maturation experiments on such feathers reveal that melanosomes are retained in degraded feathers even where visual evidence of all other colouration mechanisms has degraded completely. Given this preferential preservation of melanosomes, attempts to reconstruct colour in fossil feathers should be integrated with anatomical and geochemical data on the preservation of other pigments and biophotonic structures.

Chemistry

Traces of original cuticular biomolecules, for example chitin and amino acids, are preserved in subfossil beetles with epicuticular multilayer reflectors (Tanaka et al. 2010). Less is known about the chemical fidelity with which older structurally coloured insects are preserved. Some authors have suggested that original organic material has survived in metallic beetles from the middle Eocene of Messel, but this hypothesis is supported only by bulk elemental analysis (Parker and McKenzie 2003). Where fossil cuticles contain an epicuticular multilayer reflector, lipid extracts may be a useful proxy for the chemistry of the epicuticle and thus of the colour-producing structure. Lipid extracts of thermally matured cuticle from extant beetles analysed via py-GC-MS are dominated by lipid–protein complexes. These complexes are absent in fresh cuticle and represent reaction products of functionalized epicuticular lipids with proteinaceous moieties from the epi- or exocuticle (McNamara et al. 2013a). The composition of structurally coloured fossil cuticles has yet to be investigated using techniques that inform on the structure of preserved components.

Unequivocal traces of melanin have been reported in fossil squid (Glass et al. 2012) and fish (Lindgren et al. 2012) but have not been identified in fossil feathers (Barden et al. 2011). Recent synchrotron-aided analyses using X-ray fluorescence (XRF) show that distribution maps of certain trace elements with a melanin affinity, for example Cu, in fossil feathers may help reconstruct colour patterns in fossil plumage (Wogelius et al. 2011). EXAFS and XANES analyses demonstrate that Cu is present in fossils in organometallic form, possibly derived from original melanin (Wogelius et al. 2011). The spatial distribution of trace elements in fossil specimens, however, may also result (at least in part) from taphonomic modification of original colouration signals. Trace metal concentrations could increase in tissues as a result of adsorption by microorganisms during decay (Hitchcock et al. 2009) or chelation during later diagenesis (Shock and Koretsky 1993). Studies of melanin in fossil fish suggest that it may be concentrated in the dark regions of fossils during diagenesis due to preferential degradation of more labile organic molecules (Lindgren et al. 2012). This process does not, however, preclude diagenetic migration of melanin from source tissues. Diagnostic biomarkers for melanin were recovered from sediment adjacent to the ink sacs of fossil squid (Glass et al. 2012), although this may reflect minor leakage of ink and/or intact melanosomes from the ink sac during decay rather than later diagenesis. Trace elements within fossil feathers could also derive from endogenous sources other than melanin, for example keratin–Cu complexes in feathers (Wogelius et al. 2011)), or from external, that is, sedimentary, sources. Preservation of intact chemical moieties of the melanin molecule may result from thermally induced polymerization reactions during diagenesis (Glass et al. 2012). In situ polymerization may also explain, at least in part, the aliphatic composition of some fossil feathers (Barden et al. 2011).

Controls on the preservation of colour

Patterns in the fossil record of structural colour in insects relate to a hierarchy of taphonomic controls, including decay, burial pressure, burial temperature, the nature of diagenetic fluids, weathering and the mode of curation of fossil specimens (McNamara et al. 2012a). These factors are also likely to govern the preservation of colour in fossil feathers, although the relative importance of different taphonomic factors in preserving feather and insect colour may differ. Statistical analyses of the fidelity of the preservation of structural colours in various insect taxa from Cenozoic biotas revealed that age and taxonomic compositions do not control taphonomy; age may, however, be more important in older material as it serves as a proxy for more extensive or complex diagenesis.

Decay

Laboratory decay experiments on extant structurally coloured beetles at room temperatures and pressures did not induce colour change (Fig. 8; McNamara et al. 2013a). Some fossil insect cuticles in which metallic colours are poorly preserved or absent, however, exhibit evidence for decay by fossil or modern microbes (McNamara et al. 2012a); the latter have a particularly negative impact on the fidelity of preservation. Decay may also affect colours in fossil feathers, especially where these are structurally coloured: the keratin cortex and medulla (which house various colour-producing nanostructures) are less resistant to decay than melanin (Goldstein et al. 2004). The high preservation potential of melanin (Hollingworth and Barker 1991) does not, however, imply that it (or melanosomes) is immune to alteration during decay. Laboratory decay experiments on extant birds showed that decay-induced rupture of feathers can liberate melanosomes from the keratin matrix, obliterating original packing arrangements (McNamara unpub. data).

Burial depth

The maximum depth to which fossil insects and feathers are buried during diagenesis is an important control on the preservation of colour. It determines the maximum pressures and temperatures to which most fossils are exposed and influences the fidelity of preservation of anatomical features (McNamara et al. 2012a, 2013a); it may also influence biomolecular preservation (Höss et al. 1996). Burial depth contributes to broad-scale inter-biota patterns in the fossil record of structural colour in insects. Structurally coloured fossils can be abundant in biotas that experienced limited burial, for example Messel, Enspel, Eckfeld, but absent in biotas buried to greater depths, for example Green River (McNamara et al. 2012a). Maturation experiments confirm that increasing pressure alters insect structural colours, but the effect of pressure is secondary to that of temperature, which is the primary agent of colour change during burial (McNamara et al. 2013a). Structural colours alter progressively with increasing temperature but are lost beyond a temperature threshold; the value of this threshold temperature is likely to vary with the chemistry of the tissue and host sediment. Feathers experimentally treated to elevated pressures and temperatures lose visual and ultrastructural evidence of all colour-producing mechanisms save for melanosomes; progressive loss of anatomical detail occurs in tandem with loss of visual colour with increasing temperature (McNamara et al. 2013b). As with insects, temperature may be the primary determinant of colour (i.e. melanosome) survival in fossils.

Diagenetic fluids and weathering

Other factors that influence biota-scale patterns in the fossil record of colour (at least for structurally coloured fossil insects) include the nature of diagenetic fluid flow and the extent of weathering (McNamara et al. 2012a). Reactive hydrothermal fluids can degrade insect cuticle and lead to precipitation of authigenic minerals within colour-producing nanostructures, resulting in loss of visible colour (McNamara et al. 2012a). Extensive oxidative degradation of cuticles during weathering may also destroy structural colours. The absence of structurally coloured insects in the late Eocene of Florissant (USA), and their high abundance in other Cenozoic biotas (i.e. Messel, Clarkia (early Miocene, USA), Enspel and Eckfeld), has been attributed to variations in the extent of Recent weathering among these biotas (McNamara et al. 2012a). Host sediments at Florissant are exposed subaerially, but those from the other localities are located beneath the water table; sediments from Messel comprise approximately 40 per cent water (Schaal and Ziegler 1992). Variations in the extent of weathering can also explain patterns in the fidelity of preservation of colour within an individual biota. Specimens from Eckfeld (middle Eocene, Germany) show a statistically significant correlation between poor preservation of structural colours and preservation in extensively weathered host sediment (McNamara et al. 2012a).

Deep burial, exposure to diagenetic fluids, and subaerial or subsurface weathering may also be responsible for certain taphonomic features in fossil feathers. Melanosomes are frequently preserved as external moulds, yet the origin of this phenomenon is unclear. Melanin is resistant to microbial and chemical attack (Goldstein et al. 2004), but heat treatment can induce chemical changes in its molecular structure, rendering the molecule soluble in strongly oxidizing fluids and various acids and bases (Fox 1976). Degradation of three-dimensional fossil melanosomes could therefore result from the effects of elevated temperatures and reactive pore fluids during diagenesis, or from oxidative weathering during exhumation and exposure. Indeed, the last of these factors is considered to be an important agent of colour loss in other pigmented fossils (Hagdorn and Sandy 1998).

Mode of curation

The mode of curation of structurally coloured fossils can affect also the long-term stability of the colour-producing mechanism and the resulting hue after collection. Metallic colours in insect specimens can degrade following dehydration in air (Parker and McKenzie 2003; Schweizer et al. 2006). Loss of colour via dehydration and microbial degradation (Toporski et al. 2002) is usually, but not always (McNamara et al. 2012a), prevented by storing such specimens in liquid media, for example brine, ethanol or glycerine. Indeed, loss of nanostructural definition and visible colour can occur after only several months’ storage in brine (McNamara et al. 2012a).

Functional and evolutionary significance of fossil colours

Identification of evidence of colour (and accurate reconstructions of original colours) in fossils can inform on the function of colour, especially visual signalling mechanisms. Iridescent metallic colours in modern insects can be cryptic in foliage but conspicuous in direct sunlight and thus may function in both camouflage and sexual selection, especially in environments characterized by uneven dappled light, for example forest (Doucet and Meadows 2009; Seago et al. 2009). Metallic colours in fossil insects may have had similar functions, particularly in specimens from palaeolakes surrounded by forest (e.g. Messel (Schaal and Ziegler 1992), Clarkia (Smiley 1985)) and in specimens with original green hues. Many structurally coloured fossil insects exhibit blue hues; given that structural colours are blueshifted during fossilization (McNamara et al. 2013a), at least some of these blue fossil cuticles are likely to have been originally green. Some fossil insects preserve anatomical evidence for modification of iridescence to enhance defensive signals, that is, camouflage and aposematism (Fig. 1E; McNamara et al. 2011); cryptic functions have been inferred from monotonal colour patterns (Wang et al. 2010).

Striking colour patterns and glossy iridescence in various feathered dinosaurs suggest that sexual display or defence was important in the early evolution of plumage and feather colour (Li et al. 2010, 2012). This is supported by variations in colour within individual pennaceous fossil feathers in Anchiornis, a basal paravian, indicating that melanin-based intrafeather patterns evolved before powered flight (Li et al. 2010). Some fossil feathers exhibit evidence for the involvement of melanin in nonvisual roles. Several authors have observed trends in the intensity of dark tones within individual fossil feathers and related these to the degree of melanization and hence feather function. As in modern feathers (Lucas and Stettenheim 1972; Hill and McGraw 2006), visual tone can be pale in down feathers (McKellar et al. 2011) and in basal regions of contour feathers (Li et al. 2010) and darkest at the distal tip of contour feathers (Fig. 3B; Clarke et al. 2010; Carney et al. 2012) and in distal barbules (which overlap their proximal counterparts). Such increased melanization has been suggested to confer increased mechanical strength (Li et al. 2010; Carney et al. 2012). Other fossil feathers exhibit dark tones in proximal regions (e.g. Fig. 3C; Wogelius et al. 2011, fig. 3A). Some authors have suggested that specific melanosome geometries and configurations may affect the material properties of fossil feathers (Clarke et al. 2010), but these hypotheses are largely untested in modern material.

Identification of colour-producing structures in some fossil theropods has significant implications for our understanding of the evolution of feathers. Discoveries of melanosomes in integumentary filaments of the tail of Sinosauropteryx (Zhang et al. 2010) refute claims that the filaments are partially decayed dermal collagen fibres (Lingham-Soliar 2003; Feduccia et al. 2005; Lingham-Soliar et al. 2007) and support interpretations of the filaments as feather homologues, that is, precursors of true feathers. Although the function of colouration in such structures is still unclear, the filaments themselves are considered to occur in sufficient densities to have had important nonvisual functions in thermoregulation and protection (McKellar et al. 2011).

Future directions

Certain aspects of the taphonomy of colour in insects and feathers are poorly understood, not least that of pigmentary colours in fossil insects. Some common insect pigments (e.g. carotenoids and pterins) are also abundant in plants; the degraded remains of plant-derived pigments are often preserved in sedimentary organic matter and are the basis of many studies of lake productivity (e.g. Romero-Viana et al. 2009). It is therefore reasonable to hypothesize that evidence of some insect pigments may survive in the fossil record. Resolving the taphonomy of insect pigments will require extensive chemical analysis of fossil insect remains and taphonomic experiments. By providing insights into the relative preservation potential of different insect pigments, and into the nature of any diagnostic biomarkers for such pigments when they decay, such experimental and chemical studies are likely to prove critical to our understanding of the origins of monotonal patterning in fossil insects and will test previous suggestions (Vinther et al. 2008) that such patterning reflects the distribution of eumelanin.

Loss of structural colour via oxidative weathering has been invoked to explain inter- and intrabiota patterns in the presence or absence of metallic colours, and even within individual specimens, but the chemical processes involved are uncertain. Chemical analysis of structurally coloured cuticles at various stages of oxidative degradation could inform on this phenomenon and could also help to explain patterns in the fidelity of preservation of pigments in fossil insects and feathers.

The chemistry of melanin in fossil feathers is not fully understood. Chemical techniques such as EPR and ToF-SIMS have proved critical to our understanding of the taphonomy of melanin in fossil squid (Glass et al. 2012) and fish (Lindgren et al. 2012), but have not been applied to fossil feathers. Doing so will test whether taxonomic or tissue-related factors influence the chemical fidelity of preservation of melanin in fossils. SRS-XRF analyses have yielded intriguing insights into the elemental composition of fossil fish, squid and, in particular, feathers (Wogelius et al. 2011). The ability of this technique to resolve questions regarding the preservation of melanin in fossil feathers could be tested by comparing data on the distribution and abundance of various trace elements for specimens of various taxa (including those that do not possess melanin) from a single biota.

Other unresolved aspects of the taphonomy of colour in feathers are also amenable to experimental testing. Some authors have surmised that eumelanosomes and phaeomelanosomes may differ in their resistance to diagenetic alteration (Clarke et al. 2010). The relative preservation potential of the different melanosome types could be investigated via conventional decay- and maturation experiments; decay experiments could also assess whether autolytic and microbial decay processes affect melanosome geometry. Morphological and chemical analyses of degraded feather tissues could inform on the taphonomic processes that influence the preservation of melanosomes as external moulds; such studies could also help resolve the nature of melanin (and its derivatives), and the chromophore responsible for visual dark colouration, in feathers preserving mouldic melanosomes. Future studies of melanosome taphonomy should also consider factors that may contribute to alteration of melanosome geometry and packing arrangement in fossils. Additional maturation experiments will constrain the range of pressure–temperature conditions capable of inducing morphological change and may allow the extent of diagenetic alteration of melanosomes to be predicted.

The keratinous feather matrix is the basis of several colour-producing nanostructures, yet little is known about its physical taphonomy. Recent maturation experiments using structurally coloured feathers revealed that decay of quasi-ordered nanostructures generates diagnostic ultrastructures (McNamara et al. 2013b). Such textures could serve as a proxy for the former presence of quasi-photonic nanostructures in fossils. The chemical taphonomy of keratin has been studied in detail in various fossil reptiles (Manning et al. 2009; Edwards et al. 2011) and, to a lesser extent, in fossil birds. Immunological evidence of keratin has been reported in feathers and claw sheaths from Cretaceous theropods (Schweitzer et al. 1999a, 1999b). Amide peaks in FTIR spectra of fossil feathers may derive, in part, from feather β-keratin (Barden et al. 2011). Keratin-chelated Cu released during decay may bind to melanin during later diagenesis, enhancing Cu concentrations in fossil feathers (Wogelius et al. 2011). Sulphur associated with fossil feathers may derive from feather keratin (Wogelius et al. 2011). Future chemical analyses of fossil feathers will inform on the taphonomy of the keratin molecule and will constrain the extent to which it is linked to the taphonomy of feather melanin and melanosomes.

Conclusions

Understanding the taphonomic processes that influence the preservation of pigmentary and structural colour is of fundamental importance to palaeobiologists interested in the evolutionary history of colour and its functions. Fossils with obvious colour and colour patterns, and those lacking visible colour, have the potential to retain physical and chemical evidence of colour-generating mechanisms. Preservation of colour is controlled by a suite of taphonomic factors that are, at present, not fully resolved; key factors identified to date include the depth of burial and the extent of hydrothermal alteration and weathering. Additional taphonomic data from fossils and from controlled laboratory experiments are critical in unravelling the taphonomic history of colour in different taxa, depositional environments and diagenetic regimes. This will inform interpretations of original colour in fossils and of the role of colour in visual signalling through time.

Acknowledgements

My research on the taphonomy of colour in fossils was funded by an Irish Research Council – Marie Curie International Mobility Fellowship. I thank Derek Briggs and Patrick Orr for their advice, collaboration and many fruitful discussions of taphonomy. I also thank Mike Benton, Hui Cao, Daniel Field, Neal Gupta, Stuart Kearns, Emma Locatelli, Laura Meyer, Heeso Noh, Lin Qiu, Sonja Wedmann and Hong Yang for their significant contributions to published and ongoing collaborative studies. I am grateful to Günter Bechly, Susan Butts, Thomas Engel, Larry Gall, David Grimaldi, Kristof Kristoffersen, Herbert Lutz, Leonard Munsterman, Markus Poschmann, Michael Rasser, Greg Watkins-Colwell and Michael Wuttke for access to fossil and extant specimens, and to Daniel Field, Zhenting Jiang, Robert Patalano, Barry Piekos, Vinod Saranathan, Elizabeth Savrann, Jess Utrup, Robert Young and Shuang Zhang for assistance with laboratory techniques. Thanks also to Antónia Monteiro, Paul Nash, Jeffrey Oliver, Rick Prum, Jakob Vinther and the members of the G&G Paleo group and the staff of the Peabody Museum Entomology Division for their insights into taphonomy, evolution and the colours of extant organisms.

Ancillary