Interpreting melanin-based coloration through deep time: a critical review

1. Introduction

The astonishing diversity of colour patterns seen in extant animals testifies to the important and multifunctional roles of melanin and other biological pigments (biochromes) in nature. Moreover, organic molecules (including eumelanin) can persist and be recognized over vast spans of geological time, and thereby evince crucial information about the history of life [1]. Preserved biochromes not only yield interpretations on the appearance of ancient organisms, but can also potentially elucidate their ecology and behaviour. They are therefore an insightful data source for resolving facets of palaeobiology and evolution.

For decades, scanning electron microscopy (SEM) and, more rarely, transmission electron microscopy (TEM) have been used to detect round to oblate microstructures (about 1–2 µm long) in exceptionally preserved vertebrate fossils (e.g. [2–5]). Until 2008, these minute bodies were interpreted as the remains of microorganisms that contribute to the decomposition of organic material (e.g. [4], but also see Voigt [6] and references therein). However, that year, Vinther et al. [7] re-described them as residual melanosomes: eukaryotic, melanin-containing cellular organelles, responsible in part for the coloration of integumentary structures, including skin, hair and feathers. They further suggested that the shape and arrangement of fossil melanosomes could be used to deduce plumage colours in ancient birds and non-avian dinosaurs [7], which led to inferences about ecology and behaviour that have since been adopted by numerous studies (e.g. [8–10]).

Although information about the colour of extinct animals might provide important clues pertaining to lifestyles and habits, the reconstruction of specific tones and hues based on morphological evidence alone is not a straightforward process [11]. This is particularly pertinent when only SEM image data are considered, as these are demonstrably inadequate for discriminating between remnant melanosomes and pervasive bacteria [12]. Differentiation of melanosomes from microorganisms is necessary because they overlap in size, shape and distribution [12–14]. Moreover, microbes are always associated with decaying carcasses, and are known to fossilize as both organic (e.g. [15,16]) and inorganic traces (e.g. [17]). Another confounding issue is that microorganisms synthesize melanins [18,19], an ability that is most prevalent in bacteria inhabiting soil and marine environments [19]. Compounds arising from the degradation of melanic pigments have also been detected in microbial fossils, including fungal hyphae [20].

Defining colour from the shape of fossil microbodies (see [8,9]) has other limitations. Extant eumelanosomes (ellipsoidal organelles associated with dusky pigmentation [21], but also see below) can contain only eumelanin, but pheomelanosomes (spherical organelles associated with reddish pigmentation [21]) may not exist without a eumelanin complement [22]. Finally, the eumelanin-dominated eye pigments of vertebrates exhibit a random distribution of melanosome shapes that do not correlate with the type of melanin present [23,24].

Here, we address conceptual and methodological issues related to the interpretation of microbodies detected in fossilized animal tissues. As a framework, we briefly review melanosome formation in vertebrates (for brevity, we do not address invertebrate melanogenesis), focusing on physical and chemical properties of one of its main constituents: melanin. This moiety is thought to impart stability to the intracellular organelle, thereby enabling its preservation through deep time. Microbial melanization is also summarized for a practical character-based discrimination between fossil melanosomes and microorganismal cells. Our critical assessment of the methods commonly employed to identify fossil microstructures is intended to facilitate confident documentation and reduce the risk of insufficiently supported claims propagating in the literature.

2. Vertebrate melanogenesis

Structurally, melanins are heterogeneous biopolymers comprising a series of conjugated indole (resonant double-ring) moieties [25,26]. Several major melanin types exist in nature, but the most common are (dark brown-black) eumelanin and (red-yellow) pheomelanin [27]. Eumelanin is derived from enzyme-controlled oxidation of the amino acid tyrosine [28] to form a biochrome that is inherently resistant to degradation [25]; thus, its molecular structure is incompletely known [22,26]. Likewise, despite recognition that pheomelanin is produced from sulfur-containing benzothiazine units [28], its chemical structure is also not fully characterized [21].

In vertebrates, melanins are distributed through epidermal tissues and their derivatives, where the colour they impart plays important roles in social and predator–prey interactions, thermoregulation and ultraviolet (UV) protection [29,30]. In addition, internal organs and tissues, such as the liver, spleen, brain and inner ear, also contain melanin, which contributes to physiological processes and disease resistance [30, fig. 1].

The cellular site of melanin synthesis, storage and transportation is the melanosome—a membrane-bound organelle generated by melanocytes, melanophores and pigment-epithelial cells [29]. Melanosomes attain a broad range of sizes and shapes, ranging from sub-micrometre-sized spherical grains [21, fig. 2d] to elongate particles up to 4 μm long [31, fig. 2a]. Ellipsoidal forms are typically ascribed to eumelanosomes, whereas spherical structures are referred to pheomelanosomes (e.g. [21]). This subdivision of shape has been proposed to reflect the chemical composition of the type of melanin they contain ([21], but also see below).

Melanosome biogenesis follows four sequential stages (enumerated I–IV), where melanin deposition is initiated at stage III, and the organelle is fully melanized by stage IV [32, fig. 1a]. Polymerization of melanin within the developing melanosome results in the formation of granules (this term is inconsistently used in the literature, but we follow the definition of Simon & Peles [22] as a standard) on intraluminal fibrils, and continues until all other structures within the organelle are obliterated [28,33]. At this time (stage IV), the organelle is considered to be mature [22]. Fully grown melanin granules normally range from 10 to 30 nm in diameter [22,34,35]; however, larger grains up to 80 nm in diameter have been recorded within red human hair melanosomes [21]. The melanin granules cause the outer surface of the melanosomes to become rugose [36, figs 2 and 3], a characteristic trait that is most marked in spherical forms [23].

Whereas pure eumelanin is naturally occurring, pure pheomelanin has not yet been documented [22]. Instead, pheomelanosomes contain a mixture of eumelanin and pheomelanin [21,28,37]. Pheomelanin granules are thought to comprise a pheomelanin core surrounded by a eumelanin sheath (the casing model [22,38], but see Gorniak et al. [39] for a different interpretation), the thickness of which determines expressed colour, at least in iridal melanosomes [37].

3. Melanin synthesis in microorganisms

In addition to eumelanin and pheomelanin, bacteria and fungi also produce a third type of melanin, generically named allomelanin [18,40]. Allomelanins are synthesized from an array of sources and via different biochemical pathways. As a result, several major subtypes exist, the most ubiquitous being pyomelanin. Allomelanins usually form from nitrogen-free precursors and are hence devoid of this constituent (see Plonka & Grabacka [18] for a comprehensive review of melanin biosynthesis in microorganisms).

In contrast to vertebrates, whose melanogenesis is confined to specialized cellular organelles, melanin synthesis in fungi usually occurs within the cell wall [41,42]. Nonetheless, some fungal species deposit melanins intracellularly in the form of cytoplasmic bodies, whereas others secrete melanic pigments into the surrounding environment [43]. The melanized fungal cell wall is highly durable, and thus can be isolated by chemical treatments destructive to other cellular components [44]. The resulting melanin shells (often called melanin ghosts) are hollow, but retain both the shape and size of the original cells after degradation and removal of the non-melanized constituents [18,44]. Melanin ghosts are composed of tightly packed and occasionally laminated melanin granules between 30 and 150 nm in diameter [43,44]. These impart a nodular appearance to the external ghost wall [44, fig. 2], and are typically accompanied by diagnostic crater-like scars derived from the budding process ([44, fig. 5], [45, fig. 1]).

In prokaryotic organisms, melanogenesis normally occurs extracellularly, and hence most bacterial melanins form amorphous deposits when purified [18,46], although globular aggregates have been reported [47]. However, in some bacteria, melanins are localized to the cytoplasm and may appear as electron-dense spots [48, fig. 5e]. Melanin-like compounds have also been detected in bacterial endospore coats, where they probably protect the developing spore against harmful UV radiation (e.g. [49]).

4. Melanin and melanosomes in the fossil record

Reports describing fossil melanins and melanosomes have been published sporadically since the 1930s (e.g. [6,50,51]), yet the search for ancient melanic pigments only began in earnest in the late 2000s following the proposal that feather traces might infer evidence of original hues and shades [7,52]. Since then, a number of investigations have used chemical markers and presumed fossil melanosomes to explore aspects of the biology and ecology of extinct animals, including colour [8,9,53–57], behaviour [8,56] and physiology [10]. However, studies reporting remnant melanosomes have been met with controversy, and an alternative hypothesis has been put forth favouring a more conservative interpretation of the fossil microbodies as microbes colonizing the degrading tissues prior to burial [12,58]. Such criticism has sparked intense debate (see Edwards et al. [11] for review), that is further aggravated by the dearth of unequivocal molecular indicators for ancient melanic pigments, which thus far have only been recorded from cephalopod ink sacs [59,60], a fish ‘eye spot’ [61] and marine reptile integuments [56]. Indeed, most occurrences of fossil melanosomes reported so far (particularly in feathers) are based entirely on morphology, packing and distribution (e.g. [7–10,52–54,62]), whereas chemical studies, with few exceptions (see above), have either been inconclusive (e.g. [14,63]) or lacking in specificity and/or relevant comparative material to rule out alternative hypotheses (e.g. [55,64,65]). Most critically, many alleged melanosomes occur only as imprints (‘mouldic melanosomes’ [53]), a preservation mode that implies preferential degradation of the bodies relative to the surrounding substrate [12]. To complicate matters further, impressions indistinguishable from those attributed to melanosomes are occasionally found also in clay minerals, on silica crystals, and other sedimentary grains associated with, but clearly distinct from, the fossilized tissue structures (e.g. [9,12,65]).

These conflicts highlight the need for clear and unambiguous criteria by which remnant melanosomes can be differentiated from microbial residues. Clearly, this is vital for any accurate inference of organismal colour and its subsequent influence on ancient behaviours or ecology. We therefore outline factors contributing to the preservation potential of both melanin and melanosomes, and evaluate published methodologies employed to identify fossil microbodies. We also propose an integrated structural and molecular approach to promote rigorous interpretation of fossil melanins and melanosomes in the future. Firstly, however, we address a series of untested assumptions regarding melanosome data derived from the vertebrate fossil record.

5. Assumptions underlying assignment of colour to fossil organisms

6. A way forward: integrated morphological and geochemical approaches

The intrinsic resistance of melanin to degradation (except in the presence of cells/enzymes targeted specifically against it) and its preservation in fossils highlight significant potential for elucidating the biology of extinct organisms. Nevertheless, such interpretations are inherently equivocal, and further complicated by the fact that microbes: (1) overlap in size, shape and preservation potential with melanosomes; (2) can produce melanins; and (3) are innately associated with decaying organics [12]. Judicious elimination of alternative hypotheses for the origin of microbodies in fossils is therefore necessary prior to extrapolations of colour or function. While SEM imaging of external morphology and organisation provides a viable first step, it is not sufficient for a definitive diagnosis [12]. To augment morphology, higher resolution images should therefore be obtained from field emission gun scanning electron microscopy (FEG-SEM), which can reveal distinctive surface features, such as melanin granules (e.g. [56, fig. 2c]). TEM imaging can also yield complementary information about internal structures. Generally, microbial cells have an electron lucent core ([12, fig. 4], [79, figs 7, 9–13], [80, fig. 1]), as opposed to the electron opaque interior of mature melanosomes (see [61, electronic supplementary material, fig. S2], [81, fig. 3]). Exceptions do exist, however, such as hollow melanosomes [82] and solid, ‘carbonized’ bacteria [16, fig. 7c]. Therefore, for confident identification, chemical fingerprints of eumelanins, pheomelanins and/or their degradation products must be localized to the fossil microbodies (e.g. [56,59–61]). Ideally, these biomarkers should not occur only as trace metals because microbes and the EPS they secrete can concentrate metal ions from the environment [83]. Thus, elevated trace metal levels in fossilized animal tissues could be bacterially mediated [84,85] or artificially concentrated during diagenesis [11]. Furthermore, many enzymes employed by microbes to degrade keratinized tissues (including both white and pigmented feathers) are metalloenzymes that use a variety of metal ions which deposit on keratin surfaces during decomposition [86].

7. Summary

Organismal colour holds deep fascination because species recognition, gender differences and many other traits are intimately linked to pigmented epidermal tissues. There is no doubt that biochromes were, and are, an integral substrate for natural selection. Indeed, the various patterns, hues and shades that we observe today unquestionably also had equivalents in the distant past. Hence, characterizing pigments in extinct animals has enormous potential to shed light on evolutionary aspects of biology and ecology. Despite this, the study of colour through deep time is still very much in its infancy and can be impinged by numerous experimental pitfalls. Caution should therefore be exercised in this endeavour. Illuminating aspects of ancient organismal colour is achievable only by thorough evaluation of all plausible hypotheses. These remain valid until disproven and should not be swayed by popular opinion. Developing a comprehensive understanding of exceptional preservation processes through careful actualistic experiments must also be augmented by knowledge of diagenetic effects and how they potentially alter the morphology and chemistry of microbodies associated with decaying organics. Such approaches will ultimately facilitate more rigorous interpretations and reduce the risk of spectacular yet insufficiently supported claims propagating in the literature.

Supplementary Material

Data accessibility

FUM-N-2268 is deposited at MUSERUM. Additional supporting data can be accessed from the Department of Geology, Lund University.

Authors' contributions

J.L., A.M., M.H.S., D.E.N. and B.P.K. wrote the manuscript. J.L., A.M., P.S. and P.U. made the illustrations. J.L., A.M., P.S., P.U., J.H., A.E. and J.A.G. performed the analyses and interpreted the data. B.P.S. provided access to FUM-N-2268. All of the authors discussed the results and content of the manuscript.

Competing interests

We have no competing interests.

Funding

This research was supported by the Swedish Research Council, the Royal Physiographical Society in Lund and the Crafoord Foundation to J.L., the David and Lucile Packard Foundation to M.H.S., and the US National Science Foundation to M.H.S. and A.M.