How Do We Know What Colour Dinosaurs Were? Pigment Structures in Fossil Feathers

Melanosomes are tiny sacs within cells that contain melanin, the pigment responsible for black, grey, brown and reddish colours in feathers, hair and skin. In living birds their shape correlates with the colour they produce, with elongated rod-shaped ones associated with black and grey and rounder ones associated with reddish brown, and the correlation is strong enough across many species to be used predictively. Crucially, the structures themselves can survive fossilisation under exceptional conditions, appearing under an electron microscope as arrays of tiny bodies within the outline of a fossil feather. Finding them and measuring their shapes therefore offers a route to colour that nothing else provides.

The procedure is indirect and each step introduces uncertainty:

• •Locate exceptionally preserved feathers, generally from a small number of fossil sites
• •Examine samples under an electron microscope to find structures of the right size and arrangement
• •Demonstrate the structures are biological rather than bacteria, which look similar and was the original objection
• •Measure shapes and dimensions across many structures from different body regions
• •Compare statistically against a reference set from living birds of known colour
• •Report the resulting colour with its uncertainty, which published work generally does and popular coverage generally does not

The most serious challenge to the method was that the structures might be fossilised bacteria rather than pigment organelles, since bacteria are of similar size and shape and would be expected on a decaying carcass. The argument was pressed seriously and was answered on several grounds. The structures are confined to the feather outlines rather than scattered across the surrounding sediment, which decay bacteria would not be. Their arrangement follows the internal structure of a feather. Chemical analysis has detected melanin itself in some specimens, using methods that identify the molecule rather than inferring it from shape. And the distribution of different shapes across a single animal produces coherent patterns, including stripes and patches, which random bacterial colonisation would not generate.

Melanin-based colours are only part of what living animals display and the limits of the method should be stated. Reds, yellows and oranges produced by carotenoid pigments, which birds obtain from their diet, leave no structure and are essentially unrecoverable, so an animal that was brightly coloured by that route would be reconstructed as drab. Structural colours arising from the ordered arrangement of the structures themselves can sometimes be inferred, which is how iridescence has been proposed, and that inference depends on the arrangement surviving intact, which is demanding. Colours in bare skin, in soft tissue and in anything not preserved are unavailable. And a reconstruction describes the individual specimen, which may be a juvenile, a particular sex or a seasonal state rather than the species in general.

Several animals now have colour reconstructions with published evidence. A small feathered dinosaur from China was reported in 2010 as having a reddish crest and a striped tail, which was the first such claim. Another was reconstructed as predominantly black with iridescence, inferred from the ordered stacking of structures that produces structural colour in living birds. A larger feathered species was reconstructed with countershading, meaning darker above and lighter below, which is a widespread pattern in living animals and indicates something about the light conditions it lived in. Camouflage patterns inferred from such work have been used to argue about habitat. The reconstructions apply only to animals with preserved feathers or skin from a few localities, which is a small and unrepresentative sample of the whole group.

Pigment-containing structures survive in some exceptionally preserved feathers, and their shapes correlate with colour across living birds well enough to predict it. The objection that they were fossil bacteria was answered by their confinement to feather outlines, by direct chemical detection of melanin and by the coherent patterns they form. Only animals from a few exceptional sites can be reconstructed this way.