On its own, melanin generally produces shades of gray, brown, or black, but iridescence has the potential to create just about any type of color. This is determined by the thickness of the keratin and melanin layers, which influences the specific wavelengths of light they reflect. Given this versatility and its inherent glimmering qualities, it's no surprise that iridescence has been adopted by so many birds (and other animals) for the purpose of visual display.
 |
| Male Anna's hummingbird showing off its brilliant iridescent plumage, photographed by "Norvig", under CC BY-SA 3.0. |
The cell structures that contain melanin pigments in feathers are called melanosomes, and in recent years paleontologists have become increasingly interested in melanosome morphology. This is due to the discovery that melanosomes are often preserved in fossilized feathers and other soft tissues. Studying the correlations between melanosome shape and plumage coloration has allowed the original colors of various fossil species to be at least partly inferred.
As it turns out, melanosomes that contribute to iridescence come in several different forms. They can be solid and cylindrical, making them similar to typical melanosomes that produce blacks and grays. However, they are generally still recognizable by being relatively longer and thinner in shape compared to "standard" melanosomes. Iridescence-producing melanosomes can also be hollow and cylindrical, solid and flat, or hollow and flat. These specializations make sense for generating iridescence, as being hollow increases the number of surfaces scattering light and a flattened form allows many melanosome layers to be packed into a single feather.
 |
| Schematics showing different morphologies of iridescence-producing melanosomes with examples of bird groups that have each type, from Nordén et al. (in press). (A) shows solid, cylindrical melanosomes (found in galliforms), (B) shows hollow, cylindrical melanosomes (found in trogons), (C) shows solid, flat melanosomes (found in sunbirds), and (D) shows hollow, flat melanosomes (found in hummingbirds). |
These distinctive morphologies have led to iridescent plumage being inferred for several extinct theropods, including the dromaeosaurid Microraptor, the anchiornithid Caihong, an unnamed enantiornithine, the Eocene apodiform Eocypselus, and the Eocene gruiform Messelornis. Most of these taxa had cylindrical melanosomes, but Caihong is known to have had flat ones. Unfortunately, it is currently impossible to assess the original thickness of the keratin layers in these fossil feathers, so the exact hue of iridescence in these taxa cannot be reconstructed.
However, even in living theropods, the diversity of iridescence-producing melanosomes is not fully understood. To investigate the variety and evolution of melanosomes that contribute to iridescence, a new study assembled the largest dataset of iridescent bird feathers to date and compared their melanosome forms to those of other feathers. The authors found that iridescence-producing melanosomes were far more morphologically diverse than melanosomes that generate other colors, taking on forms that have not been observed in any other kind of melanosome. This is perhaps not a surprise, given that iridescence-producing melanosomes are the only ones known to be hollow or flattened.
 |
| Morphospace graph showing the morphological diversity of melanosomes, from Nordén et al. (in press). Note that iridescence-producing melanosomes (in purple) cover a much larger area than any of the other types. |
Despite this diversity, the iridescence-producing melanosomes of different birds tend to converge on each another in form. Iridescence has evolved many times in birds (more than three times in galliforms alone), and yet the melanosomes that produce it are typically similar in width, length, and aspect ratio. At least, that is the case when it comes to cylindrical melanosomes. Flat melanosomes (especially hollow ones) do not show such striking convergence between the bird groups that have them. This may be because their flattened shape allows for greater variation in general. For example, width and thickness are equivalent for cylindrical melanosomes but are independently variable parameters for flat ones.
As a matter of fact, it is not clear why cylindrical melanosomes show such a strong trend of convergence to begin with, given that not all variables governing melanosome form directly affect color production. The authors suggest the trend may relate to the optimal organization of the melanosomes within a feather, an interesting idea that awaits further testing.
Probably the most exciting aspect of the study (from a paleontologist's perspective) is that the authors were able to use their big dataset to test the accuracy of different models used to infer plumage coloration from melanosome morphology. They applied the most accurate model they found to two fossil birds with preserved feathers: the Eocene swift Scaniacypselus and the Oligocene trogon Primotrogon. Both of these birds have close modern relatives with iridescent plumage.
The model did predict one feather sample from Scaniacypselus as iridescent, but with low statistical confidence. Several other samples were inferred to be gray or brown, with greater statistical support. Thus, the authors concluded that the feathers they sampled from Scaniacypselus were probably not iridescent.
 |
| My amateur restoration of Scaniacypselus, which was inferred to have had non-iridescent, gray feathers across its body by Nordén et al. (in press). One feather sample from the leading edge of its wing was also inferred to have been brown, but that region probably wouldn't have been visible when the wings were folded. The authors do caution that the gray plumage they sampled might have represented down feathers (which are typically gray in most birds), in which case their results might not be so indicative of how Scanicacypselus would have looked in life. |
Gray plumage was also inferred for most of the samples from Primotrogon. However, one Primotrogon sample was strongly suggested to be iridescent! Interestingly, the melanosomes responsible for this inferred iridescence were solid, not hollow as in modern trogons. This is consistent with previous studies on the evolution of iridescence in birds: iridescence-generating melanosomes that are flat, hollow, or both readily evolve from one another, but rarely (possibly never) revert back to the ancestral solid, cylindrical form.
 |
| My amateur restoration of Primotrogon, which was inferred to have had gray and iridescent wing feathers by Nordén et al. (in press). (Okay, actually a colored-in restoration of Masillatrogon, which was morphologically similar to Primotrogon.) The hue of iridescence is unknown; I depicted it as green based on modern trogons. As with Scaniacypselus, the authors caution that they couldn't reject the possibility that the gray feathers were down. |
I look forward to seeing this new dataset used to study the origin of iridescence in other bird groups. Certainly there are quite a few bird specimens from Messel that would be interesting to look at!
Reference: Nordén, K.K., J. Faber, F. Babarović, T.L. Stubbs, T. Selly, J.D. Schiffbauer, P.P. Štefanić, G. Mayr, F.M. Smithwick, and J. Vinther. In press. Melanosome diversity and convergence in the evolution of iridescent avian feathers—implications for paleocolor reconstruction. Evolution in press. doi: 10.1111/evo.13641
