Friday, September 16, 2011

Feathers ARE Color

A lot of my posts lately have been dedicated to color in prehistoric feathers (primer here). As a  paleontographer, it's hard not to get excited about finding new ways to more accurately and naturally depict long-gone animals, and I know a lot of you reading this feel the same way. It can also be frustrating knowing the basic color palate we're dealing with, but having specific data from only a handful of species to draw inspiration from. I've already seen, not surprisingly, a few illustrations of Microraptor, Xiaotingia, etc. that basically mimic Anchiornis down to the spangles. We're so conditioned to follow phylogenetic bracketing that it's tempting to use what little info we have, and extrapolate it to close (and even not so close) relatives. This can apply to larger feather patterns as well; as I mentioned in this post, it's fine to extrapolate Microraptor-style feathering to Velociraptor, but it's not the ONLY way to do it, nor is it even necessarily the most plausible for a creature of such different size and ecosystem, as many online critics of Dinosaur Revolution would have you believe ;)

With that in mind, I want to start with a disclaimer: this kind of info only gets you so far. I'm not setting out to write the inerrant Bible of paleoart in these types of posts. Yes, all the evidence we have of dromaeosaurs shows full, lush feathering equivalent to small modern birds. But then again, we're dealing with small prehistoric birds here. Yes, we can interpret feather coloration and use inferences from modern birds and biochemistry to say what colors are likely and which are unlikely; but we can't always identify structural color in fossils, and we have yet to identify carotenoids, and many small dinosaurs we think of as "carnivorous" might well have trended more toward omnivory, insectivory, etc., or other unforeseeable pathways to those tasty, tasty bright yellow pigments.

Ok, preamble out of the way, the point of this post is to provide a general way I've used, and others can use, to try to make somewhat educated guesses about coloration in prehistoric birds which have not yet been analyzed for color pattern.

A few people pointed out to me when this recent paper by Wogelius et al. came out, in which the research team was able to identify color patterns in some Mesozoic birds using chemical markers rather than direct observation of melanosomes, that their reconstruction of Confuciusornis based on these findings was pretty similar to the restoration I'd done a while earlier.

Above: Restoration of Confuciusornis by Richard Hartley, from the press release. Below: My own earlier restoration of C. sanctus.

The two are pretty similar in the broad pattern: dark head, body, and coverts, white or light-colored wings with more black on the secondaries than the primaries. Now, so far this is only one data point, so I don't want to draw too many conclusions. But I was not simply guessing when I restored C. sanctus that way back in 2009.

The trick is to understand what you're looking at when you see a fossil feather. These are often referred to as 'impressions', but often the impressions are only part of the story, or even completely absent from the fossil. The breakthrough that led Vinther , Prum and others to figure out how to suss out feather color, was in realizing what fossil feathers are made of.

As Prum summarized on the Skeptics Guide to the Universe podcast a few years back, traditional wisdom held that the dark, 'carbonized' looking fossil feathers were the result of bacteria. Now, some old-school feather impressions, like those from Solnhoffen and the Santana formation, do contain nice, deep, actual impressions. These look very different from the stuff you find in the Jehol (compare the fossil feathers of the Berlin Archaeopteryx with the holotype of Microraptor gui). But most fossil feather,s including the famous single-feather holotype of Archaeopteryx and most Jehol stuff, are dark stains in the rock. Under the microscope, these looked to early researchers to be made up of fossil bacteria which had eaten away the feather keratin during decomposition.

Above: Photograph of an unsubscribed Anchiornis specimen by Robert Clark, from National Geographic, showing clear color patterns matching those predicted by Vinther et. al.

What Prum, Vinther, etc. showed in recent years is that this is flat-out wrong. Those granules are not bacteria--they're melanin! When you look at a fossil feather, most of the time, you're looking directly at the color pattern of the feather itself, the keratin and everything else having long since disappeared. This is especially apparent in very well-preserved fossils; for example, the beautifully preserved new Anchiornis specimen above is essentially proof of Vinther's hypothesis, which had previously been based on more obscured differences in shade. This is where my method of eyeballing it falls flat--I'd never have gotten the correct pattern from the specimen Vinther was looking at without really close physical examination. You need really nice specimens for it to conceivably work, or you have to Dave-Peters the heck out of low-res images trying to spot differences in contrast on the feathers.

Luckily, Confuciusornis was a safe bet, because so many specimens are known, and you can start to see patterns emerge. Many of the best specimens tend to have a very dark halo of feathers around the body and arms, with the wing feathers very faint, even sometimes difficult to see at all. Knowing that dark feathers means dense melanin = dark coloration, and light color = lack of melanin = light color, it was easy to come up with a good guesstimate of the life coloration. This was first inspired by Longrich's work on Sinosauropteryx, showing that the apparent bands in the tail were due to color patterning, which was later supported by published studies.

Obviously this is not a foolproof method. But it's a good place to start for artists who may want to add a little evidence-based thinking to their reconstructions, even if the evidence itself is wide open to interpretation. At the very least, if a good specimen has obvious areas of light and dark, I'd personally restore that pattern in a life illustration. It may end up being wrong, but it's better than pure speculation.