Friday 25 September 2015

What pterosaurs tell us about the evolution of feathers

2011 PR image for the 2014 description of Laquintasaura venezuelae, a basal ornithischian from Venezuela. Scales were the requested integument for this reconstruction, but how does that decision hold up today?
For the last two weeks I've been revising an image of the Jurassic ornithischian Laquintasaura venezuelae. The original (above) was produced in 2011, but a request to include it in an upcoming book was impetus to tidy up the art and update the anatomy. One significant question for updating this old piece was whether the animals should stay scaly or receive a coat of filaments. The systematic placement of Laquintasaura isn't certain, but it seems to lack features allying it to major ornithischian clades and, for now, is simply considered a basal member of Ornithischia (Barrett et al. 2014). This puts it in a controversial spot as goes interpretation of dinosaur integument: scales, filaments, or a mix of both?

The origins on filamentous integuments and feathers in reptiles remains an ongoing source of fascination and investigation for palaeontologists. It has been known that filamentous reptilian integuments extend deep into geological time since the 1800s, but research into these structures exploded in the 1990s and 2000s when fossils of many non-avian theropods - seemingly all coelurosaurs - were found adorned with feathers or filamentous feather precursors. Soon after, recovery of quills, filaments and strange, fibrous scales in ornithischians made a reality of once speculative ideas about filaments being widespread across Dinosauria. For years now, palaeontologists have been discussing the possibility that theropod filaments and feathers share ancestry with those of ornithischians. One implication of this is that bodies of dinosaur ancestors would be covered in fuzz instead of, as traditionally supposed, scales. Unravelling this conundrum is of key interest to those attempting to understand ancient reptile evolution and physiology, as well as for artists wanting to know how to credibly restore early dinosaurs. However, integument preservation, and particularly filamentous hide, is rare in the fossil record. Much as we might want to, we currently have insufficient data about the skin of early dinosaurs to address this issue directly.

All is not lost, however: some insight into dinosaur filament evolution can be provided by pterosaurs. Flying reptiles and dinosaurs are largely thought to form a more or less exclusive clade, the Ornithodira, which we now recognise as being characterised by a suite of anatomies - not just hindlimb features, as originally proposed - and commonalities of interpreted anatomy: postcranial pneumaticity, upright postures, elevated metabolisms, and filamentous integument. It's the latter which makes pterosaurs potentially useful to understanding the ancestral state of dinosaur skin. It's a little surprising that it's taken us so long to capitalise on this data, since we've had conclusive evidence of pterosaur filaments (we call them pycnofibres) since the 1970s (Sharov 1971). Suggestions that pycnofibres may have been homologous to dinosaur fuzz arrived much later, in the 2000s, when the evolutionary depth of dinosaurian filaments had become apparent and new discoveries of fuzzy pterosaur fossils were being reported (Czerkas and Ji 2002; Ji and Yuan 2002). Perhaps it was the coincidence of these events, the realisation that filaments were widespread in Pterosauria, and increased confidence in the sister relationship between dinosaurs and pterosaurs which lead to this idea finally being proposed.

Late Jurassic pterosaur Sordes pilosus, described in 1971, was one of the first pterosaurs confirmed to have a filamentous body covering. But are pterosaur filaments tied to those of dinosaurs, or independently evolved?
Studies into pterosaur and dinosaur filament homology remain thin on the ground, and much of what has been said thus far is reliant on gross filament morphology. Earlier this year, a team of researchers (Barrett et al. 2015) tackled the issue of ornithodiran filament evolution quantitatively, estimating the likelihood of homology between theropod, ornithischian and pterosaur integuments via their distribution on the ornithodiran tree. Using 18 different variations in methods, calculations and data values, they predicted the likelihood of ancestral integument states in dinosaurs and ornithodirans: were they scaly, filamentous, or feathered? The result, announced in not only the paper but also a subsequent media release, was that 12 of those 18 assessments suggested scales were ancestral to ornithodirans, and the filaments seen in pterosaurs, ornithischians and theropods were derived independently from a common scaly ancestor.

This conclusion was undoubtedly surprising to some and, indeed, a clear caveat accompanies it: scaly ancestral dinosaurs are "sensitive to the outgroup condition in pterosaurs". Support for ancestrally-scaly ornithodirans relies on the assumption that pterosaur ancestors were also scaly. This condition assumed for 50% of those 18 assessments to account for uncertain ancestral condition for pterosaur integument. In the 9 analyses where pterosaurs were treated as wholly filamentous - and thus consistent with what we see in existing pterosaur fossils - six returned results indicating an ambiguous scaly/filamentous ancestral condition for ornithodirans and dinosaurs, and only 3 supported a wholly scaly interpretation. Of those six 'ambiguous' results, most reported a strong likelihood of ornithodirans being ancestrally filamentous, and many gave dinosaurs a good chance of being ancestrally filamentous too. Moreover, treating pterosaurs as filamentous has knock-on effects through the dinosaur tree: suddenly, there are reasonable, or at least equivocal, chances that ornithichians and saurischians were also ancestrally filamentous. This is a different conclusion to the straighter story of ornithodirans and dinosaurs simply being ancestrally scaly.

What influence do fuzzy pterosaurs have on dinosaur skin evolution? Seemingly, quite a bit. The tree on the left shows integument likelihoods (pie charts) where pterosaurs are considered scaly, tree on the right shows a filamentous analysis.  Modified from Barrett et al. (2015).

Clearly, the crux of all this is the assumption that pterosaur ancestors were scaly: just how defendable is this? Because we know little about pterosaur origins, it's hard to say anything conclusive about the evolution of pterosaur integument with our current fossil record. The stratigraphically oldest pterosaur fossil with pycnofibres is from Middle/Late Jurassic deposits, and thus about 50-60 million years younger than the oldest pterosaur fossils - little help in determining if the first pterosaurs were fuzzy. Ongoing disagreements over pterosaur phylogeny complicate attempts to estimate the appearance of lineages with confirmed pycnofibres. Some schemes (those derived from Kellner 2003 and Unwin 2003) suggest pycnofibres must have appeared by the Triassic, close to or at the base of pterosaur ancestry, but others (e.g. Andres et al. 2010) indicate pycnofibres reliably extend no further than the Lower Jurassic. Of course, such assessments of filament distribution might not even be meaningful at this stage, given that pycnofibres are very rare components of pterosaur fossils. They are nowhere near as common as other soft-tissues, such as wing membranes, and we should probably be cautious about any assessment of their evolutionary pathways until we have more data. Perhaps the only significant observation we can make from our current, limited dataset is that, to date, no pterosaur is known with a scaly body covering, even when regionalised scalation - foot pads - preserves in their fossils (Frey et al. 2003).

A possible pterosaur relative with scaly hide is known: the Triassic archosaur Scleromochlus taylori. Benton (1999) described structures interpreted as thin, transversely orientated scales across the back of multiple specimens of this animal. This might provide vindication of the scaled pterosaur ancestor model, but, again, there are some caveats with this idea. For one, Scleromochlus fossils are not well preserved. The scales are feint sediment impressions, visible only in strong, low angle light, such that that they are only considered 'probable' integument impressions by Benton (1999). Previous workers have interpreted them in a different way (as gastralia). Clearly, the evidence for them being scales could be more compelling, and there's certainly not much to work with if we want to test their identification. Secondly, exactly how Scleromochlus is related to pterosaurs is not precisely agreed. Some workers consider it the sister taxon of Ornithodira, others as a member of the pterosaur branch, and others see it as more closely related to dinosaurs than pterosaurs. That might seem a minor issue, but we've already seen how sensitive models of ornithodiran integument are to changes of single variables at the base of the tree. We would probably need to run many variants of the integument probability calculations to account for all the uncertainty surrounding Scleromochlus. This might give more idea of the range of possible integuments at the base of ornithodiran evolution, but that's not much of an improvement on our current situation.

Was Scleromochlus taylori scaly? Maybe - weakly preserved structures on several specimens seem to suggest so. On this diagram, from Benton (1999), possible transverse scales can be seen on the left and middle specimen.
In all, I feel like we're hitting a bit of a wall here. It seems we just don't know enough, and have too many caveats with the limited data we have, to make even a half convincing best guess on this. Thus, how much weight we put on models of ornithodiran integument using scaly pterosaurs is almost a philosophical issue. From my end, I don't think they should be used to argue for scaly ornithodiran and dinosaurian ancestors, at least not with the same weight as tests made using a filametnous pterosaur lineage. When reconstructing ancestral states, characters objectively observed in fossils have to trump assumed character states, even if we know that our dataset is full of holes. After all, the whole point of attempting to figure out an ancestral state is establishing links between character data we have, so introducing opposing character states seems a little contrary to that objective. To be clear, I'm not saying that running models with scaly pterosaur ancestors is a waste of time. To the contrary, it's a good test of model robustness, and Barrett et al. (2015) certainly demonstrate how sensitive our models of ornithodiran integument evolution are by using this approach. Their hypothetical scaly pterosaurs demonstrate that we really do need more early ornithodiran fossils to understand ornithodiran skin evolution. However, I do not think that results of the scaled pterosaur analyses are as informative as their other assessments, as we have to overlook existing data to consider them equally valid.

With all that said, do pterosaur fossils really help us understand the evolution of dinosaur filaments? Playing the conservative card here, it seems they do not provide super strong evidence for an all-fuzzy Dinosauria, but they certainly make it difficult to defend ideas of entirely scaly dinosaur ancestors. Forcibly arguing for either scales or filaments at the base of Dinosauria seems premature at this stage, and, whatever our personal hunches are, it seems sensible to accept some ambiguity in this situation for now.

I began this article with my Laquintasaura conumdrum: how did that play out when, apparently, I can't make up my mind about this scales and filaments debate? Well, I've argued elsewhere that palaeoart can do no better than illustrate credible interpretations of the past and that, so long as the hypotheses they depict are sound, they're doing OK. When we have conflicting or ambiguous hypotheses, we just have to make a judgement call based on our own opinions, gut feelings and interpretations of existing arguments. With my own leaning being towards data showing that scales may not be ancestral to ornithodirans, but also knowing that some dinosaurs are mosaics of filaments and scales, I decided to partially enfluffen my Laquintasaura, while leaving their snouts, tails and limbs scaly. I'll leave you with the revised image.

Laquintasaura venezuelae 2015 edition: basically the same picture, but a bit fluffier, and a bit greener.

This post was brought to you by Patreon

Production of the imagery and article seen here was sponsored by my awesome Patrons, who back me at Patreon. As a bonus, they're getting to see the full version of that Sordes image, along with other rewards and benefits including more exclusive content, discount print prices and other neat stuff. If that sounds like something you're interested in, you can get involved for as little as $1 a month. A huge thanks to those who have signed on already!


  • Andres, B., Clark, J. M., & Xing, X. (2010). A new rhamphorhynchid pterosaur from the Upper Jurassic of Xinjiang, China, and the phylogenetic relationships of basal pterosaurs. Journal of Vertebrate Paleontology, 30(1), 163-187.
  • Barrett, P. M., Butler, R. J., Mundil, R., Scheyer, T. M., Irmis, R. B., & Sánchez-Villagra, M. R. (2014). A palaeoequatorial ornithischian and new constraints on early dinosaur diversification. Proceedings of the Royal Society of London B: Biological Sciences, 281(1791), 20141147.
  • Barrett, P. M., Evans, D. C., & Campione, N. E. (2015). Evolution of dinosaur epidermal structures. Biology letters, 11(6), 20150229.
  • Czerkas, S. A., & Ji, Q. I. A. N. G. (2002). A new rhamphorhynchoid with a headcrest and complex integumentary structures. Feathered Dinosaurs and the origin of flight, 1, 15-41.
  • Frey, E., Tischlinger, H., Buchy, M. C., & Martill, D. M. (2003). New specimens of Pterosauria (Reptilia) with soft parts with implications for pterosaurian anatomy and locomotion. Geological Society, London, Special Publications, 217(1), 233-266.
  • Kellner, A. W. (2003). Pterosaur phylogeny and comments on the evolutionary history of the group. Geological Society, London, Special Publications, 217(1), 105-137.
  • Ji Q., & Yuan C. (2002) Discovery of two kinds of protofeathered pterosaurs in the Mesozoic Daohugou Biota in the Ningcheng region and its stratigraphic and biologic significances. Geol. Rev. 48, 221–224.
  • Sharov A, G. (1971). New flying reptiles from the Mesozoic of Kazakhstan and Kirghizia. - Transactions of the Paleontological Institute, Akademia Nauk, USSR, Moscow, 130: 104–113 [in Russian].
  • Unwin, D. M. (2003). On the phylogeny and evolutionary history of pterosaurs. Geological Society, London, Special Publications, 217(1), 139-190.


  1. Very intersting article, a real pitty that the fossil record is so incomplete, well... Maybe be will one day get lucky and found deffinitive proof of the intergument cover of basal Ornithodirans.

  2. Is it possible that Ornithodira could be basally scaly but with strong genetic traits that make the independent evolution of filaments very easy? To put it another way, perhaps these lineages may have independently expressed pre-existing genes, but did not independently evolve them.

    1. Thanks for the comment, Andrew. I guess the whole point of the Barrett et al. study was to determine the likelihood of that hypothesis. If the ornithodiran ancestor was scaly, then, yes, we might assume that there was genetic predilection for developing fuzziness in each ornithodiran lineage (as in, it was 'easy' for them to do it, which is why it developed multiple times).

  3. Isn't Archosauria fuzzy ancestrally?

  4. Perhaps I've been reading too much Reptipage, but your post's main argument for feathered ornithodirans sounds a bit like you're instantly pouring proto-feathers into all gaps in existing data. Both hypotheses are likely at this point, but it doesn't translate into both being equally likely - to be honest, this concerns everything when it comes to reconstructing extinct animals.

    I can't access full data from Barret et al. 2015, but what would happen if you extended the model a further 50 million years back to include more lineages like crocodilians, saurischians, and ultimately *mammals* as well (assuming mammal hair=proto-feathers)? Y'know, just as control factors. We know that diverse lineages can, and do, undergo changes filamentary structures at much shorter timescales than the one involved here: turkey beards, pangolin scales, etc.

    Regardless of all that, I think Andrew Raymond Stück raises a good point. Organisms inherit genes, not features. Perhaps the crucial genes responsible for the diversity of integumentary structures in dinosaurs and pterosaurs can indeed be traced back to ornithodiran or archosaur clades, but that'd mean something different than saying they inherited the same phenotypical structures. The distinction between scales & feathers was a big leap, but so was the leap between different stages of feathers, and unless I'm mistaken, both ornitischians and coelulosaurids show evidence of "stage 2"-style feathers. Personally, I wouldn't treat is at evidence for proto-branching-feathers in their last common ancestor, not any more than evidence for proto-wings in the last common ancestor of pterosaurs, bats and birds...

    1. Oops, one correction: I meant possibly extending the model to include 'lepidosauria', not 'saurischians'. Damn all those pesky lizard-derived names.

    2. This comment has been removed by the author.

    3. This comment has been removed by the author.

    4. I think you're missing the point of the Barrett et al. study, which was testing the probability of filaments or scales occurring in the ancestors of ornithodiran clades. No-one 'poured proto-feathers into all gaps in existing data': they used current data to see what integument types were likely for different stages of ornithodiran evolution.

      "what would happen if you extended the model a further 50 million years back to include more lineages like crocodilians, saurischians, and ultimately *mammals* as well (assuming mammal hair=proto-feathers)?"

      Given that we have heaps of scaly animals between ornithodirans and mammals on the amniote tree, I'm not sure this would tell us anything. There's no evidence for filaments on the reptile lineage until we get to Ornithodira, which is why (I guess) the Barrett et al. study focused on this group alone.

    5. Thanks for the reply.

      My comment about filling blanks with proto-feathers concerned your skepticism of Scleromochlus being scaly, not Barret et al.'s study per se.

      I enjoyed your analysis, but it gave me the impression that you're holding onto every little bit of doubt concerning the suggestion that pterosaur ancestors might have been scaly just to tip the model towards the interpretation you prefer, based on your "leaning".

  5. I think it's important also to know if the alligator "feather genes" are actually correlated to filemental integrument. If they are, then it makes it unambiguous that Archosauria started as possessing "feathers"

  6. Dinosaurs had bristles.

  7. What kind of integuments that Tyrannosaurs have? Is it plumulaceous feathers like what the Chinese paleontologists guess about Yutyrannus or is it simple filaments with no branches or calamus that distributed individually like mammalian hairs? By looking at the integuments on Dilong, I doubt it look like down feathers because it look more similar like individual hairs that are clumped into one.