Friday, January 20, 2023

Are uncinate processes ancestral to Archosauria? A look at Wang et al 2023

A paper was published on January 17th (Wang et al., 2023) that I've been looking forward to since its SVP 2020 poster (Wang et al., 2020).  The basic idea is that uncinate processes leave attachment scars on ribs even if the processes themselves are unossified, and these scars show that uncinate processes were much more widespread than the record of ossified processes would suggest.  In theropods, ossified uncinate processes are only known in many pennaraptorans, Pelecanimimus and the noasaurid "Sidormimus" (Sereno, 2010; not mentioned by Wang et al., 2023) for instance, but Wang et al. found scars in Struthiomimus, Gorgosaurus and Allosaurus, all of which definitely lacked ossified processes.  Similarly, Wang et al. identified scars in Apatosaurus, when I don't think any sauropodomorph has been reported with ossified processes.

So my first question in 2020 was "do any archosaurs NOT show these scars?", and the SVP abstract and poster didn't say.  Even in the published paper, the most we get in the main section is "Using an alternate coding approach in which uncinate processes were coded as absent in taxa represented by five or more dorsal vertebral ribs that all lacked uncinate scars, nine archosaur taxa were coded as lacking uncinate processes."  Which nine taxa?  Who knows.  The Methods section at the end of the paper similarly says "Our alternative coding approach, which was used to test the stability of the results obtained under our preferred coding approach, differed in that uncinate processes were coded as absent in taxa for which at least five vertebral ribs were available, regardless of their state of preservation, and showed no sign of uncinate processes or uncinate scars. This resulted in coding uncinate processes as absent in nine taxa."  It does detail that their outgroup, the proterochampsian Chaneresuchus, lacks the scars so was assumed to lack any uncinate processes, but that's it.

Their only figure besides two showing the scars is Figure 3, an "informal consensus cladogram", whose caption reads "Major clades of Archosauria with evidence of cartilaginous uncinate processes are labelled and shaded blue; clades with evidence of ossified uncinate processes are labelled and shaded pinkish red; clade with evidence of both cartilaginous and ossified uncinate processes is labelled and shaded purple; and clades for which no evidence is available are labelled and shaded grey."  Clades shaded gray are Shartegosuchidae, Pachycephalosauria, Macronaria and Megalosauroidea, the latter being the most basal theropod clade shown.  I guess the nine mystery taxa belong to those clades?  Also note every branch in each blue clade is blue, every branch in every red clade is red, etc., implying homogeniety.  Here's the figure-


It turns out the identity of the Mystery Nine is in Table 2 of the Supplementary Information, under Alternate Coding.  In the preceding paragraph they again report "Under the alternate coding method, nine taxa were scored as lacking uncinate processes."  These taxa are- Lotosaurus, Protosuchus, Lagerpeton, Plateosaurus, Anchisaurus, Camarasaurus, Camptosaurus, Parasaurolophus, Edmontosaurus, Protoceratops and Chasmosaurus.  Wait, that's actually eleven taxa, not nine, despite the number nine being stated three times.  On the one hand, note none of these are shartegosuchids, pachycephalosaurs, or megalosauroids, so why are those clades even in the cladogram?!  On the other hand, note that Lotosaurus and Protosuchus would make two gray nodes basal to mesoeucrocodylians, Lagerpeton would be a gray node basal to dinosaurs, and Plateosaurus and Anchisaurus would be two extra gray nodes basal to sauropods.  Also all of the Ornithopoda and Ceratopsida [sic] branches are blue when they shouldn't be.  So the figure is putting in artificial discrepancy and leaving out most of the actual discrepancy.

The big methodological problem with the paper is that their Preferred Coding approach scored "the presence of uncinate processes as uncertain (?) if evidence of uncinate processes or scars was lacking."  So the Mystery "Nine" were actually all scored as unknown, even though Chaneresuchus was scored as absent (0) based on identical evidence.  So obviously if you score all the archosaurs as present or unknown and the sole non-archosaur as absent, you're going to get archosaurs ancestrally having uncinate processes.  And when they use the Alternate Coding of taxa without scars not having uncinate processes, they themselves report-

"Maximum likelihood and Bayesian inference recovered cartilaginous uncinate processes as the most likely condition at Archosauria (pml=0.61, pmb=0.55), but only when branch length estimates were incorporated. By contrast, maximum likelihood excluding branch length estimates recovered the absence of uncinate process as the most likely condition at Archosauria (pml=0.92). The ancestral condition at Dinosauria could not be recovered with confidence using either maximum likelihood or Bayesian inference (pml=0.57, pmb ≈ 0.33). Maximum likelihood recovered ossified uncinate processes as the most likely condition at Maniraptoriformes (pml=0.90) and Pennaraptora (pml=0.99), but only when branch length estimates were excluded from the analysis. Bayesian inference could not confidently recover the ancestral conditions at Maniraptoriformes (pmb=0.33) and Pennaraptora (pmb=0.33)."

So not strongly supporting "deep reptilian evolutionary roots of a major avian respiratory adaptation", and that's not even going into how the only non-avian paravians they scored were four dromaeosaurids (Microraptor, Saurornitholestes, Velociraptor and Deinonychus) with ossified processes, when at least Archaeopteryx, anchiornithines and omnivoropteryids lacked them, so anything about Pennaraptora's ancestral state is going to need more than "oviraptorosaurs, dromaeosaurids, Rhea, Gallus and Lithornis - yes, Chauna - no."

And yeah, the living bird Chauna has no uncinate processes and lacks any scars for them, so was rightfully scored 0.  Yet fossil taxa with the same morphology were scored unknown.  Which again points to the flaw in their Preferred Coding, but also suggests we might expect homoplasy in other parts of Archosauria as well.  So lambeosaurines and edmontosaurins could have lacked uncinates while kritosaurins had them, following the osteological evidence, for instance.

In conclusion, it's an excellent idea to look for osteological correlates to unossified uncinate processes, and we got some real data hidden in the supplementary information, but any use was marred by unforced errors like a misleading main figure and a nonsensical scoring methodology.  I wouldn't doubt uncinates were primitive to at least averostrans, given "Sidormimus" and Edmarka (check Figure 14A of Bakker et al., 1992; so Megalosauroidea should have ironically been blue), but the concept deserved better vetting than this.

References- Bakker, Kralis, Siegwarth and Filla, 1992. Edmarka rex, a new, gigantic theropod dinosaur from the Middle Morrison Formation, Late Jurassic of the Como Bluff outcrop, with comments on the evolution of the chest region and shoulder in theropods and birds and a discussion of the five cycles of originn and extinction among giant dinosaurian predators. Hunteria. 2(9), 1-24.

Sereno, 2010. Noasaurid (Theropoda: Abelisauroidea) skeleton from Africa shows derived skeletal proportions and function. Journal of Vertebrate Paleontology. Program and Abstracts 2010, 162A.

Wang, Sullivan and LeBlanc, 2020. Anatomical and histological data indicate uncinate processes to be homologous across Archosauria. Journal of Vertebrate Paleontology. Program and Abstracts 2020, 334.

Wang, Claessens and Sullivan. 2023. Deep reptilian evolutionary roots of a major avian respiratory adaptation. Communications Biology. 6:3.


  1. What do you think of Benito et al.'s (2022) suggestion that Apatornis & Guildavis are Ichthyornis after all?

    1. It's intriguing but I haven't examined the data yet.