Flippers and limb proportions in extinct amniotes + tooth-bone attachment in reptiles

[Ben Creisler]

Ben Creisler

bcre...@gmail.com

New fossils:

======

Caleb M. Gordon, Lisa S. Freisem, Christopher T. Griffin, Jacques A. Gauthier & Bhart-Anjan S. Bhullar (2025)
Limb proportions predict aquatic habits and soft-tissue flippers in extinct amniotes
Current Biology (advance online publication)
doi: https://doi.org/10.1016/j.cub.2025.10.068
https://www.cell.com/current-biology/abstract/S0960-9822(25)01449-6
https://www.sciencedirect.com/science/article/abs/pii/S0960982225014496

Highlights

Amniote limb proportions predict flippers and aquatic habits with >90% accuracy
Mesosaurs and other Paleozoic reptiles did not evolve highly/fully aquatic habits
Mesozoic marine reptiles show lineage-specific patterns of aquatic adaptation
Phylogenetic ROC analysis reconstructs cryptic phenotypes in extinct species

Summary

Among mammals and reptiles, the recurring evolution of fully aquatic forms from land-dwelling ancestors highlights the remarkable powers and implications of natural selection. The most aquatically specialized of these groups have limb morphologies that betray a fully marine lifestyle, but the transitional forms near the base of each lineage have more ambiguous features, making it difficult to determine which fossil species were aquatic. Here, we use a scalable phylogenetic machine-learning pipeline to test previously proposed osteological correlates of interdigital webbing, soft-tissue flippers, and aquatic habits in amniotes. We collect >11,000 original measurements from amniote limbs and use these measurements to train and test phylogenetic logistic regression models that can predict aquatic affinities in extinct species. We then interpret and select among competing predictor models with receiver-operating characteristic analysis. Ultimately, relative hand length makes the best predictions, reconstructing soft-tissue flippers and aquatic habits with >90% accuracy across amniotes and clarifying the aquatic habits of fossil species with historically ambiguous ecologies. Placing these predictions on the phylogenetic tree of amniotes reveals semi-terrestrial habits in mesosaurs and all other sampled stem reptiles, highly/fully aquatic habits in all known ichthyosauromorphs, and multiple independent origins of highly/fully aquatic habits among sauropterygians, mosasaurs, and theropod dinosaurs. Taken more broadly, these results enable a broader comparative assessment of amniote limb proportions that reveals distinct evolutionary landscapes in limb morphometry for highly/fully aquatic vs. semi-terrestrial amniotes, as well as between total-group mammals and reptiles.

====

Free pdf:

M. Šulcová, J. Dumková, B. Hutečková, M. Kavková, V. Parobková, O. Zahradníček, J. Křivánek, I. Adameyko, J. Kaiser, T. Zikmund, A. S. Tucker & M. Buchtová (2025)
Tooth–bone attachment tissue is produced by cells with a mixture of odontoblastic and osteoblastic features in reptiles
Journal of Anatomy (advance online publication)
doi: https://doi.org/10.1111/joa.70059
https://onlinelibrary.wiley.com/doi/10.1111/joa.70059

Free pdf:
https://onlinelibrary.wiley.com/doi/epdf/10.1111/joa.70059


Teeth are anchored in the jaw in a highly variable manner across vertebrates. In mammals and crocodiles, the teeth are cushioned inside bony sockets by periodontal ligaments, whereas most squamate reptiles have teeth firmly attached to the jawbone. Here, we analyzed the development of the attachment tissue in the veiled chameleon, a species with firm acrodont tooth attachment, to reveal the cellular processes establishing ankylosis and to determine the cell types contributing to the attachment. The tooth-bearing bones formed pedicles with edges fusing to the dentine via an attachment tissue produced by morphologically distinct cells exhibiting both osteoblastic and odontoblastic features. These cells were RUNX2-positive, suggesting their potential to differentiate into hard-tissue-producing cells. However, in contrast to the osteoblasts of the bony pedicles, tooth–bone interface (TBI) cells expressed elevated levels of Na+-/K+-ATPase and thus resembled odontoblasts. TBI cells were visible only temporarily, and after tooth–bone fusion they were removed by apoptosis and phagocytosis. Dynamic deposition of the hard matrix continued on both sides of the TBI and during the posthatching stages through the participation of osteoblasts. Overall, our findings demonstrate both odontoblast- and osteoblast-like characteristics of cells producing the attachment tissue at the TBI during development in chameleons, highlighting the existence of a transient intermediate cell population, which we call ankyloblasts.

====

[Mickey Mortimer]

From Gordon et al.- "We also observe two independent origins of highly/fully
aquatic habits in sampled non-avialan theropods, with Lhand/
Lzeugopodium ratios predicting at least highly aquatic habits in
both Juravenator and the common ancestor of Spinosaurus
and Suchomimus (Figure 5A). The corresponding hindlimb
model (using Lfoot/Lzeugopodium ratios) did not recover highly
aquatic habits in any sampled theropods. However, given the
greater tendency of both models toward false negatives, the
positive results of the forelimb model are more likely to reflect
an accurate interpretation of aquatic habitat."

"Among theropods, hand-zeugopodium length ratios predicted
highly/fully aquatic habits for one32 of two31,32 existing 3D Spinosaurus
aegyptiacus forelimb reconstructions, the available Suchomimus
tenerensis forelimb material,32 and for the compsognathid
Juravenator starki (Figure 5A)."

"The highly aquatic habits recovered in the compsognathid
Juravenator likewise make sense of its piscivorous
dentition and hypothesized integumentary sense organs, which
may have been adaptations for hunting underwater.71,72 These
results for Juravenator and Suchomimus suggest that forelimb
morphometry may provide evidence of highly aquatic habits
even in cases where the postcrania are otherwise minimally or
only subtly modified for subaqueous locomotion."

Juravenator, really? And not just transiently (occasionally enters water for short time) or moderately (regularly spends a lot of time in water), but at least highly aquatic (occasionally leaves water to bask or breed) like a seal or sea turtle. Because it has a long manus relative to radius length? Note despite saying they measure digit III in reptiles, their Supplementary Table 1 shows they use digit II for theropods, which is good since III is almost always the weakest (except scansoriopterygids, Xiaotingia and marginally in some taxa like Herrerasaurus). The ratio (excluding the ungual, as per their methods; and only using the radius instead of radius+ulna average to save me time; the difference in length between radius and ulna is basically the olecranon anyway, which doesn't affect limb proportions) is 1.53, matching their 1.52 from Supplementary Table 1. There are a few weird things about their Figure 5A showing which theropods they used and where they fell out ratio-wise. Sinosauropteryx and Compsognathus were both used for hindlimb measurements, but both lack manual measurements despite both preserving complete digits II (coauthor Gauthier even illustrated both of these hands in 2007). Their table shows they used the Compsognathus holotype (well, a cast of it), which has a ratio of 1.47, so pretty close to Juravenator. For Sinosauropteryx, they use USNM 618325, which they claim is an original fossil, but the USNM online catalog shows is a cast of probable Huadanosaurus NGMC 2124. It does preserve a full digit II and ulna, but we do have an actual Sinosauropteryx specimen with measured and described complete forelimbs- NIGP 127587 which has a ratio of 1.83 that blows Juravenator out of the water. 

Indeed, it's weird because theropod measurements are pretty easy to get and comparable and higher ratios than Juravenator are common among compsognathid/tyrannosauroid grade taxa- Huaxiagnathus (1.98), Huadanosaurus (1.66), Scipionyx (1.62), Haplocheirus (1.70), Coelurus (~1.95), Tanycolagreus (1.55), Guanlong (1.51), Gorgosaurus (1.53), Nanotyrannus (1.61). Something tells me Gorgosaurus wasn't swimming with its arms, and instead short lower arms are typical for this grade. Similarly, derived carcharodontosaurids shortened their lower arms but kept their hands comparatively large, with Taurovenator having a ratio of ~1.92. Two other examples of this kind of ratio are Bannykus (1.69), which started the lower arm shortening of alvarezsaurs but hadn't yet shrunk digit II that kills the ratios in parvicursorines, and Apatoraptor (~1.48; ulna used instead of incomplete radius). 

Megalosauroids are another case like carcharodontosaurids, where large hands compared to radius length have long been used as a character of the group (generally using manual ungual I). Unfortunately, no megalosauroid preserves a complete digit II and radius, or indeed a phalanx II-2 at all except a proximal portion in the Spinosaurus neotype. Dilophosaurus has a II-2/II-1 ratio of .90, Asfaltovenator 1.08 and Erectopus 1.00 (sadly ceratosaurs are also incomplete, or highly modified like Limusaurus and abelisaurids), so a good guess is a phalangeal ratio of about 1.0 in megalosauroids. But even then we're out of luck because... Afrovenator has a proximal II-1 but only proximal radius and ulna; Leshansaurus has II-1 but lacks metacarpal II and radius/ulna; and the Spinosaurus neotype has only a fragment of metacarpal II and no radius/ulna. Gordon et al. use Sereno et al.'s (2022) reconstructions for Suchomimus and Spinosaurus, which for the former is just unguals and a partial metacarpal III, with a referred I-1 (fictional ratio of ~1.72). And for Spinosaurus it's the aforementioned II-1 and base of II-2, combined with isolated ulna, I-1, III-2, III-3 and unguals (fictional ratio of 2.32). They may be decent guesses at forelimb proportions for spinosaurids, but I'd hardly call it data.

Oh yeah, and instead of using one of the numerous oviraptorids with published complete limb measurements, they use a YPM cast of an AMNH specimen with no femoral or tibiotarsal measurements that they call Oviraptor philoceratops, despite it being labeled Citipati in both the YPM and AMNH online catalogs. Are we in the 90s? And they have no hindlimb measurements for Eoraptor, either Allosaurus they used, Sciurumimus, Tanycolagreus, any of the 3 ornithomimids they used, Halszkaraptor or Deinonychus. These are basically all in the same publications that provide forelimb measurements, so I don't see why you wouldn't include them.

In any case, far from the authors' conclusion forelimb ratios in theropods reflect aquatic habitats better than the rest of the skeleton, I'd say having a forelimb completely decoupled from terrestrial locomotion let them do their own thing. Which for certain taxa like basal coelurosaurs and probably megalosauroids was leaving the hands enlarged while shortening the forearm, which I assume has positive biomechanical effects not involving swimming. 

Mickey Mortimer

[Jaime Headden]

>  In any case, far from the authors' conclusion forelimb ratios in theropods reflect aquatic habitats better than the rest of the skeleton, I'd say having a forelimb completely decoupled from terrestrial locomotion let them do their own thing. Which for certain taxa like basal coelurosaurs and probably megalosauroids was leaving the hands enlarged while shortening the forearm, which I assume has positive biomechanical effects not involving swimming. 

The reason this is being done is because of penguins, which are forelimb-propelled swimmers. But it's more accurate, as others have done, to call penguins "aquatic fliers," prompting a conflation of the "volant tendencies" with "aquatic ones." By feeding into the model for terrestrial bipeds the same assumptions that resolve penguins as bipedally aquatic, they ignore the fact that it's the other pair of limbs doing the ... erm ... leg-work here. This is important, because some other studies have suggested some limb proportions in tetanurines were exapted in maniraptorans for cursorial/volant locomotion. Whether that's true or not, it should be part of the underlying model's assumptions and adjusted to see if the data continues to resolve otherwise.

That should also include counter-indicators: very obviously terrestrial modules in the hindlimbs should render any forelimb-based signal as an absolute false negative. When it comes to Spinosaurus, a lot of these assumptions were dealt with in Hone and Holtz (2021), published over four years before this paper was submitted, and definitely before it was revised, during peer review. Meaning someone, anyone, should have caught that they were making errors in proportions, measurements, etc. And let's face it, given their proportions, I'd also accept Heyuannia yanshini has a better model for an "aquatic" theropod given the forelimb data (counterindicators being apparently irrelevant).

Cheers,

--
You received this message because you are subscribed to the Google Groups "Dinosaur Mailing Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to DinosaurMailingG...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/DinosaurMailingGroup/2d52513e-1ce2-4ce9-aef1-a322e06aaf1bn%40googlegroups.com.

--

Jaime A. Headden

The Bite Stuff: http://qilong.wordpress.com/

"Innocent, unbiased observation is a myth" - P. B. Medawar (1969)

[Mickey Mortimer]

"The reason this is being done is because of penguins"

Nope. They use two penguins, but as with all birds, do not measure their forelimbs. So anything penguin forelimb proportions had to say did not enter their data.

Btw, yanshini gets a ratio of e.g. 0.79 (IGM 100/31) and 0.83 (IGM 100/32), so scores pretty badly on that front. 

Mickey Mortimer