Tuesday, October 6, 2015

The Secret (Fossil) Origin of the TMNT

Yes, even these Ninja Turtles.

What are turtles?

Everybody knows what turtles are. They're those reptiles that have enormous shells. Most turtles live in freshwater lakes and streams, but others are entirely terrestrial (tortoises) and a few are specialized for a marine life. The upper half of the shell is called the carapace and the "belly" shell is called the plastron. They lay a ton of eggs so that a few of the babies will survive to adulthood. Snapping turtles are really bad-ass and can bite your finger off.

"I'm having an identity crisis."
That's all true. But you misunderstand the question--what ARE turtles? What other reptiles are their closest relatives?

This is where you lose most people. They don't really look like lizards, and they're certainly not snakes. Alligators and crocodiles have armor, but not like a turtle. I guess they're mostly aquatic, though, so...maybe alligators?

Well, if you believe molecular studies that have been coming out recently, turtles are either related to lizards and snakes or alligators, which is like...you don't have any other choices among modern reptiles, guys. But anyway, surely the fossil record can tell us something about turtle origins, right? I mean, we've got walking whales now!

Turns out the fossil record has been irritatingly silent on turtle origins. Baur named Proganochelys in 1887, and it is an awfully early turtle--Late Triassic! Unfortunately, Proganochelys is essentially a modern turtle, with a modern shell. Its inability to retract its horned head and armored tail seem like primitive features until you look up the recently-extinct Meiolania, who had the same features. Even the plate configuration of Proganochelys' shell is modern.

From the Greek for "Already a turtle."
The fossil record lacked something like a proto-turtle; a chelonian Archaeopteryx, as it were. Would anyone find a turtle in a "half-shell?"

As it turns out, the answer was yes! In 2008, Li et al. described Odontochelys, a Triassic aquatic Chinese turtle with exactly that--a half-shell. While Odontochelys has a fairly large, nearly modern plastron, it lacks a carapace. Well, that's not entirely accurate. The animal's ribs and vertebrae approach the usual turtle configuration but there is no horny outer armor. In other words, it's a turtle with a typical internal skeleton and an armored plastron, but no armored top half.

The original "hero in a half-shell"
It would've just looked like a really fat lizard with an armored belly. In addition to having teeth (unlike any other turtle), Odontochelys is "primitive" with respect to modern turtles in having a longer snout and a long, fairly flexible tail. However, frustratingly, Odontochelys still doesn't tell us much about turtle origins because, like other fossil and modern turtles, it has an anapsid skull.

Wait, what's that?

Anapsid skulls are opposed to diapsid skulls (or synapsid skulls). In modern and most extinct reptiles, the skull has a nose hole, eye socket, and two more holes: the temporal fenestrae (often mistaken for the "ear hole" sits behind the eye socket, and the supratemporal fenestrae is on the skull roof, also behind the eye sockets. The purpose of these holes is to allow the jaw muscles both more room to attach to bone and space to flex outward. As a result, diapsids have stronger jaw muscles than anapsids. Anapsids don't have any holes behind the eye sockets. Basal synapsids (the ancestors of mammals) have a single hole behind the eye socket--it's usually thought that they developed this hole independently of diapsids.*

A. Anapsid skull; B. Synapsid skull; C. "Eurapsid" skull; D. Diapsid skull
Turtles have anapsid skulls--there are no holes behind the eye sockets. Traditionally, that has convinced paleontologists that turtles are the sole surviving members of an ancient Permian radiation of largely herbivorous beasties like pareiasaurs and marine mesosaurs. However, aside from that anapsid skull, turtles don't share much in common with these "parareptiles." Many workers have compared turtles favorably toward an obscure group of lizard-like parareptiles called millerettids. These tiny animals have slightly broader ribs than would be expected for animals their size and paired belly ribs.

This relationship was advocated as recently as 2013, but as we shall see, things may be considerably more complicated.

Enter tiny Triassic South African fossil Eunotosaurus.

Actually, that's not right. Eunotosaurus has actually been known since 1892, when Seeley reported on a poorly-preserved fossil of the beastie. Eunotosaurus is tiny, but its extremely broad ribs and wide frame suggested a relationship with turtles as far back as 1914, with Watson's "Eunotosaurus africanus Seeley, and the ancestry of the Chelonia" paper, but the idea did not stick. In 1997, Gow redescribed Eunotosaurus based on new, better-preserved fossils and advocated a relationship with millerettids. Once more relegated to parareptile status, it was largely forgotten.

To be fair, it's not much to look at.
And the search for turtle origins continued, frustratingly fruitlessly. During this time, of course, Odontochelys was discovered, which inspired Lyson et al. (2010) to revisit Eunotosaurus.

The authors attempted to place turtles among parareptiles, using Eunotosaurus and Odontochelys as guideposts. They found that Eunotosaurus and turtles (Pan-Testudines) grouped with Milleretta. Odontochelys formed the "missing link" between Eunotosaurus and Proganochelys that was needed to secure an ancestral spot for the former.

While Gow, in his 1997 redescription of Eunotosaurus, believed its wide ribs and stiff trunk were befitting a burrowing animal, Lyson et al. (2013) saw the beginning of the turtle shell when they wrote about the evolution of such a bizarre structure. Before we continue, it might help to know exactly what the turtle shell is made of.

Turtles live inside their ribcages. I mean, we all do, but turtles take it to the extreme. In the egg, turtle babies' ribs expand laterally to the point where they contact each other and grow together. The turtle's belly ribs also fuse together. This process shoves the shoulderblades into an awkward-looking vertical orientation. At the same time, armor plating grows over the ribcage and belly. This armor plating covers the shoulderblades and pelvis, then joins the two halves of the shell along the sides. When you look at a turtle skeleton, you're basically looking at its ribcage--which has been covered with plate-like armor.

That big wall of bone? Those are a turtle's RIBS.
Eunotosaurus appears to have started the internal skeletal process. Like turtles, its ribs are ridiculously wide to the point of contacting and overlapping each other (the overlap may be an artifact of preservation). The only spaces between ribs are up by the vertebrae, where the ribs articulate. This has, predictably, resulted in vertically-oriented shoulderblades as well. The ribs are also T-shaped in cross-section, like modern turtles. Just as interestingly, the marks of muscle attachment to the ribs--called Sharpey's fibers--have almost disappeared from little Eunotosaurus' ribs (they remain on the caudal margins of ribs three, six, and seven).

Eunotosaurus: turtle in training
Apparently big fans of the taxon, Lyson et al. provided further details of Eunotosaurus' proto-shell in 2014, suggesting that it had already developed the specialized respiratory anatomy that modern turtles employ to ventilate their lungs. Non-turtle reptiles use their intercostal (essentially between-rib) muscles and hypaxial muscles to expand and contract their ribcage, thus filling or emptying the lungs. Turtles, however, including Eunotosaurus, had to figure out another solution to this problem, as they dumped their intercostal muscles in favor of a rigid, immobile trunk. Turtles use their hypaxial muscles exclusively (primarily M. tranversus and M. obliquus abdominis) to power their lungs.

(it strikes me that hupehsuchians must have gone through a similar transition)

One wonders what environmental pressures were being exerted on the Eunotosaurus lineage that would lead to a wide, stiff trunk and a wholesale renegotiation of respiratory job duties. At any rate, according to Lyson et al., turtles must have acquired their bizarre ventilation anatomy very early in their evolution--it's probably a defining feature.

But you'll recall that I mentioned at the top of this essay that molecular studies find turtles to be diapsids, sharing certain signals with lizards and/or crocodilians. How do we square that with a parareptilian origin according to paleontological evidence? Well, early in 2015, Schoch & Sues provided a possible answer: maybe turtles really are diapsids after all.

The pair described Pappochelys based on twenty incomplete specimens. Taken together, a complete picture of the tiny lizard-like creature emerges that actually looked a lot like Eunotosaurus but is more turtle-like in a few subtle ways, including the shape and partial co-ossification of the gastralia. The shape of the gastral basket looks a bit like the plastron of Odontochelys but obviously unarmored. As Pappochelys is found in a freshwater lake environment, Schoch & Sues suggest that the plastron (and possibly thickened ribs) evolved as "ballast" for easier swimming. But it's the skull of this "grandfather turtle" that is most impressive: Pappochelys is clearly a diapsid.

Pappochelys: fellow fat lizard thing
The skull preserves small, round supratemporal fenestrae and large, but (and this is crucial) ventrally open postorbital fenestrae. That is, the jugal bone does not contact the quadratojugal. The hole is not a hole--it lacks a lower border. This is actually similar to--but obviously independently derived from--modern lizards, many of whom have also lost that particular bony contact. Schoch & Sues further provide a phylogenetic analysis for a diapsid origin of turtles, They conclude that, perhaps not surprisingly, Pan-Testudines is a sister group to the Sauropterygia--a massive clade of marine reptiles that includes other broad-ribbed, armored animals like Sinosaurosphargis, placodonts, and possibly hupehsuchians!

This is exciting, because it means that many of the treats considered so strangely unique to turtles, like broad ribs and armored skin, make more sense in the context of their closest relatives. Turtles finally have a home to call their own among the rest of Sauropsida (reptiles).

Just recently, Bever et al. (2015) returned to Eunotosaurus in an effort to better understand that animal's skull, which had not (until now) been given a proper description. Using high-resolution X-rays and digital reconstructions of the skull bones of several relatively well-preserved skulls, the authors were able to create impressive cranial models of Eunotosaurus. In fact, Eunotosaurus displays the same open postorbital fenestrae that Pappochelys does. However, it displays a curious growth marker: young animals have small supratemporal fenestrae that actually close up with age due to growth of the supratemporal bone. Whether this also happens in Pappochelys is not known.

A few more years and we won't even have to dig anymore.
However, Bever et al. did not find the same turtle-sauropterygian connection that Schoch & Sues did, instead finding turtles to be the basalmost clade of living reptiles. They do suggest that turtles closed their postorbital fenestrae by rearward expansion of the jugal that met a forward expansion of the quadratojugal. Closure of the supratemporal fenestrae may have simply been a quirk of getting older in the turtle lineage (although I tend to think it's a quirk of Eunotosaurus).

So there you have it, folks. Turtles are diapsids, possibly close to sauropterygians, and they are just the weirdest.

*It's possible that the synapsid condition was ancestral, and diapsids developed a second hole on their own. In this scenario, anapsid "reptiles" are either not actually reptiles but instead animals that evolved prior to the synpasid-diapsid split. They could also be diapsids that, for whatever reason, reversed course and walled-off their skulls.

Bever, G. S., Lyson, T. R., Field, D. J. & Bhullar, B. S. (2015). Evolutionary origin of the turtle skull. Nature, doi: 10.1038/nature14900

Lyson, T. R., Bever, G. S., Bhullar, B. S., Joyce, W. G. & Gauthier, J. A. (2010). Transitional fossils and the origin of turtles. Biology Letters 6: 830-833.

Lyson, T. R., Schachner, E. R., Botha-Brink, J., Scheyer, T. M., Lambertz, M. Bever, G. S., Rubidge, B. S. & de Queiroz, K. (2014). Origin of the unique ventilatory apparatus of turtles. Nature Communications 5:5211 doi: 10.1038/ncomms6211 (2014).

Lyson, T. R., Bever, G. S., Scheyer, T. M., Hsiang, A. Y. & Gauthier, J. A. (2013). Evolutionary origin of the turtle shell. Current Biology, hppt://dx.doi.org/10.1016/j.cub.2013.05.003.

Schoch, R. R. & Sues, H. (2015). A Middle Triassic stem-turtle and the evolution of the turtle body plan. Nature, doi: 10.1038/nature14472.


  1. So this Sauropterygia + Pan-Testudines clade falls within Archosauromorpha, correct?

  2. Yes; Archosauromorpha is still a separate group.

  3. Bever et al. did find the turtle-sauropterygian connection in their Maximum Parsimony tree, just not in their Baysian tree. Pedantic phylogenetic nomenclature point- saying "Pan-Testudines is a sister group to the Sauropterygia" is incorrect since there are no living sauropterygians. In this phylogeny, sauropterygians would BE Pan-Testudines. Note Schoch and Sues' Maximum Parsimony results are like Bever et al.'s, but in their Bayesian analysis Eunotosaurus is down by millerettids.

    Before this post I didn't realize Eunotosaurus was so small. I had always pictured it as a meter or two long.