Updated July 1, 2015 with a tracing of the holotype of Pappochelys (Fig. 6). See July 1, 2015 for an update on Pappochelys.
The following notes demonstrate
the great capacity of unrelated reptiles to converge on character traits, in this case, expanded ribs and other traits. In such cases, only a large, species/specimen-based phylogenetic analysis, like the large reptile tree, can resolve such problems with great confidence, parsimony and logic. Otherwise, as in the case of Pappochelys (pah-poe-kee-luss), results can be frustrating (see below).
Fiigure 1. The turtle mimic Eunotosaurus from the Middle Permian was actually closer to Acleistorhinus.
Yesterday, a new paper in Nature
by Schoch and Sues (2015) purported to document the transitional taxon between the derived millerttid, Eunotosaurus (Fig. 1), and the basal turtle, Odontochelys (Fig. 2). They employed two cladograms (Figs.1, 2) based on Lyson et al. 2010. Both recovered topologies that are not supported by the large reptile tree. Both employ several suprageneric taxa, always a bad sign.
In the large reptile tree, now with 556 taxa, Eunotosaurus and Odontochelys are not closely related. On that note, Schoch and Sues report in their own testing, a TNT analysis (Fig. 3) produced a tree topology distinct from their own Bayesian analysis (Fig. 4), especially with regard to their key taxon, Eunotosaurus (Fig.1), which nested far from turtles in the Bayesian analysis.
Figure 2. Odontochelys, a basal soft-shelled turtle with teeth.
Odontochelys (Fig. 2) is indeed a basal turtle.
It nests with Trionyx, the extant soft-shelled turtle in the large reptile tree, so it is not as primitive as others suggest. It shared a common Early to Middle Permian ancestor with Elginia and Sclerosaurus, two more primitive horned turtle sisters (Fig. 7). Elginia nests with the giant horned turtle, Meiolania as reported earlier. Sclerosaurus had a broad flat torso with discrete osteoderms prior to carapace formation. This is how the carapace had its genesis according to the large reptile tree (Fig. 7).
Figure 3. from Schoch and Sues 2015 with colors added here to denote clades recovered by the large reptile tree. This is their TNT analysis result.
The Schoch and Sues abstract
described the 220 million-year-old, Late Triassic, Odontochelys as having a ‘partly formed shell’, but the large reptile tree nested it with the living soft shell turtle, Trionyx, so the structure was derived, not primitive. So turtles are more ancient than the Late Triassic.
Schoch and Sues listed the 260-million-year-old Eunotosaurus as a hypothetical stem turtle, but it actually nests with Acleistorhinus and Delorhynchus, convergent with turtles in several respects.
Schoch and Sues considered the new reptile, Pappochelys rosinae (“grandfather-turtle”, 20 cm in length, 240 mya, Ladinian, Middle Triassic; SMNS 91360, SMNS 90013 and other referred specimens, including a very small individual), intermediate between Eunotosaurus and Odontochelys (but only in their TNT analysis, Fig. 3).
Figure 4. Second cladogram recovered by Schoch and Sues 2015 recovered by Bayesian analysis. The use of suprageneic taxa is always dangerous due to cherry picking and taxon exclusion. Note where Eunotosaurus (in pink) nests here.
From the Schoch and Sues abstract: “The three taxa [Eunotosaurus, Pappochelys and Odontochelys} share anteroposteriorly broad trunk ribs that are T-shaped in cross-section and bear sculpturing, elongate dorsal vertebrae, and modified limb girdles. Pappochelys closely resembles Odontochelys in various features of the limb girdles. Unlike Odontochelys, it has a cuirass of robust paired gastralia in place of a plastron. Pappochelys provides new evidence that the plastron partly formed through serial fusion of gastralia. Its skull has small upper and ventrally open lower temporal fenestrae, supporting the hypothesis of diapsid affinities of turtles.”
Their analysis, based on Lyson et al. 2010,
included 198 character traits (originally 191) and generated a single MPT.
They had to add 7 traits to achieve their results
When the original data set (191 characters) was analysed using TNT, with scores for Pappochelys included, the analysis yielded three MPTs that differed in the positions of Archosauriformes, Prolacerta, and rhynchosaurs, as well as of kuehneosaurids, lepidosaurs, and the turtle-sauropterygian clade. That’s several big changes! I applaud them for their honesty. They report, in that analysis, Pappochelys was found to nest below Eunotosaurus, but still within a clade with turtles.
In the large reptile tree
deletions and addition don’t produce that sort of anarchy and large changes in tree topology.
Schoch and Sues report,
“Robustness of nodes was assessed by bootstrap, resulting in collapse of many nodes, including Diapsida and the placement of Eunotosaurus at the base of the turtle clade.”
If they only had the large reptile tree to work with, this would not have happened.
Schoch and Sues also note, |
“Although the trunk region is disarticulated in all available specimens, the maximum number of trunk vertebrae did not exceed nine.” Since each specimen was incomplete, I wonder how they came up with that number? … except that Eunotosaurus and turtles have a short dorsal series with long vertebral centra. …or no partial specimen had more than nine scattered vertebrae preserved (typically far fewer). Based on the varying sizes and shapes of the dorsal ribs, it would appear that more ribs would be necessary to fill in the shape gaps, and along with more ribs you need more vertebrae (Fig. 6). In the large reptile tree recovered sister taxa among basal enaliosaurs (Figs. 5-7) have far more than nine dorsal vertebrae.
Schoch and Sues further note,
“In ventral view, the anterior gastralia extend anterolaterally, whereas the reverse obtains on the posterior gastralia. None of the available fossils preserves undisturbed pairs of gastralia.” (Fig. 6). Not sure how Schoch and Sues came to this conclusion, based on the evidence they presented, except that appears to be the pattern in Odontochelys (Fig. 2). I know of no other examples where this also happens. Note in the related placodonts, Paraplacodus (Fig. 6) and Placocodus, the lateral gastralia tips point dorsally and crushing could have produced such a pattern as interpreted by Schoch and Sues. I hope they weren’t trying to force fit an interpretation to disarticulated remains.
Figure 5. Pappochelys skull reconstructed from colorized bone images compared to sister taxa including Palatodonta and the original reconstruction of Schoch and Sues. Pappochelys certainly looks like Palatodonta and Paraplacodus, but not Odontochelys. Note the very narrow frontals, totally unlike turtles, totally like Palatodonta.
In the Bayesian analysis
Schoch and Sues reported, “An unexpected result was the (albeit weakly supported) traditional placement of Eunotosaurus among Parareptilia and completely separate from Pappochelys, Odontochelys and Testudines, all of which were recovered as the sister-group of Sauropterygia among Diapsida. Pappochelys was firmly recovered as the sister-taxon to Odontochelys + (Proganochelys + Testudines).”
Figure 6. Pappochelys compared to placodont sister taxa and compared to the Schock and Sues reconstruction, which appears to have several scale bar errors and underestimated the number of dorsal vertebrae. Click to enlarge. So few ribs and vertebrae are known for Pappochelys that their order, size and number could vary from that shown here. Note the ribs of Paraplacodus are also expanded. The number of dorsal vertebrae is unknown and probably more than nine based on sister taxa. The pectoral girdle of Pappochelys is from several specimens.
In figure 6
note the relatively large pelvis, short torso and long legs in the Schoch and Sues version of Pappochelys. Those proportions approach those of speedy terrestrial reptiles, not what one would expect of turtle ancestors. I think their estimates were off. Certainly their scale bars were off, unless the measurements were taken from several different ontogenetic age specimens. The Schoch and Sues reconstruction also overlooks the great variety in rib shapes and sizes in Pappochelys. While creating the reconstruction I also had trouble reconciling the scale bars with their reconstruction in which certain elements are twice or half what they should be. Compare skull sizes to pelvis sizes in their reconstruction vs. mine.
Throughout the Schoch and Sues paper
the authors make note of similarities between Pappochelys, Eunotosaurus and Odontochelys.
- The large ribs bear sculpting on the dorsal surface, suggestive an intradermal origin.
- The dorsal ribs are T-shaped in cross-section
- The scapula is tall and slender
- The pelves closely resemble each other
- The pubis has a lateral process
- The S-shaped femur has an internal trochanter and an offset head.
But ALSO note that in both Schoch and Sues studies
eusaurosphargids and placodonts nest as sisters to the turtles. Schoch and Sues celebrate the fact that Pappochelys had a diapsid skull even though no turtles have temporal fenestra. Turtles have only nested with diapsids in phylogenetic analyses based on molecular data. Their interpretation of Pappochelys. therefore, comes as something of a wonderful surprise in that it appears to tie morphological and molecular study findings together. To their credit, Schoch and Sues report that those molecular studies typically nest turtles with or close to archosaurs. We all agree that on the face of it such a nesting is out of the question. No morphological study has ever replicated that result.
The authors suggest that Eunotosaurus had upper temporal openings concealed by large supratemporals. The reader is probably already aware that no sister taxa of Eunotosaurus have upper temporal fenestrae and that if a bone covers an opening, it is no longer considered an opening.
Figure 7. Two subsets of the large reptile tree focusing on Pappochelys and its enaliosaur relatives (left) and turtle relatives (right). Shifting Pappochelys to turtles adds 37 steps. Click to enlarge.
in the large reptile tree recovered a nesting close to the basal placodont, Palatodonta (skull only) and the much larger Marjiashanosaurus (post-crania only). Paraplacodus is not far removed and it has ribs with a T-shaped cross-section. Sinosaurosphargis is shaped like a turtle with a carapace and plastron of flat gastralia and it nests close by. Largocephalosaurus nests nearby and it has a tall slender scapula and a pubis with a lateral process. So reptiles near that node were experimenting with several turtle traits by convergence with actual turtles.
Of great interest in Pappochelys
is the lack of elongate dorsal transverse processes, common to eusaurosphargid and placodont sister taxa. However Anarosaurus and Pachypleurosaurus are also sisters and they, like Pappochelys and turtles, also lack elongate dorsal transverse processes.
Fingers and toes
Like Eunotosaurus, Pappochelys has relatively slender fingers and toes, unlike those of turtle and their true ancestors, like Sclerosaurus. But that’s okay, because Eunotosaurus and Pappochelys are not related to turtles.
As noted above, nearly every turtle-like trait found in Pappochelys can be found in pachypleurosaurs, eusaurosphargids and placodonts. There is no doubt that Pappochelys evolved several turtle-like traits. Unfortunately, parsimony reveals that it was not a turtle, but developed those traits by convergence. I understand the excitement that Schoch and Sues must have felt about their discovery and its apparent importance. No wonder Nature wanted to publish it. But just like Limusaurus and Yi qi, more prosaic mundane explanations and interpretations are recovered when more taxa are included in analysis.
Revisiting the new Pappochelys
If Pappochelys had the same number of dorsal vertebrae as its sister taxa, then a new, long-bodied reconstruction emerges (Fig.6). Here we have an elongate, aquatic reptile without specialized teeth. It has relatively short, weak legs and a wider than deep torso with pachystotic bones. With such traits, Pappochelys could have been a bottom-dweller in a shallow lake environment. Large eyes might have given it good night vision.
Now that we have two of these short-snouted, big-eyed placodonts, perhaps we can discard the false idea that Palatodonta was a juvenile. Rather, as in many other novel reptile clades, phylogenetic miniaturization accompanied the development of new body parts and character traits.
For the large reptile tree origin of turtles, click here and here.
The Pappochelys strata
were laid down in a shallow oligohaline or freshwater lake. It is the most common taxon in the Vellberg lake deposit and is represented by several growth stages. The authors consider Pappocehelys “critical evidence for the diapsid relationships of turtles and it provides a new stage for the evolution of the turtle body plan.”
Pappochelys is a basal placodont, unrelated to turtles.
However, Pappochelys is important to the large reptile tree because it ties a skull-only taxon (Palatodonta) to a skull-less taxon (Majiashanosaurus). So the tree is once again fully resolved, an unforeseen side-effect.
Added a day later: lots of news online about Pappochelys, some with audio
Science Magazine – reports, “So having broad, dense bones and gastralia would have acted like a diver’s weight belt, helping Pappochelys fight buoyancy and forage on the lake’s bottom. But these bones would also have had a beneficial side effect: They would have offered some degree of protection from predators, such as large amphibians or fish living in the lake, by deflecting or blunting their bites.”
Lyson TR, Bever GS, Bhullar B-AS, Joyce WG and Gauthier JA. 2010. Transitional fossils and the origin of turtles. Biology Letters 2010 6, 830-833 first published online 9 June 2010. doi: 10.1098/rsbl.2010.0371
Schoch RR and Sues H-D 2015. A Middle Triassic stem-turtle and the evolution of the turtle body plan. Nature (advance online publication) > doi:10.1038/nature14472 online