SVP abstracts – Are meiolaniform turtles stem turtles?

Kear et al. 2019 talk about
‘stem’ turtles with skull horns and club tails: the meiolaniforms.

From the abstract:
“Meiolaniforms (Meiolaniformes) are an enigmatic radiation of stem turtles with an exceptionally protracted 100 million-year evolutionary record that spans the mid-Cretaceous (Aptian–Albian) to Holocene. Their fossils have been documented for over 130 years, with the most famous examples being the derived Australasian and southern South American meiolaniids – bizarre horned turtles with massive domed shells and tail clubs that are thought to have been terrestrial and probably herbivorous.”

In the large reptile tree (LRT, 1592 taxa, subset Fig. 2) meiolaniforms (Fig. 1) are not enigmatic. They are basalmost hard-shell turtles derived from similarly-horned Elginia-type small pareiasaurs in parallel with Sclerosaurus-type small pareiasaurs basal to soft-shell turtles.

Figure 2. Another gap is filled by nesting E. wuyongae between Bunostegos and Elginia at the base of hard shell turtles in the LRT.

Figure 2. Another gap is filled by nesting E. wuyongae between Bunostegos and Elginia at the base of hard shell turtles in the LRT.

“Despite a long history of research, the phylogenetic affinities of meiolaniids have proven contentious because of ambiguous character state interpretations and incomplete fossils
representing the most ancient Cretaceous meiolaniform taxa.”

This problem is contentious only because of taxon exclusion. Prior workers have not included analyses of meiolaniforms and Elginia.

“Here, we therefore report the significant discovery of the stratigraphically oldest demonstrable meiolaniform remains, which were excavated from Hauterivian–Barremian high-paleolatitude (around 80°S) deposits of the Eumeralla Formation in Victoria, southeastern Australia. Synchrotron microtomographic imaging of multiple virtually complete skulls and shells provides a wealth of new data, which we combine with the most comprehensive meiolaniform dataset and Bayesian tip-dating to elucidate relationships, divergence timing and paleoecological diversity.”

Did the authors include Elginia, Sclerosaurus, Arganceras and Bunostegos? The abstract does not mention them.

“Our results reveal that meiolaniforms emerged as a discrete Austral Gondwanan lineage,
and basally branching sister group of crown turtles (Testudines) during the Jurassic.”

The LRT invalidated a monophyletic Testudines. Rather soft-shell and hard-shell turtles had separate parallel origins from within the small horned pareisaurs.

Figure 5. Subset of the LRT focusing on turtle origins and unrelated eunotosaurs.

Figure 5. Subset of the LRT focusing on turtle origins and unrelated eunotosaurs.

“We additionally recover a novel dichotomy within Meiolaniformes, which split into a unique Early Cretaceous trans-polar radiation incorporating apparently aquatic forms with flattened shells and vascularized bone microstructure, versus the larger-bodied terrestrial meiolaniids that persisted as Paleogene–Neogene relic species isolated in Patagonia and Australasia.”

That’s interesting. The LRT sort of separates the meiolaniform Niolama from the meiolaniform Meiolania + Proterochersis + Proganochelys. The latter taxon also has a club tail. Perhaps more meiolanforms would continue to nest with one or the other.

“Finally, our analyses resolve the paraphyletic stem of crown Testudines, which otherwise includes endemic clades of Jurassic–Cretaceous turtles distributed across the northern Laurasian landmasses. These had diverged from the Southern Hemisphere meiolaniforms by at least the Middle Jurassic, and thus parallel the vicariant biogeography of crown turtles, which likewise diversified globally in response to continental fragmentation and possibly climate.”

Outgroups are key to understanding turtle evolution in the LRT. So is taxon inclusion. Based on the dual origin of turtles from horned small pareiasaurs in the LRT, the list of stem turtles now includes pareiasaurs, if the concept of a monophyletic turtle still stands with a last common ancestor lacking a carapace and plastron within the pareiasaurs.


References
Kear BP et al. 2019. Cretaceous polar meiolaniform resolves stem turtle relationships. Journal of Vertebrate Paleontology abstracts.

Eorhynchochelys: a giant eunotosaur, not a stem turtle

Figure 1. Skull of Eorhynchochelys sinensis with DGS colors applied to bones. These differ somewhat from the original bone drawing.

Figure 1. Skull of Eorhynchochelys sinensis with DGS colors applied to bones. These differ somewhat from the original bone drawing. This is a standard eunotosaur skull, not a pareiasaur or turtle skull. I see tiny premaxillary teeth, btw.

Li, Fraser, Rieppel and Wu 2018
introduce Eorhynchochelys sinensis (Figs. 1,2), which they describe in their headline as ‘a  Triassic stem turtle’ and in their abstract as ‘a Triassic turtle.’ Unfortunately, Eorhynchochelys is not related to turtles. Instead it is a spectacular giant eunotosaur (sister to Eunotosaurus).

Figure 2. Eorhynchochelys in situ alongside manus, pes, pectoral and pelvic girdle, plus Eunotosaurus to scale. By convergence Eorhynchochelys resembles Cotylorhychus.

Figure 2. Eorhynchochelys in situ alongside manus, pes, pectoral and pelvic girdle, plus Eunotosaurus to scale. By convergence Eorhynchochelys resembles Cotylorhychus.

The problem is, once again, taxon exclusion.
Li et al. employed far too few taxa (Fig. 3) and no pertinent turtle ancestor taxa (see Fig. 4).

Figure 4. Cladogram of turtle relationships by Li et al. 2018. Yellow-green areas are lepidosauromorphs in the LRT demonstrating the mix of clades present here.

Figure 3. Cladogram of turtle relationships by Li et al. 2018. Yellow-green areas are lepidosauromorphs in the LRT demonstrating the mix of clades present here due to massive taxon exclusion. The LRT has 40x more taxa.

We know exactly from which taxa turtles arise.
In the large reptile tree (LRT, 1271 taxa, Fig. 4): 1) hard shell turtles arise from the small, horned pareiasaur, Elginia. The basalmost hard shell turtle is Niolamia, not Proganochelys. 2) soft shell turtles arise from the small, horned pareiasaurs, Sclerosaurus and Arganaceras. The basalmost soft shell turtle is Odontochelys. None of these taxa have temporal fenestrae. We looked at turtle origins earlier here. Turtle origins were published online in the form of a manuscript earlier here.

Figure 5. Subset of the LRT focusing on turtle origins and unrelated eunotosaurs.

Figure 4. Subset of the LRT focusing on turtle origins and unrelated eunotosaurs.

Unrelated
Pappochelys nests with basal placodonts. Eunotosaurus nests with the caseid clade, close to Acleistorhinus and Australothyris, all taxa with a lateral temporal fenestra. Li et al. suggested that this lateral temporal fenestra indicated that turtles were diapsids. That has been falsified by the LRT which shows that turtles never had temporal fenestra all the way back to Devonian tetrapods.

Eorhynchochelys sinensis (Li et al. 2018; Late Triassic) was considered the earliest known stem turtle with a toothless beak, but here nests as a giant aquatic eunotosaur with tiny premaxillary teeth. In size and overall build it converges with Cotylorhynchus.

References
Li C, Fraser NC, Rieppel O and Wu X-C 2018. A Triassic stem turtle with an edentulous beak. Nature 560:476–479.

Would you like to read a rejection notice, or two?

In the past week
I submitted a comment to Royal Society Proceeedings B on Foth and Joyce 2017. In it I suggested that the origin of turtles was diphyletic and that would affect the placement of the basalmost turtle in the work of Foth and Joyce.

Referee number 1 wrote:
“This paper is unsuitable for Proceedings B (or any scientific journal) and should be rejected. It is ostensibly a response to a recent paper by Foth & Joyce on the disparity of the turtle skull over time, but in reality it doesn’t address this study at all, but is a back-handed attempt by the author to publish an iconoclastic phylogenetic analysis based on an inadequate dataset riddled with errors and methodological flaws. Sorry, there is no way to be kind about this manuscript.”

Referee number 2 (Walter Joyce, one of the original authors) wrote:
“the attached manuscript by David Peters is a response to an article I published earlier this year with Christian Foth in Proceedings B regarding the evolution of cranial disparity in turtles (Foth and Joyce 2017). Although I welcome any scientific debate regarding this paper, I would like to suggest outright rejecting this contribution for one single reason: It is an open trade secret that David Peters has been developing an enormous phylogeny of reptiles that produces highly outlandish results. One such outlandish result is the polyphyletic origin of turtles. This undertaking has been submitted to many journals over the years and has been rejected every time, as basic tenants of sound cladistic analysis are not followed therein, mostly an adherence to the use of character observations that can be reproduced by people who are not David Peters. I am certain that countless scientists invested countless hours in providing sound arguments why this tree should be rejected and I will therefore save myself the work here. If anything, this phylogeny should receive full peer review in a standalone publication, and not be slipped into the sphere of published scientific literature as part of a not-quite appropriate criticism of Foth and Joyce (2017).”

And here is my reply to the editors:
“Critical thinking is a requirement in science and I’ve had a few hours now to critically think about the replies I received from the two referees. I hope these comments will help you in future endeavors.

1. You already know that referees should be unbiased when they approach a manuscript. Asking Dr. Joyce to be a referee runs counter to that ideal. After all, I was commenting on his paper. His comments should have been requested only after two unbiased referees had ok’d the manuscript for publication.

2. Some referees like to accept manuscripts knowing ahead of time they will reject them. Is there any method you use to prevent this?

3. Whenever I review a manuscript I review some of the details within the manuscript, pointing out errors, if any, congratulating insights, if any. This was not done by either referee. There is no indication that either referee actually read the manuscript, let alone tested the hypotheses that resulted with the matrix provided.

4. The paper was about taxon exclusion. Foth and Joyce excluded taxa pertinent to the origin of turtles, which affected their basalmost taxon and the rest of their phylogram. That point was ignored by both referees who described ‘an inadequate data set’ (did they actually see the dataset, or go by rumors?). No specifics were put forth. No testing of the analysis was described. That’s what I do in such cases. I run the matrix looking for mismatches. Anyone who has the same taxon list, no matter what their character list, will come to the same results as I did, unless they omit certain pertinent taxa, as Foth and Joyce did.

5. Joyce wrote: “It is an open trade secret that David Peters has been developing an enormous phylogeny of reptiles that produces highly outlandish results.”

To that point, many results of my studies follow traditional topologies: birds nest with birds, turtles with turtles, etc. When topologies shift it is virtually always because the large size of the cladogram allows taxa that have not been tested together to be tested together. That the results upset untested traditions and paradigms are THE reason why this work should be published. The origin of turtles could have been known for the last fifty years. I just included taxa that were previously excluded.

Joyce may be upset because i pointed out this oversight, after all the hours he put into his project. That’s never welcome news, especially when that correction comes from someone without a PhD. It is potentially embarassing. Nevertheless, even if the hypotheses comes from an obscure patent clerk, this is how we build our science. The present facts should be central to the case, not any disparaging rumors about the scientist.

The data presented has to be good. Otherwise there is no way for the cladogram to have high Bootstrap scores throughout. The software is unbiased with regard to output. Unfortunately, pride, shame and other emotions are involved here when it comes to the referees. Some don’t like change.

Thank you for reading this. I don’t ask for any revision to the status of my manuscript, only that you review your policies so bias does not influence the next few incoming manuscripts.

Best regards,”

Phylogenetic origin of the turtle plastron and hypoischium

Several prior workers
have attempted to explain the origin of the turtle carapace. By contrast, the plastron has been largely ignored (please let me know if otherwise), except by Rice et al. 2016, who looked at developing embryos of the pond slider, Trachemys, rather than extinct taxa. They found that condensates for each plastron bone [form] at the lateral edges of the ventral mesenchyme,” like sternal cartilage development in chicks and mice, but with the suppression of cartilage and a bias toward bone development.

Unfortunately,
Rice et al. bought into the invalid hypothesis that Pappochelys (misspelled ‘Pappachelys‘ in their paper) was related to turtles. They also mention the undocumented ‘gastralia hypothesis’ of plastron origin. However Rice et al. report, “whereas plastron bones start to mineralize from the periphery of the ventrum in a slight anterior-to-posterior preference, gastralia mineralize in a posterior-to-anterior sequence….”

The plastron in most modern turtles
is composed of nine bones (listed below) that develop between the visceral organs and ectodermal scutes. Four more appear only in the basal soft-shell turtle, Odontochelys (Fig. 1, discussed below).

In the large reptile tree (LRT, 1042 taxa) the proximal ancestors of both soft shell and hard shell turtles lack gastralia or a plastron. By contrast, all turtles from both clades have a plastron. (Yes, it is odd that so many traits developed in parallel in the two clades, but attests to the authority of the LRT that it is able to lump and separate the two clades.)

The soft shell turtle plastron
first appears in the fossil record in lake deposit specimens of the Late Triassic Odontochelys (Fig. 1). Its current proximal ancestor, Middle Triassic Sclerosaurus (Fig. 9) has no gastralia or plastron, but it does appear to have a hypoischium (novel ventral bone posterior to the ischium).

Typically the turtle plastron consists of
four sets of bones.

  1.  Anteriorly the former clavicles and interclavicle appear beneath the neck where they are renamed the epiplastra and entoplastron.
  2. Further back the hyoplastron rims the forelimbs.
  3. Posteriorly the hypoplastron rims the hind limbs.
  4. Approaching and sometimes beneath the pelvis are the xiphiplastra.

Odontochelys has two extra sets of plastra not found in extant taxa. The two mesoplastron sets are located between the hyoplastra and hypoplastra. They appear to be new structures unique to this genus given that no other known turtles have them.

FIgure 1. GIF animation of the plastron of Odontochelys. Note it only extends to the anterior pelvis. Following the pelvis is another new ventral plate, the hypoischium.

FIgure 1. GIF animation of the plastron of Odontochelys. Note it only extends to the anterior pelvis (Pu + Is). Following the pelvis is another new ventral plate, the hypoischium.

The most primitive (but not the oldest) hard shell plastron
appears in late-surviving Meiolania (Fig. 2). Proximal outgroup taxa from the Late Permian, either don’t preserve a post-crania (Elginia) or lack belly bones (Bunostegos). In the more derived Late Triassic Proganochelys and Proterochersis, the central hole is filled with bone.

Figure 2. The plastron from two specimens of Meiolania. Note the large hole in the center and the nearly complete lack of any bone shape in common with the plastron bones of Odontochelys (Fig. 1).

Figure 2. The plastron from two specimens of Meiolania. Note the large hole in the center and the nearly complete lack of any bone shape in common with the plastron bones of Odontochelys (Fig. 1).

The plastron of hard shell turtles
apparently developed in convergence with the plastron of soft shell turtles (no last common ancestor has a plastron). In basal taxa the structures are distinct from one another (Figs. 1, 2), but derived taxa converge on one another.

The soft shell plastron bones in Odontochelys
(Fig. 1) appears to radiate from the center extending to fragile lateral connections to the carapace. Note: Rice et al. did not observe any developing soft-shell turtle embryos so what they learned from Trachemys (see above) may or may not be applicable to soft shell clade.

By contrast the hard shell plastron
of Meiolania has a strong lateral connection to the carapace, underlaps the pectoral and pelvic girdles, and avoids the center. So each plastron essentially rims each limb opening. The plastra of Meiolania appear to be fused to one another, but that is not the case with other hard shell taxa (see below).

Figure 5. Meiolania, the most primitive of known turtles, has lateral forelimbs, like non turtles.

Figure 3. Meiolania, the most primitive of known hard shell  turtles, has lateral forelimbs, like non turtles. The plastron covers most of the pelvis. The neck could not be withdrawn beneath the carapace. The plastron had a large central fenestra lacking in the plastron of Odontochelys (Fig. 1). Remember, this is a model, not the actual bones.

Triassic Proganochelys
(Fig. 4) fills the central hole in the plastron and it has a hypoischium posterior to its pelvis, as seen in Odontochelys and Sclerosaurus. It’s too bad Elginia and Bunostegos preserve the post-crania so poorly. We should be able to find a hypoischium in their remains, too. Since Meiolania has never been described with a hypoischium, we should go look for it (see below).

Figure 3. The plastron of Proganochelys is solid, and is solidly connected laterally, but it also has a hypoischium posterior to the ischium and the plastron barely underlaps the pelvis.

Figure 4. The plastron of Proganochelys is solid, and is solidly connected laterally, but it also has a hypoischium posterior to the ischium and the plastron barely underlaps the pelvis.

And now, just to make things more confusing…
Compared to Odontochelys, the extant soft shell turtle, Trionyx (Fig. 5), has a reduced plastron with central fenestrae. The two midplastra are absent here. So is any ossification along the midline, convergent with hard shell turtles. The interclavicle and clavicles are not co-ossified. It’s as if ossification ceased at a certain point in the development of the plastron here.

Figure 2. Some parts of the soft-shell turtle plastron have their origins in the interclavicle and clavicle of other tetrapods. Other parts are not modified gastralia because outgroups do not have gastralia.

Figure 5. Some parts of the soft-shell turtle plastron have their origins in the interclavicle and clavicle of other tetrapods. The carapace is also shown here.

Likewise,
hard shell sea turtles, like Chelonia, do not fully ossify the plastron. Here (Fig. 6) none of the plastron elements are co-ossified. The hyoplastra and hypoplastra appear to radiate from four centers. The radiations likely point to their origins in the center of each plate. The posterior xiphiplastra likewise radiate but in a narrower pattern.

FIgure 7. Sea turtle plastron. This looks like a soft shell turtle plastron.

FIgure 6. Sea turtle plastron. Bone development ceased prior to suturing.

The predecessor to soft shell turtles, Sclerosaurus,
is known from a nearly complete and articulated skeleton (Fig. 7) that appears to preserve no plastron, but has the genesis of a hypoischium. The flexible spine composed on more than ten dorsal vertebrae and ribs was probably stiffened and reduced prior to the invention of the plastron, but some dorsal osteoderms are present along the midline.

Figure 8. Sclerosaurus insitu. This turtle ancestor still bas a flexible spine, but the pectoral girdle has migrated anterior to the dorsal ribs. A hypoischiuum is present.

Figure 7. Sclerosaurus insitu. This turtle ancestor still bas a flexible spine, but the pectoral girdle has migrated anterior to the dorsal ribs. A hypoischiuum is present.

A reconstruction of Sclerosaurus
(Fig. 8) shows the migration of the much shorter scapula anterior to the dorsal ribs and the first appearance of the hypoischium. The scapula shift is the first step toward tucking the pectoral girdle beneath the anterior dorsal ribs.

Figure 9. Sclerosaurus reconstructed. Note the placement of the narrow pectoral girdle anterior to the wide dorsal ribs.

Figure 8. Sclerosaurus reconstructed. Note the placement of the narrow pectoral girdle anterior to the wide dorsal ribs. The supratemporal horns are homologous with those of Elginia and Meiolania.

FIgure 10. Trachemys plastron and diagram. The scutes overlap the bones. The bones are impossible to understand without the diagram because they retain the impressions of the scutes.

FIgure 9. Trachemys plastron and diagram. The scutes overlap the bones. The bones are impossible to understand from photos such as this one without the diagram because the bones retain the impressions of the scutes.

Figure 10. The unidentified bone from Gaffney 19xx here imagined as the half of hypoischium attached to the posterior ischium.

Figure 10. The unidentified bone from Gaffney 1996 here imagined as the half of hypoischium attached to the posterior ischium.

Did Meiolania have a hypoischium?
Gaffney 1996 did a fantastic job of reconstructing Meiolania (Fig. 3) from bits and pieces, including a xiphiplastron from over a dozen broken bits. He also published what he called an ‘unidentified bone’ (Fig. 10). If turtle expert Gaffney was not able to identify it, I wonder if it was an unexpected bone, like a hypoischium? Let’s leave that as a big maybe for now…

References
Gaffney ES 1996. The postcranial morphology of Meiolania platyceps and a review of the Meiolaniidae. Bulletin of the AMNH no. 229.
Rice R et al. 2016. Development of the turtle plastron, the order-defining skeletal structure. PNAS 113(19):5317–5322.
.

Meiolania eggs confirm basal turtle status

Earlier the horned turtles, Meiolania and Niolamia, were nested in the large reptile tree (LRT) as basalmost hardshell turtles, closely related to the toothed horned stem turtle/pareiasaur, Elginia. This was heresy when introduced.

Now
newly discovered turtle eggs (Lawver and Jackson 2016) add evidence to the basal status of Meiolania.

From the Lawver and Jackson 2016 abstract:
“A fossil egg clutch from the Pleistocene of Lord Howe Island, Australia that we assign to Testudoolithus lordhowensis, oosp. nov. belongs to the stem turtle Meiolania platyceps.  Thin sections and scanning electron microscopy demonstrate that these eggs are composed of radiating acicular aragonite crystals. This mineral composition first evolved either before the split between Meiolaniformes and crown Testudines or prior to Proterochersis robusta, the earliest known stem turtle. Meiolania platyceps deposited its eggs inside an excavated hole nest. This nesting strategy likely evolved no later than the Early to Middle Jurassic.”

All known meiolanids
are from later, higher Late Cretaceous and Tertiary strata.

Figure 5. Meiolania, the most primitive of known turtles, has lateral forelimbs, like non turtles.

Figure 1. Meiolania, one of the most primitive of known hard-shell turtles, has lateral forelimbs, like non turtles. All extant turtles have anteriorly-directed humeri. It also had cranial horns, like the toothed pareiasaur/turtle? Elginia.

At present,
soft-shell and hard-shell turtles have a dual origin from separate small Late Permian and Middle Triassic pareiasaur ancestors, Elginia and Sclerosaurus. Both were also horned. The traditional earliest known turtles, Proganochelys and Odontochelys are both known from later, Late Triassic, strata.

Not on topic, but worth watching on YouTube:
Here’s a video about the origin of oil in the Jurassic. It runs for 90 minutes and is fascinating throughout. The video reminds us what a recent Golden Age we currently live in based on a limited supply of petroleum products. The video concludes we have long passed the tipping point for climate change based on the flood of cheap energy. And the end of the oil age is something our children will see. Ironically, climate change in the ice-free Jurassic was one factor in the Earth producing the oil we now use.

References
Lawver DR and Jackson FD 2016. A fossil egg clutch from the stem turtle Meiolania platyceps: implications for the evolution of turtle reproductive biology. Journal of Vertebrate Paleontology. DOI: 10.1080/02724634.2016.1223685.

The only problem with quality first-hand analysis…

…is that you often don’t get to ‘see’
the big picture offered by quantity second-hand analysis. That important step has to come first and unfortunately that has been largely ignored in several paleontological studies.

Both quantity and quality have their place.
But IMHO you must have access to the universe of pertinent taxa before you can say anything substantial about what is beneath your microscope. As an analogy: First the artist blocks in the composition. Later, the artist adds in the little details, like eyelashes, to the composition — which better be good to begin with, or else the little details will be ignored.

Here’s the issue:

Workers familiar with my analyses
like to caution that only first-hand quality observation can be considered scientific. In that way they insulate themselves from considering the views of workers who have not seen the material first-hand. In counterpoint, I often caution workers to consider other candidate (quantity) taxa recovered second-hand by the large reptile tree they may have overlooked.

This subject came up recently
with the continuing hypothesis that Pappochelys was the ancestor to turtles. I pointed out that other candidates share more traits and the LRT nests Pappochelys far from turtles. In counterpoint, that worker encouraged me to go visit the Pappochelys specimen before making any pronouncements. In counter-counterpoint, I encouraged the worker to broaden his inclusion set and rerun his analysis. In other words, I thought his metaphorical ‘composition’ (= inclusion set) was not yet ready to explored the finer details not already available in the literature (second-hand observation).

Alas,
I think these suggestions will come to an impasse, since we are literally and metaphorically on opposite sides of the world. And that’s too bad… No one likes to consider doing the extra work and spending the extra dollars and hours to find out they were wrong, especially after investing so much time and pride. But if they really are good scientists they should explore and refute other options before proclaiming their candidate is truly the best,  especially when those other candidates are brought to their attention.

I realize the importance of first-hand observation.
But it must be done after a wide-gamut analysis, like the large reptile tree, which sets down a working tree topology. Even that is not the final word! Everything in Science is provisional. The LRT is a useable guide to those making up their own taxon lists to explore first hand. If they don’t like particular scores, those can be ignored. If they don’t particular taxa, those can be eliminated. The LRT topology is robust enough to sustain errors, deletions and missing data. I know because I’ve been molding it and working with for 6 years and it’s better than ever. Most workers, as you know, prefer to employ the cladograms of prior workers without testing them, which, of course, perpetuates errors.

PS. I have also had the experience
of having my first-hand observations dismissed by workers who did not have first-hand observations, so personality, professional status and academic power do indeed come into play to keep some data, figures and hypotheses out of the literature. It is not fair. It is two-faced. And that’s just the way it is. Paleontology does not turn corners very easily. Attitudes like this must be placated… or played a different way…

More evidence that Meiolania is a basal turtle

Figure 5. Meiolania, the most primitive of known turtles, has lateral forelimbs, like non turtles.

Figure 1. Meiolania, the most primitive of known turtles, has lateral forelimbs, like non turtles. Extant turtle elbows point anteriorly. 

Earlier we looked at the bizarre and seeming highly derived skulls of Meiolania (Fig. 1) and Niolamia, (Fig. 2) two large late-surviving meiolanid turtles that are only known from rather recent fossil material following an undocumented origin in the Late Permian or Early Triassic.  They both nested as sisters to Elginia (Fig. 2; Late Permian), a toothed turtle sister with horns. So the horns and frills are primitive, not derived.

Figure 2. Comparing the skulls of Elginia, with teeth, and the turtle, Niolamia, toothless.

Figure 2. Comparing the skulls of Elginia, with teeth, and the turtle, Niolamia, toothless.

Here’s a review
of various turtle ancestor candidates in graphic format (Fig. 3). A candidate touted by several recent authors, Eunotosaurus, is among those shown.

Figure 1. In traditional studies Eunotosaurus nests at the base of turtles, but that is only in the absence of the taxa shown here and correctly scored. Here Eunotosaurus is convergent with turtles, but not related. Turtles arise from small pareiasaurs.

Figure 3. In traditional studies Eunotosaurus nests at the base of turtles, but that is only in the absence of the taxa shown here and correctly scored. Here Eunotosaurus is convergent with turtles, but not related. Turtles arise from small pareiasaurs.

Cervical count
Pareiasaurs have 6 cervicals. Turtles have 8, several of which are tucked inside the shell. Proganochelys, often touted as the most basal turtle, has 8 cervicals. Horned Meiolania, at the base of the hard-shell turtles has 6 cervicals with ribs and 2 without ribs according to Gaffney (1985; Fig. 4). Most living turtles do not have cervical ribs. In Proganochelys cervical ribs are much reduced.

Note that in Odontochelys (Fig. 3 a similar situation arises where the all the vertebrae anterior to the expanded ribs are considered cervicals, even though two are posterior to the scapula. Similarly, in Proganchelys (Fig. 3) the last cervical is posterior to the scapula. In other tetrapods (let me know if I am forgetting any), all the cervicals are anterior to the scapula and a few dorsal vertebrae typically appear anterior to the scapulae. The tucking of the scapula beneath the ribs of turtles is a recurring problem with many offering insight.

Figure 1. Meiolania cervicals. Did Gaffney follow tradition when he identified 8 cervicals here? Only 6 have ribs and the shape changes between 6 and 7.

Figure 4. Meiolania cervicals. Did Gaffney follow tradition when he identified 8 cervicals here? Only 6 have ribs (yellow) and the shape changes between 6 and 7.

There are several different possible nesting sites
for turtles with regard to living reptiles (including mammals and birds, Fig. 5). Only the LRT (in yellow) has not made it to the academic literature (after several tries) because it is the only tree topology that splits Archosauromorpha from Lepidosauromorpha in the Viséan, further in the past than other workers venture to place reptiles that still look like amphibians. Until we get the basic topology down and agreed upon, it is going to be difficult to nest turtles properly.

Figure 2. Various hypotheses regarding turtle origins. The LRT is added in yellow.

Figure 5. Various hypotheses regarding turtle origins. The LRT is added in yellow. Most studies show Synapsida as the basal dichotomy, whereas the LRT divides Lepidosauromorpha from Archosauromorpha together with two separate origins for diapsid reptiles.

References
Gaffney ES 1985. The cervical and caudal vertebrae of the cryptodiran turtle, Meiolania platyceps, form the Pleistocene of Lord Howe Island, Australia. American Museum Novitates 2805:1-29.

Adding the Triassic turtle Proterochersis to the large reptile tree

No surprises here.
The Late Triassic German dome-shelled turtle, Proterochersis (Fraas 1913, Szczygiellski  and Sulej 2016; ZPAL V.39/48), was added to the large reptile tree. No surprise, it nested with the other Late Triassic German dome-shelled turtle, Proganochelys. I was worried that Proterochersis would cause loss of resolution because the specimen lacks a skull, cervicals, caudals and limbs. Thus, all scores were based on the dorsal verts, ribs and girdles. And that was enough.

Proganochelys and Proterochersis, two Traissic turtles.

Figure 1. Proganochelys and Proterochersis, two Traissic turtles.

Szczygiellski and Sulej 2016
recently looked at Proterochersis together with a new Triassic turtle, Murrhardtia.

Here’s a big question
Proganochelys has a tall set of clavicles (aka epiplastra) that contacted and braced both the plastron and carapace (Gaffney 1990). Several basal dome-shelled turtles have these. In the basal dome-shelled turtle, Meiolania, Gaffney (xxxx) reports, “In the plastron the epiplastra meet on the midline and bear a short median process, apparently not homologous to that in Proganochelys and Kayentachelys, that bifurcates dorsally and articulates with the scapula. The epiplastron is a paired, curved element, meeting on the midline at the front of the plastron and forming a dorsal process. None of the specimens show a midline suture.”

Szczygiellski and Sulej 2016 reported, “the sturdy build of Proganochelys quenstedti should … be considered its own apomorphy. The presence of strong dorsal epiplastral processes contacting the carapace may be one of the consequences: although the dorsal processes themselves are interpreted by Gaffney (1990) as remnants of ancestral amniote clavicles, their additional articulation with the carapace and strengthening might have stabilized the shell, and thus serve as a more rigid point of attachment for the limb musculature (which probably was required to support the heavy body). Large dorsal epiplastral processes are present in the slightly smaller Palaeochersis talampayensis (Sterli et al., 2007), but are weaker and do not articulate with the carapace in more basal Proterochersis spp. and Keuperotesta limendorsa gen. et sp. nov. In Odontochelys semitestacea they obviously do not contact the carapace, because no suitable point of attachment was available (Li et al., 2008), but they possibly played a similar role, temporarily supporting and strengthening the limb musculature (weakened by changes in rib position), and disappeared when the torso of the animal became fully stiffened and the pectoral girdle received its derived shape.”

References
Fraas E. 1913. Proterochersis, eine pleurodire Schilderöte aus dem Keuper. Jahreshefte des Vereins für Vaterlänzische Naturkunde in Württemberg 69: 13–30.
Szczygiellski T and Sulej T 2016. Revision of the Triassic European turtles Proterochersis and Murrhardtia (Reptilia, Testudinata, Proterochersidae), with the description of new taxa from Poland and Germany. Zoological Journal of the Linnean Society 177:395-427.
Gaffney ES 1996. The postcranial morphology of Meiolania platyceps and a review of the Meiolaniidae. Bulletin of the American Museum of Naturaly Histoyr 229: 1-165.

The dual origin of turtles to scale

Earlier the large reptile tree recovered a dual origin for soft shell and hard shell turtles. Here (Figs. 1-3) we’ll put the pertinent taxa to scale as animated GIF files. These help demonstrate evolution in a crude sort of way. Unfortunately, this is the best we can do at present with known taxa and published data. More discoveries will fill in the gaps.

Figure 1. Hard shell turtle evolution with Bunostegos, Elginia, Meiolania and Proganochelys.

Figure 1. Hard shell turtle evolution with Bunostegos, Elginia, Meiolania and Proganochelys to scale. Basal hard shell turtles had horns and club tails. The anterior rotation of the forelimbs is a derived trait.

It would be nice to find some Elginia postcrania
A reduction in size and loss of teeth coincided with the appearance of the carapace and plastron in hard shell turtles. Unfortunately, this critical stage is represented at present by a skull-only taxon, Elginia. Basal turtle taxa, like Meiolania, had horns and the limbs remained oriented laterally. A club tail trailed basal turtles. Did that develop earlier? We have not seen the ribs of Bunostegos published yet. One wonders if they were different than those of other pareiasaurs. Probably not if they were unremarkable.

Figure 2. Hard shell turtle evolution featuring Bunostegos, Elgenia, Meiolania and Proganochelys - NOT to scale.

Figure 2. Hard shell turtle evolution featuring Bunostegos, Elgenia, Meiolania and Proganochelys – NOT to scale. Even the palate of Bunostegos is very close to a turtle palate.

The skull of hard shell turtles 
demonstrates the appearance and reduction of knobs/horns along with the elimination of teeth, the reduction and anterior rotation of the naris, reduction of the preorbital region relative to the postorbital region and the gradual appearance of the quadrate in lateral view. The reduction of the horns likewise reduced the dorsal exposure of the post parietals and tabulars. but the supratemporal remained a large element. Unfortunately it  has been traditionally interpreted as a squamosal.

Figure 3. Soft shell turtle evolution featuring Arganaceras, Sclerosaurus, Odontochelys and Trionyx.

Figure 3. Soft shell turtle evolution featuring Arganaceras, Sclerosaurus, Odontochelys and Trionyx.

The evolution of soft shell turtles
also begins with a size reduction from Arganaceras to Sclerosaurus. Thereafter the skull continued to shrink, as the plastron and carapace developed in Odontochelys. Teeth disappeared thereafter, as in Trionyx. Convergent with hard shell turtles the enlargement of jaw muscles in derived turtles included the enlargement of post temporal fenestra anteriorly. embaying the posterior skull. So, not listed yesterday, soft shell turtles converge (or rather developed in parallel) with hard shell turtles, given present data.

Figure 1. New cladogram of turtle systematics. Note the separation of soft shell turtles with orbits visible in dorsal view from domed hard shell turtles with laterally oriented orbits here.

Figure 4. New cladogram of turtle systematics. Note the separation of soft shell turtles with orbits visible in dorsal view from domed hard shell turtles with laterally oriented orbits here.

Small pareiasaurs from China
Since size is an issue in turtle origins, when you find a small pareiasaur, it is worthy of notice. Here (Fig. 5) are two and maybe three humeri from small  pareiasaurs, smaller than Sclerosaurus. None are slenderized nor do they develop spherical proximal articulations as seen in turtles. Apparently they just belong to small or young pareiasaurs.

Figure 5. Small pareiasaur humeri from Benton 2016. Note the scale bars. Some of these are smaller than Sclerosaurus (diagram), yet none are slenderized as in turtles.

Figure 5. Small pareiasaur humeri from Benton 2016. Note the scale bars. Some of these are smaller than Sclerosaurus (diagram), yet none are slenderized as in turtles.

Lee 1993 was correct
in putting pareiasaurs in the ancestry of turtles. That agrees with a large gamut reptile cladogram (subset Fig. 4).

However
Benton (2016) summed up current thinking when he reported, “An unusual aspect of pareiasaurs is that they were identified as an out-group, even the sister group, of turtles by Lee (1993, 1995, 1996, 1997), based on their shared characters of a rigid covering of dermal armour over the entire dorsal region, expanded flattened ribs, a cylindrical scapula blade, great reduction in humeral torsion (to 25°), a greatly developed trochanter major, an offset femoral head, and a reduced cnemial crest of the tibia.

“This was disputed by other morphological phylogenetic analyses (e.g. Rieppel & deBraga, 1996; DeBraga & Rieppel, 1997; Rieppel & Reisz, 1999; Li et al., 2009) that indicated a pairing of turtles and lepidosauromorphs among the diapsids, and by molecular phylogenetic studies of modern reptiles that repeatedly placed turtles among the Diapsida, and the Archosauromorpha in particular (e.g. Hedges & Poling, 1999; Field et al., 2014). New finds of the Triassic proto-turtles Pappochelys and Odontochelys show close links to the Middle Permian Eunotosaurus, and turtles are confirmed as archosauromorphs on the basis of fossil and molecular data, and not related to pareiasaurs (Joyce, 2015; Schoch & Sues, 2015).”

It is interesting to note what Benton does not report…
…a long list of turtle synapomorphies for Pappochelys and or diapsids and or archosauromorphs. He doesn’t because he can’t. A long list of turtle synapomorphies with these clades has not been compiled because it cannot be compiled. Unfortunately, Benton is following the latest literature, not testing it and not seeing the red flags. (Remember Benton was part of the Hone and Benton (2007, 2009) fiasco that attempted to test two origin of pterosaurs hypotheses by eliminating one of them only partly due to self-inflicted typos. The rest was a hatchet job as you can read again here).

Figure 5. Odontochelys pectoral elements reconstructed. Here the acromion process originates along the lower rim of the scapula.

Figure 5. Odontochelys pectoral elements reconstructed. Here the acromion process originates along the lower rim of the scapula. Pelociscus is an extant soft shell turtle. The coracoid of Odontochelys has been cracked at the glenoid. The green area is a hypothetical restoration. The glenoid of the scapula still had a thin veneer of matrix on it when photographed. The ? could be an acromion process. Or it could be a rib. The procoracoid of Sclerosaurus is absent here.

Morphology must trump DNA in prehistoric taxa
In the large reptile tree Pappochelys nests with basal sauropterygians, like Palatodonta, a skull-only basal placodont taxon. Several taxa near this node, including Henodus, Placochelys and Sinosaurosphargis independently developed turtle-like shells. So there was selective pressure to do so in that clade and niche at that time, convergent with extant turtles. No one knows yet why turtle DNA does not nest turtles with lizards more often or why mammal DNA does not nest mammals more closely with archosaurs in concert with the topology of the large reptile tree.

References
Benton MJ 2016. The Chinese pareiasaurs. Zoological Journal of the Linnean Society, doi: 10.1111/zoj.12389

SVP 19 – Something vague about a new clade: “Archelosauria”

Pritchard 2015
provides a teaser abstract that sets up the situation, but provides no solution.

From the abstract
“Since the earliest Triassic, saurian reptiles have been critical components of terrestrial ecosystems. However, molecular and fossil evidence indicates that the divergence between the two constituent lineages (Lepidosauria, Archelosauria [turtles + archosaurs]) took place deep in the Permian Period. A large number of early-diverging stem-archosaur and stemlepidosaur clades have been described from the Permian and Triassic, exhibiting an extraordinary range of bauplans. However, the interrelationships of these stem taxa are poorly resolved, owing to fragmentary records and poor preservation in many groups. As such, the timing of both the initial taxonomic and morphological diversifications of Sauria remain poorly understood. To resolve this phylogenetic uncertainty and the first radiation of crown reptiles, a new phylogenetic data matrix was constructed from a broad sample of Permo-Triassic diapsids. New, three-dimensionally preserved fossils from a number of poorly understood stem groups (e.g., long-necked Tanystropheidae, chameleon-like Drepanosauromorpha) allowed coding of many previously unknown morphologies. Iterations of this data matrix were subjected to both standard parsimony analysis and Bayesian tip-dating methodologies. The results of this analysis suggest that at least ten distinct lineages of Permo-Triassic diapsids survived the PTE, substantially more than went extinct at that time. They do not form a monophyletic Protorosauria clade, a group traditionally considered to include most long-necked, small-headed early archosauromorphs. Instead, these taxa include no fewer than six separate Permo-Triassic diapsid lineages. Indeed, character optimizations strongly suggest that a long-necked, lizard-like bauplan was ancestral for Archosauromorpha. The inclusion of fragmentary fossil material from Early Triassic archosauromorphs indicates that a great deal of morphological diversity existed in saurian groups within the first five million years of the Triassic.**”

*Archelosauria is not recovered in the large reptile tree. Not sure why the molecules do what they do, nesting turtles with archosaurs (hence the clade name).
** This is a teaser abstract. No conclusions are presented. I cannot compare the data here to the cladogram recovered in the large reptile tree.

References
Pritchard A 2015. Resolving the first radiation of crown reptiles. Journal of Vertebrate Paleontology abstracts