Resurrecting extinct taxa: Pareiasauria, Compsognathidae and Ophiacodontidae

Earlier we looked at
four clades thought to be extinct, but are not extinct based on their nesting in the large reptile tree (LRT, 1366 taxa). Today, three more:

Figure 1. Another gap is filled by nesting E. wuyongae between Bunostegos and Elginia at the base of hard shell turtles in the LRT.

Pareiasauria
According to Wikipedia, “Pareiasaurs (meaning “cheek lizards”) are an extinct group of anapsid reptiles classified in the family Pareiasauridae. They were large herbivores that flourished during the Permian period.”

In the LRT two clades of turtles (Fig. 1) are derived in parallel from two small horned pareiasaurs.

Figure 2.  Lately the two clades based on two specimens of Compsognathus (one much larger than the other) have merged recently.

Compsognathidae
According to Holtz 2004, “The most inclusive clade containing Compsognathus longipes but not Passer domesticsus.” Traditionally Compsognathus nests outside the Tyrannoraptora, a clade that traditionally leads to birds.

In the LRT Compsognathus specimens nest at the base of several theropod clades (Fig. 2) including the tyrannosaurs and Mirischia, Ornitholestes and the feathered theropods leading to birds.

Figure  3. Varanosaurus, Ophiacodon, Cutleria and Ictidorhinus. These are taxa at the base of the Therapsida. Ophiacodon did not cross into the Therapsida, but developed a larger size with a primitive morphology. This new reconstruction of Ophiacodon is based on the Field Museum (Chicago) specimen. Click to enlarge.

Ophiacodontidae
According to Wikipedia, “Ophiacodontidae is an extinct family of early eupelycosaurs from the Carboniferous and Permian. Ophiacodontids are among the most basal synapsids, an offshoot of the lineage which includes therapsids and their descendants, the mammals. The group became extinct by the Middle Permian.”

In the LRT Ophiacodon (Fig. 3) and Archaeothyris, neither members of the Pelycosauria, are more directly related to basal therapsids, including derived the therapsids: mammals.

References
Holtz TR 2004. Basal tetanurae. PP. 71–110 in The Dinosauria, U of California Press.

/wiki/Pareiasaur
wiki/Ophiacodontidae

 

SVP 2018: New elginiid parieasaur skull from China

Liu and Bever 2018
describe a complete pareiasaur skull close to Eliginia (Fig. 1), the proximal outgroup to hardshell turtles in the large reptile tree, and no where else. The new specimen is significantly larger than Elginia. The authors refer the specimen to Sanchuansaurus, a taxon apparently known from far less material. Evidently this abstract does not represent the earlier Eliginia wuyongae (Fig. 1), which is not significantly larger than Elginia. Liu and Bever make no mention of the eliginiid relationship to basal hard-shell turtles.

Figure 1. Another gap is filled by nesting E. wuyongae between Bunostegos and Elginia at the base of hard shell turtles in the LRT. The new specimen is larger than Eliginia.

References
Liu J and Bever GS 2018. The first complete pareiasaaur skull from China and its implications for the taxonomy of Chinese pareiasaurs.

wiki/Shihtienfenia

First evidence for elginiid pareiasaurs in the Karoo (South Africa)

In the large reptile tree (LRT, 1308 taxa) pareisaurs split after Stephanospondylus into two clades: 1) traditional pareiasaurs and 2) turtle-ancestor pareiasaurs (Fig. 1). Only the latter clade develop distinct supratemporal horns.

Figure 1. Dorsal views of bolosaur, diadectid, pareiasaur, turtle and lanthanosuchian skulls. The disappearance of the turtle orbit in lateral view occurs only in hard shell turtles. Horns only appear in elginid and sclerosaurid pareiasaurs.

On a recent trip
to the Sam Noble Oklahoma Museum of Natural History in Norman, Oklahoma, USA, I studied a pareiasaur horn, OMNH 708 (Fig. 2). For over a century elginiid pareiasaurs were only known from Scotland. This year other elginids were reported from China (Liu and Bever 2018), and others were reported in 2005 from Eastern Europe (Bulanov and Yashina 2005). OMNH 708 represents yet another specimen, perhaps the first from the Late Permian Karoo beds of South Africa. (Please, let me know of not so.)

Figure 1. OMNH 708, a Permian pareiasaur horn form the Karoo, South Africa.

The OMNH specimen
is much larger than the Scottish and Chinese specimens (Fig. 2), more in the size range of ancestral pareiasaurs and Stephanospondylus.

Figure 3. Elginia and OMNH 708 at two scales.

The only question that remains is
is this really a pareiasaur horn? Or has everyone misinterpreted it? It really looks like a cow or bison horn (Fig. 4), but its origin in Permian strata prohibits that.

Figure 4. A modern cow skull horn most closely resembles the Permian pareiasaur horn, by convergence, of course.

Say, ‘hello’, to convergence, once again.

References
Bulanov VV and Yashina OV 2005. Elginiid pareisaurs of Eastern Europe. Paleontological Journal 39(4):428–432.

Liu J and Bever GS 2018. The tetrapod fauna of the upper Permian Naobaogou formation of China: A new species of Eliginia (Parareptilia, Pareiasauria). Papers in Paleontology 2018: 1-13.
Newton ET 1893. On some new reptiles from the Elgin Sandstone: Philosophical Transactions of the Royal Society of London, series B 184:473-489.

another-turtle-with-teeth-elginia-newton-1893/

another-turtle-with-teeth-elginia-part-2/

another-turtle-with-teeth-elginia-part-3/

Not even an elevated Dimetrodon made these Dimetropus tracks

Matching tracks to trackmakers
can only ever be a semi-rewarding experience. Estimates and exclusions can be advanced. Exact matches are harder to come by. This is due to both the vagaries and varieties of sequential footprints in mud or sand, and to the rarity of having skeletal data that matches.

Figure 1. Dimetrodon adult, juvenile, skull, manus, pes. Note the asymmetry of the fingers and toes. Dimetropus tracks were named for this taxon.

Which brings us to Dimetropus
Traditionally Early Permian Dimetropus tracks (Fig. 2–8; Romer and Price 1940) have been matched to the coeval pelycosaur, Dimetrodon (Fig. 1)—but only by narrowing the gauge of the Dimetrodon feet and elevating the belly off the surface, as Hunt and Lucas 1998 showed.

Today we’ll take a look at some other solutions
not involving Dimetrodon doing high-rise pushups. Several distinctly different tracks have fallen into the Dimetropus wastebasket. Let’s look at three ichnospecimens.

Traditionally, and according to Wikipedia,
citing Hunt and Lucas 1998: “Trackways called Dimetropus (“Dimetrodon foot”) that match the foot configuration of large sphenacodontids show animals walking with their limbs brought under the body for a narrow, semi-erect gait without tail or belly drag marks. Such clear evidence for a more efficient upright posture suggests that important details about the anatomy and locomotion of Sphenacodon and Dimetrodon may not be fully understood.” Hunt and Lucas blamed traditional reconstructions of Dimetrodon for the mismatch. Instead they should have looked at other candidate trackmakers from the Early Permian. Note the asymmetric manus and pes of Dimetrodon (Fig. 1). Those don’t match the tracks no matter how high the belly is above the substrate. Dimetrodon is just fine the way it is.

Figure 2. Early Permian Dimetropus tracks matched to Middle Triassic Sclerosaurus, one of the few turtle-lineage pareiasaurs for which hands and feet are known.

A better match
can be made to the Middle Triassic pre-softshell turtle pareiasaur, Sclerosaurus (Fig. 2). Note the symmetric manus and pes like those of living turtles (Fig. 3) and the Dimetropus specimen in figure 2.

Figure 3. Snapping turtle tracks in mud. Note the relatively narrow gauge and symmetric imprints like those of Dimetropus.

Living turtle tracks
like those of the snapping turtle, Macrochelys (Fig. 3) are also symmetrical and surprisingly narrow gauge. Let’s not forget, Dimetropus tracks occur in Early Permian sediments, predating the earliest fossil turtles, like Proganochelys, first appearing in the Late Triassic. Let’s also not forget, in the large reptile tree (LRT, subset Fig. 7) Proganochelys is not the most basal turtle and valid predecessors (not eunotosaurs) had similar hands and feet.

FIgure 4. Dimetropus tracks compared to a large Dimetrodon matched to finger and toe tips. Hand too wide. Compared to a small Dimetrodon. Hand too small. Compared to a normal size Hipposaurus, good match even if not all the digits are known.

A second set of Dimetropus tracks
(Fig. 4, right), have distinctive heels behind symmetric + asymmetric imprints. A large Dimetrodon could not have made these tracks because they are too narrow. A small Dimetrodon had extremities that were too small, as the animated GIF shows.

FIgure 5. Hipposaurus compared Dimetropus. The overall and leg length is right, as are many of the digits. Unfortunately the medial digits are too short in Hipposaurus. Hipposaurus has a narrower gauge and lifted its belly of the ground, as did the Dimetropus trackmaker.

Fortunately,
we also have Middle Permian basal therapsid, Hipposaurus (Figs. 4, 5), a close relative of the last common ancestor of all pelycosaurs (see Haptodus and Pantelosaurus; Fig. 6). No doubt Hipposaurus elevated its torso on a narrow gauge track, with manus tracks slightly wider than pedal traces, as in Dimetropus. Both the carpus and tarsus are elongate, matching Dimetropus tracks.

Unfortunately,
we don’t have all the phalanges for the Hipposaurus manus and pes (Fig. 4). Drag marks can lengthen a digit trace. Flexing a claw into the substrate can shorten a digit trace. It is also important to note that during the last moment of the manus propulsion phase, the medial and lateral metacarpals can rotate axially, creating the impression of an ‘opposable thumb’ in the substrate. Note that no two ichnites are identical, despite being made one after another by the same animal.

Figure 6. Closeup of Hipposaurus manus and pes compared to random Dimetropus manus and pes tracks. Note, some digits remain unknown. Some digits might create drag marks. Others may dig in a claw or two apparently shortening the digit imprint.

At present
a more primitive sister to Hipposaurus is the best match for the Hunt et al. 1995 Dimetropus tracks and the Early Permian timing is right.

FIgure 7. Subset of the LRT focusing on Hipposaurus and its relatives, color coded to time. Hipposaurus is nearly Early Permian and probably had its genesis in the Early Permian.

In the popular press
NewScientist.com reported, “We’ve drawn iconic sail-wearing Dimetrodon wrong for 100 years. Some palaeontologists did offer an explanation – that Dimetrodon thrashed its spine from side to side so much as it walked that it could leave narrow sets of footprints despite having sprawled legs.” That hypothesis, based on omitting pertinent taxa, is no longer necessary or valid.

Abbott, Sues and Lockwood 2017 reported the limbs of Dimetrodon were morphologically closest to those of the extant Caiman, which sits on its belly, but also rises when it walks.

It is unfortunate that no prior workers considered Hipposaurus, a nearly coeval taxon with Dimetropus having matching slender digits, long legs, an erect carriage, and just about the right digit proportions.

A third ichnotaxon,
Dimetropus osageorum (Sacchi et al. 2014), was considered a possible caseid, rather than a sphenacodontid, but caseids have more asymmetric digits (= a shorter digit 2). Unfortunately, taxon exclusion also hampered the Sacchi et al. study. They did not consider Early Permian stephanospondylids, Late Permian pareiasaurs in the turtle lineage and Triassic turtles. No skeletal taxon is a perfect match for this ichnotaxon, but the Late Cretaceous turtle, Mongolochelys, is close  (Fig. 8). It took some 200 million years after the trackmaker of Dimetropus for the lateral pedal digits to shrink, but everything else is a pretty good match.

Figure 8. Dimetropus oageorum from Sacchi et al. 2014 matched to Mongolochelys, a Late Cretaceous turtle. Only pareiasaurs and turtles, among basal taxa, have such a long manual and pedal digit 2. The reduction of pedal digits 4 and 5 are derived in this late surviving basal turtle.

Also compare the hands and feet
of Early Permian Dimetropus osageorum (Fig. 8) to the Middle Triassic Sclerosaurus (Fig. 2). Dimetropus is solid evidence that turtle-ancestor pareiasaurs were present in the Early Permian (see Stephanospondylus, an Early Permian turtle and pareiasaur ancestor).

Saachi et al. conclude, “At the same time, the process of attributing ichnotaxa, on the basis of well preserved tracks and by comparison with known skeletal remains, is validated.”  True. Unfortunately all prior workers overlooked a wider gamut of skeletal taxa to compare with their ichnotaxon in their search for a ‘best match.’ Perhaps they felt restricted by time (Early Permian). As the above notes demonstrate, that is not a good excuse.

References
Abbott CP, Sues H-D and Lockwood R 2017. The Dimetrodon dilemma: reassessing posture in sphenacodonts. GSA annual meeting in Seattle, WA USA 2017. DOI: 10.1130/abs/2017AM-307190
Hunt AP and Lucas SG 1998. Vertebrate tracks and the myth of the belly-dragging, tail-dragging tetrapods of the Late Paleozoic. Bulletin New Mexico Museum of Natural History and Science. 271: 67–69.
Peters D 2011. A Catalog of Pterosaur Pedes for Trackmaker Identification. Ichnos 18(2):114-141. http://dx.doi.org/10.1080/10420940.2011.573605
Romano M, Citton P and Nicosia U 2015. Corroborating trackmaker identification through footprint functional analysis: the case study of Ichniotherium and Dimetropus. Lethaia https://doi.org/10.1111/let.12136
Romer AS and Price LI 1940. Review of the Pelycosauria: Geological Society of America, Special Paper 28:538pp
Sacchi E, Cifelli R, Citton P, Nicosia U and Romano M 2014. Dimetropus osageorum n. isp. from the Early Permian of Oklahoma (USA): A trace and its trackmaker. Ichnos 21(3):175–192. https://doi.org/10.1080/10420940.2014.933070

Another overlooked turtle ancestor just got published

Considered
congeneric with Elginia mirabilis (from Late Permian Scotland), the new elginiid comes from Late Permian China (Figs. 1, 2). The authors (Liu and Bever 2018) correctly identified the material in a specific sense, but had no idea what they had in a broader sense, because they only tested Elginia against pareiasaurs.

It’s really part of the genesis of turtles (Fig. 2), and we’re glad to see it!

Once again,
taxon exclusion raises its blind head. We’ve known Elginia was a turtle ancestor since 2014 when that went online. Unfortunately co-author Bever had earlier published on the genesis of turtles, relying on pre-turtle-mimic Eunotosaurus. Both are tested in the large reptile tree (LRT, 1152 taxa) and Eliginia nests with turtles. Eunotosaurus does not. It is more closely related to Acleistorhinus and kin. When Liu and Bever include Meiolania and Niolamia (Fig. 2) in their analyses, then they’ll see how it all plays out.

Elginia wuyongae (Figs. 1, 2) is smaller than Elginia mirabilis, lacks long horns and nests between the big desert pareiasaur, Bunostegos (Fig. 2), and its Scottish namesake at the base of hard shell turtles. Importantly, E. wuyongae preserves a few post-cranial data, including the genesis of the hard-shell turtle carapace…which is incredible news!!!

But you’re hearing that here first.
Jiu and Bever did not understand the importance and so overlooked it.

Figure 1. Elginia wuyongae was just described. It shows the genesis of shell formation in hard shell turtles. That tiny last sacral vertebra (near the four dots) suggests a tiny tail was present. 

Lacks a rostrum…
skull is pretty beaten up, parts missing, holes pocket bones, lacks a palate. Squamosal misidentified originally (repaired here). Still, you gotta love it! It has post-cranial clues lacking in other transitional taxa. And it fills a gap!

How can workers not notice the family resemblance? 

Figure 2. Another gap is filled by nesting E. wuyongae between Bunostegos and Elginia at the base of hard shell turtles in the LRT. Those other horned pareiasaurs are basal turtles, meiolaniids with substantial carapace and plastron. Both sides of the new Elginia skull are shown. The squamosal is tucked inside the overlapping supratemporal in these transitional taxa. 

The authors do mention the turtle connection, like so:
“…and their long-hypothesized, but now largely rejected, potential as the close relatives
of turtles (Rieppel & deBraga 1996; Lyson et al. 2010; Lee 2013; Lyson et al. 2013; Bever et al. 2015; Schoch & Sues 2015; Laurin & Pi~neiro 2017).” It’s not surprising how many workers think this – because they don’t test the taxa that need to be tested, as they are tested here in the LRT. Remember, a consensus of workers can be wrong.

On that note:
Liu and Bever are still clinging to the invalid clade Parareptilia.

References
Liu J and Bever GS 2018. The tetrapod fauna of the upper Permian Naobaogou formation of China: A new species of Eliginia (Parareptilia, Pareiasauria). Papers in Paleontology 2018: 1-13.

More evidence that Meiolania is a basal turtle

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.

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 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 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 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.

Bunostegos: maybe not so oddly erect in stance after all…

Earlier we nested the knobby-faced pareiasaur, Bunostegos (Sidor et al. 2003, Tsuji et al. 2013, Turner et al. 2015; Fig. 1), with spiky Elginia at the base of all hard-shelled turtles, like Meiolania and Proganochelys. Soft-shelled turtles, like Odontchelys, as you might remember, were derived from a distinct, but closely related pareiasaur clade arising from Arganceras and Sclerosaurus, indicating that living turtles are diphyletic with two clades going back to shell-less ancestors among the smaller pareiasaurs.

Figure 1. The originally published cartoon of Bunostegos with skeletal elements laid on top of it. As you can see, the right humerus was mistakenly illustrated in the right hand position. And this may have led to the error of proposing that this pareiasaur was uniquely erect in gait. Image from Brown University website (see below)

Bunostegos
was reported (Tsuji et al. 2013) to also have a parasagittal (erect, upright) gait, which is not only odd, but unique for both pareiasaurs and turtles. That put up a red flag. Sorry it took so long to get to. I think I see a mistake here in the humerus identification. Tsuji et al. might have mistaken a left humerus for a right one, based on the cartoon illustration of a complete specimen (Fig. 1). It might have been an easy mistake to make because Tsuji et al. report at least 9 individuals, several sizes, each and all represented by a short list of disarticulated bones.

Figure 2. Here’s Proganochelys in dorsal view. Note the humerus. If you look closely you’ll see a small depression lateral to the proximal articulation with the shoulder glenoid. And note the larger of the two proximal processes is lateral here, medial when the elbow is oriented posteriorly as in most other tetrapods.

Let’s start with what we know:
Everyone knows that Proganochelys (Fig. 2) nests as a basal turtle. It is the basalmost turtle in which the elbows were anterior to the shoulders in a normal configuration (in the more basal Meiolania they are primitively lateral). That rotation turns the traditional lateral condyles into medial condyles in practice. I want you to note the slight indentation lateral to the ball-like proximal humerus that fits into the socket-like shoulder glenoid in figure 2. You’ll see that again in Bunostegos (Fig. 3), but much larger.

Meiolania is an even more primitive hard-shell turtle
though this is still not the working hypothesis among traditional paleontologists. Here (Fig. 3) we’ll look at the humerus of Meiolania and other parts (Figs. 4-7) that will match what few bones were recovered from the Bunostegos site.

Figure 3. The left humerus of Bunostegos and the basal turtle Meiolania for comparison, both in dorsal view.. Colors denote homologous areas. That little dip in the medial condyle of Proganochelys (Fig. 2) is much larger here in Bunostegos and small in Meiolania.

That little dip
in the medial condyle of Proganochelys (Fig. 2) is much larger here (Fig. 3) in Bunostegos and small again in the basal turtle Meiolania. Look again at figure 1 and you’ll see the big basin in Bunostegos was incorrectly flipped in the Brown University illustration.

Figure 4. Pre-turtle pectoral girdle evolution. Here homologous areas are colorized. The acromion process is broken on all specimens of Bunostegos. Pink arrow points anteriorly. Note the lowering of the acromion process in Bunostegos. We don’t know how long it was. Also note the narrowing of the scapula. Note the maturation (ontogenetic)  changes to the glenoid in Bunostegos. The more lateral orientation is on the smaller/younger specimens, as in basal turtles.

We’ve been looking for the ancestors of turtles for some time now
And unfortunately these three papers on Bunostegos completely overlooked the possibility of a close relationship to Meiolania and other basal hard-shell turtles. You can see the evolution of the pectoral girdle and other bones provides the most gradual accumulation of derived traits known at present. At present, this blog and ReptileEvolution.com are the only studies that have recovered this heretical relationship.

Figure 5  Once again, and this time to scale, the pectoral girdles of Bunostegos. Note the more lateral orientation of the glenoid in young specimens, as in turtles (Fig. 3).

It is interesting to see
the change in the orientation of the shoulder glenoid in the Bunostegos growth series (Fig. 5). Interestingly, the smaller specimens have more laterally directed glenoids, as in basal turtles (Fig. 4), which are also smaller.

Figure 6. Turtle pelvis evolution. Here are the changes in the pelvis of pre-turtles and basal hard-shelled turtles. We don’t know how long the pubis was in Bunostegos. The ischium is narrower in the last three taxa here. Meiolania has a tall, pareiasaur-like ilium. Bunostegos has a pointed posterior ilium, as in Proganochelys.

The evolution of the turtle pelvis
is best seen in a series of pre-turtle and basal turtle pelves (Fig. 6). The acetabulum in all cases is lateral, but hard-shell turtles develop an acetabular crest that roofs over the joint and altogether form a socket shape for the ball-like femoral head (Fig. 7). This occurs concurrent with the appearance of the carapace and plastron.

Figure 7. Turtle femur evolution. Here the femoral head is interned in Bunostegos and assumes a spherical shape in the turtles, Meiolania and Proganochelys. We know the turtles held the femur horizontally, not parasagittaly.Pink arrro points anteriorly in these left femurs.

The evolution of the turtle femur
can be seen in this series of pre-turtle and basal turtle femora (Fig. 7). Note the gradual development of the ball joint on the proximal femur along with the development of the sigmoid (=’S’) shape of the femur. These developments coincide with the appearance of the carapace and plastron.

Figure 8. Even though the femur has an offset and spherical head in this basal turtle, Proganochelys, still it does not indicate a parasagittal gait, but a horizontal, sprawling one.

I was not able to find 
comparable pareiasaur humeri. They are not online and I don’t think anyone has done a large comparative study replete with a rich trove of illustrations yet. Basal turtles are smaller than most pareiasaurs. The hind limbs sprawl more.

I’d like to see
if any osteoderms or turtle-like ribs were found at the Bunostegos site. None have been reported so far. Hopefully this report will spur further studies with an eye toward gathering more pre-turtle data in Bunostegos. At present the many authors don’t know how really special their fossils are. There is a better story here than the false report of parasagittal limbs.

References
Sidor CA, Blackburn DC and Gado B 2003. The vertebrate fauna of the Upper Permian of Niger — II, Preliminary description of a new pareiasaur. Palaeontologica Africana 39: 45–52.
Turner ML, Tsuji LA, Ide O, Sidor CA 2015. The vertebrate fauna of the upper Permian of Niger—IX. The appendicular skeleton of Bunostegos akokanensis (Parareptilia: Pareiasauria). Journal of Vertebrate Paleontology: e994746. doi:10.1080/02724634.2014.994746.
Tsuji LA, Sidor CA, Steyer JSB, Smith RMH, Tabor NJ and Ide O 2013. The vertebrate fauna of the Upper Permian of Niger—VII. Cranial anatomy and relationships of Bunostegos akokanensis (Pareiasauria). Journal of Vertebrate Paleontology 33 (4): 747. doi:10.1080/02724634.2013.739537

Brown University website with news on Bunostegos

wiki/Bunostegos

 

 

Now turtles are diphyletic and finally make sense

Turtle systematics has changed for the better
Flattened softshell turtles are indeed different from domed hardshell turtles. The arose from separate, though closely related, non-shelled ancestors according to the latest data input and recovered from the large reptile tree (subset Fig. 1). The carapace and plastron in each turtle clade arose by convergence, based on present data. Earlier we looked at where other paleontologists have been looking for the ancestors of turtles.

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.

I had it wrong earlier
And that’s okay as Science marches on. We build on past successes and mistakes made by both ourselves and others (see below). I saw a Red Flag (= a logical inconsistency, see below) and reexamined my data scores. New understandings popped up, like the absence of a premaxilla in Ocepecephalon and the coincident appearance of a new secondary naris high on the skull dividing the nasal bone (fused at its midline) in two (Figs. 2). That is very weird and may be unique for all tetrapods. A sister, Trionyx, has only a vestige of a premaxilla and no ascending process. So Ocepecephalon simply took it to the next level and completely lost the premaxilla — and perhaps most of the ectopterygoid.

Fig. 2. Ocepecephalon, the siphoning turtle. Originally the long dorsal rostrum bone was considered a premaxilla, but comparisons to sister taxa, like Trionyx, indicate it is an anterior nasal separated from the posterior nasal by a new naris, unlike that of any other turtle or tetrapod. The naris and jaw opening are one here as the tiny premaxilla found in Trinoxy is completely absent here. The ectopteryogoids appear be broken here, but lacking a connection to the cheek may be yet another autapomorphy. New skull bone identities are labeled here.

This new tree topology for turtles solves the problem
of toothy Odontochelys appearing after the loss of teeth in ancestral taxa, as recovered by the old data with several incorrect scores. And this also solves the problem of soft-shell turtles with dorsally visible orbits, like Odontochelys, Trionyx and Ocepecephalon (Fig. 2), appearing after the orbits had already rotated to the lateral side of the skull in hardshell turtles derived from Elginia, Meiolania and Proganochely.

Figure 3. Dorsal views of bolosaur, diadectid, pareiasaur, turtle and lanthanosuchian skulls. The disappearance of the turtle orbit in lateral view occurs only in hard shell turtles. Now the gradual accumulation of character traits is even more gradual. 

Correcting mistakes
and seeking new insights are what ReptileEvolution.com and this blog are all about,  whether I made the mistakes or others made the mistakes (usually a combination of the two). On a grander scale, that’s what Science is all about. It also feels good to solve persistent problems.

The .nex file is available on request, as always.

Dorsal views of basal turtle skulls support the cladogram

Earlier
here, here and here we looked at turtle origins — a controversial topic in mainstream paleontology resolved quickly and surely in the large reptile tree, which gives 639 taxa the opportunity to be ancestral to turtles.

Long story short
Toothy Elginia currently nests outside the turtles (only because we don’t have any post-crania) and toothless Meiolania nests as the basalmost turtle (Fig. 1) because it retains supratemporal horns and the elbows still extend laterally, not anteriorly. These taxa are derived from pareiasaurs, which are themselves sisters to diadectids, bolosaurs and proclophonids.

Figure 1. How the large reptile tree lumps and splits the several Diadectes specimens now included here. Note that bolosaurids, including Phonodus, now nest within other Diadectes specimens.

When the skulls of pertinent taxa
are seen in dorsal view (Fig. 2) it is easier to see the reduction of the horns in  pre- and basal turtle skulls. One also gets the impression that when Proganochelys and Odontochelys arrived on the scene in the Late Triassic, they both represent a much earlier radiation of turtles, both horned and not horned. So there are many more basal turtles out there waiting for us to discover them.

Figure 2. Turtles and their ancestors among the pareiasaurs. Note the soft shell turtle clade rotates the orbits until they are visible dorsally. Click to enlarge. Odontochelys is not so primitive as once considered. AND it appears to have redeveloped teeth. Note the reduction of supratemporal horns in basal turtles.

The Odontochelys tooth problem
Odontochelys is a Late Triassic toothed turtle that originally was considered (Li et al. 2008) a very basal turtle. Not so according to phylogenetic analysis which nests it with soft shell turtles like Trionyx. The odd thing is this soft shell turtle appears to have regrown teeth. More basal and sister taxa do not have teeth (Fig. 3). Odontochelys is also unusual in having nares in the anterior lateral orientation, not completely anterior, as in Trionyx, as in virtually all other turtles, and not dorsal, as in Ocepecephalon, which is also very off for a turtle.

Figure 3. Odontochelys and Trionyx. Note the teeth in ventral view of the Odontochelys skull. Click to enlarge.

The supratemporal problem
This evolutionary sequence demonstrates that the large supratemporal bones of turtles (the supratemporal horns of pre-turtles and Meiolania) have been traditionally mislabeled. This may be part of the problem that workers have had in nesting turtles in prior studies.

The molecule problem
Some researchers have found that turtle DNA is most closely matched to that of living archosaurs: crocs and birds. Everyone knows morphology does not support that nesting. Someone somewhere will figure this out someday.

References
Li C, Wu X-C, Rieppel O, Wang L-T and Zhao L-J 2008. An ancestral turtle from the Late Triassic of southwestern China. Nature 456: 497-501.

x

 

Adding taxa to the Diadectes clade

Adding a few
and distinct Diadectes specimens (no two appear to be conspecific) opens the door to new insights into that corner of the cladogram. Some of the data are from 3D skull images with sutures delineated. Others are from firsthand observation. Some data are from drawings. Berman et al. 1992 made an interesting observation that prior authors illustrated the skull roof of Diadectes in a variety of ways (Fig. 1). The caption does not indicate that all were drawn from the same specimen. I suspect they were not.

Figure 1. How Berman et al. copied the illustrations of prior authors who each figured the skull roof of Diadectes. Perhaps these were several distinct specimens, not just one. Not sure, at this point, which illustrations represent which specimens.

The Berman et al. phylogenetic analysis
included seven taxa, including two suprageneric taxa, Pelycosauria and Captorhinomorpha. They included only nine characters. The anamniote, Seymouria, was the outgroup. The first clade included Pelycosauria + (Limnoscelis +(Tseajaia and Diadectes). The second clade included Captorhinomorpha + Petrolacosaurus. The large reptile tree includes hundreds more taxa and characters. The pertinent subset is shown here (Fig. 2). It is also clear from the Berman et al. taxon set that they thought they were dealing with a small set of basal reptiles and pre-reptiles. In 2015 it is clear that they did not include the pertinent taxa they should have as some of these taxa are not related to any of the others except distantly.

Figure 2. How the large reptile tree lumps and splits the several Diadectes specimens now included here. Note that bolosaurids, including Phonodus, now nest within other Diadectes specimens.

Now, with current data
it is becoming increasingly clear that both bolosaurids and procolophonids nest within  a fully reptilian Diadectes clade. It is also clear that the genus Diadectes needs to be further split, as Kissel (2010) started to do by renaming Silvadectes and Oradectes from former Diadectes species.

Figure 3. Skeleton of Diadectes (UC 706, UC 1078). Perhaps unnoticed are the broad dorsal ribs of this taxon, basal to Stephanospondylus, Procolophon and pareiasaurs.

Also note
the placement of Stephanospondylus as a proximal sister taxon to the diadectids nesting at the base of the pareiasaurs (including turtles). Turtles are sisters to pareiasaurs and they ARE pareiasaurs because they are derived from pareiasaurs, just as birds are derived from theropod dinosaurs.

Figure 4. Click to enlarge. Stephanospondylus based on parts found in Stappenbeck 1905. Several elements are re-identified here. Note the large costal plates on the ribs, as in Odontochelys. The pubis apparently connected to a ventral plastron, not preserved. The interclavicle was likely incorporated into the plastron.

Like everyone who studies prehistoric reptiles
there is a day when you don’t know anything about a taxon and later there is a day when you are making contributions to Science. Those days keep on coming.

References
Berman DS, Sumida SS and Lombard E 1992. Reinterpretation of the Temporal and Occipital Regions in Diadectes and the Relationships of Diadectomorphs. Journal of Vertebrate Paleontology 66(3):481-499.
Kissel R 2010. Morphology, Phylogeny, and Evolution of Diadectidae (Cotylosauria: Diadectomorpha). Thesis (Graduate Department of Ecology & Evolutionary Biology University of Toronto).