Arvinachelys: another pig-nose turtle

Arvinachelys goldeni (Lively 2015; Late Cretaceous; UMNH VP 21151; 60cm in length; Figs. 1, 2) was promoted as a unique pig-nose turtle, having twin narial openings. That trait is now shared with Kallokibotion (Nopsca 1923; Pérez-García and Vlad Codrea 2018; Fig. 2), which we looked at earlier here. This is a near-basal turtle nesting near baenids and before any extant taxa (crown group) among the hard-shell turtles.

Arvinachelys is not closely related
to the living pig-nose turtle, Carettochelys.

Lively 2015
fused many of the skull bones together in his µCT scan (Fig. 1). I was able to trace outlines for several fused elements. I also relabeled the squamosal as the supratemporal. Lively did not identify the top half of the premaxilla, separated from the bottom half in the adult skull. Though rare, such splits do occur in certain other turtles, Lively did not identify the bottom half of the jugal, recolored here.

Figure 1. Arvinachelys as originally figured, then with several fused bones segregated with colors.

Figure 1. Arvinachelys as originally figured, then with several fused bones segregated with colors.

Figure 3. Kallokibotion compared to Meiolania.

Figure 3. Kallokibotion (at right) compared to Meiolania. The former also has divided narial openings and precedes Arniachelys phylogenetically.

Publicity
Journalist call this turtle’s twin nares ‘bizarre’ and ‘weird’, even though basal turtles, like Proganochelys share this trait. When you think about it, there are several turtles that are much more weird than Arvinachelys. Superlatives, whether warranted or not, seem to drive paleo publicity.

https://www.earthtouchnews.com/discoveries/fossils/this-ancient-pig-nosed-creature-might-be-the-weirdest-turtle-that-ever-lived/

https://www.cnn.com/2015/10/24/us/goldens-bacon-turtle-fossil-feat/index.html

http://time.com/4083220/pig-snouted-turtle-fossil/

References
Lively J 2015. A new species of baenid turtle from the Kaiparowits Formation (Upper Cretaceous: Campanian) of southern Utah. Journal of Vertebrate Paleontology. doi:10.1080/02724634.2015.1009084
Pérez-García A and Vlad Codrea 2018. New insights on the anatomy and systematics of Kallokibotion Nopcsa, 1923, the enigmatic uppermost Cretaceous basal turtle (stem Testudines) from Transylvania. Zoological Journal of the Linnean Society. 182(2):419–443. doi:10.1093/zoolinnean/zlx037.

 

Advertisements

More turtles with temporal fenestrae

Figure 1. Skull of the basal hard-shell turtle, Baena. Some of these bone IDs and their sutures differ from those from Gaffney 1979. Principally, the gray/red bone is the supratemporal, considered absent by all turtle experts when they do not recognize the pareiasaur origin of the clade.

Figure 1. Skull of the basal hard-shell turtle, Baena. Some of these bone IDs and their sutures differ from those from Gaffney 1979. Principally, the gray/red bone is the supratemporal, considered absent by all turtle experts when they do not recognize the pareiasaur origin of the clade.

Yesterday we looked at several turtles with a lateral temporal fenestra. Today a few more are presented including Baena and Kayentachelys, turtles recently added to the large reptile tree (LRT, 1201 taxa).

Figure 2. Kayentachelys skull with bones colored differently than in the original drawings.

Figure 2. Kayentachelys skull with bones colored differently than in the original drawings.

These two extinct turtles
nest between basalmost forms and extant turtles.

By convergence
several turtle clades (Fig. 3) developed various skull fenestrae, including soft-shell turtles beginning with Arganaceras (not sure if it’s a turtle or not yet) and Odontochelys.

Figure 3. Subset of the large reptile tree (LRT, 1199 taxa) with the addition of three basal turtles

Figure 3. Subset of the large reptile tree (LRT, 1199 taxa) with the addition of three basal turtles. The Mongolochelys/Chubutemys clade did not develop temporal fenestrae. Foxemys and Macrochelys had tentative occipital invagination that extended further with more derived taxa in their respective clades.

Among the most striking of the fenestrated turtle skulls
are the [cryptodire = straight neck in dorsal view, S-curve in lateral view] common Eastern box turtle (genus: Terrapene, Fig. 4) and the [pleurodire = S-curve side neck in dorsal view] matamata (genus: Chelus, Fig. 5). It’s difficult to label these two ‘anapsids’ based on their skull morphology, but that’s the traditional label.

Figure 4. Terrapene, the box turtle, with skull bones colorized. Note the lack of a dermal skull and the appearance of the cranial skull, the braincase.

Figure 4. Terrapene, the box turtle, with skull bones colorized. Note the fenestrated skull. See how colors make bones so much easier to understand. You’ll note many academic papers have been following this trend lately.

Figure 2. Chelus frmbiata, the mata-mata has a temporal fenestra. Not sure if it's a lateral or upper type. Note also the mistake made by Dr. Gaffney in overlooking the squamosal and quadratojugal, and mislabeling the supratemporal.

Figure 5. Chelus frmbiata, the mata-mata has a temporal fenestra. Not sure if it’s a lateral or upper type. Note also the mistake made by Dr. Gaffney in overlooking the squamosal and quadratojugal, and mislabeling the supratemporal. This is one skull you can easily get lost in—if you don’t color the bones. Finally, note the sidesweep of the cervicals in this pleurodire turtle.

References
Gaffney ES 1979. The Jurassic Turtles of North America. Bulletin of the American Museum of Natural History 162(3):91-136.

Basal turtles with a lateral temporal fenestra

Today let’s look at
Glyptops plicatulus (Marsh 1890; AMNH 336; Late Jurassic), an associated skull, shell, and partial skeleton (Fig. 1). Gaffney (1979) reported, “The poor preservation of the skulls precludes a detailed study of the skull roof.” That may be true. Or not. Bones appear to be lost from the temporal regions, but every temporal bone can be identified, just smaller.

Apparently Glyptops had large skull openings
like other turtles. Here the temporal bones were reduced, leaving lateral and suparaoccipital openings, like other turtles. A DGS tracing (Fig. 1) and reconstruction (Fig. 2) provide one solution. Perhaps not the only solution, but one worth considering because no bones are missing here (contra Gaffney 1979).

Figure 1. Glyptops, a basal hard-shell turtle in several views. All data from Gaffney 1979 except the color overlays, which are applied here and used to make the reconstruction in figure 2.

Figure 1. Glyptops, a basal hard-shell turtle in several views. All data from Gaffney 1979 except the color overlays, which are applied here and used to make the reconstruction in figure 2.

According to Gaffney (1979), “their sole unique feature an elongate basisphenoid extending the length of and completely separating the pterygoids.”

Figure 2. Glyptops skull reconstructed from color overlays in figure 1. Note the semi-fenestrated skull mimicking the diapsid configuration.

Figure 2. Glyptops skull reconstructed from color overlays in figure 1. Note the semi-fenestrated skull mimicking the diapsid configuration that Gaffney considered poorly preserved. Gray areas are restored based on sister taxa.

Many traits presage the appearance of traits
in derived turtles, like Terrapene, the Eastern box turtle, by convergence. The two are not directly related to one another, despite sharing several traits. In Glyptops the frontals (lavender) were separated from the parietals (amber) by intervening postfrontals (orange) and postorbitals (aqua) that meet at the midline.

Figure 3. Subset of the large reptile tree (LRT, 1199 taxa) with the addition of three basal turtles

Figure 3. Subset of the large reptile tree (LRT, 1300 taxa) with the addition of three basal turtles

Other turtles that have lateral temporal fenestrae
include the leatherback sea turtle, Dermochelys (Fig. 3, we looked at yesterday), and Meiolania (Fig. 4, now basal to Proganochelys) by convergence.

Figure 2. Skull of Dermochelys adult and juvenile demonstrating the lengthening of the temporal region during maturity. The lateral temporal fenestra appears between the squamosal and quadrate.

Figure 4. From yesterday’s blogpost, the skull of Dermochelys adult and juvenile demonstrating the lengthening of the temporal region during maturity. The lateral temporal fenestra appears between the squamosal and quadrate.

So some turtles are anapsids,
(reptiles that lack temporal openings). Others are not. None are phylogenetic diapsids, despite having large skull openings (Fig. 1) from the top and the sides.

These exceptions remind us
not to define reptiles by their traits (although most of the time this method works well), but rather by their phylogenetic placement (Fig. 3), a method that always works.

Figure 1. Meiolania has a lateral temporal fenestra created by more bone encircling the tympanum (ear drum) at the quadrate. It could be that the top of the qj is actually the fused sq.

Figure 5. Meiolania has a lateral temporal fenestra created by more bone encircling the tympanum (ear drum) at the quadrate, not related to the other temporal openings that start at the back or the bottom of the skull.

Tomorrow,
more laterally fenestrated turtles.

References
Gaffney ES 1979. The Jurassic Turtles of North America. Bulletin of the American Museum of Natural History 162(3):91-136.
Marsh OC 1890. Notice of some extinct Testudinata. American Journal of Science ser. 3, vol. 40, art. 21: 177–179.

The leatherback sea turtle: alone no longer

The leatherback turtle
(genus: Dermochelys) is different from all other sea turtles. It is the only extant genus of the family Dermochelyidae, the sister family to other sea turtles.

Dermochelys coriacea (Vandelli 1761, originally Testudo; Blainville 1816) lacks a bony shell (Fig. 1), replaced by thick, oily, leathery skin. The mouth and esophagus are filled with backward pointing spikes arising from toothess jaws (Fig. 1). Here (Fig. 2) a juvenile skull demonstrates the elevation and lengthening of the postorbital region in the adult.

Figure 2. Dermochelys skeleton, ventral view. In vivo (upper left) and open mouth (lower right).

Figure 1. Dermochelys skeleton, ventral view. In vivo (upper left) and open mouth (lower right).

Note the lack of temporal emargination
in the skull of Dermochelys (Fig. 2). That turns out to not be such a big deal in turtle evolution. The invagination occurred several times in turtles by convergence. This is something one finds out by phylogenetic analysis—if you don’t have an initial bias.

Figure 2. Skull of Dermochelys adult and juvenile demonstrating the lengthening of the temporal region during maturity. The lateral temporal fenestra appears between the squamosal and quadrate.

Figure 2. Skull of Dermochelys adult and juvenile demonstrating the lengthening of the temporal region during maturity. The lateral temporal fenestra appears between the squamosal and quadrate.

Alone no longer.
In the large reptile tree (LRT, 1200 taxa), the freshwater turtle Carettochelys (Fig. 3, 4; Ramsay 1886; 70 cm) nests with Dermochelys. Brinkman, Rabi and Zhao 2017 nested Carettochelys basal to soft shell turtles, unaware that soft shell turtles had a separate origin among small horned pareiasaurs. Like the soft-shell turtle, Trionyx, the soft nose tissue of Carettochelys extends slighly from the skull (Fig. 4) by convergence. Dermochelys (Fig. 1) does not have this proboscis.

Figure 5. Carettochelys skull in two views. Bones colored here.

Figure 3. Carettochelys skull in two views. Bones colored here. Note the long, upturned premaxilla. The invagination of the temporal region is convergent with several other clades of turtles. The supratemporal is orange. The squamosal is lavender. The quadratojugal is a vestige on the posterior maxilla. Compare to figure 2.

The pig-nosed turtle
(genus: Carettochelys) is also unique. It is the only freshwater turtle with flippers. The carapace is not scaly, but leathery (hmm, where have we hear that before?) over bone (Fig. 5).

Despite their differences
Carettochelys and Dermochelys find no closer sisters in the LRT than with each other.

FIgure 1. Carettochelys, the pig-nose turtle, is a freshwater form with flippers, like marine turtles, by convergence.

Figure 4. Carettochelys, the pig-nose turtle, is a freshwater form with flippers, like marine turtles, by convergence.

Why was this not discovered earlier?
Mistaking the supratemporal for the squamosal was only part of the problem. Fusion of the skull bones in turtles (as in birds) gives paleontologists trouble. The dual origin of turtles was not previously considered a possibility.

Figure 1. Carettochelys in 3 views from Digimorph.org and used with permission.

Figure 5. Carettochelys in 3 views from Digimorph.org and used with permission. The leatherback lost its bony carapace.

When workers expand their taxon list
they will recover what the LRT recovers. Until now (Fig. 6), unfortunately, that has not happened.

Figure 3. Subset of the large reptile tree (LRT, 1199 taxa) with the addition of three basal turtles

Figure 6. Subset of the large reptile tree (LRT, 1199 taxa) with the addition of three basal turtles

References
Brinkman D, Rabi M and Zhao L-J 2017. Lower Cretaceous fossils from China shed light on the ancestral body plan of crown soft-shell turtles (Trionychidae, Cryptodira). Nature Scientific Reports 7(6719).
Gaffney ES 1979. Comparative cranial morphology of recent and fossil turtles. Bulletin of the American Museum of Natural History 164(2):65-376.
Ramsay EP 1886. On a new genus and species of fresh water tortoise from the Fly River, New Guinea. Proceedings of the Linnaean Society of New South Wales (2) 1: 158-162.

wiki/Carettochelys
wiki/Dermochelys

Dermochelys and Carettochelys in ReptiliaEvolution.com

 

Kallokibotion: a late-surviving, basal, hard-shell turtle

A recent paper by Pérez-Garcia and Codrea 2018
on the basal, but late-surviving turtle, Kallokibotion bajazidi, (Nopsca 1923a,b) brings us more specimens and more hypotheses of turtle relationships. This taxon has been described as ‘an enigma’, perhaps because all currently known specimens combine Late Triassic traits with a Latest Cretaceous occurrence The visibility of the naris in lateral view goes back to the pre-turtle, Elginia. Transitional taxa, like Meiolania (Fig. 2) and Proganochelys, do not have this trait. Essentially this taxon links the Proganochelys clade to all later hard-shell turtles. (Remember, soft-shell turtles had a separate origin).

Figure 2. Kallokibotion skull in 5 views. Note the twin nares.

Figure 1. Kallokibotion skull in 5 views. Note the twin nares. This skull retains many meiolanid traits.

Gaffney and Meylan 1992 report, “Kallokibotion is a cryptodire because it has the otic trochlea synapomorphy of cryptodires, and it is a member of the Selmacryptodira because it has a posterior pterygoid process under the middle ear. It lacks the posterior temporal emargination synapomorphic of the Daiocryptodira and lies outside that group.”

I’m not going to discuss the suprageneric taxa listed above
because no prior turtle phylogenies recognized the diphyletic nature of soft-shell and hard-shell turtles.

Figure 3. Subset of the large reptile tree (LRT, 1199 taxa) with the addition of three basal turtles

Figure 2. Subset of the large reptile tree (LRT, 1199 taxa) with the addition of three basal turtles

Kallokibotion links
Niolamia and Miolania (Fig. 3 ) with higher turtles, skipping over Proganochelys (Late Triassic, Fig. 4), the traditional baalmost turtle.

 

Figure 3. Kallokibotion compared to Meiolania.

Figure 3. Kallokibotion compared to Meiolania.

Kallokibotion was recognized as a member of Meiolaniformes
in several recently published papers (see Rabi et al., 2013b; Sterli & de la Fuente, 2013; Sterli et al., 2015a, b). Chubutemys is considered a meiolaniform without horns, but workers who proposed this clade did not realize meiolaniids were basalmost hard-shell turtles.

Figure 2. The skull of Proganochelys, a basal turtle without skull invagation and without horns.

Figure 4. The skull of Proganochelys, a basal turtle close to Kallokibation, Note the squamosal and supratemporal and compare those to Kallokibation in  Fig. 1.

Figure 2. Kallokibation carapace and plastron from Gaffney and Melton 1992.

Figure 5. BMNH R4918 specimen of Kallokibation carapace and plastron from Gaffney and Melton 1992. This is the first of the tested taxa that has a short, unarmored tail. The skull was relatively small and unarmored.

Pérez-Garcia and Codrea 2018
discovered new specimens (Fig. 1) that “not only reveal detailed cranial and postcranial elements poorly known until now, refuting previous hypotheses about the anatomy of this taxon, but also allow us to identify numerous hitherto unknown characters.” The new information, “shows Kallokibotion as the sister taxon of the crown Testudines.”

A lot of discussion about crown-group turtles goes out the window
when taxa are included that split hard from soft-shell taxa before turtles had shells. Thus all known turtles, and several non-turtle pareiasaurs, fall under the current definition of crown turtles in the LRT.

Here 
(Fig. 3) Kallokibotion does indeed nest close to the meiolanids and Kayentachelys, but other taxa intervene. The loss of horns and a long armored tail on higher hard-shell turtles, like Kallokibotion, can be attributed to neotony.

Pérez-Garcia and Codrea had no idea what turtles are.
They nest the pre-lepidosauriform, Owenetta, the pareiasaur, Anthodon, the plesiosaur, Simosaurus, and the rhynchocephalian, Sphenodon as progressively more distant outgroup taxa. Throw those out and, except for the meiolanids, their pruned turtle cladogram is similar to the same subset of the LRT.

References
Dyke GJ et al. (20 co-authors) 2014. Thalassodromeus sebesensis—a new name for an old turtle. Comment on ‘Thalassodromeus sebesensis, an out of place and out of time Gondwanan tapejarid pterosaur’, Grellet-Tinner and Codrea. Gondwana Research. doi:10.1016/j.gr.2014.08.004.
Gaffney ES and Meylan PA 1992. The Transylvanian turtle, Kallokibotion, a primitive cryptodire of Cretaceous Age. American Museum Novitates (3040).
Nopcsa F 1923a. On the geological importance of the primitive reptilian fauna of the Uppermost Cretaceous ofHungary; with a description of a new tortoise (Kallokibotion). Quart. Jour. Geol. Soc. 79(1): 100—116.
Pérez-García A and Vlad Codrea 2018. New insights on the anatomy and systematics of Kallokibotion Nopcsa, 1923, the enigmatic uppermost Cretaceous basal turtle (stem Testudines) from Transylvania. Zoological Journal of the Linnean Society. 182(2):419–443. doi:10.1093/zoolinnean/zlx037.
Zhou CF et al. 2015. A sinemydid turtle from the Jehol Biota provides insights into the basal divergence of crown turtles, Scientific Reports (2015). DOI: 10.1038/srep16299

Read more at:
https://phys.org/news/2015-11-insights-family-tree-modern-turtles.html#jCp
wiki/Kallokibotion
https://www.earthmagazine.org/article/trouble-turtles-paleontology-crossroads

Enigmatic Perochelys and a review of soft-shell turtle origins

In short:
Current turtle workers are under the mistaken assumption that Carettochelys (Fig. 1) the tube-nosed soft-shell turtle mimic with a domed hard shell and flippers is the outgroup for soft-shell turtles. That is not supported by the large reptile tree (LRT, 1176 taxa) which nests soft-shell turtles apart from hard-shell turtles, both derived from separate small, horned pareiasaurs like Scerlosaurus and Elginia respectively.

With that in mind,
it’s no wonder that two prior authors don’t know where to nest the Early Cretaceous soft-shell turtle, Perochelys (Figs. 2, 3), as derived or basal in the soft-shell clade. Li et al. 2017 and Brinkman et al. 2017 don’t even mention the basalmost soft-shell turtle, Odontochelys, let alone ancestral  taxa like Scerlosaurus and Arganaceras.

Figure 1. Carettochelys in 3 views from Digimorph.org and used with permission.

Figure 1. Carettochelys in 3 views from Digimorph.org and used with permission.

“Trionychidae plus Carettochelyidae form the clade Trionychia (Gaffney and Meylan, 1988; Meylan, 1988; Meylan and Gaffney, 1989; Shaffer et al., 1997; Joyce et al., 2004; Joyce,
2007).”

FIgure 1. Carettochelys, the pig-nose turtle, is a freshwater form with flippers, like marine turtles, by convergence.

FIgure 2. Carettochelys, the pig-nose turtle, is a freshwater form with flippers, like marine turtles, by convergence.

In the large reptile tree
(LRT, 1176 taxa) where we test as many taxa as possible and let the nodes form where they may, the tube-nosed, dome-shelled fresh water turtle with flippers, Carettochelys, nests with Foxemys in the hard-shell clade as a soft-shell turtle mimic. Only the LRT nests Sclerosaurus, Arganaceras and Odontochelys in the outgroup for soft-shell turtles.

“Molecular studies place this clade at the base of crown group Cryptodira (Shaffer et al., 1997; Krenz et al., 2005; Parham et al., 2006; Shaffer, 2009; Barley et al., 2010; Louren¸co et al., 2012), whereas unconstrained morphological studies support a more derived position nested within Cryptodira (Gaffney and Meylan, 1988; Joyce, 2007; Sterli, 2010; Anquetin, 2011; Sterli et al., 2013).”

Figure 4. The skull of Carettochelys in 5 views. This skull shares some traits with Trionyx, but more with Foxemys.

Figure 3. The skull of Carettochelys in 5 views. This skull of this dome-shell turtle shares some traits with the soft-shell Trionyx, but more with the dome-shell Foxemys. Comnpare to Trionyx in figure 4 and you’ll see why convergence has confused the issue of soft turtle origins. Don’t try to figure out turtle origins by yourself. Let the software do it without bias.

“The phylogenetic relationships among modern soft-shelled turtle species are still controversial, but it is generally accepted that Trionychidae consists of two clades, Cyclanorbinae and Trionychinae, and that Trionychinae includes some well-supported monophyletic clades (Meylan, 1987; Engstrom et al., 2004). The taxonomy and phylogenetic relationships of fossil trionychid species are far more controversial, and very little is known regarding the origin and early radiation of this group (Gardner et al., 1995; Joyce and Lyson, 2010, 2011; Vitek and Danilov, 2010; Vitek, 2012; Danilov and Vitek, 2013; Joyce et al., 2013).”

As I said… see above.

Figure 3. Trionyx, a softshell turtle with bones colorized.

Figure 4. Trionyx, a softshell turtle with bones colorized.

“The early record of soft-shelled turtles is poor, and most taxa are based either on fragmentary shells or skulls (Yeh, 1994; Hutchison, 2000; Sukhanov, 2000; Danilov and Vitek, 2013). More complete Mesozoic skull-shell-associated materials have been described only for trionychids from the Campanian and Maastrichtian of North America (Gardner et al., 1995; Brinkman, 2005; Joyce and Lyson, 2011; Vitek, 2012) or the Cenomanian–Santonian of Mongolia (Danilov et al., 2014). The new material described herein is a nearly complete skeleton and therefore represents the first complete Early Cretaceous skull shell-associated trionychid worldwide.”

Figure 1. Perochelys (Early Cretaceous) in situ

Figure 5. Perochelys (Early Cretaceous) in situ from Li et al. 2015) colors added.

Perochelys lamadongensis (Early Cretaceous)

Figure 2. Perochelys skull in dorsal and ventral views.

Figure 6. Perochelys skull in dorsal and ventral views from Li et al. 2015 with colors added.

Brinkman et al. 2017 looked at another specimen of Perochelys.

Here’s the abstract:
“Pan-trionychids or softshell turtles are a highly specialized and widespread extant group of aquatic taxa with an evolutionary history that goes back to the Early Cretaceous. The earliest pan-trionychids had already fully developed the “classic” softshell turtle morphology and it has been impossible to resolve whether they are stem members of the family or are within the crown. This has hindered our understanding of the evolution of the two basic body plans of crown-trionychids. Thus it remains unclear whether the more heavily ossified shell of the cyclanorbines or the highly reduced trionychine morphotype is the ancestral condition for softshell turtles.”

Softshell turtles never had a heavily ossified shell as demonstrated by Odontochelys and Sclerosaurus, taxa excluded from all prior soft-shell turtle studies.

Figure 7. Trionyx, an African soft-shelled turtle with fossil relatives back to the Cretaceous nests with Odontochelys.

Figure 7. Trionyx, an African soft-shelled turtle with fossil relatives back to the Cretaceous nests with Odontochelys.

“A new pan-trionychid from the Early Cretaceous of Zhejiang, China, Perochelys hengshanensis sp. nov., allows a revision of softshell-turtle phylogeny. Equal character weighting resulted in a topology that is fundamentally inconsistent with molecular divergence date estimates of deeply nested extant species. In contrast, implied weighting retrieved Lower Cretaceous Perochelys spp. and Petrochelys kyrgyzensis as stem trionychids, which is fully consistent with their basal stratigraphic occurrence and an Aptian-Santonian molecular age estimate for crown-trionychids. These results indicate that the primitive morphology for soft-shell turtles is a poorly ossified shell like that of crown-trionychines and that shell re-ossification in cyclanorbines (including re-acquisition of peripheral elements) is secondary.”

That’s what I’ve been trying to tell turtle workers.
And I presented the phylogenetic evidence in Odontochelys and Sclerosaurus. Brinkman et al. do not present these taxa.

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

Figure 8. Soft shell turtle evolution featuring Arganaceras, Sclerosaurus, Odontochelys and Trionyx.

Distinct from soft-shell turtles, hard-shell turtles have:

  1. domed carapace with scutes
  2. dorsal rib tips not visible
  3. premaxilla and maxilla curved one way or another
  4. large quadratojugal, even when fused to the squamosal above it
  5. large premaxilla (forming the ventral margin of the confluent nares
  6. nasal fused to prefrontal
  7. postorbital fused to postfrontal
  8. an ancestry with a broad, bony, convex cranium, which erodes convergent with soft-shell taxa

Like soft-shell turtles, soft-shell turtle mimics with domed hard shells often have:

  1. orbits visible in dorsal view
  2. elongate cervicals
  3. posttemporal fenestra at least half the skull length (but never(?) reaching the jugal)
  4. slender digits
  5. posteriorly elongate supraoccipital with inverted ‘T’ cross-section

Bottom line:
Don’t try to figure out turtle origins by yourself. Let the software do it without bias. 

References
Li L, Joyce WG and Liu J 2015. The first soft-shelled turtle from the Jehol Biota of China. Journal of Vertebrate Paleontology 35(2):e909450. 2015
Brinkman D, Rabi M and Zhao L-J 2017. Lower Cretaceous fossils from China shed light on the ancestral body plan of crown softshell turtles (Trionychidae, Cryptodira). Scientific Reports 2017(7):6719.

Turtles with wings

Figure 1. Manus of Carettochelys, the pig-nosed turtle, resembles the wing of other tetrapods.

Figure 1. Manus of Carettochelys, Note the crest posterior to the shoulder joint.

Yes, underwater wings.
We’re talking today about the pig-nosed turtle, Carettochelys insculpta (Figs. 1-3), which became interesting when Brinkman, Rabi and Zhao 2017 nested it basal to soft-shell turtles. The large reptile tree (LRT, 1043 taxa, subset Fig. 3) does not replicate those results. Rather the LRT nests Carettochelys with Foxemys.

Carettochelys insculpta (Ramsay 1886; 70 cm) is the extant pig-nosed turtle. Unlike any other species of freshwater turtle, the feet are flippers, like the marine sea turtle Chelonia. The carapace is not scaly, but leathery. It remains domed and the solid plastron is strongly connectedd to the carapace. Brinkman, Rabi and Zhao 2017 nested Carettochelysbasal to soft shell turtles, but the large reptile tree nests it with Foxemys. Like Trionyx, the nose extends slighly from the skull.

FIgure 1. Carettochelys, the pig-nose turtle, is a freshwater form with flippers, like marine turtles, by convergence.

FIgure 2. Carettochelys, the pig-nose turtle, is a freshwater form with flippers, like marine turtles, by convergence. The nose is tubular like soft shell turtles.

Not sure why
Brinkman, Rabi and Zhao 2017 nest Carettochelys with soft shell turtles, but I suspect it has to do with taxon exclusion (a limited gamut of tested taxa) and an improper traditional inclusion.

Figure 3. Carettochelys in 3 views from Digimorph.org and used with permission.

Figure 3. Carettochelys in 3 views from Digimorph.org and used with permission.

The skull of Carettochelys
includes large and extensive postorbital fenestrae. The jugal is quite tiny. The squamosal (blue) and quadratojugal (beige) are fused, as in sister taxa. The supratemporal (orange) has been traditionally mislabeled as a squamosal.

Figure 4. The skull of Carettochelys in 5 views. This skull shares some traits with Trionyx, but more with Foxemys.

Figure 4. The skull of Carettochelys in 5 views. This skull shares some traits with Trionyx, but more with Foxemys.

As an experiment
I deleted all taxa other than turtles (Fig. 5) and decided that Proganochelys would be the outgroup to match the analyses of other workers. Even so, soft shell turtles do not nest with Carettochelys. 

Figure 2. Subset of the LRT composed on only turtles and with Proganochelys as the outgroup.

Figure 5. Subset of the LRT composed on only turtles and with Proganochelys as the outgroup.

A subset of the LRT
(Fig. 6) shows the relationship of soft shell and hard (dome) shell turtles to pareiasaurs. Note: turtles are not monophyletic, unless you also include the pareiasaurs Bunostegos and Arganaceras, which I do here to document the clade of crown turtles. The LRT includes enough characters to separate soft shell turtles from others, despite a long list of similar traits. That should give one great confidence that the character list is sufficient at its present number.

FIgure 3. Subset of the LRT including turtles and their kin.

FIgure 6. Subset of the LRT including turtles and their kin. Pleurodires are side-neck turtles.

Marine turtles with flippers (underwater wings)
include Dermochelys, the extant leatherback turtle (Fig. 7), convergent with Carettochelys. The LRT includes enough traits to separate these two similar yet distinct taxa.

Figure 2. Dermochelys skeleton, ventral view. In vivo (upper left) and open mouth (lower right).

Figure 7. Dermochelys skeleton, ventral view. In vivo (upper left) and open mouth (lower right).

References
Brinkman D, Rabi M and Zhao L-J 2017. Lower Cretaceous fossils from China shed light on the ancestral body plan of crown soft-shell turtles (Trionychidae, Cryptodira). Nature Scientific Reports 7(6719).
Ramsay EP 1886. On a new genus and species of fresh water tortoise from the Fly River, New Guinea. Proceedings of the Linnaean Society of New South Wales (2) 1: 158-162.

wiki/Pig-nosed_turtle
http://digimorph.org/specimens/Carettochelys_insculpta/