Where does Rooneyia nest on the primate cladogram?

Rooneyia viejaensis (Kirk et al. 2014; Late Eocene, 40 mya; Fig. 2) is a basal primate known from a nearly complete skull. Everyone agrees on that. Where Rooneyia nests within the Primates is the point of contention.

FIgure 1. Subset of the LRT focusing on the primate/bat clade. Rooneyia nests between lemurs and higher primates.

FIgure 1. Subset of the LRT focusing on the primate/bat clade. Rooneyia nests between lemurs and higher primates.

Kirk et al. 2014 report: 
“Rooneyia viejaensis is a North American Eocene primate of uncertain phylogenetic affinities. Although the external cranial anatomy of Rooneyia is well studied, various authors have suggested that Rooneyia is a stem haplorhine, stem strepsirrhine, stem tarsiiform, or stem anthropoid.”

The large reptile tree (LRT, 1051 taxa) nests Rooneyia between lemur-like Notharctus and all higher primates, including tarsiers like Tarsius and Darwinius. Granted there are not very many primates on the LRT. Nevertheless, those are the current results. So the LRT indicates or suggests that Rooneyia is a stem hapolorhine, stem strerpsirrhine, stem trasiiform AND stem anthropoid. I’ll have to add more taxa in these clades to make a more precise recovery.

Figure 1. Rooneyia images from Digimorph.org and used with permission. White background and overlying DGS colors added here. The basal tree shrew/primate, Ptilocercus, is shown to scale.

Figure 2. Rooneyia images from Digimorph.org and used with permission. White background and overlying DGS colors added here. The basal tree shrew/primate, Ptilocercus, is shown to scale. The postorbital bar and canine depth on Rooneyia are imagined. Does not color help one understand the bones so much better and more quickly?

References
Kirk EC, Daghighi P, Macrini TE, Bhullar B-AS and Rowe TB 2014. Cranial anatomy of the Duchesnean primate Rooneyia viejaensis: new insights from high resolution computed tomography. Journal of Human Evolution, 74, 82-95). online here.

Advertisements

Pterosaur wing appearance when quadrupedal

Figure 1. GIF animation (2 frames) showing the original and repaired versions of this Quetzalcoatlus statue and its artisans.

Figure 1. GIF animation (2 frames) showing the original and repaired versions of this Quetzalcoatlus statue and its artisans. Pterosaur wings folded to near invisibility when folded as shown by manipulating bones and observing fossils. Only a narrow chord wing membrane, as in figure 2, works here. Note the unwarranted wrinkles in the original wing membrane. Note in the original the trailing edge of the wing membrane is directed to mid-metacarpal, rather than to the 180º metacarpophalangeal joint as in the repaired version, fossils and other illustrations below. And what is going on with those tiny anterior pteroids? That is wrong, so wrong.

Pterosaur workers
artists and filmmakers have struggled to portray pterosaur wing membrane when the wings are folded and the pterosaur is walking around (Figs. 1, 3–5). Fossils (Fig. 2) that show how the wings looked when folded are too often ignored.

Figure 4. Here's how the wing membrane in pterosaurs virtually disappeared when folded. This is CM 11426 (no. 44 in the Wellnhofer 1970 catalog),

Figure 2. Here’s how the wing membrane in pterosaurs virtually disappeared when folded. This is CM 11426 (no. 44 in the Wellnhofer 1970 catalog), Note: the left wing has been axially rotated during taphoonmy such that the folded portion of the membrane was fossilized posterior to the bony spar.

CM11426
(Fig. 2) shows how wing membranes fold down whenever the wing bones are flexed (folded). Like bats, pterosaur wing membranes fold away to near invisibility. If you think CM11426 looks a bit like Quetzalcoatlus, you’re right! It’s in the lineage in the large pterosaur tree (LPT), but it’s not larger than a typical Pterodactylus (Fig. 9).

Stan Winston Pteranodon suit for Jurassic Park 3.

Figure 3. Stan Winston Pteranodon suit for Jurassic Park 3. Those wrinkled wing membranes are a dead giveaway that Winston was lost when it came to wing folding in Pteranodon. And that’s not to mention the too short metacarpals (see fig. 4 for comparison)

For Jurassic Park 3
Stan Winston’s Pteranodon (Fig. 3) had saggy, baggy wing membranes. So did early paintings by Burian (Fig. 4). These clearly do not reflect what happens in fossils and in life.

Figure 4. Artist Z. Burian also struggled to realistically portray the folded wing membrane in pterosaurs forgetting the fossils and the fact that both birds and bats have no trouble folding their wings without wrinkling them.

Figure 4. Artist Z. Burian also struggled to realistically portray the folded wing membrane in pterosaurs forgetting the fossils and the fact that both birds and bats have no trouble folding their wings without wrinkling them.

Too often 
artists freehand their pterosaurs (Fig 5 purple), ignoring the bone and soft tissue evidence.

wo of the most completely known Pteranodon

Figure 4. Two of the most completely known Pteranodon (UALVP24238 and NMC4138) along with the skull of KUVP2212 to scale. In purple, John Conway’s Pteranodon (purple) with a much smaller skull and an inappropriate  elbow-high walking configuration.

Toy Pteranodon, ca. 1962, from the Marx Company.

Figure 5. Toy Pteranodon, ca. 1962, from the Marx Company.

Toy pterosaurs
also suffer from deep chord wing membranes (Fig. 5). Proportions here are wildly inaccurate for the toddler set. Accuracy is also absent in many professional reconstructions that include skeletons (Fig. 4), so there is enough blame to go around. The fossils (Fig. 6) document how the artists and sculptors should present those folded wing membranes. Too few artists and sculptors who claim accuracy are actually producing accuracy.

Figure 1. Click to enlarge. The plate and counter plate of the BSP AS V 29a/b specimen of Pterodactylus with color overlays of the bones and visible soft tissues.

Figure 6. The plate and counter plate of the BSP AS V 29a/b specimen of Pterodactylus with color overlays of the bones and visible soft tissues.

What do you get when you choose accuracy?
A much less monstrous awkward portrayal and a much more elegant bird-like/bat-like portrayal comes from keeping true to the bones and soft tissues as they are. Deep chord wing membranes that attach to pterosaur ankles are as outdated as tail-dragging dinosaur portrayals. And while we’re at it, keep those pteroids pointing inward, forming straight leading edges for the distal propatagia.

Pterodactylus walking. Note the foot will never plant itself in front of the hand here. And why are both hands on the ground at the same time as the back foot? Hmm.

Figure 7. Pterodactylus walking. Note the foot will never plant itself in front of the hand here. And why are both hands on the ground at the same time as the back foot? Hmm.

Pterodactylus walk matched to tracks according to Peters

Figure 8. Click to animate. Plantigrade and quadrupedal Pterodactylus walk matched to tracks

This animation frame (Fig. 7) from a walking pterosaur movie associated with the Crayssac tracks accurately portrays the wing membrane essentially invisible when folded. Artist unknown.

Another animation matched to Crayssac tracks (Fig. 8) does not include wing membranes, but they would have been nearly invisible here. This version shows a more upright quadrupedal stance, as if the pterosaur wings were used like ski poles. As noted earlier, this is essentially a bipedal pose, enabling wing opening and flapping without shifting the center of balance.

Go with the evidence, not traditional and sometimes current renderings. Follow the evidence.

The Vienna Pterodactylus.

Figure 9. The Vienna Pterodactylus.  Wing membranes in situ (when folded) then animated to extend them. There is no shrinkage here or in ANY pterosaur wing membrane. That is only an “explanation” to avoid dealing with the hard evidence here and elsewhere.

While we’re talking about Quetzalcoatlus
and its flying abilities, it is worthwhile to take another look at gracile m4.2 (second wing phalanx) on the giant Q. northropi vs. the same phalanx on the much smaller and more likely volant Q. species (Fig. 10). Sorry I didn’t bring this up during the earlier discussion, on azhdarchid flight, first published online three years ago here, but I forgot I had it, and it’s more damning evidence against giant pterosaur flight.

Figure 1 Quetzalcoatlus sp. compared to the large specimen wing, here reduced. I lengthened the unknown metacarpus to match the Q. sp. and other azhdarchid metacarpi. I offer the wing finger has reconstructed by the Langson lab and with filler reduced. Note m4.2 is narrower on the larger specimen, which doesn't make sense if Q. northropi was volant.

Figure 10. Quetzalcoatlus sp. compared to the large specimen wing, here reduced. I lengthened the unknown metacarpus to match the Q. sp. and other azhdarchid metacarpi. I offer the wing finger has reconstructed by the Langson lab and with filler reduced. Note m4.2 is narrower on the larger specimen, which doesn’t make sense if Q. northropi was volant.

Figure 1. Freehand wing planform cartoon for Quetzalcoatlus from Witton and Habib 2010. There is no evidence in any pterosaur for this wing plan.

Figure 11. Freehand wing planform cartoon for Quetzalcoatlus from Witton and Habib 2010. There is no evidence in any pterosaur for this wing plan. Such deep chord wings cannot help but create unwarranted wrinkles when folded.

Wing folding
and the muscles that enabled complete flexion (Fig. 12) were covered earlier here.

Figure 1. Pterosaur (Santanadactylus) wing folding. Note when the wing is perpendicular to the metacarpus the flexor must be off axis in order to complete the wing folding process. The insertion must be distal to the joint because the flexor process of m4.1 extends dorsally over the metacarpus during wing folding. Otherwise the ventral (palmar) flexor would be cut off from the swinging dorsal process.

Figure 12. Pterosaur (Santanadactylus) wing folding. Note when the wing is perpendicular to the metacarpus the flexor must be off axis in order to complete the wing folding process. The insertion must be distal to the joint because the flexor process of m4.1 extends dorsally over the metacarpus during wing folding. Otherwise the ventral (palmar) flexor would be cut off from the swinging dorsal process.

References
Witton MP and Habib MB 2010.
On the size and flight diversity of giant pterosaurs, the use of birds as pterosaur analogues and comments on pterosaur flightlessness. PlosOne 5(11): e13982. doi:10.1371/journal.pone.0013982

SVP abstracts 2017: Eudibamus forelimb description

Sumida et al. 2017 bring us new information
on the pectoral region of Eudibamus, (Figs. 1,2) an early likely biped in the sprawling manner of the unrelated extant iguanian lizards, Chlamydosaurus and Basiliscus by convergence.

Unfortunately,
Sumida et al. continue to cling to the invalidated tradition that Eudibamus is a bolosaurid, largely based on convergent tooth shapes and taxon exclusion in their analyses.

Figure 1. Basal diasids and proto-diapsids. Largely ignored these putative synapsids actually split from other synapsids while retaining the temporal fenestra trait that serves as the basis for the addition of upper temporal fenestra in diapsids. Included here are Protorothyris, Archaeovenator, Mycterosaurus, Heleosaurus, Mesenosaurus, Broomia, Milleropsis, Eudibamus, Petrolacosaurus, Spinoaequalis, and Tangasaurus.

Figure 1. Basal diasids and proto-diapsids. Largely ignored these putative synapsids actually split from other synapsids while retaining the temporal fenestra trait that serves as the basis for the addition of upper temporal fenestra in diapsids. Included here are Protorothyris, Archaeovenator, Mycterosaurus, Heleosaurus, Mesenosaurus, Broomia, Milleropsis, Eudibamus, Petrolacosaurus, Spinoaequalis, and Tangasaurus.

From the Sumida et al. abstract
Eudibamus cursoris, a bolosaurid parareptile, from the Early Permian Tambach Formation (approximately 290 mybp), Thüringer Wald (Thuringian Forest), of central Germany, has been interpreted as the earliest known facultative biped. This was initially proposed based on the postcranial limb proportions in the type specimen (MNG [Museum der Natur, Gotha, Germany] 8852), but the forelimb itself has never been formally described. A nearly complete left, and partial right forelimb are preserved in the type specimen. The forelimb is less than 60% the length of the hindlimb. Only a thin, blade-like scapula is visible. Brachial, antebrachial, and manual elements are slender and elongate compared to those of other basal amniotes. The humerus has two well developed distal condyles with terminally facing articular facets. Delto-pectoral attachments were along a narrow ridge. The radius and ulna are nearly subequal in length. Conspicuously, the ulna lacks a well developed olecranon process. Carpals are proximodistally elongate compared to other basal amniotes. The intermedium and lateral centrale and the radiale and medial centrale articulate end-to-end, and their combined lengths equal that of the ulnare; the intermedium and radiale, and the medial and lateral centralia are equal in length. Four distal carpals are visible, it is unclear whether whether the fifth is truly absent or simply unossified. The distal carpal associated with digit two is reduced to a tiny pebble of bone, whereas that associated with digit four is largest and somewhat wedge shaped. Four metacarpals, likely equivalent to digits two-five, are present. The proximal portion of metacarpal two is present but length of the entire element cannot be determined. No elements of digit one can be seen, though its absence could be an artifact of preservation; however, the presence of only four distal carpals suggests Eudibamus may have had only four manual digits. Three phalanges are preserved in digits three and four. Both come to blunt tips and neither exhibits a significantly elongate penultimate element. The overall limb proportions seen in Eudibamus could suggest facultative bipedality or vertical clinging and leaping. However, vertical clingers and leapers normally have at least one is proportionately elongate manual digit and well-developed manual claws. Neither phalangeal proportions, nor the two well-developed terminal phalanges show such adaptations in Eudibamus and its interpretation as a facultative biped remains the most plausible interpretation of its postcranial anatomy.”

Figure 1. Click to enlarge. Eudibamus in situ (above), traced (middle) and reconstructed (below). The revised skull retains a large orbit and has a shorter rostrum.

Figure 1. Click to enlarge. Eudibamus in situ (above), traced (middle) and reconstructed (below). The revised skull retains a large orbit and has a shorter rostrum.

First of all,
Parareptilia has been invalidated as a monophyletic clade since 2012. 

Figure 2. Eudibamus skull revised here with new data compared to bolosaurids, on the left, and basal diapsids, on the right. Post crania for bolosaurids is very fragmentary. Bolosaurids are related to pareiasaurs and turtles, all derived from millerettids. Can you see why Eudibamus was confused with bolosaurids?

Figure 2. Eudibamus skull revised here with new data compared to bolosaurids, on the left, and basal diapsids, on the right. Post crania for bolosaurids is very fragmentary. Bolosaurids are related to pareiasaurs and turtles, all derived from millerettids. Can you see why Eudibamus was confused with bolosaurids?

Since 2011
Eudibamus has nested with other slender, speedy, basalmost archoauromorph diapsids (Araeoscelis, Petrolacosaurus and kin) (Fig. 1) in the large reptile tree, far from the squat, slow, bolosaurids, like Bolosaurus and Belebey that nest with diadectids and pareiasaurs.

Let’s look again
at the pectoral region and forelimb of Eudibamus as listed by Sumida et al. above. Note how many of these traits are also present in basal archosauromorph diapsid taxa and their outgroups shown in figure 1 above. Bolosaurids, by contrast, are known chiefly by skull material, so direct comparisons to forelimbs cannot be made.

Imagine the co-authors, grad students 
who disagree with Dr. Sumida on the phylogenetic position of Eudibamus, perhaps after testing a larger gamut of taxa or by reading this blog. All co-authors sign that they agree with what is in the abstract. This is how paleontology puts on blinders, clings to traditions and generally avoids rocking the hypotheses of senior professors.

Fortunately
non-academic renegades and independent researchers have no such restrictions, but are free to explore and experiment.

References
Sumida SS et al. 2017. Structure of the pectoral limb of the early Permian bolosaurid reptile Eudibamus cursors: further evidence supporting it as the earliest known facultative biped. SVP abstracts 2017.

Sumida 2009 Ted Talk video
What is Eudibamus?

SVP abstracts 2017: Are pinnipeds (seals/sea lions) monophyletic?

Earlier the large reptile tree (LRT, 1050 taxa) invalidated the former clade Pinnipedia (seals and kin) as it split it into two clades, each derived from separate terrestrial limbed ancestors. Now comes this well written abstract from Paterson et al. 2017 that brings up all the right questions. The question is, did it have the right outgroups? I like how they say they are going to test tradition with genes and fossils. Unfortunately, they might lack a few pertinent outgroup taxa.

Figure 1. Phoca the phocid seal is most closely related to Palaeosinopa of all tested taxa.

Figure 1. Phoca the phocid seal is most closely related to Palaeosinopa of all tested taxa.

From the Paterson et al. abstract:
“Monophyly of pinnipeds is well-established. However, it is difficult to reconcile a monophyletic origin of pinnipeds with the disparate locomotor modes and associated skeletal morphologies observed between the extant families. Furthermore, the fossil record suggests many of the conventional pinniped synapomorphies arose independently, as many are not present in fossil taxa (Eotaria, Prototaria, Devinophoca) that have been firmly established as early-diverging crown members of the three extant families (e.g., homodont dentition, loss of fossa muscularis, reduction of nasolabialis fossa, loss of M2/m2, fusion of tibia and fibula, reduction of fossa for teres femoris).

“Herein, we test the hypothesis that otarioids (otariids + odobenids) and phocids share a common ancestor that was not yet fully aquatic. In the present analysis, a total evidence approach was employed to investigate the relationships of 19 extant and 37 fossil caniforme genera. Our analysis sampled five genes totalling 5490 bp and 184 morphological characters, sampled relatively evenly across morphological partitions (cranial, dental, postcranial). With Canis as an outgroup, Bayesian inference produced strong support for a monophyletic origin of pinnipeds, and recovered Puijila and Potamotherium as early-diverging pinnipedimorphs

(Ursidae(Musteloidea(Potamotherium(Puijila(Enaliarctos, (Desmatophocidae(Phocidae,(Odobenidae, Otariidae)))))))). Similar results were obtained from Bayesian and parsimony analyses of a morphology-only data set, a cranial-only data set, a craniodental-only data set, and a post-cranial-only data set. Bayesian inference of morphology-only partitions recovered Mustelavus and a sister grouping of Allocyon + Kolponomos along the stem to later-diverging pinnipedimorphs. The parsimony analysis recovered 20 synapomorphies of Potamotherium + Puijila + Pinnipedimorphs, and nine synapomorphies for a crown group Pinnipedia, to the exclusion of the pinnipedimorphs. In spite of a reinterpretation of the plesiomorphic state of many previously proposed pinniped synapomorphies, there remain more than enough pinniped synapomorphies to exclude the semi-aquatic  pinnipedimorphs, thereby challenging our hypothesis of a dual origin of flippers. However, this may be an artifact of a Bayesian model of morphological inference which, among other limitations, cannot model direction evolution, and thus, may be incapable of capturing parallel evolution in such a context.”

Figure 3. Zalophus, an otariid seal, is most closely related to Hyoposodus among tested taxa in the LRT

Figure 2. Zalophus, an otariid seal lion, is most closely related to Hyoposodus among tested taxa in the LRT. Seals and sea lions are incredibly alike. It’s a tribute to the authority of the LRT that it was able to separate these two clades, both derived from distinct and different terrestrial ancestors.

I don’t have their complete taxon list. 
If it includes the pertinent taxa that split the Pinnipedia, then we’ll have to reexamine the data. If not, it’s still worth comparing.

Their choice of
Canis (the wolf/dog) as an outgroup is a weakness. They should have let a large gamut of mammals decide the outgroup(s) for pinnipeds. I don’t see Palaeosinopa in their published taxon list, but it might be in there somewhere. In the LRT it nests with phocids, like Phoca and the limbed carnivore, PujiliaI don’t see Miacis and Hyopsodus in their taxon list. In the Paterson et al. study Enaliarctos nests basal to all extant pinnipeds, but in the LRT, Enaliarctos nests with Miacis, Hyopsodus and Zalophus, the California sea lion, an otariid. So, it looks like taxon exclusion is present here, yet again. Paterson et al. appear to be missing some pertinent outgroups. The last common ancestor of seals and sea lions goes back to something like Herpestes, the mongoose, and/or Procyon, the raccoon.

Seals and sea lions are incredibly alike.
It’s a tribute to the authority of the LRT that it was able to separate these two clades, both derived from distinct and different terrestrial ancestors.

References
Paterson RS et al. 2017. The evolution of pinnipeds from a terrestrial ancestor: the possibility of parallel evolution within a monophyletic framework. SVP abstracts 2017.

SVP 2017 abstracts: Does Malerisaurus nest with Azendohsaurus?

The short answer is
no.

You might recall
we looked at Malerisaurus earlier.

A different take comes from the Nesbitt et al. 2017 SVP abstract:
“Mediation of some of these challenges is now possible with the recently recognized early
archosauromorph clade Allokotosauria. This clade contains disparate, ecologically
diverse (faunivores and herbivores), and typically larger bodied (1-3 meters in length)
archosauromorphs (Azendohsaurus, Trilophosaurus), but to this point, plesiomorphic,
early-diverging allokotosaurians have not been identified.

“Here, we recognize specimens assigned to the enigmatic taxon Malerisaurus from both present-day India and western Texas as members of Allokotosauria, and more specifically, the Azendohsauridae. The recognition of Malerisaurus as both an allokotosaur and an azendohsaurid has also helped identify other fragmentary remains of close relatives from Triassic deposits across Pangea including India, elsewhere in North America, and Africa. As such, Allokotosauria had a near Pangean distribution for much of the Middle to Late Triassic. Allokotosauria represents one of the oldest successful clades of archosauromorphs that achieved a wide geographic distribution and both taxonomic and ecomorphological diversity.”

The Protorosaurs, Malerisaurus, Prolacerta, Protorosaurus, Pamerlaria and Boreopricea.

Figure 1. Click to enlarge. The Protorosaurs, Malerisaurus, Prolacerta, Protorosaurus, Pamerlaria and Boreopricea. It is easy to see why these taxa become confused with Trilophosaurus and Azendohsaurus. More taxa solves the problem. 

In the
large reptile tree (LRT 1050 taxa) Malerisaurus nests with other protorosaurs within the new Archosauromorpha, sharing many traits by convergence with the Allokotosauria. This clade of currently three taxa (Trilophosaurus, Azendohsaurus and horned Shringasaurus) nests within the Rhynchocephalia between the derived taxa, Noteosuchus and Mesosuchus all within the new Lepidosauromorpha.

I’m guessing,
based on past performance (I was not in Calgary), that Nesbitt et al. did not add any or many basal rhynchocephalians to their cladogram, so this new odd nestings appears to join other odd nestings, likely victims of taxon exclusion.

References
Nesbitt SJ et al. (nine co-authors) 2017. The ‘strange reptiles’ of the Triassic. The morphology, ecology, and taxonomic diversity of the clade Allokotosauria illuminated by the discovery of an early diverging member. SVP abstracts 2017.

New online course on birds and pteros from Dr. Phil Currie

This is part of a new video series on theropods and birds
Click to view. YouTube video on birds and pterosaurs from Dr. Phil Currie

Click to view. YouTube video on birds and pterosaurs from Dr. Phil Currie

YouTube caption:
Week 4, Lecture 3 for the online course “Paleontology: Theropod Dinosaurs and the Origin of Birds”, taught by Philip John Currie, Ph.D. All rights belong to Coursera and University of Alberta. For educational purposes only. Happy learning!
The new video
features all your favorites: Microraptor, Archaeopteryx, Deinonychus and some old John Ostrom hypotheses.
I added this comment:
The use of cartoons without skeletons undermines the scientific value of this project.
Microraptor does not nest with dromaeosaurids in the large reptile tree, but with Ornitholestes and tyrannosaurs.
That cladogram does not document a secondary flightless condition in dromaeosaurids.
Pterosaurs are not related to dinosaurs, but are tritosaur lepidosaurs derived from taxa like Huehuecuetzpalli, Macrocnemus and Cosesaurus.
Pterosaur flight membrane does not attach to the lower leg. See

The last teeth seen in hard-shell turtles

The ancestors of turtles had teeth.
You can clearly see them in Elginia (Newton 1893; Fig. 1), Sclerosaurus (Meyer 1859) and even in the basalmost soft-shell turtle, Odontochelys (Li et al. 2008). What we’re missing is a set of teeth in the basalmost hard-shell turtle in the large reptile tree (LRT, 1050 taxa), Meiolania (Owen 1886, Gaffney 1983). So we go looking for them (Fig. 1). 

Figure 1. Palates of Elginia (from Newton 1893) and Meiolania (from Gaffney 1983). Although the drawing at lower right erases them, the photo at upper right seems to show some tiny teeth and tooth sockets.

Figure 1. Palates of Elginia (from Newton 1893) and Meiolania (from Gaffney 1983). Although the drawing at lower right erases them, the photo at upper right seems to show some tiny teeth and tooth sockets.

And there they are.
The jaw rims of Meioliania appear to have tiny, useless teeth. These were overlooked or avoided by Gaffney 1983 who eliminated those tiny bumps and holes in his drawing (Fig. 1). In the era before software generated cladograms, Gaffney did not consider Meiolania the basalmost hard-shell turtle, but in the modern era, the LRT does.

When you evolve from
teeth to no teeth sometimes the tetrapod pattern seems to be

  1. relatively few big teeth, then
  2. relatively many tiny teeth, then
  3. toothlessness

The same pattern seems to play out in ophthalmosaurs and ornithomimosaurs, but not birds and mysticetes.

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.
Meyer H von 1859. 
Sclerosaurus armatus aus dem bunten Sandestein von Rheinfelsen. Palaeontographica 7:35-40.
Newton ET 1893. On some new reptiles from the Elgin Sandstone: Philosophical Transactions of the Royal Society of London, series B 184:473-489.
Sues H-D and Reisz RR 2008. Anatomy and Phylogenetic Relationships of Sclerosaurus armatus (Amniota: Parareptilia) from the Buntsandstein (Triassic) of Europe. Journal of Vertebrate Paleontology 28(4):1031-1042. doi: 10.1671/0272-4634-28.4.1031 online