Dear David,

thank you for your question. Pappochelys shares numerous apomorphies with other stem-turtles, especially Odontochelys. Most of these features have not been published, so it may be wise to wait until the full description is published before considering it in a phylogenetic analysis. Also, I think first-hand study of the material is essential for such phylogenetic work; this said, you are most welcome to study the material in our museum collection once it has been published.

The reason you may find Pappohelys to nest with placodonts may be based on the misunderstandig of the gatralia, which appear superficially similar in placodonts and Pappochelys but at closer sight are entirely different. There are no characters known to us that would support a placement of Pappochelys (and the very similar Odontochelys) with placodonts. If you find turtles to nest with pareiasaurs, it would be interesting to learn about your opinion about the substantial molecular and soft-anatomical evidence that supports a diapsid (archosauromorph) origin of the clade.

Dear David,

thank you for your question. Pappochelys shares numerous apomorphies with other stem-turtles, especially Odontochelys. Most of these features have not been published, so it may be wise to wait until the full description is published before considering it in a phylogenetic analysis. Also, I think first-hand study of the material is essential for such phylogenetic work; this said, you are most welcome to study the material in our museum collection once it has been published.

The reason you may find Pappohelys to nest with placodonts may be based on the misunderstandig of the gatralia, which appear superficially similar in placodonts and Pappochelys but at closer sight are entirely different. There are no characters known to us that would support a placement of Pappochelys (and the very similar Odontochelys) with placodonts. If you find turtles to nest with pareiasaurs, it would be interesting to learn about your opinion about the substantial molecular and soft-anatomical evidence that supports a diapsid (archosauromorph) origin of the clade.

Thank you for your reply, Rainer. And Happy New Year.

Not sure by your answer that you saw the PDF of the cladogram I sent along with the earlier note. For those following this question, in the cladogram Pappochelys nested with Palatodonta and that clade with Diandongosaurus, all basal sauropterygia. Placodonts are much more derived than these taxa, but all within the same clade.

Your 2015 TNT analysis nests turtles and Pappochelys with Eunotosaurus and then basal Sauropterygia. So there is some fragile agreement when it comes to Pappochelys.

In my cladogram Eunotosaurus nests with Acleistorhinus, Microleter, Delorhynchus and Eocasea, in order of increasing distance. I trust you have included these taxa in your analysis. If you have not, you should do so.

In my cladogram, not only do turtles nest with pareiasaurs, but turtles had a diphyletic origin with Sclerosaurus giving rising to Odontochelys and Elginia giving rise to Meiolania at the twin transitions - at present. As you know, pareiasaurs do not have gastralia and thus the plastron of turtles is a novel trait.

I ran turtle analyses deleting pareiasaurs and including archosaurs, archosauromorphs, diapsids (which are now diphyletic), etc. Details here:

https://pterosaurheresies.wordpress.com/2016/12/08/seriously-trying-to-nest-turtles-with-archosaurs/

Summary: when pareiasaurs and a broad selection of close and distant sisters are deleted, turtles nest with therapsids rather than Pappochelys. And that only happens when all of the better candidates are removed. Your 2015 cladogram nests such disparate morphologies as Choristodera, Kuehneosauridae, Eosauropterygia and Eunotosaurus without intervening taxa. That’s a problem. You need many more taxa or you can rely on the cladogram I sent earlier to focus your inclusion sets.

Firsthand examination of specimens is not required. Four years ago I nested Diandongosuchus with phytosaurs and Lagerpeton with Tropidosuchus, without seeing the taxa firsthand. Both relationships have been re-discovered recently by workers with firsthand access. 

I found that molecular relationships are supported by morphology when relationships are close, but do not do so with increasing phylogenetic distance. Unfortunate, but true. This also happens in mammal interrelationships. Bottom line: You’ll never find a gradual accumulation of traits among archosauromorph diapsids and turtles as I found with pareiasaurs and other more distant taxa in my cladogram. And a gradual accumulation of traits echoing evolutionary events is paramount.

This image of dorsal skulls may prove helpful:

http://www.reptileevolution.com/pareiasaur-skulls.htm

Note the gradual accumulation of traits. For others reading this, a pdf of the large genus-based cladogram is found online here:

http://www.reptileevolution.com/reptile-tree.htm

"Your 2015 cladogram nests such disparate morphologies as Choristodera, Kuehneosauridae, Eosauropterygia and Eunotosaurus without intervening taxa. That’s a problem."

It's one of two very different problems. It indicates either that the tree is wrong, as you suppose, or that the fossil record is incomplete.

Given how very poor the Permian fossil record of diapsids is, I'm strongly inclined to expect the latter.

"You need many more taxa or you can rely on the cladogram I sent earlier to focus your inclusion sets."

You, on the other hand, need many more characters, or you can rely on... :-)

"Firsthand examination of specimens is not required."

Ah, that depends on the quality of the publications. And that is very heterogeneous. Of course firsthand examination ought not to be required; of course we all ought to strive to describe and illustrate our specimens so well that firsthand examination will be unnecessary. Unfortunately, however, many publications don't reach that standard for many purposes. Nature papers, in particular, aren't really papers, they're extended abstracts with abysmally low photo resolution; they're announcements, not fully useful descriptions. Restrictions on length, photo quality and even number of citations are the price people pay for the impact factor they need in order to stay employed.

And, mind you, this problem isn't confined to extinct taxa. The skeleton of most extant tetrapods is wholly or mostly unknown to science. In my publication list you'll find a 2015 paper on an Oligocene newt; in the course of making it, I was unable to find a description or illustration of the vertebrae of good old Dicamptodon, even though I searched literature all the way down to the middle 19th century. Skulls and vertebrae of many salamandrids have been described, but not in a lot of detail or with any comments on ontogenetic or individual variation, and illustrated, but only as line drawings without any shading or any surface texture. I could go on for a long time.

It's not the case either that long descriptions are necessarily useful. If you search for "The order Microsauria" on Google Books, you'll find almost all of a book with that name, and in it you'll find a pretty long and detailed description without photos, but with specimen drawings and reconstruction drawings. I found a few things weird about it, went to see the specimen, and discovered (as now mentioned in my 2016 preprint) that some of the features in the drawings and the text are simply wrong. For instance, the supposed suture between the right parietal and the supposed postparietal is a series of discontinuous cracks! If you removed that suture from the drawings, you'd get a very asymmetric suture pattern on the skull roof, but if you tilt the specimen just a few more degrees under the microscope, you find that the suture between the parietals is just sinuous – it sweeps back across the midline on the inclined top of the occiput, restoring symmetry.

In short, sometimes firsthand examination is not required, sometimes it is, and often the only way to find out is firsthand examination. :-| Even photos often give completely misleading impressions about 3D shapes.

"And a gradual accumulation of traits echoing evolutionary events is paramount."

Only if you assume that the fossil record is very dense. This assumption is obviously untenable.

Sorry, I forgot: Diandongosuchus was indeed found as a phytosaur in a poster presented at last year's SVP meeting. But where have the dinosauromorph Lagerpeton and the proterochampsid non-archosaurian archosauriform Tropidosuchus been found anywhere near each other?

Dear David P.,

I have to say that it gives me a really bad feeling whenever I cannot study specimens of a particular taxon myself. It is not so much distrusting other people's work, but simply the fact that each and every character is a hypothesis, that is to be confirmed by first-hand observation taxon by taxon. And this should, by any means, be the observation of the author who conducts the study. This is the reason why phylogenetic analysis needs time and, sometimes at least, quite a bit of money for travelling.

Hence, dear David P., as much as I welcome your divergent point of view, which is so important in science, as much I am astonished by your bold statement, which seems rather unscientific to me. You still need to justify that by any argument (I know none, to be honest). I can only repeat my invitation to you to visit our collection, and I am looking forward to any discussion in front of the real specimens.

Thank you for your kind reply, Rainer. As you said, first-hand study is indeed important. That's why I suggest, before publishing, that you at least consider, and perhaps study first-hand the taxa recovered by my second-hand study, which considers over 900 options (minimizing a priori taxon exclusion) for turtle ancestry and finds Pappochelys no where near that node. No one's work is the final word, especially in phylogenetic analysis of fossil taxa, where new discoveries of transitional taxa can split prior sisters. I only encourage you to include taxa you may have overlooked.

Dear David,

Thank you for your interest in my project. Phylogenetic analysis is indeed a fascinating and powerful tool. Discoveries can separate former sisters with new transitional taxa. On the other hand, taxa and clades can be removed to test the remaining tree topology to see if it falls apart, or stays intact. Currently the 228 multi-state traits are capable of lumping and separating the 915 taxa in the large reptile tree. And that's with the handicap of not visiting all of the specimens first hand, relying on sometimes crappy drawings, grainy photos, and my naive introduction to each new taxa. Testing has been robust. Data can be distrusted. Taxa and clades can be removed to see if the topology remains unchanged. Phylogenetic analysis is also 'self-healing'. Autapomorphies that appear can be reexamined. Corrections can either separate sisters or cement them closer together. Since phylogenetic analysis relies on so many data points, dozens to hundreds can be missing and it does not seem to matter. Similarly, mistakes during scoring are often swamped by correct scores and the proper nesting is still recovered. I realize that your professors and textbooks have told you one thing. My testing reveals another, and, like Galileo before me, I have to report what I find, despite the break from orthodoxy. I show my work and encourage others to come up with corrections and more parsimonious sister taxa. What more can you ask? 

I realize the importance of quality observations, but in order to 'see the big picture' in this case quantity observations need to come first. 

With regard to Lagerpeton/Tropidosuchus, here is the reference: 

Novas FE and Agnolin FL 2016 Lagerpeton chanarensis Romer (Archosauriformes): A derived proterochampsian from the middle Triassic of NW Argentina. Simposio. From Eventos del Mesozoico temprano en la evolución de los dinosaurios”. Programa VCLAPV. Conferencia plenaria: Hidrodinámica y modo de vida de los primeros vertebrados. Héctor Botella (Universitat de València, España) 2016

and 

https://pterosaurheresies.wordpress.com/2016/11/13/you-heard-it-here-first-lagerpeton-is-not-a-dinosauromorph/