_Odontochelys semitestacea_ might get its shell back

[David Marjanovic <david.marjanovic@gmx.at>]

Fig. 1 contains a nice photo -- you can blow it up to 400 % in the pdf 
without it getting pixelated; this is very atypical for Nature -- and an 
interpretative drawing of that photo. Well, I would say what both the photo 
and the drawing show are not "broadened dorsal ribs" as seen in the infamous 
*Eunotosaurus*, but ribs fused to costal plates. In other words, the 
carapace is there: there's one dermal bone plate above each dorsal rib 
(except the last pair) -- these plates are just not sutured to each other, 
unlike in any other turtle known so far. To test this, we'd need more 
specimens: an ontogenetic series or a specimen that we could dare make a 
histological analysis of.

That could explain why the neural plates (those above the vertebrae) are 
apparently sutured to each other or at least make the back immobile: it's 
difficult to imagine why that would evolve before the costal plates had made 
the ribcage immobile.

It appears that the shoulder girdle is in front of the ribcage (rather than 
inside -- it's inside the carapace, but not inside the ribcage!), which is 
generally the case in turtles according to a talk at this year's SVP meeting 
and in fact generally the case in amniotes. The left cleithrum ("dorsal 
process of the epiplastron" -- but see below) is preserved reaching above 
the vertebral column, though of course the fossil is crushed, so this may 
not mean anything.

Fig. 2a appears to contradict the claim in the text (and the supplementary 
information) that, like in lepidosaurs, astragalus and calcaneum are fused 
and the perforating foramen between them is absent, but the resolution is 
_just_ too bad to be sure.

The supp. inf. contains a phylogenetic analysis: *Odontochelys* was added to 
the matrix of Joyce (2007) and the result analyzed. That wouldn't have been 
necessary. The matrix contains an all-zero ancestor, and *O.* is 
indistinguishable from it -- all scores are either 0 or ?.

Then there's another phylogenetic analysis (*O.* added to the matrix of 
Rieppel & Reisz (1999) and the result analyzed). It contains an assumption 
on the ontogenesis of the "broadened ribs" that has, see above, not been 
tested and may not matter at all in the first place -- the paper itself says 
that the neural plates are separate bones in some extant turtles (and in 
*O.*) but ossify as outgrowths of the vertebrae in others, in other words, 
ontogeny can evolve.

The supp. inf. also says that the evidence for the presence of a cleithrum 
in *Kayentachelys* (and therefore in other turtles with a "dorsal process of 
the epiplastron") is equivocal, citing the following book chapter, which I 
have yet to get.

Olivier Rieppel (2008): The relationships of turtles within amniotes, 345 --  
353 in J. Wyneken, M. H. Godfrey & V. Bels, eds.: Biology of Turtles, CRC 
Press.

The cleithrum is a bone in the shoulder girdle that is absent in all known 
crown-group diapsids.

Incidentally, *O.* is Carnian in age.