Re: New paper on fish fingers

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

> Here's the way I see it... One hypothesis for the evolution of the turtle 
> shell posits that the carapace evolved via expansion of the ribs and 
> vertebrae.  The other hypothesis proposes that the carapace evolved from 
> fusion of osteoderms, which fused with the underlying skeleton.

Yes.

> _Odontochelys_ has a plastron, but the dorsal armor is composed only of 
> neural plates (no osteoderms), and so has been cited in support of the 
> first hypothesis.

But wait. It can easily be argued that the costal plates are there and fused 
to the ribs, rather than the ribs being broadened (as they are in 
*Eunotosaurus* and *Pumiliopareia*); that would mean it has the same 
carapace morphology as *Chinlechelys*, except that it lacks the peripherals, 
the nuchal and the pygal (which are undoubted osteoderms) as well as the 
neck & tail spikes.

Li et al. take this possibility into account and then dismiss it based on 
the ontogeny of modern turtles, where (in the known cases) the costals start 
to ossify (intramembraneously) from the ribs, rather than having a separate 
ossification center. But just a few lines above, they explain that there's 
variation in how the neurals are formed in the ontogeny of extant turtles: 
in some, the neurals start to ossify from the tips of the neural spines 
rather than having a separate ossification center, while in others they have 
separate ossification centers and later fuse to the vertebrae; the latter, 
they say, must have been the way *Odontochelys* did it, because the neurals 
are disarticulated from the vertebrae in some places. Looks like the 
vertebrae reach the ossification centers of the neurals before ossification 
of the neurals starts in some turtles but not in others. My question is why 
the same can't apply to the costals: the ribs may reach the ossification 
centers of the costals before ossification of the costals starts, but that 
doesn't mean the costals are homologous to a part of the ribs.

In other words, I interpret *Odontochelys* to have the same condition as 
*Chinlechelys*, except without sutures between the costals, without 
peripherals/nuchal/pygal, without neural arches that sit between two centra 
instead on top of just one, and apparently with a bit more fusion between 
the costals and the ribs.

And besides, ontogeny can evolve. Vertebrae are endochondral 
ossifications -- except in crown-group salamanders, where they ossify 
intramembraneously and are not preformed in cartilage. Surely salamander 
vertebrae are homologous to everyone else's vertebrae?