[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]
Re: Huge morphological analysis reconfirms deep branch relationships among Squamates
John Scanlon wrote, truncatedly:
The Gauthier et al. paper is a massive milestone, but it's hard to
believe anybody thinks a morphology-only analysis is worth presenting
when there is so much molecular evidence out already.
There are several obvious reasons.
1) Do molecular and morphological evidence eventually converge on a
common solution? Only morphological evidence can answer that!
2) Molecular datasets cannot place fossils, a question that is
interesting in its own right. Not everything is about extant species.
3) Molecular data have their own problems; they're not automatically
more flawless than morphological data. One of these is that fossils can
break up long branches -- but only in datasets that contain
In my PhD (finished in 1996) I got scolecophidians nested within
anilioids on morphology, and thought the following procedure was
"The effect of convergence associated with fossorial habits must
therefore be taken into account in subsequent research cycles,
preferably by some method less ad hoc than arbitrarily deleting or
weighting characters. Reanalysis with removal of taxa (in one case
the three scolecophidian families, in the other the four anilioid
taxa) converts apparent synapomorphies of both groups into
autapomorphies of one or the other, and should allow the systematic
position of each to be determined free of the effects of convergence
As this was a pretty late thought (2nd last paragraph in the thesis)
I didn't check if it had been done before, but the answer is:
apparently not. The same method was subsequently published by
Siddall and Whiting (1999) as a way of dealing with long-branch
attraction in molecular analyses, but I've never actually used it in
a published analysis (e.g. anilioids and scolecophidians got
separated by additional data, so it wasn't necessary) and I don't
know if anyone else has either.
It has been done fairly often, just not -- apparently -- in analyses of
Conrad (2008) didn't do it, and it looks like Gauthier et al (2012)
don't do it either. I haven't re-run their analyses with the
relevant deletions (e.g. including only one of Dibamidae,
Amphisbaenia, Serpentes in Conrad) but I expect they would end up in
different places across the tree when not pulled together by
convergence. Anyone want to try it?
Another interesting thing in Gauthier et al. is the fact they cite
M=FCller et al (2011) on the stem-amphisbaenian _Cryptolacerta_ but
don't include this fossil in their analysis. (Facepalm)
Failure of peer review.