[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]

Re: molecular studies question



----- Original Message -----
From: "David Peters" <davidrpeters@earthlink.net>
Sent: Sunday, January 29, 2006 3:11 AM

I haven't found ANY molecular studies in the Reptilia that match
morphological studies. In fact molecular studies seem to bring up some
pretty odd pairings among reptiles.

Which is valid? Or am I asking for a subjective vote?

Not quite, but molecular and morphological studies suffer to similar amounts from different sets of problems:


Molecular studies can't include fossils, and even among living taxa the number of already sequenced species is very low for most genes. Morphological studies don't include as many fossils as they could.

Molecular data is prone to long-branch attraction (...and long-branch repulsion, and short-branch attraction...) under some circumstances, such as a small number of taxa (see above). Morphological long-branch attraction seems to be more rare, but it occurs just as easily when a clade full of autapomorphies first enters the fossil record when it already has most of those autapomorphies and consequently has little in common with any of its postulated sister-groups (examples are turtles and alvarezsaurs).

Morphological data matrices are prone to containing correlated characters (so that in effect they contain fewer characters than stated, of which some are weighted without reason). Molecular data matrices contain one or at most a few genes, which may well be the equivalent of using, say, skull characters only...

Molecular data can be analyzed using parsimony, maximum likelihood and Bayesian methods. Bayesian analysis for morphological data was only introduced last year, maximum likelihood for morphological data lies in the (near?) future; but even then the only scientific weighting method for morphology is a-posteriori-weighting aka reweighting.

I conclude that both kinds of analysis are steadily improving and will continue to improve throughout the foreseeable future, because -- among other things -- the data matrices of both will continue to grow!