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

AW: Dromomeron and Lagerpeton query

> Since Dromomeron and Lagerpeton are known principally from
> hind limb materials, its not surprising that the recent
> matrices of Nesbitt et al. 2009 (JVP) based on Irmis et al.
> 2007 (Sciene) had a larger than typical number of characters
> devoted to the hind limb. 
> Is that a bad thing? I'm thinking of Bennett's rejection of
> hind limb characters in a similar matrix in 1996. 

Well, from what I've seen in birds the question cannot be reliably answered 
until tested. There can be sufficient phylogenetic signal in a limb (or other 
piece of skeleton, down to a single bone), but often there is not, and 
over-emphasis of it will skew the analysis.

A test is rather easy: use a very restricted character and taxon set. In this 
case, one might use only hindlimb characters of ecologically diverse taxa 
representing the robustly-supported main lineages of 
archosauromorphs/archosaurians. Then, check in how far the analysis' proposed 
phylogeny agrees with the known consensus phylogeny.

It may be that the Bennet analysis does this, or something sufficiently 
similar. It needs to be a test by falsification; you'll want to try and show 
that there is *no* clear phylogenetic signal - that your character set will, 
for example, clade semi-aquatic ornithischians with semi-terrestrial basal 

If this fails at first, it is better to use denser taxon sampling and see if 
you still seem to get an apparently good signal, rather than to assume that 
there *is* a reliable signal after only a single test. 

If however it succeds at the first try - meaning that your character set is 
overwhelmed by homoplasies and thus a liability - but you don't want to discard 
the whole attempt, you can try and build a finer-grained character set, taking 
into account ontogenetic information (suture lines etc) for example, and test 
it anew.

The underlying idea is, in a nutshell, that character states are the way they 
are essentially due to a combination of two different evolutionary causes: 
a) the pattern of descent (phylogenetically informative - well, it *is* 
phylogeny) and 
b) the pattern of adaptation (phylogenetically risky, if not outright 
Thus, test which cause is the dominant one in a "minimalistic" but 
representative set of taxa and characters.