Re: ...and how ranks don't work was Re: Sauropodz r kewl

[evelyn sobielski <koreke77@yahoo.de>]

> > They're not measuring anything except taxonomists' mood
> swings. What you

That is true. There are workarounds (good taxon choice), but they fail at 
higher taxonomic levels. You can, for example analyze diversity within 
Passeroidea - or even Passeroidea, Muscicapoidea and Sylvioidea - by simply 
taking families and comparing them, because they originated and radiated within 
a very short time of each other, and lack significant internal structure 
between the level of "superfamily" and "family". But if you compare passeriform 
families to, say, coraciiform families, it fails.

In brief, ranks often hide crucial phylogenetic information, and anyone using 
them in ignorance of this fact deserves what they'll get.

> > need to measure biodiversity, unless you have a list of
> species under one
> > single species concept, is the sum of branch lengths

Won't work (usually). For example, in crown Dinosauria, any or all of 
psittaciforms, passeriforms, columbiforms, coliiforms and/or hoopoes* have 
extremely accelerated mutation rates, and/or unusual shifts in base 
composition. Trying to align hoopoes or mousebirds with anything is usually 
very painful, and even in mt loci it is often not really easy.

The accelerated mutation rates are trivial (it boils down to either increased 
mutagenesis and/or increased viability of mutants), but the base composition 
shifts I have hitherto failed to find any reason for.

> > organisms you look at. You need a phylogenetic tree.

That is also true. Best a robust timetree, then you can start from a point in 
the past and take the clades which separated there and compare their diversity.

You can also use a plain old tree, but this has to utilize a character set in 
which there are no drastic differences in branch length. The time from LCA to 
crown taxa in a taxon set is, after all, identical for all crown taxa. So or 
the branch-length approach to work, the branches must all be approximately the 
same length from root to crown.

I have found most loci where branch lengths are conservative to be those that 
evolve very slowly. That defeats the purpose of the branch-length approach, 
because the differences between leaves in a recent radiation are infinitesimal. 
In the end, differences in branch length between major lineages might account 
for most of the difference, compared to within-lineage diversity.



Regards,

Eike

* But not usually hornbills, which is perhaps the reason why one easily gets 
nonmonophyly for "Upupiformes"+"Bucerotiformes".