[Taxacom] Molecular shared derived characters (was, sine's, line's)

[John Grehan]

-----Original Message-----
From: Sergio Vargas [mailto:sevragorgia at gmail.com] 
Sent: Thursday, May 12, 2011 9:45 AM
To: taxacom-request at mailman.nhm.ku.edu; John Grehan
Subject: Re: Molecular shared derived characters (was, sine's, line's)

>I guess I just won't get it...

NO worries. I don't get a lot of things myself.

> I just have two brief comments.

> no clock assumption is necessary for the analysis of molecular 
>characters either. 

Perhaps so, but in practice most if not all parsimony or other
phylogenetic clustering analyeses also assume a molecular clock.

> If you use parsimony, for instance, you only hope/assume 
> characters don't evolve to fast or to slowly so that you can 
> trace the evolution of the character. 

Do molecular phylogenies rest on a hope and a prayer?

> This is also assumed when you analyze morphological characters.

Evolutionary rates have nothing to do with the identification of
uniquely shared characters in the LCA.

> The computer program actually treats the characters identically 
> no matter whether they are nucleotides or morphological 
> characters, and you can root the tree if you want or establish 
> transformation series if you want. 

The algorithm my treat them the same, but that does not mean that they
are.

> Your critique to molecular characters seem to be, precisely,
> that you cannot do any of the above (e.g. establishing 
> transformation series) with nucleotides because it is something 
> hard or impossible to do, and because you can do this with 
> morphological characters you prefer these characters. 

It's less about a 'transformation series' than it is being able to
recognize a common homology for a feature that is not identical in the
out and ingroups.

> The problem I see with this (but probably I am just not getting it) is

> that one could use (and has been used done over and over) a similar 
> argumentation line against morphological characters: morphological 
> characters are generally homoplasic or more prone to homoplasy,
labile, 
> and useless per se for phylogenetic inference. 

Yes, that's an argument, but that is all it is - an argument. Whether it
is really true or meaningful for phylogeny construction will depend on
actual instances. And of course, if true of morphology, is just as true
for molecules. But one has to be able to determine, first what stands
out as being shared derived (But which shared derived are homoplastic
will depend on how they stack up in the analysis) and it is this which
appears to be highly problematic for molecular studies. If one is not a
cladist this problem does not exist since overall similarity will be the
measure of relationship.

> IMHO both arguments share the same problem, you still need to provide
a 
> scientific explanation for the observed incongruence between
genes/genomes 
> and morphological characters. I think simply disregarding characters
does 
> not qualify as such an explanation.

Well in our orangutan paper we do not "simply disregard characters", we
give more than one explanation for the observed incongruence (actually
between sequences and morphogenetic characters) to show why we think the
morphogenetic evidence more closely conforms to cladistic requirements
and may be more accurate as a representation of human origins that is
also concordant with fossil evidence which is based totally on
morphogenetic characters that everyone (even the molecular believers)
believe are true and reliable.

John Grehan

sergio