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

John Grehan jgrehan at sciencebuff.org
Wed May 11 10:50:52 CDT 2011

-----Original Message-----
From: Sergio Vargas [mailto:sevragorgia at gmail.com] 

> I think I see the problem now. This was clarified before already. The 
> character is the position in the sequence of DNA/Protein data, the 
> character states are the specific nucleotide/amino acid recorded for 
> that position. 

Here is the additional problem in that arranging postion is not
something that can be done for each postion individually. It is just a
phenetic excercies of overall compromise of a 'best fit' based on
nothing empirical at all.

>We can trace in time the nucleotide changes for a given 
> position, i.e. we can postulate the most likely/parsimonious series of

> changes leading to the observed nucleotide/amino acid at a given 
> position in a sequence. 

One might postulate that

> If you want to define "primitive" conditions you 
> can do it based on outgroup comparison the same way you do it with 
> morphological characters, unless I just don't know how you do it with 
> morphological characters... or better said, how do you do this with 
> morphological characters? 

But one can identify the same 'character' in the outgroup and ingroup,
and identify different states in each. In the case of the base there is
no trace of what went before and no necessary relationship between the
base in the outgroup at the imagined homologous site (created not
observed in nature), and that of the ingroup.

> because I, sincerely, cannot see why DNA data 
> is so different from morphological data, 

As above

> and I cannot see why you cannot 
> take an alignment of your ingroup and your outgroup, 

Problem as above

> find your uniquely 
> shared derived characters, run a parsimony analysis with PAUP and find

> whether your orangutan-human clade is still present in the tree or
> for instance. I would really like to see something like this published

> with a clear explanation on how to do proper cladistics with molecular

> characters and a methodological account on why all previous molecular 
> analysis were wrong pointing to a chimp-human clade. Perhaps you could

> point some papers on the matter?

But can one do 'proper cladistics' with molecular (sequence as I do not
exclude other possibilities) data? That's the unanswered question.

 >Also problematic is the invocation of the molecular clock that seems 
to presuppose a clock line change in all bases overall, and the
 >model requiring bases in the outgroup to have retained their primitive


> well... retaining the primitive condition would be possible only if
> rate of evolution (morphological or molecular) in the outgroup branch
> low, independently of whether a clock holds or not. 

But there is nothing empirical to show that this is indeed the case for
the bases in the outgroup. 

> In most cases we 
> don't know anything about the rate of morphological evolution so we
> only hope or assume the rate was low enough so that the outgroup
> the "primitive" condition. 

But no clock like change of all characters is invoked in morphology

> I see most of the discrepancies may come from 
> the way in which you argument characters which seems to be
> as Pierre, said (i.e. excluding characters that doesn't fit the 
> "uniquely derived" requirement). 

But the uniquely derived requirement is cladistics - in my opinion. If
its not its phenetics which makes no such distinction in selecting
characters for analysis.

> I'm not 100% sure though, I understand 
> how you would build a matrix morphological or molecular characters for
> "cladistic" analysis, so I'm hesitating to conclude that you, indeed, 
> apply a clique method before any analysis... still don't know...
> it's me!

That's for the future to sort out I quess. And whether or not I sue a
clique method or not really does not matter so much as my contention
that humans and orangutans share more unique morphogenetic features than
either shares with any other primate.

John Grehan


More information about the Taxacom mailing list