Molecular taxonomy: on way out?

[Richard Jensen]

John Grehan wrote:

> > Or do you implement some model of morphological character evolution?
> Which
> > one and why? And if so, why reject molecular model-based analyses?
>
> Because they do not demonstrate derived states for each of their
> characters before the analysis. They do not demonstrate that a
> particular sequence in the outgroup is unchanged and therefore primitive
> with respect to the ingroup.

Dear John,

Once the sequence data are generated, it is simply not reasonable to ask
that the researcher examine every sequence visually to verify putative
homologies and apomorphies.  First, visual inspection is in all probability
more likely to result in errors and the amount of data is just too great.  A
good computer algorithm is simply much more efficient and reliable.

Of course, given your view ("The point is that my practice is not performed
by algorithms as the
algorithmis cannot study the individual character qualities to evaluate them
for inclusion. It can only
mimic the process after the characters have been presented"), any use of any
computer program for any task that a human can do by hand is is mimicry
because the algorithms cannot study the data.  I suppose that, if the data
set generates multiple solutions, you also do not rely on the computer
algorithm to find all of them - it's better done by hand to make sure that
nothing is overlooked.  In fact, I'm surprised you use e-mail; after all,
the mindless computer algorithms that make all of this possible have no
knowledge of the real intent of your words.  All the algorithms do is
concatenate a series of keystrokes, with no understanding of what any of it
means.

For the most part, people I know who employ cladistic algorithms to infer
phylogenies based on molecular data are probably just as careful about
generating their data as are those of us who examine morphological data.
Your comments certainly infer that you don't trust them to be responsible,
thoughtful scientists who gather data  and analyze it with care.  You
apparently learned cladistics in an "old school" setting (as many of us did)
in which we had to choose appropriate outgroups for our study and carefully
examine character state distributions in the outgroup to determine
putatively plesiomorphic states.  What you seem unwilling to accept is that,
regardless of the nature of the data, a computer algorithm can perform that
task for you.  All that is required is that you, as the person constructing
the data matrix, are confident that your are working with homologous
characters and that you have properly interpreted the character states and
made no errors in scoring the states for each OTU.  Once you have created
your data matrix, you are certainly free to go through and examine the
patterns for each character individually to determine what the
plesiomorphies are.  For small data sets, this is easily done.  But for
large data sets with hundreds of OTUs and thousands of characters, it is
simply a tremendous waste of time when the computer algorithm can do it for
you.

One thing you seem to overlook is this: if you claim I am wrong, it it not
up to me to prove I'm right; it's up to you to prove I'm wrong.

Cheers,

Dick J.


--
Richard J. Jensen              | tel: 574-284-4674
Department of Biology      | fax: 574-284-4716
Saint Mary's College         | e-mail: rjensen at saintmarys.edu
Notre Dame, IN 46556    | http://www.saintmarys.edu/~rjensen