The attitrion of taxonomic expertise

James Lyons-Weiler weiler at ERS.UNR.EDU
Fri Oct 18 16:09:33 CDT 1996


On Fri, 18 Oct 1996, Tom DiBenedetto wrote:

> On Fri, 18 Oct 1996 12:19:51 -0700 (PDT), James Lyons-Weiler wrote:
>
> >On Fri, 18 Oct 1996, Tom DiBenedetto wrote:
> > No need to apologize; the circularity is most certainly in my head, which
> >I recognize the problem.  The method itself (cladistic parsimony) does not
> >proceed in any such manner; parsimony algorithms first calcluate a
> >pairwise distance (purely a phenetic),
>
> HUH? Parsimony programs evaluate the number of state-tranformations
> necessary to account for the data on each topology. They do not
> evaluate alternative trees on the basis of phenetic distances.
>
>
My confusion; sorry.

They evaluate the number of transformations as steps in a single pass;
from there they go about inferring the reconstruction using a second pass.
>From data to tree to data.  Circular.


 and then go about finding the
> >shortest route possible between terminal taxa.  The problem is the same as
> >finding a Steiner graph.  Character transformations are then reported as
> >ticks on the tree.  The parsimony tree is the tree with the shortes
> >possible internodes.
>
> You are missing the concepts by trying to impose a  terribly
> misleading rewording of
> the details. There is no criterion for "shortest possible
> internodes".
>
I didn't mean to imply that this was thecriterion, but it certainly is the
outcome.

>
> Except that this is not what happens.....the tree is not used to
> infer anything about the data, the tree merely displays the results
> of the test of congruence. Congruence is a necessary test for
> hypotheses of homology, for truly homologous characters are congruent
> by definition. The parsimony algorithm merely constrains our
> statements of generality to the extent necessary in light of the
> requirement of congruence.
>
I have yet to see anything resembling a measure of congruence coming out
of analyses as you have described.  To claim that the tree is not used to
infer anything about the data ignores the second pass reconstrcution step.
I'll not bother to remind you that despite your claims of congruence,
cladistic parsimony does nothing of the sort.  It provides the tree that
minimizes homoplasy; what's left over is inferred (from the tree!) to be
synapomorphy or plesiomorphy.



> It is not an absolute test, which is what your question seems to beg;
> of course at some level NO test will assure you that your tree or the
> "facts" which it displays are TRUE.

The tree does display things then?  The discussion of truth is futile
unless the phylogeny is known.  When it is known, cladistic parsimony is
consistent under a restricted set of conditions.  It's not statistical
(nod to Atilano Contreras-Ramos).  If a statistical solution were
available in general, the question of truth would never need arise,
because the goal is to falsify appropriate null hypotheses.  To be fair to
recent contributions, some tests do define null hypotheses, at least in
the form of extreme skewness (g1), the permuted tree length distribution
(PTP), and RASA, which I'm all for :)

James Lyons-Weiler




More information about the Taxacom mailing list