[Fwd: Re: Probabilities on Phylogenetic Trees]

Tom DiBenedetto tdib at UMICH.EDU
Fri Sep 19 11:50:12 CDT 1997


 James Francis Lyons-Weiler wrote:

>Look, on my criticisms of parsimony, Tom, you hearken constantly back to
>questions about whether ANY method will be able to extract the signal that
>is, in fact, present, and then you attempt to place me in the category of
>people who think they can accurately model evolutionary processes instead.
>My criticisms stand independent of whether an alternative methods for
>stydying hypotheses of homology and getting trees exists...

ok, I will admit that it took me a while to hear exactly what you
were saying.

>Parsimony can get a leg up in the long run if such problematic data sets
>are avoided (a la signal testing), but then so can other methods.  The
>question of whether the processes of evolution have, for matrix A, left
>behind patterns that we expect to lead to decent evolutionary trees is to
>some degree independent of which method of tree estimation one adopts.

ok,,,understand that cladistics arose from a field, "traditional"
morphological systematics in which the existence of an "objective"
discernable pattern is taken somewhat for granted. The patterns were
discerned at a time when their only explanation was that we were
discovering the logical pathways in the mind of god. The question was
never really whether there was such thing as a real pattern, simply
what a valid explanation of that pattern might be. In a long and
painful process, it became generally accepted that evolution through
descent with modification was the underlying causal factor which
engendered the patterns we see. Cladistics, arising from that
tradition, seeks to extend this explanation, as a generalization, to
the extent that it remains internally consistent, in much the way
that a physicist might test their genrealizations about matter by
exploring new situations (for them, that usually means
experimentally).
        The notion of spurious patterns from "randomized" character
states is not a new concept in systematics. We have always confronted
characters which, taken by themselves, could generate bizarre
pattterns, but these were always treated (successfully) within a
parsimony context (even if that context was not precisely
formalized). If I note an equal number of scales on the side of a
species of carp and a species of perch, I will simply decline the
opportunity to code that as a potential synapomorphy. This is not
done from a position of subjective authority; I could, upon request,
outline all the characters which, when taken together, will indicate
that carps and perch are quite distantly related. And I know that if
I wander over to the perch from the position of the carp, I will
cross paths with hundreds of species with the same scale count, and
thousands with different ones. By examining the variation itself
within the context of other characters, I can see that scale counts
are potentially useful only at a low hierarchical level (as we do
with third positions).
        I realize that with the availability of sequence characters,
we are entering into a new dimension. By their nature, there is
precious little we can do in the pre-parsimony phase to test an
assertion of homology. We just read the gel or the chromatograph and
align the sequences (ok, that is problematical, but it is not the
same problem). We can still assess variation levels within the
context of other characters, but I think the potetial for being
misled is much greater than it is with morphological characters.
        And this then is my view of why these issues tend to be
overly contentious.
Most of the people who try to generate tests and other tools by which
to solve these problems, pitch their proposals in such a way as to
assert that they are solving some inherent problem with parsimony
approaches. And that is probably why it is easy for me to assume that
your proposals are wrapped up with an advocacy of non-parsimony
methods, If the proposals were pitched more accurately, as solutions
to some of the inherent problems of sequence characters, then we
probably wouldnt be so defensive. But there has been a historical
association between statisticians, advocates of model-based methods,
and sequencers such that it seems to be anathema to admit that the
problem is in the sequences-as-characters, not in the basic
principles of parsimony. I dont think that the problems of sequences
are fatal, and I am getting up out of my chair right now to go down
into the lab,,,,

>  I
>consider the bush topology in every case to represent something of a null
>model,

But before I go,,,,,I cant agree with that really,,,to me the bush
topology is a starting point (something we have before we even begin
to build the matrix)..it is simply a representation of a state of
knowing nothing. To see it as a null, would be to take the position
that the existence of any true homologies within the matrix is itself
the hypothesis to be tested. I operate from a perspective which
assumes that within my set of characters, some are true homologies
(in fact, I think they all are true homologies and this damn
parsimony criterion keeps telling me I am wrong). I recognize the
implications of making that assumption, but, (perhaps because of my
morphologist perspective), I am not troubled by them.
Even if you see it as a null for the entire systematic effort, it is
not a null for the test of congruence in isolation, because once we
get to that point, we have hypotheses which have already been
corroborated, and are assumed to be homologous. We are simply
submitting them to a test which will weed out those which fail to
exhibit congruence with the overall pattern.

>For all the volleys, and semantic difficulties, and differential focus on
>trees vs. hypotheses of homology, there are remarkably few differences
>among people (minus the pattern cladists, which you obviously do not
>identify with) with respect to purpose and interest.

I'll let you go,,but I'll just throw this in to muddy the waters
further. I think the pattern cladists tend to be greatly
misunderstood, and have been responsible for a lot of refreshingly
clear thinking on these issues. I will muddle around trying to piece
together my own unique understanding, but I feel I have learned a lot
from thinking about the issues they raise.

>(I'm not bored... just busy.)

yeah,,thanks for the workout,,,I'm ready for the new semester!




More information about the Taxacom mailing list