cladism's greatest weakness

[Richard Zander]

Tom DiBenedetto  wrote:
>
> Well, since we are discussing systematics (making classifications)
> perhaps we should leave things on this point of
> ?agreement?. I am not sure what you mean by reconstruction, or what is
> wrong with presenting our state-of-the-art reconstruction until such
> time as we learn more.

Aha. If you want to define reconstruction as the same as a classification,
that's okay. Any classification then is a reconstruction. Logically, the
best classification (particularly one based on a model of evolution...e.g.
additive tree using parsimony) should be the best reconstruction given the
data. This fails, however, because cladistic papers do not present their
results as an optimal reconstruction but as "The Reconstruction
(approximately)" with no indication of what (approximately) means. The
precision of a fully or mostly resolved tree gives false hope that you've
accomplished something better than the intuitive guesswork of the past.
    There are two optimality criteria needed for reconstruction, 1. an
optimality criterion that best groups taxa, 2. a reconstructive criterion
that best eliminates the effect of convergence. The synapomorphies of
closely related taxa on the shortest tree cannot be distinguished as
resulting from convergence or joint ancestry. This not only applies to just
sister groups, where it doesn't matter much, but also between one of a
sister group and the subtending branch, where it does matter if there is a
lot of convergence: the branches switch.
    The reconstruction criterion is, at least up to now (maybe someone has
discovered a better way), distance between taxa on an optimality tree in
terms of intermediate taxa. Only through distance can one eliminate the
confounding of detail on a cladogram by "apparent synapomorphy" (that is
convergence interpreted as synapomorphy by an optimality program. All
genuine reconstruction (as identification of shared ancestry) must eliminate
convergence as well as possible, and this "best" reconstruction requires
treating a cladogram's detail as so poorly interpretable as to be not part
of the reconstruction. Thus a "best" reconstruction is much less than the
precise cladogram as published. That's what "approximately" means.

Richard H. Zander, Curator of Botany
Buffalo Museum of Science
1020 Humboldt Pkwy
Buffalo, NY 14211 USA
email: rzander at sciencebuff.org
voice: 716-895-5200 x 351