James Lyons-Weiler weiler at ERS.UNR.EDU
Wed Oct 2 09:22:17 CDT 1996

On Wed, 2 Oct 1996, Tom DiBenedetto wrote:

> On Wed, 2 Oct 1996 07:44:02 -0700 (PDT), James Lyons-Weiler wrote:
> > The "Hennigian algoritm" is model of evolution that excludes
> >the following possible models:
> >
> [snip 6 examples]
> >When one approaches their data with cladistics (or max pars, and other
> >methods), one usally is not suspecting that any of these six alternative
> >models of evolutionary process have been at work, messing up cladistic
> >hierarchy in the distribution of character states among surviving and
> >sampled data.
> I disagree. The six examples you give do not address the question of
> whether the algortihm represents a model of evolution. Obviously
> there are certain basic assumptions in all systematic studies
> (including such things as the very existence of historical
> information); if you wish to consider these assumptions as comprising
> a minimal model,,,well ok,,but you seriously confuse the issues in
> terms of the debates between methods. Some methods use an explicit
> model of how particular processes operate, and the phylogenies are
> dependant on these models. Cladistic parsimony approaches do not use
> a particular model of evolutionary processes, but rather incorporate
> relevant knowledge at a prior stage, at the level of formulating
> hypotheses of homology. The algorithm is left to function as an
> objective pattern-extractor from a data-set of characters which have
> already been accepted as homologous.
The "accepted as homologous" part is where the assumptions I listed are
made.  I am not at all confusing the character evolution models of maximum
likelihood with the models of phylogeny that people carry aroun in their
heads.  A standard of G. Nelson is that "conclusions follow from
assumptions".  Unless the assumptions of the phylogenetic model of
diversificcation can be made explicit, they cannot be tested in specific
instances, and erroneous assumptions will lead to erroneous phylogenies.

James Lyons-Weiler

More information about the Taxacom mailing list