In article <CqAnEn.5I0 at zoo.toronto.edu> mes at zoo.toronto.edu (Mark Siddall) writes:
>I agreee with Erich here. If it doesn't do parsimony, it must be
>distance-based. If it's distance based, it's phenetic. If it's
>phenetic it's not phylogenetic.
>It's hard to believe this still has to be said after all of these years!
It does have to be said, because people like Andrey Zharkikh and me don't
get it yet. If you wish you can define distance-based methods as "phenetic".
The issue most people are interested in is whether these methods can
legitimately be used to infer a phylogeny. They can. There are circumstances
under which each of the major categories of methods work, and circumstances
under which they don't work, and the matter is not settled by calling some of
them "phenetic".
Phylogenetic systematists, particularly those close to the Willi Hennig
Society and its journal Cladistics, like to make a big point of calling
distance matrix methods "phenetic". The unstated implication is that this
means they are not making any valid estimate of the phylogeny, since they
aren't "phylogenetic". This is entirely too superficial a way of settling
the issue.
I prefer to reserve the terms "phylogenetic" and "phenetic" for description
of methods of classification (a separate issue) and to acknowledge that all
methods are attempting to infer phylogenies, and succeeding to some extent.
The useful point to discuss is what their properties and assumptions are,
not whether or not they are to be called "phenetic".
(By the way, the conclusion that "if it doesn't do parsimony, it must
be distance-based", is wrong for maximum likelihood, and may be wrong for
invariants and compatibility methods as well, unless one considers them to
also be parsimony.)
-----
Joe Felsenstein, Dept. of Genetics, Univ. of Washington, Seattle, WA 98195
Internet: joe at genetics.washington.edu (IP No. 128.95.12.41)
