Striking a balance, weighting and Cladistics

[Pierre Deleporte]

on 16/02/01 +0100, Zdenek Skala wrote:

>Apart from the weighting discussion: the parsimony criterion (in the
>cladistic sense) of course *is* an assumption about the underlying reality
>as well as methodological principle. Otherwise, phylogenetic analysis could
>not even converge to the supposed real phylogenetic tree topology. If the
>phylogeny would not be parsimonious, the pattern we are revealing has
>nothing to do with the phylogeny.

Methodological principle and (or) biological assumption?
This is an old debate, but maybe we can clarify it this way:

- the general methodological principle of parsimony is one of "optimizing
on a criterion" ("minimizing ad hoc assumptions" failing to fulfill this
criterion is just a complementary way to state it).

- now its biological aspect is : on which criterion do we optimize?
Standard cladistics optimizes the choice of the best cladogram on the basis
of homology by descent (monogeny), and I would add : homology by descent in
direct lineage (rather than generalised horizontal tranfers), under the
supposition that homologous characters are generally identifiable, i.e.
that misleading reversals or convergences are not the rule. And also the
notion of "no common mechanism" as recalled by Tom, that is the assumption
of possible "mosaïc evolution", i.e. the fate of a character on a given
branch being completely independent from the fate of this same character on
another branch of the tree of life. Recent developments by Mike Steel et
al. nicely illustrate this by the mathematical demonstration that standard
cladistics may be turned into a "maximum likekihood" approach... given that
we assume "no common mechanism" for the evolution of the characters
throughout the cladogram.

These biological assumptions are generally implicit rather than explicit in
the mind of the phylogeneticians. They are nevertheless logically necessary
to support the rule of optimizing character state contiguity (standard
cladistics), thus applying "parsimony" under THIS particular criterion.

These assumptions may be viewed as "minimal", or "careful" assumptions
about the evolutionary process by their proponents, but they are
nevertheless specific biological assumptions and we better face them and
argue.

By the way, a method without explicit assumptions about evolution cannot
pretend to reconstruct the pattern of evolution (phylogeny). Nobody can say
"this is the likely history of evolution" and at the same time : "I have no
idea of what evolution may be" (or I don't care). And "common ancestry" is
insufficient in itself to distinguish between cladistics, ML, molecular
clock approaches, all of them implying common ancestry...
The problem of possible differential weighting depends on what we know of
the evolutionary process, but so does the choice of a method.

Pierre
Pierre Deleporte
CNRS UMR 6552 - Station Biologique de Paimpont
F-35380 Paimpont   FRANCE
Téléphone : 02 99 61 81 66
Télécopie : 02 99 61 81 88