Striking a balance, weighting and Cladistics

[Pierre Deleporte]

>At 10:19 PM 2/14/01, Gabriel A. Eickhoff wrote:
>>   What I'm refering to by "proper cladistics" is post Hennigian
phylogenetic
>>systematics. Mainly, taken from that and as defined and revised by Arnold
>>Kluge and Steve Farris.
>
>[...]
>
>>  Conceptually, this understanding does not accept weighting by any means.
>
>Farris was once a proponent of a posteriori weighting. Is he still?
>


Farris '83 was very clear about weights: he did state that all characters
are likely not equally reliable as phylogenetic information. The problem
being the evaluation of these differences.
He proposed successive approximation weighting as a solution, i.e. weigting
according to the homoplasy of the different characters as shown on the most
parsimonious cladogram.
An inconvenient of this weighting scheme was that the iterative procedure
started with the most parsimonious cladogram, based on prior weights, and
thus being not obligatorily the optimal starting point.
Pablo Goloboff recently improved the method by evaluating all possible
cladograms according to the differential weighting (for homoplasy) they
imply. This is "parsimony under implied weights", a real improvement of
successive weighting (to accept or not the logics of successive weighting
is another question). The whole story can be read in the journal
"Cladistics", of course.

At least, this demonstrate that "cladists" cannot be so easily coined as
rejecting weighting, and opposed in this respect to "evolutionary
systematists". Isn't cladistics "evolutionary" in the logics of most of its
proponents? And isn't equal weight "some weight" after all? Even if it
merely reflects our ignorance of the possible differential weights.

A priori weighting is in no way theoretically at odds with "cladistics" or
"maximum parsimony approaches". The problem is simply : what strong
assumptions can we state as a basis for differential weighting? What do we
know of the evolutonary process, for some kinds of characters, in some
lineages or in every lineages, to constitute a firm basis for differenial
weighting?
How shall we convince our readers (and first ourself!) that we know enough
of the evolutionary process in order to use this knowledge for weighting?

A priori differential weighting is not surprisingly more frequent in
molecular approaches than in morpho-anatomical ones. That the present
practices are more a reflect of the immaturity of this young field than of
a fine knowledge of the molecular evolutionary processes is still matter of
hot debates... Multiple examples of conter-intuitive differential
informativeness of some kinds of data are flourishing in the recent
literature, inviting to much carefulness about a priori weighting.

(Ad hoc manipulation of the data by arbitrary weighting is of course to be
avoided, but here we quit biology for problems of intellectual integrity).

cheers
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