paraphyly

[pierre deleporte]

At 09:44 05/08/2002 +0200, Zdenek Skala wrote :
>-----Original Message-----
>From: pierre deleporte [mailto:pierre.deleporte at UNIV-RENNES1.FR]
> >I think that answering the following question could help solving our
> >discussion :
> >what exactly will a cladist classifier do with a hierarchy of gaps ? With a
> >"5 point gaps", he names a monophyletic group above the gap (the
> >eclecticist will add: putting a stop to a paraphyletic group below,...
> >I see both the cladist and the eclecticist selecting "significant gaps" for
> >naming taxa.
> >But inside this selection, the eclecticist decides which gaps (the larger
> >maybe) are worth naming a special kind of groups (paraphyletic groups),
> >when the cladist does not.
> >Isn't this a difference ?
>
>I don't think so. In fact, better description of what eclectic classifier
>does (instead of putting "stops" and "starts" of a taxon) is: (1) split
>the cladogram at the "significance level 5" and then name all the pieces
>(both paraphyletic and holophyletic); (2) split these pieces at the "level
>3" and again name all the pieces. Simple, isn't it?

Yes indeed, quite simple. And will help specifying the differences.
With a "5" gap, the cladist names the monophyletic taxon above (the
complete one).
With a "3" gap, the cladist names the monophyletic taxon above (the
complete one).
Conclusion : as for naming, "5" or "3" gap make strictly no difference to
the cladist.
To the point that you could start with the "3" gap, and finish with the "5"
gap, and obtain exactly the same result.
But would the eclecticist obtain the same result by beginning with the "3"
gap ? Certainly not (just try).
So, THIS is a difference.

>Your original point, if I remember well, was that the eclectic
>(=classical) definition of monophyly generates less parsimonious
>taxa-delimitation process

I think I never talked of "parsimony" concerning taxa-delineation criteria,
seems irrelevant.

>  since there is some additional criterion required that is not needed in
> the cladistic procedure.

This was the point. The debate was about arbitrariness involved in
delineation criteria. You agreed with Tom Di Benedetto that exhaustive
naming of all clades entailed no arbitrariness in a cladistic
classification. I further argued that even if not all groups were named,
there was additional arbitrariness (not "less parsimony") involved in the
eclecticists criteria, because the eclecticist needed a criterion for
delineating paraphyletic groups, not only complete clades.
The discussion above shows that this question (thanks to you) may be turned
into the "hierarchy of gaps" problem. Such a hierarchy of gaps is of no
concern for the cladist, but is a supplementary criterion necessary for the
eclecticist. What had to be demonstrated, and has been.

>  As a result of our discussion, we can see that the criteria are the same:

No

>  cladogram topology

Yes

>  and the significance of gaps.

No.
Cladist: gap significant or not. Eclecticist: needs a criterion for
significantly larger versus smaller gaps among this set of significant
gaps. e.g.: is 5 significantly larger than 3? The cladist doesn't care.
(Using qualitative rather than quantitative criteria for significant gaps
and their possible hierarchy would change nothing to the reasoning.)

>  Indeed, the algorithms for taxa delimitation differ between
> cladist/eclectic procedures; otherwise the output would be the same.

Not only the algorithms, also the criteria. Just begin with the "3" gap,
then the "5", in eclecticism and cladistics... and check which
classification will change, and which will not.

You cannot impose on cladistics a "hierarchy of gaps" criterion which is of
no concern in this approach (beyond the trivial "gap / no gap" of course).

Not even for nomenclature: you will name a "genus" inside a "family" in a
cladistic system because the genus clade is topologically (thus
historically) nested inside the family clade, not on the basis of a
"hierarchy of gaps". That the "family" node is possibly supported by a "3"
gap and the genus node by a "5" gap is not relevant: evolution may slow
down, and then accelerate, inside a nested hierarchy of clades, which
remains the same nested hierarchy of clades for the cladist. The possible
signalling of such varying "evolutionary distances" is external to the
cladistic classification system, not included in it, and thus requires no
specific criterion.

This is why one can, even intuitively, understand that there is a
supplementary criterion in eclecticism (there must be, logically).

This is also precisely one of the reasons why some people prefer an
eclectic classification over a strictly cladistic one: "show grades" of
some kind, some way, thus on the basis of some specific criteria, and
accept some phylogenetic inconsistence...

(Yes, Ken, OK, OK ... !   :-)   :-)

best,
Pierre