Clades are not classes/agreement

[pierre deleporte]

I think Zdenek just correctly focuses on the key point below

At 09:16 10/09/2002 +0200, you wrote :
>I will try to summarize a bit:
>
> >Pierre:
> >A snake is a tetrapod. Not in the data matrix (it has no legs), but as
> >interpreted once the phylogeny is accepted ...
> >"Lost legs" is a confirmed synapomorphy among tetrapods.
>...
> >If the physical presence of features is a requirement of the class theory,
> >of course all tetrapods do not share the presence of legs and tetrapods is
> >not a class of equivalence.
> >It can be one only if "lost legs" (reversal) can be accounted as a common
> >property of snakes and other tetrapods (having four legs, present or
> >secondarily lost).
>
>Snakes are members of tetrapods as a *clade* and four legs are perfectly a 
>property of tetrapods as a clade. However, it does not mean that all 
>members of this clade share this property (not even in some "cryptic" 
>form); clade is an entity of its own and can have properties that are not 
>shared by all members (say "plesiomorphic status" of the clade, in 
>theory). On the other hand, sharing of some property is necessary for a 
>group to be a class. This is why tetrapods are a clade but not a class.
>NB: All what can be said by cladistic analysis (of reversals) is that the 
>reversal is a new transformation event and in this respect the reversed 
>character state is independent of the plesiomorphic character state (no 
>matter that superficially identical). This is OK; problematic is the 
>statement that the reversed character state preserves in some respect the 
>earlier (apomorphic) character state. I see no reason to assume this (and 
>cladistic analysis does not need it - homoplasy was already resolved as 
>reversal).

Agreed. Clades can be "real classes" only for some highly stable characters 
(like presence of vertebrae in vertebrates, or presence of five digits in 
horses - in embryos  - with the additional, nested characters "presence of 
a hoof on the third digit" and "regression of lateral digits"). But 
reversals to absence, or complete substitution by another character state 
are problematic.
I admit that "presence of historically disappeared features" is a 
far-fetched abstraction, and that derived absence is problematic for "true 
classes of equivalence".
Clades are thus at best an imperfect approximation of classes of 
equivalence, this definition fitting only some "too beautiful" characters 
in some "too evident" clades.

As a consequence :
- the hierarchy of synapomorphies cannot be considered in the general case 
as making clades "natural classes of equivalence" based on such 
synapomorphies. I already objected in this thread to the notion of "natural 
classification" on other grounds (rejection of naive positivism, as if a 
classification could be "objectively read out of bare natural facts"; 
worse, the possible implication that nature would class some way). The 
"class" problem adds to the objection on other grounds (no "natural" true 
classes of equivalence, even under a strictly phylogenetic interpretation 
of the distribution of characters in taxa).
- as you stated before, differences between cladistic and eclectic 
classifications have to be searched elsewhere... see the "paraphyly" thread 
and the question of criteria for grades.

Best and many thanks for your patience,
I apologize for hair-splitting (was at least useful for me if nothing else!)
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