[Taxacom] Mona Lisa Smile

[pierre deleporte]

A 09:41 27/07/2006 -0500, Thomas Lammers wrote :

>Any classification seeks to express "relationships."  Items are classified
>together because they relate to one another in some fashion: appearance,
>function, use, origin, etc.

Thanks for comments. We agree that many different kinds of criteria may be 
used for classifying the same living things different ways, hence I 
consider that we agree that the notion of "the natural classification" is 
ruled out as irrealistic. Classification is a matter of convenience.

>The question then is what relationship shall we examine among living
>things?  To my thinking, the relationship that results from descent from a
>common ancestor is likely to yield the best classification.
>Things that share a common ancestor more often than not will have the maximal
>number of things in common.

The quality you're expecting from a classification seems now clear: you 
want to class together living things which "have the maximal number of 
things in common", and maybe you are right that phylogenetic relationships 
will more often than not (?) fit such a classification.
By the way, methods for achieving such a classificatory goal are well 
known: they consist, by definition, in phenetic clustering. And if you 
could specify at little bit more which kind of "overal similarity" you 
want, you'll probably be able to choose among the potentially illimited 
number of overall similarity or distance measures and the various possible 
clustering procedures.

It also seems obvious that you are looking after one single classification, 
which you try to conceive as an acceptable compromise for general use. It 
looks at first sight like a kind of Kinman system combining paraphyletic 
and monophyletic groups, but, if I understand well,  with the striking 
difference that you explicitly privilegiate overall similarity, hence 
accepting occasional monophyletic grouping as a side effect (when Ken's 
system is strictly phylogenetic, with some paraphyletic namings).

>This is not always the case, and that is where
>I differ from strict cladists.  Depending on the environment a lineage
>faces, it may diverge considerably from its sister group as it maximizes
>survivorship and fecundity.  I have no problem lopping off such discordant
>elements and leaving behind a homogeneous if paraphyletic group.  I have no
>problem with groups like "fish" or "reptiles" -- they are useful.

I have no problem with paraphyletic "fish without cows" for communicating 
with my fishmonger: it is useful in this context (social tradition and 
preservation techniques specific to fish and meat); and I have no problem 
either with monophyletic "fish including cows" for communicating with my 
phylogenetician colleagues and for teaching phylogenetic inference: it is 
useful too (non-ambiguous monophyletic naming is convenient in this context).
I really have no problem, from the moment when each classification is 
presented for what it is, so that I can't mistake about their meaning and 
context of relevance.

But my precise point is: for which purpose do you consider your system to 
be "the best classification"? The best for whom, what for, and why? Being 
"the best" has no meaning without specifying a context or relevance and the 
related, logically inferred specification for an optimal classification.
Why not everybody his own optimal system rather than an inevitably 
sub-optimal, if not far from optimal, compromise? Christopher Rogers needs 
an ecotoxicologically relevant classification, not overall similarity it 
seems...

Is your "compromise" option constrained by the a priori choice for a 
unique, universal classification ? My point is simple: why not reject this 
constraint?
(the same vails for people dreaming to impose a unique self-consistent 
classification for all purposes).

Pierre

>"Today's mighty oak is yesterday's nut that stood his ground."
>                                                                -- Anonymous

Many a nut would better change grounds under the present drought
                                                                  -- Peter


Pierre Deleporte
CNRS UMR 6552 - Station Biologique de Paimpont
F-35380 Paimpont   FRANCE
Téléphone : 02 99 61 81 63
Télécopie : 02 99 61 81 88


_______________________________________________
Taxacom mailing list
Taxacom at mailman.nhm.ku.edu
http://mailman.nhm.ku.edu/mailman/listinfo/taxacom