Predictivity vs Useful

Peter Stevens peter.stevens at MOBOT.ORG
Tue Feb 15 10:04:50 CST 2005

Not specifically addressing any of the points made, but I thought a
couple of (plant) examples might be of interest.  These concern the
New Caledonian Platyspermation (previously very uncertain
relationships) going to the SW Pacific Alseuosmiaceae (Asterales) -
and when I went to look, distinctive indumentum, corolla margins,
etc. were in full agreement with new position. At least some of the
characters are synapomorphies for Alseuosmiaceae.

Two weeks ago during lab I looked at seed morphology/anatomy of a
couple of species of the African Soyauxia (ex Medusandraceae:
relationships unclear), but very recently (Davis and Chase) placed in
Peridiscaceae (tropical American: previously Malpighiales, but now
Saxifragales).  It was very distinctive, with tanniniferous coat, the
cells being more or less collapsed, very tough endosperm with
massively thickened cells, etc.  So I looked at a seed of Peridiscus
yesterday - almost identical.  Unfortunately, there is no material of
Whittonia, the third genus.

The only problem in the latter example is that I don't know the
distribution of the endosperm character in general, or in particular
within Saxifragales. However, that simply (if one had the time) needs
a bit of library work to get the basic idea. In general, one
expects/hopes/predicts hypotheses of monophyly (molecular data were
initially responsible for both sets of associations) to translate
into finding of distinctive (apomorphic) morphologies/anatomies.  Not
that everything is "perfect"; the petiole anatomy of Peridiscus is
very different from that of Soyauxia, as are the leaves themselves,
and some specimens of Peridiscus are indeed misidentified as

Peter S.

>I do not know the groups in question well enough to comment specifically, but
>I'm quite sure that predictivity would be at its highest if the questions were
>addressed to holophyletic subsets of those groups!  Perhaps the successful
>predictions are largely based on homology, whether the investigator
>is aware of
>it or not.
>--- SK¡LA ZdenÏk <skala at INCOMA.CZ> wrote:
>>  Think about groups like (dicot) herbs/trees, for example. They are extremly
>>  predictive in *many* characters and hugely polyphyletic at the same time.
>>  Best!
>>  Zdenek Skala
>>  -----Original Message-----
>>  From: Barry Roth [mailto:barry_roth at YAHOO.COM]
>>  Does anyone seriously doubt that, because of the fact of organic evolution,
>>  the
>>  classification that maximizes predictivity will be the one most fully based
>>  on
>>  genealogy/phylogeny?
>Do you Yahoo!?
>Read only the mail you want - Yahoo! Mail SpamGuard.

More information about the Taxacom mailing list