[Taxacom] Molecular data and synapomorphies

Richard Zander Richard.Zander at mobot.org
Tue Dec 2 10:50:53 CST 2008

Alan Forrest has good comments. 

He indicated that most morphologically based taxonomic studies are not,
in fact, supported by cultivation and crossing experiments, mating
behavior, cytology, biogeography, allozymes, fossils, chemistry, and
other helpful analyses as I indicated they were. Well, yeah, okay,
right. Shame on us. I was only defending taxonomic "traditionalists"
conceived as superannuated duffers or dancing primitive indigenes
shaking rattles and intoning Linnaean binomials. Of course funding would
help, but the present emphasis on DNA analysis does not encourage
students or new doctorates to pursue that olde tyme "biosystematics." On
the other hand, I'm seeing a new interest in cultivation studies in

Regarding kinds of data, data are already massaged in many cases. For
instance, traditional classifications using morphological data are
already evolutionary classifications. Lifting the morphological traits
individually into a data set and doing parsimony analysis makes a new
classification based on all traits, not just those judged evolutionarily
important for each taxon (i.e., those involved in stabilizing selection
over millions of years), but parsimony assumes that evolutionary traits
of one taxon affect those of another taxon, somehow. 

With this in mind and judging that parsimony is only a non-ultrametric
clustering method incorporating a very simple model of evolution
(perhaps more simple than the evolutionary relationships you infer based
on viewing particular small groups), one can get a cladistic tree that
may uncover problems, which is good, but that is a tree from a data set
that was already the support for a classification. The result after
parsimony shuffling is probably worse but looks better because it is
"parsimonious" and uses real math. Parsimony if it has any analytic
value assumes no linkage between traits, yet all traits are intensely
linked into groups particular to species or higher taxa through
evolution. The linkages of parsimony may be different from those
actually observed as clusters previously when dealing with small groups,
or as clusters left over after macroevolutionary evolution including
extinction events.

Exploding the traits into a data set is a disservice to the taxonomist
who collected the taxa and determined which traits are evolutionarily
linked (that's what we do), even if that is you. If you are a taxonomist
working on a group, evaluation of species in a group should not depend
on evaluation of species in another group, and microevolution of one
species should not affect that of another unless there is lateral gene
exchange or introgression. 

Also, molecular data are quite unlike those of morphological data sets.
Sampling of morphology has already been done and analyzed as to which
traits are conservative and therefore sufficiently linked (through
ongoing selection) to be important in taxonomy. I don't see anyone doing
total evidence studies using random sampling of just any morphological
traits from the exemplars. Molecular data (ignoring their problems)
address genetic continuity and isolation, not speciation (necessarily).
Morphological data address the results of speciation and evolution of
higher taxa. The two types of data have different uses, and should not
be conflated. A combined analysis lumps data on apples and oranges (I
typed "orangs" but my spell checker changed it to oranges).

Forrest wrote:
" The more enlightened molecular systematists are well aware of the type
of data and sequences they are using (I do not class myself as a
molecular systematist either, by the way - or indeed as anything much).
Further, to state that selection may therefore strongly affect the
analyses and interpretation of molecular data suggests that
morphological trait data does not suffer from this problem."

My comment:
There indeed are enlightened molecular systematists. They do check for
selection (usually just synonymous/nonsynonymous ratios but that is
something). Most, however, accept a well-supported molecular tree as
gospel and wrench a classification out of it. And, no, no, I meant that
results from molecular data, when supporting those from morphological
data, may be responding to the same convergence pressures, or if
different from the results from morphological data, then responding to
other confounding selection pressure, and nothing is right. 

I do have a feeling that the present resolution level of evolutionary
history is far less than we think from the apparently highly supported
results of molecular phylogenetic analysis. We may have to live with

Richard H. Zander 
Voice: 314-577-0276
Missouri Botanical Garden
PO Box 299
St. Louis, MO 63166-0299 USA
richard.zander at mobot.org
Web sites: http://www.mobot.org/plantscience/resbot/
and http://www.mobot.org/plantscience/bfna/bfnamenu.htm

More information about the Taxacom mailing list