Molecular taxonomy: on way out?

[John Grehan]

> From: Les Watson [mailto:leswatson at westnet.com.au]
> Sent: Thursday, July 21, 2005 5:31 AM
> To: TAXACOM at LISTSERV.NHM.KU.EDU
> >> While it has been apparent for about forty years that comparative
DNA
> >> studies would ultimately provide conclusive evidence for
understanding
> >> phylogeny, ....
> 
> >Apparent? I would say claimed.
> 
> I thought it obvious that comparisons of organisms' COMPLETE summaries
> of themselves would offer the best evidence we are likely to get, and
> are potentially far more reliable than the relatively crude
indications
> torn from them via comparative morphology, etc. 

This is a claim, not something that is apparent. The idea that one has
'complete' data seems to reflect the pheneticist philosophy that the
more characters one has the better the phylogeny.

I still think so,
> especially now that DNA sequencing and comparisons are automated and
> rather easy to perform. 

This does not change the non-cladistic nature of the characters being
used. It also does not change the fact that one cannot individually
justify each DNA sequence match as a derived homology but it seems that
one has to take it on algorithmic faith.

It has all taken longer than one might have
> expected, but even the fascinating complications of repeated DNA,
> overlapping genes, inversions, etc. strike me as minor problems,
compared
> with the uncertainties involved in arbitrarily choosing which aspects
of
> phenotypic variation to pursue, 

The view that DNA sequence characters necessarily give a better or more
reliable phylogenetic signal is an unsubstantiated claim. The real
sequence comparisons for phylogenetic splitting might be those involved
with the development of a uniquely shared morphological structure - in
which case the genetic correlation is a series of sequences scattered
about in different locations.

The choice of which aspects of phenotypic variation is not arbitrary in
that one has to (or I should say should because oftentimes people do
not) justify the choice and polarity. Also, it is not arbitrary in a
cladistic sense in that only uniquely shared features have to be
included.

> >I take the view that even with adequate taxonomic coverage the
current
> >systematic approaches to DNA sequences will not preclude DNA trees
> >from being phenetic as the cladistic approaches are only mimics.
> 
> You may well be right, 

Thank you for at least acknowledging the possibility. That puts you at
odds with other TAXACOM people who reject the possibility.

especially re "the current systematic
> approaches"; but total DNA extracts contain ALL the available
> information for extant organisms, in relatively accessible form.

The point about cladistics is that not ALL information is necessarily
phylogenetically informative. As you admit above, if the data is
phenetic (ie. Not restricted to derived character states for the
analysis) then having more characters does not change anything. One
might have 100 measures of similarity between humans and chimpanzees
compared with two for humans and orangutans, for example, but if 99 of
those measures with the chimpanzee are actually primitive states then
the total number is not phylogenetically informative.

> I haven't been there for a long time; but to a botanist watching from
> the sidelines, the Nature referees/editor seemed exasperatingly less
> critical .... they were forever publishing
> descriptions of 'new' hominid taxa, without requiring the authors even
> to state (let alone demonstrate) that their bits of bones fell outside
> the range of variation found in living humans. Perhaps not much has
> changed?

Cheers, cheers. Crappy morphology is crappy morphology, although I would
say the same for DNA sequences as well.

John Grehan