species concepts, etc

[James Francis Lyons-Weiler]

On Wed, 22 Oct 1997, Dave Chesmore wrote:

> I have just returned from a European Science Foundation meeting on
> Systematic Biology which is aiming to develop a future programme in
> systematic biology.
>
> I presented a paper on computer-aided taxonomy which includes
> within its remit automated species identification, key systems for
> identification and all aspects of computer applications in systematics.
>  I am attempting to define the concept mroe closely.
>
> At the meeting, I was not entirely surprised that different phylogenetic
> approaches (morphology, rbcl, r-DNA) and analysis methods
> (parsimony, etc) produced different phyogenies.  What DID surprise
> me is that there appears to be no concensus as to which is best.
> Each person appeared to have his/her own favorite method!
>
> My question is - does anyone know of any research, papers, etc for
> comparative studies of the various methods giving objective
> comparisons?
>
> ====================================================

People have tried to make such comparisons, such as "morphology vs.
molecules", but in fact it is quite difficult to make objective
sense out of it.  I think perhaps the emerging consensus is that there is
no +real+ distinction among different data types, all have limitations,
each will work in some cases well and in others not at all, and that
different tree estimation procedures all have limitations, and that some
will work well (i.e., they will be consistent, i.e., lead to the right
answer with an infinite amount of data) under some conditions and
inconsistent under others.

I would recommend, for starters, and for a then up-to-date capture of the
current "knowledge" of the utility of data types and methods of inference,
Swofford et al.'s chapter in the second edition of Molecular Systematics
(Sinauer), 1996, eds are Hillis, Moritz and Mable.  For comparisons of
molecules and morphology, if you still find it an interesting question,
look up Colin Pattersons' first and second editions of Molecules and
Morphology in Evolution: Conflict or Compromise (Cambridge U Press, 1987
1st ed., 2n ed. forthcoming, I believe).  There have been numerous
simulation comparisons of the consistency of methods (see refs in
Swofford et al. for starters), but be forewarned of the many limitations
of the simulation studies, like proponents of each method seem to find
a region where their method outperforms others; that maximum likelihood
is doomed to look entirely consistent because those performing the
simulations are omniscient, and have the right model, whereas in real
life we don't know if we do; that the results are generally
optimistically interpreted (consistent over a large range of parameter
space = inconsistent over much of the range , <half full, half empty>),
and many others.

For a primer on the total evidence vs. combine vs. consensus issue, see
my past postings to TAXACOM (I have sent a complete biblio); if that
proves impossible (as it often does), see the biblio on everything on
methods in the .sea file of the rasa software (you need a mac) at

http://loco.biology.unr.edu/archives/rasa/rasa.html

There should be a word file (other.refs) that contains much of what you
might find relevant.

James Lyons-Weiler