species concepts, etc

[ROGER HYAM]

Dear all,

Dave Chesmore's question to the group started off about species
concepts and species identification but then spoke about phylogenetic
reconstruction, in a ML/parsimony sense.

James Lyons-Weiler's reply then suggested some great texts on the
problems of phylogenetic reconstruction using different types of data
and different methodologies.

I think two subjects are being confused here. My understanding is
that phylogenetics deals fundamentally with the arrangement of
entities into a hierarchy that is based on a branching diagram. The
assumption made at the beginning of the analysis is that the
relationships between these entities can be represented as a
branching diagram.

The problem arises in that one first has to decide what these
entities are. The relationships amongst sexual individual can not be
shown as a branching diagram (try and draw your family tree without
any of the parents meeting). At higher levels of taxonomy (species?)
the relationships can be drawn as a branching tree and as many
studies have shown this works very well. The question is what
happens inbetween.

There is much debate over this subject and there will continue to be
much debate.

There is no answer for Dave's original question. I can't tell
absolutely him how to tell his computer what is and is not a species.
When a working taxonomist identifies a specimen he/she works their
way through the phylogeny, using all the synapomorphies for the
different groups (i.e. using keys?!) until she/he gets close. They
then go to the herbarium (or other collection) and compare the
specimen with specimens there. They match it to a specific specimen
(perhaps even a type) and that is the det. In carrying out this
proceedure the concept of what the species is changes. The taxonomist
has learnt. The collection may have grown (if the specimen is added).
A computer carring out this task would be an expert system. It would
derive and define taxa that it 'understood' - just like sending your
specimens to a real specialist.

If we make a computer behave like a taxonomist are we really
progressing? Yes we are in terms of speed but what about in terms of
understanding the biology? If, as many would say, the delimitation of
species is often arbitrary and subjective, atleast when a human does
it it is likely to result in groupings that make sense to other
humans. If a machine makes 'arbitrary' decisions they may make no
sense to us at all and we will just dismiss them.

With regard to computer taxonomy (not computer systematics) it is
very much a case of garbage in garbage out, not in the sense of the
data but of the concepts surrounding the data model.

I am afraid I have gone on a bit - and I dislike long postings
myself!

I only get the list as a digest so please excuse me if my points are
behind the debate on this strand.

Roger.




*************************
Dr Roger Hyam
Royal Botanic Garden,
Inverleith Row,
Edinburgh, EH3 5LR
Scotland, U.K.
Tel. 0131 552 7171 x 410
Fax. 0131 552 0382
http://www.rbge.org.uk
*************************