David H. Griffin griffin at mailbox.syr.edu
Wed May 17 08:14:58 EST 1995

In article <3p2ntd$f4j at nntp3.u.washington.edu>
joe at evolution.genetics.washington.edu (Joe Felsenstein) writes:

> No, no one has ever "proved anything" with sequence data in
> this sense.  While one can find sites where a group
> shares the same state and all others have the outgroup state,
> these are scattered among many other sites showing homoplasy,
> and also showing other states that have arisen elsewhere.
> Selecting the sites that have a clear and self-consistent
> pattern is the aim of compatibility analysis (called "cliques
> analysis" by those who dislike it).  Even this is done numerically
> after looking at all states, and like all other methods is
> subject to error.
> Phylogenetic systematists in the 1980's liked to take the
> hypothetico-deductive method as their basic framework (this is
> now mostly superseded by the "logical parsimony" framework).
> Even they did not search for compatible sites but took individual
> substitutions as refutations of each other, not whole sites.
> You should therefore look in the compatibility literature.  But
> "proof" is too strong a word, whether we are talking about
> molecular or morphological characters.
I think that Joe Felsenstein has touched on a frequently used and
misused concept, "proof". In science, proof means test, as in test an
hypothesis. This is related to the common usage of the word in
labelling alcoholic beverages, as in 80 proof gin; meaning that the
product has been tested for alcohol content.

When we do experiments to prove [test] an hypothesis the results may be
contrary to the predictions of the hypothesis, and therefore disprove
it. Alternatively, the results may be consistent with the predictions
of the hypothesis and we accept the hypothesis as "true" until we can
design a more discriminating test. In this sense the hypothesis has
been proven [tested].

Molecular phylogenetic analysis has proven repeatedly the truth of the
prediction of evolutionary theory that closely related organisms have
greater similarity in genetic sequences than more distantly related
organisms. Thus, when exceptions seem to occur in any particular
analysis, we have considerable confidence in booting the offenders out
of the tree and placing them elsewhere, unless an alternative
hypothesis can be formulated and tested. Occam's Razor often comes in
to play at this point.

Lay audiences and scientists, too, often get confused over what is
meant by proof.

David H. Griffin
Department of Environmental & Forest Biology
College of Environmental Science and Forestry
350 Illick Hall
One Forestry Drive
Syracuse NY 13210-2788
e-mail: griffin at mailbox.syr.edu

More information about the Mol-evol mailing list

Send comments to us at biosci-help [At] net.bio.net