Compatibility assumes that in the tree, each
character state occupies a distinct subtree.
Equivalently, each state of each character
arises exactly once in the tree. The
morphological character vertebrate/invertebrate
presumably has this property. Finding the
tree on which the maximum number of characters
are compatible is the compatibility criterion.
Parsimony, on the other hand, does not assume
any character has this property, but seeks to
minimize the total amount of evolutionary change.
It seems obvious that many datasets are best
analyzed under parsimony rather than compatibility,
so this is a request for a discussion of whether
there are data sets for which the compatibility
criterion is as good or better than parsimony.
It would also be interesting to hear if any of you
think there is information of this sort, even if
it's not yet available in "data sets". The property
that such data would have to have is that each
state of each character would arise with such low
probability that perforce the states would occupy
subtrees. Genomic rearrangements, for example, might
be one such property. ANy thoughts on this?
Department of Computer and Information Science
University of Pennsylvania
Philadelphia PA 19104-6389
tandy at central.cis.upenn.edu