Richard Zander bryo at AGIS.AG.NET
Wed Mar 5 23:15:41 CST 1997

I thought evolution was a process. Patterns are okay for classification,
but evolutionary trees chart processes, don't they? even if they are
only nested sister groups. Since species evolve, not characters, I
figure that the number of speciation events is a measure of probability
and that is introduced by the terminal taxa selected: say a dog, a wolf
and a cat. No matter what pattern the algorithm comes up with, the
homoplasies can be evaluated as independently evolved since we can see,
from other inferential evidence, that two of the taxa share lots more
characters, phenetically, plus observations of hybridization, etc. If
the three taxa were say, x, y and z, then perhaps the comparative
numbers of autapomorphies might hint at which taxon is farthest
phylogenetically in terms of speciation events. In any case, the
algorithm gives best fit, but the terminal taxa themselves determine the
probability that you are correct in evolutionary inferences. Detail in a
cladogram is just extra resolution contributed by the mathematics,
actual evolutionary inferences are certainly better than with
omnispection or phenetics, but the probability that one is correct
depends on how the terminal taxa represent their differences in terms of
numbers of speciation events. The more speciation events (e.g. as
evidenced by many autapomorphies compared to synapomorphies in a given
lineage), the less probability that your best fit has any phylogenetic

Richard H. Zander, Buffalo Museum of Science
1020 Humboldt Pkwy, Buffalo, NY 14211 USA bryo at ag.net

More information about the Taxacom mailing list