Richard Zander bryo at AGIS.AG.NET
Sat Mar 8 01:12:00 CST 1997

I thought the same, that random data sets surely couldn't be used to
derive a single, nicely resolved tree. After some false starts, I
managed to write a successful program that generated ramdom sets (using
Borland's C randomize and random functions), and my first set of 20 X 20
matrix gave me one (1) fully resolved tree! Other random sets gave 2,
dozens or hundreds, depending on the set.

Tom DiBenedetto wrote:
> , James Francis Lyons-Weiler wrote:
> >The answer to the first question (does the parsimony algorithm reveal a
> >truly spurious pattern) is, precisely, yes.
> >How would it do that?  There exists for EVERY matrix (with variable
> >states among taxa) a set of shortest trees.
> First off, a set of shortest trees can mean a lot of trees. We have a
> convention of calculating strict consensuses of sets of MP trees
> which end up returning an unresolved topology for many such sets. The
> more pointed question is whether random data ever returns single MP
> trees, or sets that preserve resolution through strict consensus
> calculations even for large data sets.


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