Topology constraints in PAUP*

[Mike Taylor <mike@indexdata.com>]

I'm having trouble using topology constraints in PAUP*, and would
appreciate advice from anyone who's experienced in this area.

I have a matrix that generates a reasonably resolved strict consensus
tree, and I want to add some poorly representative OTUs that are
scored for only a few of the characters.  Simply adding these greatly
degrades the resolution of the consensus tree, and I end up with a big
polytomy of 80% of my OTUs.  So what I'd like to do instead is pin
down the topology recovered as strict consensus of the original
matrix, and instruct PAUP* just to place my new OTUs as best it can on
that tree.

Simple, right?  Just run the matrix without the new OTUs, then look
at the strict consensus tree, and feed that into PAUP* as a
topological constraint, and add all the poorly represented new OTUs.
Nope -- it all goes badgers up.  For example, a neosauropod OTU
resolves as a ... theropod.

Advice welcome!

 _/|_    ___________________________________________________________________
/o ) \/  Mike Taylor    <mike@indexdata.com>    http://www.miketaylor.org.uk
)_v__/\  It's hard not to suspect that SOAP was thought up late one Friday
         night after a few too many beers.