What does it mean when MacClade (through human manipulation)
comes up with a tree that is two steps shorter than the best one
PAUP can produce?
Apart from the settings... stupid question, but... have you restricted the
PAUP* analysis to, say, a certain number of replications because otherwise
the computer would be occupied for the next 20 or 30 hours? I almost always
limit my analyses, and my matrix isn't all that big (though full of question
marks). When I limit the wrong way, I don't always get any of the shortest
trees.