* OK, _if_ homoplasy is sufficiently rare, then phenetics and cladistics
give the same tree...
Er, sorry: the same _unrooted_ tree. UPGMA, WPGMA and the like always
produce rooted trees... and like them NJ produces only one tree even if
several would have the same length. So when a given data set with
sufficiently little homoplasy has only one MPT, then NJ will find it (as
will likelihood methods).