NUCLEIC ACID SEQUENCE PHYLOGENY AND RANDOM OUTGROUPS

Abstract— When divergent taxa are used to root networks, it is assumed that the character stales in the outgroup have historical similarity to those in the ingroup. Yet, if the data are nucleic acid sequences, the character stales shared by a divergent outgroup may be based not on history but on random similarity. A simple procedure is proposed to test this possibility. In the absence of an appropriate outgroup, root position can be estimated with the use of an asymmetrical character transformation matrix. If the matrix is sufficiently biased, it can supply the polarity information usually derived from an outgroup. This outgroup test and rooting procedure are demonstrated with ADH sequences from the genus Drosophila.