The outgroup problem

Peter Bell bell-peter at yale.edu
Fri Jun 4 14:07:03 EST 1993


In article <1993Jun4.110336.204196 at uctvax.uct.ac.za> harley at chempath.uct.ac.za writes:
>I have a topic which might engender some constructive debate, and can be put as
>the proposal *the outgroup is not appropriate as a way of rooting molecular
>phylogenies*. This is, of course, a bit extreme, but I am coming to the

>between the devil and the deep blue sea: if the outgroup is too close it may
>not really be a valid outgroup, if too far then the number of sites genuinly
>identifying the plesiomorphic state may be too few, and mixed with so many
>homoplasies (back & parallel mutations, esp if there is a high transition bias)
>as to be worse than useless.


Basically, you have to use multiple outgroups.  Morphologists have been hip
to this one for awhile-- it is a real good way to extend your  thesis 
indefinitely, too  "well, I think I need one more outgroup......"  It won't
solve all the issues, but it will take care of a number of them.  I think 
(someone correct me cuz I know I'm wrong) that it is a good idea to use 
4 outgoups with two taxa each from two clades to avoid a lot of these issues.
More, of course, if your "outgroups" raise questions at the sites you are 
interested in.  and there you are, the infinite "one more outgroup" loop.

Peter
bell at minerva.cis.yale.edu



More information about the Mol-evol mailing list