[Taxacom] quote of the week

[Brian O'Meara]

On Mar 21, 2008, at 6:31 PM, John Grehan wrote:
> Here's where I will probably show even more ignorance than usual.  
> If one only has to align five taxa (orang, chimp, bonobo, gorilla,  
> human) is suggested below what is the point if there is no outgroup  
> for analysis? I have seen quite a few analyses that treat the  
> orangutan as the outgroup which means, of course, that the analysis  
> contributes nothing to the orangutan question. In other analyses  
> only one or two species of gibbon or monkey are used. If that were  
> done in morphology one would be laughed out of the room (I take  
> that back, evidently that is no longer the case).
>

Yes, it would be better to have an outgroup, but even without it,  
there is information about the groupings. If orang were sister to  
human, the unrooted tree would have a bipartition between  
(human,orang) and (chimp, bonobo, gorilla). Of course, this  
bipartition is consistent with several rooted trees, such as human  
sister to everything else, but not with any rooted trees having a  
(human, chimp, bonobo) clade, and so it would be evidence against the  
standard view of great ape relationships. If the true rooted tree has  
a (human, chimp, bonobo) clade, this would show up in the unrooted  
tree as a bipartition between (gorilla, orang) and (human, chimp,  
bonobo). Thus, the unrooted tree provides some information about  
whether a particular clade is possible under any rooting of the tree.  
It's actually also possible (but probably not a good idea, in  
general) to get a rooted tree back from a tree search with no  
outgroups specified by using UPGMA, a likelihood search with  
molecular clock enforced, or with some of the search options in BEAST.

Brian