More on the 'cladistics' of sequences

[pierre deleporte]

A 16:41 08/06/2004 +0200, vous avez écrit :
>I do not want to reinvent the weels but still - there is one side fact 
>that seems to be often overlooked:
>
>Outgroup(s) is(are) used for character matrix polarization in an "absolute 
>way", i.e. the character states found in outgroup(s) are believed to 
>represent plesiomorphic states for the group in question. Consequently, 
>all the "homoplasy burden" is placed on the group studied. This seems to 
>be unbalanced approach - the character states found in outgroup should be 
>given some probability of being homoplastic too; there is no reason why 
>the homoplasies should arise only in the group under question while 
>outgroup would be homoplasy-free. (perhaps even contrary is true - 
>outgroup is by definition longer branch than any of the internal branches 
>of the group studied, so the probability of homoplasies (provided being a 
>function of the evolutionary time) is higher in the outgroup than in the 
>internal branches).
>This will of course further obscure the outgroup-ingroup relations, 
>especially when taking into account that the clade structure of the 
>multiple outgroups is unknown.
>
>Best!
>Zdenek Skala

I think that using several outgroups at the same time is a partial answer. 
Different outgroup species will have different character states for the 
same character. Hence you can't always decide for sure which is the correct 
plesiomorphioc state (unlike what is suggested by John Grehan's method as 
far as I understand it), you can have one, or SEVERAL putative 
symplesiomorphic states for some characters. Anyway if all your outgroups 
root in the same place, the congruence criterion optimizing the topology 
and rooting for all characters will have done the trick.
Possibly on the basis of few unambiguous characters (maybe those John would 
have not thrown away).
The internal topological arrangement of outgroups don't seem to be a 
problem for me: the overall optimal one is taken into account in the 
analysis, as far as I know, even if not represented on the resulting tree.

Best,
Pierre



Pierre Deleporte
CNRS UMR 6552 - Station Biologique de Paimpont
F-35380 Paimpont   FRANCE
Téléphone : 02 99 61 81 66
Télécopie : 02 99 61 81 88