[Taxacom] Evolution of human-ape relationships, remains open for investigation
sevragorgia at gmail.com
Thu Aug 11 14:45:40 CDT 2011
just my two cents:
PH: It was my distinct impression that you do not accept that cladistic analysis is possible without first assessing the direction of the transformation series. Maybe I'm right, maybe not.
JG: Yes, that is correct - or at least that it is not cladistic analysis [...]. As a subset, I am not very comfortable with multistate characters as I do not know with any confidence how clustering algorithms treat the transitions. For a multi state character with states 1, 2, 3, 4 where the outgroup is 4 - whether the program groups 1 with 2, and then (1+2) with 3, and how the presence of intermediate states may affect the grouping of taxa where the same intermediate character state codes are present in more than one taxon. I have heard, but not seen any publication on this that I recall, that the presence of intermediate states in multiple ingroup taxa can affect the analysis. I would be interested in any comment on that as I am not an algorithm theorist.
SV: Well, this is interesting and very well documented in PAUPs old (v3.1) manual. Binary transformation series, whether restricted in a way that one character state is present in the ingroup and absent in the outgroup or not, contribute the same number of steps to a parsimony analysis independently of the polarity assessment so identifying polarity in this characters prior to the analysis is irrelevant. Simplifying, in multi-state transformation series you have two options, either you tell the software you want to impose an order to the transformation series or you don't. In the first case, following JG's example, the evolution of the character is assumed to have occurred 4->3->2->1. A transformation 2->1 costs 1 but a transformation 4->1 costs 3, thus it is likely (but not necessary) that taxa with character states 1 and 2 group together. If you don't want to impose this order, which is usually a strong assumption, you just need to run the character unordered and the cost of a 4->1 step will be 1 instead 3. You can code multi-state characters in many different ways and use branching diagrams, etc. Check PAUP v3.1 manual chapter 1, is very good.
Sergio Vargas R., M.Sc.
Dept. of Earth& Environmental Sciences
tel. +49 89 2180 17929
s.vargas at lrz.uni-muenchen.de
sevra at marinemolecularevolution.org
check my webpage:
check my research ID:
More information about the Taxacom