[Taxacom] clique analysis in texbooks (was: Evolution of, , human-ape relationships...)

[Sergio Vargas]

I think I've send my previous email without a proper subject, sorry 
about that. Now this one:

SV: I don't think John Grehan is using compatibility to choose characters because his trees have CI<100, hence his matrices have homoplasic characters.

JG: Not sure I understand that because in the orangutan analysis there were character conflicts giving different possible relationships among the large bodies hominoids.

Hence the CI<100. The tree resulting from a matrix derived from a compatibility analysis has always CI=100. This is why I wrote you don't use a manual version of compatibility to choose among the characters to be included in the analysis.

JG: Perhaps I am wrong about that, in which case I retract the assertion, although I maintain my position that the inclusion of character states in the outgroup as well as the ingroup to analyze ingroup relationships is phenetics pure and simple.

I'm not sure about this one, using the same character matrix a phenetic analysis will group similar taxa together but a cladistic analysis would not necessarily do the same. I though is the method with which you analyse the character matrix what was phenetic or cladistic, not the character matrix it self.

SV: So instead of compatibility analysis I would say John Grehan restricts the character matrix to characters were the polarity can be decisively determined and discards characters that could yield equivocal polarity reconstructions.

JG: Yes

So I got it right! Finally. Then, and I've mentioned this before, the Human-Orangutan hypothesis can be viewed as a phylogenetic tree conditional only on characters which polarity can be decisively determined, often a priori.

I see several potential problems for a tree inferred using John Grehan's method of character pre-selection:

1. as an explanatory hypothesis, such a tree is worse than a tree resulting from a matrix also including characters which polarity cannot be decisively determined because the second tree would also provide an explanation for these, say bad, characters as well. Thus the second tree will have a higher explanatory power than  the first one.

2. from a testing perspective the first tree (JGs) is less thoroughly tested than the second one because characters that could potentially falsate (not sure how you write this word) the topology are not being used, even if the analytical method can handle them without problems and no cladistic textbook (unless Wiley and Brooks are "closet pheneticists" which will be scary) or published work recommend to exclude them from the analysis.

3. because of the way the character matrix is constructed, it seems that it is really hard to add new characters to the matrix, which means that the conclusions of the analysis are protected against future testing: you can always take characters out (those that do not fulfil JGs criterion of character selection) but never, or only very rarely, you can add characters to the matrix. Moreover, if you ignore JGs criterion of character selection you are "muddying the phylogenetic waters" and your matrix is phenetic, therefore irrelevant. If you use DNA characters you are a "molecular theorist", whatever that means.

I guess these problems make a tree inferred using this method a bad hypothesis when compared with an hypothesis that does not require excluding potentially relevant information and does not pose obstacles to further testing.

sergio


-- 
Sergio Vargas R., M.Sc.
Dept. of Earth&  Environmental Sciences
Palaeontology&  Geobiology
Ludwig-Maximilians-Universität München
Richard-Wagner-Str. 10
80333 München
Germany
tel. +49 89 2180 17929
s.vargas at lrz.uni-muenchen.de
sevra at marinemolecularevolution.org

check my webpage:
http://www.marinemolecularevolution.org

check my research ID:
http://www.researcherid.com/rid/A-5678-2011