More on the 'cladistics' of sequences

[pierre deleporte]

A 12:22 10/06/2004 -0400, John Grehan wrote:
>Meaning that most of the characters for orangutans and humans do not
>occur at all in the same state in any other primates.

Clear enough = uniform character state in all outgroups. Lucky if you find 
many of such characters. Risky if you discard other ones.

>I don't have any strong opinion on whether 'cladistic' programs are of
>themselves phenetic or not although at least one cladist has suggested
>to me his view that parsimony analysis is just a version of phenetics.

Never believe "one" cladist. They evolve and radiate a lot. Some "cladists" 
effectively rediscovered phenetics (Jaccard index). They belong to some 
trends of the "pattern cladistics" school, which is not evolutionary, thus 
you shouldn't bother if your own goal is phylogeny inference. Of course 
they don't perform standard cladistic analysis, but they call themselves 
"cladists". You better document and make up your own mind. Always.

>Yes there are three characters that may be apomorphic with respect to
>humans and orangutans, but also have representation in outgroup species.

Thus they can be either convergent, or synapomorphies at a deeper node. No 
other way. And unfortunately it's not written on the characters. Better 
enlarge your analysis.

>One involves thick dental enamel which is found in a couple of monkeys,

I heard of a problem of the nature of the enamel. Why would one retain 
thickness, if it's not the same enamel? How do you code: thick enamel of 
kind 1, and thick enamel of kind 2?  Unless we assume that there is a 
homologous physiological mechanism controlling for thickness of enamel, 
identical even for different kinds of enamel... Anyway you can hardly 
consider enamel thickness and ignore enamel structure. It thickness groups 
Homo with Pongo, but structure groups Homo with Pan... Matter to think about...

>the other two is the lack of estrus and the lack genital swelling during
>the menstrual cycle. I agree with your point that one could include
>these taxa to see what influence it has on the ingroup analysis - the
>latter two characters in particular if one could get an accurate enough
>documentation of the feature for outgroup species.

You get the point absolutely well.

>Now I realize that programs can sort out possibilities like this
>according to the criteria set for character matching. However, one could
>also establish an argument for characters in the first place.

Yes, but this is what everybody does while using a program: the argument 
for characters is given in the first place by feeding the analysis with the 
different outgroups, exactly as we just discussed the point, and agreed on.

You can make the argument by hand. But the program makes it for you exactly 
the same way, just because the programmer made the program to follow this 
very logic.
And for molecules too.

>As before, I do not reject the programs. Sorry if I gave that
>impression. I think all I was saying is that the programs are just
>recipes,

Yes, including your very own cladistic recipes.

>what they do follows from the assumptions about the
>characters in the first place so if the characters are not already
>limited to apomorphic features then the result may be just a 'cladistic'
>analysis of phenetic characters.

But your own characters are not limited to apomorphic features. There is 
thin enamel and thick enamel for instance in your very data set. This makes 
two features. You try to put thin enamel close to thin enamel, and thick 
enamel close to thick enamel, right? (note that this is one and the same 
action: grouping the ones automatically separates them from the others. 
Group thick ones, or group thin ones, the result is the same for the 
unrooted topology. Maybe you missed this basic point).
But you use only thick enamel to tend to support the rooting of Homo with 
Pongo as a monophyletic group, right?
Well, the programs do just that.

People put characters in the analysis. Like you. These characters have 
different states. Like yours. Outgroups are labelled a priori, like you do. 
Hence putative polarity of characters relative to outgroups is fixed a 
priori, with some possible ambiguities, like for your own characters. The 
program optimizes the solution considering all the states of affairs 
according to cladistic rules. Like you do, like you just stated it above.


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