[Taxacom] Why character-tracking doesn't happen?

Neil Bell neil.bell at helsinki.fi
Sat Sep 13 08:36:38 CDT 2008


Bob Mesibov wrote:

> That's nearly what I'm thinking. I started by asking 'How can we test a
> phylogenetic hypothesis?' The many existing 'tests' (think of
> bootstrapping, jackknifing, etc) only examine how the tree was derived
> from the input data. They don't test the *result*, i.e. the tree. I
> began to wonder whether further study of the character states involved
> and how they are related on the tree (character-tracking) could generate
> evidence to assess the plausibility of the tree.
>
> I'm still wondering that. It's always amazed me that after the
> publication of so many trees there is no big compendium 'The
> Evolutionist's Book of Homoplasies' which reviews and considers not only
> character-tracking for genuine homologies, but also the non-homologies
> that Platnick referred to.

In fact something like what you are suggesting happens informally all of 
the time. Any tree immediately reveals homoplasies of particular 
morphological features to anyone familiar with the organisms. Some of 
these will be more credible than others, and this will act as a spur for 
further research (e.g. further sampling) or critical examination of the 
data or methodology. Some people seem to think that such an "iterative" 
approach is unscientific, but like any search strategy I think it isn't 
how you get there that matters, more how you evaluate your postion once 
you are there.


Bob Mesibov wrote:
> Many thanks for the interesting nibbles. An analysis of the discussion
> so far would show incipient speciation with incomplete lineage sorting,
> but that's OK!
>
> Pierre Deleporte wrote:
>
> "if I understand well your proposition, the secondary inference of 
> homoplasy for some characters would make the phylogeny inacceptable"
>
> That's nearly what I'm thinking. I started by asking 'How can we test a
> phylogenetic hypothesis?' The many existing 'tests' (think of
> bootstrapping, jackknifing, etc) only examine how the tree was derived
> from the input data. They don't test the *result*, i.e. the tree. I
> began to wonder whether further study of the character states involved
> and how they are related on the tree (character-tracking) could generate
> evidence to assess the plausibility of the tree.
>
> I'm still wondering that. It's always amazed me that after the
> publication of so many trees there is no big compendium 'The
> Evolutionist's Book of Homoplasies' which reviews and considers not only
> character-tracking for genuine homologies, but also the non-homologies
> that Platnick referred to.
>
> Steve Manning wrote:
>
> "Second, it seems to me that probability needs to be put front and 
> center both with character tracking and with phylogenetic 
> hypotheses."
>
> I couldn't agree more, although it would be hard to develop a
> statistical-inferential methodology for evaluating homologies, which is
> what character-tracking would involve. The method might eventually arise
> from developmental genetics, which follows the links between sequences
> and phenotypes.
>
> I also agree with you that many phylogeneticists don't appreciate that
> the uncertainties in their trees need to be made explicit. Bootstrapping
> and leaving models up to MODELTEST aren't enough. I personally like the
> networks approach, e.g. with SplitsTree4 software, where the output is a
> network which graphically displays areas of uncertainty in the
> relationships.
>
> Neil Bell wrote:
>
> "Incomplete lineage sorting and 
> reticulation occur and should be identified, but they are not the 
> dominant patterns in the relationships between phylogenetically distant 
> eukaryotic organisms."
>
> You've identified an interesting methodological issue, namely the
> enthusiasm with which some methods are applied at every level of the
> lineage hierarchy. The same software is used to generate trees showing
> relationships between orders and between populations. Same reasoning,
> same assumptions, but the underlying set of objects being studied have
> different behaviours and change at vastly different time-scales. I think
> there could be a problem here.
>   





More information about the Taxacom mailing list