genetic vs morphological trace of phylogeny

[John Grehan]

At 03:06 PM 4/5/04 +0200, pierre deleporte wrote:
>But fortunately, no predetermined choice is needed. The current practice 
>is, not to use only morphological similarities (which are not in 
>themselves totally congruent) or only molecular ones (themeselves not 
>totally congruent either), but to combine all available, and potentially 
>relevant, information (molecular data sets + morphological ones), which 
>you still seem highly reluctant to do in practice.

"Current practice" of itself is not a delimiter of scientific progress. It 
was once "current practice" to regard the world as flat.

Combining morphological and molecular data sets would, in my mind, be 
justified if the characters were of 'equal' merit. In my view (admittedly 
in disagreement with others) they are not. Morphological synapomorphies are 
individually testable (the criterion for being uniquely shared can be 
corroborated or refuted case by case - and notice that on this list that 
the veracity of the orangutan characters has not come under any concerted 
refutation). DNA sequence data (which is most of what the molecular stuff 
is about) seems to be more problematic. Despite the assertion that DNA 
sequence data is 'more' reliable than morphology, there are elements of 
ambiguity. I met up with a practicing molecular phylogeneticist this week 
who agreed that there is ambiguity in molecular characters and there is no 
pre-ordained reason why molecular phylogenies are necessarily more reliable 
than those of morphology.

>This can be done through all kinds of analyses: cladistic under various 
>possible weighting schemes, and also maximum likelihood under a variety of 
>models (a few recent papers deal with the last procedure)...

Sorting procedures are sorting procedures. They don't change the nature of 
the character input.

>There are lasting problems as for which models of character evolution 
>should be implemented, both for morphoilogy or molecules (see many other 
>threads), but I never read a convincing argument that some potential 
>source of phylogenetic information should be completely discarded.

AGREED!!!!!!!! I'm really glad to see some agreement here. However it is 
the molecular primate phylogeneticists who are making just that very 
assertion - that morphology should be completely discarded!!!!!!!! They are 
saying that the problem of our nearest relative has been completely solved 
- it is a fact that the chimp is our nearest relative.

>Ignoring molecular information for the Pongo question is undefendable. You 
>can easily join with a molecularist and perform a complete analysis of all 
>available potential evidence.

It is certainly one of three main options. As above, this would have merit 
if the characters were of some equal status. At present the molecular 
characters appear to be phenetic so I would not combine them with cladistic 
characters any more than I would for phenetic morphological characters 
(chimps and humans share far more features of overall similarity than 
either does with the orangutan for example).

>If you want to. I think this would be a more adequate presentation of the 
>present state of phylogenetics than considering morphological evidence in 
>isolation.

If morphological characters are to have scientific status then they should 
have some independent status as evidence. The results of that evidence can 
certainly be compared with other corroborating or conflicting evidence, but 
not necessarily subordinated to other evidence due to incongruence.

It is my current view that molecular characters may track phylogeny, but 
not necessarily. The problem of that relationship arises when the molecular 
and morphological characters are incongruent. There is no current recipe 
for resolution. Some may opt to reject morphology as unreliable, some opt 
to see molecular characters as unreliable, some may opt to see both as 
having ambiguity (my preference). In the orangutan case my view is the 
number and characteristics of the human-orangutan are sufficiently 
extensive (as a purely personal judgement) that they become really 
problematic if we are not so closely related. There is also the predictive 
value of a theory. If the chimpanzee-genetic model is the best, it 
certainly appears to be a poor predictor of fossil hominid morphology. Lucy 
and her relatives do have a skull structure that is far more orangutan-like 
than could be predicted from a chimpanzee model.

John Grehan


>Best,
>Pierre
>
>
>>John Grehan
>
>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