Re: What's the Science of Dinosaurs?

["David Marjanovic" <david.marjanovic@gmx.at>]

> I said:
>  As an exemplar, is cladistics science?
>  The short answer is No.
>  The long answer is also No.
> You replied:
> Strange. Cladistics produces testable hypotheses based on observations of
> facts.
>
> But the test is only within a cladistic argument, using the same selected
> cladistic logic.  What I choose to include and how I choose to include it
> and what kind of cladistic logic I use will affect the result.

What "cladistic logics" do you mean? Max. parsimony vs max. likelihood vs
neighbor-joining? Testing methods like bootstrapping and jackknifing? This
will affect the result, so when given the choice (in molecular biology
only), I'd try all three methods, test them both ways and publish all
results.

What characters and what taxa you include of course affects the result. The
more taxa and (to a certain degree) the more characters, the better = the
more probable the outcome, no?

> Does this type of test sound equivalent to the scientific method's test of
a
> hypothesis by determining an implication of the hypothesis which might be
> found/not found in reality and then doing the experiment?

Sure. As I told you (HP Philidor11) offlist, HP Mickey Mortimer has thought
months ago that the loss of gastralia is one supporting character for the
clade (*Jibeinia* + Ornithothoraces). The (implicit) prediction is that no
ornithothoracine will ever be found with gastralia. Now we have
*Yixianornis* -- prediction falsified. The difference to, say, physics here
is that this isn't really an experiment, we are restricted to what we find
by chance (largely).

> Remember, this is the null hypothesis, the proposition that cladistics, as
a
> specific example, is not a science.  Before you can say your conclusions
are
> true in reality, you have to defeat the hypothesis.

Cladistics makes testable predictions. The tests may not be easy, but that's
it.

I think what bothers you is that morphological cladistics is totally
dependent on max. parsimony, while evolution need not be most parsimonious?
Well... it's all what we have. Save for testing methods. The following paper
has a most parsimonious tree and a bootstrap tree which differs from the
MPT. In the text the authors write what causes the differences, and don't
prefer one tree.

Michel Laurin & Robert R. Reisz: A new study of *Solenodonsaurus janenschi*,
and a reconsideration of amniote origins and stegocephalian evolution, Can.
J. Earth Sci. 36, 1239 -- 1255 (1999)

MPT (strongly simplified, outgroups omitted):

+--+--*Acanthostega*
|     `--*Ichthyostega*
`--+--*Tulerpeton*
     `--+--+--*Crassigyrinus*
          |     `--Baphetidae
          `--+--Colosteidae
               `--+--Temnospondyli
                    `--+--+--Gephyrostegidae
                         |     `--Embolomeri
                         `--+--Seymouriamorpha
                              `--+--*Westlothiana*
                                   `--+--Amphibia
                                        `--+--*Solenodonsaurus*
                                             `--+--Diadectomorpha
                                                  `--Amniota
Amphibia, defined as {Lissamphibia > Amniota}, contains Aistopoda,
"Lepospondyli", "Microsauria" etc. and Lissamphibia.

Bootstrap tree (strongly simplified, outgroups omitted):

+--*Acanthostega*
`--+--*Ichthyostega*
     `--+--*Tulerpeton*
          `--+--+--*Crassigyrinus*
               |     `--Baphetidae
               `--+--Colosteidae
                    `--+--Embolomeri
                         `--+--Temnospondyli
                              `--+--Gephyrostegidae
                                   `--+--Seymouriamorpha
                                        `--+--+--*Westlothiana*
                                             |    `--rest of Amphibia
                                             `--+--*Solenodonsaurus*
                                                  `--+--Diadectomorpha
                                                       `--Amniota

Does anyone who has _done_ cladistic analyses want to join this discussion?
:-)