[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]

Re: D.Marjanovics repliesto J'r pts I-II



> > DP: [...] The point of using photographs is to see ALOT of specimens,
> > a necessity when working up a cladogram.
>
> The _other_ way around. If you don't see all the specimens in person,
> you can't make a publishable cladogram.
>
> >>>>> David, David, David. Please show me one cladogram in which all of
> the specimens included were personally examined. Certainly _not_the case
> in Unwin (2003) and Kellner (2003). They confessed as much. And if the
> reader knows going into the manuscript that data was obtained in a
> variety of ways, what's the problem?

The problem is simply the quality of the data. Almost all descriptions in
the literature don't mention what some preserved detail that one would like
to code looks like. Photos are practically always too small and not always
taken from all angles one would need. (Having 2 photos from 2 different
angles of the same specimen would, BTW, give you the opportunity to test if
some of the "ephemeral features" are just faces and pyramids on Mars.)
Drawings are never sufficient. This may be an overgeneralization, but I
don't think so. Many are quite superficial and done with thick lines and in
2D.
        There are a few papers whose authors have looked at every specimen
they've coded. HP Oliver Rauhut's dissertation and follow-up paper are
examples. The authors of most analyses that have been published in the last
few years have looked at most or all specimens in person. Older papers often
had lower standards. This, along with the small size of their matrices, is
why they are so bad. :-|

> My "famous" bird analysis is entirely
> unpublishable, "even though" I recently added a few taxa and a few
> characters (I'll add a few more characters, run the analysis, and show
> you the results some weeks or so from now), and "even though" it
> contains many taxa and several characters that have never been put into
> a data matrix before.
>
> >>>>>> Maybe you ought to try to publish it. After all, it's only data.

No, it's not data. At least half is _guesswork_ -- my personal
interpretation of ambiguous wordings in descriptions or miserable
illustrations. Besides, you'll notice that I have question marks for *Anas
platyrhynchos* -- there is no publishable excuse for why I didn't look up
what states the living mallard duck has for those characters!

> Unossified transverse processes??? There are plenty of dinosaurs in
> which (of course fully ossified) vertebrae _with ordinary dorsal ribs_
> are located between the preacetabular processes.
>
> >>>>>> Actually the processes are ossified. The final connection to the
> ilium is  less obvious. Find a pterosaur with only three sacrals and
> you'll see what I'm talking about. We'll talk about it if you're still
> having trouble.

OK, I'll try to find some illustrations of pterosaur sacrals in dorsal or
ventral view... but I have very little primary literature on pterosaurs. I
don't have much literature at all besides Wellnhofer's encyclopedia.

> > DP: You missed my point. The result they wanted to get, I can
> double-dog
> guarantee you, is a single tree. Unwin got 6 trees. Kellner: 80.
>
> In the real world, you don't get a single tree more often than once a
> year.
>
> >>>>>> Not the point. What they wanted and what they got was the point.
> And yet, maybe a little more work would reveal the single tree. After
> all, Nature is a single tree. I say don't publish until you have a
> single tree. Otherwise you're showing your lack of ability to finish a
> job and do it right.

Your... idealism is breathtaking.

Yes, Nature is a single tree, so we should ideally get a single tree if we
look for shortest -- or most likely! -- trees. But to get that tree, we
would need all nature in our matrix. Not only would we need from 1000
characters upwards for a pterosaur matrix.* What proportion of _all
pterosaur species that have existed_ do you think there are in your matrix?
1 %? 0.1 %? With samples this small, the chance is high that we introduce
huge artifacts. Remember how Mickey always reports on what happens when he
adds one more coelurosaur to his matrix.
        In short, we _are_ unable to finish the job in the foreseeable
future. Period.

* Theoretical studies -- I remember an SVP meeting abstract -- show that
"enough is enough", from a certain number of characters upwards the results
don't get better anymore. That number is, however, high. On the other hand,
adding taxa always helps, if they're not _too_ scrappy (I'll certainly run
my bird analysis with and without the 7-character *Alexornis*).

> > JH: Examples exist among the salamanders:
> >
> > DP: Why didn't you just say insects?
>
> Juvenile chimps would cluster with us in most morphological matrices.
>
> >>>>>> Again, I say, show me.

Isn't that obvious by itself? Adult chimps have downright snouts. Baby
chimps, and humans, don't. The retention of this condition to adulthood is a
human autapomorphy. Adult chimps have serious canines. Baby chimps and
humans don't. Likewise an autapomorphic case of paedomorphosis. This repeats
itself when we compare adult baboons with baby baboons and chimps. You'll
certainly find more characters like these if you'll investigate.