[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]
RE: Several New Papers
evelyn sobielski wrote-
> But the 'caprimulgiformes paraphyletic to flamingos'
> arrangement wasn't even
> supported by bootstrap values of over 50%. No
> molecular worker would take
> such a result seriously, and if you thought it was
> something Fain and Houde
> advocated, you're surely mistaken.
No, I wouldn't say that. (Still, if that counts, this
http://www.bioone.org/perlserv/?request=get-abstract&doi=10.1650%2F0010-5422(2002)104%5B0725%3ASAHBOW%5D2.0.CO%3B2
should probably never have been published as it was...
and I have seen far too many of these. If node support
is bad, that's that. But one could just let it rest
there and not try to fit the evidence to the
inference.)
Try to fit the evidence to the inference? How did Fain and Houde do this?
> So you can't use
> very poorly supported
> ingroup nodes of Metaves to argue against the
> reality of Metaves itself,
> which had an 85% bootstrap (not bad, but not great)
> in Fain and Houde's
> analysis.
Tempting but insufficient for me, because it is full
of weakly-supported and otherwise spurious nodes and
as a whole is not supported well by other than
molecular data:
How does being full of weakly supported nodes make the inclusive node more
suspect? It's equivalent to arguing that since no arrangement within
Paraves (dromaeosaurids + troodontids?; troodontids + archaeopterygids?;
dromaeosaurids + archaeopterygids?; troodontids + birds?; dromaeosaurids +
birds?; archaeopterygids + birds?) is particularily strongly supported, that
Paraves itself is questionable.
As for being purely based on molecules, it's not like metavian taxa have
well supported morphology-based relationships to any coronavians.
4 characters, each one of which is *known* to occur as
homoplasy, and as such could have switched state once
or (unlikely) repeatedly since the Cenomanian,
approximately (as delimited by Galloanserae and
Syrrhaptes). Hm.
Shades of Kurochkin... ;)
The fact the characters are homoplasious doesn't suggest they are invalid.
Any character COULD have switched states any number of times.
As an example, the best supported metavian+coronavian clade in Mayr and
Clarke (2003) (metavian Eurypygidae + coronavian Ardeidae) is diagnosed by 5
unambiguous characters. Processus zygomaticus absent or vestigial is found
in 17 other taxa in their matrix. 23 or more presacral vertebrae is found
in 9 other taxa. Coracoid without foramen nervi supracoracoidei is found in
19 other taxa. 13?14 vertebrae ankylozed in synsacrum is found in 16
others. Oil gland minutely feathered or naked is found in 14 others. So if
your criterion is how often a character otherwise features homoplasy, having
only a single convergence or reversal (as in the metavian/coronavian
molecular indels) isn't really so bad, is it?
> >This being science after all, I'll just put it like
> >this: give me a good solid attempt to *falsify* the
> >Metaves-Coronaves dichotomy, and if that *fails*,
> I'll
> >think about it.
>
> Wasn't that what Ericson et al.'s analysis was?
I'd rather call it a good review/meta-analysis. They
didn't set out to refute anything, but rather to
assemble that what was lying around and see where they
arrived at. Which is also entirely valid.
Biogeographical data, the incompleteness of the fossil
record nonwithstanding, should be moe used to put
hypotheses to the test methinks.
In a way, Ericson et al. did set out to refute Fain and Houde's hypothesis -
by including the beta fibrinogen data those authors used to originally
support Metaves/Coronaves. And Metaves/Coronaves passed. Not only was it
still supported, but its support increased. Frankly, the poor fossil record
and ability of birds to fly means that I would support molecular analysis
over biogeography for birds no matter what the latter said. What would be
interesting however would be to add Mayr and Clarke's morphological
characters to Erikson et al.'s molecular ones, then include lots of fossil
taxa.
Mickey Mortimer