Re: New paper on Neoaves

[Tim Williams <twilliams_alpha@hotmail.com>]

David Marjanovic wrote:

> I favor good molecular analyses over bad morphological analyses. [snip] I 
> also favor good morphological analyses over bad molecular analyses

Well said, David.  I would agree with that.  I'd add that many 
well-supported molecular phylogenies are not as "good" as they might first 
appear.  I hate to sound like a Luddite, but some people tend to regard 
molecular analyses as automatically superior to morphology analyses, unaware 
that they are trading one set of problems for another.  At least with 
morphology analyses we have some idea of the characters that are actually 
being analyzed, so the problems are easier to spot.

> Sequencing (from organism to GenBank submission) is a matter of a week or 
> two, you download the other sequences from GenBank, and then you let the 
> computer run for a couple of weeks. That's all.

Hey, come work in my lab!  You do it a lot faster than I do.  :-)

Mickey Mortimer wrote:

> And echinoderms are crazy things that could have reversed characters.

Yes, but not even the sequence-based analyses always agree on the precise 
topology of deuterostomes.

> Lots of mammalian groups which molecular evidence reject were also 
> supported by good-looking lists of synapomorphies.

In some cases, they still are.  I don't think molecular phylogenies have 
'solved' the problem of placental phylogeny, in spite of those 
well-supported sequence-based clades.  We tend to use molecular-based 
analyses to make up for the deficiencies of morphology-based analyses 
(incomplete fossil record, convergence, etc).  But I tend to believe that 
these deficiencies carry over into molecular phylogenies too, though they're 
expressed in different ways.

With placentals, the problem of an incomplete fossil record appears to be 
compounded by a rapid radiation of "orders" at the beginning of the 
Cenozoic.  (This may have happened with birds too.)  There are very few 
mammal "orders" for which we can document their early evolution (whales are 
an exception; but they weren't part of this radiation).  But if placentals 
did undergo a phase rapid evolution, with the divergence of many lineages in 
a relatively short period of time, we may be asking too much of molecules to 
capture the exact order in which these lineages diverged.

> But I do favor molecular over morphological, so I may be biased.

I lean in the other direction - and I work with molecules.

> > The amount of homoplasy caused by biochemical constraints may be vastly 
> > underestimated.
>
> I think we're basically talking about the same thing.  I just think doing 
> this will take a LOT of time.

There are people working on it, but it's difficult research.  But I believe 
it's worth the effort.

> Well, do you find that many of the new mammal clade names to be useless?

I know some paleontological/morphological corroboration is emerging for some 
of these clades (like Afrotheria, at least in part), but I'm not sold on 
most of these clades.  I'd like to have more evidence that biases within the 
molecular dataset are not responsible for these topologies.

> Does NOBODY care how xenungulates fit in?

I do!  Poor _Carodnia_....  :-(

Cheers

Tim