Combined answer Re: Avian origins: new analysis confirms ancient beginnings

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

----- Original Message -----
From: "evelyn sobielski" <koreke77@yahoo.de>
Sent: Wednesday, February 06, 2008 4:15 AM

> The provisional PDF version is here:
> http://www.biomedcentral.com/content/pdf/1741-7007-6-6.pdf

Thanks a lot. I have now read the paper. The two external calibration points 
have both minimal and maximal ages, but all 18 internal ones have minimal 
ages only. I think that explains something (Marjanovic & Laurin 2007).

Having established the major flaw, I'll heap some praise on the paper. It is 
huge. To estimate molecular divergence date estimates, you need a data 
matrix, a tree, and a program. The paper uses a matrix (almost 5000 base 
pairs from the mitochondrial genome of no less than 135 species, plus at 
least one alligator and at least one caiman as outgroups), two trees -- one 
assembled from the literature, the other derived from the dataset --, and 
four programs*, resulting in eight analyses. Although they agree in putting 
lots of divergences into the Cretaceous and in putting the origin of 
Neornithes into the Early Cretaceous, the results differ quite 
drastically -- two programs put the first divergences in two sister clades 
at the same age, while the other two put one up to 40 Ma before the other.

The tree derived from the dataset contains numerous bizarrities, like 
Caprimulgiformes being broken into three parts that are scattered almost all 
over the tree, or having Falconiformes paraphyletic to Pterocleidae and 
Coliiformes + something else entirely or having ostrich and kiwi as 
sister-groups. But the dates derived from the two trees differ only by a 
very small amount, the sign of which depends on the program. The tree based 
on the literature unavoidably contains many polytomies, but apart from this, 
it is rather unremarkable.

In summary, this paper is almost how to do it. If the authors had used more 
maximum ages, and perhaps a few more calibration points in general, it would 
be ideal. They even went to the trouble of bootstrapping the dates that the 
program r8s found -- and that's not implemented in r8s, instead you have to 
make every single bootstrap matrix yourself and then run them all through 
the program. This huge amount of work may explain why our paper isn't in the 
enormous list of references even though it was published early enough.

Of course, as the authors mention, the addition of nuclear genes could only 
help, but that's another story...

* Including MultiDivTime, which is completely unusable unless you have l33t 
Unix sgillz. My respect.

----- Original Message ----- 
From: "evelyn sobielski" <koreke77@yahoo.de>
Sent: Wednesday, February 06, 2008 10:35 PM

> And nearly half a decade ago (which seems like an
> eternity in avian phylogeny and evolution studies) they
> quite confidently could state:
>
> "Molecular, morphological and fossil data all indicate
> that the early history of modern birds began in the
> Cretaceous and did not involve 'transitional
> shorebirds'."

The question is when in the Cretaceous -- just before its end, or around the 
time the Yixian Fm was deposited???

----- Original Message -----
From: "Tim Williams" <twilliams_alpha@hotmail.com>
Sent: Wednesday, February 06, 2008 11:10 PM

> > Whoops, my bad. At least six lineages. (Any other stem-anseriforms
> > we're forgetting?)
>
> Dromornithids?

And *Brontornis*, which used to be considered a phorusrhacoid, but was 
recently redescribed as an anseriform in a paper that I still haven't read. 
(Actually I'm waiting for Darren to blog about it. :-> )

Of course, two or three of these lineages of heavy-bodied big-skulled huge 
flightless birds could be the same, though the geography makes this rather 
improbable, I suppose.