Re: Proto-penguins lived with dinosaurs

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

> > Fordyce and friends took the [...] 62 Ma of *Waimanu manneringi*
> > (determined by traditional geological methods -- it's in a Danian
> > layer) as the date of the divergence between Sphenisciformes and...
> > whatever their molecular tree gives as the closest living relative.
>
> Storks (bizarrely), followed by loons and procellariiforms:

Bizarre indeed, but when they take the stork out of their analysis nothing 
changes, they say.

The second calibration point is *Vegavis* at 66 Ma. That's fine (give or 
take a million years maybe). However, as they say in the supplementary 
information but not the text, they used a third calibration point: the 
divergence between cassowaries and emus at 30 to 35 Ma ago. This date may 
well be correct, but unfortunately it comes out of quite thin air. The 
_entire_ fossil evidence for this age is the fact that the oldest known emu 
is _twenty_-five Ma old.

Additionally, they get the ostrich as the basalmost ratite. If this is 
wrong, it may have pushed a lot of other divergences down, especially the 
basal divergence of Neornithes. The paper that announced the complete 
mitochondrial genomes of two moas and the partial one of an aepyornithid 
found a different topology, with ostriches less basal than rheas, maybe 
because the moa and elephant-bird sequences were included in that analysis. 
They are missing this time.

In addition to the (small) standard deviations, 95 % confidence intervals 
were calculated "but are not shown".

The program for estimating divergence dates does Bayesian analyses. This 
means you can choose a so-called prior; if possible, the program will bias 
its results toward the prior. Now, "the mean of the prior distribution for 
the time separating the ingroup root and the present (rt_tm)" is 84 Ma ago, 
and "the standard deviation of rttm" is 42 Ma ago. It is not explained where 
these numbers come from.

More importantly, however, "the length of time between the root and the 
present" is set a priori as 500 Ma. How does this work? The ingroup are 
Neornithes, and the outgroup is composed of 6 "reptiles", including 2 
turtles -- so why not, like, 300 Ma???

Interestingly -- for a molecular tree -- the two crocodylians clade with the 
birds, while the turtles are in the basal trichotomy.

The internal branch lengths, especially in Neoaves minus Passeriformes, are 
_very_ short. This may mean that the used characters (complete mitochondrial 
genomes minus a few genes that were "excluded due to missing data") are not 
well suited for examining the relationships of the used taxa. Indeed some 
bootstrap values are rather low. Ratitae (with the ostrich as the 
sister-group to the rest, see above) has only 54, and Ratitae without the 
ostrich has 64. Neognathae, too, has only 67. The other bootstrap values 
below 78 are all in the following clade:

--92--85--falcon
  |   `--buzzard
  `--49--91--oystercatcher
      |   `--turnstone
      `--66--49--loon
          |   `--54--stork
          |      `--penguin
          `--99--albatross
             `--petrel

...which is arguably the most interesting one for the purposes of this 
paper. Let's put it this way... it is very honest to publish a bootstrap 
value of 49. Few people do that.