Re: Herbivory widespread in Coelurosauria

[Lindsay Zanno <lzanno@fieldmuseum.org>]

Hi Mickey,

In answer to your questions:

>Yet there are taxa like Protopteryx which only have seven or more correlated 
>herbivorous traits >because they're coded multiple times for the same thing 
>(conical premaxillary teeth and posterior >dentary toothless in this case), 
>while another taxon might actually have seven or more independent t?>traits.  
>So those cases don't really seem comparable to me.  That was my concern there.>

No, actually, non-independent characters were *not* counted more than
once when counting the number of CHTs.  You are probably confusing
"actual" with "reconstructed" CHTS.

>I suppose they're consistant with herbivory, but their correlation with 
>herbivory is more doubtful since >even a hypercarnivore doesn't lose them.  I 
>would be interested in seeing a statistical study of living >birds with known 
>diets that indicates dentary curvature, cervical number, ischial length, etc. 
>is >actually correlated with herbivory.

Our results only apply to the taxa we included in the study.  They are
not meant to be extrapolated outside the clade in question.  For
example, these traits may or may not correlate with herbivory in all
dinosaurs.  Certainly, the question in birds is wide-open.  I'd like
to see that study too :)

> You do state "Pairwise comparisons cannot process missing data; therefore, 
> nonapplicable >characters are considered absent (state 0)" in the materials 
> and methods section, but you couldn't >have run pairwise comparisons between 
> all taxa for all characters because almost every taxon does >have some 
> missing data in the matrix, often when the elements in question are 
> unpreserved.  So you >presumably didn't do a pairwise comparison between 
> Pelecanimimus and Anserimimus with regard to >dental serration, for 
> instance.  I'm no statistitian so I might be misunderstanding things, but I 
> would >think that means you could have also not done a comparison between 
> Pelecanimimus and >Struthiomimus since the latter actually has an unknown 
> state too.

Yes, I think you are misunderstanding.  The trait correlations in
pairwise are not run between taxa but between character states for all
taxa that can be coded.  The reason to code N/As as state absent is so
that they can be included in the analysis.  This actually makes it
more difficult to pick up a signal since you are essentially taking a
N/A and making it work against your data, but we felt it would add
additional rigor to our correlations.

>As I said, I don't doubt the general conclusion that most maniraptoriforms 
>were more herbivorous than >most other theropods, it's just the details I have 
>issue with.

Mickey, anyone can quibble over the details of any scientific study.
The job of any scientist worth their salt is not to be *right* but to
be *repeatable*, which is why an explicit discussion of the paper's
methodology is provided in the first place.

Finally, from one stickler for details to another: I've seen many
interesting posts by you over the years.  If you want to advance the
field you have to be brave enough to put your contributions out there,
not just your criticisms.

-- 
Lindsay E. Zanno, PhD
Department of Geology
The Field Museum
1400 S. Lake Shore Dr.
Chicago, IL 60605-2496
Ph. (312) 665-7665
lzanno@fieldmuseum.org