[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]

Re: where have all the etc.

Okay, let me repeat what you are saying in my own words as I 
understand it: There is a 'problem' in that I have insufficient
characters in my analysis. But in this partcular rabbit hole leading to
Wonderland the problem is that I have complete resolution and all sister
taxa resemble each other. My friends, that's no problem. That's a
solution.

How do you know it's a solution rather than a random quirk? Test that. Add more characters and try.

(Just a wild shot... are you sure you haven't changed your settings to neighbor-joining instead of parsimony? Neighbor-joining always finds only one tree, no matter how many of equally shortest length there are [and of course almost no matter how parsimonious they are]. I mean... sorry for implying you overlooked such a basic thing, but it's strange how all your analyses seem to find one tree.)

You'll note that each of the analyses uses some or many 
suprageneric taxa, and that each one has 13,  14, ? 11, 20, 9, ? taxa in
total. Now granted they all focused on the archosauriformes, but none
include any Choristodera, Doswellia, Cerritosuchus, Silesaurus,
Lotosaurus, Pseudhesperosuchus, Tropidosuchus, or various Youngina
[...]

That's true, and it's good that you include them all. But you have too few characters to sort them with any reasonable support. Your average Bremer support is probably 1.2.

DM wrote: What, you have only 150 or so characters in your matrix, and
in spite of this you are so certain of your conclusions?!?

When you have a single tree and everything makes sense, then the
end of the road has been reached.

Nope. You must then test whether you've _really_ reached the end of the road. If you have so few characters, the addition of just one more character is likely to change the topology. If you _have_ reached the end of the road, then the addition of 20 or even 50 characters (...or typos in the matrix!...) won't change the tree, except for 2 or 3 details and 20 or 30 precise bootstrap values. _This_ is the "enough is enough" phenomenon.

More characters mean more signal and more noise; but while the signal adds up, the noise averages out. More characters = better signal/noise ratio; high enough signal/noise ratio = enough characters.

"Everything makes sense" means that the tree is robust. Given your low number of characters, yours cannot be robust.

The ref is a JVP abstract... I'm sure several papers have been published on this meanwhile, though. I suggest googling for John J. [?] Wiens. I don't have much literature on this myself, though, and at the moment I unfortunately don't have time for anything, writing this e-mail included. I have to write up what I did in the lab for 3 weeks.

I'm sorry to hear that the same cannot
be said of your theropod work. It must be frustrating.

What _is_ frustrating is that I don't find more characters, means, that I can't hop around the world to inspect the fossils, respectively that most of them aren't described in detail with good-quality photos. I'm not frustrated that I don't find fewer MPTs with such a ridiculous ratio of characters per taxon and so many question marks in most taxa. Such a matrix is simply incapable of producing a single well-supported MPT.

Referees have brought this up, suggesting I use 20,000 taxa
and as many characters,

For 20,000 characters you'd obviously need a lot more than 20,000 characters!!!

so this ratio-thing has been on my mind. As you
can well imagine, the prospect of so many taxa is daunting and I am not
that ambitious.


(Are there even 20,000 fossil vertebrate "species" known?)

David, you mentioned you would need a week or so to get your 10
characters that will link the flat-headed, dorsal naris archosauriformes
with the tall-headed hook-nosed ones. No problem. I can wait.

I'll need that week before I can _start looking for_ such characters. I don't have most of the primary literature on that part of the tree.

Should I send you Müller's paper & supp. inf., or do you have those? That would be a matter of a minute or two for me.

DM wrote: I'll be in Paris to write my ~ Master's
Thesis under the supervision of
Michel Laurin. He could well have something.

He does. My manuscript, cladogram and I think a matrix. 

Oh!

Good luck on your thesis!

Thanks!