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

Re: Brusatte et al. 2008 question

I'm trying to understand the difference between a cladistic character
list and a database the tries to quantify overall similarity.

That's very simple. Overall similarity -- what phenetics measures -- includes autapomorphies and correlated characters. It being a phenetic rather than a phylogenetic data matrix, Brusatte et al. made no attempt to get rid of such characters; to the contrary, they added as many such characters as possible in order to represent the entire morphospace of Triassic archosaur skeletons.

Autapomorphies of single terminal taxa are useless for phylogenetics, and correlated characters distort phylogenetic trees.

Quantifying overall similarity sounds like parsimony. But evidently
there's a subtle difference. The subtlety is escaping me. Help!

It is not subtle, it is big and fundamental. Phenetics is to quantify _overall_ similarity. Phylogenetics (of which maximum parsimony is the simplest implementation) is to quantify shared _derived_ similarity.

Good that you asked. For years, as your question shows, you have been doing parsimony analyses without even knowing why. Let that sink in for a while.

...

And with regard to David M's earlier comment to my earlier note:

   DP:  3. Scleromochlus, with its teenie-tiny hands and spike like
   digit V is the sister taxon to pterosaurs. Sister taxa just outside of
    this clade don't have a long manual digit IV or a long pedal digit V
    either. In fact, those are hard to find anywhere here [Proterosuchus
    is not included]. Was there a miracle in the appearance of wings
    and toes? Maybe Hone and Benton can help here.

DM: No miracle here. *Scleromochlus* underwent reductions, and the
pterosaurs underwent expansions instead. The MRCA of *S.* +
Pterosauria did _not_ look like *Scleromochlus*.

As mentioned above, no mentioned outgroups (MRCAs) have a long manual
digit IV either, but reduced or vestigial. Same with pedal digit V.
Besides, better sister taxa are known with all the necessary
characters. I guess they mated outside their clade to get the miracle
baby with long finger four and a new toe five. So no good MRCA
candidate for many branches back. That's a red flag IMHO

Why? It's just a reversal or two. To show that the "better sister taxa" are indeed better, you must show that whatever character states they share with pterosaurs _outnumber_ those the latter share with *Scleromochlus*. That's what a parsimony analysis does.

Of course, you claim to have done just that in a recent analysis, but that's still unpublished... and given the fact that till now you didn't know how to make sure a matrix is actually fit for parsimony analysis, I doubt the results anyway.

And with regard to Scleromochlus, all you have to do is reconstruct
the poorly preserved quadrate leaning forward, rather than the
assumptive back lean and it becomes a flat-headed bipedal croc-omorph
with a particular affinity to another flathead, Gracilisuchus. Doing
that also reduces that hyper-elongated retroarticular process
proposed by Benton 1999.

Two characters for you. But if you change these two cells in a data matrix, is that enough to move everything around in the resulting trees?

And how are its ankles compatible with it being a crocodylomorph, among other things?