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

Re: Pteromimus and pterosaur origins

Having performed many analyses, and dissected others, I can tell you're talking in philosophical generalities. I'm talking specifics. Here's how it works: If I find through PAUP* a lineage, thereafter it needs to be rechecked in MacClade. For instance: Say all taxa --but one-- have a tiny quadratojugal that ascends the quadrate. Given that, I look again at that exceptional taxon for a loose quadratojugal that may have drifted slightly. Sometimes I find it. Then I reconstruct it back into place. That happened in Huehuecuetzpalli.

Except... you can't do such a thing without seeing the fossil. I have the paper; the resolution of the photo is rather pathetic as usual.

You just have to test and experiment and get the big picture before you can be sure of the validity of the smaller (14 taxa) inclusion set. Virtually all prior attempts, from Gauthier 1986 to the present, have assumed wrongly and, as a consequence, have prolonged the error. It's easily resolved via testing. I say virtually, because Benton 1985 did it right, even with suprageneric taxa, but then he never repeated the experiment.

As I have previously explained to you onlist at length, Benton 1985 does _not_ describe a cladistic analysis. That's because Benton didn't do one. If you read the paper attentively, you'll see it's just a tree with lists of autapomorphies for each clade, but no attempt whatsoever of finding out if that tree is the most parsimonious one for the data. There is no data matrix; the states of most characters was not checked in most taxa at all.

This was a pretty common habit in the mid-late 1980s and even early 90s, when the cladistics revolution had got far enough that people understood they had to use synapomorphies and pretty angular stick drawings of trees to argue for their phylogenetic hypotheses, but when most of them didn't have halfway capable computers yet.

Again: there is no experiment in Benton 1985. Benton simply presented an idea and some evidence for it, but he didn't test the idea. Such a paper couldn't get published today.

I'd slow down there for a moment - how do we make such judgements about how "evolution works"?

We make such judgments because we know evolution works step by step. Kids, for the most part, look like their parents and great- granparents. One part evolves, blends, lenghthens, shortens, then another then another. Sometimes two or three at a time, but that is within a suite of hundreds to thousands of characters, depending on your skills and patience. Even when one taxon is half the size of the predecessor most of the characters remain similar in proportion.

But we haven't _got_ great-grandparents, parents or kids!!! We've got a _tiny_ sample of a big hulky tree!

If you were right, we wouldn't need cladistics in the first place. We could simply do stratophenetics. No more agonizing over correlated characters or phylogenetic signal...

Done well you can produce 100 manual characters rather easily, for instance.

What do you mean?

No prior study has yet given us the
comprehensive overview of the Amniota using species-based taxa.

Stop being impatient, I'm working on it. OK, I use genera for the most part, because I'm too lazy to enter every single species of *Edaphosaurus* or *Ophiacodon* or burrow through the chaos that the North American species of *Diadectes* is or are, but still two of the unknown number of species of *Casea* are in as separate OTUs, for example, because Maddin et al. (2008) found them in different places.

If we knew the patterns already, there wouldn't be a need for a
cladistic analysis.

Yes, if you did. But we don't. That's why we us CA.

This "we don't" is our point: so far, and below, you act as if we did know!

The transitions above indicate is that a given ancestor generated some descendents with a short digit V, and others with a long digit V.

No. The point is you're asking a given ancestor with a short to vestigial to absent pedal digit V to give rise to one with an elongated digit V -- when you know on the next branch, there is a taxon with a long digit V with the same characters -- attached to an torso, skull and extremities with a suite of other pterosaur-ish characters. It's parsimony, pure and simple.

Parsimony applies to the whole dataset at once, not to a selected number of characters in a selected number of taxa.

That's hardly shocking, and quite plausible. More to the point, when reversals, mosaics, and other bits of messiness show up in a phylogeny it is not a sign that the answer is wrong - it just reflects the rather stochastic patterns that are biology.

Again, Mike, you're talking philosophical generalities. Get specific, test one against the other, and we'll find answers. the reversals, etc. will reveal themselves.

Yes, except you act as if he could just throw an analysis together on an afternoon...