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

Re: Reevolving bones?

In a message dated 97-05-05 10:45:39 EDT, znc14@ttacs1.ttu.edu (Jonathan R.
Wagner) writes:

<< If you read the _Alxasaurus_ paper, you will note that the first
 metatarsal of therizinosaurs is *not* "exactly" like that of other four-toed
 dinosaurs, and actually (by some miracle of evolution) records its

No, it's indeed not >exactly< like that of other four-toed dinosaurs. But it
derives from a four-toed dinosaur foot >much< more easily than from a
three-toed theropod foot. For one thing, no reversals are needed. In deriving
a segnosaur foot from a theropod foot you need something like half a dozen
simultaneous reversals to make it happen. Rejecting theropod ancestry in
favor of prosauropod ancestry is more parsimonious.

The theropod foot is remarkably conservative, being present virtually
unchanged, except for the fusion of certain elements, in all groups from
ceratosaurians to birds. >Except< in segnosaurs, where for some reason it
"reversed" to resemble a prosauropod foot.

But feet aren't the only non-theropod aspect of segnosaurs. There's the
problem of the segnosaur pelvis. It looks nothing like a theropod pelvis: the
preacetabular processes of the ilia flare way out laterally, the pubis is
opisthopubic to the point of being fused to the ischium, there is a
postacetabular lump on the ilium, and the ilia are very tall dorsally.
Although it is possible to derive such a wild pelvis from a theropod pelvis,
here again it derives more easily from the generalized dinosaurian pelvis of
prosauropod sauropodomorphs, and in several features it is suggestive of a
transitional morphology between sauropodomorphs and ornithischians. The
pelvis of _Alxasaurus_ is not well known or illustrated, and doesn't shed
much light on how it may have evolved. And in the mounted skeleton that I saw
at the CCDP exhibition, it was modeled entirely in plastic to resemble a
theropod pelvis and did >not< match the pelvic fragments of the actual type
specimen that were on display. This borders on reprehensible.

Likewise, the segnosaur carpus, as illustrated in the _Alxasaurus_ paper,
does not resemble a maniraptoran carpus in any particulars, but it >does<
resemble a prosauropod carpus in the number, shapes, and disposition of the

Combine all these observations with the prosauropod features of the skull of
_Erlikosaurus_, and you begin to suspect that segnosaurs may have a lot more
to do with prosauropods than with theropods.

<<         You really should stop getting all of your data from George ;).
 Dollo's Law is not a law in the strictest sense. The fact that ontogeny
 recapitulates phylogeny allows evolution (in some cases) to dip into the
 evolutionary history of an organism and reform structures which may be more
 useful than the ones the animals has now. Accept this now, or doom yourself
 to never understanding evolution. >>

Tracy could do worse than get all his data from George. Cladists' attitude
toward reversals is dangerously circular: cladograms show lots of reversals,
therefore they happen frequently, therefore we can expect to see them in our
cladograms, and so there they are and shut up. And I find "cladistic
infallibility" repugnant and dogmatic. Reversals are a red flag in
phylogenetic analysis--a signal that something may have gone wrong in the
analysis. >Not< that something >has< gone wrong, just that something peculiar
is going on that warrants further investigation. Although I agree that
phylogenetic analysis must precede evolutionary hypothesis, I also submit
that when no reasonable evolutionary hypotheses can accommodate the
phylogenetic analysis, then it's time to admit that the analysis itself was
incorrect, and to try to figure out what went wrong.

As far as cladistics goes, I was discussing this problem a few weeks ago with
Mickey Rowe, and I suggested the following experiment:

Using a computer program that simulates evolution, generate twenty different
phylogenies. Hand the output from these phylogenies, in the form of a
collection of pictures of hypothetical creatures with no other information
about them than that they were generated by an evolution-simulating program,
double-blind to twenty cladists. Let them analyze the creatures using
whatever methods they like to reconstruct their phylogenies. Then compare the
400 reconstructed phylogenies with the actual control phylogenies. The
question is, how many will match exactly? And--how close do they come when
they don't match? You can then repeat this experiment with non-cladists using
more homespun methods. Which group will do better? Until something like this
is actually done--testing cladistic and non-cladistic methods in a controlled
situation--I assert that cladistic methods have no claim to being better at
phylogeny than any other methods.

If the cladists were to score significantly better than non-cladists in the
above experiment, I'd be happy to adopt the cladistic methodology for
deriving my phylogenies, and I'd have much more confidence in the analyses of