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

Re: BCF in perspective (LONG!)

And here's my reply to the second posting by John R. Hutchinson,
on 95-12-03 19:33:53 EST. As before, his comments are in <<double
angle brackets>>:

<<G.O. writes:>>

<<>>The main objection I have to the BADD phylogeny is that it
tells us nothing about the critters _on the branches_ (not at the
terminal points) of the phylogeny.>>

<<Of course any cladogram only shows real taxa on the terminal
points; all others are purely hypothetical. Just filling in the

Yes, but it is in those "gaps" that everything of phylogenetic
significance--the actual evolution itself--takes place! You
should not just discard it as "purely hypothetical." Take a look
at what the phylogeny you have worked out tells you about the
nature of the animals in the gaps, see whether it makes sense.

<<Formulating hypothetical phylogenies has a few basic steps
AFAIK; for example (probably the best way):
1) Collect data from specimens (preferrably hands-on, otherwise
using published data from others). The more characters used, the
more accurate the cladogram will be. Good character choice is
absolutely imperative here; you must use any relevant characters,
not only include the characters that seem to support a
preconceived notion of the phylogeny.>>

"The more characters used, the more accurate the cladogram will
be." How do you _know_ this? If your analytic technique is
flawed, then no matter how many characters you use, your
cladogram will be inaccurate. Also, what is a "good character
choice"? What, for that matter, is a "character"? Does the
character "tail reduced" count as one character, or as 54
characters, one for each lost caudal vertebra? Or as some
intermediate number? Here is a lump on a femur in animal A, and
here is a somewhat differently shaped lump in the same place on
the femur of animal B. Is this the same character? Or did the
lump appear twice independently? How can you tell? By "analyzing"
a raft of other equally questionable characters? By "winning the
synapomorphy war"?

In BADD phylogeny, "forelimb length 50% or less than hindlimb
length" is a frequently cited character. What is magic about 50%?
Why not 54% or 62%? This character is often identified as a
synapomorphy for Theropoda. But in BCF, this character is
worthless as a synapomorphy, because I assert that it arose
independently and convergently dozens of times in dozens of
lineages from Lagosuchia to Phorusrhacidae (or whatever).
Likewise "serrated teeth." I don't bother scoring "reduced
forelimbs" and "serrated teeth"; I am not in favor of cluttering
up my analysis with easily reversed or easily regenerated

<<2) Form a data matrix, analyze it to form a cladogram (the most
parsimonious is best), a hypothetical evolutionary history of the
taxainvolved (including at least one outgroup, the closest
related taxon not included in the clade under consideration, e.g.

Let me give you a different analogy. In chess, grandmaster
players do not look at 150,000,000 moves to find the best move.
They just look at a few dozen moves--because they know that the
best move _will_ be among those few dozen. They prove their
assertion by winning the game. Chess computers, however, have to
be programmed to examine all 150,000,000 moves by brute force,
because there is as yet no known algorithm for what the
grandmaster does when he looks at the board; and the fact that
computers are still losing to grandmasters shows that brute force
and brute speed are not always the correct procedure. The best
games played by the best players are probably very close to the
best possible chess games that can be played, and no amount of
brute force computation will make them significantly better.

Likewise cladistics. You can enter a character matrix into a
computer, grind it through PAUP and get thousands of "most
parsimonious" cladograms, grind them up statistically some more
until you finally get some kind of result. But an experienced
naturalist doesn't have to do this. I'm not saying that _I'm_
doing it right; all I'm saying is that there are other methods
for doing phylogeny than grinding up character matrices and
spitting out trees, and that they have equal claims on the
validity of their results.

<<3) Compare the cladogram to the fossil record; see if your
proposed phylogeny matches the stratigraphic record of the taxa

Ha! When is the last time you heard a cladist say, "Well, my
cladogram says thus and so, but the stratigraphy says otherwise;
so its back to the drawing board for me!" The cladogram is NEVER
wrong--it's the stratigraphy, or the biogeography, or the fossil
record. Just because all the most birdlike theropods occur in the
fossil record after _Archaeopteryx_ doesn't mean such forms
weren't ancestral to _Archaeopteryx_ and casts no doubt on THE
CLADOGRAM. We just haven't FOUND the right forms yet. Well, two
can play at that game.

<<4) If desired, formulate macroevolutionary hypotheses to
explain trends - i.e. the origin of flight in that clade. Not at
all necessary, and maybe not even good falsifiable science, but
still fun.>>

A better fossil record would falsify lots of the more outre
scenarios for the origin of flight. I would say a knowledge of
physics and aerodynamics would falsify a few more.

<<The problem that I have with BCF, besides the fact that I have
not seen the phylogeny, is that:
1) Archosaurs did not start out small, unless you're redefining
the definition of Archosauria to include some small diapsids as
basal archosaurs.>>

The earliest record of an animal that has been referred to
Archosauria is of _Mesenosaurus romeri_ from the Late Permian of
Russia. It is known from several skeletons (none has been
described; only the skull) and they are small--about 30-45 cm
including tail. Now, it may be disputed that this is an archosaur
(many paleontologists do), but it has certain archosaur
characters (such as a small, slitlike antorbital fenestra in the
right place), and I'll buy it as an archosaur close to the
ancestry of the group.

Lots of diapsids close to basal Archosauria were arboreal (such
as the gliding icarosaurs). There is thus even some basis for
asserting that the ancestral archosaur itself was an arboreal
animal, and that the lineage from ancestral archosaur to robin
_never_ left the trees (or other high places).

The most primitive animals generally included in Archosauria
(stem-group definition; Archosauromorpha if you use the crown-
group definition of Archosauria) are the proterosuchids, which
were about two or three times the size of _Mesenosaurus_. This is
still pretty small, but not exactly my idea of "small" in this
context. On the other hand, they were already derived piscivores,
with specialized jaws, that were not ancestral to any other known
archosaurs (except maybe hupehsuchians). _Euparkeria_, another
near-basal archosaur, was about twice the size of _Mesenosaurus_,
smaller than most proterosuchians: still not the kind of "small"
I'm looking for but getting closer. We're not talking giant
animals here by any stretch.

Now, _Longisquama_. That's "small."

<<2) The robin is not, not, not a good representative for the
birds in general. About 140 million years of evolution separate
the modern robin and the earliest conclusively known birds.
Ratites and paleognaths are the best guesses for the earliest
non-Archaeopterygian birds; passerines such as the robin came

I'm not interested in ratites and paleognaths here. Robins
represent a bird order that is almost (if not more than) 50% of
all known birds. This is not typical?? All known Mesozoic birds
were small, according to my concept of "small." That's all I need
in this context. Besides, the lineage that led to the robin is
_identical_ with the Mesozoic portion of the lineage that led to
practically every other Cenozoic bird, because the robin branch
didn't diverge until well after the end of the Mesozoic.

<<3) The Archosaurs did not start out with large forelimbs; their
hindlimbs and forelimbs were of equivalent length (unless you're
using pterosaurs as your outgroup).>>

That's _large_ when you're looking at bipedal theropods.

<<The earliest conclusively known birds (i.e. Archaeopteryx)
had huge forelimbs compared to non-dinosaurian archosaurs.>>

Indeed! And it's easier to get to the huge forelimbs of
_Archaeopteryx_ from the large forelimbs of the early archosaurs
than from the small forelimbs of bipedal theropods.

<<4) Parsimony is used after cladograms are created, not in the
formulation of phylogenetic hypotheses without cladograms. Once
the most parsimonious cladogram is found, we have our
_hypothetical_ phylogeny (which is more than likely slightly
wrong, but if we used good data and character choice, is probably
better than anything else we have).>>

You use parsimony your way, I'll use it my way. I don't think
there is any more validity to using parsimony to sift through a
bunch of cladograms _after_ they are generated than there is in
using parsimony to eliminate a bunch of long-shot possibilities
_before_ the cladograms are created.

<<If BCF is to overturn BADD, it must play by the same rules as
Gauthier and others did when they overturned SWDKWTHBATLLRTM
(sorry, we don't know what the heck birds are, they look like
reptiles to me). It simply doesn't have a snowball's chance
otherwise. Unfair? No. Cladistics is not inherently biased
against any phylogenetic hypothesis that uses real data; it is an
objective method to resolve phylogenies, and according to the
majority of professional paleontologists, it is the best method
that we have for resolving the phylogenetic relationships of
extinct organisms.>>

No--if I play by the same rules, then I can expect to get much
the same results, and we have made no progress. The objectivity
of cladistics is compromised by the unavoidable subjectivity of
the data-gathering process: "Is this a character? How should I
score it?" And so on. How many times have cladists discarded out
of hand, behind the scenes, an "obviously incorrect" cladogram
generated by their methods?

Now let's get down to some nitty-gritty here. Get this: I _don't
really care_ whether the archosaur phylogeny that my methods and
analysis have generated overturn BADD or not. As far as I'm
concerned, BADD is history, and it's only a matter of time before
paleo comes around (if it ever does). I have arrived at a
phylogeny for archosaurs (excluding crocs and Cenozoic birds,
which I know very little about) that as far as I'm concerned fits
all the known facts and does a better and more satisying job than
BADD of explaining how and why the archosaurs evolved and
diversified the way they did. Anybody who would like to read
about my methods and see the phylogeny in detail is welcome to do
so by ordering _Mesozoic Meanderings_ #2 (third printing).

<<I'll add my mandatory disclaimer: This entire message is all
IMHO. I'm no expert in cladistics (or a BADD fanatic), I just am
trying to think clearly on this important and interesting topic.
I probably made some errors in the above statements, but someone
will scold me if they notice and we'll all be the better for it.
All right, I'll be quiet now. Sorry about the length and all. And
of course, nothing personal G.O.  8-) >>

I'm always interested in hearing from anyone and everyone who
disagrees with my phylogeny, because every so often someone
points out a flaw, a weakness, or an error of some kind, either
in my data or in my method, and also because it hones my ability
to anticipate and overcome objections.