[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]
Re: Steadman's review of Mesozoic Birds
From: "David Marjanovic" <email@example.com>
To: "DML" <firstname.lastname@example.org>
Subject: Re: Steadman's review of Mesozoic Birds
Date: Sat, 31 Jan 2004 13:58:54 +0100
> Cladistics is a tool--it is a method for generating hypotheses--it is
> an inerrant formula for producing flawless phylogenies.
Oh, I agree. But it must be noted that, presently, cladistics is the _only
existing method_ to generate phylogenetic _hypotheses_ that are not based
an a priori assumed scenario.
> To reject a
> hypothesis for the singular reaosn that it fails to produce a cladogram
> as flawed as circular logic.
The BAND doesn't just produce no cladogram. They don't produce any
phylogenetic hypothesis*. Ask them the simple question "what is the closest
known relative of birds?", and the answer will be "ABSRD"; ask them again,
and they'll say "well, you know, uh... the ancestor [note the shift of
topic!] of birds must have been pretty similar to *Longisquama*,
*Megalancosaurus* and/or *Cosesaurus*... who knows...". They offer, so to
say, a negative phylogenetic hypothesis, a hypothesis on what is _not_ the
closest known relative of birds. But they never offer a phylogenetic
* After all, you're correct, it doesn't matter how they arrived at it,
whether via PAUP*, NONA or a crystal ball -- as long as the hypothesis
itself exists and is testable.
If they would say, for example, "*Megalancosaurus* is the closest known
relative of birds, because it shares with birds the character states A, B,
and D, which dinosaurs lack", then we could say "ah, but you've overlooked
that dinosaurs, except a few specialized ones, do in fact have A; you've
misinterpreted *Megalancosaurus*, it lacks B; unlike derived birds and
*Megalancosaurus*, basal birds lack C; so D is most parsimoniously
interpreted as convergence". (To which the BAND could, in turn, answer "OK,
you've caught us on A, we'll drop it; we have correctly interpreted *M.* on
B, as this photo shows; your phylogeny of birds is wrong, so the "basal"
birds you cite aren't all that basal and can have lost C secondarily;
therefore D is most parsimoniously interpreted as a synapomorphy of *M.* +
birds". And so on, I need not explain how science works.) But no. They
come out of their snail shell. They don't tell us precisely what, if
anything, they think. Therefore it can't be falsified, and therefore it
I am well aware of this, but this criticism extends beyond just the question
of bird origins.
> Need we remind ourselves that the very
> hypothesis we are by and large united in defense of, the theropod origin
> birds, did not spring forth from cladistics, and by your own criterion
> be rendered unscientific?
:-) The hypothesis is not unscientific anymore. But before the advent of
cladistics -- the method to make phylogenetic hypotheses and to test how
well they fit the totality of the data -- it was. "Unscientific" doesn't
mean "wrong". It just means "if it's wrong, we can't find that out with
Nor was it unscientific when it was presented.
> To equate cladistics to the one true way of
> representing phylogeny, the only scientific method for the determination
> evolutionary relationships, is to effectively elevate this tool to blind
> dogma, precisely as its critics have argued.
It cannot be a dogma -- because different data sets will produce different
cladograms by the same method. What can be a dogma is the reliance on
parsimony. I hear rumors that maximum likelihood* for morphology is being
developed, and will be included in PAUP* 4.0 (currently the newest version
is 4.0b10). I can't imagine how that could work, but I'm looking forward to
it. :-) Until then, I'll stick to Occam's Razor -- of several equally good
explanations for the same phenomenon, the simplest one must be preferred,
and the simplest one is the one which makes the fewest ad hoc assumptions;
each character change is an ad hoc assumption, and the more homoplasy you
assume, the more character changes you have to assume in your tree.
* This means that different mutations of a nucleotide or protein sequence
are given different probabilities, according to what is observed -- e. g. a
C will mutate most easily to a T, a neutral amino acid will more easily
mutate to another neutral aa than to an acidic or basic one...
> I find it of lasting poignance that John Ostrom vigorously
> criticized cladistics.
I'm not sure if he has understood cladistics... :-)
Occam's Razor is one thing, but naive adherence to parsimony is another. It
is a central tenet of cladistic analysis that convergence, paralellisms, and
reversals are very rare and yet biologists working with extant and extinct
forms have usually argued just the opposite. These things are some of the
most pervasive trends in vertebrate evolution (and in invertebrate evolution
for that matter). Indeed within Aves we see astonishing convergences, time
and again. To claim for instance that anything which calls for convergence
and so on is ad hoc fails to distinguish between ad hoc hypotheses, and
While we almost all agree that shared derived characters ought to be used,
this idea that cladistic analysis in the solitary scientific method of
phylogenetic reconstruction, with a monopoloy on accuracy and objectivity,
is simply farcical. As long as cladists continue to present such a
groundless picture of their methodology, we will see equally shrill and
meaningless criticism of cladistics.
Check out the new MSN 9 Dial-up ? fast & reliable Internet access with prime