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

Re: Mononykus

In a message dated 97-03-28 16:28:28 EST, znc14@ttacs1.ttu.edu (Jonathan R.
Wagner) writes:

<< The only minimally-circular, reproducable, consistant method we have
 currently at our disposal to resolve these issues is the evaluation of
 shared dervied characteristics, and the postulation phylogeny based on that
 data (collectively cladistics/phylogenetic systematics). Only after we have
 developed our hypothetical tree can we suggest whether a character is truly
 convergent or not. >>

No, only if you think that cladistics has generated the correct phylogeny can
you point out convergent characters, homologies, and so forth. The assumption
that cladistics >must< generate the correct phylogeny is, however, certainly
open to question. Consequently, so are any putatitve homologies and

Let me construct an analogy that may clarify my personal stand on cladistics.
In chess, computers can be programmed to examine billions of possible moves
as they play. But computers, with their brute-force search algorithms, are
almost always beaten by grandmasters, who all too obviously >do not< examine
billions of possibilities when they play. As one grandmaster has put it, they
only examine a few dozen lines of play--the "best" lines of play. How do they
know which are the "best" lines and which need not be considered? Through
vast experience, and by constructing a subtle, strategic game theme soon
after the opening moves. The principles by which such a theme may be
constructed are not well articulated and have thus far defied computer
programming, but they must exist, because the grandmasters keep on winning.

Brute-force computer chess has its uses, of course. The best programs will
defeat 99.9% of all players below grandmaster rank, and computer searches
have disclosed a number of unexpected fine points in endgame analysis that
ordinary humans have overlooked.

If winning a chess game is likened to arriving at a correct phylogeny, then
cladistics is like brute-force chess algorithms, which win against most
players most of the time. And grandmaster chess represents those phylogenetic
methods that go >beyond cladistics<, experience-based methods and principles
that so far remain only vaguely articulated among evolutionary biologists. It
is these principles that set off the warning bells when cladistics produces
those startling or anomalous results.

For me, the grand theme of archosaurian phylogeny is BCF: the idea that
certain small dinosaurs climbed trees and evolved into gliding and flying
animals before they gave rise to the giant, bipedal walking and running forms
with which we're most familiar. BCF removes the paleornithological objections
to the paradigm of birds as flying dinosaurs by shifting the paradigm to one
of dinosaurs as running, flightless birds. There is no way to reach this
paradigm--or its unshifted antithesis (BADD), for that matter--from
cladistics or from direct analysis of known material. Rather, it is a
unifying theme that accounts in a clean and simple way for a number of
peculiar characters possessed by certain dinosaurs, characters whose
existence would otherwise be the product of _ad hoc_ chance.

If cladistics or any other phylogenetic analysis produces a result that seems
to conflict with BCF, this sets off warning bells that force me to re-examine
the result to see where the problem may lie. If the problem is in BCF, I
modify or discard the theme; but so far the problems seem to have lain with
the analyses.

In BCF, for example, _Mononykus_ is simply a small theropod that retained a
number of flight-related features from its flying common ancestor with
avialan birds. We know this because it possesses a number of theropod
characters lost in all avialan birds and whose presence--as acknowledged by
Novas in his recent paper on alvarezsaurids--must be considered reversals
>if< _Mononykus_ is placed above _Archaeopteryx_ with avialan birds in the
bird-theropod cladogram. Too many reversals--too many characters reappearing
in the same state they were in before being lost--set off loud warning bells
(loudness of bells is directly proportional to the number of reversals) that
the cladistics has come a cropper. I move the branch with _Mononykus_ out of
the avialan clade and farther back among the theropods, and soon the warning
bells stop ringing.

So perhaps the principle here is: Do the cladistics, but if there is an
unusual number of reversals (whatever that may mean), re-examine the
characters and see whether moving the clade will make the worst of the
reversals go away. Cladistic analysis is simply a tool--a worthy tool but
only one among several--for doing phylogeny.