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

Re: Learning cladistics (was Re: Dinosaur Web Pages' Re-Opening)



This is getting way off topic, perhaps it ought to be continued off-list?
At 08:09 PM 9/4/97 -0400, Dinogeorge wrote:
>Cladistics is trivial. What's not as trivial is recognizing characters and
>character suites worth subjecting to analysis.
        Basically, then, excluding the characters which do not agree with
your a priori interpretation of the phylogeny. That can't be good science.

>And: cladistics is >not< as testable as you think.
        I was *not* saying CLADISTICS was testable, merely that CLADISTICLY
DERIVED HYPOTHESES are testable. Personally, I don't think we can ever truly
test any means of phylogenetic reconstruction, but I may be wrong.

>I have repeatedly called for a test of the methodology
>by doing a double-blind cladistic analysis of a known phylogeny (one
>constructed, for example, by a computer).
        In order to provide a fair test, this would require us to know the
"rules" which govern evolution to such a detailed extent that we would no
longer need cladistics.

>We need to get a handle on when cladistics gives us a confirmed, correct
>phylogeny and under what circumstances cladistics >fails to do so<.
        We can never be sure that just because cladistics succeeded or
failed in your test, it would do so on real organisms. We must instead look
to the theoretical underpinnings of cladistics, parsimony. Parsimony is the
principle governing much (if not all) science, and is a basic logical
underpinning of scientific research. Note that this does *NOT* require
evolution to be parsimonious, merely it requires that, at the macroscopic
level, the laws of probability must apply today as they have in the last 4+
billion years. Given that, the explanation which requires the least numer of
unsubstantiated assumptions, and the fewest untestable hypotheses, is most
likely to be the correct one. And given that assumption, the application of
the principle of parsimony should only fail to produce the best hypothesis
in the absense of some data.
        Granted, all paleontological applications involve the absense of a
great deal of data. However, you cannot simply say that, because of the lack
of data, all cladistics is bunk, or we should disregard cladistic analyses,
even when backed up by non-cladistic evidence, simply because the results do
not fit with your conception of the phylogeny.

       
>For example, I assert that if there are a significant number of
convergences >and homoplasies in a phylogeny, then cladistics will not
perform correctly.
        This is possible. However, how can you tell whether they are
homoplaisies (convergence is a type of homoplaisy, btw) without knowing the
phylogeny. You cannot. Therefore, the rational approach is to select the
phylogeny which assumes the minimal number of homoplaisies (on the
assumption that homoplaisy is less likely than synapomorphy). It may not be
the right one, but it is the one which best fits the evidence, with the
fewest usubstantiated assumptions and the fewest untestable hypotheses. I
cannot imagine what scientist will argue with this approach.
        And do recall that just because a cladogram has been provided does
not mean that that cladogram is the *only* possible phylogeny. It is subject
to test and refutation.

>And we won't know it's performing incorrectly, because cladists will not
accept >a non-cladistic check on their cladograms.
        How often does the Federal Reserve consult astrologers before
changing the prime lending rate? There is, as far as I know, no currently
accepted "non-cladistic" method for producing a phylogeny which is testable,
reproducable, and firmly founded in evolutionary theory.
        The statement you make above presupposes that information from
non-cladistic methods is of the same quality as information from
non-cladistic methods. This is untestable. However, since no one who uses
non-cladistic methods has ever managed to codify their method into something
which is both agreeable to a large number of workers and reproducable, it
seems clear that other methods ought not to be used to falsify cladistic
analyses. THe best way to falsify a cladistic analysis is to run your own.
        Of course, although we truly can NEVER know if cladistics is
"running correctly", I draw your attention to the instances (which are
reportedly many in number) in which cladistics has upheld previous
non-cladistic work on a group.

>It takes a >lot< of faith to believe that analysis
>of an ever-larger suite of characters will in all cases eventually converge
>on the correct phylogeny instead of winding into an ever-deeper morass of
>error.
        No, it simply takes an understanding of the basic principles of
probability.
        :)
        Wagner

----------------------------------------------------------------------------
    Jonathan R. Wagner, Dept. of Geosciences, TTU, Lubbock, TX 79409-1053
               "Not the One..." -- Zathras (not Zathras)