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

Re: Cladism is not an inkblot



The Mage <themage@thedigs.dnsalias.net> wrote:

> I don't understand why everyone has been fighting back and forth for
> years on the list about the value of cladistics, or more accurately,
> the value of parsimony in cladistics, when the idea that parsimony
> (with as many characters you wish) is superior to any other method
> available today, is a falsifiable hypothesis, is it not? Why get into

    I think you may be conflating issues. In most discussions on this list,
we have been concerned with the value of phylogenetic systematics in a
general sense, not of parsimony (although parsomony has been attacked as
well. You should consider whether the problem you wish to address is with
phylogenetic systematics, the idealogy in which relationship as expressed in
internested monophyletic groups discovered through reproduceable,
falsifiable methods should be the basis of biological systematics, or with
parsimony, the logic-based process of selecting a tree which makes the
fewest assumptions about evolutionary processes. These posts may help
(although I reserve the right to disagree with any and all of them):

Holtz:
http://www.cmnh.org/dinoarch/1999Jan/msg00236.html

Brochu:
http://www.cmnh.org/dinoarch/2000Jul/msg00110.html
http://www.cmnh.org/dinoarch/1998Aug/msg00286.html
http://www.cmnh.org/dinoarch/1999Jan/msg00340.html
http://www.cmnh.org/dinoarch/1999Jan/msg00375.html
http://www.cmnh.org/dinoarch/1998Feb/msg00517.html
http://www.cmnh.org/dinoarch/2000Oct/msg00355.html
http://www.cmnh.org/dinoarch/2000Jul/msg00110.html
(these last two are just classics I personally love, not on-topic)

    Molecular data can be analyzed using maximum likelihod, distance
techniques, and Bayesian methods.Until recently, parsimony was the only
commonly accepted, reproduceable method for analyzing morphological data in
a phylogenetic framework (strictly, it is an optimality criterion, not an
analytical method). So, "any other method" in your statement above needs to
be clarified. I simply don't know what alternative you are offering for
test.

[out of order]
> standard. How much closer to the results of molecular biology than the
> best of traditional methods will the most parsimonious cladograms take
> you? A percentage, anyone? Because I see no real obstacle to obtaining
> one.

    I presume, then, you propose to convert traditional "phylogenies" to
cladograms, and test them against both morphological parsimony analyses and
molecular analyses (presumeabl ynot using parsimony)? This is presuming that
they CAN be interpreted as a cladogram. Many cannot be: they simply lack the
explicit statements of relationship necessary to do so. For those that can,
the process of comparing tree structure is actually non-trivial, and is
certainly not as simple as producing a "percentage." I suspect that, since
the former are generally formulated under a different systematic outlook,
they would contain many polytomies, which would have to be accounted for in
any comparison (I suspect this also is a non-trivial problem).  In any case,
because molecular studies are conducted under the same ideological framework
as the morphological parsimony studies, you would be effectively comparing
apples and oranges... to apples. I think I know which one will win.

[back to the order of the original post]
> all these endless he-said she-saids about it? Doesn't make any sense.
> You may never be able to falsify many particular stories of mutation
> and selection, but surely with independent molecular data, the idea of
> the applicability of parsimony itself is available to a much higher

    See my previous post on molecular studies. Molecular data are not a
panacea; gene sequencing does not give us the much-desired mainline to the
tree of life. Analyses of molecular data are fraught with just as many
assumptions as those of morphology (if not more!). It's like getting a
second opinion about a medical condition... If the doctors disagree, you
can't tell which one is right, and if they agree, they could still both be
wrong.
    In short, we cannot falsify the validity of parsimony as a
methodological tool without knowing the true tree. If we knew the true tree,
we wouldn't need parsimony. We can conduct simulation studies, and
references to some of these are in the posts linked above. However, because
simulations depend quite strongly on the model used to develop the
simulation, I'd take these with a grain of salt.

    Also, for the record, some workers use parsimony with molecular data.

    In regard to phylogenetic systematics in general, I suggest you go out
and review the current systematic literature. "Traditional" systematics
pretty much lives on only in textbooks, emeritus professors, old-school
paleontologists, and certain notable (and loveable) die-hards on the
Dinosaur listserv. The systematics of today is pervasive, far-reaching, and
thoroughly phylogenetic. The integration of population genetics, ecology and
systematics in the sequencer room at the end of the hall has caused an
explosion of phylogenetic thought and methods throughout biology. This is no
cold-fusion phenomenon, no flash-in-the-pan event, this is a classic
"paradigm shift" in the way people think about life (i.e., it has a
history). Systematists are no longer stamp collectors, they support the
framework upon which our understanding of life is hung. Systematics is now a
science, accessible to everyone, and systematists are guides, not guardians
of Special Knowledge. Cheap (hah!), reliable sequencing is translating the
bible of phylogeny into the vernacular, and the days of dictating phylogeny
from on high are over. It is only understandable that the deposed princes
should resist their usurper.

    Phylogenetic analysis, either using parsimony, likelihood, or some other
optimality criterion, is a reproducible, scientific approach to phylogeny
reconstruction (or estimation, if you prefer). It produces testable
hypotheses that make predictions about, and elucidate, the secondary
properties of the groups it delineates. "Traditional" methods rely on
authority and personal judgment, are not reproducible, and often do not
result in testable hypotheses. Is it surprising that these new methods have
been adopted by biology in general?

    I hope this helps.

    Jon Wagner