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

Re: Parsimony

At 02:52 PM 11/6/96 -0500, George O. wrote:

>But nobody has ever shown that nature obeys the principle of parsimony as
>strictly as the cladists insist it does when it comes to phylogeny. Au

        I believe that the idea is not that nature is parsimonious, mon
frair, but that with enough data, parsimony (which can be thought of as
another tool of logical analysis, the same process which drives the
phylogenies of non-cladists) will find the most likely phylogenetic
hypothesis.  Pasimony expresses the belief that, with enough data,
evoilution will show its face clearly .
        If, given a large volume of data, we can still not be confident in
our conclusions due to the processes of evolution, there doesn't seem to be
any point in trying to wade in and sort it all out by hand.

>contraire: Others maintain, with equal justification, that some
>characters--particularly morphological characters--may arise repeatedly and
>independently in multiple lineages (such characters are called homoplasies),
>and that enough such homoplasies can defeat the analysis. So, although one

        In a case where there are so many homoplasies, can you name a method
which will produce a more reliable hypothesis.  This does not sound to me
like a problem with parsimony, but a problem with elucidating phylogenetic
relationships, as I believe Mickey, amongst others, has pointed out.
Phylogeny may be discernable, but it sure covers its tracks well.
        And note I said hypothesis.  IMHO, it is foolish to assume that we
can ever be entirely certain of a phylogeny, especially were exitinct
animals are concerned.  I gather I am not the first to consider this.  ;)

>can frequently settle on a single cladogram as being "most parsimonious,"
>there is no way to show that this cladogram represents the true phylogeny of
>the group under analysis.

       Even with a time machine, we would need legions of workers to observe
every speciation event in order to know that a hypothesis represents the
ture phyologeny.  We have only the constructs of our own logic to guide us
in the evaluation of a phyologenetic hypothesis.
        If we declare that the actual phylogney cannot be approached due to
homoplasies, we abandon must abandon our claim to being scientists, and
crawl back to devining rods and religious texts for our guidance.  The
assumption that parsimony will eventually cut through all "phylogenetic
noise" is a scientific one.  I am certainly most interested in reading the
opinions of scientists who do not share this opinion.

>Cladists maintain that the way to discover homoplasies is to perform the
>cladistic analysis, then see which characters appear more than once in the
>most parsimonious cladogram.

        Is there another way?  Do they propose that a worker should simply
sit at his desk and decide which characters are convergent and which are
synapomorphic?  I'll admit that that system often produces answers which
agree with later cladistic analysis, but do they really believe that this
method works *better*?

>Others maintain that certain characters may be
>discarded as homoplasies before performing the analysis. Here again, there is
>no way to resolve the differences between these viewpoints.

        This strikes me as being very close to "old school" phylogenetics,
where everything was done by hand.  I believe Charig and Milner had some
things to say about Gauthier's use of characters which they would discard
(in _Dinosaur Systematics_).  I cannot help but think that this is the
equivalent of throwing out lab results because they don't fit your
hypothesis.  If you do this in any other science, you could find your head
handed to you (assuming you don't have tenure, and someone catches you...).

>it even allows us to attach a
>numerical figure to the confidence we can have in the truth of a particular
>cladogram, it has definite and fundamental limitations.

        I am not fully versed in the numerical aspects of cladograms, but I
was under the impression that these numbers refer to the fit of the inferred
tree and the data, not necessarily to the "truth" as you call it (how
closely the phylogenetic hypothesis fits the actual (unknowable) phylogeny).

>We should beware of
>those who invariably accept cladistic analysis ahead of all other kinds of
>analyses in creating phylogenies.

        I know of several well made arguments (I believe Ostrom did not use
cladisitics to postulate the theropodian origin of birds), but I cannot
recall any instances where workers have a priori accepted a phylogenetic
hypothesis based on cladistics over a well presented non-cladistic
hypothesis.  Yes, in some cases the former was accepted over the latter
*after* consideration of both...
        I am wondering, how often are well prepared, detailed, supported
phylogenetic hypotheses presented in the litterature which have not been
evaluated using cladistic methodology?

| Jonathan R. Wagner                    "You can clade if you want to,     |
| Department of Geosciences              You can leave your friends behind |
| Texas Tech University                  Because your friends don't clade  |
| Lubbock, TX 79409                               and if they don't clade, |
|       *** wagner@ttu.edu ***           Then they're no friends of mine." |
|           Web Page:  http://faraday.clas.virginia.edu/~jrw6f             |