Re: Gaia theropod follow-up: a "new" phylogeny

["Jeffrey Martz" <jeffmartz@earthlink.net>]

philidor11 wrote...
>I was observing that when you say a 'blind' (based on principles programmed
>into an algorithm) analysis can correctly hypothesize relationships, you
are
>assuming that the algorithm will correctly identify how evolution operated.

>If the algorithm can come to a (more likely) conclusion, then implicitly it
>has identified how evolution works because it has identified starting and
>ending points, with the explicit assertion that they are connected.


    The algorithm may have identified the particular pattern or pattern of
evolution in  the group being studied, but since the patterns are so
variable between different groups I wouldn't say that cladistic analysis can
identify The One Universal Evolutionary Law For Everything, which is the
impression I got you were looking for from your last message.  I think this
would be kind of like entering every X-cel spreadsheet in the world into a
big database, and using that to write a supersmart X-cel program that could
predict the data you would enter on a speadsheet before you type it.

>If your algorithm cannot correctly track evolution in all cases, then there
>would be little reason to believe its findings, no?

    By this do you mean "if all analyses do not agree perfectly, none of
them can be right"?  The weatherman can't track the weather prefectly in all
cases, but he can get close enough to roughly make plans for the week.

>The finding of universally applicable rules was implicit in your original
>statements, and that was what I was questioning.

    The "universally applicible rules" that a cladistic analysis is based on
are:

1. If two taxa share a character, it is more likely to represent homology
then convergence.
2. The more traits that taxa have in common, the more likely they are to be
closely related, and a taxa will be more closely related to a taxon with
more characters in common

    Now, as a lot of the previous discussion has focused on, neither of
these points is always true.  Convergence is very common.  Moreover, some
characters, usually complex ones, are probably less likely to have developed
independantly then a suite of relatively simple ones that might be likely to
be tied together, which is why I like the idea of running seperate analysis
that take out character suites that may be particularly susceptible to
homoplasy to see what happens.
    However, if two groups have a lot of characters in common, even ones
that seem pretty simple to evolve (a slightly bigger trochanter for example)
that apparantly have NOTHING TO DO WITH  EACH OTHER FUNCTIONALLY, then you
start building up an unlikely number of coincidences; AND THIS IS WHY IT IS
IMPORTANT TO HAVE AS MANY CHARACTERS AS POSSIBLE!  Even if some characters
should be weighted more then others, if you get a bunch of (probably)
unconnected light-weight characters that all seem to point in the same
direction there is an implication there.

LNJ
*****************************************************************
It is our duty to make the best of our misfortunes and not to suffer passion
to interfere withour interest and the public good.
-George Washington

It is your business when the wall next door catches fire.
-Horace
*****************************************************************
Jeffrey W. Martz
Graduate student, Department of Geosciences, Texas Tech University
3002 4th St., Apt. C26
Lubbock, TX 79415
http://illustrations.homestead.com/Illustration.html