Summing up for now (was RE: Gaia theropod follow-up: a "new" phylogeny)

["Thomas R. Holtz, Jr." <tholtz@geol.umd.edu>]

> From: owner-dinosaur@usc.edu [mailto:owner-dinosaur@usc.edu]On Behalf Of
> Jeffrey Martz
>
>Dinogeorge writes
> >First of all, I have detected an unspoken attitude among
> >dino-cladists

If it is unspoken, then perhaps you are misinterpreting the attitude.

> that
> >"more is better": "Your analysis only has 258 characters? Hah. >My<
> analysis
> >has 356 characters. Therefore it must be better than yours." But how can
> you
> >compare either analysis against the other? There is simply no metric for
> >doing this--certainly none that couldn't be challenged. So--we
> do a >third<
> >analysis, perhaps with >even more< characters. And so on, ad infinitum.
>
>
>     Having more characters in an analysis does not make it fundamentally
> better.  However, if these characters seem to be generally true
> of the taxa
> you are coding them for, are not due to distortion and reconstruction, are
> not almost synonymous with another character you have coded, and show
> variance between taxa, then more information is better.  Analyses can be
> compared by not only comparing how much information is included for the
> computer to make its decisions (how many characters there are) but by
> judging how well thought out the characters are.

Quite so.  Some people are getting more hung up over the particular numeric
value I posted, without considering a) the particular characters in question
(gee, that might require reading the paper before commenting on it... :-);
b) the fact that in order to describe the distribution of anatomy across the
wide range of taxa in that particular analysis, many different descriptors
are necessary.

Also, it might be pointed that it is perverse NOT to include characters that
others have proposed in the taxa in question, as long as you a) agree that
they describe morphologies (or genes, or whatever) that you can
independently confirm as present or absent; b) agree that they are not
synonyms of characters already in your analysis; and c) actually vary in
your study (that is, that they are present in some but not all of the OTUs).
So, yes: Xu and Sereno and Forster and Coria and others have all made
observations which they have proposed as characters within the spread of
taxa I am investigating (theropods between _Coelophysis_ and
Ornithothoraces, basically) that I have not yet considered.

I could simply ignore these characters or argue (without additional
evidence) that they are not significant.  OR I can see if I can verify these
observations for my self among the taxa in the study, enter these
observations, and see if/how they effect the topology I found previously.
This is *most* important if the characters in question do *not* support my
previous trees: instead of being a bad scientist, and only accept the
physical observations that support a previously "favored" hypothesis, I
should try and be a good scienctist, and set up conditions in which my
previous hypotheses could fail.  (This last part is one of the least well
understood aspect of scientific methodology in the public).

Additionally, various authors have proposed the non-monophyly of some taxa I
have considered as single taxa before.  In order to test this, I have to
break up some of my OTUs into component taxa.  However, in order to evaluate
the monophyly of these groups, I may (will, actually) have to add additional
characters into the matrix which have previously been used to support that
group.  (Example: consider the possibility of a paraphyletic
"Dromaeosauridae" relative to birds.  In order to test this, one has to
break up the dromaeosaurids into several different OTUs.  However, in order
to fairly test the alternative hypotheses, we must include observations that
both a) are shared between some dromaeosaurids and birds, but not other
dromaeosaurs (if any) and b) are shared among the dromaeosaurids, but are
not found in birds (if any)).

>     Some analyses do play around with the characters they include in order
> to see if they get something different, in a sense putting a
> relative value
> on them.  I think this is a good tool.  Christopher Bennet's paper on
> pterosaurs, in which he removed characters relating to locomotion to see
> what the rest of the charatcers did to the placement of pterosaurs, is a
> good example.  However, I think its probably always best to start
> off with a
> analysis that includes ALL the characters and then experiment to compare
> with that.  You then basically have a standard that lets you spot
> where the
> contradictions are.
>
Gee, sounds like an SVP presentation I should be preparing for... :-)

                Thomas R. Holtz, Jr.
                Vertebrate Paleontologist
Department of Geology           Director, Earth, Life & Time Program
University of Maryland          College Park Scholars
                College Park, MD  20742
http://www.geol.umd.edu/~tholtz/tholtz.htm
http://www.geol.umd.edu/~jmerck/eltsite
Phone:  301-405-4084    Email:  tholtz@geol.umd.edu
Fax (Geol):  301-314-9661       Fax (CPS-ELT): 301-405-0796