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

RE: Cladistics algorithm?

> From: Dinosaur Interplanetary Gazette [mailto:dinosaur@dinosaur.org]
> The question, really, is one of what algorithms are used.  I'm
> not interested in
> whether it's done in J+++ or Cobol.
> There is, as someone else pointed out, a GREAT DEAL of discussion about
> cladistics here, and it's clear that there are assumptions behind
> the various
> "positions" people take.
> Clearly, now, not only the exact software, but the exact
> algorithm used within a
> specific software colors the supposed "objective data output"
> that is used by a
> paleontologist to draw a specific and deeply held conclusion.
> In fact, it seems to me that the one of the addressable problems
> in paleontology
> is the "fudge factor."
> Which scientist uses which particular "fudge factor" in reference to which
> particular algorithm in which particular piece of software.
> In any other hard science (eg physics, chemistry) the variables
> and constants
> must be clearly defined and the experiments designed with them in
> mind. Analysis
> of medical effectiveness, for example, seems to have a much
> smaller range of
> variance than what seems to be acceptable in paleontology.
> It APPEARS to me that that the variables and constants here are
> having a lot of
> smoke blown around them and the original question was to try to
> understand the
> precise algorithms to begin with and to see if the variant
> algorithms are really
> identical, equivalent, or have inherent differences that ought to
> be considered
> in conversation.
> So, for example, a paleontologist might need to specify not only
> but also

This is standard operating procedure, often listed in the Methods &
Materials section of the paper.

Just running through the latest JVP issue, the analysis by Olsen et al. on
the new _Erpetosuchus_ uses PAUP 3.1.1, exhaustive search and Clark et al.'s
longer paper on _Hesperosuchus_ was also PAUP 3.1.1, with brach-and-bound.
If multiple iterations are employed, the author is supposed to say so.
Clark et al. have 8 pages devoted to disucssion of character choice and

> 1) the name of the software used
> 2) the setting of the software
> 3) the version of the software (do they ever update them and the
> underlying
> algorithms?  and are there proper names for these algorithms
> like, say "Fred's
> Formula" or "Manny's Method?")
> 4) in what direction the data is "weighted" during input
> 5) how much post-analytical "blurring" is done.
This is typically done (and when I review a paper, I require it!).

I hate to use my tired old refrain, but your concerns would be much reduced
if you read the papers in question to see if such is done or not.  It is
true that some authors do not do such; their analyses should indeed be
treated with caution.  On the other hand, in the field of systematics there
are whole journals (Systematic Biology and Cladistics, in particular) which
devote a majority of their pages to "methods papers".

So, to turn this discussion to an informed one, could you bring up a
particular paper or papers with a phylogenetic analysis, and we can discuss
whether (and how effectively) the methods were documented.  For purposes of
the list a dinosaur paper would be best, but any form of phylogenetic
analysis (morphological or molecular; paleo or neontological) would suffice.

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