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

RE: Cladistics algorithm? <FIN?>

At 4:40 PM -0500 01/31/2001, Thomas R. Holtz, Jr. wrote:
> From: owner-dinosaur@usc.edu [mailto:owner-dinosaur@usc.edu]On Behalf Of
 Rich Grenyer

 >However, the topic you are discussing is one of computer science and/or
 >computational phylogenetics, and not dinosaur science per se.  You should
 >seek out the various systematic discussion groups or
 computational science
 >groups for those details.

Is this true? I know this list is volatile, and that cladistic/anti-cladistic/partially-cladistic/my-version-of-cladistic threads start up as soon as you look at them, but cladistics has had such an impact of dinosaurian phylogenetics and, hopefully, will continue to do so, that surely asking what the programs actually _do_ is fair game? Is it likely anyone is going to initiate an in-depth discussion on ACCTRAN/DELTRAN optimisations, Dollo, Wagner or Fitch parsimony or Farris' Parsimony Jacknife? Goloboff's Ratchet implimentations? I hope not, because there are other lists better suited (and less interesting). But surely someone asking what fundamental cladistic algorithms actually _do_ is worthy of a reply and some pointers?

It was my impression that the question was essentially one of code: as in "what precisely is the computer code that runs the analysis?" If I was wrong, please let me know.

However, in the short form:
You take a matrix of taxa and characters, filled with 0s, 1s, 2s, ?s etcs.;
You open that matrix in the software;
You choose one of various options (exact ones vary by program: in PAUP*
there is Heuristic, Branch-and-Bound, and Exhaustive Search; NONA has
Heuristic and Parsimony Rachet; etc.);
Using the particular batch of code, the software compares potential
alternative trees under some parameter (parsimony, maximum likelihood,
etc.), calculating (for example) tree length for each tree generated;
The software saves only the shortest trees (i.e., those with the lowest tree
length) (unless you ask it to save something else);
You then have a set of shortest trees, which then you can examine (i.e.: see
the topology of the trees; get certain tree statistics; etc.).

If the original poster wants more details on the exact nature of the various
options, I again  recommend going to a particular software package and
seeing the particular algorithm used.

Hope this helps.

Now that everyone who is interested in this has no doubt read the chapter on infinite series and matrices, and have all reviewed Thom's description..... it gets even more fun! ;^)

Let's add to both Thom's and My earlier posts on this for those of you who have never really had an opportunity to do a cladogram of your own. You can build a matrix of characters - to infinity if you choose. You get to choose the characteristics and number them from 0 to infinity, from the most primitive to the most derived functional characteristic the organism has, ........ you get choose.......each and every one of us building a cladogram gets to choose.........therein lies the caveat that builds the passion so many people rise to the debate with, when it comes to clades and cladistics.

You (the person building the tree) select the characters - what is the most primitive characteristic (in your opinion) and then what is each the next most successively derived characteristic (in your opinion).

You then, the researcher, select the matrix you want to process, hit return, and hope that you get the most parsimonious tree. You can and will find however, that in some programs, like "MacClade" that you can manually often create a tree that is more parsimonious than the one the computer gave you, simply by moving the branches around.

Another person may heartily disagree with your cladogram and by selecting characteristics differently, in a different order, and create an entirely different clade.

I give this as a lab in Honors Historical Geology - and give groups of 4 students each, the same set samples to work with (Trilobites, Echinoderms, Brachiopods, etc.) You get a very good feel for how many different parsimonious cladograms can be created by a group of people, using single set of data.

You may perhaps then begin to see why cladistics raises the passions of so many people........ and you can certainly discern what is the potential for inherent bias within such a system as well as its laudable potentials.

Best regards,
Marilyn W.
                                =00=  =00=  =00=  =00=
                                Marilyn D. Wegweiser, Ph.D.
                                Adjunct Curator of Vertebrate Paleontology
                                Cincinnati Natural History Museum

                                Assistant Professor of Geology
                                Department of Geology
                                Ball State University
                                Muncie, Indiana
                                Office: 765-285-8268; 765-285-8270
                                FAX:    765-285-8265