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

taxon selection vs. character quantity



Sorry to disappoint anyone, but after adding every character in Müller 2003 
that could be widely seen and was not on my prior character list (total now: 
213), PAUP came up with another single tree. The only difference arose from a 
placement of Thadeosaurus one step higher toward Prolacerta due to addition of 
skull and cervical characters. Müller 2003 was based on earlier works by 
Dilkes, DeBraga, Rieppel and others and used 175 characters.

Müller 2004 writes in his abstract after employing 33 taxa and 184 characters 
"with the main focus on taxa that were either rarely entered... or where strong 
differing opinions towards phylogenetic placement exist": ..."the tree topology 
depends much more on the choice of taxa than on the number and definition of 
characters..." 

The only differences my cladogram has with his is the increased resolution 
accompanying the addition of 100 more taxa.

Müller 2004 (page 1 only) can be dowloaded for free at:

http://www.pfeil-verlag.de/07pala/pdf/3_52d17.pdf


Think of it this way: 

If you include 150 characters to describe an animal, you're going to get a 
pretty good picture of that animal, be it elephant or fly. Sure you could add 
more characters, but they would be important only to divide species, races, or 
family members.

On the other hand, if you don't include a certain animal as a taxon selection, 
you'll never know anything about it -- unless, of course, you've included a 
number of its closest relatives, and then you could just about do a police 
artist sketch of the missing culprit (sans autapomorphies).

Character/taxon ratios, once you get past 150 of each (22,500 matrix boxes), 
are both getting close to the peak of the graph where the probability curves 
flatten out, in my experience. Again, this all comes as news in the Diapsida 
world because people have been describing this metaphorical elephant from a 
number of different angles and no one previously has employed a wide spectrum 
of taxa in one study to figure out relationships overall. Evans (1988) and 
Müller (2003, 2004) appear to have done the best jobs to date, but even they 
excluded many taxa.

So all previous studies, even those with a dozen taxa, are correct, as far as 
their limited taxon selection resolution take them. But all saw the whole 
picture a little fuzzy and incomplete by focusing only on the bits and pieces.

Thanks for the encouragement to keep pressing,
I learned alot.

David Peters
St. Louis