Trees

Three Gene Tree:

rbcL, atpB, 18s...

The "three gene" tree includes 567 taxa, mostly angiosperms, with a carefully selected group of gymnosperms. For some reason, Cycads were not included.

The shortest trees for this data set were first found with the parsimony ratchet. The Soltises approached Kevin Nixon after several months of anlayzing this data set with PAUP* on very fast Macs. They were concerned that they were not finding the shortest tree(s). The first night of running the ratchet (with DADA and NONA) on this data set (on a much slower PC), trees 12 steps shorter than had previously found were discovered. The next day, trees one additional step shorter (13 steps shorter) were found. These are almost certainly THE shortest trees for this data set. They have been verified (hit several times) with Pablo Goloboff's new methods, sectorial searches, drift, etc., using TNT.

We provide here a WINC file that can be loaded directly into Winclada that contains the data file (stripped of uninformative characters) and the consensus tree. Anyone wishing to have sets of individual shortest trees can send an email to the address on the homepage.

The data/tree is stored in a single self-extracting executable. After downloading, run the executable and the file threegen.winc will be extracted. Open this file with Winclada to see/edit/play with the tree...

ZILLA

The tree you are about to link to is the REAL consensus of the shortest trees for the 500-taxon rbcL data set for seed plants that was originally published by Chase et al. (1993) and later reanalyzed by Rice et al. (1997) who was still not able to find the shortest trees for this data set.

The tree you are about to view is the ONLY public posting I am aware of (April 2002) in ANY form of the most parsimonious trees for this data set. Strangely, other sites have not acknowledged the existence of these shorter trees.

The trees from which this consensus tree are derived are in excess of 50,000 trees
Found with NONA using traditional searches (usually in 1-3 weeks on a 200 mhz Pentium Pro)
Found with the ratchet (usually in 3-6 hours on a 200 mhz machine)
These trees have a length of 16218 when uninformative characters are removed, 2 steps shorter than those found by Rice et al. (the Rice et al. trees have been verified to have 16220 steps when uninformative characters are removed, although massive confusion persists because of incorrect values reported by PAUP).

VIEW ZILLA TREE

If you want to read more about the ratchet, see:

Nixon, K. C. 1999. The parsimony ratchet, a new method for rapid parsimony analysis. Cladistics 15: 407-414.

This page and tree were first posted on the web on 4 February 2000 A.D., and updated in April 2002.