[Introduction] [Next screenshot]

Mesquite does not need to work with trees. For instance, there are modules that can do Principal Components Analysis without reference to phylogeny. However, this file blends PCA and phylogeny. It shows a PCA on data of 50 characters simulated live (with a Brownian motion model) on the tree shown at upper left, and the terminal taxa linked by the tree plotted into the PCA space. At lower right is an alternative ordination ("Evolutionary PCA", in development by Maddison & Dyreson), which seeks the axis that maximizes change along the branches of the tree (instead of simply variance). Note that the EPCA doesn't emphasize the distinctness of major clades nearly as much as does the PCA. If the branch lengths of the tree at upper left are changed, the characters get re-simulated accordingly and the plots change.