This file illustrates an interaction with Swofford's PAUP. Following from the previous file, assume that we have reconstructed the gene tree and right and calculate an observed s value of 3. Can we get an idea of the error on this using the methods of the parametric bootstrap? What we could do is first to estimate parameters of sequence evolution on the gene tree. Mesquite doesn't do that (yet), so let's assume the rate of mutation (etc.) has been estimated. Then, we can use Mesquite to simulate a data matrix on the tree, send the data matrix to PAUP, obtain the reconstructed trees from PAUP, and calculate the s values those reconstructed trees would suggest. If we do that many times, we get an idea of how the "observed" s would vary due to error in reconstruction of the gene tree. This can be done by requesting a Tree Block Values chart in Mesquite, and indicating that the tree blocks are to come from PAUP, PAUP is to be sent simulated data matrices, selecting the estimated model of evolution, and indicating the sample size (how many matrices). Mesquite then sends data matrices repeatedly to PAUP, calculates s from the resulting trees, and presents the results as a chart.