Statistical inference using simulations
or randomizations
Mesquite has various tools to simulate
and randomize characters and trees, including simulations of DNA
sequence evolution, tree reshuffling, coalescent gene tree simulations,
and others. With these tools, you can build your own statistical
inference methods tailored to your question.
For instance, you can test an hypothesis
about phylogenetic structure using parametric bootstrapping, simulating
many character matrices on a proposed tree, for each attempting
to reconstruct the tree and assessing the reconstruction. Or,
you can simulate gene trees within a population to test a population
genetics or phylogeographic hypothesis. Or, you can study the
effect of branch length uncertainty on calculation of Felsenstein's
contrasts by adding random noise to branch lengths on a given
tree, repeating many times to see the variation in calculated
correlation coefficient of contrasts in two characters. Many other
analyses are possible. We give an overview of the features and
some examples in the following.