Abstract

Since being pioneered by Efron (1979) and introduced to phylogenetics by Felsenstein (1985), non-parametric bootstrapping has been widely used in statistics and statistical phylogenetics to generate measures of statistical uncertainty. Less widely adopted is Felsenstein’s (1988) suggestion to use parametric bootstrapping (or more generally, simulation) to assess variability in estimates. Here I give a brief description of bootstrapping in phylogenetics, including some associated controversies, describe the general framework of parametric bootstrapping and provide some examples from tree inference and ancestral sequence reconstruction to demonstrate its flexibility and power.