Simulating Phylogenies and Evolutionary Data

Abstract

Data simulation has a critical place in evolutionary biology. Because evolutionary patterns and processes unroll over long time spans, their simulations has become an important ingredient of evolutionary research [e.g., 8]. From a statistical inference point of view, it is necessary to simulate data to assess the statistical properties of hypothesis tests and parameter estimators. This practice still needs to be generalized: it is quite common to read papers describing new data analysis methods without presenting some simulation results in order to asses the bias and variance of the estimators, or the type I and II error rates of the test. On the other hand, the recent rise in uses of simulation procedures for statistical inference (e.g., Markov chains Monte Carlo) has emphasized the need for simulating data under a wide range of situations.