CALIBRATING THE BOOTSTRAP TEST OF MONOPHYLY

It has been suggested that the bootstrap test of monophyly is too conservative, i.e. the test rejects the hypothesis of monophyly when it is true far too often. Here, a method called the iterated bootstrap is described which estimates by randomization the probabilities associated with rejecting the hypothesis of monophyly when it is true and accepting the hypothesis of monophyly when it is false. Using this method, the bootstrap test can be calibrated by taking account of the errors associated with the hypothesis test. This method is applied to the high-level phylogeny of the Platyhelminthes using 18S rRNA sequences. The analysis suggests that the data cannot unequivocally resolve the placement of the Platyhelminthes with respect to the Annelida and Insecta.