DAISIE (Dynamic Assembly of Islands through Speciation, Immigration and Extinction) is a framework for a dynamic stochastic model of island biogeography that can be used to estimate the rates of colonisation, speciation and extinction (CES rates) from phylogenetic trees of insular communities by maximum likelihood, and to simulate such data sets given a set of rates. The framework is available in an R package, and has been used in various studies to estimate CES rates for different insular systems based on molecular phylogenetic data. Simulations with estimated parameters have been used to explore the goodness-of-fit of the model and uncertai...  more

DAISIE (Dynamic Assembly of Islands through Speciation, Immigration and Extinction) is a framework for a dynamic stochastic model of island biogeography that can be used to estimate the rates of colonisation, speciation and extinction (CES rates) from phylogenetic trees of insular communities by maximum likelihood, and to simulate such data sets given a set of rates. The framework is available in an R package, and has been used in various studies to estimate CES rates for different insular systems based on molecular phylogenetic data. Simulations with estimated parameters have been used to explore the goodness-of-fit of the model and uncertainty in parameter estimates via bootstrap. We report two minor bugs in the simulation code that was used in the manuscripts published to date that have used DAISIE. One affects the branching times for some species that evolve via cladogenesis, the other affects cases when a single mainland species colonises an island multiple times. While these bugs do not have any effect on maximum likelihood estimation or model selection from empirical (non-simulated) data, they do have the potential to affect analyses based on simulated data. Here, we describe and fix the two bugs. To assess the magnitude of the error that they introduced, we re-run simulations and bootstrap analyses for the published cases that are most likely to be affected. We find that the differences between results generated with the new code and the original code are very minor, equivalent in magnitude to differences that would be obtained when using a different set of random numbers to generate the simulations. We conclude that the qualitative and to a large extent also the quantitative results of previous publications using DAISIE are robust to these bugs, and we provide a new version of the R package (v 3.2.0) with the corrected code.