Technical details

Conditioning on minimum clade size

In order to ensure the generated islands have sufficient data for inference, all simulations are conducted by conditioning the number of surviving independent clades or singletons to be at least 5. This is achieved by calling the relevant DAISIE simulation functions with the argument cond = 5. This results in simulations with less independent clades being discarded, which can result in an increase in runtime if immigration rate is low and extinction rate is high.

Simulation and likelihood constraints

Given the stochastic nature of the simulation models, and that given the very nature of these studies the properties of the simulated output are not known a priori, generated datasets may cause the maximum-likelihood estimation routine to fail due to numerical problems. Thus, whenever MLE of a given replicate, be it at the oceanic or novel stage, cannot successfully complete by non-convergence or due to an error (and assuming the error is recoverable), then the simulated data and/or MLE results of that replicate are discarded (saved for debugging purposes in objects prefixed by failed_, followed by the respective stage of the failure).