Accounting for mode of speciation increases power and realism of tests of phylogenetic asymmetry

Tree symmetry is crucial to the investigation of evolutionary processes, particularly adaptive radiation. Testing a phylogeny for asymmetry requires a null model; the most accepted to date is the Markov (random branching) model, which assumes that speciation is instantaneous and that all species have an equal and independent chance of speciation. Such assumptions result in a high probability of asymmetric trees (which makes tests of asymmetry weak) and may be unrealistic. Subsequent modification of this model incorporated parameters conforming to diversification by peripheral-isolates speciation, producing trees that were more symmetric than those generated under the Markov model. Our reanalysis of these data, and remodeling with a greater number of replicates, shows that differences between the original Markov and modified models may have been exaggerated: different measures show that initial increases in tree symmetry under peripheral-isolates speciation are transient; symmetry on small scales actually decreases when the process of speciation requires a significant fraction of the average time between speciation events. The peripheral-isolates mode of speciation is com pared with a vicariant mode, which shows steadily increasing symmetry on all scales. In cases where vicariant speciation has been common, use of a vicariant-based null model in tests of phylogenetic asymmetry is appropriate and will provide greater statistical power.