Sampling Error Does Not Invalidate the Yule-Coalescent Model for Species Delimitation. A Response to Lohse (2009)

Lohse (2009) used a simulation study to argue that the generalized mixed Yule-coalescent (GMYC) model (Pons et al. 2006; Fontaneto et al. 2007) may overestimate species numbers. He found that incomplete sampling of demes involved in the coalescence process could artificially produce clusters that are recognized as separate GMYC groups (species). The paper also criticizes our (Papadopoulou et al. 2008) simulations of the coalescent process where we found that GMYC groups are readily formed when migration among demes drops below a particular level (Nm < 0.01). We interpreted these results to indicate that divergent sequence clusters form under conditions of stringent population isolation and that these clusters resemble those widely seen in empirical data from mitochondrial DNA (mtDNA) sampled across multiple populations and species. Lohse's (2009) simulations confirmed our findings but warned that additional GMYC groups are recognizable when less than about 20% of all demes are sampled. Although this is a valid point, Lohse (2009) extrapolated these findings to dismiss the utility of DNA-based approaches to species delimitation, saying that real-world samples will be composed of “essentially random clusters.” We argue that these conclusions go well beyond the simulation results from partially isolated populations within a single species and that the dismissal of the GMYC model as the basis for delineating entities in taxonomic research is unjustified.