The process of species discovery and delimitation can be aided by the application of computational methods to genetic data. Molecular methods for species delimitation have the potential to speed up the progress of traditional taxonomy, particularly in cases where cryptic species are believed to be present.

Among the molecular methods for species delimitation, those that use a phylogenetic approach have been shown to outperform techniques based on pairwise genetic distances. At present, phylogenetic species delimitation is usually based on a single locus. This can be problematic because trees inferred from different genes can be mutually incompatible, even when there is complete linkage.

The Generalised Mixed Yule-Coalescent (GMYC) is a widely used approach to computational species delimitation. The GMYC treats the phylogenetic tree as the product of a mixture of speciation (between species) and coalescent (within species) processes. Using maximum likelihood, the method estimates the point in time that separates these two processes, yielding an estimate of the number of distinct evolutionary units or putative ‘species’ in the data set. Although the GMYC appears to exhibit robust performance on a variety of datasets, its sensitivity to the choice of genetic marker has not been evaluated.

Using a case study based on cetaceans (whales, dolphins, and porpoises), we examined the behavior of the GMYC when applied to gene trees inferred from each of the major protein-coding and ribosomal RNA genes in the mitochondrial genome. Our results indicate that the results of species delimitation can vary across different mitochondrial markers. Further development of the GMYC method should consider how the procedure can be extended to accommodate multilocus data while retaining its theoretical and computational advantages.