Examining the sensitivity of molecular species delimitations to the choice of mitochondrial marker

Abstract

Defining and understanding species diversity forms the basis of a wide range of biological and conservation work. Traditional taxonomy can be complemented and accelerated using molecular methods of species delimitation, such as the widely used Generalised Mixed Yule-Coalescent (GMYC) approach. This method uses time-calibrated phylogenetic trees in order to identify transition points between inter- and intraspecific divergence processes. Despite some important limitations, the GMYC approach appears to be robust to a wide range of dataset characteristics. It is one of the few model-based species-delimitation methods that remain practical for analysing molecular datasets with a large numbers of taxa. Most GMYC analyses have been based on datasets consisting of one or a small number of mitochondrial genes. To investigate the sensitivity of GMYC to the choice of mitochondrial marker, we compared GMYC estimates from 15 mitochondrial genes for three vertebrate datasets (cetaceans, ursids and whitefish). Despite the shared evolutionary history among mitochondrial genes, different markers exhibited substantial variation in GMYC delimitation results across all three datasets. This variability was not restricted to specific genes or taxa and extended to commonly used barcoding genes such as COI and CYTB. Using multiple concatenated markers mitigated these problems in two of the datasets, but exacerbated systematic biases present in a third. Our findings indicate the need to consider multiple markers, loci and lines of evidence when performing molecular species delimitation.