Highly polymorphic mitochondrial DNA and deceiving haplotypic differentiation: implications for assessing population genetic differentiation and connectivity

BACKGROUND: Hyperdiverse mtDNA with more than 5% of variable synonymous nucleotide sites can lead to erroneous interpretations of population genetic differentiation patterns and parameters (φ(ST), D(EST)). We illustrate this by using hyperdiverse mtDNA markers to infer population genetic differentiation and connectivity in Melarhaphe neritoides, a NE Atlantic (NEA) gastropod with a high dispersal potential. We also provide a recent literature example of how mtDNA hyperdiversity may have misguided the interpretation of genetic connectivity in the crab Opecarcinus hypostegus. RESULTS: mtDNA variation surveyed throughout the NEA showed that nearly all M. neritoides specimens had haplotypes private to populations, suggesting at first glance a lack of gene flow and thus a strong population genetic differentiation. Yet, the bush-like haplotype network, though visually misleading, showed no signs of phylogeographic or other haplotype structuring. Coalescent-based gene flow estimates were high throughout the NEA, irrespective of whether or not mtDNA hyperdiversity was reduced by removing hypervariable sites. CONCLUSIONS: Melarhaphe neritoides seems to be panmictic over the entire NEA, which is consistent with its long-lived pelagic larval stage. With hyperdiverse mtDNA, the apparent lack of shared haplotypes among populations does not necessarily reflect a lack of gene flow and/or population genetic differentiation by fixation of alternative haplotypes (D(EST) ≈ 1 does not a fortiori imply φ(ST) ≈ 1), but may be due to (1) a too low sampling effort to detect shared haplotypes and/or (2) a very high mutation rate that may conceal the signal of gene flow. Hyperdiverse mtDNA can be used to assess connectivity by coalescent-based methods. Yet, the combined use of φ(ST) and D(EST) can provide a reasonable inference of connectivity patterns from hyperdiverse mtDNA, too.