Molecular species delimitation is commonly performed by extrapolating differentiation patterns from small regions to the entire genome. Here we compare different data sources (mitochondrial COI sequences and nuclear and mitochondrial SNPs) for the marine bivalve species complex Aequiyoldia eightsii in South America and Antarctica. Whilst all data suggest that populations on either side of the Drake Passage belong to different species, the picture is less clear within Antarctic populations which harbor three distinct mitochondrial lineages (p-dist ˜ 6 %) that coexist in populations as well as in a subset of individuals. Haplotype-specific amplification and mitochondrial SNPs data demonstrate that two haplotypes occur in heteroplasmy in some individuals (h1h3 and h2h3). Standard barcoding procedures using universal primers lead to amplification bias favoring either haplotype unpredictably and thus overestimate the species richness with high confidence. In nuclear SNPs, no differentiation akin to the trans-Drake comparison can be observed inside the Southern Ocean, suggesting that 1) the Antarctic populations represent a single species and 2) their mitochondrial haplotypes h1, h2, h3 evolved during periods of temporary allopatry and survived as non-recombining mitochondrial haplotypes whereas recombination eroded similar differentiation patterns in the nuclear genome after secondary contact. Mitochondrial heteroplasmy in combination with amplification bias are possibly more common problems in molecular barcoding but remain undetected because evidence for it is eliminated during quality checking sequencing results. As a countermeasure, an active search in electropherograms for secondary peaks and reads with alternative bases in Illumina data as well as haplotype-specific primers for amplification are recommended.