Shortcomings of DNA barcodes: a perspective from the mammal fauna of Switzerland

The use of DNA barcodes is a popular approach to identify unknown environmental samples based on reference DNA databases. However, as shown previously, the success of reaching accurate species-level identification will not only depend on the taxonomic group under interest, but also on the availability of reliable and comprehensive reference databases. We applied this methodology for a sampling of vouchered specimens of wild mammals of Switzerland and demonstrated that reliable species identification with standard DNA barcodes in this reputedly well-known taxon may be challenging. The overall success of unambiguous species-level identifications with help of three commonly used mitochondrial markers barely reached 70 % in unsupervised queries submitted to BOLD or GenBank. Pitfalls were mostly due to misidentified or mislabelled sequences available in public databases, to the presence of highly divergent cryptic lineages or to missing reference sequences for the rarest species (i.e., those that are in greatest need for conservation attention). Divergent cryptic lineages (over 5 % sequence divergence) found in Swiss mammals were either due to highly distinct intraspecific haplotypes or to the existence of an overlooked cryptic biological species (Muscardinus speciosus). To circumvent part of the observed pitfalls, we provide a curated and complete reference database for the wild mammals of Switzerland based on carefully identified, vouchered specimens. We finally acknowledge that the identification of naturally hybridizing or introgressed species remains a significant challenge that is often overlooked in massive, unsupervised DNA barcode analyses.