We can’t change who they are: integrating public and new COI, NADH2 and 12S sequences to build reference libraries for cartilaginous fish species

The number of threatened cartilaginous fishes has grown at an alarming rate in the last decades. Accurate identification of species becomes a critical aspect to assess and plan accurate conservation and management measures for this important group. In the past, the identification was based on morphological methods that have been progressively integrated with molecular tools. Therefore, it becomes extremely important to maximise the sensitivity of molecular approaches improving the reliability and the accuracy of the public databases. The present work aims to a) evaluate the robustness of the public databases NCBI and BOLD for the discrimination of Elasmobranchii and Holocephali species recorded in five FAO areas and b) to produce curated reference libraries for three of the most widely used markers on cartilaginous fishes (COI, NADH2 and 12S). To reach this scope over 28000 sequences belonging to over 400 putative species were analysed, including both public and newly produced data. The results obtained revealed the presence of errors in public repositories such as: a) mislabelling and/or misidentification and b) old nomenclature and/or data not updated. The results outlined: c) presence of possible new species not yet described and d) clades which require further analysis to solve taxonomic uncertainties. COI and NADH2 showed a higher number of total sequences and a similar number of species recognizable in final datasets, suggesting the high potential for these two markers for the molecular taxonomic, particularly when compared to 12S. Additionally, the results highlighted how the discriminant power at the species level decreases using short vs long 12S sequences (usually used in metabarcoding/eDNA and phylogenetic analyses, respectively). In fact, using the shorter sequences the possibility to discriminate many species that were clearly divided using longer 12S sequences was often lost. Overall, our results highlight the 12S current limits, mainly linked to its low species coverage rate. Contrarily, both COI and NADH2 proved to be useful and valid tools, showing a solid species coverage. Anyway, further work is needed to complete and improve the reference libraries. In particular, the obtained results suggest a careful scrutiny and manual corrections of publicly obtained sequences by experts prior to their use.