Quality control of fighting fish nucleotide sequences in public repositories reveals a dark matter of systematic taxonomic implication.

Background: The number of nucleotide sequences in public repositories has exploded recently. However, the data contain errors, leading to incorrect species identification. Several fighting fish (Betta spp.) are poorly described, with unresolved cryptic species complexes masking undescribed species. Here, DNA barcoding was used to detect erroneous sequences in public repositories.

Objective: This study reflects the current quantitative and qualitative status of DNA barcoding in fighting fish and provides a rapid and reliable identification tool.

Methods: A total of 1034 barcode sequences were analyzed from mitochondrial cytochrome c oxidase I (COI) and cytochrome b (Cytb) genes from 71 fighting fish species.

Results: The nearest neighbor test showed the highest percentage of intraspecific nearest neighbors at 93.41% for COI and 91.67% for Cytb, which can be used as reference barcodes for certain taxa. Intraspecific variation was usually less than 13%, while most species differed by more than 54%. The barcoding gap, calculated from the difference between inter- and intraspecific sequence divergences, was negative in the COI data set indicating overlapping intra- and interspecific sequence divergence. Sequence saturation was observed in the Cytb data set but not in the COI data set.

Conclusion: The COI gene should thus be used as the main barcoding marker for fighting fish.