Identification of ornamental fishes for biosecurity

Abstract

Introduction: Poorly regulated international trade in ornamental fishes poses risks to both biodiversity and economic activity via invasive alien species and exotic pathogens. Border security officials need robust tools to confirm identifications, often requiring hard-to-obtain taxonomic literature and expertise. DNA barcoding offers a potentially attractive tool for quarantine inspection, but has yet to be scrutinised for many fishes in the aquarium trade. This research examines a DNA barcoding approach for ornamental cyprinid fishes (Teleostei: Cypriniformes), an important group in terms of biosecurity risk.

Methodology and results: A reference library of fishes purchased from the international aquarium trade was assembled, and the specimens were identified to species using morphological characters derived from taxonomic literature. Many species were found to be misidentified in the trade. DNA barcodes were then generated using standardised protocols, and the efficacy of the reference library in making species level identifications was assessed. A total of 172 ornamental cyprinid fish species were sampled, providing baseline molecular data for 91 species currently unrepresented in public reference libraries. DNA barcodes were found to be highly congruent with the morphological assignments, with identification success rates of up to 99%. The cyprinid fish dataset was augmented with sequences from GenBank for an additional 157 species, the benefit of which was additionally evaluated. Here, it was observed that the inclusion of GenBank data resulted in a more comprehensive library, but at a cost to success rate due to the increased number of singleton species. Identification success rates are known to be sensitive to the choice of identification criterion, and because this may be important for biosecurity applications, a specific focus of this research was to assess these procedures. Here, a variety of different techniques were applied (neighbour-joining monophyly, bootstrap, nearest neighbour, GMYC, percent threshold), and it was found that identification success rates varied between 87% and 99%, according to the method used. The appropriateness of the commonly employed Kimura two-parameter (K2P) model was also examined using an information-theoretic model-selection approach. Despite its ubiquity in the DNA barcooding literature, the K2P model was not found to be well supported as an appropriate substitution model at the species level. However, using this model did not affect identification success rates overall. Standard DNA barcoding techniques are known to be inappropriate and potentially misleading in situations where interspecific hybridisation has occurred. Similarly, where cryptic species are suspected, mitochondrial DNA is sometimes insufficient to robustly recognise lineages. As both of these situations are believed to occur in the ornamental fish trade, and using a genomic dataset, a range of candidate nuclear loci were assessed as a complementary marker to COI. The rhodopsin gene was shown to be variable between closely related species, and with 200 sequences from cyprinid fishes, interspecific hybridisation events were confirmed, and unrecognised diversity was highlighted within popular aquarium species. Traces of degraded environmental DNA present in water can now be used to detect the presence of aquatic species, so diagnostic tests for fish identification were investigated with the aim of developing a new, more efficient biosecurity quarantine tool. The COI barcode library was mined for informative short-length markers using a sliding window analysis of variation through the gene. Species-specific DNA sequences were successfully amplified from aquarium water samples, and at relatively low densities of the target species.

Conclusions: This study demonstrates that DNA barcoding can provide a highly effective biosecurity tool for rapidly identifying ornamental fishes. In the small number of cases where DNA barcodes are unable to offer a species level identification, previous studies are improved upon by consolidating supplementary information from multiple data sources in the form of specimen images, morphological characters, taxonomic bibliography, and preserved voucher material. Reference libraries can be utilised to develop new diagnostic approaches using environmental DNA, allowing quarantine facilities to capitalise on non-invasive techniques for detecting high-risk fishes. The biggest obstacles, however, to an operational implementation of DNA barcoding and any future expansions of the reference libraries, are the combined problems of misidentification of reference specimens between labs, and a lack of access to appropriate taxonomic literature to first identify the fishes. If these problems are not addressed by the barcoding and taxonomic communities respectively, this will ultimately compromise the ability of biosecurity agencies to use a DNA barcoding tool.