DNA metabarcoding improves the taxonomic resolution of visually determined diet composition of beaked redfish (Sebastes sp.)

Beaked redfishes (Sebastes fasciatus and Sebastes mentella) of the northwest Atlantic have recently reached record abundance levels in the estuary and the northern Gulf of St. Lawrence, dominated by Sebastes mentella. Knowledge of their diet composition is essential to understand the trophic role that these groundfish play in the ecosystem. The objective of the present study was to compare the performance of visual examination and DNA metabarcoding of stomach contents of the same individual redfish caught in the lower Estuary and northern Gulf of St. Lawrence. Using a universal metazoan mitochondrial cytochrome c oxidase subunit I (COI) marker, we identified a total of 24 taxonomic groups, composed of 22 species and two genera in the content of 185 stomachs with DNA metabarcoding. We compared these results to the 25 prey types, eight identified at the genus and nine at the species level, obtained with visual stomach content analysis (SCA). While both techniques revealed a similar diet composition, our results showed that the SCA and DNA metabarcoding perform differently for particular prey categories, both in terms of detectability and taxonomic resolution, as well as in the estimated relative importance of weight and occurrence in the diet. The use of DNA metabarcoding along with SCA validates and improves the taxonomic resolution of visually determined prey, which supports the concept that both techniques provide useful complementary information on the diet of these redfish species.

Methods
Redfish stomachs were thawed for visual analysis at the Maurice-Lamontagne Institute (Mont-Joli, Canada) and each prey taxon found in the stomach contents was weighed and identified to the lowest taxonomic level possible under a dissecting microscope using keys and identification guides. For metabarcoding, we annotated each unique read of each sample to the most similar species of the database if the sequences had at least 98% similarity. In cases of ties, referred to as multiple hits, when more than one species from the database had the same distance from the queried sequence, all the tied annotations were reported. Barque then produced a table of prey taxa counts (including the multiple hits) for each sample.