Fish species detection by environmental DNA metabarcoding by three surface sampling methods

Environmental DNA (eDNA) is rapidly gaining support as a tool for studying fish distributions and monitoring biodiversity. Despite its apparent ease of application, the results of metabarcoding techniques are known to be sensitive to water sampling methods. Besides, ship-bottom intake sampling holds promise over more conventional methods due to continuous collection. Thus, we assessed the species-level eDNA metabarcoding detection results among three frequently used surface sampling methods: intake from ship-bottom (4.5 m below the surface, via the ship’s faucet), bucket (0 m, dropped over the side), and Niskin bottle (5 or 10 m below the surface) from 99 stations during seven research cruises. A total of 321 species were recorded, of which 181 were common to all three methods. All three methods commonly revealed higher species detection rates at warmer temperatures in more offshore regions. Species-specific detection performance (presence/absence) showed good agreement between the three methods, except for Cololabis saira . The relative read abundance (RRA) of many frequently detected species exhibited similar tendencies among sampling methods, but there were some exceptions including Trachurus japonicus and Cololabis saira which showed a better RRA in bucket samples. Nine of the top 15 frequently detected species had significant correlations between RRA of different sampling methods, while the remaining six species displayed biases in RRA among different sampling methods. Most of the biases were likely influenced by environmental conditions that can alter the vertical distribution of the eDNA or the positioning of the fish themselves, rather than stemmed from the methods.