Comparison of fish assemblages across monitoring methods
The three approaches (dDNA, deep survey, and the 2022 WFD campaign) collectively revealed a total of 91 fish species. Dietary DNA detected nearly as many species than the 10-day deep survey (68 and 70 respectively) and three times more species than the WFD campaign. Biosampler dDNA provided a fish assemblage that was broadly consistent to the deep inventory, with almost two third species shared (n=51 species) (Figure 2). However, it also detected 17 species not recorded with traditional techniques despite the use of multiple fishing techniques. Conversely, 19 species recorded in the deep inventory were not detected in shrimp digestive contents. When excluding unique species absent from the reference databases, fish not detected by dDNA metabarcoding were significantly less abundant in field surveys, with an average of 7.3 (± 8.7) catches, compared to 21.0 (± 12.5) catches for detected species (Wilcoxon test, p = 1e-4). Additionnally, species detected exclusively through metabarcoding (n = 16) generally showed significantly lower frequencies of occurrence (mean FO = 3.27%, SD = 2.8%) compared to species detected through both metabarcoding and deep survey (mean FO, 8.01%, SD = 6.81%) (Wilcoxon rank-sum test, p-value = 0.0052).
The Venn diagram revealed limited overlap between species identified by the 2022 WFD survey and those detected through dDNA or during the deep inventory, with only 15 and 17 species shared, respectively. Additionally, fish detected during the 2022 WFD campaign were significantly larger (201.4 ± 104.8 mm) than those capture by dDNA metabarcoding (145.9 ± 129.3 mm, Wilcoxon test, p = 0.003 and exhibited a more restricted size range (Figure 3). In contrast, no significant difference in expected size distribution was observed between dDNA and deep inventory fish assemblages (Wilcoxon test, p = 0.46)