Results
In the tank experiment, regardless of BAC treatment, we observed biphasic exponential degradation of all types of Japanese jack mackerel eDNA (Fig. 2; Table 3). All types of eDNA concentrations were higher in the treatment with BAC addition than in those without BAC at time 0, which lasted throughout the sampling period. The decay rates at the initial phase (\(k_{1}\)) were substantially lower in the treatment with BAC addition (31.0 to 53.0% relative to the treatment without BAC), while those at the following slower phase (\(k_{2}\)) were not significantly different between BAC treatments (Fig. 3). In contrast, in field sampling, we observed monophasic exponential degradation of shorter fragments of eDNA (Fig. 4). Linear models showed a significant interaction between sampling time points and BAC treatments for nuS, indicating that eDNA decay rates were significantly lower in the treatment with BAC than in those without BAC (P < 0.05; Table S3). Although we did not confirm a significant interaction for mtS, target eDNA was detected for 24 hours from the seawater samples with BAC addition, whereas it was hardly detected in the samples without BAC addition (Fig. 4). We did not evaluate the effect of BAC addition on the degradation of longer eDNA fragments (mtL and nuL) because of their poor detection relative to that of shorter eDNA fragments. The overall PCR efficiencies and R2 values of the standard curves are shown in Table S4. A few filtration negative controls in the tank experiment showed PCR amplification, whose concentrations were less than one copy per PCR and much less than those of the sampling tank at the corresponding time points. No amplification was observed in any of the PCR-negative controls throughout the study.
Moreover, the number of fish species detected by eDNA metabarcoding was higher in the treatment with BAC over time (Fig. 5a). In total, 65 marine and brackish fish were detected in 18 of 1-L seawater samples, wherein 58 and 45 species were detected in the samples with and without BAC addition, respectively; 36, 40, and 38 species were detected in samples with BAC, whereas 36, 28, and 27 species were detected in samples without BAC at time 0, 6, and 24, respectively, when sampling triplicates were pooled (Table 4; the number of detected species per sample is shown in Table S5). Exact McNemer tests showed significant differences in the number of fish species between BAC treatments at time 6 and 24 (both P < 0.05), while no statistical difference was observed at time 0 (P = 1.00). In addition, PERMANOVA tests showed a significant difference in community composition between BAC treatments (P < 0.05) but not between time points (P = 0.79) (Fig. 5b). We additionally confirmed that the variances of the compositions were not statistically different among treatments (PERMDISP; both P > 0.1). After preprocessing the iSeq raw reads and removing potential contaminations, none of the eDNA reads were detected from all filtration and PCR negative controls (Table S6). All the rarefaction curves, generated byrarecurve function in the package ‘vegan’, showed that the number of species detected from each sample was saturated and the library sample was satisfactorily sequenced (Fig. S1).