Fish community monitoring in floodplain lakes: eDNA metabarcoding and traditional sampling revealed inconsistent fish community composition

• We recorded significant methodological differences in monitoring fish faunas. • Fewer eDNA samples can represent an adequate exploration of fish richness. • eDNA recorded significantly lower Shannon-wiener index in low-water phase. • eDNA recorded significantly lower β SOR and β SIM in high-water phase. Fish, acting as the dominant vertebrates in aquatic ecosystems, exhibit strong mobility and complex activity patterns, which make it challenging for the quantitative assessment of fish communities. Previous fieldwork has applied multiple fish sampling devices as well as eDNA technology respectively, with limited studies comparing the reliability, representativity, and applicability of eDNA metabarcoding and traditional sampling methods in investigating fish assemblages across spatiotemporal scales. In this study, we selected Lake Huang (a provincial nature reserve) to compare the eDNA technology and traditional sampling methods (multi-mesh gillnets and accordion-shaped traps) in revealing differences in species composition, α diversity, and β diversity of fish assemblages in both high-water phase and low-water phase. Results showed that eDNA technology detected more fish species and required fewer samples to achieve sufficient sampling compared to traditional methods. Specifically, our results demonstrated significant methodological variations in species composition, with both traditional sampling and eDNA approaches revealing the method-dependent species checklist. In the high-water phase, the eDNA approach recorded significantly lower β diversity (β SOR and β SIM ) than traditional sampling methods, which may result from the more homogeneous eDNA water samples triggered by flooding, resulting in lower between-sites variations in fish communities. However, in the low-water phase, eDNA technology reported significantly higher β diversity (β SOR and β SNE ) than traditional sampling methods, potentially attributed to the loss of eDNA information as well as heterogeneous eDNA samples accompanied by water receding. In addition, taking advantage of the separated and narrowed habitat in the low-water phase, fish individuals were more easily captured, resulting in a relatively higher Shannon-Wiener index recorded by traditional sampling methods. Based on inconsistency in species composition characterized by two methods, our results exhibited that eDNA technology is more efficient in detecting species, while there was an urgent need to design more sensitive primers, provide more complete reference databases, and popularize more emerging sequencing platforms for the advancement of eDNA technology. Comparing with the fish assemblage diversity identified by traditional sampling methods, our result exhibited the transport of DNA fragments may potentially enlarge the homogenized or heterogeneous fish community characterization influenced by flood pulses.