Comprehensive Assessment of Fish Diversity Based on Surface Water and Sediment eDNA—A Case from Sansha Bay, Southeastern China

Shansha Bay, southeastern China, is a typical semi-enclosed bay facing intense anthropogenic pressure, with mariculture production reaching approximately 660,000 tons annually, predominantly driven by the large yellow croaker (Larimichthys crocea) industry. To assess how these pressures influence biodiversity detection, we employed eDNA metabarcoding targeting the 12S rRNA gene to survey fish diversity across 10 sites, integrating both surface water and sediment matrices. A total of 40 species belonging to 16 orders, 27 families and 32 genera, were identified. Among them, L. crocea, Bostrychus sinensis, Clupanodon thrissa and Siganus fuscescens are the dominate species, which comprised over half of all OTUs. Comparison with historical trawl surveys revealed 30 newly recorded species, including the Japanese eel (Anguilla japonica) and oarfish (Regalecus glesne). Alpha diversity was low (Shannon: 0.69–1.93), with significant spatial disparities driven by freshwater runoff and hydrodynamics. While eDNA effectively captured the dominance of cultured species like L. crocea, the extreme biomass signal poses challenges for detecting wild populations. Our findings underscore that a multi-matrix eDNA approach is crucial for accurate biodiversity assessments in aquaculture-dominated coastal ecosystems, although distinguishing wild stocks from aquaculture-derived DNA remains a critical methodological hurdle.