High-density environmental DNA metabarcoding reveals fine-scale three-dimensional structure of a fish community in a semi-closed bay

Coastal fish assemblages are influenced by environmental heterogeneity across horizontal and vertical dimensions, yet their fine-scale three-dimensional structure remains poorly understood. We investigated spatial variation in fish assemblages using environmental DNA (eDNA) metabarcoding at 100 stations across three depth layers (300 samples) in Maizuru Bay, a semi-enclosed coastal bay in Japan. Concurrent hydrographic measurements were used to examine associations between fish assemblage composition and environmental conditions. We detected 216 fish species, with mean species richness decreasing from 28.4 at the surface to 16.4 at the bottom. Thirty-three species were detected exclusively below the surface. Fish assemblage composition differed significantly among depth layers and exhibited pronounced horizontal turnover along the bay axis. Although depth, geographic position, and measured environmental variables were associated with assemblage composition, they collectively explained only 13.0% of the variation. The independent contribution of spatial position (2.9%) exceeded those of depth (0.8%) and measured environmental variables (0.9%). These findings demonstrate substantial three-dimensional heterogeneity in coastal fish assemblages that cannot be fully explained by conventional hydrographic measurements. Our study highlights the importance of integrating horizontal and vertical sampling to characterize coastal fish biodiversity and improve marine ecosystem monitoring.