Land use characteristics affect the sub-basinal scale urban fish community identified by environmental DNA metabarcoding

The heterogeneity of urban landscapes has effects on the environmental characteristics and fish composition of individual urban streams, even within a single water system. It is, therefore, imperative to assess the influence of physiochemical properties on urban streams by analyzing the spatial distribution of fish communities at the local scale. However, conventional fish surveys encounter time and labor constraints when selecting and surveying dense sampling points under 2 km in stream networks. In this study, environmental DNA (eDNA) metabarcoding was used as an innovative survey methodology to identify the effects of land use and stream order on fish composition and tolerance guild in an urban area. The eDNA sampling was conducted in 31 sites of the Anyang stream network in Korea, including part of the stream undergoing ecological restoration. The eDNA survey detected 12 of 17 species (70.6%) that appeared in the historical data, and 12 of 18 species (66.7%) identified in a conventional field survey with kick nets and casting nets. The proportions of urban area, forest and grassland were positively correlated with abundance (p < 0.05) and richness (p < 0.05) in multiple regression analyses, while the proportion of agricultural area showed a negative correlation (p < 0.05). For abundance, richness, and diversity within the fish community from first- to third-order streams, there was a significant decrease in sensitive species (p < 0.05) alongside a significant increase in tolerant species (p < 0.01) across all three indices. The results of this study highlight variations in fish composition across sites within the local scale of the urban stream network, underscoring the need for detailed monitoring to understand the ecological function of urban streams.