3.5 Relationships between environmental factors and fish community composition
The correlation analysis between environmental factors (see Table S1 in the Supplementary Material) and the composition of fish communities showed that 44 (Fig. 6A), 41 (Fig. 6B), and 40 (Fig. 6C) species had significant relationships with environmental factors according to the eDNA, individual number, and biomass data types, respectively. In addition, for eDNA-based fish community composition, at the order, family, genus, and species levels, there were 5, 18, 41, and 44 units, respectively, that significantly (P < 0.05) correlated with the environmental factors (see Fig. S1 in the Supplementary Material). This result indicated that the species-level identification of the fish community compositions had the strongest response to the environmental changes, followed by genus-level identification. Interestingly, regardless of input data types, there were two environmental gradients that significantly influenced the longitudinal distribution of fish communities. The first environmental gradient was mostly composed of physical water parameters and habitat factors, including elevation (m), dissolved oxygen (mg/L), flow velocity (m/s), riffle area (%), and vegetation cover (%). The high values of these factors indicated pristine habitats in the headwaters and upper reaches, where there was less human disturbance. The second environmental gradient was mostly composed of the chemical water parameters (e.g., high CODMn [mg/L] and BOD5 [mg/L]), conductivity (μs/cm), bacteria levels (total number of bacteria and coliform), and heavy metals (e.g., Hg, As, and Cu [mg/L]). The high values of these factors indicated water pollution and extraneous interference that commonly occurred in urban and industrial areas of the lower reaches. These results indicated that eDNA-based analysis of fish community composition could reflect the longitudinal changes in environmental factors along the river, which was similar to the pattern reflected by individual number and biomass.