Protection of freshwater fish diversity is a global conservation priority in face of its alarming decline in the last decades. A crucial step to protect freshwater fish diversity is the production of prompt and precise evaluation of community composition and spatial distribution. Metabarcoding of environmental DNA (eDNA metabarcoding) generally surpasses traditional methods for documenting diversity and community composition in aquatic environments. Nevertheless, empirical evidence evaluating how eDNA transportation in water affect community composition and structure via eDNA metabarcoding data remains scarce. Using a brown trout (Salmo trutta) cage transplant experiment in the St. Lawrence River (Canada), a large fluvial system, we tested the detection and relative abundance of species’ eDNA along 15 sampling locations. We detected brown trout eDNA in five localities up to 5,000 m from the cage, but only one sampling location situated 10 m downstream and in the direct line of the cage was affected at the community composition level. This locality showed a relative abundance of brown trout eDNA of 13.1%, while the four others showed a relative abundance under 1.0%. K-means cluster analysis confirmed the impact of brown trout eDNA on community composition by separating this locality from all others. Based on species loading of a redundancy analysis, we showed that this different k-means group was associated with the high relative abundance of brown trout. No evidence of transport effect of brown trout eDNA on fish community composition was observed in any other sampling locations. Together, our results support the view that eDNA metabarcoding can be both a conveyor belt of biodiversity information and a precise tool to study the composition and structure of fish communities in river.