Reservoir proximity explains long-term transformation of riverine fish assemblage structure and function

Dams and their associated reservoirs have become so ubiquitous on the world's riverscapes that only 23% of rivers now flow freely to the ocean. These artificial structures have known ecological consequences, but ecological studies often suffer from a lack of historical baseline data and uncertainty regarding the degree to which concepts are transferable among altered river systems. We reviewed historical fish assemblage survey data collected over 373 km of the upper Sabine River, Texas, USA during 1954-1955, prior to construction of large reservoirs at the upstream and downstream extents of the study area. We then repeated surveys using identical methods in 2023 after reservoirs were in place for multiple decades. The resulting dataset provided opportunity to measure impoundment-driven deviations from historical baseline conditions and test a suite of hypotheses centered on fish assemblage changes across a gradient of proximities (i.e., distances) from reservoirs. We found support for the proximity replacement hypothesis in which fish assemblages nearest to reservoirs experience the highest temporal beta diversity; support for the longitudinal recovery gradient hypothesis in which relative abundance of periodic life history strategists returns to a natural baseline with greater downstream distance from dam tailwaters; support for the proximity host loss hypothesis in which fishes that serve as hosts to Unionid mussels decline in reservoir tailwaters; and support for the proximity host gain hypothesis in which fishes that serve as hosts to Unionid mussels increase in the river-reservoir interface upstream of a dam. This work advances knowledge of ecological consequences associated with dam construction by revealing that concepts developed using space-for-time substitutions (i.e., without historical baseline information) remain pertinent when tested against historical benchmarks and these same concepts are applicable to unstudied systems.