Unraveling the Drivers and Evolutionary Trajectory of the Flood Pulse in the Middle‐Lower Yangtze River and Its Ecological Effects on Fish

ABSTRACT Flood pulses are critical drivers of river ecosystems. However, accurately quantifying the distinct contributions of climate change and anthropogenic activities to flood pulse evolution remains a challenge. Distinct from traditional approaches, this study establishes a novel framework coupling the environmental flow components (EFCs) system with a long short‐term memory (LSTM) network to decouple nonlinear drivers in the Middle‐Lower Yangtze River (MLYR) (1965–2020). This study, utilizing flow data from four representative hydrological stations along the MLYR between 1965 and 2020, integrated ecological hydrology indicators, statistical analysis, and LSTM networks to examine the multiscale evolution of flood pulses and identify their underlying driving mechanisms. The findings demonstrated that dam operations significantly reduced both the intensity and frequency of flood pulses, with peak flow discharges decreasing by 36.9%–76.1% and flood duration shortening by 29.2%–75.7%. Attribution analysis, supported by LSTM‐based natural flow reconstruction, revealed a distinct spatial heterogeneity: while general flood metrics at downstream stations were primarily driven by climatic factors (> 70%), dam regulation was the dominant driver for the reduction of high‐intensity pulse indicators (e.g., peak flow and rise rate) in the upstream reaches. Human interventions have triggered cascading ecological consequences, including disrupted fish reproduction and degradation of floodplain functions. Fish larvae discharge declined by 65.7%, with flood pulse occurrence frequency and rising rate identified as the most influential factors. This research enhances our understanding of flood pulse dynamics and their ecological significance, providing a scientific foundation for integrated watershed management and ecological restoration strategies.