ABSTRACT Accelerating anthropogenic activity in watersheds is reshaping reservoir ecosystems and threatening their ecological integrity and services, yet the scarcity of long‐term records limits our understanding of how ecosystems respond to cumulative watershed disturbance. Here we used microeukaryotic sedimentary DNA (sedDNA) to reconstruct four decades (1981–2019) of watershed disturbance and ecological change in Duihekou Reservoir. Consistent with a concurrent shift in watershed disturbance, microeukaryotic communities underwent a clear state shift around 2000, from a moderately disturbed, low‐nutrient state to a human‐impacted, nutrient‐enriched mesotrophic regime. This shift favoured mixotrophic and phagotrophic taxa as “winners”, potentially threatening water quality and higher trophic levels. At the same time, co‐occurrence networks reorganized from a fungus‐centered web into a denser, multi‐core architecture with higher robustness metrics, motivating the hypothesis that recovery may face a higher barrier once the mesotrophic state is established. Community assembly also became increasingly deterministic, with watershed human activities emerging as the strongest correlates of succession and the regime shift. Using random forest algorithms, we developed microeukaryotic sedDNA indicator models to reconstruct watershed disturbance and clarified their applicability and limitations for hindcasting past watershed conditions. Together, these findings demonstrate that microeukaryotic sedDNA archives can serve as sensitive, integrative sentinels of cumulative watershed disturbance and reservoir regime shifts, providing a basis for applying sedDNA‐based indicator models in long‐term monitoring and adaptive management.
Sun et al. (Tue,) studied this question.