Jiawang Zhao, Wanqing Chi, Fuwei Zhang, Xiaona Zhang, Liqiang Yang
Agricultural herbicides are increasingly recognized as pervasive stressors in coastal ecosystems, yet a fundamental question remains unresolved: how temporally variable land-based inputs propagate across the land-river-sea continuum to generate ecologically meaningful thresholds in marine systems. Here, we develop a process-resolved framework that couples the Soil and Water Assessment Tool (SWAT) with coastal hydrodynamics model MIKE21, constrained by field observations and experimental validation, to quantify the transport and ecological impacts of atrazine in the Yellow River-Bohai Sea system. We show that more than 91.1% of annual atrazine export occurred during a concentrated high-export period from May to August, generating episodic riverine inputs that were subsequently transformed into more persistent nearshore exposure by coastal retention processes. Integrating phytoplankton community composition, chlorophyll a inhibition, and in situ ¹ ³C-based primary productivity measurements, we identify a narrow early-warning threshold window (0.18-0.33 nmol L⁻¹) and a conservative management threshold at the NOEC of 0.50 nmol L⁻¹ , based on the highest concentration without statistically significant functional impairment (p > 0.05). Notably, functional responses are more sensitive than structural changes, indicating that ecosystem metabolism responds earlier than community reorganization. Functional impairment preceded detectable community reorganization, consistent with the rapid suppression of photosynthetic carbon fixation and the longer timescale required for population turnover. In the semi-enclosed Bohai Sea, hydrodynamic retention transformed concentrated Yellow River inputs into persistent nearshore exposure. Linking the system-specific ecological response boundary to modeled upstream fluxes yielded a Yellow River-specific atrazine load target of ≤ 1.74 t yr⁻¹ , providing a quantitative basis for managing pulsed land-to-sea inputs.