The Effects of Agricultural Activities, Hydrology and Climate on Water Quality in Lake Dongting, China

Lake water quality is heavily influenced by agricultural activities, hydrological variations, and regional climate extremes, yet holistic assessments of their relative contributions and non-linear thresholds remain limited. We evaluated the water quality variation characteristics of Lake Dongting from 2014 to 2023 and found that total phosphorus (TP), total nitrogen (TN), and dissolved oxygen (DO) decreased significantly, while ammonia nitrogen (NH3-N) showed no significant temporal trend. CatBoost models identified water level and sediment load as important predictors of TP and DO, whereas TXx and SU25 were important predictors of TN and NH3-N. The testing R2 values were 0.62, 0.51, 0.39, and 0.58 for TP, TN, NH3-N, and DO, respectively, with testing RMSE values of 0.031, 0.480, 0.200, and 0.790, respectively. GAMs identified indicator-specific nonlinear responses: TP showed a model-estimated response point at WL = 28.36 m (95% CI: 27.74–28.89 m) and a hump-shaped response to Rx1 day, with a model-estimated point at 44.81 mm (95% CI: 40.32–49.45 mm); TN showed a response point at SU25 = 18.38 days (95% CI: 15.21–21.74 days), whereas NH3-N decreased monotonically with increasing SU25 and no accumulation threshold was inferred. Annual-scale variance partitioning and correlation analyses indicated that decadal nutrient variations co-varied with changes in agricultural activities. Our results indicate that lake water quality is associated with agricultural inputs and hydro-climatic conditions, underscoring the value of integrating agricultural non-point source control with climate-adaptive hydrological regulation for future water environment management.