Land use and hydrological seasonality regulate pollution source contributions and water-quality pattern shifts in tributaries discharging to the Yangtze River

Tributaries entering large rivers are often regarded as pollutant delivery pathways, yet how land use and hydrological season jointly shape water-quality states and pollution-source signals remains insufficiently understood. Here, 28 tributaries entering the Yangtze River in Nanjing were investigated using monthly monitoring data for 23 water-quality parameters from 2016 to 2022. The modified comprehensive water quality index (WQI-DET) and trend tests were used to characterize long-term water-quality changes, self-organizing map (SOM) clustering to identify multivariate water-quality states, and a unified positive matrix factorization–generalized estimating equation (PMF–GEE) framework to quantify source-related contributions across tributary types and hydrological seasons. Median WQI-DET increased from −8.65 to 76.36, with a marked transition around late 2018–2019, while NH 3 -N and TP remained the dominant pollution constraints. SOM identified five recurrent water-quality states, and 59.6% of paired River-Year observations changed states between flood and non-flood seasons. PMF resolved three source-related factors: a catchment-derived mixed source with agricultural influence, an urban mixed anthropogenic source with metal-associated influence, and an urban wastewater-related source. GEE showed significant effects of land use and hydrological season for all three factors, but no significant interactions. Paired-section observations further showed significant differences in BOD 5 and COD Mn , but not NH 3 -N/TN or AOU*, indicating component-specific water-quality modification during river passage. These results indicate that land use establishes the baseline source structure, whereas hydrological season alters the relative expression of source signals and reorganizes multivariate water-quality states. However, interpretation is constrained by monthly sampling and the limited availability of tributary-specific hydrological and process measurements.