When can point-source buffering be overridden? Diagnostic evidence for concentration-discharge regime transitions along a point-source intensity gradient.

Continuous point-source (PS) discharges can dampen episodic nonpoint-source (NPS) nutrient signals, but their influence on concentration-discharge (C-Q) regimes in mixed-source rivers remains poorly constrained. We analyzed TN, TP, and NH₃-N C-Q relationships at seven stations in the Yiluo River Basin, China, using high-frequency water-quality monitoring, daily discharge, and co-located PS effluent records. A multi-scale diagnostic framework combined GLS-corrected C-Q classification, seasonal decomposition, daily hydrograph-limb analysis, and XGBoost-SHAP residual attribution. Increasing point-source intensity, indexed by upstream PS count, cumulative PS load, and distance-weighted PSI, was associated with lower TN C-Q slopes and a shift from enrichment (b up to +0.32) toward chemostatic behavior (|b| < 0.10). Three station-pollutant deviations suggest candidate conditions under which this buffering may weaken or be masked: BaiMaSi TP enrichment despite 11 upstream PS, consistent with possible dissolved-particulate P superposition; TanTou TN enrichment, consistent with headwater NPS flushing relative to moderate PS inputs; and QiLiPu NH₃-N wet-season enrichment, consistent with urban wet-weather inputs. Seasonal decomposition increased explanatory power by up to ΔR² = 0.43 and revealed wet-dry regime reversals. SHAP attribution linked residuals mainly to seasonal biogeochemistry for TN, antecedent discharge for TP, and effluent variability for NH₃-N. These case-derived hypotheses support pollutant-specific nutrient management in mixed-source rivers.