Statistical significance of nutrient loading structural effects: An adaptive criterion for lake sub-watershed prioritization

Eutrophication globally impairs lacustrine ecosystem health. Clarifying trophic state (TS) heterogeneity and its driving mechanisms is key to precise nutrient control. To unravel structural disparities in nutrient loading effects and prioritize sub-watersheds for targeted lacustrine TS regulation, nine Yunnan plateau lakes are selected as typical study area. By integrating path analysis for causal inference, the random forest model for trophic state index (TSI) inversion, and the InVEST model, we systematically explored the nutrient structural impact mechanisms and identified priority sub-watersheds (PSWs) for targeted governance. The main findings are as follows: 1) TSI in nearly 90% lake water area declined during 2019-2024, with more significant trends in high-trophic regions; The TS spatial patterns of the nine lakes have been reshaped by the governance efforts. 2) Exogenous nutrients significantly affect TS in most lakes; lowering TSI by 1.00% requires a 1.15%–4.36% reduction in watershed soil erosion, 1.05%–2.94% in nitrogen load, or 1.34%–2.56% in phosphorus load. 3) TSI is affected by the differentiated structural impacts of exogenous/endogenous nutrients; but the statistical significance, as a notable criterion for watershed prioritization, is not entirely dependent on loading magnitude. 4) Four types of PSWs are identified based on effect significance: integrated-control (24.6%), total-organic-control (4.1%), total-nitrogen-control (2.8%), and total-phosphorus-control (0.8%), with targeted governance schemes proposed for each type. A PSWs identification framework is established based on causal inference to provide both scientific insights for global lake trophic state research and practical tools for shifting lake management to precise control.