Abstract: The Erhai Lake Basin, a crucial agricultural production base in Yunnan Province, has its ecological environment and regional sustainable development significantly influenced by the phosphorus (P) flow characteristics of its agricultural systems. Based on data collection and field surveys, 374 villages were classified into four major types—Crop production types, Livestock production types, Traditional crop-livestock model, and Stereoscopic fruit-livestock model—according to their primary industrial activities, with further subdivision into 17 subtypes based on dominant crops and livestock. Using a village-scale nutrient flow model (NUFER-village), P flows, use efficiency (PUE), and environmental losses(PL) were quantified. Key drivers of environmental emissions were identified via redundancy analysis (RDA). Results showed: ①the highest total P input in Crop production types (3.21-12.78 t) and the lowest in Stereoscopic fruit-livestock model (2.90-3.32 t). ②PUE ranked in descending order as follows: Crop production types (5.61%-33.68%), followed by Stereoscopic fruit-livestock model (18.35%-19.20%), then Traditional crop-livestock model (11.06%-16.03%), and finally Livestock production types (3.61%-9.49%). ③PL was highest in pig-raising subtypes (2.4 t) and lowest in Stereoscopic fruit-livestock model (0.4 t). ④SF and FI were the primary drivers of environmental emissions in Crop production types and Livestock production types, contributing 69.5% and 58.1%, respectively. In Traditional crop-livestock model and Stereoscopic fruit-livestock model, PUE and environmental emissions were co-driven by both chemical fertilizer input and feed input. These findings provide a scientific basis for P management in agricultural systems of plateau lake basins, supporting sustainable agricultural development.