Nutrient loss patterns across crop type and slope combinations in South Korean upland fields: An integrated field monitoring analysis

Agricultural non-point source pollution from upland fields poses a critical threat to water quality, yet its regional management remains limited by highly variable nutrient loss and site-specific monitoring. This study integrated 295 runoff events from South Korean upland fields as a meta-database synthesis and analyzed 274 to evaluate associations of slope and crop type with total nitrogen (TN) and total phosphorus (TP) loads and identify vulnerable field conditions. A generalized additive mixed model showed that the slope response of TN differed between crop types. Leafy vegetables exhibited a statistically significant bell-shaped response peaking at intermediate slopes, whereas root vegetables showed a comparatively flat response. TP increased approximately linearly with slope, and this response did not differ significantly between leafy and root vegetables. A generalized linear mixed model further confirmed significant crop-type differences after adjusting for rainfall and slope. Root vegetables, leafy vegetables, and oil crops showed high nutrient loads, whereas legumes and fruit crops showed consistently lower TN and TP loads, and perennial crops showed lower loads mainly for TP. K-means clustering classified runoff events into four patterns: low loss, moderate loss, high-TN loss, and high-TP loss. The high-TP cluster was associated with root and leafy vegetables on slopes of ≥ 10% under high rainfall, whereas the high-TN cluster was dominated by oil crops under high rainfall even on gentle slopes. These results indicate that nutrient-loss risk is best interpreted through combinations of crop type, slope, and rainfall, providing a practical framework for prioritizing field-level nutrient management.