Phosphorus (P) management in agroecosystems is driven by opposing requirements in agronomy, ecology, and environmental protection. The widely used maintenance P fertilization strategy relies on critical concentrations of soil test P (STP), which should cause the lowest possible impact on the environment while still ensuring optimal yield. While both soil P availability and crop yields are fundamentally related to pedoclimatic conditions, little is known about the extent to which soil and climate variables control critical STP. The official P fertilization guidelines for arable crops in Switzerland are based on empirically derived critical concentrations for two soil test methods (H2O-CO2 and AAE10). To validate those values and evaluate their relation to pedoclimatic conditions, we established nonlinear multivariate multilevel yield response models fitted to long-term data from six sites. The Mitscherlich function proved most suitable out of three functions and model fit was significantly enhanced by taking the multilevel data structure into account. Yield response to STPwas strongest for potato, intermediate for barley, and lowest for wheat and maize. Mean critical STP at 95% maximum yield ranged among crops from 0.15–0.58 mg kg−1 (H2O-CO2) and 0–36 mg kg−1 (AAE10). However, pedoclimatic conditions such as annual temperature or soil clay content had a large impact on critical STP, entailing changes of up to 0.9 mg kg−1 (H2O-CO2) and 80 mg kg−1 (AAE10). Critical STP for the AAE10 method was also affected by soil pH. Our findings suggest that the current Swiss fertilization guidelines overestimate actual crop P demand on average and that site conditions account for large parts of the variation in critical STP. We propose that site-specific fertilization recommendations could be improved on the basis of agro-climate classes in addition to soil information, which can help to counteract the accumulation of unutilized soil P by long-term P application.
