Improving phosphorus loss assessment with the apex model and phosphorus index

Abstract

Agricultural fields contribute phosphorus (P) to water bodies, which can degrade water quality. The P index (PI) is a tool to assess the risk of P-loss from agricultural fields. However, due to limited measured data, P indices have not been rigorously evaluated. The Agricultural Policy/Environmental Extender (APEX) model could be used to generate P-loss datasets for P index evaluation and revision. The objectives of the study were to i) determine effects of APEX calibration practices on P-loss estimates from diverse management systems, ii) determine fertilizer and poultry litter management effects on P-loss, iii) evaluate and update the Kansas PI using P-loss simulated by APEX and iv) determine appropriate adsorption isotherms with advection-dispersion equation with column leaching experiment. Runoff data from field studies in Franklin and Crawford counties were used to calibrate and validate APEX. Poultry litter and inorganic fertilizer application timing, rate, method, and soil test P concentration effects on P loss were analyzed using location-specific models. A column leaching laboratory study was also conducted to test the adsorption isotherms. Location-specific model satisfactorily simulated runoff, total P (TP) and dissolved P (DP) loss meeting minimum model performance criteria for 2/3 of the tests whereas management-specific models only met the criteria in 1/3 of the tests. Applying manure or fertilizer during late fall resulted in relatively lower TP loss compared to spring applications before planting. The Kansas-PI rating and the APEX simulated P-loss were correlated with r² of 0.40 (p<0.001). Adjusting the weighting factors for Prate, soil test P, and erosion improved the correlation (r² = 0.46; p<0.001. Using a component PI structure and determining the weighting factors by multiple linear regression substantially improved the correlation between the PI and TP loss (r² = 0.69; p<0.001). In the P-leaching experiment, both the linear and nonlinear adsorption isotherms did not fit the experimental data. A multi-reactional advection-dispersion model that better describes all the P processes and complexities in soils should be included in the future. These procedures can provide a roadmap for others interested P transport in soils and using computer models in evaluation, and modifying their PI.