Southern P Indices, Water Quality Data, and Modeling (APEX, APLE, and TBET) Results: A Comparison
Osmond, D., Bolster, C., Sharpley, A., Cabrera, M., Feagley, S., Forsberg, A., Mitchell, C., Mylavarapu, R., Oldham, L., Radcliffe, D., Ramirez-Avila, J. J., Storm, D., Walker, F., & Zhang, H. (2017). Southern P Indices, Water Quality Data, and Modeling (APEX, APLE, and TBET) Results: A Comparison. Journal of Environmental Quality. DOI:10.2134/jeq2016.05.0200.
Phosphorus (P) Indices in the southern United States frequently produce different recommendations for similar conditions. We compared risk ratings from 12 southern states (Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, and Texas) using data collected from benchmark sites in the South (Arkansas, Georgia, Mississippi, North Carolina, Oklahoma, and Texas). Phosphorus Index ratings were developed using both measured erosion losses from each benchmark site and Revised Universal Soil Loss Equation 2 predictions; mostly, there was no difference in P Index outcome. The derived loss ratings were then compared with measured P loads at the benchmark sites by using equivalent USDA–NRCS P Index ratings and three water quality models (Annual P Loss Estimator [APLE], Agricultural Policy Environmental eXtender [APEX], and Texas Best Management Practice Evaluation Tool [TBET]). Phosphorus indices were finally compared against each other using USDA–NRCS loss ratings model estimate correspondence with USDA–NRCS loss ratings. Correspondence was 61% for APEX, 48% for APLE, and 52% for TBET, with overall P index correspondence at 55%. Additive P Indices (Alabama and Texas) had the lowest USDA–NRCS loss rating correspondence (31%), while the multiplicative (Arkansas, Florida, Louisiana, Mississippi, South Carolina, and Tennessee) and component (Georgia, Kentucky, and North Carolina) indices had similar USDA–NRCS loss rating correspondence—60 and 64%, respectively. Analysis using Kendall’s modified Tau suggested that correlations between measured and calculated P-loss ratings were similar or better for most P Indices than the models.