Comparison of Stochastic Soybean Yield Response Functions to Phosphorus Fertilizer
Abstract: The random parameter approach of fertilizer response model was known better than fixed parameter version for determining optimum doses of fertilizer recommendation. However, the selection of functional forms suitable for certain cropping condition was also critical. The purpose of this study was to know the best model of stochastic soybean yield response function to phosphorus fertilizer. The research was conducted based on multilocation experimental data of soybean yield response to phosphorus fertilizer. The fixed parameter models (M1) of Linear plateau, Spillman-Mitscherlich, Quadratic and Logistic were compared with the random parameter models containing either 1 (M2) or 2 random effects (M3) using -2 log-likelihood, Akaike information criterion, and Bayesian information criterion. Results showed that the AIC values of M1 fixed parameter models sequentially were Linear plateau < Spillmann-Mitscherlich = Logistic < Quadratic. Meanwhile, the AIC values of M2 random parameter models sequentially were Linear plateau < Logistic < Spillmann-Mitscherlich < Quadratic. The AIC values of M3 random parameter models sequentially were Spillmann-Mitscherlich < Logistic < Linear plateau. The best model for soybean yield response function to phosphorus fertilizer was the stochastic Spillmann-Mitscherlich model with location intercept and the increase in yield by applying fertilizer random effects.
Keywords: response functions, fixed effect, random effects, multilocation trials, soybean.
Pages: 1 – 11 | Full PDF Paper