Walter Baethgen1, M. M. Alley2
1Agro‐Economic Div. Int. Fertilizer Development Ctr. Muscle Shoals AL
2Agronomy Dep. Virginia Polytechnic Inst. and State Univ. Blacksburg VA 24061
Tóm tắt
AbstractThe assessment of optimum N fertilizer rates for winter wheat (Triticum aestivum L.) production is a major unsolved problem, particularly in humid climates. Eight field experiments were conducted in the Coastal Plain and Ridge and Valley regions of Virginia during the 1981–1982 through 1985–1986 winter wheat growing seasons to determine critical N levels and optimum rates of N fertilization. The treatments in these experiments consisted of various N fertilizer rates applied at Zadoks growth stage (GS) 25 and GS 30. Plant N concentration (NCON30) and crop N uptake (NUP30) at GS 30 were measured immediately prior to the N fertilizer application. Nitrogen critical levels were defined as the NCON30 or NUP30 values that produced 90% of maximum grain yields in each experimental site with no further N fertilizer application. Two nonlinear models were developed to determine critical levels at GS 30 using NCON30 and NUP30 as the independent variable. The calculated critical levels for the models using NCON30 and NUP30 were 39.5 g N/kg and 95 kg N/ha, respectively, with R2 values for the models being 0.87 (NCON30) and 0.79 (NUP30). A family of response curves of grain yield to N fertilizer applied at GS 30 was fitted at each ex‐ perimental site. The amount of N fertilizer required at GS 30 to obtain maximum grain yield (NMAX30) was calculated for each response curve. Regression analysis was used to develop modells for estimating the amount of N fertilizer required at GS 30 to produce maximum grain yields for any given value of NCON30 or NUP30. Best models were selected considering fit, significance of the regression coefficients, and predictive ability. The best models in both cases were simple linear regressions. The model using NUP30 as the independent variable had a larger correlation coefficient (r = –0.87) and a better predicting ability than the model using NCON30. Regression analysis was also used to develop models to estimate the amount of N fertilizer required at GS 30 to produce economically optimum yields at three fertilizer/wheat price. ratios (2.0, 4.0, and 6.0) using NUP30 as the independent variable. The three simple linear regression models were similar and presented large correlation coefficients (–0.88, –0.89, and –0.88, respectively). The results indicate that further field experimentation using the same methodology described in this research should be conducted under varying soils and climatic conditions to develop a model for predicting economically optimum N fertilizer rates for grower fields.
