Nitrogen fertilizer and wheat in a semi-arid environment. 4. Empirical yield response models and economic factors

Abstract

The difficulties in predicting grain yield response to applied nitrogen under conditions of low and variable rainfall are discussed. Three models of grain yield response to applied nitrogen fertilizer for each of two strategies are proposed, based on data from field experiments carried out in the wheat growing areas of South Australia. In the first strategy (nitrogen applied at sowing) the parameters are May-August rainfall, October mean maximum temperature and one of three alternative site criteria, initial soil nitrate content (0-6 inches sampled shortly before sowing), 15-atmospheres soil moisture percentage (0-6 inches), or estimated nitrogen status. In the second strategy (nitrogen applied in late winter) the parameters are similar, except that May-July rainfall replaced May-August rainfall and a statistical relationship between yield response due to late application as compared with application at sowing was used. The response surfaces were examined using a calculated most profitable rate of application with a range of grain : fertilizer price ratios from 2 to 8. These calculations emphasize the importance of economic factors in affecting fertilizer use in areas where responses are small and variable. The limitations of the models and problems associated with the use of predicted climatic criteria are discussed. There is a need for further studies to iteratively test and modify these empirical models and ultimately to develop mechanistic models. Further study is also suggested on field aspects of late application and the possible role of both plant analysis and the single ion nitrate electrode for site characterization.