Combining data from fertilizer experiments into a function useful for estimating specific fertilizer recommendations

Abstract

The decision of greatest economic importance that farmers have to make in using fertilizers is the selection of the fertilizer treatment to employ. T f the treatment selected includes an element that is not needed or a rate of an element higher than the economic optimum, the farmer will not realize the maximum returns from his investment in fertilizers and may even suffer a loss. On the other hand, if he applies less than the optimal amount of fertilizers, he will only be partially exploiting this source of added farm income. The determination of optimal rates of fertilization is one of the most important functions of adaptive research. It is especially difficult to arrive at fertilizer recommendations for specific producing conditions as crop response to fertilization depends on the nature of the crop itself, the characteristics of the soil and climate at the location where it is grown, and the management practices employed in growing the crop. A significant change in any one of a large number of crop, soil, climatic, and management factors can greatly modify the reaction of the crop to fertilization. The optimal rate of nitrogen fertilization, for example, may depend just as much ·on the rainfall or the moisture retention characteristics of the soil as on the nature of the crop or the level of available soil nitrogen. Ideally, two types of information are needed for making specific fertilizer recommendations: (a) The general yield equation for the crop with yield expressed as a function of applied fertilizer variables and the productivity factors, and ( b) The levels of the productivity factors for the specific conditions for which a fertilizer recommendation is sought. Unfortunately there are no well defined procedures to follow in generating a general yield equation and in characterizing the productivity factors. Consequently, the results obtained in field experiments are usually averaged over broad geographical areas to produce general fertilizer recommendations. The present study was carried out with two principal objectives in mind: (a) To produce a general yield equation useful in estimating nitrogen fertilizer needs of unirrigated corn, for specific producing conditions in a region with highly variable rainfall, and ( b) To acquire a better understanding of the problems involved in measuring the productivity factors and in employing the multiple linear regression model for calculating a general yield function. The experimental part of this study was conducted during the period 1962-1965. Simple fertilizer trials were carried out with unirrigated corn in farmer's fields in the Bajio area of Central Mexico. The results obtained in the first two years of this study have already been published, so only the results obtained in 1964 and 1965 will be reported here.