Premium
A Method to Determine the Appropriate Mathematical Form for Incorporating Soil Test Levels in Fertilizer Response Models for Recommendation Purposes 1
Author(s) -
Mombiela F.,
Nicholaides J. J.,
Nelson L. A.
Publication year - 1981
Publication title -
agronomy journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.752
H-Index - 131
eISSN - 1435-0645
pISSN - 0002-1962
DOI - 10.2134/agronj1981.00021962007300060007x
Subject(s) - fertilizer , mathematics , statistics , set (abstract data type) , quadratic model , soil test , test (biology) , soil science , function (biology) , econometrics , soil water , agronomy , environmental science , computer science , ecology , response surface methodology , evolutionary biology , biology , programming language
A method is proposed for developing fertilizer recommendations based upon a statistical estimate of the amount of “plant‐available” nutrient in the soil (d), and determining the appropriate mathematical form of the relationship of this estimate to soil test values (T). A function of soil test is then substituted into the fertilizer response model for d. Some comparisons of the effects of three different models (Mitscherlich, quadratic and square root) upon d estimates and on the form of their relationships to soil test values were made. Regressions of d estimates obtained using the three models upon soil test P values were carried out to determine empirically the appropriate mathematical form [called f(T)] for a set of Irish potato data from Maine and North Carolina. The f(T) was found to be linear for that particular set of data for all three models, i.e., d = b 0 + b 1 T. From these data, it was concluded that environmental conditions between states did not affect f(T) but did affect the level of maximum response (A, in the Mitscherlich model) and a factor related to curve shape (c, in the Mitscherlich model). Once f(T) is known for a model, it may be substituted into the model for d and this equation then used for fertilizer recommendations based upon site‐specific soil test information. Equations for estimating optimal P fertilizer rates assuming different cost‐price ratios (p) and marginal rates of return (R) were given for this data set. Similar equations may be derived for other specific crop‐soil conditions using the techniques described in this paper.