|
AUTHOR(S): R. J. Oosterbaan
|
TITLE Crop Tolerance to Soil Salinity, Statistical Analysis of Data Measured in Farm Lands |
 PDF |
ABSTRACT Field data on the relation of agricultural crop yields and soil salinity, expressed in the electric conductivity and extract of saturated soil samples (ECe in dS/m) , are analysed to find the salt tolerance level. It concerns eight different crops in three countries of which wheat occurred thrice and cotton twice. A first attempt of analysis can be by means of a generalized cubic regression (GCR) to detect the trend of the relation between yield and salinity. When this trend suggests that a range of no effect may exist, this range is determined by linear regression using the condition that the regression coefficient, i.e. the slope of the regression line, does not differ significantly from zero. This method is called partial regression (PAR) because the trend of the data beyond the range is analysed separately. From the range of “no effect”, the tolerance level, i.e. the maximum salinity level at which no yield decline sets in, can be determined. This tolerance level appears as a Break-Point (BP) between the yield-salinity relations left and right of it. This breakpoint is also called threshold, tolerance or critical level. In literature, the Maas-Hofmann (MH) model has been used frequently to detect the tolerance level, but this has occurred mainly for data obtained under controlled laboratory conditions, or in pot and lysimeter experiments. Owing to the generally flat trend at the tail-end of the yield-salinity relation, the MH model, determined by the least squares method, usually produces considerably lower BP values than the PAR method as the flat tail-end draws the BP to the left. It is questionable that the trend at the tail-end should determine the salt tolerance. The van Genuchten-Gupta (vGG) model, producing a general picture of the relation between crop yield and soil salinity, has been applied less frequently. It does not yield a well defined tolerance level. In general, a polynomial or the CGR regression produces a better fit of the data to the growth curve. In all cases in this study of field data, the “range of no effect” could be clearly defined. However, in some cases the tolerance values could be higher than determined here because prolongation of the range of no effect was prohibited due to limited number of data beyond it. A comparison of results of field data and laboratory experiments is made, which shows the same orders of magnitude. |
KEYWORDS crop yield, soil salinity, farm land, statistics |
|
Cite this paper R. J. Oosterbaan. (2018) Crop Tolerance to Soil Salinity, Statistical Analysis of Data Measured in Farm Lands. International Journal of Agricultural Science, 3, 57-66 |
|
