Modeling Osmotic Salinity Effects on Yield Characteristics of Substrate-Grown Greenhouse Crops

Abstract

In a series of experiments with different osmotic potentials in the root environment, various vegetables, and ornamentals were grown in a substrate system. The osmotic potential was varied by addition of nutrients. Yield characteristics of the crop were related to the osmotic potential of the nutrient solution. Experimental results were fitted to a discontinuous piecewise linear response model introduced by Maas and Hoffman in 1977 and this study extended the investigation by introducing a polynomial and an exponential test model. The correlation coefficients for the polynomial and exponential model were generally higher than with the piecewise linear model. The polynomial model has the drawback that calculation of the salinity yield decrease (SYD) value is not possible. Salinity threshold values could be calculated with all models tested. The average values did not show great differences and the EC values found were 2.4, 2.1, and 2.2 dS m−1 for the polynomial, exponential, and peace wise linear model, respectively. It is argued that the exponential equations showed the closest relationships between the EC in the root environment and the yield characteristics of the crops. Fifth order polynomial equations yielded at least equal or even higher correlation coefficients, but these equations are very responsive to accidental positions of observations. The traditional discontinuous linear fitting gave reasonable results, but only when experimental observations below the threshold value were selected and discarded. This undermines the power of this model, because exclusion of observations is always arbitrary. Furthermore, a discontinuous relationship between crop yield and EC in the root environment is not likely, because yield response is of biological nature. Nevertheless, when the number of observations is not sufficient for the quantification of the exponential equation, mostly reasonable estimations of the salinity parameters can be obtained with the Maas and Hoffman model.