Determination of water-salt optimal production functions for yield quantitative components of Varamin and Khordad cotton cultivars

In order to determine the best water - salt - quantitative parameters (number of bolls per square meter, boll weight, plant height, lint percentage) of cotton cultivars Varamin and Khordad, a research project examined the impact of salinity and irrigation water levels on yield and yield components of these two cultivars cotton in Birjand climatic conditions. The statistical design was a factorial randomized block split under climatic dry conditions with three levels of salinity and four levels of irrigation water. Sowing was done using basin method with dimensions of 4 x 5 meter on 28 April. Simple linear, Cobb-Douglas, quadratic and transcendental functions were used for this research. First, the results of functions coefficients tables and F value showed that all functions mentioned above were able to provide a determinant and significant estimation of the parameters the number of bolls per square meter, boll weight and plant height. Moreover, coefficients of determination of the plotted regression lines and the t-student statistic in regression line between the measured and predicted values confirmed the above results. So, the next step was to rank and identify the best function among them. To achieve this, five statistical parameters including root of the mean square error, coefficient of determination, modeling efficiency, the maximum error and coefficient of residual was used. The results obtained with both cotton cultivars of Varamin and Khordad showed that for the parameter of boll weight the quadratic function with a coefficient of determination of 80 and 89%,, respectively, and for the parameters of number of bolls per square meter and plant height transcendental functions with determination coefficients of 98, 95, 99 and 90% were able to be introduced as the superior functions amongst all others in the climatic area of Birjand. Additionally, none of the studied functions not able to provide an acceptable estimation of the lint percentage, because F value which showed the significant estimation of the studied parameters was not significant for the above mentioned parameter.