Development of a Multivariate Nonlinear Regression Model for the Prediction of Soil Quality in Agricultural Soils.

The traditional agricultural use of soil causes the deterioration of its physicochemical properties, decreasing its fertility and therefore the quality and yield of crops. Maintaining the functionality and fertility of the soil is necessary to satisfy the nutritional needs of the population; therefore, it is necessary to create tools to measure, maintain and improve soil quality. In relation to the above, a multivariate nonlinear regression model (MNRL) was developed for the prediction of soil quality focused on fertility, using 75 soil samples from the Bajío of the State of Guanajuato. Twenty-two physicochemical indicators related to soil fertility were analyzed. The MNRL developed consisted of the indicators iron (Fe), sodium adsorption ratio (SAR), manganese (Mn) and exchangeable sodium percentage (ESP). The MNRL made it possible to differentiate the quality of the different soil samples analyzed, obtaining a high quadratic correlation coefficient (R² = 0.96) and a low root mean square error (RMSE= 0.05) [ABSTRACT FROM AUTHOR]