The study is aimed at identifying the most important soil properties influencing cassava yield in Akpabuyo Local Government. In order to gain a proper understanding of the relations between soil variables and the cassava parameters, and also to identify those soil properties contributing significantly to the prediction of cassava yield and its vegetative parameter in the study area, cassava yield parameters (tuber, leaves and stem) were examined and related to 21 soil parameters to statistically examine how soil properties related to yield of cassava. Consequently, bivariate and multiple-regression models were used to carry out the statistical relationship between the aforementioned variables. The results of the models form which discussions on soil parameters contributed to cassava yield were pursued. First, over-parametised model test was conducted, the essence of which is to capture the main dynamic process in the model. From this model, a parsimonious model was achieved via a reduction (selection) process, guided mostly by statistical consideration. Thus, the parsimonious reduction (selection) process made use of the step-wise regression procedure, subjecting each stage of the reduction process to several diagnostic tests before eventually arriving at an interpretable model. The results of the regression analysis show that 14 different soil properties contributed significantly to the prediction of cassava parameters in the study area. To achieve this selection/reduction process, an index of soil variables influencing the yield of cassava was computed. The index is simply the mean percentage contribution of each of the 14 significant variables to the prediction of the tree cassava parameters in the area. In order words, the percentage values representing the levels of explanation of each soil parameter on each of three multiple-regression results are summed up and the total is divided by 3. The mean value obtained is used as an index of each soil property contributing to the prediction of a cassava parameter.
 

Download Full Text - PDF