Statistical segregation of apple trees in relation to morphological characters using discriminant analysis

Apple is the most dominating fruit crop in Himachal Pradesh. It accounts for 48 per cent of the total area under fruit crops and more than 86 per cent of the total fruit production of the state. This study was focused on discriminant analysis for testing groups’ differences with respect to a set of predictors or independent variables and to identify the predictor variables that best discriminated between two groups viz High Yielder and Low Yielder. It further involved the formulation of the categorization rule for allocating an apple tree to High Yielder and Low Yielder on the basis of different morphological characters of apple for both groups by developing a linear discriminating function rule to classify observations in the defined groups. The study suggested that for assigning a new tree with observed morphological characters (variables) to one of the groups the value of discriminant function or classification score (C_j) of both High Yielder and Low Yielder models is to be computed and the observation should be assigned to the group with highest score. Thus the rule is to allocate the tree to the group giving highest discriminant score. Error rate was also computed to evaluate the performance of the classification procedure. The discriminate function revealed a significant association between groups and all predictors accounting for 43.21 per cent between group variability although closer analysis of the structure matrix revealed yield (0.737) and number of flowers (0.448) to be two most important significant predictors with height and fruit weight as poor predictors. The cross validated classification showed that overall 81 per cent observations were correctly classified.