Estimation of the proportions of offspring having soft fruit and juicy fruit as evaluated by sensory tests in Japanese persimmon

▶ The proportion of superior offspring obtained from a cross is important information for tree fruit breeders. The existing method for predicting the expected proportion of offspring with genotypic values exceeding a given critical value in crosses of Japanese persimmon was developed for data with a normal distribution and homogeneous variances in the offspring in each family. However, a normal distribution and homogeneous variances could not be assumed for categorical data evaluated by sensory tests, which are widely used for evaluation of fruit qualities because of their low cost and easy handling. In this study, we applied the generalized linear model as a method of data analysis for fruit firmness and juiciness, two attributes that are routinely evaluated by sensory evaluation in the breeding programs for Japanese persimmon. ▶ Our findings were that (i) it is possible to estimate the proportions of offspring with soft fruit and juicy fruit from the mid-parental values of sensory evaluations via logistic regression, and this method enabled us to utilize the sensory data from breeding records for predicting the expected proportion of offspring with desirable trait, (ii) firmness unilaterally influences the sensory rating for juiciness, and (iii) breeding to improve firmness and juiciness can be done with parents having moderate values for these traits, which can help mitigate the risk of inbreeding depression in Japanese persimmon. ▶ We believe these findings will contribute to an improvement in the efficiency of Japanese persimmon breeding programs.