Modelling and nondestructive quality prediction of Guanxi Honey Red Pomelo (<i>Citrus grandis</i> (L.) Osbeck) based on shape characteristics and bitterness

Currently, it is difficult to distinguish the inner quality of pachycarpous fruits using nondestructive instruments, and trace nutrients and flavor compounds are rarely considered in fruit sorting. In this study, the external features and internal quality indices of an extensively farmed citrus fruit, the Guanxi honey red pomelo, were measured. The correlation between the inner and external quality was analyzed using Pearson correlation, canonical correlation, and grey correlation analyses. This research innovatively introduced bitter substances and revealed that the height-diameter ratio could serve as a predictor for ultra-size fruits, which partially reflects inner quality indices. Several grey formulas were developed to express these correlations and were fitted to a matrix, which demonstrated accuracies at all level 1 (C < 0.35, P > 0.95). Based on the matrix, fruit quality analysis of white and red pomelo showed high prediction accuracy in TSS and bitter substance indices (R2 = 0.6 – 0.8) and indicated that mature and symmetrical pomelo (height-diameter ratio of 0.9 – 1.1) exhibit a similar quality pattern. This study achieved rapid quality assessment using easy measurable shape characteristics and explored the application of characteristic bitter substances in the quality evaluation model, providing theoretical references and potential application guidance for pachycarpous fruit sorting, as well as introducing a new idea to incorporating characteristic quality into modeling index.