Predictions of fruit shelf life and quality after ripening: Are quality traits measured at harvest reliable indicators?

Nondestructive methods such as near infrared spectroscopy (NIRS) are increasingly used in sorting lines to assess quality traits of unripe fruit, i.e. dry matter (DM) and total soluble solid (TSS) contents, in order to create homogenous batches of fruit. The use of this approach is based on the assumption that fruit quality traits at harvest are reliable indicators of their post-harvest behavior and their quality after ripening. The present study tested this assumption by analyzing the relationships between quality traits at harvest and after ripening. In parallel, models were developed to determine the capacity of NIRS measurements on unripe fruit at harvest to predict their shelf life and quality after ripening. The quality traits DM, TSS content, pulp color (PC) and titratable acidity (TA) of 92 mangoes from different harvests, production years, and orchards were compared at harvest and after ripening. Previously developed NIRS models were used to nondestructively assess the quality traits of the mangoes at harvest. New partial least squares (PLS) regressions using different variable selection procedures and preprocessing techniques were used to predict fruit shelf life and fruit quality after ripening based on NIRS measurements at harvest. Weak relationships (r² < 0.41) were found between fruit quality traits measured at harvest and after ripening, except for DM content (r² = 0.61). The PC of mango measured at harvest was found to be the best indicator of fruit shelf life. Errors of PLS regressions to predict the TSS content (RMSEV = 1.1%), titratable acidity (RMSEV = 0.52%), and the Hue angle of the flesh (RMSEV = 1.86°) were in the same range as those of linear regressions based on quality traits assessed at harvest except for PC. This work provides evidence that fruit maturity and quality should be assessed using different indicators