Predictive modeling and correlation between the sensory and physicochemical attributes in ‘Rama Forte’ astringent persimmon fruit

Abstract Correlations between quality attributes determined by destructive and non-destructive analysis methods are being investigated to enable quantification and prediction of internal quality characteristics without the need for destructive techniques. Our study correlated sensory and physicochemical attributes of 'Rama Forte' persimmons treated for astringency removal with 70 % CO2 for 18 hours or 1.70 mL Kg-1 ethanol for 6 hours, to establish predictive models for destructive analytical methods based on non-destructive ones. Physicochemical and sensory analyses were carried out daily. Principal Component Analysis (PCA), Partial Least Squares Discriminant (PLS-DA) and regression analysis by Partial Least Squares (PLS) were applied to obtain prediction models. Two models based on fruit translucency (non-destructive) were obtained for persimmons treated with CO2, one for flesh firmness, and the other for color index prediction. A model based on sensory astringency (destructive) was developed to predict the astringency index for ethanol treatment. The models show a reliable fit, particularly in predicting flesh firmness by using the translucency of 'Rama Forte' fruit treated with CO2. Using the translucency scale and the prediction model, it is possible to establish the maximum period for logistic steps to reduce losses and waste in the persimmon chain. The low correlation between sensory astringency and proanthocyanidin content points to possible other compounds in the perception of astringency. Identifying these compounds will enable advances in the development of predictive models for quality attributes and shelf life of astringent persimmons.