Multi-parameter index for durian maturity and its prediction using miniature near infrared spectrometer

Optimal maturity detection of durian is vital for the fruit to have good flavor and taste after harvest. Several indices are used by farmers to detect the proper maturity for harvesting purposes. The current study proposed a multi-parameter maturity index (MI) derived using principal component analysis from days after anthesis (DAA), total soluble solids (TSS) of the pulp, and the pulp dry matter content (DMC). In the development of the maturity prediction model using partial least squares regression, 120 durian fruit from six harvests were measured for near infrared (NIR) absorbance of the stem and the rind using a miniature spectrometer. Then, pulp samples were measured for their TSS and DMC. The cross-validated models for predicting the MI and the DMC were estab-lished and compared based on their performance levels. The prediction performance of the MI model indicated that the MI was more closely associated with the NIR spectra compared to the DMC. The best model for MI prediction of the pulp had values for the correlation coefficient, standard error of prediction (SEP), and the ratio of the standard deviation to the SEP of 0.888, 51.934 a.u., and 2.20, respectively, compared to those for DMC prediction of 0.720, 3.08%, and 1.44, respectively. In addition, MI had a linear relationship with DAA, whereas the DMC had a sigmoidal relationship. The results of this work should provide a starting point for a compre-hensive investigation to establishing an optimal maturity index. Further investigation could incorporate addi-tional maturity-related parameters and correlate the index with sensory scores.

Automated summary

Optimal maturity detection is important for durian to have good flavor. A multi-parameter maturity index (MI) was proposed using principal component analysis from days after anthesis, total soluble solids, and pulp dry matter content. Near infrared absorbance was measured to develop a maturity prediction model, which showed that MI was more associated with NIR spectra compared to pulp dry matter content. The MI prediction model had better performance with higher correlation coefficient compared to the dry matter content prediction model. Further investigation is needed to establish an optimal maturity index and correlate it with sensory scores.