The use of visible and near infrared spectroscopy for evaluating passion fruit postharvest quality

• We evaluated physico-chemical quality of passion fruit by near infrared spectroscopy.
• The competences of interactance and transmission measurements were compared.
• The PRESS analysis and VIP scores were used to explain effective PLS models.
• Reasonable prediction performances were obtained from both measurements.
• Interactance calibrations provided better prediction results than transmission ones.

Visible and short-wave near infrared spectroscopy (Vis/SWNIRS) was investigated using a non-destructive method for evaluating passion fruit quality. In this study, interactance and transmission measurements were performed and their competences were compared. Prediction models of soluble solids content (SSC), titratable acidity (TA), ascorbic acid content (ASC), ethanol concentration (EtOH), peel firmness (PF) and pulp percentage (PP) were developed based on multivariate methods of partial least square regression (PLSR) analysis. The PLS models from interactance measurements provided better prediction results than the transmission technique. The best model was obtained from interactance SSC calibration with a correlation coefficient between measured and predicted values (R) of 0.923. Furthermore, the PLS models generated from interactance and transmission spectra also provided satisfactorily prediction results for EtOH, PF and PP. However, all calibrations failed to predict ASC by providing low correlations and high root mean square errors of prediction (RMSEP).