Enhancing dry matter prediction in durian using peduncle-based NIR spectroscopy across maturation rates

Some studies have explored using near-infrared (NIR) spectra of the peduncle to predict the dry matter (DM) content of durian pulp; however, this method has not yet been adopted in practice due to its limited accuracy. This study aims to refine of peduncle-based portable NIR spectroscopy for assessing pulp DM content by examining the effect of fruit maturation rate on its predictive accuracy. Durian fruits were categorized into slow-, fast- and fastest-maturing groups based on the timeline to reach commercial maturity defined by pulp DM content of 32%. Prediction models using NIR spectral data from fresh peduncle samples showed low accuracy (R² < 0.231), regardless of maturation rate. In contrast, models based on dried peduncle powder significantly improved predictive performance. The highest accuracy (R² < 0.78) was observed in slow-maturing fruits, while the lowest (R² < 0.43) was seen in fastest-maturing fruits. Notably, slow-maturing fruits exhibited a linear increase in peduncle sugar content, whereas fastest-maturing fruits displayed a parabolic accumulation pattern. When data from all three maturation rate groups were used to construct the model, external validation confirmed the model’s ability to classify immature versus mature durian fruit accurately. The classification accuracy for identifying immature fruits reached 86% using partial least squares discriminant analysis (PLS-DA). Both portable and bench-top spectrometers produced comparable results. The findings suggest that fruit maturation rate influences the accuracy of pulp DM prediction, indicating the importance of incorporating a range of maturation rates during model development. Overall, this study highlights the potential of optimized peduncle-based portable NIR spectroscopy as a non-destructive approach for classifying immature durian fruit.