Near Infrared Spectroscopy in mango sorting lines for quality assessment and cultivar discrimination

The mango industry is subject to increasingly stringent standards concerning fruit ripeness and quality, prompting the adoption of real-time sorting technologies. This study examined the feasibility of using near infrared spectroscopy (NIRS) under laboratory conditions to assess mango quality and classify fruit by cultivar on sorting lines. A total of 227 fruit samples from Spain’s main cultivars, ‘Osteen’ and ‘Keitt’, were analysed on a conveyor belt operating at two speeds (2.3 and 7 cm s −1 ) using a diode-array spectrophotometer (spectral range: 950–1650 nm). Modified partial least squares regression models, combined with various spectral pretreatments, were developed to predict skin and pulp colour, pulp firmness, dry matter content, and soluble solids content (SSC). In addition, partial least squares–discriminant analysis was applied for cultivar classification. The results demonstrated that NIRS technology enabled accurate prediction of SSC (Ratio of Prediction to Deviation for prediction, RPD p = 2.6) and pulp firmness (RPD p = 1.7) in less than one second for fruit moving at 7 cm s −1 . At the lower belt speed (2.3 cm s −1 ), which allowed a longer scan duration (3 s), predictive performance improved for SSC (RPD p = 3.2) and firmness (RPD p = 2.1). Under both operational conditions, more than 98% of the fruit in the validation set were correctly classified by cultivar. These findings confirm the suitability of NIRS as a decision-support technology, enabling efficient, real-time classification by cultivar, ripeness, and quality.