Internal disorder evaluation of ‘Namdokmai Sithong’ mango by near infrared spectroscopy

• Published in: Horticulture, environment and biotechnology Vol. 63; no. 5; pp. 665 - 675
• Main Authors: Seehanam, Pimjai, Chaiya, Patomporn, Theanjumpol, Parichat, Tiyayon, Chantalak, Ruangwong, Onuma, Pankasemsuk, Tanachai, Nakano, Kazuhiro, Ohashi, Shintaroh, Maniwara, Phonkrit
• Format: Journal Article
• Language: English
• Published: Singapore Springer Nature Singapore 01.10.2022; Springer Nature B.V
• Subjects: Accuracy; Agriculture; Artificial neural networks; Biomedical and Life Sciences; Classification; Cultivation; Discriminant analysis; Disorders; Flowering; Fruit cultivation; Fruits; Infrared spectroscopy; Inspection; Life Sciences; Mangoes; Near infrared radiation; Neural networks; Nonlinear analysis; Plant Breeding/Biotechnology; Plant Ecology; Plant Physiology; Research Report; Signs and symptoms; Spectrum analysis; Tropical environment; Tropical environments; Vascular tissue; Artificial neural network; Black-streaked vascular; Chemometrics; Classification; Internal breakdown
• ISSN: 2211-3452; 2211-3460
• DOI: 10.1007/s13580-022-00435-5

Internal disorders are a major problem for mango growers and exporters. Internal breakdown (IBD) and black-streaked vascular tissue (BSV) are the most common symptoms found in fruit cultivated in tropical regions, especially in Thailand, where mango is considered the most important agriculture export commodity. The disorders cannot be detected by visual inspection; thus, consumers encounter these unpleasant defects when cutting into the fruit. The present study aimed to detect these internal disorders using the non-invasive technique of near infrared spectroscopy (NIRS). A total of 64 ‘Namdokmai Sithong’ mangoes were harvested at 120 days after flowering (DAF) and manually meshed with a scale of 1.5 × 1.5 cm 2 on both cheeks (totaling 1112 usable areas: intact (792), IBD (230), and BSV (90)). Spectral data were measured between 4000 and 12,500 cm − 1 for every meshed area via interactance measurement to capture intact and defective flesh. Classification models were thereafter developed using linear discriminant analysis (LDA) and an artificial neural network (ANN). IBD flesh had lower NIR absorbance than BSV and intact flesh. The LDA model discriminated intact flesh from defective flesh with a prediction accuracy of 86.25%. However, it was unable to separate IBD from intact flesh. In the case of non-linear analysis, the ANN reached a classification accuracy of 91.37%, whereby the misclassified matrix showed that intact, IBD and BSV flesh were well discriminated from each other, especially for BSV flesh. In summary, NIRS could be used to detect internal disorders in mango.