Using Intact Nuts and Near Infrared Spectroscopy to Classify Macadamia Cultivars

Abstract

Macadamia nut industry is increasingly gaining more space in the food market and the success of the industry and the quality are largely due to the selection of cultivars through macadamia nut breeding programs. Thus, the objective of this study was to investigate the feasibility NIRS coupled to chemometric classification methods, to build a rapid and non-invasive analytical procedure to classify different macadamia cultivars based on intact nuts. Intact nuts of five different macadamia cultivars (HAES 246, IAC 4-20, IAC 2-23, IAC 5-10, and IAC 8-17) were harvested in 2017. Two NIR reflectance spectra were collected per nut, and the mean spectra were used to chemometrics analysis. Principal component analysis-linear discriminant analysis (PCA-LDA) and genetic algorithm-linear discriminant analysis (GA-LDA) were used to develop the classifications models. The GA-LDA approach resulted in accuracy higher than 94.44%, with spectra preprocessed with Savitzky-Golay smoothing. Thus, this approach can be implemented in the macadamia industry, allowing the selection of cultivars based on intact nuts. However, it is recommended that more experimentation to include more data variability in order to increase the classification accuracy to 100%.