Mid-infrared spectroscopy coupled with chemometrics to identify spectral variability in Australian barley samples from different production regions. (January 2019)

• Record Type: Journal Article
• Authors: Gordon, R.; Chapman, J.; Power, A.; Chandra, S.; Roberts, J.; Cozzolino, D.
• Is Part Of: Journal of cereal science. Volume 85(2019)
• Page Start: 41
• Page End: 47
• Publication Date: 2019-01
• Subjects: Mid infrared -- Barley -- Climate -- PCA -- PLS; Grain -- Periodicals; Cereal products -- Periodicals; Céréales -- Périodiques; Produits céréaliers -- Périodiques; Cereal products; Grain; Periodicals; 664.705
• DOI: 10.1016/j.jcs.2018.11.004
• Languages: English
• ISSNs: 0733-5210

Abstract

Abstract: Mid-infrared spectroscopy coupled with chemometrics was used to identify spectral differences in malt barley associated with region (e.g. rainfall, temperature and sunlight). The observed differences in the MIR spectra confirmed that different environmental conditions of the region they are grown, explain or modulate changes in the chemical composition of the grain. Principal component analysis (PCA) was performed to identify underlying patterns and correlations in the data set. Partial least squares (PLS) regression with cross-validation was applied to develop models to predict the regions using the weather variables, the MIR fingerprint region and a combination of both. The PLS model with the combined data predicted the growing location with the standard error of cross-validation (SECV) of 3.38%, an R 2 value of 0.53, a slope of 0.52 and a bias of 0.23. The exploratory approach demonstrated the usefulness of chemometrics to identify differences and similarities based on the MIR spectra. The results provided a solid foundation for hypothesising the overall degree of biochemical similarity and or difference among the samples and a quantifiable method to screen samples for further analysis. Highlights: Barley from different Australians regions was analysed using mid infrared. Observed differences in the absorbance at specific wavenumbers related to region. Combination of climate and MIR data provided the better classification results.