Principal Component Analysis of Near Infrared Spectra as a Tool of Endosperm Mutant Characterisation and in Barley Breeding for Quality

Abstract

Near infrared technology, now widespread in quality control, makes it possible to obtain a total multivariate physical chemical fingerprint of the barley endosperm with high precision. Whole spectroscopic fingerprints of the physics and chemistry of barley seeds can be interpreted by multivariate analysis (chemometrics), by Principal Component Analysis (PCA) for classification and Partial Least Squares Regression (PLSR) for correlation. PCA classification of Near Infrared Reflectance (NIR) spectra can differentiate between mutants and alleles in the lys3 and lys5 loci. PCA on NIR can also be used as a routine in barley breeding to select for a multi-gene quality complex in barley as a whole e.g. increasing starch and reducing fibre content. This is done directly from the PCA classification plot by “data breeding” selecting the recombinants which are approaching the position of the normal high starch controls on the plot. Based on classification of NIR spectra, two alleles in the lys5 locus were characterised as a new class of (1→3,1→4)--glucan compensating starch mutants indicating a metabolic connection between starch and -glucan.