Evaluation of hill plots and near infrared reflectance techniques as breeding tools in spring malting barley

Graduation date: 1989Hill plots and near infrared reflectance (NIR) spectroscopy were\ud investigated as breeding tools to facilitate doubled haploid recurrent\ud selection for malting quality characters in spring barley. Main\ud objectives of this research were to i) compare hill and row plot\ud expression of agronomic and malting quality traits in an array of elite\ud spring growth habit barley germplasm, and ii) compare NIR prediction of\ud grain protein and malt extract in the two plot types.\ud Twenty-four elite spring barley genotypes were evaluated in\ud separate row and hill plot experiments in three environments. Hill\ud plots were evaluated at four seeding rates. A single seeding rate was\ud used for row plot evaluation.\ud There was significant genotypic variation for all agronomic traits\ud in hill plots. Genotype response to hill seeding rates was significant\ud and consistent for most agronomic traits. Spearman's rank correlations\ud between hill and row plots were high for all malting quality traits and\ud most agronomic traits, except grain yield. Comparable results were\ud obtained when the percentage of lines in common at 25 and 50% selection\ud intensities was tabulated. Therefore, selection in hill plots should be\ud effective for most malting quality and agronomic traits, with the\ud exception of grain yield. Differential expression of yield in hill and\ud row plots was attributed to differences in tillering as a consequence of\ud distinct patterns of competition in the two plot types. A seeding rate\ud of 10 seeds per hill was identified as appropriate for preliminary\ud screening of traits amenable to hill plot evaluation. Samples from this\ud seeding rate were used as validation samples to test row-plot derived\ud NIR prediction equations for grain protein and malt extract.\ud Separate of five and six wavelength calibration equations were\ud selected for grain protein and malt extract, respectively. The multiple\ud correlation coefficient (R) for the protein equation was 0.98 and that\ud for malt extract was 0.88. Three distinct sets of validation samples\ud were used to test the grain protein and malt extract calibrations.\ud Based on these analyses, NIR prediction of protein in hill and row plots\ud should be effective. NIR prediction of malt extract may only be\ud effective in row plots.\ud Due to differential trait expression in row and hill plots and the\ud error associated with NIR measurement, doubled haploid recurrent\ud selection based on NIR screening and hill plot evaluation will likely\ud only be effective for plant height, yield components other than numbers\ud of tillers per unit area, and grain protein. Selected entries should\ud then be evaluated for grain yield and malt extract in conventional row\ud plots. NIR prediction for malt extract should be effective in row plot\ud evaluations, achieving considerable efficiency compared to complete\ud chemical malt analysis