Utilization of near infrared reflectance spectroscopy for food corn hybrid evaluation

Highly significant relationships between Near Infrared Reflectance Spectroscopy (NIRS) and standard laboratory analyses were found for protein, lysine, oil, fiber, kernel hardness and starch thermal properties. NIRS prediction equations for all characters were generated by a multiple linear regressions. Regression equations, after correction for bias, with R{dollar}sp2{dollar} = 0.91 for protein, 0.81 for lysine, 0.89 for starch, 0.89 for amylose, 0.82 for oil, 0.81 for SHT-time, 0.76 for SHT-height, 0.86 for T-maximum, 0.89 for T-stop and 0.80 for T-range of gelatinization indicated good calibrations to the laboratory data. Delta enthalpy of starch thermal properties could also be estimated by an equation with R{dollar}sp2{dollar} = 0.77. Regression equations with moderate R{dollar}sp2{dollar} for fiber, ratio of coarse to fine particles, T-start, were also found.; Correlation coefficients between laboratory values and calculated NIRS values were 0.95 for protein, 0.91 for lysine, 0.95 for starch, 0.94 for amylose, 0.90 for oil, 0.73 for fiber. Correlation coefficients for hardness factors i.e. SHT-time, SHT-height, ratio of coarse to medium+fine particles and coarse to total ratio were 0.90, 0.88, 0.84, 0.84, respectively. For the starch thermal properties, the correlation coefficients were 0.89 for T-onset, 0.95 for T-maximum, 0.86 for DSC-height, 0.95 for T-stop, 0.88 for delta enthalpy, and 0.92 for T-range of gelatinization. There was no significant correlation between NIRS predicted values and standard laboratory analysis for moisture content.; The validated and bias-corrected regression equations derived from the NIRS calibration analyses were used to predict several characters of interest and to study the variation. Studies of variation of quality characters were conducted and related to food corn hybrid's evaluation. In 1991, 21 white food corn samples constituted a preliminary study, and in 1992-1993, 80 white and 12 yellow corn hybrids were full-sib mated and used for Genotype x Year evaluations. In the 1994, 56 white and 32 yellow corn hybrids planted in Lexington and Henderson were used for Genotype x Location evaluations. Another 109 hybrids from Kentucky Hybrid Corn Performance Test were full-sib mated and also used for genetic analyses. Genetic variation for most quality characters exists in the hybrid population under study. Broad sense heritability estimates are quite high. Interaction between Genotype with both Year and Location were significant. Genetic correlation coefficients among several characters were also significant.; Identification of all hybrids for nutrition factors, kernel hardness and starch thermal properties as well as constructions of index scores were facilitated by NIRS scan data. Quality scores and rankings of every food corn hybrid over all years of study and locations were obtained rapidly and might be utilized as a simple-to-use criterion for identification of superior hybrids in Lexington and Henderson. Weights assigned to individual characters were developed in conjunction with specific dry miller, wet miller and food snack processor requirements. Selection indices specific for corn dry millers, corn wet millers or food snack processor utilizations were constructed to be used in inbred development programs to select specific quality lines from segregating populations, from inbred crosses or from elite germplasms.