Christman, Justine Lee (2017-05). Development of Near Infrared Reflectance Spectroscopy (NIRS) Calibrations of Maize Kernel Phosphorus for the Identification of Useful Breeding Material. Master's Thesis.
Thesis
Analysis of grain samples for nutrient composition is useful for breeding crops with improved nutritional, industrial or agronomic value. Wet chemistry analysis for composition components can be costly and laborious; therefore, a need exists for plant breeders to rapidly screen breeding material in a non-destructive manner. This study examined the application of near infrared reflectance spectroscopy (NIRS) calibrations to predict composition components, phosphorus in particular, in whole and ground maize (Zea mays L.) kernel samples using a specific Fourier Transformed NIRS (FT-NIRS) machine. Phosphorus, although an essential plant nutrient, has the potential to be an environmental pollutant. Therefore as maize production continues to increase globally, plant breeders need the ability to rapidly analyze nutrient profiles in breeding stock in order to select lines for advancement to achieve quality and environmental goals. An initial experiment was conducted to identify the optimal NIRS scanning procedure for the FT-NIRS, specifically a Thermo-Fisher Antaris II. We determined that for maize sample analysis, the optimal number of scans for consistency, accuracy, and analysis time was 128 for whole kernel, 64 for 1 mm fineness, and 96 for 2 mm fineness. Calibration development of NIRS was facilitated through a diverse sample set in which composition components (crude protein, phosphorus, fat, and starch) were quantified by wet chemistry analysis at a commercial laboratory. The addition of other components gave a baseline for comparison with the phosphorus calibration. We found that whole kernel maize samples (performance index, an independent measure, [PI] =60, r=0.94) were nearly as predictive as ground maize kernel samples (PI=63, r=0.88) for phosphorus. Several thousand samples that were in the TAMU Maize Breeding and Quantitative Genetics program's NIRS database, were analyzed to find genotypes with high and low phosphorus levels ranging from 0.27% to 0.43%. Selected genotypes were planted in an experimental test to validate predictions on their phosphorus levels and the effects that controlled pollinations play. These results indicated that phosphorus levels could be characterized categorically from both whole kernel and UDY calibrations, although some inconsistencies with expectations were found. Consequently, FT-NIRS could easily be integrated into a breeding program for rapid selection of genotype specific composition profiles.
Analysis of grain samples for nutrient composition is useful for breeding crops with improved nutritional, industrial or agronomic value. Wet chemistry analysis for composition components can be costly and laborious; therefore, a need exists for plant breeders to rapidly screen breeding material in a non-destructive manner. This study examined the application of near infrared reflectance spectroscopy (NIRS) calibrations to predict composition components, phosphorus in particular, in whole and ground maize (Zea mays L.) kernel samples using a specific Fourier Transformed NIRS (FT-NIRS) machine. Phosphorus, although an essential plant nutrient, has the potential to be an environmental pollutant. Therefore as maize production continues to increase globally, plant breeders need the ability to rapidly analyze nutrient profiles in breeding stock in order to select lines for advancement to achieve quality and environmental goals.
An initial experiment was conducted to identify the optimal NIRS scanning procedure for the FT-NIRS, specifically a Thermo-Fisher Antaris II. We determined that for maize sample analysis, the optimal number of scans for consistency, accuracy, and analysis time was 128 for whole kernel, 64 for 1 mm fineness, and 96 for 2 mm fineness. Calibration development of NIRS was facilitated through a diverse sample set in which composition components (crude protein, phosphorus, fat, and starch) were quantified by wet chemistry analysis at a commercial laboratory. The addition of other components gave a baseline for comparison with the phosphorus calibration. We found that whole kernel maize samples (performance index, an independent measure, [PI] =60, r=0.94) were nearly as predictive as ground maize kernel samples (PI=63, r=0.88) for phosphorus. Several thousand samples that were in the TAMU Maize Breeding and Quantitative Genetics program's NIRS database, were analyzed to find genotypes with high and low phosphorus levels ranging from 0.27% to 0.43%. Selected genotypes were planted in an experimental test to validate predictions on their phosphorus levels and the effects that controlled pollinations play. These results indicated that phosphorus levels could be characterized categorically from both whole kernel and UDY calibrations, although some inconsistencies with expectations were found. Consequently, FT-NIRS could easily be integrated into a breeding program for rapid selection of genotype specific composition profiles.
