The utility of near-infrared reflectance spectroscopy for wheat quality assessment
Date
2010
Authors
Butler, Joshua Donald, author
Haley, Scott D., advisor
Brick, Mark A., committee member
Chapman, Phillip L., committee member
Seabourn, Bradford W., committee member
Journal Title
Journal ISSN
Volume Title
Abstract
End-use quality improvement is an important objective in most wheat (Triticum aestivum L.) breeding programs. Limited sample size, destructive parameter testing, and the short duration between harvest and planting of winter wheat are challenges for testing early-generation breeding material for end-use quality parameters. Near-infrared reflectance (NIR) spectroscopy is a rapid and non-destructive technique that could facilitate early-generation selection for end-use quality. The precision and accuracy of an NIR equation for prediction purposes is dependent on the construction of a reliable calibration. The objectives of this study were to: 1) develop and validate NIR calibration models for grain volume weight, kernel characteristics, and Farinograph parameters, and 2) evaluate the performance of NIR calibration models in a breeding context for grain volume weight and single kernel characteristics. Calibration models for prediction of grain volume weight and single kernel characteristics were developed using NIR spectra and laboratory reference values from up to 10,000 samples collected from breeding nurseries under multiple environments over four crop years. Models encompassing all years of data revealed R2 (validation) of 0.73 for kernel diameter, 0.74 for kernel weight, 0.70 for kernel hardness, and 0.81 for grain volume weight. Of the Farinograph parameters, only absorption was effectively predicted using NIR calibration models for whole grain and flour with R2≥0.70. Realized heritability was estimated as a response to selection using NIR predicted values and laboratory reference values and was generally larger when using the reference values when compared to predicted values (0.17-0.77 vs. 0.05-0.77), but suggested that genetic gain was possible when using NIR models for selection. Classification errors when using the NIR models were highest in the mid-range reference values (56-66%), but could allow for divergent selection of high and low reference values. The results suggest that NIR models suitable for screening grain volume weight, SKCS kernel characteristics, and Farinograph absorption could be utilized in a breeding program and could aid in the elimination of early-generation samples with unacceptable values.
Description
Department Head: Gary Peterson.
Rights Access
Subject
wheat quality
near infrared reflectance spectroscopy
heritability
Farinograph
early-generation screening
calibration
Wheat -- Quality
Wheat -- Genetics
Near infrared reflectance spectroscopy -- Calibration
Farinographs