Abstract. The main objective of this project was to determine if Near Infrared Reflectance Spectroscopy (NIRS) could be used to effectively screen for quality characteristics in hulless barley. Three hundred samples of barley were collected over three years (1993 - 1995) from various locations across Western Canada. These samples included nine hulless varieties and one hulled barley check. Samples were sent to the University of Alberta for analysis of nine quality characteristics: Ash, Lipid, -Glucan, Protein, Starch, Total Fibre, Soluble Fibre, Insoluble Fibre and Pentosans. The samples were used to develop initial NIR equations for the nine quality traits by using modified partial least squares regressions and different combinations of gap, smooth and scatter correction settings. The r² of the initial calibrations (NIR predictions vs. Lab values) ranged from 0.98 for protein to 0.65 for lipid. The values obtained for these quality characteristics show that it is possible to use NIR for a screening technique in barley breeding programs.

Introduction. In a large breeding program it is necessary to screen thousands of early generation material to determine which lines have the necessary agronomic, yield and quality traits to make successful new varieties. While it is possible to screen lines for agronomic traits quickly in the field, quality analysis is often time consuming and expensive. In addition, traditional methods of determining quality involve the use of hazardous chemicals, and destruction of the seed. NIRS can provide rapid, non-destructive analysis of whole grain samples, using a relatively small sample size. This allows the plant breeders to screen thousands of early generation lines for quality characteristics is a short time period. The main objective of this project was to determine if NIRS could be used to effectively screen for specific quality characteristics (Table 2) in whole grain hulless barley.

Materials and Methods. Ten cultivars of barley were grown in the hulless barley coop trials at various locations in Western Canada (Table 1). The samples were collected over three years (1993, 1994, and 1995). Three hundred barley samples were collected in total. The three replications from each site were bulked and a 1kg sample of cleaned seed was sent for laboratory quality testing at the University of Alberta. Harrington was the only hulled check in the test.

NIRS Scanning

The samples were scanned using a monochromator NIRSystems model 6500 (NIRSystems, Inc., Silver Springs, MD), equipped with a transport module. Samples were scanned in a large natural product cell with a removable back. A 100g sample of cleaned, whole barley kernels were used. The reflectance spectrum of 400 - 2500 nm was recorded at 2 nm intervals for each sample.

Calibration and ValidationAll calculations were done using WINISI software, version 1.02a (FOSS NIRSystems, Silver Springs, MD). The 300 samples were sorted by reference data for each constituent, and every fifth sample was extracted for an external validation group (60 samples total). Calibration equations for the remaining 240 samples were developed by using spectral data from 400-2500 nm and Modified Partial Least Squares (MPLS) regression. A repeatability file was used to minimize the effects of particle size and sample temperature. Combinations of 1^st and 2^nd derivative, gap, smooth and scatter correction were used to maximize the equation results. Cross validation was used to prevent overfitting, and the software was setup for 4 outlier elimination passes. Equations were validated using the 60 sample validation group developed earlier.

Results and Discussion. The calibration statistics for the food quality calibrations developed are summarized in Table 2. The Protein equation is the most accurate r² =0.98, and a standard error of calibration (SEC) of 0.33. The least accurate equation developed was Lipid which had an r² = 0.65 and a SEC = 0.17.

The genetic variability available for some traits was the most limiting factor in achieving high r² values. In several cases, including pentosans, Insoluble Fibre, Total Fibre and ash analysis, Harrington as a hulled sample accounted for a large amount of the variability (Temelli and Helm in press). It was expected that protein would be the most accurate parameter predicted, because the 400-2500 nm region of the electromagnetic spectrum consists primarily of absorption information on NH (protein), OH (water) and CH (fat or oil).

Out of all of the equations, the Total Fibre equations used the least number of samples for calibrations (171 samples). This indicates that the samples thrown out may not have been completely clean. Any amount of adhering hull would have profound effects on fibre content.

Even though the r² values of most of the equations were in the range of 0.65 - 0.77, these equations still indicate good quantitative information and they are quite capable of separating samples into high, medium and low groups.

# of samples = Number of samples used for the calibration, after outliers were eliminated.
SEC = standard error of calibration
SECV = standard error of cross validation

Acknowledgments. Gratitude is expressed to Alberta Agricultural Research Institute - Matching Graints Program and the Alberta Barley Commision for their financial support. We would also like to thank Dr. Temelli, Dr. Sam Jadhav and Mr. Arun Lekhi for their technical assistance.

References:

Helm, J.H., W.C. Sauer, and L.A. Helberg. 1996. The use of mobile nylon bag technique and near infrared reflectance to determine digestible energy and protein content of hulless barley for pigs. VII International Barley Genetics Symposium, Proceedings. Pp 123-125.

Shenk, J.S., and M.O. Westerhaus. 1991a. NIRS analysis of agriculture products using population structuring and modified PLS regression. Crop Sci. 31:1548-1555.

Shenk, J.S. and M.O. Westerhaus. 1995. Calibration the ISI way. 7^th International Conference on Near Infrared Spectroscopy, Montral Canada, Proceedings. Pp 198-202.

Temelli, F. , and J.H. Helm. 1999. The effect of environment on the level of Non-starch polysaccharides and protein profiles in 10 lines of hulless barley. In Press.