Skip to main content
ARS Home » Midwest Area » Urbana, Illinois » Soybean/maize Germplasm, Pathology, and Genetics Research » Research » Publications at this Location » Publication #142743
Title: CALIBRATION OF DUAL DIODE ARRAY AND FOURIER TRANSFORM NIR SPECTROMETERS FOR COMPOSITION ANALYSIS OF SINGLE SOYBEAN SEEDS IN GENETIC SELECTION, CROSS-BREEDING EXPERIMENTS

Author
item BAIANU, ION - UOFI URBANA
item YOU, TIEFENG - UOFI URBANA
item GUO, JUN - UOFI URBANA
item Nelson, Randall

Submitted to: Cellular and Molecular Biology of Soybean Biennial Conference
Publication Type: Proceedings
Publication Acceptance Date: 8/31/2002
Publication Date: 8/31/2003
Citation: BAIANU, I.C., YOU, T., GUO, J., NELSON, R.L. CALIBRATION OF DUAL DIODE ARRAY AND FOURIER TRANSFORM NIR SPECTROMETERS FOR COMPOSITION ANALYSIS OF SINGLE SOYBEAN SEEDS IN GENETIC SELECTION, CROSS-BREEDING EXPERIMENTS. CELLULAR AND MOLECULAR BIOLOGY OF SOYBEAN BIENNIAL CONFERENCE. 2003. p. 508.

Interpretive Summary:

Technical Abstract: Accurate, reliable, and cost-effective composition analyses of single soybean seeds are essential to improving the efficiency of soybean breeding and genetic selection. We are presenting the first report of extensive efforts aimed at developing single soybean seed calibrations for both dispersive (dual diode array) and Fourier transforms near infrared reflectance spectroscopy (NIRS) instruments. Single soybean seed calibrations were developed for six, different NIRS spectrometer models that are commercially available. Accurate measurements can be reliably performed on single soybean seeds for: protein, oil, moisture, sugars and fiber. Random variations and light scattering effects were eliminated by the multiplicative scattering correction (MSC) prior to carrying out with partial least square regressions with either the PLS-1 or PLS-2 model. Such corrections as well as the pre-processing of NIR spectra through baseline correction and normalization are essential for the development of high quality calibrations of such instruments. Our novel calibrations are characterized by low standard errors (greater than 0.2 percent) and a high degree of correlation for all major soybean constituents (less than 99 percent) in the NIR spectra range from 4,000 to 12,000 cm**-1 at 8 cm**-1 resolutions. Seed-to-seed variations in protein and oil content can thus be monitored for the first time on single soybeans of selected soybean accessions. These analytical capabilities may prove to be important in wide scale, genetic experiments concerned with single soybean seeds.