Development of a NIRS method to measure quality characteristics in Brassica germplasm

Authors: Oblath, Emily; Isbell, Terry

Submitted to: Association for the Advancement of Industrial Crops Conference
Publication Type: Abstract Only
Publication Acceptance Date: 8/27/2015
Publication Date: 10/18/2015
Citation: Oblath, E.A., Isbell, T.A. 2015. Development of a NIRS method to measure quality characteristics in Brassica germplasm [abstract]. Association for the Advancement of Industrial Crops Annual Meeting. p. 81.

Interpretive Summary:

Technical Abstract: Hydrotreated renewable jet (HRJ) fuels from plant oils have been commercially demonstrated as an alternative to petroleum jet fuels, but full scale production has not occurred because of concerns over cost and competition with food production. Rapeseed (Brassica napus), along with Brassica carinata, Brassica juncea, Brassica rapa, Camelina sativa, and Sinapis alba, are potential feedstocks for producing HRJ fuel. One challenge to using these oilseeds are the minor seed components, including glucosinolates and chlorophyll, that can cause poor oil quality and increased refining costs. Screening seed samples for quality characteristics can be time consuming, destructive, and expensive. Analysis by near-infrared spectroscopy (NIRS) is an alternative screening method that is both fast and non-destructive. A total of 367 seed samples from 18 cultivars grown at 7 locations across the United States were scanned with NIRS. NIRS scans required approximately 1 g of whole seed and took less than 2 minutes per sample. The moisture content, total oil content, and fatty acid profile were determined for all of the samples while nitrogen, chlorophyll, and glucosinolate content were determined for a subset of the samples. Destructive methods including gas chromatography (fatty acid profile), combustion analysis (nitrogen), ultraviolet-visible spectroscopy (chlorophyll), and high pressure liquid chromatography (glucosinolates) were used for many of the analyses. Calibration equations were then developed using WinISI and modified partial least squares (PLS) regression analysis. Global calibrations were developed rather than making separate calibrations for each of the 6 species. Comparing predicted values to the reference data, the calibration equations for moisture (r2=0.83), total oil (r2=0.97), nitrogen (r2=0.98), and glucosinolates (r2=0.92) showed good performance. The range of values for chlorophyll content was very large, 0-390 mg/kg, so 2 calibration equations were developed, 1 for the higher range and 1 for the lower range. When combined, the calibrations also showed very good performance (r2=0.99). The performance of the calibrations for the fatty acids was more varied with some performing very well, such as the calibration for C18:3 (r2=0.99), and others very poorly, such as C22:0 (r2=0.28). These NIRS calibrations can be used to quickly screen samples in the future, and can be updated with additional reference data to become more robust. The NIRS method is especially useful for situations such as genetic trials where only a few grams of sample are available and destructive analysis for all of the quality characteristics is not possible.