Development of near-infrared spectroscopy calibrations to measure quality characteristics in intact Brassicaceae germplasm

Abstract

Determining seed quality parameters is an integral part of cultivar improvement and germplasm screening. However, quality tests are often time consuming, seed destructive, and can require large seed samples. This study describes the development of near-infrared spectroscopy (NIRS) calibrations to measure moisture, oil, fatty acid profile, nitrogen, glucosinolate, and chlorophyll content in six species from the Brassicaceae family. Rapeseed and similar oilseeds are potential feedstocks for producing hydrotreated renewable jet fuel. Screening samples with NIRS would allow cultivars with desirable characteristics to be quickly identified. A total of 367 samples of six species (Brassica napus, Brassica carinata, Brassica juncea, Brassica rapa, Sinapis alba, and Camelina sativa) were scanned with NIRS. Global calibrations for all six species were developed using modified partial least squares regression with reference values obtained through wet chemistry techniques. Comparing predicted values to reference data, the coefficients of determination (r2) and ratios of performance to deviation (RPD) varied, with some calibrations performing better than others. The calibration equations for seed oil content (r2=0.98, RPD=7.3) and nitrogen (r2=0.98, RPD=5.3) performed very well while the equations for seed moisture (r2=0.93, RPD=3.8) and total glucosinolate content (r2=0.92, RPD=2.3) were more qualitative. Large variation was observed for chlorophyll content (0–390mg/kg) so two calibration equations were developed, one for the higher and one for the lower range of values. When combined, these calibrations also showed very good performance (r2=0.99, RPD=14). 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, RPD=9.9), and others, such as C22:0 (r2=0.69, RPD=1.9), showing poor correlation.