Determining Oil and Protein Content of Striped and Black Sunflower Seeds Using NIR Technology

Abstract

Applying near‐infrared (NIR) techniques to intact sunflower seeds to determine the oil content had previously been indicated to be a useful tool in the screening of samples prior to chemical analysis, but it could not replace analysis with wet chemistry. This study was conducted to develop improved oil and protein calibration models for sunflower seeds for use on a diode array‐type NIR instrument. One hundred sunflower seed calibration samples and 40 calibration test samples obtained from the National Cultivar Trials with Sunflower and growers' silos were selected to represent equal numbers of striped and black sunflower seeds. Whole and ground sunflower seeds were scanned on a diode array‐type NIR instrument. Calibration models were developed for black, striped and combined (black and striped) sunflower seeds for both whole and ground seed samples. Whole seed calibrations revealed inaccurate prediction values for oil and protein. A combined striped and black sunflower seed calibration for ground samples developed for oil and protein also displayed low predictability, but results were better than for whole seeds. Separate calibrations for the ground striped and black sunflower samples revealed D‐index values closer to 1.0 for the prediction of the oil and protein contents of the two types of seed. The D‐index values for ground striped sunflower test samples were 0.88 and 0.96 for oil and protein, respectively, whereas those for the black sunflower seed samples were 0.88 and 0.94, respectively. Calibration models for the NIR prediction of sunflower seed oil and protein content were improved by grinding the samples and developing separate calibrations for black and striped sunflower seed.