Predicting the oil contents in sunflower genotype seeds using near-infrared reflectance (NIR) spectroscopy

The aim of this experiment was to calibrate the NIR spectroscopy equation to evaluate the oil content of sunflower seeds from different genotypes produced under different environmental conditions in Brazil. The spectra of 901 standard samples obtained from 88 hybrids and 116 lines, which were evaluated in 11 locations, were collected from intact seeds (achenes) and correlated with data generated by nuclear magnetic resonance analysis. The calibration was determined by linear regression using partial least squares to estimate the parameters. The goodness of fit was evaluated using the coefficient of determination (R-2), standard error of calibration (SEC) and the standard error of performance (SEP). The wavelengths ranging from 1319 to 1760 nm were selected for the calibration. The R-2 was 0.87, the SEC was 2.39, and the SEP was 1.97. The oil content values obtained for the 19 hybrid seeds analyzed by NIR spectroscopy that were not included in the calibration were similar to the values obtained using the chemical method. The similarities between the values obtained using both methods and the R-2, SEC and SEP values indicated that it is possible to establish a calibration equation using NIR spectroscopy to determine the oil contents of sunflower seeds produced under Brazilian field conditions.