SEED COLOR ASSESSMENT IN BRASSICA-NAPUS USING A NEAR-INFRARED REFLECTANCE SPECTROMETER ADAPTED FOR VISIBLE-LIGHT MEASUREMENTS

Improved oil, protein and fibre contents are associated with light see d colour in rapeseed but the lack of reliable and efficient methods to measure seed colour has hindered breeding efforts for this trait. The feasibility of using light reflectance to assess seed colour in Brass ica napus was examined using scanning light reflectance spectrophotome try and near infrared reflectance (NIR). Light reflectance by seed sam ples from 30 doubled haploid (DH) lines segregating for seed colour in creased as the wavelength of the illuminating light in the scanning sp ectrophotometer increased between 550 and 650 nm. The largest reflecta nce values were measured for the yellow seed samples; the brown seed s amples were intermediate and the black seed samples had the lowest ref lectance values. The areas under the reflectance curves were used to t ransform the spectra to single values. Average light reflectance area values for the seed colour classes were significantly different from e ach other. The DHs and their corresponding light reflectance area valu es were also used to calibrate a NIR analyzer modified with 670 and 71 0 nm filters. The best calibration curve used three wavelengths (670, 2190 and 2208 nm) and had a multiple correlation coefficient of 0.987. Light reflectance area values determined with the calibrated NIR anal yzer for 30 randomly selected breeding lines could be used to categori ze the colour of the seed samples with no discrepancies between the vi sual and instrument classifications. The results indicate that NIR can be used to assess seed colour in rapeseed.