Abstract:Oil content and fatty acid composition in 444 ground cottonseed kernel samples were analyzed using near infrared reflectance spectroscopy (NIRS). Calibration equations were developed for oil and fatty acid contents with the modified partial least squares (MPLS) regression method. The correlations between NIRS and reference values in external validation were in agreement with the predictions in calibration. Each equation was assessed based on the relative prediction determinant for external validation (RPDv). Equations corresponding to total oil content (RPDv = 11.495) and linoleic acid (RPDv = 5.026) showed high accuracy. For palmitic acid (RPDv = 1.914), myristic acid (RPDv = 1.724) and oleic acid (RPDv = 1.999), the equations were predicted with relatively high accuracy while those for palmitoleic acid (RPDv = 0.686), stearic acid (RPDv = 0.792), linolenic acid (RPDv = 0.475) and 1-eicosenoic acid (RPDv = 0.619) were poorly predicted. The equations for traits with RPDv > 1.5 could be reliably used in screening samples for breeding programs.

Estimation of Oil Content and Fatty Acid Composition in Cottonseed Kernel Powder Using Near Infrared Reflectance Spectroscopy