Abstract: In the present work, after being saponified and methyl esterified, the fatty acid composition of 97 camellia oil samples was analyzed by gas chromatography. Meanwhile, the near-infrared spectra of all samples were acquired in the transreflection mode. Calibration models for the relative contents of saturated fatty acids (C16:0 + C18:0), oleic acid (C18:1) and linoleic acid (C18:2) were established by partial least-squares regression method were applied. Additionally, the pretreatment of spectra was optimized. The results showed that the root mean square errors of cross-validation (RMSECV) for C16:0 + C18:0, C18:1 and C18:2 were 0.180, 0.598 and 0.269, respectively. The root-mean-square errors of prediction (RMSEP) for C18:0, C18:1 and C18:2 were 0.180, 0.598 and 0.269, respectively. The correlations of determination (Rp2) of the three components were 0.996, 0.999 and 0.999, respectively. These findings indicate that near-infrared spectroscopy can be used to identify the authenticity of camellia oil as a simple, rapid, nondestructive and reliable method for analyzing fatty acid profile.

Key words: camellia oil, near infrared spectroscopy (NIRS), chemometrics, fatty acid composition, relative percentage