Genetic Diversity Of Antioxidant Property Among Rice Grains And Their Nondestructive Predictive Methods Using Near-infrared Spectroscopy

A total of 481 rice samples including 423 white rice,52 red rice and six black rice were characterized for the color of grain pericarp,grain size,total flavonoid and total phenolic contents,and antioxidant activity.Moreover,their pair-wise relationsips were analyzed.In addition,rapid and nondestrctice predictive methods based on near-infrared reflectance spectroscopy(NIRS)were established to measure phenolics content and antioxidant activity. The results indicated that highly significant genotypic differences were observed in total phenolics,flavonoids contents,2,2-azino-bis-(3-enylbenzothiazoline-6-sulphonic acid) diammonium salt(ABTS)radical cation antioxidant capacity,which displayed an increasing order in white rice,red rice and black rice,yet several white rice had higher phenolics and flavonoids contents than the red rice.The total phenolics content of different rice ranged from 108.1 to 1244.9 mg gallic acid equivalent(GAE)per 100g and averaged 195.5±144.8 mg/100g;total flavonoid content was from 88.6 to 286.3 mg rutin equivalent(RE)per 100g and the mean was 134.7±19.8mg/100g;antioxidant activity ranged from 0.012 to 5.533 mM Trolox equivalent antioxidant capacity(TEAC)per 100g and the mean was 0.413±0.696 mM TEAC.Correlation analysis showed that there was significant positive pair-wise correlation among the phenolics,flavonoids contents and antioxidant activity.Among all rice accessions,the grain pericarp color parameters had significant negative correlations with the phenolics,flavonoids contents and antioxidant activity,especially the negative correlation between a~* and antioxidant activity was consistent within the respective white rice and red rice groups.Grain length had poor correlation with flavonoids content among all rice accessions,and this relationship was consistent even between white and red groups.It was also found that antioxidant activity was negatively correlated with grain length and length to width ratio within the white rice genotypes.Calibration models for measurement of phenolics,flavonoids and antioxidant capacity were developed using principal component analysis(PCA),partial least squares regression(PLS) and modified partial least squares regression(mPLS)methods with the spectra of the dehulled grain.The results showed that NIRS could effectively predict the total phenolics contents and antioxidant capacity by PLS and mPLS methods.The standard error of prediction(SEP)were 47.1 and 45.87 mg gallic acid equivalent(GAE)/100g for phenolic content and the coefficient of determination(r~2)were 0.849 and 0.864 by PLS and mPLS method,respectively.Both PLS and mPLS methods gave similar accurate performance for prediction of antioxidant activity with SEP of 0.28 mM Trolox equivalent antioxidant capacity(TEAC)and r~2 of 0.82.However, the NIRS models was not successfully developed for flavonoid contents with the three methods above.