11个单粒花生脂肪酸近红外定量分析模型构建  

Establishment of 11 near infrared analytical models for main fatty acids in individual single peanut kernels

作　　者：韩宏伟 王传堂[1,2,3] 符明联[4] 杨军军 王志伟[2] 王秀贞 孙秀山[2] 杨珍[2] HAN Hong-wei;WANG Chuan-tang;FU Ming-lian;YANG Jun-jun;WANG Zhi-wei;WANG Xiu-zhen;SUN Xiu-shan;YANG Zhen(Shenyang Agricultural University,Shenyang 110000,China;Shandong Peanut Research Institute,Qingdao 266100,China;Shandong Luhua Group Co.Ltd.,Laiyang 265200,China;Industrial Crops Research Institute,Yunnan Academy of Agricultural Sciences,Kunming 650205,China)

机构地区：[1]沈阳农业大学,辽宁沈阳110000 [2]山东省花生研究所,山东青岛266100 [3]山东鲁花集团有限公司,山东莱阳265200 [4]云南省农业科学院经济作物研究所,云南昆明650205

出　　处：《中国油料作物学报》2023年第2期407-412,共6页Chinese Journal of Oil Crop Sciences

基　　金：山东省泰山产业领军人才工程项目(LJNY201808);国家花生产业技术体系项目(CARS-13);烟台市科技计划项目(2018ZDCX,2020XCZX046);农业农村部国家花生良种重大科研联合攻关项目;山东省农业科学院农业科技创新工程项目(CXGC2021A46,CXGC2021A09);“科创中国,一带一路”国际花生科技产业创新院(CXGC2022F13)。

摘　　要：随机选取80份不同花生品种籽仁,依托德国布鲁克光谱仪器公司MPA型傅里叶变换红外光谱仪构建了单粒花生油酸、亚油酸、棕榈酸、芥酸近红外定量分析模型,并首次构建了单粒花生一烯酸、棕榈一烯酸、山嵛酸、豆蔻酸、α-亚麻酸及长链饱和脂肪酸和不饱和脂肪酸近红外定量分析模型。经OPUS 7.5软件自动优化,花生籽仁油酸、亚油酸、棕榈酸、山嵛酸、豆蔻酸、长链饱和脂肪酸和不饱和脂肪酸的最佳光谱预处理方式均为“一阶导数+矢量归一化”,维数分别为10、10、8、10、5、10、10,决定系数(R2)分别为0.98、0.98、0.97、0.87、0.88、0.93、0.98,根均方差(RMSECV)分别为2.87%、2.18%、0.461%、0.133%、0.00416%、0.741%、0.19%;芥酸的最佳光谱预处理方式为“一阶导数+多元散射校正”,维数为9,R2为0.92;RMSECV为0.00628%;棕榈一烯酸的最佳光谱预处理方式为“一阶导数+减去一条直线”,维数为10,R2为0.80,RMSECV为0.00602%;花生一烯酸和α-亚麻酸的最佳光谱预处理方式均为“矢量归一化”,维数分别为8、10,R2分别为0.88、0.77,RMSECV分别为0.0731%、0.026%。经检验各项指标真实值和模型预测值平均相对偏差在0.12%~2.50%之间。11个花生籽仁脂肪酸近红外定量分析模型预测效果较好,可用于花生遗传育种和加工中的品质快速筛选和鉴评。Using individual single peanut kernels from 80 varieties/lines,near-infrared models were devel⁃oped based on the MPA Fourier transform infrared spectrometer(Bruker Optics,Germany),predictive of oleic acid,linoleic acid,palmitic acid,erucic acid,and predictive of eicosenoic acid,palmitoleic acid,behenic acid,myristic acid,α-linolenic acid,and long-chain saturated fatty acids and unsaturated fatty acids for the first time.Through automatic optimization with OPUS 7.5 software,the optimal spectral pretreatment mode for oleic acid,linoleic acid,palmitic acid,behenic acid,myristic acid,long-chain saturated fatty acids and unsaturated fatty acids was"first de⁃rivative+vector normalization",with dimensions of 10,10,8,10,5,10,10,the decision coefficients(R2)were(RMSECV)was 2.87%,2.18%,0.461%,0.133%,0.00416%,0.741%,and 0.19%,respectively.The optimal spec⁃tral pretreatment method for erucic acid was"first derivative+Multiplicative Scatter Correction",the dimension was 9,R2 was 0.92,and RMSECV was 0.00628%;the optimal spectral pretreatment for palmitoleic acid was"first derivative+minus a straight line",with the dimension of 10,R2 of 0.80 and RMSECV of 0.00602%.The optimal spectral pretreatment for eicosenoic acid andα-linolenic acid was"vector normalization",with dimensions of 8 and 10,R2 of 0.88 and 0.77,respectively,and RMSECV of 0.0731%and 0.026%,respectively.The average relative de⁃viation between the true value of each index and the predicted value of the model ranged from 0.12%to 2.50%.The 11 near-infrared models for fatty acid in peanut kernels had good prediction power.They could be used in rapid quality screening and evaluation for peanut breeding and processing.

关 键 词：花生 近红外反射光谱技术 Α-亚麻酸 油酸 亚油酸 

分 类 号：S565.2[农业科学—作物学]