摘要
本试验旨在分析茴香渣营养特征及构建其常规养分和矿物元素含量近红外反射光谱(NIRS)预测模型。利用NIRS技术结合改良偏最小二乘法(MPLS)分析了103份茴香渣样品中常规养分干物质(DM)、粗蛋白质(CP)、中性洗涤纤维(NDF)、酸性洗涤纤维(ADF)、粗灰分(Ash)、有机物(OM)、粗脂肪(EE)以及矿物元素钙(Ca)、铁(Fe)、钾(K)、磷(P)和镁(Mg)含量的预测模型。结果表明:茴香渣CP和EE含量较高,NDF和ADF含量较低,且矿物元素含量丰富,可作为饲料原料添加到动物饲粮中。构建的预测模型中,常规养分DM、CP、NDF、ADF、Ash、OM和EE含量的预测决定系数(RSQ)分别为0.940、0.922、0.918、0.873、0.847、0.878和0.908,验证相对分析误差(RPD)均大于2.5,预测的准确性较高,取得最佳定标效果。矿物元素Ca、Fe和K含量的RSQ分别为0.912、0.911和0.856,RPD均大于2.5,预测效果较好,可以用于实际生产;Mg含量的RSQ和RPD分别为0.794和2.203,预测效果较弱,仅能粗略预测来进行样品筛选;P含量的RSQ和RPD分别为0.654和1.667,预测效果差,预测值和实测值的误差较大,不能用于实际生产。综上所述,利用NIRS技术建立的预测模型可快速、准确地预测茴香渣中DM、CP、NDF、ADF、Ash、OM、Ca、Fe和K含量。
This experiment was conducted to analysis the nutritional characteristics of fennel residue and to construct the near infrared reflectance spectroscopy(NIRS)predictive model of conventional nutrient and mineral element contents.The predictive models for conventional nutrients of dry matter(DM),crude protein(CP),neutral detergent fiber(NDF),acid detergent fiber(ADF),crude ash(Ash),organic matter(OM),ether extract(EE)and mineral elements of calcium(Ca),iron(Fe),kalium(K),phosphorus(P),magnesium(Mg)contents in 103 fennel pomace samples were analyzed using NIRS technology coupled with modified partial least square(MPLS).The results showed that the CP and EE contents in fennel residue were relatively high,the NDF and ADF content were relatively low,and the mineral element contents were rich,which can be added as feed raw materials to animal diet.In the constructed predictive models,the coefficient of determination for validation(RSQ)for conventional nutrient contents of DM,CP,NDF,ADF,Ash,OM and EE were 0.940,0.922,0.918,0.873,0.830,0.878 and 0.908,respectively,the ratio of performance to deviation for validation(RPD)were all greater than 2.5,the accuracy of predictive was relatively high,and achieved the optimal calibration effect.The RSQ for mineral element contents of Ca,Fe and K were 0.912,0.911 and 0.856,respectively,the RPD were all greater than 2.5,the predictive effect was good,and can be used for practical production;the RSQ and RPD for Mg content were 0.794 and 2.203,respectively,the predictive effect was weak,and only roughly predicted the content for sample screening;the RSQ and RPD for P content were 0.654 and 1.667,the predictive effect was bad,and cannot be used in actual production.In summary,the predictive models constructed by using NIRS technique can quickly and accurately predict the contents of DM,CP,NDF,ADF,Ash,OM,Ca,Fe and K in fennel residue.
作者
李钰
李欣荣
李开栋
郭涛
史艳丽
李飞
年芳
徐国延
王新基
田多湖
许辉
LI Yu;LI Xinrong;LI Kaidong;GUO Tao;SHI Yanli;LI Fei;NIAN Fang;XU Guoyan;WANG Xinji;TIAN Duohu;XU Hui(College of Animal Science and Technology,Gansu Agricultural University,Lanzhou 730070,China;College of Grassland Agricultural Science and Technology,Lanzhou University,Lanzhou 730000,China;Datan Town Animal Husbandry and Veterinary Station,Minqin County Agriculture and Rural Bureau,Wuwei 733399,China;College of Science,Gansu Agricultural University,Lanzhou 730070,China;Minqin County Animal Husbandry and Veterinary Workstation,Wuwei 733399,China;Minqin County Vocational Secondary School,Wuwei 733399,China;Minqin County Defu Agricultural Technology Co,Ltd.,Wuwei 733399,China)
出处
《动物营养学报》
CAS
CSCD
北大核心
2024年第6期3973-3983,共11页
CHINESE JOURNAL OF ANIMAL NUTRITION
基金
甘肃省重点研发计划(20YF8NH158)
自治区科技支疆计划(2022E02140)。
