评定蛋鸡豆粕代谢能和氨基酸可利用率的研究

Assessment of Metabolizable Energy and Amino Acid Availability of Soybean Meals in Laying Hens

本试验旨在评定罗曼蛋鸡对不同来源豆粕表观代谢能（AME）和氨基酸可利用率（AAA）,并用傅里叶近红外光谱（NIRS）分析技术建立其预测模型。选择248只体重（1.60±0.10）kg、产蛋率85%的36周龄罗曼蛋鸡,按单因素完全随机设计,分为31组,每组8个重复,每个重复1只鸡。在训饲的基础上,采用全收粪法评定30种不同来源豆粕和1种基础饲粮的AME和AAA,然后用NIRS技术建立其生物效价的预测模型。结果如下：1）不同来源的30种豆粕AME在11.95-14.87 MJ/kg之间,平均值为（13.24±0.67）MJ/kg;总氨基酸可利用率（TAAA）在89.99%-94.96%之间,平均值为（93.73±1.23）%。2）豆粕AME的NIRS预测模型的校正决定系数（Rcal2）、交叉验证系数（Rcv^2）、外部验证系数（Rval^2）分别为99.24%、83.79%、80.73%,外部验证标准差（RMSEP）为0.22 MJ/kg;TAAA的NIRS预测模型的Rcal^2、Rcv^2、Rval^2范围分别为94.20%-99.97%、76.38%-97.32%、61.80%-99.42%,RMSEP范围为0.06%-1.00%。结果表明：1）不同来源豆粕的AME、AAA在罗曼蛋鸡上有较大差异;2）利用NIRS分析技术可建立罗曼蛋鸡豆粕的AME、AAA预测模型,模型的Rcal^2及预测的RMSEP较好。

The study was conducted to investigate apparent metabolizable energy （ AME） and amino acids a-vailability （ AAA） of different soybean meal samples, and explore feasibility of establishing prediction for the AME and AAA using Fourier near infrared spectroscopy （ NIRS） . A completely randomized design was used with a total of two hundred and forty six 36-week-old Lohmann laying hens with initial body weight of （1.60± 0.10） kg and laying rate of 85%. Hens were divided into 31 groups with 8 replicates per group and 1 hen per replicate. The experiment was conducted to use the method of trained feeding and total excreta collection. Addi-tionally, the near infrared spectroscopy （ NIRS） calibrations were established to predict AME and AAA of soy-bean meal for laying hens feeds. The results showed as follows：1） the average AME of 30 soybean meals was （13.24±0.67） MJ/kg, ranging from 11.95 to 14.87 MJ/kg. The range of 30 soybean total AAA （TAAA） of laying hens were 89.99%-94.96%, with average value of （93.73±1.23）%. 2） Adjusted determination coef-ficient （ R cal ^2 ） , cross validation factor （ R cv ^2 ） , coefficient of external verification （ R val^ 2 ） and external valida-tion standard deviation （ RMSEP ） of NIRS model for AME of soybean meals were83. 79%, 99. 24%, 80.73%, 0. 22 MJ/kg. Rcal^2, Rcv^2, Rval^2 and RMSEP of NIRS models for AAA of soybean meals were 76.38%-97.32%, 94.20%-99.97%, 61.80%-99.42%, 0.06%-1.00%. These results indicate as follows：1 ） There are significant differences among the different soybean meal sources in their AME and AAA in Lohm-ann laying hens. 2） NIRS can be used to establish equations model to predict the AME and AAA of soybean meals in Lohmann laying hens, and the predicative models have relatively higher Rcal^2 and minor RMSEP.