The Use Of Chemical Composition And Fourier Near Infrared Spectroscopy To Predict The Net Energy Value Of Soybean Meal For 0-3-week-old Yellow Plumage Broiler

Net energy (NE) values of soybean meals (SM) were measured in 0-21-day-old yellow broiler chicks by a comparative slaughter experiment, the feasibility of predictive models of NE by fourier near infrared spectroscopy and chemical composition were compared. Extend the sample content by adjusting the water content to improve the accuracy of prediction of NE by NIRS was also investigated.(1) NE was calculated as NE for maintenance (NEm) plus NE for deposition(NEp).The NEm were measured by regression method with four treatments, which were group ad libitum to food, groups restricted feeding by 30%,50%,70%, respectively. The NEp of 21 SM were measured by the method of substitution, each SM was assigned as a treatment. Chicks in both trails were 8-d-old (fasted 36 hours), with an average weight of 66.5±2.1g and randomly allotted into four treatments with six replications of two chicks. The experiments lasted 8 days. AME was also measured during the excreta collecting period. (2) Predictive equations for CP, CF, NDF, ADF, ST, AMEs with NEs were derived from the methods of one-dimensional and multivariate linear regression. (3) Each sample with measured NE value was divided into three parts and the water content was adjusted to 11%,12% and 13%. Calibration models of NE were established under the condition of different moisture by near infrared spectroscopy (NIRS). Then the 3 samples were combined into a bigger corrected group to establish the global calibration model based on different moisture contents. The results are as follows:(1) The NE values of 21 soybean meals were 6.045 to 7.829 MJ/kg DM, and efficiencies of utilization of AME for NE were 55.24 to 62.78%. (2) The coefficient of determination (R2) of regression equations from chemical composition and AME combined with chemical composition were 0.96, 0.98, and the RSDs were 0.114,0.079 MJ/kg DM, respectively. (3)The coefficient of determination in calibration (R2cal) and root mean square error of calibration (RMSEE) of NEs were 0.96/0.100,0.98/0.072,0.97/0.107,0.94/0.+105 MJ/kg; The coefficient of determination in cross validation (R2CV) and root mean square error of cross validation (RMSECV) of NEs were 0.92/0.131,0.95/0.096,0.95/0.089,0.93/0.116 MJ/kg.To sum up, (1) The best regression equation from AME combined with chemical composition is superior to the equation from chemical composition. (2) Adjusting the water content to enlarge sample content can achieve satisfactory predictive models by NIRS. (3) The accuracy of the predictive model of NIRS is similar to the best equation from chemical composition.