| The energy(NE) of feed ingredients of poultry can be evaluated by respiratory calorimetry method and comparative slaughting method. However, most of the laboratory evaluation take too much of time and require a lot of labor and capital. Besides, the energy of feed ingredients can be predicted by models established by chemical composition and near infrared reflectance spectroscopy(NIRS) analysis technique. Nowadays, the NE of com for duck was not clear. Therefore, the present study was conducted to evaluate the net energy of com for tianfu duck and establish models for predicting net NE of com using chemical composition and NIRS.36 com samples with a variety content of chemical compositions were collected, with which NE and AME were experimentally evaluated. At the age of 7 days, a total of 860 tianfu ducks with an average weight of (125.01±1.33) g were selected for the experiment which was lasted for 7 days. A factorial analysis procedure was used to estimate net energy of com that combines net energy for maintenance(NEm) and production(NEp). A complete formula feed was used for the experiment of NEm, in which 120 ducks were used. At the beginning of the experiment of NEm,20 ducks were slaughtered as a control. NEm was estimated by fasting heat production(FHP). The FHP was measured by establishing regression equation and extrapolating HP measured at five feeding levels including ad libitum and restricted feeding by 20%,32%,44%,54% to zero ME intake. The ducks in experiment of NEp were feed different com samples directly. NEp was calculated using the comparative slaughter technique.740 ducks were selected for the experiment of NEp, from which 20 ducks were killed to obtain baseline body energy at the beginning of the experiment and others were randomly allotted to 36 com samples. Each com was provided to 20 ducks housed in cages with four ducks per repetition and five replications. At the end of experiment all ducks were slaughtered, and performance were determined. Then 36 corns were measured for their CP, EE, CF, NDF, ADF and Ash. Correlation and multiple linear stepwise regression analyses were conducted to establish the models for predicting NE of com by chemical compositions. After spectra were collected, NIRS models for predicting NE of com were established using partial least square regression(PLS) by the OPUS software ver.5.5.Results showed that the variations in nutrients were large between different sources of the com samples, and AME of corn on tianfu duck were (13.47±0.28) MJ/kg. The NEm and NEp of duck provided by com were (8.49±0.35) MJ/kg and (0.95±0.05) MJ/kg, respectively. The NE of corn were (9.44=0.35) MJ/kg, which contributed (70.12±2.00)% of the apparent metabolizable energy. The best prediction model established by chemical composition was NE=6.956-0.13NDF+0.446EE-0.264ADF+0.241AME, in which R2 and RSD were 0.923 and 0.03 MJ/kg, respectively. In NIRS model, the coefficient of determination of internal cross-validation(R2cv), coefficient of determination of calibration(-R-cai) and coefficient of determination of extemal validation (R2vat) were 85.99%,90.42%and 82.62%, respectively; the root mean square error of internal cross-validation (RMSECV), the root mean square error of calibration (RMSEE) and the root mean square error of external validation (RMSEP) were 0.068 MJ/kg.0.073 MJ/kg and 0.086 MJ/kg, respectivelyIn conclusion, the NE of com on tianfu duck were (9.44±0.35) MJ/kg, and it is feasible to establish NE prediction model of corn using its chemical composition and NIRS. |
