Evaluation de methodes pour estimer la teneur en énergie nette d'aliments destinés aux porcs

The net energy (NE) content of the feed strongly determines pig performances, so that an accurate and rapid estimation is very important. We designed an experiment to evaluate the use of the Dutch feed tables (CVB, 2016), chemical analyses and in vitro organic matter digestibility (OMd) for the prediction of the in vivo NE. Twenty compound feeds with variable nutrient and ingredient composition were formulated and each feed was fed to six pens of three pigs in three weight intervals at an initial mean life weight of 52, 72 and 94 kg, respectively. The faeces were spot-sampled during five days and nutrient digestibility was determined using acid insoluble ash as marker. The compound feeds as well as their ingredients were analysed for chemical composition and in vitro OMd according to Boisen and Fernandez (1997). Besides, the NE content of the 20 feeds was calculated from the NE of the composing ingredients, using either directly the tabular NE-values (CVB, 2016) or the analyzed proximate composition in combination with the tabular digestion coefficients (CVB, 2016), which resulted in a prediction error of 0.34 and 0.19 MJ/kg DM, respectively. Further, multiple regression equations were developed using chemical parameters and in vitro OMd. The in vivo NE content of the feeds was calculated according to the Dutch feed evaluation system (CVB, 2016) and amounted to 10.8 ± 0.9 MJ/kg DM (mean ± sd), varying from 8.3 to 12.0 MJ/kg DM. The sd of the NE due to variation between pens averaged 0.08 MJ/kg DM. The calculated NE content increased with 0.0021 MJ/kg DM per kg higher bodyweight.
The best equation based on chemical parameters (crude fat, starch, ADL and crude ash) resulted in an error of 0.13 MJ/kg DM, and incorporating in vitro OMd slightly decreased this error (0.11 MJ/kg DM).