Commercial batches of soybean meal (SBM) from Argentina, Brazil, and USA (n = 9 per origin) were collected at random from different feed mills in Europe. The AMEn values of the SBM were estimated from their wet chemical composition according to published prediction models (WPSA, 1989; Rostagno et al., 2017; CVB, 2021; FEDNA, 2021). In addition, the apparent ileal crude protein digestibility (dCP) and the AMEn of the samples were determined in vivo in 21 d-old broilers. Recently, a new prediction equation based on the WPSA (1989) model was developed at the Universidad Politécnica of Madrid to improve the accuracy of the estimated AMEn values of commercial SBM. In this approach, the original equation recommended by WPSA (1989) [AMEn (kcal/kg) = 37.5 × crude protein + 46.4 × ether extract + 14.9 × nitrogen free extract] was modified based on published differences in the digestibility of the carbohydrates, protein, and ether extract fractions. The resulting UPM (2022) equation (88% DM basis) was AMEn (kcal/kg) = 69.1 + 43.1 × crude protein × dCP + 55.7× ether extract + 37.3 × (starch + sucrose) + 12.4 × oligosaccharides, where all components are presented in percentages, ether extract was analyzed after previous acid hydrolysis, and dCP values were those determined in the in vivo trial. The estimated AMEn values of the SBM, according to the origin of the beans, and the interactions between origin and the prediction equation used, were analyzed using the MIXED procedure (SAS Institute, 2018). In addition, the Pearson correlation (r) analyses was used to determine the correlation between in vivo dCP (86.5, 86.1, and 87.1% for SBM from Argentina, Brazil, and USA, respectively) and AMEn (2,307, 2,281, and 2,323 kcal/kg for SBM from Argentina, Brazil, and USA, respectively), and the estimated energy values using each individual prediction equation. The AMEn of the SBM varied according to the prediction equation used (P < 0.001) with the greatest value reported for the Rostagno et al. (2017) equation and the lowest for the CVB (2019) equation. Origin of the SBM did not affect the estimated AMEn content of the meals but the response varied according to the prediction model used (P < 0.001 for the interaction). In this respect, the in vivo dCP (r = 0.428; P < 0.05) and AMEn (r = 0.411; P < 0.05) of the SBM were positively correlated with the energy value estimated using the UPM (2022) but not with any of the other equations.
In summary, the variability of current prediction equations to estimate the AMEn of SBM have a wide variability and a poor correlation with in vivo dCP and AMEn values. Prediction equations for SBM should be based primarily on digestible protein followed by sucrose and ether extract contents.