Energia metabolizável de alimentos determinada in vivo e por métodos indiretos para diferentes categorias de aves

Abstract

The objective of this study was to propose prediction equations for different categories of birds (broilers, spare pullets and laying hens) and to evaluate different prediction methods to estimate the AMEn values of corn, soybean meal, wheat bran and sorghum. To determine in vivo values, three metabolism assays were conducted: Trial I with 390 21-day-old broilers, Trial II with 240 13-week replacement pullets, and Trial III with 156 27-week-old laying hens. The birds received reference feeding, which were replaced by 40% of energy concentrate feed and 30% of protein concentrate feed. The birds were randomly distributed in six replications in each experimental plot, which consisted of five broilers, three spare pullets and two laying hens in their respective metabolic assays. Correlations between chemical components of food and metabolizable energy were estimated and, subsequently, prediction equations were obtained for each bird category by multiple linear regression and the choice of the best-fitted model was based on the Akaike Information Criterion obtained by stepwise procedure. The chemical composition values were used to estimate AMEn by different estimation methods: AMEn Predictor neural networks, Bayesian networks and equations obtained by meta-analysis. The best-fitted overall equation for estimating AMEn of broiler feed was AMEn = 4048,731-14,087PB-196,498FB; for replacement pullets: AMEn = 4238.899-17.02PB + 120.368MM-924.087 calcium-2680.438 phosphorus; and for laying

hens: AMEn = 4156.31 + 35.54PB + 151.12FB + 97.70FDN-223.93FDA-4369.33 calcium- 451.78phosphorus. For energy feeds the best-fitted equation that estimated AMEn for broiler

chickens was: AMEn = 1983.21 + 373.20EE + 746.01 MM-481.93FB + 192.46FDA- 76.69FDN; for replacement pullets: AMEn = -6584.7 + 496.1PB + 3314.2MM + 952.5EE- 1728.5FB-548.0FDN + 1345.5FDA + 8075.7 calcium; and for laying hens: AMEn = 5319.44-

427.74MM-152.65EE + 408.26FB + 181.19FDN-378.29FDA-4630.37 calcium-6227.04 phosphorus. None of the evaluated prediction models provided efficiency for field applications. To improve the accuracy of the methods, it is necessary to add new categories of birds and to update the database used to develop these methods, ensuring that the data used refer to current genetically birds and food.