Development and validation of equations to predict the metabolizable energy value of corn for pigs

: Three experiments were conducted with the aim of developing and validating an equation to predict the ME of corn for pigs from its chemical composition, physical characteristics and particle size. Exp. 1: Eight lots of corn were ground in a hammer mill, using 5 sieves with different screen opening sizes, generating 40 batches of ground corn. The chemical composition (DM, CP, ether extract, crude fiber, ADF, NDF, and ash) and physical characteristics (bulk density- BD and 1,000-kernel weight- TKW) were determined in the 8 lots and geometric mean diameter (GMD) and N-corrected ME (AMEn) were determined in the 40 batches of corn. The AMEn values were determined in 16 metabolism assays with pigs. Mathematical models were adjusted by regression analysis, based on the Akaike Information Criterion. Based on statistical parameters ( = 0.76 and prediction error = 1.05%), number of predictor variables, and easiness of measurements, an equation with 2 segments was chosen: y = 2845.41 + 0.9385 × BD - 20.8784 × CP, if GMD ≤ 522.98 and y = 3105.75 - 0.4978 × GMD + 0.9385 × BD - 20.8784 × CP, if GMD > 522.98. Exp. 2 and 3: Sixty four gilts (Exp. 1; 29.5 ± 3.8 kg) and 64 barrows (Exp. 2; 29.3 ± 3.6 kg), 1 lot of corn, and 3 particle sizes (GMD = 483, 632, and 904 µm) were used in a 3 × 2 factorial arrangement with 2 methods of diet formulation, differing in ME value of corn: "FIX" (value from nutrient composition table) vs. "ESTIMATED" (estimated for each particle size using the equation developed in Exp. 1). In Exp. 2, ADFI was greater ( 0.10) growth performance and carcass traits of barrows. The equation developed was effective to adjust the ME value of corn, considering particle size variation. However, improvement to the proposed equation is necessary to achieve greater precision for predicting corn ME. Validation of the equation with more lots of corn of different chemical compositions and densities appears necessary to assess the efficacy of the equation regarding the variation of the other predicting variables.