Novel equations for estimating gross energy in feed ingredients for non-ruminants

Abstract

Objective: The present study aimed to evaluate the accuracy of previous equations for estimating gross energy (GE) in feed ingredients and to develop the novel equations.

Methods: A total of 2,279 ingredient samples consisted of barley (n = 58), corn (n = 319), corn distillers dried grains with solubles (n = 13), corn gluten feed (n = 583), copra expellers (n = 156), copra meal (n = 234), cottonseed meal (n = 12), palm kernel expellers (n = 504), rapeseed meal (n = 114), soybean meal (n = 138), wheat (n = 70), and wheat bran (n = 78) were analyzed for dry matter (DM), crude protein (CP), ether extract (EE), crude fiber, ash, and GE. The 2,279 ingredient samples were used for evaluating the previous equations and developing novel equations. Using data from 62 ingredients in the swine NRC publication in 2012, the old equations and the novel equations were evaluated.

Results: Based on the evaluation using 2,279 samples, the equation developed by Ewan in 1989 underestimates GE by 218 kcal/kg DM (standard error = 4 and p<0.001) on average and underestimates more for low-GE ingredients (linear bias = -0.121; standard error = 0.025 and p<0.001). The equation reported by Sauvant, Perez, and Tran in 2004 also underestimates GE by 135 kcal/kg DM (standard error = 4 and p<0.001) on average. Novel equations for estimating GE concentration (kcal/kg DM) in feeds were developed: GE = 4,299+7×CP +53×EE, with R2 = 0.342 and p<0.001; GE = 4,341+11×CP+54×EE-24×ash, with R2 = 0.372 and p<0.001, where all independent variables are in % DM. In the validation using 62 feed ingredients in the NRC publication, the equations developed in the present study were accurate whereas the previous equations were not.

Conclusion: The novel equations developed in the present study fairly accurately estimate gross energy concentrations in concentrate feeds.

Keywords: Equation; Feed Ingredient; Gross Energy; Validation.

Figure 1

Evaluation of equations reported by Ewan [10] and Sauvant et al [11] for estimating gross energy (GE) concentrations (kcal/kg dry matter; DM) in feed ingredients. A regression analysis was performed for measured GE minus predicted GE on predicted GE minus average of predicted GE. (a) For the equation of Ewan [10], the intercept (218; standard error = 4 and p<0.001) and the slope (–0.121; standard error = 0.025 and p<0.001) were different from 0 based on 2,279 ingredient samples. (b) For the equation of Sauvant et al [11], the intercept (135; standard error = 4 and p<0.001) was greater than 0 but the slope (0.041; standard error = 0.028 and p = 0.143) was not different from 0 based on 2,259 ingredient samples.

Figure 2

Validation of equations developed in the present study for estimating gross energy (GE) concentrations (kcal/kg dry matter; DM) in feed ingredients using 62 ingredients in the NRC [13]. A regression analysis was performed for measured GE minus predicted GE on predicted GE minus average of predicted GE. (a) For the Eq. 2, the intercept (–7; standard error = 47 and p = 0.884) and the slope (–0.018; standard error = 0.089 and p = 0.840) were not different from 0. (b) For the Eq. 3, the intercept (–56; standard error = 38 and p = 0.140) and the slope (0.014; standard error = 0.068 and p = 0.834) were not different from 0.

Figure 3

Evaluation of equations reported by Ewan [10] and Sauvant et al [11] for estimating gross energy (GE) concentrations (kcal/kg dry matter; DM) in feed ingredients. A regression analysis was performed for measured GE minus predicted GE on predicted GE minus average of predicted GE. (a) For the equation of Ewan [10], the intercept (101; standard error = 35 and p = 0.005) was greater than 0 but the slope (–0.066; standard error = 0.057 and p = 0.251) was not different from 0 based on 62 ingredients in the NRC [13]. (b) For the equation of Sauvant et al [11], the intercept (71; standard error = 36 and p = 0.056) tended to be greater than 0 but the slope (–0.007; standard error = 0.063 and p = 0.907) was not different from 0 based on 46 ingredients in the NRC [13].