221 Quantifying the relationship between fasting heat production and chemical empty body composition in sheep and cattle

Previous research indicates that animals of similar weight but greater protein mass have greater metabolic rate. The objective was to quantify the relationship between fasting heat production (FHP) and empty body composition in ruminants. A literature search was conducted to compile data on FHP and empty body composition. Seven studies using sheep or cattle consisting of 49 treatment means were found reporting FHP and chemical empty body composition. Data were analyzed with R statistical packages using mixed model methodology with study as a random variable. Given the strong correlations (r > 0.7) and high degree of multicollinearity (VIF > 30) among EBW, empty body protein (EBP) and empty body fat (EBF), LASSO regression was used to reveal that EBP was the important predictor of FHP. Allometric models (a*X^b) gave significant (P < 0.001) values for a and b of 74.3 ± 10.4 and 0.74 ± 0.02 for X = EBW (R2 = 0.963, RMSE = 432 kcal, AIC = 737), 227.7 ± 21.1 and 0.86 ± 0.02 for X = EBP (R2 = 0.972, RMSE = 375 kcal, AIC = 724), and 270.8 ± 75.9 and 0.75 ± 0.07 for X = EBF (R2 = 0.702, RMSE = 1219 kcal, AIC = 839), respectively. Log transformed models (lnFHP = lnX) gave significant (P < 0.001) values for intercept and slope of 4.47 ± 0.13 and 0.69 ± 0.02 for X = EBW (R2 = 0.979, RMSE = 0.088 lnkcal, AIC = -62.6), 5.61 ± 0.11 and 0.74 ± 0.03 for X = EBP (R2 = 0.973, RMSE = 0.096 lnkcal, AIC = -45.1), and 6.44 ± 0.19 and 0.33 ± 0.02 for X = EBF (R2 = 0.894, RMSE = 0.125 lnkcal, AIC = -13.8), respectively. A log transformed model including both EBP and EBF resulted in significant intercept (5.79 ± 0.14; P < 0.0001), lnEBP coefficient (0.56 ± 0.08; P < 0.0001) and lnEBF coefficient (0.09 ± 0.04; P = 0.02) with VIF of 8.3. The R2, RMSE and AIC of this model were 0.970, 0.088 lnkcal and -43.5; not improved over the model with EBP alone. In conclusion, EBP explained the variation in FHP as well or slightly better than EBW, and EBF did not significantly improve the prediction.