Submitted to: Journal of Animal Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: January 18, 2006
Publication Date: June 1, 2006
Citation: Williams, C.B., Bennett, G.L., Jenkins, T.G., Cundiff, L.V., Ferrell, C.L. 2006. Using simulation models to predict feed intake: phenotypic and genetic relationships between observed and predicted values in cattle. Journal of Animal Science. 84(6):1310-1316. Interpretive Summary: Use of biological models to predict feed intake could provide the beef cattle industry with a cost effective approach to genetically improve feed efficiency using data frequently collected by the industry. The high genetic correlations between observed and predicted feed intakes obtained in this study indicate that this technology would be potentially useful to estimate breeding values for feed efficiency with nearly the same accuracy, but at a considerable lower cost than breeding values based entirely on individual feeding records. The technology would also facilitate evaluation of a much larger number of animals, and this could potentially have a greater impact on response to selection for feed efficiency than selection based only on individual feeding records. However, high genetic correlations for final weight with observed and predicted feed intakes suggests that the technology needs to be further evaluated in populations with genetic variance in feed efficiency.
Technical Abstract: The objectives of this study were to evaluate the accuracy of the Decision Evaluator for the Cattle Industry (DECI) and the Cornell Value Discovery System (CVDS) in predicting individual DM intake (DMI), and assess the feasibility of using predicted DMI data in genetic evaluations. Observed individual animal data on the average daily DMI (OFI), ADG, and carcass measurements, were obtained from postweaning records of 504 steers from 52 sires (502 with complete data). The experimental data and daily temperature and wind speed data were used as inputs to predict average daily feed DMI (kg) required for maintenance, cold stress, and ADG; maintenance and cold stress; ADG; maintenance and ADG; and maintenance alone, with CVDS (CFRmcg, CFRmc, CFRg, CFRmg, CFRm) and DECI (DFRmcg, DFRmc, DFRg, DFRmg, DFRm). Genetic parameters were estimated by REML, using an animal model with age on test as a covariate, and genotype, age of dam, and year as fixed effects. Regression equations for observed on predicted DMI were, OFI = 1.27 +/- 0.27 + 0.83 +/- 0.04 x CFRmcg [R2 = 0.44, residual SD (sy.x) = 0.669 kg/d] and OFI = 1.32 +/- 0.22 + 0.8 +/- 0.03 x DFRmcg (R2 = 0.53, sy.x = 0.612 kg/d). Heritability of OFI was 0.27 +/- 0.12 and heritabilities ranged from 0.33 +/- 0.12 to 0.41 +/- 0.13 for predicted measures of DMI. Phenotypic and genetic correlations between OFI and CFRmcg, CFRmc, CFRg, CFRmg, CFRm, DFRmcg, DFRmc, DFRg, DFRmg, and DFRm were 0.67, 0.73, 0.41, 0.63, 0.78, 0.73, 0.82, 0.45, 0.77, and 0.86 (P < 0.001 for all phenotypic correlations); and 0.95 +/- 0.07, 0.82 +/- 0.13, 0.89 +/- 0.09 0.95 +/- 0.07, 0.91 +/- 0.09, 0.96 +/- 0.07, 0.89 +/- 0.09, 0.88 +/- 0.09, 0.96 +/- 0.06, and 0.96 +/- 0.07, respectively. Phenotypic and genetic correlations between CFRmcg and DFRmcg, CFRmc and DFRmc, CFRg and DFRg, CFRmg and DFRmg, and CFRm and DFRm were 0.98, 0.94, 0.99, 0.98, and 0.95 (P < 0.001 for all phenotypic correlations), and 0.99 +/- 0.004, 0.98 +/- 0.017, 0.99 +/- 0.004, 0.99 +/- 0.005, and 0.97 +/- 0.021, respectively. The strong genetic relationships between OFI and CFRmcg, CFRmg, DFRmcg, DFRmg, indicate that these predicted measures of DMI may be used in genetic evaluations, and that DM requirements for cold stress may not be needed, thus reducing model complexity. However, high genetic correlations for final weight with OFI, CFRmcg, and DFRmcg suggest that the technology needs to be further evaluated in populations with genetic variance in feed efficiency.