Improving dry matter intake prediction by incorporating a quadratic body weight term
Translational Animal Science · 2026-11-10
DOI: 10.1093/tas/txag181 ↗Abstract
Abstract Accurate prediction of dry matter intake (DMI) is essential for optimizing growth, using nutrients efficiently, and improving the precision of projected harvest dates in feedlot cattle; however, current equations assume a linear relationship between body weight (BW) and DMI, which may not adequately represent intake patterns in modern feedlot cattle management, heavier finishing-weight cattle. The objective of this study was to evaluate whether incorporating a quadratic effect of BW improves the prediction of DMI under high-concentrate feeding conditions. Data from 24 experiments were compiled and partitioned into training (16 out of 24) and validation (8 out of 24) datasets using a stratified random sampling procedure based on experiment within laboratory to preserve data structure. Prediction equations were developed using mixed model procedures in SAS, including experiment and pen nested within experiment as random effects. Fixed effects included BW, BW², and dietary net energy for maintenance (NEm). Three primary models were evaluated: 1) a linear BW model (model A), a quadratic BW model (BW + BW²; model B)), and a combined model including BW² and NEm (BW + BW2 + NEm; Model C). Model adequacy was assessed by comparing observed (our 3 models plus 3 presented in the nutrient requirements for beef cattle) and predicted values using mean bias, root mean square error (RMSE), coefficient of determination (R²), and concordance correlation coefficient (CCC). Additionally, mean square error of prediction (MSEP) was partitioned to evaluate sources of prediction error. Models were further compared using paired tests of squared prediction errors in an independent validation dataset. The model B reduced prediction error (RMSE = 1.87 kg/d) and minimized bias compared with the linear BW model (RMSE = 1.96 kg/d). Inclusion of NEm with BW², model C, produced similar RMSE (1.89 kg/d) but increased bias. Notably, models with greater R² did not consistently improve predictive accuracy. Overall, the results demonstrate that DMI increases at a decreasing rate as cattle approach heavier weights and that incorporation of a quadratic BW term improves prediction accuracy and better reflects the biological regulation of dry matter intake.
EUVIMED