Evaluating and developing water intake prediction equations for growing and finishing feedlot steers

Abstract

A study was conducted to evaluate 8 published water intake (WI, L/d) equations. Individual WI and DMI were collected on growing (GRW; n = 243) and finishing (FIN; n = 46) feedlot steers using an Insentec Roughage Intake Control System. Published equations were evaluated by predicting WI using 42 d DMI, BW, and weather records. Equation 5 performed with greatest accuracy (mean bias = -1.33; linear bias = -0.14) and precision (R2 = 0.41; RMSE = 6.81) for GRW. Equations 2a (R2 = 0.34; RMSE = 5.46; mean bias = 3.42; linear bias = 0.128) and 2b (R2 = 0.36; RMSE = 5.37; mean bias = 3.44; linear bias = -0.015) performed with greatest accuracy and precision for FIN. A second study was conducted to examine the impact of solar radiation (SR) and DMI on WI and water intake as a percent of body weight (WI%BW) prediction equations. Four WI and 4 WI%BW finishing equations were developed to include all variables (OVRL), include DMI without SR (DMIO), include SR without DMI (SRO), or exclude both DMI and SR (SIMP). Equations were evaluated using an independent dataset (n = 27). The OVRL equations resulted in most favorable regression statistics during development. The DMIO WI (R2 = 0.889; RMSE = 1.220; F-ratio = 74.27) and WI%BW (R2 = 0.890; RMSE = 0.255; intercept = 1.960) models produced more favorable regression statistics compared to SRO and SIMP. The WI%BW equations usually had lower prediction errors and better model fit than WI. During evaluation, DMIO (R2 = 0.67; RSME = 4.87) and SIMP (R2 = 0.64; RMSE = 5.08) WI equations performed with greatest precision. The OVRL WI (mean bias = 2.40; linear bias = -0.09) equation performed with greatest accuracy. All WI%BW equations resulted in similar levels of precision (R2 = 0.57 to 0.59; RMSE = 1.16 to 1.19) except DMIO. The OVRL (mean bias = 0.69; linear bias = -0.05) and DMIO (mean bias = 1.96; linear bias = 0.28) WI%BW equations performed with greatest accuracy. These results indicate that including DMI and SR or only DMI resulted in optimal equation performance.