Comparing the Prediction Accuracy of Models for Estimating DMI of Beef Heifers Using Surface Area, Volume, Body Measurements or Body Weight by the Elastic net Regularized Regression Method

Abstract Dry matter intake (DMI) remains one of the most variable and significant factors contributing to cost of production in cow-calf systems. Body weight (BW) has been used to estimate DMI of individual cattle as an alternative to measuring DMI, which is expensive, labor intensive, and not practically feasible in production environments. The objective of this study was to evaluate if calculated body measurements [volume (VL) 218 to 503 liters; frame score (FS) 2.1 to 7.5; surface area (SA) 4.02 to 7.12 m2) and direct body measurements (DBM: body length (BL) 89 to 136 cm; hip height (HH) 96 to 136 cm; hip width (HW) 33 to 52 cm; heart girth (HG) 149 to 191 cm; mid girth (MG) 177 to 234 cm; and flank girth (FG) 156 to 239 cm] could be used to estimate DMI of heifers as accurately as using BW (267 to 506 kg). Data from 300 heifers (9 breeds) collected over 4 years were used. Year was included in the model as a random effect. An elastic net model using the glmnet function of the R caret package was used. There were 206 training and 94 testing observations. The models were cross validated 10-fold with 5 repetitions. Models with the least minimum root mean square error and greatest median R-square (R2) values were selected as the best models. Model R2 ranged from 0.19 (SA) to 0.63 (BW), where DBM (0.40) and FS (0.43) performed similarly. As prior studies have suggested, BW in this study accounts for the greatest variation in DMI. However, calculated FS and ABM also explained some variation in DMI and more so than VL (0.22) and SA. These data indicate that FS and ABM could be used as alternatives in cases where BW records are not available to estimate DMI of heifers.