Regression model for predicting beef carcass cutability in crossbred beef cattle

Abstract

The aim of this study was to propose an equation for predicting beef carcass cutability. Records from 1,257 carcasses of crossbred beef steers and heifers were used. Data of number of permanent incisors teeth, hot carcass weight (HCW), backfat thickness (BFT), ribeye area (REA), ossification score (OS), hump height, marbling score (MS), 48-h post-mortem pH, meat color, and fat color were collected. Phenotypic (Spearman) correlations between the carcass variables were calculated and the collected data were used for the development of multiple linear regression equations to predict boneless carcass yield. Principal component (PC) analysis and non-hierarchical cluster analysis (K-means) of the data were performed to categorize the animals into yield classes of rib, loin, round, and chuck (RLRC) cuts. The model developed successfully predicted (P < 0.01) the RLRC yield (R²=0.94) of the carcasses evaluated. In addition to HCW, BFT and REA, this model included for the first time qualitative variables such as hump height, OS, MS, meat color, and fat color. The model exhibited greater accuracy and precision (R²=0.94; RMSE = 0.30; AIC = 532) than the model that included only HCW, BFT, and REA (R²=0.91; RMSE = 0.37; AIC = 1048]. The data projected on the first two components, PC1 and PC2, explained 26.9% and 10.4% of the variance, respectively, providing five categories of boneless carcass yield similar to the North American model. The proposed equation was able to predict the RLRC yield of crossbred beef carcasses (Bos taurus taurus × Bos taurus indicus) and included qualitative variables that helped to explain variations in boneless carcass yield.