Bias and future trends of pig carcass classification methods

Bias of the present formula approved for Spain using the Fat-O-Meater (FOM) grading probe applied to extremely wide subpopulations of pure lines and c rosses. A new equation was calculated adding to FOM backfat and loin depth the percentage lean of the ham (LH), as a new predictor. Biases were calcul ated when applying both equations to the different sub-populations studied. In general, when comparing the differences between estimated and actual le an percentages, the very lean breeds and crosses are underestimated and the very fat ones are overestimated. These differences were lower when the FOM + LH equation was used. The reduction of biases by linear corrections of b oth equations was analysed. The mean square error (MSE) was decomposed into error in central tendency, error due to regression, and error due to distu rbances. For the FOM equation, it was not possible to reduce MSE by means o f linear corrections. However, a reduction of 16.5% of MSE was obtained by linear correction with FOM + LH equation. In practice, LH cannot be include d in a grading scheme because it was obtained by dissection. However, these results indicate that new predictors including measurements of wider anato mical regions should be included in new technological developments. Further more, the meat industry is interested in predicting the lean percentage of the main joints in order to optimise their commercial cutting. (C) 2000 Els evier Science Ltd. All rights reserved.