Milk Component Yields Versus Concentrations as Selection Criteria to Improve Milk Revenue 1

Quadratic milk pricing formulas based on total yield with concentration differentials for fat and protein can be linearized to price milk based on yields of carrier, fat, and protein. Selection responses for a quadratic milk revenue function and an equivalent linear milk revenue function were compared in each case using in the selection indices those waits in the corresponding milk revenue function. Using observed variance and covariance estimates, selection responses for linear milk revenue functions were an average of 11% greater than responses for corresponding q~adratic milk revenue functions. However, when variances and covariances for milk component yields were predicted from those observed for milk yield and component concentrations, differences in selection responses for linear versus quadratic milk revenue functions averaged only 2.7%. Therefore, it was concluded that inconsistencies among variance and covariance estimates for milk component yields and concentrations are sufficient to explain the observed disparities in selection responses for equivalent linear and quadratic milk revenue functions. Higher levels of interhal inconsistency among trait parameter estimates were found when using averages of available literature estimates than