A stochastic simulation study for the comparison of different methods to calculate a total merit index

Absztrakt

In order to compare four different methods for calculation of a total merit index, a stochastic simulation study was conducted. Five normally distributed traits were chosen to represent the blocks dairy, beef and fitness of a simulated cattle population. The reference method was a full multivariate evaluation based on raw data. The other three methods were based on selection index theory with different approaches to calculate covariances between estimated breeding values. Additionally a focus was put on the implications of varying the residual covariances between traits. All selection index methods showed similar results. However, the method currently used in the joint genetic evaluation led to noticeable biases in EBVs especially when residual covariances between traits were high. Residual covariances seem to have an important impact when calculating a total merit index and should not be ignored. Results of the present study encourage to move towards a multitrait approach or at least to account for residual covariances when combining EBVs into a total merit index.