Comparison of Different Methods to Calculate a Total Merit Index - Results of a Simulation Study

Abstract Text: A stochastic simulation study was conducted to analyze four different methods to combine five traits into a total merit index. The reference method was a multitrait evaluation based on raw data. Two methods were variations of an approximate multitrait model; one based on yield deviations the other on de-regressed estimated breeding values (EBVs). The fourth method was an adaptation of the selection index that is used in the routine evaluation in Austria and Germany. Both approximate multitrait models turned out to be very close to the reference model. The selection index method showed good results in most cases but led to a noticeable bias in EBVs in the case of high residual covariances. The results of this study are encouraging to move towards the approximate multitrait approach, but further analyses are necessary before it can be implemented in the routine evaluation.

Keywords:

stochastic simulation

total merit index

genetic evaluation