Impact of different strategies and levels of preferential treatment on different methods of bull dam selection
Three milk, fat yield and final score type records were simulated for each cow in 20 herds of 200 cows over 13 years. At completion of the simulation, cows were ranked by different selection methods and the top 2% were chosen to be bull dams. Preferential treatment was simulated by increasing milk and fat yields by 8, 16, 32, and 48% in separate copies of the simulation. Preferential treatment was given to a limited number of cows in copies of the original herds based on 8 strategies. Cows were chosen to receive preferential treatment for 2nd and 3rd records based on phenotypic records and ETA’s alone and in combination with a phenotypic minimum for final score type. Cows were also chosen to be biased in all records based on phenotypic records of dam, parent average ETA, maternal line and final score type. Bull dam selection methods compared used 2:2:1 milk:fat:type indexes of cow Predicted Transmitting Abilities (PTA), first lactation PTA, (PTA-F), PTA after requiring phenotypic minimums, (PTA-P), 3-generation Pedigree Index (PI-3) and PTA after preselection on 3-generation Pedigree Index (PI-3/PTA). Comparison criterion was average merit on a 2:.2:.1 weighting of true transmitting abilities for milk, fat and type for cows selected in each of 3 replicates of the cow population that were started with different random number seeds. Selection methods PTA and PTA-F gave the highest average true breeding values when no bias was present, and both methods were robust to bias levels of 8 and 16% mean response, and continued to give the best results at these levels for all bias patterns studied. In general, selection on PI-3 and PTA-P gave poor results and should not be considered viable selection methods. Selection ability of PTA was greatly decreased at the 32 and 48% bias levels. Selection on PTA-F continued to be effective when bias did not occur in the first lactation or when bias was based on type score, while selection on PI-3PTA was unaffected by bias at any level. Requiring a high level of 3-generation Pedigree Index before selection on PTA appears useful for selecting bull dams when very high levels of bias are present.