Genomic selection in dairy cattle simulated populations

Abstract

Genomic selection is arguably the most promising tool for improving genetic gain in domestic animals to emerge in the last few decades, but is an expensive process. The aim of this study was to evaluate the economic impact related to the implementation of genomic selection in a simulated dairy cattle population. The software QMSim was used to simulate genomic and phenotypic data. The simulated genome contained 30 chromosomes with 100 cm each, 1666 SNPs markers equally spread and 266 QTLs randomly designated for each chromosome. The numbers of markers and QTLs were designated according to information available from Animal QTL (http://www.animalgenome.org/QTLdb) and Bovine QTL (http://bovineqtl.tamu.edu/). The allelic frequency changes were assigned in a gamma distribution with alpha parameters equal to 0·4. Recurrent mutation rates of 1·0e-4 were assumed to apply to markers and QTLs. A historic population of 1000 individuals was generated and the total number of animals was reduced gradually along 850 generations until we obtained a number of 200 animals in the last generation, characterizing a bottleneck effect. Progenies were created along generations from random mating of the male and female gametes, assuming the same proportion of both genders. Than the population was extended for another 150 generations until we obtained 17 000 animals, with only 320 male individuals in the last generation. After this period a 25 year of selection was simulated taking into account a trait limited by sex with heritability of 0·30 (i.e. milk yield), one progeny/cow/year and variance equal to 1·0. Annually, 320 bulls were mated with 16 000 dams, assuming a replacement rate of 60 and 40% for males and females, respectively. Selection and discard criteria were based in four strategies to obtain the EBVs assuming as breeding objective to maximize milk yield. The progeny replaced the discarded animals creating an overlapping generation structure. The selection strategies were: RS is selection based on random values; PS is selection based on phenotypic values; Blup is selection based on EBVs estimated by BLUP; and GEBV is selection based on genomic estimated breeding values in one step, using high (GBlup) and low (GBlupi) density panels. Results indicated that the genetic evaluation using the aid of genomic information could provide better genetic gain rates in dairy cattle breeding programs as well as reduce the average inbreeding coefficient in the population. The economic viability indicators showed that only Blup and GBlup/GBlupi strategies, the ones that used milk control and genetic evaluation were economic viable, considering a discount rate of 6·32% per year.