Realized genetic selection differentials in Canadian Holstein dairy herds.

Estimated breeding values for a selection index, production, durability, health, and fertility traits for Canadian Holstein bulls and cows born from 1950 and 1960, respectively, were used to define and determine realized genetic selection differentials (GSD) and generation intervals along the 4-path model of genetic improvement and the variability among herds in realized GSD. The effects of some non-genetic factors on realized GSD were also determined. The mean generation intervals of the sire-to-bull (SB), dam-to-bull (DB), sire-to-cow (SC), and dam-to-cow (DC) pathways reduced, respectively, from 9.7, 7.5, 7.7, and 4.7 years in 1980 to 2.3, 2.5, 4.8, and 3.6 years in 2016. The realized GSD of lifetime performance index and 305-d milk, fat, and protein yields for the SB and DB paths were greater than those of the SC and DC, and realized GSD for DB were increasing for all production traits and the index after 2009. With the exception of daughter fertility (DF), we found zero to negative selection in the DC path for all the traits studied, due to the minimal opportunity for selection in that path. No clear trends were observed in realized GSD of SB, DB, or SC paths for mammary system, dairy strength, feet and legs, or herd life. The realized GSD for DF in the SB, DB, and SC remained largely negative, except after the early 2000s when positive realized GSD were observed in the SB and DB paths. The realized GSD for DF in the DC path remained positive throughout the period of the study, though mostly non-significantly different from zero, except for 2014 and 2015. Realized GSD for somatic cell score in the SB, DB, and SC paths showed increasing and favorable trends. Year of conception, housing system, agricultural region, and their interactions had significant effects on realized GSD of some traits in the SC and DC paths. We also observed considerable variations in realized GSD among herds. The population mean realized GSD and those of the top and bottom 10% of herds could serve as benchmarks that individual herds could use to monitor their past selection and make changes if they are not moving in the intended direction.