Genotype × Environment Interaction for Grazing Versus Confinement. I. Production Traits

Abstract

The objective of this study was to investigate the possible existence of a genotype × environment interac- tion (G×E) for production traits of US Holsteins in graz- ing versus confinement herds. Grazing herds were de- fined as those that utilized grazing for at least 6 mo and were enrolled in dairy herd improvement (DHI). Control herds were confinement DHI herds of compara- ble size in similar regions. The performance of daugh- ters in grazing herds and control herds was examined using linear regression of mature equivalent milk, fat, and protein yield on the November 2000 USDA-DHI predicted transmitting abilities (PTA) of their sires for those traits. Heritabilities and genetic correlations were estimated using restricted maximum likelihood in a bivariate animal model that considered the same trait in different environments as different traits. Prod- uct-moment and rank correlations were calculated be- tweensires' estimatedbreeding values,estimated sepa- rately in both environments. For grazing herds, the coefficient of regression of milk, fat and protein on PTA were 0.78, 0.76, and 0.78, respectively. Corresponding coefficients in the control herds were 0.99, 0.96, and 0.98.Estimatesofheritabilityforthetraitsrangedfrom 0.2 to 0.25, and differences between grazing and control environments were small. Estimates of the genetic cor- relations for the traits in both environments were 0.89, 0.88, and 0.91 for milk, fat, and protein, respectively. Within-quartile analyses revealed a lower correlation for milk and protein between the upper and lower graz- ing quartiles, while the same quartiles for the control herds did not differ from unity. Rank correlation coeffi- cients between sire estimated breeding values from the 2 environments were 0.59, 0.63, and 0.66 for milk, fat, and protein, respectively. The mean rank change for the top 100 sires between the two environments was 27. The regression coefficients indicate that expected daughter differences may be overstated by current sire PTA in grazing herds. Genetic correlations less than unity suggests that there is, at least, some reranking amongsiresinbothenvironments,whiletherankcorre-