Spatial modelling in genetic evaluation of South African Holstein cattle population

African dairy production systems are characterized by small herd size and low genetic connectedness between herds. This situation makes it difficult to accurately estimate environmental and genetic effects. We evaluated how accounting for spatial relationship between neighbouring herds impacts genetic evaluation of 305-days milk yield in South African Holstein cattle population. We analysed 305-days milk yield and pedigree data for 98,632 Holstein cows collected between 1957 and 2014 from 1,145 herds. Herd and herd-year effects were modelled as independent, whereas spatial effects were modelled as correlated between regions using the Besag model. The results show a regional variation of estimated breeding values and separation of spatial effect from herd effect. The spatial modelling of herd as random effect showed similar accuracy compared to modelling herd-year. Further studies integrating the genotype-by-environment interactions will be needed to better explore the benefits from spatial modelling in African smallholder dairy farming systems.