Can genomic data enable genetic evaluation with phenotypes recorded on smallholder farms?


  • Owen Powell The Roslin Institute and Royal (Dick) School of Veterinary Studies, University of Edinburgh, Easter Bush Research Centre, Midlothian EH25 9RG, UK
  • Chris R. Gaynor
  • Janez Jenko
  • Gregor Gorjanc
  • Raphael Mrode
  • John M. Hickey


dairy cattle, genetic evaluation, genomic selection, smallholder farms, genetic connectedness


The huge increases in dairy cattle milk yield in advanced economies over the past century is a powerful example of the role of breeding in improving livestock productivity. However, breeding practices have had poor efficacy and penetrance in smallholder farming systems in regions such as East Africa.  Therefore, to meet the­­ continually growing expectations of a more discerning global population for a more varied and nutritious diet, effective dairy cattle breeding programmes need to reach smallholder dairy producers. In advanced economies, large data sets from commercial farms with modest to large herd sizes (e.g. 20 to several thousand cows) and widespread use of AI have provided sufficient animals within each herd and sufficient genetic connectedness between herds. This has enabled the genetic and environmental components of an individual animal’s phenotype to be accurately separated, thus providing accurate genetic evaluations with pedigree information. Typically, herds are neither large nor have high genetic connectedness in smallholder farming systems, such as in East Africa, which limits genetic evaluation with pedigree information. Genomic information keeps track of shared haplotypes rather than animals. This information could capture and strengthen connectedness between herds and through this may enable genetic evaluations based on phenotypes recorded on smallholder dairy farms. The objective of this study was to use simulation to quantify the power of genomic information to enable genetic evaluation under such conditions. The results show; (i) GBLUP produced higher accuracies than PBLUP at all population sizes and herd sizes, (ii) Models with herd fitted as a random effect produced equal or higher accuracies than the model with herd fitted as a fixed effect across all herd size scenarios, (iii) At low levels of genetic connectedness, with four offspring per sire and one to two animals per herd, GBLUP produced EBV accuracies greater than 0.5. Generally, a decrease in the number of sires mated showed consistently higher accuracies compared to when more sires were used. These results suggest that effective breeding programs that use data recorded on smallholder dairy farms in East Africa are possible.

Author Biography

Owen Powell, The Roslin Institute and Royal (Dick) School of Veterinary Studies, University of Edinburgh, Easter Bush Research Centre, Midlothian EH25 9RG, UK

PhD Student, Roslin Institute