Designing breeding programs in the genomic era
Item statusRestricted Access
Embargo end date31/07/2022
Increasing the rate of genetic gain of breeding programs is one route to achieve sustainable increases in food production. Breeding has been responsible for ~50% of the increases in agricultural productivity over the past 70 years. However, if we are to sustainably feed a global population of 9 billion people in 2050, breeding programs will need to double the rates of genetic gain that they deliver. Genomic selection is a breeding technology that uses a prediction equation, based on estimated associations between molecular markers and traits of interest, to estimate the genetic values of individuals. Genomic selection makes it possible to directly improve three of the four parameters of the breeder’s equation: (i) the generation interval, (ii) the selection accuracy, and (iii) the selection intensity. Genomic selection can also overcome historical scientific and infrastructural constraints that have limited the use of breeding in different species and agricultural systems. Finally, genomic selection also provides a platform to harness natural synergies between plant and animal breeding. The aim of this thesis was to develop new breeding strategies using genomic selection to increase the rates of genetic gain in both plant and animal breeding programs. Therefore, the first two research chapters of the thesis focused on the deployment of genomic selection in low to middle-income country (LMIC) smallholder dairy cattle genetic evaluations. Across a range of different breeding strategies, genomic selection enabled accurate genetic evaluations. The use of genetic markers was more powerful than pedigree information in capturing and utilising genetic connectedness between smallholder herds. Further, modelling herd as a random effect, in conjunction with genetic markers, enabled animals from very small herds to be included in genetic evaluations. However, the distribution of cattle across different LMIC smallholder dairy herds is complex. Therefore, further simulations were undertaken of three breeding scenarios that varied the distribution of sires and their use across different villages and herds. Spatial modelling increased the accuracy of genetic evaluations by improving the partitioning of the genetic and environmental effects. Therefore, the benefit of spatial modelling increased with increased confounding between genetic and environmental effects. Finally, spatial modelling provided larger increases in the accuracy of genomic evaluations compared to pedigree-based genetic evaluations. The third and fourth research chapters of the thesis focused on the deployment of genomic selection in hybrid crop breeding programs. Hybrid crop breeding programs using a two-part strategy produced the most genetic gain. The two-part strategy used outbred parents to shorten the generation interval of hybrid crop breeding programs. However, the shorter generation interval caused a higher loss of genetic variance per unit of genetic gain. Therefore, a maximum avoidance of inbreeding crossing scheme was required to managed genetic variation over time and increase long-term genetic gain. The complexity of hybrid crop breeding programs enables multiple ways to reallocate resources to implement genomic selection, and the two-part strategy offers even further opportunities. Therefore, further simulations were undertaken to quantify the impact of different resource reallocation strategies on the genetic gain of hybrid crop breeding programs. Two conclusions can be drawn from the results: (i) under a fixed budget, the number of crosses per cycle should be prioritised over the number of selection candidates per cross to maximise long-term genetic gain, and (ii) genomic selection with large levels of resources can mitigate the trade-off between the number of individuals and the number of field measurements per individual in hybrid crop breeding programs. The results from this thesis show that restructuring current breeding strategies to exploit new technology can generate large increases in the rates of genetic gain in both plant and animal breeding programs. Therefore, strategies outlined in this thesis can provide a blueprint for plant and livestock breeding programs to help meet the global demand for food of 9 billion people in 2050.