Abstract:
Genomic selection (GS) can substantially reduce breeding cycle times in forest trees compared to traditional breeding cycles. Practical implementation of GS in tree breeding requires an assessment of significant drivers of genetic gains over time, which may differ among species and breeding objectives. We present results of a GS study of growth and wood quality traits in an operational Eucalyptus grandis breeding program in South Africa. The training population consisted of 1575 full and half-sib individuals, genotyped with the Eucalyptus (EUChip60K) SNP chip resulting in 15,040 informative SNP markers. The accuracy of the GS models ranged from 0.47 (diameter) to 0.67 (fibre width). We compared a 4-year GS breeding cycle equivalent to half of a traditional 8-year E. grandis breeding cycle and obtained GS efficiencies ranging from 1.20 (wood density) to 1.62 (fibre length). Simulated over 17 years, the ratio of the accumulated genetic gains between three GS cycles and two traditional breeding cycles ranged from 1.53 (diameter) to 3.35 (wood density). To realise these genetic gains per unit time in E. grandis breeding, we show that significant adjustments have to be made to integrate GS into operational breeding steps.
