GENOTYPE X REGION INTERACTIONS AND IMPLICATIONS FOR SUGARCANE BREEDING PROGRAMS

AUSTRALIAN sugarcane breeding programs in the past have maintained fairly separate breeding and selection programs targeting different regions. This structuring implied an assumption that genotype x region interactions are important relative to genotype x environment interactions within regions, but this had not been tested. A research project was initiated in 1997 to help address this issue and the implications arising for breeding programs. It was found that genotype x region interactions were small relative to genotype main effects and GE interactions within regions for both cane yield and CCS. This is consistent with trials from different regions generally having genetic correlations that were not much less (average 75% and 85% for cane yield and CCS respectively) than correlations between trials within the same region. In general, these results support a move toward a more integrated breeding program in Australia. More specifically, this should involve data analysis methodology that takes account of genetic correlations between regions for different traits (cane yield, CCS) and whereby results from all regions are combined and clone performance is predicted for all regions. Clones with highest predicted value for regions different to those in which field trial testing has already occurred should be directly included in final assessment trials in those new regions. This procedure would contrast to the practice used in the past of placing such clones back into single row plots before deciding on advancement to final assessment trials or not. These approaches should result in faster genetic gains in the Australian sugarcane breeding program overall, and faster commercial release of cultivars across all regions to which they are suited.