Improved bivariate analysis of canola survivability against blackleg disease

Canola ( Brassica napus ) is an important oilseed crop grown extensively worldwide. It is deleteriously affected by the pathogen Leptosphaeria maculans , commonly known as blackleg, causing up to 15% yield loss in Australia annually. The most effective way to manage this disease is by growing resistant varieties. Screening genotypes for blackleg resistance has typically involved deriving percentage survivability against blackleg (from plant counts at emergence and maturity) and conducting a univariate analysis. More comprehensive approaches have involved a bivariate analysis that accounts for the correlation between plant counts. This study highlights shortcomings of the bivariate approach in the literature including independence between genotypes, the restrictive assumption of three-way separable residual models and a constrained selection measure. The research objectives were to (1) incorporate genomic marker information; (2) apply a more flexible residual model; (3) develop a novel selection measure, responsiveness to blackleg disease, from the bivariate regression. To illustrate these advancements, data from disease nurseries within a commercial breeding programme were collected over three years at four locations across Australia. Moderate to strong genetic correlations were found between traits, ranging between 0.49 and 0.91. The incorporation of genomic markers benefitted the maturity count more than emergence count. Furthermore, the more flexible residual model significantly improved model fit in five experiments. Using responsiveness as selection measure produced comparable rankings with the historic survivability measure, with some re-ranking of genotypes which reflects the improved analysis through the bivariate approach. Ultimately, these results demonstrate an improvement over historic analyses, thus encouraging their adoption in canola breeding programmes to accelerate genetic gain for blackleg resistance.