All experiment layouts were designed using DiGGer ver. 1.0.2 (Coombes, 2016 ). The two RIL population experiments that included check varieties were supra-replicated on block and sub-block basis. Residuals were examined, and if necessary, data were appropriately transformed to meet requirements for residuals to be normally distributed. Residual degrees of freedom are presented for each analysis.
Hermitage RIL population experiments: For the two whole RIL population experiments, RIL with complete data across all replicates was selected for analysis. This provided 173 RIL for the BC*susceptible population and 164 RIL for the BC*tolerant population. Analysis of the proportion of dead seedlings (dead with no pods), dead podded plants, chlorotic or senescent podded plants, non-symptomatic podded plants, and all plants with pods from the final disease assessment was made with a generalized linear mixed model (GLMM) with a binominal distribution logit link and the Wald test. The back-transformed logit values for each RIL were then used for whole-population comparisons among disease and development parameters. For the RIL from the high and low disease classes, ANOVA with RIL nested within the disease class was used to compare P. medicaginis DNA concentrations and disease parameters.
Superior RIL yield loss and inoculum production: A GLMM binominal distribution logit link and the Wald test was used for the analysis of the proportion of dead and chlorotic plants. Grain yield and height reduction data were normalised relative to the metalaxyl-protected control treatment as outlined for the determination of point tolerance responses (Pagan and Garcia-Arenal, 2020 (link)). After the evaluation of a range of models, regression with an exponential function was used to assess the relationship between the proportion of infected plants and normalised yield, and linear regression was used to assess the relationship between other parameters.
All statistical analyses were carried out with GenStat 19th edition (Anon, 2018 ).
Hermitage RIL population experiments: For the two whole RIL population experiments, RIL with complete data across all replicates was selected for analysis. This provided 173 RIL for the BC*susceptible population and 164 RIL for the BC*tolerant population. Analysis of the proportion of dead seedlings (dead with no pods), dead podded plants, chlorotic or senescent podded plants, non-symptomatic podded plants, and all plants with pods from the final disease assessment was made with a generalized linear mixed model (GLMM) with a binominal distribution logit link and the Wald test. The back-transformed logit values for each RIL were then used for whole-population comparisons among disease and development parameters. For the RIL from the high and low disease classes, ANOVA with RIL nested within the disease class was used to compare P. medicaginis DNA concentrations and disease parameters.
Superior RIL yield loss and inoculum production: A GLMM binominal distribution logit link and the Wald test was used for the analysis of the proportion of dead and chlorotic plants. Grain yield and height reduction data were normalised relative to the metalaxyl-protected control treatment as outlined for the determination of point tolerance responses (Pagan and Garcia-Arenal, 2020 (link)). After the evaluation of a range of models, regression with an exponential function was used to assess the relationship between the proportion of infected plants and normalised yield, and linear regression was used to assess the relationship between other parameters.
All statistical analyses were carried out with GenStat 19th edition (Anon, 2018 ).
Free full text: Click here
