EPIDEMIOLOGY AND MANAGEMENT STRATEGIES FOR WHEAT RUST DISEASE IN DEVELOPING COUNTRIES

Abstract

 Rapid evolving races of pathogens and worsened by climate change still pose a threat to world food security because of the diseases they cause, but the worst part is that they are mostly known to affect those developing nations where wheat is a key source of calorie. The aim of this study was to provide the description of the diversity of pathogen virulence, to test the genomic prediction models in the rust resistance as well as to know the genotype by environment interaction in three different agro-ecological zones of two growing seasons. Phenotype of 210 single-pustule isolates on 30 near-isogenic lines with major resistance genes, and genotype of 200 wheat genotypes was done by 50,000 SNP markers and area under disease progress curve in the field. A cross-validation (five-fold) and a Bayesian logistic regression modelling of environmental covariates were used to benchmark nine genomic selection models. Virulence frequencies were 0.89 or above in Yr27 and Sr31 but Yr15, Sr22 and Lr9 were very effective (nu 0.11). GBLUP model proved to be most predictively correlated (rho = 0.873, RMSE = 0.183) in contrast to random forest (rho = 0.742) and LASSO (rho = 0.811). Multi-trait genomic selection offered 4.6% improvement in prediction as compared to single-trait models and narrow-sense heritability of quantitative resistance, 0.76. The genetic correlations of the environment were 0.32-0.62 that indicates that there was a great interaction of environment genotype. The SNP content of 50,000 was narrowed down to 5,000 markers, decreasing RMSE by 36-48 and the optimal zone of disease development was identified to be between 25-28 C with 75-85 percent relative humidity. The phylogenetic analysis showed that there were three genetic clusters in relation to the agro-ecological zones. These results suggest that, with high-density genotyping, and zone-specific application of useful resistance genes, e.g. Yr15 and Sr22, the application of GBLUP-based multi-trait genomic selection can offer a viable route to long-term rust resistance, although the practical application in resource-constrained areas will involve decentralized breeding facilities and transnational information-sharing consortia on virul.