Traditional agrosystems, where humans, crops and microbes have coevolved over long periods, can serve as models to understand the eco-evolutionary determinants of disease dynamics and help the engineering of durably resistant agrosystems. Here, we investigated the genetic and phenotypic relationship between rice (Oryza sativa) landraces and their rice blast pathogen (Pyricularia oryzae) in the traditional Yuanyang terraces of flooded rice paddies in China, where rice landraces have been grown and bred over centuries without significant disease outbreaks. Analyses of genetic subdivision revealed that indica rice plants clustered according to landrace names. Three new diverse lineages of rice blast specific to the Yuanyang terraces coexisted with lineages previously detected at the worldwide scale. Population subdivision in the pathogen population did not mirror pattern of population subdivision in the host. Measuring the pathogenicity of rice blast isolates on landraces revealed generalist life histories. Our results suggest that the implementation of disease control strategies based on the emergence or maintenance of a generalist lifestyle in pathogens may sustainably reduce the burden of disease in crops.
Plant pathogens often adapt to plant genetic resistance so characterization of the architecture under-lying such an adaptation is required to understand the adaptive potential of pathogen populations. Erosion of banana quantitative resistance to a major leaf disease caused by polygenic adaptation of the causal agent, the fungus Pseudocercospora fijiensis, was recently identified in the northern Caribbean region. Genome scan and quantitative genetics approaches were combined to investigate the adaptive architecture underlying this adaptation. Thirty-two genomic regions showing host se-lection footprints were identified by pool sequencing of isolates collected from seven plantation pairs of two cultivars with different levels of quantitative resistance. Individual sequencing and phenotyping of isolates from one pair revealed significant and variable levels of correlation be-tween haplotypes in 17 of these regions with a quantitative trait of pathogenicity (the diseased leaf area). The multilocus pattern of haplotypes detected in the 17 regions was found to be highly varia-ble across all the population pairs studied. These results suggest complex adaptive architecture un-derlying plant pathogen adaptation to quantitative resistance with a polygenic basis, redundancy, and a low level of parallel evolution between pathogen populations. Candidate genes involved in quantitative pathogenicity and host adaptation of P. fijiensis were highlighted in genomic regions combining annotation analysis with available biological data.