Analyses of large-scale population structure of pathogens enable the identification of migration patterns, diversity reservoirs or longevity of populations, the understanding of current evolutionary trajectories and the anticipation of future ones. This is particularly important for long-distance migrating fungal pathogens such as Puccinia striiformis f.sp. tritici (PST), capable of rapid spread to new regions and crop varieties. Although a range of recent PST invasions at continental scales are well documented, the worldwide population structure and the center of origin of the pathogen were still unknown. In this study, we used multilocus microsatellite genotyping to infer worldwide population structure of PST and the origin of new invasions based on 409 isolates representative of distribution of the fungus on six continents. Bayesian and multivariate clustering methods partitioned the set of multilocus genotypes into six distinct genetic groups associated with their geographical origin. Analyses of linkage disequilibrium and genotypic diversity indicated a strong regional heterogeneity in levels of recombination, with clear signatures of recombination in the Himalayan (Nepal and Pakistan) and near-Himalayan regions (China) and a predominant clonal population structure in other regions. The higher genotypic diversity, recombinant population structure and high sexual reproduction ability in the Himalayan and neighboring regions suggests this area as the putative center of origin of PST. We used clustering methods and approximate Bayesian computation (ABC) to compare different competing scenarios describing ancestral relationship among ancestral populations and more recently founded populations. Our analyses confirmed the Middle East-East Africa as the most likely source of newly spreading, high-temperature-adapted strains; Europe as the source of South American, North American and Australian populations; and Mediterranean-Central Asian populations as the origin of South African populations. Although most geographic populations are not markedly affected by recent dispersal events, this study emphasizes the influence of human activities on recent long-distance spread of the pathogen. Domestication of ecosystems, climate change and expanding global trade have accelerated the pace of disease emergence, caused by their introduction into new areas with susceptible hosts or the spread of new damaging pathogen genotypes. The wheat yellow rust pathogen (PST) is a pathogen with recent reports of invasions, significantly affecting worldwide wheat production. However, its origin and ancient migration routes remained unclear and the source areas of newly spreading strains were largely unknown. This information is important to understand the trajectories of current invasions and forecast the future spread, and more generally develop risk-assessment models of pathogen emergence. We analyzed a set of worldwide representative isolates of PST, which enabled to identify six different area-specific populations. Using population genetics tools, we identified its centre of diversity in the Himalayan and near Himalayan region. We also identified the source of the recently emerged populations; Middle East-East Africa as the source of high-temperature-adapted strains spreading worldwide; Europe as the source of American and Australian populations; and Mediterranean-Central Asian populations as the origin of South African populations. We demonstrate the influence of human activities on the recent long-distance spread of the disease, though most geographic populations are not markedly affected by recent dispersal events.

We investigated whether the recent worldwide epidemics of wheat yellow rust were driven by races of few clonal lineage(s) or populations of divergent races. Race phenotyping of 887 genetically diverse Puccinia striiformis isolates sampled in 35 countries during 2009-2015, revealed that these epidemics were often driven by races from few but highly divergent genetic lineages. PstS1 was predominant in North America; PstS2 in West Asia and North Africa; and both PstS1 and PstS2 in East Africa. PstS4 was prevalent in Northern Europe on triticale; PstS5 and PstS9 were prevalent in Central Asia; whereas PstS6 was prevalent in epidemics in East Africa. PstS7, PstS8 and PstS10 represented three genetic lineages prevalent in Europe. Races from other lineages were in low frequencies. Virulence to Yr9 and Yr27 was common in epidemics in Africa and Asia, while virulence to Yr17 and Yr32 were prevalent in Europe, corresponding to widely deployed resistance genes. The highest diversity was observed in South Asian populations, where frequent recombination has been reported, and no particular race was predominant in this area. The results are discussed in light of the role of invasions in shaping pathogen population across geographical regions. The results emphasized the lack of predictability of emergence of new races with high epidemic potential, which stresses the need for additional investments in population biology and surveillance activities of pathogens on global food crops, and assessments of disease vulnerability of host varieties prior to their deployment at larger scales.

Understanding of long-term virulence dynamics of pathogen populations in response to host resistance gene deployment is of major importance for disease management and evolutionary biology. We monitored the virulence dynamics of Puccinia striiformis f. sp. tritici, the causal agent of wheat stripe rust, over 25 years in France. Virulence dynamics was explained by estimates of area associated with resistance genes carried by farmers' cultivars. The epidemics assessed through disease severity significantly correlated with the number of P. striiformis f. sp. tritici isolates collected each year, used to describe virulence dynamics. In the south, the dominance of the Mediterranean pathotype 6E16 and the cultivation of a susceptible cultivar were associated with an epidemic from 1997 to 1999. In the north, five epidemics occurred due to successive acquisition of virulence to the resistance genes Yr7, Yr6, Yr9, Yr17, and Yr32, either by acquisition of the virulence in the previous dominant pathotype or by incursion or selection of one or two new pathotypes. Frequency of pathotypes with Vr7 and Vr6 declined with the reduction in the cultivation of corresponding Yr gene cultivars, whereas the virulence Vr9 persisted longer than the cultivation of Yr9 cultivars. Although the first pathotypes carrying Vr9 decreased, this virulence persisted in other pathotypes even in the absence of Yr9 cultivars. At the regional level, Yr9 cultivars in the north caused a shift from high Vr6 frequency to high Vr9 frequency whereas, in the central region, where Yr9 cultivars were rare, Vr6 remained prevalent.