Résumé
The worldwide destructive epidemic of the fungus Mycosphaerella fijiensis on bananas started recently, spreading from Southeast Asia. M. fijiensis is an haploid and heterothallic ascomycete fungus and it spreads through three modes1. The wind-borne spread of ascospores produced during the sexual reproduction is believed to be limited to few hundred kilometers2. Conidia produced during asexual reproduction might be more involved in short-distance dispersal on the plant and to nearby plants. Populations and epidemiological studies of M. fijiensis have been undertaken (i) to provide information on the level and distribution of variability, (ii) to infer on dispersal process of M. fijiensis and (iii) to evaluate the relative importance of evolutionary factors on the pathogen. Population structure of M. fjiensis was analysed from global to plant scales using molecular markers3,4 The results indicate that a high level of genetic diversity is maintained at the plantation and plant levels. The loci were at gametic equilibrium in most of the samples analysed, supporting the hypothesis of the existence of random-mating populations of M. fijiensis, even at the plant level. Southeast Asia has the highest level of genetic diversity, supporting the idea that the pathogen originated in this region. Founder effects were detected at the global and continental scales. Genetic differentiation values between populations decreased with the geographical scale considered, from a high level at global scale (Fst = 0.52 between continents) to a non-significant level at the local scale (Fst = 0 between nearby plantations) (Figure 1). An isolation by distance analysis was conducted in Costa Rica and Cameroon within a production area (around 300 km long) to estimate gene flow and study the dispersal process of the pathogen (D. Coste and colleagues, unpublished results). A strong isolation by distance was detected in both countries, suggesting important dispersal of ascospores on short distances. These results are consistent with those obtained from a direct analysis of disease gradient with a mean dispersal distance of ascospores estimated at a few dozen meters (C. Abadie and colleagues unpublished results). The results obtained could reflect the relative importance of dispersal through infected plant materials and ascospores in relation to geographical scales (Figure 1). The relative importance, on the epidemiology of M. fijiensis, of ascospores and conidia dispersal over short distances on cannot be evaluated from the available data. At a global scale, the introduction of the disease in various continents is probably the result of rare movements of infected plant material. The spread of the disease within. a continent may result either from limited ascospores dispersal over a few hundred kilometers or from the movement of infected plant material. The dispersal of M. fjiensis over long distances appears stochastic, resulting in founder effects, a limited gene flow between established populations and consequently in a high level of genetic differentiation between them. The spread of ascopores may progressively increase when geographical scale decreases from continental to local scales. Thus, the dispersal of M. fijiensis may become more and more gradual leading to a diminution of genetic differentiation and isolation by distance at the intermediate scale (a few hundred kilometers). Population structure of M. fijiensis is now better known. However, such studies should be conducted in Southeast Asia for the different Mycosphaerella leaf spot pathogens in order to identify zones of co-evolution. The results to date also show the effects, on the population structure of pathogens, of genetic recombination, genetic drift and gene flow. A study is being developed at the local scale to evaluate the effect of the selection pressure exerted by the host on the pathogen . (Texte intégral)