Increasing plant diversity within a crop by the use of cultivar mixtures is a strategy which allows to reduce severity of windborne diseases. Potentialities of this cropping practice have still to be precisely characterized in the case of rain-borne diseases, such as septoria tritici blotch during its spring epidemiological stage. This disease, due to the pathogen fungus Mycosphaerella graminicola, is prevalent on wheat crops and it may be result in substantial yield losses, up to -40%. Field experiments were carried out during five years, from 2008 to 2012, at the Grignon location (Yvelines, France), with a mixture consisting of two wheat cultivars with contrasted resistance to M. graminicola in a 1/3 susceptible/resistant ratio. In comparison with their pure stands, we observed a severity decrease of septoria tritici blotch for the most susceptible cultivar (on average, less 45% of leaf pycnidial leaf surface on the three upper leaf levels), without significantly affect the more resistant cultivar. An original semi-automated methodology was developed to quantify the splash-dispersed spore flux in outdoor conditions. Experimental measurements allowed to correlate intensity of several rainfall events with spore dispersal within different canopies including a cultivar mixture. A mechanistic and stochastic model was implemented in order to describe disease potential progression within a heterogeneous three-dimensional plant canopy. This theoretical approach combines physics and epidemiology in order to, on one hand, (i) compute interception of raindrops with plant organs and the pathway of splash droplets within the canopy and, on the other hand, (ii) take into account cultivar resistance levels and the polycyclism of epidemics. From this model, we highlighted for two-component cultivar mixtures that the proportions and the difference between resistance levels of cultivars to mix together could be optimized in order to reduce disease severity. Furthermore, this modelling approach makes it possible to assess and identify the cultivar spatial distributions the most favourable to a decrease of progression of a splash-dispered disease. Previous studies about potentialities of cultivar mixtures to control splash-dispersed pathogen agents, led in some cases to conclusions with inconsistent and not in favour of this cropping practice. We showed here that it was possible under certain cultivar designing conditions (proportions, spatial arrangement, difference between resistance levels) and rainfall properties to obtain a consistent significant effect in terms of disease reduction.
