[en] A stochastic model was developed to predict the wheat leaf rust (Puccinia triticina Eriks.) severity (percentage of leaf area with symptoms showing uredinia) in four-replicated field experiments located in three villages (Diekirch district: Reuler; Grevenmacher district: Burmerange and Christnach), representative of the different agroclimatological zones of Luxembourg. The model was elaborated by the analysis of the night weather and leaf rust incidence. Statistical validation using regression analysis reports a strong correlation between the number of hours with specific meteorological conditions and the percentage leaf area covered by brown rust lesions for the two upper and youngest leaves, which are mostly responsible for photosynthesis activity and assimilates production filling the grains. The development of the brown rust requires a period of at least twelve consecutive hours with temperatures between 8 and 16°C and a relative humidity (RH) greater than 60%, with optimal values lying between 12 and 16°C and RH greater than 80%.
<br />During the 2004 to 2009 period, at four sites, the linear regression between simulated and observed values for Puccinia triticina was highly significant (P < 0.01) and R2 (coefficient of determination) explained 80 to 85% of the variability. Efforts are now being developed to better define thresholds for fungicide applications and to spatialize the outputs of the model over the entire Luxembourg territory.