Prediction of powdery mildew epidemic in spring barley

Abstract

Summary Several expressions of severity of leaf area infected by powdery mildew, Erysihpe graminis f. sp. hordei , were regressed on predictor variables, date of observation, growth stage of plant after Feekes, and the 7 day sums of meteorological data either before the date of observation or preceding 7 day the date of observation in spring barley. Development of epidemic better coincided with the sums of meteorological data preceding the date of observation probably due to greater influence of environmental factors before infection than during latent period of disease. We ascertained significant positive influence of date of observation, growth stage of plants and temperature, and adverse influence of sunshine within the course of epidemic. In general, the development was successfully predicted only for severity expressions with Gompertz transformations providing efficient straightening of the course of epidemic. Predictive regressions based on temperature, sunshine, and the development stage of plant were developed. Regressions differed both in significance and in necessary number of predictors according to description of infected area. Gompertz transformation of proportion of leaf area infected by powdery mildew fluffs and necrosis related to field resistance of host plant, was indicated as the best predictable value. The transformation enabled to specify a simple and highly significant linear model regressing powdery mildew occurrence only on 7 days sum of day degrees preceding 7 days the date of observation. Possibilities of further precisions of prediction equations are discussed. It is suggested that more detail analysis of phenomenons underlying regressed variables and more detailed records of environmental predictors may enhance the significance of predictions.