Forecasting the national incidence of sugar-beet powdery mildew from weather data in Britain

In a national survey of the incidence of powdery mildew (Erysiphe betae) in sugar beet in Britain, between 1980 and 1989, in which between 335 and 717 fields were inspected at the end of August each year, the average crop area infected ranged from 2% in 1986 to 68% in 1989. The effect of various climatic factors that might be expected to influence powdery mildew development over this period was examined, using meteorological records from Broom's Barn Experimental Station, Using simple linear regression analyses, a significant association was found between disease incidence and both air temperature and rainfall incidence in the period April-August, However, air temperature and frost incidence in some winter months were also closely associated with the incidence of powdery mildew in the following crop.
Combinations of these variables were examined in stepwise multiple-regression analyses, The model of best fit, incorporating winter frost incidence with summer temperature and rainfall incidence, accounted for 95% of the variation. Model validation procedures demonstrated the superiority of this model over those involving single variables. Nevertheless, for forecasting purposes, the best correlation was found between the crop area infected at the end of August each year and the number of days with a ground frost in the preceding February and March. This single climatic variable accounted for 82% of the variation and its potential use to forecast powdery mildew incidence in the national sugar-beet crop is discussed.