Predicting epidemics of yellow rust (Puccinia striiformis) on the upper canopy of wheat from disease observations on lower leaves

Data from field experiments were used to test whether disease observations on lower leaves of wheat were good predictors of future epidemic development in the upper canopy. Fungicide treatments to replicated plots in 1994 and 1995 caused variation in levels of initial inoculum of urediniospores of yellow rust ( Puccinia striiformis ). Observations of symptom severity were made on individual leaf layers throughout the season. Sporulating lesions on lower leaves were considered to be a measure of inoculum source strength for transfer to upper culm leaves. Low source strengths were associated with delays in epidemic development on the upper leaves, as quantified by a location parameter, t _m, the time at which severity reached half of the asymptote value in logistic fits to disease progress data on the upper leaves. However, there was much unexplained variation, probably due to extraneous variation in upward transfer efficiency of inoculum and rates of epidemic development. Vanderplank's sanitation ratio theory was used to account for variation in inoculum transfer and epidemic rate. The analysis revealed that if the aim is to predict disease on a newly emerged culm leaf, during the period when it can be treated effectively with fungicide, then observations of disease two leaves further down the culm were of the greatest predictive value ( R ² = 86%). These observations, however, need to be integrated with information about factors affecting upward inoculum transfer and rates of epidemic development if acceptable predictive precision is to be achieved.