Assessing the Accuracy of Simulation Model for Septoria Leaf Blotch Disease Progress on Winter Wheat.

Abstract

A mechanistic model, PROCULTURE, for assessing the development of each of the last five leaf layers and the progress of Septoria leaf blotch, caused by Septoria tritici (teleomorph Mycosphaerella graminicola), has been applied on susceptible and weakly susceptible winter wheat (Triticum aestivum) cultivars in two locations (Everlange and Reuland) in Luxembourg over a 3-year period (2000 to 2002). A double performance assessment of PROCULTURE was conducted in this study. First, the capability of PROCULTURE to correctly simulate S. tritici incidence was checked. Second, the model's ability to accurately estimate disease severity was assessed on the basis of the difference between simulated and observed levels of disease development at each leaf layer. The model accurately predicted disease occurrence in the 2000 and 2002 seasons, on susceptible and semi-susceptible cultivars, with a probability of detection (POD) exceeding 0.90. However, in 2001, even though the POD never fell below 0.90, the false alarm ratio (FAR) was too high to consider the simulations satisfactory. Concerning the evaluation of disease severity modeling, statistical tests revealed accurate simulations performed by PROCULTURE for susceptible cultivars in 2000 and 2002. By contrast, for weakly susceptible cultivars, the model overestimated disease severity, especially for the upper leaves, for the same period.