تجزيه و تحليل زماني اپيدميهاي بلايت نوزاريومي سنبله گندم در شرايط مزرعه

TEMPORAL ANALYSIS OF FUSARIUM HEAD BLIGHT EPIDEMICS OF WHEAT UNDER FIELD CONDITIONS

Epidemiological studies were conducted during two consecutive growing seasons (1999-2000 and 2000-2001), to find the best model that describe disease progress curve of wheat Fusarium head blight (FHB) caused by Fusarium graminearum. Progression of FHB on three cultivars: Falat, Tajan and Zagros and one resistant line (ShA3/SERI/Nanjing 833// Lira) was determined in a complete randomized design with four replicates at Araghi-Mahalleh Research Station in Gorgan. Experiments were performed under mist and no mist system with and without artificial inoculation. The plants were inoculated weekly for five times and the disease severity recorded at least five times every three days. Various epidemiological models including Monomolecular, Logistic, Gompertz, Log-logistic and Weibull were used to evaluate temporal analysis of the disease. The goodness-of-fit for each model was tested by calculating the correlation coefficient (R2) and mean square of errors. The correlation coefficient values for all models in 1999-2000 growing season were low indicating that the disease progression could not be pricisely deternined. In spite of this, according to R2 and error the log logistic model for Falat and Tajan, and the Gompertz model for Zagros were considered as the preferred model. In 2000-2001 experiments under mist and artificial inoculation, the disease progression on Falat and Tajan was best possible using monomolecular model with R2 of 0.966 and 0.828, respectively. However, on Zagros the Weibull (c= 2) with R2 of 0.744 was considered as the preferred model. In the experiments under mist with no inoculation, on Falat the log logistic and on Tajan the monomolecular with R2 of 0.774 and 0.597, respectively were evaluated as the most suitable model. Under the same condition on Zagros, the Logistic (R2= 0.679) exhibited high efficiency. In the experiments with no mist but artificial inoculation, the Gompertz model with R2 of 0.679 was the best fitted model for Falat. On Tajan cultivar, the R2 values were very low in all models and none of them was suitable. In spite of this, the Weibull (c=2) model was better than the others. For Zagros cultivar, the log-Logistic (R2=0.667) was considered to be the best fitted model. Resistant line showed little infection and naturally none of these models were fitted to its data.