Studies On The Epidemic And Forecasting And Chemical Control Of Grape Downy Mildew In Shenyang

Grape downy mildew caused by Plasmopara viticola is one of major diseases in grape production. This disease can spread all over the world, and the losses caused by Grape downy mildew is very large every year. Some reports on epidemiology and prediction models had been found, but these prediction models can not predict accurately in different places because of differences in these conditions,such as pathogens, meteorological conditions and grape cultivars et al. This study focuses on the temporal dynamics prediction models of grape downy mildew in Shenyang in order to clarify the law of grape downy mildew epidemic, and provide technical support for the prevention and fungicides control of the disease.The results are as follows:1. Simulated the temporal dynamics of the disease in tested grape cultivars-kyoho, Red-globe,Thompson-seedless-through SPSS19.0.The Logistic model was suitable for describe the temporal dynamics. The Logistic models was established to research the temporal dynamics.which deduced that the exponential phase of the disease of Kyoho was from early July to late July, the logistic phase was from late July to late August, the decline phase was from late August to the end of growing season, that the exponential phase of disease of Red-globe was from late June to late July, the logistic phase was from late July to mid-August, the decline phase was from mid-August to the end of growing season.. that the exponential phase of disease of Thompson-seedless was from late June to late July, the logistic phase was from late July to mid-August, the decline phase was from mid-August to the end of growing season.2. The prediction models based on BP-neural network were established, according to some predictors such as the time, daily temperature,daily relative humidity, accumulated rainfall and accumulated raining days. The number of the hidden layer nodes of the model about this disease of kyoho was 6,while the training error of which was 6.51E-08 and the test error of which was 20.73%,and the prediction accuracy was 79.27%.The number of the hidden layer nodes of the model about this disease of Red-globe was 8, while the training error of was 2.57E-08 while the test error of which was 16.95%,and the prediction accuracy was 83.05%.The number of the hidden layer nodes of the model about this disease of Thompson-seedless was three.while the training error of which was 1.30E-06 and the test error of which was 20.92%,and the prediction accuracy was79.08%.3. All 16 kinds of fungicides were tested in field experiments, the results showed that 52.5% equation-pro WDG performed best in controlling grape downy mildew and the disease control effect of which was 90.0%; followed by 60 pyrazole kresoxim·metiram WDG.and 50% Prochloraz WP performed worst.the control effect of which was 47.05%. Significance analysis showed that the control effect of these eight kinds of fungicides was superior to other tested agents,including 52.5% equation-pro WDG, and 60% pyrazole kresoxim·metiram WDG,and so on.