Remote Sensing For Wheat Powdery Mildew Mbniotring And Quantiifcation Of Conidia In Traps Using Real-time PCR

Wheat powdery mildew, caused by Blumeria graminis f. sp. triciti, is a destructive foliar disease inChina. In this study, hyperspectral canopy reflectance spectra and digital images acquired fromunmanned air vehicle (UAV) was used to detect wheat powdery mildew at two plant density levels in2010and2011. Also the using of canopy reflectance and digital images in wheat yield,1000-kernelweight and protein content estimation was studied. Dynamics in concentrations of Bgt conidia and itsrelationship to local weather conditions and disease index in wheat was analyzed, and disease predictionmodels based on field data was constructed. Airborne spores of Bgt quantification method usingreal-time PCR were developed. These results provide scientific foundation for the application of remotesensing, spore trap and molecular biology technique in Monitoring and Early Warning in wheatpowdery mildew. The main results were as follows:The canopy reflectance of wheat at two plant density levels at anthesis, early milk and late milkstage was influenced by powdery mildew in2010and2011, especially at NIR region. Red edge slope(dλred), the area of the red edge peak (SDr), DVI and SAVI have significant correlations with diseaseindexes at two plant density levels in both years. Disease detection models were constructed based onspectral parameters for both density levels at anthesis, early milk and late milk stage in two years.Moreover, there was no significant difference in slope of the constructed models between two densitylevels in both years. This indicated that plant densities didn't affect the use of spectral parameters inwheat powdery mildew estimation.At normal plant density level, color features and their combinations extracted from digital imagesfrom200m,300m and400m above the ground have significant correlations with disease indexes inboth years. While at low plant density level, only G (green) and I (intensity) have significantcorrelations with disease indexes in both years. There was no significant difference between diseaseindexes estimation models based on I in both years at normal plant density level.Loss of yield,1000-kernel weight and protein content of wheat could reach as high as40%,16%and10%for Jingshuang16(susceptible cultivar) when disease occurred. dλredand the area of the SDr atanthesis stage could be used to estimation wheat yield when powdery mildew occurred. DVI and SDr atlate milk stage could be used to estimation1000-kernel weight and protein content after powderymildew occurred. Color features extracted from digital images200m,300m and400m above theground were significantly correlated with yield,1000-kernel weight and protein content.Conidia increased gradually with time. The highest conidial concentrations in the air wereobserved at milk stage. The concentrations of Bgt conidia in the air were correlated with temperature,solar radiation and wind speed. Time series analysis, using autoregressive integrated moving average(ARIMA)(p, d, q) models, showed that each of the three season's data can be fitted with ARIMA (1,1,0) models. Three sets of models were derived by inoculum only, by weather variables only, and by both inoculum and weather variables to the disease index. And there was no significant difference betweenmodels using inoculum only in2010and2011. Model derived by inoculum are more suitable inproduction practice.After spores were eluted and disrupted from tapes, the extraction efficiency of DNA usingtraditional CTAB method was higher compared with kit method. A significant linear relationshipbetween conidial concentrations counted with a compound microscope and those determined with thereal-time PCR assay was obtained, using the same samples of spore traps. Real-time PCR was specific,accurate and more efficiency when compared with compound microscope. The results demonstrated apotential method to quantitatively determine spore inoculum potential in traps by using real-time PCR.