Validation and demonstration of existing Fusarium Head Blight disease forecasters as management decision making tools for use in wheat production in Saskatchewan
Project Code: PRR10-200
Guy Ash and Mike Grenier - Canadian Wheat Board
To transfer and promote adoption of disease forecasting technology for improving timing of fungicide applications and disease management, while reducing unnecessary fungicide use in wheat production in SK
Summary of Results
Fusarium head blight (FHB), caused by Fusarium spp. fungi, is the most destructive and economically important disease affecting wheat production across Canada. The disease reduces yield and grade of wheat grains, and also contaminates kernels with mycotoxins such as deoxynivalenol (DON), making the grain unfit for human and animal consumption Since there are no resistant wheat varieties available commercially, FHB management relies mainly on fungicides sprays leading to significant use of chemicals annually.
Consultations with various wheat industry stakeholders led by Agriculture and Agri-Food Canada's Pesticide Risk Reduction Program identified adoption of disease forecasting as a priority solution for improving the timing and efficiency of fungicide applications in FHB management, while eliminating unnecessary sprays. As such, the present project was set up as a number of on-farm strip fungicide trials and field surveys in wheat crops across Saskatchewan aimed at validating and demonstrating the use of an existing FHB forecasting tool, as part of an integrated approach to manage FHB.
The project was conducted by the Canadian Wheat Board (CWB) staff and supported by the meteorological service provided through the WeatherFarm Inc., also operated and delivered by the CWB. The validation and demonstration work focussed on a FHB risk model developed by Dr. Erick DeWolf at Kansas State University, which was already a forecasting tool available through WeatherFarm and widely used across neighbouring States of USA. It is considered a pre-flowering model that assumes that the pathogen inoculum is present in the field and in high enough concentration to create a moderate to severe outbreak. The model is modulated by three key variables: 1) class of wheat (winter or spring); 2) variety susceptibility level (very poor, poor, fair and good) and 3) the average relative humidity over the last 7 days. A very poor variety and a high average relative humidity value at flowering would lead to a high FHB risk values, i.e. > 80%.
Model validation was conducted through strip fungicide trials set up at 11 (Lemberg, Edgeley and Fairlight regions) and 14 (Indian Head and Lemberg regions) grower and demonstration field sites in 2010 and 2011, respectively. Each grower conducted two field strip trials per year and covered both hard red spring wheat and durum wheat. Each of the chosen sites had a WeatherFarm weather station in close proximity allowing site specific evaluation of the FHB risk model in relation to wheat variety and fungicide performance in the field. Both FHB risk outputs (predicted by the model) and FHB index values (assessed through visual observation in the field: % incidence x % spike area affected/100) were evaluated and compared at 50% anthesis or at the time of fungicide application to determine model performance. The performance categories indicated a ‘Correct’, ‘Under’ or ‘Over’ estimation of the actual FHB index values by the model.
Widespread surveys of wheat crops and harvested grain were also conducted in 2010 and 2011. The same validation approach was used as described above. Wheat heads were monitored for FHB incidence and severity in a total of 6 and 18 fields in respective years. Similarly, a total of 229 and 605 harvest grain samples were collected from growers in 2010 and 2011, respectively, and analysed for incidence of Fusarium Damaged Kernels (FDK), DON levels and grading. Sample data were mapped to assess observed FDK and DON levels against levels predicted by the model.
In addition, growth stage trials were conducted at three locations (Regina, Melfort and Swift Current) to test previously developed thermal time models for simulating spring wheat phenology. The goal was to improve prediction of the susceptibility period for wheat (i.e. anthesis) in order to optimize the timing of fungicide applications when warranted.
Numerous extension activities, focussing on technology and knowledge transfer, were conducted to communicate project results and increase farmer awareness of using advanced tools for informed FHB management decisions under Saskatchewan conditions.
Overall, correlations between FHB risk model outputs and FHB index values observed in the field were significant in areas where there was a previously documented history of disease pressure. In 2010, the model correctly predicted the FHB index value at about 42% of the cases in the strip fungicide trials, while 58% of the time it over-estimated the FHB index value. The tendency for the FHB risk model to over-estimate the disease severity levels in the field is consistent with a pathogen that has recently moved into a geographic region and yet to produce high enough spore concentrations for a moderate to severe outbreak to occur. In 2011, the model correctly predicted the FHB field index value at about 85% of the cases (12 of the 14 sites). In the remaining cases, the model either under-estimated (7%) or over- estimated (7%) disease levels in the field. Across surveyed commercial crops, the model correctly predicted FHB index value at 79% (2010) and 77% (2011) of the time. Under-estimation and over-estimation ranged from 3 to 20% and 18 to 4% of the time in respective years.
Results from harvest samples indicated a good agreement between observed FDK and DON levels and those predicted by the model. These analyses also indicated that fungicide application, while effective in increasing yield, did not consistently improve wheat grade quality. However, fungicide applications were associated with reduced levels of FDK and corresponding DON mycotoxin levels.
Growth stage studies resulted in high correlations, indicating that the developed models explained over 91% of the variability in wheat growth stage. In addition, previous variety performance data had demonstrated that a key management practice for mitigating the risk of FHB is to choose wheat varieties rated as Intermediate to Moderately Resistant.
The project recommends using proper variety selection in combination with FHB risk forecasts and growth models as decision support tools for determining the benefit of fungicide applications for FHB control. All three elements, the FHB risk model, the growth stage model and variety recommendations are now deployed and operational through WeatherFarm as part of an integrated FHB management package in wheat crops. This service allows farmers to customize each model for their specific field management needs. WeatherFarm has a network of about 1,000 weather stations and over 12,000 users across Western Canada. This project has demonstrated the value of a high spatial density and high resolution network to achieve more accurate forecasting performance.
Project results were continuously shared with growers at various field day tours, by presenting at grower meetings and through various media vehicles such as print, radio and webinar forums. Two YouTube videos were also produced and posted on line for grower access. Feedback from farmers indicated that the demonstrations and extension material produced through this project is leading to more growers using forecasting technology which helps them in judicious use of pesticides in their farms.
For more information please contact Guy Ash at Guy_Ash@cwb.ca or Guy_Ash@weatherfarm.com
- Date modified: