By Top Crop Manager
Retooling fusarium forecasting.
Roughly 40 percent of Ontario's 2004 wheat crop was graded feed due to fusarium
infection. Only 20 percent of the crop received a fungicide application. That
has prompted the research community to take a hard look at the DONcast prediction
model, a tool available to help growers make effective spray decisions.
In 2004, the model was wrong more than it was right. During the critical infection
period, the Fusarium Risk Maps predicted only a slight risk of fusarium development.
Given its dependency on accurate weather forecasts, that is not surprising.
"DONcast uses some forecasted weather data and considers a 70 percent chance
of precipitation a significant indicator of a 'rain day'," says Dr. David
Hooker, researcher for the University of Guelph at Ridgetown. "In 2004,
however, even a 30 percent chance of precipitation turned out to be a significant
Hooker surveyed 115 wheat fields at harvest across southern Ontario. He used
actual weather data around heading time, tillage, previous crops and wheat variety
information specific for each field. When using the actual weather, DONcast
correctly predicted DON concentrations greater than one part per million in
49 of 56 fields and correctly predicted DON concentrations less than one part
per million in 36 of 59 fields. From 1996 to 2003, he sampled more than 700
fields and prediction accuracies were between 80 and 90 percent.
Accurate weather data the missing link
In 2004, the accuracy of the prediction model was hit with a double whammy.
First, the model does not predict a fusarium outbreak at heading unless the
weather forecasts include events that would create a climate perfect for disease
development. "It's designed not to cry wolf unless it sees a wolf,"
says Ian Nichols, business manager at the Ontario Weather Network (OWN). And
secondly, the Environment Canada forecasts used in the equation failed to call
for rains when they actually occurred.
Nichols has been working with the fusarium prediction model since its inception.
"Maybe by providing an array of maps well ahead of actual heading dates,
we are doing a disservice," says Nichols. "Growers who looked at it
a week to 10 days before heading saw green and decided there was no problem
and never checked back. If they had looked again closer to the decision day,
the map would have said spray."
The model incorporates weather information from seven days before heading to
10 days after heading. Weather forecasts are only available for 10 days in the
future for temperature, five days for rain and only 36 hours for relative humidity.
For a spray decision at three days after heading, DONcast would look for weather
forecasts seven days after spray day. Unfortunately, rain data would be unavailable
for two of those days and relative humidity would be missing for six of those
days. DONcast interprets that as two days without rain and a lengthy six day
period of low relative humidity. The map would be green. Historical data suggests
there is an excellent chance of at least two of those days having high humidity.
So, because the information is unavailable to the model, the forecast is misleading.
The developers are making several changes that should dramatically improve the
accuracy of the model for next year. For 2005, OWN will be changing the model
to incorporate either better forecasting tools, perhaps using a private weather
forecaster instead of, or in addition to, Environment Canada's data. An alternate
provider can produce site specific forecasts. For example, forecasts currently
are for London airport and Windsor airport. A farm at Dutton is halfway between
and the model would use estimated data based on London and Windsor. The new
provider will actually forecast for Dutton. And when weather data is unavailable,
the model will plug in a default rain or high relative humidity day based on
"We've decided to use climatological averages for unknowns so that we're
at least running the prediction model with a complete suite of projected weather
values based on normal weather patterns," says Nichols.
In the site specific model, growers can input actual weather data at their
farms, include wheat variety, tillage system and previous crops for more accurate
predictions and ultimately make better spraying decisions. "And they can
also disagree with the weather forecast and put in their own forecast,"
The weather up to harvest time also has an impact on DON accumulation. In many
areas of Ontario in 2004, harvest took three weeks to accomplish with five days
of rainy weather followed by a few dry days and back to rain. "If you knew
at heading that harvest was going to be wet, you'd definitely apply a fungicide,"
"Folicur is free fusarium insurance," says Peter Johnson,
cereal specialist with the Ontario Ministry of Agriculture and Food. "In
years when fusarium strikes hard, a Folicur application pays in spades. In years
when it doesn't, it's a break even with less risk of fusarium damage."
Johnson says Folicur's activity on yield-robbing leaf diseases produce an average
winter wheat yield increase of about eight percent – when fusarium is not
Up until three years ago, probably only five percent of Ontario's wheat crop
was sprayed with a fungicide. Since then, growers have begun treating wheat
like a valuable cash crop. In 2004, about 20 percent of the winter wheat crop
received a fungicide application. But it needs to be higher yet, especially
if growers continue to plant susceptible varieties like Maxine.
In 2004, 55 percent of Ontario's winter wheat crop was soft red, 25 percent
was soft white and 20 percent was hard red. Of the 725,000 acres harvested,
40 percent went feed, including 75 percent of the soft white wheat crop, 25
percent of the soft red crop and 40 percent of the hard red crop. The hard red
crop was hit hardest because of the popularity of fusarium susceptible Maxine.
The results were not as bad as in 1996 when 90 percent of the crop – all
classes – went feed. That is because in 1996, most of the wheat crop was
planted to highly susceptible soft white varieties.
The Bottom Line
The DONcast prediction model is a valuable tool to farmers
but needs to be more accurate. The approach we have taken the last few
years is to spray all our cereals with a fungicide. We have found this
to be an excellent insurance tool to protect the yield and quality of
our crops. Grahame Hardy, Inkerman, Ontario.
My experience with fusarium forecasting is that it is not an exact science.
It's almost bordering on being roulette!
Therefore, because of the fragile nature of wheat and the potential yield
and crop quality loss, spraying Folicur is just as important as spraying
for weed control. It is not a perfect method and I hope a better product
will be developed in the near future. Tony Pynenburg,