Plant disease prediction apps
Known for its diverse agricultural production, the Great Lakes region offers ideal conditions for corn and soybean production. Yet, these crops are not immune to the white mold, northern corn leaf blight, and tar spot diseases that threaten yields.
Damon Smith, Extension specialist at the University of Wisconsin-Madison, is a trained epidemiologist with a passion for delivering plant disease prediction models into the hands of farmers via smartphone apps.
“I want to make sure that we can deliver research-based information into a platform that is easily accessible by as many people as possible,” Smith says. “It seems natural to go the route of mobile apps because everyone is carrying a smartphone now.”
He and his team have developed three free apps based on disease forecasting models. Two are for white mold (Sporecaster and Sporebuster) and one is for tar spot (Tarspotter). More are in the works for northern corn leaf blight and gray leaf spot.
While developing Sporecaster, Smith and his team had to rethink the disease cycle and focus on weather parameters. “This particular disease is especially in tune to the weather. During the year, not only the incidence but also the severity of the disease is dictated almost exclusively by weather,” Smith explains.
With this alternate approach, they could predict the likelihood that mushrooms, which release spores that infect the soybeans with white mold, are present in a field.
Other white mold prediction systems exist, but they are based on the presence of the disease, meaning soybeans are already sick. Fungicides are not curative, Smith says, so if fields are only sprayed after the disease is present, you’ve missed your chance to fight the disease and wasted the trip and the dollars of an application.
Sporecaster on the farm
Mike Cerny is quick to admit that, while he’s been farming most of his life, he has been surprised by white mold more than once. “When you have an app like this, you can save money and become a better farmer by knowing which conditions may trigger white mold and which ones don’t,” says the Walworth County, Wisconsin, grower.
Since using the app, which he helped Smith ground-truth, Cerny has been saved twice from making unnecessary fungicide applications to his farm.
“Sporecaster has the potential to save $20 to $40 an acre,” he says. “I have all my fields entered into the app at the beginning of the year, and I check on them at least weekly and even twice a week depending on the weather.”
Since its release in 2018, Sporecaster has been downloaded 3,500 times. During peak season in July 2019, the app ran around 600 to 800 forecasts per day.
The companion app to Sporecaster is Sporebuster, which is a return-on-investment (ROI) calculator programmed with economic models.
Farmers using Sporebuster input pricing scenarios based on their operations, and the app forecasts the ROI or breakeven probabilities to help select the most profitable fungicide treatment.
Sporebuster was developed in partnership with agriculture economists and draws on data sets from 10 different fungicide programs that are common in the Great Lakes region.
What's on the horizon
Smith and his team of graduate students continue to make improvements to the accuracy of Sporecaster’s predictions, which is currently at about 80%. Due to the widespread use of the app in 2019 and users’ willingness to share feedback and data, they’ve identified the locations where the app wasn’t successful, and they’re evaluating why.
“We’re in the process of making adjustments to help improve performance,” Smith says.
In addition, the graduate students in the lab are working to understand how varietal resistance could be incorporated into the model. Host susceptibility can affect how much white mold will be in the fields, which Smith says can also affect the accuracy of the risk.
“If we know a certain variety is susceptible or resistant to white mold, we could account for that in the predictions and improve accuracy that way, as well,” Smith explains.
Sporecaster’s framework is already being used to develop other apps with similar functionality.
Tarspotter is one of those. This new app has been through a season of testing and will undergo another in 2020.
Cerny has been involved in Tarspotter’s development, too. “Tar spot scared the heck out of us when it first came out because there were so many unknowns to me as a farmer. I had one field in 2019 that went from almost clear of tar spot to covered within about 10 days,” he recalls. “At the time, tar spot was so new that we didn’t have anywhere to turn for information.”
Smith explains that many diseases, like tar spot and white mold, are similar in terms of their need to be controlled. Fungicides are most effective when used preventively.
“Tar spot on corn fits our framework really well, because we know it’s one of these diseases that is weather-driven. With this preventive app, you really have a crystal ball to see if you’re set up for that disease later in the season,” Smith says.
Farmers like Cerny can find not only financial value from plant disease prediction apps but also education and awareness.
“I look at an app as an educational tool just as much as a predictive tool,” Cerny says. “Once you feel comfortable with an app, you return to it quite often. Tarspotter and Sporecaster do both of those for me.”