Plant Disease Forecast Models

Click here to download a PDF version of this spotlight.

» Plant disease forecast models provide information to help growers make appropriate pest management decisions.

» Many plant disease forecast models use weather data to estimate the likelihood that disease outbreaks will occur.

» Technological advances are improving the accuracy and ease of use of plant disease forecast models.  

Similar to insect degree-day models, which indicate when certain insect life stages will develop, plant disease forecast models assess risks and predict diseases. Disease forecast models may also be referred to as disease predictors, warning systems, decision aids, decision support systems, predictive systems, and risk indexes.¹

Disease forecast models are most useful when they recommend whether or not treatments are needed to prevent disease outbreaks. Some growers use forecast models with the goal of reducing the number of treatments needed for adequate disease control, while others primarily use forecast systems to improve disease management outcomes.^1,2

Plant disease epidemics are often initiated by specific environmental conditions, with temperature and moisture usually having the most impact. Many disease forecast models use weather data and other information to determine the potential for disease development and severity.³ The data used as input variables in forecast models can include meteorological variables, remote sensing images, and biological factors such as plant growth stages and pathogen spore counts. Commonly used meteorological variables include minimum and maximum air temperatures, relative humidity levels, rainfall amounts and frequencies, hours of leaf wetness, wind speed, evaporation rates, and solar radiation (intensity and duration). Remote sensing imagery can be used to detect infected plants, even before symptoms become visible to the naked eye, as selective wavelengths of light can help distinguish healthy and diseased plant foliage.⁴

The input variables in models can come from in-field weather sensors, regional weather stations, grower surveys, and regional and national governmental services.⁴ The outputs from models can include indexes of disease risk, disease incidence and severity predictions, and inoculum arrival or development. The model output may then be used to develop management recommendations, often the application of fungicides and bactericides. Growers should evaluate model outputs and recommendations to make sure that they are appropriate for their particular location and production practices. Recommendations may also vary depending on the types of fungicides used (protectant, translaminar, and systemic products) and the presence of fungicide-resistant pathogen strains.⁵

Disease forecast models may be included in the software of stand-alone, in-field weather monitoring systems. The advantage of these systems is that they provide sitespecific information that is usually more accurate for that location than is data from regional weather stations. However, some disadvantages of these systems are that they can be expensive, easily damaged (lightning, field equipment accidents), and require periodic maintenance and sensor calibration.6 Several companies such as Spectrum Technologies, Inc. and Weather INnovations (WIN) provide stand-alone systems that run disease forecast models.

Companies and organizations can also maintain and manage forecast models using data collected from regional weather stations. These centralized model systems can also be interactive, allowing users to input data. In the past, data input and dissemination of results and recommendations was done over the phone. Now, most interactive systems are web-based, with users able to access information on their computer or smartphone. The Enviro-weather system managed by Michigan State University and the Cucurbit Downy Mildew Forecasting system managed through the ipmPIPE program are examples of centralized forecast model systems. ⁶

Forecast models do not have to be complicated or require specialized equipment or software to be useful. The Stewart’s wilt severity prediction model for sweet corn is an example of a simple forecast model that is very useful and effective. This model estimates the severity of Stewart’s wilt on sweet corn from the average temperatures in December, January, and February. The bacterium that causes Stewart’s wilt overwinters in and is vectored by the corn flea beetle. When average winter temperatures are below 27°F, few flea beetles will survive, and the severity of Stewart’s wilt will be low. When the average winter temperatures are above 33°F, flea beetles and the bacteria within them are more likely to overwinter successfully; therefore, the severity of Stewart’s wilt could be high on susceptible sweet corn hybrids.⁷

The forecast model BLIGHTCAST was developed to determine the potential severity of late blight on potatoes and tomatoes based on rainfall, relative humidity (RH), and temperature data. Specifically, 24-hour rainfall totals and the maximum and minimum temperatures when the RH is above 90% are used to calculate severity values. The accumulation of severity values over time is used to indicate the likelihood of infection by the late blight pathogen and provide recommendations on the need to apply fungicides.^7,8 BLIGHTCAST was originally provided by Penn State University as a phone-in service. Derivations of the BLIGHTCAST model are now available on several platforms, including as software on stand-alone weather monitoring systems and as regional forecast systems.

The Pacific Northwest late blight warning system also provides forecasts on the severity of late blight, but this model uses long-range weather (rain) forecasts to predict the likelihood of late blight outbreaks on a regional scale. Growers who use the system are notified when a first fungicide application is needed. Recommendations on further applications are based on weather forecasts, canopy development, and the probability of late blight outbreaks.¹

The FAST (Forecasting Alternaria solani on Tomatoes) model uses hourly measurements of leaf wetness and temperature to calculate severity values for early blight on tomato. These values are then used to create recommendations for fungicide applications. The FAST model has been adapted to forecast the severity of other tomato diseases including Septoria leaf spot and anthracnose in model called TOMCAST. The TOMCAST model has also been adapted to forecast diseases of potato, asparagus, carrot, celery, and pistachio.^1,10

The Cucurbit Downy Mildew (CDM) ipmPIPE forecasting system is different from the models described above. This system forecasts the spread of CDM spores over North America. Data for the model comes from a reporting network of growers, extension specialists, researchers, and agricultural professionals who monitor commercial cucurbit fields and sentinel plots for the presence of CDM. This data is combined with weather forecast data and models on the movement of particles in the atmosphere to predict when conditions will favor the spread of inoculum and infection of plants. The program is run by the North American Plant Disease Forecast Center at North Carolina State University.1,11 The CDM forecast system generates real-time maps showing locations that are at risk for infection. The risk levels can be adjusted based on specific cucurbit crops grown and by pathogen biotypes that may be resistant to fungicide applications. It is estimated that use of the system saves growers an average of two to three fungicide applications a year compared to using a calendarbased spray program. The maps can be viewed on the CDM ipmPIPE website, and alerts can be sent by text or email to the user, based on geographic area.^12,13 

Managing anthracnose on watermelon starts with using clean, disease-free seed and transplants to reduce the chances of introducing the pathogen into the field. The application of seed-treatment fungicides can help control seedborne inoculum.^1,2,4 Crop rotation to non-cucurbit hosts for at least one year (two to three years are better) can help keep inoculum levels in the field low. Watermelon crop residue should be destroyed (plowed in) promptly after harvest to speed the process of decomposition.

Avoid overhead irrigation or apply irrigation early in the day so that the foliage can dry before nightfall. Do not allow workers or equipment to enter fields when plants are wet from rain, irrigation, or dew. Carefully inspect fruit at harvest and discard any diseased fruit to minimize the spread of the pathogen and the development of fruit rot after harvest. A post-harvest wash with chlorinated water followed by refrigerated storage can also help slow the development of fruit rot after harvest.^1,2,4

As mentioned, most commercial watermelon varieties are resistant to races 1 and 2b of C. orbiculare. The resistance gene Ar-2-1 was found in 1959. So far, no commercial watermelon varieties have resistance that is effective against race 2.^1,7

In addition to cultural practices, watermelon anthracnose is managed with fungicides. Fungicides may not be cost-effective when cultural control practices are used.² Regular scouting for anthracnose should begin shortly after canopy closure. One recommendation is to examine ten leaves and five fruit in each of five random areas in the field. Fungicide applications should begin as soon as the disease is detected.⁵

Both protectant and systemic fungicides are registered for use on watermelon to manage anthracnose. Protectant fungicides that are effective against anthracnose include chlorothalonil, mancozeb, and thiophanate-methyl. Some newer fungicides, including tebuconazole and a combination product containing difenconazole and cyprodinil, are being used to replace the older fungicides. However, these newer products may not provide the same level of protection.^1,8,9

The QoI (strobilurin) class of fungicides (FRAC group 11) have been particularly effective for managing anthracnose on cucurbits, including watermelon. This group includes the fungicides azoxystrobin, fluoxastrobin, pyraclostrobin, and trifloxystrobin. The fungicides used most often to manage gummy stem blight include the DMI fungicides (FRAC group 3), such as tebuconazole and difenoconazole; the carboxamides (FRAC group 7), such as fluopyram; the anilino pyrimidines (FRAC Group 9), such as cyprodinil, and the phenyl pyrroles (FRAC Group 12), such as fludioxonil. Unfortunately, these fungicides are not as effective against anthracnose.^3,8 Some pathogens, such as the fungus that causes gummy stem blight, have developed resistance to the QoI class of fungicides. These fungicides are no longer effective for controlling these diseases, which has led some growers to stop using the QoI fungicides and switch to the fungicides mentioned above. It is thought that this change in the types of fungicides use has resulted in an increase in the severity of anthracnose on watermelon over the past few years.^5,8 However, growers should not rely exclusively on the QoI fungicides for managing anthracnose, as it is possible that the anthracnose fungus could also develop resistance to this group of fungicides. The fungicides trifloxystrobin, azoxystrobin, pyraclostrobin, thiophanate-methyl, chlorothalonil, and mancozeb continue to provide consistent control of anthracnose on watermelon. Alternating or combining applications of the QoI fungicides with the protectant fungicides will help prevent or delay resistance development.³

A disease forecast model, MelCast, was developed to help growers determine when fungicides need to be applied to manage anthracnose on watermelon.^10,11 The MelCast model uses temperature and leaf wetness data to estimate when conditions are favorable for disease spread and infection.  

1 Gent, D., Mahaffee, W., McRoberts, N., and Pfender, W. 2013. The use and role of predictive systems in disease management. Annu. Rev. Phytopathol. 51:267–289.
2 Fabre, F., Plantegenest, M., and Yuen, J. 2007. Financial benefit of using crop protection decision rules over systematic spraying strategies. Phytopathology 97:1484-1490.
3 Shah, D., Paul, P., De Wolf, E., Madden, LV. 2019. Predicting plant disease epidemics from functionally represented weather series. Phil. Trans. R. Soc. B 374:20180273.
4 Fenu, G. and Malloci, F. 2021. Forecasting plant and crop disease: an explorative study on current algorithms. Big Data Cogn. Comput. 2021, 5, 2.
5 University of California and Resources. California PestCast: Disease Model Database. http://ipm.ucanr.edu/DISEASE/DATABASE/diseasemodeldatabase.htm.
6 van Bruggen, A. 2015. Epidemiology of plant diseases: Decision support systems. https://plantpath.ifas.ufl.edu/classes/plp6404/lectures/powerpoints/Epidemiology%20 2015%20lecture29%20Decision%20support%20systems.pdf.
7 Pataky, J. 2004. Stewart’s wilt of corn. The Plant Health Instructor. DOI:10.1094/ PHI-I-2004-0113-01.
8 Krause, R., Massie, L., and Hyre, R. 1975. BLIGHTCAST: A computerized forecast of potato late blight. Plant Dis. Rep. 59:95-98.
9 MacKenzie, D. 1981. Scheduling fungicide applications for potato late blight with BLIGHTCAST. Plant Disease 65:394-399.
10 Pitblado, R. 1988. Development of a weather-timed fungicide spray program for field tomatoes. Can. J. Plant Pathol. 10:371.
11 Main, C., Keever, T., Holmes, G., Davis, J. 2001. Forecasting long-range transport of downy mildew spores and plant disease epidemics. APSnet Features. https://www.apsnet.org/edcenter/apsnetfeatures/Pages/DownyMildew.aspx.
12 Cucurbit Downy Mildew Forecasting – IPM PIPE. https://cdm.ipmpipe.org/.
13 Ojiambo, P., Holmes, G., Britton, W., Keever, T., Adams, M., et al. 2011. Cucurbit downy mildew (CDM) ipmPIPE: A next generation web-based interactive tool for disease management and extension outreach. Plant Health Progress. doi:10.1094/PHP-2011-0411- 01-RV.
Websites verified 11/11/2021

For additional agronomic information, please contact your local seed representative.

Performance may vary from location to location and from year to year, as local growing, soil and weather conditions may vary. Growers should evaluate data from multiple locations and years whenever possible and should consider the impacts of these conditions on the grower’s fields. The recommendations in this article are based upon information obtained from the cited sources and should be used as a quick reference for information about vegetable production. The content of this article should not be substituted for the professional opinion of a producer, grower, agronomist, pathologist and similar professional dealing with vegetable crops.

BAYER GROUP DOES NOT WARRANT THE ACCURACY OF ANY INFORMATION OR TECHNICAL ADVICE PROVIDED HEREIN AND DISCLAIMS ALL LIABILITY FOR ANY CLAIM INVOLVING SUCH INFORMATION OR ADVICE.

9050_SE_S14 Published 12/03/2021

Bayer, Bayer Cross Design, and Seminis® are registered trademarks of Bayer Group. All other trademarks are property of their respective owners. © 2021 Bayer Group. All rights reserved.