UC IPM Online UC ANR home page UC IPM home page

UC IPM Home

SKIP navigation

 

Research and IPM

Models: Diseases

Potato late blight

Crop: Potato

Disease: Late Blight
Pathogen: Phytophthora infestans

Note: Before using a model that was not field tested or validated for a specific location, the model should be tested for one or more seasons under local conditions to verify that it will work in the desired location. See "Validation Work" below.

Model 1 of 16 (See also late blight on tomato.)

Model developer and citation

Hansen J. G., Andersson, B. and Hermansen, A. 1995. NEGFRY- A system for scheduling chemical control of late blight in potatoes. In: Phytophthora infestans 150: European Association for Potato Research (EAPR)--Pathology Section Conference, held in Trinity College, Dublin, Ireland, September 1995 to mark the one hundred and fiftieth anniversary of the first record of potato blight in Ireland and the subsequent famine. L. J. Dowley, et al. (Eds). Boole Press, Ltd. Dublin. pp. 201-208.

Sensor location

Meteorological stations.

Input variables

Environmental: Hourly temperature, relative humidity, and rainfall.

Model description

NEGFRY is a personal computer-based model developed in Denmark. It uses the "negative prognosis" model of Ullrich and Schrodter (1966) to forecast risk of late blight outbreak on potato, and the model of Fry et al. (1983) to time subsequent fungicide applications during the season. Parameterization of the model was based on biological and meteorological data obtained from Foulom, DK. First fungicide application is recommended once accumulated risk values exceed 160 and the daily risk value, calculated according to the "negative prognosis" model, is above 8. After the initial spray, favorable weather for disease development is expressed as blight units according to the method of Fry et al. (1983). All parameters in the model are saved in a setup file and can be changed by the user.

Action threshold

Fungicide sprays are started when the accumulated risk value is greater than 160, and the risk value for the previous night is greater than 8. Subsequent treatments should be done when accumulated rainfall is greater than 20 mm or risk value of the previous night is 8, and a blight unit threshold of 40 (susceptible cultivar), 45 (moderately susceptible cultivar) or 50 (moderately resistant cultivar) has been exceeded. Blight units and rainfall value is set to zero after each fungicide treatment.

Model validation

This model has been validated in Denmark, Norway, and Sweden in 1993 and 1994 for late blight of potato. It is being validated by Adcon Telemetry and the California Tomato Research Institute in northern California.

Model implementation

This model has been implemented in Europe.

Top of page

Model 2 of 16

Model developer and citation

Ullrich J. and Schrodter H., 1966. Das problem der vorhersage des aufretens der kartoffelkrautfaule ((Phytophthorainfestans) und die moglichkeit seiner losung durch eine negativprognose. Nachrichtenblatt Dt. Pflanzenschutzdienst (Braunschweig) 18:33-40.

Sensor location

The model was developed with temperature sensors mounted 2 meters above the ground.

Input variables

Environmental: Hourly temperature, relative humidity, and rainfall.

Calculated: Number of hours that fit pre-defined environmental conditions, risk values, and accumulated risk values.

Model description

This "negative prognosis" model uses measurements of temperature, relative humidity, and rainfall to predict when late blight (Phytophthora infestans) epidemics are not likely to occur. It has been used in Germany to predict the timing of the first treatment. Daily and accumulated risk values over a week are calculated starting at crop emergence. Risk values are calculated according to the following table:

Table 1. Late blight risk values
Multiplication factor (r) Number of hours hourly temperature averages are in this range (h), or other conditions to be met RH or Precipitation Requirements, or other conditions to be met
0.899 10.0 - 11.9 Only count hours that co-occur with 4 or more consecutive hours at RH>=90% or rain>=0.1 mm/hr
0.4118 14.0 - 15.9
0.5336 16.0 - 17.9
0.8816 18.0 - 19.9
1.0498 20.0 - 21.9
0.5858 22.0 - 23.9
0.3924 10.0 - 11.9 Only count hours that co-occur with 10 or more consecutive hours at RH>=90% or rain>=0.1 mm/hr
0.0702 14.0 - 15.9
0.1278 16.0 - 17.9
0.9108 18.0 - 19.9
1.4706 20.0 - 21.9
0.855 22.0 - 23.9
0.1639 15.0 - 19.9 Do not consider RH or rain, add 7.5479 to the product of r x h
0.0468 Number of hours with average RH < 70% Subtract 7.8624 from the product of r x h

To calculate risk values, count the number of hours (h) that the average temperature falls within the temperature range in the second column and the conditions of relative humidity or rainfall indicated in the third column. Since there are 14 rows in the table, there should be 14 values of h, some of which may be zero. Then, multiply each of these 14 h-values by the corresponding r-value of the first column and sum up the 14 products (r x h). The sum of the products is the risk value. For example, count all hours within a week during which hourly average temperature was between 10.0-11.9 C and relative humidity was greater than or equal to 90% or precipitation was 0.1 mm/hour or greater. Then multiply that number of hours by the corresponding value of the first column (r=0.8990). Do the same procedure for the remaining 13 rows of the table and then add up all products to get the risk value.

Action threshold

Disease is expected when accumulated risk value has exceeded the threshold of 150. Users have adjusted this threshold to fit local conditions.

Model validation

This model has been validated in Germany.

Model implementation

This model has been implemented in Europe.

Current limitations of model

The model assumes the same amount of initial inoculum is present every year, which may not be justified all times.

Top of page

Model 3 of 16

Model developer and citation

Fry, W. E., Apple, A. E., and Bruhn, J. A. 1983. Evaluation of potato late blight forecasts modified to incorporate host resistance and fungicide weathering. Phytopathology 73:1054-1059.

Bruhn, J. A. and Fry, W. E. 1981. Analysis of potato late blight epidemiology by simulation modelling. Phytopathology 71: 597-601.

Sensor location

Within crop canopy.

Input variables

Environmental: Daily rainfall (mm), hourly average temperature and relative humidity.

Calculated: Number of consecutive hours with relative humidity greater than or equal to 90%, at five intervals of temperature (< 3, 3-7, 8-12, 13-22, 23-27 and >27 C) in 24 hour periods (from noon to noon the following day).

Host: Cultivar resistance (susceptible, moderately susceptible or moderately resistant).

Fungicide: Time (days) since last fungicide application (Note: this model was developed for chlorothalonil).

Model description

This forecast model was derived from two simulation models. One model describes weather effects on fungicide distribution and amount. The second model describes effects of host resistance and weather on development of Phytophthora infestans on potato. Blight units are calculated according to the number of consecutive hours that relative humidity is greater than or equal to 90%, and average temperature falls within any of six ranges (< 3, 3-7, 8-12, 13-22, 23-27 and >27 C). Fungicide units are calculated based on daily rainfall (mm) and time since last fungicide application (note: fungicide units were developed for chlorothalonil). Decision rules about when fungicide should be applied were generated based on cumulative blight units or fungicide units since last spray. Use Tables 1-3 to calculate blight and fungicide units and apply decision rules.

Table 1. Blight units for the simulation forecast as determined by temperature and periods of high relative humidity /a.
Average temperature Cultivar resistance Consecutive hours of relative humidity >=90% that should result in blight units of:
(C)   0 1 2 3 4 5 6 7
>27 S /b 24              
MS /c 24              
MR /d 24              
23-27 S 6 7-9 10-12 13-15 16-18 19-24    
MS 9 10-18 19-24          
MR 15 16-24            
13-22 S 6         7-9 10-12 13-24
MS 6 7 8 9 10 11-12 13-24  
MR 6 7 8 9 10-12 13-24    
8-12 S 6 7 8-9 10 11-12 13-15 16-24  
MS 6 7-9 10-12 13-15 16-18 19-24    
MR 9 10-12 13-15 16-24        
3-7 S 9 10-12 13-15 16-18 19-24      
MS 12 13-24            
MR 18 19-24            
<3 S 24              
MS 24              
MR 24              
/a High relative humidity >=90%. Blight unit estimation period is 24 hr (noon to noon).
/b S = susceptible cultivars.
/c MS = moderately susceptible cultivars.
/d MR = moderately resistant cultivars
Table 2. Fungicide units (for chlorothalonil) for the potato late blight simulation forecast as determined by rainfall and the number of days since the last fungicide application.
Time (days) since fungicide application Daily rainfall amounts (mm) that result in fungicide units of
1 2 3 4 5 6 7
1 <1     1 2-3 4-6 >6
2 <1   1 2-4 5-8 >8  
3 <1   1-2 3-5 >5    
4-5 <1   1-3 3-8 >8    
6-9 <1   1-4 >4      
10-14 <1 1 2-8 >8      
>14 <1 1-8 >8        

Action threshold

Use Table 3 to determine when fungicide should be sprayed.

Table 3. Decision rules for the simulation forecast.
Logic statements Cultivar resistance
Susceptible Moderately susceptible Moderately resistant
Fungicide should be applied if fungicide has not been applied within 5 days      
AND cumulative blight units since last spray exceed: 30 35 40
OR cumulative fungicide units since last spray exceed: 15 20 25

Model validation

This model was evaluated during 1980 and 1981 using the susceptible cultivar Hudson and the moderately resistant cultivar Rosa. It also has been validated as a part of NEGFRY (Hanson, 1993) in Europe, and is currently being evaluated by Adcon Telemetry and the California Tomato Research Institute in northern California.

Model implementation

Not known.

Current limitations

This model was developed to be used with chlorothalonil. It was not developed to predict timing of the first fungicide application.

Future directions

This model has been modified to time metalaxil applications and to include weather forecasts.

Related work

Doster, M. A., and Fry, W. E. 1991. Evaluation by computer simulation of strategies to time metalaxyl applications for improved control of potato late blight. Crop Protection 10: 209-214.

Raposo, R., Wilks, D. S., and Fry, W. E. 1993. Evaluation of potato late blight forecasts modified to include weather forecasts: A simulation analysis. Phytopathology 83: 103-108.

Fohner, G. R., Fry, W. E., and White, G. B. 1984. Computer simulation raises question about timing protectant fungicide application frequency according to a potato late blight forecast. Phytopathology 74: 1145-1147.

Top of page

Model 4 of 16

Model developer and citation

Winstel, K. 1993. Kraut- und knollenfaule der kartoffel eine eeue prognosemoglichkeit-sowie bekampfungsstrategien. Med. Fac. Landbouww. Univ. Gent, 58/3b.

Sensor location

Two meters above ground.

Input variables

Environmental: Temperature, relative humidity.

Calculated: Daily average, minimum and maximum temperatures, hours of temperatures greater than 10 C and relative humidity greater than 90%.

Model description

This model is composed of two phases. Phase 1 predicts infection, which is predicted after the following requirements are met: After the daily average temperature is between 10 and 23 C and then 10 hours or more of temperatures greater than 10 C and relative humidity greater than 90% occur (such periods are considered to be the same as leaf wetness). Phase 2 sets criteria for pathogen growth. Phase 2 occurs when the maximum daily temperature for two consecutive days is between 23 C and 30 C. Phase 2 must occur at least 24 hours but not later than 10 days after phase 1.

Action threshold

According to the model, initiate treatment when phase 1 occurs and is followed by phase 2. Subsequent treatments should be based on calendar and material.

Current limitations

This model was developed for early potato varieties.

Future directions

To modify the model for later varieties.

Model validation

This model is being validated by the California Tomato Research Institute. A modification of the model is being validated by Adcon Telemetry.

Model implementation

Unknown.

Top of page

Model 5 of 16

Model developer and citation

Johnson, D. A., Alldredge, J. R. and Vakoch, D. L. 1996. Potato late blight forecasting models for the semiarid environment of south-central Washington. Phytopathology 86:480-484.

Sensor location

Air temperature is measured at a height of 1.5 m above soil surface.

Input variables

Environmental: Daily total rainfall, daily minimum temperature.

Calculated: Number of days with rain >= 0.25 mm during April and May, number of days with rain > = 0.25 during July and August, total precipitation during May when daily minimum temperature was greater or equal to 5 C.

Pathogen: Occurrence of a late blight outbreak during the preceding year ( 0 = no; 1 = yes)

Model description

Two discriminant functions and two logistic regression models were developed to forecast outbreaks of late blight (Phytophthora infestans) on potato (mainly Russet Burbank) in the south-central area of Washington State, a semiarid environment where late blight occurs sporadically.

The first discriminant function uses the following variables: late blight outbreak during the preceding year (Yp = 0 = no; Yp =1 = yes), number of days with rain >= 0.25 mm during April and May (Ram) and total precipitation during May when daily minimum temperature was greater or equal to 5 C (Pm). The second discriminant function uses Yp, Ram and number of days with rain > = 0.25 during July and August (Rja). The logistic regression models use the same variables as each of the discriminant functions. They were calculated as an alternative to the discriminant functions to overcome the assumption that the variables have a multivariate normal distribution.

The first discriminant function provided a lower percentage of correct classifications (88%) of outbreak and nonoutbreak years than the second discriminant function (92%), but the second function cannot be used until the end of August. The first function can be used starting the first of June. If no late blight has been observed by August 31, then the second function is used. According to 25 years of disease records, late blight outbreaks in the south-central area of Washington occur after June 14.

First discriminant function is:

Nonoutbreak: -4.426 + 2.052(Yp, 1=yes, 0=no) + 0.863(Ram) + 0.052(Pm)

Outbreak: -11.886 + 6.191(Yp) + 1.462(Ram) - 0.033(Pm).

Second discriminant function is:

Nonoutbreak: -5.636 + 1.774(Yp, 1=yes, 0=no) + 0.974(Ram) + 0.5(Rja)

Outbreak: -14.546 + 5.776(Yp, 1=yes, 0=no) + 1.506(Ram) + 0.711(Rja)

To estimate the risk of late blight outbreak during a given year, values of each variable are multiplied by their coefficients and added to the constant. If the equation for outbreak gives a higher score than the equation for nonoutbreak, then that year is considered a late blight year (outbreak), otherwise no disease is expected (nonoutbreak).

Logistic regression models are as follows:

Model 1:

L = 7.548 - 3.553(Yp, 1=yes, 0=no) - 0.629(Ram) + 0.09(Pm)

Model 2:

L = 11.470-3.88(Yp, 1=yes, 0=no)- 0.716(Ram) - 0.259(Rja)

After calculation of the above L values, the probability of an outbreak is estimated as follows:

P = 1/[1 + exp(L)]

If P >= 0.5 then the year is classified as a late blight (outbreak) year; otherwise the year is classified as nonoutbreak.

If criteria for the occurrence of a late blight outbreak according to the first discriminant function or the first logistic regression model are met before June 1, a forecast for the potential occurrence of an outbreak could be made earlier.

The second discriminant function or the second logistic regression model could be used through July and August 31 by solving for the value Rja needed for an outbreak to occur and comparing it to the normal and expected occurrences of rainy days during July and August based on weather forecasts.

Action threshold

When a late blight outbreak is forecast, monitor the fields closely and treat areas with a history of late blight.

Model validation

Unknown.

Model implementation

Unknown.

Current limitations of model

This model has not been tested under California conditions.

Top of page

Model 6 of 16

Model developer and citation

Forsund, E. 1983. Late blight forecasting in Norway 1957-1980. EPPO Bull. 13(2)255-258.

Sensor location

Off-site, at regional weather stations.

Input variables

Environmental: Daily maximum and minimum air temperature (C), relative humidity at noon, and daily rainfall (mm).

Model description

This forecasting model was developed in Norway and has been used since 1957. The model contains four criteria used to assess the risk of potato late blight disease caused by the fungus Phytophthora infestans. The following conditions are considered favorable for the disease:

Daily maximum temperature is 17-24 C

Minimum temperature >= 10 C

Relative humidity at noon is >= 75 %

Daily rainfall >= 0.1 mm

Warnings are issued daily when these conditions occur for two consecutive days. The risk of disease outbreak increases with the number of favorable days. Warnings are issued by the Norwegian Institute of Meteorology and distributed through press and broadcasting systems.

Action threshold

According to disease records from 1957 to 1980, in Norwegian locations where more than three warnings have been issued, moderate to severe potato late blight has occurred. Lower disease severity than expected can occur in locations where low or no inoculum is present due to low disease pressure in previous years.

Model validation

This model has been validated in Norway.

Model implementation

This model has been implemented in Norway.

Current limitations

This model does not differentiate between initial infection conditions and subsequent infection events; it provides a warning of the risk of infection anytime during the season.

Top of page

Model 7 of 16

Model developer and citation

Forrer, H.R., Gujer, H.U. and Fried, P.M. 1993. PhytoPRE - A comprehensive information and decision support system for late blight in potatoes, from the Workshop on Computer-based Decision Support System (DSS) in Crop Protection, Parma, Italy, November 23-26 1993.

Sensor location

Off-site, at regional weather stations, although rainfall data is collected in potato fields.

Input variables

Environmental: Hourly temperature, relative humidity and rainfall.

Calculated: Risk values (RV) according to Ullrich and Schroedter, (1966), and infection potential (IP).

Pathogen: Date and location of First Late Blight record (FLB).

Model description

PhytoPRE is a computer-based information and decision support system (DSS) for potato late blight (Phytophthora infestans) in Switzerland. PhytoPRE consists of an epidemiological forecast model, a set of decision rules, and an information system.

The epidemiological forecast model was developed from national and cantonal records (1990-1992) of disease progress curves over time that were collected from a network of about 150 unsprayed (organic) potato plots. Since the slope and the shape of the curves were the same in almost all cases, and only the start date of the epidemics varied, a logistic model was used to forecast the probability of infection (IP). The logistic model variable for initial disease was assumed to be 0.03% and the apparent infection rate was set at 0.2. Forecasts are provided after the first national late blight (FLB) record is reported. To adapt the forecast to specific regions, Ullrich and Schroedter's Risk values (RV) for that region are compared to the RV in the location with the FLB. The lower the RV in a certain region, the more the start of the epidemic is delayed.

The infection probability (IP) is calculated as follows:

e{r*t + ln[y/(1-y)]}

IP = _____________________

1 + e{r*t + ln[y/(1-y)]}

where

r = 0.2

y = 0.03

t = (dF - dOR) + L

dOR = Julian day of the FLB

dF = Day to forecast

L = Latent period (10 d)

For regions with lower RV than the RV of the region with the FLB, dOR is calculated as follows:

dOR = dO + [(RVFLB / RVREG) - 1] * K

where

dO = Julian day of the FLB

RVFLB = RV of the region with the FLB on dO

RVREG = RV on dO for the region to forecast

K = constant = 14 d

If FLB occurs after the RV of the region to forecast equals 90, RVFLB / RVREG is calculated for the day when RVFLB equals 90.

The timing of the first and subsequent fungicide treatments depends on host susceptibility, IP values, amount of rainfall, fungicide used and the amount of time since the last fungicide application. Basic decision rules are indicated in Table 1.

PhytoPRE is comprised of a database of participants and their meteorological and disease records, a catalogue of epidemiological parameters, cultivars, fungicides, meteorological stations, recommendations, and the coordinates of all Swiss communities (about 3000), and the decision rules.

PhytoPRE involves three types of farmer participants. Type A participants follow an interactive procedure to receive spray advisories. These participants return a weekly report with their personal fungicide application data, rainfall and disease and crop observations. Type B participants receive a letter with a list of the 10 nearest records of late blight, a simulation of the IP curve, disease risk based on the participant's field data (such as cultivar), and meteorological data from the nearest station. Type C participants receive a weekly bulletin with a table of all late blight records and rainfall data of all meteorological stations (about 30).

PhytoPRE produces three kinds of outputs; field specific letters with application recommendations, lists with lateblight records and graphs with IP, rainfall and fungicide applications, weekly bulletins with lateblight records, rainfall, RVs for the whole country, and personal comments from the PhytoPRE manager, and a lateblight risk map with IPs for the different regions of Switzerland and all actual lateblight records.

Action threshold

Timing of fungicide sprays is determined according to the following decision rules:

Table 1. Timing of fungicide sprays.
Potato Susceptibility First Treatment Subsequent Treatments
High IPa > 1% TIME> 6-9 daysb since last application
and
RAINc > 29 mm since last application or TIME > 14 days
Medium IP > 1%

TIME > 7-11 daysb since last application
and
RAINc > 29 mm since last application

Low IP > 25% and RAINd last 10 days > 29 mm and GS > 69 TIME > 14 days since last application
and
RAINd > 29 mm since last application
Where IP = infection probability; TIME = time since last application; GS = Growth Stage
a Except for fields near (>1 km) early potatoes
b Exact number of days depends on fungicides used
c Amount of rain since last application
d Amount of rain within last 10 days

Model validation

This model has been validated in Switzerland.

Model implementation

This model has been implemented in Switzerland.

Current limitations

This model was developed based on observed disease progress curves in Switzerland and would need to be modified for use in California, but a similar approach could be taken. This would involve a network of growers communicating about the first occurrence of late blight, on-site monitoring of environmental conditions, and the development and validation of California-based decision rules.

Top of page

Model 8 of 16

Model developer and citation

Gutsche, V. 1993. PROGEB - a model-aided forecasting service for pest management in cereals and potatoes. EPPO Bull. 23:577-581.

Sensor location

Unspecified.

Input variables

Environmental: Hourly temperature, relative humidity and rainfall.

Host: Cultivar

Model description

PROGEB is an integrated group of forecasting models for the main pests of potatoes and cereals in Germany. One of their components, PHYTEB, forecasts Phytophthora infestans. PHYTEB simulates symptomatic stages of the host (noninfected, latent, preinfection, number of infections, amount of dead tissue) by means of state variables. State variables are calculated every 3 hours. Temperature, relative humidity, rainfall, cultivar, and plant protection actions taken are the independent variables of the functions or determine their parameters.

PHYTEB consists of two submodels, SIMPHYT 1 and 2. The first submodel forecasts the beginning of the epidemics 7 -10 days ahead. It takes into account the cultivars, date of emergence and agrometeorological phenomena. The second submodel simulates the course of the epidemic for two cultivar classes and different fungicide application practices, including no fungicide sprays. Two special features of this submodel are a detailed mathematical representation of fungicide action, and a function to calculate how long fungicide applications can be delayed without any risk.

Action threshold

Fungicide sprays recommended according to the output of submodel 2, which calculates how long fungicide sprays can be delayed without any risk.

Model validation

This model has been validated in Germany.

Model implementation

This model has been implemented in Germany.

Top of page

Model 9 of 16

Model developer and citation

IPM University of Wisconsin. 1995. Integrated Pest Management WISDOM. Professional software for Agricultural Systems. 101 pp.

Stevenson, W. R. 1993. IPM for potatoes: a multifaceted approach to disease management and information delivery. Plant Disease. 77:309-311.

Sensor location

In-field weather stations, sensors within the canopy.

Input variables

Environmental: Daily rainfall, maximum and minimum temperature, hourly temperature and relative humidity.

Calculated: Number of hours of relative humidity 90% or greater. Minimum and maximum temperature during periods of relative humidity 90% or greater.

Host: Presence of blight symptoms, cultivar.

Model description

The potato disease management (PDM) program is one of the five components of the University of Wisconsin Potato Crop Management (PCM) system. The latest version of this software is also called WISDOM. PDM was initially released alone in 1983. In 1989, PDM was released as an integral component of the PCM.

The first PDM component forecasts early blight (Alternaria solani) of potatoes based on the work of Madden et al., (1978) and Pscheidt (1983, 1985). The second PDM component forecasts late blight (Phytophthora infestans) of potatoes based on the BLITECAST program.

Action threshold

To control late blight, PDM recommends initiation of the first fungicide spray when 18 severity values have accumulated. Subsequent treatments are recommended every 5 days when conditions favor late blight, and every 10 days when conditions are not favorable for late blight.

Model validation

This model has been validated in Wisconsin by the University of Wisconsin, in Ohio by Campbell Soup Company, and in Mexico by Campbell Soup Company and Agridiagnostics.

Model implementation

This model has been implemented in Wisconsin, Minnesota and Illinois, North Dakota and Mexico. By 1987, it was being used on at least 27,645 acres in Wisconsin, Minnesota and Illinois. It is currently being evaluated under California conditions.

Top of page

Model 10 of 16

Model developer and citation

Krause, R. A., Massie, L. B., and Hyre, R. A. 1975. BLITECAST, a computerized forecast of potato late blight. Plant Disease Reporter 59: 95-98.

Sensor location

In-field weather stations, sensors within the canopy.

Input variables

Environmental: Daily rainfall, maximum and minimum temperature

Calculated: Number of hours of relative humidity 90% or greater, maximum and minimum temperature during periods of relative humidity 90% or greater.

Model description

BLITECAST is an integrated computerized version of both the Hyre and the Wallin model. The first part of the program forecasts the initial occurrence of late blight 7-14 days after the first accumulation of 10 rain-favorable days according to Hyre's criteria, or the accumulation of 18 severity values according to Wallin's model. The second part of the program recommends fungicide sprays based on the number of rain-favorable days and severity values accumulated during the previous seven days. Accumulation of rain-favorable days and severity values begins when distinct green rows can be seen in the potato field, and ends at vine kill. The first spray is recommended when the first late blight forecast is given. Subsequent sprays are recommended according to an adjustable matrix which correlates rain-favorable days with severity values.

Action threshold

First spray is recommended when the first forecast is given. Subsequent treatments are based on the following table:

Adjustable matrix used to relate severity values and rain-favorable days and generate spray recommendation for Blitecast.
Total number of rain-favorable days during the last 7 days Severity values during the last seven days
  <3 3 4 5 6 >6
  Message number
<5 -1 -1 0 1 1 2
>4 -1 0 1 2 2 2
Meaning of message numbers is:
-1 = No spray is recommended.
0 = A late blight warning (treat or review conditions in 2 to 3 days. If the short range forecast is for blight favorable weather, follow a 7-day spray schedule)
1 = A 7-day spray schedule is recommended.
2 = A 5 day spray schedule is recommended.
Note: Some researchers have advised reverting to a regular spray schedule at 1% disease severity.

Model validation

This model was validated in the eastern United States in 1972, and in other parts of the world subsequently.

Model implementation

This model has been implemented extensively. This model was modified and incorporated into a computerized crop management program, WISDOM, in 1983.

Related work

MacKenzie, D. R. 1981. Scheduling fungicide applications for potato late blight. Plant Disease 65: 394-399.

MacKenzie, D. R. 1984. Blitecast in retrospect a look at what we learned. FAO Plant Protection Bulletin 32:45-49.

Model 11 of 16

Model developer and citation

Hyre, R. A. 1954. Progress in forecasting late blight of potato and tomato. Plant Disease Reporter. 38: 245-253.

Sensor location

Off-site, at regional weather stations.

Input variables

Environmental: Daily rainfall, maximum and minimum temperature.

Model description

This model is based on daily rainfall and maximum and minimum temperatures. It forecasts the outbreak of potato late blight 7-14 days after the occurrence of ten consecutive blight favorable days. Days are considered blight favorable when the 5-day average temperature is below 25.5 C and the total rainfall for the last 10 days is 3 cm or greater. Days with minimum temperatures below 7.2 C are considered unfavorable.

Model validation

This model has been extensively evaluated.

Model implementation

This model has been implemented by growers in the northeastern United States.

Top of page

Model 12 of 16

Model developer and citation

Wallin, J. R. 1951. Forecasting tomato and potato late blight in the northcentral region (Abstr) Phytopathology 41: 37.

Wallin, J. R. 1962. Summary of recent progress in predicting the late blight epidemics in United States and Canada. American Potato Journal 39: 306-312.

Wallin, J. R., and P. E. Waggoner. 1950. The influence of climate on the development and spread of Phytopthora infestans in artificially inoculated potato plots. Plant Disease Reporter Suppl. 190: 19-33.

Sensor locations

Off-site, at regional weather stations.

Input variables

Environmental: Hourly relative humidity and temperature.

Calculated: Minimum and maximum temperature during periods of relative humidity >= 90%. Number of hours of relative humidity >= 90%.

Model description

This model predicts the first occurrence of potato late blight and its subsequent spread based on season accumulation of severity values. These severity values are based on various combinations of the hours with a relative humidity of 90% or greater and the average temperature during those periods. Accumulation of severity values starts at plant emergence.

The following table indicates how severity values are assigned:

Relationship of temperature and relative humidity (RH) periods as used in the Wallin late blight forecasting system.
Average Temperature Range Severity values Hours of 90% RH or greater
  0 1 2 3 4
7.2 - 11.6 C 15 16-18 19-21 22-24 25+
11.7 - 15.0 C 12 13-15 16-18 19-21 22+
15.1 - 26.6 C 9 10-12 13-15 16-18 19+

Action threshold

The first occurrence of blight is predicted 7-14 days after 18-20 severity values have accumulated.

Model validation

This model has been evaluated extensively world-wide.

Model implementation

This model has been implemented by growers in the northeast US.

Top of page

Model 13 of 16

Model developer and citation

Bourke, P.M. 1953. Potato blight and the weather: A fresh approach. Technical Note No. 12. Irish Meteorological Service.

Keane, T. Potato blight warning practice in Ireland. In: Phytophthora infestans 150: European Association for Potato Research (EAPR)-Pathology Section Conference, held in Trinity College, Dublin, Ireland, September 1995 to mark the one hundred and fiftieth anniversary of the first record of potato blight in Ireland and the subsequent famine. L. J. Dowley, et al. (Eds). Boole Press, Ltd. Dublin. pp. 191-200.

Sensor location

Off-site, at regional weather stations.

Input variables

Enivronmental: Hourly temperature, relative humidity and rainfall.

Calculated: Number of hours with temperature not below 10 C (50 F) and relative humidity of at least 90%.

Effective Blight Hours (EBH), the effective period which follows the initial wetness period, EBH start to accumulate after 12 hours with rain and conducive temperatures and relative humidity, or after 16 hours without rain but with conducive temperatures and relative humidity.

Model description

This model was developed by Austin Bourke and is known as the 'Irish Rules'. It is currently used in Ireland. For Irish conditions, sporangia of Phytophthora infestans may germinate and infect new potato leaves with a wetness period of at least 12 hours and air temperature not below 10 C and relative humidity of at least 90% if precipitation occurs in the interval from 7 to 15 hours after the start of the period. The minimum wetness period required is increased to 16 hours if no precipitation takes place during the wetness interval.

To determine the importance of blight-weather spells, the number of effective blight hours (EBH) is calculated starting as 1 EBH at the 12th hour (or at the 16th hour if no precipitation occurs). EBH are accumulated as long as the criteria for temperature and relative humidity are sustained. If two consecutive favorable periods are separated by 5 hours or less, no new lead period is required for the second spell and both periods are coalesced for the calculation of the duration of the effective period.

The current Government run Irish system starts issuing warnings after mid-June based on the prediction of upcoming favorable weather for late blight spread. Prior to mid-June, the Irish rules assume a "zero date" and blight is rarely seen before this date and no warnings are issued. Synoptic weather maps and computer weather models are used to predict up to a week in advance the time of arrival of these conditions. Suitability of the weather conditions for fungicide sprays (i.e. ability to enter fields, fungicide wash-off potential) in potato fields are also reported. In addition, potato inspectors report first signs of blight every season.

Since fungicide sprays generally protect crops for two weeks, warnings are normally issued at intervals no less than two weeks apart even if further blight-weather conditions occur. Intervals between warnings may be reduced if a major spell of blight-weather is imminent.

Action threshold

When warnings of upcoming blight-weather conditions occur. EBH values have also been used to evaluate action thresholds, Keane suggests that significant development of blight was likely to follow blight-weather spells of conditions amounting to 25 EBH or more during the preceding 10 days.

Model validation

This model has been used in Ireland since the early 1950s.

Current limitations

Since this model has been developed for the temperate conditions of Ireland and it may not be suited for use in California, however, equivalent criteria to identify blight-weather spells and EBH action thresholds could be developed and evaluated.

Top of page

Model 14 of 16

Model developer and citation

Smith, L. P. 1956. Potato blight forecasting by 90% humidity criteria. Plant Pathology 5:83-87.

Sensor location

Off-site, at regional weather stations.

Input variables

Enivronmental: Minimum temperature and hourly relative humidity.

Calculated: Number of hours of relative humidity 90% or higher.

Model description

This model was proposed by L. P. Smith as an alternative to reduce the number of invalid late blight forecasts by the Beaumont rules in England and Wales. According to the Smith criteria, two consecutive days with minimum temperature of 10 C (50 F) with at least 11 hours of relative humidity 90% or higher are favorable for the development of Phytophthora infestans on potato.

Action threshold

Initiate treatments when two consecutive days with minimum temperature of 10 C (50 F) with at least 11 hours of relative humidity 90% or higher have occurred.

Model validation

This model is currently under validation in England and Wales (Hims et al., 1995). Preliminary results indicate that spray programs guided by the Smith criteria achieved a comparable degree of control as calendar based spray programs. In high risk areas for late blight, the model advised the same number of sprays as the routine spray program, but in low risk areas it advised 3 to 5 fewer sprays.

Current limitations

The Smith criteria was developed for the temperate climatic conditions of England and Wales and may not be directly usable under California conditions. However, the use of the Smith criteria plus the use of a maximum temperature threshold should be explored.

Related work

Hims, M. J., M. C. Taylor, R. F. Leach, N. J. Bradshaw, and N.V. Hardwick, 1995. Field testing of blight risk prediction models by remote data collection using cellphone analogue networks, p. 220-225 In: Phytophthora infestans 150: European Association for Potato Research (EAPR)-Pathology Section Conference, held in Trinity College, Dublin, Ireland, September 1995 to mark the one hundred and fiftieth anniversary of the first record of potato blight in Ireland and the subsequent famine. L. J. Dowley, et al. (Eds). Boole Press, Ltd. Dublin. pp. 220-225.

Top of page

Model 15 of 16

Model developer and citation

Cook, H. T. 1949. Forecasting late blight epiphytotics of potatoes and tomatoes. Journal of Agricultural Research, 78:54-563.

Nugent, T. J. 1950. Three years experience forecasting late blight in Tidewater, Virginia. In: Plant Disease forecasting: A symposium. Plant Disease Reporter Suppl., 190:9-13.

Sensor location

Off-site, at regional weather stations.

Input variables

Environmental: Daily average temperature and rainfall.

Calculated: 7-day means of average daily temperatures. Cumulative rainfall up to the earliest recorded late blight attack.

Model description

This model was developed by H. T. Cook for use in Virginia and later was applied by T.J. Nugent. It was developed for fields were late blight occurs in less than 50 per cent of the years. The critical temperature above which late blight is not considered to be important is 75 F. Cumulative rainfall is calculated starting at a date prior to the earliest recorded late blight attack. Rainfall data from late blight and disease free years are plotted for a season. A straight median line (the 'critical rainfall line') is drawn between the two rain fall lines generated either by least squares regression or by visual inspection. Conditions are considered favorable for late blight development when cumulative rainfall goes above the critical rainfall line and the 7-day mean temperature remains below 75 F. At least two consecutive weeks of favorable weather are considered necessary for a serious outbreak of blight to develop.

Action threshold

Fungicide sprays are recommended after at least two consecutive weeks of favorable weather have occurred. Favorable weather has occurred when cumulative rainfall goes above the critical rainfall line and the 7-day mean temperature remains below 75 F.

Model validation

This model has been used in Virginia, Connecticut and British Columbia, but was not found to be effective in England, Wales, Iowa, Wisconsin and Indiana (Bourke, 1955).

Current limitations

Favorable periods of high humidity may be independent of the amount of rainfall. Early season rainfall may be so deficient or so excessive as to overly influence the character of the cumulative rainfall curves for the entire growing season.

Related work

Bourke, P. M. Austin, 1955. The forecasting from weather data of potato blight and other plant diseases and pests, World Meteorological Organization, Technical Note # 10, WMO No. 42 TP 16, Geneva, Switzerland.

Top of page

Model 16 of 16

Model developer and citation

Schepers, H. T. A. M., 1995. ProPhy: a computerized expert system for control of late blight in potatoes in the Netherlands. Proceedings XIII International Plant Protection Congress p. 948, (abs.).

Ridder, J. K., Bus, C. B. and Schepers, H. T. A. M. 1995. Experimenting with a decision support system against late blight in potatoes (ProPhy) in The Netherlands. In: Phytophthora infestans 150: European Association for Potato Research (EAPR)-Pathology Section Conference, held in Trinity College, Dublin, Ireland, September 1995 to mark the one hundred and fiftieth anniversary of the first record of potato blight in Ireland and the subsequent famine. L. J. Dowley, et al. (Eds). Boole Press, Ltd. Dublin. pp 191-200.

Sensor location

In-field weather stations, sensors within the canopy.

Input variables

Environmental: Temperature, relative humidity, rainfall and leaf wetness.

Calculated: Hours with high relative humidity, maximum and minimum temperatures, hours of leaf wetness, and hours of rain. Number of favorable days for development of Phytophthora infestans.

Host: Cultivar resistance, time until the crop reaches a height of 15 cm, new leaf area formed during the protection period of a fungicide.

Model description

ProPhy is a computerized expert system developed in the Netherlands for farmer decision support. It integrates current knowledge of control measures for Phytophthora infestans on potato in that country.

ProPhy recommends a first fungicide spray in sensitive varieties when the crop reaches a height of 15 cm or ten days later if a more resistant cultivar is used. When late blight is observed in the region, the first spray is recommended earlier. Subsequent treatments are scheduled by combining the fungicide protection status of a crop with the weather conditions.

It is assumed that a sensitive potato cultivar is protected for eight days after spraying with a protectant fungicide. In more resistant cultivars this period may be extended for 1, 2, or 3 days depending on the degree of resistance and several other parameters. Factors such as whether a grower has used a reduced dosage of fungicide, whether or not new leaves have been formed, occurrence of rain, and the number of favorable days that occurred in a period of three to ten days from the present time may shorten or lengthen the protection period of the fungicide. The assumed protection period of a fungicide can only be shortened to a maximum of three days.

A favorable day for the development of Phytophtora infestans occurs when at least six hours with high relative humidity plus at least two hours of leaf wetness or rain occur in a period from 8 PM on the previous day until 12 AM on the day of evaluation. The temperature must be in the range of 8 to 25 C. If viable lesions are observed close to the site, calculation of favorable days is overruled and a spray is advised.

ProPhy formulates recommendations on the basis of the calculated protection periods and the weather forecast. If the crop is no longer present and weather conditions are favorable for infection, a cymoxanil-containing fungicide is recommended.

Action threshold

Sprays are recommended when the crop is not protected by fungicides and favorable weather for late blight development is forecasted.

Model validation

Results of trials done in 1993 and 1994 in the Netherlands showed that growers using the ProPhy model used fewer fungicide applications than those using a 7-day calendar based model yet they achieved similar disease control.

Current limitations

This model was developed for weather conditions in the Netherlands and is based on the use of cymoxanil fungicides, which are not registered for use in California. In validation trials done by Ridder et al., 1996, the higher cost of cymoxanil fungicides appeared to eliminate the economic advantages of using fewer applications.

Top of page


Statewide IPM Program, Agriculture and Natural Resources, University of California
All contents copyright © 2005 The Regents of the University of California. All rights reserved.

For noncommercial purposes only, any Web site may link directly to this page. FOR ALL OTHER USES or more information, read Legal Notices. Unfortunately, we cannot provide individual solutions to specific pest problems. See our Home page, or in the U.S., contact your local Cooperative Extension office for assistance.

Agriculture and Natural Resources, University of California

Accessibility   /DISEASE/DATABASE/potatolateblight.html revised: March 28, 2005. Contact webmaster.