Research and IPM

Models: Diseases

Crop: Pear and Apple

Disease: Fire Blight
Pathogen: Erwinia amylovora

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

Model developer and citation

Steiner, P. W. and G. W. Lightner. 1996. Maryblyt^TM 4.3. A Predictive Program for Forecasting Fire Blight Disease in Apples and Pears. University of Maryland, College Park.

Steiner, P.W. 1990. Predicting Apple Blossom Infections by Erwinia amylovora Using the Maryblyt Model. Acta Horticulturae 273:139-148.

Steiner, P.W. 1990. Predicting Canker, Shoot and Trauma Blight Phases of Apple Fire Blight Epidemics Using the Maryblyt Model. Acta Horticulturae 273:149-158.

Sensor location

On-site, within canopy. Data from weather stations can also be used for general forecasts.

Input variables

Environmental: Daily minimum and maximum temperature, rainfall, and leaf wetness.

Calculated: Degree-hours and degree-days.

Host: Bud phenology.

Description of model

Maryblyt is a comprehensive computer program for predicting specific infection events and symptom development for most phases of fire blight epidemics in apples and pears. The program predicts specific infection events and the appearance of four distinct types of fire blight symptoms: blossom, canker, shoot and trauma blight. It does not predict rootstock blight. The program can be operated in real time to assess the current risks or progress of an epidemic, or in a simulation mode for predicting future events using forecasted weather data. Information generated in both modes of the program provides a basis for making decisions concerning when to make specific control treatments and when it is reasonably "safe" to delay them.

Maryblyt uses three cumulative heat unit measures to indirectly monitor the development of the host, pathogen populations, insect vector availability, and symptom development.

• Cumulative DD>40F (4.4C) are used to monitor the age of flowers and the appearance of insect vectors.
• Cumulative DH > 65F (18.3 C) are used to establish the epiphytic infection potential (EIP), which is an index for infection risk. EIP is calculated by expressing the number of DH>65F accumulated over the last 80 (apple) or 120 (pear) DD>40F as a percentage of 198 DH>65F, which represents the threshold for infection. For example, if EIP is below 100 then few infections will occur, if EIP is 100-150 this is sufficient to support a blossom blight epidemic, and if EIP exceeds 200-250 then a large number of infections can be expected with any wetting event.
• Cumulative DD > 55F (12.7C) are used to predict symptom development, once infection has occurred.

DD and DH are calculated by the program using a sine wave function with a 90F (32C) maximum and various minimum temperature thresholds.

Blossom Blight: The Maryblyt model assumes an abundance of inoculum. For a blossom infection to occur, four conditions must be met in sequence:

• flowers must be open with stigmas and petals intact.
• accumulation of at least 198 DH > 65F (110 DH > 18.3C) within the last 120 DD > 40F for pears and within the last 80 DD > 40F (44.4 DD > 4.4C) for apples.
• occurrence of either dew, a rain of 0.01 inch (0.25 mm) or greater during the current day, or 0.10 inches (2.5mm) during the previous day.
• daily average temperature greater than or equal to 60 F (15.6 C).

When all of these minimum conditions are met in sequence, infection occurs and first blossom blight symptoms are expected after an accumulation of 103 additional DD > 55F (57 DD > 12.7 C). DH are reduced by one-third, one-half or reset to zero if the temperature does not surpass a threshold of 64 F (17.8) during one, two or three days, respectively. However, once DH exceeds 400 (EIP = 200), no negative adjustments are made.

Canker Blight: Canker blight is predicted with the accumulation of at least 196 DD > 55F (109 DD > 12.7 C) after green tip.

Shoot Blight: Maryblyt forecasts only the first early shoot blight symptoms. Forecasts are based on the assumption that insect vectors are present. These early symptoms usually develop with the accumulation of 103 DD > 55F (57 DD > 12.7 C) after the first appearance of either blossoms or canker blight symptoms in the immediate area when average daily temperature is 60F (15.6 C) or above. Maryblyt can also be used to identify different insects contributing to shoot blight by subtracting 103 DD from the total of DD > 55 F shown by the program on the day of first shoot blight symptoms.

Trauma Blight: Trauma blight symptoms are based on infections associated with late frosts, hail or high winds. Symptoms can be expected when the EIP reaches 100, but are generally more severe when the EIP exceeds 200-250.

Action threshold

According to the model, risk is high if three of the four conditions for blossom infection are met. One antibiotic spray is recommended at bloom stage if the risk is high and an infection event is predicted for the next day.

Insecticide sprays to control insect vectors associated with fire blight (leafhoppers, plants bugs, psylla) should be maintained during the rapid vegetative growth phase to reduce the incidence of shoot blight. Treatments should begin when blossom or canker blight symptoms appear and, preferably, as the suspect insect species that vector the pathogen begin to appear.

Maryblyt can help signal when specific types of early symptoms begin. Removal of infected tissue may be effective if fire blight incidence is low or small outbreaks are localized in the orchard.

In addition to weather and crop development based timing of bactericide applications, a copper spray at bud break to reduce the efficiency of over-wintering inoculum plus a second spray not later than green tip stage may provide protection against fire blight.

Integrated fire blight management involves several other actions in addition to copper and antibiotic sprays. Dormant pruning of blighted limbs, shoots and cankers, fertility management to control the amount and duration of succulent tissues, stress avoidance, insecticide sprays and removal of infected shoots and limbs should also be considered.

Model validation

By Brent Holtz, UCCE Madera County and Beth Teviotdale, Plant Pathology, UC Davis (at Kearney Agricultural Center), for apples as part of the Pestcast Project during the years 1997-1999.

van der Zwet, T., A. R. Biggs, R. Heflebower, and G. W. Lightner. 1994. Evaluation of the MARYBLYT Computer Model for Predicting Blossom Blight on Apple in West Virginia and Maryland. Plant Disease 78:225-230.

Bonn, W. G., and T. Leuty. 1993. An Assessment of the Maryblyt Computer Program for the Prediction of Fire Blight in Ontario, Canada. Acta Horticulturae 338:145-152.

Jones, A . L. 1992. Evaluation of the Computer Model MARYBLYT for Predicting Fire Blight Blossom Infection on Apple in Michigan. Plant Disease 76:344-347.

Model implementation

MARYBLYT is used by growers and in research, extension and teaching programs in 32 US States and in at least 36 countries.

Current limitations of model

The model does not always explain why fire blight does or does not develop in some locations.

The model stops the blossom blight prediction phase as soon as petal fall stage is entered. Therefore this stage should not be entered until the very last blossom has dropped its petals.

Certain wetness situations and frost damages have to be judged subjectively, which may influence model output.

Model vendor

Gempler’s, Inc.

Model developer and citation

Smith, T. J. 1993. A Predictive Model for Forecasting Fire Blight of Pear and Apple in Washington State. Acta Horticulturae 338:153-157.

Smith, T . J. A Risk Assessment Model for Fire Blight of Apple and Pear (Fahrenheit version). Washington State University Cooperative Extension document.

See the author's Web site for more information.

Sensor location

Within canopy.

Input variables

Environmental: Rainfall, leaf wetness and temperature.

Calculated: Degree-hours (number of hours above a 60 F temperature threshold).

Host: Variety, age, vigor, presence of blossoms.

Pathogen: Potential for pathogen presence due to proximity of fire blight cankers.

Description of model

The "Cougarblight" model was developed for fire blight of pear and apple in Washington state. It uses temperature data to estimate the growth rate of fire blight bacteria (Erwinia amylovora) over the past three days plus the present day, if wetting occurs in the afternoon of evening, or the previous four days if wetting occurs in the morning. Each day blossoms are open, the degree-hours for the noted days are added to obtain a four day degree-hour total. The goal is to determine what sort of growing conditions the bacteria have had while on the stigma during the approximately 96 hours prior to a 3+ hour blossom wetting period. Degree-hours are calculated from average hourly temperatures, or from daily minimum/maximum temperatures based on an estimated degree-hour look up chart (see Table 1). Calculation of the "four-day degree-hour total" must be done for each day. If blossoms are wetted by rain, four or more hours of dew, or any significant wetting, refer to the degree-hour total and Table 2 below to evaluate the potential risk of infection.

Action threshold

Control actions are advised if a high or extreme risk infection period is detected. Factors determining potential for fire blight damage to the trees (flower numbers, tree age, vigor and variety) should also be considered.

Model validation

This model was validated in Washington state by T.J. Smith throughout the 1980’s.

This model is being validated by Brent Holtz, UCCE Madera County and Beth Teviotdale, Plant Pathology, UC Davis (at Kearney Agricultural Center), for apples as part of the Pestcast Project during the years 1997-1999.

Model implementation

The Cougarblight model has been implemented by growers and advisors in Washington state since 1991. Growers and advisors who used the model in 1993 and 1994 sprayed far less frequently and with better control than those who followed more traditional control methods.

Future directions

Degree-hour totals relative to daily high and low temperatures vary from region to region, therefore degree-hour accumulation tables should be developed for each location of intended use based on regional weather station data. As an alternative, individual growers can use their own on-site weather stations to calculate degree-hours, and then refer to Table 2 to assess the risk level.

Model developer and citation

Gubler, W.D, Lindow, S., Zoller, B., and Duncan, R. 1999. Pear Diseases in Production and Handling of California Pears. University of California, Division of Agriculture and Natural Resources Publication.

Zoller, B.G. 1990. Controlling Fire Blight of Pear Using Heat Summation to Predict Blossom Blight. Proceedings for Pear Short Course. University of California Cooperative Extension, Konocti Harbor Resort, November 27-29.

van der Zwet, T., Zoller, B. G., and Thomson, V.S. 1988. Controlling Fire Blight of Pear and Apple by Accurate Prediction of the Blossom Blight Phase. Plant Disease 72: 464-472.

Sensor location

Standard weather shelter adjacent to the orchard.

Input variables

Environmental: Temperature, relative humidity, precipitation.

Calculated: Degree-hours. 1 F-degree-hour = 1 F-DH = 1 degree above 65 F for 1 hour.

Host: European and Asian pear.

Description of the model

This model was developed from five years of field observations on the development of Erwinia amylovorain Bartlett pear blossoms. It is based on the correlation between the number of accumulated F-degree-hours and the incidence of random new blight infections per holdover infection during an 11 year period. Degree-hours accumulate each hour of the day until three consecutive days occur where the temperature is below 66F. In this case the accumulation of degree-hours is reset to zero, unless the accumulated total exceeds 400 F hours and either, it has rained or warm, humid conditions have occurred with the temperature at least 57F and the humidity at least 90%. If the orchard is being irrigated, the humidity threshold is reduced to 80% instead of 90%. Irrigation of the orchard should be timed so that it does not occur during the bloom period.

Action threshold

Use Table 1 to determine the appropriate action.

Model validation

This model was developed using E. amylovorapopulation from 1972-76. The model was validated in California using data of disease observations from 1976-1986.

Model implementation

This model has been used in commercial orchards of California for more than 23 years.

Model limitations

Even when improved disease control had been achieved with the use of this model, the incidence of fire blight was 8.7 times greater in years when accumulated degree-hours during infection periods approached 1000 F-DH than when only 100 F-DH had accumulated during the 1976-86 period. Subsequent experience has shown that increasing treatment frequency during critical situations in high degree-hour seasons will further improve control, as noted in Table 1.

Different locations in California may experience differences in fire blight disease dynamics due to interactions of factors other than temperature.

Model developer and citation

Thomson, S.V., Schroth, M.N., Moller, W.J., and Reil, W.O. 1982. A Forecasting Model for Fire Blight of Pear. Plant Disease 66:576-579.

Sensor location

Temperature probe placed in the warmest part of the orchard.

Input variables

Environmental: Minimum and maximum temperature.

Calculated: Mean temperature (average of maximum and minimum temperature from midnight to midnight).

Description of model

Prediction of flower colonization by bacteria (Erwinia amylovora) is based on the daily mean temperature rising above a line drawn from 16.7 C on March 1 to 14.4 C on May 1. Subsequent applications should be made every five days until the end of bloom.

Fig. 1. Populations of Erwinia amylovora during bloom are usually detected in flower samples taken shortly after the mean temperature exceeds the prediction line.

Action threshold

Initiate applications when the mean daily temperature rises above the action threshold, represented by a line drawn from 16.7 C. on March 1 to 14.4 C on May 1, as shown in Figure 1.

Model validation

This model was validated in Lake, Mendocino, Solano, Santa Clara, Contra Costa, El Dorado, Glenn, Napa, Sacramento, Yolo and Yuba counties in a total of 132 orchards, using data from the 1970’s.

Model implementation

In California, implemented on 16,200 ha in 1978, with an estimated savings of $1,200,000 per year due to a reduced number of bactericides.

Current limitations of model

The mean temperature line frequently is exceeded before bacterial detection in the North Coast Range area due to lower nighttime temperatures than in the Sacramento area.

The use of daily mean temperature to predict the need for bactericide applications is a conservative approach as it does not take into account the presence of the pathogen, indicating that bactericides may be applied when there is no pathogen present and hence no risk of disease.

Model developer and citation

Billing, E. 1996. BIS95, an Improved Approach to Fire Blight Risk Assessment. Acta Horticulturae 411:121-126.

Sensor location

Official regionally based weather stations, field located data loggers, or minimum and maximum thermometers plus a rain gauge.

Input variables

Environmental: Daily minimum, maximum and mean temperature (C) , rain (mm), mist and heavy dew, storms with strong winds (30 kt, 15 m/sec or more), hail and/or heavy rain, thunderstorms in the area and frosts (-2 C or lower). Recording time should be 7-9 a.m.

Calculated: Degree-days above two different temperature thresholds: 18 C and 13 C.

Host: Bud burst (green tip), blossom periods (first open flowers, full bloom, 80-90% petal fall), late flower production, secondary blossom production, shoot growth periods, tissue damage due to frost or wind.

Pathogen: New disease (pre-bloom, blossom, shoot and fruit blight) and signs of stem invasion.

Vector: Numbers and activity of bees or other insects that can spread the bacteria during bloomtime.

Description of model

Billing’s integrated system (BIS) was designed as a substitute for Billing’s revised system (BRS) to assess the risk of fire blight in apples, pears and related hosts. It is flexible enough to be applied to all hosts in all climatic areas. The model uses two types of degree-day calculations, which are counted and summed on a daily basis, to help assess the risk of fire blight.

• DD18 = the sum of daily values of 1.0 C or more above 18 C for the maximum temperature. If the maximum daily temperature is 21 C then 3 is added to the DD18 sum. DD18 calculations begin on the first day of bloom, and continue throughout the bloom period. If the maximum temperature falls to 16-17 C for two days or to 15 C or lower for one day, the DD18 sum is reset to zero.
• DD13 = the sum of daily values of 0.5 C or more above a 13 C mean. DD13 calculations begin on the day after each infection risk (IR) day.

The model also includes a calculation of relative risk of direct infection, of which there are two kinds, blossom infection risk (BIR) and infection risk (IR). In general, infection risk depends on inoculum potential (IP), host/target susceptibility (Su) which is increased by tissue damage (TD), and warmth and wetness (WW) at the time of infection, such that

IR = IP+Su(TD)+WW.

Blossom infection risk (BIR) occurs in situations where IP levels depend on flower colonization by the pathogen and spread by insects. A day is classified as a BIR day if :

1. The DD18 sum is >= 17, the open flowers are wet by heavy dew, mist, or rain, and the mean temperature is 15C or more on the day of wetting.
2. The WW score for a day based on the mean temperature (C) on the wet day or the day before the wet day is between 2 and 6 as indicated in Table 1.
3. There is no wetting event but the maximum temperature for the day is >= 27C and/or the mean temperature for the day is >= 20C.

Infection risk (IR) events occur in situations where IP levels at the targets depend on the spread of ooze by rain. A day is classified as IR if:

1. The IP is principally coming from ooze and is spread by rain during pre-bloom, blossom, young shoot and fruit growth stages.
2. The WW score for the day (based on the mean temperature (C) on the wet day or the day before) is between 2 and 6 as indicated in Table 1.

In addition to degree-day and infection risk calculations based on weather data, the model also involves intensive orchard scouting and systematic recording of host growth stages, insect activity, and disease incidence to determine risk of new disease.

Action threshold

According to the model, blossom blight risk is moderate to high if the DD18 sum is 17 or more (47 for apples), while the risk is high if the DD18 sum is 34 or more. Blossom blight has not been observed with high WW scores alone. Shoot blight infection is common and the disease incidence highest when there is tissue damage by strong winds and/or hail during storms with wind and driven rain.

The DD13 sum is used to time orchard scouting for signs of new disease. From several years of observation the BIS95 model predicts that new disease will be seen in pears within 0 to 3 days of the DD13 sum reaching 17 (47 for apples).

Table 2 shows the values of DD18, maximum temperature, mean temperature, and rainfall which tend to correspond to each risk level.

Model validation

This model has been tested in England.

Model implementation

No information available on this new version.

Current limitations of model

Usefulness of the model depends on the quantity and quality of the data input and the knowledge and experience of the user.

Top of page

Model developer and citation

Billing, E. 1992. Billing's Revised System (BRS) for Fire Blight Risk Assessment. Bulletin OEPP/EPPO 22: 1-102.

Billing, E. 1990. Fire Blight Concepts and a Revised Approach to Risk Assessment. Acta Horticulturae 273: 163-170.

Sensor location

Official regionally based weather stations.

Input variables

Environmental: Daily maximum and minimum temperature, rainfall and storm activity recorded at 9 a.m. and credited to the previous day (thrown back).

Host: Bud phenology, in particular blossom periods (first open flowers, full bloom, 80-90% petal fall), late flower production, secondary blossom production, shoot growth periods, tissue damage due to frost or wind.

Pathogen: Potential daily doubling of bacteria (based on temperature), observed disease incidence.

Vector: Number and activity of bees or other insects that can spread the bacteria during bloom-time.

Description of model

Billing’s revised system (BRS) is a modification of Billing’s original system (BOS, see model 7), which was intended to make it more simple, logical, and precise. BRS is intended to offer guidance to pathologists and growers with knowledge of the essentials of fire blight epidemiology and of local orchard disease incidence. This model uses weather data to assess the risk of fire blight infection and estimate the disease development rate. It also uses intensive scouting and systematic recording of host growth stages and disease incidence to determine the risk of new disease.

The model has the following components:

• Potential doublings (PD) are determined according to Schouten (1987), as specified in Table 1 below, instead of Billing’s (1980) original table.
• Rain is assigned a score of 1.0 if the quantity of rainfall is 2.5 mm or higher, 0.5 if the quantity of rain is 2.4 mm or less, 0.5 if there is heavy dew or irrigation during the bloom period, 0.5 if PD>=11 during the bloom period only, or 0 if there is no rain.
• The infection risk score (WIR) depends on the PD score and the rain score, as given in Table 2.
• The method for determining the length of the disease development period (D) is the same as in BOS. Calculations begin on days when the WIR score is greater than or equal to 2. This is day zero. The disease development period ends when PD>=(36t/R)-6. In this equation PD is the sum of the PD scores since day zero, t is the number of days, and R is the cumulative rain score since day zero. The equation shows that when the weather is wet, less heat is needed to complete the incubation period, and when the weather is dry, more heat is needed to complete the incubation period.
• Various graphical representations of the weather data and calculated variables are available to assess the risk level both within and across years.

Note that in Table 2 a trace amount of rain is equivalent to less than 0.1 mm, mist, heavy dew, or irrigation. When the PD score is higher than 11.0, the WIR score is 2 when there is no rain during bloom only.

Action threshold

The model provides information on when to look for signs of new disease and when to apply sprays during bloom. From bud-break, search for signs of new disease when D-periods end, particularly D-periods which began on a day with the WIR score >=4. In low risk years, BRS can reduce the time spent searching for signs of disease.

Spray guidelines include the following:

• In the pre-bloom period, spray the day before the WIR score is forecasted to be >=2.
• Spray at least once at full bloom regardless of weather conditions.
• During bloom, sprays are based on estimated inoculum levels using D-period analysis, and estimated spread of inoculum by insects, as suggested by the frequency of temperatures >=21C.

The WIR score alone is not a reliable guide to infection risk because it does not in itself take into account inoculum level or host susceptibility. A low score may still be associated with high risk, or a high score associated with low risk, depending on the level of inoculum present, as estimated by the D-period length.

Model validation

BOS, and subsequently BRS, was developed and tested over a period of 23 years (1968-1990) using data from England. The model has also been tested using weather and disease data from the USA. The model seems to be useful for diverse climatic areas.

Model implementation

Information not given.

Model limitations

This model is intended to be used by people with previous experience and knowledge of the epidemiology of fire blight. Other limitations of BRS, according to E. Billing, are mainly due to a lack of knowledge of the disease and the quality of the data used as input into the model, problems that are also true for other disease models.

Related work

Schouten, H.J. 1987. A Revision of Billing's Potential Doubling Table for Fire Blight Prediction. Netherlands Journal of Plant Pathology 93:55-60. 

Model developer and citation

Billing, E. 1984. Principles and Applications of Fire Blight Risk Assessment Systems. Acta Horticulturae 151:15-22.

Billing, E. 1980a. Fire Blight in Kent, England in Relation to Weather (1955-1976). Annals of Applied Biology 95:341-464.

Billing, E. 1980b. Fire Blight (Erwinia amylovora) and Weather: a Comparison of Warning Systems. Annals of Applied Biology 95:365-377.

Sensor location

Regionally based weather stations.

Input variables

Environmental: Daily rainfall and daily maximum and minimum temperature.

Pathogen: Daily potential doublings of bacteria (PD) (based on temperature).

Description of model

This model assesses the potential for fire blight of pears and apples based on daily minimum and maximum temperatures and rainfall, and can be used to help determine the most effective times to apply chemicals.

The model uses daily potential doublings (PD) of the bacteria plus rain scores to estimate potential infection days (PIF), theoretical incubation period ( I ), and potential for fire blight activity (PFA).

• PD is the number of times the pathogen would be expected to divide each day if temperature was the only limiting factor, as given in Table 1.
• The rain score is 1.0 if daily rainfall is 2.5 mm or more, 0.5 if daily rain is below 2.5 mm, and 0 if there is no rain. R scores were defined empirically based on the correlation between rain and the number of infected trees.
• In spring, moisture may not be a limiting factor, so R is given a score of 1.0 up through May 21, regardless of rainfall. This does not count as a PIF day.
• A PIF day is any day when there are 2.5 mm of rain or more, or through June 5 any day when PD equals nine or more (R is scored as 1.0 on such days for purposes of calculating the I period).
• The incubation period (I) is the time between infection and early symptom expression, which is also the time of production of fresh inoculum. This is the time it takes for an infected cell to reach, through doubling, a population of 10⁹, or 30 doublings. The I period is complete when PD >=(36t/R)-6. In this equation PD is the sum of the PD scores since the day of infection, t is the number of days since infection occurred (t=0 on the day of infection), and R is the cumulative rain score since the day of infection. The equation shows that when the weather is wet, less heat is needed to complete the incubation period, and when the weather is dry, more heat is needed to complete the incubation period.
• Various graphical representations of the weather data and calculated variables are available to assess the risk level both within and across years.

Action threshold

Table 2 can be used to assess infection risk during spring bloom time. Initiate sprays during the first medium or high risk day, or during a low risk day if high inoculum levels or future favorable weather are suspected. Continue sprays according to observed inoculum levels, rate of new blossoms opening, dates of completion of I periods, occurrence of medium to high risk infection days, and likely insect activity. Spray immediately after any damaging storms.

Model validation

This model has been tested for a range of conditions. Adjustments may be needed for specific locations, particularly dry climates.

Dinesen, I., Friis, E., Oleson, J.E., 1983. Climate and Fire blight: Billing's "System 1" Tested Under Danish Conditions and Computerized for Operational Use. Acta Horticulturae 151: 79-83.

Model implementation

Extensively implemented in Europe. It has been replaced by Billing’s revised system (BRS).

Current limitations of model

Usefulness of PD tables should be tested for each region. Suggestions offered by the model should not be taken as final solutions but as a working basis for further tests, used in conjunction with local knowledge and experience and adjusted if necessary to suit local conditions.

Related work

Billing, E. 1979. Fire Blight: the Development of a Predictive System. In Plant Pathogens pp. 51-59, edited by D.W. Lovelock. Academic Press, London.

Model developer and citation

Jacquart-Romon, C. and Paulin, J.P. 1991. A Computerized Warning System for Fire Blight Control. Agronomie 11: 511-519.

Sensor location

Within orchard using an automatic weather station.

Input variables

Environmental: Daily minimum and maximum temperature, rainfall, and occurrence of storms (strong winds and/or heavy rain and hail).

Calculated: Daily rain above or below a 2.5 mm threshold, forecast of minimum and maximum temperature and rainfall for the next day.

Host: Stage of development (D, E₂ or G, according to Fleckinger, 1945), presence of succulent shoots and occurrence of secondary blossoms.

Pathogen: Nature of symptoms in time and space (old or recent lesions in the orchard to be protected or in adjacent orchards).

Description of model

This IBM-PC computer based model was developed as a warning system for fire blight of pears and apples. The model is intended for use between February 1st and the end of July, and uses climatic and inoculum potential values and weather forecasts to evaluate fire blight risk.

Potential doublings of bacterial populations and the end of incubation periods are calculated according to Billing (see model 7), and are used to define scales of climatic potential (CP) for specific phenological stages of the plant, as is given in Tables 1 to 3. Inoculum potential (IP) during the pre-bloom and blossom periods (Table 4), and during the post-bloom period (Table 5) is estimated based on records of disease in the field under protection and nearby fields. Fire blight risks and recommended actions during the pre-bloom and blossom periods are defined for each combination of CP and IP.

Action threshold

Table 6 indicates the type of action required for each CP and IP combination, where N suggests no action is required, V suggests a visit to the plot to detect and remove symptoms, and S suggests spraying with a suitable chemical.

Decisions noted with a superscripted 1 suggest a spray if very active pollinating insects and symptoms are present in the orchard. Decisions noted with a superscripted 2 suggest a visit only if secondary blossoms are present. Decisions noted with a superscripted 3 suggest spraying and/or removing secondary blossoms.

Model validation

The model has been used under different climatic situations in France over several years. Results are published in:

Jacquart-Romon, C. and J.P. Paulin. 1990. Preliminary Experimentation of a Computerized Warning System for the Control of Fire Blight. Acta Horticulturae 273: 131-137.

Lecomte. P., Jacquart-Romon, C. and Paulin, J.P. 1996. Evaluation and Utilization of FIRESCREENS, a Computer Program for the Control of Fire Blight. Bulletin OEPP/EPPO 26:549-553.

Model implementation

No data available.

Current limitations of model

CP is based on Billing’s original system, which was validated under local conditions before implementation. This system should be studied under local climatic conditions before implementation.

Model developer and citation

Timmermans, Y. 1990. A Warning System for Fire Blight on Pears in Belgium: Preliminary Model and Practical Prospects. Acta Horticulturae 273: 121-137.

Sensor location

Information not given.

Input variables

Environmental: Temperature, rain and dew.

Calculated: Degree-hours above 18.3 C.

Host: Presence of flowers, mean number of flowers, shoots and fruit, susceptibility of cultivar, tissue age, host vigor.

Pathogen: Number of primary inoculum sources (cankers) in the orchard.

Description of model

This simulation model was constructed to adjust existing warning systems of pear fire blight to the conditions prevailing in Belgium. The model consists of three successive stages:

• Assessment of disease risk.
• Calculation of incubation time.
• Determination of symptom expression.

Each day, the simulation expresses the relative number of infected flowers, shoots and fruits and the relative quantities of symptoms resulting from previous infections.

In this model, the assessment of the number and spatial distribution of inoculum sources is considered more important than the evaluation of bacterial populations in the estimation of inoculum potential in an orchard. The model considers primary (INOCS) and secondary (INOCS2) inoculum sources, which are specified as

INOCS = ACS*INOCH

INOC2 = INOC21+INOC22, where

ACS = canker reactivation associated to the accumulation of degree hours > 18.3C. INOCH = number of primary sources of inoculum.

INOC21 = share of total symptoms removed daily from the orchard.

INOC22 = symptoms left unnoticed by the grower.

Inoculum dispersal (DISP) is modeled as a function of maximum temperature (TX), presence of flowers (FL), rain (RR), and storms (ST).

DISP = f(TX, RR, ST, FL)

Infection risk (INFx) is estimated from weather (INFCLIM) and host dependent factors (INFORG):

INFx = INFCLIM*INFORG

INFCLIM is a function of potential doublings (PD) of Erwinia amylovora, calculated according to Schouten (see reference under related work for model 6) and rain (RR).

INFCLIM = f (PD, RR)

INFORG is calculated from the number (Nx) and susceptibility (Sx) of host targets:

INFORG = Nx * Sx, where

x = flowers, shoots , pears per tree.

The total number of infected targets on the day of infection (INFOx) is calculated from the total number of infected trees on the day of infection (INOCO) and INFx.

INFOx = INOCO * INFx

INOCO = (1+DISP) * (INOCS +INOC2)

Use of INOCO assumes that one tree is bearing only one inoculum source.

Incubation period (INCx) is calculated as the sum of internal multiplication time (MULTx) and latency before external visualization of the disease (EXPR):

INCx = MULTx + EXPR

Multiplication of Erwinia amylovora is a function of temperature and moisture (PD and RR) inside infected tissues. The model assumes that multiplication varies beyond a minimal value corresponding to optimal host conditions for each type of infection:

MULTx = ; (PD* (1+y*RR)/z)

After completion of internal multiplication, symptom expression is modeled as a function of temperature:

EXPR = f (TX)

Total symptoms for each infection type (SYMPTx) are equivalent to INFOx. Daily symptoms (SYMPDx) are the proportions of SYMPTx appearing each day, which is a function of INCx:

SYMPDx = f(INCx)

The total daily number of symptoms represents the sum of all the symptoms resulting from previous infection days whose symptom expression period overlap.

Action threshold

No action thresholds were reported.

Model validation

The model was tested during 1982-1988. Good agreement in the dates of appearance and the quantities of symptoms was observed.

Model implementation

Not specified.

Current limitations of model

Additional data is needed to validate the model. Assessment of factors was done using working hypothesis on scoring values.

Model developer and citation

Mills, W. D. 1955. Fire Blight Development on Apple in Western New York. Plant Disease Rep. 39:206-207.

Sensor location

Weather stations

Input variables

Environmental: Maximum temperature, rainfall and relative humidity.

Description of the model

This model was developed by W. D. Mills from disease records and weather data from counties bordering Lake Ontario. According to this model, blossom blight of apple only occurs with maximum temperatures during bloom over 65 F, with precipitation or very high humidity.

Action threshold

Sprays should be applied on the first day after the beginning of bloom when the maximum temperature exceeds 65 F and precipitation or high humidity has been forecasted. If these conditions do not arrive before full bloom, streptomycin should be applied and a protection period of up to seven days should be considered to be in effect if no rain occurs. If rain follows a streptomycin spray, the treatment should be repeated after the rain as soon as temperatures over 65 F with precipitation or high humidity occur. If the first spray is applied early in bloom, further applications every 4 to 7 days are indicated if favorable conditions for disease are met. No sprays should be made afterwards if these conditions do not occur.

Model validation

This model was validated for five years by Luepschen, Parker and Mills (1961). They found that a minimum of two favorable days were required for severe blossom infections.

Model implementation

This model has been used in California, Ohio, Virginia and Pennsylvania. Partial control and some failures have been reported from Michigan and New York.

Top of page