Research and IPM

Models: Diseases

Crop: Tomato

Disease: Powdery Mildew
Pathogen: Leveillula taurica

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 this location. See "Validation Work" below.

Model 1 of 1

Model developer and citation

Guzman-Plazola, R. A. 1997. Development of a spray forecast model for tomato powdery mildew (Leveillula taurica (Lev) Arn). Ph.D. Thesis. University of California, Davis.

R. Michael Davis and J .J. Marois. Development of a spray forecast model for powdery mildew of tomato. California Tomato Research Institute 1995 Final Report. p. 52-59. In conjunction with R. Guzman-Plazola.

(See the UC IPM Web version of this model.)

Sensor location

One or more sensors of temperature (T) and relative humidity (RH) are located inside the foliage and toward the top of the plants. A T sensor and its corresponding RH sensor must be kept together and protected with a shield to avoid direct exposure to the sun and water. The shield must allow good air flow around the sensors. One or more leaf wetness (LW) sensors must also be used and located on top of the canopy on fully developed leaves. All sensors must be relocated periodically according to the dynamics of plant growth and checked after periods of high speed winds. Each sensor should make several measurements during an hour.

Input variables

Environmental: Hourly average, maximum, and minimum temperatures (T), hourly average, maximum, and minimum relative humidity (RH), hourly average leaf wetness (LW).

Calculated: Daily values of the following: average T and RH; average hourly range of T and RH; range of Tmax and RHmax; average of the amount by which T>27.4 C; number of hours where 27.5<=T<=32.4 C; number of hours 5<=LW<=10; number of hours 17.5<=T<=22.4 C and RH>=40%; number of hours T>=32.5C; average of the amount by which RH<40%; number of hours T<12.5 C. If two sets of sensors are used, the hourly values are averaged before these calculations are made.

Model description

Daily environmental conditions are classified according to their conduciveness for powdery mildew development, in California typically from June to September. Conditions are measured several times every hour and values are averaged over the hour. Twenty-four hourly values are transformed into additional variables which are fed into a linear discriminant function. Using "No disease," "Moderate," and "Severe" coefficients, the function is evaluated three times. For each disease severity class, values of the 15 calculated daily variables are multiplied by their respective coefficients, summed, then added to the appropriate constant. The disease severity class sum with the largest value indicates whether the weather that day was nonconducive, moderately conducive, or conducive for powdery mildew development.

After this, management decision rules (see Action Threshold below) are applied on the basis of prevailing conditions on a set of six consecutive days. No fungicide spray is recommended when the six-day period is composed only of nonconducive days, moderate days, or any combination of them. A spray is recommended when at least three conducive days accumulate within the six-day period and no period of more than one nonconducive day has occurred. Occurrence of a minimum of two consecutive nonconducive days can cause 70-75% reduction in the final number of powdery mildew lesions. In the cases where no spray was recommended, the new set of six days for evaluation is modified by the elimination of the oldest day(s) and inclusion of the most recent day(s).

The six-day evaluation period corresponds to approximately half a latent period of Leveillula taurica under favorable conditions for disease. In greenhouse trials, sulfur or myclobutanil have controlled the disease when sprayed one to six days after inoculation. Making decisions after evaluating the microclimate for a period of six days is a strategy to assess the risk of disease severity by assuming what would happen if conditions remain the same during the next few days. Since in the absence of fungicides powdery mildew lesions start becoming visible about 13 days after inoculation under conducive conditions, the six-day waiting period provides a reasonable time to allow weather changes to suppress disease development.

The model assumes the following:

• Unlimited inoculum of Leveillula taurica is present in the field.
• The tomato cultivar in use is susceptible to Leveillula taurica.
• Fungicide sprays can reduce infection in progress and provide effective protection against the disease for 10 days.
• Location of sensors int he field and sensor measurements are representative of tomato canopy microclimate in the rest of the field.

(See the UC IPM Web version of this model.)

Action threshold

According to the model, daily conditions evaluated over a six-day period yield the disease assessments and recommended actions listed in Table 1.

Table 1. Decision rules, expected disease severity, and recommended actions based on evaluation of six-day period of daily conditions.

Model validation

In 1997, model validation will continue in the northern San Joaquin Valley by UC Farm Advisor Kent Brittan.

In 1995 and 1996, the moded was validated in several locations in the southern Sacramento Valley and northern San Joaquin Valley, by R. M. Davis, R. A. Guzman-Plazola, G. Miyao, B. Mullen, J. Valencia. Results are reported in R. Michael Davis and J. J. Marois. Development of a spray forecast model for powdery mildew of tomato. California Tomato Research Institute 1995 Final Report. P. 52-59.

Model implementation

Daily output from the model will be available through the California Tomato Weather Network in the 1998 field season.

Limitations of the model

Characterization of environmental conditions was done from microclimate data collected in 1993 and 1994 from a total of five tomato fields in California's Central Valley. Validation of the model has been done during two years (1995 and 1996) only. Efficacy of the discriminant function to correctly classify weather patterns for conduciveness to powdery mildew under a wider range of conditions must still be tested.

Future directions of the model

Creation of a larger data set of high resolution microclimate data from tomato canopies, with corresponding data on disease progress on several cultivars, would allow additional tests of the robustness of the discriminant function and possible adjustment of the function's coefficients.

Model detail (from Model description)

Linear Discriminant Function used to characterize weather data for conduciveness to powdery mildew of tomato on a daily basis. To classify the daily weather conditions, multiply the value of each variable by its respective coefficient in each of the three severity classes. Sum the resulting products in each column and add the constant at the top. Classify the day based on which column (severity class) has the highest value.

Top of page