RSS | Register/註冊 | Log in/登入
Site search:
Home>FFTC Document Database>Extension Bulletins>The Application of the Epidemiologic Simulation Model &Quot;Melan&Quot; to Control Citrus Melanose Caused by Diaporthe Citri (Faw.) Wolf.
The Application of the Epidemiologic Simulation Model &Quot;Melan&Quot; to Control Citrus Melanose Caused by Diaporthe Citri (Faw.) Wolf.
Shigematsu Kuhara*
Former Chief, Laboratory of Plant Pathology,
Kuchinotsu Branch, National Fruit Research Institute,
MAFF, Japan, 1999-12-01


A melanose control project, which was strongly supported by the simulation model "MELAN", was conducted jointly by research stations, agricultural cooperatives and citrus growers in Kyushu, southwest Japan. Several years' practice proved that the new control system supported by the simulation model is superior to conventional systems, particularly in providing a stable protective effect. The computer simulation model "MELAN" contributed much to the success of the project by providing a systems analysis of the ecology of the causal fungus, the infection process and disease development, as well as the protective effect of applied chemicals.

Abstracts in Other Languages: 中文(1087), 日本語(1194), 한국어(1178)


Integrated pest management is now widely accepted as the basic strategy for plant protection. Its final goal will be how to manage pests to keep numbers below the economic injury level, without any pesticide application. However, even IPM is forced to depend upon some pesticides. Thus, it is a major concern for plant protection how to minimize chemical inputs while maintaining crop quality at an economically acceptable level. In order to realize this aim, it is essential when practicing pest control to understand the state of a disease epidemic or pest outbreak at various points in time, and be able to predict its future development.

Advances in ecology and related sciences have revealed the mechanism of disease epidemics and insects outbreaks for many crops. Both generally involve a large number of factors which are mutually related. Most mechanisms are so complicated that extention staff and growers cannot be expected to understand the whole picture and use it as the basis for chemical control. Moreover, each step or component is generally expressed as numerical figures or mathematical equations. A practical way of utilizing this new scientific information for pest control might be to integrate the components and equations into a simulation model that is capable of giving a quantitative picture of the problem in the recent past, the present, and the near future.

Following the success of the "EPIDEM" model for Tomato Early Blight (Waggoner 1969), many models have been developed for various pests. Many of these are widely recognized as useful tools for pest management. The bottleneck in applying these models to pest control is how to complete a huge number of calculations in a short enough period of time for prompt decision-making.

Computerized simulation models offer a solution to the problem. Such a model contains all the necessary equations and parameters, and can provide timely information in response to inputs of data about weather and chemical applications. The key in using a simulation model for pest control is a high-performance computer. In the 1970s, when we began this project, it was not possible for a local pest control project to have access to such a large computer. Our project began on a computer time-share basis, using the business processing computers of agricultural cooperatives.

Nowadays, the situation has changed. Even in less industrialized countries, there is access to low-cost PCS with high performance. Many countries have also developed a computer network for data processing and storage. Encouraged by these developments, we feel it is now time to publish our work.

The Joint Project to Control Citrus Melanose


The target disease, melanose, is very important in citrus production in the southwest part of Japan, as is scab. Melanose causes serious economic loss to growers by spoiling the appearance of the fruit. Most of the fruit affected is graded as fit only for processing, and the price is only 1/5 of that paid for fruit eaten fresh. Fruit which is only lightly affected can be sold for table use, but is graded as poor and its price is 1/4 of that graded excellent ( Fig. 1(1160)).

Warm and humid summer weather is characteristic of citrus production in Kyushu. Annual precipitation reaches 2,000 mm, of which 80% falls over summer from March to October, particularly during the rainy season of May to mid July.

The average temperature at this time is 20.3°C, optimum for melanose occurrence. Favored by such a warm and humid climate, melanose is so widespread that conventional control practices often fail to suppress it, although these same controls give excellent results in other parts of Japan.

Consequently, citrus growers urgently needed a new control system that could cope flexibly with melanose outbreaks in Kyushu. Application of the simulation model "MELAN" (Koizumi 1980) was expected to be a breakthrough in this difficult situation.

The joint project was conducted from 1980 to 1985. Local agricultural cooperatives shared the work of collecting weather data, testing the simulation and issuing alarms about potential melanose outbreaks in their area. The size of the land unit area for simulation was generally several hundred hectares. Growers applied chemical pesticides as recommended by their local cooperative. National and prefectural research stations surveyed disease incidence in orchards and the extent of damage to fruit at packing facilities, to examine the correlation between results of the simulation and disease occurrence. They also provided technical assistance to both cooperatives and growers. The Kyushu Regional Office of the National Agricultural Cooperative Federation supported the project financially.

The model was originally written in Fortran, but was late rewritten in Cobol to adapt it for use by office computers. The project first began as a demonstration pilot at two cooperatives, Higashi-matsuura in Saga Prefecture and Misumi in Kumamoto Prefecture. Members of these two cooperatives, both of which were in a major orange-producing area, had suffered greatly from the disease. Newly developed orchards occupied most of the area and growers had relatively little experience of melanose control. With the success of the project, it was extended into other major orange-producing areas in Kyushu.

Infection Process of Melanose

The infection process is briefly described, since the main function of the model "MELAN" is to simulate melanose development. The infection cycle is shown in Fig. 2(1186) and Fig. 3(1413). The disease attacks the branches, leaves and fruit of citrus trees. Because it gives rise to diverse symptoms, it has a variety of names e.g. gummosis, melanose and fruit end rot. Of these symptoms, melanose on fruit is economically the most important. Numerous minute black lesions cover the fruit, spoiling its appearance and causing a marked fall in price. The project focused on protecting fruit from infection by dispersed spores, so as to keep the disease index below 16 (See Appendix 1).

The major source of infection is the pycnidia which form on dead, diseased branches. Perithecia also work as an infection source, but are less important because there are fewer of them. Lesions on leaves and fruits do not work as an infection source, because they do not produce any reproductive organs.

The process of infection by spores follows four steps: Dispersal, adherence, germination and invasion into the epidermal tissue of fruit. Rain and a wet fruit skin are indispensable for spores to complete the infection process. Spore dispersal occurs when raindrops strike the spore horn ( Fig. 2(1186)). Spores which adhere to the outer skin of fruit can only germinate, develop a germination tube and invade epidermal tissue while the surface is wet. All spores dry out and die once the skin is dry. The time needed to complete infection becomes shorter as temperatures rises towards the optimum of 23°C. No infection occurs below 17°C or above 35°C.

Outline of the Model &Quot;Melan&Quot;

The "MELAN" model is capable of calculating the probability of disease occurrence, and of recommending preventative action by applying chemicals on any date for which weather parameters are available. Fig. 4(1373) shows the flow of the simulation.

Since the infection process proceeds rapidly when temperatures are favorable, the model uses the hourly infection rate as the unit for simulation, and first calculates it. The daily infection rate is obtained by adding the hourly infection rates over 24 hours. Similarly, adding up daily infection rates gives the accumulated infection rate over longer periods. To complete this calculation, the model has the function of generating hourly data on temperatures, relative humidity and how long the fruit surface is wet. These data are available by routine meteorological observations at 9 A.M.. This saves having to buy sophisticated and expensive instruments. Disease occurrence when chemical pesticides are used is calculated by multiplying the protective effect of the chemicals. The properties of major fungicides, including their protective effect, the extent to which they resist rainfall, and details of their chemical decomposition after application, are included in the model as input for simulation, together with application data.

According to the results of the simulation, the model then issues a message to growers on disease incidence, the residual effect of chemicals and a warning about when the next chemical application will be needed. It is desirable for growers to get the warning at least 10 days before chemical applications are due. Otherwise they often fail to apply chemicals in time, because of the tight schedule of various cultivation practices. However, the function of the model, by its nature, is to simulate infections which have occurred in the past. The problem was solved by inputting of unlikely but possible weather data for the next 10 days, namely, daily rainfall of 10 mm and temperatures which remained the same every day. This made possible short-range forecasts which satisfied growers in terms both of their accuracy and because they gave enough time for growers to arrange chemical applications.

The following data and parameters are needed as input for the simulation ( Fig. 4(1373)).

Dates of Blooming and Initiation of Fruit Coloring

These two dates are required to determine the period in which infection is simulated. Infection of fruit begins when blossoms fall, and ceases when the average temperature falls below 17°C. This fall in temperature coincides with the initiation of fruit coloring of early varieties of Unshu orange, which in Kyushu occurs around the end of September. Once coloration begins, no new lesions appear even at favourable temperatures, since infecting mycelia cannot grow in colored fruits. Medium and late varieties of fruit are still green at that time, and require further chemical applications.

Chemical Application Data

"MELAN" needs the data of each chemical application. It needs the application date, the name of the chemical, and the dose, as well coefficients of the control effect, wash-out and chemical decomposition. The model uses this data to calculate the amount of pesticide remaining and its residual effect.

Daily Weather Data

The necessary data are the temperature and relative humidity at 9 A.M., the maximum and minimum daily temperature, and the rainfall (frequency, amount, and the time rain began and ended). Hourly weather data are generated from these data.

Required Facilities


The program needs computers with a CPU of at least 20 KB, and preferably more than 64 KB. The time required to complete the simulation for one growing season is several minutes, even with a CPU of 128 KB. A compiler for Fortran is desirable.

Meteorological Instruments

A thermo-hygro meter and rain gauge (self-registering type) are generally used. To save costs, the rain gauge can be replaced by a plastic container of 20 liter capacity with a funnel. Computer simulation showed that residual fungicidal action was generally lost by the time accumulated precipitation since the previous spray reached 250-300 mm. Chemical sprays at such times always gave excellent results.

Results and Discussion

The results of the program were examined each year, and are summarized below:

Advantages of the New System Compared to Conventional Control

Application of the model "MELAN" produced a marked improvement in chemical applications. The new control system, which was fully supported by "MELAN", showed excellent performance in giving stable yields of high-quality fruit. Fruit produced under the new system were always graded as good or excellent, with a disease index below 35. The control effect was consistently good, even in unfavorable circumstances e.g. when applied by inexperienced growers, or at periods of heavy rainfall or with high inoculum potential. Conventional systems, on the other hand, often failed to suppress the disease to acceptable levels.

A simulation of the infection sequence in the new and the conventional systems showed why the former gave better control, even though the frequency of chemical applications was the same for both systems (three times per season). The difference in the control effect was the result of the timing of applications. The new system always applied chemicals while the residual effect of previously applied chemicals still remained, so there was constant protection from infection. The conventional system, however, which could not calculate the amount of chemicals remaining, often generated chemical-free periods during which trees were subject to infection. A chemical-free period of only one day is sometimes fatal for melanose control. Simulation proved that infection with 100 or more spores per fruit can occur in a single day, if the weather is highly favorable for infection. The significance of this is readily understood, since fruits with more than 100 lesions are usually rejected by the markets as being unsuitable for table use, irrespective of their flavor.

The number of meteorological observation posts has a marked effect on the accuracy of disease simulation. Hence, when determining the unit area where the simulation is made, the climatic homogeneity should be taken into account. Otherwise, the results of the simulation will much less accurate if the weather differs in various parts of the simulation area. In this project, each local cooperative made a simulation in the area for which it was responsible, an average of several hundred hectares.

The weather within this area differed from place to place, particularly with regard to the amount, time and frequency of rain. The main difference was between hillsides and flat locations. The results of the simulation often did not fit the disease occurrence on hillslopes if the weather observation post was down in the plain, and vice versa. This problem was solved by setting up multiple observation posts, and basing the simulation separately on weather data from each post.

Another problem was how to simplify the demands of the simulation program without losing an acceptable level of accuracy. It is rather difficult in practice to satisfy perfectly the demands of simulation work in the field, from the viewpoint of cost, labor and facilities. We solved this problem to some extent by partial modification of the model.

For example, the constant value, N, was introduced as the amount of inoculum, even though in practice this varies according to the number of dead branches and the season. "MELAN" was originally equipped with a sub-program to calculate the dead branches from each tree in terms of their volume, but this data required much time and labor to collect. In the course of the project, a detailed study was made of the factors closely correlated to the optimum timing of chemical applications and the intervals between them, as indicated by the simulation.

The study showed that a total rainfall of 250 - 300 mm since the previous application could indicate appropriate timing. Encouraged by these results, some growers determined the best time to spray according to their own data of accumulated rainfall, using home-made apparatus (a 20-liter plastic container fitted with a funnel) that was simpler and much cheaper than a self-registered hygrometer.

Any modification which may facilitate control practices should be tried, but its control effect must be examined carefully. Meanwhile, the Misumi local agricultural cooperative rewrote "MELAN" into Basic, so that an ordinary PC could perform the simulation. Some growers began to use the simulation model for themselves. The information provided was used not only by themselves, but by their neighbors.

Use of the simulation model in melanose control has been intermittent for several years, even though a questionnaire showed that more than 90% of growers were satisfied with the results of the project. This is because systematic support, both technical and in terms of facilities, ceased when the project terminated. Local cooperatives and growers then had to conduct the simulation work for themselves and provide timely information. They found this a heavy burden.

The problem could be resolved by establishing a network of growers, local cooperatives and prefectural or regional offices of the national cooperative movement. The regional or prefectural offices would serve as information centers. Using this network, growers or local cooperatives could have direct access to the host computer to input weather or chemical application data, and obtain information for pest control. New simulation programs are also needed which respond quickly to progress in agricultural technology, e.g. new varieties, crop mana-gement methods and chemicals.

Nowadays, environmental deterioration due to excessive use of chemicals has become a global concern. We must replace most of the conventional chemical application programs, which are used widely but designed so as to withstand unusual weather. Instead, systems that are more friendly to the environment should be extended. Pest control systems assisted by simulation models are a very promising alternative. Such systems are primarily designed to minimize chemical inputs, but their flexibility also enables them to cope with severe outbreaks of pests under unusual climatic conditions.

Appendix 1. Grading of Citrus Fruit According to the Number of Melanose Lesions

Citrus fruits are examined at the shipment facilities of local agricultural cooperatives, and given a disease score of 0, 1, 2 or 3, in accordance with the number of lesions, as shown in Fig. 1(1160). A score of 1 is given to fruits with roughly 100 lesions. The disease index is then given to each shipment by the equation:

Here, n 0, n 1, n 2, n 3 are the number of fruits which score 0, 1, 2, and 3, respectively.

Lots with a disease index of 100 are graded as poor, and their price is only a quarter as high as that of fruit with an excellent grade. Those with a disease index of more than 100 are graded as material for juice processing, and sold at a price about one fifth that of fruit for table use.

The present project aims to suppress melanose to below a Disease Index of 16, in response to the keen concern of growers over quality control.

Index of Images

Figure 1 Illustrated Standard for Disease Score of Citrus

Figure 1 Illustrated Standard for Disease Score of Citrus

Figure 2 Infection Cycle of <I>Diaporthe Citri </I>(Faw.) Wolf

 Figure 2 Infection Cycle of Diaporthe Citri (Faw.) Wolf

Figure 3 Infection Cycle of <I>Diaporthe Citri</I>

 Figure 3 Infection Cycle of Diaporthe Citri

Figure 4 Flow of Simulation in the Model &Quot;Melan&Quot;

 Figure 4 Flow of Simulation in the Model &Quot;Melan&Quot;

Formula 1

Formula 1

Download the PDF. of this document(720), 105,002 bytes (103 KB).