Figure 3. Disease assessment key used in the Danish monitoring network for late blight.
Web-blight
Jens Grønbech Hansen, Manfred Röhrig, Poul Lassen and Iver Thysen,
Danish Institute of Agricultural Sciences (DIAS)
Dept. of Agricultural Systems
Research Group for Informatics and Agrometeorology
Research Centre Foulum, Denmark
Foulum May 10, 2000
1. Introduction
In 1996 the Danish Institute of Agricultural Sciences (DIAS) developed a prototype of an Internet based information and decision support system for agriculture. The system was called Pl@nteInfo. Currently this system is in commercial operation in close collaboration between DIAS and the Danish Agricultural Advisory Centre (DAAC).
In Pl@nteInfo comprehensive information and decision support is available for potato late blight such as: late blight monitoring and disease forecasting, searchable potato variety database, information about fungicides and control strategies, NegFry homepage, animated weather radar pictures, local weather prognosis etc.
The existing Internet based monitoring system was made very general and applicable for other countries than Denmark, and the system was in use by Denmark, Sweden, Norway, Finland and Lithuania in 1999. Several countries have expressed their interest in utilising the applications for late blight monitoring, and more collaborative components in preparation. Therefore, it was decided to develop an International web site to manage and present the results of existing and future Internet based decision support components for potato late blight. The domain name of this site will be www.web-blight.net.
Objectives for Web Blight:
In a first version of Web Blight, applications will be developed for International collaboration in the areas of late blight monitoring, evaluation of crop resistance in variety observation trials and evaluation of results from field trials for validation of late blight decision support systems.
2. Web Blight
2.1 Introduction
The general approach for management of Web-blight is that the system should be self sustainable in a way that each country will be responsible for updating of national relevant biological data and user data in the system. Based on user names and passwords information and results in the system can be restricted to closed user groups or they can be public determined by a national country administrator. Web Blight will be developed based on experiences and existing tools from the Danish Pl@nteInfo system.
2.2 Methods
Web Blight components
A first draft of the system will contain the following components:
2.3 Organisation and administration of Web Blight
Country administrators and country reporters
A country administrator (CA) will be responsible for management of national basic data and for appointment of national country reporters (CR) in the three networks (Pi-DSSTrial-Net, Pi-Monitoring-Net and Pi-ObsTrial-Net) (Figure 1). It will be possible to participate in each network separately. The country reporters enter data into the system via PC-program interfaces called Pi-Monitoring, Pi-ObsTrial and Pi-DSSTrial respectively. Data are stored on the local PC and automatically transferred via FTP to the Web Blight web server in Denmark. In the Web Blight system data are quality controlled, data handled with statistical methods using SAS and presented on maps, figures and tables on the Internet. National results on the Internet can be public or they can be restricted, administrated by the country administrator.
The Country administrator (CA) will be responsible for:
To update trial identifications, regions and variety information and user identification, the country administrator use a PC program called PI-CountryAdministrator. For description of the national web-blight networks on the Web-Blight web site the CA will use the News service facility in Web-Blight for upload of HTML documents. For drawing the EU-map with late blight monitoring data each country has to be divided into regions.
User identification
Input from Pi-programs
Based on Country code and initials (three digit) of the CR, the Pi-CountryAdministrator program can create a binary file which has to be send to relevant country reporters(CR). The CR download this file into the directory containing other Pi-programs. When sending data to the Web-Blight server the Pi-Programs will integrate this CR identification code into the file name of the data file. Before updating the Web Blight databases, data files are checked for identification. If the data files are not recognised, the data file will automatically be deleted from the server.
Access to output from Web-Blight
Each user of Web-Blight needs to have a user number, a password and an ID. Via a registration web page in Web-Blight a new user automatically receive his user number, password and an ID. The user number is used by the CA to appoint which national outputs in Web Blight should be available for national users (closed information based on user number or public). This will be managed by the CA via a web page in Web Blight and not via the Pi-CountryAdministrator program.
2.4 Mailing lists
Different mailing lists will be established on the Web Blight server e.g. Mails from the system administrator to all country administrators or internal between country administrators. Mails from the country administrator to country reporters on national level or mails from the system administrator to all country reporters.
2.5 Discussion group
The system is intended to be self sustainable in a way that all problems and ideas are discussed via a closed Internet based discussion group on the Internet. All country administrators and country reporters have access to the discussion group. Discussions on national level can be done via the mailing lists mentioned above
3. Monitoring of potato late blight
3.1 Introduction
During the last 10 years a new Phytophthora infestans population has been established in Europe. There are indications that the sexual recombination by P. infestans has caused a change in the behaviour of the fungus e.g. more aggressive genotypes, stem blight and (early?) infections from oospores in the soil. The change in epidemiological behaviour of P. infestans is a threat for the control of late blight in general, but also for the use of existing warning and forecasting systems, especially if sub-models are based on data collected before the new population was established.
Therefore, development of an "Internet based monitoring system for potato late blight" is prepared with the following objectives:
The system will be based on experiences from a system developed earlier (Hansen et al, 1999).
3.2 Methods
Disease survey network
Each country is responsible for organising a late blight disease survey network and describe what kind of fields that are surveyed, by whom and how intensive, quality control of assessed data etc. One or more country reporters are responsible for data collection, quality control, entrance and transmission of data to the Web Blight server. Organising a late blight disease survey network can be done in many ways, and there are advantages and constraints in each way to do it. This is a short description of the Danish approach:
The survey network is organised by the Danish Agricultural Advisory Service in Denmark. About 50 advisors participate in a formal survey network, and they are paid 3-500 DKr for doing the work. We have discussed to make observations in fixed fields, field trials etc. but this cause the first problem: "When the monitoring map contain no recordings of late blight we cannot be sure that late blight is not established in farmers fields somewhere around." Therefore all conventional and organic fields, field trials and experimental trials are the target for monitoring and the scouts are all the farmers and the appointed advisors in the network. For Danish conditions home gardens and potatoes under plastic were excluded from the network. This was done because late blight in home gardens is not recognised as a threat for conventional fields and potatoes under plastic are grown in special areas in Denmark. Secondly attacks in potatoes under plastic are very early when conventional grown potatoes are in a growth stage when they are at the highest level of age dependent crop resistance (BBCH 10-30). In 1998 and 1999 only one early attack was found before BBCH 30. We do not know if this situation will change in the future caused by a build-up of oospores in the soil!!
When late blight is found by the farmer (or somebody), he will contact his local advisor. If the advisor recognise symptoms as late blight (or probably late blight) he will send a plant sample to the Danish advisory service in Århus including background information (This is what he is paid for). The sample will be verified by a specialist, if necessary under microscope, and the result will be entered into the system at this point.
The advantage of this system is that all information on the Internet is certainly late blight. We know that even experienced potato advisors can be wrong in the interpretation of "late blight" symptoms. In 1998, four out of fifty samples received by DAAC was not late blight. In earlier years, rumours about early attacks could get many farmers in a region on the tractor and start their fungicide control strategy. Later maybe it showed up not to be late blight or it was found in one home garden. Farmers in Denmark are very satisfied about the verification method and more and more farmers support the survey system by looking carefully in own fields and by making contact to the local advisors when symptoms are found.
The disease forecast is used to identify the time when scouting should be intensified. This can differ up to three weeks (mid June - early July) for Danish conditions.
The constraints in the "central verification method" is, that the information will be one day old (time from observation- postal mail-verification). If the observation is done on Fridays the observation can be 2-3 days old.
In the Danish Pl@nteInfo system the map with forecasting data and the map with monitoring data are put side by side on one page (Figure 2)
Figure 2. Maps in PlanteInfo on June 18, 1999. The disease forecast is shown to the left and observations of late blight to the right. A click with the mouse on a dot will give some background information about the specific recording. A general warning for late blight attacks was issued on June 14 based on the forecast, but at the time when no attacks was found yet.
Disease assessments
When late blight is found, the recording is entered into the system via a PC-program called Pi-Monitoring. Data to enter are divided into two parts. Part A is information that are necessary to identify the reporter, where and what to put on the map and in the table produced by the system e.g. geographical position, variety name, field type and how much disease that was found. Part B contain information that are important for evaluation of the data (owner of the field, phone or fax, crop emergence, last year potatoes were grown on the field, seed quality etc.). It should be possible to operate the system only based on data from part A.
Disease assessment key
A special disease assessment key was developed, that focus on very early attacks of late blight (Figure 3). Use of this key facilitates the interpretation of the recording in regard to the time when the attack probably was established as a primary attack.
Figure 3. Disease assessment key used in the Danish monitoring network for late blight.
Maps
It will not be possible to identify single field observations on a map covering whole Northern Europe. Therefore we need a two level approach
EU Map
The countries are divided into regions e.g. Counties in Denmark and Bundesländer in Germany. When late blight is found in the region the first recording will produce a small dot representing the region. When more recordings of late blight is found, the dot will increase in size (scale 1-5: 1-2 recordings, 3-5 recordings, 6-10 recordings, 10-15 recordings and 15-20 recordings). This map will give an overview when late blight is found in different regions of EU and in how many fields.
Country maps
When selecting a single country, the system will zoom to country level and show the exact geographical position of each recording. A click with the mouse on the dot on the map will give some background information about the recording: variety name, crop emergence, field type and disease level.
Tables
All recordings can be evaluated via a table including more background information including a possibility to sort the information ascending or descending according to each variable shown in the table (Figure 4). There will be produced a table for each country, but also a table including recordings in all countries. Based on all data, evaluation tables and graphics will describe where the first recordings were done, in which varieties, field types etc. This project will evaluate if the data can be used to evaluate the reason for very early attacks of late blight.
Figure 4. Table including late blight observations in Denmark, 1999. Data were sorted by "Date of recording"
Organisation and data flow
See section 2.3 and figure 1.
4. Evaluation of disease progress curves from untreated observation trials (and other trials) for evaluation and quantification of crop resistance to late blight (Pi-ObsTrial-Net)
4.1 Introduction
In more and more EU countries goals are decided to minimise the use of fungicides to control potato late blight caused by Phytophthora infestans (Mont.) de bary. From field trials it is well known that moderate resistant varieties can be controlled with less amount of fungicide than susceptible varieties (delay of first spray, lower dosage and/or longer spraying intervals) but this knowledge is not widely used in conventional potato production. This is to some extent due to the fact that field resistance components are not very well described and implemented in existing decision support systems. Therefor farmers are not willing to take the risk of reducing the amount of fungicide input in moderate resistant varieties compared to the input levels in susceptible varieties. The use of more resistant varieties is more or less considered as an "extra security" for a successful control of late blight. Therefor the field resistance in potato varieties should be quantified in more detail, components of field resistance should be implemented in decision support systems and the potential of more resistant varieties in reducing the input level of fungicides should be documented in field and demonstration trials.
In the Pi-ObsTrial-Network several countries in the Nordic and Baltic countries will do observation trials based on harmonised methods. Intensive disease assessments will be done in untreated local cultivars including standard cultivars in different groups of maturity and type of potatoes to facilitate the comparison of trial results from different countries. Weather data will be included to evaluate weather influence on late blight development in different years and in different regions in Europe. If possible one or more weather based parameter will be used to calibrate between years and regions.
The major goal is to collect quantitative knowledge of crop resistance components under field conditions for implementation in decision support systems. Secondly the obtained primary data and interpretations will be presented on the Internet during the season. In this way the data can be used directly as a part of existing warning and decision support systems.
4.2 Methods
Tasks
In 2000 data from Denmark, the Baltic countries and Poland are used for the development of a prototype of the system.
Biological methods
The trials are conducted based on modified EPPO guidelines for fungicide trials. The late blight key will be the one from Anon Trans Brit Mycological society, 1947.
Potato varieties will be classified into groups based on Maturity class (early, medium, and late), Resistance (susceptible, moderate susceptible and moderate resistant) and type of potato (seed, ware and starch). In the first place this will be done on national level. Harmonisation of methods will be needed.
In the evaluation of the results on the Internet the user can select how the cultivars should be divided into groups (maturity, resistance or type) and what cultivars should be the reference in each group. When this information is submitted, a local SAS application then make all the graphics and data calculations e.g. Delay of first symptoms (relative to the reference cultivar), Disease rating when reference disease level is 90%, relative AUDPC, apparent infection rate( r ) etc. Based on data from all the countries several methods for describing field resistance components will be evaluated (Figure 5). As data will be available on the Internet shortly after disease assessment in the field, the data can be used directly as a part of existing warning and decision support systems.
Figure 5. Examples of theoretical disease progress curves related to crop resistance.
Organisation and data flow
See section 2.3 and figure 1
Output Pi-ObsTrial in Web-Blight
Settings
Divide varieties in groups according to: (radio buttons)
1. Maturity class (Early, medium late)Select reference variety name:(drop down box)
Output
|
Delay of first symptoms (days) é sort ê Graphics |
Disease when reference disease is 90 % é sort ê Graphics |
Final disease rating [%] é sort ê Graphics |
Apparent infection rate ( r ) é sort ê Graphics |
Relative area under the disease progress curve (AUDPC) é sort ê Graphics |
|
|
Bintje (reference) Sava Folva |
0 5 8 |
90% 55% 65% |
97 95 95 |
0.28 0.19 0.22 |
0.62 0.45 0.52 |
It will be possible to sort the table ascending or descending based on each variable in the table. A click with the mouse on Graphics will show the data on a graph.
5. Evaluation and presentation of results from validation trials of late blight decision support systems in Europe (Pi-DSSTrial-Net)
5.1 Introduction
In 1999 six different late blight decision support systems (DSS) were validated in six different EU countries in the frame of the concerted action EU.NET.ICP. For the DSS validation a common field trial guideline was agreed and used based on the EPPO guidelines for test of fungicides in experimental trials. One of the major constraints in this project was a lack of continuos backup and help from the "model builders". Due to this fact, some mistakes were made in the interpretation and proper use of the decision support systems. The validation project will continue during the coming years and using Web Blight may be a help for management, quality control and interpretation of the results in this project.
A prototype of Pi-DSSTrial will be used in year 2000 in a Danish/Baltic/Polish project about late blight decision support and control strategies.
5.2 Methods
Tasks
Biological and field trial methods
The trials are conducted based on modified EPPO guidelines for fungicide trials. The late blight key will be the one from Anon Trans Brit Mycological society, 1947.
In the evaluation of the results on the Internet components based on the disease progress curves will be used to compare results of different control strategies. The same epidemiological parameters as used in the interpretations of disease progress curves in ObsTrial-Net will be used (delay of first symptoms (relative to first symptoms in untreated plots), Disease rating when a reference control strategy reach a certain level, relative AUDPC, apparent infection rate( r ) etc. Based on data from all the countries several methods for evaluation of the results will be evaluated.
As data will be available on the Internet shortly after disease assessment in the field, the data can be used directly as a part of existing warning and decision support systems.
Organisation and data flow
See section 2.3 and figure 1
Output DSSTrial in Web-Blight
Output
|
Number of Fungicide applications |
Treatment Frequency index |
Yield |
Tuber blight |
Delay of first symptoms (days) |
Final disease rating [%] |
Apparent infection rate (r) |
Relative area under the disease progress curve (AUDPC) |
|
|
Untreated (reference) NegFry PlantPlus |
It will be possible to sort the table ascending or descending based on each variable in the table. A click with the mouse on Graphics will show the data on a graph.