|
A Technical Guide for Wine Producers
|
|
RECENT ARTICLES | WYNBOER HOME
New And Future Early Warning Models For Downy Mildew
1. Pieter Haasbroek
The Metos downy mildew early warning model that is currently being used at Nietvoorbij previously had certain shortcomings in terms of accuracy and reliability, but has now been improved. At the same time the model has also been made more user friendly and a new downy mildew early warning model (DMEW) model has been developed. Future warnings or prediction models can be further improved thanks to rapid technological progress.
Disease warning models for downy mildew, powdery mildew and to a lesser extent Botrytis have been intensively used in most viticultural regions of the Western Cape over the past few years. Downy mildew (Plasmopara viticola) is probably the most devastating disease to afflict vines (Vitis vinifera) and is dependent on the presence of water for its survival (Figures 1 and 2). All vine varieties are susceptible to downy mildew, but the resistance to the pathogen differs considerably (Perold and Phil, 1926). Very serious infections can even cause a total crop loss, especially if it occurs during the flowering phase when the vineyard is at its most susceptible (Magarey, Wachtel and Emmet, 1994). The disease does not occur every year, but due to the epidemic proportions in which it may occur, it is considered to be one of the most important diseases to be controlled. Each year approximately 1500 tons of chemicals, valued at R136 million, are sprayed to control downy mildew.
To effectively control downy mildew in the Western Cape, most producers make use of a preventative spraying programme, while others also use a pre-infection and/or a post-infection spraying programme (Fourie, 2002). Apart from the various chemical spraying programmes followed by producers, the early warning models for downy mildew play a critical role. ARC Infruitec-Nietvoorbij, in collaboration with ARC Institute for Soil, Climate and Water (ARC ISCW), has been making information available to the industry in recent years regarding possible downy mildew infection periods as well as actual outbreaks. The model which calculates the infection times derives from the overseas Metos model (Pessl, 2000), which has been adapted to South African weather stations and conditions. Certain shortcomings of this model have been observed in due course and have been addressed by a new downy mildew early warning model (DMEW). In future the new (DMEW model) downy mildew early warning model will be further expanded and improved. This means that producers in the Western Cape will then be able to access the calculated infection times of downy mildew quickly and easily and act accordingly.
Shortcomings of the adapted Metos model
In some years the adapted Metos model did not entirely live up to expectations. The model firstly makes use of measured leaf wetness values. Up to now the sensor that creates the leaf wetness figures has not been able to realistically simulate the wetness of a vine leaf and consequently from time to time the measured figures are far from the truth. Furthermore there is no predetermined standard for setting up the leaf moistness sensor. The height above the ground, aspect, slope and whether the sensor must be placed in or near the vineyard, are consequently still being determined by each researcher.
Probably one of the biggest shortcoming of the adapted model is that it does not give a sufficiently sensitive indication of infections. For example, outbreaks of downy mildew are reported, but in that period no infection periods are calculated by the model itself. Lastly the current adapted model only provides a qualitative "yes or no" warning, which indicates whether primary and/or secondary infection has occurred or not. Consequently the model does not indicate the infection as a percentage of possibility, which would have been much more useful. Producers therefore have no indication of whether the disease pressure which may come about will be low, medium or high and confidence in the model is therefore still variable. Many producers prefer to use a preventative spraying programme, which requires regular spraying, rather than relying on a model that only indicates infections at certain times, whereafter control should be applied.
New downy mildew early warning model (DMEW model)
The shortcomings of the adapted model were adressed firstly by replacing the leaf wetness values being measured by a mathematical regression, which makes use of relative humidity and temperature. Research that has been done up to now has achieved good results with the new technique. To increase the sensitivity of the new model, the variables of the adapted model were increased and reduced in the DMEW model, in order to obtain more infection periods for the calculation of both the primary and the secondary infection.
The "yes or no" infection warnings of the adapted model have also been replaced by a percentage of possible infection. To this end values were allocated to each weather element used in the DMEW model. The closer the given weather element to the optimum for downy mildew infection conditions, the bigger the weight allocated to the particular element, culminating in a bigger calculated weight in the end. For the weather data of each day that the model ran, the weights of the individual weather elements were added and the sum total expressed as a percentage of downy mildew infection that may occur.
Once the percentage of possible downy mildew infection was known, disease infection was divided into four classes that could indicate the degree of infection. Four classes of infection were compiled for both the primary and the secondary infections and consisted of a zero, low, medium and a high degree of possible infection. The zero class was 0%, the percentage of the low infection ranged from 0% - 35% (green colour), the medium infection from 35% - 75% (yellow colour) and the high infection from 75% - 100% (red colour). The model was developed so that low calculated infections usually go hand in hand with a low disease pressure, medium infections with a moderate disease pressure and if a high infection is calculated, it is as a result of a high disease pressure that reigned for some time in the vineyard. Similar to the "yes or no" line of the adapted model, each of the lines of the DMEW model indicates one hour, in which the conditions were sufficiently favorable for downy mildew infection. For increased thoroughness the hourly wind speed was also made part of the new DMEW model, seeing that wind speed plays a big role in evaporation and drying of foliage. Figures 3 and 4 indicate the differences and similarities between the adapted model and the DMEW model.
Testing of adapted Metos and DMEW model with historical weather data
For both models Nietvoorbij's hourly weather data for 8 - 28 December 2003 was taken and applied to the models. Figure 3 indicates the visual representation of calculated downy mildew infections of the adapted model for the period 8 - 28 December 2003, while Figure 4 indicates the representation of the calculations of the DMEW model for the same period. In the latter two figures it is clear that the adapted model indicated far fewer infections for both primary and secondary infections. The DMEW model also very clearly shows, in contrast to the adapted model (Figure 3), the degree of infections by means of the three colours mentioned above. More calculated leaf wetness values occur in the DMEW model (Figure 4), and give rise to the many secondary infections, compared to the few of the adapted model.
Testing of adapted and DMEW models with actual disease occurrence data
A block of SA Riesling at Nietvoorbij in Stellenbosch was used to determine actual disease occurrence. The vineyard block was visited on a weekly basis from October 2003 to January 2004. During these visits the vineyard was monitored for the presence of downy mildew. The disease occurrence was determined by evaluating 100 leaves and visually estimating the percentage of infected leaf surface. The percentage of the disease occurrence obtained in this way, was then divided according to the following scale: 0% - 35%, 35% - 75% and 75% - 100%. This represents a low, medium and high degree of disease occurrence respectively.
Four different treatments against downy mildew were applied in the trial: 1) Untreated control - no spraying against downy mildew, 2) Two-weekly preventative spraying, 3) Spraying according to the adapted model and 4) Spraying according to the DMEW model. Disease warnings of both models were calculated on a daily basis and used during the visits to determine whether spraying against downy mildew should take place or not. After each spraying a fourteen day period was considered to be a protected period. This means that after each spraying no applications were given for fourteen days, even if one or more warnings were calculated in this period by the two models. Applications were only given if rainfall of >10mm occurred in the fourteen day period, given that rainfall of >10mm was reckoned to be sufficient to rinse off the fungicide.
Table 1 indicates the four treatments, the number of applications and the percentage occurrence of downy mildew. The control sites with no applications had the most downy mildew, namely 69%. The latter percentage is considered to be a medium occurrence of downy mildew. For the three remaining treatments the percentage occurrence of downy mildew was considerably less and ranged from 10% to 20%. The difference came about in the number of applications given to each. By February the preventative treatment, in which altogether seven applications were given, showed the second least amount of disease occurrence (15%). According to the DMEW model an additional infection warning was noted and consequently an additional application was required (Table 1), compared to the adapted model for the same period. Despite the one additional application for the DMEW model, the disease occurrence by 1 February 2004 was eight per cent less for the DMEW model (11%) as opposed to the treatment of the adapted model (19%). The lower occurrence of downy mildew as a result of the warnings by the DMEW model might justify the one extra application that had to be made to treatment number four.
Table 1. Percentage occurrence of downy mildew by 1 February 2004 at Nietvoorbij after spraying according to the various treatments.
Future models
Despite the improvements to the DMEW model, there are still other problems that have to be addressed by future research. The biggest of these is surely the fact that the DMEW model, just like the adapted model, is still based on point weather station information. The early warnings of favourable infection conditions for downy mildew are still being done according to weather station and the error of the models' calculations becomes increasingly bigger the further away they are from the particular weather station. Future model/s should therefore be able to give early warnings for an entire area or region, together with increased accuracy for any section within an area that has few or no weather stations.
Further research that is currently being undertaken investigates a Geographic Information System (GIS) as a solution to address the above problem. The DMEW model can, for example, be programmed with GIS software. The parameters of the model can remain unchanged, while simulated weather data will be used for areas between existing weather stations with no weather data. Results of the DMEW model can then be represented in colour charts of favourable infection conditions for a relatively small or bigger area. Colour charts can be compiled by using all the measured weather data from weather stations in an area, as well as the simulated weater data between the stations, to run the model. In future the producer will then be able to obtain an early warning risk chart which indicates infections for a specific area for the previous few days.
True prediciton models
At the moment downy mildew early warning disease models in the Western Cape only make use of historic and contemporary weather data to give warnings of downy mildew infections. Contemporary weather data is weather data from the aforegoing hours and any warnings/predictions based on this type of data cannot in fact be considered a true prediction. Furthermore the present method of an early warning service using the warnings of existing models has limitations in the sense that sometimes they leave the producer too little time for quick and effective control of downy mildew, if producers rely simply on the infections of the warning model to combat the disease.
The South African Weather Bureau Service (SAWS) can play a significant role in this regard if its forecasts of weather data may be used. The SAWS has three and five day forecasts of inter alia rainfall and temperatures which may be used in a GIS model. The only real disadvantage of this weather data is that the figures are daily and unfortunately the amount of rainfall per region cannot be predicted. A GIS model will have to be adapted so as to make use of predicted as well as measured historical weather data. The biggest advantage of this is surely that potential disease infections may then be predicted up to two weeks in advance, compared to the adapted Metos and DMEW models. This GIS model will therefore give the producer maximum opportunity to apply effective disease control in good time, as soon as favourable downy mildew infection conditions are predicted.
Summary
Downy mildew is an extremely destructive and consequently a very important disease all over the world. The disease rears its head annually, but fortunately the occurrence thereof is not always extremely severe and it may be fairly well controlled by means of good management, so that large scale and serious outbreaks seldom occur. The disease will never be eradicated completely. Improved control measures, combined with aids such as mathematical models, are essential and should play an increasingly significant role in future.
The DMEW model is the first step in positioning existing chemical applications in such a way as to obtain better and more effective control. In the light of other in depth research that has already been undertaken (report available in September), it looks as if the new model is more reliable, accurate and user friendly. This enables the producer to form a better seasonal picture of the actual downy mildew disease conditions that are present in the vineyards.
The three classes of possible infection (low, medium and high) also assist the producer in taking more sensible decisions with regard to the control of downy mildew, versus a simple "yes or no" warning given by the adapted model. Reliable mathematical models using automatic weather stations (hourly weather data) to calculate downy mildew infection periods, are undoubtedly one of the best instruments to be used in the optimal management of spraying against downy mildew.
Acknowledgement
Our thanks to the Western Cape Agriculture Department for their financial support.
For more information about the new downy mildew model or any queries, contact Pieter Haasbroek at (021) 887-4690 and Abraham Vermeulen at (021) 809-3161.
References
Fourie, P., 2002. Previously senior researcher in the Disease Management Division, ARC Infruitec-Nietvoorbij. Personal communication about downy mildew in the Western Cape.
Magarey, P.A., Wachtel, M.F. and Emmett, R.W., 1994. Downy mildew. Grape Production series Number 1: Diseases and pests. P.R. Nicholas, P.A. Magarey and M.F. Wachtel (eds.). Winetitles, Adelaide. 106, 5 - 11.
Perold, A.I. en Phil, B.A., 1926. Handboek oor Wynbou. Pro Ecclesia Drukkery Bpk. Stellenbosch, pp 645.
Pessl, G., 2000. Instruments and software for agricultural climate monitoring and electronic sprayer calibration. Internet site: http://www.metos.at/
|
|
Wynboer is incorporated in WineLand, magazine of the SA wine producers.
Visit our sister sites:
|
| UP |
COPYRIGHT (C) 2000 WineLand |