DISEASE FORECASTING - The Epidemiology of Plant Diseases - page 248

Blitecast (Krause et al ., 1975). The scheme takes account of cultivar susceptibility,
emergence date and irrigation to advise on the date of the first fungicide application
and the timing of subsequent sprays. It is distributed as an easy-to-use computer
programme which growers can customise for their location and the level of infection
they are willing to accept. In order to work effectively, the programme requires
weather readings at three-hour intervals, or better, and preferably from sensors in or
near the crop. In validation trials, the number of sprays was reduced by 50% in
comparison with routine treatment. The reduction was based mainly on delays to the
start of the spray programme and the longer interval between spraying as
recommended by the system.
9.6.10 Developments
The introduction to Europe of new populations of the late blight fungus from Mexico
in the late 1970s and particularly the A2 mating types in the early 1990s (Shattock,
2002) has cast doubt on the validity of current forecasting schemes. European blight
forecasting systems were developed before oospore progeny were detected in field
situations (Flier and Turkensteen, 1999). This has prompted moves to include more
widespread monitoring of outbreaks and incorporating them within schemes and
presenting them on maps using internet applications (Hansen et al ., 2000).
Some of the above schemes have been converted to computer programmes and are
available for personal computers which can be fed data from 'in-crop' weather stations
that provide read-outs giving an indication of when weather criteria have been satisfied
and when sprays should be applied.
In England and Wales, blight forecasting has been revisited to determine whether,
with the advent of inexpensive in-crop meteorological data capture equipment, it is
possible to improve on blight forecasting from the current Smith periods. Detailed
meteorological data, together with the record of first blight occurrence were obtained
from a series of experiments conducted at five sites covering a range of infection
pressures over four years (Taylor et al ., 2003) and tested against a number of schemes
(Blitecast, NEGFRY, Schrödter and Ullrich, Smith and Sparks). The initial results
were surprising, in that Smith was the only scheme to consistently predict the outbreak
of blight in advance of it appearing in the crop. However, results from the 1997
season, a major blight year for the UK, indicated that the forecasting systems worked,
in that they all triggered a spray application at least 14 days before the occurrence of
blight. There would seem to be justification for confidence in forecasting in high risk
years but less so where there was low risk.
The multiplicity of schemes for forecasting blight, even within single countries
and the continued development of further schemes, indicates that forecasting for this
disease is complex. It is highly unlikely that a standard scheme will be applicable to
all countries, or even all regions within a country. The increasing sophistication of
the data capture equipment may be providing the imperative to devise even more
precise schemes. However, Royle and Shaw (1988) suggested that with an annual
and substantial build-up of inoculum with a disease such a potato blight, then a
“relatively complex tactical spray programme based on forecasting is unlikely to be