AGRICULTURAL RESEARCH ENTER L RAR
Hastings, Flori a AN 1 8 1979
Hastings Research Report PR77- 5 /.~ S.- Dec er 1977
POTATO LATE BLIGHT FORECASTING AND S S~$rida
D. P. Weingartner, Assoc. Plant Pathologis
Accurate forecasting of late blight in northeast Florida (NEF) is intended to
eliminate needless applications of protective fungicides and to provide more
effective disease control by more accurately timing necessary fungicide
A plant disease must fulfill several criteria before a practicable forecast
system can be developed: (i) the disease must cause economic damage; (ii) the
disease must vary in impact from season to season; (iii) effects of weather on
the disease must be known; and (iv) effective control measures must be available.
Late blight in NEF meets each of these criteria. Although late blight is
usually found somewhere in NEF each season, it is severe only approximately
one year in ten. If the disease is left unchecked, however, complete crop losses
can occur. There are several fungicides, such as maneb, which effectively con-
trol late blight. In addition, the effects of weather on spread of late blight
have been studied extensively.
There have been a number of disease forecast systems developed for various
crops. Parameters used in the various systems include rainfall, dew point,
relative humidity, leaf wetness, temperature, and density of spores in the
atmosphere. Spore density sampling is not used in forecasting late blight
because the disease spreads so rapidly that it would be too late to protect
the crop with fungicides by the time spores were detected. A number of late
blight forecast systems have been developed in other potato producing countries
and in various sections of the U.S. Each of these share several features:
(i) they were developed by trial and error fitting of late blight epidemic
data to weather phenomena; (ii) they are "late warning" systems which advise
that conditions favoring late blight have occurred (i.e. they forecast the
first occurence of symptoms rather than the first infection); (iii) all make
the assumption that inoculum is universally present. When evaluating such
systems, therefore, one is not certain whether failure to observe disease is
due to failure of the forecast system or due to the lack of inoculum.
Until recently, the two most successful late blight forecast systems used in
theLLS. were those of -allin and Hyre (Table 1). Neither, however, were
extensively used by growers. Hyre's system is based on temperature averages
and accumulated rainfall. The allin'systemise u tUratifi of relative humidity
12. Q and me n temperatures durnl this period to generate "severity values'."
When 18 severity units are accumulated, a late blight warning is issued.
Recently, Blitecast was developed. Blitecast uses features of both the Wallin
and Hyre systems. The initial late blight warning is issued when either the
Wallin or the Hyre system advise that blight conditions have occurred. Once
a warning is issued, Blitecast schedules sprays according to a matrix which
uses Hyre's rain favorable day (any day on which the 10-day precipitation
total Z 1.20 inches) and Wallins' severity values. In addition Blitecast is
programmed for computerized use thereby providing a means of rapid calculation.
December 6, 1977
SUsing Hyre's system as a basis, Eddins developed a forecast system for NEF which
lB used accumulated rainfall and moving temperature averages. Depend ing upon
the severity of blight, Eddins' system forecasted late blight before the first
occurrence 94- 100% of the time, however, it often forecasted late blight long
before it occurred.
Several factors prompted renewed interest in late blight forecasting at Hastings.
First, it seemed desirable to shorten the time interval between the initial late
blight warning and the first occurence of blight. The rainfall methods are
based on the amount of precipitation rather than the duration of precipitation
periods. One-two inches of rain received within a short period of time would
have a considerably different effect on a late blight epidemic than an identical
amount of rain received via a light mist over several days. It therefore
seemed likely that the relative humidity method of Wallin might provide more
reliable forecasts of initial late blight occurrences than systems based on
rainfall. Preliminary study of existing data suggested that this was indeed
Also, it was apparent that very few fungicide applications may be needed for
blight control during exceptionally dry seasons. The potential impact of other
foliage pathogens such as Alternaria, Sclerotinia, and Botrytis, which might
not be controlled by a fungicide program timed specifically for late blight, was
The primary objectives of late blight forecasting research at Hastings, there-
fore, became: (i) to determine the most effective late blight forecast system
for NEP; (ii) to determine the effect of other diseases on production of potatoes
when no or few fungicide applications are needed to control late blight.
In 1972 cooperative efforts were initiated with R. A. Krause, one of the authors
of Blitecast. Blitecast requires that spray schedules be timed from data
obtained within the crop canopy. Prelimnary observations in NEF suggested
that microclimatic differences among potato fields may not be so important in
NEF. Since 1973 a third objective has been to determine whether late blight
forecasts made from data collected in centrally located standard weather
shelters (i.e. synoptic forecasts) are as accurate as those made from micro-
climatic data accumulated within growers' fields.
Eight weather stations have been operated during the past three seasons. Relative
humidity and temperature were recorded on hygrothermograph charts at each location.
Total rainfall was determined using U. S. Weather Bureau approved rain gauges.
The locations and distance from the ARC at Hastings, and from the St. John's
River (which could moderate temperature extremes) are shown in Table 2. Initial
data aquistion was begun when a faint line of green could be seen along the
length of the potato row (i.e. green row).
Summary of Observations
(i) Providing that accurate green row dates were known, one or more of the
forecast systems forecasted the first occurrence of late blight in NEF each year
since 1970 (Table 3).
(ii) The relative humidity method has been more reliable than the rainfall
(iii) A modification of Blitecast which substitutes Eddin's rain favorable
day for Hyre's has been the most reliable of all the methods studied.
(iv) If the green row dates at individual locations were used as the star-
ting points for analyzing humidity and temperature data, accurate forecasts of
the first occurrence of blight were made from data acquired at the central
(v) Attempts to schedule spray applications from the central location
varied in reliability depending upon the season and the distance from the central
station. During 1977, a very dry season, microclimatic differences among fields
would have drastically affected spray schedules and localized data were
necessary to accurately schedule sprays. In 1976, the differences among locations
were less than the differences among instruments and sprays could have been
(vi) Whenever late blight has occurred, sprays timed according to the
modified Blitecast forecast system and based on microcllmatic weather data have
always preceded detectable spread of the disease (Fig. 1).
(vii) Preliminary evidence from 1976 and 1977 experiments suggest that
early blight and/or other diseases may affect potato yields during some seasons
if very few applications of fungicides are needed for late blight control
(Tables 4, 5, and 6).
During the past three seasons spray advisories based on the modified Blitecast
forecast system have been formulated from weather data acquired at the eight
different locations. The advisories were recorded on a tape connected to a
self-answering telephone (i.e. Code-a-phone). In addition, during the past
season, four day advance weather forecasts and spray advisories were placed on
Late blight spray advisories were updated weekly or as often as needed. Data
from the central weather station were used to determine how often the individual
weather stations had to be monitored. Weather advisories were updated daily by
the Environmental Sciences Service Center (ESSC) at Auburn, Alabama. Data on
the tapes included: late blight spray advisories for eight different locations,
anticipated overnight low temperatures, expected duration of freezing temperatures,
precipitation probabilities, hours of leaf wetness, the number of Wallin severity
units expected during each of the next four nights, and the anticipated direction
and velocity of the wind.
During the season (3/9/77 5/13/77) 337 calls were received. The number of
daily calls varied from 0 34, with 6 10/day being typical. Few to no calls
were received on weekends and the greatest number were usually received on Monday
The telephone advisory service will be provided during the 1977-78 season. The
phone number will be 692-2882. Essentially the same data will be placed on the
tape as during 1976.
Growers who follow the spray advisories on the tape should do so for the entire
season. In addition, the tape should be consulted daily, including weekends.
It is possible for a spray advisory to change from no spray to a five-day
recommendation within 24 hours. Timing of fungicide applications is important
and decisions to spray should be made on a daily basis.
Table 1. A comparison of the late blight forecast systems evaluated
Forecast is issued after 10 con-
secutive days having:
(a) 5-day x temperature (78F
(b) 10-day total rainfall
Forecast is issued after 8 con-
secutive days having:
(a) 7-day x temperature 77F
(b) 10-day rainfall 2 1.02 in.
RELATIVE HUMIDITY-TEMPERATURE METHODS
Range in mean temper-
ature during periods
when R. H.290%
Hours of relative humidity,90%
45-53 15 16-18 19-21 22-24 25+
54-49 12 13-15 16-18 19-21 22+
60-80 9 10-12 12-15 16-18 19+
Z Late blight warning is issued following an accumulation of 18 severity
values. Additional sprays are called for whenever three or more
severity values are accumulated during any seven day period.
Pennsylvania State University Blitecast.
First advisory issued by either Hyre or Wallin system. Sprays scheduled
according to the following matrix.
Total Hyre rain
favorable days Wallin's severity values
in past seven during past seven days
3 3 4 5 6 6
<5 no no 7 7 5
spray spray warning day day day
>4 no 7 5 5 5
spray warning day day day day
Identical to Pennsylvania State Blitecast except: (i) Uses Eddins rather
than Hyre rain favorable day. (ii) Accumulates severity values on days
when temperature : 88F. Penn State Blitecast does not.
Table 2. Years in operation and distance of the various weather
stations from the central weather station located on the ARC Lab-
oratory farm at Hastings,
and from the St. Johns River or Crescent
Sta- Location direction from Distance from Years
tion 4-foot weather river or lake in
No. shelter (miles) (miles) operation
1 Mill Creek 10.3 N 5.3 E 75-77
2 Molasses Junction 6.6 N 2.2 E 75-77
3 Elkton 4.7 NE 4.9 E 74-77
4 Federal Point 2.8 NNW 0.1 E & S 74-77
5 ARC Main Lab 0.1 NNW 2.4 NE 70-77
6 ARC Yelvington 3.4 SE 5.3 E & N 73-77
7 S. Hastings-Boulevard 3.2 S 3.9 E & NE 74-77
8 Bunnell 13.4 S 5.0 E 75-77
Table 3. Comparison of dates on which initial late blight
warning were made by the Eddins, Hyre and Wallin forecast
methods in relation to the first reported occurence of
late blight during the 1970-77 seasons.z
Late blight forecast system First report
Year Hyre Eddins Wallin of late blight
1970 4/7 3/16 4/1 4/13
1971 4/14 4/3 3/14 3/19 (3/30)Y
1972 2/10 2/7 1/27 2/27 (2/13)y
1973 3/18 2/7, 3/16 3/3 2/21x 5/9
1974 3/1 2/27 1/29 3/19w
1975 2/15 2/13 2/5 4/4
1976 None 4/15 2/29 3/24
1977 None 3/14 3/19 3/19
z Warnings were based on weather data collected in a
centrally located standard instrument shelter. Data
accumulation was started with the first general
emergence of potato plants in the area.
y Dates in parentheses are the dates of first observa-
tions of blight in Bunnell area. Other date is for
x 2/21 observation of blight was not confirmed
w Emergence date in this field was not known, however,
it was considerably earlier than in other fields in
Table 4. Percent defoliation and yields of Red La Soda
potatoes as affected by different spray schedules in 1976.1/
Spray Fungicide % Defoliation cwt/acre
schedule Applications 5/10 5/13
Seven-day 7 15 30 96
Blitecast 5 15-20 50 80
No Spray 0 50 95 66
1/ Defoliation was principally due to early blight,
however, some foliage was lost to air pollution
injury, leaf scald due to high temperature and
vascular wilt pathogens. Under normal growing
conditions a dessicant would have been applied
- -------------------------- - -
Table 5. Early blight severity, defoliation ratings, and yields
of Atlantic, Sebago, and Green Mountain potatoes as affected by
fungicide applications. 1977.-/
Early blight rated 1-10 with 1 being no leaves affected and
10 being all leaves 100% covered with lesions. Defoliation
rated 1-10 with 1 being 0 defoliation and 10 being all plants
completely defoliated. Early blight was the only foliage
pathogen identified in the test area. Sclerotium rolfsii
caused extensive tuber decay in one cultivar.
This fungicide controls late blight, but not other potato
Table 6. Yields, early blight severity and degree
of defoliation in 1977 pest management study as
affected by different fungicide application schedules.
Spray Fungicide Blightl/ Defoliationsl/ Yield
Schedule Applications 5/18/77
Seven-day 7 3.8 4.1 270
Blitecast 5 4.0 4.4 271
CG1-821/ 5 6.9 6.8 271
SDefoliation and early blight rated on a scale of
1 10; 1 = no symptoms and 10 = complete defoliation.
2/ This chemical controls late blight, but no other foliage
diseases of potato.
Fig. 1. Spread of late blight in a field of Sebago potatoes during
1974. Numbers in Field and Synoptic rows at base of figure indicate
the respective number of Wallin severity values accumulated in the
field and in the centrally located instrument shelter. Advisory
indicates the spray advisory (No spray, 7-day or 5-day schedule, or
warning). Curves A and B in the figure represent the high (A) and
the low (B) estimates of the number of leaflets in the observational
C O b- 0
sS 3 S o o
passajdx3 (x) sjalea1- paoaJuui0/ X
U I I -