Agenda for Project Overview Meeting

ACO301 - Defra Final Project Report – March 2008 (ANNEX 8)
ANNEX 8
VULNERABILITY OF UK AGRICULTURE TO EXTREME EVENTS
Case studies
Index
Page
Case study 1
Winter cauliflower
2
Case study 2
Vining Peas
3
Case study 3
Lifting potatoes
5
Case study 4
Bone-dry soil
7
Case study 5
Leafy Salad production
9
Case study 6
Peach-potato aphid Myzus persicae
13
Case study 7
Cutworm Agrotis segetum
15
Case study 8
Diamondback Moth Plutella xylostella
16
Case study 9
Late Blight Phythophthora infestans
17
Page 1 of 17
ACO301 - Defra Final Project Report – March 2008 (ANNEX 8)
Case study 1 Winter Cauliflower
The winter cauliflower industry in the UK faces difficulties in predicting when crops will reach maturity. The
crop is very sensitive to temperature and the unpredictable weather means that production patterns so
carefully planned are rarely achieved. Models of the phases of juvenility, curd induction and curd growth,
driven by temperature have been developed for winter cauliflower types and were linked to enable
prediction of the time of curd maturity. The models are described fully by Wurr et al. (2004).
For this study, a model for the Roscoff X Walcheren cross CV Renoir was used. In order to investigate
effects of changes in temperature on production pattern the model was run from seven sequential starting
dates defining the end of the juvenile phase of growth and thus the beginning of the induction phase,
beginning on 1 July and ending on 1 October. The induction phase is described by a rate function defined
as a gamma curve with an optimum temperature of 9.0 oC. Once a curd is initiated, it enters the curd growth
phase, which is described by a quadratic relationship between the natural logarithm of curd diameter and
accumulated ambient day-degrees >0oC from curd initiation. This sequence of two models was run until
curds with a target diameter of 120 mm, a typical supermarket specification, were produced. The model
sequence was run for 150 years of data for all timeslices.
As predictions were made further into the future there was a trend for the differences in maturity time to
diminish, so that by the 2050s the continuity of production was predicted to be lost as the example Figure
1.1 shows for a Renoir model run with synthetic weather for Camborne in Cornwall. The predicted dates
of maturity for each of the 150 years of synthetic weather data have been summarised in the box plots
which also show the 5 and 95% outliers among the predicted dates.
Predicted day of maturity
Case study Figure 1.1 Predicted day of maturity using models for CV Renoir and weather data for
Camborne in Cornwall. The predicted dates of maturity for each of the 150 years of synthetic weather data
have been summarised in the box plots. The boxes represent the 25th and 75th percentiles and the line
within the box represents the median. The whiskers represent the 10th and 90th percentiles and the dots
show the maximum and minimum values.
1 May
1 Apr
1 Mar
1 Feb
1 Jan
Base
1 Jul
1 Aug
1 Sep
1 Oct
2020s
1 Jul
1 Aug
1 Sep
1 Oct
2050s
1 Jul
1 Aug
1 Sep
1 Oct
Model starts at the end of the juvenile phase
Wurr, D.C.E., Fellows, J.R. & Fuller, M.P. (2004). Simulated effects of climate change on the production
pattern of winter cauliflower. Scientia Horticulturae 101, 359-372.
Page 2 of 17
ACO301 - Defra Final Project Report – March 2008 (ANNEX 8)
Case study 2 Vining Peas
In the UK vining peas are harvested at a specified stage of maturity which is determined by measurement
of samples using a tenderometer. An ideal crop for freezing will measure up to 100 tenderometer readings
(TR) while a higher reading of 130 will indicate suitability for canning. The model described by Holloway et
al. (1995) was used to predict TR of 100 and 130 so defining a predicted harvesting window for a pea
crop. This model estimates daily TR from day-degrees above a base temperature of 4.5oC, where TR
increases by 1 unit for each 1.49 day-degrees above the base. The model was run for 150 years of
synthetic data for all timeslices. As suggested by Holloway et al. (1995) the model was run from a very
early sowing date of 14 February and a late sowing date of 8 May to span the range of commercial sowing
dates in the UK. The model was run with synthetic data for Kirton, Leuchars and Wye to represent
traditional areas of production. The predicted harvesting windows were summarised using box plots
shown in Figure 2.1. For all sites the time of average harvesting window was reduced by the 2080s and
the variation in the duration of the harvest window over the 150 years of data was also reduced. This
predicted reduction may have implications for the logistics of harvesting and processing of the UK crop in
the future. Figure 2.2 shows that in future the predicted time of the first harvest at 100 TR may also be
earlier and so may possibly extend the growing season for vining peas.
Harvesting window in days
from 100 to 130 tenderometer units
Kirton Lincolnshire
20
early sowing 14 February
20
15
15
10
10
5
5
0
late sowing 8 May
Case study
Figure 2.1
0
base
2020s
2050s
base
Timeslice
2020s
2050s
Timeslice
Harvesting window in days
from 100 to 130 tenderometer units
Leuchars Fife
early sowing 14 February
20
15
15
10
10
5
5
0
late sowing 8 May
20
0
base
2020s
2050s
base
Timeslice
2020s
2050s
Timeslice
Harvesting window in days
from 100 to 130 tenderometer units
Wye Kent
20
early sowing 14 February
20
15
15
10
10
5
5
0
late sowing 8 May
0
base
2020s
2050s
base
Timeslice
2020s
Timeslice
Page 3 of 17
2050s
Predicted duration
of the harvest
window. The
predictions for each
of the 150 years of
synthetic weather
data have been
summarised in the
box plots. The
boxes represent
the 25th and 75th
percentiles and the
line within the box
represents the
median. The
whiskers represent
the 10th and 90th
percentiles and the
dots show the
maximum and
minimum values.
ACO301 - Defra Final Project Report – March 2008 (ANNEX 8)
Case study Figure 2.2 Predicted day of first harvest. The predictions for each of the 150 years of
synthetic weather data have been summarised in the box plots. The boxes represent the 25th and 75th
percentiles and the line within the box represents the median. The whiskers represent the 10 th and 90th
percentiles and the dots show the maximum and minimum values.
Kirton Lincolnshire
165
early sowing 14 February
Predicted harvest day (1 = 1 Jan)
at 100 tenderometer units
Predicted harvest day (1 = 1 Jan)
at 100 tenderometer units
140
130
120
110
100
90
80
70
late sowing 8 May
160
155
150
145
140
135
base
2020s
2050s
base
Timeslice
2020s
2050s
Timeslice
Leuchars Fife
165
early sowing 14 February
130
120
110
100
90
Predicted harvest day (1 = 1 Jan)
at 100 tenderometer units
Predicted harvest day (1 = 1 Jan)
at 100 tenderometer units
140
80
late sowing 8 May
160
155
150
145
140
base
2020s
2050s
base
Timeslice
2020s
2050s
Timeslice
Wye Kent
130
165
Predicted harvest day (1 = 1 Jan)
at 100 tenderometer units
Predicted harvest day (1 = 1 Jan)
at 100 tenderometer units
140
early sowing 14 February
120
110
100
90
80
70
60
late sowing 8 May
160
155
150
145
140
135
base
2020s
2050s
Timeslice
base
2020s
2050s
Timeslice
Holloway, L., Ilbery, B. & Gray, D. (1995). Modelling the impact on navy beans and vining peas of
temperature changes predicted from global warming. European Journal of Agronomy 4(3), 281-287.
Page 4 of 17
ACO301 - Defra Final Project Report – March 2008 (ANNEX 8)
Case study 3 Lifting potatoes
During 2000, heavy rainfall late season caused widespread problems for potato lifting (British Potato
Council, pers. comm.). By Christmas 2000, 20% of the crop (25000 ha) were still in the ground. In a year
where lifting goes to plan, by mid-September 70% of the UK crop is defoliated and ready to lift and go into
storage. In order to identify threshold values of cumulative rainfall, data for Kirton in Lincolnshire for 2000,
the problem season and 2003 a season where lifting went to plan were compared. Figure 3.1 shows these
data.
Case study Figure 3.1 Daily rainfall and cumulative percentage of the UK potato crop lifted in 2000 and in
2003.
25
Daily rainfall (mm)
20
rain in 2000
National cumulative lifting % /10
15
10
5
0
1 Aug
1 Sep
1 Oct
1 Nov
1 Dec
25
rain in 2003
National cumulative lifting % /10
Daily rainfall (mm)
20
15
10
5
0
1 Aug
1 Sep
1 Oct
1 Nov
1 Dec
Examination of the rainfall data suggested that accumulations of 10mm or greater rainfall would be likely to
cause problems with lifting as shown by the dotted red line in Figure 3.2.
Synthetic rainfall data for Boulmer, Brooms Barn, Camborne, Cawood, Kirton, Leuchars, Rothamsted and
Shawbury were used to investigate how conditions for lifting might change. For all eight locations, the
incidence of runs of seven consecutive days with accumulated rainfall reaching or exceeding 10mm
between 1 August and 31 October declines for the 2020s and the 2050s. This suggests that conditions for
lifting the UK potato crop may improve. Figure 3.3 summarises the data for Rothamsted in Hertfordshire
and Boulmer in Northumberland.
Page 5 of 17
ACO301 - Defra Final Project Report – March 2008 (ANNEX 8)
Accumulated weekly rainfall (mm)
Case study Figure 3.2 Weekly accumulated rainfall and percentage of the UK potato crop lifted in 2000
40
National cumulative lifting % / 10
accumulated weekly rainfall
30
20
10
0
1 Aug
1 Sep
1 Oct
1 Nov
1 Dec
Case study Figure 3.3 Incidence of runs of 7 days with an accumulated rainfall total of 10 mm or greater
between 1 August and 31 October under baseline, 2020HI and 2050HI scenarios. (Note: 10 consecutive
days count as 4 runs). The boxes represent the 25th and 75th percentiles and the line within the box
represents the median. The whiskers represent the 10th and 90th percentiles and the dots show the
maximum and minimum values.
Incidence
Rothamsted Hertfordshire
80
80
60
60
40
40
20
20
0
Boulmer Northumberland
0
base
2020s
2050s
base
Timeslice
Page 6 of 17
2020s
2050s
ACO301 - Defra Final Project Report – March 2008 (ANNEX 8)
Case study 4 Bone-dry soil
During Summer 2006, there were periods when it was impossible to transplant horticultural crops because
the soil was bone-dry and it was very hot. Even where irrigation was available the demand for water from
the transplants made it impossible to proceed. Figure 4.1 shows the weather observed during June, July
and August in 2006 at Brooms’s Barn.
minimum air
maximum air
smd sandy
smd sandy loam
smd clay loan
smd silt loam
total rainfall
40
30
20
o
Temperature ( C) OR smd (mm) OR daily rainfall (mm)
Case study Figure 4.1 Observed weather at Brooms’s barn in 2006 and estimated soil moisture deficits
(SMD) for four different soil types
10
0
1 June
15 June
1 July
15 July
1 Aug
15 Aug
31 Aug
In Figure 4.1 lines indicate the estimates of soil moisture deficit (SMD) in the top 20cm. The maximum soil
moisture deficit is 22mm for sandy soil, 30mm for sandy loam, 36mm for clay loam and 42mm for silt loam
soil. So, if the soil moisture deficit reaches 22mm for sandy soil, the soil would be bone-dry, as can be
seen on 3 July. The soil became bone-dry on 11, 15 and 20 July respectively for the other soil types.
Average predicted incidence of runs of 7 days
To investigate whether there would be an increasing trend towards more of these periods, estimates of
SMD for all four soil types, all four locations and for 150 years of synthetic weather date for all timeslices
were made. The frequencies of occurrence of runs of seven consecutive days when the soil was
estimated to be bone-dry over the period from 1 March to 31 October were counted. Figure 4.2 shows the
frequencies for Kirton in Lincolnshire averaged over 150 years as an example. Increasing trends with time
were seen for all locations
30
25
20
Case study Figure 4.2
sand
sand loam
clay loam
silt loam
Average predicted incidence of runs of 7
days with soil bone-dry between 1 March
and 31 October at Kirton Lincolnshire.
(Note: 10 consecutive days count as 4
runs).
15
10
5
0
base
2020s
2050s
Timeslice
Page 7 of 17
ACO301 - Defra Final Project Report – March 2008 (ANNEX 8)
In the East of England growers are unlikely to have access to irrigation currently and in future may not
have the water resources to supply irrigation systems if they have them. To investigate this further, the
estimates of SMD for silt loam at Kirton were summarised. Figure 4.3 shows a box plot of these data and
shows that incidence of these periods of bone-dry soil is likely to increase with time.
Case study Figure 4.3 Predicted incidence of runs of 7 consecutive days with bone-dry silt loam soil over
the period from 1 March to 31 October for Kirton Lincolnshire. (Note: 10 consecutive days count as 4
runs). The predictions for each of the 150 years of synthetic weather data have been summarised in the
box plots. The boxes represent the 25th and 75th percentiles and the line within the box represents the
median. The whiskers represent the 10th and 90th percentiles and the dots show the maximum and
minimum values.
Frequency of predicted incidence of
runs of 7 days bone dry soil
60
50
40
30
20
10
0
base
2020s
Timeslice
Page 8 of 17
2050s
ACO301 - Defra Final Project Report – March 2008 (ANNEX 8)
Case study 5 Leafy Salad production
Quality
Leafy salads, grown over a long season for continuous supply are vulnerable to extreme weather events
causing quality thresholds to be exceeded. Discussion with industry indicated that heavy rainfall, (defined
as >12mm in a day, >25mm in a week, or >50mm in a month); high temperatures, (defined as >28°C
during the day and >14°C at night); low temperatures, (defined as <5°C); and differences in day/night
temperature <8°C would be expected to lead to a negative impact on quality. To assess the impact of
extreme weather events on the production of leafy salads, the risks of exceeding quality thresholds were
estimated for the period 1 March to 31 October based on synthetic weather data simulated for 16 locations
across the UK. An example for Brooms Barn is shown below in Figure 5.1 summarised as box plots.
night >14oC
day >28oC
30
25
20
15
10
5
0
base
2020s
80
60
40
20
0
2050s
base
2020s
2050s
night <5oC
day >28oC & night >14oC
30
40
25
Number of days when...
Number of days when...
Case study Figure 5.1
100
Number of days when...
Number of days when...
35
20
15
10
5
0
base
2020s
30
20
The likelihood of temperatures
a) exceeding 28°C during the
day, b) exceeding 14°C at
night, c) falling below 5°C and
d) the difference between day
and night being less than 8°C.
Data are based on 150 years
of synthetic data simulated for
Kirton (1 May to 31 Oct). The
boxes represent the 25th and
75th percentiles and the line
within the box represents the
median.
The
whiskers
represent the 10th and 90th
percentiles and the dots show
the maximum and minimum
values.
10
0
2050s
base
2020s
2050s
difference between day & night <8oC
Number of days when...
120
100
80
60
40
20
base
2020s
2050s
Overall the risk of exceeding the thresholds for rainfall remained relatively similar in the short to medium
term. However, there would appear to be a much greater risk of exceeding 28°C during the day or 14°C at
night as a result of climate change (Figure 5.1); this may have a detrimental effect on quality. Conversely,
the number of occasions when night temperatures are expected to fall below 5°C (between 1 May to 31
Oct) is predicted to decrease and the number days when the difference between day and night
temperatures is less than 8°C is predicted to decrease at most, but not all sites.
Page 9 of 17
ACO301 - Defra Final Project Report – March 2008 (ANNEX 8)
Continuity of supply
In order to investigate possible disruption to continuity, a crude and unpublished lettuce model for CV
Saladin was used to calculate a planting schedule for iceberg lettuce. The model is based on data from 24
transplanted crops of Saladin grown at Wellesbourne between 1980 and 1986. The data were described
by Wurr, Fellows & Suckling (1988). The model is as follows
y  a  bx
where y is accumulated day-degrees above a base of 0oC from 1 March to day of cutting at maturity and x
is accumulated day-degrees above a base of 0oC from 1 March to day of planting and a is 816.3 and b is
1.0775.
This model is crude and is inherently flawed because the y and x variables are not independent of one
another. However, from a pragmatic angle, the model has been used successfully to design planting
schedules for a commercial grower over several years, though these data are confidential and so may not
be quoted here. The earliest and latest date of cutting in the data used to derive this model were 18 June
and 16 October respectively. Target cuttings of one unit of lettuce were established for 60 dates cutting
one unit on each Monday, Wednesday and Friday from week 23 to week 42. The planting schedule was
estimated using Wellesbourne observed weather data for 2005. The data for the most recent season 2006
were not used as the season was hotter than average. The planting schedule was then run using
observed weather data for 2001, 2002, 2003, 2004 and 2006. Figure 5.2 below shows the results of the
model runs.
Case study Figure 5.2 Target and predicted day of cutting for iceberg lettuce in 2001, 2002, 2003, 2004
and 2006. Predictions were not made for 2005 because the weather for 2005 was used to estimate the
planting schedules.
320
320
Day of cutting
300
TARGET cut
predicted 2001
280
280
260
260
240
240
220
220
200
200
180
180
160
160
140
140
60 80 100120 140160180 200220240 260
60 80 100120 140160180 200 220240 260
320
Day of cutting
300
320
TARGET cut
predicted 2003
280
260
260
240
240
220
220
200
200
180
180
160
160
140
140
60 80 100120 140160180 200 220240 260
320
60 80 100120 140160180 200220240 260
320
TARGET cut 2005
300
280
280
260
260
Day of cutting
Day of cutting
TARGET cut
predicted 2004
300
280
300
TARGET cut
predicted 2002
300
240
220
200
TARGET cut
predicted 2006
240
220
200
180
180
160
160
140
140
60 80 100 120 140 160 180 200 220 240 260 60 80 100120 140160180 200220240 260
Day of planting
Day of planting
Figure 5.3 shows the same data as a weekly production plot which is targeted to produce three units of
iceberg lettuce per week over the season.
Page 10 of 17
ACO301 - Defra Final Project Report – March 2008 (ANNEX 8)
Case study Figure 5.3 Target and predicted units of iceberg lettuce in 2001, 2002, 2003, 2004 and 2006.
Predictions were not made for 2005 because the weather for 2005 was used to estimate the planting
schedules
6
6
2001
5
4
4
3
3
2
2
1
1
0
0
25
30
35
40
6
25
30
35
40
6
2003
5
Predicted units to cut
2002
5
4
4
3
3
2
2
1
1
0
0
25
30
35
40
6
2004
5
25
30
35
40
6
Target 2005
5
4
4
3
3
2
2
1
1
0
0
25
30
2006
5
35
40
Week number
25
30
35
40
Week number
The planting schedule was then run with synthetic data for Kirton generated by the weather generator.
The first five years of data within each data set of 150 years were used as these had been randomly
generated. The resulting predicted production did not vary greatly into the future other than predicting
earlier cutting so shifting the cuts in time. Figure 5.4 shows the predicted production for the base and
2020s and 2080s.
Page 11 of 17
ACO301 - Defra Final Project Report – March 2008 (ANNEX 8)
Case Study Figure 5.4 Predicted production shown by the black bars and the green area showing the target production of 3 units of lettuce per week.
Predicting iceberg lettuce cuts Kirton 2020s
and Saladin model 1986
Predicting iceberg lettuce cuts Kirton base
and Saladin model 1986
7
Year 1
6
Year 2
6
Predicting iceberg lettuce cuts Kirton 2050s
and Saladin model 1986
7
Year 1
6
7
Year 2
6
7
Year 1
6
5
5
5
5
5
4
4
4
4
4
4
3
3
3
3
3
3
2
2
2
2
2
2
1
1
1
1
1
1
0
0
0
0
0
0
30
35
40
25
30
35
25
40
5
4
4
3
3
2
2
1
1
0
0
25
30
35
40
7
Year 4
6
5
25
30
35
Year 5
25
30
35
40
25
7
Year 3
4
4
3
3
2
2
1
1
0
0
25
30
35
40
Year 4
25
30
35
40
Year 5
25
Year 3
5
4
4
3
3
2
2
1
1
0
0
25
30
35
40
6
5
5
4
4
4
4
4
4
3
3
3
3
3
3
2
2
2
2
2
2
1
1
1
1
1
1
0
0
0
0
0
0
40
25
30
35
Week number
40
25
30
35
Week number
40
25
30
35
Week number
Wurr, D.C.E., Fellows. Jane R. & Suckling, R.F. (1988) J. agric.Sci., Camb. 111, 481-486
Page 12 of 17
40
30
35
40
25
30
35
Week number
Year 5
6
Target based on
Wellesbourne 2005
5
35
Year 4
25
5
30
40
7
5
Week number
35
6
5
5
25
30
7
7
6
Target based on
Wellesbourne 2005
35
6
40
7
6
30
7
6
5
7
6
Target based on
Wellesbourne 2005
40
5
40
7
6
35
6
Predicted units to cut
Year 3
6
30
7
7
Predicted units to cut
25
Year 2
6
5
7
Predicted units to cut
7
7
40
25
30
35
Week number
40
ACO301 - Defra Final Project Report – March 2008 (ANNEX 8)
Case study 6 Peach-potato aphid Myzus persicae
The peach-potato aphid, Myzus persicae, is a pest of potato, sugar beet, brassica and several other
outdoor and protected crops. It is important as a vector of several viruses, as a contaminant, and for its
effect on crop growth, if present in large numbers. Population development is affected strongly by
temperature. The overwintering population consists largely of anholocyclic nymphs and adults. Analyses
of long-term data sets from the Rothamsted Insect Survey have shown that there is a negative correlation
between winter temperature and the date the first alate of a given species is captured in suction traps
(Harrington et al., 1990). There is also often a strong positive correlation between winter temperature and
aphid abundance up until early July.
This case study uses an equation developed from Rothamsted Insect Survey data to predict the initial
migration of alate M. persicae from overwintering hosts to new crops in the spring (Collier & Harrington,
2001) and then published day-degree relationships to predict the number of generations that will be
completed during the summer/autumn period. All models were run for four locations, Camborne, Kirton,
Leuchars and Wye. The models used 150 years of synthetic weather data.
For all models, the first aphid migration was predicted using the following equation (Collier & Harrington
2001).
D = 256.6 - 11.616 JFT + 4.44 JFR - 44.78 X - 9.65 Y + 4.304 X2 + 0.947 Y2 + 2.308 XY + 0.1353 Z
where
D is the date of first record expressed as 1 = 1 January
JFT is the mean temperature in January and February (oC)
JFR is the total rainfall in January and February (loge mm)
X is the longitude (as a four figure grid reference / 1000)
Y is the latitude (as a four figure grid reference / 1000)
And Z is the altitude
For all locations the predictions of first migration was earlier as time increased and Figure 6.1 show box
plots summarising the predictions of the first sighting for Kirton in Lincolnshire as an example.
Predicted day where 1 = 1 Jan
180
Case study Figure 6.1
Predicted day of first sighting for Kirton
Lincolnshire. The predictions for each of the
150 years of synthetic weather data have been
summarised in the box plots. The boxes
represent the 25th and 75th percentiles and the
line within the box represents the median. The
whiskers represent the 10th and 90th percentiles
and the dots show the maximum and minimum
values.
160
140
120
100
base
2020s
2050s
Timeslice
Three models predicting generation time were selected from the Phenology Model Database at Davis at
http://www.ipm.ucdavis.edu/MODELS/index.html and are outlined below.
Model 1 Whalon & Smilowitz (1979)
This model predicts the generation time from nymph to nymph using a day-degree sum with a base
temperature of 4 oC and an upper limit of 30oC. The model was developed with a host crop of potato and
the thermal sum describing completion of a generation was 129.8 day-degrees.
Page 13 of 17
ACO301 - Defra Final Project Report – March 2008 (ANNEX 8)
Model 2 Weed,(1927)
This model also predicts the generation time from nymph to nymph using a day-degree sum with the same
base temperature of 4 oC as for model 1, but with a lower upper limit of 28oC. The model was developed
with a host crop of spinach and the thermal sum describing completion of a generation was 152.5 daydegrees.
Model 3 El Din(1976).
The third model also predicts the generation time from nymph to nymph using a day-degree sum, but with
a base temperature of 3.3 oC and an upper limit of 25oC. The model was developed with a host crop of
Brussels sprout and the thermal sum describing completion of a generation was 132.1 day-degrees.
All three models predicted that the number of generations within the calendar year would increase. Figure
6.2 shows frequency distributions for the predicted number of generations completed using 150 years of
synthetic weather data and the first model of Whalon & Smilowitz (1979).
Case study Figure 6.2 frequency distributions for the predicted number of generations completed using
150 years of synthetic weather data and the first model of Whalon & Smilowitz (1979).
75
75
Camborne Cornwall
50
base
2020s
2050s
Frequency of synthetic years
25
Kirton Lincolnshire
50
25
0
0
6
75
8
10 12 14 16 18 20 22 24 26 28
Wye Kent
6
75
50
50
25
25
0
6
8
8
10 12 14 16 18 20 22 24 26 28
Leuchars Fife
0
10 12 14 16 18 20 22 24 26 28
6 8 10 12 14 16 18 20 22 24 26 28
Number of generations before 31 December
Collier & Harrington (2001). Outdoor lettuce: Refinement and field validation of forecasts for the aphid pests of
lettuce foliage. HDC Final Report FV 162e.
El Din, N. S. 1976. Effects of temperature on the aphid Myzus persicae (Sulz.), with special reference to
critically low and high temperature. Zeitschrift fur Angewandte Entomologie 80: 7-14.
Harrington, R., Tatchell, G. M. and Bale, J. S. (1990) Weather, life cycle strategy and spring populations of
aphids. Acta Phytopathologica et Entomologica Hungarica 25, 423-432.
Weed, A. 1927. Metamorphosis and reproduction in apterous forms of Myzus persicae Sulzer as influenced by
temperature and humidity. J. Econ. Entomol. 20: 150-157.
Whalon, M. E., and Z. Smilowitz. 1979. Temperature-dependent model for predicting field populations of green
peach aphid, Myzus persicae (Homoptera: Aphididae). Can. Ent. 111: 1025-1032.
Page 14 of 17
ACO301 - Defra Final Project Report – March 2008 (ANNEX 8)
Case Study 7 Cutworm Agrostis segetum
A descriptive population model for cutworm (Bowden et al., 1983) was run using 150 years of synthetic
weather data for two locations, Camborne, Kirton. It is suggested that cutworm infestations are usually
worse in East Anglia and in the South West than in other areas of England and Wales and that the species
is scarce in Scotland. The model estimates larval survival to third instar based on temperature-rate
relationships and on mortality attributed to daily rainfall. In 2006 there were incidences of a second
generation of moths laying more eggs but for this study it was assumed that there would be only one
generation and that moths would lay egg batches nightly from 1 June to 31st July. The mortality factor
states that 0.1mm of rainfall will kill 1% of surviving first and second instar larvae but that rainfall will not
affect the older larvae. So if 10mm of rain was recorded all larvae reaching third instar on that day would
be killed. The number of surviving third instar larvae has been counted up to 31 July.
Figure 7.1 shows box plots of the predicted numbers of batches of larvae surviving to third instar
accumulated up to the end of July at both locations. In both locations, the numbers of surviving batches
increases into the future.
Predicted numbers of surviving batches
of instar3 larvae on 31 July
Case study Figure 7.1 Predicted number of batches o larvae surviving to thirst instar up to 31 July for
Camborne in Cornwall and Kirton in Lincolnshire. The predictions for each of the 150 years of synthetic
weather data have been summarised in the box plots. The boxes represent the 25th and 75th percentiles
and the line within the box represents the median. The whiskers represent the 10 th and 90th percentiles
and the dots show the maximum and minimum values.
50
50
Kirton
Camborne
40
40
30
30
20
20
10
10
0
0
base
2020s
2050s
base
Timeslice
2020s
2050s
Timeslice
Bowden, J., Cochrane, J., Emmett, B. J., Minall, T. E. & Sherlock, P. L. (1983). A survey of cutworm
attacks in England and Wales, and a descriptive population model for Agrotis segetum (Lepidoptera:
Noctuidae). Annals of Applied Biology 102 29-47.
Page 15 of 17
ACO301 - Defra Final Project Report – March 2008 (ANNEX 8)
Case study 8 Diamondback Moth Plutella xylostella
Models predicting generation time from egg to adult moth were selected from the Phenology Model
Database at Davis at http://www.ipm.ucdavis.edu/MODELS/index.html
The model of Harcourt (1954) predicted the generation time from egg to adult using a day-degree sum with
a base temperature of 7.3 oC. The model was developed with a host crop of cabbage and the thermal sum
indicating completion of a generation was 283 day-degrees. A further model (Butts & McEwen, 1981) was
identified with the same base temperature but with a thermal sum 10 day-degrees greater than that of the
previous model. This model, based on a host crop of Brussels sprouts, was not used because the
difference in thermal sum was so small that the trends would be very similar.
Predictions were made from three different times of egg-laying. Firstly, the assumption was made, that the
moths would successfully over-winter in the UK, and start laying eggs from 1 February onwards. The
second and third times of egg-laying were based on moths migrating into the UK from abroad, so eggs
were laid on either 1 May or 1 June. For all three prediction model runs the generations were counted until
either the first Autumnal frost, or if there was none, to the end of the calendar year. Predictions were
made for four locations, Camborne Cornwall, Kirton, Lincolnshire, Wye, Kent and Leuchars, Fife. As the
predictions were made further into the future, the number of generations within the calendar year
increased as Figure 8.1 shows.
Case study Figure 8.1 Predicted number of generations completed by Plutella xylostella (diamondbackmoth) in different locations in the UK using synthetic weather data.
base
2020s
2050s
Camborne Cornwall
Kirton Lincolnshire
Wye Kent
Leuchars Fife
2
3
7
2
3
4
5
6
7
2
3
4
5
6
7
2
3
4
5
6
7
2
3
4
5
6
7
2
3
4
5
6
7
2
3
4
5
6
7
2
3
4
5
6
7
2
3
4
5
6
7
2
3
4
5
6
7
2
3
4
5
6
7
2
3
4
5
6
7
150
moths
over-winter
first sighting
on 1 February
100
50
moths
migrate in
first sighting
on 1 May
Frequency of synthetic years
0
4
5
6
150
100
50
0
Camborne Cornwall
150
moths
migrate in
first sighting
on 1 June
100
50
0
Number of generations before either the first Autumnal frost or 31 December
Butts, R. A. and F. L. McEwen. 1981. Seasonal populations of the diamondback moth, Plutella xylostella
(Lepidoptera: Plutellidae), in relation to day-degree accumulation. Can. Ent. 113: 127-131.
Harcourt, D. G. 1954. The biology and ecology of the diamondback moth, Plutella maculipennis (Curtis), in
eastern Ontario. Ph.D. thesis, Cornell University, Ithaca, N.Y. 107p.
Page 16 of 17
ACO301 - Defra Final Project Report – March 2008 (ANNEX 8)
Case study 9 Late Blight (Phythophthora infestans)
Late blight is the current single greatest threat to potato production in the UK. There are many forecast
systems for late blight but most of these require measurements of humidity. In order to look at possible
effects
on
late
blight
a
single
model
was
identified
on
the
Davis
website
http://www.ipm.ucdavis.edu/DISEASE/DATABASE/potatolateblight.html
The model of Hyre (1954) required only daily rainfall and minimum and maximum temperatures {EE???}.
The model forecasts the outbreak of potato late blight 7-14 days after the occurrence of ten consecutive
blight favourable days. Days are considered blight favourable when the 5-day average temperature is
below 25.5oC and the total rainfall for the last 10 days is 30mm or greater. Days with minimum
temperatures below 7.2o C are considered unfavourable.
The model was run to estimate the number of blight favourable days and the number of days when blight
was forecast within a window from 1 May to 31 October for all 150 years at each location and for each
timeslice. There was very little difference between the estimates and forecasts for the timeslices as the
box plots in Figure 9.1 for Camborne in Cornwall show.
100
80
60
40
20
0
base
2020s
2050s
Number of days when blight was forecast
Number of days favourable to blight
Case study Figure 9.1 A model was run to estimate blight favourable days and to forecast blight within a
window from 1 May to 31 October. The figure shows box plots of the predicted numbers of days at
Camborne in Cornwall. The predictions were made using 150 years of synthetic weather data. The
symbols show the 5 and 95th percentile outliers.
40
30
20
10
0
base
2020s
2050s
Timeslice
Hyre, R. A. 1954. Progress in forecasting late blight of potato and tomato. Plant Disease Reporter. 38: 245-253.
Page 17 of 17