Temperature and lettuce growth
330 ppmv was assumed, in accordance with data
of
Baker and Allen (1993). Dry weight accounting for
C0
2
concentration (W
3)
was calculated from W
2
of
equation (4), such
that:
(5)
in which C is
C0
2
concentration (ppmv) and
z
is a constant calculated from the regression
of
percentage increase in dry weight and
C0
2
concentration which has the value 0.0684 (s.e.
=
0.014). Daily increase in dry weight
of
lettuces (allowing 2 days growth check after transplanting
of
plants
of
0.13 g dry weight) were calculated from daily records
of
mean air temperature and incident radiation for Rothamsted (51.8°N latitude, 0.5°W longitude) in
15
of
the years between 1968 and 1991 using equations (2)(5) (the years 1970, 1976, and 19811987 were omitted due to the absence
of
radiation data). The daily increase in temperature calculated near to this site
(51
.25°N latitude, 0.5°W longitude) for six scenarios
of
climate change were then added to these baseline temperature data. The scenarios used were : three composite timedependent scenarios for the years 2010, 2030, and 2050 (assuming a Businessasusual (A) policy emissions scenario (Houghton
et
al.,
1990)), and three GCM equilibrium 2 x
C0
2
scenarios (derived from the Geophysical Fluid Dynamics Laboratory,
GFDL;
United Kingdom Meteorological Office low resolution, UKMOL ; and the Goddard Institute
of
Space Studies, GISS, general circulation models). More details
of
these scenarios are provided by Barrow (1993). The mean temperature increase and
C0
2
concentration calculated using these six scenarios for this growth period at a site near
to
Rothamsted are shown in Table
1.
The daily radiation integral for the baseline period was also used for each scenario. The dry matter content
of
lettuces
of
more than 50 g fresh weight was found to be constant at 4.5 per cent. The calculated daily increase in dry weight
of
lettuce was then used to forecast maximum plant fresh weight at 4.5 per cent dry matter, and time to attain a fresh weight
of
200 g for each
of
the
15
years using each
of
the six scenarios.
Table
1.
Summary
of
the mean temperature increase relative to the baseline climate, and
COz
concentration
for
the present and climate change scenarios
for
the growing
period
15
August
to 2
7
September.
Baseline A2010 A2030 A2050 GFDL UKMO GISS Temperature 0 0.5
1.2 1.9
5.6 5.6 4.3 increase (°C)
C0
2
concn. 353 400 458 539 560 560 560 (ppmv)
Vol. 2,
0°
4
1993
307
RESULTS Growth
o
lettuce
Mean air temperature during crop growth differed between plots at each end
of
the tunnels by 5.5 and 4.1 °C (range 15.6 to 21.1 °C, and 14.7° to 18.8 °C) in tunnels 1 and 2, respectively. Mean air temperature for the harvest intervals was between 10.7 and 22.3 °C. Mean ambient air temperature over this period was 16.6 °C. The initial increase in dry weight was more rapid in the warmer than the cooler
plots:
dry weight (calculated from each fitted function) was a positive function
of
mean temperature at
14
d from transplanting
p
<
0.01). However, the maximum dry weight
of
lettuce grown at the cooler end
of
the tunnel, although achieved more slowly, was ultimately greater than that grown under warmer conditions. Accordingly, the relationship between dry weight and mean temperature was negative for final maximum dry weight
p
<
0.01). Mean relative growth rate, calculated for each harvest interval, declined linearly with time from transplanting at all temperatures. Also, there was an interaction between time from transplanting and mean temperature for each harvest interval such that the relationship between mean relative growth rate and temperature changed during ontogeny from a positive function soon after transplanting to a negative one at about 40 d from transplanting (Figure
1).
Thus, the
i:l
0,3
bl
0,2
Q)
§
.s:::
0,1
e
Ol
Q)
>
0
Oi
a:
Time from transplanting
d)
Temperature
(oC)
Figure
1.
Relationship
between mean
relative growth rate
(•),
time
from
transplanting
(t),
and
mean temperature
en
for
that harvest
interval. The
fitted
response plane is
described
by
R
=
0.179
+
0.39
X
10
3
t
+
10.39
X
10
3
T 0.613
X
l0
3
t.T, R
2
=
0.835,
91
df
The
position
of
each observation
from
the
plane
is denoted by a bar.
308
Table
2.
Optimum values
of
the parameters
of
equations 2)
and
3).
Parameter Value
Ro
0.35
g
g
1 d1
ef
685 °Cd a 0.037
oc l
b
0.005
MJ
1
m
d1
optimum temperature for relative growth rate appeared to decline from more than
23
oc
soon after transplanting to less than 10
oc
at final harvest. The optimum values for the parameters
of
equations (2) and (3) for plants grown in two plots are shown in Table 2. The increase in dry weight predicted using these equations with these parameter values reproduced both the rapid increase in plant weight observed in the warmer plot, and the higher final dry weight observed in the cooler plot (Figure 2a). The mean daily relative growth rates predicted were initially larger in the warmer plots, but then declined more rapidly compared with values predicted for the cooler plot. This trend followed that observed for estimates
of
mean relative growth rate (Figure 2b). Parameter values for equations (2) and (3), optimised for two plots
of
lettuce, provided an accurate prediction
of
the dry weights observed independently in the remaining 14 plots (Figure 3). No systematic deviations between observed and predicted weights were detected over the range
of
observations from 2.0 g to 14.8 g dry weight, except for some
of
the heaviest plants for which the model overestimated plant weight. The main variable in the model was time. Predictions using the model based on time alone (given
by
the fitted relationship between the logarithm
of
dry weight and time for all plants grown in the two plots from which model parameters were estimated) accounted for 84.5 per cent
of
the variance
of
the independent observations, compared with 87.1 per cent once temperature and radiation were added ; a decrease in the residual sum
of
squares of
17
per cent (83 df).
Effect
o
climate change
Mean maximum fresh weight estimated using the baseline temperatures for
15
years and the current
C0
2
concentration was 262 g (CV
=
4.0 per cent). This mean increased progressively using the timedependent scenarios, although increase in yield was not significant with the A2010 and A2030 scenarios (Figure 4). Calculated maximum fresh weight for the year 2050 was only 4.7 per cent greater compared with the baseline climate, with a small decrease in variation about the mean (CV
=
3.8 per cent). In contrast, maximum fresh weight was less for all equilibrium 2 x
C0
2
scenarios compared with the baseline
b
bl
12
9
8
Q)
§
r:
Cl
4
Q)
>
jj
a:
0 0,3
b
0,2
bl
Q)
§
r:
0,1
0
0,
Q)
>
jj
a:
0
T.R. Wheeler
et
al.
0 10 20
30
40
Time from transplanting
Figure
2.
Relationship
between
plant
dry
weight
a), relative growth rate b)
and
time
from
transplanting
for
lettuces grown in
plots
1
•) and
7
.A.)
of
tunnel
1.
The
fitted
functions
repre
sent predictions
using equations 2)

4)
and
the values
of
the parameters shown in Table 2
for
plots
1
7


). Each
observation
for
dry
weight
is the
mean
of
nine
plants.
§
:c
Cl
Qi
C:
0
0
Q)
tl
6
Q)
::
16
12 8
4
0
4 8 12
16
Observed dry weight
g)
Figure
3.
Comparison
of
actual
dry
weight
of
lettuces
from
14
independent
plots
within the tunnels with those
predicted
using equations 2)

4) with the value
of
the
parameters presented
in Table
2.
The line shown represents
exact agreement
between
observation
and
prediction.
Eur.
J
Agron.
Temperature
and
lettuce growth
climate. The greatest reduction in yield was 5.5 per cent using the GFDL scenario, although differences among these three GCM 2 x
C0
2
scenarios were not significant (Figure 4
).
Variation about these means was increased compared to the baseline (for example, for the GFDL scenario, CV
=
5.6 per cent).
300
§
.E
200
l
iii
: :
.c
/)
lL
100 Baseline A2010 A2050
UKMOL
A2030 GFDL GISS
Figure
4.
Mean
maximum fresh
weight
of
lettuces
calculated
using the
baseline
temperature
and
radiation
record
for
15 years
at Rothamsted
between 1968
and
1991)
and
six scenarios
of
climate
change. Each
bar
represents the
mean±
one s.e.
The mean estimated time for lettuces to reach a fresh weight greater than 200 g was
35
days from transplanting using the baseline climate and current
C0
2
concentration (CV
=
6.1
per cent).
30
Ol
0 0
C\J
20
.8
Q)
E
i=
10
Baseline
A201
0 A2050
UKMOL
A2030 GFDL GISS
Figure
5.
Mean time
from
transplanting to
200
g
fresh
weight
calculated
using the baseline temperature
and
radiation
record
for
15
years
at
Rothamsted
between
1968
and
1991)
and six
scenarios
of
climate change. Each
bar
represents the
mean
±
one s.e.
Vol.
2,
n°
4
1993
309
This time was significantly reduced using all six climate change scenarios except A2010 (Figure 5). Using the three timedependent scenarios, the mean time to 200 g declined progressively from
33
d for the year 2010 to 30 d for the year 2050. Mean time to 200 g in all three equilibrium 2
x
C0
2
scenarios was 26 d (CVs were 4.7, 4.9 and 4.9 per cent, respectively). DISCUSSION Increasing temperature resulted in a rapid increase in early dry weight accumulation in lettuce but reduced final weight. This response accounts for both the positive correlation reported between temperature and lettuce weight during early growth (Hicklenton and Wolynetz, 1987), and the negative relationship reported for crisphead lettuce between temperature and head weight and head density at final harvest (Wurr and Fellows, 1991 ; Wurr
et al.,
1992). Similar examples have been reported for soyabean (Hadley
et al.,
1994) and pearl millet
Pennisetum typhoides
S.
and H.) (Squire
et al.,
1984). The effect
of
temperature on final crop yield is dependent on whether a crop is determinate or indeterminate (Goudriaan and Unsworth, 1990). The response observed here for lettuce is characteristic
of
that for determinate crops, in which an increase in temperature usually results in a reduction in final yield (Squire, 1990 ; Goudriaan and Unsworth, 1990). The relationship between relative growth rate and temperature changed during ontogeny. Although
Ri
was a function
of
temperature, equation (3), the decline in potential relative growth rate
(RP)
during ontogeny resulted in an apparent decline in the temperature optimum for relative growth rate
of
lettuce during crop growth. This is exactly the response predicted previously using equations (2) and (3) to describe the growth
of
soy abean (Hadley
et al.,
1994
,
and thus provides evidence in lettuce to support the hypothesis that instantaneous relative growth rate is a function of current temperature and developmental thermal time. The equations for the effects
of
temperature and incident radiation on crop growth described and validated here for lettuce may provide a framework for a simple and robust model to assess the effects
of
potential climate change on crops. The model, as optimized here, can only be used at this plant density for a well watered and fertilized crop ; further experiments at different sowing dates are required to extend the application
of
the model. However, the potential effect
of
different climate change scenarios on outdoor butterhead lettuce production can be assessed within the limits
of
these assumptions. The overall effect
of
potential climate change on the yield
of
many determinate crops will partly be a