You are on page 1of 10

AGEE-3025; No of Pages 10

Agriculture, Ecosystems and Environment xxx (2007) xxx–xxx


www.elsevier.com/locate/agee

Impact of elevated CO2 and temperature on rice yield and methods


of adaptation as evaluated by crop simulation studies
P. Krishnan a,*, D.K. Swain b, B. Chandra Bhaskar b, S.K. Nayak a, R.N. Dash b
a
Division of Biochemistry, Plant Physiology and Environmental Sciences, Central Rice Research Institute, Cuttack 753006, India
b
Division of Soil Science and Microbiology, Central Rice Research Institute, Cuttack 753006, India

Received 7 February 2006; received in revised form 20 December 2006; accepted 19 January 2007

Abstract

Impact of elevated CO2 and temperature on rice yield in eastern India was simulated by using the ORYZA1 and the INFOCROP rice
models. The crop and weather data from 10 different sites, viz., Bhubaneswar, Chinsurah, Cuttack, Faizabad, Jabalpur, Jorhat, Kalyani, Pusa,
Raipur and Ranchi, which differed significantly in their geographical and climatological factors, were used in these two models. For every
1 8C increase in temperature, ORYZA1 and INFOCROP rice models predicted average yield changes of 7.20 and 6.66%, respectively, at
the current level of CO2 (380 ppm). But increases in the CO2 concentration up to 700 ppm led to the average yield increases of about 30.73%
by ORYZA1 and 56.37% by INFOCROP rice. When temperature was increased by about +4 8C above the ambient level, the differences in the
responses by the two models became remarkably small. For the GDFL, GISS and UKMO scenarios, ORYZA1 predicted the yield changes of
7.63, 9.38 and 15.86%, respectively, while INFOCROP predicted changes of 9.02, 11.30 and 21.35%. There were considerable
differences in the yield predictions for individual sites, with declining trend for Cuttack and Bhubaneswar but an increasing trend for Jorhat.
These differences in yield predictions were mainly attributed to the sterility of rice spikelets at higher temperatures. Results suggest that the
limitations on rice yield imposed by high CO2 and temperature can be mitigated, at least in part, by altering the sowing time and the selection
of genotypes that possess higher fertility of spikelets at high temperatures.
# 2007 Elsevier B.V. All rights reserved.

Keywords: Climate change; CO2; INFOCROP; ORYZA; Oryza sativa L.; Simulation; Temperature; Yield

1. Introduction their potential threat to rice productivity and the associated


impact on the socioeconomic structure of many rice-growing
The climatic variability and the predicted climatic changes countries. Among the global atmospheric changes, the incre-
are of major concern to the rice crop scientists because of asing concentrations of greenhouse gases such as CO2 may
have significant effect on rice productivity due to increase in
Abbreviations: BVP, Basic Vegetative Phase; DLAI, Death Rate of both the average surface temperature and the amount of CO2
Leaf Area Index; FACE, Free Air Concentration Enrichment; GCMs, available for photosynthesis (Aggarwal, 2003). In the absence
General Circulation Models; GFDL, General Fluid Dynamics Laboratory; of temperature increase, many studies have shown that the net
GFP, Grain Filling Phase; GISS, Goddard Institute of Space Studies; GLAI,
Leaf Area Growth Rate; IPCC, Intergovernmental Panel on Climate
effect of doubling of CO2 was increase in the yield of rice
Change; LAI, Leaf Area Index; NATP, National Agricultural Technology (Kim et al., 2003). It becomes necessary to assess the effects of
Project; PFP, Panicle Formation Phase; PLTR, Net loss of LAI due to potential interactive changes of CO2 and temperature in order
transplanting; PSP, Photoperiod-Sensitive Phase; RLAI, Net Leaf Area to determine the future agricultural strategies that would
Growth Rate; RUE, Radiation Use Efficiency; RWLVG, Increment in Leaf maintain higher rice productivity.
Weight; SLA, Specific Leaf Area; SUBLAI, LAI is simulated in the
subroutine SUBLAI; UKMO, United Kingdom Meteorological Office
The simulations by different models and many field
* Corresponding author. experiments have shown the potential impact of climatic
E-mail address: prameelakrishnan@yahoo.com (P. Krishnan). change and the variability in rice productivity (Baker et al.,

0167-8809/$ – see front matter # 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.agee.2007.01.019

Please cite this article in press as: Krishnan, P. et al., Impact of elevated CO2 and temperature on rice yield and methods of adaptation as
evaluated by crop simulation studies, Agric. Ecosyst. Environ. (2007), doi:10.1016/j.agee.2007.01.019
AGEE-3025; No of Pages 10

2 P. Krishnan et al. / Agriculture, Ecosystems and Environment xxx (2007) xxx–xxx

1992b; Peng et al., 2004; Kim et al., 2003). The modeling


studies from Bangladesh (Karim et al., 1994), Japan (Horie
et al., 2000), China (Bachelet et al., 1995) and India (Mall
and Aggarwal, 2002) reported the country-wise variations in
rice production, anticipated due to the climatic changes. The
simulated yields increased when temperature increases were
small, but declined when the decadal temperature increase
was more than 0.8 8C, with the greatest decline in crop
yields occurring between the latitudes of 108 and 358N.
Similar results were obtained by Penning de Vries (1993).
Many uncertainties exist in modeling studies, partly due
to the quality of the predictions by the models, from the use
of limited sites for which historical weather data are
available, due to the quality of the crop simulation models,
Fig. 1. Comparison of yield simulated by ORYZA1 and INFOCROP rice.
especially when applied under the rain-fed conditions
(Bachelet et al., 1995), and due to the quality of the climate
models used to predict future weather scenarios. These matter production as a function of light, CO2 and temperature
uncertainties may be reduced only when a large number of by considering photosynthetic processes at the leaf level and
scenarios for different locations are compared and evaluated. integrating these over the canopy to obtain crop-level values.
In order to overcome the predicted limitations for rice Respiration is also modeled explicitly as a function of
production in the future, there is also a need to identify and temperature and partitioning of dry matter is according to
evaluate the suitable agronomic practices such as altered phenology-dependent functions. Thus, ORYZA calculates
sowing date and selection of improved varieties with dry matter production as a function of gross canopy
increased spikelet fertility at high temperature and other photosynthesis, depending on the detailed calculations of
useful traits. the distribution of light within the canopies, the radiation
Attempts have been made earlier to assess the general absorbed by the canopy and photosynthesis light response
effects of global environmental changes on rice yield using curve of leaves. Growth and maintenance respiration are
simulation models, but little attention has been given to the calculated as a function of tissue N content, temperature
potential interactive effects between high temperature and and crop-specific coefficients. This methodology, although
increasing CO2 levels. Lal et al. (1998) used the CERES rice yields very accurate results, poses practical difficulties
and predicted a 20% decline in rice yields in the because of its requirement for detailed and careful measure-
northwestern India due to elevated CO2 and temperature. ments (Kropff et al., 1994). The INFOCROP model, however,
Eastern India accounts for about 63% (26.5 million ha) of uses a simpler radiation use efficiency (RUE) relationship
the total rice-growing area in India. The rice ecosystems in between intercepted solar radiation and growth, in which
these regions show characteristic differences with respect to respiration is implicit. CO2 effects are accounted for by using
the environmental factors as well as the cultural practices. a curvilinear function relating RUE to CO2 concentration.
There is an urgent need to characterize the impact of future Temperature is assumed not to affect RUE. More or less
climatic changes on rice yield in these regions for sustaining similar results can be obtained under normal radiation
the productivity. The present paper discusses the outcomes situations by calculating the net dry matter production as a
of two rice growth simulation models, ORYZA and function of RUE. Pre-determined values of RUE were input in
INFOCROP rice, when applied to different rice-growing the model as a function of crop/cultivar (RUEMAX), and
sites in the eastern India for studying the potential interactive RUE was further modified by the development stage
effects between high temperature and increasing CO2 levels (Aggarwal et al., 2004).
and for the different thermal climate change scenarios. The rice-growing regions included for the present study lie
in eastern India, and these sites (Bhubaneswar, Chinsurah,
Cuttack, Faizabad, Jabalpur, Jorhat, Kalyani, Pusa, Raipur
2. Materials and methods and Ranchi), which are geographically apart at varied
altitudes ranging from 7.8 to 86.56 m above mean sea level,
Two popular models of rice growth ORYZA1 (Kropff show characteristic features with respect to the weather and
et al., 1994) and INFOCROP rice (Aggarwal et al., 2006) are crop factors. The crop parameters included in the study were
used in this study. Prior to their use, both were evaluated obtained from the field experiments conducted at these sites
and compared (Fig. 1) and then these crop models were under National Agricultural Technology Project (NATP)
calibrated for the indica variety IR 36 at all sites. In general, RRPS 25 during 2001–2003 (Table 1).
the two models differ in the manner in which dry matter The input parameters used for the two models are given in
production and partition, leaf area development, and pheno- Table 2a (parameters such as varietal data, soil data and
logical development are calculated. ORYZA1 calculates dry weather data), Table 2b (physical and chemical properties of

Please cite this article in press as: Krishnan, P. et al., Impact of elevated CO2 and temperature on rice yield and methods of adaptation as
evaluated by crop simulation studies, Agric. Ecosyst. Environ. (2007), doi:10.1016/j.agee.2007.01.019
AGEE-3025; No of Pages 10

P. Krishnan et al. / Agriculture, Ecosystems and Environment xxx (2007) xxx–xxx 3

Table 1 were made for the main rice-growing season for each of
Sites and institutes where the field observation and weather data were these sites. Generally, the dates of sowing and transplanting
collected
were supplied along with the weather data.
Sites Institutes
Bhubaneswar Orissa University of Agriculture & Technology (OUAT) 2.1. Climate change scenarios
Chinsurah Rice Research Station (RRS)
Cuttack Central Rice Research Institute (CRRI)
Faizabad Narendra Deva University of Agriculture & For each of these rice-growing regions, the potential yield
Technology (NDUAT) of rice was simulated under 30 different combinations of
Jabalpur Jawaharlal Nehru Krishi Vishwa Vidyalaya (JNKVV) CO2 and temperature, including with the ‘fixed increment’
Jorhat Assam Agricultural University (AAU) changes in CO2 (380, 400, 500, 600 and 700 ppm) and
Kalyani Bidan Chandra Krishi Viswa Vidyalaya (BCKV)
temperature (ambient, +1, +2, +3, +4 and +5 8C) individu-
Pusa Rajendra Agricultural University (RAU)
Raipur Indira Gandhi Agricultural University (IGAU) ally, and with all the combinations of these levels of CO2 and
Ranchi Birsa Agricultural University (BAU) temperature. The actual daily weather data, collected for
Indian Meteorological Department (IMD), three consecutive years from each of these sites were used
Pune (weather data) for simulation. For temperature changes, the daily weather
data from each site were exported to Excel spread sheets and
soils) and Table 2c (geographical information). The weather then the daily maximum and minimum temperatures were
data containing measurements of sunshine hours, tempera- increased by 1–5 8C individually. Later, the modified
ture and rainfall from all these sites were used as the baseline weather data were used as the inputs for ORYZA1. In case
data for input into these crop models. Later, the simulations of INFOCROP, for a given daily weather data, the climate

Table 2a
Combined input parameters for both the models used
Varietal data Soil data Weather data (daily data)
1. Base temperature for sowing to germination 1. Soil type 1. Maximum and minimum
temperatures
2. Thermal time for sowing to germination 2. Soil texture 2. Relative humidity
3. Base temperature for germination to 50% flowering 3. pH of soil 3. Sun shine hours
4. Thermal time for germination to 50% flowering 4. EC 4. Precipitation
5. Base temperature for 50% flowering to physiological maturity 5. Sand %
6. Thermal time for 50% flowering to physiological maturity 6. Silt %
7. Optimal temperature 7. Clay %
8. Maximum temperature 8. Saturation fraction
9. Sensitivity to photoperiod 9. Field capacity
10. Relative growth rate of leaf area 10. Wilting point
11. Specific Leaf Area 11. Saturated hydraulic conductivity
12. Extinction coefficient of leaves at flowering 12. Bulk density
13. Radiation Use Efficiency 13. Organic carbon
14. Root growth rate
15. Index of greenness of leaves
16. Sensitivity of crop to flooding
17. Index of N fixation
18. Slope of storage organ number/m2 to dry matter during storage
organ formation stage
19. Potential storage organ weight at maximum temperature
20. Nitrogen content of storage organ
21. Sensitivity of storage organ setting to low temperature
22. Sensitivity of storage organ setting to high temperature

Table 2b
Model inputs of physical and chemical properties of soil at different locations
Soil properties Bhubaneswar Kalyani Faizabad Ranchi Pusa Raipur Jabalpur Jorhat Cuttack
Texture Sandy clay loam Sandy clay loam Loam Loam Sandy clay loam Silt Clayey Sandy clay loam Sandy loam
B.D.a (g/cc) 1.35 1.45 1.3 1.41 1.3 1.23 1.34 1.48 1.4
O.C.b (%) 0.73 0.51 0.3 0.68 0.57 0.54 0.69 0.67 0.47
Total N (%) 0.07 0.04 – 0.07
Available N (kg/ha) 128.4 355 258.6 – 267 265.8
a
B.D., bulk density.
b
O.C., organic carbon.

Please cite this article in press as: Krishnan, P. et al., Impact of elevated CO2 and temperature on rice yield and methods of adaptation as
evaluated by crop simulation studies, Agric. Ecosyst. Environ. (2007), doi:10.1016/j.agee.2007.01.019
AGEE-3025; No of Pages 10

4 P. Krishnan et al. / Agriculture, Ecosystems and Environment xxx (2007) xxx–xxx

Table 2c
Model inputs of geographical information of different locations
Soil properties Bhubaneswar Kalyani Faizabad Ranchi Pusa Raipur Jabalpur Jorhat Cuttack
0 0 0 0 0 0 0 0
Latitude 20814 N 22857 N 22887 N 23817 N 2285 N 25859 N 23809 N 2688 N 208300 N
Longitude 858520 E 888210 E 88840 E 858190 E 868E 85840 E 798580 E 95850 E 868E
Altitude (m) 25.9 7.8 8.62 625 23 51.84 411 86.56 23

change scenarios were specified as +1 to +5 8C for relative increase predicted for each site was weighed by its
temperatures and the percent changes in precipitation. current production observed in the field trials of an earlier
The inbuilt weather generator of that model generated the study (Annual Report of NATP RRPS 25, 2002–2003). The
modified daily weather data accordingly, and these data were differences in the production capacities among these sites
used as the daily weather inputs. The climate change were evaluated.
scenarios under different levels of CO2 were applied by
changing the ambient CO2 concentration parameter in these
two models. 3. Results
Besides, the outputs from the climate models were used
in combination with these crop models. The coarse grid from 3.1. Effect of temperature and CO2 levels at fixed
each GCM was interpolated using a four point inverse- increments on yields
distance-squared algorithm to a 0.58 latitude  0.58 long-
itude grid using a raster-based Geographical Information At all the CO2 levels tested (380, 400, 500, 600 and
Systems (GIS) software package. Scenarios were produced 700 ppm), both the models predicted the declining yields of
by applying ratios of precipitation or differences in tempe- rice due to an increase in temperature. On the contrary, an
rature predicted for the 2  CO2 and 1  CO2 simulation to increase in CO2 level at any particular temperature increased
the baseline present climate data set for different sites the rice yields (Table 4). At the current level of CO2
(Mathews and Wassmann, 2003). The important features of (considered at 380 ppm), ORYZA predicted a mean change
three General Circulation Models (GCMs) used such as of 7.20% in yields for every 1 8C increase in temperature,
the General Fluid Dynamics Laboratory (GFDL) Model, while INFOCROP predicted 6.66%. But increasing CO2
Goddard Institute of Space Studies (GISS) model and the concentration (700 ppm) resulted in increases of 30.37 and
United Kingdom Meteorological Office (UKMO) model are 56.37% in yield by ORYZA and INFOCROP, respectively.
provided in Table 3. However, with temperature increase of +4 8C above
The GCM scenarios were produced by applying the ratios ambient, the differences in the yield predictions by the
of precipitation or differences in temperature predicted for two models became remarkably small (Table 4).
the 1  CO2 (380 ppm) and 2  CO2 (760 ppm) simulations
to the baseline daily weather data set. The changes between 3.2. Effect of predicted GCM scenarios on rice yields
1  CO2 and 2  CO2 conditions were representative of the
differences between the present and the future climate The predicted changes in overall production for each site
scenarios following an equivalent doubling of CO2. under different climate scenarios using the two crop models
are provided in Tables 5a and 5b. In general, the ORYZA
2.2. Simulation model suggested the decreases of 7.63, 9.38 and
15.86% in yield for the GDFL, GISS and UKMO
For the simulation analysis, only runs that terminated scenarios, respectively. For the corresponding scenarios,
normally or by crop deaths as a result of temperature were INFOCROP indicated larger reductions at 9.02, 11.30
included. The relative yield changes under the scenarios, and 21.35%, respectively. When each site was analysed
referred to yields predicted for the current climate, were individually under three GISS and UKMO scenarios, almost
used in the analysis rather than absolute yields. To provide all the sites except one showed the declining trend in yields.
an estimate of the overall effect of climate change on rice The decreases in the yield of rice, when the mean of the
production under the three GCMs scenarios, the average corresponding values in Tables 5a and 5b were considered,

Table 3
The major features of the General Circulation Models used in this study
GFDL GISS UKMO
Source laboratory Geophysical Fluid Dynamics Goddard Institute for United Kingdom
Laboratory Space Studies Meteorological Office
Base 1  CO2 (ppm) 380 380 380
Change in global temperature (8C) +4.0 +4.2 +5.2
Change in global precipitation (%) 8 11 15

Please cite this article in press as: Krishnan, P. et al., Impact of elevated CO2 and temperature on rice yield and methods of adaptation as
evaluated by crop simulation studies, Agric. Ecosyst. Environ. (2007), doi:10.1016/j.agee.2007.01.019
AGEE-3025; No of Pages 10

P. Krishnan et al. / Agriculture, Ecosystems and Environment xxx (2007) xxx–xxx 5

Table 4
Mean predicted changea (%) in the potential yield under the ‘‘fixed’’ temperature and CO2 scenarios
CO2 concentration Temperature increments (8C) b
0 +1 +2 +3 +4 +5 Average
INFOCROP
380 0 2.68 18.04 28.15 36.36 42.49 20.39
400 5.23 2.90 12.03 23.21 33.05 40.32 18.12
500 25.92 13.90 3.60 12.32 25.08 35.11 4.85
600 40.61 29.39 16.74 1.04 15.80 29.17 7.14
700 56.37 41.21 27.96 9.85 7.67 23.15 17.43
Average 25.14 16.86 3.65 10.56 23.59 34.05 3.76
ORYZA
380 0 4.27 15.32 26.78 29.45 38.96 19.13
400 2.78 0.93 8.44 19.74 23.67 33.65 13.22
500 18.57 11.73 0.24 11.75 15.43 26.57 3.87
600 26.8 18.54 5.03 6.86 10.87 21.84 1.80
700 30.73 23.77 8.67 3.68 7.32 19.04 5.52
Average 16.27 10.14 1.96 13.76 17.35 28.01 5.78
a
Changes are averaged across all sites and at all the available years.
b
Temperature increments are above the current temperatures at each site.

Table 5a
Estimated changes in rice yield predicteda by the INFOCROP rice model for each observation site in the eastern India under the three GCM scenarios
Sites Rice yield GFDL GISS UKMO
(t/ha) Predicted Predicted Predicted Predicted Predicted Predicted
change (%) yield (t/ha) change (%) yield (t/ha) change (%) yield (t/ha)
Bhubaneswar 4.46 23.87 3.40 27.45 3.24 37.22 2.80
Chinsurah 5.18 7.03 4.82 7.38 4.80 8.11 4.76
Cuttack 4.93 25.44 3.68 27.67 3.57 40.87 2.92
Faizabad 4.72 13.55 4.08 17.65 3.89 28.34 3.38
Jabalpur 7.54 10.7 6.73 14.04 6.48 25.66 5.61
Jorhat 3.83 13.51 4.35 12.32 4.30 7.55 4.12
Kalyani 3.55 8.73 3.24 11.65 3.14 22.38 2.76
Pusa 3.82 3.74 3.68 4.35 3.65 5.26 3.62
Raipur 3.75 1.71 3.69 5.11 3.56 18.01 3.07
Ranchi 4.5 8.89 4.10 12.01 3.96 35.15 2.92
Average change (%) 4.63 9.02 4.18 11.50 4.06 21.35 3.60
a
Predicted rice yield is adjusted by the simulated changes in the experimental rice yield obtained.

Table 5b
Estimated changes in rice yield predicteda by the ORYZA1 model for each observation site in the eastern India under the three GCM scenarios
Sites Rice yield GFDL GISS UKMO
(t/ha) Predicted Predicted Predicted Predicted Predicted Predicted
change (%) yield (t/ha) change (%) yield (t/ha) change (%) yield (t/ha)
Bhubaneswar 4.46 17.33 3.69 20.36 3.55 27.53 3.23
Chinsurah 5.18 8.03 4.76 8.72 4.73 9.59 4.68
Cuttack 4.93 19.67 3.96 20.32 3.93 30.75 3.41
Faizabad 4.72 9.02 4.29 11.27 4.19 18.82 3.83
Jabalpur 7.54 11.05 6.71 14.08 6.48 21.05 5.95
Jorhat 3.83 12.13 4.29 12.64 4.31 8.31 4.15
Kalyani 3.55 7.75 3.27 9.76 3.20 16.51 2.96
Pusa 3.82 4.93 3.63 6.31 3.58 6.58 3.57
Raipur 3.75 2.79 3.65 5.22 3.55 10.09 3.37
Ranchi 4.50 7.87 4.15 10.35 4.03 25.98 3.33
Average change (%) 4.63 7.63 4.24 9.38 4.16 15.86 3.85
a
Predicted rice yield is adjusted by the simulated changes in the experimental rice yield obtained.

Please cite this article in press as: Krishnan, P. et al., Impact of elevated CO2 and temperature on rice yield and methods of adaptation as
evaluated by crop simulation studies, Agric. Ecosyst. Environ. (2007), doi:10.1016/j.agee.2007.01.019
AGEE-3025; No of Pages 10

6 P. Krishnan et al. / Agriculture, Ecosystems and Environment xxx (2007) xxx–xxx

suggested a significant decrease in rice yield in Cuttack,


while an increase in Jorhat under all three GCM scenarios.
The changes in yield components for both the sites are
provided in Table 6. In general, the number of spikelets
formed at both the locations was greater under the changed
climates. However, there was higher spikelet sterility at
Cuttack (>60%) than at Jorhat (13.25%). In addition, the
harvest indices at Cuttack decreased by about 25%, largely
due to the decreases in grain yield caused by supra-optimal
temperatures. On the contrary, there were only small
differences in harvest indices at Jorhat (Table 6).

3.4. Methods of adaptation for higher rice yield under


Fig. 2. Changes in monthly average temperature (8C) obtained from the climate change scenarios
daily weather data of each site during the main rice season (kharif) at
different sites in eastern India.
3.4.1. Adjustments in sowing date
Previous studies had suggested that adjusting sowing
varied significantly: Cuttack (27.45%), Bhubaneswar dates might be a simple and powerful tool for mitigating the
(25.63%), Ranchi (16.71%), Faizabad (16.44%), effects of a potential global warming (Baker and Allen,
Jabalpur (16.10%), Kalyani (12.80%), Chinsurah 1993). The potential outcomes by adjusting the sowing time
(8.14%), Raipur (7.16%) and Pusa (5.20%). On the in two sites (Cuttack and Jorhat) were examined by
contrary, there were increases in yield for Jorhat (11.08%). simulating the crop growth under different climate change
Variation in average temperature during the normal growing scenarios. Under the GCMs scenarios, temperature at the
season is shown in Fig. 2. All the sites had higher maximum time of flowering in the main season was found to be high,
average temperature at the time of sowing, followed by a and there were considerable variations when simulated for
decrease in average temperature during tillering stage and an different climate change scenarios under different sowing
increase during panicle initiation and flowering. The effect dates. Among the different sowing dates tested, the sowing
of high temperature on spikelet sterility seems to be limited on July 15 at Cuttack led to the yield changes of +6.6, +4.1
to time of flowering and grain filling. During the flowering and 9.8%, respectively, under the GFDL, GISS and
stage, the maximum temperature was about 34 8C in Cuttack UKMO model scenarios (Fig. 3). Interestingly, the sowing
and Bhubaneswar, around 33 8C in Jabalpur, 32 8C in on July 1 at Jorhat resulted in yield increases of +27.1, +24.3
Faizabad, Kalyani and Chinsurah, 30 8C in Raipur and Pusa, and +13.4%, respectively, for the corresponding scenarios
and 28 8C in Jorhat, which was the least of all. (Fig. 4). Any further delay in sowing at both the sites, which
had different dates for the maximum response, was not
3.3. Yield components at two contrasting sites beneficial in terms of crop yield (Figs. 3 and 4).

Among the different sites included for the present study, 3.4.2. Increased tolerance of spikelet fertility to
two sites such as Cuttack, which showed the maximum yield temperature
decrease, and Jorhat, which had the maximum increase in The hypothesis that high temperature induces spikelet
yield, were selected to examine their responses under injury was evaluated by enhancing the tolerance level in
different scenarios in detail. Interestingly, the results the crop models. The equation used in the ORYZA1 model

Table 6
Simulated yield and yield components for the current climate and the climate change scenarios of the three GCMs at Cuttack and Jorhat
Yield No. of spikelets No. of grains Filled grain 1000 grain Crop duration Total dry Harvest
(kg ha1) (m2) (m2) fraction (%) weight (g) (days) matter (kg ha1) index (%)
(i) Cuttack
Present 4930 29,984 26,779 89.31 18.41 110 10564 46.67
GFDL 3683 55,010 19,242 35.21 19.14 103 12191 30.21
GISS 3572 50,104 18,624 38.12 19.18 103 12653 28.23
UKMO 2924 56,637 14,850 26.48 19.69 102 11734 24.92
(ii) Jorhat
Present 3835 21,636 20,640 95.40 18.58 120 8351 45.92
GFDL 4357 25,945 22,401 87.12 19.45 105 9318 46.76
GISS 4398 25,900 22,600 87.53 19.46 105 9312 47.23
UKMO 4197 25,702 22,148 86.75 18.95 103 9493 44.21

Please cite this article in press as: Krishnan, P. et al., Impact of elevated CO2 and temperature on rice yield and methods of adaptation as
evaluated by crop simulation studies, Agric. Ecosyst. Environ. (2007), doi:10.1016/j.agee.2007.01.019
AGEE-3025; No of Pages 10

P. Krishnan et al. / Agriculture, Ecosystems and Environment xxx (2007) xxx–xxx 7

Fig. 5. Changes in yield (%) under the GCMs scenarios of the rice variety
Fig. 3. Changes in yield (%) under the GCMs scenarios of the rice variety IR 36 and IR 36 with improved temperature tolerance grown during the
IR 36 under different sowing dates grown during the kharif season at kharif season at Cuttack.
Cuttack.

to describe the response of spikelet fertility to temperature tolerance of the variety by not adjusting the value of Tmp,
is large decreases in yield due to spikelet sterility were pre-
dicted. But with the adaptation of variety by improved
100 temperature tolerance of the spikelet, the yield increased

½1 þ e0:853ðT max T mp Þ  higher than that of the current scenario level, at about +10.7,
+13.6 and 8.4, respectively, under the GFDL, GISS and
where d is the fertility percentage, Tmax the average daily UKMO model scenarios (Fig. 5).
maximum temperature (8C) during the flowering period and
Tmp the average daily maximum temperature (8C) at which
50% of the spikelets are fertile. For the indica variety, Tmp 4. Discussion
had a value of 36.5. To simulate the possible effect of an
increase in tolerance of spikelet to high temperatures, it was Both the crop simulation models predicted that any
assumed that this response was shifted by 2 8C by increasing increase in temperature at all the CO2 levels tested would
the value of Tmp to 38.5 8C. This adaptation in the spikelet cause declines in yields but an increase in CO2 level at each
trait was examined in Cuttack site. With the available temperature increment would increase yields. These results
weather data for this site, and with a constant sowing date corroborated with that of Bachelet et al. (1995). Summariz-
of June 15, a comparative study using the ORYZA1 model ing the data from several experimental studies on different
was made for the current climate and other GCM scenarios agricultural crops, Kimbal et al. (2002) found a 30%
as obtained by the GFDL, GISS and UKMO (Fig. 5). Under increase in growth rate with a doubling of CO2 levels, which
the GCMs scenarios, temperature at the time of flowering for was midway between the predicted values of the two models
the main season was already high. Without any temperature in the present study. Nevertheless, the experimental findings
from the growth chamber studies (Baker et al., 1992a,b)
showed a 32% increase in rice grain yield due to doubling of
the CO2 concentration from 330 to 660 mmol CO2 mol1 air
(ppm). The increased growth response with increasing CO2
concentration was attributed to greater tillering and more
grain-bearing panicles. The net assimilation rate and canopy
net photosynthesis also increased with increasing CO2
concentration. The elevated CO2 concentration was found to
accelerate the development but shorten the total growth
duration of rice.
There are many indirect influences of elevated CO2 on
rice growth and development. When photosynthesis is
enhanced by increased CO2, the C/N ratio also increases in
the plants, which can reduce the nutritional quality of leaves
and increase feeding by the herbivorous insects (Johnson
and Lincoln, 1990). There can be considerable changes in
Fig. 4. Changes in yield (%) under the GCMs scenarios of the rice variety the nutrient-cycling processes in soils also (Strain, 1985).
IR 36 under different sowing dates grown during the kharif season at Jorhat. Since the current crop growth simulation models have not

Please cite this article in press as: Krishnan, P. et al., Impact of elevated CO2 and temperature on rice yield and methods of adaptation as
evaluated by crop simulation studies, Agric. Ecosyst. Environ. (2007), doi:10.1016/j.agee.2007.01.019
AGEE-3025; No of Pages 10

8 P. Krishnan et al. / Agriculture, Ecosystems and Environment xxx (2007) xxx–xxx

taken these factors into account, there are still limitations on In both the models, the phenological phases are
their predictive value. characterized by the thermal time and day length. In the
The average yield changes of 8.23 and 7.31% by ORYZA1 model, the phenological development of the rice
ORYZA and INFOCROP, respectively, due to the effect of crop is divided into four main phases, namely Basic
temperature when simulated on per degree Celsius basis, Vegetative Phase (BVP), Photoperiod-Sensitive Phase
were comparable with that of 10% measured in the (PSP), Panicle Formation Phase (PFP) and Grain Filling
controlled environment experiments (Baker and Allen, Phase (GFP) (Kropff et al., 1994). In INFOCROP model, the
1993). The low responses at 400 ppm and 1 8C in both the phenological development is divided into three main phases,
models ORYZA (0.93%) and INFOCROP (2.90%) clearly namely sowing to seedling emergence, seedling emergence
show the positive effects of temperature increase, simulated to anthesis and storage organ filling phase. The seedling
by step-wise 1 8C increase with corresponding rise in CO2 to emergence to anthesis phase is further subdivided into three
400 ppm from the present ambient condition. However, the major sub-phases depending on the environmental factors
CO2 concentration of 700 ppm resulted in increases of about affecting them and the organs formed, namely basic juvenile
30.73 and 56.37% by ORYZA and INFOCROP, respectively. phase, PSP and storage organ formation phase (Aggarwal
In the light of recent experimental evidence (Kim et al., et al., 2004, 2006).
2003), these values appeared to be very high, probably For each crop model, the GFDL scenario was the most
because the simulation models predicted the crop yield benign and the UKMO the most severe, corresponding to
mathematically from either RUE or net photosynthesis. In the severity of temperature increases predicted by each GCM.
ORYZA1, the hyperbolic relationship between the max- The predictions across both crop models and the three GCM
imum rate of leaf photosynthesis at 1 g N/m2, and the scenarios indicated a 12.45% decline in the overall regional
external CO2 concentration during rice growth has been rice yield. Averaged across all three GCM scenarios, the mean
used. The rate of photosynthesis increased from change in yield predicted by INFOCROP to be 13.95% and
34 kg CO2 ha1 h1 at the 350 ppm CO2 concentration to by ORYZA to be 10.96%. Nevertheless, these values were
47 kg CO2 ha1 h1 at the 700 ppm CO2 concentration. The lesser than the average values for the scenarios in which
CO2 fertilization factor is applied in INFOCROP to reflect temperature and CO2 were varied at the fixed increments,
the direct physiological stimulation by elevated CO2 independently or in combination, above the current tempera-
concentration. When compared with the results from the ture for each site. It is likely that the GCM scenarios have
FACE experiments (Kim et al., 2003), the fertilization appropriate temperature corrections associated with the
effects used in these two models are probably overestimated. elevated CO2 concentration, resulting in a better predictive
When simulated for the climate change scenarios, the value compared to that of the scenarios with arbitrary
ORYZA model predicted changes of 7.63, 9.38 and combinations of elevated CO2 and temperatures.
15.86% for the GDFL, GISS and UKMO scenarios, Among the different sites tested, both the models pre-
respectively, and INFOCROP predicted changes of 9.02, dicted the maximum loss in yield at Cuttack (27.45%),
11.30 and 21.35%, respectively (Tables 5b and 5a). The while the maximum gain in yield was at Jorhat (+11.08%).
main cause for the differences in the predictions of the two These differences in yield predictions were mainly due to the
crop models was the way in which the leaf area development rice spikelet sterility at high temperature. The temperature at
and crop growth rate were calculated, and in the routines the time of flowering affects the spikelet fertility and hence
describing phenological events in the crop. In ORYZA1, the the yield (Krishnan and Surya Rao, 2005). The rice-growing
leaf area is calculated from leaf dry matter using the Specific sites such as Cuttack and Bhubaneswar (hot, moist
Leaf Area (SLA). LAI is simulated in the subroutine subhumid climate type) had high maximum temperature
SUBLAI. For a closed canopy, the LAI is calculated from of about 34 8C and minimum temperature of 25 8C during
the leaf dry weight using SLA. When the canopy is not the flowering period. Other sites such as Jabalpur, Faizabad
closed, the plants grow exponentially as a function of the and Ranchi (hot dry and moist subhumid type) had a high
temperature sum. The temperature sum is calculated using maximum temperature of about 31 8C and the minimum
the same procedure used to calculate the heat units for the temperature of about 21 8C, which were lower than that of
phenological development. The relative death rate of leaves Bhubaneswar and Cuttack. The low minimum temperature
is applied to the leaf weight to calculate the weight loss of probably helps to reduce the respiration at night. Likewise,
leaves. The reduction in leaf area is calculated from the loss the rice-growing sites such as Kalyani, Pusa, Raipur and
of leaf weight using SLA (Kropff et al., 1994). In Chinsurah (hot subhumid type) had a maximum temperature
INFOCROP, the Leaf Area Index changes proportionally of 30 8C and a minimum temperature of 20 8C during the
with Leaf Area Growth Rate (GLAI); its value is obtained by flowering period. But Jorhat (warm moist perhumid type)
multiplying the Increment in Leaf Weight (RWLVG) by the had the maximum temperature of about 28 8C and a
SLA. The Net Leaf Area Growth Rate (RLAI) was minimum temperature of 19 8C only, which probably
calculated based on the Initial Leaf Area Index (LAII), contributed to the benefits from the predicted effects of
GLAI, death rate of LAI (DLAI) and net loss of LAI due to climate change scenarios. The predicted declines in the
transplanting (PLTR) (Aggarwal et al., 2004). overall rice yield by both cop models for the GFDL, GISS

Please cite this article in press as: Krishnan, P. et al., Impact of elevated CO2 and temperature on rice yield and methods of adaptation as
evaluated by crop simulation studies, Agric. Ecosyst. Environ. (2007), doi:10.1016/j.agee.2007.01.019
AGEE-3025; No of Pages 10

P. Krishnan et al. / Agriculture, Ecosystems and Environment xxx (2007) xxx–xxx 9

and UKMO scenarios showed the need to increase the rice account while evaluating the model predictions about the
production than what was achieved at present. The effect of climate change on rice production. Further studies
differences in the yield predicted due to different scenarios on the adjustments to the management practices may help to
were largely due to the differences in the temperature at the offset any detrimental effects of climate change on rice
time of flowering (Fig. 2). production.
Adjustment of management practices may help to offset Some considerations are necessary, when interpreting
any detrimental effects of climate change on rice production. results from the scenarios predicted by the GCMs. The most
Probably the easiest adaptation is to adjust the sowing dates. significant limitations are their poor resolution, inadequate
Adjustment of sowing dates options was explored to coupling of atmospheric and oceanic processes, poor
investigate a suitable agronomic option for adaptation simulation of cloud processes and inadequate representation
under the future climate change scenarios. In this way, for of the biosphere and its feedbacks. The poor resolution is
the Cuttack site, the average yield changes of +6.6, +4.1 and likely to be significant in northeastern parts of India where
9.8% were predicted during July 15 sowing for the GFDL, the relief is varied and local climate may be quite different
GISS and UKMO scenarios, respectively, considerably from the average across the area used by a GCM. Most
higher than those of 19.67, 20.32 and 30.75% observed GCMs have difficulty in even describing the current climate
during June 15 sowing for the GFDL, GISS and UKMO adequately (Bachelet et al., 1995). The current GCMs are
scenarios, respectively. Likewise, the Jorhat site showed able to predict neither the changes in the variability of the
+27.1, +24.3 and +13.4% changes for the sowing on July 1 weather nor the frequency of catastrophic events such as
under the GFDL, GISS and UKMO model scenarios, hurricanes, floods or even the intensity of monsoons, all of
respectively; these changes were considerably higher than which can be important in determining crop yields as the
those of +12.13, +12.64 and +8.31% for the June 15 sowing average climatic data. It seems, therefore, that GCMs can at
for the corresponding scenarios. Further postponement in best be used to suggest the likely direction and rate of change
sowing did not improve the grain yield, probably due to low of future climates.
incident solar radiation and temperature. According to Long et al. (2005), fertilization effect of
There were striking differences in the predicted yield [CO2] has probably been overestimated, while omission of
changes among the three scenarios. The large differences [O3] effects from most models could have led to a 20%
were mainly due to the sensitivity of spikelet sterility to overestimation of crop production in the Northern Hemi-
temperature. Even a small difference of just 1 8C could result sphere. Database of chamber studies are the mechanistic
in a large yield decrease due to lower number of grains being basis for crop yield models. Hence, these models over-
formed (Sheehy et al., 2006). This was illustrated in the two estimate the yield gain due to elevated [CO2] compared to
examples used in the present study (Table 6). This was further those observed under fully open-air condition (FACE)
justified by the current changes in the temperature at different experiments in the field. The current FACE experiments are,
sites during the main cropping season (Fig. 2). The imposed however, not adequate enough to reparameterize the existing
climate change scenarios further enhanced this temperature models (Long et al., 2005). In a recent study, Bannyayan
effect. The increased spikelet sterility at Cuttack was mainly et al. (2005) evaluated ORYZA 2000 (Bouman and Van
due to the predicted increment in temperature above the Laar, 2006) against the observed growth and yield of rice in a
already high daily maximum temperature (34 8C) (Prasad 3-year field experiment in Japan where rice plants were
et al., 2006; Krishnan and Surya Rao, 2005) under the climate subjected to the elevated CO2 with FACE under varying N
change scenarios reaching levels (38 8C) where spikelet fertilizer rates. The simulation results showed that the model
damage was considerable. Although the predicted tempera- overestimated the increases in green leaf area indices due to
ture increments were similar to Cuttack, the lower average the elevated CO2 concentration but the enhancement of total
temperature (28 8C) at Jorhat was well below the level (30 8C) biomass was only a minor overestimation. While the model
at which spikelet fertility is affected. At both the locations, the was successful in simulating the increase in rice yield due to
number of spikelets formed was greater under the changed the CO2 enrichment, it failed to reproduce the observe
climates. This could be due to the enhanced growth rate of the interaction with N in the rice yield response to elevated CO2
crop between panicle initiation and flowering as the concentration. Thus, the lack of complete understanding of
consequences of fertilizing effect of higher CO2 level. the effects and the potential interactions of environment
The limitation on yields imposed by the increased variables on plant processes preludes the definitive
spikelet sterility can be largely overcome by the selection of predictions of the effects of global climate change.
genotypes that possess a higher potential of spikelet fertility Despite the limitations imposed by the assumptions made
at high temperatures. The fertilizing effect of increased in both the GCM and the crop simulation models, the current
atmospheric CO2 level is then likely to offset the changes in study provides significant progress in our understanding of
crop development rate brought about by the increased how future climates are likely to affect rice production in the
temperatures, so that significant yield increases may also be eastern India. The use of simulation models to predict the
obtained (Horie et al., 2000). Thus, the sensitivity of spikelet likely effects of climate change on crop production is an
sterility to temperature is a factor that must be taken into evolving process. Our study in view of the findings of the

Please cite this article in press as: Krishnan, P. et al., Impact of elevated CO2 and temperature on rice yield and methods of adaptation as
evaluated by crop simulation studies, Agric. Ecosyst. Environ. (2007), doi:10.1016/j.agee.2007.01.019
AGEE-3025; No of Pages 10

10 P. Krishnan et al. / Agriculture, Ecosystems and Environment xxx (2007) xxx–xxx

recent FACE studies clearly shows the need for modification Bouman, B.A.M., Van Laar, H.H., 2006. Description and evaluation of rice
of the existing models. Other levels of production such as the growth model ORYZA 2000 under nitrogen limited conditions. Agric.
Syst. 87, 249–273.
influences of water, nutrients and pests, and diseases due to Horie, T., Baskar, J.T., Nakagawa, H., 2000. Crop ecosystem responses to
climate change are to be included in the refined models. climate change: Rice. In: Reddy, K.R., Hodges, H.F. (Eds.), Climate
Some of these limitations in the use of present models can be change and Global crop productivity. CABI Publishing, Wallingford,
addressed so that increasingly more accurate predictions can Oxon, pp. 81–106.
Johnson, R.H., Lincoln, D.E., 1990. Sagebrush and grasshopper responses
be made in future.
to atmospheric carbon dioxide concentration. Oecologia 84, 103–110.
Karim, Z., Ahmed, M., Hussain, S.G., Rashid, Kh.B., 1994. Impact of
climate change on the production of modern rice in Bangladesh.
Acknowledgement Bangladesh Agricultural Research Council, Dhaka, Bangladesh.
Kim, H.Y., Lieffering, M., Kobayashi, K., Okada, M., Miura, S., 2003.
Seasonal changes in the effects of elevated CO2 on rice at three levels of
Our acknowledgements are to the collaborating members nitrogen supply: A free air CO2 enrichment (FACE) experiment. Global
from Orissa University of Agriculture & Technology Change Biol. 9, 826–837.
(OUAT), Bhubaneswar; Rice research Station (RRS), Kimbal, B.A., Kobayashi, K., Bindi, M., 2002. Responses of agricultural
Chinsurah; Narendra Deva University of Agriculture & crops to free air CO2 enrichment. Adv. Agron. 77, 293–368.
Krishnan, P., Surya Rao, A.V., 2005. Effects of genotypic and environmental
Technology (NDUAT), Faizabad; Jawaharlal Nehru Krishi
on seed yield and quality of rice. J. Agric. Sci. 143, 283–292.
Vishwa Vidyalaaya (JNKVV), Jabalpur; Assam Agricultural Kropff, M.J., Van Laar, H.H., Mathews, R.B., 1994. ORYZA1: An eco-
University (AAU), Jorhat; Bidan Chandra Krishi Viswa physiological model for irrigated rice production. In: SARP Research
Vidyalaya (BCKV), Kalyani, Rajendra Agricultural Uni- Proceedings, IRRI, Wagningen.
versity (RAU), Pusa; Indira Gandhi Agricultural University Lal, M., Singh, K.K., Rathore, L., Srinivasan, G., Saseendran, S.A., 1998.
(IGAU), Raipur; Birsa Agricultural University (BAU), Vulnerability of rice and wheat yields in NW India to future changes in
climate. Agric. Forest Meterol. 89, 101–114.
Ranchi; for providing the crop parameter and weather data Long, S.P., Ainsworth, E.A., Leakey, A.D.B., Morgan, P.B., 2005. Global
under the NATP RRPS 25 and Indian Meteorological food insecurity. Treatment of major food crops with elevated carbon
Department (IMD), Pune, for weather data. dioxide or ozone under large-scale fully open-air conditions suggests
recent models may have over estimated future yields. Phil. Trans. R.
Soc. B 360, 2011–2020.
Mall, R.K., Aggarwal, P.K., 2002. Climate change and rice yields in diverse
References agro-environments of India. I. Evaluation of impact assessment models.
Clim. Change 52, 315–330.
Aggarwal, P.K., 2003. Impact of climate change on Indian agriculture. J. Mathews, R., Wassmann, R., 2003. Modeling the impacts of climate change
Plant Biol. 30, 189–198. and methane emission reductions on rice production: A review. Eur. J.
Aggarwal, P.K., Kalra, N., Chander, S., Pathak, H., 2004. INFOCROP A Agron. 19, 573–598.
Generic Simulation model for annual crops in tropical environments. Peng, S.B., Huang, J.L., Sheehy, J.E., Laza, R.C., Visperas, R.M., Zhong,
IARI, New Delhi, p. 132. X., Centeno, G.S., Khush, G.S., Cassman, K.G., 2004. Rice yield
Aggarwal, P.K., Kalra, N., Chander, S., Pathak, H., 2006. InfoCrop: A decline with higher night temperature from global warming. Proc. Natl.
dynamic simulation model for the assessment of crop yields, losses due Acad. Sci. U.S.A. 101, 9971–9975.
to pests, and environmental impact of agro-ecosystems in tropical Penning de Vries, F.W.T., 1993. Rice production and climate change. In:
environments. I. Model description. Agric. Syst. 89 (1), 1–25. Penning de Vries, F.W.T., Teng, P., Metselaar, K. (Eds.), Systems
Bachelet, D., Kern, J., Tolg, M., 1995. Balancing the rice carbon budget in Approaches for Agricultural Development. Kluwer Academic Publish-
China using spatially-distributed data. Ecol. Model 79 (1/3), 167–177. ers, Dordrecht, The Netherlands, pp. 175–189.
Baker, J.T., Allen Jr., L.H., 1993. Effects of CO2 and temperature on rice: A Prasad, P.V.V., Boote, K.J., Allen Jr., L.H., Sheehy, J.E., Thomas, J.M.G.,
summary of five growing seasons. J. Agric. Meterol. 48 (5), 575–582. 2006. Species, ecotype and cultivar differences in spikelet fertility and
Baker, J.T., Allen Jr., L.H., Boote, K.J., 1992a. Effects of CO2 and harvest index of rice in response to high temperature stress. Field Crops
temperature on growth and yield of rice. J. Exp. Bot. 43, 959–964. Res. 95 (2/3), 398–411.
Baker, J.T., Allen Jr., L.H., Boote, K.J., 1992b. Response of rice to carbon Sheehy, J.E., Mitchell, P.L., Ferrer, A.B., 2006. Decline in rice grain yields
dioxide and temperature. Agr. Forest Meteorol. 60, 153–166. with temperature: Models and correlations can give different estimates.
Bannyayan, M., Kobayashi, K., Kim, H., Lieffering, M., Okada, M., Mirza, Field Crops Res. 98, 151–156.
S., 2005. Modelling the interactive effects of atmospheric CO2 and N on Strain, B.R., 1985. Physiological and ecological controls on carbon seques-
rice growth and yield. Field Crops Res. 93, 237–251. tering in terrestrial ecosystems. Biogeochemistry 1, 219–232.

Please cite this article in press as: Krishnan, P. et al., Impact of elevated CO2 and temperature on rice yield and methods of adaptation as
evaluated by crop simulation studies, Agric. Ecosyst. Environ. (2007), doi:10.1016/j.agee.2007.01.019