Effect of Elevated Air Temperature and Carbon Dioxide Levels on Dry Season Irrigated Rice Productivity in Bangladesh ()
1. Introduction
Bangladesh is a deltaic small country in South Asia with the 13th highest world population density and the population would be 202 million by 2050 [1] . All suitable lands are under cultivation, which is decreasing due to increasing demand for residential and industrial uses [2] . Besides, Bangladesh either enjoys food self-sufficiency or scarcity depending on natural calamities that are increasing in recent times [3] [4] [5] . Climate change associated sea level rise is expected to increase the risk for flooding and salinization of agricultural lands, especially near the southern coast [6] [7] . Increase of global mean surface temperatures for 2081-2100 relative to 1986-2005 is projected to be in the range of 0.3˚C to 4.8˚C depending on different Representative Concentrate Pathways [8] . Such changes will also be found in Bangladesh where average day temperature increase is likely to be 2.0˚C to 4.0˚C by 2100 [8] .
Throughout the last 150 years, atmospheric CO2 concentration has increased from 280 ppmv to 385 ppmv in 2008 (https://www.esrl.noaa.gov/gmd/ccgg/trends/) due to widespread human activities such as fossil fuel burning, cement production, and modified land-use patterns [9] [10] . At the current rate of atmospheric CO2 increase it will be double before 2100 causing dramatic effects on global and regional-scale climate. Rainfall has become increasingly variable and has demonstrated an uneven distribution. This erratic pattern produces extreme events, such as floods and drought, which have shown noticeable adverse effects on rice yields [11] [12] . As a result, rice production is likely to decline by 8% - 17% by 2050 [13] [14] .
Crop agriculture in Bangladesh is dominated by rice monoculture. Almost 80% of the total cropped area is planted with rice in different seasons, which accounts for more than 90% of total grain production [15] [16] . Among the rice growing seasons, dry season rice (Boro) contributed more than 53% of the total rice production in Bangladesh [15] . Food security is synonymous to rice security in Bangladesh. So, there is a crucial need to assess effects of climate change on rice productivity and economic growth of this country.
Some studies have recently investigated the economic effects of climate change on agricultural production in developing countries [17] - [23] . Although these studies showed susceptibility of crop agriculture to climate change, but the limited literatures on these are available for Bangladesh [24] [25] [26] [27] [28] . Ali [26] has discussed the influence of climate change by considering cyclones, storm surges, coastal erosion and backwater effects and reported large land losses in eastern part of Bangladesh because of beach erosion. Rashid & Islam [28] identified droughts, floods, soil salinity and cyclones as the major extreme climatic events that have affected agricultural production adversely. The CERES-Rice model and the DSSAT model was used to study the influence of higher air temperature and higher atmospheric CO2 concentration on rice yield [29] [30] [31] with old varieties that are not found in the field now a day. Moreover, future climate change scenario delineation has been changed also. Therefore, the objective of this study was to assess the impact of climate change on dry season irrigated rice (Boro) production because of high temperatures and CO2 concentrations in Bangladesh for futuristic adoption indication in agriculture.
2. Materials and Methods
2.1. Site Characterization
The study was conducted in six locations across the country of Bangladesh having diverse soil and weather conditions. The study locations were Gazipur (23˚45'N latitude, 90˚22'E longitude, 8.4 m above mean sea level [AMSL]), Rangpur (24˚41'N latitude, 89˚16'E longitude, 33.04 m AMSL), Rajshahi (24˚22'N latitude, 88˚22'E longitude, 17.24 m AMSL), Barisal (22˚41'N latitude, 90˚21'E longitude, 2.54 m AMSL), Comilla (23˚28'N latitude, 91˚09'E longitude, 6.54 m AMSL) and Habiganj (24˚25'N latitude, 91˚25'E longitude, 22.54 m AMSL) districts of Bangladesh.
2.2. Model Inputs Parameters
Selected input data used for CERES-Rice model are shown in Table 1.
2.2.1. Weather Data
Weather data for the study regions were collected from the Bangladesh Meteorological Department (BMD) for the period of 1981-2015. Base year daily average (to minimize unusual phenomenon) of maximum and minimum temperatures, rainfall and sunshine hours were calculated and created two successive
Table 1. Selected input data requirements for the CERES-Rice model (modified from [32] [33] ).
years’ weather file for DSSAT format, because Boro rice grows from November in one year and ends at April or May in the next year. The seasonal model simulation was run in SIMMETEO mode for 30 years to capture temperature and CO2 effects on yield and growth duration of selected varieties.
Extreme weather parameters like yearly number of cold spell duration indicator by taking annual count of days with at least 6 consecutive days when minimum temperature (<10th percentile) and number of warm spell indicator by taking annual count of days with at least 6 consecutive days when maximum temperature (>90th percentile) was estimated by using RClimDex v1.0 model software [34] . These indicators stated the cooler or warmer events of a particular area over the time spell.
2.2.2. Soil Parameters
Location-wise soil parameters used in DSSATv4.6 model is thickness of three layers, layer-wise sand, clay, bulk density, soil organic carbon and soil hydraulic characters. Soil pH, EC and slope also are required as inputs for different locations (Table 2). The terms “lower limit” and “drained upper limit” correspond to the permanent wilting point and field capacity, respectively [35] .
2.2.3. Crop Parameters
BRRI dhan28, BRRI dhan29 and BRRI dhan58 were used for grain yield and growth duration under varying levels of increased temperatures and CO2 concentrations. In all locations, selected varieties were tested against 15 November sowing, the optimum sowing date for Boro rice cultivation. In sowing date experiment, seeds were sown on 15 October to 30 December at 7 days interval to find out the optimum sowing date and to overcome the climatic effects for Boro rice growing.
Table 2. Physical and chemical properties of soils at the selected locations.
*WP-Wilting point, FC-field capacity, Ks-Saturated hydraulic conductivity, BD-Bulk density, OC-Organic carbon.
2.3. Yield Simulations
Potential yield is defined as the maximum yield of a variety restricted only by season-specific climatic conditions. This assumes that other inputs (nutrient, water, etc.) are non-limiting and cultural practices are optimal. Thus, the potential yield of a crop depends on the temporal variation in CO2 level in the atmosphere, solar radiation, maximum and minimum temperatures during the crop growing season and physiological characteristics of the variety. Mechanistic crop growth models are routinely used to estimate potential yield and assess the effects of climate change [36] [37] [38] [39] .
To simulate potential rice yield CERES-Rice v4.6 was used. This mechanistic model simulates crop growth and development processes, net photosynthesis based on radiation use efficiency, leaf area index, extinction coefficient and light absorption by the canopy [40] . The model can also simulate the effect of CO2 on photosynthesis and water use based on stomatal conductivity [41] .
Potential yields of BRRI dhan28, BRRI dhan29 and BRRI dhan58, the most popular varieties in dry season of Bangladesh, were simulated. Genetic coefficients used for those varieties are provided in Table 3. The coefficients were determined based on experiments by repeated iterations until a close match between simulated and observed phenology and yield was obtained. The data of field experiments were used for calibration and validation. There was a good agreement between simulated and observed phonological development and grain yield. The performance of the model has been well validated in Bangladesh [42] .
2.4. Selection of Scenarios for Model Applications
In future, the rise in temperature is likely to be 2˚C to 4˚C by 2100 in South Asia including Bangladesh [8] . So, the CERES-Rice model was applied for normal (no temperature rise) and 1˚C, 2˚C, 3˚C and 4˚C rise over normal weather condition.
Table 3. Genetic coefficients of rice varieties used in DSSAT model.
aGDD, growing degree days (˚C).
During initial growth stage, Boro rice plants grow under cool and dry conditions but exposed to hot environment in flowering and harvesting stages. Although Boro rice growing season is generally uneventful, few localized thunder showers take place sometimes. It is found that temperature stress and resultant crop damage is a significant reason for reductions in Boro rice yield, especially with delayed established crop. According to CMIP5 and Earth System Model, predicted CO2 concentrations will be reaching 421 ppm (RCP2.6), 538 ppm (RCP4.5), 670 ppm (RCP6.0), and 936 ppm (RCP 8.5) by the year 2100 [8] . So, we have selected those CO2 levels and compared with ambient CO2 concentration (380 ppm).
3. Results and Discussion
3.1. Grain Yields
Grain yields of selected rice varieties varied among regions mainly due to climatic variability and soil properties. In general, the highest potential grain yield was found in Rangpur and the lowest in Barisal region irrespective of varieties. Grain yield varied from 6.50 to 7.36, 7.29 to 8.08 and 6.84 to 7.80 t・ha−1 for BRRI dhan28, BRRI dhan29 and BRRI dhan58, respectively based on long-term (1981-2015) weather parameters (Figure 1). These variations are related with weather parameters and soil properties. Pathak [43] reported 7.7 to 10.7 t・ha−1 yield of wet season rice in Indo-Gangetic Plains (IGP). Aggarwal et al. [39] reported about 10 t・ha−1 rice yields in Punjub and Haryana. Mohandas et al. [44] also reported 10.5 t・ha−1 potential yield in Kapurthala district, Punjab but 7.1 t・ha−1 in Cuttack, Orissa. These clearly indicate that grain yield of rice varies across locations.
3.2. Weather Data and Its Extremes
Long-term (1981-2015) monthly mean weather data showed that mean maximum temperature varied from 23.0˚C to 36.1˚C for all the study regions. The
Figure 1. Grain yields of selected varieties under normalized weather conditions in different regions of Bangladesh.
lowest mean maximum temperature was found in January and the highest mean maximum temperature in April (Figure 2). The extreme high temperature (>35˚C) was recorded at Rajshahi location. Long-term mean minimum temperature varied from 10.4˚C to 25.0˚C. The lowest temperature was found at Rajshahi in January and the highest temperature at Barisal in May (Figure 2).
Climatic extreme events like five years’ total number of cold spell duration were in decreasing trends in most of the regions, except a slight deviation in Rangpur and Rajshahi areas (Table 4), where cold spell was decreasing in 1996-2000 and then in increasing trend. Five years total number of warm spell duration was in increasing trends in most cases except in Rangpur region (Table 4). Since trend in warm spell is increasing, Boro rice crop might be exposed to unusual temperature in future and thus reduced production is most likely. Boro rice starts its journey in cooler and drier conditions and continue grow up to vegetative. Its flowering, ripening and harvesting stages are exposed to hot summer weather. So, Boro rice faces winter extreme cool and warm spells in its growing periods. Initially crop growth is hampered because of cold spell, but temperature is increasing in most regions and if temperature goes beyond 35˚C during heading and flowering stages, sterility increased resulting in reduced grain yield. Generally, any temperature below 20˚C and above 30˚C cause yield reduction in rice [45] . Horie et al. [46] also reported that as the average temperature increases above 22˚C - 23˚C, rice yield declined linearly up to 30˚C and decline sharply thereafter. The initial linear decrease was due to the shorter crop
Figure 2. Long-term (1981-2015) monthly mean maximum and minimum temperature of the study regions during winter season. Error bar indicate the standard deviation.
Table 4. Number of cold and warm spells duration during 1981 to 2015 in the study regions of Bangladesh.
duration caused by increased temperature and the sharp decline after 30˚C was because of spikelet sterility from high temperature damage.
3.3. Effect of CO2 and Temperature on Yield
Grain yields decreased with increase in temperatures at 380, 421, 538, 670 and 936 ppm CO2 levels; although it improved with increasing CO2 levels for a specific temperature rise (Figures 3-5). Increase in grain yields were 2.6% - 2.8% for RCPs 2.6, 10.8% - 11.8% for RCPs 4.5 and 19.4% - 21.2% for RCPs 6.0 and 32.7% - 37.1% for RCPs 8.5 with varied CO2 levels. The highest yield increase was 40.3% with BRRI dhan28, 33.6% with BRRI dhan29 and 40.3% with BRRI dhan58 rice varieties at 936 ppm CO2. In all regions of Bangladesh, predicted mean yield reduction was 4.77%, 3.44% and 5.03% per degree rise in temperature for BRRI dhan28, BRRI dhan29 and BRRI dhan58, respectively at 380 ppm CO2 level. Mahmood [24] reported 9.7% - 22.7% yield reduction at Mymensingh and 7.3% - 17.0% at Barisal with BR3 depending on temperature rise. This indicates that rice production will be in decreasing trends in future if climate smart practices are not adopted. Grain yield reduction rates were higher with short and medium growth duration rice varieties compared to long duration BRRI dhan29. Among the study studied regions, grain yield reductions were the lowest (0.7% - 16.4%) in Barisal irrespective of temperature rise with all the tested varieties and CO2 levels (Tables 6-8). Since minimum temperature range is generally higher in Barisal regions during Boro season (Figure 2), attainable yield is low (Figure 1) compare to other tested regions and thus less grain yield reduction is likely because of temperature rise in future. Long duration variety (BRRI dhan29) had the lowest grain reduction (12.2% - 15.4%) compared to other tested varieties irrespective of regions and CO2 levels (Tables 6-8). Crop duration reduces with increased temperature, but its influence is more prominent with short duration variety. So, higher grain yield is expected in future with existing long duration varieties than others. The scenario could be changed in future with the development of cool variety. In Rangpur region, grain yield reduction with BRRI dhan29
was minimum (0.4% - 3%) because of temperature rise up to 2˚C. This part of the country generally enjoys cooler environment for a longer period in winter season and thus predicted temperature increase by 2˚C might not be hazardous for Boro rice production in future.
Grain yield reduction as predicted through the use of ORYZA1 and INFOCROP rice models with 1˚C increase in temperature was 7.20% and 6.66%, respectively at 380 ppm CO2 level [47] . Increased CO2 concentration up to 700 ppm, however, showed 30.73% and 56.37% higher yield by ORYZA1 and INFOCROP crop model, respectively. When temperature was increased by about 4˚C above the ambient level, ORYZA1 predicted yield reductions of 7.63%, 9.38% and 15.86%, respectively based on General Fluid Dynamics Laboratory (GFDL) Model, Goddard Institute of Space Studies (GISS) model and the United Kingdom Meteorological Office (UKMO) model; while INFOCROP predicted reductions of 9.02%, 11.30% and 21.35%. We have found less yield reduction than their findings. Moreover, IPCC [14] and BBS [48] estimated 8% - 17% decline in rice production by 2050 in South Asian region which is similar to our findings.
An increase in air CO2 level generally increased crop yield because of stimulated photosynthetic processes and improved water use efficiency [49] . In general, the effect of increased temperature would be negative because of increasing respiration [50] and a shortened vegetative and grain filling period [46] . Although major rice models indicate about 5% yield reduction for every degree rise in mean temperature [51] and Dias et al. [52] estimated 25% - 35% yield reduction in Sri Lanka using DSSAT. Basak et al. [31] reported 20% and 50% yield reduction by 2050 and 2070, respectively in Boro season with BR3 and BR14 rice varieties. These greater predicted yield reductions could be because of response behavior of studied varieties or use of inappropriate genetic coefficients in the model.
All the study regions had lower minimum average temperature at sowing (middle of November) and transplanting (January) time and it increased during tillering, panicle initiation and flowering stages. After nineties, cold spell at Rangpur and Rajshahi areas was is increasing trend that might have inflicted cold injury of Boro rice seedlings (Table 4). In Gazipur, Rajshahi and Habiganj
regions, warm spells were in increasing trend resulting in reduced rice production in those areas (Table 4). At flowering stage, maximum temperature was about 33.3˚C - 36.1˚C in Rajshahi, 31.9˚C - 33.6˚C in Gazipur, 30.4˚C - 31.7˚C in Rangpur, 32.4˚C - 33.6˚C in Barisal, 30.9˚C - 32.5˚C in Comilla and 31.7˚C - 32.9˚C in Habiganj regions and sometimes it goes beyond the critical level (>35˚C). High temperature generally prevails in March through May (Figure 2) and thus spikelet sterility and increased respiratory loss are most likely depending on growth stages of rice crops. Rice pollen viability and its production decline as daytime maximum temperature exceeds 33˚C and ceases beyond 40˚C [53] . In grain filling stage, kernel weight of rice decreases greatly below and above 24˚C temperature [54] . In the projected climate change scenarios, the maximum temperature at flowering stage of rice will be beyond the critical limit and thus reduction of rice yield is a must. To avoid this situation, shifting the sowing window of Boro rice and to develop cold and heat tolerant rice varieties may be the option for sustaining food security in Bangladesh.
3.4. Effects of CO2 and Temperature on Growth Duration
In all studied regions, growth duration of rice varieties was not significantly influenced by CO2 levels but it was reduced by 7.45, 7.49 and 7.88 days for BRRI dhan28, BRRI dhan29 and BRRI dhan58, respectively with every degree increase in temperature compared to ambient temperature (Figure 6 and Table 5). Mahmood [24] also reported 7 - 8 days reduced growth duration of BR3 per 1˚C rise in temperature. Biswas et al. [55] reported that both growth duration and grain yield of rice reduces depending on planting time. The lowest growth duration reduction was predicted for Barisal and Rajshahi regions for all temperature levels (Table 5). Since growth duration influences grain yield in most cases, reduced life span might be the cause of reduced yield in future.
Figure 6. Growth durations of BRRI dhan28, BRRI dhan29 and BRRI dhan58 as affected by temperature under different regions of Bangladesh; Error bars indicate the standard deviation.
Table 5. Growth duration reductions (days) with different varieties at varied regions of Bangladesh.
3.5. Combined Effects of Temperature and CO2 on Yield
At the ambient CO2 concentration, grain yield of rice is likely to be reduced by 16.4% - 21.3% for BRRI dhan28 (Table 6), 12.2% - 15.4% for BRRI dhan29 (Table 7) and 14.8% - 22.3% for BRRI dhan58 (Table 8) if temperature increases by 4˚C. But 936 ppm CO2 concentration (RCPs 8.5) might increase yield by 36.6% - 41.1% for BRRI dhan28 (Table 6), 32.3% - 33.6% for BRRI dhan29 (Table 7) and 33.1% - 41.1% for BRRI dhan58 (Table 8) in different regions of Bangladesh with existing temperature pattern. In most areas, BRRI dhan28 showed a competitive yield advantage under ambient conditions (no temperature rise and 380 ppm CO2) that extended up to 2˚C temperature rise with 538 ppm CO2 level except in southern Bangladesh. Grain yield reduction due to 3˚C temperature rise could be compensated up to 670 ppm CO2 level in all studied regions except in Rangpur areas. Similarly, if CO2 936 ppm level might be able to compensate grain yield reduction due to 4˚C rise in all test regions of Bangladesh (Table 6). BRRI dhan29 showed similar trends and are more tolerant to adverse conditions than BRRI dhan28. In Rangpur and Habiganj regions, grain yield reductions due to 3˚C rise were compensated with 538 ppm CO2 level; but in Gazipur, Rajshahi and Comilla regions, yield compensation is likely for 2˚C under same CO2 level. Grain yield reductions because of temperature rise up to 4˚C are likely to be compensated by the presence of 670 ppm CO2 for all the studied regions for BRRI dhan29 (Table 7). BRRI dhan58 showed more vulnerable in response to elevated temperature. In most of the regions, yield compensation is likely for 1˚C temperature rise with 538 ppm CO2 except Barisal and Comilla regions. Yield compensation could be extended for 3˚C temperature rise by 670 ppm CO2 level except in Rangpur areas and 936 ppm CO2 for 4˚C temperature rise in all test regions (Table 8). Baker et al. [49] and Figueiredo et al. [56] also reported that yield reduction due to higher temperature can be compensated with increased CO2 level. Horie et al. [57] reported that 100 μmol・mol−1 increased in CO2 concentration gave 7% - 8% greater rice yield in Japan, but 2˚C rise in temperature significantly reduced such beneficial effect.
Table 6. Percent changes in grain yield as affected by increased temperature and elevated CO2 concentration over control (no temperature rise, 380 ppm CO2) for BRRI dhan28 at various test regions of Bangladesh.
Table 7. Percent changes in grain yield as affected by increased temperature and elevated CO2 concentration over control (no temperature rise, 380 ppm CO2) for BRRI dhan29 at various test regions of Bangladesh.
Table 8. Percent changes in grain yield as affected by increased temperature and elevated CO2 concentrations over control (no temperature rise, 380 ppm CO2) for BRRI dhan58 at various test regions of Bangladesh.
4. Conclusion
This study was conducted to determine the effect of increased daily maximum and minimum temperatures and elevated CO2 levels using CERES-Rice model on grain yield and growth duration of dry season rice (Boro) at six representative locations across Bangladesh. Long-term normalized weather data were used to predict grain yields of those varieties under variable increased temperature and CO2 levels. Temperature increase rate considered was 0, +1˚C, +2˚C, +3˚C and +4˚C with the elevated CO2 concentrations of 380, 421, 538, 670 and 936 ppm based on different RCPs. Rice grain yield was reduced by 256 - 403 kg・ha−1 for BRRI dhan28, 172 - 370 kg・ha−1 for BRRI dhan29 and 268 - 432 kg・ha−1 for BRRI dhan58 per 1˚C temperature rise. On the contrary, CO2 concentration compensated yield by 225 - 275 kg・ha−1 for BRRI dhan28, 230 - 267 kg・ha−1 for BRRI dhan29 and 198 - 275 kg・ha−1 for BRRI dhan58 per 50 ppm CO2 rise. In general, grain yield reduction and compensation rate were lower in warmer region and higher in cooler region. Growth duration was reduced by about 9 days irrespective of locations and varieties with an exception for BRRI dhan58, which has comparatively less growth duration reduction per degree temperature rise. In the projected climate change scenarios, maximum temperature at flowering stage of rice might cross the critical limit and thus reduction in rice yield is expected. To avoid this situation, shifting of sowing window for Boro rice and to develop cold and heat tolerant rice varieties would be the options for sustaining food security in Bangladesh and similar environments in other parts of the world.
Acknowledgements
The authors acknowledge the Modeling Climate Change on Agriculture Project (CRP-II, KGF) for funding this study. This finding is the outcome of the collaborative research of BRRI, BARI and BSMRAU.