Article Request Page ASABE Journal Article
Integrated Assessment of Climate Change Impacts on Corn Yield in the U.S. Using a Crop Model
Published in Transactions of the ASABE 60(6): 2123-2136 (doi: 10.13031/trans.12314). Copyright 2017 American Society of Agricultural and Biological Engineers.
Submitted for review in February 2017 as manuscript number NRES 12314; approved for publication as part of the “Crop Modeling and Decision Support for Optimizing Use of Limited Water” collection by the Natural Resources & Environmental Systems Community of ASABE in July 2017.
The author is Kenichi Tatsumi, Associate Professor, Department of Environmental and Agricultural Engineering, Tokyo University of Agriculture and Technology, 3-5-8, Saiwai-cho, Fuchu, Tokyo, Japan; phone: +81 (42)367-5679; e-mail: firstname.lastname@example.org.
Abstract. A detailed analysis was conducted of the effects of climate change and increased carbon dioxide (CO2) concentrations on corn yield in the U.S. with a crop model using outputs from multiple general circulation models (multi-GCMs). Corn yield was simulated for 1999-2010, for the 2050s (average for 2041-2060), and for the 2070s (average for 2061-2080) under the representative concentration pathway 8.5 (RCP8.5) climate scenario. Results indicated a shortening of the growing period (GP), decreased water use efficiency (WUE) in almost all regions, and increased evapotranspiration (ET) during GP in almost all regions except for the southern U.S. Using multi-GCMs, the simulations under the RCP8.5 scenario resulted in negative effects of climate change on yield in almost all regions during both future periods. Especially strong negative impacts were reported south of latitude 40° N due to less optimal growing conditions. On the other hand, there were relatively smaller negative impacts in high-latitude regions (approximately north of latitude 40° N) due to more optimal growing conditions because of larger temperature changes compared to low-latitude and mid-latitude regions. Higher CO2 concentrations have the potential to increase corn yield. CO2 effects resulted in an approximately 0.04% to 0.05% increase in yield per 1 ppm increase in CO2 concentration under the RCP8.5 scenario, but the negative impacts of increased temperatures fully outweighed the CO2-fertilization effects.
Keywords.Climate change impacts, CO2 effects, Corn yield, Multiple GCMs, Uncertainty.
Many studies related to climate change impacts on agricultural productivity have been conducted since the early 1990s. Rising temperatures and changes in precipitation amount and pattern, the frequency of extreme hydrologic events, and atmospheric carbon dioxide (CO2) concentration can have significant impacts on agricultural productivity (e.g., Rosenzweig et al., 2014; Johnston et al., 2015).
In North America, Izaurralde et al. (2003) showed that corn yields would increase (on average 3% to 6% without CO2-fertilization effects and 16% to 18% with CO2-fertilization effects) for the Great Lakes, Corn Belt, and Northern Plains regions in 2090-2099 from baseline levels under HadCM2 climate scenarios. Reilly et al. (2003) showed that climate change would be beneficial to C3 and C4 crop productivity in the U.S., except for the southern U.S., and the benefits would increase in 2090 compared with 2030 for two climate scenarios even though temperature increases are quite large by 2090 in the Canadian Center Climate Model. Ko et al. (2012) indicated that the projected negative effects of rising temperatures on C3 and C4 crop production dominated any positive impacts of CO2 increases in the U.S. Central Great Plains and that simulated adaptation via changes in planting dates does not significantly mitigate the yield losses. Islam et al. (2012) found that the mean decrease in corn yield for four emission scenarios (SRES B1, A1B, A2, and a combination of all emission scenarios) ranged from 17.1% to 21.0% and from 20.7% to 27.7% during the 2050s and 2080s, respectively. Moreover, Long et al. (2006) pointed out that elevated CO2 levels enhance global crop yields ~50% less than in enclosure studies and cast serious doubt on projections that rising CO2 concentrations will fully offset losses due to climate change. Ultimately, quantitative assessments of climate change impacts on agricultural productivity have a degree of uncertainty, including contrasts in results arising from a range of differing crop and climate models (Rosenzweig et al., 2014). These facts indicate that research using models and model inputs characterized by less uncertainty are needed to accurately project the impacts of climate change on the production of important crop species, such as corn, wheat, and rice, for food security. A necessary approach for the generation of climate projections and the reduction of related uncertainties is to use the ensemble mean over multiple general circulation models (multi-GCMs) (Giorgi et al., 2008). However, only limited work has been carried out on simulations by optimized crop models using the latest multi-GCMs and its ensemble to estimate the uncertainty in climate change impact predictions.
Drawing from the above insights, the objectives of the present study are to (1) establish the impacts of climate change on corn yield in the U.S. for the 2050s (average of 2041-2060) and the 2070s (average of 2061-2080) under the representative concentration pathway 8.5 (RCP8.5) scenario across multi-GCMs, and (2) evaluate changes in corn yield in response to changes in CO2 concentration.
Materials and Methods
The Environmental Policy Integrated Climate (EPIC) model version 0810 (Williams, 1995) was used to simulate corn yields in the U.S. (23° to 50° N and 120° to 60° W). The EPIC model simulates crop growth in response to weather, soil, and crop management on a daily time step and has been continuously extended into a comprehensive soil-water-atmosphere-crop-management model. This model has been used in the U.S. and worldwide for various purposes, including climate change impacts on crops (e.g., Liu et al., 2007; Causarano et al., 2008; Chavas et al., 2009; Niu et al., 2009; Tingem and Rivington, 2009; Balkovic et al., 2013; Rosenzweig et al., 2014; Bhattarai et al., 2017). Radiation use efficiency (RUE) is modeled as being dependent on atmospheric CO2 concentration. RUE values are adjusted for ambient vapor pressure deficit and calculated by the daily intercepted photosynthetically active radiation to estimate potential biomass accumulation (Erickson, 1993). Potential biomass is primarily calculated from measurements of active radiation availability for photosynthesis, photosynthetic efficiency, and vapor pressure deficit on a daily basis and is adjusted to actual biomass through stress factors related to temperature, water, soil, aeration, and nutrients. In the present study, potential biomass for day i of the year is described as follows:
DM = daily potential increase in biomass (t ha-1 d-1)
PARelev = photosynthetically active radiation under elevated atmospheric CO2 (MJ m-2 d-1)
PARamb = photosynthetically active radiation under ambient atmospheric CO2 (MJ m-2 d-1)
BE = crop parameter for converting energy to biomass (kg MJ-1)
RUE = radiation use efficiency factor for converting energy to biomass (kg ha-1 (MJ m-2 d-1)-1)
VPD = vapor pressure deficit (kPa)
RA = solar radiation (MJ m-2 d-1)
LAI = leaf area index (-)
CO2 = atmospheric CO2 level (ppm)
a and b = crop parameters relating RUE and CO2 (a = 5.543 and b = 0.001442), which are calculated analytically as described by Stockle et al. (1992) and Leakey et al. (2006).
The relationship between RUE and CO2 using equation 3 is shown in figure A1 in the Appendix. Crop yield is estimated as the proportion of economic yield to total actual aboveground biomass at maturity, referred to as the harvest index. The final crop yield is adjusted according to the plant stress factors related to climate, soil properties, and crop management regime. Crop yield is estimated by using the harvest index and plant stress factors concept as follows:
AY = actual crop yield (t ha-1)
HI = harvest index (-)
WS = water stress (-)
TS = temperature stress (-)
NS = nutrient stress (-)
AS = aeration stress (-)
AE = plant water use (mm d-1)
EP = potential plant water evaporation (mm d-1)
TX = mean daily air temperature (°C)
Tb = base temperature for corn (8.0°C)
To = optimal temperature for corn (25.0°C)
SNS = scaling factor for nutrient stress: SNS = 200(Ua/Uo), where Ua is the actual N or P content of the corn (kg ha-1), and Uo is the optimal N or P content for the corn (kg ha-1)
SAT = saturation factor: SAT = 100.0(ST/PO – CAF) / (1.0 – CAF), where ST is the water content minus field capacity of the top soil (mm), PO is the porosity minus field capacity of the top soil (mm), and CAF is the critical aeration factor (CAF = 0.85).
The model computes evaporation from plants and soils separately. Plant water evaporation is simulated as a linear function of potential evaporation and LAI (Williams, 1995). Actual soil water evaporation is estimated by using exponential functions of soil depth and water content. The Penman-Monteith method is used for estimating plant water evaporation and the effects of CO2 changes in the present study and is expressed as follows:
Eo = potential evaporation (mm d-1)
RN = net radiation (MJ m-2 d-1)
d = slope of saturation vapor pressure curve (kPa °C-1)
AD = air density (kg m-3)
VPD = vapor pressure deficit (kPa): VPD = Ea(1.0 – RH), where Ea is saturation vapor pressure at mean air temperature (kPa), and RH is relative humidity (-)
U10 = mean wind speed at 10 m height (m s-1)
HV = latent heat of vaporization (MJ kg-1): HV = 2.501 – 0.0022TX (°C)
GMA = psychrometer constant (kPa °C-1)
Ep = potential plant water evaporation (mm d-1)
AR = aerodynamic resistance for heat and vapor transfer (s m-1)
CR = canopy resistance for vapor transfer (s m-1)
RA = solar radiation (MJ m-2 d-1)
AB = soil albedo (-)
RLO = net outgoing longwave radiation (MJ m-2 d-1): RLO = [0.34 – 0.14sqrt(ED)] × 4.9E-9TK4, where ED is vapor pressure at mean air temperature (kPa)
RAmax = clear-day radiation at the surface (MJ m-2 d-1)
TK = mean daily air temperature (K)
ZD = displacement height of the crop (m)
Z0 = surface roughness parameter (m)
CPheight = crop height (m)
SMLA = sum of LAI of the plant stand at the time
MSC = maximum stomatal conductance (0.0070 m s-1)
FVPD = fractional vapor pressure deficit (kPa)
VPth = threshold vapor pressure for the corn (0.50 kPa).
If soil water is limited, plant water evaporation is reduced as follows:
Ep' = actual plant water evaporation (mm d-1)
Eo' = adjusted potential evaporation (mm d-1).
RFI = intercepted rainfall (mm d-1).
Potential soil water evaporation for a layer is estimated by taking the difference between total potential soil water evaporation values at the layer boundaries:
SEV = potential soil evaporation for layer l (mm d-1).
EVZ = total potential soil water evaporation (mm d-1) from soil depth Z (m)
ES = potential soil water evaporation remaining after snow and litter evaporation (mm d-1).
The depth-distributed estimate of soil water evaporation is reduced if soil water is limited in a layer. The final step in adjusting the soil water evaporation estimate is to ensure that the soil water supply is adequate to meet the demand. The adjusted soil water evaporation (ASE, mm) is determined as follows:
SEVl' = adjusted soil water evaporation estimate (mm d-1)
ST = soil water content in the root zone (mm)
WP = wilting point water content (mm)
FC = field capacity water content (mm).
Detailed explanations of EPIC are beyond the scope of the present work. More technical information on EPIC and detailed descriptions of the model parameter optimization can be found in Texas (2016) and Tatsumi et al. (2016), respectively.
The automatic heat unit schedule (HUS) was determined as follows: (1) determination of heat units (HU) (calculated from daily temperatures above a base temperature threshold) and potential heat units (PHU) (calculated by accumulating daily temperature above a base temperature threshold, summed over all phenological stages under present planting operation); (2) calculation of HU/PHU values on current planting dates; (3) establishment of planting date based on HU/PHU values obtained from (2) under future climate conditions; and (4) harvesting date is set when HU/PHU equals 1.15 (Balkovic et al., 2013). It was assumed that HUS operation in a warmer climate will occur earlier compared to present climate conditions.
Present Meteorological Forcing Data
Simulation by the EPIC model requires six daily meteorological forcing datasets: minimum and maximum air temperature at 2 m above the ground surface, precipitation, solar radiation, relative humidity, and wind speed. These forcing datasets were obtained for the period 1999-2010 (baseline period) from a global meteorological forcing dataset for land surface modeling (Sheffield et al., 2006), which was constructed with a 0.25° spatial resolution by combining a suite of global observation-based datasets with NCEP/NCAR reanalysis.
Reported corn yields at the county level from 1999 to 2010 were obtained from the USDA National Agricultural Statistics Service (USDA, 2014). This dataset contains annual irrigated and rainfed corn yield for each county. Soil inputs and corresponding hydrological properties are required by the EPIC model in the process of simulating soil water dynamics. Soil physical and chemical property data, including bulk density, sand/silt/and clay contents, pH, sum of the bases (Ca + Mg + K + Na), base saturation, organic carbon concentration, calcium carbonate content, and cation exchange capacity, were obtained from the Harmonized World Soil Database (FAO, 2012).
The principal field management practices data required by the EPIC model include planting and harvesting dates; irrigation operation dates, amounts, and times; and fertilizer application dates, amounts, and times. Planting and harvesting dates data were obtained from the USDA at the state level (USDA, 2010). Irrigation schedules have a large impact on crop growth; however, the availability of detailed information relating to operation schedules is very limited. In this study, rainfed and irrigated regions were simulated separately using the annual harvested area of irrigated and rainfed corn obtained from Portmann et al. (2010). Irrigation occurred on days when the ratio of dry biomass produced to potential corn production, given adequate water, fell below 1. In other words, if the area equipped for irrigation, expressed as a ratio of the total area, was 1, then there was no water stress. Irrigation water was applied between the minimum (20 mm) and maximum (50 mm) volumes allowed for automatic irrigation in a single application. The amount and timing of applied nitrogen (N) and phosphorus (P) fertilizer for each grid cell during the growing period was determined by the RUSLE2 database, version 126.96.36.199 (USDA, 2016). The total amounts of N and P during the growing period never exceeded the values obtained from Mueller et al. (2012).
Model Parameter Optimization
The MOCOM-UA optimization method is based on the strengths of the complex shuffling strategy, competitive evolution, Pareto ranking, and the multi-objective downhill simplex search. Detailed descriptions of the MOCOM-UA algorithm, including multi-objective downhill simplex search (MOSIM) can be found in Yapo et al. (1998) and Tatsumi et al. (2016).
The use of more calibration parameters rather than fewer can contribute to a reduction in uncertainty and an increase in the number of the best attainable parameter values obtained, but it comes with a heavy computational cost. Therefore, efficient selection of calibration parameters is important to enable accurate modeling of corn yield with a reasonable computational load. Two parameters important for simulating corn growth were optimized using MOCOM-UA to match the behavior of the real agricultural system: biomass-energy ratio (BE) and harvest index (HI) (Xiong et al., 2014). The domains of these calibrated parameters and other key parameter and input values are given in table A1 in the Appendix.
Automatic optimization procedures typically need an objective criterion that is efficient in searching for the best parameter values for crop models to optimize a given objective function. In the present study, the targeted parameters were simultaneously calibrated using four objective functions to emphasize the minimization of errors of the average and the maximum yield over the simulation periods (1999-2010), and the inter-annual variation of simulated yield against reported yield, in a multi-objective context using MOCOM-UA. The objective functions were defined as follows:
where Savg and Oavg are the means of the simulated and reported annual corn yields over the simulation period, respectively, St and Ot are the yearly simulated and reported yields in the corresponding year t, respectively, and Smax and Omax are the maximum simulated and reported yields in the simulation period, respectively. The number of years included in the simulation (N) was 12 in the present study. The multi-objective problem for the optimization process was defined as follows:
where F(?) is the multi-objective vector, and ? denotes the optimized parameters. The predictive combination of the four parameters, which is defined as the minimum uncertainty in the parameters that can be achieved without stating a subjective relative preference for minimizing one specific competent of F(?) at the expense of another, was selected based on the minimization of F(?). In this study, the number of initial and final parameter sets were both set to 120, taking into account the computational load. The initial parameter set was randomly assigned across a generated population size of 120, with values drawn from a standard normal distribution. The detailed algorithm for the parameter optimization used in this study can be found in Tatsumi et al. (2016).
After the crop model was calibrated and validated for the U.S. region, crop yield simulations were implemented using each final parameter set and multi-GCMs for the 2050s and 2070s for future climate scenario.
Future Meteorological Forcing Data and Climate Change Scenario Generation
The climate change GCM projections were obtained from WorldClim - Global Climate Data (www.worldclim.org; Hijmans et al., 2005). Future meteorological forcing data for the 2050s and 2070s were obtained from WorldClim at 30 s spatial resolution. These data are based on Coupled Model Intercomparison Project Phase 5, and original GCM outputs assessed by the Intergovernmental Panel on Climate Change (IPCC) 5th Assessment Report (AR5) climate projections were downscaled and calibrated by bias correction as a baseline current climate. In the present study, emission paths RCP8.5 (the highest-emission scenario, seventeen GCM members) (table A2) were used to simulate the impacts of projected climate change on corn yield. Detailed descriptions of this scenario are given in van Vuuren et al. (2011).
Monthly average minimum and maximum temperature and monthly total precipitation anomalies from GCMs were calculated relative to the baseline period per Sheffield et al. (2006) and were applied to this baseline while conserving observed variability (Gerten et al., 2011; Waha et al., 2013). Linear interpolation was used to calculate daily minimum and maximum temperatures from monthly average minimum and maximum temperatures. Daily precipitation data were provided by a weather generator (Geng et al., 1986) that distributes monthly precipitation to the number of days with precipitation >10 mm d-1 and maximum wet spell duration (daily rainfall >0.1 mm) in a month, considering transition probabilities between wet and dry phases (Gerten et al., 2004; Waha et al., 2013). These precipitation indices were assumed constant at their average value from the baseline period. Geng et al. (1986) showed that the expected number of wet days and daily total precipitation from this weather generation method are generally very similar to observations in various regions. The Mountain Microclimate Simulation Model (MTCLIM) was used for estimating daily solar radiation and relative humidity (Thornton and Running, 1999; Thornton et al., 2000; Bohn et al., 2013).
To account for the effects of increased CO2 concentrations on corn yield, CO2 levels of 545 ppm (2050s) and 682 ppm (2070s) were used for the RCP8.5 scenario (Meinshausen et al., 2011). Simulations for the baseline period (1999-2010) and simulations without CO2 effects were run with a CO2 concentration of 380 ppm, the value for the year 2005.
The corn yield was calculated for each grid cell by weighting the proportion of the rainfed and irrigated crop area obtained from Portmann et al. (2010) as follows:
where Sj is the simulated corn yield in grid cell j, Sr,j is the rainfed yield in grid cell j, aj is the ratio of rainfed area to total crop area in grid cell j, Si,j is the irrigated yield in grid cell j, and N is the total number of grid cells (N = 2,586). Next, the county-level corn yield was estimated as a weighted average:
where is the county-level averaged corn yield for the cth county, Sc,j is the corn yield in grid cell j in county c, Ac,j is the corn-harvested area in grid cell j in county c, n is the total number of grid cells in county c, and m is the number of counties (m = 974).
The optimized ensemble mean RUE and HI values obtained using the final parameter sets are presented in figure A2. The difference in corn yield between the simulated ensemble mean yield obtained using the final model parameter sets and the average reported yield for the period 1999-2010 is presented in figure A3. The temporal ensemble mean simulated and reported yields for the period 1999-2010 are shown in figure A4. The spatial pattern of simulated yield is generally consistent with the pattern of reported yield, and the differences in yield are within ±2 t ha-1 in almost all regions. The spatial averaged RMSE is 1.91 t ha-1. However, the simulated yield tends to be underestimated in the Corn Belt, particularly in Illinois, Iowa, and Minnesota, and in the lower Mississippi River Valley in the southern U.S. On the other hand, yield is overestimated in areas with relatively lower reported yields in the eastern U.S. and western Corn Belt. The coefficient of determination between simulated and reported yields during the period 1999-2010 for each state is 0.36 (p < 0.05; two-tailed probability) (figs. A5 and A6). Although the reproducibility of the interannual variability of yield is low compared with the reproducibility of the long-term average yield, mainly due to lack of reliable information about fertilizer applications, soil, and agricultural management, these results demonstrate that, with the use of the converged optimal model parameter sets obtained using MOCOM-UA optimization, the EPIC model is sufficiently accurate to perform an analysis of the impact of climate change on corn yield at the continental scale.
Climate Change Impacts on Growing Period and Water Environment
Changes in temperature and precipitation compared to the baseline period have a significant impact on crop growth and physiological development (Muchow et al., 1990). The sim-ulation results show a decrease in growing period (GP) in almost all regions in both future periods under the RCP8.5 scenario (fig. 1a; table 1). GP changes from 138.2 days for the baseline period to 123.0 and 120.0 days for the 2050s and 2070s, respectively, averaged over areas currently under corn cultivation. The large predicted shift to shorter GP is especially remarkable in the northern part of the U.S. In the north-central portion of the U.S. Corn Belt, GP decreases by more than 20 days in both future periods.
Evapotranspiration (ET) during GP, which is derived using the Penman-Monteith equation, depends on temperature, humidity, solar radiation, wind, and total leaf area. In spite of a shortened GP due mostly to rising temperatures, the simulated ET during GP increases in almost all regions except for the southern U.S. under the RCP8.5 scenario (fig. 1b; table 1). Evapotranspiration increases by 29.4% in the 2050s and 40.1% in the 2070s without CO2 effects (table 1). When simulations were made considering CO2 effects, ET increases by 21.0% in the 2050s and 22.5% in the 2070s. The increase in ET is higher in mid-latitude and high-latitude regions than in low-latitude regions with and without CO2 ef-fects, which is a result of relatively large increases in minimum and maximum temperatures compared to the present climate condition. The presence of CO2 effects reduces the projected increase in ET by 6.5% (2050s) and 12.6% (2070s), mainly due to stomatal closure.
Table 1. Simulated evapotranspiration, water use efficiency, and growing period for the baseline (1999-2010) and future periods. Results are averaged across all corn-growing areas for each period. Terms in parentheses are spatial standard deviations. Parameter[a] Baseline RCP8.5 1999-2010 2050s 2070s Without CO2 effects ET 482.9 (93.1) 625.3 (106.1) 677.0 (119.1) WUE 17.1 (3.3) 11.5 (2.0) 9.7 (1.5) GP 138.2 (20.4) 123.0 (13.5) 120.0 (13.6) With CO2 effects ET 482.9 (93.1) 584.5 (99.8) 591.6 (104.0) WUE 17.1 (3.3) 13.6 (2.3) 12.8 (1.8) GP 138.2 (20.4) 123.0 (13.5) 120.0 (13.6)
[a] ET = evapotranspiration during growing season (mm), WUE = water use efficiency (defined as crop yield / ET, kg mm-1), and GP = growing period (days).
Under the RCP8.5 scenario, water use efficiency (WUE), which is defined as the crop yield per unit of ET during GP, decreases compared to WUE under present climate conditions, assuming no change in crop cultivars, because of increases in ET and decreases in yield (fig. 1c; table 1). Values of WUE are higher with CO2 effects than without CO2 effects because increased CO2 concentrations cause decreases in stomatal conductance and transpiration (table 1). Spatially averaged, WUE is 17.1 kg mm-1 under the baseline, WUE with CO2 effects is 13.6 kg mm-1 (2050s) and 12.8 kg mm-1 (2070s), and WUE without CO2 effects is 11.5 kg mm-1 (2050s) and 9.7 kg mm-1 (2070s) (table 1).
Yield Changes in Future Climate Scenarios with and without CO2 Effects
The results of the present study suggest that corn yields in almost all regions are sensitive to climate change and CO2 effects. Projections of U.S. relative yield changes differ substantially not only between both future periods but also between simulations with and without CO2 effects (fig. 2). Significant yield reductions are expected in the RCP8.5 scenario and in both periods in almost all regions, with these reductions somewhat offset by CO2 effects (figs. 2 and 3).
In the 2050s, future climate change across multi-GCMs without CO2 effects has a negative impact on yield, ranging from 10.0% to 32.4% (average 18.9%). On the other hand, when considering CO2 effects, decreases in yield vary from 1.8% to 25.3% (average 11.1%). In the 2070s, the decrease in ensemble corn yield is 26.3% without CO2 effects and 15.0% with CO2 effects. The negative impacts of climate change and the standard deviations of yield across multi-GCMs increase over time (fig. 2). Increased CO2 concentrations result in higher yields, but the negative impacts due to increased temperatures outweigh the CO2-fertilization effect.
Spatial Distribution of Change in Yield
When using a multi-GCM ensemble, the relative changes in simulated corn yield under the RCP8.5 scenario with and without CO2 effects are negative in most grid cells and exhibit a geographic distribution along broad latitudinal bands (figs. 3 and 4). The state-level corn yield decreases by 1.6% to 37.4% for the 2050s and by 5.7% to 43.5% for the 2070s relative to the baseline period without CO2 effects (table A3). Simulation results without CO2 effects indicate relatively strong negative effects of climate change on corn yield, especially with large temperature changes in the 2070s. When the CO2 effects are included, there is a smaller reduction in yield compared to the effects of climate change alone. Although the yield decreases by as much as 20% throughout most of the corn-growing region, especially in low-latitude and mid-latitude regions, CO2 effects mostly offset the negative impacts of climate change in some high-latitude regions (fig. 4).
For the RCP8.5 climate scenario, the most significant negative changes in yield are found south of latitude 40° N, a region with life-cycle shortening and increasing temperature stress. The relatively lower yield decreases in high-latitude regions, most likely due to a reduced occurrence of low-temperature stress during GP under warmer-than-current conditions. Conversely, low-latitude and mid-latitude regions tend to experience greater reductions in yield due to higher temperatures. Moreover, significant temperature stress occurs in almost all regions for the 2070s.
The primary conclusion of this study was that the GP is shortened under climate change. The cause of the decline in WUE (table 1) is mainly the increase in evapotranspiration due to a hotter environment. Higher temperatures in the future tend to shorten phenology duration and decrease corn yield, mainly because of a shorter growing period that decreases the duration of photosynthesis. In this study, yield decreases were demonstrated for the 2050s and 2070s under the RCP8.5 scenario because the rising temperature caused a more rapid temperature accumulation, hastening phenological development, creating greater evaporative demand, and shortening the time to crop maturity. The average corn yield would decrease compared to present conditions if the planting date did not change under the RCP8.5 scenario. Corn growers in the U.S. can respond to climate change by the advancing planting date, and a shift in preferred planting from April-May to March-April may occur. Based on the simulated outcomes, advancement of the planting date by two weeks was shown to mitigate the reductions in yield of 8% in the 2050s and 23% in the 2070s under the RCP8.5 scenario with and without CO2 effects, in comparison with the results obtained with the current planting date. Advancing the planting date by two weeks was also shown to lead to higher crop-available water, with 8% in the 2050s and 12% in the 2070s under the RCP8.5 scenario, in comparison with the results obtained with the current planting date. Moreover, corn-growing regions would change with future climate change, possibly shifting northward in response to less favorable conditions at the southern boundary.
Some discussion of the implications of this adaption for total corn yield production would be helpful. After all, a shift in the life cycle of a crop cultivar by moving northward or by genetic modification (e.g., new varieties with heat stress tolerance, and longer-maturing cultivars) and advancing the planting date can help focus efforts on the decrease in evapotranspiration and prevent excessive reduction of the cultivation period. Adaptation measures in the agriculture sector that address food insecurity need to increase productivity, as well as avoid yield reduction. This approach will help farmers adapt to climate change and increase their corn yields. In particular, for an existing cultivar with a longer cycle, moving it northward can be easily done for most locations and is one of the most promising potential adaptations.
Uncertainty of Changes in Yield
Increasing temperatures shorten the plant growing cycle and the duration of the reproductive phase, causing a reduction in corn yield (Muchow et al., 1990). Muchow et al. (1990) showed that simulated yields in the central Corn Belt decrease by 5% to 8% per 2°C temperature increase. Runge (1968) reported that observed corn yields are affected by interactions between maximum daily temperature and rainfall from 25 days before to 15 days after anthesis. Izaurralde et al. (2003) indicated that dryland corn yields in the Corn Belt are predicted to increase by about 8.3% in the period around 2095 with CO2 effects under HadCM2 scenarios. Including the historical study described at the beginning of this article (Izaurralde et al., 2003), some previous research is not consistent with the findings of the present study. Therefore, multiple GCM outputs are needed to accurately conduct crop yield simulations.
In the present study, the standard deviations of the change in yield over all seventeen GCMs, which indicate the degree of uncertainty in the multi-GCMs, increase over time with and without CO2 effects. In other words, although the magnitude of the change in yield varies widely across all GCMs, the probability that yield will decrease is relatively high (fig. 5). When the effect of elevated CO2 is considered, the probability of a yield decrease (increase) in high-latitude regions decreases (increases) compared to not taking into consideration CO2 effect (fig. 5). GCM outputs have a wide range of uncertainty; therefore, studies of estimated climate change impacts using a single GCM or a few GCMs depend strongly on the model structure, scenarios, and initial conditions included in the GCM or GCMs (Rosenzweig et al., 2014). For example, even if a yield increase is predicted using a single GCM, the use of different GCMs may predict a decrease in yield. Therefore, yield simulations using multi-GCMs, as well as the optimization of model parameters, can help in the comprehensive consideration of uncertainties in future climate projections. Moreover, using multiple crop models will produce a more robust prediction of potential climate change impacts because there is considerable uncertainty in crop model predictions, especially in relation to the effects of climate change.
Yield Responses to CO2
The quantitative effects of rising CO2 concentrations on wild plant growth remain unclear (Reich et al., 2014). There are inconsistent reports on the effects of elevated CO2 on corn yield, varying from little positive effect (Ghannoum et al., 2000; Leakey et al., 2004), to no effect (Kim et al., 2007), or a yield increase of 50% (Prins et al., 2007; Vanaja et al., 2015). In general, the effects of higher CO2 concentrations tend to counteract the negative impacts of rising temperature and reduced soil moisture (e.g., Hatfield et al., 2011; Lobell and Gourdji, 2012). Higher CO2 concentrations reduce the stomatal openings of both C3 and C4 crops and transpiration per unit leaf area while enhancing photosynthesis. As a result of these complex interactions, higher CO2 concentrations would have the benefit of reducing stomatal conductance (Lammertsma et al., 2011), thereby increasing WUE by reducing transpirational water loss of crops (table 1). Moreover, CO2-fertilization effects may reduce the nutritional quality of crops, reduce nitrate assimilation, and lower the protein concentration (Taub et al., 2008). Parry et al. (2004) showed that corn yield increases by approximately 0.02% to 0.03% for a 1 ppm increase in CO2 concentration. Using values for C3 grains, Lobell and Gourdji (2012) suggested that a likely value for the change in yield per 1 ppm change in CO2 concentration is 0.07%, and the plausible range is 0.05% to 0.09% based on an extensive review of previous simulation studies. In the present study, CO2 effects cause an approximately 0.04% to 0.05% increase in yield per 1 ppm increase in CO2 concentration under the RCP8.5 scenario between the present time and the future.
The projections of future corn yields in the present study used a model that accepted that rising CO2 will directly affect photosynthesis and, therefore, may likely be overly op-timistic. These results, such as the positive effects of higher CO2 concentration, are consistent with previous modeling research, although the reported results of CO2 effects in the scientific literature vary widely because of differences among models in their assumed parameters. A variety of techniques can assess CO2 effects on crop yield, but all of these techniques have considerable uncertainty in their results. Model parameterization, especially the RUE-CO2 curve and ET-CO2 response, would have to be updated to account for uncertainty about the corn response to CO2. Therefore, experimental and theoretical research related to CO2 effects on crop growth is important for further advancement of crop modeling.
Limitations of the Present Study
Generally, the accuracy and reliability of the simulation results obtained in the present study depend heavily on the accuracy of the model input parameters. However, this study does not consider fertilizer management, farmers’ production decisions, and future technology improvements related to breeding and agronomy. In reality, field operations and production methods depend on individual farmers, and their management decisions, in turn, depend on several factors, such as weather conditions, food supply and demand, market forces, and traditional culture. In addition, uncertainties about the further mechanization of corn farming and other labor-saving developments also contribute to the uncertainty of the present results. Assumptions made for the model simulations in the present study could cause differences between simulated and reported yields. Uncertainties still exist regarding the effects of climate change, CO2, and technology development, as well as their interactions, on crop yield. Nonetheless, the analysis by the optimized model in the pre-sent study provides better information than previously available to analyze the effects of climate change and CO2 on corn yields, and it enables discussions of these yields in regard to food security and production within the conceptual framework of future climate scenarios.
This detailed analysis of the effects of climate change impacts and increasing CO2 concentrations on corn yield in the U.S. by a crop model using multi-GCMs reached the following conclusions:
- Simulations for the RCP8.5 future climate scenario resulted in the growing period (GP) shortening by 15.2 days in the 2050s and by 18.2 days in the 2070s compared to the baseline period. Despite the shortened GP, which was caused by higher minimum and maximum temperatures, evapotranspiration (ET) during the GP substantially increased in almost all regions, except for the southern U.S., and the magnitude of the ET increase was smaller with CO2 effects than without CO2 effects. Water use efficiency (WUE) decreased compared to present climate conditions due to increases in evapotranspiration and decreases in yield, and the largest decrease in WUE was for the 2070s without CO2 effects compared to the other scenarios and periods.
- Results for simulations with multi-GCMs indicated negative effects of climate change on yield in almost all regions under the RCP8.5 scenario and for both periods. Especially strong negative effects were predicted for regions south of latitude 40° N due to less optimal growing conditions compared to the baseline period. Conversely, smaller negative impacts were predicted for high-latitude regions due to better growing conditions caused by larger temperature changes compared to low-latitude and mid-latitude regions. However, temperature stress was notably present in almost all regions for the 2070s under the RCP8.5 scenario. Future CO2 concentrations increased corn yield compared to the yield predicted without CO2 effects, but the negative impacts of increased temperature fully outweighed the increases by the CO2-fertilization effect in almost all regions.
Projections of future climate change are highly uncertain. Therefore, the use of multi-GCMs for assessments of climate change impacts is essential not only for considering particular scenarios and bounding-case scenarios, but also for conducting a synthetic study of uncertainties in GCMs.
Balkovic, J., van der Velde, M., Schmid, E., Skalsky, R., Khabarov, N., Obersteiner, M., ... Xiong, W. (2013). Pan-European crop modelling with EPIC: Implementation, up-scaling, and regional crop yield validation. Agric. Syst., 120, 61-75. http://dx.doi.org/10.1016/j.agsy.2013.05.008
Bhattarai, M. D., Secchi, S., & Schoof, J. (2017). Projecting corn and soybeans yields under climate change in a Corn Belt watershed. Agric. Syst., 152, 90-99. https://doi.org/10.1016/j.agsy.2016.12.013
Bohn, T. J., Livneh, B., Oyler, J. W., Running, S. W., Nijssen, B., & Lettenmaier, D. P. (2013). Global evaluation of MTCLIM and related algorithms for forcing of ecological and hydrological models. Agric. Forest Meteorol., 176, 38-49. https://doi.org/10.1016/j.agrformet.2013.03.003
Causarano, H. J., Doraiswamy, P. C., McCarty, G. W., Hatfield, J. L., Milak, S., & Stern, A. J. (2008). EPIC modeling of soil organic carbon sequestration in croplands of Iowa. J. Environ. Qual., 37(4), 1345-1353. https://doi.org/10.2134/jeq2007.0277
Chavas, D. R., Izaurralde, R. C., Thomson, A. M., & Gao, X. (2009). Long-term climate change impacts on agricultural productivity in eastern China. Agric. Forest Meteorol., 149(6), 1118-1128. http://dx.doi.org/10.1016/j.agrformet.2009.02.001
Erickson, J. D. (1993). From ecology to economics: The case against CO2 fertilization. Ecol. Econ., 8(2), 157-175. http://dx.doi.org/10.1016/0921-8009(93)90042-5
FAO. (2012). Harmonized world soil database. Ver. 1.2. Rome, Italy: United Nations FAO. Retrieved from http://webarchive.iiasa.ac.at/Research/LUC/External-World-soil-database/HTML
Geng, S. (1986). A simple method for generating daily rainfall data. Agric. Forest Meteorol., 36(4), 363-376. http://dx.doi.org/10.1016/0168-1923(86)90014-6
Gerten, D., Heinke, J., Hoff, H., Biemans, H., Fader, M., & Waha, K. (2011). Global water availability and requirements for future food production. J. Hydrometeorol., 12(5), 885-899. https://doi.org/10.1175/2011jhm1328.1
Gerten, D., Schaphoff, S., Haberlandt, U., Lucht, W., & Sitch, S. (2004). Terrestrial vegetation and water balance-hydrological evaluation of a dynamic global vegetation model. J. Hydrol., 286(1), 249-270. http://dx.doi.org/10.1016/j.jhydrol.2003.09.029
Ghannoum, O., Von Caemmerer, S., Ziska, L. H., & Conroy, J. P. (2000). The growth response of C4 plants to rising atmospheric CO2 partial pressure: A reassessment. Plant Cell Environ., 23(9), 931-942. https://doi.org/10.1046/j.1365-3040.2000.00609.x
Giorgi, F., Diffenbaugh, N. S., Gao, X. J., Coppola, E., Dash, S. K., Frumento, O., ... Zakey, A. S. (2008). The regional climate change hyper-matrix framework. Eos, 89(45), 445-446. https://doi.org/10.1029/2008EO450001
Hatfield, J. L., Boote, K. J., Kimball, B. A., Ziska, L. H., Izaurralde, R. C., Ort, D., ... Wolfe, D. (2011). Climate impacts on agriculture: Implications for crop production. Agron. J., 103(2), 351-370. https://doi.org/10.2134/agronj2010.0303
Hijmans, R. J., Cameron, S. E., Parra, J. L., Jones, P. G., & Jarvis, A. (2005). Very high resolution interpolated climate surfaces for global land areas. Intl. J. Climatol., 25(15), 1965-1978. https://doi.org/10.1002/joc.1276
Islam, A., Ahuja, L. R., Garcia, L. A., Ma, L., Saseendran, A. S., & Trout, T. J. (2012). Modeling the impacts of climate change on irrigated corn production in the central Great Plains. Agric. Water Mgmt., 110, 94-108. http://dx.doi.org/10.1016/j.agwat.2012.04.004
Izaurralde, R. C., Rosenberg, N. J., Brown, R. A., & Thomson, A. M. (2003). Integrated assessment of Hadley Center (HadCM2) climate-change impacts on agricultural productivity and irrigation water supply in the conterminous United States: Part II. Regional agricultural production in 2030 and 2095. Agric. Forest Meteorol., 117(1), 97-122. http://dx.doi.org/10.1016/S0168-1923(03)00024-8
Johnston, R. Z., Sandefur, H. N., Bandekar, P., Matlock, M. D., Haggard, B. E., & Thoma, G. (2015). Predicting changes in yield and water use in the production of corn in the United States under climate change scenarios. Ecol. Eng., 82, 555-565. http://dx.doi.org/10.1016/j.ecoleng.2015.05.021
Kim, S.-H., Gitz, D. C., Sicher, R. C., Baker, J. T., Timlin, D. J., & Reddy, V. R. (2007). Temperature dependence of growth, development, and photosynthesis in maize under elevated CO2. Environ. Exp. Bot., 61, 224-236. http://dx.doi.org/10.1016/j.envexpbot.2007.06.005
Kiniry, J. R., Major, D. J., Izaurralde, R. C., Williams, J. R., Gassman, P. W., Morrison, M., Bergentine, R., & Zentner, R. P. (1995). EPIC model parameters for cereal, oilseed, and forage crops in the northern Great Plains region. Canadian J. Plant Sci., 75(3), 679-688. http://dx.doi.org/10.4141/cjps95-114
Ko, J., Ahuja, L. R., Saseendran, S. A., Green, T. R., Ma, L., Nielsen, D. C., & Walthall, C. L. (2012). Climate change impacts on dryland cropping systems in the central Great Plains, USA. Climatic Change, 111(2), 445-472. https://doi.org/10.1007/s10584-011-0175-9
Lammertsma, E. I., de Boer, H. J., Dekker, S. C., Dilcher, D. L., Lotter, A. F., & Wagner-Cremer, F. (2011). Global CO2 rise leads to reduced maximum stomatal conductance in Florida vegetation. Proc. Natl. Acad. Sci., 108(10), 4035-4040. https://doi.org/10.1073/pnas.1100371108
Leakey, A. D. B, Bernacchi, C. J., Dohleman, W. F. G., Ort, D. R., & Long, S. P. (2004). Will photosynthesis of maize (Zea mays) in the U.S. Corn Belt increase in future [CO2] rich atmospheres? An analysis of diurnal courses of CO2 uptake under free-air concentration enrichment (FACE). Global Change Biol., 10(6), 951-962. https://doi.org/10.1111/j.1529-8817.2003.00767.x
Leakey, A. D. B., Uribelarrea, M., Ainsworth, E. A., Naidu, S. L., Rogers, A., Ort, D. R., & Long, S. P. (2006). Photosynthesis, productivity, and yield of maize are not affected by open-air elevation of CO2 concentration in the absence of drought. Plant Physiol., 140(2), 779-790. https://doi.org/10.1104/pp.105.073957
Liu, J., Williams, J. R., Zehnder, A. J. B., & Yang, H. (2007). GEPIC: Modelling wheat yield and crop water productivity with high resolution on a global scale. Agric. Syst., 94(2), 478-493. http://dx.doi.org/10.1016/j.agsy.2006.11.019
Lobell, D. B., & Gourdji, S. M. (2012). The influence of climate change on global crop productivity. Plant Physiol., 160(4), 1686-1697. https://doi.org/10.1104/pp.112.208298
Long, S. P., Ainsworth, E. A., Leakey, A. D. B., Nosberger, J., & Ort, D. R. (2006). Food for thought: Lower-than-expected crop yield stimulation with rising CO2 concentrations. Science, 312(5782), 1918-1921. https://doi.org/10.1126/science.1114722
Meinshausen, M., Smith, S. J., Calvin, K., Daniel, J. S., Kainuma, M. L., Lamarque, J.-F., ... van Vuuren, D. P. P. (2011). The RCP greenhouse gas concentrations and their extensions from 1765 to 2300. Climatic Change, 109(1), 213. https://doi.org/10.1007/s10584-011-0156-z
Muchow, R. C., Sinclair, T. R., & Bennett, J. M. (1990). Temperature and solar radiation effects on potential maize yield across locations. Agron. J., 82(2), 338-343. https://doi.org/10.2134/agronj1990.00021962008200020033x
Mueller, N. D., Gerber, J. S., Johnston, M., Ray, D. K., Ramankutty, N., & Foley, J. A. (2012). Closing yield gaps through nutrient and water management. Nature, 490(7419), 254-257. https://doi.org/10.1038/nature11420
Niu, X., Easterling, W., Hays, C. J., Jacobs, A., & Mearns, L. (2009). Reliability and input-data induced uncertainty of the EPIC model to estimate climate change impact on sorghum yields in the U.S. Great Plains. Agric. Ecosyst. Environ., 129(1), 268-276. http://dx.doi.org/10.1016/j.agee.2008.09.012
Parry, M. L., Rosenzweig, C., Iglesias, A., Livermore, M., & Fischer, G. (2004). Effects of climate change on global food production under SRES emissions and socio-economic scenarios. Global Environ. Change, 14(1), 53-67. http://dx.doi.org/10.1016/j.gloenvcha.2003.10.008
Portmann, F. T., Siebert, S., & Doll, P. (2010). MIRCA2000 - Global monthly irrigated and rainfed crop areas around the year 2000: A new high-resolution data set for agricultural and hydrological modeling. Global Biogeochem. Cycles, 24(1). https://doi.org/10.1029/2008GB003435
Prins, A., Verrier, P., Kunert, K. J., & Foyer, C. H. (2007). Acclimation of the maize transcriptome to CO2 enrichment. South African J. Bot., 73(2), 307-308. http://dx.doi.org/10.1016/j.sajb.2007.02.103
Reich, P. B., Hobbie, S. E., & Lee, T. D. (2014). Plant growth enhancement by elevated CO2 eliminated by joint water and nitrogen limitation. Nature Geosci., 7(12), 920-924. https://doi.org/10.1038/ngeo2284
Reilly, J., Tubiello, F., McCarl, B., Abler, D., Darwin, R., Fuglie, K., ... Rosenzweig, C. (2003). U.S. agriculture and climate change: New results. Climatic Change, 57(1), 43-67. https://doi.org/10.1023/a:1022103315424
Rosenzweig, C., Elliott, J., Deryng, D., Ruane, A. C., Muller, C., Arneth, A., ... Jones, J. W. (2014). Assessing agricultural risks of climate change in the 21st century in a global gridded crop model intercomparison. Proc. Natl. Acad. Sci., 111(9), 3268-3273. https://doi.org/10.1073/pnas.1222463110
Runge, E. C. (1968). Effects of rainfall and temperature interactions during the growing season on corn yield. Agron. J., 60(5), 503-507. https://doi.org/10.2134/agronj1968.00021962006000050018x
Sheffield, J., Goteti, G., & Wood, E. F. (2006). Development of a 50-year high-resolution global dataset of meteorological forcings for land surface modeling. J. Climate, 19(13), 3088-3111. https://doi.org/10.1175/jcli3790.1
Stockle, C. O., Williams, J. R., Rosenberg, N. J., & Jones, C. A. (1992). A method for estimating the direct and climatic effects of rising atmospheric carbon dioxide on growth and yield of crops: Part I. Modification of the EPIC model for climate change analysis. Agric. Syst., 38(3), 225-238. http://dx.doi.org/10.1016/0308-521X(92)90067-X
Tatsumi, K. (2016). Effects of automatic multi-objective optimization of crop models on corn yield reproducibility in the U.S.A. Ecol. Model., 322, 124-137. http://dx.doi.org/10.1016/j.ecolmodel.2015.11.006
Taub, D. R., Miller, B., & Allen, H. (2008). Effects of elevated CO2 on the protein concentration of food crops: A meta-analysis. Global Change Biol., 14(3), 565-575. https://doi.org/10.1111/j.1365-2486.2007.01511.x
Texas. (2016). EPIC and APEX models: Model executables. Temple, TX: Texas A&M AgriLife Research. Retrieved from http://epicapex.tamu.edu/model-executables/
Thornton, P. E., & Running, S. W. (1999). An improved algorithm for estimating incident daily solar radiation from measurements of temperature, humidity, and precipitation. Agric. Forest Meteorol., 93(4), 211-228. http://dx.doi.org/10.1016/S0168-1923(98)00126-9
Thornton, P. E., Hasenauer, H., & White, M. A. (2000). Simultaneous estimation of daily solar radiation and humidity from observed temperature and precipitation: An application over complex terrain in Austria. Agric. Forest Meteorol., 104(4), 255-271. http://dx.doi.org/10.1016/S0168-1923(00)00170-2
Tingem, M., & Rivington, M. (2009). Adaptation for crop agriculture to climate change in Cameroon: Turning on the heat. Mitig. Adapt. Strat. Global Change, 14(2), 153-168. https://doi.org/10.1007/s11027-008-9156-3
USDA. (2010). Usual planting and harvesting dates for U.S. field crops. Washington, DC: USDA. Retrieved from http://usda.mannlib.cornell.edu/usda/current/planting/planting-10-29-2010.pdf
USDA. (2014). Quick Stats database. Washington, DC: USDA National Agricultural Statistics Service. Retrieved from http://quickstats.nass.usda.gov
USDA. (2016). Revised Universal Soil Loss Equation, Ver. 2 (RUSLE2). Washington, DC: USDA Natural Resources Conservation Service. Retrieved from http://fargo.nserl.purdue.edu/rusle2_dataweb/RUSLE2_Index.htm
van Vuuren, D. P., Edmonds, J., Kainuma, M., Riahi, K., Thomson, A., Hibbard, K., ... Rose, S. K. (2011). The representative concentration pathways: An overview. Climatic Change, 109(1), 5. https://doi.org/10.1007/s10584-011-0148-z
Vanaja, M., Maheswari, M., Jyothi Lakshmi, N., Sathish, P., Yadav, S. K., Salini, K., ... Razak, A. (2015). Variability in growth and yield response of maize genotypes at elevated CO2 concentration. Adv. Plants Agric. Res., 2(2), 42.
Waha, K., Muller, C., Bondeau, A., Dietrich, J. P., Kurukulasuriya, P., Heinke, J., & Lotze-Campen, H. (2013). Adaptation to climate change through the choice of cropping system and sowing date in sub-Saharan Africa. Global Environ. Change, 23(1), 130-143. http://dx.doi.org/10.1016/j.gloenvcha.2012.11.001
Wang, X., He, X., Williams, J. R., Izaurralde, R. C., & Atwood, J. D. (2005). Sensitivity and uncertainty analyses of crop yields and soil organic carbon simulated with EPIC. Trans. ASAE, 48(3), 1041-1054. http://dx.doi.org/10.13031/2013.18515
Widdicombe, W. D., & Thelen, K. D. (2002a). Row width and plant density effect on corn forage hybrids. Agron. J., 94(2), 326-330. http://dx.doi.org/10.2134/agronj2002.3260
Widdicombe, W. D., & Thelen, K. D. (2002b). Row width and plant density effects on corn grain production in the northern Corn Belt. Agron. J., 94(5), 1020-1023. http://dx.doi.org/10.2134/agronj2002.1020
Williams, J. R. (1995). The EPIC model. In V. P. Singh (Ed.), Computer models of watershed hydrology (pp. 909-1000). Littleton, CO: Water Resources Publications.
Xiong, W., Balkovic, J., van der Velde, M., Zhang, X., Izaurralde, R. C., Skalsky, R., ... Obersteiner, M. (2014). A calibration procedure to improve global rice yield simulations with EPIC. Ecol. Model., 273, 128-139. http://dx.doi.org/10.1016/j.ecolmodel.2013.10.026
Yapo, P. O., Gupta, H. V., & Sorooshian, S. (1998). Multi-objective global optimization for hydrologic models. J. Hydrol., 204(1), 83-97. http://dx.doi.org/10.1016/S0022-1694(97)00107-8
Figure A1. Relationship between radiation use efficiency (RUE) and CO2 level using equation 3.
(a) (b) Figure A2. Spatial patterns of ensemble mean value using final parameter sets: (a) biomass-energy ratio (BE) and (b) harvest index (HI).
Figure A3. Difference between simulated ensemble mean corn yield and average reported corn yield during 1999-2010 (t ha-1).
Figure A4. Temporal ensemble mean simulated corn yield and reported corn yield 1999-2010 for areas currently under corn cultivation.
Figure A5. Comparison between reported and simulated corn yields for each state for 12 years (1999-2010).
Figure A6. Spatial patterns of ensemble mean values of R2 between simulated and reported yield for 1999-2010.
Table A1. Initial parameters and inputs used in EPIC for corn yield simulations (biomass-energy ratio and harvest index were calibrated). Parameter Symbol Domain Source Biomass-energy ratio (kg MJ-1) BE 30 to 50 Based on Wang et al. (2005) Harvest index HI 0.45 to 0.60 Wang et al. (2005) Planting dates (DOY) PDAY Median of “most active” USDA (2010) Harvesting dates (DOY) HDAY Median of “most active” USDA (2010) Planting density (plant m-2) PDENS 5.6 to 9.0 Based on Widdicombe and Thelen (2002a, 2002b) Optimal temperature (°C) OPT 25 Kiniry et al. (1995) Base temperature (°C) BT 8 Kiniry et al. (1995) Fraction of season when leaf area declines (-) DLAI 0.7 Kiniry et al. (1995) Potential leaf area index (-) DMLA 6.5 Kiniry et al. (1995) First point on optimal leaf area development curve (-) DLAP1 15.05 Kiniry et al. (1995) Second point on optimal leaf area development curve (-) DLAP2 50.95 Kiniry et al. (1995) Biomass-energy decline rate (-) RBMD 1.0 Kiniry et al. (1995) Leaf area index decline rate parameter (-) RLAD 1.0 Kiniry et al. (1995) Maximum rooting depth (m) RDMX 2.0 Kiniry et al. (1995)
Table A2. GCMs used for analysis in this study. “Lat” and “Lon” are the latitudinal and longitudinal spatial resolutions (°) of the models. Institute ID Model Modeling Center or Group Lat Lon CSIRO-BOM ACCESS1.0 CSIRO (Commonwealth Scientific and Industrial Research Organisation) and BOM (Bureau of Meteorology), Australia 1.24 1.88 BCC BCC-CSM1.1 Beijing Climate Center, China Meteorological Administration 2.81 2.81 NCAR CCSM National Center for Atmospheric Research 0.94 1.25 CNRM-CERFACS CNRM-CM5 Centre National de Recherches Météorologiques / Centre Européen de Recherche et Formation Avancée en Calcul Scientifique 1.41 1.41 NOAA GFDL GFDL-CM3 NOAA Geophysical Fluid Dynamics Laboratory 2.00 2.50 NASA GISS GISS-E2-R NASA Goddard Institute for Space Studies 2.00 2.50 MOHC
HadGEM2-AO Met Office Hadley Centre (additional HadGEM2-ES realizations contributed by Instituto Nacional de Pesquisas Espaciais) 1.25 1.88 HadGEM2-CC 1.25 1.88 HadGEM2-ES 1.24 1.88 INM INM-CM4 Institute for Numerical Mathematics 1.50 2.00 IPSL IPSL-CM5A-LR Institut Pierre-Simon Laplace 1.88 3.75 MIROC MIROC-ESM-CHEM Japan Agency for Marine-Earth Science and Technology, Atmosphere and Ocean Research Institute (University of Tokyo), and National Institute for Environmental Studies 2.81 2.81 MIROC-ESM 2.81 2.81 MIROC5 Atmosphere and Ocean Research Institute (University of Tokyo), National Institute for Environmental Studies, and Japan Agency for Marine-Earth Science and Technology 1.41 1.41 MPI-M MPI-ESM-LR Meteorological Research Institute 1.88 1.88 MRI-CGCM3 1.13 1.13 NCC NorESM1-M Norwegian Climate Centre 1.88 2.50
Table A3. Simulated spatially averaged changes in yield (%) at the state level under the RCP8.5 scenario relative to average reported yield for 1999-2010. State 1990s to 2050s 1990s to 2070s Without
Alabama -37.1 -30.5 -41.4 -31.2 Arkansas -23.7 -17.8 -29.5 -21.6 California -21.6 -11.4 -32.5 -19.4 Colorado -26.0 -20.2 -29.9 -21.2 Delaware -7.0 -0.9 -22.8 -13.9 Georgia -22.4 -15.8 -29.4 -20.1 Illinois -22.5 -15.4 -32.0 -22.2 Indiana -18.9 -10.9 -27.0 -15.5 Iowa -16.2 -9.5 -26.5 -17.8 Kansas -24.9 -16.1 -31.3 -18.1 Kentucky -20.0 -12.9 -32.3 -22.2 Louisiana -23.1 -17.0 -30.2 -22.1 New Jersey -6.2 3.1 -11.0 3.8 Maryland -11.0 -3.6 -23.7 -13.1 Michigan -9.4 -0.2 -12.1 1.8 Minnesota -8.8 -2.7 -17.7 -10.2 Mississippi -22.3 -15.2 -28.9 -19.0 Missouri -25.5 -18.1 -34.0 -23.5 Montana -3.7 3.0 -5.69 3.5 Nebraska -22.3 -15.2 -29.1 -18.8 New York -6.0 2.9 -10.4 3.0 North Carolina -36.1 -27.8 -42.3 -29.5 North Dakota -10.0 -0.5 -16.8 -3.0 Ohio -14.6 -5.9 -18.0 -4.7 Oklahoma -30.3 -20.5 -36.6 -21.3 Pennsylvania -11.6 -0.4 -12.3 5.7 South Carolina -24.9 -14.9 -30.4 -14.8 South Dakota -18.2 -8.6 -25.9 -11.6 Tennessee -37.4 -30.3 -43.5 -32.6 Texas -23.8 -15.4 -33.8 -21.9 Virginia -25.7 -18.0 -36.2 -24.8 Washington -7.3 -3.0 -10.1 -5.2 West Virginia -1.6 8.8 -9.98 6.0 Wisconsin -9.8 -2.3 -16.6 -6.3 Wyoming -7.3 2.6 -14.9 -0.1