Content uploaded by Jane Southworth
Author content
All content in this area was uploaded by Jane Southworth
Content may be subject to copyright.
CHANGES IN SOYBEAN YIELDS IN THE MIDWESTERN UNITED
STATES AS A RESULT OF FUTURE CHANGES IN CLIMATE,
CLIMATE VARIABILITY, AND CO2FERTILIZATION
JANE SOUTHWORTH 1,R.A.PFEIFER
2, M. HABECK 3, J. C. RANDOLPH 1,
O. C. DOERING 3, J. J. JOHNSTON1and D. G. RAO4
1School of Public and Environmental Affairs, 1315 E. Tenth Street, Indiana University,
Bloomington, IN 47405-1701, U.S.A.
E-mail: jsouthwo@indiana.edu
2Agronomic consultant, New Palestine, IN, U.S.A.
3Department of Agricultural Economics, Purdue University, West Lafayette, IN, U.S.A.
4Central Research Institute for Dryland Agriculture, Hyderabad, India
Abstract. This modeling study addresses the potential impacts of climate change and changing
climate variability due to increased atmospheric CO2concentration on soybean (Glycine max (L.)
Merrill) yields in the Midwestern Great Lakes Region. Nine representative farm locations and six
future climate scenarios were analyzed using the crop growth model SOYGRO. Under the future
climate scenarios earlier planting dates produced soybean yield increases of up to 120% above current
levels in the central and northern areas of the study region. In the southern areas, comparatively small
increases (0.1 to 20%) and small decreases (–0.1 to –25%) in yield are found. The decreases in yield
occurred under the Hadley Center greenhouse gas run (HadCM2-GHG), representing a greater warm-
ing, and the doubled climate variability scenario - a more extreme and variable climate. Optimum
planting dates become later in the southern regions. CO2fertilization effects (555 ppmv) are found
to be significant for soybean, increasing yields around 20% under future climate scenarios. For the
study region as a whole the climate changes modeled in this research would have an overall beneficial
effect, with mean soybean yield increases of 40% over current levels.
1. Introduction
In recent years concerted international efforts have been undertaken to address the
problem of global climate change (Intergovernmental Panel on Climate Change
(IPCC) (1995, 1990) although debate continues as to its causes and potential im-
pacts. Consensus has been reached however, recognizing the serious implications
of such potential changes in climate to many sectors. Any change in climate will
have implications for climate-sensitive systems such as agriculture, forestry, and
natural resources. The combined effect of elevated temperatures, increased at-
mospheric CO2concentrations, increased probability of extreme events (droughts,
floods), and reduced crop-water availability, is expected to have significant impacts
on the agricultural sector (Chiotti and Johnston, 1995).
With respect to agriculture, changes in solar radiation, temperature, and precipi-
tation will produce changes in crop yields, crop mix, cropping systems, scheduling
Climatic Change 53: 447–475, 2002.
© 2002 Kluwer Academic Publishers. Printed in the Netherlands.
448 JANE SOUTHWORTH ET AL.
of field operations, pest conditions, and grain moisture content at harvest. Climate
change also will have effects on the economics of agriculture, including changes
in crop production practices, farm profitability, trade, and regional or national
comparative advantage. The impacts of climate change will depend on both the
magnitude of the change in climate and how well agriculture can adapt to these
changes (Kaiser et al., 1995). Our research addresses such issues for nine agri-
cultural farm locations across the Midwestern Great Lakes region (Figure 1), a
five-state area including Indiana, Illinois, Ohio, Michigan, and Wisconsin. This
region is one of the most productive and important agricultural regions in the
world, with over 50% of the land-use devoted to agriculture (National Agricultural
Statistics Service, 1997).
While most studies of climate change impacts on agriculture have analyzed ef-
fects of mean changes of climatic variables on crop production, impacts of changes
in climate variability have been studied much less (Mearns et al., 1997; Mearns,
1995). Yet the consequences of changes in climate variability may be as important
as those that arise due to changes in mean climatic variables (Hulme et al., 1999;
Carnell and Senior, 1998; Semenov and Barrow, 1997; Liang et al., 1995; Rind,
1991; Mearns et al., 1984). Climate variability, and its potential changes under fu-
ture climates, is therefore essential to evaluate, specifically for its possible impacts
on agriculture (Semenov and Barrow, 1997; Barrow et al., 1996; Semenov et al.,
1996; Rind, 1991). Properly validated crop simulation models can be used to model
multiple effects and see their interaction in terms of the environmental effects on
crop physiological processes and to evaluate the consequences of such influences
(Peiris et al., 1996).
The major production region for soybeans lies between 25◦and 45◦latitude at
altitudes of less than 1000 m where temperature ranges are generally favorable to
soybean production. Currently, the United States produces over 50% of the world’s
soybeans (Wittwer, 1995), making it one of the nation’s most important crops.
Any change in soybean yield brought about by climate change could have major
economic consequences in many regions of the United States (Jones et al., 1999).
Research by Ferris et al. (1998) suggests that the impact of climatic extremes
on determinate crops depends on the stage of crop development. In soybeans, pod
development and seed filling are the most sensitive phases of crop development.
High temperatures at the time of flowering or early seed filling, and water deficit
during the reproductive stage, reduces photosynthesis and seed yield of soybean
(Ferris et al., 1998).
Modeling studies have found specific relationships between temperatures and
crop growth or development. Hoogenboom et al. (1995) found the number of days
from planting to flowering or physiological maturity was shortest at temperatures
between 25◦and 30 ◦C. This number of days will increase at higher temperatures,
due to the decrease in reproductive development rate. In addition, total biomass
growth shows a strong cultivar based response to temperature. Jones et al. (1999)
reported optimum temperatures for soybean seed yield with decreases at higher
CHANGES IN SOYBEAN YIELDS IN THE MIDWESTERN UNITED STATES 449
Figure 1. Location of nine representative agricultural areas in the Midwestern Great Lakes Region.
temperatures. Higher temperatures were found to have direct physiological effects
as well as indirect effects through increased evapotranspiration and hence, water
stress. Lal et al. (1999) concluded that both maximum and minimum temperatures
are important to soybean production.
Moisture stress will also impact soybean yields. Varietal differences to moisture
stress can be dramatic, and short season cultivars tend to be most affected during
an extended period of drought stress during the warmer part of the growing season
(Specht et al., 1986). In addition, the stage of growth of the plant during which
moisture stress occurs directly influences the impact upon the plant (Shaw and
Laing, 1966). Specifically, stress in early growth stages reduces the number of pods
450 JANE SOUTHWORTH ET AL.
produced by the plant, while moisture stress in later periods of growth reduces the
number of seeds per pod and bean size, moisture stress at either time eventually
contributed to lower yields. Significantly different temperature or precipitation
regimes in these regions could have dramatic effects on soybean production, even
given the soybean plant’s great ability to recover from these stresses.
Another important factor for crop production under a changed climate is the fer-
tilizing effect of increased atmospheric concentrations of CO2. This was reported
over 100 years ago and may have been observed much earlier. Numerous studies
have been conducted throughout western and northern Europe and all reported sig-
nificant increases in plant growth and productivity (Wittwer, 1995). However, plant
species differ in their response to CO2fertilization. C3species, such as soybeans
and wheat, benefit from increased atmospheric concentrations of CO2. The primary
reason is that the increased atmospheric CO2will reduce photorespiratory loss of
carbon in the C3plant, enhancing plant growth and productivity (Allen et al., 1987).
Wittwer (1995) reported that 93% of over 1000 studies of CO2fertilization
effects found an increase in plant productivity, with a mean increase in yield of
52%. Recent studies of climate change impacts on soybean yields in the U.S. corn
belt also reported increases as a result of CO2fertilization (Kaiser et al., 1995;
Phillips et al., 1996; Ritchie, 1989). Yield increases ranged from 15 to 83%. Ritchie
(1989) reported yield decreases when the fertilization effect of CO2was not taken
into account.
Jones et al. (1999) reported a north-south trend in simulated soybean yields
under future climate scenarios. Increase in temperatures of up to 2◦C had little
effect on yields at most northern locations, and decreased yields by 16% in southern
locations. When the CO2fertilization effect was added, all yields increased unless
precipitation decreased. Across all locations average increases in yield ranged from
about 1% in Mississippi to 35% in Michigan. In addition, Jones et al. (1999)
found that the best planting date under future climates was about 17 days later
than present. The change in planting date apparently resulted in cooler conditions
during the reproductive period of growth under the warmer environment and pro-
vided more favorable water availability during the critical growth phase. The later
planting date also resulted in the need for an earlier maturing cultivar.
For this project, soybean growth was modeled at nine representative farm loca-
tions across the five-state study area. Seven climate scenarios were created: the
current climate, two future climates based on mean climate changes, and four
variability analyses as part of a sensitivity study. Research questions we address in
this paper relate to the impacts of possible future climates, under the assumption of
increased atmospheric CO2concentration, on current agricultural practices across
the midwestern Great Lakes Region. Specifically, what are: (1) the impacts of mean
climate change on soybean yields in the Midwestern United States, (2) the impacts
of changing climate variability in conjunction with changes in the mean climate
on soybean yields, (3) the impact of CO2fertilization and changing future climate
(both mean and variability) on soybean yields, (4) the spatial variability of yield
CHANGES IN SOYBEAN YIELDS IN THE MIDWESTERN UNITED STATES 451
Tab l e I
Soybean maturity group for early, mid and late maturing soybeans selected for each
location
Area Maturity group #
Late-maturing Mid-maturing Early-maturing
Eastern Illinois (EIL) IV aIII II
East-central Indiana (ECI) IV III II
North-west Ohio (NOW) III II I
South-central Michigan (SCM) III II I
Southern Illinois (SIL) V IV III
South-west Indiana (SWI) V IV III
Eastern Wisconsin (EWI) II I 0
South-west Wisconsin (SWW) III II I
Michigan thumb (MTH) III II I
aRoman numerals indicate the maturity group at each location. Higher numbers indicate
an increased number of days to maturity.
changes and implications of such changes on midwestern agriculture, and (5) the
possible adaptation strategies for farmers in the Midwest to potential changes in
future climate.
2. Methods
2.1. STUDY REGION
The midwestern Great Lakes region (Indiana, Illinois, Ohio, Michigan, and Wis-
consin) was divided into nine ecological regions based on climate, soils, land use,
and current agricultural practices. Representative farms were created in each area
based upon local characteristics and farm endowments (Figure 1). At each location,
yields are simulated for three soybean cultivars, which are early, mid, and late
maturing. The soybean maturity groups used in each location are given in Table I.
2.2. SOYBEAN CROP MODEL
The Decision Support System for Agrotechnology Transfer (DSSAT) software is
a suite of crop models that share a common input-output data format. The DSSAT
itself is a shell that allows the user to organize and manipulate data and to run
crop models in various ways and analyze their outputs (Thornton et al., 1997;
Hoogenboom et al., 1995). DSSAT version 3.5 was used in this analysis.
We selected the SOYGRO model (Jones et al., 1988) for this research because:
(1) the daily time step of the model allows analysis from different planting dates to
452 JANE SOUTHWORTH ET AL.
meet adaptations and farm management requirements, (2) plant growth dependence
on both mean daily temperatures and the amplitude of daily temperature values was
desired (not just a daily mean temperature growth dependence), (3) the model is
physiologically oriented and simulates crop response to major climate variables,
including the effects of soil characteristics on water availability, (4) the model can
simulate CO2fertilization adequately (Hoogenboom et al., 1995; Siqueira et al.,
1994; Jones et al., 1988), (5) the model is developed with compatible data struc-
tures so that the same soil and climate datasets can be used for all cultivars of crops
which helps in comparison (Adams et al., 1990), and (6) comprehensive validation
has been done across a wide range of different climate and soil conditions, and for
different crop hybrids (Semenov et al., 1996; Wolf et al., 1996; Hoogenboom et al.,
1995).
Phasic development is quantified according to the plant’s physiological age.
The input data required to run the SOYGRO model includes daily weather infor-
mation (maximum and minimum temperatures, rainfall, and solar radiation); soil
characterization data (data by soil layer on extractable nitrogen and phosphorous
and soil water content); a set of genetic coefficients characterizing the soybean
maturity group being grown (Table I); and crop management information, such as
emerged plant population, row spacing, seeding depth, and fertilizer and irrigation
schedules (Thornton et al., 1997). The soil data were obtained from the U.S. Nat-
ural Resources Conservation Service (USDA, 1994) and were selected to represent
the dominant local soil at each of the nine representative farm sites. The model
apportions the rain received on any day into runoff and infiltration into the soil,
using the runoff curve number technique. We assigned a runoff curve number to
each soil, based on the soil type, depth and texture as obtained from the STATSGO
database (USDA, 1994). We chose not to make nitrogen and phosphorous limiting
to crop growth, so these modules are turned off within our model runs.
Temperature is one of the main controls on the processes in plant growth and
development. SOYGRO includes two different temperature response functions, one
for vegetative development and one for reproductive development. The former has
a minimum temperature for development of 7 ◦C, an optimum temperature between
30◦and 35 ◦C, with decreasing development at higher temperatures, until 45 ◦C
is reached. The latter has a curve-linear response curve; with a base temperature
of 6.4 ◦C, maximum development between 21.1◦and 26.8 ◦C, and then a sharp
decline at higher temperatures until 41 ◦C, above which there is no reproductive
development (Hoogenboom et al., 1995).
The SOYGRO model (Jones et al., 1988) includes the capability to simulate
the direct physiological effects of increased atmospheric CO2concentrations on
plant photosynthesis and water use. The atmospheric CO2concentration for the
future climate scenarios for the years 2050 through 2059 used in this research is
555 ppmv. This is the value modeled in the HadCM2 models for this time period,
assuming a 1% increase in CO2production per year (D. Viner, pers. comm.). The
photosynthetic enhancement of CO2results in a photosynthesis multiplier of 1.21
CHANGES IN SOYBEAN YIELDS IN THE MIDWESTERN UNITED STATES 453
and a transpiration ratio of 0.96 for soybeans (Hoogenboom et al., 1995; Siqueira
et al., 1994).
2.3. CROP MODEL VALIDATION
The DSSAT models have been well validated across a large range of environmental
conditions and crops, and are not specific to any particular location or soil type
(Fischer et al., 1995). SOYGRO has been extensively validated at several locations
(Dhakhwa et al., 1997; Hoogenboom et al., 1995). Furthermore, because manage-
ment practices may be varied (e.g., changing planting dates, selecting different
cultivars), it is possible to simulate potential farm-level adjustments in response to
climate change (Schimmelpfennig et al., 1996; Rosenberg, 1992).
Detailed farm-level data were used to ensure that yields produced by the model
reflected the actual yields in each representative agricultural area. Experiments
conducted during 1981 and 1982 at the Purdue University Agronomy Farm pro-
vided data for eight planting dates (Figure 2). Overall agreement between the
observed data and simulated results was good, with an R2of 0.76. The deviation
in the observed values and simulated yield is in part due to the model being a
simplified reality and as such it will not capture every relevant phenomenon. How-
ever, despite this lack of perfect agreement due to some model imperfection the
overall agreement between the observed and simulated yields is high, and as such
is appropriate for use here.
Each representative farm location was also validated using historical yield and
climate data from as close an area as possible to ensure the model could replicate
past yields for mid-maturing (currently grown) soybean cultivars as well as both
late- and early-maturing cultivars that might be used in the future. These analy-
ses thus ensured that each location could mimic reality. Our future yields should
therefore indicate changes due to climate change not due to model or location
(Siqueira et al., 1994). While this is an imperfect assumption we cannot validate
the model for climate conditions that do not exist and we make the assumption
that the responses we see under future climate runs are representing changes based
on the climate conditions, as the climate is the only data that changes across these
runs.
2.4. CURRENT CLIMATE ANALYSIS:VEMAP
The VEMAP dataset includes daily, monthly, and annual climate data for the
United States including maximum, minimum, and mean temperature, precipitation,
solar radiation, and humidity (Kittel et al., 1996). The VEMAP baseline (30-year
historical mean) climate data was used for each of our nine representative agri-
cultural areas in our study region. The weather generator SIMMETEO (as used
in DSSAT version 3.5) used these climate data to stochastically generate daily
weather data in model runs. This approach of using monthly data to generate daily
data allowed us to generate variability scenarios as well. The climate variables
454 JANE SOUTHWORTH ET AL.
Figure 2. Validation of SOYGRO results for a 2-year, 5-planting date simulation of soybean yield at
Purdue University Agronomy Farm.
distribution patterns reflect the current climate variables distributions, as there is
no way to know future distributions. Various combinations of planting and harvest
date were used under each climate scenario to determine yield sensitivity to these
factors.
2.5. FUTURE CLIMATE SCENARIOS
Future climate experiments performed recently at the Hadley Center in England
used the new Unified Model. The Unified Model was modified slightly to pro-
duce a new, coupled ocean-atmosphere GCM, referred to as HadCM2, which has
been used in a series of transient climate change experiments using historic and
future greenhouse gas and sulfate aerosol forcing. Transient model experiments
are considered more physically realistic and complex than equilibrium scenarios,
and allow atmospheric concentrations of CO2to rise gradually over time (Harrison
and Butterfield, 1996). HadCM2 was also one of the climate models used in the
U.S. National Assessment (National Assessment Synthesis Team, 2001).
HadCM2 has a spatial resolution of 2.5◦×3.75◦latitude by longitude and the
representation produces a grid box resolution of 96 ×73 grid cells, which produces
CHANGES IN SOYBEAN YIELDS IN THE MIDWESTERN UNITED STATES 455
a surface spatial resolution of about 417 ×278 km reducing to 295 ×278 km at
latitude 45◦. The atmospheric component of HadCM2 has 19 levels and the ocean
component 20 levels. The equilibrium sensitivity of HadCM2, that is the global-
mean temperature response to a doubling of effective CO2concentration, is 2.5 ◦C,
somewhat lower than most other GCMs (IPCC, 1990).
The greenhouse-gas-only version, HadCM2-GHG, used the combined forcing
of all the greenhouse gases as an equivalent CO2concentration. HadCM2-SUL
used the combined equivalent CO2concentration plus a negative forcing from
sulfate aerosols. The addition of the negative forcing effects of sulfate aerosols
represents the direct radiative forcing due to anthropogenic sulfate aerosols by
means of an increase in clear-sky surface albedo proportional to the local sulfate
loading (Carnell and Senior, 1998). Our research used the period of 2050 to 2059
for climate scenarios. The results from HadCM2-GHG and HadCM2-SUL cannot
be viewed as a forecast or prediction, but rather as two possible realizations of
how the climate system may respond to a given forcing. A comparison of the main
three climate datasets (Table II) highlights the differences in projected climate for
the study region. A climate variability analysis was also conducted on these two
scenarios, thus increasing the number of future climate scenarios to six. Conse-
quently, we have examined a range of probable climate changes and a range of
their impacts on soybean growth.
2.6. CLIMATE VARIABILITY ANALYSIS
In order to separate crop response to changes in climatic means from crop response
to changes in climate variability it is necessary first to model the impacts of mean
temperature changes on crop growth. Then a time series of climate variables with
changed variability can be constructed and added to the scenarios of mean change.
Hence, when the analysis is undertaken on future mean and variability changes it
is possible to infer what type of climate change caused changes in yield (Mearns,
1995).
The variance of maximum and minimum temperature, and precipitation values
for each month were altered separately, according to the following algorithm from
Mearns (1995):
X
t=µ+δ1/2(Xt−µ), (1)
and
δ=σ2/σ 2,(2)
where X
t=new value of climate variable Xt(e.g., monthly mean maximum
February temperature for year t); µ=mean of the time series (e.g., the mean of
the monthly mean maximum February temperatures for a series of years); δ=ratio
of the new to the old variance of the new and old time series; Xt=old value of
456 JANE SOUTHWORTH ET AL.
Tab l e I I
Comparison of HadCM2-GHG and HadCM2-SUL climate scenario mean monthly maximum and minimum temperature and precipitation values,
compared to VEMAP current mean monthly climatic conditions for August
Area aVEMAP HadCM2-GHG HadCM2-SUL
Maximum Minimum Total monthly in max. in min. in total monthly in max in min. in total monthly
temp. (◦C) temp. (◦C) precip. (mm) temp. (◦C) temp. (◦C) precip. (mm) temp. (◦C) temp. (◦C) precip. (mm)
EIL 28.3 16.1 82.0 +13.3 +11.8 –4.1 +10.4 +10.3 –14.2
ECI 27.6 14.5 79.0 +11.5 +11.2 +12.5 +8.3 +9.5 +8.2
NWO 26.8 14.2 77.0 +9.9 +9.5 –8.8 +6.4 +7.4 +3.9
SCM 26.3 14.3 79.0 +9.5 +8.5 –8.7 +6.1 +6.7 –11.4
SIL 30.5 17.3 85.0 +11.1 +10.6 –7.1 +8.2 +9.1 –17.2
SWI 29.8 17.1 90.0 +9.3 +8.6 –0.7 +6.1 +6.9 –2.8
EWI 26.4 13.8 93.0 +12.5 +10.7 –16.9 +9.3 +8.8 –7.9
SWW 26.8 15.2 98.0 +12.1 +9.3 –21.9 +8.9 +7.4 –12.9
MTH 26.4 13.1 74.0 +9.8 +9.7 –3.7 +6.4 +7.9 –6.4
aWhere EIL is eastern Illinois, ECI is east-central Indiana, NWO is north-west Ohio, SCM is south-central Michigan, SIL is southern Illinois, SWI
is south-west Indiana, EWI is eastern Wisconsin, SWW is south-west Wisconsin, and MTH is the Michigan thumb.
CHANGES IN SOYBEAN YIELDS IN THE MIDWESTERN UNITED STATES 457
climate variable (e.g., the original monthly mean February temperature for year t);
σ2=new variance; and σ2=old variance.
To change the time series to have a new variance σ2, the variance and mean
of the original time series was calculated and then a new ratio (δ) was chosen.
For this research we halved the variance and then doubled the variance. From
the parameters µ,δ, and the original time series, a new time series with variance
was calculated using equation one. This algorithm was used to change both max-
imum and minimum temperatures and the precipitation time series. This simple
method, developed by Mearns (1995), was used although more complex method-
ologies are being developed to allow comparisons of variability techniques. Mearns
et al. (1996, 1997) found the results obtained from more computationally ad-
vanced statistical techniques were similar to results from more statistically simple
methodologies. The approach we used permits the incorporation of changes in both
mean and variability of future climate in a computationally simple, consistent, and
reproducible manner.
2.7. MODEL LIMITATIONS
The SOYGRO model, as with all models, involves a large number of assumptions.
Our simulation results do not take into account possible effects of plant diseases,
pest damage or weed competition (Hoogenboom et al., 1995). Competitive crop-
weed interactions could change, particularly if the crop and weed have different
photosynthetic pathways (C3versus C4), which may be differentially affected by
elevated CO2or climate changes. Nutrients were non-limiting in our model runs,
which may not be the case, especially under the elevated atmospheric CO2concen-
trations (Phillips et al., 1996). In addition, the model does not simulate extreme soil
conditions such as salinity, acidity, compaction, or extreme weather events such
as floods, tornadoes, hail, droughts, and hurricanes (Hoogenboom et al., 1995).
Other limitations relate to the simplified representation of the farms and the use
of a single dominant soil type at each location. The climatic tolerance of cultivars
was assumed to be unchanged from the base runs through the simulation. Finally,
the experimental conditions used to examine increased CO2effects on photosyn-
thesis may overestimate yields (Rosenzweig et al., 1994). While the assumptions
discussed may tend to either overestimate or underestimate simulated yields (Fig-
ure 2), our extensive validation and analysis at the farm level acts to increase the
validity of the model runs, at least under current climate conditions.
3. Results
3.1. CLIMATE CHANGE SCENARIOS:TEMPERATURES AND PRECIPITATION
A comparison of the climate scenarios provides a background for the soybean
yield results. The most notable change in the climate under the HADCM2-GHG
458 JANE SOUTHWORTH ET AL.
and HADCM2-SUL scenarios is the increase in average annual temperatures. The
number of days during the growing season with temperatures above 35 ◦Cand
above 40 ◦C for each of the locations is given in Table II. The increase is dramatic
at all sites, particularly under the HadCM2-GHG scenario. Those very high tem-
peratures may cause some stress on the growing soybeans, hasten senescence, and
reduce yields, particularly in the southern reaches of the study area. A comparison
of climatic conditions for the month of August (Table III) shows that this particular
month may be drier and much warmer under the changed climate at each of the
locations.
However, a full year’s climate data may be more meaningful for comparison.
Two sites, one northern (eastern Wisconsin) and one southern (southern Illinois),
were chosen for comparison (Figures 3a,b). In every month at these sites, the
average monthly minimum temperatures are significantly higher under both the
HadCM2-GHG and HadCM2-SUL scenarios. The average monthly maximum
temperatures also increase, but most significantly in the summer months. The pat-
terns at both locations are similar. The precipitation pattern, however, is not the
same. At southern Illinois, the late fall to early spring months tend to be wetter
than currently, while at the eastern Wisconsin site, this period is drier. Impacts on
yields in the northern and southern areas, then, may be expected to be different.
Extrapolation of results from research at northern sites to more southern areas and
vice versa would not be advised.
3.2. CONSEQUENCES OF CHANGING MEAN CLIMATE AND CLIMATE
VARIABILITY ON SOYBEAN YIELDS
Changes in soybean yields under climate change are presented in Figures 4–
6. These figures show the change in average maximum yield for the optimum
planting date under each scenario. Percent changes in yield were derived by com-
paring the future soybean yields to the current maximum yields. Increases in yield
were greatest in HadCM2-SUL runs and for the halved and unchanged variability
analyses. Decreases in soybean yields occurred in some southern locations for the
HadCM2-GHG runs, most significantly in the doubled variability analysis.
Late-maturing soybean cultivars showed increases in yields, for all future cli-
mate scenarios, in all northern and central locations (Figure 4). The increases
range from 0.1 to 120%. In all future climate scenarios the largest increases in
yield occurred in south-central Michigan and the Michigan thumb. In contrast, the
southernmost locations showed slight decreases in yields under some of these same
future climate scenarios. These decreases in yield range from –0.1 to –20%. The
changes in soybean yield under the HadCM2-GHG scenarios at the four south-
ernmost agricultural sites (southern Illinois, southwest Indiana, eastern Illinois,
and east-central Indiana) ranged from +10% to –20%. Under the HadCM2-SUL
scenarios, only southern Illinois under the doubled variability run experienced any
decreases in yield, and these only of –0.1 to –5%. The HadCM2-GHG scenarios
CHANGES IN SOYBEAN YIELDS IN THE MIDWESTERN UNITED STATES 459
Table III
Number of days in the growing season (1 May–30 September) with maximum daily temperatures 35.0◦–39.9 ◦C, and 40.0◦–44.9 ◦C under
1 year of VEMAP current climate and future HadCM2-GHG (halved variability (0.5)/normal variability (1.0)/doubled variability (2.0)) and
HadCM2-SUL (halved variability (0.5)/normal variability (1.0)/doubled variability (2.0)) climate scenarios for the year 2055
Area aVEMAP: current HadCM2-GHG: 2055 HadCM2-SUL: 2055
Max. temp Max. temp. Max. temp. >35 ◦C Max. temp. >40 ◦C Max. temp. >35 ◦C Max. temp. >40 ◦C
0.5 1.0 2.0 0.5 1.0 2.0 0.5 1.0 2.0 0.5 1.0 2.0
EIL 5 0 928095 132828 444537 0 5 0
ECI 1 0 62 69 56 6 9 3 12 18 14 0 0 0
NWO1 0 203435 011 656 000
SCM0 0 151210 000 220 000
SIL 9 0 898393 222830 324444 0 1 0
SWI 4 0 606569 4 9 7 81719 0 0 0
EWI0 0 433544 140 244 000
SWW0 0 554845 132 171 000
MTH0 0 151622 010 010 000
aWhere EIL is eastern Illinois, ECI is east-central Indiana, NWO is north-west Ohio, SCM is south-central Michigan, SIL is southern Illinois,
SWI is south-west Indiana, EWI is eastern Wisconsin, SWW is south-west Wisconsin, and MTH is the Michigan thumb.
460 JANE SOUTHWORTH ET AL.
Figure 3. Average maximum and minimum temperatures and change in precipitation at (a) southern
Illinois and (b) eastern Wisconsin under future climate conditions.
CHANGES IN SOYBEAN YIELDS IN THE MIDWESTERN UNITED STATES 461
resulted in reductions in yields in southern areas and less pronounced increases
in yield across the central and northern areas of the study region. The doubled
variability runs of both scenarios resulted in decreases in yield in the southern
locations and smaller increases in yield in the central and northern locations as
compared to unchanged or halved variability runs. The greatest gains in yield of
late maturing cultivars occurred in the HadCM2-SUL scenarios.
Mid-maturing cultivars (Figure 5) under future scenarios produced soybean
yields significantly higher than under VEMAP conditions across most of the
central and northern locations. Under the HadCM2-SUL scenarios southern lo-
cations also showed predominantly increasing patterns of soybean yields. Under
the HadCM2-SUL scenarios all areas experienced increases in yield ranging from
0.1 to 90%, except for southern Illinois under doubled variability, where yields de-
creased from –0.1 to –5%. Under the HadCM2-GHG scenarios southern locations
(southern Illinois, southwest Indiana, eastern Illinois, and east-central Indiana) and
south-western Wisconsin (for doubled and normal variability runs) show minimal
increases in yield from 0.1 to 5% or alternatively show yield decreases ranging
from –0.1 to –25%. Under this same scenario the central and northern loca-
tions (east Wisconsin, south-central Michigan, north-west Ohio, and the Michigan
thumb) show yield increases of 0.1 to 60% as compared to yields using VEMAP.
The HadCM2-GHG scenario resulted in greater reductions, or smaller increases
in yield than did the HadCM2-SUL scenarios. In addition, the doubled variability
runs of both scenarios resulted in more extreme yield decreases and much smaller
yield increases, as compared to the unchanged or halved variability runs for each
scenario. The greatest gains in yield (and also results with no yield decreases any-
where) occurred for the HadCM2-SUL scenarios for the unchanged and halved
variability scenarios.
Early-maturing soybean cultivars (Figure 6) showed the greatest yield increases
under the HadCM2-GHG scenarios with only minimal yield decreases in the south-
ern areas under doubled variability. With no change in variability yields at the
east-central Indiana site decreased –5.1 to –10%. Results across all other locations
for all variability scenarios showed yields increasing from 0.1 to 80%. Under the
HadCM2-SUL scenarios all results across all scenarios and locations show yields
increasing from 0.1 to 120%.
Across locations increases in yields of late-maturing soybean are greater than
early-maturing soybean cultivars for HadCM2-SUL scenarios. However, early-
maturing cultivars had higher yields than late- and mid-maturing cultivars under
more intensive warming, as represented by the HadCM2-GHG scenario. The
HadCM2-GHG scenario resulted in greater yield decreases and smaller yield in-
creases than did the HadCM2-SUL scenario. The doubled variability runs of both
scenarios resulted in smaller yield increases and for the HadCM2-GHG scenario
some yields decreased. The greatest increases in yield occurred in the halved
variability HadCM2-SUL scenario, by as much as 120% in south-central Michigan.
462 JANE SOUTHWORTH ET AL.
Figure 4. Percent change in mean maximum decadal yield for late-maturing soybean, compared to VEMAP yields for (a) halved variability HadCM2-GHG,
(b) HadCM2-GHG, (c) doubled variability HadCM2-GHG, (d) halved variability HadCM2-SUL, (e) HadCM2-SUL, and (f) doubled variability
HadCM2-SUL.
CHANGES IN SOYBEAN YIELDS IN THE MIDWESTERN UNITED STATES 463
Figure 5. Percent change in mean maximum decadal yield for mid-maturing soybean, compared to VEMAP yields, for (a) halved variability HadCM2-GHG,
(b) HadCM2-GHG, (c) doubled variability HadCM2-GHG, (d) halved variability HadCM2-SUL, (e) HadCM2-SUL, and (f) doubled variability
HadCM2-SUL.
464 JANE SOUTHWORTH ET AL.
Figure 6. Percent change in mean maximum decadal yield for early-maturing soybean, compared to VEMAP yields, for (a) halved variability
HadCM2-GHG, (b) HadCM2-GHG, (c) doubled variability HadCM2-GHG, (d) halved variability HadCM2-SUL, (e) HadCM2-SUL, and (f) doubled
variability HadCM2-SUL.
CHANGES IN SOYBEAN YIELDS IN THE MIDWESTERN UNITED STATES 465
Figure 7. Soybean yield by planting date at southern Illinois and eastern Wisconsin sites under
HadCM2-GHG and HadCM2-SUL scenarios as percent of maximum VEMAP yield.
3.3. SHIFTS IN FUTURE PLANTING DATES FOR OPTIMUM YIELDS
Figure 7 shows soybean yields for two locations, southern Illinois and eastern Wis-
consin. The yields are plotted on separate axes, as a percent of the maximum yield
for each site. Under current (VEMAP) conditions, the optimum planting period
for mid-maturing soybeans at the southern Illinois site is from approximately the
110th to 160th day of the year (Figure 7). Soybeans planted during this same period
under the HadCM2-GHG scenario can be expected to yield only 65% as much. The
optimum planting period under the HadCM2-GHG scenario at this site is much
later, around days 170–200. Extremely high temperatures and moisture stress dur-
ing seed fill limits yield. If climate change is more similar to the sulfate scenario,
then soybean yields in southern Illinois will be expected to increase slightly over
the range of planting dates.
Results from eastern Wisconsin (Figure 7) show a very different pattern. The
optimum planting period under current conditions is approximately day 116–160.
Soybeans planted during this same period under HadCM2-GHG conditions would
show an increase in yield of about 15%. The optimum planting period under
HadCM2-GHG conditions is extended and earlier, from approximately day 102
to day 150, with soybeans planted during this time period yielding 30% higher
466 JANE SOUTHWORTH ET AL.
Figure 8. Soybean yield by planting date at southern Illinois and eastern Wisconsin sites under
HadCM2-GHG and HadCM2-SUL scenarios as percent of southern Illinois maximum VEMAP
yield.
than the current maximum. Results from the HadCM2-SUL scenario indicate even
higher yield increases throughout much of the growing season due to slightly lower
temperature increases.
The results are presented slightly differently in Figure 8. This chart shows the
yields as a percent of the maximum current yields in southern Illinois. Clearly,
soybean yields in eastern Wisconsin under current (VEMAP) conditions are less
than those in southern Illinois. However, under both future scenarios soybean yields
in eastern Wisconsin compare favorably to those in southern Illinois. In fact, under
the HadCM2-GHG scenario, yields at eastern Wisconsin are much higher for a
large portion of the planting season.
Patterns of maximum yields by planting date vary spatially across the study
region under the future climate scenarios. For southern areas (eastern Illinois,
southern Illinois, and southwest Indiana) maximum yields resulted from later plant-
ing dates. East-central Indiana showed no discernable pattern in maximum yields
or planting dates under future climate scenarios. For central and northern locations
(northwest Ohio, southwest Wisconsin, south-central Michigan, eastern Wisconsin,
and the Michigan thumb) earlier planting dates produced maximum yields. This
CHANGES IN SOYBEAN YIELDS IN THE MIDWESTERN UNITED STATES 467
north-south gradient results in a shift to predominantly earlier planting dates, with
late- and mid-maturing soybean cultivars producing maximum yields, and allowing
large yield increases in the northern locations. This is a potentially important factor
for the logistics of spring planting on farms with competing labor and machinery
requirements.
3.4. THE IMPACT OF EXTREME MOISTURE AND TEMPERATURE REGIMES ON
SOYBEANS
DSSAT results did not indicate any extreme moisture stress at either of these lo-
cations that represent the northern and southern extent of our study region, under
the HadCM2-SUL scenarios, even though Table II does indicate decreased precip-
itation in the month of August. The temperature effect is much more important in
causing these shifts in yields. However, for southern Illinois, under the doubled
variability HadCM2-GHG run, increased moisture stress is evident.
As Table III indicates, the number of days with extremely high temperatures at
each site will increase dramatically under the future climate scenarios. Incidents
of extremely high temperatures at the eastern Wisconsin site under the HadCM2-
SUL conditions will approach that of southern Illinois under current conditions.
Of particular interest, though, is the temperature during the various growth stages
of the plant. To analyze this more clearly we compared growth stages under cur-
rent climate conditions (VEMAP) and under the most extreme climate scenario
(doubled variability HadCM2-GHG). The average maximum and minimum tem-
peratures by growth stage at the eastern Wisconsin and southern Illinois sites for
these climate scenarios are given in Figure 9. The planting dates presented for the
southern Illinois site (Figure 9a) indicate the optimum period currently (144) and
under the doubled variability HadCM2-GHG scenario (193). The doubled variabil-
ity HadCM2-GHG conditions force higher maximum temperatures throughout the
growing season, exacerbating moisture stress, particularly during seed fill. Equally
dramatic is the increase in minimum temperatures.
The opposite effect is seen at eastern Wisconsin (Figure 9b). Higher maximum
temperatures and higher minimum temperatures, particularly late in the season,
result in significant yield increases. In fact, the temperatures under the doubled
variability HadCM2-GHG scenario at eastern Wisconsin are in the range of those
of current conditions at southern Illinois.
3.5. CO2FERTILIZATION EFFECTS
To evaluate the effect of CO2fertilization within our study region we selected
a single location (eastern Illinois) and simulated yields for ambient CO2levels
(360 ppmv) and under future CO2levels (555 ppmv) for VEMAP, HadCM2-GHG,
and HadCM2-SUL climate scenarios (Figure 10). The difference in yields between
360 ppmv CO2and 555 ppmv CO2is greatest under the HadCM2-SUL scenario.
The SOYGRO model results did not indicate any significant water stress during
468 JANE SOUTHWORTH ET AL.
Figure 9. Mean maximum and minimum temperatures by developmental stage for (a) Southern Illinois, and (b) Eastern Wisconsin, for current (VEMAP)
climate conditions and for the doubled variability HadCM2-GHG climate scenario.
CHANGES IN SOYBEAN YIELDS IN THE MIDWESTERN UNITED STATES 469
Figure 10. A comparison of soybean yields (bu/ac) for 20 different planting dates for Eastern Illi-
nois, under current (360 ppmv) and future (555ppmv) CO2levels, for VEMAP, HadCM2-GHG, and
HadCM2-SUL climate scenarios.
the growing season. With sufficient water availability and no nutrient stress, the
CO2fertilization effect resulted in soybean yield increases of about 30% at this
site. Where soybeans are planted early and under the HadCM2-GHG scenario, the
difference in yields was not as great. Other conditions, such as the extremely high
temperatures, were limiting and so the full effect of CO2fertilization was not seen.
4. Discussion
4.1. CROP YIELD CHANGES BY REGION
At southern locations, where later planting dates produced higher yields, and later-
maturing soybean were the dominant cultivars producing maximum yields, there
appears to be a limit in the amount of yield increase. In these locations there also
were yield decreases under more extreme future climate scenarios, in part due to
the limits in yield of the late-maturing cultivars planted later in the season. The
yield responses to increasing temperatures found in this research matches those
found by previous researchers (Jones et al., 1999; Lal et al., 1999; Ferris et al.,
1998; Hoogenboom et al., 1995). High temperatures limit soybean growth in the
southern locations, and especially under HadCM2-GHG runs, due to the greater
frequency of higher temperature events (Table III) and concomitant increases in
470 JANE SOUTHWORTH ET AL.
moisture stress under the doubled variability HadCM2-GHG scenario. Addition-
ally, the higher yields produced by later planting dates reflect the results of Jones
et al. (1999).
In the central and northern locations increases in yield are much greater due in
part to (i) more mid-maturing cultivars as the producers of maximum yields, and
(ii) to the shift to earlier planting dates. The maturity group influences the flexibility
of length of growing period. These changes in planting dates and cultivars allow for
a greater potential yield, as represented in this research, by significantly increased
potential future yields in these locations. Such shifts in planting dates reflect the
same north-south patterns as yields, and thus help to explain the shifts in yields.
This research indicated a dominance of mid-maturing cultivars in central and
northern regions, and we also found a simultaneous shift in planting dates to earlier
dates. The differences in our central and northern location from the results of Jones
et al. (1999) are likely because their sites were predominantly in southern locations,
whereas our study included more northern locations. The differential impacts upon
the climate, and hence crop yields from north to south illustrate the need for such
analyses to be conducted across larger spatial areas, as results are highly variable
spatially.
4.2. IMPACTS OF CHANGES IN CLIMATE VARIABILITY
Increasing the variability of the future climate scenarios, by using the doubled
climate variability scenarios, decreases the mean decadal crop yields. Overall,
yields associated with these doubled climate variability scenarios have lower mean
decadal yields due to some years having very low yields, i.e., the year-to-year
variability of yields is much greater under these scenarios. The greater variance
of yields associated with the doubled variability scenarios, and especially for the
HadCM2-GHG climate scenario, is due to the more extreme increases in temper-
atures and moisture stress associated with these climate scenarios (Tables II and
III), compared to the lesser increases for the HadCM2-SUL climate scenarios.
4.3. IMPACTS OF CO2FERTILIZATION AND CLIMATE CHANGE
Our results are consistent with those reported elsewhere for increased tempera-
ture and CO2concentrations (Jones et al., 1999; Lal et al., 1999; Rosenzweig and
Hillel, 1998; Phillips et al., 1996; Kaiser et al., 1995; Wittwer, 1995; Rosenzweig
et al., 1994; Siqueira et al., 1994), suggesting that the impacts of climate change
resulting in a warmer, wetter Midwestern Great Lakes Region with atmospheric
CO2concentrations of 555 ppmv, will result in large increases in soybean yields
across more central and northern locations, and some limited decreases in yields in
more southern locations. These yield decreases are more likely under conditions of
extreme heat (HadCM2-GHG) and increased climate variability. As these scenarios
are much more extreme, and considered less likely to occur, moderate warming and
CHANGES IN SOYBEAN YIELDS IN THE MIDWESTERN UNITED STATES 471
increased CO2concentrations in this region will result in greatly increased soybean
yields.
5. Conclusions
The main conclusions of this research are:
•Under future climate scenarios central and northern locations in the study re-
gion will experience large increases in soybean yields, and southern locations
will experience moderate decreases in yield.
•High temperatures appear to limit soybean yields under the future climate
scenarios.
•For soybeans, increased atmospheric CO2concentration and changing climate
variability appear to be major drivers of yield change.
•CO2fertilization produces a yield increase of approximately 20%, increasing
to 30% under moderate (HadCM2-SUL halved and normal variability runs)
future climate scenarios.
•Only under extreme climate change scenarios, as represented by the HadCM2-
GHG and doubled variability scenarios, will soybean yields decrease in all
locations in this study region due to high temperatures and, at southern
Illinois, due to high temperatures and the resultant moisture stress.
•Increasing the variability of the future climate scenarios increases the variabil-
ity of the year-to-year crop yields, and results in lower mean decadal yields,
compared to unchanged or halved variability analyses.
•Planting dates may shift. To obtain maximum yields planting dates must
shift to earlier in the central and northern locations, and later in the southern
locations. Higher yields come from earlier maturity groups.
Farmers can adapt and adjust their management practices as a result of climate
change (Schimmelpfennig et al., 1996; Rosenberg, 1992). Typical options include
changing planting dates, using cultivars better adapted to new conditions (increased
heat tolerance, increased drought tolerance, etc.,), planting alternate crops which
may be better adapted, and using supplemental irrigation. While farmers in the
southern locations of our study region may require some or all of these options
(e.g., more heat tolerant cultivars), farmers in the central and northern areas of this
region may not require adaptation strategies because their soybean yields tend to
increase under conditions of future climate. However, there will be incentives to
take advantage of increased yields from earlier planting dates. Our results demon-
strate the importance of including planting dates and other production concerns,
as well as climate variability and CO2fertilization effects, in climate change and
potential adaptation integrated research assessments.
The DSSAT suite of models provides an effective tool to study the potential
of climate change on crop production (Hoogenboom et al., 1995). Our research
472 JANE SOUTHWORTH ET AL.
also examined the impact of climate change and changing climate variability on
maize (Southworth et al., 2000) and wheat in this study region. We are currently
evaluating the impact of changes in yields and optimum planting periods due to
climate change upon the economics and logistics of these farms. Future results will
allow us to discern likely changes in current production practices and in crop mix
within this very important agricultural study region.
Acknowledgements
This research was funded by grant number R 824996-01-0 from the Science to
Achieve Results (STAR) Program of the United States Environmental Protection
Agency. The overall quality of this manuscript was greatly improved by two anony-
mous reviewers whose comments we greatly appreciate. We thank Dr. David Viner
for his help and guidance with the use of the HadCM2 data, which was provided
by the Climate Impacts LINK Project (DETR Contract EPG 1/1/68) on behalf of
the Hadley Center and the United Kingdom Meteorological Office. We thank Dr.
Joe Ritchie at Michigan State University for his help and advice on SOYGRO
throughout this project and Dr. Susan Riha at Cornell University for her comments
on the temperature sensitivity of SOYGRO. We thank Drs. Jim Jones and Ken
Boote of the University of Florida for sharing a manuscript and several stimu-
lating discussions with us. We thank Dr. Timothy Kittel at the National Center
for Atmospheric Research (NCAR), Boulder, Colorado for recommendations, help
and guidance with VEMAP data acquisition and use. We greatly appreciate the
assistance of Dr. Linda Mearns at NCAR in discussions on climate variability and
for her comments regarding this manuscript and our research. In addition we thank
Michael R. Kohlhaas for his editing of this manuscript.
References
Adams, R. M., Rosenzweig, C., Peart, R. M., Ritchie, J. T., McCarl, B. A., Glyer, J. D., Curry,
R. B., Jones, J. W., Boote, K. J., and Allen, Jr., L. H.: 1990, ‘Global Climate Change and U.S.
Agriculture’, Nature 345, 219–224.
Allen, Jr., L. H., Boote, K. J., Jones, J. W., Jones, P. H., Valle, R. R., Acock, B., Rogers, H. H.,
and Dahlman, R. C.: 1987, ‘Response of Vegetation to Rising Carbon Dioxide: Photosynthesis,
Biomass, and Seed Yield of Soybean’, Global Geochem. Cycles 1, 1–14.
Barrow, E., Hulme, M., and Semenov, M.: 1996, ‘Effect of Using Different Methods in the
Construction of Climate Change Scenarios: Examples from Europe’, Clim. Res. 7, 195–211.
Carnell, R. E. and Senior, C. A.: 1998, ‘Changes in Mid-Latitude Variability Due to Increasing
Greenhouse Gases and Sulfate Aerosols’, Clim. Dyn. 14, 369–383.
Chiotti, Q. P. and Johnston, T.: 1995, ‘Extending the Boundaries of Climate Change Research: A
Discussion on Agriculture’, J. Rural Studies 11, 335–350.
Dhakhwa, G. B., Campbell, C. L., LeDuc, S. K., and Cooter, E. J.: 1997, ‘Corn Growth: Assessing
the Effects of Global Warming and CO2Fertilization with Crop Models’, Agric. For. Meteorol.
87, 253–272.
CHANGES IN SOYBEAN YIELDS IN THE MIDWESTERN UNITED STATES 473
Ferris, R., Wheeler, T. R., Hadley, P., and Ellis, R. H.: 1998, ‘Recovery of Photosynthesis after
Environmental Stress in Soybean Grown under Elevated CO2’, Crop Sci. 38, 948–955.
Fischer, G., Frohberg, K., Parry, M. L., and Rosenzweig, C.: 1995, ‘Climate Change and World
Food Supply, Demand, and Trade’, in Climate Change and Agriculture: Analysis of Potential
International Impacts, ASA Special Publication, No. 59 Madison, Wisconsin, U.S.A.
Harrison, P. A. and Butterfield, R. E.: 1996, ‘Effects of Climate Change on Europe-Wide Winter
Wheat and Sunflower Productivity’, Clim. Res. 7, 225–241.
Hoogenboom, G., Tsuji, G. Y., Pickering, N. B., Curry, R. B., Jones, J. W., Singh, U., and Godwin,
D. C.: 1995, ‘Decision Support System to Study Climate Change Impacts on Crop Production’,
in Climate Change and Agriculture: Analysis of Potential International Impacts, ASA Special
Publication No. 59, Madison, Wisconsin, U.S.A., pp. 51–75.
Hulme, M., Barrow, E. M., Arnell, N. W., Harrison, P. A., Johns, T. C., and Downing, T. E.: 1999,
‘Relative Impacts of Human-Induced Climate Change and Natural Climate Variability’, Nature
397, 688–691.
IPCC (Intergovernmental Panel on Climate Change): 1990, ‘First Assessment Report’, in Houghton,
J. T., Jenkins, G. J., and Ephraums, J. J. (eds.), Scientific Assessment of Climate Change – Report
of Working Group I, Cambridge University Press, U.K.
IPCC (Intergovernmental Panel on Climate Change): 1995, in Houghton, J. T., Meira Filho, L. G.,
Callender, B. A., Harris, N., Kattenburg, A., and Maskell, K. (eds.), Second Assessment Report –
Climate Change. The Science of Climate Change, Cambridge University Press, U.K.
Jones, J. W., Jagtap, S. S., and Boote, K. J.: 1999, ‘Climate Change: Implications for Soybean
Yield and Management in the U.S.A.’, in Kaufman, H. E. (ed.), Proceedings of World Soybean
Research Conference VI, August 4–7 1999, Chicago, University of Illinois, Urbana-Champaign,
IL, pp. 209–222.
Jones, J., Boote, K., Jagtap, S., Hoogenboom, G., and Wilkerson, G.: 1988, SOYGRO v5.41 Soybean
Crop Growth Simulation Model, User’s Guide, Florida Agricultural Experiment Station Journal
No. 8304, IFAS, University of Florida, Gainesville, FL.
Kaiser, H. M., Riha, S. J., Wilks, D. S., and Sampath, R.: 1995, ‘Adaptation to Global Climate Change
at the Farm Level’, in Kaiser, H. M. and Drennen, T. E. (eds.), Agricultural Dimensions of Global
Change, St. Lucie Press, Florida.
Kittel, T. G. F., Rosenbloom, N. A., Painter, T. H., Schimel, D. S., Fisher, H. H., Grimsdell, A.,
VEMAP Participants, Daly, C., and Hunt, Jr., E. R.: 1996, The VEMAP Phase I Database: An
Integrated Input Dataset for Ecosystem and Vegetation Modeling for the Conterminous United
States, CDROM and World Wide Web (URL = http://www.cgd.ucar.edu/vemap/).
Lal, M., Singh, K. K., Srinivasan, G., Rathore, L. S., Naidu, D., and Tripathi, C. N.: 1999, ‘Growth
and Yield Responses of Soybean in Madhya Pradesh, India, to Climate Variability and Change’,
Agric. For. Meteorol. 93, 53–70.
Liang, X., Wang, W., and Dudek, M. P.: 1995, ‘Interannual Variability of Regional Climate and its
Change Due to the Greenhouse Effect’, Global Planetary Change 10, 217–238.
Mearns, L. O.: 1995, ‘Research Issues in Determining the Effects of Changing Climate Variability on
Crop Yields’, in Climate Change and Agriculture: Analysis of Potential International Impacts,
ASA Special Publication No. 59, Madison, Wisconsin, U.S.A., pp. 123–143.
Mearns, L. O., Rosenzweig, C., and Goldberg, R.: 1997, ‘Mean and Variance Change in Climate
Scenarios: Methods, Agricultural Applications, and Measures of Uncertainty’, Clim. Change 35,
367–396.
Mearns, L. O., Rosenzweig, C., and Goldberg, R.: 1996, ‘The Effect of Changes in Daily and Inter-
Annual Climatic Variability on Ceres-Wheat: A Sensitivity Study’, Clim. Change 32, 257–292.
Mearns, L. O., Katz, R. W., and Schneider, S. H.: 1984, ‘Extreme High-Temperature Events: Changes
in their Probabilities with Changes in Mean Temperature’, J. Clim. Appl. Meteorol. 23, 1601–
1613.
474 JANE SOUTHWORTH ET AL.
National Agricultural Statistics Service: 1997, Census of Agriculture, http://www.nass.usda.gov/
census.
National Assessment Synthesis Team: 2001, Climate Change Impacts on the United States: The
Potential Consequences of Climate Variability and Change, Report for the U.S. Global Change
Research Program, Cambridge University Press, Cambridge, U.K., p. 620.
Peiris, D. R., Crawford, J. W., Grashoff, C., Jeffries, R. A., Porter, J. R., and Marshall, B.: 1996,
‘A Simulation Study of Crop Growth and Development under Climate Change’, Agric. For.
Meteorol. 79, 271–287.
Phillips, D. L., Lee, J. J., and Dodson, R. F.: 1996, ‘Sensitivity of the U.S. Corn Belt to Climate
Change and Elevated CO2: I. Corn and Soybean Yields’, Agric. Systems 52, 481–502.
Rind, D.: 1991, ‘Climate Variability and Climate Change’, in Schlesinger, M. E. (ed.), Greenhouse-
Gas-Induced Climate Change: A Critical Appraisal of Simulations and Observations, from the
series: Development in Atmospheric Science 19, Elsevier, New York.
Ritchie, J. T., Baer, B. D., and Chou, T. Y.: 1989, ‘Effect of Global Climate Change on Agriculture:
GreatLakesRegion’,inSmith,J.B.andTirpack,D.A.(eds.),The Potential Effects of Global
Climate Change on the United States, EPA-230-05-89-053.
Rosenberg, N. J.: 1992, ‘Adaptation of Agriculture to Climate Change’, Clim. Change 21, 385–405.
Rosenzweig, C. and Hillel, D.: 1998, Climate Change and the Global Harvest: Potential Impacts of
the Greenhouse Effect on Agriculture, Oxford University Press, New York.
Rosenzweig, C., Curry, B., Ritchie, J. T., Jones, J. W., Chou, T. Y., Goldberg, R., and Iglesias, A.:
1994, ‘The Effects of Potential Climate Change on Simulated Grain Crops in the U.S.’, in Rosen-
zweig, C. and Iglesias, A. (eds.), Implications of Climate Change for International Agriculture:
Crop Modeling Study, EPA-230-8-94-003.
Shaw, R. H. and Laing, D. R.: 1966, ‘Moisture Stress and Plant Response’, in Plant Environment and
Efficient Water Use, Agronomy Society of America, Madison, Wisconsin, U.S.A., pp. 73–94.
Schimmelpfennig, D., Lewandrowski, J., Reilly, J., Tsigas, M., and Perry, I.: 1996, Agricultural
Adaptation to Climate Change: Issues of Long-Run Sustainability, Natural Resources and
Environment Division, USDA, AER-740.
Semenov, M. A. and Barrow, E. M.: 1997, ‘Use of a Stochastic Weather Generator in the
Development of Climate Change Scenarios’, Clim. Change 35, 397–414.
Semenov, M. A., Wolf, J., Evans, L. G., Eckersten, H., and Iglesias, A.: 1996, ‘Comparison of Wheat
Simulation Models under Climate Change. II. Application of Climate Change Scenarios’, Clim.
Res. 7, 271–281.
Siqueira, O. J., Farias, J. R. B., and Sans, L. M. A.: 1994, ‘Potential Effects of Global Climate
Change on Brazilian Agriculture: Applied Simulation Studies for Wheat, Maize, and Soybeans’,
in Implications of Climate Change for International Agriculture: Crop Modeling Study,EPA
230-B-94-003.
Southworth, J., Randolph, J. C., Habeck, M., Doering, O. C., Pfeifer, R. A., Rao, D. G., and Johnston,
J. J.: 2000, ‘Consequences of Future Climate Change and Changing Climate Variability on Maize
Yields in the Midwestern United States’, Agric. Ecosyst. Environ. 82, 139–158.
Specht, J. E., Williams, J. H., and Weidenbenner, C. J.: 1986, ‘Differential Responses of Soybean
Genotypes Subjected to a Seasonal Soil Water Gradient’, Crop Sci. 26, 922–934.
Thornton,P.K.,Bowen,W.T.,Ravelo,A.C.,Wilkens,P.W.,Farmer,G.,Brock,J.,andBrink,J.E.:
1997, ‘Estimating Millet Production for Famine Early Warning: An Application of Crop Simu-
lation Modeling Using Satellite and Ground Based Data in Burkina Faso’, Agric. For. Meteorol.
83, 95–112.
United States Department of Agriculture (USDA): 1994, Data Users Guide: State Soil Geographic
Data Base (STATSGO), Nat. Res. Cons. Serv., Misc. Pub. 1492.
Wittwer, S. H., 1995: Food, Climate, and Carbon Dioxide – The Global Environment and World Food
Production,LewisPublishers,NewYork,U.S.A.
CHANGES IN SOYBEAN YIELDS IN THE MIDWESTERN UNITED STATES 475
Wolf, J., Evans, L. G., Semenov, M. A., Eckersten, H., and Iglesias, A.: 1996, ‘Comparison of Wheat
Simulation Models under Climate Change. I. Model Calibration and Sensitivity Analyses’, Clim.
Res. 7, 253–270.
(Received 15 August 2000; in revised form 20 August 2001)