United States
Environmental Protection
Agency
Policy, Planning,
And Evaluation
(2122)
EPA 230-B-94-003
June 1994
Implications Of Climate Change
For International Agriculture:
Crop Modeling Study

-------

-------
Implications Of Climate Change For
       International Agriculture:
               Crop Modeling Study
       Cynthia Rosenzweig and Ana Iglesias, Editors
          United States Environmental Protection Agency
            Office of Policy, Planning and Evaluation
                 Climate Change Division
                   Adaptation Branch
          Ronald Benioff and Joel Smith, Project Officers
                     June 1994

-------

-------
IMPLICATIONS OF CLIMATE CHANGE FOR INTERNATIONAL AGRICULTURE:
                       CROP MODELING STUDY
                                Editors

                      Cynthia Rosenzweig and Ana Iglesias

-------

-------
                              TABLE OF CONTENTS
 PREFACE

 EXECUTIVE SUMMARY

 SECTION 1: INTRODUCTION

       THE USE OF CROP MODELS FOR INTERNATIONAL CLIMATE CHANGE IMPACT
       ASSESSMENT: STUDY DESIGN, METHODOLOGY, AND CONCLUSIONS
       Cynthia Rosenzweig and Ana Iglesias


 SECTION 2: NORTH AMERICA

       EFFECTS OF GLOBAL CLIMATE CHANGE ON WHEAT YIELDS IN THE CANADIAN
       PRAIRIE
       M. Brklacich, R. Stewart, V. Kirkwood, and R. Muma

       THE EFFECTS OF POTENTIAL CLIMATE CHANGE ON SIMULATED GRAIN CROPS IN
       THE UNITED STATES
       C. Rosenzweig, B. Curry, J.T. Ritchie, J.W. Jones, T.-Y. Chou, R. Goldberg, and A. Iglesias

       POSSIBLE IMPACTS OF CLIMATE CHANGE ON MAIZE YIELDS IN MEXICO
       Diana Liverman, Max Dilley, Karen O'Brien, and Leticia Menchaca

SECTION 3: SOUTH AMERICA

       POTENTIAL EFFECTS OF GLOBAL CLIMATE CHANGE FOR BRAZILIAN AGRICULTURE:
       APPLIED SIMULATION STUDIES FOR WHEAT, MAIZE, AND SOYBEANS
       Otavio Joao Fernandes de Siqueira, Jose Renato Boucas Farias, and Luis Marcelo Aguiar Sans

       IMPACTS OF GLOBAL CLIMATE CHANGE ON MAIZE PRODUCTION IN ARGENTINA
       O. E. Sala and J.M. Paruelo

       IMPACT OF CLIMATE CHANGE ON BARLEY IN URUGUAY: YIELD CHANGES AND
       ANALYSIS OF NITROGEN MANAGEMENT SYSTEMS
       Walter E. Baethgen

SECTION 4: EUROPE

       POSSIBLE EFFECTS OF INCREASING C02 CONCENTRATION ON WHEAT AND MAIZE
       CROPS IN NORTH AND SOUTHEAST FRANCE
       R. Del6colle, D. Ripoche, and F. Ruget, G. Gosse

       POTENTIAL EFFECTS OF GLOBAL WARMING AND CARBON DIOXIDE ON WHEAT
       PRODUCTION IN THE FORMER SOVIET UNION
       Gennadiy V. Menzhulin, Larisa A. Koval, Alexander L. Badenko

-------
SECTION 5: AFRICA
      IMPACT OF CLIMATE CHANGE ON SIMULATED WHEAT AND MAIZE YIELDS IN EGYPT
      H.M. Bid

      IMPLICATIONS OF CLIMATE CHANGE FOR MAIZE YIELDS IN ZIMBABWE
      Paul Muchena

SECTION 6: ASIA

      IMPLICATIONS OF GLOBAL CLIMATE CHANGE FOR AGRICULTURE IN PAKISTAN:
      IMPACTS ON SIMULATED WHEAT PRODUCTION
      Ata Qureshi, and Ana Iglesias

      IMPACT OF CLIMATE CHANGE ON SIMULATED WHEAT PRODUCTION IN INDIA
      D.Gangadhar Rao and S.K.Sinha

      IMPACT OF  CLIMATE CHANGE  ON THE PRODUCTION OF MODERN RICE  IN
      BANGLADESH
      Z. Karim, M. Ahmed, S.G. Hussain, and Kh.B. Rashid

      IMPACT OF CLIMATE CHANGE ON SIMULATED RICE PRODUCTION IN THAILAND
      C. Tongyai

      CLIMATE IMPACT ASSESSMENTFOR AGRICULTURE IN THE PHILIPPINES: SIMULATION
      OF RICE YIELD UNDER CLIMATE CHANGE SCENARIOS
      Crisanto R. Escano and Leandro V. Buendia

      EFFECTS OF CLIMATE  CHANGE ON RICE  PRODUCTION AND STRATEGIES FOR
      ADAPTATION IN SOUTHERN CHINA
      Zhiqing Jin, Daokou Ge, Hua Chen, and Juan Fang

      IMPLICATIONS OF CLIMATE CHANGE FOR JAPANESE AGRICULTURE: EVALUATION
      BY SIMULATION OF RICE, WHEAT, AND MAIZE GROWTH
      Hiroshi Seino

 SECTION 7: AUSTRALIA

      POSSIBLE  EFFECTS OF GLOBAL  CLIMATE  CHANGE ON WHEAT  AND RICE
      PRODUCTION IN  AUSTRALIA
      Brian D. Baer, Wayne S. Meyer, and David Erskine

-------
PREFACE

       This publication presents the crop modeling research conducted over the period 1989-1992 for the
U.S. Environmental Protection Agency Climate Change Division project "Implications of Climate Change for
International Agriculture: Global Food Production, Trade and ^Vulnerable Regions." Additional support was
provided by the US Agency for International Development. Principal Investigators of the project were Cynthia
Rosenzweig, of Columbia University and Goddard Institute for Space Studies, and Martin  Parry,  of the
Environmental Change Unit at Oxford University.
       At U.S. EPA Climate Change Division, we thank Dennis Tirpak, Joel Smith, Ron Benioff, and Joel
Scheraga; and at U.S. AID, Tej Gill. Dr. Roy Jenne provided the GCM climate change scenarios. At Columbia
University, we thank Ruth Levenson, Chris Shashkin, and Rich Goldberg. We especially acknowledge the
IBSNAT crop modeling team, Bruce Curry, Gerrit Hoogenboom, Jim Jones, Joe Ritchie, Upendra Singh, and
Gordon Tsuji, for worldwide scientific and technical support.
       Finally, the editors and authors are grateful to the reviewers of the reports: E. Alocilja, D. Bachelet,
T. Carter, S. Cohen, Y. Congyi, E. Cooler, G. Edmeades, J. Gerber, J. Goudriaan, A Hall, H. Hamdy, T.
Hodges, S. Hollinger, E.  Kanemasu, J. Kiniry, G. Knapp, R. Loomis, R. Luxmore, H.  Nix, S. Obrien, D.
Pottker, R. Puluektov, G.  Rocca da Cunha, J. Russell, B. Smit, Z. Uchijima, J. Wescoat, E. Wheaton, and D.
Wilkes.
       Special thanks to Shelly Preston for her editing prowess and dedication to the project.

-------

-------
EXECUTIVE SUMMARY

        The central aim of the study was to provide an assessment of potential climate change impacts on
world crop production, including quantitative estimates of yield and water-use changes  of major crops.
Agricultural scientists in  18 countries estimated potential changes in water use and crop growth, using
compatible crop models and consistent climate change scenarios. The crops modeled were wheat, rice, maize
and soybean. Wheat, rice and maize account for approximately 85% of world cereal exports; soybean accounts
for about 67% of trade in  protein cake equivalent. Site-specific estimates of yield changes were aggregated to
national levels for the modeled major crops for use in a world food trade model, the Basic Linked System.
        The study assessed the implications of climate change for world crop yields taking into account:
(a) Uncertainty in the level  of climate change expected. Sensitivity tests were conducted with arbitrary increases
in temperature and changes in precipitation. The effects of three climate change scenarios were also tested
using doubled CO2 equilibrium climates from three general circulation models: the Goddard Institute for Space
Studies (GISS), the Geophysical Fluid Dynamics Laboratory (GFDL) and the United Kingdom Meteorological
Office (UKMO) models. These climates are assumed to occur in 2060. Finally, a transient projection of climate
change  was tested, based on the GISS transient run A, for the 2010s, 2030s, and 2050s.
(b) Physiological effects of CO2. The climate change scenarios were tested with and without the direct effects
of CO2 on crop growth and water use, as reported in experimental literature.
(c) Different adaptive  responses. Climate change impacts on crop yields incorporating farm-level adaptation
were simulated, based on  different assumptions about shifts in crop planting dates, changes in crop variety,
level of irrigation, etc.

        The principal results that emerged from the study were:

        Sensitivity tests. A 2  C temperature rise increased aggregated crop yields on a global basis, while a 4  C
rise  led to decreases in globally aggregated crop yields. Nevertheless, in semi-arid and subtropical regions, a
2C temperature increase caused yield declines. The greatest yield decreases are caused by a 4C temperature
increase and 20% precipitation decrease.

        GCM climate change scenarios without adaptation. Without physiological CO2 effects, production of
all three crops  decreased  compared to baseline climate conditions on a global basis. With CO2 effects, yields
were positive at middle  and high latitudes,  and negative at low  latitudes for the scenarios with lower
temperature  increases (~4C global surface  air temperature increase).  For the warmest scenario  (~5C
temperature  increase),  crop yields declined almost everywhere. Thus, increases in potential yield  depend
strongly on full realization of the direct effects of CO2 on crop growth.

        GCM climate change scenarios with adaptation. Farm-level adaptation (shifts in planting dates, changes
in crop variety, application of irrigation) reduces the negative effects of climate change. However, even when
farmer adaptation is taken into account, climate change may decrease yields in semi-arid, subtropical regions.
Successful adaptation often implies significant changes to current agricultural systems.

         In order to minimize possible adverse consequences to climate change worldwide, the agricultural
sector should continue to develop crop breeding and management programs for heat and drought conditions
(these  will be  immediately  useful in improving productivity in marginal environments  today). Another
important activity is to  enlarge, maintain, and screen crop  genetic resources at established seedbanks.
Resilience of the agricultural production sector  also depends on improved use of systems  for monitoring
weather, soil moisture, nutrient requirements, and pest infestations. Finally, strong communication links among
the  agricultural research,  production, and policy sectors are essential.

-------


-------
SECTION 1: INTRODUCTION

-------

-------
             THE USE OF CROP MODELS FOR
INTERNATIONAL CLIMATE CHANGE IMPACT ASSESSMENT:
   STUDY DESIGN, METHODOLOGY, AND CONCLUSIONS


                      Cynthia Rosenzweig

       Columbia University and Goddard Institute for Space Studies
                        New York, USA

                            and

                         Ana Iglesias

            Inst. Nacional de Investigaciones Agrarias (INIA)
                        Madrid, Spain
                          INTRO-1

-------
TABLE OF CONTENTS

SUMMARY
INTRODUCTION
BACKGROUND AND PREVIOUS STUDIES
CLIMATE CHANGE SCENARIOS
       Sensitivity Tests
       GCM Equilibrium and Transient Scenarios
       Limitations
CROP MODELS
       Description
       Physiological CO2 Effects
       Limitations
CALIBRATION AND VALIDATION
CROP MODELING PROCEDURES
ESTIMATION OF NATIONAL YIELD CHANGES
DESIGN OF THE ADAPTATION STUDY
       Planting Date
       Irrigation
       Fertilizer
       Crop Variety
       Cropping Area
SOURCES OF UNCERTAINTY
RESULTS
       Sensitivity Tests
       GCM Scenarios
       Transient Scenarios
       Adaptation Studies
CONCLUSIONS AND FUTURE RESEARCH NEEDS
                                     INTRO-2

-------
SUMMARY

        The methodology for an assessment of potential climate change impacts on world crop production,
including quantitative estimates of yield and water use changes for major crops, is described. Agricultural
scientists in 18 countries estimated potential changes in crop growth and water use using compatible crop
models and consistent climate change scenarios. The crops modeled were wheat, rice, maize and soybean. Site-
specific estimates of yield changes for the major crops modeled were aggregated to national levels for use in
a world food trade model, the Basic Linked System. The study assessed the implications of climate change for
world crop yields for arbitrary and GCM equilibrium and transient climate change scenarios. The climate
change scenarios were tested with and without the direct physiological effects of CO2 on crop growth and water
use, as reported in experimental literature.  Climate change impacts on crop yields incorporating farm-level
adaptation were simulated, based on different assumptions about shifts in crop planting dates, changes in crop
variety,  and level of irrigation.
INTRODUCTION

   Scientists predict significant global warming in the coming decades due to increasing atmospheric carbon
dioxide and other trace gases (IPCC 1990a; 1992). Substantial changes in hydrological regimes are also forecast
to occur. Understanding the potential effects of these changes on agriculture is an important task, because
agriculture provides food for the world's population, now estimated at 5 billion and projected to rise to 10
billion in the coming century. Despite technological advances such as improved crop varieties and irrigation
systems, weather and climate are still key factors in agricultural productivity. For example, weak monsoon rains
in 1987 caused large shortfalls in crop production in India, Bangladesh, and Pakistan, contributing to reversion
to wheat importation by these countries (World Food Institute 1988). Despite adequate supplies elsewhere,
the 1980s also saw the  continuing deterioration of food production in Africa, caused in part by persistent
drought and low production potential. This resulted in international relief  efforts to prevent widespread
famine. These examples emphasize the close links between agriculture and climate, the international nature
of food trade and food security, and the need to consider the impacts of climate change in a global context.
    Recent research has been focused on regional and national assessment of the potential effects of climate
change on agriculture (IPCC 1990b). The methodology for regional and national climate impact studies has
thus been developed and tested.  However, the studies have, for the most part, treated each region or nation
in isolation, without relation  to changes in production in other places. The purpose of this study was to
increase understanding of potential simultaneous changes in production in all  major food-producing regions,
because such changes may lead to altered world supply and demand (and prices), and hence competitiveness
in any given region. Such understanding should aid in the meaningful interpretation of regional climate change
impact studies.
        The study1 was  an international collaborative effort of agricultural scientists in 18 countries. A suite
of dynamic process models, climate change scenarios, and simulation experiments was assembled, comprised
of climate sensitivity tests and climate change scenarios devised from global climate models (GCMs), dynamic
crop growth models, and a world food trade model (Figure 1). Common methodology was developed for the
    'The study was commissioned by the U.S. Environmental Protection Agency Climate Change Division.
 The U.S. Agency for International Development provided support and additional funding for the crop
 modeling simulations.

                                              INTRO-3

-------
  simulation of climate change impacts on major agricultural crops and for the analysis of the results. The
  International Benchmark Sites Network for Agrotechnology Transfer (IBSNAT1989) compatible crop growth
  models were utilized at over 100 sites (Figure 2). The crop models can simulate the direct physiological effects
  of increased atmospheric CO2 on crop growth and water use. They allow for the simulation of both rainfed
  and irrigated agricultural systems and other potential farmer adaptations  to climate change. Results of the
  dynamic crop growth simulations were then used to estimate global changes in crop yields for use in a world
  food trade model, the Basic Linked System (Fischer et al. 1988). Results of the entire study are described in
  Rosenzweig and Parry (1994).

  BACKGROUND AND PREVIOUS STUDIES

         Research on the effects of climate on agriculture has been extensive for many years. Much of the
  previous research on climate impact assessment sought to isolate the effects of climate on agricultural activity,
  whereas lately there has been a growing emphasis on understanding the interactions of climatic, environmental,'
  and  social factors in a wider context (Parry et al. 1988). An example of the earlier approach is  the U.S.
  Department of Transportation's study on the possible effects of atmospheric ozone depletion which (in part)
  used regression models to determine statistical relationships between climate and agriculture, a method which
  does not explain the processes underlying the relationships. The National Defense University also studied long-
  term effects of climate change on crop yields and agricultural production with a relatively simple cause-and-
  effect approach (NDU 1980).
         More integrated approaches to climate change impact assessments have been described by Callaway
 et al. (1982), the Carbon Dioxide Assessment Committee, Nix (1985), and Warrick et al. (1986). These studies
 and reviews advocate comprehensive research methods that integrate economic and political factors as well
 as biophysical ones. Some integration has been achieved in national impact studies completed in the United
 States (Adams et al. 1990; Smith and Tirpak 1989), Canada (Smit 1989), Australia (Pearman 1988), the  UK
 (UK Department of the Environment 1991), and New Zealand (Martin et al. 1990), and in regional studies
 including high-latitude and semi-arid agricultural areas (Parry et al. 1988) and the U.S. Midwest (Rosenberg
 and Crosson 1991).                                                                               5
        These regional and national studies have been summarized in the Intergovernmental Panel on Climate
 Change (IPCC) Working Group II Report (IPCC 1990b), but integrated global assessments of climate change
 impacts have been few to date. Kane et al. (1991) have analyzed the sensitivity of world agriculture to potential
 climate changes and found that  the overall effect of moderate climate change on world and domestic
 economies may be small as reduced production in some areas is balanced by gains in others.  These estimates
 did not consider the agricultural impacts of climate change associated with the higher end of the predicted
 IPCC-predicted range of warming  (1.5 to 4.5 C mean global surface air temperature rise) (IPCC 1990a).
 Leemans and Solomon (1993) found that climate change would affect the yield and distribution of world crops
 for one GCM doubled CO2 climate change scenario, leading to production increases at high latitudes and
 production decreases in low latitudes.

 CLIMATE  CHANGE SCENARIOS

        Because of the uncertainties surrounding prediction of climate change, it is common to employ climate
scenarios (Wigley 1987; Lamb 1987), in order to estimate the impacts of climate change on a system (in this
case agricultural production, both potential and actual, and world food trade). Climate scenarios are sets of
climatic perturbations which are used with impact models to test the sensitivity of the system to the projected
changes. The design of an impact study should include more than one scenario, so that a range of possible
effects may be defined. Realism is augmented when the climate change scenarios are internally consistent, i.e.,
                                            INTRO-4

-------
the climate variables within the scenario should vary in a physically realistic way (Wigley 1987). Common
doubled CO2 GCM climate change scenarios and sensitivity tests were devised for the global crop modeling
analysis.

Sensitivity Tests

       An approach to analyze the possible impacts of different climate on crop yield  is to specify
incremental changes to temperature and precipitation and to apply these changes uniformly to the baseline
climate. Arbitrary climate  sensitivity tests were conducted to  test crop model responses to a range  of
temperature (+2 and +4C) and precipitation changes (+/- 20%). The scenario, +2C with a precipitation
decrease of -20%, is interesting because it assumes a possible temperature increase combined with a large
decrease in the amount of rain, which is not simulated under the GCM climate scenarios. Sensitivity studies
allow the consideration of the question: "What type, magnitude, and rate of climate change would seriously
perturb the agricultural system in question?"

GCM Equilibrium and Transient Scenarios

       Scenarios are often  devised by changing an original set of climatological data by prescribed anomalies.
These anomalies may be derived from historical climate or from global climate models. GCMs provide the
most advanced tool for predicting the potential future climatic consequences of increasing radiatively active
trace gases in a consistent manner. However, climate models have not yet been validated to project changes
in climate variability, such as changes in the frequencies of drought and storms, even though these could affect
crop yields significantly.
       Mean annual changes in climate variables from doubled CO2 simulations of three GCMsGoddard
Institute for Space Studies (GISS, Hansen et al  1983), Geophysical Fluid Dynamics Laboratory (GFDL,
Manabe and Wetherald 1987), and United Kingdom Meteorological Office (UKMO, Wilson and Mitchell
1987)were applied to observed (baseline) daily climate to create climate change scenarios for each site (Table
1). GCMs were used to create climate change scenarios because they produce climate variables that are
internally consistent; thus they allow for comparisons between or among regions.
       The method used the differences between lxCO2 and 2xCO2 monthly GCM temperatures, and the ratio
between 2xCO2 and lxCO2 monthly precipitation  and solar radiation amounts (lxCO2 refers to modeled
current climate conditions and 2xCO2 refers to the modeled climate that would occur with an equivalent
radiative forcing of doubled CO2 in the atmosphere). Temperature ranges of the three GCMs are near the high
end of the IPCC range of predicted warming for doubled CO2 (IPCC 1990a).  In general, the GCMs predict
increases in global precipitation associated with warming because warmer air can hold more water vapor.
       Most of our knowledge concerning the climate response to greenhouse-gas forcing has been obtained
from equilibrium response GCM experiments. These are experiments which consider the steady-state response
of the model's climate to step-function changes in atmospheric CO2. Recent  evidence from a few GCM
experiments incorporating  time-dependent greenhouse-gas forcing suggest that  there may be important
differences between the equilibrium and transient  responses (Hansen et  al. 1988, Bryan et al. 1988; IPCC
1992).
       This  study also considered a set of transient climate scenarios (as opposed to the atmospheric
equilibrium scenarios),  derived from the GISS transient climatic simulations (Hansen et al. 1988)  for the
2010s, 2030s, and  2050s, and assuming CO2 concentrations of 405, 460, and 530 ppm, respectively. Transient
scenarios for each site were developed by using the same procedure as that used for the equilibrium scenarios.

Limitations

                                            INTRO-5

-------
        Current global climate models which have been used for CO2 studies employ grids on the order of 4
 latitude by 5 longitude, or greater. At this resolution, many smaller scale elements of climate are not properly
 represented, such as warm and cold fronts and hurricanes, as well as the diversities of ecosystems and land-use.
 Accurate modeling of hydrological processes is particularly crucial for determining climate change impacts on
 agriculture, but GCM simulation of infiltration, runoff, and evaporation, and other hydrological processes is
 highly simplified. Precipitation, in particular, is sometimes poorly represented in GCMs results.
        There  is also uncertainty in the prediction of rate and magnitude of climate change. Ocean heat
 transport is a key, but not well understood, process that affects how fast the climate may warm. The doubled
 CO2 climate change scenarios used in this study have assumed an abrupt doubling of the CO2 concentration
 in the atmosphere and then have allowed the simulated climate to come to a new equilibrium. This step
 change in the atmosphere is unrealistic, since trace gases are increasing gradually. The transient scenario does
 simulate the response of gradually increasing radiatively active gases and is more realistic in this regard.

 CROP MODELS

 Description

        The crop modeling study estimated how climate change and increasing levels of carbon dioxide may
 alter yields and water use of world crops in both major production  areas and vulnerable regions. The crops
 modeled were wheat, rice, maize, and soybeans. Table 2 shows the percentages of world production modeled
 in this study for wheat, rice, maize, and soybean. Even though only two countries (Brazil and USA) simulated
 soybean production, their combined  output  accounts for 76% of world total. Less of the  total world rice
 production was simulated than total production of the other crops. This is because India, Indonesia, and
 Vietnam have significant rice production not included in the study.  Together, these crops account for more
 than 85% of the world traded grains and legumes, although only approximately 4% of rice produced is traded
 compared to about 20% of wheat. Rice is included in the study because of its importance to the food security
 of Asia.
        The IBSNAT models employ simplified functions to predict the growth of crops as influenced by the
 major factors  that affect yields,  i.e., genetics,  climate  (daily solar radiation, maximum and minimum
 temperatures, and precipitation), soils, and management  (IBSNAT 1989). The models used were CERES-
 Wheat (Ritchie and Otter 1985; Godwin et al. 1989), CERES-Maize (Jones and Kiniry 1986; Ritchie et al.
 1989), CERES-Rice (both paddy and upland) (Godwin et al.  1993), and SOYGRO (soybean)  (Jones et al.
 1989).
        The IBSNAT models were selected for use in this study because they have been validated over a wide
 range of environments (e.g., Otter-Nacke et al. 1986) and are not specific to any particular location or soil
 type. Thus they are suitable for use in international studies in which crop growing conditions differ greatly.
 The validation  of the CERES and SOYGRO models  over different environments also serves to enhance
 predictive capability concerning the climate change scenarios, in cases when predicted climates are similar to
 existing climates in other regions. Furthermore, because management practices, such as cultivar, planting date,
 plant population, row spacing, and sowing depth, may be varied in the models, they permit experiments that
 simulate management adjustments by farmers to climate change.
        Modeled processes include phonological development, i.e., duration of growth stages, growth of
vegetative and reproductive plant parts, extension growth of leaves  and stems, senescence (aging) of leaves,
 biomass production and partitioning among plant parts, and root system dynamics. The CERES and SOYGRO
 models also  have the capability to simulate the effects  of nitrogen deficiency and  soil-water deficit on
 photosynthesis and pathways of carbohydrate movement in the plant.
                                             INTRO-6

-------
Physiological CO2 Effects

        Most plants growing in atmospheric CO2 that is higher than ambient levels exhibit increased rates of
net photosynthesis (i.e., total photosynthesis minus respiration). High CO2 also reduces the stomatal openings
of some crop plants. By so doing, CO2 reduces transpiration per unit leaf area while enhancing photosynthesis.
Thus it often improves water-use efficiency (the ratio of crop biomass accumulation or yield and the amount
of water used in evapotranspiration). Experimental effects of CO2 on crops have been reviewed by Acock and
Allen (1985) and Cure (1985). In a compilation of greenhouse and other experimental studies, Kimball (1983)
estimated a mean crop yield increase of 33 +/- 6% for a doubling of CO2 concentration from 300 to 600 ppm
for a range of important agricultural crops.
        In order to project the impact of increasing CO2 on agricultural production, these beneficial direct
effects should be considered along with the climatic effects of the radiatively active trace gases. The assessment
of the relative contributions of the direct effects of CO2 and the predicted climate changes to agricultural crop
responses remains a crucial research  question. GCM predictions show that climate warms in virtually all
regions, while hydrologic regimes may become either wetter or drier. Thus, the climatic effects on crop yields
may thus be either negative or positive depending on location; this study tested whether the beneficial direct
effects (as estimated from experimental studies) may compensate for negative climate change impacts in a
variety of crop-growing environments  around the world.
        The IBSNAT  models  have been modified to  simulate  the changes  in photosynthesis and
evapotranspiration caused by higher levels of CO2. These modifications (based on methods derived from Peart
et al. (1989)) were used in the crop yield/climate change scenario modeling to study the relative magnitudes
of the direct physiological and the climatic effects of increased CO2. Ratios were calculated between measured
daily photosynthesis and evapotranspiration rates for a canopy exposed to a range of high CO2 values, based
on published experimental results (Allen et al. 1987; Cure and Acock 1986, and Kimball 1983). Instantaneous
midday values were then modified to give daily integrated increases, allowing for lower light intensities in
morning and evening. In the crop models, the photosynthesis ratios (Table 3) were applied to the maximum
amount of daily carbohydrate production which is based on incoming solar radiation.
        To account for the effect of elevated carbon dioxide on stomatal closure and increased leaf area index,
and hence on potential transpiration, the evapotranspiration formulation of the IBSNAT models was changed
to include a ratio of transpiration under elevated CO2 conditions to that under ambient conditions. To derive
the ratio, Peart et al. (1989) applied the Penman-Monteith equation (as written in France and Thornley 1984)
to the same canopy and environment, except  for differing  CO2 concentrations. The leaf resistances were
calculated as a function of the differing CO2 concentrations using equations developed by Rogers et al. (1983)
based on experimental data for maize and soybean (used for all C3 crops) (Table 3). The ratio procedure
results in a lower transpiration rate  for higher  CO2  levels on  a daily per unit leaf area basis.  Seasonal
evapotranspiration, however, may not change proportionately, and may even increase, because of the greater
leaf area grown under elevated CO2 conditions.
        The simulation of direct CO2 effects for soybeans, wheat, and maize under current climate conditions
have been compared to experimental  results (Peart et al. 1989; Jones and Allen, pers. comm.; Rosenzweig
1990). The wheat and soybean results compare  well with experimental results, but maize simulations tended
to overestimate yield increases due to  high CO2 at sites with low annual precipitation.
        Rates of future emissions of trace gases,  as well  as when the full magnitude of their effects will be
realized, are unknown. For this study, CO2 concentrations are estimated to be 555 ppm in 2060 (based on
Hansen et al. 1988). Because other greenhouse  gases  besides CO2 (e.g., methane  (CH4), nitrous oxide (N2O),
and the chlorofluorocarbons (CFCs)) are also increasing, an "effective CO2 doubling" has been defined as the
combined radiative forcing of all greenhouse gases having the same forcing as doubled CO2 (usually defined
as 600 ppm). The effective CO2 doubling will occur around the year 2030, if current emission trends continue.


                                             INTRO-7

-------
The climate change caused by an effective doubling of CO2 may be delayed by 30 to 40 years or longer, hence
the projections for 2060 in this study.

Limitations

        The IBSNAT models contain many simple, empirically-derived relationships that do not completely
mimic actual plant processes. These relationships may or may not hold under differing climatic conditions,
particularly the higher temperatures predicted for global warming. For example, most of the data used to
derive the relationships  in the  crop models were obtained with temperatures  below 35 C whereas  the
projected temperatures for doubled  CO2 are often 35  or even  40 C during the growing  period. Other
simplifications of the crop models are that weeds, diseases,  and insect  pests are controlled; there are no
problem soil conditions such as high salinity or acidity; and there are no catastrophic weather events such as
heavy storms. The crop models simulate the current range of agricultural technologies available around the
world; they do not include potential improvements in such technology, but may be used to test the effects of
some potential improvements, such as improved varieties and irrigation schedules.

CALIBRATION AND VALIDATION

        Individual investigators calibrated and validated the IBSNAT crop models using local experimental
data, where possible. Where such procedures were not possible, previous calibrations of cultivars based on the
IBSNAT minimum dataset methods were relied on, as well as previous validations. The IBSNAT crop models
have been created with the express purpose of broad applicability across a wide range of environments.
        The validation procedures and experimental data used for the 18 country studies are described in the
following chapters.  In general, genetic coefficients for different crop genotypes were  estimated from  data
gathered in local agricultural experimental stations. Each parameter was calibrated directly from observed
results (phase duration, biometric ratio, and growth rates) in order to obtain rough parameter values. These
values were then used in model runs and adjusted in order to attain a pseudo-best fit of observed data.  Overall,
validation showed acceptable results in the experiments conducted.

CROP MODELING  PROCEDURES

        The participating agricultural scientists carried  out a set of crop modeling simulation experiments for
baseline climate, GCM doubled CO2 and transient climate change scenarios with and without the physiological
effects of CO2 and sensitivity tests. This involved the following tasks:

1.      Define the geographical boundaries of the major production regions of the country, and estimate the
        current production of major crops in  those regions.

2.      Provide observed climate data for representative sites within these  regions, for the baseline period
        (1951-1980), or for as many years of daily data as  are  available,  and specify the soil, crop, and
      .  management inputs necessary to run the crop models at the selected sites.

3.      Validate the crop models with experimental data from field trials.

4.      Run the crop models with baseline data and climate  change scenarios, with and without the direct
        effects of CO2 on crop growth, with irrigated production, sensitivity tests, and adaptation responses
          for example, shifts in  planting date and crop varieties. Report modeled yield changes and other
                                             INTRO-8

-------
        results, e.g., changes in crop growing season arising from climate change.

5.      Identity and evaluate alterations in agricultural practices that would lessen any adverse consequences
        of climate change.

        The chapters that comprise the following sections of this volume describe the agricultural system that
was modeled, the methods and results of the crop modeling work, including adaptation responses, and the
implications of the projected  climate changes, yield and water use changes, and adaptation strategies for
agriculture in the 18 individual countries.

ESTIMATION  OF NATIONAL YIELD CHANGES

        In order for the crop modeling results at individual sites to be used in the world food trade study, the
initial task was to scale up from site results to changes in national crop yields. This was done first for the crops
and countries in the crop modeling study. Then these results were extended to other crops based on agronomic
characteristics,  and to  other countries and regions based on similarities in agro-ecological zones, previous
climate change impact studies, and on comparison of GCM climate change scenarios for a full complement
of global estimates of potential impacts of climate change on crop yields.
        Crop model results from over 100 sites in the 18 countries were  aggregated by weighting current
regional production to produce national yield change estimates. This is an essential intermediate step that
permits the extrapolation of results from crop modeling experiments at individual sites  to national yield
changes for the food trade model.  The agricultural scientists in each country selected sites representative of
major  agricultural regions, described the  agricultural practices of the regions, and provided  regional and
national production data for estimation of regional contributions to the national yield changes. All the crop
modeling  aggregation results used to develop the BLS estimates were either calculated by the agricultural
scientists themselves or developed jointly with them.
        A region is defined as an "area within a country where there are homogeneous agricultural practices,
soils, and  climate." In the most complete national studies, enough sites were modeled to represent all the
major agroecological regions. In other cases, modelers were asked to analyze the sites modeled and the regions
in their country in order to extend the results as appropriately as possible. The regional yield estimates
represent  the current mix of rainfed and irrigated acreage, the current crop varieties, nitrogen management
and soils,  as provided  by the  country participants. In most cases only one crop variety and one soil were
modeled at each site.
        A database was created for the current regional and national production for the 18 countries with crop
model  results, primarily provided by the participants. Another source of production data was the FAO (1988).
For the USA,  the  data source was the USDA (Crop Production Statistical Division); for the former Soviet
Union the data source was the USDA International Service.
        Results were aggregated similarly for the 11 countries where wheat was modeled (Table 4). There are
large differences among national results; for example, Brazil, Egypt, Pakistan, and Uruguay show significant
decreases  in wheat yields even  with the direct effects of CO2, while simulated wheat yields in Canada, and the
USSR  primarily increase under the climate change scenarios with the direct effects of CO2. The other crops
were aggregated using the same methodology.
        The crop  yield estimates  incorporate some major  improvements: 1) consistent  crop simulation
methodology and climate change scenarios; 2) weighting of model site results by contribution to regional and
national, and rainfed and irrigated production; and 3) quantitative foundation for estimation of physiological
CO2. effects on crop yields. Another set of estimates incorporating the effects of farmer adaptation to climate
change was also produced. All results forwarded to the world food trade model were in  terms of percent


                                             INTRO-9

-------
change from current yields, rather than absolute values. Analysis of relative changes  in yields are more
appropriate given the many uncertainties involved in analysis of climate change impacts. It is important to
note, however, that percent change in yield depends on absolute value of base yields which are different for
the three crops, e.g., soybean base yields are low, so that a high percentage change does not represent a large
absolute decrease in yield.

DESIGN OF THE ADAPTATION STUDY

        Farmers will react dynamically to changing environmental conditions. Country participants tested the
efficacy of several types of adaptations in crop simulation  experiments. The adaptation strategies  tested
involved changes in current management practices, e.g., planting date, fertilizer, and irrigation; and changes
in crop variety either to existing varieties or hypothetical new varieties. Several participants considered the
expansion of crops (e.g., winter wheat in the former Soviet Union and Canada and rice in China) to areas that
are temperature limited under the current conditions. The primary adaptation strategies tested are described
below.

Planting Date

        The most likely response of farmers to warmer temperature would be to plant earlier to utilize the
cooler early season and  to avoid high  temperatures during the grain-filling period. All the crop model study
participants suggested earlier planting as a strategy for adapting to global warming. A relatively small change
in planting date, perhaps up to four weeks, should be easily supportable, but longer shifts in sowing date may
alter the soil moisture (it may be either too wet or too dry) and change the solar radiation (it may be lower)
at planting.  In some country studies (Argentina, Philippines, and China) the sensitivity tests on planting  date
suggested large changes in seasonal agricultural production, implying major changes in the agricultural systems.
In Argentina, planting date shifts of up to 4 months earlier or one month later were suggested by adaptation
simulations.

Irrigation

        The crop model simulations demonstrated that climate change may bring significant increases in the
need for irrigation. In Petrolina, PE, Brazil, where soybeans are currently produced under rainfed conditions,
soybean response to the UKMO climate change was negative even with direct CO2 effects, but positive  with
full irrigation. Climate change may increase demand for irrigation water for crops already under irrigation and
may encourage installation of new irrigation systems, if economic resources are available. Currently only about
15% of the world's agriculture is under irrigation. The potential problems associated with increased irrigation
as an adaptive  strategy are the questionable availability of  water resources, the associated costs, and the
environmental problems of soil salinization and water pollution.

Fertilizer

        An increase in the amount of fertilizer applied can compensate in some cases for yield losses caused
by climate change. Studies in Uruguay and Mexico used the nitrogen module in the IBSNAT crop models to
test adaptation to climate change via fertilizer applications. In Uruguay,  the barley response to  nitrogen
fertilizer under baseline and UKMO conditions was compared. For the baseline runs the currently available
cultivar was used at  the normal planting date. The same cultivar was used for the UKMO runs, but sown at
an earlier date with the physiological effects of increased CO2. The planting date was changed for the UKMO

                                             INTRO-10

-------
runs, because the warmer temperatures of the scenario resulted in a shorter growing season. Four nitrogen
fertilizer rates were used, and the response curve was adjusted with linear regression analysis. Consequently,
the baseline maximum yield was more than 1 t ha"1 higher than the corresponding yield under UKMO
conditions. Also, the amount of N fertilizer needed to attain the maximum grain yield under UKMO was 2.6
times larger than the amount required to attain the same yield under current climatic conditions. These results
may have significant implications for future fertilizer use under climate change conditions at the high end of
the range of predicted warming and indicate an important area for future research.

Crop Variety

        Since most regions are predicted to experience substantial increases in growing season temperature,
country participants tested substitution of existing varieties with higher thermal requirements for currently
grown varieties.  In Mexico, use  of existing cultivars  produces slightly higher yields with the GISS climate
change scenario  (with the direct  effects of CO2), but does not overcome the negative climate change effects.
Some researchers also tested hypothetical new varieties in the crop model simulations, a technique useful for
establishing new breeding objectives. The crop models also allow testing of differing crop types, such as winter
and spring wheats. In the former USSR, a comparison of winter and spring wheat simulations indicates that
winter wheat will respond more  favorably under the GCM climate change  scenarios tested.

Cropping Area

        With further analysis of crop-climate classification, estimation of changes in cropping systems may be
made. In China, climate change as projected by the three GCM scenarios would bring significant shifts in the
rice cropping pattern, based on the extension of the growing period and increased thermal regime during the
rice growing season (Gao et al. 1987). The regions where triple, double, and single-rice crops per year could
be grown would move northward and there would be increased sowing of "indica" rice now grown in southern
China, replacing the current "japonica" types.

SOURCES OF  UNCERTAINTY

        The primary uncertainties in the crop yield modeling depend on the assumptions embedded in the
IBSNAT crop models, as discussed earlier; the methods by which the IBSNAT models were used in the climate
change impact study; and the difficulty of estimating future technological improvements  in agriculture.
        Regarding the use of the IBSNAT models for the climate change study, there are several key points.
Climate data were taken from differing numbers of years and of differing quality in the various country studies,
and artificially generated daily solar radiation were used in the absence of observed  data.  Changes in climate
variability were not simulated.
        Furthermore, the generalized soil characteristics do not encompass all  the wide variety of global
agricultural  soils. The resetting  of the initial profile of soil water  content, in  most cases to  full, at the
beginning of each cropping season leads to underestimation of the impacts of changes in the hydrological cycle
on crop production. Varying levels of nitrogen fertilization were not considered in most of  the country studies,
exceptions  being Argentina, Uruguay and Mexico.  Consistent high levels of fertilization are especially
unrealistic in developing countries. Limited (often only one) cultivars were simulated at  each site, although
in common  practice several to many cultivars are planted in most cropping regions, which might respond
differently to climate change.
        Finally, the crop models simulate the current range of available agricultural technologies. They do not
include potential improvements in such technology, although they may be used (as shown in this study) to test

                                             INTRO-11

-------
the effects of some potential improvements, such as improved varieties and irrigation schedules.

RESULTS

Sensitivity Tests

        While the arbitrary sensitivity tests are dissociated from the processes that influence climate, they
simulate a controlled experiment  and provide better understanding of the factors affecting crop model
responses. They can also help to identify climatic thresholds of critical impacts. Climate sensitivity tests were
carried out in 13 countries for combinations of 0,2, and 4C temperature increases coupled with precipitation
changes of 0%, +20%, and -20%. Changes were considered relative to the baseline yield at 330 ppm CO2.
        Without the direct effects of CO2, crops averaged over all sites showed an increasingly negative
response to  increased temperatures, with percent decreases in yields approximately doubling from the +2 to
+4C cases. When direct CO2 effects are included, wheat, soybean, and rice yields increase about 15% with
a 2C temperature rise, but turn negative at +4C, indicating a possible threshold of compensation of direct
CO2 effects for temperature increases between 2 and 4C as simulated in the IBSNAT crop models (Figure
3).

GCM Scenarios

        Climate change scenarios without the direct physiological effects of CO2 caused decreases in simulated
crop yields in many cases, while the direct effects of CO2 mitigated the negative effects primarily in mid and
high latitudes (Table 4).  Potential changes in national yields (averaged over all commodities in the BLS) varied
for the GISS, GFDL, and UKMO climate change scenarios with the physiological effects of CO2 (Figure 4).
However, latitudinal  differences were apparent in all the scenarios; high latitude changes were less negative
or even positive in some cases, while lower latitude suffered more detrimental effects of climate change on
agricultural  production.
        The GISS and GFDL climate change scenarios produced a range of yield changes from +30 to -30%,
although there were regional differences. The GISS scenario is, in general, more detrimental to crop yields
in Asia and  S. America, and GFDL is more harmful in North America (USA and Canada) and the former
USSR. The UKMO climate change scenario  which has the greatest warming  (5.2 C global surface air
temperature increase) generally causes the largest yield declines (up to -50%).
        The magnitudes of the estimated yield changes vary  by crop. Maize production is most negatively
affected, probably due to its lower response to the physiological effects of CO2 on crop growth, while soybean
is least affected because it responds significantly to increased CO2, at least in the climate change scenarios with
lower estimated mean global surface air temperature warming.           Simulated yield losses are caused
by a combination of factors, depending on location and nature of the climate change scenario. Primary causes
of detrimental impacts on yield are:

1.       Shortening of the growing period (especially grain filling stage)  of the crop. This occurred at some
        sites in all countries.

2.       Decrease of water availability caused by increased evapotranspiration and loss of soil moisture and
        in some cases a decrease in precipitation in the climate change scenarios. This occurred in Argentina,
        Brazil, Canada,  France, Japan, Mexico, and USA.

3.       Poor vernalization. Many temperate crops require a period of low temperature in winter to initiate

                                             INTRO-12

-------
        or accelerate the flowering  process. Low vernalization results in low flower bud initiation and
        ultimately reduced yields. This caused decreases in winter wheat yields in Canada and the former
        USSR.

        Simulated yield increases in the mid- and high-latitudes were caused primarily by:

1.       The positive physiological effects of CO2. At sites with cooler initial temperature regimes, increased
        photosynthesis more than compensated for the shortening of the growing period caused by warming.

2.       The lengthened growing season and the amelioration of cold temperature effects on growth. At some
        sites near the high latitude boundaries of current agricultural  production, increased  temperatures
        extended the frost-free growing season and provided  regimes more conductive to  greater crop
        productivity.

Transient Scenarios

        When the crop models were run with transient climate  changes projected for the 2010s, 2030s, and
2050s from the  GISS transient run  A,  yield responses were  non-linear over time (Figure 5). Aggregated
national wheat yields exhibited the widest range  of effects and were the most non-linear of the three crops,
displayed differing trajectories of change in different regions of the world. Soybean yield changes in the U.S.
and Brazil were the most positive of the three crops, while maize yield changes tended to be the most negative.

Adaptation Studies

        The adaptation studies conducted by the project participants suggest that ease of adaptation to climate
change is likely to vary with latitude (Figure 6). With the existing pool  of cultivars and current resources of
water and fertilizer, agricultural adaptation to climate change seems likely in high and mid-latitude countries,
but seems out of reach for nations in the low latitudes. In tropical and semi-tropical regions, especially semi-
arid zones, the  temperature changes suggested by the GCM scenarios tested are problematic. While soil
moisture deficits can easily be made up with simulated automatic irrigation, the economic and environmental
costs of establishing irrigation systems can be high.

CONCLUSIONS AND FUTURE  RESEARCH NEEDS

        Climate change induced by increasing greenhouse gases  is likely to affect crop yields differently from
region to region across the globe. Under  the climate change scenarios adopted in this study, the effects on crop
yields in mid- and high-latitude regions appeared to be less adverse  than those  in low-latitude  regions.
However, the more favorable effects on yield  in temperate regions depended to  a large extent on  full
realization of the potentially beneficial direct effects of CO2 on crop growth. Decreases in potential crop yields
are likely to be caused by shortening of the crop  growing period, decrease in water availability  due to higher
rates of evapotranspiration, and poor vernalization of temperate cereal  crops. When adaptations at the farm
level were tested (e.g., change in planting date, switch of crop variety,  changes in fertilizer application and
irrigation), compensation for the detrimental effects of climate change was found to be more successful in
developed countries.
        Future research needs include determining how countries, particularly developing countries, can and
will respond to  reduced yields. More detailed adaptation studies in many different locations will help to
address this  need. In order to minimize possible adverse consequences  to climate change worldwide, the

                                              INTRO-13

-------
agricultural sector should be encouraged to continue to develop crop breeding and management programs for
heat and  drought conditions (these  will be  immediately  useful in improving  productivity  in marginal
environments today). Another important activity is to enlarge, maintain, and screen crop genetic resources at
established seedbanks. Resilience of the agricultural production sector also depends on  improved use of
systems for monitoring weather, soil moisture, nutrient requirements, and pest infestations. Finally, strong
communication links among the agricultural research, production, and policy sectors are essential.
                                            INTRO-14

-------
                                         REFERENCES

 Acock, B. and L.H. Allen Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
        and J.D. Cure (eds). Direct Effects of Increasing Carbon Dioxide on Vegetation. U.S. Department of
        Energy. DOE/ER-0238. Washington, D.C. pp. 33-97.

 Adams, R.M., C. Rosenzweig, R.M. Peart, J.T. Ritchie, B.A McCarl, J.D. Glyer, R.B. Curry, J.W. Jones, K.J.
        Boote, and L.H. Allen, Jr. 1990. Global climate change and US agriculture. Nature 345(6272) :219-22.

 Allen, L.H., Jr., K.J. Boote, J.W. Jones, P.H. Jones, R.R. Valle, B. Acock, H.H. Rogers, and R.C. Dahlman.
        1987. Response of vegetation to rising carbon dioxide: Photosynthesis, biomass and seed yield of
        soybean. Global Biogeochemical Cycles 1:1-14.

 Callaway, J.N., FJ. Cronin, J.W. Currie, and J. Tawil. 1982. An Analysis of Methods and Models for Assessing
        the Direct and Indirect Impacts ofCO2-induced Environmental Changes in the Agricultural Sector of U.S.
        Economy. Pacific Northwest Laboratory, Battelle Memorial Institute. PNL-4384. Richland, WA.

 Cure, J.D. 1985. Carbon dioxide doubling responses: A crop survey. In B.R. Strain and J.D. Cure  (eds). Direct
        Effects of Increasing  Carbon Dioxide  on Vegetation.  U.S. Department of Energy. DOE/ER-0238.
        Washington, D.C. pp. 33-97.

 Cure, J.D. and B. Acock.  1986. Crop responses to carbon dioxide doubling: A literature survey. Ag. and For.
        Meteor. 38:127-145.

 Fischer, G., K. Frohberg, M.A Keyzer, and K.S. Parikh. 1988. Linked National Models: A Tool for International
        Food Policy Analysis. Kluwer. Dordrecht. 227 pp.

 FAO. 1988.1987 Production Yearbook. Food and Agriculture Organization. United Nations. Statistics Series
        No. 82. Rome. 351 pp.

 France, J. and J.H.M. Thornley. 1984. Mathematical Models in Agriculture. Butterworths. Boston. 335 pp.

 Gao, L., L. Lin, and Z. Jin. 1987. A classification for rice production in China. Ag. and For. Meteor. 39:55-65.

 Godwin, D., J.T. Ritchie, U. Singh, and L. Hunt. 1989. A User's Guide to CERES-Wheat -  v2.10. International
        Fertilizer Development Center. Muscle Shoals.

 Godwin, D., U. Singh, J.T. Ritchie, and E.C. Alocilja.  1993. A User's  Guide to CERES-Rice. International
        Fertilizer Development Center. Muscle Shoals.

Hansen, J., G. Russell, D.  Rind, P. Stone, A. Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983. Efficient three-
       dimensional global models for climate studies: Models I and II. Monthly Weather Review  111(4):609-
       662.

Hansen, J., I. Fung, A. Lacis, D. Rind, G. Russell, S. Lebedeff, R. Ruedy, and P. Stone. 1988.  Global climate
       changes as forecast by the GISS 3-D model. Journal of Geophysical Research  93(D8):9341-9364.
                                            INTRO-15

-------
IPCC, 1990a. Climate Change: The IPCC Scientific Assessment. Houghton, J.T., G.J. Jenkins, and J.J. Ephraums
        (eds). Intergovernmental Panel on Climate Change. Cambridge University Press. Cambridge. 365 pp.

IPCC, 19905. Climate Change: The IPCC Impacts Assessment. Tegart, W.J. McG., G.W. Sheldon, and D.C.
        Griffiths (eds). Intergovernmental Panel on Climate Change. Australian  Government Publishing
        Service. Canberra.

IPCC, 1992. Climate Change  1992. The Supplementary Report to the IPCC Scientific Assessment. J.T.
        Houghton, B.A. Callander, and S.K. Varney (eds). Intergovernmental Panel on Climate Change.
        Cambridge University Press. Cambridge. 200 pp.

International Benchmark Sites Network for Agrotechnology Transfer (IBSNAT). 1989. Decision Support System
       for Agrotechnology Transfer Version 2.1 (DSSATvZl). Dept. of Agronomy and Soil Science. College
        of Tropical Agriculture and Human Resources. University of Hawaii. Honolulu.
Jones, C.A. and J.R. Kiniry. 1986. CERES-Maize: A Simulation Model of Maize Growth and Development. Texas
       A&M Press. College Station. 194 pp.

Jones, J.W., K.J. Boote, G. Hoogenboom, S.S. Jagtap, and G.G. Wilkerson. 1989. SOYGRO v5.42: Soybean
       Crop  Growth Simulation Model.  Users' Guide.  Department of Agricultural Engineering  and
       Department of Agronomy. University of Florida. Gainesville.

Kane, S., J. Reilly, and J. Tobey. 1991. Climate Change: Economic Implications for World Agriculture. U.S.
       Department of Agriculture. Economic Research Service. AER-No. 647. 21 pp.

Kimball,  B.A. 1983.  Carbon dioxide and  agricultural yield.  An assemblage and analysis of 430 prior
       observations. Agronomy Journal 75:779-788.

Lamb,  P.J. 1987. On the development of regional climatic scenarios for policy-oriented climatic impact
       assessment. Bull Amer. Met. Soc. 68:1116-1123.

Leemans, R. and AM. Solomon. 1993. Modeling the potential change in yield and distribution of the earth's
       crops under a warmed climate. Climate Research 3:79-96.

Manabe, S. and Wetherald, R.T.  1987. Large-scale changes in soil wetness induced by an increase in CO2.
       Journal of Atmospheric Science, 44, 1211-1235.

Martin, R.J., M.J. Salinger, and W.M. Williams. 1990. Agricultural Industries in Climate Change: Impacts on
       New Zealand. Ministry for the Environment. Wellington, pp. 165-173.

NDU. 1980. Crop Yields and Climatic Change for the Year 2000. Vol.  1. Fort Lesley and J.  McNair (eds).
       National Defense University. Washington, DC.

Nix, H.A. 1985. Agriculture. In R.W. Kates (ed). Climate Impact Assessment. SCOPE 27. John Wiley & Sons.
       pp. 105-130.

Otter-Nacke,  S., D.C. Godwin, and J.T. Ritchie. 1986.  Testing and Validating the  CERES-Wheat Model in


                                           INTRO-16

-------
       Diverse Environments. AgGRISTARS YM-15-00407. Johnson Space Center No. 20244. Houston.

Parry, M.L., T.R. Carter and N.T. Konijn (eds). 1988. The Impact of Climatic Variations on Agriculture. Vol 1.
       Assessments in Cool Temperate and Cold Regions. Vol 2. Assessments in Semi-arid Regions. Kluwer.
       Dordrecht. Netherlands. 876 pp. and 764 pp.

Pearman, G. 1988. Greenhouse: Planning for Climate Change. CSIRO. Canberra. 752 pp.

Peart, R.M., J.W. Jones, R.B. Curry, K. Boote, and L.H. Allen, Jr. 1989. Impact of climate change on crop
       yield in the southeastern U.S. A. In J.B. Smith and D.A. Tirpak (eds). The Potential Effects of Global
       Climate Change on the United States. Report to Congress. U.S. Environmental Protection Agency.
       EPA-230-05-89-050. Appendix C. Washington, D.C.

Ritchie, J.T. and S. Otter. 1985. Description and performance of CERES-Wheat: A user-oriented wheat yield
       model. In W.O.  Willis (ed). ARS Wheat Yield Project.  Department  of Agriculture, Agricultural
       Research Service. ARS-38. Washington, D.C. pp. 159-175.

Ritchie, J.T., U. Singh, D. Godwin, and L. Hunt. 1989. A User's Guide to CERES-Maize - v2.10. International
       Fertilizer Development Center. Muscle Shoals.

Rogers, H.H.,  G.E. Bingham, J.D. Cure, J.M. Smith, and K.A. Surano. 1983.  Responses of selected plant
       species to elevated carbon dioxide in the field. Journal of Environmental Quality 12:569-574.

Rosenberg, N.J. and  P.R. Crosson. 1991. Processes for Identifying Regional Influences of and Responses  to
       Increasing Atmospheric CO2 and Climate Change: the MINK Project. An Overview. Resources for the
       Future. Dept. of Energy. DOE/RL/01830T-H5. Washington, D.C. 35 pp.

Rosenzweig, C. 1990. Crop response to climate change in the Southern Great Plains: A simulation study. Prof.
       Geog.  42:20-39.

Rosenzweig, C. and M.L. Parry. 1994. Potential  impact of climate  change  on world food supply. Nature
       367:133-138.

Smit, B. 1989.  Climatic warming and Canada's comparative position in agricultural production and trade.  In
       Climate Change Digest. CCD 89-01. Environment Canada, pp. 1-9.

Smith, J.B. and D.A. Tirpak (eds). 1989. The Potential Effects of Global Climate Change on the United States.
       Report to Congress. U.S. Environmental Protection Agency.  EPA-230-05-89-050. Washington, D.C.
       423 pp.

UK Dept of the Environment. 1991. The Potential Effects of Climate  Change  in the United Kingdom. United
       Kingdom Climate Change Impacts  Review Group. HMSO. London. 124 pp.

Warrick,  R.A., R.M.  Gifford, and M.L.  Parry. 1986. CO2, climatic change and agriculture. Assessing the
       response of food  crops to  the direct effects of increased CO2 and climatic change. In B. Bolin, B.R.
       Dos, J. Jager and R.A. Warrick (eds). The Greenhouse Effect,  Climatic  Change, and Ecosystems.
       SCOPE 29. John Wiley & Sons. New York. pp. 393-473.


                                           INTRO-17

-------
Wigley, T.M.L. 1987. Climate Scenarios. Prepared for the European Workshop in Interrelated Bioclimate and
       Land Use Changes. National Center for Atmospheric Research. NCAR 3142-86-3.

Wilson, C.A. and Mitchell, J.F.B. 1987. A doubled CO2 climate sensitivity experiment with a global climate
       model including a simple ocean. Journal of Geophysical Research 92(13) :315-343.

World Food Institute. 1988. World Food Trade and U.S. Agriculture, 1960-1987. Iowa State University. Ames.
       90pp.
                                           INTRO-18

-------
Table 1.
GCM climate change scenarios.
    GCM
  Year*
Resolution**
    Change in Average Global

Temperature         Precipitation
GISS1
GFDL2
UKMO3
1982
1988
1986
7.83" x 10
4.4 x 7.5
5.0 x7.5
4.2C
4.0 C
5.2C
11%
8%
15%
  *When calculated.
 **Latitude x longitude.
 'Hansen, J. et al. 1983.
 2Manabe, S. and R.T. Wetherald. 1987.
 3Wilson, C.A. and J.F.B. Mitchell. 1987.
                                           INTRO-19

-------
Table 2.        Current world crop yield, area, production, and percent world production aggregated for
               countries participating in study.
 Crop
Yield
Area
Production
Study Countries

Wheat
Rice
Maize
Soybeans
tha'1
2.1
3.0
3.5
1.8
ha x 1000
230,839
143,603
127,393
51,357
txlOOO
481,811
431,585
449,364
91,887
%
73
48
71
76
Source: FAO, 1988.
                                            INTRO-20

-------
Table 3.        Photosynthetic ratios and stomatal resistances used to simulate direct physiological CO2
               effects in the IBSNAT models (555 ppm CO2/330 ppm COJ.

Soybean
Wheat
Rice
Maize
Photosynthesis*
Ratio
1.21
1.17
1.17
1.06
Stomatal Res.**
s m"1
49.7/34.4
49.7/34.4
49.7/34.4
87.4/55.8
               *Based on experimental work reviewed by Cure (1985).
               **Based on experimental work by Rogers et al. (1983).
                                          INTRO-21

-------
     Table 4.        Current production and changes in simulated wheat yields under GCM 2 x CO2 climate
                    change scenarios, with and without the direct effects of CO2l.
                      CURRENT
                    PRODUCTION
CHANGE IN SIMULATED YIELDS
Country
Australia
Brazil
Canada
China
Egypt
France
India
Japan
Pakistan
Uruguay
Former
USSR
winter
spring
USA
WORLD4
Yield
tha'1
1.38
1.31
1.88
2.53
3.79
5.93
1.74
3.25
1.73
2.15

2.46
1.14
2.72
2.09
Area
haxlOOO
11,546
2,788
11,365
29,092
572
4,636
22,876
237
7,478
91

18,988
36,647
26,595
231
Prod.
txlOOO
15,574
3,625
21,412
73,527
2,166
27,485
39,703
772
12,918
195

46,959
41,959
64,390
482
% GISS2 GFDL2 UKMO2 GISS3 GFDL3 UKMO3
Total % % % % % %
3.2
0.8
4.4
15.3
0.4
5.7
8.2
0.2
2.7
0.0

9.7
8.7
13.4
72.7
-18
-51
-12
-5
-36
-12
-32
-18
-57
-41

-3
-12
-21
-16
-16
-38
-10
-12
-28
-28
-38
-21
-29
-48

-17
-25
-23
-22
-14
-53
-38
-17
-54
-23
-56
-40
-73
-50

-22
-48
-33
-33
8
-33
27
16
-31
4
3
-1
-19
-23

29
21
-2
11
11
-17
27
8
-26
-15
-9
-5
31
-31

9
3
-2
4
9
-34
-7
0
-51
-9
-33
-27
-55
-35

0
-25
-14
-13
'Results for each country represent the site results weighted according to regional production.  The world estimates
represent the country results weighted by national production.
*GCM 2xCO2 climate change scenario alone.
3GCM 2xCO2 climate change scenario with direct CO2 effects.
4World area and production x 1,000,000.
                                                 INTRO-22

-------
                                   Crap Mod**

                               ViMl.Rle.Miiz, Soy
                                       CropYMd
                                   by Slta and Scenario

                                      ET. Irrigation
                                     Seaton Length
                        c
    Aggregation of SRaRacuni
   Agroecotogical Zone Analyila
Pravtou* Impact Studiee, QCM Output
J
                                   YWdCtwngEstbnatM

                                   Commodfty Group and
                                      Cou*y/Rglan
                                    WortdFood
                                    Trad* Moda)
                                      ShfftsinTnria
                                   Kldanc* of Food Pwarty ,
Figure 1.        Key elements of crop yield and world food trade study.

-------
Figure 2.       Crop model sites.

-------
                % YIELD CHANGE
                                                           WITH DIRECT CO2 EFFECTS
                               T +2
                      T +4
                            Wheat
Rice
Soybean
Maize
Figure 3.      Aggregated ffiSNAT crop model yield changes for +2C and +4C temperature increase.
              Country results are weighted by contribution of national production to world production.
              Direct effects of CO2 on crop growth and water use are taken into account.

-------
                                                                   WITH DIRECT CO2 EFFECTS
                  GFDL 2XC02
                                    WITH DIRECT COj EFFECTS
                                                                            <*>
                  UKM02XC02
                                    WITH DIRECT COj EFFECTS
                 -4510-30
-2910-15
-14 to 0
1to15
cn
 18 to 30
Figure 4.       Estimated change in average grain yield (wheat, rice, coarse grains, and protein feed) for the

               GISS, GFDL, and UKMO climate change scenarios with direct CO2 effects.

-------
    POTENTIAL CHANGE IN WHEAT YIELD

           G1SS TRANSIENT A and 2XCO2
   PERCENT
                WITH DIRECT CO2 EFFECTS
-30
                                                     CANADA
                                                     =SU
                                                    CHINA


                                                     USTRALIA
                                                    FRANCE

                                                     NDIA
                                                     APAN
                                                    USA
                                        'AKISTAN

                                        JRUGUAY


                                        EGYPT
                                        RAZIL
 BASE
2010s
2030s
2050s
2XCO2
     Figure Sa   Estimated change in average national yields of wheat for the GISS transient
             run A and doubled CO2 climate change scenarios. Direct effects of CO2 on
             crop growth and water use are taken into account

-------
     POTENTIAL CHANGE IN MAIZE YIELD

           GISS TRANSIENT A and 2XCO2
   PERCENT
                WITH DIRECT CO2 EFFECTS
-30
                                                    i JAPAN
                                                     RANGE
                                                     RGENTINA
                                                     RAZIL
                                                     HINA
                                                     MBABWE
                                                     SA
                                                     GYPT
                                                     EXICO
 BASE
2010s
2030s
2050s
2XCO2
     Figure 5b.   Estimated change in average national yields of maize for the GISS transient
             run A and doubled CO2 climate change scenarios. Direct effects of CO2 on
             crop growth and water use are taken into account

-------
POTENTIAL CHANGE IN SOYBEAN YIELD
        G1SS TRANSIENT A and 2XCO2
 PERCENT
WITH DIRECT CO2 EFFECTS
                                                JSA
                                                RAZIL
                                  2050s
                    2XCO2
          Estimated change in average national yields of soybean for the GISS transient
          run A and doubled CO2 climate change scenarios. Direct effects of CO2 of
          crop growth and water use are taken into account

-------
                   UKUO 2XCO2
                                                                       WITH DIRECT COj EFFECTS
                   UKMO 2xC02
                   UKMO 2XC02
                                                                             WITH ADAPTATION 1
                                                                           WITH ADAPTATION 2
                 -4510-30
-29 to -15
                                               -14 to 0
                                                              1 to 15
I       I      %

18 to 30
Figure 6.        Estimated change in average grain yield (wheat, rice, coarse grains, and protein feed) under
                two levels of adaptation for the UKMO climate change with direct CO2 effects.

-------
SECTION 2: NORTH AMERICA

-------

-------
EFFECTS OF GLOBAL CLIMATE CHANGE ON WHEAT YIELDS
                  IN THE CANADIAN PRAIRIE
                            M. Brklacich

              Department of Geography, Carleton University
                           Ontario, Canada
                             R. Stewart

                 Bureau for Environmental Sustainability
                  Agriculture Canada, Ontario, Canada

                            V. Kirkwood

         Centre for Land & Biological Resources Research (CLBRR)
                  Agriculture Canada, Ontario, Canada

                             R. Muma

                  Horticulture & Special Crops Division
                  Agriculture Canada, Ontario, Canada
                            CANADA-1

-------
ACKNOWLEDGEMENTS

       The authors wish to express their collective thanks to Cynthia Rosenzweig, the IBSNAT
team (especially Jim Jones, Upendra Singh, Bruce Curry, and Gerrit Hoogenboom), Anne Mane
Label, and Awegechew Teshome. Our appreciation is also extended to Julian Dumanski, Jim Dyer,
and three anonymous reviewers for their comments on an earlier version of this report.
                                    CANADA-2

-------
TABLE OF CONTENTS

SUMMARY
INTRODUCTION
       Background
       Objectives
DATA, PROCEDURES, AND ASSUMPTIONS
       Baseline Climate Data
       Climatic Change Scenarios Derived from GCMs
       Transient Scenarios
       Sensitivity Analysis
       Crop Model, Cultivars, Management Variables, and Soils
       Simulation of the Physiological Effects of CO2
       Performance of the CERES-Wheat Model in the Canadian Prairie
IMPLICATIONS OF GLOBAL WARMING AND CO2 INCREASES FOR SPRING WHEAT
       Global Warming in Isolation
       CO2 Increases and Global Warming
       Sensitivity Analysis
       The GISS Transient Scenarios
ADAPTIVE STRATEGIES: RESPONDING TO GLOBAL CLIMATE CHANGE
       Irrigation
       Earlier Seeding
       Winter Wheat Conversion
CONCLUSIONS
REFERENCES
                                   CANADA-3

-------
 SUMMARY

         Climate change, as  projected by  three General Circulation Models (GCMs),  caused
 simulated spring wheat yields to decrease at all sites selected for the study in the Canadian prairie.
 Yield decreases were caused primarily by significant increases in temperature, which shortened the
 growing season and resulted in less time for biomass accumulation. When the direct effects of CO2
 were  considered, the results varied with the  scenario. Under the GISS scenario, wheat yields
 increased at most sites, while under the more severe UKMO scenario (hotter and drier), wheat
 yields declined at most sites.
        Adaptation strategies designed to offset the negative effects  of climate  change were
 successful in some areas. Higher temperatures allowed regions in southern Canada to plant winter
 wheat instead of spring wheat, taking advantage of more favorable climatic conditions  to produce
 higher yields, but northern  regions were still too cold to grow winter wheat. Irrigation was found
 to be  an  effective method of adaptation at the driest sites.
 INTRODUCTION

 Background

        Agriculture in Canada is highly diversified, an important influence on the national economy,
 and a major contributor to international export markets. In the Atlantic provinces, livestock, dairy,
 poultry, and potatoes are the major commodities. Central Canada benefits from many advantages,
 including superior soils and climatic conditions, proximity to urban markets, and highly developed
 processing and transportation sectors. The agriculture in this region is diversified and is a major
 contributor to Canadian livestock, field crop, and vegetable production. In  British Columbia,
 agriculture is a relatively small industry, focusing on dairy, poultry, vegetable, and fruit production.
 In the Canadian prairie provinces (Manitoba, Saskatchewan, and Alberta), the agri-food industry is
 a  major  component of  the  economy (51% of  the  total  Canadian  cash  farm receipts). In
 Saskatchewan, agriculture accounts for about 16% of the provincial gross domestic product  (the
 highest percentage of any Canadian province). In the prairie provinces, spring wheat is the main
 commodity (96% of the national production), and most of the crop  is destined for export markets.
 Canada is the third largest wheat exporter in the world.
        Extensive dryland agriculture is the dominant characteristic of farming systems in the prairie
 region. Annual production for the 1981-85 period averaged more than 21  million t, and more than
 11 million ha are used for spring wheat production each year. Spring wheat is important in all three
 prairie provinces, but approximately 55% of total production is from Saskatchewan.
        The potential  impacts of a greenhouse-induced global  climate warming on agroclimatic
 resources, crop yield, and  regional production potential have been previously reported in regional
 studies (Arthur 1988; Bootsma et al. 1984; Brklacich and Smit 1991; Singh and Stewart 1991; Smit
 et al. 1989; Stewart 1990; Williams et al. 1988; Bootsma and de Jong 1988b). Current agroclimatic
 conditions for many regions of Canada are characterized by relatively short periods without frost,
 and therefore a majority of the Canadian research on the agricultural impacts of global warming has
 focused on possible alterations in the growing season properties for annual crops with concomitant
adjustments to productivity levels. As knowledge of climatic processes, climatic change, and climate-
agriculture relationships improve, these assessments need to be  refined. In addition, a study that
goes beyond regional impact studies, considers  the agri-food sector in the broader context of

                                       CANADA-4

-------
national and international economies, and analyzes the extent to which climatic change might alter
comparative advantages among various regions in Canada and its agricultural trading partners is
essential.

Objectives

       This  study is the  Canadian contribution to a project designed to assess the potential
implications of global warming on major world agricultural commodities and their international
trade.  It was coordinated by the U.S. Environmental Protection Agency and the U.S. Agency for
International Development. Contributing countries  were requested to appraise the impacts of a
range of scenarios for global warming on production opportunities for a major export or import
crop. This study focussed on spring wheat production  in the Canadian prairie provinces. The primary
analytical tool used in this study was the CERES-Wheat model (Godwin et al.  1989). Spring wheat
was selected because it has regional and national significance to the Canadian economy and because
Canada is one of the top three wheat exporters in the world. Consequently, changes in climatic (or
other) conditions which would alter Canada's potential for wheat production or the country's
comparative advantage, could have repercussions on national and international levels.
       Wheat production occurs over a vast area in the prairie region which  transverses several soil
and climatic zones. The projected global climatic change scenarios suggest that long-term climatic
alterations may not be distributed uniformly across the prairie region. In order to capture some of
the spatial variability in the region's biophysical properties, seven indicator sites, which include all
major biophysical zones in the prairie provinces, are considered in this study (Figure 1 and Table
1). The extent to which these indicator sites represent the region as a whole is not known, and
therefore this report does not attempt to generalize the findings across the  region.

DATA, PROCEDURES, AND ASSUMPTIONS

Baseline Climate  Data

       Daily maximum and minimum temperatures and daily total precipitation data (from 1951-
80) for the seven sites were obtained from the Canadian national weather archive maintained by the
Atmospheric Environment Service (AES), Environment Canada. Daily solar radiation data were not
available for  the full baseline  period at all sites and  therefore were estimated using the procedure
described in Doorenbos and Pruitt (1977). The method uses daily observations of bright sunshine
hours  (obtained  from the national  archive),  estimates of solar  radiation  at  the  top of  the
atmosphere, and daylength (according to Russelo et al  1974). The observed  and estimated solar
radiation values obtained by this  method are comparable.

Climate Scenarios Derived from GCMs

       Climate change scenarios for each site were generated from three equilibrium General
Circulation Models: the Goddard Institute for Space Studies Model (GISS), (Hansen et al 1983),
the Geophysical Fluid Dynamics  Laboratory Model  (GFDL), (Manabe and Wetherald 1987), and
the United Kingdom  Meteorological Office Model (UKMO), (Wilson and Mitchell 1987). The
method for creating scenarios for each site involved the following procedures: (a) the GCM grid
point located closest to each indicator site was identified; (b) for each of the identified grid points,
the changes in mean monthly temperature were calculated as the difference between the 2XCO2 and

                                       CANADA-5

-------
the control (IXCOj) GCM runs, and for mean monthly precipitation and solar radiation, the change
was calculated as the ratio of the 2XCO2 run to the control run; (c) changes in the monthly mean
values derived under step (b) were then applied to the observed daily baseline record.
        Table 2 shows the temperature and precipitation changes used in this study. Temperature
increases considerably under the three GCMs at all sites, with the largest increases occurring in the
winter. The GISS and GFDL scenarios present similar temperature changes.  In the  southern
Canadian prairies, temperature increases are 3C-6.5C, and the largest increases correspond to the
northern site in Alberta (Fort Vermillion). Summer temperature increases are 2C-3C lower than
winter increases. The UKMO scenario is characterized by the largest temperature increases (5C-
10C). Winter increases are 3C-5C higher than summer increases and are largest in the Alberta
and Manitoba sites.
        The three GCMs estimate precipitation increases for most seasons, but the magnitude of
the increases varies considerably among GCMs and sites. For the critical growing period of spring
wheat (May to August), the GISS model predicts precipitation increases up to 25% above the
current average; winter precipitation  also increases.  The direction of the precipitation changes
predicted by the UKMO model are similar to those projected  by the GISS model, but the increases
in the winter are larger than those in the summer. The GFDL model predicts smaller precipitation
increases than the other two GCMs, and no trend was discernable for the Canadian prairie sites.
Late winter, spring, and summer precipitation changes for the sites in Manitoba tracked 50%-70%
above current averages. For the sites in Saskatchewan, winter  and spring precipitation increases up
to 50%, but there are no significant changes in summer precipitation.
        Solar  radiation changes under the 2XCO2 scenarios (not shown) are not very large in
comparison to the current levels.

Transient Scenarios

        This study also considered a set of transient climate  scenarios  (as  opposed to  the
atmospheric equilibrium scenarios), derived from the GISS transient climatic simulations (Hansen
et al.  1988) for the 2010s, 2030s, and 2050s, and assuming CO2 concentrations of 405, 460, and 530
ppm,  respectively. Transient scenarios for each site were developed by using the same procedure as
that used for the equilibrium scenarios.
        In general, each step in the transient scenario implies a further temperature increase, with
winter increases higher than the summer increases. Except for Fort Vermillion, the largest estimated
increase in temperature occurs between the 2030s and  the 2050s. For the sites in Saskatchewan and
Alberta, precipitation changes mostly during the 2010s.  Solar radiation changes little under the
transient scenarios, with only slight decreases during the winter.

Sensitivity Analysis

        An alternative approach for incorporating changes  to the current  climate is to specify
incremental adjustments to selected climatic variables and to apply these changes  uniformly to the
daily observed weather record. While this approach is removed from the processes that influence
climate, it has the advantages of simulating a controlled experiment and thereby providing a better
understanding of the factors affecting responses. In addition, the approach can also identify climatic
thresholds that could ultimately imply substantial  impacts for agriculture. Combinations of
temperature increases of 0C, +2C, and +4C and precipitation changes of 0%, -20%, and +20%
were considered in this study.

                                       CANADA-6

-------
Crop Model, Cultivars, Management Variables, and Soils

        The crop model used in this simulation study is the CERES-Wheat model (Godwin et al.
1989).  The model simulates  crop responses to  the  major factors driving  plant  growth and
development.  Simulated processes include soil moisture balance, phonological development, yield
and biomass production.
        Cultivars. The area sown with different spring wheat varieties has changed over the past fifty
years. In 1988, the most important varieties were Katepwa (49% of the total area of spring wheat
in the prairie), Neepawa (21%), and Columbus (18%) (Prairie Pools, Inc. 1988). In the first half of
the 1980s, Neepawa and Benito were the most popular varieties. The differences among varieties are
usually related to resistance to pests and diseases, resistance to shattering, and susceptibility to root
rot. For example, the major difference between Katepwa and Neepawa is that the Katepwa "has
better stem and leaf rust resistance and is easier to thresh"  (Saskatchewan Agriculture and Food
1990).
        These important factors, which ultimately affect crop yield,  are not considered by the
CERES-Wheat model. Therefore, given that a few varieties account for the majority of the area
sown with spring wheat and that many of the differences among varieties are beyond the scope of
the CERES-Wheat model, a single variety can adequately represent a substantial proportion of the
varieties currently used in the Canadian prairie. We selected the Manitou variety for the simulation
study, with the genetic coefficients derived by Godwin  et al. (1989). Manitou was an important
variety used in the 1960s and 1970s, and many of the varieties used in the 1980s are derived from
it. The major differences between Manitou and other varieties used in the 1980s are confined to
characteristics not considered in the CERES-Wheat model, and therefore, the genetic coefficients
associated with the Manitou are still applicable (Morrison, personal communication).
        The genetic coefficients for winter wheat used in the adaptation section of this study were
the representative winter wheat genetic coefficients for varieties grown in the northern plains of the
United States (Godwin et al. 1989).
        Management variables.  The seeding dates for spring wheat were derived from Bootsma and
de Jong (1988a) (Table 3). These seeding  dates were estimated using the observed weather record
(1951-80) and  represent the average dates for  this period.  Yearly  variability in the weather
conditions result in a considerable range of seeding dates. For example, the earliest and the latest
seeding dates  estimated  for Lethbridge  between 1951 and 1980 were April 24 and  May 23,
respectively. At Winnipeg the  estimated seeding dates ranged from April 24 to June 9. The
importance of an early or late spring on seeding date, crop growth, and grain yield is beyond the
scope of this study.
        For winter wheat (considered in the Adaptation section of this study), seeding dates were
derived from provincial field crop production guides (Manitoba Agriculture 1988; McLelland 1985a;
Saskatchewan 1981).
        Field crop production guides and agronomists determined the planting densities for the
spring and winter wheats (Manitoba Agriculture 1988; McLelland 1985a,  1985b; Saskatchewan
Agricultural Services Coordinating Committee 1981). The midpoint in the seeding rate was selected,
and seeding rates in kg ha"1 were converted to plants m"2 using an average seed weight of 30 mg (Fei
and Ripley 1985).
        The adaptation study considers the effects of irrigation on wheat yields. In this study,
irrigation was triggered when the soil moisture estimated for the 1.2-meter rooting zone dropped
below 50% of the moisture-holding capacity. It was assumed that the amount of water required to
                                       CANADA-7

-------
return the rooting zone to field capacity was applied at that time, with an irrigation efficiency of
100%.
        Soils. A representative soil for each site was identified in consultation with a prairie-soil
expert (Shields, personal communication). For all sites except for Fort Vermillion, data on horizon
depths, texture, bulk density, organic carbon, coarse fractions, pH, and soil classification were
extracted from the Canada Soil Information System (CanSIS) (Canada-Alberta  1989; Canada-
Manitoba 1989; Canada-Saskatchewan 1989). Permeability, drainage, slope, quantity of roots, and
soil color were estimated by Shields (personal communication). Soil surveys for the Fort Vermillion
area are incomplete, and data on soils for this area are not available from CanSIS. The Donnelly
series has been characterized for  the neighboring Grinshaw and Notikewin areas (Scheelar and
Odynsky 1968), and these data were used in this study.
        The CERES-Wheat model estimates the soil moisture storage capacity from the input soil
profile. This estimate represents the maximum amount of soil moisture that could  be available to
the crop at seeding. The version of the CERES-Wheat model used in this study does not estimate
soil moisture recharge between harvest and seeding, but it does allow users to specify soil moisture
conditions at seeding as a percentage of the estimated storage capacity. Since soil moisture reserves
of prairie soils at seeding are typically well below the moisture-holding capacity, the initial soil water
conditions were adjusted. The estimates of soil-moisture conditions at seeding presented in Table
3 were derived from de Jong and Bootsma (1987). These estimates are long-term, average values for
the entire soil profile and were calculated using the observed weather record for 1951-80, assuming
a continuous wheat-farming system.

Simulation of the  Physiological Effects of CO2

        Higher levels of atmospheric CO2 have been shown to increase photosynthesis and water-use
efficiency, resulting in yield increases in experimental settings (Acock and Allen, 1985). Because the
climate change scenarios are associated with levels of CO2 that are higher than  the current climate
GCM simulations (330 ppm), the physiological effects of alternative CO2 levels  on crops were
included in the crop model simulations.
        Other gases such as CH4, N2O, and  CFCs are expected to contribute to the greenhouse
effect, and therefore,  the equivalent of a 2xCO2 atmosphere would be reached before  the actual
doubling of CO2. To account for this effect, an atmospheric CO2 concentration of 555 ppm was used
to represent the equivalent of a 2xCO2 atmosphere.

Performance of the CERES-Wheat Model in the Canadian Prairie

        Previous studies. The CERES-Wheat model has been used previously to estimate wheat
yields in the Canadian prairies. Fei and Ripley (1985) used the model to estimate wheat yields from
1964-84 for the Saskatoon crop-reporting district (more than 2 million ha of  cropland). After
accounting for technological advances, their main conclusions were:  (a) for the entire period, the
CERES model overestimated observed yields by 24%; and (b) the model tended to overestimate
yields in the good years and underestimate yields in the poor years.
        Fei and Ripley suggested that the following factors could have contributed to the observed
and simulated yield discrepancies: (a) the weather and soil data used as input for the CERES-Wheat
model did not accurately represent the range of weather and soil conditions occurring in the area;
(b)  the model does not account for yield losses due to pest and disease damage; and (c)  the model
underestimates root growth, and therefore, moisture-deficit stress was exaggerated in dry years, as

                                       CANADA-8

-------
there was insufficient root growth to exploit water reserves from the deeper layers. In wet years, the
model would produce excessive above-ground growth, greater leaf area development, and higher
yields. In addition, there are the usual concerns regarding the  reliability of the reported yields. The
estimates are based on samples of farmers'  estimates of yields, and the extent to  which these
estimates are influenced by unreported crop  failures and/or  portions of fields not used for crop
production is not known.
        Current study.  For this study, observed  and simulated yields  and season lengths were
compared (Table 4). These parameters were closely related at all sites, although the modeled season
lengths  tended to be  longer,  with  the greatest  discrepancies occurring at Prince Albert and
Letherbridge. The simulated grain yields at Winnipeg, Dauphin, Swift Current, and Prince Albert
were comparable to previous estimates (Fei and Ripley 1985). On average, the  model tended to
overestimate yield by about 25%. The  model  was  considerably less reliable for the three Alberta
sites, especially for the central and northern Alberta sites (Ellerslie and Fort Vermillion), suggesting
that the version of the model used in this study has not adequately captured the influence of longer
days and lower light intensities associated with these northern latitudes.
        For all the sites, the CERES-Wheat model  simulated season length reasonably well. For
Winnipeg, Dauphin, Swift Current, Prince Albert, and Letherbridge, the model provides reasonable
estimates of long-term grain yields. However, for Ellerslie and Fort Vermillion, the yield-predicting
capability of the CERES-Wheat model needs to be refined.
        These performance characteristics suggest that results from the Canadian analysis should
be applied in similar fashion to the results obtained from GCMs, i.e., as percent differences between
baseline and scenario runs. Relative changes from baseline simulations rather than absolute values
from crop model runs should be used. Impacts on yield in a particular region thus can  be estimated
and compared to results from other scenarios  and regions. The value of the CERES-Wheat model
is that it provides estimates of the direction and magnitude of yield change among different sets of
conditions.  It is not intended to provide an exact  measure of yield nor absolute predictions of
climate  change impacts.

IMPLICATIONS OF GLOBAL WARMING AND CO2 INCREASES FOR SPRING
WHEAT

Global Warming in Isolation

        The 2xCO2 scenarios for climate change provide a basis for comparative static assessments
of the implications of an altered climate on crop yields. This section evaluates the effect  of three
GCM global climate change scenarios on agro-climatic conditions and wheat yields at seven sites
throughout the prairies.
        Maturation Time.  The large temperature increases associated with the GISS, GFDL, and
UKMO scenarios imply a decrease in the time required for spring wheat to mature (Figure 2). For
the indicator sites in the southern portion of the Canadian prairies, the GISS and GFDL scenarios
simulated a growing period that is 11 to 14 days shorter than the current growing period. The
impacts were more  pronounced for  the northern sites, with the GISS and  GFDL scenarios
shortening the growing period by approximately 3 and 4 weeks, respectively. The UKMO scenario
showed the largest decreases in the growing  period due to its higher temperature predictions. A
three-week reduction in the amount of time required for spring wheat to mature was typical in the
southern prairies under this scenario, with a  4-5-week reduction estimated for the northern parts
of the Canadian prairie.

                                       CANADA-9

-------
        Crop Moisture Stress. Each of the climate change scenarios had a different impact on crop
 moisture stress (Table  5). (Crop moisture stress is  the difference between  precipitation  and
 simulated evapotranspiration during the crop growing period.) Under the current conditions, deficits
 in  crop moisture characterize wheat  production throughout the Canadian prairies. The GISS
 scenario had little impact on the magnitude of the moisture deficit accumulated over the maturation
 period of the crop. However, these accumulated deficits, coupled with considerable declines in the
 time required for wheat to mature, resulted in an increase in average deficits per day. Under the
 GFDL and UKMO scenarios, precipitation increased in the eastern sites and decreased in the
 western sites. These altered precipitation patterns eliminated moisture deficits for the sites  in the
 eastern prairie, while the deficits  became somewhat more severe in the western sites. Under the
 UKMO scenario, precipitation increases provided some relief to moisture stress, but the shortened
 maturation period of the crop offset this potential benefit, and an increase in the  average daily crop
 moisture deficit was estimated.
        Yields. The overwhelming trend of the  impacts of all three climate change scenarios on
 wheat yields was negative, but the magnitude of the impacts varied with scenario and site (Figure
 3). The simulated shortening of the season length caused a decrease in the time available for the
 grain filling, and thereby contributed to a decline in crop yields. The GISS and  UKMO scenarios
 had considerably smaller impacts on the driest part of the prairie (Swift Current and Lethbridge),
 with larger declines in crop yields elsewhere. The yield impacts under the GISS scenario in the dry
 areas were negligible, but in the rest of the regions yield decreased about 20%-30%. Under the
 UKMO scenario yield decreases showed similar trends, but the substantially shorter simulated season
 length had a larger effect on yields. Yields decreased about 20% in the dry areas and 45%-60%
 elsewhere.
        In the eastern sites of Winnipeg and Swift Current, there were modest yield increases  under
 the  GFDL scenario, probably due to  the scenario precipitation increases. In  the western sites,
 however,  precipitation decreases, coupled with higher temperatures, created a less favorable regime
 for wheat and led to substantial yield declines.

 CO2 Increases and Global Warming

        This section evaluates the combined effects  of global climate change with the beneficial
 effects  of increased CO2  on crop  yield-improved water-use  efficiency  and increased  net
 photosynthesis (Figure 3).
        Under the GISS scenario, the direct effects  of CO2 compensated  for the yield decreases
under the scenario of climate change alone at all sites. The most noticeable changes were at the
driest sites (Swift Current and Lethbridge), where the benefits of increased water-use efficiency were
larger, and the direct effects of CO2 caused simulated yields to increase by about 40%-50% above
the current level. Yield increases were  about 15% at the other sites.
        The GFDL scenario projected precipitation increases for the eastern sites. The combination
of a more favorable moisture regime and additional CO2 caused simulated wheat yields to increase
above  the current level. The increase was more pronounced in areas that are currently very  dry.
However, at the sites where the GFDL scenario projects precipitation decreases (the western sites),
simulated wheat yields decreased, even with the direct effects of CO2.
        Under the UKMO scenario, yields increased at the driest sites and decreased elsewhere.
Once again, the benefits of increased CO2 on enhanced water-use efficiency were more noticeable
in the driest areas. At the rest of the sites, the benefits of elevated CO2 levels only partially  offset
                                       CANADA-10

-------
the negative impacts of the UKMO scenario conditions on simulated yields, and yields decreased
25%-40%.

Sensitivity Analysis

        Maturation Time. A 2C temperature increase caused the simulated maturation period of
the crop to decrease 7-12 days in the southern sites, and 2-3 weeks in the northern sites (Figure 4).
A 4C increase reduced season length by 2-3 weeks in the southern sites and by more than 30 days
in the northern sites.
        Crop Moisture  Stress. Increased evapotranspiration  rates caused  by the 2C  and 4C
temperature  increases  added to  crop moisture stress  at all  sites  (Table  5).  These  higher
temperatures,  coupled with  a 20% reduction in precipitation, increased the severity of  moisture
deficits  accumulated over the crop-growing period by up to 41%. On the other hand, a  20%
precipitation increase more than offset the negative impacts of a 2C temperature increase, but
seasonal moisture deficits persisted. Increases  in evapotranspiration  due to  a 4C temperature
increase tended to offset a 20% increase in precipitation.
        Yields. A temperature increase alone caused decreases in simulated yields at all sites (Table
6). At Swift Current and Lethbridge, relatively severe moisture deficits currently cause wheat yields
to be low. As a result, the impacts of a shorter time for crop growth and drier conditions  were not
as pronounced at these sites. Of course, the opposite trend is anticipated if the moisture regime
becomes more favorable for wheat production. The yield benefits from a 20% precipitation increase
were greater at the driest sites.
        For all sites except those in the driest areas, a 2C temperature increase offset the benefits
of a 20% precipitation increase, and therefore, simulated yields decreased. A 2C temperature
increase tempered the yield benefits of additional moisture in the driest areas, and simulated yields
increased about 20% under these conditions.
        A 20% precipitation decrease caused substantial declines in wheat yields at all sites. Prairie
agriculture currently suffers from a deficit  of moisture and the simulated yield losses due to the
additional stress associated with a 20% reduction in precipitation ranged from 22%-39%. Higher
temperatures exacerbated the consequences  of reductions in precipitation (-40% and -60% if
precipitation declines are coupled with a +2C  and a +4C, respectively).
        The direct beneficial CO2 effects  offset the yield decreases in some of the scenarios
considered in this sensitivity analysis (Table 6). In the least favorable of the scenarios considered (a
+4C increase combined with a 20% precipitation decrease), yield decreases were  estimated at all
sites, even with  the direct effects of CO2.  In contrast, under the +2C and +20% precipitation
scenario, wheat yields increased substantially at all sites.

The GISS Transient Scenarios

        Simulated wheat yields under the transient scenarios (Figure 5) were nonlinear. For all sites
except Fort Vermillion, there was an initial increase in yields through the 2010s, followed by a
decrease from the 2010s to the 2030s, and then a recovery of the yields from the 2030s to the 2050s.
This suggests that under the GISS transient scenarios, the temperature increase  until the 2010s
would have a relatively small impact on yield, and the beneficial effects of CO2 (405 ppm) would
stimulate  grain  productivity compared to  base levels. However, the influence  of temperature
increases with concomitant reductions in the maturation period of the crop  dominated the impact
by  the  2030s, and in some cases outweighed the benefits  of  the CO2 increase  to 460 ppm.

                                       CANADA-11

-------
 Temperature increases between the 2030s and the  2050s  were small, and  therefore, the yield
 increases associated with this period were a consequence of the increased beneficial effects of rising
 CO2 levels (530 ppm in the 2050s).

 ADAPTIVE STRATEGIES: RESPONDING TO GLOBAL CLIMATE CHANGE

        The analysis in the previous section has isolated the impacts of climate change on wheat
 yields, but it is reasonable to assume that many other biophysical and socioeconomic conditions will
 also change during this time period, and that these adjustments will act in concert with the climate
 change. In this section three possible responses to climate change are considered: irrigation, changes
 in  the planting date, -and  a shift from spring to  winter wheat production. These  adjustments
 represent different levels of economic adaptation, with irrigation being the most expensive. The
 results presented in this section include the beneficial CO2 effects on simulated wheat yields.  An
 appraisal of economic and technical feasibility of each option is beyond the scope of this study, as
 are the changes in other conditions, including the technological advances or adaptive measures taken
 by  farmers or public institutions.

 Irrigation

        Under all three climate change scenarios and at all sites, irrigation was an effective adaptive
 strategy to improve wheat yields under the climate change scenarios. The sites that benefited  the
 most from irrigation are the driest sites: Swift Current and Lethbridge.

 Earlier Seeding

        Sowing spring wheat earlier in the season  would take advantage of cooler temperatures
 during the earlier part of the frost-free period and therefore lessen the negative impacts of climate
 change on yield. This option was most effective at the driest sites. Under the UKMO scenario, which
 predicts the largest yield decreases, earlier seeding compensated for the yield decreases only in the
 driest areas. Under the GISS and GFDL scenarios, earlier seeding compensated for yield losses at
 all  sites except Fort Vermillion and Ellerslie. Overall, the earlier seeding option is the easiest to
 implement, but it was the least effective of the three options considered, and in some cases it did
 not fully offset the negative impacts of climate change on wheat yields.

 Winter Wheat  Conversion

        Winter survival is the main, factor that currently deters winter wheat production in the
 Canadian prairies. Global warming (especially if there is a substantial increase in winter temperature
 and sufficient snow cover to  ensure winter survival) would allow cereal  grains  to  take better
 advantage of spring moisture reserves and thereby contribute to a more favorable set of conditions
 for  winter cereals. A shift from spring to winter wheat  increased yields under GCM climate change
scenarios at the southern sites. Winter wheat would allow a better use of spring moisture supplies
and would diminish the impacts of relatively high summer temperatures. For the northern locations
 (Fort Vermillion, Ellerslie, and Prince Albert), the temperature increases under the GCM scenarios
during the winter were not enough to eliminate cold damage due to winter kill and therefore limited
the effectiveness of a conversion to winter wheat at  these sites.
                                      CANADA-12

-------
CONCLUSIONS

        This study considered the extent to which global climate change, increases in atmospheric
CO2 concentrations to 555 ppm, and selected adaptive strategies would alter simulated wheat yields
at seven sites in the Canadian prairie. The climate  scenarios were derived  from three General
Circulation  Models (GCMs); additional climate scenarios were created by altering the observed
temperature and precipitation records by fixed amounts. An assessment of the crop model used
(CERES-Wheat) under the Canadian prairie conditions indicated that the model can provide an
estimate of the direction and magnitude of yield shifts stemming from a potential  climatic change.
However, the model is not intended to predict reliable yield estimates for a particular year.

        The conclusions of the simulation study were:

1.      Each of the three GCM scenarios used (GISS,  GFDL, and UKMO)  imply a different set
        of agro-climatic conditions for the region.

2.      There is no uniform response to climate change throughout the Canadian prairie.  Each
        climate change scenario  results in different wheat yield response at each site.

3.      Without the beneficial effects of increased CO2, wheat yields decreased under the climate
        change scenarios in comparison with current yields. The shortening of the time required for
        wheat to mature under the climate change scenarios was the main factor responsible for the
        yield decreases.

4.      When the direct effects of CO2  are included  in the simulation, wheat  yields  increased
        compared to current yields at all sites under the GISS scenario; yields  increased only in the
        northern sites under the  GFDL scenario; and yields decreased at all sites except two under
        the  UKMO scenario.

5.      The beneficial effects of increased CO2 have the greatest effect on the driest sites in the
        prairie under the climate change scenarios.

6.      Although the overall impact of global climate change on Canadian wheat yields may not be
        negative, it appears that a considerable redistribution in yield patterns is  possible.

7.      The effectiveness  of possible adaptation strategies varies  from region to region, and
        therefore it is reasonable to expect that no single response strategy will adequately mitigate
        the  possible negative impacts  of climate change on all of the Canadian prairie provinces.
                                       CANADA-13

-------
REFERENCES

Acock, B., and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In
        B.R. Strain and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation.
        DOE/ER-0238. U.S. Department of Energy. Washington, D.C. pp. 53-97.

Arthur, L. 1988. The implication of climatic change for agriculture in the prairie provinces. CCD 88-01,
        Atmospheric Environment Service, Environment Canada, Downsview.

Bootsma, A., and R. de Jong. 1988a. Estimates of seeding Dates of Spring Wheat on the Canadian
        Prairies From Climatic Data. Canadian Journal of Soil Science, 68:513-517.

Bootsma, A., and R. de Jong. 1988b. Climate Risk Analyses of the Prairie Region,  in J. Dumanski
        and V. Kirwood (eds). Crop Production Risks in the Canadian Prairie Region in Relation to
        Climate and Land Resources. Technical Bulletin 1988-5E, Research Branch, Agriculture
        Canada, Ottawa.

Bootsma, A., W. Blackburn, R.  Stewart, R.  Muma,  and  J. Dumanski. 1984. Possible Effects of
        Climatic Change  on Estimated Crop Yields in Canada. LRRI Contribution No. 83-64,
        Research Branch, Agriculture Canada, Ottawa.

Brklacich, M., and B. Smit. 1991.  Implications of Changes in Climatic Averages and Variability on
        Food Production Opportunities in Ontario, Canada, Climatic Change.

Canada-Alberta 1989. Soil inventory Map Attribute File - Alberta: Soil Layer Digital  Data. CLBRR,
        Research Branch, Agriculture Canada, Ottawa.

Canada-Manitoba 1989. Soil inventory Map Attribute File - Manitoba: Soil Layer Digital Data.
        CLBRR, Research Branch, Agriculture Canada, Ottawa.

Canada-Saskatchewan 1989. Soil inventory Map Attribute File - Saskatchewan: Soil Layer Digital Data.
        CLBRR, Research Branch, Agriculture Canada, Ottawa.

de Jong, R., and A. Bootsma. 1987. Estimated Long-term Soil Moisture Variability on the Canadian
        Prairies. Canadian Journal of Soil Science, 68:307-321.

Doorenbos, J., and W. Pr uitt. 1977. Guidelines for Predicting Crop Water Requirements. FAO Irrigation
        and Drainage Paper 24, FAO, Rome.

Fei, Q., and E. Ripley. 1985. Simulation of Spring Wheat Yields in the Saskatoon Crop District 1960
        to 1984  Using  the  CERES-Wheat Growth  Model. SRC Publication No.  E-906-46-B-85,
        Saskatchewan Research Council, Saskatoon.

Godwin, D., Ritchie, J., Singh, U., and Hunt, L. 1989. A User's Guide to CERES Wheat-V2.W.
        International Fertilizer Development  Center. Muscle Shoals.  AL
                                      CANADA-14

-------
Hansen, J., G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983.
       Efficient Three-Dimensional  Global Models for Climate Studies: Models  I and II. April
       Monthly Weather Review, Vol HI, No. 4:609-662.

Hansen, J., I. Fung, A. Lacis, D. Rind,  G. Russell, S. Lebedeff, R. Ruedy, and P. Stone. 1988. Global
       climate changes as forecast  by the GISS 3-D model. Journal of  Geophysical Research
       93(D8):9341-9364.

Manabe, S., and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in
       atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.

Manitoba  Agriculture. 1988. Field Crop  Production  Guide for Manitoba  1988-1990. Manitoba
       Agriculture, Winnipeg.

McLelland, M.  1985a. Winter Wheat Production. AGDEX 112/20-3, Field Crops Branch, Alberta
       Agriculture, Lacombe.

McLelland, M. 1985b. Soft White Wheat Production. AGDEX 112/20-2, Field Crops Branch, Alberta
       Agriculture, Lacombe.

Prairie Pools Inc.  1988.1988 Prairie Grain Variety Survey. Prairie Pools Inc.,  Regina.

Russelo, D., S. Eddy, and J. Godfrey.  1974. Selected Tables & Conversions Used in Agrometeorology
       & Related Fields. Publication  1922, Agriculture Canada, Ottawa.

Saskatchewan Agricultural Services Co-ordinating Committee. 1981.1981 Guide to Farm Practices
       in Saskatchewan. The University of Saskatchewan, Saskatoon.

Saskatchewan Agriculture and Food. 1990. Varieties of Grain Crops for Saskatchewan 1990. Soils and
       Crops branch, Saskatoon.

Scheelar, M., and W. Odynsky. 1968. Reconnaissance  Soil Survey of the Grimshaw and Notikewin
       Area. Report No. 88, Research Council of Alberta, Edmonton.

Singh, B., and R. Stewart. 1991.  Potential  Impacts of a CO2-Induced Climate Change Using the
       GISS Scenario on Agriculture in Quebec, Canada. Ecosystems and Agriculture, 35:327-347.

Smit, B.,  M. Brklacich, R.B. Stewart, R. McBridge, M. Brown, and D. Bond. 1989. Sensitivity of
       Crop Yields and Land Resource Potential to Climatic Change in Ontario, Canada, Climatic
       Change 14 (2):153-174.

Stewart, R. 1990. Possible Effects  of Climatic Change on Estimated Crop  Yields in Canada: A
       Review. In G. Wall and M. Sanderson (Eds.) Climate Change:  Implications for Water and
       Ecological Resources.  Department of Geography, Occasional Paper  No.  11, University of
       Waterloo, Waterloo (pp  275-284).
                                      CANADA-15

-------
Williams, G., H. Jones, E. Wheaton, R. Stewart, and R. Fautley. 1988. Estimating the Impacts of
        Climatic Change on Agriculture in the Canadian Prairies, the Saskatchewan Case Study, in
        M. Parry, T. Carter, and N. Konjin (eds). The Impact of Climatic Variations on Agriculture.
        Volume 1. Assessment in Cold Temperate and Cold  Regions, Kluwer, Dordrecht. (PP221-
        379).

Wilson, C.A., and J.F.B.  Mitchell. 1987. A doubled CO2 Climate Sensitivity Experiment with a
        Global Model Including a Simple Ocean. Journal of Geophysical Research, 92:13315-13343.
                                     CANADA-16

-------
Table 1.        Recent production and land use data (1981-85) for Canadian spring wheat in the
               areas represented by the indicator sites in the Canadian prairie provinces.
 Province/Site
                        Estimated
          Wheat       contribution
  Lat.   Production     to National
Long.    T x 1000    Production (%)
Average
 Yield
 (T/Ha)        Soil
MANITOBA
Winnipeg

Dauphin

SASKATCHEWAN
Swift Current

Prince Albert

ALBERTA
Lethbridge

Ellerslie

Fort Vermillion

3,882
+49.5 N 15
-97.1 W
+51.1 N 4
-100.3 W
11,869
+50.2 N 32
-107.4 W
+53.1 N 23
-105.4 W
5,661
+49.4 N 16
-112.5 W
+53.3 N 7
-113.3 W
+58.2 N 3
-116.0 W

2.24 Black Soils

2.1 Other Soils


1.65 Brown Soils

1.95 Black Soils


1.93 Brown Soils

2.06 Black Soils

1.9 Northern
Soils

                                       CANADA-17

-------
Table 2.        Temperature differences (C) and precipitation changes (%) between GCM lxCO2
               and 2xCO2 climate change scenarios, at selected sites in Canada.

Site/GCM
Winnipeg
GISS
GFDL
UKMO
Dauphin
GISS
GFDL
UKMO
Swift Current
GISS
GFDL
UKMO
Prince Albert
GISS
GFDL
UKMO
Lethbridge
GISS
GFDL
UKMO
Ellerslie
GISS
GFDL
UKMO

Spr.

4.6
5.0
9.3

4.6
5.0
9.5

3.9
5.5
7.1

3.9
5.5
7.1

3.9
5.2
5.5

3.9
5.2
9.0
Temp.
Sum.

2.8
2.9
7.4

2.8
2.9
6.5

3.2
3.6
5.9

3.2
3.6
5.9

3.2
4.2
5.7

3.2
4.2
5.4
Diff. (
Fall

4.8
5.4
8.5

4.8
5.4
8.0

5.9
5.3
7.6

5.9
5.3
7.6

5.9
5.0
7.3

5.9
5.0
7.2
:c)
Win.

5.5
5.0
10.2

5.5
5.0
9.6

6.2
4.9
7.5

6.2
4.9
7.5

6.2
4.9
6.7

6.2
4.9
5.9

Ann.

4.4
4.6
8.8

4.4
4.6
8.4

4.8
4.8
7.0

4.8
4.8
7.0

4.8
4.8
6.3

4.8
4.8
6.9

Spr.

7
70
33

7
70
55

19
59
30

19
59
30

19
38
27

19
38
58
Precip.
Sum.

17
46
-3

17
46
15

9
13
9

9
13
9

9
1
18

9
1
20
Changes
Fall

5
27
2

5
27
24

5
15
17

5
15
17

5
0
19

5
0
34
(%)
Win.

12
-3
22

12
-3
42

30
18
25

30
18
25

30
6
24

30
6
28

Ann.

11
36
9

11
36
27

15
24
17

15
24
17

15
9
21

15
9
31
Fort Vermillion
GISS
GFDL
UKMO
2.9
5.2
8.6
2.7
3.6
5.3
5.1
5.5
7.2
6.9
6.3
7.3
4.4
5.1
7.1
28
19
53
13
-9
23
26
13
41
56
22
47
26
12
35
                                    CANADA-18

-------
Table 3.        Management variables for spring and winter wheat used in the simulation at
               different sites.
                                                                % Soil moisture
Site
Spring Wheat
Winnipeg
Dauphin
Swft Current
Prince Albert
Lethbridge
Ellerslie
Ft. Vermillion
Winter Wheat
Winnipeg
Dauphin
Swft Current
Prince Albert
Lethbridge
Ellerslie
Ft. Vermillion
Seeding

May 13
May 11
May8
May 6
May 7
May8
May 13

Sept 15
Sept 6
Sept 21
Sept 8
Sept 29
Sept 8
Sept 8
Plants/m2

333
333
190
111
172
390
390

283
283
224
277
224
283
283
at seeding1

63
71
49
57
42
61
63








        % of water holding capacity
                                      CANADA-19

-------
Table 4.
Simulated and observed yield (kg ha"1) and season length (days) at selected sites.
       Site
                             Season Length
                           Days      % Diff.
        Yield
kg ha'1     % Diff.
Winnipeg

Dauphin

Swift Current

Prince Albert

Lethbridge

Ellerslie

Ft. Vermillion

Observed
CERES
Observed
CERES
Observed
CERES
Observed
CERES
Observed
CERES
Observed
CERES
Observed
CERES
89
93
90
98
98
100
96
106
90
102
106
111
119
119

+6

+9

+2

+10

+13

+4

0
2,275
2,682
2,531
3,210
1,385
1,612
2,223
3,020
1,930
1,550
2,060
5,450
1,900
3,850

+18

+21

+16

+36

-20

+164

+102
                                       CANADA-20

-------
Table 5.
Impacts of changes in climate on crop moisture stress* (mm).
 Scenario
   WINN
DAUP
      Crop Moisture Stress (mm)
P.A.      S.C.      LETH       ELLE
                                                                                     F.V.
BASE
GISS
GFDL
UKMO
T+2, PO
T+4, PO
T 0, P+20
T+2, P+20
T+4, P+20
T 0, P-20
T+2, P-20
T+4, P-20
-49
-48
15
-44
-50
-59
-13
-26
-40
-61
-64
-70
-67
-60
2
-55
-66
-69
-38
-48
-54
-77
-78
-79
-53
-55
-45
-56
-57
-61
-27
-41
-50
-60
-64
-65
-48
-47
-36
-40
-50
-51
-35
-43
-44
-53
-54
-55
-35
-37
-30
-32
-40
-43
-14
-27
-32
-43
-47
-48
*Precipitation during the crop season length less evapotranspiration simulated
model during the crop season length).
-69
-66
-77
-39
-64
-71
-19
-34
-46
-83
-84
-86
by the CERES- Wheat
-55
-54
-68
-46
-57
-61
-30
-42
-49
-62
-66
-69

                                    CANADA-21

-------
Table 6.
Sensitivity analysis of CERES-Wheat to climate and CO2 changes at selected sites.
VARIABLE
CHANGE
Temp. Precip.
change change
330
+2
+4
0
+2
+4
0
+2
+4
555
+2
+4
0
+2
+4
0
+2
+4
ppm CO2
0
0
+20
+20
+20
-20
-20
-20
ppm CO2
0
0
+20
+20
+20
-20
-20
-20
SIMULATED YIELD CHANGES FROM CURRENT (%)
WINN DAUP P.A. S.C. LETH ELLE F.V.

-22
-44
24
-4
-30
-31
-45
-59

11
-16
71
33
0
3
-19
-39

-29
-30
24
-10
-36
-27
-48
-63

4
-24
69
26
-7
9
-23
-45

-21
-43
27
-2
-28
-32
-45
-58

13
-17
74
33
2
5
-18
-40

-14
-27
40
19
-4
-39
-43
-48

35
10
112
79
42
-1
-12
-24

-12
-26
31
14
-6
-39
-43
-48

31
7
95
62
33
-2
-13
-25

-25
-43
14
-13
-34
-23
-43
-57

2
-21
46
13
-13
11
-16
-37

-25
-47
23
-9
-37
-26
-43
-59

5
-26
66
25
-13
7
-16
-42
                                     CANADA-22

-------
Table 7.
Impact on simulated wheat yield of adaptation strategies to GCM climate change.
      Strategy
                Changes in simulated yield (% from current)
        WINN   DAUP   P.A.     S.C.    LETH    ELLE
IRRIG = changes from dryland to irrigation
-2WK, -4WK and -6WK = changes in the planting date two, four and six weeks earlier
WIN WH = shift to winter wheat.
F.V.
GISS
no change
IRRIG
-2WK
-4WK
-6WK
WIN WH
GFDL
no change
IRRIG
-2WK
-4WK
-6WK
WINWH
UKMO
no change
IRRIG
-2WK
-4WK
-6WK
WINWH

10
68
14
22
24
39

40
60
51
56
58
62

-40
11
-28
-15
-2
13

2
56
3
5
9
21

31
48
39
41
45
49

-32
9
-31
-28
-25
2

14
82
11
12
16
23

13
63
20
27
31
35

-26
47
-25
-25
-24
-25

50
219
66
75
84
132

66
183
103
117
131
167

19
155
25
24
25
48

40
234
72
92
110
200

-7
218
46
90
116
145

17
181
55
68
84
139

-2
18
-6
-10
-12
-1

-26
11
-23
-20
-21
-23

-25
-7
-36
-41
-43
-39

6
76
3
6
13
22

-35
57
-33
-31
-25
-25

-31
39
-37
-42
-46
-42
                                     CANADA-23

-------
                                                                        CANADA
                        FT. YERMILLION
                           ELLERSLIE
                                       PRINCE ALBERT
                      LETHBRIDGE 
                                                DAUPHIN
                                    SWIFT CURRENT __ WINNIPEG
Figure 1.       Location  of the indicator sites in the Canadian prairie provinces: Manitoba
               (Winnipeg, Dauphin), Saskatchewan (Swift Current, Prince Albert) and Alberta
               (Lethbridge, Ellerslie, Fort Vermillion).

-------
                        IMPACT OF CLIMATE CHANGE ON
                                SEASON LENGTH
        CHANGE FROM CURRENT (DAYS)
        I WINN

        OOH LETH
Figure 2.      Impacts of GCM climate change on simulated wheat season length at selected sites
             in the Canadian prairie provinces.

-------
                             CLIMATE CHANGE ALONE

           CHANGE FROM CURRENT YIELD {%)
          i WlNN

          OH LETH
    DAUP

CD ELLE
    P.A.
CD F.V.
                             CLIMATE CHANGE WITH
                         PHYSIOLOGICAL CO2 EFFECTS
          CHANGE FROM CURRENT YIELD (%)
Figure 3.      Impacts of GCM climate change on simulated spring wheat yields (% change from
             current) at selected sites in the Canadian prairie provinces.

-------
                         SENSITIVITY OF DAYS TO MATURITY TO
                         INCREMENTAL TEMPERATURE CHANGE
            CHANGE FROM CURRENT (DAYS)
                        TEMP +2
                            TEMP +4
                 WINN
                 LETH
    DAUP
CD ELLE
P.A.
F.V.
S.C.
Figure 4.       Sensitivity of simulated wheat season length (days to maturity) to incremental
              temperature changes at selected sites in the Canadian prairie provinces.

-------
                     QISS TRANSIENT SCENARIOS: IMPACTS ON
                                  WHEAT YIELDS
          CHANGE FROM CURRENT (%)
            WINN    DAUP
P.A.
                          2010s
        !2030s
LETH    ELLE     F.V.
    I 2050s
Figure 5.      Impact of GISS transient scenarios on simulated wheat yield changes at selected
             sites in the Candian prairie provinces.

-------
THE EFFECTS OF POTENTIAL CLIMATE CHANGE ON SIMULATED
               GRAIN CROPS IN THE UNITED STATES
                              C. Rosenzweig

           Columbia University and Goddard Institute for Space Studies
                             New York, USA

                                B. Curry

                    University of Florida, Gainesville, USA

                               J.T. Ritchie

                       Michigan State University, USA

                               J.W. Jones

                    University of Florida, Gainesville, USA

                               T.-Y. Chou

                      Taiwan National University, Taiwan

                               R. Goldberg

           Columbia University and Goddard Institute for Space Studies
                             New York, USA


                                   and


                                A Iglesias

              Int. Nacional de Investigaciones Agrarias (INIA), Spain
                                  USA-1

-------
 TABLE OF CONTENTS

 SUMMARY
 INTRODUCTION
       Background
       Aims and Scope of Study
       Agricultural Regions and Crops
 STUDY DESIGN
       Climate
       Baseline Climate
       Climate Scenarios
       Crop Models
       Input Data for Crop Models
       Physiological CO2 Effects
       Crop Model Validation
       Simulations
       Limitations of the Study
EFFECTS ON CROP YIELD AND SEASON LENGTH
       Sensitivity Analysis
       GCM Climate Change Scenarios
       Transient Scenarios
EFFECTS ON EVAPOTRANSPIRATION AND IRRIGATION WATER DEMAND
EFFECTS UNDER DIFFERENT MANAGEMENT STRATEGIES- ADAPTATION
CONCLUSIONS                                             ^riAiiuiN
REFERENCES
                                      USA-2

-------
SUMMARY

        The study considered the potential effects of global climate change on wheat, maize, and soybean
production in the United States. Climate scenarios derived from three General Circulation Models (GCMs)
were used in combination with crop growth models to characterize yield and irrigation water demand changes
of the three main crops in major agricultural regions. Under the present management system, projected climate
change caused simulated wheat, maize, and soybean yields to decrease at most sites even when the direct
effects of CO2 were included. These decreases were caused primarily by temperature increases which shortened
the duration of the crop  life cycles, particularly the grain-filling  periods. At some northern sites, yields
increased, probably because crop growth is temperature-limited at these high latitudes. Yield decreases varied
among GCM scenarios. Adaptation strategies were  identified that compensated for the negative effects of
climate change at  some but not all sites.  These strategies included changing planting date and shifting to
cultivars more adapted to  the projected future climate. Patterns of agriculture in the U.S. are likely to shift
as a result of changes in regional crop yields and in  crop irrigation requirements.
INTRODUCTION
Background

        The enhanced greenhouse effect of increased atmospheric concentration of CO2 and other trace gases
could lead to higher global surface temperatures and changed hydrological cycles (IPCC 1990a). Most previous
climate impact assessments suggest significant consequences for agriculture, including shifts in agricultural
zones, changes in irrigation demand, and loss of fertile lands in deltas due to sea-level rise (Parry et al. 1988;
IPCC 1990b). The heavy dependence on North America for world grain reserves (almost 80% of the 1975-77
global marketable surplus) has increased the sensitivity of the world food supply to the climate of that region.
Crop impact studies in the U.S. suggest a range of outcomes depending on scenario and region (Smith and
Tirpakl989.

Aims and Scope of the Study

        The purpose of this study was to investigate the potential effects of climate change on crop yield and
irrigation demand of the three major U.S. crops, based on 19 crop modeling sites that characterize the main
agricultural regions. The results generated from  this study have been used in an economic world trade model
to determine the possible impact of climate change on world cereal production and prices (Rosenzweig and
Parry, 1994). The approach taken  here is  to  compare output of crop  simulations under three equilibrium
GCM, transient, and sensitivity climate scenarios to simulations of present climate and growing conditions.
        The study incorporated simulation  results of the three major crops using consistent methodology so
that comparisons can be made between crops  and regions. Since agricultural production and systems are not
regionally isolated processes, projecting differential regional effects is important. While some crops are not
significant in a particular region at present, this study  tests whether climate change may alter their zonation
in the future. The simulations also  test the differential effects of climate change on winter (wheat) and spring
(maize and soybean) crop production.
        Because the driving force for a greenhouse-gas-induced climate change is the observed increase in CO2
and other gases, the crop models have been modified to simulate the physiological effects of higher
atmospheric CO2 on crop growth and water  use, based on experimental literature. Thus the simulated changes
in crop parameters under the climate change scenarios are driven by two interacting effects, changes in climate

                                               USA-3

-------
 and  CO2 enrichment. It is interesting to compare the responsiveness of each crop  to  higher levels of
 atmospheric CO2. Finally, this study evaluates changes in crop management that may represent ways in which
 farmers adapt to changed climate.

 Agricultural Regions and Crops

        The three most important U.S. crops both in domestic and export markets are wheat, maize, and
 soybean. The U.S. ranks first in world maize and soybean production and third in wheat production. The U.S.
 raises about half of the world total of both maize and soybean. We selected  19 sites in major agricultural
 regions to represent different  agroclimatic and production conditions (Figure 1 and Table 1).
        Although wheat is produced in most states, the most productive regions is the Great Plains  (Figure
 2). About 80% of U.S. wheat production comes from winter wheat; therefore we only simulated winter wheat
 in the study. Two different practices were simulated: rainfed and irrigated. The only major irrigated zone for
 wheat is the southern area of the Pacific Northwest; irrigated wheat areas in California and Arizona  are not
 significant contributors to the national production (Table 2).
        U.S. maize production is centered in the Midwest (Figure 2 and Table 2). Soybean production is
 mainly located in the  eastern two-thirds of  the country, with the heaviest concentration in the Corn Belt
 (Figure 2). Most soybean production occurs  in a rotation plan with maize on the best soils in a given area;
 the highest yields are in Iowa. We estimate that in the soybean and maize growing areas only about  10% of
 the land area is irrigated (Table 2). Although only a small area of grain production in the U.S. is irrigated,
 some regions (especially  the  southeastern part of the Pacific Northwest and California) rely heavily on
 irrigation (Table 2).

 STUDY DESIGN

        The approach  taken was to compare output from the crop models under climate change scenarios
 (either arbitrary changes in climate variables or derived from global climate models) to model output from
 simulations of current  climate. The scenarios derived from GCM results included both equilibrium doubled
 CO2 and transient projections of climate change. Results for both yield and irrigation water use are analyzed
 as  percent change from baseline simulations in order  to provide indications of possible direction and
 magnitude of responses to changed climate conditions.  Additionally the study evaluated possible adaptation
 strategies to climate  change.

 Climate

       Baseline climate. Observed daily climate data (1951-80) were obtained for the  19 sites, consisting of
 daily  maximum and minimum temperatures and precipitation. Daily solar radiation was simulated using a
 weather-generating program, WGEN (Richardson and Wright, 1984). Figure 3 shows the baseline (observed)
 monthly precipitation and temperature regimes at six of the 19 sites.
       Sensitivity scenarios. In order to test the crop models' sensitivity to climate change, fixed combinations
 of temperature (0C, +2C, +4C) and precipitation changes (-20%, 0%, +20%) were tested.
       GCM equilibrium scenarios. Climate change scenarios were created from GCMs because they produce
 climate variables which are internally consistent and because they allow for comparisons  among regions.
 Output from the GCMs is not accurate enough for direct use in crop models on a daily basis. Therefore, mean
monthly changes in climate variables from doubled  CO2 simulations of three GCMs: Goddard Institute for
Space Studies (GISS); Geophysical Fluid Dynamics Laboratory (GFDL); and United Kingdom Meteorological
Office (UKMO), were applied  to observed daily climate records to create climate change scenarios for each

                                              USA-4

-------
site. The GCMs compute mean climatic variables for large (100,000 km2) gridboxes and they do not explicitly
account for variation of these quantities within the gridbox. Therefore no interpolation was made of the GCM
output.
        The projected monthly GCM precipitation and temperatures changes for the U.S. are shown in Figure
3.  All three models  predict significant  warming: the GISS and GFDL scenarios produce comparable
temperature increases (3C to 5C), while the UKMO scenario produces more drastic rises (6c to 9C).
Annual precipitation  generally increases at most sites. (GCMs  predict increases in global precipitation
associated with warming since warmer air can hold more water vapor.) However, the projected precipitation
shows considerable seasonal and geographical variation and is very uncertain.
        Transient scenarios. Most of our knowledge concerning the climate response to greenhouse-gas forcing
has been obtained from equilibrium response GCM experiments. These are experiments which consider the
steady-state response of the model's climate to step-function changes in atmospheric CO2. Recent evidence
from  a few GCM experiments  incorporating time-dependent greenhouse-gas forcing suggest that there may
be important differences between the equilibrium and  transient responses (Hansen et al. 1988, Bryan et al.
1988; IPCC 1992).

Crop Models and Inputs

        Potential changes in crop yields at  the selected sites in the U.S.  were estimated with the CERES-
Wheat, CERES-Maize, and SOYGRO crop models (Ritchie and Otter 1985; Jones and Kiniry 1986; Jones
et al.  1988). The models simulate physiological  crop responses  (water  balance,  phenology, and growth
throughout the season) on a daily basis to the major factors of climate (daily solar radiation, maximum and
minimum temperature, and precipitation), soils, (albedo and  a variety of measures relating to water in the
profile) and management (cultivar, planting date, plant population, row spacing and sowing depth).
        The CERES and SOYGRO models have been validated with experimental data from many locations
encompassing a wide range of environments (Otter-Nacke et al. 1986; Jones and Kiniry 1986; Egli 1992; Jones
and Ritchie 1991).
        All simulations were made using DSSAT, v2.5 (Jones et al.  1990). Changes were computed in the
mean and standard deviation of yield, evapotranspiration (ET), water applied for irrigation, and crop maturity
date.
        Soils. Representative agricultural soils for each site were chosen with reference to the Major Land
Resource Area descriptions; State Soil Conservation Stations, and information provided by County Agricultural
Extension Agents. Soil characteristics for the representative soils were specified by 12 generic soil types (Table
1). The generic soil most representative of each site was used. Soil characteristics included in the model are:
albedo, water drainage, soil evaporation, runoff, and characteristics describing each layer such as depth, lower
and upper limit of plant extractable water, saturated water content, organic carbon, ammonium, nitrate, and
pH values. Since the present model did not simulate a soil water balance during the entire year, we assumed
that the soil moisture level was full at the beginning of each growing season. While this assumption represents
crop planting conditions at most sites in most years in the  U.S., it overestimates soil moisture at planting for
dry years.
        Crop varieties and management. In the CERES and SOYGRO models, crop varieties are defined by
a set  of coefficients that represent characteristics such as  photoperiodism, vernalization, and crop maturity
type. The varieties used in the simulations are representative of common varieties grown in each region (Table
3).  The management variables for the CERES and SOYGRO models were determined  for each location
according to information on current practices provided by  the County Agricultural Extension Agents as well
as state publications (Table 3).
        For the irrigation simulation, the water demand was calculated assuming 100% efficiency of the
                                              USA-5

-------
automatic irrigation system;  a 1-meter irrigation management depth; and automatic irrigation when the
available soil water is 50% or less of capacity.

Physiological Effects of CO2

        Higher levels of atmospheric CO2 have been found to increase photosynthesis and stomatal resistance,
resulting in yield increases in experimental settings (Acock and Allen 1985). Because the climate change
scenarios are associated with concomitant higher levels of CO2 and other trace gases, we have included the
physiological effects of 555 ppm CO2 in the crop model simulations (Peart et al. 1989).

Limitations of the Study

        Climate change scenarios. While GCMs are useful for climate change studies, current climate models
oversimplify many aspects of the climate system, especially ocean dynamics, cloud physics, and land-surface
hydrology. GCMs  do not simulate current climate well at regional scales and they were not designed for
ptedictive regional studies. Therefore, the climate change scenarios created from GCM output must not be
considered as predictions, but only as examples of possible future climates for the regions under study.
        As configured, the climate change scenarios do not alter the patterns of events in the base climate.
Therefore they  dp not  simulate changes in  the underlying variability  (e.g., extended  periods of  high
temperature, droughts, etc.) that can be vital for crops. Dryland yields may be considerably different depending
on whether a change in precipitation results from a change in mean, frequency or intensity. However, the
scenarios created for this study do result in an increased frequency of temperatures above certain thresholds.
        Crop models. In the crop model simulations, technology is held constant; nutrients are not limiting;
weeds, diseases and insect pests are controlled; and there are no problem soil conditions or catastrophic
weather events.  All these assumptions tend to overestimate simulated yields. In  this study an important
assumption is that climatic tolerances of cultivars do not change. It is also important to note that the
physiological effects of CO2  in the crop model  may be overestimated because experimental results from
controlled environments, used to calibrate the model, may not be accurate under windy, and pest-infected field
conditions.

EFFECTS ON CROP YIELD AND SEASON LENGTH

Sensitivity Analysis

        The results of the sensitivity analysis for wheat, maize and soybean for three sites without the direct
effects of CO2 are presented in Figure 4. The main conclusions of the sensitivity study are:

        Increased temperatures cause simulated yield decreases. The greatest percentage yield decreases caused
        by high temperature are found in soybean.

        The three crops are slightly more sensitive to a decrease in precipitation than to an increase.

        In general, a 20% increase in precipitation mitigates yield decreases caused by  a 2C temperature
        increase, resulting in  yield changes that are close to zero. With the same 20% precipitation increase
        and a 4C temperature rise, yield decreases are still significant.

        The greatest percent yield decreases are caused by a 4C temperature increase and a 20% precipitation

                                              USA-6

-------
        decrease.

GCM Climate Change Scenarios

        Tables 4,5, and 6 show the simulated changes in crop yields under the GCM climate change scenarios,
both with and without the direct effects of CO2. Since the driving force for a greenhouse temperature change
is the increase in CO2 and other gases, the most important comparison from an agricultural system response
point of view is between the base (current climate under 330 ppm CO2) and the GCM scenarios with increased
CO2 (555 ppm CC-2).
        Soybeans. Yield changes under the climate change scenario vary with location and scenario considered
(Table 4), but in general soybean yields decrease under  the three climate change scenarios. The largest
decreases occurred under the UKMO scenario and the smallest decreases under the GISS scenario. In some
cases the effect of increased CO2 on photosynthesis was enough to compensate for the yield decreases under
climate change alone. When  the crop was simulated under irrigated conditions, many locations showed an
increase in yield compared to the base (see section on Adaptation). The main cause of the yield decrease is
the shortening of the growing season (by about two weeks) (Table 7).
        Maize. Maize yield decreased significantly at almost every site with the climate change scenarios alone
(Table 5). When the direct effects of CO2 were included, significant yield decreases were still projected for the
GFDL and UKMO scenarios, but negative yield effects of the GISS scenario were completely mitigated. The
yield decreases were caused by the combined effects of high temperature shortening the grain-filling period
and by increased moisture stress. Maturity dates advanced by an average of over two weeks under the three
scenarios (Table 7). In the highest  latitude site (Spokane) there was  a large increase in maize yield as
simulated with all three GCM scenarios; the maize crop appears to be  temperature-limited in the current
climate at this relatively high latitude.
        In the case of climate change with the physiological effects of CO2, simulated maize yields increased
only slightly compared to climate change alone; this is consistent with the lower photosynthetic response of
C4 crops to increased CO2 levels (Acock and Allen 1985).  At all sites except Spokane, irrigated maize yields
decreased significantly even when the direct effects of CO2 were considered.
        Wheat. Wheat was simulated under dryland and irrigated conditions at all locations except Fresno
(California), where dryland wheat produces negligible yields under current climate. Overall, there were large
changes in simulated wheat yields under the climate change scenarios compared to the baseline yield (Table
6). Under the GISS  scenario, modeled wheat yields decreased in every location except in the northernmost
sites, with the largest decreases at the lower latitude sites. Some of the major areas of wheat production, (e.g.,
Dodge City, Abilene) showed very negative yield changes under each  of the climate change scenarios.
        The yield decreases were driven primarily by  increased  temperatures, which caused the duration of
crop growth stages (particularly the grain-filling period) to be shortened (Table 7). Shortening of the  grain-
filling period reduces the amount of carbohydrates available for grain formation and harvestable yield. Maturity
dates of wheat occurred, on average, about two weeks earlier under the scenarios. In the sites  of the Pacific
Northwest region (Spokane, Yakima and Boise), the unusually large simulated dryland wheat yield increases
were probably due to the large increase in winter precipitation projected by the GISS climate change scenario.
Under the GFDL and UKMO scenarios, yields decreased at all but  a few sites. The yield increases in the
northern sites may be due to the positive effects of higher temperatures on growth in colder, high-latitude
sites. In contrast, in the Southern Great Plains,  the temperatures are already high to begin with, and the
additional stress of higher temperatures can lead to drastic yield decreases.
        The addition of the physiological CO2 effects to the simulations caused a significant increase in wheat
yield at all sites. This was due to the beneficial increases in simulated dry matter conversion and stomatal
resistance. Nevertheless, in the lower latitude sites substantial yield decreases were simulated even when the
                                               USA-7

-------
 direct effects of CO2 were taken into account.

 Aggregated Results

        The simulated changes in crop yield at different sites suggest that regional and national production
 may be significantly affected by climate change. In order to estimate these possible changes, we developed an
 aggregation scheme based on actual regional production from 1979-1989  as reported by the USDA Crop
 Reporting Service. A relative "production weighting factor", based on the amount of production of a particular
 crop, was assigned to each site. We also produced a "dryland/irrigation weighting factor" for each site, which
 is an estimation of the percent of harvested area associated with each site which is dryland production (Table
 2).
        Aggregated U.S. crop yield results are shown in Figure 5. The magnitudes of the estimated yield
 changes varied by crop and scenario. When the direct CO2 effects were included, maize (a C4 crop) production
 was most negatively affected, probably due to its lower growth response to higher levels of atmospheric CO2;
 soybean and wheat (C3 crops) production responded significantly to increased CO2. The aggregated production
 results also imply:

        Without physiological CO2 effects, production of all three crops decreased. The warmest scenario
        (UKMO) produced  the largest decreases, with soybean the most severely effected.

        Under the GISS and GFDL scenarios with direct CO2 effects included, soybean production increased,
        wheat production remained unchanged,  while maize production was reduced. Under the UKMO
        scenario all three crops showed decreases in yields.

        Simulated U.S. maize production was negatively affected in all scenarios, both with and without the
        physiological CO2 effects.

 Transient Scenarios

        When the crop models were run with transient climate changes projected for the 2010s, 2030s, and
 2050s from the GISS transient run A, yield responses were non-linear over time (Figure 6). Aggregated wheat
 yield was the most non-linear of the three crops, showing a positive response to climate change in the  2010s
 and 2030s, and a negative response in the 2050s.  Soybean yield changes  were positive throughout the time
 trajectory, and maize yield changes were slightly negative.

 EFFECTS ON EVAPOTRANSPIRATION AND IRRIGATION WATER DEMAND

        The temperature increases  under the GCM climate change scenarios are likely to lead to increased
 daily potential evapotranspiration  and therefore changes in crop water use  and  irrigation demand. The
 following simulation results considered both the effect of climate change and the effect of increases in CO2
 level on crop growth, water use, and irrigation demand. Because the calculation of the irrigation water in the
 CERES and  SOYGRO models assumes nonlimiting situations, only relative changes in crop water use
 requirements are analyzed (Tables 8 and 9).
        Soybeans. Water us.e efficiency (WUE), defined as yield/total crop evapotranspiration (kg/ha mm'1),
declined at most locations under the GFDL and  UKMO scenarios, because yield reductions are relatively
greater than ET reductions (Table 8). Table 9 shows that irrigation demand increased substantially under the
climate change scenarios, with the warmer scenarios producing the largest increases.

                                              USA-8

-------
        Maize. The maize WUE increased almost everywhere under the GISS scenario, in about half of the
sites under the GFDL scenario, but decreased under the UKMO scenario except in the northernmost sites
(Table 8). Under the GISS scenario, irrigation demand decreased at most of the study sites, while under the
more severe GFDL and UKMO scenarios, irrigation demand increased at some sites, while decreasing at other
sites. The decreases in irrigation water requirements (Table 9) have to be carefully interpreted, since they
primarily reflect the decreases in total crop evapotranspiration due to the significant shortening of the crop
growing season and are often accompanied by significant yield reductions.
        Wheat.The wheat WUE values decreased at most sites under the GISS scenario, but increased under
the GFDL and UKMO scenarios. Irrigation water demand for wheat shows very mixed results (Table 9).

EFFECTS UNDER DIFFERENT MANAGEMENT STRATEGIES: ADAPTATION

        Farmers and agricultural systems will, in reality, try to adjust to changing environmental conditions.
This study evaluated changes  in  planting dates, cultivars, and  irrigation as possible  changes  in crop
management  under climate change  conditions. Estimates  of yield changes were  based on crop model
simulations. Changes in economics or domestic agricultural policies were beyond the scope of the adaptation
estimates. Neither the costs of adaptation nor changes in water availability under the climate change scenarios
were considered.
        Soybeans.  Rainfed soybean yields decreased by more than 50% under climate change scenarios (Table
4), but adding irrigation tripled yields. Even though irrigation was found to increase yield, it may not be
possible to use it  as a widespread amelioration tool, since  the  demand would probably  exceed the water
resources available. Only about 10% of U.S. cropping regions is currently irrigated. Establishing irrigation as
an adaptation to climate change could be costly.
        The soybean adaptation results using different cultivars and planting dates showed that adaptation to
temperature changes can be met with existing cultivars in the U.S. Midwest. In the southern part of the U.S.
where temperatures are higher, potential temperature changes suggested by  some GCMs may exceed crop
physiological  tolerance and increase water stress.
        Maize. Changes in planting  date (10,  20, and  30  days  earlier), changes  in variety, or combined
strategies of both were examined for Des Moines. With an earlier  planting date, the yield, with the direct CO2
effect included, could not be restored to current levels. The selection of alternative varieties helped ameliorate
some of the yield  losses, but not completely (Figure 7). The combination of earlier planting dates and a
different variety resulted in  yield increases for the GISS scenario, and for  the irrigated GFDL scenario.
However,  under the UKMO  scenarios, yields still decreased.
        Wheat. When the automatic irrigation is applied, wheat yields improve  over the  baseline in all
locations, except in the southernmost latitudes, with combined climatic and direct effects of CO2 in both GISS
and GFDL scenarios.

DISCUSSION

        Simulated yield decreases occurred in the Southern Great Plains and the Southeast in  response to the
climate change scenarios tested in this study, even when the direct effects of CO2 were included.  In those
regions, high temperatures shortened the duration of crop  stages, especially the  grain-filling period,  and
increased water stress, resulting in moderate to severe yield decreases. The hotter UKMO model produced the
most severe yield  reductions, while the GISS  model was the most benign  of the three GCM scenarios.
Increased temperatures had a beneficial effect (due to a longer frost-free period) on simulated yields at the
northern sites.
        Farmer and agricultural system adaptations (change in planting date, crop variety,  and  irrigation)

                                              USA-9

-------
helped mitigate yield losses. However, successful adaptation to climate change often implied significant changes
to current agricultural systems, and so some adaptations to climate change may be costly. A potential need
for increased irrigation is suggested, particularly at lower latitudes. Currently only a small percentage of the
area of wheat, maize and soybean is irrigated. Whether this would actually change depends on many factors
such as the availability and cost of water, which are beyond the scope of this study.
        Adjusting to a need for increased irrigation water may be difficult. The country's hydrological system
has a complicated legal framework that limits flexibility; in addition the existing irrigation systems may be
subjected to climatic conditions for which they were not designed. Even without climate change, U.S. water-
allocation institutions need to evolve toward programs that encourage a more efficient use of water. Currently
there is little incentive to irrigate without wasting water. The environmental problems (salinity, erosion and
water pollution) derived from increased irrigation also need to be further addressed.
        This study has not addressed the issue of changes in climate variability. Possible changes in climatic
variability under GCM climate change scenarios (such as the magnitude and frequency of droughts, storms,
and heat waves) are important factors in determining production amounts. Since crop yields exhibit nonlinear
responses to heat  and cold stress, changes in the probability of extreme temperature events beyond critical
thresholds can be  significant. Differential effects on minimum and maximum temperatures also need to be
studied. A farmer can adapt to gradual changes in mean  weather variables, but he or she is highly vulnerable
economically to increased variability, e.g., crop failure in one or several years in a row with only adequate yield
in others. Government policies regarding crop insurance also need to be examined.
        Overall, while national production does not appear  to  be at risk, shifts in regional  patterns of
agricultural  are implied by this study. The northern states could  become much more productive for annual
crops such as corn and soybeans because of the lengthening of the frost-free period, while the southern states
could become less productive, due to heat and moisture stress.
                                               USA-10

-------
REFERENCES

Acock, B., and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R.
       Strain and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation. DOE/ER-0238.
       U.S. Department of Energy. Washington, D.C. pp. 53-97.

Egli, D.B. and W. Bruening. 1992. Planting date and soybean yield: evaluation of environmental effects with
       a crop simulation model: SOYGRO. Agric. Forest. MeteoroL 62, 19-29.

Godwin,  D., J. Ritchie, U. Singh, and L. Hunt. 1989. A User's Guide to CERES-Wheat - v2.10. International
       Fertilizer Development Center. Muscle Shoals, USA

Hansen, J., I. Fung, A Lacis, D. Rind, S. Lebedeff, R. Ruedy and G. Russell. 1988. Global Climate Changes
       as  Forecasted by Goddard Institute for  Space Studies Three-Dimensional Model. Journal  of
       Geophysical Research, 93:9341-9364.

Hansen,  J., G. Russell, D. Rind, P. Stone, A Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983. Efficient
       Three-Dimensional Global Models for Climate Studies: Models I and II. Monthly Weather Review, Vol.
       Ill, No.4.

IPCC,  1992. Climate Change 1992. The Supplementary Report to the IPCC Scientific  Assessment.  J.T.
       Houghton, B.A  Callander, and S.K. Varney (eds). Intergovernmental Panel on Climate Change.
       Cambridge University Press. Cambridge. 200 pp.

IPCC. 1990a. Climate Change: The IPCC Scientific Assessment. J.T. Houghton, G.J. Jenkins,  and J.J. Ephraums
       (eds). Intergovernmental Panel on Climate Change. Cambridge University Press.  Cambridge.

IPCC,  1990b. Climate Change: The IPCC Impacts Assessment.  Tegart, W.J. McG., G.W. Sheldon, and D.C.
       Griffiths  (eds).  Intergovernmental Panel on Climate Change. Australian Government Publishing
       Service. Canberra.

Jones, C.A, and J.R. Kiniry, eds. 1986. CERES-Maize: A Simulation Model of Maize Growth and Development
       College Station. Texas A&M University Press.

Jones,  J.W. and J.T. Ritchie. 1991. Crop growth models, p.69-98. Management of Farm  Irrigation Systems.
       American Society of Agricultural Engineers, St. Joseph, Michigan.

Jones,  J.W., K.J. Boote,  S.S. Jagtap, G. Hoogenboom, and G.G. Wilkerson. 1988. SOYGRO v5.41: Soybean
       Crop Growth Simulation Model User's Guide. Florida Agr. Exp. Sta. Journal No. 8304, IFAS. Univ. of
       Florida.

Jones,  J.W., S.S. Jagtap, G. Hoogenboom, and G.Y. Tsuji. 1990. The structure and function of DSSAT. In
       International Benchmark Sites Network for Agrotechnology  Transfer. Proceedings of the IBSNAT
       Symposium: Decision Support System for Agrotechnology Transfer. University of Hawaii.

Manabe, S.,  and R.  Wetherald. 1987. Large-scale changes  of  soil wetness  induced by an  increase in
       atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.
                                             USA-11

-------
Otter-Nacke, S., D.C. Godwin, and J.T. Ritchie. 1986. Testing and Validating the CERES-Wheat Model in
       Diverse Environments. AgGRISTARS YM-15-00407. Johnson Space Center No. 20244. Houston.

Parry, M.L., T.R. Carter and N.T. Konijn. 1988. The Impact of Climatic Variations on Agriculture.   Vol 1
       Assessments in Cool  Temperate and Cold Regions. Vol 2 Assessments in Semi-arid Regions. Kluwer,
       Dordrecht, Netherlands. 876 pp. and 764 pp.

Peart, R.M., J.W. Jones, R.B.  Curry, K.T. Boote, and L.H. Allen. 1989. Impact of climate change on crop yield
       in the southeastern United States: A simulation study. In: J.B. Smith and D.A.  Tirpak (eds). The
       Potential Effects of Global Climate Change on the United States, Appendix C, Agriculture, Vol. 1, EPA-
       230-05-89-053. U.S. Environmental Protection Agency, Washington, DC.

Richardson, C.W., and D.A. Wright.  1984. WGEN: A Model for Generating Daify Weather Variables. ARS-8.
       U.S. Department of Agriculture, Agricultural Research Service. Washington, DC. 83 pp.

Ritchie J., U. Singh, D. Godwin, and L. Hunt. 1989. A User's Guide to CERES-Maize v2.10. Michigan State
       University-IFDC-IBSNAT, Muscle Shoals, AL.

Ritchie, J.T., and S. Otter. 1985. Description and performance of CERES-wheat: A user-oriented wheat yield
       model. In Willis, W.O., ed. ARS Wheat Yield Project. Washington, DC: USDA Agricultural Research
       Service. ARS-38.

Rosenzweig, C. and M.L. Parry. 1994. Potential impact of climate change on world food supply. Nature, 367,
       133-138.

Smith, J.B., and D. Tirpak (eds.): 1989. The Potential Effects of Global Climate Change on the United States.
       Report to Congress.  EPA-230-05-89-050. U.S. Environmental Protection  Agency. Washington, DC.
       423pp.

Wilson, C.A., and J.F.B. Mitchell. 1987. A doubled CO2 Climate Sensitivity Experiment with a Global Model
       Including a Simple Ocean. Journal of Geophysical Research, 92: 13315-13343.
                                             USA-12

-------
Table 1.
Sites selected for the study and representative soils for each site.
Site
Spokane
Yakima
Boise
Fresno
Phoenix
Fargo
North Platte
Topeka
Dodge City
Abilene
Des Moines
Indianapolis
Memphis
Baton Rouge
Columbia
Macon
Tallahassee
Lynchburg
Williamsport
Long.
+47.63
+46.57
+43.57
+36.77
+33.43
+46.90
+41.08
+39.00
+37.46
+32.43
+41.53
+39.73
+35.05
+30.53
+33.95
+32.42
+30.38
+37.33
+41.25
Lat.
-117.53
-120.53
-116.21
-119.72
-112.02
-96.97
-100.41
-96.00
-99.58
-99.68
-93.65
-86.28
-89.98
-91.15
-81.12
-83.39
-84.37
-79.20
-79.92
Soil
DSL
DSL
DSL
DSL
DSnL
DSC
DSn
DSL
DSL
DSnL
DSL
MSL
MSL
MSC
MSnL
MSnL
OrSnL
MSL
MSC
DSL: deep silty loam; DSnL: deep sandy loam; DSC: deep silty clay; DSn: deep sand; MSL: medium silty loam;
MSC: medium silty clay; OrSnL: Orangesburgh sandy loam.
                                            USA-13

-------
Table 2.         Regional wheat, maize and soybean production (% of total national production) and area
                irrigated (% of total grain area irrigated at site or in the region).
WHEAT
Region
Pacific NW


South West

N Mnt Plains
N Great Plains


S Great Plains
Corn Belt

Delta

Southeast



Northeast
Site % p
Spokane 12.0
Yakima
Boise
Fresno 3.0
Phoenix
Fargo 27.0
RPlatte 20.0
Topeka
Dod. C.
Abilene 13.0
Des Mo. 11.0
Indiana.
Memph. 3.0
B.R.
Colum. 2.0
Macon
Tallahas.
Lynch. 2.0
William.
%l
0
100
100
100
100
0
0


0
0
0
0
0
0
0
0
0
0
MAIZE
%P
0.4


0.6

0.4
1.5


2.0
71.0
16.0
0.5

1.5


4.1
2.5
%I
100
100
100
100
100
20
50
50
50
10
10
10
0
0
0
0
0
0
0
SOYBEANS
%P





7.0

7.0

1.0
37.0
28.0
8.0
5.0
2.0
2.0
0.5
2.0
2.0
%I





20

25

25
25
5
15
15
10
15
10
5
0
                                              USA-14

-------


W
fcjj

O
CO



S
s



*-
q


1
03
Q

AH
el
3
o








EH
variables.
WHE

1
w

s ^~
< -^

J i
M ^-
|
* 
Q
PM

1
4-*
"3
0
s

CO
 0
s
t
a

o o
AH AH


 
1 1
s-S

^^
i
T-H


ll
1
s
o
T-H

0 3





cs
VN
C3

CM
vq
t~-
T-H
s
O
' VO

l.a
 "2
on S


0 
11

>,
T-H




1^
^5
T i

3





CM
CN
C3

CM
t>

T-H
o
VO
T-H

.s
00


Phoenix

03 Cd CO
0 O <0
CM C^ rH

en
6
8 o i
66^
oJ"
^H *3
T-H ^,
T-H 10

u 0




f~ oo
C5 C5

rJ tD
C*^ "4"

C3 f"^
S S
8- 8-
C/3 00
T-H T-H

03 CA
.s .s
C3 CO
AH C
^ 5S


1 JS
1 ^ 1
CH Z EH

C3 P3 CC
>o >o >o
T-< T-H CM


6 t2 6
S, -8 &
H^ ^T ( ^3
T ) *O f^j
T-I CM >O

CM CO CM
0 0 0




oo oo oo
c5 c5 d

* vq o

CM oo oo
oo vo o
CM T-H CM

111
co co "Z


o 0|
^^ O O
 c 5
OJJ ^ l*-4

0? ^
>0 >0
T-H T-4

CA
William
Forrest
^ .
u ^
^1 */^
rH 0
d d

oo co
CO ^H

S R
CO CM
VO VO
T-H T-I

S 1
U p


Indianapolis
Memphis

C5 CO
o >o
CM T-H


5^ 18
cQ cd
E^ 
"8 iS
[T j ^q
>O >O
CM T-H

3 0




>0 <0
d d

O\ 10
00 CM

>o o
CM 00
CM T-H
11
vo vo
T-I T-H

1 1
w w


B. Rouge
Columbia

03 W
>o o
T-H CM


II
OQ O
8
PH
OO

^ **^
T-H f^4
CM T-H

8 Q




10 10
f^i f^l

>o oq
CM CO

O O
oo vo
T-H CO
1 
Z CO
VO \D
T-I T-H


t2 w

4_*
Lynchburg
Williamspor
.
3
s
OH
OH
1
1
2
T3
Ji
s 
 ^ *r~*
.2 
5 ?
'S 'fe
51
gO ***
O w
3S
CO
-o t--
Cj ri
3 
S *o if*
3 <3 T-H
OH 03 L
It |
H1^
ir
3 a
O _03
O OH
^ of
a **
o, 
2 
-------
Table 4.
Effects of GCM climate change on dryland and irrigated soybean yields.
Site
DRYLAND
Spokane
Yakima
Boise
Fresno
Phoenix
Fargo
North Platte
Topeka
Dodge City
Abilene
Des Moines
Indianapolis
Memphis
Baton Rouge
Columbia
Macon
Tallahassee
Lynchburg
Williamsport
BASE
(t ha'1)

0.79
0.34
0.38
0.45
0.30
1.35
0.74
2.49
1.42
0.92
2.72
2.43
2.06
2.49
2.41
1.75
3.44
2.42
2.11
GCM
GISS

-11
-26
-16
-53
-57
-10
-12
-31
-49
-62
-7
-12
-40
-43
-19
-24
-23
2
-5
Scenario
GFDL

-32
-35
-47
-49
-93
71
-18
-36
-57
-63
-26
-37
-63
-43
-35
-61
-21
-65
-55
alone
UKMO

-51
-65
-29
-98
-77
-46
-31
-40
-58
-84
-76
-43
-80
-82
-22
-86
-69
-71
-37
GCM Scenario +
GISS

52
12
26
-33
-40
58
50
-6
-26
-32
26
16
-21
-17
4
1
-3
34
34
GFDL

13
-9
-32
-24
-93
150
42
-11
-40
-41
8
-14
-46
-17
-14
-45
0
-48
-29
D.E. CO,
UKMO

-20
-47
-5
-96
-57
-4
16
-18
-42
-73
-64
-23
-72
-72
7
-74
-50
-58
-9
D.E, = Direct physiological effects of 555 ppm CO2
                                             USA-16

-------
Table 4.
(Cont.)
Site
IRRIGATION
Spokane
Yakima
Boise
Fresno
Phoenix
Fargo
North Platte
Topeka
Dodge City
Abilene
Des Moines
Indianapolis
Memphis
Baton Rouge
Columbia
Macon
Tallahassee
Lynchburg
Williamsport
BASE
(t ha'1)

2.66
3.24
2.99
3.32
3.93
2.48
2.55
2.90
3.03
4.15
3.42
3.47
4.03
3.60
3.90
3.90
3.69
3.61
2.86
GCM
GISS

16
-7
-12
-37
-38
15
-4
-14
-65
-36
-1
-2
-27
-15
-12
-14
-13
0
8
Scenario alone
GFDL

13
-3
-28
-36
-37
22
-10
-13
-25
-40
-5
-1
-18
-15
-6
-9
-10
-2
7
UKMO

-2
-40
-31
-68
-37
2
22
-20
-36
-63
-46
-15
-48
-41
-12
-25
-30
-16
-5
GCM Scenario +
GISS

58
20
28
-18
-20
65
47
8
-26
-18
25
21
-10
5
6
4
5
24
38
GFDL

58
28
14
-14
-3
66
44
10
-5
-19
21
22
1
5
14
10
9
22
39
D.E. CO2
UKMO

39
-19
-13
-56
-7
50
30
0
-19
-46
-30
5
-36
-26
7
-6
-2
6
23
D.E. = Direct physiological effects of 555 ppm CO2.
                                          USA-17

-------
Table 5.
Effects of GCM climate change on dryland and irrigated maize yields.
Site
DRYLAND
Fargo
North Platte
Dodge City
Abilene
Des Moines
Indianapolis
Memphis
Baton Rouge
Columbia
Tallahassee
Lynchburg
Williamsport
BASE
(t ha'1)

10.25
6.25
8.36
5.89
11.58
9.74
7.82
6.58
7.00
9.31
8.30
8.55
GCM Scenario alone
GISS

-10
-22
-8
15
-21
-7
-6
10
-28
-5
-58
-21
GFDL

-6
-17
-14
-9
-27
-59
-27
-45
-90
-41
-61
-51
UKMO

-47
-57
-18
-17
-42
-20
-44
-14
-28
-34
-21
-23
GCM Scenario +
GISS

2
11
7
22
-18
1
5
26
-3
0
-20
7
GFDL

3
16
1
8
-21
-32
-13
-23
-75
-22
-35
-29
D.E. CO,
UKMO

-28
-33
-10
-6
-34
-12
-34
-4
-14
-21
-3
-2
 D.E. = Direct physiological effects of 555 ppm CO2.
                                              USA-18

-------
Table 5. (Cont.)
Site
IRRIGATION
Spokane
Yakima
Boise
Fresno
Phoenix
Fargo
North Platte
Dodge City
Abilene
Des Moines
Indianapolis
Memphis
Baton Rouge
Columbia
Tallahassee
Lynchburg
Williamsport
BASE
(t ha1)

9.88
12.92
14.58
12.19
7.78
13.37
12.94
12.66
10.49
13.68
13.04
11.53
11.47
10.75
11.65
12.67
12.86
GCM Scenario alone
GISS

40
-5
-22
-4
-35
-8
-22
-9
-2
-25
-20
-19
-9
-16
-10
-24
-21
GFDL

59
7
-20
1
-28
-12
-22
-15
-1
-28
-6
-14
4
9
-13
-3
-8
UKMO

32
-28
-24
-3
-57
-24
-38
-21
-36
-38
-25
-42
-39
-23
-9
-11
-14
GCM Scenario +
GISS

48
0
-17
-3
-31
-3
-17
-5
1
-21
-15
-14
-4
-11
-5
-19
-17
GFDL

62
13
-16
1
-24
-7
-17
-10
0
-24
-3
-10
5
12
-9
0
-14
D.E. C02
UKMO

39
-24
-20
-2
-55
-21
-34
-21
-32
-34
-21
-39
-36
-19
-6
-7
-10
D.E. = Direct physiological effects of 555 ppm CO2.
                                           USA-19

-------
Table 6.
Effects of GCM climate change on dryland and irrigated wheat yields.
Site
DRYLAND
Spokane
Yakima
Boise
Phoenix
Fargo
North Platte
Dodge City
Abilene
Des Moines
Indianapolis
Memphis
Baton Rouge
Columbia
Tallahassee
Lynchburg
Williamsport
BASE
(t ha'1)

2.62
1.48
1.38
0.89
4.13
2.79
3.56
2.60
4.76
4.95
4.41
3.86
4.38
3.58
4.72
4.88
GCM Scenario alone
GISS

76
103
97
-99
-18
-18
-48
-58
-4
-3
-25
-57
-22
-56
-6
-2
GFD
L

-6
-37
-20
-93
-14
-36
-44
-33
-12
-6
-10
-58
-19
-80
-2
0
UKM
O

29
-7
43
-99
-38
-33
-32
-67
-15
-16
-37
-98
-35
-100
-25
-6
GCM Scenario +
GISS

126
180
172
-99
8
11
-29
-43
16
14
-12
-48
-8
-46
13
17
GFDL

33
-84
14
-91
10
-15
-26
-10
7
10
4
-50
-5
-76
17
19
D.E. CO,
UKMO

71
31
102
-99
-13
-10
-10
-57
1
-1
-26
-97
-22
-100
-10
12
 D.E. = Direct physiological effects of 555 ppm CO2.
                                             USA-20

-------
Table 6.
(Cont.)
Site
IRRIGATION
Spokane
Yakima
Boise
Fresno
Phoenix
Fargo
North Platte
Dodge City
Abilene
Des Moines
Indianapolis
Memphis
Columbia
Tallahassee
Lynchburg
Williamsport
BASE
(t ha'1)

5.41
5.34
5.67
7.29
4.85
5.12
4.91
5.59
5.24
4.90
5.26
4.42
4.53
3.93
5.15
5.35
GCM Scenario alone
GISS

6
-3
2
-37
-92
-8
8
-7
-30
0
-8
-24
-22
-45
-12
-8
GFD
L

-6
-4
8
-39
-67
-15
12
-8
-24
4
-12
-12
-20
-80
-9
-6
UKM
O

10
-3
13
-63
-96
-8
1
-12
-48
-11
-20
-36
-34
-100
-27
-9
GCM Scenario 4
GISS

25
14
24
-16
-91
8
28
9
-18
17
7
-11
-9
-35
4
8
GFDL

10
12
27
-15
-62
0
31
8
-11
13
3
4
-6
-76
7
11
D.E. C02
UKMO

29
14
32
-45
-96
9
19
3
-40
4
-6
-25
-22
-100
-15
7
D.E. = Direct physiological effects of 555 ppm CO2.
                                           USA-21

-------






<

CJ
S
6
1
I w
.1 M
g S
1
i
i
&
o
H. 5
1 w
> P"!
^ S
S O
00 CO
a
_
e
o

CO

_>
Q
rS


j
0
VI
E
O
W
^ji
03

0
1

CO
S2
0
ttl
0
i
J
O

8




23



CO
O OO CM TH c^ l> *
IO CO *H rH ^- VO U")
CM CM CM TH CM CM CM



V> OO CO CM t^ IO TH
CM CM CM TH csi CM CM

CO TH 00 00 CO >n VO
vo * TH CM co oo w->
CM CM CM TH CM CM CM
^ vg vo tj ^ TH 00
CS C^l C^ TH rH CO 0s!


TH
OO 00 C"** O\ 00 Ox O\

00 00 O\ IO Tj* rH t^
TH oo oo o\ CT\ ON o\

TrH rH . ( . i
O\ O ON O O 00 OO O

gNCQc^cNi^cot^oo
ONTHOCOOOONOOO
^_f ^r( i, ^ _^


CSrH^cOOOTHOOOO
OTHOCO^-ONOOO




CO CD TH "* O\ TH CP CO
CO *^* C*^ CO *O TH C3 t~H



| 1 0 S | 0, I |
o-i-^So?^a-
CU 45 5 r"H J3 rI ^ O
COr^OQCi^OnbZE-1
CM TH CM



* 00 VO
CM TH CM

Tj* -H^ VO
co ON in
CM TH CM
CO CM "*
not*-
CM CM CM


"si* co in
ON co 00
TH
TH

S C ON

oo in r-
O TH O
* ON o
CM CM O

* VO CM
CM CM O


in in co
CM CM O




00 CM ON
CM CM O


^
 CH
o a 1
So S 2>
in CM TH ON



VO * O O
in vo oo i>
CM TH TH TH

* O O ON
>n c~- oo vo
CM TH TH TH
f^ f^^ C1^ C1^
CM TH TH TH


TH
s * * s

VO U") CO VO
ON O\ O ON
rH

B 8 TH cB
ON O CO OO O
O CO CO Tf in

ON in r~ i> vp
O CM CM * -^


 8 8  $




vo >n >n TH c~-
TH CM CM in <*


.53
1 a & 1 a
1 S 1 1 8
s i S g s
rH VO
r> co
rH CS
i
i
i
 in TH
TH TH CM

I> 00 ON
VO O in
TH TH CM
g g CM
TH TH CM


O ON ON
CO vo C"^
o oo oo
TH

t- CO rf
tO ON C3
TH TH

TH 2 TH
in CM o

S S 2


5 3 S
^H T^ ^H



O TH in
in co TH


C
 P EL
8 5 g
2 > a
s -s a
151
<
s

-------

.
1
o
u
1
o
^3
'
s  s
g
e
o
i
cl
en
C
fi
0
OH
I"
b
T3
a
,2
o*
3
2
^^
'c3
si
00
^
o
"w
5.
a
CO

^
s>.
e
.22
'o
1
3
2
1
oo
n c^
CM rH rH rH

0 ca
W '^ ' CQ 
0
o


CM
0

0
c5

^
CM


ON
0




CM
CM



O
r-5

vq




VO
O




rH
O



t~
0
ON
C3

Phoenix
oo
vd


oo

VO
oo

oo



t^
t>
i i



CM



in
vd
CM
H
CM



n
CO




VO
00



oo
in
m
CO

$
PH
o\ in
in vd


ON in

oo 1>
t> 10

* oq
vd vd


rH CO
CO *
1 1 1 1



in t>
CJ ON
CM rH



5t VO
ON ON
i ( . i i
10 ON
vd vd
rH rH



0 ON Tt-
CO CO rH




ON Tf VO
CO ** r-5



rH OO ON
^ Tf rH
l> 10 t-
CM 10 CM

&
2  
*  -S a
  nj &0
o o
a


VO
IO

VO
CO

ON
n


vq
i5
i i



vq
i i



ON

i i
CO
rH



VO
O




CM
r-5



CO
rH
0
CM

Abilene
CM
oo


<*
00

^H
o
l-H
i 1
00


CM
f^
1 1



ON
CM



IO
CO
CM
3
C*^



oo
rH




-^
10'



00
VO
VO
10'

1
1
CO
oo


ON

CM
c5
rH
-*
OO


CM
ON
i i



oo
n
i i



vq

CM
vq
CM



l>
CO




in




ON
n
i i
iri

Indianapolis
00
in


p
r-5
rH

CO
00

00
00


oo
ON




CM
in
rH



l>

CM
vq
rH



0
l-H




CM
CM



oo
CM
0


.a
1
1
CM
0


rj-
CO

in
CO

VO
t>


oo
CO
l-H



00
o
1 1



CM

CM
m
>*
rH



CM
rH




^f
CO



VO
co
vq


u
I
CH
CQ
in
vd


oo
00

CM
00

00
00


ON
rH
rH



f;



CM

rt
VO
in
rH



0
*




^f
CO



CD
"*
CM
*

Columbia






























ON
0




f  ,
r-5



ON
CM
CM
CO

i



r-5

^J.
CO

f-
vd


^.
^
rH



CO
rH



CM

CM
CM
ON
l-H



^
CM




l-H
10



CM
n
oo
n

Tallahassee
od


oq
r-5
rH

^J.
r-5
rH
CM
ON


O
ON
i 1



CO
CO
1 1



CM
ON
rH
in
oo
rH



C3
CM




f^
CM



CO
vd
o
in

Lynchburg
in
ON


ON

Tj.
C3
rH
Tt-
00


p
33




vd
rH



CM

CM
O
ON
rH



CM
^




f^
CO



VO
vd
p
n

Williamsport

-------
g
O
"o
.s

s
flj
0
O 
|j

O

S


j
I


CO
22

S
o
o
5
'E
1
.9 W
s> 2
:i s
.5
o
>

d

O
1

^
O

CO
CO
s


Ov
4J o
S K

VOrHOt^cOCMCM |-rJ--<*VOOr>OOCM |OOOO
rH OOO\CO'<* <4-rHCOCOCMrH1OrH'
' rH ' CO ' rH !
i


CM^CMC^t-OiO 'COTtvOCOOOOOCM JiO^fVO
'rHrHt~-O\CMTfrH rHVOTfOOrH1 T'T*


rHVOOCO^O^ "CMCMCMrHCMVOOX IOOCMOO
COCMCOt^CO rH'cMCOrHTfCMCM' ->*COrH
I I I i T-H ' ' rH ' '


C~rHrfcOO\CMCT\ 'COT*COOAOQO'* 'f-iOO
rHrH VOCOCMrHlrHCOCMCN^lOCMrH




VOt^CMOCM^OO |OC~-rHVOrH'^-'^)- |r-VOCM
CM li i3 ;=i ^ .3. i '^-HrHCMCOCMCOVO '^CMIO
v 1 1 1 1 III r-l 1


TfCTvOOt^OOlO 'VOCMVOVOONOOrH |rHTj-O\
2* : ^4 l< C ft) rl 'cOrplOvSvOlOCM 'OOrHIO
CM
CO
^


\OCOOOOT-Ht^-lOC**iQ\'^'flr-ITfCSOOt1-VOC^O
0s! CO CQ VO CO ^1 CO VO *O 00 00 VO OK VO VO ^H ^H ON Cj
^-( i ( C^ ^f *~* i~t

t*^- ON C^- O\ ^3 OQ VO ON ^t" VO VO ^3 ^t" C^^ O3 r~i 00 VO t-H
co oi *o '^ i^ vo vo *o ^" v^ vo vo vo i"** ON ON *o
1 T-H

1 CNCO T-HTj-'^'CO fs) VO r-l i 4 VO ^H CN
X 1

S & S 3 s  S
5  eow0 ^S?^
0 _ ... a* O ., 'S O-c -a So 'o  S e
1 1 s 1 1 a H  | * 1 1 1 1 1 1 ! 1
O'5'aSo'-1S&-
co>^ca.[i,PHCHlz;E-iQ
-------
                                               UNITED STATES
Figure 1.     Map of the United States and location of the study sites.

-------
      US REGIONAL PRODUCTION
                WHEAT
               SOYBEAN
Figure 2.    Regions of wheat, maize and soybean production.

-------
 TEMPERATURE (C)
                                          SPOKANE
                                                          TEMPERATURE (C)
                                                                                                     FAROO
JAN FEB  MAR APR  MAY JUN  JUL  AUQ SEP OCT  NOV DEC
                                                         JAN FEB  MAR APR  MAY  JUN  JUL  AUQ SEP  OCT NOV  DEC
 TEMPERATURE (C)
                                        DODGE CITY
                                                          TEMPERATURE (C)
                                                                                                 DES MCHNES
JAN FEB  MAR APR  MAY JUN  JUL  AUG SEP OCT  NOV DEC
                                                         JAN  FEB  MAR APR  MAY JUN  JUL  AUG SEP  OCT NOV  DEC
  TEMPERATURE (C)
                                           ABILENE
                                                           TEMPERATURE (C)
                                                                                                    MEMPHIS
JAN  FEB  MAR APR  MAY JUN  JUL  AUQ SEP  OCT NOV DEC
                                                         JAN  FEB  MAR APR  MAY JUN  JUL AUG  SEP OCT NOV DEC
	 OBSERVED
---oiss 
	 QFDL
-S-UKMO
         Figure 3a.   Observed and 2 x CO2 scenario temperature at selected sites in the U.S.

-------
 200
    PRECIPITATION (MM/MONTH)
                                               SPOKANE
                                                                PRECIPITATION (MM/MONTH)
  JAN  FEB MAR APR  MAY JUN  JUL AUQ SEP  OCT NOV  DEC
                                                              JAN  FEB MAR APR  MAY JUN  JUL AUQ SEP  OCT NOV  DEC
    PRECIPITATION (MM/MONTH)
                                            DODGE CITY
                                                             200
                                                                PRECIPITATION (MM/MONTH)
                                                                                                        DES KOINES
  JAN FEB  MAR  APR MAY  JUN  JUL AUG  SEP OCT  NOV  DEC
                                                              JAN  FEB MAR  APR MAY JUN  JUL AUQ  SEP OCT NOV  DEC
200
   PRECIPITATION (MM/MONTH)
                                                                PRECIPITATION (MM/MONTH)
                                                                                                           MEMPHIS
  0
  JAN
      FEB  MAR APR MAY  JUN JUL  AUQ  SEP OCT  NOV DEC
FEB MAR APR  MAY JUN  JUL AUQ SEP  OCT NOV DEC
	 OBSERVED
---OISS -
	 QFDL
-e-UKMO
                Figure 3b.  Observed and 2 x CO2 scenario precipitation at selected sites in the U.S.

-------
 CO
 0
CO
LJJ
DC
LLJ

O
                      CD
                                                                      CO
                                                                     03
                                                                     "^

                                                                     CD
                                                                     m
                                                                     cO
                                                                     03
                           CO   CO

                           C\j   T-^


                           
-------
 0
 N

"CO
LLJ
DC
LLJ
O
        I   I    I   I
OJ
CO
U)
                                                               0)
                                                               0)
co  co
o  o
 C/)   CO   C/J

000
    X  I
    p  p
    S
;!  o>  oo
                                                                   o
         II
        o
         o
    o   o
 .   CO   CO
 CO  CO   cO
CQ  CQ  CQ
                                                                               O

                                                                                 i
                                                                               CL
                                                                               O
                                                                               CM
                                                                               o
                                                                               C\J
                 0
                 CM
                 CL
                 O
                 CM
                 CL
                 CM
                O
                CM
                CL
                                                                               o
                                                                               CL
                                                                               o
                                                                               CL
                        8
                          tf
                        I!
                                                                                      <= '-
                                                                                      00
      oooooooooo
      CMr-            T-CMCO^LOCO^
                                p|9!A

-------
o
cc
o
                                                                       O
                                                                       CM

                                                                       DL
                                                                       o
                                                                       CM
                                                                       Q_
                                                                       CM
                                                                       O
                                                                       CM
                                                                       QL
                                                                       O
                                                                       o
                                                                       CM
                                o
                                CM
                                CL
                                CM
                                                                       O
                                                                       CM
                                                                       Q.
                                                                       O
                                                                       O
                                                                       CL
                                                                       O
                                                                       Q_
                                        8
                                        w
                                        i
                                        I
                                        .a
                                        t
                                        i
                                        J
                           .s

                           1
                           >> -5
                           O S
                                                                               ca 9
                                                                               a o
                                                                               a
                                                                                 O
                                                                                 o
                                                                               eo
                                                                               E
      O
      CM
o
CM
O
CO
O
LO
O
CO
                             PI9IA

-------
                    20
                    15
                    10
                     S
                     0
                    -5
                    -10-
                    -15-
                    -20-
                    -25-
                    -30-
                    -35-
                    40-
                    -45-
                    -50-
                    -55-
                    -60
                                 Annual US Wheat Production
                                            [ Without C02
                                             WHhCO2
Current Production - 64 mfl MT
                              GISS
                                                QFDL
                                                                  UKMO
                   20
                   15
                   10
                    5
                    0
                   -5
                   -10
                   -15-
                   -20-
                   -25-
                   -30-
                   -35-
                   40-
                   -45-
                   -50-
                   -55-
                   -60
                                  Annual US Corn Production
                                           | Without CO2
                                            WHhCOZ
Current production -183 mil MT
                             GISS
                                               GFDL
                                                                 UKMO
                               Annual US Soybean Production
                                                                   Without CO2
                                                                   WXhCO2
                        Currant production - 49 mB MT
                            GISS
                                              GFDL
                                                                UKMO
Figure 5.       Change in annual national production of wheat, maize and soybean with the GCM climate
               change scenarios.

-------
O
10
O
CM

CO
CO

O
O
CO
O
01

CO
CO

O
 03
O

O
CM

CO
CO

CD
CL
Q_

O
CO
CO-

LU
  .O
  'l-l

  a
  c



 I
  W5

 I
 oo
 00

 3
 CD
        I
        C5
 5s
 flS *T3

 II

 |i
  a
 a S
 "" ts
 c -S
 6 ^
E
CD

-------
     CO
 C0.0>

 8,1

 S

JC CO
O  u.
     o
T3 ^
 05-8
 BM^BHMH  HtanB
     CO
     c
     CD
 CD  o
.NJ CO

 CC CO
~~ CO
    O
o
CM
I
o
CO
     ^
     f **
              to
              10
              10

              o
              
              o
              CO
              CO
               o
               O)
              IO
              10
              IO

              o
              a>
              o
              CO
              CO
              o
              a>
              CC
             o
             CO
             h-
             CO

             g
             Q_
                   CO

                   O

                   CL
                   
-------
POSSIBLE IMPACTS OF CLIMATE CHANGE ON MAIZE YIELDS IN MEXICO






                    Diana Overman, Max Dilley, and Karen O'Brien




                   Pennsylvania State University, Pennsylvania, USA






                               Leticia Menchaca




                  Centra de Ciencias de la Atmosfera, UNAM, Mexico
                                  MEXICO-1

-------
TABLE OF CONTENTS

SUMMARY
INTRODUCTION
       Background
       Maize Production
       Analysis of Climate Change Scenarios
       Study Design
       Site Characteristics
METHODS
       Base Climate and Climate Change Scenarios
       Soils and Management Variables
       Crop Model Calibration and Validation
RESULTS AND DISCUSSION
       Sensitivity Analysis of Baseline Yields to Inputs
       Sensitivity Analysis of Maize Yields to Climate and CO2
       Maize Yields with GCM Climate Change Scenarios
       Maize Yields with the GISS Transient Scenarios
       Adaptive Responses
CONCLUSION
REFERENCES
                                        MEXICO-2

-------
SUMMARY

        This study used global climate models and a crop growth model to estimate the potential impacts of
climate change on maize yields in Mexico. Projected climate change caused simulated maize yields to decrease
dramatically in two main Mexican regions, but the magnitude of the decrease varied with climatic scenario and
with the initial set of management variables selected for the simulation. The decreases in yield were caused
primarily by temperature increases, which shortened the duration of the crop's life cycle, particularly the grain-
filling period. The duration of the crop's growth period became shorter, exerting a dramatic negative pressure
on yields. These decreases were slightly counteracted by the beneficial physiological CO2 effects, as simulated
in this study. A sensitivity analysis indicated that decreases in crop yields would be severe under global
warming unless irrigation expands, fertilizer use increases, or new varieties that are more tolerant of heat are
developed.
INTRODUCTION

Background

        With production of 25 million metric tons per year, Mexico is one of the world's top fifteen cereal
producers. Although Mexico  has exported cereals in the past, net cereal imports have exceeded 5 million
metric tons in recent years. Mexico imported $284 million worth of maize in 1987, $394 million in 1988, and
$441 million in 1989. The country is continually striving to support a growing population with an agricultural
system that relies on relatively low  and variable rainfall.  For Mexico, an increase in temperature  could
aggravate the existing nutritional and economic problems. More than one-third of Mexico's rapidly growing
population works in agriculture, a sector whose prosperity is critical to the nation's debt-burdened economy.
Although only  one-fifth of Mexico's  crop land is irrigated, this area accounts for half of the value of the
country's agricultural production, including many export crops. Many irrigation districts rely on small reservoirs
or wells that deplete rapidly in dry years. The remaining rainfed crop land supports many subsistence farmers
and provides much of the domestic food supply. Frequent droughts reduce harvest and increase hunger and
poverty in many areas of Mexico.

Maize Production

        Maize is the most important crop grown in  Mexico. In 1988, production totalled 14.4 million tons
(52% of the total production  of major crops), grown on 11.5 million hectares of land (56% of the total crop
land). Almost every region in Mexico produces maize, but the states of Veracruz, Tamaulipas, and Michoacan
tend to be the largest producers. Yields are generally low, ranging from about 350 kg ha"1 in Quintana Roo
in 1986 to 3,670 kg ha"1 in Sonora in 1986. Average yields are 1,250 kg ha"1. Maize is produced in a wide
variety of environments, from highlands to coasts, dry to humid climates, and poor to fertile soils.
        Maize was probably domesticated in Mexico, and many of its wild ancestors can still be found. It was
the staple crop of pre-Columbian populations and continued to be an important component of the diet under
Spanish colonialism. In contemporary Mexico, maize  is the basis of subsistence and the primary cash crop for
the majority of poorer and smaller farms. It also provides the corn flour for the tortillas eaten with almost
every meal in both urban and rural Mexico. Each Mexican  consumes approximately  250 kilograms of maize
per year. Most  maize is rainfed. In recent years, 11% of the maize area has been irrigated, providing about
20% of the production. Thus, maize production is extremely vulnerable to climatic variation and drought. The
rapid increase in the Mexican population has contributed to the growth in the demand for maize in recent

                                            MEXICO-3

-------
years. Coupled with environmental and economic problems, the situation has led the Mexican government to
import considerable amounts of maize in recent years.
        In 1950, the expansion of the area planted to maize combined with the use of improved seeds and
fertilizer led to a large increase in Mexican maize production. Mexico was one of the leading participants in
the "Green Revolution" and considerable attention has been focused on breeding  improved varieties of both
rainfed and irrigated maize. The use of improved seeds and fertilizer varies widely in Mexico. In some regions,
such as the Valley of Puebla, improved seeds, agricultural chemicals, and agricultural extension work has led
to much higher yields in good years. However, in other regions, some farmers cannot afford additional
chemical fertilizer or irrigated land, and yields remain low.

Analysis  of Climate Change Scenarios

        Liverman and O'Brien (1992) analyzed the possible impacts of global warming on climate and water
availability in  Mexico. Their study  used five climate  models to construct scenarios for  temperature,
precipitation, and evaporation at a range of stations in Mexico, and highlighted some of the problems that
arise from using climate models to predict future climate. Their model versions are  the same as those used
in this study, and they used the same methodology that we have  used to construct the climate scenarios.
Climate models are unable to adequately represent current climate in many parts of Mexico, and climate
change scenarios created with different models vary  widely. The principal conclusion of the study is that
Mexico  is likely to  be warmer and drier under global warming conditions. With all of the climate models
analyzed, potential evaporation increased and moisture availability decreased, even in those cases where the
model projected an increase  in precipitation. A sensitivity analysis of our calculations of evaporation and
moisture deficit indicates that water availability could increase only with higher rainfall and relative humidities,
or significantly less  solar radiation and windiness.

Study Design

        In this study, we used a crop simulation model (CERES-Maize) (Jones  and Kiniry 1986;  IBSNAT
19S9) and climate change scenarios generated from three General Circulation Models to analyze the possible
impacts of global warming on Mexican  maize yields.  The CERES-Maize model  uses daily weather data to
estimate the  effects of climate on yields of rainfed or irrigated maize. The model is physiologically based,
simulating the influence of solar radiation, temperature, nutrients, and water availability on maize development
through major  phonological stages.
        It  has  been suggested that crops will benefit from  higher levels of atmospheric carbon dioxide,
resulting  in  higher yields. The  CERES-Maize model  also  includes  an  option to simulate the  direct
physiological effects of elevated CO2 levels on crop growth and the efficiency of water use (Acock and Allen
1985). However, the model does not account  for losses from, and changes in, pests and diseases. It may,
therefore, overestimate yields.

Site Characteristics

        Our  choice of sites for this study was limited by the availability of long time-series of daily climate
data, and by  the need to find sites where experimental data were available for calibrating and validating the
CERES model. With the cooperation of CYMMIT, we were able to obtain daily climate data and limited crop
experimental data for Poza Rica, a site on the Gulf Coast of Mexico,  (+2032' N, 9726' W; 60 m altitude)
and Tlaltizapai, a site in the highlands south of Mexico City (+1841' N, 9908' W;  940  m altitude) (Figure
1). These sites are representative of important maize-growing regions in Mexico.  The Tlaltizapan site is

                                             MEXICO-4

-------
representative of maize in the central plateau and might be used to approximate conditions in the states of
Mexico and Puebla. Poza Rica is more typical of the Gulf coast states such as Veracruz,  Tabasco, and
Tamaulipas.

METHODS

Base Climate and Climate Change Scenarios

        We were able to obtain climate data from 1973 to 1989 for Tlaltizapan and Poza Rica, with missing
data for 1986 at the first  site. Because observed solar radiation data were only available for a small number
of years, we used sunshine hours to generate the solar radiation for the two sites using WGEN (Richardson
and Wright 1984).
        The climate change scenarios for each site were generated from three GCMs: GISS (Goddard Institute
for Space Studies, Hansen et al. 1983); GFDL (Geophysical Fluid Dynamics Laboratory Model, Manabe and
Wetherald  1987); and UKMO (United Kingdom Meteorological  Office Model, Wilson and Mitchell 1987).
Changes in temperature, precipitation, and solar radiation projected by the GCMs were applied to the
observed (base) climate variables to create the climate change scenarios for Tlaltizapan and Poza Rica. The
base climate sets were also modified with the GISS transient model to create transient scenarios for the 2010s,
2030s, and 2050s (Hansen et al. 1989). Table 1 presents the seasonal and annual temperature and precipitation
changes projected by the GCMs.
        The mean  annual observed temperature at Tlaltizapan is 24C; and the average observed rainfall is
about 96 mm from November to May, and 729 mm from June to  October. The mean annual temperature at
Poza Rica is 23C, and the average rainfall is about 380 mm from  November to May, and 834 mm from June
to October.

Soils and Management Variables

        The soils in Tlaltizapan are silt-clay soils 1.0 to 1.8 m deep ("isothermic udic pellusterts"). Because
we were unable to obtain detailed soil information, we used an isothermic udic pellusert from Guatemala from
a soil data base. Soils in Mexico tend to be poor in nutrients and extremely  dry at the beginning of the growing
season. Therefore, we modified the initial soil conditions to represent low soil nitrogen and water content. We
also reduced the organic content of the soil by reducing the amount of crop residue in the initial conditions.
        In Poza Rica, the site with a lower altitude, soils are typically sandy loam of moderate depth. Because
of the lack  of detailed soil information, we used a generic medium-depth  sandy loam soil from the IBSNAT
data base and modified the initial soil conditions to reflect low soil fertility  and water content at the beginning
of the growing season.
        In rural Mexico, maize is  typically grown by poorer farmers who cannot afford chemical fertilizers.
Rather than use the high levels of fertilization and plant densities practiced at the experimental stations, we
selected agronomic conditions more typical of average farmers in the region. We used 50 kg of nitrogen
fertilizer at planting in the model  runs arid soils of relatively low fertility  and moisture content to represent
Mexican conditions. In addition, we have simulated maize yields  in the absence of nitrogen limitations for
purposes of comparison. We used  a planting date of May 20 at Tlaltizapan and June 15 at Poza Rica. In the
cases where we simulated irrigated maize, we assumed a 0.5 m irrigation depth at a rate of 50% efficiency.

Crop Model Calibration and Validation

       The CERES-Maize  crop  model (Ritchie et al. 1989) was used for the simulation of maize yield,

                                           MEXICO-5

-------
season length duration, and crop growing season evapotranspiration. To calibrate the model for the varieties
and conditions of this study, we used several databases obtained from CYMMIT experimental plots between
1987  and 1989. Five genetic coefficients may be modified to calibrate the model phenology and yield with
observed phenology and yield data (Table 2). We began by running the model using the genetic coefficients
of the SUWAN-1 variety that was listed in the IBSNAT database (IBSNAT 1989). This variety has genetic
characteristics  similar  to  some of the  Mexican varieties grown at both Tlaltizapan and Poza Rica. We
compared the predicted values for anthesis date, maturity date, yield, kernel weight, grains per square meter,
grains per ear, and total biomass, with experimental data from CYMMIT at both stations. Coefficients P5 and
G3 modify the grain-filling duration and kernel-filling  rate, respectively. We repeatedly ran the model and
adjusted P5 (within the ranges suggested by the CERES-Maize User's  Guide) to obtain as close a match as
possible between the simulated and predicted phenologies. We reduced the coefficients G2 (maximum number
of kernels per plant) and G3 (kernel-filling rate) to get closer to observed yields. Since we had data for more
than one experimental plot at each station, we validated our calibrated variety with data from a second plot
and made small modifications to the coefficients based on the results. The results of the validations are shown
in Table 2.

RESULTS

Sensitivity Analysis of the Baseline Yield to  Inputs

        Table 3 shows the influence of management variables on simulated maize yields under the baseline
climate. Simulated yields at the two sites were rather sensitive to our assumptions about soil, fertilizer use,
management practices, sowing date, and initial conditions. Our baseline results for 16 years at Tlaltizapan
produced an average rainfed yield of 4.02 t ha"1 with nutrient stress conditions. Without nitrogen constraints,
rainfed yields are 4.55 t ha"1. At Poza Rica, 17 years of baseline  climate produced an average rainfed yield of
3.18 t ha"1 with  nutrient stress. Without nitrogen constraints, rainfed yields average 3.911 ha"1. The sensitivity
of yields to our assumptions of nutrient-poor, dry soils and limited fertilizer use is clear. Yields increased at
both sites with  increases in fertilizer application and with deeper, moister, more fertile soils.
         A delay in the planting date produced lower yields at  both  sites, while planting 10 days earlier
reduced yields  at Tlaltizapan, but slightly increased yields at Poza Rica. The unmodified  SUWAN variety
produced slightly higher yields at both sites than the calibrated variety. Yields were lower with a  planting
density of two plants per square meter and higher at Tlaltizapan with four plants per square meter.
        These variations in responses to current practices suggest that a full set of practices should be tested
for better projections of the impacts of climate change. Due to lack of detailed site information, we made
some rather arbitrary decisions about soils, initial conditions, planting dates, and genetic coefficients.

Sensitivity Analysis of Maize Yields to Climate and CO2 Changes

        We also conducted a sensitivity analysis of  the CERES-Maize model to changes in climate of +2C
and +4C, +20% and -20% precipitation,  and 330  ppm and 555 ppm of CO2 (Table 4). At Poza Rica, base
yields decreased with temperature increases and rainfall decreases; base yields increased under the +20%
rainfall scenario. The simulation of the physiological CO2 effects results in a slight increase of scenario yields
in all  cases.

Maize Yields under GCM Climate Change Scenarios

       The effect of GCM climate change scenarios on simulated maize yields at Tlaltizapan and Poza Rica

                                            MEXICO-6

-------
is presented in Table 5; For1 each climate scenario, yields were simulated under conditions with nutrient stress
(to represent the actual-situation) and under conditions without nutrient limitations. The physiological effects
of elevated atmospheric CO2 (555 ppm) were evaluated in all simulations. In this experiment we assumed that
the planting dates and varieties remain unchanged under the climate change scenarios.
        In Tlaltizapan, base rainfed yields decreased under all climate change scenarios, even when the
physiological effects of GO2 were included in the simulation. The decreases in yield ranged from 20% for the
GFDL scenario, to 61% for the UKMO scenario. These low scenario yields occurred  because the  higher
temperature accelerates the phenology and the crop matures faster with less time for gain-filling. There is also
drought stress early in the growing season under the scenarip conditions.
        Baseline  rainfed yields of maize at Poza Rica also decreased with global warming, even with the
physiological effects of 555 ppm CO2- The yield decreases at this site were smaller than those at Tlaltizapan,
but nevertheless, they are significant ranging from -6% (GISS) to -26% (UKMO).

Maize Yields under the GISS  Transient Scenarios

        Yields of rainfed maize have  also been simulated at Tlaltizapan and Poza Rica for  three GISS
transient scenarios, corresponding to the decades of the 2010s, 2030s, and  2050s (Table 6). At Tlaltizapan,
yields decreased consistently as the  climate changes, while at Poza Rica the yield responses varied through
time. The latter result indicates that the trajectory of agricultural response to climate change may be non-
linear.

Adaptive Responses

        The sensitivity analyses of the baseline yields to management variables indicate the importance of
some of the current uncertainties in estimating the impacts of future climate changes. In the previous sections
we have presented results assuming no change in the average planting date or in the varieties used. But clearly,
as climate changes, farmers and governments will try to adapt to change. In this section, we examine whether
changes in maize  varieties and nutrient levels may offset the negative impacts of a warmer climate.
        The amount of fertilizer used is critically important to the maize yields predicted by the CERES-Maize
model (Tables 3 and 5). At present, many Mexican producers can only afford to use small doses of nitrogen
fertilizer at planting. If more fertilizer becomes available to more farmers, some of the yield reductions under
the climate change scenarios might be offset (Table 5). With full fertilization, simulated maize yields increased
under irrigation at Poza Rica, but the increase does not fully offset the negative effects of climate change on
maize yields. However, given the environmental and economic constraints and trends in agricultural inputs in
Mexico, unlimited water and nutrients are extremely unlikely.
        We also simulated planting varieties more suited to the new climate conditions, since changing the
variety or the genetic characteristics  of crops may also help farmers to adapt to global warming. For example,
planting the unmodified SUWAN or PIONEER varieties, rather than the calibrated varieties, produced slightly
higher yields under the GISS scenario at both sites.

CONCLUSION

        In this study we found that rainfed and irrigated  maize yields in Mexico decreased  with  global
warming, in spite of differences among the climate  model results.  Sensitivity analyses indicated  that the
decreases  in crop yields  will be severe under global warming unless irrigation expands, use  of fertilizer
increases, or new varieties of plants are developed.
                                            MEXICO-7

-------
REFERENCES

Acock, B., and L.H. Allen Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
        and J.D. Cure (eds.), Direct Effects of Increasing Carbon Dioxide on Vegetation. U.S. Department of
        Energy. DOE/ER-0238. Washington, D.C. pp. 33-97.

Hansen, J., G. Russell, D. Rind,  P. Stone, A. Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983. Efficient
        Three-Dimensional Global Models for Climate Studies: Models I and II. April Monthly Weather
        Review, Vol III, No. 4: 609-662.

Hansen, J., I. Fung, A. Lacis, D. Rind, G. Russell, S. Lebedeff, R. Ruedy, and P. Stone. 1988. Global climate
        changes as forecast by the GISS 3-D model. Journal of Geophysical Research, 93(D8):9341-9364.

IBSNAT (International Benchmark Sites Network for Agrotechnology Transfer Project). 1989. Decision
        Support System for Agrotechnology Transfer Version 2.1 (DSSAT v2.1). Dept. Agronomy and Soil Sci.,
        College of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii 96822.

Jones, C.A., and J.R. Kiniry. 1986. CERES-Maize: A Simulation Model of Maize  Growth and Development.
        Texas A&M Press. College Station. 194 pp.

Liverman, D., and O'Brien. 1992. Global Warming and Climate Change  in Mexico. Global Environmental
        Change, in press.

Manabe, S., and R. Wetherald. 1987. Large-scale changes of soil wetness  induced  by an increase in
        atmospheric carbon dioxide. Journal of Atmospheric Science, 44:1601-1613.

Richardson, C.W., and D.A. Wright. 1984. WGEN: A Model for Generating Daify Weather Variables. ARS-8.
        U.S. Department of Agriculture, Agricultural Research Service. Washington, DC. 83 pp.

Wilson, C.A., and J.F.B. Mitchell. 1987. A doubled CO2 Climate Sensitivity Experiment with a Global Model
        Including a Simple Ocean. Journal of Geophysical Research, 92: 13315-13343.
                                           MEXICO-8

-------
Table 1.       Temperature differences (C) and precipitation changes (%) between GCM lxCO2 and
              2xCO2 climate change scenarios.
                      Temp. Diff. (C)                     Precip. Changes (%)
Site/GCM
Tlaltizapan
GISS
GFDL
UKMO
Poza Rica
GISS
GFDL
UKMO
Spr.

4.1
2.3
3.7

4.1
2.3
4.1
Sum.

4.1
2.5
3.5

4.1
2.5
4.0
Fall

4.3
2.9
2.9

4.3
2.9
4.0
Win.

4.5
2.9
3.0

4.5
2.9
3.6
Annual

4.2
2.7
3.3

4.2
2.7
3.9
Spr.

-6
14
-33

-6
14
-5
Sum.

-6
-19
42

-6
-19
-13
Fall

-2
-16
11

-2
-16
16
Win.

-3
10
-24

-3
10
-29
Annual

-4
-9
5

-4
-9
-6
                                          MEXICO-9

-------
Table 2.
Simulated and observed (1989) yield and season length at Tlaltizapan and Poza Rica.
  Site
     Experiment
      Yield (T/Ha)

Obs.         Sim.
   Season Length (days)

Obs.             Sim.
Tlaltizapan
Tlaltizapan
Poza Rica
Poza Rica
Poza Rica
1
2
1
2
3
6.20
5.83
5.27
6.98
6.56
6.58
5.72
9.12
6.53
7.50
122
122
112
138
138
127
127
109
133
133
Coefficients of SUWAN maize variety and modified coefficients of the varieties used at Tlaltizapan (TLA) and
Poza Rica (PR):
SUWAN:      PI (380); P2 (0.6); P5 (780); G2 (750); G3 (7.0)
TLA:          PI (320); P2 (0.6); P5 (780); G2 (500); G3 (8.0)
PR:            PI (350); P2 (0.6); P5 (780); G2 (550); G3 (8.5)
                                          MEXICO-10

-------
Table 3.
Sensitivity of CERES-Maize yield to changes in management variables.
               Management variables
                                           Simulated Yield (T/Ha)
                                           Tlaltizapan    Poza Rica
Base scenario*
Planting 10 days later
Planting 10 days earlier
Variety SUWAN
100 kg ha'1 N fertilizer
200 kg ha"! N fertilizer
More productive soil**
V. good initial conditions
Medium initial conditions
2 plants m"2
4 plants m'2
4.02
1.41
3.94
4.75
4.56
4.55
4.05
4.66
4.38
2.86
4.41
3.18
1.12
3.26
3.31
3.81
3.91
4.75
4.53
3.66
2.70
3.16
                *Tlaltizapan planting date May 20; Poza Rica planting date June
                **Medium silt clay at Tlaltizapan; Deep sand at Poza Rica.
                                                           15.
                                           MEXICO-11

-------
Table 4.
Sensitivity of CERES-Maize yield to changes in temperature, precipitation, and atmospheric
              CO2.
                       Scenario
                                    Poza Rica (t ha'1)
                       BASE
                       T+2 C, 330 ppm CO2
                       T+2 C, 555 ppm CO2
                       T+4 C, 330 ppm CO2
                       T+4 C, 555 ppm CO2
                       P+20%, 330 ppm CO2
                       P+20%, 555 ppm CO2
                       P-20%, 330 ppm CO2
                       P-20%, 555 ppm CO2
                                              3.18
                                              2.90
                                              3.10
                                              2.58
                                              2.89
                                              3.26

                                              3.06
                                              3.30
                                      MEXICO-12

-------
Table 5.
Effect of GCM climate change scenarios on CERES-Maize yield.
                                                  Simulated Yield (T/Ha)
                                          Nutrient stress*      No nutrient limitations
Scenario
BASE
GISS, 330 ppm CO2
GISS, 555 ppm CO2
GFDL, 330 ppm CO2
GFDL, 555 ppm CO2
UKMO, 330 ppm CO2
UKMO, 555 ppm CO2
Tlaltizapan
4.02
3.11

2.74
3.20
2.34

Poza
Rica
3.18
2.76
2.97
2.26
2.70
1.83
2.35
Tlaltizapan
4.49
3.49
3.77
3.07
3.47
3.92
3.93
Poza
Rica
3.98
2.80
3.30
2.62
3.18
1.98
2.67
               *Actual conditions
                                           MEXICO-13

-------
Table 6.
Effect of GISS transient scenarios on CERES-Maize yield.
                   Scenario
                                          Simulated Yield
                                             (T/Ha)
                                      Tlaltizapan   Poza Rica
                   BASE, 330 ppm CO2
                   GISS 2010s, 405 ppm CO2
                   GISS 2030s, 460 ppm CO2
                   GISS 2050s, 530 ppm CO2
                                           4.02
                                           3.67
                                           3.32
                                           2.95
3.18
3.27
2.48
2.95
                                       MEXICO-14

-------
                                                                      MEXICO
Figure 1.       Map of Mexico and location of the sites selected for the study.

-------
LU
o:
=)
CEO
LLJQ
r
LU
                        RXXXXXXXXXXXXXXXXXXS^^
o


Z)
                                                               Q
                                                               LJ_

                                                               O
                                                               CO
                                                               CO

                                                               o
LU

S8
CQ
                                *      A


                        0 S33H93Q

-------
o


Is
LJJ
   LLI
or
DL
LU
            R%gggg?Sggg8gg%888SSg?SSSSggg$^^
i .Q
         Q
         u_
         O
                                                               CO
                                                               CO

                                                               o
         LLI

         S8
         DQ
                             IAI1AI

-------
LU
DC
LUO
LU
                                                                  Q
                                                                  Li-
                                                                  CD
                                                                  CO
                                                                  CO
                                                                  O
LU

58
00

-------
o

I
9=0
00
   s
LJJ
Q
U_
O
CO
CO

o
LLJ

S
CD
                                                                                     a

-------

-------
SECTION 3: SOUTH AMERICA

-------

-------
POTENTIAL EFFECTS OF GLOBAL CLIMATE CHANGE FOR BRAZILIAN
 AGRICULTURE: APPLIED SIMULATION STUDIES FOR WHEAT, MAIZE,
                            AND SOYBEANS
                       Otavio Joao Fernandes de Siqueira

                      National Research Center for Wheat
                          CPACT/EMBRAPA, Brazil
                          Jose Renato Boucas Farias

                      National Research Center for Soybean
                           CNPS/EMBRAPA, Brazil
                          Luis Marcelo Aguiar Sans

                 National Research Center for Maize and Sorghum
                          CNPMS/EMBRAPA, Brazil
      Research partially supported by CNPq  (National Council for Technical and Scientific
      Development), Contract 040-2941/90.
                                BRAZIL-1

-------
ACKNOWLEDGEMENTS

The authors address thanks to the following institutions: EMBRAPA (CNPT, CNPMS, CNPS,
CNPMF,  CPAA,  CPATU,  CPATB,  CPATSA,  NMA),  AC-SA/SP,   IPAGRO-SAA/RS,
DNEMET/CRMC, CNPq, US EPA, US AID, and collaborators: Dr. Rogerio Remo Alfonsi, Dr. Ivo
Pierozzi Jr., Dr. Aristides CSmara Bueno, Dr. Jaime Ricardo Tavares Maluf, Dr. Fernando Silveira
da Mota, Dr. Moacir Antdnio Berlato, Dr. Delmar Pottker, Dr. Sirio Wietholter, Dr. Gilberto Rocca
da Cunha, Dr. Douglas Godwin, Dr. Cynthia Rosenzweig, Dr. Ron Benioff, Dr. Ana Iglesias. Also
to local support, Adriano de Mello e Silva and Pitagoras Ribeiro, and to family, Marisa, Otavio, and
Fabiana.
                                     BRAZIL-2

-------
TABLE OF CONTENTS

SUMMARY
INTRODUCTION
       Agroecological Regions and Sites
METHODS
       Baseline Climate
       Sensitivity Scenarios
       Climate. Change Scenarios
       GISS Transient Scenarios
       Crop Models and Inputs'
       Calibration and Validation of the Crop Models
       Limitations of the Study
RESULTS AND DISCUSSION
       Sensitivity Analysis
       Crop Changes under GCM Climate Change Scenarios
       National Grain Yields
       Transient Results
       Adaptation Strategies to Climate Change
CONCLUSIONS
REFERENCES
                                     BRAZIL-3

-------
 SUMMARY

        Wheat, maize and soybean production were simulated with the CERES-Wheat, CERES-
 Maize and SOYGRO crop models to estimate the possible impact of global climate change at 13
 sites in Brazil. Climate change scenarios for each site were created by combining observed climate
 data with  the output of the GISS, GFDL and UKMO  General  Circulation Models  (GCMs).
 Historical  climate data were used as the base scenarios.  The three  GCMs project  temperature
 increases, changes in precipitation, and small variations of solar radiation. The sensitivity of the
 simulated crop yield to changes in temperature, precipitation, and atmospheric CO2 was also tested.
        Temperature increases resulted in lower grain yield, biomass, and season length for wheat
 and maize, but  soybean was less affected. The simulated physiological effects of CO2 partially
 compensated for the yield decreases under the scenarios of climate change alone in wheat and maize,
 and fully compensated in the case of soybean. Wheat yields declined at all sites under the three
 GCM climate change scenarios, although the magnitude of the decreases varied with scenario and
 region; the largest decreases occurred in the Central region (24 and 46% under  the GFDL and
 UKMO scenarios respectively) and in the Central South region (43% under the GISS scenario).
 Maize yields also declined at all sites. The largest decreases  were in the Northern region (26%
 under  the GISS scenario), while the decreases in the South and Central South region varied but
 were less than 20%. In contrast to wheat and maize, simulated soybean yield remained  the same (in
 the Northern regions) or increased under the GCM climate change scenarios.
        Adaptation strategies such as irrigation, changes in planting date, and  increased nitrogen
 fertilization helped improve yields, but not enough to compensate for all of the losses under climate
 change. The crop simulation model was used to design a hypothetical more heat-resistant cultivar
 that showed promising results for potential adaptation to warmer  climates, but the feasibility of this
 strategy still needs to be tested through breeding programs.
INTRODUCTION

        The rising concentration of greenhouse gases in the atmosphere may lead to increased
global temperatures (IPCC 1990) and the study of the potential impacts of global climate change
on ecosystems and agriculture is a relatively new area of research (Smith and Tirpak 1989a, 1989b).
In Brazil, very few studies have been conducted in this area. Some results have been presented by
Mota et al (1984) using statistical models. In this study, we analyze the possible impact of global
climate change on wheat, maize, and soybean production in Brazil using crop growth simulation
models and climate change scenarios created with GCM results. These crops  are the major
agricultural commodities in Brazil (along with rice), and cover approximately 25 million ha of
cultivated land. In addition, this study analyzes adaptation  strategies to minimize the impact of the
projected climate changes on crop production.

Agroecological Regions and  Sites

        Thirteen sites were selected for the simulation study (Figure 1 and Table 1). The locations
of the sites  range from latitude 30 degrees South to close to the equator. Nine sites are concentrated
in the subtropical and tropical/high-elevation regions and the other four sites are distributed among
the tropical, semi-arid,  and equatorial/sub-equatorial regions. The sites  represent the  main
agricultural regions of Brazil, and they were selected based on previous agroclimatic studies (Mota

                                        BRAZIL-4

-------
1989; Mota and Agendes 1986; Alfonsi et al 1981; Queiroz et al. 1979). Nine sites are located in the
most important agricultural region of the country, the South and Central-South regions. Almost
99% of the national wheat production and more than 80% of the maize and soybean production are
concentrated in these regions.

METHODS

Baseline Climate

        The baseline climate is represented by the historical data available for each site during the
period 1951-1980 (Table 2) and includes daily maximum and minimum temperatures, precipitation,
and hours of sunshine. Daily solar radiation data were generated using hours of sunshine, and the
results were compared to actual data where  available.
        The lowest mean annual temperature is found in Vacaria, which has the  most significant
seasonal temperature variation. In contrast, Belem shows the highest mean annual temperature with
the smallest variation during  the  year. Precipitation  varies among the sites, with  the highest
precipitation occurring in the northern sites and the lowest in the northeast. There is also variability
in seasonal precipitation in some central sites, where the lowest monthly rainfall occurs during the
winter. Seasonal differences in solar radiation are more apparent in the southern  sites.

Sensitivity Scenarios

        To analyze the sensitivity of the crop models to temperature, precipitation, and CO2 levels,
sensitivity scenarios were created by combining step changes in  the climate variables (0, +2C,
+4 C temperature changes; 0, +20%, -20% precipitation changes). The physiological effects of 555
ppm CO2 were also considered for each scenario.

Climate Change Scenarios

        This study used climate change scenarios generated by three equilibrium general circulation
models: Goddard  Institute for Space Studies (GISS) (Hansen  et al. 1983), Geophysical Fluid
Dynamics Laboratory (GFDL) (Manabe and Wetherald 1987), and United Kingdom Meteorological
Office (UKMO) (Wilson and Mitchell 1987). These GCMs are  three-dimensional models which
incorporate physical knowledge of the processes involved in the transfer of the energy among the
earth,  oceans, and atmosphere.
        The climate change scenarios for each site were created by applying the changes between
the  lxCO2 and 2xCO2 monthly  GCM-simulated climate variables (differences were  used for
temperatures and ratios for precipitation and solar radiation) to the corresponding daily baseline
climate variables.  The scenario changes in mean annual temperature,  precipitation and  solar
radiation are shown in Table 3 for each site.
        The annual temperature increases range from +2C to +6C, but there are significant
differences among the scenario projections. In general, the largest increases correspond to the
UKMO scenario. The sites  included in the Central-South region show the largest temperature
changes from March to  November, which is the growing period of wheat. For the sites in the
Northeast region, the UKMO scenario projects higher temperatures in the winter (June to August).
        Under the climate change scenarios, precipitation projections vary greatly, especially for the
more southern sites. For the most part, the projected annual precipitation increased in comparison

                                        BRAZIL-5

-------
with the current (observed) climate. The GFDL scenario projects the largest precipitation increases
for sites in the South region from September to November, and for sites in the Central-South region
from March to May. The precipitation projected by the GFDL and UKMO scenarios for December
for most of the Southern sites is lower than the current precipitation, indicating a higher probability
of drought problems for summer crops. Smaller annual increases in precipitation are projected for
the sites in the Northeast and North regions, where the UKMO scenario projects decreases of about
10% to 15%. The precipitation reductions projected by the UKMO scenario during the summer
(December to February) and winter (June to August) might bring additional stress for the crops in
the northern sites, especially considering the currently low water supply in the winter.

GISS Transient Scenarios

        The GISS transient scenarios were used to assess the effect of gradual changes in  climate
for the decades of the 2010s, 2030s, and 2050s on  crop production  (Hansen et al.  1988). The
atmospheric CO2 concentrations considered were 405 ppm, 460 ppm, and 530 ppm, respectively, for
these decades.

Crop  Models and Inputs

        Crop-growth models developed  by IBSNAT (Jones et al.  1990) were  selected for  this
simulation study: CERES-Wheat version 2.10 (Godwin et al. 1989; Ritchie and Otter 1985), CERES-
Maize  (Jones  and Kiniry  1986), and SOYGRO (Jones et al. 1988).  The IBSNAT crop  models
simulate plant development  and growth by integrating soil, plant, climate (daily maximum  and
minimum temperatures, precipitation, and solar radiation), and .management factors.
        The IBSNAT crop models include an option to simulate the direct physiological effects of
CO2 atmospheric concentrations on plant photosynthesis and water use, based on experimental
results (Rose 1989; Curry et al 1990). The photosynthetic enhancement of 555 ppm CO2 is 1.17 for
wheat, 1.21 for soybeans, and 1.06 for maize.  Increases in stomatal resistance are also simulated.
        Soil data were obtained from regional soil survey studies (Brasil 1971, 1973; Larach et al.
1984; Oliveira et al. 1984; Mothei et al. 1979; Santos et al 1983).  Soils were chosen to represent the
local soils of the sites selected. The soil profile for Passo Fundo (crop  model calibration site)  was
created with local data.
        The cultivars, plant  population, and planting date used as  input for  the crop  growth
simulation models are shown in Appendix A. The information was obtained from published crop
management reports (QueiiozetaL 1979; Miyasaka and Medina 1981; Vernetti 1983; Reuniao 1987;
CSBPT 1988; CCSBPT 1989; CCBPT 1988).
        Although there are water deficits in some regions, almost all maize and soybean crops are
cultivated without irrigation. For winter wheat, irrigation is  recommended for some regions,
including Campinas, Campo Grande, and Sete Lagoas. In Campo Grande, wheat could be produced
in a rainfed  system, if it is  planted at a different planting date.

Calibration  and Validation of the Crop Models

        CERES-Wheat. The local wheat cultivar BR 14 was chosen for most dryland sites,  and its
genetic coefficients were determined using experimental field data. The cultivar  ANZA, with the
original genetic coefficients included in the DSSAT database, was selected for the irrigation sites.
The model was validated in Passo Fundo with data from several field experiments, (Figure 2) and

                                       BRAZIL-6

-------
there is a satisfactory agreement between the observed and estimated grain yields (Siqueira 1991).
In Brazil, the CERES model has  also been validated for wheat in Sao Paulo  (South region)
(Anunciacao and Liu 1991).

       CERES-Maize.  The genetic coefficients for the local cultivar used in the South and Central-
South regions (PIO 3230) were determined by comparing data from a field experiment conducted
in Taquari involving irrigation, nitrogen, and population levels (Matzenauer et al. 1988). Figure 2
shows selected results of the validation. Cultivar SUWAN-1, with the original genetic coefficients
included in the DSSAT data base, was used for the other regions because the simulated results are
in agreement with the regional observed crop parameters.

       SOYGRO. The genetic coefficients of the cultivar DAVIS were calibrated  using data from
a field experiment conducted in Passo Fundo in 1989, and the model was validated with data from
several field experiments (Siqueira and Berg 1991) (Figure 2). The cultivars VICOJA and JUPITER
were used for the sites in the Northeast and North regions. In Belem, simulated and observed yields
and anthesis dates show a close relationship.

Limitations of the Study

       The crop models have not been validated in all of the regions in this study.  Technology and
land use are assumed to be constant, even though it is certain that they will change in the future.
The direct physiological effects of CO2 on crop development and yield may be different than the
simulated effects.
        The GCM climate change scenarios do not include changes in climatic variability that might
represent a very important factor for crop production, especially in the more vulnerable regions. The
spatial resolution of the climate change scenarios as created in this study is low.

RESULTS AND DISCUSSION

Sensitivity Analysis

        A sensitivity analysis was conducted at each site to evaluate the effect of step changes in
temperature and  precipitation on wheat, maize, and soybean (Tables  4, 5, and 6). For wheat, an
increase  in temperature resulted in significant reductions in crop season length  and grain yields
(Table 4). An increment of +4C resulted in a shortening of the crop season length by about  15%,
and a 40% to 50%  decrease  in grain yields.  With the physiological effects of 555 ppm CO2 on the
crop included in the +4C scenario, wheat yield increased in comparison with the +4C scenario
alone, but the effects did not completely compensate for the negative impact of higher temperature
on yields.
        For maize, the effect of warmer temperatures  on grain yield varied  among the regions
(Table 5), ranging  from about -20% in the South and Central-South to -28% in the Northeast
region. Warmer temperatures reduced the maize crop season length by an average of 15%, but the
effect was greater for the more southern latitude sites. There were no significant  effects on maize
grain yield from increased precipitation except in the Northeast region. Lower precipitation may be
beneficial for maize in the North region and detrimental in the Northeast. The physiological effects
of 555 ppm of CO2 were smaller on maize grain yields than on wheat yields, as expected from the
lower  response of maize, a C4 crop, to higher CO2.

                                        BRAZIL-7

-------
        Temperature and precipitation changes had a different effect on soybean yields than on
maize and wheat yields (Table 6). A 2C temperature increase with the physiological effects of
doubled CO2 resulted in small increases in yield. In addition, the temperature increase did not
significantly affect the length of  the crop season. With the physiological effects of CO2, soybean
yields generally increased under the scenarios with higher temperatures.

Crop Changes under GCM Climate Change Scenarios

        The results of the simulation of wheat, maize, and soybean growth under the GCM climate
change scenarios are shown in Table 7. Tables 8, 9, and 10 and Figure 3  show the results for each
region.

        Wheat.  GCM scenarios projected wheat biomass, grain yield, and season length decreases
in comparison with baseline data for all sites (Table 7).  Wheat yield reductions were the largest
with the UKMO scenario. All scenarios  projected a shorter wheat crop  season, especially for the
South region. These changes were driven  by the temperature increases of the scenarios, as described
in the sensitivity analysis.
        Figure 3 shows the percent change in wheat grain yield for each region under the different
climate  scenarios. With the physiological effects of CO2,  the negative effect of the climate change
alone was partially diminished (Figure 3). Wheat yields in the Central region were most vulnerable
to future climate changes under the GFDL and the UKMO scenarios: yield reductions were near
24 and 46%. Under the GISS scenario, losses were projected to be about 43% for the Central-South
region. For the most part, the South region appeared to be less vulnerable to the climate changes,
with projected average losses in yield of  22%.

        Maize.   All GCM scenarios projected reductions in maize biomass production, grain yield,
and season length, when compared to the present climate (Tables 7 and 9  and Figure 3). Decreases
in season length under the climate change scenarios varied in each region, but they averaged about
15%.
        The effect of climate change on crop season length and yield was a  consequence  of the
increases in temperature  projected by  the GCMs. With the physiological effects  of CO2, the
projected reductions in grain yield were diminished in comparison with the yields projected under
the climate change scenarios alone. Under the GFDL and the UKMO scenarios with physiological
CO2 effects, the largest decreases  in yield were in the South and the Central-South regions, ranging
from 11% to 20%. Under the GISS scenario, the largest reductions were found in the North region
(24%).

        Soybean. In general,  reductions in soybean biomass and grain yield were smaller than
reductions in wheat and maize projected by the GCM scenarios (Tables 7 and 10 and Figure 3).
With the GCM scenarios alone, there were soybean yield and biomass reductions in  almost all of
the regions. Mean reductions in grain yield varied from 5% to 31% (Table 7).  W i t  h    the
physiological effects of CO2 on yield, the SOYGRO model simulated significant yield increases: gains
averaged 22% (Table 7). The results were consistent in all regions except the Northeast where yields
decreased, even when the physiological effects of CO2 were considered (Figure 4). A slightly shorter
crop season length was simulated for soybean with all GCM scenarios for  the sites in the South and
the Central-South regions, but this effect  was very small in comparison with the shorter crop season
of wheat and maize under the same conditions.
                                        BRAZIL-8

-------
        These results agree with the aforementioned sensitivity analysis and indicate that soybean
production in Brazil might not be as adversely affected by climate change as wheat or maize. The
more positive results projected for soybean, however, were dependent on the beneficial physiological
effects of CO2 concentration.

National Grain Yields

        The effects of the climatic changes on national crop production wheat, maize, and soybean
yields were estimated by aggregating the regional results weighted by cultivated area. For example,
the yield changes from a region that represents 50% of the national cultivated wheat area was given
a weight of 50% in the national  results (Table 11).  All results reported  include  the beneficial
physiological CO2 effects on crop yield. Crop management, technology, and distribution of cultivated
land were assumed to be constant, although these will change in the future.
        Wheat is currently grown on 3.6 million ha in Brazil. Using the changes in regional wheat
yields under the GCM scenarios and the contribution of these regions to the national wheat area,
we estimate  that the possible impact of climate change scenarios on national wheat  production
would be large (reductions of 33%, 18% and 34% under the GISS, GFDL, and UKMO scenarios,
respectively). Although  significant reductions in wheat yield are projected for the Central region,
the impact of these regional reductions on the national yield is not highly significant because of the
small acreage that is cultivated (1%) in that region. Maize is the most widely cultivated crop  in
Brazil, now cultivated on about 22 million ha. National maize yields were reduced 11%, 11%, and
16% under the GISS, GFDL, and UKMO scenarios, respectively.
        Soybean is the second most widely cultivated crop in Brazil, occupying about 15.5 million
ha. Only in the Northeast region under UKMO were losses projected for simulated soybean yields.
In all other regions, soybean yields increased under the conditions of climate change, as simulated
in this study. The estimated changes in national yields under the GISS, GFDL, and UKMO climate
scenarios are +26%, +23%, and +18%, respectively. The 17% decrease in yield under the UKMO
scenario in the Northeast region did not have a large impact on the national yield estimate since
that region only contributes about 1% of the total soybean cultivation area in the country.

Transient Results

         Wheat and maize yields and season lengths were simulated  under the GISS transient
scenarios in Passo Fundo (South region) (Figure 4). Linear increases in temperature were projected
from  1990 to the year 2050. Wheat yield  and season length decreased  under GISS transient
scenarios; the rate of reduction decreased after the year 2030. In contrast to wheat, the simulated
maize grain yield increased until the year 2030 with the physiological effects of CO2 (460 ppm) and
then decreased. The length of the maize crop season decreased linearly over this same period.

Adaptation Strategies to Climate Change

         Wheat and maize development and yield were affected by the climate changes projected by
 the GCM scenarios, and the changes were largest under the UKMO scenario. This part of the study
 aims to define possible alternatives that would compensate for the negative impact of the climate
 changes on wheat and maize. The simulation was carried out in Passo Fundo because of the careful
 calibration and validation of the CERES model at that site. Additional and preliminary strategies
 were also simulated for maize and soybeans in

                                         BRAZIL-9

-------

-------
 Petrolina in the Northeast region, a vulnerable area to climate change.
        Crop management alternatives (irrigation, new cultivars, and nitrogen management) were
 evaluated as possible adaptation strategies for wheat under the climate change conditions (Figure
 13). The UKMO scenario was run  at Passo Fundo with various combinations of wheat genetic
 coefficients and management practices. When tested separately, the adaptation strategies failed to
 improve yields  significantly. Only when  several high-cost adaptation strategies (irrigation and
 nitrogen management) were combined were wheat yields restored to 90% of their baseline values.
        The changes in maize crop management that were tested (irrigation and changes in planting
 date) did not compensate for the decrease in maize yield under the UKMO scenario in Passo Fundo
 (South region). The crop growth model can test the performance of a hypothetical new cultivar
 under the  conditions of climate change. The development of a hypothetical new cultivar with a
 different P5 "genetic coefficient" would improve maize yield production under the climate change
 scenarios. The P5 coefficient is the growing degree days from flowering to maturity (Ritchie et al
 1989). A higher P5 coefficient extends the simulated duration of the grain-filling  period  under
 warmer climate conditions. An increase of 20% in the actual value of the P5 coefficient compensated
 for the projected grain yield losses caused by the UKMO scenario at Passo Fundo. The feasibility
 of breeding for extended grain-filling periods should be further explored by crop geneticists.
        At Petrolina, an improvement in the crop management practices such as irrigation and
 increased nitrogen fertilization could compensate for the yield decreases under the UKMO scenario.
 However it is important to notice that in  this case, base yields also increase  substantially. At this
 site, irrigation also increases soybean yields under the base case and the UKMO scenario and would
 fully compensate for any negative impact of climate change on soybean yields in the vulnerable
 Northeast region.

 CONCLUSIONS

        Projected climate change with the physiological effects of CO2 reduced modeled wheat and
 maize yields significantly, while soybean yields increased. Most of the wheat and maize yield  losses
were associated with a significant shortening of the growing season due to increased temperatures,
which allowed less time for biomass accumulation. In contrast, soybean growing season was not
reduced significantly, because soybean responds to temperature either  positively or negatively
depending  on phenological stage. The variable season length response probably contributed to the
more positive soybean yield responses to climate change. Soybean temperature responses at a wide
range of sites require further study.
        Adaptation strategies such as irrigation, changes in planting date, and increased nitrogen
fertilization helped improve wheat and maize yields, but not enough to compensate  for all of the
losses projected for  the climate change scenarios. The development of a hypothetical new,  more
heat-resistant cultivar showed promising results, although the  feasibility of this strategy still  needs
to be tested through breeding programs.
        These results imply substantial reduction in the national production of wheat and maize and
increases in the production of soybean under climate change. Differential impacts on the country's
regions are projected, with the Northeast especially vulnerable to maize and soybean yield decreases
and the Central region vulnerable to wheat yield declines.
                                        BRAZIL-10

-------
REFERENCES

Alfonsi, R.R., M.J. Pedro Junior, and AP. de Camargo. 1981. Zoneamento agroclimatico da soja
        no Brasil. In Miysaka, S. and Medina, J.C., eds. A soja no Brazil. s.L, Banco do Estado de
        Sao  Paulo/FINEP-FNDCT/ Ministerio   da  Agricultura/Ministerio  da  Industria  e
        Comercio/Coqp. Agricola de Cotia.

Anunciacao, Y.M.T.,  and W.T. Liu. 1991. Estimativa  da produtividade do trigo em  campo
        experimental, utilizando o modelo fisiologico CERES-Wheat V.2.10. In Congresso Brasileiro
        de Agrometeorologia, 7th. Vicosa. Universidade Federal de Vicosa.

Brasil. 1971. Ministerio da Agricultura. Departamento Nacional de Pesquisa Agropecuaria. Divisao
        de Pesquisa Pedologica. Levantamento de reconhecimento dos solos do sul do estado do Mato
        Grosso. Rio de Janeiro. (Tech. Bull. 18).

Brasil. 1973. Ministerio da Agricultura. Departamento Nacional de Pesquisa Agropecuaria. Divisao
        de Pesquisa Pedologica. Levantamento de reconhecimento dos solos do estado do Rio Grande
        do Sul  Recife. (Tech. Bull. 30).

CCBPT. 1988. Reuniao da Comissao Centre Brasileira de Pesquisa de Trigo, 4th. Sao Paulo, SP. In
        Recomendacoes da Comissao Centra Brasileira de Pesquisa de Trigo para o ano de 1988. Sao
        Paulo, CAC-CC.

CCSBPT. 1989. Reuniao  da Comissao Centro-Sul Brasileira de Pesquisa de Trigo, 5th. Cornelio
        Procopio, PR. In Recomendacoes da Comissao Centro-Sul Brasileira de Pesquisa de Trigo
        para 1989. Londrina, PR. CAC-CC.

CSBPT. 1988. Reuniao da Comissao Sul-Brasileira de Pesquisa de Trigo, 20th.  Porto Alegre, RS.
        In Recomendacoes da Comissao Sul-Brasileira de Pesquisa  de  Trigo para a cultura do Trigo
        em 1988. Passo Fundo, EMBRAPA-CNPT.

Curry, R.B., R.M.  Peart,  J.W.  Jones, K.J. Boote, and L.H. Allen. 1990. Simulation as a tool for
        analyzing crop response to climate change. American Society of Agricultural Engineers,
        33(3):981-990.

Godwin, D., J. Ritchie, U. Singh,  and  L. Hunt.  1989. A user's guide to CERES wheat -  V2.10.
        International Fertilizer Development Center. Muscle Shoals, USA

Hansen, J., I. Fung, A Lacis, D. Rind, S. Lebedeff, R. Ruedy, G. Russel, and P. Stone. 1988. Global
        Climate Changes as Forecast by Goddard Institute for Space Studies Three-Dimensional
        Model. /. Geophys. Res. 93:9341-9364.

Hansen, J., G. Russell, D. Rind, P. Stone, A  Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983.
        Efficient Three-Dimensional Global Models for Climate Studies: Models I and II. Monthly
        Weather Review, Vol. Ill, No.4.
                                       BRAZIL-11

-------
 IPCC. 1990. Climate Change: The IPCC Scientific Assessment. J.T. Houghton, G.J. Jenkins, and J.J.
        Ephraums (eds). Intergovernmental Panel on Climate Change. Cambridge.

 Jones, C.A., and J.R. Kiniry, eds. 1986. CERES-Maize: A Simulation Model of Maize Growth and
        Development. College Station. Texas A&M University Press.

 Jones, J.W., K.J. Boote, S.S. Jagtap, G.  Hoogenboom, and G.G. Wilkerson. 1988. SOYGRO v5.41:
        Soybean Crop Growth Simulation Model User's Guide. Florida Agr. Exp. Sta. Journal No.
        8304, IFAS. Univ. of Florida.

 Jones, J.W., S.S. Jagtap, G.  Hoogenboom, and G.Y. Tsuji. 1990. The structure and function of
        DSSAT. In International Benchmark Sites Network for Agrotechnology Transfer. Proceedings
        of the IBSNAT Symposium: Decision Support System for Agrotechnology Transfer. University
        of Hawaii.

 Larach, J.O.I., A. Cardoso, A.P. de Carvalho, D.P. Hochmuller, P.J. Fasolo, and M. de Rauen. 1984.
        Levantamento   de  reconhecimento   dos  solos   do   estado   do   Parana.
        EMBRAPA-SNCLS/SUDESUL/IAPAR, Curitiba. 2t. (Tech. Bull, 27).

 Manabe, S., and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in
        atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.

 Matzenauer, R., AL. Pons, J.R.T. Maluf, and A.C. Bueno. 1988. Efeito da irrigacao e densidade de
        plantas na cultura do milho: I. Rendimento de graos e componentes de rendimento.  In
        Reuniao tecnica anual do milho, 33. Porto Alegre, RS.

 Miyasaka, S. and J.C. Medina. 1981. A Soja no BrasiL Banco do  Brasil/FINEP/MA/MIC.  Secao de
        Divulgacao, ITAL. 1062p.

Mota, F.S. da 1989. Clima, technologia  e produtividade do trigo no Brasil. In Agrometeorologia do
        trigo no BrasiL Campinas, Sociedade Brasileira de Agrometeorologia

Mota, F.S. da and M.O. de O. Agendes. 1986. Clima e agricultura no Brasil. Porto Alegre, SAGRA.

Mota, F.S. da, Z.B. Berny, and M.O. de O. Agendes. 1984. Perspectivas dimaticas da agricultura ate
        o ano  2000,  no sul  do Brasil,sob  uma provavel  agao  do   efeito  estufa.  Pelotas,
        EMBRAPA-DDT. 70p. (EMBRAPA-UEPAE de Pelotas. Doc. 11).

Mothei, E.P., J.AM. do  Amaral,  and  R.D. dos Santos. 1979. Levantamento de reconhecimento
        detalhado e aptidao agricola dos solos da area do Centra  Nacional de Pesquisa de Gado de
        Corte, Mato Grosso do Sul. EMBRAPA-SNLCS, Rio de Janeiro, 225p. (Tech. Bull, 59).

Oliveira, J.B. de and J.R.F. Menk. 1984.  Latossolos  roxos do estado de Sao Paulo.  Institute
        Agronomico, Campinas. (Tech. Bull. 82).
                                      BRAZIL-12

-------
Queiroz, E.F., N. Neumaier, and E. Torres. 1979.  Ecologia e manejo da cultura.  In Empresa
       Brasileira de pesquisa agropecuaria. Centro Nacional de Pesquisa de Soja, Londrina, PR.
       Ecologia, manejo e adubacao da soja. Londrina. (EMBRAPA-CNPSO. Tech. Circ. 2).

Reuniao de Pesquisa de Soja da Regiao Sul, 15th. Cruz Alta, RS. In Recomendacoes da Reuniao de
       Pesquisa de Soja da Regiao Sul-1987/88. Cruz Alta, FECOTRIGO-CEP. 75P.

Ritchie, J.T., and S. Otter. 1985. Description and performance of CERES-wheat: A user-oriented
       wheat yield model. In Willis, W.O., ed. ARS Wheat Yield Project. Washington, DC: USDA
        Agricultural Research Service. ARS-38.

Ritchie, J., U. Singh, G. Godwin, and L. Hunt. 1989. A user's guide to CERES-maize - V2.10 Muscle
        Shoals, USA. International Fertilizer Development Center.

Rose, E. 1989. Direct (physiological) effects of increasing CO2 on crop plants and their interactions
        with indirect (climatic) events. In: The Potential Effects of Climate Change in  the United
        States. Smith, J.B. and D.A. Tirpak, eds. EPA. Washington, DC. V2.

Santos, P.CT. dos, L.S. Vieira, M. de N.F. Vieira, and A Cardoso. 1983. Os solos da Faculdade de
        Ciencias Agrarias do Para. FCAP. Belem. (FCAP. Infom. 5).

Siqueira, O.J.F. de 1991. Avancos na informatica, aplicados as relacoes solo- planta-clima: Exemplo,
        utilizando  o sistema de suporte  de decisao  "DSSAT" -  Nitrogenio x Trigo. In  Reuniao
        national de pesquisa de trigo, 16. Dourados, MS. Resumes. Dourados: EMBRAPA/UEPAE

Siqueira, OJ.F. de and M. van den Berg. 1991. Validacao do modelo de  simulacao SOYGRO-Soja
        -  Passo  Fundo, RS. In Reuniao de pesquisa de soja da regiao sul,  19. Pelotas, RS. Soja;
        resultados de pesquisa 1990-1991. Passo Fundo: EMBRAPA/CNPT. (Doc. 3).

Smith, J.B. and D.A. Tirpak. 1989a. The potential effects of global climate change on the United States:
       Appendix C - Agriculture.  EPA. Washington, DC. VI and V2.

Smith, J.B. and D.A. Tirpak. 1989b. The potential effects of global climate change on the United States.
        Report to Congress. EPA. Washington, DC. 413p.

Vernetti, F. de  Jesus. 1983. Soja. Campinas, Fundacao Cargill.

Wilson, C.A., and J.F.B. Mitchell. 1987. A doubled CO2 Climate Sensitivity Experiment with a
        Global Model Including a Simple Ocean. Journal of Geophysical Research, 92:13315-13343.
                                       BRAZIL-13

-------
Table 1.
 Site
Characteristics of sites selected for the simulation study.
               Lat.
Long.
Elevation Weather
      (m)       Data
Soil
Pelotas, RS
Passo Fundo, RS
Sao Borja, RS
Vacaria, RS
Ponta Grossa, PR
Londrina, PR
Campinas, SP
Campo Grande, MS
Sete Lagoas, MG
Cruz das Almas, BA
Petrolina, PE
Manaus, AM
Belem, PA
31.47S
28.15S
28.39S
28.33S
25.06S
23.19S
22.53S
20.27S
19.28S
12.40S
9.23S
3.08S
1.28S
52.29W
52.24W
56.00W
50.42W
50.10W
51.19W
47.06W
54.37W
44.15W
39.06W
40.30W
60.01W
48.27W
13
667
99
955
868
566
669
530
732
226
366
48
24
52/80
51/80
56/80
51/80
54/80
58/80
51/80
74/80
60/80
71/80
65/80
71/80
67/80
Hapludult
Haplorthox
Paleudalf
Haplohumox
Haplorthox
Haplorthox
Eutrorthox
Haplorthox
Haplusthox
Haplorthox
Eutrusthox
Acrorthox
Haplorthox
                                        BRAZIL-14

-------
Table 2.
Observed baseline climate at selected sites.
 TEMPERATURE
           Dec-Feb    Mar-May     Jun-Aug
                                                               Sep-Nov
Annual
Pelotas
Passo Fundo
Sao Borja
Vacaria
Ponta Grossa
Londrina
Campinas
Campo Grande
Sete Lagoas
Cruz das Almas
Petrolina
Manaus
Belem
PRECIPITATION
(mm)
Pelotas
Passo Fundo
Sao Borja
Vacaria
Ponta Grossa
Londrina
Campinas
Campo Grande
Sete Lagoas
Cruz das Almas
Petrolina
Manaus
Belem
22.9
22.6
25.0
20.2
22.0
24.3
24.0
25.3
23.4
25.9
26.6
26.6
27.0
Dec-Feb

297
474
348
384
480
627
648
678
738
273
255
708
960
18.7
18.4
20.5
15.9
18.6
21.4
21.6
23.3
21.8
24.9
25.6
26.6
27.0
Mar-May

264
348
372
291
309
324
267
390
201
342
255
903
1101
13.5
14.2
15.5
11.8
15.1
18.0
18.9
21.6
19.1
22.1
24.0
26.6
27.2
Jun-Aug

366
441
306
369
276
213
114
123
33
303
24
348
450
17.5
18.4
20.0
15.9
18.8
21.8
22.1
24.0
22.6
24.2
27.0
27.6
27.4
Sep-Nov

315
501
399
408
381
420
330
507
393
243
69
420
366
18.2
18.4
20.2
15.9
18.6
21.4
21.6
23.6
21.7
24.3
25.8
26.8
27.2
Annual

1,242
1,764
1,425
1,452
1,446
1,584
1,359
1,698
1,365
1,161
603
2,379
2,877
                                   BRAZIL-15

-------
SOLAR RAD. (MJ/m2)    Dec-Feb    Mar-May    Jun-Aug     Sep-Nov
Annual
Pelotas
Passo Fundo
Sao Borja
Vacaria
Ponta Grossa
Londrina
Campinas
Campo Grande
Sete Lagoas
Cruz das Almas
Petrolina
Manaus
Belem
18.4
17.7
17.8
17.1
15.1
16.3
16.1
16.2
15.9
15.7
163
11.7
13.1
11.2
11.6
11.5
11.4
11.4
13.3
13.4
13.7
14.5
13.2
14.3
11.2
12.7
8.0
8.6
8.2
8.5
9.6
11.3
11.6
12.1
13.6
11.3
13.4
13.7
15.6
14.6
14.7
15.0
14.4
13.7
15.2
15.3
15.6
15.0
14.3
17.2
14.3
16.2
13.0
13.1
13.1
12.8
12.4
14.0
14.1
14.4
14.7
13.6
15.3
12.7
14.4
                                BRAZIL-16

-------
Table 3.        Annual GCM scenario changes in temperature, precipitation, and solar radiation
               at selected sites.
                    CHANGES IN
                 TEMPERATURE (C)
PRECIPITATION
    RATIOS
SOLAR RADIATION
     RATIOS
SITE
Pelotas
Passo Fundo
Sao Borja
Vacaria
Ponta Grossa
Londrina
Campinas
Campo
Grande
Sete Lagoas
Cruz das
Almas
Petrolina
Manaus
Belem
GISS
4.0
4.5
4.8
4.5
4.5
4.5
4.5
4.5

4.1
4.3

43
3.5
4.0
GFDL
4.2
3.5
3.5
3.5
3.1
3.1
3.2
3.0

4.2
2.5

2.5
2.7
2.7
UKMO
4.5
6.1
5.6
6.0
6.0
4.8
4.8
5.7

6.0
4.5

4.6
3.9
3.9
GISS
1.00
0.98
1.65
0.98
0.98
1.24
1.24
1.24

1.08
1.03

1.03
1.25
1.09
GKDL
1.00
1.20
1.20
1.20
1.21
1.21
1.25
1.17

1.07
1.08

1.08
1.20
1.00
UKMO
1.14
1.05
1.15
1.05
1.05
1.20
1.20
1.08

1.26
0.83

0.84
0.97
0.80
GISS
1.06
1.02
0.98
1.02
1.02
1.00
1.00
1.00

1.05
1.02

1.02
1.02
1.02
GFDL
1.01
0.99
0.99
0.99
0.99
0.99
0.96
0.98

0.98
1.01

1.01
1.01
1.01
UKMO
1.03
1.04
1.02
1.04
1.04
1.03
1.03
1.03

1.14
1.02

1.04
1.11
1.00
                                       BRAZIL-17

-------
Table 4.       Sensitivity of the CERES-Wheat model to changes in temperature, precipitation,
              and CO2 levels (330 ppm and 555 ppm CO2). Simulated yields and season length
              for different regions.
                         SIMULATED GRAIN YIELD (t ha'1)
SOUTH
Precip.
Diff.
(%)
0%
0%
0%
+20%
+20%
+20%
-20%
-20%
-20%
Temp.
Diff.
(C)
+0
+2
+4
0
+2
+4
0
+2
+4
330
ppm
2.30
1.88
1.47
2.26
1.86
1.46
2.32
1.89
1.48
555
ppm
2.66
2.27
1.87
2.61
2.24
1.85
2.71
2.30
1.89
C. SOUTH
330
ppm
2.36
1.63
1.03
2.35
1.64
1.04
2.36
1.63
1.03
555
ppm
2.91
2.22
1.54
2.90
2.21
1.54
2.96
2.24
1.54
CENTRAL
330 ppm
2.93
2.18
1.44
2.93
2.17
1.44
2.94
2.18
1.44
555
ppm
3.36
2.71
2.00
3.34
2.70
2.00
3.32
2.67
1.97
                      SIMULATED SEASON LENGTH (DAYS)*

Temp.
Diff (C)
0
+2
+4
SOUTH
330
ppm
124
116
108
C. SOUTH
330
ppm
103
95
89
CENTRAL
330
ppm
100
91
85
*Season length changes (as simulated with the CERES-Wheat model) were affected only by
temperature.
                                    BRAZIL-18

-------
Table 5.        Sensitivity of the CERES-Maize model to changes in temperature, precipitation,
               and CO2 level (330 ppm and 555 ppm of CO^). Simulated yields and season length
               for different regions.
                          SIMULATED GRAIN YIELD (t ha'1)
Precip.
Diff.


0%
0%
0%
+20%
+20%
+20%
-20%
-20%
-20%
Temp.
Diff.

0
+2
+4
0
+2
+4
0
+2
+4
SOUTH
330 555
ppm
7.78
7.14
6.31
7.92
7.26
6.41
7.57
6.92
6.11
ppm
8.49
7.81
6.90
8.51
7.85
6.95
8.37
7.63
6.70
SIMULATED
SOUTH




Temp.
Diff.
0
+2
+4
330
ppm
134
118
103




C.SOUTH
330 555
ppm
6.66
6.15
5.36
6.58
6.10
5.33
6.70
6.17
5.35
SEASON
ppm
7.05
6.56
5.75
6.94
6.47
5.70
7.14
6.62
5.78
LENGTH
C.SOUTH
330
ppm
117
106
98




N.EAST
330 555
ppm
4.87
4.11
3.49
5.04
4.32
3.67
4.54
3.79
3.16
(DAYS)*
ppm
5.44
4.68
4.14
5.52
4.74
4.18
5.30
4.52
3.98

N.EAST
330
ppm
106
98
94




NORTH
330 555
ppm
4.39
3.90
3.42
4.19
3.72
3.28
4.64
4.10
3.55

ppm
4.36
3.91
3.43
3.99
3.60
3.19
4.67
4.15
3.62

NORTH
330
ppm
104
96
92




*Season length changes (as simulated with the CERES-Maize model) were affected only by
temperature.
                                     BRAZIL-19

-------
Table 6.        Sensitivity of the SOYGRO model to changes in temperature, precipitation, and
               CO2 levels (330 ppm and 555 ppm of COJ. Simulated yields and season length for
               different regions.
                          SIMULATED GRAIN YIELD (t ha'1)
Prec.
Diff.


0%
0%
0%
+20%
+20%
+20%
-20%
-20%
-20%
Temp.
Diff.


0
+2
+4
0
+2
+4
0
+2
+4
SOUTH
330
ppm
2.91
2.99
2.79
3.08
3.18
3.00
2.63
2.67
2.47
555
ppm
3.92
4.14
3.77
4.26
4.17
3.95
3.68
3.72
3.46
SIMULATED






Temp.
Diff.
(C)

0
+2
+4
SOUTH
330
ppm
143
137
134
555
ppm



C.SOUTH
330
ppm
3.10
2.83
2.35
3.29
2.98
2.46
2.90
2.65
2.16
SEASON
555
ppm
4.60
4.36
3.88
4.71
4.48
4.01
4.43
4.18
3.70
LENGTH
C.SOUTH
330
ppm
125
118
114
555
ppm



N.EAST
330
ppm
3.26
3.06
2.79
3.36
3.18
2.92
3.06
2.84
2.54
(DAYS)*
555
ppm
4.08
3.81
3.45
4.18
3.92
3.60
3.86
3.56
3.19

N.EAST
330
ppm
110
112
116
555
ppm



NORTH
330
ppm
2.17
1.94
1.79
2.21
2.00
1.85
2.10
1.88
1.71

555
ppm
3.08
2.93
2.82
3.10
2.96
2.88
3.05
2.89
2.76

NORTH
330
ppm
91
90
92
555
ppm



 *Season length changes (as simulated with the SOYGRO model) were affected only by
 temperature.
                                      BRAZIL-20

-------
Table 7.        Effects of GCM climate change scenarios on wheat, maize, and soybean yields in
               selected sites.
 Site
BASE
                 Simulated Yield (t ha"1)
                                   Climate Scenario
Climate Scenario Alone               + Physiological CO2 Effects
  GISS    GFDL   UKMO    BASE    GISS    GFDL    UKMO
WHEAT
Pelotas
Passo Fundo
Vacaria
Sao Borja
Ponta Grossa
Londrina
Campinas (*)
Campo Grande (*)
Sete Lagoas (*)

Pelotas
Passo Fundo
Vacaria
Sao Borja
Ponta Grossa
Londrina
Campinas
Campo Grande
Sete Lagoas
Cruz das Almas
Petrolina
Bclcm
Manaus
2.54
1.99
2.66
2.01
2.28
2.02
2.93
2.12
2.93

6.26
8.17
9.20
7.60
7.64
7.36
7.16
6.72
5.41
5.10
4.64
4.00
4.79
1.73
1.21
1.89
0.93
1.50
0.93
1.14
0.59
1.51

5.01
6.83
7.49
5.54
5.88
6.21
6.34
5.34
4.46
3.77
2.59
3.02
3.82
1.52
1.25
1.94
1.21
1.52
1.35
1.57
1.23
1.72

3.93
6.91
7.24
5.24
6.08
5.78
5.93
5.83
4.66
3.94
3.09
3.45
3.92
1.62
0.91
1.69
1.01
1.17
0.95
1.17
0.76
1.10
MAIZE
5.01
5.85
6.93
4.21
5.29
5.76
5.78
4.28
4.67
4.07
2.81
3.64
3.96
2.85
2.32
2.89
2.54
2.70
2.42
3.52
2.79
3.36

7.58
8.93
9.64
8.25
8.07
7.77
7.56
7.08
5.79
5.64
5.25
3.65
5.06
2.14
1.62
2.18
1.33
1.91
1.33
1.72
1.00
2.04

6.06
7.42
7.92
5.99
6.26
6.59
6.72
5.63
4.92
4.49
3.06
2.54
4.10
1.94
1.65
2.22
1.62
1.88
1.78
2.24
1.84
2.24

5.12
7.54
7.68
5.94
6.42
6.10
6.33
6.17
5.12
4.92
3.73
3.94
4.23
2.05
1.32
2.02
1.44
1.57
1.34
1.74
1.23
1.57

6.02
6.64
7.39
5.10
5.72
6.17
6.18
4.56
5.05
4.73
3.36
3.65
4.30
SOYBEAN
Pelotas
Passo Fundo
Vacaria
Sao Borja
Ponta Grossa
Londrina
Campinas
Campo Grande
Sete Lagoas
Cruz das Almas
Petrolina
Belcm
Manaus
2.84
3.03
2.94
2.48
3.28
2.16
3.77
3.43
3.05
3.01
3.52
1.87
1.89
2.20
2.38
3.68
2.36
3.24
1.73
3.44
2.97
2.53
2.57
3.18
2.38
1.49
2.12
2.39
3.52
1.83
3.23
1.60
3.10
2.91
2.62
2.63
2.83
3.17
1.50
3.66
1.71
3.65
1.55
2.76
0.94
3.22
1.76
2.66
2.33
2.51
2.36
1.19
2.95
4.38
3.78
3.52
4.28
4.64
4.80
4.92
4.05
3.74
4.42
3.14
3.01
2.99
3.67
4.64
3.23
4.38
4.12
4.36
4.41
3.62
3.19
3.44
3.16
2.84
3.35
3.76
4.48
2.73
4.37
4.01
4.07
4.29
3.50
3.28
3.99
3.02
2.80
3.30
2.82
4.76
2.36
4.18
3.76
4.20
3.63
3.57
2.93
3.17
3.14
2.55
SOUTH REGION: Pelotas, Passo Fundo, Sao Borja, Vacaria, and Ponta
SOUTH REGION: Londrina, Campinas, Campo Grande, and Sete Lagoas (S.
region for wheat). NORTHEAST REGION: Cruz das Almas and Petrolina.
Manaus and Belem. (*) Irrigated simulation.
                                              Grossa. CENTRAL
                                             L. in the CENTRAL
                                              NORTH REGION:
                                     BRAZIL-21

-------
Table 8.
Effect of the GCM climate change scenarios on simulated wheat.
                               Climate Scenario Alone
                                                          Climate Scenario
                                                     + Physiological CO2 Effects
Region
BIOMASS (t ha'1)
BASE
GISS
GFDL
UKMO
GRAIN YIELD (t ha'1)
BASE
GISS
GFDL
UKMO
SEASON LENGTH (Days)
BASE
GISS
GFDL
UKMO
EVAPOTRANSPIRATION
BASE
GISS
GFDL
UKMO
South

8.82
6.18
6.33
5.50

2.30
1.43
1.49
1.28

124
107
109
104
(mm)
274
247
243
246
CSouth

7.30
3.01
4.51
3.28

2.36
0.89
1.38
0.96

103
87
93
87

264
248
250
252
Central

7.72
4.21
4.74
3.22

2.93
1.51
1.72
1.10

100
85
89
82

303
294
277
305
South

10.30
7.87
7.96
7.27

2.66
1.84
1.86
1.68

124
107
109
104

248
227
222
227
CSouth

9.04
4.55
6.37
4.89

2.91
1.35
1.95
1.43

103
87
93
87

233
219
224
222
Central

8.76
5.63
6.10
4.57

3.36
2.04
2.24
1.57

100
85
89
82

260
253
238
259
                                       BRAZIL-22

-------
Table 9.
Effect of the GCM climate change scenarios on simulated maize.
 Region
        Climate Scenario Alone
   South  CSouth    NEast    North
       Climate Scenario
  + Physiological CO2 Effects
South   CSouth   NEast   North
BIOMASS (t ha'1)
BASE
GISS
GFDL
UKMO
GRAIN YIELD
BASE
GISS
GFDL
UKMO
14.03
12.54
12.06
11.60
(tha-1)
7.78
6.15
5.88
5.46
13.32
11.85
12.03
11.42

6.66
5.59
5.55
5.12
10.64
7.62
7.86
8.28

4.87
3.16
3.52
3.44
10.82
8.78
8.52
9.72

4.39
3.42
3.68
3.80
14.73
13.27
12.90
12.63

8.50
6.73
6.54
6.18
13.72
12.32
12.46
11.94

7.05
5.96
5.93
5.49
11.50
8.75
9.34
9.34

5.45
3.78
4.32
4.04
10.70
8.60
9.33
10.17

4.36
3.32
4.08
3.97
SEASON LENGTH (days)
BASE
GISS
GFDL
UKMO
134
105
109
101
EVAPOTRANSPIRATION
BASE
GISS
GFDL
UKMO
448
407
394
405
117
100
101
97
(mm)
414
378
378
400
106
88
94
90

316
258
268
381
104
92
94
92

305
306
294
297
134
107
109
101

377
347
339
353
117
100
101
97

347
318
318
338
106
88
94
90

268
264
233
244
104
92
94
92

254
256
244
249
                                     BRAZIL-23

-------
Table 10.
Effect of the GCM climate change scenarios on simulated soybean.
 Region
           Climate Scenario Alone
      South   CSouth   NEast   North
       Climate Scenario
   + Physiological CO2 Effects
South   CSouth   NEast  North
BIOMASS (t ha1)
BASE
GISS
GFDL
UKMO
YIELD (t ha'1)
BASE
GISS
GFDL
UKMO

6.53
5.89
5.44
5.08

2.92
2.83
2.70
2.44

5.69
5.04
4.74
3.92

3.10
2.71
2.55
2.02

6.06
5.18
5.33
4.28

3.20
2.60
2.75
2.01

3.86
3.37
3.33
3.08

2.17
1.94
1.90
1.78

7.77
8.19
7.72
7.41

3.63
3.84
3.74
3.48

7.02
7.80
7.52
7.05

3.70
4.13
3.95
3.63

6.17
6.76
6.91
5.81

3.56
3.32
3.47
2.80

4.82
5.21
5.12
4.95

2.76
3.00
2.91
2.85
SEASON LENGTH (days)
BASE
GISS
GFDL
UKMO
EVAPOTRANSP.
BASE
GISS
GFDL
UKMO
143
133
135
133
(mm)
462
495
453
496
123
119
120
114

432
453
442
462
109
110
109
110

381
433
400
407
91
92
90
92

302
345
324
336
135
133
135
133

454
494
454
502
121
121
122
122

446
462
446
500
110
110
109
110

334
432
396
408
92
92
90
92

324
330
310
322
                                       BRAZIL-24

-------
 Table 11.
Aggregated yield changes under GCM climate change scenarios. Results include the
physiological effects of CO2 on yield.
                                                                   Simulated Yield Change (%)
Production
Region
SOUTH
CENTRAL-SOUTH
CENTRAL
NATIONAL
SOUTH
CENTRAL-SOUTH
CENTRAL
NORTHEAST
NORTH
NATIONAL
SOUTH
CENTRAL-SOUTH
CENTRAL
NORTHEAST
NORTH
NATIONAL
Source of Production data:
Crop
WHEAT
WHEAT
WHEAT
WHEAT
MAIZE
MAIZE
MAIZE
MAIZE
MAIZE
MAIZE
SOYBEAN
SOYBEAN
SOYBEAN
SOYBEAN
SOYBEAN
SOYBEAN
IBGE and Bank of Brazil
(t x 1000)
1,573
2,028
24
3,625
6,695
11,131
2,495
1,126
330
21,778
6,408
6,943
2,135
96
0
15,582

GISS
-21
-43
-30
-33
-13
-10
**
-22
-24
-11
30
32
**
9
38
30

GFDL
-19
-17
-24
-18
-16
-11
**
-11
-7
-11
25
28
**
12
34
23

UKMO
-27
-39
-46
-34
-20
-18
**
-17
-10
-16
20
22
**
-6
31
18

(**) not simulated
                                      BRAZIL-25

-------
Table 12.
Adaptation strategies.
(A)     Effect of changes in planting date, irrigation, and nitrogen stress on simulated maize and
        soybean yields. The climate change scenario simulations include the physiological effects of
        CO2 on yield.
                                                                                       Yield
                                                                                     Change
                                                      Simulated Yield (t ha'1)      from Base
Site
Passo Fundo





Petrolina



Petrolina

Crop Strategy
Maize Oct 15, rainfed(*)
Sep 15, rainfed
Nov 15, rainfed
Dec 15, rainfed
Jan 15, rainfed
Oct 15, irrig.
Maize Rainfed, N stress(*)
Rainfed, N
Irrig., N stress
Irrig., N
Soybean Rainfed
Irrig.
BASE
8.17




8.48
4.64
6.72
5.37
7.05
3.39
3.73
UKMO
6.69
6.54
5.75
5.72
6.62
6.79
3.98
5.34
5.10
6.37
2.64
4.19
(%)
-18
-20
-30
-30
-19
-17
-14
15
10
37
-22
24
(*) current practice
N stress: 80 Kg N/Ha
N: N for maximum yield (N balance off in the model)
(B)     Sensitivity of the CERES-Maize model to changes^in the P5 coefficient of the cultivar PIO
        3230. Scenario simulations include the physiological effects of CO2 on yield.
Site Crop
Passo Fundo Maize




Strategy
Change in P5
P5=995*
P5=795
P5=1195
P5=1395
Simulated Yield (T/Ha)
Base UKMO

8.18 6.86
5.21
8.51
10.09
Yield
Change from
Base (%)

-16
-36
4
23
 (*) calibrated coefficient for the cultivar PIO 3230 in Passo Fundo.
                                        BRAZIL-26

-------
Appendix A.   Cultivars and crop management data used to run the crop models.
Crop
WHEAT












MAIZE

















SOYBEAN






Region
SOUTH
SOUTH

SOUTH
SOUTH
SOUTH

C.SOUTH
C.SOUTH
C.SOUTH

CENTRAL

SOUTH
SOUTH

SOUTH
SOUTH
SOUTH

C.SOUTH
C.SOUTH
C.SOUTH

C.SOUTH

N.EAST

N.EAST
NORTH
NORTH
SOUTH
SOUTH

SOUTH
SOUTH
SOUTH

Site
PELOTAS
PASSO
FUNDO
SAO BORJA
VACARIA
PONTA
GROSSA
LONDRINA
CAMPINAS
CAMPO
GRANDE
SETE
LAGOAS
PELOTAS
PASSO
FUNDO
SAO BORJA
VACARIA
PONTA
GROSSA
LONDRINA
CAMPINAS
CAMPO
GRANDE
SETE
LAGOAS
CRUZ
D.ALMAS
PETROLINA
MANAUS
BELEM
PELOTAS
PASSO
FUNDO
SAO BORJA
VACARIA
PONTA
GROSSA
Cultivar
BR14
BR14

BR14
BR14
BR14

BR14
ANZA*
BR 14/ANZA*

ANZA*

PIO-3230
PIO-3230

PIO-3230
PIO-3230
PIO-3230

PIO-3230
PIO-3230
PIO-3230

PIO-3230

SUWAN-1

SUWAN-1
SUWAN-1
SUWAN-1
DAVIS
DAVIS

DAVIS
DAVIS
DAVIS

Plant
Pop.
(pl/m2
330
330

330
330
330

330
350
350

350

5
5

5
5
5

5
5
5

5

5

5
5
5
40
40

40
40
40

Planting
Date
JUN. 15
JUN. 15

MAY. 31
JUL. 15
JUN. 15

APR. 15
APR. 30*
APR.30*

APR. 30*

OCT. 15
OCT. 15

OCT. 15
NOV. 15
OCT. 15

OCT. 15
OCT. 15
OCT. 30

OCT. 30

OCT. 15

OCT. 15
NOV. 15
NOV. 15
NOV. 15
OCT. 15

NOV. 15
NOV. 15
NOV. 15

                                     BRAZIL-27

-------
C.SOUTH
C.SOUTH
C.SOUTH

C.SOUTH

N.EAST

N.EAST
NORTH
NORTH
LONDRINA
CAMPINAS
CAMPO
GRANDE
SETE
LAGOAS
CRUZ
D.ALMAS
PETROLINA
MANAUS
BELEM
DAVIS
DAVIS
DAVIS

DAVIS

VICOJA

VICOJA
JUPITER
JUPITER
40
40
40

40

40

40
40
40
NOV. 15
NOV. 15
NOV. 15

OCT. 15

NOV. 30

NOV. 30
NOV. 30
NOV. 30
"Irrigated. Others: rainfed.
Local cultivars: BR 14: PIV=1.9; P1D=1.5; P5=6.0; Gl=3.2; G2=0.6; G3=3.9.
           PIO-3230: PI =220; P2=0.85; P5=995; G2=720; G3=5200.
                                     BRAZIL-28

-------
        I.BELCM
         . MANAUS
        3. PCTROUNA
         , CRUZ OAS ALMAS
        s. sere  LAOOAS
        . CAMPO ORANDE
          CAMPMAS
         . LONDKINA
         . PONTA WtOSS
       10. PAS50 PUNDO
       II. VACARIA
          SAO oruA
       13. PCUOTAS
                                      EquatorMl and
Figure la.     M^> of Brazil; climatic regions and sites selected for the study.

-------
Figure Ib.      Wheat agroecological regions; C=Central, CS=Central-South, S=South.

-------
              MAIZE - SOYBEAN

Figure Ic.      Maize and soybean agroecological regions; N=North, NE=Northeast,
               CS=CentraI-South, S=South.

-------
             6000
             5000 -
             4000
   Observed
     Grain
     Yield  3000
     (kg/ha)

             2000 -
1000
   1000
              6500
     Observed '
       Grain
       Yield  800
      (kg/ha)
              7000
                 7000
              5000
              4000
    Observed
      Grain   3000
      Yield
     (kg/ha)

              2000
                           2000      3000      4000       5000
                             Simulated Grain Yield (kg/ha)
                                                                    6000
                       CERES-Maize
                                   -K
                             7600         8000         8600

                               Simulated Grain Yield (kg/ha)
                                                                    9000
                      SOYGRO
              1000^
                 1000
                2000        3000         4000
               Simulated Grain Yield (kg/ha)
                                                     5000
Figure 2.   Observed and simulated grain yields (kg ha'1) using the CERES-Wheat (Siqueira 1991),
          CERES-Maize (adapted from Matzenauer et al. 1988), and SOYGRO (Siqueira and
          Berg 1991) at Rio Grande do Sul.

-------
    WHEAT
    Relative Yield Changes (%)
          SOUTH
                     CENTRAL-SOUTH
                    330 ppm CO2
     MAIZE
     Relative Yield Changes (%)
        SOUTH
                  C.8OUTH    NORTHEAST    NORTH
                    330 ppm CO2
     SOYBEAN
     Relative Yield Changes (%)
  -40
        SOUTH     C.SOUTH   NORTHEAST
                    330 ppm CO2
                                        NORTH
WHEAT
Relative  Yield Changes (%)
                                                             SOUTH
                CENTRAL-SOUTH
                555 ppm CO2
                                                                                         CENTRAL
 MAIZE
 Relative Yield Changes (%)
              C.SOUTH    NORTHEAST
                555 ppm CO2
                                                                                           NORTH
 SOYBEAN
 Relative Yield Changes (%)
    SOUTH     C.SOUTH   NORTHEAST
                555 ppm CO2
                                                                                           NORTH '
                   CDGFW.  BUKMO
Figure 3.    Regional yield changes for wheat, maize, and soybean under GCM climate change
             scenarios.

-------
Temperature Differences (Celsius)
                                                              Precipitation Ratio (GISS/Base)
             1010   (YEAR)    2030

          GISS Transient Scenarios
                                         20M
                                                   o
                                                   mo
                                                                          2010   (YEAR)   2030

                                                                       GISS Transient Scenarios
          WHEAT


          Grain Yield (kg/ha)    Season Length (days)
                                              B
             -- *MIN YIELD -~ (CATCN LCNOTN
                                                      MAIZE


                                                 	Grain Yield (kg/ha)   Season Length (days)
                                                 ioooo,e			1130
           rioei          i4soi
        GISS Transient Scenarios
                                                  eooo
                                      ,100        5000
                                      2060      yeAB  w
                                                                          * *EA*ON LIN*TH
                                                                           2010          2030          2050

                                                                       _1_J
-------
             Relative Yield (%)
             ~BRl4	
                                       P1V.2.6JBfWoH
                                     P1V-2.6.IBR
                                  P1V-2.6
                                P1V-19
        Season Length (days)
              8           10
           P-5 Genetic Coefficient
                                 P1V-2.6.IRR.No(f
                               P1V-J.6.IRR
                             P1V-2.6
                           P1V-1.9
                   10
      P-5 Genetic Coefficient
            Grain Yield (t/ha)
                                     P1V-2.5JRR.Noff
                                      2JJSH
                                                                 Grain Yield (t/ha)
      <                 10
          P-5 Genetic Coefficient
<           a     10
    P-5 Genetic Coefficient
                                                                                    P1V-1.8
Figure 5.    Adaptation studies for two cultivars of wheat (BR 14 and BR 23) in Passo Fundo to
             the UKMO climate change scenario. The results include the physiological CO2 effects.
             Changes in the genetic coefficients and crop management. IRR: Irrigation; Noff: full
             nitrogen fertilization.

-------
IMPACTS OF GLOBAL CLIMATE CHANGE ON MAIZE PRODUCTION IN
                           ARGENTINA
                       O. E. Sala and J.M. Paruelo

                       Universidad de Buenos Aires
                         Buenos Aires, Argentina
                            ARGENTINA-1

-------
TABLE OF CONTENTS

SUMMARY
INTRODUCTION
       Agricultural Systems in Argentina
       Objectives
METHODS
       Baseline Climate Data and Climate Change Scenarios
       Crop Model Calibration
       Soils and Management Variables
RESULTS AND DISCUSSION
       The Effects of Climate Change
       Sensitivity Analysis
       Adaptation Strategies
       Simulations with Hypothetical New Hybrids
IMPLICATIONS  OF THE RESULTS
REFERENCES
                                   ARGENTINA-2

-------
SUMMARY

        Climate change is predicted by three GCMs to increase temperature and precipitation in the Rolling
Pampa. These  climatic changes may decrease maize yields, if all  other variables remain constant. The
simulation results suggest that decreases in maize yields may be offset by (1) changing the sowing date of
current maize varieties, or (2) using hybrids more adapted to the climate change conditions. However, at the
warmer boundaries of the Pampa, it is unlikely that decreases in maize yields could be completely offset by
changes in management practices. Under climate change conditions, maize production may also be constrained
in the northern edge of its current distribution due to extreme high temperatures. At the same time, maize
production could expand  into areas to the south that are currently limited by frost.
INTRODUCTION

Agricultural Systems in Argentina

        The Pampa region (Figure 1) covers approximately 34 million hectares of agricultural land (Hall et
al. 1991).  One-third of the area is used to grow grain crops; the rest, which is comprised of meadows and
natural grasslands, is used to fatten steers  and conduct cow/calf operations. The Pampa region  has been
subdivided according to ecological criteria (Le6n et al. 1991). Of these subregions, the Rolling Pampa is the
most productive. Here, 50 - 75% of the region is devoted to grain crops.
        The Pampa region has a temperate, humid climate and lacks a fully developed dry season. There is
a gradient of rainfall from 1,000 mm in the NE of the region to 600 mm in the SW. Mean annual temperatures
range from 17 C in the northern part of the region to 14 C in the SE. Similarly, there is a gradient of the
frost-free period which ranges from 180 to 260 days.
        Soils in the region are mainly Mollisols developed on a deep mass of Pampean loess (Frenguelli 1925).
This is a quaternary sediment which was originally transported by wind and later redistributed by water. There
is a clear gradient of soil  texture, from coarse soils in the West to fine-textured soils in the East.
        Maize is one of the most important crops of the Pampa, occupying 2.3 millon hectares (Hall et al.
1991), with production concentrated in the Rolling Pampa in the NE portion of the region. Maize is alternated
with wheat and soybean crops. Sunflower and grain sorghum are also important (1.5 and 1.1 millon hectares,
respectively, from 1983 to 1985). The average yield of maize in the region from 1980 to 1986 was 4.2 t ha'1.

Objectives

        The objective of  this work is to  assess the possible effect of climate change on maize yields in the
Pampa of Argentina. Climate change driven by an increase of greenhouse gases, mainly CO2, will affect crops
in two ways. One, an elevated CO2 level may enhance net photosynthesis and change stomatal  conductance
(making the water use per unit leaf area more efficient). These responses are called the "direct CO2 effects".
Two, the expected  temperature increases and changes in precipitation patterns would alter the growth and
development of the crop. In this study, we explore the effects of climate change alone and climate change with
the direct effects of increasing CO2 on maize. We use climate change scenarios developed from three General
Circulation Models (GCMs) and apply them to a dynamic process crop simulation model to determine the
potential yield changes under future climate conditions.

METHODS
                                         ARGENTINA-3

-------
Baseline Climate Data and Climate Change Scenarios

        The climatic data used for the modeling experiments are from the town of Pergamino in the Rolling
Pampa  (lat. 33 S, long. 60 W) (Figure  1). We used daily  precipitation and  maximum and minimum
temperature data for the period 1960-1984. Daily radiation was estimated using sunshine hours data.
        Climate change scenarios were generated from three  GCMs: GISS  (Goddard Institute for Space
Studies, Hansen et aL 1983); GFDL (Geophysical Fluid Dynamics Laboratory Model, Manabe and Wetherald
1987); and UKMO (United  Kingdom Meteorological Office Model, Wilson and Mitchell 1987). Changes in
temperature, precipitation, and solar radiation projected by the GCMs were applied to the observed (baseline)
climate  to create the climate change scenarios for Pergamino.
        The annual and seasonal changes in  temperature and  precipitation projected by  the GCMs at
Pergamino are shown  in Table 1. The three scenarios predict  significant increases in annual temperatures
ranging  from 4.5 C to 5.2 C; the temperature increases are larger during the summer months. The projected
precipitation changes vary. The UKMO and  GISS scenarios predict precipitation increases (29% and 10%,
respectively), while the GFDL scenario predicts an annual decrease in precipitation (-5%).

Crop Model Calibration

        The CERES-Maize crop model (Ritchie  et al. 1989; Jones and Kiniry, 1986) was chosen for the
simulations. This model has been widely validated in different agroecological conditions (Hodges et al. 1986).
We further calibrated and validated the CERES-Maize model for the conditions of the Pampa to determine
its suitability as a simulating tool. The calibration was performed for two single flint type hybrids, DAF11 and
DAF12, from Dekalb  Argentina. Table 2 presents the derived genetic coefficients that define the maize
varieties. A summary of the simulated and observed validation  results is presented in Table 3.

Soils and Management Variables

        The main soil type  of the Rolling Pampa is the Pergamino series. It is  a typical argiudol with no
serious constraints for agriculture. It is comparable to soils of the mid-western U.S. or the Ukraine, but it does
not freeze in winter; thus tillage is feasible year-round (Hall et al 1991) (Appendix 1).
        Maize in the Rolling Pampa is mostly rainfed, and long-cycle hybrids account for 80% of the cultivated
area. There are some medium-cycle hybrids grown. Maize is sown in early September and harvested in early
February the following year. For this simulation, the sowing date of long-cycle hybrids was the 277th day of
the Julian calendar and for medium-cycle hybrids, the 267th day. The duration of the fallow is directly
correlated with maize yields in the region. A fallow period longer than 120 days is reported for 37% of the
crops.

RESULTS AND DISCUSSION

The Effects of Climate Change

        The changes in temperature and precipitation predicted for the doubled CO2 scenarios produced
decreases in maize yields (Table  4).  Under the GISS scenario, yields simulated with nitrogen stress were
reduced about 19% in comparison with base yields. Under the GFDL and UKMO scenarios, the yield
reductions were larger (about 36% in both cases). These simulated yield reductions are mainly a result of a
shorter  growing period. The higher temperatures  under the climate change scenarios trigger the onset of
maturity stages earlier than under the present climate.

                                         ARGENTINA-4

-------
        The direct effect of the increased atmospheric CO2 on the crop resulted in very small yield increases.
These did not compensate for the yield decreases simulated under the GCM scenarios alone.
        Yearly variability of the yields decreased in the doubled-CO2 runs. Since precipitation variability was
not modified in this modeling exercise we suggest that the observed reduction in yield variability is a result
of the shortened growing season, with the crop being constrained to a portion of the year where precipitation
variability was lower. Results simulated without nitrogen stress are also shown in Table 4.

Sensitivity Analysis

        The  purpose of the  sensitivity analysis  is to assess  the  maize  model  responses to  changes in
temperature, precipitation, and CO2. The experiment consisted of running the model for three temperature
conditions (control, +2C, and +4C); three precipitation conditions (control, +20%, -20%); and two CO2
conditions (330  ppm and  555 ppm).  The  response variables analyzed are yield, season length, and
evapotranspiration (Table 5).
        Yields decrease as a result of increases in temperature, due  to a decrease in the season length which
particularly affects the critical grain-filling period. The simulated increases in precipitation do not produce
increases in simulated yields as we had expected. A possible explanation may be that the modeled crop was
under more nitrogen stress under the higher precipitation scenario, due to nutrient leaching. Alternatively,
with seasonal rainfall, percentage increases of precipitation may still be quite small. Further analysis of the
nitrogen balance in the CERES-Maize model and the simulation conditions are needed.
        The direct effects of 555 ppm CO2 result in a small increase  in simulated yield, but this yield increase
did not compensate for the negative effects of a +2 C temperature increase. The direct effects of CO2 on yield
are larger under the lower precipitation conditions because the beneficial  effect of CO2 on simulated water
use is more apparent  under low precipitation conditions.

Adaptation Strategies

        The major effect of climate change predicted by this modeling exercise is that temperature increases
result  in yield reductions due to  a shortening of the growing period. We  suggest two possible adaptive
strategies to climate change: (1) a shift in the sowing date; and (2)  a change in the hybrid.
        Changes in the sowing date had large effects on simulated maize yields (Figure 3) and on the length
of the growing period (Figure 2). Under the baseline climate, the optimum simulated sowing date is close to
the date most frequently used in the  region. However, under the GISS climate change scenario, the curve
relating sowing dates and yields shows a bimodal response, with maxima occurring with very early or very late
sowing dates. Both dates avoid the high-stress months of midsummer  (January and February) and consequently
decrease the water stress.
        Therefore, one possible adaptive strategy to climate change conditions would be to sow maize very
early so that the growing period occurs mostly during the cooler part of the year. Planting two months earlier
than the present sowing date can fully compensate for the yield decreases under climate change conditions.
An alternative strategy is to sow very  late and avoid the hot months of midsummer.
        These strategies have some risk associated with them. The yield variability increases markedly as  the
sowing date moves away from October/November (Figure 2), because of extremely low temperatures in some
years.  Although these strategies simulate an optimum yield under climate change conditions, they may  not
represent a practical optimum because of the biological uncertainties and the economic risks associated with
them.  Each of these alternatives implies major changes in the agricultural system of the region.

Simulations with Hypothetical New Hybrids

                                          ARGENTINA-5

-------
        A possible strategy for adaptation to climate change could be to replace the currently available hybrid
with a hypothetical new hybrid better adapted to climate change conditions.  To analyze this alternative,
hypothetical new hybrids were created under the GISS scenario by increasing the genetic coefficients, PI} P2,
and P5, from 10% to 50% (Figure 4). Pj represents the time period during which the plant is not responsive
to changes in the photoperiod. P2 defines the photoperiod sensitivity of the cultivar. P5 is the number of degree
days above a base of 8C from silking to physiological maturity.
        The figure shows that an increase in the genetic coefficients of the potential new hybrids of between
10% and 20% is enough to restore the GISS climate change yields to their baseline level. This simulation
exercise suggests that hypothetical new hybrids may be able to take advantage of a prolonged growing season
and higher precipitation amounts. The extent to which this could be possible in practice needs to be further
investigated with maize breeders.

IMPLICATIONS OF THE RESULTS

        This study suggests that projected climate changes resulting from an increase in greenhouse gases may
result in decreases in maize yields in the Rolling Pampa.  The yield decreases may be compensated for by
changing the planting date of the hybrids presently used. But in the warmer boundaries of the rolling Pampa,
it is unlikely that maize yields could be fully restored to their previous level. Nevertheless, maize production
could expand into areas that are currently limited by the length of the frost-free period.
        Using the same tools that we used in this exercise, we plan to conduct further model simulations for
different locations in order to estimate the geographical location of the potential new maize-producing region
in Argentina under climate change conditions. A qualitative analysis suggests that a large portion of the maize-
producing area will be located on soils of poorer quality than present soils. Therefore, the average soil fertility
of the maize area may decrease, possibly resulting in decreased yields, unless additional fertilizer is added to
the soil. These hypotheses can be tested further by using these same modeling tools  on a larger number of
sites.
                                          ARGENTINA-6

-------
REFERENCES

Frenguelli, J. 1925. Loess y limos pampeanos. Anales de la Sociedad Argentina de Estudios Geogrdficos GAEA
        1:7-91.

Hall, A.J., CM. Rebella, CM. Ghersa, and J.Ph. Culot. 1993. Field crops systems of the pampas. In C.J.
        Pearson  (ed)  Ecosystems  of the World, Field Crop Ecosystems  of the World. Elsevier  Scientific
        Publishing Company, Amsterdam, The Netherlands.

Hansen, J., G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983. Efficient three-
        dimensional global models for  climate studies: Models I and II. Monthly Weather Review 3:609-662.

Hodges, T., D. Botner, C Sakamoto, and J. Hays-Haug.  1986. Using the  CERES Maize model to estimate
        production for the US corn belt. Agricultural and Forest Meteorology 40:293-303.

Jones, C.A. and J.R.  1986. CERES-Maize: A Simulation  Model of Maize Growth and Development. College
        Station. Texas A&M Press.

Le6n, R.J.C. 1991. Natural Grasslands of South America. In R.T.Coupland (ed) Ecosystems of the World,
        Volume 8, Natural Grasslands. Elsevier Scientific Publishing Company, Amsterdam, The Netherlands.

Manabe, S., and  R.  Wetherald.  1987. Large-scale changes  of soil  wetness  induced by an increase  in
        atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.

Ritchie, J., U. Singh,  D. Godwin, and L. Hunt. 1989. A user's guide to  Ceres Maize- v2.10. International
        Fertilizer Development Center. pp86.

Wilson, C.A., and J.F.B. Mitchell. 1987. A doubled CO2 Climate Sensitivity Experiment with a Global Model
        Including a Simple Ocean. Journal of Geophysical Research, 92:13315-13343.
                                         ARGENTINA-7

-------
Table 1.       Average seasonal changes of temperature and precipitation for the GISS, GFDL, and UKMO
              GCM climate change scenarios in Pergamino.
      Scenario/variable
SPRING   SUMMER
          FALL   WINTER   ANNUAL
      GISS
        Temperature C
        Precip. (%)
      GFDL
        Temperature C
        Precip. (%)
      UKMO
        Temperature C
        Precip. (%)
    3.8
     -3

    4.0
      0

    4.8
     51
5.4
30

5.5
20

5.9
14
5.5
 10

4.3
-34

5.1
 40
4.5
 15

4.4
-25

5.0
 33
4.8
 10

4.5
 -5

5.2
 29
                                      ARGENTINA-8

-------
Table 2.
Genetic coefficients derived for single flint type hybrids.
              Genetic coefficient
                                       DAF11 hybrid     DAF12 hybrid
              PI  (Juvenile)
              P2  (Photoperiodism)
              P5  (Grain filling duration)
              G2   (Kernel number)
              G3   (Kernel weight)
                                                260
                                                 1.0
                                                700
                                                625
                                                 9.6
260
 1.0
700
710
 8.6
                                          ARGENTINA-9

-------
Table 3.
Validation of the CERES-Maize model for Pergamino.
 Crop variable
                             DAF11 hybrid
                              Sim.          Obs.
DAF12 hybrid
 Sim.          Obs.
Emergence
Anthesis
Beginning grain filling
End grain filling
Yield (t ha'1)
Kernel weight (g)
Grain/ear
Biomass (t ha'1)
3-11
10-1
20-1
20-2
10.12
0.275
383
20.57
3-11
13-1
-
-
9.96
0.281
381
22.68
3-11
10-1
20-1
20-2
10.15
0.246
429
20.58
3-11
9-1
~
~
10.25
0.251
436
20.25
                                     ARGENTINA-10

-------
Table 4.         Effects  of climate change on  simulated rainfed maize yield,  season length, season
                precipitation, and evapotranspiration with and without nitrogen stress.
                                                                         Climate Scenario With Physiological
                                         Climate Scenario Alone
CO2 Effects
Management
Rainfed
Nitrogen
Stress









Rainfed
Fertilized



Simulated
Variable
Yield
(t ha'1)

SD
SXength
(d)
SD
S. PP
(mm)
SD
ET (mm)
SD
Yield
(t ha -1)
SD
ET (mm)
SD
BASE
3.69


0.85
126

8
611

142
487
20
9.42

1.44
494
22
GISS
3.00


0.70
102

4
498

102
460
21
7.89

0.95
471
21
GFDL
2.52


0.61
101

4
530

134
458
25
7.39

1.13
458
25
UKMO
2.40


0.48
99

3
594

115
448
22
8.20

0.95
475
22
BASE
3.77


0.82
126

8
611

142
398
16
9.95

1.30
419
19
GISS
3.04


0.51
102

4
498

102
387
24
8.52

0.85
414
23
GFDL
2.52


0.53
101

4
530

134
408
18
8.29

0.91
408
19
UKMO
2.54


0.42
99

3
594

115
367
27
8.88

0.85
417
22
                                          ARGENTINA-11

-------
Table 5.
Sensitivity analysis of CERES-Maize to climate and CO2 changes.
                                                     Simulated variable*
Changes
Precip.
(%)
0%


20%


-20%


Temp.
(C)
0
2
4
0
2
4
0
2
4
Yield
(t ha'1)
3.69
3.35
3.31
3.52
3.19
3.18
3.76
3.49
3.35
330 ppm CO2
Season L.
(days)
126
113
104
126
113
104
126
113
104
ET
(mm)
486
457
437
487
456
437
482
453
433
Yield
(t ha'1)
3.77
3.38
3.35
3.49
3.18
3.12
3.92
3.59
3.52
555 ppm CO2
Season L.
(days)
126
113
104
126
113
104
126
113
104
ET
(mm)
397
376
364
395
373
360
399
378
366
* Simulations with nitrogen stress.
                                        ARGENTINA-12

-------
                      ARGENTINA
Figure 1.     Map of Argentina and location of Pergamino

-------
        150
        140

   *">
   cn


   <9   130
    tn   120
    o
    (A
    CO
    cu
    en
        110
        100
         90
                                                        Baseline

                                                        CLIMATE CHANGE
             ISAug       ISSep       150ct       IBNov


                                      Sowing  date
IBDec
15Jan
Figure 2.      Effect of sowing date on maize yield for the baseline and the climate change conditions.
             Vertical bars represent standard deviations.

-------
    tti
    ID
    H
                     Baseline
                     CLIMATE CHANGE
             15Aug
IBSep
150ct      IBNov

 Sowing date
                                                          ISDec
Figure 3.     Effects of sowing date on the season length of maize for the baseline and climate change
            conditions.

-------
       6 -
cd

EH
       5 -
   ID
  :H    4 H
                 Climate  Change  condition
                                  DAF12  (baseline  condition)
               Control   Hyb  1    Hyb 2    Hyb 3   Hyb 4    Hyb 5
Figure 4.     Simulated season length of hybrids with genetic coefficients PI, P2, and P5 ranging from
           10% to +50% (Hybl to Hyb5) higher than the control hybrid.

-------
       IMPACT OF CLIMATE CHANGE ON BARLEY IN URUGUAY:
YIELD CHANGES AND ANALYSIS OF NITROGEN MANAGEMENT SYSTEMS
                          Walter E. Baethgen

                International Fertilizer Development Center (IFDC)
                               USA
                             URUGUAY-1

-------
TABLE OF CONTENTS

SUMMARY
INTRODUCTION
       Background
       Objectives
       Representative climate and soils
       Managements practices and nitrogen available for the crop
METHODS
       Baseline climate data
       Climate change scenarios for the region
       Crop model and simulation strategies
       Validation and calibration of the crop model
RESULTS AND DISCUSSION
       Effects of 2xCO2 GCM climate change on barley production
       Sensitivity analysis
       Transient analysis
       Adaptation strategies to climate change
CONCLUSIONS
REFERENCES
                                      URUGUAY-2

-------
SUMMARY

        This study utilized global climate models (GCMs) and dynamic crop growth models to estimate the
potential agricultural effects of climate change on barley production and nitrogen management systems in
Uruguay. The barley crop simulation model was  calibrated and validated in great detail  in the study,
considering seven parameters of growth and development of the crop. Since barley management practices vary
widely in Uruguay, several crop management scenarios were considered combining three planting dates and
two levels of available nitrogen. Projected climate change caused simulated barley yields to decrease in all
strategies considered. The decrease in modeled grain yields were caused primarily by temperature increases
which shorten the duration of the crop life cycle, particularly the grain-filling period. These decreases were
partially counteracted by the beneficial physiological CO2 effects on crop growth and water use as simulated
in this study. The negative effects of GCM climate change were worse when no nitrogen fertilizer was applied,
and/or when planting was delayed. The variability of grain yields was larger under the GCM scenarios and
when planting dates were delayed, but it was not affected by nitrogen fertilization. A possible  adaptation of
barley management systems in Uruguay to climate change conditions is an optimization of the soil nitrogen
available for the crop, but even considering this adaptation strategy, significant production losses were still
associated with the climatic conditions projected by the UKMO scenario.
INTRODUCTION

Background

        The climatic characteristics of Uruguay (Table 1) allow the production of subtropical and temperate
species. The most important crops in Uruguay are wheat, barley, rice, maize, sorghum and sunflower. The
country is self-sufficient in food, and 30-40% of the country's total exports are textiles, beef, hides and cereals.
Most of the wheat, maize, sorghum and sunflower are consumed in the country; barley and rice are primarily
export crops. Due to the economic importance of barley as an export, it was selected for this study.
        The enhanced greenhouse effect of increased atmospheric concentration of CO2 and other trace gases
could lead to higher global surface temperatures and changed hydrological cycles (IPCC 1990). Global climate
change may affect crop production in some areas of the world, as previous impact studies have shown (Parry
et al. 1988). A change in climatic conditions could affect crop production in Uruguay.
        Barley  crops in  Uruguay are sometimes  damaged by excess soil water that causes  leaching and
denitrification, with consequent increased requirements for nitrogen fertilization, lower economic returns, and
greater potential for nitrate contamination of groundwater. A possible change in climatic conditions may have
an important impact on  barley production and  on the environmental consequences of the  use of nitrogen
fertilizers.

Objectives

        This study had two main objectives.  One goal  was to study the potential impact of global climate
change on barley production in Uruguay under a wide range of management practices, in particular available
nitrogen levels and planting dates. Another goal was to evaluate soil and crop management practices (nitrogen
fertilizer use, cultivar characteristics,  and planting  dates) that may be better adapted to the possible climate
change conditions.

Representative Climate and Soils

                                           URUGUAY-3

-------
         The crop-growing area of Uruguay is small (about 700,000 hectares) and the major agricultural regions
 are located in the west and southwest of the country; the barley production is concentrated in those regions.
 Given the relatively small size of the barley growing area in Uruguay, the climate variables recorded in the
 different meteorological stations within the area are very similar; therefore, one site was selected for the study:
 "La Estanzuela" Experimental Station (34 27' S, 57 46' W) (Figure 1).
         The soils are heterogeneous and important variability exists with respect to soil water-holding capacity,
 the ability to supply nitrogen through mineralization, and the ease with which the soils can be tilled. The
 dominant  soils in the area  are Mollisols and Vertisols. Typical ranges of the most important chemical
 characteristics of unfertilized soils of the area are: pH 5.5-6.5; organic mater 25-60 g/kg; 2.0-4.0 mg/kg available
 phosphorus (Bray 1); 0.25-0.60 cmol(+)/kg of potassium. Most Mollisols in the area have more than 25% of
 clay in the superficial  horizon and  usually a heavy textured B  horizon (more than 40% clay). The Vertisols
 are less  differentiated and usually present  more than 30% of clay  through all the profile. Although soil
 variability in the region is important, for this simulation study we selected one representative soil profile that
 accurately represents  the  textures  and water capacities of the  main agricultural soils  in the region. The
 available nitrogen content in the soil was varied to create different  nitrogen conditions  that represent the
 agricultural soils and management  practices in the area.

 Management Practices and Nitrogen Available for the Crop

        Typical farms in the area  produce crops and raise livestock. Consequently, the grain crops and
 pastures are rotated. For example, three years of crops (such  as wheat, barley, sunflower and sorghum)
 alternate with three years of pasture (typically a mixture of white clover, red clover, birdsfoot trefoil and tall
 fescue). As a result of  this system, soils have a variable ability to supply nitrogen to the crops, depending on:
 (a) time since the last pasture was plowed, (b) soil tillage, and (c) soil  type. Soil tillage is especially important
 for barley production because the planting time of the recommended cultivars coincides with the time of the
 year when soil tillage is the most difficult (Figure 2). Thus, years with low rainfall during summer and fall
 usually result in good soil tillage, high nitrogen levels, and high yields. Years with heavy rainfall  during soil
 preparation often result in inadequate seedbeds, low plant populations, poor natural soil nitrogen supply, and
 low grain yields.
        Winter crops in the west and southwest can be planted as early as April (cultivars with photoperiod
 sensitivity) and as late as August. With the exception of a few newly developed cultivars, barley cannot be
 planted earlier due to  lodging problems. This simulation study considered three possible planting dates and
 two levels of available nitrogen to construct simulations that represent the wide variety of conditions used for
 barley production  in Uruguay.

 METHODS

 Baseline  Climate Data

        The baseline climate  data were obtained from Mr. Ricardo Romero, Agrometeorologist of the La
 Estanzuela Experimental Station of INIA (National Agricultural Research Institute of Uruguay). The period
 included in the dataset was 1 January, 1966 to 31 December 1989. The solar radiation values were obtained
 from the sunshine hours based on the total possible hours and the observed hours  using the WGEN program
 (Richardson and Wright 1984). Mean monthly temperature, precipitation and solar radiation for  the period
are shown in Table 1. The temperature regime is seasonal and the largest monthly precipitation corresponds
 to the late summer.
                                           URUGUAY-4

-------
Climate Change Scenarios for the Region

        Global Climate Models (GCMs) were used to derive climate scenarios. Runs of these models were
used to estimate change in climate variables (temperature, precipitation, and solar radiation). These changes
were then used to modify the baseline climate and produce the climate change scenarios at each site. The
equilibrium GCMs used were: Goddard Institute for Space Studies (GISS, Hansen et al. 1983); Geophysical
Fluid Dynamics Laboratory (GFDL, Manabe and Wetherald 1987); and  United Kingdom Meteorological
Office (UKMO, Wilson and Mitchell 1987). Transient scenarios were created from the GISS transient model
simulations (Hansen et al. 1988).
        The following climate scenarios were run: (a) baseline; (b) GCMs (GISS, GFDL, and UKMO)
doubled CO2 climate change; (both (a) and (b) with and without the direct effects of CO2 on photosynthesis);
(c) sensitivity  studies with combinations of 0, +2C,  +4C, and 0, +20%, -20% precipitation, over the
baseline, each with and without direct CO2 effects; (d) transient climate change for the 2010s, 2030s and 2050s
using the GISS model; and (e) adaptive responses, e.g. the use of soil and crop management practices that are
better adapted to the possible new climatic conditions.
        The three GCMs used to predict climate changes under doubled CO2 conditions show similar trends
with respect to the average mean temperature (Figure 3a). All three models predict an increase in the monthly
average of about 5C. The UKMO model predict the largest increase (about 6C); GFDL produces the
smallest change (approximately 4C).
        The GCMs show contrasting trends for precipitation (Figure 4a). Both the GISS and UKMO models
predict a general increase in total precipitation. Increases in rainfall are predicted for spring and fall, although
the GISS model also predicts a considerable increase in rainfall for the winter months. The GISS and UKMO
models predict higher means for annual precipitation than the baseline scenario - approximately 150 and 300
mm higher, respectively. In contrast, the GFDL model predicts a small decrease in total precipitation for most
of the fall and winter months except for September (Figure 4). Excluding this month, the mean annual
precipitation predicted with the GFDL model is about 50 mm lower than the baseline.
        All three models predict very small changes in solar radiation (Figure 5a). The UKMO model projects
a slightly larger solar radiation value for January and February, approximately 2 MJ/m2.
        Because the UKMO model predicts the most unfavorable climate changes (largest temperature and
precipitation increases) for barley production in  the  area it was used for the adaptive response studies
discussed earlier. It was also the only GCM included in this study in which a model gridbox included the entire
land area of Uruguay.
        The GISS model was also used to predict  transient climate changes for the 2010s, 2030s, and 2050s.
Figures 3b, 4b and 5b show the results of these predictions  for temperature, rainfall, and solar radiation,
respectively. The results indicate a gradual temperature increase through the 2010s, 2030s, and 2050s, reaching
the maximum at the doubled CO2 scenario. The variation in the total precipitation was less systematic and
showed a very high rainfall value for March in the 2030s (GCM regional precipitation projections are highly
uncertain). As expected, the predicted solar radiation values showed almost no change from  the baseline data
over the three decades.

 Crop Model and Simulation Strategies

         An International Fertilizer Development Center (IFDC) - United Nations Development Programme
 (UNDP) global project was started in 1990, which will include the validation and regional adaptation of the
wheat, rice, maize, soybean, sunflower, and barley CERES models  for further studies. The CERES-Barley
 model, based on the CERES generic model, was used for all the simulation activities reported here (IBSNAT
 1989). The soil used for the modeling activities, (fine, messic, typic, Argiudoll) was the same soil that was used

                                           URUGUAY-5

-------
for the model validation. A complete description of the chemical and physical properties for 10 cm layers
throughout the soil profile is shown in Table 2.
        The simulations are described in Table 3. Three possible fertilization strategies were used: (a) no N
fertilizer; (b) 60 kg N/ha at planting + 60 kg/ha at the end of tillering (a common practice in the region); and
(c) optimal N fertilization. This last strategy simulates the situation of having a diagnostic tool available to
ensure optimal N fertilizer application, one of the key objectives of the barley research mentioned above. Also,
given the long planting season for winter crops, three different planting dates were used to run the model: 1
June (early), 21 July (normal), and 28 August (late).
        Frequent rains in summer and in fall result in problems with the soil tillage for winter crops. Common
consequences are inadequate seedbeds and a very compacted layer at 20-25 cm. The contraction is caused by
numerous tillage operations  performed in a short period of  time. An attempt was made to simulate this
situation by setting the value of the weighting factor  to determine new root  growth (WR in the CERES
models) equal to zero for all soil layers below 30 cm. However, the model was insensitive to the simulated
poor soil preparation. Setting the value of the weighing factor to determine new root growth (WR in the
CERES model) equal to 0.0 for all soil layers below 30 cm did  not affect the results in crop production (data
not shown). Therefore, the model was not used to perform any further simulations changing soil preparation.
        Since the climate change scenarios are associated with higher levels of CO2 (designated as 555 ppm
in the figures, representing an equivalent doubling of CO2) than the current climate GCM simulations (330
ppm in the figures), the study includes simulations of the physiological effects of CO2 on barley. Higher levels
of atmospheric CO2 have been found to increase photosynthesis and water use efficiency, resulting in yield
increases in experimental settings (Acock and Allen 1985; Cure 1985).

Validation and Calibration of the Crop Model

        A complete dataset is available in Uruguay that is suitable for calibrating and validating the CERES-
Barley model. All the crop production variables (dry matter production, grain yield, kernel weight, grains/m^
grains/ear, etc.) of the CERES model were validated  with this dataset. However, the phenology variables
(anthesis dates and maturity dates) were also tested with results from three years of cultivar trials conducted
by the Department of Plant Breeding of INLA, The phonological variables were calibrated and validated for
planting dates similar to the early, normal and late dates used in the simulation studies.
        The CERES-Barley model validation procedures for  this study were performed in two stages: (a)
determination of the genetic coefficients and validation of the phonological variables of the model; and (b)
validation of the crop production variables. Data from several barley planting date trials were used for the first
stage. The genetic coefficients, P1V and P1D, were adjusted  for the cultivar CLE-116, a newly developed
cultivar from INIA, using three years of data. The model's accuracy in predicting anthesis and maturity dates
was excellent. The difference between observed and simulated  dates for all years and planting dates included
in the study was always less than three days. The CERES model also accurately predicted the end of tillering
in most situations.
        The validation of the crop production variables was performed using data from 1991 research. The
most relevant results of this validation are shown in Figures 2a-2g. For example, there was a high correlation
(r>0.9, P<0.01) between the simulated and observed results for grain yield, total biomass production, straw
production, grains/m2, and kernel weight. However, the model did not simulate grains/ear or for nitrogen
content of the grain as well.  Considering the grains/ear, the principal problem was that the model did not
correctly predict the number of spikes per m2 (data not shown). In all treatments, the  predicted number of
spikes/m2 was always lower than the observed data, indicating a problem with the model routines that calculate
tiller production, or the number of tillers that produce spikes.  Observation of the predicted number of tillers
showed that, in most cases, the simulated values were similar to the ones observed, and in some cases, the


                                           URUGUAY-6

-------
predicted values were higher than those observed. This suggests that a problem may exist predicting the
proportion of tillers that will produce spikes, a topic to be studied in further validation research.
        The reason for the poor performance of the model in predicting the nitrogen content of the grain is
not clear. Problems may exist during the grain-filling stage because the length of this stage greatly affects the
N content of the grain. Finally, the kernel weight and the N content of the grain simulated by the model
showed a smaller sensitivity to N fertilizer than that observed in the field research. The kernel  weight
depended only on the value of G2  (kernel weight genetic coefficient) and  was constant for all the N
treatments. Similarly, the N content of the grain observed in the field varied more than the values predicted
by the model. Although these limitations of the model justify  further work  on the routines and in the
validation procedures, the  general performance  of the CERES-Barley  model  was very good. This was
particularly true for simulating  grain yield, biomass production, and dates of anthesis and maturity.

RESULTS AND DISCUSSION

Effects of 2xCO2 GCM Climate Change on Barley Production

        Yield. Figures 6a and 6b show the effects of the climate change predicted with the three GCMs on
barley production. Without nitrogen fertilization, the GCMs predicted a 50% mean  decrease in grain yield.
When nitrogen fertilizer was applied at rates and timings that are similar to Uruguayan farmers' practices (60
kg N/Ha at planting + 60 kg N/ha at tillering), the yield was reduced by 40%. In all of the climate scenarios
tested, N fertilizer affected the grain yields, especially in the normal (NP) and late (LP) planting dates. Figure
6 indicates that normal planting date yields doubled and that late planting yields  tripled with N fertilizer.
        The interaction of planting date and nitrogen fertilizer for the baseline and the three GCMs resulted
in a substantial difference between early planted (EP) and NP barley yields when no fertilizer was applied. In
contrast, the yields of EP and NP were veiy similar when N fertilizer was used. A possible explanation of these
results is that early planted barley remains in the vegetative growth stage for a longer  period of time than NP
barley. This gives the soil more time to supply larger quantities of N (through mineralization) at a stage when
N requirements are relatively low. When barley is planted at a normal date, the crop is much more dependent
on fertilizer N.
        Direct CO2 effects. Figure 6b shows the physiological effects of CO2 on crop growth.  The results
indicate that the baseline yields increased by 20% and the GCM climate change scenario yields increased by
30%. The yields simulated with the GCM scenarios that included the direct effects of  CO2 were 30% lower
than the yields of the baseline climate at 330 ppm of CO2 concentration.
        Yield variability. Baseline yields were less variable than yields simulated with three GCM scenarios
(Figures 7a and 7b). However, the main source of variability in grain yields was planting date. Figure 7 shows
that the CV increased from 20% to almost 50% when the planting time was delayed to a later date. Also, the
CVs for the GCMs were much larger those for the baseline at late planting dates. Finally, N fertilization did
not affect variability of the grain yield: the CVs with or without fertilizer were  similar.
        Biomass and season length. Decreases in barley grain yield can be caused by  a reduction in the total
biomass production (which is highly correlated with grain yields), and/or by a  reduction in the grain filling
period. To explain the observed reductions of grain yield, we studied the effect  of the GCMs on the total
production of barley biomass (Figures 8a and 8b). The experiments showed that the biomass production
changed in proportion to grain yields, and as a result, the barley harvest index (grain yield/total biomass,
expressed as a percent) did not change. Therefore, reductions in grain yield were mainly due to reductions in
the production of total dry matter. The decrease in total biomass production was caused by a shorter growing
season (Figure 9). The figure indicates that: (a) the period from emergence to maturity was shortened by a
delay in the planting date; and  (b) for any given planting date, all of the GCM  scenarios caused a reduction

                                            URUGUAY-7

-------
of the period equal to 30 days for EP, 20 days for NP, and 15 days for LP - about 20% in all cases.

Sensitivity Analysis

        The results in Figures 3 and 4 indicate that the GCMs predicted important changes in the temperature
and total precipitation. A sensitivity analysis was performed to try to identify which of these variables was
responsible for the negative effect on the grain yields (Figures lOa and lOb). The analysis indicates that the
barley grain yields were much more sensitive to temperature changes than to precipitation changes. This is
expected since Uruguay does not have periods of water shortages in the winter. As mentioned above, one
factor causing reductions in winter crop yields is the excess of water in the soil. In all of the simulation runs,
there were no years with periods of water stress in the vegetative stage and very few years with periods of mild
water stress during the reproductive stage.
        The results of the sensitivity analysis indicate that the mean grain yields were reduced by 0.45 t ha"1
for each 1C of mean temperature increase. This reduction  is similar  for the three tested precipitation
scenarios. Grain yields decreased 0.101 ha"1 for each 10% decrease in total precipitation and did not decrease
with the potential period  of excess of water in the soil.
        Figure 11 shows  the results of the sensitivity analysis on total biomass production. The  effects of
temperature and precipitation changes were similar to the effects on the grain yields, suggesting that the
reduction in grain yield - caused by an increase in the mean temperature - is a consequence of a reduction in
the total biomass produced. Figure 12 shows the effect a of temperature increase on the length of the period
from emergence to maturity.  The results indicate that the crop growing season was shortened by five days for
each  1C increase in the mean temperature.

Transient Analysis

        The analysis of the transient climate change using the GISS model indicates a gradual increase in the
mean temperature through the 2010s, 2030s and 2050s. As expected, the transient analysis of barley production
indicated a reduction in grain yields and an increase in yield variability (Figures 13a and 13b). Barley grain
yields were reduced by 1.7 t ha"1 from the baseline for the 2010s, and by an additional 0.35 t ha"1 for each of
the following decades. There is no apparent reason for this differential reduction since the temperature change
predicted for the barley-growing season was gradual.

Adaptation Strategies to Climate Change

        Adaptive responses were tested to mitigate the negative effects of the predicted climate changes of
theUKMO model, including the improvement of nitrogen fertilizer management and the creation of a cultivar
with a longer grain-filling period. An additional cultivar which is more adapted to earlier planting dates was
also simulated. The results of these runs indicate the importance of the these alternatives  (Figures 14a and
14b). The  use  of a more adequate fertilizer regime produced an increase in yield  of  1 t  ha"1 over the
unfertilized crop. An additional 11 ha"1 increase was achieved without N limitations. Finally, another 11 ha"1
increase was gained by using  the optimum N management with an improved "hypothetical cultivar" designed
with a higher number of kernels/m2, and  a longer grain filling period.
        The adaptive response analysis shows that using improved crop management practices would increase
currently attainable grain yields of 2.5-3.0 T/Ha (baseline with 60 kg N/ha at planting + 60 kg N/ha at tillering,
in Figure 14a), and results in  a mean grain yield of 5.0 T/Ha, even under the unfavorable weather conditions
predicted by the UKMO model. However, Figure 14a also shows that the mean grain yield achieved with these
same crop management practices would be 6 t  ha"1 with 330 ppm CO2,  or 7.5 T/ha with direct effects of

                                           URUGUAY-8

-------
doubling the CO2 concentration - indicating a potential loss of 1.0-2.5 t ha"1 under the predicted climate
changes.
        A second set of simulation runs to compare the barley response to nitrogen fertilizer under baseline
and UKMO conditions. For the baseline runs, CLE-116 was used at the normal planting date. The same
cultivar was used for the UKMO runs at the early planting date. The planting date was changed for the
UKMO runs because the model had predicted an increase in the mean temperature, which in turn resulted
in a shorter growing season. The probable response of farmers to this change would be to plant at earlier dates
and avoid high temperatures during the grain-filling period. The results presented for the climate change
scenario included the physiological CO2 effects.
        Figure 15 shows the results of testing four nitrogen fertilizer rates. The response curve was adjusted
with linear regression analysis. Because the baseline response curve was much steeper than the UKMO curve,
the baseline maximum yield was more than 11 ha"1 higher than the UKMO maximum yield. Also, the amount
of N fertilizer needed to attain the maximum grain yield  under UKMO was 2.6 times more than the amount
required to attain the same yield under current climatic  conditions.

CONCLUSIONS

        The climate changes predicted by the three GCMs are increased mean temperature, unchanged or
increased precipitation, and unchanged solar radiation.  These climate conditions resulted in reduced crop
growing seasons and lower  total biomass production and grain yields for barley in Uruguay. The negative
effects of the GCMs were worse when no nitrogen fertilizer was applied, and/or when the planting date was
delayed. The variability of grain yields was greater under the GCM conditions and when the planting dates
were delayed, but it was not affected by N fertilization.
        The sensitivity analysis indicated that the decrease in grain yields is related to the reduction of the
barley-growing season. This reduction results from the temperature increase predicted by the GCM climate
change scenarios. In the major barley-growing region of Uruguay, the estimated yield effect of increasing the
mean temperature by 1C is equivalent to the estimated effect of a 40% reduction of the total precipitation.
The adaptive response studies indicated that although  there is a good potential for developing adaptive
management practices for the predicted climate changes, significant losses in crop production can still be
expected.
                                           URUGUAY^

-------
REFERENCES

Acock, B. and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
        and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation. DOE/ER-0238. U.S.
        Department of Energy. Washington, D.C.

Cure, J.D. 1985. Carbon dioxide doubling responses: A crop survey. In B.R. Strain and J.D. Cure (eds.). Direct
        Effects of Increasing Carbon Dioxide on  Vegetation. DOE/ER-0238. U.S.  Department of Energy.
        Washington, D.C.

Hansen, J., I. Fung, A. Lascis, D. Rind, S. Lebedeff, R. Ruedy and G. Russell. 1989. Global Climate Changes
        as  Forecasted  by Goddard  Institute for Space  Studies  Three-Dimensional Model. Journal of
        Geophysical Research, 93: 9341-9364.

Hansen, X, G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R. Ruedy and L. Travis. 1983. Efficient Three-
        Dimensional Global Models for Climate Studies: Models I and II. April Monthly Weather Review, Vol
        III, No. 4: 609-662.

IBSNAT. 1989. International Benchmark Sites Network for Agrotechnology Transfer Project. Decision Support
        System for Agrotechnology Transfer Version 2.1 (DSSAT V2.1). Dept. Agronomy and Soil Sci., College
        of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii 96822.

IPCC 1990. Intergovernmental Panel on Climate Change. First Assessment Report.

Manabe, S. and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in atmospheric
        carbon dioxide. Journal of Atmospheric Science, 44:1601-1613.

Parry, M.L., T.R. Carter and N.T. Konijn. 1988.  The  impact of climatic variations on agriculture.  Vol 1
        Assessments in cool temperate and cold regions.   Vol 2 Assessments in semi-arid  region. Kluwer,
        Dordecht, Netherlands. 876 pp. and 764 pp.

Richardson, CW. and D.A. Wright.  1984. WGEN: A Model for Generating Daify Weather Variables. ARS-8.
        U.S. Department of Agriculture, Agricultural Research Service. Washington, DC. 83 pp.

Wilson, C.A. and J.F.B. Mitchell. 1987. A doubled CO2  Climate Sensitivity Experiment with a Global Model
        Including a Simple Ocean. Journal of Geophysical Research, 92: 13315-13343.
                                          URUGUAY-10

-------
Table 1.         Mean temperature, precipitation, and solar radiation for baseline data in La Estanzuela,
                Uruguay (1966-1989).
Mean Total
Month Temperature Precipitation

JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
DEC
C
21.1
20.3
18.5
15.7
12.6
9.6
9.8
10.5
12.2
14.4
16.6
19.8
mm
77.6
85.8
107.0
58.7
59.5
62.3
67.2
73.0
76.6
96.7
80.7
78.1
Solar
Radiation
MJ/m2
23.0
20.8
16.9
13.0
9.4
7.3
8.0
10.7
14.4
18.5
21.6
23.3
                                          URUGUAY-11

-------
Table 2.
Description of the soil used for modeling activities.












Layer
cm
0-5
5-10
10-20
20-30
30-40
40-50
50-60
60-70
70-80
80-90
90-100
LL
DUL
SAT
?D
OC
SW
LL

0.124
0.124
0.119
0.127
0.186
0.200
0.199
0.192
0.193
0.185
0.186
DUL
cm3/cm
0.320
0.320
0.290
0.300
0.290
0.310
0.310
0.300
0.290
0.280
0.270
= Lower limit of soil
=
Drained
SAT
3
0.350
0.350
0.320
0.350
0.350
0.360
0.360
0.350
0.340
0.320
0.310
SW

0.320
0.320
0.290
0.300
0.290
0.310
0.310
0.300
0.290
0.280
0.270
BD
g/cm3
1.28
1.28
1.32
1.32
1.36
1.40
1.40
1.42
1.42
1.42
1.42
OC
%
2.08
2.08
2.08
2.08
.78
.78
.95
.95
.14
.14
.14
NH4

5.0
5.0
5.0
2.0
2.0
1.0
1.0
1.0
1.0
1.0
1.0
NO3
mg/kg
20.0
15.0
10.0
5.0
1.0
1.0
1.0
1.0
1.0
1.0
1.0
PH

5.6
5.6
5.6
5.6
6.1
6.1
6.6
6.6
6.6
6.8
6.8
extractable water
upper limit of soil water content
= Saturated water content
= Bulk density
=
Organic
Carbon







= Initial soil water content
                                             URUGUAY-12

-------
Table 3.
Management practices used in the simulations.
 a)      Baseline and GCMs (x2 CO2) with and without direct effects of doubled CO2 concentration
         TOTAL: 48 runs
 Planting Date
 Early (1-June)
 Normal (21-July)
 Late (28-August)
 Early (1-June)
 Normal (21-July)
 Late (28-August)
              N fertilizer
              No N Fertilizer
              No N Fertilizer
              No N Fertilizer
              60 + 60 Kg N/ha (Plant. + Till.)
              60 + 60 Kg N/ha (Plant. + Till.)
              60 + 60 Kg N/ha (Plant. + Till.)
 b)      Sensitivity analysis (0,+2,+4 C X 0,-20,+20% mm) with and without 2xCO2 TOTAL: 18 runs
 Planting Date                N fertilizer
 Normal (21-July)             60 + 60 Kg N/ha
 c)      Transient climate change with and without 2xCO2 TOTAL: 6 runs
 Planting Date                N fertilizer
 Normal (21-July)             60 + 60 Kg N/ha
 d)      Adaptive Response with and without 2xCO2 TOTAL 12 runs
 Planting Date
 Normal (21-July)
 Normal
 Early
 Normal
 Normal
 Early
              N fertilizer
              60 + 60 Kg N/ha
              Optimal
              Optimal
              Optimal
              Optimal
              Optimal
Cultivar
CLE-116
CLE-116
CLE-116
Improved #!(*)
Improved #2
Improved #3
Improved cultivar #1 = CLE-116 with increased grains/m2
Improved cultivar #2 = Improved #1 with longer grain filling period
Improved cultivar #3 = Improved #2 better adapted to early planting
                                        URUGUAY-13

-------
 LLJ
 O
OC
111  CO
O  O
~Z.  LL.
O
 D)
L.
            E
            
                                            O
                                            LLJ
                                            Q


                                            O
                                            O
                                            O
                                            CL
                                            LJJ
                                            CO
                                            O
                                           DC
                                           Q_
                                           <
                                           DC
                                           CO
                                           LLJ
                                           LL









CO
CO
LJJ
O
X
LJJ
CO
1
LJJ
CO




I
O
LJJ
DC
*^
CO
1
CL
Z> O
^ LJJ
5 DC






CO

H
LU
DC
1
1

CL
L
L
f
L

^
L
LI
~\
-1
>
>


              ooooooooo
              CO'^CMOCOCO'sJ-CM

-------
O
> c
3S
go
CO
     ra w 
    I
                     rt
                     -C
                     1  I
f

3
o
a
V








v-
*
\
O
J)
O
C
.g
i *
cd
33
15
"o
a
o
m
CO
LU
DC
LJLJ
O
di
s
1
a
si
a-3
 e

_ w S
-  ^^

  o
                                                     I
                                                  5
                     I

-------
II
  "
I
TJ -
H 0)
^2 O.

_ to

i

^6
      10
      a
          m
          10
at
t
     g
                         2



                         sl
                         S
        


        R
                                   O
                                   O
                                  CM


                                   D)

                                  LL
                                    I
                                    O)
                                                              
                                                            K> O.

-------
LJJ
o
I
o

UJ  O

fe  =
< 
*sr  co
*>.  v_
~"!  0
UJ

CO
DC
H

"3
111
O
X  0
Ob
Hi 'ra


II
-J ,0
o -
o
o

"co
o


Q.



I

C
CO
o
                                                Q_
                                                UJ
                                                CO
                                                _J  X
                                                Z>  H
                                                -^  z
                                                    o
                                  -- <
  o
  CO
                    ID
                    CM
               O    ID    O
               CM    T-    T-
            O
S
o>
Q.


I

C
CO
o
                                     o
                                    z.
                                                 Q.
                                                 CD
                                                CO
                                              _.  3
                                                C
                                             -- CO
                                                                 o
                                                                 CM
_j   CO   O

LU   52   o
Q   O   CM
O
2
                                                                 \    \
                                                                LJJ   O
                                                                CO   CO
                                                                <   O
                                                                m   CM
                                                                 C
                                                                 0)

                                                                "co
                                                                 C
                                                                 cc
                                                                "O
                                                                 C
                                                                 03
                                                                            O
                                                                             
-------
LLJ
O
z
<

o

LU
H
 UJ
 CO

 <
 DC
c
g
*s

t?
'g.
'o
                                                  Q.
                                                  Ill
                                                  CO
                                                  _J I
                                                  13 V-
                                                 cc
              o
              10
              CM
         o
         o
         CM
                                 o
                                 o
                                         o
                                         ID
                                                                  o
                                                                  CM
                                                             Q   O   CM
                                                             o
                                                                  UJ   O
                                                                  CO   CO
                                                                  <   O
                                                                  m   CM
                                                                  c
                                                                  0)

                                                                 'co
                                                                  C
                                                                  05
                                                                              C
                                                                              03
O
o

 c
 o
+-
 03
LJLJ
O


<
X  c
0.9
< .9-
 O


 O
 O

'co
            E

           v^>

            c
            g
           *-
            CO
            o
            o
                                     O
                                      Q.
                                      UJ
                                      CO


                                      _J  I
                                      Z)  H
                                     DC
                                                                 CO
                                                                 CO
                                                          o
                                                                 O  ID
                                                             H   !    I
                                                                 o


                                                                 ^   Q
                                                                 CO   '-'
                                                                 CO   U.
                                                                 m   O
                                                                             g.
                                                                            "o
                                                                             0)

                                                                            ol
                                                                              0
                                                                              O)
                                                                              C
                                                                              03

                                                                             6
                                                                 D)

                                                                 LL
              o    o     o     o     o
              ID    O     ID     O     IO
              CM    CM     i-     -i-

-------
LLJ
O
LJLJ .2
    DC
    v_

h- -^
5 
dgco

CO

<
DC
H

JQ
LLI
O

<  C
I  .2

0  
HI  TJ
H  3
<  DC
2  V.
-   co
O
O
          CM
C
O
CO
DC
O
CO
  O
  CO
10
CM
O   U>   O
CM   T-   T-
IO
CM
           C
           O
          u->
           CO
          O
          CO
                                           |
                                 DL
                                 LU
                                 CO
        _J X
        13 I-
        -> Z
          O

        Is


        DC
        <
                       t <
                                 O
                                 Q.
                                 LU
                                 CO
                          _l  I
                          3  H
                          ~3  Z
                             O
                                        --  <
                                          _
                                          LU
                                          Q
                                          O
                                          LU
                                          Q

                                          O
                                              O
                                              CM
                                        CO
                                        CO

                                        o
                                       o
                                       ID
                                       o
                                       CM
                                              I   !
                                              LU  O
                                              CO  CO
                                              <  O
                                              CO  CM
                                       CO  5
                                       CO  6
                                   I   !
                                              o


                                              I  o
                                              CO  U-
                                              m  O
                                                  co
                                                  C
                                                  03
                                                 O
                                             c
                                             O
                                                                  TJ
                                                                   03
                                                                  QC
                                                         O
                                                        CO
                               

                                                        LJL
            O
            co
      10
      CM
    o   10   o
    CM   T-   T-

-------
 Fig.  6a:  BASELINE AND GCMs
           MEAN GRAIN  YIELDS
   GRAIN YIELD (T/ha)
    EP-NO  NP-NO  LP-NO  EP-N  NP-N   LP-N
                  STRATEGY
                   MODEL
        BASE

        GFDL
                   GISS

                   UKMO
Fig. 6b:   BASELINE AND GCMS
           MEAN GRAIN YIELDS
   GRAIN YIELD (T/ha)
 6-
 5-
	-40%	
   EP-NO  NP-NO  LP-NO  EP-N  NP-N  LP-N
                 STRATEGY
               MODEL (2 X CO2)

    EZ1  BASX          8SS3 GISX

    Hi!  GFDLX          UKMX

-------
Fig. 7a: BASELINE AND GCMs
           CV % GRAIN YIELDS
   YIELD C.V. (%)
   EP-NO NP-NO  LP-NO  EP-N
                 STRATEGY
   NP-N
LP-N
                  MODEL
        BASE
        GFDL
GISS
UKMO
Fig. 7b: BASELINE AND GCMS
           CV  % GRAIN YIELDS
   YIELD c.v. (%)
so -ri
   EP-NO NP-NO  LP-NO  EP-N   NP-N  LP-N
                 STRATEGY
               MODEL (2 X CO2)
        BASX          K\\\\N GISX
        GFDLX         HH UKMX

-------
 Fig. 8a: BASELINE AND GCMS
      MEAN BIOMASS PRODUCTION
   BIOMASS (T/ha)
16 -d
   EP-NO  NP-NO  LP-NO   EP-N  NP-N  LP-N
                STRATEGY

cm


BASE
GFDL
MODEL
l\\\\\l GISS
Wf& UKMO
Fig. 8b:  BASELINE AND GCMS
      MEAN BIOMASS PRODUCTION
  BIOMASS (T/ha)
16 -d
14-
   EP-NO NP-NO  LP-NO  EP-N  NP-N   LP-N
                STRATEGY
             MODEL (2 X CO2)

       BASEX        ESS3 GISSX

       GFDLX        W% UKMOX

-------
CO
<
CD

    LU
LU
O
DC
LU
E  LU
                              LU
                                 LU
                                 0
                              5
                              QC
                              LU
              o  o o o o o
              CD  ^f CM O 00 CD
                                         CO
                                         CO

                                         o

                                    LU
                                    Q

                                    O
LU
CO
<
CD

-------
Fig. 10a: SENSITIVITY ANALYSIS
            MEAN GRAIN  YIELDS
    GRAIN YIELD (T/ha)
         +2
+4   0+2+40
 TEMPERATURE CHANGE
+2   +4
              CO2 CONCENTRATION

              M YSENS  SSSI YSENX
Fig. 10b: SENSITIVITY ANALYSIS
             C.V. GRAIN YIELDS
     C.V. YIELD (%)
             +4   0+2+40
              TEMPERATURE CHANGE
                    +2  +4
              CO2 CONCENTRATION
              ^ SSOppm  ^^ 555ppm

-------
    O
CO  h-
<  O
    Q_

>-  CO
\-  CO
=:  <
  o
UJ
CO  <
 ..  LU
 D)
LL
CO
CO
g
CQ
C02 CONCENTRATION
555ppm
I


Q.
Q.
O
CO
CO
1


               \   i  I  I   \  I   \
            CD ^t CV! O CO CD -sf CM O

-------
Fig. 12: SENSITIVITY ANALYSIS
 EMERGENCE - MATURITY LENGTH
 DAYS EM - MAT
     0         +2        +4
        TEMPERATURE CHANGE

-------
 Fig. I3a: TRANSIENT ANALYSIS (GISS)
               MEAN GRAIN YIELDS
      GRAIN YIELD (T/ha)
     5-
     4-
     3-
     2-
     1
     0
1
        BASELINE    2010'S
                   2030'S
2050's
                 CO2 CONCENTRATION
                 S! 330ppm     555ppm
Fig. I3b:  TRANSIENT ANALYSIS (GISS)
                C.V. GRAIN YIELDS
       c.v. (%)
        BASELINE   2010'S
                   2030's
2050's
                 CO2 CONCENTRATION
                 S 330ppm      555ppm

-------
Fig. I4a: ADAPTIVE RESPONSE (UKMO)
               MEAN GRAIN  YIELDS
     8
     7-
     6-
     5-
     4-
     3-
     2-
     1-
      GRAIN YIELD (T/HA)
                           NO GC
            IMPR. N + CULTIVAR
IMPROVE is!
        V-O N 0 V-O 60+60 V-O Nopt  V-1 Nopt V-1 Nopt
                 CO2 CONCENTRATION

                 H 330ppm  K\\\\M 555ppm
Fig. I4b:  ADAPTIVE RESPONSE (UKMO)
                C.V. GRAIN YIELDS
       c.v. (%)
                          NOGC
           IMPROVE N PERT.
                       IMPR. N + etlLTIVAR
                                  I
        V-O N 0 V-O 60+60 V-O Nopt V-1 Nopt V-1 Nopt
                  CO2 CONCENTRATION

                 H 330ppm   K\\\\N 555ppm

-------
 Fig. 15
 Grain Yield (T/ha)
4 -  3.48 T/ha
1
0
                 96 kg N/ha
                  37 kg N/ha
   0
 40          80
Fertilizer (kg N/ha)
120
       BASELINE    -a- UKMO (2xCO2)
       NP              P

-------

-------
SECTION 4: EUROPE

-------

-------
     POSSIBLE EFFECTS OF INCREASING C02 CONCENTRATION
ON WHEAT AND MAIZE CROPS IN NORTH AND SOUTHEAST FRANCE
                       R. Delecolle, D. Ripoche
                  Station de Bioclimatologie INRA, France

                         F. Ruget, G. Gosse
                  Station de Bioclimatologie INRA, France
                           FRANCE-1

-------
TABLE OF CONTENTS

SUMMARY
INTRODUCTION
      Agricultural regions and systems
METHODS
      Climate
      Crops Models and Management Variables
      Calibration of the crop models
      Climate change scenarios
RESULTS WITHOUT ADAPTATION
      Wheat
      Maize
RESULTS WITH ADAPTATION
      Wheat
      Maize
DISCUSSION
REFERENCES
                                   FRANCE-2

-------
ACKNOWLEDGMENTS

       The authors are grateful to Claude Varlet-Grancher and Pierre Pluchard (both from INRA) for
providing wheat data.
                                       FRANCE-3

-------
SUMMARY

        This study evaluated the potential effects of climate change due to increased trace gas concentrations
in the atmosphere on wheat and maize yields in France. Sensitivity studies showed that yields generally
decreased with increasing temperature, probably due to a shorter crop-growing season. When the direct effects
of CO2 on the crop were included, yield decreases were somewhat counteracted.
        Under both temperate and mediterranean climates in France, winter cereal yields did not decrease
under future climate conditions (provided that irrigation supply was not limiting under dry conditions). Under
temperate climate, maize  yields increased, while under mediterranean conditions, the reduction of phase
duration induced a drastic yield decrease, even under optimal irrigation. Adaptation simulations (change in
planting date) produced slightly higher yields. If these results are realized with global warming, maize may be
farmed in France more extensively in the future.
INTRODUCTION

        This study was part of an international endeavor to assess the possible effects of global climate change
on world food production in the coming century. Any study of this type must account for the climate
modifications induced by increasing trace gas concentrations in the atmosphere, which are simulated by general
circulation models  (GCMs), and the direct effects of increasing carbon dioxide concentration  on plant
physiology. These studies are particularly interesting in northern and middle Europe, where cropping systems
will likely shift, within a few decades, from high-input production to more extensive practices, due to an
increasing concern  for pollution costs and over-production. It is therefore necessary to estimate whether
climate change will create extra costs (irrigation, pesticides), or if increasing CO2 will promote crop production
at steady costs. In this study,  we used the CERES plant process  models to estimate the response of wheat and
maize crops in France and to address the effects of climate change on yield and water demand.

Agricultural Regions and Systems

        The research was conducted in two different agricultural areas in France (Figure 1). These regions
were chosen because they represent two extremes of French agroclimatologic conditions.  The northern plain
is characterized by intensive cropping practices, semi-oceanic climate, and little water stress. The soils are deep
silts with high field capacities. The most frequently grown crops are wheat and sugarbeet, with several
secondary crops such as peas, potatoes, and beans. Maize was introduced in the northernmost areas 15 years
ago.
        The southeast has diverse crops. Fruits and vegetables, maize, sunflower, and sorghum are grown in
irrigated areas; wheat (soft and durum) and barley, are usually rainfed except on very shallow soils. The climate
is mediterranean, with drought periods during the summer, characterized by a large interannual variability. Soil
types  are highly variable and generally shallow with low field capacities. In river valleys, however, alluvial
deposits and water tables insure better water supplies.

METHODS

Climate

        For northern France, we used a timeseries of climate data from 1951 to 1980 (1984 for wheat) that
was available from Versailles-INRA (lat. 49N). This  weather  station, albeit located in the southern part of

                                            FRANCE-4

-------
the study area, provides data that are representative of the area. For the South, the Avignon-INRA weather
station (lat. 44.9N) provided data from 1968 to 1984.

Crop Models and Management Variables

       Wheat and maize were the sample crops for this study. These crops provide a physiological comparison
between C3 and C4 crop responses to the direct effects of increased COV and a comparison between winter
and spring crops for testing the influence of indirect climatic effects on season length and water requirements.
Both crops were tested at each site.
       The choice of genotypes was guided by the availability of calibration data. For wheat, we chose one
genotype per location to account for the specific conditions and farmers' practices at each site. We chose the
genotype Arminda (winter type, long cycle) for northern France, and the genotype  Fidel (half-spring,
medium-length cycle) for the South. For maize, we used only one genotype, Dea, which can be grown in both
climates, with special regard to its insensitivity to photoperiod. It is often used as a control in productivity
tests.
       Crop models. The CERES-Wheat (Godwin et al. 1989) and CERES-Maize (Ritchie et al. 1989) models
provided by the DSSAT package were used to simulate wheat and maize crops. Modified versions accounted
for higher-than-ambient CO2 concentration changes. Photosynthesis enhancement factors for various CO2
concentrations came from Allen et al. (1987).
       Soils. For northern France, we used a standard "Deep Silt Loam" soil (IBSNAT 1989) tp represent
the soils found in this area. For the Southeast, we created a specific soil file from the soil properties that exist
in a typical alluvial area near Avignon-although farmers would grow mostly vegetables and other high-return
crops on this soil.
       Irrigation. For all of the simulations,  the initial soil water was set to 100% capacity for each layer.
This is a realistic assumption for both locations under present climatic conditions. When irrigation was applied,
the "automatic" option was chosen to provide the crop with a non-limiting situation.

Calibration of the Crop Models

        Wheat. The CERES-Wheat  genetic coefficients for both wheat  genotypes were estimated  from a
dataset from the Mons-en-Chaussee-INRA experiment station (49.8 N). The data were representative of the
wheat area in the North. We used a  sowing date  experiment with  three years of data and with eight
independent treatments. Experimental data were split into  two  sets-one for calibration and the other for
validation. The calibration was carried out in two phases. First, each parameter was calibrated directly from
observed results (phase durations, biometric ratios, growth rates) to obtain rough parameter values. Second,
these values were used in model runs and arbitrarily adjusted to attain a pseudo-best fit of observed data
(Table 1). The validation procedure showed acceptable results for Fidel, but underestimated yields for Arminda
by 13%, due to underestimation of final kernel weights.
        Durations from sowing to anthesis were simulated adequately most of the time, but simulated anthesis-
to-maturity periods were generally too short in comparison to reality. These results were attained with values
of Gx coefficients which are different from what is suggested in DSSATs CERES-Wheat guide for European
wheats (Godwin et al. 1989). All growth coefficients are highly correlated, thus leading to potentially large
instabilities in estimated values-a crucial point for study in the future. Figure 2 displays the sensitivity of
simulated yields to values of P1D, Gl, and G2.
        Maize. Two  sets of data were  used  to study maize. The results  from a planting-density  trial in
Mons-en-Chauss6e were used for the calibration of genetic parameters and initial values (sowing depth). The
results from another density trial in Grignon (near Versailles) allowed us to test the consistency of estimated

                                            FRANCE-5

-------
 parameters. PI and sowing depth bave been adjusted to attain satisfactory anthesis dates and maximum leaf
 area indices, and P2 was set to zero because Dea has been shown to be a non-photoperiodic genotype.
        The calibration of parameters linked to grain number and grain filling was problematic. In France,
 maize kernels rarely reach "physiological maturity" (P5). Also, the potential grain number was dramatically
 increased to insure an adequate simulation of the final number. Finally, the grain-filling rate (G3) was set to
 far below the potential value. Therefore, the set of parameters obtained (Table 2) is different from the values
 indicated for our latitude (Ritchie et al. 1989).

 Climate Change Scenarios

        Climate change ratios from three GCMs have been applied to both weather series. Three equilibrium
 GCMs were used: GISS (Goddard Institute for Space Studies; Hansen et al. 1983); GFDL (Geophysical Fluid
 Dynamics Laboratory Model;  Manabe and Wetherald 1987); and UKMO (United Kingdom Meteorological
 Office Model; Wilson  and Mitchell 1987).  Transient climate scenarios were also created with the GISS
 transient model (Hansen et al. 1988). Figure 3  shows the monthly averages of changes in required weather
 variables for the three doubled CO2 and transient (2010s, 2030s, 2050s) scenarios.
        Temperature. For  both regions, temperatures simulated by GISS and GFDL lie 3-4 C above the
 present mean values, whereas  UKMO predicts more drastic changes-more than 10C during the summer in
 Avignon, and more than 6C  during the winter in Versailles.
        Precipitation. According to all scenarios, precipitation increases during the winter  in Avignon and
 throughout the year in Versailles. UKMO simulates a dramatic precipitation decrease during the summer in
 the South. Individual monthly values for precipitation change are very chaotic in the Southeast for the GISS
 and UKMO simulations, and in the North for GFDL.
        Solar radiation. For UKMO and GISS, the simulations of solar radiation suggest an increase  of about
 10% over the year for both sites. GFDL simulations are very close to baseline values.

 RESULTS WITHOUT ADAPTATION

 Wheat

        The following results correspond to current management practices: sowing on November 1 and plant
 population of 250 plants m"2 for both sites.
        Yield and season length. To synthesize the effects  of global change on crop behavior (yield, days to
 maturity), Figures 4-9 show scenarios arranged by the corresponding yearly mean increase in temperature. We
 simulated both irrigated and rainfed production at each  location. For the North (Versailles), the effect of
 irrigation is negligible, so irrigated and rainfed treatments are represented by one set of points in Figures 4a
 and 4b. The effects of climate change alone tended to decrease yields up to  almost 30% (GFDL), probably
 due to the shortened season length of up to one month  (Figure 4b). When CO2 physiological effects were
 added, this effect was counteracted; the yield changes were almost nonexistent  (at  least for the transient and
 equilibrium GISS scenarios).
       Figures 5a and  5b  show the results for irrigated and  rainfed wheat in Avignon. The  irrigated results
were similar to those obtained from Versailles. Rainfed wheat was less  adversely affected by climate change,
with the exception of the UKMO climate change scenario. When the direct effects of CO2 were included, all
scenarios produced increased yields except UKMO. Table 3 summarizes results of water consumption by the
crops as a percent variation from the baseline for Versailles (irrigated) and for Avignon (irrigated and rainfed).
       Water consumption. The simulation results show the difference between the effects of climate change
alone, due to a shortening of the crop season, and the effects of climate change and increased CO2, due to

                                           FRANCE-6

-------
increasing stomatal resistance. In both cases, the major reduction in evapotranspiration (ET) (Table 3) was
mostly attributable to climate effects (up to about two thirds), with a magnitude that varies according to
scenario. See also the sensitivity analysis (Appendix A). In the North, transient scenarios induced a small
reduction in ET, whereas equilibrium scenarios had a greater influence (up to a 22% reduction). For irrigated
wheat in the South, the overall effect on evapotranspiration could range from -10% to -20%. Rainfed crop
ET showed less dramatic reductions.

Maize

        The following results correspond to sowings on May 1 in Versailles and April 15 in Avignon. The
plant population was 9 plants m'2 in both cases.
        Yield and season length. Figures 6a and 6b summarize the results for yield and season length for maize
in Versailles, with the same representation as for wheat. Irrigation did not modify season length, which showed
a significant decline when the mean temperature roses, especially in the extreme case of UKMO. Surprisingly,
the yield response to climate change alone shows a marked optimum for conditions in 2010. This could be
interpreted as a consequence of the shift in development phases, which positioned the grain-filling period more
favorably in terms of available solar radiation. Even  if season length was  reduced, the yield was not
substantially decreased below the baseline level even in the least favorable conditions (equilibrium scenarios).
This trend was confirmed  by sensitivity analysis results (Appendix B). Adding the CO2 physiological effect
increased yields up to 15-20% above the baseline.
        Figure 7 shows the results from irrigated and rainfed maize at Versailles. Irrigation had only a slight
influence on the yields. (Irrigated baseline was 8.521 ha"1 compared with 8.431 ha"1 under rainfed conditions.)
Although changing the climate reduced the length of the development phases, it also increased the yields in
the rainfed case (Figure 8a), probably due to the same phenomenon that occurred in Versailles (better timing
of development to radiation availability). Adding direct CO2 effects led to a large increase in grain dry-matter
production.  Rainfed baseline yields were very low (simulated 2.3 t  ha"1). Yet if the simulation  results are
reliable, a  high  CO2-high temperature  environment would lead to high  yields under rainfed  conditions
(simulated 5.91 ha"1 for UKMO). With irrigation, maize yields showed a decline from 10 to 30% with climate
change alone,  and adding CO2 effects did not consistently modify the results (Figure 9). The worst yields in
this case were still higher than the best rainfed simulations.
        Water consumption. Table 4 summarizes simulated ET for maize. In Versailles, decreases in water use
largely depended on the assumed scenario. When direct CO2 effects  were added, a 20% to 30% decrease in
water consumption occurred in the equilibrium and 2050 scenarios for irrigated and rainfed crops. These
results were generally similar to results  for irrigated crops in Avignon.

RESULTS WITH ADAPTATION

        Possible adaptations to future climate conditions were tested with a range of sowing dates to adjust
the timing of crop phonological events for photothermal time and radiation availability.

Wheat

        For wheat, three sowing dates were tested for both locations: two of them earlier (September 1 and
October 1) than the regular date (November 1) and one later (December 1). A four-point response curve for
grain yield appears in Figure 10. Figure 10 also illustrates the grain  dry-matter response to the sowing date,
representing (in percent of lxCO2 baseline value for November 1 sowing) the transient scenarios including
direct CO2 effects. For Versailles, the sowing date ded not help to counteract the effect of season length in

                                            FRANCE-7

-------
the most pessimistic case (2050). Whatever the scenario, the yield difference between the first and last sowing
was about the largest that can be expected in an agronomic experiment at potential production level. In
Avignon, even earlier sowing dates improved the situation in the near future projections (compare September
1, October 1, November 1,2010s, and 2030s), but the initial soil water conditions would probably hinder this
solution.

Maize

        Figure 11 illustrates the effects of two more sowing dates, 15 days earlier and 15 days later than the
regular date. For the North, it appears that an earlier sowing date would not modify the yield changes, but
delaying the planting date clearly reduced the benefit  of climatic  change. Earlier sowing dates present a
problem for spring crops that are limited by soil temperature. Additional studies on the possible warming of
upper soil layers under changing weather conditions are needed to study this adaptation.
        In the South, postponing the planting dates would counterbalance the negative influence of climate
change, at least in the near future. This effect may be due to a better synchrony between phasic development
and available energy, as was suggested for Versailles. For the 2050 scenario, further study of an earlier sowing
(April 1) is necessary, assuming again that more information on soil temperature will be available.

DISCUSSION

        The overall trends that can be deduced from this study are: (1) Season lengths will be shortened by
climate change; (2) Yields will decrease in proportion  to the magnitude of warming, but may be partially
mitigated by direct CO2 effects (up to an increase of 5C in temperature); and (3) Water use will decrease.
        Two major questions remain:  (1)  Do the multiplicative coefficients used to  represent high-CO2
photosynthetic response account for  long-term plant behavior especially with possible acclimation? and (2)
How realistic is the simulation of the water budget in the crop models?
        The calibration problems we encountered preclude straightforward interpretation of absolute values,
but the relative or differential results probably deserve more credit. Assuming that complete water repletion
of all soil layers at sowing may strongly temper the following statements, it may be concluded that: (1) Under
both temperate and mediterranean climates in France,  winter cereal yields will not be decreased by future
conditions (provided that irrigation supply  is not limiting under dry conditions); and (2) Under  temperate
climate, maize could take advantage of development-phase shrinkage and improve its radiation-use efficiency.
The traditional assumption that C3 crops would take precedence over C4 crops because they use CO2 more
efficiently would become questionable in this case. Inversely, under mediterranean conditions, the reduction
of the phase duration may induce a  drastic decrease in yields, even under optimal irrigation. Could maize
become a potential crop for extensive fanning in the two regions, as the simulations would suggest?  The
answer depends in part on the ability of CERES-Maize to account for the effect of water stress on ear fertility
and the potential grain number.
        Finally, the diversity of French climates and soils prohibits generalizing these results for the entire
country. The two sites chosen for this study represent extremes for intensive crops in France;  more continental
situations should be studied to investigate the behavior of crops in the eastern agricultural plains regions under
future climate.
                                            FRANCE-8

-------
REFERENCES

Allen L.H., KJ. Boote, J.W. Jones, P.H. Jones, R.R. Valle, B. Acock, H.H. Rogers, and R.C. Dahlman. 1987.
        Response of vegetation to rising carbon dioxide: photosynthesis, biomass and seed yield of soybean.
        Global Biogeochimical Cycles, 1:1-14.

Hansen, J., G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R. Ruedy and L. Travis. 1983. Efficient Three-
        Dimensional Global Models for Climate Studies: Models I and II. April Monthly Weather Review, Vol
        III, No. 4:609-662.

Hansen, J., I. Fung, A Lascis, D. Rind, S. Lebedeff, R. Ruedy and G. Russell. 1988. Global Climate Changes
        as  Forecasted by Goddard Institute for Space Studies Three-Dimensional Model. Journal  of
        Geophysical Research, 93:9341-9364.

IBSNAT 1989. DSSAT (Decision Support System for Agrotechnology Transfer) Version  2.1 User's Guide.
        Honolulu,: Univ. of Hawaii.

Godwin D., J. Ritchie, U. Singh, and L. Hunt. 1989. A user's guide to CERES Wheat V. 2.10. Michigan State
        University-IFDC-IBSNAT, Muscle Shoals, AL.

Manabe, S. and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in atmospheric
        carbon dioxide. Journal of Atmospheric Science, 44:1601-1613.

Ritchie J., U. Singh, D. Godwin, and L. Hunt. 1989. A user's guide to CERES Maize V. 2.10. Michigan State
        University-IFDC-IBSNAT, Muscle Shoals, AL.

Wilson,  C.A. and J.F.B. Mitchell. 1987. A doubled CO2 Climate Sensitivity Experiment with a Global Model
        Including a Simple Ocean. Journal of Geophysical Research, 92:13315-13343.
                                           FRANCE-9

-------
                                      FIGURES CAPTIONS
Figure 1.              Location  of test  sites   (Versailles  and  Avignon)  and  calibration  station
                       (Mons-en-Chauss6e) used for France.

Figure 2.              Sensitivity analysis of wheat yield to parameters P1D, Gl and G2. Mean values (solid
                       line) and confidence intervals (dotted lines) are computed for all available years in
                       each weather series.

Figure 3.              Changes in weather variables as simulated by GCMs (a,b: temperature; c,d: rainfall;
                       e,f: radiation).

Figure 4.              Yield variation  (in percent of lxCO2 baseline) and season  length  for wheat  in
                       Versailles  (squares  =  climate change  alone;  circles = climate  change with
                       physiological CO2 effects).

Figure 5.              Yield variation  (in percent of lxCO2 baseline) and season  length  for wheat  in
                       Avignon  (squares  =  climate change alone;  circles  =  climate  change with
                       physiological CO2 effects).

Figure 6.              Yield variation (in percent of lxCO2 baseline) and season length for rainfed maize
                       in Versailles (squares =  climate change alone; circles = climate change with
                       physiological CO2 effects).

Figure 7.              Yield variation  (in percent of lxCO2  baseline) for irrigated maize in Versailles
                       (squares = climate change alone; circles = climate change with physiological CO2
                       effects).

Figure 8.              Yield variation (in percent of lxCO2 baseline) and season length for irrigated maize
                       in Avignon  (squares = climate change  alone;  circles  =  climate  change with
                       physiological CO2 effects).

Figure 9.              Yield variation  (in percent of lxCO2 baseline)  for irrigated maize in  Avignon
                       (squares = climate change alone; circles = climate change with physiological CO2
                       effects).

Figure 10.      Wheat-yield response to sowing date  (in percent of lxCO2 baseline) for GISS transient
               scenarios in Versailles and Avignon.

Figure 11.      Maize-yield response to sowing date (in  percent  of lxCO2 baseline) for GISS transient
               scenarios in Versailles and Avignon.
                                            FRANCE-10

-------
Table 1.
Values of genetic coefficients used for wheat.
Genotype P1V P1D
Fidel

4.5 3.0
(0.5) (3.5)
Arminda 6.0 4.5

P1V
P1D
P5
Gl
G2
G3
(6.0) (3.5)
Vernalization coefficient
Photoperiodism coefficient
Grain filling duration coefficient
Kernel number coefficient
Kernel weight coefficient
Spike number coefficient
(Values recommended in DSSAT are
P5
4.5
(2.5)
4.5
(4.0)


shown between
Gl G2
4.5 1.5
(4.0) (3.0)
4.6 1.2
(4.0) (3.0)


brackets)
G3
2.5
(2.0)
1.7
(2.0)



                                        FRANCE-11

-------
Table 2.
Values of genetic coefficients used for maize.
  Genotype
     PI
P2
P5
Gl
G2
Dea

PI
P2
P5
Gl
G2
180
(130)
Juvenile phase coefficient
Photoperiodism coefficient
0.0
(0.2)


640 800
(680) (780)


7
(8)


Grain-filling duration coefficient
Kernel number coefficient
Kernel weight coefficient


(Values recommended in DSSAT are


shown between brackets)



                                        FRANCE-12

-------
Table 3.
Simulated total evapotranspiration for wheat (in percent base
             Scenario
                Climate Change Alone
Climate Change and
CO2 Physiol. Effects

BASE
GISS
GFDL
UKMO
2010
2030
2050

BASE
GISS
GFDL
UKMO
2010
2030
2050

BASE
GISS
GFDL
UKMO

0
-5
-16
-16
4
3
-1

0
-12
-12
-10
-6
-5
-10

0
-7
-9
-7
Versailles (irrigated)
-6
-12
-22
-16
1
-1
-7
Avignon (irrigated)
-8
-8
-18
-16
-8
-10
-16
Avignon (rainfed)
-4
-11
-13
-10
                                          FRANCE-13

-------
Table 4.
Simulated total evapotranspiration for maize (in percent of base
             Scenario
                Climate Change Alone
Climate Change and
CO2 Physiol. Effects

BASE
GISS
6FDL
&KMO
2010
2030
2050

BASE
GISS
GFDL
UKMO

BASE
GISS
GEDL
UKMO
2010
2030
2050

0
-11
-16
-11
0
-4
-6

0
-10
-16
-11

0
-8
-9
-7
-6
-8
-9
Versailles (rainfed)
-16
-24
-29
-25
-5
-12
-19
Versailles (irrigated)
-17
-25
-29
-26
Avignon (irrigated)
-16
-23
-23
-20
-11
-16
-21
                                          FRANCE-14

-------
Appendix A.    Sensitivity analysis of GERES-Wheat to climate and CO2 changes.
    CO,
Changes
 Precip.    Temp.
     VERSAILLES
Yield   Seas.L.     ET
Yield
 AVIGNON
Seas. L.        ET
(ppm) (%) (C)
330 0 +0
+2
+4
+20 +0
+2
+4
-20 +0
+2
+4
555 0 +0
+2
+4
+20 +0
+2
+4
-20 +0
+2
+4
(tfaa-1)
6.4
5.9
5.3
6.5
6.0
5.3
6.3
5.9
5.2
7.7
7.0
6.2
7.7
7.0
6.2
7.6
6.9
6.2
(days)
252
241
231
252
241
231
252
241
231
252
241
231
252
241
231
252
241
231
(mm)
483
460
437
491
466
442
470
449
429
454
431
410
458
435
412
445
424
405

6.32
6.3
5.8
6.8
6.6
5.9
5.5
5.8
5.5
7.9
7.7
6.9
8.2
7.9
7.0
7.1
7.3
6.7

216
205
194
216
205
194
216
205
194
216
205
194
216
205
194
216
205
194

421
396
375
440
410
387
394
375
357
403
376
357
418
387
366
380
361
343
                                       FRANCE-15

-------
Appendix B.    Sensitivity of CERES-Maize to climate and CO2 changes.
      CO,
Changes
 Precip.    Temp.
     VERSAILLES
Yield      Seas. L.
ET
       AVIGNON
Yield   Seas. L.      ET
(ppm) (%) (C)
330 0 +0
+2
+4
+20 +0
+2
+4
-20 +0
+2
+4
555 0 +0
+2
+4
+20 +0
+2
+4
-20 +0
+2
+4
(t ha'1)
8.4
9.3
8.6
8.5
9.3
8.6
9.4
9.1
8.5
10.0
9.9
9.1
9.0
9.9
9.1
9.0
9.9
9.1
(days)
159
128
108
159
128
108
159
128
108
159
128
108
159
128
108
159
128
108
(mm)
442
416
384
450
423
389
432
405
376
369
349
323
375
354
328
361
342
318

10.9
9.5
8.5
10.9
9.5
8.5
10.8
9.5
8.5
11.6
10.2
9.1
11.6
10.1
9.0
11.5
10.2
9.1

109
97
89
109
97
89
108
97
89
109
97
89
109
97
89
108
97
89

458
428
408
461
432
411
453
424
405
385
361
345
389
363
346
381
356
341
                                      FRANCE-16

-------
                                                OMons

                                                   Versailles
Figure 1.
Location  of  test  sites  (Versailles  and  Avignon)  and  calibration  station
(Mons-en-Chausse) used for France.

-------
                        AVIGNON cv. FIDEL
                                           VERSAILLES cv. ARMINDA
    o
    JC
                 10
                                   4
                                  P1D
                                                             101-
                                                              0
                                         2       34       5
                                                       P1 D
    I
    o
                 10
                                  4
                                  G1
                                                             10
                                        234
                                                       G1
    I
   a
   
-------
                        Versailles
                                            10     12
                - G1SS     GFDL   -a- UKMO
                - Trans 2010 -*- Trans 2030 -*- Trans 2050
               GISS    - GFDL   -*- UKMO
              -s- Trans 2010 -K- Trans 2030 -*r Trans 2050
                 -GISS    --GFDL    -x-UKMO   I
                 - Trans 2010 -*- Trans 2030 -*- Trans 2050
                                                      Avignon
                                               - QSS    - GFDL   -- UKMO
                                                Trans2010 -K- Trans2030 -*- Trars2050
                                               - GISS    - GFDL   -- UKMO
                                               - Trans2010-*- Trans2030-*- Trans2050
                                               - GISS    - GFDL   -*- UKtVO
                                               - Trans 2010 -x- Trans 2030 -*- Trans 2050
Rgure3.
Changes in weather variables as simulated by GCMs (a,b: temperature; c,d: rainfall;
e,f: radiation).

-------
         Versailles wheat (rainfed or irrigated)
         -5





         I3'





         I'I

          	0	2030
         4	O	\ 3" I
           30  -25  -20  -15   -10   -5  0   5   10   15
                          Yield change 06)
V(
6
ff
|5
& 4
1
I3
c 
s2
!'
jrsailles wheat (rainfed or irrigated)
 	 	 1 UK


 	 [5
" 	 [Ja
. nr
* 	 |2K
_ 	 _ 	 Jlo

220 225 230 235 240 245 250 L^
Season length (days)

JW3[
a]

SJ
*]
o]
o]
E]
Rgure4.
Yield variation (in percent of lxCO2 baseline)  and season  length for wheat in
Versailles  (squares  =  climate  change alone;  circles  = climate change  with
physiological CO2 effects).

-------
               Avignon wheat (irrigated)
          7-i
        I
                                               IUKMO
                                                GFDL
                                                GISS
                                                2050
                       2030|


         ~W	12010 I
          -30  -25  -20  -15  -10  -5   0   5   10   15
                         Yield change (%)
                                           Avignon wheat (rainfed)
arly
                                                                            ^
                                     04
                                     -25
                                         -20  -15  -10  -50   5
                                                     Yield change (%)
                                                                 10   15  20

o
o
CD

5
i.
S

0
f




Avignon wheat (rainfed or irrigated)
'! " 	 [U
5-,
!  [I


3'


i ' 
. 1 m 1 
11 	 [a
i
185 190 195 200 205 210 215 4=
Season length (days)


-------
               Versailles maize (rainfed)





	 J
"  Trim.. A


,....... ...... , ,^, 	
- 	 --O-

K. ., ....^~
i 	 ~ 	 r-m 	 . 	 . r
                                              UKMO!
                         5       10
                         rieU change {%)
6
&
 5
s
j
o 3-
a
S
i 2
E
ff
>
10
Versailles maize (rainfed)
 	  	 IUKMO |
" 	 	  	 IGFDL |
	 IGISS
	 J2U50 1
" "" 	 ' 	 ' 	 "" 	 |2030|
 	 J2010
0 110 120 130 140 150 160 I^EI
Season length (days)
Figure 6.
Yield variation (in percent of lxCO2 baseline) and season length for rainfed maize
in Versailles  (squares  =  climate change  alone;  circles  s  climate change  with
physiological CO2 effects).

-------
                                           Versailles maize (irrigated)
                                    _ 0


                                   Jh




                                    i 2
                                      0    2   4   6   8   10   12   14  16  18 -
                                                     Yield change (%)
Figure 7.
Yield variation (in percent of lxCO2 baseline)  for irrigated  maize in Versailles
(squares = climate change alone; circles = climate change with physiological CO2
effects).

-------
                 Avignon maize (rainfed)
                20   40  60   80  100   120   140  160
                           YieU change fr)
                                                             S6
                                                             s
                                                             i
                                                             s.
                                                             i3-
                                            Avignon maize (rainfed)
                                                                                               ^UKMO
                                                                           GFDLj

                                                                           6ISS!
                                                               80
                                                                      SS
                                                     90      95
                                                    Season length (days)
                                                                    100
RgureS.
Yield variation 'n percent of lxCO2 baseline) and season length for irrigated maize
in Avignon  (squares  =  climate change  alone;  circles = climate change with
physiological CO2 effects).

-------
        7

      6
      I3
             Avignon maize (irrigated)
 	  	
, 	  	 : 	
 	  	

 	 o 	 "

	 1



* 	 r^j
                                             GFDL|
                                             GISSl
        -30   -25   -20   -15   -10   -5
                        Yeti change <%)
Figure 9.
Yield variation (in percent of lxCO2 baseline) for irrigated maize in Avignon
(squares = climate change alone; circles  = climate change with physiological CO2
effects).

-------
                Versailles wheat (irn'gated)
I 5
I
I *
i
I 2
i
 1-
 0-
                                                  2010

                                                  2030

                                                  2050
Septl
                        Oct1     Ncw1
                           Sowing da!e
                                       Dec1
                                                             1 0-
                                                             t
                                                             I'21
                                                             o
                                                               -4-

                                                               -&
                                                   Avignon wheat (irrigated)
                                                                                     2010

                                                                                     2030
                                                                                     -*~
                                                                                     2050
                                                           Septl
OcM     Nov1
   Sowing date
                                                                                            Dec)
Figure 10.       Wheat-yield response to sowing date (in percent of lxCO2 baseline) for GISS transient
                 scenarios in Versailles and Avignon.

-------
  1&
  18

2 17
o
 16

I15'
  1+
I 13

-------

-------
 POTENTIAL EFFECTS OF GLOBAL WARMING AND
CARBON DIOXIDE ON WHEAT PRODUCTION IN THE
              FORMER SOVIET UNION
          Gennadiy V. Menzhulin and Larisa A. Koval

               Department of Climatic Changes
              State Hydrological Institute, Russia
                  Alexander L. Badenko

               Laboratory of Plant Biophysics
                Agrophysical Institute, Russia
                        FSU-1

-------
TABLE OF CONTENTS

SUMMARY
INTRODUCTION
       Background
       Crop Production
       Objectives
METHODS
       Crop Model
       GCM Climate Change Scenarios
       Transient Scenarios
INPUT INFORMATION
       Climate Data
       Soil Data
       Initial Soil Conditions
       Management Variables
       Calibration
       Validation
RESULTS AND DISCUSSION
       Sensitivity Analysis
       Description of the GCM Climate Change Scenarios
       Wheat Crop Changes
       Projection of National Wheat Yield
ADAPTATION
SOURCES OF UNCERTAINTY
       Models
       Input Parameters
       Technological Adjustments
       Variability
FUTURE RESEARCH NEEDS
REFERENCES
                                     FSU-2

-------
SUMMARY

        This study combines the CERES-Wheat crop growth model and climate change scenarios
to estimate the possible impacts of climate change on wheat production (the main food crop of the
country) in 19 sites in Russia and the former Soviet Republics. This is the first crop modeling study
in the former Soviet Union (FSU) that uses future climate change scenarios generated by General
Circulation Models (GCMs). The results will make a significant contribution to the understanding
of possible yield changes under future climate.

1.      The sensitivity of the crop model to arbitrary incremental temperature and precipitation
        changes was estimated in two regions. A 2C increase in temperature decreased yields by
        an average of  approximately 20%,  and a temperature increase of 4C resulted in yield
        decreases of more than 30%. These yield decreases probably result from a shortening of the
        wheat season under a higher  temperature.  Changes in the precipitation  by  20% had a
        smaller impact on simulated wheat yields than changes in temperature.

2.      GCM climate  change alone resulted in considerable regional differences  in  simulated
        changes in wheat yields compared to baseline yields. The simulation study considered the
        beneficial physiological effects of CO2 on wheat yields. Under the GISS scenario, spring and
        winter wheat yield increases exceeding 50% were obtained in the northwestern, central, and
        eastern regions of Kazakhastan. These regions are presently fairly arid, but with the increase
        in precipitation in the climate change scenarios they are predicted to become very favorable
        to wheat production. In contrast, under the GFDL scenario, wheat  yields were not favored
        in these regions. In general, the climate change scenarios used were more  favorable for
        winter than for spring wheat growth and production.

4.      The GCMs forecast significant changes in temperature. These  changes would result in the
        contraction of  the crop growing season and the cold period of the year,  and could  be of
        great importance for agriculture in Russia and the republics of the FSU.  In  some regions
        these changes would raise the possibility of new cropping techniques, especially the ability
        to grow a second crop during the calendar year. The potential decrease in duration of the
        cold period is particularly significant in the case of vegetable crops which currently have a
        short growing season.

5.      The results of  the simulated yield changes for each site were aggregated to estimate the
        possible impact of climate change on national wheat yield. Under  the GISS scenario and
        considering the direct physiological effects of CO2 on crop growth and water use, winter and
        spring wheat yields increased by 41%  and 21% respectively. The GFDL and UKMO
        scenarios were significantly less favorable for the simulated wheat production. The largest
        yield decreases19%  -were simulated under the UKMO scenario.

6.      Changes in sowing date and irrigation were studied as possible adaptation strategies to
        climate change in two regions. While small changes in the sowing  date did not  appear to
        have a large impact on yields, irrigation had a dramatic effect  on spring wheat yields.
                                          FSU-3

-------
INTRODUCTION

Background

        Russia and the former Soviet Republics have the world's largest area of arable land.
Agriculture has been predominant throughout all periods of the country's history, and there is
important ongoing research on the impacts of climate on crop yields (Golzberg 1967; Mishchenko
1984; Shashko 1985). Scientists in Russia and the former Soviet Republics have developed several
empirical statistical  models  for  forecasting crop  yields in different regions  of the  country.
Publications of the Agrometeorological Forecast Department of the Moscow Hydrometeorological
Centre (Ulanova 1984) illustrate the developments in this field. Recently, scientists have developed
prognostic models with a physical and physiological  basis.  For example, the St. Petersburg
Agrophysical Institute has developed a special Supporting Agrotechnological Decisions System for
the study of prognostic crop models (The Modeling...., 1982).
        In the last fifteen years, a number of important studies have been done in the former Soviet
Union related to the possible impacts of climate change on agricultural systems and crop production
(Zhukovskiy and Belchenko 1988, Zhukovskiy et al.  1992, Sirotenko et al. 1984). These studies did
not include the physiological (direct) effects of increased CO2 on crop yield.
        Menzhulin (1976 a,b, 1984), Koval and Sawateyev (1982), and Menzhulin and Nikolayev
(1987) have developed crop models that consider both the climate impacts and the statistical analysis
of annual variation, and can be  combined with different  climate and  technological scenarios to
evaluate the CO2 physiological effects on yield.  These crop models have been used to study the
impact of climate on wheat yield in Russia and the  former Soviet Republics, North America, and
Europe (Menzhulin and Sawateyev 1980, 1981; Menzhulin et al.  1987 a,b; Anthropogenic... 1987;
and Prospects... 1990; Rosenzweig et al. 1993).
        Historically, Russian scientists and their colleagues have not used climate change scenarios
generated by the General Circulation Models (GCMs) to study the possible impact of climate change
on agriculture. The climate change scenarios were paleoclimatic analogues  of global warming: up
to 1C  (year 2000)  used the Holocene Optimum; up to 2C (year 2025) used the  Eemian
Interglaciar; and up  to 4C (year  2050)  used the Pliocene  Optimum. The  carbon dioxide
concentrations corresponding to  these three periods would be  380 ppm, 420 ppm, and 560 ppm,
respectively.

Crop Production

        Today, agriculture presents a complex  set of challenges in the  former  Soviet Union.
Agricultural and economic policies and the monopoly of the agricultural industry by the State have
led agriculture to its current crisis. These past policies have resulted in a country that cannot provide
for itself with adequate agricultural production. The low effectiveness of Soviet agriculture is evident
by the following statistics. In 1991, more than 22 million people were involved in the agricultural
sector of the former USSR, which is more than in any country in the Organization for Economic
Co-operation and Development (OECD) comprised of the countries of Europe, North America,
Australia, New Zealand, and Japan. Yet agricultural production in the former Soviet Union is five
times smaller than that of the OECD as a whole. The efficiency of Soviet agriculture is ten times
lower than that of the USA, the Netherlands, Canada, and Belgium.
        The general agricultural crisis is particularly  manifested in grain production. Average yields
of winter and spring wheat, the main  components of national grain production, increased slightly

                                         FSU-4

-------
before the 1970s, but leveled off during the last two decades. Although Russia holds a leading
position in the total amount of wheat production, it ranks 35th in yield level among the 45 wheat-
producing countries.

Objectives

       The actual crisis in Soviet agriculture is sufficient cause for alarm, but the alarm is greater
if the problem has to be solved during a period of possible climate change. This is the first crop-
modeling study in the FSU that uses future climate change scenarios generated  by GCMs. The
results will make a significant contribution to the understanding of possible yield changes under
future climate.
       An analysis of Soviet agricultural production (Appendix 1) suggests that the best crop for
the simulation study is wheat. Wheat (winter and spring) is the main food crop in Russia and the
other republics. Its mean annual production is 100 million t. Other crops such as sugarbeet, potato,
and barley are also major components of national agricultural production, and subsequent studies
of the impacts of climate change on their production would be valuable. The production of wheat
in Russia and the Soviet Republics has been described and analyzed in more detail than that of any
other crop, providing empirical information for the validation of the simulation model.
        Simulations of wheat yields and production changes require meteorological, soil, and crop
management information  about the  zone  of production.  We chose 113 administrative regions
(oblasts) in the main grain-producing area of the country (Figure 1 and Appendix 4) that include
65 regions of Russia, 25 regions of Ukraine, 6 regions of Belarus, 13 regions of Kazakhastan, the
Baltic Republics, and Moldova. The Republics of the Caucasus and Middle Asia, and the regions
of East Siberia and Russian Far East were not included because their wheat production does not
significantly contribute to the total national wheat production.

METHODS

Crop Model

       The simulation model used was the CERES-Wheat v2.10 model (Ritchie and Otter 1985)
developed by IBSNAT (International Benchmark Sites Agrotechnology Transfer). The model has
been  calibrated and validated over  a wide range of geographical locations, and thus the results of
the simulation are comparable to previous and current studies in other regions of the world.

GCM Climate Change Scenarios

       We used the scenarios of future climate generated by three equilibrium GCMs: the Goddard
Institute for Space Studies (GISS) (Hansen et al. 1983), the Geophysical Fluid Dynamics Laboratory
of  Princeton University (GFDL)  (Manabe and Wetherald 1987), and  the  United Kingdom
Meteorological Office (UKMO) (Wilson and Mitchell 1987). The climate scenarios  provided by
these GCMs correspond to the global warming resulting from an increase of the greenhouse gas
content in the atmosphere equivalent to the doubling of CO2 concentration. It is projected that CO2
would account for 83.3% (550 ppm) of the equivalent concentration (660 ppm) and that other trace
gases (methane, nitrogen oxides, etc.) would account for 16.7%.
                                          FSU-5

-------
         The  local climate scenarios were  developed from  the GCM  output (temperature,
 precipitation, and solar radiation) that is associated with a particular grid network for each GCM.
 We developed the local climate change scenarios by two methods.
         Simplified method. Local scenarios for each site were created by the direct extrapolation
 (attribution) of the changes in climate variables at the nearest gridpoint. This method has been used
 in previous climate change impact studies (Smith and Tirpak 1989), and therefore the results of
 simulation under such scenarios can  be compared to other studies. We created scenarios for 19
 selected sites with this method (Appendix 2).
         Multi-point interpolation. We developed this more complex method to create climate change
 scenarios that simulate local climate  more accurately (Appendix 3). This method used the GCM
 output of changes in  temperature, precipitation, and solar radiation occurring in several of the
 nearest gridpoints to each site. This  method was applied to generate scenarios in all of the 113
 selected regions. With this complete set of scenarios we estimated the national changes in spring and
 winter wheat production.

 Transient scenarios

         In addition to the equilibrium GCM scenarios, we used scenarios generated by the GISS
 transient model for the decades of the 2010s, 2030s, and 2050s (Hansen et al. 1988). We calculated
 the local changes of climate variables in 113 selected regions (as described above) and applied them
 to the crop model.

 INPUT INFORMATION
 Climate Data

        Ten years (1970-79) of daily maximum and minimum temperatures and precipitation data
 were obtained from the Union  Research Institute of Meteorological  Information (World Data
 Center "B" in Obninsk, Moscow Region) at 51 locations. When necessary, extrapolation of data
 based on geographical proximity and environmental and climatic conditions was done to create data
 for the entire set of 113 selected locations included in this study.
        Daily solar radiation values were generated using long-term monthly radiation normals
 (obtained from the Solar Radiation Data Bank in the Geophysical Observatory (St. Petersburg) and
 the solar radiation generator included in the CERES-Wheat v2.10 model. Appendix 2 shows the
 stations selected to create scenarios for the 19 representative sites discussed in this report.

 Soil Data

        Soil data were obtained from the largest soil information archive at the Dokuchayev Soil
 Scientific  Museum, with the help of  the Soil Departments at the St. Petersburg and  Moscow
 Universities. All the information in these archives is uniformly systematized (Ivanova and Rozov
 1967, Glazovskaya 1966). The large territory included in this study and the wide variety of soils
within each region required careful study for the selection of the representative soils  and the
determination of the input parameters for the simulation (Aderikhin 1964; Kovda and Lobova 1964;
Liverovskiy  1965;  Dobrovolskiy  1968; Glazovskaya and Friedland 1980; and Kovda 1973). We
selected 26 soil types  in the wheat-producing  region. For a complete representation  of grain
production in each region, it was necessary to include more than one soil in each region (Table 1).

                                         FSU-6

-------
Initial Soil Conditions

        Initial soil moisture in the profile, acidity, and the content of nitrates and ammonium must
be defined for the simulation model. Unfortunately, there is not enough reliable annual information
on initial soil moisture and nitrogen content for each region. Therefore, we used the following
standard values. For arid regions, the soil water content at the end of the previous season was
assumed to be equal to the lower limit (LOL) of water availability to the plant. In the regions where
precipitation is always sufficient to fill the soil water store to the drained upper limit (DUL), the
initial value was taken as equal to DUL just before sowing (Ritchie et al. 1989). Although there is
nitrogen stress in some areas, we assumed that the nitrogen was not limiting.

Management Variables

        For spring wheat, the simulated planting date occurs when the average temperature is about
5C. For winter wheat, the simulated planting date occurs in autumn, about 60 days before the mean
temperature reaches 0C.

Calibration

        The CERES-Wheat model has been calibrated and validated extensively with a number of
wheat cultivars (Ritchie and Otter 1985). In this study, however, we used specific cultivars that were
representative of production in Russia and the other republics,  and that were not previously
included in the CERES model. Therefore, we had to determine the genetic coefficients and calibrate
the model for the cultivars used in the FSU. (The Manual...  1982).
        According to specifications developed by the National Seed Inspection, all varieties of wheat
cultivars recommended for use in national grain production are divided into eight adaptation groups:
East European  Forest; East  European Forest-Steppe; East European Southern Steppe;  Volga
Steppe; West-Siberian Forest Steppe; East Siberian Forest Steppe; East Siberian Forest; and Middle
Asian Rainfed.
        The genetic coefficients were determined  based  on experimental data provided  by  the
Agrophysical Research Institute. We calibrated the model with  Russian  cultivars so that  the
simulated  values correspond  effectively to  the experimental observations on  duration and
phenophases, weight and number of grains in the ear, growth of phytomass and leaf area and yield.
        Table 1 shows the adaptation group number (1 to 8) and the wheat cultivars selected for
the 19 regions. The actual wheat production in any region of the FSU involves  more than one
cultivar. The number of cultivars  used and their relative contribution to the  total regional
production  is subject to yearly changes and the introduction of new cultivars.  We used a set of
nonvariable genetic parameters that represents only one cultivar and therefore we do not completely
represent current practices. This assumption  could introduce possible  error into  the results that
could be significant because we have simulated only ten years of wheat production.

Validation

        The model  was  validated for  the  eight  adaptation groups by comparing  observed
experimental values of yields from the National Bureau of Agricultural Statistics to simulated values.
                                          FSU-7

-------
 RESULTS AND DISCUSSION

 Sensitivity Analysis

         We estimated the sensitivity of the CERES-Wheat model to step changes in temperature
 and precipitation. The mean daily temperature was increased by 2C and 4C, and the precipitation
 was changed by +20% and by -20%. All possible combinations were evaluated. The physiological
 effects of increased CO2 were also considered in each scenario. The sensitivity study was simulated
 for two  important wheat producing regions, Zhitomir and North Bashkiria (for both spring and
 winter wheat).
         Table 2 shows the results of simulations conducted with climate change alone and with
 climate change and physiological effects of CO2. The results show large decreases in simulated yields
 due to changes in temperature. In the case of climate change alone, a 2C increase in temperature
 decreased yields by approximately 20%; a temperature increase of 4C resulted in yield decreases
 of more than 30%  (the  largest decrease  was  45%  for winter wheat in North Bashkiria). The
 physiological effects of CO2 partially compensated for the adverse impact of temperature increases
 in simulated yields. For example, for spring wheat in North Bashkiria, a 4C increase produced a
 40% decrease in yield for temperature change alone (330 ppm) and a 17% decrease for temperature
 change with physiological effects of CO2 (550 ppm).  The impact of changes in precipitation on
 winter and spring wheat yields was noticeably smaller than the impact of changes in temperature.
        The simulated yield increases due to the beneficial direct effects of  CO2 were close  in
 magnitude to the yield decreases produced by a  +2C temperature increase. In all cases, when
 considering the physiological CO2 effects, the simulated yields of winter and spring wheats increased
 more than 20% compared to the baseline yields with a maximum of 35% in the case of winter wheat
 in North Bashkiria.
        In the CERES-Wheat model, the phenological stages are influenced by temperature, and
 our results show that the length of the growing season was not affected by changes in precipitation.
 In addition, the physiological CO2 effects did not have any influence on the length of the growing
 period in this case. A temperature increase of 2C resulted in a decrease in the length of the spring
 wheat growth period by 8 days  in both Zhitomir and North Bashkiria. With temperature increases
 of 4C, the decrease was 14  and 15 days for spring wheat and 10 and 18 days  for winter wheat  in
 Zhitomir and North Bashkiria  respectively.
        The most significant evapotranspiration  (ET) change (-18%)  was for spring wheat  in
 Zhitomir under the warmest and driest scenario (temperature +4C and precipitation -20%). ET
 was not significantly affected by the  simulated physiological effects of CO2. Under the baseline
 climate, ET was reduced only 6% due to the physiological CO2 effects. It is important to note that
 this estimate is dramatically smaller than estimates from experimental data (Rose 1989).

 Description of the GCM Climate Scenarios

       Tables 3,4, and 5 show the seasonal and annual changes in temperature, precipitation, and
solar radiation under the  three GCM scenarios for the selected sites. The temperature changes
 (Table 3) simulated by the GISS and GFDL scenarios are comparable in all sites and average 4 -
5C higher than current temperatures. The changes under the UKMO scenario are almost twice as
large as those simulated by the other two scenarios. In most cases, the differences between scenario
and baseline temperatures are more noticeable in the cold season.
                                         FSU-8

-------
        The average annual and seasonal precipitation under the climate change scenarios is
significantly greater than the baseline precipitation in most cases (Table 4). In general, the largest
amounts of precipitation are projected by the GISS and UKMO scenarios.
        Table 5 shows changes in solar radiation under the three scenarios. There is a general
tendency for increased solar radiation under the GFDL and UKMO scenarios.  Under the GISS
scenario, solar radiation decreases slightly  in most sites.

Wheat Crop Changes

        The simulated changes in yield, season length, and evapotranspiration under GCM climate
change  scenarios are given in  Tables 6,  7, and 8 for the 19 representative sites of the grain
production area of Russia, the Ukraine, Belarus, and Kazakhstan.
        Yield.  Climate change alone  resulted in  considerable regional differences in simulated
changes in wheat yield under the GCM scenarios. The maximum increases in yield were projected
under the GISS  scenario.       The  following results  are for climate change including  the
physiological effects of CO2.
        Under the GISS scenario, increases in spring wheat yields exceeding 50% were obtained in
the northwestern, central, and eastern regions of Kazakhastan (Aktyubinsk, Zelinograd, Karaganda).
In the same regions, the changes in winter wheat yields were also characteristically high. These
regions are fairly arid, but with the increase in precipitation under the climate change scenarios,
were very  favorable to  wheat production. The GFDL scenario was not as favorable for wheat
production in  the regions listed  above.  Under  this scenario,  the decreases in yields were
characteristic of the sites in Kazakhstan, as in most of the other sites.
        A systematized  interpretation of the meaning of climate change for wheat production is
complicated because of the considerable differences in yield changes estimated under the three GCM
scenarios. Yet we can conclude that,  on average, the climate change scenarios used were more
favorable for winter than for spring wheat yields.
        Length of the growing season.    To analyze the possible impacts of global climate change
on agriculture in the FSU, it is important to consider the changes in the lengths of the growing
period and the  warm seasons under the climate  scenarios (Table 7). All three GCM scenarios
resulted in a shortening of the growing season for both winter and spring wheat at all sites. Under
the UKMO scenario, which is characterized by the highest temperature increases, the growing season
in the Baltic regions for winter wheat decreased by more than two months. For spring wheat the
decrease was about one month. The shortening of the simulated growing season for winter and
spring wheats occurred in all regions of the FSU.
        The GCMs forecast significant changes in the temperature regime, which would result in
the contraction of the crop-growing season and of the cold period of the year, and could be of great
importance for agriculture in Russia and the republics of the former Soviet Union. In some  regions
these changes would raise the possibility of new cropping techniques, especially the growth of a
second crop during the calendar year. The decrease in  the duration of the cold period is especially
significant in the case of vegetable crops, which have currently have a short growing season in many
regions.
        Evapotranspiration. The evapotranspiration changes under the GCM scenarios were generally
insignificant (Table 8). In most cases, the total crop ET compared to baseline data. We believe that
the CERES-Wheat model  needs to be improved in order to simulate the crop  water  regime
accurately. In the simulation of evapotranspiration of the wheat crop, it is necessary to account for
                                          FSU-9

-------
 the interactions among leaf temperature, moisture level, and the physiological effects of CO2 that
 increase stomatal resistance (Rose 1989).

 Projections of National Wheat Yields

        The national production changes for equilibrium and transient scenarios were estimated
 using the 16-point interpolation scheme for all 113 sites,  not the nearest gridpoint extrapolation
 method.
        Aggregation of the GCM-scenario results.  The possible impact of climate change on wheat
 production in Russia, the Ukraine, Belarus, and Kazakhstan can be estimated from the simulated
 changes in yield under the climate change scenarios. Table 9 shows  estimates in yield changes
 calculated from the results of 113 subregions of the grain production  zone. Under the GISS
 scenarios and considering the physiological effects of CO2, winter wheat production increased by
 41%  and spring wheat production increased by 21%. The GFDL and  UKMO scenarios were
 significantly less favorable for the simulated wheat production. Under the UKMO scenario and
 considering the physiological effects of CO2, the simulated wheat production decreased by 19%.
        Aggregation of the transient scenario results. Wheat yield changes were also simulated under
 the GISS transient scenarios. The first two transient scenarios correspond to the intermediate levels
 of climate change, which are simulated for the decades of the 2010s and 2030s, with a corresponding
 increase in greenhouse gas concentration in the atmosphere equivalent to  405ppm and 430ppm of
 CO2.  The third GISS transient scenario can be used as another alternative  scenario of climatic
 change for the decade of the 2050s.
        Winter and spring wheat yields were simulated under the three GISS transient scenarios for
 the 113 selected regions. Table 9 shows the results of the aggregation of the yield changes. The
 results suggest that the production of both winter and spring wheat would gradually increase due
 to climate change.

 ADAPTATION

        A realistic approach to  the problem  of climate change impacts  on  agriculture requires
 studying possible measures to avoid unfavorable consequences. A detailed investigation  of this
 question was beyond the scope of this study, but we analyzed some possible adaptation  responses,
 such as changes in the sowing date and in irrigation on spring wheat yields at Zhitomir and Saratov
 regions (Table 10). The adaptation responses were analyzed under the UKMO scenario  because it
 is the most unfavorable scenario. This small adaptation study shows approaches and methods that
 can be  and should be  applied to a more detailed study of adaptation responses to unfavorable
 climate changes in agriculture in Russia and in the former Soviet Republics.
        The most significant result of the analysis was that changes in the  sowing date (both later
 and earlier) by 15 days and even by 30 days did not noticeably affect crop yields. Under the UKMO
scenario, there is a considerable precipitation increase in the winter and  spring seasons at these two
sites (Table 4). However, these precipitation changes did  not fully compensate for the moisture
 limitation of wheat in the summer season due to a temperature increase. In  addition, in the Saratov
region, the UKMO scenario predicts a 24% decrease in the summer precipitation, which is already
limited. Irrigation could have a large impact on spring wheat yields in these two regions, especially
in Saratov.

SOURCES OF UNCERTAINTY
                                         FSU-10

-------
Models
        The use of the simulation crop model (CERES-Wheat) and the climate change scenarios
(GISS, GFDL, and UKMO) introduce uncertainties. In principle, we could quantify the uncertainty
associated with the use of GCMs in the agricultural zones, applying the methods proposed in Global
Comparison of Selected  GCM: Control Runs and Observed Data (Kalkstein 1991).

Input Parameters

        Here we include uncertainties derived from the limited availability of current information
on input parameters of the models. As discussed above, the possible error in the calculations of the
climate scenarios could  be smaller when we use the procedure for daily climate data  generation
described in Appendix 3. The standard procedure used to  develop climate scenarios for important
agricultural sites  that are situated far from  the GCM gridpoints could be incorrect. With the
multipoint  interpolation system we can simulate more accurately the current climate than when we
use the GCM output from the nearest gridpoint. Therefore, additional uncertainty may be generated
when using the conventional system to create local scenarios.
        Another potential error is associated with the large size and climatic diversity of some of
the study regions that would require data from more  than one station to represent their climate
more accurately.

Technological Adjustments

        There  are  unavoidable  uncertainties in estimating the future changes of agricultural
technology and management in response to climatic change. These are significant factors limiting
the reliability of the results. Climate change can alter the geography of the current regions and
influence soil processes, and will eventually create demand for newly developed technologies.

Variability

        Another major factor that limits the reliabiity  of these results is the direct extrapolation to
the future of the current variation oif daily temperatures and precipitation. In order to obtain more
reliable estimates of future  changes in  wheat yields, it would be necessary to consider potential
differences in climate variability.

FUTURE RESEARCH NEEDS

        The present study is a contribution to the understanding of the possible impacts of climate
change on  the FSU's agricultural systems. However, future research is still needed. With improved
models and a more accurate set of input parameters, some of the uncertainties could be eliminated.
        An  adaptation  study with a  larger   scope-one  that  would  consider  the complex
agrotechnological decisions  that  might  decrease the negative consequences of climate change on
wheat productionis essential for continued progress in this field.  In  addition, further research
should consider the impact of climate change on the production of other major crops.  Finally, the
frequency  of extreme climate  events, especially the occurrence of drought periods, needs to be
investigated.
        The impact of drought variability on agriculture  was addressed during the preparation of
the Intergovernmental  Panel on Climate Change  (IPCC) Working Group II report. However,

                                          FSU-11

-------
methods for estimating such variability were not clearly established within the framework of the
IPCC or other projects. It will require considerable work to compile and analyze GCM output and
to prepare appropriate input data, but such an effort is justified by the importance of potential
impacts of future climate variability in the FSU and other agriculturally important regions of the
world.
                                        FSU-12

-------
REFERENCES

Anthropogenic  Climatic  Changes.   1987.  Eds: M.I.  Budyko and Yu.A.  Izrael.  Leningrad,
        Gidrometeoizdat. 406 p. (Russian). English Translation: Arizona University Press, 1990.

Aderikhin, P.G. 1964. Chernozems of the USSR. In: Genesis, Classification and Cartography of Soviet
        Soils: Proceeding of the 8th International Congress of the Soil Science. Moscow, Nauka Publ.
        House. (Russian).

Dobrovolskiy, V.V. 1968. Soil Geography. Moscow,  High Education Publ. House. 350 p. (Russian).

Glazovskaya, M.A. 1966. The Principles of the Soil Classification. Soviet Soil Science Journal, Ne 8.
        p. 14-25. (Russian).

Glazovskaya, M.A., and V.M. Friedland 1980. The Map of the USSR Soils (1:4000000).  Moscow,
        Geodesic and Geographic Publ. House. (Russian).

Golzberg, LA. 1967. Agroclimatological and Microclimatological Research in the Soviet Union.
        Transactions of the Main Geophysical Observatory, vol. 218. p.75-88. (Russian).

Hansen, J., G. Russell,  D. Rind, P. Stone, A. Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983.
        Efficient Three-Dimensional Global Models for Climate Studies: Model 1 and 2. April
        Monthly Weather Review, Vol. Ill, Na 4: 609-662.

Hansen J., I.  Fung, A Lacis, D. Rind, S. Lebedeff, R. Ruedy, and G. Russell. 1988. Global Climate
        Changes as Forecasted by Goddard Institute for Space Studies Three- Dimensional Model.
        Journal of Geophysics Research, 92: 9341-9364.

Ivanova, E.N. and N.N. Rozov 1967. The Classification of the Soils in the USSR. Soviet Soil Science
        Journal, N2 2. p. 34-43.

Kalkstein. L.A.  (ed.) 1991.  Global Comparisons of Selected  GCM: Control Runs and  Observed
        Climatic Data. Centre of Climatic Research University of Delaware. Report Prepared for
        US EPA, Office of Policy,  Planning, and Evolution. Climate Change Division. 251 p.

Koval, L.A., and C.P. Sawateyev. 1982. The Using of Crop Productivity Models for Estimations of
        Climatic Changes Influence on the Crops Productivity. In: The Problems of Land Hydrology,
        eds. I.V. Popov and S.A. Kondratyev. Leningrad, Hydrology and Meteorology Publ. House.
        p. 215-233. (Russian).

Kovda, V.A., 1973. The Fundamentals of Soil Science. Moscow, Nauka Publishing Co., Book 1: 380
        p., Book 2: 467 p. (Russian).

Kovda, V.A., and E.V. Lobova. 1964. The Soils Map of Asia. Moscow, Soviet Academy of Science
        Publ. House. (Russian).
                                         FSU-13

-------
 Liverovskiy, Yu.A. Soils of the Soviet Union. Transactions of Moscow State University, Vol. 1, N2 2.
        p. 51-121. (Russian).

 Manabe, S., and R.T. Wetherald. 1987. Large-Scale Changes in Soil Wetness Induced by an Increase
        in Carbon Dioxide. Journal of Atmospheric Sciences, 44: 1211-1235.

 The Manual on Selection of Wheat Cultivars for USSR Cereal Regions.  1982. Moscow, Agricultural
        Ministry Publ. House. 86 p.

 Menzhulin, G.V. 1976a. Biosphere Aspects of Climate Modeling: Climate Variations Effect on the
        Crops Productivity. Papers ofSov. Amer.  Symp., Tashkent. 12 p.

 Menzhulin, G.V. 1976b. The Influence of Climatic Changes on Crop Plant Productivity. Transactions
        of the State Hydrological Institute, Vol. 365. p. 41-48. (Russian).

 Menzhulin, G.V.  1984. The  Influence of Contemporary Climate Changes and CO2 Content
        Evaluation on the Crop Productivity. Soviet Meteorology and Hydrology, Nfi 4. p. 95-101.

 Menzhulin, G.V., and M.V. Nikolayev. 1987. Methods for Estimating the Year-to-year Variability
        of Cereals Crop Yields. Transactions of the State Hydrological Institute, Vol. 327. p. 113-131.
        (Russian).

 Menzhulin, G.V., and S.P. Sawateyev. 1980. The Impact of Expected Climate Changes on Crop
        Productivity. In: The Problems of Atmospheric Carbon Dioxide, ed. M.I. Budyko. Transactions
        of Sov. Amer. Symp., Leningrad, Gidrometeoizdat. p. 186-197. (Russian).

 Menzhulin, G.V., and S.P. Sawateyev. 1981. The  Present-day Climatic Changes and the Problems
        of Crop Productivity. Transactions of the State Hydrological Institute. Vol. 271. p. 90-103.
        (Russian).

 Menzhulin, G.V., L.A. Koval, M.V. Nikolayev,  and S.P.  Sawateyev. 1987a.The Estimating of
        Climatic Changes Agroclimatic Consequences: Scenario for Northern America. Transactions
        of the State Hydrological Institute, Vol. 327. p. 132-146. (Russian).

 Menzhulin, G.V.,  L.A.  Koval, and M.V. Nikolayev. 1987b. The Agroclimatic Consequences of
        Present Day Global Climate Changes. In: The Problems of Agroclimatic Providing for  the
        USSR Foodstuff Programme, ed.  I.G. Gringhof. Leningrad, Gidrometeoizdat.  p.  72-81.
        (Russian).

Mishchenko, Z-A. 1984. The Bioclimate of the Day and the Night. Leningrad, Gidrometeoizdat. 280
        p. (Russian).

The Modeling of Agroecosystems Productivity. 1982. Eds. N.F. Bondarenko, N.G. Zhukovskiy, N.G.
        Mushkin, S.V. Nerpin, R.A. Poluektov, and I.B. Uskov. Leningrad, Gidrometeoizdat. 264
        p. (Russian).
                                         FSU-14

-------
Prospects of Future Climate. 1990. Eds. M. MacCracken, AD. Hecht, M.I. Budyko, Yu.A Izrael. A
        Special US/USSR Report on Climate and Climate Change, Lewis Publishers. 270 p.

Rose,  E.  1989.  Direct (Physiological)  Effects  of Increasing CO2 on Crop Plants  and their
        Interactions with Indirect (Climatic) Effects. In: The Potential Effects of Global Climate
        Change on the United States. Appendix C, Agriculture, Vol. 2, eds. J.B. Smith, D.A. Tirpak.
        Office of Policy, Planning and Evaluation US EPA, Washington, DC.

Ritchie, J.T., and S. Otter. 1985. Description and Performance of CERES-Wheat: A User-oriented
        Wheat Yield Model.  In: Willis W.O., ed. ARS Wheat Yield Project. Washington D.C.: US
        DOA, Agricultural Research Service. Ars-38. p. 159-175.

Ritchie, J.T., D.C. Godwin and U. Singh. 1989. Soil and Weather Inputs for the IBSNAT Crop
        Models. In: IBSNAT Symposium Proceedings: 81st Annual Meeting of the American Society
        of Agronomy, Las Vegas, Nevada, October 1989. p. 31-45.

Rosenzweig, C, G. Menzhulin, and L Koval.  1993.  Climate Change and Agriculture in the former
        Soviet Union. In: Science, Agriculture and Environment in the FOrmer Soviet Union, M.S.
        Strauss and S.M.B. Thomson, eds. AAAS,  Washington, DC.

Shashko,  D.I. 1985. Agroclimatic Resources of the USSR. Leningrad, Gidrometeoizdat.  247 p.
        (Russian).

Sirotenko, O.D., E.V. Abashina, and V.N. Pavlova. 1984. The Estimates of Climate Variations of
        Crops Productivity. The Proceedings of the USSR Academy of Science: Physics of Atmosphere
        and Ocean, Vol. 20, N 11. p. 1104-1110. (Russian).

Smith, J.B. and  D.A. Tirpak.  1989. The Potential Effects of Global Climate Change on the  United
        States. US Environmental Protection Agency, Washington, DC.

Ulanova, E.S. 1984. Meteorological Factors and the Cereal Crops Productivity, p. 95-100. (Russian).

Wilson, C.A., and J.RB. Mitchell. 1987.  A Doubled-CO2 Climate Sensitivity Experiment with a
        Global Model Including a Simple Ocean. Journal of Geophysical Research, 92:13315-13343.

Zhukovskiy, E.E., and G.G. Belchenko. 1988. The Stochastic System of Crop Yields Forecasting.
        Transactions of the Agrophysical Institute, Vol. 70. p. 2-7. (Russian).

Zhukovskiy, E.E., G.G. Belchenko, and T.N. Brunova.  1992. The Stochastic Analysis the Climate
        Change Influence on  The Crop Productivity Potential. Meteorology and Hydrology Journal,
        Vol.1, (in press)  (Russian).
                                         FSU-15

-------
APPENDIX 1.       Production (x 100,000 t)/Area (x 100,000 ha) of the principal food crops.

Crops
W.Wheat
S. Wheat
Sugar Beet
Potatoes
Barley
Maize
Vegetables
Oats
Rye
Beans
Sunflower
Millet
Buckwheat

1940
145/143
172/260
180/12
759/77
121/105
51/36
137/15
168/202
210/231
15/24
26/35
44/60
13/21

1950
114/125
197/260
208/13
886/85
64/81
66/48
93/13
130/162
180/236
12/20
18/36
17/38
13/30
Years
1960
182/121
461/483
577/30
844/91
160/110
98/51
166/15
120/128
182/162
27/33
40/
32/38
6/14

1970
427/185
576/467
789/34
968/81
383/213
94/34
212/15
142/92
130/100
76/51
61/48
21/27
11/19

1987
462/153
371/314
904/34
760/62
584/306
323/46
292/17
219/118
181/97
100/64
61/42
39/28
13/16
                                      FSU-16

-------
APPENDIX 2.
Meteorological and solar radiation stations used for the 19 representative
sites.
No.
1
2
3
4
5
6

7

8

9

10

11

12

13
14

15
16
17
18
19
Region
W Russia
C Chernozem of
Russia
Central Russia
Middle Volga
South Volga
Southern Russia

East Euro
Russia
Western
Ukraine
Northwestern
Ukraine
Eastern Ukraine

Southern
Ukraine
Moldova

Western Belarus
Central Belarus

Baltic
NW Kazakhstan
N Kazakhstan
C Kazakhstan
E Kazakhstan
Site
(Lat/Long)
Bryansk
52.9/33.4
Kursk
51.7/36.0
Yaroslavl
57.6/38.9
Saratov
51.6/46.6
Astrakhan
46.6/48.0
Rostov-Don
47.4/40.8
N Bashkiria
55.0/55.6
LVov
49.5/23.6
Zhitomir
50.0/28.4
Kharkov
49.5/36.3
Kherson
46.7/33.5
Kishinev
47.0/28.9
Brest
52.0/24.2
Minsk
53.3/26.0
Lithuania
55.0/23.0
Aktyubinsk
49.9/57.9
N Kazakhstan
54.6/68.3
Zelinograd
51.6/70.1
Karaganda
50.0/72.4
Meteorological
Station
(Lat/Long)
Smolensk
54.7/32.0
Kursk
51.7/36.0
Kostroma
57.8/41.0
Saratov
51.3/46.0
Astrakhan
48.2/46.2
Rostov-Don
47.0/40.0
Ufa
54.9/55.9
L'vov
49.5/23.6
Vinnitsa
49.3/28.3
Kharkov
50.1/36.3
Askania-Nova
46.3/33.5
Kishininev
47.0/28.9
Brest
52.2/23.9
Minsk
54.0/27.3
Kaunas
54.5/24.0
Aktyubinsk
50.2/57.1
Petropavlovsk
54.5/69.1
Zelinograd
51.5/68.2
Karaganda
49.1/73.1
Distance +
(km)
290
0
125
75
270
100

35

0

80

75

45

0

40
165

125
95
90
210
125
Solar
Station
(Lat/Long)
Br.forest
53.1/34.5
Ostrogozhsk
50.9/39.0
Vologda
59.2/40.0
Saratov
51.3/46.0
Or. Promysel
45.9/47.5
Rostov-Don
47.0/40.0
Ufa
54.9/55.9
L'vov
49.5/23.6
Kiev
50.6/30.3
Kharkov
50.1/36.3
Kherson
46.4/32.4
Kishinev
47.0/28.4
Vasilyevichi
52.2/30.0
Minsk
54.0/27.3
Zarasay
55.7/26.3
Orenburg
51.4/55.1
Petropavlovsk
54.5/69.1
Atbasar
51.1/71.2
Karkaralinsk
49.5/75.5
+ Distance from the agricultural region to the meteorological station
                                         FSU-17

-------
APPENDIX 3.
Method for calculating local climate change scenarios.
        The procedure for developing the local climate change scenarios using the nearest GCM
gridpoint and the local climate data set may not always be the best representation of the local future
climate. In certain cases there may be discrepancies between the observed climate variables and the
simulated  baseline climate  for a site.  Problems may arise  when the climate variables  are
nonuniformly distributed on the region, and when two grain-producing sites are closely situated or
within the same GCM gridbox. This  is a particular problem in the creation of the scenarios in the
113 grain-producing regions selected for this study, because the gridpoint resolution of the GCMs
used  (GISS, GFDL, and UKMO) is  lower than ideal. The  distance between gridpoints in the
latitudinal direction for any of the GCMs used is two times the size of the average region selected
for the study; Furthermore,  when we want to develop a map of  the spatial  distribution of the
changes of climatic variables  and the productivity in the agricultural territory.
        These problems arise because of the large agricultural region used in this study. To obtain
potential representative estimates of changes of national agricultural production under the climate
change scenarios it is important to study the responses of crops yields in relatively small regions. In
addition, it would be best to  represent the simulation results  as territorial distributions of output
parameters in the forms of maps with isolines.
        We developed a method called Multipoint Interpolation for developing local climate
scenarios based on a modification of the standard method. The  method is based on the interpolation
of climate variables from several gridpoints for each GCM to each  of the centers of the wheat-
producing regions (113). For  reasons of symmetry, the 16 nearest surrounding gridpoints were used
in this modified procedure. To calculate the values of the three climate variables (temperature,
precipitation, and solar radiation)  for each agricultural site we used the formula:

        X(in the site)  = Sum[p(i) *  X(i)];  i=l,2,...16

where X(i) are the climatic variables in each of the 16 gridpoints, and p(i) are their relative weights.
        To estimate the relative weight values (p(i)), we assumed that the weight of each of the 16
gridpoints that surround the agricultural site for each variable  must  be proportional to the value of
the correlation coefficient between its magnitudes in the gridpoint  and of the site. The coefficient
of correlation for temperature (K(T,i)), precipitation (K(P,i)), and solar radiation (K(S,i)) have been
plotted as a function of distance by the formulas:
               = exp[:r(i)/R(T)]
               = exp[-r(i)/R(P)]
               = exp[-r(i)/R(S)]
             i= 1,2,.... 16
where R(T), R(P), and R(S)  are  so-called  radii of correlation,  (e.g., the distances  for which
correlation has decreased e = 2.7183). The final formulas for weights after normalization can be
represented as follows:

              = K(T,i)/Sum[K(T,i)];
              = K(P,i)/Sum[K(P,i)];
              = K(S,i)/Sum[K(S,i)].
                                          FSU-18

-------
        To carry out the calculations in accordance with this procedure, the values of these three
parameters must be known: radii of temperature, precipitation, and solar radiation correlation. In
this particular study we considered these values to be:

        R(T) = 1500 km;  R(P)  = 600 km;  R(S) = 700 km.

        The use of this procedure establishes a better correlation between the local climate and the
climate scenarios generated by the GCMs. But in spite of the use of this more accurate technique
of interpolation, there is still a degree of uncertainty associated with  the local climatic scenarios.
Specifically, there is  also uncertainty  associated  with the  extrapolation of present spatial
meteorological fields into the future. For more accurate  use of this multipoint interpolation
procedure, it is necessary to have information on the changes of correlation radii R(T), R(P), and
R(S) during climate change.
                                          FSU-19

-------
APPENDIX 4.
            A complete list of sites in the FSU used in the simulation study.  The
            numbers below each site indicate its latitude/longitude. An asterisk (*)
            indicates the sites selected as examples for this report. In these sites the
            climate change scenarios were created by the simplified method.
     1.

     3.

    *5.

     7.

     9.

    11.

    13.

   *15.

    17.

   *19.

   *21.

   *23.

    25.

   *27.

    29.

   *31.

    33.

    35.

    37.

    39.
Vinnitsa
48.7/28.6
Dnepropetrovsk
48.4/34.7
Zhitomir
50.0/28.4
Kiev
50.1/31.0
Lugansk
49.0/38.8
Odessa
46.6/30.4
Rovno
50.4/26.0
Kharkov
49.5/36.3
Cherkassy
49.0/31.3
Moldova
47.1/28.6
Bryansk
52.9/33.4
Kursk
51.7/36.0
Oryol
52.8/36.2
Brest
52.0/24.2
Gomel
52.0/29.4
Minsk
53.3/26.0
Vladimir
56.0/40.9
Ivanovo
57.2/41.4
Kostroma
58.2/43.2
Mordovia
54.4/44.3
  2.      Volyn
         50.8/24.4
  4.      Donetsk
         47.9/37.5
  6.      Zaporozhye
         47.3/35.7
  8.      Kirovograd
         48.3/32.2
 10.      Nikolaev
         47.2/32.0
 12.      Poltava
         49.6/33.8
 14.      Sumy
         51.0/33.9
*16.      Kherson
         46.7/33.5
 18.      Chernigov
         51.1/32.0
 20.      Belgorod
         50.7/37.7
 22.      Voronezh
         51.1/40.0
 24.      Lipetsk
         52.6/38.8
 26.      Tambov
         52.6/41.5
 28.      Vitebsk
         55.2/28.3
 30.      Grodno
         53.0/24.2
 32.      Mogilyev
         53.6/30.4
 34.      Nizhniy Novgorod
         55.9/44.2
 36.      Kaluga
         54.3/35.0
 38.      Mary
         56.4/47.8
 40.      Ryazan
         54.2/40.7
                                           FSU-20

-------
41.

43.

45.

47.

49.

51.

53.

55.

57.

*59.

61.

63.

65.

67.

*69.

*71.

73.

*75.

77.

79.

81.

83.

85.

Tula
53.9/37.4
Chuvashia
55.4/47.0
Moscow
55.4/38.1
Volgograd
49.9/44.1
Krasnodar
45.2/39.2
Stavropol
44.8/43.2
West Orenburg
51.8/54.3
Penza
53.1/44.7
Tataria
55.3/51.0
North Bashkiria
55.0/55.6
Kurgan
55.4/65.0
South Yekaterinburg
57.2/61.9
Region "B" of Mid.Ural
54.8/58.9
South Chelyabinsk
53.1/60.2
North Aktyubinsk
49.9/57.9
Karaganda
50.0/72.4
Kustanaiy
52.4/63.5
North Kazakhstan
54.6/68.3
C'trl Semipalatinsk
48.7/80.0
North Ural
50.2/52.9
Tver
57.1/34.5
St. Petersburg
59.2/29.0
Novgorod
58.2/32.0
42.

*44.

*46.

48.

*50.

52.

54.

*56.

58.

60.

62.

64.

66.

68.

70.

72.

74.

76.

*78.

80.

82.

84.

86.

Udmurtia
56.6/52.2
Yaroslavl
57.6/38.9
Astrakhan
46.6/48.0
Kalmykia
45.9/44.4
Rostov
47.4/40.8
Samara
53.0/50.6
East Orenburg
51.3/59.6
Saratov
51.6/46.6
Ulyanovsk
53.9/48.0
South Bashkiria
52.9/57.3
Perm
57.5/55.9
Region "A" of Mid.Ural
57.8/59.9
North Chelyabinsk
55.0/60.5
South Urals Region
53.2/59.1
North Guryev
57.5/53.4
Kokchetav
53.4/70.3
Pavlodar
52.3/76.0
North Semipalatinsk
50.3/79.6
Zelinograd
51.6/70.1
South Ural
49.3/49.4
Smolensk
54.6/32.3
Pskov
57.1/28.5
Vologda
59.5/40.4
FSU-21

-------
87.

89.

91.

93.

95.

97.

99.

101.

103.

105.

*107.

109.

111.

113.

Latvia
56.2/23.8
Estonia
58.2/25.5
Kaliningrad
54.2/20.6
South Tomsk
58.0/83.9
South- West Novosibirsk
55.0/77.9
South-East Kemerovo
54.1/87.0
North-East Altai
53.6/84.4
North Omsk
57.2/73.4
North Tyumen
58.9/68.6
Vyatka
57.9/49.8
Lvov
49.5/23.6
Carpathians
47.7/22.3
Ternopol
49.0/25.1
Crimea
45.4/34.0
*88.

90.

92.

94.

96.

98.

100.

102.

104.

106.

108.

110.

112.



       Lithuania
       55.0/23.0
       Karelia
       61.9/32.4
       North Tomsk
       59.5/80.0
       NE Novosibirsk
       55.6/82.1
       North-West Kemerovo
       56.0/86.0
       North-West Altai
       52.1/79.8
       South-East Altai
       51.9/83.0
       South Omsk
       55.1/73.0
       South Tyumen
       56.9/68.4
       East Kazakhstan
       50.7/82.3
       Ivano-Frankovsk
       48.4/24.0
       Chernovtsy
       48.0/26.1
       Khmelnitsky
       49.1/26.0
FSU-22

-------
Table 1.   Soil types and wheat cultivars used in the simulation.
     SITE
SOIL TYPE
WHEAT CULTIVAR (Spring,
Winter)
     Bryansk       Gray Wooded!
     Kursk        Chernozem:
                    Leached, Modal
     Yaroslavl     Dernopodzolic
     Saratov       Chernozem: Steppe, Modal,
                   Southern; Deep Chestnut
     Astrakhan     Light Chestnut, Brown Steppe,
                   Desert
     Rostov-Don   Deep Chestnut Chernozem:
                   Southern, Deep
     N. Bashkiria   Chernozem: Modal,
                   Leached
     Lvov         Leached Chernozem,
                   Dernopodzolic
     Zhitomir      Modal Chernozem
                   Dernopodzolic
     Kharkov       Modal Chernozem
     Kherson       Chernozem:
                    Deep, Steppe, Southern

     Kishinev       Gray Wooded,
                   Modal Chernozem
     Brest         Podzolic, Ferruginous, Humic
     Minsk        Dernopodzolic Podzolic
     Lithuania     Glee Wooded
     Aktyubinsk    Deep Chestnut, Brown Steppe
                   Desert
     N. Kazakhstan Brown Meadow Steppe
     Zelinograd     Chernozem: Steppe, Southern;
                   Deep Chestnut
     Karaganda     Chestnut: Light, Deep
                                 Palmira, Bezostaya-1
                                 Albidum-43, Mironovskaya-808

                                 Palmira, Bezostaya-1
                                 Albidum-43, Mironovskaya-808

                                 Albidum-43, Mironovskaya-808

                                 Lutescens-62, Krymskaya

                                 Albidum-43, Mironovskaya-808

                                 Ostka, Lutescens-17

                                 Ostka, Lutescens-17

                                 Albidum-43, Mironovskaya-808
                                 Lutescens-62, Krymskaya


                                 Lutescens-62, Krymskaya

                                 Palmira, Bezostaya-1
                                 Palmira, Bezostaya-1
                                 Palmira, Bezostaya-1
                                 Albidum-43, Mironovskaya-808

                                 Albidum-43, - NO -
                                 Albidum-43, - NO -

                                 Albidum-43, Mironovskaya-808
                                      FSU-23

-------
Table 2.   Sensitivity analysis of CERES-Wheat to climate and CO2 changes in Zhitomir and N.
          Bashkiria.
                                                Changes from baseline
CO2 level Precip.
(ppm) (%)
330 0%


20%


-20%


555 0%


20%


-20%



330 0%


20%


-20%


555 0%


20%


-20%


Temp. Yield
(C) (%)
0
2
4
0
2
4
0
2
4
0
2
4
0
2
4
0
2
4

0
2
4
0
2
4
0
2
4
0
2
4
0
2
4
0
2
4
0
-16
-33
5
-12
-27
-8
-26
-44
26
8
-9
30
12
-4
20
0
-19
N. Bashkiria
0
-22
-40
2
-18
-36
-4
-26
-44
21
1
-17
27
3
-14
13
-2
-21
Season ET
L.(days) (mm)
0
-8
-14
0
-8
-14
0
-8
-14
0
-8
-14
0
-8
-14
0
-8
-14
Spring Wheat
0
-8
-15
0
-8
-15
0
-8
-15
0
-8
-15
0
-8
-15
0
-8
-15
0
-4
-9
3
0
-4
-6
-11
-18
-6
-9
-11
-4
-6
-8
-9
-13
-18

0
-4
-9
4
-2
-4
-8
-12
-16
-6
-6
-10
-4
-4
-6
-10
-12
-16
Yield Season
(%) L.(days)
0
-23
-38
8
-13
-27
-18
-36
-52
27
0
-17
36
14
-3
16
-13
-34
N. Bashkiria
0
-25
-45
13
-12
-34
-20
-40
-56
35
10
-15
48
22
-3
17
-9
-32
0
-10
-18
0
-10
-18
0
-10
-18
0
-10
-18
0
-10
-18
0
-10
-18
Winter
0
-10
-18
0
-10
-18
0
-10
-18
0
-10
-18
0
-10
-18
0
-10
-18
ET
(mm)
0
-3
-6
4
2
0
-8
-12
-13
-6
-7
-9
-2
-3
-5
-10
-14
-10
Wheat
0
-3
-5
5
2
1
-7
-10
-11
-5
-5
-7
1
-1
-3
-9
-11
-13
                                       FSU-24

-------
Table 3.   Temperature change (C) predicted by GISS, GFDL, and UKMO climate change
          scenarios.
            Site
Spring  Summer Autumn Winter  Annual
Bryansk GISS
GFDL
UKMO
Kursk GISS
GFDL
UKMO
Yaroslavl GISS
GFDL
UKMO
Saratov GISS
GFDL
UKMO
Astrakhan GISS
GFDL
UKMO
Rostov-on-Don GISS
GFDL
UKMO
N. Bashkiria GISS
GFDL
UKMO
Lvov GISS
GFDL
UKMO
Zhitomir GISS
GFDL
UKMO
Kharkov GISS
GFDL
UKMO
Kherson GISS
GFDL
UKMO
Kishinev GISS
GFDL
UKMO
4.3
4.6
11.0
4.6
4.6
11.0
4.4
5.3
9.6
4.7
4.5
10.1
3.6
4.0
10.2
4.6
4.0
9.0
5.0
4.4
9.8
3.7
4.6
10.4
4.3
4.6
11.4
4.6
4.6
10.6
4.8
4.1
10.6
4.8
4.1
10.6
1.0
4.8
8.6
1.4
4.1
8.6
0.7
4.1
5.8
3.0
3.6
8.3
4.3
3.6
8.1
1.4
3.8
8.9
3.9
3.8
8.5
2.3
5.2
8.2
1.0
4.8
7.1
1.4
4.1
8.5
3.5
4.5
8.5
3.5
4.5
8.5
4.3
4.7
7.5
4.6
4.2
7.5
4.4
4.8
6.5
5.0
3.7
7.1
4.8
3.6
7.3
4.6
4.0
7.5
4.6
4.3
7.0
4.2
4.8
7.7
4.3
4.7
7.1
4.6
4.2
7.2
4.7
4.3
7.2
4.7
4.3
7.2
6.3
4.2
8.5
6.6
4.1
8.5
6.8
4.6
9.9
6.9
4.1
8.4
4.6
4.2
8.2
6.6
3.9
9.5
6.4
4.2
8.0
5.8
4.6
8.7
6.3
4.2
8.7
6.6
4.1
9.7
4.7
3.9
9.7
4.7
3.9
9.7
4.0
4.6
8.9
4.3
4.2
8.9
4.1
4.7
7.9
4.9
4.0
8.9
4.3
3.8
8.7
4.3
3.9
8.7
5.0
4.2
8.3
4.0
4.8
8.8
4.0
4.6
8.6
4.3
4.2
8.0
4.4
4.2
8.0
4.4
4.2
8.0
                                       FSU-25

-------
Brest


Minsk


Lithuania


Aktyubinsk


N. Kazakhstan


Zelinograd


Karaganda


GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
3.7
4.6
11.4
4.3
4.6
11.4
4.0
5.3
11.5
5.0
4.8
10.1
5.0
5.2
10.5
5.0
5.3
10.5
5.0
5.4
10.5
2.3
5.2
7.1
1.0
5.2
7.1
1.6
4.7
6.2
3.9
4.0
9.1
4.1
4.0
8.3
4.1
4.1
8.3
4.1
4.5
8.3
4.2
4.8
7.1
4.3
4.8
7.1
3.5
5.0
6.0
4.6
5.8
7.0
4.1
6.1
7.7
4.1
6.4
7.7
4.1
5.9
7.7
5.8
4.6
8.7
6.3
4.6
8.7
5.3
5.3
9.5
6.4
5.0
9.0
7.5
6.0
6.9
7.5
6.0
6.9
7.5
6.1
6.9
4.0
4.8
8.6
4.0
4.8
8.6
3.6
5.1
8.3
5.0
4.9
8.8
5.2
5.3
8.3
5.2
5.4
8.3
5.2
5.5
8.3
FSU-26

-------
Table 4.   Precipitation change (%)  predicted by GISS, GFDL, and UKMO climate change
          scenarios.
          Site
Spring  Summer Autumn   Winter  Annual
Bryansk


Kursk


Yaroslavl


Saratov


Astrakhan


Rostov-on-Don


N. Bashkiria


Lvov


Zhitomir


Kharkov


Kherson


Kishinev


GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
+24
+36
+71
+28
+15
+72
+34
+8
+79
+16
+28
+70
+22
+10
+68
+30
+3
+44
+40
+18
+77
+27
+25
+49
+24
+36
+60
+28
+15
+52
+15
-1
+52
+15
-1
+52
+35
-3
-24
+37
+20
-24
+74
+2
-12
+38
+57
-24
-98
+31
-20
+37
+8
-32
+45
+18
-31
+9
-16
-16
+35
-3
+6
+37
+20
-35
-11
+9
-35
-11
+9
-35
+33
+8
+25
+18
-3
+25
+35
+23
+33
-6
+6
+18
-8
-5
+14
+18
-27
+21
-22
+44
+21
+6
+2
+19
+33
-3
+6
+18
-3
+28
+6
+15
+28
+6
+15
+28
+16
+6
+34
+38
+18
+34
+36
+5
+47
+41
+26
+30
+19
+19
+22
+38
+38
+37
+31
+39
+28
+14
-6
+26
+16
+6
+47
+37
+18
+23
+5
+29
+23
+5
+29
+23
+27 '
+12
+27
+30
+12
+27
+45
+9
+37
+22
+29
+23
+33
+13
+21
+30
+5
+17
+24
+29
+23
+14
+2
+19
+27
+12
+38
+30
+12
+ 16
+4
+12
+16
+4
+12
+16
                                       FSU-27

-------
Site
Spring Summer Autumn  Winter  Annual
Brest


Minsk


Lithuania


Aktyubinsk


N. Kazakhstan


Zelinograd


Karaganda


GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
+27
+25
+70
+24
+25
+70
+29
+2
+84
+40
+12
+59
+51
+33
+84
+51
+16
+84
+51
+47
+84
+9
-16
+6
+35
-16
+6
+49
-9
+11
+45
+7
-22
+99
+37
-26
+99
+ 11
-26
+99
+24
-26
+6
+2
+33
+33
+2
+33
+28
+24
+29
-22
+94
+24
+52
+38
+2
+52
+33
+2
+52
-4
+2
+ 14
-6
+47
+16
-6
+47
+ 14
+23
+50
+31
+46
+32
+78
+76
+30
+78
+22
+30
+78
+23
+30
+14
+2
+38
+27
+2
+38
+32
+ 10
+42
+24
+40
+23
+70
+46
+22
+70
+20
+22
+70
+22
+22
                          FSU-28

-------
Table 5.   Solar rdiation change (%) predicted by GISS, GFDL, and UKMO climate change
          scenarios.
          Site
Spring  Summer Autumn  Winter  Annual
Bryansk


Kursk


Yaroslavl


Saratov


Astrakhan


Rostov-on-Don


N. Bashkiria


Lvov


Zhitomir


Kharkov


Kherson


Kishinev


GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
0
+10
+22
-1
+14
+22
-5
+47
+8
-3
+11
+31
+1
-2
+29
-1
+1
+18
-4
+10
+29
-2
+2
+26
0
+10
+23
-1
+14
+21
0
-1
+21
0
-1
+21
-3
+6
+9
-3
-2
+9
-11
+1
+5
-2
-3
+15
0
-1
+16
-3
0
+9
-1
-1
+1
+1
+7
+6
-3
+6
0
-3
-2
+4
0
0
+4
0
0
+4
-6
+6
+10
-4
-9
+10
-9
-2
+6
0
+2
+15
+5
+6
+14
-4
+11
+6
0
+9
+11
-4
+7
+12
-6
+6
+6
-4
+9
+5
+7
+4
+5
+7
+4
+5
-16
+83
+4
-20
+62
+4
-18
+37
-1
-20
+24
+6
+1
+25
+3
-20
+14
0
-20
+10
+ 10
-12
+51
+7
-16
+83
+ 1
-20
+62
-6
+1
+16
-6
+1
+16
-6
-6
+26
+11
-7
+20
-11
-8
+20
+4
-6
+8
+16
+2
+7
+17
-7
+4
+7
-6
+ 15
+ 15
-4
+ 17
+ 12
-6
+26
+8
-7
+20
+6
+2
+5
+6
+2
+5
+6
                                      FSU-29

-------
Site
Brest


Minsk


Lithuania


Aktyubinsk


N.Kazakhstan


Zelinograd


Karaganda


Spring Summer Autumn
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
-2
-2
+23
0
+2
+23
-2
+9
+14
-5
+21
+13
-5
+6
+1
-5
+10
+11
-5
+10
+11
+1
+7
0
-3
+7
0
-2
+3
-3
-1
+2
+8
0
0
+14
0
-1
+14
+0
-1
+14
-4
+7
+6
-6
+7
+6
-4
-1
+14
0
+12
+3
-2
+12
+15
-2,
+6
+15
-2
+9
+15
Winter Annual
-12
+51
+1
-16
+51
+1
-1
+37
-2
-20
+7
-2
-21
+1
+1
-21
+21
+1
-21
+35
+ 1
-4
+17
+8
-6
+17
+8
-4
+12
+6
-6
+10
+5
-7
+5
+10
-7
+9
+10
-7
+12
+ 10
FSU-30

-------
Table 6.    Simulated change in wheat yield under projected GCM climate scenarios for selected
           sites. The physiological effects of 555 ppm CO2 were considered in each climate change
           scenario.
Base Yields
(t ha'1)
Site
Bryansk
Kursk
Yaroslavl
Saratov
Astrakhan
Rostov-Don
N. Bashkiria
Lvov
Zhitomir
Kharkov
Kherson
Kishinev
Brest
Minsk
Lithuania
Aktyubinsk
N. Kazakhstan
Zelinograd
Karaganda
Spr
5.0
3.0
3.5
2.1
0.7
3.2
2.8
6.7
5.4
2.9
1.9
2.8
4.2
4.0
4.3
1.6
4.6
2.4
2.0
Win
6.2
6.0
5.7
3.2
1.0
3.7
4.6
5.8
4.0
5.2
2.1
2.4
3.9
4.6
4.7
1.3
-
-
1.7
Simulated yield changes
under GCM scenarios %
GISS GFDL UKMO
Spr
+12
+25
+18
+ 4
+ 9
+12
-5
+ 6
+16
+ 4
+30
+31
+ 5
+10
+17
+61
-5
+55
+59
Win
+23
+23
-28
+45
+13
+51
+25
+18
+13
+34
+52
+95
+22
+23
+42
+68
-
-
+102
Spr
-31
+ 1
+ 1
+ 6
+51
-9
-3
-20
-18
-11
-3
+ 3
-25
-32
-26
+16
-1
-12
+ 6
Win
0
+ 9
-34
+45
+28
+30
+11
-8
-4
+12
+25
+71
+ 6
+ 2
-38
+ 12
-
-
+ 5
Spr
+34
-31
-1
-48
-7
-30
-38
-19
-33
-44
+ 1
+11
-35
-32
-27
-36
-38
-28
-18
Win
-12
-16
0
-44
+32
+30
-53
-16
-33
-17
+91
+ 130
-33
0
+47
- 1
-
-
+ 9
      Base yields are simulated considering a CO2 concentrationof 330 ppm.
                                         FSU-31

-------
Table 7.    Simulated changes in wheat season length for GCM climate change scenarios at selected
           sites. The physiological effects of 555 ppm CO2 were considered in each climate change
           scenario.
Season Length Changes (days)
GISS GFDL UKMO
Site
Bryansk
Kursk
Yaroslavl
Saratov
Astrakhan
Rostov-Don
N-Bashkiria
Lvov
Zhitomir
Kharkov
Kherson
Kishinev
Brest
Minsk
Lithuania
Aktyubinsk
N.Kazakhstan
Zeli-grad
Karaganda
Spr
-8
-9
-1
-8
-9
-8
-13
-10
-8
-8
-12
-11
-10
-8
-8
-8
-14
-10
-9
Win
-23
-22
-72
-20
-14
-20
-20
-25
-23
-24
-22
-22
-20
-18
-33
-16
-
-
-16
Spr
-19
-11
-6
-6
-7
-8
-11
-19
-15
-10
-10
-11
-18
-20
-18
-8
-14
-11
-12
Win
-25
-20
-20
-15
-15
-19
-16
-25
-20
-20
-18
-18
-23
-23
-38
-16
-
-
-19
Spr
-27
-24
0
-16
-14
-13
-22
-23
-26
-21
-19
-19
-26
-25
-25
-13
-22
-18
-17
Win
-55
-49
-16
-35
-39
-34
-37
-52
-44
-44
-50
-48
-52
-50
-70
-32
-
-
+32
                                         FSU-32

-------
Table 8.   Simulated changes in wheat evapotranspiration for projected GCM climate scenarios at
          selected sites. The physiological effects of 555 ppm CO2 were considered in each climate
          change scenario.
                                        Change in ET (mm)

Site
Bryansk
Kursk
Yaroslavl
Saratov
Astrakhan
Rostov-Don
N.Bashkiria
Lvov
Zhitomir
Kharkov
Kherson
Kishinev
Brest
Minsk
Lithuania
Aktyubinsk
N.Kazakhstan
Zeli-grad
Karaganda

Spr
-9
-5
-2
0
-4
-2
-4
-7 
-8
-7
-4
-8
-10
-8
-4
+7
+2
+15
+15
GISS
Win
-6
-3
-30
+1
0
-2
-1
-5
-8
-4
-3
-2
-12
-10
-2
+8
-
-
+22

Spr
-12
-4
-10
-1
-4
-10
-8
-13
-9
-6
-8
-10
-15
-12
-16
-4
-5
-12
-2
GFDL
Win
+4
+4
+8
+7
+2
-3
-1
-1
+6
0
-3
-2
-9
-3
-12
+9
-
-
+12
UKMO
Spr
-14
-10
+20
-10
-5
-10
-13
-9
-8
-11
0
-2
-11
-9
-4
-6
-13
-10
-6
Win
0
+2
+27
+8
0
+1
+4
-7
0
+3
-4
-1
-24
-3
-6
+11
-
-
+18
                                         FSU-33

-------
Table 9.
Aggregated national wheat yield change  (%) for GCM equilibrium and the GISS
transient climate change scenarios.
             Scenario         Carbon Dioxide         Yield changes (%)
             	                 (PPm)     Spring Wheat Winter Wheat
             GISS-Trans-2010        405
             GISS-Trans-2030        460
             GISS-Trans-2050        530

             GISS                  555
             GFDL                 555
             UKMO                555
                                             +8
                                            +12
                                            +18

                                            +21
                                             -4
                                             -18
+16
+24
+38

+41
+12
 +9
                                     FSU-34

-------
Table 10.   Simulated wheat yield response to changes in sowing date and irrigation under the
           UKMO  climate change  scenario. The physiological effects of 555 ppm CO2 were
           considered in the simulation.
          Site
Management     Sowing date
          Saratov
Rainfed
Rainfed
Irrigated
Irrigated
15 days early
30 days early
15 days early
30 days early
                     % Change from
                       Baseline yield
Zhitomir



Rainfed
Rainfed
Irrigated
Irrigated
15 days late
30 days late
15 days late
30 days late
-1
-3
+50
+40
  +1
  -1
 +70
+100
                                         FSU-35

-------

-------
SECTION 5: AFRICA

-------

-------
IMPACT OF CLIMATE CHANGE ON SIMULATED WHEAT AND MAIZE YIELDS
                               IN EGYPT
                                H.M. Bid

                        Soils & Water Resources Institute,
                       ARC, Ministry of Agriculture, Egypt
                                 EGYPT-1

-------
TABLE OF CONTENTS

SUMMARY
INTRODUCTION
METHODS
       Climate Data and Climate Change Scenarios
       Crop Models
       Validation of the Crop Models
RESULTS AND DISCUSSION
       Crop Sensitivity to Temperature Increases
       Yields under GCM Climate Change Scenarios
       Yields under GISS Transient Scenarios
       Adaptation Strategies for Wheat and Maize to Climate Change
       Future Research Needs
REFERENCES
                                      EGYPT-2

-------
SUMMARY

        The potential impact of climate change on maize and wheat production in Egypt was evaluated by
simulating crop production under different climatic scenarios and by analyzing crop sensitivity to temperature
increases in the two major agricultural regions of Egypt. Increases of 2C and 4C reduced wheat and maize
yields in the Delta  and Middle Egypt regions. Under GCM climate change scenarios, yields decreased in
comparison to current climate conditions, even when the beneficial effects of CO2 were taken into account.
The UKMO scenario produced the largest yield decreases. According to this simulation study, the impact of
climate change on national wheat and maize production would be severe. Future adaptation strategies to
climate change may involve the development of new, more heat-tolerant cultivars.
INTRODUCTION

        Major crops in Egypt include wheat (used as a staple food crop), maize (used primarily as coarse grain
for animal feed), clover, cotton, rice, sugar cane, fava bean, and soybean. The national wheat and maize
production do not meet the current demand for these crops, and each year additional amounts have to be
importedup to 60% of total consumption in the case of wheat. The rapid growth of the country's population,
the economic stress of reliance on food imports, and the limited area for agriculture (most of the country is
a desert)  require Egyptians to find  new ways to increase agricultural productivity. If climate  change as
projected  by atmospheric scientists (IPCC 1990) adversely affects crop production, Egypt would have to
increase its reliance on costly food imports.
        The purpose of this study was to investigate the potential effects of climate change on wheat and
maize yields and irrigation demands in Egypt. Wheat and maize are irrigated under the flood irrigation system,
using water from the Nile River. Two important agricultural regions were selected for the study (Figure 1).
The Sakha region, at the north of the Nile delta, is represented by the site Sakha (Khafr El-Sheik governorate)
(31.07N; 30.57E). This region of the delta is the most fertile area in Egypt and produces about 60% of the
national wheat and 75% of the total maize crop (Table 1). The Giza region, near Cairo, is represented by the
site Giza (30.03N; 31.13E).

METHODS

Climate Data and Climate Change Scenarios

        Daily maximum and minimum temperatures, precipitation, and solar radiation data were obtained
for Sakha from 1975 to  1989 and for Giza from 1960 to 1989 (Table 2).
        The climate change scenarios used in this simulation study were created using three equilibrium
General Circulation Models (GCMs) combined with the observed daily climate data for each site (Table 2).
Three GCMs  were used: Goddard Institute for Space Studies, (Hansen  et al.  1983),  Geophysical Fluid
Dynamics Laboratory, (Manabe and Wetherald 1987), and United Kingdom Meteorological Office, (Wilson
and Mitchell 1987). Table 3 shows the projected seasonal and annual GCM climate changes for these sites.
In Giza, the GISS and UKMO climate change scenarios presented consistent, unexplained errors in the years
1968  (GISS scenario) and 1974 and 1989 (UKMO scenario). These years were not included in  the study.
        The study also includes: (a) a sensitivity study to arbitrary  changes in temperature (+2 C and +4 C);
and (b) a transient scenario study, using the GISS transient run A for the years 2010, 2030, and 2050 (Hansen
et al.  1988). Atmospheric CO2 concentrations of 405 ppm, 460 ppm, and 530 ppm were used for the years 2010,
2030, and 2050, respectively.

                                             EGYPT-3

-------
 Crop Models

        Crop yields and demand for irrigation water were estimated with the CERES-Wheat (Ritchie and
 Otter  1985)  and CERES-Maize  (Jones and Kiniry  1986) models using  the  version developed by  the
 International Benchmark Sites Network for Agrotechnology Transfer (IBSNAT 1989). The IBSNAT crop
 models can simulate the physiological effects of CO2, i.e., increase in rate of photosynthesis and change in
 evapotranspiration due to increases in stomatal resistance (Acock and Allen 1985). Simulations for the GCM
 scenarios and sensitivity analyses included climate change alone and climate change with the physiological
 effects of CO2 at both sites.

 Management Variables for the Crop Models

        Typical soils at Giza and Sakha are montmorillonitic, thermic, slightly calcareous, and deep (Abdel
 Wahed 1983). The texture, albedo, and water-related specific characteristics of these soils are adequately
 represented by  the generic soil (Medium Silty Clay) provided for the study (Jones et al. 1990).
        Wheat and maize are grown using flood irrigation (Gad-El Rab et al.  1988; Eid 1977) . For  the
 simulations, the automatic irrigation option was chosen to provide  the crops with nonlimiting water;  the
 models do not include an option that simulates flooding. Maize and wheat are fertilized in the regions in this
 study, and therefore the simulations did not consider nitrogen stress.

 Crop  Model Validation

        The CERES-Wheat and CERES-Maize models were validated by comparing observed data on biomass,
 yields, and maturity dates to simulated values (Table 4). The results of the validation experiment indicate that
 the CERES crop  models can be used at the selected sites in Egypt. The observed data on grain yield and
 season length were very close to the corresponding simulated values. The observed total biomass was slightly
 smaller than the simulated one. According to these results, the models were considered validated for  the
 conditions of the study.

 RESULTS AND DISCUSSION

 Crop  Sensitivity to Temperature and CO2 Increases

        Table 5 shows simulated grain yield, season length, evapotranspiration (ET), and irrigation demand
 in response to arbitrary changes in temperature. The physiological effects of CO2 were considered in each
 scenario (shown as 555 ppm in the tables). Increases in temperature resulted in  lower grain yields for both
 crops at the two sites; yield reductions showed a linear relation to temperature increases. Yield decreases were
 larger for maize, a summer crop, than for wheat, a winter crop. The temperature-induced reductions in maize
 and wheat may be due to a shortening of the grain-filling periods. Although not simulated in CERES-Maize,
 high temperatures as projected by the GCM scenarios may also cause pollinization failure. When the direct
 effects of CO2 were considered, simulated grain yields increased in all scenarios compared with the results in
 the scenarios of climate change alone. These increases, however, did not compensate for the yield decreases
 under the higher temperature scenarios compared to base yields.
        Total crop evapotranspiration and irrigation demand increased with temperature increases in the case
 of maize, but decreased in the case of wheat, due to the shortening of the crop  growing season. When the
 physiological effects of CO2 were considered, ET and irrigation water demand were reduced for both  crops
compared to the case of climate change alone. These reductions are the consequence of the CO2-induced

                                            EGYPT-4

-------
decreases in the transpiration rate per unit leaf area (Acock and Allen 1985) and the shortening of the crop
growing season included in the crop models.

Yields under GCM Climate Change Scenarios

        All climate change scenarios considered resulted in simulated decreases in maize and wheat yields at
both sites (Table 6 and Figures 2 and 3). The largest decreases in yield for both wheat and maize were under
the UKMO scenario in Giza. The physiological effects of CO2 caused simulated maize yields to increase
slightly and wheat yields to increase substantially compared to yields under the climate change scenario alone.
The relative responses of wheat and maize are due to the greater CO2 photosynthetic response in C3 crops
(wheat, barley, soybean, cotton, rice, and many others) than in C4 crops (maize, sugarcane, and sorghum)
(Cure 1985).
        Considering the simulated negative impacts of climate change on simulated wheat and maize yields
in the Delta and Middle  Egypt regions, it is possible to conclude  that climate change may bring about
substantial reductions in the national grain production.

Transient Scenarios

        The crop responses to  the transient  GISS scenarios are  shown  in Table 7. Simulated  grain yield
decreases were not linear with time; the largest reductions correspond to  the 2030s.

Adaptation  Strategies for Wheat and Maize to Climate Change

        Changes of maize and wheat cultivars were considered as  possible adaptation strategies to climate
change (Table 6). Nevertheless, for the two locations, all cultivars tested had similar losses in yield under the
climate change scenarios in comparison to baseline yields. It will be important for Egypt to develop new
cultivars that  are more adapted to higher temperatures. Since the crop models can be used  to identify
appropriate crops, varieties, and management strategies to maximize benefits and minimize risks associated
with future climatic change, further simulation studies would be valuable in assessing the risks associated with
given production strategies.
                                             EGYPT-5

-------
 REFERENCES

 Abdel Wahed, A.  1983. Research  report on  soil survey. Soil  Survey EMCIP  Res.  Extension  Centers.
        Publication No.62. Consortium for International Development. ARC, Cairo, May 1983.

 Acock, B., and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
        and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation. DOE/ER-0238. U.S.
        Department of Energy. Washington, D.C. pp. 53-97.

 Cure, J.D. 1985. Carbon dioxide doubling responses: A crop survey. In B.R. Strain and J.D. Cure (eds.), Direct
        Effects of Increasing Carbon Dioxide on  Vegetation. U.S. Department of Energy. DOE/ER-0238.
        Washington, D.C. pp. 33-97.

 Bid, H.M. 1977. Effect of some climatic factors on evapotranspiration, growth, and yield of wheat. Ph.D.
        Thesis, Fac. of Agric., Al-Azhar Univ., Egypt.

 Gad-El Rab, G.M., N.G. Ainer, and H.M. Bid. 1988. Water stress in relation to yield of wheat and some water
        relations in wheat. Egypt. J. Soil. Sci. 28, No.4, pp.433-445.

 Hansen, J., G. Russell, D. Rind, P. Stone, A Lascis, S. Lebedeff, R. Ruedy, and L.Travis. 1983. Efficient three-
        dimensional global models for climate studies: Models I and II. Monthly Weather Review 3:609-622.

 Hansen, J., I. Fung, A. Lascis, D. Rind, G. Russell, S. Lebedeff, R.  Ruedy, and P. Stone. 1988. Global climate
        changes as forecasted by the GISS 3-D model. Journal of Geophysical Research 93:9341-9364.

 IBSNAT.  International Benchmark Sites Network for Agrotechnology Transfer Project. 1989. Decision
        Support System for Agrotechnology Transfer Version 2.1  (DSSAT V2.1). Dept.  Agronomy and Soil
        Sci., College of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii 96822.

Jones, J.W., S.S. Jagtap, G.Hoogenboom, and G.T. Tsuji. 1990. The structure and function of DSSAT. pp 1-14.
        In: Proceedings of IBSNAT SYMPOSIUM: Decision Support System for Agrotechnology Transfer.
        Las Vegas. N.V. 16-18 Oct. 1989. University of Hawaii, Honolulu, Hawaii.

Jones, C.A, and J.R. Kiniry.  1986. CERES-Maize: A Simulation Model of Maize  Growth and Development.
       Texas A&M University Press. College Station, TX. 194 pp.

IPCC 1990. The IPCC Scientific Assessment, eds. J.T. Houghton, G.J. Jenkins, and JJ. Ephraums. Cambridge
       University Press.

Manabe, S.,  and  R. Wetherald.  1987. Large-scale changes of  soil wetness  induced  by an increase  in
       atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.
Ritchie, J.T., and S. Otter. 1985. Description and performance of CERES-Wheat: A user-oriented wheat yield
        model.  In W.O. Willis  (ed.). ARS Wheat Yield Project. Department of Agriculture, Agricultural
        Research Service. ARS-38. Washington, D.C. pp. 159-175.
                                            EGYPT-6

-------
Wilson, C.A., and J.F.B. Mitchell. 1987. A Doubled CO2 Climate Sensitivity Experiment with a Global Climate
       Model Including a Simple Ocean. Journal of Geophysical Research 92:13315-13343.
                                            EGYPT-7

-------
Table 1.
Wheat and maize production in Egypt.
                                           Area (ha)
                                            Production (t)
Yield (t ha'1)
WHEAT
Giza Region (1984)
Sakha Region (1984)
Egypt total
(avg. 1980-1988)
MAIZE
Giza Region (1984)
Sakha Region (1984)
Egypt total
(avg. 1980-1988)

26,111
39,395
572,000


54,269
36,264
754,000


107,055
162,110
2,166,000


309,290
204,557
3,386,000


4.15
4.12
3.79


5.66
6.64
4.49

                                            EGYPT-8

-------
Table 2.        Observed average temperature,  total precipitation, and solar radiation during the crop
               growing season at Giza and Sakha.
                       Maize at Giza

                           Temperature
                           Precipitation
                           Solar radiation
                       Maize at Sakha

                           Temperature
                           Precipitation
                           Solar radiation
                       Wheat at Giza

                           Temperature
                           Precipitation
                           Solar radiation
                       Wheat at Sakha

                           Temperature
                           Precipitation
                           Solar radiation
25.6 C
0.0mm
549 MJ m'2
23.9 C
2.0mm
546 MJ m'2
16.5 C
17.0 mm
408 MJ m'2
15.4 C
42'.0 mm
399 MJ  m'2
                                             EGYPT-9

-------
Table 3.       Seasonal and annual changes in temperature, precipitation, and solar radiation under GCM
              climate change scenarios at Giza and Sakha.
                      GCM CLIMATE CHANGES AT SAKHA AND GIZA

                                   Temperature Change (C)

Spring
Summer
Autumn
Winter
Annual


Spring
Summer
Autumn
Winter
Annual
GISS
5.1
3.2
4.4
4.0
4.2
Precipitation
GISS
-7.1
350.0
27.3
5.9
55.7
GFDL
4.5
4.4
4.1
3.7
4.2
Change (%)
GFDL
-19.2
0.0
-20.0
-10.0
-15.3
UKMO
4.7
4.1
4.5
4.5
4.4

UKMO
-12.5
-37.0
1.2
-8.9
-13.8
Solar Radiation Change (%)

Spring
Summer
Autumn
Winter
Annual
GISS
-0.3
-4.2
-1.2
0.0
1.7
GFDL
2.0
-0.6
0.6
0.8
0.6
UKMO
6.2
6.1
1.3
8.7
5.5
                                         EGYPT-10

-------
Table 4.        Simulated and observed wheat and maize yields, biomass, and season length at Giza and
                Sakha.
Yield (t ha1)
Cultivar

Giza-2
H.204
Cairo-1
Pio.514

D.C.202
Year

1985
1985
1958
1985

1984
Sim

9.95
10.70
9.77
9.92

9.20
Obs
Maize
9.90
10.54
9.61
9.83
Maize
8.93
Biomass
Sim
(Sakha)
23.47
24.77
24.75
24.11
(Giza)
31.59
(tha1)
Obs

26.73
28.45
25.94
26.53

24.44
S. length
(days)
Sim

120
122
122
122

128
Obs

121
121
121
121

124
        Sakha 8
1983
        Giza-156       1975

        Mexipak65     1975
          Wheat (Sakha)

4.30     4.35          18.72     14.41.

          Wheat (Giza)

6.22     6.14          14.77     18.27

6.24     6.27          15.10     16.73
153      156


132      133

132      133
Maize experiments: Sakha Clayey soil; planting June 1; row spacing 70 cm; 5.7 plants m"2; initial soil water
(depth, water content) (5,.246) (10..246) (15,.208) (15,.208) (15..208) (15,.318) (15,318) (15,.318); irrigation
dates and amounts (Julian day, mm) (169,158) (192,37) (199,39) (204,35) (210,41) (216,39) (222,36)  (228,37)
(234,37)   (240,35) (246,36)  (253,37)  (260,39) (267,37)  (274,38) (282,36) (290.35)  =  751  for H.204;
N-fertilization dates and amounts (Julian day, kg ha'1} (169,69) (190,133).

Wheat experiments: Sakha Clayey soil; planting December 24; row spacing 10 cm; 350 plants"2; irrigation dates
and amounts (Julian day, mm) (385,146) (29,35) (42,34) (73,35)  (84,34) (95,37)  (103,34) (111,36)  (119,34)
(126,40)  (132,36) (137,35) (142,38) (147,37); N-Fertilization dates and amounts (Julian  day, kg ha'1) (18,63)
(46,63).
                                              EGYPT-11

-------
Table 5.        Sensitivity analysis of yield, season length (SL), total evapotranspiration (ET), and irrigation
               water (Irrig.) simulated with CERES-Wheat and CERES-Maize.
                              WHEAT (SAKHA)
WHEAT (GIZA)
CO2
level
(ppm)
330


555


CO2
level
(ppm)
330


555


Temp
Increase
(C)
0
2
4
0
2
4
Temp.
Incr.
(C)
0
2
4
0
2
4
Yield
(t ha'1)
3.80
3.30
3.20
4.40
3.89
3.46

Yield
(T/Ha)
10.25
8.73
7.54
10.33
9.11
7.96
SL
(days)
141
133
125
141
133
125
MAIZE
SL
(days)
115
107
104
115
107
104
ET
(mm)
552
547
511
517
498
498
(SAKHA)
ET
(mm)
569
565
609
480
481
521
Irrig.
(mm)
475
456
430
432
418
404

Irrig.
(mm)
524
527
566
433
438
475
Yield
(t ha'1)
5.60
4.40
3.20
6.00
4.80
3.58

Yield
(T/Ha)
9.80
8.70
7.50
10.40
9.20
8.00
SL
(days)
118
106
96
118
106
96
MAIZE
SL
(days)
120
114
114
120
114
114
ET
(mm)
288
246
206
260
219
181
(GIZA)
ET
(mm)
583
608
665
500
525
577
Irrig.
(mm)
250
209
174
220
183
151

Irrig.
(mm)
543
567
624
462
482
534
        The cultivar used at Sakha was "Giza-2" and at Giza, "D.C. 202." Similar results were obtained with
        cultivars "H-204," "Cairo 1," and "Pioneer 514."

        The cultivar used at Sakha was "Sakha-8" and at Giza, "Giza-156." Similar results were obtained with
        the cultivar "Mexipak-65" at Giza.
                                            EGYPT-12

-------
Table 6.         Simulated maize and wheat yields under baseline climate and GCM climate change scenarios
                at Sakha and Giza.
                                     Simulated maize yield (t ha'1}
Site
Sakha



Cultivar
Giza-2
H.204
Cairo-1
Pio. 514
BASE
330
10.25
10.60
7.61
10.15
GISS
330
7.96
8.16
7.61
7.79
GISS
555
8.41
8.62
8.04
8.23
GFDL
330
7.56
7.85
7.24
7.44
GFD
L
555
7.98
8.28
7.65
7.86
UKMO
330
8.38
8.63
8.01
8.21
UKMO
555
8.46
8.71
8.07
8.28
     Giza      D.C.202
                9.83
            6.96    7.35       7.73     8.16
                                      3.20
                              3.38
                                     Simulated wheat yield (t ha"1)
     Sakha     Sakha-8
                4.16
            3.03    3.20       3.21     3.41
                                      3.27
                              3.46
     Giza
Giza 156
Mexipak65
5.57
5.42
3.22   3.59
3.17   3.52
3.44
3.38
3.82
3.74
1.32
1.30
1.49
1.46
330: Effects of climate alone on crop yield
555: Physiological effects of CO2on crop yield
                                              EGYPT-13

-------
Table 7.         Simulated wheat and maize yields under the GISS transient scenarios. The physiological
                effects of CO2 on crop yield were included.
                      Crop
Site
Scenario
Yield (t ha'1)
MAIZE Sakha* BASE
2010
2030
2050
Giza** BASE
2010
2030
2050
WHEAT Sakha* BASE
2010
2030
2050
Giza** BASE
2010
2030
2050
10.25
8.76
7.93
8.15
9.80
8.99
8.29
7.74
4.40
4.79
4.47
3.89
5.60
5.48
4.31
3.58
       Maize cultivar shown at Sakha "Giza-2". Similar results were obtained with "H-204," "Cairo-1," and
       "Pioneer-514." Wheat cultivar simulated at Sakha "Sakha 8."
       Maize cultivar simulated at Giza "D.C.202."  Wheat cultivar simulated at  Giza "Giza-156"; similar
       results were obtained with "Mexipak-65."
                                            EGYPT-14

-------
                                           EGYPT
Figure 1.      Map of Egypt and location of the crop modeling sites.

-------
    YIELD (T/HA)
              SAKHA
                             SITE
GIZA
                     CLIMATE SCENARIOS
        IBASE330  ^GISS330   E^GISS555   ^GFDL330
        IGFDL555  OUKMOSSO  ^UKMO555
Irrigated simulation
       Figure 2.    Simulated maize yield under GCM 2xCO2 climate change scenarios.

-------
  YIELD (T/HA)
0
             SAKHA
                             SITE
GIZA
                     CLIMATE SCENARIOS
        IBASE330 I^GISS330  ^GISS 555  MGFDL330
        IGFDL555 OUKMO330 ^UKMO 555
 Irrigated simulation
     Figure 3.    Simulated wheat yield under GCM 2xCO2 climate change scenarios.

-------

-------
IMPLICATIONS OF CLIMATE CHANGE FOR MAIZE YIELDS IN ZIMBABWE






                              Paul Muchena




                  Plant Protection Research Institute, Zimbabwe
                              ZIMBABWE-1

-------
TABLE OF CONTENTS

SUMMARY
INTRODUCTION
       Description of the Agricultural Regions
       Farming Sectors, Management Systems and Crops in Zimbabwe
       Food Trade and Vulnerabilities
       Objectives
METHODS
       Baseline climate data
       Climate Change Scenarios
       Crop Model, Inputs and Simulations
       Validation of the crop model
RESULTS
       Sensitivity analysis
       Maize yields under GCM climate change scenarios
       Adaptations to climate change
CONCLUSIONS
REFERENCES
                                     ZIMBABWE-2

-------
SUMMARY

       The potential impact of climate change on maize production in Zimbabwe was evaluated by simulating
crop production under climate scenarios generated by General Circulation Models (GCMs). The baseline
climate data for each site was also modified by increasing daily air temperatures by 2C and 4C for a
sensitivity analysis. Maize yields decreased under the GCM climate change scenarios, even when the direct
beneficial effects of CO2 were included in the simulation. Temperature increases of 2C and 4C reduced
maize yields at all sites.
       If these results are realized, the impact on national maize production in Zimbabwe would be severe.
Many farmers would not find maize economical to produce. Adaptation results suggest that major changes in
the farming system might ameliorate some losses, but the costs of additional fertilizer, seed supplies, and
irrigation could be high.
INTRODUCTION

Description of the Agricultural Regions

        Zimbabwe is situated in southern Africa between the latitude of 15 and 23 S and the longitude of
25 and 33 E, and occupies a land area of 390,759 square kilometers. Although Zimbabwe lies within the
tropics, only one-fifth of the country's area experiences typical tropical climate. The rest  of the country
includes areas of altitude between 600 and 1,200 meters above sea level (three-fifths of the  total area) and
areas above 1,200 meters (one-fifth of the total area). Agricultural land constitutes 85% of the land resources
in  the country. The land is divided into  five natural agroecological  regions on the basis  of rainfall and
temperature (Figure 1).

Farming Sectors, Management  Systems and Crops in Zimbabwe

        The agricultural land is used by large- and  small-scale commercial sectors, and by communal and
resettlement sectors. The large-scale commercial sector includes private and state enterprises (39% of the total
agricultural land). Private enterprises (about 4,500 large-scale farmers) hold title deeds to the land. The large-
scale cropping system is intensive and mechanized, using modern production technologies over large areas of
monoculture. Summer crops are irrigated  when necessary, and winter  crops are always irrigated. The small-
scale commercial sector consists of small private farms in which the farmers hold title deeds to their farms (4%
of the total agricultural land).
        The communal sector comprises about 900,000 farming families (4.5 million people; 49% of the total
 agricultural  land).  Individual farmers hold family rights to limited  areas for cultivation  (2.5 ha/family).
 Monoculture of maize, the staple crop, generally dominates the system. The cropping system  is similar to the
 small-scale sector, although the farms are smaller and the production capacity is comparatively  less. About
 40,000 families have been resettled on about 8% of the total agricultural land. As in the case of the communal
 sector, the resettlement sector has socioeconomic constraints due to poor planning and management.
         Maize is the primary food crop food in Zimbabwe and occupies about half of the total  agricultural
 crop land. It is grown by all of the farming  sectors, although the productivity varies largely among them (mean
 yields of 3.21 ha'1 and 1.11 ha'1 in the commercial and communal sectors, respectively). Eighty percent of the
 area sown with maize is communal. Other important crops are small grain cereals (sorghum, finger millet, and
 pearl millet); oil-seed crops (soybeans, groundnut, and sunflowers); and important cash crops (tobacco, cotton,
 coffee, tea, and horticultural crops). The small grains use about 20% of the total crop land and are generally

                                           ZIMBABWE-3

-------
 grown by communal farmers.

 Food Trade and Vulnerabilities

        Crop production in Zimbabwe usually exceeds demand, especially in years when rainfall is not limiting.
 However, when rainfall is below average, production in the marginal regions is often not enough to meet local
 demand, and the population can experience food deficits.

 Objectives

        During the last 10 years, the frequency of drought has increased in Zimbabwe, causing crop yields to
 be below average. If climate change brings about warmer and dryer conditions in the region, crop production
 may be seriously damaged. For an adequate impact study, it is important to  select sites representing the
 different agroecological regions of Zimbabwe. For this study, these three sites were selected: Banket, Gweru,
 and Chisumbanje. The sites represent high, low, and marginal productivity regions, respectively. They were also
 chosen because the long-term,  daily weather data sets (solar radiation, maximum  and  minimum air
 temperatures, and precipitation) were available to run the crop simulation model.
        The specific objectives of the study are to evaluate: (1) potential climate changes that might occur
 from doubled CO2; (2) the possible impact of climate change on maize yields; and (3) the sensitivity of maize
 yields to temperature increases (+2C or +4C).

 METHODS

 Baseline Climate Data

        Figure 2 shows the daily observed climate data (precipitation, solar radiation, maximum and minimum
 air temperatures) for Banket  (1968 to 1988),  Gweru  (1960 to 1987), and Chisumbanje (1962  to 1988).
 Chisumbanje has the warmest average annual temperature. The seasonal precipitation regime is similar at all
 sites (wet in the summer and dry in the winter), but there are large differences in the total precipitation among
 the sites. Chisumbanje is the driest  site with average annual precipitation of about 560 mm.

 Climate Change Scenarios

        Using the GCMs, the observed climate data were modified to create climate change scenarios for each
 site. The GCMs used were developed by the Goddard Institute for Space Studies (GISS) (Hansen et al. 1983),
 the Geophysical Fluid Dynamics Laboratory (GFDL) (Manabe and Wetherald 1987), and the United Kingdom
 Meteorological  Office (UKMO) ((Wilson and Mitchell  1987). Daily observed climate values were modified
with the monthly outputs of the GCMs for the particular gridbox where the site is located.
        The climate change  scenarios  created  for each site show major changes from the current climate
 (Table 1). Temperature increases are very similar in the  three sites. The GFDL scenario produces the lowest
 temperature increase (3.6C), while the UKMO scenario produces the largest (5C). Precipitation changes
vary among the sites: the GFDL scenario is the driest. Considering the temperature increases projected by the
 GCMs and  their possible effects on evapotranspiration, the climate change scenarios  could imply water
shortages, especially in the sites that currently have low precipitation, such as Chisumbanje.
        The baseline climate data for each site was also modified by uniformly increasing the values of the
daily air temperatures by +2C and +4C.
                                          ZIMBABWE-4

-------
Crop Model, Inputs and Simulations

       The maize simulation model used was the CERES-Maize model (Jones and Kiniry 1986). The model
simulates crop responses to changes in climate, management variables, soils, and different levels of CO2 in the
atmosphere. The  software used to run the programs was developed by the  Decision Support System for
Agrotechnology Transfer (DSSAT) and includes database management, crop models, and application programs
(Jones et al. 1990).
       Management variables.  Maize in  Zimbabwe is  mainly  rainfed and partially fertilized.  For the
simulations, maize was not irrigated and nitrogen was applied four and eight weeks after planting (60 kg ha'1
each time). Additional simulations to evaluate potential adaptive strategies considered possible changes  in
management variables. For the irrigation simulation, the water demand was  calculated assuming 100%
efficiency of the automatic irrigation system; a 1-meter irrigation management depth; automatic irrigation when
the available soil water is 50% or less of capacity; and the soil water for each layer is re-initialized to 100%
capacity at the start of each growing season. The plant population was higher in the irrigated simulations (7
plants m"2) than in the rainfed simulations (5 plants m"2).
        Soils. Zimbabwe's soils are predominantly derived from granite and are often sandy and light-textured,
with low agricultural potential due to low nutrient content (especially nitrogen and phosphorus). Nevertheless,
in all regions, there is a significant portion of soils with a heavier clay content that is  more suitable for crop
growth. The representative soils in Banket and Gweru are medium sandy loam soil and in the smaller region
of Chisumbanje are medium silt loam (as described in Jones et al. 1990).
        Physiological CO2 effects. Since the scenario climate change has higher levels of CO2 than the current
climate,  the CERES-Maize model includes an option to simulate the physiological  effects of CO2  on
photosynthesis and water-use efficiency that results in higher yields (Acock and Allen 1985).
        Simulations. For all climate scenarios included in this study (GCMs, sensitivity, GISS transient, and
UKMO scenarios for adaptation), maize growth was simulated under the conditions of climate change alone
and under the conditions of climate change with the physiological effects of CO2 on crop growth and yield.

Validation of the Crop Model

        The CERES-maize model was validated at Banket and Gweru using local experimental and climate
data. The experimental data included soil fertility before planting, cultivar, planting date, phonological growth-
stage components and growth analysis, harvesting date, and final yield components. Experimental  crop data
and climate were used for the 1987-88 season in Banket, and for the 1985-86 and 1986-87 seasons in Gweru.
In Banket, the observed yield was 13.6% lower than the simulated yield, and the observed season length was
1.6% shorter than the simulated season length.  In Gweru, the mean observed yield was 2% lower than  the
simulated yield and the observed season length was 1% longer than the simulated season length. The results
indicate that CERES-Maize is adequate to simulate maize growth under the conditions of this study, especially
to evaluate changes in phenology and relative changes in crop yields.

 RESULTS

 Sensitivity Analysis

         A temperature increase of 2C over the baseline climate reduced maize yields by 8% to 14% and by
 24% to 27% when the baseline temperature was increased 4C (Table 2). The largest yield decreases were in
 Chisumbanje, suggesting that, at this site, maize is already close to the upper temperature limit for growth.
 The increase in temperature reduced maize yields because the growing season was shortened and the crop  had

                                           ZIMBABWE-5

-------
 less time for grain-filling and biomass accumulation. When the direct effects of CO2 were added, maize yields
 increased slightly, but not enough to compensate for the adverse effects of the temperature increase.

 Maize Yields under GCM Climate Change Scenarios

        In Banket, simulated maize yields under the GISS, GFDL, and UKMO scenarios were reduced by
 21%, 14%, and 28%, respectively. In Gweru, the reductions were 28%, 26%, and 50%, and in Chisumbanje,
 they were 27%, 25%, and 39% (Figure 3). These yield reductions were a consequence of the shortening of the
 growing season induced by higher temperatures. The physiological effects of CO2 caused maize yields to
 increase at all sites, due to the higher photosynthetic rate of the crop with the elevated CO2. In Chisumbanje,
 the probability of obtaining an acceptable yield (e.g., 2.51 ha"1) decreased with an increase of +2C and under
 the UKMO scenario (Figure 4).

 Adaptation  to Climate Change

        We tested  possible adaptation strategies to climate change at Gweru because, under the UKMO
 scenario, it projects the largest yield decreases. Two high-cost adaptation strategies were tested: (1) increased
 fertilization (double the amount of nitrogen) and (2) a combination of increased fertilization and irrigation
 (simulated under the automatic irrigation option of the CERES model)  (Figure 5). Increased fertilization
 alone  had a small,  positive effect on the. yield  under the UKMO scenario. However, even if increased
 fertilization was combined with irrigation (nonlimiting water), the combined positive effect was not enough
 to fully compensate for the yield losses under the UKMO  scenario.

 CONCLUSIONS

        The results show that even when the physiological effects of CO2 were included, modeled maize yields
 decreased significantly at  all sites, as a result of climate change. The effect of climate change on  maize
'production may force some farmers to switch to other crops with higher thermal requirements and drought
 tolerance, such as sorghum, pearl millet, and finger millet.  Although the results suggest that farmers may be
 able to offset some of the yield losses by adaptation, this  solution may not be a very realistic one. This is
 because the costs of additional fertilizer, seed supplies, and irrigation may be high in a country with very
 limited financial resgurces.
                                           ZIMBABWE-6

-------
REFERENCES

Acock, B., and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
       and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation. DOE/ER-0238. U.S.
       Department of Energy. Washington, D.C. pp. 53-97.

Hansen, J., G. Russell, D. Rind, P. Stone, A Lascis, S. Lebedeff, R. Ruedy, and L.Travis. 1983. Efficient tree-
       dimensional global models for climate studies: Models I and II. Monthly Weather Review 3:609-622.

Jones, J.W., S.S. Jagtap, G.Hogenboom, and G.T. Tsuji. 1990. The structure and function of DSSAT. pp 1-14.
       In: Proceedings of IBSNAT SYMPOSIUM: Decision Support System for Agrotechnology Transfer.
       Las Vegas. N.V. 16-18 Oct. 1989. University of Hawaii, Honolulu, Hawaii.

Jones,  C.A., and J.R. Kiniry (eds.).  1986. CERES-Maize:  A Simulation Model  of Maize Growth and
       Development. Texas A&M University Press. College Station, TX. 194 pp.

Manabe,  S., and R. Wetherald. 1987. Large-scale changes of soil wetness induced by  an increase in
       atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.

Wilson, C.A, and J.F.B. Mitchell. 1987. A Doubled CO2 Climate sensitivity Experiment with a Global Climate
       Model Including a Simple Ocean. JGR 92:13315-13343.
                                         ZIMBABWE-7

-------
Table 1.        Annual temperature differences and precipitation ratios under GCM scenarios at selected
               sites in Zimbabwe.
Temperature Changes

BANKET
GWERU
CHISUMBANJE
GISS
4.5
4.6
4.5
GFDL
3.7
3.7
3.6
UKMO
4.8
5.0
3.9
                                                 Precipitation Ratios
         BANKET
         GWERU
         CHISUMBANJE
GISS
 1.18
 1.22
 1.24
GFDL
 1.04
 0.93
 1.05
UKMO
  1.10
  1.59
  1.09
                                         ZIMBABWE-8

-------
Table 2.
Sensitivity analysis of CERES-Maize to temperature and CO2 changes in Zimbabwe.
                                 Simulated Yield (T/Ha)
                                                     Simulated Season Length
                                                             (days)
CO2 level
(ppm)
330


555


Temp.
(Q
0
2
4
0
2
4
Banket
4.93
4.40
3.68
5.25
4.70
3.95
Gweru
3.72
3.41
2.83
4.41
3.99
3.29
Chisumb.
2.86
2.47
2.09
3.09
2.63
2.22
Banket
120
107
96
120
107
96
Gweru
135
115
101
135
115
101
Chisumb.
97
90
87
95
88
85
                                        ZIMBABWE-9

-------
ZIMBABWE

-------
                            BANKET
           TEMPERATURE (C)
                                        PRECIP (MM/MONTH)
                                                       200
                                                       150
                                                       100
           JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
                             MONTH

                 	TEMPERATURE  ESi PRECIPITATION
                            GWERU
           TEMPERATURE (C)
                                        PRECIP (MM/MONTH)
                                                       200
                                                      - 160
                                                        100
           JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
                             MONTH
                     TEMPERATURE
                                   3 PRECIPITATION
                         CHISUMBANJE
            TEMPERATURE (C)
                                          PRECIP (MM/MONTH)
                                                         200
                                                         150
                                                         100
             JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
                               MONTH

                   	TEMPERATURE   ES3 PRECIPITATION
Figure 2.        Baseline climate for Banket, Gweru, and Chisumbanje.

-------
                                          BANKET, ZIMBABWE
                            YIELD (T/HA)
                                   GISS
        GFDL           UKMO


I BASE  CD CC  CD CC'OE
                                             GWERU. ZIMBABWE
                              YIELD (T/HA)
                                    GISS
                                                   GFDL            UKMO


                                           I BASE  CUCC  CD CODE
                                        CHISUMBANJE, ZIMBABWE
                             YIELD (T/HA)
                                    GISS
          GFDL            UKMO


 I BASE  CD CC  CD CC'DE
Figure 3.       Simulated maize yields under base climate and GCM climate change scenarios. CC indicates
               simulations under climate change alone; CC+DE indicates simulations under climate change
               including the direct effects of CO2 on maize yield.

-------
m
O
cc
a.
o
             DISTRIBUTION OF MAIZE YIELDS

              FOR CHISUMBANJE, ZIMBABWE
                                i i  i i  i i  i i  i i  i i  i i  i i  i i  i i  i i  t i
                                                                     BASEUNE


                                                                      *;

                                                                     +2degC


                                                                      X

                                                                     +2 deg C, -20% P
      0.00    0.40    0.80
                                YIELD, T/HA
 Figure 4.     Cumulative probability of maize yields under baseline, +2C and -20% precipitation, and the

            UKMO GCM scenario at Chisumbanje. Acceptable yield equals 2.5 T/ha.

-------
                           CHANGE IN  MAIZE YIELD
                       WITH  HIGH COST ADAPTATIONS
             % CHANGE (BASE-3.72 MT/HA)
                                                                  GWERU
                    UKMO
+FERT
+FERT+IRRIG
Figure 5.      Maize yield changes from baseline under the UKMO scenario considering possible adaptation
             strategies at Gweru. The direct effects of CO2 on yield were included  in the simulation.
             +FERT indicates  simulations with double  the amount of nitrogen fertilizer (120 kg/ha
             applied at 4 and 8 weeks after planting). +FERT +IRRIG indicates simulations with double
             the amount of nitrogen fertilizer combined with full irrigation.

-------
SECTION 6: ASIA

-------

-------
IMPLICATIONS OF GLOBAL CLIMATE CHANGE FOR AGRICULTURE IN
     PAKISTAN: IMPACTS ON SIMULATED WHEAT PRODUCTION
                             Ata Qureshi
                            Climate Institute
                          Washington, DC, USA

                                and

                             Ana Iglesias
                Inst. Nacional de Investigaciones Agrarias (INIA)
                             Madrid, Spain
                            PAKISTAN-1

-------
TABLE OF CONTENTS

SUMMARY
INTRODUCTION
       Background
       Description of the Regions and Sites
       Aims of the Study
METHODS
       Observed Climate
       Climate Change Scenarios
       Wheat Growth Simulation
       Limitations of the Study
RESULTS
       Sensitivity Analysis
       Climate Change Scenarios
       Continuous vs.  Noncontinuous Soil Water Balance
       Changes in Management under Climate Change
DISCUSSION
REFERENCES
                                      PAKJSTAN-2

-------
SUMMARY

        This study used global climate models  (GCMs) and dynamic crop growth models to estimate the
potential agricultural effects of climate  change in  Pakistan. Under present climate conditions, wheat is
currently under stress due to high temperatures and arid conditions. Projected climate change caused simulated
wheat yields to decrease dramatically in the major regions of agricultural production, even under fully irrigated
conditions. Decreases in modeled grain yields were caused primarily by temperature increases that shortened
the duration of the life cycle of the  crop, particularly the grain-filling period, exerting a strong negative
pressure on yields. These decreases were  somewhat counteracted by the beneficial  physiological CO2 effects
on crop growth, as simulated in this study. Some adaptation strategies were tested that partially offset the
negative impacts of climate change on wheat yields.
INTRODUCTION

Background

        Pakistan is located between the Tropic of Cancer (23 N) and latitude 38 N and longitudes 61 E and
77 E (Figure 1). The total area of the country is 803,943 km2. The main river, the Indus, flows across the
regions of the Northwestern frontier (NWFP), Punjab, and Sind; the Baluchistan region is traversed by the
Zhob river. The Punjab region (in the great Indogangetic plain) accounts for the major part of agricultural
production (Wescoat and Leichenko 1990). Pakistan's climate is generally hot and arid, except some locations
have a distinct southwest monsoon rainy season.
        Agriculture is a main contributor to the Pakistani economy. Agricultural systems in Pakistan (arable
crops, rangelands, forestry, and fisheries) employ 52% of the labor force and account for 32% of the Gross
National Product (GNP).  Crop production accounts for about 70% of agricultural revenues and contributes
to export earnings. Cotton textile output is the most important industry. Food processing is also  a major
industry, and wheat is the staple food crop in Pakistan. The principal crops grown in Pakistan are cotton, rice,
wheat, sugarcane, fruits, and vegetables. Cotton and rice are the main export agricultural commodities. Wheat
is of very high economic value and represents a large proportion of the food in the region. Therefore, wheat
was selected for this climat change simulation study. Wheat production occurs over a large area of Pakistan
and under very different climatic and management conditions. Temperature, precipitation, and soil moisture,
as well as frequency of heat waves and droughts, are significant factors in crop productivity.
        Crop production in Pakistan is heavily dependent on the irrigation systems that have been installed
over the past four decades. Pakistan boasts the largest system of integrated irrigation in the world, covering
75% (195,000 km2) of the cultivated land (Wescoat and Leichenko 1990). Irrigated cropland accounts for about
90% of Pakistan's agricultural production. The irrigated areas in the Indus basin and plains (154,000 km2)
account for the bulk of  the  national harvest. Another 10 million hectares are cultivated under rainfed
conditions, also known as the "barani" method of crop production. Irrigation management practices in Pakistan
vary, but in most cases  the system involved is "partial irrigation", whereby water is provided only to ensure
limited production. The potential need for increased irrigation is a concern for some areas of the region under
present circumstances. Overapplication and inadequate drainage has aggravated waterlogging in other areas.
In 1985, Pakistan suffered critical water shortages, leading to inadequate irrigation water supplies for the main
crop growing season and causing power blackouts (Wescoat and Leichenko 1992).
        In addition to irrigation, crop production in Pakistan is also increasingly dependent on chemicals, such
as fertilizers and biocides. The environmental damage resulting from this dependence is one of Pakistan's most
severe problems. The inadequate attention to water drainage has resulted in soils that suffer from waterlogging

                                            PAKISTAN-3

-------
 and salinization, reducing soil fertility on about 100,000 km2 of land. At least 400 km2 of irrigated land is lost
 to agriculture each year. In addition, more than 12,000 km2 of agricultural lands suffer from soil erosion.
        Uncertain water supplies, soil problems, and floods along the entire Indogangetic plain also cause crop
 production losses. Natural disasters such as earthquakes and landslides also account for destruction of crops
 and damage to the country's agricultural infrastructure. In the first half of the 1980s, floods caused $3 billion
 of damage to crops, buildings, roads, and other installations. Sedimentation from upper watersheds could
 completely fill several reservoirs in the not-too-distant future.
        At present, Pakistan is self-sufficient in grain production and is a net grain exporter. The decline of
 the  agricultural resource base could change the net balance. In addition, as  the current 2.9% population
 growth rate continues, the amount of cultivated land per rural habitant is expected to decrease from 0.32
 ha/person  (1983) to a projected 0.14 ha/person in the year 2010. The 1990 population was 122.7 million, and
 the projected population for 2025 is 267.1 million (Wescoat and Leichenko 1990).
        Climate change could have a critical impact in a country that already has limitations on agricultural
 and natural resources (Myers 1989; Qureshi 1989).  The enhanced greenhouse effect, due to an increased
 atmospheric concentration of CO2 and other trace  gases, is projected to lead to higher global surface
 temperatures and changed hydrological cycles  (IPCC 1990). Evidence of the potential impacts of climate
 change in the Indian subcontinent has been mostly limited to studies in northern India, where a temperature
 increase of 0.5C is estimated to reduce wheat yields by about 10%; similar increases in central India would
 probably lead to larger percentage reductions from a lower base yield (IPCC 1990; Gadgil et al 1988). Few
 studies have projected potential agricultural impacts in Pakistan.

 Description of the  Regions and Sites

        To assess the  potential impacts of climate change, four sites  were selected to represent the broad
 differences in the agroecological zones of Pakistan (Figure  1).
        The upper Indus plain in the Punjab region varies from 150-300 meters in altitude and is a fertile
        agricultural area that accounts for about two-thirds of the total national wheat production (Table 1).
        The site of Jhelum represents wheat production in the northern plateau of Punjab's agroecological
        region. A large portion of the production is dryland, but irrigation is also important.
        The lower Indus plain corresponds to the Sind province,  is lower in altitude than the Punjab,  and
        declines to sea level at the coast. There are large tracts of deserts and marshes  to the southeast, but
        there are also  agricultural areas that account for about 20% of the national wheat production. The
        site of Khanpur represents wheat production in the region,  which is conducted mainly under
        irrigated conditions.
        The northern area of Pakistan is mountainous, and wheat  production is limited. It is represented by
        the site of Gilgit in this study.
        The central west area of Pakistan is represented by the site of D.I. Khan, which is also sandy  and
        mountainous. The wheat production is  dryland.
        The region of Baluchistan falls in the great plateau (914-1220 meters in altitude). Relatively small
        amounts of wheat are produced in this region, which is unimportant  from an international trade
        perspective. Therefore, a site from this region was not included in this study.
        Since Jhelum and Khanpur together account for nearly 90% of the country's wheat production, some
of the discussion that follows will pertain  to these two sites only.

Aims of the Study
                                           PAKISTAN-4

-------
       This study analyzed the results of the simulation of wheat production, crop water use, and irrigation
demand at four contrasting locations under current climate and under climate change scenarios generated with
GCMs. The study also included a sensitivity analysis of arbitrary temperature and precipitation changes. The
physiological effects of elevated CO2 on wheat yields were simulated in all climate change scenarios.

METHODS

Observed Climate

       Climate data for the four sites were obtained from the Pakistan Meteorology Department and include
monthly temperatures and precipitation. Daily climate variables were generated from monthly temperature and
precipitation data using the weather generator supplied by IBSNAT (IBSNAT 1989). Daily solar radiation
values were generated from observed monthly sunshine values using the WGEN program (Richardson and
Wright 1984).
       Pakistan experiences extreme temperature fluctuations both  seasonally and daily, and its climate is
generally arid. The three seasons are: November to February (cool and dry), March to May (hot and dry), and
June  to October (hot and humid).  The latter is the southwest monsoon season. In the Indus valley, the
climate becomes progressively more arid from north to south. In Sind, temperatures sometimes rise to 50 C
in the summer, and precipitation can drop to below 100 mm per year.
       Figure  2 shows the observed climate for the selected sites.  The temperature regime shows strong
seasonally at all four of the sites, with the summers being very hot. Khanpur has the warmest average annual
temperatures. Gilgit is  the coolest of the sites. There are large differences in the amounts of precipitation.
Jhelum, in the southwest monsoon region, has the highest precipitation and Khanpur has the lowest, with less
than  10 mm during 10  months of the year. Gilgit is also very dry, but is slightly wetter  than Khanpur, with
maximum precipitation during the spring. D.I. Khan has a similar precipitation  pattern to Jhelum, but it has
less rainfall.

Climate Change Scenarios

       GCM Climate Change Scenarios. The GCMs compute climatic variables for different longitude and
latitude gridboxes; the predicted variables do not reflect variations within the gridbox. Mean annual changes
in climate variables from doubled CO2 simulations of three GCMs-Goddard Institute for Space Studies
(GISS, Hansen et al 1983),  Geophysical Fluid Dynamics  Laboratory (GFDL, Manabe and Wetherald 1987),
and  United Kingdom  Meteorological Office (UKMO,  Wilson and Mitchell  1987)-were  applied to the
generated daily climate to create climate change scenarios for each site. GCMs were used to create climate
change scenarios because they produce climate variables which are internally consistent; thus they allow for
comparisons between or among regions. The method used the differences between lxCO2 and 2xCO2 monthly
GCM temperatures, and the ratio between 2xCO2 and lxCO2 monthly precipitation  and solar radiation
amounts (lxCO2 refers to current climate conditions and 2xCO2 refers to the climate that would occur with
an equivalent radiative forcing of doubled CO2 in the atmosphere).
        The annual averages of the climate change scenarios show some major differences when compared
with  the observed climate (Table 2). The temperature increases  are similar at the four  sites. The GISS and
GFDL scenarios produce comparable temperature increases (3  C to 5  C) while the UKMO scenario produces
more drastic changes (5C to 7C). The precipitation changes are more variable among the sites. In general,
the GCMs predict increases in global precipitation associated with warming because warmer air can hold more
water vapor. Khanpur, the more humid site  under current conditions, shows  the largest increase  in
precipitation (UKMO  scenario). In Gilgit, all scenarios show very small precipitation increases. In Jhelum

                                           PAKISTAN-5

-------
 (Punjab region), the GISS and GFDL scenarios show very small increases in precipitation amounts, while the
 UKMO scenario shows decreases. The climate change scenarios in D.I. Khan are comparable to those in
 Jhelum.
         Because the water supply in the Indus basin depends entirely upon climate conditions within the basin,
 any decrease in precipitation could affect the quantity of irrigation water available for the crops. Wescoat and
 Leichenko (1992) used a river-basin model to show that flows in the Indus delta are sensitive to changes in
 precipitation. When the GISS (+3.2C, +30% precipitation) and the GFDL (+3.6C, +20% precipitation)
 climate changes were put into their model, water flow increased almost 30%. However, a 2C temperature
 increase combined with a 20% decrease in precipitation decreased flows into the delta by close to 50%, which
 would result in economic and ecological devastation.
        Sensitivity Scenarios. An alternative approach to analyze the possible impacts of different climate on
 crop  yield is to specify incremental changes to temperature and precipitation and to apply these changes
 uniformly to the baseline climate. We used combinations of temperature increases of 0C, +2C, and +4C
 and precipitation changes of-20%, 0%, and +20%. The scenario, +2C with a precipitation decrease of-20%,
 is interesting because it assumes a possible temperature increase combined with a large decrease in monsoon
 rain,  which is not simulated under the GCM climate scenarios.

 Wheat  Growth Simulation

        CEKES-Wheat Model. Potential changes in wheat physiological responses (yield, season length, ET,
 and irrigation demand) were estimated with the CERES-Wheat model (Ritchie and Otter 1985) under
 different climate scenarios. The CERES crop model has been validated with experimental data from many
 locations that encompass a wide range  of environments (Ritchie  and Otter  1985). The  model simulates
 physiological crop responses (water balance, phenology, and growth throughout the season) on a daily basis
 to the major factors of climate (daily solar radiation, maximum and minimum temperature, and precipitation),
 soils  (albedo and a variety of measures relating to water in the profile), and management (cultivar, planting
 date,  plant population, row spacing, and sowing depth). The choice of the CERES-Wheat model allowed us
 to compare the simulation results with other studies in different regions of the world, and therefore, view
 Pakistan's results from a global perspective.
        Simulation of the Direct COZ Effects. Higher levels of atmospheric CO2 have been found to increase
 photosynthesis and water-use efficiency, resulting in yield increases in experimental settings (Acock and Allen
 1985). Because the climate change scenario has higher levels of CO2 than the current climate GCM simulations
 (330  ppm), the CERES-Wheat model includes an option to  simulate the physiological effects of CO2 on
 photosynthesis and water-use efficiency (Peart et al. 1989) that result an increase in yield. A level of 555 ppm
 CO2 was used in the crop models. The physiological effects of CO2 were considered in all scenarios and at all
 sites.
       Management Variables. The cultivar selected for the study in the four  sites was Mexipak (Table 3)
 because it accurately represents the most common cultivars in the area. Mexipak is a classic wheat variety
 introduced in Pakistan during the Green Revolution. Mexipak (or a cultivar with similar characteristics) is a
variety that performs well almost anywhere in Pakistan, especially in the irrigated areas that produce the bulk
 of the national wheat production.  Mexipak has been calibrated and validated for many different climate
 conditions. The "genetic coefficients'1 of Mexipak were taken from  the DSSAT data base (IBSNAT  1989).
Season length and yield were compared with experimental data from the Barani Agricultural Research Institute
for the site of Jhelum. A number of new wheat varieties are currently grown in Pakistan: Chakwal-86, Rewal-
87, and Pak-81.
       The management variables used for the CERES-Wheat model were determined for each location
according to information on current practices. Wheat in Pakistan is grown under partial or deficit irrigation,
                                          PAKISTAN-6

-------
in which farmers provide water to the crop but do not seek to optimize yield per hectare. This subirrigation
phenomenon is difficult to simulate exactly with the CERES model, since the input irrigation has to be pre-
scheduled before starting the simulation. Therefore, we simulated wheat  production  under dryland and
irrigated conditions to provide a range of possible scenarios and to analyze the production changes. We
simulated irrigation  under the "automatic option" to provide  the  crop with a hypothetical,  nonlimiting
situation. The amount of irrigation water used and consequent yields obtained are overestimates, but this
approach permits us to compare the  relative changes in each site and in each scenario. For the irrigation
simulation, the water demand was calculated assuming: 100% efficiency of the automatic irrigation system; a
1-meter irrigation management depth; automatic irrigation when the available soil water is 50% or less of
capacity; and the soil water for each layer is reinitialized to 100% capacity at the start of each growing season.
        To account for the supplementary type of irrigation practiced in D.I. Khan and Khanpur, these sites
are best represented by irrigated simulations. In Jhelum, simulation of both irrigated and dryland production
is appropriate. Here, most wheat is grown under irrigation, but certain marginal areas are dryland. In Gilgit,
wheat production is marginal and basically  dryland. For dryland multi-year simulations, it is desirable to
establish the initial soil-moisture levels at planting because soil moisture is a major factor in determining final
yields. Unfortunately, we had no data available on initial soil-moisture conditions, so we tested the sensitivity
of simulated wheat yields to initial conditions. In all areas of wheat production in Pakistan, the level of
fertilization are relatively low. In Jhelum and Khanpur the soils are deeper with higher fertility and better
management practices than at the other two sites.
        Simulations. The CERES-Wheat model was run for 29 years of baseline climate and the GISS, GFDL,
and UKMO climate change scenarios under dryland and irrigated conditions.  It was also run for the scenarios
created by superimposing arbitrary temperature and precipitation changes on the baseline climate (sensitivity
scenarios). Simulations  for all  scenarios included  climate change alone  and climate change with  the
physiological effects of CO2 at all  sites. The mean and standard deviation of  the yield,  evapotranspiration,
water applied for irrigation, and crop-maturity date are simulated in the study.

Limitations of the Study

        Climate Change Scenarios.  While GCMs are useful for climate change studies, current climate models
oversimplify many aspects of the climate system, especially ocean dynamics,  cloud physics, and land-surface
hydrology. GCMs do not accurately simulate the current climate at regional scales, and they were not designed
for predictive regional studies. Therefore, the climate change scenarios created from GCM output must not
be viewed as predictions, but as examples of possible future climates for the regions under study. GCM output
analyzed by the U.S. Environmental Protection Agency indicated that the GCMs perform least well in the
Indus Valley and in four other regions of the world. Given the  seriousness  of these deficiencies, a detailed
examination of the GCM output was  undertaken in Pakistan. Different gridboxes of the GISS, GFDL, and
UKMO GCMs were compared with historic temperatures and precipitation levels of sites in the region, and
recommendations were made for the use of climate scenarios in the area (Wescoat and Leichenko 1990). Mean
annual changes of each variable were used in this study, instead of the usual monthly changes. The justification
for the use of the climate change scenarios generated by GCMs  is to have comparability among agricultural
production in different countries, and among river basins and  climate change impacts projects.
        As configured, the climate change scenarios do not alter the patterns  of events in the base climate.
Therefore, they do  not simulate  changes in the underlying variability (e.g., extended periods of high
temperature, droughts, etc.) that can be vital for crops. Dryland yields may be considerably different depending
on  whether a change in precipitation results from a  change in mean, frequency, or intensity. However, the
scenarios created for this study do result in an increased frequency of temperatures above certain thresholds.
                                            PAKISTAN-7

-------
              Also, the weather generator used in this study does not simulate observed daily weather with complete
              accuracy.
                      Crop Models. For this study, we assumed that nutrients are nonlimiting, pests are controlled, and that
              there are no catastrophic weather events. These assumptions tend to overestimate the simulated yields. We
              also assumed that technology and climatic tolerance of cultivars do not change under the conditions of climate
              change, although this is not a realistic assumption (see discussion on adaptation). The physiological effects
              of CO2 in the crop model may be overestimated because experimental results from controlled environments,
              used to calibrate the model, may not be accurate under windy and pest-infected field conditions.

              RESULTS

                      The specific objectives of this study were to measure the change in simulated crop yields under the
              three GCM scenarios and to use the data as part of a world trade economic model.

              Sensitivity Analysis

                      A sensitivity approach to climate change impact analysis is removed from  the processes which
              influence climate. However, it has the advantage of simulating a controlled experiment and therefore provides
              a good understanding  of the factors  affecting responses; furthermore, it can identify temperature and
              precipitation thresholds for crop production. We used combinations of temperature increases of 0C, +2C,
              and +4Cand precipitation changes of-20%, 0%, and +20% and superimposed them on the baseline climate.
              For each scenario, the wheat model was run under current (330 ppm) and elevated (555 ppm) CO2 levels
              (Table 4).
                      Table 4 shows that the  decrease in yields associated with a 2C temperature increase could be
              compensated  for when the direct effects of CO2 were taken into account (this result occurred at three of the
              four sites). With a 4C temperature increase, however, wheat yields were significantly lower (particularly at
              Jhelum), even with the direct effects of CO2. These yield decreases were a consequence of a shorter season
              length, and in some cases, water stress to the crop. A temperature of 2C above the baseline implied that the
              season length would be reduced by 8-14 days at all sites tested; a 4C increase reduced it by about 20 days at
              all sites.
                      An increase in  the temperature also increased the moisture stress of the wheat crop (i.e.,  the
              difference between precipitation and evapotranspiration). The simulated evapotranspiration (ET) decreased
              in most cases, despite large increases in temperature and in potential evaporation. This was due to the shorter
              growing season, which reduced the total amount of ET, and by decreased demand by the crop because it was
              not growing as well. Irrigation demand generally decreased for the same reasons. Most of the sites' modeled
              wheat yields were not very sensitive to precipitation changes (see section on continuous vs. noncontinuous
              water balance).

              Climate Change  Scenarios

                      Currently, high temperatures and low precipitation limit present-day wheat production in Pakistan.
              The sensitivity study showed that temperature increases of 2C and higher resulted in a reduction of crop
              yields throughout the country,  especially  in the main wheat-producing region of Punjab. In this section we
              consider the  possible  impact of the climate change scenarios generated  by the three  GCMs on wheat
              production in four sites distributed throughout Pakistan. The results show the degree to which the impacts
              may vary within the regions. Table 5 shows the yield changes under doubled CO2 climate change scenarios for
              all of the sites for both dryland and irrigated conditions with and without the direct effects of CO2. At Jhelum

                                                        PAKISTAN-8
_

-------
and Khanpur, the two major sites, irrigated and dryland results were very similar, showing major reductions-up
to 80%in yields. Reductions in yields were projected with all three scenarios, but the reductions were more
significant with the UKMO scenario. Even when the direct effects of CO2 are included, reductions in yields
were still generally significant. At D.I. Khan and Gr/gzY, results were mixed, with some increases in yields.
        Yield decreases were driven primarily by the increase in temperature, which causes the duration of
the crop-growth stages (particularly the grain-filling period) to be shortened. The season length was greatly
reduced by about four weeks under the three scenarios (Table 6). Tables 7 and 8 show that simulated ET and
irrigation demand generally decreased under the climate change scenarios.

Continuous vs. Noncontinuous Soil Water Balance

        The assumption that the soil-moisture profile was full at the beginning of each growing season is one
that may be acceptable under  moist conditions, but it is unrealistic in dry or marginal areas, leading to an
overestimation  of the  available soil water.  To determine how sensitive the yield results are to soil-water
conditions, a series of tests was run using a continuous water balance in addition to the normal method of
resetting the soil moisture to full every growing season. Under the continuous water balance method, the soil
moisture is never reset and is allowed to continue, as computed, from one year to the next. These experiments
were carried  out at Jhelum and D.I. Khan for the 29 years under  both the current climate and the GISS
climate change  scenario.
        At Jhelum, the continuous water balance made very little difference on crop yields. This is because
Jhelum receives copious amounts of monsoon rainfall just prior to planting. At D.I.  Khan, where precipitation
amounts are lower, the continuous water balance resulted in decreased yields under current climate conditions.
However, the GISS scenario yields also decreased, so the relative changes from base yields remained similar
(+16.3% for the normal reset run; +21.6% for the continuous soil-moisture run). Only relative yield changes
were contributed to the world food trade study.

Changes in Management under Climate Change

        This  study determined the possible causes of losses in wheat yields under different climatic conditions
(GCM climate  change scenarios and sensitivity analysis)  a$ the first step in the  analysis of adaptation to
climate change. Adaptation strategies included changes  in  the management practices of the wheat cultivars
currently used  (planting date and irrigation) and changes in the cultivar.  All adaptation strategies were
simulated in Jhelum  under UKMO, the most unfavorable climate change scenario (Table 9). The results of
the simulation show  that there was no significant benefit from  full  irrigation at this site under the UKMO
scenario (a -61% yield decrease in dryland conditions and a -56% yield decrease under full irrigation). This
result was not very surprising, since Jhelum was found to be insensitive to changes in precipitation.
        A likely response of farmers to warmer temperatures would be to shift the planting dates to avoid high
temperatures during the grain-filling period and increase  the crop-growing season. In Jhelum, a simulated shift
in the planting  date to 10 days earlier (from Oct. 20 to Oct. 10) increased the yield losses under the UKMO
scenario (Table 9), probably because the new planting date falls in a very hot  period. Nevertheless, a delay in
the planting date to  Nov. 1 or Nov. 15 was beneficial for wheat yields in Jhelum (a -61% yield loss under
present management conditions and a -30% yield loss if the planting is postponed to Nov. 15).
        Two  cultivars were tested as possible alternatives to Mexipak: Sonalika (widely used in India) and
Anza. Sonalika, planted on Nov. 15, resulted in larger yield losses than Mexipak planted on Nov. 15 (a -42%
yield decrease with Sonalika and a -30% with Mexipak). Anza,  planted on Nov. 15,  was the most favorable
strategy tested.  In this case, yields decreased under the UKMO scenario by -20%.
                                           PAKISTAN-9

-------
DISCUSSION

        Conclusions. Projected climate change caused simulated wheat yields to decrease dramatically both in
dryland and irrigated production, due to a shorter season length caused by temperature increases. Yield
decreases were somewhat counteracted by the physiological effects of CO2, as simulated in this study, but the
effect was not enough to offset yield decreases under the climate scenarios. The overwhelming impact of
temperature increases, especially when coupled with precipitation decreases, would reduce the time required
for crop maturation and thus reduce wheat yields. If global climate change brings warmer and drier conditions
to Pakistan, additional problems would arise for water supplies for irrigation in the region.
        In comparison to simulated climate change agricultural impacts in  other hot and dry regions, the
Pakistan study found more severe yield decreases  in response to the same climate change scenarios tested
elsewhere (Adams et aL 1990; Smith and Tirpak 1989; Rosenzweig 1990). It is possible that Pakistan may be
one of the regions more severely affected by climate  change, and national wheat production may decrease
substantially in the future  if climate change  occurs as predicted by GCMs. This study suggests that the
development of new cultivars that are more heat-resistant than the present ones is essential to improve the
resiliency of Pakistani wheat production to possible increases in temperature.
        The adaptation strategies tested in Jhelum under the UKMO scenario did not compensate fully for
the yield reductions, but a shift in cultivar combined with a later planting date decreased yield losses under
irrigated conditions. With the known pool of cultivars and current resources  (water and  fertilizer), full
adaptation seems unlikely in Pakistan since the temperature changes suggested by some GCM scenarios would
exceed the temperature the wheat plant can tolerate physiologically.
        Further research needs. Possible future adaptations need to be analyzed so that farmers will be able
to adapt their practices  in  response to altered climate conditions. These studies should include detailed
changes in planting date, fertilizer, and  irrigation. Studies should also focus on the effect  of the  altered
calendar crop on the agricultural system in the region, particularly where more than one crop is harvested each
year.
        The possible increase in climatic variability under future climate change scenarios may be an important
factor in determining production losses. Since crop yields exhibit a nonlinear response to heat and cold stress,
any change in the probability of extreme temperature events can be significant. Therefore, a variability study
on possible future yields is important. Because insect pests and diseases are also an important component of
production losses under the present climate, any possible change in pest and disease patterns under climate
change conditions should be analyzed.
        Finally, it is important to analyze the potential extension of, or shifts, in the areas of crop production,
the soil constraints in these potential agricultural regions, economic tendencies, and other  environmental
factors. Based on the results  of the present study, the effects of potential climate change in Pakistan could have
dramatic consequences for agricultural production and it is important to analyze the impact in a larger number
of sites and for other major crops, such as rice. Finally, further research should be done  on  simulating the
annual water balance more accurately in order to estimate future water availability for irrigation.
                                           PAKISTAN-10

-------
REFERENCES

Acock, B., and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
        and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation. DOE/ER-0238. U.S.
        Department of Energy. Washington, D.C. pp. 53-97.

Adams, R.M., C. Rosenzweig, R.M. Peart, J.T. Ritchie, B.A McCarl, J.D. Glyer, R.B. Curry, J.W. Jones, KJ.
        Boote, and L.H. Allen, Jr. 1990. Global climate change and U.S. agriculture. Nature 345(6272):219-
        224.

Gadgil, S., AK.S. Huda, N.S. Jodha, R.P. Singh, and S.M. Virmani. 1988. The effects of climatic variations on
        agriculture in dry tropical regions of India. M.L. Parry, T.R. Carter and N.T. Ronijn (eds.) The Impact
        of Climatic Variations on Agriculture.  Chapter 5. pp. 497-576.

Hansen, J., G. Russell, D. Rind, P. Stone, A Lascis, S. Lebedeff, R. Ruedy, and L.Travis. 1983. Efficient three-
        dimensional global models for climate studies: Models  I and II. Monthly Weather Review 3:609-622.

IBSNAT.  International Benchmark Sites Network for Agrotechnology Transfer Project. 1989.  Decision
        Support System for Agrotechnology Transfer Version 2.1 (DSSAT V2.1).  Dept. Agronomy and Soil
        Sci., College of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii 96822.

IPCC. 1990. Climate Change: The IPCC Scientific Assessment. J.T. Houghton, G.J. Jenkins, and J.J. Ephraums
        (eds).  Intergovernmental Panel on Climate Change. Cambridge.

Manabe, S., and R. Wetherald.  1987.  Large-scale changes of soil wetness induced by an increase in
        atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.

Myers, N. 1989. Environmental Security: The Case of South Asia. International Environmental Affairs, 1:138-
        154.

Peart, R.M., J.W. Jones,  R.B. Curry, K. Boote, and L.H. Allen, Jr. 1989. Impact of climate change on crop
        yield in the southeastern U.S.A. In J.B. Smith, and D. Tirpak (eds.). The Potential Effects of Global
        Climate Change on the United States. Report to Congress. EPA-230-05-89-050. U.S. Environmental
        Protection Agency. Washington, DC. Appendix C-l, pp. 2-1 to 2-54.

Qureshi, A 1989. Mitigating climate change: strategies to finance retention of tropical forests. In Coping with
        Climate  Change-Proceedings  of the  Second North American Conference  on Preparing for Climate
        Change: A Cooperative Approach. J.C. Topping, Jr., ed. The Climate Institute, Washington, D.C.

Richardson, C.W., and D.A Wright. 1984. WGEN: A Model for Generating Daity Weather Variables. ARS-8.
        US Dept. of Agriculture, Agricultural Research Service. Washington, DC. 83 pp.

Ritchie, J.T., and S. Otter. 1985. Description  and performance of CEKES-Wheat: A user-oriented wheat yield
        model. In  W.O. Willis (ed..). ARS Wheat Yield Project.  ARS-38. U.S. Department of Agriculture,
        Agricultural Research Service. Washington, DC. pp. 159-175.
                                           PAKISTAN-11

-------
Rosenzweig, C. 1990. Crop response to climate change in the Southern Great Plains: A simulation study.
        Professional Geographer 42(l):20-37.

Smith, J.B., and D. Tirpak (eds.). 1989. The Potential Effects of Global Climate Change on the United States.
        Report to Congress. EPA-230-05-89-050. U.S. Environmental Protection Agency. Washington, DC.
        423pp.

Wescoat, J.L., and R. Leichenko. 1990. Climate change scenarios for the Indus basin, Pakistan: Modification
        of Scenario Research Design. Report to "Complex River Basin Project Participants".

Wescoat, J.L., and R. Leichenko. 1992. Complex River Basin Management in a Changing Global Climate: The
        Indus Basin in Pakistan, A National Assessment. Collaborative Paper No. 5. Center for Advanced
        Decision Support for Water and Environmental Systems and the Institute of Behavioral Science. The
        University of Colorado. 143 pp.

Wilson, C.A., and J.F.B. Mitchell. 1987. A Doubled CO2 Climate Sensitivity Experiment with a Global Climate
        Model Including a Simple Ocean. Journal of Geophysical Research 92:13315-13343.
                                         PAKISTAN-12

-------
Table 1.
Wheat production (1981-89).
          Region
               Site
Production
Area
                              Gilgit
Yield

Punjab
Sind
NWFP

Jhelum
Khanpur
D.I. Khan
txlOOO
9,345
2,189
939
ha x 1000
5,404
1,032
791
tha-1
1.73
2.12
1.19
          Baluchistan
                                        445
                      252
               1.79
          Pakistan
                                     12,918
                     7,478
               1.73
                                          PAKISTAN-13

-------
Table 2.        Average annual changes in temperature, precipitation, and solar radiation in the GCM
               climate change scenarios.
                    Temp, changes (C)
Precip. changes (%)
SR changes (%)
Site
Jhelum
Khanpur
D.LKhan
Gilgit
GISS
4.7
3.2
4.7
4.7
GFDL
4.5
4.5
4.9
4.8
UKM
O
7.0
6.0
7.0
5.7
GISS
13
25
13
13
GFDL
16
19
11
10
UKM
O
-18
40
-18
13
GISS
-1
-2
-1
-1
GFDL
8
0
8
17
UKM
0
5
-1
5
20
                                        PAKISTAN-14

-------
Table 3 Test sites, soils, and management variables.
Site
Jhelum
Khanpur
Gilgit
D.LKhan
Location
+32.97 N
+73.75 E
+28.63 N
+70.67 E
+35.90 N
+74.33 E
+31.82 N
+70.93 E
Representative
Soil
medium
sandy loam
deep
sandy loam
medium
sand
medium
sandy loam
Cultivar
mexipak
mexipak
mexipak
mexipak
Practice
dryland/
irrigation
irrigation
mainly
dryland
irrigation
Planting
date
Oct. 20
Nov. 12
Oct. 30
Nov. 30
     Plant population for irrigation practice = 250 plants m":
     Plant population for dryland practice = 125 plants m'2
                                            PAKISTAN-15

-------
Table 4.
Sensitivity of irrigated CERES-Wheat yields to temperature and CO2 changes.
                                                         Simulated Yields (t ha"1)
CO2 level
(ppm)
330


555


Temp.
Change (C)
0
2
4
0
2
4
Jhelum
5.28
3.55
2.02
6.60
4.89
3.22
Khanpur
4.79
4.19
3.23
5.80
5.28
4.29
D.I. Khan
3.73
3.50
3.16
4.41
4.19
3.86
Gilgit
4.54
3.97
3.94
5.38
4.85
4.42
                                         PAKISTAN-16

-------
Table 5.
Effects of climate change on wheat yields.
   Site
      Rainfed/  Base
         Irrig.  Yield
Percent changes in simulated yields from base

  GCM Scenario Alone       GCM Scenario + DE


 GISS   GFDL   UKMO    GISS   GFDL  UKMO

Jhelum
Khanpur
D.LKhan
Gilgit
Jhelum
Khanpur
D.LKhan
Gilgit

rainfed
rainfed
rainfed
rainfed
irrigated
irrigated
irrigated
irrigated
tha'1
4.81
4.19
1.94
0.6
5.28
4.79
3.73
4.75

-67
-45
16
20
-60
-43
-20
-17

-29
-54
-55
8
-24
-54
33
-5

-80
-61
-11
-47
-74
-57
-17
-12

-41
-15
61
67
-36
-19
-2
-6

-5
-33
-45
50
-5
-37
60
9

-61
-35
29
-20
-56
-37
6
5
   DE = direct effects of CO2 on wheat yield included.
                                       PAKISTAN-17

-------
Table 6.
Effects of climate change on the season length of wheat.
                                              Simulated season lengths (days)
Site
Jhelum
Khanpur
D.I. Khan
Gilgit
Base (days)
128
122
122
150
GISS
100
99
104
132
GFDL
105
96
98
131
UKMO
94
93
97
133
                                      PAKISTAN-18

-------
Table 7.
Effects of climate change on wheat evapotranspiration (ET).
 Site
                                          Percent changes in simulated ET from base
                                        GCM Scenario Alone          GCM Scenario +DE
    Rainfed/  Base ET
       Irrig.  (mm)
GISS    GFDL   UKMO
GISS    GFDL   UKMO
Jhelum
Khanpur
D.LKhan
Gilgit
Jhelum
Khanpur
D.I. Khan
Gilgit
rainfed
rainfed
rainfed
rainfed
irrigated
irrigated
irrigated
irrigated
207
226
226
166
232
267
331
474
-39
-41
-9
-5
-31
-31
-22
-17
1
-36
-10
-7
25
-29
21
-9
-50
-47
-20
-5
-42
-36
-19
13
-34
-35
-11
-5
-31
-32
-27
-23
2
-33
-10
-7
18
-31
14
-17
-44
-40
-19
-4
-38
-34
-24
3
 DE = direct effects of CO2 on wheat ET included.
                                       PAKISTAN-19

-------
Table 8.
Effects of climate change on amounts of irrigation.
     Site
       Base
Percent changes in simulated irrig. amounts from
                       base
  GCM Scenario Alone         GCM Scenario +DE
 GISS   GFDL  UKMO    GISS   GFDL  UKMO

Jhelum
Khanpur
D.LKhan
Gilgit
(mm)
128
199
215
337

-45
-31
-28
-8

57
-32
-43
4

-56
-36
-16
33

-43
-33
-33
-16

43
-33
-33
-7

-47
-34
-20
20
     DE = direct effects of CO2 on wheat irrig. water demand included.
                                      PAKISTAN-20

-------
Table 9.        Effect of changes in wheat cultivar and management on simulated yield under the UKMO
               climate change scenario in Jhelum. The physiological effects of CO2 were included in all
               UKMO scenario simulations.
Scenario

BASE*
BASE*
UKMO**
UKMO
UKMO
UKMO
UKMO
UKMO
UKMO
Cultivar

Mexipak
Mexipac
Mexipak
Mexipak
Mexipak
Mexipak
Mexipak
Sonalika
Anza
Management

rainfed
irrigation
rainfed
irrigation
irrigation
irrigation
irrigation
irrigation
irrigation
Planting Date

Oct. 20
Oct. 20
Oct. 20
Oct. 20
Oct. 10
Nov. 1
Nov. 15
Nov. 15
Nov. 15
Yield
T/Ha
4.81
5.28
1.88
2.35
0.64
2.56
3.71
3.08
4.25
% Change
from Base



-61
-56
-88
-52
-30
-42
-20
  *Base conditions using current climate as scenario.
 **Simulations under UKMO scenario without adaptation.
                                          PAKISTAN-21

-------

-------
         Location  of Test Sites for Pakistan
                   NORTH-WEST
                     FRONTIER
      BALUCHISTAN
                                              PUNJAB
                                         SIND
                       Arabian Sea
Figure 1.      Map of Pakistan with sites selected for this study. The denominations used and the
            boundaries shown do not imply any judgement on the legal status of any territory or any
            endorsement or acceptance of such boundaries.

-------
  CC


  CL
  o
  10
  CM
                   o      o
                   o      10      _
                   CM      T-      

         0.

         o
         UI
         en
         o.
               f
        LU
        IT
        111
                O    IO   O    IO    O    IO    O
                CD    CM   CM    -    1-
                                                       o.

                                                       o


                                                      I
                                                      o


                                                      E
                                                                               3
                                                                               C3
                                                                                                                                             E
                                                                                                                                             O


                                                                                                                                              o.
                                                                                                                                                "8


                                                                                                                                                H
                                                                                                                                                


                                                                                                                                                1
                                                                                                                                                Bk
o      o     o      o
10      o     10      o
CM      CM     T-      *-
                                         O

                                         IO
                                                       o      o      o      o
                                                       10      O      10      O

                                                       CM      CM      *-      1-
       Q.

       O
_i
UJ
       o

       ui

       
                                                               E


                                                               a.
                                                               o

                                                               s
                                                              a
                                                              o


                                                               CL
                                                                                 <
                                                                                                                                     s
                                                                                                                                     a.
                                                                                                                                               O
                                                                                                                                      a
                                                                                                                                                i
                                                                                                                                                f
                                                                                                                                                          
-------
IMPACT OF CLIMATE CHANGE ON SIMULATED WHEAT
               PRODUCTION IN INDIA
                    D.Gangadhar Rao

                 GRID A, Hyderabad, India


                       S.K.Sinha

                  IARI, New Delhi, India
                       INDIA-1

-------
TABLE OF CONTENTS

SUMMARY
INTRODUCTION
METHODOLOGY
       Baseline Climate Data
       GCM Climate Change Scenarios
       The CERES-Wheat Model, Crop Inputs and Soil Data
       Model Validation
       Simulations
RESULTS AND DISCUSSION
       Sensitivity Analysis of the CERES-Wheat model
       Wheat Responses to Climate Change at Delhi
       Wheat Responses to Climate Changes at Hyderabad
       Adaptation Strategies
       Implications of Climate Change for Wheat Production in India
REFERENCES
                                       INDIA-2

-------
SUMMARY

        This study assessed the impact of climate change on wheat production at two contrasting locations
in India, Delhi and Hyderabad, using the CERES-Wheat simulation model  under rainfed and  irrigated
conditions.  Climate change scenarios for each location were created  from three equilibrium  General
Circulation Models (GISS, GFDL, and UKMO) and from the transient GISS model (for the 2010s, 2030s, and
2050s). The simulation study considered the physiological effects of increased CO2 levels on the crop in all
scenarios. In all climate change simulations, wheat yields were smaller than those in the current climate, even
with the direct beneficial effects of CO2 on crop yield considered. Yield  reductions are primarily  due to a
shortening of the wheat-growing season, resulting from the scenario temperature  increases.
INTRODUCTION

        Agriculture remains an important economic sector in India, despite the development of industry and
services in the post-independence nation. Seventy percent of the population is rural and depends directly or
indirectly on agriculture. Agricultural failures due to natural disasters have been an important factor in  the
migration of the population from rural to urban areas. Today, agriculture represents 35% of India's Gross
National Product (GNP). This percentage is likely to decline due to further industrialization, but agriculture
will continue to occupy an important place in the Indian economy. Therefore, global climate change that could
influence agriculture in the Indian subcontinent should be a cause for concern (IPCC 1990).
        The Indian agricultural system includes the production of crops, animal products, fisheries, and forest
products. It is impossible to present a holistic picture of the effect of climate change on the agricultural system
because of the many interdependent and interactive factors. Therefore, for the purpose of the present analysis,
we considered the potential effects of climate change on the primary sector of the agricultural system-crops.
        Because of the different  agroclimatic environments in India, it is possible to grow almost all of  the
important cereal, leguminous, oilseed, horticultural, and plantation crops,  and it would be interesting to
simulate their potential production under climate change. A simulation study, however, requires reliable input
data and a reasonable model. Among the available crop models, CERES-Wheat has been widely used over
different environments, and therefore is probably suitable for the agroclimatic conditions in India. Wheat is
important in India because it constitutes the basis of its food security system. Although rice simulation models
are also available, ecological niches for rice production  in India are highly variable and would require a more
elaborate study. Therefore, the present study was confined to wheat as a test system.
        There is a need to have a long-term weather record and high-quality experimental data for validating
any model. The two locations selected for this study (Delhi and Hyderabad),   representing contrasting
agroclimates, may be a  good starting  point (Figure 1). Delhi represents the highly productive regions of
Haryana, Punjab, and Uttar Pradesh  (contributing  approximately 70% of the national  production), and
Hyderabad represents  the  less  productive regions of Andhra  Pradesh, Karnataka,  and  Maharashtra
(contributing approximately 2.6% of the national production) (Table 1).

METHODOLOGY

Baseline Climate Data

        For the simulation study  daily minimum and maximum temperature, solar radiation, and rainfall data
were used for twenty-eight years (1953-80) for Delhi  (28N;78E) and for seventeen years (1964-80)  for
Hyderabad (17 N; 78 E). The monthly averages for  these parameters are given in Figure 2.

                                             INDIA-3

-------
        Solar radiation was unavailable, but was computed using sunshine hours calculated from the following
formula:

         Rs  = Ra (a + b * n / N)   where

         Rs  = Solar Radiation in Megajoules m"2
         Ra  = Extraterrestrial Radiation
         n/N = % Sunshine Hours
         a,b = Location-specific Constants

        The constants a and b were established by the India Meteorology Department.

GCM Climate Change Scenarios

        Climate change scenarios for Delhi and Hyderabad were created by combining daily observed climate
data with the monthly output of three General Circulation Models: the Goddard Institute for Space Studies
(GISS) (Hansen et al. 1983), the Geophysics Fluid Dynamics Laboratory (GFDL) (Manabe and Wetherald
1987), and the United Kingdom Meteorological Office (UKMO) (Wilson and Mitchell  1987) (Figure 2).
Similarly, transient climate change scenarios were created for the two locations for the decades of 2010s, 2030s,
and 2050s with monthly output from the GISS transient run A (Hansen et al. 1988) (Figure 2).

The CERES-Wheat Model, Crop Inputs and Soils

        Growth and development of wheat were simulated using the CERES-Wheat model (Ritchie and Otter
1985). A common wheat genotype (Sonalika) was chosen to represent the most common cultivars used at both
Delhi and Hyderabad. In all the simulations the crop was sown on November 1, and a plant population of 200
plants per square meter was maintained. The irrigated simulations considered three irrigation treatments of
40 mm at the time of sowing, 30 days after sowing (vegetative stage), and 60 days after sowing (reproductive
stage).
       The soil characteristics used for the simulations are listed in Appendix A. For Delhi, the soils include
deep, well-drained loamy soils that exhibit a regular decline in the organic carbon with depth, indicating no
disturbance in the profile  development. The soil characteristics used for Hyderabad included deep, mildly
alkaline, moderately well-drained clayey soils.

Model Validation

       The CERES-Wheat model has been validated using datasets from various wheat-growing areas of the
world (Otter and Ritchie 1985). For the sites tested in India, the simulated phenology did not always replicate
the observed. The average yield of the variety "Kalyansona" is about 4.3 t ha"1, with adequate irrigation at New
Delhi (Sinha and Swaminathan 1991); with the same management conditions, the averaged simulated yield was
5.2 t ha"1. At Hyderabad, the average wheat yield is 2.6 t ha"1 and the average simulated yield was 2.2 t ha"1.

Simulations

       The simulations were conducted using the CERES-Wheat model with DSSAT (Decision  Support
System for Agrotechnology Transfer) (IBSNAT1989). Both rainfed and irrigated simulations were performed
under all of the  climate scenarios described above.  The physiological effects of CO2 on the crop were

                                            INDIA-4

-------
considered in each simulation.  In addition a sensitivity analysis was performed with arbitrary changes in
precipitation (+20% and -20%) and in temperature (+2C and +4C) from the baseline climate.

        The CO2 levels and the photosynthetic factors used for the climate change scenarios are:

Climate Change Scenario               CO2 Level             Photosynthetic Factor

GISS, GFDL, UKMO                  555                           1.17

2010                                  405                           1.05
2030                                  460                           1.10
2050                                  530                           1.15


RESULTS AND DISCUSSION

Sensitivity Analysis

        This section presents the sensitivity of the CERES-Wheat model to changes in precipitation (+20%
and -20%)  and temperature (+2C  and +4C). Figure 3 shows that temperature is the main  factor
responsible for yield changes. Both grain yields and biomass decline with increasing temperatures, although
there is a small yield increase when  the  rainfall is increased by  20%. Field  experiments (at the Indian
Agricultural Research Institute and other sites) indicate that wheat yields  are very sensitive to  elevated
temperatures, particularly during the grain-filling period. The simulation analysis shows similar results.

Wheat Responses to Climate Change at  Delhi

        At Delhi, the GISS and GFDL climate change scenarios project an increase in the amount of rainfall
from June to October (Figure 2). The air temperature increases under all GCM scenarios throughout the year
and particularly during the post-rainy season. No appreciable change in the solar radiation is observed under
climate change conditions. The GISS transient scenarios show gradual increases in monthly temperature of
at least 5C by 2050 and sharp increases in rainfall during July (Figure 2). Solar radiation increases slightly
in January and February during the 2030s and 2050s.
        Under the GCM equilibrium scenarios alone (not considering the direct CO2 effects), simulated wheat
grain yields were largely reduced in  comparison to the base values. Under the transient scenarios, yield
reductions were smaller (Figure 4). When the physiological effects of CO2 are included in the yield simulations,
they partially offset the  yield decreases (Figure 4). In fact, under the GISS scenario  and  the 2010 transient
climate, there was a marginal increase in the simulated grain yields under rainfed conditions. Nevertheless,
under irrigated conditions, yield decreases from the base were apparent even when the direct CO2 effects were
included. The transient scenarios predicted a greater reduction in the response of the crop to irrigation as we
pass from the 2010s to the 2050s. Simulated biomass followed a similar trend, but in this case the marginal
increase in the dryland simulations under the GISS scenario with CO2 effects were not observed (Figure 4).
        Temperature increases  under the  scenarios decreased the length of the growing period and total
evapotranspiration in the crop model simulations (Figure 4). The length of the growing season was not
affected by the physiological effects of CO2. In general, the climate change scenarios predict a greater percent
reduction in evapotranspiration under irrigated conditions.
                                             INDIA-5

-------
Wheat Responses to Climate Changes at Hyderabad

        All GCMs show an increase in the average monthly air temperature at Hyderabad. The UKMO model
projects the largest increases-up to 5C from January to March (Figure 2). The rainfall increases during the
rainy season (May to November) under all scenarios. The solar radiation shows considerable reduction in the
GFDL scenario from June to December (Figure 2) because this scenario predicts higher precipitation during
those months.  The transient scenarios show an increase in the mean average temperatures, which reach an
average of 35 C (Figure 2) during August. Rainfall shows no particular trend and solar radiation shows small
increases from June to December in 2050.
       As in the case of Delhi, simulated wheat yields and season length decreased under all GCM scenarios
of climate change (Figure 5). The largest reductions were under the GISS and UKMO scenarios. The transient
scenarios  projected a 20% decrease in the 2010s and a 60%  decrease in the 2050s. The same trend was
observed with respect to biomass production, except under the GFDL scenario. As in the case of Delhi, the
evapotranspiration was lower.

Adaptation Strategies

       Possible strategies to increase wheat yields under climate change conditions are: increasing irrigation,
using more fertilizer, or increasing the plant population. These strategies were tested at Delhi under the
equilibrium GCM scenarios, but only slightly reduced the negative impacts of climate change on simulated
yield (data not shown).
       To further evaluate possible adaptive strategies to climate change, two approaches could be tried:
(a) Evaluate the  impacts of the climate change scenarios on  other Indian  wheat cultivars. However, the
tremendous variability in germoplasm requires accurate estimates of the genetic coefficients that describe each
cultivar in the CERES-Wheat model, and these coefficients are not readily available for many of the Indian
genotypes.
(b) Design hypothetical new cultivars that will adapt to climate change conditions, using the CERES-Wheat
model as a tool. Possible breeding objectives could be established by analyzing the genetic coefficients in order
to define a theoretical cultivar that is adapted to climate change.

Implications of Climate Change for Wheat Production in India

       The results presented in this report suggest that wheat yields would decrease under equilibrium and
transient climate change  conditions at Delhi and Hyderabad. These yield decreases could have  a  serious
impact on food security, especially in view of the increasing population and its demand of food grains. Since
most of the wheat production comes from the northern plains, where it is almost impossible to increase the
present area of wheat under irrigation, climate change could lead to shortages  in food supplies. Furthermore,
the simulated changes in wheat yields could be indicative of the possible impacts of climate change on other
similar crops grown during the same season.
study:
        Before considering the above scenario, it is important to evaluate some important limitations of this
        (a) Poor simulation of the present climate by the GCMs in this region.

        (b) Poor validation of the CERES-Wheat model for some of the Indian wheat cultivars. For example,
        the cultivar "Kalyansona", under irrigation at Delhi, has an average growing period of 150-160 days,
                                            INDIA-6

-------
        but the simulated growing period was always under 116 days in the present study. In addition, the yield
        components are not simulated adequately.

        Because of these uncertainties, it would be premature to draw final conclusions on the impact of
climate change on India's food security system. However, the results of this simulation study suggest possible
negative consequences of climate change for the crops in India, and  thus, further research needs to be
encouraged through both national and international efforts. It is essential to conduct further research using
the wheat model, and efforts are needed to develop  models for crops such as chickpeas, rapeseed, mustard,
and barley. This would require extensive field experiments to generate crop phenology data.
                                              INDIA-7

-------
 REFERENCES

 Hansen, J., G. Russel, D. Rind, A. Lacis, S. Lebedeff,  R. Ruedy, and L. Travis. 1983. Efficient Three
        Dimensional Global Models for Climate Studies: Models I and II. Monthly Review, Vol III, No. 4:609-
        662.

 Hansen, J., L Fung, D. Rind, S. Lebedeff, R. Ruedy, and G. Russel.  1988. Global Climate Changes as
        Forecasted by Goddard Institute For Space Studies Three Dimensional Model. Journal of Geophysical
        Research, 93:9341-9364.

 BBSNAT. International Benchmark Sites Network for Agrotechnology Transfer Project  (IBSNAT).  1989.
        Decision Support System for Agrotechnology Transfer Version 2.1 (DSSAT V 2.1). Dept. Agronomy
        and Soil Science., College of Trop. Agr. and Human Resources, University of Hawaii, Honolulu,
        Hawaii 96822.

 IPCC. 1990. Intergovernmental Panel on Climate Change. First Assessment Report.

 Manabe, S., and R.T. Wetherald. 1987. Large-Scale Changes in Soil Wetness Induced by an Increase in Carbon
        Dioxide. Journal of Atmospheric Sciences, 44:1211-1235.

 Otter, S., and J.T. Ritchie. 1985. Validation of the CERES-Wheat Model in Diverse Environments. In: Wheat
        Growth and Modeling, eds. W. Day and R.K. Atkins. Plenum Press. New York, pp. 307-310.

Ritchie, J.T., and S. Otter.  1985. Description and Performance of CERES-Wheat: A User-Oriented Wheat Yield
        Model In: W.O. Wills, ed. ARS Wheat Yield Project. Washington, DC:  U.S. Dept. of Agriculture,
        Agricultural Research Service. ARS-38.pp. 159-175.

Sinha, S.K., and M.S. Swaminathan. 1991. Deforestation, Climate Change and Sustainable Nutrition Security:
        A Case Study of India. Climate Change, 19:201-209.

Wilson, CA, and J.F.B. Mitchell. 1987. A Doubled CO2 Climate Sensitivity Experiment with a Global Model
        Including a Simple Ocean. Journal of Geophysical Research, 92:13315-13343.
                                           INDIA-8

-------
Table 1.
Area and production of wheat in various states of India.
State
Andhra Pradesh
Arunachal Pradesh
Assam
Bihar
Gujarat
Haryana
Himachal Pradesh
Jammu&Kashmir
Karnataka
Madhya Pradesh
Maharastra
Meghalaya
Orissa
Punjab
Rajastan
Sikkim
Tamilnadu
Tripura
Uttar Pradesh
West Bengal
Delhi, Dadra &
Nagar Haveli
TOTALS
86-87
Area
10.60
2.10
120.00
1839.70
315.10
1782.00
377.10
222.40
306.50
3251.30
735.50
4.80
46.80
3189.00
1843.20
10.20
0.40
2.70
8311.50
397.70
41.20
22820.30
Prodn.
6.90
6.90
125.80
2863.10
661.70
5055.00
492.00
272.10
148.80
3865.00
536.40
6.70
83.30
9458.00
3401.60
16.50
0.20
4.30
16078.50
682.70
119.60
45576.50
87-88
Area
11.80
2.10
98.30
1934.90
192.40
1731.00
374.90
238.40
262.40
3549.10
732.70
4.80
45.70
3126.00
1533.80
11.20
0.10
3.20
8339.80
374.20
37.30
22604.20
Prodn.
4.20
6.90
105.80
2776.60
351.20
,4861.00
351.20
212.00
133.90
4328.80
633.40
6.70
80.90
11066.00
2909.80
17.70
0.10
2.70
16462.90
673.90
107.20
45095.50
                Area (x 1000 ha)
                Prodn. (million t)
                                              INDIA-9

-------
Appendix A   Soil physical properties used for CERES-Wheat simulations at Delhi and Hyderabad.
SOIL AT
DLAYR
10.
15.
15.
15.
30.
30.
30.
30.
30.
-1.
SOIL AT
DLAYR
10.
15.
15.
15.
15.
30.
30.
-1.
DLAYR
LL
DUL
SAT
SW
DB
DELHI
LL
.156
.166
.167
.157
.148
.110
.111
.112
.112
.00

DUL SAT SW
.262 .362 .262
.262 .362 .262
.262 .362 .262
.262 .362 .262
.261 .361 .261
.260 .360 .260
.259 .359 .259
.258 .358 .258
.258 .358 .258
.00 .00 .00

BD
1.37
1.37
1.37
1.37
1.38
1.38
1.39
1.39
1.39
.00
HYDERABAD
LL
.313
.313
.314
.316
.316
.318
.320
.00
DUL SAT SW
.480 .560 .480
.479 .559 .479
.479 .559 .479
.477 .557 .477
.477 .557 .477
.476 .556 .476
.474 .554 .474
.00 .00 .00
BD
1.35
1.36
1.36
1.37
1.37
1.37
1.38
.00
= Thickness of the soil layer
= Lower limit plant extractable
water
= Drained upper limit soil water content
= Saturated water content
= Defaulted soil water content
= Moist bulk density of soil
                                           INDIA-10

-------
                                                INDIA
Figure 1.      Map of India and location of the sites selected for the study.

-------
                                   o
                                   a
                                   u.
                                   o

                                   t

                                   CO
                                   CO
                                   5

                                   t
                                   IU
                                   5
                                   m
o   o   o   o   o-
o   o   o   o
X   CT   CM   t-
                                                                                                       DO

                                                                                                       (E
u>
CM
O    O   O    O
O    IO   O    IO
CM    T-   T-

-------
o
10
o
CM
O
CO
o
CM
O
en

\
111
CO

CO
                                                                        CO
                                                                        S

-------
                Grain Yield (t/ha)
                                                T*2
                                                        T4
               Biomass (t/ha)
                                                T*2
                   Avg.Rain 330 ppm
                   20% Rain 555 ppm
                                                        T*4
                            Avg.Rain 555 ppm   f&k-*i  20% Rain *330 ppm
                            -20% Rain *330 ppm d)  -20% Rain +655 ppm
Figure 3.
Sensitivity analysis of CERES-Wheat to temperature, precipitation, and CO2 at Delhi.

-------
    40
       (% change over ttae value)
                                  CLIMATE CHANGE ALONE
          G/SS     GFDL     UKMO     2010     2030      2050
S
    40
    20
       (% change over base value)
                              CLIMATE CHANGE + P.E. OF CO2
   -40



   -60



   -80








     40




     20
  G/SS      GFDL     UKMO     2010      2030     2050
* change over baa. va,u.)  CLIMATE CHANGE + P.E. OF CO2
 ce

 S,
 2
 M  -
    -40
    -eo


           6/SS     GFDL    UKMO     2010     2030      2050
                         Painted
                               Irrigated
  Figure 4a.      Effect of climate change on simulated yield and biomass at Delhi.

-------
          (% ch*ng over b vmlut)
% SEASON LENGTH
+v
20
0
-20
-40
-fin


Ti 1 1 "" u I

i i i i i i
             G/SS      GFDZ.     i/KMO     2070      2030      2050
     40
     20
    -20
    -40
        (% chango ovtr t>*t valut)  CLIMATED Cifl&NGE + P.E. OF CO
                                  i
   -60 '	-
           G/SS      GFDL     UKMO     2O10      2030     2050
                         RAINFED   ^^ IRRIGATED
Figure 4b.      Effect of climate change on season length and total crop evapotranspiration
              (ET) at Delhi.                                          F

-------


   40
   20

  -20
  -
  -60
  -80
 -100
     (% change over baa* value)
                                 CLIMATE CHANGE ALONE
        GISS     GFDL.    UKMO     2010     2030      2050
 40

 20

   0

 -20

 -40

 -60

 -80
    (% ching* over i> value)   CLIMATE CHANGE + P.E. OF CO2
1
       GISS     GFDL     UKMO     2010     2030      2050
  40
  20
     (K change over b value)
  -20
  -40
                                              ii
        GISS     GFDL    UKMO    2010      2030     2050
                     RAINFED
                   i IRRIGATED
Figure 5.      The effect of climate change on simulated yield and season length at
            Hyderabad.

-------

-------
IMPACT OF CLIMATE CHANGE ON THE PRODUCTION
         OF MODERN RICE IN BANGLADESH
       Z. Karim, M. Ahmed, S.G. Hussain, and Kh.B. Rashid

           Bangladesh Agricultural Research Council
                    Dhaka, Bangladesh
                    BANGLADESH-1

-------
TABLE OF CONTENTS

SUMMARY
INTRODUCTION
     Background
     Climate Vulnerabilities in Bangladesh
     Aims and Design of the Study
METHODS
     Climate Data and Climate Change Scenarios
     Crop Models, Cultivars and Management Variables
RESULTS
     Sensitivity of Rice Yields to Temperature Increases
     Rice Yields under GCM Climate Change Scenarios
IMPLICATIONS OF THE'RESULTS
REFERENCES
                                   BANGLADESH-2

-------
SUMMARY

      Bangladesh is located in a region that is vulnerable to environmental hazards, frequent floods, droughts,
cyclones, and storm surges that damage life, property, and agricultural production. This study uses climate
models combined with crop  simulation models to determine the possible effects of climate change on rice
production in major agricultural regions of the country.
      Sensitivity simulations showed that rice yields decreased significantly with temperature increases in the
two sites considered. The rice yields under the GCM climate scenarios alone decreased at both sites. When the
physiological CO2 effects were considered, the yield decreases under the climate change scenarios were offset.
If the physiological CO2 effects are not as positive as simulated in this study, rice production in Bangladesh
could be damaged under climate change conditions. A decrease in rice production, combined with the rapidly
increasing population, would threaten the country's food security.
INTRODUCTION

Background

      Bangladesh occupies an area of about 55,598 square miles near the Tropic of Cancer between latitudes
20.25N-26.38N and longitudes  88.01E-92.40E  (Figure  1). For its  relatively small size,  Bangladesh
encompasses a wide range of environmental conditions. Land, soil, hydrology, and climatic variability occur at
the national level and also in small areas of the country. The country is divided into 30 agroecological regions
(FAO 1988) according to physiography, characteristics and  length of seasonal flooding, length of the cool
winter period, and frequency of extremely high (>40C) summer temperatures.
      A variety of crops are grown in the country (Statistical Yearbook 1990), but local and modern varieties
of rice represent the dominant crop. Rice is the staple food in the country, currently grown on about 10.3
million hectares of land with production of about 18 million  tons. Other important crops are jute and wheat.
Wheat is becoming increasingly significant especially in the northern regions. This study considers the possible
impacts of climate change on modern rice production in two contrasting locations in Bangladesh (Figure 1)
that represent major rice-growing regions.

Climate Vulnerabilities in Bangladesh

      Monsoon rainfall occurs everywhere in the country, although the amount and intensity of the rain varies
greatly. About 90% of the precipitation occurs during four months (June-September), but the distribution
within these months differs at different sites (Figure 2)., Annual rainfall ranges from 1,400 mm in the  dry
Rajshahi (Northwest) region to more than 5,000 mm in the wet Sylhet  (Northeast)  region. Regional cropping
patterns are primarily determined by the seasonal monsoon flooding regime. There are three  seasons with
different hydrological regimes: the wet season from June to November  (called "kharif"), the dry season (called
"rabi"), and a transitional season (called "prekharif"). The winter temperatures are moderate, but the summer
temperatures are extremely .high and may exceed 40C.
      The country is highly vulnerable to climatic events; floods and droughts are very frequent and damaging.
The recent severe  floods  of 1987 and 1988 produced colossal damages to crops and property. In 1988, about
2.5 million tons of food grain were damaged by floods. Due to the uncertain and uneven distribution of rainfall,
droughts of different intensities may occur and cause large crop losses! During the  wet season, the crop most
affected by drought is rice. In the areas  that are most severely affected  (about 2.3 million ha), yield losses can


                                          BANGLADESH-3

-------
represent a 45% decrease from the achievable yield (Karim et al. 1990; Karim and Akhand 1982). During the
dry season, wheat, potato, and mustard are most affected by drought. Possible climate changes (IPCC 1990)
may bring more unfavorable conditions for agriculture in Bangladesh, as has been previously suggested (Mahtab
1990).

Aims and Design of the Study

      Since climatic variability is the most important factor determining agricultural production in Bangladesh,
it is important to study the potential impacts of climate changes on rice production. This study uses climate
change scenarios generated by General Circulation Models (GCMs) and is part of a global research project
funded by the U.S. Environmental Protection Agency.
      Two contrasting locations, Mymensingh (Lat. 24.46N and Long. 90.23E) and Barisal (Lat. 22.42N and
Long. 90.23E), were selected for the study (Table 1). Mymensingh represents the old Brahmaputra, young
Brahmaputra, and Jamuna floodplain regions. This location includes a relatively large proportion of high and
medium  highlands ideally suitable for transplanted modern rice cultivars. Barisal represents 2.85 million
hectares of the coastal areas of Bangladesh and constitutes about 30% of the net cultivable area of the country.
The Ganges tidal floodplain is the major agroecological region represented by Barisal, and part of this region
is seasonally inundated by tides.

METHODS

Climate Data and Climate Change Scenarios

      Daily maximum and minimum temperatures, precipitation, and hours of sunshine data were obtained
for the two sites. Daily solar radiation values were calculated from sunshine hours using the weather-generating
program  included in DSSAT (Jones et al. 1990). Figure 2 shows the monthly average temperatures and
precipitation for Mymensingh and Barisal.
      The climate change scenarios used in this simulation study were created from three  GCMs: Goddard
Institute  for Space Studies (GISS)  (Hansen et al. 1983), Geophysical Fluid Dynamics Laboratory (GFDL)
(Manabe and Wetherald 1987), and United Kingdom Meteorological Office (UKMO) (Wilson and Mitchell
1987). The UKMO scenario projects the largest annual temperature increases (4.3C) (Table 2), while GISS
and GFDL predict smaller  annual temperature changes. Annual precipitation is  projected to increase
considerably at the two sites (up to 41%) except at Barisal, where annual precipitation is projected to decrease
slightly under the GISS scenario. The study also includes a sensitivity study to changes in temperature (+2C
and +4C).

Crop Model, Cultivars and Management Variables

      The simulation of rice growth and yield was performed using the CERES-Rice model (Godwin et al.
1992). Since the climate change scenarios have higher levels of CO2 than the current climate GCM simulations,
the CERES-Rice model includes an option to simulate the physiological effects of CO2 on photosynthesis and
water-use efficiency that results in an increase in yield, as documented experimentally (Acock and Allen 1985).
The physiological effects of CO2 were also considered at both sites.
      The Bangladesh Rice Research Institute has developed a number of crop varieties; BR11 and BR3 are
the two most extensively used. The cultivar BR11 is grown during the wet season (June-November) and is
transplanted (called "transplanted Aman"). The cultivar BR3 is the main cultivar grown during the dry season
(called "Boro" rice) and is generally irrigated. Occasionally the cultivar BR3 is grown during the wet season

                                        BANGLADESH-4

-------
as transplanted Aman.
      The management variables for the simulation model were determined according to current practices in
the two regions considered in this study. Rice growth was simulated under the "automatic irrigation" option
of the model to provide the crop with a nonlimiting water situation. A limitation of this choice of simulation
is that the yield losses due to insufficient water for the crop are not simulated, and therefore the yield could
be overestimated. Nevertheless, relative yield changes under climate change conditions compared to baseline
climate could still be analyzed.

RESULTS

Sensitivity of the Rice Yields to Temperature Increases

      Table 3 shows that simulated rice yields decreased significantly with increases in temperature at the two
sites considered. During the transplanted Aman season, rice yields were reduced about 21 ha"1 by a temperature
increase of 4C at both locations. Similar effects were also observed with the cultivar BR3  in the Boro rice
season. In all cases, the high temperatures result in a reduced season length that partially accounts for  the
reduced yields. The crop may also experience heat stress due to the very high temperatures considered in these
sensitivity scenarios, especially during the transplanted Aman season (warmest season). During the Boro season,
base temperatures are lower and rice yields were somewhat less reduced under the high-temperature scenario.
When the direct effects of 555 ppm CO2 were included in the yield simulations, they compensated for  the
negative effects of a 2C temperature increase but not for the effects of a 4C temperature increase.

Rice Yields Under GCM Climate Change Scenarios

      Rice yields decreased significantly at both sites under the three climate change scenarios compared with
the base yields under current climate (Table 4). During the Boro season, the UKMO scenario projected  the
largest decreases in yields at  the two sites because of the higher temperature projections. In  the transplanted
Aman season, simulated yield losses under the GCM scenarios were larger than in the Boro season. Simulated
season length decreased under the three GCM scenarios considered due to the acceleration of phonological
growth stages by high accumulated temperature.  When the direct effects of CO2 (555 ppm) were included in
the yield simulation, they compensated for the yield decreases under the climate change scenarios in most cases.

IMPLICATIONS OF THE RESULTS

      The current rice production could be damaged under climate change conditions as projected by  the
GCMs. Although the simulated direct effects of CO2 may compensate for yield reductions in general, the  full
extent of these effects has not been clearly established under  field conditions. Another cause for concern in
regard to climate change is  the potential for increases in sea-level leading to flooding of low-lying coastal
regions. These regions support high populations and extensive agricultural production. Any possible decrease
in rice production in Bangladesh would have severe consequences on the food  resources available to  the
country and thus threaten food security. The Bangladesh Planning Commission estimates that the country's
population will increase to 137 million by the year 2000 and to 155 million by the year 2010, leading to further
reduction of the land-person ratio from the low  current level  of 0.13 ha per person.
      In view of the results of this study, serious consideration should be given to reduce population growth,
to develop  research strategies to increase rice production,  and to limit greenhouse gas  emissions.  The
development of rice cultivars that are more tolerant to heat and require a shorter growing season is one of the
possible strategies for adaptation to climate change. Policymakers should seriously consider population growth

                                         BANGLADESH-5

-------
regulations as well as qualitative improvements in agronomic research and technology to sustain the country's
economic development.
                                       BANGLADESH-6

-------
REFERENCES

Acock, B., and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
      and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation. DOE/ER-0238. U.S.
      Department of Energy. Washington, D.C. pp. 53-97.

FAO. 1988. Land Resources and Appraisal of Bangladesh. Technical Report 3, Vol. 1. (BGD/81/035).

Godwin, D., U. Singh, J.T. Ritchie, and E.C. Alocilja. 1992. A User's Guide to Ceres-Rice. International
      Fertilizer Development Center. Muscle Shoals, AL.

Hansen, J., G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983. Efficient Three-
      Dimensional Global Models for Climate Studies: Models I and II. Monthly Weather Review, Vol III, No.
      4:609-662.

IPCC. 1990. Climate Change: The IPCC Scientific Assessment. J.T. Houghton, G.J. Jenkins, and J.J. Ephraums
      (eds). Intergovernmental Panel on Climate Change. Cambridge.

Jones, J.W., S.S. Jagtap, G. Hoogenboom, and G.Y. Tsuji. 1990. The structure and function of DSSAT. pp 1-
      14. In: Proceedings of IBSNAT Symposium: Decision Support  System for Agrotechnology Transfer, Las
      Vegas, NV. 16-18 Oct. 1989. Part I: Symposium Proceedings. Dept. of Agronomy and Soil Science,
      College of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii 96822.

Karim,  Z., and N.A. Akhand. 1982. Net irrigation Requirement of Rice and Evapotranspiration of Wheat and
      Potato for Different Locations of Bangladesh.  BARC Publication No.ll. Dhaka.

Karim,  Z., A.M. Ibrahim, A. Iqbal, and M. Ahmed. 1990.  Drought in Bangladesh Agriculture and irrigation
      Schedules for major crops. BARC publication N34.

Mahtab, F.U.  1990. Implication of Global Change for Bangladesh. Paper presented at the Workshop on the
      Environmental Implication of Global  Change,  18th IUCN General Assembly, Perth,  Australia,
      November 1990.

Manabe, S., and R. Wetherald. 1987. Large-scale changes of soil wetness induced by an increase in atmospheric
      carbon dioxide. Journal of Atmospheric Science 44:1601-1613.

Statistical Yearbook 1990. 1990. Bureau  of Statistics. Government of Bangladesh.

Wilson, C.A.,  and J.F.B. Mitchell. 1987. A doubled CO2 climate sensitivity experiment with a global model
      including a simple ocean. Journal of Geophysical Research, 92: 13315-13343.
                                         BANGLADESH-7

-------
Table 1.  Characteristics of the land resources and inundation land type in Bangladesh.
                                                   Land Type (ha)

Bangladesh Total
Mymensingh Region
Barisal Region
Highland
4,199,952
156,022
12,117
Medium
Highland
5,039,724
138,303
135,580
Medium
Lowland
1,771,102
63,243
23,593
Lowland
1,101,560
7,821
4,566
Very Lowland
193,243
-
-
 Highland: Land above normal inundation level.
 Medium Highland: Land normally inundated up to about 90 cm deep.
 Medium Lowland: Land normally inundated up to 90-180 cm deep.
 Lowland: Land normally inundated up to 180-300 cm deep.
 Very Lowland: Land normally inundated deeper than 300 cm.
                                       BANGLADESH-8

-------
Table 2. Annual temperature and precipitation changes from base under GCM scenarios.
 Site
             Temp, Changes (C)
Lat/Long   GISS    GFDL   UKMO
                                 Precip. Changes (%)
                               GISS   GFDL   UKMO
 Mymensingh
 Barisal
24.75 N
90.38 E
22.68 N
90.33 E
2.8
4.0
2.8
2.8
4.3
4.3
27
-7
33
33
                                                                               41
                                                                               41
                                       BANGLADESH-9

-------
Table 3. Sensitivity of CERES-Rice to temperature and CO2 changes at selected sites.
CO2 level Temp.
(ppm) (C)


330 0
2
4
555 0
2
4


330 0
2
4
555 0
2
4
Yield Season L.
(t ha'1) (days)
Rice Cultivar BR11, transplanted Aman
Mymensingh
6.28 139
5.14 128
4.39 127
7.69 140
6.51 131
5.74 129
Rice cultivar BR3, Boro Rice
Mymensingh
6.39 156
5.77 150
4.94 147
7.61 156
6.98 150
6.09 148
Yield Season L.
(t ha-1)

Barisal
7.74
6.59
5.71
9.43
8.17
7.29

Barisal
6.81
6.31
5.65
8.13
7.61
6.91
(days)


152
140
138
153
141
139


138
131
129
138
131
129
                                      BANGLADESH-10

-------
Table 4.  Simulated rice yield (a) and season length (b) under current climate and yield changes under climate
         change scenarios in Bangladesh.
 (a) Yield (t ha'1)
 Site
Type of rice      BASE    GISS
        % change from base

GISS    GFDL    GFDL   UKMO   UKMO

Mymensingh


Barisal



BR 11 Aman
BR 3 Aman
BR 3 Boro
BR 11 Aman
BR 3 Aman
BR 3 Boro
tha'1 330 555 330 555
6.28 -22 0 -27 -6
4.61 -23 0 -27 -6
6.37 -14 5 -18 1
7.74 -19 2 -24 -3
5.7 -18 4 -24 -3
6.81 -11 8 -14 5
330 555
-22 0
-21 1
-25 -6
-20 1
-19 2
-23 -4
(b) Season Length (days)
Site
Mymensingh


Barisal


Type of Rice
BR 11 Aman
BR 3 Aman
BR 3 Boro
BR 11 Aman
BR 3 Aman
BR 3 Boro
BASE GISS GFDL
139 130 127
138 130 127
156 147 148
152 138 139
152 138 138
138 131 129
UKMO
127
127
145
137
138
125
330 - Climate change alone
555 - Climate change with physiological CO2 effects.
                                        BANGLADESH-11

-------

-------
                                 BANGLADESH
Figure 1. Map of Bangladesh and location of the sites used in the simulation study.

-------
                            BARISAL
     TEMPERATURE (C)
               PRECIP (mm)
                                                           500
  10
     Jan  Feb Mar  Apr May  Jun  Jul  Aug Sep  Oct Nov  Dec
                            MONTH
                     Temp (C)
 I Precip (mm)
                        MYMENSINGH
    TEMPERATURE (C)
              PRECIP (mm)
                                                           600
     Jan  Feb  Mar Apr  May Jun  Jul  Aug Sep  Oct  Nov Dec
                           MONTH
                    Temp (C)
I Precip (mm)
Figure 2. Observed monthly averaged temperature and precipitation for selected sites.

-------
IMPACT OF CLIMATE CHANGE ON SIMULATED RICE
            PRODUCTION IN THAILAND
                      C. Tongyai

           Soil Science Division, Dept. of Agriculture
        Ministry of Agriculture and Cooperatives, Thailand
                     TrtAILAND-1

-------
             TABLE OF CONTENTS

             SUMMARY
             INTRODUCTION
             METHODS
                   Observed Climate
                   Climate Change Scenarios
                   Crop Model and Management Variables
                   Validation of the Crop Model
             RESULTS AND DISCUSSION
                   Sensitivity Analysis
                   Effect of GCM Climate Change Scenarios on Rice Growth
             REFERENCES
                                                THAILAND-2
_

-------
SUMMARY

        This study used global climate models and crop dynamic models to estimate the potential effects of
climate change on the rice production of Thailand. Projected climate change caused simulated rice yields (both
upland and paddy rice) to decrease dramatically, due to temperature increases. Yield decreases were partially
counteracted by the physiological effects of CO2, as simulated in this study. The  high temperatures of the
climate change scenarios had a negative effect on crop growth even when an adequate water supply was
available. We found that in spite of the differences among the GCM scenarios, the locations  simulated, and
the agricultural practices of the regions, rice yields under climate change scenarios dramatically decreased in
comparison with base-line yields. The results may be indicative of the potential impacts of climate change on
the agricultural systems in Thailand.
INTRODUCTION

        Thailand, located in Southeast Asia, extends over a large latitudinal range (latitude 540'N to 2030'N)
and from longitudes 9720'E to 10545'E (Figure 1). The total area of the country comprises 513,115 km2. Six
regions are defined and characterized by their landforms: the Northern Ranges and Valleys, the Northeast
Plateau or Khorat Plain, the Central Plain, the Western mountains, the Southeast coast, and the Peninsular
South. Agriculture is a main contributor to  the Thailand economy. Agricultural systems (arable crops,
rangelands, forestry, and fisheries) employ more than 60% of the labor force during the cropping season and
account for 20% of GNP. Crop production accounts for more than 50% of the agricultural revenues. Rice is
the main food crop and the main agricultural export commodity. It is grown in all regions of the country
during both the wet and dry seasons. In. some  regions additional water  for irrigation is provided when
necessary. Although both long-grain (Thai) and glutinous rice are grown throughout the country, the long-
grain rice is the main crop in most regions and accounts for all rice exports (Table 1).
        Floods are an important cause of crop production losses. Natural  disasters (tropical or depression
storms and landslides) also account for the destruction of crops and damage to agricultural infrastructure. The
greenhouse effect could lead to higher global surface temperatures and changed hydrological cycles (IPCC
1990). Previous research has suggested that the possible impacts of climate change in Southeast Asia could
be severe (Parry et al 1988). In Thailand a change in the rainfall pattern or an increase in temperature could
lead to soil moisture deficits for the rainfed rice crop,  and result in limited irrigation water supply for the
second crop. This study analyzed the simulation of rice production with scenarios of climate change, generated
by General Circulation Models (GCMs), and compared the results to simulations under the current climate.
The study also evaluated the physiological effects of CO2 on rice yields.
        To represent the varied range of agricultural practices and regions  of the country, simulations were
carried out at four contrasting locations and included a combination of different management practices and
soils. The regions selected for the simulation study are the upper and lower Northern regions and the Central
Plain. The lower Northern region encompasses broad differences in climatic conditions, therefore two sites
were selected in this region (Phitsanulok and Nankhon Sawan). The upper Northern region is represented by
Chiang Mai, and the Central Plain is represented by Bangkok (Figure 1). These sites are representative of the
major rice production areas in Thailand.

METHODS

Observed Climate
                                           THAILAND-3

-------
        Climate data for the four sites included in this study were obtained from the Thailand Meteorological
Observatory (1956-1989), and included daily maximum and minimum temperatures and daily precipitation.
Daily solar radiation was estimated from sunshine hours by the WGEN program (Richardson and Wright
1984).
    Thailand experiences extreme seasonal climate fluctuations, although its climate is generally humid. Figure
2 shows the monthly temperature and precipitation averages for the 1956-89 period for the site of Bangkok.
Two seasons can be distinguished: wet (May to October) and dry. The mean temperature is always over 25C
except in the coldest regions (represented by Chiang Mai), and the maximum temperature often exceeds 30C.

GCM Climate Change Scenarios

        The General Circulation Models compute climatic variables for spatial gridboxes across the earth's
surface, but they do not account for the variations within the gridbox. Climate change scenarios for each site
were created based on the GCM gridbox output of climate variables applied to the daily observed climate at
the study sites. The method employs the differences between lxCO2 and 2xCO2 monthly temperatures, and the
ratios between 2xCO2 and lxCO2 monthly precipitation and solar radiation amounts. (lxCO2 refers to current
climate conditions, and 2xCO2 refers to the climate which would occur with an equivalent doubling of CO2.)
Three  equilibrium GCMs were  used in this study to create the climate change scenarios for each site: the
Goddard Institute for Space Studies model (GISS)  (Hansen et al. 1983), the Geophysical Fluid Dynamics
Laboratory model (GFDL) (Manabe and Wetherald 1987), and the United Kingdom Meteorological Office
model (UKMO)  (Wilson and Mitchell 1987). Scenarios were also generated with output from  a  transient
climate change scenario (GISS transient run A, Hansen et al. 1988).
    The climate change scenarios show some major differences when compared with the observed climate.
Figure 3 shows monthly differences  between observed climate and climate change scenarios for each site. The
temperature increases are very similar at the four sites and in general larger temperature increases are
simulated by the UKMO climate change scenario. With all climate change scenarios, the largest changes  in
the seasonal temperatures occurred  during the dry season. The scenario precipitation changes vary among the
sites and the distribution of the projected seasonal rainfall was very different from site to site. The largest
increases in projected precipitation usually correspond to the early wet season.

Crop  Model and Management Variables

        Potential changes in rice  physiological responses (yield, season length, evapotranspiration,  and
irrigation water demand) were estimated with the CERES-Rice model (Godwin et al 1992) under  different
climate scenarios. The software for the climate change simulations of the rice crop was developed by  IBSNAT
(IBSNAT1989, Jones et al 1990). The model was run using 30 years of baseline climate and GISS, GFDL, and
UKMO climate change scenarios under dryland and irrigated conditions.
       Cultivars. The cultivar RD23 was selected for the study because it represents the most common long-
grain cultivar in Thailand (Table 2). The genetic coefficients of RD23 were taken from  the DSSAT  database
(Jones et al 1990). In the upper North region (Chiang Mai), the glutinous cultivar NSPT was also used in this
study (Table 2). These cultivars have been calibrated for many different climate conditions, including Thailand.
       Management Variables. Rice was planted in  early to mid- August, and the plant population was 25
plants m"2. Both upland and paddy rice practices were simulated. For the irrigated simulations, the automatic
option of the CERES-Rice model was used to provide the crop with a nonlimiting water situation. The water
demand was calculated assuming:  100%  efficiency of the automatic irrigation system;  1 m irrigation
management depth; and automatic  irrigation when the available soil water is 50% or  less of capacity. The
initial soil water content for each layer was determined  for each type of soil (see below).
                                          THAILAND-4

-------
       Soils and Initial Soil Water Conditions. The input soil parameters, nutrient levels, and initial soil
conditions (determined for each layer of soil and for all soils employed in this study) were obtained from a
soil data survey from the Soil Physics Research Group at the Department of Agriculture (Table 3).
       Physiological Effects ofCO2. Many crops may benefit from the higher levels of atmospheric CO2 (Acock
and Allen 1985). Because the climate change scenarios have higher amounts of CO2 than the present climate,
the CERES-Rice model includes an option to simulate the physiological effects of CO2 on  the rice (i.e., an
increase in photosynthesis and water use efficiency). The physiological effects of CO2 were considered in all
scenarios and at all sites.

Validation  of the CERES-Rice Model

       The CERES-rice model was validated at Bangkok using observed crop and weather datasets for the
years 1985,1986 and 1987.  The rice cultivar used was RD23 (long-grain Thai). The initial soil water conditions
were set to realistic values for the years of the simulation. Observed and simulated yield, biomass, and season
length were compared. Table 3 shows the close correspondence between the  simulated and the observed
parameters during the years of the validation experiments.

RESULTS

Sensitivity Analysis

        This section characterizes the sensitivity of the CERES-Rice model to changes in temperature at the
four sites selected for the study in Thailand. The direct beneficial CO2 effects on rice growth were simulated
for each set of conditions. Table 4 shows the sensitivity results under irrigated conditions. The  temperature
increases resulted in decreases in crop yield and season length at all sites. In the rainfed simulations  (not
shown) the  relative changes in yields  and season length produced by the  temperature increases were
comparable to the irrigated simulation.

Effect of GCM Climate Change Scenarios on Rice Growth

        This section analyzes changes in simulated crop yield for the GCM climate change scenarios, using
the current  climate and management  variables as  the base reference. Each climate scenario  included  crop
simulations  with 330 ppm CO2 (representing the effects of climate change alone  on the crop) and with 555
ppm of CO2 (representing the effects of climate change with the direct beneficial effects of CO2 on yield and
crop water use).
        Simulations were conducted to represent different crop management conditions, soils, and initial water
conditions of the soils. There were large decreases in simulated yields under the climate change scenarios alone
compared to the baseline yield (Table 5).  These were comparable at all sites and with the different  crop
conditions, probably because the negative effect of increased temperature on season length  (as shown in the
sensitivity analysis) is the dominant factor driving yield changes in each case. The performance of the rice
cultivar RD23 was compared with the cultivar NSPT, which is a typical photoperiod-sensitive, glutinous indica
cultivar planted in the upper northern region.  Generally, the yield of the RD23 cultivar decreased more than
that of the NSPT under the conditions that simulate rice growth in paddies (Table 5).
        When the direct CO2 effects were included in the climate change simulations, rice yields increased at
all sites in comparison to the climate change scenarios alone. The direct CO2 effects offset the yield decreases
simulated under the climate change scenarios in some but not all cases (Table 6  and Figure 4).    The reaite
of these simulations imply the potential for significant changes for  Thailand rice production  with climate

                                           THAILAND-5

-------
change. Yield decreases for two of the GCM scenarios tested ranged from -20% to -40%; further studies are
planned to test possible adaptation strategies to yield decreases of such magnitude. The third GCM scenario
implied little change. Better GCM simulations of both current climate and of regional predictions of climate
change are needed to narrow the uncertainties of crucial importance to rice production in Thailand.
                                          THAILAND-6

-------
REFERENCES

Acock, B., and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
       and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation. DOE/ER-0238. U.S.
       Department of Energy. Washington, D.C. pp. 53-97.

Godwin,  D., U. Singh, J.T. Ritchie, and E.C. Alocija. 1992. A  User's  Guide to CERES-Rice. International
       Fertilizer Development Center. Muscle Shoals, AL.

Hansen, J., I. Fung, A. Lascis, D. Rind, G. Russell, S. Lebedeff, R. Ruedy, and P. Stone. 1988. Global climate
       changes as forecasted by the GISS 3-D model. Journal of Geophysical Research 93:9341-9364.

Hansen, J., G. Russell, D. Rind, P. Stone, A Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983. Efficient
       Three-Dimensional Global Models for  Climate Studies:  Models I  and II. April Monthly Weather
       Review, Vol III, No. 4: 609-662.

IPCC. 1990. Intergovernmental Panel on Climate Change. 1990. First  Assessment Report. Working Group
       II. Tegart, W.L. McG., G.W. Sheldon, and D.C. Griffiths (eds.).

IBSNAT. 1989. International Benchmark Sites Network for Agrotechnology Transfer  Project.   Decision
       Support System for Agrotechnology Transfer Version 2.1 (DSSAT V2.1). Dept. Agronomy and Soil
       Sci., College of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii 96822.

Jones, J.W., S.S. Jagtap, G. Hoogenboom, and G.Y. Tsuji. 1990. The structure and function of DSSAT. pp 1-
       14. In Proceedings of IBSNAT Symposium: Decision Support  System for Agrotechnology Transfer,
       Las Vegas, NV. 16-18 Oct. 1989. Part I: Symposium Proceedings. Dept. of Agronomy and Soil Science,
       College of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii 96822.

Manabe,  S., and  R. Wetherald. 1987. Large-scale changes of soil  wetness induced  by  an increase  in
       atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.

Parry, M., T.R. Carter, and N.T. Konijn (eds.) 1988. The Impact of Climatic Variations on Agriculture.
       Kluwer, Dordecht, Netherlands. 876 pp. and 764 pp.

Richardson, C.W., and D.A Wright. 1984. WGEN: A Model for Generating Daify Weather Variables. ARS-8.
       U.S. Department of Agriculture, Agricultural Research  Service. Washington, DC. 83 pp.

Wilson, C.A, and J.F.B. Mitchell. 1987. A doubled CO2 Climate  Sensitivity Experiment with a Global Model
       Including a Simple Ocean. Journal of Geophysical Research, 92: 13315-13343.
                                          THAILAND-7

-------
Table 1.
Regional rice production (1984-87).
Region
THAI RICE
North Upper

North Lower

Northeast Upper

Northeast Lower

Central East

Central Mid

Central West

South Upper

South Lower

GLUTINOUS RICE
North
Northeast Upper
Northeast Lower
Others
Season

Wet
Off
Wet
Off
Wet
Off
Wet
Off
Wet
Off
Wet
Off
Wet
Off
Wet
Off
Wet
Off





Area (Ha)

55,200
12,544
1,484,224
65,456
304,704
16,800
611,408
13,872
510,144
62,528
1,184,256
447,824
224,480
15,120
475,040
27,152
111,248
3,840

576,656
196,752
910,976
24,944
Yield (T/Ha)

2.96
3.11
2.14
3.69
1.56
3.02
1.61
2.16
1.84
3.25
2.29
3.95
2.22
3.43
1.55
2.53
1.61
3.37

2.96
1.54
1.39
1.85
                Representative Sites:
                Chiang Mai (N. Upper); Phitsanulok and Nankhon Sawan
                (N. Lower); Bangkok (C. East).
                                         THAILAND-8

-------
Table 2.
Sites, cultiVars, and soil types used in the simulations.
                   Site
                        Lat/Long
   Rice
  Cultivar
Soil Type
                   Chiang Mai
                   Phitsanulpk
                   Nankhon Sawan
                   Bangkok
                        +18.5 N
                        +98.6 E
                        +16.5 N
                        +100.2 E
                        +15.5 N
                        +100.1 E
                        +13.4 N
                        +100.3 E
  NSPT
(glutenous)
  RD23
(long grain)
  RD23
  RD23
  RD23
  RD23
  RD23
  RD23
  RD23
  RD23
 Hd Clay

 Hd Clay

Wangtong
 Phimai
 Hd Clay
 Rb Clay
 Np Clay
 Suphan
 Rangsit
 Bangkok
                                         THAILAND-9

-------
Table 3.
Validation of the CERES-Rice model for Bangkok.
      Yr.
                    Yield (t ha'1)       Biomass (t ha'1)

                     Obs.     Sim.       Obs.      Sim.
Season Length (d)

   Obs.      Sim.
1985
1986
1987
4.23
4.53
4.45
4.40
4.85
4.77
8.93
9.42
8.57
7.89
8.58
8.60
93
90
91
105
106
99
      Locations: Klong Luang, Rangsit, Bangkok
      Crop Management Conditions:
        Soil type; Rangsit Lowland fine, mixed, acid sulfic Tropaquepts
        Planting date: August 15
        Start of simulation: July 15
        Soil water initial condition: 0.456, 0.456, 0.447, 0.414, 0.414, 0.386
        Plant population: 25 plants/m2
                                        THAILAND-10

-------
Table 4.        Sensitivity analysis of CERES-Rice to climate and CO2 changes. Simulations are for irrigated
               production of long-grain Thai rice (BR23).
CHANGES
CO2 level Temp.
(ppm)

330


555


(oQ

0
2
4
0
2
4
Yield
(tha1)
Chiang
(soil: Hd
5.09
3.83
2.74
6.58
5.31
4.29
SIMULATED VARIABLES
Season L. ET Yield Season L.
(days)
Mai
clay)
115
105
102
115
105
102
(mm)

587
574
574
577
569
580
Phitsanulok
(soil: Hd clay)
330


555


0
2
4
0
2
4
4.39
3.32
2.44
5.85
4.84
3.95
104
100
103
104
100
103
591
594
620
582
591
623
(tha-1)
Nankhon
(soil: Rb
4.71
3.62
2.64
6.16
5.07
4.11
(days)
Sawan
clay)
105
101
103
106
101
103
ET
(mm)

572
574
587
561
569
595
Bangkok
(soil: Np clay)
4.63
3.36
2.44
6.14
4.82
3.91
103
97
99
103
98
99
635
625
642
624
621
651
                                         THAILAND-11

-------
Table 5.        Effects of climate change on rice yield under GCM climate change scenarios at four sites with
               different cultivars and soils. Initial soil water was set to characteristic observed values. (DE
               = direct effects of CO2 on rice yield included.)
                                                     % CHANGE IN SIMULATED YIELD
                                                GCM Scenario Alone
GCM Scenario +DE of
         CO,

Site:




Site:






Site:


Site:








Cult.
Soil
Rainfed/
Irrig.
Chiang Mai
RD23
RD23
NSPT
NSPT
Hd clay
Hd clay
Hd clay
Hd clay
Rainfed
Irrig
Rainfed
Irrig
Base
Yield
tha'1
2.28
5.09
2.24
4.93
GISS

-36
-41
-34
-29
GFDL

-11
-25
-11
-12
UKMO

-25
-39
-25
-28
GISS

-6
-10
-7
-2
GFDL

25
6
18
12
UKMO

10
-10
4
-4
Phitsanulok
RD23
RD23
RD23
RD23
RD23
RD23
Nankhon
RD23
RD23
Bangkok
RD23
RD23
RD23
RD23
RD23
RD23
RD23
RD23
Wangtong
Wangtong
Phimai
Phimai
Hd clay
Hd clay
Sawan
Rb clay
Rb clay

Np clay
Np clay
Suphan
Suphan
Rangsit
Rangsit
Bangkok
Bangkok
Rainfed
Irrig
Rainfed
Irrig
Rainfed
Irrig

Rainfed
Irrig

Rainfed
Irrig
Rainfed
Irrig
Rainfed
Irrig
Rainfed
Irrig
1.98
4.38
2.61
4.55
2.28
4.39

3.48
4.71

2.58
4.63
2.08
4.68
1.83
4.93
2.82
4.76
-35
-41
-39
-35
-38
-40

-40
-40

-42
-44
-42
-43
-44
-37
-44
-40
-15
-26
-16
-24
-16
-27

-20
-26

-16
-31
-16
-33
-17
-29
-17
-31
-30
-36
-31
-31
-31
-35

-36
-38

-47
-48
-48
-47
-48
-43
-48
-46
-6
-6
-11
-3
-11
-4

-14
-8

-14
-12
-12
-7
-10
-11
-16
-9
23
7
18
5
22
7

11
4

20
0
23
-1
25
-2
18
-1
1
-4
-2
-1
0
-1

-9
-7

-23
-18
-21
-15
-19
-18
-23
-17
                                         THAILAND-12

-------
Table 6.        Effects of climate change on rice yields under GCM climate change scenarios in four selected
               sites. The water soil initial conditions were set to characteristic observed values. The results
               presented are an average of simulations with different soils under irrigated conditions using
               cultivar RD23 (long gram  Thai rice).
                                  Percent Changes in Simulated Yields from Base
                           GCM Climate Scenario Alone
 GCM Climate Scenario
Including the Physiological
       CO2 Effects
Site
Chiang Mai
Phitsanulok
Nankhon Sawan
Bangkok
GISS
-41
-39
-40
-41
GFDL
-25
-26
-26
-31
UKMO
-39
-34
-38
-46
GISS
-10
-4
-8
-10
GFDL
6
6
4
-1
LKMO
-10
-2
-7
-17
                                          THAILAND-13

-------

-------
                                   THAILAND
Figure 1.      Map of Thailand showing the sites selected for the study.

-------
                   BASELINE CLIMATE
                   BANGKOK, THAILAND
      TEMPERATURE (C)
PRECIPITATION (MM/MONTH)
                                                        400
      JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC

                           MONTH
             TEMPERATURE (C)
PRECIP (MM/MONTH)
Figure 2.     Observed mean monthly temperature and precipitation in Bangkok.

-------
CLIMATE IMPACT ASSESSMENT FOR AGRICULTURE IN THE PHILIPPINES:
   SIMULATION OF RICE YIELD UNDER CLIMATE CHANGE SCENARIOS
                     Crisanto R. Escafto and Leandro V. Buendia

       Philippine Council for Agriculture, Forestry and Natural Resources Res. and Dev.
                     (PCARRD), Los Banos, Laguna, Philippines
                               PHILIPPINES-1

-------
TABLE OF CONTENTS

SUMMARY
INTRODUCTION
       Rice Production in the Philippines
       Climate Restrictions to Crop Production in the Philippines
METHODS
       Baseline Climate Data
       Climate Change Scenarios
       Crop Model and Management Variables
       Validation of the Crop Model
       Limitations of the Study
RESULTS AND DISCUSSION
       Sensitivity of Simulated Rice Yields to Temperature Increases and CO2
       Rice Yields Under GCM Equilibrium and Transient Climate Change Scenarios
       Adaptation Strategies to Climate Change
REFERENCES
                                     PHILIPPINES-2

-------
SUMMARY

        This  study used General Circulation Models (GCMs) and a crop growth model to estimate the
potential impact  of  climate change on rice yields in  two main  agricultural regions of the Philippines,
represented by the sites of Batac and Los Banos. Projected climate change caused simulated rice yields to
decrease at both  locations, but the decreases were larger at Los  Banos, which has a lower latitude. The
decrease in simulated rice yield at these tropical sites was caused primarily by the GCM-projected temperature
increases which shorten the crop-growing cycle. When the direct CO2 effects on crop growth were included
in the simulation, the yield losses under the climate change scenarios were completely offset in Batac, but only
partially offset in Los Banos. If climate change brings  conditions similar to  those projected by the GCM
scenarios, rice production in Los Banos, an important agricultural region, could be significantly reduced.
Testing of possible adaptation strategies to climate change suggested dramatic changes in the cropping patterns
of rice and in the agricultural systems of the Philippines.
INTRODUCTION

Rice Production in the Philippines

        The Philippines is an archipelago of about 7,000 islands scattered between latitudes 4 30' and 21 30'
north and longitudes 116 and 127 east (Figure 1). The islands are located in the humid tropical belt with
twelve of the 7,000 islands having individual areas of more than 1,000 km2. About 50% of the country's 29
million hectares are cultivated lands. Agriculture is considered the economic lifeline of the Philippines. More
than 50% of the working population is engaged in agriculture and more than 70% of the foreign exchange
earnings are derived from exports of agricultural products. Rice is the staple food of the Filipino people (in
1980 the consumption of rice was 93 kg per capita), and the current population (more than 61 million people)
depends on the rice supply from the 3.4 million ha of cultivated rice each year.
        The major areas of rice production (Table 1) are in the regions of Ilocos (represented in this study
by  the site of Batac;  1803'N  and 12032'E),  Cagayan  Valley, Central Luzon, and Southern Tagalog
(represented in this study by the site of Los Banos; 1411'N and 12115'E). Rice management practices vary:
24% of the total area of rice is irrigated during the wet season; 19% is irrigated during the dry season; 38%
is rainfed; and 19% is upland rice. The first three practices are, included in the "wetland" system and the fourth
practice takes place on dry land. Wetland rice production in the Philippines takes place mainly in low-elevation
areas (50 m or lower), in a few areas of low-terraced paddies on intermediate upland slopes, and in isolated
spots on terraces on mountain slopes. Wetland areas are typically the coastal plain regions, the river basins,
and the broad depositional plains. Rice areas on dry land are located on steeper slopes and higher topographic
positions.
        Fully irrigated wetland rice areas produce two or more crops  of rice annually. Partially irrigated
wetland rice regions produce one crop of rice a year, and are frequently seeded with dryland crops during the
drier part of the year. Dryland rice regions generally produce one rice  crop a year, and are seeded with other
crops during the drier part of the year. A shift from rice to other dryland crops, or vice versa, is dictated* by
the relative economic advantages of growing each crop.

Climate Restrictions to  Crop Production  in the Philippines

        Mean  annual rainfall ranges from 1,200 mm to 4,600 mm and mean annual temperature at middle-
to low-elevation ranges from 25.8 C to 27.9. C. Climatic differences at middle and low elevations are primarily

                                          PHILIPPINES-3

-------
due to rainfall distribution from typhoons and monsoon rains. Agriculture is very vulnerable to climate
hazards, such as the occurrence of tropical cyclones and floods, and to the variability in the dates of the rainy
season that may cause drought during some periods. A large proportion of the potential rice production is lost
each year due to tropical cyclones, floods, and droughts. For example, during the severe drought in 1973, 0.5
million t of potential rice harvest was lost. In 1979 about 1 million t of rice was lost due to floods and
typhoons. In January of 1991, the Philippine weather bureau (PAGASA) confirmed that several provinces were
hit by a drought related to the "El Nino" oscillation. Only 10 weeks after the onset of the El Nino drought,
rice and corn crops suffered estimated damages of $752.8 million.
       Scientists predict global climate changes from an increase of atmospheric CO2 and other gases (IPCC
1990). The objective of this study was to evaluate potential climate change impacts on rice production in the
Philippines.

METHODS

Baseline Climate  Data

       Daily maximum and minimum temperature and precipitation data for the period 1980-89 for Batac
and Los  Bafios were used to estimate 30 years  of daily weather data for each location  using a weather
generator program (Richardson and Wright 1984, Jones et al. 1990).

Climate Change Scenarios

       Climate change scenarios were created with the output of three equilibrium GCMs combined with the
daily climate data for each site. The GCMs used were:  Goddard Institute for Space Studies (Hansen et al.
1983), Geophysical Fluid Dynamics  Laboratory (Manabe and Wetherald  1987),  and United Kingdom
Meteorological Office (Wilson and Mitchell 1987). The mean seasonal temperature and precipitation changes
projected by the GCM scenarios are presented in Table 2. The  mean annual temperatures are projected to
increase from 2.1 C to 3.7 C, while annual precipitation is projected to decrease by as much as 14% under
the GISS and UKMO scenarios, and to increase slightly under the GFDL scenario.
       Transient scenarios were created with the GISS transient model for the decades of the 2010s, 2030s,
and 2050s (Hansen et al. 1988). The atmospheric CO2 concentrations considered were 405 ppm, 460 ppm, and
530 ppm, for the respective decades. This study also included  a sensitivity test of simulated rice yield to
arbitrary  changes in temperature (+2C and +4C).

Crop Model and Management Variables

       Crop model The simulation of rice growth was performed with the CERES-Rice model developed by
the International Benchmark Sites Network for Agrotechnology Transfer (IBSNAT) (Jones et al. 1990, Godwin
et al.  1992).  The  CERES-Rice  model  can  simulate  increases  in photosynthesis and  changes  in
evapotranspiration driven by increases in atmospheric CO2; these responses are known as the "direct" effects
of CO2 (Acock and Allen 1985).
       Cultivars and planting dates. The rice cultivar IR64 was used for the simulation at both sites with an
observed growing cycle of 110 to 120 days. The population density was 25 plants m'2; the row spacing was 0.2
m; and the  planting dates were May 15 for Batac, and September 15 for Los Banos.
       Soils. The parent materials  of wetland rice soils are mainly recent alluvium and Pleistocene marine
sediments. In the higher topographic positions, the soils are composed of Miocene and Pliocene sediments.
The typical  soil in Batac is classified as fine, montmorillonitic, isohyperthermic Udorthentic Pellustert; at Los

                                         PHILIPPINES-4

-------
Baftos the typical soil is fine, mixed, isohyperthermic Typic Haplaquoll. The initial soil conditions for the
simulation are: (a) soil water initial condition = full; (b) soil pH = 6.8 for all layers; and (c) soil nitrogen
amendment = 50 kg N ha"1.
       Simulations. Simulations for the GCM scenarios and sensitivity analysis included climate change alone
and climate change with the physiological effects of CO2 (Ackok and Allen 1985) at the two sites. Simulations
of possible adaptive  strategies were  performed under the most  unfavorable  climate scenario in each
location-GFDL for Batac, and UKMO for Los Baftos.

Validation of the Crop Model

       The CERES-Rice model was validated for Los Baftos, comparing simulated parameters to the
observed experimental data  from the Los Bafios Irrigation and Nitrogen Study conducted in 1980 at the
International Rice Research Institute. The rice cultivar was IR36, and the observed and simulated yields were
closely related (6.57 t  ha"1 simulated and 6.90 t ha"1 observed).

Limitations of the Study

       The 30 years  of daily weather data used in this simulation study were generated from 10 years of
available data. Generated weather data, although statistically comparable with observed 10-year data, may not
project the actual daily occurrence of weather events over 30 years. Simulations under irrigated conditions used
the automatic irrigation option of the model, but may not represent the actual practice in the Philippines and
may lead to overestimated rice yields.  However, because all of the simulations were done under the same
conditions, the relative changes in yield under the different climate scenarios compared to the base yields may
still be considered significant. The rice yields obtained in the simulation are "potential" yields because the
simulation considered  that nutrients are nonlimiting; weeds, insect pests, and diseases are controlled; and there
are no typhoons over  the simulated period.

RESULTS AND DISCUSSION

Sensitivity of Simulated Rice Yields to Temperature Increases and CO2

       A temperature increase of 2C caused decreases in yield at both locations, a 15% reduction in Batac,
and a 27% reduction in Los Baftos (Table 3). An increase of 4 C above the base temperatures resulted in yield
decreases of 27% in Batac and 53% in Los Baftos. The physiological effects of CO2 (555 ppm in the Table)
in the +2C scenario  compensated for the yield losses due to the increase in  temperatures at both locations.
Nevertheless, in the +4C scenario, the physiological effects of CO2 only partially compensated for the yield
losses, and significant  yield decreases were still evident, especially in Batac (-25%).

Rice Yields Under GCM Equilibrium and Transient Climate Change Scenarios

       Simulated rice yields decreased under all climate change scenarios in comparison to baseline yields
(Table 4 and Figure 2). The largest yield decreases were simulated in Los Baftos under the GISS and UKMO
scenarios. These yield decreases may be partially a consequence of a shortened crop season due to higher
temperatures. In Los Baftos, the length of the rice season decreased more than two weeks under the GISS and
UKMO scenarios. The shortening of the rice-growing season in Los Bafios was also responsible for a decrease
in simulated total crop evapotranspiration.
       When the direct effects of increased CO2 (555 ppm) were considered in the simulations with the

                                         PHILIPPINES-5

-------
climate change scenarios, the negative impact of climate change on rice yield was offset in Batac. In Los Banos,
the rice yield decreased significantly under the GISS and UKMO scenarios, even when the direcfeffects of CO2
were considered.
        A linear decrease in yields occurred under the GISS transient scenarios (2010s, 2030s, and 2050s)
(Table 5). In Batac, the physiological CO2 effects offeet the yield decreases, but not in Los Banos.

Adaptation Strategies to Climate Change

        Because our results indicated taht climate change may cause decreases in rice yields in some regions
in the Philippines, we considered possible adaptation strategies. The strategies were evaluated under the most
severe climate change scenario (the one that resulted in the largest yield decreases) for each site: GFDL in
Batac, and UKMO in Los Banos. We tested a range of rice cultivars and various planting dates. Table 6 shows
the results of the adaptation simulation strategies; all simulations include the direct CO2 effects on rice yields.
        In Batac, planting rice one month earlier increased simulated rice yield of all cultivars tested under
the GCM climate change scenario. In Los Banos, the optimum planting date of the original cultivar (IR64)
under climate change conditions was one or two months earlier than the current planting date. The highest
simulated yields under climate change conditions for Los Banos were obtained by planting the cultivar IR43
one or two months before the current planting date.
        September is currently the recommended planting month for rice in Los Banos for the dry season.
September planting dated allow a farmer to harvest rice 4-5 times in two years. If climate conditions  force
farmers to plant earlier, the rice cropping pattern could be dramatically affected. Additional problems may
arise when shifting the crop calendar. For example, in Batac, farmers report that a July planting date may not
be appropriate for rice because strong  winds in October and November affect the grain-filling stage. Thus,
additional climate factors such as wind  need to be considered in the development of adaptation strategies.
        These adaptation experiments suggest that the negative impact of climate change on rice yields can
be overcome by planting earlier and changing cultivars, but they also imply major changes to the current
farming system of the Philippines.  Such  changes  may lead to widespread  and varied alterations in  rural
regions.
                                          PHILIPPINES-6

-------
REFERENCES

Acock, B., and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
       and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation. DOE/ER-0238. U.S.
       Department of Energy. Washington, D.C. pp. 53-97.

Godwin, D., U. Singh, J.T. Ritchie, and E.C. Alocilja. 1992. A User's Guide to CERES-Rice. International
       Fertilizer Development Center. Muscle Shoals, AL. (in press).

Hansen, J., G. Russell, D. Rind, P. Stone, A. Lascis, S. Lebedeff, R. Ruedy, and L. Travis. 1983. Efficient
       three-dimensional global models for climate studies: Models I and II. Monthly Weather Review 3:609-
       622.

Hansen, J., I. Fung, A. Lascis, D. Rind, G. Russell, S. Lebedeff, R. Ruedy, and P. Stone. 1988. Global climate
       changes as forecasted by the GISS 3-D model. Journal of Geophysical Research 93:9341-9364.

Jones, J.W., S.S. Jagtap, G.Hogenboom, and G.T. Tsuji. 1990. The structure and function of DSSAT. pp 1-14.
       In: Proceedings of IBSNAT SYMPOSIUM: Decision Support System for Agrotechnology Transfer. Las
       Vegas. N.V. 16-18 Oct. 1989. University of Hawaii, Honolulu, Hawaii.

IPCC1990. The IPCC Scientific Assessment, eds. J.T. Houghton, G.J. Jenkins, and J.J. Ephraums. Cambridge
       University Press.

Manabe,  S., and R. Wetherald.  1987. Large-scale changes  of soil  wetness induced  by an  increase in
       atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.

Richardson, C.W., and D.A. Wright. 1984. WGEN:A model for generating daify weather variables. ARS-8. U.S.
       Department of Agriculture, Agricultural Research Service. Washington, DC. 83  pp.

Wilson, C.A, and J.F.B. Mitchell. 1987. A Doubled CO2 Climate Sensitivity Experiment with a Global Climate
       Model Including a Simple Ocean. Journal of Geophysical Research 92:13315-13343.
                                         PHILIPPINES-7

-------
Table 1.
National and regional rice production in the Philippines (1980-89).
Region
CAR
Ilocos * BATAC
Cagayan Valley
Central Luzon
S Tagalog * LOS BANGS
Bicol
W. Visayas
C. Visayas
E. Visayas
W. Mindanao
N. Mindanao
S. Mindanao
C Mindanao
NATIONAL
Area
haxlOOO
78
299
324
472
378
310
464
110
198
136
120
186
243
3,317
Production
txlOOO
170
725
826
1,461
869
646
1,112
158
370
321
325
558
706
8,246
Yield
T/Ha
2.21
2.42
2.64
3.09
2.30
2.08
2.40
1.45
1.86
2.36
2.71
3.03
2.90
2.49
*Sites modeled
                                         PHILIPPINES-8

-------
Table 2.        Seasonal and annual temperature and precipitation changes in the GCM climate change
               scenarios.
   SITE/
   Season
                          Temperature changes (C)
                                       Precipitation changes (%)
GISS
GFDL     UKMO
          GISS
                                              GFDL
                      UKMO
   BATAC
    Winter
    Spring
    Summer
    Fall

    Annual
  2.3
  3.0
  2.7
  3.5

  2.9
   4.3
   2.4
   2.5
   2.3

   2.9
3.0
0.9
2.3
3.1

2.3
-33
-17
-25
 20

-14
-12
  3
 37
-13
 31
-49
 -9
 12

 -2
   LOS BANGS
    Winter
    Spring
    Summer
    Fall

    Annual
  3.3
  3.2
  3.2
  3.8

  3.4
   2.5
   2.1
   2.1
   1.7

   2.1
4.2
4.0
2.9
3.5

3.7
 13
 11
-47
 10

 -5
-40
 13
 27
 25
-13
-27
 -5
  7

 -5
                                         PHILIPPINES-9

-------
Table 3.
Sensitivity of simulated rice yields to temperature increases and CO2
                                                        SIMULATED YIELD (t ha'1)
SITE
Los Banos


Batac


TEMPERATURE
CHANGE
0 (base)
+2C
+4C
0 (base)
+2C
+4C
Climate Scenario
Alone
(330 ppm CO2)
5.92
5.03
4.34
4.03
2.92
1.89
Climate Scenario +
P.E. of CO2 (555
ppm CO2)
7.01
6.23
5.59
5.34
4.12
3.00
P.E. - Physiological effects of CO2: the direct beneficial effects of 555 ppm CO2 on rice yield are included in
the simulation.
                                        PHILIPPINES-10

-------
Table 4.
Simulated yield, season length, and evapotranspiration (ET) under GCM climate changes
               scenarios.
                             SIMULATED YIELD (t
                           	ha'1)	
                            Scenario     Scenario +
                           Alone (330    PE of CO2
                                         Simulated
Season
Simulated
SITE

Batac



SCENARI
O
BASE
GISS
GFDL
UKMO
ppm CO2)

5.91
4.99
4.93
5.51
ppm CO2

7.19
6.19
6.11
6.77
(days)

121
118
118
118
(mm)

1,659
1,263
2,151
1,409
(mm)

586
637
631
343
Los Banos



BASE
GISS
GFDL
UKMO
3.86
2.09
2.75
2.04
5.15
3.22
3.91
3.14
120
103
109
103
928
1,057
982
983
421
402
409
403
P.E. - Physiological effects of CO2.
                                        PHILIPPINES-11

-------
Table 5.
Simulated rice yields under the GISS transient scenarios and the GISS equilibrium scenario.
                                                    SIMULATED YIELD (t ha'1)
SITE
Batac




SCENARIO
BASE
GISS 2010s
GISS 2030s
GISS 2050s
GISS
Scenario
Alone
5.91
5.49
5.31
4.81
4.99
Scenario
+ P.E. of CO2*
~
5.86
6.03
5.86
6.19
          Los Banos            BASE                3.86
                             GISS 2010s             3.21
                             GISS 2030s             2.82
                             GISS 2050s             2.12
                               GISS                2.09

* See METHODS for the CO, levels under each scenario.
                                                          3.58
                                                          3.54
                                                          3.20
                                                          3.22
                                       PHILIPPINES-12

-------
Table 6.        Effect of changes in rice cultivar and planting dates on rice yields simulated with the
               CERES-Rice model under base climate and GCM climate change scenarios.
Site Scenario
Batac BASE
GFDL
GFDL
GFDL
GFDL
GFDL
GFDL
GFDL
Los Banos BASE
UKMO
UKMO
UKMO
UKMO
UKMO
UKMO
UKMO
UKMO
UKMO
Cultivar
IR64
IR64
IR64
IR64
IR43
IR43
IR43
UPLR 15
IR64
IR64
IR64
IR64
IR64
IR43
IR43
IR43
IR43
UPLR 15
Planting Date
May 15
May 15
Apr 15
June 15
May 25
Apr 15
June 15
May 15
Sept 15
Sept 15
Aug 15
July 15
Oct 15
Sept 15
Aug 15
July 15
Oct 15
Sept 15
SIMULATED
YIELD t ha'1
5.91
6.11
6.48
5.75
6.42
6.83
6.19
5.01
3.86
3.14
3.93
4.87
2.69
3.39
4.17
5.03
2.79
2.02
                                         PHILIPPINES-13

-------

-------
                                  PHILIPPINES
Figure 1.      Map of the Philippines and location of the sites selected for the simulations.

-------
                % CHANGE FROM BASE YIELD
                     BATAC
                              LOS BANOS
GISS.330
GFDL.555
                                    GISS.555     HH GFDL.330
                                    UKMO.330    CHI UKMO.5S5
Figure 2.       Simulated rice yield changes under GCM climate change scenarios.

-------
EFFECTS OF CLIMATE CHANGE ON RICE PRODUCTIONAND STRATEGIES FOR
                     ADAPTATION IN SOUTHERN CHINA
                   Zhiqing Jin, Daokou Ge, Hua Chen, and Juan Fang

                          Department of Agrometeorology
                       Jiangsu Academy of Agricultural Sciences
                            People's Republic of China
                                  CHINA-1

-------
ACKNOWLEDGEMENTS

The authors would like to express their profound appreciation to Dr. C. Rosenzweig, Prof. L.Z. Gao, Dr. J.T.
Ritchie, and Dr. T.Y. Chou for their suggestions, encouragement, assistance, and advice in this work.
                                          CHINA-2

-------
TABLE OF CONTENTS

SUMMARY
INTRODUCTION
       Agroecological System
       Literature Review
METHODS
       Climate Data
       Climate Change Scenarios
       CERES-Rice Crop Model
       Management Variables and Soils
       Calibration and Validation
       Simulation Scenarios
       Crop Index
RESULTS AND DISCUSSION
       Impacts on Rice Yields
       Impacts on Season Length
       Impacts on Irrigation
       Impacts on Cropping Systems
       Adaptation Strategies
REFERENCES
                                         CHINA-3

-------
SUMMARY

        The CERES-Rice model was calibrated and validated for nine sites in Southern China to examine its
suitability to model rice production in the region, using agronomic data from three or more successive years.
After determining the genetic coefficients for the cultivars, the CERES-Rice model was run a second time for
the same locations for a time period of 20-30 years. The model used local climate data (1958-86) and doubled-
COZ climate change scenarios generated from the GISS, GFDL, and UKMO General Circulation Models
(GCMs), with and without supplemental irrigation (to model paddy and upland rice, respectively). This study
assessed the direct physiological effects of CO2 on rice growth for each scenario. Finally, the study examined
several strategies for adapting rice production to climate change.
    The results of the  study are listed below. They should not be regarded as predictions, but as plausible
assessments of the potential effects of climate change on rice production in Southern China.

(1)     Climate change alone

        Simulated rainfed rice yields decreased  under climate change alone due to increases in temperature
        that shorten the growing season for rice. For some sites, however, sharp decreases in precipitation
        were also an important factor in the decreased yield of rainfed rice.

        Rice yields simulated under "automatic" irrigation also decreased.  Although irrigation did not fully
        compensate for the  negative effects of climate change, it significantly improved simulated rice yields,
        especially in regions where precipitation decreased under climate change conditions.

(2)     Climate change with the direct effects of CO2.

        In rainfed rice, the direct effects of CO2 compensated for the negative effects of climate change alone
        in most sites, except in sites where rainfall sharply decreased in the climate change scenarios.

        In irrigated rice, the three GCM scenarios produced increases in modeled rice yields in comparison
        with the baseline yields in the northern sites, but decreases in the central and southern sites. These
        findings suggest that there is less compensation by the physiological effects of CO2 in areas with high
        temperature.

(3)     General results.

        Simulated irrigated  rice yields are higher  and have less year-to-year variability than rainfed yields.

        Under all climate change scenarios studied, the amount of water needed for automatic irrigation
        greatly increased in  areas where precipitation sharply decreased.

        Evapotranspiration  (ET) for rainfed rice was usually less than that for automatic irrigated rice.
        Therefore, cultivation of upland  rice may  be extended into areas where irrigation water is  not
        available.

       An increase in temperature would increase China's rice-based cropping system. The northern  limits
        for double-rice and  triple-rice cropping systems could be moved northward about 5-10 degrees of
       latitude, depending on the climate scenario.
                                             CHINA-4

-------
        Introducing upland rice cultivars to areas where precipitation sharply decreased under the climate
        change scenarios significantly improved rainfed rice yields at some sites, but not at others.

        For paddy rice, adjusting planting dates ameliorated the negative effects of climate change on modeled
        yields in the northern part of the study region, but not in the southern part.
INTRODUCTION

        Since 1949, China has made great progress in food production. With only 9% of the world's cultivated
land, China supports more than one-fifth of the world's population. However, with increasing population, a
sharp decrease in cultivated land,  lack of water resources, environmental pollution, and frequent natural
disasters, it is difficult for China to continue to increase its food production.
        Recently, scientists have suggested that the addition of greenhouse gases to the atmosphere will alter
global climate, increasing temperatures and changing rainfall and other climate patterns (IPCC 1990). The
combined effects of climate change and physiological effects of CO2 on crops may result in a net increase in
crop yields in some cases, especially in the high latitudinal regions. In low latitudes, agriculture could be
negatively affected (Parry et al. 1988).
        The objective of this study is to characterize the direction, magnitude, and degree of uncertainty of
the potential impacts of global climate change on rice production in Southern China. This study is important
because: (1)  China is the largest rice producer and consumer in  the  world; (2) Southern China plays an
important role in the nation's rice production; and (3) Southern China appears to be more vulnerable to
climate change than the more northern regions of China (Jin et al.  1990).
        The study is based on simulations with the CERES-Rice model. The model used actual daily weather
data (about 30 years) from nine sites in seven provinces of Southern China (Figure  1) and regional climate
change  scenarios generated from General Circulation Models  (GCMs). Rice growth is simulated for both
rainfed  and automatic irrigation conditions to represent some mountain rice areas without irrigation and the
plains rice areas with an extensive irrigation system. Changes in rice yield, growth duration, evapotranspiration,
and irrigation are estimated  under baseline and climate change conditions. The physiological effects of CO2
are also analyzed under each scenario.
        The results of this study should be considered as indicators of the possible impacts of global warming.

Agroecological System

        Southern China is subject to the influence of the monsoonal wind system of eastern Asia, which makes
the climate wet and hot in the summer, and cold and dry in the winter. Ample rainfall combined with high
temperatures and high intensity of solar  radiation  during the rice-growing season  are generally favorable
conditions for rice production.
        About 1,000 rivers and more than 200 large lakes provide a copious irrigation system, and rich soil
and intensive  farming have made this  region  one of the world's  greatest centers  of rice production.
Nevertheless, some unfavorable climate events constrain rice production. Summer and autumn droughts are
rather common, and typhoons, floods, and heat waves occur frequently.
        The  topography in Southern China is varied and complex. Altitude drops from west to east, leaving
only about 15% of the land for agricultural purposes.  Consequently, the farmers  must adopt a multiple-
cropping system.
        Southern China is comprised of 16 provinces: Anhui, Fugian, Guangdong, Guangxi, Guizhou, Hainan,
Hubei,  Hunan, Jiangsu, Jiangxi, Sichan, Taiwan, Tibet, Yunnan,  Zhejian, and Shangai.  Rice is the most

                                              CHINA-5

-------
 important crop in all of them, except in Tibet.
        Southern China can be divided into four rice subregions: the Middle and Lower reaches of the Yanzi
 River; the Southwest; South China; and the Yellow-Hui Rivers valleys. Table 1 shows information on land use
 and rice production for each subregion.

 Literature Review

        The impacts of 2xCO2 climate change on agricultural crops have been studied mainly by scientists from
 developed countries. In North America, Robertson et al. (1987) analyzed the impacts of climate change on
 yields and soil erosion for selected crops in the Southern U.S., Central Prairie, and Northern Plains using the
 Erosion Productivity Impact Calculator (EPIC);  Allen et al. (1987) reported that the soybean yield could
 increase about 32% with  a doubling of CO2; Smit (1989) studied how crop yields in Ontario (Canada) would
 be affected by climate change (GFDL and GISS scenarios); and Jones et al. (1986) used the Soybean Integrated
 Crop Management (SICM)  model to estimate potential changes in earworm damage to soybean under climate
 change scenarios, concluding that earworm damage may increase in the Grain Belt region under a warmer
 climate. Using an agroclimatic approach, Rosenzweig (1990) suggested that the wheat cultivars in the Southern
 Great Plains would change from winter to spring wheats due to the lack of cold winter temperatures under
 the climate change scenarios, and that climate change would  bring about an increased demand for irrigation.
        The CERES and Soygro models have been widely used  in the US: in the Great Lakes (Ritchie et al.
 1989), the Southeast (Peart  et al 1989) and Great Plains (Rosenzweig 1990), to simulate change in crop yields
 under different climate change scenarios. Rosenzweig (1990) characterized the direction, magnitude, and
 uncertainty of potential climate change-induced alterations in wheat and corn yields. The study suggested that
 adjusting  the planting dates of winter wheat and corn would  not significantly ameliorate the negative effects
 of climate change and that change of cultivars only could be a possible adaptation strategy to compensate for
 yield loss at some selected sites but not others. Her study also analyzed the direct effects of CO2 on crops (i.e.
 increased  photosynthesis and improved water use efficiency). Ritchie et al. (1989)  suggested that  climate
 change induced maize yield  reductions  in the Great Lakes could be partly compensated by using new cultivars
 with a longer  growing cycle. Parry et al. 1988 reported possible yield decreases of rice in India, and some
 increases in Japan under climate change conditions; Parry et al. (1988) reported that  climate change may cause
 rice yield  to decrease in Thailand.
        At the present time there is no published comprehensive study that evaluates the potential impacts
 of climate change on large-scale rice production in China and  that includes the beneficial direct effects of CO2
 on rice  production.

 METHODS

 Climate Data

        Daily climate data  (maximum and minimum temperatures, precipitation, and solar radiation) for a
period of 20-30 years (about 1957-1986) were taken from the Monthly Report of Chinese Meteorological
Record on the Surface (China Meteorological Bureau) and the Daily Solar Radiation Record of China (Beijing
Meteorological Bureau). The daily solar radiation at Xuzhou was not available and was calculated according
 to Gao  and Lu (1982).
        Climate  data  for  the calibration and  validation  experiments  were  collected from the  local
meteorological stations in the following institutions: China National Rice  Research  Institute, the National
Academy  of Agricultural Sciences, Guangdong  Academy  of  Agricultural Sciences, Jiangxi Agricultural
University, Jiangsu  Agricultural Bureau, Jiangsu Academy  of Agricultural Sciences, Shaoguan Prefecture

                                            CHINA-6

-------
Institute of Agricultural Sciences in Guangdon province, and Lixiahe Prefecture of Agricultural Sciences in
Jiangsu province.

Climate Change Scenarios

        Mean monthly temperature differences, precipitation and solar radiation ratios (2xCO2/lxCO2 GCM-
generated climate) were combined with daily historic climate data to generate climate change scenarios for the
various sites. The GCMs used were:  Goddard  Institute for Space Studies (GISS)  (Hansen et al. 1987);
Geophysical Fluid Dynamics Laboratory (GFDL) (Manabe and  Wetherald 1987);  and United  Kingdom
Meteorological Office (UKMO) (Wilson and Mitchell 1987).
        GCMs have several advantages over other approaches for creating climate scenarios (Smith and Tirpak
1989): (1) the models estimate how global climate may change  in response to increased CO2, and therefore,
regional outputs are internally consistent with global warming associated with doubled CO2; (2) the climate
variables estimated are in agreement with physical laws; and (3)  the GCMs provide the information necessary
to run the crop simulation models.
        GCMs, however, have some disadvantages: (1) the ocean  models currently used in most GCMs are
relatively simple (most GCMs treat oceans as "swamps" or only simulate their upper layers); (2) they have low
spatial resolution and do not provide a good representation of major geographic features, such as plateaus,
mountains and basins, which have large impacts on local climate; and (3) GCMs include  some simplified
assumptions about other factors, such  as cloud cover, albedo, and  the hydrology of the land surface.
        Table 2 presents annual mean  temperature and precipitation changes generated by the GISS, GFDL,
and UKMO scenarios for different locations in Southern China.

CERES-Rice Crop Model

        The CERES-Rice model (Godwin et al. 1992; IBSNAT 1989; Alocija and Ritchie 1988) was used for
the study. The model simulates yield and other parameters of crop growth and development under different
management and climatic conditions.
        The CERES-Rice model was  chosen because  (1)  it simulates the effects of major factors, such as
climate and soil management, on rice growth, development, and yield; (2) the model also simulates the direct
effects of CO2 on crop photosynthesis and evapotranspiration; (3) the genetic coefficients that are used to
characterize the different cultivars  can be calibrated based on the  experimental data; and (4) the model has
been available and documented for several years and has been validated under a wide range of soil and climatic
conditions.

Management Variables  and Soils

        Crop data of the local rice cultivars (sowing date, soil type, sowing depth, transplanting density, row
spacing, maturity date, biomass, and grain yields) were taken from local  field experiments. The following
institutions provided data: China National Rice Research Institute, the National Academy of Agricultural
Sciences, Guangdong Academy of Agricultural Sciences, Jiangxi Agricultural University, Jiangsu Agricultural
Bureau, Jiangsu Academy of Agricultural Sciences, Shaoguan Prefecture Institute of Agricultural Sciences in
Guangdon province, and Lixiahe Prefecture of Agricultural Sciences in Jiangsu province. Table 3 lists the main
rice production areas of Southern China, the sites selected for the study, the local rice cultivars used for
modeling, and the soil types at the various sites.
        Soil data included soil type, albedo, organic matter, texture, structure, and bulk density. Representative
soil types and profiles were  chosen according to the Soil Atlas  of China (Institute of Soil Science  1986) and

                                             CHINA-7

-------
 Chinese Soils (Xiong Yi and Li Qingkui 1988).

 Calibration of Genetic Coefficients

         Eight genetic  coefficients define a  rice cultivar in the  CERES-Rice model and  characterize
 quantitatively how that particular rice cultivar responds to environmental factors such as climate, management,
 and soils. The genetic coefficients for the rice cultivars used in this study were determined by contrasting the
 simulated  results  to  local  field  experimental  observations, and  included growth duration, dry-matter
 accumulation (if available), and yield. The management variables (planting date, plant density,  and spacing,
 eta) for each calibration site were taken from the field experiments and local climate data for the calibration
 years was used. The specific genetic coefficients of a local cultivar were determined by trial-and-error so that
 the simulated yield and maturation date were as close as possible to the observed values.

 Simulation Scenarios

        The three GCM climate change scenarios were simulated  at  each  site, and each simulation was
 completed under rainfed and irrigated conditions. Automatic irrigation was done to provide the crop with a
 nonlimiting water situation, which is  similar to practice. Since the climate change scenarios involve an
 equivalent doubling of CO2, the physiological CO2 effects on crop yield and water use were also included in
 simulations at each site (indicated as 555 ppm CO2 in the tables).
        In addition, the study evaluated the possible adaptive strategies to rice management under future
 climate. Scenarios were created by altering current agricultural practices in order to maximize yields under the
 conditions of climate change.

 Crop Index

        A Crop Index has been used  to evaluate different cropping systems in  China  (Gao et al.  1987).
 Possible changes in the index were analyzed by using simulated rice growth output under climate change
 conditions.
        According to Gao et al (1987), the rice growing season in China is defined as the number of days
 from the safe sowing date to the safe maturity date. It is also equal to the number of days from the safe sowing
 date to the safe heading period, plus 40 and 30 days for the Japonica and Indica varieties, respectively. The
 safe sowing date is defined as the time when the mean daily temperature is constantly above 10C  and the safe
 heading period is defined as the time when the mean daily temperature is above 20C, with no temperature
 lower than 20C for three consecutive days. Based on this index, it is easy to compute  the lengths of the
 growing season for rice under the current climate and for the three GCM climate change scenarios, and to
 evaluate the changes in rice cropping systems in the future.
        According to the  China National Rice  Research Institute (1988), a value of the accumulated
 temperatures above 10C for different rice-based cropping systems is:

        (1)     2,000C to 4,500C for single-rice cropping;
        (2)     4,500C to 7,000C for double-rice cropping (the northern limit for double-rice  is 5,300C);
               and
        (3)     more than 7,000C for triple-rice cropping.

        These thermal values can also be used to estimate potential changes in the rice cropping systems under
climate change scenarios.
                                             CHINA-8

-------
RESULTS AND DISCUSSION

Impacts on Rice Yields

        Doubled CO2 climate change would directly influence rice yields through physiological processes and
indirectly through climate.

        Climate Change Alone. The simulated rice yields decreased under the three scenarios at all locations.
Under the GISS scenario, rainfed yields decreased significantly (10%-78%) from baseline yield (Table 4).
Under the GFDL and the UKMO scenarios, yields decreased 6%-33% and 7%-34%, respectively. The yield
decreases were a result of the higher temperature under the climate change scenarios which shortened the
growing cycle of rice and also caused water stress in some regions. More severe yield decreases  (-78% at
Chengdu) under the GISS scenario were due in part to significant growing season precipitation decreases. For
example, the July rainfall at Chengdu under the GISS scenario declined 90% in comparison with the current
rainfall, resulting in a growing season with total rainfall of less than 400 mm (Jin et al 1990). Although the
enhancement of respiration due to high temperatures may be another reason to decrease rice yields, it was not
taken into account in the CERES-Rice model.
        The automatic  irrigation simulations under the three climate change scenarios also resulted in
decreases in rice yields in comparison  with the base data at all sites (Table 4): 15%-33% under the GISS
scenario; 14%-37%  under the GFDL  scenario;  and 19%-33% under the UKMO scenario. These results
indicate that full irrigation did not completely offset the negative effects of increases in temperature on rice
yields because of the shorter rice growth duration, particularly of the grain-filling stage. Nevertheless, the
negative effects of climate change on rice yields were partially offset by full irrigation at  the sites where the
rainfall during the rice growing season greatly decreases under the climate  change scenarios. For example, at
Chengdu and Guangzhou under the GISS scenario, rainfed yields decreased 78% and 45%, respectively, and
the irrigated yields declined only 32% and 16%. These results suggest that an effective strategy for adapting
to climate change would be to improve the irrigation systems in such regions.
        Table 4 also shows that the standard deviations of the percent yield changes-an indication of the
variability (Rosenzweig 1990)of irrigated rice yields were much lower than the standard deviations of rainfed
rice yields.

        Climate Change with Physiological CO2 Effects. Table 5 shows the physiological effects of CO2 on crop
yield and water use under the climate change scenarios. Under the GISS scenario, rainfed rice yields increased
at the northern sites (Xuzhou and Nanjing) and the eastern sites (Fuzhou and Nanchang). At these sites, the
scenario precipitation was not a limiting factor on rice production. In Wuhan, Changsha, and Shaoguan, the
direct effects of CO2 ameliorated,  to a certain degree, the negative effect of increased temperature on rainfed
rice yield. However, in Chengdu and Guangzhou, because of large decreases in precipitation during the rice
growing season, the yields remained significantly lower than base case yields, even with  the direct effects of
CO2.  In these two southern sites, the GISS scenario yields were very low in  comparison with the baseline, and
there were very small differences in yields  simulated with  and without CO2 physiological effects (Table 4).
Under the GFDL and UKMO scenarios, the direct effects of CO2 largely compensated for the climate change
impacts on rice yields in many but not all sites.
        In the irrigated simulations, rice yields declined in the central-southern region (Nanchang, Changsha,
Wuhan, and Shaoguan) under the three climate change scenarios (Table  5). This was a result of the large
annual temperature increases projected by the GCMs in that region (4C-7C). In contrast, the direct effects
of CO2 resulted in an increase of the modeled irrigated yield at Xuzhou under the  three  climate change
scenarios. These results suggest that increased temperatures would benefit rice production in the northern

                                              CHINA-9

-------
areas of the studied region.
        As in the case of climate change alone, the yield stability (standard deviation) under the rainfed
conditions was less than that under irrigated conditions.

Impacts on Season  Length

        The increment of temperature under all climate change scenarios caused the rice growth duration to
decrease at all sites (Table 6). In the northern areas (Xuzhou, Nanjing, and Chengdu), the mean rice growth
duration decreased more than in the southern areas. Because the current temperatures in the northern sites
are relatively low, rice has a long growing period which is more sensitive to the temperature increase projected
by the GCM climate change scenarios. In contrast, the current temperatures at the southern sites already cause
the rice crop duration to be very short, and a further increase in temperature did not have a very large effect
on simulated growth duration. The physiological effects of CO2do not influence season length, as simulated
with the CERES-Rice model.

Impacts on Irrigation

        An adequate water supply is one of the most important factors in rice production. The amount of
irrigation required for rice depends on many factors, such as growing season rainfall, temperature, solar
radiation, and evapotranspiration (Table 7), as well as crop and soil characteristics, depth of water table, and
topography. Among these factors,  rainfall is the most important. Table 8 shows the percent change in
irrigation demand for rice under the climate change scenarios, including the direct effects of CO2. Simulations
were done with automatic irrigation (100% efficiency of application and availability of water).
        Under the GISS scenario, the demand for irrigation rose at six of the nine sites. At Chengdu and
Guangzhou, the irrigation demand  increased six and two times, respectively, over the present levels. This
increase was caused by large decreases in the summer rainfall and increases in temperature and solar radiation.
        Under the GFDL scenario the irrigation demand decreased in most sites where precipitation during
the local rice growing season increases. Under the UKMO scenario irrigation demand results were mixed: they
increased in the sites where projected precipitation decreased, and  increased at the rest of the sites.

Impacts on  Cropping Systems

        Under all the doubled CO2 scenarios, the temperature, as well as the >10C accumulated temperature
index, increased and could result in an extension of the growing season for rice over large areas of Southern
China. However, the increased temperature shortened the simulated  lifecycle of rice and had a negative impact
on yields. The combined effects of a prolonged growing season and shortened growth duration would shift  the
northern limits of the various rice-based cropping systems towards higher latitudes. As a result, the Crop Index
in China would increase and  the varieties and management systems would have to be adjusted to the new
conditions.
        The current climatic classification for rice production in China would no longer be applicable under
climate change conditions. Table 10 lists the onset of the rice cropping season, length of the growing season,
and the >10C accumulated temperatures during the rice growing season at selected sites. Under the GISS
scenario, the rice maturity dates in China advanced by an average of 19 days, the length of the rice growing
season was prolonged by an average of 45 days, and the 10C accumulated temperature increased by an average
of 1,522C. Figure  2 shows the northern limits for double- and triple-rice crops.
        According to the GISS scenario, the thermal conditions at Beijing were more favorable for rice than
the current thermal conditions  at Nanjing, and  it would be suitable to grow rice after wheat. The thermal
                                            CHINA-10

-------
conditions at Shenyang would be more favorable than those at Beijing, without any low temperature problems
to sustain the growth pattern of the three crops in two years. In the most northern area of China, Harbin, the
>10C accumulated temperature would reach 3,696C, with a rice-growing season of 159 days, and it would
be safe to grow single-rice; early Japonica with a late maturity date.
        Similar results were obtained with the GFDL scenario since the temperature scenario increases are
comparable to those under the GISS scenario. The UKMO scenario resulted in an extreme situation: the
northern limit for double-rice would shift to the Shenyang region, and that for the triple-rice would move to
a line extended from  Jinan  to  Zhengzhou.  In  the southern parts of China, triple-rice could be grown
throughout the year.
        Currently, Indica rice varieties are grown in the south and Japonica varieties in  the north.  With
climate change, the patterns of rice varieties used throughout China would shift and the traditional practice
of "South Indica" and "North Japonica" would no longer hold. The high-temperature-tolerant varieties would
not only dominate in the southern parts of China, but would be likely to be more widely used in the north.
        Based on the above analysis, a 2xCO2 climate change would result in a tremendous change in China's
thermal conditions. Therefore, effective strategies for adapting to climate change need to be developed, such
as improving crop breeding (especially the use of biotechnology to develop more heat- and drought-resistant
rice varieties); expanding the use of chemical fertilizers, pesticides, and herbicides; and improving irrigation
systems. These strategies might be quite costly, as they imply an increase in agriculture-related inputs.

Adaptation Strategies

        This section presents and  discusses some possible strategies for adaptation to climate change. The
GISS scenario was chosen because it had the most negative effects on crop yield. All adaptation simulations
included the physiological effects of CO2 was analyzed (Table 9).
        Using crop model simulations, the  following strategies  for adaptation  to climate  change were
evaluated.

        Change in rice cultivar. This strategy considers whether new cultivars would improve rice yield under
climate change conditions. An upland rice cultivar (UPLR15) was simulated at the sites where precipitation
declined dramatically under the scenario conditions (Chengdu and Guangzhou) (Table 9). With the new
cultivar, the yield at Chengdu increased in comparison with the yield of the originally used cultivar. Compared
with the baseline yield the decrease under climate change conditions was still significant (-54% with the new
cultivar and -72% with the original cultivar). At Guangzhou, however, introducing the same cultivar did not
improve the yield under the scenario conditions at all.
        The  cultivar IR43 was  simulated at seven locations. At five sites  (Nanchang, Fuzhou, Xuzhou,
Changsha, and Wuhan), the yields of the new cultivar were higher that those of the original cultivar under the
conditions of climate change. In  the other two sites  (Nanjing and Shaoguan), the change of cultivar did not
improve modeled yields.

       Adjusted planting dates. Changes in the planting dates (by 10 day intervals) of the present cultivars may
increase yields under the climate change conditions. Changing the planting date caused rice yields to increase
at the northern sites (Xuzhou and Nanjing), but not at the southern sites.

        Changes in both cultivars and planting dates. Rice yields increased significantly  when a combination
and change in planting date was simulated at six of the seven locations tested (all except Nanjing). The yield
increases ranged from +3% to +43%, and the mean increase was +22% (the calculations assume equal area).
                                             CHINA-11

-------
        Improvement of the irrigation system. Yields simulated under rainfed and irrigated conditions were
compared at each site to determine if irrigation may mitigate the negative effects of high temperatures on
yields.
                                            CHINA-12

-------
REFERENCES

Allen Jr., L.H., K.J. Boote, J.W. Jones, P.H. Jones, R.R. Valle, B. Acock, H.H. Rogers, and R.C. Dahlman.
        1987. Response of vegetation to rising carbon dioxide: photosynthesis, biomass, and seed yield of
        soybean. Global Biogeochemical Cycles.  1:1-14.

Alocilja, E.C., and J.T. Ritchie. 1988. Rice Simulation and its use in Multicriteria Optimization. IBSNAT
        Research Report Series 01.

China National Rice Research Institute. 1988. A Classification for Rice Planting in China. Zhejing Scientific
        and Technological Publishing House. Hangzhou.

Gao,  L., L. Lin, and J. Zhiqing. 1987. A classification for rice production in China. Ag. and Forest Meteor.
        39:55-65.

Gao,  G., and Y.  Lu. 1982. Radiation and Heat Balance on the  ground surface in  China.  Scientific and
        Technological Publishing House. Beijing.

Godwin, D., U. Singh,  J.T.  Ritchie, and E.C.  Alocilja. 1992. A user's guide to Ceres-rice. International
        Fertilizer Development Center. Muscle Shoals, AL.

Hansen, J., G. Russell,  D. Rind, P. Stone, A Lacis, S. Lebedeff, R. Ruedy, and L. Travis. 1983. Efficient
        Three-Dimensional Global Models for Climate Studies: Models I and II. April Monthly Weather
        Review, Vol HI, No. 4: 609-662.

Institute of Soil Science, Academic Sinica. 1986. The Soil Atlas of China. Cartographic Publishing House.
        Beijing.

IBSNAT. International Benchmark Sites Network for Agrotechnology Transfer Project (IBSNAT).  1989.
        Decision  Support System for Agrotechnology Transfer Version 2.1 (DSSAT V2.1). Dept. Agronomy
        and Soil Sci., College of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii
        96822.

IPCC. Intergovernmental Panel on Climate Change. 1990. First Assessment Report

Jin, Z.,  X. Zhen, J. Fang. D. Ge, and H. Chen. 1990. Potential effects of global climate change on winter wheat
        production in China. Internal Report.

Jones,  C.A., and J.R.  Kiniry  (eds.). 1986. CERES-Maize: A Simulation  Model of Maize Growth and
        Development. Texas A&M University Press. College Station, TX. 194 pp.

Jones, J.W., S.S. Jagtap, G. Hoogenboom, and G.Y. Tsuji. 1990. The structure and function of DSSAT. pp 1-
        14.  In: Proceedings of IBSNAT Symposium: Decision Support System for Agrotechnology Transfer,
        Las Vegas, NV. 16-18 Oct. 1989. Part I: Symposium Proceedings. Dept. of Agronomy and Soil Science,
        College of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii 96822.

Manabe, S., and R.  Wetherald.  1987.  Large-scale changes  of soil  wetness induced by  an increase  in


                                            CHINA-13

-------
        atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.

Parry, M.L., T.R. Carter, and N.T. Konijn. 1988. The impact of climatic variations on agriculture.  Vol 1
        Assessments in cool temperate and cold regions.  Vol 2 Assessments in semi-arid region. Kluwer,
        Dordecht, Netherlands. 876 pp. and 764 pp.

Peart, R.M., J.W. Jones, R.B. Curry, K. Boote, and L.H. Allen, Jr. 1988. Impact of climate Change on Crop
        Yield in the Southeastern U.S.A.  In J.B. Smith and D.A. Tirpak (eds). Report to Congress. The
        Potential Effects  of Global Climate Change  on the  United States. U.S. Environmental Protection
        Agency. Washington, DC. EPA-230-05-89-050.

Ritchie, J.T., B.D. Baer, and T.Y. Chou. 1989. Effect of global climate change on agriculture: Great Lakes
        region.  In  Smith, J.B. and D. Tirpak (eds.). The Potential Effects of Global Climate Change on the
        United States. Report to Congress.  EPA-230-05-89-050. U.S. Environmental Protection Agency.
        Washington, DC. Appendix C-l, pp. 1-1 to 1-30.

Rosenzweig, C. 1990. Crop response to climate change in the Southern Great Plains: A simulation study.
        Professional Geographer 42(l):20-37.

Smit, B. 1989. Climatic warming and Canada's comparative position in agricultural production and trade. In
        Climate Change Digest CCD 89-01. Environment  Canada, pp. 1-9.

Smith, J.B., and D.  Tirpak (eds.). 1989. The Potential Effects of Global Climate Change on the United States.
        Report to Congress. EPA-230-05-89-050. U.S. Environmental Protection Agency. Washington, DC
        423pp.

Wilson, C.A., and J.F.B. Mitchell. 1987. A doubled CO2 Climate Sensitivity Experiment with a Global Model
        Including a  Simple Ocean. Journal of Geophysical Research, 92: 13315-13343.

Xiong Yi, and Li Qinqkui. 1987. Chinese Soils (Second Edition). Scientific Publishing House, Beijing.
                                            CHINA-14

-------
Table 1.
Regional contribution to rice production in China (1980-86).
      Regions *
                Contribution to Total Rice
                       Production2
Contribution to Total Rice
          Area
Yangzi River
South China
South West China
Yellow and Huir
69.9
15.7
7.6
6.0
68.1
17.7
7.8
6.0
               1 Study sites includes in each region: Nanjing, Whuan and Changsha in Yangzi River
               Region; Fuzhozi, Sshaoguan and Guangzou, in South China Region; Chengdou in South
               West China; Xuzhou and Xiam in Yellow and Huir Region.
               2 National Rice Production in China: 32,975,000 ha,  161,770,000 t, 4.92 t ha'1 (average
               yield).
                                           CHINA-15

-------
Table 2.
Annual temperature changes (C) and precipitation differences (%) projected by the GISS,
GFDL and UKMO climate change scenarios for sites in southern China.
                           Temp. Diff. (C)
                                                  Precip. Diff. (%)
Site
Nanjing
Xuzhou
Nanchang
Wuhan
Changsha
Chengdu
Fuzhou
Shaoguan
Guangzhou
GISS
4.1
4.1
3.5
6.0
6.0
4.7
3.5
6.0
3.5
GFDL
4.6
4.7
4.4
4.4
4.4
3.2
4.2
4.0
4.0
UKMO
6.5
6.5
5.7
6.5
5.7
6.7
5.7
3.3
3.4
GISS
43.7
45.2
17.8
-9.1
-7.3
-20.4
18.6
-7.5
79.9
GFDL
9.3
4.6
34.9
35.0
30.3
89.1
-11.4
6.8
9.3
UKMO
13.1
14.0
0.5
12.3
-0.6
34.0
1.6
26.2
31.8
                                          CHINA-16

-------
Table 3.
Representative rice cultivars and soil types.
Region
Yellow & Huir
Yangzi R. Valley



South West
South China


Site
Xuzhou
Nanjing
Nanchan
Wuhan
Changsha
Chengdu
Fuzhou
Shaoguan
Guangzhou
Latitude
34 19' N
3200' N
2836' N
303T N
2812'N
3040' N
26 05' N
24 48' N
23 08' N
Cultivar
Y.G.#2
S.U.#63
W.U.#64
Z.X.#26
Z.X.#26
G.C.#2
G.L.A.#4
S.U.G.#33
G.C.#2
Soil type
Fluoro-aquic
Yellow brown loam
Red earth
Yellow brown loam
Red earth
Yellow earth
Red earth
Lateric red earth
Lateric red earth
                                             CHINA-17

-------
Table 4.        Simulated yield changes (% of base yields) under GCM climate change scenarios without the
               direct effects of CO2.
RAINFED
Site
Nanjing
Xuzhou
Nanchang
Wuhan
Changsha
Chengdu
Fuzhou
Shaoguan
Guangzhou
Mean
GISS
Y SD1
-20
-10
-23
-27
-27
-78
-21
-33
-45
-31
10
8
12
8
12
5
15
9
12
9
GFDL
Y
-33
-31
-5
-14
-6
-14
-19
-22
-27
-21
SD
10
8
14
9
13
2
6
9
4
7
UKMO
Y
-33
-23
-35
-28
-11
-28
-18
-8
-25
-21
SD
10
8
12
9
13
3
6
10
4
8
AUTOMATIC IRRIGATION
GISS
Y
-17
-15
-24
-22
-33
-32
-28
-25
-16
-24
SD
4
3
4
5
5
2
5
5
4
4
GFDL
Y
-20
-18
-19
-26
-37
-14
-25
-30
-21
-21
SD
4
3
4
5
5
2
5
5
4
4
UKMO
Y
-20
-19
-33
-26
-31
-20
-27
-26
-22
-24
SD
4
3
4
5
5
2
5
5
4
4
 1 expressed as %
                                            CHINA-18

-------
Table 5.        Simulated yield changes (% of base yields) under GCM climate change scenarios with the
               direct effects of CO2 included in the simulation.
RAINFED
Site
Nanjing
Xuzhou
Nanchang
Wuhan
Changsha
Chengdu
Fuzhou
Shaoguan
Guangzhou
Mean
GISS
Y SD1
4
20
2
-9
-2
-72
5
-14
-32

11
15
14
10
13
2
7
10
5

GFDL
Y
-13
-9
39
11
27
4
7
5
-6

SD
11
8
15
10
14
3
7
10
4

UKMO
Y
-12
3
-13
-6
21
-3
10
9
19

SD
11
8
13
10
14
4
7
10
4

AUTOMATIC IRRIGATION
GISS
Y
3
6
0
-3
-12
-6
-4
-3
3

SD
4
3
5
6
6
3
6
5
4

GFDL
Y
-2
2
4
-5
-15
3
0
-7
-2

SD
4
3
4
6
6
3
6
6
6

UKMO
Y
-2
1
-11
-5
-9
3
-3
-5
-5

SD
4
3
4
6
6
3
5
4
4

 1 expressed as %
                                            CHINA-19

-------
Table 6.
Season length changes (days??) with simulation of automatic irrigation.
                                330 ppm CO2
                                               555 ppm CO2

Nanjing
Xuzhou
Nanchang
Wuhan
Changsha
Chengdu
Fuzhou
Shaoguan
Guangzhou
Mean
GISS
-16
-22
0
-7
-8
-16
-12
-7
-13
-13
GFDL
-15
-20
-6
-14
-14
-15
-11
-17
-21
-16
UKMO
-17
-23
-6
-15
-13
-26
-15
-12
-16
-17
GISS
-15
-20
-1
-7
-8
-15
-11
-5
-11
-11
GFDL
-14
-19
-5
-13
-14
-14
-10
-15
-20
-15
UKMO
-16
-22
-5
-15
-13
-26
-14
-10
-15
-16
                                         CHINA-20

-------
Table 7.
Simulated changes in evapotranspiration.
                                             Percent ET Changes
                                 RAINFED               AUTOMATCIRRIG.

Nanjing
Xuzhou
Nanchang
Wuhan
Changsha
Chengdu
Fuzhou
Shaoguan
Guangzhou
Mean
GISS
15
10
-1
0
3
-27
3
2
-13
-1
GFDL
0
-5
-10
-7
-2
-8
-6
-4
-8
-5
UKMO
8
5
-1
3
13
1
3
-7
-16
2
GISS
12
10
-5
12
8
25
0
21
7
10
GFDL
4
2
-18
-14
-17
-9
-7
-5
-5
-8
UKMO
14
12
-8
-3
-7
16
-4
-5
-6
1
                                         CHINA-21

-------
Table 8.
Simulated changes in irrigation. The direct effects of CO2 were included in the simulation.
Site
Nanjing
Xuzhou
Nanchang
Wuhan
Changsha
Chengdu
Fuzhou
Shaoguan
Guangzhou
% Irrigation water
changes
GISS GFDL UKMO
-2
8
11
39
25
603
-16
69
194
4
14
-12
-41
-48
53
-9
-8
35
16
22
18
-6
-22
69
-22
-31
0
                                            CHINA-22

-------
Table 9.        Simulated yield changes for different adaptation strategies with the GISS climate change
               scenario. The direct effects of CO2 were included in the climate change scenario simulation.
                                                    Strategy
Site
Nanjing
Xuzhou
Nanchang
Wuhan
Changsha
ChengduH-
Fuzhou
Shaoguan
Guangzhou+
CC% CC+PD% CC+C% CC+PD+C%
3
6
-1
-3
-12
-72
-4
-3
-32
14
10
3
-1
-10
~
-2
5
-
-2
10
40
6
10
-54
29
-1
-47
0
11
43
21
25
-41
30
3
-36
                   + Rainfed, the rest are automatic irrigation.

                   CC - Climate change effect (GISS scenario).
                   CC+PD - Effect of climate change plus change in the sowing
                   date.
                   CC+C - Eeffects of climate change plus change in the cultivar.
                   CC+PD+C - Combination effect of climate change, changes in
                   the sowing date and change
                   in cultivar.
                                           CHINA-23

-------
Table 10.       Crop Index indicators for baseline and GCM climate change scenarios. Initial possible date
               for growing rice (Month/Day), number of possible days per year suitable for growing rice
               (days), and index of more than 10 C accumulated temperature (T).
BASE
Site
Harbin
Shenyan
Beijing
Jinan
Zhengzhou
Xuzhou
Nanjing
Chwngdu
Wuhan
Nanchang
Changsha
Fuzhou
Shaoguan
Guangzhou
Mean
(M/D)
5/14
4/24
4/10
4/8
4/5
4/9
4/5
3/22
3/29
3/30
3/26
3/9
3/13
2/26
4/1
Days
133
167
189
189
212
221
226
231
239
234
240
289
263
300
224
T
2848
3537
4216
4979
4822
4714
5076
5361
5433
5831
5731
6740
7011
7745
5289
GISS
(M/D)
4/24
4/14
3/30
3/24
3/20
3/22
3/14
2/27
3/7
3/10
3/9
2/15
2/10
2/4
3/13
Days
159
198
225
248
246
259
265
300
281
274
270
331
346
359
269
T
3696
4376
5215
6291
6185
6041
6565
7324
7634
7237
7905
8294
9465
9148
6812
GFDL
(M/D)
4/24
4/8
3/27
3/16
3/12
3/16
3/14
2/23
2/27
2/27
2/27
-
2/6
-
3/12
Days
191
204
233
261
240
258
254
281
276
300
278
365
326
365
274
T
3992
4772
5497
6469
6325
6274
6793
6761
7168
7722
7551
8839
8912
9558
6902
UKMO
(M/D)
4/9
2/24
3/9
3/3
3/8
3/8
2/25
2/9
2/26
2/13
2/18
-
2/25
2/3
2/27
Days
216
230
275
291
293
276
304
340
321
313
310
365
321
359
301
T
4815
5785
6353
7310
7197
7022
7598
8247
8038
8290
8103
9404
8470
9170
7557
                                            CHINA-24

-------
                                                                    CHINA
Figure 1.       Map of China and location of sites selected for the simulation study.

-------
cd e
e c


B
  II

 t"~-
  X
S "
o P
 II
5*.
C 9


Si
 c
.2 
H  a3
 t 3
u s fi
oo g U
"S ii ^
 ". rf
111
  f a
  r-< 00
 I
 E

-------
IMPLICATIONS OF CLIMATE CHANGE FOR JAPANESE AGRICULTURE:
EVALUATION BY SIMULATION OF RICE, WHEAT, AND MAIZE GROWTH


                            Hiroshi Seino

               National Institute of Agro-Environmental Sciences
                              Japan
                             JAPAN-1

-------
TABLE OF CONTENTS

SUMMARY
INTRODUCTION
       Aims of the Study
       Description of the Agricultural Regions and Systems
       Food Trade and Vulnerabilities
METHODS
       Baseline Climate Data and Climate Change Scenarios for the Region
       Crop Models and Management Variables
       Crop Model Validation
RESULTS AND DISCUSSION
       Sensitivity Analysis
       Crop Changes under GCM Climate Change Scenarios
       Adaptation Strategies to Climate Change
IMPLICATIONS OF THE RESULTS
REFERENCES
                                        JAPAN-2

-------
SUMMARY

        This study used climate change scenarios derived from three General Circulation Models (GCMs) to
assess the  possible impacts of climate  change on rice, maize, and wheat  production in Japan.  Higher
temperature decreased simulated crop yields in many regions under the current management system. While
the direct beneficial effects of CO2 compensated for the yield decreases in central and northern Japan, they
did not compensate for the larger yield decreases in the southwestern part of the island, especially in Kyushu.
Early planting and irrigation are possible adaptation strategies to climate change. In most cases, simulated
yields increased under climate change conditions if an earlier planting date was adopted. However, in Kyushu,
because of high temperature stress, an earlier planting date did not improve simulated yields; the introduction
of new cultivars would be required. In Hokkaido, the major upland production area of Japan, climate change
increased simulated crop yields. However, these increases were dependent on the precipitation projections of
individual climate change scenarios and on the irrigation  system practiced.
INTRODUCTION

Aims of the Study

        The purpose of this study was to estimate changes in the yields of rice, wheat, and maize crops-in
both major production areas and vulnerable regions of Japandue to climate change projected for increasing
levels of atmospheric greenhouse gases (IPCC 1990).

Description of the Agricultural Regions and Systems

        The major crops cultivated in Japan are rice (2.11 million ha), wheat (0.40,million ha), potatoes (0.2
million ha), maize (0.33 million ha), and soybeans (0.16 million ha). The major rice production region is the
Tohoku district, located in the north area of the main island (Honshu). Two sites, Niigata and Miyagi, were
selected to represent this region (Figure 1). The production in Kyushu, iii southwestern Japan, is much lower
than in the northern area, but could be potentially affected by any increase in temperature as a result of global
warming. Therefore, a site, Miyazaki, was selected in this region for the simulation study. In Japan, most rice
fields are fully irrigated  during the growing season. The soils in paddy fields are grey,  lowland soils.
        Major wheat production regions include Hokkaido (northern Japan), Tohoku (northern part of the
main island) and Kyushu (western Japan). In Hokkaido, wheat could be sown both in the spring and autumn.
In Tohoku and Kyushu, wheat is often planted in rice fields as a winter crop (i.e., sown in the autumn). The
sites selected to represent the main wheat-producing regions in this study are: Kitami  (Hokkaido), Morioka
(Tohoku), and Fukuoka  (Kyushu). While most wheat is rainfed, there are occasional, severe droughts that limit
production. The representative soils in these sites are grey lowland soils  and andosols (volcanic ash soils).
        Maize is primarily cultivated in Hokkaido (northern Japan), in central Japan, and in Kyushu. The sites
selected for this study are Obihiro (Hokkaido), Matsumoto (central Japan), and Miyakonojo (Kyushu). Most
of the maize is rainfed. The main  soil type of these regions is andosols (volcanic ash soils).

Food Trade and Vulnerabilities

        Japan is a net food importer (Appendix 1). Since Japan, depends on food imports, changes in world
crop yields and prices would directly affect domestic Japanese food prices.
                                             JAPAN-3

-------
METHODS

Baseline Climate Data and Climate Change Scenarios for the Region

        Climate data-daily maximum and minimum temperatures, precipitation, and solar radiation-from
1951 to 1988 were used as the baseline climate. Climate change scenarios were created based on the output
of three General Circulation Models (GCMs). GCMs provide the most advanced  tool for assessing  the
potential climatic consequences of increasing radiatively active trace gases. Climate models have been used to
simulate the equilibrium climate effects of an equivalent doubling of CO2. Some models have also been  run
with gradually increasing trace gases to predict a set of transient responses. However, GCMs have not yet been
validated to project changes in climate variability, such as changes in the frequencies of drought and storms,
even though these could affect crop yields significantly. Therefore, the scenarios used in this study did  not
include changes in variability.
        Climate scenarios, which are sets of climatic perturbations, are commonly used to estimate the impacts
of potential climate changes on agricultural production. In this study,  GCMs were used to derive climate
change scenarios for the crop models to estimate future changes in  yield and other agronomically important
variables. The three climate change scenarios used were the Goddard Institute for  Space Studies (GISS)
(Hansen el al. 1983); the Geophysical Fluid Dynamics Laboratory Model (GFDL) (Manabe and Wetherald
1987); and the United Kingdom Meteorological Office Model (UKMO) (Wilson and Mitchell 1987).
        The characteristics of the climate change scenarios  for Japan are summarized in Table 1. The GISS
model predicts a 2C to 3.5C rise in air temperature and a 1% to 5% increase in precipitation, except on the
southwestern islands. The precipitation increases are larger during the  summer  season (June-August). The
GFDL model predicts a 4C rise in air temperature and a 10%-20% increase in precipitation, except on the
southwestern islands. The UKMO model  predicts the highest temperature  rise-3.5C to 6C. This model
predicts a decrease in precipitation in many regions, except during the summer season in eastern Hokkaido.
The solar radiation changes under the three  GCM scenarios are less than 10%.

Crop Models and Management Variables

        Crop models. The crop models used are the Crop Environment Resource Synthesis Models: CERES-
Rice (Ritchie et al. 1986), CERES-Maize (Ritchie et al. 1989), and CERES-Wheat (Ritchie and Otter 1985).
        Soils and management. All soil data (type, initial soil water condition, etc.) were taken from soil survey
data (National Institute of Agricultural Science 1976). The organic carbon content, bulk density, and aluminum
saturation were estimated for each soil by the soil-data-retrieval program of the Decision Support System for
Agrotechnology Transfer (DSSAT). The planting date, row spacing, plant population, fertilizer, and cultivar
were taken from published data.
        Irrigation. In Japan, most rice is well irrigated, and therefore, the option for automatic irrigation in
the crop model was chosen to provide the simulated crop with a nonlimiting water supply. Maize and wheat
are grown under rainfed conditions in all locations. Irrigation was tested as a possible adaptation to climate
change for these crops.

Crop Model Validation

        The crop models were validated at nine locations by comparing experimental field data to simulated
crop yields and season lengths (Table 2). The simulated data correspond well with experimental observed data,
suggesting that the CERES models can be used for the simulation of rice, wheat, and maize growth in Japan.
                                            JAPAN-4

-------
RESULTS AND DISCUSSION

Sensitivity Analysis

        Arbitrary scenarios were created by applying changes in temperature and precipitation to the baseline
climate series. The physiological effects of 555 ppm of CO2 were considered in each of the scenarios. Tables
3, 4, and 5 show the changes in yield and season length for the three crops under the scenarios.
        Rice.  Because most rice fields are fully irrigated, the automatic irrigation option was selected in the
CERES model. Therefore, changes in precipitation did not affect rice. Rice yields decreased with a 2C to 4C
increase  in temperature, and the  length of the season from planting to maturity also decreased. With the
physiological effects of CO^ the simulated yield increased in each scenario. These increases compensated for
the negative effects of a 2C increase in temperature and resulted in  a net yield increase (6% to 10%) in
comparison with the baseline yields. However, a 4C increase resulted in decreases in rice yields, even when
the physiological effects of 555 ppm of CO2 were taken into account.
        Maize. In Japan, maize is grown under rainfed conditions. The effects of temperature increases on
maize production (Table 4) were quite  different from those on rice production.  Maize yields at Obihiro (in
Hokkaido) increased under a +2C scenario, and even under a +4C scenario there was a small increase. The
yield increases were intensified both by increases in precipitation and by the additional physiological effects
of CO2. These results suggest that current low  temperatures  limit the growth  of the variety  cultivated in
Obihiro, and that a moderate warming would have beneficial effects  on maize production in that region.
However, associated precipitation (especially during the growing season) is an important factor in determining
the crop's final yield.
        Wheat. Most wheat in Japan is cultivated under rainfed conditions. Wheat yields decreased under the
+2C and +4C  scenarios at all locations. However, a precipitation increase combined with  the higher
temperature scenarios produced increases in yields.

Crop  Changes  under GCM Climate Change Scenarios

        The climate variables under each GCM are different, and therefore, the effects of climate change on
crop production depend strongly on GCM climate change scenario.
        Rice.  The rice yields at the three sites decreased under the 2xCO2 climate change scenarios  with
climate change alone (Table 6). Under  the GISS scenario, the  physiological effects of CO2 compensated for
the negative effects of climate change alone at the three sites. However, under the GFDL scenario in Miyagi
and Niigata, and the UKMO scenario in Miyazaki, rice yields remained below the baseline yield, even with the
direct effects of CO2. At all locations, the length of the growing season was shortened by the temperature rise
associated with the climate change scenarios. The UKMO scenario produced the shortest season length of rice.
        In fully irrigated paddy fields, there are very small differences among rice yields at each site. The direct
effects of CO2 increased simulated rice yields up to 20%. In rainfed upland fields, the direct  effects of an
increased concentration  of CO2 depended on the cultivar used and the site selected.
        Maize. Table 7 shows changes in maize yields under the climate change scenarios considered. Maize
yields  in Obihiro (Hokkaido) greatly increased under the  GISS and GFDL scenario conditions, but
dramatically decreased under the  UKMO scenario due to the lack of rain during the summer season. The
yields in Matsumoto (Kanto) decreased under climate change scenarios alone,  but recovered if the direct
effects of CO2 were included in the simulation. In contrast, in Miyakonojo (Kyushu), maize was subject to high
temperature stress under all scenarios, and the yield decreases were not compensated when the the direct CO2
effects were included in the climate change simulation. Maize season length was shortened everywhere due to
the temperature increases  associated with the GCM scenarios.

                                             JAPAN-5

-------
             t The results of the simulation of wheat under the GCM climate change scenarios (Table 8) were
similar to those of maize. The severe yield decreases under the UKMO scenario were caused by precipitation
decreases in the summer season in eastern Hokkaido. The yield decreases in Morioka and Fukuoka were not
offset by the direct CO2 effects.

Adaptation  Strategies to Climate Change

        In this study, early planting dates (15 and 30 days earlier) and irrigation were analyzed as possible
adaptation strategies to counteract the negative effects of climate change. The amounts of irrigation needed
to maintain soil moisture were calculated with the water balance submodel included in the crop models. In
the case of rice, we analyzed only the effect of early planting. We did not test irrigation, because most of the
rice production currently occurs without water restrictions, and it was assumed that water would be available
for irrigation in the future.
        Table 9 shows that early planting was a very effective method of increasing rice yields under climate
change  in northern Japan, but not effective in southern Japan (e.g., Miyazaki in Kyushu). Maize yields in
Obihiro (Hokkaido) increased with an early planting date and full irrigation, while the same practices did not
recover yield decreases in the  Matsumoto region. In Miyakonojo (Kyushu), early planting improved yields;
however, irrigation did not have any effect since water is not limiting during the maize growing season at that
site (both in the current climate and in the climate change scenarios). We considered irrigation and early
planting date for spring wheat and late planting date for winter wheat. These strategies improved yields for
spring wheat in Kitami (Hokkaido), but they were not effective for winter wheat yields in Morioka and
Fukuoka.

IMPLICATIONS OF THE RESULTS

        The results of this study suggest that climate change could affect regional agricultural production in
Japan. Increases in temperature decreased crop yields in many regions in the simulations with the present
management system. While the direct beneficial effects of CO2 compensated for yield decreases in central and
northern Japan, they did not compensate for the larger yield decreases in southwestern Japan,  especially in
Kyushu. These results imply shifts in regional production patterns.
        Early planting and irrigation are possible adaptation strategies to climate change in Japan. In most
cases, simulated rice, wheat, and maize yields increased under climate change conditions if an earlier planting
date was adopted. However, in Kyushu, because  of high temperatures, an earlier planting did not improve
simulated yields, and the introduction of new cultivars better adapted to the climate change conditions would
be required. In Hokkaido, the major upland production area of Japan, climate change increased simulated crop
yields. However, the simulated yield increases are dependent on the scenario's precipitation level and on
regional irrigation systems. Potential future water availability under changed climate needs to be investigated
to improve agricultural projections.
                                             JAPAN-6

-------
 REFERENCES

 Godwin, D., U. Singh, J.T. Ritchie, and E.C. Alocilja. 1992. A user's guide to Ceres-rice. International
        Fertilizer Development Center. Muscle Shoals, AL.

 Godwin, D.C., U. Singh, R.T. Buresh, and S.K. DeDadda. 1990. Modeling of nitrogen dynamics in relation to
        rice growth and yield. In 14th International Soil Science Congress. Kyoto, Japan.

 Hansen, J., G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R Ruedy, and L. Travis. 1983. Efficient three-
        dimensional global models for climate studies. Model I and II. Monthly Weather Reviews., 3, 609-662.

 Hansen, J., I. Fung, A. Lacis, D. Rind, S. Lebedeff, R. Ruedy, and G. Russell. 1988. Global climate changes
        as forecasted by Goddard Institute  for Space Studies three-dimensional model. /. Geophy. Res.,
        93,9341-9364.

 Horie, T. 1987. The effects on rice yields in Hokkaido. In The Impact of Climatic Variations on Agriculture. Vol.
        1 Assessments in Cool Temperate and Cold Regions. M.L. Parry, T.R. Carter, and N.T. Konijn eds.809-
        825.

 IPCC. 1990. Intergovernmental Panel on Climate Change. 1990. First Assessment Report

 Jager, J., and W.W. Kellog. 1983. Anomalies in temperature and rainfall during warm arctic seasons. Climate
        Change, 5,39-60.

 Jones, C.A., and J.R. Kiniry, eds. 1986. CERES-Maize: A Simulation Model of Maize Growth and Development.
        College Station, TX: Texas A&M University Press.

 Jones, J.W., S.S. Jagtap, G. Hoogenboom, and G.Y. Tsuji. 1990. The structure and function of DSSAT. pp 1-
        14. In: Proceedings of IBSNAT Symposium: Decision Support System for Agrotechnology Transfer,
        Las Vegas, NV. 16-18 Oct. 1989. Part I: Symposium Proceedings. Dept. of Agronomy and Soil Science,
        College of Tropical Agr. and Human Resources, University of Hawaii, Honolulu, Hawaii 96822.

 Manabe, S., and R.T. Wetherald. 1987. Large-scale changes in soil wetness induced by an increase in carbon
        dioxide. /. Atmos. ScL, 44,1211-1235.

 Manabe, S., and R.T. Wetherald. 1986.  Reduction in summer soil wetness induced  by  an increase in
        atmospheric carbon dioxide. Science., 232,626-628.

 Newman, J.E. 1980. Climate change impacts on the growing season of the North American corn belt. Biomet.,
        7, Part 2, 128-142.

Ritchie, J.T., and S. Otter. 1985. Description and performance of CERES-Wheat: A user-oriented wheat yield
        model.  In:  Willis,  W.O.  ed. ARS  Wheat  Yield Project.  Washington,  DC:  U.S. Department of
        Agriculture, Agriculture Research Service, ARS-38,159-175.

Ritchie, J.T., and S. Otter. 1985. Description and performance of CERES-Wheat: A user-oriented wheat yield
        model. In W.O. Willis  (ed.). ARS Wheat Yield Project.  ARS-38. US Department of Agriculture,


                                            JAPAN-7

-------
        Agricultural Research Service. Washington, DC.

Ritchie J., U. Singh, D. Godwin, and L.  Hunt. 1989. A user's guide to Ceres Maize- v2.10. International
        Fertilizer Development Center. pp86.

Rosenzweig, C. 1985. Potential CCyinduced climate effects on North American wheat-producing regions.
        Climate Change., 7,367-389.

Uchijima, T. 1987. The effects on altitudinal shift of rice yield and cultivable area in Northern Japan. In M.L.
        Parry, T.R. Carter and N.T. Konijn eds.  The Impact of Climatic Variations on Agriculture. Vol.  1
        Assessments in Cool Temperate and Cold Regions. 797-808.

Uchijima, Z., and H. Seino. 1988. Probable effects of CCyinduced climatic change on agroclimatic resources
        and net primary productivity in Japan. Bull Natl Inst. Agro-Environ. ScL, 4,67-88.

Wigley, T.M.L., P.D. Jones, and P.M. Kelley. 1980. Scenarios for a warm, high CO2 world. Nature, 283,17-21.

Willson, C.A., and J.F.B. Mitchell. 1987. A doubled CO2 climate sensitivity experiment with a global model
        including a simple ocean. /. Geophy. Res., 92,13315-13343.
                                             JAPAN-8

-------
Table 1.       Average seasonal change in temperature and precipitation for the GCM climate change
              scenarios.
                             SPRING   SUMMER
                              (MAM)        (JJA)
FALL   WINTER   ANNUAL
(SON)       (DJF)
Temperature changes (C)
Myyazaki, Fukuoka, Miyakonojo
GISS
GFDL
UKMO
Matsumoto, Morioka,
GISS
GFDL
UKMO
Obihiro, Kitami
GISS
GFDL
UKMO
4.0
4.6
4.1
Miyagi,
3.2
4.5
7.0

4.3
4.2
5.8
3.8
3.6
5.2
Niigata
2.9
3.7
4.8

3.1
3.5
7.0
3.3
4.1
4.3

2.8
4.4
4.9

3.1
5.4
6.9
3.0
4.6
3.7

2.3
4.0
5.3

3.5
4.2
5.0
3.6
4.2
4.3

2.8
4.2
5.5

3.5
4.3
6.2
Precipitation Changes (%)
Myyazaki, Fukuoka, Miyakonojo
GISS
GFDL
UKMO
Matsumoto, Morioka,
GISS
GFDL
UKMO
Obihiro, Kitami
GISS
GFDL
UKMO
15
7
-25
Miyagi,
7
19
2

12
37
-8
-6
18
35
Niigata
-4
14
7

-4
38
-18
1
5
15

22
24
-23

9
1
9
-5
9
-8

3
7
-13

2
1
7
1
10
4

7
16
-7

5
19
-3
                                         JAPAN-9

-------
Table 2.
Simulated and observed yields and season lengths.
S. Length (d)
Crop
Rice
Rice
Rice
Maize
Maize
Maize
Wheat
Wheat
Wheat
Site
Miyagi
Niigata
Miyazaki
Obihiro
Matsumoto
Miyakonojo
Kitami
Morioka
Fukuoka
Data
(year)
75-87
80-88
80-88
76-86
68-82
70-82
67-83
70-86
72-87
DBS
145.8
134.7
118.0
136.3
125.6
114.3
103.3
288.1
196.8
SIM
143.1
136.8
119.8
134.0
127.3
113.5
102.8
285.6
191.6
Yield (t ha'1)
OBS
4.95
5.74
5.64
11.16*
16.64*
3.95
2.59
3.64
3.83
SIM
4.99
5.34
5.49
11.33*
16.67*
3.82
2.53
3.18
3.81
          * Total biomass
                                          JAPAN-10

-------
Table 3.        Sensitivity of simulated rice yields and season length to temperature and CO2. Rice was
               simulated under the automatic irrigation option of the CERES model.
                                    330 ppm CO2
555 ppm CO2
Site
Miyagi

Niigata

Miyazaki

Temp.
Change
(C)
+2
+4
+2
+4
+2
+4
Yield
Change (%)
-13
-22
-9
-16
-9
-16
S. Length
(d)
-15
-23
-12
-12
-7
-12
Yield
Change (%)
+6
-5
+10
+1
+9
+1
S. Length
(d)
-14
-22
-11
-17
-7
-12
Yield = percent changes from base; Season Length = differences from base.
                                          JAPAN-11

-------
 Table 4.        Sensitivity of simulated maize yield, season length and evapotranspiration to temperature
                (C), precipitation (%), and CO2 changes.
Site
Obihiro







Matsumoto







Miyakonojo







T
+2
+4
+2
+2
+4
+4
0
0
+2
+4
+2
+2
+4
+4
0
0
+2
+4
+2
+2
+4
+4
0
0
PP
0
0
-20
+20
-20
+20
-20
+20
0
0
-20
+20
-20
+20
-20
+20
0
0
-20
+20
-20
+20
-20
+20
330 ppm
Yield
+43
+17
+34
+48
+8
+21
-8
+3
-12
-20
-27
-3
-31
-14
-18
+ 11
-20
-32
-20
-19
-32
-31
-1
+1
C02
SL
-4
-19
-4
-4
-19
-19
0
0
-13
-18
-15
-12
-19
-18
-4
+1
-7
-13
-7
-7
-13
-13
0
0

ET
+1
-10
-3
+3
-14
-7
-4
+2
-8
-10
-13
-4
-15
-6
-7
+4
-7
-12
-7
-6
-13
-11
-1
+1
555
Yield
+63
+35
+60
+66
+30
+37
+11
+14
+8
-5
-3
+15
-12
-1
+11
+36
-15
-27
-15
-15
-27
-27
+6
+6
ppm
SL
-4
-19
-4
-4
-19
-19
0
0
-12
-18
-13
-11
-19
-18
0
+3
-7
-13
-7
-7
-13
-13
0
0
CO2
ET
-14
-22
-17
-13
-24
-20
-18
-14
-17
-19
-21
-14
-22
-16
-14
-8
-19
-24
-20
-19
-24
-23
-14
-13
xicia    percent changes  from base; SL  =  season  length differences from  base;  ET  =  percent
evapotranspiration from base.
                                            JAPAN-12

-------
 Table 5.
Sensitivity of simulated maize yield, season length and evapotranspiration to temperature
(C), precipitation (%) and CO2 changes.
                               330 ppm CO2
                                              555 ppm CO2
Site
Kitami
(SW)






Morioka
(WW)






Fukuoka
(WW)






T
+2
+4
+2
+2
+4
+4
0
0
+2
+4
+2
+2
+4
+4
0
0
+2
+4
+2
+2
+4
+4
0
0
PP
0
0
-20
+20
-20
+20
-20
+20
0
0
-20
+20
-20
+20
-20
+20
0
0
-20
+20
-20
+20
-20
+20
Yield
(%)
-17
-30
-27
-12
-38
-25
-10
+4
-7
-15
-10
-7
-18
-15
-4
+1
-14
-27
-18
-11
-31
-24
-4
+2
SL
(d)
-8
-14
-8
-8
-14
-14
0
0
-4
-7
-4
-4
-7
-7
0
0
-3
-5
-3
-3
-5
-5
0
0
ET
(%)
-5
-9
-12
+1
-17
-4
-7
+5
+2
+4
0
+4
+1
+5
-2
+1
-2
-2
-5
0
-6
0
-3
+2
Yield
(%)
+10
-7
0
+14
-14
-1
+22
+31
+9
0
+8
+9
-1
0
+16
+18
+3
-13
-1
+5
-17
-10
+15
+22
SL
(d)
-8
-14
-8
-8
-14
-14
0
0
-4
-7
-4
-4
-7
-7
0
+3
-3
-5
-3
-3
-5
-5
0
0
ET
(%)
-9
-13
-16
-5
-19
-8
-11
-2
-5
-2
-7
-5
-5
-1
-10
-9
-8
-7
-11
-7
-11
-6
-9
-5
Yield  =  percent  changes from  base;  SL  =  season  length  differences  from base; ET  =  percent
evapotranspiration from base.
                                           JAPAN-13

-------
Table 6.        Changes in rice yield, season length and total season precipitation under GCM climate change
               scenarios.
Site
Miyagi


Niigata


Miyazaki



GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
330 ppm
Yield
-12
-27
-17
-7
-21
-11
-12
-17
-19
CO2
SL
(d)
-19
-22
-25
-15
-18
-21
-12
-12
-14

PP
-19
-6
-24
-16
0
-21
+5
-3
-5
555
Yield
+7
-11
+1
+12
-5
+6
+6
0
-2
ppm
SL
(d)
-18
-22
-25
-15
-18
-21
-12
-12
-14
CO2
PP
-19
-5
-24
-16
+1
-21
+5
-3
-5
Yield = percent changes from base; SL= season length differences from base; PP= recipitation percent from
base.
                                            JAPAN-14

-------
Table 7.        Change in maize yield, season length, evapotranspiration and total season precipitation under
               GCM climate change scenarios.
                         330 ppm CO2
555 ppm CO2

Site
Obihiro


Matsumoto


Miyakonojo



GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
Yield
+24
+25
-32
-19
-14
-15
-28
-34
-31
SL
(d)
-15
-16
-29
-17
-17
-20
-12
-13
-15
PP
-16
+9
-45
-19
+ 18
-17
+6
-4
-11
ET
-2
-9
-16
-7
-15
-7
-11
-16
-8
Yield
+51
+38
+5
+3
-5
-2
-23
-31
-27
SL
(d)
-14
-16
-29
-16
-17
-21
-12
-13
-15
PP
-16
+9
-45
-18
+18
-17
+6
-4
-11
ET
-15
-23
-22
-16
-26
-16
-23
-28
-20
Yield = percent changes from base; SL= season length differences from base; ET= percent cahnge of total
season evapotranspiration PP= recipitation percent from base.
                                            JAPAN-15

-------
Table 8.       Change in wheat yield, season length, evapotranspiration and total season precipitation under
               GCM climate change scenarios.
                         330 ppm CO2
555 ppm CO2

Site
Kitami


Morioka


Fukuoka

-.,...

GISS
GFDL
UKMO
GISS
GFDL
UKMO
GISS
GFDL
UKMO
Yield
-21
-17
-59
-16
-20
-31
-24
-33
-40
SL
(d)
-11
-12
-21
-7
-7
-12
-8
-5
-5
PP
-18
+8
-43
-1
+2
-23
-8
-6
-28
ET
-7
-5
-22
+5
+3
+6
-4
0
-2
Yield
+6
+8
-41
-1
-7
-18
-8
-20
-27
SL
(d)
-11
-12
-21
-7
-7
-12
-8
-5
-5
PP
-18
+8
-45
-1
+2
-23
-8
-6
-28
ET
-10
-11
-23
-2
-3
0
-9
-4
-7
    1= percent changes from base; SL= season length differences from base; ET= percent cahnge of total
season evapotranspiration PP= recipitation percent from base.
                                           JAPAN-16

-------
Table 9.
Evaluation of potential adaptation strategies: changed planting dates and levels of irrigation.


Miyagi

Niigata

Miyazaki


Obihiro



Matsumoto



Miyakonojo




Kitami



Morioka



Fukuoka



GCM Rainfed Planting
RICE
GFDL irrigated 15 d. early
irrigated 30 d. early
GFDL irrigated 15 d. early
irrigated 30 d. early
UKMO irrigated 15 d. early
irrigated 30 d. early
MAIZE
UKMO rainfed 15 d. early
rainfed 30 d. early
irrigated 15 d. early
irrigated 30 d. early
GFDL rainfed 15 d. early
rainfed 30 d. early
irrigated 15 d. early
irrigated 30 d. early
GFDL rainfed 15 d. early
rainfed 30 d. early
irrigated 15 d. early
irrigated 30 d. early
WHEAT
UKMO rainfed 15 d. early
rainfed 30 d. early
irrigated 15 d. early
irrigated 30 d. early
UKMO rainfed 15 d. late
rainfed 30 d. late
irrigated 15 d. late
irrigated 30 d. late
UKMO rainfed 15 d. late
rainfed 30 d. late
irrigated 15 d. late
irrigated 30 d. late
Base
Yield

4.30

5.44

5.56


5.81



4.87



2.64




1.60



2.69
2.69
2.69
2.69
2.94
2.94
2.94
2.94
Scenario Yield
with Adaptation

4.62
4.90
5.72
5.99
5.60
5.68

5.82
6.02
8.37
8.64
4.61
4.35
4.64
4.36
3.05
3.38
3.05
3.38

2.02
2.22
3.14
3.36
2.69
2.68
2.71
2.70
2.88
2.80
2.88
2.80
amoun










114.68
116.34


38.39
38.32


4.18
6.24



169.29
167.05


74.49
74.16


2.24
2.24
                                              JAPAN-17

-------
Appendix 1. Food
Commodity
Cereals
Rice
Wheat
Barley
Naked Barley
Sweet corn
Sweet potatoes
Potatoes
Starches
Pulses
Vegetables
Fruits
Meat
Hen eggs
Milk&milk products
Fish&Shellfish
balance in Japan in
Domestic
Production
11,870
10,627
864
326
27
1
1,423
3,955
2,357
469
16,598
5,974
3,607
2,394
7,427
11,800
1987 (x 1000 t).
Imports
28,187
39
5,133
1,988
0
16,602
0
323
119
4,797
1,114
2,260
1,171
36
1,767
3,299
Exports
0
0
0
0
0
0
0
0
0
0
4
48
4
0
0
1,583
JAPAN-18

-------
                                               JAPAN
Figure 1.      Map of Japan and location of the sites selected for the study.

-------

-------
SECTION 7: AUSTRALIA

-------

-------
POSSIBLE EFFECTS OF GLOBAL CLIMATE CHANGE
ON WHEAT AND RICE PRODUCTION IN AUSTRALIA
                     Brian D. Baer

            Department of Crop and Soil Science
              Michigan State University, USA
                    Wayne S. Meyer

               Division of Water Resources
                   Griffith Laboratory
                    CSIRO, Australia
                     David Erskine

               Division of Water Resources
                   Griffith Laboratory
                    CSIRO, Australia
                    AUSTRALIA-1

-------
TABLE OF CONTENTS

SUMMARY
INTRODUCTION
       Aims
       Sites and Current Climate
METHODS
       Climate Change Scenarios
       Crop Models and Management Practices
       Calibration and Validation of the Crop Models
       Simulations
RESULTS
       Crop Yield and Irrigation Demand
       Sensitivity Analysis
       Adaptation to Climate Change
DISCUSSION
       Limitations of the Crop Models
       Impact of Climate Change on Crops
REFERENCES
                                    AUSTRALIA-2

-------
SUMMARY

        This  simulation study uses crop models  and climate  change scenarios generated from General
Circulation Models (GCMs) to determine the possible impacts of climate change on rice and wheat production
in Australia. The beneficial direct effects of CO2 on crop yield and water use are taken into account in the
simulations. In most sites, dryland wheat yields increase when the scenario projects a rainfall increase.
However, in the scenarios with the largest GCM temperature increases, yields generally decreased due to a
shortening of the crop growing season. Irrigated wheat yields also decreased due to the temperature increases.
Paddy rice yields decreased slightly under climate change conditions.
        Simulation experiments showed that the most successful adaptation strategy to climate change was
changing the variety of rice to a more  tropical one. Also, adjusting the sowing dates for dryland wheat to
obtain maximum water availability was helpful.
INTRODUCTION

Aims

        The purpose of this study is to estimate the possible impacts of climate change on yield and irrigation
water demand of wheat and rice in different sites in Australia using simulation crop growth models and climate
change scenarios generated from GCMs.

Sites and Current Climate

        Wheat is grown in  the semi-arid regions of southern and eastern Australia, while rice is  mainly
confined to inland southern New South Wales (NSW) (Figure 1). The main constraint on the extent of these
agricultural regions is water availability. In the southern part of the country, the rainfall pattern has a winter
maximum; in the middle regions of the eastern wheat belt, rainfall is evenly distributed throughout the year.
Winter rainfall is stored in the soil and is available to the wheat crop during the following winter until early
summer. In the north of Australia, the dominance of summer rainfall and the limited areas of suitable soils
restrict agriculture.  Winter temperatures in the mainland are mild and do not present any constraint to crop
growth.
        Five sites were chosen to represent the major wheat regions of Australia (Figure 1). Wheat is grown
under rainfed conditions in most areas; only in the Murrumbidgee Irrigation Area, represented by Griffith in
NSW, is wheat grown under irrigated conditions. Griffith is also central to the major rice-producing  area of
southern New South Wales. The country's aridity and strong commitment of available water supplies to diverse
users  make it  unlikely that other grain-growing regions of the country will be able to acquire water for
irrigation in the future.

METHODS

Climate Change Scenarios

        A series of climate data available for the period 1951-80 was obtained for each site (Wongan Hills
 1966-80; Roseworthy 1957-80; Horsham 1957-59, 1961-80; Griffith 1951-80; Narrabi 1962, 1964-66, 1968-69,
 1971-80). Daily maximum temperature, minimum temperature, and rainfall were available for all sites. Missing


                                          AUSTRALIA-3

-------
 values were estimated based on the values before and after that day. Daily solar irradiance data was available
 for the Griffith site only. For the rest of the sites, we generated values based on maximum possible irradiation
 modified with observed monthly average hours of sunlight (Richardson and Wright 1984).
         Climate change scenarios for each site were generated from three GCMs: the Goddard Institute for
 Space Studies Model (GISS) (Hansen et al. 1983), the Geophysical Fluid Dynamics Laboratory Model (GFDL)
 (Manabe and Wetherald 1987), and the United Kingdom Meteorological Office Model (UKMO) (Wilson and
 Mitchell 1987). The scenarios for each site were created based on the GCM output of climate variables applied
 to the daily  observed climate. This method uses the  difference between lxCO2 and 2xCO2 monthly GCM
 temperatures and the ratio between 2xCO2 and lxCO2 monthly GCM precipitation amounts. (lxCO2 refers
 to current climate conditions, and 2xCO2 refers to the climate that would occur with a doubling of greenhouse
 gases.) Table 1 shows the seasonal changes projected under the climate change scenarios. All GCMs project
 temperature  increases for all seasons. The GISS and GFDL models predict an average annual  temperature
 rise of around 4.5C, and the UKMO model predicts temperature increases of approximately 5.7C. The GISS
 and UKMO models predict a slight increase in annual rainfall, while GFDL predicts a slight decrease. In most
 sites and seasons, the three GCMs predict a slight increase in solar radiation.

 Crop Models and  Management Practices

        The CERES-Rice (Godwin et al. 1992) and the CERES-Wheat (Godwin et al. 1989) models were used
 for the crop simulations. In addition to the climate data, the crop models require a well-defined set of inputs
 to simulate actual crop  conditions. The inputs include soil parameters and management practices. The soil
 parameters include  data on texture, water-holding capacity, and nitrogen status of the soil. A representative
 soil was determined and defined for each of the five sites. In general, the representative soils chosen were clay
 loam soils, with the exception  of a heavy clay soil for the rice growing area, a sandy loam soil for western
 Australia, and a cracking clay soil for Narrabri.  Information on soils in the wheat-growing areas was taken
 from  McGarity (1975). The crop management practices include sowing dates, plant population, irrigation,
 nitrogen applications, and plant variety. These variables were determined according to current practices.
        Wheat  is grown in the winter at all sites to take advantage of the low evaporation and to maximize
 available water. We simulated nonirrigated wheat with low  inputs  of nitrogen  and relatively low  plant
 populations at all sites. In addition, we simulated irrigated wheat and paddy rice in Griffith. Table 2 shows the
 conditions that  were used for the crop simulations.

 Calibration and  Validation of the Crop Models

        We determined the  genetic coefficients  that define a  variety  in the CERES  models by comparing
 model runs to actual field data and calibrating the coefficients by the method of trial and error. After defining
 the genetic coefficients, the models were validated for  use in the selected  sites.
        General data  from a wide range of dryland wheat experiments in Australia were obtained from the
 data base of Rimmington et al (1987) and were used to validate the wheat model for Roseworthy. For eastern
 Australia, validation data were compiled from the data sets of MacKenzie et al. (1985) and supplemented with
 more specific information from Angus et al. (1980)  for an area similar  to Narrabri.  For Griffith,  data for the
validation were taken from Mason and Fischer (1986); for Horsham, from  Rimmington et al. (1987); and for
 Wongan Hills,  from  Anderson and Smith  (1990).  Validation  data for rice were obtained from the
comprehensive experiments of Muirhead et al (1989) and Humphreys et al. (1987). For irrigated wheat, local
data published by Meyer et al (1985) were used for validation.

Simulations

                                         AUSTRALIA-4

-------
       The simulations include the following sets of scenarios: (a) baseline climate (in all sites); (b) GCM
climate change scenarios alone (in all sites); (c) GCM climate change scenarios including the direct effects of
555 ppm CO2 in the crop growth simulation (in all sites) (Acock and Allen 1985); (d) sensitivity analysis,
where base daily temperature and precipitation were modified by fixed amounts (in Roseworthy and Griffith);
and (e) scenarios that incorporated possible changes in management conditions to analyze adaptive strategies
to climate changes. For dryland wheat at Horsham and Roseworthy we changed the current planting dates;
for irrigated rice, we replaced the current cold-resistent variety, Calrose, with a more tropical variety, IR-36.

RESULTS

Crop  Yield and Irrigation Demand

       Table 3 shows the yields  (and standard deviations) and irrigation amounts simulated by the CERES
models for dryland wheat and irrigated wheat and rice for the baseline climate and the GCM climate change
scenarios. The direct effects of CO2, as well as the climate effects alone, are shown.
       Crop Yield. Climate change scenarios had varied effects on wheat yields at each of the sites (Table 3).
While  the UKMO model showed a decrease in yields in all cases but one, the other two models showed no
general trend. From the climate effects alone, the yield changes ranged from a 15% increase in Horsham under
the GFDL model to a 45% decrease in Wongan Hills under the GISS scenario. The direct effects  of CO2
increased the yield  under base climate from  25% to 39%. The beneficial direct CO2 effects on yield fully
compensate for the yield losses under the climate change scenarios in most cases.
       Table 3 also shows the effects of climate change scenarios on irrigated wheat at Griffith. Simulated
yields decreased due to increased temperatures and a shortening of the growing season. The greatest yield
losses  corresponded to  the scenario that projects  the largest  increases in temperature.  In Griffith, the
temperature increase was lowest  under the GISS scenario and largest under the UKMO scenario. Simulated
wheat  yields decreased 17% and 45% under  the GISS and UKMO scenarios, respectively. With the direct
effects of CO2,  the corresponding yield decreases were only 2% and 29%.
        Flooded rice (Table 3) was much less susceptible to temperature increases than was wheat. Under
climate change alone, yields decreased from 9% (GISS) to 16% (UKMO). The direct CO2 effects compensated
for the rice yield decreases under the climate change scenarios alone.
       Irrigation Demand. Two  irrigated crops were simulated at Griffith. The irrigation needed for rice
increased or, in the case of GISS, did not significantly change (Table 3). However, the irrigation demand for
wheat  showed an opposite trend under the climate scenarios, showing a decrease of more than 50% in the
UKMO scenario  due to a shorter growing period. In all cases, irrigation demands were smaller with the
physiological CO2 effects because of increased stomatal resistance.

Sensitivity Analysis

        Table 4 shows the results of the sensitivity analysis for dryland wheat in Roseworthy. Changes in yields
due to temperatures are relatively small in comparison with changes due to precipitation. There is an increase
in yields with a 2C temperature increase, but a  decrease from the baseline yields with a 4C temperature
increase.

Adaptation to Climate Change

        Table 5 shows the effect of changing planting dates on dryland wheat yields  under climate change
scenarios at two sites. Planting dates of 15 and 30 days before and after May 30 (the most common planting

                                          AUSTRALIA-5

-------
 date) were chosen. Yields were higher with an earlier sowing date and they decreased with later planting dates.
 Table 5 also shows the results of the flooded rice model runs at Griffith using variety IR-36. According to the
 CERES model, this variety of rice produced higher average yields than Calrose (the current cultivar), even in
 the current climate. However, the standard deviation of the current simulated yield increased significantly.
 Under the climate change scenarios, yields increased substantially when using the variety IR-36, but without
 an increase in standard deviation.

 DISCUSSION

 Limitations of the Crop Models

        For this  study the simulations lacked two components that might have affected the outcome. First,
 the soil moisture was reset at the beginning of every season rather than being simulated for the entire year.
 Since there is not enough water to supply the demands of the dryland wheat from rainfall during the growing
 season, the water stored in the soil is critical to the development of the plants. The stored water comes from
 rain that fell  during earlier months when  there were few or no plants growing. In this  study we partially
 compensated for  this by starting the simulation two months before the sowing date. However, a full-year water
 balance would have given more accurate predictions of water in the soil during the growing season. This
 procedure might  possibly underestimate the impacts of climate change on soil moisture.
        Second, the rice model is not sensitive to cool temperatures. The rice model was mainly developed
 with data from tropical areas where the temperatures are warm throughout the growing season. However, in
 less tropical areas, such as southern Australia, there can be fairly cold nights  during critical stages of the rice
 growing cycle. These cold temperatures can induce sterility in the plants and reduce the yield substantially. The
 CERES-Rice model does not account for this problem, and therefore, it probably overestimates yields in years
 with cold nights.  This effect should be tested in future climate change impact studies.

 Impact of Climate Change on Crops

        Dryland wheat yields in Australia increased when the scenario projected an increase in  the amount
 of  rainfall (compared to the current  rainfall).  The sensitivity analysis  at  Roseworthy  clearly shows the
 importance of this variable. In most Australian regions, the water stored in the soil prior to sowing provides
 a large percentage of the water needed by the crop, so the extra rainfall in the summer months contributes
 directly to increases  in yields. In contrast, at  Wongan Hills, under the GFDL scenario  yields decrease
 substantially (about 30%) with the effects  of climate change alone, due in part to projected decreases in
 rainfall.
        However, in the situations where the GCMs predict large temperature increases, the effect of rainfall
 increases can be overcome by the effects of higher temperatures. In the cases of irrigated wheat and flooded
 rice crops this high temperature effect is the only one to which the models are sensitive. Warmer temperatures
 during  the growing season  decrease the length of time that the plant takes to reach maturity. This is
 particularly  critical during the grain-filling period, the duration of which  is determined by accumulated
 temperature. An  increase in temperature stimulates the plants to mature more quickly, but with less grain
yield. Figure 2 shows  the effect of temperature on the length of the growing season on simulated irrigated
wheat at Griffith.
        An increase in atmospheric CO2 results in larger yields due to greater net photosynthesis and better
 use of water by the plant due to increased stomatal resistance. In the plants that  are not suffering from
drought stress these effects are particularly pronounced.
        The rainfall changes associated with the climate change scenarios influenced the amount of irrigation

                                          AUSTRALIA-6

-------
required for the crop. For irrigated wheat, the decrease in water demand is related to the shorter growing
season and smaller biomass accumulation. The plants that do poorly require less water, and therefore, less
irrigation. Rice, however, showed less of a decrease in yield and therefore, the water demand did not decrease
as much as in the case of wheat.
        A change in management practices may compensate for the negative impact of climate change on
yields, as simulated  in this study. In the case of dryland wheat, it is critical to grow the crop when there is
water available, and therefore, sowing dates may be adjusted to the most favorable moisture regime under the
climate change conditions. Changing the variety of rice may be a successful strategy for adaptation to climate
change.
        Australia's farmers produce more than enough wheat and rice for internal consumption, so that none
of the predicted changes would threaten the food security of the country. However, farm products are major
export commodities, and yield variations could have large effects on farm export earnings, leading to significant
impacts on the  national economy.
                                          AUSTRALIA-?

-------
 REFERENCES

 Acock, B., and L.H. Allen, Jr. 1985. Crop responses to elevated carbon dioxide concentrations. In B.R. Strain
        and J.D. Cure (eds.). Direct Effects of Increasing Carbon Dioxide on Vegetation. DOE/ER-0238. U.S.
        Department of Energy. Washington, D.C. pp. 53-97.

 Anderson, W.K., and W.R. Smith. 1990. Field advantage of two semi-dwarf compared to two tall wheats
        depends on sowing time. Aust. J. Agric. Res. 41, 811-826.

 Angus, J.F., H.A. Nix, J.S. Russell, and J.E. Krruizinga. 1980. Water  use, growth and yield of wheat in a
        subtropical environment. Aust. J. Agric. Res. 31, 873-886.

 Godwin, D., J. Ritchie, U. Singh, and L. Hunt. 1989. A User's Guide to CERES Wheat-V2.10. International
        Fertilizer Development Center.  Muscle Shoals. AL

 Godwin, D., U.  Singh, J.T. Ritchie, and E.C. Alocilja. 1992. A  User's  Guide to CERES-Rice. International
        Fertilizer Development Center.  Muscle Shoals. AL

 Hansen, J., G. Russell, D. Rind, P. Stone, A. Lacis, S. Lebedeff, R. Ruedy, and  L. Travis. 1983. Efficient
        Three-Dimensional  Global  Models for Climate Studies: Models I and  II. April Monthly Weather
        Review,  Vol III, No. 4: 609-662.

 Humphreys, K, W.A. Muirheas, F.M. Melhuish, and R.J.G. White. 1987. Effects of time of urea application
        on combine sown calrose rice in south east Australia. Aust. J. Agric. Res. 38, 101-112.

 McGarity, J.W. 1975. Soils of the Australian wheat-growing areas. In:  Australian Field Corps 1: Wheat and
        Other Temperate Cereals. Eds. Lazenby, A. and Matheson, E.M. P 227-255. Angus and Robertson
        Sydney.

 MacKenzie, D.H., M.E. Robertson, M.F. Hutchinton,  M.P. O'Connor,  J.P. McMahon, and H.A. Nix. 1985.
        Matched sets of crop, soil, and weather data from wheat lands in eastern Australia. Report CSIRO,
        Division of Water Resources, Camberra.

 Manabe, S., and R. Wetherald.  1987.  Large-scale changes of soil wetness  induced by  an increase in
        atmospheric carbon dioxide. Journal of Atmospheric Science 44:1601-1613.

 Mason, I.E., and R.A. Fischer. 1986. Tillage practices  and the growth and yield of wheat in southern New
        South Wales: Lockhart, in a 450 mm rainfall region. Aust. J.  Exp. Agric. 26, 457-468.

 Meyer, W.S., F.X. Dunin, R.C.G.  Smith, G.S.G. Shell, and N.S. White. 1987. Characterizing water use by
        irrigated wheat at Griffith, New South Wales. Aust. J. Soil Res.  25, 499-515.

Muirhead, W.A., J. Blackwell, E. Humphreys, and R.J.G. White. 1989. The growth and nitrogen economy of
        rice under sprinkler and flood irrigation in South East Australia. Irrig. ScL 10, 183-199.

Richardson, C.W., and D.A. Wright. 1984. WGEN: A Model for Generating Daify Weather Variables. ARS-8.
        U.S. Department of Agriculture, Agricultural Research Service. Washington, DC. 83 pp.


                                         AUSTRALIA-8

-------
Rimmington, G.M., T.A. McMahon, and D.J. Connor. 1987. The Australian wheat field trial database (interim
       report). Agric. Engin. Misc. Report. University of Melbourne.

Wilson, C.A., and J.F.B. Mitchell. 1987. A doubled CO2 Climate Sensitivity Experiment with a Global Model
       Including a Simple Ocean. Journal of Geophysical Research, 92: 13315-13343.
                                         AUSTRALIA-9

-------
Table 1.        Temperature differences and precipitation and solar radiation ratios between lxCO2 and
               2xCO2 climate change scenarios at selected sites.
                       Temp. Diff. (C)
Precip. Ratio
Solar Rad. Ratio
Site/Season
Wongan Hill
Summer
Autumn
Winter
Spring
Roscworthy
Summer
Autumn
Winter
Spring
Horsham
Summer
Autumn
Winter
Spring
Griffith
Summer
Autumn
Winter
Spring
Narrabri
Summer
Autumn
Winter
Spring
OISS

3.68
4.93
5.61
5.67

4.87
437
4.34
4.27

4.87
4.37
4.34
4.27

434
3.93
4.00
3.77

3.34
4.83
5.07
4.91
GFDL

4.63
4.32
3.77
4.84

5.27
4.53
4.09
4.00

4.86
4.46
4.46
4.29

5.27
4.53
4.09
4.00

2.36
3.82
4.12
4.41
UKMO

4.37
4.51
4.60
4.47

6.79
7.37
5.62
5.84

5.13
5.53
5.11
5.04

6.03
6.55
6.61
5.91

3.35
7.07
6.32
6.26
GISS

1.57
1.38
0.80
1.69

1.50
0.91
1.22
1.14

1.50
0.91
1.22
1.14

1.13
0.96
1.03
1.25

0.75
1.25
0.89
0.98
GFDL

1.01
0.96
0.77
0.50

0.71
1.01
1.06
0.98

0.92
0.87
1.12
0.79

0.71
1.01
1.06
0.98

0.75
0.78
0.95
1.08
UKMO

1.00
1.08
1.16
1.06

1.21
0.86
1.17
1.09

1.12
1.23
1.25
1.12

1.29
1.05
1.04
1.19

0.99
0.79
1.16
1.18
GISS

1.00
1.00
1.02
1.00

1.01
1.03
1.06
1.00

1.01
1.03
1.06
1.00

1.02
1.05
1.09
1.01

0.66
1.01
1.05
1.01
GFDL

1.01
1.02
1.07
1.07

1.02
0.97
1.04
1.07

1.05
1.02
0.95
1.09

1.02
0.97
1.04
1.07

0.66
1.09
1.05
1.05
UKMO

1.02
1.08
1.01
1.02

1.06 :
1.12
1.03
1.11

1.09
1.08
1.09 :
1.10

1.03
1.07
1.05 ;
1.04

0.67
1.10
1.04
1.04
                                        AUSTRALIA-10

-------
Table 2.
Management variables used for the CERES models to simulate crop growth.
 Site
     Crop/Variety
                                               Soil
                    Sowing    Plant Population
                    Dates      (plants m"2)
 Wongan Hills
 Roseworthy
 Horsham
 Griffith
 Griffith
 Griffith
 Narrabri
     Dryland wheat/Gamenya
     Dryland wheat/Egret
     Dryland wheat/Egret
     Dryland wheat/Egret
     Irrigated wheat/Egret
     Flooded rice/Calrose
     Dryland wheat/Egret
Deep sandy loam     1 May     150
Red brown earth     30 May    150
Red brown earth     30 May    150
Red brown earth     15 May    150
Red brown earth     15 May    180
Gray cracking        17 Oct     128
Vertisol             25 Jun     150
                                         AUSTRALIA-11

-------
Table 3.        Effects of climate change on simulated wheat and rice; (a) yield and (b) irrigation water.

  (a) Yield (t ha'1)
                                     Climate Scenario Alone
 Climate Scenario with
Physiological CO2 Effects
Site



Wonhan Hills Yield

Roscworthy

Horsham

Griffith

Narrabri


Griffith


Griffith

(b) Irrigation

Griffith


Griffith

SD
Yield
SD
Yield
SD
Yield
SD
Yield
SD

Yield
SD

Yield
SD
water (mm)

mm
SD

mm
SD
BASE

3.24
1.22
2.48
1.17
2.64
1.08
2.73
1.23
3.55
1.96

6.65
0.55

9.16
0.42


240
83

693
84
GISS

2.19
1.10
2.57
1.24
2.73
1.08
2.71
1.48
1.96
1.43

5.46
0.53

8.30
0.77


177
63

689
105
GFDL

2.27
1.08
2.45
1.42
2.60
1.23
2.62
1.44
2.47
1.73

5.08
0.49

7.80
0.58


169
65

761
93
UKMO
Dryland Wheat
3.14
0.86
1.94
1.15
3.00
1.05
1.80
1.36
2.27
1.40
Irrigated Wheat
3.79
0.45
Flooded Rice
7.66
0.51

Irrigated Wheat
118
48
Flooded Rice
747
110
BASE

4.08
0.90
3.40
1.01
3.28
0.90
3.44
1.15
4.74
2.15

7.22
0.61

10.10
0.54


202
80

680
87
GISS

3.04
1.34
3.26
1.22
3.40
0.87
3.38
1.48
2.82
1.74

6.41
0.56

9.12
0.77


136
55

675
104
GFDL

3.18
1.12
3.11
1.40
3.32
1.14
3.27
1.39
3.43
2.10

5.99
0.53

8.60
0.82


133
50

749
94
UKMO

3.85
0.83
2.64
1.39
3.59
0.94
2.38
1.69
3.13
1.63

4.64
0.52

8.54
0.70


98
50

731
112
                                           AUSTRALIA-12

-------
Table 4.       Sensitivity analysis of the CERES-Wheat model to changes in temperature, precipitation and
              CO2 levels. Dryland wheat at Roseworthy.
                                        330 ppm CO2
555 ppm CO2
Changes in
Precip. Temp.
(%) (C)
0% 0
2
4
20% 0
2
4
-20% 0
2
4
Yield
(t ha1)
2.55
2.82
2.51
3,29
3,57
3,05
1.59
1.86
1.66
SD
(t ha1)
1.24 .
1.32
1.49
1.26
1.11
1.27
0.75
1.03
1.12
Yield
(t ha-1)
3.45
3.69
3.18
4.00
4.19
3.75
2.69
2.85
2.51
SD
(t ha'1)
1.10
1.11
1.45
0.85
0.88
1.15
1.08
1.23
1.45
                                       AUSTRALIA-13

-------
Table 5.
Adaptation strategies for climate change scenarios; effects of change in planting date on
simulated wheat yield at Horsham and Roseworthy (a); effects of change in cultivar on
simulated irrigated rice yield (b) and irrigation water (c) at Griffith.
 (a) Dryland Wheat Yield (t ha'1)
                                        Climate Scenario Alone
                                                                 Climate Scenario with
                                                               Physiological CO2 Effects

Horsham




Roseworthy




(b) Irrigated
Griffith


30 Apr.
15 May
30 May'
15 June
30 June
30 Apr.
15 May
30 May*
15 June
30 June
Rice Yield (t ha"1) 
Calrose*
IR-36
BASE
3.94
3.19
2.64
2.29
1.80
3.72
3.15
2.48
1.91
1.37

9.16
9.30
GISS
2.24
2.77
2.73
2.66
1.77
1.60
2.40
2.57
2.53
1.79

8.30
10.17
GFDL
1.64
2.37
2.60
2.45
1.49
1.70
2.38
2.45
2.11
1.18

7.80
9.47
UKMO
2.09
2.92
3.00
2.75
1.72
0.80
1.39
1.94
2.09
1.32

7.66
9.63
GISS
3.11
3.53
3.40
3.20
2.37
2.22
3.19
3.26
3.22
2.39

9.12
12.05
GFDL
2.36
3.13
3.32
3.21
2.12
2.32
3.09
3.11
2.90
1.84

8.60
11.19
UKMO
3.04
3.78
3.59
3.28
2.29
1.16
2.03
2.64
2.85
1.97

8.54
11.40
(c) Water Used for Irrigation (mm)
Griffith

Calrose*
IR-36
693
825
689
728
761
783
747
769
675
707
749
760
731
741
                                           AUSTRALIA-14

-------
                                         NSW
                                  Roseworthy
Figure 1.      Australian study sites and major grain-growing regions.

-------
y







7


6


5


4


3


2
 	  	 __r_
/
'V
X* /
T T/ /
T v^.//
 T J> x v
 /o V /
 / 9 V
"h^ /j^TV^W
"^ ,/ W^ T
"> An xX *" V T V
m Ai D / jfv
&E *"/ / v^ v
  Bjr3^-i X  T v
^f^vx * ' 9
*t Jfjfeb**'
'*5RA . O*
:x/ if% 
* xitl -#^A 
/V** * 6 D A
/ \ ^8,/r ^
/ * / A n
- /x * /^o
/ -w%
X / v
/ o o *
/ <>o
~xX
/
/* 
1 1 1 1












330 555

v T Base

D  GISS

A A GFDL
0 * UKMO





100 120 140 160 180 200
          Season length (days)
Figure 2.      Effects of season length on irrigated wheat yield at Griffith, NSW.

-------