68774 Impacts of Climate Change on Brazilian Agriculture Impacts of Climate Change on Brazilian Agriculture [P118037] Dr. Eduardo Assad, EMBRAPA1, Brazil Prof. Hilton S. Pinto, UNICAMP2, Brazil Dr. Andre Nassar, ICONE3, Brazil Dr. Leila Harfuch, ICONE, Brazil Dr. Saulo Freitas, I-PE4, Brazil Barbara Farinelli, World Bank, Brazil Mark Lundell, World Bank, Brazil Erick C.M. Fernandes, Task Team Leader, LCSAR, World Bank, USA 1 The Brazilian Enterprise for Agricultural Research 2 University of Campinas, Sao Paulo 3 The Brazilian Institute for International Trade Negotiations 4 The Brazilian Institute for Space Research © 2013 International Bank for Reconstruction and Development / International Development Association or The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contri- butions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorse- ment or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2422; e-mail: pubrights@ worldbank.org. Contents Foreword 7 Executive Summary 13 Introduction 17 The Evolution of the Farming Sector in Brazil and Implications to 2030 20 Threats to Brazilian Agriculture from climate variability and eventual climate change 23 BOX 1 - Climate Change and Agriculture in Latin America, 2020-2050 25 Scope of this study 29 Climate Change and Agricultural Impact Projections in Brazil to 2030 and Beyond 31 Refining climate change impact projections via global, regional and local scale modeling 32 Regional and Brazilian Approaches to Selecting and Using Climate Models 32 Emissions Scenarios Used for this Study 32 Climate Models Used for this Study 33 Model Testing, Climate Projections and Model Calibrations 35 The Agro Climatic Risk and Vulnerability Zoning Model 38 The Water Needs Index (ISNA) of the Agro Climatic Zoning Approach 40 Soil Classification and Map of the Agro Zoning Approach. 41 Identifying Cropping Areas that are less Vulnerable to Climate Change Impacts 42 Projected Climate Change Impacts on Area of Crop Suitability to 2020 and 2030 45 Maps of projected climate change impacts on major grain crops, sugarcane, and pastures in Brazil to 2020 and 2030 47 Projected climate change impacts on commodity supply and demand and land use dynamics 57 Methodology for the economic simulations of climate change scenarios and projected agricultural impacts 57 The Brazilian Land Use Model – BLUM 58 The supply and demand projections 58 The Components of Land Use Dynamics 60 Accounting for Land Use Dynamics in BLUM 61 EMBRAPA’s Agricultural Impact Projections as Inputs to BLUM 64 Simulation Results from the Brazilian Land Use Model (BLUM) 68 Land Use and Production 69 Domestic Consumption, Prices and International Trade 73 The Projected Real Prices of Commodities to 2020 and 2030 as Impacted by Climate Change 74 Projected Production Value of Agriculture as Impacted by Climate Change 2020 and 2030 75 Conclusions 77 Bibliography 83 List of Figures Figure 1. Grain production and area increase in Brazil from 1991 to 2010. 18 Figure 2. World and Brazil Agricultural Production: Expansion from 1990 to 2009 (1990=100) 19 Figure 3. Brazil’s and World Meat and Milk Production: Expansion from 1990 to 2009 (1990=100) 19 Figure 4. Brazilian Agricultural Sector: Total Production and Use of Labor, Land, and Capital and Total Factor Productivity (1975=100) Source: Gasques et al. (2009) 19 Figure 5. The BRAMS land-use map for the analysis of model simulation results 35 Figure 6. Spatial distribution of the Brazilian hydromet stations [Source: EMBRAPA] 36 Figure 7. Variation of the temperature estimated by the seven models for the weather station “DFUNBFAL”, located in Brasila, DF, Brazil (lat: -15,79; long: -47,9227). 37 Figure 8. Monthly maximum and minimum temperatures estimated from the seven models for the weather station “DFUNBFAL”, located in Brasília, DF, Brazil. (Lat:-15,79; Long:-47,9227). 38 Figure 9. Legally permitted area for agriculture based on land and environmental legal frameworks and landuse restrictions 40 Figure 10. Flowchart of components and biophysical, climatic, and plant growth processes used for zoning 41 Figure 11. Representation of the process of handling models to the generation of scenarios. 43 Figure 12. Example of low and high risk areas for planting corn in Brazil considering the sowing date in the first ten days of January, considering the pessimistic scenario. 43 Figure 13. Impact of climate change on area suitable for soybean (2010 – baseline and 2030 optimistic and pessimistic) 48 Figure 14. Projected losses in pasture productivity (%) relative to 2010 baseline under optimistic and pessimistic scenarios (2020 & 2030) 49 Figure 15. Impact of climate change on area suitable for beans – summer season (2010 – baseline and 2030 optimistic and pessimistic) 50 Figure 16. Impact of climate change on area suitable for beans – autumn season (2010 – baseline and 2030 optimistic and pessimistic) 51 Figure 17. Impact of climate change on area suitable for wetland rice (2010 – baseline and 2030 optimistic and pessimistic) 52 Figure 18. Impact of climate change on area suitable for sugarcane (2010 baseline and 2030 optimistic and pessimistic) 53 Figure 19. Impact of climate change on area suitable for cotton (2010 baseline and 2030 optimistic and pessimistic) 54 Figure 20. Impact of climate change on area suitable for beans – summer season (2010 – baseline and 2030 optimistic and pessimistic) 55 Figure 21. Impact of climate change on area suitable for beans – autumn season (2010 – baseline and 2030 optimistic and pessimistic) 56 Figure 22. Regions considered in the Brazilian Land Use Model – BLUM 59 Figure 23. Interactions between BLUM sectors 60 List of Tables Table 1. Evolution of the Structure of the Brazilian Agricultural Sector 20 Table 2. Crops and Area Planted in Brazil 42 Table 3. Percent change in area at low risk from climate change 46 Table 4. Simulated scenarios for soybeans, aggregated in BLUM regions (in 1,000 ha) 64 Table 5. Simulated planted area for crops and pasture for Brazil (in 1000 ha) 65 Table 6. Land available and suitable for agricultural expansion for each scenario (1000 ha) 66 Table 7. Land used in 2009 and potential projected for 2030 for each scenario (in 1000 ha) 67 Table 8. Land available and suitable for agricultural production, comparing scenarios for 2030 (1000 ha) 67 Table 9. Pastureland suitable for crop production, comparing scenarios for 2030 (1000 ha) 68 Table 10. Land used by pasture and first season crops (1000 ha) 69 Table 11. Land allocated to pasture (million hectares) 70 Table 12. Land allocated to cropsi (1000 ha) 71 Table 13. Grain production, first season crop only* (thousand tons) 72 Table 14. Beef production (thousand tons) 73 Table 15. Domestic consumption of each product analyzed (1000 tons and billion liters) 73 Table 16. Commodities’ real prices (2011=100) 74 Table 17. Production Value in R$ million (2011=100) 75 Table 18. Net trade results for each scenario (1,000 tons and billion liters for ethanol) 76 Foreword T his study is the result of an exemplary collaborative approach involving lead- ing edge Brazilian agencies - The Brazilian cultural area via deforestation. The Brazil- ian Government is also concerned about the potential adverse impacts of climate Enterprise for Agricultural Research (EM- change on Brazilian agriculture and asso- BRAPA), the University of Campinas (UNI- ciated livelihoods. For example, the Inter- CAMP), the Brazilian Institute for Inter- national Panel on Climate Change (IPCC) national Trade Negotiations (ICONE), and Assessment Report 4 (2007) noted that cli- the Brazilian Institute for Space Research mate change in Latin America will affect a 7 (INPE). We are honored to have been asso- number of ecosystems and sectors over the ciated to this effort. coming decades, with specific impacts on Agriculture is a major sector of the Bra- agriculture resulting in: zilian economy and is critical for econom- •• Reduction in the quantity and quality of ic growth and foreign exchange earnings. water flows and thus irrigation poten- Between 1996 and 2006 the total value of tial; the country’s crops more than quadruple, •• Increasing aridity, land degradation, and from 23 billion reais (~U$ 11 billion) to desertification; 108 billion reais (~U$ 53 billion). In 2009, •• Increasing incidence and impacts of agriculture accounted for 19.3 percent of crop pests and diseases; the labor force, or 19 million people, thus •• Decreasing plant and animal species di- strongly contributing to poverty reduction. versity and changes in biome boundar- Agribusiness employs 35 percent of the ies; and labor force and accounted for over 38 per- cent of Brazil’s exports, and a $77.5 billion •• Perturbations to ecosystem services (e.g. trade surplus in 2011. carbon sequestration, functional biodi- versity, environmental flows) needed to Against the backdrop of a vibrant and sustain the productivity of current agri- productive agricultural sector, Brazil con- cultural areas. tinues to pioneer agricultural intensifica- tion to increase agricultural productivity In order to mitigate the impacts of cli- even further to meet growing national, mate change, the world must drastically regional, and global food demands while reduce global greenhouse gas (GHG) emis- at the same time not increasing the agri- sions in the coming decades. To date, Brazil has led many key domestic and internation- Global and Regional Climate Models as well Impacts of Climate Change on Brazilian Agriculture al initiatives to reduce emissions from de- as significantly better hydro meteorolog- forestation and land use change, aggressive- ical and land suitability data were used ly promoted renewable energy, particularly to assess the vulnerability and impacts of bioenergy, and adopted a National Climate climate change on Brazilian agriculture. Change Policy, which includes an ambitious In addition, the study includes a coupled voluntary national GHG reduction target synthesis of climate-agricultural impacts for 2020. There is growing concern in Bra- with robust economic simulations to proj- zil and in the Latin American Region that ect economic impacts of climate change on increasing short term climate variability Brazilian agriculture to 2030. and medium to long term climate change The network of professionals estab- will have significant negative impacts on lished for this study can now contin- the Brazilian landscape and on Brazilian ue to improve and refine the integrated agriculture, national economic growth, and agro-ecological, biophysical, and economic associated livelihoods. The Government of modeling and analysis developed for this Brazil is developing proactive adaptation study. The inclusion of UNICAMP also lays 8 measures to counter the emerging risks of the foundation for capacity building of the climate change impacts on the major sec- next generation of climate modelers in Bra- tors underpinning the Brazilian economy zil and Latin America. The integrated and with special attention to agriculture. multidisciplinary expertise and knowledge This study builds on the findings of base strengthened by this study will serve previous World Bank studies, in particu- Brazil well as it enhances the productivity lar the 2010 World Development Report and resilience of its agricultural sector that to support growth and poverty reduction is critical not only for national food securi- efforts in the face of climate change, a re- ty, but also for the global supply of key agri- gional flagship report on Low Carbon, High cultural commodities. Growth Latin American Responses to Cli- It is our hope that Brazil will also serve mate Change (2009), and a regional fo- as a model and mentor for many developing cused on climate change impacts to Latin countries seeking to enhance their insti- America’s agriculture (2011). This study tutional capacities and strategic planning also integrates the methods and findings processes to combat climate change and to of several other Brazilian studies for ex- continue to develop sustainably in the face ample, Impact assessment study of climate of looming challenges from climate change. change on agricultural zoning (2006), Cli- mate change and extreme events in Brazil Deborah L. Wetzel World Bank Director for Brazil (2010), Assessment of regional seasonal Región de Latinoamérica y el Caribe predictability using the PRECIS regional climate modeling system over South Amer- ica (2010). In this study, a range of tested More than ten years ago, the Kyoto Proto- Although some scientists believe that col established the tolerable limits for car- global warming proceeds at a lower rate bon emissions resulting from the economic than so far estimated and, thus, that the and social activities of the countries, with a countries would have more time for cor- view to slowing down global warming and rections and adjustments, most scientists other climatic changes. Scholars have since agree that global warming is ocurring. The demonstrated that the effects of climate climate-change scenarios point out to an change are already being felt as higher av- average temperature rise in excess of 2°C erage air temperatures and the impacts of by 2050 and that the impact of such high- extreme temperature and rainfall events on er temperatures would cause major imbal- the world’s populations. ances in ecosystems essential to human- In the ensuing period the concentration kind survival. of carbon dioxide, the main greenhouse Nevertheless, the significant changes gas, has risen to four hundred parts per predicted for the Amazon Forest and its million. Data from the World Meteorologi- biodiversity, the meaningful losses of the cal Organization showed that 12 out of the glaciers in the Andes and Himalayas and 9 last 13 hottest years since the measure- the fast acidification of the oceans and con- ments started in 1850 occurred between sequent break down of marine ecosystems 2001 and 2012. Nevertheless, as Professor and death of coral reefs are still plausible Ed Hawkins of the University of Reading events, all of which would condemn count- has demonstrated the average temperature less species to extinction and considerably has surprisingly remained the same since affect world food supply. The speed and 1988, when the Kyoto Protocol was opened magnitude of the change can doom many to accession by the countries. species to extinction and significantly af- Such apparently conflicting data flash- fect food supply in the planet. Some people es a warning light to the world population are still sceptical.   because they remind the fact that Earth is Scientist are free to chose the hypoth- much more complex than the human mind eses in their research, but are enjoined to can conceive and its climate does not al- go beyond the initial evidence and not re- ways provide constant and predictable an- nounce their duty to delve ever more deep- swers. They also forewarn the people that ly in the mysteries of Nature. The authors science is entering a nebulous area where of this book acknowledge and rigorous current scientific knowledge does not en- comply with such tenets. able to understand the phenomenon and Ten years ago, Embrapa’s Eduardo Assad that new scientific breakthroughs are nec- and Unicamp’s Hilton S. Pinto researchers essary to reduce the uncertainties about developed a series of studies as an attempt Earth’s future climate. to associate the various global warming Foreword scenarios with possible impacts on climatic risks zoning and their ultimate influence on The results of the simulations pointed Impacts of Climate Change on Brazilian Agriculture agricultural production in Brazil. They used out to the same direction, confirming tem- a single climate model for estimating pro- perature increases predicted for 2030. The duction area fof each municipality in Brazil. climatic-risk agricultural zoning technolo- They also showed the different consequenc- gy was then applied using those results to- es of higher temperatures on various crops gether with information about the land use cultivated in the  regions of Brazil. effectively altered by anthropogenic activi- Assad and Pinto took up the challenge ties. In a further step the impact on Brazil- in this book, with the fruitful and diligent ian agriculture was calculated by means of support of André Nassar and Leila Harfuch economic models. from the Institute for International Trade The work makes clear that agriculture Negotiations – ICONE; Saulo Freitas from is vulnerable to higher temperatures giv- the National Institute of Space Research en the predicted levels of global warming. – INPE; and from Barbara Farinelli, Mark There can be losses in yield and conse- Lundell and Erick Fernandes, from the quently in production. There can have crop World Bank. migration from one region to another. The 10 A more precise analysis of the vulner- regional production profile can  change. ability of Brazilian agriculture is possible The outcome of the current work, which using new tools for model climate and the clearly goes beyond previous studies, em- dynamics of soil use. The researchers used phasizes the need to assign priority to bio- seven climate analysis models proposed by technology  research, particularly in the the Intergovernmental Panel on Climate regions that will be strongly affected, such Change-IPCC, three of which in a detailed as Northeastern and Southern Brazil. The resolution and adapted to tropical condi- search for genes that increase plant toler- tions. Brazilian Regional Climate Model - ance to high temperatures and water stress RCM - BRAMS developed by CPTEC/INPE should be a routine endeavour in the next was also analysed. few years. The implementation of pub- The researchers were then able to re- lic policies associated to more balanced duce the degree of uncertainty of the re- production systems should be supported sults by considering the maximum and throughout the country in order to reduce minimum deviations of the estimated tem- the impact of global warming by either mit- peratures by the models for both optimistic igating greenhouse gas emissions or inte- and pessimistic scenarios considering the grating production systems better adapted temperature ranges obtained by the differ- to abiotic stresses. ent models. The most important outcome Only studies like this one, carried out is the possibility of appraising the degree by a consortium of institutions and based of uncertainty of the results by using more on advanced scientific and technological models. knowledge, can help identify the solutions required by such a challenging future. The to help in the development of Brazilian ag- possibility of a two-degree Celsius increase riculture. It is unlikely that we will ever be in the average temperature of Earth re- able to express the true importance and quires that tropical agriculture be prepared scope of international scientific coopera- for such conditions assuring the guarantee tion in the evolution of Brazilian agricultur- food security during the next few decades.   al science. Embrapa is also greatly indebt- The first approach to the vulnerabili- ed to the generosity of institutions such as ty of Brazilian agriculture vis-à-vis global Unicamp, INPE and ICONE, always willing warming relied on material support from to contribute to network studies, an asset the British Government, through the Brit- that guarantees quality work. ish Council. The more recent study received Maurício Antônio Lopes support from the World Bank, always ready President of Embrapa 11 Foreword 12 Impacts of Climate Change on Brazilian Agriculture Executive Summary 13 A griculture is a major sector of the Bra- zilian economy accounting for about 5.5% of GDP (25% when agribusiness is ingly vulnerable to climate variability and change. To meet national development, food included) and 36% of Brazilian exports. security, climate adaptation and mitiga- As per the 2006 agricultural census, Brazil tion, and trade goals over the next several has 5 million farms of which 85% are small decades, Brazil will need to significant- holders and 16% are large commercial ly increase per area productivity of food farms occupying 75% of the land under cul- and pasture systems while simultaneous- tivation. In 2009, Brazil enjoyed a positive ly reducing deforestation, rehabilitating agricultural trade balance of $55 billion. In millions of hectares of degraded land, and the second quarter of 2010, Brazil’s econo- adapting to climate change. Because of the my recorded 8.8 percent growth with agri- projected magnitude of the agricultural im- culture making a major contribution (11.4 pacts and investments required, and the percent) relative to the industrial (13.8 decadal response time of best adaptation percent) and services (5.6 percent) sectors. options available, there is an urgent need Because agriculture is vital for national for a state of the art, assessment of climate food security and is a strong contributor to change impacts on agriculture to guide pol- Brazil’s GDP growth, there is growing con- icy makers on investment priorities and cern that Brazilian agriculture is increas- phasing. The current projections of climate change impacts for Brazil are based on cli- In the absence of climate change, Bra- Impacts of Climate Change on Brazilian Agriculture mate models that were available prior to zilian cropland is projected to increase to 2008. Since then, not only has the science 17 million hectares in 2030 compared to and quality of global, regional, and local observed area of cropland in 2009. Due to modeling advanced significantly, but also climate change impacts, however, all the improved land quality and climate data are scenarios simulated in this study, result in now available. a reduction of ‘low risk – high potential’ This study built upon several recent flag- cropland area in 2020 and 2030. More spe- ship studies on climate change impacts by cifically, our findings suggest that the South using a range of tested Global and Regional Region of Brazil, currently an agricultural Climate Models as well as significantly bet- powerhouse, could potentially lose up to 5 ter hydrometerological and land suitability million ha of its highly suitable agricultural data that were unavailable to previous such land due to climate change while Brazil as a studies to assess the vulnerability and im- whole could have around 11 million ha less pacts of climate change on Brazilian agri- of highly suitable agricultural land by 2030. culture. Fortunately, our findings also show that 14 In addition to confirming and the results the bulk of the loss of high potential agri- from previous climate change impact stud- cultural land could be allocated to current- ies that projected significantly negative im- ly occupied by poorly productive pastures. pacts on Brazilian crops to 2020 and 2030, The displacement of pastures by grains and our findings help to further extend the sugarcane in the Center-West, and North- knowledgebase not only on the extent of east Cerrado Regions could potentially impacts to different crops but also the level compensate for the projected loss of suit- of impacts in the different regions of Brazil. able cropland and especially the grain loss- es in the South (~9 million tons) by about For example, this study showed that half. while for some crops (soybean and cotton) the projected negative climate impacts to It is especially noteworthy that despite 2020 are likely to be more moderate than the projected reduction in the pasture area, previously projected, for other crops (beans beef production is projected to decrease by and corn), however, the impacts could be a much lower amount due to technological significantly more severe that than project- intensification. So although pasture pro- ed in previous studies. These differences ductivity in Brazil might decrease by 7% in illustrate, at least partially, the value of har- all scenarios simulated in 2030 compared nessing more complete and geographically to the baseline, our simulations project distributed climate, terrain characteristics, that beef production may continually grow water, and climate data sets for more nu- until 2030 in all scenarios compared to the anced analytical power of climate change observed production in 2009, and could in- modeling approaches. crease by more than 2 million tons. Although pasture intensification po- ple products. It is noteworthy that beef and tentially compensates for displacement of soybean oil account for almost 50% of the the pastures by grain and sugarcane in the projected total production value for Brazil- central regions of Brazil, this study projects ian agriculture. that beef producer prices are projected to The projected impacts of climate change increase by more than 25% in all scenarios, on rainfall and soil moisture deficits at crit- suggesting that that intensification of pas- ical phases of crop growth from this study ture use and cattle production might lead suggest that there is an urgent need for to a price increase in order to compensate more detailed analysis for priority crop for the investments to increase yields. zones to develop an integrated improved, In general, the production declines can drought-tolerant (deeper rooted) varieties be expected to impact prices, domestic de- coupled with good land and water manage- mand, and net exports of these products. ment strategies to mitigate the projected Relative to 2009 and in the absence of cli- effects. In addition to extending access to mate change, domestic consumption of all efficient irrigation technology, manage- commodities is projected to increase in ment strategies that conserve and enhance 15 2020 and 2030. However, our simulations soil carbon will increase soil moisture re- across all the climate change scenarios tention capacity. For example, suggest that when compared to the 2009 a. The Brazilian Government and the pri- baseline, climate change is likely to reduce vate sector have been steadily facilitat- consumption of almost all commodities, ing the adoption of improved conserva- specially grains and ethanol. The major tion agriculture practices, such as no-till driver of this projected reduced consump- planting, and more resource-efficient tion is the higher real price of all commod- systems, such as integrated crop-live- ities when land availability for agricultural stock systems that are inherently more production is reduced as a function of cli- resilient to climate shocks than some in- mate change. tensive cropping systems. The production impact estimates from b. The Government is providing credit and our study show that unlike previous esti- financing for the newly-launched “Low mates of declining agricultural production Carbon Agriculture” program with ap- value, the negative impacts on supply of proximately US$ 1 billion available for agricultural commodities is expected to low interest credit in the 2011 season result in significantly increased prices for alone. some commodities, especially staples like c. The buildup of agricultural soil carbon rice, beans, and all meat products. This will Executive Summary may also be eligible for carbon pay- counter the effect of declining productivi- ments in voluntary and (future) formal ty on value of agricultural production but markets. could have major negative effects on the poor and their consumption of these sta- In our study, the efforts to access the adaptation measures to counteract the pro- Impacts of Climate Change on Brazilian Agriculture latest available hydrometeorological and jected negative impacts of climate change land use data significantly improved our on agricultural productivity. The need for ability to undertake more robust modeling improved and integrated climate change and impact projections. Nevertheless, the impact assessments is especially urgent for lack of good quality and long term climate the agricultural sector. A recent survey car- data continues to hamper regional and lo- ried out by the Brazilian Enterprise for Ag- cal climate modeling efforts as well as the riculture and Animal Research (EMBRAPA), calibration and validation of current pro- revealed that even with advanced breeding jections that are being used to inform pol- techniques, it takes approximately 10 years icy and investment decisions to 2030 and of R&D and costs at around US$6 -7million beyond. Because the climate forcing factors to develop, test, and release (including 2-3 operate both within and external to na- years for scaling up seed production) a new tional frontiers, there is an urgent need for crop cultivar or variety that is heat and/ coordinated and targeted climate change or drought tolerant. The review synthesis investments over the next 1-5 years for in- from this report suggests that within the 16 strumentation, data assembly, data sharing next decade, Brazilian agriculture will al- and data access systems. National, bilateral, ready be dealing with a significant level of and multilateral investments agencies need climate induced crop and livestock produc- to coordinate their investment strategies to tivity stresses. Much of the crop improve- support this specific and urgent need. ment work to date has focused on drought The need for improved and integrated tolerance and a great deal still remains to climate change impact assessments is es- be done for heat tolerance. pecially urgent for the agricultural sector. The findings of this study will be incor- A recent survey carried out by the Brazil- porated in the EMBRAPA/UNICAMP Agro- ian Enterprise for Agriculture and Animal ecozone Model to improve the simulation Research (EMBRAPA), revealed that even and climate impact projections that under- with advanced breeding techniques, it pin the national rural credit and insurance takes approximately 10 years of R&D and programs in Brazil. This means that the costs at least US$6 million to develop, test, study will immediately begin having far and release a new crop cultivar or variety reaching operational and policy implica- that is heat and/or drought tolerant. tions in Brazil. The experiences from Brazil It is important to note that this study are highly relevant for other regions and did not simulate the potential impact of countries where similar work is on-going. technological advances (new varieties, ex- panded and enhanced access to irrigation, improved land and water management) as Introduction 17 A griculture is a major sector of the Bra- zilian economy accounting for about 5.5% of GDP (25% when including agri- mainly the production of beef and poultry, pork, milk, and seafood. Brazil is currently the world’s largest exporter of beef, poul- business) and 36% of Brazilian exports. try, sugar cane and ethanol. Brazil is the world’s largest producer of Between 1996 and 2006 the total value sugarcane, coffee, tropical fruits, frozen of the country’s crops rose from 23 billion concentrated orange juice, and has the reais to 108 billion reais, or 365%. Brazil in- world’s largest commercial cattle herd at creased its beef exports tenfold in a decade 210 million head. Brazil is also an import- overtaking Australia as the world’s largest ant producer of soybeans, corn, cotton, co- exporter. It is also the world’s largest ex- coa, tobacco, and forest products. Between porter of poultry, sugar cane and ethanol. 1996 and 2006 the total value of the coun- Since 1990 its soybean output has risen try’s crops rose 365 percent from 23 billion from barely 15m tons to over 60m. Brazil reais to 108 billion reais (US$ 64 billion). accounts for about a third of world soybean Brazil accounts for about a third of world exports, second only to America. In 1994 soybean exports and supplies a quarter of Brazil’s soybean exports were one-seventh the world’s soybean trade from 6% of the of America’s; now they are six-sevenths. country’s arable land. The remainder of ag- Moreover, Brazil supplies a quarter of the ricultural output is in the livestock sector, world’s soybean trade on just 6% of the velopment of the agriculture technology Impacts of Climate Change on Brazilian Agriculture country’s arable land. (Figure 1). In 2009, Brazil enjoyed a posi- From 1991 to 2010 the grain production tive agricultural trade balance of US$55 bil- of the country (cotton, peanut, rice, bean, lion. In the second quarter of 2010, Brazil’s sunflower, corn, soybean, sorghum, wheat, economy recorded 8.8 percent growth with oat, barley, castor bean, rye and rapeseed) agriculture making a major contribution increased 147% and the cultivated area (11.4 percent) relative to the industrial only 25%, or 4,8%/year and 1,7%/year (13.8 percent) and services (5.6 percent) respectively, that represents an strong de- sectors. Figure 1. Grain production and area increase in Brazil from 1991 to 2010. PRODUCTION MILLION TON 144.1 145.9 131.8 135.1 PRODUCTION INCREASE 123.2 119.1 122.5 114.7 +153,7% = 4,8%/year 100.3 96.8 18 82.4 83.0 81.1 78.4 76.6 76.0 73.6 68.4 68.3 57.9 CULTIVATED AREA (MILLION HA) INCREASE = 25,4% = 1,7%/year 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09** 10*** Fonte: CONAB - AGE/Mapa In recent decades, this growth in the Total output has grown 2.5 times since Brazilian agricultural sector has increas- the 70s while the use of labor is decreased ingly been driven by productivity gains in and the use of capital and land has slightly cereals, coarse grains, sugarcane, oilseeds increased. More importantly, the produc- and milk sectors. Brazilian production tivity of all production factors has strongly has grown more than 1.5 times the rate of increased for the same period (Figure 4) world production (figure 2 and 3). In meat (Source of data: FAO/FAOSTAT) sectors, the average growth has been 1.8 times faster than the world production. Figure 2. World and Brazil Agricultural Production: Expansion from 1990 to 2009 (1990=100) 300 300 250 250 200 200 150 150 100 100 50 50 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Brazil_Sugarcane Brazil_Coarse grains World_Sugarcane World_Coarse grains Brazil_Cereals Brazil_Oilcakes World_Cereals World_Oilcakes Figure 3. Brazil’s and World Meat and Milk Production: Expansion from 1990 to 2009 (1990=100) 450 450 19 400 400 350 350 300 300 250 250 200 200 150 150 100 100 50 50 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Brazil_Beef Brazil_Milk World_Beef World_Milk Brazil_Chicken Brazil_Pork World_Chicken World_Pork Figure 4. Brazilian Agricultural Sector: Total Production and Use of Labor, Land, and Capital and Total Factor Productivity (1975=100) Source: Gasques et al. (2009) 400 450 350 400 350 300 300 250 250 200 200 150 150 100 100 50 50 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 Introduction Capital Land Total factor of productivity Productivity of land Production Labor Productivity of labor Productivity of capital The Evolution of the Farming Sector in Brazil Impacts of Climate Change on Brazilian Agriculture and Implications to 2030 The 2006 Agricultural Census5 indicated between 1995 and 2006, the number of that the number of rural households is in- smaller households (less than 100 hect- creasing again despite the reduction ob- ares) has increased. In terms of land occu- served from the 80s to 90s. With respect to pied, both the smallest and the largest size the distribution of households according to classes have decreased from 1995 to 2006, farm size, the agricultural sector is clearly while the middle classes (10 to 1,000 ha) becoming less concentrated. For example, have increased. Table 1. Evolution of the Structure of the Brazilian Agricultural Sector   NUMBER OF RURAL HOUSEHOLDS (units) LAND OCCUPIED (HA)   1970 1975 1980 1985 1995 2006 1995 2006 20 Total 4,924,019 4,993,252 5,159,851 5,801,809 4,859,865 5,175,489 353,611,246 329,941,393 Less than 10 ha 2,519,630 2,601,860 2,598,019 3,064,822 2,402,374 2,477,071 7,882,194 7,798,607 10 to 100 ha 1,934,392 1,898,949 2,016,774 2,160,340 1,916,487 1,971,577 62,693,586 62,893,091 100 to 1000 ha 414,746 446,170 488,521 517,431 469,964 424,906 123,541,517 112,696,478 More than 1000 ha 36,874 41,468 47,841 50,411 49,358 46,911 159,493,949 146,553,218 Source: IBGE Agricultural Census 2006 The total factor productivity (TFP) of the zil’s effort to prioritize intensification-led Brazilian agriculture has increased steadi- productivity gains as opposed to expanding ly over the last 35 years. Relative to 1970 farm areas (Contini et al. 2010).5 (=100), production has increased by 243%, Will the Brazilian agricultural landscape inputs by 53%, and TFP by a corresponding in 2030 and beyond resemble current ag- 124%. Investments in R&D have been fun- ricultural landscapes in Australia, Can- damental for these increases and it has been ada, and the USA that are dominated by estimated that a 1% increase in agricultur- few large, technologically advanced farms al R&D has resulted in a 0.2% increase in TFP (Gasques et al. 2009). Positive trends 5 The last agricultural census for which data is currently in the productivity indices underscore Bra- available with national value added derived from could result in major economic windfall for land, capital, and skilled labor? The data smallholder farmers. Many of the function- presented in Figures 2-4 suggest that pro- al foods are “low volume, high value” prod- ductivity gains and production growth is ucts that are well suited to smallholder taking place in the agricultural sector as a cropping systems in the Brazilian farming whole. Interestingly, the rural sector in Bra- landscape. zil is becoming more capital and labor in- Another issue that is often overlooked tensive across all the scales of farms small in the discussions on future scenarios for and large (see Figure 4 above). It is difficult Brazilian agriculture is the prevailing le- to envisage the structure of Brazilian farms gal and land administration framework in 2030 but the above trends suggest that in Brazil (Sparovek et al., 2010). The legal simulations on the long term future of the framework dictates what can and cannot Brazilian agricultural sector need to focus be done in the rural and agricultural land- on the sustainable intensification of the scape (e.g. maintaining riparian zones, le- production rather than the likely changes gal (forest) reserves, securing indigenous in the structure of the households and the lands etc.) and the Brazilian Government 21 size of farms. is aggressively enforcing the legal aspects Small holder farmers are generally more via a range of monitoring actions, policies, vulnerable to economic and environmental and fiscal instruments. The two main legal shocks and have access to fewer resources frameworks are (a) the Forest Law and (b) to adapt to climate variability and change Preservation Areas such as state and na- when compared to large scale farmers. tional parks, and indigenous reserves. However, relative to large scale producers The Forest Law, currently being revised that rely on one or two crops planted over and under discussion in the Brazilian Sen- thousands of hectares, small holder farm- ate, covers all natural vegetation (the Am- ers can play a vital role in providing land- azon, the Atlantic Forest, the Cerrado (sa- scape scale resilience through a diversity of vanna), the Caatinga (the scrub woodland production approaches that harness a wid- in northeastern Brazil), the Pantanal, and er spectrum of agrobiodiversity while also the Pampas (grassland of southern Brazil). preserving and harnessing ecosystem ser- The law delineates rural private land into vices and the emerging markets for these land for production and land that must be services (carbon and biodiversity offsets, preserved. The land that must be preserved hydrological flows for reduced floods and/ with natural vegetation on all private farm- or improved water quality). In addition to land is further subdivided into (a) conser- emerging markets for ecosystem services, vation areas (Legal Reserves) and (b) Areas the increasing global demand for “func- for Permanent Preservation (APP) that in- tional foods” (foods that have direct health Introduction clude (i) riparian zones defined as vege- benefits like reducing cholesterol, improv- tation strips along rivers and other water ing liver function, reducing hypertension) bodies with varying width depending on type and size of the water body, (ii) any It is possible to envisage a ‘paradigm Impacts of Climate Change on Brazilian Agriculture land with slopes >45°, (iii) hill tops, and shift’ for a productive, resilient, culturally (iv) any land above 1800m above sea level. appropriate and inclusive Brazilian rural The goal for the APP is to protect parts and farming landscape that has both large of the landscape with strategic value for farmers ensuring efficient high volume freshwater recharge and thus the APP can- growth and smallholders ensuring resil- not be used for any type of production ac- ience to climate change shocks via a range tivities and must be maintained with the of cropping systems that are productive original native vegetation. Changes to this and profitable on the basis of payments aspect of the law that allow the planting for ecosystem services (e.g. Reduced Emis- of exotic tree species (e.g. eucalyptus, Af- sions from Deforestation and Degradation rican oil palm) and the reduction in the – REDD plus) and the emerging markets to prescribed area to be preserved are cur- meet the growing global demand for func- rently the subject of intense discussions. tional foods and feedstock for industry! Legal Reserves are established to pro- Understanding the evolving intensification, mote biodiversity conservation. Although vulnerability, resilience, and investment is- 22 the primary goal is to maintain the native sues and better mapping projected climate vegetation, Legal Reserves can be used for change impacts across relevant spatial and some low-impact production systems, such time scales, will be critical to enhancing as managed low-impact forest extraction, and sustaining Brazil’s agriculture and ru- selected agroforestry systems, and apicul- ral sectors and their competitive regional ture. These are suitable for smallholder and global advantage. This short section is family agriculture and possibly alternative included to highlight the importance of the production schemes aiming at niche mar- legal aspects on future expansion, intensi- kets. Conventional mechanized agriculture fication, and diversification of Brazilian ag- employing intensive inputs or forestry op- riculture. The Forest Law (Legal Reserves erations employing complete forest remov- and APP) is currently under debate in the al are not allowed. Ideally, any proposed Brazilian senate with the objective of revis- changes in the forestry law should also con- ing the law. A full discussion of the evolving sider the implications of projected impacts legal framework and its potential influence from climate change on the agroecological on the future structure of Brazilian agricul- landscapes. ture is beyond the scope of this report (see Sparovek et al., (2010) for a detailed review and discussion). Threats to Brazilian Agriculture from climate variability and eventual climate change There is growing concern in Brazil and in the soybean sector alone accounting for the Latin American Region that increasing almost 50% of the losses; short term climate variability and medi- ii. Under a pessimistic Climate Change sce- um to long term climate change will have nario (A2), the best current coffee pro- significant negative impacts on the Brazil- duction (“low risk”) areas are expected ian landscape and on Brazilian agriculture, to shrink by at least 30%, which could national economic growth, and associated result in losses of close to US$ 1 billion livelihoods (Assad and Pinto, 2008; Margu- by 2050. Interestingly, even under a pes- lis and Dubeux, 2010). simistic A2 scenario, the area suitable •• The study by Assad and Pinto (2008) 35 for sugarcane could double by 2020. crops were assessed in terms of climate The study by Margulis and Dubeux (2010) risks but nine major crops (cotton, rice, used the Assad and Pinto (2008) study coffee, sugarcane, beans, sunflower, cas- 23 methods based on a single GCM-RCM com- sava, maize and soybean, as well as pas- bination and the A2 and B2 IPCC Third tures and beef cattle) representing 86% Assessment Report scenarios. The climate of the planted area in Brazil, received modeling outputs were used to drive a com- special focus. putable general equilibrium (CGE) model to •• Based on a 2007 baseline, climate risk better assess the likely economic impacts zone mapping in 5,000 municipalities due to projected climate change to 2020, for these crops, the agricultural scenari- 2050, and 2070. The simulations showed os in Brazil were simulated for the years that Brazil’s GDP in 2050 will approximate 2010 (closest representation to the cur- US$9.4 trillion and that in the worst case rent conditions), 2020, 2050 and 2070 (IPCC Scenario A2) the country could lose and two IPCC Third Assessment Report about 2.5% every year due to temperature scenarios: A2 the most pessimistic, and increase impacts. At a discount rate of 1 B2, slightly more optimistic. In scenario percent per year, this is equivalent to the A2, the estimated temperature rise vari- loss of one whole year’s GDP over the next ation is between 2°C and 5.4°C; and in 40 years. The study’s findings also project- B2, between 1.4ºC and 3.8ºC. ed a significant reduction in the best crop The results showed that: areas currently characterized by low pro- i. Projected climate change impacts on duction risk, for 8 of the 9 major food and all currently produced food grains will export crops (Table 1). Introduction amount to US$ 4 billion by 2050 with Impact of climate change on current “low risk” areas suitable Impacts of Climate Change on Brazilian Agriculture for cultivation (Margulis et al., 2010) Crops Variation relative to current productive area (%) SRES B2 SRES A2 (+1.4ºC to +3.8ºC) (+2°C to +5.4°C) 2020 2050 2070 2020 2050 2070 Cotton -11 -14 -16 -11 -14 -16 Rice -9 -13 -14 -10 -12 -14 Coffee -7 -18 -28 -10 -17 -33 Sugar cane 171 147 143 160 139 118 Beans -4 -10 -13 -4 -10 -13 Sunflower -14 -17 -18 -14 -16 -18 Cassava -3 -7 -17 -3 -13 -21 24 Maize -12 -15 -17 -12 -15 -17 Soybean -22 -30 -35 -24 -34 -41 The projected reductions in cultivation et al., 2010 are sobering (Table 2 below - 1 area of low risk and associated economic US $ = Br$ 1.8) losses to 2050 as summarized by Margulis Reduction in “low risk” Scenario A2 Annual Economic loss Crop cultivation area (%) (Millions Reais)* Rice -12 530 Cotton -14 408 Coffee -17.5 1,597 Beans -10 363 Soybean -32 6,308 Maize -15 1,511 Sugar cane 145 0 The Margulis and Dubeux (2010) study was sity, energy, and hydrological resources) a pioneering contribution to the Brazilian and the macroeconomic growth implica- knowledgebase on climate change impacts tions at a national scale. The authors nev- on a range of sectors (agriculture, biodiver- ertheless identified the following opportu- nities for improving future climate change A regional study on climate change impacts economic impact assessments: to Latin America’s agriculture (Fernandes 1. The use of a suite of GCMs and RCMs et al. 2011) also found that agricultural for improving the robustness of climate productivity is likely to be significantly and change projections rather than the sin- negatively impacted, albeit with different gle GCM and RCM used for the study. sub-regional intensities (BOX 1 below). 2. The improvement in projected rainfall The study projected that Brazilian soy- impacts as there was no consensus in bean, production could decline by as much the magnitude and direction of the pro- as 30% in 2020 and even more so by 2050 jected rainfall impacts – a problem that with significant decreases also likely in continues to plague most other studies. maize, and wheat. Encouragingly, however, the study reported that simulation of adap- 3. An explicit treatment of uncertainty and tation interventions (short/long cycle vari- the magnitude and frequency of extreme eties, deeper rooted/drought tolerant vari- events eties, moderate irrigation at critical growth 4. Improvement in the data density (crop phases, and a shift in planting dates shoed area, land quality, rainfall, temperature, the possibility of mitigating a significant 25 runoff, infiltration, biodiversity, land amount of the yield declines in all impacted cover dynamics) and data accessibility crops. for model parameterization, calibration, and validations. BOX 1 - Climate Change and Agriculture in Latin America, 2020-2050 The World Bank’s 2011 regional study (Fernandes et al. 2011) reported that the prevailing and often expressed view that Latin America and the Caribbean will continue to be the breadbasket of the future—stepping in to supply grain to other regions affected by climate change—needs to be tempered and subjected to fur- ther rigorous testing. Key findings include: •• For wheat, yields could be significantly affected by climate change, regardless of the emission scenario or general circulation model. Percentage yield declines are projected to be deeper in Mexico, Colombia, and Brazil. Yield reductions due to the shortening of the crop cycle resulting in fewer days to fill grains. The projected yield declines due to disease in 2020 and 2050 could also be signif- icant. With few exceptions, insufficient water could affect wheat productivity more than other factors. •• For soybean, yields could be reduced by climate change in 2020 and more so in 2050, though with different magnitudes throughout the region. Yield loss- es could be large in Brazil (more than 30% from the baseline) but less pro- Introduction nounced in Argentina, Bolivia, Colombia, and Uruguay. This can be explained by the greater impact of climate change in Brazil, where the crop cycle is projected to be shorter than in other parts of Latin America, and likely to result in a mark- Impacts of Climate Change on Brazilian Agriculture edly reduced soybean grain-filling period. •• For maize, climate change could reduce yields throughout Latin America, re- gardless of the emission scenario or GCM. This is mainly due to the shorter grain-filling period not being compensated for by the higher daily biomass ac- cumulation rates and the CO2 fertilization effect. The countries most affected are likely to be Brazil, Ecuador, Mexico, and Caribbean countries, where maize is one of the main crops. •• For rice, the AZS estimates show that productivity could, on average, increase across the region. A major reason for this positive outlook appears to be relat- ed to the fact that rice is a wetland/irrigated crop. Except for Brazil, Mexico, and the Caribbean, the 2020 and 2050 projections are encouraging, with high- er productivity projected in most cases. In low-temperature areas (especially Uruguay and southern Brazil) climate change could reduce the incidence of pre-flowering cold shocks inducing sterility. Except for Brazil and the Caribbe- an, rice ‘blast’ disease pressure could ease, because temperature and rainfall conditions become less favorable for the blast causing pathogen Pyricularia grisea, 26 A key challenge and opportunity for Brazil 1. the simulations regarding climate is the need to better understand, quantify, change were based on scenarios of uni- and map the locations of projected impacts form increase in temperature and pre- on currently productive agriculture and to cipitation and did not use geographical- better quantify the magnitude and uncer- ly differentiated climate projections. tainty associated with current projections 2. the studies used climate data sets that of both positive and negative impacts. were significantly less comprehensive In 2011, a World Bank report (Assad et in terms of geographical distribution as al. 2011) presented a detailed review of well as precision than what is currently the literature and outputs of recent empir- available. ical studies that had developed projections 3. The studies were based exclusively on of the likely impacts of climate change on Global Climate Models (GCMs) for pro- Brazilian agriculture. Generally speaking, jections of future climate change im- the empirical evidence suggests that the pacts. Although simulations with GCMs net impact of climate change on Brazilian are appropriate tools to address global agriculture is negative, although there are to sub-continental scale climate change varying regional consequences. However, and impacts (Giorgi et al. 2001), the most of the studies to assess likely impacts results of long-term multimodel GCM of climate change on Brazilian agriculture simulations must still be treated with were constrained by several of the follow- caution as they do not capture the de- ing limitations: tail required for regional impact assess- ments, due in part to the coarse resolu- sults for Brazil demonstrate that RCMs tion (~300 km x 300 km) in the majority show good skill in the simulation of the of the models used. The concern about present-day climate, they still require the low spatial resolution of GCMs is es- adjustments and calibration (based on pecially relevant for heterogeneous re- local data and field observations) to the gions, such as South America, where the settings used by the model in order to distributions of surface variables such correct for the systematic errors inher- as temperature and rainfall are often in- ited from the GCM from which they were fluenced by local effects of topography, derived and ultimately to produce useful and thermal contrasts, which can have estimates of regional, and seasonal to in- a significant effect on the climate (Alves ter-annual climate projections (Maren- and Marengo, 2009). go et al., 2009b). 4. To address country, sub-country and lo- cal scale climate change consequences or impacts, higher resolution (e.g. 50 km x 50 km) regional climate models (RCMs) 27 have been employed. It is important to note, however, that although the re- Introduction 28 Impacts of Climate Change on Brazilian Agriculture Scope of this study 29 T his study builds upon the findings of a decade of research on climate change impacts on Brazilian agriculture and pro- •• changes in the distribution of land use and production within Brazil for given supply and demand scenarios. vides the latest findings of new modeling This report highlights the outputs of target- approaches and simulations of: ed modeling on major Brazilian crops by re- i. projected climate change in Brazil to gions within Brazil to provide more robust 2020 and 2030 and quantitative information on how and ii. the likely impact of climate change on where the drivers of agricultural produc- existing agroecological zones and their tion growth are more likely to be impacted suitability for major grain crops, sugar- by changing climate. The goal is to empow- cane, cotton, and pastures, and er policy makers to ensure that the farm- iii. the economic impacts of changes in ing sector has access to the knowledge and agroecological zone suitability for the resources to undertake the adaptation that various crops and the: will be necessary to cope with unavoidable climate changes while simultaneously con- •• induced changes in supply and de- tributing to mitigating GHG emissions. mand of agricultural products at a national level, Four key integrated and linked interven- tions were used to attempt to significantly •• the economic effects on agricultural improve currently available assessments production and profitability, and of climate change impact on Brazilian ag- that incorporates aerosol and land Impacts of Climate Change on Brazilian Agriculture riculture and to guide policy makers with cover/land use feedbacks for much the priorities and phasing of needed invest- improved local weather and climate ments. This study: (especially rainfall) projections. 1. Accessed and incorporated the best 3. Coupled the best GCM, RCM, and BRAMS available hydrometerological data from suite of models identified above with all calibrated and validated ground sta- the EMBRAPA/UNICAMP Agro Zoning tions of the Brazilian Water Agency model and recently available highly dis- (ANA) in all the sub-regions in Brazil to aggregated land (soil) quality data at significantly reduce the identified “cli- municipal level to develop an updated mate data deficiency” of previous stud- EMBRAPA AZM. ies. 4. Coupled the EMBRAPA AZM with the 2. Refined climate change projections via Brazilian Land Use Model (BLUM) for an coupling global, regional and local scale improved Climate-Sensitive BLUM to as- models to provide more robust climate sess: 30 change projections for Brazil. This was a. Climate change induced changes in achieved via: supply and demand of agricultural a. An analysis of the best ensemble of products at a national level global and regional climate models b. Changes on the distribution of land (GCMs and RCMs) that have been test- use and production within Brazilian ed for Brazilian climate conditions territory for given supply and de- b. Integration of the best available mand scenarios. GCMs and RCMs with a the state of c. Economic effects on agricultural pro- the art Brazilian developments in Re- duction and profitability gional Atmospheric Model (BRAMS) Climate Change and Agricultural Impact Projections in Brazil to 2030 and Beyond 31 B ased on the literature review findings highlighted in the preceding sections, and the emerging outputs of on-going al and international programs have used RCMs to help quantify better regional cli- mate change and provide regional climate work in Brazil, there are significant oppor- scenarios for assessing climate change im- tunities to improve both the quality and pacts and vulnerability. These have all fol- the robustness of the currently available lowed a standard experimental design of projections for climate change impacts on using one or two GCMs to drive various re- Brazilian agriculture over the next three to gional models from meteorological services four decades. One option has been to devel- and research institutions in the regions to op regional climate models (RCMs) nested provide dynamically downscaled regional within a GCM to facilitate more robust pro- climate projections over Central and South jections at national to sub-national scales America (Marengo et al., 2009; Soares and (Christensen et al. 2007). Various nation- Marengo, 2009; Urrutia and Vuille, 2009). Refining climate change impact projections via Impacts of Climate Change on Brazilian Agriculture global, regional and local scale modeling As discussed in previous sections, the rel- tionale for the choice of global model Ha- atively coarse resolutions of GCMs pose dAM3P is because (a) the model adequately limitations to the explicit simulation of me- reproduces the seasonal distribution and soscale climate processes and to the rep- variability of rainfall over large areas of resentation of topography, land cover, land South America, even though some system- use, and land–sea distribution. This study atic errors persist, (b) the model has been undertook the following key steps in refin- investigated quite thoroughly in various re- ing climate change projections and impact gions in previous downscaling experiences. assessments for major Brazilian crops. Emissions Scenarios Used Regional and Brazilian for this Study Approaches to Selecting Due to resource (funding and time) con- 32 and Using Climate Models straints, this study refined previous work We reviewed the ongoing Latin American by Brazil and the World Bank by using sim- regional effort (Cenarios Regionalizados ilar emissions scenarios and modeling ap- de Clima Futuro da America do Sul (CREAS) proaches. [The Regional Climate Change Scenarios Previous work (Assad and Pinto 2008, for South America], where three RCMs: (1) Margulis and Dubeux, 2010) was based on Eta for Climate Change Simulations—Eta a 2007 baseline, climate risk zone mapping CCS—(Pisnichenko and Tarasova 2009, (2) in 5,000 municipalities for major Brazil- RegCM3 (Seth and Rojas 2003, Pal et al., ian crops, and the agricultural scenarios in 2007) and (3) the public version 3 of the UK Brazil were simulated for the years 2010 Met Office Hadley Centre HadRM3P (Jones (closest representation to the current con- et al. 2004; Alves and Marengo 2009) were ditions), 2020, 2050 and 2070 and for two nested within the public version of the at- IPCC Third Assessment Report scenarios: mospheric global model of the UK Met Of- A2 the most pessimistic, and B2, slightly fice Hadley Centre HadAM3P (Marengo and more optimistic. In scenario A2, the esti- Ambrizzi 2006; Marengo 2009). mated temperature rise variation is be- The CREAS effort aims to provide high tween 2°C and 5.4°C; and in B2, between resolution climate change scenarios in 1.4ºC and 3.8ºC. South America for raising awareness among Based on the previous work in Brazil in government and policy makers in assessing the context of the A2, B2 scenarios, we se- climate change impact, vulnerability and lected the A2 emission (more pessimistic in designing adaptation measures. The ra- with projected temperature rise variation between 2°C and 5.4°C) scenario as most •• INCM3 (INM-CM3.0) – Institute for Nu- closely resembling the estimated increased merical Mathematics – Russia future heterogeneity with continued pop- In addition to the above GCMs, we select- ulation growth. Economic development is ed 3 Regional Climate Models (RCMs) that primarily regionally oriented, the per cap- have already been extensively tested and ita economic growth and technological de- calibrated in Brazil: velopment are more fragmented and slow- •• PRECIS (Providing Regional Climates er when compared with other scenarios for Impact Studies) developed by the (IPCC, 2007). Hadley Center (UK) and was initially de- In this study, we focused on refining cli- nominated HadRM3P. The contour con- mate-agricultural impact assessments for ditions are defined by the projections of the periods 2020-2030 and for the scenar- the model HadRM3P and HadAM3P. The io A2 because these decades are of greatest model is indicated for the South Amer- concern to current investments and policy ica and adjacent ocean conditions. Pre- makers. More importantly, the reliability vious work developed by EMBRAPA and of available data and projection capability CEPAGRI showed an excellent suitability 33 is also greatest for the period to 2030. Be- for temperature projections to 2050 but yond 2030 and based on available climate with problems in simulating rainfall. and other relevant data, projections be- 1. Eta for Climate Change Simulations— come increasingly uncertain. Eta CCS—(Pisnichenko and Tarasova 2009) developed at Belgrade University Climate Change and Agricultural Impact Projections in Brazil to 2030 and Beyond Climate Models Used for and implemented by the National Cen- this Study ter for Environmental Prediction. The Brazilian Center for Weather Forecasts We used the findings of a study by Mace- and Climate Studies (CPTEC) has used do (2011) that evaluated GCMs for Brazil, the Eta model operationally since 1996 and selected the 4 most appropriate GCMs to provide weather forecasts over South using IPCC SRES A2 based on climate (tem- America. Due to its vertical coordinate perature) projection congruence for differ- system, the Eta Model is able to produce ent regions of Brazil. The selected GCMs satisfactory results in regions with steep included: orography such as the Andes range. •• NCCCSM (CCSM3) – National Center for The CPTEC GCM forecasts comparisons atmospheric Research – USA with Eta showed that the model pro- •• GIER (GISS-ER) – NASA Goddard Insti- vided considerable improvement over tute for Space Studies – USA the driver model. The assessment of the •• CSMK3 (CSIRO – Mk 3.0) – Common- Eta Model seasonal forecasts against cli- wealth Scientific and Industrial Re- matology showed that, in general, the search Org – Australia model produced additional useful infor- mation over climatology. The Eta Model biophysical parameters maximum Impacts of Climate Change on Brazilian Agriculture exhibited better results in simulations of stomatal conductivity, leaf area index, upper- and lower level circulation and albedo, roughness, biomass and soil precipitation fields. heat capacity, soil porosity, hydraulic 2. BRAMS Brazilian developments on Re- conductivity and moisture potential gional Atmospheric Modeling System. at saturation and root distribution (Freitas et al., 2009; Longo et al., 2010). associated with the vegetation and The BRAMS model is based on the Re- soil parameterizations of RAMS were gional Atmospheric Modeling System adapted for tropical and sub-tropical - RAMS - with specific parameteriza- biomes and soils, using observations tion for the tropics and sub tropics. The or estimations obtained in recent model has a set of modules to simulate Brazilian field campaigns, mostly as- processes of radioactive transfer, water sociated with the LBA (Large Scale and heat exchange between surface and Biosphere-Atmosphere Experiment atmosphere, microphysics of clouds and in Amazonia – www.lba.cptec.inpe. turbulent transfer in the boundary layer. br) program. 34 a. The BRAMS system is able to incor- c. Overall, the BRAMS model is able to porate aerosol effects on radiation replicate the seasonal cycle of pre- balance and the hydrological cycle cipitation over Brazil with good skill thereby helping to overcome a signif- across most of regions. In Figure x be- icant source of inconsistencies in the low, for regions 1 and 2 the projection rainfall projections. skill is very good, while for regions b. BRAMS has also high resolution and 3, 4 and 5 it is satisfactory. Region 6 updated topography, land use, soil is where the model underestimates type and normalized difference veg- rainfall. etative index (NDVI) data sets. The Figure 5. The BRAMS land-use map for the analysis of model simulation results 35 Model Testing, Climate Once the temperature and precipitation Climate Change and Agricultural Impact Projections in Brazil to 2030 and Beyond simulations were conducted, they were cal- Projections and Model ibrated against hydrometerological data Calibrations from a range of Brazilian agencies (ANA, CPETEC, EMBRAPA, INMET, UNICAMP). After selecting the seven climate models - The AGRIPEMPO hydromet system has a four GCMs and three RCMs - the tempera- network of 1,200 meterological stations ture and precipitation were simulated for to and 4,000 rain gauges nationally (see Fig- 2020 and 2030 with 2010 as the baseline. ure 6 below) with at least 25 years of data The detailed methods for climate simula- records that have been quality checked to tion and the accompanying mathematical 2007. treatment of data are available on request. Figure 6. Spatial distribution of the Brazilian hydromet stations [Source: EMBRAPA] Impacts of Climate Change on Brazilian Agriculture 36 In order to derive the variation of tem- monthly temperatures for each of the tar- perature across the seven climate models get years (2020 and 2030). over time we used the method proposed 1. So for any given spatial coordinate (x, y) by Gleik (1986). Using 2000 as the baseline in a given climate model M the tempera- we derived the difference of the value for ture variation is estimated by: ∆TM M(m,a,x,y) = TMM(m,a,x,y) - TMM(m,2000,x,y) where TM M(m,a,x,y) is the moving av- then for each year/month the ∆TM MAX erage of 11 years for the month m , year and the ∆TM MIN. a for the point (x,y). 3. To obtain the temperature used in the 2. For each hydrometerological station crop impact simulations (TS) for each in the national AGRIMET database we geographical coordinate (x,y) corre- were then able to determine the value of sponding to a hydromet station, TS was ∆TM for each of the climate models and calculated as follows TS(m,a,x,y) = TR(m,x,y) + ∆TMM(m,a,x,y) Where TR(m,x,y) is the real tempera- temperature for a given model M for the ture at the location (x,y) for the month month m and year of interest a. m, and ∆TMM(m,a,x,y) is the variation of To assess the representativeness of the low shows the good congruence of the sim- simulated temperatures Ts with observed ulated Ts across the 7 models (4 GCMs and values at hydrometerological stations, we 3 RCMs) for a hydromet station located in simulated the 2010 temperatures for se- Brasilia. lected stations across Brazil. Figure 7 be- Figure 7. Variation of the temperature estimated by the seven models for the weather station “DFUNBFAL”, located in Brasila, DF, Brazil (lat: -15,79; long: -47,9227). Brasília - 2010 (DFUNBFAL) 28,0 CSMK3 26,0 GIER Temperature (°C) INCM3 24,0 NCCCSM 22,0 PRECIS 20,0 ETA 18,0 BRAMS jan feb mar apr may jun jul aug sep oct nov dec 37 Brasília - 2020 (DFUNBFAL) CSMK3 28,0 GIER Temperature (°C) 26,0 INCM3 24,0 NCCCSM 22,0 PRECIS Climate Change and Agricultural Impact Projections in Brazil to 2030 and Beyond 20,0 ETA 18,0 BRAMS jan feb mar apr may jun jul aug sep oct nov dec Brasília - 2030 (DFUNBFAL) CSMK3 28,0 GIER Temperature (°C) 26,0 INCM3 24,0 NCCCSM 22,0 PRECIS 20,0 ETA 18,0 BRAMS jan feb mar apr may jun jul aug sep oct nov dec In addition to deriving ∆TM we also ob- ed at all other stations tested across Bra- tained the maximum and minimum values zil thereby allowing the development of an for the models (∆TM MAX e ∆TM MIN) that OPTIMISTIC (∆TM MIN) and PESSIMISTIC when averaged over the 7 climate models (∆TM MAX) temperature increase scenari- were consistent for the Brasilia hydromet os 2010-2030. station (Figure 8). The pattern was repeat- Figure 8. Monthly maximum and minimum temperatures estimated from the seven mod- Impacts of Climate Change on Brazilian Agriculture els for the weather station “DFUNBFAL”, located in Brasília, DF, Brazil. (Lat:-15,79; Long:- 47,9227). Brasília - 2010 (DFUNBFAL) 28,0 MAX Temperature (°C) 26,0 MIN 24,0 22,0 20,0 18,0 jan feb mar apr may jun jul aug sep oct nov dec Brasília - 2020 (DFUNBFAL) 28,0 MAX Temperature (°C) 26,0 MIN 24,0 22,0 20,0 38 18,0 jan feb mar apr may jun jul aug sep oct nov dec Brasília - 2030 (DFUNBFAL) 28,0 MAX Temperature (°C) 26,0 MIN 24,0 22,0 20,0 18,0 jan feb mar apr may jun jul aug sep oct nov dec The Agro Climatic Risk and Vulnerability Zoning Model In Brazil, zoning of agricultural risks is a tures at critical phases of crop growth. For public policy since 1996 and each of the example, drought stress at flowering or 5,564 municipalities in the country has grain filling can significantly impact yields. been zoned for suitability of crop cultiva- Excessive rain at harvest time can ruin a tion in terms of at least 80% probability crop. The incidence of extreme tempera- for harvesting an economically viable crop tures can cause the loss of production due yield. The zoning is based on the growth to flower loss in the case of high tempera- phases of each crop (phenology), drought tures or frost by low temperatures. stress, flood risk, and extreme tempera- In 2001, EMBRAPA and UNICAMP de- which guarantees seed germination and es- veloped a simulator to project the agri- pecially flowering and grain filling – factors cultural risks as a function of climate and that are critical to final crop yield. This risk soil. The simulator was then used to pro- must not exceed 20%. duce 500,000 simulated observations for This study is contributing to upgrade the beans, 600,000 for soybean, 400,000 for current Brazilian agricultural zoning sys- rice, 2,500,000 for maize and 450,000 for tem to include future climate scenarios and wheat. These simulations to reflect the dif- projections of climate and once completed ferent soil, plant, climate characteristics of will begin to have an immediate operation- the different municipalities in Brazil result- al and policy impact nationally. ed in an advanced knowledge base of the The Agro Climatic Risk and Vulnerabil- agricultural geography of the country. ity Zoning Model (Assad and Pinto, 2008) In addition to information on crop developed by EMBRAPA and UNICAMP cur- needs, terrain characteristics, soil quality, rently underpins all financial lending to the and weather data, the zoning has been fur- agricultural sector in Brazil. The Central ther fine-tuned to include specific indices Bank of Brazil requires mandatory agricul- 39 of sensitivity of crops to extreme tempera- tural zoning throughout the country for ac- ture and moisture events during critical cess to rural credit and the EMBRAPA/UNI- growth phases of crop growth based on CAMP model indicates “what, where and known agricultural calendars. For example, when” to plant a crop variety according to the crop risk indices are based on agro-me- a zoning system. Three types of zoning are Climate Change and Agricultural Impact Projections in Brazil to 2030 and Beyond teorological water balance, calculated from defined: crop evapotranspiration, which is the sum a. Agro-ecological - uses the data base between leaf transpiration and soil evapo- of soil, topography, climate, and the ration. Each crop has optimal soil moisture current land and environmental legal characteristics for optimal levels of pho- framework. For example, Figure 9 be- tosynthesis, growth, and yield. Critical cli- low is the available land area (in green) mate factors for this process are tempera- for agriculture at municipal scale reso- ture and soil moisture that can be used to lution that can be legally accessed for delineate the area in which any crop could farming. This study used high resolution be produced in Brazil and the associated soils, vegetation, and terrain character- climate related risks. istics data sets and included all restric- By incorporating IPCC global warming tions on the types of land use that can be scenarios, the projected temperature and practiced as mandated in Brazil’s legal any rainfall/soil moisture impacts can be framework to produce this high resolu- introduced in the simulations on the ba- tion map as a baseline for analyzing fu- sis of temperature and moisture risk indi- ture climate change impacts. ces for any given crop. The areas of lowest risk are those where there is water stress, Figure 9. Legally permitted area for agriculture based on land and environmental legal frame- Impacts of Climate Change on Brazilian Agriculture works and landuse restrictions 40 b. Agroclimatic – based simply on climate crop water needs, and crop phenology (see information without evaluating the po- Figure 10). tential crop risk. c. Climatic - uses climate, soil, and crop The Water Needs Index culture by assessing the risk analysis (ISNA) of the Agro Climatic taking into account mainly the informa- tion about rainfall, temperature and wa- Zoning Approach ter balance of derivatives that indicate The basis for the zoning is a crop water sup- the deficiencies and surpluses of water ply (Vulnerability) index based on the ratio for agricultural crops. of actual to maximum evapotranspiration Agro Climatic Zoning integrates crop per crop is used to derive a crop risk and growth models with refined climate sim- suitability zoning. The risk zones set for ulations described above and uses a crop each municipality in the country indicate risk matrix based on a state of the art soil which of the 9 major food and export crops and land quality typology, weather data, that are at least 80 percent likely to pro- vide an economically acceptable harvest. Each crop or variety has a pre-defined set Soil Classification and of climate conditions based on long term research and field observations. The com- Map of the Agro Zoning plete length of a crop cycle is divided into Approach. four phenological (growth) phases (Ini- The soils are classified into three types - tial Development, Vegetative Growth, Re- sandy, medium and clayey – or with low, production and Maturity) where the third medium or high capacity for water reten- phase is normally considered as critical tion capacity respectively. The crop coeffi- mainly due to the high sensitivity of flower- cient (Kc) is defined according to the typical ing to dry spells and/or high temperatures. soil and is a measure of water consumption The length of each phenological phase is for each phase of the crop development. defined by degree-days or heat units. The The ISNA values are based on the rainfall incidence of extreme temperatures can stations and estimated by a specific sowing cause the loss of production due to flow- period produced by the water balance for a er loss in the case of high temperatures or fixed combination of soil type and pheno- frost by low temperatures. logical cycle. 41 Figure 10. Flowchart of components and biophysical, climatic, and plant growth processes used for zoning Meteorological stations Climate Change and Agricultural Impact Projections in Brazil to 2030 and Beyond Average temperature (month and years) Average monthly Sowing temperature date Determination of Evapotranspiration Determination of Potential (ETP) phenological phases Soil Type Daily Rain data Evapotranspiration Duration of Cc LAI or Potential (ETP) phenological phases Kc index Water balance Crop Water Stress Index Climatic Risks by municipality Final ZONING Identifying Cropping Areas in Brazil. In addition to soil moisture, the Impacts of Climate Change on Brazilian Agriculture projected temperatures for 2020, 2030, that are less Vulnerable to and 2050 are also used to refine risk as- Climate Change Impacts sessments. Based on temperature effects to 2020, 2030 c. The major advance of the above ap- vulnerable areas are identified and the area proach relative to the previous studies is quantified. The principles for determining that each low risk agroecological zones climate risk are as follows: are also screened for soil types, steep a. Areas with the least risk are those slopes, legal reserve area, riparian zones that do not have a soil water deficien- (APPs), indigenous areas, and protected cy that results in good germination as areas thereby greatly increasing the pre- well as flowering and grain filling. This cision of the estimates of crop produc- risk should not exceed 20%. The risk is tivity and likely climate impacts. based on an evapotranspiration index of d. For current modeling efforts, the base- the crops. line for the crops planted, area planted, 42 b. Using the above criteria, it is possible to and value of production is the 2009 IBGE assess the risk of planting any crop with- survey (Table 2 below). Table 2. Crops and Area Planted in Brazil (2009) Crop Planted Area (ha) Cotton 814,700 Rice 2,905,700 Sugarcane 8,845,650 Bean Summer Season 1,201,600 Bean Autumn Season 675,000 Maize Summer Season 9,463,200 Maize Autumn Season 4,799,650 Soybean 21,761,800 Rainfed Wheat 2,345,500 The process of integrating the GCM and pessimistic climate scenarios to 2020 and RCM outputs to generate optimistic and 2030 is highlighted below (Figure 11). Figure 11. Representation of the process of handling models to the generation of scenarios. GIER INCM3 NCCCSM CSMK3 Optimistic Scenario Global 2010 Scenarios and Models Generation Pessimistic of scenarios Scenario Moving Average Climatic Output simulated 2020 and DELTA Models temperatures definition Analysis BRAMS Model Regionais 2030 BRAMS Model Model - simulates ETA rarinfall PRECIS BRAMS BRAMS C/ CHUVA To obtain the impact of projected climate data, land, water, and crop characteristics/ 43 change impacts on temperature and pre- requirements based on nationally tested cipitation to 2020 and 2030 on target field tested datasets. SCenAgri can be used crops and pastures, we used EMBRAPA to simulate future agricultural production (C-PTIA’s) Simulator of Agricultural Sce- scenarios based on regional climate projec- Climate Change and Agricultural Impact Projections in Brazil to 2030 and Beyond narios (SCenAgri) that integrates climate tions (Figure 12 below). Figure 12. Example of low and high risk areas for planting corn in Brazil considering the sow- ing date in the first ten days of January, considering the pessimistic scenario. 44 Impacts of Climate Change on Brazilian Agriculture Projected Climate Change Impacts on Area of Crop Suitability to 2020 and 2030 45 T his study assessed the likely impacts of climate change projected by an ensem- ble of GCMs and RCMs on the major grain and pessimistic (temperature increase) scenarios of climate change thereby facil- itating a more nuanced interpretation of (soybean, maize, wheat, and beans) and likely risks to and impacts on the major biofuel (sugarcane) crops as well as cotton Brazilian crops. Table 3 below presents the and pastures. In addition to an innovative results of our climate impact projections on approach that allowed a high resolution suitable area (relative to 2010 baseline) for disaggregation not only of agroecological- major Brazilian grain crops. The values for ly suitable but also legally accessible farm- pasture are estimated decreases in the pro- land, this study also developed optimistic ductivity of pastures. Table 3. Percent change in area at low risk from climate change Impacts of Climate Change on Brazilian Agriculture 2020 2030 Crops Optimistic % Pessimistic % Optimistic % Pessimistic % Cotton -4.6 -4.8 -4.6 -4.9 Rice -10 -7.4 -9.1 -9.9 Sugarcane¹ 107 101 108 91 Soybean -13 -24 -15 -28 Rainfed wheat -41 -15.3 -31.2 -20 Bean (summer season) -54.2 -55.5 -54.5 -57.1 Bean (autumn season) -63.7 -68.4 -65.8 -69.7 Maize (summer season) -12 -19 -13 -22 Maize (autumn season) -6.1 -13 -7.2 -15.3 46 Pasture² -34.4 -37.1 -34.9 -38.3 ¹Sugarcane includes potential (new) areas not just current areas of production ²Pasture value = productivity. For soybean, bean (summer and autumn For sugarcane, we included ‘potentially seasons), maize (summer and autumn suitable areas’ rather than just the current seasons), and cotton the results indicate a area where it is grown, which resulted in significant loss in the low risk area due to a significant increase in areas low risk (or increasing temperature. As expected, more high suitability) suggesting that sugarcane pronounced losses were observed in the is naturally better adapted to cope with pessimistic scenario where the tempera- increasing ambient temperatures. Unlike ture increase is projected to be higher than for sugarcane, however, our simulations in the optimistic scenario. suggest that pasture productivity with be Interestingly, for rice and rain fed wheat, increasingly negatively impacted with in- the pessimistic temperature scenario creasing temperatures. seems to have less severe impacts than in Figures 13-21 (below) show the geo- the pessimistic scenario. This could be due graphical distribution and extent of the to the higher temperatures in the pessi- climate change impacts on “low risk” agri- mistic scenario offsetting damage to these cultural land across Brazil for the crops in crops from cold temperatures and/or frost. Table 3. For example, it is well known that cold tem- peratures can result in flower sterility in rice. Maps of projected climate change impacts on major grain crops, sugarcane, and pastures in Brazil to 2020 and 2030 47 Figure 13. Impact of climate change on area suitable for soybean (2010 – baseline and 2030 Impacts of Climate Change on Brazilian Agriculture optimistic and pessimistic) 48 Figure 14. Projected losses in pasture productivity (%) relative to 2010 baseline under opti- mistic and pessimistic scenarios (2020 & 2030) 49 Maps of projected climate change impacts on major grain crops Figure 15. Impact of climate change on area suitable for beans – summer season (2010 – Impacts of Climate Change on Brazilian Agriculture baseline and 2030 optimistic and pessimistic) 50 Figure 16. Impact of climate change on area suitable for beans – autumn season (2010 – base- line and 2030 optimistic and pessimistic) 51 Maps of projected climate change impacts on major grain crops Figure 17. Impact of climate change on area suitable for wetland rice (2010 – baseline and Impacts of Climate Change on Brazilian Agriculture 2030 optimistic and pessimistic) 52 Figure 18. Impact of climate change on area suitable for sugarcane (2010 baseline and 2030 optimistic and pessimistic) 53 Maps of projected climate change impacts on major grain crops Figure 19. Impact of climate change on area suitable for cotton (2010 baseline and 2030 opti- Impacts of Climate Change on Brazilian Agriculture mistic and pessimistic) 54 Figure 20. Impact of climate change on area suitable for beans – summer season (2010 – baseline and 2030 optimistic and pessimistic) 55 Maps of projected climate change impacts on major grain crops Figure 21. Impact of climate change on area suitable for beans – autumn season (2010 – base- Impacts of Climate Change on Brazilian Agriculture line and 2030 optimistic and pessimistic) 56 Projected climate change impacts on commodity supply and demand and land use dynamics 57 I n order to estimate the economic impacts of the simulated yield effects as a function of different climate change scenarios on the sections highlight the methodology used to integrate the scenarios in the models and also describe the results of four scenari- agricultural sector, ICONE used EMBRAPA’s os simulated: (i) the baseline projections suitable area or yield impact results for (without any climate change impact); (ii) crops and pasture as i-Puts to the Brazil- the pessimistic, (iii) the optimistic, and (iv) ian Land Use Model (BLUM). The following BRAMS without precipitation scenarios. Methodology for the economic simulations of climate change scenarios and projected agricultural impacts. This section describes the methodology cultural impact projections were adapted used to simulate the economic impacts of as i-Puts to the BLUM model and an alloca- climate change scenarios. The Brazilian tion model was then used to distribute the Land Use Model – BLUM was the main tool BLUM outputs across 558 micro-regions used for the simulations. EMBRAPA’s agri- nationally. The Brazilian Land Use is formed by national production (which is Impacts of Climate Change on Brazilian Agriculture regionally projected) and beginning stocks Model – BLUM (again considered only for grains and final BLUM is a one-country, multi-regional, sugarcane-based products) and responds multi-market, dynamic, partial equilibri- to expected profitability of each commod- um economic model for the Brazilian ag- ity, which depends on costs, prices and ricultural sector which comprises two sec- yields. tions: supply and demand and land use. Land allocation for agriculture and The model includes the following prod- livestock is calculated for six regions6, as ucts: soybeans, corn (two crops per year), showed in Figure 22 (below). cotton, rice, dry beans (two crops per •• South (states of Paraná, Santa Catarina, year), sugarcane, wheat, barley, dairy, and and Rio Grande do Sul); livestock (beef, broiler, eggs and pork). •• Southeast (states of São Paulo, Rio de Ja- Commercial forests are considered as ex- neiro, Espírito Santo, and Minas Gerais); ogenous projections. In total, the selected •• Center-West Cerrado (states of Mato 58 products account for 95% of total area Grosso do Sul, Goiás and part of the state used for agricultural production in 2008. of Mato Grosso inside the biomes Cerra- Although second (winter) crops, such as do and Pantanal); corn, dry beans, barley and wheat do not •• Northern Amazon (part of the state of generate additional need for land (they Mato Grosso inside the Amazon biome, are smaller and planted in the same fields Amazonas, Pará, Acre, Amapá, Rondônia, as first season crops, in double cropping and Roraima); areas), their production is accounted for •• Northeast Coast (Alagoas, Ceará, Paraí- in the national supply. ba, Pernambuco, Rio Grande do Norte, and Sergipe); •• Northeast Cerrado (Maranhão, Piauí, To- The supply and demand cantins, and Bahia). projections National supply and demand and regional In the supply and demand section, the de- land use of each product respond to prices. mand is projected at the national level and Consequently, for a given year, equilibrium formed by domestic demand, net trade is obtained by finding a vector of prices that (exports minus imports) and final stocks clears all markets simultaneously. Year by (which are not considered for dairy and year, a sequence of price vectors are found, livestock sectors and sugarcane), which re- which allows the market trajectory to be spond to prices and to exogenous variables such as gross domestic product (GDP), 6 The main criteria to divide the regions were agricultural population and exchange rate. The supply production homogeneity and individualization of biomes with especial relevance for conservation. followed through time. The outputs of the supply and demand sections of the mod- model are: regional land use and change, el, considering that the following identity national production, prices, consumption must be satisfied: and net trade. Beginning stock + Production + Imports Annual production in each region comes = Ending Stock + Consumption + Exports from the product of allocated land and or, considering that Net Trade = Exports yields. National production is the sum of all - Imports: regions’ production, in addition to begin- Beginning stock + Production = Ending ning stocks. This relationship guarantees Stock + Domestic Consumption + Net Trade the interaction between the land use and Figure 22. Regions considered in the Brazilian Land Use Model – BLUM Tropical Forest 59 Grasslands Savannas Savannas Savannas and North Amazon Atlantic Forest Projected climate change impacts on commodity supply and demand Center-West Cerrado Northeast Cerrado Northeast Coast Southeast Atlantic Forest and Grasslands South Source: ICONE, IBGE and UFMG. BLUM also takes into account interactions mestic demand for corn and soybeans. In among the analyzed sectors, and among the case of the soybean complex, the com- one product and its sub-products. For ex- ponents soybean meal and soybean oil are ample, the interaction between the grain parts of the domestic demand for soybeans and livestock sectors is the feed consump- and are determined by the crush demand. tion (basically corn and soybean meal) that Similarly, ethanol and sugar are the compo- comes from the supply of meat, milk and nents of sugarcane demand (Figure 23). eggs, which is one component of the do- Figure 23. Interactions between BLUM sectors Impacts of Climate Change on Brazilian Agriculture Rice Corn Sugarcane Ethanol Cotton Sugar Industry and biodiesel Soybean Drybean Pork oil Soybean Poultry (eggs Soybean meal and chicken) Pasture Beef 60 Source: ICONE The Components of Land al land, the increase in the profitability of one activity will result in an increase in the Use Dynamics share of area dedicated to this activity and The land use dynamics is divided in two ef- reduce in the share of area of competing ac- fects: competition and scale. Intuitively, the tivities. competition effect represents how the dif- The regularity conditions (homogeneity, ferent activities compete for a given amount symmetry and adding up) are imposed so of available land, and the scale effect refers that the elasticity matrices (and associated to the way that the competition among dif- coefficients) are theoretically consistent. ferent activities generates the need for ad- For any set of these coefficients we calcu- ditional land. This need is accommodated late individual, cross impacts, and com- by the expansion of total agricultural area petition among activities. Then, using this over natural vegetation. structure, simulations in BLUM allow the The competition effect is accounted for calculation not only of land allocation, but via a set of equations that allocates a share also land use changes. In other words, the of agricultural area to each crop and pas- conditions allow the identification of the ture in each region as a function of its own exchanged area for each activity, consider- and “cross” price-profitability. It establish- ing the amount of total allocated agricul- es that, for a given amount of agricultur- tural area. In order to ensure coherence of the two components of the return elasticities above mentioned conditions, pasture area of each activity. Considering a ceteris pari- is regionally and endogenously deter- bus condition, the increase in profitability mined, but modeled as the residual of to- of one activity has three effects: increase tal agricultural area minus crop area. In in total agricultural area (through average the context of the Brazilian agriculture, it return), increase in its own share of agri- is particularly relevant to project pasture cultural area and, therefore, reduction in both endogenously and regionally, since it the share of agricultural area of other activ- represents around 77% of total land used ities. For competing activities, cross effects for agricultural production. of profitability on area are negative. Although the competition among activi- As mentioned previously, the elasticities ties may represent regions where the agri- of each crop are the sum of competition and cultural area is stable and near its available scale elasticities. At the same time, regional potential, this is an insufficient analysis for elasticity of land use with respect to total Brazil. Recent Brazilian agricultural trends agricultural returns (total Agland elastici- show that crops, commercial forests and ty) is the sum of the scale elasticities of each 61 pastures all respond to market incentives activity. Therefore, competition elasticities by contributing to an expansion of the to- can be calculated directly after total Agland tal area allocated to agriculture (Nassar et elasticity while total individual elasticities al., 2010a)7. This effect is captured in the were obtained through econometric analy- scale section of the BLUM. This method- sis and literature review. ological improvement is essential to adjust the model skills to the specific reality of the Accounting for Land Use Brazilian agricultural land use dynamics. Dynamics in BLUM Projected climate change impacts on commodity supply and demand The scale effect refers to the equations that define how the returns of agricultural In the BLUM land use section, the area a of activities determine the total land allocated crop i of each region l (l=1,…,6) in year t is to agricultural production. More precisely, defined by the following equation: total land allocated to agriculture is a share (1) of total area available for agriculture, and AT l is the total area available for agricul- this share responds to changes in the aver- tural production in the region l; mlt is the age return of agriculture regionally. share of AT l that is currently used for agri- Scale and competition effects are not cultural production (all crops and pasture), independent. In conjunction, they are the and is the share of the area used by agricul- ture that is dedicated to crop i. ATl is an ex- 7 Nassar, A. M.; Antoniazzi, L. B.; Moreira, M. R.; Chiodi, L.; ogenous variable defined by GIS modeling. Harfuch, L. 2010a. An Allocation Methodology to Assess GHG Emissions Associated with Land Use Change: Final The variable mlt is endogenous to the Report. ICONE, September 2010. Available at . tural market return (profitability) index of smaller than one and reduces the effect of Impacts of Climate Change on Brazilian Agriculture region l (rlt), so the share of area allocated . The opposite occurs when current agri- to agriculture can be defined as: cultural land is smaller than (Alo), increas- ing the land supply elasticity. (2) The rlt is calculated through evidences that indicate which activities most expand where k is a constant parameter; is in the agricultural frontier defined as: the land supply elasticity (with respect to the average return) for region l (results for (4) the Brazilian average is presented in Barr where dli is a weighting vector of defor- et al. 2010). The parameter αlt is positive, estation rate caused by each agricultural higher or lower than one and can be de- activity obtained by satellite imagery and fined as: GIS modeling. We can then calculate the (3) weighting vector dli as follows: where Al0 is the land used for agriculture (5) 62 in a defined base period. When agricultur- al land in period t is close to the base peri- According to Holt (1999) the cross area od, αlt is close to 1 and it does not affect . elasticity of crop i with respect to the re- However if agricultural land in t is larger turn of other crops j can be defined as: than in the base period, the parameter αlt is (6) Which by rearranging terms leads to: (7) The first term on the right hand side of The competition effect of the cross area equation (6) can be defined as the scale ef- elasticity is the last part in the right fect of the cross area elasticity : hand side of equation (6): (8) (9) By analogy, the area elasticity of crop i the scale and competition effects and can related to its own return is also formed by be written as: (10) Where is the scale effect and is Which, from equation (14) and with the land competition component of the some calculation, can be rewritten as: area elasticity of crop i with respect to its own return8. (15) The land competition component can From equation (4), equation (15) can be then be calculated as: rewritten as: (11) (16) The link between the regional land sup- Using equation (15), if the land supply ply elasticity ( ) and the scale effect of elasticity is known, the scale effect of ac- 63 each activity ( ) can be observed. The tivity i can be easily calculated. As a result, land supply elasticity can be defined as: the vector containing all land competition (12) component elasticities represents the diagonal of the competition matrix (one for And, rearranging: each region l). Along with other restrictions (such as the regularity conditions and neg- (13) ative cross elasticities) the diagonal terms Projected climate change impacts on commodity supply and demand The elasticity with respect to the varia- are then used to obtain the cross elastici- tion in return of a given crop i in region l is: ties in the competition matrix, as repre- sented in equation (9). (14) 8 Also explained in Nassar et al. (2009) available at http:// www.iconebrasil.com.br/arquivos/noticia/1872.pdf EMBRAPA’s Agricultural Impact Projections as Impacts of Climate Change on Brazilian Agriculture Inputs to BLUM We used the results for each crop and pas- the models and for each scenario by munic- tures from simulated scenarios by EM- ipality, for 2009, 2020 and 2030. BRAPA as i-Puts in the Brazilian Land Use In the case of BLUM model, we aggre- Model. The baseline for the EMBRAPA pro- gated the results in terms of impacts on ar- jections is the cropped area in 2009 and the eas for each activity into the six Brazilian simulations project the area that will con- regions (BLUM regions). Since EMBRAPA tinue to be suitable for future production used the planted area for crops from the activities. Thus, for each simulated scenar- Municipality Agricultural Survey, IBGE – io there is a set of results for pasture and Brazilian Institute of Geography and Statis- the following crops: rice, cotton, corn (1st tics, we calculated the impacts in percent- and 2nd crop), soybeans, dry beans (1st and age points over the planted area used in 2nd crops), sugarcane and wheat. BLUM (from CONAB – Companhia Brasile- 64 However, in order to adapt EMBRAPA ira de Abastecimento) for 2009. results to BLUM and the micro-regional As an example of the set of data simulat- allocation models, we made some assump- ed by EMBRAPA, Table 4 below shows the tions. The database received was for total results for soybeans. planted area for each activity considered in Table 4. Simulated scenarios for soybeans, aggregated in BLUM regions (in 1,000 ha) Planted BRAMS BRAMS Pessimistic Optimistic   area (- precipitation) (+precipitation) Region 2009 2020 2030 2020 2030 2020 2030 2020 2030 South 8.286 4.626 4.272 6.196 5.826 4.824 4.233 8.285 4.233 Southeast 1.424 1.161 1.156 1.233 1.233 1.162 1.160 1.160 1.160 Center-West Cerrado 7.676 6.676 6.540 7.307 7.296 6.690 6.540 6.540 6.540 North (Amazon) 2.422 2.420 2.420 2.420 2.420 2.420 2.420 2.420 2.420 Northeast Coast 1 0 0 0 0 0 0 0 0 Northeast Cerrado 1.953 1.589 1.247 1.727 1.659 1.589 1.264 1.264 1.264 Brazil 21.762 16.473 15.634 18.883 18.434 16.686 15.617 15.617 15.617 Source: IBGE and EMBRAPA. [Source: EMBRAPA and ICONE] Comparing the results for each scenario and pasture results that had negative im- with the observed planted area in 2009 pacts on area for the scenarios. Some mu- (baseline), the impact of climate change to nicipalities presented positive impacts on 2030 is evident as a reduction of suitable pastureland and sugarcane, due to climate area for soybeans in all scenarios. Impor- change scenarios. In the case of the im- tantly, the most severely impacted region pacts on pastureland, EMBRAPA simulated is the South (a major soybean producing the impacts in terms of percentage change area) where the projected suitable area de- related to a starting point to 2010, 2020 cline is almost 50% by 2030. On average, and 2030 for each climate change scenario. the area that can be used to produce soy- BLUM has a mixed source for pasture area, beans in Brazil reduces by 28% in the sim- which was used to derive impact values as ulated pessimistic and BRAMS (no precipi- a proportion of the EMBRAPA results. Table tation) scenarios in 2030. 5 shows the compiled results for each crop In order to use suitable area projections and each scenario simulated by EMBRAPA as inputs in BLUM we combined all crops and used as inputs in BLUM. Table 5. Simulated planted area for crops and pasture for Brazil (in 1000 ha) 65 BLUM Pessimistic Optimistic BRAMS (-P) BRAMS (+P) 2009 2030 2030 2030 2030 Soybean 21.743 15.634 18.434 15.617 21.588 Corn 1st crop 9.463 7.620 8.361 7.796 9.135 Rice 2.909 2.617 2.640 2.614 2.560 Projected climate change impacts on commodity supply and demand Cotton 843 776 777 776 812 Sugarcane 8.846 16.922 18.419 17.125 11.997 Dry Beans 1st crop 2.894 1.122 1.188 1.137 1.923 Wheat 2.396 1.877 1.614 1.561 0 Corn 2nd crop 4.901 4.064 4.456 4.122 4.500 Dry Beans 2nd crop 1.254 519 587 525 970 Pasture 183.485 183.320 183.489 183.478 162.915 Note: BRAMS (-P) refers to the BRAMS scenario with no precipitation change; BRAMS (+P) includes precipitation changes [Source: EMBRAPA and ICONE] The BRAMS (with precipitation) scenar- sion of the report, the results relating to the io results were found to have an anomaly BRAMS +P observations are omitted. They that was traced to a programming error will be included as soon as the recalculated that has since been rectified but in this ver- values are available in a week or so. Because BLUM is an annual projection mod- have the same impact of that considered on Impacts of Climate Change on Brazilian Agriculture el, the impacts on planted area for 2020 and crops and pasture for each scenario simu- 2030 computed by EMBRAPA were distrib- lated. In other words, we used the share of uted along the period from 2013 to 2030. An each crop on total area used for agriculture assumption was made in order to calculate and its percentage variation for each sce- the impacts of each scenario on land avail- nario in order to calculate the impact over able and suitable for agricultural expan- natural vegetation available and suitable for sion (remaining vegetation). We assumed agriculture, as shown in Table 6 for the pes- that the land available for expansion will simistic and optimistic scenarios. Table 6. Land available and suitable for agricultural expansion for each scenario (1000 ha)i Original BRAMS Pessimistic Optimistic (- precipitation) Database in BLUM 2020 2030 2020 2030 2020 2030 South 2.081 1.788 1.763 1.924 1.898 1.816 1.761 66 Southeast 4.324 4.256 4.272 4.289 4.288 4.276 4.275 Center-West Cerrado 8.872 8.686 8.697 8.815 8.814 8.723 8.698 North Amazon 16.108 15.949 15.997 16.051 16.051 16.049 16.047 Northeast Coast 68 56 56 58 57 57 56 Northeast Cerrado 12.066 11.555 11.474 11.672 11.643 11.605 11.482 Brazil 43.519 42.289 42.258 42.809 42.751 42.525 42.318 Considering only the impacts on the following products: soybeans, corn (1st crop), rice, dry beans i (1st crop), sugarcane and pasture. [Source: ICONE] This assumption is necessary because it is the impacts on crops and pasture are con- unrealistic to expect that total area allocat- sidered together, the impacts are signifi- ed for agriculture will be reduced and there cantly higher, as shown in Table 7. Out of will be no deforestation in areas suitable approximately 230 million hectares used for expansion. for grains (first crop), sugarcane and pas- For the pessimistic scenario, total area ture in 2009, climate change scenarios available for agricultural expansion is pro- could reduce this amount by more than jected to decrease by more than 1 million 10 million hectares in the Pessimistic and hectares. This impact is much lower than BRAMS (without precipitation) scenarios, that presented for crops in Table 5. The ex- while in the optimistic scenario, the area planation is that pasture area considered reduction could amount to 7 million hect- separately will have a much lower impact ares in 2030. on area reduction. On the other hand, when Table 7. Land used in 2009 and potential projected for 2030 for each scenario (in 1000 ha) Baseline Pessimistic Optimistic BRAMS (-P) Region 2009 2030 2030 2030 2030 South 30.281 29.823 25.084 27.031 25.114 Southeast 37.193 37.317 36.784 36.963 36.835 Center-West Cerrado 58.998 59.678 58.698 59.396 58.725 North Amazon 51.629 58.688 58.003 58.054 58.165 Northeast Coast 14.790 14.911 12.672 14.384 12.725 Northeast Cerrado 37.100 38.255 36.871 37.224 36.903 Total 229.990 238.671 228.112 231.640 228.467 Note: Only first crops for corn and dry beans and excluding winter crops (wheat and barley) Source: ICONE In terms of total land available and suitable able for agriculture will be in the South. 67 for agriculture, which is the sum of areas Brazil is likely to have 12.5 and 12.2 million with natural vegetation suitable for pro- hectares less land suitable for agricultural duction and land currently used for these production in the pessimistic and BRAMS activities, as also presented in Table 2, the (-P) scenarios in 2030. For the optimistic South region will be the most affected in scenario, the potential area for agriculture all scenarios. According to Table 8, more could be reduced by 8 million ha compared than 50% of total reduction on land avail- to the original baseline. Projected climate change impacts on commodity supply and demand Table 8. Land available and suitable for agricultural production, comparing scenarios for 2030 (1000 ha) Original Pessimistic Optmistic BRAMS (-P) South 32.362 27.412 29.513 27.380 Southeast 41.517 41.015 41.169 41.044 Center-West Cerrado 67.870 66.535 67.425 66.536 North Amazon 67.737 67.271 67.495 67.480 Northeast Coast 14.859 12.066 12.475 12.128 Northeast Cerrado 49.165 46.753 47.445 46.787 Brazil 273.509 261.053 265.523 261.357 Source: ICONE Based on past trends for land use and espe- 42.2 million hectares of pastureland suit- Impacts of Climate Change on Brazilian Agriculture cially pasture dynamics, it is very likely that able for crop production, where 32% is con- a significant proportion of current pasture centrated in the Center-West Cerrado, 22% land could be converted to cropland un- in the South, 16% in the Southeast, 16% der all climate change scenarios. In BLUM in the North Amazon, 9% in the Northeast projections, for example, beef production Cerrado and 4% in the Northeast Coast, as increases even with less land allocated to shown in Table 9. pasture in the future. Currently, Brazil has Table 9. Pastureland suitable for crop production, comparing scenarios for 2030 (1000 ha) Original Pessimistic Optimistic BRAMS (-P) South 8.528 3.870 5.856 3.841 Southeast 6.043 5.593 5.729 5.618 Center-West Cerrado 12.306 11.134 11.915 11.134 68 North Amazon 5.983 5.850 5.850 5.850 Northeast Coast 1.652 -63 247 -45 Northeast Cerrado 3.547 1.644 2.172 1.669 Brazil 38.060 28.028 31.769 28.067 Source: Sparovek and ICONE Despite the projected reduction in the area The next section presents the prelim- of pastureland highly suitable for crop pro- inary results for three scenarios in BLUM: duction by almost 10 million hectares rela- baseline, pessimistic, optimistic and BRAMS tive to the 2009 baseline, pastureland can (without precipitation). As described above, continue to be converted to crop produc- the dynamic variable in the model for each tion in all scenarios via increased intensifi- scenario in 2020 and 2030 was the land cation of beef production in all the simulat- available and suitable for agriculture, com- ed scenarios. bined with the amount of pastureland that can be converted into cropland. Simulation Results from the Brazilian Land Use Model (BLUM) The results are presented in three sub-sec- results of the pessimistic, optimistic and tions: land use and production, domestic BRAMS (no precipitation) scenarios for consumption, production value, and inter- 2020 and 2030 with the baseline scenario national trade and prices. We compared the (without climate change). Land Use and Production crops and pastureland together, for each scenario in 2009, 2020 and 2030. Table 10 shows the results for land allocat- ed to agricultural production, considering Table 10. Land used by pasture and first season cropsi (1000 ha) Baseline Pessimistic Optimistic BRAMS (-P) Region 2009 2020 2030 2020 2030 2020 2030 2020 2030 South 30.281 29.807 29.823 25.369 25.084 27.353 27.031 25.766 25.114 Southeast 37.193 37.317 37.317 36.650 36.784 36.978 36.963 36.843 36.835 Center-West Cerrado 58.998 59.442 59.678 58.402 58.698 59.165 59.396 58.617 58.725 North Amazon 51.629 55.629 58.688 54.421 58.003 54.486 58.054 54.697 58.165 Northeast Coast 14.790 14.912 14.911 12.772 12.672 13.023 14.384 12.861 12.725 Northeast Cerrado 37.100 37.752 38.255 36.584 36.871 36.800 37.224 36.692 36.903 69 Total 229.990 234.858 238.671 224.198 228.112 227.804 231.640 225.476 228.467 Only first crops for corn and dry beans and excluding winter crops (wheat and barley) i Source: ICONE In the baseline scenario, total area allocat- For the optimistic scenario the impacts ed to crops and pasture in Brazil increases were much lower. The South region re- by 2% in 2020 and 4% in 2030, relative to duced total area used by agricultural pro- 2009. duction by 2.5 and 2.8 million hectares in Projected climate change impacts on commodity supply and demand Comparing the results from the pessi- 2020 and 2030, respectively, relative to the mistic scenario with the baseline for 2020 baseline. Similarly for Brazil as a whole, and 2030, in terms of total area allocated the total area reduction is projected to be to agriculture, the South region is likely around 7.1 million hectares for 2030 com- to be the most affected, due to the climate pared to the baseline. Interestingly, howev- change restrictions for this scenario. In er, most of this reduction was allocated to 2020, total area might be reduced by 4.4 pasture area, as shown in Table 11. This is million hectares compared to the baseline, the result of cattle raising intensification, increasing to 4.7 million hectares in 2030. since there is pastureland with high suit- In general, Brazil might have 10.6 million ability for crops. hectares less land allocated to agriculture in 2030. BRAMS scenario without precipi- tation presented similar results as the pes- simistic scenario. Table 11. Land allocated to pasture (million hectares) Impacts of Climate Change on Brazilian Agriculture Baseline Pessimistic Optimistic BRAMS (-P) Region 2009 2020 2030 2020 2030 2020 2030 2020 2030 South 16.19 14.64 13.79 12.13 11.31 13.24 12.32 12.36 11.33 Southeast 27.47 25.67 24.29 24.96 23.68 25.29 23.89 25.13 23.72 Center-West Cerrado 49.00 45.66 42.78 44.20 41.39 45.08 42.18 44.44 41.42 North Amazon 47.83 51.08 53.62 49.64 52.58 49.81 52.74 49.93 52.74 Northeast Coast 10.85 10.70 10.44 9.11 8.82 9.31 9.05 9.18 8.86 Northeast Cerrado 32.15 30.69 29.44 29.43 27.99 29.74 28.42 29.55 28.02 Total 183.48 178.44 174.36 169.47 165.78 172.46 168.61 170.58 166.09 Source: ICONE The projections show that for Brazil, to- displaced by crops. All other regions, ex- 70 tal pastureland could decrease by 8.6 and cept the North Amazon, also present a de- 8.3 million hectares in the pessimistic and crease trend on pastureland in the baseline BRAMS (no precipitation) scenarios in scenario. 2030, and 5.8 million ha in the optimistic Table 12 shows that total crop area was scenario for the same year, compared to the reduced, but not as substantially as pas- baseline. Despite the high level of reduc- turelands. Again, most of the reduction was tion, in relative terms the impacts were 5% concentrated in the South, since this region for the pessimistic and BRAMS scenarios was the most affected by the climate change and 3% for the optimistic scenario, com- scenarios. That is, it will not be possible pared to the baseline. to displace pasture in the same amount of Regionally, as expected due to climate crop demand for area in the South region, change impacts, the South region was the which requires a regional reallocation of most affected in terms of pastureland dis- production. The Center-West Cerrado and placed by crops. For both pessimistic and the Northeast Cerrado increased crop area BRAMS (no precipitation) scenarios, pas- in the simulated climate change scenarios, tureland reduced by 2.5 million hectares, compared to the baseline. which represent 18% reduction compared to the baseline in 2030. However, even for the baseline scenario area allocated to pas- ture reduced by 2.4 million hectares com- pared to observed pastureland in 2009. This shows the decrease trend of pasture- land in the South region, which have been Table 12. Land allocated to cropsi (1000 ha) Baseline Pessimistic Optimistic BRAMS (-P) Region 2009 2020 2030 2020 2030 2020 2030 2020 2030 South 14.090 15.171 16.034 13.236 13.771 14.116 14.710 13.405 13.783 Southeast 9.727 11.646 13.030 11.687 13.104 11.690 13.071 11.716 13.115 Center-West Cerrado 9.994 13.779 16.901 14.204 17.305 14.088 17.214 14.178 17.302 North Amazon 3.798 4.553 5.065 4.778 5.419 4.677 5.310 4.771 5.429 Northeast Coast 3.945 4.213 4.468 3.661 3.850 3.711 3.921 3.681 3.867 Northeast Cerrado 4.951 7.059 8.810 7.159 8.878 7.061 8.805 7.144 8.880 Total 46.506 56.421 64.308 54.726 62.328 55.343 63.031 54.896 62.376 Only summer season crops corn and dry beans and excluding winter crops (wheat and barley) i Source: ICONE The baseline (in the absence of climate cropped area in all other regions is pro- 71 change) shows that cropland is projected jected to increase thereby partially com- to increase to 17 million hectares in 2030 pensating for the potential climate change compared to observed cropland in 2009. impacts. In essence these land use trends Due to climate change impacts, howev- appear to represent autochthonous ad- er, all the scenarios simulated, result in a aptation strategies – displacement of less reduction of cropland in 2020 and 2030 suitable cropping systems and relocation of compared to the baseline. It is important cropping systems to more favorable areas Projected climate change impacts on commodity supply and demand to note, however, that the displacement of relative to current locations. pastures by grains and sugarcane partially With respect to Brazilian grain produc- compensates for projected cropland losses tion, as shown in Table 13, our simulations hence the lower initial impacts presented project a reduction of around 4.6 million in Table 2. As a result, as presented in Ta- tons in 2030 in the pessimistic and BRAMS ble 12, land allocated to crops is projected (no precipitation) scenarios relative to to decrease by almost 2 million hectares in the baseline. As expected, the optimistic 2030 for the pessimistic and BRAMS sce- scenario projects a reduced impact from narios and 1.3 million ha for the optimistic climate change with production projected scenario. to decline by 2.7 million tons in 2030 com- The cropland simulations (Table 12) pared to the baseline. In general, the pro- present an interesting trend in regional duction declines can be expected to impact land use change dynamics in Brazil. While prices, domestic demand, and net exports cropland is projected to decrease in the of these products. South and Northeast (coastal) regions, Table 13. Grain production, first season crop only* (thousand tons) Impacts of Climate Change on Brazilian Agriculture Baseline Pessimistic Optimistic BRAMS (-P) Region 2009 2020 2030 2020 2030 2020 2030 2020 2030 South 42.160 59.428 67.849 52.159 58.973 55.476 62.687 52.788 58.996 Southeast 14.622 17.900 23.372 18.042 23.775 17.985 23.574 18.082 23.793 Center-West Cerrado 28.853 41.175 50.561 42.905 52.634 42.338 51.979 42.769 52.601 North Amazon 10.323 14.609 18.301 15.461 19.748 15.083 19.286 15.424 19.779 Northeast Coast 2.310 3.197 3.671 2.781 3.178 2.815 3.226 2.795 3.190 Northeast Cerrado 10.222 20.471 30.247 21.091 31.079 20.650 30.557 21.007 31.063 Total 108.492 156.781 194.001 152.440 189.389 154.346 191.310 152.865 189.422 *Only summer crops (corn and dry beans) and excluding winter crops (wheat and barley) Source: ICONE 72 Despite the projected decrease in grain It is especially noteworthy that despite production in the South region by around the projected reduction in the pasture area 8.9 million tons in 2030 under the pessi- of pasture, beef production will decrease mistic scenario relative to the baseline, the by a much lower amount due to technolog- Center-West, North Amazon and Northeast ical intensification as shown in Table 14. So Cerrado regions will increase grain produc- although beef production in Brazil might tion by 4.4 million tons in 2030 under the decrease by 7% in all scenarios simulated same scenario, compared to the baseline. in 2030 compared to the baseline, our sim- That is, regional production re-allocation ulations project that beef production will will reduce the climate change negative im- continually grow until 2030 in all scenar- pacts on grains by almost half. ios compared to the observed production in 2009, and could increase by more than 2 million tons. Table 14. Beef production (thousand tons) Baseline Pessimistic Optimistic BRAMS (-P) Region 2009 2020 2030 2020 2030 2020 2030 2020 2030 South 1.072 1.596 1.942 1.453 1.700 1.492 1.748 1.460 1.700 Southeast 2.483 2.894 3.292 2.823 3.158 2.820 3.144 2.824 3.157 Center-West Cerrado 2.997 4.473 4.927 4.349 4.594 4.367 4.629 4.353 4.597 North Amazon 1.381 1.474 1.891 1.404 1.733 1.403 1.725 1.408 1.736 Northeast Coast 388 532 627 511 588 512 590 512 588 Northeast Cerrado 839 911 1.012 886 954 887 956 887 954 Total 9.161 11.881 13.691 11.426 12.726 11.482 12.793 11.443 12.733 Source: ICONE Domestic Consumption, Prices and International Trade 73 In terms of domestic consumption, Table 15 summarizes the results for each product and scenario analyzed. Table 15. Domestic consumption of each product analyzed (1000 tons and billion liters) Baseline Pessimistic Optimistic BRAMS (-P) Activities 2009 2020 2030 2020 2030 2020 2030 2020 2030 Projected climate change impacts on commodity supply and demand Grains 106.940 143.375 175.286 141.488 173.643 142.386 174.466 141.713 173.665 Ethanol 23.960 40.891 67.599 39.809 65.260 40.211 66.024 39.954 65.321 Soybeans meal 12.000 16.022 18.922 15.826 18.807 15.929 18.898 15.854 18.810 Soybeans oil 4.341 6.783 8.260 6.739 8.220 6.761 8.240 6.745 8.220 Sugar 10.341 14.288 19.055 14.185 18.975 14.225 19.003 14.200 18.977 Beef 7.433 9.400 10.089 8.997 9.250 9.045 9.304 9.012 9.255 Broiler 7.294 10.791 12.088 10.695 12.160 10.770 12.216 10.715 12.161 Pork 2.598 3.017 3.434 3.006 3.401 3.015 3.449 3.009 3.440 Source: ICONE In the absence of climate change, domestic vere reductions of domestic consumption Impacts of Climate Change on Brazilian Agriculture consumption of all commodities is project- compared to the baseline. ed to increase in 2020 and 2030 compared to 2009. However, our simulations across The Projected Real Prices all the climate change scenarios suggest that when compared to the 2009 baseline, of Commodities to 2020 climate change is likely to reduce consump- and 2030 as Impacted By tion of almost all commodities, specially Climate Change grains and ethanol. The main cause of this The real prices of commodities are present- reduction is the higher real prices faced ed in Table 16. Competition among crops by all commodities when land availability and pasture lead to higher prices in the sce- for agricultural production is reduced as a narios with land availability for agriculture function of climate change. The pessimistic restriction. As expected, the pessimistic and BRAMS scenario project the most se- scenario presented the higher impacts on prices than other scenarios. 74 Table 16. Commodities’ real prices (2011=100) Baseline Pessimistic Optimistic BRAMS (-P) Region Unit 2011 2030 2030 2030 2030 Corn R$/ton 395.79 359.29 385.58 374.91 385.08 Soybeans R$/ton 712.41 815.42 865.34 843.37 864.88 Cotton R$/ton 1,667.91 1,415.15 1,454.70 1,437.42 1,453.63 Rice R$/ton 420.10 571.22 671.88 629.49 671.07 Dry Beans R$/ton 1,178.37 1,523.41 1,691.32 1,638.99 1,688.96 Soybean meal R$/ton 568.52 814.73 841.80 832.22 841.65 Soybean oil R$/ton 2,427.65 2,463.22 2,558.26 2,510.82 2,556.95 Wheat R$/ton 420.04 480.60 480.60 480.60 480.60 Barley R$/ton 496.33 368.41 368.41 368.41 368.41 Sugar R$/ton 986.40 343.08 374.25 363.54 373.38 Ethanol R$/liter 1.35 0.66 0.71 0.69 0.71 Beef R$/kg 6.35 9.43 12.09 11.83 12.07 Broiler R$/kg 1.64 2.57 2.80 2.74 2.80 Pork R$/kg 2.13 3.90 4.21 4.12 4.20 Source: ICONE Interestingly, beef producer prices in- Based on the above price projections, cli- creased more than 25% in all scenarios, mate change impacts are likely to lead to showing that intensification of pasture use higher values of production in the climate and cattle production might lead to a price change scenarios due to higher prices and increase in order to compensate for the in- impacts on production (Table 17). So as vestments to increase yields. Costs of pro- production declines in one region, supply duction increase with increasing intensifi- will be lower than demand, prices will in- cation of livestock production. crease and production in other regions will respond positively. It is noteworthy that Projected Production Value beef and soybean oil account for almost 50% of the projected total production val- of Agriculture as Impacted ue for Brazilian agriculture. by Climate Change 2020 and 2030 Table 17. Production Value in R$ million (2011=100) 75 Baseline Pessimistic Optimistic BRAMS (-P) 2009 2020 2030 2020 2030 2020 2030 2020 2030 Corn (total) 16,678 24,675 31,889 26,020 33,717 25,459 33,037 25,854 33,684 Soybeans 47,550 68,639 96,181 73,162 101,625 71,048 99,258 72,620 101,577 Cotton 3,413 8,168 10,047 8,338 10,266 8,263 10,171 8,314 10,260 Rice 7,831 6,812 9,254 7,485 10,131 7,207 9,789 7,427 10,124 Projected climate change impacts on commodity supply and demand Dry Beans (total) 4,562 5,829 9,268 6,517 10,138 6,303 9,870 6,466 10,126 Soybean meal 18,350 20,673 33,022 21,255 33,832 21,009 33,589 21,196 33,830 Soybean oil 71,615 128,603 150,100 132,665 155,060 130,581 152,464 132,171 155,002 Wheat 2,819 4,452 3,600 4,362 3,516 4,182 3,463 4,215 3,452 Barley 86,208 78,166 42,852 77,813 42,658 77,949 42,726 77,862 42,665 Sugar 28,248 29,066 20,906 30,806 22,684 30,142 22,076 30,563 22,634 Ethanol 28,102 55,802 52,179 57,824 54,621 57,053 53,787 57,542 54,553 Beef 51,963 85,111 129,121 93,785 153,907 92,198 151,305 93,306 153,678 Broiler 19,287 30,448 50,568 32,674 54,999 31,942 53,994 32,463 54,938 Pork 6,897 11,862 18,843 12,599 20,298 12,347 19,926 12,526 20,270 Total 393,523 558,304 657,832 585,304 707,454 575,683 695,453 582,526 706,794 Increasing prices also explain domestic Climate change scenarios increased prices Impacts of Climate Change on Brazilian Agriculture consumption decrease, as showed in Ta- and, consequently, reduced the demand. ble 15, and also on net trade (Table 18). Table 18. Net trade results for each scenario (1,000 tons and billion liters for ethanol) Baseline Pessimistic Optimistic BRAMS (-P) Region 2009 2020 2030 2020 2030 2020 2030 2020 2030 Grains 30,471 50,218 68,654 48,629 67,044 49,013 67,556 48,582 66,932 Ethanol 2,897 9,944 11,983 9,944 11,855 9,879 11,924 9,806 11,859 Soybeans meal 12,210 17,569 21,722 17,382 21,540 17,474 21,622 17,304 21,384 Soybeans oil 1,579 1,613 1,885 1,559 1,834 1,585 1,856 1,576 1,862 Sugar 24,088 32,674 41,814 32,370 41,588 32,553 41,710 32,415 41,595 Beef 1,728 2,517 3,549 2,428 3,477 2,510 3,536 2,431 3,478 76 Broiler 3,635 5,570 7,479 5,570 7,479 5,543 7,409 5,544 7,481 Pork 592 2,598 3,439 2,598 3,439 913 1,378 913 1,386 Source: ICONE Net trade had lower effects than the domes- tons in 2030, relative to the baseline, for all tic consumption, but grain exports were simulated scenarios. the most affected by more than 1 million Conclusions 77 T he value added of this study relative to the other studies carried out in the region over the last decade can be summa- al impacts, this study used a combination of global and higher resolution regional models and long term hydrometerolog- rized as follows: ical and land use data sets to improve 1. The study was conceptualized in the calibration of the climate model out- context of an on-going national-region- puts. The integration of the different cli- al-global Cenarios Regionalizados de Cli- mate models and data sources allowed ma Futuro da America do Sul (CREAS) a more refined analysis and synthesis at effort to improve robustness of climate sub-national scales and also allowed the change projections and likely impacts identification of key data (e.g. hydromet on agriculture. We used some of the density) gaps. same Global and Regional Climate Mod- 3. The implementation of this study re- els (GCMs and RCMs) used in CREAS so quired active collaboration among re- the results from this study will both con- searchers, agronomists, and professors tribute to and benefit from the on-going and students from leading Brazilian CREAS effort. national agencies (EMBRAPA-Agri- 2. Whereas previous studies had used a culture, UNICAMP-climate modeling, single Global Climate Model and single INPE-mesoscale weather and spatial RCM to project national and sub-nation- modeling, ICONE-economic modeling). The network of professionals can now the previous study by Assad and Pinto Impacts of Climate Change on Brazilian Agriculture continue to improve and refine the in- (2008) that used one GCM and RCM and tegrated agroecological, biophysical, projected substantial negative impacts and economic modeling and analysis to soybean, wheat, maize, and pasture developed for this study. The inclusion systems, this study using a range of of UNICAMP also lays the foundation for GCMs and RCMs and significantly better capacity building of the next generation hydrometerological and land suitability of climate modelers in Brazil and the data, showed that while for some crops LCR region. (soybean and cotton) the projected cli- 4. This study assessed the vulnerability mate impacts are likely to be more mod- and impacts of climate change on Bra- erate, for other crops (beans and corn) zilian agriculture by building on valu- the impacts could be significantly more able work done in the last decade in severe that than projected in the 2008 Brazil and in the LAC region. The results study. The Table below highlights for from this study confirm and extend the 2020, these differences and illustrates, findings of pervious work that climate at least partially, the value of harnessing 78 change is likely to have increasingly sig- more robust climate, land, water, and cli- nificant and mostly negative impacts on mate data sets for more nuanced analyt- the major grain and pasture systems in ical power of climate change modeling Brazil. For example, in comparison with approaches. 2020 COMPARISON PRECIS model by Assad and Pinto Series of GCM & RCMs low risk area (%) (2008) Optimistic Pessimistic Optimistic Pessimistic Cotton -11.4 -11.7 -4.6 -4.8 Rice -8.41 -9.7 -9.9 -7.4 Sugarcane 170.9 159.7 107 101 Soybean -21.62 -23.59 -13 -24 Bean (summer season) -54.3 -55.5 -4.3 -4.3 Bean (autumm season) -63.7 -68.4 Maize (summer season) -12 -19 -4.3 -4.3 Maize (autumm season) -6.1 -13 5. Coupling the above climate impact on Model (BLUM), revealed the following agriculture data with an econometric likely outcomes at sub-regional scales simulation tool – the Brazilian Land Use and geographic locations: a. In the absence of climate change, by 2.7 million tons in 2030 compared cropland is projected to increase to to the baseline. 17 million hectares in 2030 com- e. Despite the projected decrease in pared to observed area of cropland in grain production in the South region 2009. Due to climate change impacts, by around 8.9 million tons in 2030 however, all the scenarios simulated, under the pessimistic scenario rela- result in a reduction of cropland in tive to the baseline, the Center-West, 2020 and 2030. Northeast Cerrado and North Ama- b. In the pessimistic scenario Brazil zon regions are projected to increase could have 10.6 million hectares less grain production by 4.4 million tons land allocated to agriculture in 2030 in 2030 under the same scenario, as a result of climate change with the compared to the baseline. That is, re- South Region being the worst im- gional production re-allocation will pacted losing close to 5 million ha by reduce the climate change negative 2030. impacts on grains by almost half. c. It is important to note, however, f. Although the pasture area is project- 79 that the displacement of pastures by ed to be reduced, beef production grains and sugarcane partially com- is projected to decrease by a much pensates for the projected cropland lower amount than the pasture area and grain losses. Framers and the due to technological intensification. market will partially drive adaptation Pasture productivity in Brazil might to loss of suitable crop land due to cli- decrease by 7% in all scenarios sim- mate change via displacement of cur- ulated to 2030, but our simulations rent, poorly producing pastures with project that compared to the 2009 grain crops and sugarcane. The pro- baseline, beef production may con- jections suggest that there could also tinue to increase until 2030 in all sce- be a regional relocation with some narios, and could increase by more of the grain crops moving out of the than 2 million tons. south to the central regions of Brazil. g. Beef producer prices are projected to d. With respect to Brazilian grain pro- increase by more than 25% in all sce- duction, our simulations project a narios, showing that intensification reduction of around 4.6 million tons of pasture use and cattle production in 2030 in the pessimistic scenarios might lead to a price increase in order relative to the baseline. As expected, to compensate for the investments to the optimistic scenario projects a re- increase yields. duced impact from climate change h. In general, the production declines Conclusions with production projected to decline can be expected to impact prices, do- mestic demand, and net exports of these products. In the absence of cli- expanded and enhanced access to irriga- Impacts of Climate Change on Brazilian Agriculture mate change, domestic consumption tion, improved land and water manage- of all commodities is projected to in- ment) as adaptation measures to coun- crease in 2020 and 2030 compared teract the projected negative impacts of to 2009. However, our simulations climate change on agricultural produc- across all the climate change scenari- tivity. For example, os suggest that when compared to the a. The Brazilian Government and the 2009 baseline, climate change is like- private sector have been steadily fa- ly to reduce consumption of almost cilitating the adoption of improved all commodities, specially grains and conservation agriculture practices, ethanol. The main cause of this reduc- such as no-till planting, and more tion is the higher real prices faced by resource-efficient systems, such as all commodities when land availabil- integrated crop-livestock systems ity for agricultural production is re- that are inherently more resilient to duced as a function of climate change. climate shocks than some intensive The pessimistic and BRAMS scenario cropping systems. 80 project the most severe reductions of b. The Government is providing credit domestic consumption compared to and financing for the newly-launched the baseline. “Low Carbon Agriculture” program i. Our estimates show that unlike pre- with approximately US$ 1 billion vious estimates of declining agricul- available for low interest credit in the tural production value, the negative 2011 season alone. impacts on supply of agricultural 7. In our study, our efforts to access the commodities is expected to result in latest available hydrometerological and significantly increased prices for land use data significantly improved our some commodities, especially staples ability to undertake more robust mod- like rice, beans, and all meat prod- eling and impact projections. Neverthe- ucts. This will counter the effect of less, the lack of good quality and long declining productivity on value of ag- term climate data is hampering regional ricultural production but could have and local climate modeling efforts as well major negative effects on the poor as the calibration and validation of cur- and their consumption of these staple rent projections that are being used to products. It is noteworthy that beef inform policy and investment decisions and soybean oil account for almost to 2050 and beyond. Because the climate 50% of the projected total production forcing factors operate both within and value for Brazilian agriculture. external to national frontiers, there is an 6. It is important to state that our study urgent need for coordinated and target- did not simulate the potential impact of ed climate change investments over the technological advances (new varieties, next 1-5 years for instrumentation, data variety that is heat and/or drought tol- assembly, data sharing and data access erant. systems. National, bilateral, and multi- 9. The findings of this study will be in- lateral investments agencies need to co- corporated in the EMBRAPA/UNICAMP ordinate their investment strategies to Agroecozone Model to improve the sim- support this specific and urgent need. ulation and climate impact projections 8. The need for improved and integrated that underpin the national agricultural climate change impact assessments is credit and insurance programs in Brazil. especially urgent for the agricultural This means that the outputs of the study sector. A recent survey carried out by the will begin having immediate and far Brazilian Enterprise for Agriculture and reaching operational and policy impli- Animal Research (EMBRAPA), revealed cations in Brazil. The experiences from that even with advanced breeding tech- Brazil are highly relevant for other re- niques, it takes approximately 10 years gions and countries where similar work of R&D (including 2-3 years of scaling is on-going and could both enrich and up and distribution of seed) and costs in benefit from other regional experiences 81 the range of US$6-7 million to develop, via south-south exchange programs. test, and release a new crop cultivar or Conclusions 82 Impacts of Climate Change on Brazilian Agriculture Bibliography 83 Anderson, K and E. Reis. (2007). The Ef- Costa, C. et al. Evolução das pastagens cul- fects of Climate Change on Brazilian tivadas e do efetivo bovino no Brasil. Agricultural Profitability and Land Arq. Bras. Med. Veter. Zoot., v. 15, n. 1, Use: Cross-Sectional Model with Cen- p. 8-17, 2008. sus Data. Final report to WHRC/IPAM De Gouvello, C. (2010) Brazil Low-carbon for LBA project Global Warming, Land Country Case Study. The World Bank, Use, and Land Cover Changes in Brazil. Washington, DC. Assad, E and H. Pinto. 2008. Global warm- Decker, W.L., V. Jones, and R. Achtuni. ing and future scenarios for Brazilian (1986). The Impact of Climate Change Agriculture. EMBRAPA and CEPAGRI/ from Increased Atmospheric Carbon UNICAMP. - www.climaeagricultura. Dioxide on American Agriculture. org.br EMBRAPA and CEPAGRI/ UNI- DOE/NBB-0077. Washington, DC: U.S. CAMP. Department of Energy. Cline, W.R. (2007), Global Warming and Dêschenes, Olivier and Michael Greenstone Agriculture: Impact Assessment by (2007). “The Economic Impacts of Country, Washington DC: Centre for Climate Change: Evidence from Agri- Global Development and Peterson In- cultural Output and Random Fluctua- stitute for International Economics. tions in Weather”. American Economic Review, 97(1): 354-85. Doorenbos, J.; Kassan, A.H. Yield response Golub, A. T.W. Hertel, H. Lee, S. Rose and B. Impacts of Climate Change on Brazilian Agriculture to water. Rome: FAO, 1979. 193p. (Ir- Sohngen (2009), “The Opportunity rigation and Drainage Paper, 33) Cost of Land Use and the Global Po- Evenson, R.E. & D.C.O. Alves (1998). Tech- tential for Greenhouse Gas Mitigation nology, climate change, productivity in Agriculture and Forestry”, Resource and land use in Brazilian agriculture. and Energy Economics 31(4): 299- Planejamento e Políticas Públicas, 18. 319, November. Fernandes, ECM, Soliman, A., Confalonieri, Grell, G. A. and Dezso Devenyi. A general- R., Donatelli, M, and Tubiello, F. (2011) ized approach to parameterizing con- Climate change and agriculture in Lat- vection combining ensemble and data in America (2020-2050). The World assimilation techniques. Geophys. Bank, Washington, DC. Res. Lett., 29, 14, 2002. Freitas, S. R., Longo, K. M., Silva Dias, M. A. IPCC (2001). Climate change 2001: Synthe- F., Chatfield, R., Silva Dias, P., Artaxo, sis Report. Summary for Policymak- P., Andreae, M. O., Grell, G., Rodrigues, ers. Approved in detail at IPCC Plena- 84 L. F., Fazenda, A., and Panetta, J.: The ry XVIII (Wembley, United Kingdom, Coupled Aerosol and Tracer Transport 24-29 September 2001). World Me- model to the Brazilian developments teorological Organization and United on the Regional Atmospheric Model- Nations Environmental Programme. ing System (CATT-BRAMS) – Part 1: Kjellstrom, T., R.S. Kovats, S.J. Lloyd, T. Holt Model description and evaluation, At- and R.S.J. Tol (2009), “The Direct Im- mos. Chem. Phys., 9, 2843-2861, 2009. pact of Climate Change on Regional Frizzone, J.A. & Olitta, A.F.L. Efeitos da su- Labour Productivity”, Archives of En- pressão de água em diferentes fases vironmental and Occupational Health do crescimento e na produção do tri- 64(4): 217-27, December go. Engenharia Rural, Piracicaba, v.1, Lobell, D. B., M. Banziger, C. Magorokosho n.1, p.23-36; 1990. and B. Vivek (2011), ‘Nonlinear Heat Giorgi, F., B. Hewitson, J. Christensen, M. Effects on African maize as Evidenced Hulme, H. von Storch, P.Whetton, R. by Historical Yield Trials’, Nature Cli- Jones, L. Mearns and C. Fu, 2001. Re- mate Change 1: 42-45, March. gional climate information – evalua- Longo, K., S. R Freitas, M. Andreae, A. Setzer, tion and projections. In: J.T. Houghton, E. Prins and P. Artaxo . The Coupled Y. Ding, D.J. Griggs, M. Noguer, P.J. van Aerosol and Tracer Transport mod- der Linden, X. Dai, K. el to the Brazilian developments on Gleick, P.H., 1986. Methods for evaluating the Regional Atmospheric Modeling the regional hydrologic impacts of System (CATT-BRAMS). Part 2: Mod- global climatic changes. Journal of Hy- el sensitivity to the biomass burning drology, 88(1-2), pp.97–116. inventories. Atmos. Chem. Phys., 2010 Nassar, A. M.; Antoniazzi, L. B.; Moreira, M. (in press). R.; Chiodi, L.; Harfuch, L. 2010a. An Al- Margulis, S. and Dubeux, S.B.C. (2010) The location Methodology to Assess GHG Economy of Climate Change in Brazil: Emissions Associated with Land Use Costs and Opportunites. Change: Final Report. ICONE, Septem- Margulis, S. e Dubeaux, C. B. S. The Econom- ber 2010. http://www.iconebrasil. ics of Climate Change in Brazil: Costs org.br/arquivos/noticia/2107.pdf and Opportunities. Coord. Marcovitch, Nelson, G.C., M.W. Rosegrant, A. Palazzo, J. 84 p. 2011. FEA/USP. Embaixada I. Gray, C. Ingersoll, R. Robertson, S. Britânica. Tokgoz, T. Zhu, T.B. Sulser, C. Ringler, Marin, F. R. ; Pandorfi, H.; Ferreira, A. S. . Es- S. Msangi and L. You (2010), Food Se- timativa das temperaturas máximas, curity, Farming, and Climate Change mínimas e médias para o Brasil . In: to 2050: Scenarios, Results, Policy Congresso Brasileiro de Agrometeo- Options, Washington DC: Internation- rologia, 2003, Santa Maria, RS. Anais al Food Policy Research Institute, De- do Congresso Brasileiro de Agromete- cember. 85 orologia, 2003. v. 2. p. 761-762. Pinto, Hilton S. and Assad, Eduardo D. Maskell and C.A. Johnson (eds.). pp. 583– Aquecimento Global e a Nova Geogra- 638. Climate Change 2001: The Sci- fia da Produção Agrícola no Brasil. entific Basis. Contribution of Working Embaixada Britânica do Brasil. 82 pp. Group I to the Third Assessment Re- 2008. port of the Intergovernmental Panel Pinto, Hilton S. and Assad, Eduardo D. on Climate Change. Cambridge Uni- (2008) Global Warming and the New versity Press. Geography of Agricultural Production Mendelsohn, R., W. Nordhaus, e D. Shaw in Brazil. http://www.cpa.unicamp.br (1994). The Impact of Global Warming /aquecimento_ agricola_en.html on Agriculture: A Ricardian Analysis. Sanghi, A., D. Alves, R. Evenson, and R. American Economic Review. 84(4): Mendelsohn (1997). Global warming 753-71 impacts on Brazilian agriculture: esti- Mendelsohn, Robert, William D. Nordhaus mates of the Ricardian model. Econo- and Daigee Shaw (1999). “The Impact mia Aplicada, v.1,n.1,1997. of Climate Variation on US Agricul- Santos, M. E. R. et al. Produção de bovinos ture”. In The Impact of Climate Change em pastagens de capim-braquiária on the United States Economy, ed. diferidas. R. Bras. Zootec.[online]. Robert Mendelsohn and James E. Neu- 2009, vol.38, n.4, pp. 635-642. ISSN mann, 55-74. Cambridge: Cambridge 1806-9290. Bibliography University Press. Schlehuber, A.M.; Tucker, B.B. Culture of wheat. In: QUISENBERRY, K.S.; REITS, L.P., (Ed.). Wheat and wheat improve- Center for International Economic Impacts of Climate Change on Brazilian Agriculture ment. Madison: Am. Soc. Of Agronomy, Studies, University of Adelaide. April. 1967. p.154-160. van der Mensbrugghe, D. and R. Roson Siqueira, O.J.F. de, J.R.B. de Farias, and (2010), “Climate, Trade and Develop- L.M.A. Sans (1994). Potential effects of ment”, Paper presented at the 13th global climate change for Brazilian ag- Global Economic Analysis Conference, riculture, and adaptive strategies for Penang, 9-11 June. wheat, maize, and soybeans. Revista Walko R., Band L., Baron J., Kittel F., Lam- Brasileira de Agrometeorologia, Santa mers R., Lee T., Ojima D., Pielke R., Tay- Maria, v.2 pp. 115-129. lor C., Tague C., Tremback C., Vidale P. Valenzuela, E. and Anderson, K. (2011) Coupled Atmosphere-Biophysics-Hy- Climate Change and Food Security to drology Models for Environmental 2030: A Global Economy-wide Per- Modeling. J. Appl. Meteorol. 39: (6) spective. Discussion paper No. 1102. 931-944, 2000. 86 Acknowledgment This work was funded by the Program on Forests (PROFOR), a multi-donor partnership managed by a core team at the World Bank. PROFOR finances forest-related analysis and processes that support the following goals: improving people’s livelihoods through better management of forests and trees; enhancing forest governance and law enforcement; financing sustainable forest management; and coordinating forest policy across sectors. PROFOR’s donors include the European Commission, Finland, Germany, Italy, Japan, the Netherlands, Switzerland, the United Kingdom and the World Bank. Learn more at www.profor.info. The World Bank 1818 H Street, NW, Washington, DC 20433, USA. www.worldbank.org