92704 v3 Latin America and the Caribbean Turn Down Heat the Confronting the New Climate Normal Latin America and the Caribbean Turn Down Heat the Confronting the New Climate Normal Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal © 2014 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved 1 2 3 4 17 16 15 14 This work was prepared for The World Bank by the Potsdam Institute for Climate Impact Research and Climate Analytics. The find- ings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Ex- ecutive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this commissioned 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 endorsement or acceptance of such boundaries. This is an extract from the World Bank Group’s “Turn Down the Heat Report – Confronting the New Climate Normal” with a focus on the Latin America and the Caribbean Region. For full bibliographic references please refer to the complete report. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions This work is available under the Creative Commons Attribution—NonCommercial—NoDerivatives 3.0 IGO license (CC BY-NC-ND 3.0 IGO) http://creativecommons.org/licenses/by-nc-nd/3.0/igo. Under the Creative Commons—NonCommercial—NoDerivatives license, you are free to copy, distribute, and transmit this work, for noncommercial purposes only, under the following conditions: Attribution—Please cite the work as follows: World Bank. 2014. Turn Down the Heat: Confronting the New Climate Normal. Washing- ton, DC: World Bank. License: Creative Commons Attribution—NonCommercial—NoDerivatives 3.0 IGO (CC BY-NC-ND 3.0 IGO). Noncommercial—You may not use this work for commercial purposes. No Derivative Works—You may not alter, transform, or build upon this work. Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank therefore does not warrant that the use of any third-party-owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to re-use a component of the work, it is your responsibility to determine whether permission is needed for that re-use and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to the Publishing and Knowledge Division, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. The following items from the full report are used with permission and require further permission for reuse. Please refer to the cap- tion or note corresponding to each item: Figures 2.2, 2.4, 2.9, 3.10, 3.14, 3.15, 3.21, 4.13, 4.14, 4.19, 4.21, 4.22, 5.11, 5.12, 5.13, 5.14, 5.15, 5.16, 5.17, 5.18, 5.19, 5.20, 5.21. ISBN: 978-1-4648-0437-3 Cover photos: photos 1, 2, 3, 5, and 7 © The World Bank Group; photo 4 (forestry), © istockphoto, used with permission, further permission for reuse; photos 6 and 8, © Erick Fernandes (floating houses in Peru and jaguar in Amazon)/The World Bank Group. Cover design: Gregory Wlosinski/General Services Department—Printing and Multimedia, The World Bank Group. B Contents Latin America and the Caribbean 13 1 Regional Summary 13 1.1 Regional Patterns of Climate Change 14 1.2 Regional Sea-Level Rise 15 1.3 Sector-based and Thematic Impacts 15 1.4 Overview of Regional Development Narratives 18 2 Introduction 18  ocial, Economic and Demographic Profile of the Latin America 2.1 S and Caribbean Region 20  ulnerabilities to Climate Change in the Latin America 2.2 V and Caribbean Region 20 2.3 Vulnerabilities Faced by Rural Populations 20  rban Settlements and Marginalized Populations 2.4 U 21 3 Regional Patterns of Climate Change 26 3.1 Projected Temperature Changes 26 3.2 Heat Extremes 26 3.3 Regional Precipitation Projections 29 3.4 Extreme Precipitation and Droughts 29 3.5 Aridity 31 3.6 Tropical Cyclones/Hurricanes 32 3.7 Regional Sea-level Rise 35 4 Regional Impacts 37 4.1 Glacial Retreat and Snowpack Changes 37 4.2 Water Resources, Water Security, and Floods 41 4.3 Climate Change Impacts on Agriculture 47 4.4 Climate Change Impacts on Biodiversity 52 4.5 Amazon Rainforest Dieback and Tipping Point 54 4.6 Fisheries and Coral Reefs 57 4.7 Human Health 62 4.8 Migration 64 4.9 Human Security 66 4.10 Coastal Infrastructure 67 4.11 Energy Systems 69 5 Regional Development Narratives 74 5.1 Overarching Development Narratives 74 5.2 Sub-regional Development Narratives 78 6 Synthesis Table – Latin America and the Caribbean 81 C Figures Figure 1.1: Multi-model mean temperature anomaly for Latin America and the Caribbean for RCP2.6 (2°C world, left) and RCP8.5 (4°C world, right) for the austral summer months (DJF) 14 Figure 1.2: Multi-model mean of the percentage change in the aridity index 15 Figure 1.4: Temperature projections for the Latin American and Caribbean land area 27 Figure 1.5: Multi-model mean temperature anomaly for Latin America and the Caribbean 28 Figure 1.6: Multi-model mean of the percentage of austral summer months (DJF) in the time period 2071-2099 with temperatures greater than 3-sigma (top row) and 5-sigma (bottom row) 29 Figure 1.7: Multi-model mean and individual models of the percentage of Latin American and Caribbean land area warmer than 3-sigma (top) and 5-sigma (bottom) 30 Figure 1.8: Multi-model mean of the percentage change in austral summer (DJF, top), winter (JJA, middle) and annual (bottom) precipitation 32 Figure 1.9: Multi-model mean of the percentage change in the annual-mean of monthly potential evapotranspiration for RCP2.6 (2°C world, left) and RCP8.5 (4°C world, right) for Latin America and the Caribbean by 2071-2099 relative to 1951-1980. 33 Multi-model mean of the percentage change in the aridity index Figure 1.10:  35 Change in average rate of occurrence of Category 4 and 5 Figure 1.11:  tropical cyclones per hurricane season (August-October) at about 2.5°C warming globally above pre-industrial levels by the end of the 21st century compared to the present-day 36 Patterns of regional sea-level rise Figure 1.12:  37 Regional anomaly pattern and its contributions in the Figure 1.13:  median RCP8.5 scenario 38 Sea level projections for selected cities Figure 1.14:  39 Compilation of mean annual area loss rates for different Figure 1.15:  time periods for glaciated areas between Venezuela and Bolivia 40 Ice loss from outlet glaciers on the Patagonian Ice Field Figure 1.16:  in southern South America since the Little Ice Age 41 Cumulative regional surface mass balance relative to the Figure 1.17:  1986-2005 mean from the model forced with CMIP5 projections up to the year 2100. SLE = Sea-level equivalent 44 Changes in seasonal total runoff in 4 IPCC climate-change Figure 1.18:  scenarios with respect to the 1961-1990 mean monthly runoff 50 Aggregate impacts on crop yields in the LAC region with Figure 1.19:  adaptation, computed by the AZS-BioMA platform under 2020 and 2050 NCAR GCM for A1B scenario. 50 Meta-analysis of crop yield reductions Figure 1.20:  56 Simulated precipitation changes in Eastern Amazonia from Figure 1.21:  the 24 IPCC-AR4 GCMs with regional warming levels of 2-4.5 K (left panel). Simulated changes in biomass from LPJmL forced by the 24 IPCC-AR4 climate scenarios assuming strong CO2 fertilization effects (middle panel, CLIM+CO2) and no CO2 fertilization effects (CLIM only, right panel) 60 Change in maximum catch potential for Latin American Figure 1.22:  and Caribbean waters 75 D Figure 1.23: Sub-regional risks for development in Latin America and the Caribbean (LAC) under 4°C warming in 2100 compared to pre-industrial temperatures. 111 Tables Table 1.1:  Basic Socioeconomic Indicators of LAC Countries 19 Table 1.2:  Total Population and Indigenous Population Census 2000 23 Table 1.3:  Percentage of Latin American and Caribbean Population Living in Urban Areas and Below Five Meters of Elevation 25 Table 1.4:  Multi-model mean of the percentage of land area in Latin America and the Caribbean which is classified as hyper arid, arid, semi-arid and sub-humid 33 Table 1.5:  Sea-level rise between 1986-2005 and 2081-2100 for the RCP2.6 (1.5°C world) and RCP8.5 (4°C world) in selected locations of the LAC region 35 Table 1.6:  Projected Changes in Yields and Productivity Induced by Climate Change 49 Table 1.7:  Summary of Crop Yield Responses to Climate Change, Adaptation Measures, and CO2 Fertilization 50 Table 1.8:  Projected losses from sea-level rise 68 Table 1.9:  Cumulative loss for the period 2020-2025 for Latin American and Caribbean sub-regions exposed to tropical cyclones 69 Table 1.10: Electricity production from hydroelectric and thermoelectric sources 70 Table 1.11: Projected temperature and hydrologic changes in the Rio Lempa River 71 Table 1.12: Maximum hydropower energy potential 72 Table 1.13: Climate change-related stressors projected to affect hydroelectricity generation 73 Table 1.14: Natural gas production for LAC countries in 2012 and oil production in 2013 73 Table 1.15: Synthesis table of climate change impacts in LAC under different warming levels 81 Boxes Box 1.1: Hurricane Mitch’s Impact in Urban Areas 21 Box 1.2: The Case of Mexico City 22 Box 1.3: Water Security in the Mexico City Metropolitan Area 42 Box 1.4: Glacial Lake Outbursts 43 Box 1.5: Water Security in Quito, La Paz, Bogotá, and Lima 44 Box 1.6: Water from the Cordillera Blanca 44 Box 1.7: Water Security in the Central Andes 45 Box 1.8: Water Security and Glacial melt in La Paz and El Alto, Bolivia 45 Box 1.9: Surface Ozone Concentrations 48 Critical Ecosystem Services of High Andean Mountain Ecosystems Box 1.10:  52 Freshwater Fisheries – Vulnerability Factors to Climate Change Box 1.11:  59 Distress Migration during Hurricane Mitch Box 1.12:  66 E Foreword from the Full Report Dramatic climate changes and weather extremes are already affecting millions of people around the world, damaging crops and coastlines and putting water security at risk. Across the three regions studied in the report, record-breaking temperatures are occurring more fre- quently, rainfall has increased in intensity in some places, while drought-prone regions like the Mediter- ranean are getting dryer. A significant increase in tropical North Atlantic cyclone activity is affecting the Caribbean and Central America. There is growing evidence that warming close to 1.5°C above pre-industrial levels is locked-in to the Earth’s atmospheric system due to past and predicted emissions of greenhouse gases, and climate change impacts such as extreme heat events may now be unavoidable. As the planet warms, climatic conditions, heat and other weather extremes which occur once in hundreds of years, if ever, and considered highly unusual or unprecedented today would become the “new climate normal” as we approach 4°C—a frightening world of increased risks and global instability. The consequences for development would be severe as crop yields decline, water resources change, diseases move into new ranges, and sea levels rise. Ending poverty, increasing global prosperity and reduc- ing global inequality, already difficult, will be much harder with 2°C warming, but at 4°C there is serious doubt whether these goals can be achieved at all. For this report, the third in the Turn Down the Heat series, we turned again to the scientists at the Potsdam Institute for Climate Impact Research and Climate Analytics. We asked them to look at the likely impacts of present day (0.8°C), 2°C and 4°C warming on agricultural production, water resources, cities and ecosystems across Latin America and the Caribbean, Middle East and North Africa, and parts of Europe and Central Asia. Their findings are alarming. In Latin America and the Caribbean, heat extremes and changing precipitation patterns will have adverse effects on agricultural productivity, hydrological regimes and biodiversity. In Brazil, at 2°C warming, crop yields could decrease by up to 70 percent for soybean and up to 50 percent for wheat. Ocean acidification, sea level rise, tropical cyclones and temperature changes will negatively impact coastal livelihoods, tour- ism, health and food and water security, particularly in the Caribbean. Melting glaciers would be a hazard for Andean cities. In the Middle East and North Africa, a large increase in heat-waves combined with warmer average tem- peratures will put intense pressure on already scarce water resources with major consequences for regional food security. Crop yields could decrease by up to 30 percent at 1.5–2°C and by almost 60 percent at 3–4°C. At the same time, migration and climate-related pressure on resources might increase the risk of conflict. In the Western Balkans and Central Asia, reduced water availability in some places becomes a threat as temperatures rise toward 4°C. Melting glaciers in Central Asia and shifts in the timing of water flows F Fo r e wo r d fr o m the Fu ll Rep or t will lead to less water resources in summer months and high risks of torrential floods. In the Balkans, a higher risk of drought results in potential declines for crop yields, urban health, and energy generation. In Macedonia, yield losses are projected of up to 50 percent for maize, wheat, vegetables and grapes at 2°C warming. In northern Russia, forest dieback and thawing of permafrost threaten to amplify global warming as stored carbon and methane are released into the atmosphere, giving rise to a self-amplifying feedback loop. Turn Down the Heat: Confronting the New Climate Normal builds on our 2012 report, which concluded the world would warm by 4°C by the end of this century with devastating consequences if we did not take concerted action now. It complements our 2013 report that looked at the potential risks to development under different warming scenarios in Sub-Saharan Africa, South East Asia and South Asia, and which warned that we could experience a 2°C world in our lifetime. Many of the worst projected climate impacts outlined in this latest report could still be avoided by holding warming below 2°C. But, this will require substantial technological, economic, institutional and behavioral change. It will require leadership at every level of society. Today the scientific evidence is overwhelming, and it’s clear that we cannot continue down the current path of unchecked, growing emissions. The good news is that there is a growing consensus on what it will take to make changes to the unsustainable path we are currently on. More and more voices are arguing that is possible to grow greener without necessarily growing slower. Today, we know that action is urgently needed on climate change, but it does not have to come at the expense of economic growth. We need smart policy choices that stimulate a shift to clean public transport and energy efficiency in factories, buildings and appliances can achieve both growth and climate benefits. This last report in the Turn Down the Heat series comes at a critical moment. Earlier this year, the UN Secretary General’s Climate Summit unleased a new wave of optimism. But our reports make clear that time is of the essence. Governments will gather first in Lima and then Paris for critical negotiations on a new climate treaty. Inside and outside of the conference halls, global leaders will need to take difficult decisions that will require, in some instances, short term sacrifice but ultimately lead to long term gains for all. At the World Bank Group we will use our financial capacity to help tackle climate change. We will innovate and bring forward new financial instruments. We will use our knowledge and our convening power. We will use our evidence and data to advocate and persuade. In short, we will do everything we can to help countries and communities build resilience and adapt to the climate impacts already being felt today and ensure that finance flows to where it is most needed. Our response to the challenge of climate change will define the legacy of our generation. The stakes have never been higher. Dr. Jim Yong Kim President, World Bank Group G Abbreviations °C degrees Celsius JJA June, July, and August (the summer season of the $ United States Dollars northern hemisphere; also known as the boreal AI Aridity Index summer) AOGCM Atmosphere-Ocean General Circulation Model LAC Latin America and the Caribbean AR4 Fourth Assessment Report of the LDC Least Developed Countries Intergovernmental Panel on Climate Change MAGICC Model for the Assessment of Greenhouse Gas AR5 Fifth Assessment Report of the Intergovernmental Induced Climate Change Panel on Climate Change MCMA The Mexico City Metropolitan Area BAU Business as Usual MENA Middle East and North Africa CaCO3 Calcium Carbonate MGIC Mountain Glaciers and Ice Caps CAT Climate Action Tracker NAO North Atlantic Oscillation CMIP5 Coupled Model Intercomparison Project Phase 5 NDVI Normalized Differenced Vegetation Index (used as CO2 Carbon Dioxide a proxy for terrestrial gross primary production) DGVM Dynamic Global Vegetation Model NH Northern Hemisphere DIVA Dynamic Interactive Vulnerability Assessment NPP Net Primary Production DJF December, January, and February (the winter OECD Organization for Economic Cooperation and season of the northern hemisphere) Development ECA Europe and Central Asia PDSI Palmer Drought Severity Index ECS Equilibrium Climate Sensitivity PgC Petagrams of Carbon (1 PgC = 1 billion tons of ENSO El-Niño/Southern Oscillation carbon) FAO Food and Agricultural Organization ppm Parts Per Million FPU Food Productivity Units PPP Purchasing Power Parity (a weighted currency GCM General Circulation Model based on the price of a basket of basic goods, GDP Gross Domestic Product typically given in US dollars) GFDRR Global Facility for Disaster Reduction and Recovery RCM Regional Climate Model GLOF Glacial Lake Outburst Flood RCP Representative Concentration Pathway HCS Humboldt Current System SCM Simple Climate Model IAM Integrated Assessment Model SLR Sea-level Rise IEA International Energy Agency SRES IPCC Special Report on Emissions Scenarios IPCC Intergovernmental Panel on Climate Change SREX IPCC Special Report on Managing the Risks of Extreme ISI-MIP Inter-Sectoral Impact Model Intercomparison Events and Disasters to Advance Climate Change Project Adaptation ITCZ Intertropical Convergence Zone H Ab breviations TgC Teragrams of Carbon (1 TgC = 1 million tons of UNHCR United Nations High Commissioner for Refugees carbon) USAID United States Agency for International UNCCD United Nations Convention to Combat Development Desertification WBG World Bank Group UNDP United Nations Development Programme WGI Working Group I (also WGII, WGIII) UNEP United Nations Environment Programme WHO World Health Organization UNFCCC United Nations Framework Convention on Climate Change I Glossary Aridity Index:  The Aridity Index (AI) is an indicator designed for CO2 fertilization:  The CO2 fertilization effect refers to the effect identifying structurally arid regions; that is, regions with a long- of increased levels of atmospheric CO2 on plant growth. It may term average precipitation deficit. AI is defined as total annual increase the rate of photosynthesis mainly in C3 plants and increase precipitation divided by potential evapotranspiration, with the water use efficiency, thereby causing increases in agricultural pro- latter a measure of the amount of water a representative crop type ductivity in grain mass and/or number. This effect may to some would need as a function of local conditions such as temperature, extent offset the negative impacts of climate change on crop yields, incoming radiation, and wind speed, over a year to grow, which although grain protein content may decline. Long-term effects is a standardized measure of water demand. are uncertain as they heavily depend on a potential physiological long-term acclimation to elevated CO2 and other limiting factors, Biome:  A biome is a large geographical area of distinct plant and including soil nutrients, water, and light. animal groups, one of a limited set of major habitats classified by climatic and predominant vegetative types. Biomes include, CMIP5:  The Coupled Model Intercomparison Project Phase 5 for example, grasslands, deserts, evergreen or deciduous forests, (CMIP5) brought together 20 state-of-the-art GCM groups, which and tundra. Many different ecosystems exist within each broadly generated a large set of comparable climate-projection data. The defined biome, all of which share the limited range of climatic project provided a framework for coordinated climate change experi- and environmental conditions within that biome. ments and includes simulations for assessment in the IPCC AR5. C3/C4 plants:  C3 and C4 refer to two types of photosynthetic Development narratives:  Development narratives highlight the biochemical pathways. C3 plants include more than 85 percent implications of climate change impacts on regional development. of plants (e.g., most trees, wheat, rice, yams, and potatoes) and The Turn Down the Heat series, and in particular this report, discuss respond well to moist conditions and to additional CO2 in the the potential climate change impacts on particularly vulnerable atmosphere. C4 plants (e.g., savanna grasses, maize, sorghum, groups along distinct storylines—the so called development nar- millet, and sugarcane) are more efficient in water and energy use ratives. These development narratives were developed for each and outperform C3 plants in hot and dry conditions. region in close cooperation with regional World Bank specialists. They provide an integrated, often cross-sectoral analysis of climate CAT:  The Climate Action Tracker is an independent, science- change impacts and development implications at the sub-regional based assessment that tracks the emissions commitments of and or regional level. Furthermore, the development narratives add actions by individual countries. The estimates of future emissions to the report by allowing the science-based evidence of physical deducted from this assessment serve to analyze warming scenarios and biophysical impacts to be drawn out into robust develop- that could result from current policy: (i) CAT Reference BAU: a ment storylines to characterize the plausible scenarios of risks lower reference business-as-usual scenario that includes existing and opportunities—showcasing how science and policy interface. climate policies but not pledged emissions reductions; and (ii) CAT Current Pledges:  a scenario additionally incorporating reductions GCM:  A General Circulation Model is the most advanced type currently pledged internationally by countries. of climate model used for projecting changes in climate due to J Glos s ary increasing greenhouse gas concentrations, aerosols, and external ISI-MIP:  The first Inter-Sectoral Impact Model Intercomparison forcing (like changes in solar activity and volcanic eruptions). Project (ISI-MIP) is a community-driven modeling effort which These models contain numerical representations of physical pro- provides cross-sectoral global impact assessments based on the cesses in the atmosphere, ocean, cryosphere, and land surface newly developed climate Representative Concentration Pathways on a global three-dimensional grid, with the current generation and socioeconomic scenarios. More than 30 models across five of GCMs having a typical horizontal resolution of 100–300 km. sectors (agriculture, water resources, biomes, health, and infra- structure) were incorporated in this modeling exercise. GDP:  Gross Domestic Product is the sum of the gross value added by all resident producers in the economy plus any product taxes Pre-industrial Level (what it means to have present 0.8°C warm- and minus any subsidies not included in the value of the product. ing):  Pre-industrial level refers to the level of warming before/at It is calculated without deductions for depreciation of fabricated the onset of industrialization. The instrumental temperature records assets or for depletion and degradation of natural resources. show that the 20-year average of global-mean, near-surface air temperature in 1986–2005 was about 0.6°C higher than the aver- GDP PPP:  This is GDP on a purchasing power parity basis divided age over 1851–1879. There are, however, considerable year-to-year by population. Whereas PPP estimates for OECD countries are variations and uncertainties in the data. In addition, the 20-year quite reliable, PPP estimates for developing countries are often average warming over 1986–2005 is not necessarily representa- rough approximations. tive of present-day warming. Fitting a linear trend over the period 1901–2010 gives a warming of 0.8°C since “early industrialization.” Highly unusual and Unprecedented:  In this report, highly Global mean, near-surface air temperatures in the instrumental unusual and unprecedented heat extremes are defined using records of surface-air temperature have been assembled dating thresholds based on the historical variability of the current local back to about 1850. The number of measurement stations in the climate. The absolute level of the threshold depends on the natural early years is small and increases rapidly with time. Industrializa- year-to-year variability in the base period (1951–1980), which is tion was well on its way by 1850 and 1900, which implies using captured by the standard deviation (sigma). Highly unusual heat 1851–1879 as a base period, or 1901 as a start for linear trend extremes are defined as 3-sigma events. For a normal distribution, analysis might lead to an underestimate of current and future 3-sigma events have a return time of 740 years. The 2012 U.S. warming. However, global greenhouse-gas emissions at the end of heat wave and the 2010 Russian heat wave classify as 3-sigma the 19th century were still small and uncertainties in temperature and thus highly unusual events. Unprecedented heat extremes reconstructions before this time are considerably larger. are defined as 5-sigma events. They have a return time of several million years. Monthly temperature data do not necessarily fol- RCP:  Representative Concentration Pathways are based on low a normal distribution (for example, the distribution can have carefully selected scenarios for work on integrated assess- long tails, making warm events more likely) and the return times ment modeling, climate modeling, and modeling and analysis can be different from the ones expected in a normal distribution. of impacts. Nearly a decade of new economic data, informa- Nevertheless, 3-sigma events are extremely unlikely and 5-sigma tion about emerging technologies, and observations of such events have almost certainly never occurred over the lifetime of environmental factors as land use and land cover change are key ecosystems and human infrastructure. reflected in this work. Rather than starting with detailed socio- economic storylines to generate emissions scenarios, the RCPs Hyper-aridity:  This refers to land areas with very low Aridity Index are consistent sets of projections of only the components of (AI) scores, generally coinciding with the great deserts. There is radiative forcing (the change in the balance between incoming no universally standardized value for hyper-aridity, and values and outgoing radiation to the atmosphere caused primarily by between 0 and 0.05 are classified in this report as hyper-arid. changes in atmospheric composition) that are meant to serve as IPCC AR4, AR5:  The Intergovernmental Panel on Climate Change inputs for climate modeling. These radiative forcing trajectories (IPCC) is the leading body of global climate change assessments. are not associated with unique socioeconomic or emissions It comprises hundreds of leading scientists worldwide and on a scenarios; instead, they can result from different combina- regular basis publishes assessment reports which provide a com- tions of economic, technological, demographic, policy, and prehensive overview of the most recent scientific, technical, and institutional futures. RCP2.6, RCP4.5, RCP6 and RCP8.5 refer, socioeconomic information on climate change and its implications. respectively, to a radiative forcing of +2.6 W/m², +4.5 W/m², The Fourth Assessment Report (AR4) was published in 2007. The +6 W/m² and +8.5 W/m² in the year 2100 relative to pre- Fifth Assessment Report (AR5) was published in 2013/2014. industrial conditions. K Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal RCP2.6:  RCP2.6 refers to a scenario which is representative of the SREX:  The IPCC published a special report on Managing the literature on mitigation scenarios aiming to limit the increase of Risks of Extreme Events and Disasters to Advance Climate Change global mean temperature to 2°C above pre-industrial levels. This Adaptation (SREX) in 2012. The report provides an assessment of emissions path is used by many studies that have been assessed the physical and social factors shaping vulnerability to climate- for the IPCC 5th Assessment Report and is the underlying low related disasters and gives an overview of the potential for effective emissions scenario for impacts assessed in other parts of this disaster risk management. report. In this report, the RCP2.6 is referred to as a 2°C world (with the exception of sea-level rise, where the subset of model Tipping element:  Following Lenton et al. (2008), the term tipping used actually leads to 1.5°C world—). element describes large scale components of the Earth system pos- sibly passing a tipping point. A tipping point “commonly refers to RCP8.5:  RCP8.5 refers to a scenario with a no-climate-policy base- a critical threshold at which a tiny perturbation can qualitatively line with comparatively high greenhouse gas emissions which is alter the state or development of a system” (Lenton et al. 2008). used by many studies that have been assessed for the IPCC Fifth The consequences of such shifts for societies and ecosystems are Assessment Report (AR5). This scenario is also the underlying likely to be severe. high-emissions scenario for impacts assessed in other parts of this report. In this report, the RCP8.5 is referred to as a 4°C world Virtual water:  A measure of the water resources used in the pro- above the pre-industrial baseline period. duction of agricultural commodities. International trade in such commodities thereby implies a transfer of virtual water resources Severe and extreme:  These terms indicate uncommon (negative) from one country to another embedded in the products. consequences. These terms are often associated with an additional qualifier like “highly unusual” or “unprecedented” that has a WGI, WGII, WG III:  IPCC Working Group I assesses the physical specific quantified meaning. scientific aspects of the climate system and climate change. IPCC Working Group II assesses the vulnerability of socio-economic SRES:  The Special Report on Emissions Scenarios, published by the and natural systems to climate change, negative and positive IPCC in 2000, has provided the climate projections for the Fourth consequences of climate change, and options for adapting to it. Assessment Report (AR4) of the Intergovernmental Panel on Climate IPCC Working Group III assesses the options for mitigating climate Change. The scenarios do not include mitigation assumptions. The change through limiting or preventing greenhouse gas emissions SRES study included consideration of 40 different scenarios, each and enhancing activities that remove them from the atmosphere. making different assumptions about the driving forces determining future greenhouse gas emissions. Scenarios were grouped into four families (A1FI, A2, B1 and B2), corresponding to a wide range of high- and low-emissions scenarios. L Latin America and the Caribbean The Latin America and the Caribbean region encompasses a huge diversity of landscapes and ecosystems. The region is highly heterogeneous in terms of economic development and social and indigenous history. It is also one of the most urbanized regions in the world. In Latin America and the Caribbean, temperature and precipitation changes, heat extremes, and the melting of glaciers will have adverse effects on agricultural productivity, hydrological regimes, and biodiversity. In Brazil, without additional adaptation, crop yields could decrease by 30–70 percent for soybean and up to 50 percent for wheat at 2°C warming. Ocean acidification, sea level rise, and more intense tropical cyclones will affect coastal livelihoods and food and water security, particularly in the Caribbean. Local food security is also seriously threatened by the projected decrease in fishery catch potential. Reductions and shifts in water availability would be particularly severe for Andean cities. The Amazon rainforest may be at risk of large scale forest degradation that contributes to increasing atmospheric carbon dioxide concentration and local and regional hydrological changes. 1  Regional Summary of the region are mainly rain-fed and, as a result, susceptible to variable rainfall and temperatures. In the Andean regions, houses The Latin America and Caribbean region is highly heterogeneous built on the often steep terrain are critically exposed to storm in terms of economic development and social and indigenous surface flows, glacial lake outbursts, and landslides. Coastal history with a population of 588 million (2013), of which almost residents, particularly in the Caribbean region, face the risks of 80 percent is urban. The current GDP is estimated at $5.655 loss of ecosystem services and livelihoods from degrading marine trillion (2013) with a per capita GNI of $9,314 in 2013. In 2012, ecosystems, loss of physical protection from degrading reefs, and approximately 25 percent of the population was living in poverty coastal flooding, as well as from damages to critical infrastructure and 12 percent in extreme poverty, representing a clear decrease (especially in the beach front tourism sector) and threats to fresh- compared to earlier years. Undernourishment in the region, for water from sea water intrusion due to sea level rise. example, declined from 14.6 percent in 1990 to 8.3 percent in 2012. Despite considerable economic and social development progress 1.1  Regional Patterns of Climate Change in past decades, income inequality in the region remains high. The region is highly susceptible to tropical cyclones and strong 1.1.1  Temperatures and Heat Extremes El Niño events, as well as to rising sea levels, melting Andean By 2100, summer temperatures over the region will increase by glaciers, rising temperatures and changing rainfall patterns. The approximately 1.5°C under the low-emissions scenario (a 2°C rural poor who depend on a natural resource base are particularly world) and by about 5.5o C under the high-emissions scenario vulnerable to climate impacts on subsistence agriculture and (a 4°C world) compared to the 1951–1980 baseline (Figure 3.1). ecosystem services; the urban poor living along coasts, in flood Along the Atlantic coast of Brazil, Uruguay, and Argentina, the plains, and on steep slopes are particularly vulnerable to extreme warming is projected to be less than the global average, ranging precipitation events and the health impacts of heat extremes. The between 0.5–1.5°C in a 2°C world and 2–4°C in a 4°C world. intensive grain-producing cropping systems in the southern part In the central South American region of Paraguay, in northern 13 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal Figure 1.1: Multi-model mean temperature anomaly for Latin America and the Caribbean for RCP2.6 (2°C world, left) and RCP8.5 (4°C world, right) for the austral summer months (DJF). Temperature anomalies in degrees Celsius are averaged over the time period 2071–2099 relative to 1951–1980. Argentina, and in southern Bolivia, warming is likely to be more under climate change, most dry regions will get drier and most pronounced, up to 2.5°C in a 2°C world and up to 6°C in a 4°C wet regions will get wetter. The exception is central Brazil. The world by 2071–2099. Similar levels of warming are projected for annual mean precipitation here is projected to drop by 20 percent the equatorial region, including eastern Colombia and southern in a 4°C world by the end of the century. In general, more intense Venezuela. Projections indicate that in a 4°C world almost all and frequent extreme precipitation events also become more likely. land area (approximately 90 percent) will be affected by highly In a 4°C world, the Amazon basin, the full land area of Brazil unusual,28 and more than half of the land area (approximately except the southern coast, southern Chile, the Caribbean, Central 70 percent) by unprecedented, summer heat extremes. America, and northern Mexico, are expected to be under severe to extreme drought conditions relative to the present climate by the end 1.1.2  Precipitation, Drought, and Aridity of the 21st century. The total area of land classified as hyper-arid, arid, In general, in a 2°C world, precipitation changes are relatively small or semi-arid is projected to grow from about 33 percent in 1951–1980 (+/–10 percent) and models exhibit substantial disagreement on to 36 percent in a 2°C world, and to 41 percent in a 4°C world. the direction of change over most land regions. In a 4°C world, the models converge in their projections over most regions, but 1.1.3  Tropical Cyclones inter-model uncertainty remains over some areas (such as north- Observations over the last 20–30 years show positive trends in ern Argentina and Paraguay) (Figure 3.2). Tropical countries on tropical cyclone frequency and strength over the North Atlantic the Pacific coast (Peru, Ecuador, and Colombia) are projected to but not over the eastern North Pacific. While Atlantic tropical see an increase in annual mean precipitation of about 30 percent. cyclones are suppressed by the El Niño phase of ENSO, they are Similarly, Uruguay on the Atlantic coast (and bordering regions enhanced in the eastern North Pacific. Under further anthropogenic in Brazil and Argentina) will get wetter. Regions which are pro- climate change, the frequency of high-intensity tropical cyclones jected to become drier include Patagonia (southern Argentina and is generally projected to increase over the western North Atlantic Chile), Mexico, and central Brazil. These patterns indicate that, by 40 percent for 1.5–2.5°C global warming and by 80 percent in a 4°C world. Global warming of around 3°C is associated with In this report, highly unusual heat extremes refer to 3-sigma events and unprec- 28  an average 10 percent increase in rainfall intensity averaged over edented heat extremes to 5-sigma events (see Appendix). a 200 km radius from a tropical cyclone’s center. Although there 14 Lati n Ame r i ca and the Caribbean Figure 1.2: Multi-model mean of the percentage change in the aridity index under RCP2.6 (2°C world, left) and RCP8.5 (4°C world, right) for Latin America and the Caribbean by 2071–2099 relative to 1951–1980. Hatched areas indicate uncertain results, with two or more out of five models disagreeing on the direction of change. Note that a negative change cor- responds to a shift to more arid conditions.29 is some evidence from multiple-model studies for a projected to experience above-average sea-level rise (Recife: median estimate: increase in frequency of tropical cyclones along the Pacific coast 0.63 m, low estimate: 0.41 m, high estimate: 1.14 m; Rio de Janeiro: of Central America, overall projections in this region are currently median estimate: 0.62 m, low estimate: 0.46 m, high estimate: inconclusive. Despite these inconclusive projections, however, any 1.11 m). Sea-level rise is exacerbated at low latitudes due to both increase in Pacific and Atlantic storms (not necessarily cyclones) increased ocean heat uptake and the gravity-induced pattern of ice making landfall simultaneously would potentially entail more sheets and glaciers. As an example, Guayaquil on the Pacific Coast damaging impacts than increasing frequency of any individual of Ecuador is projected to experience 0.62 m (low estimate: 0.46 m, Pacific or Atlantic cyclone.29 high estimate: 1.04 m) of sea-level rise in a 4°C world. In contrast, Puerto Williams (Chile) at the southern tip of the South American 1.2  Regional Sea-Level Rise continent is projected to experience only 0.46 m (low estimate: 0.38 m; high estimate: 0.65 m). Port-Au-Prince (Haiti) is projected Sea-level rise is projected to be higher at the Atlantic coast than at the to experience 0.61 m (low estimate: 0.41 m, high estimate: 1.04 m) Pacific coast. Valparaiso (median estimate: 0.55 m for a 4°C world) of sea-level rise in a 4°C world (Figure 3.11); it serves as a typical is projected to benefit from southeasterly trade wind intensification example for sea-level rise in other Caribbean islands. over the Southern Pacific and associated upwelling of cold water leading to below-average thermosteric (due to ocean temperature 1.3  Sector-based and Thematic Impacts rise) sea-level rise. In contrast, the Atlantic coast of Brazil is projected 1.3.1  Glaciers and Snowpack Changes Some individual grid cells have noticeably different values than their direct neigh- 29  Glacial recession in South America has been significant. The bors (e.g., on the border between Peru and Bolivia). This is due to the fact that the tropical glaciers in the Central Andes in particular have lost major Aridity index is defined as a fraction of total annual precipitation divided by potential portions of their volume in the course of the 20th century. A clear evapotranspiration (see Appendix). It therefore behaves in a strongly non-linear way, and thus year-to-year fluctuations can be large. Since averages are calculated over a trend of glacial retreat is also visible for glaciers in the southern relatively small number of model simulations, this can result in these local jumps. Andes, which have lost about 20 percent of their volume. 15 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal The recession of the tropical glaciers in the Central Andes will 1.3.3  Climate Change Impacts on Agriculture, continue as rapidly as it has in recent decades. Even for low or Livestock, and Food Security intermediate emissions scenarios inducing a global warming of The results of the climate change impact projections on crop yields 2–3°C above pre-industrial levels, two comprehensive studies con- differ among studies, but most authors agree that climate change sistently project a glacial volume loss of 78–97 percent. Both studies will very likely decrease agricultural yields of important food predict an almost complete deglaciation (93–100 percent) for a 4°C crops in the Latin America and Caribbean region. An exception world. Other studies are slightly less dramatic; irrespective of the is the projected increase in yield of irrigated/flooded rice in some temperature evolution in the next decades, however, large parts of regions. The few available studies on climate change impacts on the glaciers of the tropical Andes will be gone long before the end of livestock indicate that beef and dairy cattle production will decline the century. In the Southern Andes, the model spread for the 2–3°C under increasing temperatures, as heat stress is a major influenc- global warming ranges from 22–59 percent glacier volume loss; a ing factor of cattle productivity. Sheep seem to cope better with comparison for individual scenarios is difficult. In a 4°C world, warmer and drier conditions than cattle and pigs. models project a glacier volume retreat of 44–74 percent by 2100. Monitoring of snow cover in the high altitudes of Chile and 1.3.4  Climate Change Impacts on Biodiversity Argentina since 1950 shows no significant trend (possible trends Climate change-induced negative effects on biodiversity, from are hard to identify in the records, since the inter-annual variabil- range contractions to extinctions, are very likely in a warmer than ity is large and clearly modulated by ENSO). The lack of reliable 2°C world. As the adaptive capacity of affected species and eco- projections for snowpack and snow cover changes in the Andes systems is hard to project or quantify, models need to use simpli- is an important research gap. fied approaches as implemented in bioclimatic envelope models, species-distribution models, and dynamic global vegetation models. 1.3.2  Water Resources, Water Security, and Floods One clear trend regarding future warming levels is that the Although the magnitude of the change varies, there is a high more temperature is projected to increase, the more species diver- agreement on decreasing mean annual runoff and discharge in sity is affected. Mountainous regions in the tropics (e.g., cloud Central America. Water stress may increase, especially in arid forests) are projected to become very vulnerable due to the high areas with high population densities and during the dry season. number of endemic and highly specialized species which might In the Caribbean, runoff projections are of low confidence due to face mountaintop extinction. Most models do not take biotic lack of data. However, freshwater availability may decrease for interactions (e.g., food-web interactions, species competition) or several reasons, such as sea-level rise leading to an intrusion of resource limitations into account. Therefore, the realized ecological sea water into coastal aquifers. Regionally, the risk of flooding niche of species within an ecosystem might become much smaller and mudslides with high mortality rates is high. Although floods than what is potentially possible according to climatic and other often seem to be associated with land-use change, more severe environmental conditions, leading to shifts in ecological zones. flooding events may also occur in the context of climate change. Higher variability of seasonal discharge is projected for the 1.3.5  Amazon Rainforest Degradation, Dieback, Tropical Andes. Decreased streamflow during the dry season has and Tipping Point already been observed, and may decrease further as a result of Overall, the most recent studies suggest that the Amazon dieback is an ongoing glacier retreat. However, streamflow during the wet sea- unlikely, but possible, future for the Amazon region. Projected future son may increase. The Andean region could experience a higher precipitation and the effects of CO2 fertilization on tropical tree growth flood risk in a 4°C world (e.g. due to accelerated glacier melting). remain the processes with the highest uncertainty. Climate-driven In the Amazon Basin, runoff and discharge projections for most changes in dry season length and recurrence of extreme drought parts of the Amazon basin are diverging. For the western part of years, as well as the impact of fires on forest degradation, add to the basin a likely increase in streamflow, runoff, flood zone, and the list of unknowns for which combined effects still remain to be inundation time are projected. In southern most South America, investigated in an integrative study across the Amazon. A critical a decrease in mean runoff is projected. tipping point has been identified at around 40 percent deforesta- Although the Latin America and Caribbean region has an tion, when altered water and energy feedbacks between remaining abundance of freshwater resources, many cities depend on local tropical forest and climate may lead to a decrease in precipitation. rivers, aquifers, lakes, and glaciers that may be affected by climate A basin-wide Amazon forest dieback caused by feedbacks change—and freshwater supplies might not be enough to meet between climate and the global carbon cycle is a potential tipping demand. For example, Guadalajara (Mexico) and many Andean point of high impact if regional temperatures increase by more cities are expected to face increasing water stress and, if the current than 4°C and global mean temperatures increase by more than demand continues, low-income groups who already lack adequate 3°C toward the end of the 21st century. Recent analyses have, how- access to water will face more challenges. ever, downgraded this probability from 21 percent to 0.24 percent 16 Lati n Ame r i ca and the Caribbean for the 4°C regional warming level when coupled carbon-cycle including droughts, floods, landslides, and tropical cyclones; all climate models are adjusted to better represent the inter-annual of these extreme events can induce migration. variability of tropical temperatures and related CO2 emissions. Examples indicate that drought-induced migration is already This holds true, however, only when the CO2 fertilization effect is occurring in some regions. The largest level of climate migration realized as implemented in current vegetation models. Moreover, is likely to occur in areas where non-environmental factors (e.g., large-scale forest degradation as a result of increasing drought may poor governance, political persecution, population pressures, and impair ecosystem services and functions, including the regional poverty) are already present and putting migratory pressures on hydrological cycle, even without a forest dieback. local populations. The region is considered to be at low risk of armed conflict. 1.3.6  Fisheries and Coral Reefs However, in the context of high social and economic inequality Together with ocean acidification and hypoxia, which are very likely and migration flows across countries, disputes regarding access to to become more pronounced under high-emissions scenarios, the resources, land, and wealth are persistent. Climate change could possibility of more extreme El Niño events poses substantial risks increase the risk of conflict in the region through more resource to the world’s richest fishery grounds. Irrespective of single events, scarcity, more migration, increasing instability, and increasing the gradual warming of ocean waters has been observed and is frequency and intensity of natural disasters. further expected to affect fisheries (particularly at a local scale). Generally, fish populations are migrating poleward toward 1.3.9  Coastal Infrastructure colder waters. Projections indicate an increase in catch potential By 2050, coastal flooding with a sea-level rise of 20 cm could of up to 100 percent in the south of Latin America. Off the coast of generate approximately $940 million of mean annual losses in Uruguay, the southern tip of Baja California, and southern Brazil the 22 largest coastal cities in the Latin America and Caribbean the maximum catch potential is projected to decrease by more than region, and about $1.2 billion with a sea-level rise of 40 cm. The 50 percent. Caribbean waters and parts of the Atlantic coast of Central Caribbean region is particularly vulnerable to climate change due America may see declines in the range of 5–50 percent. Along the to its low-lying areas and the population’s dependence on coastal coasts of Peru and Chile, fish catches are projected to decrease by and marine economic activity. In a scenario leading to a 4°C world up to 30°percent, but there are increases expected toward the south. and featuring 0.89–1.4 m of sea-level rise, tropical cyclones in Irrespective of the sensitivity threshold chosen, and irrespective the Caribbean alone could generate an extra $22 billion by 2050 of the emissions scenario, by the year 2040, Caribbean coral reefs (and $46 billion by 2100) in storm and infrastructure damages and are expected to experience annual bleaching events. While some tourism losses, compared to a scenario leading to a 2°C world. species and particular locations appear to be more resilient to such The potential increase in tropical cyclone intensity may increase events, it is clear that the marine ecosystems of the Caribbean port downtime for ships and, therefore, increase shipping costs. are facing large-scale changes with far-reaching consequences for Beach tourism is particularly exposed to direct and indirect climate associated livelihood activities as well as for the coastal protection change stressors, including sea-level rise, modified tropical storm provided by healthy coral reefs. patterns, heightened storm surges, and coastal erosion. Coastal 1.3.7 Health tourist resorts are potentially two-to-three times more exposed The Latin America and Caribbean region faces increased risks of to climate change-related stressors than inland touristic resorts. morbidity and mortality caused by infectious diseases and extreme 1.3.10 Energy weather events. Observed patterns of disease transmission associ- The assessment of the current literature on climate change impacts ated with different parts of the ENSO cycle offer clues as to how on energy in Latin America and the Caribbean shows that there changes in temperature and precipitation might affect the incidence are only a few studies, most of which make strong assumptions of a particular disease in a particular location. Projections of how about such key issues as seasonality of water supply for hydro- malaria incidence in the region could be affected by climate change power. These studies are more qualitative than quantitative and over the rest of the century are somewhat inconsistent, with some important gaps remain. There is also a lack of studies with respect studies pointing to increased incidence and others to decreased to the impacts of climate change impacts on renewable energies. incidence. Such uncertainty also characterizes studies of the rela- In general, the impacts of climate change on energy demand tionship between climate change and malaria globally and reflects are less well studied than those on energy supply—and, yet, the complexity of the environmental factors influencing the disease. demand and supply interact in a dynamic way. For example, the 1.3.8  Migration and Security concomitant increase in energy demand during heat extremes While migration is not a new phenomenon in the region, it is and the decrease in energy supply through reduced river flow and expected to accelerate under climate change. There are many areas low efficiencies may put existing energy systems under increasing in the Latin America and Caribbean Region prone to extreme events, pressure in the future. 17 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal 1.4  Overview of Regional Development Narratives The development narratives build on the climate change impacts ana- MEXICO DOMINICAN HAITI REPUBLIC ST. KITTS AND NEVIS JAMAICA lyzed in this report (see Table 1.15: Synthesis table of climate change BELIZE ANTIGUA AND BARBUDA DOMINICA GUATEMALA HONDURAS ST. VINCENT AND THE GRENADINES impacts in LAC under different warming levels) and are presented ST. LUCIA BARBADOS EL SALVADOR NICARAGUA GRENADA in more detail in Section 5. Climate change impacts have manifold COSTA RICA R.B. DE TRINIDAD AND TOBAGO PANAMA VENEZUELA direct and indirect implications for development in the region. These GUYANA COLOMBIA SURINAME impacts occur on a continuum from rural to urban; not only are there many climate impacts directly affecting rural spaces leading for ECUADOR example to reduced agricultural productivity or altered hydrological regimes, but these impacts also affect urban areas through changing PERU BRAZIL ecosystem services, migration flows, and so forth. Development will likewise be impacted as the challenges of a changing climate mount BOLIVIA and interact with socioeconomic factors. In particular, glacial melt and PARAGUAY changing river flows, extreme events, and risks to food production systems will put human livelihoods under pressure. Climate change impacts are and will continue to affect devel- CHILE ARGENTINA URUGUAY opment across the region in several ways. First, changes to the hydrological cycle endanger the stability of freshwater supplies and ecosystem services. An altered hydrological system due to changing runoff, glacial melt, and snowpack changes will affect the ecosystem GSDPM Map Design Unit services that the rural population depends on, freshwater supplies in IBRD 41281 OCTOBER 2014 This map was produced by the Map Design Unit of The World Bank. Falkland Islands (Islas Malvinas) The boundaries, colors, denominations and any other information cities, and such major economic activities as mining and hydropower. shown on this map do not imply, on the part of The World Bank Group, any judgment on the legal status of any territory, or any endorsement or acceptance of such boundaries. A DISPUTE CONCERNING SOVEREIGNTY OVER THE ISLANDS EXISTS BETWEEN ARGENTINA WHICH CLAIMS THIS SOVEREIGNTY AND THE U.K. WHICH ADMINISTERS THE ISLANDS. Second, climate change places at risk both large-scale agricultural production for export and small-scale agriculture for regional food production. Third, a stronger prevalence of extreme events affects Grenada, Guatemala, Guyana, Haiti, Honduras, Jamaica, Mexico, both rural and urban communities, particularly in coastal regions. Nicaragua, Panama, Paraguay, Peru, Puerto Rico, St Kitts and Nevis, At the sub-regional level, the following climate-development St Lucia, St Vincent and the Grenadines, Suriname, Trinidad and interactions are particularly important. In Central America and Tobago, Uruguay, and Venezuela. the Caribbean, extreme events threaten livelihoods and damage The region is very large and has a very diverse range of distrib- infrastructure. In the Andes, changes in water resource availability uted ecosystems from the Andean mountains that stretch for about challenge the rural and urban poor. In the Amazon, the risks of 8,850 kilometers, to mountain glaciers, vast rainforests, savannas, a tipping point, forest degradation, and biodiversity loss threaten grasslands, wetlands, islands, deserts and a coastline that is over local communities. Hydrological changes may affect the wider 72,000 kilometers long. There are broad differences in development region. The Southern Cone faces risks to export commodities from levels both within and among countries (Table 3.1); these factors loss of production from intensive agriculture. In the Mexican dry influence the social vulnerability of the population. In addition, subtropical regions and northeastern Brazil, increasing drought current and projected climate change impacts vary strongly within stress threatens rural livelihoods and health. the region, with some key impacts relating to changing temperatures and precipitation. Changes in extreme events (e.g., heatwaves, 2 Introduction droughts, tropical cyclones, and changing ENSO patterns) (Sec- tion 2.3.2 of the full report, El-Niño/Southern Oscillation) and This report defines Latin America and the Caribbean (LAC) as sea-level rise are also projected to vary across the region. These the region encompassing the South American continent, Central physical risk factors trigger biophysical impacts on hydrological America,30 the Caribbean islands, and Mexico. It is constituted flows, agricultural productivity, biodiversity in general, and forest by the following countries: Antigua and Barbuda, Argentina, the dynamics in the Amazon in particular, coral reefs, and fisheries, as Bahamas, Barbados, Belize, Bolivia, Brazil, Chile, Colombia, Costa well as social impacts on human health, security, energy systems, Rica, Cuba, Dominica, Dominican Republic, Ecuador, El Salvador, and coastal infrastructure. This report analyses these physical, biophysical, and social The World Bank Central America subregion includes the following countries: Costa 30  30  impacts in an integrated way using data analysis, model projec- Rica, El Salvador, Guatemala, Honduras, Nicaragua, and Panama. tions, and an intensive review of the scientific literature. Wherever 18 possible, the results are regionally stratified. Table 1.1: Basic Socioeconomic Indicators of LAC Countries Urban Life Urban Population Agriculture, Expectancy Indicator Population Population Growth GDP per capita value added1 at Birth2 % of current 1000 Unit million population annual % US$ % of GDP years Year 2012 2012 2012 2012 2011 2011 SP.URB.TOTL ID SP.POP.TOTL .IN.ZS SP.URB.GROW NY.GDP.PCAP.CD NV.AGR.TOTL.ZS SP.DYN.LE00.IN Argentina 41.1 92.6 1.03 11.6 10.7 75.8 Antigua and Barbuda 0.1 29.9 1.01 12.7 2.5 75.5 Bahamas, The 0.4 84.4 1.75 21.9 2.3 74.8 Belize 0.3 44.6 2.01 – 13.1 73.5 Bolivia 10.5 67.2 2.27 2.6 12.5 66.6 Brazil 198.7 84.9 1.19 11.3 5.5 73.3 Barbados 0.3 44.9 1.65 14.9 1.5 75.0 Chile 17.5 89.3 1.13 15.5 3.7 79.3 Colombia 47.7 75.6 1.68 7.7 6.9 73.6 Costa Rica 4.8 65.1 2.12 9.4 6.5 79.5 Cuba 11.3 75.2 –0.07 0.0 5.0 78.9 Dominica 0.1 67.3 0.57 6.7 13.5 – Dominican Republic 10.3 70.2 2.07 5.7 6.0 73.0 Ecuador 15.5 68.0 2.43 5.4 10.4 75.9 Grenada 0.1 39.5 1.25 7.3 5.3 72.5 Guatemala 15.1 50.2 3.43 3.3 11.8 71.3 Guyana 0.8 28.5 0.88 3.6 21.3 65.9 Honduras 7.9 52.7 3.12 2.3 15.3 73.2 Haiti 10.2 54.6 3.85 0.8 – 62.3 Jamaica 2.7 52.2 0.36 5.4 6.6 73.1 St. Kitts and Nevis 0.1 32.1 1.41 14.3 1.8 – St. Lucia 0.2 17.0 –3.03 6.8 3.3 74.6 Mexico 120.8 78.4 1.60 9.7 3.4 76.9 Nicaragua 6.0 57.9 1.98 1.8 19.7 74.1 Panama 3.8 75.8 2.42 9.5 4.1 77.2 Peru 30.0 77.6 1.68 6.8 7.0 74.2 Puerto Rico 3.7 99.0 –0.64 27.7 0.7 78.4 Paraguay 6.7 62.4 2.58 3.8 21.4 72.1 El Salvador 6.3 65.2 1.40 3.8 12.5 71.9 Suriname 0.5 70.1 1.47 9.4 10.0 70.6 Trinidad and Tobago 1.3 14.0 2.26 17.4 0.5 69.7 Uruguay 3.4 92.6 0.45 14.7 9.4 76.8 St. Vincent and the 0.1 49.7 0.80 6.5 6.4 72.3 Grenadines Venezuela, RB 30.0 93.7 1.73 12.7 – 74.3 Note: Agriculture corresponds to ISIC divisions 1–5 and includes forestry, hunting, and fishing, as well as cultivation of crops and livestock production. Value added is the net 1 output of a sector after adding up all outputs and subtracting intermediate inputs. It is calculated without making deductions for depreciation of fabricated assets or depletion and degradation of natural resources. 2 Life expectancy at birth indicates the number of years a newborn infant would live if prevailing patterns of mortality at the time of its birth were to stay the same throughout its life. Source: World Bank (2013b). 19 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal 2.1  Social, Economic and Demographic Profile in densely populated low-income settlements (Ravallion et al. 2008) of the Latin America and Caribbean Region are more likely to be adversely affected by climate extremes. The Latin America and Caribbean Region comprises a population 2.2  Vulnerabilities to Climate Change in the of 588 million (in 2013) of which almost 80 percent is urban. The Latin America and Caribbean Region current GDP is estimated at US$ 5.655 trillion (in 2013) with a GNI per capita of US$ 9,314 in 2013. In 2012, approximately 28.2 per- In the LAC region, climate change is expected to accentuate pre- cent of the population were living in poverty and 11.3 percent existing socioeconomic vulnerabilities. People living in low-lying in extreme poverty or deprivation (ECLAC 2014). These figures coastal areas, slums (Douglas et al. 2008), and certain popula- represent a decrease of about 1.4 percent in the poverty rate with tion groups (such as poor (Ahmed et al. 2009; Hertel et al. 2010) respect to 2011 (ECLAC 2014). Although the number of people and women-led households (Kumar and Quisumbing 2011)), are living in poverty in the region has been going down slowly, in particularly exposed to shocks and future climate change risks. absolute terms, this means that 164 million people were poor—of Several socioeconomic and physical factors can contribute to whom 66 million were extremely poor (ECLAC 2014). increasing the vulnerability of populations to climate change. For Despite progress in the past decade and a growing middle example, poverty hinders households’ adaptive capacity (Kelly and class now surpassing the number of poor, inequality in the region Adger 2000). According to Calvo (2013), the following character- remains high, and may be stagnating. Thirty eight percent of istics of the population in the LAC region increase the exposure the population live just above the poverty line on an income of to climate change impacts and the likelihood of being affected by $4–10 per day (Fereira et al. 2013). economic shocks: (1) one third of the population can be classified Income inequalities in the region affect vulnerability to cli- as poor or extremely poor, so that any shock can push them into mate change, as the poor are more likely to be exposed to both further poverty; (2) there are more children in poor and extremely climate and economic shocks and limited ability to prepare for or poor households given the higher fertility rates amongst the poor, mitigate impacts. Haldén (2007) notes that disparities and divi- so that shocks can have particularly adverse consequences for sions could impede growth and undermine adaptation strategies, children, given that they are at a stage of life with greater needs with the additional risk that substantial inequality might also and dependency; and (3) poor households have members with destabilize societies and increase the likelihood of conflict in the fewer years of formal education, which can limit their capacity light of climate change and variability (see Section 2.5, Social to adapt to climate change impacts or macroeconomic shocks. Vulnerability to Climate Change). These income inequalities are further exacerbated by gender, spatial, and ethnic inequalities. 2.3  Vulnerabilities Faced by Rural Populations Ethnicity correlates closely with poverty. In seven countries for which data are available, the poverty rate is 1.2–3.4 times higher Even though urban and rural areas have both experienced poverty for indigenous and afrodescendent groups than for the rest of the reduction, the gap between the rural and urban experience is still population (ECLAC and UNFPA 2009). Furthermore, while indig- wide. In 2010, the rural poverty rate was twice as high as that enous peoples in LAC represent 10 percent of the population, their of urban areas; when considering extreme poverty, it was four income levels and human development indicators (e.g., education times as high (IFAD 2013). Close to 60 percent of the population and health status) have consistently fallen behind those of the in extreme poverty lives in rural areas (RIMISP 2011). Many rural rest of the population (Hall and Patrinos 2005). people in the region continue to live on less than $2 per day and The region’s population is expected to rise to 622 million in have poor access to financial services, markets, training, and 2015 and to 700 million by 2030. The distribution of the population other opportunities. The rural poor are thus more likely to feel the is increasingly urban. In 2010, the urban population accounted for impacts of climate change and variability given their dependency on 78.8 percent of the total; this number is projected to rise to 83.4 per- small-hold, rain-fed agriculture and other environmental resources cent by 2030 (ECLAC 2014). The concentration of poverty in urban that are particularly susceptible to the effects of climate change settlements is a central determinant of vulnerability to climate (Hoffman and Grigera 2013). Moreover, these populations have change. In addition, differences in fertility levels of social groups in limited political influence and are less able to leverage govern- Latin America and the Caribbean show that the poor segments of the ment support to help curb the effects of climate change (Hardoy urban population contribute most to urban growth, exacerbating the and Pandiella 2009). The dependence of the rural population on contribution of predominantly poor rural-urban migrants. Residents land as a source of food and income, coupled with lack of physi- cal and financial adaptive capacity, means that poor farmers are 20 Lati n Ame r i ca and the Caribbean also at increased risk of harm from slow-onset change (Rossing and Rubin 2011). Box 1.1: Hurricane Mitch’s Impact in Urban Areas 2.4  Urban Settlements and Marginalized In 1998, Hurricane Mitch cut a swath across Central America, hitting Populations Honduras especially hard. Overall, 3 percent of the central district of Honduras, including the cities of Tegucigalpa, was destroyed. Most of In addition to high levels of urbanization, in many countries in the damage was concentrated around the four rivers that cross the the region a high proportion of the urban population lives in a cities; as a result, 78 percent of Tegucigalpa’s drinking water supply few very large cities. National economies, employment patterns, pipelines were destroyed. Factors that increased the vulnerability of and government capacities—many of which are highly centralized— the city included obsolete and inadequate infrastructure, especially are also very dependent on these large cities. This makes them regarding water, sanitation, and drainage; a lack of zoning codes; extremely vulnerable (Hardoy and Pandiella 2009). Based on two concentration of services and infrastructure in only a few areas; a lack global model studies, Ahmed et al. (2009), Hertel et al. (2010), and of official prevention and mitigation strategies; and inappropriate man- Skoufias et al. (2011) estimate that urban salaried workers will be agement of the river basins. Source: Hardoy and Pandiella (2009). the most affected by climate change given the increase in prices of food resulting from reduced agricultural production. Increasing pressure on rural economic activities induced by droughts, heat waves, or floods—also driven by future climate change impacts— ill-suited to settlement, such as areas prone to flooding or affected could result in a greater rural exodus and add further pressure on by seasonal storms, sea surges, and other weather-related risks. human and economic development in cities (Marengo et al. 2012, Such land is cheap or is state-owned land and relatively easy for 2013; Vörösmarty et al. 2002). low-income groups to occupy. In most cases, the poor have no Spatial vulnerability within urban centers is a major source of formal tenure of the land and face not only environmental risks risk. There are particularly hazardous areas within Latin American but also the risk of eviction. Left with few options, low-income cities where settlements have been built, including flood plains groups live in overcrowded houses in neighborhoods with high (Calvo 2013). These settlements already face infrastructural population densities (Hardoy and Pandiella 2009). All these factors problems that affect water supplies, sanitation, and solid waste contribute to a high level of vulnerability to floods and landslides. management as they were built for less populated cities. This leaves In most LAC cities there are concentrations of low-income these areas at greater risk of flooding and other disasters (Hardoy households at high risk from extreme weather (Hardoy et al. 2001). and Pandiella 2009) (Box 3.1). In 2004, for example, 14 percent of For example, an estimated 1.1 million people live in the favelas of the population in the LAC region (more than 125 million people) Rio de Janeiro that stretch over the slopes of the Tijuca mountain did not have access to improved sanitation, and an even-higher range (Hardoy and Pandiella 2009). Most low-income groups live percentage lacked good quality sanitation and drainage. Limited in housing without air-conditioning or adequate insulation; dur- access to sanitation and freshwater sources is also a key source of ing heat waves, the very young, the elderly, and people in poor vulnerability as this increases the risk of the spread of water-borne health are particularly at risk (Bartlett 2008; see also Section 4.7, diseases (McMichael and Lindgren 2011; McMichael et al. 2012). Human Health). In northern Mexico, for example, heat waves Houses in informal settlements are built incrementally with have been correlated with increases in mortality rates; in Buenos deficient materials and no attention to building or zoning regula- Aires, 10 percent of summer deaths are associated with heat strain; tions. As a result, a significant share of the population is exposed and records show increases in the incidence of diarrhea in Peru to flooding, contamination of groundwater by salt water, and (Mata and Nobre 2006). constraints on the availability and quality of drinking water, as Although LAC has an abundance of freshwater resources, well as to a rising sea level (Magrin et al. 2007). In addition, the many cities depend on local rivers, underground water, lakes, and impacts of extreme weather events are more severe in areas that glaciers that may be affected by climate change (see Section 4.1, have been previously affected and have not yet been able to recover Glacial Retreat and Snowpack Changes and Section 4.2, Water properly, with cumulative effects that are difficult to overcome. Resources, Water Security, and Floods). Considering city growth, Limited disaster preparation and a lack of planning compound environmental deterioration, and possible climate change impacts, the problems (Martí 2006). the supply of fresh water might not be enough to meet demand. A great deal of urban expansion in the region has taken Guadalajara in Mexico (Von Bertrab and Wester 2005) and many place over floodplains, on mountain slopes, and in other zones Andean cities may face increasing water stress and, if the current 21 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal situation continues, low-income groups who already lack adequate and Krellenberg 2011). A similar situation exists in Sao Paulo, access to water will be even less likely to obtain it. Quito is likely where the local catchment of Alto Tiete provides just 10 percent to face water shortages as a result of glacier retreat (Hardoy and of the water supply for 11 million people and where urban areas Pandiella 2009). In Santiago de Chile, an estimated 40 percent are expanding over agricultural and natural areas; this is impact- reduction in precipitation would impact water supply in a city that ing the area’s storm-water retention capacity, thereby making the is expecting a 30 percent population growth by 2030 (Heinrichs city more prone to flood events (Heinrichs and Krellenberg 2011). 2.4.1  Gender and Age-specific Vulnerabilities In the context of a male-dominated, patriarchal society, gender and Box 1.2: The Case of Mexico City age are important aspects of vulnerability in the LAC region. Many women and children are particularly vulnerable to the effects of Mexico City provides a good case study for examining the potential climate change as they have limited access to resources and fewer future impacts of climate change on urban areas in the region. The capabilities and opportunities for participating in decision and main climate-related risk factors in Mexico City stem from increased policy making (Hardoy and Pandiella 2009). The most vulnerable dry-spell periods and heat waves (see Section 3.2, Heat Extremes). groups seldom have an influential voice with regard to disaster Greater Mexico City, with about 20 million inhabitants, is among the most highly populated cities on the planet. Despite a very high preparedness or response, and their needs receive little attention. GDP per capita, the city exhibits a very large income inequality, Economic dependency places women and children in a with about 13 percent of the population lacking enough money to particularly disadvantaged situation, and climate change could meet minimum food needs and approximately 23 percent unable to exacerbate the problem. According to ECLAC and UNFPA (2009), access education or affordable health care (Ibarrarán 2011). poverty is 1.7 times higher among minors under 15 than in adults, According to UN-Habitat, Mexico City is already exposed and 1.15 times greater among women than men. For example, in to several environmental challenges: The urban region is rapidly Uruguay poverty is 3.1 times higher among children than adults; expanding, increasing the demand for space, infrastructure, water, in Chile, it is 1.8 times greater; and, in Nicaragua, it is 1.3 times and energy. This taxes already-deficient water supplies and an inad- greater. Sudden-onset disasters or a worsening of drought condi- equate sewage system. Waste management is similarly challenging, tions have the potential to trigger severe acute malnutrition with as collection, transportation, and adequate final disposal are limited greater effects on women and children (see Section 2.5, Social compared to the daily volume of waste produced by the 20 million Vulnerability to Climate Change). inhabitants of the city. In addition to these preexisting socioeconomic and environmen- There are various ways in which women can be affected by tal vulnerabilities, climate change stressors are projected to increase climate change differently than men. One way is through the rise Mexico City’s overall vulnerability. Four principal climate-related in domestic violence in the context of environmental disasters. stressors are projected to affect Mexico City: (1) the higher fre- Gender-based violence is already a significant problem for women quency of heat waves and hotter days; (2) the decreased instance of in LAC, where most studies estimate that prevalent physical vio- cooler days; (3) the increased occurrence of flash floods; and (4) the lence between intimate partners affects between 20–50 percent of extension of summer droughts (Ibarrarán 2011). An increase in the women. While there are important differences in the estimates, frequency of heat waves could have two potential consequences studies similarly find that 8–26 percent of women and girls report in Mexico City. First, the steadily growing elderly population will be having been sexually abused (Morrison et al. 2004). Moser and particularly exposed as they are more sensitive to heat extremes Rogers (2005) indicate that rapid socioeconomic changes—such than the rest of the population (Gasparrini and Armstrong 2011). as those that can occur as a result of climate shocks—might have Second, in response to heat waves, the population could purchase destabilizing effects within families, leading to an increased risk more air conditioning and cooling systems. This may put power plants under severe stress, particularly as they work less efficiently of domestic violence. Although evidence of the effects of climate- under higher temperatures. The extension of the summer droughts induced disasters in the region remains mixed and limited, accounts is projected to increase Mexico City’s water stress situation (Novelo of gender-based violence have been found in Nicaragua after Hur- and Tapia 2011; Romero Lankao 2010). Furthermore, the increased ricane Mitch, in the Dominican Republic after Tropical Storm Noel, occurrence and extension of summer droughts may disproportion- and in Guatemala after Tropical Storm Agatha (Bizzarri 2012). ately impact the rural population, who may then be more inclined to Some groups of indigenous women are also particularly migrate to cities to find less climate-dependent economic activities vulnerable given their involvement in specific activities. Within (Ibarrarán 2011). As a consequence, the population of Mexico’s indigenous populations in the Colombian Amazon, for example, urban areas is expected to grow, putting more pressure on the impacts on horticulture would affect mainly women as they are urban environments and resources. traditionally in charge of this activity (Kronik and Verner 2010). However, not all gender differences are necessarily worse for women and children. For example, in the case of indigenous 22 Lati n Ame r i ca and the Caribbean groups, impacts in the availability of fish and game will affect urbanization, indigenous populations in LAC are found both in mainly young men (Kronik and Verner 2010). urban and rural areas (Del Popolo and Oyarce 2005). Popolo et Changes in migration patterns as a result of climate change are al. (2009) found that on average 40 percent of the indigenous also likely to have important effects on women. While traditionally population in 11 countries in Latin America were living in urban it is young males who have migrated domestically or abroad, over areas in 2000/2001. Whereas the ratio varies from country to the past two decades rural indigenous women have also started country, recent census information highlights that 21.4 percent of migrating, generally with the support of their social and family indigenous peoples in Colombia (Paz 2012), 54 percent in Bolivia networks. Studies indicate that the experiences of migrant indig- (Molina Barrios et al. 2005), 55.8 percent in Peru (Ribotta 2011), enous women tend to be less favorable, however, as they become 64.8 percent in Chile, (Ribotta 2012a), and 82 percent in Argentina vulnerable and disadvantaged by discrimination, lack of previous (Ribotta 2012b) are currently living in towns and cities. crosscultural experiences, illiteracy, and language barriers. Because Understanding the climate change impacts on indigenous of these barriers, their only option for work tends to be low-wage populations requires an understanding of the cultural dimension employment in the informal sector (Andersen et al. 2010). of their livelihood strategies and the social institutions that sup- port them (Kronik and Verner 2010). In rural areas, indigenous 2.4.2  Indigenous People groups are particularly vulnerable to climate change because of There are about 40 million indigenous people within the LAC their reliance on natural resources, traditional knowledge systems, region, with the majority located in the cooler high regions of and culture (Kronik and Verner 2010) and due to poor access to the Andes and in Mesoamerica (Kronik and Verner 2010). The infrastructure and technology (Feldt 2011). Indigenous popula- indigenous population is made up of about 400 indigenous tions with greater territorial autonomy, and with their livelihoods groups (Del Popolo and Oyarce 2005), of which about 30 percent more intertwined with forest and water resources, are therefore are afrodescendant (Rangel 2006). Bolivia is the country with more affected by climate change when compared to indigenous the highest share of indigenous people (66 percent) and Mexico populations with restricted territorial autonomy (whose livelihoods has the highest absolute number (Table 3.2). When compared to are more diversified, and include wage labor, tourism, and other non-indigenous groups, the profile of indigenous peoples shows income-generating activities) (Kronik and Verner 2010). Indig- that they have higher levels of poverty and infant and maternal enous groups with territorial autonomy are normally located in mortality, lower levels of life expectancy, income, and schooling, the Amazon region whereas those without are more likely to be and less access to water and sanitation; together, this highlights found in the Andes (Kronik and Verner 2010). the exclusion of and discrimination against these groups (Del Kronik and Verner (2010) studied the impacts of climate change Popolo and Oyarce 2005; World Bank 2014). As a response to on indigenous populations in the LAC region, in particular on those Table 1.2: Total Population and Indigenous Population Census 2000 Country and Total Indigenous % Indigenous Census Year Population Population Population Recognized Peoples Groups Bolivia (2001) 8,090,732 5,358,107 66.2 36 groups (49.5% Quechua, 40% Aymara) Brazil (2000) 169,872,856 734,127 0.4 241 groups Costa Rica (2000) 3,810,179 65,548 1.7 Chile (2002) 15,116,435 692,192 4.6 9 groups (83% Mapuche) Ecuador (2001) 12,156,608 830,418 6.8 Guatemala (2002) 11,237,196 4,433,218 39.5 21 groups (all Maya) Honduras (2001) 6,076,885 440,313 7.2 Mexico (2000) 97,014,867 7,618,990 7.9 62 groups Panama (2000) 2,839,177 285,231 10.0 3 groups (Ngöbe-Buglé, Kuna, and Embera-Wounan) Paraguay (2002) 5,183,074 87,568 1.7 Please note that data are from 2000 but used here to provide a comprehensive overview of the share of indigenous population. Sources: Del Popolo and Oyarce (2005); Rangel (2006). 23 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal living in the Amazon, the Andes, the Caribbean, and in Central vulnerability of indigenous groups living in urban areas is more America. In the Colombian Amazon, they found the biggest direct related to the social conditions they face (e.g., discrimination and impacts related to changes in the seasonal cycle (i.e., floods, social exclusion) than linked to their livelihoods. and dry and rainy periods): river flooding affects fish and turtle reproduction, thereby impacting the food security of indigenous 2.4.3  Risk for Populations in Coastal Areas populations; changes in periods during which important local Coastal communities in the region are particularly exposed to fruits ripen and the succession of dry and rainy seasons affect the climate change extremes and sea-level rise (Trab Nielsen 2010). harvests of wild fruits; and changes in the length of the dry season The region’s 64,000 km coastline is one of the most densely affects agriculture productivity, particularly in alluvial plateaus gar- populated in the world (Sale et al. 2008). Coastal states have more dens. Increases in temperature and changes in precipitation affect than 521 million residents, of whom two-thirds (348 million) live mainly horticulture, favoring specific crops such as cassava but within 200 km of the coastline. More than 8.4 million people in threatening harvest diversity (Echeverri 2009). Climate change and LAC live in the path of hurricanes, and roughly 29 million live in climate variability is also apparent in the ‘decoupling’ of ecological low-elevation coastal zones where they are highly vulnerable to markers from seasonal changes (whereby seasons appear to occur sea-level rise, storm surges, and coastal flooding (McGranahan et al. significantly late or early), affecting livelihood decision-making in 2007; UNEP 2007). For example, several countries have a large particular for indigenous peoples. This may affect the credibility section of their urban population living in areas where elevation of elders and traditional leaders, as their authority to predict the is below five meters above sea level (CIESIN 2011). In Belize, the natural seasonality is challenged (Kronik and Verner 2010). Bahamas, Antigua and Barbuda, and Suriname, between 15 and In the indigenous Andes, rising temperatures can increase 62 percent of the urban population live below five meters above demand for water. At the same time, higher evapotranspiration sea level (Table 3.3). This low elevation significantly increases the rates and glacial retreat are expected to reduce the water supply; urban population’s exposure to sea-level rise, storm surges, and restricting pasture land availability in the dry season, and poten- modified tropical storm patterns. tially provoking conflict over land use (Kronik and Verner 2010). Furthermore, human activities such as overfishing, marine In the Bolivian Altiplano, however, Aymara communities have pollution, and coastal development have eroded the ecosystems declared that high-elevation zones have now become productive, in many coastal areas to a level where they no longer provide as changing climatic conditions have turned the area into arable buffers to climate extremes. Climate change and variability are land (Kronik and Verner 2010). likely to compound the damage to ecosystems and to human Indigenous populations in the Andes are not only subjected to settlements, directly through more intense and frequent storms biophysical vulnerability. In the rural Andes, social marginalization and sea-level rise and indirectly through the further degradation and social determinants that limit the ability to improve terms of of the ecosystems (Trab Nielsen 2010). labor, education and access to technical assistance, undermine Coastal communities at greatest risk from climate change the adaptive capacity of the indigenous population (McDowell and variability are generally those that rely on natural resources and Hess 2012). In Palca (Bolivia), for example, farmers are not for a living, occupy marginal lands, and have limited access to only vulnerable due to the retreat of the Mururata glacier and the the livelihood assets that are necessary for building resilience to resulting impact on water supply but also to historical margin- climate change. They include communities that rely on coastal alization due to the lack of official identity cards, land titles, or tourism and on fisheries. They also include much of the region’s access to bilingual (Aymara-Spanish) basic education (McDowell large population of urban slum dwellers (Trab Nielsen 2010). and Hess 2012). More than 50 percent of the Caribbean population lives along In the Caribbean and Central America, an increase in the the coastline, and around 70 percent live in coastal cities (Mimura frequency of some natural disasters (e.g., hurricanes) could limit et al. 2007; UNEP 2008). Many economic activities (e.g., tourism) the access of indigenous populations to key crop, forest, and fish are also concentrated in coastal areas (UNEP 2008). Pressures resources (Kronik and Verner 2010); slow onset changes, meanwhile, arise on the islands over limited land resources as people are could decrease the productivity of traditional varieties of maize, dependent on these natural resources for economic development generating pressure to switch to more commercial varieties (Kronik and their livelihoods. The GDP of the region is generated mainly and Verner 2010). Given that rural areas are mainly populated by from two sectors—tourism and agriculture. Both are highly vul- indigenous groups—especially those that are most remote—means nerable to climate-induced hazards, including flooding, sea-level that they are the most likely to be affected. This is exacerbated by rise, storms, and coastal erosion (Karmalkar et al. 2013). Small the strong dependence of indigenous groups on natural resources islands are especially vulnerable to extreme events (UNEP 2008). as well as by their reliance on traditional farming techniques. The east coast of Mexico and Central America, and the In contrast to the situation of rural indigenous populations, the Caribbean, are strongly affected by wind storms and cyclones 24 Lati n Ame r i ca and the Caribbean Table 1.3: Percentage of Latin American and Caribbean Population Living in Urban Areas and Below Five Meters of Elevation. Urban Population Percentage Percentage of Percentage of Urban in Percentage of of Population Land Area Below Population Living Total Population Living in Informal 5 Meters of Below 5 Meters of Countries (in 2012) Settlements (2005) Elevation Elevation (2010) Caribbean Countries Antigua and Barbuda 29.87 47.9 10.30 15.50 Bahamas, The 84.45 -- 1.61 23.55 Barbados 44.91 -- 0.92 0.92 Belize 44.59 47.3 0.56 17.36 Cuba 75.17 -- 0.38 2.66 Dominica 67.30 -- 1.39 3.05 Dominican Republic 70.21 17.6 0.20 0.90 Grenada 39.49 59.0 1.77 1.92 Haiti 54.64 70.1 0.20 2.44 Jamaica 52.16 60.5 2.05 3.08 St. Kitts and Nevis 32.11 -- 9.25 9.46 St. Lucia 16.97 11.9 0.76 0.84 St. Vincent and the Grenadines 49.70 -- 0.00 0.00 Trinidad and Tobago 13.98 24.7 1.68 2.85 Latin American Countries Argentina 92.64 26.2 0.07 3.29 Bolivia 67.22 50.4 0.00 0.00 Brazil 84.87 28.9 0.06 3.04 Chile 89.35 9.0 0.02 0.65 Colombia 75.57 17.9 0.09 1.35 Costa Rica 65.10 10.9 0.08 0.26 Ecuador 67.98 21.5 0.29 4.68 El Salvador 65.25 28.9 0.10 0.11 Guatemala 50.24 42.9 0.02 0.04 Guyana 28.49 33.7 0.22 11.81 Honduras 52.73 34.9 0.05 0.49 Nicaragua 57.86 45.5 0.03 0.31 Panama 75.78 23.0 0.13 1.90 Paraguay 62.44 17.6 0.00 0.00 Peru 77.58 36.1 0.02 0.81 Suriname 70.12 38.9 0.27 62.04 Uruguay 92.64 -- 0.14 3.65 Venezuela, RB 93.70 32.0 0.16 2.63 Mexico 78.39 14.4 0.15 1.30 Source: Data from CIESIN (2011); UN-HABITAT (2013); and World Bank (2013b). 25 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal (Maynard-Ford et al. 2008). Coastal areas are prone to storm surge of Paraguay, northern Argentina, and southern Bolivia will see floods and sea-level rise (Woodruff et al. 2013). Floods and hur- more pronounced warming, up to 2.5°C in a 2°C world and up to ricanes present both a high risk of death (Dilley et al. 2005) and a 6°C in a 4°C world by 2071–2099. Similar levels of warming are threat to regional development and economic stability in Central projected for eastern Colombia and southern Venezuela. America and the Caribbean (Mimura et al. 2007). The normalized warming (that is, the warming expressed in terms of the local year-to-year natural variability—see Appendix) is 3  Regional Patterns of Climate Change plotted in the lower panels of Figure 3.4. The normalized warming indicates how unusual the projected warming is compared to the 3.1  Projected Temperature Changes natural fluctuations a particular region has experienced in the past, here the period 1951–1980 (Coumou and Robinson 2013; Hansen Figure 3.3 shows projected austral summer (December, Janu- et al. 2012; Mora and Frazier et al. 2013). The tropics will see the ary, February—or DJF) temperatures for the LAC land area (see strongest increase in normalized monthly summer temperatures Appendix). By 2100, summer temperatures over the LAC land area since historic year-to-year fluctuations are relatively small. In the will increase by ~1.5°C under the low-emissions scenario (a 2°C eastern part of the equatorial region between 15°S and 15°N, monthly world) and by ~5.5°C under the high-emissions scenario (a 4°C temperatures will shift by 3–4 standard deviations in a 2°C world world) compared to the 1951–1980 baseline. This is about 0.5°C and by 6–7 standard deviations in a 4°C world. A shift of 3–4 stan- less than the projected global mean land warming which is typical dard deviations implies that an average monthly temperature in the for the Southern Hemisphere (see Figure 2.5 in Schellnhuber et al. future will be as warm as the most extreme monthly temperatures 2013). in a 2°C world, warming of 1.5°C (multi-model mean) is experienced today (i.e., events in the tail of the current distribution). reached by mid-century. Summer temperatures will continue to A shift twice as large (i.e., 6–7 standard deviations) implies that increase beyond mid-century under the high-emissions scenario, extremely cold summer months by the end of the 21st century will causing the multi-model mean warming for the 2071–2099 period be warmer than the warmest months today. Thus, in a 4°C world, to be about 4.5°C (Figure 3.3. and Figure 3.4). monthly summer temperatures in tropical South America will move The regional maps (Figure 3.5) show rather uniform patterns to a new climatic regime by the end of the century. Subtropical of summer warming, with regions in the interior of the continent regions in the south (northern Argentina) and the north (Mexico) generally projected to see a somewhat stronger temperature increase. are expected to see a much less pronounced shift. Nevertheless, Along the Atlantic coast of Brazil, Uruguay, and Argentina, the a shift by at least 1-sigma (in a 2°C world) or 2-sigma (in a 4°C warming remains limited, with about 0.5–1.5°C in a 2°C world world) is projected to occur here over the 21st century. and 2–4°C in a 4°C world. The central South American region 3.2  Heat Extremes Figure 1.3: Temperature projections for the Latin American Figure 3.5 and Figure 3.6 show a strong increase in the frequency and Caribbean land area compared to the 1951–1980 of austral summer months (DJF) warmer than 3-sigma and 5-sigma baseline for the multi-model mean (thick line) and individual (see Appendix) over LAC by the end of the century (2071–2099). The models (thin lines) under RCP2.6 (2°C world) and RCP8.5 (4°C tropics, which are characterized by relatively small natural variability, world) for the months of DJF. will see the largest increase in such threshold-exceeding extremes. Especially along the tropical coasts, including Peru, Ecuador, and Colombia, summer month heat extremes will become much more frequent, consistent with the large shift in the normalized temperature distribution here (see Figure 3.5). The 5-sigma events, which are absent under present-day climate conditions, will emerge in these countries even in a 2°C world, and are projected to occur in roughly 20 percent of summer months. At the same time, 3-sigma events, which are extremely rare today, will become the new norm (i.e., this threshold will be exceeded in roughly half of the summer months during 2071–2099). In a 4°C world, almost all summer months will be warmer than 3-sigma and, in fact, most will be warmer than 5-sigma as well (70 percent). Thus, under this scenario, the climate in tropical South America will have shifted to a new hot regime. Compared to the tropics, the subtropical regions in the north The multi-model mean has been smoothed to give the climatological trend. (Mexico) and south (Uruguay, Argentina, and southern Chile) 26 Lati n Ame r i ca and the Caribbean Figure 1.4: Multi-model mean temperature anomaly for Latin America and the Caribbean for RCP2.6 (2°C world, left) and RCP8.5 (4°C world, right) for the austral summer months (DJF). Temperature anomalies in degrees Celsius (top row) are averaged over the time period 2071–2099 relative to 1951–1980, and normalized by the local standard deviation (bottom row). are projected to see a more moderate increase in the frequency warmer than 3-sigma by 2071–2099 (i.e., this will have become of threshold exceeding extremes. In fact, in a 2°C world, 5-sigma the new norm). Furthermore, 5-sigma events will also emerge events will remain absent and 3-sigma events will still be rare (less and occur typically in about 20 percent of summer months over than 10 percent of summer months). In a 4°C world, however, a subtropical regions. substantial increase in frequency is projected. In most subtropi- The strong increase in frequency of summer months warmer cal regions, at least half of all summer months are expected to be than 3- and 5-sigma in the tropics, as reported here, is quantitatively 27 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal Figure 1.5: Multi-model mean of the percentage of austral summer months (DJF) in the time period 2071–2099 with temperatures greater than 3-sigma (top row) and 5-sigma (bottom row) for scenario RCP2.6 (2°C world, left) and RCP8.5 (4°C world, right) over Latin America and the Caribbean. consistent with published results from analyses using the full The duration of warm spells is projected to increase most in the CMIP5 dataset of climate projections (Coumou and Robinson 2013; tropics—already by 60–90 days in a 2°C world and by 250–300 days Sillmann et al. 2013a; b). In addition, minimum night-time and in a 4°C world. In the tropics, temperatures experienced during the maximum day-time temperatures in the summer are projected to 10 percent warmest summer nights of the 1961–1990 period will increase by 1–2°C in a 2°C world and by 5–6°C in a 4°C world occur most nights (50–70 percent) in a 2°C world and almost all (Sillmann et al. 2013b), in good agreement with the projected nights (90–100 percent) in a 4°C world by the end of the century seasonal mean temperatures in Figure 3.3. (Sillmann et al. 2013b). 28 Lati n Ame r i ca and the Caribbean Figure 1.6: Multi-model mean and individual models of exhibit substantial disagreement on the direction of change over the percentage of Latin American and Caribbean land area most land regions. With a more pronounced climatic signal (i.e., a warmer than 3-sigma (top) and 5-sigma (bottom) during 4°C world, RCP8.5), the models converge in their projections over austral summer months (DJF) for scenarios RCP2.6 (2°C most regions, but inter-model uncertainty remains over some areas world) and RCP8.5 (4°C world). (hatched shading in the maps). Nevertheless, a well-defined pat- tern of change in annual precipitation can be extracted for defined sub regions. For example, tropical countries on the Pacific coast (Peru, Ecuador, and Colombia) are projected to see an increase in annual mean precipitation of about 30 percent. This enhanced rainfall occurs year-round and can be detected in both the austral winter and summer seasons. Similarly, Uruguay on the Atlantic coast (and bordering regions in Brazil and Argentina) are projected to get wetter. Again, this increase in annual rainfall is year-round, though it is most pronounced during the summer (DJF). Regions which are projected to become drier include Patagonia (southern Argentina and Chile), Mexico, and central Brazil. These patterns indicate that, under climate change, most dry regions may get drier and most wet regions may get wetter in the future (but see Greve et al. 2014 for a discussion of this concept for past climate). The exception is central Brazil (i.e. the region from 0–20°S and 50–65°W), which contains the southeastern part of the Amazon rainforest. The annual mean precipitation here is projected to drop by 20 percent in a 4°C world by the end of the century. This drop in annual rainfall is entirely due to a strong and robust decrease in winter (JJA) precipitation (–50 percent), with essentially no change in summer (DJF) precipitation. In fact, this reduction in winter precipitation appears already in a 2°C world. These projected changes in annual and seasonal temperatures generally agree well with those provided by the IPCC AR5 based on the full set of CMIP5 climate models (Collins et al. 2013). However, there is one important difference in that the full set of CMIP5 models shows significant JJA Amazon drying over northern Changes in heat extremes in subtropical regions are less dramatic Brazil only. Over central Brazil, the multi-model mean of the full but nevertheless pronounced. In the Southern Hemisphere subtrop- set of CMIP5 models projects drying, as also seen in Figure 3.7, ics, the length of warm spells increases by roughly 0–15 days (2°C but the magnitude of change is small. Instead, significant drying world) or 30–90 days (4°C world). In the Northern Hemisphere over the full Amazon region primarily occurs during austral spring subtropics (Mexico) these values roughly double, but they are still (September-October-November). less than the increase in the tropics (Sillmann et al. 2013b). Night- time temperatures experienced during the 10 percent warmest austral 3.4  Extreme Precipitation and Droughts summer nights in 1961–1990 will occur in roughly 30 percent of nights (2°C world) and 65 percent of nights (4°C world). Analysis of the observational record since the 1950s indicates a robust increase in overall precipitation and in intensity of extreme 3.3  Regional Precipitation Projections precipitation events for South America, particularly over southern South America and the Amazon region (Skansi et al. 2013). Long- Projected changes in annual and seasonal precipitation (see Appen- term trends in meteorological droughts are not statistically robust dix) are plotted in Figure 3.7 for the LAC region for 2071–2099 over 1950–2010. Over the recent decade, however, two severe relative to 1951–1980. Note that projections are given as percent- droughts (2005 and 2010) have affected the Amazon, likely con- age changes compared to the 1951–1980 climatology and thus, nected to an anomalous warm tropical North Atlantic (Marengo especially over dry regions, large relative changes do not neces- et al. 2011; Zeng et al. 2008). sarily reflect large absolute changes. In general, in a 2°C world Dai (2012) finds a statistical significant increase in drought these changes are relatively small (+/–10 percent) and models conditions for Central America and the Caribbean for the 1950–2010 29 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal Figure 1.7: Multi-model mean of the percentage change in austral summer (DJF, top), winter (JJA, middle) and annual (bottom) precipitation for RCP2.6 (2°C world, left) and RCP8.5 (4°C world, right) for Latin America and the Caribbean by 2071–2099 relative to 1951–1980. Hatched areas indicate uncertain results, with two or more out of five models disagreeing on the direction of change. 30 Lati n Ame r i ca and the Caribbean period, although the significance of this trend depends on the refer- except the southern coast, southern Chile, and Central America, ence period and the formulation of the underlying drought index and in particular northern Mexico, is expected to be under severe (Trenberth et al. 2014). Fu et al. (2013) report a significant increase to extreme drought conditions relative to the present climate by in the length of the dry season over southern Amazonia since 1979. the end of the 21st century under the RCP4.5. These results are Using an ensemble of CMIP5 models, Kharin et al. (2013) inves- confirmed by a multi-model impact analysis under a 4°C scenario tigated extreme precipitation events based on annual maximum that also reveals a strong increase in drought risk in the Caribbean, daily precipitation with 20-year return values. In a 4°C world, although uncertainties remain substantial (Prudhomme et al. 2013). these events are found to intensify by about 25 percent over LAC The increase in future drought risk in Central America and the with a large uncertainty range.31 In addition, the return time of a Caribbean is generally related to an extension and intensification 20-year extreme precipitation event from the 1985–2005 period of the so-called midsummer drought period (Rauscher et al. 2008). would reduce to about 6 years by the end of the 21st century While an overall reduction in precipitation during the dry season is (2081–2100) in a 4°C world (Kharin et al. 2013). robustly projected by regional and global models alike (Campbell These increases are not, however, homogeneous over the full et al. 2011; Karmalkar et al. 2013; Taylor et al. 2013), it is not clear continent. This is consistent with the variable seasonal precipita- if this will lead to an increase in meteorological drought conditions tion projections in Figure 3.7. While little-to-not statistically sig- (e.g., an increase in the number of consecutive dry days) (Hall nificant, an increase in frequency is projected for the Caribbean, et al. 2012). This illustrates the added value of a comprehensive Meso-America, Southern Argentina, and Chile, and hotspots impact model analysis, as undertaken by Prudhomme et al. (2013), with extreme precipitation increases of more than 30 percent are that accounts for the full change in the regional water cycle in projected in the Serra do Espinhaco in Brazil, the Pampas region investigating future drought risk. in Argentina, and the Pacific coastline of Ecuador, Peru, and Changes over the Amazon basin and eastern Brazil are found Colombia (Kharin et al. 2013). The latter may be related to an to be particularly pronounced during the dry season (from July to increase in frequency of future extreme El Niño events (Cai et al. September), which amplifies the risk of large-scale forest degradation; 2014; Power et al. 2013). These regions are also found to show this contrasts with Central America, Venezuela and southern Chile, the strongest rise in compound maximum 5-day precipitation where the drought risk is projected to increase during the austral sum- (which is most relevant for flooding events) by the end of the mer (Prudhomme et al. 2013). Drought risks are found to be strongly 21st century in a 4°C world (Sillmann et al. 2013b). Increases in scenario-dependent and to be much less pronounced in a 2°C world, extreme precipitation in southern Brazil and northern Argentina in particular for Meso-America and the Caribbean; a substantial risk are in line with results from regional climate models (Marengo remains for South America under this scenario (Prudhomme et al. 2013). et al. 2009) and might be dominated by intensification of the In the Amazon region, climate change is not the only anthropo- South American monsoon system (Jones and Carvalho 2013). genic interference expected over the decades to come; deforestation Projections of extreme precipitation for Meso-America and the will be at least equally important. The link between large-scale Caribbean discussed above do not comprehensively account for deforestation and reduced precipitation is well established (e.g., the risk of extreme precipitation related to tropical cyclones (that Davidson et al. 2012; Medvigy et al. 2011; Runyan 2012), and Bagley are discussed in Section 3.6, Tropical Cyclones/Hurricanes). In a et al. (2014) used a regional climate model to demonstrate that 2°world, changes in heavy precipitation would be greatly reduced deforestation might have amplified the severe droughts over the last and barely significant over most parts of the continent. decade. However, none of the projections given above accounts for While an increase in extreme precipitation represents a poten- the possible adverse effects of deforestation and forest degradation on tial threat for some regions, increase in duration and intensity of the climate of the Amazon region, which, in the presence of possible droughts might represent the bigger threat over all of Latin America self-amplifying feedbacks between reduced forest cover and extreme and the Caribbean. An increase and intensification in meteorological droughts, represent a substantial risk of large-scale Amazon dieback droughts is projected for large parts of South and Central America (see Section 4.5, Amazon Rainforest Dieback and Tipping Point). in a 4°C world (Sillmann et al. 2013b), although large model 3.5 Aridity uncertainties remain in particular for Central America (Orlowsky and Seneviratne 2013). A more comprehensive analysis of future Apart from a reduction in precipitation, warming can also cause droughts accounting for the effects of runoff and evaporation as a region to shift toward more arid conditions as enhanced surface well as local soil and vegetation properties was undertaken by Dai temperatures trigger more evapotranspiration—thereby drying the (2012). He found that the Amazon basin, the full land area of Brazil soil. This long-term balance between water supply and demand is captured by the aridity index (AI), which is shown in Figure 3.9 for The lower and upper limits of the central 50 percent inter-model range are 14 and 31  the Latin American region. The AI is defined as the total annual 42 percent respectively. precipitation divided by the annual potential evapotranspiration 31 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal Figure 1.8: Multi-model mean of the percentage change in the annual-mean of monthly potential evapotranspiration for RCP2.6 (2°C world, left) and RCP8.5 (4°C world, right) for Latin America and the Caribbean by 2071–2099 relative to 1951–1980. Hatched areas indicate uncertainty regions with two or more out of five models disagreeing on the direction of change. (see Appendix); it fundamentally determines whether ecosystems model uncertainty in these regions and no robust statements can and agricultural systems are able to thrive in a certain area. A be made about whether conditions will become more or less arid. A decrease in the value of the AI thus indicates that water becomes prime reason for this is that both annual precipitation and potential more scarce (i.e., more arid conditions), with areas classified as evapotranspiration in these regions have upward trends, and it is hyper-arid, arid, semi-arid, and sub-humid as specified in Table 3.4. the relative magnitude of these trends which determines whether Potential evapotranspiration is a measure of the amount of water a region becomes more or less arid. In other words, it is unclear a representative crop type would need over a year to grow, (i.e. whether or not warming-driven drying outpaces the increase in a standardized measure of water demand; see Appendix). Under annual precipitation projected for these regions. most circumstances, changes in potential evaporation are governed Outside these uncertain regions, the LAC land area is projected to by changes in temperature. become more arid other than for an isolated region in the southern Changes in annual-mean potential evapotranspiration (Fig- tip of Chile. Again three major drying regions can be identified: ure 3.8) broadly follow those of absolute warming (Figure 3.4), (1) Northern Hemisphere subtropics (Mexico); (2) the interior of with regions in the continental interior generally experiencing the South American continent (southern Amazonia, Bolivia, and the strongest increase. Thus, though potential evapotranspiration Paraguay); and (3) central Chile and Patagonia. Over the first two depends on several meteorological variables, it seems primarily regions, the AI is projected to decrease by up to 20 percent in a 2°C driven by future temperature changes. The signal is weak in a 2°C world and up to 40 percent in a 4°C world. For the third region, world, with relative changes in potential evapotranspiration smaller the decrease in AI is especially pronounced in a 4°C world, drop- than 20 percent everywhere except for some isolated regions in the ping up to 60 percent (note that this region is already arid today). continental interior. In a 4°C world, changes become much more The shift in AI in Figure 3.9 causes some regions to be clas- pronounced, with countries like Paraguay and Bolivia projected to sified in a different aridity class (see Table 3.4). The total area of see an increase in potential evapotranspiration of up to 50 percent. land classified as either hyper-arid, arid, or semi-arid is projected Consistent with this result, these regions are also projected to see to grow from about 33 percent in 1951–1980 to 36 percent in a 2°C the strongest absolute warming (Figure 3.4). world (i.e., an increase of nearly 10 percent) and to 41 percent in Over almost all of the LAC land area, the multi-model mean a 4°C world (i.e., an increase of nearly 25 percent). projects more arid conditions under future climate change. Still, over extended areas, notably near the equator, at the Pacific tropical 3.6  Tropical Cyclones/Hurricanes coast (Peru), and at the sub-tropical Atlantic coast (southern Brazil, Uruguay, and northern Argentina), the AI changes little and mod- In Central America and the Caribbean, tropical cyclones occur els disagree on the direction of change. Thus, there is substantial regularly and have severe impacts, especially when making landfall; 32 Lati n Ame r i ca and the Caribbean Figure 1.9: Multi-model mean of the percentage change in the aridity index under RCP2.6 (2°C world, left) and RCP8.5 (4°C world, right) for Latin America and the Caribbean by 2071–2099 relative to 1951–1980. Hatched areas indicate uncertain results, with two or more out of five models disagreeing on the direction of change. Note that a negative change cor- responds to a shift to more arid conditions.32 Table 1.4: Multi-model mean of the percentage of land area gas concentrations and/or aerosol concentrations, increases the in Latin America and the Caribbean which is classified as amount of heat that is absorbed by the atmosphere and can increase hyper-arid, arid, semi-arid and sub-humid for 1951–1980 and the potential for tropical cyclones to form. Changes in the frequency 2071–2099 for both the low (2°C world, RCP2.6) and high (4°C and intensity of tropical storms are modulated, however, by other world, RCP8.5) emissions scenarios. factors, including vertical wind shear and humidity. Of particular importance is the vertical wind shear (i.e., the difference between 2071–2099 2071–2099 1951–1980 (RCP2.6) (RCP8.5) wind speeds near the surface and higher up in the troposphere). High wind shear disrupts the process of tropical cyclone formation and Hyper-Arid 8.6 10.1 12.8 intensification, so that increases in wind shear counter increases in Arid 10.3 11.2 12.7 sea surface temperature—impacting tropical cyclone formation and Semi-Arid 14.3 14.8 15.9 intensity. El Niño events (see Section 2.3.2 of the full report, El-Niño/ Sub-Humid 5.5 5.8 6.0 Southern Oscillation) tend to enhance wind shear over the Gulf of Mexico and the Caribbean Sea and thus suppress Atlantic tropical cyclones (Aiyyer and Thorncroft 2011; Arndt et al. 2010; Kim et al. 2011). On the other hand, El Niño events have been shown to increase the impacts on marine ecosystems, transport, and infrastructure tropical cyclone activity in the eastern North Pacific (Kim et al. 2011; can also be severe. Strobl (2012), for instance, derived an average Martinez-Sanchez and Cavazos 2014). Observational evidence, how- 0.83 percent drop in economic output after tropical cyclones strikes ever, suggests atmospheric patterns tend to steer tropical cyclones in this region, with big variations between countries.32 away from the Mexican coast during El Niño years (and toward the The energy of tropical cyclones is derived from the ocean surface coast in La Niña years), so that the net effect on the Pacific coastlines and lower atmosphere. Warming, along with increased greenhouse of the Americas remains unclear. In both regions, changes in the El-Niño/Southern-Oscillation (ENSO) due to climate change and the Some individual grid cells have noticeably different values than their direct neighbors 32  associated uncertainties affect tropical cyclone projections. In addi- (e.g., on the border between Peru and Bolivia). This is due to the fact that the Aridity tion to such dynamic changes, thermodynamic processes alone can index is defined as a fraction of total annual precipitation divided by potential evapo- also work to suppress tropical cyclone formation and intensification transpiration (see Appendix). It therefore behaves in a strongly non-linear fashion, and thus year-to-year fluctuations can be large. Since averages are calculated over a (Mallard et al. 2013). relatively small number of model simulations, this can result in these local jumps. 33 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal These factors make projecting changes in tropical cyclone Short-term variability in tropical cyclones is large and GCM frequency and intensity difficult. The recent IPCC WGI AR5 report resolution too low to resolve high-intensity tropical cyclone struc- found in relation to observed changes that “there is low confidence tures. Projections therefore rely on proxies of tropical cyclone char- in attribution of changes in tropical cyclone activity to human influ- acteristics, or a cascade of low- to high-resolution models. Using ence owing to insufficient observational evidence, lack of physical CMIP5 models (50 percent uncertainty range across 17 GCMs), understanding of the links between anthropogenic drivers of cli- Villarini and Vecchi (2013) projected that the Power Dissipation mate and tropical cyclone activity, and the low level of agreement Index would increase by 100–150 percent in a 2°C world over the between studies as to the relative importance of internal variability, North Atlantic. A considerably larger increase and a much wider and anthropogenic and natural forcings” (Bindoff et al. 2013). range of about 125–275 percent were projected for a 4°C world. Observational records show little or no historical global trend Bender et al. (2010) used a variety of models to initialize a very in tropical cyclone frequency or intensity, in particular in light of high-resolution operational hurricane-prediction model, noting an uncertainties resulting from the potential undercounting of tropical increase of 80 percent in the frequency of the strongest category 4 cyclones in early parts of the record predating satellite observations and 5 Atlantic tropical cyclones in a 4°C world (compared to the (before about 1970). The North Atlantic, however, is an exception. present.) Knutson et al. (2013) also found an 80 percent increase Tropical cyclone frequency has increased in the North Atlantic in the strongest category tropical cyclones for the same scenario sharply over the past 20–30 years, but uncertainty is large over and class of models and around a 40 percent increase for a lower longer time-periods (Bindoff et al. 2013). Emanuel (2008) noted an emissions scenario and for the most recent generation of global increase in the Power Dissipation Index (a combination of frequency models included in IPCC AR5 (2013b) at roughly 1.5–2.5°C warm- and intensity) of North Atlantic tropical cyclones over a 1949–2004 ing (early and late 21st century RCP4.5). The largest increase (see observational period. Using a new record of observations, Kossin Figure 3.10) occurred in the Western Atlantic, north of 20°N (i.e., et al. (2013) showed a strong and statistically significant increase to the north of Haiti), a pattern confirmed by Emanuel (2013). in lifetime maximum intensity of tropical cyclones over the North Using a very different statistical downscaling method, Grinsted et al. Atlantic of 8 m.s–1 per decade, over the period 1979–2010, particu- (2013) projected a twofold to sevenfold increase in the frequency larly for mid- to high-intensity storms. This can be compared to the of “Katrina magnitude events” regarding storm surge (not wind median lifetime maximum intensity of around 50 m.s–1 of tropical speed) for a 1°C rise in global temperature. For context: in terms cyclones across the region in this historical time series. Such observed of wind speeds, the 2005 Gulf of Mexico tropical cyclone Katrina changes were shown to be linked to both anthropogenic climate was a class 5 tropical cyclone. change and internal climate variability (Camargo et al. 2012; Villarini The eastern North Pacific is less well represented in the scien- and Vecchi 2013; Wang and Wu 2013). Differential warming of the tific literature. Based on variations of one high-resolution model, tropical Atlantic, with historically observed warming higher than Murakami et al. (2012, 2011) projected no significant trends for average for the tropics, tends to enhance tropical cyclone intensi- this region under future climate change. By contrast, Emanuel fication in this region (Knutson et al. 2013). No significant trends (2013), using an ensemble of 6 different CMIP5 models, projected have been observed over the Eastern North Pacific (Kossin et al. an increase in frequency of tropical cyclones along the Pacific coast 2013). In general however, tropical cyclones haven been observed of Central America (particularly large near the coast of southeast to migrate polewards (Kossin et al. 2014). Mexico); the author notes, however, that the method does not In the long term, model simulations from a range of models capture well the currently observed storm frequency in this region. lead to the expectation that tropical cyclone frequency will not With projected increased intensity and frequency of the most be affected much by continued global warming but that mean intense storms, and increased atmospheric moisture content, intensity, as well as the frequency of the most intense tropical Knutson et al. (2013) estimated an increase of 10 percent in the cyclones, are projected to increase (Knutson et al. 2010; Tory et al. rainfall intensity averaged over a 200 km radius from the tropical 2013). IPCC AR5 WGI found that: cyclone center for the Atlantic, and an increase of 20–30 percent for the tropical cyclone’s inner core, by the end of the 21st century “Projections for the 21st century indicate that it is likely for roughly 2.5–3.5°C global warming. This confirms the earlier that the global frequency of tropical cyclones will either results of Knutson et al. (2010) reported in IPCC AR5 WGI (2013b). decrease or remain essentially unchanged, concurrent This effect would greatly increase the risk of freshwater flooding with a likely increase in both global mean tropical cyclone from tropical cyclones making landfall. maximum wind speed and rain rates . . . The influence of The projections above focus on changes in frequency of tropi- future climate change on tropical cyclones is likely to vary cal cyclones and wind and rainfall intensity. Colbert et al. (2013) by region, but there is low confidence in region-specific projected a shift in tropical cyclone migration tracks in the tropi- projections. The frequency of the most intense storms cal North Atlantic, with more frequent ocean recurving tropical will more likely than not increase substantially in some cyclones and fewer cyclones moving straight westward toward land. basins.” (Stocker et al. 2013) 34 Lati n Ame r i ca and the Caribbean Figure 1.10: Change in average rate of occurrence of Category 4 and 5 tropical cyclones per hurricane season (August–October) at about 2.5°C warming globally above pre-industrial levels by the end of the 21st century compared to the present-day. Source: Knutson et al (2013). Together with changes in genesis area, but assuming constant tropi- on the climate-driven factors of local heat uptake of the ocean, cal cyclone frequency, this lead to a projected increase in tropical ocean current changes, and the far-reaching influence of changing cyclones per season in the central Atlantic and a decrease in the gravity from the ice sheets. In LAC, regional sea-level rise largely Gulf of Mexico and the Caribbean. Murakami and Wang (2010), reflects the rise in global mean sea level. Still, consistent regional however, found no change in trajectories. Current literature does features exist across both the 1.5°C and 4°C world scenarios in not provide evidence for a (change in) risk of synchronized landfall both the median and high estimate (Figure 3.11, Table 3.5). These of different tropical cyclones (e.g., in Central America from both features are more pronounced for stronger overall sea-level rise. the Pacific and the Atlantic). While individual tropical cyclones may not be strong, their compound impact may be more severe. Moreover, any increase in trend of Pacific and Atlantic storms Table 1.5: Sea-level rise between 1986–2005 and 2081–2100 (not necessarily cyclones) making landfall simultaneously would for the RCP2.6 (1.5°C world) and RCP8.5 (4°C world) in selected locations of the LAC region (in meters). potentially entail more damaging impacts than increasing frequency of any individual Pacific or Atlantic cyclone alone. RCP2.6 (1.5°C world) RCP8.5 (4°C world) In summary, observations show historical positive trends in Acapulco 0.38 (0.23, 0.61) 0.6 (0.42, 1.01) tropical cyclone frequency and strength over the North Atlantic Antofagasta 0.37 (0.22, 0.58) 0.58 (0.42, 0.98) but not over the eastern North Pacific. While Atlantic tropical Barranquilla 0.39 (0.22, 0.65) 0.65 (0.43, 1.12) cyclones are suppressed by the El Niño phase of ENSO, they are Buenos Aires 0.34 (0.24, 0.52) 0.56 (0.45, 0.97) enhanced in the eastern North Pacific. Under further anthropogenic climate change, the frequency of high-intensity tropical cyclones Cristobal 0.39 (0.22, 0.65) 0.66 (0.44, 1.07) is generally projected to increase over the western North Atlantic Guayaquil 0.39 (0.25, 0.62) 0.62 (0.46, 1.04) by 40 percent for 1.5–2.5°C global warming and by 80 percent in Lima 0.38 (0.24, 0.61) 0.6 (0.45, 1.02) a 4°C world. Global warming around 3°C is associated with an Port-au-Prince 0.38 (0.21, 0.61) 0.61 (0.41, 1.04) average 10 percent increase in rainfall intensity averaged over a Puerto Williams 0.27 (0.19, 0.37) 0.46 (0.38, 0.65) 200 km radius from the tropical cyclone center. Although there is Recife 0.39 (0.23, 0.65) 0.63 (0.41, 1.14) some evidence from multiple-model studies for a projected increase Rio de Janeiro 0.37 (0.24, 0.61) 0.62 (0.46, 1.11) in frequency of tropical cyclones along the Pacific coast of Central America, overall projections in this region are currently inconclusive. Tumaco 0.38 (0.24, 0.6) 0.61 (0.44, 1.01) Valparaiso 0.35 (0.21, 0.54) 0.55 (0.41, 0.91) 3.7  Regional Sea-level Rise Numbers in parentheses indicate low and high bounds (see Section 6.2, Sea-Level Rise Projections for an explanation of the 1.5° world). Regional sea-level rise will vary in a large and geographically diverse region such as Latin America and the Caribbean, and will depend 35 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal Figure 1.11: Patterns of regional sea-level rise. Median (left column) and high (right column) estimates of projected regional sea-level rise for the RCP2.6 scenario (1.5°C world, top row) and the RCP8.5 scenario (4°C world, bottom row) for the period 2081–2100 relative to the reference period 1986–2005. Associated global mean rise is indicated in the panel titles. Representative cities are denoted by black dots and discussed in the text with numbers provided in Sea-level rise is projected to be higher at the Atlantic coast estimate: 0.62 m for a 4°C world), where a large number of people than at the Pacific coast. Valparaiso is projected to benefit from are at risk (Nicoldi and Mueller Petermann 2010). this effect (median estimate: 0.55 m for a 4°C world), where Sea-level rise is enhanced at low latitudes due to both increased southeasterly trade wind intensification over the Southern Pacific ocean heat uptake in the region (Figure 3.12; middle) and the and associated upwelling of cold water (Merrifield and Maltrud gravity-induced pattern of ice sheets and glaciers (Figure 3.12 bot- 2011; Timmermann et al. 2010) lead to below-average thermoste- tom). As an example, Guayaquil on the Pacific Coast of Ecuador is ric sea-level rise (Figure 3.13). In contrast, Recife on the Atlantic projected to experience a sea-level rise of 0.62 m (median estimate; coast of Brazil is projected to experience above-average sea-level low estimate: 0.46 m; high estimate: 1.04 m) of sea-level rise in rise (median estimate: 0.63 m for a 4°C world). A similar rise a 4°C world (Figure 3.13). Guayaquil is among the most vulner- is projected for the Rio de Janeiro region (Figure 3.11) (median able coastal cities in terms of relative GDP losses (Hallegatte et al. 36 Lati n Ame r i ca and the Caribbean Figure 1.12: Regional anomaly pattern and its contributions in 2013). In contrast, Puerto Williams (Chile) at the southern tip the median RCP8.5 scenario (4°C world). of the South American continent is projected to experience only 0.46 m (median estimate for a a 4°C world) (low estimate: 0.38 m; high estimate: 0.65 m). The upper bound differs between the two locations since it is largely determined by the risk of high ice-sheet-driven sea-level rise. Port-Au-Prince (Haiti) is projected to experience 0.61 m (low estimate: 0.41 m, high estimate: 1.04 m) of sea-level rise in a 4°C world (Figure 3.13); it serves as a typical example for sea-level rise in other Caribbean islands. The sea-level rise at the conti- nental Caribbean Coast exceeds the projection for the Caribbean islands (Barranquilla, median estimate: 0.65; low estimate: 0.43; and high estimate: 1.12 in a 4°C world). The difference may be linked to a weakening of the Caribbean Current that is connected to the Atlantic meridional overturning circulation33� (Pardaens et al. 2011). The high upper bounds are due to the strong influence of the Antarctic ice sheet. The ocean is predicted to warm, with high rates in the Southern Ocean off Buenos Aires (Kuhlbrodt and Gregory 2012). The coastal waters are, however, dominated by the cold Malvinas Current, a branch of the Antarctic Circumpolar Current (ACC). Since minor warming is projected for the ACC and the Malvinas Current, the strong warming signal in the Southern Ocean does not lead to additional sea-level rise at the Rio de la Plata estuary—which is projected to rise slower than the global mean (Buenos Aires, median estimate: 0.56 m; low estimate: 0.42 m; and high estimate: 1.03 m in a 4°C world). Sea-level rise in the region will be influenced by the future strength of the Brazil and Malvinas Currents and the position of their confluence zone (Lumpkin and Garzoli 2011). 4  Regional Impacts 4.1  Glacial Retreat and Snowpack Changes 4.1.1  Topography of Glaciers in the Andes The Andes are the longest continental mountain range in the world, stretching about 7,000 km along the coast of South America. There are major ice masses in the Patagonian Andes and on Terra del Fuego in the Southern Andes. A much smaller amount of ice (about 217 Gt), covering an area of about 4,900 km2, is stored in the Central Andes; this region hosts more than 99 percent of the world’s glaciers that are located in tropical latitudes (Tropical Andes). These glaciers exist at high altitudes between 4,000 and 6,500 m above sea level and are of crucial importance for the liveli- hood of the local populations as they act as critical buffers against Total sea-level rise (top), steric-dynamic (middle), and land-ice (bottom) highly seasonal precipitation and provide water during the dry contributions to sea-level rise, shown as anomalies with respect to the global mean sea-level rise. Global mean contributions to be added on top of the spatial anomalies are indicated in the panel titles. This ocean current system transports a substantial amount of heat energy from 33  the tropics and Southern Hemisphere toward the North Atlantic, where the heat is then transferred to the atmosphere. Changes in this ocean circulation could have a profound impact on many aspects of the global climate system. 37 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal Figure 1.13: Sea level projections for selected cities. Time series for sea-level rise for the two scenarios, RCP2.6 (1.5°C world, blue) and RCP8.5 (4°C world, green). Median estimates are given as full thick lines and the lower and upper bound given as shading. Full thin lines are global median sea-level rise with dashed lines as lower and upper bound. Vertical and horizontal black lines indicate the reference period and reference (zero) level. 38 Lati n Ame r i ca and the Caribbean season for domestic, agricultural, and industrial use. In addition, Observed Glacier Recession the water and energy supplies (see Section 4.11) of the capital cities As a general trend, the Andean glaciers are shrinking. This is caused of Lima (Peru), La Paz (Bolivia), and Quito (Ecuador) depend on by increased melt rates, decreased accumulation, changes in the the glacial melt water (see Section 4.2, Water Resources, Water ice dynamics, and/or a combination of all these factors. For the Security, and Floods). Andean snowpack is also a crucial natural period 1980–2011, Giesen and Oerlemans (2013) calculated, for the water resource, particularly in the semi-arid regions of Southern tropical glaciers, a relative volume change of 7.3 percent (39 Gt) South America (Masiokas et al. 2012, 2013). with respect to a total glaciated area of 4,940 km2 (99 percent of which is located in the Central Andes). Over a much longer 4.1.2  Current Situation and Observed Changes historical period (1901–2009) and for a larger region (account- Characteristics of Tropical Glaciers ing for 82 percent of all tropical glaciers), Marzeion et al. (2012) Tropical glaciers are particularly threatened by climate change due estimated an area reduction of glaciers of about 79±2 percent to their high altitude, the high level of radiation, and the tropical (15,900±500 km2), which corresponds to a volumetric ice loss climate dynamics. Observations have shown that the variability of 90 percent (1,740 Gt). The Southern Andes contain a much of the surface temperature of the Pacific Ocean is the governing larger ice mass of 11,430 Gt (1980) extending over a glaciated factor, explaining the dramatic glacier recession of the 20th century, area of 33,700 km2. This huge ice mass decreased in volume by although the precipitation trend has not been significant during that 6.1 percent (695 Gt) between 1980–2011 (Giesen and Oerlemans period. The impact of the ENSO phenomenon (see Section 2.3.2 of 2013). Over the 20th century (1901–2009), Marzeion et al. (2012) the full report, El-Nino/Southern Oscillation) on the inter-annual infer a reduction of about 32 percent in area (15,500±200 km2), mass balance is consequently high in the tropical glacier zone, which can be associated with a volume loss of 22±5 percent with low temperatures, high precipitation, high wind speeds, high (1,340±290 Gt). albedo, and a nearly balanced or positive mass balance during La These global projections, however, rely on coarse resolution Niña events and a strongly negative mass balance during El Niño models that cannot adequately simulate glacier dynamics in the events (Chevallier et al. 2011). steep topography of the narrow mountain chain of the Andes. In particular, the strong recession rate of the comparably small tropical glaciers is likely overestimated by the global-scale methodology. Regional models, in contrast, can provide more comprehensive analyses with resolutions of up to a few hundred meters. Figure 1.14: Compilation of mean annual area loss rates for Rabatel et al. (2013) reviewed the various studies on the current different time periods for glaciated areas between Venezuela state of Tropical Andes’ glaciers, considering a variety of different and Bolivia. measurement techniques (e.g., monitoring of the mass balance, aerial photography, and remote sensing). Generally, a clear change in glacier evolution can be seen after the late 1970s, accelerating in the mid-1990s and again in the early 2000s (Figure 3.14). This is different from the glaciers located at mid or high latitudes, where accelerated melting started in the 1990s. The glaciers in the tropi- cal Andes appear to have had more negative mass balances than glaciers monitored worldwide. In the Peruvian Andes, glacial areas have been well documented and multiple reports found on average a retreat of 20–35 percent between the 1960s and the 2000s; most of that retreat occurred after 1985 (Vergara et al. 2011). A similar pattern of glacial reces- sion is found in the Bolivian Andes. A rapid decline has also been reported for the Ecuadorian Andes, where glaciers on Chimborazo shrunk by 57 percent during the period 1962–1997, while glaciers on the Cotopaxi and Antisana volcanoes shrunk in area during the period 1979–2007 by 37 percent and 33 percent respectively. For the Andes of Colombia, a moderate glacier area loss of 11 percent has been documented in the period of the 1950s to the 1990s, with a fourfold acceleration in retreat during the period of the 1990s to The grey box around the average represents the uncertainty corresponding 2000s. In Venezuela, glacial retreat has been even more dramatic, to ±1 standard deviation. Source: Rabatel et al. (2013), Figure 4. with a loss of about 87 percent between 1953–2003. 39 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal Although the conditions in the Southern Andes are very dif- and Southern Patagonia, Ivins et al. (2011) inferred ice loss rates ferent (e.g., in terms of climate or sun angle) the trend in glacier of 26±6 Gt per year between 2003–2009, which explains a total retreat is obvious here as well (Figure 1.15). Lopez et al. (2010) loss of about 154±36 Gt over six years. Jacob et al. (2012) reached investigated changes in glacier length in 72 glaciers in the Chilean similar estimates using the same technique, with mass balance rates Southern Andes (Northern and Southern Patagonian Ice Field for the period 2003–2011 of 23±9 Gt per year in the Patagonian and Cordillera Darwin Ice Field) between 1945–2005, based on glaciers and of 6±12 Gt per year in the rest of South America aerial photographs and satellite images (ASTER, Landsat). They (including the tropical glaciers). However, the spatial resolution concluded that the observed general trend in glacial retreat is of about 300 km is extremely coarse and difficulties arise in dis- likely controlled by atmospheric warming. In the Northern Pata- tinguishing signals from hydrological storage and glacial isostatic gonian Ice Field, glaciers retreated in length by 4–36 percent, in adjustment (Gardner et al. 2013). the Southern Patagonian Ice Field by 0–27 percent, and further south in the Cordillera Darwin Ice Field by 3–38 percent. However, Snowpack and Snow Cover Changes glacial length fluctuations provide only limited insight into the In the tropical glacier region, due to the high solar radiance, with the imbalance of glaciers, and the large heterogeneity of glacial retreat sun close to the zenith, albedo appears to be a major determinant is very much influenced by such local conditions as exposition, in attenuating the melting process. Consequently, the frequency basin geometry, glacier dynamics, and response times. and intensity of snowfall plays a major role in determining the net A different way of measuring glacier mass loss rates is by radiation over the entire year, modulated by wet and dry seasons space gravimetry (GRACE)—by measuring the changing gravity (Rabatel et al. 2013). In the subtropical Andes of Chile and western field from satellites in regions with large continuous ice extent (a Argentina, where snowpack has been monitored for more than method available since 2003). For the large ice caps of Northern 50 years (1951–2004), there is no significant trend over this period (Masiokas et al. 2006, 2012). However, the data display a marked inter-annual variability ranging from 6–257 percent around the Figure 1.15: Ice loss from outlet glaciers on the Patagonian 1966–2004 mean, with a clear influence from the warm phases Ice Field in southern South America since the Little Ice Age. of ENSO (El Niño). Studies about snowpack in the Southern Andes are rare. It can generally be stated that changes in snowpack extent magnify changes in the seasonality of the water availability by a reduction of the flows in dry season and an increase in flows in wet seasons (Vicuña et al. 2013). 4.1.3  Projections of Glacial Change As the IPCC confirms with high confidence, glaciers worldwide are out of balance with current climatic conditions. Furthermore, it is very likely that anthropogenic forcing played a statistically significant role in the acceleration of the global glacier loss in the last decades of the 20th century (Bindoff et al. 2013). Various model projections for different future emissions sce- narios indicate that glaciers will continue to shrink in the future, even without further temperature increases. Confidence in these models is supported by their ability to reproduce past observed glacier changes using corresponding climate observations as forc- ing. Model validation is challenging, however, due to the scarcity of independent observations (currently available for only a small fraction of well-observed glaciers). In a 2°C world, Marzeion et al. (2012) expect a reduction in ice volume of tropical glaciers by 78–94 percent based on the period 1986–2005. This signal is less drastic in the Southern Andes, with an expected 21–52 percent volume reduction of the 4,700 Gt ice mass by 2100 for the same warming level. Marzeion et al. (2012) project the amount of tropical glaciers to be lost in a 3°C world Source: Glasser et al. (2011), Figure 1. at 82–97 percent, very similar to the 2°C world scenario. For the 40 Lati n Ame r i ca and the Caribbean much larger glaciers in the Southern Andes, the same study expects dammed behind the moraines of the last maximum extent of the a loss of 33–59 percent in a 3°C world. For the 21st century, with Little Ice Age in the mid-1800s. Carey et al. (2012) performed an a warming of 3°C above pre-industrial levels by 2100, Giesen and interdisciplinary case study on a glacial lake outburst flood (GLOF), Oerlemans (2013) estimate a volumetric loss of 66 percent in the which happened in Peru’s Cordillera Blanca mountain range in tropical glaciers (325 Gt) and of 27 percent in the Southern Andes 2010. Based on their analysis, they provide advice for effective (2,930 Gt). Marzeion et al. (2012) estimate an almost complete glacier hazard management. Hazard management is of high con- deglaciation (91–100 percent) of the remaining 280 Gt tropical cern, as projected warming will continue to promote glacial lake glaciers in a 4°C world (Figure 3.16.) This signal is less drastic formation (see Box 3.4: Glacial Lake Outbursts). in the Southern Andes, where a 44–72 percent deglaciation is estimated. An almost complete deglaciation for the 195 Gt tropi- 4.1.5 Synthesis This section indicates that glacial recession in South America cal glaciers (93–100 percent) in a 4°C world is also projected by has been significant. The tropical glaciers in the Central Andes in Radic´ et al. (2013). However, for the Southern Andes, the response particular have lost major portions of their volume in the course is much slower, and Radic ´ et al. (2013) expect 50 percent glacial of the 20th century. Regional studies show that the retreat has losses (3,080 Gt). All these models use a scaling methodology accelerated, with the strongest recession rates after 1985. A clear which may overestimate the recession of the small remnant tropi- trend of glacial retreat is also visible for glaciers in the southern cal glaciers. Regarding the Patagonian ice fields in the Southern Andes, which have lost about 20 percent of their volume. Regional Andes, Schaefer et al. (2013) estimate a glacial volume loss for studies highlight that the individual recession rate is very much the Northern Patagonian Ice Field of 590±50 Gt with 4°C global influenced by local conditions, which cause a large heterogeneity of warming with respect to pre-industrial levels. Their projections rates. Space gravimetry (GRACE) confirms that the declining trend of the future surface mass balance of the Northern Patagonian Ice in glacier volume has continued in the last decade. Monitoring Field predict a strong increase in ablation (refers to all processes of snow cover in the high altitudes of Chile and Argentina since that remove snow, ice, or water from a glacier or snowfield) from 1950 shows no significant trend (i.e., possible trends are hard to 2050 onward and a decrease in accumulation from 2080, both due identify in the records, since the inter-annual variability is large, to increasing temperatures. and clearly modulated by ENSO). The accelerated melting will lead to increasing runoff; when The recession of the tropical glaciers in the Central Andes the glacier reservoirs disappear, runoff will tend to decrease, will continue as rapidly as it has in recent decades. Even for low particularly in the dry season (see Section 4.2, Water Resources, or intermediate emissions scenarios inducing a global warming Water Security, and Floods). Following the trend in the tropical of 2–3°C above pre-industrial levels, two comprehensive stud- Andes (Poveda and Pineda 2010), this peak is expected within the ies consistently project a glacial volume loss of 78–97 percent next 50 years (Chevallier et al. 2011) if it has not already occurred (Marzeion et al. 2012; Radic ´ et al. 2013). Both studies predict an (Baraer et al. 2012). almost complete deglaciation (93–100 percent) for a 4°C world. 4.1.4  Glacial Hazards In contrast, Giesen and Oerlemans (2013) project a loss of only Glacier hazards are a serious risk to populations in mountain 66 percent of the glacial volume of the year 2000 in the Central regions worldwide, where a general trend of glacial retreat has Andes with a global warming of about 3°C by 2100. Thus, irre- supported the formation of glacial lakes that were precariously spective of the temperature evolution in the next decades, large parts of the glaciers of the tropical Andes will be gone before the end of the century. In the Southern Andes, the model spread for Figure 1.16: Cumulative regional surface mass balance the 2–3°C global warming ranges from 22–59 percent; a com- relative to the 1986–2005 mean from the model forced with parison for individual scenarios is difficult. In a 4°C world, three CMIP5 projections up to the year 2100. SLE = Sea-level models project a glacier volume retreat of 44–74 percent by 2100. equivalent An important research gap is the lack of reliable projections for snowpack and snow cover changes in the Andes. 4.2  Water Resources, Water Security, and Floods LAC has abundant overall water resources, but their distribution is temporally and regionally unequal (Magrin et al. 2007). ENSO- related rainfall anomalies play a major role in many areas and Source: Modified after Marzeion et al. (2012), Figure 21. determine much of the inter-annual discharge variability (Baraer 41 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal In many parts of the region there is no clear trend in future Box 1.3: Water Security in the Mexico discharges due to the uncertainties in rainfall projections in dif- City Metropolitan Area ferent GCMs (see also Section 3.3, Regional Precipitation Projec- tions) and the diverging results of different impact models (Bravo The Mexico City Metropolitan Area (MCMA) faces frequent climate- et al. 2013; Davie et al. 2013; Döll and Schmied 2012; Hidalgo et related hazards. These include extremes of water and heat, with al. 2013; Imbach et al. 2012; Krol and Bronstert 2007; Malhi et al. floods on the one side and heat waves and droughts on the other. 2009; Rowell 2011; Schewe et al. 2013). The most common extreme events from 1980–2006 were floods resulting from heavy precipitation events (Romero Lankao 2010). 4.2.1  Central America and Mexico Up to 42 percent of the population in MCMA was estimated to be Milly et al. (2005) modeled a decrease in river runoff for Central vulnerable to climate change and natural hazards. Forty percent of America of up to 10 percent for the 20th century. Hidalgo et al. (2013) those live in so-called “high-risk areas” which are characterized by projected mean annual runoff to decrease by 10–30 percent by the very steep slopes of over 15 degrees where landslides can occur end of the 21st century with a median of 3°C regional warming. The after heavy precipitation events (Baker 2012). Exponential popula- same tendencies were shown in Imbach et al. (2012), who found tion growth in MCMA during the 20th century also contributed to the high vulnerability to flooding and water shortages (Brun 2007). decreases in annual runoff in 61–71 percent of the area (notably in Currently water is provided from the Mexico City basin aquifer, central Yucatan Peninsula, the mountains of Nicaragua, Honduras, with one-third transferred from external water basins (Romero and Guatemala) in a 2°C warmer world, depending on the sub- Lankao 2010). Overextraction of groundwater in combination with region and the emissions scenario. Increases were projected only locally unfavorable soil conditions (heavily saturated clay) have for just one percent of the area, mainly along the southern edge. caused parts of the city to subside and thus suffer more frequent Evapotranspiration was projected to increase more than 20 percent flooding (Baker 2012; Romero Lankao 2010). Another problem for in more humid areas (e.g., Costa Rica, Panama) whereas northern long-term water security is related to groundwater contamination of areas were projected to experience no change. Fabrega et al. (2013) the Mexico City basin aquifer, most probably due to surface waste- found precipitation increases by 5 percent or more for most regions water (Brun 2007) of Panama—but with no statistically significant changes in total The projection of temperature increases, more frequent and runoff in a 3°C world. Maurer et al. (2009) modeled the inflow of prolonged dry spells, and a (probable) precipitation decrease will two reservoirs in the Rio Lempa basin. They found decreases of harm water security and increase water dependencies in the grow- 13 percent in total annual reservoir inflow in a 2°C world and of ing MCMA. The water security of the water-providing external areas will also be affected (Magrin et al. 2007; Romero Lankao 2010; 24 percent in a 4°C world, implying potential reductions in hydro- Sosa-Rodriguez 2013). power capacities. Low flow years might occur more frequently, especially under a higher warming (Maurer et al. 2009). Global studies mostly confirm this picture. Milly et al. (2005) projected a decrease in river runoff in Central America of 5–20 per- et al. 2012; Cortés et al. 2011; Krol and Bronstert 2007; Mata et al. cent for the middle of the 21st century with 3°C global warming. 2001; Poveda 2004; Ronchail et al. 2005; Shi et al. 2013; Vicuña Nakaegawa et al. (2013) also found that the total annual runoff et al. 2010; Vuille et al. 2008). Large parts of the region are char- decreases for Mexico and Central America from 2075–2099 with 3°C acterized by inter-annual/seasonal rainfall variability through the warming. For the Rio Grande, annual discharge decreases by more oscillation of the Intertropical Convergence Zone (Garreaud et al. than 20 percent. In a 3°C world, Mesoamerica also experiences a 2009). Due to the unreliable rainfall, groundwater resources and strong decrease in discharge in the study of Schewe et al. (2013). water from glacier and snowmelt play a crucial role in supply- Portmann et al. (2013) reported a mean decrease across several ing local water (Chevallier et al. 2011; Hirata and Conicelli 2012; GCMs of more than 10 percent in groundwater recharge in a 4°C Vuille et al. 2008). world for Central America. The projected changes were much LAC suffers from widespread floods and landslides (Maynard- less pronounced when assuming lower global mean temperature Ford et al. 2008) which result from different origins (Dilley et al. increases of 2–3°C. 2005). Heavy precipitation events in the context of ENSO or tropical cyclones can lead to disastrous floods, especially in regions with 4.2.2 Caribbean steep terrains such as in the Andes and Central America (IPCC 2012; The assessment of water resources in the Caribbean relies more on Mata et al. 2001; Mimura et al. 2007; Poveda et al. 2001). Coastal assumptions and extrapolations from climatological data than on areas in the Caribbean and Central America suffer from flooding long-term hydrometric measurements, especially for the smaller as a result of storm surges and tropical cyclones (Dilley et al. 2005; islands (Cashman 2013; FAO 2003). Water provisioning is espe- Woodruff et al. 2013). In the Andes, glacial lake outbursts pres- cially difficult on islands which rely mainly on a single source of ent a permanent hazard for Andean cities (Chevallier et al. 2011). water (such as groundwater in Barbados, Bahamas, Antigua and 42 Lati n Ame r i ca and the Caribbean Barbuda, and Jamaica, or surface water in Trinidad and Tobago, Grenada, St. Vincent and the Grenadines, St. Lucia, Dominica, Box 1.4: Glacial Lake Outbursts and elsewhere)(Cashman 2013; Gencer 2013). The lack of long-term measured stream flows in the Caribbean Glacial lake outburst floods (GLOFs) originate from various causes. First, increasing glacier melting raises the water levels of lakes, renders the evaluation of hydrological models in the region dif- eventually resulting in an overflow of water or the breaking of dams. ficult, and future projections of runoff have only low confidence Second, ice instability may cause an avalanche of seracs into a lake, (Hidalgo et al. 2013). A combination of lower precipitation, high leading to suddenly higher water levels and the breaking of dams. abstraction rates, and sea-level rise may lead to intrusion of saline Ice instability might increase with increasing temperatures. Third, sea water into coastal groundwater aquifers (Cashman et al. 2010; glacier retreat may trigger major rock slides. It is important to note Cashman 2013). Another hydrological hazard regarding climate that the Andes belong to a region of high seismic activity, which can change is more severe flooding events related to tropical cyclones contribute to GLOFs. (Chevallier et al. 2011; Kaser et al. 2003). (Cashman et al. 2010). 4.2.3  Northern South America (Colombia) this is around 30 percent higher than what was measured during Restrepo et al. (2014) found significant discharge increases for the the 1930–2009 period (Baraer et al. 2012). Mulatos, Magdalena (at Calamar), Canal del Dique, and Fundación Kinouchi et al. (2013) simulated the glacier melt and runoff rivers, especially from 2000 to 2010. Regional studies of climate- in a headwater catchment of the Cordillera Real in Bolivia. They related hydrologic impacts are limited, as access to and quantity of applied different 1–1.5°C temperature increase scenarios by 2050 observational climate data is limited (Hoyos et al. 2013). Reanalysis and found only small changes in annual runoff. The seasonal or simulated/reconstructed datasets have been used, but Hoyos variation was, however, modified significantly. Under their projec- et al. (2013) reported substantial differences between climatologi- tions, streamflow during the dry and early wet season was reduced cal datasets and observed values. Nakaegawa and Vergara (2010) (e.g., because of snowmelt decrease, and during the wet season found a trend of decreasing mean annual river discharge due to it increased, especially in January and February). Baraer et al.(2012) increased evapotranspiration in the Magdalena Basin in a 3°C projected that once the glaciers have melted, the average dry season world. Monthly mean river discharge decreased significantly in discharge may decrease more than 60 percent in Parón and Llanga- April, October, and November at Puerto Berrio. It is important, nuco and up to 70 percent at La Balsa—with serious implications however, to note that mean precipitation was overestimated by for water supplies during the dry season. Juen et al. (2007) found about 35 percent for the GCMs used. little changes in total annual discharge by 2050 and 2080 in the Llanganuco catchment, but they did find a bigger amplitude of 4.2.4 Andes discharge seasonality (with a risk of very low flows during the In mountainous regions, winter precipitation accumulated as snow dry season). They concluded that a smaller glacier size causes and ice, similar to groundwater reserves, helps to buffer water short- decreasing glacier melting, but that this decrease is supplemented ages resulting from little or seasonal rainfall (Masiokas et al. 2006; by an increase in direct runoff from non-glaciered areas. Wet sea- Viviroli et al. 2011; Vuille et al. 2008). Downstream regions with son discharge was projected to increase from 10–26 percent and low summer precipitation in particular benefit from this temporal dry season discharge to decrease from 11–23 percent for warming water storage (Masiokas et al. 2013; Viviroli et al. 2011). Glacier >1.5°C in 2050 and >2°C in 2080 depending on the emissions retreat thus endangers water security in these areas (Vuille et al. scenario and timeframe (see Figure 3.17). For the northern half 2008). Current accelerated melting rates, however, imply a short- of the Andes, a very likely increase in flood frequency in a 4°C term local surge in water—and higher river flow peaks can cause world was projected (Hirabayashi et al. 2013). landslides and floods. Massive flood events have been associated with glacial lake outburst (Chevallier et al. 2011) (see Box 3.4). 4.2.6  Central Andes The Andes in Central Chile and central western Argentina are 4.2.5  Tropical Andes characterized by a direct relationship between the amount of Baraer et al. (2012) analyzed historical streamflow records for the snow accumulated in winter and river discharge released during Cordillera Blanca over the period 1990–2009; they showed that spring-summer (Masiokas et al. 2006), and around 85 percent of discharge was decreasing annually and during the dry season. the observed river flow variance over 60 years in the area can The trends were attributed to glacier retreat. Meltwater contrib- be explained by snowpack records (Masiokas et al. 2010, 2013). utes 10–20 percent to the total annual discharge of the Río Santa The inter-annual variability of snowpack extent is enormous, (Cordillera Blanca), but may rise to over 40 percent during the varying from zero percent to over 400 percent of the long-term dry season (Baraer et al. 2012). In the period from 1990–2009, the mean (Masiokas et al. 2006, 2010, 2013). Freshwater availability overall glaciered area was decreasing by 0.81 percent annually; is thus strongly dependent on mountain snowpack. In very dry 43 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal Figure 1.17: Changes in seasonal total runoff in 4 IPCC Box 1.5: Water Security in Quito, La climate-change scenarios with respect to the 1961–1990 Paz, Bogotá, and Lima mean monthly runoff. Precipitation patterns are different throughout the tropical Andes. The Pacific slopes in Colombia receive on average more than 8000 mm per year of rainfall, whereas large parts of the highlands of Bolivia and the Peruvian coast get less than 100 mm per year. Water security is a pressing issue because the capital cities experienced population growth rates of 11.9 percent in Quito and 20.6 percent in La Paz for the period 2000–2010. Bogotá and Quito are situated on steep mountain terrain at 2650 m and 2850 m above sea level respectively. High popula- tion densities provoke local water stresses, and both cities require inter-basin water transfers from the wet Amazonian slopes to meet their water demand. Sixty-two percent of Quito’s water is currently provided by the Amazonian basin. Lima, meanwhile, is the second- Source: Juen et al. (2007). largest desert city in the world and its water is supplied almost solely by the western slope of the Andes. Competition exists with other water users, including the agricultural sector (Buytaert and De Cortés et al. (2011) found that, for the period 1961–2006, river Bièvre 2012). regimes in the dry north were driven by snowmelt whereas those further south were more rainfall-dominated. The southern river systems (to the south of 35°) have displayed consistently earlier timing in peak annual flow rates. Vicuña et al. (2013) found indica- Box 1.6: Water from the Cordillera tions of a shift in the last 30 years to an earlier annual snowmelt Blanca season of around 15 days for the Mataguito basin. More high-flow discharges were also observed during the last The arid coastal area of Peru is home to approximately half of the 10 years. They occurred mostly during autumn months when high country’s population. Water is provided mainly by rivers coming rainfall and high minimum temperatures decreased the fraction down from the western slopes of the Cordillera of the Andes. Con- of precipitation falling as snow and cause a faster rainfall-runoff tributors to runoff during the rainy season are rainfall, groundwater, and glacier melt. In contrast, during the dry season rivers are fed by response (Vicuña et al. 2013). Annual low-flow levels during groundwater and glacier melting at higher than 5000 m (Chevallier spring and summer decreased significantly in the Mataguito basin et al. 2011). (Vicuña et al. 2013) and for some stations in the Limay River The water from the Cordillera Blanca supports human activities Basin (Seoane and López 2007). This trend in river flow variability at different altitudes. Irrigated agriculture is practiced between might endanger electrical power generation in the region, as the 2000–4000 m and at the foot of the Andes. Below 2000 m, Limay River basin contains many hydropower stations which yield electricity is generated (Chevallier et al. 2011; Kaser et al. 2003). around 26 percent of Argentina’s total electrical power generation To maintain the full capacity of electricity production, a discharge (Seoane and López 2007). of 60 m3 per second is required. As the minimum flow of the Río Projections of the mean number of snowy days decrease by Santa usually falls below that level, water management is needed to about 9 percent in a 2°C world, and 26 percent in a 4°C world, in guarantee the minimum discharge. the Mataguito basin (Demaria et al. 2013). In addition, the center The population of the Andes area is increasing. Due to this timing of mass of annual flow was projected to occur earlier, by population trend, and to the possibility of expanding cultivation into higher areas of the Andes under increasing temperatures, water 12 days in a >1.5°C world and by 16 days in a >3°C world. In demand is expected to increase. This might lead to conflicts with the Limarí basin, reductions in annual streamflow are probably hydroelectric power generation (Juen et al. 2007; Mark et al. 2005). intensified by increased evapotranspiration because a 19 percent rainfall decrease resulted in a 21 percent streamflow decrease in a 4°C world (Vicuña et al. 2010). Vicuña et al. (2010) also found years, with no or little solid precipitation in the upper water- increasing winter flows (28.8–108.4 percent), decreasing summer shed, glacier melting gains in importance, although there is little flows (–16.5 to –57.8 percent), and earlier center timing of mass information about the contribution of ice masses to river flow of annual flows for different sub-basins of the Limarí basin in a (Masiokas et al. 2013). >3°C world. For northeastern Chile, Arnell and Gosling (2013) 44 Lati n Ame r i ca and the Caribbean 4.2.7  Amazon Basin Box 1.7: Water Security in the Espinoza Villar et al. (2009) found significant decreasing mean and Central Andes minimum annual runoffs from 1990–2005 for southern Andean rivers (Peru, Bolivia) and increasing mean and maximum annual In the Andes, between latitudes 30 and 37° lie two major cities, runoffs for northern Andean Rivers (Ecuador) draining into the Santiago de Chile (Chile) and Mendoza (Argentina). Human activi- Amazon. For the southern Amazon, Li et al. (2008) found more ties in these cities are almost completely dependent on meltwater, dry events during 1970–1999. Espinoza Villar et al. (2009) reported especially in the drier Argentinean foothills which receive only around decreasing mean annual discharge and monthly minimum dis- 200 mm of annual precipitation (Masiokas et al. 2013). charge from 1974–2004 for Tapajós in the southeastern Amazon, Snowmelt is very important for water supply, hydroelectric the Peruvian Amazon Rivers, and the upstream Madeira. generation, and viniculture in large parts of Chile (Demaria et al. Guimberteau et al. (2013) analyzed the impacts of climate change on 2013). The central valley of Chile contains the majority of the coun- try’s reservoir storage and supplies water to several large towns. extreme streamflow over several Amazonian sub-basins by the middle Water demand results as well from agriculture, as 75 percent of of this century for a 2°C global warming scenario. They found that low the irrigated area in Chile is located here (Demaria et al. 2013). The flows would become more pronounced. The trend is significant at the Central Valley suffers from high inter-annual rainfall variability, with Madeira and Xingu rivers, with JJA precipitation decreases of 9 percent conditions being wetter during El Niño years and drier during El and 22 percent respectively. At Porto Velho, the decrease in median Niña years (Cortés et al. 2011). In years with above-average rainfall, low flows is about 30 percent; at Altamira, it is about 50 percent. In farmers irrigate annual crops (e.g., orchards, vineyards) with surface addition, Tosiyuki Nakaegawa et al. (2013) found total annual runoff water, whereas in below-average rainfall years they are forced to use decreases in the southern half of the Amazon River in a 3°C world. groundwater. The use of groundwater has recently reached unsus- Average annual runoff varied from –72 percent to +6 percent in a 3°C tainable levels, and the Chilean water authorities have therefore world for the Bolivian part of the Amazon (Alto Beni), assuming no restricted water extraction. Decreasing annual rainfall as a result of land use change (Fry et al. 2012). Nevertheless the projected ground- climate change would put even more pressure on agriculture and water recharge was consistently negative (–96 percent to –27 percent) water resources (Arumí et al. 2013). because potential evapotranspiration increases. Malhi et al.(2009) found an increase in dry-season intensity in eastern Amazonia in a 3° world and seasonal increased water stress because of climate change and deforestation. Langerwisch Box 1.8: Water Security and Glacial et al. (2013) found shifts in flood patterns in a 3° world. The Melt in La Paz and El Alto, Bolivia duration of flooding at the end of the 21st century was projected to be 0.5–1 months shorter than for 1961–1990. The probability of La Paz and El Alto receive 80 percent of their water from the Tuni three successive extreme wet years decreased by up to 30 percent Condoriri range. The contribution of glacier ice melt could be from 30–40 percent (World Bank 2008) up to 60 percent (Painter 2007). (Langerwisch et al. 2013). Since water demand has risen in recent years, the water manage- Median high flows in the western part of the Amazon basin ment of both cities is very much challenged (Jeschke et al. 2012; increase by 5–25 percent by the middle of this century for a warming Shi et al. 2013). Almost the entire energy supply of La Paz is sup- of 2°C; this trend is not, however, significant (Guimberteau et al. plied by hydroelectric power which comes mainly from two glacier 2013). In a 2°C world, the increase in high flow was projected to ranges—in the Zongo valley and Charquiri (Painter 2007). The be lower than in a warmer than 3°C world; low flows increase glaciers of the Cordillera Real encompass 55 percent of the Bolivian 10–30 percent under a 4° warming scenario (Guimberteau et al. glaciers. Between 1963–2006 they lost more than 40 percent of 2013). The flood zone is consistently projected to increase with a their volume (Soruco et al. 2009) and they are further declining (Liu 2–3 month longer inundation time in a 3°C world over several GCMs et al. 2013). It has been postulated that water demand might soon (Langerwisch et al. 2013). The average runoff and the maximum surpass water supply in El Alto (Shi et al. 2013). Future water and runoff increased in two subcatchments of the Paute basin for a 2°C energy supply will be increasingly critical due to rising demand, in global warming scenario from 2045–65 (Mora and Campozano et al. combination with decreasing tropical glacier volumes (Rabatel et al. 2013; Vuille et al. 2008). 2013). Exbrayat et al. (2014) also found increases in annual runoff for a catchment in the Ecuadorian Andes by 2100; they also showed a high variability of runoff projections depending on the choice of simulated a decreasing mean annual runoff for warming higher GCM, emissions scenario, and hydrological model. Similarly, Buytaert than 1°C over 21 GCMs. Projections by Döll (2009) also showed et al. (2009) showed that, due to the wide range of GCM projections, a reduced groundwater recharge for the central Andes region by the projected average monthly discharges diverge considerably in the 2050s with 2°C warming. the Paute River system in Ecuador under a 1°C increase by 2030. 45 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal For the northernmost Amazon and the river mouth region, from January–May at Corrientes on the Paraná River—but not at river flow and runoff coefficients decrease with a global warm- Posadas which lies further upstream. Nóbrega et al. (2011) found ing of 2°C in 2045–2065 (Guimberteau et al. 2013). In the same that for the Río Grande, a tributary of the Paraná, for every 1°C study, median low flows decrease by 20 percent for the Japura temperature rise annual flow increased by 8–9 percent in relation and Negro river and 55 percent at the Río Branco. to 1961–1990. Assuming 2°C warming, mean river flow ranged At the main stem of the Amazon River the runoff coefficient from –20 percent to +18 percent. is projected to slightly decrease at Óbidos (the last station of the Camilloni et al. (2013) projected an increase in frequency and Amazon before the mouth) with a warming of 2°C by 2050 duration of river flooding in a >3°C world in the Uruguay and (Guimberteau et al. 2013). Median low flow is projected to decrease Paraná basins. Hirabayashi et al. (2013) showed a decrease in the by 10 percent, but this trend is uncertain. In a 4°C world, Guim- 20th century 100-year return period for floods for the Parana in a berteau et al. (2013) projected that low flows and high flows would 4°C world, but there was little consistency across the 11 GCMs used. each increase by five percent at Óbidos. Döll and Schmied (2012), Besides river floods, storm surge floods present a major hazard however, projected the mean river discharge of the downstream for Buenos Aires. Barros et al. (2005; 2008) found a greater inland part of the Amazon to increase for under a 2°C warming by 2050 reach of recurrent storm surge floods by 2070 under a 3°C global in one GCM but decrease in another GCM. warming scenario. Assuming no changes in population distribu- tion, permanent coastal flooding due to sea-level rise will play a 4.2.8  Northeast Brazil minor role and will affect rather sparsely populated areas at the Krol and Bronstert (2007) found that a decrease in precipitation coast of Buenos Aires and its surroundings. In contrast, Pousa et al. by the end of the 21st century would significantly decrease the (2013) projected that sea-level rise could aggravate the impact of runoff of the Jaguaribe River and the stored volume in the Ceará storm surge floods in Buenos Aires. reservoir. In contrast, an increase of precipitation by 50 percent did not significantly increase river runoff because of an accompany- 4.2.10  Southernmost South America ing increase in water demand. Döll and Schmied (2012) projected Milly et al. (2005) simulated a decrease in mean relative runoff of the seasonality of river discharge in northeastern Brazil to remain up to 10 percent, which is in agreement with observed 20th century stable but also that mean river discharge would decrease by the trends for southernmost South America. They projected a decrease middle of the 21st century under a 2°C global warming scenario. in mean relative runoff of 10–30 percent for southernmost South Due to uncertainty in the GCM projections, there is no clear America for the middle of the 21st century with 3°C global warm- signal about the relative change of annual discharge for northeastern ing. Schewe et al. (2013) found similar results for a >2°C world. South America under 2°C warming (Schewe et al. 2013). Portmann et al. (2013) projected both strong decreases and increases in mean 4.2.11 Synthesis groundwater discharge for northeastern Brazil in a 4°C world ENSO-related rainfall anomalies play a major role in many areas depending on the GCM. Assuming different warming scenarios in LAC and determine much of the inter-annual discharge variabil- with varying levels of decreasing rainfall, Montenegro and Ragab ity. In Central America, there is a high agreement on decreasing (2010) projected strong decreases in groundwater recharge of up mean annual runoff and discharge, although the magnitude of to 77 percent and streamflows of up to 72 percent for a subcatch- the change varies (Arnell and Gosling 2013; Hidalgo et al. 2013; ment of the Sao Francisco River Basin. Imbach et al. 2012; Maurer et al. 2009; Milly et al. 2005; Nakae- gawa et al. 2013; Schewe et al. 2013). The trend seems to be more 4.2.9  Río de la Plata pronounced for the northern than for the southern part of Central The Río de la Plata region experienced a 10–30 percent increase America (Hidalgo et al. 2013; Imbach et al. 2012). Therefore, water in river runoff during the 20th century (García and Vargas 1998; stress may increase, especially in arid areas with high population Jaime and Menéndez 2002; Menéndez and Berbery 2005; Milly densities and during the dry season. et al. 2005). There is no consensus, however, among river runoff The Caribbean lacks long-term measured stream flow data. projections for the Río de la Plata and its tributaries because the Runoff projections are therefore of low confidence (Cashman 2013; projected direction of rainfall trend varies among GCMs. Milly FAO 2003; Hidalgo et al. 2013). However, freshwater availability et al. (2005) modeled an increase in mean relative runoff for the may decrease for several reasons. Sea-level rise (Mimura et al. Rio de la Plata region of 20–50 percent for the middle of the 21st 2007) may lead to an intrusion of sea water into coastal aquifers century and using a 3°C warming scenario. River flow projec- (Cashman et al. 2010; Cashman 2013) and summer precipitation tions in the upper Paraguay River basin varied from ±10 percent is projected to decrease (Mimura et al. 2007). Regionally the by 2030 for a 1.5° warming scenario and by ±20 percent by risk of flooding and mudslides with high mortality rates is high 2070 for a 2°C warming scenario (Bravo et al. 2013). For a 3°C (Cashman 2013; Edwards 2011; Williams 2010). Although floods world, Nakaegawa et al. (2013) projected increasing discharge often seem to be associated with land-use change, more severe 46 Lati n Ame r i ca and the Caribbean flooding events may also occur in the context of climate change phase (Hatfield et al. 2011). Generally, warmer temperatures act (Cashman et al. 2010; IPCC 2012). to decrease the development phase of perennial crops, resulting In Northern South America (Colombia), there are only a in earlier crop flowering and reduced seed sets (Craufurd and limited number of regional hydrological impact studies available Wheeler 2009). When temperatures increase above the maximum, for northern South America, and rainfall projections are uncertain. plant growth and yields can be drastically reduced (Ackerman and Conclusions about projected hydrological impacts are therefore Stanton 2013; Berg et al. 2013; Luo 2011). of low confidence. The optimum seasonal average temperature for maximum In the Andes, higher discharge seasonality is projected for the grain yield is 15°C for wheat, 18°C for maize, 22°C for soybeans, Tropical Andes. Streamflows during the dry season may decrease and 23°C for rice (Hatfield et al. 2011; Lobell and Gourdji 2012). because of ongoing glacier retreat (Baraer et al. 2012; Juen et al. Hatfield et al. (2011) also identified average temperatures leading 2007; Kinouchi et al. 2013). Lower dry season discharge has already to a total crop failure: 34°C for wheat, 35°C for maize, 39°C for been observed during the past two decades (Baraer et al. 2012). soybeans, and 35°C for rice. Short intervals of a few days above the However, streamflow during the wet season may increase (Juen optimum average temperature can lead to strong yield decreases et al. 2007; Kinouchi et al. 2013). The region has a high flood risk (Ackerman and Stanton 2013). Teixeira et al. (2013) project an (e.g., due to accelerated glacier melting; see Box 3.4: Glacial Lake increasing occurrence of heat stress for maize, rice, and soybeans Outbursts) (Carey 2005; Hirabayashi et al. 2013). For the Central in Latin America. Lobell and Gourdji (2012) estimate global yield Andes, more streamflow was observed and projected to occur at declines of 3–8 percent per °C of temperature increase based on earlier dates locally (Cortés et al. 2011; Vicuña et al. 2013; Demaria a literature review. It is, however, important to note that there et al. 2013). Lower dry season discharges may cause significant are numerous knowledge gaps concerning plant reactions to tem- water supply problems in urban areas. peratures above their optimum averages (Craufurd and Wheeler Amazon Basin: Runoff and discharge projections for most parts 2009; Porter et al. 2014). Moreover, plants are somewhat capable of the Amazon basin are diverging, especially for the southern of adapting to changing climatic conditions—and it is unclear if and eastern areas. The main reasons for this are the high vari- climate change will alter growing conditions too fast for crops to ability of rainfall projections using different GCMs and uncertain- adapt on their own (Ackerman and Stanton 2013). ties introduced by hydrological impact models. However, for the western part of the basin a likely increase in streamflow, runoff, 4.3.2  Plant Diseases flood zone, and inundation time was projected (Guimberteau et al. How pests and diseases will spread under future climate condi- 2013; Langerwisch et al. 2013; Mora and Campozano et al. 2013). tions, and how severe the effects will be on yields and production Northeast Brazil: The direction of discharge and groundwater quantities, is unclear. Already today crop diseases are respon- recharge trends vary due to diverging rainfall projections under sible for losses of 10 percent or more of global food production different GCMs (Döll and Schmied 2012; Krol and Bronstert 2007; (Chakraborty and Newton 2011; Ghini et al. 2011; Luck et al. 2011). Portmann et al. 2013; Schewe et al. 2013). Climate change is expected to alter the geographic distribution Río de la Plata: There are no consistent river runoff projections of insects and diseases in much of the world (Porter et al. 2014). for the basin because the directions of rainfall projections vary The knowledge on climate change and plant diseases, however, is among the GCMs (Bravo et al. 2013; Milly et al. 2005; Nakaegawa still very limited (Ghini et al. 2011; Luck et al. 2011). The impacts et al. 2013; Nóbrega et al. 2011). differ greatly between crops and pathogens as do the interactions Southernmost South America: A decrease of mean runoff among hosts, pathogens, microorganisms, and the climate (Ghini was projected with a high confidence (Milly et al. 2005; Schewe et al. 2011; Bebber et al. 2014). The existing knowledge base is et al. 2013). inadequate to make generalizations about the behavior of crop diseases under a changing climate (Luck et al. 2011). Factors that 4.3  Climate Change Impacts on Agriculture are most likely to influence the development of plant diseases are increasing atmospheric CO2, increasing winter temperatures, and 4.3.1  Temperature Sensitivity Crop Thresholds increasing humidity (Luck et al. 2011). Agriculture is one of the most climate dependent human activities, One recent example of the impact of plant diseases on agricul- and the development and growth of plants is affected to a very ture in the LAC region is the outbreak of coffee leaf rust (Hemileia large extent by temperature. Every plant has a range between a vastratix), considered the most destructive coffee disease, in Central maximum and minimum temperature in which the plant can exist America during the 2012–13 growing season. Around 50 percent of and an optimum temperature at which growth is at its optimal rate the roughly one million ha under coffee production in the region (Hatfield et al. 2011). Crops often require different temperatures were affected by the disease, reducing the production quantity in their numerous development stages and are very sensitive to by an estimated 17 percent in comparison to the previous year temperatures above the optimum, especially during the pollination (Ghini et al. 2011; ICO 2013). The outbreak devastated small holder 47 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal Box 1.9: Surface Ozone Concentrations Surface ozone concentrations have negative impacts on agricultural yields. The impact on crop yields strongly depends on the seasonal and regional distribution of surface ozone, as it is not distributed evenly in the atmosphere (Teixeira et al. 2011). Declines in yield levels currently range from 7–125 percent for wheat, from 6–16 percent for soybeans, from 3–4 percent for rice, and from 3–5 percent for maize; wheat and soybeans are especially sensitive to surface ozone (Van Dingenen et al. 2009; Teixeira et al. 2011). Jaggard et al. (2010) noted that the impact of ozone on crop yields has been neglected in many climate impact projections and found that the benefits of the CO2 fertilization effect (see Box 2.4) could be offset by the negative effect of increased ozone concentrations on C3 plants (and even lead to a yield reduction of five percent in C4 plants). By 2030, increasing surface ozone could lead to yield declines in Latin America by up to 7.8 percent for wheat, 2.9 percent for maize, and 7.5 percent for soybeans depending on the emissions levels of ozone precursors (Avnery et al. 2011). coffee growers and possibly contributed to rising coffee prices southern Brazil, bean and maize productivity would decline by globally (NYT 2014). Coffee leaf rust, together with soybean rust 15–30 percent in comparison to 1971–2000 levels under a global (Phakopsora pachyrhizi), are expected to move further south and mean warming of 2°C by 2050 and by 30–45 percent with 4°C affect South American countries with global temperatures increas- warming by 2080 without CO2 fertilization but with technological ing by approximately 3.5°C by 2080 compared to pre-industrial progress (Costa et al. 2009). Including CO2 fertilization for beans levels (Alves et al. 2011). leads to productivity increases of up to 15 percent (Costa et al. 2009). Because maize is a C4 crop, including CO2 fertilization has 4.3.3  Projected Changes in Crop Yields only a limited impact and productivity keeps decreasing (Costa et al. Climate change impacts on crop yields vary depending on crop 2009). Rain-fed sugarcane yields could increase by 15–59 percent type and location. Fernandes et al. (2012) projected changes in with global warming of 1.5–2.3°C by 2050, including CO2 fertil- crop yields in 2050 (compared to 1989–2010) under global warming ization and technological improvement (Marin et al. 2012). In scenarios of between 1.7°C and 2.3°C. Table 3.6 and Figure 3.18 the Brazilian Amazon, soybean yields decline by 44 percent with present some of their key results. It is important to note that, 4°C mean global warming by 2050 and by 1.8 percent with a 2°C when considering adaptation measures, yield declines are less temperature increase (Lapola et al. 2011). On average and over all pronounced but still negative for wheat, soybeans, and maize. analyzed crop types, yields are projected to decline by 31 percent Yield projections for rice show a different picture. With the when temperatures increase by 4°C without CO2 fertilization and exception of Brazil, Mexico, and the Caribbean, where temperatures increase by 14 percent when temperatures increase by 2°C with are already high, rice yields could increase by up to 12 percent CO2 fertilization (Lapola et al. 2011). by 2020 and by 17 percent by 2050 as average conditions for rice In Ecuador, ECLAC (2010) projects yield declines of 53 percent photosynthesis would improve with increasing temperatures for maize, 9 percent for beans, 41 percent for bananas, 36 percent (Fernandes et al. 2012). for sugarcane, 23 percent for coffee, and 21 percent for cocoa; Nelson, Rosegrant and Koo et al. (2010) project yield changes ECLAC also projects yield increases of up to 37 percent for rice for for different crops in LAC with a 1.8–2.5°C global temperature the year 2080 with 3.5°C warming. Colombian agriculture, mean- increase by 2050. Their key results, shown in Table 3.6, show that while, is projected to be severely impacted by climate change. Up yields generally decline without CO2 fertilization; this is most pro- to 80 percent of agricultural crops currently cultivated in Colombia nounced for irrigated maize, soybeans, and wheat. CO2 fertilization in 60 percent of the cultivation areas of the country would be nega- increases yields for rice, soybean, and maize by over 10 percent tively affected by 2–2.5°C global temperature increases by 2050 if besides irrigated maize (Nelson, Rosegrant, Koo et al. 2010). no adaptation measures are introduced (Ramirez-Villegas et al. In Chile, even when including the CO2 fertilization effect, 2012). Perennial crops (notably such high-value crops as tropical yields could be reduced by 2050 by 5–10 percent for maize and fruit, cocoa, bananas, and coffee) could be particularly affected by 10–20 percent for wheat in comparison to 1971–2000 levels with climate change (Ramirez-Villegas et al. 2012). Coffee farming might 2.7°C global warming if no adaptation measures are implemented have to migrate to higher altitudes or other cultivation regions to (Meza and Silva 2009). In Argentina, yields for wheat, maize, and maintain present yields, a problem also relevant in other parts of soybeans are projected to decline by 16, 24, and 25 percent respec- Latin America (Camargo 2010; Laderach et al. 2011; Zullo et al. 2011). tively by 2080 under a 3.5°C global warming scenario without In Panama, yield changes for maize range from –0.8 to CO2 fertilization (ECLAC 2010). Yield declines are less pronounced +2.4 percent for global warming of 1.7–1.9°C in 2055, and from with only 2.7°C global warming, with declines of 11 percent for +1.5 to +4.5 percent for global warming of 2.2–3.3°C in 2085 wheat, 15 percent for maize, and 14 percent for soybeans; includ- including CO2 fertilization (Ruane et al. 2013). Accelerated crop ing CO2 fertilization increases yields slightly (ECLAC 2010). In development helps to complete the grain-filling phase before the 48 Lati n Ame r i ca and the Caribbean Table 1.6: Projected Changes in Yields and Productivity Induced by Climate Change. Yield or Source Scenario Time Horizon Region Crop Productivity Effect Fernandes et al. (2012) A1B / B1 2050 Brazil Soybeans –30 to –70 % Brazil, Ecuador, Maize up to –60 % Brazil Wheat –13 to –50 % LAC Rice up to +17 % Meza and Silva (2009) A1F1 2050 Chile Maize –5 to –10 % Wheat –10 to –20 % Costa et al. (2009) A2 2050 Brazil Beans –15 to –30 % Maize –15 to –30 % Ruane et al. (2013) A2 / B1 2050 Panama Maize –0.8 to +2.4 % A2 /B1 2080 Panama Maize +1.5 to +4.5 % Lapola et al. (2011) A2 2050 Brazilian Amazon Soybeans –1.8 to –44 % Marin et al. (2013) A2 / B2 2050 Southern Brazil Sugarcane +15 to +59 % ECLAC (2010) A2 2080 Ecuador Maize –53 % Beans –9 % Bananas –41 % Sugarcane –36 % Coffee –23 % Cocoa –21 % Rice +37 % A2 / B2 Argentina Wheat –11 to –16 % Maize –15 to –24 % Soybeans –14 to –25 % Nelson et al. (2010) A2 2050 LAC Maize –3.0 to +2.2 % Rice –6.4 to +12.7 % Soybeans –2.5 to +19.5 % Wheat –5.6 to +12.2 % beginning of dry periods with high levels of water stress (Ruane A significant positive relationship between crop yield change and et al. 2013). In Mexico, wheat yields decline with global tempera- temperature is revealed, however, when CO2 fertilization is con- tures rising between 1.6–2.1°C by 2050 across several crop models sidered (see Table 3.7 and Figure 3.19), although the beneficial and GCMs (Rosenzweig et al. 2013b). Yield declines are more effects of CO2 fertilization are highly uncertain (Ainsworth et al. pronounced with stronger warming, but they remain relatively 2008)(see Box 2.4) The interpretation of these results therefore small because CO2 fertilization reduces the negative yield effect requires some caution, as model assumptions made regarding in the crop models (Rosenzweig et al. 2013b). CO2 fertilization may not hold in an actual crop production envi- A meta-analysis of the impacts of climate change on crop yields ronment. If the effects of CO2 fertilization are not considered, for the LAC region (see Section 6.3, Meta-analysis of Crop Yield the relationship remains significant but becomes negative, with Changes with Climate Change,) reveals no significant influence of increasing temperature leading to considerable yield declines (see temperature increase over crop yields across all available studies. Figure 3.19). 49 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal Figure 1.18: Aggregate impacts on crop yields in the LAC region with adaptation, computed by the AZS-BioMA platform under 2020 and 2050 NCAR GCM for A1B scenario. Source: Fernandes et al. (2012), Figure 4.1. Table 1.7: Summary of Crop Yield Responses to Climate Figure 1.19: Meta-analysis of crop yield reductions. Change, Adaptation Measures, and CO2 Fertilization. Slope R2 T-stat P-value Full dataset 0.0023 0 0.1255 0.9 Crop yield change with 0.07 0.266 2.81 0.009** effect of CO2 fertilization Studies not considering –0.065 0.24 –2.65 0.0145* the effects of adaptation measures or CO2 fertilization Results of a general linear model applied to all studies with reported values for changes in yield and changes in temperature, to studies considering the effect of CO2 fertilization, and to studies not considering the effects of adaptation measures nor those of CO2 fertilization. Significance levels: *P<0.05, **P<0.01, ***P<0.001. Best-fit line for LAC studies not considering the effects of adaptation mea- sures or those of CO2 fertilization (blue line) and for studies considering the effects of CO2 fertilization (but no adaptation, orange) and their 95 percent confidence intervals of regressions consistent with the data based on 500 bootstrap samples (patches). 50 Lati n Ame r i ca and the Caribbean To conclude, the possible effects of climate change on crop 2020, by 16.2 percent by2050, and by 22.1 percent by 2080 with yields in the region are very diverse. Yield impacts differ among 2.9°C regional warming (ECLAC 2010). regions and crops and also among different GCMs, emissions sce- narios, and crop models (see Table 3.6) (Berg et al. 2013). Most of 4.3.6  Climate Change Impacts on Food Security the effects of rising temperatures are expected to be negative, even Nelson, Rosegrant and Koo et al. (2010) project that international if lessened CO2 fertilization (which introduces large uncertainties crop prices will increase significantly even when ignoring climate into the impact projections). For some crops, however, increasing change—mainly driven by population growth, income growth, temperatures might have positive effects, such as increasing yields and demand for biofuels. The price of wheat is projected to for rice and sugarcane. increase by 39 percent, rice by 62 percent, maize by 63 percent, and soybeans by 72 percent. Including climate change with global 4.3.4  Climate Change Impacts on Livestock mean temperature increasing by 2.5 °C by 2050, and without CO2 The livestock sector in the LAC region is of high economic fertilization, would accelerate price increases by an additional importance, especially in major livestock producing and export- 94–111 percent for wheat, 32–37 percent for rice, 52–55 percent for ing countries Brazil and Argentina (ECLAC et al. 2012), and the maize, and 11–14 percent for soybeans. Including CO2 fertilization impacts of climate change on livestock systems in developing would lead to less severe price increases by 2050 (Nelson, Rose- countries are diverse (Thornton et al. 2009). Climate change can grant, Koo et al. 2010). These results are confirmed by the IPCC severely impact the quantity and quality of feed, as rising tem- AR5 report: Increasing food prices as a consequence of changing peratures, increasing atmospheric CO2 concentrations, and changes climatic conditions are to be expected by 2050 without taking the in precipitation patterns influence the availability of nutrients, CO2 fertilization effect into account; including elevated CO2 will the productivity of grasslands, and the composition of pastures. temper price increases (Porter et al. 2014). Furthermore, heat stress directly affects livestock productivity. Climate change poses great risks to the economic develop- Cattle, in particular, are susceptible to high temperatures. Heat ment of Latin America and the Caribbean; it not only threatens stress is known to reduce food intake and milk production and economic growth but also poverty reduction and food security also to affect reproduction, growth, and cattle mortality rates (ECLAC 2010). Without climate change, calorie availability would (Porter et al. 2014). Higher temperatures are also closely linked to be expected to increase by 3.7 percent, up to 2,985 calories per growing water demand for livestock, increasing the competition capita in 2050 in LAC (Nelson, Rosegrant, Koo et al. 2010). How- and demand for water in water-scarce regions. More scientific ever, with climate change and without CO2 fertilization, per capita research is needed, meanwhile, on the effects of climate change calorie availability in 2050 is expected to drop below the value on livestock diseases and livestock biodiversity. for the year 2000 (2,879 calories per capita) (Nelson, Rosegrant, Koo et al. 2010). These projections show that climate change 4.3.5  Projected Impacts on Livestock threatens food security, especially for people with low incomes, With a 2.7°C warming by 2060, livestock species choice (i.e., the as access to food is highly dependent on income (FAO 2013). The adoption of new livestock) is projected to decline across Argen- cascading impacts of warming that reduce productivity in other tina, Brazil, Chile, Colombia, Ecuador, Uruguay, and Venezuela sectors apart from agriculture can further reduce economic output by 3.2 percent for beef cattle; by 2.3 percent for dairy cattle; by and negatively affect incomes (Porter et al. 2014). Results from 0.9 percent for chicken; and by 0.5 percent for pigs. Meanwhile, a Brazilian study (Assad et al. 2013) on climate change impacts the adoption of sheep species is projected to increase by an aver- on agriculture to 2030 project that Brazil could face a reduction age of 7 percent across the region, and by more in Colombia of approximately 11 million hectares of high quality agricultural (11.3 percent), Chile (14.45 percent), and Ecuador (19.27 percent) land as a result of climate change with the South Region (current (Seo et al. 2010). grain belt) being the worst impacted losing ~5 million ha of ‘low With a lower warming of 1.3–2.3°C by 2060, the pattern of climate risk’ crop land. The increase in climate risk in the south declining livestock species choice for beef cattle, dairy cattle, could be partially offset by transferring grain production to the chicken, and pigs, and the increasing choice of sheep, remains the central region currently occupied by low productivity pastures same but is less pronounced (Seo et al. 2010). According to Seo et al. (sub regional reallocation). Intensification of livestock and pasture (2010), the choice of sheep increases with increasing temperatures systems will also offset projected losses due to climate change. and decreasing precipitation because sheep are better adapted to In general, however, the production declines can be expected to these conditions than other livestock species. In Paraguay, beef impact prices, domestic demand, and net exports of most crops/ cattle production is projected to increase by 4.4 percent by 2020, livestock products. Simulations from this study across all the but then decline by 7.4 percent by 2050 and 27.1 percent by 2080 climate change scenarios suggest that rising staple and export in a scenario leading to 3.5°C regional warming (ECLAC 2010). crop and beef prices could double the agricultural contribution Beef cattle production is projected to decline by 1.5 percent by to Brazil’s economy. 51 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal Box 1.10: Critical Ecosystem Services of High Andean Mountain Ecosystems A highly critical ecosystem service provided by the LAC region is that of carbon storage. For example the ecosystems of the Andean mountains, including tropical montane cloud forests, the high-altitude wetlands, and the páramos ecosystem, store large amounts of carbon. Despite the fact that they cover a mere 3 percent of global land area they store about 30 percent of the global carbon stock of terrestrial ecosystems (Peña et al. 2011). Further, numerous large cities (such as Quito, Bogota or La Paz) extract part of their water supply from páramos areas. 4.3.7 Synthesis levels. Warren et al. (2013) found that, globally, 57 percent of The results of the climate change impact projections on crop yields plants and 34 percent of animals will lose greater than 50 percent differ among studies, but most authors agree that climate change of their habitat in a 4°C world. will very likely decrease agricultural yields of important food A comparative review of different model predictions across taxa crops in LAC (see Table 3.6) (ECLAC 2010; Fernandes et al. 2012; and regions revealed a large variability in the predicted ranges of Nelson, Rosegrant, Koo et al. 2010). An exception is the possible biodiversity loss, especially at the local level (Bellard et al. 2012). yield development of rice in some regions (ECLAC 2010; Fernandes One reason for this variability is that there is still high uncertainty et al. 2012; Nelson, Rosegrant, Koo et al. 2010). Although studies about the capacity of species to buffer the effects of climate change on climate change impacts on livestock are scarce (Thornton et al. (Moritz and Agudo 2013). Nonetheless, Scholes et al. (2014) state 2009), the few studies that are available indicate that beef and that there is “high confidence that climate change will contribute dairy cattle production will decline under increasing temperatures, to increased extinction risk for terrestrial and freshwater species as heat stress is a major influencing factor of cattle productivity over the coming century.” (Seo et al. 2010; Thornton et al. 2009). Sheep production could 4.4.3  Projections of Potential Future Shifts become more important in the future, as sheep are better adapted to in Ecosystems and Ecoregions warmer and drier conditions than cattle and pigs (Seo et al. 2010). The G200 ecoregions (Olson and Dinerstein 2002) located in Latin America and the Caribbean may experience severe climate 4.4  Climate Change Impacts on Biodiversity change in the future (Beaumont et al. 2010). Li et al. (2013) found 4.4.1  Current Status and Current Threats strong local climatic changes in the ecoregions Coastal Venezuela to Biodiversity Montane Forests, Amazon River and Flooded Forests, and Atlantic Biodiversity, the diversity of genes, populations, species, communi- Dry Forests for 2–4°C global warming. Further, 38.4 percent of the ties, ecosystems, and biomes, is the foundation for all ecosystem surface of the biodiversity hotspot of Tumbes-Choco-Magdalena processes (MEA 2005). Climate change is a major threat to biodi- and 11.5 percent of the Mesoamerican biodiversity hotspot will be versity, as species have evolved to live within specific temperature experiencing no-analogue climates in a warmer than 2°C world ranges that may be surpassed faster than species are able to adapt. (Garcia-Lopez et al. 2013). South America is a biodiversity hotspot, particularly due to Heyder et al. (2011) find a range of small to severe eco- the large extent of tropical rainforests (MEA 2005; Myers et al. system changes for the whole South American continent in 2000) and the continent’s long geographical isolation until approxi- their projections for a 2°C and warmer world. In a 4°C world, mately 3 million years ago—which together have nurtured a high results of one dynamic vegetation model show severe ecosystem number of endemic species. Habitat destruction and fragmenta- changes for more than 33 percent of the area in 21 out of 26 tion by land-use change as well as the commercial exploitation distinct biogeographic regions in South America (Gerten et al. of species groups are currently larger threats to biodiversity than 2013). Warszawski et al. (2013), meanwhile, projected such climate change (e.g., Hof et al. 2011). Land-use change is expected severe ecosystem changes in a 3°C world in South America to have a greater impact on plants than climate change by 2050, (notably in Amazon, Guyana moist forests, and Brazilian Cer- after which climate change becomes increasingly important for rado) when applying an ensemble of seven dynamic vegeta- species loss (MEA 2005; Vuuren et al. 2006). tion models. Imbach et al. (2012) projected that such severe ecosystem changes at global mean warming levels greater than 4.4.2  Impacts of Future Climate Change 3°C would lead to a considerable decrease in tree cover, indi- on Biodiversity cated by a change in leaf area index of more than 20 percent Forecasts of future changes in biodiversity are generally alarming across 77–89 percent of the area. Bellard et al. (2014) projected (e.g., Bellard et al. 2012; Foden et al. 2013). Using a global meta- that out of 723 Caribbean islands, 63 and 356 of them will be analysis, MacLean and Wilson (2011) found a mean extinction entirely submerged under one and six meters of sea-level rise, probability of 10 percent by 2100 across taxa, regions, and warming respectively. They also found that 165 of the islands will be at 52 Lati n Ame r i ca and the Caribbean least half-submerged (i.e., having lost more than 50 percent of to experience at least 50 percent species turnover, so that future their area) under one meter of sea-level rise, and 533 under communities would bear little resemblance to the currently estab- six meters of sea-level rise. While a six meter sea-level rise is lished ones (Lawler et al. 2009). In a greater than 3°C world, the not realistically expected to happen within this century, a one entire LAC region would experience high bioclimatic unsuitability meter sea-level rise is within the range of sea-level rise projected for amphibians in general (many grid cells between 50–80 percent under a global mean warming of 4°C at the end of this century loss). In the Northern Andes, 166 frog species (73 percent of local (see Section 3.7, Regional Sea-level Rise). frog fauna) and, in Central America, 211 species (66 percent of local salamander fauna), would lose their local climatic suitability 4.4.4  Projections of Habitat Changes, Species between 2070–2099 (Hof et al. 2011). Range and Distribution Shifts, and Extinction Risks Based on historical data, Sinervo et al. (2010) assume that if the for Species and Species Groups rate of change in maximum air temperature at 99 Mexican weather Microorganisms stations continues unabated by 2080, 56 percent of the viviparous Little is known about the consequences of future climate change on lizard species would go extinct by 2050 and 66 percent by 2080; of the microbial biodiversity due to the complex microbial feedback loops oviparous species, 46 percent would go extinct by 2050 and 61 percent within the climate system (Singh et al. 2010). The ratio between by 2080. By 2080, the predicted loss of suitable areas for the royal heterotrophic soil bacteria and fungi will likely be affected (Rinnan ground snake (Liophis reginae) is 30 percent (Mesquita et al. 2013). et al. 2007). Generally, temperature increase stimulates microbial Sea-level rise will affect the reproductive behavior of sea turtles, growth and accelerates decomposition, which leads to an increase which return to the same nesting sites every breeding season and in heterotrophic respiration (Davidson and Janssens 2006). therefore rely on relatively constant shorelines for laying their eggs. Invertebrates Fish et al. (2005) predict a 14/31/50 percent habitat loss of nest- Insects act as pollinators to ensure plant fertilization, but they may also ing sites for endangered sea turtles by 2050 under 0.2/0.5/0.9 m emerge as pests. Climate change affects temperature-driven reproductive sea-level rise respectively on Bonaire Island. Narrow and shallow cycles of many insect populations. In a 4°C world, Deutsch et al. (2008) beaches are predicted to be most vulnerable, but turtles seem to projected a range contraction of 20 percent for tropical insects, because prefer steep slopes which might to some extent alleviate climate tropical insects will face near-lethal temperatures much faster than change impacts at their preferred nesting sites. those in temperate climates. Estay et al. (2009) projected an increase Birds in insect population densities of grain pests in Chile of 10–14 percent Birds are most diverse in the tropics where they typically have in a 3°C world and 12–22 percent in a 4°C world. smaller home ranges than migratory birds in temperate zones (Jetz et al. 2007). This renders tropical bird diversity especially vulner- Amphibians and Reptiles able to extinctions caused by climate change and accompanying Due to the difficulties in entangling the relative contributions habitat destruction. Anciaes et al. (2006) projected that 50 percent of climate versus land use change, Scholes et al. (2014) stated of 49 neotropical manakin (passerine bird) species will have lost that “due to low agreement among studies, there is only medium more than 80 percent of their current habitat by around 2055 with confidence in detection of extinctions and attribution of Central mean global warming of 2°C. For a similar time frame and warming American amphibian extinctions to climate change.” Amphibians scenario, Souza et al. (2011) projected that 44 of 51 endemic Brazil- are particularly vulnerable because, due to their permeable skin, ian Atlantic forest bird species would lose their distribution area they depend on constant water availability at least during some by 2050, which corresponds to a habitat reduction of 45 percent periods of their life cycle. Loyola et al. (2013) found that most of the of the original area. The study assumes that the entire area of the 444 amphibian species in the Atlantic Forest Biodiversity hotspot Atlantic forest is suitable for these bird species. However, about in Brazil could increase their range, while 160 species would face 80 percent of the Atlantic forest is already deforested, and most range contractions with 1.9°C global warming in 2050. A more remaining forest areas are fragmented and isolated. Twenty-six recent projection for 2050 includes different dispersal scenarios for Cerrado bird species face 14–80 percent range contractions under the amphibians in this region and projects a majority of the 430 a no-dispersal scenario, and they face a 5 percent range increase amphibian species would face range contractions accompanied to a 74 percent range decrease under a full-dispersal scenario in by an overall species loss with 1.9°C global warming (Lemes et a 3°C world (Marini et al. 2009). al. 2014). Already in a 2°C world, 85–95 percent of species face net loss in range size, and 13–15 percent of species would lose Marsupials 100 percent of their current range depending on the modeled Most of the 55 marsupial species found in Brazil inhabit forested dispersal limitation (Lawler et al. 2009). In a 4°C world, most areas and are therefore exposed to both climate change and ecoregions are projected to experience at least 30 percent species land use change caused by deforestation. Loyola et al. (2012) turnover, and many in western South America and Central America found that marsupial species in Brazil face range contractions of 53 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal 67 percent of their original habitat with mean global warming of 2–5 percent of mammal species, 2–4 percent of bird species, and approximately 2°C by 2050. 1–7 percent of butterfly species in Mexico, as well as 38–66 per- cent of plant species in the Brazilian Cerrado, would go extinct Mammals (Thomas et al. 2004). At a warming of 1.8–2.0°C, these values Schloss et al. (2012) projected that up to 39 percent of the mammals increase to 2–8 percent, 3–5 percent, and 3–7 percent respectively. in the Neotropics would be unable to keep pace with climate-change With an increase in global mean temperature of greater than 2°C, velocity due to their limited dispersal abilities in a 4°C world. 44–79 percent of plant species in Amazonia are projected to go Torres, Jayat and Pacheco (2012) projected an at least 33 percent extinct (Thomas et al. 2004). habitat loss for the maned wolf (Chrysocyon brachyurus) in Central South America with 2°C global warming by 2050. Synthesis Climate change induced negative effects on biodiversity, from Plant Species range contractions to extinctions, are very likely in a warmer Plants are especially vulnerable to climate change because indi- than 2°C world. Climate change impacts on local biodiversity by vidual plants cannot migrate to avoid thermal stress. As a result, 2100 will depend on the balance between the number of species their dispersal mode will largely determine to what extent they abandoning an area and those facing local extinctions versus the may be able to adapt to changing climatic conditions. Moreover, number of species invading that same area due to thermal stress. plants directly respond to elevated atmospheric CO2 levels (see Species and species communities are possibly threatened by range Box 2.4)—but the degree to which some plant species may ben- contractions, extinctions, predator/prey disruptions, and phenology efit from rising CO2 levels is still being debated (Cox et al. 2013; changes due to climate and land use changes. Their opportunity Rammig et al. 2010). to survive in this changing environment lies in their capacity to Brazil is the country with the largest number of vascular plant adapt to these new conditions or to migrate to avoid them. As the species (>50,000) on Earth (ICSU-LAC 2010, p.57). Most future adaptive capacity of affected species and ecosystems is hard to projections paint a bleak picture for plant biodiversity, mostly due project or quantify, models need to use simplified approaches as to land-use change as a result of deforestation and, increasingly, the implemented in bioclimatic envelope models, species-distribution impacts of climate change. Simon et al (2013) projected a reduction models, and dynamic global vegetation models. in geographic distribution of 78 percent (±7 percent) in a >2°C One clear trend regarding future warming levels is that the world for 110 Brazilian Cerrado plant species. Feeley et al (2012) more temperatures are projected to increase, the more species projected a loss of suitable habitat area in the Amazon region of diversity is affected. Mountainous regions in the tropics (e.g., cloud between 8.2–81.5 percent in a 2°C world and 11.6–98.7 percent forests) are projected to become very vulnerable due to their high in a 4°C world, and a change in plant species richness between number of endemic and highly specialized species which might –4.1 percent to –89.8 percent in a 2°C world and –25.0 percent to face mountaintop extinctions. Most models do not take biotic complete loss for the studied species in a 4°C world. In Mexico, interactions (e.g., food-web interactions, species competition) and even common species are under threat, and great differences in resource limitations into account. Therefore, the realized ecologi- species response (0.1–64 percent loss) to regional warming above cal niche of species within an ecosystem might be much smaller 1.5°C in 2050 even among related tree species (e.g., oak trees) are than what is potentially possible according to climatic and other being projected (Gómez-Mendoza and Arriaga 2007). environmental conditions leading to shifts in ecological zones. Species Groups 4.5  Amazon Rainforest Dieback Most studies on range contractions focus on single species or spe- and Tipping Point cies groups; fewer studies have attempted to project the impact of future climate change at the community or biome level. Rojas- Old-growth rainforests in the Amazon basin store approximately Soto et al. (2012) projected a reduction of 54–76 percent in the 100 billion tons of carbon in their biomass (Malhi et al. 2006; extent of the Mexican cloud forest with 2°C global warming by Saatchi et al. 2011). Through evapotranspiration, Amazon rainfor- 2050. They concluded that this reduction forces tree communities ests recycle 28–48 percent of precipitation and contribute to local to move about 200 m to higher elevations. Similarly, Ponce-Reyes rainfall (van der Ent et al. 2010). A loss of these forest ecosystems et al. (2012) projected a 68 percent loss of suitable area for cloud due to climate change would release an enormous amount of forests in Mexico with a global warming of 3°C by 2080. Alarm- carbon into the atmosphere and reduce their evapotranspiration ingly, 90 percent of the cloud forest that is currently protected potential (thereby reducing atmospheric moisture); this would lead will not be climatically suitable for this ecosystem by 2080. As a to strong climate feedbacks (Betts et al. 2004; Costa and Pires 2010; consequence, climate change may lead to the extinction of 9 of Cox et al. 2004). These climate feedbacks, in combination with the 37 vertebrate species restricted to Mexican cloud forests. With large-scale deforestation, put the Amazon rainforest on the list of an increase of global mean temperature of 0.8–1.7°C by 2050, potential tipping elments in the Earth system (Lenton et al. 2008). 54 Lati n Ame r i ca and the Caribbean Factors Leading to Forest Dieback to cause large-scale forest dieback and to increase atmospheric and Potential Feedbacks CO2 concentrations. Observations of the Current Period Extreme drought events in combination with land use changes Current observations show that forests in Amazonia are adapted lead to an increased frequency in forest fires because the flamma- to seasonal drought (Davidson et al. 2012) mainly due to the bility of forests increases with a more open forest canopy (i.e., as ability to access deep soil water through deep rooting systems it allows more radiation to dry out the forest surface and enhance (Nepstad et al. 1994). It has been long debated, however, whether fire spread) (Ray et al. 2005). Fires were twice as frequent in 2005 the productivity of tropical rainforests during the dry season is as during the average of the previous seven years and they were more limited by precipitation or by cloud cover. Depending on the spatially concentrated in the arc of deforestation in the southern method used, remote sensing or modeling, seasonal droughts were Amazon (Zeng et al. 2008). Increasing fires resulting from defor- thought to enhance productivity either by more light entering the estation, pasture renewal, and other land-use-related activities canopy through reduced cloud cover, or by the combined effects increase the vulnerability of the Amazon rainforest to fire and of several interconnected processes (Brando et al. 2010; Huete et al. cause changes in forest composition and productivity (Brando et al. 2006). These findings have been challenged by remote sensing 2012; Morton et al. 2013). This interplay of factors is thought to experts who ascribe greening effects to saturation of the satellite initiate a positive bidirectional feedback loop between fire and sensor used (Samanta et al. 2011) or to changes in the optical forest which could initialize forest transformation into savannahs constellation of the sensor (near-infrared reflectance) (Morton and contribute to the Amazon tipping point (Nepstad et al. 2001, et al. 2014). Recent evidence from a large-scale and long-term 2008). Basin-wide measurements show that the combined effects experiment suggests that the feedbacks between climatic extreme of fire and drought can change the Amazon into a carbon source events such as droughts and forest fires increase the likelihood of (e.g., with 0.48 PgC emitted in 2010); it remains carbon-neutral, an Amazon dieback (Brando et al. 2014). meanwhile, during wet years (Gatti et al. 2014). Extreme weather events in the Amazon may have several causes. Deforestation is feared to influence the lateral moisture transport The drought events in 2005 and 2010 were not related to El Niño from coastal to inland areas because convective precipitation is but rather to high Atlantic sea-surface temperatures (Marengo et al. responsible for recycling precipitation locally. Walker et al. (2009) 2011). Cox et al. (2008) found the gradient between northern and showed that the current distribution of conservation areas in the southern tropical Atlantic lead to a warmer and drier atmosphere Amazon basin, which cover approximately 37 percent of the area, over the Amazon. Atypically low rainfall inflicted water stress on would be sufficient to maintain regional moisture transport and 1.9 million km² (2005) and 3.0 million km² (2010) of forest area recycling of precipitation when considering different deforesta- (Lewis et al. 2011). As a result, approximately 2.5 million km² tion rates. Influences of dynamic vegetation on water fluxes and (2005) and 3.2 million km² (2010) of forest area were affected by eventual carbon-climate feedbacks were not part of this study. increased tree mortality and reduced tree growth due to water Future Projections stress (Lewis et al. 2011). These two droughts are thought to have Water Stress reversed the currently assumed role of the intact forest as a carbon The Maximum Climatological Water Deficit (MCWD) (Aragão et sink and lead to decreased carbon storage of approximately 1.6 Pg al. 2007) is an indicator for drought intensity and plant water stress carbon (2005) and 2.2 Pg carbon (2010) compared to non-drought and correlates to tree mortality. For the period 2070–2099, 17 out years (Lewis et al. 2011; Phillips et al. 2009).34 The 2005 drought of 19 GCMs project increased water stress for Amazon rainforests reversed a long-term carbon sink in 136 permanent measurement in a 3°C world (which implies a mean regional warming of 5°C) plots (Phillips et al. 2009). (Malhi et al. 2009). Ten of 19 GCM projections passed the Two multi-year rainfall exclusion experiments in Caxiuanã and approximate bioclimatic threshold from rainforest to seasonal forest Tapajós National Forest generated remarkably similar results of (MCWD<–200 mm). Similarly, Zelazowski et al. (2011) projected water drought-induced tree mortality. These experiments demonstrated stress for forests to increase from 1980–2100 with an increase in global that once deep soil water is depleted, wood production is reduced mean temperature of 2–4°C above pre-industrial levels. They found by up to 62 percent, aboveground net primary productivity declines that humid tropical forests of Amazonia would retreat by 80 percent by 41 percent, and mortality rates for trees almost double (Brando for two out of 17 GCMs. Seven other models projected at least a et al. 2008; Costa and Pires 2010; Nepstad et al. 2007). Thus, an 10 percent contraction of the current extent of humid tropical forest. increase in extreme droughts in the Amazon region (medium con- Changes in Forest Cover fidence for Central South America in the IPCC SREX)(IPCC 2012) Hirota et al. (2010) simulated potential changes in Amazon for- or a prolonged dry season (Fu et al. 2013) may have the potential est cover. At a regional temperature increase of 2°C, forest cover decreased by 11 percent along with a 20 percent reduction in The combination of reduced uptake of carbon due to the drought and loss of carbon 34  precipitation in the Western Amazon forest (66°W). At a regional due to drought induced tree mortality and decomposition committed over several years. 55 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal temperature increase of 4°C, forest cover loss was 80 percent Biomass Loss independent of precipitation reduction. With fire included, tree Huntingford et al. (2013) showed that Amazon rainforest vegeta- cover was even further reduced. tion carbon generally increases in a 4°C world (and with regional Cook and Vizy (2008) projected a 69 percent reduction in temperature increases of up to 10°C). They conclude that there is rainforest cover extent in a 4°C world. Cook et al. (2012) showed evidence for forest resilience despite considerable uncertainties. that, with a regional warming of 3–4°C (which corresponds to Previous studies, however, projected considerable losses in a mean global warming of 3°C), soil moisture was reduced by biomass. In a 4°C world, Huntingford et al. (2008) found that eight percent, leaf area index (i.e., corresponds to forest cover) with a regional temperature increase of approximately 10°C from decreased by 12.6 percent, and the land-atmosphere carbon flux 1860–2100, vegetation carbon was reduced by about 7 kgC per m2. increased by about 27.2 percent due to fire from 2070–2099 com- Similarly, Fisher et al. (2010) simulated decreasing carbon stocks pared to 1961–1990. Cox et al. (2004) showed that a 4°C world of 15–20 kgC per m2 in 1950, 2.6–27 kgC per m2 in 2050, and could lead to a forest cover decrease of 10–80 percent. This find- 1–10 kgC per m2 in 2100 with a regional temperature increase of ing was recently challenged by Good et al. (2013), who showed about 2–5°C from 1900–2100. Galbraith et al. (2010) found that that an improved version of the Hadley model (HadGEM2-ES) vegetation carbon may either decrease or increase depending on projected only minimal changes in the Amazon forest extent due the emissions scenario and vegetation model. In their simulations, to forests surviving better in warmer and drier climates than pre- vegetation carbon changed by –10 to +35 percent relative to viously thought. However, about 40 percent of the difference in 1983–2002 with a regional warming of 4–8°C from 2003 to 2100. forest dieback projections was associated with differences in the Rammig et al. (2010) showed that including the CO2 fertiliza- projected changes in dry-season length in the new simulations, tion effect resulted in an increase in biomass in Eastern Ama- as a result of improvements in the simulated autotrophic plant zonia (EA) of 5.5–6.4 kgC per m2, in Northwestern Amazonia respiration, the soil moisture component, and a reduced gradient in (NWA) of 2.9–5.5 kgC per m2, and in Southern Amazonia (SA) of the tropical Atlantic sea surface temperatures. This does not mean 2.1–4.3 kgC per m2 in a 3°C world (see also Vergara and Scholz that the forest became more resilient in the updated model, but 2010). The probability of a dieback was zero percent for all regions rather that the improved vegetation-climate feedback mechanisms in this case. In contrast, climate-only effects without the buffering impose less stress on the simulated forests so that tree mortality CO2 fertilization effect resulted in biomass reduction in EA (–1.8 is reduced under warmer and drier climates (Good et al. 2013). In to –0.6 kgC per m2), NWA (–1.2 to 0.6 kgC per m2), and SA (–3.3 line with this, Cox et al. (2013) quantified a smaller risk (between to –2.6 kgC per m2) (Figure 3.20). The probability of biomass 1–21 percent) of Amazon dieback when constraining projections loss without CO2 fertilization was projected to 86.4 percent (EA), based on current observations of the atmospheric CO2 growth rate 85.9 percent (NWA), and 100 percent (SA). The probability of a and assuming that the CO2 fertilization effect is large. dieback (>25 percent biomass loss) was projected to be 15.7 percent Figure 1.20: Simulated precipitation changes in Eastern Amazonia from the 24 IPCC-AR4 GCMs with regional warming levels of 2–4.5 K (left panel). Simulated changes in biomass from LPJmL forced by the 24 IPCC-AR4 climate scenarios assuming strong CO2 fertilization effects (middle panel, CLIM+CO2) and no CO2 fertilization effects (CLIM only, right panel). Source: Calculated from Rammig et al. (2010). 56 Lati n Ame r i ca and the Caribbean (EA), 1.1 percent (NWA), and 61 percent (SA) (Rammig et al. 2010). 4.5.1 Synthesis These results imply that understanding the still uncertain strength Intensive research efforts over the past decades have enormously of the CO2 fertilization effect is critical for an accurate prediction improved the understanding and interaction of processes linking of the Amazon tipping point; it therefore urgently requires further climate, vegetation dynamics, land-use change, and fire in the empirical verification by experimental data from the Amazon region. Amazon. However, the identification of the processes and the Deforestation and forest degradation, for example from selec- quantification of thresholds at which an irreversible approach tive logging (Asner et al. 2005), are also factors which crucially toward a tipping point is triggered (e.g., a potential transition influence future changes in vegetation carbon. Gumpenberger et from forest to savannah) are still incomplete. al. (2010) found relative changes in carbon stocks of –35 percent Overall, the most recent studies suggest that the Amazon dieback to +40 percent in a protection scenario without deforestation and is an unlikely, but possible, future for the Amazon region (Good –55 percent to –5 percent with 50 percent deforestation in a 4°C et al. 2013). Projected future precipitation and the effects of CO2 world. Poulter et al. (2010) found a 24.5 percent agreement of fertilization on tropical tree growth remain the processes with the projections for a decrease in biomass in simulations with 9 GCMs highest uncertainty. Climate-driven changes in dry season length in a 4°C world. and recurrence of extreme drought years, as well as the impact of fires on forest degradation, add to the list of unknowns for which Large-Scale Moisture Transport combined effects still remain to be investigated in an integrative Several studies show that changes in moisture transport and regional study across the Amazon. A critical tipping point has been identified precipitation are strongly linked to deforestation. Costa and Pires at around 40 percent deforestation, when altered water and energy (2010) found in simulation runs with a coupled climate-vegetation feedbacks between remaining tropical forest and climate may lead model that precipitation was reduced in 9–11 of 12 months under to a decrease in precipitation (Sampaio et al. 2007). different deforestation scenarios (based on Soares-Filho et al. 2006). A basin-wide Amazon forest dieback caused by feedbacks Sampaio et al. (2007) performed simulations with a coupled climate- between climate and the global carbon cycle is a potential tipping vegetation model for different agricultural regimes and deforesta- point of high impact. Such a climate impact has been proposed tion scenarios of 20–100 percent for the Amazon basin (based on if regional temperatures increase by more than 4°C and global Soares-Filho et al. 2006). When replacing forest with pasture, they mean temperatures increase by more than 3°C toward the end of projected a 0.8°C increase in regional temperature and a 0.2 percent the 21st century. Recent analyses have, however, downgraded the reduction in precipitation at 20 percent deforestation levels. At 40 per- probability from 21 percent to 0.24 percent for the 4°C regional cent deforestation levels, regional temperatures increased by 1.7°C warming level when coupled carbon-cycle climate models are and precipitation was reduced by –2.2 percent. At 50–80 percent adjusted to better represent the inter-annual variability of tropical deforestation, regional temperatures increased by 1.8–2–1°C and temperatures and related CO2 emissions (Cox et al. 2013). This precipitation decreased by 5.8–14.9 percent. When replacing forest holds true, however, only when the CO2 fertilization effect is with soybeans at 50 percent deforestation, regional temperatures realized as implemented in current vegetation models (Rammig increased by 2.9°C and precipitation decreased by 4.6 percent. At et al. 2010). Moreover, large-scale forest degradation as a result 80–100 percent deforestation, regional temperatures increased by of increasing drought may already impair ecosystem services and 3.7–4.2°C and precipitation decreased by 19.2–25.8 percent. functions without a forest dieback necessarily to occur. Fire 4.6  Fisheries and Coral Reefs Studies projecting future fires in the Amazon are still scarce. Fires are projected to increase along major roads in the southern and 4.6.1  Vulnerability to Climate Change southwestern part of Amazonia with a 1.8°C global warming by Significant impacts of human origin, such as changes in tempera- 2040–2050 (Silvestrini et al. 2011; Soares-Filho et al. 2012). High ture, salinity, oxygen content, and pH levels, have been observed rates of deforestation would contribute to an increasing fire occur- for the oceans over the past 60 years (Pörtner et al. 2014). Such rence of 19 percent by 2050, whereas climate change alone would changes can have direct and indirect impacts on fishery resources account for a 12 percent increase (Silvestrini et al. 2011). Drought and food security (for example, as fish prey reacts sensitively to and anthropogenic fire incidences could significantly increase the ocean acidification or habitat is lost due to coral reef degradation) risk of future fires especially at the southern margin of the Amazon (Turley and Boot 2010). (Brando et al. 2014). If deforestation can be excluded in protected The Humboldt Current System off the coast of Peru and Chile areas, future fire risk would be evenly reduced, emphasizing the sustains one of the richest fisheries grounds in the world and management option to increase carbon storage when avoiding or is highly sensitive to climate variability such as that resulting reducing forest degradation (Silvestrini et al. 2011; Soares-Filho from ENSO. The Eastern Pacific region’s fishery is dominated by et al. 2012). catches of small pelagic fish which respond sensitively to changes 57 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal in oceanographic conditions. Peru and Colombia are among the ranges include 500–600 μatm, which is associated with warming eight countries whose fisheries are most vulnerable to climate of more than 2°C in 2100 and 851–1,370 μatm, which is associated change (Allison et al. 2009; Magrin et al. 2014). with 4°C warming. They found differential responses for coral The Caribbean Sea and parts of the South Atlantic, in contrast species, with 38 percent and 44 percent of all species studied to the Eastern Pacific, sustain vast coral reefs (see Section 4.6.2 exhibiting sensitivity to both scenarios. Echinoderms and mollusks Coral Reefs). The Caribbean Sea sustains a more diverse but less both exhibit high sensitivity to ocean acidification as they have productive fishery (UBC 2011). low metabolic rates and depend on calcium carbonate for shell formation. Wittmann and Pörtner (2013) stated that most species Uncertain Climate Change Effects on the Intensity of Coastal Upwelling studied will be affected in a 4°C world, but that the effects will be Several effects of climate change on upwelling and related ecosystem visible before that. In fact, nearly 50 percent of all species studied functioning have been hypothesized—and they point in opposite show sensitivity to a 2°C warming. While crustaceans appear directions. One hypothesis is that a decrease in productivity may to be relatively resilient, with about a third of species affected be driven by a globally warmer ocean in the future, as observed in a 4°C scenario, the effects on fish are already significant in a under El Niño conditions. However, sea-surface temperatures have relatively low CO2 concentration scenario; more than 40 percent not been observed to increase in the Humboldt Current System over of species are affected. This figure nearly doubles for the high the last 60 years (Hoegh-Guldberg et al. 2014). The hypothesis is scenario. However, Wittmann and Pörtner (2013) stress that their further contradicted by data indicating higher productivity during investigation into fish species´ sensitivity is biased toward reef fish. warm interglacial periods (Chavez and Messié 2009) and a projected Species Interaction and Ecosystem Effects weakening of trade winds and associated El Niño events (Bakun It is important to note that the effects of ocean acidification do not et al. 2010). However, particularly the latter projections are highly act in isolation; rather, they act in concert with such changes as uncertain. As described in Section 2.3.2 of the full report, El-Niño/ rising sea-surface temperatures, changes in salinity, and decreas- Southern Oscillation, projections on the frequency and intensity of ing nutrient availability due to enhanced stratification. These future El Niño events are uncertain. further interact with non-climatic pressures such as pollution and An increase in productivity, in turn, has been hypothesized to overfishing (Hoegh-Guldberg et al. 2014). For example, increasing occur due to a stronger land-sea temperature gradient—the land temperatures may lead to a drastic narrowing of species’ thermal surface warms faster than the ocean waters—leading to stronger tolerance window—with effects such as delayed spawning migra- winds driving stronger upwelling (Bakun 1990; Chavez and Messié tion or mortality (Pörtner and Farrell 2008). 2009). However, analysis has been unable to determine whether Sensitivity to ocean acidification can further narrow this toler- or not the recorded intensification of winds at the eastern sides ance window (Wittmann and Pörtner 2013). Such combined effects of the world’s oceans is due to inconsistencies in measurement are to date little understood and knowledge remains limited due to techniques over long time scales. Comparison with non-upwelling limits on experimental settings and the limited ability to discern regions, however, indicates that the wind intensification is more anthropogenic effects in a setting of high natural variability such pronounced in upwelling regions (Bakun et al. 2010). Analysis of as the Humboldt Current System (Hoegh-Guldberg et al. 2014). The changes in coastal biomass shows an upward trend of waters within expected synergistic effects of multiple pressures mean, however, the Humboldt Current System for 1998–2007, particularly for the that assessments remain conservative for a single or a subset of Peruvian coast. In contrast, the southern part of the California Cur- pressures (Wittmann and Pörtner 2013). rent System (south of 30°N off the coast of Baja California) shows A further potential risk factor for biological productivity and a negative trend (Demarcq 2009). Moreover, whether increased fisheries is the effect of climate-induced changes on species interac- upwelling would lead to higher productivity depends on nutrient tion, which can occur due to the differential responses of species availability—and changing physical conditions may disturb the to changing environmental cues. For example, phytoplankton and natural food web structure. zooplankton biomass changes may affect fish biomass (Taylor et al. Physiological Effects of Ocean Acidification 2012b). Different sensitivities to increasing CO2 concentration on Marine Species may lead to remarkable shifts in species composition (Turley and The IPCC states with high confidence that rising CO2 levels will Gattuso 2012). Similarly, asynchronous responses to warming increasingly affect marine organisms (Pörtner et al. 2014). A meta- may lead to mismatches in the predator-prey relationship. Such analysis of 228 studies revealed overall negative impacts of ocean effects have been detected to modify species interactions at five acidification not only on calcification but also on survival, growth, trophic levels (Pörtner et al. 2014). For the Humboldt Current development, and abundance (Kroeker et al. 2013). Wittmann and System, changes in the intensity of coastal upwelling add yet Pörtner (2013) conducted a meta-analysis of existing studies to another factor that may endanger the balance of various species determine responses for a range of future CO2 concentrations. The interactions. It has been hypothesized that the predator-prey 58 Lati n Ame r i ca and the Caribbean relationship between phyto- and zooplankton could be disrupted due to excessive offshore transportation of zooplankton. If in such Box 1.11: Freshwater Fisheries— a scenario overfishing would not allow for small pelagic fish to Vulnerability Factors to Climate control phytoplankton growth, sedimentation of organic matter Change may contribute to hypoxia, red tides, and the accumulation of methane (Bakun et al. 2010). The freshwater fishery of the Amazon River is an important source Projections of Changes to Coastal Upwelling of protein for the local population. According to the FAO, annual per The direction and magnitude of upwelling changes remains capita fish consumption in the Amazon basin may exceed 30 kg, which is significantly higher than in areas remote from freshwater uncertain, particularly for the Humboldt Current System. Wang sources where consumption has been estimated at around 9 kg per et al. (2010) showed that results diverge for different models. person per year (FAO 2010). As such, the river and its hydrological With approximately 1.5°C global warming in 2030–2039, projec- network provide an important source of proteins and minerals for tions show an overall increase in the decadal averaged upwell- the local population. ing index (July) for the California Current when compared to Those resources may be under threat from climate change 1980–1989 for most of the GCMs analyzed. For the Humboldt (Ficke et al. 2007) as rising water temperatures may exceed species’ Current System, however, there is very little agreement among temperature tolerance window. In addition, warmer waters are asso- models both in terms of direction and magnitude of change. ciated with higher toxicity of common pollutants (e.g., heavy metals) Wang et al. (2010) observed that the factors driving coastal and lower oxygen solubility; this may negatively affect exposed upwelling systems are too local to be captured by the coarse organisms. In addition, “blackwaters” such as the Amazon varzèa resolution of global models. lakes, depend on seasonal flooding for nutrient replenishment and for toxins to be flushed out. Reduced river flow and a reduced size Projected Changes to Fisheries Catch Potential of the flood plain may further lead to a reduced habitat for spawning In response to changing oceanic conditions, including seawater (Ficke et al. 2007). temperature and salinity, fish stocks have been observed in, and are further expected to shift to, higher latitudes (Perry et al. 2005). This ultimately affects local fisheries in the tropics and subtropics. fish catch is projected to decrease by up to 30°percent, but there A further climate impact on the productivity of fisheries is the are also increases towards the south. reduction in productivity at the base of the food chain due to the There are, however, inherent uncertainties in the projections stronger stratification of warming waters (Behrenfeld et al. 2006). presented here. Cheung et al (2010) pointed out that important In fact, primary productivity has been shown to have declined by factors, such as expected declines in ocean pH (ocean acidifica- 6 percent since the early 1980s (Gregg 2003). Declining pH levels tion), direct human pressures, and local processes which escape and increasing hypoxia may further negatively impact fisheries the coarse resolution of global models, are not taken into account. (Cheung et al. 2011). Incorporating the effects of decreasing ocean pH and reduced oxygen No regional projections of future fishery catches appear to availability in the northeast Atlantic yields catch potentials that exist. A global study that considers the habitat preference of are 20–30 percent lower relative to simulations not considering 1,066 commercially caught species and projects changes to pri- these factors (Cheung et al. 2011). mary productivity computes the expected changes in fish species Taking into consideration the effects of species interaction distribution and regional patterns of maximum catch potential by on redistribution and abundance, Fernandes et al. (2013) report 2055 in a scenario leading to warming of approximately 2°C in latitudinal shifts in the North Atlantic to be 20 percent lower than 2050 (and 4°C by 2100) (Cheung et al. 2010). reported by the bioclimatic envelope model developed by Cheung Results of Cheung et al. (2010) for LAC indicate a mixed picture et al. (2010). A further limitation of the studies is that results are (see Figure 3.21). Concurrent with the expectation of fish popula- given as 10-year averages and consequently do not take into con- tions migrating poleward into colder waters, the waters further sideration abrupt transitions as observed under El Niño conditions. offshore of the southern part of the Latin American continent are It should also be noted that not all captured species are included expected to see an up to 100 percent increase in catch potential. in these calculations, with small-scale fisheries possibly not taken Catch potentials are expected to decrease by 15–50 percent along into account (Estrella Arellano and Swartzman 2010). Finally, it the Caribbean coasts and by more than 50 percent off the Ama- needs to be taken into account that local changes in fish population zonas estuary and the Rio de la Plata. Furthermore, the Caribbean distribution are likely to affect the small-scale sector most severely, waters and parts of the Atlantic coast of Central America are as artisanal fishers will not have the means to capture the benefits expected to see declines in the range of 5–50 percent, with the of higher productivity at higher latitudes further offshore. waters around Cuba, Haiti, the Dominican Republic, and Puerto For the Exclusive Economic Zone of the Humboldt Current Rico, as well as Trinidad and Tobago, St. Lucia, and Barbados, System, Blanchard et al. (2012) projected a 35 percent decline in particularly severely affected. Along the coasts of Peru and Chile, 59 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal Figure 1.21: Change in maximum catch potential for Latin American and Caribbean waters. Source: Cheung et al. (2010), Figure 1a. phytoplankton and zooplankton density and similar magnitudes Vulnerability of Coral Reefs to Climate Change of change in the overall biomass of fish under 2°C global warming Coral reefs are particularly vulnerable to the double effects of by 2050. Comparing the impacts of climate change with fishing climate change on the oceans: rising temperatures and declin- pressure shows that climate impacts drive ecosystem change under ing pH levels. This vulnerability is particularly visible in events low fishing rates (0.2yr–1); under heavy fishing pressure (0.8yr–1), of coral bleaching, where external stresses cause corals to expel climate effects become secondary. their symbiotic algae (Hoegh-Guldberg 1999). Severe or prolonged Fisheries are in many cases at risk due to high fishing pressure, bleaching events are often followed by disease outbreaks and can with many commercially caught species already showing signs of cause coral mortality (Eakin et al. 2010). Bleaching events on a overexploitation. Climate change, by locally limiting productivity, large scale (“mass bleaching”) have been linked to unusually has the potential to further aggravate this situation. While sustain- high sea-surface temperatures, which exceed the temperature able fisheries management can significantly reduce the risks of threshold of affected species. Other factors exerting stresses on fisheries collapse, the uncertainty of climate impacts adds to the coral reef systems which have been identified as causes for coral challenge of establishing the quantity of fish that can be caught bleaching include pollution, overfishing, and the related shift in at sustainable levels (maximum sustainable yield). species composition (De’ath et al. 2012). A prolonged period of unusually high sea-surface temperatures 4.6.2  Coral Reefs across the Caribbean reefs for more than seven months in 2005 Coral reefs provide ecosystem services which are particularly caused the most extensive and most severe bleaching event recorded important at the local level for subsistence fisheries and tourism to date. Analyzing the concurrently high hurricane season during sector income (Hoegh-Guldberg et al. 2007). Healthy coral reefs that period, Trenberth and Shea (2006) found that about half of the also help to dampen the impact of coastal storm surges through observed sea-surface temperature anomaly was linked to global the reduction of wave energy (Villanoy et al. 2012). In the face of warming. Following the warming events, bleaching continued in rising sea-surface temperatures and declining pH-levels (see Section 2006 and was accompanied by disease and mortality. Mortality 2.3.8 of the full report) as well as in concert with local stressors reached 50 percent in a number of locations, with the strongest such as pollution, coral reefs and the services they provide are effects recorded in the northern and central Lesser Antilles and particularly vulnerable to climate change. less severe cases in the waters of Venezuela. 60 Lati n Ame r i ca and the Caribbean Hurricanes, while posing a direct threat to coral reef structures, The study from Meissner et al. (2012) is based on a single Earth have also been found to cool the surrounding waters and thereby System Model. Van Hooidonk et al. (2013) used a large ensemble reduce the warming signal and the risk of severe bleaching (Eakin of climate models to analyze the onset of bleaching conditions et al. 2010). The effect of hurricanes on coral reefs may thus be for different emissions scenarios. With warming leading to a 2°C positive in the sense that vertical mixing and upwelling caused world, the median year in which bleaching events start to occur by tropical cyclones may reduce heat stress for coral reefs. This annually is 2046. While this median applies to most regions within effect was reconstructed for the 2005 anomaly, for which Carrigan the Caribbean, some parts experience bleaching 5–15 years ear- and Puotinen (2014) found that nearly 75 percent of the assessed lier. These include the northern coast of Venezuela and Colombia area experienced cooling from tropical cyclones. They estimated as well as the coast of Panama. With warming leading to a 4°C that this lead to around a quarter of reefs not experiencing stresses world, the median year in which annual bleaching starts to occur above critical thresholds, outweighing the negative effects of direct is 2040 (with no earlier onset in the Caribbean region). Generally, damage (e.g., through breakage). While they pointed out that the the reefs in the northern waters of the Caribbean Sea appear to be relatively frequent occurrence of tropical cyclones may have sup- less sensitive than those in the south. However, as Caldeira (2013) ported the development of relatively resistant coral reef species, it points out, those reefs at the higher latitude fringes of the tropical remains unclear whether such resistance will persist in the face coral range (both north and south of the tropics) are likely to be of a projected increase in the intensity and frequency of tropical more heavily affected by ocean acidification. cyclones, particularly as such developments would be concur- Buddemeier et al. (2011) computed coral losses in the Caribbean rent with changes to ocean chemistry deleterious to coral reefs. for three different scenarios, which would lead to a 2°C, 3°C, and Dove et al. (2013) point out that expected reductions in reef net 4°C world by 2100. Temperature trajectories diverge around 2050, calcification, associated with changes in ocean chemistry under a by which time warming reaches about 1.2°C. A comparison of high atmospheric CO2 concentration, will significantly hinder the all trajectories shows little difference between scenarios in terms recovery of coral reefs after damages related to extreme events. of coral cover. By 2020, live coral reef cover is projected to have halved from its initial state. By the year 2050, live coral cover is Projections of Climate Change Impacts on Coral Reefs less than five percent; in 2100, it is less than three percent, with Based on observations and laboratory experiments, thresholds have no divergence among emissions scenarios. Notably, a 5–10 per- been identified which enable the projection of risk of bleaching cent live coral cover is assumed as the threshold below which events in the future. The decrease in calcium carbonate saturation the ecosystem no longer represents a coral reef (but is instead a and concurrent pH levels poses another threat to reef-building shallow-water ecosystem that contains individual coral organ- corals (Section 2.3.8 of the full report). Taking into consideration isms). Assuming a scenario in which corals are able to adapt by both the projected decrease in the availability of calcium carbonate gaining an additional 1°C of heat tolerance, the loss of live coral and increase in sea-surface temperatures, Meissner et al. (2012) cover below five percent is prolonged by around 30 years. Bud- projected that most coral reef locations in the Caribbean sea and demeier et al. (2011) noted that results can be extrapolated for western Atlantic will be subject to a 60–80 percent probability of the wider Southeast Caribbean, albeit with the caveat that the annual bleaching events with 2°C warming by 2050, with areas at assumed high mortality rate of 50 percent was lower in regions the coast of Guyana, Suriname, and French Guiana being exposed outside the Virgin Islands. to a 100 percent probability. In contrast, under 1.5°C warming by The modeling projections of future bleaching events and loss 2050, most locations in the Caribbean sea have a comparably low of coral cover presented above are based on changes in marine risk of 20–40 percent probability of annual bleaching events, with chemistry and thermology. While the adaptive potential of coral the waters of Guyana, Suriname, French Guiana, and the north reef species remains uncertain, it should be noted that further Pacific being at slightly higher risk (up to 60 percent probability). impacts impeding the resilience of coral reefs are not included By the year 2100, almost all coral reef locations are expected in these future estimates. These include potential impacts that to be subject to severe bleaching events occurring on an annual are likely to change in frequency and/or magnitude under future basis in a 4°C world. Exceptions are major upwelling regions, warming, such as hurricanes and the variability of extreme tem- which experience a risk of 50 percent. Compared to impacts in the peratures. Taking these uncertainties into account, Buddemeier year 2050, the Caribbean sea experiences more locations under et al. (2011) concluded that the presented projections are likely risk in 2100 despite no significant further increase in emissions unduly optimistic, which leads the authors to predict that the and temperatures, highlighting the long-term impact of climate “highly diverse, viable reef communities in the Eastern Caribbean change on marine ecosystems even under emissions stabiliza- seem likely to disappear within the lifetime of a single human tion. A potentially limiting assumption made by Meissner et al. generation.” According to some estimates, a 90 percent loss of (2012) is that no changes in the frequency or amplitude of El Niño coral reef cover would lead to direct economic losses of $8.712 events is expected. billion (2008 value) (Vergara et al. 2009). 61 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal Overall, while there are limitations to the projections of the leishmaniasis, and fascioliasis, and food- and water-borne diseases state of coral reefs in the future, a bleak picture emerges from the such as cholera and childhood diarrheal disease. Many of these available studies. Irrespective of the sensitivity threshold chosen, diseases have been found to be sensitive to changes in weather and indeed irrespective of the emissions scenario, by the year 2040 patterns brought about by ENSO. This indicates that disease Caribbean coral reefs are expected to experience annual bleaching transmission in LAC could prove highly responsive to changes in events. This is in accordance with Frieler et al. (2012) showing temperature and precipitation patterns induced by climate change. that, at the global scale, the global mean temperature at which Extreme weather events, including heat waves, hurricanes, floods, almost 90 percent of coral reefs are at risk of extinction is 1.5°C and landslides, also cause injuries and fatalities in Latin America, above pre-industrial levels. and these in turn can lead to outbreaks of disease. 4.6.3 Synthesis 4.7.1  Vector-Borne Diseases The rich fishery grounds of the Humboldt Current System in the Dengue fever is widespread in Latin America, with much of the Eastern Pacific react strongly to fluctuations in oceanic condi- region providing highly suitable climatic conditions to the primary tions related to the El Niño/Southern Oscillation (ENSO), during mosquito vector, Aedes aegypti. There has recently been a reemer- which the upwelling of nutrient-rich waters is suppressed by the gence and marked increase in the incidence of dengue fever and influx of warm surface waters. Together with ocean acidification dengue haemorrhagic fever in countries that had been declared and hypoxia, which are very likely to become more pronounced free of the illness following successful elimination programs in under high-emissions scenarios, the possibility of more extreme the 1950s and 1960s (Tapia-Conyer et al. 2009). El Niño events pose substantial risks to the world’s richest fishery Climate change is expected to play a contributing role in grounds. Irrespective of single events, the gradual warming of determining the incidence of the disease (Confalonieri et al. 2007), ocean waters has been observed and is further expected to affect although it is often difficult to separate the impact of climate fisheries particularly at a local scale. Generally, fish populations change from the impacts of urbanization and population mobil- are migrating poleward towards colder waters. Projections taking ity (Barclay 2008). In Brazil, the country with the largest number into consideration such responses indicate an increase of catch of cases in the world, a greater intensity of transmission of the potential by up to 100 percent in the south of Latin America. Off the disease has been observed during the hot, rainy months of the coast of Uruguay, the southern tip of Baja California and southern year (Teixeira et al. 2009). Between 2001–2009 in Rio de Janeiro, Brazil the maximum catch potential is projected to decrease by a 1°C increase in monthly minimum temperature was associated more than 50 percent. Caribbean waters may see declines in the with a 45 percent increase in dengue fever cases the following range of 5–50 percent, with the waters around Cuba, Haiti, the month, and a 10 mm increase in precipitation with a 6 percent Dominican Republic, and Puerto Rico, as well as Trinidad and increase (Gomes et al. 2012). Analysis from Mexico points to a Tobago, St Lucia, and Barbados, particularly severely affected. Along correlation between increases in the number of reported cases the coasts of Peru and Chile, fish catch is projected to decrease by and increases in rainfall, sea-surface temperature, and weekly up to 30°percent, but there are also increases towards the south. minimum temperature (Hurtado-Diaz et al. 2007). A study in It stands to reason that the communities directly affected by Puerto Rico, based on analysis of a 20-year period, likewise finds local decreases in maximum catch potential would not gain from a positive relationship between monthly changes in temperature catch potentials increasing elsewhere. They may also be the ones and precipitation and monthly changes in dengue transmission whose livelihoods would be most affected by the expected deleteri- (Johansson et al. 2009). Projections by Colon-Gonzalez et al ous effects of ocean acidification and warming on tropical coral (2013), holding all other factors constant, point to an upsurge in reefs. Irrespective of the sensitivity threshold chosen and indeed dengue incidence in Mexico of 12 percent by 2030, 22 percent by irrespective of the emissions scenario, by the year 2040 Caribbean 2050, and 33 percent by 2080 with a warming scenario leading to coral reefs are expected to experience annual bleaching events. a 3°C world in 2100; or 18 percent by 2030, 31 percent by 2050, While some species and particular locations appear to be more and 40 percent by 2080 with a warming scenario leading to a 4°C resilient to such events than others, it is clear that the marine world by 2100. In general, increases in minimum temperatures ecosystems of the Caribbean are facing large-scale changes with play the most decisive role in influencing dengue incidence, with far reaching consequences for associated livelihood activities as a sharp increase observed when minimum temperatures reach or well as for the coastal protection provided by healthy coral reefs. exceed 18°C (Colon-Gonzales et al. 2013). It appears, however, that climatic conditions alone cannot 4.7  Human Health account for rates of disease occurrence. Based on temperature-based mechanistic modeling for the period 1998–2011, Carbajo et al.(2012) The main human health risks in Latin America and the Caribbean found that temperature can estimate annual transmission risk include vector-borne diseases such as malaria, dengue fever, but cannot adequately explain the occurrence of the disease on a 62 Lati n Ame r i ca and the Caribbean national scale; geographic and demographic variables also appear the ENSO cycle and annual incidence of cutaneous leishmaniasis in to play a critical role. Colombia (Gomez et al. 2006) and visceral leishmaniasis in Brazil Malaria is endemic in Latin America, and rates of transmission (Franke et al. 2002). A study from Colombia of both types of the have increased over recent decades. This resurgence is associated disease also identified an increase in occurrence during El Niño and in part with local environmental changes in the region, such as a decrease during La Niña (Cardenas et al. 2006). These findings extensive deforestation in the Amazon basin (Moreno 2006). suggest that an increased frequency of drought conditions is likely Periodic epidemics have also been associated with the warm to increase the incidence of leishmaniasis (Cardenas et al. 2006). phases of ENSO (Arevalo-Herrera et al. 2012; Mantilla et al. 2009; Fascioliasis, a disease caused by flatworms and carried by snails Poveda et al. 2011). as an intermediate host, is a major human health problem in the It is possible that high temperatures could cause malaria to Andean countries of Bolivia, Peru, Chile, and Ecuador (Mas-Coma spread into high altitude cities (e.g., Quito, Mexico City) where 2005). Cases have also been reported in Argentina, Peru, Venezuela, it has not been seen for decades (Moreno 2006). Evidence shows Brazil, Mexico, Guatemala, and Cuba (Mas-Coma et al. 2014). The an increasing spread of malaria to higher elevations in northwest host infection incidence of fascioliasis is strongly dependent on Colombia during the last three decades due to rising temperatures, weather factors, including air temperature, rainfall, and/or potential indicating high risks under future warming (Siraj et al. 2014). evapotranspiration. Temperature increases associated with climate The connection between malaria and climate change, however, change may lead to higher infection and transmission rates and cause is unclear given the complexity of factors involved. Indeed, it is an expansion of the endemic zone, while increases in precipitation likely that the effect of climate change on malaria patterns will could, for example, increase the contamination risk window presently not be uniform. While incidence could increase in some areas, it linked to the November-April rainy season (Mas-Coma et al. 2009). is also possible that it may decrease in others—for example, in the Amazon, in Central America, and elsewhere where decreases in 4.7.2  Food- and Water-Borne Diseases precipitation are projected (Haines et al. 2006) (see Section 3.3, Cholera is transmitted primarily by fecal contamination of food Regional Precipitation Projections). and water supplies. Outbreaks are therefore often associated with Caminade et al. (2014) projected a lengthening of the malaria warm temperatures, flooding, and drought, all of which can aid transmission season in the highlands of Central America and contamination. Climatic variables have been shown to be decisive southern Brazil by the 2080s, but a shortening in the tropical in determining the extent of outbreaks (Koelle 2009). A recent regions of South America. This spatial differentiation is signifi- study of the relationship between rainfall and the dynamics of the cantly more pronounced with warming leading to a 4°C world cholera epidemic in Haiti, for example, shows a strong relationship than to a 2°C world. whereby increased rainfall is followed by increased cholera risk Earlier projections, however, offer mixed results. Béguin et al. 4–7 days later (Eisenberg et al. 2013). In South America, ENSO can (2011) projected an expansion of malarial area by 2050 in Brazil be a driving factor in cholera outbreaks in coastal areas because and isolated areas near the west coast of the continent under the El Niño phase provides warm estuarine waters with levels of approximately 2°C of global warming, although this is only if salinity, pH, and nutrients suitable for the blooming of the V. Chol- climatic and not socioeconomic changes are taken into account. erae pathogen (Martinez-Urtaza et al. 2008; Salazar-Lindo 2008). Van Lieshout et al. (2004), in contrast, found reductions in the Rates of childhood diarrheal disease have also been shown to size of the population exposed to malaria for at least three months be influenced by ambient temperature—and by ENSO in particular. of the year in all the scenarios they considered, and a reduction This was observed during the 1997–1998 El Niño event in Peru. in exposure to malaria for at least one month of the year in a 4°C During that particularly warm winter, in which ambient tempera- world (but not in a 3°C world). This study, meanwhile, projects tures reached more than 5°C above normal, hospital admissions for an expansion of the malarial zone southward beyond its current diarrheal disease among children increased by 200 percent over the southernmost distribution in South America—a finding consistent previous rate (Checkley et al. 2000). The relative risk of diarrheal with Caminade et al. (2014). disease in South America is expected to increase by 5–13 percent for Leishmaniasis is a skin disease carried by sandflies that takes the period 2010–39 with 1.3°C warming, and by 14–36 percent for the two main forms: cutaneous and visceral. Both are found through period 2070–99 with 3.1°C warming (Kolstad and Johansson 2011). much of the Americas from northern Argentina to southern Texas, 4.7.3  Impacts of Extreme Temperature Events excluding the Caribbean states (WHO 2014). A spatial analysis by Unusually high or low temperatures can potentially increase Valderrama-Ardilla et al. (2010) of a five-year outbreak of cutaneous morbidity and mortality, particularly in vulnerable groups such as leishmaniasis in Colombia beginning in 2003 identified temperature the elderly and the very young. A strong correlation has been found as a statistically significant variable. The study concluded, however, between unusually cold periods and excess deaths in Santiago, that climatic variables alone could not explain the spatial variation Chile, for example. In a time-series regression analysis, Muggeo of the disease. A positive association has been reported between 63 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal and Hajat (2009) estimated a 2.4 percent increase in all-cause to how changes in temperature and precipitation might affect the deaths among the above-65 age group for every 1°C decrease incidence of a particular disease in a particular location. below a cold threshold identified in their model. Cold-related risks Projections of how malaria incidence in LAC could be affected to human health would therefore be reduced if climate change by climate change over the rest of the century are somewhat results in a reduction in extreme cold events. inconsistent, with some studies pointing to increased incidence and Urban populations tend to be the most vulnerable to extreme others to decreased incidence. Such uncertainty also characterizes heat events due to the urban heat island effect, in which the studies of the relationship between climate change and malaria built environment amplifies temperatures. In northern Mexico, globally and reflects the complexity of the environmental factors heat waves have been correlated with increases in mortality rates influencing the disease. Little quantitative data is available on the (Mata and Nobre 2006); in Buenos Aires, 10 percent of summer future impacts of extreme weather events on human health, although deaths are associated with heat strain (de Garin and Bejaran studies based on historical data, such as that of Muggeo and Hajat 2003). Excessive heat exposure can cause or exacerbate a range (2009), have revealed a link between extreme temperatures and of health conditions, including dehydration, kidney disease, and increased rates of mortality in vulnerable sub-populations. cardiovascular and respiratory illnesses (Kjellstrom et al. 2010). Increased rates of hospital admissions of kidney disease patients 4.8 Migration have been documented during heat waves (Kjellstrom et al. 2010). While migration is not a new phenomenon in the region, it is Heat stress has been identified as a particular danger for work- expected to accelerate under climate change. There are many ers in Central America and one that coincides with high rates of areas in LAC prone to extreme events, including droughts, floods, kidney disease in some populations (Kjellstrom and Crowe 2011). landslides, and tropical cyclones, all of which can induce migra- 4.7.4  Impacts of Flooding and Landslides tion. Faced with severe impacts, migration might seem like the Torrential rain and resulting floods are among the main natural only option for finding alternative livelihoods (Andersen et al. hazards in the region and cause widespread injury and loss of 2010). However, migration typically causes an economic strain on life, livelihood, and property (Mata and Nobre 2006). Catastrophic both internal and external migrants (Raleigh et al. 2008), and not flooding has affected Mexico, Venezuela, Colombia, Brazil, Chile, everyone whose livelihood is threatened can afford to migrate. Argentina, and Uruguay in recent years (WHO/WMO 2012). Flood- The very poorest who do not have the necessary resources to ing can have multiple indirect health impacts, including the spread migrate can get trapped in a situation of ever-increasing poverty. of water-borne disease through water supply contamination and The transition from temporary to permanent migration resulting via the creation of stagnant pools that serve as habitats for disease from climate events can be facilitated by the existence of strong vectors such as malaria and dengue mosquitoes. Landslides and migration ties and networks (e.g., between LAC and the United mudslides can also be a consequence of flooding; these tend to States) (Deprez 2010). There are also important pull factors, which be exacerbated by factors such as deforestation and poor urban have been a major determinant of emigration in the past, including planning. Flash floods and landslides are a particular danger for more and better paid job opportunities and better access to services. informal settlements located on steep slopes and on alluvial plains They incentivize migration, particularly to North America or to (Hardoy and Pandiella 2009). countries within LAC with stronger economies. The question to Glacial lake outburst floods (see Box 3.4) also pose a risk to be assessed over time is whether climate change will make push populations located in the Andean region (Carey et al. 2012). The factors more important than pull factors. historical impacts of glacial lake outburst floods in Peru’s Cordillera The scientific literature on the interaction between migration and Blanca mountain range illustrate the potential for catastrophic loss of climate change is limited in terms of future projections. There is, life during periods of glacial retreat; many thousands of deaths have however, a growing body of literature on the demographic, economic, resulted from flooding, most notably in incidents in 1941, 1945, and and social processes of the interactions of climate and migration 1950 (Carey et al. 2012). The projection of further glacial melting in (Piguet et al. 2011; Tamer and Jäger 2010). Migration is considered the Andes (see Section 4.1, Glacial Retreat and Snowpack Changes) an adaptive response to maintain livelihoods under conditions of means that flooding continues to pose a risk to human populations. change. Assunção and Feres (2009) show that an increase in poverty levels by 3.2 percent through changes in agricultural productivity 4.7.5 Synthesis induced by regional warming of 1.5°C in 2030–2049 is reduced to The literature on the potential climate change impacts on human two percent if sectoral and geographic labor mobility is allowed health in the LAC region shows increased risks of morbidity for. This means that migration can reduce the potential impact of and mortality caused by infectious disease and extreme weather climate change on poverty (Andersen et al. 2010). events. Observed patterns of disease transmission associated with The projections of environmentally-induced migration agree different parts of the ENSO cycle seem to offer valuable clues as that most of the movement is likely to occur within the same 64 Lati n Ame r i ca and the Caribbean country or region (Deprez 2010). The largest trend in migration events to consider permanent domestic, regional, or international continues to be major movements from rural areas to urban areas. migration (ECLAC 2001). In 1998, Hurricane Mitch affected sev- Given the well-established migration channels between most LAC eral Central American countries and displaced up to two million countries and the United States, however, the impacts of climate people either temporarily or permanently. The impact was highly change may increase South-North migration flows. differentiated by country, with much lower displacement rates in There are no official statistics to show how many migrants Belize compared to Nicaragua, Honduras, and El Salvador, and a in LAC are moving in response to climate-related or other envi- 300 percent increase in international emigration from Honduras ronmental factors (Andersen et al. 2010). Although functioning (Glantz and Jamieson 2000; McLeman and Hunter 2011). Although as an adaptive strategy, environmentally-induced migration has the number of migrants has decreased over time, it has so far strong negative impacts on transitory areas and final destinations. remained above the level prior to the hurricane (McLeman 2011). For example, the International Organization for Migration reports that “rapid and unplanned urbanization has serious implications 4.8.3  Exacerbating Factors for urban welfare and urban services”, particularly in cities with The largest level of climate migration is likely to occur in areas “limited infrastructure and absorption capacity” (IOM 2009). For where non-environmental factors (e.g., poor governance, political areas of origin, the general consensus for LAC seems to be that the persecution, population pressures, and poverty) are already present impact of environmentally induced migration is overwhelmingly and already putting migratory pressures on local populations. In addi- negative (Deprez 2010). tion, poverty and an unequal geographical population distribution heighten people’s vulnerability to biophysical climate change impacts, 4.8.1 Drought thus compounding the potential for further migration (Deprez 2010). The effects of drought on migration have not been fully researched. Perch-Nielsen et al. (2008) explained that drought is the “most 4.8.4  Social Effects of Climate-Induced Migration complex and least understood natural hazard,” and that there Similar to traditional migrants, climate migrants with more edu- are a number of adaptive measures households might take before cation and skills are able to benefit the most from migration. resorting to migration. Nevertheless, scenarios based on projec- Benefits to migrants and their families can include, for example, tions of Mexico–United States migration rates (Feng et al. 2010) the possibility of finding better jobs. But migration can also have and of Brazilian internal migration (Barbieri et al. 2010) suggest a strong negative social impact on those who stay behind, par- that drought will lead to increased emigration along established ticularly on the poorest who typically do not have the resources migration routes and the depopulation of rural areas (Faist and to migrate and therefore risk being trapped in an adverse situation Schade 2013). Other examples of drought-induced migration with limited coping strategies (Andersen et al. 2010). In addition, include the flow of migrants from Brazil’s and Argentina’s north- climate-change-induced labor migration can have implications east regions to the state capitals and to the south-central regions on families left behind (e.g., challenges to children resulting (Andersen et al. 2010). Examples indicate that drought-induced from being raised in single parent homes with limited economic migration is already occurring in some regions. In Northeastern resources). In addition, climate change may induce greater levels Brazil, a primarily agricultural region, spikes in the rate of migra- of female migration; in the context of gender-based discrimination, tion to rapidly growing coastal cities or to the country’s central these women may face more challenges settling down and finding and southern regions have been observed following decreases in adequate housing and stable jobs (Deprez 2010). crop yields in years of severe drought (Bogardi 2008). Barbieri Labor migration can also provide benefits to migrants and et al. (2010) projected emigration rates in Brazil from rural areas their families. Migration can generate an increase in a family’s and found that depopulation is expected to occur—especially with financial assets, as work in the new location often pays better. increasing temperatures. This study, meanwhile, finds the biggest This contributes to better living conditions if the family is able to increase in migration coming from productive agricultural areas migrate together or generates remittances that can be sent back that support a large labor force. to help the family left behind. Despite some benefits, climate migrants face significant risks. 4.8.2  Sea-Level Rise and Hurricanes For example, the cost of migration (including travel, food, and Projections considering the impacts of sea-level rise on migration housing) can be very high and result in a worse financial situa- in Latin America and the Caribbean are sparse (Deprez 2010). tion for the family. There is also evidence that migrants’ working There is more research, however, on the impact of hurricanes. and housing standards can, in some cases, be very poor (such Although projections posit that migration resulting from hurricanes as in marginalized areas, informal settlements, and slums) with will continue to be mostly temporal and internal (Andersen et al. possible negative effects on health (Andersen et al. 2010). Further, 2010), stronger hurricane impacts in the Caribbean will increas- migrants who do not have networks or social capital in their new ingly drive households that have repeatedly suffered from these location can be socially isolated or discriminated against, resulting 65 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal Several countries in the region have also faced political Box 1.12: Distress Migration during instability in the past few years. For example, Bolivia has seen Hurricane Mitch some internal social movements calling for the independence of some regions; and there are important challenges related to the A typical example of distress migration took place when Hurricane drug trade which has become increasingly violent, particularly in Mitch struck Central America in 1998. Honduras evacuated 45,000 Mexico (Necco Carlomagno 2012). In fact, the activities of criminal citizens from Bay Island. The government of Belize issued a red alert groups and organized crime syndicates in countries like Brazil and asked citizens on offshore islands to leave for the mainland. and Mexico are a major source of some of the most significant Much of Belize City was evacuated. Guatemala issued a red alert conflicts (Rubin 2011). as well. By the time Mitch made landfall those evacuated along Climate change could aggravate these situations, further the western Caribbean coastline included 100,000 in Honduras, increasing conflicts over the use of resources. Socioeconomic 10,000 in Guatemala, and 20,000 in the Mexican state of Quintana disparities could be exacerbated, and in the worst case scenario, Roo. Despite this, nearly 11,000 people were killed and more than government capacities could be insufficient to face these along 11,000 were still missing by the end of 1998. In all, 2.7 million were left homeless or missing. The flooding caused damage estimated with natural disasters and climate-related challenges (McLeman at over $5 billion (1998 dollars; $6.5 billion in 2008 dollars). Source: 2011). In some Latin American countries where criminal organi- Andersen et al. (2010). zations already have significant power, such security gaps can enable them to increase their influence, further weakening the capacity of the state (Carius and Maas 2009). It is important to note that in the past, environmental degradation has often in tensions or conflict. In addition, ties to families and networks been used as a pretext for conflicts that are in fact caused by in the communities of origin may deteriorate while they are away underlying ethnic tensions and injustices associated with an (Andersen et al. 2010). unequal geographical distribution of the population and income In the case of climate-related evacuations, the social effects inequality (Deprez 2010). are mostly negative (see Box 3.12: Distress Migration during Hur- In this context, Rubin (2011) suggests four ways in which ricane Mitch). These include serious damage to physical assets climate change could increase the risk of conflict in LAC: (e.g., housing and livestock), and to other natural resources in the community of origin. In many cases, natural disasters can also • More resource scarcity. Climate change is likely to exacerbate contribute to financial and health problems (Andersen et al. 2010). resource scarcities. Increasing scarcity of food, water, forests, Migration contributes strongly to structural and sociodemo- energy, and land could intensify competition over the remaining graphic change in LAC cities. Migrants tend to come from similar resources, triggering internal unrest and even border conflicts. locations and settle in the same areas which usually are marginal • More migration. The LAC region has important migration areas in urban areas where they might have social capital or dynamics that are being exacerbated by climate change as social networks (Vignoli 2012). This contributes to creating social households face increased resource scarcities, rising sea levels, vulnerability to climate change by increasing spatial segregation and more (and more intense) natural disasters. Larger flows at the destination, or by modifying social networks of migrant of migrants could potentially destabilize destination countries. households in their origin (Pinto da Cunha 2011; Vignoli 2012). As a result, immigrant populations may lack the knowledge • Increasing instability. Climate change and variability may of disaster risk management plans, especially if they are new undermine the capacity of the state by increasing the cost of to urban areas and had not encountered disaster management infrastructure in remote rural areas, limiting the reach of the plans in the rural areas where they migrated from (Adamo 2013). state. This could be aggravated by the rising costs of disaster management (e.g., an increase in the level of agricultural 4.9  Human Security subsidies needed to maintain adequate food production) as well as by the general need for increased adaptation spending. The LAC region is considered to be at low risk of armed conflict, These limits in state capacity might result in a weakening of with the incidence of armed conflicts declining substantially in the relationship between the state and its citizens. the last 15 years (Rubin 2011). With the downfall of many military • Increasing frequency and intensity of natural disasters. The regimes in the 1980s, and continued economic integration, the chaotic conditions that follow in the wake of natural disasters region has achieved relative stability (Rubin 2011). However, in may provide opportunities for rebel groups to challenge the the context of high social and economic inequalities and migration government’s authority. flows across countries, disputes regarding access to resources, The empirical literature is inconclusive regarding the linkages land, and wealth are persistent. between climate change and an increasing risk of conflict globally. 66 Lati n Ame r i ca and the Caribbean Rubin (2011) suggests, however, that while resource scarcity in isola- 4.10  Coastal Infrastructure tion from other socioeconomic factors does not necessarily increase the risk of conflict, it often acts as a catalyst or driver, amplifying the Coastal areas, infrastructure, and cities are all vulnerable to cli- existing (often traditional) causes of conflict. In this sense, Haldén mate change. This is particularly true for the Caribbean region (2007) notes that disparities in standards of living and income can due to its low-lying areas and the population’s dependence on be problematic for several reasons: (1) disparities and divisions coastal and marine economic activities (Bishop and Payne 2012). might by themselves impede growth and undermine adaptation Tropical cyclones and sea-level rise represent the main risks as strategies; (2) substantial inequality might also destabilize societ- their combination can severely affect economic development (and ies and increase the risk of conflict in the light of climate change have generated significant losses and damages in past decades). and variability; and (3) the differences between large segments of For example, category 5 Tropical Cyclone Ike generated approxi- the populations imply that climate change will have very unequal mately $19 billion in damages, including $7.3 billion in Cuba impacts on the population, further exacerbating tensions. This is alone (Brown et al. 2010). particularly relevant in the LAC region, which has one of the most Overall losses induced by climate change stressors such as unequal income distributions in the world (Fereira et al. 2013). increased wind speed, storm surge, and coastal flooding could Large inequalities among groups (differentiated along ethnic, amount to 6 percent of GDP in some Caribbean countries (CCRIF religious, political, or geographical lines) increase the risk of violent 2010). Climate change-related impacts, including local sea-level conflict and high individual income inequality is a driver of crime rise, increased hurricane intensity, and modified precipitation and (Dahlberg and Gustavsson 2005; Fajnzylber et al. 2002; Østby 2007). temperature patterns could increase current economic losses by In the case of Bolivia, for example, highly unequal distributions of 33–50 percent by the 2030s (CCRIF 2010). natural resources create tensions among regions. In response to the nationalization of gas and oil reserves, the resource-rich regions 4.10.1  Impacts of Sea-Level Rise on Coastal Cities of Santa Cruz, Tarija, Beni, and Pando (comprising 35 percent of Several studies (Brecht et al. 2012; Hallegatte et al. 2013; Hanson the Bolivian population) unsuccessfully sought autonomy. Violent et al. 2011) have recently estimated the potential costs of sea-level disputes erupted between the government and the regions seeking rise, and the modification of storm patterns and land subsidence autonomy (Rubin 2011); this has since subsided due to the 2010 (which is not induced by climate change), for coastal cities in LAC. Autonomies Law and other agreements. Hallegatte et al. (2013) found that, by 2050, coastal flooding could Relatively small populations can have a tremendous impact on generate approximately $940 million of mean annual losses in the the environment (Hoffman and Grigera 2013). There are examples– 22 largest coastal cities in the region with a sea-level rise of 20 cm, notably in the Amazon basin—where the rural poor have turned and about $1.2 billion with a sea-level rise of 40 cm (Table 3.8). to illicit extractive activities (e.g., illegal logging) because they The study likely underestimates the overall impact as it only lack legal or formal alternatives. The effects of climate change assesses the costs of climate change on the largest coastal cities. and environmental degradation, along with the rapid growth of the extractive industry, is expected to take the greatest toll on the 4.10.2  Impacts on Port Infrastructure most vulnerable—small-hold farmers, indigenous populations, and Port infrastructures are particularly vulnerable to the direct and the poor (Hoffman and Grigera 2013). The resulting increase in indirect consequences of climate change (Becker et al. 2013). resource competition, (particularly for water and land)—together Becker et al. (2013) identified sea-level rise, higher storm surges, with increasing market pressure on landholders with tenuous legal river floods, and droughts as the main direct impacts, and coastal tenure—is expected to exacerbate existing inequities and tensions erosion, which could undermine port buildings and construction, surrounding the proper and equitable allocation of the region’s as one of the indirect impacts of climate change. The potential natural wealth (Hoffman and Grigera 2013). increase in tropical cyclone intensity may increase ships’ port Climate change could also increase violence in small com- downtime and, therefore, increase shipping costs (Chhetri et al. munal or household settings. One example is gender-based vio- 2013; Esteban et al. 2012). lence, which is already widespread in Latin America (Morrison Port infrastructure is crucial to economic development as et al. 2004). Although there is little research on this topic, there international trade is principally channeled through ports. Fur- are some studies (e.g., Harris and Hawrylyshyn 2012) which thermore, in Caribbean countries, port infrastructure plays a very indicate that climate change (by transforming livelihoods and significant role as they often are the only vector for trade in goods social structures) could spur social violence in non-conflict and assets (Bishop and Payne 2012). Impacts on seaports will situations. Moser and Rogers (2005) showed how rapid socio- also have indirect consequences on local economies as import economic changes might have a destabilizing effect not only on disruptions generate price increases for imported goods and societies but also within families, leading to an increased risk export disruptions lead to revenues and incomes decreasing at of domestic violence. the national level (Becker et al. 2012). 67 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal Table 1.8: Projected losses from sea-level rise under two different sea-level-rise scenarios and land subsidence in the largest LAC cities. 20 cm Sea-level Rise and Subsidence (no 40 cm Sea-level rise and Subsidence   adaptation) (no adaptation) Mean Increase due Mean Increase due to to SLR and Subsidence SLR and Subsidence Mean Annual Compared to Current Mean Annual Compared to Current Urban Agglomeration Loss (M$) Losses Loss (M$) Losses La Habana (Cuba) 9 5939% 21 13660% Port-au-Prince (Haiti) 8 1090% 11 1482% San Juan (Puerto Rico) 1.680 2365% 4.238 6118% Santo Domingo (Dominican 263 1166% 410 1880% Republic) Baixada Santista (Brazil) 274 3041% 467 5256% Barranquilla (Colombia) 87 1782% 102 2106% Belém (Brazil) 93 698% 586 4955% Buenos Aires (Argentina) 161 268% 592 1257% Panama City (Panama) 431 916% 451 962% Fortaleza (Brazil) 52 2762% 108 5814% Grande Vitória (Brazil) 2.643 1289% 10.096 5208% Guayaquil (Ecuador) 31.288 1012% 32.267 1047% Lima (Peru) 39 1009% 48 1254% Maceió (Brazil) 54 887% 283 5025% Maracaibo (Venezuela) 67 1086% 588 10238% Montevideo (Uruguay) 50 258% 180 1181% Natal (Brazil) 150 1505% 487 5100% Porto Alegre (Brazil) 71 641% 483 4918% Recife (Brazil) 259 1279% 970 5063% Rio de Janeiro (Brazil) 411 1088% 1.803 5108% Salvador (Brazil) 245 4903% 262 5248% San Jose (Costa Rica) 10 551% 67 4133% Total 2769.6 6164.4 Source: Hallegatte et al. (2013). 4.10.3  Impacts on Tourism Activities Beach tourism is particularly exposed to several direct and Tourism in the region, especially beach tourism in the Caribbean, is indirect climate change stressors, including sea-level rise, modi- projected to be affected by the impacts of climate change (Hyman fied tropical storm pattern, heightened storm surges, and coastal 2013). The total contribution of travel and tourism in the Caribbean erosion (Simpson et al. 2011). In a study comparing the vulner- was about 14 percent of the regional GDP and directly supported ability of four different tourist destinations in Jamaica, Hyman approximately 650,000 jobs (World Travel and Tourism Council 2013). (2013) found that coastal tourist resorts are two-to-three times As a result, the impact of climate change on tourism could detrimen- more exposed to climate change-related stressors than inland tally affect regional economic development (Simpson et al. 2011, 2010). touristic resorts. 68 Lati n Ame r i ca and the Caribbean 4.10.4  Impacts of Tropical Cyclones into account GDP and population projections but not potential Although projections on tropical cyclone frequency are still uncertain, adaptation measures. they indicate an augmentation of the number of Category 4 and 5 high- 4.11  Energy Systems intensity tropical cyclone on the Saffir-Simpson scale (see Chapter 3.6, Tropical Cyclones/Hurricanes). The losses and damages associated Energy access is a key requirement for development, as many with tropical cyclones making landfall are also projected to change economic activities depend on reliable electricity access (Akpan (Hallegatte 2007; Mendelsohn et al. 2011). Quantifying the future et al. 2013). At the individual and household level, electricity impact of tropical cyclones and their associated costs is complex as access enables income-generating activities, increases safety, and it involves not only climate model projections but also projections of contributes to human development (Deichmann et al. 2011). In socioeconomic conditions and potential adaptation measures. 35 36 37 LAC, the population generally has extended access to electricity in In a scenario leading to a 4°C world and featuring a 0.89–1.4 m rural and urban areas (apart from Haiti, where only 12 percent of sea-level rise, tropical cyclones in the Caribbean alone could gen- the rural population and 54 percent of the urban population had erate an extra $22 billion and $46 billion in storm and infrastruc- access to electricity in 2010) (World Bank 2013z). ture damages and tourism losses by 2050 and 2100, respectively, Climate change is projected to affect electricity production and compared to a scenario leading to a 2°C world (Bueno et al. distribution both globally and in the region (Sieber 2013). This 2008). The sea-level rise assumed in this study is based on challenges the LAC countries, which will have to increase or at semi-empirical sea-level rise projections (Rahmstorf 2007) and is least maintain electricity production at the current level to support higher than the upper bound projected in Section 3.7, Regional economic development and growing populations. Sea-level Rise. Curry et al. (2009) project that cumulative losses The effects of extreme weather events and climate change induced by tropical cyclones in the Caribbean, Central America, could lead to price increases and/or power outages (Ward 2013). and Mexico are going to increase to about $110 and $114 billion Thermal electricity and hydroelectricity are projected to be most during the period 2020–2025. These numbers assume an increasing vulnerable. Three types of climate-change-related stressors could tropical cyclone intensity of 2 percent and 5 percent, respectively, potentially affect thermal power generation and hydropower gen- compared to average values from 1995–2006. The majority of the eration: Increased air temperature (which would reduce thermal costs, approximately $79 billion in cumulative losses, would incur conversion efficiency); decreased available volume and increased in Mexico (Table 3.9). The estimates of Curry et al. (2009) take temperature of cooling water; and extreme weather events (which could affect the production plants, the distribution systems, and grid reliability) (Han et al. 2009; Sieber 2013). Table 1.9: Cumulative loss for the period 2020–2025 for Latin American and Caribbean sub-regions exposed to tropical 4.11.1  Current Exposure of the cyclones under scenarios A1 (constant frequency, intensity LAC’s Energy Systems increased by 2 percent) and A2 (constant frequency, intensity LAC countries have a diverse energy mix (Table 3.10). The majority of increased by 5 percent). the South American countries heavily rely on hydroelectricity(almost 100 percent, for example, in Paraguay); Central American countries Scenario A1 Scenario A2 use thermal electric sources and hydroelectricity. Caribbean coun- Sub-region (in million US$) (in million US$) tries, meanwhile, rely on thermal electric sources for electricity Mexico 79.665 79.665 production. Between 91 percent (for Jamaica) and 55 percent (for Central America Cuba) of the electricity consumed is generated from these sources. and Yucatan35 5.128 5.847 With a projected change in water availability, from decreas- Greater Antilles 36 22.771 26.041 ing precipitation and river runoff and/or increasing seasonality Lesser Antilles37 1.813 2.073 and shrinking snow caps and decreasing snow fall in the Latin Bahamas, The 985 1.241 American mountainous regions, thermal electricity plant cooling Total 110.362 114.867 systems may become less efficient and electricity production could be affected (Mika 2013; Sieber 2013). Hydroelectric power The data and calculations are based on Curry et al. (2009). Please note generation is similarly affected (Hamududu and Killingtveit 2012). that the scenarios named here ‘A1’ and ‘A2’ are not SRES scenarios but based on Emanuel (2005) and Webster et al. (2005). 4.11.2  Impacts of Climate Change on Energy Supply There are limited studies specifically quantifying the impacts of 35  Belize, Costa Rica, El Salvador, Guatemala, Honduras, and Nicaragua. 36  Cuba, Dominican Republic, Haiti, Jamaica, and Puerto Rico. climate change on thermal electricity and hydroelectricity genera- 37  Antigua and Barbuda, Barbados, Dominica, Grenada, St. Kitts and Nevis, St. Lucia, tion in LAC. As the larger share of the electricity produced in the and St. Vincent and the Grenadines. 69 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal Table 1.10: Electricity production from hydroelectric and thermoelectric sources, including natural gas, oil, coal, and nuclear in 2011 in the Latin American and Caribbean countries. Electricity Electricity Production from Production from Electricity Electricity Power Hydroelectric Thermoelectric Production from Consumption Sources Sources Other Sources Country or Region (kWh per capita) (% of total) (% of total) (% of total) Caribbean Cuba 1326.6 0.56 54.89 44.55 Dominican Republic 893.31 11.79 87.99 0.21 Haiti 32.49 16.71 78.97 4.32 Jamaica 1549.23 1.96 91.81 6.22 Trinidad and Tobago 6331.94 – 100 – Latin America Argentina 2967.39 24.36 73.97 1.66 Bolivia 623.37 32.50 64.10 3.41 Brazil 2437.96 80.55 12.77 6.68 Chile 3568.08 31.97 60.40 7.63 Colombia 1122.73 79.06 17.64 3.30 Costa Rica 1843.94 72.56 8.78 18.66 Ecuador 1192.28 54.93 42.27 2.79 El Salvador 829.57 34.64 34.06 31.30 Guatemala 539.08 39.84 33.10 27.07 Honduras 707.76 39.50 56.51 3.99 Nicaragua 521.58 11.61 65.99 22.41 Panama 1829.01 52.16 47.55 0.29 Paraguay 1228.19 100.00 0.00 0.00 Peru 1247.75 55.00 43.13 1.87 Uruguay 2810.12 62.64 28.10 9.26 Venezuela, RB 3312.68 68.55 31.45 0.00 Mexico 2091.69 12.26 84.13 3.62 Sources: World Bank (2013e, f, g, h, i, j). – means not available. region originates from hydropower, general impacts of climate would result in a decrease in annual power output of approxi- change on thermal electric generation plants are discussed in mately 10 percent, from 1540 gigawatt hours (GWh) to 1250 GWh Section 4.4.6 of the full report, Energy Systems. (Vergara et al. 2007). Hamududu and Killingtveit (2012) found that production will Hydropower increase by 0.30 TWh (or 0.03 percent) in the Caribbean compared Hydropower produces the larger share of electricity in the LAC to 2005 production levels, and by 0.63 TWh (or 0.05 percent) in region (see Table 3.10). The core natural resource for hydroelec- South America, under 2°C global warming by the middle of the tricity is river runoff, which has to be inter- and intra-annually 21st century. Maurer et al. (2009) projected the impacts of climate stable to allow hydropower installations to produce electricity most change on the Rio Lempa Basin of Central America (flowing through efficiently (Hamududu and Killingtveit 2012; Mukheibir 2013). In Guatemala, Honduras, and El Salvador and into the Pacific—see Peru it is estimated that a 50 percent reduction in glacier runoff Table 3.11). They concluded that an increase in frequency of 70 Lati n Ame r i ca and the Caribbean Table 1.11: Projected temperature and hydrologic changes basin, average river flow could be between –20 and +18 percent in the Rio Lempa River during the period 2040–2069 and with a global warming of 2.1°C depending on the different GCM 2070–2099 relative to the period 1961–1990 for hydrological chosen. This difference between the lowest and the highest esti- change and pre-industrial levels for temperature changes. mates highlights the limitations of the current models to project the potential hydropower production from dams built on this river Impact/Period Scenario 2040–2069 2070–2099 basin and reflects that projections vary from one GCM to another Temperature increase B1 +1.8°C +2.2°C (Nóbrega et al. 2011). Popescu et al. (2014) showed an increase above pre-industrial A2 +2.1°C +3.4°C levels in the maximum hydropower energy potential for the La Plata Basin of between 1–26 percent with a global warming of 1.8°C Precipitation change B1 – –5% relative to 1961–1990 by 2031–2050 (see Table 3.12). There were also great disparities A2 – –10.4% (median change) in the sub-basin projections depending on the model used. The Reservoir inflow relative B1 – –13% La Plata basin is one of the most economically important river to 1961–1990 (median A2 – –24% basins in Latin America (being part of Argentina, Bolivia, Brazil, change) Paraguay, and Uruguay). It has a major maximum hydropower Frequency of low flow B1 +22% +33% energy potential, producing on average 683,421 GWh per year relative to 1961–1990 A2 +31% +53% (median change) during the period 1991–2010 and 76 percent of the 97,800 MW total electricity generation capacity of the five countries in the La Source: Maurer et al. (2009). Plata basin (Popescu et al. 2014). The results of these studies, however, need to be interpreted with care. For example, the significant decrease in hydropower capacity at the micro-level as projected by Maurer et al. (2009) low-flows in scenarios leading to a 2°C world and a 3°C world strongly contrasts with the results of Hamududu and Killingtveit implies a proportional decrease in hydropower capacity for the (2012), who projected an increase in hydropower generation at two main large reservoirs used for hydroelectricity generation in El the macro-level. The Hamududu and Killingtveit study may be Salvador (Cerron Grande and 15 Setiembre). Low-flow frequency limited for several reasons. First, it does not take into account is a key indicator of the economic viability of hydropower infra- seasonality and the impacts of climate change on the timing of structures as it determines the firm power, which is the amount river flows. Second, changes in hydrology and temperatures are of “energy a hydropower facility is able to supply in dry years” accounted for at the country level but not at the river basin level; (Maurer et al. 2009). The projected increase in low-flow frequency this does not take into account potential spatial variability and could therefore reduce the economic return from the existing facil- changes occurring over short distances. Third, the study does not ity and reduce the return on investments in future hydroelectric consider the potential impacts of floods and droughts, which have infrastructures (Maurer et al. 2009). very significant impacts on hydroelectricity generation and are For Brazil, de Lucena et al.(2009) project that average annual projected to occur more frequently and with a greater intensity in river flows will decrease by 10.80 percent with 2.9°C global warm- the coming decades (Marengo et al. 2012, 2013; Vörösmarty et al. ing, and by 8.6 percent with 3.5°C global warming, by 2071–2100. 2002) (see also Section 4.2, Water Resources, Water Security, This decrease in annual flow will lead to a decrease in firm power and Floods). Finally, the study does not consider the impacts on of 3.2 and 1.6 percent, respectively, during this time period com- river runoff from decreasing snow cover and snowfall in the Latin pared to production level for 1971–2000. For the Rio Grande river American mountainous regions (Barnett et al. 2005; Rabatel et al. Table 1.12: Maximum hydropower energy potential for the La Plata Basin with present, near future, and end-of-century climate conditions for two climate models (PROMES-UCLM and RCA-SMHI). Present Climate Future Climate 2031–2050 End of Century 2079–2098 Scenario 1991–2010 (1.8°C in Scenario A1B) (3.2°C in Scenario A1B) Energy Variation to Energy Variation to Energy (GWh/year) (GWh/year) Present (GWh/year) Present PROMES-UCLM 688,452 1.01 715,173 1.05 683,421 RCA-SMHI 861,214 1.26 838,587 1.23 Source: Popescu et al. (2014). 71 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal 2013; Vuille et al. 2008). These limitations could potentially explain Table 1.13: Climate change-related stressors projected to why their results are different from those of Maurer et al. (2009), affect hydroelectricity generation. who use monthly precipitation rates to calculate annual inflows. Further research is needed to adequately inform decision mak- Category of Climate- change-related Climate-change- ers in the region on the impacts of climate change on hydropower Stressors related Stressors generation. Similarly to Hamududu and Killingtveit (2012), de Long term trends or gradual Reduction in average precipitation Lucena et al. (2009) only accounted for the average behavior of changes induced by climate Increase in average precipitation flows and did not integrate potential change in seasonality or the change Increase in average temperature effects of extreme dry or wet events on hydropower generation. In Increase in extreme climate Drought this context, projections for hydropower production in Brazil by variability Flooding de Lucena et al. (2009) may underestimate the potential impacts of climate change. The projections by Popescu et al. (2014) only Indirect climate change impacts Water scarcity estimated a maximum hydropower potential, but this does not Siltation through land degradation mean that more hydropower electricity will be produced from Source: Mukheibir (2013). existing or future installations. For example, the specifications for existing dams (e.g., reservoir size, dam height, and so forth) may not be sufficient to efficiently manage projected excess flows. Despite these uncertainties, there are some clear climate change impacts on hydropower. Mukheibir (2013) inventoried the Oil and Gas climate-related stressors that are projected to affect hydroelectric- Some LAC countries, such as the Republica Bolivariana de Ven- ity generation. He separated climate-change-related stressors into ezuela, Brazil, and Mexico, benefit from significant oil and/or three categories: (1) long-term trends or gradual changes induced gas reserves. For example, Venezuela is the world’s tenth largest by climate change; (2) increases in extreme climate variability; exporter of oil and Mexico was the ninth biggest producer in and (3) indirect climate change impacts (Table 3.13). These 2013 (EIA 2014a). The production of gas and oil in LAC countries stressors could potentially reduce firm energy and increase vari- contributed 7.38 percent in 2012 and 12.01 percent in 2013 to ability and uncertainty of supply in the energy sector (Ebinger global production (Table 3.14). Furthermore, some countries in and Vergara 2011). the region have a very significant share of their GDP originating Table 1.14: Natural gas production for LAC countries in 2012 and oil production in 2013. Natural Gas Production in 2012 Oil Production in 2013 (in (in billion cubic feet) thousands of barrels per day) Argentina 1,557.39 707.91 Bolivia 652.27 64.46 Brazil 910.77 2,712.03 Chile 42.98 15.57 Colombia 1,110.30 1,028.47 Cuba 38.00 48.73 Ecuador 54.39 527.03 Mexico 1,684.42 2,907.83 Peru 639.91 174.96 Trinidad and Tobago 1,504.74 118.12 Venezuela, RB 2,682.81 2,489.24 Total LAC region 10,878.68 10,851.42 LAC percentage of global production 7.38% 12.01% The LAC countries not displayed in the table produce little oil or natural gas. Source: EIA (2014a; b). 72 Lati n Ame r i ca and the Caribbean from oil and natural gas rents (defined as the difference between Solar energy installations are subject to two types of climate- value of natural gas or oil production and the total cost of produc- related impacts: reduced insulation induced by cloudiness, which tion); in Venezuela and Trinidad and Tobago, for example, about decreases heat or electricity output, and extreme weather events 30 percent of 2012 GDP in 2012 came from oil and gas rents (World such as windstorms or hail, which could damage production units Bank 2013e; f). The assessment of climate change impacts on oil and their mounting structures (Arent et al. 2014). According to and gas presented here focuses on direct impacts and does not Arent et al. (2014), gradual climate change and extreme weather consider the possible decrease in fossil fuel assets value induced by events “do not pose particular constraints to the future deploy- future mitigation policies, which would contribute to a reduction ment of solar technologies.” Studies estimating projected impacts in fossil fuel demand at the global level (IPCC 2014d). of climate change and extreme weather events on solar energy Off-shore platforms and on-shore infrastructures are susceptible outputs are not available. to climate related impacts, such as sea-level rise and coastal ero- sion that could damage extraction, storage and refining facilities 4.11.3  Impacts of Tropical Cyclones (Dell and Pasteris 2010) and also to extreme weather events such on Power Outages as tropical cyclones which could lead to extraction and production The Caribbean and Central American regions are particularly disruption and platform evacuation (Cruz and Krausmann 2013). exposed to the impacts of tropical cyclones and a higher frequency For example, in the aftermath of tropical cyclones Katrina and of high-intensity tropical cyclones is projected (see Section 3.6, Rita in 2005, 109 oil platforms and five drilling rigs were damaged Tropical Cyclones/Hurricanes). Strong winds, heavy precipitation, leading to interruptions in production (Knabb et al. 2005). Cozzi and floods associated with tropical cyclones have the capacity to and Gül (2013) identify two key climate related risks for the LAC disrupt and even damage essential power generation and distribu- region: sea-level rise and an increase in storm activity (particularly tion infrastructures leading to power outages. A growing number of for Brazil). These risks would mainly lead to an increase in the studies have developed models to estimate and forecast the risks shutdown time of coastal refineries and an increase in offshore of power outages to energy systems in order to improve disaster platform costs, which will have to be more resistant to high-speed assistance and recovery (Cao et al. 2013; Han et al. 2009; Nateghi winds associated with tropical cyclones (Cozzi and Gül 2013). et al. 2013; Quiring et al. 2013). However, studies and models spe- cifically quantifying or taking into account the projected effect of Wind and Solar Energy climate change on tropical cyclones intensity and frequency and Solar and wind energy sources play an important role in climate the potential disruptions to power generation and distribution in change mitigation strategies to reduce global emissions from fossil the Caribbean and Central American countries are lacking. fuel combustion. Even though wind and solar energy still play a very minor role in LAC, significant development of the sector is 4.11.4  Effects of Climate Change projected (Bruckner et al. 2014). In this context, a more precise on Energy Demand understanding of the effect of climate change on these energy Climate change will also affect energy demand. Increasing tempera- sources is of great significance. tures and heat extremes (see Sections 3.1, Projected Temperature For wind energy, the main climate change impact relates to Changes, and 3.3.2, Heat Extremes) lead to a higher demand for changing wind patterns and how climate change will affect inter- air conditioning (Cozzi and Gül 2013); on the other hand, demand and intra-annual variability and geographical distribution of wind for heating may decrease. At the global level, Isaac and van (Arent et al. 2014). Despite significant progress, GCMs and RCMs Vuuren (2009) estimated that by 2100 in a 4°C world the number still do not produce very precise projections for inter-annual, sea- of cooling degree days will rise from 12,800 during the period sonal, or diurnal wind variability (Arent et al. 2014). Furthermore, 1971–1991 to 19,451, (a 51.9 percent increase) while demand for specific studies estimating the effects of climate change on wind heating (measured in heating degree days) was projected to remain patterns and therefore wind energy in LAC are still missing (Pryor almost constant. At the regional level, they project that by 2100 and Barthelmie 2013). However, drawing on conclusions from Pryor demand for heating is going to decrease by 34 percent compared and Barthelmie (2013) for the United States and Europe stating to the 1971–1991 period—from 364 to 240 heating degree days. that “generally, the magnitude of projected changes over Europe They project the demand for cooling to increase by 48 percent, and the contiguous USA are within the ‘conservative’ estimates from 1802 to 2679 cooling degree days. embedded within the Wind Turbine Design Standards,” it can be assumed that future climate change may not significantly affect 4.11.5 Synthesis The assessment of the current literature on climate change impacts wind energy supply. Pryor and Barthelmie (2013) nonetheless on energy in LAC shows that there are only a few studies, most of highlight the need for more research in this area to better quantify which make strong assumptions about key issues such as seasonality the effects of long-term climate change and extreme events on of water supply for hydropower. These studies are more qualitative wind energy supplies. 73 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal than quantitative, and important gaps remain. However, in general will add water stress to an area in which agriculture has already substantial climate change impacts can be expected for the energy shifted toward more water-intensive irrigated crop production since sector. There is also a lack of studies with respect to the impacts the 1990s (Bury et al. 2011). As the glacier reservoirs gradually of climate change impacts on renewable energies. In general, the disappear, however, runoff will tend to decrease (particularly in impacts of climate change on energy demand is less well studied the dry season). The peak in runoff is expected to be reached in than those on energy supply—and, yet, demand and supply interact about 20–50 years from now (Chevallier et al. 2011) if it has not in a dynamic way. For example, the concomitant increase in energy yet peaked already (Baraer et al. 2012). demand during heat extremes and the decrease of energy supply Changes to river flow translate into risks to stable water through reduced river flow and low efficiencies may put existing supplies for much of the region. Diminishing downriver water energy systems under increasing pressure in the future. flow will also undermine hydropower generation; crucial to the economic development of the continent (Hoffman and Grigera 5  Regional Development Narratives 2013) (Section 4.11). Current studies project that by 2050, up to 50 million people in the vast lowland area fed by Andean glacial In this section, implications of climate change for regional devel- melt, will be affected by the loss of dry season water for drink- opment are discussed in order to relate climate change impacts to ing, agriculture, sanitation, and hydropower (Cushing and Kopas existing and future vulnerabilities in the LAC region. The develop- 2011). Deforestation and land degradation can furthermore alter ment narratives are split into overarching development narratives the water cycle and possibly endanger water availability (Buytaert across the region and in sub-regional development narratives. It et al. 2006; Viviroli et al. 2011). is important to note that each development narratives presents Changes to the seasonal cycle of water availability affect the only one of the many possible ways in which climate change can ecosystems that rely on a stable water supply. Consequently eco- put key development trajectories at risk. Table 1.15 summarizes system services are put at risk. For example, freshwater fisheries the key climate change impacts under different warming levels may be exposed to climate-related risks as decreasing river flows in the Latin American and the Caribbean region and Figure 3.22 reduce the floodplain for spawning and the natural seasonal flood- summarizes the key sub-regional impacts. ing of lakes is reduced. A projected decrease in annual precipita- tion and increasing risk of drought will in turn increase the risk 5.1  Overarching Development Narratives of large-scale forest degradation, not only in the Amazon, with a loss of associated ecosystem services. 5.1.1  Changes to the Hydrological Cycle Endanger In the Andes in particular, water stress will reduce pasture the Stability of Freshwater Supplies and Ecosystem land availability in the dry season and increase the potential for Services conflict over land use (Kronik and Verner 2010). Social conflicts An altered hydrological system due to changing runoff, glacial melt, over water rights and water access may increase in the Peruvian and snowpack changes will affect the ecosystem services that the Andes between farming communities and mining companies. rural population depends on, freshwater provisioning in cities, Moreover, higher river flow peaks can lead to landslides and devas- and major economic activities such as mining and hydropower. tating floods associated with glacial lake outburst (Chevallier et al. Throughout the 20th century the tropical glaciers in the Central 2011)—with direct consequences for human lives and settlements. Andes have lost large amounts of their volume (see Section 4.1, Cities are highly vulnerable as continuous urbanization and Glacial Retreat and Snowpack Changes). As land surface tempera- population growth increases water demand (Hunt and Watkiss 2011) tures rise, this trend is expected to accelerate possibly leading to and as they depend on ecosystem services provided by surround- an almost complete deglaciation of 93–100 percent in a 4°C world. ing areas. The high Andean moorlands (known as páramos—see In concert with decreasing snowpack, changes to precipitation pat- Box 3.10, Critical Ecosystem Services of High Andean Mountain terns, and higher evaporation, increasing glacial melt will impact Ecosystems), which are key ecosystems able to stock large amounts the timing and magnitude of river flows. In general, runoff is of carbon on the ground and act as water regulators, are threatened projected to increase during the wet season, increasing flood risk by temperature rise, precipitation changes, and increasing human (see Section 4.2, Water Resources, Water Security, and Floods). activity. Major population centers, such as Bogota and Quito, rely Accelerated melting rates may lead to a localized short-term surge on páramo water as a significant supply source. The melting of in water supply that might lead to unsustainable dependency (Vuille the Andean glaciers, increasingly unpredictable seasonal rainfall 2013). For example, in the area known as Callejon de Huaylas in the patterns, and the overuse of underground reserves are affecting the central highlands of Peru, climate-change-induced glacier retreat urban centers of the highlands (e.g., La Paz, El Alto, and Cusco), 74 Lati n Ame r i ca and the Caribbean Figure 1.22: Sub-regional risks for development in Latin America and the Caribbean (LAC) under 4°C warming in 2100 compared to pre-industrial temperatures. Central America & the Caribbean Higher ENSO and tropical cyclone frequency, precipitation extremes, drought, and heat Dry Regions waves. Risks of reduced water availability, crop yields, food security, and coastal safety. Poor exposed to landslides, coastal erosion Caribbean with risk of higher mortality rates and migration, Central America negative impacts on GDP where share of coastal tourism is high. The Andes Glacial melt, snow pack changes, risks of flooding, and freshwater shortages. Amazon Rainforest In high altitudes women, children, and indigenous people particularly vulnerable; and agriculture at risk. In urban areas the poor living Dry Regions on steeper slopes more exposed to flooding. The Andes Amazon Rainforest Increase in extreme heat and aridity, risk of forest fires, degradation, and biodiversity loss. Population Density [People per sqkm] Risk of rainforest turning into carbon source. Southern Cone Shifting agricultural zones may lead to 0 conflict over land. Risks of species extinction threatening traditional livelihoods and cultural 1–4 losses. 5–24 Dry Regions 25–249 Increasing drought and extreme heat events 250–999 Falkland Islands (Islas Malvinas) leading to cattle death, crop yield declines, and 1000+ A DISPUTE CONCERNING SOVEREIGNTY OVER THE ISLANDS EXISTS BETWEEN ARGENTINA WHICH CLAIMS THIS SOVEREIGNTY AND THE U.K. WHICH ADMINISTERS challenges for freshwater resources. THE ISLANDS. Risks of localized famines among remote indigenous communities, water-related health problems. Stress on resources may lead to conflict and urban migration. Southern Cone Decreasing agricultural yields and pasture productivity, northward migration of agro- ecozones. Risks for nutritious status of the local poor. Risks for food price increases and cascading impacts beyond the region due to high export share of agriculture. 75 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal which rely to some extent on glacial melt for dry season water has increased by roughly three percent per annum since the supplies and are already facing dire shortages. The arid coastal early 1990s (IFPRI 2012). In LAC, large parts of agriculture are plain of Peru faces similar challenges. Water shortage has become rain-fed and therefore very vulnerable to climatic variations such a huge risk and a source of tension in Lima, which is dependent as droughts and changing precipitation patterns. Only 10.5 mil- on water from the Andes. In Santiago de Chile, meanwhile, an lion ha of agricultural area are irrigated, amounting to roughly estimated 40 percent reduction in precipitation will impact water 0.6 percent of the total agricultural area. Of those 10.5 million ha supplies in a city that is expecting a 30 percent population growth of irrigated area, 3.5 million are located in Brazil, amounting to by 2030 (Heinrichs and Krellenberg 2011). Quito is another city 1.3 percent of Brazil’s agricultural area (FAOSTAT 2013; Oliveira that will face water shortages as a result of glacier retreat (Hardoy et al. 2009). Changing precipitation patterns and extreme events and Pandiella 2009). could therefore affect important parts of the economy. Moreover, Freshwater in coastal areas is particularly exposed to the Hoffman and Grigera (2013) calculate that for the Amazon and risks associated with sea-level rise. Here, a substantial section of Cerrado regions, shifting rainfall patterns and temperature rises due the population is exposed to repeated flooding, contamination of to climate change will lead to more frequent droughts and forest groundwater by salt water, and constraints on the availability and fires in the dry season and floods in the rainy season, threatening quality of drinking water (Magrin et al. 2007). Low-income groups the growth of monoculture agribusiness and the livelihoods of who already lack adequate access to water will be even less able small-holder farmers and ranchers. likely to obtain it unless there is a considerable improvement in Besides the implications of climate change for large-scale the provision of basic services. agriculture, there is also evidence that climate change will strongly affect small-to-medium-scale agriculture and regional 5.1.2  Climate Change Places at Risk Both Large- food security as well as indigenous communities. This is par- Scale Agricultural Production for Export and Small- ticularly true for rural communities who heavily rely on sub- Scale Agriculture for Regional Food Production sistence farming and the urban poor who are most hard-hit by Latin America and the Caribbean is a climatically highly heteroge- rising food prices. A projected 15–50 percent decline in fishery neous region. As a result, agricultural production systems and their catch potential along the Caribbean coast, and by more than outputs differ greatly among climatic zones and countries, as will 50 percent off the Amazonas estuary and the Rio de la Plata the impacts of climate change on agricultural production. Despite (Cheung et al. 2010), together with widespread coral reef loss, the relatively low contribution of agriculture (10–12 percent) to further adds to the challenge of maintaining a healthy diet for the total GDP, agriculture plays a vital role for the LAC economy, the poorest in the region. Coral reef loss and more frequent with 30–40 percent of the labor force engaged in the agricultural extreme events could also affect the viability of the tourism sector (IAASTD 2009). However, the numbers and proportion of industry, with significant implications for livelihoods across the sector comprised by subsistence and commercial agriculture, different socio-economic groups. differ greatly among the LAC countries. Overall undernourishment in the region has decreased. In Climate change is expected to have different impacts over 1990, roughly 65 million people (14.6 percent of the population) different timeframes (see Section 4.3, Climate Change Impacts on were undernourished; by 2012, the number decreased to 49 mil- Agriculture). In the short run, expected changes in agricultural lion people (8.3 percent of the population) (FAO 2012a). The LAC outputs in the region are likely to be heterogeneous, with some countries most affected by undernourishment are Haiti, Bolivia, regions and crops seeing gains and others losses (Table 1.15). In Guatemala, Nicaragua, and Paraguay. In all five countries more the long run, however, larger reductions in agriculture are expected than 20 percent of the population are undernourished (FAO 2012a). with important impacts on livelihoods (Calvo 2013; Sanchez and However, population growth and changing nutritional patterns Soria 2008; Samaniego 2009) despite uncertainties with regard are expected to increase global food demand by 60 percent by to the importance of CO2 fertilization and potential adaptation. 2050 (FAO 2012a). As a result, increased agricultural production Overall, the potential risks estimated for the agricultural sector is essential to maintain the currently positive trend of decreasing in LAC are substantial, particularly during the second half of the undernourishment. The expected negative effects of climate change 21st century. on agriculture (Table 1.15) will make the challenge of achieving By putting agricultural production at risk, climate change food security in LAC all the more difficult. According to one study, threatens an important regional export. LAC plays a vital role without adaptation measures climate change is likely to stall the in global agriculture (IFPRI 2012). The two biggest exporters of projected decline in child undernourishment in LAC by 5 percent agricultural products in Latin America are Brazil and Argentina by 2050 (Nelson et al. 2009). This study does not take into account (Chaherli and Nash 2013). Agricultural production in LAC countries the impacts on food resources other than agricultural crops; it 76 Lati n Ame r i ca and the Caribbean therefore may underestimate the multidimensional impacts that urban populations, including social marginalization and limited climate change has on food security. access to resources. However, conditions such as dense and poorly There are potentially direct consequences of climate change on constructed housing in urban areas or high, direct dependence the levels of poverty and food security in the region. According to on ecosystem services among rural indigenous populations result an exploratory modeling study by Galindo et al. (2013) an average in specific vulnerability patterns for different population groups. decline of six percent in agricultural production due to climate Despite these differences in vulnerabilities, climate impacts also change by 2025 would result in 22.6 percent and 15.7 percent act along an urban-rural continuum. For example cities depend fewer people overcoming the $1.25 and $2 per day poverty lines on the surrounding landscape to provide ecosystem services and respectively, given losses in livelihoods. This means a total of the rural population benefits from remittances sent from urban to 6.7–8.6 million people who would remain under the poverty line rural areas. However, the effectiveness of remittances in support- as a result of climate change impacts on agriculture. In addition, ing adaptive capacity under rising impacts is open to question, as important indirect effects resulting from reductions in agricultural both demand on the receiving end and exposure to climate risks yields include risks to agro-industrial supply chains. Given the on the sending end are expected to rise. exploratory nature of this study, however, exact numbers have to The rural poor are likely to feel the impacts of climate change be interpreted with care. and variability most directly given their dependency on rain-fed agriculture and other environmental resources (e.g., forests and 5.1.3  A Stronger Prevalence of Extreme Events fish) which are particularly susceptible to the effects of climate Affects Both Rural and Urban Communities, change in general and extremes in particular. Moreover, these Particularly in Coastal Regions populations have limited political voice and are less able to A changing frequency and intensity of extreme events, such as leverage government support to help curb the effects of climate drought, heat extremes, tropical cyclones, and heavy precipitation, change (Prato and Longo 2012; Hardoy & Pandiella, 2009). Rural will have strong implications for the urban and rural populations poverty in the LAC region has declined considerably over the past of the region, with particular vulnerability patterns shaping the two decades—both in terms of the numbers of people who live in risks of different population groups. poverty and the rate of poverty among rural populations—with The LAC region is heavily exposed to the effects of strong many countries in the region showing positive trends both in ENSO events, including extreme precipitation and disastrous poverty reduction and in a better distribution of income. That flooding, especially in the Andes and Central America where said, many rural people in the region continue to live on less than steep terrains are common (IPCC 2012; Mata et al. 2001; Mimura $2 per day and have poor access to financial services, markets, et al. 2007; Poveda et al. 2001). Glacial lake outbursts present training, and other opportunities. There is a strong concentra- a further permanent hazard for Andean cities (Chevallier et al. tion of extreme poverty among landless farmers and indigenous 2011). Along the Caribbean and Central American coasts, tropical peoples, particularly among women and children; indeed, close cyclones and rising sea levels expose the population to storm to 60 percent of the population in extreme poverty live in rural surges and coastal inundation (Dilley et al. 2005; Woodruff et areas (RIMISP 2011). al. 2013). Although the scientific evidence is limited, there are Extreme events will also strongly impact the urban poor as studies indicating an increase of 80 percent in the frequency urban areas are also a focal point of climate change impacts from of the strongest category 4 and 5 Atlantic Tropical Cyclones extreme events (Vörösmarty et al. 2013). In 2010, the urban popu- (Bender et al. 2010; Knutson et al. 2013) and a doubling in the lation accounted for 78.8 percent of the total population (ECLAC frequency of extreme El Niño events above 20th century levels 2014). National economies, employment patterns, and government (Cai et al. 2014). The latter two projections are especially wor- capacities—many of which are highly centralized—are also very risome as they concur with a sea-level rise up to 110 cm (see dependent on large cities; this makes them extremely vulnerable Section 3.7, Regional Sea-level Rise). A poleward migration to the effects of extreme events (Hardoy and Pandiella 2009). of tropical cyclones as recently observed (Kossin et al. 2014) Urbanization in the region includes unplanned, haphazard could potentially lead to less damage to tropical coasts but expansion of cities (UN Habitat 2012) over floodplains, mountain countries would also benefit less from the water replenishment slopes, or areas prone to flooding or affected by seasonal storms, that cyclone rainfall brings and areas currently less exposed to sea surges, and other weather-related risks (Hardoy and Pandiella tropical cyclones would face additional risks. 2009). Houses in informal settlements are frequently built with The vulnerability of people exposed to extreme events is inadequate materials, which make them damp and cold in the shaped by a multitude of non-climatic factors. Some socioeco- winter and very hot in the summer (Hardoy and Pandiella 2009). nomic factors shaping vulnerability are the same for rural and Hence, there are concentrations of low-income households at high 77 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal risk from extreme weather (Hardoy et al. 2001). For example, an protection provided by healthy coral reefs. Altogether these impacts estimated 1.1 million people live in the favelas of Rio de Janeiro that may augment impacts on coastal infrastructure (including beach sprawl over the slopes of the Tijuca mountain range, making them erosion), thus threatening transport, settlements, and tourism. particularly at risk from mudslides (Hardoy and Pandiella 2009). In combination with an up to 50 percent decrease in fish catch Moreover, most low-income people live in housing without air potential under a 4°C world (Cheung et al. 2010), damage to coral conditioning or adequate insulation; during heat waves, the very reefs threatens artisanal fisheries that support local livelihoods. young, pregnant women, the elderly, and people in poor health Infrastructure in the Caribbean is already highly vulnerable are particularly at risk (Bartlett 2008) (see also Section 4.7, Human to natural hazards, and important assets (including airports) are Health). In northern Mexico, heat waves have been correlated often low-lying. Further climate change impacts may affect the with increases in mortality rates; in Buenos Aires, 10 percent of condition of infrastructure, increasing failures and maintenance summer deaths are associated with heat strain; in Peru, records costs. The high vulnerability to hurricanes and tropical storms in show a correlation between excessive heat and increases in the low-lying states could further increase with a growing population, incidence of diarrhea (Mata and Nobre 2006). Such effects may putting growing numbers of assets at risk, exacerbating pervasive be compounded by a climate-change-related increase in the broad poverty/inequality and potentially leading to displacement of a geographic areas and microclimates in which certain vector-borne greater proportion of the population, as was observed in the wake diseases, such as malaria and dengue fever, can flourish (Costello of Hurricane Mitch (Glantz and Jamieson 2000; McLeman and et al. 2009). Hunter 2011). Ultimately, due to the small size of many Caribbean Adverse socioeconomic conditions in concert with exposure to islands, more frequent natural disasters may cause severe setbacks climate change impacts undermine the development of adaptive to the overall economy. However, not only coastal areas are at high capacity. People living in informal urban settlements without legal risk. In Central America and the Caribbean, the poor are often tenure rights—who are generally from poor and socially excluded living on steep slopes or close to rivers; they are therefore espe- communities (including marginalized ethnic groups)—in principle cially exposed to landslides and floods. Such hydro-meteorological have limited means or incentive to attempt to climate-proof their events may damage the poor quality (often informal) residential houses (Moser et al. 2010). A lack of accountability to the citizens structures of vulnerable communities, which could in turn lead to and a very limited scope for public participation in decision making higher mortality rates and population displacement. More generally, means that poorer areas are deprioritised for infrastructural upgrad- intensified non-climate stressors related to land use change and ing, and thus frequently have inadequate infrastructure (e.g., storm ecosystem degradation could impair the resilience and the ability drains) to cope with extreme events (Hardoy and Pandiella 2009). to cope with the impacts of extreme hydro-meteorological events. Furthermore, the impacts of extreme weather events are often more Climate extremes (e.g., drought, heat waves) in combination severe in areas that have been previously affected or have not yet with long-term decreases in precipitation may reduce crop yields been fully recovered from previous a previous extreme event; these in Central America and Caribbean countries and affect food secu- cumulative effects are difficult to overcome (Hardoy and Pandiella rity and market prices. This is particularly relevant in the case of 2009; Hardoy and Romero Lankao 2011). Damage to housing as a coffee crops, which are important for the livelihoods of workers result of extreme weather can lead to loss of key assets used in urban and small farmers in Central America. informal sector businesses (Moser et al. 2010), further undermining There are several studies projecting reduced runoff and ground- the buildup of resilience and increasing the risk of poverty traps. water recharge in a 4°C world (Table 1.15), which will reduce water availability. This may disproportionately affect the lives of 5.2  Sub-regional Development Narratives women responsible for managing household water resources, as well as the health and wellbeing of vulnerable members of poor 5.2.1  Central America and the Caribbean—Extreme households (e.g., infants, the chronically ill, and the elderly). Events as a Threat to Livelihoods Ultimately, water stress could increase conflicts over land, affect In a 4°C world, the Central American and Caribbean countries food security, and provoke climate-induced migration. Additionally, are projected to be at risk from higher ENSO and tropical cyclone Central America is heavily dependent on hydropower to generate frequency, drought and heat extremes, and precipitation extremes electricity; it is expected that energy security could become an issue. (Table 1.15). The impacts of tropical cyclones will be exacerbated by rising sea levels fostering storm surges. Moreover, by the year 5.2.2  The Andes—Changing Water Resources 2040, Caribbean coral reefs are expected to experience annual Challenge the Rural and Urban Poor bleaching events due to sea-level rise, ocean warming, and Climate change already affects and will further affect water resources sedimentation from flood events, which will diminish the coastal in the Andes (Table 1.15). These resources are already scarce as 78 Lati n Ame r i ca and the Caribbean a result of insufficient management and degradation of critical and forest-fringe communities would be at particular risk, which ecosystems (including the páramo and cloud forests). Increas- could, in turn, spur additional migration of affected groups to ing temperatures leading to higher evapotranspiration, changing cities, and could also open up forest regions to settlers, spurring precipitation patterns, and more extreme precipitation events all additional deforestation. directly affect river runoff. Moreover, glacial melt and snow pack Besides the risk of a tipping point, climate change is expected changes are important components of the regional hydrological to contribute to forest degradation and biodiversity loss. There balance. Increased glacial melt may increase water availability are variable rainfall patterns and significant differences in rain- in the next few decades while reducing it thereafter. Both glacial fall projections between the northern and southern zones of the melt and snowmelt affect the amount and seasonality of water Amazon. In a 4°C world, in the southern zone, winter annual flows. These changes threaten the water supply for hydropower, precipitation is likely to decrease while evapotranspiration and agriculture, and domestic use. This is particularly relevant because aridity is expected to increase (see Section 3.3). This puts the many large cities and populations are located at high altitudes or southern part of the forest at increased fire risk. in arid regions in the lowlands where alternative sources of water Increasing fires will not only lead to large emissions of CO2 are not abundant and where the urban poor are already suffering but, in combination with deforestation, declining rainfall, and from limited access to water. Moreover, the regional energy mix forest drying, may also draw the agricultural frontier northwards. strongly depends on hydropower; higher risks of power outages This threatens the livelihoods of forest-dependent communities may impact household and community welfare. In addition, and could lead to land-use conflicts between existing communi- subsistence farming and cattle herding in the highlands, as well ties and newly arriving farmers. In addition, timber harvest from as large-scale agriculture in the coastal areas, depend on water concessions could be negatively affected. Moreover, increasing coming from the mountains. While a decreasing water supply is fires in the Amazon threaten rural and urban settlements and the an important risk to food security and poverty levels in general, resulting smoke/haze could aggravate respiratory disease for both women and children who are often in charge of agriculture in high- forest dwellers and urban residents in central Brazil. altitude communities are at particular risk of increasing poverty Negative effects of climate change on biodiversity resulting from and water-scarcity-related conflicts. The same applies to indigenous habitat contractions and extinctions are very likely in a warmer people whose traditional water management systems are likely to than 2°C world. In combination with increasing forest degrada- be affected and whose livelihoods are already threatened. Other tion, changes in the range of certain species will affect resource water-dependent activities, such as large-scale or artisanal mining, availability for indigenous populations that are very reliant on may be affected as well. These stresses could exacerbate the cur- native plants and animals. This could increase malnutrition among rent urbanization trend, leading to further rural-urban migration children and the elderly and undermine traditional knowledge of and amplifying the risks to the urban poor. ecosystems, impacting the community social structure and the Besides these more gradual changes, extreme hydrological value placed on traditional knowledge. Altogether, these changes events (such as an intensification of ENSO, extreme precipitation, may push local communities to expand subsistence agriculture high flows, and glacial lake outburst floods) increase the risk for as an alternative livelihood strategy or to migrate to other forest natural disasters, erosion, and landslides. While such events may areas, thereby amplifying forest degradation and threatening generally decrease GDP, the impact across different layers of the existing protected areas. population is uneven, with the urban poor living on steep slopes typically at the highest risk. 5.2.4  Southern Cone—Risks to Export Commodities from Intensive Agriculture 5.2.3  The Amazon—Risk of Tipping Point, Forest The Southern Cone countries are currently a major grain and Degradation, and Biodiversity Loss Threatens Local livestock producing region for local and global markets (Chaherli Communities and Nash 2013). The region has experienced significant climate Despite an improved understanding of processes linking climate, shocks, mainly related to ENSO, which have resulted in floods vegetation, land-use change, and fire in the Amazon, the identifica- or droughts at critical phases of the crop cycle. Moreover, despite tion of the processes and the quantification of thresholds at which small and uncertain increases in precipitation, the region faces an irreversible approach toward a tipping point is triggered (i.e., increasing evapotranspiration rates under a 4°C world, (see Sec- a potential transition from forest to savannah) is still incomplete. tion 3.3). This highlights the high risks to agricultural production Overall the most recent studies suggest that forest dieback is an from climate change in a 4°C world; this is particularly true for unlikely, but possible, future for the Amazon region (Good et al. rain-fed agriculture, which is prevalent in more than 98 percent 2013). Should such die-back occur, the livelihoods of forest-dwelling of Brazil’s agricultural area (FAOSTAT 2013; Oliveira et al. 2009). 79 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal The results of agricultural modeling studies differ in the severity “drought polygon,” an area characterized by a semi-arid climate of the climate change impact, but most agree that climate change that suffers from recurrent droughts (Krol and Bronstert 2007). will very likely decrease agricultural yields for important food Parts of this region in Brazil have been identified as having socio- crops in Latin America in the absence of adaptation measures and climatic hotspots, given the naturally limited water availability, persistent CO2 fertilization (ECLAC 2010; Fernandes et al. 2012; a relatively low human development index, a high population Nelson, Rosegrant, Koo et al. 2010) (see also Section 4.3, Climate density (Torres and Lapola et al. 2012), and existing conflicts over Change Impacts on Agriculture). Moreover, while CO2 fertiliza- water (Araújo and Knight 2005; Krol et al. 2006). Especially in a tion may increases yields, there is some evidence of it decreasing 4°C world, dry regions in Mexico and Brazil face strong increases protein contents in major grains (Müller et al. 2014; Myers et al. in highly unusual heat extremes and aridity leading to more 2014). For sugarcane, there might be beneficial effects with yield intense and longer drought events (Table 1.15). Northeast Brazil increases (Table 1.15). Moreover, agro-ecozones in Brazil, includ- is particularly impacted by ENSO-related droughts; these may ing major grain belts, may move northward (to central Brazil) to become more frequent in a 4°C world. In dryland Brazil, urban already cleared lands in the Cerrado region; Assad et al. (2013) migration to rapidly growing coastal cities in the northeastern project displacement of poorly productive and degraded pasture states is highly likely to be the result of the loss of agricultural with intensive multicrop grain cropping and intensified pastures. income (Mendelsohn 2007). The impacts on agriculture and livestock (Table 1.15) may In these dry regions, increasing drought events may lead to lead to increasing food prices that could entail trade impacts and problems for urban water supply or widespread cattle deaths. In stresses on other regions’ food production systems and may alter addition, small-hold family farmers in rural areas may experience dietary patterns (especially of the poor). Alongside price risks, lower productivity or even lose entire harvests, threatening their reduced nutrient contents (especially protein) could also raise the livelihoods. A decline in agricultural productivity may cause local- risk of malnutrition in children. Opportunities may arise from the ized famines, especially among remote indigenous communities reshuffling of agricultural zones as plantation forestry, horticultural (particularly in Northern Mexico), and possibly result in long-term crops, and biofuel production from sugarcane may be able to impacts on the household nutritional status. expand on lands that become unsuitable for grain crops. Reduced Increases in irrigated agriculture, if not well integrated with crop and livestock productivity can be moderated via adaptation long-term water resource planning and management, pose another measures and climate-smart technologies (e.g., improved varieties risk as they will exacerbate issues of water availability and also and breeds, irrigation, conservation agriculture, liming, and fertil- concentrate wealth. A diminished drinking water supply in rural izers to enhance crop rooting depth). These intensification and communities may also lead to an increasing reliance on water trucks climate-smart innovations would, however, require a significant that occasionally deliver contaminated water (resulting in illness upgrading of knowledge and extensive field testing. and death). Moreover, the need to search for drinking water and the associated health problems associated with low-quality drinking 5.2.5  Dry Regions (Mexican Dry Subtropics and water could decrease the work force and income in rural areas, North Eastern Brazil)—Increasing Drought Stress leading to increased crime, social exclusion, and other problems Threatening Rural Livelihoods and Health related to rural-urban migration during drought events. Increasing There has been significant development progress in these regions water stress may also lead to further over-exploitation of aquifers in recent decades, which has lifted a number of communities in the Northern part of Mexico. This in turn would lead to the out of extreme poverty. The possibility of increasing droughts, release of groundwater minerals, affecting groundwater quality, however, threatens to force many of these populations back into and, in coastal aquifers, lead to sea water intrusion. In general, extreme poverty. hydropower and energy systems will be stressed across these dry The Central and Northern arid areas in Mexico and the semi- regions. Direct damages from droughts and secondary impacts arid areas in Mexico and Northeast Brazil are already under water on the agriculture sector and related labor markets may result in stress and are sensitive to inter-annual climate variability. For negative GDP growth rates in the agriculture sector. example, parts of northeast Brazil are situated within the so-called 80 6  Synthesis Table—Latin America and the Caribbean Table 1.15: Synthesis table of climate change impacts in LAC under different warming levels. Observed Around 4°C Vulnerability Around 1°C Around 1.5°C Around 2.0°C Around 3.0°C and above Risk/Impact or Change (≈2010s1) (≈2030s) (≈2040s) (≈2060s) (≈2080s) Heat Highly Unusual Absent Around 10% of Up to 30% of land area 30–40% of land area Around 65% of land area Around 90% of land area Extremes Heat Extremes land area affected affected in DJF affected in DJF affected in DJF affected in DJF in DJF Unprecedented Absent Absent Around 5% of land area Around 15% of land area Around 40% of land area Around 70% of land area Heat Extremes affected in DJF affected in DJF affected in DJF affected in DJF Regional Warming (austral 0.8°C 1.5°C, warming limited 5.5°C, warming limited summer temperatures) along the Atlantic coast along the Atlantic coast of Brazil, Uruguay, and of Brazil, Uruguay, and Argentina with about Argentina with about 0.5–1.5°C. The central 2–4°C. The central South-American region South-American region of Paraguay, northern of Paraguay, northern Argentina, and southern Argentina, and southern Bolivia with more Bolivia with more pronounced warming, pronounced warming, up up to 2.5°C* to 6°C* Precipitation Relatively small changes Peru, Ecuador, and and disagreement among Colombia on the Pacific climate models. Peru, coast increase in annual Ecuador, and Colombia mean precipitation of about on the Pacific coast with 30%, most pronounced a small increase in annual during the summer. The mean precipitation of up Caribbean, Patagonia to 10%. Reduction in (southern Argentina winter precipitation over and Chile), Mexico, and southeastern Amazon central Brazil become drier rainforest* (10–40%). Central America becomes drier in winter (up to 60%). The annual mean precipitation in southeastern Amazon rainforest is projected to drop by 20% mostly because of a strong decrease in winter precipitation (–50%)* 81 82 Table 1.15: Continued. Observed Around 4°C Vulnerability Around 1°C Around 1.5°C Around 2.0°C Around 3.0°C and above Risk/Impact or Change (≈2010s1) (≈2030s) (≈2040s) (≈2060s) (≈2080s) Extreme Precipitation Robust increase in Annual extreme daily Annual extreme daily Annual extreme daily intensity of extreme precipitation with 20–year precipitation with 20–year precipitation with 20–year precipitation events for return interval increases return interval increases return interval increases South America2 by 7% and the 20–year by 11% and the 20–year by 25% and the 20–year return value of maximum return value of maximum return value of maximum precipitation returns precipitation returns every precipitation returns every every 15 years*3 12 years*3 6.5 years. Important 5%, 7%, and 3% 9%, 7%, and 8% increase in areas are the Caribbean, increase in maximum maximum 5–day precipitation Meso-America, Southern 5–day precipitation in in the Amazon, Central Argentina, and Chile as well the Amazon, Central America, and Southern as parts of Brazil and the America, and Southern South America respectively*4 Pacific coastline of Ecuador, South America Peru, and Colombia*3 respectively*4 16%, 8%, and 12% increase in maximum 5-day precipitation in the Amazon, Central America, and Southern South America respectively*4 Drought Severe droughts in 2005 4-, 1- and 2-days 1%, 4% and 9% increase 8-, 2- and 2-days longer 17-, 10- and 8-days and 2010 in the Amazon5 longer droughts in the in days under drought droughts in the Amazon, longer droughts in the Increase in drought Amazon, Central America conditions in Caribbean, Central America and Amazon, Central America, conditions in Central and Caribbean, and Meso-America, Caribbean, and Southern and Caribbean and America6 Southern South America and South America South America respectively*4 Southern South America respectively*4 respectively*7 11.5%, 12%, and 12.5% respectively*4 increase in days under 22%, 25%, and 22% drought conditions in increase in days under Caribbean, Meso-America, drought conditions in and South America Caribbean, Meso-America, respectively*7 and South America Reduction by 5 to 9% in respectively*7 annual soil moisture content in Amazon and Central America*8 Increase in extreme droughts in the Amazon, Brazil, Central America, northern Mexico, and Southern Chile*8 Observed Around 4°C Vulnerability Around 1°C Around 1.5°C Around 2.0°C Around 3.0°C and above Risk/Impact or Change (≈2010s1) (≈2030s) (≈2040s) (≈2060s) (≈2080s) Aridity 33% of land area hyper- 36% of land area hyper- 41% of land area hyper- arid, arid, or semi-arid arid, arid, or semi-arid arid, arid, or semi-arid (increase of about 10%)* (increase of about 25%)* Sea-level Rise Above Median estimate across Median estimate across the Present (1985–2005) the region 0.27–0.39 m, region 0.46–0.66 m, with with highest sea-level highest sea-level rise on the rise on the Atlantic Coast Atlantic Coast and lowest and lowest on the tip of on the tip of the American the American continent. continent. Maximum Maximum 0.65 m sea- 1.14 m sea-level-rise and level rise in Recife* 1.4 m in Rio de Janeiro and Barranquilla on the Atlantic Coast by 2100* El Niño Southern ENSO has never been as Doubling of frequency of Oscillation (ENSO) variable as during the last extreme El Niño events*10 few decades9 Tropical Cyclones Tropical cyclone Power Dissipation Power Dissipation Index frequency increase in the Index increasing by increasing by 125–275%*12 North Atlantic11 100–150%*12 Increase of 80% in the Increase of 40% in frequency of the strongest the frequency of the category 4 and 5 Atlantic strongest Atlantic Tropical Tropical Cyclones*14 Cyclones*13 Glaciers Southern Up to 22% loss of glacial 21–52% loss of glacial 27–59% loss of glacial 44–72% loss of glacier Andes volume15,16 volume*15 volume*15, 16, 20 volume*15, 20 Reduction in glacier length by 3.6–36% (Northern Patagonian Ice Field), 0.4–27% (Southern Patagonian Ice Field), and 2.5–38% (Cordillera Darwin Ice Field)17 31.7% loss of glacial area15 23–26.6 Gt/yr glacial mass loss rate over Patagonian Ice Fields18 1.88 Gt/yr annual calving loss in Northern Patagonian Ice Field19 83 84 Table 1.15: Continued. Observed Around 4°C Vulnerability Around 1°C Around 1.5°C Around 2.0°C Around 3.0°C and above Risk/Impact or Change (≈2010s1) (≈2030s) (≈2040s) (≈2060s) (≈2080s) Tropical Up to 90% loss of glacial 78–94% loss of glacial 66–97% loss of glacial 91–100% loss of glacier Glaciers volume15, 16 volume*15 volume*15, 16, 20 volume*15, 20 79% loss of glacial area15, 87% in Andes of Venezuela over 1952–2003, 11% in Andes of Colombia over 1950–1990s, 57% in Chimborazo over 1962– 1997, 37% in Cotopaxi and 33% in Artinsana over 1979–2007, and 20–35% in Peruvian Andes over 1960–2000s21 6 Gt/yr glacial mass loss rates22 Water Central Up to 10% less runoff23 13% decrease in total 10–30% decrease of mean 24% decrease in total America & annual reservoir inflow in annual runoff*25 annual reservoir inflow in Caribbean Rio Lempa*24 5–20% decrease in river Rio Lempa*24 Around up to 15–45% runoff23 10% in groundwater reduction of annual 20% decrease in discharge recharge*27 discharge39 from Rio Grande*26 15–45% reduction of annual discharge39 Andes Discharge in Cordillera Decreasing mean Wet season discharge Wet season discharge in 21% streamflow decrease Blanca decreasing annual runoff in the Llanganuco the Llanganuco catchment In the Limarí basin and annually and during the for northeastern catchment increases increases from 10–26% increasing winter flows dry season28 Chile29 from 10–26% and and dry season discharge (28.8–108.4%), decreasing dry season discharge decreases from 11–23%30 summer flows (–16.5 to decreases from –57.8%), and earlier center 11–23%30 timing of mass of annual Reduced groundwater flows for different sub- recharge for the central basins of the Limarí basin32 Andes region31 Likely increase in flood frequency33 Observed Around 4°C Vulnerability Around 1°C Around 1.5°C Around 2.0°C Around 3.0°C and above Risk/Impact or Change (≈2010s1) (≈2030s) (≈2040s) (≈2060s) (≈2080s) Water Amazon Decreasing mean annual Low-flows become more Total annual runoff decreases Low flows increase discharge and monthly pronounced over several in the southern half of the between 10–30% in minimum discharge Amazonian sub-basins35 Amazon River36 the western part of the for Tapajós in the Median high-flows Duration of inundation Amazon35 southeastern Amazon, increase by 5–25% in 0.5–1 month shorter in Low flows and high flows the Peruvian Amazon the western part of the eastern Amazonia*37 would increase each by Rivers, and the upstream Amazon basin35 Inundation area will increase 5% at Óbidos35 Madeira34 Median low-flows with a 2–3 month longer decrease significantly, by inundation time in the 20% for the Japura and western part of the Amazon Negro river and 55% at basin*37 the Río Branco35 Northeast Seasonality of river Strong decreases and Brazil discharge remains stable increases in mean but mean river discharge groundwater discharge decreases38 depending on the GCM27 No clear signal in relative change of annual discharge39 Rio de la Plata Increase in river runoff of Mean river flow from Increase in mean relative Increase in frequency and 10–30%40 –20% to +18% for the runoff for the Rio de la Plata duration of fluvial floods in Río Grande, a tributary of region of 20–50%23 the Uruguay and Paraná the Paraná41 basin42 Decrease in the 20th century 100–year return period for floods for the Parana43 Southernmost Decrease in mean relative Decrease in mean relative 15–45% reduction of South America runoff up to 10%23 runoff by 10–30%23, 39 annual discharge39 Crop Wheat Brazil: –23%44 Brazil: up to –50%44 Brazil: –41% to –52%44 Argentina: –16%48 yield Central America and Central America and Central America and Caribbean: –43%44 Caribbean: –56%44 Caribbean: –58% to –67%44 LAC: 6.5–12%45 and LAC: 0.9–12%45 and –5.5 0.3–2.3%46 to 4%46 Chile*: up to –10%45,47 Argentina: –11%48 85 86 Table 1.15: Continued. Observed Around 4°C Vulnerability Around 1°C Around 1.5°C Around 2.0°C Around 3.0°C and above Risk/Impact or Change (≈2010s1) (≈2030s) (≈2040s) (≈2060s) (≈2080s) Crop Maize Panama: up to México: –29%44 Ecuador and Brazil up to Brazil: –30 to –45%50 Argentina: –24%48 yield –0.5%45,49 Panamá: 0.8%45,49 –64%44 Panama: 4.5%45,49 Ecuador: –54%48 México: up to –45%44 Panama: 1.5–2.4%45,49 LAC: –2.3 to +2.2%45 and –0.4 to –2.8%46 Brazil: –15 to –30%50 Soybean Brazil: –45%44 Brazil: up to –70%44 Brazil: –66% to –80%44 Argentina: –25%48 Brazilian Amazon: LAC: 18–19%45 and –2.5 Brazilian Amazon: –44%51 –1.8%45,51 to 4%46 LAC: 19.1–19.5%45 and Argentina: –14%48 –1.2 to 2.3%46 Rice Central America and Central America and Central America and Ecuador: 37%48 Caribbean: +3%44 Caribbean: –4%44 Caribbean: +1.5–4%44 LAC: –1.2 to +13%45 and LAC: 6.7–745 and –0.8 to –6.4 to 5%46 –1.8%46 Beans Brazil: –15 to –30%50 Brazil: –30 to –45%50 Ecuador: –9%48 Coffee Ecuador: –23%48 Cocoa Ecuador: –21%48 Bananas Ecuador: –41%48 Sugarcane Southern Brazil: 15%45,52 Southern Brazil: 59%45,52 Ecuador: –36%48 Livestock Livestock choice in Livestock choice in 7 to 16% decrease in 22 to 27% decrease in beef Argentina, Brazil, Argentina, Brazil, Chile, beef cattle numbers in cattle numbers in Paraguay48 Chile, Colombia, Colombia, Ecuador, Paraguay48 Ecuador, Uruguay, Uruguay, and Venezuela: Livestock choice in and Venezuela: Beef Cattle: –1.6 to 5% Argentina, Brazil, Chile, Beef Cattle: –12.5 Dairy cattle: –6.7 to 2.5% Colombia, Ecuador, to 5.7% Pigs: –0.8 to 0.0% Uruguay, and Venezuela: Dairy cattle: –6.6 Sheep: 0.0 to 7.0% Beef Cattle: –11.0 to to 1.2% Chicken: –1.0 to 1.3%53 0.3% Pigs: –1.6 to 0.2% Dairy cattle: –10 to 5% Sheep: –5 to Pigs: –0.9 to 0.1% 20.1% Sheep: 0.0 to 19% Chicken: –2.9 to Chicken: –1.5 to –0.3%53 1.4%53 Observed Around 4°C Vulnerability Around 1°C Around 1.5°C Around 2.0°C Around 3.0°C and above Risk/Impact or Change (≈2010s1) (≈2030s) (≈2040s) (≈2060s) (≈2080s) Biodiversity Extinction rates of Extinction rates of 68% loss of suitable area for In most LAC ecoregions, species: 2–5% for species: 2–8% for cloud forest and extinction of amphibian species mammals, 2–4% for mammal, 3–5% for 9 of 37 vertebrate species in experience at least 30% birds, 1–7% for butterfly birds, 3–7% for butterfly Mexico63 turnover; in western South species in Mexico, and species in Mexico, and 78% reduction in geographic America and Central 38–66% for plant species 48–75% of plant species distribution of 110 Brazilian America at least 50%*57 in Cerrado54 in Cerrado54 Cerrado plant species64 Up to 21 out of 26 Changes in amphibian biogeographic ecoregions species ranges in in South America faces the Atlantic Forest severe ecosystem Biodiversity Hotspot55 change*58 Marsupial species ranges Loss of habitat between declining in Brazil56 11.6% to 98.7% and 85–95% of LAC change in species richness amphibian species face between –25% to –100% net loss in range size*57 for plants in Amazon61 1 out of 26 biogeographic ecoregions in South America faces severe ecosystem change*58 Climatically suitable areas for cloud forest reduced by 54–76%59 44 of 51 birds species lose distribution area in Brazilian Atlantic forest60 Loss of habitat between 8.2% to 81.5% and change in species richness between –4.1% to –89.8% for plants in Amazon61 Majority of 430 amphibian species would face range contractions accompanied by an overall species loss in the Atlantic Forest Biodiversity Hotspot62 87 88 Table 1.15: Continued. Observed Around 4°C Vulnerability Around 1°C Around 1.5°C Around 2.0°C Around 3.0°C and above Risk/Impact or Change (≈2010s1) (≈2030s) (≈2040s) (≈2060s) (≈2080s) Amazon Dieback Model agreement on Carbon loss (kg C/m2) by 10–80% forest cover above-ground live –1.8 to –0.6 in Eastern loss*68 biomass loss: 14.3% Amazonia, –1.2 to 0.6 in –35% to +40% change for climate change Northwestern Amazonia and of carbon without only but increasing to –3.3 to –2.6 in Southern deforestation and 43.1–58.6% with different Amazonia*66 –55% to –5% with 50% deforestation scenarios65 Carbon increase (kg C m2) by deforestation*69 5.5–6.4 in Eastern Amazonia, 10–80% forest cover loss*70 2.9–5.5 in Northwestern Carbon losses: 70 GtC Amazonia and 2.1–4.3 in (Vegetation carbon), Southern Amazonia*45,56 150 GtC (Soil carbon)*71 Decrease of LAI by 12.6% 69% reduction in rainforest and increase in land- extent*72 atmosphere carbon flux of Model agreement on about 27.2% due to fire*67 above-ground live biomass loss: 25.5% for climate change only but increasing to 48.1–65.9% with different deforestation scenarios65 Coral Reefs Strong bleaching event 20–40% and up to 60% <60% and >60% >60% probability of annual in 2005, less severe in probability of annual probability of annual bleaching events in all 2010, in the Caribbean bleaching events in bleaching events in regions*73 Sea, Guyana, Suriname, Caribbean Sea and Caribbean Sea and French Guiana, and north Guyana, Suriname, Guyana, Suriname, Pacific Ocean73 French Guiana, and and French Guiana north Pacific Ocean respectively73 respectively73 60–80% and up to Coral cover halved from 100% probability of initial state in Virgin annual bleaching events Islands and Eastern in Caribbean Sea and Caribbean74 Guyana, Suriname, Onset of bleaching and French Guiana events starts 2046*75 respectively*73 Coral cover less than 3–5% in Virgin Islands and Eastern Caribbean74 Onset of bleaching events starts 2040*75 Observed Around 4°C Vulnerability Around 1°C Around 1.5°C Around 2.0°C Around 3.0°C and above Risk/Impact or Change (≈2010s1) (≈2030s) (≈2040s) (≈2060s) (≈2080s) Marine Fisheries Species shifts toward 35% decline in Large increase in catch higher latitudes76 phytoplankton, potential in the south zooplankton, and fish (up to 100%), strong density77 decrease in parts of the Caribbean Sea (up to 50%)76 Health 5–13% increase in Expansion of malarial 12–49 million people less 19–169 million people less relative risk of diarrheal areas mostly in Brazil80 exposed to risk of malaria exposed to risk of malaria diseases in South No net changes in for at least three months of for at least three months of America78 increased malaria length the year*82 the year*82 12–22% increase in of transmission season 1–16 million people more 5–42 million people less Dengue incidence in except in southernmost exposed to risk of malaria exposed to risk of malaria Mexico79 Brazil and Uruguay*81 for at least one month of the for at least one month of 31–33% increase in year*82 the year*82 Dengue incidence in Increased malaria length Increased malaria length Mexico79 of transmission season in of transmission season in southern Brazil, Uruguay, some highland areas of and parts of Mexico*81 southern Brazil, Uruguay, Decreased malaria length Argentina, Bolivia, Peru, of transmission season Ecuador, Colombia, and in parts of the Amazon Mexico*81 basin in Brazil, Bolivia, and Decreased malaria length Paraguay*81 of transmission season for 14–36% increase in relative tropical Latin America*81 risk of diarrheal diseases in South America*78 40% increase in Dengue fever incidence in Mexico*79 89 90 Table 1.15: Continued. Observed Around 4°C Vulnerability Around 1°C Around 1.5°C Around 2.0°C Around 3.0°C and above Risk/Impact or Change (≈2010s1) (≈2030s) (≈2040s) (≈2060s) (≈2080s) Energy 683,421 GWh/yr 688,452–861,214 GWh/ 715,173–838,587 GWh/ Decrease in firm power by maximum hydropower yr maximum hydropower yr maximum hydropower 1.58%*86 energy potential in La energy potential in La energy potential in La Plata Energy demand for 2,679 Plata River Basin83 Plata River Basin83 River Basin83 cooling degree days in Energy demand for 0,63TWH (or 0.05%) Decrease in hydropower South America84 1,802 cooling degree and 0,3TWH (or 0.03%) capacity for the two main days in South America 84 increase in electricity large reservoirs used for production in South hydroelectricity generation in America and in the El Salvador: Cerron Grande Caribbean respectively85 and 15 Setiembre*24 Decrease in hydropower Decrease in firm power by capacity for the two 3,15%*86 main large reservoirs used for hydroelectricity generation in El Salvador: Cerron Grande and 15 Setiembre*24 The impacts reported in several impact studies were classified into different warming levels (see Appendix for details) Lati n Ame r i ca and the Caribbean Endnotes 1 Years indicate the decade during which warming levels are exceeded with a 50 percent or greater change (generally at start of decade) in a business-as-usual scenario (RCP8.5 scenario) (and not in mitigation scenarios limiting warming to these levels, or below, since in that case the year of exceeding would always be 2100, or not at all). Exceedance with a likely chance (>66 percent) generally occurs in the second half of the decade cited. Impacts are given for warming levels irrespective of the timeframe (i.e., if a study gives impacts for 2°C warming in 2100 then the impact is given in the 2°C column). If a study refers to a warming level by the end of the century, this is marked with an asterisk (*). Impacts given in the observations column do not necessarily form the baseline for future impacts. Impacts for different warming levels may originate from different studies and therefore may be based on different underlying assumptions; this means that the impacts are not always fully comparable (e.g., crop yields may decrease more under a 3°C than a 4°C scenario because underlying the impact at 3° warming is a study that features very strong precipitation decreases). Moreover, this report did not systematically review observed impacts. It highlights important observed impacts for current warming but does not conduct any formal process to attribute impacts to climate change. 2 Skansi et al. (2013). 3 Kharin et al. (2013); 20–year return value of maximum precipitation refers to 1986–2005. 4 Sillmann et al. (2013b). 5 Marengo et al. (2011); Zeng et al. (2008). 6 Dai (2012). 7 Prudhomme et al, (2013) increase in days under drought conditions refers to 1976–2005. 8 Dai (2012). 9 Li et al. (2013). This is a tree-ring-based reconstruction of ENSO strength over the last 700 years, but attribution to climate change is uncertain. 10 Cai et al. (2014). 11 IPCC AR5 WGI (2013). Frequency increase in the North Atlantic over the past 20–30 years. 12 Villarini et al. (2013). The Power Dissipation Index is a combination of frequency and intensity. 13 Knutson et al. (2013). 14 Bender et al. (2010); Knutson et al. (2013). 15 Marzeion et al. (2012). Past period for glacial volume loss and area loss refers to 1901–2000. 16 Giesen and Oerlemans (2013). For past: 6.1 percent (southern) and 7.3 percent (tropical) loss of glacial volume over 1980–2011 compared to 1980. 17 Lopez et al. (2010). 18 Ivins et al. (2011); Jacob et al. (2012) refers to 2000s. 19 Schaefer et al. (2013) refers to 1990–2011. 20 Radic et al. (2013). 21 Rabatel et al. (2013). Andes of Venezuela over 1952–2003; Andes of Colombia over 1950–1990s; Chimborazo over 1962–1997; in Cotopaxi and in Artinsana over 1979–2007; and Peruvian Andes over 1960–2000s. 22 Jacob et al. (2012) for past refer to 2000s. 23 Milly et al. (2005). 24 Maurer et al. (2009). 25 Hidalgo et al. (2013). 26 Nakaegawa et al. (2013). 27 Portmann et al. (2013). 28 Baraer et al. (2012). 29 Arnell and Gosling (2013). 30 Juen et al. (2007). No differentiation possible for changes in warming for >1.5°C in 2050 and >2°C in 2080. 31 Döll (2009). 32 Vicuña et al. (2010). 33 Hirabayashi et al. 2013). 34 Espinoza Villar et al. (2009). 35 Guimberteau et al. (2013). 36 Nakaegawa et al. (2013). 37 Langerwisch et al.(2013). 38 Döll and Schmied (2012). 39 Schewe et al. (2013). 40 García and Vargas (1998); Jaime and Menéndez (2002); Menéndez and Berbery (2005); Milly et al. (2005). 41 Nóbrega et al. (2011). 42 Camilloni et al. (2013). 43 Hirabayashi et al. (2013). There was little consistency across the 11 GCMs used. 44 Fernandes et al. (2012). 45 With CO2 fertilization. 46 Nelson, Rosegrant, Koo et al. (2010). 47 Meza and Silva (2009). 48 ECLAC (2010). 49 Ruane et al. (2013). 50 Costa et al. (2009), including technological progress. 51 Lapola et al. (2011). 52 Marin et al. (2012), including technological progress. 53 Seo et al. (2010). 91 Turn Do w n Th e H e at: Conf r ont i ng t he n e w cli mate no r mal 54 Thomas et al. (2004). Mammal species (n=96), bird species (n=186), and butterfly species (n=41) in Mexico all with dispersal; plant species in Cerrado (n=163) without dispersal. The study was criticized by Harte et al. (2004) for overestimating potential extinction rates by using a common species-area exponent z for all species which may not be justified. 55 Loyola et al. (2013). 56 Loyola et al. (2012). 57 Lawler et al. (2009). 58 Gerten et al. (2013). 59 Rojas-Soto et al. (2012). 60 Souza et al. (2011). 61 Feeley et al. (2012). Large range stems from different assumption about deforestation, land-use, and adaptation and migration potentials. 62 Lemes et al. (2014). 63 Ponce-Reyes et al. (2012). 64 Simon et al. (2013). 65 Poulter et al. (2010). 66 Rammig et al. (2010). 67 Cook et al. (2012). 68 Zelazowski et al. (2011). 69 Gumpenberger et al. (2010). 70 Cox et al. (2004). 71 Betts et al. (2004). 72 Cook and Vizy (2008). 73 Meissner et al. (2012). 74 Buddemeier et al. (2011). 75 Van Hooidonk et al. (2013). 76 Cheung et al. (2010). 77 Blanchard et al. (2012). 78 Kolstad and Johansson (2011), compared to 1961–1990 levels. 79 Colon-Gonzalez et al. (2013), compared to 2000. 80 Beguin et al. (2011). 81 Caminade et al. (2014). 82 Van Lieshout et al. (2004). 83 Popescu et al. (2014). 84 Isaac and van Vuuren (2009). 85 Hamududu and Killingtveit (2013), compared to 2005 levels. 86 De Lucena et al. (2009), compared to 1971–2000. 92