Climate and southern Africa's water-energy-food nexus Declan Conway1, Emma Archer van Garderen2,3, Delphine Deryng4, Steve Dorling5, Tobias Krueger6, Willem Landman2,7, Bruce Lankford8, Karen Lebek4, Tim Osborn3, Claudia Ringler9, James Thurlow9, Tingju Zhu9, Carole Dalin1 1 Grantham Research Institute on Climate Change and the Environment, London School of Economics and Political Science, Houghton Street, London, UK; 2 Council for Scientific and Industrial Research, Natural Resources and the Environment, Pretoria, South Africa; 3 School of Geography, Archaeology and Environmental Studies, University of the Witwatersrand, Johannesburg, South Africa; 4 Climatic Research Unit, School of Environmental Sciences, University of East Anglia, Norwich Research Park, Norwich, UK; 5 Centre for Ocean and Atmospheric Sciences, School of Environmental Sciences, University of East Anglia, Norwich Research Park, Norwich, UK; 6 IRI THESys, Humboldt-Universität zu Berlin, Unter den Linden 6, 10099 Berlin Germany; 7 Department of Geography, Geoinformatics and Meteorology, University of Pretoria, Pretoria, South Africa; 8 School of International Development, University of East Anglia, Norwich Research Park, Norwich, UK; 9 Environment and Production Technology Division, International Food Policy Research Institute, 2033 K Street, NW, Washington DC Abstract In southern Africa, the connections between climate and the water-energy-food nexus are strong. Physical and socioeconomic exposure to climate is high in many areas and in crucial economic sectors. Spatial interdependence is also high, driven for example, by the regional extent of many climate anomalies and river basins and aquifers that span national boundaries. There is now strong evidence of the effects of individual climate anomalies, but associations between national rainfall and Gross Domestic Product and crop production remain relatively weak. The majority of climate models project decreases in annual precipitation for southern Africa, typically by as much as 20% by the 2080s. Impact models suggest these changes would propagate into reduced water availability and crop yields. Recognition of spatial and sectoral interdependencies should inform policies, institutions and investments for enhancing water, energy and food security. Three key political and economic instruments could be strengthened for this purpose; the Southern African Development Community, the Southern African Power Pool, and trade of agricultural products amounting to significant transfers of embedded water. Introduction Numerous challenges coalesce to make southern Africa emblematic of the connections between climate and the water-energy-food nexus which has important economic influence throughout the 1 region. Physical and socioeconomic exposure to climate is high in socioeconomically vulnerable areas and crucial sectors, such as agriculture, but also in energy generation and mining. For example, almost 100% of electricity production in the Democratic Republic of Congo (DRC), Lesotho, Malawi, and Zambia is from hydropower. Hydropower further comprises a major component of regional energy security through extensive sharing as part of the Southern African Power Pool (SAPP). The region’s population is concentrated in areas exposed to high levels of hydro-meteorological variability1 and is projected to roughly double by 20502. Of the thirteen mainland countries and Madagascar (Table 1) that comprise the Southern African Development Community (SADC), six are defined as low income, three as lower-middle income and four as upper-middle income, according to the World Bank Classification (using 2012 GNI per capita). There are few quantified examples of the linkages between climate and economic activity in the region, though South Africa experienced a decrease in GDP in the 1983 El Niño year3 and economic modelling studies in Malawi and Zambia indicate that the severe 1992 drought caused an approximately 7-9% drop in GDP and adversely affected household poverty4. Climate variability has important consequences for resource management in the region including for non- equilibrium production systems such as rangeland ecology5, irrigation6 and lakes7. Hence, southern Africa is a region where seasonal climate forecasts have potential benefit in areas where sustained forecast skill is demonstrated. Seasonal climate forecasting has been the subject of many studies in sub-Saharan Africa (SSA)8,9; and the Southern African Regional Outlook Forum (SARCOF) provides advance information about the likely character of seasonal climate. Yet, despite over a decade of research on hydrological applications of seasonal forecasts there is limited evidence of their operational use in the water sector9. With ongoing climate change, annual precipitation, soil moisture and runoff are likely to decrease, while rising temperatures could increase evaporative demand in large parts of the region10 (Figure 1). The last decade saw rapid growth in research and policy interest in natural resource scarcity, with water-energy-food interdependencies increasingly framed as a nexus, or resource trilemma. The Bonn Nexus conference in 201111 is notable in this process of recognising the complex interactions between sectors and resource systems and the need to minimise the trade-offs and risks of adverse cross- sectoral impacts11,12. The nexus is increasingly prominent on policy-makers’ agendas, partly in relation to the post-2015 development agenda for the Sustainable Development Goals13. The private sector was another early promoter of the nexus concept14 due to growing associated risks affecting production security along supply chains, such as (but not exclusively) for water15. In southern Africa, for example, South African Brewers SABMiller are seeking better approaches to handling trade-offs between water, energy and food by attempting to make business decisions through a resource nexus lens16. Strong interdependencies at a range of scales give rise to a large number of trade-offs and co-benefits, according to the heterogeneous configurations of societal uses of water across river basins and aquifers. The region’s many transboundary basins require that actions among upstream and downstream water uses are reconciled between countries. Previous nexus studies have concentrated on global interdependencies17, problem framing18 or case studies of specific systems such as islands19 and irrigation and hydropower production20. Here, we examine southern Africa’s nexus from the perspective of climate and modify Hoff’s nexus framework11 2 which integrates global trends (drivers) with fields of action, to highlight the role of climate as a driver in southern Africa’s nexus (Figure 2). Climate encompasses average (i.e. 30-year) conditions, variability over years to decades (i.e. as observed) and anthropogenic climate change. In terms of the nexus, we consider the main elements of intra-regional linkages in water-energy-food at a national level, while highlighting connections at the river basin scale and drawing attention to case studies of the many examples of specific trade-offs and synergies21. We base our review on published studies, complemented by empirical analysis of available national-level data on climate, water resources, crop production, trade and GDP. We first consider national-level exposure of water, energy and food production to climate variability in aggregate economic terms and analyse the relationship between inter-annual and multi-year climate variability and economic activity, focusing on GDP and agricultural production. We then outline the potential for seasonal climate forecasting in areas with high forecasting skill and socially and economically important nexus related activities, and summarise studies that model the impact of anthropogenic climate change on elements of the nexus. Finally, we describe three key intra-regional mechanisms for balancing nexus components and conclude by identifying knowledge gaps in southern Africa’s climate and water-energy-food nexus. National level exposure of nexus sectors to climate We characterise exposure as the interaction between characteristics of the climate system (particularly inter-annual rainfall variability) and a country’s dependence on climate-sensitive economic activities such as the share of agriculture in GDP, the proportion of rain-fed agricultural land and the energy contribution from hydroelectric sources (Table 1, Figure 3). South Africa’s GDP is larger than that of the other 12 southern African economies combined. The direct contribution of agriculture to the economy is lowest (<10%) in South Africa, Botswana, Swaziland, Namibia, Angola and Lesotho, 13% in Zimbabwe, and over 20% in the other countries. If agricultural processing were included in agricultural GDP, the shares would be substantially larger in most, if not all, SADC countries. The share of cropland equipped for irrigation is low in most of the region, with the exception of Madagascar, South Africa and Swaziland (Table 1). The contribution of hydropower to energy production is very high overall (Figure 3), but varies considerably across the region, from 1.5% in South Africa, to over 30% in Madagascar, Swaziland and Zimbabwe, and to almost 100% in DRC, Lesotho, Malawi, and Zambia. Reliable electricity production is at risk during prolonged droughts, and also during extreme flood events, when dam safety is an additional risk. Over 90% of South Africa's energy generation is coal-based22, well above the rest of the region. Coal-fired power plants with wet cooling systems consume far more water than most other energy technologies22. Thus, South Africa’s main energy utility Eskom uses about 2% of the country’s freshwater resources, mainly for coal-fired power stations23. Coal mining and energy generation from coal both substantially impact water quality and availability24. To reduce these impacts, Eskom has implemented a dry-cooling system in two existing and all new power stations25, enabling a 15-fold reduction in water use. Overall, there are strong contrasts (Table 1) in energy (8-84% of energy consumption imported) and food (5-90% of cereal food imported) self-sufficiency, and in the intensity of freshwater use, expressed 3 as freshwater withdrawals relative to total actual renewable water resources (TARWR) (0.1-24%). Countries facing most water shortage, expressed as share of TARWR withdrawn (Table 1), are South Africa (24%), Swaziland (23%), and Zimbabwe (21%), well within categories defined as physically water-scarce (ratio larger than 20%26). We interpret this indicator with caution, noting its failure to capture the complex spatial and temporal distribution of water, political-economic access, differences in water needs and socioeconomic capacity to support effective water utilisation27,28. Sub-national areas of high demand relative to availability include southern Malawi, Namibia and Botswana. Low ratios of water withdrawal to TARWR (such as 0.05% in DRC28) could also indicate economic water scarcity due to inadequate investments to harness and deliver water. The cereal import dependency ratio (Table 1) reflects the importance of imports in the volume of grains available for consumption in the country (i.e. Production + Imports - Exports). It is particularly high for the small countries of Swaziland and Lesotho, and more strikingly so for larger nations like Botswana (90%), Namibia (65%) and Angola (55%). Dependency ratios are lowest in Zambia and Malawi. Total food aid received by the region (260,000 tons in 2012, Figure S1) was equivalent to about 2-3% of food imported by the region from the rest of the world (9 million tons in 2008). Thus, chronic and episodic food insecurity remain important problems in the region. The causes are multiple and, at the household and individual level, are dominated by poverty, environmental stressors and conflict, often underpinned by structural elements in the lives of communities, intensified by sudden shocks which can be climate related such as decrease in cereal availability and food price spikes29,30. Climate signals in nexus sectors Multi-year rainfall variability in southern Africa is higher than in many other parts of the world31,32. Inter- annual variability, expressed as the coefficient of variation (CoV), is not particularly high at national scales: < 20% for most countries, except for the driest two countries Botswana and Namibia (Figure 3). However, rainfall displays much greater local variability (local CoV exceeds 20% across much of the SADC region), strong seasonality, and a range of multi-annual to decadal variations33. At the national level, long-term trends in rainfall between 1901 and 2012 are modest (the linear trend is insignificant relative to the long-term average) without evidence of any clear spatial pattern (Table S1). Linear trends during the last two decades show varied behaviour; three countries with wetting trends above 20% of the long-term mean annual rainfall (Botswana, Namibia and Zambia) and Tanzania with a drying trend of 21% (Table S1). National level analysis is likely to obscure local trends and the results are highly sensitive to the period chosen for analysis, particularly in regions with strong multi-annual variability. National variations in rainfall and temperature have been found to exert major influence on agricultural production in all of SSA, but with considerable regional heterogeneity in the response to rainfall34. Another study for SSA used panel regressions to explore the effects of temperature and precipitation variability on country level economic growth indicators and found drought was the most significant climate influence on GDP per capita growth35. We use correlation analysis to explore, for each country, the associations between annual rainfall and national economic activity (GDP annual growth rate) and agricultural production (all cereals and maize - the most significant crop in the region). Fifteen year 4 sliding correlations are used to examine the temporal stability of associations between variables (see SI Methods and data). There are no statistically significant relationships between annual rainfall and GDP growth rate (Table S2). Correlation of rainfall with total production of cereals and maize shows three countries with significant relationships at the 1% level and three at the 5% level (although for DRC, it is negative and possibly spurious). The average sliding correlations are somewhat higher (Table S3). Time series data of hydropower production are not publically available and not easily comparable between sites/countries, making it difficult to assess the importance of climate variability as a driver of energy production fluctuations. A study of the effects of modified reservoir operation to enhance downstream environmental flows of the Zambezi shows considerable variability in observed hydropower production at three sites, but does not consider the role of climate36. Electricity insecurity is known to negatively affect total factor productivity and labour productivity of small and medium-sized enterprises but the relationship is as yet not straightforward, with differences between countries and measurement effects37. Studies of specific events highlight major consequences of drought-induced reductions in electricity production38. Ref. 18 cites examples of drought impact on the Kariba Dam (Zambezi basin), during 1991-92, resulting in an estimated $102 million reduction in GDP and $36 million reduction in export earnings; and Kenya where, during 2000, a 25% reduction in hydropower capacity resulted in an estimated 1.5% reduction in GDP. A review of the economics of climate change in Tanzania profiled the consequences of the 2003 drought, which brought the Mtera dam reservoir levels close to the minimum required for electricity generation39. This prompted Tanzania Electric Supply Company (TANESCO) to approach a private provider to use gas turbine units at huge cost. A more recent World Bank estimate put costs of power shortages in Tanzania at $1.7 million per day with an average 63 days a year with power outages39. Early warnings from the climate system Given the linkages between climate and the water-energy-food nexus in the region, seasonal forecast information can play an important role in guiding nexus-related decision-making, depending on forecast skill and utility. Seasonal to inter-annual variability in southern Africa is high, but so is its predictability relative to other regions, depending on location, time of year40 and phase of the El Niño-Southern Oscillation41 (ENSO). This can be seen by considering the association (Figure 4a) between Nino3.4 sea surface temperatures (SST) - as a representation of ENSO - and gridded rainfall over southern Africa south of 15°S42. A state-of-the-art coupled ocean-atmosphere model has some skill in predicting seasonal (December to February, DJF) rainfall over the region at a 1-month lead-time (DJF forecasts produced in November, Figure 4b shows areas with statistically significant correlation41, see SI Methods and data). Stronger ENSO associations and the best model performance are found for maximum temperatures (Figure S2). The areas where ENSO impacts significantly and where forecast skill levels are relatively high include the river basins of the Limpopo, Orange, Umgeni and lower Zambezi. The Limpopo basin is particularly notable as having both high economic productivity and strong ENSO associations and forecast skill. Comprising 408,800km2, and including the countries of South Africa, Botswana, Mozambique and Zimbabwe, the Limpopo basin is one of the most water stressed in sub- 5 Saharan Africa, and features some of the largest urban conglomerations (including Pretoria, Johannesburg, Gaborone, Francistown and Bulawayo). Irrigation comprises more than 50% of basin water use and other infrastructure (including industry and mining) is also highly dependent on basin water. There are significant mining activities in the basin, particularly in South Africa and Zimbabwe42, that generate major water pollution downstream43. The Limpopo is heavily regulated, with extensive plans for further development. Despite forecast skill and potential utility in economic productivity hotspots such as the Limpopo basin, a comprehensive review of seasonal forecasting status in SSA identified persistent barriers in realizing the benefits of forecast products, which were generally insufficient to inform response actions, such as production decisions and institutional actions44. If these barriers can be overcome, seasonal forecasting has the potential to contribute to anticipating fluctuations in nexus sectors and could inform guidance on reservoir multi-use, water allocation, early targeting of or access to agricultural inputs and credit, design of interventions during food crises, and improvements to trade and agricultural insurance45. Modelling nexus sectors in a changing climate The challenges for the water-energy-food nexus posed by inter-annual variability occur in the context of a gradually changing climate. Even if an international agreement to limit global warming to 2°C above pre-industrial conditions is successfully developed, climate models project significant changes that exceed the range of natural climate variability (Figure 1). According to the majority of climate models, most southern African countries warm more than the global-mean, with annual-mean temperatures rising by 2 to 3°C in most cases. Precipitation changes are more uncertain, with both increases and decreases possible. Nevertheless, for most countries the majority of models project decreases in annual precipitation, typically by as much as 20% though more for some models and countries. Except for the southernmost countries, there is a tendency for models that warm most to simulate stronger reductions in precipitation. Analysis of extreme precipitation in the climate models used for IPCC AR4 shows a marked delay in rainy season onset over most of the region and an early end to the season in parts of the region46. Most nexus studies for southern Africa have been motivated by climate change and assess biophysical impacts for specific sectors, e.g., rainfall and irrigation water availability on crop production, or river flow changes on hydropower generation. Some crop models simulate sizable yield losses for southern Africa47, suggesting the region’s food system could be particularly vulnerable to climate change48. However, differences in climate scenarios, impact models, spatial and temporal scales and processes represented, restrict our ability to reliably define impacts for specific sectors and, importantly, secondary effects across the water-energy-food nexus. Nevertheless, an estimate of the range of potential impacts on maize yield (and the wide uncertainty range) can be determined from the 30-member ensemble of global gridded crop models run by the ISI-MIP programme49 (see SI). The simulated maize yield averaged across southern Africa decreases by 15.7±16.3% (rain-fed) and 8.3±20.4% (irrigated) by the 2080s relative to the 2000s, i.e. a median yield reduction with a substantial range of different outcomes. The wide range is due to climate uncertainties and uncertainties in our understanding of crop response 6 to climate change, particularly the role of elevated atmospheric CO concentration on photosynthesis. 2 Median impacts in the top five southern African producers are relatively small in the 2020s and 2050s, becoming more substantially negative by the 2080s, with a stronger level of agreement in the sign of change among simulations (Figure 5). Among these countries, rain-fed cultivation is most negatively impacted, highlighting that water stress is an important limiting factor to crop yield in the region. Average crop water use decreases, resulting in a 5.9±20.7% increase in estimated crop water productivity (see SI and Figure S3) by the 2080s. An ensemble of global hydrological models driven by five climate scenarios from the CMIP5 programme shows reductions in annual discharge from 0 to 50% for the multi-model mean across much of southern Africa, excluding Southwest Botswana50. River basin and water management models indicate higher risks for Zambezi hydropower generation51; while regional and global water and food models suggest lower runoff raises risks for water and food security in southern Africa in general52. The economic dimensions of the nexus in southern Africa can be studied using general equilibrium models that translate biophysical impacts into economic outcomes. This approach simulates economies as adapting to shocks, albeit imperfectly, through market and resource adjustments. Incorporating economic adaptation generally leads to smaller impacts than those from biophysical studies. Since global models rarely separate southern Africa from SSA, country-level studies are the region’s main evidence base. Historical climate variability imposes high costs on low-income agrarian economies53 and climate change is likely to have adverse effects on food security54. Long-term change in annual precipitation and temperature may impact less than historical variability until 205055;4. Historical data show substantial variability in smallholder farm yields and incomes. Increase in future variability of smallholder farm yields from climate change is therefore likely to increase the livelihood and food insecurity risks for farmers who are already at high risk54. Although most studies focus on agriculture, this is not always the main climate impact channel. For example, nexus studies find that road damages from flooding and weather stress are equally or more important drivers of the economic losses associated with climate change in Mozambique and South Africa56. More integrated multi- sector/country-level studies are needed to guide adaptation responses. A second strand of economic research focuses on climate and energy policy. A high proportion of SADC greenhouse gas emissions are from South Africa due especially to its reliance on coal-fired power. Curbing these emissions may reduce national income and employment, because financing domestic renewable options requires higher electricity tariffs57; 58. Lifting South Africa’s restrictions on hydropower imports would reduce investment costs and economic losses59. Climate change will have considerable indirect impacts on electricity generation with positive feedbacks. Higher water and air temperatures make cooling processes in coal-fired power plants less effective and potentially reduce water availability during longer dry periods24; this could result in an overall reduction of power plant efficiency and higher carbon emissions. Within its climate change strategy, Eskom aspires to diversify its energy generation mix to lower carbon-emitting technologies60. Solar photovoltaic and wind energy are considered to be the most viable renewable options in terms of water withdrawal and consumption compared with biofuel and hydropower25. Biofuels may reduce the region’s imported fossil fuels and 7 reduce rural poverty, but have potential food security trade-offs61. The research indicates that continued climate change, economic development and urbanization will strengthen inter-dependencies in the water, energy and food nexus in southern Africa and that climate and associated energy policy will further reinforce the costs of trade-offs and complementarities across the nexus, especially so if expansionist regional hydropower and biofuel strategies are adopted. Intra-regional instruments for the water, energy and food nexus Southern Africa can be characterised as a single economic block of strongly interlinked economies where water, energy and food flow between producers and consumers, while also displaying considerable heterogeneity in its natural resource endowments and infrastructure distribution, its socio- political cohesion and its economic development. For both the region and individual nations, this implies significant challenges in attempting to balance supply and demand while maintaining coherent policies towards integrated management of water-energy-food resources. The region is well placed to transfer resources intra-regionally to meet energy and food shortfalls. However, rising demand for electricity, food and water throughout southern Africa may sharpen the region’s sensitivity to climate-induced shocks. Fifteen trans-boundary river basins transect the region, including the large Congo and Zambezi basins, shared by nine and eight countries, respectively, as well as many smaller shared catchments. Surface catchments are underlain by an estimated 16 trans-boundary aquifers62. The origin of the southern African economic block can be tied to the dominant position of South Africa and its history alongside other ex-South African and British colonies such as Swaziland, Zimbabwe, Botswana, Namibia and Zambia. South Africa in particular has great cultural, economic and political influence over its neighbours making its role as a source (and sometimes a sink) of energy, water, and food hegemonic63. This alliance and influence is also evidenced via the SAPP (South Africa has 77% of SAPP’s installed power supply capacity64), the SADC and other agreements. In responding to the distribution of and demand for water-energy-food resources, three key instruments have emerged. First, the SADC, based in Botswana, addresses how member countries sharing rivers might resolve water allocation priorities through a Protocol on Shared Watercourses65; 66. The presence of significant water demands arising from irrigated agriculture and the Gauteng urban industrial complex in South Africa has led to relatively sophisticated water sharing agreements such as the Joint Development and Utilization of the Water Resources of Komati River Basin67 and the Lesotho Highlands Development Project. Large-scale dams and inter-basin, often trans-boundary, transfers (ref. 68 reports 27 existing ones)68 form part of national water-energy-food security strategies. South Africa and Zimbabwe, which have the largest numbers of dams, use these predominantly for irrigation and water supply, whilst Mozambique, which has one of the largest total dam capacities, concentrates on hydropower production (Table S4). Notwithstanding these institutional and physical structures, in some instances water sharing still suffers from a lack of institutional integration (particularly between agricultural and water institutions) and incomplete efforts to increase stakeholder participation and decentralise water management (ref. 69; reviewing South Africa, Zimbabwe and Mozambique)69. Coordination during flood events can also be challenging. For example, the persistent 2010/11 summer 8 rainfall in the Zambezi basin resulted in high water levels of Lake Kariba. Opening of spillway gates raised downstream water levels which increased flooding and compromised effective reservoir management at Cahora Bassa further downstream in Mozambique70. Second, the SAPP is a remarkable alliance of 12 energy-generating bodies from 12 countries inter- connected through a grid to help smooth spatial and temporal shortfalls in electrical capacity. It was established in 1995 by the member governments of SADC (excluding Mauritius) to develop an interconnected electrical system, coordinate and enforce common regional standards, harmonise relationships, develop expertise across member utilities, and promote sustainable development71. The SAPP electricity generating mix in 2012-13 was 54,923 MW, comprising a significant proportion from hydropower (17.4%), but dominated by coal (72.9%). The network is intended to function as a competitive market in which surpluses and deficits are resolved via trades and negotiations and therefore has potential to serve as a buffering mechanism for climate-induced river basin scale electricity insecurity. Third, food trade in southern Africa naturally results from regional variability in production, especially of maize. Large and efficient producers in South Africa induce a trade surplus with other SADC members. Importantly, trade of agricultural products corresponds to significant transfers of embedded water resources, or “virtual water trade” (VWT, see SI Data and Methods). Water resources embedded in South Africa’s and Zamia’s regional food exports (0.9 and 1.2 km3 in 2011, respectively, Figure 6a)72 account for two thirds of the total intra-regional flow (3.2 km3). The dominant link is from Zambia to Zimbabwe, with a volume of 0.8 km3/y of virtual water, followed by Mozambique to Malawi and South Africa to Zimbabwe (both 0.5 km3/y). Zimbabwe is the region’s major virtual water importer in 2011, importing 2.0 km3/y from other southern African nations. Considering all international food trade, southern Africa is largely a net importer of virtual water. Indeed, international imports from outside the region (10 million tons of food, or 20.5 km3 of virtual water) dominate the VWT flows of southern Africa (27.9 km3/y, Figure 6b). In return, smaller volumes to outside the region are exported mainly from South Africa (3.2 km3/y). Most of South African virtual water exports via food are embedded in maize, of which less than 10% is irrigation (blue) water (0.066 km3). This represents almost all of the intra-regional blue VWT (0.067 km3)73. This small percentage reflects the dominance of rain fed (green water) agriculture in the region. Although strong open trade is an important tool to alleviate climate-induced food deficits74, 75, and virtual water trade openness tends to reduce undernourishment76, southern African countries have varying levels of trade connectivity and trade link strengths, both for intra- and extra-regional food trade links. Thus, the potential benefits of food trade to alleviate production shocks are likely uneven across the region, and require further investigation. One of SADC’s main goals for regional integration is to promote trade across member countries. Efforts are ongoing to reduce major existing barriers, such as trade regulations and lack of reliable transportation infrastructure79, notably via the Protocol on Trade80, including facilitation of customs processes, and a regional infrastructure plan for the transport sector81. SADC is currently exploring opportunities for greater cross-sectoral coordination in the SADC Climate Change and Green Economy strategy (under revision at time of writing82) in which key recommendations focus on implementations 9 that will ensure actions do not take place in a single directorate. Such recommendations have arguably comprised the most demanding area of work for the strategy development, reflecting the importance of ensuring cross-sectoral coordination, as well as finding agreement on how to achieve it at the regional scale. Conclusion and outlook Climate plays an important role in determining medium-term water availability, potential agricultural production, and some components of energy production and demand. Climate variability drives fluctuations in WEF elements with secondary effects across the whole nexus (Figure 1). Exposure to climate variability and climate change are high across nexus sectors that include substantial areas of economic activity in southern Africa and there is strong evidence of the effects of individual climate events. For example, South Africa experienced a 7% drop in GDP in the 1983 El Niño year, and climatic fluctuations resulted in GDP variations of up to US $5 billion3. The 2000 floods in Mozambique led to devastating impacts on livelihoods, electricity supplies and basic infrastructure83. Yet our analysis of associations between rainfall, GDP and crop production using available data shows mostly weak and statistically insignificant correlations, in contrast to other studies for SSA based on panel regressions (Brown, Barrios). This is likely to be partly a function of scale, where national and annual scales obscure stronger relationships that may exist at finer levels of analysis. Data availability (e.g. absence of publically available hydropower production time series) and quality also play a role. The country climate estimates are often based on sparse station coverage, particularly since the 1980s84 and recent scrutiny of GDP data for SSA has highlighted lack of transparency in data sources and collection methods, lack of metadata and lack of detail on methods of aggregation85. This leads to differences between GDP estimates, non-random errors, adjustments to historical data, and inhomogeneity in time series. National statistical offices are woefully under-resourced in SSA, while the need for good quality data is paramount and urgent; to underpin reliable physical and economic modelling and detailed narrative of the causal linkages between climate and nexus sectors86. River flows in the region are strongly linked to seasonal rainfall and temperature variations, and the information reviewed here provides evidence that seasonal forecasting of river flows in some basins has application potential. However, the benefits from seasonal forecasting for reducing net food and energy imports through enhanced agricultural and hydropower production/energy mix have yet to be studied and, even more importantly, implemented in practice. For the future, climate models show fairly strong agreement that the southern countries in the region may become drier and the secondary impacts, though very uncertain, are likely to be substantial across the water-energy-food nexus. Water, energy and food are linked across different scales in southern Africa. Spatial interdependence is high and climate anomalies can produce regional scale effects, for example ENSO related droughts and river basin scale floods. At the national level, water and energy are closely coupled through significant hydropower production in several countries. Water use for biofuels and cooling for electricity generation remains relatively modest except for cooling in South Africa. In South Africa policies rarely cross sectoral boundaries of water and energy at all governance levels, yet integration of renewable 10