EmmaLi Johansson [email protected], Marianela Fader, Jonathan W. Seaquist, and Kimberly A. NicholasAuthors Info & Affiliations
Edited by Dieter Gerten, Potsdam Institute for Climate Impact Research, Potsdam, Germany, and accepted by Editorial Board Member Hans J. Schellnhuber August 24, 2016 (received for review December 18, 2015)
September, 2016
Significance
Freshwater appropriation can have vast impacts, depending on management and scale of water use. Since 2000, foreign investors have contracted an area the size of the United Kingdom in Africa, leading to increased pressure on water resources. Here we couple site-specific water demand for the crops planted there to the efficiency of different irrigation systems, while relating these estimates to local water availability. This approach enables us to identify “hotspot” areas of freshwater use where crops demand more water from irrigation than can be supplied by soil moisture, where the potential water demands from large-scale land acquisitions pose a risk for increased competition over water resources. Of these land acquisitions, 18% would be hotspots even with the most efficient irrigation system implemented.
Abstract
In the last decade, more than 22 million ha of land have been contracted to large-scale land acquisitions in Africa, leading to increased pressures, competition, and conflicts over freshwater resources. Currently, 3% of contracted land is in production, for which we model site-specific water demands to indicate where freshwater appropriation might pose high socioenvironmental challenges. We use the dynamic global vegetation model Lund–Potsdam–Jena managed Land to simulate green (precipitation stored in soils and consumed by plants through evapotranspiration) and blue (extracted from rivers, lakes, aquifers, and dams) water demand and crop yields for seven irrigation scenarios, and compare these data with two baseline scenarios of staple crops representing previous water demand. We find that most land acquisitions are planted with crops that demand large volumes of water (>9,000 m3⋅ha−1) like sugarcane, jatropha, and eucalyptus, and that staple crops have lower water requirements (<7,000 m3⋅ha−1). Blue water demand varies with irrigation system, crop choice, and climate. Even if the most efficient irrigation systems were implemented, 18% of the land acquisitions, totaling 91,000 ha, would still require more than 50% of water from blue water sources. These hotspots indicate areas at risk for transgressing regional constraints for freshwater use as a result of overconsumption of blue water, where socioenvironmental systems might face increased conflicts and tensions over water resources.
Sign up for PNAS alerts.
Get alerts for new articles, or get an alert when an article is cited.
Increased Competition Over Freshwater Resources
Freshwater is becoming increasingly scarce in many regions of the world, a result of both unsustainable land management and changes in rainfall patterns as a consequence of global and regional climate change (1). Moreover, the demand for water is increasing because of population growth, higher food demand, and changing dietary preferences, as well as increased industrialization and urbanization. Water, food, and energy are closely linked, and fundamental for human well-being, poverty alleviation, and sustainable development (2). As demand for water, food, and energy increases, there is an increased competition for water resources between agriculture, livestock, fisheries, forestry, energy, and other sectors, with unpredictable impacts for livelihoods and the environment.
Globally, agriculture is the most water-consuming sector, responsible for 70% of global freshwater withdrawals and more than 90% of consumptive water use (3). Agriculture’s freshwater use is causing severe environmental degradation in many parts of the world (4). This in turn affects local ecosystems and people, especially in countries where the population directly depends on the surrounding environment for their livelihoods. For example, Lake Chad has shrunk by 95% since 1963 as a result of large-scale irrigation projects in Chad, Nigeria, Niger, and Cameroon together with climatic changes (4). This is just one example of how large-scale irrigation has contributed to local water scarcity, and in turn harmed societies and ecosystems.
Large-scale conversion of land to agriculture to provide food, fiber, and energy needs to balance trade-offs between agricultural production, and other societal and ecosystem needs (5). It is important to weigh the benefits of increasing yields through irrigation with the consequences those water extractions might have on local and regional scales. The cumulative effect of local land-use changes also have regional to global consequences, to the degree that regional boundaries of freshwater use are transgressed, thereby increasing the risk for abrupt and irreversible environmental change (6), potentially creating new challenges for food, fiber, and energy supplies.
Green and Blue Water in Agricultural Production
Water embedded in agricultural production can be divided into site-specific precipitation stored in soils and consumed by plants through evapotranspiration (green) and surface and ground water in aquifers, rivers, lakes, and dams that can be extracted from renewable and nonrenewable sources for irrigation (blue) (7). Blue water is sometimes diverted from nonlocal sources to enable agricultural production (8). The volume of blue water required by agricultural systems differs, depending on crops planted, agricultural management, and water lost through evaporation from the water source to the field. The green and blue water concept can help estimate site-specific water demand of agriculture, and refine our understanding of human impacts on freshwater resources. Distinguishing between blue and green water indicates the volume of freshwater needed to meet human demands in addition to what is available through precipitation. These blue water extractions in turn might pose increased competition or conflicts between other water-using sectors.
Large-Scale Land Acquisitions and Freshwater Appropriation
Large-scale land acquisitions are areas larger than 200 ha contracted for commercial agriculture, for the purpose of timber extraction, carbon trading, food, feed, and renewable energy production (9). In 2014 the land-monitoring initiative Land Matrix had registered about 47 million ha of land contracted globally since 2000 under such large-scale land acquisitions. All deals in the database have at least one transnational investor from the public or private sector (including individuals, companies, investment funds, and state agencies), and may also include one or more domestic investors. These investors are currently key players in the modernization of African agriculture, and imply a conversion from smallholder production or community use to a commercial use of land and water (10).
The reasons for the rush for land are many, but were partly triggered by the food and energy crisis of 2007–2008 (11). Globalization, market liberalization, and commodification of land and natural resources in combination with the support of international donors have facilitated the implementation of these land contracts (11). Governments in the targeted countries may see foreign investments in land as an opportunity for agricultural modernization (10), as investors often motivate and legitimize their business proposals with rural and national development goals, typically including improved infrastructure, technological transfer, job opportunities, and financial benefits. However, research and nongovernmental organization reports (12–14) point out that large-scale land acquisitions rarely benefit local people, and that the proposed infrastructure is often not developed on the local scale.
Africa is the continent where most land has been contracted (about 22 million ha) because of cheap land and labor costs (15), but also has the potential to boost yields and reduce yield gaps with modern agricultural techniques and irrigation systems (16). However, many of these land deals have been abandoned or are not yet in production (17), and only about 3% of the contracted deals (0.7 million ha) are currently in production (Dataset S1). These numbers are constantly changing, as land acquisitions are expanding, abandoned, or were never implemented. An example of this is the belief that Chinese investors are major actors acquiring large tracts of land in Africa, which has recently been shown to be on a smaller scale than first reported (18).
The rush for water might be just as important for investors as the rush for land (10, 19, 20). Land contracts rarely indicate any limits to water use, which means that investors might choose inexpensive and inefficient irrigation for their operations. The lack of water regulations thereby increases the risk of unsustainable water use, which in turn has the potential to alter the availability and accessibility for local communities, ecosystems, and other water-intensive sectors.
Human appropriation of freshwater can have vast impacts depending on the management and scale of water use (21), highlighting the importance to estimate the growing water demand associated with land transformations in Africa. No study has yet connected the site-specific water demand to water-use efficiencies of different irrigation systems. This connection is vital because it can indicate areas that might experience increased water stress or conflicts over water resources. The objectives for our study, therefore, are: (i) to estimate and identify site-specific green and net blue water demand of crops grown on acquired land in production; (ii) to calculate yields, as well as green and gross blue water demand, for crops grown on acquired land under seven irrigation scenarios, and for staple crops as a baseline; and (iii) to develop a Blue Water Index (BWI) to identify hotspot areas of increased competition for freshwater resources where demand for blue water exceeds green water supply.
We note that previous continental- to global-scale studies of land acquisitions have met serious critique for issues with data selection biases and quality of data sources, therefore producing results of questionable accuracy (22, 23). One notable example found that 310 km3⋅y−1 of green water and 140 km3⋅y−1 of blue water are appropriated globally for crop and livestock production (24). However, this study included contracted global land deals, which likely overestimate the water use on acquired land because few projects are currently in production.
As a response to these critiques, and to meet our objectives, we focus on land deals in production. We model green and blue crop water demand with the dynamic agro-ecosystem and hydrology model Lund–Potsdam–Jena managed Land (LPJmL), and provide a clarification of model assumptions and parameterization for the given crops planted. The model output includes: (i) green water demand met by rainfall; (ii) net blue water demand that plants need to grow, in addition to rainfall; and (iii) gross blue water demand that has to be extracted to fulfill plant requirements, accounting for losses between the water source and the field. Water losses depend on the efficiency of irrigation and their distribution systems (see description in Table 1). Finally, we validate the data by cross-referencing the 54 largest land deals in Google Earth, responsible for 95% of acquired land area in production (SI Materials and Methods). Because there is a lack of information about water management of acquired land, we model seven different irrigation systems to obtain a full range of plausible water-use efficiency scenarios.
Table 1.The seven different irrigation scenarios that were run with LPJmL for crops planted on large-scale land acquisitions in Africa
| Irrigation scenario | Description |
| Rainfed | Rainfed agriculture (modeled for crops currently in production on acquired land, and for the staple crop baseline) |
| Drip (pipelines) | Micro (drip) irrigation with pressurized pipelines for distribution |
| Sprinkler (pipelines) | Irrigation with sprinklers supplied by pressurized pipes |
| Mixed | Irrigation with a mix of surface and sprinkler irrigation systems with both open canals and pressurized pipes |
| One-step improvement | Irrigation and distribution systems that are one step higher in efficiency than current national irrigation efficiencies (e.g., moving from sprinkler to drip systems). |
| Current irrigation efficiencies | Irrigation under current national irrigation and distribution systems in every country (39) (modeled for crops currently in production on acquired land, and for the staple crop baseline) |
| Surface (open canals) | Surface irrigation systems (flooding) with water diverted from open canals |
EXPAND FOR MORE
Scenarios are presented from most (top) to least (bottom) efficient, based on gross blue water use. The current irrigation efficiency, and therefore also the one-step improvement, varies by country.
SI Materials and Methods
Additional Information About LPJmL Modeling.
LPJmL is an agro-ecosystem and hydrology model that uses gridded (0.5° resolution) monthly climate inputs (temperature, cloudiness, rainy days, and precipitation from Climate Research Unit 3.10), soil textures, and global atmospheric CO2-concentrations to model hydrological variables, phenology, agricultural yields (ton per hectare), and the carbon cycle.
LPJmL has a detailed hydrology module, with a river routing and irrigation scheme (8), management of dams and reservoirs (40), and a five soil-layer hydrology (41). The performance of LPJmL’s hydrology, including the simulation surface and subsurface runoff, soil evaporation, plant transpiration, infiltration, and percolation, has been demonstrated in numerous studies and validation efforts (42, 43). In LPJmL, the soil water content depends on climatic variables (notably temperature and rainfall), vertical and horizontal water movements, soil evaporation, and water extraction by plant roots. Irrigation is triggered in irrigated areas when soil water content is lower than 90% of field capacity in the upper 50 cm. Net irrigation water requirements for specific plants are modeled as the amount of water that plants need according to atmospheric demand, taking into account the relative soil moisture and the water holding capacity of the irrigated layer. Water withdrawal or extraction is obtained by dividing net irrigation water requirements by the overall irrigation efficiencies. Gross irrigation efficiencies are the result of the combination of distribution efficiency (transporting water from source to field), field application efficiency (how much water is lost to evaporation upon storage and application), and a management factor that represents higher water efficiencies in pressurized systems.
These parameters were compiled by Rohwer et al. (39), who also provided a map with the current dominant irrigation and distribution systems in every country. We used this dataset for the run called “national irrigation efficiencies” (see main text). Because Land Matrix data do not indicate whether the land acquisitions under production are being irrigated, we simulated the full range of possibilities, including rainfed (nonirrigated) agriculture, and irrigated agriculture with: (i) current national irrigation systems, (ii) improved systems (“one-step” improvement), (iii) surface irrigation and open channels, (iv) sprinklers and pressurized pipes, and (v) drippers and pressurized pipes (see Scenarios for Green and Blue Water Demand).
An agricultural management module represents the combined influence of management practices on plant and stand development by means of parameters indicating the planting density of trees for perennial crops, the harvest index, homogeneity of the fields for annual crops, and maximum attainable leaf area index. The combination of these parameters represents the influence of management, including fertilizer inputs, mechanization, use of high-yielding varieties, weed and pest control, and so forth (see refs. 37 and 38 for more details). Normally, the parameters are calibrated to best match reported yield from the Food and Agriculture Organization of the United Nations (FAO). However, for the present study it is assumed that land acquisitions will be managed intensively and, thus, plant growth will not be limited by pest attacks, lack of nutrients, and so forth. Consequently, the runs were performed with optimized management parameters. Further details about the parameterization of crops are found in Bondeau et al. (36) for annual crops, Fader et al. (37) for perennial crops, and Beringer and Lucht (44) for bioenergy crops.
Sowing dates for annual crops are calculated on the base of temperature and rainfall seasonality (45). LPJmL’s phenology approach follows the heat unit theory where the growth and maturity are simulated by accumulating temperatures higher than a designed base temperature until reaching a prescribed or calculated sum of degree-days. The dynamic phenology approach and calculation of sowing dates is one of the strengths of the model because it allows adjusting the growing periods (sowing and harvest dates) to changes in climate.
Carbon allocation to different parts of the plant is a function of the development of accumulation of degree-days. The fruits of agricultural trees are represented by a carbon accumulation equal to a prescribed portion of the net primary productivity.
Chilling requirements of fruit trees are compared with 20-y running average of the coldest-month maximum and minimum temperatures. Hence, temperature warming above these limits would inhibit the establishment and survival of the perennial crops.
The vegetables and fodder grasses (alfalfa) are modeled as C3 grasses, and photosynthesis is assumed to be optimal between 10 and 30 °C. Once their phenology is complete (i.e., the growing degree-day accumulation determined by a parameter was reached) and the biomass increment is equal or greater than 200 g C m−2, 50% of the aboveground biomass is transferred to the harvest compartment. The class “managed grasslands” functions in a very similar way, with the only difference that it is represented by a mixture of C3 and C4 grasses.
The extended version of LPJmL from Fader et al. (37) that was used in this study represents 26 crops or crop groups, of which the following were used in this study: 12 annual crops (temperate cereals, rice, maize, pulses, tropical cereals, tropical root, sunflower, soybeans, groundnuts, oil seeds, sugar cane, potatoes), five agricultural trees, shrubs, and vines (date palms, citrus, olives, grapes, and cotton), and managed grasslands.
For this study, we included bananas in LPJmL for the first time. Bananas were represented by an evergreen, broadleaved plant with a C3 photosynthetic pathway and with a maximum height of 5 m. Bananas are planted with an average density of 3,000 plants per hectare (46) and have a harvest ratio of 0.5 (47). Bananas are assumed not to grow in areas where the coldest monthly mean temperature is below 5 °C because the FAO assumes damage already at temperatures slightly below zero (48); their optimal photosynthesis temperature range is 26–30 °C (47).
For the crop groups named above, the parameterization linked to biological and agronomical processes is implemented by means of one representative crop per group (Table S4): wheat for temperate cereals, dry pea for pulses, millet for tropical cereals, cassava for tropical root, and orange tree for citrus. The categories “vegetables” and “fodder grass” are modeled as C3 grasses, as described in Sitch et al. (49) (Table S4).
Most of the crops described for the production of the land acquisitions are included in LPJmL, but some, including acacia, alfalfa, cacao, castor oil plant, coffee, flowers, jatropha, oil palm, pongamia pinnata, rubber, sesame, tea, and teak were represented through the class “managed grasslands” to give an estimate for the behavior of these crops. These crops represent 31% of the total area (Table S4). Future model development could focus on parameterizing oil palm (14.3% of planted area) and rubber (7.5% of planted area). The largest crop group in the Land Matrix data (43.3%) is listed only as “trees,” which we have modeled as citrus trees here.
Baseline for Water Requirements of Staple Crops.
To evaluate the added effect on freshwater use by land acquisitions, we modeled a baseline of water use for the five most common staple crops in Africa: wheat, rice, cassava, maize, sorghum (retrieved from GIEWS Country Briefs, www.fao.org/giews/countrybrief/, FAO, May 2016). This was done by modeling the water requirements for rainfed and national irrigation efficiency scenarios (Table S3). The management intensity was changed to represent agricultural management and yields in 2000. The model runs for the past are therefore not based on optimal agricultural management, but a calibrated version to best match reported yield from the FAO.
Cross-Referencing the Land Matrix Data with Google Earth.
To get an idea of past land use, assumptions for irrigation, and current production area of land acquisitions, we cross-referenced the data by observing the 54 largest land acquisitions in Google Earth, representing 95% of acquired land area. First, we consulted the investor webpage for information on type of production, irrigation practices, and location. Thereafter we looked at the satellite images to see if the land acquisitions are visible by looking for areas of agricultural fields. Of 54 land deals, 32 were positively identified, 9 were uncertain, and 13 could not be located (2 because of poor spatial resolution). We found that 14 of the 54 land deals were likely irrigated, based on webpage information, or from the field shape [clear linear or circular fields, greener areas than the surrounding vegetation (examples shown in Fig. S2)]. It was difficult to discern if large-scale tree plantations (like oil palm) were irrigated, even if the fields showed linear structure (Fig. S2B). We also tried to identify past land use, but this was difficult becaues of poor spatial resolution, but we did ascertain that some areas previously consisted of either natural, or no vegetation (Fig. S2 C and D), or a mix of natural vegetation and small-scale farming (Fig. S2A). We also show that the Land Matrix dataset might underestimate the current state of land in production with an example of a deal (#1798) registered as 44,000 ha contracted and 0 ha in production when the data were retrieved (Fig. S2E).
Results
Land Acquisitions in Africa for Water-Intense Crop Production.
More than 60% of the acquired land in production (>418,000 ha) is for forestry purposes (Fig. 1). Most tree species are not specified, but rubber, eucalyptus, pine, and teak are commonly grown for timber or pulp and, in some cases, carbon sequestration (Fig. 1). The next largest crop group is flexible crops, covering 244,000 ha (35% of acquired land). Flexible crops can be used for food, feed, or biofuel; here, these include sugarcane, oil palm, soybean, maize, wheat, sorghum, and cotton. The remaining 5% are acquired for food and beverage crops (tea, coffee, fruits, vegetables), biofuel crops (jatropha), feed, and flowers (Dataset S1).
Fig. 1.
Table S1.Country names and ISO3 codes for the countries where land acquisitions are in production in Africa
| ISO3 | Country Name |
| AGO | Angola |
| BFA | Burkina Faso |
| CAF | Central African Republic |
| CIV | Cote d’Ivoire |
| COD | Democratic Republic of the Congo |
| COG | Congo |
| EGY | Egypt |
| ETH | Ethiopia |
| GAB | Gabon |
| GHA | Ghana |
| GNB | Guinea-Bissau |
| KEN | Kenya |
| LBR | Liberia |
| MDG | Madagascar |
| MOZ | Mozambique |
| NGA | Nigeria |
| SDN | Sudan |
| SEN | Senegal |
| SLE | Sierra Leone |
| SSD | South Sudan |
| TZA | United Republic of Tanzania |
| UGA | Uganda |
| ZMB | Zambia |
| ZWE | Zimbabwe |
EXPAND FOR MORE
Fig. S1.
Our results from LPJmL show that some crops require more water than others (Table S2), but also that the same crop varies in water demand depending on temperature and rainfall. It is possible to distinguish between two crop groups, one with lower water demand (sorghum, soybean, wheat, maize, rice) and one with higher water demand (cotton, eucalyptus, jatropha, oil palm, pine, rubber, sugarcane, teak, trees) (Fig. 1). Within each group, there is a large variation in the amount of green and blue water required to meet the total water demand. For example, sugarcane in the Sudan (green bubble in upper left corner of Fig. 1) has an average net water demand of 13,390 m3⋅ha−1 (bubble size) of which 90% is blue water and 10% is green, whereas in Gabon the average net water demand is 15% lower, of which 11% is required from blue water sources and 89% supplied from green water (green bubble in the lower right corner of Fig. 1).
Table S2.Average green, net blue, and total water demand of crops grown on 121 land acquisitions currently in production in Africa (Dataset S1), sorted in descending order according to total water demand
| Crop | Green water demand (m3⋅ha−1) | Net blue water demand (m3⋅ha−1) | Total water demand (m3⋅ha−1) | Area in production (ha) |
| Alfalfa | 0 | 17,291 | 17,291 | 1575.0 |
| Tomatoes | 1,764 | 14,628 | 16,392 | 100.0 |
| Cotton | 2,115 | 12,913 | 15,028 | 6504.0 |
| Castor oil plant | 5,584 | 8,704 | 14,288 | 500.0 |
| Palms | 1,499 | 12,697 | 14,196 | 7.5 |
| Pongamia innata | 6,378 | 7,350 | 13,728 | 1250.0 |
| Sugar cane | 7,734 | 5,756 | 13,490 | 81275.0 |
| Sesame | 8,627 | 4,810 | 13,437 | 16.7 |
| Eucalyptus | 8,455 | 4,773 | 13,228 | 29033.7 |
| Citrus fruits | 7,578 | 5,491 | 13,069 | 42.9 |
| Pineapple | 8,013 | 5,008 | 13,021 | 2125.0 |
| Trees | 7,044 | 5,085 | 12,129 | 301387.2 |
| Coffee plant | 8,251 | 3,874 | 12,125 | 3075.3 |
| Jatropha | 6,741 | 5,349 | 12,090 | 10963.5 |
| Accacia | 7,723 | 4,296 | 12,019 | 825.0 |
| Olives | 3,301 | 8,716 | 12,017 | 800.0 |
| Food crops (no specification) | 8,237 | 3,724 | 11,961 | 1070.0 |
| Cacao | 7,281 | 4,411 | 11,692 | 1362.5 |
| Pine | 8,200 | 3,413 | 11,613 | 22301.2 |
| Teak | 8,134 | 3,306 | 11,440 | 12970.3 |
| Tea | 7,155 | 4,272 | 11,427 | 4635.3 |
| Onion | 5,793 | 5,607 | 11,400 | 125.0 |
| Vegetables | 6,561 | 4,811 | 11,372 | 1124.7 |
| Flowers | 5,996 | 5,289 | 11,285 | 343.0 |
| Pepper | 7,958 | 3,022 | 10,980 | 42.9 |
| Oil Palm | 8,351 | 2,433 | 10,784 | 99541.3 |
| Grapes | 2,978 | 7,245 | 10,223 | 127.5 |
| Banana | 6,725 | 2,848 | 9,573 | 1614.1 |
| Rubber | 8,234 | 1,232 | 9,466 | 51859.8 |
| Fruit | 7,087 | 1,595 | 8,682 | 539.7 |
| Cassava (Maniok) | 6,186 | 802 | 6,988 | 1384.5 |
| Peanut | 5,648 | 1,227 | 6,875 | 264.7 |
| Maize | 4,811 | 2,005 | 6,816 | 10121.2 |
| Rice (hybrid) | 3,986 | 625 | 4,611 | 432.0 |
| Oil seeds | 3,651 | 627 | 4,278 | 1591.7 |
| Barley | 2,458 | 1,754 | 4,212 | 2604.6 |
| Peas | 2,405 | 1,801 | 4,206 | 500.0 |
| Wheat | 3,162 | 1,015 | 4,177 | 7265.4 |
| Rice | 3,915 | 203 | 4,118 | 9524.5 |
| Potatoes | 3,524 | 472 | 3,996 | 125.0 |
| Bean | 3,355 | 249 | 3,604 | 42.9 |
| Grains | 3,304 | 45 | 3,349 | 116.7 |
| Sun Flower | 2,801 | 358 | 3,159 | 1734.3 |
| Cereals (no specification) | 3,082 | 0 | 3,082 | 500.0 |
| Pulses | 2,802 | 275 | 3,077 | 333.3 |
| Sorghum | 2,511 | 540 | 3,051 | 7558.3 |
| Soya beans | 2,259 | 410 | 2,669 | 13829.0 |
EXPAND FOR MORE
Scenarios for Green and Blue Water Demand and Crop Yields.
Simulating water demand with LPJmL enables us to compute water demand of land deals for different irrigation scenarios. Irrigating all land acquisitions has the potential to almost double yields for the crops planted on acquired land (from 17 to 28 megatons) compared with purely rainfed management (Fig. 2 and Table S3). However, this would come at the cost of blue water extractions, which in turn require infrastructure for freshwater appropriation from either local or distant water sources, causing negative impacts on freshwater systems. Total water demand for land acquisitions in production ranges between 5.4 (rainfed) and 8.5 km3⋅y−1, depending on the water use efficiency of the irrigation system (Fig. 2). If all land acquisitions were irrigated with the most efficient system (drip irrigation with pressurized pipes), the annual gross blue water use would be 2.1 km3⋅y−1, compared with 3.5 km3⋅y−1 if the least-efficient irrigation system (surface irrigation with open canals) were used, leading to a water efficiency improvement of up to 40%.
Fig. 2.
Table S3.Total green, net and gross blue water demand of all land acquisitions in production in Africa, and two baseline scenarios for staple crops, as well as production and water content (m3⋅ton−1) for the different irrigation scenarios
| Code | Irrigation type | Green water (km3⋅y−1) | Net blue water (km3⋅y−1) | Gross blue water (km3⋅y−1) | production (million tons) | Total gross water demand (km3⋅y−1) | Virtual water (m3⋅ton−1) |
| a | Rainfed | 5.43 | 0 | 0 | 17.23 | 5.43 | 315 |
| b | Current irrigation efficiencies | 5.02 | 1.88 | 3.18 | 28.09 | 8.20 | 292 |
| c | Surface irrigation and open canal distribution | 5.04 | 1.87 | 3.47 | 28.12 | 8.50 | 302 |
| d | Mixed irrigation | 4.90 | 1.93 | 3.01 | 28.08 | 7.91 | 282 |
| e | Sprinkler irrigation and pipeline distribution | 4.82 | 2.06 | 2.75 | 28.11 | 7.57 | 269 |
| f | Drip irrigation and pipeline distribution | 5.05 | 1.86 | 2.07 | 28.11 | 7.12 | 253 |
| g | One step improvement from current efficiencies | 4.90 | 2.01 | 3.02 | 28.09 | 7.93 | 282 |
| Baseline 1 | Rainfed (year 2000) | 3.3 | 0 | 0 | 2.7 | 3.3 | 1,263 |
| Baseline 2 | National irrigation efficiencies (year 2000) | 3.3 | 0.28 | 0.49 | 2.8 | 3.8 | 1,372 |
EXPAND FOR MORE
Producing maximum crop yield (Fig. 2) requires less water for all irrigated cases, ranging from 255 m3 of water per ton of crop yield for drip irrigation, to 300 m3⋅ton−1 for open canal surface irrigation, compared with 315 m3⋅ton−1 for rainfed agriculture (Fig. 2 and Table S3). This means that establishing irrigated agriculture in areas of production would demand more water in total, but require less water per unit of production compared with the rainfed case.
To estimate the added pressure on water resources by land acquisitions compared with previous land use, we provide a baseline of water demand for five staple crops widely grown under small-scale farming systems in the affected countries: maize, wheat, rice, sorghum, and cassava (Dataset S2). We model the most common staple crops grown in that country, at the location of the acquired land, using both rainfed and national irrigation efficiency scenarios for the year 2000. The water requirements for staple crops varies between 2,500 and 14,500 m3⋅ha−1 (average 5,500 m3⋅ha−1), summing up to a total green water use of 3.3 km3⋅y−1 under rainfed conditions, with an additional 0.5 km3 of blue water if the staple crops were irrigated with national irrigation efficiencies (Table S3). This finding suggests that green and blue water use is 39% and 76–86% greater, respectively, for crops grown on acquired land compared with the baseline of common staple crops, showing that land acquisitions substantially increase water demands.
Mapping Blue Water Use Hotspots with the BWI.
To assess what areas might face increased water scarcity as a result of land acquisitions, we relate crop water demand to the water supply of the specific area in production by calculating the ratio between the gross blue water demand to the total (green + gross blue water) water demand for these crops. We call this the Blue Water Index, which indicates the fraction of water added from irrigation needed to generate maximum yields. An index of 1 indicates that all crop water comes from irrigation, whereas an index of 0 indicates that precipitation is sufficient to achieve maximum yields. The BWI helps identify those land acquisitions that might have large impacts on freshwater availability.
For all irrigation scenarios, land acquisitions in production with a BWI lower than 0.5 (less than 50% of water demand from blue water sources) are distributed in tropical and temperate climate zones of sub-Saharan Africa (Fig. 3 and Dataset S3), whereas land acquisitions with a BWI above 0.5 (more than 50% of water demand from blue water sources) are scattered throughout all climate zones from dry to tropical (Fig. 3).
Fig. 3.
These blue water hotspots are mapped in Fig. 4, which shows the effect of irrigation system on blue water use. Under current national irrigation efficiencies, 35% of all land deals in production would be hotspots (Dataset S3), with 33 land deals using 50–75% blue water and 9 using >75%. The remaining 46 deals using 25–50% blue water and 33 deals using <25% (Dataset S3) may still stress local water systems. If more efficient sprinkler or drip irrigation were applied, hotspot areas would drop to 22% and 18% of total deals in production, respectively (red and yellow dots in Fig. 4). Even under the most efficient drip irrigation system, there will still be 22 land acquisitions where more than 50% of water would be drawn from blue water sources to meet demand, most in Central and Eastern Africa (red dots in Fig. 4).
Fig. 4.
Discussion
Water Intense Crop Production of African Land Acquisitions.
As shown in this study, most acquired land in production is for forestry and flexible crop production for crops with high water demand (e.g., sugarcane, jatropha, trees, and eucalyptus). It is relevant to consider site-specific green and blue water demands of individual land acquisitions to identify land deals that might induce water stress, and cause water-related conflicts between different water users. Blue water demand depends on crop choice and location. From a water-efficiency perspective, for example, it is better to grow sugarcane in the Congo than in the Central African Republic, but in the context of food security it might be better to develop the land for food crops that require less water, like maize, rice, sorghum, and wheat (shown in the light gray zone in Fig. 2). However, in reality, low water demand is not the primary driver of crop choice, but rather local to global demand, market prices, and nutrient calorie content play more dominant roles in deciding crop production (25).
Scenarios for Green and Blue Water Demand.
Green water demand from crops now planted on acquired land (5.4 km3⋅y−1) is substantially higher than it would be for traditional staple crops (3.3 km3⋅y−1). It is important to consider the scale of production when calculating blue water use, and to assess how blue water demands differ depending on the irrigation system implemented. Irrigating all crops currently in production for land acquisitions on a continental scale would require 2.1–3.5 km3 of blue water per year in addition to what is supplied naturally from rainfall. By adding this amount of blue water, it is possible to maximize and almost double yields compared with rainfed agriculture. It is reasonable to assume that investors irrigate acquired land because they want to guarantee high agricultural productivity and reduce the risk of crop failure because of erratic rainfall (10). Land acquisitions in semiarid regions are, however, more likely to be irrigated than in tropical regions, as a result of crop type and relative availability of green water. Note that this is accounted for in LPJmL, as blue water requirements are only added if needed to avoid soil water deficit.
Improving water-use efficiency, while also considering the purpose of production (food, feed, or fuel) and the location of consumption, is essential for developing more sustainable agricultural systems. Efficiently irrigated agriculture contributes to increased yields and also allows allocation of water to other sectors, like sanitation and health, but for already water-scarce regions, the additional extraction of blue water might be substantial even if the most efficient irrigation system is implemented. If water is available and free of charge, investors will probably prefer cheap and inefficient irrigation systems, such as surface irrigation ($600–800/ha) or sprinklers ($3,000–5,000/ha) rather than expensive but efficient drip irrigation systems ($10,000/ha) (26). In reality, the irrigation scenarios are linked to factors like economic costs, labor intensities, water rights, and governance. However, these factors could not be considered in the present biophysical framework, where our goal was to bracket the range in water use given a variety of irrigation scenarios.
Blue Water-Use Hotspots.
The BWI was developed to delineate hotspot areas of blue water demand, and to indicate areas where increased freshwater use potentially creates tensions and conflicts between different water users. We find that 22% of land acquisitions in production (for the sprinkler irrigation scenario) require more than 50% of their water from blue water sources. Land acquisitions with a BWI above 0.5 are scattered over all climate zones, from dry to tropical, which indicates that it is not only the lack of rainfall that gives rise to water-scarcity hotspots, but also crop choice and scale of production. For example, both West Africa and Madagascar are tropical zones where there are land deals with both high and low BWI.
Beyond Water.
Just because a crop planted on acquired land is suitable to grow in that area does not mean that the area is suitable for large-scale agriculture. There are several other risks to be considered, including biodiversity loss from land conversion, and loss of land-rights for people who are engaged in small-scale farming (among others). For example, Central Africa has the second-largest rainforest in the world and is rich in biodiversity (27). Until recently, forests there have remained largely intact because of low demographic pressure and limited accessibility (28), but deforestation has increased in recent years as a result of the rush for farmland (29). Consequently, large areas of forests and people’s access to land are threatened (30).
Many investors claim to stimulate local and national development, thereby reducing rural poverty and food insecurity (11, 25); however, optimizing yields for timber or biofuel crops for export might not be the most suitable option to do so (31). Even though socioeconomic benefits in terms of infrastructure and employment might contribute to food security on the local scale (25), case studies have found that this has not been realized on the ground (32, 33). Therefore, there is a need to further examine local implications for rural societies and ecosystems, and whether the crop production is of benefit to the national or local population. This approach would shed light on the trade-offs between the purpose of production and increased yields at the cost of ecosystem health (e.g., water pollution, reduction of wildlife, and deforestation), as well as local to national socioeconomic trade-offs regarding infrastructure development and employment.
Data Limitations and Key Assumptions.
There are several data limitations for the land deals themselves, as well as current water management in the study area. Although Land Matrix is the most extensive dataset currently available, it is continuously being updated as a result of the rapidly changing nature of the land deals, highlighting issues of uncertainty (34, 35). Nevertheless, these data are suitable for showing general trends and patterns.
Additionally, there is a lack of information about water management for current land deals, which is why we modeled different irrigation scenarios. To refine this measure, there is a need for additional research on the types of irrigation systems that are implemented, the source of blue water, as well as on how the water is diverted to the irrigation system. It is also a challenge to estimate the added pressure on freshwater use by land acquisitions, because there is a lack of data about previous land use. This is a research gap that needs to be filled to assess changes in water use with greater confidence.
This study is an estimate of how much water the plantations on acquired land might require for different types of irrigation systems. We assume (using LPJmL) that irrigation requirements can always be met. This assumption is reasonable, given the additional assumption that investors are likely to assess the availability of water (and potential for profit) associated with leasing or purchasing land. It is worth noting that the aggregate figures of land and water use from land acquisitions are likely to be an underestimate because of the conservative assumptions made in this analysis.
Finally, crops that are not specifically parameterized in LPJmL were modeled as crops with similar behavior (SI Materials and Methods). The class “managed grasslands” was used as a proxy for 21 crops covering 31% of acquired land (Table S4). Although uncertainties introduced by this procedure may be low for crops like alfalfa, uncertainties will be higher for tree crops. Consequently, estimates for water use and crop production should be treated with care for these crops. Future model development should focus on parameterizing the most widespread crops not currently in LPJmL: oil palm (14.3% of planted area) and rubber (7.5% of planted area).
Table S4.List of crops from Land Matrix and the area in production, together with the representative crop or crop group used to represent the crops in LPJmL
| LPJ name | Land Matrix crop | Area (ha) | Land Matrix crop (% of total area) | LPJ name (% of total area) |
| Bananas | Banana | 1,614.1 | 0.23 | 0.31 |
| Fruit | 539.7 | 0.08 | ||
| Citrus | Citrus fruits | 42.9 | 0.01 | 47.46 |
| Eucalyptus | 29,033.7 | 4.17 | ||
| Trees | 301,387.2 | 43.28 | ||
| Cotton | Cotton | 6,504.0 | 0.93 | 0.93 |
| Date palms | Palms | 7.5 | 0.00 | 0.00 |
| Grapes | Grapes | 127.5 | 0.02 | 0.02 |
| Groundnuts | Peanut | 264.7 | 0.04 | 0.04 |
| Maize | Corn (maize) | 10,121.2 | 1.45 | 1.45 |
| Managed grass | Accacia | 825.0 | 0.12 | 31.17 |
| Alfalfa | 1,575.0 | 0.23 | ||
| Cacao | 1,362.5 | 0.20 | ||
| Castor oil plant | 500.0 | 0.07 | ||
| Coffee plant | 3,075.3 | 0.44 | ||
| Flowers | 343.0 | 0.05 | ||
| Food crops (no specification) | 1,070.0 | 0.15 | ||
| Jatropha | 10,963.5 | 1.57 | ||
| None | 1,260.0 | 0.18 | ||
| Oil palm | 99,541.3 | 14.30 | ||
| Onion | 125.0 | 0.02 | ||
| Pepper | 42.9 | 0.01 | ||
| Pine | 22,301.2 | 3.20 | ||
| Pineapple | 2,125.0 | 0.31 | ||
| Pongamia pinnata | 1,250.0 | 0.18 | ||
| Rubber | 51,859.8 | 7.45 | ||
| Sesame | 16.7 | 0.00 | ||
| Tea | 4,635.3 | 0.67 | ||
| Teak | 12,970.3 | 1.86 | ||
| Tomatoes | 100.0 | 0.01 | ||
| Vegetables | 1,124.7 | 0.16 | ||
| Olives | Olives | 800.0 | 0.11 | 0.11 |
| Potatoes | Potatoes | 125.0 | 0.02 | 0.02 |
| Pulses | Bean | 42.9 | 0.01 | 0.13 |
| Peas | 500.0 | 0.07 | ||
| Pulses | 333.3 | 0.05 | ||
| Rice | Rice | 9,524.5 | 1.37 | |
| Rice (hybrid) | 432.0 | 0.06 | 1.43 | |
| Soybeans | Soya beans | 13,829.0 | 1.99 | 1.99 |
| Sugar cane | Sugar cane | 81,275.0 | 11.67 | 11.67 |
| Sunflower | Sun flower | 1,734.3 | 0.25 | 0.25 |
| Mean value of sunflower, soybeans, groundnuts and rapeseed | Oil seeds | 1,591.7 | 0.23 | 0.23 |
| Temperate cereals | Barley | 2,604.6 | 0.37 | 1.51 |
| Cereals (no specification) | 500.0 | 0.07 | ||
| Grains | 116.7 | 0.02 | ||
| Wheat | 7,265.4 | 1.04 | ||
| Tropical cereals | Sorghum | 7,558.3 | 1.09 | 1.09 |
| Tropical roots | Cassava (Maniok) | 1,384.5 | 0.20 | 0.20 |
EXPAND FOR MORE
Conclusions
Our study quantifies water demand of land acquisitions in Africa as a function of crop choice, local climate, and irrigation scenarios. As such, it advances the field by detailing the implications of crop choice and irrigation techniques on water demand. It also highlights areas that might experience conflicts and tensions over freshwater use between sectors, especially hotspots using more than 50% blue water for crop production.
We show that there is potential to boost yields through irrigation, but that blue water demand varies with irrigation system (because of water use efficiencies). Even if the most efficient irrigation system is used for land acquisitions in production, 18% would require more than 50% of water from blue water sources. If land acquisitions are to benefit local communities, investors need to re-evaluate the purpose of production together with local decision makers and communities while also considering crop water demand to minimize negative trade-offs between water users and ecosystems.
Materials and Methods
Data on Land Acquisitions.
We used the collection of large-scale (>200 ha) land deals from Land Matrix (Retrieved in July 2014; www.landmatrix.org/en/). The Land Matrix database contained a total of 1,795 land-deals with an emphasis on food, fuel, and forestry crops. Of these, 747 deals were contracted in Africa, of which 121 were currently in production (Dataset S1). The dataset has geographical coordinates for each specific deal. We cross-referenced the Land Matrix data by observing the 54 largest land acquisitions in Google Earth, representing 95% of acquired land area (Fig. S2).
Fig. S2.
Simulation of Agricultural Production and Water Demand with LPJmL.
There is no information about the irrigation systems the investors use in the existing datasets, prompting the use of the LPJmL (36, 37) to estimate the green and blue water demand for seven different irrigation scenarios, at the site-specific locations given by land-deal coordinates. All scenarios include simulations of vegetation growth, phenology, and agricultural yield (SI Materials and Methods).
For LPJmL simulations, we assume that deals are managed intensively; that is, efficient pest and disease control, high-yielding varieties, mechanization, homogenous fields, and no nutrient limitations (see ref. 38 for details on the management parameters). Second, we assume that irrigation water is always available in the irrigated scenarios, if not locally, then by developing water infrastructure that would divert water from local or nonlocal sources. Finally, many crops that are grown on acquired land in Africa were not specifically parameterized. Instead, they were simulated by using a proxy crop with similar characteristics (SI Materials and Methods).
Acknowledgments
This study was supported in part by LUCID (lucid.lu.se.webbhotell.ldc.lu.se/), a Linnaeus Centre of Excellence at Lund University funded by the Swedish Research Council Formas (Grant 259-2008-1718), as well as two Formas-funded projects LUsTT (Land Use Today and Tomorrow) (Grant 211-2009-1682) and The Rush for Land in Africa (Contract 2012/7689). M.F. was supported by the Labex OT-Med (no ANR-11-LABX-0061) and the A*MIDEX Project 467 ANR-11-IDEX-0001-02.
References
1 CJ Vörösmarty, et al., Global threats to human water security and river biodiversity. Nature 467, 555–561 (2010).
2 FAO The Water–Energy–Food Nexus: A New Approach in Support of Food Security and Sustainable Agriculture (Food and Agriculture Organization of the United Nations, Rome, 2014).
3
P Steduto, J-M Faurès, J Hoogeveen, J Winpenny, J Burke Coping with Water Scarcity: An Action Framework for Agriculture and Food Scecurity (Food and Agriculture Organization of the United Nations, Rome, 2012).
4
; UNEP Vital Water Graphics: An Overview of the State of the World’s Fresh Marine Waters (United Nations Environment Programme, Nairobi, Kenya, 2008).
5
RS DeFries, JA Foley, GP Asner, Land-use choices: Balancing human needs and ecosystem function. Front Ecol Environ 2, 249–257 (2004).
6
W Steffen, et al., Sustainability. Planetary boundaries: Guiding human development on a changing planet. Science 347, 1259855 (2015).
7
M Falkenmark, J Rockstrom, The new blue and green water paradigm; breaking new ground for water resources planning and management. J Water Resour Plan Manage 132, 129–132 (2006).
8
S Rost, et al., Agricultural green and blue water consumption and its influence on the global water system. Water Resour Res 44, W09405 (2008).
9
W Anseeuw, et al., Transnational Land Deals for Agriculture in the Global South. Analytical Report based on the Land Matrix Database. (CDE/CIRAD/GIGA, Bern/Montpellier/Hamburg.). (2012).
10
P Woodhouse, New investment, old challenges. Land deals and the water constraint in African agriculture. J Peasant Stud 39, 777–794 (2012).
11
A Zoomers, Globalisation and the foreignisation of space: Seven processes driving the current global land grab. J Peasant Stud 37, 429–447 (2010).
12
A Mittal, F Mousseau, A Tajdin, D Farrell-Bryan, B Young Irresponsible Investment: Agrica’s Broken Development Model in Tanzania (Oakland Institute, Oakland, CA, 2015).
13
J Baxter, Who is benefitting? The social and economic impact of three large-scale land investments in Sierra Leone: A cost-benefit analysis. Report for the Action for Large-Scale Land Acquisition Transparency (ChristianAid, London). (2013).
14
M Curtis, R Mbunda, Take action: Stop EcoEnergy’s land grab in Bagamoyo, Tanzania. Available at www.actionaidusa.org/publications/take-action-stop-ecoenergys-land-grab-bagamoyo-tanzania. Accessed September 7, 2016. (2015).
15
O De Schutter, Large-scale land acquisitions and leases: A set of core principles and measures to address the human rights challenge. Briefing Note (UN Office of the High Commissioner for Human Rights, Geneva). (2009).
16
ND Mueller, et al., Closing yield gaps through nutrient and water management. Nature 490, 254–257 (2012).
17
L Cotula, et al., Testing claims about large land deals in Africa: Findings from a multi-country study. J Dev Stud 50, 903–925 (2014).
18
D Brautigam Will Africa Feed China? (Oxford Univ Press, New York, 2015).
19
C Smaller, H Mann, A thirst for distant lands: Foreign investment in agricultural land and water. Foreign Investment for Sustainable Development Program (International Institute for Sustainable Development, Winnipeg, Canada, 2009).
20
J Skinner, L Cotula Are Land Deals Driving ‘Water Grabs’? Briefing: The Global Land Rush (International Institute for Environment and Development, London, 2011).
21
SL Postel, GC Daily, PR Ehrlich, Human appropriation of renewable fresh water. Science 271, 785–788 (1996).
22
I Scoones, R Hall, SM Borras, B White, W Wolford, The politics of evidence: A response to Rulli and D’Odorico. J Peasant Stud 40, 911–912 (2013).
23
I Scoones, R Hall, SM Borras, B White, W Wolford, The politics of evidence: Methodologies for understanding the global land rush. J Peasant Stud 40, 469–483 (2013).
24
MC Rulli, A Saviori, P D’Odorico, Global land and water grabbing. Proc Natl Acad Sci USA 110, 892–897 (2013).
25
KW Deininger, D Byerlee Rising Global Interest in Farmland: Can It Yield Sustainable and Equitable Benefits? (World Bank Publications, Washington, DC, 2011).
26
M Smith, G Muñoz, JS Alvarez Irrigation Techniques for Small-Scale Farmers: Key Practivces for DRR Implementers (Food and Agriculture Organization of the United Nations, Rome, 2014).
27
ND Burgess, JDA Hales, TH Ricketts, E Dinerstein, Factoring species, non-species values and threats into biodiversity prioritisation across the ecoregions of Africa and its islands. Biol Conserv 127, 383–401 (2006).
28
C Megevand Deforestation Trends in the Congo Basin: Reconciling Economic Growth and Forest Protection (World Bank, Washington, DC, 2013).
29
L Feintrenie, Agro-industrial plantations in Central Africa, risks and opportunities. Biodivers Conserv 23, 1577–1589 (2014).
30
L Cotula Land Grab or Development Opportunity? Agricultural Investment and International Land Deals in Africa (IIED/FAO/IFAD, Buenos Aires, 2009).
31
L Cotula The Great African Land Grab? (Zed Books, London, 2013).
32
KF Davis, P D’Odorico, MC Rulli, Land grabbing: A preliminary quantification of economic impacts on rural livelihoods. Popul Environ 36, 180–192 (2014).
33
, eds KE Havnevik, PBE Matondi, AE Beyene (Zed Books/Nordiska Afrikainstitutet, London Biofuels, Land Grabbing and Food Security in Africa, 2011).
34
M Edelman, Messy hectares: Questions about the epistemology of land grabbing data. J Peasant Stud 40, 485–501 (2013).
35
C Oya, Methodological reflections on ‘land grab’ databases and the ‘land grab’ literature ‘rush’. J Peasant Stud 40, 503–520 (2013).
36
A Bondeau, et al., Modelling the role of agriculture for the 20th century global terrestrial carbon balance. Glob Change Biol 13, 679–706 (2007).
37
M Fader, W von Bloh, S Shi, A Bondeau, W Cramer, Modelling Mediterranean agro-ecosystems by including agricultural trees in the LPJmL model. Geosci Model Dev 8, 3545–3561 (2015).
38
M Fader, S Rost, C Müller, A Bondeau, D Gerten, Virtual water content of temperate cereals and maize: Present and potential future patterns. J Hydrol (Amst) 384, 218–231 (2010).
39
J Rohwer, D Gerten, W Lucht, Development of functional irrigation types for improved global crop modelling. PIK Report, ed Gerstengarbe, F-W (Potsdam, Germany). (2006).
40
H Biemans, et al., Impact of reservoirs on river discharge and irrigation water supply during the 20th century. Water Resour Res 47, W03509 (2011).
41
S Schaphoff, et al., Contribution of permafrost soils to the global carbon budget. Environ Res Lett 8, 014026 (2013).
42
J Elliott, et al., Constraints and potentials of future irrigation water availability on agricultural production under climate change. Proc Natl Acad Sci USA 111, 3239–3244 (2014).
43
M Fader, S Shi, W von Bloh, A Bondeau, W Cramer, Mediterranean irrigation under climate change: More efficient irrigation needed to compensate for increases in irrigation water requirements. Hydrol Earth Syst Sci 20, 953–973 (2016).
44 T Beringer, W Lucht, Nachhaltiges globales Bioenergiepotenzial. Commissioned expert study for the German Advisory Council on Global Change (WBGU) as a contribution to the flagship report World in Transition—Future Bioenergy and Sustainable Land Use (German Advisory Council on Global Change, London). (2008).
45 K Waha, L Van Bussel, C Müller, A Bondeau, Climate‐driven simulation of global crop sowing dates. Glob Ecol Biogeogr 21, 247–259 (2012).
46 Banana Planters, Banana Cultivation Guide. Available at mahaprisons.gov.in/Uploads/Dockets_Files/635259935664912504Banana_Cultivation_Guide_%C2%AB_Banana_Planters.pdf, Accessed September 8, 2016. (2009).
47 J van Velzen, Estimation of the Potential and Actual Yield of Banana in the Atlantic Zone of Costa Rica (Atlantic Zone Programme, Turrialba, Costa Rica), Field Rep 136. (1994).
48 RL Snyder, JP Melo-Abreu, Frost Protection: Fundamentals, Practice and Economics, Vol 1. (Food and Agriculture Organization of the United Nations, Rome). (2005).
49 S Sitch, et al., Evaluation of ecosystem dynamics, plant geography and terrestrial carbon cycling in the LPJ dynamic global vegetation model. Glob Change Biol 9, 161–185 (2003).
Reference
https://s7.addthis.com/static/sh.f48a1a04fe8dbf021b4cda1d.html#rand=0.5942748691603401&iit=1670928249515&tmr=load%3D1670928249250%26core%3D1670928249383%26main%3D1670928249487%26ifr%3D1670928249560&cb=0&cdn=0&md=0&kw=land%20grabbing%2Cwater%20scarcity%2CLPJmL%2Cwater%20footprints%2Cirrigation&ab=-&dh=www.pnas.org&dr=http%3A%2F%2Fm.facebook.com%2F&du=https%3A%2F%2Fwww.pnas.org%2Fdoi%2F10.1073%2Fpnas.1524741113&href=https%3A%2F%2Fwww.pnas.org%2Fdoi%2F10.1073%2Fpnas.1524741113&dt=Green%20and%20blue%20water%20demand%20from%20large-scale%20land%20acquisitions%20in%20Africa%20%7C%20Proceedings%20of%20the%20National%20Academy%20of%20Sciences&dbg=0&cap=tc%3D0%26ab%3D0&inst=1&jsl=1&prod=undefined&lng=en&ogt=description%2Csite_name%2Cimage%2Curl%2Ctype%3DArticle%2Ctitle&pc=men&pub=xa-4faab26f2cff13a7&ssl=1&sid=63985779cfe66e14&srf=0.01&ver=300&xck=0&xtr=0&og=title%3DGreen%2520and%2520blue%2520water%2520demand%2520from%2520large-scale%2520land%2520acquisitions%2520in%2520Africa%2520%257C%2520Proceedings%2520of%2520the%2520National%2520Academy%2520of%2520Sciences%26type%3DArticle%26url%3Dhttps%253A%252F%252Fwww.pnas.org%252Fdoi%252Fabs%252F10.1073%252Fpnas.1524741113%26image%3Dhttps%253A%252F%252Fwww.pnas.org%252Fcms%252Fasset%252Fbcac357b-ffbf-4a1c-8ab3-330205214686%252Fpnas.2016.113.issue-41.largecover.png%26site_name%3DPNAS%26description%3DIn%2520the%2520last%2520decade%252C%2520more%2520than%252022%2520million%2520ha%2520of%2520land%2520have%2520been%2520contracted%2520to%2520large-scale%250Aland%2520acquisitions%2520in%2520Africa%252C%2520leading%2520to%2520increased%2520pressures…&csi=undefined&rev=v8.28.8-wp&ct=0&xld=1&xd=1
