Hydrological sustainability of international virtual water trade

International virtual water (VW) trade helps to balance water stress across regions. However, it can be questioned whether such trade can remain sustainable as water resources are redistributed across regions resulting from changes in our climate. A conceptual framework to compare VW trade volumes with water fluxes within the water cycle is introduced. We evaluate the distribution of traded water surpluses and deficits associated with crop, animal, and industrial products over 157 countries and 182 global watersheds. About 7% of the countries are identified to conduct VW trade unsustainably. Regions within Africa, North America, central Asia, and Europe exhibit unfeasible VW trading resulting from higher appropriation of freshwater resources than availability influenced by precipitation and evaporation. Assessment at the watershed scale captures overexploitation at finer resolution, generally overlooked in country level analysis. An evaluation into the future reveals more watersheds becoming vulnerable to water storage depletion under future climate trends.


Introduction
Since industrialisation, anthropogenic interactions with the global water cycle have modified water flows across both space and time [1][2][3].This is not limited to physical transfer of water stocks, but extendable also to the hidden movement of water through trade, a concept referred to as virtual water (VW) flow [4][5][6][7][8][9].VW represents the volume of water theoretically embedded within goods.While studies have highlighted the existence of global VW systems with flows comparable to real water fluxes within the hydrological cycle [10], the question of whether such systems are sustainable (especially when both virtual and actual water volumes are changing under climate change) remains underexplored.With the establishment of international trading links, VW trading volumes actually doubled, increasing by more than 50% during the period of 1986-2008 [11][12][13].A similar trend with increasing VW trade volumes is expected in the present century [14].The global VW trade is supported by localised water flows available from the water cycle, as illustrated in figure 1.The appropriation of water, more than its replenishment, depletes both surface and sub-surface water storage, thus comparing the real and VW fluxes at an annual scale is necessary to ensure water resources for the future remain accessible and utilisable across the planet.
The global water trade reduces environmental costs (in terms of land and water savings and pollution reduction) of producing goods locally [15].For example, more than 100 billion cubic meters (BCM) of water related to agriculture and animal husbandry was saved globally in the year 2008 [16].However, international trade has led some regions to become deeply dependent on imported resources [17].Importing water intensive products provides relief to physical water stress, especially in arid regions [4,18,19].However, meeting the collective demands of the population can create pressure on critical hydrological reserves for regions supplying these products [7,20].For instance, the United States, China, India, and Australia are major VW Figure 1.IVW flows and the hydrological cycle.Pictorial depiction of relation between utilisation of water for international VW trade and the hydrologic cycle.VW trade with crop, animal and industrial products is supported by availability of real water primarily in the surface and subsurface of earth influenced by the water cycle.With an increasing population, escalation of VW exports is imminent, leading to unsustainable exploitation of these water resources.Major factors impacting the status of unsustainability are mentioned within circles representing sets of systems (countries or watersheds) conducting unsustainable exports of blue, green, and grey VW.Systems landing within the intersection regions are intensively exploiting water reserves.
exporters that are partly under water stress [21,22], a situation undergoing more complex change as the world warms and water demands increase with time.This raises questions about the sustainability of water management to satisfy external water demands.Previous studies have mostly focused on evaluating the utilisation of surface water, non-renewable ground water and soil moisture for crops [23][24][25][26][27] and dairy products [28], and highlighted overexploitation of these resources via incorporation of hydrological, agricultural, and earth observation system models [29][30][31][32][33][34].Therefore, a comprehensive evaluation of the sustainability of VW trading by combining animal and industrial products along with crop products (CP) is imperative.
VW trading encompasses three components, namely green, blue, and grey water.A green water footprint is the consumption of precipitation (P) that evaporates (including transpiration) back to the atmosphere.A blue water footprint defines the surface and ground water consumed in the form of evaporation or direct incorporation into products.A grey water footprint is the amount of water required to dilute polluted water based on water quality standards.Regions with limited freshwater availability that fulfil VW exports are likely to face sustainability challenges in these three categories.Moreover, systems with riverine ecosystems require adequate freshwater to maintain ecological balance and human health [33,35].This indicates the need to ensure environmental flows (EF B ) that are, theoretically, unavailable for human consumption (relevant to blue and grey water).Similarly, green environmental flow (EF G ) denotes green water required to support natural ecosystems, which creates a limit on the permissible green water availability [36] that can be translated to limited land availability for agriculture and product manufacturing [37].A system with fewer land options might breach this constraint (e.g.appropriation of forest lands) and witness unsustainable green water exports.Furthermore, immoral socioeconomic policies and demands exacerbate each of these scenarios.
Overall, water sustainability can be assessed using indices defining socio-economic, water scarcity and environmental aspects [19,[38][39][40][41][42][43][44][45] (the watershed sustainability index and the economic water scarcity index to name a few) that are data intensive and require mostly continuous spatial and temporal data, or adopting a comprehensive study that requires knowledge and understanding on uncertainties involved within production steps, such as the life cycle analysis [46,47].The present study introduces a framework that examines sustainability of international virtual water (IVW) flows associated with 146 crops and 200 derived-CP [48], animal products (AP) [49], and industrial products (IP; single category) during 1996-2005 obtained from a previous study [50] against real water flows derivable from the water cycle.Unlike traditional data driven approaches, we intend to compare water trade amounts to hydrological water fluxes, highlighting the extent of water appropriation by human interference.The concept is extended to watersheds around the globe with rational assumptions in defining their trade and economic status.Implications of historical trade practices in a future scenario are similarly performed, implementing climatic shifts derived from the IPCC report findings.Furthermore, table A1 in the appendix lists all abbreviations used in the present study.

Sustainable-vulnerable space
A conceptual framework is established that places a system according to its VW export and import magnitudes and local restricted water availability (RWA) (figure 2(a)).RWA is defined as the maximum water available for human appropriation without interfering with ecosystem services.Hydrological flows within the hydrological cycle are stored on the surface as lakes, reservoirs, and rivers and within the sub-surface as soil moisture and groundwater, thus recognised as water availability (figure 2(b)).Any appropriation higher than RWA makes the system vulnerable.
The study was conducted with 5 classes of sustainability assessments, individually addressing green, blue (further divided into 2 classes), grey, and a combination of grey and blue VW.In quantifying VW of CP and IP, evapotranspiration (ET) demands of primary crops and animal feed materials maintained by the P that infiltrates into soil and persists in root zone layer (green water) and water abstracted from the surface and ground water (blue water) for irrigation purposes (figure 1), is already considered [48,51].Therefore, total green VW and blue VW exports (Blue1) associated with CP and AP in the form of ET are compared with ET volumes after deductions for EF G (referred to as effective ET; figure 2(b)).Presenting effective ET as RWA highlights what proportions of it are appropriated for human use.Total grey VW exports associated to CP, AP, and IP, and blue VW exports (blue2) linked with AP (since blue water is also used in animal husbandry) and IP are compared with the difference between mean annual P and ET (runoff in hydrologic sense) after deductions for EF B (i.e.P-ET-EF B ), hereby denoted as effective P (figure 2(b)).Furthermore, total VW exports are modified by subtracting green and parts of blue VW (related to ET) contents associated with CP and IP from total water exports, while comparing with effective P. The modified export volume, referred to as effective export, is the total grey and partial blue VW (figure 2(b)).
The framework displays the relationship between export volumes and RWA, both normalised with respective VW imports.Green, blue, and grey VW exports and corresponding water availabilities are normalised with green, blue, and grey VW imports respectively.In the case of effective VW exports, total VW imports are considered.The space within the framework is divided into four sections using two limits described as the water appropriation line (WA-line) and the water trade balance line (WTBline).The WTB-line is equivalent to exports being equal to imports.Therefore, it demarcates regions with exports more (less) than import volumes above (below) the line.The WA-line demarcates regions based on the extent of exports relative to permissible water availability.Here, the distance of a point from the WA-line is defined as the water allowance, suggesting a surplus/shortage of water after export.Sections I and II represent a sustainable state, describing sustainable usage of water reserves since exports are lower than restricted limits of water availability.Sections III and IV describe vulnerability as appropriation of water is higher than restricted limits.The overall space is referred to henceforth as the Sustainability-Vulnerability (SV) space.A Net water importer within section IV of SV space would pose a serious threat to hydro-socio-ecological stability as the system is already dependent on foreign water, yet exports domestic water at a higher rate than its replenishment.

IVW trade and meteorological data
The IVW volumes incorporated in the study are retrieved from the report [50], which discusses the VW flows in CP, AP, and IP for the period of 1996-2005.It includes the water footprint of 146 crops and crop-derived products at a higher resolution of 5 by 5 arc minute grid [48] along with different farm AP constituting beef cattle, dairy cattle, pigs, sheep, goats, broiler chicken, layer chicken, and horses, considering different production systems [49].The water footprint per dollar of IP is calculated per country by dividing the total national industrial water footprint by the value added to the industrial sector obtained from the United Nations Statistical Divisions database [7].
The gross domestic product (GNI) per capita of countries is obtained from the World Bank database (https://data.worldbank.org/indicator/NY.GNP.PCAP.CD).Global watersheds are acquired from the WRI Major Watersheds of the World Delineation layer from the FAO-GeoNetwork.A further selection of watersheds and calculations of distributed IVW flows, global maps of croplands and pasturelands for the year circa 2000, were prepared using the agricultural inventory data of 57 countries and two satellite derived products, i.e.Moderate Resolution Imaging Spectrometer (MODIS) land cover and Global Land Cover 2000 (GLC2000).The maps are retrieved from the Socioeconomic Data and Applications Centre (SEDAC) [52].The global human footprint (GHF) dataset is used to assess distributed import volumes.GHF is constructed using the Human Influence Index (created from nine global datasets including population density, land use/land cover, human access datasets, etc) normalized by biome and realm.The gridded map can be accessed from SEDAC [53].For blue water environmental flow calculations, a gridded map of environmental flow percentage for class A is accessed from the International Water Management Institute (IWMI) [54].
Monthly P, ET, and potential evapotranspiration (PET) datasets are obtained from ERA5-Land reanalysis of the European Centre for Medium-Range Weather Forecasts (ECMWF) Integrated Forecasting System.The datasets are provided at a spatial resolution of 0.1 × 0.1 degrees (9 km on a reduced Gaussian grid) and subsequently converted to a mean annual scale for conducting the study.

Assessment of environmental flows
EF B represents the water resources required to maintain riverine ecosystem health.This is calculated by producing a gridded map of environmental flows using spatial datasets of environmental flow percentages (percentage of annual flow required; class A selected indicating minor modifications where ecosystem services are intact despite minimal disturbances) provided by IWMI [54] (supplementary text 1) and natural annual runoff considered as P-ET (devoid of direct anthropogenic interference such as dam development, etc) constructed with gridded maps of P and ET and extracting for countries and watersheds.Here, the map is resampled to the constructed maps of P-ET of 0.099 × 0.099 degrees.
Similarly, EF G represents the green water flows needed to support natural ecosystems that produce biomass to sustain the economy.This can be attributed to restricted use of lands (e.g.lands with protected status) to support natural terrestrial ecosystems.Global assessments documented about 22% of green water flow emerges from biodiversity conservation areas and 17% is from lands inaccessible for human use due to unfavourable topographical or climatic conditions [37].Locally, permissible maximum sustainable green water would vary across regions and thus, for simplicity, 30% of ET for respective countries or watersheds are reserved for nature.

Defining VW volumes
Total IVW exports (W exp ) can be defined as follows: Here, W CP expG , W CP expB, and W CP expGy are green, blue, and grey VW related to exported CP.Similarly, W AP expG , W AP expB, and W AP expGy are green, blue, and grey VW related to exported AP.W IP expB and W IP expGy are blue and grey VW related to exported IP.International exported VW of CP and AP include green water and parts of total blue water that support ET.During the cultivation of crops, green water available from P and stored in the root zone is utilised.By means of irrigation, blue water abstracted from surface and ground water reservoirs supports transpiration of crops and animal feed (involving various crops).Thus, while comparing water flux (after meeting ET and EF B demands) for a country or watershed with water exports, it is necessary to deduct crop ET losses from export amounts.After the deduction, the remainder is referred to as effective export.For the aforementioned reasons, 70% of exported blue water in AP is included in the deduction.Effective export is calculated as follows: ) . ( Here, W * exp is effective export.The blue water included in the deduction for the calculation of effective export is referred to as ET portion of blue water exports (W expB1 ) and expressed as follows: The rest of blue water exports include water utilised for servicing and drinking purposes, feed mixing of cattle, and manufacturing of IP.This is denoted as runoff portion of blue water exports and represented as: Thus, total blue VW export can be stated as: Contrary, total grey and green VW are simply defined as follows: In a similar fashion, W imp is total VW volumes imported expressed as: Here, W CP impG , W CP impB, and W CP impGy are green, blue, and grey VW related to imported CP.Similarly, W AP impG , W AP impB, and W AP impGy are green, blue, and grey VW related to imported AP.W IP impB and W IP impGy are blue and grey VW related to imported IP.
Total blue, grey, and green VW are defined as follows:

Calculation of water trade magnitudes for watersheds
In preparation of VW flows datasets for watersheds, IVW volumes are disaggregated at grid scale using raster datasets; cropland and pastureland raster for VW involved in exported CP and AP respectively, and GHF raster for VW import and exported IP.A similar approach is adopted for assessing deductions for CP and IP in calculating effective exports.The GHF raster is resampled to 0.083 × 0.083 degrees and rescaled (normalised) to 0-1 from 0-100 before incorporating in the assessment to be consistent with cropland and pastureland maps.A common weightage factor is associated with each grid value encircled within a country boundary, such that the summation of ensuing grid values constitutes the country's VW volumes.For instance, if there are N grids (x 1 , x 2 , x 3 ….xN ) within the boundary of a particular country with Y VW volume, then the weightage factor (β) is quantified as: Following the construction of the gridded datasets, IVW volumes for watersheds are extracted and expressed in heights as follows: Here, subscripts 'imp' and 'exp' are water imported and exported (CP, AP, and IP) for the watershed C. Notation x c denotes the grids inside the watershed boundary.

Reliability-resilience-vulnerability (RRV)
The well-known RRV concept [55] is applied to examine the all-round performance of watersheds while maintaining IVW flows under changing climatic conditions [56][57][58][59][60]. A moving window of 10 years is considered to perform the analysis.In the context of the present study, reliability (Rel) indicates the probability of the watershed to remain in a sustainable state; resilience (Res) indicates how quickly the watershed recovers from a vulnerable state; vulnerability * (Vul) indicates the severity of damage caused to the watershed when approaching within a vulnerable state ( * represents the index and not the general vulnerability idea discussed here).A watershed coming into a sustainable (vulnerable) state is considered a success (S) (failure; F).
Considering X t as the point position for tth window in the conceptual space, reliability is assessed in a way that if X t ϵ S then Z t = 1, else 0. Rel is expressed as: Here, N denotes the total number of moving windows, which is 22 in this case.
Similarly, resilience is assessed in such a way that if X t ϵ F and X t+1 ϵ S then W t = 1, else 0. Res is expressed as: Let M represent the number of times a watershed falls in vulnerable state, then vulnerability is calculated as: A is water allowance which is defined as the distance of the point at any time t from the WA-line.Figure 3 shows the complete flowchart of steps undertaken before the RRV analysis.

Relationship between water trade and the hydrological cycle
Datasets demonstrate that countries tend to import VW more to alleviate stress on limited real water reserves (figure 4).Despite the intuitive idea of water savings via the VW trade, studies have highlighted various drivers of VW trade other than water endowments, such as GDP, human population, agricultural land, economic land use productivity, government policies, etc [61,62].Oceania receives spatially variable P with average annual P of Australia and New Zealand (1041.6 mm yr −1 ) lower than Polynesia, Micronesia, and Melanesia (2581.5 mm yr −1 ).Despite this, Australia and New Zealand export more VW, which is explainable from their economic deregulation reforms [63,64] and exports of AP with higher VW footprints than most CP [51].Australia is one of the highest exporters of beef [65] which contains water footprints per ton (m 3 /ton) approximately 15 and 50 times more than fruits and vegetables respectively.New Zealand and Australia are major exporters of dairy products [66,67] with high VW footprints.We also observed an increase in VW trade volumes per cap GNI, although considerable variance is noted (supplementary text 2 and supplementary figure 1).Considering other hydrologic parameters would help explain the intricacies of the relationship between hydrological and VW cycles.Nevertheless, VW trade entails an efficient consumption of water resources that offers a viable solution to the global food crisis and socio-economic security.

Sustainable water trade for countries and watersheds
The SV space is divided into 4 zones depending on water allowance; zone-1: 0-1 mm mm −1 , zone-2: 1-10 mm mm −1 , zone-3: 10-100 mm mm −1 , zone-4: >100 mm mm −1 .Zone-1 is the nearest region to WA-line; thus, countries or watersheds within this zone are susceptible to vulnerability.Around 7% (11/157) of all the considered countries are exploiting national water reserves in an unsustainable manner, casting them in a vulnerable state (negative water allowance; figure 5, supplementary figure 2).Singapore, the Netherlands, Belgium, and Mauritius (all are net VW importers, i.e. total VW imports are higher than total VW exports) appeared in the vulnerable state for the green VW trade (all in section IV), explainable with adequate croplands and pastureland proportions within the countries (Singapore: 34% with one of the highest EF G ; the Netherlands: 58%; Belgium: 46%; obtained from croplands and pasturelands gridded maps [52]).These countries are also conducting grey VW exports unsustainably (figure 5(e)).Similarly, the high consumption of blue water as ET for agriculture and animal feed cast Singapore in a vulnerable state for the blue VW trade (section III; figure 5(a)).Ten countries appeared in the vulnerable state for exporting grey VW, of which four countries are also conducting unsustainable blue VW trade related to industrial exports and animal husbandry (figure 5(d) and (e)).Distribution of total water trade volumes exported and imported per year during the same time period.The y-axis is presented in a log scale.Low P in Southern Africa, North Africa, and Western Asia resulted in importing more VW than its export.However, Australia and New Zealand export more than other regions in Oceania even with low P due to land availability and productivity, technological advancements, and socio-economic policies.Points for Australia and New Zealand are denoted with 'x' and 'o' respectively to highlight differences in comparison to other countries in Oceania.This is due to higher environmental flow demands required to sustain riverine ecosystems, which led to less RWA for consumption.For effective export (total grey VW and blue VW modified for ET), all the mentioned countries (ten) and Moldova appeared to be in a vulnerable state.Out of the 11 countries, 4 are net VW exporters (Hungary, Kazakhstan, Mali, and Moldova).Water allowances of countries within Europe and Asia are low compared to other continents (figures 6(a)-(e); supplementary table 1) due to high traded water volumes (Asia and Europe are the top two continents in exporting green, blue and grey VW, exporting 862.8 and 123.5 mm/year/country of average total VW respectively) and lower RWA (high average environmental flow fractions of 63% and 51% respectively for Europe and Asia).This is expected as Asia is home to rich and diverse riverine ecosystems that need water to sustain them.We observed low green water allowance in North America due to huge exports of green water as CP and AP (average green water exported: 50.5 mm/year/country, the highest after Asia and Europe).No countries from South America, North America, and Oceania have VW exports higher than RWA (supplementary table 2); however, North America and Oceania have countries appearing in zone-1 of the SV space (supplementary table 3).
Major exporters and importers of VW are mostly in a sustainable state.Figure 7 displays water availability surplus/deficits (in BCM) after exports.Positive magnitudes (water surplus) mean limited usage of real water reserves.Conversely, higher appropriation than RWA follows water deficit.The United States, Pakistan, India, Australia, China, and Brazil are the major exporters of VW (net exporters) that appeared in section I or II of the SV space for all categories and, thus, have surplus water.Major net VW importers like Japan, Germany, Italy, France, and Mexico are also in section I of the sustainable state (all categories).However, net VW importers like Belgium, the Netherlands, and Singapore and net VW exporters, namely Hungary and Kazakhstan, are in section III or IV (vulnerable) of the SV space for effective, grey, and green VW categories and effective, blue2, and grey VW categories respectively.Focussing on rainwater stressed regions (figure 4), no country falls into a vulnerable state.However, Azerbaijan and Yemen from Western Asia, and South Africa and Swaziland from Southern Africa fall into zone-1 of the SV space for effective exports, with Morocco (Northern Africa) falling into zone-1 for effective and grey water exports.Moreover, Lebanon (Western Asia) falls into zone-1 for effective, blue2, grey, and green water exports (supplementary table 4).All are net VW importers.Most of the countries within regions of low P are net importers of VW (85%; 11/13) highlighting external water dependency due to lower real water availability.
The study is extended to the watershed scale to observe the impacts of IVW flows at a sub-country level.It is ensured that the watersheds are large (> 10 000 km 2 ) and at-least 10% of the total area contains crop or pastureland (supplementary text 3).In total, only 2 watersheds are in vulnerable state from Europe (figures 8 and 9, supplementary table 5, supplementary figure 3).However, supplementary table 6 shows 29 watersheds appearing in zone-1, of which 27, 21, 13, 1, and 1 are in effective, grey, blue2, blue1, and green water categories respectively.Only 42% (16/38), 23% (7/30), 23% (5/22), 0% (0/3), and 40% (2/5) of the overlying countries appeared in vulnerable or zone-1 states during country-scale analysis, highlighting additional evidence of adverse trading practices at the sub-country level.Countries like the United States, India, China, Australia, Russia, Mexico, Turkey, etc. emerged in sustainable zone-2, 3, or 4 (all categories) in country-scale analysis but hold watersheds appearing in vulnerable or zone-1 (for at-least one category; figures 9(a), (b) and (d)).We observe a similar pattern for groundwater, river, and lake depletion displayed in figure 1 from [27], when compared with effective water allowance maps, particularly in the United States, south-western Australia, Southern India, Mexico, Spain, and Iraq.Watersheds within Western Europe and Central Asia have shown low water allowance magnitudes, conveying similar outcomes as in country-scale studies.Moreover, the combined outflow of water (ET and effective export) from countries/watersheds compared to PET (potential atmospheric demand of ET) is less, except Singapore (supplementary text 4 and supplementary figure 4).To examine the implications of historical water trading practices (1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005) for watersheds against climatic variability (water availability), RRV indices are incorporated [55] by taking a 10 year moving window across 1990-2020 period with constant import/export volumes.It is conducted with only effective VW exports where maximum watersheds appeared in zone-1 and vulnerable state during 1996-2005 period.EF B magnitudes changed with P-ET magnitudes (keeping EF fractions unchanged), whereas EF G changed depending on ET (30% of ET).Twelve watersheds encountered a vulnerable state at least once during the RRV analysis (supplementary table 7), of which eight are from Asia and Europe (four each).Africa, South America, North America, and Oceania have 2, 1, 1, and 1 watershed respectively.For the future Representative Concentration Pathway 8.5 (RCP8.5)emission scenario, 16 watersheds (supplementary table 8; Increased in Europe, Asia, and North America with 6, 5, and 4 watersheds respectively) appeared in a vulnerable state at least once (increased from 12 at historical time), although water availability increased for 63% of the watersheds, also suggested in the previous study [70].Details are provided in supplementary text 5 and supplementary figure 5.

Uncertainties
The present assessment inherits uncertainties on the quantification of water footprints of products.A major assumption in assessing water footprints is that the origin of products is traced by one step, i. e. importing a product from a country means the product is manufactured in that country [7].Although tedious, recent studies have used bilateral trade statistics to consider water utilised from its production source to consumption destination [71,72].We assume 30% of ET for EF G calculations for simplicity.Disaggregating VW trade for watershed-scale analysis is based upon various underlying assumptions.The present study assumed appropriation of real water to produce CP and IP occurred at locations of crop and pasturelands respectively.Similarly, it is assumed that the GHF represents the appropriation of water for industrial goods and also defines demand for import volumes.Nevertheless, any finer information on import/export volumes (which would require localised manufacturing and demand information) will not significantly impact the present sustainability assessments.

Conclusions
About one-fifth of the global water footprint is exported through trade [7].For water stressed regions, external water dependency is inevitable.However, technological advancements and foreign policies, etc. influence the correlation between trading patterns and domestic water availability.We present a comparison of VW export and import volumes to water availability for countries and global watersheds.Countries and watersheds, mostly within Africa, Central Asia, and Europe, are prone to vulnerabilities pertaining to unsustainable trading practices.We acknowledge the relatively shorter time-length (10 years) of IVW flows used for the assessments, particularly, in RRV analysis of watersheds in historical and future scenarios where, additionally, VW trading volumes are kept unchanged; thus, we expect an underestimation of the implications of trading since population growth and an increase in demand are evident.Furthermore, RRV analysis of global watersheds with current and future climatic scenarios (RCP8.5)reinforced the idea of adverse implications of inappropriate IVW flows.
Our study discusses the global crisis in the availability of limited freshwater reserves due to problems of insensitive human appropriation.Intra-national VW flows will further expand this essence of water sustainability at the sub-country level.

Figure 2 .
Figure 2. Assessing sustainability of IVW flows.(a) Conceptual diagram of the relationship of water trade flows with hydrological water fluxes.Sections I-II represent a sustainable state and III-IV represent vulnerable states.For a system within the SV space, increasing water imports (via water intensive products) and lowering export volumes would trigger a downward shift, approaching the sustainable state.(b) Comparison of VW trade flows against restricted water availability derived from the mean annual water balance and assessed differently depending on its comparison with export water volumes.ET-EFG is effective ET and presented in the SV space as availability, merely to highlight the appropriation of ET flow translated as green and blue VW into CP and IP through agriculture (rainfed and irrigation) and animal feed.Effective P (P-ET-EFB) signifies water available for animal husbandry (nursing and food mixing), industrial manufacturing, and dilution of waste (grey water) generated from these crop, animal, and industrial products.Effective VW describes the portion of total VW devoid of ET flows into the atmosphere, thus compared with effective P.

Figure 3 .
Figure3.Flowchart of the steps performed in achieving the primary objective of assessing hydrological sustainability of IVW flows.The chart displays the incorporation of climatic datasets to assess environmental flows and calculations of VW trade volumes for watersheds and countries that eventually feed into the SV space.The colour fills in the boxes denote the type of water (blue, green, or grey).RRV analysis is then conducted using the SV space with historical and climatic variabilities in examining various aspects of sustainability.

Figure 4 .
Figure 4. Virtual water trade and Precipitation.(a) Distribution of mean annual P and P after accounting for environmental flows across landmasses for the period of 1996-2005.It is noticed that annual P (average) for Southern Africa (688.5 mm yr −1 ), Northern Africa (172.4 mm yr −1 ), and Western Asia (374.2 mm yr −1 ) are low compared to other parts of the world.(b) Distribution of total water trade volumes exported and imported per year during the same time period.The y-axis is presented in a log scale.Low P in Southern Africa, North Africa, and Western Asia resulted in importing more VW than its export.However, Australia and New Zealand export more than other regions in Oceania even with low P due to land availability and productivity, technological advancements, and socio-economic policies.Points for Australia and New Zealand are denoted with 'x' and 'o' respectively to highlight differences in comparison to other countries in Oceania.

Figure 5 .
Figure 5. Relationship between IVW and water availability for countries.(a)-(e) Scatter plot of 157 countries in the conceptual SV space for blue1 (a), green (b), effective (c), blue2 (d), and grey (e) virtual water during the period 1996-2005.The SV space (sustainable region) is divided into 4 zones (zone-1, zone-2, zone-3, and zone-4) depending on the distance of the points from the WA-Line (defined as water allowance).Countries mentioned within the circles (at the back) are conducting trades unsustainably as export volumes exceed restricted water availability (negative water allowance in the SV space).Countries appearing within the intersections of the circles exhibit unsustainable trading practices in more than one category.Countries conducting unsustainable effective VW trades are shown separately.The Net flow of VW considering total VW export and import volumes (Importer/Exporter) is denoted by shape types within the SV space.The x-axis is presented in a logarithmic scale (Thus curved WA-line).

Figure 6 .
Figure 6.Water allowance among countries.(a)-(e) Spatial map of 157 countries with water allowance for effective (a), blue2 (b), blue1(c), grey (d), and green (e) virtual water exports during the period 1996-2005.The majority of countries in Western Asia and Northern Africa are not considered in the analysis because of negative magnitudes of P-ET at least once during the 10-year moving windows taken from 1990 to 2020 and displayed with solid grey fill.This generally happens when non-traditional sources of water are exploited (like sea water desalination) that intensify water evaporation[68,69].The colour codes represent vulnerability (colour red) and different zones depending on water allowances within the sustainability segment of the SV space.

Figure 7 .
Figure 7. Balance of sustainable water trade.(a)-(e) Bar plot of major VW importing and exporting countries having a positive (water surplus; sustainable) and negative (water deficit; vulnerable) balance of sustainable effective (a), blue2 (b), blue1 (c), grey (d), and green (e) water trades.The water volumes are assessed with water allowance and corresponding import magnitudes as per the VW export type and expressed in billion cubic meters (BCM).

Figure 8 .
Figure 8. Relationship between IVW and water availability for watersheds.(a)-(e) Scatter plot of 182 watersheds in the conceptual SV space for blue1 (a), green (b), effective (c), blue2 (d), and grey (e) virtual water during the period 1996-2005.The SV space (sustainable region) is divided into 4 zones (zone-1, zone-2, zone-3, and zone-4) depending on the distance of the points from the WA-Line (defined as water allowance).Watersheds mentioned within the circles (at the back) are conducting trades unsustainably as export volumes exceed restricted water availability (negative water allowance in the SV space).No watershed appeared within the intersections of the circles that represent unsustainable trading practices in more than one category.Watersheds conducting unsustainable effective VW trades are shown separately.The Net flow of water (Importer/Exporter) is denoted by shape types within the SV space.The x-axis is presented in a logarithmic scale (Thus curved WA-line).

Figure 9 .
Figure 9. Water allowance among watersheds.(a)-(e) Spatial map of 182 watersheds with water allowance for effective (a), blue2 (b), blue1 (c), grey (d), and green (e) virtual water exports during the period 1996-2005.The colour codes represent vulnerability (colour red) and different zones depending on water allowances within the sustainability segment of the SV space.