REVIEW SUMMARY

SUSTAINABILITY

Systems integration forglobal sustainabilityJianguo Liu Harold Mooney Vanessa Hull Steven J Davis Joanne GaskellThomas Hertel Jane Lubchenco Karen C Seto Peter GleickClaire Kremen Shuxin Li

BACKGROUND Many key global sustain-ability challenges are closely intertwined (ex-amples are provided in the figure) Thesechallenges include air pollution biodiver-sity loss climate change energy and food sec-urity disease spread species invasion andwater shortages and pollution They are inter-connected across three dimensions (orga-nizational levels space and time) but areoften separately studied and managed Sys-tems integrationmdashholistic approaches to in-tegrating various components of coupled

human and natural systems (for example social-ecological systems and human-environmentsystems) across all dimensionsmdashis necessa-ry to address complex interconnections andidentify effective solutions to sustainabilitychallenges

ADVANCES One major advance has beenrecognizing Earth as a large coupled humanand natural system consisting ofmany smallercoupled systems linked through flows of in-formation matter and energy and evolv-

ing through time as a set of interconnectedcomplex adaptive systems A number of in-fluential integrated frameworks (such as eco-system services environmental footprintshuman-nature nexus planetary boundariesand telecoupling) and tools for systems in-tegration have been developed and testedthrough interdisciplinary and transdiscipli-

nary inquiries Systemsintegration has led to fun-damental discoveries andsustainability actions thatare not possible by usingconventional disciplinaryreductionist and compart-

mentalized approaches These include find-ings on emergent properties and complexityinterconnections among multiple key issues(such as air climate energy food land andwater) assessment of multiple often conflict-ing objectives and synergistic interactionsin which for example economic efficiency canbe enhanced while environmental impacts aremitigated In addition systems integrationallows for clarification and reassignment ofenvironmental responsibilities (for exampleamong producers consumers and traders)mediation of trade-offs and enhancement ofsynergies reduction of conflicts and design ofharmonious conservation and developmentpolicies and practices

OUTLOOK Although some studies have rec-ognized spillover effects (effects spilling overfrom interactions among other systems) orspatial externalities there is a need to simul-taneously consider socioeconomic and envi-ronmental effects rather than consideringthem separately Furthermore identifyingcauses agents and flows behind the spill-over effects can help us to understand betterand hence manage the effects across multi-ple systems and scales Integrating spilloversystems with sending and receiving systemsthrough network analysis and other advancedanalytical methods can uncover hidden in-terrelationships and lead to important in-sights Human-nature feedbacks includingspatial feedbacks (such as those among send-ing receiving and spillover systems) are thecore elements of coupled systems and thusare likely to play important roles in globalsustainability Systems integration for glob-al sustainability is poised for more rapiddevelopment and transformative changesaimed at connecting disciplinary silos areneeded to sustain an increasingly telecoupledworld

RESEARCH

SCIENCE sciencemagorg 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 963

ON OUR WEB SITE

Read the full articleat httpdxdoiorg101126science1258832

The list of author affiliations is available in the full article onlineCorresponding author E-mail liujimsueduCite this article as J Liu et al Science 347 1258832 (2015)DOI 101126science1258832

Integratinghuman and natural

systems

Illustrative representation of systems integration Among Brazil China the Caribbean and theSahara Desert in Africa there are complex human-nature interactions across space time andorganizational levels Deforestation in Brazil due to soybean production provides food for people andlivestock in China Food trade between Brazil and China also contributes to changes in the global foodmarket which affects other areas around the world including the Caribbean and Africa that alsoengage in trade with China and Brazil Dust particles from the Sahara Desert in Africamdashaggravated byagricultural practicesmdashtravel via the air to the Caribbeanwhere they contribute to the decline in coralreefs and soil fertility and increase asthma ratesThese in turn affect China andBrazilwhich have bothinvested heavily in Caribbean tourism infrastructure and transportation Nutrient-rich dust fromAfrica also reachesBrazilwhere it improves forest productivity [Photo credits clockwise fromright topphoto Caitlin Jacobs Brandon Prince Rhett Butler and David Burdick used with permission]

on

Feb

ruar

y 26

201

5w

ww

sci

ence

mag

org

Dow

nloa

ded

from

o

n F

ebru

ary

26 2

015

ww

ws

cien

cem

ago

rgD

ownl

oade

d fr

om

on

Feb

ruar

y 26

201

5w

ww

sci

ence

mag

org

Dow

nloa

ded

from

o

n F

ebru

ary

26 2

015

ww

ws

cien

cem

ago

rgD

ownl

oade

d fr

om

on

Feb

ruar

y 26

201

5w

ww

sci

ence

mag

org

Dow

nloa

ded

from

o

n F

ebru

ary

26 2

015

ww

ws

cien

cem

ago

rgD

ownl

oade

d fr

om

on

Feb

ruar

y 26

201

5w

ww

sci

ence

mag

org

Dow

nloa

ded

from

o

n F

ebru

ary

26 2

015

ww

ws

cien

cem

ago

rgD

ownl

oade

d fr

om

on

Feb

ruar

y 26

201

5w

ww

sci

ence

mag

org

Dow

nloa

ded

from

o

n F

ebru

ary

26 2

015

ww

ws

cien

cem

ago

rgD

ownl

oade

d fr

om

REVIEW

SUSTAINABILITY

Systems integration forglobal sustainabilityJianguo Liu1 Harold Mooney2 Vanessa Hull1 Steven J Davis3 Joanne Gaskell4

Thomas Hertel5 Jane Lubchenco6 Karen C Seto7 Peter Gleick8

Claire Kremen9 Shuxin Li1

Global sustainability challenges from maintaining biodiversity to providing clean air andwater are closely interconnected yet often separately studied and managed Systemsintegrationmdashholistic approaches to integrating various components of coupled humanand natural systemsmdashis critical to understand socioeconomic and environmentalinterconnections and to create sustainability solutions Recent advances include thedevelopment and quantification of integrated frameworks that incorporate ecosystemservices environmental footprints planetary boundaries human-nature nexuses andtelecoupling Although systems integration has led to fundamental discoveries andpractical applications further efforts are needed to incorporate more human and naturalcomponents simultaneously quantify spillover systems and feedbacks integrate multiplespatial and temporal scales develop new tools and translate findings into policy andpractice Such efforts can help address important knowledge gaps link seeminglyunconnected challenges and inform policy and management decisions

The goal of achieving global sustainabilityis to meet societyrsquos current needs by usingEarthrsquos natural resources without compro-mising the needs of future generations (1)Yet many disparate research and manage-

ment efforts are uncoordinated and unintention-ally counterproductive toward global sustainabilitybecause a reductionist focus on individual com-ponentsof an integratedglobal systemcanoverlookcritical interactions across system componentsAlthough our planet is a single system compris-ing complex interactions between humans andnature research and management typically iso-late system components (such as air biodiversityenergy food land water and people) As a re-sult the compounding environmental impacts ofhuman activities have too often been missed be-cause they gobeyond the organizational level spaceand time of focus For example large amountsof affordable and reliable energy are available infossil fuels but concomitant emissions of carbondioxide (CO2) will alter global climate and affectother human and natural systemsmdasha trade-off thatcurrent policies have not adequately addressed(2) Likewise attention to growing more food on

land may inadvertently result in excess use offertilizers and in turn eutrophication of down-stream coastal waters that compromises foodproduction from the ocean Progressing towardglobal sustainability requires a systems approachto integrate various socioeconomic and environ-mental components that interact across organi-zational levels space and time (3ndash5)Systems integration generates many benefits

comparedwith isolated studies including under-standing of interconnectivity and complexity(Table 1) Here we review recent advances in de-veloping and quantifying frameworks for systemsintegration of coupled human and natural sys-tems illustrate successful applications focusingon unexpected impacts of biofuels and hiddenroles of virtual water and discuss future direc-tions for using systems integration toward globalsustainability

Framework developmentand quantification

The development and quantification of frame-works are critical steps in integrating humanand natural systems (6ndash9) For instance inter-actions between sectors and stakeholders in thehuman system or between biotic and abiotic fac-tors in the natural system at different organiza-tional levels (for example government agenciesfrom local to national levels and food trophiclevels from producers to consumers in ecosys-tems) lead to emergent properties that individualcomponents do not have (10) All coupled sys-tems evolve over timeas complex adaptive systems(11) Their interactions emergence evolutionand adaptation also vary with spatial scales (12)Accordingly integration along organizational spa-

tial and temporal dimensions is needed to avoidsustainability solutions in one system that causedeleterious effects in other systems Such in-tegration can also enhance positive and reducenegative socioeconomic and environmental effectsacrossmultiple systems at various organizationallevels over time (Table 1)Integration requires blending and distilling

of ideas concepts and theories from multiplenatural and social science disciplines as well asengineering and medical sciences (4 13) varioustools and approaches (such as simulation re-mote sensing and life cycle assessment) anddifferent types and sources of biophysical andsocioeconomic data (14) For example integratedassessment models such as those used by the In-tergovernmental Panel on Climate Change (IPCC)analyze information from diverse fields to under-stand complex environmental problems (such asacid rain climate change energy shortages andwater scarcity) (15 16) The Global Trade Analy-sis Project has recently evolved from a databasefor analyzing global trade-related economic issuesto a platform for integrating trade with globalland use and associated greenhouse gas (GHG)emissions (17) The Global Biosphere ManagementModel analyzes and plans land use among sec-tors (agriculture forestry and bioenergy) acrossthe globe in an integrated way (18) Below we il-lustrate the development and quantification ofsome important integration frameworks that haveled to substantial advances

Ecosystems services environmentalfootprints and planetary boundaries

Human and natural systems interact in a multi-tude of ways Several integration frameworksbring multiple aspects of human-nature interac-tions together (Fig 1) Quantifying the servicesthat ecosystems provide (Fig 1A) for societalneeds (such as clean water nutrient cycling andrecreation) (6) helps assign value to natural com-ponents for humans Recent advances considera variety of ecosystem services simultaneously inorder to evaluate trade-offs and synergies amongthem (19) Environmental footprint (20) andplanetary boundary (21) frameworks attempt toquantify the negative effects that human activitieshave on natural systems The environmentalfootprints framework quantifies resources (suchas natural capital) consumed and wastes gener-ated by humans (Fig 1B) (20) Recent manifes-tations of the concept go beyond the previouslydeveloped ecological footprints framework byincluding more diverse types of footprints [forexample water carbon and material footprints(20)] Planetary boundaries are threshold levelsfor key Earth system components and processes(such as stratospheric ozone global freshwaterand nitrogen cycling) beyond which humanitycannot safely be sustained (21) (Fig 1C)Quantifying the above frameworks relies on

systems integration For instance organizationalintegration in environmental footprint analysisdemonstrates how different human activities con-tribute to human impacts at local to global levels(20) Spatial integration is illustrated in integrated

RESEARCH

SCIENCE sciencemagorg 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 1258832-1

1Center for Systems Integration and Sustainability Departmentof Fisheries and Wildlife Michigan State University EastLansing MI USA 2Department of Biology Stanford UniversityStanford CA USA 3Department of Earth System ScienceUniversity of California Irvine CA USA 4World BankWashington DC USA 5Department of Agricultural EconomicsPurdue University West Lafayette IN USA 6Department ofIntegrative Biology Oregon State University Corvallis OR USA7School of Forestry and Environmental Studies Yale UniversityNew Haven CT USA 8The Pacific Institute Oakland CA USA9Department of Environmental Science Policy andManagement University of California Berkeley CA USACorresponding author E-mail liujimsuedu

1258832-2 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 sciencemagorg SCIENCE

Table 1 Example benefits of systems integration

Benefit Illustrative study

Revealing mechanisms ofecological degradation inprotected areas

Socioeconomic factors (such as forest harvesting fuelwood collection and increases inhousehold numbers) are responsible for ecological degradation in protected areas forgiant pandas (which are supposed to be protected from human activities) (100)

Understanding complexity Agricultural intensification schemes may promote further agricultural expansion over thelong term responses varied across space and were nonlinearly related to agriculturalinputs (101)

Improving economic efficiency Integrated assessment modeling shows specific cost estimates for delaying climate changemitigation with respect to geophysical technological social and political factors (102)

Reducing environmentalimpacts in distant places

Integrated cross-boundary management suggests ways of decreasing the spread of pollutionand spillover of climate-change effects to distant places around the globe (15)

Addressing multipleissues simultaneously

The climate changendashhealthndashfood security nexus demonstrates ways that managementmeasures can improve all three key issues at the same time (103)

Assessing the feasibility ofmultiple and conflicting goals

Integrated coastal zone management allows for multiorganizational management forcompeting interests such as recreation fisheries and biodiversity conservation (104)

Developing priorities forresearch and sustainability action

Integrated modeling of global water agriculture and climate change pinpoints areasvulnerable to future water scarcity and puts forth actionable strategies for mitigation (16)

Identifying complementaryconservation and managementstrategies

Coupling global energy security policy with climate change and air pollution policies(the air-climate-energy nexus) would decrease oil consumption compared to implementingenergy policy alone (46)

Enhancing synergiesamong factors

Cross-site integration of natural resource management approaches in response todisturbances shows opportunities for reframing ecosystem management to enhancecollaboration among institutions (such as NGOs government agencies researchorganizations businesses) (105)

Anticipating feedbacks A lag between fire control management and the response of the forests to such changesaffects the eagerness of landowners to continue implementing control measures (106)

Detecting latencies The latent effect of mosquito ditch construction on fish populations only emerged duringnew pressures from residential development and recreational fisheries (107)

Maximizing economic gainsand minimizing environmental costs

Integrated soil-crop management system could maximize grain yields while minimizingapplications of fertilizers and GHG emissions (108)

Fig 1 Examples of ecosystem services environmental footprints andplanetary boundaries (A) Ecosystem services (B) Environmental footprints(C) Planetary boundaries Outward arrows in (A to C) indicate increases in thevalues inward arrows indicate decreases and dashed lines indicate no dataIn (B) and (C) the inner green shading represents maximum sustainablefootprints (20) and safe operating space for nine planetary system variables(21) respectively Redwedges refer to the estimated current positions for thevariables Most of the ecosystem services in (A) decreased between the1950s to early 2000s (94) In (B) at least three types of footprints (eco-

logical carbon and material) have exceeded maximum sustainable footprints(20) Blue water footprint was 1690 billion m3year (1985ndash1999) (81) graywater footprint increased during 1970ndash2000 (95) and green water footprint is6700 billion m3year (without reference point) (20) Question marks indicatethat the information is uncertain Carbon footprint increased during 1960ndash2009 (96) With every 10 increase in gross domestic product the averagenational material footprint increases by 6 (97) For (C) all planetary systemvariables have increased in values between preindustry and 2000s and threeboundaries have been crossed (21)

RESEARCH | REVIEW

landscape planning for ecosystem services whichallows for coordination across space For exampleit can promote afforestation and reforestation inupland areas above irrigated agricultural systemsthus reducing erosion protecting waterways min-imizing flooding providing drinking water andfacilitating sustainable agricultural production(22) Temporal integration is crucial to quantifythe planetary boundaries framework as short-term fluctuations in key Earth system processesare scaled up to predict long-term trends manyof which cannot be accurately predicted with-out a systems approach (21) Temporal inte-gration can also reveal legacy effects of priorhuman-nature couplings For example carbonfootprints are driven in large part by past landuse (23) A condition termed ldquocarbon lock-inrdquo hasbeen used to describe systems that have evolvedover long time frames to become dependent onfossil fuels (24 25) The fossil energy system iscomprised of long-lived infrastructure such aspower plants which represent an investment infuture CO2 emissions Retiring this infrastruc-ture before the end of its economic or physicallifetimewould entail substantial costs As of 2013it is estimated the global committed emissionsrelated to existing fossil infrastructure are rough-ly 700 billion tons of CO2 (26)

Human-nature nexuses

In contrast to the conventional decision-makingthat takes place within separate disciplines orsectors the human-nature nexus framework rec-ognizes the interdependency between two ormore issues (or nodes) and addresses them to-gether For example the energy-foodnexus consid-ers both the effects of energy on food productionprocessing transporting and consumption andthe effects of food (such as corn) production onthe generation of energy (such as ethanol) (27)The nexus framework can help anticipate other-

wise unforeseen consequences evaluate trade-offsproduce co-benefits and allow the different andoften competing interests to seeka commonground(28) and co-optimization (29) The vast majorityof the 229 human-nature nexus studies recordedin the Web of Science (as of 16 August 2014)analyzed two-node nexuses (80) with only 16and 4 of the nexus studies including three andfour nodes respectively Although the concept offood-energy nexus first appeared in 1982 therewas no paper recorded in the Web of Science formany years The interest in the nexus frameworkhas reemerged recently and has grown rapidlysince 2010 Several two-nodenexuses have receivedspecial focus including energy-water nexus food-waternexus energy-foodnexus air-climate nexushealth-water nexus and energy-national securitynexus Among themore commonly examined three-and four-node nexuses are economy-environment-land nexus and climate-energy-food-water nexusAdding more nodes to a nexus framework leadsto more complexity but also captures greaterreality For instance the climate-energy-foodnexus considers not only the interrelationshipsbetween energy and food but also the relation-ships between energy and climate (for example

energy use emits CO2 and climate change affectsenergy demand such as heating and cooling) andinterrelationships between food and climate (forexample climate changes affect food productionand CO2 is emitted throughout the food produc-tion processing transporting and consumption)The nodes in the nexus framework are me-

diated by and influencemany organizational lev-els For example energy production and use areshaped by international markets and policies atdifferent government organizations and at thesame time influence many trophic levels of ani-mals plants and microorganisms (30) Buildingon the increasing recognition of conceptual in-terconnections among various nodes efforts areunder way to quantify their relationships suchas via hydro-economic modeling (31) structuraland nonstructural economicmodels (32) and lifecycle assessments (33) Scenario analysis is partic-ularly promising for teasing out roles of differentorganizations functioning at different scales(34 35) For example the Agrimonde model hasbeen used to examine intersections between foodand numerous other sectors (including energy andwater) worldwide under different growth and con-sumption scenarios (35) and illustrates the ef-fects of diverse individual governments on globalcross-sector dynamicsTemporal integration is also a key element of

the nexus framework For example recent long-term quantitative integration studies on theeconomy-energy nexus show that reductions inenergy use can have negative impacts on grossdomestic product (GDP) in the short-term butlittle detectable effect over the long term (36)Alternatively one model predicted that a smallincrease in foreign trade in Indonesia will leadto substantial increases in long-term per capitaCO2 emissions although its contribution to CO2

emissions is negligible in the short term (37)

Telecoupling

Many studies on sustainability have been place-based even if they look at coupled systems [forexample the energy-water nexus in the UnitedStates (38)] However economic production andresource use in different regions or countriesmaylead to very different consequences Furthermorethere are increasing distant interactions aroundthe world so that local events have consequencesglobally (39) For example each year several hun-dred million tons of dust from Africa (especiallythe Sahara desert) travel via the air across theAtlantic Ocean to distant places such as theCaribbean where it causes severe impacts in-cluding decline in coral reefs increase in asthmadisease spread and loss of soil fertility (40 41)Greenhouse gases emitted into the atmospherefromapoint source becomemixed and transportedglobally affecting societies and ecosystems fardistant from the point sources of origin Manyof the changes to the biotic composition of localplaces can also affect society regionally and glob-ally through ever-increasing global trade as wellas the often dramatic impact of invasive speciesand disease transmission In other words pat-terns and processes at one place may enhance or

compromise sustainability in other places (42)Human actions [such as production of biofuels(43)] in one place may create unintended con-sequences elsewhere [such as carbon leakage(44) biodiversity losses (45 46) and pollution(47)] Although external factors originating fromother systems are sometimes considered in sus-tainability research and practices they are typi-cally treated as one-way drivers of changes inthe system of interest with little attention to thefeedbacks between the system of interest andother systems (6 42)The framework of telecoupling (socioeconom-

ic and environmental interactions over distances)has been developed to tie distant places together(42) It is a natural extension of the frameworksof coupled human and natural systems and builton disciplinary frameworks such as climate tele-connections (distant interactions between cli-mate systems) urban land teleconnections (landchanges that are linked to underlying urbaniza-tion dynamics) (7) and economic globalization(distant interactions between human systems)So far the telecoupling framework has been ap-plied to a number of important issues acrossspatial scales such as global land-use and land-change science (39 48 49) international landdeals (39) species invasion (39 42) payments forecosystem services programs (50) and trade offood (42) and forest products (9)The framework is particularly effective for un-

derstanding socioeconomic and environmentalinteractions at international scales For examplethe flow of coal from Australia (sending system)to a number of countries and regions (receivingsystems for example Japan the EuropeanUnionand Brazil) reflects abundant Australian coal sup-plies and the demand for coal in receiving sys-tems (Fig 2) The coal trade is facilitated bymany agents in receiving and sending systems(such as government agencies that make coaltrade policies) and international organizations(such as shipping companies) Many other coun-tries (such as those inAfrica) are spillover systemsmdashsystems that may be affected by the coal flowsbecause of financial flows between sending andreceiving systems as well as the CO2 emissionsproduced when the coal is burned Global effortsto address this and similar feedbacks such as theREDD+ program tomitigate CO2 emissions fromdeforestation (51) and the Green Climate Fundto facilitate low-emission projects in developingnations (52) should focus on all countriesThe telecoupling framework can also be use-

ful at regional or national scales For examplethe 20million residents in Chinarsquos capital city ofBeijing (receiving system) receive clean waterfrom the Miyun Reservoir watershed (sendingsystem) more than 100 km away from the city(50) The framework explicitly links agents causesand effects in the sending and receiving sys-tems For instance the quantity and quality ofthe water flows are made possible through thePaddy Land-to-Dry Land program an ecosystemservices payment program that Beijing establishedwith the farmers in the watershedwho convertedrice cultivation in paddy land to corn production

SCIENCE sciencemagorg 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 1258832-3

RESEARCH | REVIEW

in dry land to provide clean water for Beijingin exchange for cash payments (53) Throughsystematic analysis the framework also helpsidentify research and governance gaps such asspillover systemsmdashregions that are affected bywater and cash flows between thewatershed andBeijing but have received little attention fromresearchers and government agencies (50)The telecoupling framework also emphasizes

temporal dynamics A lack of temporal integra-tion may miss key dynamics and create a misun-derstanding of infrequent but drastic changessuch as disasters wars outbreaks of deadly dis-eases such as Ebola (54) regime shifts and pro-found policy changes For instance in theWolongNature Reserve of China designated for conserv-ing the endangered giant pandas the devastat-ing earthquake in 2008 substantially altered thetelecouplings between Wolong and outside sys-tems [for example collapse of tourism and agricul-tural trade (55)] Studies omitting the earthquakeimpacts could misrepresent the mechanisms be-hind the system dynamics (such as increases inlandslides and relocation of households)

Applications of systems integration

Systems integration has been applied success-fully to many sustainability issues IntegratedCoastal ZoneManagement (56) Marine SpatialPlanning (57) andEcosystem-BasedManagement(58) all integratemultiple dimensions for naturalresource management Although some of thesepractices have existed for several decades therehave been continued efforts formore integrationnew advances and novel insights For exampleEcosystem-Based Management has expanded totackle issues not traditionally thought of within

natural resourcemanagement such as food secu-rity (59) politics (60) and disease (61) The ex-amples of biofuels and virtual water below alsoillustrate the importance of systems integrationin detail We chose to focus on these two exam-ples because they are emerging and contentiousglobal phenomena that represent challengingsustainability issues and they have unexpectedand hidden socioeconomic and environmentaleffects that were impossible to reveal withoutsystems integration

Unexpected impacts of biofuels

The environmental and socioeconomic impactsof biofuels have been among the most hotlycontested policy issues over the past decade TheUnited States EuropeanUnion and nearly threedozen other countries in Africa Asia and theAmericas have developed biofuel mandates ortargets (62) This enthusiasm was buoyed by theprospect of displacing high-priced oil importsgenerating rural incomes and contributing toclimate change mitigation As of 2006 it wassuggested that biofuels could be both economi-cally and environmentally beneficial Howeverwith the implementations of these policies andsystems integration research serious concernshave arisen about their geospatial impacts andthe temporal viability of these mandatesBiofuels are a prime topic for systems integra-

tion research because biofuel production andconsumption as well as their impacts vary acrosstime space and organizational levels For in-stance the carbon fluxes after conversion of newcroplands depend not only on below- and above-ground carbon at present but also on the legacyeffects stemming fromhistorical land use such as

land clearing as well as subsequent cultivationand cropping practices (63) The United Statesand Brazil were responsible for 90 of the globalbiofuel production of 105 billion liters in 2011 butseveral other countries with new mandates suchas China Canada and Argentina are increasingthe spatial extent of production (64) Assessingorganizational impacts on biofuel production en-compasses analysis that integrates across institu-tions For example when and where additionalcropland is converted for biofuel production de-pends critically on local national and internationalinstitutions as well as the global supply chainThe systems integration frameworks discussed

above have direct relevance to biofuels For ex-ample the environmental footprint frameworkhas been applied to assess the impacts of bio-fuels It is estimated that the global footprintfrom biofuels was sim072 billion gha in 2010 andexpected to rise by 73 in 2019 (consisting ofland use carbon embodied energy materialsand waste transport and water) (65) From theperspective of ecosystem services biofuels haveboth positive (for example energy) and negative(for example loss of food and freshwater services)impacts Biofuel production also affects severalplanetary boundaries including climate changeland-use change (proportion of cropland) nutri-ent cycles (increased use of phosphorus and nitro-gen) and biodiversity loss (66) For instance it isestimated that corn-based ethanol nearly doublesgreenhouse emissions across the world over a30-year period because of land-use change (43)Biofuels have also been studied under the

human-nature nexus frameworkmdashin particularthe energy-food nexus and energy-food-waternexus Rising demand for ethanol feedstocks bid

1258832-4 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 sciencemagorg SCIENCE

Fig 2 Illustrative example of sending receiving and spillover systemsas well as flows under the telecoupling framework In the case of tradein Australian coal in 2004 [measured in megatons (Mt) of CO2 emissions]Australia is a sending system (in blue) the main receiving systems are ingreen (destinations of the coal are Japan South Korea Taiwan MalaysiaIndia the European Union and Brazil) and the spillover systems are in lightred (all other countries and regions are affected by CO2 emissions from

using the coal produced in Australia and consumed in the receivingsystems) The arrows show the magnitude of the flows Countries thatreceive less than 10 Mt of emissions of Australian coal are not includedFlows to Europe are aggregated to include all 28 member states of theEuropean Union [data from (44)] Financial flows take place betweensending and receiving systems but may also affect financial conditions inspillover systems indirectly [data from (98)]

RESEARCH | REVIEW

up food price (67) which has major implicationsfor food security (68) And questions have beenraised about the adverse interplay between bio-fuel mandates and increased interannual varia-bility in crop production anticipated under futureclimate change (69) Water also comes into playas limits on the future availability of water forirrigated agriculture will shift the location ofcropland conversion owing to biofuel expansiontoward regions with carbon-rich rainfed agricul-ture Overall accounting for hydrological con-straints boosts estimated GHG emissions fromland use by 25 (70)Telecoupled processes such as international

trade and flows of information (for exampleglobal market prices) cause biofuel programs inone part of the world to translate spatially intoland conversion in other regions (indirect landuse) They have already contributed to croplandexpansion in the United States and overseasand to cascading and spillover effects over longdistances (Fig 3) National biofuel programswhich looked environmentally beneficial at first

blush might in fact lead to increased environ-mental damage when viewed over time and atthe global scale (42 71) Unlike early analyses(42) that assumed that higher prices effectivelyinfluenced all agents in the market equally sub-sequent research has revealed that some agricul-tural suppliers (such as the United States andArgentina) are more closely telecoupled than oth-ers (72) The spatial pattern and extent of landconversion stemming from biofuels are also af-fected by geophysical characteristics such as poten-tial productivity of the newly converted lands

Hidden roles of virtual water

Although many sustainability studies have fo-cused on flows of real material and energy suchas biofuels there has been increasing interest inthe flows of ldquovirtualrdquo material and energy suchas ldquovirtual waterrdquo ldquovirtual energyrdquo ldquovirtual landrdquoand ldquovirtual nutrientsrdquo (73) Virtual resources arethose resources used for production and incor-porated into goods and services in the same waythat related pollution and impacts are embodied

(or hidden) in these products In the case ofwater for example it is used to grow crops raiselivestock and grow their food and produce mar-ketable goods Virtual water is traded amongcountries as goods are traded Globally the vol-ume of virtual water trade and the number oflinks (pairs of trading countries) have bothdoubledfrom 1986 to 2010 (74)With roughly 27 trillionm3

of water traded virtually worldwide in 2010 (74)virtual water trade fromwater-rich countries hashelped mitigate water shortages in water-poorcountries (75) The concept of virtual resourceshas helped analysts think more clearly about thereal risks of resource scarcity and the role thattrade plays in mitigating or worsening those risks(76 77) Targeted trade policies may help to fur-ther prevent water scarcity by encouraging morewater-efficient trade links (78)Virtual water is a good target for systems in-

tegration research because the issues involvedaredynamic across time space andorganizationallevels Global water scarcity issues are inherent-ly temporally sensitive with cumulative effects

SCIENCE sciencemagorg 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 1258832-5

Fig 3 Cascading and spillover effects of biofuel production on landconversion and CO2 emissions as revealed by systems integrationMeeting the Renewable Fuel Standard mandate [to produce 50 Gigaliters(GL) of additional ethanol on top of the 2001 production level] in the UnitedStates reduces the use of petroleumbut requires additional corn area (99)Theexpansion in US corn area leads to reduction in harvested area of oilseeds andother crops in the United StatesThis boosts world prices for these crops andencourages more production of oilseeds and corn in the rest of the worldTheexpansion of cropland in the United States and the rest of the world leads to

more emissions of CO2 and the conversion of pasture and forest lands aroundthe world This land conversion also releases CO2 offsetting the reduction ofCO2 emissions from using less fossil fuels and more biofuels (assuming a 23ethanolpetroleum energy conversion rate) Estimates are on an annual basisover a 30-year production period for the biofuel facilities and in approximateamounts derived from (99) Arrows indicate the direction of influencesSymbols ldquondashrdquo and ldquo+rdquo refer to decrease and increase respectively in land areaethanol fossil fuel or CO2Tg teragrams Mha million ha [Graphics are usedwith permissions from Fotoliacom]

RESEARCH | REVIEW

stemming from legacies of overuse of water in-teractingwithnewdrivers such as climate changeEstimates suggest that global virtual water trademay decrease with climate change because of thedifficulty of growing crops in warmer drier cli-mates [water savings from reduction in growingrice soybeans and wheat may amount to up to15 trillion m3 in 2030 (78)] Also there was aprofound shift in the spatial distribution of hu-man populations (thus water demand) relativeto water distribution over the past few decadesIn 1986 68 of the worldrsquos population was inwater-exporting countries but by 2010 the dis-tribution was almost completely reversed with60 of the global population in water-importingcountries (74) One of the greatest organizationalconcerns related to virtual water is that a fewcountries control themajority of the global tradewhich leaves the market vulnerable to the deci-sionsmade by a few key players (74) In additionthere is unequal distribution of resources withincountries and a tendency for local agrarian com-munities to be marginalized owing to trade dic-tated by country-level agencies (79)The ecosystem services framework has con-

tributed to virtual water research in many waysFor example Canada is a major exporter of vir-tual water worldwide (95 Gm3 year) exportingvirtual water affects the ecosystem services pro-vided by the nationrsquos boreal forests which makeup nearly 60 of Canadarsquos territory (80) Pro-duction of commodities through processes suchas hydroelectric power generation oil extractioncrop irrigation and livestock-rearing contributesto virtual water exports which in turn threatenfreshwater resources that are a key part of theboreal forests Boreal freshwater comprises 80 to90 of Canadarsquos lakes and 25 of the entireEarthrsquos wetlands (80) Removal of water compro-mises the estimated annual gain of $703 billionin ecosystem services that the boreal forests pro-

vide including carbon storage flood control andwater filtering biodiversity conservation andpest control (80)The environmental footprints framework has

informed virtual water research by depicting thewater resources used for production of goodsand services (ldquowater footprintsrdquo) For example2320 Gm3year of the total global water foot-print of 9087 Gm3year comes from virtual watertrade (81) There is a close relationship betweenvirtual water and planetary boundaries becauseone of the nine key planetary boundaries iden-tified is the limit to global freshwater use (21) Arelated conceptmdashldquopeakwaterrdquomdashhelps to illustratehow close global freshwater bodies are to thisthreshold Global water consumption has alreadyreached a peak and begun to decline in manyareas because of limited remaining water (13)Furthermore scarcity in global freshwater is inlarge part linked to the virtual water embeddedin agricultural production and trade (81) Thehuman-nature nexus framework is also usefulfor virtual water research For example a studyon water-food nexus indicates that 76 of virtualwater trade is attributed to crops or crop-derivedproducts (81)Virtual water trade varies spatially and is an

important telecoupling process Themain virtualwater exporters (sending systems) are water-richregions inNorth and South America andAustraliawhereas Mexico Japan China and water-poorregions in Europe are the main importers (re-ceiving systems) (Fig 4) (75) Sending and re-ceiving systems involved in virtual water tradehave dynamic roles Asia recently switched itsvirtual water imports from North America toSouth America (82) On the other hand NorthAmerica has engaged in an increased diversifi-cation of intraregional water trade while tradingwith distant countries in Asia (82) China hasundergone a dramatic increase in virtual water

imports since 2000 via products such as soy-beans from Brazil (nearly doubling from 2001 to2007 and amounting to 13 of the total globalworld water trade) (82) The spatial shift in theuse of soybean products in Brazil from domesticto international has led to water savings in othercountries but at the cost of deforestation inBrazilian Amazon (82) Within-country virtualwater transfer is also common For example vir-tual water flow through grain trade from NorthChina to South China goes in the opposite direc-tion of real water transfer through large projectssuch as the South-to-North Water Transfer Pro-ject that aim to alleviate water shortages inNorth China

Future directions

Despite the substantial progress in systems inte-gration illustratedabovemany important challengesremain For example the integrated frameworkshave been studied largely in isolation althoughthey are interconnected throughhuman activities(for example usingmore ecosystem services maylead to larger environmental footprints) Achiev-ing a greater degree of integration would involveanalyzing and managing coupled human andnatural systems over longer time periods largerspatial extents (for example macrosystems andultimately the entire planet) and across morediverse organizations at different levels Belowwe suggest several ways to advance systems in-tegration with the intent of improving its theo-retical foundations expanding its tool box andproviding broad implications for managementand policy

Incorporate more human and naturalcomponents simultaneously

Although some previous studies have consideredmultiple components of coupled human andnatu-ral systems many components are either not

1258832-6 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 sciencemagorg SCIENCE

Fig 4 Balance and flows of virtual water related to trade of agricultural and industrial products during 1996ndash2005 Net exporters (sending systems)are in green and net importers (receiving systems) are in red The arrows indicate the relative sizes of large gross virtual water flows between sending andreceiving systems (gt 15 Gm3year) Countries without arrows are potential spillover systems of the large virtual water flows Data are from (81)

RESEARCH | REVIEW

considered or treated as exogenous variablesleading to biases and even incorrect conclusionsFor example a food-water nexus study withoutconsidering GHG emissions during groundwaterextraction for irrigation of crops in China under-estimated GHG emissions by as much as 331MtCO2e (83) One way to correct this problem isto convert more variables from exogenous toendogenousmdashinternalize all important relevantvariablesmdashso that their dynamics and feedbackeffects are explicitly studied For instance con-sidering multiple telecoupling processes (suchas species invasion trade disease spread andtechnology transfers) at the same time can helplinkmany seemingly disconnected distant inter-actions Unifying evolutionary approaches suchas seen with complex adaptive systems (84) mayprovide productive ways to integrate disparateideas and understand temporal dynamics andsustainability of coupled systems

Identify and quantify spillover systems

Previous research on issues such as trade oftenfocused on sending and receiving systems (forexample trade partners) with little attention tospillover systems (for example nontrade partners)mdashother systems affected by the interactions betweensending and receiving systems Although someprevious studies have recognized some spillovereffects [such as spatial externalities (85 86)] theywere often on either socioeconomic or environ-mental effects rather than all effects simulta-neously Furthermore they rarely consider othercomponents of spillover systems (causes agentsand flows) as articulated in the telecouplingframework (41) Identifying and quantifying oth-er components of spillover systems related tospillover effects may help understand the mech-anisms behind the spillover effects and developmore effective management strategies Connect-ing spillover systems with sending and receivingsystems through network analysis (87) may gen-erate fruitful outcomes such as the appreciationof dynamic interrelationships among differentsystems

Explicitly account for feedbacks

Human-nature feedbacks are a core componentof coupled systems For instance an importantnegative feedback in Wolong Nature Reserve forgiant pandas in China occurred when deforesta-tion and panda habitat degradation by localhouseholds prompted the government to devel-op and implement new conservation programsthat provide subsidies for local households tomonitor forests and thus reduce deforestationand improve panda habitat (88) This feedbackhas helped forest and habitat recovery whileincreasing income for local households Moreinnovative measures such as this are needed toidentify and use feedbacks as mechanisms forsustainability

Integrate multiple temporal andspatial scales

Human and natural processes and patterns atmultiple scales may be different and they can

interact with each other For example food pro-duction at the local scale may create jobs at thelocal scale but may not affect overall job creationat the global scale Many urban sustainabilityefforts focus on locally specific solutions thatmay not be scalable (7) Thus considering mul-tiple spatial scales at the same time can helpidentify all important factors their interdepen-dence and their effects and nonlinear relation-ships Temporally short-term studies should becombined with long-term studies in order tomaximize the strengths of each approach Forinstance short-term studies may capture morenuanced immediate changes in system behaviorbut long-term studies may account for temporaldynamics time lags cumulative effects legacy ef-fects and other phenomena (such as rare events)that cannot be seen over shorter terms Moresystematic incorporation of human dimensionsto long-term studies such as the Long-term Ecol-ogical Research sites and the National EcologicalObservatory Network is needed

Develop and use new tools

More effective integration requires developingand using powerful tools to overcome difficultbarriers (for example mathematical and com-putational challenges quantification of impactsat one scale on other scales and relationshipsamong patterns and processes across scales andacross borders) and to predict emergence of un-expected threats for sustainability policy andmanagement Examples include spatially explicitlife cycle assessment supply chain analysis andmultilevel modeling Agent-based models areparticularly promising tools because they takeinteractions (such as human adaptation to en-vironmental changes) at different scales into ac-count and model coupled systems as complexadaptive systems Agent-basedmodels create vir-tual worlds thatmimic the real world in contrastto traditional empirical statistical models (suchas econometric models) that are fitted to pastdata and fail when the future differs from thepast and dynamic stochastic general equilibriummodels that presume a perfect world and ignoredisturbances or crises (89) Many agent-basedmodels have beendeveloped in various disciplinesto provide insights on complexities and informa-tion for policymaking in issues such as economicdevelopment andmanagement of common spaces(90 91) However new models are needed toaccount for telecoupled systems Also increasingcomputational powerwill allowagent-basedmod-els to include more agents in larger areas andultimately all important agents across the worldAs more high-resolution data become availableit is necessary to develop and use big data tools(such as distributed databases massively parallelprocessing and cloud computing) for effectiveand efficient searching retrieving analysis andintegration (92)

Translate findings into policy and practice

Systems integration can provide more unbiasedinformation for policy and practice to help clarifyresponsibilities mediate trade-offs reduce con-

flicts and anticipate future trends It is necessaryto foster coordination among multiple nationaland international policies andminimize situationsin which different policies offset one another be-cause of conflicting goals and counterproductiveimplementation Unfortunately institutions andregulations have traditionally focused on singleissues and often do not have the mandate or in-frastructure to address the organizational con-nections and detrimental spillovers The WorldTrade Organization for example has the princi-pal mandate of promoting global trade One ofthe spillover effects of this mission is the globaltransmission of invasive species but the meansto address invasive species are weak comparedwith the forces driving the global market (93)Adopting the telecoupling framework can helpassign responsibilities of addressing spillover ef-fects (such as CO2 emissions and species inva-sion) to consumers and producers (for examplevia regulation at the source of extraction or con-sumption) as well as others such as traders ofgoods and products across space Last govern-ments need to incorporate long-term studies intotheir policies to account for the complex dynam-ics of coupled systems (such as time lags) Moreapplications of systems integration frameworksandmethods such as those discussed in this papercan accelerate understanding and solving globalsustainability challenges

REFERENCES AND NOTES

1 World Commission On Environment and DevelopmentOur Common Future (Oxford Univ Press Oxford 1987)

2 M A Hanjra M E Qureshi Global water crisis and futurefood security in an era of climate change Food Policy 35365ndash377 (2010) doi 101016jfoodpol201005006

3 J Liu et al Complexity of coupled human and naturalsystems Science 317 1513ndash1516 (2007) doi 101126science1144004 pmid 17872436

4 H A Mooney A Duraiappah A Larigauderie Evolution ofnatural and social science interactions in global changeresearch programs Proc Natl Acad Sci USA 110(suppl 1) 3665ndash3672 (2013) doi 101073pnas1107484110pmid 23297237

5 W Steffen et al Global Change and the Earth SystemA Planet Under Pressure IGBP Science 4 (Springer VerlagHeidelberg Germany 2004)

6 Millennium Ecosystem Assessment Ecosystems amp HumanWell-being Synthesis (Island Press Washington DC 2005)

7 K C Seto et al Urban land teleconnections andsustainability Proc Natl Acad Sci USA 109 7687ndash7692(2012) doi 101073pnas1117622109 pmid 22550174

8 E Ostrom A general framework for analyzing sustainabilityof social-ecological systems Science 325 419ndash422 (2009)doi 101126science1172133 pmid 19628857

9 J Liu Forest sustainability in China and implications for atelecoupled world Asia Pacific Pol Stud 1 230ndash250 (2014)

10 J Norberg G S Cumming Complexity Theory for aSustainable Future (Columbia Univ Press New York 2013)

11 S A Levin Ecosystems and the biosphere as complexadaptive systems Ecosystems (NY) 1 431ndash436 (1998)doi 101007s100219900037

12 E F Lambin Conditions for sustainability of humanndashenvironment systems Information motivation andcapacity Glob Environ Change 15 177ndash180 (2005)doi 101016jgloenvcha200506002

13 P H Gleick M Palaniappan Peak water limits to freshwaterwithdrawal and use Proc Natl Acad Sci USA 107 11155ndash11162(2010) doi 101073pnas1004812107 pmid 20498082

14 J Cools et al Coupling a hydrological water qualitymodel and an economic optimization model to set upa cost-effective emission reduction scenario for nitrogenEnviron Model Softw 26 44ndash51 (2011) doi 101016jenvsoft201004017

SCIENCE sciencemagorg 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 1258832-7

RESEARCH | REVIEW

1258832-2 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 sciencemagorg SCIENCE

Table 1 Example benefits of systems integration

Benefit Illustrative study

Revealing mechanisms ofecological degradation inprotected areas

Socioeconomic factors (such as forest harvesting fuelwood collection and increases inhousehold numbers) are responsible for ecological degradation in protected areas forgiant pandas (which are supposed to be protected from human activities) (100)

Understanding complexity Agricultural intensification schemes may promote further agricultural expansion over thelong term responses varied across space and were nonlinearly related to agriculturalinputs (101)

Improving economic efficiency Integrated assessment modeling shows specific cost estimates for delaying climate changemitigation with respect to geophysical technological social and political factors (102)

Reducing environmentalimpacts in distant places

Integrated cross-boundary management suggests ways of decreasing the spread of pollutionand spillover of climate-change effects to distant places around the globe (15)

Addressing multipleissues simultaneously

The climate changendashhealthndashfood security nexus demonstrates ways that managementmeasures can improve all three key issues at the same time (103)

Assessing the feasibility ofmultiple and conflicting goals

Integrated coastal zone management allows for multiorganizational management forcompeting interests such as recreation fisheries and biodiversity conservation (104)

Developing priorities forresearch and sustainability action

Integrated modeling of global water agriculture and climate change pinpoints areasvulnerable to future water scarcity and puts forth actionable strategies for mitigation (16)

Identifying complementaryconservation and managementstrategies

Coupling global energy security policy with climate change and air pollution policies(the air-climate-energy nexus) would decrease oil consumption compared to implementingenergy policy alone (46)

Enhancing synergiesamong factors

Cross-site integration of natural resource management approaches in response todisturbances shows opportunities for reframing ecosystem management to enhancecollaboration among institutions (such as NGOs government agencies researchorganizations businesses) (105)

Anticipating feedbacks A lag between fire control management and the response of the forests to such changesaffects the eagerness of landowners to continue implementing control measures (106)

Detecting latencies The latent effect of mosquito ditch construction on fish populations only emerged duringnew pressures from residential development and recreational fisheries (107)

Maximizing economic gainsand minimizing environmental costs

Integrated soil-crop management system could maximize grain yields while minimizingapplications of fertilizers and GHG emissions (108)

Fig 1 Examples of ecosystem services environmental footprints andplanetary boundaries (A) Ecosystem services (B) Environmental footprints(C) Planetary boundaries Outward arrows in (A to C) indicate increases in thevalues inward arrows indicate decreases and dashed lines indicate no dataIn (B) and (C) the inner green shading represents maximum sustainablefootprints (20) and safe operating space for nine planetary system variables(21) respectively Redwedges refer to the estimated current positions for thevariables Most of the ecosystem services in (A) decreased between the1950s to early 2000s (94) In (B) at least three types of footprints (eco-

logical carbon and material) have exceeded maximum sustainable footprints(20) Blue water footprint was 1690 billion m3year (1985ndash1999) (81) graywater footprint increased during 1970ndash2000 (95) and green water footprint is6700 billion m3year (without reference point) (20) Question marks indicatethat the information is uncertain Carbon footprint increased during 1960ndash2009 (96) With every 10 increase in gross domestic product the averagenational material footprint increases by 6 (97) For (C) all planetary systemvariables have increased in values between preindustry and 2000s and threeboundaries have been crossed (21)

RESEARCH | REVIEW

landscape planning for ecosystem services whichallows for coordination across space For exampleit can promote afforestation and reforestation inupland areas above irrigated agricultural systemsthus reducing erosion protecting waterways min-imizing flooding providing drinking water andfacilitating sustainable agricultural production(22) Temporal integration is crucial to quantifythe planetary boundaries framework as short-term fluctuations in key Earth system processesare scaled up to predict long-term trends manyof which cannot be accurately predicted with-out a systems approach (21) Temporal inte-gration can also reveal legacy effects of priorhuman-nature couplings For example carbonfootprints are driven in large part by past landuse (23) A condition termed ldquocarbon lock-inrdquo hasbeen used to describe systems that have evolvedover long time frames to become dependent onfossil fuels (24 25) The fossil energy system iscomprised of long-lived infrastructure such aspower plants which represent an investment infuture CO2 emissions Retiring this infrastruc-ture before the end of its economic or physicallifetimewould entail substantial costs As of 2013it is estimated the global committed emissionsrelated to existing fossil infrastructure are rough-ly 700 billion tons of CO2 (26)

Human-nature nexuses

In contrast to the conventional decision-makingthat takes place within separate disciplines orsectors the human-nature nexus framework rec-ognizes the interdependency between two ormore issues (or nodes) and addresses them to-gether For example the energy-foodnexus consid-ers both the effects of energy on food productionprocessing transporting and consumption andthe effects of food (such as corn) production onthe generation of energy (such as ethanol) (27)The nexus framework can help anticipate other-

wise unforeseen consequences evaluate trade-offsproduce co-benefits and allow the different andoften competing interests to seeka commonground(28) and co-optimization (29) The vast majorityof the 229 human-nature nexus studies recordedin the Web of Science (as of 16 August 2014)analyzed two-node nexuses (80) with only 16and 4 of the nexus studies including three andfour nodes respectively Although the concept offood-energy nexus first appeared in 1982 therewas no paper recorded in the Web of Science formany years The interest in the nexus frameworkhas reemerged recently and has grown rapidlysince 2010 Several two-nodenexuses have receivedspecial focus including energy-water nexus food-waternexus energy-foodnexus air-climate nexushealth-water nexus and energy-national securitynexus Among themore commonly examined three-and four-node nexuses are economy-environment-land nexus and climate-energy-food-water nexusAdding more nodes to a nexus framework leadsto more complexity but also captures greaterreality For instance the climate-energy-foodnexus considers not only the interrelationshipsbetween energy and food but also the relation-ships between energy and climate (for example

energy use emits CO2 and climate change affectsenergy demand such as heating and cooling) andinterrelationships between food and climate (forexample climate changes affect food productionand CO2 is emitted throughout the food produc-tion processing transporting and consumption)The nodes in the nexus framework are me-

diated by and influencemany organizational lev-els For example energy production and use areshaped by international markets and policies atdifferent government organizations and at thesame time influence many trophic levels of ani-mals plants and microorganisms (30) Buildingon the increasing recognition of conceptual in-terconnections among various nodes efforts areunder way to quantify their relationships suchas via hydro-economic modeling (31) structuraland nonstructural economicmodels (32) and lifecycle assessments (33) Scenario analysis is partic-ularly promising for teasing out roles of differentorganizations functioning at different scales(34 35) For example the Agrimonde model hasbeen used to examine intersections between foodand numerous other sectors (including energy andwater) worldwide under different growth and con-sumption scenarios (35) and illustrates the ef-fects of diverse individual governments on globalcross-sector dynamicsTemporal integration is also a key element of

the nexus framework For example recent long-term quantitative integration studies on theeconomy-energy nexus show that reductions inenergy use can have negative impacts on grossdomestic product (GDP) in the short-term butlittle detectable effect over the long term (36)Alternatively one model predicted that a smallincrease in foreign trade in Indonesia will leadto substantial increases in long-term per capitaCO2 emissions although its contribution to CO2

emissions is negligible in the short term (37)

Telecoupling

Many studies on sustainability have been place-based even if they look at coupled systems [forexample the energy-water nexus in the UnitedStates (38)] However economic production andresource use in different regions or countriesmaylead to very different consequences Furthermorethere are increasing distant interactions aroundthe world so that local events have consequencesglobally (39) For example each year several hun-dred million tons of dust from Africa (especiallythe Sahara desert) travel via the air across theAtlantic Ocean to distant places such as theCaribbean where it causes severe impacts in-cluding decline in coral reefs increase in asthmadisease spread and loss of soil fertility (40 41)Greenhouse gases emitted into the atmospherefromapoint source becomemixed and transportedglobally affecting societies and ecosystems fardistant from the point sources of origin Manyof the changes to the biotic composition of localplaces can also affect society regionally and glob-ally through ever-increasing global trade as wellas the often dramatic impact of invasive speciesand disease transmission In other words pat-terns and processes at one place may enhance or

compromise sustainability in other places (42)Human actions [such as production of biofuels(43)] in one place may create unintended con-sequences elsewhere [such as carbon leakage(44) biodiversity losses (45 46) and pollution(47)] Although external factors originating fromother systems are sometimes considered in sus-tainability research and practices they are typi-cally treated as one-way drivers of changes inthe system of interest with little attention to thefeedbacks between the system of interest andother systems (6 42)The framework of telecoupling (socioeconom-

ic and environmental interactions over distances)has been developed to tie distant places together(42) It is a natural extension of the frameworksof coupled human and natural systems and builton disciplinary frameworks such as climate tele-connections (distant interactions between cli-mate systems) urban land teleconnections (landchanges that are linked to underlying urbaniza-tion dynamics) (7) and economic globalization(distant interactions between human systems)So far the telecoupling framework has been ap-plied to a number of important issues acrossspatial scales such as global land-use and land-change science (39 48 49) international landdeals (39) species invasion (39 42) payments forecosystem services programs (50) and trade offood (42) and forest products (9)The framework is particularly effective for un-

derstanding socioeconomic and environmentalinteractions at international scales For examplethe flow of coal from Australia (sending system)to a number of countries and regions (receivingsystems for example Japan the EuropeanUnionand Brazil) reflects abundant Australian coal sup-plies and the demand for coal in receiving sys-tems (Fig 2) The coal trade is facilitated bymany agents in receiving and sending systems(such as government agencies that make coaltrade policies) and international organizations(such as shipping companies) Many other coun-tries (such as those inAfrica) are spillover systemsmdashsystems that may be affected by the coal flowsbecause of financial flows between sending andreceiving systems as well as the CO2 emissionsproduced when the coal is burned Global effortsto address this and similar feedbacks such as theREDD+ program tomitigate CO2 emissions fromdeforestation (51) and the Green Climate Fundto facilitate low-emission projects in developingnations (52) should focus on all countriesThe telecoupling framework can also be use-

ful at regional or national scales For examplethe 20million residents in Chinarsquos capital city ofBeijing (receiving system) receive clean waterfrom the Miyun Reservoir watershed (sendingsystem) more than 100 km away from the city(50) The framework explicitly links agents causesand effects in the sending and receiving sys-tems For instance the quantity and quality ofthe water flows are made possible through thePaddy Land-to-Dry Land program an ecosystemservices payment program that Beijing establishedwith the farmers in the watershedwho convertedrice cultivation in paddy land to corn production

SCIENCE sciencemagorg 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 1258832-3

RESEARCH | REVIEW

in dry land to provide clean water for Beijingin exchange for cash payments (53) Throughsystematic analysis the framework also helpsidentify research and governance gaps such asspillover systemsmdashregions that are affected bywater and cash flows between thewatershed andBeijing but have received little attention fromresearchers and government agencies (50)The telecoupling framework also emphasizes

temporal dynamics A lack of temporal integra-tion may miss key dynamics and create a misun-derstanding of infrequent but drastic changessuch as disasters wars outbreaks of deadly dis-eases such as Ebola (54) regime shifts and pro-found policy changes For instance in theWolongNature Reserve of China designated for conserv-ing the endangered giant pandas the devastat-ing earthquake in 2008 substantially altered thetelecouplings between Wolong and outside sys-tems [for example collapse of tourism and agricul-tural trade (55)] Studies omitting the earthquakeimpacts could misrepresent the mechanisms be-hind the system dynamics (such as increases inlandslides and relocation of households)

Applications of systems integration

Systems integration has been applied success-fully to many sustainability issues IntegratedCoastal ZoneManagement (56) Marine SpatialPlanning (57) andEcosystem-BasedManagement(58) all integratemultiple dimensions for naturalresource management Although some of thesepractices have existed for several decades therehave been continued efforts formore integrationnew advances and novel insights For exampleEcosystem-Based Management has expanded totackle issues not traditionally thought of within

natural resourcemanagement such as food secu-rity (59) politics (60) and disease (61) The ex-amples of biofuels and virtual water below alsoillustrate the importance of systems integrationin detail We chose to focus on these two exam-ples because they are emerging and contentiousglobal phenomena that represent challengingsustainability issues and they have unexpectedand hidden socioeconomic and environmentaleffects that were impossible to reveal withoutsystems integration

Unexpected impacts of biofuels

The environmental and socioeconomic impactsof biofuels have been among the most hotlycontested policy issues over the past decade TheUnited States EuropeanUnion and nearly threedozen other countries in Africa Asia and theAmericas have developed biofuel mandates ortargets (62) This enthusiasm was buoyed by theprospect of displacing high-priced oil importsgenerating rural incomes and contributing toclimate change mitigation As of 2006 it wassuggested that biofuels could be both economi-cally and environmentally beneficial Howeverwith the implementations of these policies andsystems integration research serious concernshave arisen about their geospatial impacts andthe temporal viability of these mandatesBiofuels are a prime topic for systems integra-

tion research because biofuel production andconsumption as well as their impacts vary acrosstime space and organizational levels For in-stance the carbon fluxes after conversion of newcroplands depend not only on below- and above-ground carbon at present but also on the legacyeffects stemming fromhistorical land use such as

land clearing as well as subsequent cultivationand cropping practices (63) The United Statesand Brazil were responsible for 90 of the globalbiofuel production of 105 billion liters in 2011 butseveral other countries with new mandates suchas China Canada and Argentina are increasingthe spatial extent of production (64) Assessingorganizational impacts on biofuel production en-compasses analysis that integrates across institu-tions For example when and where additionalcropland is converted for biofuel production de-pends critically on local national and internationalinstitutions as well as the global supply chainThe systems integration frameworks discussed

above have direct relevance to biofuels For ex-ample the environmental footprint frameworkhas been applied to assess the impacts of bio-fuels It is estimated that the global footprintfrom biofuels was sim072 billion gha in 2010 andexpected to rise by 73 in 2019 (consisting ofland use carbon embodied energy materialsand waste transport and water) (65) From theperspective of ecosystem services biofuels haveboth positive (for example energy) and negative(for example loss of food and freshwater services)impacts Biofuel production also affects severalplanetary boundaries including climate changeland-use change (proportion of cropland) nutri-ent cycles (increased use of phosphorus and nitro-gen) and biodiversity loss (66) For instance it isestimated that corn-based ethanol nearly doublesgreenhouse emissions across the world over a30-year period because of land-use change (43)Biofuels have also been studied under the

human-nature nexus frameworkmdashin particularthe energy-food nexus and energy-food-waternexus Rising demand for ethanol feedstocks bid

1258832-4 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 sciencemagorg SCIENCE

Fig 2 Illustrative example of sending receiving and spillover systemsas well as flows under the telecoupling framework In the case of tradein Australian coal in 2004 [measured in megatons (Mt) of CO2 emissions]Australia is a sending system (in blue) the main receiving systems are ingreen (destinations of the coal are Japan South Korea Taiwan MalaysiaIndia the European Union and Brazil) and the spillover systems are in lightred (all other countries and regions are affected by CO2 emissions from

using the coal produced in Australia and consumed in the receivingsystems) The arrows show the magnitude of the flows Countries thatreceive less than 10 Mt of emissions of Australian coal are not includedFlows to Europe are aggregated to include all 28 member states of theEuropean Union [data from (44)] Financial flows take place betweensending and receiving systems but may also affect financial conditions inspillover systems indirectly [data from (98)]

RESEARCH | REVIEW

up food price (67) which has major implicationsfor food security (68) And questions have beenraised about the adverse interplay between bio-fuel mandates and increased interannual varia-bility in crop production anticipated under futureclimate change (69) Water also comes into playas limits on the future availability of water forirrigated agriculture will shift the location ofcropland conversion owing to biofuel expansiontoward regions with carbon-rich rainfed agricul-ture Overall accounting for hydrological con-straints boosts estimated GHG emissions fromland use by 25 (70)Telecoupled processes such as international

trade and flows of information (for exampleglobal market prices) cause biofuel programs inone part of the world to translate spatially intoland conversion in other regions (indirect landuse) They have already contributed to croplandexpansion in the United States and overseasand to cascading and spillover effects over longdistances (Fig 3) National biofuel programswhich looked environmentally beneficial at first

blush might in fact lead to increased environ-mental damage when viewed over time and atthe global scale (42 71) Unlike early analyses(42) that assumed that higher prices effectivelyinfluenced all agents in the market equally sub-sequent research has revealed that some agricul-tural suppliers (such as the United States andArgentina) are more closely telecoupled than oth-ers (72) The spatial pattern and extent of landconversion stemming from biofuels are also af-fected by geophysical characteristics such as poten-tial productivity of the newly converted lands

Hidden roles of virtual water

Although many sustainability studies have fo-cused on flows of real material and energy suchas biofuels there has been increasing interest inthe flows of ldquovirtualrdquo material and energy suchas ldquovirtual waterrdquo ldquovirtual energyrdquo ldquovirtual landrdquoand ldquovirtual nutrientsrdquo (73) Virtual resources arethose resources used for production and incor-porated into goods and services in the same waythat related pollution and impacts are embodied

(or hidden) in these products In the case ofwater for example it is used to grow crops raiselivestock and grow their food and produce mar-ketable goods Virtual water is traded amongcountries as goods are traded Globally the vol-ume of virtual water trade and the number oflinks (pairs of trading countries) have bothdoubledfrom 1986 to 2010 (74)With roughly 27 trillionm3

of water traded virtually worldwide in 2010 (74)virtual water trade fromwater-rich countries hashelped mitigate water shortages in water-poorcountries (75) The concept of virtual resourceshas helped analysts think more clearly about thereal risks of resource scarcity and the role thattrade plays in mitigating or worsening those risks(76 77) Targeted trade policies may help to fur-ther prevent water scarcity by encouraging morewater-efficient trade links (78)Virtual water is a good target for systems in-

tegration research because the issues involvedaredynamic across time space andorganizationallevels Global water scarcity issues are inherent-ly temporally sensitive with cumulative effects

SCIENCE sciencemagorg 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 1258832-5

Fig 3 Cascading and spillover effects of biofuel production on landconversion and CO2 emissions as revealed by systems integrationMeeting the Renewable Fuel Standard mandate [to produce 50 Gigaliters(GL) of additional ethanol on top of the 2001 production level] in the UnitedStates reduces the use of petroleumbut requires additional corn area (99)Theexpansion in US corn area leads to reduction in harvested area of oilseeds andother crops in the United StatesThis boosts world prices for these crops andencourages more production of oilseeds and corn in the rest of the worldTheexpansion of cropland in the United States and the rest of the world leads to

more emissions of CO2 and the conversion of pasture and forest lands aroundthe world This land conversion also releases CO2 offsetting the reduction ofCO2 emissions from using less fossil fuels and more biofuels (assuming a 23ethanolpetroleum energy conversion rate) Estimates are on an annual basisover a 30-year production period for the biofuel facilities and in approximateamounts derived from (99) Arrows indicate the direction of influencesSymbols ldquondashrdquo and ldquo+rdquo refer to decrease and increase respectively in land areaethanol fossil fuel or CO2Tg teragrams Mha million ha [Graphics are usedwith permissions from Fotoliacom]

RESEARCH | REVIEW

stemming from legacies of overuse of water in-teractingwithnewdrivers such as climate changeEstimates suggest that global virtual water trademay decrease with climate change because of thedifficulty of growing crops in warmer drier cli-mates [water savings from reduction in growingrice soybeans and wheat may amount to up to15 trillion m3 in 2030 (78)] Also there was aprofound shift in the spatial distribution of hu-man populations (thus water demand) relativeto water distribution over the past few decadesIn 1986 68 of the worldrsquos population was inwater-exporting countries but by 2010 the dis-tribution was almost completely reversed with60 of the global population in water-importingcountries (74) One of the greatest organizationalconcerns related to virtual water is that a fewcountries control themajority of the global tradewhich leaves the market vulnerable to the deci-sionsmade by a few key players (74) In additionthere is unequal distribution of resources withincountries and a tendency for local agrarian com-munities to be marginalized owing to trade dic-tated by country-level agencies (79)The ecosystem services framework has con-

tributed to virtual water research in many waysFor example Canada is a major exporter of vir-tual water worldwide (95 Gm3 year) exportingvirtual water affects the ecosystem services pro-vided by the nationrsquos boreal forests which makeup nearly 60 of Canadarsquos territory (80) Pro-duction of commodities through processes suchas hydroelectric power generation oil extractioncrop irrigation and livestock-rearing contributesto virtual water exports which in turn threatenfreshwater resources that are a key part of theboreal forests Boreal freshwater comprises 80 to90 of Canadarsquos lakes and 25 of the entireEarthrsquos wetlands (80) Removal of water compro-mises the estimated annual gain of $703 billionin ecosystem services that the boreal forests pro-

vide including carbon storage flood control andwater filtering biodiversity conservation andpest control (80)The environmental footprints framework has

informed virtual water research by depicting thewater resources used for production of goodsand services (ldquowater footprintsrdquo) For example2320 Gm3year of the total global water foot-print of 9087 Gm3year comes from virtual watertrade (81) There is a close relationship betweenvirtual water and planetary boundaries becauseone of the nine key planetary boundaries iden-tified is the limit to global freshwater use (21) Arelated conceptmdashldquopeakwaterrdquomdashhelps to illustratehow close global freshwater bodies are to thisthreshold Global water consumption has alreadyreached a peak and begun to decline in manyareas because of limited remaining water (13)Furthermore scarcity in global freshwater is inlarge part linked to the virtual water embeddedin agricultural production and trade (81) Thehuman-nature nexus framework is also usefulfor virtual water research For example a studyon water-food nexus indicates that 76 of virtualwater trade is attributed to crops or crop-derivedproducts (81)Virtual water trade varies spatially and is an

important telecoupling process Themain virtualwater exporters (sending systems) are water-richregions inNorth and South America andAustraliawhereas Mexico Japan China and water-poorregions in Europe are the main importers (re-ceiving systems) (Fig 4) (75) Sending and re-ceiving systems involved in virtual water tradehave dynamic roles Asia recently switched itsvirtual water imports from North America toSouth America (82) On the other hand NorthAmerica has engaged in an increased diversifi-cation of intraregional water trade while tradingwith distant countries in Asia (82) China hasundergone a dramatic increase in virtual water

imports since 2000 via products such as soy-beans from Brazil (nearly doubling from 2001 to2007 and amounting to 13 of the total globalworld water trade) (82) The spatial shift in theuse of soybean products in Brazil from domesticto international has led to water savings in othercountries but at the cost of deforestation inBrazilian Amazon (82) Within-country virtualwater transfer is also common For example vir-tual water flow through grain trade from NorthChina to South China goes in the opposite direc-tion of real water transfer through large projectssuch as the South-to-North Water Transfer Pro-ject that aim to alleviate water shortages inNorth China

Future directions

Despite the substantial progress in systems inte-gration illustratedabovemany important challengesremain For example the integrated frameworkshave been studied largely in isolation althoughthey are interconnected throughhuman activities(for example usingmore ecosystem services maylead to larger environmental footprints) Achiev-ing a greater degree of integration would involveanalyzing and managing coupled human andnatural systems over longer time periods largerspatial extents (for example macrosystems andultimately the entire planet) and across morediverse organizations at different levels Belowwe suggest several ways to advance systems in-tegration with the intent of improving its theo-retical foundations expanding its tool box andproviding broad implications for managementand policy

Incorporate more human and naturalcomponents simultaneously

Although some previous studies have consideredmultiple components of coupled human andnatu-ral systems many components are either not

1258832-6 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 sciencemagorg SCIENCE

Fig 4 Balance and flows of virtual water related to trade of agricultural and industrial products during 1996ndash2005 Net exporters (sending systems)are in green and net importers (receiving systems) are in red The arrows indicate the relative sizes of large gross virtual water flows between sending andreceiving systems (gt 15 Gm3year) Countries without arrows are potential spillover systems of the large virtual water flows Data are from (81)

RESEARCH | REVIEW

considered or treated as exogenous variablesleading to biases and even incorrect conclusionsFor example a food-water nexus study withoutconsidering GHG emissions during groundwaterextraction for irrigation of crops in China under-estimated GHG emissions by as much as 331MtCO2e (83) One way to correct this problem isto convert more variables from exogenous toendogenousmdashinternalize all important relevantvariablesmdashso that their dynamics and feedbackeffects are explicitly studied For instance con-sidering multiple telecoupling processes (suchas species invasion trade disease spread andtechnology transfers) at the same time can helplinkmany seemingly disconnected distant inter-actions Unifying evolutionary approaches suchas seen with complex adaptive systems (84) mayprovide productive ways to integrate disparateideas and understand temporal dynamics andsustainability of coupled systems

Identify and quantify spillover systems

Previous research on issues such as trade oftenfocused on sending and receiving systems (forexample trade partners) with little attention tospillover systems (for example nontrade partners)mdashother systems affected by the interactions betweensending and receiving systems Although someprevious studies have recognized some spillovereffects [such as spatial externalities (85 86)] theywere often on either socioeconomic or environ-mental effects rather than all effects simulta-neously Furthermore they rarely consider othercomponents of spillover systems (causes agentsand flows) as articulated in the telecouplingframework (41) Identifying and quantifying oth-er components of spillover systems related tospillover effects may help understand the mech-anisms behind the spillover effects and developmore effective management strategies Connect-ing spillover systems with sending and receivingsystems through network analysis (87) may gen-erate fruitful outcomes such as the appreciationof dynamic interrelationships among differentsystems

Explicitly account for feedbacks

Human-nature feedbacks are a core componentof coupled systems For instance an importantnegative feedback in Wolong Nature Reserve forgiant pandas in China occurred when deforesta-tion and panda habitat degradation by localhouseholds prompted the government to devel-op and implement new conservation programsthat provide subsidies for local households tomonitor forests and thus reduce deforestationand improve panda habitat (88) This feedbackhas helped forest and habitat recovery whileincreasing income for local households Moreinnovative measures such as this are needed toidentify and use feedbacks as mechanisms forsustainability

Integrate multiple temporal andspatial scales

Human and natural processes and patterns atmultiple scales may be different and they can

interact with each other For example food pro-duction at the local scale may create jobs at thelocal scale but may not affect overall job creationat the global scale Many urban sustainabilityefforts focus on locally specific solutions thatmay not be scalable (7) Thus considering mul-tiple spatial scales at the same time can helpidentify all important factors their interdepen-dence and their effects and nonlinear relation-ships Temporally short-term studies should becombined with long-term studies in order tomaximize the strengths of each approach Forinstance short-term studies may capture morenuanced immediate changes in system behaviorbut long-term studies may account for temporaldynamics time lags cumulative effects legacy ef-fects and other phenomena (such as rare events)that cannot be seen over shorter terms Moresystematic incorporation of human dimensionsto long-term studies such as the Long-term Ecol-ogical Research sites and the National EcologicalObservatory Network is needed

Develop and use new tools

More effective integration requires developingand using powerful tools to overcome difficultbarriers (for example mathematical and com-putational challenges quantification of impactsat one scale on other scales and relationshipsamong patterns and processes across scales andacross borders) and to predict emergence of un-expected threats for sustainability policy andmanagement Examples include spatially explicitlife cycle assessment supply chain analysis andmultilevel modeling Agent-based models areparticularly promising tools because they takeinteractions (such as human adaptation to en-vironmental changes) at different scales into ac-count and model coupled systems as complexadaptive systems Agent-basedmodels create vir-tual worlds thatmimic the real world in contrastto traditional empirical statistical models (suchas econometric models) that are fitted to pastdata and fail when the future differs from thepast and dynamic stochastic general equilibriummodels that presume a perfect world and ignoredisturbances or crises (89) Many agent-basedmodels have beendeveloped in various disciplinesto provide insights on complexities and informa-tion for policymaking in issues such as economicdevelopment andmanagement of common spaces(90 91) However new models are needed toaccount for telecoupled systems Also increasingcomputational powerwill allowagent-basedmod-els to include more agents in larger areas andultimately all important agents across the worldAs more high-resolution data become availableit is necessary to develop and use big data tools(such as distributed databases massively parallelprocessing and cloud computing) for effectiveand efficient searching retrieving analysis andintegration (92)

Translate findings into policy and practice

Systems integration can provide more unbiasedinformation for policy and practice to help clarifyresponsibilities mediate trade-offs reduce con-

flicts and anticipate future trends It is necessaryto foster coordination among multiple nationaland international policies andminimize situationsin which different policies offset one another be-cause of conflicting goals and counterproductiveimplementation Unfortunately institutions andregulations have traditionally focused on singleissues and often do not have the mandate or in-frastructure to address the organizational con-nections and detrimental spillovers The WorldTrade Organization for example has the princi-pal mandate of promoting global trade One ofthe spillover effects of this mission is the globaltransmission of invasive species but the meansto address invasive species are weak comparedwith the forces driving the global market (93)Adopting the telecoupling framework can helpassign responsibilities of addressing spillover ef-fects (such as CO2 emissions and species inva-sion) to consumers and producers (for examplevia regulation at the source of extraction or con-sumption) as well as others such as traders ofgoods and products across space Last govern-ments need to incorporate long-term studies intotheir policies to account for the complex dynam-ics of coupled systems (such as time lags) Moreapplications of systems integration frameworksandmethods such as those discussed in this papercan accelerate understanding and solving globalsustainability challenges

REFERENCES AND NOTES

1 World Commission On Environment and DevelopmentOur Common Future (Oxford Univ Press Oxford 1987)

2 M A Hanjra M E Qureshi Global water crisis and futurefood security in an era of climate change Food Policy 35365ndash377 (2010) doi 101016jfoodpol201005006

3 J Liu et al Complexity of coupled human and naturalsystems Science 317 1513ndash1516 (2007) doi 101126science1144004 pmid 17872436

4 H A Mooney A Duraiappah A Larigauderie Evolution ofnatural and social science interactions in global changeresearch programs Proc Natl Acad Sci USA 110(suppl 1) 3665ndash3672 (2013) doi 101073pnas1107484110pmid 23297237

5 W Steffen et al Global Change and the Earth SystemA Planet Under Pressure IGBP Science 4 (Springer VerlagHeidelberg Germany 2004)

6 Millennium Ecosystem Assessment Ecosystems amp HumanWell-being Synthesis (Island Press Washington DC 2005)

7 K C Seto et al Urban land teleconnections andsustainability Proc Natl Acad Sci USA 109 7687ndash7692(2012) doi 101073pnas1117622109 pmid 22550174

8 E Ostrom A general framework for analyzing sustainabilityof social-ecological systems Science 325 419ndash422 (2009)doi 101126science1172133 pmid 19628857

9 J Liu Forest sustainability in China and implications for atelecoupled world Asia Pacific Pol Stud 1 230ndash250 (2014)

10 J Norberg G S Cumming Complexity Theory for aSustainable Future (Columbia Univ Press New York 2013)

11 S A Levin Ecosystems and the biosphere as complexadaptive systems Ecosystems (NY) 1 431ndash436 (1998)doi 101007s100219900037

12 E F Lambin Conditions for sustainability of humanndashenvironment systems Information motivation andcapacity Glob Environ Change 15 177ndash180 (2005)doi 101016jgloenvcha200506002

13 P H Gleick M Palaniappan Peak water limits to freshwaterwithdrawal and use Proc Natl Acad Sci USA 107 11155ndash11162(2010) doi 101073pnas1004812107 pmid 20498082

14 J Cools et al Coupling a hydrological water qualitymodel and an economic optimization model to set upa cost-effective emission reduction scenario for nitrogenEnviron Model Softw 26 44ndash51 (2011) doi 101016jenvsoft201004017

SCIENCE sciencemagorg 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 1258832-7

RESEARCH | REVIEW

landscape planning for ecosystem services whichallows for coordination across space For exampleit can promote afforestation and reforestation inupland areas above irrigated agricultural systemsthus reducing erosion protecting waterways min-imizing flooding providing drinking water andfacilitating sustainable agricultural production(22) Temporal integration is crucial to quantifythe planetary boundaries framework as short-term fluctuations in key Earth system processesare scaled up to predict long-term trends manyof which cannot be accurately predicted with-out a systems approach (21) Temporal inte-gration can also reveal legacy effects of priorhuman-nature couplings For example carbonfootprints are driven in large part by past landuse (23) A condition termed ldquocarbon lock-inrdquo hasbeen used to describe systems that have evolvedover long time frames to become dependent onfossil fuels (24 25) The fossil energy system iscomprised of long-lived infrastructure such aspower plants which represent an investment infuture CO2 emissions Retiring this infrastruc-ture before the end of its economic or physicallifetimewould entail substantial costs As of 2013it is estimated the global committed emissionsrelated to existing fossil infrastructure are rough-ly 700 billion tons of CO2 (26)

Human-nature nexuses

In contrast to the conventional decision-makingthat takes place within separate disciplines orsectors the human-nature nexus framework rec-ognizes the interdependency between two ormore issues (or nodes) and addresses them to-gether For example the energy-foodnexus consid-ers both the effects of energy on food productionprocessing transporting and consumption andthe effects of food (such as corn) production onthe generation of energy (such as ethanol) (27)The nexus framework can help anticipate other-

wise unforeseen consequences evaluate trade-offsproduce co-benefits and allow the different andoften competing interests to seeka commonground(28) and co-optimization (29) The vast majorityof the 229 human-nature nexus studies recordedin the Web of Science (as of 16 August 2014)analyzed two-node nexuses (80) with only 16and 4 of the nexus studies including three andfour nodes respectively Although the concept offood-energy nexus first appeared in 1982 therewas no paper recorded in the Web of Science formany years The interest in the nexus frameworkhas reemerged recently and has grown rapidlysince 2010 Several two-nodenexuses have receivedspecial focus including energy-water nexus food-waternexus energy-foodnexus air-climate nexushealth-water nexus and energy-national securitynexus Among themore commonly examined three-and four-node nexuses are economy-environment-land nexus and climate-energy-food-water nexusAdding more nodes to a nexus framework leadsto more complexity but also captures greaterreality For instance the climate-energy-foodnexus considers not only the interrelationshipsbetween energy and food but also the relation-ships between energy and climate (for example

energy use emits CO2 and climate change affectsenergy demand such as heating and cooling) andinterrelationships between food and climate (forexample climate changes affect food productionand CO2 is emitted throughout the food produc-tion processing transporting and consumption)The nodes in the nexus framework are me-

diated by and influencemany organizational lev-els For example energy production and use areshaped by international markets and policies atdifferent government organizations and at thesame time influence many trophic levels of ani-mals plants and microorganisms (30) Buildingon the increasing recognition of conceptual in-terconnections among various nodes efforts areunder way to quantify their relationships suchas via hydro-economic modeling (31) structuraland nonstructural economicmodels (32) and lifecycle assessments (33) Scenario analysis is partic-ularly promising for teasing out roles of differentorganizations functioning at different scales(34 35) For example the Agrimonde model hasbeen used to examine intersections between foodand numerous other sectors (including energy andwater) worldwide under different growth and con-sumption scenarios (35) and illustrates the ef-fects of diverse individual governments on globalcross-sector dynamicsTemporal integration is also a key element of

the nexus framework For example recent long-term quantitative integration studies on theeconomy-energy nexus show that reductions inenergy use can have negative impacts on grossdomestic product (GDP) in the short-term butlittle detectable effect over the long term (36)Alternatively one model predicted that a smallincrease in foreign trade in Indonesia will leadto substantial increases in long-term per capitaCO2 emissions although its contribution to CO2

emissions is negligible in the short term (37)

Telecoupling

Many studies on sustainability have been place-based even if they look at coupled systems [forexample the energy-water nexus in the UnitedStates (38)] However economic production andresource use in different regions or countriesmaylead to very different consequences Furthermorethere are increasing distant interactions aroundthe world so that local events have consequencesglobally (39) For example each year several hun-dred million tons of dust from Africa (especiallythe Sahara desert) travel via the air across theAtlantic Ocean to distant places such as theCaribbean where it causes severe impacts in-cluding decline in coral reefs increase in asthmadisease spread and loss of soil fertility (40 41)Greenhouse gases emitted into the atmospherefromapoint source becomemixed and transportedglobally affecting societies and ecosystems fardistant from the point sources of origin Manyof the changes to the biotic composition of localplaces can also affect society regionally and glob-ally through ever-increasing global trade as wellas the often dramatic impact of invasive speciesand disease transmission In other words pat-terns and processes at one place may enhance or

compromise sustainability in other places (42)Human actions [such as production of biofuels(43)] in one place may create unintended con-sequences elsewhere [such as carbon leakage(44) biodiversity losses (45 46) and pollution(47)] Although external factors originating fromother systems are sometimes considered in sus-tainability research and practices they are typi-cally treated as one-way drivers of changes inthe system of interest with little attention to thefeedbacks between the system of interest andother systems (6 42)The framework of telecoupling (socioeconom-

ic and environmental interactions over distances)has been developed to tie distant places together(42) It is a natural extension of the frameworksof coupled human and natural systems and builton disciplinary frameworks such as climate tele-connections (distant interactions between cli-mate systems) urban land teleconnections (landchanges that are linked to underlying urbaniza-tion dynamics) (7) and economic globalization(distant interactions between human systems)So far the telecoupling framework has been ap-plied to a number of important issues acrossspatial scales such as global land-use and land-change science (39 48 49) international landdeals (39) species invasion (39 42) payments forecosystem services programs (50) and trade offood (42) and forest products (9)The framework is particularly effective for un-

derstanding socioeconomic and environmentalinteractions at international scales For examplethe flow of coal from Australia (sending system)to a number of countries and regions (receivingsystems for example Japan the EuropeanUnionand Brazil) reflects abundant Australian coal sup-plies and the demand for coal in receiving sys-tems (Fig 2) The coal trade is facilitated bymany agents in receiving and sending systems(such as government agencies that make coaltrade policies) and international organizations(such as shipping companies) Many other coun-tries (such as those inAfrica) are spillover systemsmdashsystems that may be affected by the coal flowsbecause of financial flows between sending andreceiving systems as well as the CO2 emissionsproduced when the coal is burned Global effortsto address this and similar feedbacks such as theREDD+ program tomitigate CO2 emissions fromdeforestation (51) and the Green Climate Fundto facilitate low-emission projects in developingnations (52) should focus on all countriesThe telecoupling framework can also be use-

ful at regional or national scales For examplethe 20million residents in Chinarsquos capital city ofBeijing (receiving system) receive clean waterfrom the Miyun Reservoir watershed (sendingsystem) more than 100 km away from the city(50) The framework explicitly links agents causesand effects in the sending and receiving sys-tems For instance the quantity and quality ofthe water flows are made possible through thePaddy Land-to-Dry Land program an ecosystemservices payment program that Beijing establishedwith the farmers in the watershedwho convertedrice cultivation in paddy land to corn production

SCIENCE sciencemagorg 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 1258832-3

RESEARCH | REVIEW

in dry land to provide clean water for Beijingin exchange for cash payments (53) Throughsystematic analysis the framework also helpsidentify research and governance gaps such asspillover systemsmdashregions that are affected bywater and cash flows between thewatershed andBeijing but have received little attention fromresearchers and government agencies (50)The telecoupling framework also emphasizes

temporal dynamics A lack of temporal integra-tion may miss key dynamics and create a misun-derstanding of infrequent but drastic changessuch as disasters wars outbreaks of deadly dis-eases such as Ebola (54) regime shifts and pro-found policy changes For instance in theWolongNature Reserve of China designated for conserv-ing the endangered giant pandas the devastat-ing earthquake in 2008 substantially altered thetelecouplings between Wolong and outside sys-tems [for example collapse of tourism and agricul-tural trade (55)] Studies omitting the earthquakeimpacts could misrepresent the mechanisms be-hind the system dynamics (such as increases inlandslides and relocation of households)

Applications of systems integration

Systems integration has been applied success-fully to many sustainability issues IntegratedCoastal ZoneManagement (56) Marine SpatialPlanning (57) andEcosystem-BasedManagement(58) all integratemultiple dimensions for naturalresource management Although some of thesepractices have existed for several decades therehave been continued efforts formore integrationnew advances and novel insights For exampleEcosystem-Based Management has expanded totackle issues not traditionally thought of within

natural resourcemanagement such as food secu-rity (59) politics (60) and disease (61) The ex-amples of biofuels and virtual water below alsoillustrate the importance of systems integrationin detail We chose to focus on these two exam-ples because they are emerging and contentiousglobal phenomena that represent challengingsustainability issues and they have unexpectedand hidden socioeconomic and environmentaleffects that were impossible to reveal withoutsystems integration

Unexpected impacts of biofuels

The environmental and socioeconomic impactsof biofuels have been among the most hotlycontested policy issues over the past decade TheUnited States EuropeanUnion and nearly threedozen other countries in Africa Asia and theAmericas have developed biofuel mandates ortargets (62) This enthusiasm was buoyed by theprospect of displacing high-priced oil importsgenerating rural incomes and contributing toclimate change mitigation As of 2006 it wassuggested that biofuels could be both economi-cally and environmentally beneficial Howeverwith the implementations of these policies andsystems integration research serious concernshave arisen about their geospatial impacts andthe temporal viability of these mandatesBiofuels are a prime topic for systems integra-

tion research because biofuel production andconsumption as well as their impacts vary acrosstime space and organizational levels For in-stance the carbon fluxes after conversion of newcroplands depend not only on below- and above-ground carbon at present but also on the legacyeffects stemming fromhistorical land use such as

land clearing as well as subsequent cultivationand cropping practices (63) The United Statesand Brazil were responsible for 90 of the globalbiofuel production of 105 billion liters in 2011 butseveral other countries with new mandates suchas China Canada and Argentina are increasingthe spatial extent of production (64) Assessingorganizational impacts on biofuel production en-compasses analysis that integrates across institu-tions For example when and where additionalcropland is converted for biofuel production de-pends critically on local national and internationalinstitutions as well as the global supply chainThe systems integration frameworks discussed

above have direct relevance to biofuels For ex-ample the environmental footprint frameworkhas been applied to assess the impacts of bio-fuels It is estimated that the global footprintfrom biofuels was sim072 billion gha in 2010 andexpected to rise by 73 in 2019 (consisting ofland use carbon embodied energy materialsand waste transport and water) (65) From theperspective of ecosystem services biofuels haveboth positive (for example energy) and negative(for example loss of food and freshwater services)impacts Biofuel production also affects severalplanetary boundaries including climate changeland-use change (proportion of cropland) nutri-ent cycles (increased use of phosphorus and nitro-gen) and biodiversity loss (66) For instance it isestimated that corn-based ethanol nearly doublesgreenhouse emissions across the world over a30-year period because of land-use change (43)Biofuels have also been studied under the

human-nature nexus frameworkmdashin particularthe energy-food nexus and energy-food-waternexus Rising demand for ethanol feedstocks bid

1258832-4 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 sciencemagorg SCIENCE

Fig 2 Illustrative example of sending receiving and spillover systemsas well as flows under the telecoupling framework In the case of tradein Australian coal in 2004 [measured in megatons (Mt) of CO2 emissions]Australia is a sending system (in blue) the main receiving systems are ingreen (destinations of the coal are Japan South Korea Taiwan MalaysiaIndia the European Union and Brazil) and the spillover systems are in lightred (all other countries and regions are affected by CO2 emissions from

using the coal produced in Australia and consumed in the receivingsystems) The arrows show the magnitude of the flows Countries thatreceive less than 10 Mt of emissions of Australian coal are not includedFlows to Europe are aggregated to include all 28 member states of theEuropean Union [data from (44)] Financial flows take place betweensending and receiving systems but may also affect financial conditions inspillover systems indirectly [data from (98)]

RESEARCH | REVIEW

up food price (67) which has major implicationsfor food security (68) And questions have beenraised about the adverse interplay between bio-fuel mandates and increased interannual varia-bility in crop production anticipated under futureclimate change (69) Water also comes into playas limits on the future availability of water forirrigated agriculture will shift the location ofcropland conversion owing to biofuel expansiontoward regions with carbon-rich rainfed agricul-ture Overall accounting for hydrological con-straints boosts estimated GHG emissions fromland use by 25 (70)Telecoupled processes such as international

trade and flows of information (for exampleglobal market prices) cause biofuel programs inone part of the world to translate spatially intoland conversion in other regions (indirect landuse) They have already contributed to croplandexpansion in the United States and overseasand to cascading and spillover effects over longdistances (Fig 3) National biofuel programswhich looked environmentally beneficial at first

blush might in fact lead to increased environ-mental damage when viewed over time and atthe global scale (42 71) Unlike early analyses(42) that assumed that higher prices effectivelyinfluenced all agents in the market equally sub-sequent research has revealed that some agricul-tural suppliers (such as the United States andArgentina) are more closely telecoupled than oth-ers (72) The spatial pattern and extent of landconversion stemming from biofuels are also af-fected by geophysical characteristics such as poten-tial productivity of the newly converted lands

Hidden roles of virtual water

Although many sustainability studies have fo-cused on flows of real material and energy suchas biofuels there has been increasing interest inthe flows of ldquovirtualrdquo material and energy suchas ldquovirtual waterrdquo ldquovirtual energyrdquo ldquovirtual landrdquoand ldquovirtual nutrientsrdquo (73) Virtual resources arethose resources used for production and incor-porated into goods and services in the same waythat related pollution and impacts are embodied

(or hidden) in these products In the case ofwater for example it is used to grow crops raiselivestock and grow their food and produce mar-ketable goods Virtual water is traded amongcountries as goods are traded Globally the vol-ume of virtual water trade and the number oflinks (pairs of trading countries) have bothdoubledfrom 1986 to 2010 (74)With roughly 27 trillionm3

of water traded virtually worldwide in 2010 (74)virtual water trade fromwater-rich countries hashelped mitigate water shortages in water-poorcountries (75) The concept of virtual resourceshas helped analysts think more clearly about thereal risks of resource scarcity and the role thattrade plays in mitigating or worsening those risks(76 77) Targeted trade policies may help to fur-ther prevent water scarcity by encouraging morewater-efficient trade links (78)Virtual water is a good target for systems in-

tegration research because the issues involvedaredynamic across time space andorganizationallevels Global water scarcity issues are inherent-ly temporally sensitive with cumulative effects

SCIENCE sciencemagorg 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 1258832-5

Fig 3 Cascading and spillover effects of biofuel production on landconversion and CO2 emissions as revealed by systems integrationMeeting the Renewable Fuel Standard mandate [to produce 50 Gigaliters(GL) of additional ethanol on top of the 2001 production level] in the UnitedStates reduces the use of petroleumbut requires additional corn area (99)Theexpansion in US corn area leads to reduction in harvested area of oilseeds andother crops in the United StatesThis boosts world prices for these crops andencourages more production of oilseeds and corn in the rest of the worldTheexpansion of cropland in the United States and the rest of the world leads to

more emissions of CO2 and the conversion of pasture and forest lands aroundthe world This land conversion also releases CO2 offsetting the reduction ofCO2 emissions from using less fossil fuels and more biofuels (assuming a 23ethanolpetroleum energy conversion rate) Estimates are on an annual basisover a 30-year production period for the biofuel facilities and in approximateamounts derived from (99) Arrows indicate the direction of influencesSymbols ldquondashrdquo and ldquo+rdquo refer to decrease and increase respectively in land areaethanol fossil fuel or CO2Tg teragrams Mha million ha [Graphics are usedwith permissions from Fotoliacom]

RESEARCH | REVIEW

stemming from legacies of overuse of water in-teractingwithnewdrivers such as climate changeEstimates suggest that global virtual water trademay decrease with climate change because of thedifficulty of growing crops in warmer drier cli-mates [water savings from reduction in growingrice soybeans and wheat may amount to up to15 trillion m3 in 2030 (78)] Also there was aprofound shift in the spatial distribution of hu-man populations (thus water demand) relativeto water distribution over the past few decadesIn 1986 68 of the worldrsquos population was inwater-exporting countries but by 2010 the dis-tribution was almost completely reversed with60 of the global population in water-importingcountries (74) One of the greatest organizationalconcerns related to virtual water is that a fewcountries control themajority of the global tradewhich leaves the market vulnerable to the deci-sionsmade by a few key players (74) In additionthere is unequal distribution of resources withincountries and a tendency for local agrarian com-munities to be marginalized owing to trade dic-tated by country-level agencies (79)The ecosystem services framework has con-

tributed to virtual water research in many waysFor example Canada is a major exporter of vir-tual water worldwide (95 Gm3 year) exportingvirtual water affects the ecosystem services pro-vided by the nationrsquos boreal forests which makeup nearly 60 of Canadarsquos territory (80) Pro-duction of commodities through processes suchas hydroelectric power generation oil extractioncrop irrigation and livestock-rearing contributesto virtual water exports which in turn threatenfreshwater resources that are a key part of theboreal forests Boreal freshwater comprises 80 to90 of Canadarsquos lakes and 25 of the entireEarthrsquos wetlands (80) Removal of water compro-mises the estimated annual gain of $703 billionin ecosystem services that the boreal forests pro-

vide including carbon storage flood control andwater filtering biodiversity conservation andpest control (80)The environmental footprints framework has

informed virtual water research by depicting thewater resources used for production of goodsand services (ldquowater footprintsrdquo) For example2320 Gm3year of the total global water foot-print of 9087 Gm3year comes from virtual watertrade (81) There is a close relationship betweenvirtual water and planetary boundaries becauseone of the nine key planetary boundaries iden-tified is the limit to global freshwater use (21) Arelated conceptmdashldquopeakwaterrdquomdashhelps to illustratehow close global freshwater bodies are to thisthreshold Global water consumption has alreadyreached a peak and begun to decline in manyareas because of limited remaining water (13)Furthermore scarcity in global freshwater is inlarge part linked to the virtual water embeddedin agricultural production and trade (81) Thehuman-nature nexus framework is also usefulfor virtual water research For example a studyon water-food nexus indicates that 76 of virtualwater trade is attributed to crops or crop-derivedproducts (81)Virtual water trade varies spatially and is an

important telecoupling process Themain virtualwater exporters (sending systems) are water-richregions inNorth and South America andAustraliawhereas Mexico Japan China and water-poorregions in Europe are the main importers (re-ceiving systems) (Fig 4) (75) Sending and re-ceiving systems involved in virtual water tradehave dynamic roles Asia recently switched itsvirtual water imports from North America toSouth America (82) On the other hand NorthAmerica has engaged in an increased diversifi-cation of intraregional water trade while tradingwith distant countries in Asia (82) China hasundergone a dramatic increase in virtual water

imports since 2000 via products such as soy-beans from Brazil (nearly doubling from 2001 to2007 and amounting to 13 of the total globalworld water trade) (82) The spatial shift in theuse of soybean products in Brazil from domesticto international has led to water savings in othercountries but at the cost of deforestation inBrazilian Amazon (82) Within-country virtualwater transfer is also common For example vir-tual water flow through grain trade from NorthChina to South China goes in the opposite direc-tion of real water transfer through large projectssuch as the South-to-North Water Transfer Pro-ject that aim to alleviate water shortages inNorth China

Future directions

Despite the substantial progress in systems inte-gration illustratedabovemany important challengesremain For example the integrated frameworkshave been studied largely in isolation althoughthey are interconnected throughhuman activities(for example usingmore ecosystem services maylead to larger environmental footprints) Achiev-ing a greater degree of integration would involveanalyzing and managing coupled human andnatural systems over longer time periods largerspatial extents (for example macrosystems andultimately the entire planet) and across morediverse organizations at different levels Belowwe suggest several ways to advance systems in-tegration with the intent of improving its theo-retical foundations expanding its tool box andproviding broad implications for managementand policy

Incorporate more human and naturalcomponents simultaneously

Although some previous studies have consideredmultiple components of coupled human andnatu-ral systems many components are either not

1258832-6 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 sciencemagorg SCIENCE

Fig 4 Balance and flows of virtual water related to trade of agricultural and industrial products during 1996ndash2005 Net exporters (sending systems)are in green and net importers (receiving systems) are in red The arrows indicate the relative sizes of large gross virtual water flows between sending andreceiving systems (gt 15 Gm3year) Countries without arrows are potential spillover systems of the large virtual water flows Data are from (81)

RESEARCH | REVIEW

considered or treated as exogenous variablesleading to biases and even incorrect conclusionsFor example a food-water nexus study withoutconsidering GHG emissions during groundwaterextraction for irrigation of crops in China under-estimated GHG emissions by as much as 331MtCO2e (83) One way to correct this problem isto convert more variables from exogenous toendogenousmdashinternalize all important relevantvariablesmdashso that their dynamics and feedbackeffects are explicitly studied For instance con-sidering multiple telecoupling processes (suchas species invasion trade disease spread andtechnology transfers) at the same time can helplinkmany seemingly disconnected distant inter-actions Unifying evolutionary approaches suchas seen with complex adaptive systems (84) mayprovide productive ways to integrate disparateideas and understand temporal dynamics andsustainability of coupled systems

Identify and quantify spillover systems

Previous research on issues such as trade oftenfocused on sending and receiving systems (forexample trade partners) with little attention tospillover systems (for example nontrade partners)mdashother systems affected by the interactions betweensending and receiving systems Although someprevious studies have recognized some spillovereffects [such as spatial externalities (85 86)] theywere often on either socioeconomic or environ-mental effects rather than all effects simulta-neously Furthermore they rarely consider othercomponents of spillover systems (causes agentsand flows) as articulated in the telecouplingframework (41) Identifying and quantifying oth-er components of spillover systems related tospillover effects may help understand the mech-anisms behind the spillover effects and developmore effective management strategies Connect-ing spillover systems with sending and receivingsystems through network analysis (87) may gen-erate fruitful outcomes such as the appreciationof dynamic interrelationships among differentsystems

Explicitly account for feedbacks

Human-nature feedbacks are a core componentof coupled systems For instance an importantnegative feedback in Wolong Nature Reserve forgiant pandas in China occurred when deforesta-tion and panda habitat degradation by localhouseholds prompted the government to devel-op and implement new conservation programsthat provide subsidies for local households tomonitor forests and thus reduce deforestationand improve panda habitat (88) This feedbackhas helped forest and habitat recovery whileincreasing income for local households Moreinnovative measures such as this are needed toidentify and use feedbacks as mechanisms forsustainability

Integrate multiple temporal andspatial scales

Human and natural processes and patterns atmultiple scales may be different and they can

interact with each other For example food pro-duction at the local scale may create jobs at thelocal scale but may not affect overall job creationat the global scale Many urban sustainabilityefforts focus on locally specific solutions thatmay not be scalable (7) Thus considering mul-tiple spatial scales at the same time can helpidentify all important factors their interdepen-dence and their effects and nonlinear relation-ships Temporally short-term studies should becombined with long-term studies in order tomaximize the strengths of each approach Forinstance short-term studies may capture morenuanced immediate changes in system behaviorbut long-term studies may account for temporaldynamics time lags cumulative effects legacy ef-fects and other phenomena (such as rare events)that cannot be seen over shorter terms Moresystematic incorporation of human dimensionsto long-term studies such as the Long-term Ecol-ogical Research sites and the National EcologicalObservatory Network is needed

Develop and use new tools

More effective integration requires developingand using powerful tools to overcome difficultbarriers (for example mathematical and com-putational challenges quantification of impactsat one scale on other scales and relationshipsamong patterns and processes across scales andacross borders) and to predict emergence of un-expected threats for sustainability policy andmanagement Examples include spatially explicitlife cycle assessment supply chain analysis andmultilevel modeling Agent-based models areparticularly promising tools because they takeinteractions (such as human adaptation to en-vironmental changes) at different scales into ac-count and model coupled systems as complexadaptive systems Agent-basedmodels create vir-tual worlds thatmimic the real world in contrastto traditional empirical statistical models (suchas econometric models) that are fitted to pastdata and fail when the future differs from thepast and dynamic stochastic general equilibriummodels that presume a perfect world and ignoredisturbances or crises (89) Many agent-basedmodels have beendeveloped in various disciplinesto provide insights on complexities and informa-tion for policymaking in issues such as economicdevelopment andmanagement of common spaces(90 91) However new models are needed toaccount for telecoupled systems Also increasingcomputational powerwill allowagent-basedmod-els to include more agents in larger areas andultimately all important agents across the worldAs more high-resolution data become availableit is necessary to develop and use big data tools(such as distributed databases massively parallelprocessing and cloud computing) for effectiveand efficient searching retrieving analysis andintegration (92)

Translate findings into policy and practice

Systems integration can provide more unbiasedinformation for policy and practice to help clarifyresponsibilities mediate trade-offs reduce con-

flicts and anticipate future trends It is necessaryto foster coordination among multiple nationaland international policies andminimize situationsin which different policies offset one another be-cause of conflicting goals and counterproductiveimplementation Unfortunately institutions andregulations have traditionally focused on singleissues and often do not have the mandate or in-frastructure to address the organizational con-nections and detrimental spillovers The WorldTrade Organization for example has the princi-pal mandate of promoting global trade One ofthe spillover effects of this mission is the globaltransmission of invasive species but the meansto address invasive species are weak comparedwith the forces driving the global market (93)Adopting the telecoupling framework can helpassign responsibilities of addressing spillover ef-fects (such as CO2 emissions and species inva-sion) to consumers and producers (for examplevia regulation at the source of extraction or con-sumption) as well as others such as traders ofgoods and products across space Last govern-ments need to incorporate long-term studies intotheir policies to account for the complex dynam-ics of coupled systems (such as time lags) Moreapplications of systems integration frameworksandmethods such as those discussed in this papercan accelerate understanding and solving globalsustainability challenges

REFERENCES AND NOTES

1 World Commission On Environment and DevelopmentOur Common Future (Oxford Univ Press Oxford 1987)

2 M A Hanjra M E Qureshi Global water crisis and futurefood security in an era of climate change Food Policy 35365ndash377 (2010) doi 101016jfoodpol201005006

3 J Liu et al Complexity of coupled human and naturalsystems Science 317 1513ndash1516 (2007) doi 101126science1144004 pmid 17872436

4 H A Mooney A Duraiappah A Larigauderie Evolution ofnatural and social science interactions in global changeresearch programs Proc Natl Acad Sci USA 110(suppl 1) 3665ndash3672 (2013) doi 101073pnas1107484110pmid 23297237

5 W Steffen et al Global Change and the Earth SystemA Planet Under Pressure IGBP Science 4 (Springer VerlagHeidelberg Germany 2004)

6 Millennium Ecosystem Assessment Ecosystems amp HumanWell-being Synthesis (Island Press Washington DC 2005)

7 K C Seto et al Urban land teleconnections andsustainability Proc Natl Acad Sci USA 109 7687ndash7692(2012) doi 101073pnas1117622109 pmid 22550174

8 E Ostrom A general framework for analyzing sustainabilityof social-ecological systems Science 325 419ndash422 (2009)doi 101126science1172133 pmid 19628857

9 J Liu Forest sustainability in China and implications for atelecoupled world Asia Pacific Pol Stud 1 230ndash250 (2014)

10 J Norberg G S Cumming Complexity Theory for aSustainable Future (Columbia Univ Press New York 2013)

11 S A Levin Ecosystems and the biosphere as complexadaptive systems Ecosystems (NY) 1 431ndash436 (1998)doi 101007s100219900037

12 E F Lambin Conditions for sustainability of humanndashenvironment systems Information motivation andcapacity Glob Environ Change 15 177ndash180 (2005)doi 101016jgloenvcha200506002

13 P H Gleick M Palaniappan Peak water limits to freshwaterwithdrawal and use Proc Natl Acad Sci USA 107 11155ndash11162(2010) doi 101073pnas1004812107 pmid 20498082

14 J Cools et al Coupling a hydrological water qualitymodel and an economic optimization model to set upa cost-effective emission reduction scenario for nitrogenEnviron Model Softw 26 44ndash51 (2011) doi 101016jenvsoft201004017

SCIENCE sciencemagorg 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 1258832-7

RESEARCH | REVIEW

in dry land to provide clean water for Beijingin exchange for cash payments (53) Throughsystematic analysis the framework also helpsidentify research and governance gaps such asspillover systemsmdashregions that are affected bywater and cash flows between thewatershed andBeijing but have received little attention fromresearchers and government agencies (50)The telecoupling framework also emphasizes

temporal dynamics A lack of temporal integra-tion may miss key dynamics and create a misun-derstanding of infrequent but drastic changessuch as disasters wars outbreaks of deadly dis-eases such as Ebola (54) regime shifts and pro-found policy changes For instance in theWolongNature Reserve of China designated for conserv-ing the endangered giant pandas the devastat-ing earthquake in 2008 substantially altered thetelecouplings between Wolong and outside sys-tems [for example collapse of tourism and agricul-tural trade (55)] Studies omitting the earthquakeimpacts could misrepresent the mechanisms be-hind the system dynamics (such as increases inlandslides and relocation of households)

Applications of systems integration

Systems integration has been applied success-fully to many sustainability issues IntegratedCoastal ZoneManagement (56) Marine SpatialPlanning (57) andEcosystem-BasedManagement(58) all integratemultiple dimensions for naturalresource management Although some of thesepractices have existed for several decades therehave been continued efforts formore integrationnew advances and novel insights For exampleEcosystem-Based Management has expanded totackle issues not traditionally thought of within

natural resourcemanagement such as food secu-rity (59) politics (60) and disease (61) The ex-amples of biofuels and virtual water below alsoillustrate the importance of systems integrationin detail We chose to focus on these two exam-ples because they are emerging and contentiousglobal phenomena that represent challengingsustainability issues and they have unexpectedand hidden socioeconomic and environmentaleffects that were impossible to reveal withoutsystems integration

Unexpected impacts of biofuels

The environmental and socioeconomic impactsof biofuels have been among the most hotlycontested policy issues over the past decade TheUnited States EuropeanUnion and nearly threedozen other countries in Africa Asia and theAmericas have developed biofuel mandates ortargets (62) This enthusiasm was buoyed by theprospect of displacing high-priced oil importsgenerating rural incomes and contributing toclimate change mitigation As of 2006 it wassuggested that biofuels could be both economi-cally and environmentally beneficial Howeverwith the implementations of these policies andsystems integration research serious concernshave arisen about their geospatial impacts andthe temporal viability of these mandatesBiofuels are a prime topic for systems integra-

tion research because biofuel production andconsumption as well as their impacts vary acrosstime space and organizational levels For in-stance the carbon fluxes after conversion of newcroplands depend not only on below- and above-ground carbon at present but also on the legacyeffects stemming fromhistorical land use such as

land clearing as well as subsequent cultivationand cropping practices (63) The United Statesand Brazil were responsible for 90 of the globalbiofuel production of 105 billion liters in 2011 butseveral other countries with new mandates suchas China Canada and Argentina are increasingthe spatial extent of production (64) Assessingorganizational impacts on biofuel production en-compasses analysis that integrates across institu-tions For example when and where additionalcropland is converted for biofuel production de-pends critically on local national and internationalinstitutions as well as the global supply chainThe systems integration frameworks discussed

above have direct relevance to biofuels For ex-ample the environmental footprint frameworkhas been applied to assess the impacts of bio-fuels It is estimated that the global footprintfrom biofuels was sim072 billion gha in 2010 andexpected to rise by 73 in 2019 (consisting ofland use carbon embodied energy materialsand waste transport and water) (65) From theperspective of ecosystem services biofuels haveboth positive (for example energy) and negative(for example loss of food and freshwater services)impacts Biofuel production also affects severalplanetary boundaries including climate changeland-use change (proportion of cropland) nutri-ent cycles (increased use of phosphorus and nitro-gen) and biodiversity loss (66) For instance it isestimated that corn-based ethanol nearly doublesgreenhouse emissions across the world over a30-year period because of land-use change (43)Biofuels have also been studied under the

human-nature nexus frameworkmdashin particularthe energy-food nexus and energy-food-waternexus Rising demand for ethanol feedstocks bid

1258832-4 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 sciencemagorg SCIENCE

Fig 2 Illustrative example of sending receiving and spillover systemsas well as flows under the telecoupling framework In the case of tradein Australian coal in 2004 [measured in megatons (Mt) of CO2 emissions]Australia is a sending system (in blue) the main receiving systems are ingreen (destinations of the coal are Japan South Korea Taiwan MalaysiaIndia the European Union and Brazil) and the spillover systems are in lightred (all other countries and regions are affected by CO2 emissions from

using the coal produced in Australia and consumed in the receivingsystems) The arrows show the magnitude of the flows Countries thatreceive less than 10 Mt of emissions of Australian coal are not includedFlows to Europe are aggregated to include all 28 member states of theEuropean Union [data from (44)] Financial flows take place betweensending and receiving systems but may also affect financial conditions inspillover systems indirectly [data from (98)]

RESEARCH | REVIEW

up food price (67) which has major implicationsfor food security (68) And questions have beenraised about the adverse interplay between bio-fuel mandates and increased interannual varia-bility in crop production anticipated under futureclimate change (69) Water also comes into playas limits on the future availability of water forirrigated agriculture will shift the location ofcropland conversion owing to biofuel expansiontoward regions with carbon-rich rainfed agricul-ture Overall accounting for hydrological con-straints boosts estimated GHG emissions fromland use by 25 (70)Telecoupled processes such as international

trade and flows of information (for exampleglobal market prices) cause biofuel programs inone part of the world to translate spatially intoland conversion in other regions (indirect landuse) They have already contributed to croplandexpansion in the United States and overseasand to cascading and spillover effects over longdistances (Fig 3) National biofuel programswhich looked environmentally beneficial at first

blush might in fact lead to increased environ-mental damage when viewed over time and atthe global scale (42 71) Unlike early analyses(42) that assumed that higher prices effectivelyinfluenced all agents in the market equally sub-sequent research has revealed that some agricul-tural suppliers (such as the United States andArgentina) are more closely telecoupled than oth-ers (72) The spatial pattern and extent of landconversion stemming from biofuels are also af-fected by geophysical characteristics such as poten-tial productivity of the newly converted lands

Hidden roles of virtual water

Although many sustainability studies have fo-cused on flows of real material and energy suchas biofuels there has been increasing interest inthe flows of ldquovirtualrdquo material and energy suchas ldquovirtual waterrdquo ldquovirtual energyrdquo ldquovirtual landrdquoand ldquovirtual nutrientsrdquo (73) Virtual resources arethose resources used for production and incor-porated into goods and services in the same waythat related pollution and impacts are embodied

(or hidden) in these products In the case ofwater for example it is used to grow crops raiselivestock and grow their food and produce mar-ketable goods Virtual water is traded amongcountries as goods are traded Globally the vol-ume of virtual water trade and the number oflinks (pairs of trading countries) have bothdoubledfrom 1986 to 2010 (74)With roughly 27 trillionm3

of water traded virtually worldwide in 2010 (74)virtual water trade fromwater-rich countries hashelped mitigate water shortages in water-poorcountries (75) The concept of virtual resourceshas helped analysts think more clearly about thereal risks of resource scarcity and the role thattrade plays in mitigating or worsening those risks(76 77) Targeted trade policies may help to fur-ther prevent water scarcity by encouraging morewater-efficient trade links (78)Virtual water is a good target for systems in-

tegration research because the issues involvedaredynamic across time space andorganizationallevels Global water scarcity issues are inherent-ly temporally sensitive with cumulative effects

SCIENCE sciencemagorg 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 1258832-5

Fig 3 Cascading and spillover effects of biofuel production on landconversion and CO2 emissions as revealed by systems integrationMeeting the Renewable Fuel Standard mandate [to produce 50 Gigaliters(GL) of additional ethanol on top of the 2001 production level] in the UnitedStates reduces the use of petroleumbut requires additional corn area (99)Theexpansion in US corn area leads to reduction in harvested area of oilseeds andother crops in the United StatesThis boosts world prices for these crops andencourages more production of oilseeds and corn in the rest of the worldTheexpansion of cropland in the United States and the rest of the world leads to

more emissions of CO2 and the conversion of pasture and forest lands aroundthe world This land conversion also releases CO2 offsetting the reduction ofCO2 emissions from using less fossil fuels and more biofuels (assuming a 23ethanolpetroleum energy conversion rate) Estimates are on an annual basisover a 30-year production period for the biofuel facilities and in approximateamounts derived from (99) Arrows indicate the direction of influencesSymbols ldquondashrdquo and ldquo+rdquo refer to decrease and increase respectively in land areaethanol fossil fuel or CO2Tg teragrams Mha million ha [Graphics are usedwith permissions from Fotoliacom]

RESEARCH | REVIEW

stemming from legacies of overuse of water in-teractingwithnewdrivers such as climate changeEstimates suggest that global virtual water trademay decrease with climate change because of thedifficulty of growing crops in warmer drier cli-mates [water savings from reduction in growingrice soybeans and wheat may amount to up to15 trillion m3 in 2030 (78)] Also there was aprofound shift in the spatial distribution of hu-man populations (thus water demand) relativeto water distribution over the past few decadesIn 1986 68 of the worldrsquos population was inwater-exporting countries but by 2010 the dis-tribution was almost completely reversed with60 of the global population in water-importingcountries (74) One of the greatest organizationalconcerns related to virtual water is that a fewcountries control themajority of the global tradewhich leaves the market vulnerable to the deci-sionsmade by a few key players (74) In additionthere is unequal distribution of resources withincountries and a tendency for local agrarian com-munities to be marginalized owing to trade dic-tated by country-level agencies (79)The ecosystem services framework has con-

tributed to virtual water research in many waysFor example Canada is a major exporter of vir-tual water worldwide (95 Gm3 year) exportingvirtual water affects the ecosystem services pro-vided by the nationrsquos boreal forests which makeup nearly 60 of Canadarsquos territory (80) Pro-duction of commodities through processes suchas hydroelectric power generation oil extractioncrop irrigation and livestock-rearing contributesto virtual water exports which in turn threatenfreshwater resources that are a key part of theboreal forests Boreal freshwater comprises 80 to90 of Canadarsquos lakes and 25 of the entireEarthrsquos wetlands (80) Removal of water compro-mises the estimated annual gain of $703 billionin ecosystem services that the boreal forests pro-

vide including carbon storage flood control andwater filtering biodiversity conservation andpest control (80)The environmental footprints framework has

informed virtual water research by depicting thewater resources used for production of goodsand services (ldquowater footprintsrdquo) For example2320 Gm3year of the total global water foot-print of 9087 Gm3year comes from virtual watertrade (81) There is a close relationship betweenvirtual water and planetary boundaries becauseone of the nine key planetary boundaries iden-tified is the limit to global freshwater use (21) Arelated conceptmdashldquopeakwaterrdquomdashhelps to illustratehow close global freshwater bodies are to thisthreshold Global water consumption has alreadyreached a peak and begun to decline in manyareas because of limited remaining water (13)Furthermore scarcity in global freshwater is inlarge part linked to the virtual water embeddedin agricultural production and trade (81) Thehuman-nature nexus framework is also usefulfor virtual water research For example a studyon water-food nexus indicates that 76 of virtualwater trade is attributed to crops or crop-derivedproducts (81)Virtual water trade varies spatially and is an

important telecoupling process Themain virtualwater exporters (sending systems) are water-richregions inNorth and South America andAustraliawhereas Mexico Japan China and water-poorregions in Europe are the main importers (re-ceiving systems) (Fig 4) (75) Sending and re-ceiving systems involved in virtual water tradehave dynamic roles Asia recently switched itsvirtual water imports from North America toSouth America (82) On the other hand NorthAmerica has engaged in an increased diversifi-cation of intraregional water trade while tradingwith distant countries in Asia (82) China hasundergone a dramatic increase in virtual water

imports since 2000 via products such as soy-beans from Brazil (nearly doubling from 2001 to2007 and amounting to 13 of the total globalworld water trade) (82) The spatial shift in theuse of soybean products in Brazil from domesticto international has led to water savings in othercountries but at the cost of deforestation inBrazilian Amazon (82) Within-country virtualwater transfer is also common For example vir-tual water flow through grain trade from NorthChina to South China goes in the opposite direc-tion of real water transfer through large projectssuch as the South-to-North Water Transfer Pro-ject that aim to alleviate water shortages inNorth China

Future directions

Despite the substantial progress in systems inte-gration illustratedabovemany important challengesremain For example the integrated frameworkshave been studied largely in isolation althoughthey are interconnected throughhuman activities(for example usingmore ecosystem services maylead to larger environmental footprints) Achiev-ing a greater degree of integration would involveanalyzing and managing coupled human andnatural systems over longer time periods largerspatial extents (for example macrosystems andultimately the entire planet) and across morediverse organizations at different levels Belowwe suggest several ways to advance systems in-tegration with the intent of improving its theo-retical foundations expanding its tool box andproviding broad implications for managementand policy

Incorporate more human and naturalcomponents simultaneously

Although some previous studies have consideredmultiple components of coupled human andnatu-ral systems many components are either not

1258832-6 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 sciencemagorg SCIENCE

Fig 4 Balance and flows of virtual water related to trade of agricultural and industrial products during 1996ndash2005 Net exporters (sending systems)are in green and net importers (receiving systems) are in red The arrows indicate the relative sizes of large gross virtual water flows between sending andreceiving systems (gt 15 Gm3year) Countries without arrows are potential spillover systems of the large virtual water flows Data are from (81)

RESEARCH | REVIEW

considered or treated as exogenous variablesleading to biases and even incorrect conclusionsFor example a food-water nexus study withoutconsidering GHG emissions during groundwaterextraction for irrigation of crops in China under-estimated GHG emissions by as much as 331MtCO2e (83) One way to correct this problem isto convert more variables from exogenous toendogenousmdashinternalize all important relevantvariablesmdashso that their dynamics and feedbackeffects are explicitly studied For instance con-sidering multiple telecoupling processes (suchas species invasion trade disease spread andtechnology transfers) at the same time can helplinkmany seemingly disconnected distant inter-actions Unifying evolutionary approaches suchas seen with complex adaptive systems (84) mayprovide productive ways to integrate disparateideas and understand temporal dynamics andsustainability of coupled systems

Identify and quantify spillover systems

Previous research on issues such as trade oftenfocused on sending and receiving systems (forexample trade partners) with little attention tospillover systems (for example nontrade partners)mdashother systems affected by the interactions betweensending and receiving systems Although someprevious studies have recognized some spillovereffects [such as spatial externalities (85 86)] theywere often on either socioeconomic or environ-mental effects rather than all effects simulta-neously Furthermore they rarely consider othercomponents of spillover systems (causes agentsand flows) as articulated in the telecouplingframework (41) Identifying and quantifying oth-er components of spillover systems related tospillover effects may help understand the mech-anisms behind the spillover effects and developmore effective management strategies Connect-ing spillover systems with sending and receivingsystems through network analysis (87) may gen-erate fruitful outcomes such as the appreciationof dynamic interrelationships among differentsystems

Explicitly account for feedbacks

Human-nature feedbacks are a core componentof coupled systems For instance an importantnegative feedback in Wolong Nature Reserve forgiant pandas in China occurred when deforesta-tion and panda habitat degradation by localhouseholds prompted the government to devel-op and implement new conservation programsthat provide subsidies for local households tomonitor forests and thus reduce deforestationand improve panda habitat (88) This feedbackhas helped forest and habitat recovery whileincreasing income for local households Moreinnovative measures such as this are needed toidentify and use feedbacks as mechanisms forsustainability

Integrate multiple temporal andspatial scales

Human and natural processes and patterns atmultiple scales may be different and they can

interact with each other For example food pro-duction at the local scale may create jobs at thelocal scale but may not affect overall job creationat the global scale Many urban sustainabilityefforts focus on locally specific solutions thatmay not be scalable (7) Thus considering mul-tiple spatial scales at the same time can helpidentify all important factors their interdepen-dence and their effects and nonlinear relation-ships Temporally short-term studies should becombined with long-term studies in order tomaximize the strengths of each approach Forinstance short-term studies may capture morenuanced immediate changes in system behaviorbut long-term studies may account for temporaldynamics time lags cumulative effects legacy ef-fects and other phenomena (such as rare events)that cannot be seen over shorter terms Moresystematic incorporation of human dimensionsto long-term studies such as the Long-term Ecol-ogical Research sites and the National EcologicalObservatory Network is needed

Develop and use new tools

More effective integration requires developingand using powerful tools to overcome difficultbarriers (for example mathematical and com-putational challenges quantification of impactsat one scale on other scales and relationshipsamong patterns and processes across scales andacross borders) and to predict emergence of un-expected threats for sustainability policy andmanagement Examples include spatially explicitlife cycle assessment supply chain analysis andmultilevel modeling Agent-based models areparticularly promising tools because they takeinteractions (such as human adaptation to en-vironmental changes) at different scales into ac-count and model coupled systems as complexadaptive systems Agent-basedmodels create vir-tual worlds thatmimic the real world in contrastto traditional empirical statistical models (suchas econometric models) that are fitted to pastdata and fail when the future differs from thepast and dynamic stochastic general equilibriummodels that presume a perfect world and ignoredisturbances or crises (89) Many agent-basedmodels have beendeveloped in various disciplinesto provide insights on complexities and informa-tion for policymaking in issues such as economicdevelopment andmanagement of common spaces(90 91) However new models are needed toaccount for telecoupled systems Also increasingcomputational powerwill allowagent-basedmod-els to include more agents in larger areas andultimately all important agents across the worldAs more high-resolution data become availableit is necessary to develop and use big data tools(such as distributed databases massively parallelprocessing and cloud computing) for effectiveand efficient searching retrieving analysis andintegration (92)

Translate findings into policy and practice

Systems integration can provide more unbiasedinformation for policy and practice to help clarifyresponsibilities mediate trade-offs reduce con-

flicts and anticipate future trends It is necessaryto foster coordination among multiple nationaland international policies andminimize situationsin which different policies offset one another be-cause of conflicting goals and counterproductiveimplementation Unfortunately institutions andregulations have traditionally focused on singleissues and often do not have the mandate or in-frastructure to address the organizational con-nections and detrimental spillovers The WorldTrade Organization for example has the princi-pal mandate of promoting global trade One ofthe spillover effects of this mission is the globaltransmission of invasive species but the meansto address invasive species are weak comparedwith the forces driving the global market (93)Adopting the telecoupling framework can helpassign responsibilities of addressing spillover ef-fects (such as CO2 emissions and species inva-sion) to consumers and producers (for examplevia regulation at the source of extraction or con-sumption) as well as others such as traders ofgoods and products across space Last govern-ments need to incorporate long-term studies intotheir policies to account for the complex dynam-ics of coupled systems (such as time lags) Moreapplications of systems integration frameworksandmethods such as those discussed in this papercan accelerate understanding and solving globalsustainability challenges

REFERENCES AND NOTES

1 World Commission On Environment and DevelopmentOur Common Future (Oxford Univ Press Oxford 1987)

2 M A Hanjra M E Qureshi Global water crisis and futurefood security in an era of climate change Food Policy 35365ndash377 (2010) doi 101016jfoodpol201005006

3 J Liu et al Complexity of coupled human and naturalsystems Science 317 1513ndash1516 (2007) doi 101126science1144004 pmid 17872436

4 H A Mooney A Duraiappah A Larigauderie Evolution ofnatural and social science interactions in global changeresearch programs Proc Natl Acad Sci USA 110(suppl 1) 3665ndash3672 (2013) doi 101073pnas1107484110pmid 23297237

5 W Steffen et al Global Change and the Earth SystemA Planet Under Pressure IGBP Science 4 (Springer VerlagHeidelberg Germany 2004)

6 Millennium Ecosystem Assessment Ecosystems amp HumanWell-being Synthesis (Island Press Washington DC 2005)

7 K C Seto et al Urban land teleconnections andsustainability Proc Natl Acad Sci USA 109 7687ndash7692(2012) doi 101073pnas1117622109 pmid 22550174

8 E Ostrom A general framework for analyzing sustainabilityof social-ecological systems Science 325 419ndash422 (2009)doi 101126science1172133 pmid 19628857

9 J Liu Forest sustainability in China and implications for atelecoupled world Asia Pacific Pol Stud 1 230ndash250 (2014)

10 J Norberg G S Cumming Complexity Theory for aSustainable Future (Columbia Univ Press New York 2013)

11 S A Levin Ecosystems and the biosphere as complexadaptive systems Ecosystems (NY) 1 431ndash436 (1998)doi 101007s100219900037

12 E F Lambin Conditions for sustainability of humanndashenvironment systems Information motivation andcapacity Glob Environ Change 15 177ndash180 (2005)doi 101016jgloenvcha200506002

13 P H Gleick M Palaniappan Peak water limits to freshwaterwithdrawal and use Proc Natl Acad Sci USA 107 11155ndash11162(2010) doi 101073pnas1004812107 pmid 20498082

14 J Cools et al Coupling a hydrological water qualitymodel and an economic optimization model to set upa cost-effective emission reduction scenario for nitrogenEnviron Model Softw 26 44ndash51 (2011) doi 101016jenvsoft201004017

SCIENCE sciencemagorg 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 1258832-7

RESEARCH | REVIEW

up food price (67) which has major implicationsfor food security (68) And questions have beenraised about the adverse interplay between bio-fuel mandates and increased interannual varia-bility in crop production anticipated under futureclimate change (69) Water also comes into playas limits on the future availability of water forirrigated agriculture will shift the location ofcropland conversion owing to biofuel expansiontoward regions with carbon-rich rainfed agricul-ture Overall accounting for hydrological con-straints boosts estimated GHG emissions fromland use by 25 (70)Telecoupled processes such as international

trade and flows of information (for exampleglobal market prices) cause biofuel programs inone part of the world to translate spatially intoland conversion in other regions (indirect landuse) They have already contributed to croplandexpansion in the United States and overseasand to cascading and spillover effects over longdistances (Fig 3) National biofuel programswhich looked environmentally beneficial at first

blush might in fact lead to increased environ-mental damage when viewed over time and atthe global scale (42 71) Unlike early analyses(42) that assumed that higher prices effectivelyinfluenced all agents in the market equally sub-sequent research has revealed that some agricul-tural suppliers (such as the United States andArgentina) are more closely telecoupled than oth-ers (72) The spatial pattern and extent of landconversion stemming from biofuels are also af-fected by geophysical characteristics such as poten-tial productivity of the newly converted lands

Hidden roles of virtual water

Although many sustainability studies have fo-cused on flows of real material and energy suchas biofuels there has been increasing interest inthe flows of ldquovirtualrdquo material and energy suchas ldquovirtual waterrdquo ldquovirtual energyrdquo ldquovirtual landrdquoand ldquovirtual nutrientsrdquo (73) Virtual resources arethose resources used for production and incor-porated into goods and services in the same waythat related pollution and impacts are embodied

(or hidden) in these products In the case ofwater for example it is used to grow crops raiselivestock and grow their food and produce mar-ketable goods Virtual water is traded amongcountries as goods are traded Globally the vol-ume of virtual water trade and the number oflinks (pairs of trading countries) have bothdoubledfrom 1986 to 2010 (74)With roughly 27 trillionm3

of water traded virtually worldwide in 2010 (74)virtual water trade fromwater-rich countries hashelped mitigate water shortages in water-poorcountries (75) The concept of virtual resourceshas helped analysts think more clearly about thereal risks of resource scarcity and the role thattrade plays in mitigating or worsening those risks(76 77) Targeted trade policies may help to fur-ther prevent water scarcity by encouraging morewater-efficient trade links (78)Virtual water is a good target for systems in-

tegration research because the issues involvedaredynamic across time space andorganizationallevels Global water scarcity issues are inherent-ly temporally sensitive with cumulative effects

SCIENCE sciencemagorg 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 1258832-5

Fig 3 Cascading and spillover effects of biofuel production on landconversion and CO2 emissions as revealed by systems integrationMeeting the Renewable Fuel Standard mandate [to produce 50 Gigaliters(GL) of additional ethanol on top of the 2001 production level] in the UnitedStates reduces the use of petroleumbut requires additional corn area (99)Theexpansion in US corn area leads to reduction in harvested area of oilseeds andother crops in the United StatesThis boosts world prices for these crops andencourages more production of oilseeds and corn in the rest of the worldTheexpansion of cropland in the United States and the rest of the world leads to

more emissions of CO2 and the conversion of pasture and forest lands aroundthe world This land conversion also releases CO2 offsetting the reduction ofCO2 emissions from using less fossil fuels and more biofuels (assuming a 23ethanolpetroleum energy conversion rate) Estimates are on an annual basisover a 30-year production period for the biofuel facilities and in approximateamounts derived from (99) Arrows indicate the direction of influencesSymbols ldquondashrdquo and ldquo+rdquo refer to decrease and increase respectively in land areaethanol fossil fuel or CO2Tg teragrams Mha million ha [Graphics are usedwith permissions from Fotoliacom]

RESEARCH | REVIEW

stemming from legacies of overuse of water in-teractingwithnewdrivers such as climate changeEstimates suggest that global virtual water trademay decrease with climate change because of thedifficulty of growing crops in warmer drier cli-mates [water savings from reduction in growingrice soybeans and wheat may amount to up to15 trillion m3 in 2030 (78)] Also there was aprofound shift in the spatial distribution of hu-man populations (thus water demand) relativeto water distribution over the past few decadesIn 1986 68 of the worldrsquos population was inwater-exporting countries but by 2010 the dis-tribution was almost completely reversed with60 of the global population in water-importingcountries (74) One of the greatest organizationalconcerns related to virtual water is that a fewcountries control themajority of the global tradewhich leaves the market vulnerable to the deci-sionsmade by a few key players (74) In additionthere is unequal distribution of resources withincountries and a tendency for local agrarian com-munities to be marginalized owing to trade dic-tated by country-level agencies (79)The ecosystem services framework has con-

tributed to virtual water research in many waysFor example Canada is a major exporter of vir-tual water worldwide (95 Gm3 year) exportingvirtual water affects the ecosystem services pro-vided by the nationrsquos boreal forests which makeup nearly 60 of Canadarsquos territory (80) Pro-duction of commodities through processes suchas hydroelectric power generation oil extractioncrop irrigation and livestock-rearing contributesto virtual water exports which in turn threatenfreshwater resources that are a key part of theboreal forests Boreal freshwater comprises 80 to90 of Canadarsquos lakes and 25 of the entireEarthrsquos wetlands (80) Removal of water compro-mises the estimated annual gain of $703 billionin ecosystem services that the boreal forests pro-

vide including carbon storage flood control andwater filtering biodiversity conservation andpest control (80)The environmental footprints framework has

informed virtual water research by depicting thewater resources used for production of goodsand services (ldquowater footprintsrdquo) For example2320 Gm3year of the total global water foot-print of 9087 Gm3year comes from virtual watertrade (81) There is a close relationship betweenvirtual water and planetary boundaries becauseone of the nine key planetary boundaries iden-tified is the limit to global freshwater use (21) Arelated conceptmdashldquopeakwaterrdquomdashhelps to illustratehow close global freshwater bodies are to thisthreshold Global water consumption has alreadyreached a peak and begun to decline in manyareas because of limited remaining water (13)Furthermore scarcity in global freshwater is inlarge part linked to the virtual water embeddedin agricultural production and trade (81) Thehuman-nature nexus framework is also usefulfor virtual water research For example a studyon water-food nexus indicates that 76 of virtualwater trade is attributed to crops or crop-derivedproducts (81)Virtual water trade varies spatially and is an

important telecoupling process Themain virtualwater exporters (sending systems) are water-richregions inNorth and South America andAustraliawhereas Mexico Japan China and water-poorregions in Europe are the main importers (re-ceiving systems) (Fig 4) (75) Sending and re-ceiving systems involved in virtual water tradehave dynamic roles Asia recently switched itsvirtual water imports from North America toSouth America (82) On the other hand NorthAmerica has engaged in an increased diversifi-cation of intraregional water trade while tradingwith distant countries in Asia (82) China hasundergone a dramatic increase in virtual water

imports since 2000 via products such as soy-beans from Brazil (nearly doubling from 2001 to2007 and amounting to 13 of the total globalworld water trade) (82) The spatial shift in theuse of soybean products in Brazil from domesticto international has led to water savings in othercountries but at the cost of deforestation inBrazilian Amazon (82) Within-country virtualwater transfer is also common For example vir-tual water flow through grain trade from NorthChina to South China goes in the opposite direc-tion of real water transfer through large projectssuch as the South-to-North Water Transfer Pro-ject that aim to alleviate water shortages inNorth China

Future directions

Despite the substantial progress in systems inte-gration illustratedabovemany important challengesremain For example the integrated frameworkshave been studied largely in isolation althoughthey are interconnected throughhuman activities(for example usingmore ecosystem services maylead to larger environmental footprints) Achiev-ing a greater degree of integration would involveanalyzing and managing coupled human andnatural systems over longer time periods largerspatial extents (for example macrosystems andultimately the entire planet) and across morediverse organizations at different levels Belowwe suggest several ways to advance systems in-tegration with the intent of improving its theo-retical foundations expanding its tool box andproviding broad implications for managementand policy

Incorporate more human and naturalcomponents simultaneously

Although some previous studies have consideredmultiple components of coupled human andnatu-ral systems many components are either not

1258832-6 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 sciencemagorg SCIENCE

Fig 4 Balance and flows of virtual water related to trade of agricultural and industrial products during 1996ndash2005 Net exporters (sending systems)are in green and net importers (receiving systems) are in red The arrows indicate the relative sizes of large gross virtual water flows between sending andreceiving systems (gt 15 Gm3year) Countries without arrows are potential spillover systems of the large virtual water flows Data are from (81)

RESEARCH | REVIEW

considered or treated as exogenous variablesleading to biases and even incorrect conclusionsFor example a food-water nexus study withoutconsidering GHG emissions during groundwaterextraction for irrigation of crops in China under-estimated GHG emissions by as much as 331MtCO2e (83) One way to correct this problem isto convert more variables from exogenous toendogenousmdashinternalize all important relevantvariablesmdashso that their dynamics and feedbackeffects are explicitly studied For instance con-sidering multiple telecoupling processes (suchas species invasion trade disease spread andtechnology transfers) at the same time can helplinkmany seemingly disconnected distant inter-actions Unifying evolutionary approaches suchas seen with complex adaptive systems (84) mayprovide productive ways to integrate disparateideas and understand temporal dynamics andsustainability of coupled systems

Identify and quantify spillover systems

Previous research on issues such as trade oftenfocused on sending and receiving systems (forexample trade partners) with little attention tospillover systems (for example nontrade partners)mdashother systems affected by the interactions betweensending and receiving systems Although someprevious studies have recognized some spillovereffects [such as spatial externalities (85 86)] theywere often on either socioeconomic or environ-mental effects rather than all effects simulta-neously Furthermore they rarely consider othercomponents of spillover systems (causes agentsand flows) as articulated in the telecouplingframework (41) Identifying and quantifying oth-er components of spillover systems related tospillover effects may help understand the mech-anisms behind the spillover effects and developmore effective management strategies Connect-ing spillover systems with sending and receivingsystems through network analysis (87) may gen-erate fruitful outcomes such as the appreciationof dynamic interrelationships among differentsystems

Explicitly account for feedbacks

Human-nature feedbacks are a core componentof coupled systems For instance an importantnegative feedback in Wolong Nature Reserve forgiant pandas in China occurred when deforesta-tion and panda habitat degradation by localhouseholds prompted the government to devel-op and implement new conservation programsthat provide subsidies for local households tomonitor forests and thus reduce deforestationand improve panda habitat (88) This feedbackhas helped forest and habitat recovery whileincreasing income for local households Moreinnovative measures such as this are needed toidentify and use feedbacks as mechanisms forsustainability

Integrate multiple temporal andspatial scales

Human and natural processes and patterns atmultiple scales may be different and they can

interact with each other For example food pro-duction at the local scale may create jobs at thelocal scale but may not affect overall job creationat the global scale Many urban sustainabilityefforts focus on locally specific solutions thatmay not be scalable (7) Thus considering mul-tiple spatial scales at the same time can helpidentify all important factors their interdepen-dence and their effects and nonlinear relation-ships Temporally short-term studies should becombined with long-term studies in order tomaximize the strengths of each approach Forinstance short-term studies may capture morenuanced immediate changes in system behaviorbut long-term studies may account for temporaldynamics time lags cumulative effects legacy ef-fects and other phenomena (such as rare events)that cannot be seen over shorter terms Moresystematic incorporation of human dimensionsto long-term studies such as the Long-term Ecol-ogical Research sites and the National EcologicalObservatory Network is needed

Develop and use new tools

More effective integration requires developingand using powerful tools to overcome difficultbarriers (for example mathematical and com-putational challenges quantification of impactsat one scale on other scales and relationshipsamong patterns and processes across scales andacross borders) and to predict emergence of un-expected threats for sustainability policy andmanagement Examples include spatially explicitlife cycle assessment supply chain analysis andmultilevel modeling Agent-based models areparticularly promising tools because they takeinteractions (such as human adaptation to en-vironmental changes) at different scales into ac-count and model coupled systems as complexadaptive systems Agent-basedmodels create vir-tual worlds thatmimic the real world in contrastto traditional empirical statistical models (suchas econometric models) that are fitted to pastdata and fail when the future differs from thepast and dynamic stochastic general equilibriummodels that presume a perfect world and ignoredisturbances or crises (89) Many agent-basedmodels have beendeveloped in various disciplinesto provide insights on complexities and informa-tion for policymaking in issues such as economicdevelopment andmanagement of common spaces(90 91) However new models are needed toaccount for telecoupled systems Also increasingcomputational powerwill allowagent-basedmod-els to include more agents in larger areas andultimately all important agents across the worldAs more high-resolution data become availableit is necessary to develop and use big data tools(such as distributed databases massively parallelprocessing and cloud computing) for effectiveand efficient searching retrieving analysis andintegration (92)

Translate findings into policy and practice

Systems integration can provide more unbiasedinformation for policy and practice to help clarifyresponsibilities mediate trade-offs reduce con-

flicts and anticipate future trends It is necessaryto foster coordination among multiple nationaland international policies andminimize situationsin which different policies offset one another be-cause of conflicting goals and counterproductiveimplementation Unfortunately institutions andregulations have traditionally focused on singleissues and often do not have the mandate or in-frastructure to address the organizational con-nections and detrimental spillovers The WorldTrade Organization for example has the princi-pal mandate of promoting global trade One ofthe spillover effects of this mission is the globaltransmission of invasive species but the meansto address invasive species are weak comparedwith the forces driving the global market (93)Adopting the telecoupling framework can helpassign responsibilities of addressing spillover ef-fects (such as CO2 emissions and species inva-sion) to consumers and producers (for examplevia regulation at the source of extraction or con-sumption) as well as others such as traders ofgoods and products across space Last govern-ments need to incorporate long-term studies intotheir policies to account for the complex dynam-ics of coupled systems (such as time lags) Moreapplications of systems integration frameworksandmethods such as those discussed in this papercan accelerate understanding and solving globalsustainability challenges

REFERENCES AND NOTES

1 World Commission On Environment and DevelopmentOur Common Future (Oxford Univ Press Oxford 1987)

2 M A Hanjra M E Qureshi Global water crisis and futurefood security in an era of climate change Food Policy 35365ndash377 (2010) doi 101016jfoodpol201005006

3 J Liu et al Complexity of coupled human and naturalsystems Science 317 1513ndash1516 (2007) doi 101126science1144004 pmid 17872436

4 H A Mooney A Duraiappah A Larigauderie Evolution ofnatural and social science interactions in global changeresearch programs Proc Natl Acad Sci USA 110(suppl 1) 3665ndash3672 (2013) doi 101073pnas1107484110pmid 23297237

5 W Steffen et al Global Change and the Earth SystemA Planet Under Pressure IGBP Science 4 (Springer VerlagHeidelberg Germany 2004)

6 Millennium Ecosystem Assessment Ecosystems amp HumanWell-being Synthesis (Island Press Washington DC 2005)

7 K C Seto et al Urban land teleconnections andsustainability Proc Natl Acad Sci USA 109 7687ndash7692(2012) doi 101073pnas1117622109 pmid 22550174

8 E Ostrom A general framework for analyzing sustainabilityof social-ecological systems Science 325 419ndash422 (2009)doi 101126science1172133 pmid 19628857

9 J Liu Forest sustainability in China and implications for atelecoupled world Asia Pacific Pol Stud 1 230ndash250 (2014)

10 J Norberg G S Cumming Complexity Theory for aSustainable Future (Columbia Univ Press New York 2013)

11 S A Levin Ecosystems and the biosphere as complexadaptive systems Ecosystems (NY) 1 431ndash436 (1998)doi 101007s100219900037

12 E F Lambin Conditions for sustainability of humanndashenvironment systems Information motivation andcapacity Glob Environ Change 15 177ndash180 (2005)doi 101016jgloenvcha200506002

13 P H Gleick M Palaniappan Peak water limits to freshwaterwithdrawal and use Proc Natl Acad Sci USA 107 11155ndash11162(2010) doi 101073pnas1004812107 pmid 20498082

14 J Cools et al Coupling a hydrological water qualitymodel and an economic optimization model to set upa cost-effective emission reduction scenario for nitrogenEnviron Model Softw 26 44ndash51 (2011) doi 101016jenvsoft201004017

SCIENCE sciencemagorg 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 1258832-7

RESEARCH | REVIEW

stemming from legacies of overuse of water in-teractingwithnewdrivers such as climate changeEstimates suggest that global virtual water trademay decrease with climate change because of thedifficulty of growing crops in warmer drier cli-mates [water savings from reduction in growingrice soybeans and wheat may amount to up to15 trillion m3 in 2030 (78)] Also there was aprofound shift in the spatial distribution of hu-man populations (thus water demand) relativeto water distribution over the past few decadesIn 1986 68 of the worldrsquos population was inwater-exporting countries but by 2010 the dis-tribution was almost completely reversed with60 of the global population in water-importingcountries (74) One of the greatest organizationalconcerns related to virtual water is that a fewcountries control themajority of the global tradewhich leaves the market vulnerable to the deci-sionsmade by a few key players (74) In additionthere is unequal distribution of resources withincountries and a tendency for local agrarian com-munities to be marginalized owing to trade dic-tated by country-level agencies (79)The ecosystem services framework has con-

tributed to virtual water research in many waysFor example Canada is a major exporter of vir-tual water worldwide (95 Gm3 year) exportingvirtual water affects the ecosystem services pro-vided by the nationrsquos boreal forests which makeup nearly 60 of Canadarsquos territory (80) Pro-duction of commodities through processes suchas hydroelectric power generation oil extractioncrop irrigation and livestock-rearing contributesto virtual water exports which in turn threatenfreshwater resources that are a key part of theboreal forests Boreal freshwater comprises 80 to90 of Canadarsquos lakes and 25 of the entireEarthrsquos wetlands (80) Removal of water compro-mises the estimated annual gain of $703 billionin ecosystem services that the boreal forests pro-

vide including carbon storage flood control andwater filtering biodiversity conservation andpest control (80)The environmental footprints framework has

informed virtual water research by depicting thewater resources used for production of goodsand services (ldquowater footprintsrdquo) For example2320 Gm3year of the total global water foot-print of 9087 Gm3year comes from virtual watertrade (81) There is a close relationship betweenvirtual water and planetary boundaries becauseone of the nine key planetary boundaries iden-tified is the limit to global freshwater use (21) Arelated conceptmdashldquopeakwaterrdquomdashhelps to illustratehow close global freshwater bodies are to thisthreshold Global water consumption has alreadyreached a peak and begun to decline in manyareas because of limited remaining water (13)Furthermore scarcity in global freshwater is inlarge part linked to the virtual water embeddedin agricultural production and trade (81) Thehuman-nature nexus framework is also usefulfor virtual water research For example a studyon water-food nexus indicates that 76 of virtualwater trade is attributed to crops or crop-derivedproducts (81)Virtual water trade varies spatially and is an

important telecoupling process Themain virtualwater exporters (sending systems) are water-richregions inNorth and South America andAustraliawhereas Mexico Japan China and water-poorregions in Europe are the main importers (re-ceiving systems) (Fig 4) (75) Sending and re-ceiving systems involved in virtual water tradehave dynamic roles Asia recently switched itsvirtual water imports from North America toSouth America (82) On the other hand NorthAmerica has engaged in an increased diversifi-cation of intraregional water trade while tradingwith distant countries in Asia (82) China hasundergone a dramatic increase in virtual water

imports since 2000 via products such as soy-beans from Brazil (nearly doubling from 2001 to2007 and amounting to 13 of the total globalworld water trade) (82) The spatial shift in theuse of soybean products in Brazil from domesticto international has led to water savings in othercountries but at the cost of deforestation inBrazilian Amazon (82) Within-country virtualwater transfer is also common For example vir-tual water flow through grain trade from NorthChina to South China goes in the opposite direc-tion of real water transfer through large projectssuch as the South-to-North Water Transfer Pro-ject that aim to alleviate water shortages inNorth China

Future directions

Despite the substantial progress in systems inte-gration illustratedabovemany important challengesremain For example the integrated frameworkshave been studied largely in isolation althoughthey are interconnected throughhuman activities(for example usingmore ecosystem services maylead to larger environmental footprints) Achiev-ing a greater degree of integration would involveanalyzing and managing coupled human andnatural systems over longer time periods largerspatial extents (for example macrosystems andultimately the entire planet) and across morediverse organizations at different levels Belowwe suggest several ways to advance systems in-tegration with the intent of improving its theo-retical foundations expanding its tool box andproviding broad implications for managementand policy

Incorporate more human and naturalcomponents simultaneously

Although some previous studies have consideredmultiple components of coupled human andnatu-ral systems many components are either not

1258832-6 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 sciencemagorg SCIENCE

Fig 4 Balance and flows of virtual water related to trade of agricultural and industrial products during 1996ndash2005 Net exporters (sending systems)are in green and net importers (receiving systems) are in red The arrows indicate the relative sizes of large gross virtual water flows between sending andreceiving systems (gt 15 Gm3year) Countries without arrows are potential spillover systems of the large virtual water flows Data are from (81)

RESEARCH | REVIEW

considered or treated as exogenous variablesleading to biases and even incorrect conclusionsFor example a food-water nexus study withoutconsidering GHG emissions during groundwaterextraction for irrigation of crops in China under-estimated GHG emissions by as much as 331MtCO2e (83) One way to correct this problem isto convert more variables from exogenous toendogenousmdashinternalize all important relevantvariablesmdashso that their dynamics and feedbackeffects are explicitly studied For instance con-sidering multiple telecoupling processes (suchas species invasion trade disease spread andtechnology transfers) at the same time can helplinkmany seemingly disconnected distant inter-actions Unifying evolutionary approaches suchas seen with complex adaptive systems (84) mayprovide productive ways to integrate disparateideas and understand temporal dynamics andsustainability of coupled systems

Identify and quantify spillover systems

Previous research on issues such as trade oftenfocused on sending and receiving systems (forexample trade partners) with little attention tospillover systems (for example nontrade partners)mdashother systems affected by the interactions betweensending and receiving systems Although someprevious studies have recognized some spillovereffects [such as spatial externalities (85 86)] theywere often on either socioeconomic or environ-mental effects rather than all effects simulta-neously Furthermore they rarely consider othercomponents of spillover systems (causes agentsand flows) as articulated in the telecouplingframework (41) Identifying and quantifying oth-er components of spillover systems related tospillover effects may help understand the mech-anisms behind the spillover effects and developmore effective management strategies Connect-ing spillover systems with sending and receivingsystems through network analysis (87) may gen-erate fruitful outcomes such as the appreciationof dynamic interrelationships among differentsystems

Explicitly account for feedbacks

Human-nature feedbacks are a core componentof coupled systems For instance an importantnegative feedback in Wolong Nature Reserve forgiant pandas in China occurred when deforesta-tion and panda habitat degradation by localhouseholds prompted the government to devel-op and implement new conservation programsthat provide subsidies for local households tomonitor forests and thus reduce deforestationand improve panda habitat (88) This feedbackhas helped forest and habitat recovery whileincreasing income for local households Moreinnovative measures such as this are needed toidentify and use feedbacks as mechanisms forsustainability

Integrate multiple temporal andspatial scales

Human and natural processes and patterns atmultiple scales may be different and they can

interact with each other For example food pro-duction at the local scale may create jobs at thelocal scale but may not affect overall job creationat the global scale Many urban sustainabilityefforts focus on locally specific solutions thatmay not be scalable (7) Thus considering mul-tiple spatial scales at the same time can helpidentify all important factors their interdepen-dence and their effects and nonlinear relation-ships Temporally short-term studies should becombined with long-term studies in order tomaximize the strengths of each approach Forinstance short-term studies may capture morenuanced immediate changes in system behaviorbut long-term studies may account for temporaldynamics time lags cumulative effects legacy ef-fects and other phenomena (such as rare events)that cannot be seen over shorter terms Moresystematic incorporation of human dimensionsto long-term studies such as the Long-term Ecol-ogical Research sites and the National EcologicalObservatory Network is needed

Develop and use new tools

More effective integration requires developingand using powerful tools to overcome difficultbarriers (for example mathematical and com-putational challenges quantification of impactsat one scale on other scales and relationshipsamong patterns and processes across scales andacross borders) and to predict emergence of un-expected threats for sustainability policy andmanagement Examples include spatially explicitlife cycle assessment supply chain analysis andmultilevel modeling Agent-based models areparticularly promising tools because they takeinteractions (such as human adaptation to en-vironmental changes) at different scales into ac-count and model coupled systems as complexadaptive systems Agent-basedmodels create vir-tual worlds thatmimic the real world in contrastto traditional empirical statistical models (suchas econometric models) that are fitted to pastdata and fail when the future differs from thepast and dynamic stochastic general equilibriummodels that presume a perfect world and ignoredisturbances or crises (89) Many agent-basedmodels have beendeveloped in various disciplinesto provide insights on complexities and informa-tion for policymaking in issues such as economicdevelopment andmanagement of common spaces(90 91) However new models are needed toaccount for telecoupled systems Also increasingcomputational powerwill allowagent-basedmod-els to include more agents in larger areas andultimately all important agents across the worldAs more high-resolution data become availableit is necessary to develop and use big data tools(such as distributed databases massively parallelprocessing and cloud computing) for effectiveand efficient searching retrieving analysis andintegration (92)

Translate findings into policy and practice

Systems integration can provide more unbiasedinformation for policy and practice to help clarifyresponsibilities mediate trade-offs reduce con-

flicts and anticipate future trends It is necessaryto foster coordination among multiple nationaland international policies andminimize situationsin which different policies offset one another be-cause of conflicting goals and counterproductiveimplementation Unfortunately institutions andregulations have traditionally focused on singleissues and often do not have the mandate or in-frastructure to address the organizational con-nections and detrimental spillovers The WorldTrade Organization for example has the princi-pal mandate of promoting global trade One ofthe spillover effects of this mission is the globaltransmission of invasive species but the meansto address invasive species are weak comparedwith the forces driving the global market (93)Adopting the telecoupling framework can helpassign responsibilities of addressing spillover ef-fects (such as CO2 emissions and species inva-sion) to consumers and producers (for examplevia regulation at the source of extraction or con-sumption) as well as others such as traders ofgoods and products across space Last govern-ments need to incorporate long-term studies intotheir policies to account for the complex dynam-ics of coupled systems (such as time lags) Moreapplications of systems integration frameworksandmethods such as those discussed in this papercan accelerate understanding and solving globalsustainability challenges

REFERENCES AND NOTES

1 World Commission On Environment and DevelopmentOur Common Future (Oxford Univ Press Oxford 1987)

2 M A Hanjra M E Qureshi Global water crisis and futurefood security in an era of climate change Food Policy 35365ndash377 (2010) doi 101016jfoodpol201005006

3 J Liu et al Complexity of coupled human and naturalsystems Science 317 1513ndash1516 (2007) doi 101126science1144004 pmid 17872436

4 H A Mooney A Duraiappah A Larigauderie Evolution ofnatural and social science interactions in global changeresearch programs Proc Natl Acad Sci USA 110(suppl 1) 3665ndash3672 (2013) doi 101073pnas1107484110pmid 23297237

5 W Steffen et al Global Change and the Earth SystemA Planet Under Pressure IGBP Science 4 (Springer VerlagHeidelberg Germany 2004)

6 Millennium Ecosystem Assessment Ecosystems amp HumanWell-being Synthesis (Island Press Washington DC 2005)

7 K C Seto et al Urban land teleconnections andsustainability Proc Natl Acad Sci USA 109 7687ndash7692(2012) doi 101073pnas1117622109 pmid 22550174

8 E Ostrom A general framework for analyzing sustainabilityof social-ecological systems Science 325 419ndash422 (2009)doi 101126science1172133 pmid 19628857

9 J Liu Forest sustainability in China and implications for atelecoupled world Asia Pacific Pol Stud 1 230ndash250 (2014)

10 J Norberg G S Cumming Complexity Theory for aSustainable Future (Columbia Univ Press New York 2013)

11 S A Levin Ecosystems and the biosphere as complexadaptive systems Ecosystems (NY) 1 431ndash436 (1998)doi 101007s100219900037

12 E F Lambin Conditions for sustainability of humanndashenvironment systems Information motivation andcapacity Glob Environ Change 15 177ndash180 (2005)doi 101016jgloenvcha200506002

13 P H Gleick M Palaniappan Peak water limits to freshwaterwithdrawal and use Proc Natl Acad Sci USA 107 11155ndash11162(2010) doi 101073pnas1004812107 pmid 20498082

14 J Cools et al Coupling a hydrological water qualitymodel and an economic optimization model to set upa cost-effective emission reduction scenario for nitrogenEnviron Model Softw 26 44ndash51 (2011) doi 101016jenvsoft201004017

SCIENCE sciencemagorg 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 1258832-7

RESEARCH | REVIEW

considered or treated as exogenous variablesleading to biases and even incorrect conclusionsFor example a food-water nexus study withoutconsidering GHG emissions during groundwaterextraction for irrigation of crops in China under-estimated GHG emissions by as much as 331MtCO2e (83) One way to correct this problem isto convert more variables from exogenous toendogenousmdashinternalize all important relevantvariablesmdashso that their dynamics and feedbackeffects are explicitly studied For instance con-sidering multiple telecoupling processes (suchas species invasion trade disease spread andtechnology transfers) at the same time can helplinkmany seemingly disconnected distant inter-actions Unifying evolutionary approaches suchas seen with complex adaptive systems (84) mayprovide productive ways to integrate disparateideas and understand temporal dynamics andsustainability of coupled systems

Identify and quantify spillover systems

Previous research on issues such as trade oftenfocused on sending and receiving systems (forexample trade partners) with little attention tospillover systems (for example nontrade partners)mdashother systems affected by the interactions betweensending and receiving systems Although someprevious studies have recognized some spillovereffects [such as spatial externalities (85 86)] theywere often on either socioeconomic or environ-mental effects rather than all effects simulta-neously Furthermore they rarely consider othercomponents of spillover systems (causes agentsand flows) as articulated in the telecouplingframework (41) Identifying and quantifying oth-er components of spillover systems related tospillover effects may help understand the mech-anisms behind the spillover effects and developmore effective management strategies Connect-ing spillover systems with sending and receivingsystems through network analysis (87) may gen-erate fruitful outcomes such as the appreciationof dynamic interrelationships among differentsystems

Explicitly account for feedbacks

Human-nature feedbacks are a core componentof coupled systems For instance an importantnegative feedback in Wolong Nature Reserve forgiant pandas in China occurred when deforesta-tion and panda habitat degradation by localhouseholds prompted the government to devel-op and implement new conservation programsthat provide subsidies for local households tomonitor forests and thus reduce deforestationand improve panda habitat (88) This feedbackhas helped forest and habitat recovery whileincreasing income for local households Moreinnovative measures such as this are needed toidentify and use feedbacks as mechanisms forsustainability

Integrate multiple temporal andspatial scales

Human and natural processes and patterns atmultiple scales may be different and they can

interact with each other For example food pro-duction at the local scale may create jobs at thelocal scale but may not affect overall job creationat the global scale Many urban sustainabilityefforts focus on locally specific solutions thatmay not be scalable (7) Thus considering mul-tiple spatial scales at the same time can helpidentify all important factors their interdepen-dence and their effects and nonlinear relation-ships Temporally short-term studies should becombined with long-term studies in order tomaximize the strengths of each approach Forinstance short-term studies may capture morenuanced immediate changes in system behaviorbut long-term studies may account for temporaldynamics time lags cumulative effects legacy ef-fects and other phenomena (such as rare events)that cannot be seen over shorter terms Moresystematic incorporation of human dimensionsto long-term studies such as the Long-term Ecol-ogical Research sites and the National EcologicalObservatory Network is needed

Develop and use new tools

More effective integration requires developingand using powerful tools to overcome difficultbarriers (for example mathematical and com-putational challenges quantification of impactsat one scale on other scales and relationshipsamong patterns and processes across scales andacross borders) and to predict emergence of un-expected threats for sustainability policy andmanagement Examples include spatially explicitlife cycle assessment supply chain analysis andmultilevel modeling Agent-based models areparticularly promising tools because they takeinteractions (such as human adaptation to en-vironmental changes) at different scales into ac-count and model coupled systems as complexadaptive systems Agent-basedmodels create vir-tual worlds thatmimic the real world in contrastto traditional empirical statistical models (suchas econometric models) that are fitted to pastdata and fail when the future differs from thepast and dynamic stochastic general equilibriummodels that presume a perfect world and ignoredisturbances or crises (89) Many agent-basedmodels have beendeveloped in various disciplinesto provide insights on complexities and informa-tion for policymaking in issues such as economicdevelopment andmanagement of common spaces(90 91) However new models are needed toaccount for telecoupled systems Also increasingcomputational powerwill allowagent-basedmod-els to include more agents in larger areas andultimately all important agents across the worldAs more high-resolution data become availableit is necessary to develop and use big data tools(such as distributed databases massively parallelprocessing and cloud computing) for effectiveand efficient searching retrieving analysis andintegration (92)

Translate findings into policy and practice

Systems integration can provide more unbiasedinformation for policy and practice to help clarifyresponsibilities mediate trade-offs reduce con-

flicts and anticipate future trends It is necessaryto foster coordination among multiple nationaland international policies andminimize situationsin which different policies offset one another be-cause of conflicting goals and counterproductiveimplementation Unfortunately institutions andregulations have traditionally focused on singleissues and often do not have the mandate or in-frastructure to address the organizational con-nections and detrimental spillovers The WorldTrade Organization for example has the princi-pal mandate of promoting global trade One ofthe spillover effects of this mission is the globaltransmission of invasive species but the meansto address invasive species are weak comparedwith the forces driving the global market (93)Adopting the telecoupling framework can helpassign responsibilities of addressing spillover ef-fects (such as CO2 emissions and species inva-sion) to consumers and producers (for examplevia regulation at the source of extraction or con-sumption) as well as others such as traders ofgoods and products across space Last govern-ments need to incorporate long-term studies intotheir policies to account for the complex dynam-ics of coupled systems (such as time lags) Moreapplications of systems integration frameworksandmethods such as those discussed in this papercan accelerate understanding and solving globalsustainability challenges

REFERENCES AND NOTES

1 World Commission On Environment and DevelopmentOur Common Future (Oxford Univ Press Oxford 1987)

2 M A Hanjra M E Qureshi Global water crisis and futurefood security in an era of climate change Food Policy 35365ndash377 (2010) doi 101016jfoodpol201005006

3 J Liu et al Complexity of coupled human and naturalsystems Science 317 1513ndash1516 (2007) doi 101126science1144004 pmid 17872436

4 H A Mooney A Duraiappah A Larigauderie Evolution ofnatural and social science interactions in global changeresearch programs Proc Natl Acad Sci USA 110(suppl 1) 3665ndash3672 (2013) doi 101073pnas1107484110pmid 23297237

5 W Steffen et al Global Change and the Earth SystemA Planet Under Pressure IGBP Science 4 (Springer VerlagHeidelberg Germany 2004)

6 Millennium Ecosystem Assessment Ecosystems amp HumanWell-being Synthesis (Island Press Washington DC 2005)

7 K C Seto et al Urban land teleconnections andsustainability Proc Natl Acad Sci USA 109 7687ndash7692(2012) doi 101073pnas1117622109 pmid 22550174

8 E Ostrom A general framework for analyzing sustainabilityof social-ecological systems Science 325 419ndash422 (2009)doi 101126science1172133 pmid 19628857

9 J Liu Forest sustainability in China and implications for atelecoupled world Asia Pacific Pol Stud 1 230ndash250 (2014)

10 J Norberg G S Cumming Complexity Theory for aSustainable Future (Columbia Univ Press New York 2013)

11 S A Levin Ecosystems and the biosphere as complexadaptive systems Ecosystems (NY) 1 431ndash436 (1998)doi 101007s100219900037

12 E F Lambin Conditions for sustainability of humanndashenvironment systems Information motivation andcapacity Glob Environ Change 15 177ndash180 (2005)doi 101016jgloenvcha200506002

13 P H Gleick M Palaniappan Peak water limits to freshwaterwithdrawal and use Proc Natl Acad Sci USA 107 11155ndash11162(2010) doi 101073pnas1004812107 pmid 20498082

14 J Cools et al Coupling a hydrological water qualitymodel and an economic optimization model to set upa cost-effective emission reduction scenario for nitrogenEnviron Model Softw 26 44ndash51 (2011) doi 101016jenvsoft201004017

SCIENCE sciencemagorg 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 1258832-7

RESEARCH | REVIEW

15 S Reis et al From acid rain to climate change Science338 1153ndash1154 (2012) doi 101126science1226514pmid 23197517

16 M Hejazi et al Long-term global water projections using sixsocioeconomic scenarios in an integrated assessmentmodeling framework Technol Forecast Soc 81 205ndash226(2014) doi 101016jtechfore201305006

17 GTAP Global Trade Analysis Project wwwgtapageconpurdueedu (2014)

18 IIASA Global Biosphere Management Model (GLOBIOM)wwwglobiomorg (2014)

19 E M Bennett G D Peterson L J Gordon Understandingrelationships among multiple ecosystem services Ecol Lett12 1394ndash1404 (2009) doi 101111j1461-0248200901387xpmid 19845725

20 A Y Hoekstra T O Wiedmann Humanityrsquos unsustainableenvironmental footprint Science 344 1114ndash1117 (2014)doi 101126science1248365 pmid 24904155

21 J Rockstroumlm et al A safe operating space for humanity Nature461 472ndash475 (2009) doi 101038461472a pmid 19779433

22 C Kremen A Miles Ecosystem services in biologicallydiversified versus conventional farming systems Benefitsexternalities and trade-offs Ecol Soc 17 40 (2012)

23 J Fargione J Hill D Tilman S Polasky P Hawthorne Landclearing and the biofuel carbon debt Science 319 1235ndash1238(2008) doi 101126science1152747 pmid 18258862

24 P J Vergragt N Markusson H Karlsson Carbon captureand storage bio-energy with carbon capture and storage andthe escape from the fossil-fuel lock-in Glob Environ Change 21282ndash292 (2011) doi 101016jgloenvcha201101020

25 G C Unruh Understanding carbon lock-in Energy Policy 28817ndash830 (2000) doi 101016S0301-4215(00)00070-7

26 M R Raupach et al Sharing a quota on cumulative carbonemissions Nature Clim Change 4 873 (2014) doi 101038nclimate2384

27 R L Naylor The Evolving Sphere of Food Security(Oxford Univ Press Oxford 2014)

28 J I Agboola A K Braimoh Strategic partnershipfor sustainable management of aquatic resourcesWater Resour Manage 23 2761ndash2775 (2009) doi 101007s11269-009-9407-4

29 A Santhosh A M Farid K Youcef-Toumi The impact ofstorage facility capacity and ramping capabilities on thesupply side economic dispatch of the energyndashwater nexusEnergy 66 363ndash377 (2014) doi 101016jenergy201401031

30 S C Doney et al Climate change impacts on marineecosystems Annu Rev Mar Sci 4 11ndash37 (2012)doi 101146annurev-marine-041911-111611

31 J Granit M Fogde S Holger Hoff J Joyce Unpacking theWater-Energy-Food Nexus Tools for Assessment andCooperation Along a Continuum (Stockholm InternationalWater Institute Stockholm Sweden 2013) pp 45ndash50

32 G Oladosu S Msangi Biofuel-food market interactions Areview of modeling approaches and findings Agriculture 353ndash71 (2013) doi 103390agriculture3010053

33 S Nair B George H M Malano M Arora B NawarathnaWaterndashenergyndashgreenhouse gas nexus of urban watersystems Review of concepts state-of-art and methodsResour Conserv Recycling 89 1ndash10 (2014) doi 101016jresconrec201405007

34 P Criqui S Mima European climatendashenergy security nexusA model based scenario analysis Energy Policy 41 827ndash842(2012) doi 101016jenpol201111061

35 B Hubert M Rosegrant M A J S van Boekel R Ortiz Thefuture of food Scenarios for 2050 Crop Sci 50 (suppl 1)S33ndashS-50 (2010) doi 102135cropsci2009090530

36 R Coers M Sanders The energyndashGDP nexus Addressing anold question with new methods Energy Econ 36 708ndash715(2013) doi 101016jeneco201211015

37 B Saboori J B Sulaiman S Mohd An empirical analysis of theenvironmental Kuznets curve for CO2 emissions in Indonesia Therole of energy consumption and foreign trade Int J Econ Fin 4243 (2012)

38 A S Stillwell D C Hoppock M E Webber Energy recoveryfrom wastewater treatment plants in the United StatesA case study of the energy-water nexus Sustainability 2945ndash962 (2010) doi 103390su2040945

39 J Liu et al in Rethinking Global Land Use in an Urban EraK Seto A Reenberg Eds (MIT Press Cambridge MA 2014)pp 119ndash139

40 J M Prospero O L Mayol-Bracero Understanding thetransport and impact of African dust on the Caribbean basinBull Am Meteorol Soc 94 1329ndash1337 (2013) doi 101175BAMS-D-12-001421

41 D W Griffin C A Kellogg V H Garrison E A ShinnThe global transport of dust Am Sci 90 228 (2002)doi 10151120023228

42 J Liu et al Framing sustainability in a telecoupled worldEcol Soc 18 art26 (2013) doi 105751ES-05873-180226

43 T Searchinger et al Use of US croplands for biofuelsincreases greenhouse gases through emissions from land-use change Science 319 1238ndash1240 (2008) doi 101126science1151861 pmid 18258860

44 S J Davis G P Peters K Caldeira The supply chain of CO2

emissions Proc Natl Acad Sci USA 108 18554ndash18559(2011) doi 101073pnas1107409108 pmid 22006314

45 M Lenzen et al International trade drives biodiversitythreats in developing nations Nature 486 109ndash112 (2012)doi 101038nature11145 pmid 22678290

46 S L Pimm et al The biodiversity of species and theirrates of extinction distribution and protection Science344 1246752 (2014) doi 101126science1246752pmid 24876501

47 J Bollen S Hers B Van der Zwaan An integratedassessment of climate change air pollution and energysecurity policy Energy Policy 38 4021ndash4030 (2010)doi 101016jenpol201003026

48 H Eakin et al in Rethinking Global Land Use in an Urban EraK Seto A Reenberg Eds (MIT Press Cambridge MA 2014)pp 141ndash161

49 B Wicke in Rethinking Global Land Use in an Urban EraK Seto A Reenberg Eds (MIT Press Cambridge MA 2014)pp 163ndash181

50 J Liu W Yang Integrated assessments of payments forecosystem services programs Proc Natl Acad Sci USA110 16297ndash16298 (2013) doi 101073pnas1316036110pmid 24072648

51 M C Thompson M Baruah E R Carr Seeing REDD+ as aproject of environmental governance Environ Sci Pol 14100ndash110 (2011)

52 F I Khan D S Schinn Triple transformation Nature ClimChange 3 692 (2013) doi 101038nclimate1965

53 H Zheng et al Benefits costs and livelihood implicationsof a regional payment for ecosystem service programProc Natl Acad Sci USA 110 16681ndash16686 (2013)doi 101073pnas1312324110 pmid 24003160

54 L O Gostin D Lucey A Phelan The Ebola epidemicA global health emergency JAMA 312 1095ndash1096 (2014)doi 101001jama201411176 pmid 25111044

55 A Vintildea et al Effects of natural disasters on conservationpolicies The case of the 2008 Wenchuan earthquake ChinaAmbio 40 274ndash284 (2011) doi 101007s13280-010-0098-0pmid 21644456

56 M E Portman L S Esteves X Q Le A Z KhanImproving integration for integrated coastal zonemanagement An eight country study Sci Total Environ439 194ndash201 (2012) doi 101016jscitotenv201209016pmid 23063925

57 C White BS Halpern CV Kappel Ecosystem servicetradeoff analysis reveals the value of marine spatial planningfor multiple ocean uses Proc Natl Acad Sci USA 1094696ndash4701 (2012)

58 R C Szaro W T Sexton C R Malone The emergence ofecosystem management as a tool for meeting peoplersquos needsand sustaining ecosystems Landsc Urban Plan 40 1ndash7(1998) doi 101016S0169-2046(97)00093-5

59 R T Munang I Thiaw M Rivington Ecosystemmanagement Tomorrowrsquos approach to enhancing foodsecurity under a changing climate Sustainability 3 937ndash954(2011) doi 103390su3070937

60 T Hahn Self-organized governance networks for ecosystemmanagement Who is accountable Ecol Soc 16 18 (2011)

61 S E Shackleton C M Shackleton Linking povertyHIVAIDS and climate change to human and ecosystemvulnerability in southern Africa Consequences forlivelihoods and sustainable ecosystem management Int JSustain Dev World Ecol 19 275ndash286 (2012) doi 101080135045092011641039

62 B Digest Biofuels Mandates around the World 2012wwwbiofuelsdigestcombdigest20121122biofuels-mandates-around-the-world-2012 (2014)

63 S J Davis J A Burney J Pongratz K Caldeira Methods forattributing land-use emissions to products Carbon Management5 233ndash245 (2014) doi 101080175830042014913867

64 Worldwatch Institute Biofuels make a comeback despitetough economy (2011) available at wwwworldwatchorgbiofuels-make-comeback-despite-tough-economy

65 G P Hammond S M Seth Carbon and environmentalfootprinting of global biofuel production Appl Energy 112547ndash559 (2013) doi 101016japenergy201301009

66 V Galaz et al lsquoPlanetary boundariesrsquomdashExploring thechallenges for global environmental governance Curr OpinEnviron Sustain 4 80ndash87 (2012) doi 101016jcosust201201006

67 P Westhoff The Economics of Food (Financial Times PressNJ 2010)

68 M Ivanic W Martin Implications of higher global food pricesfor poverty in low-income countries Agric Econ 39 405ndash416(2008) doi 101111j1574-0862200800347x

69 N S Diffenbaugh T W Hertel M Scherer M VermaResponse of corn markets to climate volatility underalternative energy futures Nat Clim Chang 2 514ndash518(2012) pmid 23243468

70 F Taheripour T W Hertel J Liu The role of irrigation indetermining the global land use impacts of biofuels EnergySustain Soc 3 1 (2013)

71 J M Melillo et al Indirect emissions from biofuels Howimportant Science 326 1397ndash1399 (2009) doi 101126science1180251 pmid 19933101

72 N B Villoria T W Hertel Geography matters Internationaltrade patterns and the indirect land use effects of biofuelsAm J Agric Econ 93 919ndash935 (2011) doi 101093ajaeaar025

73 J N Galloway et al International trade in meat The tip ofthe pork chop Ambio 36 622ndash629 (2007) doi 1015790044-7447(2007)36[622ITIMTT]20CO2 pmid 18240675

74 J A Carr P DrsquoOdorico F Laio L Ridolfi Recent history andgeography of virtual water trade PLOS ONE 8 e55825(2013) doi 101371journalpone0055825 pmid 23457481

75 S Suweis A Rinaldo A Maritan P DrsquoOdorico Water-controlledwealth of nations Proc Natl Acad Sci USA 110 4230ndash4233(2013) doi 101073pnas1222452110 pmid 23359709

76 J A Allan Virtual Water A Long Term Solution for WaterShort Middle Eastern Economies (School of Oriental andAfrican Studies University of London 1997)

77 A Y Hoekstra P Hung ldquoVirtual water trade A quantificationof virtual water flows between nations in relation tointernational crop traderdquo Value of Water Research ReportSeries 11 (IHE Delft Delft Netherlands 2002)

78 M Konar Z Hussein N Hanasaki D MauzerallI Rodriguez-Iturbe Virtual water trade flows and savingsunder climate change Hydrol Earth Syst Sci 17 3219ndash3234(2013) doi 105194hess-17-3219-2013

79 R Boelens J Vos The danger of naturalizing water policyconcepts Water productivity and efficiency discourses fromfield irrigation to virtual water trade Agric Water Manage108 16ndash26 (2012) doi 101016jagwat201106013

80 D Schindler P Lee Comprehensive conservation planning toprotect biodiversity and ecosystem services in Canadianboreal regions under a warming climate and increasingexploitation Biol Conserv 143 1571ndash1586 (2010)doi 101016jbiocon201004003

81 A Y Hoekstra M M Mekonnen The water footprint ofhumanity Proc Natl Acad Sci USA 109 3232ndash3237(2012) doi 101073pnas1109936109 pmid 22331890

82 C Dalin M Konar N Hanasaki A Rinaldo I Rodriguez-IturbeEvolution of the global virtual water trade network Proc NatlAcad Sci USA 109 5989ndash5994 (2012) doi 101073pnas1203176109 pmid 22474363

83 J Wang et al Chinarsquos waterndashenergy nexus Greenhouse-gasemissions from groundwater use for agriculture Environ Res Lett7 014035 (2012) doi 1010881748-932671014035

84 S A Levin J Lubchenco Resilience robustness and marineecosystem-based management Bioscience 58 27 (2008)doi 101641B580107

85 B S Halpern S E Lester J B Kellner Spillover from marinereserves and the replenishment of fished stocks EnvironConserv 36 268ndash276 (2009) doi 101017S0376892910000032

86 K S Andam P J Ferraro A Pfaff G A Sanchez-AzofeifaJ A Robalino Measuring the effectiveness of protected areanetworks in reducing deforestation Proc Natl Acad Sci USA105 16089ndash16094 (2008) doi 101073pnas0800437105pmid 18854414

87 B D Anderson S Vongpanitlerd Network Analysis andSynthesis A Modern Systems Theory Approach (CourierDover Publications Mineola NY 2006)

88 W Yang et al Nonlinear effects of group size on collectiveaction and resource outcomes Proc Natl Acad Sci USA110 10916ndash10921 (2013) doi 101073pnas1301733110pmid 23776222

1258832-8 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 sciencemagorg SCIENCE

RESEARCH | REVIEW

89 J D Farmer D Foley The economy needs agent-basedmodelling Nature 460 685ndash686 (2009) doi 101038460685a pmid 19661896

90 M A Janssen E Ostrom Empirically based agent-basedmodels Ecol Soc 11 37 (2006)

91 E Ostrom V Ostrom Choice Rules and Collective ActionThe Ostroms on the Study of Institutions and Governance(ECPR Press Colchester UK 2014)

92 D Agrawal S Das A El Abbadi in Proceedings of the 14thInternational Conference on Extending Database Technology(ACM New York 2011) pp 530ndash533

93 M Margolis J F Shogren Disguised protectionism globaltrade rules and alien invasive species Environ Resour Econ51 105ndash118 (2012) doi 101007s10640-011-9490-x

94 S R Carpenter et al Science for managing ecosystemservices Beyond the Millennium Ecosystem AssessmentProc Natl Acad Sci USA 106 1305ndash1312 (2009)doi 101073pnas0808772106 pmid 19179280

95 C Liu C Kroeze A Y Hoekstra W Gerbens-Leenes Pastand future trends in grey water footprints of anthropogenicnitrogen and phosphorus inputs to major world rivers Ecol Indic18 42ndash49 (2012) doi 101016jecolind201110005

96 CO2Noworg co2noworgCurrent-CO2CO2-Nowglobal-carbon-emissionshtml (2014)

97 T O Wiedmann et al The material footprint of nations ProcNatl Acad Sci USA 101073pnas1220362110 (2013)

98 F Bingham B Perkins Trade Competitiveness amp AdvocacyBranch Australiarsquos coal and iron ore exportsmdash2001ndash2011(2012)

99 T W Hertel et al Effects of US maize ethanol on globalland use and greenhouse gas emissions Estimatingmarket-mediated responses Bioscience 60 223ndash231 (2010)doi 101525bio20106038

100 J Liu et al Ecological degradation in protected areasThe case of Wolong Nature Reserve for giant pandasScience 292 98ndash101 (2001) doi 101126science1058104pmid 11292872

101 J Phelps L R Carrasco E L Webb L P KohU Pascual Agricultural intensification escalates futureconservation costs Proc Natl Acad Sci USA 1107601ndash7606 (2013) doi 101073pnas1220070110pmid 23589860

102 J Rogelj D L McCollum A Reisinger M MeinshausenK Riahi Probabilistic cost estimates for climate changemitigation Nature 493 79ndash83 (2013) doi 101038nature11787 pmid 23282364

103 D Shindell et al Simultaneously mitigating near-termclimate change and improving human health and foodsecurity Science 335 183ndash189 (2012) doi 101126science1210026 pmid 22246768

104 J Reis T Stojanovic H Smith Relevance of systemsapproaches for implementing Integrated Coastal Zone

Management principles in Europe Mar Policy 43 3ndash12(2014) doi 101016jmarpol201303013

105 R Biggs F R Westley S R Carpenter Navigating the backloop Fostering social innovation and transformation inecosystem management Ecol Soc 15 9 (2010)

106 T A Spies et al Examining fire-prone forest landscapes ascoupled human and natural systems Ecol Soc 19 art9(2014) doi 105751ES-06584-190309

107 T C Coverdale N C Herrmann A H Altieri M D BertnessLatent impacts The role of historical human activity incoastal habitat loss Front Ecol Environ 11 69ndash74 (2013)doi 101890120130

108 X Chen et al Producing more grain with lowerenvironmental costs Nature 514 486ndash489 (2014)doi 101038nature13609 pmid 25186728

ACKNOWLEDGMENTS

We thank the US National Science Foundation the InternationalNetwork of Research on Coupled Human and Natural SystemsMichigan State University and Michigan AgBioResearch forfinancial support We are grateful to J Broderick W McConnellJ McCoy S Nichols and W Yang for assistance and to twoanonymous reviewers for helpful comments

101126science1258832

SCIENCE sciencemagorg 27 FEBRUARY 2015 bull VOL 347 ISSUE 6225 1258832-9

RESEARCH | REVIEW

DOI 101126science1258832 (2015)347 Science

et alJianguo LiuSystems integration for global sustainability

This copy is for your personal non-commercial use only

clicking herecolleagues clients or customers by you can order high-quality copies for yourIf you wish to distribute this article to others

herefollowing the guidelines

can be obtained byPermission to republish or repurpose articles or portions of articles

) February 26 2015 wwwsciencemagorg (this information is current as of

The following resources related to this article are available online at

httpwwwsciencemagorgcontent34762251258832fullhtmlversion of this article at

including high-resolution figures can be found in the onlineUpdated information and services

httpwwwsciencemagorgcontent34762251258832fullhtmlref-list-1 27 of which can be accessed freecites 86 articlesThis article

httpwwwsciencemagorgcgicollectionengineeringEngineering

httpwwwsciencemagorgcgicollectionecologyEcology

subject collectionsThis article appears in the following

registered trademark of AAAS is aScience2015 by the American Association for the Advancement of Science all rights reserved The title

CopyrightAmerican Association for the Advancement of Science 1200 New York Avenue NW Washington DC 20005 (print ISSN 0036-8075 online ISSN 1095-9203) is published weekly except the last week in December by theScience

on

Feb

ruar

y 26

201

5w

ww

sci

ence

mag

org

Dow

nloa

ded

from