Water saving potentials and possible trade-offs for future food and energy supply
Kerstin Damerau
a,*, Anthony G. Patt
a,b, Oscar P.R. van Vliet
aaDepartmentofEnvironmentalSystemsScience,SwissFederalInstituteofTechnologyZurich(ETHZ),Universitätstrasse22,CHNJ72.1,8092Zurich, Switzerland
bInternationalInstituteforAppliedSystemsAnalysis(IIASA),Schlossplatz1,2361Laxenburg,Austria
ARTICLE INFO Articlehistory:
Received31July2015
Receivedinrevisedform21March2016 Accepted30March2016
Availableonline23April2016 Keywords:
Water-energy-foodnexus Biofuels
Naturalresourcemanagement
ABSTRACT
Thesufficientsupplyoffoodandenergyrequireslargeamountsoffreshwater.Mainlyrequiredfor irrigation,butalsoprocessingandcoolingpurposes,waterisoneoftheessentialresourcesinboth sectors.Risingglobalpopulationnumbersandeconomicdevelopmentcouldlikelycauseanincreasein naturalresourcedemandoverthecomingdecades,whileatthesametimeclimatechangemightleadto loweroverallwateravailability.Theresultcouldbeanincreasedcompetitionforwaterresourcesmainly inwater-stressedregionsoftheworldinthefuture.Inthisstudyweexploreasetofpossiblechangesin consumptionpatternsintheagriculturalandenergysectorthatcouldbeprimarilymotivatedbyother goalsthanwaterconservationmeasures—forexamplepersonalhealthandclimatechangemitigation targets,andestimatetheindirecteffectsuchtrendswouldhaveonglobalwaterrequirementsuntil2050.
Lookingatfiveworldregions,weinvestigatedthreepossiblechangesregardingfuturefoodpreferences, andtwopossiblechangesinfutureresourcepreferencesforelectricityandtransportfuels.Wefindthat whileanincreaseinfoodsupplyasaresultofhigherproteindemandwouldleadtoanincreaseinwater demandaswell,thistrendcouldbecounteractedbyotherpotentialdietaryshiftssuchasareductionin grainsandsugars.Intheenergysectorwefindthatanincreasingwaterdemandcanbelimitedthrough specificresourceandtechnologychoices,whileasignificantgrowthoffirst-generationbiofuelswould leadtoadrasticriseinwaterdemand,potentiallyexceedingthewaterrequirementsforfoodsupply.
Lookingatthetwosectorstogether,weconcludethatanoverallincreaseinwaterdemandforbothfood andenergyisnotinevitableandthatchangesinfoodandenergypreferencescouldindeedleadtoan alleviationofwaterresourceusedespiterisingpopulationnumbers.
ã2016TheAuthors.PublishedbyElsevierLtd.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).
1.Introduction
Thetypesoffoodsandenergyweconsumehaveconsiderable directandindirect effectsonglobalfreshwateruse.Byfarmostwater resourcesgetusedforirrigationpurposesintheagriculturalsector, mainlyforfoodproduction.Inanothersector,energy,electricityand fuelproductionrequiresincreasingamountsofwater,mainlyfor resourceextractionandcooling(Macknicketal.,2012;Mielkeand Anadon,2010). In some regionsthis trend has alreadyled to a competitionbetweendifferentwaterusers(TheWorldBank,2014).
Risingglobalpopulationnumbersandsocio-economicdevelopment couldleadtoafurtherincreaseinwaterdemandinbothsectorsover thecomingthreetofourdecades.Atthesametimeenvironmental
changeslikeclimatechangemightdecreasethewateravailability andqualityinmanypartsofthe world(JiménezCisnerosetal.,2014).
Hence,thesourceandtypeoffood,electricity,andtransportfuelwe chooseinthefuturecaneitheracceleratearisingwaterdemandor offset increasing resource needs, depending on the effects of consumer preferences and policy initiatives on consumption patternsinbothsectors.Waterisoneofthemostimportantnatural resourcesandtheinteractionsbetweenwateruse,energydemand andfoodproductionarecomplex,aschangesinthedemandofone resourceinonesectorcanchangeitsavailabilityandthatofanother resourceinanothersectorandviceversa.Waterisusedandre-used forfood,electricity,andfuelproduction,whileenergyisrequiredfor agricultureandwatersupply,creatingpositivefeedbackloopsthat canaggravatealready existing water shortages orgeneratenew ones.
Overthelastdecadeanumberofscientificpapersandpolicy reportshaveexaminedtheinteractionsbetweentheagriculture andenergysectorfromanaturalresourceperspective.Resources
* Correspondingauthor.
E-mailaddress:kerstin.damerau@usys.ethz.ch(K.Damerau).
http://dx.doi.org/10.1016/j.gloenvcha.2016.03.014
0959-3780/ã2016TheAuthors.PublishedbyElsevierLtd.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/).
ContentslistsavailableatScienceDirect
Global Environmental Change
j o u r n al h o m ep a g e: w w w . el s e v i e r . c o m / l o c at e / g l o e n vc h a
that received highest attention with regard to their regionally interrelatedavailabilityarewater,energy,andtosomeextentland.
Atermthatisoftenusedtodescribethisinterconnectionistheso- calledwater-energy-(land)-food(WE(L)F)nexus,i.e.theinterac- tionregardingwaterthatisrequiredforfoodandenergy,energy requiredforwaterandfood,andlandrequiredforfoodandenergy supply.Therearequalitativeandquantitativeapproaches,aswell asglobalandregionalstudiescoveringeitherspecificpartsofthe WE(L)Fnexusortryingtointegrateseveralresourceinterdepen- denciesatthesametime,searchingfor trade-offsand potential conflicts.Existingstudiesonthistopicdiscussagrowingscarcityof naturalresourcesduetorisingpopulationnumbersandeconomic development,andtheirpotentialsocialimplications,whilemostof themfocusonthewater-foodorwater-energynexus.
Withinthiscontext,animportantissuethathasnotyetbeen examinedinthescientificliteraturearetheeffectsthatpotential changesinconsumerpreferencescouldhaveonnaturalresource use.Theamountsofwaterthatgetconsumedforsupplyingfood, electricity,andtransportfuelcanvaryvastlydependingontypeof foodandenergysourcechosen.Inthisstudy,weaddressthisvery question:howanincreasingglobalpercapitaandoveralldemand forfoodandenergywouldpotentiallybeinfluencedthroughaset ofdifferentconsumptiontrendsregardingchangesindietaryand energy source preferences. In the form of a global high-level quantification for water consumption in the agricultural and energysector,wemodelthewateruseforirrigation,coolingand processingpurposesinfiveworldregionsasdefinedfortheshared socio-economicpathways(SSPs)usedforthelatestIPCCassess- mentreport(Fieldetal.,2014).Ouraimistocompareandevaluate thewaterconsumptionsharesforfood,electricity,andtransport fuelsuntil2050anddetectglobalandregionalpatternsinwater demandacrossthesetwosectors.Throughthisintegratedanalysis wewillbeabletoidentifyasetofrelativeandcombinedeffectsof resource preference changes on the presumablysteadily rising waterdemandinbothsectors.
2.Background
A numberofrecent qualitativeandquantitative papershave discussed theWE(L)F nexusin general and particular resource interactions,oftenfocusingonspecificpartsoftheworldwhichare characterizedbysignificantnaturalresourcescarcityandcompe- tition.Afirstsetofstudieshaslookedat(aspectsof)theWE(L)F nexusonaqualitative basis.Ringler etal.(2013) discussedthe linkagesofwaterandfood,energyandwater,energy-food,land- energy,and energy-land, and underlined theimportance of an integratedmanagementapproach.Halsteadetal.(2014)reviewed thecurrentliteratureonthe WEFnexus,though didnotrelate water use shares of both sectors to each other. FAO (2014) examined the WEF nexus as a new approach tosupport food securityandsustainableagriculture.Bogardietal.(2012)analyzed theinterconnectedchallengesforwatersecurityforaplanetfacing increasingregionalwaterstressduetorisingpopulation,climate change,urbanizationanddevelopment,callingforanintegrated managementframeworkinordertoaddressallofthesechallenges simultaneously.DeFraitureetal.(2010)discussedcomprehensive assessmentmethodsforwatermanagement inagriculture.Also Rosegrantetal.(2009)focusedonthewateruseintensityofthe agriculturalsectorandhowtomaintainfoodsecuritywhilewater stress increases with an emphasis on improving efficiencies.
Hellegerset al. (2008) presented a debate on the interactions betweenwater, energy, foodand environment witha focus on water-relatedpolicyissues.Allouche(2011) lookedatwaterand foodsecuritypredominantlyfromasocialandpoliticalperspec- tive,doingsoonaglobal,regionalandnationalscale.Harveyand Pilgrim (2011) explored the “new competition for land”,
integratingfood,energyandclimatechangeintotheirdiscussion.
All of these studies have envisioned a drastic rise in natural resourcedemandbased onanextrapolationof currentrequire- mentstofuturepopulationnumbersandongoingsocio-economic developmenttrends,andhencehavecalledforanintegratedpolicy andmanagementframework.
Anothersetof studieshastriedtoquantifynatural resource interconnections on a global level. Hanjra and Qureshi (2010) analyzed expectedreduced global water availabilityand future foodsecurity,reviewingquantitativeresultsfrompreviousstudies tounderlinetheseverityoflimitedwaterresourcesforagriculture overthecomingdecades.ChartresandSood(2013)undertooka global quantitative analysis for the water demand for food productionuntil2050.UsingtheWATERSIMmodeltheydeveloped threescenarioswithdifferingassumptionsonpopulationandGDP growthrateswheretheyextrapolatedcurrentdietarypatterns,but did not integrate a discussion on potential changes in future consumerpreferences.Allscenarios showanincrease in global water demand for agriculture from 2400km3/yr in 2010 to between3820and7230km3/yrin2050.Sulseretal.(2010)used IFPRI’sIMPACTmodelfortheiranalysisoftheNileandGangesriver basins, including a set of global scenarios that illustrate the potentialgrowthratesofconsumptivewateruseintheagricultural sector until the mid-century depending on global per capita income growth. They projected an increase from 1425km3/yr irrigation (blue)water demand for cropproduction in 2000to 1785km3/yrin2050intheirbaselinescenario.
AthirdsetofstudiesfollowedaregionalapproachtotheWE(L)F nexus. Lele et al. (2013) debated governance issues when integrating food, water and energy security, including a case studyforwatermanagementinChinaandIndia.Gulatietal.(2013) presenteda national WEF studyforSouth Africa, exploring the interdependenciesofthesethreeresources,includinganeconomic analysis.Hardyetal.(2012)undertookaquantitativeanalysisof the water-energy nexus for Spain, calculating a potentially increasing waterdemand for energy supply. Scottet al. (2011) looked at the policyand institutional dimensionof thewater- energy nexus including cases studies from the United States, highlightingtheroleofintegratedlocalwatermanagement.Khan etal.(2009)presentedwaystoreducewaterandenergydemand forgrainproductioninAustralia.Larson(2013)analyzedthewater demandforalternativefoodsecuritypoliciesintheMiddleEast andNorthAfrica,focusingonwheatproductionandtrade.Rasul (2014) studied food, water and energy security in South Asia.
Lawfordetal.(2013)gaveabasinperspectiveontheWEFsecurity nexus,usingresultsfromcasestudiesfromdifferentlargeriver basins.Perroneetal. (2011)presented anintegratedqualitative analysisframeworkforthewater-energynexusonthecommunity level.Inalloftheseregionalanalysesnaturalresourceavailability isexpectedtodeclinedue torisingdemands andsimultaneous adverseecologicalchanges.
Therehavealsobeenseveralregionalandglobalstudieslooking particularlyatthewater(andland)demandofenergyintheform ofbiofuels,andtheirpotentiallynegativeimpactsonfoodsecurity andwateravailabilitywhenscalingupbiofuelproductioninthe future.Dominquez-Fausetal.(2009)analyzedthewaterrequire- mentsformaizeasenergycropintheUS,concludingthatamajor shifttosuchanenergysourcewouldhavelargedetrimentaleffects regardingwateravailabilityandenvironmentalhealth.Fingerman etal.(2010)examinedthewaterimpactsofproducingbioethanol inacomprehensiveenvironmentalassessmentwithacasestudy forCalifornia,findingthattheproductionofethanolfrommaizeor sugarbeetswouldrequireenormousamountsofwaterwithupto 5100L/Lethanol.Yangetal.(2009)calculatedthelandandwater requirementsforbiofuelproductioninChinaanditspotentially adverseconsequencesforfoodsupplyandtheenvironment.Using
theWATERSIMmodel,Fraitureetal.(2008)lookedatinternational biofuelpoliciesandtheirimplicationsfor waterdemandin the agriculturalsectorona globallevel. Theyputemphasis onthe countriesChinaandIndia,whereafastgrowingenergydemand and limited water resources could lead to strong resource competition in the future were biofuels utilized as one of the main transport fuels. Globally they estimated irrigation water withdrawalsforbioethanolof30.6km3/yrin2005,anamountthat couldriseto128.4km3/yrin2030.
Givencurrentconsumptionpatterns,ahighpercapitasupplyof foodand energy, rising global population numbers, and socio- economicdevelopment,allcallingforhighnaturalresourceinputs, andtheirresultingecologicalconsequences likeclimatechange aggravatingregionalresourcescarcity, everyone of theWE(L)F studiesundertakensofarpictureanincreasingresourcedemand forthecomingdecades,andconsequentlyunderlinethenecessity forbetter,integratedmanagementmeasurestoavoidoralleviate resourcecompetition. Theirresults show thatcurrent practices anddevelopmenttrendswouldleadtoanincreaseddemandfor food,water,energy,andland,andthattargetingmultipleresource usegoalsatoncecanleadtohighermanagementefficiencywith regardtosustainability.Whatnoneofthemhasdone,however,is to examine the effects that sectoral specific changes – both technologicalandbehavioral–couldhaveonsuchfutureresource demands.Thisisimportant,bothbecausesector-specificchanges mayrepresentthebestleveragepointsforpolicy,andbecauseit maybethattheopportunitiesfor resourceconservationinone sectormaydominatethoseinallothersectors.Itistheissuewe nowaddress.
3.Methods
3.1.Modelingframework
For our own quantitative approach we focused on water consumption(heresynonymouswithwaterdemand)forfoodand energy at the supply stage. We chose not to include water withdrawalsofthesetwosectors,asthismightleadtoamultiple accounting of the same water resources used and re-used for variouspurposesinbothsectors.Ratherthanonlyextrapolating current trends and consumption patterns as done in previous
studies,whichnecessarilyleadtoanincreaseinresourceusein absence of policy interventions that directly target water use efficiencyaswellastechnologicalimprovements,weexploredthe variability within those patterns. As this variability might potentiallyinfluencewaterdemandwithinandtrade-offsbetween the twosectors, we testedtheextent towhich preferences for certain food sources as well as electricity and transport fuel sourcescouldindirectlydriveoverallfuture regionalandglobal waterdemand.
Wedevelopedascenarioapproachforwhichweusepopulation projectionsuntilthemid-centuryandbuiltatwo-partaccounting and linear optimization model calculating water consumption associated withfood and energy demand.To beable todetect potentialdrivers,watersavingopportunitiesandpossibletrade- offs between the agricultureand energy sector withregard to futurewateruse,wetestedthreepotentialdietaryandtwoenergy demand trends in the form of changed global consumption patterns in 2050 compared to today’s foodand energy source preferences. Fig.1 displays anoverview of our methodological approach.
3.2.Underlyingscenariosanddata
The Shared Socio-economic Pathways (SSPs) constitute a frameworkforclimatechangeresearchthatdescribesplausible alternative developments in society and economy without integrating climate change or new climate policies. For our studytheyserveasreferencepointmainlyregardingpopulation growth as well as for assumptions on general socio-economic development.Wetooktheaveragepopulationprojectionsfrom the framework’s five global world regions, Asia (ASIA), Latin America (LAM), the Middle East and Africa (MAF), the OECD countries (OECD), andcountries from reforming economies of Eastern Europe and the former Soviet Union (REF). In all SSP projections overall growth of population numbers as well as GDP is projected in ten-year steps until 2100 (O’Neill et al., 2013; IIASA, 2013), we selected the year 2050 as projection pointforourownanalysisforwhichweestimatedatotalglobal populationrisetoroughlyninebillionpeople.Fig.2presentsan overview of the SSP world regions and their associated population projections.
Food: FAOSTAT Energy: The World Bank Water: Water Footprint Network
accounting and optimization
model
combined and relative water demand scenarios Food supply - 3 trends
Energy supply - 2 trends Input resource
data 2011
Frame: SSP demographic projections until 2050
trade-offs
(1) increase protein demand (2) replace foods (3) macro-nutrient shift
(1) increase renewables/
electric transport (2) increase first-generation
biofuels
Fig.1.Methodologicalapproachfortestingasetofpotentialdietaryandenergysourcedemandtrendswithregardtowaterconsumption,startingin2011.
Wechose2011asreferenceyearasthisyearmarksthemost recentconsistentpointintimefordatacollectiononglobalfood andenergysupply.FAO’sonlinedatabase(FAOSTAT2014)offers dataonannualfoodsupply(foodsoldonmarketsandinstores)on acountrylevelformajorfoodsandfoodgroups.Thisinformation reflectsactualfood consumption onlytosome extent, aspost- supplyfoodwasteratesandsharesvaryfromfoodgrouptofood groupandregiontoregion(Gustavsonsonetal.,2011)anddoesnot includesuppliesfromsubsistencefarming.Ofcourse,foodwaste occursalreadybetweenproductionstagesandfinalsupplyandthis alsovariesbetweenregions,asshowninthedatabaseaswell,but sharesdonotdistinguishwasteassociatedtotheediblepartofthe productandnon-ediblebutotherwiseusedparts.Foreachofthe fiveSSPregionswecalculatedtheaveragefoodsupply(weightand energycontent)basedonpopulationshareswithintheregionfor thefollowingmainfoodgroupsandtheirindividuallylistedfoods:
cereals, starchy roots, sweeteners, pulses, nuts, vegetable oils,
vegetables, fruit, meat, animal fats, eggs, dairy, and fish— 43 products in total. The World Bank energy database offers annualdataonelectricityandtransportfuelsupply,givingmain energysourcestechnologysharesonacountrylevel(TheWorld Bank,2015).Weaggregatedthesedataforeachworldregionand adaptedglobalassumptionsonthesharesofenergyplantcooling technologiesfromDaviesetal.(2013)fortheelectricitysectorof eachregion.
Forcalculatingthewaterconsumptionoftheglobalfoodsupply aswellasforfirst-generationbiofuelproductionwecollecteddata fromtheglobalWaterFootprint(WFP)Network.Itformsanoften- applied approach toassess the waterconsumption that occurs whenproducingacertaingood(Mekonnenetal.,2011;Mekonnen and Hoekstra, 2012; Gerbens-Leenes et al., 2008). The Water Footprintisdefinedasthevolumeoffreshwaterappropriatedto produce a product, taking into account the volumes of water consumedandpollutedinthedifferentstepsofthesupplychain Fig.2. SSPworldregionsandpopulationgrowthasprojectedforeachregion.ThemapdisplaysthefiveworldregionsAsia(ASIA),LatinAmerica(LAM),MAF(MiddleEastand Africa),OECDcountries(OECD)andthecountriesfromreformingeconomiesofEasternEuropeandtheformerSovietUnion(REF).Thegraphontherightpresentsthe projectedpopulationgrowthforeachworldregionuntil2050.
Table1
Modelconstraintsforregionalfoodsupplyandtrade.ThisTablelistsandexplainsthemodel’srestrictionsandboundarieswithregardtowaterconsumption(blueandgreen water),regionaldietpatternsandnutritionalassumptionsconcerningregionalfoodsupply.Singlefoodsincludedintheanalysisarewheatandwheatproducts,rice,barley, maizeandmaizeproducts,rye,oats,sorghum,othercereals(cereals);cassava,potatoesandpotatoproducts,sweetpotatoes,yams,otherroots(tubers);sugarand sweeteners;beans,peas,soybeans,otherpulses(pulses);nuts;soybeanoil,groundnutoil,sunflowerseedoil,rapeseedoil,cottonseedoil,palm(kernel)oil,coconutoil, sesameseedoil,oliveoil(plantoils);vegetables;fruit;beef,goat,pig,poultry,offal,othermeats(meat);animalfatsincl.butter;eggs;dairy;fish.
Objective Limitation
Noincreaseofcurrentwaterstressthroughfoodimports(Hoekstraand Mekonnen,2012)
Noincreaseinbluewatertrade
Increaseingreenwatertradelimitedtoamaximumof10%
(New)foodsaddedtothecurrentdietonlyconsumewaterresourcesstemmingfrom withintheregion
Nofundamentalchangesindietarypatternsassumed(compareLastetal.,2015) Mustkeepatleast50%ofaregion’sstaplefood(e.g.riceandwheatinASIA) Noincreaseinuncommonfoodswithinaregion,e.g.sorghuminOECD Mitigatepotentialhealthrisksfromsugaroverconsumption(FriedandRao,
2003;Shapiroetal.,2011)
Noincreaseinsugarandsweeteners Mitigatepotentialhealthrisksfromdairyconsumption:onlyabout30%ofthe
globalpopulationareabletodigestlactose(Lomeretal.,2007)
Noincreaseindairy Mitigatepotentialhealthrisksfromsoyoverconsumption(Gilanietal.,2012;
Cederrothetal.,2012)
Soyandsoyproductsarelimitedtoamaximumof100kcal/cap/d Limitbiodiversityloss(KohandWilcove,2008;Burgessetal.,2013) Noincreaseinpalm(kernel)oilconsumption
Noincreaseinseafoodconsumption Limitpotentialwaterdemandchangesformeat,assoybeanoilcakeiswidely
usedasanimalfodder
Limitpotentialincreaseofsoybeanoilto10%
Ensurevarietyinnutrientsupply(Footeetal.,2004) Keepallmainfoodswithineachregion’stypicaldiet Keepcurrentvegetableandfruitconsumptionstable Includequalityassumptionswhencomparingplantandanimalproteinsources
(Friedman,1996;Sarwar,1997)
Combinegrainsandlegumestoprovidesufficientproteinsource,includinglower qualityassumptionsofaboutathirdcomparedtoaverageanimalprotein
(directandindirectwaterconsumption).Itismostlyusedwhen assessingthevirtualwatertradethataccompaniesinternational producttrade.Thedatabaseliststheaverageblue,grayandgreen waterconsumptionforagriculturalproductsbyproductonasub- national level, calculated using the global CROPWAT model (Hoekstraetal.,2011).ForbioenergyproductionGerbens-Leenes etal.(2008)listblueandgreenwaterconsumption.Bluewateris definedasthefreshsurfaceandgroundwater.Graywateriswater that is requiredtodilute pollutants tosuchan extentthat the qualityofthewaterremainsaboveagreedwaterqualitystandards.
Greenwateristheprecipitationonlandthatdoesnotrunofforre- chargethegroundwaterbut isstoredinthesoilor temporarily stays on top of the soil or vegetation. Eventually, this part of precipitation evaporates or transpires through plants. For this analysiswechosetofocusprimarilyontothebluewaterdemand ofdifferentagriculturalproducts(RidouttandPfister,2012;Sulser etal.,2010).AstheWFPNetworkoffersdataonthenationaland sub-nationallevel,wecalculatedtheaveragewaterconsumption (bluewater)ofaproductforeachregionwhenproducedwithin thisregion.Thesenumbershoweverdonotincludetheamountof waterrequiredforassociatedfertilizerandpesticideproduction.
Forbiofuelsweselectedthethreemostprominentenergyplantsin eachregion.Regardingglobalfoodtrade,wecombinedtradedata from the FAO and ITC databases, providing trade shares and informationontradingpartners(FAOSTAT,2014;ITC,2014).We determinedthetwotothreemaintradingpartners(worldregions) foreachimportedproductandhencewereabletoestimatethe amountofwaterthatisimportedthroughacertainfoodproduct (virtualwatertrade).TheWFPNetworkdatabaseprovideswater useforunprocessedagriculturalproductsaswellasprocessedfood products.Forestimatingtheamountofwaterthatgetsattributed tothefinalfoodproduct,wechoseaveragesforfinaluncooked foodsthatalignwiththefoodsupplydatafromFAO.Regardingthe water demand of specific electricity and fuel technologies, we appliedasetofdatacollectedbyDamerauetal.(2015).
3.3.Modelandconstraints
Inorder tointegrateallcollectedresourcedemanddata,we developed a two-part accounting model that allowed us to calculate the water demand (blue, green and gray water separately) for each selected foodper kcal within each region, including the regional water amount from imported products.
Listing the associated food group and specific macro-nutrient content of each food created the basis for a rough qualitative comparisonbetweensinglefoodsandfoodgroups.Inanextstep, weaddeda modelfunctionintheformofalinearoptimization moduleusingtheprogramminglanguageR(Venablesetal.,2015).
This made it possible to limit the amount of energy, macro- nutrients,foodgroups,andsinglefoodsaslistedinthecaptionto Table1,aswellasgreenandgraywateruse,whenoptimizing,i.e.
minimizingthewaterdemand(bluewater)ofagivenorassumed dailynutritionalintake withina region. Wefollowed thesame methodologyfortheenergysector,whereinsteadoffoodsupplyin kcalwelistedthewaterdemandperGJforelectricityandtransport fuelsupply,includingvirtualwaterimportsfromimportedfossil fuelsforthelatter.
Tobeabletoruntheoptimizationmodelforfoodsupplyand potential future dietarypatternswithoutcompromising variety and health or excluding staple foods typically consumed in a certainworldregion,aswell aslimitingvirtualwatertrade,we compiled a listof assumptions and restrictionsas presentedin Table1.
Forourenergymodelwealsodefinedasetofconstraints.After calculatingthespecificelectricitytechnologysharesforcoal,gas, nuclear,oil,combinedcycle,biomass,concentratingsolarpower
(CSP), photovoltaic, wind, geothermal, and hydropower—all technologies that are representedin the World Bank database (TheWorldBank,2015),andifapplicabletheirassociatedcooling technologies for each of the five world regions, we made the baselineassumptionthatthespecificenergytechnologysharesdo notchangeovertimewithineachregion.Thisfirststeprepresents asimpleextrapolationfromcurrentenergysupplyconditionsthat laterallowsacomparisontopossiblychangingtechnologymixes and theirassociated waterdemand inthe energysectorin the future.Suchshiftsintechnologysharesarereflectedinpossible changing demands for certain energy technologiesthat do not directly target water saving goals. We accounted for water consumptionfromhydropowerseparatelyasthiswaterusestems mainlyfromevaporationlossesathydropowerreservoirs,which areoftenusedformultiplepurposesandthusmakeassumptions onattributablewaterlossesduetopowerproductionproblematic.
Electricitygenerationfrombiomassishereassumedtobeprovided bywastematteranddoesnotrequiretheadditionalplanting,and thereforeirrigationofenergycrops.
Regardingtransportfuels,weincludedvirtualwaterimports fromfossilfuelsextractedandexportedfromtheMiddleEastand NorthAfrica,aregioncontributingabout40%toglobaloilexports today(BP,2013).Forbiofuels,weusedassumptionsonconven- tionalfirstgenerationbiofuelssuchasbioethanolfromsugarcane or biodieselfromrapeseed oil. FromtheWFPNetworkdataon bioenergywedeterminedthe(partiallyweighted)averagewater consumption for three main biofuelcrops planted withineach worldregion.Tothesenumbersweaddedthewaterdemandfor processingandconvertingthesecropsintoliquidtransportfuels (VanVlietetal.,2009).
3.4.Alternativescenariosincorporatingshiftsinfoodandenergy consumptionpatterns
Globaldevelopmentgoalsincludefoodandenergysecurityfor alargenumberofpeopleforwhichbothfoodandenergydemand (absoluteand percapita) arelikelytoincrease overthecoming decades.FAO’sfoodsecuritydefinitionstates thatfoodsecurity existswhenallpeopleatalltimeshaveaccesstosufficient,safe, nutritiousfoodtomaintainahealthyandactivelife(FAO,1996).
From there, one can make very different assumptions about potential changes in future consumer preferences and their motivation. Weassumedanoveralldesiredtrendtowardsmore balanceddiets(sufficientandproportionallyadequatesupplyofall essentialmacro-andmicro-nutrients)thattargethealth,longevity, and optimal physical and cognitive performance. We did not assume extreme changes in global dietary patterns,but rather examine theeffectofhow moremoderate shiftstowards more nutritiousandsafefoodscanhaveonthewaterdemandforfuture foodsupply.Forthispurposewespecifiedthreeconcurrentdietary shifts.
Thefirstisanincreaseinproteinsupplyinallregionsexceptthe OECD regiontolevelscomparable tothose inOECD, which we assumed to be nutritionally sufficient for supporting basic metabolicprocessesandphysicalperformance.InOECDcountries wecalculatedanaveragesupply(notconsumption)ofproteinof approximately110g/cap/dwithashareofabout40%plantand60%
animalprotein.Forclosingthis‘proteingap’globallywecompared potentialanimalandplantproteinsourcesforeachselectedregion, foodsthatarealreadyavailableandconsumedwithineveryregion, usinglinearoptimizationtoidentifypossibleproteinsourcesthat show lowest regional water demand. Given the lower overall nutritionalqualityofmostplantproteinsincomparisontoanimal protein,andhencethenecessitytocombinedifferentplantfoods fora sufficientamino acidsupply,wetookaveragedigestibility datafromvariousstudies,assuminga 50/50proteinsharefrom
grains and legumes and an average digestibility factor of 1.5comparedtoanimalproteinsources;i.e.oneneedstoconsume 50%moreplantproteintoreachsimilarbioavailabilityasaverage animalprotein(Friedman,1996).
Inthesecondshift,withoutchangingthemacro-nutrientshares typical for today’s diets on a global level roughly 60%
carbohydrates,10% protein,and 30% fat (FAOSTAT, 2014) we lookedforpossibilitiestoswaptosomeextentcertainfoodswith eachother.Stayingwithinthemainfoodgroupsandoverallmacro- nutrientsharesofanaverageregionaldiet,anexampleforsuchan exchangewouldbethereplacementofoneplantoilinthedietwith another,potentiallymorenutritiousonewhencompareddirectly (USDA,2014;Siri-Tarinoetal.,2010;Deoletal.,2015).
Inthethirdshift,weexaminedthepotentialeffectsadecrease inabsoluteandrelativetotalcarbohydratesharefromroughly60%
todayto40%andhenceanincreaseinthefatshareofadiet.Sucha trendwouldbedrivenbycurrentempiricalandclinicalevidence on the potential negative health effects of long-term high- carbohydratediets (Sondike et al., 2003; Bazzano et al., 2014;
Westmanetal.,2007).Thisshiftcanbeconsideredasaprofound changeof the averagediet of a large number of people,while carbohydrateswould still represent thehighest macro-nutrient sharewithinsuchadiet.Inthisstepwealsoincludedincreased proteinlevelsinASIA,LAM,MAFandREFascalculatedfortrend oneandkeptoverallenergysupplystableineachregion,asthe averageenergeticsupplyofeachworldregion’sdietappearstobe sufficientifnotexcessiveinsomeregions,thoughcertainmacro-
andmicro-nutrientneedsmightnotbemetbymodern(Western) diets(Gosbyetal.,2014;Hunt,2003).Inallofthesethreepotential basictrends,watersavingsarenotassumedtobetheprimarygoal, but can be supportedby smart choices regarding the resource intensityofdifferentfoods.
Fortheenergysector,weenvisionedtwoconcurrentdevelop- ments reflecting potential consumer preference changes until 2050.Thefirstisanincreasedawarenessofclimatechangeand engagementtomeetclimatemitigationgoals,leadingtoahigher demand for renewable energy sources such as solar and wind power. The secondis growing health concerns associated with noiseandairpollutionfromtrafficrelyingmainlyonfossilfuels (Andersonetal.,2012;Curranetal.,2013),leadingtohighershares of electric transport and/or biofuels, such as bioethanol and biodiesel.Overthelastdecadeanumberofcountrieshavedefined variousgoalsforfuturebiofuelsharesintheirtransportfuelmix, oftenrangingfrom10to20%(Lane,2014).Producingtheirown biofuels would increase those countries’ energy independence, thoughacompetitionofbioenergywithfoodproductioncouldbe oneofthepotentialdownsides.Besides,first-generationbiofuels canhavelargenegativeecologicalimpacts,notonlywithregardto water(Creutzigetal.,2014).TheEuropeanUnionthereforerevised theirbiofuelstargetsuntil2020,limitingfirst-generationbiofuels toashareof7%(TheEconomist,2015).Weadoptedthisgoalforour globalestimatesand testedbothpossibletrends,estimatingthe effect they would haveonregional and globalwater resources without directly targeting future water availability. Additional
Fig.3.Totalcurrentandpotentialfuturewaterdemandforfood,electricity,andliquidfuelsupplybyregion.ExtrapolatingthewaterdemandofbothsectorsinASIA,LAM, MAF,OECDandREFtomeetthegoaloftoday’sOECDconsumptionpatternsfortheentirefuturepopulationresultsinasubstantialincreaseinwaterdemand.Dietaryshifts andtechnologicalchangescanleadtoimprovementsinwaterefficiencyforbothfoodandenergysupply.However,anincreaseinbiofuelsupplytomeet7%ofthefuture population’stransportfuelneedswouldleadtoadrasticriseinwaterdemand,potentiallyexceedingwaterrequirementsforfoodsupply.
factorsthatmightinfluenceorcounteractthetrendswedetected willbeevaluatedinthediscussionsectionofthisstudy.
4.Results
Fig.3displaysourfirstsetofresultsregardingcombinedand relative water demands on a global level, comparing water consumption for food and energy supply by region for 2011, 2050inabaselinescenarioaswellas2050inanimprovedscenario withandwithoutamajorexpansionoffirst-generationbiofuels.As our baseline scenario shows, extrapolating current food and energy consumption to a global population in 2050 would inevitablyleadtoalargeincreaseinwaterdemand,muchmore sointheagriculturalsectorthanintheenergysector.Anincrease in food supply in the form of a higher global average protein demandcomparable toOECDlevelsin 2050wouldresultsin a highercaloriedemandof40–60%dependingontheproteinsource chosen.Ifatthesametimeenergydemandweretoincreasetoper capitalevels wecurrently seein OECD countries, assuming no changesintheenergytechnologyshares,wewouldseeatotalrise inenergydemandby180%.Bothrisingresourcedemandswould lead to an increase of overall freshwaterconsumption by 50%
compared to current global water requirements, 15% of which wouldberequiredintheenergysector.Inourimprovedscenario, where we consider three shifts regarding food consumption patterns, and one shift in the energy sector towards more renewables (and/or dry-cooled thermal power production in general)andelectrictransportuntil2050,weseeaslightdecrease forthecombinedwaterdemandofbothsectorsby4%despitea globalpopulationgrowthtoninebillionpeople.Thewatersavings for food supply outweigh growing water requirements for electricityandtransportfuels.Comparedtothebaselinescenario,
thisprojectionshowsanintotal35%lowerwaterconsumptionin 2050.Onecaveat,however,ispresentedinthelastscenariowithan expansion of first-generation biofuels to globally 7% of total transportfuels.Thiswouldresultinawaterdemandforenergy supplyhigherthanforthecurrentfoodsupply,totalglobalwater demandinthisscenariowouldmorethandouble.
In Fig. 4 we present a second set of resultsillustrating the specificeffectssingletrendswouldhaveonthewaterdemandfor foodsupplyineachworldregion.Dailypercapitawaterintensity offoodsupplyiscurrentlyhighestintheREFregion,andlowestin MAF.Thispresentwaterconsumptionisputintorelationwith(1)a potential driver ofwaterconsumption in theformofincreased proteindemandinfouroutoffiveworldregions,andtwowater- saving trends: (2) more nutritious foodsources could tosome extentreplacecurrentfooditems,and(3)acombinationoffood replacements and a macro-nutrient shift from 60 to 40%
carbohydratesintheaveragedietby2050.Overallresultsshow thatinallfiveregionsaconsiderablereductioninwaterdemand couldbeachievedindirectlythroughdietarychanges.
4.1.Increasingproteinsupply
Reaching the level of protein supply as observed in OECD countriestoday,includingahighanimalproteinshare,wouldlead toanincreaseindietaryproteinandcaloriesassociatedwiththese proteinsourcesinallotherworldregions,ASIA,LAM,MAF,andREF.
In ASIAand LAManimal proteinsourceswouldlead toslightly strongerwaterdemandincreasethanplantproteinsources,while in REF plant protein requires slightly more water. The biggest difference betweenthewaterrequirementsfordifferentprotein sources was foundin MAF, where animal protein(goat) would requireconsiderablylesswaterthanamaize/peamix.Inallregions
Fig.4. Currentregionalwaterdemandforfoodsupplyandeffectsofpotentialdietaryshifts.Waterdemandinliterspercapitaanddayincreaseswhenanimal(orange)or plant(yellow)proteinsourcesareaddedtotheaveragediet.Inthreeregions,ASIA,OECD,andREF,substitutinghalfoftheamountofcertainfoodswithlesswater-intensive onesreducedoverallwaterdemandsignificantly.Anevengreatereffectcanbereachedinallregionsthroughashiftfromcarbohydratesourcestowardsmorefatsources.(For interpretationofthereferencestocolorinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)
anincreaseofproteinsupplythroughplantproteinsourceswould alsoleadtoaconsiderablyhigherincreaseinenergysupplythan thatforanimalprotein,exceedingcurrentOECDlevelsofroughly 3500kcal/cap/d.
4.2.Replacingfoods
InLAMandMAFwedidnotfindsinglesignificantfoodswhere anexchangewouldleadtosubstantiallylowerwaterrequirements.
InASIAhowever,ahypotheticalreplacementofhalfitswheatand rice consumption with more nutritious tubers such as sweet potatoesandyamswouldleadtoan18%reductionofoverallwater intensityoftheaverageAsiandiet.InOECDandREFwefindsimilar savingpotentials.Giventherelativelyhighdairyconsumption,a 50% replacement of dairy products with either eggs in OECD countriesorgoat/sheepmeatinREFcountrieswouldlowerwater demandby6–10%;also,replacing50% oftheseregions’current soybeanand safflower oil supply withrapeseed or coconut oil wouldleadtoareducedwaterdemandofanother5%.Suchshifts could potentially offset water demand increases from rising population demands as discussed above, while increasing the micro-nutrientcontentoftheaveragediet.
4.3.Shiftingmacro-nutrientcomposition
ThistrendincludesaslightincreaseinproteintoOECDlevels (ascalculatedfor trendone)as wellaspotential watersavings describedfortrendtwo.Weassumedadditionalproteinsourcesto besupplied byanimal sources,whichshowlowercarbohydrate loads.Inallfiveregionsatrendawayfromveryhighcarbohydrate suppliestowardsdietshigherinfat,infourregionsanimalprotein,
andalsomicro-nutrientcontent,wouldleadtoapercapitawater demandoftheaverageglobaldietlowerthanseentoday(from 560to400l/cap/d).Dependingontheregion,wedetectanumber ofdriversforthistrendincludinga shiftawayfromgrains(and sugar)towardstubers(thoughtheotherwayaroundinLAM),more plantoilssuchascoconutoil,lessdairy butmoremeatsources suchasgoat,sheepandinMAFalsopoultry,andmoreeggsand animalfatsinOECDcountries. Wefindadecreaseinpercapita waterconsumptionbetween12%inLAMand45%inREF.Addingup thesepotentialsavingoverthewholeglobalpopulation,weseea decreaseintotalwaterdemandforglobalfoodsupplyin2050by 10%despitethedemographicgrowth.
Fig.5 illustratesa third setof resultsbycomparingcurrent waterdemand forelectricity and transportfuel supplyto(1) a globalincreasetopercapitaenergyintensityascurrentlyobserved inOECDcountries,(2)ascenarioinwhich50%ofthethisenergy supplygoalcouldbemetbyrenewablesand/ordry-cooledenergy technologies,includinga50%shareofelectrictransport,and(3)an increaseoffirst-generationbiofuelsharetoglobally7%.
4.4.Extrapolatingcurrenttechnologyshares
Meeting the potential future electricity and transport fuel demandinASIA,LAM,MAF,andREFwouldresultinanincreased waterdemandbyafactorof2.5inREFtofactorof8inASIA.When compared to today’s consumption levels, a considerable share wouldberequiredforincreasedfossilfuelproduction,alsoleading toasignificantriseinassociatedvirtualwatertradewithMAF.Not includedintheseestimatesiswatercontributingtohydropower production.Withouthydropowerwewouldseeaglobalincreasein waterconsumptionforelectricityandtransportfuelproductionby
Fig.5. Currentregionalwaterdemandforelectricityandliquidfuelsandeffectsofchangingdemandpatterns.Increasingthepercapitaenergydemandinallregionsto currentOECDlevelsalsoleadstoanincreaseinwaterdemandoftheenergysector.Watersavingscanbeachievedthroughshiftstowardswindorphotovoltaicbutalsodry- cooledthermalpowercapacities.Asignificantincreaseinfirst-generationbiofuels,however,wouldincreasetheoverallwaterdemandforenergydramatically.Water demandforhydropowerisnotincludedinthisgraph.
afactoroffour,from74km3/yrto298km3/yr.Thisisasignificant increasethatmightindeedleadtoincreasedcompetitionforwater resource in water-stressed areas. Compared to the potential increaseinwateruseforfoodsupply,i.e.increasedproteinsupply, from 1449 to 1942km3/yr this trend in the energy sector still appears minor, though not trivial. If hydropower needs were includedinthisextrapolation,thiswouldaddanother1580km3/yr ofwaterconsumption, analmost12 timeshigheramountthan today’s estimates. In this scenario highest increases in water consumptionassociatedwithlargehydropowerappearintheMAF region,followedbyASIAandREF.
4.5.Newenergytechnologiesintheelectricitymix
New power plant capacities are required either to replace outdated plants or increase overall electricity supply. Some technologiesrequirenegligibleamountsofwatertooperatesuch asphotovoltaicandwindturbines.Inthecaseofthermalpower plantsdryorseawatercoolingtechnologiescanbeemployedto reduce the water demand by about 90%. Hence, if 50% of the electricityplantsin2050weretoeitherusewindorphotovoltaic, ordry/seawater-cooledtechnologiessuchas CSP,geothermalor biomass/wasteplants,thewaterdemandoftheelectricitysector couldbealmostcutinhalf.Thesameholdstrueforthetransport sector,iffossilfuelsweretobereplacedwithelectricityfromthose water-savingtechnologies.
4.6.Increaseinfirst-generationbiofuels
Ifin2050globalaveragepercapitafueldemandwouldreach OECD levels and 7% of this demand would be met by first- generationbiofuels,thatareproducedwithineachworldregion, using the currently most common energy crops, total water consumptionforenergywouldincreasefrom74km3/yrtodayto possibly 2012km3/yr, 97% of which for growing biomass. This amountofwaterwouldequaltheamountofwaterrequiredfor increasedfoodsupplywhennotassumingpotentialdietaryshifts.
5.Discussion
Incontrasttopreviousstudies,ourworkisabletoshowthatan increase in waterdemand for food production in future is not inevitable,whileariseinwaterconsumptionintheenergysector appearsineveryscenarioweexamined.Becausetheuseofwater for food is currently in most cases more than one order of magnitudelargerthanforenergydependingonworldregion,there isanoverallpotentialtosavewateracrossthetwosectors.Atthe sametime,increasedrelianceonbiofuelscouldeasilychangethis story,makingenergythelargerwaterconsumer,overshadowing anypotentialgainsinthefoodsector.
Agloballyconsiderableintensificationinbluewaterdemandof 50%asestimatedbyusinthefirststepofthisstudyiscomparable tofindingsofotherauthors(OECD,2012),andpotentialmitigation measuresarediscussedin manyWE(L)Fstudiesas citedin the Background section. Indeed, if we were to simply extrapolate currentpercapitaOECDconsumptionpatternstothegloballevel in2050,regionalandlocalwatercompetitionislikelytoincrease, andmightevenleadtopotentialresourceconflicts(Bogardietal., 2012).Still,onaregionalandglobalscalewaterdemandforenergy remainsminorwhencomparedtoresourcesusedforfoodsupply.
Interconnectionsbetweenthetwosectors withregardtowater thereforesofarappear asa potential environmentaland social issuepredominantlyonasub-regionalandlocalscale(Apipalakul etal.,2015).Additionalsectoralwaterdemandscanstemfromfor example hydrogen generation, an element that is required for fertilizerproductionbut alsooiland naturalgasrefining.These
waterneedsarenotincludedinourglobalanalysis,andcurrently amounttoavolumeaboutfourmagnitudessmallerincomparison tooverallbluewaterdemandforfoodandenergysupply.However, in areas that already experience water stress, such auxiliary requirements have the potential to further increase resource competition.
Inthepartofourstudylookingatwaterrequirementsforfood supply,weinvestigatedhowpotentialchangesinfoodconsump- tion,i.e.changingdietarypatterns,andenergypreferencescould affectregionalandglobalfreshwaterconsumption.Wechoseto lookatthreepotential, overlappingtrendsinfooddemandover almostfourdecades.Dietarytrendswithinmodernsocietiescan onlybeobservedoverarelativelylongtimespanofseveraldecades beforetheybecomestatisticallyvisible(USDAERS,2008).Ourgoal wastoillustratetheeffectssuchtrendscouldhaveonoverallwater demand;thesetrendsdonotconstitutedietaryrecommendations.
Wefindthatpossibleandplausiblechangesinfoodpreferences, partly in combination witha shiftto less water-intensivefood sources, both potentiallydriven bypersonal health and perfor- mancegoals,couldindeedresultinaloweroverallwaterdemand forfoodsupplythantodaydespiterisingpopulationnumbers.We focusedonbluewaterdemandforthisanalysis,aswemadethe assumption that potential savings in green water use would practicallynotincrease actuallyavailablewater supplyfor both food or energy production (Wichelns, 2004). However, we restrictedthepotentialincreaseofgreenwateruseandtradeto a maximum of 10% inour scenarios toavoida significant, and potentiallyunfeasibleincreaseofthosewaterresources.
The trends investigated in this study do not fundamentally affectregionalandlocalcuisinesandtraditions,asmostlybroad averages for regional food supply were used that do not compromise foodvariety and traditional choice of meal ingre- dients.However,thedemandformorenutrientdensedietscould alsoleadtootherplausiblechangesinfoodpreferencessuchasan increaseinvegetableandfruitconsumption.Suchatrendwould counteractpotentialwatersavings,asbothfoodgroupsshowhigh freshwaterconsumptionrates.Itisalsoworthtolookateachworld region separately as water footprints can vary significantly between regions. Regardingfood sources that provide protein, adequateplantproteindoesnotnecessarilyrequirelesswaterthan comparableanimalproteinsources.
Anotherimportantpointtomakeisthatfoodsupplydoesnot equalagriculturalproduction(compareresultsfromChartresand Sood (2013)). Plant and animal products often satisfy multiple purposesbesidesdeliveringfood,suchasprovidingseeds,fodder, leather,oringredientsforpersonalhygieneproducts.Thereforethe losses and waste that occur between the production of the agriculturalproductandthefinalfoodproductinretailaredifficult toallocate.Thisis thereasonwechosetofocusonfoodsupply ratherthenagriculturalproductiondataconcerningwaterrequire- mentsforfoodproductionwithinthebroadercontextofthewater- energy-foodnexus.Thelargestshareoffoodwaste(onaverage 30%)occursafterthesupplystageatretailpointsandinprivate households(Gustavsonsonetal.,2011).Thedataweappliedfor foodsupplythereforedonotreflectaveragefoodconsumption,and do not include private food production on a household level.
Above,inhigh-incomecountriesfoodwastesharesafterretailare often higher compared to those in low-income countries. A reduction in food wastecould thereforeadditionally lowerthe intensityofnaturalresourceusewithoutassuminganydemandor technologicalchangesintheglobalfoodsystem.
Incontrast,intheenergysectorwaterdemandwilllikelygrow, even when considering an increasing share of renewable technologies. Whenassuming high per capita energy intensity inthefutureonaglobalaverage,energysupplycapacitieshaveto be expended drastically to satisfy the growing demand. This
