Original
articles
Water
footprint
benchmarks
for
crop
production:
A
fi
rst
global
assessment
Mes
fi
n
M.
Mekonnen
*
,
Arjen
Y.
Hoekstra
1TwenteWaterCentre,UniversityofTwente,P.O.Box217,7500AEEnschede,TheNetherlands
ARTICLE INFO
Articlehistory: Received7January2014
Receivedinrevisedform28May2014 Accepted10June2014
ABSTRACT
Inthecomingfewdecades,globalfreshwaterdemandwillincreasetomeetthegrowingdemandforfood,
fibreandbiofuelcrops.Raisingwaterproductivityinagriculture,thatisreducingthewaterfootprint (WF)perunitofproduction,willcontributetoreducingthepressureonthelimitedglobalfreshwater resources.ThisstudyestablishesasetofglobalWFbenchmarkvaluesforalargenumberofcropsgrown intheworld.Thestudydistinguishesbetweenbenchmarksforthegreen–blueWF(thesumofrain-and irrigationwaterconsumption)andthegreyWF(volumeofpollutedwater).Thereferenceperiodis1996– 2005.Weanalysedthespatialdistributionofthegreen–blueandgreyWFsofdifferentcropsascalculated ataspatialresolutionof5by50withadynamicwaterbalanceandcropyieldmodel.Percrop,weranked theWFvaluesforallrelevantgridcellsfromsmallesttolargestandplottedthesevaluesagainstthe cumulativepercentageofthecorrespondingproduction.Thestudyshowsthatifwewouldreducethe green–blueWFofcropproductioneverywhereintheworldtothelevelofthebest25thpercentileof current globalproduction,global watersavingin cropproductionwouldbe 39%comparedtothe referencewaterconsumption.WithareductiontotheWFlevelsofthebest10thpercentileofcurrent globalproduction,thewatersavingwouldbe52%.Inthecasethatnitrogen-relatedgreyWFsincrop productionarereduced,worldwide,tothelevelofthebest25thpercentileofcurrentglobalproduction, waterpollutionisreducedby54%.IfgreyWFspertonofcroparefurtherreducedtothelevelofthebest 10thpercentileofcurrentproduction,waterpollutionisreducedby79%.Thebenchmarkvaluesprovide valuableinformationforformulatingWFreductiontargetsincropproduction.Furtherstudieswillbe requiredtotestthesensitivityofthebenchmarkvaluestotheunderlyingmodelassumptions,tosee whetherregionalizationofbenchmarksisnecessaryandhowcertainWFbenchmarklevelsrelateto specifictechnologyandagriculturalpractices.
ã2014TheAuthors.PublishedbyElsevierLtd.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/3.0/).
1.Introduction
Agricultureisthelargestfreshwateruser,accountingfor99%of theglobalconsumptive(greenplusblue)waterfootprint(Hoekstra and Mekonnen, 2012). Growing populations, coupled with changingpreferencesindietsandrisingdemandforbiofuels,will put increasing pressure on the globe’s freshwater resources (Falkenmarket al.,2009;Gleick,2003; Rosegrantet al., 2009). Theconsumptivewateruse(frombothprecipitationand irriga-tion)forproducingfoodandfoddercropsisexpectedtoincreaseat 0.7%peryearfromitsestimatedlevelof6400billionm3/yearin
2000to9060billionm3/yearinordertoadequatelyfeedtheglobal
populationof 9.2 billion by2050 (Rosegrant et al., 2009). The growingfreshwaterscarcityisalreadyevidentinmanypartsofthe world(Gleick,1993;Hoekstraetal.,2012;OkiandKanae,2006; Postel,2000;Vörösmartyetal.,2010;Wadaet al.,2011).
Raisingwaterproductivityinagriculture(“morecropperdrop”) cancontributetoreducingthepressureontheglobalfreshwater resources(Passioura,2006;Rockström,2003).Thewaterfootprint (WF)offersaquantifiableindicatortomeasurethevolumeofwater consumption per unit of crop, as wellas the volumeof water pollution(HoekstraandChapagain,2008;Hoekstraetal.,2011). ThegreenWFmeasuresthevolumeofrainwaterconsumedduring thegrowingperiodofthecrop;theblueWFmeasuresthevolume ofsurfaceandgroundwaterconsumed.ThegreyWFmeasuresthe volumeoffreshwaterthatisrequiredtoassimilatethenutrients and pesticides leaching and running off from crop fields and reaching groundwater or surface water, based on natural *Correspondingauthor.Tel.:+31534892080.
E-mailaddresses:m.m.mekonnen@utwente.nl(M.M. Mekonnen), a.y.hoekstra@utwente.nl(A.Y. Hoekstra).
1
Tel.:+31534893880.
http://dx.doi.org/10.1016/j.ecolind.2014.06.013
1470-160X/ã2014TheAuthors.PublishedbyElsevierLtd.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/3.0/). ContentslistsavailableatScienceDirect
Ecological
Indicators
background concentrations and existing ambientwater quality standards(Hoekstraetal.,2011).
WFbenchmarksforcropproductioncanbeaninstrumentto compareactualWFsin certainregions oreven atfieldlevel to certain referencelevels and can form a basis to formulateWF reduction targets, aimed to decrease water consumption and pollutionperunitofcrop(Hoekstra,2013a,b).WFsofcropsvary enormously acrossregions and withinregions (Braumanet al., 2013;Faderet al.,2011;Finger,2013;Hoekstraand Chapagain, 2007; Mekonnen and Hoekstra, 2011; Siebert and Döll, 2010). Therearenopreviousstudiesthataimedtodevelopbenchmarks fortheWFofcrops,butanumberofstudiesexistonbenchmarking water productivities. The water productivity (ton/m3) in crop
productionisinfacttheinverseofthegreen–blueWF(m3/ton)of
cropproduction.Waterproductivitystudiescanbegroupedinto four classes: field studies, modelling studies, studies based on remote sensing, and studies employing a combination of field measurementandmodellingorsatellitedata.Infieldstudies,the relationship between seasonal water use and crop yield is determinedfromfieldmeasurements(Oweisetal.,2000;Rahman etal.,1995;Sadrasetal.,2007;Sharmaetal.,1990,2001;Zhang et al., 1999, 1998). Water productivity studies based on field measurements are boundto experiments on a relativelysmall number of fields, so that results are always limited to local conditions such as climate, soil characteristics and water managementpracticesandcannoteasilybescaledupforlarger areas.Inmodellingstudies,soilwaterbalanceandcropgrowth modelsareusedtoestimatethecomponentsoftheseasonalcrop waterbalance(AmirandSinclair,1991;Assengetal.,1998,2001). The limitation of model studies is that they generally do not accountfor allconstraining factorsand mayexcludeimportant factorssuchaspests,diseasesandweedsandtheiruseislimitedby dataavailabilityandquality(Grassinietal.,2009).Remotesensing studiesusesatellitedatatoestimatethespatialvariationofwater productivity (Biradar et al., 2008; Cai et al., 2009; Zwart and Bastiaanssen, 2007; Zwart et al., 2010a,b). The use of remote sensingallowsestimatingthewaterproductivityoverlargeareas. A numberof studies combinedmeasured data withsimulation models(Grassini et al.,2009; Robertson and Kirkegaard, 2005; Sadras et al., 2003) and others combined measured data with remotesensing data(Caiand Sharma, 2010).Whilecropwater productivity is receiving an increasing amount of attention, minimizing water pollution (the grey WF) per unit of crop productionreceivesmuchlessattention.Itisclear,though,thatthe grey WF per unit of crop varies greatly from place to place depending on agricultural practices (Chapagain et al., 2006; MekonnenandHoekstra,2010,2011).
Toourknowledge,therehasbeennopreviousstudyproviding globalbenchmarkvaluesforgreen–blueandgreyWFsofcrops.The studiescitedabovearelimitedtoeitherafewcropsorspecific locations.Theobjectiveofthecurrentstudyhasbeentodevelop globalWFbenchmarkvaluesfor 124cropsbasedonthespatial variability of crop WFs as found in our earlier global WF assessmentofcropproduction(MekonnenandHoekstra,2011). 2.Methodanddata
Thestudydistinguishesbetweenbenchmarks forthegreen– blue WFand thegrey WF of crops. The approach hasbeen to analysethespatialdistributionofthegreen–blueandgreyWFsof differentcropsascalculatedataspatialresolutionof5by50witha dynamicwaterbalanceandcropyieldmodel.Detailsonthemodel usedhavebeenreportedinMekonnenandHoekstra(2010,2011). Basically, the model computes a daily soil water balance and calculatescropwaterrequirements,actualcropwateruse(both greenandblue)andactualyields.Green–blueWFsarecalculated
bydividingtheevapotranspirationofgreenandbluewaterover the growing period bythecropyield. Grey WFsare calculated basedonnitrogenapplicationrates,leaching-runofffractionsand waterqualitystandardsfornitrate.Wedidnotconsiderthegrey WF fromother nutrients (likephosphorous) or pesticides. The modelwasappliedataglobalscalefortheperiod1996–005.In total,124cropswerestudied.
WefirstanalysedtheWFofwheatintermsofm3/tonatthree
differentspatialresolution levels–country, provincialand grid level – in order to identify the proper spatial resolution for developing WFbenchmarks forcropproduction.Afterchoosing thegridlevelasthebestoptionforfurtheranalysis,thevariability inWFsofcropsoverallcropgrowinggridcellsintheworldwas usedfordevelopingthebenchmarks.Percrop,werankedtheWF valuesforallrelevantgridcellsfromsmallesttolargestandplotted thesevaluesagainstthecumulativepercentageofthe correspond-ingproduction.Fromthegraph,wecouldthusreadtheWFvalues atdifferentproductionpercentiles.
ForananalysisofdifferencesinWFsbetweendevelopingversus industrialisedcountries,weusedthecountryclassificationbased onincomefromtheWorldBank(2012);inwhichcountriesare dividedaccordingtothe2007percapitagrossnationalincome. The groupsare: lowincome (USD 935),lowermiddleincome (USD 936–3705), upper middle income (USD 3706–11455) and highincome(USD11,456).
In order to analyse differences in WFs between different climaticregions,weusedtheKöppen–Geigerclimateclassification (Kotteketal.,2006)togrouptheworldintofourmajorclimate classes:tropics(aridandequatorial),temperate,boreal(snow)and tundra(polar).Sincelittleornocropcultivationexistsintheboreal andtundraregionsoftheworld,wehavefocusedonthetropics andtemperateregions.
3.Results
3.1.Thedistributionofthegreen–bluewaterfootprintofwheatat threespatialresolutions.
The distribution of the green–blue water footprint (WF) of wheatwasanalysedatthreedifferentspatialscalesbyconsidering theaveragegreen–blueWFinm3/tonandproductiondatainton/ year at country, provincial and grid level. Figs.1–3 have been obtainedbyplottingthegreen–blueWF,sortedfromsmallestto largest,againstthecumulativepercentageofproduction.Although thefigures for thethree spatialscalesof analysisshowsimilar patterns,thepointsinthecountry-scaleanalysis(Fig.1)do not formasmoothcurvelikethepointsintheprovincial-scale(Fig.2) andgrid-scaleanalysis(Fig.3),causedbythelimitednumberof pointsinthecountry-scaleanalysis.Thegreen–blueWFvaluesat therespectiveproductionpercentilesdecreasewhenmovingfrom thecountrytothegridlevel.Inaddition,weseethattheWFatthe 50thpercentileofproductionisnotnecessarilyequaltotheglobal averageWF,whichisacharacteristicofanyskeweddistribution.As wecanseefromthegrid-basedanalysis,thevariabilityoftheWFof thebesthalfoftheglobalwheatproductionissmallerthanthe variabilityoftheWFoftheworsthalf,sothattheglobalaverage WF(1620m3/ton)turnsouttobelargerthantheWFatthe50th percentileofproduction(1391m3/ton).Thelattervaluemeansthat
50% of globalwheat production occurs at a green–blue WF of 1391m3/tonorless.
Therearetworeasonsthatfavourthegridovertheprovincialor countrylevelanalysis.First,particularlythecountrylevelanalysis isweakasitprovidesaverydispersedcurveandtheanalysiswill getevenweakerforcropswhicharegrowninonlyafewcountries. Second,therecanbesignificantWFdifferenceswithinprovinces andcountries,whicharehiddenintheanalysisatthoselevels.The M.M.Mekonnen,A.Y.Hoekstra/EcologicalIndicators46(2014)214–223 215
averages at provincial and even more so at country level are generally biased towards the worse footprints (because of the skeweddistributions),sothattheWFsatthevariousproduction percentilesfoundwiththegrid-basedanalysisareclosertoreality than in case of the analyses at the lower resolution levels. Therefore, for the remainder of this study, we analyse the distributionofcropWFsatthegridlevel.
3.2.Thegreen–bluewaterfootprintofdifferentcropsatdifferent productionpercentiles
Thedistributionofthegreen–blueWFfortenselectedcropsat differentproductionpercentiles isshowninTable1.Thevalues werederivedbyplottingthegreen–blueWFoftherespectivecrops fromsmallesttolargestWFagainstthecumulativepercentageof
cropproduction.Thecurvesarerelativelyflatinthefirst(best)half of theglobal production.The second (worst)half ofthe global productionshowsasteepercurve,withverylargeWFvaluesfor thelast 10–20% of production.As a result,the WFat the50th percentile of production is generally smaller than the global averageWF,aswasalreadyexplainedintheprevioussectionfor thecaseofwheat.SupplementaryMaterial–AppendixAprovides thegreen–blueWFatdifferentproductionpercentilesforall124 cropsstudied.
ThemapsinFig.4showthespatialvariabilityofthegreen–blue WF of the tenselected cropsacross theworld.The ranges are chosensuchthatonecaneasilyseeinwhichpartsoftheworld, productionoccursatWFsintherangeofthebest10%ofglobal production,etc.OnecanimmediatelyseethatrelativelysmallWFs arenotinherenttohigh-incomecountriesorhumidregionsand Fig.1.Green–blueWFofwheat(inm3
/ton)forallwheat-producingcountriesintheworld,plottedfromsmallesttolargestWF.
Fig.2.Green–blueWFofwheat(inm3
that large WFs are not intrinsically connected to low-income countriesor(semi-)aridregions.Thisismorepreciselyshownin Table2.AlthoughlowWFsasfoundinthebest10–20%ofglobal production are mostly found in high income and temperate regions,wecanfindthedifferentpercentiles inallpartsof the world, also in low income and tropical regions. High-income countries have a greater capacity to implement best available technologyandbestpracticesthanlessdevelopedcountries,but the presence of the best percentiles of production in the less developedandtropicalcountriesindicatesthatreductionofWFsto thebest10thpercentileofcurrentglobalproductionistechnically feasibleeverywhere.
In order to compare the results from this study with the literature,wecollecteddatafromanumberofwater productivi-ty studies for different crops and locations. We used four publications(DoorenbosandKassam,1979;Hatfieldetal.,2001; Sadras et al., 2007; Zwart and Bastiaanssen, 2004) that summarizecropwaterproductivitiesfromvariousstudies.Since the different studies relate to dissimilar climate and soil conditionsandwatermanagementpractices,thewater produc-tivityvaluesforagivencropvaryoverawiderange.Fig.5shows the inverse of the water productivity ranges collected from literaturetogetherwiththegreen–blueWFatdifferent produc-tionpercentilesfromthecurrentstudy.Inmostcases,theranges foundintheliteratureoverlapwellwiththevaluesfoundinthis
study.Insomecases,thelowestvaluefoundintheliteratureis substantiallysmallerthantheWFatthebest10thpercentileof global production (millet, sorghum, cotton, soybean, chickpea, maize, banana), while other cases show the reverse (barley, green bean, pepper, potato and sugar beet). The values from literature are too random and probably not representative enoughtoreflectglobalvariabilitytodrawanyconclusionshere based onthecomparison.Ingeneralthoughit canbesaidthat thisstudyisthefirstinitssortandthatitwillbeusefultostudy the sensitivityof thebenchmark values presentedhere tothe underlyingmodelanddata.
3.3.Thegreywaterfootprintofdifferentcropsatdifferentproduction percentiles
Thenitrogen-relatedgreyWFfortenselectedcropsatdifferent productionpercentilesispresentedinTable3.Thevariabilityinthe greyWFacrosscropsand spaceismainlydue todifferences in nitrogen(N)application(kg/ha)andcropyield(ton/ha).Thegrey WFatdifferentproductionpercentilesforallcropsisprovidedin SupplementaryMaterial–AppendixB.
ApplicationofNfertilizerinfluencescropwaterproductivityby affectingtherateofphotosynthesis,canopysizeandtheharvest index(Sadrasetal.,2007).Napplicationgenerallyincreasesgrain yieldandwaterproductivitysignificantly(Belderetal.,2005),but
Table1
Green–bluewaterfootprintforafewselectedcropsatdifferentproductionpercentiles.
Crop Green–bluewaterfootprint(m3/ton)atdifferentproductionpercentiles Globalaverage
10th 20th 25th 50th Barley 447 516 546 1029 1292 Cotton 1666 1821 1898 2880 3589 Maize 503 542 562 754 1028 Millet 2292 2741 2905 3653 4363 Potatoes 92 137 154 216 224 Rice 599 859 952 1476 1486 Sorghum 1001 1082 1122 1835 2960 Soybean 1553 1605 1620 1931 2107 Sugarcane 112 123 128 175 197 Wheat 592 992 1069 1391 1620
Fig.3. Green–blueWFofwheat(inm3
/ton)forallwheat-producinggridcellsintheworld,plottedfromsmallesttolargestWF.
theincreaseincropyieldandwaterproductivityisachievedonly uptoacertainleveloffertilization(Sandhuetal.,2012).Ensuring adequateNsupplyiscriticalforgoodwaterproductivity,butonlya fractionoftheappliedNfertilizerisrecoveredbyplants(Addiscott, 1996;Kingetal.,2001;Maetal.,2008;Noulaset al.,2004)andon averageabout16%oftheappliedNispresumedtobelosteitherby denitrificationorleaching(Addiscott,1996).Therefore,thereisa trade-offbetweenhighercropwaterproductivityandincreasing water pollution resulting fromthe loss of N to thefreshwater system.Thistrade-off needstobeconsideredcarefullybecause maximizingwaterproductivitymayresultindeterioratingwater qualitythroughnutrientpollution.
3.4.Watersavingandreducedwaterpollutionwhenreducingwater footprintsdowntobenchmarkvalues
Table4presentstheglobalgreen–bluewatersavingthatcould beachievedwhen,worldwide,theWFincropproductionwouldbe broughtdowntocertainbenchmarkvalues.Asbenchmarkvalues, wehaveusedtheWFsassociatedwiththebest10th,20th,25thand 50th percentileof current production. The global water saving related to improved water productivity in crop production increaseswhentheWFbenchmarkvaluesgetsmaller(fromthe 50thtothe10thpercentile).IfthegapbetweencurrentWFlevels andtheglobalbenchmarkvaluesatthe25thpercentileofcurrent Fig.4. Spatialdistributionofthegreen–bluewaterfootprintofselectedcrops(inm3
productioniseliminated,theglobalwatersavingwouldbe39%.In absoluteterms, thelargestWFreduction is observedfor cereal crops: wheat (375 billionm3/year), rice (350 billionm3/year),
maize (296 billionm3/year), sorghum (111 billionm3/year) and
barley(110billionm3/year).Inthecaseoffurtherreductiontothe levelsofthebest10thpercentilesofcurrentglobalproduction,the globalwatersavingwouldbe52%comparedtotoday.Thepotential reductionofthegreen–blueWFrelatedtocropproductionforall cropsispresentedinSupplementaryMaterial–AppendixC.
Thepossiblereductionsinwaterpollutionareevengreaterthan thepossiblereductionsinconsumptivewateruse(Table5).Inthe casethatgreyWFsincropproductionarereduced,worldwide,to thelevelofthebest25thpercentileofcurrentglobalproduction, waterpollutionisreducedby54%.IfgreyWFspertonofcropare furtherreducedtothelevelofthebest10thpercentileofcurrent production,waterpollutionisreducedby79%comparedtotoday's pollution level. The potential reduction of the water pollution related tocropproduction forall cropsis presentedin Supple-mentaryMaterial–AppendixD.
4.Discussion
WedevelopedWFbenchmarkvaluesforcropproductionbased onthe spatialvariabilityin WFsof cropsworldwide, usingthe global assessment published earlier (Mekonnen and Hoekstra, 2011). The current study is the first proposing global WF benchmarks forcropsbasedonsuchspatialvariabilityanalysis. Itwillbeusefultocarryoutsimilaranalyseswithothermodels thantheoneusedinMekonnenandHoekstra(2011)totestthe sensitivity of the outcomes tothe model used. In addition, as proposedby(Hoekstra,2013a,b),itwouldbeusefultodevelopWF benchmarksfrominsightsonwhatcanbereachedbasedonbest available technologyandpractice.The currentstudy showsthe spatialdistributionofWFsintermsofm3/tonbasedonregional
differences in evapotranspiration andyields, but itprovides no insightinwhyWFsarerelativelysmallorlargeinspecificregions andhowWFscanactuallybeloweredinthoseregionswherethey arelarge.
WehaveestablishedglobalWFbenchmarkvaluesinsteadof specific benchmarks for different agro-climatic or economic regions.Onemayarguethatclimaticorsoilconditionscanbea limiting factor for reducing the WF and different regional benchmarkvaluesshouldbeestablisheddependingongrowing conditionsperregion.However,althoughclimaticandsoilfactors areimportantindeterminingevapotranspirationfromcropfields and yields, the green–blue WF of crops in m3/ton is largely
determined by agricultural management rather than by the environment in which the crop is grown (Rockström et al., 2007; Mekonnen and Hoekstra, 2011). The higher evaporative demand in tropical climates compared to temperate regions is largelycompensatedforby moreefficient photosynthetic path-ways,whichresultsinhighercropgrowthperunitoftranspiration
flow, sothat thetotal cropwater requirementsare onaverage similar between hydro-climatic zones (Rockström, 2003). A largeincreaseincropyields,withoutanincreaseorevenwitha decreaseinfieldevapotranspiration,isachievableformostcrops acrossthedifferentclimateregionsoftheworldthroughproper nutrient, water and soil management (Mueller et al., 2012). Therefore, water productivities as shown in the best 10th percentileofglobalcropproductioncanbeachievedirrespective ofclimate,whichisalsoshowninourcomparisonofWFsofcrops acrossdifferent climateregions. Weacknowledge, though, that furtherstudymayrefineourglobalbenchmarksintoclimate– soil-specificbenchmarks.Anotherdiscussionreferstosetting bench-marks for low versushigh-incomecountries. One maypropose another(lessstrict)WFbenchmarkforlow-incomecountries,with
T able 2 Per centag e o f grid cells in differ ent income and climate regio ns in which cro ps ha v e a W F belo w the WF at the 1 0th, 20th, 25th, or 50th per centile of globa l p roduction. Cr op Per centag e o f grid cells with WF below the 1 0th per centile Percenta g e of grid cells with WF belo w the 20th percentile Per centag e o f grid cells with WF belo w the 25th per centile Per centag e o f grid cells with WF belo w the 50th per centile Income class Climat e class Income class Climat e class Income class Climat e class Income class Climat e class Lo w income Middle income High income T e mper at e T ropics Low income Middle income High income T e mperat e T ro pics Low income Middle income High income T e mper at e T ro pics Lo w income Middle income High income T emper at e T ro pics Bar ley 1 4 9 9 1 1 6 1 7 1 5 2 1 7 22 1 8 3 1 0 2 6 6 3 4 7 11 Cot ton 1 9 3 1 2 2 2 1 8 7 2 1 5 2 22 8 2 5 6 7 4 8 5 2 5 6 1 8 Maize 0 0 2 7 8 0 0 0 38 1 2 0 0 0 4 4 1 4 0 2 8 75 29 3 Millet 8 3 4 6 9 4 0 9 1 8 43 78 50 1 8 22 49 84 54 22 3 7 79 1 0 0 7 3 3 4 Po tato es 8 0 1 7 6 8 1 4 3 5 9 1 8 1 3 1 5 5 7 1 23 1 6 26 28 95 4 7 30 Rice 0 5 5 7 0 0 1 3 29 1 9 2 0 1 8 4 1 26 3 8 35 86 50 1 3 Sorghum 1 1 6 4 9 3 2 4 2 2 4 5 7 4 0 8 2 2 6 6 2 4 5 9 5 3 9 9 6 6 6 2 2 Soybe an 0 1 0 2 6 1 2 9 0 11 3 6 1 5 1 0 0 11 39 1 6 1 0 1 2 0 7 3 2 7 2 1 Sugar cane 8 1 0 1 0 1 5 6 10 14 1 8 2 0 9 11 1 7 2 2 2 2 1 2 2 0 4 8 7 7 5 5 3 7 Wheat 0 2 11 5 3 5 8 23 1 3 9 7 1 4 25 1 7 11 20 33 46 40 22
theargumentthatachievingacertainwaterproductivityina low-incomecountryismoredifficultthaninahigh-incomecountry, buttherearetwoargumentsagainstthat,first,realityshowsthat formostofthecropsstudied,thewaterproductivitiesthatcanbe associatedwiththebest10thpercentileofglobalcropproduction canbefoundinboth lowand high-incomecountries(Table2). Second, there seems little reason to set other environmental standardsfordevelopingandindustrialisedcountries,eventhough itcan indeedbea greaterchallengein developing countriesto achievecertainimprovements.
When developing green–blue WF benchmarks we have not distinguishedbetween rain-fedand irrigated agriculture. Intui-tively one would assume that water productivities in irrigated agriculturearegenerallyhigher,thusWFbenchmarksshouldbe lower, but this appears not to be the case. As Mekonnen and Hoekstra (2011) have shown, the global average consumptive (green–blue)WFpertonofcropislowerinirrigatedthanin rain-fedagricultureforsomecrops(e.g.maize,rice,coffee,cotton)but higher for other crops (e.g.soybean, sugarcane, rapeseed). The formercasecanbeexplainedbythefactthatirrigatedyieldsare Fig.5. Comparisonofthegreen–blueWFofselectedcropsatdifferentproductionpercentileswithvaluesreportedinliterature.Sourcesoftheliteraturevalues:wheat,maize, riceandcottonfromZwartandBastiaanssen(2004);sorghum,millet,soybean,sunflowerfromSadrasetal.(2007);barleyandchickpeafromHatfieldetal.(2001);therest fromDoorenbosandKassam(1979).
Table3
Greywaterfootprintatdifferentproductionpercentiles. Crop Greywaterfootprint(m3
/ton)atdifferentproductionpercentiles Globalaverage
10th 20th 25th 50th Barley 23 53 64 121 131 Cotton 0 63 175 469 440 Maize 71 128 138 171 194 Millet 0 0 0 63 115 Potatoes 16 22 24 38 63 Rice 71 129 162 215 187 Sorghum 0 0 0 40 87 Soybean 9 9 10 11 37 Sugarcane 3 7 8 11 13 Wheat 27 82 99 144 208 Table4
Globalgreen–bluewatersavingifeverywherethewaterfootprintofcropproductionisreducedtothelevelofthebest10th,20th,25thor50thpercentileofcurrent production.
Crop Globaltotalgreen–bluewaterfootprint(billionm3
/year) Green–bluewatersaving(%)inthecaseofworldwideWFreductiontothelevelofthebestxth percentileofcurrentproduction
10th 20th 25th 50th Barley 184 66 61 60 36 Cotton 207 54 50 49 30 Maize 648 51 48 46 35 Millet 126 49 39 36 25 Potatoes 70 59 42 36 17 Rice 881 60 44 40 18 Sorghum 177 67 64 63 50 Soybean 363 26 24 23 15 Sugarcane 254 43 38 35 21 Wheat 964 64 43 39 25 Others 2750 47 40 37 23 Total 6625 52 42 39 25
generally larger than rain-fed yields, not just because of the additional water, but also because other differences (e.g. crop varieties grown, nutrient supply). The reverse case can be explainedbythefactthatmarginalwaterproductivitydecreases withincreasingwatersupply,sothatbeyondacertainpointmore irrigationmay still increase land productivity (ton/ha),but not water productivity (ton/m3). In general, the higher water productivities that can be observed in some of the rain-fed agriculturearenotspecifictorain-fedagriculture;theycanalsobe achieved in irrigated agriculture (by applying smart irrigation technologyandpractice).Reversely,thehigherwater productiv-itiesfoundinsomeoftheirrigatedagriculturearenotspecificto irrigated agriculture; provided that green water resources are sufficient,theycouldalsobeachievedinrain-fedagriculture(by bettersoilandnutrientmanagement).
We have developed global benchmarks for the combined green–blue WF of crops, no separate green WF and blue WF benchmarks.Thereasonisthattheratiogreentobluewilldepend onlocalgreenwaterresourcesavailability.Location-specificblue WF benchmarks can bedeveloped asa function of theoverall consumptiveWFbenchmarksandlocalgreenwateravailability.Ifa certainlocationreceivessufficientraintoachieveacertainglobal benchmark on consumptive WF, the location-specific blue WF benchmarkwillbezero.Thelessrain,belowacertainpoint,the higherthefractionbluewillneedtobetoachieveacertainwater
productivityorWFbenchmark.Thisexercisehasnotbeencarried outaspartofthecurrentstudy.
Thereareseveralstrategiestoincreasecropwaterproductivities andreducetheWFofcrops(Table6).Itwouldbehighlyvaluableto developinsightinhowvarioustechniquesandpracticesaffectgreen, blueandgreyWFsintermsofm3/ton,andhowcertaincombinations
oftechniquesandpracticeswillberequiredtoreduceWFstothe benchmarkvaluesproposedinthisstudy.
The useof fertilizers will often improvewater productivity, becauseyieldswillincreasewhilewaterconsumptioncanremain moreorlessequal.However,aboveacertainfertilizerapplication rate, yieldsmaystill slightlyincrease,but theeffectof nutrient leaching and runoff to the freshwater system will start to dominate.Whenapplyingfertilizers,thetrade-offbetweenhigher cropwater productivity (smaller green–blue WF)and potential pollutionofthegroundwaterandstreamsthroughnutrients(grey WF)shouldbeconsideredcarefully.SettingagreyWFbenchmark valueasdoneinthisreportmayhelptointegratetheissueofwater pollutionintothediscussiononwateruseefficiencyinagriculture, adiscussionthatisusuallyfullyfocusedontheconsumptivesideof freshwaterappropriation,leavingoutthepollutionside.
WhenapplyingWFbenchmarkvaluesastargetlevels, trade-offsmayberequiredwhensettingspecificgreen,blueandgreyWF targets. Particularly, grey WFs can often be easily reduced by reducing theuseof fertilizersand pesticides (andapplyingthe amountsstillusedintheoptimalwayatthebesttimesothatyields
Table6
Technologyandpracticestoreducethewaterfootprintincropproduction.
Sources:Hatfieldetal.(2001);Kijneetal.(2003);Sadrasetal.(2007);Hoekstraetal.(2011). Strategies Technologyandpractices
Increasingyield Soilnutrientsmanagement(optimizingcroprotation,theuseofcropresidues,erosioncontrol,appropriatetillage,proper applicationandtimingofmanureorartificialfertilizer)
Precisionirrigation:synchronizingwaterapplicationwithcropwaterdemand Weedandpestcontrol(throughcroprotation,propertillage,biologicalpestcontrol) Breedingofsuperiorcropvarietieswithhigheryieldandbetterdiseaseresistance Reducingnon-beneficial
evapotranspiration
Cropschedulingtoreduceevaporationduringfallowperiod Plantspacingandroworientation
Affectingcanopydevelopmentthroughagronomyandbreeding
Minimumtillagetoreducesoilwaterevaporationandconservesoilwaterduringfallowperiods Useofcropresidueandmulchestoreducesoilwaterevaporationandimprovenutrientrecycling Improvedirrigationtechniques(drip&subsurfaceirrigation)
Effectivecontrolofweedstoreducetranspirationfromweeds Enhancingeffectiveuseofrainfall Synchronizingcropschedulingandrainfall
Waterharvestingandsupplementalirrigation Table5
Reducedwaterpollutionifeverywhereintheworldthegreywaterfootprintofcropproductionisreducedtothelevelofthebest10th,20th,25thor50thpercentileofcurrent production.
Crop Globaltotalgreywaterfootprint(billionm3
/year) Reducedwaterpollution(%)inthecaseofworldwidegreyWFreductiontothelevelofthebestxth percentileofcurrentproduction
10th 20th 25th 50th Barley 19 83 63 57 27 Cotton 25 100 88 68 31 Maize 122 65 40 36 23 Millet 3 100 100 100 70 Potatoes 20 76 67 64 50 Rice 111 64 38 24 4 Sorghum 5 100 100 100 70 Soybean 6 76 76 74 73 Sugarcane 17 78 52 46 30 Wheat 123 88 65 58 43 Others 280 85 73 68 42 Total 732 79 61 54 33
arenotaffected),butatsomepointthismayreduceyieldand– sincetheevapotranspirationrate remainsequal– thusincrease thegreen–blueWFintermsofm3/ton.Asimilarthingcanhappen
whenreducingtheblueWFbyapplyinglessirrigationwater,for instancebydeficitprecisionirrigationusingdriptechnology,since atsomepointfurtherreductionofirrigationmaylowertheyieldso thattheblueWFperhectaremaystilldiminish,butthegreen,blue andgreyWFperunitofcropwillincrease.
5.Conclusion
Withincreasingwaterscarcity,thereis agrowinginterest in improvingcropwaterproductivityinordertomeetthegrowing global food demandwith the limitedfreshwater resources.The challenge is thus to produce more crops with less water, thus reducingtheWFperunitofcropproduced.Thisstudyhasdeveloped WF benchmarkvalues for a largenumber of crops grown in the world. Thestudyshowsthatwatersavingsandreducedwaterpollutioncan beverysubstantial–39%ofglobalwatersavingand54%ofreduced waterpollution–ifWFsperunitofcroparereducedtolevelssimilar to thebest quarterof global production. Our estimationof the potentialreductionintheglobalWFofcropproductionisnotmeant toimplythatthisreductioniseasilyattainable.Raisingyieldsin low-incomecountrieswillrequirelargeinvestmentsincapacitybuilding andappropriatetechnologies.
WFbenchmarksforcropsasdevelopedinthisstudycanbeused toprovideanincentiveforfarmerstoreducetheWFoftheircrops towards reasonable levelsand thus use water moreefficiently. When granting water consumption permits to farmers and developing regulations on fertiliser use, it makes sense for governmentstotakeintoaccounttherelevantWF benchmarks forthespecificcropsgrown.Thebenchmarksareequallyrelevant forthefood-processingindustry,whichincreasinglyfocusseson theefficient andsustainableuseof waterin theirsupplychain (Sikirica,2011;TCCC,TNC,2010;Unilever,2012).Thesameholds fortheapparelsector,particularlyregardingcotton(Frankeand Mathews,2013), forthecosmetics industry,whichusesvarious sortsof agriculturalinputs(Francke and Castro,2013), and the biofuelsector(Gerbens-Leenesetal.,2009).WFbenchmarkswill enablethe actors along supply chains – from farmers through intermediate companies to final consumers – to compare the actualWFofproductsagainstcertainreferencelevels(Hoekstra, 2014).Thebenchmarkvaluescanbeusedtomeasureperformance, tosetWF reduction targets and monitor progressin achieving thesetargets.
AppendixA.Supplementarydata
Supplementarydataassociatedwiththisarticlecanbefound,in the online version, at http://dx.doi.org/10.1016/j.eco-lind.2014.06.013.
AppendixA–D.Supplementarydata
ThefollowingareSupplementarydatatothisarticle: References
Addiscott,T.M.,1996.Fertilizersandnitrateleaching.In:Hester,R.E.,Harrison,R.M. (Eds.), Agricultural Chemicalsand the Environment.The Royal Societyof Chemistry,pp.1–26.
Amir,J.,Sinclair,T.R.,1991.Amodelofwaterlimitationonspringwheatgrowthand yield.FieldCropsRes.28,59–69.
Asseng,S.,Keating,B.A.,Fillery,I.R.P.,Gregory,P.J.,Bowden,J.W.,Turner,N.C.,Palta,J. A.,Abrecht,D.G.,1998.PerformanceoftheAPSIM-wheatmodelinwestern Australia.FieldCropsRes.57,163–179.
Asseng,S.,Turner,N.C.,Keating,B.A.,2001.Analysisofwater-andnitrogen-use efficiencyofwheatinaMediterraneanclimate.PlantSoil233,127–143.
Belder,P.,Spiertz,J.H.J.,Bouman,B.A.M.,Lu,G.,Tuong,T.P.,2005.Nitrogeneconomy andwaterproductivityoflowlandriceunderwater-savingirrigation.Field CropsRes.93,169–185.
Biradar,C.M.,Thenkabail,P.S.,Platonov,A.,Xiao,X.,Geerken,R.,Noojipady,P.,Turral, H.,Vithanage,J.,2008.Waterproductivitymappingmethodsusingremote sensing.J.Appl.RemoteSens.2,023522–023544.
Brauman,K.A.,Siebert,S.,Foley,J.A.,2013.Improvementsincropwaterproductivity increasewatersustainabilityandfoodsecurity—aglobalanalysis.Environ.Res. Lett.8,024030.
Cai,X.,Thenkabail,P.S.,Biradar,C.M.,Platonov,A.,Gumma,M.,Dheeravath,V., Cohen,Y.,Goldlshleger,N.,Dor,E.B.,Alchanatis,V.,Vithanage,J.,Markandu,A., 2009. Water productivity mappingusing remote sensingdata of various resolutionstosupport“morecropperdrop”.J.Appl.RemoteSens.3,033523– 033557.
Cai,X.L.,Sharma,B.R.,2010.Integratingremotesensing,censusandweatherdata foranassessmentofriceyield,waterconsumptionandwaterproductivityin theIndo–Gangeticriverbasin.Agric.WaterManag.97,309–316.
Chapagain,A.K.,Hoekstra,A.Y.,Savenije,H.H.G.,Gautam,R.,2006.Thewaterfootprint ofcottonconsumption:an assessment ofthe impact ofworldwideconsumptionof cottonproductsonthewaterresourcesinthecottonproducingcountries.Ecol. Econ.60,186–203.
Doorenbos,J.,Kassam,A.H.,1979.Yield ResponsetoWater,FaoDrainageand IrrigationPaper.FoodandAgricultureOrganization,Rome,Italy.
Fader,M.,Gerten,D.,Thammer,M.,Heinke,J.,Lotze-Campen,H.,Lucht,W.,Cramer, W.,2011. Internalandexternalgreen–blueagriculturalwaterfootprints of nations,andrelatedwaterandlandsavingsthroughtrade.Hydrol.EarthSyst. Sci.15,1641–1660.
Falkenmark, M., Rockström, J., Karlberg, L., 2009. Present and future water requirementsforfeedinghumanity.FoodSecur.1,59–69.
Finger,R.,2013.Morethanthemean—anoteonheterogeneityaspectsinthe assessmentofwaterfootprints.Ecol.Indic.29,145–147.
Francke,I.C.M.,Castro,J.F.W.,2013.Carbonandwaterfootprintanalysisofasoapbar producedinBrazilbyNaturaCosmetics.WaterResour.Ind.1-2,37–48. Franke,N.,Mathews,R.,2013.C&A’sWaterFootprintStrategy:CottonClothing
SupplyChain.WaterFootprintNetwork/C&AFoundation, Netherlands/Zug, Switzerland.
Gerbens-Leenes,W.,Hoekstra,A.Y.,vanderMeer,T.H.,2009.Thewaterfootprintof bioenergy.Proc.Natl.Acad.Sci.106,10219–10223.
Gleick,P.H.,1993.Waterandconflict:freshwaterresourcesandinternational security.Int.Secur.18,79–112.
Gleick,P.H.,2003.Globalfreshwaterresources:soft-pathsolutionsforthe21st century.Science302,1524–1528.
Grassini, P., Hall, A.J., Mercau, J.L., 2009. Benchmarking sunflower water productivityinsemiaridenvironments.FieldCropsRes.110,251–262. Hatfield,J.L.,Sauer,T.J.,Prueger,J.H.,2001.Managingsoilstoachievegreaterwater
useefficiency.Agron.J.93,271–280.
Hoekstra,A.Y.,2013a.TheWaterFootprintofModernConsumerSociety.Routledge, London,UK.
Hoekstra,A.Y.,2013b.Wisefreshwaterallocation:waterfootprintcapsbyriver basin,benchmarksbyproductandfairwaterfootprintsharesbycommunity. Value of Water Research Report Series. UNESCO-IHE, Delft, The Netherlands.
Hoekstra,A.Y.,2014.Waterscarcitychallengestobusiness.Nat.Clim.Change4,318– 320.
Hoekstra,A.Y.,Chapagain,A.K.,2007.Waterfootprintsofnations:wateruseby peopleasafunctionoftheirconsumptionpattern.WaterResour.Manag.21,35– 48.
Hoekstra,A.Y.,Chapagain,A.K.,2008.GlobalizationofWater:SharingthePlanet’s FreshwaterResources.BlackwellPublishing,Oxford,UK.
Hoekstra,A.Y.,Chapagain,A.K.,Aldaya,M.M.,Mekonnen,M.M.,2011.TheWater FootprintAssessmentManual:SettingtheGlobalStandard.Earthscan,London, UK.
Hoekstra,A.Y.,Mekonnen,M.M.,2012.Thewaterfootprintofhumanity.Proc.Natl. Acad.Sci.109,3232–3237.
Hoekstra,A.Y.,Mekonnen,M.M.,Chapagain,A.K.,Mathews,R.E.,Richter,B.D.,2012. Global monthly water scarcity: blue water footprints versus blue water availability.PLoSOne7,e32688.
Kijne,J.W.,Tuong,T.P.,Bennett,J.,Bouman,B.,Oweis,T.,2003,2003.Ensuringfood securityviaimprovementincropwaterproductivity.In:CGIAR(Ed.),Challenge Program on Water and Food: Background Papers to the Full Proposal. InternationalWaterManagementInstitute,Colombo,SriLanka.
King,J.A.,Sylvester-Bradley,R.,Rochford,A.D.H.,2001.Availabilityofnitrogenafter fertilizerapplicationstocereals.J.Agric.Sci.136,141–157.
Kottek,M.,Grieser,J.,Beck,C.,Rudolf,B.,Rubel,F.,2006.WorldmapoftheKöppen– Geigerclimateclassificationupdated.MeteorologischeZeitschrift15,259–263. Ma,W.,Li,J.,Ma,L.,Wang,F.,I,Sisák,Cushman,I.,Zhang,G.,F,2008.Nitrogenflow anduseefficiencyinproductionandutilizationofwheat,rice,andmaizein China.Agric.Syst.99,53–63.
Mekonnen,M.M.,Hoekstra,A.Y.,2010.Aglobalandhigh-resolutionassessmentof thegreen,blueandgreywaterfootprintofwheat.Hydrol.EarthSystSci.14, 1259–1276.
Mekonnen,M.M.,Hoekstra,A.Y.,2011.Thegreen,blueandgreywaterfootprintof cropsandderivedcropproducts.Hydrol.EarthSyst.Sci.15,1577–1600. Mueller,N.D.,Gerber,J.S.,Johnston,M.,Ray,D.K.,Ramankutty,N.,Foley,J.A.,2012.
Closingyieldgapsthroughnutrientandwatermanagement.Nature490,254– 257.
Noulas,C.,Stamp,P.,Soldati,A.,Liedgens,M.,2004.Nitrogenuseefficiencyofspring wheatgenotypesunderfieldandlysimeterconditions.J.Agron.CropSci.190, 111–118.
Oki,T.,Kanae,S.,2006.Globalhydrologicalcyclesandworldwaterresources. Science313,1068–1072.
Oweis,T.,Zhang,H.,Pala,M.,2000.Wateruseefficiencyofrainfedandirrigated breadwheatinaMediterraneanenvironment.Agron.J.92,231–238. Passioura,J.,2006.Increasingcrop productivitywhenwateris scarce–from
breedingtofieldmanagement.Agric.WaterManag.80,176–196.
Postel,S.L.,2000.Enteringaneraofwaterscarcity:thechallengesahead.Ecol.Appl. 10,941–948.
Rahman,S.M.,Khalil,M.I.,Ahmed,M.F.,1995.Yield-waterrelationsandnitrogen utilizationbywheatinsalt-affectedsoilsofBangladesh.Agric.WaterManage. 28,49–56.
Robertson,M.J.,Kirkegaard,J.A.,2005.Water-useefficiencyofdrylandcanolainan equi-seasonalrainfallenvironment.Aust.J.Agric.Res.56,1373–1386. Rockström,J.,2003.Waterforfoodandnatureindrought-pronetropics:vapour
shiftinrain-fedagriculture.Philos.Trans.RoyalSoc.London.SeriesB:Biol.Sci. 358,1997–2009.
Rockström,J.,Lannerstad,M.,Falkenmark,M.,2007.Assessingthewaterchallenge ofanewgreenrevolutionindevelopingcountries.Proc.Natl.Acad.Sci.104, 6253.
Rosegrant,M.W.,Ringler,C.,Zhu,T.,2009.Waterforagriculture:maintainingfood securityundergrowingscarcity.Annu.Rev.Environ.Resour.34,205–222. Sadras,V.,Baldock,J.,Roget,D.,Rodriguez,D.,2003.Measuringandmodellingyield
andwaterbudgetcomponentsofwheatcropsincoarse-texturedsoilswith chemicalconstraints.FieldCropsRes.84,241–260.
Sadras,V.O.,Grassini,P.,Steduto,P.,2007.StatusofWaterUseEfficiencyofMain Crops,SOLAWBackgroundThematicReport.FoodandAgricultural Organiza-tion,Rome,Italy.
Sandhu,S.S.,Mahal,S.S.,Vashist,K.K.,Buttar,G.S.,Brar,A.S.,Singh,M.,2012.Crop andwaterproductivityofbedtransplantedriceasinfluencedbyvariouslevels ofnitrogenandirrigationinnorthwestIndia.Agric.WaterManag.104,32–39. Sharma,D.K.,Kumar,A.,Singh,K.N.,1990.Effectofirrigationschedulingongrowth, yieldandevapotranspirationofwheatinsodicsoils.Agric.WaterManag.18, 267–276.
Sharma,N.K.,Samra,J.S.,Singh,H.P.,2001.Influenceofboundaryplantationof poplar(PopulusdeltoidesM.)onsoil–wateruseandwateruseefficiencyof wheat.Agric.WaterManag.51,173–185.
Siebert,S.,Döll,P.,2010.Quantifyingblueandgreenvirtualwatercontentsinglobal cropproductionaswellaspotentialproductionlosseswithoutirrigation.J. Hydrol.384,198–217.
Sikirica,N.,2011.Waterfootprintassessmentbananasandpineapples.DoleFood Company,Soil&MoreInternational,DriebergenTheNetherlands.
TCCC,TNC,2010.Productwaterfootprintassessments:Practicalapplicationin corporatewaterstewardship.TheCoca-ColaCompany/TheNatureConservancy, Atlanta,USA/Arlington,USA.
Unilever,2012.SustainableLivingPlan:Progressreport2012.UnileverRotterdam, TheNetherlands.
Vörösmarty,C.J.,McIntyre,P.B.,Gessner,M.O.,Dudgeon,D.,Prusevich,A.,Green,P., Glidden,S.,Bunn,S.E.,Sullivan,C.A.,Liermann,C.R.,Davies,P.M.,2010.Global threatstohumanwatersecurityandriverbiodiversity.Nature467,555–561. Wada,Y.,vanBeek,L.P.H.,Viviroli,D.,Dürr,H.H.,Weingartner,R.,Bierkens,M.F.P.,
2011.Globalmonthlywaterstress:2.Waterdemandandseverityofwater stress.WaterResour.Res.47,W07518.
World Bank, 2012. Countries Classification: Income Group. The World Bank, Washington,D.C.,USA.
Zhang,H.,Wang,X.,You,M.,Liu,C.,1999.Water-yieldrelationsandwater-use efficiencyofwinterwheatintheNorthChinaPlain.Irrig.Sci.19,37–45. Zhang,J.,Sui,X.,Li,B.,Su,B.,Li,J.,Zhou,D.,1998.Animprovedwater-useefficiency
forwinterwheatgrownunderreducedirrigation.FieldCropsRes.59,91–98. Zwart,S.J.,Bastiaanssen,W.G.M.,2004.Reviewofmeasuredcropwaterproductivity values forirrigated wheat,rice, cotton andmaize. Agric. Water Manag. 69,115–133. Zwart,S.J.,Bastiaanssen,W.G.M.,2007.SEBALfordetectingspatialvariationofwater productivityandscopeforimprovementineightirrigatedwheatsystems.Agric. WaterManag.89,287–296.
Zwart, S.J.,Bastiaanssen,W.G.M., deFraiture, C.,Molden,D.J., 2010a.Aglobal benchmarkmapofwaterproductivityforrainfedandirrigatedwheat.Agric. WaterManag.97,1617–1627.
Zwart,S.J.,Bastiaanssen,W.G.M.,deFraiture,C.,Molden,D.J.,2010b.WATPRO:a remotesensingbasedmodelformappingwaterproductivityofwheat.Agric. WaterManag.97,1628–1636.
