ContentslistsavailableatSciVerseScienceDirect
Field
Crops
Research
j o ur n a l ho me p a g e :w w w . e l s e v i e r . c o m / l o c a t e / f c r
Use
of
agro-climatic
zones
to
upscale
simulated
crop
yield
potential
Justin
van
Wart
a,∗, Lenny
G.J.
van
Bussel
b, Joost
Wolf
b, Rachel
Licker
c, Patricio
Grassini
a,
Andrew
Nelson
d,
Hendrik
Boogaard
e,
James
Gerber
f,
Nathaniel
D.
Mueller
f,
Lieven
Claessens
g,
Martin
K.
van
Ittersum
b,
Kenneth
G.
Cassman
aaDepartmentofAgronomyandHorticulture,UniversityofNebraska-Lincoln,Lincoln,NE68583-0915,USA bPlantProductionSystemsGroup,WageningenUniversity,P.O.Box430,6700AK,Wageningen,TheNetherlands cWoodrowWilsonSchoolofPublicandInternationalAffairs,PrincetonUniversity,Princeton,NJ08544,USA dInternationalRiceResearchInstitute(IRRI),LosBa˜nos4031,Philippines
eAlterra,WageningenUniversityandResearchCentre,P.O.Box47,6700AA,Wageningen,TheNetherlands
fInstituteontheEnvironment(IonE),UniversityofMinnesota,325LearningandEnvironmentalSciences,1954BufordAvenue,SaintPaul,MN55108,USA gInternationalCropsResearchInstitutefortheSemi-AridTropics(ICRISAT),P.O.Box39063,00623Nairobi,Kenya
a
r
t
i
c
l
e
i
n
f
o
Articlehistory:Received23October2012 Receivedinrevisedform 27November2012 Accepted30November2012 Keywords: Agroecologicalzone Climatezone Yieldpotential Water-limitedyield Yieldgap Extrapolationdomain Globalfoodsecurity
a
b
s
t
r
a
c
t
Yieldgapanalysis,whichevaluatesmagnitudeandvariabilityofdifferencebetweencropyieldpotential (Yp)orwaterlimitedyieldpotential(Yw)andactualfarmyields,providesameasureofuntappedfood productioncapacity.Reliablelocation-specificestimatesofyieldgaps,eitherderivedfromresearchplots orsimulationmodels,areavailableonlyforalimitednumberoflocationsandcropsduetocostandtime requiredforfieldstudiesorforobtainingdataonlong-termweather,croprotationsandmanagement practices,andsoilproperties.Giventheseconstraints,wecompareglobalagro-climaticzonationschemes forsuitabilitytoup-scalelocation-specificestimatesofYpandYw,whicharethebasisforestimating yieldgapsatregional,national,andglobalscales.Sixglobalclimatezonationschemeswereevaluated forclimatichomogeneitywithindelineatedclimatezones(CZs)andcoverageofcroparea.Anefficient CZschemeshouldstrikeaneffectivebalancebetweenzonesizeandnumberofzonesrequiredtocovera largeportionofharvestedareaofmajorfoodcrops.ClimateheterogeneitywasverylargeinCZschemes withlessthan100zones.Oftheotherfourschemes,theGlobalYieldGapAtlasExtrapolationDomain (GYGA-ED)approach,basedonamatrixofthreecategoricalvariables(growingdegreedays,aridityindex, temperatureseasonality)todelineateCZsforharvestedareaofallmajorfoodcrops,achievedreasonable balancebetweennumberofCZstocover80%ofglobalcropareaandclimatehomogeneitywithinzones. WhileCZschemesderivedfromtwoclimate-relatedcategoricalvariablesrequireasimilarnumberof zonestocover80%ofcroparea,within-zoneheterogeneityissubstantiallygreaterthanfortheGYGA-ED formostweathervariablesthataresensitivedriversofcropproduction.SomeCZschemesare crop-specific,whichlimitsutilityforup-scalinglocation-specificevaluationofyieldgapsinregionswithcrop rotationsratherthansinglecropspecies.
©2012ElsevierB.V.Allrightsreserved.
1. Introduction
Growingdemandforfoodincomingdecadeswillrequire sub-stantialincreaseincropproduction(Godfrayetal.,2010).Given disadvantagesand limitations ofmassive expansion of existing cropland,suchaslossofbiodiversity andincreasingGHG emis-sions,itisofcriticalimportancetoknowwhereandhowbestto increasecropyieldonexistingcroplandarea(Foley etal.,2005; Tilmanetal.,2002).Yieldgap(Yg)analysis,anevaluationofthe differencebetweencropyieldpotentialandactualfarmers’yields
∗Correspondingauthor.
E-mailaddress:[email protected](J.vanWart).
(Lobell etal.,2009),providesaquantitativeestimateofpossible increaseinfoodproductioncapacityforagivenlocation,whichis acriticalcomponentofstrategicfoodsecurityplanningatregional, nationaland globalscales.For irrigated croppingsystems,yield potential(Yp)isdefinedastheyieldofcropcultivarwhengrown withoutlimitationsfromwater,nutrients,pestsanddiseases;in rainfedcroppingsystem,water-limitedyieldpotential(Yw)isalso determinedbywatersupplyamountanddistributionduringthe croppingseason(vanIttersumetal.,2013).Atagivenlocation,Yg isthedifferencebetweenYporYwandactualyield.
BothYpandYwaresite-specificbecausetheyaredetermined byweather,management,lengthofgrowingseason,andsoil prop-ertiesthataffectroot-zonewaterstoragecapacity(thelatterfor Yw only).Bothcanbeestimated fromresearchplots, in which
0378-4290/$–seefrontmatter©2012ElsevierB.V.Allrightsreserved. http://dx.doi.org/10.1016/j.fcr.2012.11.023
J.vanWartetal./FieldCropsResearch143(2013)44–55 45
thecropisgrownwithoutlimitations,orbysimulationusingcrop models(Lobelletal.,2009).Inarecentcomparisonofthesetwo optionsacrossarangeofcroppingsystemsandenvironments,van Ittersumetal.(2013)concludedthatuseofcropsimulationwitha long-termweatherdatabaseprovidesamorerobustestimateofYp andYwthanresearchplotsbecausesimulationbetteraccountsfor theimpactofvariationintemperature,solarradiation,andrainfall overtime.Butuseofcropmodelsrequiresreliablelocation-specific dataonsowingdate,cultivarmaturity,plantpopulation,soilsand weatherandsuchdataarenotgenerallyavailableformostlocations (Ramirez-VillegasandChallinor,2012).Obtainingthesedataata largenumberoflocationsistime-consuming,costly,andoften sim-plynotfeasible.Therefore,anupscalingmethodisneededtoextend coverageofestimatesofYpandYwbasedonlocation-specific infor-mationtoanappropriateextrapolationdomainusingaprotocol thatminimizesthenumberoflocation-specificsimulations.Ideally, extrapolationdomainswouldbesmallenoughtominimize varia-tioninclimateandcropmanagementpracticeswithinthedomain, andlargeenoughtominimizedatacollectionrequirementsto esti-mateYgatregionalandnationalscales.Likewise,relevanceofa zonationschemeforsimulationofYpandYwisdeterminedbythe quality,resolution,extentandchoiceofvariablesusedtodelineate boundaries.
Previousstudieshavedistinguishedgeographicalspaceby cli-mateandsoilclassificationschemesasabasisforextrapolating and applying agricultural information and research to broader spatialscales(Wood andPardey, 1998;Padburyet al.,2002).A region can be divided into agro-climatic zones (CZs) based on homogeneity in weather variables that have greatest influence oncropgrowthandyield,whileagro-ecologicalzones(AEZs)are definedasgeographicregionshavingsimilarclimateandsoilsfor agriculture(FAO,1978).Such zonationschemeshavebeenused toidentifyyieldvariability andlimiting factorsforcropgrowth (Caldiz et al., 2002; Williams et al., 2008), to regionalize opti-malcropmanagementrecommendations(Seppelt,2000),compare yieldtrends(GallupandSachs,2000),todeterminesuitable loca-tionsfornewcropproductiontechnologies(Geertsetal.,2006; Arayaetal.,2010),andtoanalyzeimpactsofclimatechangeon agriculture(Fischeretal.,2005).Table1includesadescriptionof previouslypublishedzonationschemesusedtoevaluate extrapola-tiondomainsforagriculturaltechnologiesandinyieldgapanalysis. OurreviewfocusesonCZschemesandtheclimaticcomponents of AEZschemes withthe goalof identifyingan appropriate CZ schemefor upscalinglocation-specificestimates ofYporYwto regionalandnationallevels.Toourknowledge,nosuchreviewhas beenpreviouslypublishedwiththisgoalinmind.Specific objec-tivesofthisreviewareto:(1)evaluatezonationschemesbasedon thedegreeofvariabilityinweathervariableswithinzones,and(2) evaluatetheusefulnessandlimitationsofthesezonationschemes forupscalinglocation-specificestimatesofYpandYwtonational levels.
2. Agro-climaticandagro-ecologicalzonationschemes
Zonationschemesessentiallyfallintotwocategories:matrix andcluster.Inthissectiondifferencesbetweenmatrixand clus-termethodologiesareexplained,andsixglobalmatrixandcluster zonationschemesusefulforextrapolationofestimatesofYporYw aredescribed.
2.1. Matrixmethodologies
Perhapsthebestknownandearliestexampleofamatrix zona-tionschemeisdescribedbyKöppen(1900).Köppendevelopeda climateclassificationsystembasedonmultiplevariablesrelated
totemperatureandprecipitation,andusedhissystemtoidentify thetypeofvegetation,includingsomecrops,thatcouldgrowin eachzone.Ina matrixzonation,each variableusedtodelineate zonesisdividedintoclassesorclass-ranges.Classcutoffvaluesfor eachvariablecanbebasedonexpertopinionorfrequency distri-butionsofthevariable’srangeofvalues.Zonesareformedbythe matrix“cells”ofintersectingclasses.Forexample,amatrixzone cellmightbeageographicareainwhichmeanannualtemperature isbetween20and25◦Candmeanannualprecipitationisbetween 300and400mm.
Matrixzonationschemesareadvantageousinthattherangeof inputparametersforallzonesisknownandspecificallydefined by theresearchers. The size of thezones in a matrix zonation resultsfromthenumber ofinputvariablesusedandthedegree ofspecificityinclassesforeachvariable,i.e.moreclassvariables andmoresub-divisionswithineachvariableresultinalarger num-berofzoneswithsmallerarea.Thus,matrixmethodologyallows for highdegreeof controlover thenumbertheresulting zones asdetermined byintendeduseof thezonationscheme.Robust matrixschemesforuscalingYpandYwwouldusethemost sensi-tiveweathervariablesforsimulationofcropyieldsunderirrigated andrainfedconditions.
2.2. Clustermethodologies
Clustermethodologies[alsoreferredtoasstatistical stratifica-tion(Hazeuetal.,2011)]reliesonmultivariatestatisticalanalyses to separatecells into a researcher-specified number of distinct zones. Clustering essentiallyinvolves assigning grid-cell values derived from mathematical or statistical modeling of categori-calvariables. Groupingor“clustering”grid-cellsbasedonthese derivedvaluesisaccomplishedusingavarietyoftechniquessuch as assigning a certain value or range of values as a class or cluster,minimizingthesumofthedifferencebetweengrid-cells within clusters, or more sophisticated Iterative Self-Organizing DataAnalysis(ISODATA).Inthelatter,thenumberofcluster cen-tersisspecified,randomlyplaced,andthenclustersaredivided ormergedbasedonstandarddeviationofgrid-cellsassigned to eachcluster(TouandGonzález,1974).Theprocesscontinuesuntil reassignment of grid cells nolonger improvescluster standard deviation.Duetothestatisticalnatureof“clustering,” subjectiv-ityisavoidedinselectionofclassrangesforeachvariable.Though class ranges may be more objective in clustering compared to matrixmethodology,sizeofzonesispartiallydependenton num-ber of zonesspecified by theresearcher, which may introduce subjectivity.Unlike matrixzonation,thenumberofzonesisnot determinedbythenumberofweathervariables thatdetermine thezonation.Therefore,arelativelylargenumberofvariablescan beconsideredwithoutnecessarilyreducingthesizeoftheresulting zones.
Oneofthebetterknownexamplesofaclusterzonationwas createdthroughclimate-basedmodelingofnaturalvegetationon grid-cells,whichwerethengroupedintoregionsbasedon domi-nantplanttypes(Prenticeetal.,1992).Clustermethodologiesalso havebeenusedtodeterminetheapplicabilityoffarmmanagement researchin differentregions (Seppelt, 2000), tostudy potential impacts ofclimatechangeonecosystemsand theenvironment (Metzgeretal.,2008),and toidentifypotentialnewproduction areasforbio-energycrops(EEA,2007).
2.3. Zonationschemesthatcanbeusedinestimationofyield
potential
2.3.1. TheGlobalAgro-EcologicalZonemodelingframework
TheGlobalAgro-EcologicalZonemodelingframework(GAEZ) was developed to spatially analyze agricultural systems and
Table1
Previouslypublishedglobalzonationschemes(AEZ).
AEZscheme Numberofzones TypeofAEZ Variablesconsidered,methodology Reference
FAOa 14 Matrix Meangrowingperiodtemperatureandlengthofgrowingperiod, determinedbyannualprecipitation,potentialevapotranspirationand thetimerequiredtoevapotranspire100mmofwaterfromthesoil profile
FAO(1978)
CGIAR-TACb 9 Matrix Meanannualandgrowingperiodtemperature,andlengthofgrowing period(determinedthesameasintheFAOzonationscheme)
SivakumarandValentin(1997) Prentice 17 Cluster Soiltexturebasedwater-storagecapacity,monthlyprecipitation,
sunshinehours,potentialevapotranspiration,growingdegreedays, minimumtemperature,meantemperature.Thesevariableswereused inamodelwhichcalculatedmostlikelyvegetationtypeforthe environmentofthisgridcellandcellsweregroupedbasedon vegetationtype.
Prenticeetal.(1992)
Pappadakis 74 Matrix Precipitationandtemperatureareusedincalculationsofavarietyof seasonalstatistics.Rangesofvariablesforeachzonearebasedoncrop requirements.
Papadakis(1966)
Köppen-Geiger 31 Matrix Meanannualtemperature,minimumandmaximumtemperatureof warmestandcoolestmonths,accumulatedannualprecipitation, precipitationofdriestmonth,lowestandhighestmonthly precipitationforsummerandwinterhalfyears,andadryness thresholdbasedonseasonalityofprecipitation
Kotteketal.(2006)
Holdridge 100 Matrix Meanannualtemperature,meanannualprecipitation,elevation (evaporativedemandandfrostwerealsoconsideredindetermining climaterangesofzones).
Holdridge(1947)
GAEZ-LGPc 16 Matrix Temperature,precipitation,potentialevapotranspirationandsoil characteristicsareusedtocalculatelengthofgrowingseason.
Fischeretal.(2012) HCAEZd 21 Matrix Meantemperatures,elevation,andGAEZ-LGPareusedtodefine
thermalregimesandtemperatureseasonality.
Woodetal.(2010) SAGEe 100 Matrix Growingdegreedays(GDD;
Tmean–crop-specificbasetemperature)andsoilmoistureindex(actualevapotranspiration dividedbypotentialevapotranspiration).
Lickeretal.(2010)
GLIf 25 Matrix Harvestedareaoftargetcrop,crop-specificGDDandsoilmoisture index(actualevapotranspirationdividedbypotential
evapotranspiration).
Muelleretal.(2012)
GEnSg 115 Cluster 4variables(GDDwithbasetemperatureof0◦C,anaridityindex, evapotranspirationseasonality,temperatureseasonality)usedin iso-clusteranalysisto“cluster”grid-cellsintozonesofsimilarity.
Metzgeretal.(inpress)
aFoodandAgriculturalOrganization.
b ConsultativeGrouponInternationalAgriculturalResearch– TechnicalAdvisoryCommittee. c GlobalAgro-EcologicalZoneLengthofGrowingPeriod.
d HarvestChoiceAgro-ecologicalZone.
eCenterforSustainabilityandtheGlobalEnvironment. f GlobalLandInitiative.
gGlobalEnvironmentalStratification.
evaluate the impacts of agricultural policies at a global scale (Fischer,2009).DelineationofAEZswithinGAEZaredetermined by monthly weather data with a resolution of 10 (roughly 20km×20kmattheequator,or400km2).Theweatherdatawere
obtainedfromtheClimateResearchUnit(Newetal.,2002)andthe GlobalPrecipitationClimatologyCentre(Rudolfetal.,2005). Cate-goricalvariablesused,orderived,fromthesedatatodefineanAEZ include:(a)accumulatedtemperaturesumsformeandaily temper-atureaboveabasetemperature[growingdegreedays(GDD)],(b) annualtemperatureprofiles,basedonmeanannualtemperature andwithin-yeartemperaturetrends,(c)delineationofcontinuous, discontinuous,sporadicandnopermafrostzones,(d)quantification ofsoilwaterbalanceandactualevapotranspirationforareference crop,(e)lengthofgrowingperiod(LGP),definedasthesumofdays whenmeandailytemperatureexceeds5◦Candevapotranspiration forthereferencecropexceedshalfof potential evapotranspira-tion,(f)multiplecroppingclassification,whichindicateswhether annualsingle,doubleortriplecroppingispossibleinagivenzone, basedontheLGP and assuminga growth durationper cropof 120days(Fischeret al., 2012).This GAEZframework hasbeen adaptedtoassessthepotentialproduction ofallmajorbio-fuel crops(FischerandSchrattenholzer,2001),toanalyzethe poten-tialimpactofacceleratedbiofuelproductiononfoodsecurityto 2050,andtoevaluatetheresultingsocial,environmentaland eco-nomicimpacts(Fischeretal.,2009).Additionalassessmentshave
usedaGAEZframeworktoevaluatescenariosoffuturelanduse andproductionofmajorcropsataglobalscale(Fischeretal.,2002, 2006).OfthevariousAEZschemesusedintheGAEZframework,we selectedtheonebasedonLGPinwhichLGPisderivedfrom temper-ature,precipitation,andsoilwaterholdingcapacityascategorical variables.TheGAEZ-LGPwasselectedbecauseitutilizesthemost agronomicallyrelevantcategoricalvariablesandhasthesmallest, andpresumablymostclimaticallyhomogenouszones,withinthe GAEZ-familyofAEZschemes(Figs.1a–5a).
2.3.2. CenterforSustainabilityandGlobalEnvironmentzonation
scheme
TheCenterforSustainabilityandtheGlobalEnvironment(SAGE) zonationschemewasgeneratedusingglobal,griddeddatafortwo variables knownto beimportant driversfor crop development andcropgrowth(Lickeretal.,2010):growingdegree-days(GDD) andacropsoilmoistureindex,thelattercalculatedastheratio ofactualtopotentialevapotranspirationfollowingtheapproach ofPrentice etal. (1992)and Ramankutty etal. (2002). Calcula-tionsutilized a 33-y monthlyaveraged weather databasefrom theClimateResearchUnit(Newetal.,2002)witha10 resolu-tion.Soiltexturedata usedtoestimatethesoilmoistureindex weretaken from theInternationalSoil Referenceand Informa-tionCenter witha 5 resolution (Batjes, 2006).By downscaling theweatherdatafroma10 toa5resolution,calculationswere
J.vanWartetal./FieldCropsResearch143(2013)44–55 47
Fig.1.ZonationofAfricafor(a)GlobalAgro-EcologicalZoneforlengthofgrowingseason(GAEZ-LGP),(b)CenterforSustainabilityandtheGlobalEnvironment(SAGE) zonationscheme(crop-specific,derivedusingGDDwithbasetemperatureof8◦Casusedformaize),(c)HarvestChoiceAgroecologicalZone(HCAEZ,d)GlobalLandscapes Initiative(crop-specific,derivedusingGDDwithbasetemperatureof8◦Casusedformaize),(e)GlobalEnvironmentalStratification(GEnS),(f)GlobalYieldGapAtlas ExtrapolationDomain(GYGA-ED).
carried outona 5 gridbasis (approximately 10km×10km, or 100km2attheequator).Theglobalrangesofthetwocategorical
variableswereeachdividedintotenclasses,whichwerethenused todevelopamatrixof100uniquecombinationsofgrowing degree-dayandsoilmoistureconditions.Separatezonationschemeswere developed for each of 18 cropspecies using crop-specificbase temperaturesforcalculationofgrowingdegree-days(e.g.,8◦Cfor maize,5◦Cforrice).Thezonationschemeformaizeisshownin
Figs.1b–5b.
Thiszonationschemewasdevelopedtodeterminewithin-zone maximumyieldachievedforaspecificcropwithineachofthe100 zones.Ifthezonal-maximumyieldwaslargerthanobservedyields foraparticularregionwithinthezonetheauthorsconsideredthisa Ygandidentifiedtheregionashavinganopportunityforincreasing yields(Lickeretal.,2010).TheSAGEzonationwasalsoemployed byJohnstonetal.(2011)toexamineopportunitiestoexpandglobal biofuelproductionthroughagriculturalintensificationinregions withsimilargrowingconditions.
Fig.2. ZonationofAsiafor(a)GlobalAgro-EcologicalZoneforlengthofgrowingseason(GAEZ-LGP),(b)CenterforSustainabilityandtheGlobalEnvironment(SAGE)zonation scheme(crop-specific,derivedusingGDDwithbasetemperatureof8◦Casusedformaize),(c)HarvestChoiceAgroecologicalZone(HCAEZ,d)GlobalLandscapesInitiative (crop-specific,derivedusingGDDwithbasetemperatureof8◦Casusedformaize),(e)GlobalEnvironmentalStratification(GEnS),(f)GlobalYieldGapAtlasExtrapolation Domain(GYGA-ED).
2.3.3. ModificationsofGAEZandSAGEzonationschemes
AspectsofboththeSAGEandGAEZhavebeenutilizedor mod-ifiedtodevelopimprovedAEZschemesforyieldgapanalysis.The HarvestChoice1 AEZ scheme(HCAEZ), developed foranalysis in
sub-SaharanAfrica,isanexample(Woodetal.,2000,2010).Itis
1 HarvestChoiceisalargecollaborativeefforttoprovideknowledgeproducts
aimedatguidinginvestmentstoimprovewell-farethroughmoreprofitable agri-cultureinSub-SaharanAfricaledbyscientistsfromtheUniversityofMinnesotaand
amatrixwith21zonesbasedonGAEZ-LGPandthermalregime classesfor thetropics, sub-tropics,temperate, andborealzones distinguishedbyhighlandandlowlandregions.Essentially,HCAEZ isacombination,orintersection,ofseveraldistinctand indepen-dentzonationschemesusedintheGAEZframework.Althoughit usesdataofmorerecentorgin,theHCAEZresemblesanearlier
theInternationalFoodPolicyResearchInstitute(IFPRI).Severalzonationschemes havebeenusedatHarvestChoice,basedonthesameunderlyingmethodology.
J.vanWartetal./FieldCropsResearch143(2013)44–55 49
Fig.3.ZonationofEuropefor(a)GlobalAgro-EcologicalZoneforlengthofgrowingseason(GAEZ-LGP),(b)CenterforSustainabilityandtheGlobalEnvironment(SAGE) zonationscheme(crop-specific,derivedusingGDDwithbasetemperatureof8◦Casusedformaize),(c)HarvestChoiceAgroecologicalZone(HCAEZ,d)GlobalLandscapes Initiative(crop-specific,derivedusingGDDwithbasetemperatureof8◦Casusedformaize),(e)GlobalEnvironmentalStratification(GEnS),(f)GlobalYieldGapAtlas ExtrapolationDomain(GYGA-ED).
AEZschemedevelopedbytheTechnicalAdvisoryCommittee(TAC) oftheConsultativeGrouponInternationalAgriculturalResearch (CGIAR)(TAC/CGIAR,1992;SivakumarandValentin,1997).
TheSAGEzonationschemewasmodifiedbytheGlobal Land-scapesInitiative(GLI)groupattheUniversityofMinnesota,keeping theclassificationbasedoncrop-specificGDDbutreplacingthecrop soilmoistureindexbyannualtotalprecipitation.Another modifica-tionwasthatonlyterrestrialsurfacecoveredbyharvestedareafora specificcropwasconsideredbasedongeospatialcropdistribution
mapsofMonfredaetal.(2008).Climatezonesweredevelopedfor eachcropbydividingGDDandprecipitationeachintotenclasses, theintersectionofwhichformedamatrixof100individualCZs. Insteadofusingequal rangesfor theclasses,zoneswere deter-minedusinganalgorithmsuchthat1%oftheglobalharvestedarea ofthatspecificcropwasineachzone,amethodologyknownas the‘equal-areaapproach’(Figs.1d–5d).ThisrevisionoftheSAGE zonationschemeformedthebasis oftheyieldgapestimatesin
Fig.4.ZonationofNorthAmericafor(a)GlobalAgro-EcologicalZoneforlengthofgrowingseason(GAEZ-LGP),(b)CenterforSustainabilityandtheGlobalEnvironment (SAGE)zonationscheme(crop-specific,derivedusingGDDwithbasetemperatureof8◦Casusedformaize),(c)HarvestChoiceAgroecologicalZone(HCAEZ,d)Global LandscapesInitiative(crop-specific,derivedusingGDDwithbasetemperatureof8◦Casusedformaize),(e)GlobalEnvironmentalStratification(GEnS),(f)GlobalYieldGap AtlasExtrapolationDomain(GYGA-ED).
2.3.4. TheGlobalEnvironmentalStratificationmethodology
(GEnS)
TheGlobalEnvironmentalStratification(GEnS)byMetzgeretal. (inpress)isthefirstclustermethodologyaimingatestablishing aglobal,climate-explicit zonation system.GEnS wasdeveloped withintheGrouponEarthObservationsBiodiversityObservation Network(GEOBON,Scholesetal.,2008)andwillbeavailableto assistfurtherresearchonglobalecosystems.Thisclusterzonation
usesmonthlygriddedclimatedatafromtheWorldClimdatabase (Hijmans et al., 2005) and annual aridity and potential evapo-transpirationseasonalityderivedfromtheCGIARConsortiumfor SpatialInformation(CGIAR-CSI,Trabuccoetal.,2008;Zomeretal., 2008), with 30 resolution (approximately 1km2 at the
equa-tor).GEnSwasconstructedinthreestages.Inthefirststage,42 categorical variables werescreened toremove thosethat were auto-correlated.Amongthevariableswithhighauto-correlation,
J.vanWartetal./FieldCropsResearch143(2013)44–55 51
Fig.5.ZonationofSouthAmericafor(a)GlobalAgro-EcologicalZoneforlengthofgrowingseason(GAEZ-LGP),(b)CenterforSustainabilityandtheGlobalEnvironment (SAGE)zonationscheme(crop-specific,derivedusingGDDwithbasetemperatureof8◦Casusedformaize),(c)HarvestChoiceAgroecologicalZone(HCAEZ,d)Global LandscapesInitiative(crop-specific,derivedusingGDDwithbasetemperatureof8◦Casusedformaize),(e)GlobalEnvironmentalStratification(GEnS),(f)GlobalYieldGap AtlasExtrapolationDomain(GYGA-ED).
researchersselectedthemostsensitiveparametersandeliminated theothers topreventover-weighting thezonation byco-linear variables. In thesecond step, statisticalclustering analysiswas performedonremainingvariables:annualcumulativeGDDusing basetemperature=0◦C,temperatureandpotential evapotranspi-ration seasonalities(month tomonthvariation), and anannual aridityindex (calculatedastheratio ofmean annualtotal pre-cipitationtomeanannualtotalpotentialevapotranspiration).The statisticalclusteringwascarriedoutusing principlecomponent analysisanditerativeself-organizingdataanalyses,resultingin125 zones(Figs.1e–5e).A climaticstratificationofEurope(Metzger etal.,2005)hasbeenusedin modelingeffortstoquantifycrop production potential and yield gaps in Europe (Hazeu et al., 2009).
2.3.5. TheGlobalYieldGapAtlasExtrapolationDomain
(GYGA-ED)
The goal of the Global Yield Gap Atlas (GYGA) project (www.yieldgap.org)is toestimatetheyield gapformajor food cropsin allcrop-producingcountriesbasedonlocallyobserved data. Unlike past efforts to estimate Yg that rely on gridded weatherdataasdescribedabove,GYGAseekstousea “bottom-up” approach withlocation-specific observed weather data.To extrapolateresultsfromlocation-specificobserveddata,theGYGA approach utilizes a hybrid zonation scheme, called the GYGA ExtrapolationDomain (GYGA-ED),which combines components of otherzonationschemesas reviewedin thispaper.The chal-lenge of usinga bottom-up approach is thetime, expense and access to acquire observed weather data as well as associated
location-specificinformationaboutcroprotations,soilproperties andfarmmanagement,whicharerequiredforrobustestimatesof YpandYw(vanIttersumetal.,2013).Therefore,theGYGAapproach strivesforazonationschemethatbalancesneedtominimizethe numberoflocation-specificsitesrequiringweather,soils,andcrop managementdatawiththegoalofminimizingclimatic heterogene-itywithintheCZs.
GYGA-EDisconstructedfromthreecategoricalvariablesalso usedbytheGEnS:(1)GDDwithbasetemperatureof0◦Cand(2) temperatureseasonality(quantifiedasthestandarddeviationof monthlyaveragetemperatures),and(3)anaridityindex(annual totalprecipitationdividedbyannualtotalpotential evapotranspi-ration).Gridcellsizefortheunderpinningweatherdatawasthe sameasforGLIbasedontheSAGEframework(5grid,orroughly 100km2 at the equator). BothGDD and temperature
seasona-litywerecalculatedusingclimatedatafromWorldClim(Hijmans etal.,2005);thearidityindexvaluesweretakenfromCGIAR-CSI (Trabuccoetal.,2008;Zomeretal.,2008).FollowingMuelleretal. (2012),onlyterrestrialsurfacecoveredbyatleastoneofthemajor foodcrops(maize,rice,wheat,sorghum,millet,barley,soybean, cassava,potato,yam,sweetpotato,bananaandplantain, ground-nut,commonbeanandotherpulses,sugarbeets,sugarcane)was consideredinthiszonationscheme.Toavoidinclusionofareaswith negligiblecropproduction,onlygridcellswithsumoftheharvested areaofmajorfoodcrops>0.5%ofthegridcellareawereaccounted for,based onHarvestChoiceSPAM cropdistributionmaps(You etal.,2006,2009),whichupdategeospatialcropdistributiondata ofMonfredaetal.(2008).TheresultingrangeinvaluesforGDDand aridityindexweredividedinto10intervals,eachwith10%ofgrid cellswithharvestedareaofthemajorfoodcrops,andcombined inagridmatrixwith3rangesoftemperatureseasonalitytogivea totalof300AEZclasses.Ofthese,only265occurinregionswhere majorfoodcropsaregrown.
3. Comparisonoftheagro-climaticandagro-ecological zonationschemes
Zonationschemesvarywidelyindefiningthesizeand bound-ariesofregionswithsimilarclimate(Figs.1–5).Forexample,each oftheschemesrecognizesthesignificanceoftheSaharadesert,but theydifferbyasmuchas2◦or3◦ (roughly250–350km)in loca-tionofthesouthernborderinsomeareas.Differencesamongthe zonationschemesareconsideredinthefollowingsections accord-ingtorelevanceforassessingperformanceofcropsandcropping systemswithinazone,andinthedegreeofhomogeneityofthe underpinningweathervariables.
3.1. Keyvariablesusedwithinthezonationschemes
Allglobalzonationschemesanalyzedinthepresentstudyare associatedwithtemperatureandwateravailabilitybutthey dif-ferinselection of specificweather variables todelineatezones (Table1).Forexample,toaccountforthermalconditions,GDDis calculatedwithintheSAGEandGLIschemesusingcrop-specificbase temperaturesresultinginadifferentsetofCZsforeachcropwhile GEnSandGYGA-EDuseasingle,non-crop-specificbase tempera-ture(0◦C)tocalculateGDD,whichgivesasinglesetofCZsforall crops.Creatingadifferentzonationschemeforeachcrop,however, limitsopportunitiestoanalyzeYgforcroprotationsandmuchof theworld’scroplandproducesmorethanonemajorfoodcrop.For example,crop-specificschemesmakeitdifficulttoreconcile per-formanceofcropswithinaspecificcroppingsystem(e.g.doubleor triplericeorrice-wheatcroppingsystemsinAsia).Inadditionto GDD,GEnSandGYGA-EDincludeameasureoftemperature varia-tionduringtheyearbasedontemperatureseasonality.
Differentindexeshavebeenusedtoquantifythedegreeofwater limitation.WatersupplyintheGLIzonationiscalculatedastotal annualrainfall.However,thisapproachdoesnotaccountforthe degreeofwaterlimitationtocropgrowth,whichvaries depend-ingonthebalancebetweencropwaterdemand,hereaftercalled potentialevapotranspiration,andwatersupply.Incontrast, GAEZ-LGP,HCAEZ,andSAGEtrytoaccountforbothwatersupplyand demandusingactualandpotentialevapotranspiration.Specifically, thenumberofdaysinwhichactualevapotranspirationisgreater than50%ofpotentialevapotranspirationareusedbyGAEZ-LGP andHCAEZtodeterminewhencropgrowthispossibleduetolack ofwaterstress.SAGEconsiderstheratioofactual evapotranspi-ration topotential evapotranspirationas a soil moistureindex. Estimationofactualevapotranspirationisderived fromdataon soiltexture,bulkdensity,anddepthofrootzone(whichdefines plant-available water-holding capacity),temperature, precipita-tion,andleafarea.Thesoilcomponentsofthisestimatearederived fromspatiallyexplicitglobaldatabasesandrequireanumberof assumptionsinordertocalculatehydraulicconductivity.Finally, GEnSand GYGA-ED consider anaridityindex calculated asthe ratioofannualtotalprecipitationtoannualtotalpotential evapo-transpiration.WhilenotassophisticatedastheGAEZ-LGPorSAGE schemes,thisaridityindexisderiveddirectlyfromvariablesinthe weatherdatabaseanddoesnotrequiresoildataandtheassociated uncertaintiesofassumptionsusedtoestimatesoilwaterholding capacity.
Oneofthemostinfluentialdifferencesamongzonationschemes iswhethertheydefinezonesovertotalterrestrialareaoronlythe fractionofthatareainwhichcropsaregrown.Forexample,GEnS, GAEZ-LGP,HCAEZand SAGEallconsidertotal terrestrialareain constructingtheirzonationschemes.Incontrast,GLIconsidersonly harvestedareaofindividualmajorfoodcropspeciestogive sep-aratezonationschemesforeach cropwhileGYGA-ED considers oneschemebasedonharvestedareaofallmajorfoodcrops.Asa resulttheareaoverwhichzonesaredefinedistherefore signifi-cantlyreducedforthoseAEZschemesthatonlyconsiderharvested croparea(Figs.1–5).
3.2. Climaticvariabilitywithinthezones
Climatehomogeneityforagivenzonationschemewas evalu-atedbycalculatingfrequencydistributionsoftherangeofgrid-cell values found within each zone for mean annual temperature, cumulative annual water deficit (precipitationless evapotrans-piration),temperatureseasonality,and precipitationseasonality (monthtomonthcoefficientofvariationinprecipitation)basedon WorldClimdataat5resolution(Hijmansetal.,2005).Inadditionto calculatingrangesofthesevariablesforeachzoneinagiven zona-tionscheme,rangesofmeanannualtemperatureandcumulative annualwaterdeficitwerecalculatedonlyforthosecellsinwhich wheatis grownbased onspatialcropdistributionofPortmann etal.(2010),inordertominimizebiasforthosezonationschemes thatarenotcrop-specific.ThegeospatialdistributionofPortmann etal.(2010)waschosenforuseinthisanalysisovertheSPAMor
Monfredaetal.(2008)databecausethesetwodatasetswereused inthederivationofoneormoreofthezonationschemesexamined. However,itshouldbenotedthatclimatedatausedforthisanalysis arethesameasthoseusedintheGEnS,GYGA-ED,andHCAEZ.
3.2.1. Temperaturevariability
ZonesizewaslargestinGAEZ-LGPandHCAEZ(Table2).Large zoneareawithschemesthatconsidercompleteterrestrial cover-ageresultsinawiderangeofwithin-zonetemperatureasindicated by thecumulative frequency distributionof mean annual tem-perature (Fig. 6a). For example, 50% of the GAEZ and HCAEZ zoneshavearangeofmeanannualtemperature>29◦Cand24◦C,
J.vanWartetal./FieldCropsResearch143(2013)44–55 53
Table2
AEZschemecoverageofglobal,ChinaandUSArainfedwheatandmaizebasedondatafromPortmannetal.(2010).Valuesinparenthesisindicate(±SD)ofthemean.
AEZscheme Numberofzones Averagezonearea(Mkm2) Rainfedmaizeareaper
zone(Mha)
Numberofzonestocover80%of rainfedmaizeharvestedarea
Global China USA
GAEZ-LGPa 16 20.2(18.2) 7.5(7.2) 7 6 4 HCAEZb 21 15.3(28.0) 5.8(8.2) 6 3 2 SAGEc 100 2.7(4.7) 1.2(2.1) 28 11 5 GLId 100 2.9(2.0) 1.2(0.7) 66 37 25 GEnSe 125 2.6(2.5) 1.0(1.7) 30 13 5 GYGA-EDf 265 0.3(0.3) 0.4(0.7) 49 21 9
aGlobalAgro-EcologicalZoneLengthofGrowingPeriod. bHarvestChoiceAgro-ecologicalZone.
c CenterforSustainabilityandtheGlobalEnvironment. d GlobalLandInitiative.
eGlobalEnvironmentalStratification. f GlobalYieldGapAtlasExtrapolationDomain.
respectively. In contrast, zonation schemes with smaller zone sizehaveconsiderablyless within-zonetemperaturevariability. For example,therangeof meanannualtemperaturefor 50%of theGLIandGEnSzonesis>4◦C.Whenonlycroppedterrestrial areaisevaluated(whetherforaspecificcropormultiplecrops), within-zonetemperaturevariabilitydecreasessubstantially.The clusteringmethodologyofMetzgeretal.(inpress)alsoresulted in zones with small ranges in temperature variability despite consideringtotalterrestrialareawithinzones.Apparentlythelarge numberofcategorical variablesconsideredintheGEnS cluster-ingschemeresultsinrelativelyhomogeneoustemperatureregime despitecompleteterrestrialcoverage.Whenonlywheatharvested areaisconsideredinallzonationschemes,thefrequency distribu-tionnarrowssubstantially(Fig.6b).
3.2.2. Wateravailability
Similartotemperaturevariabilitywithinzones,schemeswith thelargestzonearea(GAEZ-LGPandHCAEZ)havegreatestrange ofcumulativewaterdeficit(Fig.6c).Likewise,crop-areazonation schemes,suchas GYGA-EDand GLIhave greatesthomogeneity withinzones.Consideringonlyharvestedwheatareawithin zona-tionschemesthathavecompleteterrestrialcoveragedecreasesthe within-zonerangeofwaterdeficitofthezonalschemessomewhat, buttherangeisstillrelativelylarge(Fig.6d).
3.2.3. Temperatureandprecipitationseasonality
TheGYGA-ED, which considers three rangesof temperature seasonality as categorical variables, and the GEnS scheme, for whichtemperatureseasonalityisanexplicitinputparameter,have smallestrangeintemperatureseasonalitywithinzones.Whilethe HCAEZ,whichalsoaccountsfortemperatureseasonality,hasless heterogeneityforthisvariablethanzonationschemesthatdonot explicitlyconsiderit,itslargezonesizeresultsinagreaterrange thanforGYGA-ED.TheGAEZ-LGPhasthelargestwithin-zonerange oftemperatureseasonalitybecauseitsdelineationisbasedmoreon wateravailabilityandmanyofitszoneshaverelativelylargenorth tosouthextension,capturingawiderangeoftemperatureregimes. Rangeofprecipitationseasonalitywasalsosmallestinthe GYGA-EDschemeeventhoughthisparameterisnotexplicitlyconsidered initsderivation.
3.3. Balancingnumberofzonesandwithin-zoneclimatic
heterogeneity
Anappropriatezonationschemeforextrapolatingpoint-based estimatesofyieldpotentialwhilelimitingrequirementsfordata collectionisonewhichoptimizesthetrade-offbetweenachieving climatichomogeneitywithinzonesandminimizingthenumberof
zonesnecessarytocapturelargeportionsofharvestedareaoftarget crop.Whilezonationschemeswithfewzonesandlargezonearea, suchasGAEZ-LGPandHCAEZ,require<10zonestocover80%of globalrainfedmaizeharvestedarea(Table2),theyhavelarge vari-abilityinweathervariablesthatinfluencecropgrowthandyield (Fig.6).Amongschemeswithatleast100zonesandsmallerzone size,thoseschemesthatusetheclusteringmethodology(GEnS) orathree-parametermatrix(GYGA-ED)appeartohavethebest balancebetweennumberofzonesfor80%coverageofharvested area(Table2)andhomogeneityinweathervariableswithinzones (Fig.6).Whilethecrop-specificGLIzonationschemehasrelatively homogeneousweatherwithinitszones,itrequiresthelargest num-berofzonestoachieve80%coverageofrainfedmaizearea,andit requiresaseparatezonationschemeforeachcropspecies.In con-trast,theSAGEschemerequiresthesmallestnumberofzonesfor 80%coverageofrainfedmaizeareabuthashighdegreeof variabil-ityinweathervariableswithinitszonesdespiteuseofcrop-specific basetemperaturesusedtoderiveGDD.
4. Discussion
TheGAEZ-LGPand HCAEZschemesaresimplytoocoarsefor usein estimatingand extrapolating yieldgap analysesbecause climateheterogeneitywithinzonesistoolarge. BothSAGEand GLIschemesarecrop-specificanduseatwo-parameterzonation matrix.Ofthetwo,theGLIapproachgivesmuchgreater homo-geneity of weather variables within zones, but it requires the largestnumberofzonestocovercroparea.Bothschemesrequire separatezonation schemes foreach cropwhich would make it cumbersometoestimateYp,Yw,andyieldgapsinregionswhere morethan onecropwasgrowninrotation. BothGYGA-ED and GEnSapproachesarenotcropspecificandachieverelativelylow within-zoneheterogeneityinkeyweathervariables.WhereasGEnS requiresfewerzonestoachieve80%coverageofrainfedmaizearea andhasslightlylessheterogeneityinmeantemperature, GYGA-EDhassubstantiallylesswithin-zoneheterogeneityincumulative waterdeficitandinseasonalityoftemperatureandprecipitation. Bothmethodsappear tobewell-suitedfor up-scalingyieldgap analysis.
Severalconclusionsfollowfromthisevaluation.Climatezones usedasextrapolationdomainsforyieldgapanalysisofcurrent pro-ductionshouldfocusonareaswherecropsaregrowntominimize within-zoneweather variability.While theclustermethodology alsoappearsefficientatlimitingthenumberofzonesrequiredto covercropareaandminimizingwithin-zoneheterogeneity,they are less intuitivethanmatrix zonation schemes becauseof the sophisticatedmathematicsandlargenumberofweathervariables considered.However,formatrix-basedzonationschemesithasnot
Fig.6. Frequencydistributionofwithin-zonerangeofmeanannualtemperature,annualwaterdeficit(precipitationlessevapotranspiration),temperatureseasonalityand monthtomonthcoefficientofvariationinprecipitationbasedonWorldClimdataat5resolution(Hijmansetal.,2005)for6climatezonationschemes.Allterrestrialarea coveredbythezonesareconsidered(panelsa,c,e,f);meanannualtemperaturesandannualwaterdeficitwasalsocalculatedconsideringonlywherezonesoverlapwheat harvestedarea(bandd).Thelatterevaluationeliminatesbiasofgenericzonationschemesthatevaluateallterrestrialarea(GAEZ-LGP,GEnS,SAGE,HCAEZ)andallmajor crops(GYGA-ED).
beentestedhowtobestdeterminetherange-boundaries,whether byequaldistributions(Lickeretal.,2010),frequencydistributions (GYGA-ED),oranothersetofcriteriasuchasquantityofharvested areawithinzones(GLI).Beneficialfutureworkwouldbevalidation and comparison of zonation schemes using weather data from differentweatherstationswithinazoneorbyperformingand com-paringyieldgapanalysisforseveralsiteswithinazone.
Allzonationschemesarelimitedbychoiceandqualityofthe underpinningdatausedtoderivethem.Thisincludesavailability anddistributionofhigh-quality,locationspecificweatherstation
data.UsinganyzonationschemetoestimateYp,Ywandyieldgaps atlargerscalesalsorequiresdataonsoilsandmanagement varia-tionwithinzones(vanIttersumetal.,2013),andqualityofthose datawillalsoaffecttheaccuracyanduncertaintyinsuchlargescale estimates(vanWartetal.,2013).
Acknowledgments
TheauthorswouldliketothanktheBillandMalindaGates Foun-dationwhosesupportoftheGlobalYieldGapAtlasprojectmade
J.vanWartetal./FieldCropsResearch143(2013)44–55 55
thisresearchpossible.Secondly,wethankDr.MarcJ.Metzgerfrom theUniversityofEdinburghforprovidinguswiththedataonGEnS.
References
Araya,A.,Keesstra,S.D.,Stroosnijder,L.,2010.Anewagro-climaticclassificationfor cropsuitabilityzoninginnorthernsemi-aridEthiopia.Agric.ForestMeteorol. 150,1057–1064.
Batjes,N.H.,2006.ISRIC-WISEDerivedSoilPropertiesona5by5Arc-minutesGlobal Grid.ISRIC–WorldSoilInformation,Wageningen.
Caldiz,D.O.,Haverkort,A.J.,Struik,P.C.,2002.Analysisofacomplexcropproduction systemininterdependentagro-ecologicalzones:amethodologicalapproachfor potatoesinArgentina.Agric.Syst.73,297–311.
EuropeanEnvironmentalAgency(EEA),2007.EstimatingtheEnvironmentally Com-patibleBioenergyPotentialfromAgriculture.EEATechnicalReport12.European EnvironmentalAgency,Copenhagen.
FoodandAgriculturalOrganization(FAO),1978.Reportontheagro-ecologicalzones project.FAO,Rome.
Fischer,G.,2009.Worldfoodandagricultureto2030/50:howdoclimatechangeand bioenergyalterthelongtermoutlookforfood,agricultureandresource avail-ability?In:ProceedingsoftheExpertMeetingonHowtoFeedtheWorldin 2050,FAORome,Italy.June24–262009.
Fischer,G.,Hizsnyik,E.,Prieler,S.,Shah,M.,VanVelthuizen,H.,2009.Biofuelsand FoodSecurity.OFID/IIASA,Vienna,p.40.
Fischer,G.,Nachtergaele,F.O.,Prieler,S.,Teixeira,E.,Tóth,G.,VanVelthuizen,H., Verelst,L.,Wiberg,D.,2012.GlobalAgro-EcologicalZones–Model Documenta-tionGAEZv.3.0.IIASA/FAO,Laxenburg,Austria/Rome,Italy,p.179.
Fischer,G.,Schrattenholzer,L.,2001.Globalbioenergypotentialsthrough2050. BiomassBioenerg.20,151–159.
Fischer,G.,Shah,M.,Tubiello,F.N.,VanVelhuizen,H.,2005.Socio-economicand climatechangeimpactsonagriculture:anintegratedassessment,1990–2080. Phil.Trans.R.Soc.Lond.B360,2067–2083.
Fischer,G.,Shah,M.M.,VanVelthuizen,H.,Nachtergaele,F.O.,2006. Agroecologi-calzonesassessment.In:Verheye,W.H.(Ed.),LandUse,LandCoverandSoil Sciences.EOLSSPublishers,Oxford.
Fischer,G.,VanVelthuizen,H.,Shah,M.,Nachtergaele,F.O.,2002.Global agro-ecologicalassessmentforagricultureinthe21stcentury:methodologyand results.ResearchReportRR-02-02.InternationalInstituteforAppliedSystems Analysis,Laxenburg,Austria,pp.xxii,119pp.andCD-Rom.
Foley,J.A.,DeFries,R.,Asner,G.P.,Barford,C.,Bonan,G.,Carpenter,S.R.,Chapin,F.S., Coe,M.T.,Daily,G.C.,Gibbs,H.K.,Helkowski,J.H.,Holloway,T.,Howard,E.A., Kucharik,C.J.,Monfreda,C.,Patz,J.A.,Prentice,I.C.,Ramankutty,N.,Snyder,P.K., 2005.Globalconsequencesoflanduse.Science309,570–574.
Foley,J.A.,Ramankutty,N.,Brauman,K.A.,Cassidy,E.S.,Gerber,J.S.,Johnston,M., Mueller,N.D.,O’Connell,C.,Ray,D.K.,West,P.C.,Balzer,C.,Bennett,E.M., Car-penter,S.R.,Hill,J.,Monfreda,C.,Polasky,S.,Rockstrom,J.,Sheehan,J.,Siebert, S.,Tilman,D.,Zaks,D.P.M.,2011.Solutionsforacultivatedplanet.Nature478, 337–342.
Gallup,J.L.,Sachs,J.D.,2000.Agriculture,climate,andtechnology:whyarethe trop-icsfallingbehind?Am.J.Agric.Econ.82,731–737.
Geerts,S.,Raes,D.,Garcia,M.,DelCastillo,C.,Buytaert,W.,2006.Agro-climatic suit-abilitymappingforcropproductionintheBolivianAltiplano:acasestudyfor quinoa.Agric.ForestMeteorol.139,399–412.
Godfray,H.C.J.,Beddington,J.R.,Crute,I.R.,Haddad,L.,Lawrence,D.,Muir,J.F.,Pretty, J.,Robinson,S.,Thomas,S.M.,Toulmin,C.,2010.Foodsecurity:thechallengeof feeding9billionpeople.Science327,812–818.
Hazeu,G.,Elbersen,B.,Andersen,E.,Baruth,B.,VanDiepen,C.A.,Metzger,M.J., 2009.Abiophysicaltypologyforaspatially-explicitagri-environmental mod-elingframework.In:Brouwer,F.,VanIttersum,M.K.(Eds.),Environmentaland AgriculturalModelling:IntegratedApproachesforPolicyImpactAssessment. SpringerAcademicPublishing,NewYork/London/Heidelberg/Dordrecht. Hazeu,G.W.,Metzger,M.J.,Mücher,C.A.,Perez-Soba,M.,Renetzeder,C.,Andersen,
E.,2011.Europeanenvironmentalstratificationsandtypologies:anoverview. Agric.Ecosyst.Environ.142,29–39.
Hijmans,R.J.,Cameron,S.E.,Parra,J.L.,Jones,P.G.,Jarvis,A.,2005.Veryhigh res-olutioninterpolatedclimatesurfacesforgloballandareas.Int.J.Climatol.25, 1965–1978.
Holdridge,L.R.,1947.Determinationofworldplantformationsfromsimpleclimatic data.Science105,367–368.
Johnston,M.,Licker,R.,Foley,J.,Holloway,T.,Mueller,N.D.,Barford,C.,Kucharik,C., 2011.Closingthegap:globalpotentialforincreasingbiofuelproductionthrough agriculturalintensification.Environ.Res.Lett.6,1–11.
Köppen,W.,1900.VersucheinerKlassifikationderKlimate,vorzugsweisenachihren BeziehungenzurPflanzenwelt.GeographieZeitschrift6,657–679,593–611. Kottek,M.,Grieser,J.,Beck,C.,Rudolf,B.,Rubel,F.,2006.WorldMapofthe
Köppen-Geigerclimateclassificationupdated.MeteorologischeZeitschrift15,259–263. Licker, R., Johnston, M., Foley, J.A., Barford, C., Kucharik, C.J., Monfreda, C., Ramankutty, N., 2010. Mind the gap: how do climate and agricultural managementexplainthe‘yieldgap’ofcroplandsaroundtheworld?GlobalEcol. Biogeogr.19,769–782.
Lobell,D.B.,Cassman,K.G.,Field,C.B.,2009.Cropyieldgaps:theirimportance, mag-nitudes,andcauses.Annu.Rev.Environ.Resour.34,179–204.
Metzger,M.J.,Bunce,R.G.H.,Jongman,R.H.G.,Mücher,C.A.,Watkins,J.W.,2005.A climaticstratificationoftheenvironmentofEurope.GlobalEcol.Biogeogr.14, 549–563.
Metzger,M.J.,Bunce,R.G.H.,Leemans,R.,Viner,D.,2008.Projectedenvironmental shiftsunderclimatechange:Europeantrendsandregionalimpacts.Environ. Conserv.35,64–75.
Metzger M.J., Bunce R.G.H, Jongman R.H.G, Sayre R., Trabucco A., Zomer R. A high-resolution bioclimate map of the world: a unifying framework for global biodiversity research and monitoring. Global Ecol. Biogeogr. http://dx.doi.org/10.1111/geb.12022,inpress.
Monfreda,C.,Ramankutty,N.,Foley,J.A.,2008.Farmingtheplanet:2.geographic dis-tributionofcropareas,yields,physiologicaltypes,andnetprimaryproduction intheyear2000.GlobalBiogeochem.Cy.22,1–19.
Mueller,N.D.,Gerber,J.S.,Johnston,M.,Ray,D.K.,Ramankutty,N.,Foley,J.A.,2012. Closingyieldgaps:nutrientandwatermanagementtoboostcropproduction. Nature490,254–257.
New,M.,Lister,D.,Hulme,M.,Makin,I.,2002.Ahigh-resolutiondatasetofsurface climateovergloballandareas.ClimateRes.21,1–25.
Padbury,G.,Waltman,S.,Caprio,J.,Coen,G.,McGinn,S.,Mortensen,D.,Nielsen, G.,Sinclair,R.,2002.AgroecosystemsandlandresourcesofthenorthernGreat Plains.Agron.J.94,251–261.
Papadakis,J.,1966.ClimatesoftheWorldandtheirAgriculturalPotentialities. DAPCO,Rome.
Portmann,F.T.,Siebert,S.,Doll,P.,2010.MIRCA2000-Globalmonthlyirrigatedand rainfedcropareasaroundtheyear2000:anewhigh-resolutiondatasetfor agriculturalandhydrologicalmodeling.GlobalBiogeochem.Cy.24,GB1011. Prentice,I.C.,Cramer,W.,Harrison,S.P.,Leemans,R.,Monserud,R.A.,Solomon,A.M.,
1992.Specialpaper:aglobalbiomemodelbasedonplantphysiologyand dom-inance,soilpropertiesandclimate.J.Biogeogr.19,117–134.
Ramankutty,N.,Foley,J.A.,Norman,J.,McSweeney,K.,2002.Theglobaldistribution ofcultivablelands:currentpatternsandsensitivitytopossibleclimatechange. GlobalEcol.Biogeogr.11,377–392.
Ramirez-Villegas,J.,Challinor,A.,2012.Assessingrelevantclimatedatafor agricul-turalapplications.Agric.ForestMeteorol.161,26–45.
Rudolf,B.,Beck,C.,Grieser,J.,Schneider,U.,2005.Globalprecipitationanalysis prod-ucts.GlobalPrecipitationClimatologyCentre(GPCC),DWD,Internetpublication, pp.1–8.
Scholes,R.J.,Mace,G.M.,Turner, W.,Geller,G.N.,Jürgens,N.,Larigauderie,A., Muchoney,D.,Walther,B.A.,Mooney,H.A.,2008.Towardaglobalbiodiversity observingsystem.Science321,1044–1045.
Seppelt,R.,2000.Regionalisedoptimumcontrolproblemsforagroecosystem man-agement.Ecol.Model.131,121–132.
Sivakumar,M.V.K.,Valentin,C.,1997.Agroecologicalzonesandtheassessmentof cropproductionpotential.Phil.Trans.R.Soc.Lond.B352,907–916.
TAC/CGIAR(TechnicalAdvisoryCommittee)ConsultativeGrouponInternational AgriculturalResearch,1992.ReviewofCGIARprioritiesandstrategies.PartI. WashingtonDC:CGIAR.
Tilman,D.,Cassman,K.G.,Matson,P.A.,Naylor,R.,Polasky,S.,2002.Agricultural sustainabilityandintensiveproductionpractices.Nature418,671–677. Tou,J.T.,González,R.C.,1974.Patternrecognitionprinciples.Addison-Wesley
Pub-lishingCo.Reading,MA.
Trabucco,A.,Zomer,R.J.,Bossio,D.A.,vanStraaten,O.,Verchot,L.V.,2008. Cli-matechangemitigationthroughafforestation/reforestation:aglobalanalysis ofhydrologicimpactswithfourcasestudies.Agric.Ecosyst. Environ.126, 81–97.
vanIttersum,M.K.,Cassman,K.G.,Grassini,P.,Wolf,J.,Tittonell,P.,Hochman,Z., 2013.Yieldgapanalysiswithlocaltoglobalrelevance–areview.FieldCropRes. 143,4–17.
vanWart,J.,Kersebaum,K.C.,Peng,S.,Milner,M.,Cassman,K.G.,2013.Aprotocol forestimatingcropyieldpotentialatregionaltonationalscales.FieldCropsRes. 143,34–43.
Williams,C.L.,Liebman,M.,Edwards,J.W.,James,D.E.,Singer,J.W.,Arritt,R., Herz-mann,D.,2008.Patternsofregionalyieldstabilityinassociationwithregional environmentalcharacteristics.CropSci.48,1545–1559.
Wood,S.,Pardey,P.G.,1998.AgroecologicalaspectsofevaluatingagriculturalRand D.Agric.Syst.57,13–41.
Wood,S.,Sebastian,K.L.,Scherr,S.,2000.Pilotanalysisofglobalecosystems: agroe-cosystems.In:JointStudybytheInternationalFoodPolicyResearchInstitute andtheWorldResourcesInstitute.WorldResourcesInstitute,Washington,DC. Wood,S.,Sebastian,K.L.,You,L.,2010.Spatialperspectives.In:Pardey,P.G.,Wood, S.,Hertfor,R.(Eds.),ResearchFutures:ProjectingAgriculturalR&DPotentialsfor LatinAmericaandtheCaribbean.InternationalFoodPolicyResearchInstitute, Washington,DC.
You,L.,Wood,S.,Wood-Sichra,U.,2006.Generatingglobalcropmaps:from cen-sustogrid.In:Selectedpaper,IAAE(InternationalAssociationofAgricultural Economists)AnnualConference,GoldCoast,Australia.
You,L.,Crespo,S.,Guo,Z.,Koo,J.,Sebastian,K.,Tenorio,M.T.,Wood,S., Wood-Sichra,U.2009.SpatialProductionAllocationModel(SPAM)2000Version3 Release6.
Zomer,R.J.,Trabucco,A.,Bossio, D.A.,Verchot,L.V.,2008.Climatechange mit-igation: aspatialanalysisofgloballandsuitabilityfor cleandevelopment mechanism afforestation and reforestation. Agric. Ecosyst. Environ. 126, 67–80.