The role of rainfed agriculture in securing food production in the Nile Basin

The role of rainfed agriculture in securing food production in the Nile Basin

C. Siderius

*, P.E.V. Van Walsum

, C.W.J. Roest

, A.A.M.F.R. Smit

, P.J.G.J. Hellegers

, P. Kabat

, E.C. Van Ierland

aAlterra,WageningenUniversityandResearchCentre(WUR),POBox47,6700AAWageningen,TheNetherlands

bEnvironmentalEconomicsandNaturalResourcesGroup,WageningenUniversityandResearchCentre(WUR),POBox47,6700AAWageningen,The Netherlands

cGranthamResearchInstituteonClimateChangeandtheEnvironment,LondonSchoolofEconomicsandPoliticalScience,WC2A2AE,London,UK

dInternationalInstituteforAppliedSystemsAnalysis(IIASA),Schlossplatz1,A-2361Laxenburg,Austria

ARTICLE INFO Articlehistory:

Received10September2015

Receivedinrevisedform15March2016 Accepted18March2016

Availableonline31March2016 Keywords:

Nilebasin Waterdispute Rainfedagriculture Foodsecurity Waterallocations

ABSTRACT

Abetteruseoflandandwaterresourceswillbenecessarytomeettheincreasingdemandforfoodinthe Nilebasin.Usingahydro-economicmodelalong thestorylineofthree futurepoliticalcooperation scenarios,weshowthatthefutureoffoodproductionintheBasinliesnotintheexpansionofintensively irrigatedareasandthedisputedreallocationofwater,butinutilizingthevastforgottenpotentialof rainfedagricultureintheupstreaminterior,withsupplementalirrigationwhereneeded.Ourresults indicatethatrainfedagriculturecancovermorethan75%oftheneededincreaseinfoodproductionby theyear2025.ManyofthemostsuitableregionsforrainfedagricultureintheNilebasin,however,have beendestabilizedbyrecentwarandcivilunrest.Stabilizingthoseregionsandstrengtheningintra-basin cooperationviafoodtradeseemtobebetterstrategiesthanunilateralexpansionofupstreamirrigation, asthelatterwillreducehydropowergenerationandrelocate,ratherthanincrease,foodproduction.

ã2016ElsevierLtd.Allrightsreserved.

1.Introduction

Major socioeconomic and geopolitical transformations are affecting the allocation of one of the world’s most disputed resources:thewateroftheNileRiver.Atpresent,mostwaterinthe LowerNileisbeingutilized,mainlyforirrigationbydownstream Egypt.Attempts to convert existing water allocation, primarily basedonthe1959treaty betweenEgyptand Sudan,toa more equitableshareforallcountrieshavenotbeensuccessful(Nicol and Cascão, 2011). The regional balance of power is, however, changing: (i) the political upheaval after the Arab spring has weakenedthedominanceofEgypt(NicolandCascão,2011);(ii)in anincreasinglymulti-polarworld,accesstoinfrastructureloansto builddamsandirrigationinfrastructureupstreamhasdiversified (Broadman,2008;Fosteretal.,2009);and(iii)foreigninvestors havetakenarenewedinterestinthebasin’sagriculturalresources, buyingandleasingagriculturallandalloverthebasin(Cotulaetal., 2009; von Braun and Meinzen-Dick, 2009). Amid these

transformations,reallocationofNilewaterisahotissue(Cascão, 2009; Waterbury, 2002; Whittington et al., 2005), with many countriesseekingtoutilizemorewaterforhydropowerandfood production.

Increasedfoodavailabilityinthebasinisurgent.Accordingto the2012reportoftheUnitedNations,“TheStateofFoodInsecurity intheWorld”(FAOetal.,2012),100millionpeopleinthecountries ofthebasinareundernourished,whichamountstoalmostathird of the local population. Undernourishment has increased in northernand sub-Saharan Africaoverthepast decade,bucking the world-wide trend. Except for Egypt, none of the 11 Basin countriesareself-sufficientinfood(Omitietal.,2011).Withinthe context of high and volatile commodity prices that favour net producers over buyers (Breisinger et al., 2010; Swinnen and Squicciarini,2012),thisrelianceonglobalmarketsisadangerous gamble:recent political instabilityin the Nile regionhas been directlylinkedtofoodpricehikes(ArezkiandBruckner,2011),and theseriskswillonlyincrease.ThepopulationoftheBasincountries is expected to grow by a third, from 367 million in 2012 to 488millionin2025(UNDP,2011).Atthesametime,world-wide competition for land, water, energy, and, ultimately, food is increasing(Godfrayetal.,2010).Developingcountrieslikethosein theNile,withpurchasingpowersmuchlowerthanthatofother

*Correspondingauthorat:Alterra,WageningenUniversityandResearchCentre (WUR),POBox47,6700AAWageningen,TheNetherlands.

E-mailaddress:christian.siderius@wur.nl(C. Siderius).

http://dx.doi.org/10.1016/j.envsci.2016.03.007 1462-9011/ã2016ElsevierLtd.Allrightsreserved.

ContentslistsavailableatScienceDirect

Environmental Science & Policy

j o u r n a lh o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / e n vs c i

majorfood importing countries, are most vulnerable toglobal shortages(Ruttenetal.,2013).

WeaimtosupportthecomplexpolicychallengeoftheNile basin by clarifying the science behind the discourse onwater, energyandfoodsecurity,exploringthepossibilityofnationalto regional food self-sufficiency as alternatives to an increasing reliance on global markets. We approach this from a hydro- economicperspectiveandarguethatwiththewaterresourcesof theNile itself almost fullyand productivelyallocated, the real solutiontofuturefoodself-sufficiencyfortheBasinliesoutsidethe domainofwaterallocation and irrigatedagricultureand in the rainfed areas of South Sudan and the Lake Victoria region.

AccordingtotheUnitedNationsFoodandAgricultureOrganization (FAO),thepotentialareasuitableforcultivation inSouthSudan alone isas highas30 million hectares,which is tentimes the croppedareaofEgypt.Onlyabout10%ofthatpotentialiscurrently beingusedforagriculture.Recentworld-wideassessmentsoffood productionhavestressedintensificationinexistingareas,rather thanexpansiontonewareas,asthebestwayofincreasingfood production(Foleyetal.,2011;Godfrayetal.,2010;Tilmanetal., 2011).TheNilebasinseemstobeanimportantexception,witha combination of both intensification and expansion being war- ranted.

2.Methods 2.1.Approach

Forourresearch,wederivedabaselineofwateruse(Fig.2), agriculturalcropproductionand grossmargin(GM)intheNile basinaroundtheyear2005,usinganarea-basedhydro-economic modelinsimulationmode(WaterWise(Sideriusetal.,2016);see Section2.2andSI1).Forthis,apresent–dayspatialdistributionof landusesystems(FAO,2009)wasmadeconsistentwithcountry- specificFAO cropstatistics (FAO,2004)onactual croppedarea (SI2).Cropproductionandagriculturalgrossmargin(GM)ofthe water-limitedproductionwasthencalculatedforbothrainfedand irrigatedcrops.

Next,weestimatedfoodrequirementsinthebasinfortheyear 2025.Future foodself-sufficiencycorrectionfactorspercountry were based on the projected population increase up to 2025 (UNDP, 2011) and a population-average calorie requirementof 2300kcal/personperday(TontisirinanddeHaen,2001).Assuch,a minimum intake was imposed, without regard for household access,dietarypreferences,ornutritionalvalue.Weassumedthat agriculturalproductionintheNilecatchmentpartofeachcountry willgrowatthesamepaceaseachcountry’saverageandthatthe proportionoffoodcropstocashcropsremainsthesame.Future food self-sufficiency targets for the Nile basin could then be derivedbymultiplyingbaselineagriculturalproductionwiththese correctionfactors(Table2).

Finally,weappliedthehydro-economicmodelinoptimization mode, to select those investments in agriculture (area-wise expansion or intensification of rainfed agriculture and new irrigationschemes)andhydropower(newreservoirs)thatgener- atethehighestGMusingtheavailablelandandwaterresources.

Weexploredwhereandhowfoodproductioncanbestbeincreased andwhetherfoodself-sufficiencyforthebasinanditsindividual countriescanbeachievedbytheyear2025.

2.2.WaterWisemodel

Ourmodelresemblesexistinghydro-economicmodelsdevel- opedfor theNile(BlockandStrzepek,2010;Blocket al.,2007;

Jeuland, 2010; Whittington et al., 2005; Wu and Whittington,

2006). Similarly to the model of Whittington et al. (2005) it describes thewholeNile basin, including allexisting irrigation schemes and hydropowerreservoirs,and mostoftheproposed hydropower plans. Water gets transmitted through the river networkusingaroutingschemeincombinationwiththevariable storagemethodforthedynamicsoflargewaterbodies(swamps, reservoirs),withuseinonelocationlimitingoptionselsewhere.

Economic parameters, like the pricing of hydropower, are like thoseinearlieroptimizationstudies.However,incontrasttothe latterwedidnotlimitouranalysistotheriversystemalone,i.e.

optimizing hydropowerand irrigationyields,but includedyield fromrainfedlanduse.Landuseisanendogenousvariableinour modelandland-usechangesandtheimpactondownstreamflows are thereby integrated into the optimization. The general idea behindthemodelisthatitshouldbecapableofexploringawide rangeoflandandwatermanagementoptions,forvariousscenarios withrespecttobasincooperation.Suchanexploratoryfunction- alitynecessitatesarelativelysimplemodelformulationforboth hydrologyandagronomy.Itshouldthenberealizedthatthemodel resultsarejustindicativeofasearchdirection.Furtherstudiesare neededformoreaccurateassessments.

ThemodeloptimizesGMbychoosingtheoptimalcombination of land and water use options for each of 1371 so-called hydrotopes,unitsofsimilarsoilandmeteorologicalcharacteristics, givenavailablewaterresources:

YTOT¼YLUþYHPCLWM

with

YLU¼

S

CLWM¼

S

whereYTOTrepresentstotalgrossmargin(inUSD/yr),YLUtheprofit fromlanduse(USD/yr)basedonproduction(Prod,inton)times priceofproduct(P,USD/ton)minusnon-watercosts(CLU,USD/ha) times thecroppedarea(Ac,in ha), in yearyperland useu in hydrotopez.YHPistheGMofhydropower(USD/yr).CLWMarethe costsof local water-managementmeasuresfor supportingland use,i.e.,thevariablecostsoflocalirrigationmeasures(inUSD/ha), dependingontheamountofwaterused.Variablecostsofwater relatetopumpingcosts,whichisacombinationoflabor,capital andenergycosts.Forthevariablecostsofwaterweusedaregional estimateof0.01USD/m³(HellegersandPerry,2006).

Cropproductionandrelatedwaterfluxesforalllandandwater useoptionsin eachhydrotopeare pre-processedbywater-crop modulesruninanofflinemode(SI2).IntheNileapplicationasoil moistureaccountingmodelofthebuckettypeisused,verysimilar totheAQUACROPmodeloftheFAO(Raesetal.,2011),butmore advancedinsimulatingsoilstorageanddrainage,whilesimplify- ing thedynamiccropgrowth. Rainfallcan contributetorunoff, drainage,orgroundwaterstorage,aftercorrectingforevapotrans- piration. The calculation scheme for the evapotranspiration follows the FAO single crop coefficient method (Allen et al., 1998),appliedseparatelytothevegetatedandnon-vegetatedpart.

Cropproductionissimulatedwithaslightlymodifiedformofthe K_yapproachofFAO(DoorenbosandKassam,1979),wheretheratio betweenactualandpotentialevapotranspirationistranslatedinto ameanyieldratio.Actualyieldineachhydrotopeisthencalculated bymultiplyingthismeanyieldratiowithapredefinedpotential yield.Thisrelativelysimplemethodhastheadvantageofbeing robustandrequiringaminimumofdata.

WaterWiseoptimizesGMoffoodproductionbyi.converting non-arablelandintoarableland,byii.convertingexistingarable landintohigh-intensivevariantsand/oriii.byincreasingthearea

under irrigation in predefined existing and new large-scale irrigation areas, depending on irrigation water availability and availabilityofinvestments.GMfromhydropowercanbeincreased byroutingmorewaterthroughexistinghydropowerschemes,if turbinecapacityallows,orbyinvestinginnewones.

Investmentcostsfortheconversiontoirrigatedareawerebased on a comprehensive study on the cost of irrigation by IFPRI (Inocencioetal.,2007).Wetookthevaluefor‘success’projects, undertheoptimisticassumptionthatnewirrigationsystemswill bedesigned,constructedandmaintainedaccordingtothelatest knowledge and standards. Thereis a clear difference between north Africa and sub-Saharan Africa: the latter having, at 3552USD/ha,onlyabouthalftheconversioncostsasNorthAfrica.

Conversiontoarablelandwasmadepossibleataninvestmentof 2174USD/ha,assumingthatconversiontorainfedarable landis similar to land preparation for irrigation, but without the additionalhardwarecosts.Investmentscostsinnewhydropower weremainlybasedongreyliterature(seeSI2).Allmajorplanned hydropower plants, including Ethiopia’s highly controversial GrandRenaissanceDam,wereofferedasoptions.

The optimization was performed on the basis of two representativeclimateyears—arelativelywetyear(1999)followed byadryyear(2000).Wedidnotexplicitlyincludewaterdemand fromothersectorslikehouseholdandindustry,beingrelatively smallcomparedtoagriculturaldemand,northeeconomicbenefits of flood or sediment control, or environmental flows. Climate changewas left out from the analysis. Within the time-frame considered, we expect that any climate change trend will be overshadowedby existing natural variability. However, rainfall projectionsforEastAfricadoshowalargespreadbetweenclimate models for the periods beyond 2025, adding considerable uncertaintytoanylong-terminvestmentdecision.

2.3.Dataandschematization

Rainfall from the tropical rainfall measurement mission (TRMM)(Kummerowetal.,1998)anddailyreferenceevapotrans- piration from ECMWF (Uppala et al., 2005) were used as meteorological inputs, with soil properties coming from FAO- UNESCO’s1974SoilMap oftheWorld(1:5,000,000),soilclasses aggregatedbasedonmaximumsoilmoisturestorageandsurface slope.Apresent-dayspatialdistributionoflandusesystems(FAO, 2009)at5arcminutesspatialresolutionwasmadeconsistentwith country-specificFoodandAgricultureOrganizationcropstatistics (FAOSTAT)(FAO,2004)onactualcroppedareabycorrectingfor fallow area. Estimations of arable land were only available at nationallevel.Simplycorrectingbasedonlandareawouldleadto anunderestimationofarablelandwithintheNilebasin,theBasin partbeingwetter,ingeneral.ANilebasinestimatewasderivedby multiplyingthenationalaveragewiththerelativeproportionof humid zone within the Basin area, as proposed by the FAO (Appelgrenetal.,2000).Amoredetailedmappingoftheirrigated areaswasachievedbyasupervisedclassificationofLandsatimages incombinationwithaFAOmapindicatingregionswithacertain percentage of irrigation (Occurrenceof irrigated areas (FGGD);

(FAO,2007;Siebertetal.,2005)).Trainingsites(areasinthemap thatareknowntoberepresentative)fortheirrigatedareaclass weredeterminedbasedonpriorknowledgeofthelocationofsome ofthemajorirrigatedareasinEgyptandSudan.Theirrigatedarea wasthenbasedontheclassificationofthehigh-resolution(30m) Landsat images, but only in regions where irrigation hasbeen reportedaccordingtothelower-resolution(10m)FAOirrigated areamap.ThemajorirrigatedareasintheNiledelta,intheNile valley and in Sudan could be well identified, with their area matchingtheareaasreportedinFAOSTAT(Fig.1).

Forarablelandineachcountrywedefinedauniquecountry- specificcroppingsystem,CCSs,representingarangeofcrops.Only cropsthatoccupyeachatleast10%ofthearableareainatleastfive countriesaccordingtoFAOSTAT(FAO,2004)wereincluded.This resultedin sevenmain crops: bananas,beans, maize, sorghum, sweetpotatoes,vegetablesandwheat.Becauseoftheimportance of groundnuts for Sudanese agriculture and rice for Egyptian agriculturethese two cropswere added.For each country, five dominantcropswereselectedfromthissubsetandbasedonthese fivecropsanaveragepricepertonproducedandcostperhawere derivedforeachcountry-specificcroppingsystem.Thisresultedin atotalofsevenrainfedandtwoirrigatedCCSsforthebasinasa whole(Table1).Cropgrowthperiodsandmonthlycropfactors,to multiplythedailyreferenceevaporationwith,werederivedfrom Allenet al.(1998).Exceptfor thesorghum croppingpatternin Sudanweassumedadoublecroprotationineachcountry.

A uniform region-specific potential yield of 4ton/ha was derived by correlating country-specific crop yields on rainfed arableland(inton/ha,fromFAOSTAT,2004)withtheETa/ETpratio of each country (AQUASTAT) (R²=0.7). While this is a gross simplificationofthediversityincropproduction,apotentialyield of 4 ton/ha does correspond well with earlier estimates (e.g.

PenningdeVriesetal.,1997).Byusingaregion-specificpotential yield,limiting factorsotherthanwater,for example,phosphate shortages,pests,orNileregion-specificrestrictionsintheagro- food chain infrastructure, are implicitly taken into account.

Economicparametersintermsofcroppricesandcostsperhectare dodifferpercountry.AveragecostsandpricesforeachCCSwere calculated using area averaging of the FAOSTAT data, thereby ignoringanypricefluctuationsoruncertainty.

Largescaleirrigationwasseparatelyschematizedandparame- terized. This type of irrigation in the Nile Basin is currently concentratedinEgyptandSudan.EspeciallyinSudanandEthiopia thereisthelandpotentialtoincreasetheareairrigated(Blockand Strzepek,2010;Blocketal.,2007).Irrigationfromthemainwater courses was only allowed in predefined large-scale irrigation schemes,currentlylocatedinEgyptandSudan. Yield,priceand cost data per hectarefor Egyptcould bederived directly from FAOSTATdata(FAO,2004),butforSudanthesewereavailableonly as anaverage of irrigatedand rainfed areascombined. Sudan’s irrigated agriculture is known tounderperform because of the siltationofirrigationcanals,waterlogging,andgeneraldeteriora- tion of operationand maintenance (Plusquellec,1990).Sudan’s

- 2 4 6 8 10 12

- 2 4 6 8 10 12

WW arable area (in million ha)

FAOSTAT arable area (in million ha) Uganda

Egypt (irrigated) Sudan (irrigated) Rwanda

Burundi

Sudan

Ethiopia

Kenya Tanzania

Fig.1.WW-NilearablelandvsFAOSTATarablelandestimates(withirrigatedarea fromAQUASTAT).‘Sudan’includesbothSudan andSouthSudan,butexcludes irrigatedareainSudan.

yieldperhectarewasassumedtobehalfthatofEgypt,butwiththe samecosts,andcroppingintensityatonlyhalfofitspotential.With regard to new irrigation schemes in Sudan and Ethiopia, we assumethatinvestors,watermanagersand irrigationengineers havelearnedfrompastmistakesandthatproductivitywillmatch thatofirrigatedagricultureinEgypt.

2.4.Scenarios

We evaluatedthetargetof foodself-sufficiencyunder three transformative scenarios with varying degrees of cooperation, whicharecurrentlyunderdebate.Ahydro-economicmodellike WaterWisesearches, ifunrestricted,for abasin-wide optimum, thus reflecting complete cooperation and sharing of GM. This Fig.2. ComparisonofWaterWise-Nilerunoffcontributionandabstractionwithmodelledrunoff(MWRI,2005)andrunoffderivedfromwaterbalanceestimates(Sutcliffe andParks,1999)forthemainwaterbalanceareasoftheNile(inbillionm³/yr).ThewaterdemandoftheDeltaandValleywasnotavailableforthelattertwostudiesand thereforeomitted.NofiguresfortheWhiteNileareavailablefromSutcliffe&Parksbecauseofthedifferentcatchmentschematization.WaterWise-Nilewasvalidatedonthe wet,average,anddryyearsof1999–2001;SutcliffeandParkshavedeterminedtherunoffbasedonmeasurementdataoftheperiod1905–1995.TheperiodoftheMWRIstudy represents1991–2001.Waterabstractionsof70billionM³toEgyptsupportunofficialestimates,suggestingthatactualreleasesatAswanarehigherfortheperiodevaluated thanthe,oftenreported,officiallyallocated55.5billionM³(NicolandCascão,2011)evenaftercorrectionforreturnflows.

cooperationcanthenberestrictedbyspecificboundaryconditions orobjectivetargets.Withthemodelwefocusontheallocationof landand waterresources.Thethird productionfactor,labour,is assumedtobeavailableandwasnottakenintoaccount.

Thebackgroundofthe“NationalFoodSelf-Sufficiency”scenario isafuturewherecooperationandtradeofagriculturalproduceis limited and food self-sufficiency is a target of each country individually; GM will drop once supply exceeds demand since transactioncostswillincrease once productshaveto betrans- portedtoothermarkets.Tomimicthisbehaviourtotheextremein themodel,theweightoflanduserevenuesaboveacountry’starget is reduced to nil in the objective function. In the “Upstream Hegemony”scenario,EthiopiaandSudanmaximizetheiragricul- turalGMforinternationalexport,irrespectiveofanydownstream demands. All newirrigation schemes and the rehabilitation of existingirrigationschemesinEthiopiaandSudan areforcefully implementedin themodel at the investmentcost required. In addition, the model maximizes agricultural GM of the major irrigation schemes in these two countries via the objective function. The “Basin Cooperation” scenario represents a future ofenhanced tradeinagriculturalcommoditieswithinthebasin, underpinned by infrastructural developments and political, economic, and financial cooperation. In the model this is implementedbysolvingtheobjectivefunctionforthebasinasa whole,givingtotalfreedomtomaximizelandusethroughoutthe basintoreachthefoodself-sufficiencytargetfor thebasin asa whole.Onecountrycanoffsetshortagesinanother.

Our model includesboth expansion of agriculturalareaand intensification with higher profits and costs per hectare, with investments in agriculture competing with investments in hydropower.Thedifferencebetweenexpansionandintensification inthemodelneedstobeinterpretedwithcare.Especiallysmall- scaleagricultureislikelytobeclusteredwithnon-agriculturalland usesinthepresentdaylanduseclassification.Inaddition,inwar- tornregions,manyfieldshavebeentemporarilyabandonedorleft fallow. In theseareas, ‘expansion’ will refer more to a leap in production fromlow-yield agriculturetoa form ofcommercial agriculture connected to regional markets, rather than an agriculturaldevelopmentfromscratch.

Livestock was not explicitly included in ouranalysis, as we focusedonarablefarming,whichhasafarlargerclaimonlandand waterresources.We assumed livestockraising tobeintegrated witharablefarminginmixedagriculturalsystems,withoutexplicit additionallandandwaterdemands.AnexceptiontothisintheNile BasincouldbethelargegrazingareasinSudanandSouthSudan.

Conversionoftheseexistingpastorallandstoarablelandswasnot restrictedinthemodel.However,ingeneral,themodeldidnot select these areas for arable expansion. The mere existence of pastoral lands can, in itself, be an indication that biophysical circumstancesmakesuchlandslesssuitableforarablefarmingfor example,becauseoferraticorstrongseasonalityinrainfall.

Wefocusedonthenearfuture,inwhichwe assumegradual autonomoustechnologicalprogressinrainfedfarmingpracticesin those countries currently producing at a GM level below the Table1

CroppingsystemcharacteristicsforthedominantcropsasderivedfromFAOSTAT(current)andestimatesforfuture.

CurrentYield Potentialyield Price Cost PotentialGM Dominantcropsinthe Doublecropping

ton/ha ton/ha US$/ton US$/ha US$/ha croppingsystem

Current

Burundi 3.6 4.0 216 122 736 Beans–Bananas Yes

Egypt(irrigated) 19.5 19.5 122 183 2197 Wheat–maize Yes

Ethiopia 1.5 4.0 149 92 499 Sorghum–wheat Yes

Kenya 2.4 4.0 125 106 390 Beans–maize Yes

Rwanda 3.9 4.0 218 136 730 Beans–bananas Yes

Sudan 0.6 4.0 152 92 510 Sorghum No

Sudan(irrigated) 9.7 9.7 122 183 1007 Sorghum–wheat Yes

Tanzania 1.7 4.0 147 96 486 Maize–s.potatoes Yes

Uganda 2.9 4.0 218 123 743 Beans–Bananas Yes

Future

Futureintensive 4.0 218 123 743 Notspecified Yes

Ethiopia(newlyirrigated) 19.5 122 123 2256 ,, Yes

Sudan(newlyirrigated) 19.5 122 183 2196 ,, Yes

Table2

Foodself-sufficiencyandthecontributionofirrigatedagriculturetofoodself-sufficiencytargetsforthemainfood-producingcountriesintheNilebasin(Nilebasinarea);

baseline(2005)andthreefuturescenarios.

Baseline FutureTarget

2025

Scenarios NationalFoodSelf- Sufficiency

Upstream Hegemony

BasinCooperation Country Contributionof

irrigationtofood self-sufficiency

Overall foodself- sufficiency

Needed increasein agriculture GM

Contributionof irrigationtofood self-sufficiency

Overall foodself- sufficiency

Contributionof irrigationtofood self-sufficiency

Overall foodself- sufficiency

Contributionof irrigationtofood self-sufficiency

Overallfood self- sufficiency

Egypt 100% 135% 17% 85% 85% 57% 57% 78% 78%

Ethiopia 0% 78% 117% 16% 100% 16% 80% 16% 80%

Sudan 28% 92% 83% 0% 100% 122% 237% 93% 223%

Uganda 0% 102% 91% 0% 100% 0% 111% 0% 111%

Other 0% 75% 165% 0% 66% 0% 96% 0% 98%

Basin 48% 111% 74% 27% 92% 36% 103% 38% 107%

SudanincludesbothSudanandSouthSudan,butchangesinthecontributionofrainfedproductiontoGMrefermainlytoSouthSudan,whilechangesinirrigationare restrictedtoSudan,whichcontainsallthelarge-scaleirrigatedareas.

Fig.3.Increaseinannualagriculturalgrossmargin(inUSD/ha)betweenbaseline(2005)and2025(inascenariooffull“BasinCooperation”oninvestmentsinlandusechange andwaterresourceallocationforagricultureandhydropower).TheregionsindarkgreenrepresentincreaseingrossmarginintherehabilitatedirrigatedareasofSudanand thenewirrigatedareasinEthiopia,undertheassumptionthattheyreachthesameproductivityasEgypt’sirrigatedareas.Thedrawnriverwidthisproportionaltoannual meandischargeinthisscenario,withamaximumof2622m³/safterconfluenceofthemainNilewiththeAtbarainSudan.(Forinterpretationofthereferencestocolourinthis figurelegend,thereaderisreferredtothewebversionofthisarticle.)

regionalmaximum(Uganda,accordingtoFAO).Suchproductivity- basedgrowth,currentlyestimatedat1.3%forSub-SaharanAfrica (Fuglie andRada,2013), wasrepresentedbyanoptional‘future intensive’croppingsystem,activatedunderconditionsofsufficient water availability (SI). No investments were required for con- versionstomoreintensecroppingsystems,astheyareassumedto beanautonomousdevelopmentwithintheboundariesofcurrent agronomicpracticesintheBasin.

3.Results

ComparisonofWW-Nilerunoffandabstractionwithmodelled runoff (MWRI, 2005) and runoff derived from water balance estimates(Sutcliffe andParks,1999)for themainwaterbalance areasoftheNileshowsthereisquiteagoodmatchbetweenthe studies, especially considering the complexity ofthehydrologyin the basin(Fig.2).Justlikeourstudy,theseotherstudiesareconstrained by data availability and conceptual limitations. Despite these limitations,therunoffpatternissimilarinthevarioussubcatch- ments—eventheevaporativelossesintheSuddmatchratherwell.

The largest differences can be found in subcatchments with significantirrigationwaterabstractions.Thetwootherstudieslack dataonirrigationabstractions,focussingprimarilyonnaturalflow, i.e.runoff,fromthedifferentsubcatchments.

OurbaselinevalueofannualagriculturalGMof15.4billionUSD peryearisabout35%lowerthanthesingleavailableFAOestimatefor thebasin(Appelgrenetal.,2000).Theinclusionoflivestockinthe latterfigure,estimatedat18–35%ofAfricanagriculturalGDP(Ehui etal., 2002;Sansoucy,1995), can explainalargepart ofthedifference.

Toaccommodatethegrowthofthepopulationandmeetaminimum foodsupplyof2300kcal/p/day,totalfoodrequirementsareexpected to rise by 75% over the 2005–2025 period, according to our calculations.AmajorshiftoccursinEgypt,whichgoesfromfood surplustoshortage.

Ourresultsshowthatunderthe“NationalFoodSelf-Sufficien- cy” scenario,when noneof thecountriesisstimulatedtohave surpluses due to lack of trade, investments shift towards generatinghigherhydropowerrevenuesandthebasinasawhole willfailtobecomefoodself-sufficient(Table2).Egypt,Rwanda, andEritreaareunabletoproduceenoughfoodfortheirgrowing populations because of the restricted availability of water or agricultural lands. Under the “Upstream Hegemony” scenario, whenthereisnorestrictionontradewithinthebasin,foodself- sufficiencycan be realizedin 2025in the Nile basin at a total investmentcostof100billionUSD.Asimposedinthescenario, EthiopiaandSudanexpandtheirirrigatedagriculture.However, thisisachievedattheexpenseofincreasingthevulnerabilityof Egypt,withtheflowofwaterdownstreambeingreducedbyalmost 40%,asSudanandEthiopiafullydeveloptheirirrigationpotential.

Egyptwillbeabletoproduceonlyhalfitsneededfoodrequire- ments, increasing inequality in food self-sufficiency among countries.

Underthe“BasinCooperation”scenario,thebasinattainsself- sufficiencyina mannerthatisprofoundlydifferentfromthatof

“UpstreamHegemony”.Here,theLakeVictoriaregionandSouth Sudanareresponsibleforthebulkoftheincreaseinfoodproduction through intensification and expansion of the areas of rainfed agriculture (Fig. 3), while allowing Egypt’s highly productive irrigation schemes still to receive a large amount of water.

Interestingly,Ethiopiacanbefoodself-sufficient,butdoesnotneed to besounderthe ‘Basin Cooperation’ scenario, whereclimatic circumstancesforrainfedagriculturearemorefavourableinSouth Sudan and investments there are prioritized.Alimited reallocationof irrigation watertowardEthiopiaiswarrantedthough,asthecountry hasthe comparative advantage of more favourable rainfall and temperatureconditions than Egypt or Sudan. Rehabilitating the

currentlyunderperformingschemesofSudanisalsoprioritized,but additionalexpansionfurthernorthneartheMeroweDamisnot,as irrigationtherehasnoadvantageovertheexistingschemesinEgypt.

Waterallocationsof59billionm³toEgyptremainaboveitsshareof 55.5 billion m³ of the1959 treaty,a number oftenquoted. The construction of large hydropower reservoirs, like the Grand Renaissance Dam, does not affect Egypt’s share, neither does conversionoflandtorainfedagriculture.

Rainfedagriculturecontributesover75%oftheadditionalfood requirementsinallscenarios.Expansionof rainfedagricultureis suggestedprimarily inunstableregionsofSouthSudanandnorthern Uganda,wherethecausesofunderdevelopmentarelargelysocio- political as opposed to biophysical. Many parts of Africa are characterizedbyhighinter-annualandintra-annualvariabilityin rainfall (Cooper et al., 2008).A reliable rainfed agriculture will require investments in local water harvesting and site-specific supplementalirrigation(RockströmandFalkenmark,2015),inthe longrunsupportedbymoreaccurateregionalweatherforecasting andsmartformsofcrop,waterandsoilmonitoringandmanage- ment.However,thepessimisticviewofthewholeofEastAfrican agriculturebeingdrought-strickenneedsrefinementaswell.Fig.4, whichcomparesseasonalrainfalltotalswithcropwaterdemand, indicatessuitabilityforrain-fedagriculture,withcountryregions lyingwithintheNilebasinbeingwetterthanthecountries’total averages.Potentialnewagriculturalareasidentifiedinthisstudy haveatotalcropseasonprecipitationofabout900mm,morethan doublethecountry’saverageandwellabovecropwaterrequire- ments.Ourmodelsuggestsinvestmentsinatotalareaofaround 11millionhainSouthSudan,aboutathirdofthepotentialidentified.

Ifnotallcountriesareself-sufficientinfood,asisthecaseunder the“BasinCooperation”scenario,thenregionaltradeisrequiredto deliverfoodtowhereitisneeded.Foodsurplusregionsinthebasin aresituatedinthesouth,whereasthelargestshortageswilloccur in the north: in Egypt and Eritrea. While basic transport infrastructure is present in the form of river connections and railroads,historictraderoutesneedtoberevived.Tomakeoptimal use of comparative advantages, staple food suitable for long- distancetransporttoEgyptcouldbeproducedinupstreamareas, while Egypt could specialize in fresh produce for its urban populationandEuropeanmarkets(Wichelnsetal.,2003).Export ofagriculturalproducefromSouthSudan,which,accordingtoour calculations,couldamountto1.8billionUSDayearatfarm-gate level,willprovidediversificationtothisyoungeconomy,lessening itsdependenceonoil.Ethiopia’shydropowerrevenuescouldgive thecountryaccesstofoodmarkets,shoulditchoosenottodevelop its vulnerable highland regions to the maximum. The recent integration of energy grids in the region shows that such cooperationispossible.

4.Discussion

Thisstudyfocussesonthepotentialtoreachnationaltoregional food self-sufficiency in the Nile basin, as an alternative to an increasing reliance on global markets. This focus on food self- sufficiencygaveusaframeworktoassessthecontributionofrainfed agriculturecomparedtothatofirrigatedagricultureandtheimpact ofdifferentscenariosontheallocationofNilewaters.Wedonot, however,wishtoadvocatebasincooperationandself-sufficiencyas theonlysolutionorcriticisearelianceonglobalmarkets.Forthis,a differenttypeofstudyincludingananalysisofthecostsandbenefits ofregionaltoglobalfoodimportsandexportswouldberequired.

Similar to earlier studies that focused solely on irrigated agricultureand hydropower (Whittingtonet al.,2005;Wu and Whittington,2006),wefindthatbasincooperationwillprovidethe mostbenefittothebasin—aresulttobeexpectedgiventhenature

of the model used. Integration of rainfed agriculture in the objectivefunction, however,greatlychanges thesolutionspace available.Earlier(non-)cooperationstudieswiththeirstrongfocus on Nile water allocation tend to emphasize potential conflicts betweenEgypt,SudanandEthiopiaandhighlighttheroleofEgypt as the main hegemon and the unequal distribution of water (Cascão,2008,2009;Whittingtonetal.,2005;WuandWhitting- ton,2006).Inourstudyweshowthatadifferentdistributionwill merely shiftproduction, which will not be sufficient tofeed a growingpopulation.Inrainfedagricultureareas,however,there stillispotentialtoincreaseproduction.

An integrated analysis of this kind faces numerous data uncertainties.Several (theyieldofhydropower,theyieldof the currentirrigationsysteminSudanandtheinvestmentcostofland coverchange)wereassessedinapartialsensitivityanalysis(SI5).

Changesinthesevariablesdidnotchangetheresultssuchthatthe mainconclusionshadtoberevised.Inevitably,caveatsremain.Our studyaimstoexploredifferentsolutionsfromahydro-economic perspective,therebysimplifyingthediversityofcropproduction.

Limitations, e.g. in terms of soil nutrient conditions, farmers’ knowledge levels and access to markets, were not explicitly addressed,butimplicitlyincludedinapotentialrainfedyieldthat is lower thanwhat would beexpectedfrom cropand soil and meteorologicalcharacteristicsalone.

For agriculture in the political and socio-economic unstable regions of South Sudan and north Uganda to approach this potential yield requires considerable effort in creating the infrastructuretomakeknowledge,technologyandinputs–seeds, fertilizersandpesticides–availabletofarmers.Thisstudydidnot assessthelikelihoodofsuchdevelopments,butratheradvocates Fig.4. Satellite-derivedcountry-specificrainfall(source:TropicalRainfallMeasurementMission[TRMM]data(Kummerowetal.,1998))forvariousspatialdelineationsfor themaincroppingseasons(JJASOforSudan,SouthSudan,EthiopiaandEritrea;MAMandSONforallothercountries)inrelationtoaveragecropwaterrequirementsofrainfed agricultureduringthesemonths(setequaltopotentialcropevapotranspiration,basedonaverageECMWFreferenceevapotranspiration(Uppalaetal.,2005)andFAOcrop factors,seeSI).Greenshadesindicatearangebetween75%and100%ofcropwaterrequirements.(Forinterpretationofthereferencestocolourinthisfigurelegend,thereader isreferredtothewebversionofthisarticle.)

increasedefforttomakethishappen.Environmentalconsequences ofsuchdevelopmentshouldbethoroughlyassessed.Wedidnot includeenvironmentallimitationstoagriculturalintensificationor expansion.Butasustainableintensification(Godfrayetal.,2010), withproperlandmanagementtoreducenegativeexternalitiesof increasedproductionwillberequiredintheNilebasinasmuchas elsewhere.

Agricultural intensification and expansion did not lead to significantchangesindownstreamrunoff.Wefoundthatseasonal evapotranspirativedemandfromarablelandswassimilartothatof theoriginalvegetation in mostlocations.Cropfactors between naturalvegetationandarablecropsdonotdiffermuchduringthe peak growing season, when it rains most. Small variations in evapotranspirativedemandcanbebufferedbythesoilcolumn, resultinginevensmallerdifferencesinrunoffattheaggregated timestepsofacoupleofdaysasisusedinthemodel.Inaddition, swampsinthemainsurfacewatersystem,locatedmainlyinSouth Sudanandevaporatingalargefractionofrunoff,furtherattenuate anychangeinrunofffromthewhiteNilepartofthebasin.Itisin thisregionwheremostlandusechangeswereprojectedbythe model. In literature, increases in runoff after deforestation are reported, but mainly for temperate regions. Results from the tropicsaremixed(Brownetal.,2005;Bruijnzeel,1989).Oneofthe fewextensiveempiricalstudiesfromtheNilebasinitself,byHurni etal.(2005),suggeststhatintensificationoflanduseonsmalltest plots haslead toincreased surface runoff, but possiblyalso to decreasedbaseflow,whilesoilconservationmeasuresmighthave ledtolessrunoffinsemi-aridregions,butnotinthehumidparts,in theEthiopianhighlands.Moreingeneral,anychangeinlanduseon less than20% of thecatchment areaappearshard to detectin runoff (Bosch and Hewlett,1982; Brown etal., 2005;Stednick, 1996). Still, further study on the local and regional impact of upstreamlandusechangesand/withoutassociatedsoilandwater conservation measures using a model with a more detailed vegetation and land management parameterization would be requiredtoverifytheseinitialfindings.

In addition, a moredetailed analysisof theimpactof intra- seasonaldroughtsonfoodproductionisneededtofurtherverify whetherrainfedagricultureissustainable.Thisshouldideallybe supplementedwithananalysisoftherobustnessofagricultureand hydropower development under a range of future climate scenarios, given the diversity in both magnitude and direction ofchangeinprojectionsforthispartoftheworld.Ultimately,a comparison of regional versus global climate variability would shed more light on whether the region would be better off cooperating rather than depending onvolatile global markets.

Althoughregionalcooperationmakescountriesmorevulnerable toregionalclimateextremes,theregionwouldstillhaveasafety netduringsuchperiodsofbasin-widescarcity:theglobalmarket.

IftheNileregionweretorelyontheglobalmarketinthefirstplace, itcouldnolongeractasasafetynet.

Finally,ourstudyexploredthepossibilityofdifferentformsof cooperationfroma hydro-economic point of view, but didnot assessthelikelihoodofoneformofcooperationversustheotheror therequiredpoliticalandinstitutionalsetting.Onesuchinstitu- tionaloptionwouldbetoreinvigorateandbroadenthescopeofthe NileBasinInitiative(NBI),whichaimstostimulatecooperationby thenine Nile riparian countries. For theNBI totransit from a project towards a River Basin Organisation, a Cooperative FrameworkAgreement(CFA)hasbeenoutlinedandapermanent institutional mechanism should be established, the Nile River BasinCommission(NRBC).Inrecentyears,however,theNBIhas beenstrugglingtodefineandagreeontheCFAandtoestablishthe NRBC.Anotheroptionwouldbetoembednegotiationsonsharing theNilewaterswithinEastAfricantradeblocks,likeCOMESA,the Common Market for Eastern and Southern Africa. A focus on

broadereconomiccooperation,asalsosuggestedbyWichelnsetal.

(2003)andHilhorstandSchütte(2010)andfurtherexploredand quantifiedinthispaper,canprovideanewperspectiveontheissue ofsharingwaterresourcesbydefiningcommonbenefitsandanew angleforcooperationwithinexistinginitiatives.

5.Conclusion

WearguethatrainfedagricultureinunstableregionslikeSouth SudanandNorthUgandaiskeytofoodself-sufficiencyintheNile basinandthattheheateddebateonwaterallocationshouldbeput into perspective. Conflicts over allocation can only hinder cooperation on foodproduction and trade, thereby hampering theBasin’sdevelopment.

Egypt’spolicy standinparticular seemstoresemble a risky strategy:obstructingcooperationwithinthebasinandhindering upstreamwaterinfrastructuredevelopment,asithasdoneinthe past,givesEgyptthemostwater. Butifthislackofcooperation leads to unilateralism, increased and uncoordinated upstream abstractionswillhaveseriousconsequencesforEgypt’sagriculture andhydropowersectors.Theresultingmoreunequaldistribution of food self-sufficiency among basin countries will jeopardize regionalstability.However,wealsoshowthatamoreequitable solution is available, should countries choose to cooperate on basin-widefoodproductionandtrade,albeitwithsome,butrather limited,lossofwaterallocationsforEgypt.Thiswillrequireold policydogmastoberelinquishedandachangeofperspectiveboth onthebasin itself and ontheutilization of itsland and water resources.

Such a change in perspective asks for a different, more integrativeapproachtobasingovernanceandinvestments,away from the current focus on large water infrastructure projects.

Investmentsforsupportingatransitiontowardsaclimate-smart sustainableagricultureareneeded,withtechnologyimprovement and technologyadaptationand transferessential toreduce the environmentalimpactsofincreasedproductioninthebasin.The alternativeisanincreaseddependenceofNilebasincountrieson volatileglobalfoodmarkets.

Acknowledgments

OurworkontheNilebasinhasbeensupportedbythestrategic research program KBIV “Sustainable spatial development of ecosystems, landscapes, seas and regions” which is funded by the Dutch Ministry of Economic Affairs, and carried out by WageningenUR.WeespeciallywanttothankProfessorDavidGrey forofferingthoughtfulandhelpfulsuggestionsforimprovingan early draft and the anonymous reviewers for reviewing the manuscript.

AppendixA.Supplementarydata

Supplementarydataassociatedwiththisarticlecanbefound,in the online version, at http://dx.doi.org/10.1016/j.

envsci.2016.03.007.

References

Allen,R.G.,Pereira,L.S.,Raes,D.,Smith,M.,1998.Cropevapotranspiration—

guidelinesforcomputingcropwaterrequirements.FAOIrrigationandDrainage Papers,56.FAO—FoodandAgricultureOrganizationoftheUnitedNations, Rome.

Appelgren,B.,Klohn,W.,Alam,U.,2000.WaterandAgricultureinTheNileBasin.

FAO,Rome.

Arezki,R.,Bruckner,M.,2011.FoodPricesandPoliticalInstability.Leibniz InformationCentreforEconomics,Munchen.