Maize productivity and profitability in Conservation Agriculture systems across agro-ecological regions in Zimbabwe: A review of knowledge and practice

Full text

(1)

Review

Maize

productivity

and

pro

tability

in

Conservation

Agriculture

systems

across

agro-ecological

regions

in

Zimbabwe:

A

review

of

knowledge

and

practice

Paramu

Mafongoya

a

,

Leonard

Rusinamhodzi

b

,

Shephard

Siziba

c

,

Christian

Thierfelder

d

,

Brighton

M.

Mvumi

e

,

Brighton

Nhau

f

,

Lewis

Hove

g

,

Pauline

Chivenge

a,

*

aSchoolofAgricultural,EarthandEnvironmentalSciences,UniversityofKwaZulu-Natal,PrivateBagX01,Scottsville,Pietermaritzburg3209,SouthAfrica bCIMMYT,P.O.Box1041VillageMarket-00621,Nairobi,Kenya

cDepartmentofAgriculturalEconomics&Extension,UniversityofZimbabwe,P.O.BoxMP167,MountPleasant,Harare,Zimbabwe dCIMMYTSouthernAfricanRegionalOffice,12.5KmPegMazoweRoad,P.O.BoxMP163,MountPleasant,Harare,Zimbabwe eDepartmentofSoilScienceandAgriculturalEngineering,UniversityofZimbabwe,BoxMP167,MountPleasant,Harare,Zimbabwe

fFood&AgricultureOrganizationoftheUnitedNations(FAO)oftheUnitedNations,SubRegionalOfficeforSouthernAfrica,Block1,TendesekaOfficePark, Eastlea,Harare,Zimbabwe

gFoodandAgricultureOrganizationoftheUnitedNations(FAO),SubRegionalOfficeforSouthernAfrica,MerafeHouse,11NaivashaRoad,Sunninghill,2157 Johannesburg,SouthAfrica

ARTICLE INFO Articlehistory:

Received2December2014

Receivedinrevisedform7January2016 Accepted11January2016 Availableonlinexxx Keywords: Yieldadvantage Maize-basedsystem No-tillage Directseeding

Residueretentionandweeddynamics

ABSTRACT

Conservationagriculture(CA)isincreasinglypromotedinsouthernAfricaasastrategytoimprovefood securityandreversesoildegradationinthefaceofclimatechange.However,theperformanceofCA underdifferentenvironmentsanditsabilitytoimproveecosystemservicesisstillunclear.Theeffectsof theCAoptions;directseeding,rip-lineseeding,andseedingintoplantingbasinsonmaizegrainyield,soil healthandprofitabilityacrossagro-ecologicalregionsinZimbabwewereevaluatedthroughareviewof literatureincombination withmeta-analysis. Overall,CAimprovedmaizeyieldoverconventional agriculture.Comparedtoconventionalagriculture,directseeding,rip-lineseeding,andseedinginto plantingbasinsincreasedyieldby445,258and241kgha!1,respectively.However,therewasaninitial yielddeclineinthefirsttwoyears.CApracticesreducedsoilerosionandbulkdensity,andincreasedsoil watercontentinmoststudies.Underhighlevelsofresidueretention(6Mgha!1),CAsystemsexhibited greatermacrofaunaabundanceanddiversitythanconventionalagriculture,particularlytermites.Weed pressure tended to increase labour requirement for hand-hoe weeding under CA compared to conventionalagriculture.However,theuseofherbicidesreducedweedinglabourdemandduringthe earlyseason.ThebenefitsofCAaretiedtothefarmers’managementintensityincluding:timeofplanting, weeding,fertiliserandherbicideapplication,andadequatetrainingonequipmentuse.Economicanalysis resultsshowedthatonaverage,afarmerincurslossesforswitchingfromconventionalagriculturetoCA inthemainmaizegrowingregionsofZimbabwe.Basedonthesixseasons’data,thelosseswereleast withtheripperindrierareasandworstwiththedirectseederinwetterareas.Incorporationofchemical herbicidesworsenstheeconomicreturnsofCAtillageoptionsinalltheagro-ecologicalzones.Overall, thestudyshowedthattherip-lineseedingismoreattractiveinthedrierareasthandirectseeding. Althoughnotcostedinthisstudy,criticalisthecumulativereversalofsoildegradationassociatedwith consistentCApracticewhichcansustainagriculture.Resultsfromthisreviewsuggestthatthebenefitsof CAdependlargelyonthetypeandcontextofCAbeingpractised.Itisthusimperativetoprofilethe technology,thefarmersocio-economiccircumstancesandthebio-physicalenvironmentinwhichthe farmeroperatesforproper geographicalandbeneficiarytargetingtoachievegreater impact.More longer-termstudiesarerequiredtofullyelucidatethebenefitsandcontextofCAoptionsandpractice. ã2016ElsevierB.V.Allrightsreserved.

*Correspondingauthor.Presentaddress:ICRISAT,MatoposResearchStation,P.O.Box776,Bulawayo,Zimbabwe. E-mailaddress:pchivenge@gmail.com(P.Chivenge).

http://dx.doi.org/10.1016/j.agee.2016.01.017 0167-8809/ã2016ElsevierB.V.Allrightsreserved.

ContentslistsavailableatScienceDirect

Agriculture,

Ecosystems

and

Environment

(2)

Contents

1. Introduction ... 212

2. Materialsandmethods ... 214

2.1. Datasources ... 214

2.2. Sitedescriptions ... 214

2.3. Summaryoffieldexperiments... 214

2.4. Dataanalyses ... 214

2.4.1. Cropproductivity ... 214

2.4.2. EconomicsofswitchingtoCA ... 215

3. Resultsanddiscussion ... 215

3.1. YieldbenefitsofCAoverconventionaltillage ... 215

3.2. ConservationagricultureandsoilhealthinZimbabwe ... 218

3.2.1. Soilbiologicalproperties ... 218

3.2.2. Soilphysicalproperties ... 219

3.2.3. Soilchemicalproperties ... 219

3.3. WeeddynamicsinCAsystems ... 219

3.4. EconomicprofitabilityofCAoptions ... 220

3.4.1. EconomicreturnsofCAoptions ... 220

3.4.2. Breakevenmaizeyieldgains ... 221

3.4.3. Breakevenlaboursavings ... 221

3.4.4. OverallprofitabilityofCA ... 221

3.5. CAinthebroaderfarmingsystem ... 222

3.6. Governmentpolicy ... 222

3.7. Futureresearchneeds ... 222

4. Conclusions ... 222

Acknowledgements ... 223

References ... 223

1.Introduction

Increasing population pressure and competing global demands for food and feed, and in recent years, renewable energy, warrant the need for options that optimise crop production per unit area in the face of the changing climate (e.g.Lobell,2014).ThisisparticularlyimportantinSub-Saharan Africa (SSA) and the rest of the developing world, where an increasingdemandforfoodhasbeenworsenedbyavariableand changingclimate(Challinoretal.,2007).Conservationagriculture (CA)hasbeenpromotedinSSAinrecentyearsasanalternativeto conventional agriculture because of its ability to reduce soil degradation,enhancesoilhealth,improveresourceuseefficiency and sustainlong-term crop productivity (Gwenziet al.,2009; Vanlauwe et al., 2014). CA is based on three principles: (i) continuousminimummechanicaldisturbanceofthesoilthrough no-till,(ii)maintenanceofapermanentorganiccoverusingcrop residues, and (iii) diversification of crop species grown in sequences e.g. cereal-legume rotations (Kassam et al., 2009; FAO,2011).CAhasbeenpromotedinsouthernAfrica,aregion whichisvulnerabletoclimatechangeandvariabilityimpactsthat frequently result inlowcropyields andsometimes totalcrop failure(Hobbs,2007;Wall,2007).

ResearchonCAinZimbabwecanbetracedbacktoasfaras the 1920s, with the original focus mainly on minimum or reducedtillage.Highcostsofdieselandsparepartsasaresultof economic sanctions applied on the then colonial government acceleratedthedevelopmentofminimumtillageequipmentand practices.It isestimatedthatatindependencein1980,30% of commercial farmers in Zimbabwe were using reduced tillage systems(Smith,1988).From1996to1998,acollaborativeproject betweentheDepartmentofAgriculturalTechnicalandExtension Services and the German Technical Cooperation Agency was implemented focusing on reduced tillage for sustainable crop productioninsmallholderfarmingsystems.Themajorobjective ofthatprojectwastodevelopanumberoftillagetechnologiesto addressproblemsofsoilloss,runoffanderosionanddeclining

cropyields(Nyagumbo, 2002). In that project, mulchripping, cleanrippingandtiedridgesweretestedagainstaconventional control treatment, i.e. moldboard ploughing or hand-hoeing. Mulchrippinginvolvedtheuseofatinerippertorip("25cm depth)betweenplantingrowswithcropresiduesbeinglefton thesurface whereas cropresidues were removed underclean ripping. Tied ridges were permanent ridges with ties in the furrowsapproximatelyevery50cm,creatingabasin-likesystem withcropsplanted onthe ridges.Afterseveral seasonsof on-farm and on-station research, it was concluded that mulch rippingwithitshighwateruseefficiencywasthemostviable conservation tillage practice in the semi-arid regions of Zimbabwe (Nyagumbo, 2002). However, there was limited uptake of reduced tillage practices by smallholder farmers during and after the project. This was largely due to the institutionalframeworkof theproject andlack of appropriate equipment(Johansenetal.,2012).

In 2003, there was renewed interest to promote CA in Zimbabwewithitsthreemainprinciplesofno-till,mulchcover andcroprotationsordiversification,mainlydrivenbysubstantial donorfundingtargetingfoodsecurityforvulnerablehouseholds withnodraughtpowertofacilitateearlyplantingandincreased cropyields(Mashingaidzeetal.,2006).Aconsortiumof organiza-tionsincludingFoodandAgricultureOrganizationoftheUnited Nations(FAO),DepartmentforInternationalDevelopment(DFID), European Union, International Crop Research Institute for the Semi-arid Tropics (ICRISAT), International Maize and Wheat ImprovementCentre(CIMMYT),UniversityofZimbabwe,various non-governmentalorganizations(NGOs)andextension services were involved in research and promotion of CA technologies throughout the country. The CA technologies promoted were primarilyseeding intohand-hoebasins,animaltractionrip-line and animal traction direct seeding. Tractor powered seeding systemswereleftoutbecausemostdonorfundsweremeantto targetonlythosecommunitiesconsideredvulnerableduetolackof draught power, labour and chronic illnesses such as HIV/AIDS (Nyamangaraetal.,2014).Thebasin/mulchpackagerepresenteda

(3)

disconnection with earlier science based experimentationwith reducedtillagepracticesthatwerepreviouslytestedinZimbabwe. Oflate,thefocusofpromotioninZimbabwebytheDepartmentof Agricultural andExtension Services has shifted to mechanized systems(Marongweetal.,2011).

Whilethecommongoalistoimprovecropproductivity,the term CA means different things to different stakeholders e.g. farmers,NGOs,media,policymakers,amongothers.Termssuch CA,conservationtillage,no-tillage,conservationfarming,precision conservation agriculture, resource conserving technologies and minimumtillagepracticesarecommonlyusedinCAnomenclature (Table1).HenceCAinsmallholderfarmingsystemsinZimbabweis asetofprincipleswherereducedtillageisappliedindifferentways accordingtofarmercircumstances.

CropproductivityonmanysmallholderfarmsinZimbabweand much ofAfrica isconstrainedby acombination of interrelated factors, e.g. low soil fertility, insufficient and inappropriate

fertiliserapplication, erraticrainfall,lackofimproved cultivars, labour constraints,andinsome situations inappropriatetillage practices.Duetothis,manyfarmersaretrappedinabjectpoverty, experience food insecurity and poor nutrition (Sanginga and Woomer, 2009). CA has been promoted to address these constraints (FAO, 2002; Wall, 2007), with approximately 903,000haoflandunderCAinsouthernAfrica(Friedrichetal., 2012).ThecomponentsofCA;no-till,mulchingwithlivingordead plantsandcroprotationanddiversificationaffectprocessessuch asorganicmatterdecomposition(Chivengeetal.,2007;Gwenzi etal.,2009),rainfallinfiltration,soilmoisture,soilerosion(Yimer et al., 2008; Thierfelder andWall, 2009), and ultimately crop productivity(Stevensonetal., 2014).CAcanpotentiallymakea differenceinsituationsoferraticrainfalldistributionandseasonal dryspellswherehighermoistureconservationduringcriticalcrop phasesmayincreasecropyieldsatharvestoratleastreducethe

risk ofcrop failure(Thierfelder andWall, 2010; Rusinamhodzi etal.,2012).ThissuggeststhatCAhaspotentialtoimprovecrop productivityinthesemi-aridareas, andwillbecomeespecially important in Zimbabwe and southern Africa where drought occurrence is projected toincrease as theclimate changes.CA alsoimprovessoilfertility,associatedwiththebuild-upoforganic carboninthesoil,althoughtheamountandthetimeneededto reach appreciableamounts of soil organic carbon varygreatly between sites and cropping systems (Govaerts et al., 2009; ThierfelderandWall,2012).Theclimaticandedaphicconditions insouthernAfricavarygreatlywithinshortdistancesduetorelief andsoil parent material (Lewis andBerry,1988), andit isnot sufficienttodrawlessonsfromfewlocalities.

In the smallholder farming systems of Zimbabwe land preparationhaslargelybeenox-powerbased,butwiththedecline in livestock numbers as a result of droughts and economic challenges,therehavebeennotableincreasesinmanualplanting. TheCAoptionsavailableinmanualsystemsare:useofthedibble stick,jabplanter,hoeplanterandplantingbasins,whereas animal-traction rippers and direct seeders are the more mechanized systems accessible to smallholder farmers. The plantingbasin, ripperandthedirectseederarethemostpracticedCAoptionsin thecountryatpresent.ThebenefitsofCAarediverseanddependto alargeextentonthenatureoftheagro-ecosysteminquestionand howwellCAtechnologiesareadaptedtothelocalenvironmental and, socioeconomicandculturalconditions.Moststudiesshow thatCAhasyieldadvantageovertheconventionalagriculturein mostagro-ecologicalzones(Marongweetal.,2011).Moststudies alsoshowthatlabourrequirementwithsomeoftheCAoptions also increases (Siziba, 2007; Mazvimavi and Twomlow, 2009; Thierfelderetal.,2013),mostlyduetoincreasedweedpressure,if noherbicidesareused.Perennialweedsbecomedifficulttocontrol

when soil is not tilled. Although the labour demand forland preparationandcropseedingaresubstantiallyreducedwiththe ripperanddirectseeder,basinsareinfact,morelaboriousasthey are dugmanually with hand-hoes.To afarmer thedecisionto switchfromtheconventional plough totheCA optionslargely depends on trade-offs between gains in crop yield and the opportunitycostofadditionallabour.Withtheripperanddirect seeder, implementation costs are additionally considered. If herbicides are used, the trade-off is between the cost of the chemicalandthereductioninlabourneededformanualweeding. AreviewofCApracticesinZimbabwewasconductedwiththree mainobjectives.Thefirstobjectivewastoquantitatively synthe-sise the yield advantages of practising CA over conventional agricultureinthemaize-basedsmallholderfarmingsystemsacross agro-ecologicalenvironmentsinZimbabweusingpublisheddata. The second was toevaluate theimpact ofCApractices onsoil health.Thethirdobjectivewastoevaluatetheeconomicbenefits (profitability) of switching to CA options for farmers. Current estimates suggest a total of over 300,000 smallholder farmers implementingsomecomponentsofCA(mostlyno-tillageplanting basins) overan areajustabove 100,000 hectares (Nyamangara etal.,2014).However,theestimateswerenotbasedonthestrict definitionsuggestedbyFAO(2011).Thus,inordertoretainenough

datatobeabletoconducttheanalysis,CAsystemsincludingthose not complying with the strict definition were used in this synthesis.Maize wasusedas themaintest cropas itaccounts for 50-90% ofthe population's caloric intake (Dowswell et al., 1996). Yieldadvantageswere analysedbycomputingthemean difference between CA and conventional agriculture. As an incentive forfarmerstoadopt anewproductionsystem short-term benefits especially in the form of yield gains or other economicbenefitsareimportant.Finally,thestudyanalysedhow thebiophysical,socio-economicandinstitutionalconditionsinthe

Table1

Typologies of conservation agriculture in smallholder farming systems in Zimbabwe.

Technology/practiceused Characteristics

Conventionaltillage Intensivetillageusingploughs,discs,harrows andrippers

Noresiduesleftonthesurface

Minimumtill/tillage #Minimumsoilmanipulation,thiscausesalot confusion

Striptill Usesminimumtillage,leavessomecropresidues Conservationtillage Objectivetoreducesoilerosionand30%cover

withcropresidue

Notill/notillage/zerotillage Growingcropsyeartoyearwithoutsoil disturbance

Directseeding/mulchbased croppingsystems #

Minimumsoildisturbance

#Soilcoverthroughcovercropsinrotationwith themaincrop

Conservationfarming Usingabasinandmulch Plantingbasins UsingpitssmallerthantheZaipits Zaisystems OriginatedfromWestAfrica

#Harvestingwater

#Applyingorganicinputs

#Nomulchisused Conservationagriculture

(CA)

3principles

#Minimumsoildisturbance

#Atleast30%cover

#Croprotation Fertiliserapplication

PrecisionCA #Applicationoffertiliserinbasin

(4)

selectedsitescansupportandsustainthebenefitsrealisedunder CA.

2.Materialsandmethods

Forthepurposesofthisstudy,CAsystemsevaluatedwerewith thefollowingno-tillpractices;rippertines,directseedersandjab planters.Residueswereleftonthesurfaceunderno-till.

2.1.Datasources

Maizegrain yield datawere obtained from tillageandcrop residue management studies established under rain-fed condi-tionsinZimbabwe.Treatmentshadtobefromrandomisedplots withatleastthreereplicates.Studieswereobtainedfromrefereed journals,bookchaptersorpeerreviewedconferenceproceedings (see Appendix A) through online searches in OvidSP, Scopus, Google Scholar and Web of Science as well as personal communicationwithkeyexpertsandresearcherswhoareworking onCA.Theonlinesearchwascomprehensive,usingthefollowing keywords and their combinations: conservation agriculture; reducedtillage;no-tillage;maizeyield;rain-fed;Zimbabwe.The dataincludedintheanalyseshadtosatisfythefollowingminimum requirements:

(a)The studies were established under rain-fed conditions in three agro-ecologicalzones(II,IIIandIV),locallyknownas naturalregions1inZimbabwe,

(b)Thetreatmentsincludedaconventionaltillage(control)and no-till,

(c)The experimental design was randomised and treatments replicatedatleastthreetimes,

(d)Thetestcropwasmaizeandthesamevarietywasappliedtoall treatmentsundercomparison,and

(e)Nutrientmanagementwasthesameacrosstreatments.

2.2.Sitedescriptions

Mostsoilsinthestudysiteshadaloamytosandyloamtexture, dominated by Lixisols and Arenosols while some sites were characterisedbyLuvisols.Inallthesites,maizeisthemajorfood cropwhilesorghum(Sorghumbicolor(L.)Moench)andpearlmillet (Pennisetumglaucum(L.)R.Br.)areimportantcereals.Grainlegume crops such as groundnuts(Arachis hypogaea L.), cowpea(Vigna

unguiculata(L.)Walp),drybeans(PhaseolusvulgarisL.)areoften

growninrotationorassociationwiththecerealcrops.Thelarge partsofZimbabwesmallholderfarmingsystemsarecharacterised by agreater integrationofcropandlivestockproduction. Most smallholderfarmersusecattlefordraughtpowerandmostlyuse themouldboardploughforlandpreparationandalsocultivators for weed control (Riches et al., 1997). During cultivation, the ploughcuts,breaks,loosens,invertsthesoilandburiesweeds,crop residuesandmanure.

2.3.Summaryoffieldexperiments

Thetillageexperimentswereestablishedasfollows:

Conventionaltillage (i.e.control treatment) withoutresidue retention,achievedusinga mouldboard plough atshallow soil depth(10–15cm)andplantingintofurrowscreatedbytheplough. Remainingcrop stubbles and weeds were incorporated by the plough,

Rip-line seeding (Ripper) using an animal traction ripper (Magoyeripper)withresidueretention.Maizewasplantedin rip-linesof5–10cmdepthatseeding.Cropresidueswereretainedin situafterharvest,

(a)Animal drawn direct seeder where maize was fertilised at planting,cropresidueswereretainedinsituafterharvest,and (b)Plantingbasinsdugusingahand-hoe,eachbasinwas15cm (length),15cm(width)and15cm(depth),andspacedat90cm betweenrowsand60cmwithinrows.

Weedcontrolonconventionalagriculturewasmanualthrough hand-hoes,andonripperanddirectseeder,chemicalherbicides wereused,mainlyglyphosateappliedatplanting.

2.4.Dataanalyses

2.4.1.Cropproductivity

Ameta-analysiswasperformedusingthe‘meta’packageinRto assesstheyieldbenefitsofCAoverconventionaltillagepractices acrosssitesandseasons.Theeffectsizewasobtainedbycomputing theweightedmeandifference(WMD)usingtherandom-effects model(DerSimonianandLaird,1986;Eggeretal.,1997;Borenstein etal.,2009).Weusedthemeandifferenceinyieldbetweenthe treatmentandcontrolbecauseofitseaseofinterpretationandthe relevanceforcomparingpotentialgains(Ried,2006;Sileshietal., 2008).Intherandomeffectsmodel,itisassumedthatthetrue effectofCAoncropyieldvariedfromsitetositeandfromseasonto season,thuscontributionsofeachstudytotheoveralleffectsize wasconsideredtobeindependent.Weight(wi)assignedtoeach

studywascalculatedastheinverseofthevariance(1/vi)whereviis

thewithin study variance forstudy (i). The weighted mean is calculatedas the product of mean andweight divided by the overallweight.Theoverallweightistheinverseofvarianceofthe wholestudy(Eqs.(1.1)–(1.5)).

MeandifferenceðMDÞ¼meantreated!meancontrol ð1:1Þ

weighti¼variance1 i¼ 1 SD2 i ð1:2Þ

WeightedmeandifferenceðWMDÞoverall

¼ Xi¼n i¼1 ðweighti'MDÞ Xi¼n i¼1 weighti ð1:3Þ

CI95%¼meanoverall(ð1:96'ðvarianceoverallÞ0:5Þ ð1:4Þ

Varianceoverall¼ 1

Xi¼n i¼1

weighti

ð1:5Þ

1Naturalregionsareagro-ecologicalzonesclassiedmainlyonthebasisof amountofrainfallandtemporaldistribution.Therearesixmainzonesandthe othersareI()1000mm;mostofwhichfallsthroughouttheyear;goodsoils),IIa (750–1000mm;goodtemporaldistribution;generallygoodsoils),IIb(sameasIIa butlessreliable);III(500–750mm;subjecttoseveremid-seasondrought);IV(450– 650mm;Severedryspellsduringtherainyseason,andfrequentseasonaldroughts; normallyconsideredunsuitablefordrylandcropproduction);V(<450mm;Highly erraticrainfall;poor soilsnormallyconsideredunsuitableforcropproduction [Adaptedfrom:VincentandThomas,1960;DepartmentoftheSurveyorGeneral, 1984].

(5)

Studies(seasonsandsites)withlowervariancecontributedmore weighttotheoveralleffect.Theanalysisincludedhowthesemean differenceswereaffectedbysoiltypeandtheamountofrainfall received in a season. All the data were derived from studies established inNaturalRegionsII, IIIandIVas described under Section2.1.Nitrogeninputwasnotconsideredbecauseitdidnot varymuchacrosstheexperimentsusedforthemeta-analysis.

2.4.2.EconomicsofswitchingtoCA

Literature shows that economic incentive (profitability) is central to sustainable uptake of a new technology (Cary and Wilkinson, 1997; Pannell, 1999). A partial budget approach (CIMMYT, 1988) was adopted in evaluating the net economic effectofswitchingfromtheconventionalploughtoprospectiveCA options.Thefocuswasontheeconomicimplicationsofmakingthe switch for each CA option. The financial benefits and costs

associatedwith each optionwereevaluated against anaverage resourced smallholder farmer in Zimbabwe mostly cultivating maizeonaonehectareplot,usingthetraditionalox-drawnplough andfamilylabour.Itwasassumedthattheinputuse(fertilisersand seeds)levelsbetweenCAandtheconventionalsystemswerethe same—necessaryforafaircomparison.Inthiscontext,switchingto CAimpliedpossiblechangesinthefollowingparameters:achange inmaizeyield(gainsareexpected),changeinlabourdemand(an increase—duetoextraweedingordecrease—duetoreducedtillage canoccur),andequipmentcostsofnewimplements.Ifherbicides wereusedtheircostwasincorporated.Intheanalysisweattached valuestothesecostsandbenefitstocomputetheneteconomic effectofswitchingtoCAforfarmers.Ourpriceswerebasedonthe 2013/14season.

Incrementalmaizegain,themajorexpectedbenefitofCAwas evaluatedusingthe2013/14maizepriceofUS$375permetricton. Theannuitymethodwasusedtospreadtheinvestmentcostsof equipmentovertheanalysisperiodofsixyears.Forallequipment, annualmaintenancecostswerecomputedas5%ofthepurchase valueoftheequipment.Costinglabourinthesmallholdersectorin Zimbabweisnoteasybecausemostfarmersusefamilylabour,and alsothelabourmarketisnotfullydeveloped.TheGovernmentof ZimbabwegazettedwagerateofUS$3perdaywasusedtoattach value to the opportunity cost of family labour. However, the authorswerecognisantofthefactthatthewageratescanfluctuate acrosstheseasontendingtobehighduringpeakperiods.Under herbicideuse,amixtureofglyphosate,atrazineandmetolachlor; identifiedasthemosteffectiveunderCA(Muonietal.,2013)was assumed.Againmarketpriceswereusedincostingthisherbicide regime.Theeconomiceffectisthenetdifferenceofthevalueof benefitsandvalueofcostsassociatedwitheachCAoption.Thenet economiceffectforthesixyearsafarmeradoptCAwascomputed. ThisallowsthecaptureofthedynamiceffectsofCAyieldsover time. TheNetPresent Value(NPV)was usedtosummarisethe economic effect over the six years of switching from the conventional plough tillage to CA options of direct seeding or ripping.TheNPVisasummationofthediscountedneteconomic effectsoverthesixyears.Aninterestrateof14%,theprevailingreal interestrateatthetime,wasusedtodiscountthestreamofnet economiceffects.

Breakevenyields andlaboursavings requiredfortheswitch fromtheconventionalploughtoCAwerealsoderived.Theseare thevaluesoflaboursavingsandmaizeyieldgainsrequiredtomake the NPV equal to zero. This is the situation when CA and conventionalploughingwouldgivethesameeconomicreturns, a threshold for the switch. For example, a farmer considering uptakeofthedirectseederwithoutuseofherbicideswillhaveto purchasethedirectseederandincurinvestmentandmaintenance costofthenewequipment.Sincethefarmeralreadyownsaplough whichalsosuffersthesamekindofcosts,thenetequipmentcost

wouldbethedifferenceinthecostofthetwoequipment.Sincethe directseederatUS$550ismoreexpensivethantheplough(US$ 125) the farmer incurs higher equipment cost for the switch. Assuming that overall labour use does not change, then the incremental equipment cost should be off-set by the value of additionalmaizeyieldforthefarmertoswitchtothedirectseeder. The minimum maize grain yield gain required to off-set the incrementalequipmentcostisthebreakevenyieldgain. Alterna-tively,assumingnoyieldchange,theminimumlabourreduction necessary to off-set the incremental equipment cost could be computed—thebreakevenlaboursavings.Incorporationof chemi-calherbicidesinCAoptions,potentiallyreduceslabourdemand, butalsoaddscosttofarmers.Herbicideusecostsincludethecost oftheherbicidechemicalsand,theinvestmentandmaintenance costs of the knapsack sprayer. Currently, few farmers use herbicides, even with CA systems. Exploring the economic implications ofherbicideuseare limitedby thelackofstudies thatexaminedthelabourdynamicswhenherbicidesareused.The trade-offbetweentheadditionalcostsoftheherbicidechemicals andtheextenttowhichweedinglabourisreduceddeterminesthe neteconomiceffect.Muonietal.(2013)exploredweedcontrol strategiesinCAsystems.Itistheonlyavailablepublishedmaterial trackinglabourusechangesunderdifferentherbicideregimesin Zimbabwe.Fromthiswork,itwasderivedthatthebestherbicide controlregime(amixtureofglyphosate,atrazineandmetolachlor) canreducemanualweedinglabourinCAbyasmuchas8daysha!1. Below arethe economic effectandNPV equations usedinthe analyses:

i)Theeconomiceffectequation

EE=

D

Y'P+

D

EC+

D

L'W+

D

HC (1.6)

whereEEistheneteconomiceffectoftheswitch(US$);

D

ECis change in equipment cost (US$);

D

L is change in labour use (daysha!1);Wisthewagerate(US$day!1);

D

Y=changeinmaize yield(kgha!1);Pisthemaizegrainprice(US$kg!1);and

D

HCis changeincostofherbicides(US$ha!1)

iiÞNPV¼X 5 t¼0

EEt

ð1þrÞt ð1:7Þ

whereNPVisthenetpresentvalue(US$);EEistheeconomiceffect ineachsubsequentyearofCAadoption;tisthetimeperiod(0in thefirstyearofadoptionand5forthe6thyear);risthe real interestrate(0.14).

Theminimummaizeyieldgainsandlaboursavingsthatwould benecessaryforfarmerstoatleastfindtheCAtillageoptionsas profitableastheox-ploughtillagewereexplored.Todothisallelse washeldconstantandonlytheparameterofinterest(labouror maizeyieldgain)changedandthepointatwhichtheNPVwas equal to zero observed; a situation when farmers would be indifferent betweenchoosingtheCAtillageoptionandthe ox-ploughtillage.Thisanalysiswasnotdoneforherbicide incorpo-rationbecauseofthepooreconomicperformance.

3.Resultsanddiscussion

3.1.YieldbenefitsofCAoverconventionaltillage

Overall,thedirectseedertreatmentwasnotdifferentfromthe ripper(sub-soiler)butout-yieldedconventionaltillagebyalmost 300kgha!1(Fig.1).Largeyieldbene

fitsofCAwereobservedin favourablegrowingconditionsofNaturalRegionIIwiththedirect seederwhereastheripperwassuperiorinthedriestregion,i.e.in NaturalRegionIV(Fig.2).Basinsandjab-plantertreatmentswere excludedfromthebox-plotduetolimitedsamplesizesfromthe availabledata(Fig.1).However,similaranalyseshaveshownthat

(6)

basins out yielded conventional tillage in 64% of the cases considered in the semi-arid regionsofZimbabwe (Fig. 3), and theripperwasonlysuperiortoconventionaltillagein8%ofthe casesunderthoseconditions(Nyamangaraetal.,2013a).

Theweightedmeandifferenceshowedthatdirectseeding out-yielded conventional tillage by 445kgha!1 overall (Fig. 2a). However, yield advantages across the agro-ecological regions tended tovary;theyielddifferencetendedto belargestinthe favourable growing conditions of Natural Region II, but were significantinNaturalRegionIV.The95%confidenceintervalfor weightedmeandifferenceinNaturalRegionsIIandIIIextended into the negativeindicating no significantdifferences between directseederandconventionaltillageintheseregions.

Theoverallyieldadvantageofripperoverconventionaltillage was284kgha!1,muchlessthanthatofthedirectseeder(Fig.2b). There were no significant differences between the ripper and conventional tillageinNaturalRegionsIIandIII,butinNatural RegionIV,characterisedbylowerrainfall,theyieldadvantageof theripperwas400kgha!1.Thisisbecausetheripperallowsdeeper waterinfiltrationandgreatermoisturecaptureandconservation than conventional tillage, which is beneficial under dry

environments and within-season dry-spells (Rao et al., 1998; Fanetal.,2013).

LargeyieldbenefitsofCAwereobservedinNaturalRegionII withthedirectseeder(thoughtherearechancesoflowyields) whereas,itwaswiththeripperinthedrierregions.Theresults suggestthatthehighrainfallinNaturalRegionIIisfavourableto thedirectseederwhereseedsareplantedinfairlyshallowplanting holesandislikelytooffsetthedevastatingeffectsoflongdryspells. Thehighyieldpotentialinthisregionalsosuggestsanincreasing likelihoodofprovidingsufficientbiomassformulch.Itshouldalso benotedthatasignificantproportionofthedatashowednegative effectofdirectseedingonyield.Thiscouldsuggesttheeffectsof waterloggingduetoexcessiverainfall(Griffithetal.,1986).

InthedrierNaturalRegion IV,thesub-soilingimpactofthe rippercouldbeanimportantfactorinincreasinginfiltrationand

storageof rainwater, and thus wateruse efficiency. In Natural

RegionIV,therewasnonegativeweightedmeandifferenceinboth thedirectseederandrippersuggestingthepositiveeffectofthese tillagetreatmentsonthewaterbalance(Nyamangaraetal.,2013a). ThisisinsupportoftheviewthatCAimprovescropyieldsrelative to conventional agriculture in dry areas because of greater moistureconservationleadingtoenhancedcropwater productiv-ity(Farooqetal.,2011).Arecentglobalmeta-analysis,showedthat althoughtheoverallresultsuggestthatCAreducescropyields,in

Fig.1.Summarystatisticsformaizegrainyielddatafromthetillagetreatments usedinthemeta-analysis.Valuesabovetheplotsarethemeansforeachtreatment. Meanyieldsfollowedbyadifferentletteraresignificantlydifferentatp<0.05.The boxplotrepresentsthestandardfivenumbersummary:minimum(lowerwhisker), firstquartile,median,thirdquartile,andmaximum(upperwhisker).Thepoints beloworabovetheminimumandmaximumrespectivelyareoutliers.

Ripper

Region II Region III Region IV Overall -600 -400 -200 0 200 400 600 800 1000 Direct Seeder

Region II Region III Region IV Overall

W eig ht ed m ea n di ffe re nc e ( kg h a -1 ) -600 -400 -200 0 200 400 600 800 1000 n = 368 n = 284 n = 124 n = 168 n = 76 n = 98 n = 112 n = 74

Fig.2.Pooledandnaturalregion-specificweightedmeandifferences(WMD)inyieldabetweendirectseederandconventionaltillage,andbbetweenripperand conventionaltillage,inZimbabwe.RegionII,IIIandIVrepresentthreeofthesixnaturalagro-ecologicalregionsofZimbabwe1.Theverticalbarsrepresentthe95%confidence intervals. Experiment 0 20 40 60 80 100 We ig ht ed m ea n di ffe re nc e (k g ha -1) -3000 -2000 -1000 0 1000 2000 3000 Planting Basins n = 81 WMD = 241 kg ha-1

Fig.3.Theweightedmeandifference(WMD)inyieldbetweenplantingbasinsand conventionaltillageforthedriersouthernpartofZimbabwe.Theverticalbars respresentstandarddeviation.FigureoriginallyreportedbyNyamangaraetal. (2013b).

(7)

drierregionsCAincreasescropyields,duetothegreaterbenefit derivedthroughimprovedmoistureconservation(Pittelkowetal., 2014),dependingonthetypeofCAsystemsandsoiltypes.Thishas importantimplications onthe potential ofCAto improvecrop productivityand its sustainability in Zimbabwegiven that the southernAfricaregionisprojectedtogetdrierwithmoreerratic rainfallandheatstress(Challinoretal.,2007;Cairnsetal.,2012; Cairnsetal.,2013;).

Negativeweightedmeandifferenceswererecordedinthefirst two yearsof theexperiments inboth directseeder andripper

treatments in both Natural Region II and III, and increased thereafterbutdiminishedafteryearfiveinNaturalRegionIIbut maintainedinNaturalRegionIIIandIV(Fig.4).Theinitialyearsof shiftingtoCAarefraughtwithchallengesincluding:inadequate amounts ofcrop residues,poor management skillsby farmers, weed pressureinthe absenceof herbicidesandor insufficient labourforoptimumweedcontrolcandepressyield(Muonietal., 2013;Muonietal.,2014;Stevensonetal.,2014;Thierfelderetal., 2015a).Inthemediumterm,yieldscanimproveduetothelikely improvement insoil quality, e.g. soil organic matterover time

Dire

ct seeder

1 2 3 4 5 6 We ig ht ed m ea n di ffe re nc e (k g ha -1) -4000 -2000 0 2000 4000

Re

gi

on

II

I

1 2 3 4 5 6 We ig ht ed m ea n di ffe re nc e (k g ha -1 ) -4000 -2000 0 2000 4000

Re

gi

on

IV

Seasons of CA (years) 1 2 3 4 5 6 7 We ig ht ed m ea n di ffe re nc e (k g ha -1 ) -4000 -2000 0 2000 4000

Ripper

1 2 3 4 5 6 -4000 -2000 0 2000 4000

Re

gi

on

II

1 2 3 4 5 6 -4000 -2000 0 2000 4000 Seasons of CA (years) 1 2 3 4 5 6 7 -4000 -2000 0 2000 4000 (a) (b) (c) (d) (e) (f)

Fig.4.Weightedmeandifferencesinyieldbetweendirectseederandconventionaltillageontheleft;andripperandconventionaltillageontherightintimeacrossnatural regionsinZimbabwe.Thecomparisonsbetweenthedirectseederandconventionaltillageareshownin(a),(c),and(e)forRegionsII,IIIandIV,respectively.Thecomparions betweentheripperandconventionaltillageareshownin(b),(d),and(f)forRegionsII,IIIandIV,respectively.Theverticalbarsrepresentthe95%confidenceintervals.

(8)

(Chivenge et al., 2007; Thierfelder andWall, 2012; Thierfelder etal.,2015b).Thedepressedyieldsinlateryearscouldbeduetothe negative monoculture effects on maize productivitypreviously observedinasimilarstudy(Rusinamhodzietal.,2011).

Afterasimilarmeta-analysisofICRISATdatafromtillageand crop residuemanagement “CA” trialsbetween2004 and2010, Nyamangaraetal.(2013b)observedthat:plantingbasinCAhasa greater chance of increasing maize yield above conventional agriculture dueto precision application ofnutrientsandwater collection in the basins. Planting basins were superior to conventional tillage in 59% of the experiments. The overall reported weighted mean difference for planting basins was 241kgha!1 (Fig. 3). However,

firstly, this yield advantage was

reportedtodecreaseto95kgha!1ifrainfallwasintherangeof 500–830mm per season. Secondly, CA does not stabilise yield

under conditions of poor rainfall distribution and should be implemented in combination with other drought mitigation technologies and the use of drought-tolerant varieties (Cairns et al., 2013). Thirdly, the performance of CA under semi-arid conditionsisenhancedbytheadditionofsmallamountsofmineral Nfertiliserandcattlemanurebutisdepressedbysurfacemulching with crop residuesof highC:N ratios.Nutrientmanagement is important for improving crop production underCA (Rusinam-hodzi, 2015b).Lastly,basinsshouldberecommendedforsandy soilswithgooddrainage.Thedatashowingthisevidencewerenot includedinthecurrentreviewbutthefindingsarerelevantasthey coveredmostofthesemi-aridregions(Nyamangaraetal.,2014). The general conclusion from the meta-analysis was that yield performanceunderCAisinfluencedbysoiltype,rainfallamount anddistribution,inorganicfertiliserandmanureapplication.These

findingsarefurthersupportedbyanotherglobalmeta-analysisof maizeyielddataunderrain-fedconditions(Rusinamhodzietal., 2011)(Fig.5).

3.2.ConservationagricultureandsoilhealthinZimbabwe

ThebasisforyieldincreasesinCAsystemsisincreasedwater useefficiencyandimprovednutrientcycling.Thesederivefrom soilsurfacecover,increasedmicroandmacrofauna,andinsome instances, accumulation of soil organic carbon (Brouder and Gomez-Macpherson, 2014) that improve soil physical and biologicalproperties.Inthefollowing sections,theeffectofCA on the soil biological, physical, and chemical properties in Zimbabweareexplored.

3.2.1.Soilbiologicalproperties

Soilfaunaareimportantregulatorsofdecomposition,nutrient cycling,soil organic matterdynamics andimprovementof soil physical properties. CA using planting basins combined with mulching have a significant effect on soil fauna diversity and abundance(Table2)(Nhamo,2007;Mutemaetal.,2014).Mutema et al. (2014) observed high fauna population (termites, ants, centipedes and beetle larvae) in CA systems compared to conventionalpracticesinstudiesconductedonmultiplesitesin NaturalRegionIIIandIVinZimbabwe.Inthatstudy,termiteswere the most dominant fauna group across all sites with the CA treatmenthavingatleast1600%greaternumbersoftermitesthan conventional agriculture (Table 2). Similarly, Nhamo (2007) observedatleast 120%more termites underdirect seederand ripperbasedCAthantheconventionalpractice.Theabundanceof

Table2

Asynthesisoftheeffectsofconservationagricultureonsoilproperties.

Property Units Conventional CA %difference Tillage TimeunderCA(years) Soiltexture Source Biological Number/monolith

Termites numberm!2 750 12925 1623 Basin 2 Loam Mutemaetal.(2013)a numberm!2 50 110 120 Directseeder 2 Sandyloam Nhamo(2007)

numberm!2 50 140 180 Ripper 2 Sandyloam Nhamo(2007)

Ants numberm!2 95 350 268 Basin 2 Loam Mutemaet al.(2013)

Centipede numberm!2 0 290 Basin 2 Loam Mutemaetal.(2013)

Beetlelarvae numberm!2 100 685 585 Basin 2 Loam Mutemaetal.(2013) numberm!2 10 20 100 Directseeder 2 Sandyloam Nhamo(2007)

numberm!2 10 22 120 Ripper 2 Sandyloam Nhamo(2007)

Earthworms numberm!2 5 8 60 Directseeder 2 Sandyloam Nhamo(2007)

numberm!2 5 10 100 Ripper 2 Sandyloam Nhamo(2007)

Physical

Bulkdensity gcm!3(05cm) 1.45 1.35 !7 Basin 5 Sandy Nyamangaraet al.(2014) gcm!3(05cm) 1.25 1.15 !8 Basin 5 Clayloam Nyamangaraetal.(2014) Infiltration mm 25.3 78.2 209 Directseeder 6 Sandysoil ThierfelderandWall(2012)

Porevolume %(0–5cm) 7 16 128 Basin 5 Sandy Nyamangaraetal.(2014)

%(0–5cm) 11 19 73 Basin 5 Clayloam Nyamangaraetal.(2014)

Soilmoisture mm 66.6 80.7 21 Directseeder 3 Sandy ThierfelderandWall,(2009) Erosion tha!1 68.9 29.9 !130 Directseeder 8 Sandy Thierfelderetal.(2012a,b)b

Run-off mm 361 165 !119 Directseeder 3 Sandy ThierfelderandWall(2009)

mm 1.7 1.5 !12 Ripper 1 Sandy Mupangwaetal.(2008)c

mm 1.7 1.0 !41 Basin 1 Sandy Mupangwaetal.(2008)

Chemical

pH H2O(0–20cm) 4.6 5.1 10 Basin 9 Sandy Nyamangaraetal.(2013)d

OrganicC tha!1(030cm) 6.9 13.3 93 Directseeder 4 Sandy ThierfelderandWall(2012) gkg!1(020cm) 5.9 6.8 15 Basin 9 Sandy Nyamangaraetal.(2013) tha!1(015cm) 9.4 15.9 69 Basin 5 Sandy Nyamangaraetal.(2014) tha!1(015cm) 20.5 28.4 38 Basin 5 Clayloam Nyamangaraetal.(2014) TotalP gkg!1(020cm) 0.17 0.19 12 Basin 9 Sandy Nyamangaraetal.(2013) aAveragedacrossvesitesfortheCAtreatmentwhere2tha!1residueswereadded.

bCummulativeerosionloadover8years.

cBasinbasedCAaveragedacross450farmsin15districts. dAveragedacrossfoursites.

(9)

termitesconfirmsfarmers’beliefthatretentionofcropresidues will increase prevalence of termites, which may subsequently damagecrops.Mutemaetal.(2013)observedanincreaseinfauna abundance with increasing amounts of crop residues applied under CA, in the order of 6tha!1>4tha!1>2tha!1>0tha!1. However, results showedlimitedoccurrence of millipedes and earthworms.Incontrast,Nhamo(2007)observed60%and100% more earthworms under CA than conventional agriculture. Additionally, there were greater abundances of beetle larvae underCAthanconventionalagricultureandinallcasestheripper hadmorefaunathanthedirectseeder.Theincreasedabundanceof soil fauna under CAimproved soil physical properties such as infiltration,porosity,aggregatestabilityandhydrological

proper-ties.

3.2.2.Soilphysicalproperties

Increasedrainfallinfiltrationandsoilwaterstorageappeartobe

the most obvious and immediate benefits associated with CA providedthatthereissufficientsoilcover(Thierfelderetal.,2014). Thierfelder and Wall (2012) observed up to 209% greater infiltration under direct seeded CA treatments with surface residuesretentioncomparedwithconventionaltillagetreatments withresidueremoval(Table2)inastudyconductedfrom 2005-2010withamini-rainfallsimulatoratHendersonResearchStation inZimbabwe.ThissupportsotherstudiesfromMalawi, Zambia andMozambiquewhereincreasedinfiltrationhasbeenshownto be desirablewhen mid-seasondry-spellsare common(Ngwira etal.,2013;ThierfelderandWall,2010;Thierfelderetal.,2013), otherwise waterlogging can arise leading to depressed yields (ThierfelderandWall,2010).

Inconventionalagriculturalsystems,soillossesoftheorderof 50Mgha!1 peryearthroughsoil erosionandlossesof30% of received rainfall has been reported in Zimbabwe (Elwell and Stocking, 1988). In contrast, CA reduces soil erosion and was initially introduced to regulate both wind and water erosion (Rosenstocketal.,2014).Manytrialshaveshownthepositiveeffect ofCAonwaterproductivity,higherinfiltrationratesandwater content.Inaddition,reducedsoillossandrun-offwererecordedin CAsystems comparedtoconventional tillagesystems(Table2; Thierfelder andWall, 2009). Greater infiltrationrate underCA comparedtoconventionalagriculturewasobservedalreadyafter thesecond yearofcontinuous treatment(ThierfelderandWall, 2009). The abundance of macro fauna in CA systems also contributes to improving soil physical properties. Increased infiltrationwaseitheraresultofaggregatestability(Thierfelder etal.,2014),greaterbiologicalactivityandreducedbulkdensity (Table2;Nyamangaraetal.,2014).

Similarly,CAhasbeenshowntoimprovesoilqualitythrough theimprovementofsoilphysicalpropertiesinfournaturalregions inZimbabwe,butwithgreaterbenefitsinhighclaysoils(Brouder andGomez-Macpherson,2014).Incontrast,however,therewasno significant correlation between improved bulk density and

infiltration with maize grain yields in a study conducted in

Murehwadistrict,Zimbabwe(Rusinamhodzietal.,2013).Thismay implythatsoilphysicalpropertiesmaynotbethemajorlimitation tomaizegrainyieldsbutnutrientsupplyandavailabilityinstead (Nyamangaraetal.,2013a).

3.2.3.Soilchemicalproperties

Globally,theimpactofCApracticesonsoilorganiccarbonis inconclusive(cf.Govaertsetal.,2009).InZimbabwe,Nyamangara etal.(2013a)rejectedthehypothesisthatCAincreasessoilorganic carbonafterastudybasedonsoilsamplescollectedfromabout 450farmsin15districtsacrossthecountry(Table2).Similarly, Chivengeetal.(2007)afterassessingorganiccarboninsoilderived from a 9-year tillage experiment concluded that soil organic

carbonishardlyaffectedby tillagepracticesinsandysoilsand decisionstoenhanceorganiccarboninthesesoilsshouldfocus moreonmanagingorganicinputs.Lackofsufficientcarboninputs intheformofmulchwhichisoftenthecaseonmostsmallholder farms canlead tonoorvery slowsoilorganiccarbonincrease (Nyamangara et al., 2013a). This observation is particularly importantgiven thatsmallholder farmersin Zimbabweoccupy mostlythepoorsandysoils.WherepositiveeffectsofCAonsoil organiccarbonarereported,themagnitudeofchangeareoften verysmallorinsignificant(Thierfelderetal.,2012a,b;Pannelletal., 2014). Nyamadzawoetal. (2009)reported greater soil organic carbon sequestration in no-till systems because of improved aggregation which protected carbon frommineralization com-pared toconventionaltillagealthoughthesystemwasdifferent because continuous maize was compared with maize-legume fallows.Itappearsbiomassproductionduringthefallowperiod wasthemajordriverofSOCincreasesinthemaize-fallowsystems underno-till(Nyamadzawoetal.,2008).

TherewerenosignificantdifferencesinsoilpHandphosphorus between CA and conventional systems (Supplementary Fig. 4; Nyamangaraetal.,2013a).Thiscouldbeattributedtoapplication of low amounts of mulch with high C:N ratios, which could potentiallyleadtoimmobilizationofmineralnutrientsandlow cropyields.However,theseresultswereobtainedfromshort-term researchprojects.Thereisaneedtoinvestinlong-term experi-ments(beyond10years)tomonitorchangesinsoilpropertiesand relatethesetocropproductivity.

3.3.WeeddynamicsinCAsystems

Although CA systems improve crop yields for smallholder farmers,challengeswithweedmanagementintheearlyyearsof adoptionhasbeencitedasamajorreasonforlowadoptionofthese systems on small plots by farmers (Mazvimavi andTwomlow, 2009).Increasedweedgrowthhasbeenreportedtobeproblematic inthefirstfewyearsbutwilldeclineandbecomeeasiertocontrol

withtimeinCAsystems(Wall,2007).Insupportoftheseearlier claims,Muonietal.(2014)recentlyfoundthatweedpopulation declinedexponentiallyinfourcroppingseasonsonasandysoilsite atDomboshawaTrainingCentre,Zimbabwe.Other studieshave shownincreasedweedpressureaftersixyearsofpractisingCAand relatedreducedtillagesystems(Nyamangaraetal.,2013b). Long-termstudiesbyMashingaidzeetal.(2012)showedthatCAsystems hadhigherearlierweedgrowthcomparedtoconventionaltillage (SupplementaryFig.3).Thiswouldimplytheneedforearlyand morefrequentweedinginCAsystemscomparedtoconventional tillagesystemsandthusincreasethelabourdemand.Mabasaetal. (1998)showedthat theseedbedunderthemouldboardplough wasweed-freeforuptofourweeksafteraploughingoperationand reduced the need for early weed control. Most of the weeds observedinCAsystemswereperennial,whichgrowrapidlywith thefirstrainsduetothedeeprootsystemandtheirperennial

growth habit that enabled them to tolerate long dry seasons (Mashingaidzeetal.,2012). Perennialweedsregeneraterapidly afterhand-hoeweedingunderwetconditions,suggestingmore labourisneededforweedinginCAsystemsifnoherbicidesare used(Mashingaidzeetal.,2012).

Weedcontrolmethodsusedbymostsmallholderfarmersrely onhand-hoeweedingandtoalesserextentmulching.Theneedfor earlyandfrequentweedinginCAsystemwillresultincompetition forlabourforploughingandplantingoperationsatthebeginning of the season. Hence, there is a need to evaluate less-labour-demandingweedcontrolmeasures.Mulchingwithcropresidues hasbeenshowntobeanappropriatepracticetoreduceearlyweed growth.Mulchingwithmaizeresiduesat4Mgha!1reducedweed growth (Mashingaidzeetal., 2013). Weedsuppressionthrough

(10)

mulchingmayreducelabourdemandintheearlyseason.However, themajorityofthesmallholderfarmersmaynotbeabletomulch at 4Mgha!1 because of mulch scarcity.The amount of mulch available to smallholder farmers is limited by low biomass production in semi-arid conditionsunder poorsoils. The little cropresiduesavailableisalsousedforlivestockfeedingduringthe dryseason,hencethereislittlemulchwhichremainsforsufficient mulchingtoeliminatetheneedforearlyweedinginCAsystemsas suggested previously (e.g. Valbuena et al., 2012; Rashid etal., 2013).

LabourconstraintsandmulchlimitationsforweedcontrolinCA systemsdictatethatalternativeweedcontrolsystemsbeevaluated andpromoted.Weedcontrolusingahand-hoeinCAsystemis inadequateduetohighweedpressureanddiversity(Arslanetal., 2014). Severalauthorshavesuggestedtheuseofherbicidesfor weed controlin CAsystems (Mashingaidze etal.,2012; Muoni etal.,2013).Pre-emergenceherbicidessuchasatrazine,cynazine and alachlor have been used to control weeds effectively in conventionaltillagesystems.Untilrecently(Muonietal.,2014), herbicideeffectivenessinCAsystemswithhigherweeddensity anddiversityhasnotbeenwell-documentedinZimbabwe.There were significant lower weed densities under pre-emergence herbicides comparedto hand-hoeing atthreeweeksaftercrop emergence(SupplementaryFig.4;Arslanetal.(2014)).

Theeffective weedsuppressionby pre-emergence herbicide during thefirstsixweeks aftercrop emergence hashelped to reducelabourrequiredforearlyweedinginCAsystems.Mixtures oftwo pre-emergenceherbicidesorthreeeffectivelycontrolled weedscomparedtoindividualherbicidesapplication(Muonietal., 2013;Arslanetal.,2014).Applicationofmixturesofherbicides

effectivelycontrolledbothgrassandbroadleavedweedsandalso changedweedcommunitystructurebyreducingweeddiversity indices.Thishassignificantimplicationsforweedmanagement strategies.Althoughapplicationofherbicidescansavelabourfor weedcontrolinCAsystems,useofherbicidesbysmallholdersis problematic. Earlier studies have shown that only 2% of smallholder farmersuse herbicidesin Zimbabwe(Vogel,1994) althoughthishasincreasedinthe2000s.Thisismainlybecause farmersin lowproductivity areaslack the capital to purchase herbicidesandassociatedequipment.Herbicidetechnologyisalso knowledge-intensive and thus requires substantial capacity development of smallholder farmers on the correct herbicide applicationequipment,rate,techniques,timingandaswellassafe useofthechemicals(Siziba,2007).Moreresearchisneededthat focussesonalternativeweedcontrolstrategiestoreducecosts.

3.4.EconomicprofitabilityofCAoptions

3.4.1.EconomicreturnsofCAoptions

Table3illustratesthedetailsofhowNPVwascomputedusing thecase ofNaturalRegionIIandTable4givesthesummarised resultsforallnaturalregions.Thenetpresentvalues(NPVs)were negativeforbothCAtillageoptionsinallthethreenaturalregions (Table 4). Thus the value of yield gains in the six years was outweighed by the additional costs of equipment and labour associatedwith switchingtoCAoptions.TheprofitabilityofCA tillage options was further reduced by the incorporation of herbicides. This means that the evaluated herbicide option (mixture of glyphosate, atrazine, and metolachlor) adds more coststhan thebenefits to farmers. Herbicidesare supposed to

Table3

NeteconomiceffectsofswitchingtoCAoptionsandnetpresentvalues(NPV)oversixyears—computationexampleinNRII. Year 1 2 3 4 5 6 NPV Rippernoherbicide Grain(kg) 16.38 !276.93 348.78 678.05 650.86 !433.97 Grainvalue($) 6.14 !103.85 130.80 254.27 244.07 !162.74 Labourcost($) !72.00 !72.00 !72.00 !72.00 !72.00 !72.00 Equipmentcost($) 3.19 3.19 3.19 3.19 3.19 3.19

Neteconomiceffect($) !62.67 !172.66 61.98 185.46 175.26 !231.55 !$4.92 Directseedernoherbicide

Grain(kg) 336.00 !152.84 808.35 1365.30 473.02 497.11

Grainvalue($) 126.00 !57.31 303.13 511.99 177.38 186.42

Labourcost($) !171.00 !171.00 !171.00 !171.00 !171.00 !171.00 Equipmentcost($) !90.44 !90.44 !90.44 !90.44 !90.44 !90.44

Neteconomiceffect($) !135.44 !318.75 41.69 250.55 !84.06 !75.02 !$10.43

Ripperwithherbicide 1 2 3 4 5 6

Grainvalue($) 6.14 !103.85 130.79 254.27 244.07 !162.74

Labourcost($) !48.00 48.00 48.00 48.00 48.00 48.00

Equipmentcost($) !3.62 !3.62 !3.62 !3.62 !3.62 !3.62

Herbicidecost($) !63.95 !63.95 !63.95 !63.95 !63.95 !63.95

Neteconomiceffect($) !109.43 !219.42 15.22 138.70 128.50 !278.31 !$8.26 Directseederwithherbicide

Grainvalue($) 126.00 !57.31 303.13 511.99 177.38 186.42

Labourcost($) !147.00 147.00 147.00 !147.00

Equipmentcost($) !97.25 !97.25 !97.25 !97.25 !97.25 !97.25 Herbicidecost($) !63.95 !63.95 !63.95 !63.95 !63.95 !63.95

Neteconomiceffect($) !182.20 !365.51 !5.07 203.79 !130.82 !121.78 -$13.77 Assumptionsandnotes:

Allequipmenthaveausefullifespanof15years.

Theannuitiesmethodswasusedtocomputetheannualcostofallequipmentusinganinterestrateof14%. HireratefordraughtpowerisUS$50perha.

WagerateisUS$3permandayandamandayis8hlong. MaizepriceisUS$0.375perkg.

Equipmentpurchaseprices:plough=US$125;ripper=US$110;directseeder=US$550. NPVcomputedusingarealinterestrateof14%.

(11)

substituteforthelaborioushandweeding.Thevalueofdisplaced labour,wasthereforelessthanthedirectcostofpurchasingthe herbicidesplustheindirectcostofacquiringtheknapsacksprayer attheprevailingmarketpricesofalltheseinputs.Thusgiventhe prevailingyieldresponses,maizegrainprices,equipmentcosts, labour prices andinterest rate, farmerswould makeeconomic lossesbyswitchingfromtheconventionalsystemtotheCAoption (directseederandripper)inthethreeNaturalRegions(II,III,and IV). The ripper option fared better in all the natural regions comparedtothedirectseederbecauseofitslowerequipmentcost. The bestperformanceforthe ripperwithoutherbicideswasin NaturalRegionIV(NPV=$!1.79)andlargelydrivenbyrelatively

highermaizeyieldgainsovertheox-ploughtillagerealisedinthe region.IntherelativelydryNaturalRegionIV,useoftheripper triggeredsubstantial maizeyieldgainsthatnearlyequalledthe cost of additional weeding labour and the relatively lower equipmentcost.Therelativelyequipmentcostheavydirectseeder performedbestinthewetterNaturalRegionII.

3.4.2.Breakevenmaizeyieldgains

Generally,lessmaizeyieldgainsarerequiredwiththeripper thanwiththedirectseederinallnaturalregionstomakeCAtillage matchox-ploughtillagereturns(Fig.5VsFig.6).Forexample,in Natural Region IV, an additional maize yield gain of 125kg is requiredineachofthesixyearswiththeripper.Anadditionalyield gainofasmuchas1000kgperyearisrequiredforthedirectseeder togiveequaleconomicreturnsastheox-ploughinNaturalRegion III (Fig. 6). Generally, the additional yield gains, ranging 125–

1000kg, seem relatively high. This is because in addition to covering the cost of labour and equipment, there is need to compensatefortheobservednegativeyieldgainsthatoccurinthe secondyearofCAadoptionandinsomecasesevenbeyondthe secondyear.

3.4.3.Breakevenlaboursavings

Literatureshowsthatlabouruseincreasesby24mandaysha!1 and57mandaysha!1fortheripperanddirectseederrespectively (Siziba,2007).Fig.7depictlaboursavingsneededtoforCAtillage optionstorealiseequaleconomicreturnsastheox-ploughatthe observed maize yield gains. The less costly ripper generally requiredlesslaboursavingsthanthedirectseederinallnatural regions.Anincrementinlabourofupto16labourdaysha!1was allowableto makethe ripperas profitable as theox-plough in NaturalRegionIV;inNaturalRegionIIthecriticalpointwaszero; andinNaturalRegionIIIalaboursavingof30mandaysha!1was required.Forthedirectseeder,theNaturalRegionIIhadthelowest labour savings requirement (!27mandaysha!1); this was fol-lowedbyNaturalRegion IV(!10mandaysha!1);andinNatural Region III a positive labour saving (60mandaysha!1) was necessary. It is noteworthy that the breakeven labour savings forbothCAtillageoptionsarelowerthanactualmarginsobserved inliterature,exceptinNaturalRegionIII.

3.4.4.OverallprofitabilityofCA

At currentaverage CAyieldresponsesandmarket prices in Zimbabwe,theCAoptionsdonotpresenteconomicincentivesfor farmerstoadopt. Theripperfaredbetter, becauseofthelower equipment cost. The labour requirement increases, largely for weeding, present a huge opportunity cost of farmers' labour. Although,thisisanimplicitcost,itisimportantbecauseevenif farmersdonothingbutrest,theopportunitycostoftheirlabouris

Table4

Netpresentvalues(NPV)forswitchingfromconventionaltillagetoCAtillage options.

Naturalregion

CAtillageoption II III IV

NPV Noherbicide Ripper !$4.92 !$14.09 !$1.79 Directseeder !$10.43 !$32.21 !$14.82 Withherbicide Ripper !$8.26 !$17.43 !$5.13 Directseeder !$13.77 !$35.55 !$18.16 -20 -15 -10 -5 0 5 10 -30 -20 -10 0 10 20 30 NPV ($)

Labour savings (mandays)

Region II Region III Region IV

Fig.7.RipperbreakevenlaboursavingsacrossnaturalregionsII,IIIandIVin Zimbabwe.NPV=netpresentvalue.

-40 -30 -20 -10 0 10 20 30 0 500 1000 1500 NPV ($)

Maize yield gain (kg ha-1)

Region II Region III Region IV

Fig.6.DirectseederbreakevenmaizeyieldgainacrossnaturalregionsII,IIIandIV inZimbabwe.NPV=netpresentvalue.

-15 -10 -5 0 5 10 15 20 25 30 35 0 500 1000 1500 NPV ($)

Maize yield gain (kg ha-1)

Region II Region III Region IV

Fig.5.RipperbreakevenmaizeyieldgainsacrossnaturalregionsII,III,andIVin Zimbabwe.NPV=netpresentvalue.

(12)

not zeroaspeopleattach somevaluetoleisuretime (CIMMYT, 1988).Thefourcriticalparametersthatdriveneteconomiceffects of switching to CA are the yield gain advantage, and thecost elements of equipment, herbicides and labour. While we are confidentoftheyieldimpactestimatebecauseoftheconsiderable number ofstudies that quantified CA yieldresponses, and the relativelylargesamplesandwidergeographiccoverage(different rainfall,soil type),dataontheaccompanyinglabour changesis scanty.Thepaucityoflabourusedatacouldbeduetothedifficulty ofmeasuringlabour.Inthecurrentpaper,asinglestudybySiziba (2007),whichisbasedononlysevenon-farmtrialswithinone locality inonedistrict,wasreliedonforlabourdata.However, throughthebreakevencurves,theanalysisallows evaluationof economiceffectsunderdifferentyieldandlabourresponselevels. Thoughattheaveragesoftheeconomiceffectequationparameters (maizeyieldresponses,prices,andlabourusechanges)thereturns werenegative,therearestillconditionsunderwhichreturnscould bepositive.Thesewouldgenerallybecaseswhereyieldresponses are maximisedandlabourincreasesminimised.Theyieldgains maybeincreasedbygoodfarmermanagementorwheresoilsare fertile butmoisture isvery limiting such inNaturalRegion IV. Labourincreasecanbeminimisedunderbiophysicalenvironments whereweedpressureisnaturallylow.Therearemanyfarmerswho maybeabletoenjoythisnicheandrealisepositivereturns.Such prospects aremore likelywith theripper inNaturalRegion IV wherebestNPVswererealised.

3.5.CAinthebroaderfarmingsystem

In-situcropharvestresiduemanagementisoneofthepillarsof

conservation agriculture (CA) but also the most pronounced barriertoitswidespreadpracticeespeciallyonsmallholderfarms in thetropics(Rusinamhodzi,2015a). Thisisbecausecrop and livestockproductionarecloselyintegrated(ThorntonandHerrero, 2001;Rufinoetal.,2011)i.e.cropresiduesareneededtoprovide

livestockfeedduringthedryseasonwherefeedisseverelylimited whilemanureisstronglyneededforcropproduction(Rufinoetal., 2011; Rusinamhodziet al., 2013). Due tothe communal open-grazingsysteminthedryseason,non-cattleownersneedtoeither investinfencingorcarrytheircrop residuesforprotectionand bring them back at the start of cropping incurring significant labourcostsintheprocess.Thustheabsenceofalternativegrazing in the dry season due to degraded communal rangelands characterizedbypoorqualityfodder(Rufinoetal.,2011),coupled withrejectionofpotentialfoddergrassandtreespeciesbyfarmers (Giller, 2001) means smallholder farmers at present cannot achievesufficientmulchcovertosuccessfullypracticeCA(Giller etal.,2009).

3.6.Governmentpolicy

Implementation of CA in southern African countries is at different stages, driven by significant policy shifts in some

countriesandindifferenceinothers.TheNGOcommunityplays a majorrolethroughadvocacyaswell asstrategiclobbyingfor external funding for the technology. The government policies aroundCAarestillfragmented.Theconditionaltechnical perfor-mance ofCA often reportedhave not done much to influence agriculturalpolicyinsouthernAfrica.InZimbabwe,the govern-menthasadoptedCAasoneofthesustainabletechnologiesthat can increase productivity and productionand a CAup-scaling frameworkwhichtargetsatleast500,000farmerspractisingCAon atleast250,000habytheendof2015withanaverageyieldof 1.5tha!1onCAeld(AMID,2011).CAisnotexclusivelypromoted atpolicylevelthussuggestlimitedmechanismstoencourageits widespreadpractice.

3.7.Futureresearchneeds

The CAinitiatives and experiences inZimbabwe are largely focussedonthetechnologyandlessontheusers.Muchdataare primarilyonbiophysicalimpacts,andlittleeffortiscommittedto understandinganddescribingthetechnologyuserorthe“farmer”. Inaddition,privatesectorengagementmodelshavebeenlimited. Basedonthesescenarios,thefollowingaspectswereidentifiedas importanttoshapethefutureresearchandpromotionaleffortson CAinZimbabwe:

#Thereisneedforstudiesonfarmers’knowledge,attitudeand

perceptionson CAondifferentsoiltypesandagro-ecological zonesinordertodevelopappropriatelocallyadapted technolo-gies

#Thereisneedtodevelopandevaluatemachinerytoeaselabour

demandsexperiencedespeciallyintheinitialstagesofCA.

#Public-Private Partnerships, including local level CA land preparation services that provide sustainable access to CA equipmentby smallholder farmers, needto be explored and promoted.

#Needmoreinformationoncost-effectivenessofherbicideuse underdifferentcircumstances.

#Research is needed to explore various options to generate sufficient plantbiomassforeffectivecoverinCAsystems. As mostsmallholderfarmsaremixedcrop-livestock,moreresearch isneededtobetterunderstandthetrade-offsbetweenlivestock andcrop productionwith specificsonhowto makeCAfit in systemswithhighdemandsforcropresidues.

#BecauseCAisnotamaize-onlycroppingsystem,researchefforts

should target adaptation of CA systems to other crops (e.g. cassava,sweetpotatoes,tobacco,cotton,smallgrainsetc.)

#Soilhealth indicatorsinCA basedsystems onsoil biological,

chemical and physical properties should be developed and downscaled for easier understanding by field practitioners,

developmentagentsandfarmers

#Long-termstudiesontheimpactofCAsystemsonsoilbiological, chemicalandphysicalpropertiesareneededtounderstandthe sustainabilityofthesystems

#There isneed to establish comprehensive andwell-designed studieson thebiophysical andsocioeconomic boundary con-ditionsforCAadoption.

#Alternativeland-usepracticesneedtobeexploredwhereCAis not adequate or not possible due to biophysical and socio-economicconstraints.

#Thereisneedformorefarm-levelstudiestotracktheactualyield gains andlabour use changes realised by farmers to enable empiricalevaluationofCAprofitability.

#The niche for CA in the current farming systems could be

increasedifbetterfarmplanningprocedureswereappliedand implementedbyfarmers.

#ThepotentialeffectsofCAonclimateadaptationandmitigation

needtobebetterunderstoodandquantificationoffutureeffects

couldbeassessedthroughmodelling

#Research on policy and institutional studies to support CA

adoptiononawidescaleisneeded.

#Adoptionanddis-adoptionstudiesonCAsystemsontemporal andspatialscalesareurgentlyneeded.

#The occurrence androle of termites in mulched land needs furtherinvestigation.

4.Conclusions

InZimbabwe,well-designedlong-termCAexperimentsbased onsoundagronomicmanagementpracticesarestillscarce.

(13)

Field-basedresultsshowthatCAhasgreaterpotentialtoincreaseyields insomeareasmorethaninothers.Theresultsalsoshowthat rip-lineseedingisamoreattractiveoptionthandirect-seedinginthe drierareas.Soilqualityassessmentsshowageneralincrease in biologicalactivityinCAsystemsbutduetothelimitedstudiesthat haveaddressedthesubject,resultsarenotconclusive.Therelative profitability ofCA issmall, largely becauseofincreased labour costsforweedcontrol.ThereturnsofCAdependonspecificyield gainsandlabourincreasesperfarm,whichareinfluencedbythe interactionsbetweenthebiophysicalconditionsandthefarmer's management.However,dataonlabourdemandinCAsystemsis very scanty. Prospects of improving profitability are there,

especially if labour demand is reduced. The success of CA implementationwilllargelydependonaddressingthechallenges together with farmers to find local solutions andto adapt CA

systemtolocalconditions.Resultsfromthisreviewsuggestedthat thebenefitsofCAdependlargelyonthetypeofCAbeingpracticed,

where,whenandby whom.Itisthusimperativetoprofilethe technology,theendusers(farmers)andthebio-physical environ-ment in which they operate for proper targeting and greater impact.Theevidencepresentedinthisstudysuggeststhatmore longer-term studies beyond ten years are required to fully elucidatethebenefitsandcontextofCApractice.

Acknowledgements

The organisations such as CIMMYT, ICRISAT and FAO, and individualssuchasProf.JusticeNyamangara,ChinhoyiUniversity ofTechnology,Zimbabwe(formerlywithICRISAT), andMsSepo LungoweMarongwe,Department ofAgricultural, Technicaland ExtensionServices,Zimbabwe(AGRITEX)arespecifically acknowl-edgedforprovidingdataaswellasothertechnicalreportsthat allowedustocompletethemanuscript.

AppendixA.Studiesusedinthemeta-analysis.

Study Source Treatments

Mashingaidze etal.(2012)

SoilandTillageResearch CP,NT Mavunganidze

etal.(2014)

CropProtection CP,NT

Nyamangara etal.(2013a,b)

Agriculture,EcosystemsandEnvironment CP,NT Muonietal.

(2013)

CropProtection CP,NT

Thierfelder (2006)

14thInternationalSoilConservation OrganisationConference—Morocco.

CP,NT Thierfelderetal.

(2006)

CPWDInternationalForumonWaterand Food

CP,NT Walland

Thierfelder (2008)

SSSAAnnualMeeting,Houston,2008 CP,NT

Thierfelderand Wall(2009)

SoilandTillageResearch CP,NT Walland

Thierfelder (2009)

ChallengeProgramonWaterandFood,2008. CP,NT

Thierfelderand Wall(2010)

JournalofCropImprovement CP,NT,NTR Thierfelderand

Wall(2010)

ExperimentalAgriculture CP,NT,NTR Thierfelderand

Wall(2012)

InnovationskeyfortheGreenRevolutionin Africa

CP,NT,NTR Marongweetal.

(2011)

InternationalJournalofAgricultural Sustainability

CP,NT Thierfelderetal.

(2012a,b)

InternationalJournalofAgricultural Sustainability

CP,NT,NTR Thierfelderetal.

(2012a,b)

FieldCropsResearch CP,NT Thierfelderetal.

(2012a,b)

SoilUseandManagement CP,NT,NTR

(Continued)

Study Source Treatments

Thierfelderetal. (2013)

FieldCropsResearch CP,NT,NTR Thierfelderetal.

(2013)

SoilandTillageResearch CP,NT,NTR Thierfelderetal.

(2013)

InternationalJournalofAgricultural Sustainability

CP,NT Mupangwaetal.

(2007)

PhysicsandChemistryoftheEarth CP,NT,NTR Mupangwaetal.

(2012)

FieldCropsResearch CP,NT,NTR

CP=conventional ploughing; NT=no tillage, NTR=no tillage withlegumerotation.

AppendixB.Supplementarydata

Supplementarydataassociatedwiththisarticlecanbefound,in theonlineversion,athttp://dx.doi.org/10.1016/j.agee.2016.01.017. References

AMID,2011.ConservationAgricultureanInvestmentFrameworkforZimbabwe (2010–2015).In:DevelopmentM.a.I(Ed.),GovernmentofZimbabweMinistryof Agriculture,Harare,Zimbabwe.

Arslan,A.,McCarthy,N.,Lipper,L.,Asfaw,S.,Cattaneo,A.,2014.Adoptionand intensityofadoptionofconservationfarmingpracticesinZambia.Agric. Ecosyst.Environ.187,72–86.

Borenstein,M.,Hedges,L.V.,Higgins,J.P.T.,Rothstein,H.R.,2009.Introductionto Meta-Analysis.Wiley.

Brouder,S.M.,Gomez-Macpherson,H.,2014.Theimpactofconservationagriculture onsmallholderagriculturalyields:ascopingreviewoftheevidence.Agric. Ecosyst.Environ.187,11–32.

Cairns,J.E.,Sonder,K.,Zaidi,P.H.,Verhulst,N.,Mahuku,G.,Babu,R.,Nair,S.K.,Das,B., Govaerts,B.,Vinayan,M.T.,Rashid,Z.,Noor,J.J.,Devi,P.,SanVicente,F.,Prasanna, B.M.,2012.Maizeproductioninachangingclimate:impacts,adaptation,and mitigationstrategies.In:Sparks,D.(Ed.),AdvancesinAgronomy,vol.114. AcademicPress,Burlington,pp.1–58.

Cairns,J.E.,Hellin,J.,Sonder,K.,Araus,J.L.,MacRobert,J.F.,Thierfelder,C.,Prasanna, B.,2013.Adaptingmaizeproductiontoclimatechangeinsub-SaharanAfrica. FoodSecur.5,345–360.

Cary,W.J.,Wilkinson,R.L.,1997.Perceivedprofitabilityandfarmers'conservation behaviour.J.Agric.Econ.48,13–21.

Challinor,A.,Wheeler,T.,Garforth,C.,Craufurd,P.,Kassam,A.,2007.Assessingthe vulnerabilityoffoodcropsystemsinAfricatoclimatechange.Clim.Change83, 381–399.

Chivenge,P.P.,Murwira,H.K.,Giller,K.E.,Mapfumo,P.,Six,J.,2007.Long-term impactofreducedtillageandresiduemanagementonsoilcarbonstabilization: Implicationsforconservationagricultureoncontrastingsoils.SoilTillageRes. 94,328–337.

CIMMYT,1988.FromAgronomicDatatoFarmerRecommendations:anEconomics TrainingManual.CIMMYT,Mexico(completelyrevisededition).

DepartmentoftheSurveyorGeneral,1984.MapofZimbabwe(1:1000000): NaturalFarmingRegionsandFarmingAreas.GovernmentPrinters,Harare. DerSimonian,R.,Laird,N.,1986.Meta-analysisinclinicaltrials.Control.Clin.Trials7,

177–188.

Dowswell,C.R.,Paliwal,R.L.,Cantrell,R.P., 1996.MaizeintheThirdWorld.Westview Press,Colorado,USA.

Egger,M.,Smith,G.D.,Phillips,A.N.,1997.Meta-analysis:principlesandprocedures. BMJ315,1533–1537.

Elwell,H.A.,Stocking,M.,1988.Lossofsoilnutrientsbysheeterosionisamajor hiddenfarmingcost.ZimbabweSci.News22,79–82.

Fan,R.,Zhang,X.,Yang,X.,Liang,A.,Jia,S.,Chen,X.,2013.Effectsoftillage managementoninfiltrationandpreferentialflowinablacksoil,Northeast China.Chin.Geog.Sci.23,312–320.

FAO,2002.Conserationagriculture:CasestudiesinLatinAmericaandAfrica.FAO SoilsBulle-tin.FAO,Rome.

FAO,2011.WhatisConservationAgriculture?FAO,Rome.http://www.fao.org/ag/ ca/1a.html.

Farooq,M.,Flower,K.C.,Jabran,K.,Wahid,A.,Siddique,K.H.M.,2011.Cropyieldand weedmanagementinrainfedconservationagriculture.SoilTillageRes.117,172–

183.

Friedrich,T.,Derpsch,R.,Kassam,A.,2012.OverviewoftheGlobalSpreadof ConservationAgriculture.FieldActionsScienceReports SpecialIssue6. Giller,K.E.,2001.NitrogenFixationInTropicalCroppingSystems.CABIPublishing,

NewYork.

Giller,K.E.,Witter,E.,Corbeels,M.,Tittonell,P.,2009.Conservationagricultureand smallholderfarminginAfrica:theheretics’view.FieldCropRes.114,23–34.

Figure

Fig. 2. Pooled and natural region-specific weighted mean differences (WMD) in yield a between direct seeder and conventional tillage, and b between ripper and conventional tillage, in Zimbabwe
Fig. 2. Pooled and natural region-specific weighted mean differences (WMD) in yield a between direct seeder and conventional tillage, and b between ripper and conventional tillage, in Zimbabwe p.6
Fig. 3. The weighted mean difference (WMD) in yield between planting basins and conventional tillage for the drier southern part of Zimbabwe
Fig. 3. The weighted mean difference (WMD) in yield between planting basins and conventional tillage for the drier southern part of Zimbabwe p.6
Fig. 4. Weighted mean differences in yield between direct seeder and conventional tillage on the left; and ripper and conventional tillage on the right in time across natural regions in Zimbabwe
Fig. 4. Weighted mean differences in yield between direct seeder and conventional tillage on the left; and ripper and conventional tillage on the right in time across natural regions in Zimbabwe p.7
Table 3 illustrates the details of how NPV was computed using the case of Natural Region II and Table 4 gives the summarised results for all natural regions

Table 3

illustrates the details of how NPV was computed using the case of Natural Region II and Table 4 gives the summarised results for all natural regions p.10
Fig. 7. Ripper breakeven labour savings across natural regions II, III and IV in Zimbabwe
Fig. 7. Ripper breakeven labour savings across natural regions II, III and IV in Zimbabwe p.11
Fig. 6. Direct seeder breakeven maize yield gain across natural regions II, III and IV in Zimbabwe
Fig. 6. Direct seeder breakeven maize yield gain across natural regions II, III and IV in Zimbabwe p.11
Fig. 5. Ripper breakeven maize yield gains across natural regions II, III, and IV in Zimbabwe
Fig. 5. Ripper breakeven maize yield gains across natural regions II, III, and IV in Zimbabwe p.11

References

Updating...