of weed cover ﬂ uence Net ecosystem CO exchange in an irrigated olive orchard of SE Spain:In Agriculture, Ecosystems and Environment

(1)

Net ecosystem CO

2

exchange in an irrigated olive orchard of SE Spain:

In ﬂuence of weed cover

Sonia Chamizo

^a,

*, Penélope Serrano-Ortiz

^b,c

, Ana López-Ballesteros

^d,c

,

Enrique P. Sánchez-Cañete

^e,c

, José Luis Vicente-Vicente

^f

, Andrew S. Kowalski

^a,c

aDepartamentodeFísicaAplicada,UniversidaddeGranada,18071Granada,Spain

bDepartamentodeEcología,UniversidaddeGranada,18071Granada,Spain

cInstitutoInteruniversitariodeInvestigacióndelSistemaTierraenAndalucía,CentroAndaluzdeMedioAmbiente(IISTA–CEAMA),18071Granada,Spain

dEstaciónExperimentaldeZonasÁridas(EEZA–CSIC),04120Almería,Spain

eBiosphere2,UniversityofArizona,85623AZ,USA

fDepartamentodeBiologíaAnimal,BiologíaVegetalyEcología,UniversidaddeJaén,23071Jaén,Spain

ARTICLE INFO

Articlehistory:

Received3August2016

Receivedinrevisedform14November2016 Accepted11January2017

Availableonlinexxx

Keywords:

Olivetree Carbonuptake Conservationagriculture Eddycovariance Sustainablemanagement Mediterraneanclimate

ABSTRACT

No-tillmanagementandtheestablishmentofplantcoverhavebeenimplementedinolivecropsinrecent yearsinordertopreventsoilerosionandincreasesoilorganiccarbon.However,theeffectofthese conservationpracticesonthenetCO2exchangeattheecosystemscalehasnotbeenexploredsofar.In thisstudy,weanalyzetheinﬂuenceofresidentvegetationcover(hereafterweeds)onthenetecosystem CO2exchange(NEE)inanirrigatedoliveorchardlocatedinJaén(SESpain)byusingtheeddycovariance technique.NEEwasmeasuredintheoliveorchardundertwotreatments,onewithweedcoverinthe alleysfromautumntospring,andanotherwhereweedgrowthwasavoidedbytheapplicationofa glyphosateherbicide.Ourstudydemonstratesthatthepresenceofweedsinthealleysincreasedcarbon assimilationintheweed-covertreatmentduringtheweedgrowingperiod(fromDecembertoApril).

However,thenetecosystemCO2uptakedecreasedintheweed-covertreatmentduringlatespring(May andJune),afterweedswerecutandleftonthesoil,comparedtotheweed-freetreatment,probablydue toanincreaseinsoilrespiration.Onanannualbasis,weedremovaldecreasednetcarbonuptakeby50%

compared totheweed-covertreatment. The annual NEE was 140gCm²y¹ in theweed-cover treatmentand70gCm²y¹intheweed-freetreatment.Insummary,ourstudydemonstratesthat, duringtheﬁrstyearofdifferentialtreatment,maintenanceofweedcoverinolivegroveshasapositive effectonCO2uptakeandenhancesthecapacityoftheagro-systemtoactasanetCO2sink.

1.Introduction

Soilcultivationandanthropogenicclimatechangehavecaused a greatimpactontheglobalsoil carbon(C)cycleover thelast century.Inadequatemanagementofagriculturallandhasledto accelerated ratesof soil erosionand hasexposed trapped C to decomposition,acceleratingmineralizationofsoilorganiccarbon (SOC;Lal,2004).Asaconsequence,thesepracticeshavemodiﬁed gains and losses of soil C, altering the natural C balance and increasing greenhouse gas emissions (Aguilera et al., 2015;

Amundsonetal.,2015).SomeestimatespointtoglobalSOClosses byagriculturalerosionof404TgCy¹(Doetterletal.,2012)andto globalCreleasestotheatmosphereassociatedwitherosionthat

rangefrom0.8to1.2GtCy¹(Lal,2003).TheseCemissionsare equivalentto12%ofglobalCemissionsbyfossilfuelsandindustry (9.80GtCin2014;LeQuéréetal.,2015).Therefore,theapplication of sustainable practices aimed to increase C sequestration in agriculturehasbecomearelevantsubjectofinterest.Thiscanbe especiallyimportantinSpain,whereSOCcontentslowerthan1%

are frequent, mostly in southern areas and agricultural soils (RodríguezMartínetal.,2016).

Olive trees(OleaeuropaeaL.) areoneof themostimportant crops in the Mediterranean basin, where they cover around 9.5Mhaandaccountfor98%oftheworld’solivecultivationarea (Repullo-Ruibérriz de Torres et al., 2012). The largest area dedicated to this crop is found in Spain, where it occupies 2.6Mhaandrepresents72%ofworld’soliveproduction(datafor 2013–2014; IOOC, 2015). Around 60% (1.5Mha) of the olive cultivation in Spain is located in Andalusia (southern Iberian

* Correspondingauthor.

E-mailaddress:[email protected](S.Chamizo).

ContentslistsavailableatScienceDirect

Agriculture, Ecosystems and Environment

j o u r n a l h o m e p a g e : w w w . e l s e vi e r . c o m / l o c a t e/ a g e e

(2)

Peninsula; MAGRAMA, 2012). Thus, olive groves represent an importantagriculturalsystem inthis regionduetoitsenviron- mental,socialandeconomicbeneﬁts.However,olivegrovesare subject to several environmental problems due to inadequate conventionalsoil-managementpracticessuchasintensivetillage andovergrazing,whichhavecausedhighrunoffanderosionrates, highsoilandSOClosses,andthelossofsoilfertility(Álvarezetal., 2007;FranciaMartínezetal.,2006;Martinez-Menaetal.,2008;

Gómezetal.,2009).Inordertomitigatetheseproblems,research hasbeencarriedoutinrecentdecadestoimprovesoilmanage- mentpractices,andpreventthemineralizationoforganicmatter andthelossofsoilstructureandfertility(FAO,2004).

Oneofthemostwidespreadconservationpracticesappliedin olive-groveplantationshasbeenthemaintenanceofspontaneous residentvegetation cover(hereafter“weeds”)inthealleysfrom autumntospring(Marquez-Garciaetal.,2013;Nietoetal.,2013).

Weedcovers,in additiontoprotecting thesoilagainst erosion, offer a number of well-known beneﬁts for soil properties:

improvement of soil physicochemical properties (Ramos et al., 2010);increasesintheinterceptionandstorageofrainfallwater,as wellasin soilwatercontentand wateravailabilityindeepsoil (Celanoetal.,2011;Paleseetal.,2014);increasesinatmosphericC ﬁxationand SOC content,therebyimproving soil structure and fertility(Hernándezetal.,2005;Castroetal.,2008;Gómezetal., 2009;Repullo-Ruibérriz de Torreset al., 2012; Marquez-Garcia etal.,2013;Sorianoet al.,2014;Herencia,2015);andincreased

biodiversity (Plaza-Bonilla et al., 2015). In this regard, some estimatespointtoSOCincreasesbetween44%and85%intopsoil (0–15cm)inolivegrovesafter100yearsofcovercropmanage- ment(Nietoetal.,2013),andpreliminaryestimationssuggestan increaseinsoilCsequestrationofaround1tonCha¹y¹inolive orchardsunderMediterraneanconditionsduetotheadoptionof plant covers (Vicente-Vicente et al., 2016). Thus, agricultural systemscanfunctionasCsinksifadequatemanagementpractices areapplied.

Althoughnumerousstudieshaveexaminedtheeffectofweed coveronsoilpropertiesandSOCcontent,littleresearchhasbeen focusedontheireffectonsoilCO₂fluxesorhowtheyaffectthe ecosystemCbalanceinoliveorchards.Indeed,fewstudieshave reportedinformationonCO2fluxesfromolivegrovesorquantified the ecosystem C uptake accounting for total CO2 inputs and outputs(seeTestietal.,2008;Nardinoetal.,2013).Sofar,mostCO2

exchange measurements have been conducted at the tree (Villalobosetal.,2012; Pérez-Priegoetal.,2010)and soillevels (Bertollaetal.,2014)byusingchambers,andsoilCO2emissions havebeenalsoestimatedviamodellingapproaches(Nietoetal., 2010).Intheabsenceofweedcover,netecosystemCO2exchange (NEE)fromolivegroveswillresultfromthebalancebetweenCO2

inputsbytreephotosynthesisand CO2 outputsbyaboveground autotrophic respiration (olive leaves, trunks and branches), belowgroundautotrophicrespiration(oliveroots)andheterotro- phic soil respiration. However, in the presence of weeds, it is

Fig.1.Locationoftheoliveorchardandpictureoftheeddytowerinstalledateachtreatmentconsideredatthissite:maintenanceofspontaneousweeds(weedcover)and weedremovalbyapplicationofanherbicide(weedfree).Pointsindicatethelocationoftheeddycovariancetowers.Thecoloredareaindicatesthefetchforeachtreatment.

(3)

necessarytoaccountforCO2uptakeviaweedphotosynthesisand CO2 emissionsvia weed and weed-covered soil respiration for quantiﬁcation of NEE. Knowledge of how conservation versus traditionalpracticesmayaffectthenetCO2uptakeinolivegrovesis lackingandthisinformationisnecessarytoelucidatetherolethat thesepracticesplayin Csequestrationandthus,theirpotential regardingclimatechangemitigation.

Non-destructive, ecosystem-scale and long-term measure- mentsofNEEarepossiblethankstothetechnologicaldevelopment of robust tools such as the eddy covariance (EC) technique (Dabberdtetal.,1993;Baldocchi,2003).Whilethistechniquehas been used to characterize CO₂ and water vapor exchanges in natural(Baldocchietal.,2001;Reichsteinetal.,2007)aswellas agriculturalecosystemsunderdifferingmanagement(Bakerand Griffis, 2005; Chi et al., 2016), its application to agricultural systemssuchasoliveorchardsispracticallynon-existent.Some reasonsfortheabsenceofinformationonthesewidespreadcrops intheMediterraneanregionare:i)thesteepslopeswherethese cropsareusuallylocated,whichcomplicatetheimplementationof these micrometeorological techniques; and ii) the intensive managementincludingirrigation,fertilizationandpruning,which reducesstressforwater,nutrientsorlight,andstronglymodifies CO2exchangescomparedtootherMediterraneanecosystemsor rainfedcrops(Testietal.,2008;Nardinoetal.,2013).Therefore, quantificationofCO₂exchange inolivegrovesattheecosystem scale is necessary tounderstand how they contributetothe C balanceandhowdifferentmanagementpracticescanamplifyor diminishtheircapacitytoactassinksofCO2.Toourknowledge, onlyafewstudieshavemeasuredNEEinolivegrovesatecosystem scale using the eddy covariance technique (Testi et al., 2008;

Nardinoet al.,2013; López-Bernal etal., 2015).However, these studies wereconducted either duringshort time periods or in youngoliveorchards,andnoneanalyzedtheinﬂuenceofno-till practicessuchasmaintenanceofplantcoverontheecosystemC uptake.

In this study, we measure NEE in an irrigated mature olive orchardofSESpainundertwomanagementregimes,maintenance ofweedcoverandweedsuppression,usingtheeddycovariance technique. The objective of this study was two-fold: i) to characterizemonthlyandannualpatternsofNEEinanirrigated, matureoliveorchard;andii)toanalyzetheeffectofweedcoveron theecosystemCuptakeinolivegrovesascomparedtomanage- mentforweedsuppression.

2.Materialsandmethods 2.1.Studysite

This research has been conducted in “Cortijo Guadiana” (3754⁰39.30⁰⁰N, 313⁰42.40⁰W), anirrigatedolive(Olea europaea L.“Arbequina”)orchardinÚbeda (Jaén,Spain),which belongsto theoilgroup“CastillodeCanena,S.L.”(Fig.1).Thesiteissituatedat 370mabovesealevel.TheclimateisMediterranean,withamean annual temperature of 16C, a mean annual precipitation of 495mm,andameanannualpotentialevapotranspiration(calcu- latedusingthePenman-Monteithequation)of1220mm(from15- year records at the Agroclimatic Station of Úbeda, Junta de Andalucía, http://www.juntadeandalucia.es/agriculturaypesca/

ifapa/ria/servlet/FrontController).Predominantwindscomefrom thenorthwestduringthedayandfromthesouthandsoutheastat night. The farmland has a total extension of 1500ha, but our experimentwasdevelopedinaﬂatarea,wheretwohomogeneous plotsweredelimitedof29.3ha(weedcover)and20.2ha(weed free).Soilorganicmattercontentis2.9%from0to5cmand2.4%

from5to15cm.Soiltextureisclayloam,with24%sand,32%silt and 44% clay. Trees are irrigated bydrip 3 timesa week from

FebruarytoOctober,atarateof32Lh¹pertreefor8h(atnight).

Withinirrigation,40gofNPKfertilizerpertreeisappliedtogether withwater(0.156gNPKL¹water,everyirrigationnight).Theolive treesare80yearsoldwitha77mspacingbetweenthem(204 treesha¹)andtreeheightisapproximately4m.ThePlantArea Index(PAI)of treeswasdeterminedfromtheindirect measure- mentofthegapfractionusingupwardhemisphericimagestaken with a 4.5mm F2.8 EX DC HSM circular ﬁsheye lens (Sigma CorporationofAmerica).Imageswereprocessedwiththesoftware CAN-EYEv6.1(INRA-CSE,Avignon).PAIofthetrees(correctedby clumpingeffect)was8.130.83m²vegetation/m²groundsurface.

In thetwo areas selected in the olive orchard (Fig.1), two treatments were applied: 1) weed-free treatment, in which a glyphosate-based herbicide was applied to avoid spontaneous weed growth (September 2014), and 2) weed-cover treatment, whichisthemanagementcommonlyappliedintheorchardand consistsofmaintenanceofspontaneousweedcoverinthealleys from autumn to spring. In spring (29–30 April), weeds were mechanicallywhackedandleftonthesurfacetoavoidexcessive waterconsumptionandcompetitionforwaterwithtrees.

2.2.Eddycovarianceandmeteorologicalandsoilmeasurements

Duringthehydrologicalyear2014(October)–2015(September), ﬂuxes of CO₂ and latent(LE) and sensible (H) heat have been determinedfromfast-response(10Hz)instrumentsmountedatop 10m-towers, one in each treatment (Fig.1). The towers were placed in thecenter ofeach treatmentand separatedbyabout 500mtoavoidinterferencefromonetreatmenttoanother.Wind vector components and sonic temperature were measured by three-axissonicanemometers(CSAT-3,CampbellScientiﬁc,Logan, UT,USA;hereafterCSI),whiledensitiesofCO2andH2O,together withtemperatureandpressure,weremeasuredbyenclosedpath infraredgasanalyzers(IRGA,Li-Cor7200;Lincoln,NE,USA).The stainless steel intaketubes are 1min lengthand haveoutside diametersof6.35mm.Flowratesare15Lmin¹andpassthrough 2

m

^mﬁltersthatreducedustenteringthegasanalyzeropticalcell.

Calibrations oftheIRGAs weredoneeverysixmonthsusingan ultra-highpurityN2zerogas,anda500ppmCO2spangas(inN2).

High-speed (10-Hz) mixing ratios of CO2 and water vapor (calculatedfromtheIRGAs measurements),windvectorcompo- nentsandsonictemperatureswereregisteredinLI-7550Analyzer InterfaceUnits.

At each treatment, additional instrumentation measures environmental and soil states. Air temperature and humidity weremeasuredat6mbyathermo-hygrometer(HC2S3,Rotronic AG,Bassersdorf,Switzerland),fromwhichvaporpressuredeficit (VPDs) was calculated. Incoming and outgoing short-wave and long-waveradiationcomponentsweremeasuredbya4-compo- nent radiometer (CNR-4, Kipp & Zonen, Delft, Netherlands), installed at2mfromthemast ata heightof7m,allowing the determination of net radiation (Rn) and albedo. Incident and reflected photosynthetic photon flux densities (PPFDs) were measured usingphotodiodes at7m(Li-190, Li-Cor,Lincoln,NE, USA). To monitorthe temporal evolution of soil moisture, soil watercontent(SWC)wasmeasuredinanalleyofeachtreatment usingtwosoilmoistureprobesinstalledat0.10mdepth(CS616, CSI).Oneachtreatment,twothermocouples(TCAV,CSI)measured soiltemperatures at0.04msoil depthand twoheatflux plates (HFP01,Hukseflux,Delft,theNetherlands)wereinsertedat0.08m.

Environmental and soil measurements were stored as 30min averagesbyadatalogger(CR3000,CSI).Finally,precipitationdata wereobtainedfromtheÚbedaAgroclimaticStationoftheJuntade Andalucía (http://www.juntadeandalucia.es/agriculturaypesca/

ifapa/ria/servlet/FrontController)locatedat7kmfromourstudy site.

(4)

2.3.Eddydataprocessingandstatisticalanalysis

FluxesofCO2(NEE)andLEandHfluxeswerecalculatedonhalf- hourbasesusingtheEddyPro5.2.0software(LI-CORInc.,Lincoln, Nebraska, USA). Raw 10-Hz data were filtered for spikes and compensationfortimelagsbetweentheairsamplingpointandthe analyzerwasdonebymaximizingthecorrelationbetweenvertical windspeedandmixingratiosofCO2andwatervapor.Half-hour covariancesbetweentheverticalwindcomponentandCO₂,water vapor and sonic temperature were calculated using block averaging, double coordinate rotations and spectral corrections for high frequency range (Moncrieff et al.,1997). Without the spectralcorrection,theCO2fluxeswere,onaverageforthewhole studyperiod, 7%and8%lessfor theweed-coverand weed-free treatments,respectively,thanthecorrectedCO2fluxes.

Duetohighpowerrequirementsbytheairpump,thesystem suffered frequent energy losses that caused data gaps, mainly duringnighttime. During Mayand September,continuous data losseswerefoundfrom4amto7amintheweed-freetreatment due to energy loss. Nighttime fluxes measured during weak turbulencewererejectedbyfilteringwithafrictionvelocity(u*) below0.15ms¹(Reichsteinetal.,2005).Inaddition,dataquality checkofthehalf-hourlyNEE,H2Ofluxandsensibleheatflux(H) wasappliedbyfilteringaccordingtothefollowingparameters:1) ForCO₂:i)qualityofdata=0or1(MauderandFoken,2004);ii)CO₂ variance<50ppm²;iii)12<pitch<12;iv)4<skewness<4;

v)Kurtosis<10; 2)For H2O: i) quality ofdata=0 or 1;ii) H2O variance<0.5ppt²;iii)10<pitch<10;iv)4<skewness<4;

v) Kurtosis<9; and for H: i) quality of data=0 or 1; ii) H variance<2(Wm²)²; iii)10<pitch<10; iv)4<skewness

<4;v)Kurtosis<9.Datagapsdue toenvironmentalconditions, instrumentmalfunctionandnighttimelowturbulenceledtoadata coverageof41%intheweed-covertreatment(69%duringdaytime and22%duringnighttime),and38%intheweed-freetreatment (62%duringdaytimeand22%duringnighttime).Datalossesare frequentin eddycovariancestudies (datagapaverage35%, see Falgeetal.,2001).Despitehighdatagapsinoursite,mostdata losses occurred during night when GPP is absent and Reco is generallylow,and low frictionvelocities (u*<0.15) lead tonot validCO2fluxes duringmanynighttimeperiods.Thus,frequent datalossesduringnighttimeatoursitelikelyhadlittleinfluence onmonthlyandannualCO2budgets.Datagapswerefilledusing the marginal distribution sampling technique described by Reichsteinetal.(2005).Thistechniquealsocalculatesuncertain- tiesforactualmeasurementsbysimulatinggapsandapplyingthe gap-fillingprocedure.Twicethestandarddeviationofsumsofthe 30-minuncertaintiesderivedfromthegap-fillingprocedurewas consideredas our NEE errorfor thedifferent time periods we analyzed (monthly and annual NEE). Positive values of NEE indicatenetCO2releasetotheatmosphere,whilenegativevalues representnetCO₂uptake.Half-hourNEEvalueswereintegratedto obtainCexchange(gCm²)atdaily,monthlyandannualscales.

An estimation of the flux footprint during daytime and nighttimeperiodswas determinedusingthe methoddescribed byKljunet al.(2004) toverifythatfluxesoriginatedfromwell withinthe fetch(higher than 200mfromthe tower). Daytime periods were defined when net radiation was higher than 10Wm².

Flux partitioning into Gross Primary Production (GPP) and Ecosystem Respiration (Reco) was performed according to the method by Reichstein et al. (2005) and Lasslop et al. (2010).

However,unexpectedseasonal behavior and unreliable estima- tionsofannualGPPandRecowereobtainedforbothtreatments, suggesting these partitioning methods are not suitable for applicationatoursite.Forthisreason,thelight-responsecurves wereused tomodel GPP and Reco.The rectangular hyperbolic

light-responsefunction(Falgeetal.,2001)wasappliedtomonthly averagesof30-mindaytimedata,andmonthlyparameterization coefﬁcientswereobtainedaccordingtotheequation:

NEE¼b1PPFD

b2þPPFD þb3 ð1Þ

wherePPFDistheincidentphotosyntheticphotonﬂuxdensity,the coefﬁcientb1isthemaximumgrossprimaryproduction(GPPmax,

m

^mol^CO²^m²^s¹^);^b²^represents^the^level^of^PPFD^for^which^GPP

ishalfofGPPmax(

m

^mol^photons^m²^s¹^);^and ^the^parameter^b3

represents the ecosystem respiration (Reco,

m

^mol^CO²^m²^s¹^).

For determination of the parameterization coefficients, only measured data (quality=0 or 1) until noon was considered in ordertoaccountfortheeffectofPPFDonNEEandavoidtheeffect ofhighVPDonNEE.Inordertofitthedatatothelight-response modeldescribedinequation1,wefirstlygeneratetheinitialvalues (uniform distribution) of the model coefficients randomly by delimiting realisticranges foreverycoefficient:0–50forb1,0– 2000

m

^mol^photons^m²^s¹^for^b²^and⁰–20

m

^mol^CO²^m²^s¹^for

b3.Theseinitialvaluesarenecessarytostartthefittingprocedure, which consists of calculating the nonlinear (weighted) least- squares estimates of the parameters of the non-linear model (Eq.(1)).WeusedRsoftwareversion3.1.3(RDevelopmentCore Team,2015)toperformthisanalysis.Coefficientswereconsidered significantwhenp<0.05.

Toassesstheaccuracyoftheeddycovariancemeasurements, weanalyzedlinearregressionsbetweenthesumoflatentheat(LE) andsensibleheat(H)versusnetradiation(Rn)minussoilheatflux (G,calculatedasthesumofthesoilheatfluxat0.08mandtheheat storageterm(Q)inthe0–0.08msoildepth):RnG=LE+H.We determinedtheenergybalanceclosureusing30-mintimeseriesof Rn,H,LEandGfortheperiodbetweenAprilandJune.Thisperiod wasselectedinordertoaccountfortheperiodofmaximumweed growth and the later period when weeds were cut, and also becausethereweresimultaneousmeasurementsofsoilheatflux, soiltemperature,andsoilwatercontentatbothtreatments.

2.4.SoilCO2efﬂuxmeasurements

In addition to the eddy covariance measurements, soil CO2

efﬂuxesweremeasuredincylindricalPVCcollars(10cmdiameter

5cm height) inserted into the soil in the alleys of the two treatments.Fivecollarswereinsertedpertreatmentandthesoil CO2effluxwasmeasuredatmidday,onceamonthfromMarchto July, with a manual chamber system model EGM-4/SRC-1 (PP- Systems,Hitchin,UK).Eachcollarwasmeasuredthreetimesand theaveragewasusedasthesoilCO2effluxoftheplot.Thefluxwas determinedfromtheslopeoftheCO2molarfraction(referencedto dryair)measuredevery5sduring120safterchamberclosureand was corrected for atmospheric pressure and the chamber air temperature.Significant differences (P<0.05)in soil CO2 efflux betweenthetwo treatments(weed-coveredsoil andweed-free soil) were analyzed using a one-way ANOVA. Analyses were conducted usingR softwareversion 3.1.3 (RDevelopment Core Team,2015).

2.5.Weedbiomassandweedorganiccarbondetermination

WeedsamplingwasconductedatthebeginningofApril(before weedwhacking)inordertoquantifyabovegroundweedbiomass andtheorganicCinputcontributedbyweedbiomass.Fivesquare plotsof0.5m0.5m(0.25m²)wereselectedandweedswerecut andharvestedfordeterminationofdryweight.OrganicCreleased byweedswas determinedusingtheWalkleyand Blackmethod modiﬁedbyMingoranceetal.(2007).Samplesof30mgofplant

(5)

materialwere weighedand 5mLof potassium dichromateand 7.5mLofsulfuric acidwereadded. Afterdigestionat155C for 30min,10mLof distilledwaterwasadded andabsorbancewas measuredat600nminaspectrophotometer.TheorganicCcontent wasdeterminedfromthecalibrationcurvebuiltwithincreasing concentrationsofglucose.

2.6.Cropproductivityquantiﬁcation

Oliveharvestingwas carriedoutinDecember2015.Wooden sticksanda trunk-shakermachinewereusedtodislodgeolives from14treesselectedrandomlyateach treatment.Oliveswere collectedonnetsplacedonthegroundandthenweighed.Samples ofolivesweretransportedtothelaboratoryanddriedinanovenat 60C inordertodeterminethedryweight.Fromthisvalue,we calculated the average crop productivity for each treatment, expressedaskilogramsofolivespertree,aswellastheCexportby

oliveyieldingCm²byusingtherelation:1gdrymatter=0.4782 gofC(Paleseetal.,2013).

3.Results

3.1.Meteorologicalconditionsandsoilvariables

Meteorologicalconditionsandevolutionofsoilvariablesinthe twotreatmentsduringthestudyyearareshowninFig.2.Annual rainfallduringthestudyyearwas381mm,mainlyconcentrated fromNovembertoApril,andlowerthantheclimatologicalaverage forthissite(495mm;Fig.2a).Themeanannualtemperaturewas 17C,andthemaximumandminimumaveragedailytemperatures were 32.4C (in July) and 0.4C (in December) (Fig. 2b). The maximum and minimum averaged daily values of VPD were 42.5hPaand0.4hPa,recordedattheendofJuneandinDecember, respectively. PPFD was the highest during the dry season.

0.00.20.40.60.8

Alle y SWC (m

3

m

−3

)

020406080100

Rain (mm)

Weed−covered soil Weed−free soil Rain

a)

01020304050

T e mper ature (ºC)

Weed−covered soil Weed−free soil Air

b)

02004006008001000

Oct14 Nov14 Dec14 Jan14 Feb14 Mar14 Apr14 May15 Jun15 Jul15 Aug15 Sep15

PPFD (µmol m

−2

s

−1

)

PPFD VPD

c)

01020304050

VPD (hP a )

Date

Fig.2.Dailyaveragesofenvironmentalandsoilvariablesduringthehydrologicalyear:a)Soilwatercontent(SWC,m³m³;averageofthetwomoistureprobes)measuredin analleyat0.10mintheweed-coveredandweed-freesoil,andrain(mm)duringthestudiedperiod.b)Airtemperature(dataaverageofthetwotreatments)andsoil temperature(C)intheweed-coveredandweed-freesoil(averageofthetwothermocouples).c)Photosyntheticphotonﬂuxdensity(PPFD,mmolphotonsm²s¹)andvapor pressuredeﬁcit(VPD,hPa)(dataaverageofthetwotreatments).

(6)

MaximumaverageddailyPPFDwas888

m

^mol^photons^m²^s¹ⁱⁿ

Juneand minimumdailyvaluewas 30

m

^mol^photons^m²^s¹ⁱⁿ

December (Fig. 2c). There were large differences in alley SWC betweenthesoilswithandwithoutweedcovers(Fig.2a)(standard deviationofSWCateachtreatmentwasverylow,withaverage valuesforthewholeperiodof0.03andmaximumandminimum valuesof0.05and0.004,respectively).Duringwetperiods,SWC wasupto0.2m³m³higherintheweed-coverthanintheweed- freesoil.However,duringthedrysoilperiod,soilmoisturewas similarforbothsoils.Therewerealsomarkeddifferencesinsoil temperaturebetweentreatments(Fig.2b).FromOctobertoMarch, soil temperature was similar at both treatments and strongly coupledwithairtemperature.However,duringspring(April,May and June) and summer (July, August, September) months, the temperaturewashigherinthesoilwithnoweeds,reachingdaily averagesupto13C aboveairtemperatureand12C abovethe weed-coversoiltemperature.

3.2.Validityofeddymeasurements:ﬂuxfootprintanalysisandenergy balanceclosure

Thefootprintanalysisshowedthatupwinddistancescontrib- utingtothemeasuredCO2ﬂuxwereinallcaseswithinthefetchfor each treatment. The median of the x_90% (distance from anemometerdelimiting90%oftheﬂux)duringthestudiedperiod was164mintheweed-covertreatmentand172mintheweed-free treatmentatnight,andmuchlessduringdaytime.

Regardingtheenergybalanceclosure,resultsweresimilarat bothtreatments.Theclosuredeﬁcitwas 27%intheweed-cover treatmentand29%intheweed-freetreatment,withR²of0.90and 0.87, respectively.Theenergybalanceclosure improvedatboth treatments when only thedrier period fromMay to June was considered,withclosuredeﬁcitsof26%and23%andR²of0.91and 0.90attheweed-coverandweed-freetreatments,respectively.

−10−505

October

NEE (µmol m−2 s−1 )

November December January

−10−505

February

March April May

−10−505

0 4 8 12 16 20 24 June

0 4 8 12 16 20 24 July

0 4 8 12 16 20 24 August

Hour

0 4 8 12 16 20 24 Weed−cover Hay−cover Weed/hay−free September

Fig.3. AveragemonthlydiurnaltrendsinnetecosystemCO2exchange(NEE)inthetwotreatments.MonthlyaverageswerecalculatedfrommeasuredhourlydataofNEE.

(7)

3.3.TemporalvariabilityofNEEbetweentreatments

Forboth treatments,as expected,monthlydiurnal curvesof NEE showed positive values at night and increasingly negative valuesaftersunriseasincomingsolarradiationincreased,uptoa maximumafterwhichNEEincreases,thenreachingpositivevalues aftersunset(Fig.3).Inaddition,achangeisobservedthroughout theyearinthetimeofdaywhenthemaximumnetCO2uptake occurs.While thehighestvalues ofnetCO₂ uptake occurredat midday (12pm–1pm, solar hour) during autumn and winter, maximumCO2uptakeoccurredatearlierhoursinspring(10am– 11am,s.h.)andsummer(8am–9am,s.h.;Fig.3).

Thus,despiteirrigation,somecontrollingeffectsofVPDwere foundindiurnaltrendsofNEE.Fig.4showsmonthlydiurnaltrends ofPPFD,VPDandNEEatbothtreatmentsduringthegrowthperiod inMarchand thehot dryperiod in August.During thegrowth periodand underlow water stress(maximummonthlydiurnal VPDwas16hPa), NEEwasstronglycoupledwithlightintensity, and maximum netCO2 uptake coincided with maximum light intensity (maximum monthly diurnal PPFD was 1280

m

^mol

photons m²s¹; Fig. 4a and c). By contrast these variables showed lags during periods of high water stress (maximum monthlydiurnalVPDinAugustwas41hPa),whennetCO2uptake peakedseveralhoursbeforethetimeofmaximumlightintensity (maximummonthlydiurnal PPFD was 1630

m

^mol^photons ^m²

s¹; Fig.4b andd). ThisnetCO2 uptake peakusually occurred beforethetimeofmaximumVPDinbothperiods(lowandhigh water stress), but the delay between both was greater during periodsofhighwaterstress(Fig.4bandd).Itcanbealsoseenthat duringthegrowthperiod,netCO2uptakewasmuchhigherinthe weed-cover (maximum monthly diurnal net CO2 uptake was

9.6

m

^mol^m²^s¹⁾^thanⁱⁿ^the^weed-free^treatment^(maximum

monthlydiurnalnetCO2uptakewas4.7

m

^mol^m²^s¹^),^but^both

showedsimilarNEEduringAugustwhenweedshadbeenalready cut (maximum monthly diurnal net CO2 uptake was 3.5 and

4.5

m

^mol^m²^s¹^,respectively;Fig.4candd).

Throughouttheyear,importantdifferenceswereobservedin monthlyNEEbetweenthetwotreatments.ResultsshowthatNEE was similar in the two treatments in the initial conditions (October), when there were no weeds in either of the two treatments(Fig.3).Bothtreatmentsshowedpositivevaluesduring this month, indicating a net CO₂ emission to the atmosphere (Table1).Forthisperiod,dailyNEEvaluesrangedfrom0.69to 1.26gCm²andfrom1.07to1.07gCm²intheweed-coverand

0500100015002000

PPFD (µmol m

−2

s

−1

)

a) b)

0102030405060

VPD (hP a )

PPFD

VPD

0 4 8 12 16 20 24

−12−8−4024

c)

NEE (µmol m

−2

s

−1

)

Hour

0 4 8 12 16 20 24

d)

Weed−cover Weed−free

Fig.4.MonthlydiurnaltrendofPPFDandVPDduringMarch(a)andAugust(b)andmonthlydiurnaltrendofNEEinbothtreatmentsduringMarch(c)andAugust(d).

Monthlyaverageswerecalculatedfrommeasuredhalf-hourlydataofNEE.

Table1

MonthlyNEE(gap-ﬁlleddata)(anderror)foreachmonthatthetwotreatments.

MonthlyNEE(gCm²)

Weed-cover Weed-free

Oct 3.68(3.63) 4.59(3.92)

Nov 7.30(3.88) 5.17(2.94)

Dec 35.61(3.48) 5.51(2.74)

Jan 18.06(3.10) 17.74(1.80)

Feb 36.66(4.98) 9.07(2.26)

Mar 74.43(6.83) 28.09(3.61)

Apr 26.76(6.06) 19.40(4.90)

May 7.43(2.43) 14.85(4.18)

Jun 2.49(2.91) 21.84(5.78)

Jul 25.51(3.17) 23.26(4.87)

Aug 6.82(3.00) 6.91(4.02)

Sep 12.71(3.52) 16.16(4.26)

(8)

weed-freetreatment,withdailyaveragesof0.12and0.15gCm², respectively.

FromNovembertoApril,negativemonthlyvaluesofNEEwere foundat both treatments, indicating net CO2 uptake (Table 1).

However,asweedsgrew,Cuptakewasmuchhigherintheweed- coverthantheweed-freetreatment(Fig.3),withthemaximum differencein March.During thismonth, NEEintheweed-cover treatmentwas upto15.5

m

^mol^m²^s¹^and ^daily ^NEE ^ranged

from3.49to0.06gCm²,withanaveragevalueof2.40gCm², whileintheweed-freetreatment,NEEwasupto11.0

m

^mol^m²

s¹ and daily NEE ranged from 2.44 to 0.10gCm², with an averagevalueof0.91gCm².

In April,weedsreachedtheirmaximumsize.Averageabove- ground weed biomass was 22058gm², of which 36% was organicCcontent(79.4gOCm²).Duringthismonth,althoughnet CO2 uptake was still higher in theweed-cover treatment, NEE valuesbecamemoresimilaratbothtreatments(Fig.3).DailyNEE rangedfrom2.92to1.20gCm²intheweed-covertreatment, withanaverageof 0.89gCm²,while dailyNEE rangedfrom

1.49to0.50gCm²intheweed-freetreatment,withanaverage of0.65gCm².ThisperiodcoincidedwiththehighestsoilCO2

efﬂuxrecordedinthesoilcoveredbyweeds atmidday(Fig.5).

FromMarchtoMay,thesoilCO2effluxwassignificantlyhigherin theweed-coveredthanweed-freesoil.However,differenceswere especiallymarkedinApril,whenthesoilCO₂effluxintheweed- coveredsoilwasupto5.6timeshigherthanintheweed-freesoil.

AttheendofApril,weedswerecutandleftonthesoiltoallow weedresidues(hereafter,‘hay’)todecomposeandincorporateinto thesoil.FromMaytoJune,contrarytothepatternobservedduring the weed growth period, net CO2 assimilation in the hay-free treatmentsurpassed that observed in the hay-cover treatment (Fig.3).NegativemonthlyvaluesofNEE (netCO2 uptake)were obtained for the hay-free treatment, whereas the hay-cover showed positive values (net CO2 emission to the atmosphere;

Table 1). Daily average NEE in the hay-cover and hay-free treatments were, respectively, 0.24 and 0.48gCm² in May, and0.08and0.73gCm²inJune.

During thesummer months(July toSeptember), both treat- mentsshowedsimilarandpositivemonthlyvaluesofNEE(Table1).

AveragedailyvaluesinJuly,AugustandSeptemberwere0.82,0.22 and0.42gCm²inthehay-covertreatment, and0.75,0.22 and 0.54gCm² in the hay-free treatment. Soil respiration rates measuredduringthedryseason(JuneandJuly)werelow(Fig.5) andsoilsbothwithandwithouthayshowedsimilarrespiration rates(averagesoilrespirationratewas0.870.17

m

^mol^m²^s¹ⁱⁿ

thesoilcoveredbyhay,and0.780.02

m

^mol^m²^s¹ⁱⁿ^the^bare

soil).

3.4.Functionalrelationshipsbetweenenvironmentalvariablesand NEE

The light-response curves showed a significant relationship betweenNEEandPPFDduringwinterandearlyspring(Table2), whereas no significant relationship was found during the dry period. Consequently, significant parameterized coefficients of monthlyGPPmaxandRecoweregenerallyobtainedfromthelight- response equation forboth treatments duringwinterand early spring(Table2),butnosignificantvalueswereobtainedforeither ofthemduringsummer.ModeledvaluesofRecowerehigherinthe weed-coverthanintheweed-freetreatment(withtheexceptionof February, where modeled Reco was unexpectedly higher in the latter),and modeledGPPmaxwas upto4.3 times higherin the weed-coverthanintheweed-freetreatment(maximummodeled GPPmaxwas28.30and15.15

m

^mol^m²^s¹^,respectively).Also,a betterrelationshipbetweenNEEandPPFDwasfoundduringthe weedgrowthperiod(fromDecembertoMarch)intheweed-cover treatmentcompared totheweed-free treatment.Concretely, in MarchwhenGPPmaxwashighestintheweed-covertreatment,a betterﬁtwasfoundinthistreatment(R²=0.98)comparedtothe weed-free treatment (R²=0.71). For the same PPFD level (1216

m

^mol ^photons ^m²^s¹ ^at ^noon), ^maximum ^net ^CO²

assimilationwasdoubleintheweed-coverthatoftheweed-free treatment(Fig.6).

Nevertheless, a worse ﬁt was observed for the weed-cover treatmentduringAprilandMay,whenweedswerecutandnetCO2

fixation decreased in the hay-cover treatment. During these months,significantvaluesofGPPmaxwereobtainedinthehay-free treatment,butnosignificantvaluesofGPPmaxorRecowerefoundin thehay-covertreatment.

Contrarytoexpectations,nosigniﬁcantrelationshipwasfound betweennighttimeNEEandtemperatureeitheratdailyormonthly scales.NighttimeNEEwaslowatbothtreatmentsduringthestudy period,withaveragesof0.77and0.85gCm²night¹intheweed- cover and weed-free treatments, respectively. Nonetheless, we couldobservesomeseasonaltrendsinnighttimeNEErelatedto temperaturefor both treatments.Nighttime NEEwas low from December to March (average nighttime NEE was 0.29 and 0.42gCm²night¹intheweed-coverandweed-freetreatments, respectively), coinciding with periods of low air temperature (averagenighttimetemperaturewas5.4C,andrangedfrom2.5C inJanuaryto9.0CinMarch),whilehighervaluesofnighttimeNEE werefoundfromApriltoNovember(averagenighttimeNEEwas 1.01 and 1.11gCm² night¹in theweed-cover and weed-free treatments, respectively), coinciding withperiods of higher air temperatures (average nighttime temperature was 18.5C, and rangedfrom11.2CinNovemberto24.8CinJuly).

3.5.AnnualNEEandoliveproductivitybetweentreatments

Although higher net CO2 emissions were found during late springintheweed-covertreatmentcomparedtotheweed-free treatment,apositiveeffectofweedcoverwasfoundinannualnet ecosystemexchange.ThecumulativeNEEvaluesduringthestudy year(Fig.7)resultedinanannualNEEvalueof14020gCm²in theweed-coverand7010gCm²intheweed-freetreatment.

Thus, although the weed-free treatment acted as a net C sink during longerperiod (from NovembertoJune)than the weed- covertreatment(fromNovembertoApril),thehigherassimilation rateinthelatterduringtheweedgrowthperiodwasabletooffset thehigheremissionsfoundinthistreatmentduringlatespring.As aresult,annualnetecosystemCuptakewasreducedby50%inthe 0

2 4 6 8 10 12

Soil CO2efflux (µmol m-2s-1) ^Weed^{-free s}^oil_Weed_-co_{vered so}_il

7-April 12-May 3-June 30-July 5-March

a b

a

b b

a

a a a

Fig.5. SoilCO2efﬂux(meansd,n=5)measuredatmiddayinthealleysofthetwo treatments.

(9)

weed-freetreatment.Despitefrequentdatagapsduringthestudy year,uncertaintyintheestimationofannualcarbonbudgetsfor both treatments was low (14%), making differences between treatmentsnoteworthy.

Regardingproductivity,somedifferenceswerefoundbetween bothtreatmentsduringthestudiedyear.Onaverage,oliveyield (dry weight) was 34.2kg of olives per tree in the weed-cover treatmentand28.0kgofolivespertreeintheweed-freetreatment, thus indicating productivity was 22% higher in the former.

Accordingtotheseresults,theC exportbyoliveharvestingwas

estimated in 334gCm² in the weed-cover treatment and 273gCm²intheweed-freetreatment.

4.Discussion

Largedifferences inNEE wereobservedin theoliveorchard under the two treatments. Although plant covers are able to enhancesoilrespirationbyincreasingSOCcontentandmicrobial activity,alternatively, theycanincreaseCﬁxationthroughtheir photosyntheticactivity.Hence,wefoundthatthemaintenanceof spontaneous weeds from autumn to early spring strongly increased net C ﬁxation compared tothe weed-free treatment (Fig.3).InMarch,whennetCuptakeintheoliveorchardunder both managements was the highest, thetreatment with weed cover showed up to 2.7 times higher monthly NEE than the treatment without weed covers, with values of 74.43 and

28.09gCm² month¹, respectively (Table 1). Assuming that thedifferencebetweenthesevaluesrepresentsthenetCuptakeby weeds,theresultingvalueis46.34gCm²month¹,whichis1.7 timeshigherthanthenetCuptakebyolivetreesintheweed-free treatment(28.09gCm²month¹).Coincidingwiththis,Palese et al. (2013) reported that spontaneousvegetation (weeds and grasses)wasthemostimportantcropcomponentforCfixationin anirrigatedoliveorchardinsouthernItaly,contributingto35%of totalCO2fixation(therestbeingpruningmaterialandyield).This high C assimilation by weeds can be explained by their short lifetimeandtheneedforhigherefficiencyinCO₂uptaketoinvest in biomassgrowth, before thebeginningof thesenescence dry period.Inaddition,weedspeciesusedifferentwater-andlight-use strategiesthanolivetrees,whichcanexplaindifferencesinNEE trendsbetweentreatments.UnderincreasingPPFD,olivetreescan limittheirphotosyntheticactivitybyclosingstomatainresponse toincreasedwaterstress(Testietal.,2008;Villalobosetal.,2012).

Table2

CoefficientsforGPPmax(b1),Reco(b3)andPPFDwhenGPPmaxwashalf(b2)obtainedbyapplyingtheFalgeetal.(2001)equationusingmonthlyaveragesofdaytime30-min datauntilnoonforeachmonth.ThecoefficientofdeterminationoftherelationshipbetweenNEEandPPFDisalsoshown.Onlymonthswithatleastonesignificant parameterizationcoefficientareshown.

Estimate Standarderror pvalue Estimate Standarderror pvalue

December b1 19.3 2.9 p<0.001 11.3 6.6 0.138

b2 757.7 272.3 0.032 654.6 1012.1 0.542

b3 4.2 0.5 p<0.001 2.8 1.6 0.124

R² 0.99 0.83

January b1 17.2 3.7 0.006 8.7 1.8 0.005

b2 820.4 510.9 0.169 596.4 515.8 0.3

b3 3.7 1.1 0.019 2.5 1.1 0.075

R² 0.98 0.96

February b1 20.4 2.9 p<0.001 15.2 2.5 p<0.001

b2 812.4 279.5 0.023 150 75.2 0.093

b3 3.6 0.6 p<0.001 9.6 3.1 0.022

R² 0.99 0.98

March b1 28.3 3 0 6.6 4.4 0.174

b2 241.3 89.6 0.027 300.4 713.8 0.686

b3 15.4 4 0.005 0.9 6 0.882

R² 0.98 0.71

April b1 17.5 14.6 0.245 13.5 4.9 0.013

b2 192.1 357 0.596 256 254.2 0.326

b3 9 16.4 0.589 6.7 5.9 0.273

R² 0.53 0.75

May b1 10.9 48.7 0.825 7.3 1.9 0.001

b2 55.8 320.6 0.863 601.8 1071.7 0.581

b3 9.2 49.2 0.853 2.5 3.4 0.478

R² 0.35 0.48

R² = 0.98 R² = 0.71

-14 -10 -6 -2 2 6 10 14

0 200 400 600 800 1000 1200 1400 NEE (µmol m-2 s-1)

PPFD (µmol m^-2s^-1)

Fig.6. Light-responsecurvesinMarchforthetwotreatments,usingmonthly averagesofmeasuredhalf-hourlydaytimeNEE.Barsrepresentstandarddeviation ofmonthlyaverages.

(10)

In contrast, weeds are able to maintain their photosynthetic activityunderhighlightintensities,despiterelativelyhighairVPD (LongandHällgren,1993;Pérez-Priegoetal.,2015).Thisbehavior wasreﬂectedinthebetterrelationshipfoundbetweenNEEand PPFDintheweed-coverthantheweed-freetreatment(Fig.6).

Inouroliveorchard,despiteirrigation,theeffectoflightand VPD on diurnal trends of NEE was visible at both treatments (Fig.4).Duringthespringgrowthperiod,netCO₂uptakeincreased withincreasingPPFDuptoathreshold,coincidingwithmaximum light intensity, after which net CO2 assimilation decreased, coincidingwithmaximum VPD duringday(Fig.4a and c). The relationshipbetweenNEEandPPFDduringthisperiodwasbetter intheweed-coverthantheweed-freetreatment, attributednot onlytoCO2uptakebyolivetreesbutalsotohighCO2uptakeby weedsandtheirrapidresponsetoincreasingPPFD,ascomparedto theweed-free treatment, where the response of olivetrees to increasingPPFDissubjecttostomatalcontrol.Duringthesummer period(August),anincreaseddelaybetweentheCO2fixationpeak and maximum VPD was observed, and there was also a slight decouplingbetweenthenetCO2uptakepeakandmaximumPPFD (Fig.4bandd).Thisisindicativeofthemechanismsofstomatal controlusedbyolivetreesforreducingCO2fixationinorderto regulatewater losses by transpirationunder high water stress conditions(highVPD).Indeed,upto80%oftotalCuptakeoccurred beforemiddayduringsummermonths,whileonly44%occurred beforemiddayduringwintermonths.Consistentwithouranalysis, Testietal.(2008)reported60%oftotalCwasfixedbeforemidday insummerinanirrigatedoliveorchard,whilethisdecreasedto 40%inthecoolseason,whenVPDexertedaminoreffect.

The presenceof weedsnotonlyincreasesC uptake butalso signiﬁcantlyincreasessoilCO2efﬂux(Table2,Fig.5).Bertollaetal.

(2014)foundthatsoilrespirationwashigherinanoliveorchard withweedscomparedtothatwithoutweedsandestimatedannual emissionsduetorespirationof1179and784gCm²inthetwo treatments,respectively.Nonetheless,thehigherCO2efﬂuxfound inthesoilwithweedswasmorethanoffsetbyweedphotosyn- theticactivityduringthegrowthperiod,thusresultinginhigher annualnetCassimilationcomparedtothemanagementforweed suppression (Table 1). Modeled parameters using the light- responseequationalsoshowedthat bothGPP_maxand R_ecowere higher in the treatment withweed thanwithout weed covers (Table 2).These parametersweresigniﬁcant duringwinter and earlyspringbutnotduringsummer,probablybecauseoftheeffect ofhighVPDthatcouldmasktherelationshipbetweenNEEand PPFD.AerialandrootweedbiomasslargelycontributestosoilC

enrichment(Guzmánetal.,2014).Inthisstudy,abovegroundweed biomassrepresented79.4gOCm², which iscomparable tothe valuesreportedbyotherauthorswhohavefoundorganicCinputs byspontaneousvegetationbetween46.2and50.9gOCm²during yearsofnormalprecipitationregime(Repullo-RuibérrizdeTorres etal.,2012).TheCﬁxedbyweedsispartlyrespiredbacktothe atmospherebydecompositionofmorelabileC compounds,and partlyremainsinthesoilandisincorporatedasresistantorganic matter in the uppermost layer of the soil, contributing to increasing SOC (Hernández et al., 2005; Castro et al., 2008;

Sorianoetal.,2014).

Theapplicationofglyphosatetocontrolweedemergencewas expectedtohavelittleeffectonCO2ﬂuxesofolivetreesorbaresoil.

Previousstudieshaveshownnoeffectofglyphosateapplicationto weedsonthephotosyntheticactivityofyoungolivetrees(Cañero etal.,2011)oronthesoilmicrobial community(Weaveret al., 2007). However, some studies have shown that glyphosate increases microbial biomass-C, soil enzymatic activity, and microbial respiration (Zabaloy et al., 2008; Panettieri et al., 2013), as glyphosate could beused by microbesas a C source (Eseretal.,2007).Duetotherelativelyshortdegradationtimeand lowpersistenceofglyphosateinthesoil,itseffectsonthesoilcan benegligibleafteraboutsixweeks,dependingonsoilcharacter- istics(mainlytexture),croptypeandclimaticconditions(Tejada, 2009;Panettierietal.,2013).TheslightlyhighernighttimeNEEin theweed-freetreatmentinOctobermighthavebeencausedbyan increase in microbial respirationdue toglyphosate application, justonemonthbeforethebeginningofmeasurements.Onceits effectdisappeared,similarnighttimeNEEwasfoundinthetwo treatments(November–April).Inthelongterm,weedremovalby the herbicide in the weed free treatment could provoke soil compaction,thereductionofSOCandtheincreaseofbulkdensity (Castroetal.,2008),therebydecreasinginﬁltration.

Afterweedswerecutandthehayleftonthesoil(MayandJune), oppositetothepatternobservedduringthegrowthperiod,netC uptakewashigherinthehay-freetreatmentthaninthehay-cover treatment (Table 1). Thislower net C uptake in the hay-cover treatment during late spring can be attributed to the higher respirationpromotedbythehay.First,haydecompositionfavors the formation of stable microaggregates that are enriched in organic C, enhancing earthworm and soil microfauna activity, which in turnaffectsrespirationand thesoil Cpool (Pulleman etal.,2005;Plaza-Bonillaetal.,2015).Second,hayalsoincreases theamountoflabileCwhichisreadilyusedforrespirationbysoil microorganisms. Third, although respiration usually increases Fig.7. CumulativeNEE(continuousline)anduncertainty(dashedline)ofthegap-ﬁlleddata,aswellasannualnetCuptakeinthetwotreatmentsduringthestudyyear.

(11)

with soil temperature, very high temperatures can constrain respiration by exceeding the optimum temperature for some microorganismactivity(O'Connell,1990).Thehightemperatures registeredinthehay-freesoilduringthedryseason(Fig.2)could beresponsibleforlowerrespirationinthesesoilsrelativethehay- covered soil. Fourth, photodegradation of the hay can also contribute to enhancing CO2 emissions (Brandt et al., 2009;

Rutledge et al., 2010). Although soil chamber measurements supportthishigherCO₂effluxinthehay-coveredsoilduringMay, wefoundnosignificantdifferencesinsoilCO2effluxbetweensoils with or without hay in June, when higher net ecosystem CO2

uptakewasstillobservedinthesitewithouthay.AssoilCO₂efﬂux wasmeasuredatmidday,itispossiblethathighsoiltemperatures limitedrespiratoryactivityinbothsoilsduringthistime.Further researchondiurnaltrendsofsoilCO2 efﬂuxunderthetwosoil managementswillhelptoelucidatetheroleofweedcoversinCO2

emissionsandtheirrelativecontributiontoecosystemNEE.

ContrarytopublishedstudiesthathavereportednetCuptake duringsummerandthroughouttheyearinirrigatedoliveorchards underclimateconditionssimilartoours(Testietal.,2008;Nardino etal.,2013),wefoundmonthlynetCO2releaseintheoliveorchard underthetwomanagementsduringthesummerperiod(fromJuly toSeptember;Table 1).Thehighevaporativedemand recorded duringthisyear(maximumdailyairtemperatureduringJulyand Augustonaveragewas36.9CandmaximumdailyVPDonaverage was 52.8 hPa) could constrain tree photosynthesis, making respiratory processes the main contributors to the ecosystem CO2ﬂux.Inthisregard,althoughsoilrespirationinthealleysofour twotreatmentswaslowduringthisperiod(around0.9

m

^mol^CO²

m²s¹), in accordance with other studies (between 1.1 and 1.6

m

^mol^CO2m²s¹, see Testi et al., 2008), due to low soil moisturecontent,highrespirationratesmightbeexpectedinthe drip-irrigatedzoneswherewateravailabilitywasnotlimited.For instance,Testietal.(2008)reportedrespirationrateswereupto 5.7

m

^mol ^CO²^m²^s¹ⁱⁿ^irrigated^olive^groves^beneath ^the^tree

canopyduringsummer. Thus,in theirrigated zones,CO2 efflux frombothsoilandabovegroundrespirationfromolivetreesare expected to significantly contribute to NEE. Leaves and fruits appeartobethemaincontributorstoabovegroundrespirationin olive trees, while respiration of wood biomass (trunk and branches) represent a very small fraction of CO2 flux (Pérez- Priegoetal.,2014).Thus,thepositivevaluesofNEEfoundinboth treatments during summer may be due to leaf, fruit, and soil respiration,thelatteroriginatingunderthetreecanopy.

Nighttime NEE values at our site were higher than those reported by Testi et al. (2008) for winter periods (0.7 versus 1.4

m

^mol^CO2m²s¹,onaverage,inoursite),butlowerthanthose reportedby the mentioned authors during non-winter periods (4.8

m

^mol^CO²^m²^s¹^versus^2.2

m

^mol^CO²^m²^s¹^,^on^average,ⁱⁿ

oursite).Similartotheresultsoftheseauthors,wefoundnoclear relationshipbetweennighttimeNEEandsoilorairtemperature.

Thisisprobablyduetoacombinationofdifferentcauses:i)copious missingdataduringnighttimeperiodsduetobatterymalfunction- ing, and predominance of stable conditions and low friction velocities (u*<0.15ms¹); ii) influence of soil moisture on ecosystemrespiration,sincewater-limitedecosystemsaremois- ture-pulsedependent(Chenetal.,2009;López-Ballesterosetal., 2016). Thus, while in mid-high latitudes temperature is a key driver for CO2 fluxes, its relativeimportance coulddecrease in semiaridenvironmentswherewateristhemostimportantdriver forvegetationproductivity;andiii)strongseasonalityofweedand olivetreeactivity,whichcouldmasktheeffectofsoiltemperature onnighttime(aswellasdaytime)NEE.Insupportofthis,Pérez- Priego et al. (2014) reported a good relationship between aboveground respiration in olive trees and both temperature andphenological stage(i.e.periodsof dormancy,floweringand

fruitsetting),sothattheeffectsoftemperatureonCO2ﬂuxescould be confounded by plant phenology and/or productivity during ﬂoweringandfruit-developmentperiods.

Despite small differences in nighttime NEE between treatments,wecanobservethatnighttimeNEEwasslightlyhigherin theweed-freetreatmentfromMaytoSeptember(Fig.3).Possible causes forthis higherNEEduring nighttimecanbe: i)frequent dewfallepisodesduringnighthaveagreatereffectonactivating soilmicrobialrespirationinthesoilwithouthay,whichisdirectly exposedtodewfall,whiledewshouldberatherdepositedonthe hay in the hay-cover treatment, making this water input less accessibletosoil;orii)highernighttimetemperatureinthesoil without hay (on average,4.9C higher than the hay-cover soil during the period from May to August) can greater stimulate microbialrespiration.

Ingeneral,NEEvaluesrecordedinthisstudywerelowerthan thosereportedinotherirrigatedoliveorchardsundersimilarsoil (clay/clay loam soils) and climate conditions (precipitation regime). While we found NEE values up to 0.7gCm²day¹ duringsummer,Testietal.(2008)reportedNEEvaluesinayoung oliveorchard(LAIofthetreeswas1.9m²m²)inSouthernSpainof

2.7gCm²day¹duringthisperiod.MaximumdaytimeNEEin oursitewas4gCm²day¹,withanaveragevalueof 2.5gC m²day¹ during the period of maximum CO2 assimilation (March) and 0.8gCm²day¹ during summer. In contrast, López-Bernal et al. (2015) reportedan averagedaytime NEE of

4.5gCm²day¹inanirrigatedoliveorchard(LAI=1.5m²m²) insouthernSpainintheperiodfromlateJunetolateSeptember.In anoliveorchard(LAI3m²m²)insouthernItaly,Nardinoetal.

(2013) reportedmaximummonthlyNEEvalues of170gCm² month¹,whilewefoundamaximummonthlyvalueof74.4g Cm²month¹(Table1).Differences inNEEbetweenourstudy siteandtheresultsreportedinotherirrigatedoliveorchardscanbe attributedtothedifferentageanddensityoftheolivetrees,andthe inherentinter-annualvariabilityofsemiaridecosystems, among other factors. Contrary to the young age of the olive orchards reportedinthepreviouslycitedstudies,ourstudywasconducted ina matureoliveorchard(80yearsoldtrees),wheregrowthof treesislimitedandincreaseoftreebiomassislowcomparedtothe rapidgrowththatcanbeexpectedinyoungoliveplantations(5–6 years old). Plant density was also lower in our study (200 treesha¹)thaninthementionedstudies(400treesha¹).The annualNEEinouroliveorchardwasalsolowerthanthatreported foryoungoliveorchardswithplantcovermanagement.Whilewe found annual C uptake of 140gCm² y¹ in the weed-cover treatment(Fig.7),valuesfrom1160gCm²y¹to1345gCm²y¹ havebeenreportedin12–16yearoldoliveorchards(Nardinoetal., 2013). In a 50year old olive orchard on sandy loam soils in Southern Italy and taking into account C inputs bycover crop residues, pruning material, senescent leaves, yield and root biomass ofolivetrees,Paleseet al.(2013)estimated anannual NEE of1550and 1020gCO2m²y¹(419and 275gCm²y¹) undersustainable(irrigationwithurbanwastewatertreatedand conservationofspontaneousweedsandgrasses)andconventional management (rainfedconditions,intensive tillageand pruning), respectively.Our resultsweresimilartothosereportedbyBrilli et al. (2016), who found anannual NEE ranging from137 to

667gCm²y¹(averageof3years,364gCm²y¹)inarain-fed olive orchard in central Italy withsurface tillage management.

Unfortunately,thelackofliteraturereportingdirectmeasurements onCuptakeinoliveorchardsundersimilarconditionstoours(crop characteristic,soilmanagement)makescomparisonsdifﬁcult.

Somestudieshavereportedareductionofcropproductivityin oliveorchardswithplantcoversduetocompetitionforwaterand nutrient resourceswithtrees (Gucci etal.,2012; Ferreiraet al., 2013).Incontrast,otherauthorshavefounda positiveeffectof