Spatio-temporal analysis of factors controlling nitrate dynamics and potential denitrification hot spots and hot moments in groundwater of an alluvial floodplain

(1)

O

pen

A

rchive

T

OULOUSE

A

rchive

O

uverte (

OATAO

)

OATAO is an open access repository that collects the work of Toulouse researchers and

makes it freely available over the web where possible.

This is an author-deposited version published in :

http://oatao.univ-toulouse.fr/

Eprints ID : 15830

To link to this article : DOI:10.1016/j.ecoleng.2015.12.031:

URL :

https://dx.doi.org/10.1016/j.ecoleng.2015.12.031

To cite this version :

Bernard-Jannin, Léonard

and Sun, Xiaoling

and Teissier,

Samuel

and Sauvage, Sabine

and Sanchez-Pérez, José-Miguel

Spatio-temporal analysis of factors controlling nitrate dynamics and

potential denitrification hot spots and hot moments in groundwater

of an alluvial floodplain. (2017) Ecological Engineering vol. 103

(Part B). pp. 372-384. ISSN 0925-8574

Any correspondence concerning this service should be sent to the repository

administrator:

staff-oatao@listes-diff.inp-toulouse.fr

(2)

Please cite this article in press as: Bernard-Jannin, L., et al., Spatio-temporal analysis of factors controlling nitrate

dynam-Ecological

Engineering

j ou rn a l h o m epa g e :w w w . e l s e v i e r . c o m / l o c a t e / e c o l e n g

Spatio-temporal

analysis

of

factors

controlling

nitrate

dynamics

and

potential

denitrification

hot

spots

and

hot

moments

in

groundwater

of

an

alluvial

floodplain

Léonard

Bernard-Jannin

∗

_,

_Xiaoling

_Sun,

_Samuel

_Teissier,

_Sabine

_Sauvage,

José-Miguel

Sánchez-Pérez

∗

ECOLAB,UniversitédeToulouse,CNRS,INPT,UPS,France

Keywords: Groundwater Denitrification Nitrate Wetland Floodplain

a

b

s

t

r

a

c

t

Nitrate(NO3−)contaminationoffreshwatersystemsisaglobalconcern.Inalluvialfloodplains,riparian

areashavebeenproventobeefficientinnitrateremoval.Inthisstudy,alargespatio-temporaldataset collectedduringoneyearatmonthlytimestepswithinameanderareaoftheGaronnefloodplain(France) wasanalysedinordertoimprovetheunderstandingofnitratedynamicanddenitrificationprocessin floodplainareas.TheresultsshowedthatgroundwaterNO3−concentrations(mean50mgNO3−L−1)were

primarilycontrolledbygroundwaterdilutionwithriverwater(explaining54%ofNO3−variance),but

alsobynitrateremovalprocessidentifiedasdenitrification(explaining14%ofNO3−variance).Dilution

wascontrolledbyhydrologicalflowpathsandresidencetimelinkedtoriver-aquiferexchangesandflood occurrence,whilepotentialdenitrification(DEA)wascontrolledbyoxygen,highdissolvedorganic car-bon(DOC)andorganicmattercontentinthesediment(31%ofDEAvariance).DOCcanoriginateboth fromtheriverinputandthedegradationoforganicmatter(OM)locatedintopsoilandsedimentsofthe alluvialplain.Inaddition,riverbankgeomorphologyappearedtobeakeyelementexplainingpotential denitrificationhotspotlocations.Lowbankfullheight(LBH)areascorrespondingtowetlandsexhibited higherdenitrificationratesthanhighbankfullheight(HBH)areaslessoftenflooded.Hydrology deter-minedthetimingofdenitrificationhotmomentsoccurringafterfloodevents.Thesefindingsunderline theimportanceofintegratingdynamicwaterinteractionsbetweenriverandaquifer,geomorphology,and dualcarbonsource(riverandsediment)whenassessingnitratedynamicsanddenitrificationpatternsin floodplainenvironments.

1. Introduction

Nitratecontaminationofgroundwaterthroughpointsourceand diffusepollutionhaslongattractedworld-wideattention(Power andSchepers,1989;Bijay-Singhetal.,1995;Zhangetal.,1996; Arrateetal.,1997;Carpenteretal.,1998;Jégoetal.,2012).Nitrate pollutionfromagriculturalsourcesisconsideredtobethemain causeofgroundwaterdegradationintheEuropeanUnion(Sutton etal.,2011).InEuropeandNorthAmerica,upto90%offloodplain areasarecultivated(TocknerandStanford,2002).Agriculturalland use,incombinationwithfactorssuchasshallowgroundwater,high

∗ Correspondingauthorsat:Avenuedel’Agrobiopole,31326CastanetTolosan Cedex,France.

E-mailaddresses:l.bernardjannin@gmail.com(L.Bernard-Jannin),

jose-miguel.sanchez-perez@univ-tlse3.fr(J.-M.Sánchez-Pérez).

permeabilityof alluvialdepositsand interconnectionswith sur-facewater,makealluvialaquifersparticularlyvulnerabletonitrate diffusepollution(Arauzoetal.,2011).Asaresult,nitrate concentra-tionsexceedingthelimitof50mg-NO3−L−1setforgroundwater

systemsinEuropebytheNitrateDirective(91/676/EEC)andthe GroundwaterDirective(2006/118/EEC)havebeenreportedfor sev-eralshallow aquifers infloodplain areas (Baillieux etal., 2014; Sánchez-Pérezetal.,2003).

Floodplain environmentsarecharacterised bystrongsurface water-groundwater interactions that are important for aquifer watercomposition(Amorosetal.,2002;Sunetal.,2015).Nitrate contaminationinshallowaquiferscanthereforebemitigatedby thedilutionresultingfrommixingwithriverwatercontaininglow nitrateconcentrations(Pinayetal.,1998;Baillieuxetal.,2014). Nitratemassremovalalsooccursthroughnatural biogeochemi-calprocesses suchasplantuptake, denitrification,dissimilatory nitrate reduction to ammonium and microbial immobilisation,

(3)

amongwhichdenitrificationisreportedtobethemostimportant ingroundwater(Korom,1992;Burtetal.,1999;Rivettetal.,2008). Denitrificationistheanaerobicreductionofnitrate(NO3−)into

gaseouscompounds(nitrousoxideordinitrogen)by microorgan-isms.Denitrificationingroundwaterislinkedto:(i)presenceof nitrate,denitrifyingbacteriaandorganiccarbon(OC)aselectron donor,(ii)anaerobicconditionsand(iii)favourable environmen-talconditionsinterms ofe.g.temperatureandpH(Rivettetal., 2008).However,among allthese factors,availability of OChas beenidentifiedasthemajorlimitingfactorinnitrate-contaminated groundwater(seeRivettetal.(2008)forareview).

Riparianzonesarelocatedattheinterfacebetweenaquaticand terrestrialenvironments(Vidonetal.,2010).Intheseareas,where surfacewaterrichinorganicmatter(OM)meetsgroundwater con-tainingabundantnutrients,denitrificationispromoted(Hilletal., 2000;Sánchez-Pérezetal.,2003).Thereforeriparian areashave beenshowntobeefficientinnitratepollutionmitigation(Vidon andHill,2006;Dosskeyetal.,2010;Naimanetal.,2010).Inthese systems, the hydrological exchanges at the river/groundwater interfacecanhaveasignificantimpactondenitrificationratesin theaquifer(Baker and Vervier,2004;Lamontagneetal., 2005). SuchexchangesrechargeaquiferwithriverwaterrichinOM, stim-ulatingdenitrificationingroundwater(Iribar,2007;Sánchez-Pérez etal.,2003).Thelocationofthedenitrificationprocessisdrivenby advectiveflux,whereflowpathsleadorganicmatterandnitrate tomeet(Seitzingeretal.,2006),atpointsdefinedashotspotsby

McClainetal.(2003).Atthescaleofthefloodplainsection, den-itrificationisusuallytriggeredbyNO3−inputfromuplandareas

andhotspotscanbelocatedattheinterfacebetweentheupland andriparianzones,betweentheriparianzoneandthestreamor withintheriparianzone(McClainetal.,2003).Atthisscale,the efficiencyoftheriparianareainnitrateremovalisreportedtobe relatedtohydrogeomorphic characteristicsassociatedwith geo-logicalandhydrologicalsettings(Groffmanetal.,2009),suchas waterresidencetime(Seitzingeretal.,2006).Inaddition,the occur-renceofhotspotsmayvaryintime,especiallyinenvironments withstrong temporal variations in hydrological conditions.For example,denitrificationrateshavebeenfoundtobehigherinhigh flowconditions(BakerandVervier,2004;Iribar,2007;Peteretal., 2011), within periodscorresponding to hot moments (McClain etal.,2003).Therefore,studiesondenitrificationprocessesneed totakeintoaccountbothspatialandtemporalscalewithsuitable resolutioninordertoaccuratelydescribetheprocessesatstakeand theirimpactonnitratedynamics.

Themain objectiveof this studywastoexaminethe occur-rence of potential denitrification hot spots and hot moments and theirrelationship toenvironmental conditions,in order to improve knowledge on nitrate dynamics in floodplains. Based onahighspatialresolutiondatasetcollectedduring12monthly samplingcampaigns, wesought to:(i) identify thefactorsbest explaining nitrate concentrations variations and exploredtheir spatio-temporalpatterns;(ii)identifythefactorsbestexplaining potentialdenitrificationratesmeasuredinaquifersediment sam-ples;and(iii)analysetheoccurrenceofdenitrificationhotspots/hot momentsaccordingtothecontrollingfactorsinordertodevelop aconceptualdiagramofdenitrificationprocessinfloodplain envi-ronments.

2. Materialsandmethods 2.1. Studysite

Thestudyarea,whichcoversabout50ha,islocatedwithina 2km long meanderin the middle sectionof theGaronne river watershed,closetothevillageofMonbéquiinsouth-westFrance

(Fig. 1). Mean annual precipitationin the areais 660mm. The drainage area of the Garonne watershed at the study site is about13,730km2_,_with_annual_average_flow_of₁₉₀_m3_s−1 _and_a

rangefrom98m3_s−1 ₍₁₉₈₉₎_to₃₁₅_m3_s−1 ₍₁₉₇₈₎_over_the_past

41 years. The driest month is August,with 76m3_s−1 _on

aver-age, and the wettest is May, with 343m3_s−1 _on _average. _The

dailyflow ishighlyvariable,rangingfrom10m3_s−1 _during_the

severedroughtinAugust1991to2930m3_s−1_during_the_largest

floodeventrecorded,on6November2000(www.hydro.eaufrance. fr).Thetwo-year return period floodcorresponds todaily flow of 1400m3_s−1_. _The _floodplain _comprises _4–7_m _of _quaternary

sandand gravel deposits (meansaturated hydraulic conductiv-ity10−3_m_s−1_),_overlying_impermeable_molassic_bedrock._In_this

areatheGaronneRiverfullypenetratesthealluvialformation,so thattheriverbedliesontheimpermeablesubstratum.The allu-vialdepositsarecovered witha siltysoillayer1–2mdeepand containing1.5%OMonaverage(Jégoetal.,2012).Theconnection betweenaquiferandtheGaronneriverisstronglyinfluencedby hydrologicalconditions(Sunetal.,2015).Thefloodplainisheavily cultivated(irrigatedmaize,sunflower,sorghum,wheat),leading tomajornitrateinflux intothegroundwater.Concentrationsof 100mg-NO3−L−1arecommon(Sánchez-Pérezetal.,2003).Asmall

areaofriparianforest,mainlycomposedofwillow(Salixalba)and ashtrees(Fraxinusexcelsior),islocatedclosetotheriveratalower elevationthantherestofthestudyarea,andisseparatedfromthe agriculturalfieldsbyplantationsofpoplar(Populusalba)(Fig.1). Thegeomorphologyoftheriverbankontherightsideoftheriver canbeseparatedintotwotypes:Lowbankfullheighttype(LBH) correspondingtotheprofileA–A′₍_Fig.₁_),_which_is_regularly_flooded

asthegroundwaterlevelisoftenclosetothesurfaceandcanbe designatedaspermanentwetland;andhighbankfullheighttype (HBF),correspondingtotheprofileB–B′_,_which_is_only_flooded

dur-ingthehighestfloods(greaterthantwo-yearreturnperiodflood events).Threepiezometers(P6,P9andP13)arelocatedwithina LBHarea.

Previousstudiescarried outonthisareahave demonstrated therole of DOC borne bythe river intothe aquiferin denitri-fication(Sánchez-Pérez et al.,2003)and theimportance ofthe sediment-attachedbacterialcommunityforaquiferdenitrification (Iribaretal.,2008).Inaddition,modellingstudieshaveshownthe importanceofriver-aquiferexchangesforgroundwater composi-tion(Wengetal.,2003;Peyrardetal.,2008;Sunetal.,2015)andthe impactofagriculturalpracticesonnitrateleachingintotheshallow aquifer(Jégoetal.,2012).

2.2. Sampling

Anetworkof22piezometers(internaldiameter51mm,with 1mmslots)wasinstalledthroughoutthestudysitebetweenthe Garonneriverandtheagriculturalfields(Fig.1).Watersamples werecollectedwithineachpiezometerduringmonthlysampling campaigns from April 2013 to March 2014, providing a high-spatialresolutiondatasetwitharound50–100mbetweensampling points.Infourofthesecampaigns,calledfullcampaigns,additional sedimentsamplingwasperformed(Fig.2).

After measuring water table depth (WD), as the distance betweensoilsurfaceandwatertablesurface,waterwaspumped withathermalmotorpumpandphysico-chemicalparametersin waterweremeasuredonceelectricalconductivity(EC)had sta-bilised (Sánchez-Pérez et al., 1991a, 1991b). Dissolved oxygen (DO),temperature(T),pHandECweremeasuredusingaportable metre(WTWMulti3420)andspecificprobes.Watersampleswere thenfilteredthrough0.45mmcelluloseacetatemembranefilters. Anionconcentrations,i.e.nitrate(NO3−),chloride(Cl−),sulphate

(SO42−)andphosphate(PO43−),andcationconcentrations,i.e.

(4)

Please cite this article in press as: Bernard-Jannin, L., et al., Spatio-temporal analysis of factors controlling nitrate dynam-Fig.1.(a)Geographicallocation;(b)digitalterrainelevationandlandusemapofthestudyarea,showingthenetworkofpiezometers(PA-P22)usedinthepresentstudy; and(c)verticalprofilesattwolocations(A–A′_and_B–B′_).

andammonium(NH4+),weredeterminedbyionchromatography

(DionexICS-5000+ _and_DX-120)._Silica_{concentration}_was

deter-minedbyspectrophotocolorimetry(Alpkem)andalkalinity(Alk) by acidtitration. Watersamples collected for determination of DOC concentrationwere filteredthrough rinsed0.45-mm cellu-loseacetatemembranefilters,storedincarbon-freeglasstubes, acidifiedwithHClandcombustedusingaplatinumcatalyser (Shi-madzu,ModelTOC5000)at650◦_C._For_the_four_full_campaigns,

sediment was sampled after water sampling by increasingthe pumpingvelocityfor5–10min,withthewaterflowingintoa50-L tankwheresedimentssettledbeforebeingcollectedtogetherwith 100mLwaterinsealedsterilebags.Intheriver,insitu physico-chemicalmeasurements(DO,TandpH)weremadeandwaterwas sampleddirectlywithinthestream.Allthesampleswere immedi-atelystoredat4◦_C_after_collection.

Oxygen isotope composition wasmeasured byisotope ratio massspectrometryafter CO2–H2Oequilibrationusingthe

tech-niquedescribedbyEpsteinandMayeda(1953).Theresultswere expressedrelativetoastandard(V-SMOW,Gonfiantini,1978)using deltanotation(ı),whereı18_O_is_given_by:

ı18_O = R18_O sample−R18OV-SMOW R18_O V-SMOW whereR18_O

sampleandR18OV-SMOW arethe18O/16Oratiosforthe

sampleandV-SMOW,respectively.Asthedifferencebetween sam-plesandstandardissmall,the‘ı-value’isusuallyexpressedinparts per1000[‰].

2.3. Denitrifyingenzymeactivity(DEA)

Assessmentsofdenitrifyingenzymeactivity(DEA)were con-ducted in the laboratory within the week following sediment samplecollection.Eachsedimentsamplewasanalysedintriplicate. A25mLportionofwetsedimentand50mLdeoxygenatedmilliQ watercontainingKNO3−andsodiumacetate,togivefinal

concen-trationsof100mg-NL−1_and₅₀_mg-C_L−1_,_{respectively,}_were_added

toagas-tight150-mLserumbottle.Aftercompletedeoxygenation (N2sparging),inhibitionofN2Oreductaseswasachievedby

inject-ing15mLC2H2.Incubationswereperformedinthedarkat14◦C,

correspondingtotheaveragemeasuredtemperatureofthe ground-waterinthestudyarea.Atmosphericpressureintheserumbottles wasensuredduringincubation(removalofgasphasebeforeC2H2

additionandadditionofN2tocompensatedsamples).Gassamples

weretakenfromthegasphaseoftheserumbottlesaftervigorous shaking,at30minand6h30minafterC2H2 injection.N2Owas

(5)

Fig.2.Waterdischarge(red)andrain(blue)inthestudyarea.Arrowsand num-bersindicatesamplingperiods(simplearrowsforregularsamplingcampaignsand doublearrowswithboldnumbersforfullcampaigns).(Forinterpretationofthe ref-erencestocolourinthisfigurelegend,thereaderisreferredtothewebversionof thisarticle.)

fittedwithanelectroncapturedetector.Calculationswere per-formedwithN2OsolubilitycoefficientstakenfromWeissandPrice

(1980).TheDEAresultswereexpressedinmgofNpergofdry sed-imentperhour[mg-Ng-sed−1_h−1_]_by_averaging_the_rates_of_N₂_O

productionfor thetriplicatesamples.DEArepresents the maxi-mumrateofdenitrificationthatcanoccurwhenCandNarefreely availableandwasusedinthisstudyasaproxyforthe denitrifi-cationcapacityofthedifferentsamplinglocationsatthesampling times.

Afterincubation,eachsedimentsamplewasdried(105◦_C,₂₄_h)

andcombusted(550◦_C,₄_h)_in_order_to_determine_sediment_OM

content,which wasexpressedasa percentage of dry sediment weight.

2.4. Bacterialrichness

DNAwasextractedaftercentrifugationofthesedimentsamples and PCRamplificationof 16SrRNAgenes wasperformed. Elec-trophoresiswasusedtoproduceT-RFLPprofiles.Bacterialrichness (BR)correspondedtothenumberofT-RF(correspondingtopeaks ontheT-RFLPpatternanddefinedhereafterasOTU,Operational TaxonomicUnit)persample.Presence/absencedatawereusedin theanalyses.

2.5. Dataanalysis

Datatreatmentinvolved20parameters,includingadescriptor ofthehydrologicalconditions(WD),physico-chemicalparameters ofthegroundwatersamples(DO,T,pH,DOC,Alk,SiO2,NO3−,Cl−,

SO42−,PO43−,Ca2+,Mg2+,K+,Na+,NH4+,ı18O),biological

param-eters(BRandDEA)and asediment-relatedparameter(OM).EC wasusedtoensurethestabilityofwatercharacteristicsduringthe sampling,butitwasnotusedindataanalysisasitwasalready rep-resentedthroughthemajorionsconcentrationdata.Variableswere transformedtoensurenormalityoftheirdistributionwhen neces-sary.Partialleastsquaresregression(PLSR),whichisaverysuitable methodforecological studiesanalysinga largearrayof related predictors(Carrascaletal.,2009),wasperformedheretorelate NO3−andDEAtotheotherenvironmentalfactors.PLSRisa

tech-niquethatcombinesfeaturesfromprincipalcomponentanalysis andmultipleregressiontoanalysetheeffectsoflinear combina-tionsofseveralpredictorsXonaresponsevariableY.Associations areestablishedwithlatentfactorsdefinedaslinearcombinations ofpredictorvariablesthatmaximisetheexplainedvarianceofthe dependentvariable (Carrascal et al., 2009; Abdi, 2010).One of

Table1

Hydrological characteristics of the study area during the four full sampling campaigns.

Campaign Discharge [m3_s−1_]

Timefromlast peakflow (Q>1400m3_s−1₎ [days] Samplingdate 1 337 79 8–10April2013 4 314 13 1–3July2013 7 53 104 30September–2 October2013 10 209 54 13–15January2014

theinterestingfeaturesofPLSRisthattherelationshipsbetween thepredictorsandtheresponsefunctioncanbeinferredfromthe weightsandregressioncoefficientsofindividualpredictorsinthe mostexplanatorycomponents(Yanetal., 2013).Leave-one-out (LOO)cross-validationwasperformedtoselectthenumberof com-ponentsandjack-knifeestimationwasusedtotestthesignification ofthecoefficients(MevikandWehrens,2007).ForeachPLSR,the percentageofvarianceexplainedfortheresponsevariableandthe cross-validatedrootmeansquareerror(RMSECV)werecalculated. Afterthepre-analysisofthedata,NH4+andPO43−werenotselected

forthestatisticalmethodbecausetheirconcentrationswerevery lowandbelowthelimitofdetectionformorethan,respectively, 25%and66%ofthemeasurements.PLSRwasperformedonthe14 parametersavailableforthe12campaignsforNO3−(WD,DO,T,

pH,DOC,Alk,SiO2,Cl−,SO42−,Ca2+,Mg2+,K+,Na+,ı18O)andonthe

17parametersavailableforthefourfullcampaignsforDEA(WD, DO,T,pH,DOC,Alk,NO3−,SiO2,Cl−,SO42−,Ca2+,Mg2+,K+,Na+,

ı18_O,_OM,_BR)._All_statistical_analyses_were_performed_using_R_3.1.1

software(RCoreTeam,2012).

Interpolationofgroundwaterlevel,NO3−,DEA,DOC,OM,DO

andPLSRcomponentscorevaluestoproducemapswasperformed usingtheInverseDistanceWeight(IDW)methodimplementedon ArcGis10.0(ESRI,2011).

3. Results

3.1. Hydrologyofthestudyarea

Inallthesamplingcampaigns,thegroundwaterlevelsindicated groundwaterflowfromthealluvialplaintotheriver. Groundwa-terlevelforthefourfullsamplingcampaignsisshowninFig.3.In campaigns1and10,whichtookplacemorethan50daysafterthe lastpeakflow(Q>1400m3_s−1₎_and_in_relatively_high_flow

condi-tions,thehydraulicgradientwassmallerthanfortheothertwofull campaigns.Campaign4tookplaceafewdaysafterthelargestflood recordedduringthestudyperiodandthehydraulicgradientwas thehighestobserved.Campaign7tookplaceduringalowwater period,afewmonthsafterthesamelargefloodbeforecampaign4, andthehydraulicgradientwaslower(Table1andFig.3).

Forallfourcampaigns,themainwaterflowdirectionwasfrom thecentreofthemeandertothenorth-west.However,somewater flow tothewest and south-westof themeanderwasdetected incampaigns4,7and10(Fig.3).Overall,thehydraulicgradient ranged from 1‰ to 6‰, which represents water flow veloci-ties rangingfrom 0.09 to0.52mday−1_,_considering _a _saturated

hydraulicconductivityof10−3_m_s−1_._Depending_on_the_location,

theresidencetimeofthewaterflowingfromtheagriculturalfield totheriverwasestimatedtovaryfrom100daysinthenorthof themeander(locationA,Fig.3)to3000daysinthesouthofthe meander(locationB,Fig.3).

(6)

Please cite this article in press as: Bernard-Jannin, L., et al., Spatio-temporal analysis of factors controlling nitrate dynam-Table2

Characteristicofthe20parametersmeasuredinthepiezometers:meanvalue,standarddeviation(SD)minimum(min),maximum(max)andcoefficientofvariation(CV). Meanvaluesfortheparametersmeasuredintheriverarealsopresented(rivermean).nindicatesthenumberofcampaignforwhichtheparameterisavailable.

Mean SD Min Max CV Rivermean n

WD[m] 3.19 1.20 0.16 5.67 38% – 12 DO[mgL−1_] _3.45 _2.44 _0.06 _9.69 _71% _10.6 ₁₂ T[◦_C] _13.72 _1.12 _10.70 _16.00 _8% _13.27 ₁₂ pH[–] 6.94 0.14 6.40 7.88 2% 8.01 12 DOC[mgL−1_] _1.03 _0.65 _0.26 _4.87 _63% _1.65 ₁₂ Alk[mgL−1_] _6.26 _1.16 _2.80 _9.90 _19% _2.26 ₁₂ NO3−[mgL−1] 49.57 36.29 0.60 139.40 73% 8.06 12 Cl−_[mg_L−1_] _53.77 _29.62 _7.57 _196.63 _55% _9.11 ₁₂ SO42−[mgL−1] 73.60 35.82 5.70 283.10 49% 19.96 12 SiO2[mgL−1] 11.79 3.17 4.10 26.70 27% 4.90 12 Ca2+_[mg_L−1_] _132.89 _42.12 _32.60 _231.47 _32% _48.66 ₁₂ Mg2+_[mg_L−1_] _19.99 _7.31 _5.00 _37.10 _37% _4.93 ₁₂ K+_[mg_L−1_] _2.52 _0.99 _0.77 _6.02 _39% _1.50 ₁₂ Na+_[mg_L−1_] _24.84 _10.32 _7.40 _65.00 _42% _8.06 ₁₂ NH4+[mgL−1] 0.07 0.28 <d 2.78 368% 0.05 12 PO43−[mgL−1] 0.02 0.07 <d 1.16 352% 0.03 12 ı18_O_[‰] −8.88 1.02 −11.54 −5.62 12% −10.81 12 BR[numberofOTU] 13.15 6.50 2.00 36.00 49% – 4

DEA[mg-Ng-sed−1_h−1_] _2.15E−02 _3.63E−02 _2.42E−04 _1.83E−01 _169% _– ₄

OM[%] 0.55 0.17 0.16 1.05 31% – 4

3.2. Biogeochemicalcharacteristicsofthestudyarea

Mean values of the 20 parameters recorded during the 12 campaigns (fourfull campaignsfor BR,DEAand OM)in the22 piezometers and average river characteristics are summarised in Table 2. The average WD was typical of a shallow aquifer, 3.19monaverage,andrangedfrom5.67mtoalmostatthe sur-face(0.16m).Nitrateconcentrationsshowedgreatvariation,from 0.6 to 139mgL−1 _(coefficient _of _variation _(CV) _73%). _The _Ca2+_,

SO42− and Cl− concentrations were relativelyhigh (>50mgL−1

on average). The average NH4+ and PO43− concentration were

ratherlow(<0.1mgL−1_),_but_on_occasion_reached_higher

concen-trations.Biologicalandsediment-relatedparametersalsoexhibited strongvariability.DEAvaluesvariedbyseveralordersofmagnitude (2.42×10−4–1.83×10−1mg-Ng-sed−1h−1, CV=169%). Bacterial richness and OM were also very variable (CV 49% and 31%,

respectively),whileT,pHandı18_O_were_the_parameters_with_the

smallestvariability.Overall,DO,NO3−,DEA,PO43−andNH4+were

themostvariableparameters.

3.3. Nitratedynamicinthefloodplain

Nitrateconcentrationmapsshowedacleardifferencein concen-trationbetweennorthandsouthofthemeander,withthenorth beinghighlyconcentrated andthesouthshowinglower nitrate concentrations(Fig.4).However,theNO3−concentrationswere

lowerinP6andP13,bothlocatedinthelow riparianareathan inthenorth-westpartofthemeander,closetotheagricultural fields,infullsamplingcampaigns1,4and10.Onatemporalscale, NO3−concentrationswerehigherincampaign7,afteralonglow

waterperiod.Inaddition,inthiscampaignsomepoints(PG,PE,

Fig.3. Groundwaterlevelinthestudyareaandgroundwaterflowpathsfromtwolocationsinthemeander(oneinthenorth,A,andoneinthesouth,B)observedinthefour fullcampaigns(1,4,7,and10).

(7)

Fig.4.Mapsofinterpolated(IDW)nitrateconcentrationsandofthescoresofthefirst(comp1)andsecondcomponents(comp2)ofthePLSRperformedonNO3−forthefour

(8)

Please cite this article in press as: Bernard-Jannin, L., et al., Spatio-temporal analysis of factors controlling nitrate dynam-Table3

SummaryofthePLSRanalysisonNO3−.RMSECVistherootmeansquareerrorof

thecross-validation. Component Cumulative%of explained variabilityinNO3− %ofexplained variabilityin NO3− RMSECV 1 56.44 56.44 0.668 2 70.82 14.38 0.545 3 72.65 1.83 0.543 4 74.19 1.54 0.537 5 75.90 1.71 0.525 Table4

ResultsofthePLSRonNO3−with14predictorsavailableforthe12sampling

cam-paigns.p-Valuesarebasedonjack-knifeestimates.W1andW2aretheloadings weightofeachpredictoroncomponents1and2(values>0.3areshowninboldand values<0.1arenotshown).

Estimate pvalue W1 W2 Mg2+ _0.16 _<2.20E−16 _0.402 ı18_O _0.25 _<2.20E−16 _0.418 _0.288 Cl− _0.16 _<2.20E−16 _0.398 DO 0.23 1.91E−15 0.272 0.458 DOC −0.15 4.27E−14 −0.113 −0.407 Na+ _0.06 _6.72E−06 _0.304 _−0.268 K+ _0.10 _9.66E−06 _0.29 T 0.10 4.20E−05 0.158 0.113 Alk −0.09 1.94E−05 −0.427 SiO2 0.09 1.02E−04 0.253 SO42− 0.05 3.95E−03 0.292 −0.257 pH 0.08 6.24E−03 0.382 Ca2+ _0.06 _0.0129 _0.248 −0.174 WD 0.03 0.134 0.146

PC)locatedclosetotheagriculturalfieldsexhibitedhighnitrate concentrations.

ThefactorsexplainingtheNO3−concentrationvariationswithin

thestudyareawereinvestigatedwithaPLSRanalysis(Table3).The twofirstcomponentswereselected,whichexplained,respectively, 56%and14%ofthevarianceofNO3−.Thenextcomponentonly

added2%oftheexplainedvarianceandthedecreaseinRMSECV whenusingthreecomponentswasrelativelylowcomparedwith theRMSECobtainedwithtwocomponents(<1%).

Overall,themostsignificantpredictors(p<0.001),in decreas-ingorder,wereMg2+_,_ı18_O,_Cl−_,_DO,_DOC,_Na+_,_K+_,_T,_Alk_and_SiO₂

(Table4).Theleastsignificantandonlyparameterwithp>0.1was WD.ThecorrelationwaspositivebetweenNO3−andtheregression

coefficientsofallthevariablesexceptDOCandAlk.Themost impor-tant predictors contributing tothefirst componentwere Mg2+_,

ı18_O,_Cl−_and_Na+_,_which_accounted_for_59%_of_the_component₁

composition.SO42−andK+werealsoimportantandwhen

consid-eringthesevariables,thecontributiontocomponent1roseto76%. Allthesevariableshadapositiveweightonthefirstcomponent. Consideringthesecondcomponent,themostimportantvariables contributingtoitwereDOandpH(positivecontribution)andDOC andAlk(negativecontribution).Thesefourvariablescontributed 70%of component2. Thisvalueroseto92% whenadding ı18_O

(positivecontribution)andNa+_and_SO

42−(negativecontribution).

Thescore ofeach componentwaspositivelycorrelated with NO3−andwascalculatedforeachpiezometerandforeach

cam-paign.Theresultinginterpolatedmapsforthefourfullcampaigns are shown in Fig.4. Component1 exhibited spatialdifferences betweenthenorthandsouthofthemeander,beinghigherinthe norththaninthesouth.Inaddition,thecomponent1scoreswere higherinsummerafterthedryingstagephase(campaign7)and theextentofthenorthareawithhighvalueswasgreaterthanfor theothercampaigns.

Thesecondcomponentalsodisplayedspatialdifferencesbutfor thiscomponent,itwasthenorth-westzoneofthemeanderwhich

Table5

SummaryofthePLSRanalysisonDEA.RMSECVistherootmeansquareerrorofthe cross-validation. Component Cumulative%of explained variabilityinDEA %ofexplained variabilityin DEA RMSECV 1 41.32 41.32 0.831 2 45.00 3.68 0.807 3 47.00 2.00 0.826 4 48.06 1.06 0.837 5 48.88 0.82 0.872 Table6

ResultsofthePLSRonDEAwith17predictorsavailableforthefourfullsampling campaigns.p-Valuesarebasedonjack-knifeestimate.W1aretheloadingsweight ofeachpredictoroncomponent1(values>0.3areshowninboldandvalues<0.1 arenotshown).

Estimate pvalue W1 DOC 0.20 2.33E−06 0.494 DO −0.20 5.30E−06 −0.493 OM 0.20 2.13E−04 0.489 Alk 0.10 2.52E−03 0.246 NO3− −0.11 0.018 −0.264 K+ −0.11 0.036 −0.275 Ca2+ _0.06 _0.19 _0.148 ı18_O −0.05 0.29 −0.119 WD −0.04 0.35 Mg2+ −0.04 0.41 T −0.03 0.50 BR 0.03 0.59 Na+ _0.02 _0.67 SiO2 0.01 0.79 Cl− _−0.01 _0.85 SO42− 0.01 0.86 pH 0.00 0.91

hadlowerscoresthantheremainingareaexceptforthecampaign 7.Forthatcampaign,component2hadlowerscoresintheeastand fortwoisolatedpiezometersinthesouth.Overall,thecomponent scorewaslowerafterthelargeflood(campaign4)thanfortheother samplingcampaigns.

3.4. LocalisationandfactorscontrollingDEAhotspotsandhot moments

ThespatialheterogeneityofDEAisclearlyapparentfromthe mapsinFig.5.HighDEAvaluesgreaterthan0.1mg-Ng-sed−1h−1 (correspondingtoDEAhotspots)wereobservedintheLBHarea (P6,P9andP13)incampaigns1,4and7.Inaddition,relativelyhigh values(>0.05mg-Ng-sed−1h−1)wereobservedinthepiezometers closetotheriverincampaign4(P3,P10,P14,P17,PD).In cam-paign7,DEAremainedhighinpiezometerslocatedintheLBHarea (especiallyP6andP9).Finally,DEAappearedtobelowerduring campaign10 and only P3 had a DEA greater than0.05mg-N g-sed−1_h−1_.

ThefactorsexplainingtheDEAvariationswithinthestudyarea wereinvestigatedwithaPLSRanalysis.Onlythefirstcomponent wasselectedand explained41% ofDEA variance(Table5).The nextcomponentonlyadded4%oftheexplainedvarianceandthe decreaseinRMSECVwhenusingtwocomponentswasrelatively low comparedwiththeRMSECcalculated withonecomponent (<4%).

Themostsignificantpredictors(p<0.001),indecreasingorder, wereDOC,DO,andOM(Table6).DOCand OMwerepositively correlated withDEA, while DO was negatively correlated with DEA.Thesethreeparameterscontributed76%ofthecomposition ofthefirstcomponents.Otherparametersrelativelyimportantin explainingDEAvariance(p<0.05)wereAlk(positivecorrelation), NO3−andK+(negativecorrelation).ParameterssuchasWD,Mg2+,

(9)

Fig.5.Mapsofinterpolated(IDW)denitrifyingenzymeactivity(DEA)forthefourfullcampaigns.

T,BR,Na+_,_SiO

2,Cl−,SO42−andpHweretheleastimportant(weight

<0.1).

3.5. DOCandOMdynamicsinthefloodplain

AsOCisakeyelementforthedenitrificationprocess,the vari-ability in DOC and OMat the studysite wasinvestigated. The DOCmapsshowedstrongtemporalvariationsbetweensampling campaigns(Fig.6).TheDOCconcentrationswerethehighestover theentiremeanderincampaign4,correspondingtotheperiod afterthe largestflood event recorded during thestudy period. Spatially,DOCconcentrationswereratherhomogeneousoverthe meander except for one piezometer where values were higher (P13).

AmapofOMcontentofthesedimentisalsoshowninFig.6.In thefirstcampaign,twopiezometersexhibitedhighvalues(P3and P11),whiletheotherswererelativelylowcomparedwithinthe othercampaigns(especiallycampaign4,whereOMvalueswere higherthaninothercampaigns).Overall,OMwasratherspatially andtemporallyhomogenous(CV=31%),butsomelocations(P3,P7, P11)showedparticularlyhighOMcontent(>0.8%).

3.6. Bankgeomorphologyeffectonvariablescontrolling denitrification

BankgeomorphologyinthestudyareadifferedbetweenLBH and HBH areas. Boxplots of variables related to DEA and WD betweenLBHandHBHareshowninFig.7.LBHareashadagreater DEA(p<0.01),awatertableclosertothesurface,lowerDOvalues andhigherDOCvalues(p<0.001)thanHBHareas.However,the OMcontentwasnotsignificantlydifferentbetweenthetwobank geomorphologytypes(p=0.31).

4. Discussion

4.1. Processescontrollingspatio-temporalvariationsinnitrate concentrationsatthestudysite

ThePLSRperformedonNO3−concentrationsindicatedthatthe

twofirstcomponentsexplained56%and14% ofNO3− variance.

Thefirstcomponentwasmainlyrelatedtomajorions concentra-tions,especiallyMg2+_,_Cl−_,_Na+_,_but_also_to_highı18_O_values._This

componentcanberelated tothedegreeofmixingbetweenthe

groundwater,characterisedbyhighconcentrationsofions (Cl−_,

SO42−,Ca2+,Mg2+,Na2+)resultingfromagriculturalactivitiesand

fertiliserinput(Böhlke,2002)andwater-aquiferinteractions,and theriverwater,withlowerionconcentrations.Inaddition,ı18_O_is

knowntobeagoodtracerofdifferentwaterbodiessuchassurface andsubsurfacewater(KendallandMcDonnell,2012).Therefore, thefirstcomponentcanbeseenasanindicatoroftheintensityof mixingandoftheconnectivitybetweentheriverandthe ground-water,primarilycontrollingnitrateconcentrationinthearea,with highscoresrepresentinghighgroundwaterproportion and low scoreshighriverwaterproportion.Spatialmappingofthescore ofthefirstcomponentfordifferentcampaignsindicatedthatthe degreeofmixingwaslinkedtothehydrologyofthearea.According tothegroundwaterlevel(Fig.3),thedominantgroundwaterflow directionwasfromtheagriculturalareanorthofthemeander,and thewaterinthisareathereforehadalowerresidencetimethan inthesouthofthearea,wheregroundwaterflowvelocitieswere lower.Thisisingood agreementwithhighscoreofcomponent 1inthenorthofthemeanderandthelowscoreinthesouthof themeander.Therefore,theinfluenceoftheriverwasgreaterin thesouthofthemeander,wherethefloodingoccurrencecouldbe higherthantheresidencetimeofthegroundwater,leadingtoan areaofpermanentlylowionconcentrations.Inaddition,temporal variationsinthedegreeofmixingbetweengroundwaterandriver waterwereobserved.Theinfluenceofthegroundwateraffected agreaterareaincampaign7,morethanthreemonthsafterthe lastsignificantfloodevent,thanfortheotherdates.Thiscanbe explainedbythelonggroundwaterdrainingperiodassociatedwith thedryingstage,duringwhichnoriverwaterenterstheaquifer.

Whilepreviousstudieshaveshownthatmixingbetweenthe riverandthegroundwatermostlyexplainsthedecreaseinaquifer NO3−concentrations(Pinayetal.,1998;Baillieuxetal.,2014),

oth-ersstudieshaveidentifieddifferentsourcesleadingtotheNO3−

dilution,suchasgroundwaterflowingupwardsfromtheshallow bedrockaquifer(Craigetal.,2010)orfromuplandterracerecharge (Pfeifferetal.,2006).Inthisstudy,theriver–groundwaterwater exchangeswerefoundtobeessentialinexplainingNO3−

spatio-temporal patternsobservedin theshallowaquiferofthestudy area,whichismainlycontrolledbytheresidencetimeofthehighly loadedgroundwaterflowingfromtheagriculturalareatotheriver. Temporalvariationsinthedegreeofmixingcanalsobeobserved andarecontrolledbythehydrologicalcycle,especiallydryingstage durationandfloodoccurrences.

(10)

Please cite this article in press as: Bernard-Jannin, L., et al., Spatio-temporal analysis of factors controlling nitrate dynam-The second component of the PLSR (14% of NO3− variance

explained)ismainlyrelatedtoDOCandalkalinity(negative contri-bution)andDOandpH(positivecontribution).Inriverinesystems, riverwater infiltratingthroughthehyporheiczonecanprovide groundwaterwithDO(Boultonetal.,1998;Zarnetskeetal.,2012). However, in the study area, the lower DOconcentration areas arelocatedalongtheriverchannel andthereforetheselowDO concentrationscanmorelikelybelinkedtomicrobialactivity con-sumingoxygenbroughtbytheriverwaterenteringthehyporheic zone (Mermillod-Blondin et al., 2005). Alkalinity is a parame-terthatcanbegeneratedinriparianareasthroughOCoxidation (Vidonetal.,2010)anddenitrification(LiandIrvin,2007).In addi-tion, DOC is a key element for heterotrophicmicrobial growth andcanbelinkedtoheterotrophicactivity.DOCisknowntobe relatedtoheterotrophicactivity(Wetzel,1992),aswellas rela-tivelylowpHthatmaybeanindicatorofhighrespirationlevel (Henze,1997).Theresultscouldindicatethatnitrate concentra-tionsareaffectedbyabiologicalprocessdenotedbycomponent 2ofthePLSRand,interactingwithpH,fuelledbyOC,consuming oxygenand producing alkalinity. Amongall thebiological pro-cessesaffectingnitrate,thiscombinationofparameterscouldbe related todenitrification.Indeed, denitrificationleadstonitrate removalunderanaerobicconditionsandusesOCasanelectron

donor(Rivettetal.,2008)andhasbeenrelatedtoalkalinity pro-duction(LiandIrvin,2007).Nitrificationoccursinriparianareas, butinthisstudyitwasdifficulttolinkcomponent2with nitrifica-tion,sincegroundwaterinthestudyareaexhibitedlowammonium concentrations.Anotherprocessthatcouldbeconsideredisthe assimilationof nitrate into microbial biomass, but this process canonlyremoveasmallamountofnitrateandtherapid bacte-rialdie-offmayleadtoNreleaseasammoniumandthennitrate in groundwater(Rivett etal.,2008).In additiontheprobability that component 2is linked todenitrificationis accentuatedby thefactthat riparian areasareknowntobeplaceswhere high denitrification rates are found(Pinay and Decamps, 1988; Hill et al., 2000; Ranalli and Macalady, 2010)and this has already beenobservedatthestudysite(Sánchez-Pérezetal.,2003;Iribar et al., 2008). Denitrification had a lower impacton NO3−

con-centrationsthanwatermixing,butunlikemixingdenitrification leadstoeffectiveNremovalfromthegroundwater(Pinayetal., 1998).

4.2. EnvironmentalfactorscontrollingDEA

Inthisstudy,DEAwasusedasanindicatorofdenitrification underoptimalandstandardisedcarbonandnitratesupply.While

(11)

Fig.7. BoxplotofDOC,DO,OM,DEAandWDaccordingtothetwobankgeomorphologytypes(LBHandHBH).DOC,DEAandDOweretransformedtoensurenormalityof thedatadistribution.p-ValuesarecalculatedwithStudentt-test.

DEAmeasuresthepotentialcapacityofdenitrifyingcommunities underoptimalconditionsanddoesnotrepresentactualinsiturate ofdenitrification,ithasbeenrelatedtothefunctional response (denitrification)ofdenitrifyingcommunities(Iribaretal.,2008). DEA measured in the studyarea varied over several orders of magnitude.Theaverageratewas0.022mg-Ng-sed−1h−1,which isconsistentwiththeratesreportedbyPetersonetal.(2013)for subsoil(0.015mg-Ng-sed−1h−1)andslightlyhigherthantherates theymeasuredindeepergravellayer(0.005mg-Ng-sed−1h−1)of analluvialfloodplain.However,thehighestraterecordedinthe area(0.18mg-Ng-sed−1h−1)waslowerthantheratesmeasuredin topsoilbyMilleretal.(2008),whichreached1.3mg-Ng-sed−1h−1. Furthermore,theresultsshowthatDEAratesinthestudyareacould beverylow(2.2×10−4mg-Ng-sed−1h−1).

Inthepresentstudy,highDEAwasrelatedtohighDOC con-centrationandOMcontentandtolowDOconcentrations,these factorsexplaining31%ofDEAvarianceoverthestudyarea.OM contenthasbeenlinkedtobacterialdensityandcanbe consid-eredagoodproxyforbacterialbiomassinsediment(Iribar,2007). Therefore,theoccurrenceatthesamespotofanoxiaandhighOM couldindicateimportantbacterialactivity.Theaboverelationship canthusbeexplainedbythefactthatincreasedheterotrophic bac-terialactivityinthepresenceofDOCleadstoanoxia(Arangoetal., 2007)andgrowthofbacterialbiomass(increasingOM),promoting conditionsfordenitrificationtooccur.Inadditiontobeingan indi-catorofbacterialbiomass,OMcanindicatepatcheslocatedwithin sedimentandcanbeasourceofOCfuellingthedenitrification pro-cess(Hilletal.,2000).Therefore,OCdynamicscanbeinterpreted asakeyfactorexplainingspatio-temporalvariationsinDEAandOC sourcesneedtobethoroughlyinvestigated.Indeedcarbonsupply isoftenconsideredasthemajorlimitingfactorfordenitrification (StarrandGillham,1993;Jacintheetal.,1998;Devitoetal.,2000; Rivettetal.,2008;Zarnetskeetal.,2011;Peteretal.,2012). How-ever,bacterialrichness,whichwastheonlyvariabledescribingthe bacterialcommunity,wasnotrelatedtoDEAinthisstudy.Thetotal

DEAvarianceexplainedbytheothersvariableswasrelativelylow. Othersdescriptorssuchascarbonbioavailability(Bakeretal.,2000) orabetterdescriptorofbacterialcommunities(Iribaretal.,2008) couldhaveimproveexplainedDEAvariance.

OCinalluvialaquifercanoriginatefromtwosources:(i) degra-dationoftheOMcontainedinthesedimentand(ii)transported bywaterflows(Zarnetskeetal.,2012).Thefirstsourceincludes thedegradationofOMcontainedinthetopsoillayerduringhigh groundwaterperiodsorleachingfromleaflitterduringinundation (BaldwinandMitchell,2000).IsolatedOMpatchessuchasburied channeldepositscanalsoprovideasourceofOCwithinsediments (Hill etal.,2000).OCcanalsoberelatedtotheDOCbroughtby theriverthroughlateralrechargeandoverbankflowinfiltration duringfloodevents(Peyrardetal.,2011;Zarnetskeetal.,2012). TheroleofDOCcomingfromtheriverinthestudyareahasbeen reportedpreviously(Sánchez-Pérezetal.,2003),butrecent mod-ellingstudiesshowthatDOCinputintheriparianareathroughthe riverplaysa smallroleincomparisonwiththeinsitusediment particulateorganiccarbon(POC)content(Sun,2015).Inthisstudy DOCconcentrationswerehigherneartheriverandintheperiod followingafloodandcanbelinkedtorivertransport.However,the DOCconcentrationsarenotrelatedtoconservativetracerssuchas Cl−_,_suggesting_that_the_DOC_can_also_originate_from_another_source,

suchassedimentOMdissolution.

DifferencesinDEAandrelatedfactorswereinvestigatedfortwo differenttypesofbankgeomorphology,LBHandHBH.DEA,in rela-tiontoDOCandDO,wasmoreimportantinLBHthaninHBHareas, makingtheLBHareasDEAhotspots.Geomorphologyisknownto affectdenitrificationinriparianareas(Goldetal.,2001;Vidonand Hill,2004;RanalliandMacalady,2010).Hill(1996)indicatedthat denitrificationispromotedinriparianareaswithanimpermeable layerclosetothesurfaceforcingthegroundwatertoflowintoareas whereconditionsareoptimumfordenitrification.Thissetting cor-respondstotheLBHareas ofthisstudy.Thelowerwater table, increasingthecontactbetweenwaterandtopsoillayercontaininga

(12)

Please cite this article in press as: Bernard-Jannin, L., et al., Spatio-temporal analysis of factors controlling nitrate dynam-Fig.8.Conceptualdiagramofdenitrificationprocessinthefloodplain.Highbankfullheight(HBF)correspondstotheleftbankandlowbankfullheight(LBF)totheright bankinthediagram.(a)Duringtheflood,DOCenterstheaquiferwithriverwaterandisdegradedfromOMofthetopsoillayer;(b)aftertheflood,denitrificationhotspots occurinHBFandLBF;(c)afteralongdryingstage,denitrificationislowerandonlyoccursinLBF.

highOMcontent,mayexplainthehighDEAmeasuredinLBHareas. InadditiontheOMcontentwasnotdifferentbetweenthetwotypes ofgeomorphologyandsomehighvaluesfoundwererelatedtoDEA hotspotslocatedoutsidetheLBHarea.

4.3. Conceptualmodelofdenitrificationinalluvialfloodplains Analysisofthespatio-temporalvariationsindenitrificationat the study site allowed a conceptual diagram of denitrification occurrenceinalluvialfloodplainstobeshaped(Fig.8).Thetwo geo-morphologicalconfigurationsofthebanks(LBHandHBH)showed differencesindenitrificationprocesses.Toexplainthedynamics of denitrificationandnitratein a floodplain,a typicaltemporal sequencestartingfromafloodeventtotheendofthedryingstage hastobetakenintoconsideration.Duringflooding,DOCcoming fromtheriverflowenterstheaquiferthroughbankrechargein HBHandLBH.Inaddition,OCfromthetopsoillayerisincorporated intothegroundwaterinLBH,wherethegroundwaterlevelreaches thesurfaceandthroughleachingduetotheinfiltrationof over-bankflow(Fig.8a).ThereisconsequentlymoreDOCinLBHdueto itsdualoriginthaninHBH.Then,followingtheflood, denitrifica-tionoccursatasignificantrateinHBHandLBH,andnitrateflowing fromtheagriculturalareatotheriverisconsumedatboth loca-tions(Fig.8b).Afteradryingperiod,thedurationofwhichdepends ontheriverfloodplaingeomorphologicalsettings,DOCisentirely consumedin HBHandnolongersupportsdenitrification. How-ever,DOCconcentrationsarehigherinLBHthaninHBHafterthe floodandthegroundwaterremainsclosetothesurfaceforalonger period,makingOCinthesoilstillavailableinLBH.Therefore,in LBH,denitrificationremainsactiveforalongerperiodduringthe dryingstage(Fig.8c).DenitrificationinLBHdecreasesasthedrying stagecontinuesandeventuallyceases.

Somelocaldeviationsfromthegeneralisednitratedynamics describedabovewereobservedinthedataandmaybeduetolocal heterogeneityofthesubstrateaffectingflowpathsorlocalOM con-tent.However,theseobservationswereratherrareinviewofthe highdensityofsamplingpoints,indicatingthatsignificant conclu-sionscanbedrawnfromthisstudy.Thisworkwasbasedonthe hypothesisthat denitrificationisthemostimportantprocessof nitrateremoval,buttheshallowwatertableinLBHareaswould allowplantuptaketooccur(Pinayetal.,1998).Thisprocesswould resultingreaternitrateremovalratesandshouldbeincludedin furtherstudiesforbetterevaluationofnitrateremovalpotentialof theriparianarea.

5. Conclusions

Thisstudyexaminedthecharacteristicsofprocessesoccurring inafloodplainarea,withthefocusondenitrificationpatternsand nitratedynamics.PLSRanalysisrevealedthatNO3−concentrations

in groundwaterweremainlycontrolledbythemixing between river water and groundwater and denitrification also played a significantrole innitrateremoval.Potentialdenitrificationrates were strongly linked to the presence of DOC, OM and anoxia. However,manyfactors,includinggeomorphologyandhydrological cycle,appearedtocontrolpotentialdenitrificationhotspotandhot momentpatterns.Thecombinationofthesefactorsledtocomplex andheterogeneousnitratedynamicsinthestudyarea.These find-ingsshowthataccurateassessmentsofdenitrificationratesand nitrateloadsinfloodplainenvironmentsfromfieldstudiescanbe madefromdetaileddatasets.Duetothedifficultyofgettinga com-pletefielddatasetforeverylocation,modellingworkintegrating allthesefactorsmightbeavaluabletoolinevaluatingfloodplain

(13)

nitratemitigationpotentialunderdifferentfloodplain configura-tions.

Acknowledgments

ThisstudywasundertakenaspartoftheEUInterregSUDOEIVB programme(ATTENAGUA–SOE3/P2/F558project)andwasfunded byERDF.Itwascarriedoutaspartof“ADAPT’EAU” (ANR-11-CEPL-008),aprojectsupportedbytheFrenchNationalResearchAgency (ANR)withintheframeworkoftheGlobalEnvironmentalChanges andSocieties(GEC&S)programme.Wearegratefultovarious col-leaguesatECOLABfortheirhelp,especiallytoOusamaChamsiwith fielddatacollectionandVirginiePayré-SucandIssamMoussawith laboratoryanalyses.

References

Abdi,H.,2010.Partialleastsquaresregressionandprojectiononlatentstructure regression(PLSRegression).WileyInterdiscip.Rev.Comput.Stat.2,97–106.

Amoros,C.,Bornette,G.,Cedex,V.,2002.Connectivityandbiocomplexityin water-bodiesofriverinefloodplains.Freshw.Biol.47,761–776,http://dx.doi.org/10. 1046/j.1365-2427.2002.00905.x.

Arango,C.P.,Tank,J.L.,Schaller,J.L.,Royer,T.V.,Bernot,M.J.,David,M.B.,2007.

Benthicorganiccarboninfluencesdenitrificationinstreamswithhighnitrate concentration.Freshw.Biol.52,1210–1222.

Arauzo,M.,Valladolid,M.,Martínez-Bastida,J.J.,2011.Spatio-temporaldynamicsof nitrogeninriver-alluvialaquifersystemsaffectedbydiffusepollutionfrom agri-culturalsources:implicationsfortheimplementationoftheNitratesDirective. J.Hydrol.411,155–168,http://dx.doi.org/10.1016/j.jhydrol.2011.10.004. Arrate,I.,Sanchez-Perez,J.-M.,Antiguedad,I.,Vallecillo,M.A.,Iribar,V.,Ruiz,M.,

1997.GroundwaterpollutioninQuaternaryaquiferofVitoria–Gasteiz(Basque Country,Spain).Environ.Geol.30,257–265.

Baillieux,A.,Campisi,D.,Jammet,N.,Bucher,S.,Hunkeler,D.,2014.Regionalwater qualitypatternsinanalluvialaquifer:directandindirectinfluencesofrivers. J.Contam.Hydrol.169,123–131,http://dx.doi.org/10.1016/j.jconhyd.2014.09. 002.

Baker,M.A.,Valett,H.M.,Dahm,C.N.,2000.Organiccarbonsupplyandmetabolism inashallowgroundwaterecosystem.Ecology81,3133–3148.

Baker,M.A.,Vervier,P.,2004.Hydrologicalvariability,organicmattersupplyand denitrificationintheGaronneRiverecosystem.Freshw.Biol.49,181–190,http:// dx.doi.org/10.1046/j.1365-2426.2003.01175.x.

Baldwin,D.S.,Mitchell,A.M.,2000.Theeffectsofdryingandre-floodingonthe sedimentandsoilnutrientdynamicsoflowlandriver–floodplain systems: a synthesis. Regul. Rivers Res. Manage. 16, 457–467, doi:10.1002/1099-1646(200009/10)16:5<457::AID-RRR597>3.0.CO;2-B.

Bijay-Singh,Yadvinder-Singh,Sekhon,G.S.,1995.Fertilizer-Nuseefficiencyand nitratepollutionofgroundwaterindevelopingcountries.J.Contam.Hydrol. 20,167–184,http://dx.doi.org/10.1016/0169-7722(95)00067-4.

Böhlke,J.-K.,2002.Groundwaterrechargeandagriculturalcontamination. Hydro-geol.J.10,153–179.

Boulton,A.J.,Findlay,S.,Marmonier,P.,Stanley,E.H.,Valett,H.M.,1998.The func-tionalsignificanceofthehyporheiczoneinstreamsandrivers.Annu.Rev.Ecol. Syst.29,59–81,http://dx.doi.org/10.1146/annurev.ecolsys.29.1.59.

Burt,T.P.,Matchett,L.S.,Goulding,K.W.T.,Webster,C.P.,Haycock,N.E.,1999. Deni-trificationinriparianbufferzones:theroleoffloodplainhydrology.Hydrol. Pro-cesses13,1451–1463, doi:10.1002/(SICI)1099-1085(199907)13:10<1451::AID-HYP822>3.0.CO;2-W.

Carpenter,S.R.,Caraco,N.F.,Correll,D.L.,Howarth,R.W.,Sharpley,A.N.,Smith,V.H., 1998.Nonpointpollutionofsurfacewaterswithphosphorusandnitrogen.Ecol. Appl.8,559–568,doi:10.1890/1051-0761(1998)008[0559:NPOSWW]2.0.CO;2.

Carrascal,L.M.,Galván,I.,Gordo,O.,2009.Partialleastsquaresregressionasan alternativetocurrentregressionmethodsusedinecology.Oikos118,681–690,

http://dx.doi.org/10.1111/j.1600-0706.2008.16881.x.

Craig,L.,Bahr,J.M.,Roden,E.E.,2010.Localizedzonesofdenitrificationinafloodplain aquiferinsouthernWisconsin,USA.Hydrogeol.J.18,1867–1879.

Devito,K.J.,Fitzgerald,D.,Hill,A.R.,Aravena,R.,2000.Nitratedynamicsinrelation tolithologyandhydrologicflowpathinariverriparianzone.J.Environ.Qual. 29,1075–1084.

Dosskey,M.G.,Vidon,P.,Gurwick,N.P.,Allan,C.J.,Duval,T.P.,Lowrance,R.,2010.The roleofriparianvegetationinprotectingandimprovingchemicalwaterquality instreams.JAWRAJ.Am.WaterResourcesAssoc.46,261–277,http://dx.doi. org/10.1111/j.1752-1688.2010.00419.x.

Epstein,S.,Mayeda,T.,1953.VariationofO18contentofwatersfromnatural sources.Geochim.Cosmochim.Acta4,213–224.

ESRI,R.,2011.ArcGISDesktop:Release10.EnvironmentalSystemsResearch Insti-tute,CA.

Gold,A.J.,Groffman,P.M.,Addy,K.,Kellogg,D.Q.,Stolt,M.,Rosenblatt,A.E.,2001.

LandscapeAttributesasControlsonGroithdWaterNitrateRemovalCapacityof RiparianZones.

Gonfiantini,R.,1978.StandardsforStableIsotopeMeasurementsinNatural Com-pounds.

Groffman,P.,Butterbach-Bahl,K.,Fulweiler,R.,Gold,A.,Morse,J.,Stander,E.,Tague, C.,Tonitto,C.,Vidon,P.,2009.Challengestoincorporatingspatiallyand tempo-rallyexplicitphenomena(hotspotsandhotmoments)indenitrificationmodels. Biogeochemistry93,49–77,http://dx.doi.org/10.1007/s10533-008-9277-5. Henze,M.,1997.Basicbiologicalprocesses.In:WastewaterTreatment.Springer,pp.

55–112.

Hill,A.R.,1996.Nitrateremovalinstreamriparian zones.J.Environ.Qual.25, 743–755.

Hill,A.R.,Devito,K.J.,Campagnolo,S.,Sanmugadas,K.,2000.Subsurface denitrifica-tioninaforestriparianzone:interactionsbetweenhydrologyandsuppliesof nitrateandorganiccarbon.Biogeochemistry51,193–223.

Iribar,A.,2007.Compositiondescommunautésbactériennesdénitrifiantesausein d’unaquifèrealluvialetfacteurscontrôlantleurstructuration:relationentre structuredescommunautésetdénitrification.UniversitédeToulouseIII–Paul Sabatier.

Iribar,A.,Sánchez-Pérez,J.,Lyautey,E.,Garabétian,F.,2008.Differentiated free-livingandsediment-attachedbacterialcommunitystructureinsideandoutside denitrificationhotspotsintheriver–groundwaterinterface.Hydrobiologia598, 109–121,http://dx.doi.org/10.1007/s10750-007-9143-9,LA–English. Jacinthe,P.-A.,Groffman,P.M.,Gold,A.J.,Mosier,A.,1998.Patchinessinmicrobial

nitrogentransformationsingroundwaterinariparianforest.J.Environ.Qual. 27,156–164.

Jégo,G.,Sánchez-Pérez,J.M.,Justes,E.,2012.Predictingsoilwaterandmineral nitrogencontentswiththeSTICSmodelforestimatingnitrateleachingunder agriculturalfields.Agric.WaterManage.107,54–65,http://dx.doi.org/10.1016/ j.agwat.2012.01.007.

Kendall,C.,McDonnell,J.J.,2012.IsotopeTracersinCatchmentHydrology.Elsevier.

Korom,S.F.,1992.Naturaldenitrificationinthesaturatedzone:areview.Water ResourcesRes.28,1657–1668,http://dx.doi.org/10.1029/92WR00252. Lamontagne,S.,Herczeg,A.L.,Dighton,J.C.,Jiwan,J.S.,Pritchard,J.L.,2005.Patterns

ingroundwaternitrogenconcentrationinthefloodplainofasubtropicalstream (WollombiBrook,NewSouthWales).Biogeochemistry72,169–190,http://dx. doi.org/10.1007/s10533-004-0358-9.

Li,B.,Irvin,S.,2007.ThecomparisonofalkalinityandORPasindicatorsfor nitrifi-cationanddenitrificationinasequencingbatchreactor(SBR).Biochem.Eng.J. 34,248–255,http://dx.doi.org/10.1016/j.bej.2006.12.020.

McClain,M.E.,Boyer,E.W.,Dent,C.L.,Gergel,S.E.,Grimm,N.B.,Groffman,P.M.,Hart, S.C.,Harvey,J.W.,Johnston,C.A.,Mayorga,E.,McDowell,W.H.,Pinay,G.,2003. Biogeochemicalhotspotsandhotmomentsattheinterfaceofterrestrialand aquaticecosystems.Ecosystems6,301–312, http://dx.doi.org/10.1007/s10021-003-0161-9.

Mermillod-Blondin,F.,Mauclaire,L.,Montuelle,B.,2005.Useofslowfiltration columnstoassessoxygenrespiration,consumptionofdissolvedorganiccarbon, nitrogentransformations,andmicrobialparametersinhyporheicsediments. WaterRes.39,1687–1698,http://dx.doi.org/10.1016/j.watres.2005.02.003. Mevik,B.-H.,Wehrens,R.,2007.Theplspackage:principalcomponentandpartial

leastsquaresregressioninR.J.Stat.Softw.18,1–24.

Miller,M.N.,Zebarth,B.J.,Dandie,C.E.,Burton,D.L.,Goyer,C.,Trevors,J.T.,2008.Crop residueinfluenceondenitrification,N2Oemissionsanddenitrifiercommunity

abundanceinsoil.SoilBiol.Biochem.40,2553–2562,http://dx.doi.org/10.1016/ j.soilbio.2008.06.024.

Naiman,R.J.,Decamps,H.,McClain,M.E.,2010.Riparia:Ecology,Conservation,and ManagementofStreamsideCommunities.AcademicPress.

Peter,S.,Rechsteiner,R.,Lehmann,M.F.,Tockner,K.,Durisch-Kaiser,E.,2011. Deni-trificationHotSpotandHotMomentsinaRestoredRiparianSystem.IAHS-AISH Publication,pp.433–436.

Peter,S.,Rechsteiner,R.,Lehmann,M.F.,Brankatschk,R.,Vogt,T.,Diem,S.,Wehrli, B.,Tockner,K.,Durisch-Kaiser,E.,2012.Nitrateremovalinarestoredriparian groundwatersystem:functioningandimportanceofindividualriparianzones. Biogeosci.Discuss.9,6715–6750.

Peterson,M.E.,Curtin,D.,Thomas,S.,Clough,T.J.,Meenken,E.D.,2013. Denitrifi-cationinvadosezonematerialamendedwithdissolvedorganicmatterfrom topsoilandsubsoil.SoilBiol.Biochem.61,96–104,http://dx.doi.org/10.1016/j. soilbio.2013.02.010.

Peyrard,D.,Sauvage,S.,Vervier,P.,Sánchez-Pérez,J.-M.,Quintard,M.,2008.A cou-pledverticallyintegratedmodeltodescribelateralexchangesbetweensurface andsubsurfaceinlargealluvialfloodplainswithafullypenetratingriver.Hydrol. Processes22,4257–4273,http://dx.doi.org/10.1002/hyp.7035.

Peyrard,D.,Delmotte,S.,Sauvage,S.,Namour,P.,Gerino,M.,Vervier,P., Sánchez-Pérez,J.M.,2011.Longitudinaltransformationofnitrogenandcarboninthe hyporheiczoneofanN-richstream:acombinedmodellingandfieldstudy.Phys. Chem.EarthA/B/C36,599–611,http://dx.doi.org/10.1016/j.pce.2011.05.003. Pfeiffer,S.M.,Bahr,J.M.,Beilfuss,R.D.,2006.Identificationofgroundwater

flow-pathsanddenitrificationzonesinadynamicfloodplainaquifer.J.Hydrol.325, 262–272,http://dx.doi.org/10.1016/j.jhydrol.2005.10.019.

Pinay,G.,Decamps,H.,1988.Theroleofriparianwoodsinregulatingnitrogen fluxesbetweenthealluvialaquiferandsurfacewater:aconceptualmodel.Regul. Rivers2,507–516.

Pinay,G.,Ruffinoni,C.,Wondzell,S.,Gazelle,F.,1998.Changeingroundwaternitrate concentrationinalargeriverfloodplain:denitrification,uptake,ormixing?J.N. Am.Benthol.Soc.17,179–189,http://dx.doi.org/10.2307/1467961.

Power,J.F.,Schepers,J.S.,1989.NitratecontaminationofgroundwaterinNorth America.Agric.Ecosyst.Environ.26,165–187, http://dx.doi.org/10.1016/0167-8809(89)90012-1.

RCoreTeam,2012.R:ALanguageandEnvironmentforStatisticalComputing.R FoundationforStatisticalComputing,Vienna,Austria.

(14)

Please cite this article in press as: Bernard-Jannin, L., et al., Spatio-temporal analysis of factors controlling nitrate dynam-Ranalli,A.J.,Macalady,D.L.,2010.Theimportanceoftheriparianzoneandin-stream

processesinnitrateattenuationinundisturbedandagriculturalwatersheds–a reviewofthescientificliterature.J.Hydrol.389,406–415,http://dx.doi.org/10. 1016/j.jhydrol.2010.05.045.

Rivett,M.O.,Buss,S.R.,Morgan,P.,Smith,J.W.N.,Bemment,C.D.,2008.Nitrate atten-uationingroundwater:areviewofbiogeochemicalcontrollingprocesses.Water Res.42,4215–4232,http://dx.doi.org/10.1016/j.watres.2008.07.020. Sánchez-Pérez,J.M.,Tremolieres,M.,Carbiener,R.,1991a.Unestationd’épuration

naturelledesphosphatesetnitratesapportésparleseauxdedébordementdu Rhin:laforêtalluvialeàfrêneetorme.C.R.Acad.Sci.Sér.3,Sci.Vie312,395–402.

Sánchez-Pérez,J.M.,Tremolieres,M.,Schnitzler,A.,Carbiener,R.,1991b.Evolution delaqualitéphysico-chimiquedeseauxdelafrangesuperficielledelanappe phréatiqueenfonctionducyclesaisonnieretdesstadesdesuccessiondesforêts alluvialesrhénanes(Querco-UlmetumminorisIssl.24).ActaOecol.(1990)12, 581–601.

Sánchez-Pérez,J.M.,Vervier,P.,Garabétian,F.,Sauvage,S.,Loubet,M.,Rols,J.L., Bariac,T.,Weng,P.,2003.Nitrogendynamicsintheshallowgroundwaterofa riparianwetlandzoneoftheGaronne,SWFrance:nitrateinputs,bacterial den-sities,organicmattersupplyanddenitrificationmeasurements.Hydrol.Earth Syst.Sci.7,97–107,http://dx.doi.org/10.5194/hess-7-97-2003.

Seitzinger,S.,Harrison,J.A.,Böhlke,J.K.,Bouwman,A.F.,Lowrance,R.,Peterson,B., Tobias,C.,VanDrecht,G.,2006.Denitrificationacrosslandscapesand water-scapes:asynthesis.Ecol.Appl.16,2064–2090.

Starr,R.C.,Gillham,R.W.,1993.Denitrificationandorganiccarbonavailabilityintwo aquifers.Groundwater31,934–947.

Sun,X.,Bernard-Jannin,L.,Garneau,C.,Volk,M.,Arnold,J.G.,Srinivasan,R.,Sauvage, S.,Sánchez-Pérez,J.M.,2015.Improvedsimulationofriverwaterand ground-waterexchangeinanalluvialplainusingtheSWATmodel.Hydrol.Processes.

Sun,X.,2015.Modélisationdeséchangesnappe-rivièreetduprocessusde déni-trificationdanslesplainesalluvialesàl’échelledubassinversant.Université ToulouseIII–PaulSabatier.

Sutton,M.A.,Howard,C.M.,Erisman,J.W.,Billen,G.,Bleeker,A.,Grennfelt,P.,van Grinsven,H.,Grizzetti,B.,2011.TheEuropeanNitrogenAssessment:Sources, EffectsandPolicyPerspectives.CambridgeUniversityPress.

Tockner,K.,Stanford,J.A.,2002.Riverinefloodplains:presentstateandfuturetrends. Environ.Conserv.29,308–330.

Vidon,P.G.F.,Hill,A.R.,2004.Landscapecontrolsonnitrateremovalinstream ripar-ianzones.WaterResourcesRes.,40.

Vidon,P.G.,Hill,A.R.,2006.Alandscape-basedapproachtoestimateriparian hydro-logicalandnitrateremovalfunctions.JAWRAJ.Am.WaterResourcesAssoc.42, 1099–1112.

Vidon,P.,Allan,C.,Burns,D.,Duval,T.P.,Gurwick,N.,Inamdar,S.,Lowrance,R., Okay,J.,Scott,D.,Sebestyen,S.,2010.Hotspotsandhotmomentsin ripar-ianzones:potentialforimprovedwaterqualitymanagement.JAWRAJ.Am. WaterResourcesAssoc.46,278–298,http://dx.doi.org/10.1111/j.1752-1688. 2010.00420.x.

Weiss,R.F.,Price,B.A.,1980.Nitrousoxidesolubilityinwaterandseawater.Mar. Chem.8,347–359,http://dx.doi.org/10.1016/0304-4203(80)90024-9. Weng,P.,Sánchez-Pérez,J.M.,Sauvage,S.,Vervier,P.,Giraud,F.,2003.Assessment

ofthequantitativeandqualitativebufferfunctionofanalluvialwetland: hydro-logicalmodellingofalargefloodplain(GaronneRiver,France).Hydrol.Processes 17,2375–2392,http://dx.doi.org/10.1002/hyp.1248.

Wetzel,R.G.,1992.Gradient-dominatedecosystems:sourcesandregulatory func-tionsof dissolvedorganicmatter infreshwater ecosystems.In:Dissolved OrganicMatterinLacustrineEcosystems.Springer,pp.181–198.

Yan,B.,Fang,N.F.,Zhang,P.C.,Shi,Z.H.,2013.Impactsoflandusechangeon water-shedstreamflowandsedimentyield:anassessmentusinghydrologicmodelling andpartialleastsquaresregression.J.Hydrol.484,26–37,http://dx.doi.org/10. 1016/j.jhydrol.2013.01.008.

Zarnetske,J.P.,Haggerty,R.,Wondzell,S.M.,Baker,M.A.,2011.Labiledissolved organiccarbonsupplylimitshyporheicdenitrification.J.Geophys.Res. Bio-geosci.(2005–2012),116.

Zarnetske,J.P.,Haggerty,R.,Wondzell,S.M.,Bokil,V.A.,González-Pinzón,R.,2012.

Coupledtransportandreactionkineticscontrolthenitratesource-sinkfunction ofhyporheiczones.WaterResourcesRes.,48.

Zhang,W.L.,Tian,Z.X.,Zhang,N.,Li,X.Q.,1996.Nitratepollutionofgroundwaterin northernChina.Agric.Ecosyst.Environ.59,223–231,http://dx.doi.org/10.1016/ 0167-8809(96)01052-3.