Biogeosciences,12,67–78,2015 www.biogeosciences.net/12/67/2015/ doi:10.5194/bg-12-67-2015 ©Author(s)2015.CCAttribution3.0License. Technical Note: Large overestimation of pCO calculated from pH 2 and alkalinity in acidic, organic-rich freshwaters G.Abril1,2,S.Bouillon3,F.Darchambeau4,C.R.Teodoru3,T.R.Marwick3,F.Tamooh3,F.OchiengOmengo3, N.Geeraert3,L.Deirmendjian1,P.Polsenaere1,andA.V.Borges4 1LaboratoireEPOC,EnvironnementsetPaléoenvironnementsOcéaniquesetContinentaux,CNRS, UniversitédeBordeaux,France 2ProgramadeGeoquímica,UniversidadeFederalFluminense,Niterói,RiodeJaneiro,Brazil 3KatholiekeUniversiteitLeuven,DepartmentofEarth&EnvironmentalSciences,Leuven,Belgium 4Unitéd’OcéanographieChimique,UniversitédeLiège,Belgium Correspondenceto:G.Abril([email protected]) Received:23June2014–PublishedinBiogeosciencesDiscuss.:31July2014 Revised:18November2014–Accepted:24November2014–Published:6January2015 Abstract. Inland waters have been recognized as a signifi- Owingtothewidespreaddistributionofacidic,organic-rich cantsourceofcarbondioxide(CO )totheatmosphereatthe freshwaters, we conclude that regional and global estimates 2 globalscale.FluxesofCO betweenaquaticsystemsandthe of CO outgassing from freshwaters based on pH and TA 2 2 atmospherearecalculatedfromthegastransfervelocityand dataonlyaremostlikelyoverestimated,althoughthemagni- thewater–airgradientofthepartialpressureofCO (pCO ). tudeoftheoverestimationneedsfurtherquantitativeanalysis. 2 2 Currently,directmeasurementsofwaterpCO remainscarce Direct measurements of pCO are recommended in inland 2 2 infreshwaters,andmostpublishedpCO dataarecalculated watersingeneral,andinparticularinacidic,poorlybuffered 2 from temperature, pH and total alkalinity (TA). Here, we freshwaters. compare calculated (pH and TA) and measured (equilibra- torandheadspace)waterpCO inalargearrayoftemperate 2 and tropical freshwaters. The 761 data points cover a wide range of values for TA (0 to 14200µmolL−1), pH (3.94 to 1 Introduction 9.17), measured pCO (36 to 23000ppmv), and dissolved 2 organic carbon (DOC) (29 to 3970µmolL−1). Calculated Inlandwaters(streams,rivers,lakes,reservoirs,wetlands)re- pCO were >10% higher than measured pCO in 60% ceive carbon from terrestrial landscapes, usually have a net 2 2 of the samples (with a median overestimation of calculated heterotrophic metabolism, and emit significant amounts of pCO2comparedtomeasuredpCO2of2560ppmv)andwere CO2totheatmosphere(Kempe1984;Coleetal.,1994;Ray- >100% higher in the 25% most organic-rich and acidic mondetal.,2013).Thisterrestrial–aquatic–atmospherelink samples (with a median overestimation of 9080ppmv). We intheglobalcarboncycleiscontrolledbycomplexbiogeo- suggesttheselargeoverestimationsofcalculatedpCO with graphical drivers that generate strong spatial and temporal 2 respecttomeasuredpCO areduetothecombinationoftwo variations in the chemical composition of freshwaters and 2 cumulativeeffects:(1)amoresignificantcontributionofor- the intensity of CO2 outgassing at the water–air interface ganicacidsanionstoTAinwaterswithlowcarbonatealka- (e.g. Tamooh et al., 2013; Dinsmore et al., 2013; Abril et linityandhighDOCconcentrations;(2)alowerbufferingca- al.,2014;Borgesetal.,2014).Hence,largedatasetsarenec- pacityofthecarbonatesystematlowpH,whichincreasesthe essaryinordertodescribetheenvironmentalfactorscontrol- sensitivity of calculated pCO2 to TA in acidic and organic- lingtheseCO2 emissionsandtoquantifyglobalCO2 fluxes richwaters.Noempiricalrelationshipcouldbederivedfrom from inland waters (Sobek et al., 2005; Barros et al., 2011; ourdatasetinordertocorrectcalculatedpCO forthisbias. Raymond et al., 2013). Dissolved inorganic carbon (DIC) 2 concentration and speciation in freshwaters greatly depend PublishedbyCopernicusPublicationsonbehalfoftheEuropeanGeosciencesUnion. 68 G.Abriletal.:OverestimationofpCO calculatedfrompHandalkalinityinfreshwaters 2 on the lithological nature of watersheds (Meybeck 1987). Alimitednumberofstudieshavecompareddirectlymea- For instance, rivers draining watersheds rich in carbonate suredpCO tocomputedpCO .Earlierexamplesprovided 2 2 rocks have a high DIC concentration, generally well above acomparisonbetweenpCO measuredbyheadspaceequili- 2 1000µmolL−1. Bicarbonate ions contribute to most of the brationcoupledtogaschromatography(GC)andpCO cal- 2 total alkalinity (TA) in these waters, which have high con- culated from pH and DIC (Kratz et al., 1997; Raymond et ductivitiesandhighpH.Inthesehardwaters,dissolvedCO al., 1997). Reports by these authors in Wisconsin lakes and 2 representsaminorfraction(5–15%)oftheDICcomparedto the Hudson River show that the pCO values were linearly 2 bicarbonates. In rivers draining organic-rich soils and non- correlated but showed a variability of ±500ppmv around carbonate rocks, DIC concentrations are lower (typically a the 1:1 line, over a range of measured pCO from 300 2 fewhundredµmolL−1)butdissolvedorganiccarbon(DOC) to 4000ppmv. Later, Frankignoulle and Borges (2001) re- concentrations are higher, and commonly exceed the DIC ported the first comparison of pCO calculated from pH 2 concentrations. Organic acid anions significantly contribute and TA and pCO measured by equilibration coupled to 2 to TA of these soft waters (Driscoll et al., 1989; Hemond an IR analyser in an estuary in Belgium. In this high TA 1990),whichhavelowconductivitiesandlowpH.Dissolved (2500–4800µmolL−1) and high pH (>7.4) system, they CO representsalarge,generallydominant,fractionofDIC found a good agreement between the two approaches, cal- 2 intheseacidic,organic-richwaters. culatedpCO beingeitheroverestimatedorunderestimated, 2 Fluxes of CO between aquatic systems and the atmo- butalwaysbylessthan7%.In2003,concomitantmeasure- 2 sphere can be computed from the water–air gradient of the mentsofpH,TAandpCO wereperformedinacidic,humic- 2 concentrationofCO andthegastransfervelocity(Lissand rich(“black”type)watersoftheSinnamaryRiverinFrench 2 Slater, 1974) at local (e.g. Raymond et al., 1997), regional Guiana (Abril et al., 2005, 2006). Calculation of pCO 2 (e.g. Teodoru et al., 2009), and global scales (e.g. Cole et from pH (∼5) and TA (∼200µmolL−1) gave unrealisti- al., 1994; Raymond et al., 2013). The partial pressure of callyhighvaluescomparedtothosemeasureddirectlywitha CO (pCO ) is relatively constant in the atmosphere com- headspacetechnique(typically30000ppmvvs.5000ppmv). 2 2 pared to surface freshwaters pCO that can vary by more DirectmeasurementsofCO andCH outgassingfluxeswith 2 2 4 than4 ordersofmagnitudespatiallyand temporally(Sobek floatingchambersandthecomputationoftherespectivegas et al., 2005; Abril et al., 2014). Consequently, water pCO transfer velocities of these two gases (Guérin et al., 2007) 2 controls the intensity of the air–water flux, together with confirmedthatpCO valuescalculatedfrompHandTAwere 2 thegastransfervelocity.Atpresent,bothmeasuredandcal- overestimated compared to direct measurements in the Sin- culated water pCO data are used to compute CO fluxes namary River. More recently, Hunt et al. (2011) and Wang 2 2 fromfreshwatersystems,althoughcalculatedpCO isover- et al. (2013) provided evidence that organic acid anions in 2 whelmingly more abundant than directly measured pCO DOC may significantly contribute to TA in some rivers and 2 (e.g.Coleetal.,1994;Raymondetal.,2013).pCO canbe generateanoverestimationofcalculatedpCO .Butmanand 2 2 calculated from the dissociation constants of carbonic acid Raymond (2011) reported higher calculated than measured (whichareafunctionoftemperature)andanyofthefollow- pCO insomeUSstreamsandrivers,butnoinformationwas 2 ingcouplesofmeasuredvariables:pH/TA,pH/DIC,DIC/TA available on the potential role of organic acids in this over- (Park, 1969). In a majority of cases, calculated pCO is estimation.Theseauthorsconcludedthatthelownumberof 2 basedonthemeasurementsofpH/TAandwatertemperature. samples in their study reflected the need for more research Thesethreeparametersareroutinelymeasuredbymanyen- onthistopic. vironmental agencies, and constitute a very large database WiththegrowinginterestonpCO determinationinfresh- 2 availableforthescientificcommunity.CalculationofpCO waters globally, and given the apparent simplicity and low 2 from pH and TA was initiated in world rivers in the 1970s costofpHandTAmeasurements,thenumberofpublications (Kempe, 1984) and relies on the dissociation constants of thatreportcalculatedpCO infreshwatershasincreaseddra- 2 carbonic acid, and the solubility of CO , all of which are maticallyinthepastdecade.Someofthesepublicationsre- 2 temperature-dependent (Harned and Scholes, 1941; Harned port extremely high and potentially biased pCO values in 2 andDavis,1943;Millero,1979;StummandMorgan,1996). low-alkalinity and high-DOC systems. It has thus become Measured pCO is based on water–air phase equilibration necessary to pay attention to this issue and investigate the 2 eitherondiscretesamples(headspacetechnique,e.g.Weiss, occurrence of such potential bias and its magnitude in the 1981) or continuously (equilibrator technique, e.g. Frankig- differenttypesoffreshwaters.Here,wepresentalargedata noulle et al., 2001) using various systems and devices, fol- set of concomitant measurements of temperature, pH, TA, lowed by direct, generally infrared (IR), detection of CO pCO , and DOC in freshwaters. This is the first compre- 2 2 intheequilibratedgas.CommercialIRgasanalysersarebe- hensivedatasettoinvestigatethemagnitudeofthebiasbe- comingcheaperandmoreaccurate,stableandcompact,and tweencalculatedandmeasuredpCO ,asitcoverstheentire 2 providealargerangeoflinearresponsewelladaptedtovari- range of variation of most parameters of the carbonate sys- abilityofpCO foundinfreshwaters. teminfreshwaters.Theobjectiveofthispaperistoalertthe 2 scientific community to the occurrence of a bias in pCO 2 Biogeosciences,12,67–78,2015 www.biogeosciences.net/12/67/2015/ G.Abriletal.:OverestimationofpCO calculatedfrompHandalkalinityinfreshwaters 69 2 Table1.Summaryofthepresenteddataset.Average,minimum,andmaximumvaluesoftemperature,DOC,pH(measuredontheNBS scale),totalalkalinity(TA),andmeasuredpartialpressureofCO2(pCO2)inthedifferentfreshwaterecosystems. Country Watersheds Temperature DOC pH TA MeasuredpCO2 N (◦C) (µmolL−1) (NBSscale) (µmolL−1) (ppmv) Av. Min. Max. Av. Min. Max. Av. Min. Max. Av. Min. Max. Av. Min. Max. Brazil Amazon 30.3 27.4 34.3 352 118 633 6.60 4.53 7.60 385 30 1092 4204 36 18400 155 Kenya Athi-Galana-Sabaki 25.9 19.8 36.0 307 29 1133 7.69 6.49 8.57 2290 407 5042 2811 608 10405 44 DRC Congo 26.3 22.6 28.2 1002 149 3968 6.01 3.94 7.22 212 0 576 6093 1582 15571 97 DRC/Rwanda LakeKivu 24.0 23.0 24.7 162 142 201 9.05 8.99 9.17 13037 12802 13338 660 537 772 53 France Leyre 12.5 7.9 19.2 588 142 3625 6.20 4.40 7.41 280 38 1082 4429 901 23047 92 France Loire 15.5 8.8 19.3 195 167 233 8.70 8.07 9.14 1768 1579 1886 284 65 717 18 Belgium Meuse 18.1 13.3 25.9 229 102 404 7.89 6.95 8.59 2769 360 7141 2292 176 10033 50 Madagascar RianilaandBetsiboka 25.4 20.2 29.5 138 33 361 6.84 5.83 7.62 233 76 961 1701 508 3847 36 Kenya ShimbaHills 25.1 21.9 31.8 214 36 548 7.37 6.22 8.93 1989 227 14244 2751 546 9497 9 FrenchGuiana Sinnamary 27.1 24.1 28.7 419 213 596 5.50 5.08 6.30 143 66 290 7770 1358 15622 49 Kenya Tana 26.6 25.0 27.9 321 193 651 7.65 7.32 8.02 1619 1338 2009 2700 845 6014 51 Zambia/Mozambique Zambezi 26.9 18.8 31.8 252 103 492 7.59 5.06 9.08 1245 52 3134 2695 151 14004 107 Entiredataset 24.6 7.9 36.0 408 29 3968 7.00 3.94 9.17 1731 0 14244 3707 36 23047 761 calculation from pH and TA in acidic, poorly buffered and withasmallproportionofcarbonateandevaporiterocksfor organic-richfreshwaters,tobrieflydiscussitsorigininterms theCongoriver. ofwaterchemistry,andtoprovidetherangeofpH,TA,and DOC values where pCO2 calculation should be abandoned 2.2 Fieldandlaboratorymeasurements andtherangewhereitstillgivesrelativelyaccurateresults. Although pH measurements might seem almost trivial, highlyaccurateandprecisepHdataareinfactnoteasytoob- 2 Materialandmethods tain,especiallyinlow-ionicstrengthwaters,whereelectrode readingsaregenerallylessstable.EventhoughpHmeasure- 2.1 Samplecollection ments in the laboratory might be more accurate, it is cru- cialtomeasurepHinsituorimmediatelyaftersampling,as Our data set consists of 761 concomitant measurements of pH determination several hours or days after sampling will temperature,pH,TA,waterpCO ,andDOCin12contrast- be affected by CO degassing and/or microbial respiration 2 2 ingtropicalandtemperatesystemsinEurope,Amazonia,and (Frankignoulle and Borges, 2001). In this work, water tem- Africa(Fig.1;Table1).Thesesampleswereobtainedinthe perature and pH were measured in the field with different CentralAmazonRiverandfloodplainssysteminBrazil,the probesdependingontheoriginofthedataset.However,all Athi-Galana-SabakiRiverinKenya,theTanaRiver(Kenya), thepHdatawereobtainedwithglasselectrodesandrelyon smallriversdrainingtheShimbaHillsinsoutheasternKenya, daily calibration with two-point United States National Bu- the Congo River and tributaries in the Democratic Repub- reauofStandards(NBS)standards(4and7).Measurements lic of the Congo (DRC), Lake Kivu in Rwanda and DRC, wereperformeddirectlyinthesurfacewater,orincollected theLeyreRiverandtributariesinFrance,theLoireRiverin waterimmediatelyaftersampling. France, the Meuse River in Belgium, the Rianila and Betsi- Several techniques were used to measure water pCO . 2 bokariversinMadagascar,theSinnamaryRiverdownstream Water–gasequilibrationwasperformedwithamarbles-type of the Petit Saut Reservoir in French Guiana, and the Zam- equilibrator (Frankignoulle et al., 2001) for the Amazon, bezi River in Zambia and Mozambique (Fig. 1). Details on Loire,Leyre,Sinnamary,andCongorivers(December2013) someofthesamplingsitescanbefoundinAbriletal.(2005, aswellforLakeKivu,orwithaLiqui-CelMiniModulemem- 2014), Borges et al. (2012, 2014), Marwick et al. (2014a, branecontactorequilibrator(seeTeodoruetal.,2009,2014) b), Polsenaere et al. (2013), Tamooh et al. (2013), Teodoru fortheZambeziandsomesiteswithintheCongobasin(De- et al. (2014). These watersheds span a range of climates cember 2012): water was pumped either continuously from and are occupied by different types of land cover, which a ship, or on an ad hoc basis from the bank of the rivers include tropical rainforest (Amazon, Congo, Rianila), dry after waiting ∼15min for complete equilibration; air was savannah (Tana, Athi-Galana-Sabaki, Betsiboka, Zambezi), continuously pumped from the equilibrator to the gas anal- temperate pine forest growing on podzols (Leyre), mixed yser (see e.g. Abril et al., 2014 for a more detailed descrip- temperateforest,grassland,andcropland(Meuse),andcrop- tion of the system). A syringe-headspace technique (Kratz land(Loire).Lithologyisalsoextremelycontrastedasitin- et al., 1997; Teodoru et al., 2009) was used in the field in cludesforinstancecarbonate-rocks-dominatedwatershedas all African rivers and in the Meuse River: 30mL volume for the Meuse, sandstone-dominated silicates (Leyre), and of atmospheric air was equilibrated with 30mL volume of precambrian crystalline magmatic and metamorphic rocks river water by vigorously shaking during 5–10min in four www.biogeosciences.net/12/67/2015/ Biogeosciences,12,67–78,2015 70 G.Abriletal.:OverestimationofpCO calculatedfrompHandalkalinityinfreshwaters 2 Figure1.LocationofthesamplingsitesinAfrica,Amazonia,andEurope. replicate gas-tight syringes. The four replicates 30mL of 2014). The gas analysers were calibrated before each field equilibrated gas and a sample of atmospheric air were in- cruise,withaircirculatingthroughsodalimeorpureN for 2 jectedinanIRgasanalyser(Li-Cor® models820or840,or zeroandwithacertifiedgasstandardforthespan.Depend- PP systems® model EGM-4); the first gas injection served ing on the cruises and expected pCO ranges, we used gas 2 as a purge for the air circuit and cell and the three other standardconcentrationof1000–2000ppmv,orasetofcali- injections were used as triplicate pCO determination (av- brationgasesat400,800,4000and8000ppmv.Stabilityof 2 erage repeatability of ±1%). The pCO in the river water theinstrumentwascheckedafterthecruise,anddeviationof 2 was deduced from that measured in the headspace account- thesignalwasalwayslessthan5%.Theseinstrumentsoffer ing for the initial pCO in the air used for equilibration, alargerangeoflinearresponse,dependingonmanufacturer 2 water temperature in the river and in the water at equilib- and model: 0–20000ppmv or 0–60000ppmv. The linearity riumin thesyringe,and basedonHenry’s law.Comparison ofanLi-COR®Li-820gasanalyserwasverifiedbyconnect- betweensyringe-headspaceandmarblesormembraneequi- ing it to a closed circuit of gas equipped with a rubber sep- librator was made during two cruises on the Congo River tumtoallowinjectionofpureCO withasyringe.Linearity 2 and three cruises in the Zambezi basin and gave very con- wascheckedbyinjectingincreasingvolumesofCO inorder 2 sistentresults,deviationfromthe1:1linebeingalwaysless tocoverthewholerangeofmeasurementandwasexcellent than15%(seeFig.2).Thishighlightstheconsistencyofthe betweenzeroand∼20000ppmv.InadditiontotheIRanal- presentdatasetofdirectpCO measurementsalthoughdif- ysers generally used in this work, in the Sinnamary River, 2 ferenttechniqueswereused.Aserumbottle-headspacetech- pCO was also measured with an INNOVA® 1312 optical 2 nique (Hope et al., 1995) was also used on the Sinnamary filter IR photoacoustic gas analyser (range 0–25000ppmv) River; surface water was sampled in 120mL serum bottles connected to an equilibrator and with a Hewlett Packard® thatwerepoisonedwithHgCl andsealedexcludingairbub- 5890gaschromatographequippedwithathermalconductiv- 2 bles.Backinthelaboratory,a40mLheadspacewascreated itydetector(TCD);bothanalyserswerecalibratedwithagas withpureN (Abriletal.,2005).TheCO concentrationof mixture of 5000ppmv of CO . Both methods gave results 2 2 2 equilibrated gas in the headspace was analysed by injecting consistentat±15%inthe0–13000ppmvrange(Abriletal., smallvolumes(0.5mL)ofgasinagaschromatographcali- 2006).Sinnamarydatareportedherearefromheadspaceand bratedwithcertifiedgasmixtures. GCdetermination. Immediatelyafterwater–gasphaseequilibration,CO was TAwasanalysedbyautomatedelectro-titrationon50mL 2 detectedandquantifiedinmostsampleswithnon-dispersive filteredsampleswith0.1NHClastitrant.Equivalencepoint IR gas analysers (Frankignoulle et al., 2001; Abril et al., was determined with a Gran method from pH between 4 Biogeosciences,12,67–78,2015 www.biogeosciences.net/12/67/2015/ G.Abriletal.:OverestimationofpCO calculatedfrompHandalkalinityinfreshwaters 71 2 15000 and Harned and Davis, 1943) and the CO2 solubility from Marbles equilibrator Weiss(1974)asimplementedintheCO2SYSprogram.Hunt Membrane equilibrator etal.(2011)reporteddiscrepancylowerthan2%forpCO 2 v) computedthiswaywiththoseobtainedwiththePHREEQC m program (Parkhurst and Appelo, 1999). Differences in soft- p p 10000 ware or dissociation constants cannot account for the large or ( bias in calculated pCO compared to measured pCO we at 2 2 r reportinthispaper. b uili q e 2 5000 3 Results O C p 3.1 Datarangesandpatternsintheentiredataset Measured pCO varied between 36ppmv in a floodplain 2 0 of the Amazon River and 23000ppmv in a first-order 0 5000 10000 15000 stream of the Leyre River (Table 1). Minimum values of pCO2 syringe (ppmv) pH and TA occurred in the Congo River (pH=3.94 and TA=0) and maximum values in Lake Kivu (pH=9.16 Figure 2. Comparison of results of different water–air equilibra- and TA=14200µmolL−1). Highest DOC concentrations tiondesignsfordirectpCO2 measurements;pCO2 measuredwith (>3000µmolL−1) were observed in small streams in the a marbles equilibrator (Congo) and with a membrane equilibrator Congo basin and in first-order streams draining podzolized (Congo and Zambezi) are plotted against pCO2 measured with a syringe headspace technique. Detection was made with an IR gas soils in the Leyre basin. Lowest DOC concentrations analyser. (<40µmolL−1) occurred in some tributaries of the Athi- Galana-Sabaki,intheRianilaandBetsibokarivers,andinthe ShimbaHillsstreams.Whenconsideringthewholedataset, and 3 (Gran, 1952). Precision based on replicate analyses measured pCO and DOC were negatively correlated with 2 was better than ±5µmolL−1. TA measurements should be pH, whereas TA was positively correlated with pH (Fig. 3, done on filtered samples; otherwise some overestimation p<0.0001forthethreevariables).Thisillustratesthelarge would occur in turbid samples, which may contain signifi- contrast in acid–base properties between acidic, organic- cantamountsofacid-neutralizingparticles(e.g.calciumcar- rich, and poorly buffered samples on one hand, and basic, bonate).IncontrasttoTAmeasurementsbasedontitrationto carbonate-bufferedsamplesontheother. an endpoint of 5.6 (e.g. Wallin et al., 2014), the Gran titra- tion method allows the determination of TA values in sam- 3.2 Comparisonbetweenmeasuredandcalculated pleswithinsitupHdownto∼4.5,i.e.veryclosetothedis- pCO2 − sociation constant of HCO /H CO . In most acidic sam- 3 2 3 CalculatedpCO wasmorethan10%lowerthanmeasured ples with low TA, reproducibility was improved by slightly 2 pCO in 16% of the samples; the two methods were con- increasing the pH by up to 0.2 units by vigorously stirring 2 during ∼15min in order to degas as much CO as possi- sistent at ±10% in 24% of the samples; calculated pCO2 2 was more than 10% higher than measured pCO in 60% blebeforestartingthetitration.DOCwasmeasuredonsam- 2 ples filtered through pre-combusted (490◦C) glass fibre fil- of the samples and more than 100% higher in 26% of the samples. Absolute values, as expressed inppmv, were terwithaporosityof0.7µmandstoredacidifiedwithultra- largely shifted towards overestimation, calculated vs. mea- pure H PO in borosilicate vials capped with polytetraflu- 3 4 suredpCO databeingwellabovethe1:1line,andcalcu- oroethylene stoppers. Analysis was performed with a Shi- 2 latedminusmeasuredpCO valuesrangingbetween−6180 madzu TOC5000 analyser based on high-temperature cat- 2 and +882022ppmv (Fig. 4). The largest overestimation alytic oxidation, after removal of dissolved CO for sam- 2 of calculated pCO occurred in the most acidic samples, plesfromAmazon,Loire,Leyre,andSinnamaryrivers.DOC 2 whereas underestimations of calculated pCO occurred in concentrationsweremeasuredwithacustomizedwetoxida- 2 neutralorslightlybasicsamples(Fig.4b).Rankingthedata tionTOCanalyser(ThermoHiperTOC,orIOAnalyticalAu- rora1030W)coupledtoaDelta+XLorDeltaVIRMS. according to the pH, TA and DOC reveals that overestima- tion of calculated pCO compared to measured pCO in- 2 2 2.3 pCO calculationfrompHandTA creased in acidic, poorly buffered waters in parallel with an 2 increase in the DOC concentration (Table 2). Discrepancies We calculated pCO from TA, pH, and temperature mea- between calculated and measured pCO were very differ- 2 2 surements using carbonic acid dissociation constants of ent from one system to another, depending on the chemical Millero(1979)(basedonthoseofHarnedandScholes,1941 status of the waters. On average at each sampled site, the www.biogeosciences.net/12/67/2015/ Biogeosciences,12,67–78,2015 72 G.Abriletal.:OverestimationofpCO calculatedfrompHandalkalinityinfreshwaters 2 (a) (c) (e) 25 000 14 5 000 12 v) 20 000 4 000 m asured pCO (pp2 1105 000000 -1TA (mmol L)14680 -1DOC (µmol L) 23 000000 e m 5 000 1 000 2 0 0 0 4 5 6 7 8 9 10 4 5 6 7 8 9 10 4 5 6 7 8 9 10 (b) (d) (f) 100 000 100 10 000 mv)10 000 10 sured pCO (pp21 000 -1TA (mmol L)0.1001 -1DOC (µmol L)1 010000 a e 100 m 0.010 10 0.001 10 4 5 6 7 8 9 10 4 5 6 7 8 9 10 4 5 6 7 8 9 10 pH pH pH Amazon - Brazil Leyre - France Shimba Hills - Kenya Athi Galana Sabaki - Kenya Loire - France Sinnamary - French Guiana Congo - Democratic Republic of Congo Meuse - Belgium Tana - Kenya Kivu - DRC / Rwanda Rianila Betsiboka - Madagascar Zambezi - Zambia / Mozambique Figure3.Plotofcarbonvariablesvs.pHinthestudiedfreshwatersystems.Toppanelsareshownwithalinearscaleandbottompanelswith alogarithmicscale;(a,b):measuredpCO2;(c,d)totalalkalinity;(e,f)DOC.ZeroTAvaluesareplottedas0.001inordertobevisibleon thelogpCO2scale.RianilaandBetsibokaareplottedtogetheralthoughtheybelongtodifferentwatershedsinMadagascar. relative overestimation of calculated pCO decreased with 4 Discussion 2 pHandTAandincreasedwithDOC(Fig.5).Overestimation ofcalculatedpCO2wasonaverage<10%intheKivuLake, 4.1 OriginofoverestimationofcalculatedpCO2 and the Meuse, Loire, Shimba Hills and Tana rivers, which all have neutral or basic pH, TA >1000µmolL−1 and low Ourdataset(Fig.3;Table1)probablycoversthefullrange tomoderateDOCconcentrations(<400µmolL−1)(Fig.5). of conditions of carbon speciation that can be encountered Incontrast,calculatedpCO wasoverestimatedby>200% 2 in continental surface waters. A pCO overestimation neg- 2 on average in the Congo, Leyre, Sinnamary and Amazon atively correlated with pH (p=0.001) and TA (p=0.005) rivers,whichhaveacidicpH,TA<500µmolL−1 andhigh- and positively correlated with DOC (p<0.001) (Fig. 5) is estDOCconcentration,reaching1000µmolL−1 onaverage consistent with the observations of Cai et al. (1998) in the intheCongo.ThecasesofAthi-Galana-Sabaki,Rianila,Bet- freshwaterend-membersofsomeestuariesinGeorgia,USA, siboka,andZambeziriverswereintermediateinpH,TAand andofHuntetal.(2011)inriversinNewEngland(USA)and DOC,andwithaverageoverestimationsofcalculatedpCO 2 New Brunswick (Canada). These authors performed NaOH of50–90%(Fig.5). back-titration in order to measure non-carbonate alkalinity (NCA).TheyfoundthatNCAaccountedforalargefraction (insomecasesthegreaterpart)ofTA;inaddition,thecontri- butionofinorganicspeciesotherthancarbonatewasassumed negligible and most of the NCA was attributed to organic Biogeosciences,12,67–78,2015 www.biogeosciences.net/12/67/2015/ G.Abriletal.:OverestimationofpCO calculatedfrompHandalkalinityinfreshwaters 73 2 Table2.MedianandaveragevaluesofDOC,pH(measuredontheNBSscale),totalalkalinity(TA),andcalculatedminusmeasuredpCO2 inthedataset. N %of cal–measpCO2 cal–measpCO2 pH TA DOC samples (ppmv) (%ofmeaspCO2) (µmolL−1) (µmolL−1) Med. Av. Med. Av. Med. Av. Med. Av. Med. Av. Allsamples 761 100% +611 +10692 +23% +194% 6.94 7.00 467 1731 315 408 Rankedbycalculated–measured pCO2as%ofmeasuredpCO2 <−10% 122 16% −540 −890 −34% −36% 7.89 7.85 1269 1766 259 275 ±10% 174 23% +15 +50 +2% +1% 7.67 7.78 1576 3735 228 273 >+10% 465 61% +2430 +17710 +72% +327% 6.52 6.49 308 972 360 497 >+50% 280 37% +5490 +28660 +162% +526% 6.18 6.14 192 460 375 567 >+100% 199 26% +9080 +39120 +270% +710% 5.89 5.96 166 364 389 602 RankedbypH pH>7 368 48% +1 +82 +1% +15% 7.82 7.92 1572 3284 231 255 pH<7 393 52% +3280 +20630 +71% +362% 6.30 6.13 232 277 413 558 pH6–7 256 34% +1580 +2710 +40% +96% 6.58 6.55 334 370 350 427 pH<6 136 18% +18410 +54486 +308% +864% 5.50 5.35 93 101 487 828 pH<5 25 3% +115580 +209910 +1645% +3180% 4.53 4.53 41 45 1427 1,843 RankedbyTA TA>2000µmolL−1 110 14% +20 +340 +2% +12% 8.58 8.47 7023 8326 163 202 TA1000–2000µmolL−1 157 21% −8 −163 −2% −9% 7.81 7.83 1566 1534 271 295 TA500–1000µmolL−1 99 13% +1307 +1900 +28% +72% 6.97 7.11 651 697 304 318 TA<500µmolL−1 395 52% +2070 +20090 +64% +350% 6.30 6.24 222 232 400 538 TA<100µmolL−1 82 11% +6840 +60560 +230% +1040% 5.50 5.35 59 56 603 988 RankedbyDOC DOC<200µmolL−1 179 24% +40 +776 +5% +62% 7.89 7.92 1579 4807 163 149 DOC200–300µmolL−1 167 22% +102 +2755 +5% +69% 7.56 7.37 1132 1259 258 252 DOC300–400µmolL−1 165 22% +887 +4473 +25% +101% 6.90 6.93 499 866 341 344 DOC>400µmolL−1 250 33% +3070 +27197 +59% +434% 6.15 6.14 200 415 555 765 DOC>800µmolL−1 79 10% +4995 +62784 +92% +886% 5.80 5.62 94 180 1099 1438 acid anions. Hunt et al. (2011) also showed that in the ab- WecomparedthemodelofDriscolletal.(1989),whichas- sence of direct titration of NCA, which is labour-intensive sumesasinglepKvalueforallorganicacids,andthetriprotic and whose precision may be poor, this parameter could be modelofHruskaetal.(2003),whichassumesthreeapparent calculatedasthedifferencebetweenthemeasuredTAandthe pK values for organic acids. These two models applied to alkalinitycalculatedfrommeasurementsofpHandDICand our pH and DOC gave very similar organic alkalinity val- the dissociation constants of carbonic acid. Using the latter ues,whichcouldbesubtractedfromthemeasuredTA.Inthe approach,Wangetal.(2013)obtainedapositivecorrelation mostacidicsamples(e.g.somesitesfromtheCongobasin), betweenNCAandDOCconcentrationsintheCongoRiver, modelledorganicalkalinitieswerelargerthanmeasuredTA evidencingthepredominantroleoforganicacidsinDICspe- and the difference was thus negative. Nevertheless, we then ciation and pH in such acidic system. Because we did not recalculated pCO from the measured pH and the TA cor- 2 directly measure DIC in this study, we could not calculate rectedfromorganicalkalinity.CalculatedpCO valuescor- 2 NCAwiththesameprocedureasthesestudies.Weattempted rected with that method were, however, still very different to calculate TA from our measured pH and pCO with the from those measured in the field, being sometimes higher 2 CO2SYSprogram.However,TAvaluescalculatedthisway andsometimeslowerthanthemeasuredpCO ,withoutany 2 wereinconsistentwithothermeasuredvariables(withsome- meaningfulpattern(indeed,correctedpCO wasnegatively 2 timesnegativevalues).Indeed,becausepHandpCO aretoo correlated(p<0.001)withmeasuredpCO ).Consequently, 2 2 interdependent in the carbonate system, very small analyti- we were unable to derive any empirical relationship to cor- calerrorsonthesevariablesleadtolargeuncertaintiesinthe rectforthebiasinpCO calculationfrompHandTA.Nev- 2 calculatedTA(CullisonGrayetal.,2011).Asecondattempt ertheless,thenegativecorrelationbetweenpHandDOCand to correct our TA data from NCA consisted in calculating positive correlation between pH and TA (Fig. 3) confirm a organic alkalinity using pH and DOC as input parameters. www.biogeosciences.net/12/67/2015/ Biogeosciences,12,67–78,2015 74 G.Abriletal.:OverestimationofpCO calculatedfrompHandalkalinityinfreshwaters 2 (a) (b) 1 000 000 900 000 700 000 800 000 500 000 600 000 300 000 v) m 100 000 p 400 000 mv) 25 000 O (p2 O (pp2 ed pC 200 000 Calculated pC 1250 000000 1:1 line ulated - measur 150 000000 c 10 000 al C 0 5 000 -5000 0 0 5000 10000 15000 20000 25000 4 5 6 7 8 9 10 Measured pCO (ppmv) pH 2 Figure4.ComparisonbetweenmeasuredandcalculatedpCO2 forthewholedataset:(a)calculatedvs.measuredpCO2,thelineshows whenmeasuredpCO2equalscalculatedpCO2;(b)thedifferencebetweencalculatedandmeasuredpCO2asafunctionofpH;samesymbols asinFig.3. (a) (b) (c) 600% 600% 600% 2 500% 500% 500% O C of p 400% 400% 400% n o 300% 300% 300% ati m sti 200% 200% 200% e ver 100% 100% 100% o % 0% 0% 0% -100% -100% -100% 4 5 6 7 8 9 10 0.01 0.10 1 10 0 200 400 600 800 1000 1200 pH TA (mmol L-1) DOC (µmol L-1) Amazon - Brazil Leyre - France Shimba Hills - Kenya Athi Galana Sabaki - Kenya Loire - France Sinnamary - French Guiana Congo - Democratic Republic of Congo Meuse - Belgium Tana - Kenya Kivu - DRC / Rwanda Rianila Betsiboka - Madagascar Zambezi - Zambia Mozambique Figure5.AveragepercentagesofpCO2 overestimation,calculatedas100×(calculatedpCO2 –measuredpCO2)/measuredpCO2,asa functionof(a)pH,(b)TA,and(c)DOC,forthe12studiedsites.Errorbarsindicatethestandarddeviationfromthemeanforeachfreshwater system. strong control of organic acids on pH and DIC speciation culated pCO with the CO2SYS program, or with any pro- 2 acrosstheentiredataset. gram that accounts only for the inorganic species that con- AsdiscussedbyHuntetal.(2011),asignificantcontribu- tribute to TA. It is thus obvious that the observed increase tionoforganicacidstoTAleadstoanoverestimationofcal- in pCO overestimation when pH decreases (Figs. 4b and 2 Biogeosciences,12,67–78,2015 www.biogeosciences.net/12/67/2015/ G.Abriletal.:OverestimationofpCO calculatedfrompHandalkalinityinfreshwaters 75 2 (a) calculation to the TA, increases exponentially when pH de- creases (Fig. 6a), i.e. it is proportional to the H+ concen- 1010 tration. To go further in this theoretical analysis, we com- -1ol) 109 dpCO2 / dTA putedthedifferencebetweenthepCO2 calculatedatagiven m mv 108 TvaAlueva(lTueA+anXdµtmheoloLn−e1)c.alTchuelasteedcaalctualatsiloignhstlryevheaiglhaenr eTxA- p TA (p 107 t(rFeimg.e6bse).nFsiotriviintsytaonfcec,ailnccurleaatesidngpTCAOb2yto5µTmAolaLt−a1ci(daivcalpuHe d 106 /2 close to the precision of TA titrations) increases the calcu- O pC 105 latedpCO2by31ppmvatpH7,by307ppmvatpH6andby d 3070atpH5.IncreasingTAby100µmolL−1(atypicalvalue 104 ofNCAfoundinfreshwaters;Driscolletal.,1994;Caietal., (b)107 1998; Hunt et al., 2011), increases the calculated pCO by 2 TA + 0.005 mmol L-1 615ppmvatpH7,by6156ppmvatpH6andby61560ppmv TA + 0.020 mmol L-1 at pH5. Note that this increase in calculated pCO is in- 106 TA + 0.100 mmol L-1 2 dependent of the chosen initial TA value. The difference TA + 0.500 mmol L-1 between calculated and measured pCO from our data set mv) showsthatanNCAcontributionaround1200µmolL−1issuf- p 105 O (p2 fimcoiestnstatmopelxepslaatinpHth<e6o,vwerheesrtiemasaatinonNCofAccaolcnutlraibteudtiopnChOig2hoefr d pC 104 than 500µmolL−1 would be necessary for several samples e at circumneutral and slightly basic pH (Fig. 5b). Samples ur s requiring this high NCA contribution are from the Athi- a me 103 Galana-Sabaki and Zambezi watersheds, and correspond to d - TA values well above 1000µmolL−1. An NCA value of ate 500µmolL−1inthesesamplesisthusplausible. cul 102 We have no definitive explanation for lower calculated Cal thanmeasuredpCO2,whichisobservedmainlyatneutralto slightlybasicpH,forexampleintheZambeziRiver(Fig.4). 101 In most of these samples, owing to the relatively high TA value, an overestimation of pH of less than 0.2 units is suf- ficient to account for the low calculated pCO compared 2 100 to measured values. In general, it is not easy to judge the 4 5 6 7 8 9 accuracy of pH measurements, especially when data come pH fromenvironmentalagencies.Thus,onefactorofvariability throughoutthedatasetaswellasinliteraturedataistheac- Figure6.SensitivityofpCO2overestimationtopH:(a)theoretical curacyofpHmeasurements–despitethecaretaken(e.g.cal- factor dpCO2/dTA, which describes the sensitivity of calculated pCO2totheTAvalue;(b)thesolidlinesshowtheincreaseincal- ibrations with NBS buffers for each day of measurements), culatedpCO2 inducedbyvariousincreasesinTA,asfunctionsof wecannotruleoutthatdriftormalfunctionofpHelectrodes pH;theselinesmimictheoverestimationofcalculatedpCO2 gen- contributes to the observed variability, constituting an addi- eratedbyincreasingcontributionsoforganicalkalinitytotheTA; tionaldisadvantagecomparedtodirectpCO measurements 2 field data(as calculated– measuredpCO2) havebeen plottedfor withverystablegasanalysers. comparison;samesymbolsasinFig.3.Notethatnegativevalues donotappearinthelogarithmicscale. 4.2 ImpactonestimatesofCO emissionsfrom 2 freshwaters 5; Table 2) is due to an increasing contribution of organic According to our analysis, overestimation of calculated acidanionstoTA.However,thiseffectisnottheonlydriver pCO is largest in acidic, poorly buffered and organic-rich 2 of the observed overestimation of pCO , which is also due waters. Consequently, the overestimation of regional and 2 toadecreaseinthebufferingcapacityofthecarbonatesys- global CO emissions computed from calculated pCO de- 2 2 tem at acidic pH. To investigate the magnitude of this sec- pends on the relative contribution of these types of waters ond effect, we calculated the factor dpCO /dTA (inppmv worldwide.Intheiranalysis,Raymondetal.(2013)havedis- 2 mol−1),whichdescribesthechangeincalculatedpCO in- carded all calculated pCO values with a pH value of less 2 2 duced by a change in TA. This factor, which is the oppo- than 5.4, as well as all pCO values above 100000ppmv. 2 site of a buffer factor as it reflects the sensitivity of pCO These criteria would exclude only 8% of samples from our 2 www.biogeosciences.net/12/67/2015/ Biogeosciences,12,67–78,2015 76 G.Abriletal.:OverestimationofpCO calculatedfrompHandalkalinityinfreshwaters 2 data set. Indeed, from our analysis, it appears that overesti- culated pCO (and hence, CO degassing rates) are over- 2 2 mation of calculated pCO occurs at pH much higher than estimated by 50 to 300% relative to direct, in situ pCO 2 2 5.4 (Figs. 4, 5 and 6; Table 2). The two techniques were measurements.Sincealargemajorityoffreshwatersystems consistent at ±10% on average in only 5 of the 12 studied globally have characteristics outside the range of applica- systems, which combine a circumneutral to basic pH with bility of pCO calculation, it appears reasonable to assume 2 a TA concentration well above 1000µmolL−1 (Fig. 5). Al- thatrecentestimatesofglobalCO emissionfromlakesand 2 though it would not be sufficient for the cases of the Zam- rivers,whicharebasedexclusivelyoncalculatedpCO data, 2 beziandAthi-Galana-Sabakirivers,whereoverestimationis are too high. We propose that while TA and pH measure- still significant, a TA value above 1000µmolL−1 appears ments remain useful to describe the aquatic chemistry, data as a more robust criterion than a pH threshold to separate onpCO shouldinthefuturerelyondirectmeasurementsof 2 calculated pCO values affected by bias from those consis- pCO .Evenifsomestudiesreportrelativelyrobustcalcula- 2 2 tent with measured pCO (Table 2). In fact, pCO calcu- tionofpCO frompHandDICmeasurements(Raymondet 2 2 2 lation from pH and TA in freshwaters historically relies on al.,1997;Kratzetal.,1997;AbergandWallin,2014),direct theoreticalbackgroundandvalidationdatainhigh-alkalinity pCO values in the field are stable, precise and straightfor- 2 waters(Nealetal.,1998),includingkarsticwaters(Kempe, wardanddonotdependonthequalityofpHmeasurements, 1975).Attheglobalscale,highTAtypicallyoccursinrivers which are often uncertain. Further, high-quality DIC mea- drainingwatershedswithasignificantproportionofcarbon- surementsareverytime-consuming,fairlycomplicatedtoset aterocks,typically>30%oftheirsurfaceareaifthecrite- upanddonotallowcontinuousmeasurementstobecarried rion of TA>1000µmolL−1 is chosen and the normalized out in a simple and straightforward fashion. Although there weathering rates of Meybeck (1987) are applied. Accord- are some practical limitations to their use in the field, sub- ingtoMeybeck(1987),theaverageanddischarge-weighted mergedIRsensors,whichallowhightemporalresolution,are TA is around 900µmolL−1 for world rivers and around alsopromising(Johnsonetal.,2010).Long-terminstrument 600µmolL−1fortropicalrivers.Amongthe25largestrivers stability and accuracy based on newly developed off-axis intheworld,15haveaTA>1000µmolL−1accordingtoCai integrated cavity output spectroscopy and cavity ring-down etal.(2008).Thetwolargestriversintheworldintermsof spectroscopytechnologiesseemtoimproveincomparisonto discharge, the Amazon and the Congo, are also well below traditional IR instruments, although the latter are more af- thislimitof1000µmolL−1 andhavelargeoverestimationin fordable,morecompactandhavelowerpowerrequirements. calculated pCO (on average 200 and 360%, respectively). Jointinternationaleffortsarenecessarytodefinethemostap- 2 Very low TA and pH and high DOC values have also been propriate protocols for the measurement of DIC parameters reportedinborealstreamsandrivers(Humborgetal.,2010; infreshwaters. Dinsmoreetal.,2012;Wallinetal.,2014). In lakes, the highest pCO values in the literature come 2 from tropical black water lakes and were also calculated rather than directly measured (Sobek et al., 2005). Calcu- Acknowledgements. The data set used in this study was collected in the framework of projects funded by the Cluster of Excel- lated pCO was 65250ppmv in Lago Tupé in the Brazil- 2 lence COTE at the Université de Bordeaux (ANR-10-LABX-45, ianAmazon,aRialakeconnectedtotheRioNegro,where, CNP-Leyre project), the European Research Council (ERC-StG according to our own data set, pH is below 5 and TA is 240002, AFRIVAL: African river basins: Catchment-scale carbon around70µmolL−1.Itwas18950ppmvinKambanainLake fluxes and transformations, http://ees.kuleuven.be/project/afrival/), inPapuaNewGuinea,correspondingtoapHvalueof6.1and the Fonds National de la Recherche Scientifique (FNRS, a TA value of 350µmolL−1 (Vyverman, 1994). This sug- CAKI, 2.4.598.07, TransCongo, 14711103), the French national gests a widespread overestimation of calculated pCO2 that agency for research (ANR 08-BLAN-0221, CARBAMA project significantlyimpactstheestimationofglobalCO2emissions http://carbama.epoc.u-bordeaux1.fr/), the Research Foundation frominlandwaters.However,apreciseanalysisbasedonex- Flanders (FWO-Vlaanderen), the Belgian Federal Science Policy act quantitative information on the relative contribution of (BELSPO-SSDprojectsCOBAFISHandEAGLES),theResearch acidic and high- and low-alkalinity waters to the total sur- Council of the KU Leuven, and the Institut de Radioprotection et SuretéNucléaire,France(FLOREproject).WethanktheHydreco face area of inland waters is necessary in order to evaluate LaboratoryinFrenchGuiana,andPatrickAlbéric(ISTOOrléans) theexactmagnitudeoftheoverestimation. whoanalysedsomeofthedatareportedhere,AuroreBeulen(ULg) forcollectionofMeusedataset,Marc-VincentCommarieu(ULg) for analytical support, two anonymous reviewers and C. W. Hunt 5 Conclusions (reviewer) for constructive comments on the previous version of thepaper.AVBisaseniorresearchassociateattheFNRS. From our analysis, it appears that the validity of calculat- ing pCO2 from pH, TA and temperature is most robust in Editedby:J.Middelburg freshwaterswithcircumneutraltobasicpHandwithTAex- ceeding1000µmolL−1.AtlowerTAandpH,however,cal- Biogeosciences,12,67–78,2015 www.biogeosciences.net/12/67/2015/
Description: