Available online at www.sciencedirect.com

(2008) 148–158www.elsevier.com/locate/marchem

Marine Chemistry 108

Copper speciation in estuarine waters by forward andreverse titrations

Juan Santos-Echeandía a,b, Luis M. Laglera a,Ricardo Prego b, Constant M.G. van den Berg a,⁎

a Department of Earth and Ocean Sciences, University of Liverpool, Liverpool, L69 3GP, UKb Instituto de Investigaciones Marinas, CSIC. 36208 Vigo, Spain

Received 14 June 2007; received in revised form 2 November 2007; accepted 13 November 2007Available online 22 November 2007

Abstract

The chemical speciation of copper in the estuarine waters of the Vigo Ria was determined by titrations with salicylaldoxime(reverse copper titrations) and with copper (forward titrations). The forward titrations quantified the concentrations of ligandspresent in excess whereas the reverse titrations demonstrated the presence of low concentrations of very strong binding ligands,approximately matching the copper concentration. The data obtained by the reverse titrations indicated that copper was about 10×stronger bound than data based on the usual forward titrations.

The copper concentration in these ria waters was low at 5 nM with a minor mid-estuarine maximum of 8 nM. These copperlevels are amongst the lowest reported for estuarine waters and therefore represent uncontaminated waters. The concentration ofinorganic copper was very low across the ria at ∼10–100 fM, except at Bouzas harbour (salinity 35.5) where it was raised to∼1 pM due to copper contamination, in waters affected by the port facilities, to total levels of 15 to 20 nM copper, exceeding theconcentration of the very strong ligand detected by the reverse titrations.© 2007 Elsevier B.V. All rights reserved.

Keywords: Copper complexation; Cathodic stripping voltammetry; Metal speciation; Organic complexing ligands

1. Introduction

Copper speciation in coastal waters is importantbecause the free copper ion is toxic to some marinemicroalgae at pM levels (Brand et al., 1986), muchlower than ambient levels of total dissolved copperwhich are typically in the nM range. Ambient levelsmay well increase because copper is increasingly usedas antifoulant on ships (Comber et al., 2002) and fish

⁎ Corresponding author.E-mail address: Vandenberg@liv.ac.uk (C.M.G. van den Berg).

0304-4203/$ - see front matter © 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.marchem.2007.11.004

farms (Braithwaite and McEvoy, 2005) since a ban onorganotins (Champ, 2000) is being enforced. Copperspeciation has been determined in various waters frommarine (Donat et al., 1994; Dryden et al., 2007; Duinkerand Kramer, 1977; van den Berg, 1984b) and freshwater(Xue and Sigg, 1993) origin. Enhanced levels of copperhave been linked to the release complexing ligands byalgae (Moffett et al., 1997), and these ligands have beenassociated with thiols (Boulegue et al., 1982; Drydenet al., 2007; Leal and Van den Berg, 1998). The ligandshave been shown to be produced by prokaryotic andeukaryotic marine microorganisms in surface waters

149J. Santos-Echeandía et al. / Marine Chemistry 108 (2008) 148–158

(Dupont et al., 2004; Gordon et al., 2000; Moffett et al.,1997) and high concentrations are known to occur insediments (Skrabal et al., 2000).

Copper speciation is complicated by the presence ofseveral, perhaps many, ligands in natural waters. For thisreason it has been advocated (Hudson et al., 2003) thattitrations are carried out over extended detection win-dows (Buckley and van den Berg, 1986), and to includea high detection window, to improve detection of strongligands.

Estuarine waters are subject to variations in the sa-linity, which affect the complex stability, and dilution ofhigh ligand concentrations in the freshwater end-mem-ber with low ligand concentrations in the seawater end-member (Gerringa et al., 1998; Shank et al., 2004; Tanget al., 2001; van den Berg et al., 1986). Total ligandconcentrations for low salinity estuarine water can be ashigh as 300 nM (Gerringa et al., 1996; Laglera and vanden Berg, 2003; Shank et al., 2004), which makes itdifficult to detect the presence of much lower concen-trations of specific ligands, even though these bind thecopper more strongly.

The concentration of copper complexing ligands andthe stability of the complexes can be determined by titra-tions with copper with detection of reactive copper byanodic stripping voltammetry (Chau et al., 1974; Ndunguet al., 2005), or cathodic stripping voltammetry (CSV)with ligand competition (Shank et al., 2004; van denBerg,1984a).

Previous work has shown that different ligands aredetected when the detection window of the analyticalmethod is varied: because of large differences in reactivitybetween weak (log K′CuL∼9–11 or αCuL∼100) andstrong (log K′CuL∼13–15 or αCuL∼106) ligands,different ligands are detected depending on the detectionwindow of the method used. The width of the detectionwindow of CSV methods is about two to three orders ofmagnitude inαCuL (Apte et al., 1988) because of the effectof the analytical error on the determination when reactiveconcentrations are b10% or N90% of the total. Theanalytical error alsomakes it difficult to resolve more thana single ligand within a detection window which meansthat ligand concentrations and stability constants are tosome extent weighted averages of all ligands within thedetection window (Miller and Bruland, 1997). The de-tectionwindow inCSVis defined by the added competingligand and its concentration (van den Berg and Donat,1992) and the detection window readily varied by chang-ing the concentration of the added ligand or choosing adifferent one.

For this study of copper speciation in Galician riawaters, we used copper titrations with detection by CSV,

and “reverse titrations” (Nuester and van den Berg, 2005).Reverse titrations work by drawing the copper out of thenatural complexes by gradually increasing the concentra-tion of the added competing ligand whilst keeping thecopper concentration constant. Salicylaldoxime (SA) wasused as the added competing ligand in all titrations. Thecomplexwith SAwas detected byCSVafter adsorption onthe electrode and was used to measure the change in thecopper concentration as it was drawn out of the naturalcomplexes. The α-coefficient by which copper is boundto ligands occurring in the natural water (αCuL) was quan-tified by comparison to αCuSA, and variation in αCuLduring the titrationwas used to calculate the concentrationof the natural ligand L and the conditional stabilityconstant, K′uL. Normal, forward, titrations were carriedout at two detection windows to obtain the excess ligandsin addition to the strong ligands detected by the reversetitrations.

2. Materials and methods

2.1. Equipment

The voltammetric system consisted of a static mercury dropelectrode (VA663, Metrohm, Switzerland) connected via anIME-663 module to a computer-controlled voltammeter(PGSTAT 10, Eco Chemie, Netherlands). The working electrodewas a hanging mercury drop electrode (HDME). The referenceelectrode was a double-junction Ag/AgCl (3 M KCl) electrode,and the counter electrodewas a platinumwire. The solutionswerestirred by a rotating PTFE rod during the deposition step of themeasurements. Sample bottles (LDPE, Nalgene) were cleaned bysoaking 1 week in 50% HCl, followed by at least 1 week in 10%HCl, and were subsequently stored partially filled with 0.01 MHCl. The voltammetric measurements were performed in a glasscell cleaned after high copper additions with 0.01 M HCl.

2.2. Reagents

Water was purified by reverse osmosis (Milli-RO, Milli-pore) followed by ion-exchange (Milli-Q). Ammonia andhydrochloric acid (both Analar grade, Merck) were purified bysubboiling distillation from a quartz condenser. Solutions ofcopper were prepared by dilution of a standard solution (BDH,SpectrosoL grade) for atomic absorption spectrometry in0.01 M HCl. The pH was fixed at 7.9 by addition of 100 μlto 10 ml of sample of a solution 1 M HEPES buffer (Analargrade) and 0.35 M ammonia. A stock solution 0.1 M of sali-cylaldoxime (SA) (Sigma) was prepared in 0.1 M HCl.

2.3. Study area

The Vigo Ria is the most southern of the Western Galicianrias (the four rias to the south of Finisterre Cape), on the

150 J. Santos-Echeandía et al. / Marine Chemistry 108 (2008) 148–158

Atlantic coast of the Iberian Peninsula (Fig. 1). It has apartially mixed estuary with a tidal range up to ∼3 m. TheGalician rias are influenced by strong coastal upwelling(Blanton et al., 1987), which brings in high concentrations ofnutrients (Prego et al., 1999) resulting in highly productivewaters allowing intensive mariculture (Tenore et al., 1995).

The Vigo Ria can be divided in three areas: inner, middleand outer. The inner part is a narrow channel formed by theestuary of the river Oitaven covering salinities from 0 toapproximately 25 at high tide. The first six samples were fromthis area, covering salinities 0–25.

The middle part is constituted by the San Simon Inlet(17 km2), a semi-closed system of low hydrodynamic energy(Romero and Prego, 1995) delimited by the River Oitavenoutflow and the Rande strait (Fig. 1). Here sample G wascollected, salinity 34. The bottom slopes down from a depthof ∼5 m near the end of the channel to a depth of 25 m nearthe Rande Strait (station H, salinity 35) where there is a

Fig. 1. Map showing the sampling stations in the Viga Ria. Stations A–F werehigh salinity waters affected by the port facilities of Bouzas.

motorway bridge. At high tide the Inlet has salinities between25 and 35. The flow through the Inlet is dominated by thetides but with a net outflow mostly from the Oitaven discharge(Gomez-Gesteira et al., 1999).

The third and outer part comprises a stratified bay where thewater composition is controlled by upwelling. Its inner boun-dary is Rande Strait and connects to the Atlantic Ocean. StationH is in this bay whilst station I is close to Bouzas harbour.

2.4. Sampling

Eight surface water samples (salinities 0–35) werecollected along the axial salinity gradient of the Ria of Vigoon 16th December 2004 following the estuarine mixing duringhigh tide (Fig. 1). A telescopic pole with a bottle holder andsensors for in situ analysis of salinity, pH and temperature wasattached to a raft and used to collect water samples. Samplelocations were chosen to reflect the whole salinity gradient.

in the estuary of the Oitaven river, and I was on the margin of the Ria in

151J. Santos-Echeandía et al. / Marine Chemistry 108 (2008) 148–158

On the same day, another sample (salinity 35.5) was takenwith the same telescopic pole in the southern shore of themiddle part of the Ria. The location was close to a shipyardarea (Bouzas port) and to the outlet of a sewage treatmentplant, which are sources of local contamination (Point I inFig. 1) (Canario et al., 2007). High copper and organic matterconcentrations are released in these waters without noticeablyaffecting the salinity.

All samples were taken in low-density polyethylene bottles(Azlon) which had been acid-washed and rinsed with Milli-Qwater. Prior to filling with sample, the bottles were conditionedthree times with the same water. On the same day, sampleswere filtered in the IIM-CSIC laboratories of Vigo through a0.2 μM polypropylene nuclepore filter (Pall Corporation)inside a “clean” laboratory (HEPA filtered air, class 1000 lab,with a class 100 laminar flow hood) to avoid contamination.Filtrates were frozen and stored at −20 °C until being thawedat room temperature for the analysis of total dissolved copperand copper titrations.

2.5. Total dissolved copper determination

The copper concentration in the filtered samples wasdetermined as before (Campos and van den Berg, 1994).Seawater was UV-digested (1 h) after acidification to pH 2.2 byaddition of 10 μl 6 M HCl per 10 ml of sample in acid cleanedsilica tubes. A 10 ml sample aliquot was pipetted into thevoltammetric cell and immediately were added in this sequence:a volume of ammonia enough to approximately neutralise thepH, HEPES buffer (final concentration 0.01 M) and SA (finalconcentration of 25 μM SA). The solution was deaerated bypurging (5 min) with nitrogen. Voltammetric parameters were:deposition 60 s at −1.1 V whilst stirring; 8 s quiescence at−0.1V; potential scan using the square-wavemodulation: 10Hz,step height 2.5 mV, pulse height 25 mV, from 0 to −0.8 V. Thesensitivity was calibrated by two standard copper additions toeach sample and each sample was measured three times.

2.6. Forward titrations of copper complexing ligands presentin excess

The procedure used for the copper titrations was similar tothat described before (Campos and van den Berg, 1994). A170-ml filtered sample was transferred to a polyethylene bottle(Nalgene) and HEPES buffer (0.01 M, pH 7.9) and SA (2 or10 μM) were added. Aliquots (10 ml) were pipetted into 15polystyrene vials (30 ml, Bibby, Sterilin), previously spikedwith copper to give a concentration range between 0 and up to400 nM (actual range depending on the initial copperconcentration and the determined ligand concentration), andallowed to equilibrate overnight (17 h) at room temperature(20 °C). Prior to the first titration the tubes were conditionedtwice with seawater containing the same range of copperconcentrations. The concentration of copper bound by SA afterequilibration was determined by CSV using an adsorptionpotential of −0.15 V.

2.7. Reverse titrations of complexing ligands already boundby copper

The procedure used for the reverse titrations of ligandscomplexed with copper was similar to that described before(Nuester and van den Berg, 2005), differing only in that SAadditions were made in individual sample cups (holding 10 mlwater) and were equilibrated overnight, whilst previously thetitrations had been carried out in a single, large, volume in thevoltammetric cell and had been equilibrated for ∼1 h aftereach SA addition. This change was made for reasons ofconvenience, to ensure equilibration, and to carry out themeasurements within a shorter time. A 140-ml filtered samplewas transferred to a polyethylene bottle and HEPES buffer(0.01 M, pH 7.9) was added. Aliquots (10 ml) were pipettedinto 13 polystyrene vials previously spiked with SA to give aconcentration range between 4 and up to 600 μM. The sampleswere allowed to equilibrate overnight at room temperature(20 °C). Prior to the first titration the tubes were conditionedtwice with seawater containing the same range of SAconcentrations. The concentration of the Cu(SA)2 complexwas determined by CSV using the square-wave modulationwith a frequency of 10 Hz and a step potential of 2.5 mV; thedeposition time was 120 s at −0.2 V, the quiescence period was10 s, and the potential scan was from −0.12 to −0.8 V.

2.8. Data handling

The forward titrations of copper complexing ligands wereevaluated as before (Campos and van den Berg, 1994) bylinear least squares regression of [Cu]labile as a function of[Cu]labile/[CuL] according to the following equation similar tothe linearised version of the Langmuir equation (Ruzic, 1982;van den Berg, 1982):

Cu½ �labileCuLn½ � ¼ Cu½ �labile

CLþ a0

CLK0CuL

ð1Þ

where α′=αCu′+αCuSA and where αCu′=[Cu′] / [Cu2+]=7 at

the high salinity end and 4 at the low salinity end at theanalytical pH (pHNBS 7.9) of the measurements (calculatedusing an ion-pairing model and constants corrected for thesalinity (Byrne and Miller, 1985; Soli and Byrne, 1989; Turneret al., 1981; van den Berg, 1984c). Ligand concentrations (CL)were computed based on the assumption that the ratio ofcopper/ligand is unity.

The plot according to Eq. (1) is straight in case a single typeof ligand is present, but it is curved when two or more ligandsbind copper simultaneously. Titrations performed with 10 μMSA gave linear plots indicating that a single, strong, liganddominated the data. The ligand concentration and theconditional stability constant were then calculated from theslope and Y-axis intercept of the linear least squares regression.The sensitivity required to calculate concentrations of Culabilewas calibrated from the slope at the end of the titration andconfirmed by an unequilibrated copper addition at the end ofthe titration.

152 J. Santos-Echeandía et al. / Marine Chemistry 108 (2008) 148–158

Values for the stability constants of the complexes CuSA andCu(SA)2, required to calculate α′ in Eq. (1) at various salinities,were calculated from (Campos and van den Berg, 1994):

log K VCuSA ¼ 10:12F0:03ð Þ � 0:37F0:02ð Þ log salinity ð2Þ

log BVCu SAð Þ2 ¼ 15:78F0:08ð Þ � 0:53F0:07ð Þ log salinity: ð3Þ

Plots of copper titrations in the presence of 2 μM SAwerefound to be curved and were fitted to a model with two ligands(L1 and L2) using an iterative calculating method (Laglera andvan den Berg, 2003; van den Berg, 1984a). Briefly, initialestimates for K1′ and CL1

were obtained through linearisationof the few first data points and K2′ and CL2

similarly from therest of the data points. Then the calculation was repeated aftercorrection for the contribution to the complexation of L1 by L2

and of L1 to L2 until convergence.Measurements of the current during reverse titrations using

SA were converted to ratios of ip/ipmax and fitted to Eq. (4)(below) as a function of the SA concentration by non-linearleast squares regression using Sigmaplot 10 (Systat Software,Inc.). An advantage of this procedure is that it is not necessaryto know the sensitivity of the method at any given SAconcentration, however, it is necessary to obtain a value foripmax which is the maximum current that is approached at highSA concentrations (Nuester and van den Berg, 2005). ipmax

was fitted along with the usual parameters (CL and K′CuL)using the following equation:

ipipmax

¼ bCuSA2SA½ �2

aCuVþ KCuSA SA½ � þ bCuSA2SA½ �2 þ K VCuLCL

1þK VCuL Cu2þ½ �� � :

ð4Þ

Free copper ion concentrations ([Cu2+]) were computed foreach iteration from a quadratic equation using the estimates forCL and K′CuL) (Nuester and van den Berg, 2005).

The data fitting using Sigmaplot of values for K′CuL andCL was tested by varying the starting conditions for the threeparameters over orders of magnitude, which demonstrated thatthe calculations were not biased.

3. Results and discussion

3.1. Reverse and forward titrations of copper complexing ligands

3.1.1. CopperThe copper concentration was 4 nM in the freshwater end,

which is low for river water. For instance, there is a con-siderably higher copper concentration (25 nM) in the outflowof the Mississippi (Shiller and Boyle, 1991).

There was a mid-estuarine maximum of 8 nM copper atsalinities 20–25. A mid-estuarine maximum is not unusual forcopper (Flegal et al., 1991; Zwolsman et al., 1997) and can bedue to benthic re-mineralisation or anthropogenic contamina-

tion. In this case anthropogenic contamination did not occur atmid-estuarine locations, however it did occur in the area ofinfluence of the shipyards and port of Bouzas causing highercopper levels (15 and 20 nM) at salinities≥35 (stationsH and I,Fig. 1) which did not extend into the estuary. Overall the lowcopper concentrations at salinities of 34 and below indicate thatthese estuarine waters are uncontaminated with copper. Theseestuarine copper levels are amongst the lowest reported, lowerfor instance than the minimum of 13 nM in Elizabeth River,Virginia (Dryden et al., 2007), or than the 8–21 nM in the Loireestuary (Waeles et al., 2004). The lack of copper contami-nation in these waters is important as the Galician rias produce250,000 ton y−1 mussels, 50% of the European mollusc pro-duction (Smaal, 2002). The low ambient copper levels explainwhy molluscs grown in this ria have less copper accumulationthan those grown in other rias (Beiras et al., 2003).

3.1.2. Reverse titrations of copper with SALigand concentrations and conditional stability constants

calculated from the reverse and forward titrations aresummarised in Table 1. Titrations of one of the samples(salinity 25) are shown in Fig. 2A (reverse) and B (forwardtitrations). A typical S-shaped curve was obtained for theincrease in the current, representing the concentration ofCuSA2, (Fig. 2A) as a function of the SA concentration. Amodelled reverse titration of seawater without natural ligandshas been plotted next to the actual titration to show the shift tohigher SA concentrations due to complexation of copper bynatural ligands. To a first estimation, the curve has been shiftedby about 1.5 log units of [SA], which is equivalent to anoverall α-coefficient of copper ([CuL]/[Cu2+]) of about 3 logunits (3 instead of 1.5 because of complexation as CuSA2). Atthe half way point of the titration in the absence of naturalligands, the concentration of the CuSA2 species is the same asthat of Cu′ and CuSA combined. From known values of αCu′,K′CuSA and β′CuSA2 it follows that log αCuSA∼4.1, whichgives an estimated value of 7.1 for log αCuL (for organiccomplexation). This means that only about 1 in 106 copperspecies is inorganic (predominantly CuCO3

0 and hydroxidespecies, and some Cu2+), the remainder being organicallybound.

The reverse titrations (ip/ipmax values) were fitted to Eq. (2)to calculate values for the ligand concentration (CL) and theconditional stability constant (log K′CuL), which are shown inTable 1 and Fig. 2. Individual values for CL and log K′CuLwere calculated rather than a value for αCuL as αCuL varieswith increasing additions of SA. This is caused by theincreasing release of free L by competition for copper by theSA additions. This calculation is inaccurate at ligandconcentrations much greater than the copper concentration,which means that the method of reverse titrations is mostsuitable for ligands with concentrations similar to that of themetal. The ligand concentrations detected by the reversetitrations were similar to the copper concentrations, i.e. theligand concentrations were low.

The complex stability of the ligands detected by the reversetitrations was very high, with log K′CuL values from 14.4 at the

Table 1Reverse titrations at constant copper

S CCu nM CL nM SD Log K′CuL SD ipmax SD pCu Log αCuL

0 4 4.11 0.18 16.7 0.17 3.6 0.47 15.2 8.35 4.5 4.09 0.49 15.8 0.13 0.79 0.08 10.8 7.410 5.6 5.14 0.82 15.6 0.25 1.27 0.15 10.7 7.315 6 5.16 0.26 16.3 0.18 1.67 0.18 10.5 8.020 7.8 6.83 0.39 15.9 0.16 1.39 0.11 10.4 7.725 7.6 7.1 0.39 15.5 0.09 2.41 0.08 10.7 7.434 5.9 6.68 1.01 14.4 0.13 1.65 0.03 13.5 6.2

Forward titrations in the presence of 10 μM SA

S Cu L SD Log K′CuL SD pCu Log αCuL

0 4 14.9 0.8 14.5 0.3 14.9 6.75 4.5 16.7 1.1 14.9 1.2 15.1 7.110 5.6 13.3 1.1 14 0.4 14.2 6.115 6 12.9 0.5 14.3 0.4 14.4 6.420 7.8 13.6 1 14.1 0.5 13.9 6.225 7.6 11.5 0.8 14.3 0.6 13.9 6.434 5.9 12.8 0.3 13.9 0.1 13.9 6.035 14.7 10.6 0.3 14.6 0.5 9.8 6.635.5 19.7 10.2 0.8 13.7 0.2 9.4 5.7

Forward titrations in the presence of 2 μM SA

S L1 SD Log K′CuL1SD L2 SD Log K′CuL2

SD pCu Log αCuL

0 32.8 2.8 13.7 0.09 96 8.4 12.1 0.24 14.6 6.25 30 6.9 13.5 0.13 92 9.4 11.5 0.12 14.2 6.010 29.8 1.1 13.1 0.03 63 10.9 11.1 0.13 13.7 5.615 31 1.9 13.1 0.05 62 16.1 11.3 0.25 13.8 5.620 34 1.5 12.9 0.04 53 6.6 10.8 0.6 13.4 5.425 21.4 8.4 13.3 0.32 49 2.9 11.7 0.17 13.6 5.734 20.2 1.8 12.9 0.08 10 3.8 11.8 0.5 13.3 5.235 15 0.5 13.4 0.05 19.6 1.6 11.4 0.1 12.5 5.635.5 18.2 0.6 13.7 0.3 29.6 3.8 11.6 0.18 12.6 6.0

Average copper and ligand concentrations (nM), and conditional stability constants, for complexes in the Viga Ria

10 uM SA 2 uM SA

S CCu nM CL nM L Log K′ L1 log K′1 L2 log K′2 Log αCuL

5.9 5.6 15.7 13.7 14.3 28.5 13.2 60.7 11.5 7.5

ipmax, CL and K′CuL calculated as indicated. Concentrations of copper (CCu) and ligands (CL) are in nM. SD are standard deviations. αCuL calculatedon basis of the ligand concentration instead of the free ligand concentration.

153J. Santos-Echeandía et al. / Marine Chemistry 108 (2008) 148–158

high salinity end to 16.7 at the low salinity end. It has beensuggested (Town and van Leeuwen, 2005) that the kinetics ofvery stable complexes should be prohibitively slow, making itunlikely that equilibrium between the natural complexes andthe added ligands is obtained. This assumes that the naturalcomplex needs to dissociate prior to the formation of thecomplex with the added ligand (SA), and that the reactionfollows the Eigen mechanism which means that the rate offormation of the natural complex (CuL) is limited by the lossof water from the inner coordination sphere of the metal(Morel and Hering, 1993). This would put an upper limit onlog KCuL of ∼14 in seawater for overnight equilibrium, withan uncertainty of about a unit in log KCuL. The averagestability of the copper complexes (log K′CuL) found here using

the reverse titrations was 15.7, and the average stability usingthe forward titrations with 10 μM SAwas 14.3, so the averageof the reverse titrations would not be in agreement with theEigen mechanism, whilst the average of the forward titrationswould be on the margin. Some of the individual log K′CuLvalues are definitely not in agreement with the Eigenmechanism.

Shortfalls of the suggested Eigen mechanism have beendiscussed (Nagai et al., 2007), indicating that if the reaction ofcomplexation proceeds via an adjunctive pathway (Morel andHering, 1993), an outer-sphere complex is not produced, andtherefore this theory would not be valid. Our data suggest thateither the Eigen mechanism is not applicable and that theexchange reaction between L and SA follows the adjunctive

Fig. 2. Reverse (A) and forward (B) titrations of copper complexationin a sample with a salinity of 25. A: The “no-ligand” curve is themodelled response for a titration of seawater without natural organicligands. The S-shaped lines are the fitted regression curves. B: Forwardtitrations at two concentrations of SA; the SA concentrations areindicated on the plot.

154 J. Santos-Echeandía et al. / Marine Chemistry 108 (2008) 148–158

route, or that the kinetics were so slow that the reaction did notreach equilibrium within 17 h. For instance, according to theEigen mechanism the reaction time should be 140 h for logKCuL=15. Experiments have shown that equilibrium in reversecopper titrations is reached within 30 min (Nuester and vanden Berg, 2005) (this experiment was at a low-ish detectionwindow), whilst in forward titrations equilibrium has variouslybeen found within 5 min (Kogut and Voelker, 2001) (again fora low detection window), 3 h (Campos and van den Berg,1994) (at a high detection window), and within 24 h (Kogutand Voelker, 2003) (at a high detection window), suggestingthat the Eigen mechanism may be inconsistent with theexperimental data for copper, as it is for iron (Nagai et al.,2007). However, to fully settle this question is may benecessary to carry out equilibration experiments at a very highdetection window to verify the reaction time of the very strongfraction of copper species.

In view of the variation in some of the samples in this seriesit is likely that not all the variability in the log K values here

was due to the salinity gradient and that some of the variationmay be due to other competing reactions or to differences inthe ligands. However, the range found happens to be verysimilar to that in at least some other waters confirming thatmajor ion competition does have a major effect on the stabilityof the organic copper species. For instance, the log K′CuLvalues for strong copper complexing ligands in the Scheldtestuary showed a similar range from 16 at low salinity to 14.5at high salinity (Laglera and van den Berg, 2003). It appearstherefore that the strong ligands in the freshwater end-memberhave a complex stability (log K′CuL) of 16–16.7, which islowered to about 14.4 in the seawater end-member as a resultof major cation competition.

3.1.3. Forward titrations with copperThe plots for the forward titrations (peak height, equivalent

to the concentration of CuSA2, as a function of the copperconcentration, Fig. 2B) showed one or two approximatelylinear sections depending on the heterogeneity of the ligandslying into the detection window. When more than one ligandwas present, the plots according to Eq. (1) showed a shortlinear section at low concentrations followed by a gradualincrease as the ligand became saturated followed by a secondlinear section. The first linear section was absent at 10 μM SAand only a strong ligand could be resolved. The first linearsection at the bottom of the curved section was apparent in allthe titrations with copper in the presence of 2 μM SA. Ligandsof low log K'CuL that were out competed by SA at aconcentration of 10 μMmay at a lower concentration competewith SA for the copper additions and be resolved. Similarly,complexes of very high αCuL that were not disturbed by the SAadditions, do not contribute to the calculation of K′CuL1. In thepresence of heterogeneous ligands, this fraction is greater at alower centre of the detection window, leading to higher L1

concentrations and a lower value for K′CuL1.

The titrations in the presence of 10 μM SAwere evaluatedon basis of a single natural ligand (L1). The concentration ofthis ligand decreased from ∼15 nM at low salinity to 10 nM athigh salinity, and the complex stability (log K′CuL) from 14.5to 13.7.

The titrations in the presence of 2 μM SAwere interpretedon basis of two natural ligands because the plots according toEq. (1) were curved. The concentration of the stronger ligandof the two (L1 type) varied between 33 nM (at low salinity) and15 nM (high salinity), and the complex stability (log K′CuL)between 13.7 and 13.4. The concentration of the weaker ligand(L2 type) varied between 96 (high salinity) and 20 nM (lowsalinity), with log K′CuL values between 12.1 and 11.6.

The overall alpha coefficients (αCuL) of the ligands forcopper complexation were highest for the ligands detected bythe reverse titrations (average log αCuL=7.5), followed by theligands detected at 10 μM SA (6.4 same salinity range) andthose at 2 μM SA (5.7 same salinity range). In spite of thehigher α-coefficients, the calculated concentration of freecopper was greater when estimated on basis of the reversetitrations alone, because the copper concentration in some ofthe samples was slightly greater than the ligand concentration.

Fig. 3. Variation in the concentrations of copper and the complexingligands as a function of the salinity in the VigoRia. The ligands detectedby the reverse and forward titrations (at 10 and 2 μM SA) are labelledon the plot. Only one ligand was calculated from the reverse titrationsand from the 10 μMSAdata, whilst two ligands were apparent from the2 μM SA data. The increases in the concentrations of copper and L2 inthe high salinity water were in waters affected by port facilities.

155J. Santos-Echeandía et al. / Marine Chemistry 108 (2008) 148–158

This excess copper of course became paired with the excessligands detected by the forward titrations, but that was notapparent from the reverse titrations alone. The copper that wasbound by the strong ligands detected by the reverse titrationswas complexed more strongly than the remainder of the copperthat was bound by the excess ligands in the same samples.

3.2. Comparison reverse and forward titrations

The ligands detected by the reverse titrations were notapparent in the usual, forward, titrations with copper,suggesting that some of them bound copper so strongly thatthey were outside the detection window even at 10 μM SA.The average log αCuL of the ligands detected by the reversetitrations was 7.5, whereas it was 6.4 for the ligands detectedby the forward titrations in the presence of 10 μM SA. Thedifference was greater when the detection window was shiftedlower at 2 μM SA when the detected log αCuL was 5.7 for L1

type ligands and 4.2 for L2 type ligands.The systematic shift in the detected complex stabilities

demonstrate that copper speciation in estuarine waters is go-verned by several, perhaps many, ligands with decreasingcomplex stability at increasing ligand concentration.

The physical meaning of the differences in complexationdetected by the different methods is to a first approximation,that the reverse titrations detect ligands already complexedwith copper, whilst the forward titrations include the complex-ing ability of ligands present in excess to (incremental to) thecopper originally present in the water. This means that, ac-cording to the reverse titrations, the copper in the ria wasbound with an average log αCuL of 7.5, whereas any incre-mental copper, up to an average ligand concentration of ∼14nM, would be bound with a log αCuL of 6.4 by ligands detectedby the forward titrations in the presence of 10 μM SA. Copperconcentrations N∼14 nM up to ∼29 nM would be bound witha log αCuL of 5.7 by ligands of type L1 detected in the presenceof 2 μM SA. At even higher copper concentrations, thesubsequent 61 nM of copper up to a total of 90 nM (L2

detected with 2 μM SA) would be bound with a log αCuL of4.2. This representation is valid to a first approximation asthere is a smooth transition in the α-coefficient for coppercomplexation when a strong ligand is nearly saturated.

The data suggests that the stability of complexed copper inestuarine waters is underestimated if it is estimated only usingthe usual forward titrations. On the other hand, the overallligand concentration, specifically the weaker ligands present inexcess to the copper concentration, is underestimated by thereverse titrations.

The complexing effect of the weaker ligands on copperspeciation is minor compared to that of the stronger ligands, so ifthe main aim is to determine copper speciation then it is notnecessary to use forward titrations with a low detection window.

3.3. Ligand distribution in the Vigo Ria

The most obvious variation in the ligand concentrations isthe systematic decrease in L2 with increasing salinity (Fig. 3)

from 95 nM at low salinity to 10 nM at salinity 34. Thedistribution of L2 is not driven by copper as it is in great excess tothe copper concentration. The concentration of L2 increased atsalinities 35 and 35.5 near the outflow of sewage and shipyardand port effluents of Bouzas harbour. This increase was eitherdue to in situ ligand production by the local community ofplanktonic algae as shown for algae in culture (Dupont et al.,2004), or if the local copper source (either the port facilities orsewage inputs) were to be associated with a ligand source. Thecopper input from the Vigo sewage effluent amounted to136mol d−1 andwas associated with a similar amount of ligands124 mol d−1. The flux of copper was therefore matched by asimilar ligand flux which explains the increase in both in theseawater. The total copper concentration (dissolved andparticulate) was about 700 nM in the Vigo sewage (Santos-Echeandia et al., in press), whichmeans that thiswas diluted by afactor of ∼100–200 by seawater to obtain the increase seen inthese waters. This large dilution factor would correspond withthe lack of an apparent decrease in the salinity.

The complex stability of L1 (average log K′CuL1∼13.2) wassimilar to that for thiols in the Scheldt estuary (logK′CuThiol∼13(Laglera and van den Berg, 2006)) indicating that the L1 ligandsmay be thiol compounds.

The decrease in the concentration of L1 with increasingsalinity from 33 to 20 nM was less than that of L2 (from 96 to10 nM), causing the concentration of L1 to be greater than that ofL2 at the high salinity end (salinity 34) of the ria (not includingthe contaminated waters at Bouzas in this discussion).

The concentration of L detected by 10 μM SA decreasedslightly from 15 to 13 nM at salinity 34, and the concentrationof L (reverse), the ligand detected by the reverse titrations, didnot show a systematic trend as it was co-varying with copper.

156 J. Santos-Echeandía et al. / Marine Chemistry 108 (2008) 148–158

Concentrations of ligands in the samples of salinities 35 and35.5 were determined by the forward titrations, but not by thereverse method due to a lack of sample. In spite of their highsalinity, these samples were affected by localised coppercontamination due to the proximity of port facilities. The con-centration of L2, but not that of L1 and L, was greater in thecopper contaminated samples. Ligands have been shown to beproduced by phytoplankton in response to increased copperlevels (Croot et al., 2000; Moffett and Brand, 1996), probablyof thiol type (Dupont et al., 2004; Leal et al., 1999). L2 in thisstudy has a lower complex stability (log K′CuL∼11.5) thanexpected for thiols in estuarine waters (log K′CuThiol∼13(Laglera and van den Berg, 2006)) suggesting that perhaps theirsource is related to sewage outlets coming into the ria at theport, rather than from localised production. Thiols would haveshown up as a peak in the CSV scans (Laglera and van denBerg, 2006) so the lack of this peak suggests that thiols were notpresent in these ria waters at nanomolar concentration.

3.4. Influence of the various ligands on copper speciation inria waters

The ligand detected by the reverse titrations has the lowestconcentration, but binds the majority of copper because of thehigh stability of its complex. Part of these ligands was notapparent from the usual forward titrations even at high SAconcentration. Higher detection windows cannot be used inforward titrations because the analytical error becomes greaterthan the ligand concentration at high additions of copper.Therefore, the reverse titration is the more suitable method todetermine those ligands which have a very high complex sta-bility and which are present at the same or lower concentrationthan that of total copper but dominate its speciation. For in-stance for average ria water containing 6 nM copper, 5.4 nM(91%) occurs as CuL (reverse), whilst most of the remainder isbound by the fraction defined as CuL (10 μM SA) which wasnot resolved by the reverse titration. The concentration ofinorganic copper was around 10–100 fM (equivalent to a pCuof 14–15).

Acknowledgements

The visit by Juan Santos-Echeandía to LiverpoolUniversity was funded by a Marie Curie Host Fellowship(Contract No HPMT-CT-2001-00218). The research ofvan denBerg and Laglerawas funded byNERCgrant NE/B000176/1 and that of Prego by CICYT (“Biogeochem-ical Budget and Fluxes of Metals in a Ría”, REN2003-04106-C03).

References

Apte, S.C., Gardner, M.J., Ravenscroft, J.E., 1988. An evaluation ofvoltammetric titration procedures for the determination of tracemetal complexation in natural waters by use of computer simu-lation. Analytica Chimica Acta 212, 1–21.

Beiras, R., Bellas, J., Fernandez, N., Lorenzo, J.I., Cobelo-Garcia, A.,2003. Assessment of coastal marine pollution in Galicia (NWIberian Peninsula); metal concentrations in seawater, sedimentsand mussels (Mytilus galloprovincialis) versus embryo-larval bio-assays using Paracentrotus lividus and Ciona intestinalis. MarineEnvironmental Research 56 (4), 531–553.

Blanton, J.O., et al., 1987. The relationship of upwelling to musselproduction in the rias on the western coast of Spain. Journal ofMarine Research 45 (2), 497–511.

Boulegue, J., Lord, C.J., Church, T.M., 1982. Sulfur speciation andassociated trace metals (Fe, Cu) in the pore waters of Great Marsh,Delaware. Geochimica et Cosmochimica Acta 46, 453–464.

Braithwaite, R.A., McEvoy, L.A., 2005. Marine biofouling on fishfarms and its remediation, advances in marine biology. Advancesin Marine Biology 47, 215–252.

Brand, L.E., Sunda, W.G., Guillard, R.R.L., 1986. Reduction of marine-phytoplankton reproduction rates by copper and cadmium. Journal ofExperimental Marine Biology and Ecology 96 (3), 225–250.

Buckley, P.J.M., van den Berg, C.M.G., 1986. Copper complexationprofiles in the Atlantic Ocean: a comparative study usingelectrochemical and ion exchange techniques. Marine Chemistry19 (3 SU-), 281–296.

Byrne, R.H., Miller, W.L., 1985. Copper(II) carbonate complexation inseawater. Geochimica et Cosmochimica Acta 49, 1837–1844.

Campos, M.L.A.M., van den Berg, C.M.G., 1994. Determination ofcopper complexation in sea water by cathodic stripping voltam-metry and ligand competition with salicylaldoxime. AnalyticaChimica Acta 284, 481–496.

Canario, J., Prego, R., Vale, C., Branco, V., 2007. Distribution ofmercury and monomethylmercury in sediments of Vigo Ria, NWIberian Peninsula. Water Air and Soil Pollution 182 (1–4), 21–29.

Champ, M.A., 2000. A review of organotin regulatory strategies,pending actions, related costs and benefits. Science of the TotalEnvironment 258 (1–2), 21–71.

Chau, Y.K., Gachter, R., Lum-Shue-Chan, K., 1974. Determination ofthe apparent complexing capacity of lake waters. Journal of theFisheries Research Board of Canada 31, 1515–1519.

Comber, S.D.W., Gardner, M.J., Boxall, A.B.A., 2002. Survey of fourmarine antifoulant constituents (copper, zinc, diuron and Irgarol1051) in two UK estuaries. Journal of Environmental Monitoring 4(3), 417–425.

Croot, P.L., Moffett, J.W., Brand, L.E., 2000. Production of extracellularCu complexing ligands by eucaryotic phytoplankton in response toCu stress. Limnology and Oceanography 45 (3), 619–627.

Donat, J.R., Lao, K.A., Bruland, K.W., 1994. Speciation of dissolvedcopper and nickel in South San Francisco Bay: a multi-methodapproach. Analytica Chimica Acta 284 (3), 547–571.

Dryden, C.L., Gordon, A.S., Donat, J.R., 2007. Seasonal survey ofcopper-complexing ligands and thiol compounds in a heavilyutilized, urban estuary: Elizabeth River, Virginia. Marine Chem-istry 103 (3–4), 276–288.

Duinker, J.C., Kramer, C.J.M., 1977. Experimental-study on specia-tion of dissolved zinc, cadmium, lead and copper in River Rhineand North-Sea water, by differential pulsed anodic-strippingvoltammetry. Marine Chemistry 5 (3), 207–228.

Dupont, C.L., Nelson, R.K., Bashir, S., Moffett, J.W., Ahner, B.A.,2004. Novel copper-binding and nitrogen-rich thiols produced andexuded by Emiliania huxleyi. Limnology And Oceanography 49(5), 1754–1762.

Flegal, A.R., Smith, G.J., Gill, G.A., Sanudowilhelmy, S., Anderson,L.C.D., 1991. Dissolved trace-element cycles in the San FranciscoBay estuary. Marine Chemistry 36 (1–4), 329–363.

157J. Santos-Echeandía et al. / Marine Chemistry 108 (2008) 148–158

Gerringa, L.J.A., Poortvliet, T.C.W., Hummel, H., 1996. Comparisonof chemical speciation of copper in the Oosterschelde andWesterschelde estuaries, The Netherlands. Estuarine Coastal andShelf Science 42 (5), 629–643.

Gerringa, L.J.A., Hummel, H., Moerdijk-Poortvliet, T.C.W., 1998.Relations between free copper and salinity, dissolved andparticulate organic carbon in the Oosterschelde and Westerschelde,Netherlands. Journal of Sea Research 40 (3–4), 193–203.

Gomez-Gesteira, M., et al., 1999. A two-dimensional particle track-ing model for pollution dispersion in A Coruna and Vigo Rias(NW Spain). Oceanologica Acta 22 (2), 167–177.

Gordon, A.S., Donat, J.R., Kango, R.A., Dyer, B.J., Stuart, L.M.,2000. Dissolved copper-complexing ligands in cultures of marinebacteria and estuarine water. Marine Chemistry 70 (1–3),149–160.

Hudson, R.J.M., Rue, E.L., Bruland, K.W., 2003. Modelingcomplexometric titrations of natural water samples. EnvironmentalScience & Technology 37 (8), 1553–1562.

Kogut, M.B., Voelker, B.M., 2001. Strong copper-binding behavior ofterrestrial humic substances in seawater. Environmental Science &Technology 35 (6), 1149–1156.

Kogut, M.B., Voelker, B.M., 2003. Kinetically inert Cu in coastalwaters. Environmental Science & Technology 37 (3), 509–518.

Laglera, L.M., van den Berg, C.M.G., 2003. Copper complexation bythiol compounds in estuarine waters. Marine Chemistry 82 (1–2),71–89.

Laglera, L.M., van den Berg, C.M.G., 2006. Photochemical oxidationof thiols and copper complexing ligands in estuarine waters.Marine Chemistry 101 (1–2), 130–140.

Leal, M.F.C., Van den Berg, C.M.G., 1998. Evidence for strong copper(I) complexation by organic ligands in seawater. AquaticGeochemistry 4 (1), 49–75.

Leal, M.F.C., Vasconcelos, M.T.S.D., van den Berg, C.M.G., 1999.Copper-induced release of complexing ligands similar to thiols byEmiliania huxleyi in seawater cultures. Limnology and Oceanography44 (7), 1750–1762.

Miller, L.A., Bruland, K.W., 1997. Competitive equilibration techniquesfor determining transition metal speciation in natural waters:evaluation using model data. Analytica Chimica Acta 343 (3),161–181.

Moffett, J.W., Brand, L.E., 1996. Production of strong, extracellularCu chelators by marine cyanobacteria in response to Cu stress.Limnology and Oceanography 41 (3), 388–395.

Moffett, J.W., Brand, L.E., Croot, P.L., Barbeau, K.A., 1997. Cuspeciation and cyanobacterial distribution in harbors subject toanthropogenic Cu inputs. Limnology and Oceanography 42 (5),789–799.

Morel, F.M.M., Hering, J.G., 1993. Principles and Applications ofAquatic Chemistry. J. Wiley & Sons, New York. 588 pp.

Nagai, T., Imai, A., Matsushige, K., Yokoi, K., Fukushima, T., 2007.Dissolved iron and its speciation in a shallow eutrophic lake and itsinflowing rivers. Water Research 41 (4), 775–784.

Ndungu, K., Hurst, M.P., Bruland, K.W., 2005. Comparison of copperspeciation in estuarine water measured using analytical voltam-metry and supported liquid membrane techniques. EnvironmentalScience & Technology 39 (9), 3166–3175.

Nuester, J., van den Berg, C.M.G., 2005. Determination of metalspeciation by reverse titrations. Analytical Chemistry 77 (1), 11–19.

Prego, R., Barciela, M.D., Varela, M., 1999. Nutrient dynamics in theGalician coastal area (Northwestern Iberian Peninsula): do the RiasBajas receive more nutrient salts than the Rias Altas? ContinentalShelf Research 19 (3), 317–334.

Romero, B., Prego, R., 1995. La ensenada de San Simon:caracter-ísticas generales de una cuenca costera. Series de QuimicaOceanografica 1, 35–61.

Ruzic, I., 1982. Theoretical aspects of the direct titration of naturalwaters and its information yield for trace metal speciation.Analytica Chimica Acta 140, 99–113.

Santos-Echeandia, J., Laglera, L.M., Prego, R. and van den Berg, C.M.G.,in press. Copper speciation in continental inputs to the Vigo Ria:Sewage discharges versus river fluxes. Marine Pollution Bulletin,doi:10.1016/j.marpolbul.2007.10.021.

Shank, G.C., Skrabal, S.A., Whitehead, R.F., Kieber, R.J., 2004.Strong copper complexation in an organic-rich estuary: theimportance of allochthonous dissolved organic matter. MarineChemistry 88 (1–2), 21–39.

Shiller, A.M., Boyle, E.A., 1991. Trace elements in the MississippiRiver Delta outflow region: behavior at high discharge. Geochi-mica et Cosmochimica Acta 55 (11), 3241–3251.

Skrabal, S.A., Donat, J.R., Burdige, D.J., 2000. Pore water distributionsof dissolved copper and copper-complexing ligands in estuarineand coastal marine sediments. Geochimica Et Cosmochimica Acta64 (11), 1843–1857.

Smaal, A.C., 2002. European mussel cultivation along the Atlanticcoast: production status, problems and perspectives. Hydrobiologia484 (1–3), 89–98.

Soli, A.L., Byrne, R.H., 1989. Temperature dependence of Cu(II)carbonate complexation in natural seawater. Limnology and Ocean-ography 34, 239–244.

Tang, D.G., Warnken, K.W., Santschi, P.H., 2001. Organic complexa-tion of copper in surface waters of Galveston Bay. Limnology andOceanography 46 (2), 321–330.

Tenore, K.R., et al., 1995. Fisheries and oceanography off Galicia, NwSpain — mesoscale spatial and temporal changes in physicalprocesses and resultant patterns of biological productivity. Journalof Geophysical Research-Oceans 100 (C6), 10943–10966.

Town, R.M., van Leeuwen, H.P., 2005. Measuring marine iron(iii)complexes by CLE-AdSV. Environmental Chemistry 2 (2), 5.

Turner, D.R., Whitfield, M., Dickson, A.G., 1981. The equilibriumspeciation of dissolved components in freswater and seawater at25 °C at 1 atm. pressure. Geochimica et Cosmochimica Acta 45,855–882.

van den Berg, C.M.G., 1982. Determination of copper complexationwith natural organic ligands in seawater by equilibration withMnO2. I. Theory. Marine Chemistry 11, 307–322.

van den Berg, C.M.G., 1984a. Determination of the complexing capacityand conditional stability constants of complexes of copper(II) withnatural organic ligands in seawater by cathodic stripping voltammetryof copper–catechol complex ions. Marine Chemistry 15, 1–18.

van den Berg, C.M.G., 1984b. Organic and inorganic speciation ofcopper in the Irish Sea. Marine Chemistry 14 (3), 201–212.

van den Berg, C.M.G., 1984c. Speciation of boron with Cu2+, Zn2+,Cd2+ and Pb2+ in 0.7 M KNO3 and in sea-water. Geochimica etCosmochimica Acta 48 (12), 2613–2617.

van den Berg, C.M.G., Buckley, P.J.M., Huang, Z.Q., Nimmo, M.,1986. An electrochemical study of the speciation of copper, zincand iron in two estuaries in England. Estuarine, Coastal and ShelfScience 22 (4), 479.

van den Berg, C.M.G., Donat, J.R., 1992. Determination and dataevaluation of copper complexation by organic ligands in sea waterusing cathodic stripping voltammetry at varying detectionwindows. Analytica Chimica Acta 257, 281–291.

Waeles, M., Riso, R.D., Maguer, J.F., Le Corre, P., 2004. Distributionand chemical speciation of dissolved cadmium and copper in the

158 J. Santos-Echeandía et al. / Marine Chemistry 108 (2008) 148–158

Loire estuary and North Biscay continental shelf, France. EstuarineCoastal and Shelf Science 59 (1), 49–57.

Xue, H., Sigg, L., 1993. Free cupric ion concentration and Cu(II)speciation in a eutrophic lake. Limnology and Oceanography 38,1200–1213.

Zwolsman, J.J.G., Van Eck, B.T.M., Van der Weijden, C.H., 1997.Geochemistry of dissolved trace metals (cadmium, copper, zinc) inthe Scheldt estuary, southwestern Netherlands: impact of seasonalvariability. Geochimica Et Cosmochimica Acta 61 (8), 1635–1652.