www.elsevier.com/locate/marpolbul

Marine Pollution Bulletin 54 (2007) 1634–1644

Air–sea gaseous exchange of PCB at the Venice lagoon (Italy)

L. Manodori a, A. Gambaro a,b,*, I. Moret a,b, G. Capodaglio a,b, P. Cescon a,b

a Environmental Sciences Department, Ca’ Foscari University of Venice, 30123 Venice, Italyb Institute for the Dynamics of Environmental Processes, C.N.R., 30123 Venice, Italy

Abstract

Water bodies are important storage media for persistent organic pollutants (POPs) such as polychlorinated biphenyls (PCBs) and thisfunction is increased in coastal regions because their inputs are higher than those to the open sea. The air–water interface is extensivelyinvolved with the global cycling of PCBs because it is the place where they accumulate due to depositional processes and where they maybe emitted by gaseous exchange. In this work the parallel collection of air, microlayer and sub-superficial water samples was performed inJuly 2005 at a site in the Venice lagoon to evaluate the summer gaseous flux of PCBs. The total concentration of PCBs (sum of 118 cong-eners) in air varies from 87 to 273 pg m�3, whereas in the operationally defined dissolved phase of microlayer and sub-superficial watersamples it varies from 159 to 391 pg L�1. No significant enrichment of dissolved PCB into the microlayer has been observed, although apreferential accumulation of most hydrophobic congeners occurs. Due to this behaviour, we believe that the modified two-layer modelwas the most suitable approach for the evaluation of the flux at the air–sea interface, because it takes into account the influence of themicrolayer. From its application it appears that PCB volatilize from the lagoon waters with a net flux varying from 58 to 195 ng m�2 d�1

(uncertainty: ±50–64%) due to the strong influence of wind speed. This flux is greater than those reported in the literature for the atmo-spheric deposition and rivers input and reveals that PCB are actively emitted from the Venice lagoon in summer months.Crown Copyright � 2007 Published by Elsevier Ltd. All rights reserved.

Keywords: Air–water interface; Gaseous fluxes; Polychlorinated biphenyls; Pressurized solvent extraction; Venice lagoon

1. Introduction

Persistent organic pollutants (POPs) are xenobioticchemical substances characterized as being persistent, bio-accumulating, toxic with endocrine disrupting properties,and prone to undergo long-range atmospheric transport(Lerche et al., 2002). The air–water interface is the placewhere they are exchanged between the atmosphere andthe aquatic systems, via different processes such as diffusivevapour exchange, the precipitation scavenging of vapoursand particle-sorbed chemicals and dry deposition with par-ticles. These phenomena are fundamental in the global cyc-lic of POP and are largely controlled by temperature(Wania et al., 1998) which controls their seasonal and lat-

0025-326X/$ - see front matter Crown Copyright � 2007 Published by Elsevi

doi:10.1016/j.marpolbul.2007.06.012

* Corresponding author. Address: Environmental Sciences Department,Ca’ Foscari University of Venice, 30123 Venice, Italy. Tel.: +390412348950; fax: +39 41 2348549.

E-mail address: gambaro@unive.it (A. Gambaro).

itudinal distribution (Wania and Mackay, 1996). The pres-ence of a sea surface microlayer (SML) may influence thedynamics of the exchange as Wurl et al. (2006a) recentlyobserved especially for the most hydrophobic compounds.

The water masses act as important store medium ofPOPs and may interact with other environmental compart-ments to transport or sink them (Iwata et al., 1993; Juradoet al., 2004; Wania and Daly, 2002). In this framework thecoastal areas are very critical regions because they tend toreceive higher depositions than those in the open sea due tothe close land-based sources; they also receive loads fromriverine input and from direct emissions (Wania et al.,1998). The Venice lagoon is a typical example of this kindof environment since it is surrounded by a highly industri-alized and populated mainland the impact of which hasbeen extensively demonstrated. Inputs deriving from atmo-spheric depositions have been quantified by Rossini et al.(2001) but there is a general lack of knowledge about thegaseous transfers that occur at the air–water interface.

er Ltd. All rights reserved.

L. Manodori et al. / Marine Pollution Bulletin 54 (2007) 1634–1644 1635

The net flux deriving from this process is demonstrated tobe not negligible, especially for hydrophobic, low vapourpressure organic chemicals such as polychlorinated biphe-nyls (PCB) (Baker and Eisenreich, 1990; Mackay and Pat-erson, 1986; Tolosa et al., 1997; Wurl et al., 2006a). Theyare a well-known class of persistent organic compoundswidely detected in all environmental matrices although theywere banned in the late Seventies and their worldwide pro-duction presumably ended in 1993 (Breivik et al., 2002). Inthis study the concentrations of PCBs have been investi-gated in the air, sea microlayer and underlying water ofthe Venice lagoon with the aim of evaluating their diffusiveflux at the air–water interface.

2. Experimental

2.1. Sampling and analysis

The air and water samplings were performed from 6 Julyto 5 August 2005 in the Northern part of the Venice lagoonnear the island of Mazzorbo (Fig. 1). The high volumeaerosol sampler (Tisch Environmental Inc; flow: 0.32 ±0.08 m3/min) was positioned at ground level in a meadowfacing the lagoon (N 45�29 009.700 E 12�24 012.700) andequipped with a quartz fibre filter (QFF; size 102 mm,SKC) followed by one and half polyurethane foam plugs

Fig. 1. Map showing the locati

(PUF; height 75 mm, diameter 65 mm, SKC). Filters andPUFs were replaced every 2–3 d, individually wrapped inaluminium foil and stored at �20 �C until analysis.

The sea microlayer (thickness: about 50 lm) and sub-superficial water (SSW; depth about 30 cm from the sur-face) samples were parallel collected twice weekly in thenearby area (average depth: 50–80 cm) by means of anautonomous floating device previously described (Mano-dori et al., 2006a). Full details of the sampling campaignsand the main meteorological information are reported inTable 1.

Before use, QFFs were furnace-treated at 400� for 5 hand PUFs were pre-cleaned by pressurized solvent extrac-tion (One-PSE, Applied Separations). The extraction wasdone in a stainless steel vessel (33 mL) using toluene (twocycles) and an n-hexane and dichloromethane mixture(1:1; one cycle); the working conditions were: temperature:100 �C; pressure: 100 bar; static duration: 5 min; solventflow: 20 s; gas flow (nitrogen): 2 min. The same operatingconditions were applied for sampled PUFs, which wereextracted with the mixture n-hexane and dichloromethane(three cycles).

Immediately after sampling, SML and SSW samples(10 L) were filtered with pre-heated (400 �C for 5 h) glassfibre filters (GF/F Whatman International Ltd.) andliquid/liquid extracted in continuous for about 24 h with

on of the sampling station.

Table 1Summary of the campaigns performed, with relative sampled volumes, average air temperature and average wind speed

Water sampling Air sampling

Sampleno.

Mean windspeed (m s�1)

Sampleno.

Volumecollected (m3)

Mean airtemperature (�C)

Mean windspeed (m s�1)

A1 6 July–8 July 2005 670 20.7 3.5A2 8 July–11 July 2005. 1194 20.6 3.0A3 11 July–13 July 2005 575 23.1 2.6

W1 13 July 2005 2.7 A4 13 July–15 July 2005 399 25.0 2.8A5 15 July–18 July 2005 1524 25.2 2.8

W2 18 July 2005 3.6 A6 18 July–20 July 2005 1009 25.5 3.5W3 20 July 2005 3.6 A7 20 July–22 July 2005 1026 24.1 3.4

A8 22 July–25 July 2005 1581 23.3 3.2W4 25 July 2005 4.2 A9 25 July– 27 July 2005 1020 25.2 3.3W5 27 July 2005 2.9 A10 27 July–29 July 2005 976 26.7 2.0W6 1 August 2005 4.7 A11 29 July–1 August 2005 1547 27.5 3.0

A12 1 August–3 August 2005 1055 25.8 3.7W7 3 August 2005 4.0 A13 3 August–5 August 2005 1189 22.1 2.9

To evaluate the fluxes, the air and SML samples have been paired as follows: W1–average A3–A4, W2–average A5–A6, W3–average A6–A7, W4–averageA8–A9, W5–A10, W6–average A11–A12, W7–average A12–A13, and named S1, S2, S3, S4, S5, S6, S7, respectively.

1636 L. Manodori et al. / Marine Pollution Bulletin 54 (2007) 1634–1644

an n-pentane and dichloromethane mixture (2:1) asreported elsewhere (Manodori et al., 2006a; Moret et al.,2005). For quantitative purposes, all water and air sampleswere spiked before extraction with 600 pg and 10 ng of a13C-labeled standard mixture (PCB 28, 52, 101, 138, 153and 180; Cambridge Isotope Laboratories) respectively.The extracts were dehydrated with anhydrous sodium sul-phate, their volume was reduced to 5 mL under nitrogenflow at 23 �C (Turbovap II Zimark) and cleaned by anautomated multi-column system (Power-Prep, FluidManagement System Inc.). Samples were loaded on packedneutral silica column (flow: 2 mL min�1) previously condi-tioned with 50 mL of n-hexane (flow: 10 mL min�1). PCBswere eluted with 30 mL of n-hexane (flow: 10 mL min�1)followed by 30 ml of an n-hexane and dichloromethanemixture (1:1, flow: 5 mL min�1). The final volume wasreduced to 500 lL, added with 100 lL of iso-octane andreduced again to 100 lL.

One MAT 95XP (Thermo Finnigan) high-resolutionmagnetic mass spectrometer, equipped with a Hewlett–Packard Model 5890 series II gas chromatograph, was usedto analyze all samples. Gas chromatographic separationwas performed on a fused silica capillary column (J&WScientific DB-5MS, 60 m · 0.250 mm · 0.25 lm) and datawere acquired in the electron impact (EI) mode (45 eV).The operating conditions were: injector temperature300 �C; transfer line temperature 300 �C; oven temperatureprogram 120 �C (1 min), 20 �C min�1 to 150 �C, 8 min at150 �C, 4 �C min�1 to 235 �C, 10 min at 235 �C,12�C min�1 to 290 �C, 34 min at 290 �C (post-run); carriergas (helium), 1.2 mL min�1; injection mode, splitless (splitvalve open after 1 min) with purge flow 50 mL min�1.Quantification of 118 congeners was performed by compar-ing the area of the chromatographic peak of the PCB withthose of the same 13C-labeled homolog; results were cor-rected by periodically evaluated individual responsefactors.

2.2. Quality assurance

2.2.1. Breakthrough from adsorbentThe risks of losses of semivolatile organic compounds

from the PUF when large volumes of air are collectedand temperature is high are known (Burdick and Bidleman,1981). To evaluate this potential artefact, the air samplingswere performed using an entire PUF followed by anotherhalf one, which were analyzed separately. Results showthat breakthrough ranged on average from 18% to 49%for the homologs with one to three chlorines and it wasnegligible for all the others (615%). These results agreewith those reported by other authors (Hermanson andHites, 1989) and to correct this preferential underestima-tion the concentrations of PCBs were calculated by addingthe quantities found in both PUFs.

2.2.2. Repeatability and recovery

Pressurized solvent extraction was applied to these sam-ples instead of the more time- and solvent-consumingSoxhlet extraction, as well as the automatic clean-up, sothe reliability of the method was strictly tested by analyzingthree pre-cleaned PUFs spiked with known amounts ofPCB mixture (2 ng). The recovery of PCBs for the entireprocedure was on average 75% ranging from 62% (PCB77) to 93% (PCB 172) and the repeatability, expressed asrelative standard deviation, varied from 0.2% (PCB16 + 32) to 25% (PCB 194); for R118PCB the average recov-ery was 77% and the relative standard deviation was 4%.These results are similar to those obtained with the previ-ous traditional method (Gambaro et al., 2004) and thosereported by Hornbuckle et al. (1993), Lee and Jones(1999) and Kim and Masunaga (2005).

The procedure for water samples was previously tested(Manodori et al., 2006a; Moret et al., 2005) and here mod-ified only in the clean-up step. Its control was performed byadding a known amount of surrogate standard mixture

L. Manodori et al. / Marine Pollution Bulletin 54 (2007) 1634–1644 1637

(PCBs 30, 65, 96, 166, 189, 199) to every sample beforeextraction. The relative errors are about ±30% while therelative standard deviation ranges from 13% to 36%.

2.2.3. Blanks and limit of detection

Intense control of external contamination was appliedby analyzing many laboratory and field blanks, which con-stituted of pre-cleaned PUF plugs left in the sampler for afew minutes without air flowing and then analyzed asusual. In total, the blanks represented about the 30% ofthe sample processed. For the air samples the amount ofR118PCB in the laboratory blanks is 552 ± 172 pg, rangingfrom 0.2 ± 0.3 pg (PCB 176) to 51 ± 22 pg (PCB 31 + 28)for individual congeners, and 1044 ± 129 pg in the fieldblanks, varying from 0.2 ± 0.1 pg (PCB 197) to80 ± 10 pg (PCB 31 + 28) for single ones. These values,which testify the occurrence of contamination derivingfrom transport and storage, are considerably lower thanthose reported for the traditional method (Gambaroet al., 2004) probably due to the different extraction systemused to pre-clean the PUFs and the clean-up automaticallyperformed. Also the PCB amounts detected in the waterprocedural blanks, as defined by Moret et al. (2005), arelower than the values generally reported in the literature(Bamford et al., 2002a; Hornbuckle et al., 1995; Manodoriet al., 2006a); in fact the single congener quantities foundrange from 0.3 pg (PCB 131) to 105 pg (PCB 31 + 28)and R118PCB is 1535 pg.

Table 2Summary of PCB concentrations in SSW, SML and air samples

PCB SSW dissolved concentration (pg L�1) SML dissolv

Min Max Average Standarddeviation

Min Max

5 + 8 < 21.2 6.6 8.8 < 30.918 < 19.5 10.0 7.5 < 25.531 + 28 < 29.2 19.5 9.8 < 34.852 11.8 19.7 15.8 3.1 6.4 14.874 2.0 5.3 3.6 1.0 1.0 5.184 + 101 + 90 15.1 29.6 21.5 5.2 13.4 26.187 + 115 < 5.5 3.0 2.1 3.3 4.6110 9.1 17.7 13.2 3.0 7.5 16.6149 8.1 14.6 11.2 2.2 7.3 16.1118 6.9 14.2 9.6 2.5 5.6 14.3153 11.2 18.7 14.1 2.9 10.2 20.0164 + 138 9.3 18.4 14.1 3.1 9.7 20.0128 + 167 1.1 2.8 1.9 0.5 < 3.1180 5.2 10.1 6.8 1.8 5.7 11.3170 + 190 2.1 3.9 2.7 0.6 2.6 4.9201 < 1.4 0.3 0.5 < 2.1Mono-CB < < < < <Di-CB 3.3 43.1 18.0 15.3 3.2 56.7Tri-CB 38.9 94.2 61.5 22.5 2.0 110.6Tetra-CB 41.5 97.9 71.8 17.4 16.7 96.4Penta-CB 53.2 113.0 79.6 21.4 46.2 98.4Hexa-CB 41.7 75.4 58.0 12.2 36.6 83.4Epta-CB 17.9 34.8 24.8 7.0 23.7 36.6Octa-CB 2.2 3.4 2.8 0.8 2.1 4.8Nona- and deca-

CB0.2 0.9 0.4 0.3 0.3 1.2

R118PCB 228.0 390.5 312.2 65.8 158.6 388.4

The limit of detection was calculated as the average fieldblank plus three times its standard deviation. For air sam-ples it ranges from 0.5 pg (PCB 197) to 110 pg (PCB31 + 28) and it is 1762 pg for the R118PCB. For water sam-ples it varies from 0.9 pg (PCB 34) to 248 pg (PCB 31 + 28)and it is 2790 pg for the R118PCB. The comparison betweenthe PCB amounts found in the samples with the LODshows that quantities detected in every PUF are always sig-nificant while those detected in SML and SSW samples aresometimes lower, especially for mono-, di-, octa- and nona-CB. On average, blank accounts for about 29% in waterand 2.5% in air samples; because of these reasons, all sam-ples have been corrected for the corresponding averagefield blank.

3. Results and discussion

3.1. Water samples

The minimum, maximum, average and standard devia-tion of the operationally defined dissolved (truly dissolvedplus colloidal fraction) concentrations of R118PCB andmore representative congeners found in SSW and SMLsamples are reported in Table 2.

The average concentration of R118PCB found in theSSW samples is 312 pg L�1, with values ranging from228 pg L�1 to 391 pg L�1, and main contributing congen-ers (greater than 5%) are 31 + 28, 52, 84 + 90 + 101. A

ed concentration (pg L�1) Gaseous concentration (pg m�3)

Average Standarddeviation

Min Max Average Standarddeviation

7.9 11.8 1.8 9.5 5.5 2.68.5 9.4 3.0 72.5 13.6 18.3

13.3 13.9 7.9 28.9 17.3 5.69.9 2.9 5.3 14.2 9.6 2.52.8 1.4 0.9 2.7 2.0 0.6

18.7 4.4 5.7 13.5 10.0 2.54.1 0.6 1.7 5.0 3.3 1.0

11.2 3.4 2.4 7.6 4.8 1.512.6 3.6 0.03 5.3 3.6 1.38.9 3.2 1.4 4.7 3.0 0.9

15.2 3.7 0.3 5.0 3.5 1.314.6 3.4 0.04 3.9 2.8 1.01.7 1.1 0.2 0.6 0.4 0.18.4 2.1 0.4 1.2 0.8 0.23.6 0.7 0.2 0.4 0.3 0.11.1 0.9 0.1 0.2 0.12 0.03

< 0.1 0.6 0.2 0.225.8 23.3 3.4 17.3 9.7 4.439.7 40.8 20.8 120.9 54.8 27.448.9 24.8 28.3 82.0 50.7 13.168.4 18.6 20.7 55.4 38.8 10.761.3 16.7 10.5 24.3 18.5 3.829.5 5.7 2.5 6.5 4.3 1.13.4 1.5 0.2 0.5 0.3 0.10.8 0.4 0.0 0.2 0.1 0.0

265.0 90.1 86.5 273.4 177.3 51.0

1638 L. Manodori et al. / Marine Pollution Bulletin 54 (2007) 1634–1644

lower concentration is found in the SML samples, whichrange from 159 pg L�1 to 388 pg L�1 (average 273 pg L�1);the most frequently present congeners are 84 + 90 + 101,153, 164+138. These values are typical for the Venicelagoon (Manodori et al., 2006a; Moret et al., 2005) andare similar to the Baltic Sea (Bruhn et al., 2003) but arelower than those reported for the Mediterranean (Garcıa-Flor et al., 2005a,b) and South China Sea (Wurl andObbard, 2006b; Wurl et al., 2006c) coastal waters.

Although in both phases penta-CB is the predominatinggroup, the homolog distribution is rather different betweenSSW and SML, as shown in Fig. 2, which highlights thepreferential accumulation of the most hydrophobic cong-eners into the microlayer phase, as found previously(Manodori et al., 2006a; Garcıa-Flor et al., 2005b; Wurland Obbard, 2005a). As a consequence, the enrichment fac-tor EF, defined as the ratio of SML versus SSW concentra-tions, rises with the degree of chlorination but in any case itmaintains values generally lower than 2. The R118PCBappears equally distributed in both matrices with anEF = 0.8 ± 0.2, in agreement with the results reported forthe dissolved phase by Wurl et al. (2006a) at Hong Kong,where the EF was 1.4–2.3, and by Garcıa-Flor et al.(2005b) at an oligotrophic bay (EF = 1) and a heavy pol-luted site (EF = 2.3) in the NW Mediterranean zone.

The dissolved PCB concentrations in SSW and SMLsamples are well correlated with each other (Pearson coef-ficient: 0.93 ± 0.04) and show a similar decreasing trend(a < 0.05: R2 SSW = 0.7; R2 SML = 0.52), which suggeststhat they derive from a common source of contamination,probably consisting of PCBs buried the sediment andremobilized in the suspended particulate matter (Manodoriet al., 2006a). In fact the lower water depth, the tidalexchange and the heavy shipping traffic in the Venice

Fig. 2. Relative abundances of PCB homologs in SSW an

lagoon lead to very intensive sediment resuspension pro-cesses, which may increase the PCB concentration in thewater column masking the SML enrichment. Sedimentremobilization is large problem of this area and it producessome other strong impacts, for example the changes in thehydrologic order and in the biological communities (Bian-chi et al., 2000; Facca et al., 2002).

3.2. Air samples

The minimum, maximum, average and standard devia-tion of the gaseous concentrations of R118PCB and themore representative congeners and found in air samplesare reported in Table 2.

The concentration of R118PCB in the air samples rangesfrom 87 pg m�3 to 273 pg m�3 and its average value is177 pg m�3. The most frequently present congeners(>5%) are PCB 18, 31 + 28, 52, 44 + 59 + 42, 84 + 90 +101 and tri-, tetra- and penta-CB predominate over theother homologs, respectively, constituting 30%, 29% and22% of the total concentration. This pattern is in agreementwith the results reported for other areas (Jaward et al.,2004; Park et al., 2002; Wurl and Obbard, 2005b). Forexample, Yeo et al. (2004) found that PCB homologconcentrations in the urban area of Seoul (Korea) were dis-tributed in the order of tri-CBs (28%) > tetra-CBs(25%) > penta-CBs (24%) > other homologs. They sup-posed that the heaviest homologs (>penta-CBs) easilydeposit on plants, soil and layers of water and evaporateless into the atmosphere due to their lower vapour pressurecompared with tri- and tetra-CBs, which mostly exist asgases in the atmosphere (Simcik et al., 1997; Yeo et al.,2003). This kind of pattern is compatible with the emissions

d SML samples and average enrichment factor (EF).

Fig. 3. Temporal trend of RPCB and homolog distribution in air samples.

L. Manodori et al. / Marine Pollution Bulletin 54 (2007) 1634–1644 1639

from secondary sources and consequently long-rangetransport, enhanced by the summery temperature.

The concentrations found in this study are higher thanthose reported for scarcely contaminated environmentssuch as the Swedish coast as reported by Sundqvist et al.(2004) but agree well with areas influenced by the sea(Lee and Jones, 1999; Wurl and Obbard, 2005b) and valuesreported for other Italian locations (Manodori et al.,2006b).

Two peaks with PCB concentration >250 pg m�3 weredetected, one of them characterized also by a great abun-dance of 3-CB, as shown in Fig. 3. Although in those dayswinds blew mainly from the S–SE direction, which was typ-ical for the marine source and generally characterized bypoor concentrations of PCBs (Manodori et al., 2006b),the study of the back-trajectories calculated by applyingthe HYSPLIT model (http://www.arl.noaa.gov/ready.html) reveals that in both cases the air comes from themainland of northern Italy. Another event of air comingfrom Northern Italy occurred and coincides with the A5sample collection. Its homolog distribution differs consid-erably from the others due to an anomalous prevalenceof tetra- and penta-CB and it might reveal the presenceof a peculiar source. The lowest concentration was mea-sured in sample A8, and it is probably due to the concur-rence of two factors: the wash-out of the atmosphereconsequent to the raining event (Simcik, 2004) thatoccurred in those days, and the air coming from the Adri-atic sea, as revealed by the back-trajectories study.

3.3. Fluxes at the interface

The importance of the SML in the air–sea gaseousexchange of PCB has been recently evaluated by Wurlet al. (2006a), by applying the modified two-layer model

proposed by Zhou and Mopper (1997), which considersthe POP enrichment at the water–air interface. They foundthat the relative difference between this flux and that calcu-lated with the traditional model (Liss and Slater, 1974) issubstantially high for hydrophobic compounds such asPCB, indicating that the SML plays an important role inthe air–sea gas exchange. In agreement with this approach,we evaluated the fluxes of every PCB congener by using thesame model as proposed (Zhou and Mopper, 1997), and wecompared them with those obtained by applying the con-ventional approach. The gaseous flux expression is:

F ¼ kolðCw � CaRT=HÞ; ð1Þ

where kol (m d�1) is the overall mass-transfer rate coeffi-cient, Cw and Ca are the SML (or SSW for the traditionalmodel) and air concentrations (pg m�3) of each congener,respectively, R is the gas constant (8.314 J mol�1 K�1), T

is the average air temperature over the sampling period(298 K) and H is the Henry’s law constant (Pa m3 mol�1).This last parameter is very important in the air–water fluxestimation, and we agree with some authors (Rowe et al.,2007; Wurl et al., 2006a; Totten et al., 2001) that the H val-ues measured by Bamford et al. (2002b) are the best dataset available at the state of the art; so we used them forour calculations.

The complete details about the flux calculations arereported elsewhere (Hornbuckle et al., 1994; Wurl et al.,2006a); briefly, kol has been derived from

1=kol ¼ 1=kw þ RT =Hka; ð2Þ

where ka is the rate coefficients across the stagnant air layerand kw is the rate coefficient across the stagnant waterlayer, which were calculated considering the average windspeed at a height of 10 m over each sampling period, as re-ported in Table 1. The diffusivity of PCB homologs has

1640 L. Manodori et al. / Marine Pollution Bulletin 54 (2007) 1634–1644

been calculated by applying the Fuller method (Reid et al.,1987), the Schmidt numbers considered were those re-ported by Hornbuckle et al. (1994) at 25 �C. The valuesof ka, kw and kol obtained for every sample are listed inTable 3; they are in good agreement with those reportedfor other summer sampling campaigns (Hornbuckleet al., 1994), with some discrepancies due to the differentwind and temperature conditions.

From Eq. (1), a positive value of F indicates the ten-dency of PCBs to move from water to air (volatilization)and, on the contrary, a negative result indicates the ten-dency to transfer from air to water. The uncertainty asso-ciated to our flux measurements has been analyzed by thepropagation of the errors considering the contributionderived from kol, H and the measured concentrations Cw

and Ca; it ranges from 50% to 64% and fall in the widerange typically found by other authors (Hornbuckleet al., 1994; Nelson et al., 1998; Wurl et al., 2006a). How-ever, neither significant possible systematic errors nor theimproper quantification of the concentration gradient dueto the inclusion of colloidal bound contaminants in the‘operationally defined’ dissolved phase measurements, theunderestimation of the wind effects on kol and the inaccu-rate values of H can be evaluated with this analysis (Bam-ford et al., 2002a). In any case, we think that these resultscan help to complete the mass balance of PCB in the Venicelagoon, by filling the gap at the air–water interface.

The flux of R118PCB, calculated as the sum of singlecongener fluxes, is very similar by considering both the tra-

Table 3Value of kw and ka for every homologs (m d�1) and average value on chlorin

Sample 1 2

Average wind speed 2.70 3.15Kw,PCB Di- 0.383 0.493

Tri- 0.374 0.482Tetra- 0.364 0.469Penta- 0.357 0.459Hexa- 0.348 0.448Hepta- 0.343 0.441Octa- 0.335 0.431Nona- 0.327 0.421

Ka,PCB Di- 267 296Tri- 261 289Tetra- 256 283Penta- 251 278Hexa- 247 273Hepta- 242 268Octa- 239 264Nona- 235 260

Kol,PCB Di- 0.333 0.420Tri- 0.334 0.423Tetra- 0.328 0.415Penta- 0.330 0.420Hexa- 0.327 0.417Hepta- 0.318 0.405Octa- 0.283 0.356Nona- 0.299 0.381

The wind speed (m s�1) considered for calculation is averaged over the sampl

ditional (called TM) and the modified two-layers (calledMM) models, in fact it is 111 ± 40 and 104 ± 44 ngm�2 d�1, respectively. In both cases, all congeners volatizefrom the water to the air, and the largest contributionderives from homologs with penta-chlorines, mainlybecause of their prevalence in water samples. Despite thatthe homologs relative abundances are different by applyingthe two models, in both cases about 80% of the total fluxderives from tri- to hexa-CB homologs, as previously foundby Bamford et al. (2002a) at Baltimore Harbour and in theNorthern Chesapeake Bay.

Generally PCB volatilization occurs in tropical watersand loading in the high latitudinal area (Iwata et al.,1993). For example, fluxes from the open MediterraneanSea range from �35 to 26 lg m�2 y�1 indicating that vola-tilization might be as important as absorption, while forthe industrial coastal areas they are 0.5–73 lg m�2 y�1

(Tolosa et al., 1997), in the same order of magnitude asthose calculated in this study. Similar rates have beenfound also for the Chesapeake Bay region (Nelson et al.,1998; Bamford et al., 2002a; Hornbuckle et al., 1995), theLake Superior (Hornbuckle et al., 1994) and the Swedishwest coasts (Sundqvist et al., 2004). The flux of the mostrepresentative congeners, which minimum, maximum,average and standard deviation are reported in Table 4,are comparable with those reported by Wurl et al.(2006a) for the Singapore coastal marine environment byapplying the TM and referring to July 2004, but they aresubstantially lower than those calculated with the MM.

ation degree basis of kol (m d�1) for every pair of water and air samples

3 4 5 6 7

3.45 3.25 2.00 3.35 3.300.573 0.519 0.234 0.546 0.5320.559 0.507 0.229 0.533 0.5200.545 0.494 0.223 0.519 0.5060.533 0.483 0.218 0.508 0.4960.521 0.472 0.213 0.496 0.4840.512 0.464 0.209 0.488 0.4760.501 0.454 0.205 0.477 0.4660.489 0.443 0.200 0.466 0.455

315 302 223 309 306308 296 218 302 299302 289 213 295 292296 284 209 290 287291 279 205 285 282286 274 202 280 277281 270 199 275 273277 266 196 271 269

0.481 0.440 0.211 0.460 0.4500.486 0.444 0.210 0.465 0.4540.477 0.436 0.206 0.456 0.4460.483 0.441 0.206 0.462 0.4510.481 0.438 0.203 0.459 0.4480.467 0.426 0.198 0.446 0.4360.408 0.373 0.180 0.390 0.3820.438 0.400 0.187 0.419 0.409

ing period.

Table 4Calculated fluxes for the most representative congeners (ng m�2 d�1)

PCB congeners (number of chlorine) Modified two layers model Traditional model

Min Max Average Standard deviation Min Max Average Standard deviation

8 + 5 (2) nd 15.0 7.5 6.6 nd 10.3 6.3 4.118 (3) nd 12.1 5.7 4.3 nd 9.3 5.7 2.831 + 28 (3) nd 17.0 8.9 5.6 nd 14.2 8.3 6.152 (4) 1.7 6.1 3.9 1.4 2.4 8.9 6.3 2.174 (4) 0.2 1.7 1.2 0.6 0.4 2.3 1.5 0.684 + 90 + 101 (5) 3.2 10.2 7.6 2.4 3.1 12.6 8.8 3.087 + 115 (5) nd 1.8 1.7 0.2 nd 2.3 1.8 0.4110 (5) 1.7 5.9 4.5 1.5 2.1 7.4 5.3 1.8118 (5) 1.3 5.2 3.5 1.3 1.4 5.9 3.8 1.4149 (6) 2.6 7.7 5.0 1.8 2.5 5.8 4.4 1.1153 (6) 2.2 8.3 6.1 2.1 2.6 7.8 5.5 1.6164 + 138 (6) 2.5 7.6 5.7 1.7 2.6 7.6 5.5 1.7128 + 167 (6) nd 1.1 0.8 0.3 nd 1.1 0.7 0.4180 (7) 1.3 4.3 3.2 1.0 1.0 4.0 2.6 0.9170 + 190 (7) 0.6 2.0 1.3 0.4 0.4 1.1 0.9 0.3201 (8) nd 0.9 0.6 0.2 nd 0.6 0.4 0.3

L. Manodori et al. / Marine Pollution Bulletin 54 (2007) 1634–1644 1641

Although there is this apparent discrepancy, our results arein agreement with the Wurl and co-authors’ observation, infact the influence of microlayer (evaluated as the differencebetween the fluxes calculated with the MM and TM)appears for the most hydrophobic congeners (positive dif-ference) and it is not relevant for the lightest PCB (negativedifference), as depicted in Fig. 4. Although to a minorextent with respect to the Singapore coasts, the gaseous fluxat the air–water interface of the Venice lagoon is influencedby the presence of the superficial microlayer film, so webelieve that the conceptual approach proposed by the Zhouand Mopper’s model is suitable for our aims.

The range of variation of MM total flux found is quitelarge, 58 ng m�2 d�1 being the minimum and 195 ng m�2

d�1 the maximum. These values correspond to the lowest

Fig. 4. Comparison between the homologs fluxes calculated with

(2.0 m s�1) and highest (3.5 m s�1) wind speeds, respec-tively, as shown in Fig. 5, and they agree with the stronginfluence of wind speed in determining the flux (Iwataet al., 1993). However, when the difference between waterand air concentrations is small, the corresponding fluxdrops, even when a fairly high wind speed is present, asoccurs for sample S6 (Fig. 4).

To assess the total balance of PCBs in the Venicelagoon, the MM gaseous fluxes at the air–water interfacehave been compared (as orders of magnitude) with thosederiving from atmospheric deposition and basin drainage.Rossini et al. (2001) reported that the bulk depositionranges from 378 to 1026 ng m�2 y�1, i.e., they are about1.04–2.81 ng m�2 d�1 while Collavini et al. (2005) esti-mated that the annual load from the twelve major tributar-

the traditional (TM) and modified two-layers (MM) models.

Fig. 5. Temporal trend of gaseous flux of PCB, compared with the air and water concentrations and wind speed.

1642 L. Manodori et al. / Marine Pollution Bulletin 54 (2007) 1634–1644

ies is 3359 g (expressed as Aroclor 1254 + 1260, 48) whichcorresponds to about 21 ng m�2 d�1 by considering thewater surface open to the tidal exchange (430 km2). Fromthis, it appears that in summer months the output due tovolatilization from the water surface exceeds the directinput from the atmosphere and the rivers, and the Venicelagoon water acts as secondary source of contaminationfor PCBs.

4. Conclusions

This study has provided the first data about the gaseousexchange of PCB at the air–water interface of the Venicelagoon. As recently proposed by the literature, a modelthat includes the SML has been used to calculate the fluxes.In fact, although the total dissolved concentration of PCBin the SML is not significantly enriched with respect to theunderlying water, mainly due to the resuspension processeswhich affect the lagoon shallow waters, a preferential accu-mulation of the most hydrophobic homologs into themicrolayer phase has been observed. The influence of themicrolayer presence, evaluated by comparing the resultsfrom the traditional and modified two-layers models, isnot displayed with respect to the total flux increase, butwith regard to the heaviest homologs flux magnitude, asfound by other authors (Wurl et al., 2006a) who consideredchemicals with different hydrophobicity.

The fluxes calculated over the sampling period of July2005 are in the same order of magnitude as those reportedfor the industrial coastal areas of the Mediterranean seaand they reveal that PCB volatilize from the SML intothe air. Therefore, in the summer months the Venicelagoon acts as secondary source of these pollutants, to a

larger extent than the input it receives from atmosphericdeposition and the riverine load.

Acknowledgements

This work was supported by CORILA under the project‘Pollutant flows in the lagoon carried by aerosols andatmospheric fall-out’ and by the National Research Coun-cil of Italy (CNR). The authors are grateful to Dr. ItaloOngaro and Dr. Angela M. Stortini for their support dur-ing sampling activities and the NOAA Air Resources Lab-oratory (ARL) for the provision of the HYSPLITtransport and dispersion model (http://www.arl.noaa.gov/ready.html) used in this publication.

References

Baker, J.E., Eisenreich, S.J., 1990. Concentrations and fluxes of polycyclicaromatic hydrocarbons and polychlorinated biphenyls across the air–water interface of Lake Superior. Environmental Science and Tech-nology 24, 342–352.

Bamford, H.A., Ko, F.C., Baker, J.E., 2002a. Seasonal and annual air–water exchange of polychlorinated biphenyls across Baltimore Har-bour and the Northern Chesapeake Bay. Environmental Science andTechnology 36, 4245–4252.

Bamford, H.A., Poster, D.L., Huie, R.E., Baker, J., 2002b. Using extra-thermodynamic relationships to model the temperature dependence ofHenry’s law constants of 209 PCB congeners. Environmental Scienceand Technology 36, 4395–4402.

Bianchi, F., Acri, F., Alberighi, L., Bastianini, M., Boldrin, A., Cavalloni,B., Cioce, F., Comaschi, A., Rabitti, S., Socal, G., Turchetto, M.M.,2000. Biological variability in the Venice lagoon. In: Lasserre, P.,Marzolla, A. (Eds.), The Venice Lagoon Ecosystem. Inputs andInteractions between Land and Sea. Man and the Biosphere Series,vol. 25. UNESCO, pp. 97–125.

L. Manodori et al. / Marine Pollution Bulletin 54 (2007) 1634–1644 1643

Breivik, K., Sweetman, A., Pacyna, J.M., Jones, K.C., 2002. Towards aglobal historical emission inventory for selected PCB congeners – amass balance approach. Science of the Total Environment 290, 181–198.

Bruhn, R., Lakaschus, S., McLachlan, M.S., 2003. Air/sea exchange ofPCBs in the southern Baltic Sea. Atmospheric Environment 37, 3445–3454.

Burdick, N.F., Bidleman, T.F., 1981. Frontal movement of hexachloro-benzene and polychlorinated biphenyl through polyurethane foam.Environmental Science and Technology 53, 1926–1929.

Collavini, F., Bettiol, C., Zaggia, L., Zonta, R., 2005. Pollutant loads fromthe drainage basin to the Venice lagoon (Italy). Environment Inter-national 31, 939–947.

Facca, C., Sfriso, A., Socal, G., 2002. Changes in abundance andcomposition of phytoplankton and microphytobenthos due toincreased sediment fluxes in the Venice lagoon, Italy. Estuarine,Coastal and Shelf Science 54, 773–792.

Gambaro, A., Manodori, L., Moret, I., Capodaglio, G., Cescon, P., 2004.Determination of polychlorobiphenyls and polycyclic aromatic hydro-carbons in the atmospheric aerosol of the Venice lagoon. Analyticaland Bioanalytical Chemistry 378, 1806–1814.

Garcıa-Flor, N., Guitart, C., Bodineau, L., Dachs, J., Bayona, J.M.,Albaiges, J., 2005a. Comparison of sampling devices for the determi-nation of polychlorinated biphenyls in the sea surface microlayer.Marine Environmental Research 59, 255–275.

Garcıa-Flor, N., Guitart, C., Bodineau, L., Dachs, J., Bayona, J.M.,Albaiges, J., 2005b. Enrichment of organochlorine contaminants in thesea surface microlayer: an organic carbon-driven process. MarineChemistry 96, 331–345.

Hermanson, M.H., Hites, R.A., 1989. Long-term measurements andatmospheric polychlorinated biphenyls in the vicinity of superfunddumps. Environmental Science and Technology 23, 1253–1258.

Hornbuckle, K.C., Achman, D.r., Eisenreich, S.J., 1993. Over-water andover land polychlorinated biphenyls in Green Bay, Lake Michigan.Environmental Science and Technology 27, 87–98.

Hornbuckle, K.C., Jeremiason, J.D., Sweet, C.W., Eisenreich, S.J., 1994.Seasonal variation in air–water exchange of polychlorinated biphenylsin Lake Superior. Environmental Science and Technology 28, 1491–1501.

Hornbuckle, K.C., Sweet, C.W., Pearson, R.F., Swackhamer, D.L.,Eisenreich, S.J., 1995. Assessing annual water–air fluxes of polychlo-rinated biphenyls in Lake Michigan. Environmental Science andTechnology 29, 869–877.

Iwata, H., Tanabe, S., Sakal, N., Tatsukawa, R., 1993. Distribution ofpersistent organochlorines in the oceanic air and surface seawater andthe role of ocean on their global transport and fate. EnvironmentalScience and Technology 27, 1080–1098.

Jaward, F.M., Barber, J.L., Booij, K., Dachs, J., Johmann, R., Jones,K.C., 2004. Evidence of dynamic air–water coupling and cycling ofpersistent organic pollutants over the open Atlantic ocean. Environ-mental Science and Technology 38, 2617–2625.

Jurado, E., Lohmann, R., Meijer, S., Jones, K.C., Dachs, J., 2004.Latitudinal and seasonal capacity of the surface oceans as a reservoirof polychlorinated biphenyls. Environmental Pollution 128, 149–162.

Kim, K.-S., Masunaga, S., 2005. Behaviour and source characteristic ofPCBs in urban ambient air of Yokohama, Japan. EnvironmentalPollution 138, 290–298.

Lee, R.G.M., Jones, K.C., 1999. The influence of meteorology and airmass on daily atmospheric PCB and PAH concentrations at a UKlocation. Environmental Science and Technology 33, 705–712.

Lerche, D., van de Plassche, E., Schwegler, A., 2002. Selecting chemicalsubstances for the UN-ECE POP protocol. Chemosphere 47, 617–630.

Liss, P.S., Slater, P.G., 1974. Flux of gases across the air–sea interface.Nature 247, 181–184.

Mackay, D., Paterson, S., 1986. Model describing the rates of transferprocesses of organic chemicals between atmosphere and water.Environmental Science and Technology 20, 810–816.

Manodori, L., Gambaro, A., Piazza, R., Ferrari, S., Stortini, A.M.,Moret, I., Capodaglio, G., 2006a. PCBs and PAHs in microlayer andsub-surface water samples of the Venice lagoon (Italy). Science of theTotal Environment 52, 184–192.

Manodori, L., Gambaro, A., Moret, I., Capodaglio, G., Cairns, W.R.L.,Cescon, P., 2006b. Seasonal evolution of gas-phase PCB concentra-tions in the Venice lagoon area. Chemosphere 62, 449–458.

Moret, I., Gambaro, A., Piazza, R., Ferrari, S., Manodori, L., 2005.Determination of polychlorobiphenyl congeners (PCBs) in thesurface water of the Venice lagoon. Marine Pollution Bulletin 50,167–174.

Nelson, E.D., McConnell, L.L., Baker, J.E., 1998. Diffusive exchange ofgaseous polycyclic aromatic hydrocarbons and polychlorinated biphe-nyls across the air–water interface of the Chesapeake Bay. Environ-mental Science and Technology 32, 912–919.

Park, J.-S., Wade, T.L., Sweet, S.T., 2002. Atmospheric deposition ofPAHs, PCBs and organochlorine pesticides to Corpus Christi Bay,Texas. Atmospheric Environment 36, 1707–1720.

Reid, R.C., Prausnitz, J.M., Poling, B.E., 1987. The properties of gasesand liquids, fourth ed. McGraw-Hill International Editions, NewYork, p. 587.

Rossini, P., De Lazzari, A., Guerzoni, S., Molinaroli, E., Rampazzo, G.,Zancanaro, A., 2001. Atmospheric input of organic pollutant to theVenice lagoon. Annali di Chimica – Rome 91, 491–501.

Rowe, A.A., Totten, L.A., Xie, M., Fikslin, T.J., Eisenreich, S.J., 2007.Air–water exchange of polychlorinated biphenyls in the Delawareriver. Environmental Science and Technology 41, 1152–1158.

Simcik, M.F., 2004. The importance of surface adsorption on the washoutof semivolatile organic compounds by rain. Atmospheric Environment38, 491–501.

Simcik, M.F., Zhang, H., Eisenreich, S.J., Franz, T.P., 1997. Urbancontamination of the Chicago/coastal Lake Michigan atmosphere byPCBs and PAHs during AEOLOS. Environmental Science andTechnology 31, 2141–2147.

Sundqvist, K.L., Wingfors, H., Brorstom -Lunden, E., Wiberg, K., 2004.Air–sea gas exchange of HCHs and PCBs and enantiomers of a-HCHin the Kattegat Sea region. Environmental Pollution 128, 73–83.

Tolosa, I., Readman, J.W., Fowler, S., Villeneuve, J.P., Dachs, J., Bayona,J.M., Albaiges, J., 1997. PCBs in the western Mediterranean. Tempo-ral trends and mass balance assessment. Deep-Sea Research Part II 44,907–928.

Totten, L.A., Brunciak, P.A., Gigliotti, C.L., Dachs, J., Glenn IV, T.N.,Nelson, E.D., Eisenreich, S.J., 2001. Dynamic air–water exchange ofpolychlorinated biphenyls in the New York–New Jersey HarborEstuary. Environmental Science and Technology 35, 3834–3840.

Wania, F., Daly, G.L., 2002. Estimating the contribution of degradationin air and deposition to the deep sea to the global loss of PCBs.Atmospheric Environment 36, 5581–5593.

Wania, F., Mackay, D., 1996. Tracking the distribution of persistentorganic pollutants. Environmental Science and Technology 30, 390A–396A.

Wania, F., Axelman, J., Broman, D., 1998. A review of processes involvedin the exchange of persistent organic pollutants across the air–seainterface. Environmental Pollution 102, 3–23.

Wurl, O., Obbard, J.P., 2005a. Chlorinated pesticides and PCBs in the sea-surface microlayer and seawater samples of Singapore. MarinePollution Bulletin 50, 1233–1243.

Wurl, O., Obbard, J.P., 2005b. Organochlorine compounds in the marineatmosphere of Singapore. Atmospheric Environment 39, 7207–7216.

Wurl, O., Obbard, J.P., 2006b. Distribution of Organochlorine com-pounds in the sea-surface microlayer, water column and sediment ofSingapore’s coastal environment. Chemosphere 62, 1105–1115.

Wurl, O., Karuppiah, S., Obbard, J.P., 2006a. The role of the sea-surfacemicrolayer in the air–sea gas exchange of organochlorine compounds.Science of the Total Environment 369, 333–343.

Wurl, O., Obbard, J.P., Lam, P.K.S., 2006c. Distribution of organochl-orines in the dissolved and suspended phase of the sea-surface

1644 L. Manodori et al. / Marine Pollution Bulletin 54 (2007) 1634–1644

microlayer and seawater in Hong Kong, China. Marine PollutionBulletin 52, 768–777.

Yeo, H.-G., Choi, M., Chun, M.-Y., Kim, T.-W., Cho, K.-Chul, Sunwoo,Y., 2003. Gas/particle concentrations and partitioning of PCBs in theatmosphere of Korea. Atmospheric Environment 37, 3561–3570.

Yeo, H.-G., Choi, M., Chun, M.-Y., Kim, T.-W., Cho, K.-Chul, Sunwoo,Y., 2004. Concentration characteristics of atmospheric PCBs for urban

and rural area, Korea. Science of the Total Environment 324,261–270.

Zhou, X., Mopper, K., 1997. Photochemical production of low-molecular-weight carbonyl compounds in seawater and surface microlayer andtheir air–sea exchange. Marine Chemistry 56, 201–213.