Article pubs.acs.org/est
PCBs and OCPs on a East-to-West Transect: The Importance of Major Currents and Net Volatilization for PCBs in the Atlantic Ocean Rainer Lohmann,*,†,‡ Jana Klanova,§ Petr Kukucka,§ Shifra Yonis,† and Kevyn Bollinger† †
Graduate School of Oceanography, University of Rhode Island, South Ferry Road, Narragansett, 02882 Rhode Island, United States Geowissenschaften, Universität Tübingen, Hölderlinstrasse 12, 72074 Tübingen, Germany § Research Centre for Toxic Compounds in the Environment (RECETOX), Faculty of Science, Masaryk University, Kamenice 3/126, 625 00 Brno, Czech Republic ‡
S Supporting Information *
ABSTRACT: Air−water exchange gradients of selected polychlorinated biphenyl (PCB) congeners across a large section of the tropical Atlantic suggested net volatilization of PCBs to the atmosphere. Only for the higher chlorinated PCB 153 and hexachlorobenzene (HCB) were gradients near equilibrium detected. The use of passive samplers also enabled the detection of dichlorodiphenyltrichloroethane (DDT) and its transformation products across the tropical Atlantic, indicating net deposition. There were clear differences between the southern and northern hemisphere apparent in terms of atmospheric concentrations: Once the ship moved from the southern into the northern hemisphere air, concentrations of HCB and other organochlorine pesticides increased several-fold. For large swaths of the tropical Atlantic Ocean, neither PCB nor organochlorine pesticide dissolved concentrations varied much longitudinally, probably due to efficient mixing by ocean currents. In selected samples, dissolved concentrations reflected the influence of river plumes and major ocean currents far away from the continents. Dissolved concentrations of PCBs 28, 52, 101, 118, and HCB increased in the Amazon plume and the Gulf Stream. While the Amazon plume flushed only a few kg of PCBs and HCB, the Gulf Stream is potentially delivering tons of PCBs into the North Atlantic annually.
■
INTRODUCTION The fate of persistent organic pollutants (POPs) in the open Oceans is under debate. In 1989/1990, a first global study covering the world’s oceans reported polychlorinated biphenyls (PCBs) and other POPs mostly being taken up via air-to-water exchange.1 This was backed up by theoretical work that predicted air−water exchange (via net deposition) to be the main delivery pathway for PCBs into the Atlantic Ocean.2 The initial transfer of PCBs into the mixed surface ocean layer would lead to removal fluxes to deep oceans and sediments, probably by coupling to the “biological” pump.3 Shelf sediments were identified as major repositories of PCBs, but the exact pathway of how the PCBs reached the sediments was not investigated.4 Removal of PCBs by deep water formation seems to be of regional importance but represents only a small fraction of total PCB losses from the oceans.5 Most previously published ship-based transects that reported on POPs and their cycling were performed on north−south transects on the eastern side of the Atlantic Ocean.6−11 These studies provided valuable insights into regional emissions, e.g., of PCBs8 and polycyclic aromatic hydrocarbons (PAHs)9 off Africa, and the decline of polychlorinated dibenzo-p-dioxins and furans (PCDD/Fs) 10 and hexachlorocyclohexanes (HCHs)11 in the remote Atlantic. However, the sampling © 2012 American Chemical Society
regime was invariably affected by continental emissions off Europe and Africa, making extrapolations across the entire Atlantic Ocean difficult and fraught with uncertainties. In key modeling studies by Jurado et al., the authors had to assume that there were no significant east−west gradients of POPs such as PCBs and PCDD/Fs to calculate basin-wide deposition fluxes.2,12 Of major importance and interest in studying air−water exchange of POPs across the oceans is the question whether they serve as secondary sources or continue to take up POPs. Recent work in the Pacific Ocean suggested widespread net volatilization for PCBs in the oligotrophic Pacific,13 while results from the Atlantic Ocean were mixed,14 and PCBs in the Arctic Ocean were reported to undergo net deposition.15 However, oceans are not homogenously well-mixed water bodies but display strong gradients in temperature, productivSpecial Issue: Marine Boundary Layer: Ocean Atmospheric Interactions Received: Revised: Accepted: Published: 10471
September 30, 2011 January 24, 2012 February 2, 2012 February 2, 2012 dx.doi.org/10.1021/es203459e | Environ. Sci. Technol. 2012, 46, 10471−10479
Environmental Science & Technology
Article
collected in the ship’s laboratory from the ship’s seawater pipe using a sampling train, equipped with a precombusted glass fiber filter (GFF, Whatman GF/F, 0.7 μm retention) and 3 polyurethane foam (PUF) plugs in series. For samples 1−33 (Namibia−Barbados) 600−1100 L of water was collected, and 200−500 L of water was collected for samples 34−57 (Barbados−Rhode Island, U.S.) (see Table S9, Supporting Information). High Volume Air Samples. While water sampling proceeded continuously, unfavorable winds restricted air sampling at times (see page S3, Supporting Information). A total of 47 air samples were collected using a high-volume air sampler, equipped with a precombusted GFF (Whatman GF/A, 1.6 μm retention) and 2 pre-extracted PUF plugs. The active air sampler was setup at the front of the monkey’s level (above the bridge), facing the wind. For samples 1−23 between Namibia and Barbados, volumes of ∼400−570 m3 (∼12 h each) were collected. For samples between Barbados and Rhode Island, volumes of ∼230−350 m3 (∼8 h each) were collected (see Table S11, Supporting Information). PE Sheet Samplers. PEs were prepared as detailed on page S4. PE samplers were deployed in 4 different modes: 12 shorttime deployments (48 h each) in air and flow-through cell for water (see Figure S5, Supporting Information). The air PE samplers were exposed in inverted stainless steel bowls (“UFO disk”) on the ship’s main (chimney) mast located 17.5 m above sea level. The water PE was fixed in a steel pipe connected to the flowing seawater inside the ship’s special purpose laboratory, at a nominal flow rate of 10 L min−1. PE samplers were exposed continuously, even when hi-volume air sampling was paused due to adverse winds. Another PE sampler was towed on a line via the A-frame, ca. 100 m behind the ship, similar to water deployments suggested recently.18 Seven such towed PEs were deployed, for 40−70 h each. Lastly, several PE samplers were deployed inside the ship at several locations for 12−14 days each (see Tables S8−S10, Supporting Information). Sample Analysis. High Volume Samples (PUFs, GFFs). Standards for indicator PCBs and OCPs were purchased from LGC Standards (UK). PUFs and GFFs were extracted separately using automated warm Soxhlet extraction (40 min warm Soxhlet followed by 20 min of solvent rinsing) with dichloromethane (DCM) in a B-811 extraction unit (Büchi, Switzerland). The concentrated extracts were split into 2 portions; 1/4 of the extract was used for PAHs analysis, and the remaining 3/4 were used for PBDEs, indicator PCBs, and OCPs analysis. The extracts of water PUF and GFF samples were first dried using Na2SO4. Due to space constraints, results for PAHs and PBDEs will be reported elsewhere. Analysis. The second portion of the extract was cleaned-up on a H2SO4 modified (30% w/w) silica column, and the analytes were eluted with 40 mL of DCM/n-hexane mixture (1:1). The elute was concentrated under nitrogen in a TurboVap II concentrator unit and transferred into a smallvolume vial insert. The syringe standards (13C BDEs 77 and 138 and native PCB 121) were added to all samples; the final volume was 50 μL. PCBs and OCPs were analyzed via GC-MS (see page S3, Supporting Information). PE Sheet Samplers. Blanks and exposed sheets of PE were rinsed with Milli-Q water (Millipore, Billerica, MA), dried with a disposable tissue, and soaked for 24 h in 200 mL of n-hexane followed by 24 h in 200 mL of DCM. The two solvents were then pooled and concentrated under nitrogen in a TurboVap II
ity, and influx of terrestrial materials, including POPs. Numerous currents move water around and induce mixing both horizontally and, less efficiently, vertically. In light of these properties and features, it would be surprising if the same ocean did not have (adjacent) regions that would, at the same time, display net volatilization next to net uptake of POPs. A research cruise of the R/V Endeavor in July−August 2009 from Namibia via Barbados back to her home port in Narragansett (RI), USA, offered a unique opportunity to determine POPs’ gradients in air, water, and their air−water exchange (Figure 1). The cruise track covered both large east−
Figure 1. Cruise track of EN 464 from Namibia to Rhode Island (U.S.) in 2009 with major ocean currents and wind patterns encountered.
west and north−south gradients and was mostly far away from shore. In addition to the traditional POPs sampling on ships using active high-volume air and water sampling, we attempted also to deploy passive samplers as complementary approaches to measure truly dissolved concentrations. The potential for the use of passive samplers to determine POPs in the remote ocean had been earlier demonstrated by Booij et al. using semipermeable membrane devices (SPMDs).16 We were hoping that we could shorten deployment times of polyethylene (PE) passive samplers and thus get an improved spatial resolution of both concentrations and PE-based net air−water exchange gradients.17 In summary, we collected and analyzed samples collected across the tropical Atlantic to deduce whether (i) there were significant east−west and north−south gradients of selected PCB congeners and OCPs; (ii) the Atlantic Ocean was a net sink or secondary source of PCBs and OCPs; (iii) ocean current and river plumes would affect the presence and air− water gradients of selected POPs; and (iv) to compare active and passive sampling approaches for POPs.
■
METHODS
Sampling. High Volume Water Sampling. Sampling was conducted from July 2−29, 2009 between Namibia and Rhode Island (U.S). Active air and water sampling procedures were as described elsewhere.13 A total of 57 water samples were 10472
dx.doi.org/10.1021/es203459e | Environ. Sci. Technol. 2012, 46, 10471−10479
Environmental Science & Technology
Article
Figure 2. Dissolved concentrations of selected PCB congeners, PeCB, HCB (pg L−1), water temperature (°C), and salinity (psu) as a function of longitude. Major ocean currents are highlighted.
partitioning (Kaw) values, and aqueous solubility at saturation of the subcooled liquid (Ciwsat (L)), were taken from internally consistent data compilations whenever possible (see Tables S1 and S2, Supporting Information).21 For all compounds, average values were taken for enthalpies of PE−water (25 kJ mol−1) and PE−air exchange (80 kJ mol−1), close to calculated values of internal octanol−water and octanol−air exchange.21,22 The Setschnow constant was taken as 0.35, as reported elsewhere.23 Kaw was not corrected for the influence of dissolved organic carbon (DOC), as DOC concentrations in the open Ocean24 are in general too low to sorb more than a few % of the most hydrophobic PCBs, assuming general DOC affinities to POPs.25 Calculation of Concentrations from PE Samplers. Truly dissolved concentrations, Cdiss (pg L−1 H2O) (and gasphase, Cgas in pg m−3) were derived from PE-normalized concentrations, CPE (pg L−1 PE):
concentrator, and the extract was split into 2 portions and processed using the same procedure as the high volume samples. Quality Assurance, Quality Control, Data Treatment. Method performance was tested prior to sample preparation (See Tables S3−S5, Supporting Information). Recoveries were higher than 75% for all samples for PCBs+OCPs. Recovery factors were not applied to any of the data. Recovery of native analytes measured for a certified reference material (ASLAB soil standard, Czech Republic) varied from 88 to 100% for PCBs and from 75 to 98% for OCPs. Four field blanks each were analyzed for water and air PUFs and GFFs; 5 were analyzed for PEs. Field blanks were mounted as real samples, shortly turned on and recovered. The final concentrations were blank corrected. The method detection limit (MDL) was calculated as 3 standard deviations of blank concentrations (see Table S11, Supporting Information). Microsoft Excel’s Data Analysis was used for regression analysis. Only correlations significant at P = 0.05 are reported here. Concentrations < MDL were substituted with 1/2 MDL in cases where ≥70% of data were > MDL.19 Physicochemical Properties. PE−water (KPEw) and PE− air (KPEa) equilibrium partitioning constants were taken as recently reviewed, corrected for the corresponding sampling temperature and salinity.20 Octanol−water (Kow), air−water
CPE
Cdiss/gas =
KPEw(a)(1 − e 3
R st PEw(a)v )
−K
(1)
−1
where Rs (m day ) is the overall sampling rate, t is time (days), v is the PE volume (m3), and KPEw (or KPEa) is corrected for the average temperature and salinity of the deployment (see pages S6−S10, Supporting Information). For 10473
dx.doi.org/10.1021/es203459e | Environ. Sci. Technol. 2012, 46, 10471−10479
Environmental Science & Technology
Article
observed, longitude accounted for only 26% (significant at P = 0.05). Water and air temperature were also significant predictors for changes in [HCB]diss, with Twater explaining 30% of the observed variance and air temperature explaining 34% (r2 = 0.34, n = 30; P = 7.5e−4):
water exposures, Rs was derived on the basis of the ratios of the performance reference compound (PRC) concentrations prior to (CiPRC,t=0) and post deployment (CiPRC,t), while typical sampling rates were used for atmospheric deployments due to interferences with PRC-based results. Meteorological and Sea Surface Auxiliary Measurements. From the ship’s routine measurements, we averaged values of latitude, longitude, surface water temperature (Twater), salinity, fluorescence of the flow-through seawater, air temperature (Tair), relative humidity (RH), and relative and absolute wind speed and direction recorded every minute for each sampling period (see Tables S6−S10, Supporting Information). Back-trajectories were back-calculated for 5 days with 6 h steps at 300 m above sea level using HYSPLIT.26 Cruise Track. The cruise crossed several different major currents of the tropical Atlantic Ocean (Figure 1). In the southern hemisphere (SH), these were the Benguela, South Equatorial, and North Brazil Current. Discharges from the Amazon (and Orinoco River), the North Equatorial Current, and Gulf Stream affected samples in the Northern Hemisphere (NH). The cruise began with the atmospheric influence of the southeasterly trade winds which move air masses in a westerly direction toward the equator and into the intertropical convergence zone (ITCZ). The cruise then continued in the northeasterly trade winds which move air masses toward the west along the equator in the (NH). The last few samples were affected by the westerlies which move air masses eastwards across the Atlantic Ocean.
[HCB]diss = − 0.16 × Twater (°C) + 5.2 (r 2 = 0.30, n = 30; P = 1.6e−3)
(2)
A similar inverse correlation of [HCB]diss and Twater was observed in samples taken across the North Atlantic and Arctic Ocean, albeit with a steeper slope (−0.57).27 These correlations imply that, within the northern hemisphere, colder water (and air) carried higher concentrations of dissolved HCB. These correlations support the concept of enhanced partitioning of POPs into water at colder temperatures (“cold condensation”).28 For PCBs 101 and 118, concentrations increased 2−3-fold from the southern hemisphere toward the U.S. East coast. Correlations were significant with latitude (PCB 101: r2 = 0.19, n = 55; P = 1.0e−3; PCB 118: r2 = 0.30, n = 55; P = 1.7e−5) and longitude (PCB 101: r2 = 0.26, n = 55; P = 5.8e−5; PCB 118: r2 = 0.29, n = 55; P = 1.9e−5). Latitude and longitude played only a minor role in explaining changes in dissolved PCB 52 (r2 < 0.12, n = 55; P = 0.01), which was mostly
