Concentrations, Fluxes and Residence Time of PBDEs Across the Tropical Atlantic Ocean

Little is known about the fate of polybrominated diphenylethers (PBDEs) across the Oceans. 10 Air and water were sampled using both active and passive polyethylene samplers on an East- 11 West transect across the tropical Atlantic Ocean in 2009, and analyzed for PBDEs. Typical 12 particle-bound concentrations of PBDEs in the surface water were low, at < 1 pg L -1 . Truly 13 dissolved concentrations from passive samplers were ~ 0.5 pg L -1 for BDE 47 and around 0.1 14 pg L -1 for BDEs 28, 99 and 100 (results from active samples were compromised). In the 15 atmosphere, particle-bound BDE 209 dominated overall concentrations (median 1.2 pg m -3 ), 16 followed by BDE 99 (0.13 pg m -3 ). Gas-phase concentrations based on passive samplers were 17 1–8 pg m -3 for BDE 47, and ≤ 4 pg m -3 for BDE 99. Net air-water exchange strongly favoured gas-phase deposition of PBDEs into the water. Net gas-phase deposition fluxes ranged from 10s of pg m -2 day -1 for BDEs 28 and 85 to around 1 ng m -2 day -1 for BDE 20 47, 99 and 209. Settling fluxes of particle-bound PBDEs in atmosphere and surface water 21 were around 50 pg m -2 day -1 for BDE 47, and < 10 pg m -2 day -1 for the other


INTRODUCTION 25
Oceans have emerged as an important buffer and final sink for a wide range of persistent 26 organic pollutants (POPs). [1][2][3][4] There are four major pathways affecting POPs in the Oceans. 27 The most important pathway for persistent hydrophobic and lipophilic compounds is the 28 biological pump, which moves POPs to depth by partitioning into phytoplankton and settling 29 to depth 4,5 . POPs can also be moved to depth via the subduction of surface water 3 . Another 30 'physical' pathway is the movement to depth via eddy diffusion, which will be most important 31 for polar compounds which do not sorb strongly to organic carbon in the water column 6 . 32 Lastly, compounds can be prone to degradation, either biodegradation 7 or by direct and 33 indirect photolysis in the surface water 8,9 . Potential sources of POPs to the Atlantic combine 34 atmospheric deposition and riverine transport from terrestrial sources. In the case of 35 polybrominated diphenylethers (PBDEs), there is the possibility that debromination reactions 36 results in the production of lower brominated congeners in situ 10 . Due to their low vapour 37 pressure, and high lipophilicity, their transport on particles in atmosphere and seawater is 38 likely the dominating process. 39 This is particularly true for the Atlantic Ocean, which is affected by emissions from America, 40 Africa and Europe. 11,12 It has been extensively studied on several transects for legacy 41 pollutants, such as polychlorinated biphenyls (PCBs) 11 and polycyclic aromatic hydrocarbons 42 (PAHs) 13 , polychlorinated dibenzo-p-dioxins and furans 1 and hexachlorocyclohexanes [14][15][16] . 43 Most of these transects occurred on European research vessels on their biannual migration 44 from the Arctic to the Antarctic and vice versa. These transects typically follow the western 45 side of the North Atlantic. Sampling on these transects was invariably affected by continental 46 emissions off Europe and Africa, making extrapolations across the entire Atlantic Ocean 47 HRGC/HRMS instrumental analysis for PBDEs was performed on 7890A GC (Agilent, USA) 122 equipped with a 15 m x 0.25 mm x 0.10 µm DB5 column (Agilent J&W, USA) coupled to 123 AutoSpec Premier MS (Waters, Micromass, UK). The MS was operated in EI+ mode at the 124 resolution of >10 000. For BDE 209, the MS resolution was set to >5 000. Injection was 125 splitless 1 µL at 280°C, with He as carrier gas at 1 mL min -1 . The GC temperature program 126 was 80°C (1 min hold), then 20°C min -1 to 250°C, followed by 1.5°C min -1 to 260°C (2 min 127 hold) and 25°C min -1 to 320°C (4.5 min hold). 128 129 PE sheet samplers 130 Blanks and exposed sheets of PE were rinsed with Millipore water, dried with a disposable 131 tissue and soaked for 24 hours in 200 mL of n-hexane followed by 24 hours in 200 mL of 132 DCM. The two solvents were then pooled; concentrated using stream of nitrogen in a 133 TurboVap II concentrator and the extract was split into 2 portions and processed using the 134 same procedure as the high volume samples. 135 136 Quality assurance, Quality control 137 The results for PBDEs in high volume samples are recovery corrected (recoveries ranged 138 from 34 to 110%, Table S7). Method performance was tested prior to sample preparation. 139 Four field blanks were extracted for water and air GFFs each; Four PUF blanks were 140 extracted for air and water samples each (one PUF blank was excluded for air samples) and 141 combined for blank correction and method detection limit (MDL) determination. The MDL 142 was calculated as 3 standard deviations of blank concentrations (for more details, see SI and  143   Tables S2-S6).  144 145

Physicochemical properties
Best-fit values of PE-water (K PEw ) equilibrium partition constants were taken from Lohmann 147 (2012) 27 , and corrected for average sampling temperature and salinity (Table S1) (1) 160 Where R s (m 3 day -1 ) is the overall sampling rate, t is the deployment time (days), V PE the 161 volume of the PE sampler (m 3 ) and K PEw (or K PEa ) is corrected for the average temperature 162 and salinity of the deployment (for details, see SI). Sampling rates were estimated based on 163 PRC loss and typical uptake rates. 21  On the water side, the cruise covered different major currents of the tropical Atlantic Ocean 182 ( Figure 1). Currents were identified based on a combination of changes in temperature (and 183 salinity for the Amazon plume) and typical current fields 35 (see Table S9). In the southern 184 hemisphere, these were the Benguela, South Equatorial and North Brazil Current. Discharges 185 from the Amazon, the North Equatorial Current and Gulf Stream affected samples in the 186 northern hemisphere. We obtained back-trajectories to confirm air mass origin. 21 Initially, we 187 encountered southeasterly trade winds, moving air masses in a westerly direction towards the 188 equator. We then passed the intertropical convergence zone (ITCZ), which was situated at 189 approximately 10 ° N during our cruise. The cruise then continued in the northeasterly trade 190 winds moving air masses towards the west along the equator in the NH. The last few samples 191 were affected by the westerlies, moving air masses eastwards across the Atlantic Ocean. 192

Particle-bound PBDE water concentrations 197
Typical particle-bound concentrations of PBDEs in the surface water were low, at < 1 pg L -1 , 198 with the exception of BDE 209 (mean 7 pg L -1 , median 1.8 pg L -1 ) ( Table S18) frequency of detection for all congeners was only 21% in the southern hemisphere, but 49% 208 in the northern hemisphere. Clearly, the southern tropical Atlantic Ocean is much cleaner (on 209 average 5-times) with respect to particle-bound PBDEs in the surface water than the northern 210 hemisphere tropical Atlantic. We note that the increase in particle-bound PBDEs began at 5 -211 10 °N, not the equator itself (Table S18). This was probably the result of the ITCZ having 212 shifted northwards during the northern summer, as reflected in the back-trajectories. This also 213 implies that particle-bound PBDEs reflected fairly recent deposition events. 214 A closer look at the spatial distribution of particle-bound PBDEs showed that highest

Truly dissolved PBDE concentrations from passive sampling 227
PBDEs were detected in the towed passive samplers deployed in the water, likely due to their 228 extremely high K PEw values (causing strong enrichment in the polyethylene films) (see SI). 229 Total mass of PBDEs accumulated in the towed PE samplers were at least 10 times higher 230 than those in the flow-through water exposed PE samplers, highlighting the boundary-layer 231 limitations we encountered in the laboratory. This was also evident in the detection frequency 232 of PBDEs in both types of deployments. Whereas almost all PBDE congeners were regularly 233 detected in the towed passive sampler (overall detection frequency was 90%, except BDE 66,  (Table S12) sampling is reason for concern. Apparently dissolved PBDE concentrations from active 275 sampling on our study were elevated by 10 -1,000 fold. We examined whether the collection 276 of colloids or microplastics in our PUF-based sampling approach could explain the difference. 277 We assumed 1 mg DOC L -1 during the cruise. DOC-water partition constants, K DOC , were 278 taken either from Burkhard (2000)  There were no significant differences between BDE concentrations on particles between both 295 hemispheres (Table S10). 296 Concentrations were slightly greater than those reported by Xie et al. (2011)  Most PBDEs were detected routinely in the air passive samplers (Table S14)

Air-water exchange 378
Air-water exchange gradients were calculated based on simultaneous passive sampler 379 deployments in air and water. Gradients were based on PBDE concentrations in PE (ng g -1 380 PE) at equilibrium. 46 In short, passive samplers, such as the PE we used, reflect the chemical 381 activity of the BDEs in their respective matrix (air and water in our case). The ratio of those 382 BDE activities (corrected for non-equilibrium) is the activity gradient across the air-water 383 interface. 384 In the water, results from towed PE samplers were used; in the air, the equilibrium-corrected 385 PBDE concentrations from the passive sampler deployed on the ship's mast were used. In 386 approximately 50% of possible cases, PBDEs were > MDL in both phases simultaneously 387 (Table S19) Air-water exchange velocities ranged from < 1 to 36 cm day -1 , and decreased with increasing 394 MW (Table S22). Net gas-phase deposition fluxes ranged from 10s of pg m -2 day -1 for BDEs 395 28 and 85 to around 1 ng m -2 day -1 for BDE 47, 99 and 209 (Table S23) north of 20 °N. The product of MLD and total (sum of truly dissolved + particle-bound) 465 PBDEs yielded the mass loading of PBDEs on particles (ng m -2 ) in the surface water. The mass loading divided by the PBDE removal flux (ng m -2 day -1 ) gave the residence time (days) 467 of particle-bound PBDEs in the surface water of the Atlantic Ocean, assuming particle settling 468 was the only removal pathway. The tight coupling of particle-bound PBDE fluxes from 469 atmosphere and out of the mixed layer could indicate that different fates of particle-bound 470 PBDEs (settling) and those derived from gaseous diffusion, probably due to chemical and 471 biological degradation of dissolved molecules in the water column. In the northern 472 hemisphere, median PBDEs' residence were several weeks, but were around one year in the 473 southern hemisphere ( Figure S5