PERSISTENT ORGANIC POLLUTANTS IN THE ARCTIC, ATLANTIC AND PACIFIC OCEANS

Persistent organic pollutants (POPs) are a group of compounds that are persistent, toxic, bioaccumulative and undergo long range transport. Polychlorinated biphenyls (PCBs), organochlorine pesticides (OCPs) and polybrominated diphenyl ethers (PBDEs) are three groups of POPs. They were widely used as dielectric and coolants fluids, pesticides and flame retardants, respectively. Polycyclic aromatic hydrocarbons (PAHs), normally produced by incomplete combustion of carbonaceous materials, were not in the list but have similar properties as POPs. Even though restrictions and bans over POPs started decades ago (PCBs and OCPs in the 1970s and BDEs in the 2000s), POPs are still detected in the environment and could negatively affect human and wildlife health. Previous studies of POPs monitoring were mostly focused on land or coastal areas; POPs data in remote oceans are lacking. In this study, polyethylene passive samplers (PEs) were used for measuring POPs in the Atlantic and Pacific. The Atlantic study involved deep ocean measurements, while the Pacific study used surface seawater and atmosphere measurements. Deep ocean measurements were conducted by deep moorings at two locations, in the North and Tropical Atlantic Ocean. Results revealed the presence of POPs in the deep ocean with concentrations up to 10 pg L -1 for PCBs, OCPs and BDEs. Oceanic current study suggested that the concentration maximum around 800 m at the Tropical Atlantic site could be potentially originated from the Mediterranean Sea. Mass balance calculation indicated that deep ocean is an important storage for POPs (4.8-26 % of the global HCB environmental burdens). The Pacific study measured both gas-phase and dissolved phase POPs in the Pacific. Atmospheric and oceanic concentrations of PCBs were detected at the magnitude of 1 pg m -3 and 0.1 pg L -1 respectively, except for PCB-8 which was detected at concentrations 10 times higher. HCB dominated in the gas phase (~300 pg m -3 ), while α -HCH dominated in the dissolved phase (~100 pg L -1 ). Large variations of BDEs and PAHs concentrations in either phase were found and were higher than most reported values. Close to equilibrium state of PCBs and OCPs were found in the oligotrophic Pacific. Absolute air-water exchange fluxes were <0.4 ng m -2 day -1 for PCBs and <4.5 ng m -2 day -1 for OCPs. Net air-water exchange gradients strongly favored gas-phase deposition of PBDEs into the water, while mixed gradients were found for PAHs.


INTRODUCTION
Open ocean seawater measurements of persistent organic pollutants (POPs) are scarce due to the difficulties associated with the sampling procedure, contamination and costs of cruises (Gioia et al., 2013). Even scarcer are the data on POPs in the deep ocean, since most of the measurements of POPs in the open ocean were limited to surface seawater.
Little is known of the role played by deep ocean compartments in containing POPs from the surface (Booij et al., 2014).
There are very few previous measurements of deep oceanic POPs: (1) PCBs and PAHs in deep waters in the North Atlantic (near the South-Western edge of the Porcupine Abyssal Plain and around Iceland) using active sampling (Schulz-Bull et al.,1988); (2) PCBs in the central Arctic Basins (Nansen, Amundsen, and Makarov) again relying on active sampling (Sobek & Gustafsson, 2014); (3) PCBs, PAHs, HCB and DDE at 0.1-5 km depth in the Irminger Sea, the Canary Basin and the Mozambique Channel relying on passive sampling (Booij et al., 2014). These studies revealed the existence of POPs in the deep ocean and the significance of deep ocean as a compartment for storing POPs.
The study by Booij et al (2014) was the first to use passive sampling (semipermeable membrane devices, SPMD) to study the POPs vertical distribution in the ocean. Even though active sampling has been traditionally used, it bears the disadvantage of extensive labor and extreme care of controlling blank levels (Booij et al., 2014). Polyethylene sheets (PEs) is one common form of passive sampling devices. It has many advantages compared to other forms of passive samplers, such as simplicity in chemical makeup, low cost, easy handling and a high enrichment of POPs (Lohmann, 2012a). Passive sampling was recently suggested a useful tool used monitoring POPs in open ocean (Lohmann & Muir, 2010).
Under the circumstances that the time to reach equilibrium is not known, the sampling rate (R s ) can be used to derive the state of equilibrium upon retrieval. One way to calibrate R s is to use performance reference compounds (PRCs). PRCs are chemicals that are artificially made which share similar properties to target compounds. However, due to the extensive long time of pre-spiking samplers with PRCs, another approach has also been suggested by Bartkow et al. (2004) using different thickness to confirm that equilibrium has been reached.
In this study, polyethylene passive samplers of different thickness were deployed at two deep ocean sites (eastern Fram Strait and tropical Atlantic) in the Atlantic Ocean to determine vertical distributions of truly dissolved concentrations of several classes of POPs (Polychlorinated biphenyls (PCBs), organochlorine pesticides (OCPs), polybrominated diphenyl ethers (PBDEs) and polycyclic aromatic hydrocarbons (PAHs)). Concentrations derived from this study were compared to previous measurements to validate the reliability of using polyethylene passive samplers. The objectives of this study were to (i) assess the potential of using PEs for monitoring POPs in remote ocean; (ii) contrast various approaches to derive R s (iii) contrast north-south and surface-to-deep gradients and (iv) improve the knowledge of fate and transport of POPs to the deep ocean.

Sampling
during the two cruises. After samples were collected, they were wrapped in clean aluminium foil, shipped back to lab and stored at -4˚C until analysis.

Sample extraction
PEs were wiped clean using Kimwipes with Milli-Q water (Millipore, Billerica, MA).
1 ml extracts were then transferred into amber auto sampler GC-vials with another 2 ml rinse of Turbovap tubes in hexane. Extracts were further blown down to 100 μl by minivap and transferred into spring-bottom glass inserts with a 100 μl hexane rinse of GC-vials. 50 ng of injection standard (p-terphenyl-d 14 and polybrominated biphenyl 30) was spiked into the final extracts for instrument analysis.

Instrumental Analysis
PCBs, OCPs and PBDEs were analyzed on a 6890N GC (Agilent, USA) -Quattro Micro tandem MS (Waters, Micromass, UK) on a DB5-MS fused silica capillary column (30 m  0.25 mm, J&W Scientific). Injection was splitless 1 μL at 3 0 ˚C with Helium as carrier gas at 1 mL min -1 . PCBs and OCPs were analyzed together during the same GC temperature program: 100 ˚C, hold 1 min, 8 ˚C/min to 180 ˚C, 3 ˚C/min to 60 ˚C, 10 ˚C/min to 300 ˚C, and hold for 6 min; PBDEs were run using: 100 ˚C, hold 3 min, ˚C/min to 31 ˚C, and hold for min . PAHs were analyzed on an Agilent 6890 Series GC -Agilent 5973 Network MS Detector on a DB-6 MS fused silica capillary column (30 m 0.25 mm, J&W Scientific). Injection was splitless 1 μL at 80 ˚C with Helium as carrier gas at 1 mL min -1 . GC temperature program was: 90 ˚C, hold 3 min, ˚C/min to 110 ˚C, 8 ˚C/min to 00 ˚C, ˚C/min to 31 ˚C, and hold for min.

Quality assurance/Quality control
Matrix spike and lab blanks were performed for each batch of approximately 10 samples.
Field blanks were taken during the North Atlantic deployment. All blanks were extracted in the same method as samples. Limits of detection (LOD) were derived from field blanks and determined by three times the standard deviation of field blank samples.
Detailed QA/QC information on detection limits are in the supporting information.

PE concentration (C PE ) conversion to environmental concentration
Concentration of target compounds in the PEs (C PE , ng g -1 ) were converted to concentration freely dissolved in water (C w , kg L -1 ) by equation (1) PE w PEw 1 = %equilibrium where PEw K is the compound specific partitioning coefficient between PE and water (L kg -1 ) whose temperature correction for K PEw was done using equation (2) and %equilibrium is the percentage of equilibrium achieved by individual compound in the sampling period which is given by equation (3) PEw where K PEw (T) and K PEw (298)  where R s is the sampling rate (L/day) for one whole sampling sheet; t is sampling time (days); and V PE is the volume of the PE (L).

Using Different Thickness
Sampling rates were estimated based on the assumption that PEs deployed at the same depth/cage were exposed to the same truly dissolved concentration (C w ), and that the same sampling rate applies to all compounds. R s pairs were assumed to be from different combinations of integers from 1~X L/day (X=100, 50, 40, 30 20, 10). For each combination, %equilibrium was calculated using equation (6) and the environmental concentration was further determined using equation (1). The aim was to find the exact pair of R s that minimizes the total of standard deviations of all detected compounds. The total of standard deviations is defined as where n is the number of pairs of data, s is the standard deviation of the derived environmental concentration (x j ) by the two PEs in the same cage

Using PRCs
Site specific sampling rates were also calculated via a nonlinear least-squares method adapted from (Booij & Smedes, 2010). This method only applies to PEs that were pre spiked with PRCs, i.e., 50 μm thick ones for tropical Atlantic deployment.

Estimation of water age using CFCs
Transient tracers, including CFCs are useful as a water mass tracer because their atmospheric concentration can be uniquely related to a calendar year. In turn, water at the ocean surface records this unique concentration based upon air-water gas partitioning.
The equilibrium partitioning between CFC concentration in the air and water is described by Henry's Law. Dissolved CFCs concentration C w (mol kg -1 ) in the interested locations were obtained from CLIVAR and Carbon Hydrographic Data Office (CCHDO).
Temperature and salinity data were obtained from the same origin and were used for deriving the Henry's Law Constant H (mol kg -1 ). The molar ratio of CFCs in the atmosphere  (mol CFC mol -1 air) were then calculated using  data were from CDIAC (Bullister, 2014). SF 6 data were of the highest priority if available; if not, CFC-12 data were used instead.

Equilibration of target analytes during Deployments
Passive samplers of different thickness were deployed to gauge the equilibration of target compounds during deployments. Obtaining similar (mass-normalized) concentrations of target analytes would indicate equilibrium had been attained during the field deployments. Yet the deployment times (1 year each) were not long enough to reach equilibrium for most target analytes, hence we relied on other approaches to deduce truly dissolved concentrations of target compounds in the Atlantic Ocean. PRC derived sampling rate (4.9 and 8.7 L/day) at two depths indicated the necessity to deduce siteand depth-specific sampling rates for each deployment.

Estimating Sampling Rate and state of Equilibrium
Sampling rates were obtained by the modeling approach mentioned above. From previous observations, Rs are typically in the range of 1~100 L day -1 . Scenarios (1) to (6) Figure S1 & S2) and thus large variations of Rs were not expected between these two sites. The same Rs range (1~10 L/day) was chosen for the Tropical Atlantic.
The PRC derived Rs were 8.7 and 4.9 for 84 m and 251 m at depth which were less than the variance of currents between these two depths, indicating that deployment cages effectively buffered samplers from current velocity. This limiting factor can also be shown in Table S5 that no strong and significant positive correlations of Rs with current can be found in either scenario. Given the small range of current speed at the five sampling depths and the current limiting factor from the deployment cages, scenarios (6) were selected. Comparison of the result of the scenario (6) with PRC derived Rs were shown in Table 1.1 suggests good agreement between these two methods. Rs changes within a factor of 2 only gives difference in final concentration (North Atlantic PCBs) within a factor of 2, as shown in Table S7.
The percentage of equilibrium which target compounds achieved is shown in Table S8 and All concentrations discussed below are based on modeling results from scenario 6 (Rs = 1~10 L day -1 ).
HCB concentration was found at 6.0 pg L -1 near surface, higher than what was found in the North Atlantic (0.1-3 pg L -1 ) by Lohmann et al. (2012b) but within the range of Northern Hemisphere average (2-9 pg L -1 ) by Booij et al. (2007). p,p'-DDT, o,p'-DDT and p,p'-DDE were all within the range of 0.2-0.5 pg L -1 , which was in agreement with North Atlantic (0.1-3 pg L -1 ) by Lohmann et al. (2012b) and close to the lower end of Northern Hemisphere average (0.3-1.4 pg L -1 ) by Booij et al. (2007) and detected near 30˚N Atlantic (0.5 pg L -1 ) by Iwata et al. (1993).
The only detectable PBDEs were 47, 100 and 99. Concentrations were 1.4, 0.3 and 1.6 pg L -1 , slightly higher than (~ 0.5 pg L -1 for 47 and around 0.1 pg L -1 for 99 and 100) by Lohmann et al. (2013a) and (0.02-1.05, nd-0.11 and 0.01-0.53 pg L -1 respectively) by . In the first study, PEs were used for collecting samples; on the second cruise, XAD resins and GFF were used as active sampling. Nonetheless, results from all three studies were fairly similar to each other.
Detected ∑ 7 PAHs was 83 pg L -1 near surface; which is within the range of (∑ 10 PAHs, 58-1070 pg L -1 ) by Nizzetto et al. (2008) and lower than (∑ 3 PAHs , average of 270 pg L -1 ) by Lohmann et al. (2013b). The higher end of the previous range occurred near the northwest coast of Africa, which was potentially influenced by emerging oil industry, biomass burning and natural source of PAHs in Africa. Therefore the unexpected large numbers (0.7-1 ng L -1 ) were not representative of remote ocean concentrations in tropical Atlantic. The second study was near the coast of America thus detected ∑PAHs were not comparable to remote ocean, either. As to individual PAHs, phenanthrene, fluroanthrene and pyrene were dominant, which is in agreement with  and (Lohmann et al., 2013b).
Taken together, the concentrations of various POPs near the surface of the tropical Atlantic were in agreement with previous measurements reported for the remote Atlantic.

North Atlantic
In the North Atlantic, surface ∑PCBs concentration of the dissolved phase was 0.8 pg L -1 in this study, which was comparable to what was observed for ∑ 7 PCBs (0.7 pg L -1 ) (Sobek & Gustafsson, 2014), ∑ 6 PCBs (< 1 pg L -1 ) (Gioia et al., 2008a). As for individual congeners, the highest concentration was determined by PCB-28, followed by 18, 44, 52 and then 101, 138, 153. This result is in agreement with result from Gioia et al. (2008a). 28,52,101,118,138,153 were the most detected congeners in previous studies; they were also present in this study.
Hexachlorobenzene (HCB) was the OCP detected at the highest concentration. The near surface concentration was 10 pg L -1 , higher than those reported for NABE (0.1-0.8 pg L -1 ) (Zhang et al., 2012) but close to results from the ARKXX cruise (high Arctic , 4-10 pg L -1 ) (Lohmann et al., 2009), results for the East Atlantic Ocean (2-9 pg L -1 ) (Booij et al., 2007) and for the Arctic Ocean (7 pg L -1 ) (Strachan et al., 2000). It was mentioned by Lohmann et al. (2009)  There was no HCH detected, due to the less hydrophobic nature of HCHs which reduced their uptake by PEs. p,p'-DDE and p,p'-DDD are two breakdown products of p,p'-DDT.
Fluoranthene was detected with the highest concentration among all PAHs.
Concentration of fluoranthene at 213 m (94 pg L -1 ) fell in the range of other studies (Lohmann et al., 2009;Booij et al., 2014) in the similar area. There was no significant gradient of concentration with latitude. The ratio of fluoranthene over fluoranthene and pyrene ranged between 0.6 and 0.8. Lohmann et al. (2009) reported this ratio of 0.9-1.0 for the North Atlantic and Arctic Ocean; Schulz-Bull et al. (1998) reported 0.6 for the surface water of surface water in Irminger Sea. All these results suggested a combustion origin for the PAHs in the North Atlantic/Arctic region.
In summary, the concentrations of various POPs near the surface of the North Atlantic Ocean were in agreement with previous measurements reported for the remote Atlantic.
Overall, this validates the use of PE samplers and the derived concentrations.

Certainty of derived truly dissolved concentrations
We compared how the use of different Rs (PRC-approach versus model-derived) affected truly dissolved concentrations of POPs. Maximum differences of 19% and 7% in equilibrium for PCBs were obtained (equilibrium using PRCs versus the equilibrium using modeling). PCBs cover a wide range of properties and are thus representative of the other compounds. Therefore, uncertainties associated with R s were within a factor of 2 at most. The absolute concentrations of each individual compound between two PEs in one cage were also compared. The relative standard deviation of each detected compound was calculated (Table S10). For North Atlantic samples, averaged relative standard deviation between two PEs from one cage at the same depth varied between 35-61 %; for tropical Atlantic samples, this range was 27-68 %. Variation was largely associated with the different detection limits between PEs of different thickness. The surface area of 800 μm PEs were twice as much as 1600 μm PEs, resulting in a faster uptake (Bartkow et al., 2004) and more contaminants above detection limits. Similarly, the surface area of 800 μm PEs and 0 μm PEs were the same, while the former one weighed 6~7 times more than the latter. Not surprisingly, more contaminants were above detection limits for 800 μm PEs than for the 50 μm PEs. The deviation of POP concentrations between 800 μm PEs and 50 μm PEs were, on average, larger (61 ± 9 %) than between 800 μm PEs and 1600 μm PEs (45± 12 %). For a better consistency between two sites as well as more

Latitudinal Fractionation
The ratio of the concentration of each individual compound near the surface layer (231 m) at 79˚ N divided by the surface concentration at 24˚ N (26 m) was plotted as a function of their log vapor pressure (P L ) (Figure 1.5). The overall correlation was not strong (R 2 = 0.25), due to an outlier (PCB-28). Outlier was defined as values outside of 1.5 times of interquartile distance (IQD = Q3-Q1) subtracted from or added to the first quartile (Q1) and the third quartile (Q3) (Tukey, 1977). Similarly, an increasing trend of concentration ratios of PCBs (88˚ N : 62˚ N) along with log vapor pressure was found by Sobek & Gustafsson. (2004). A significant higher concentration of PCB-28 in the north could be resulted from an unknown emission source near the sampling area. After removing PCB-28, the correlation between north-south ratio of POP concentrations and log P L was much improved (R 2 = 0.95) and became significant (p<0.05). The results imply that higher mobility compounds (higher log P L ) display a relatively greater abundance up north than compounds with lower mobility. This supports the ideas formulated in the cold condensation theory. We note, however, that the two samples we used for the comparison were not real surface samples, especially for North Atlantic site.
Therefore, the conclusions reached here have to be interpreted with caution, and are obviously affected by compounds not adhering to the trend.

Comparison with the other deep ocean POP measurements
Depth profiles were plotted as in Figure

Depth profile shapes
PCBs, OCPs and PAHs congeners displayed similar depth profiles; mostly with a maximum at 800 m. Vertical profiles for PBDE are significantly different from the other compounds. All BDEs congeners exhibited a drastic decline in truly dissolved concentrations below 251 m. Little BDEs were detected at depth, possibly linked to the fact that production of BDEs peaked 20~30 years later than those for PCBs and OCPs.
Thus, PCBs and OCPs had more time to penetrate the deeper layers of the oceans, while PBDEs have only touched the surface ocean.

Explanations for depth profile
No measurements were done above 1400 m in the study of (Booij et al., 2014) and due to the similar concentrations found near the surface at other locations in the Atlantic, the authors concluded that no large concentration gradients existed in the upper 1400 m at Canary Basin. However, in the present study, a concentration maximum existed at 800 m which was significantly different from the other depths. We note that neither study was able to fully resolve depth profiles satisfactorily as only a few samplers were deployed, potentially missing important vertical features in POPs concentration.
Two hypotheses were tested to explain the shapes of the PCBs, OCPs and PAHs depth profiles: i) Particle binding/sinking; ii) Water current transport. Chemicals with higher K ow (partitioning coefficient between octanol and water) tend to bind to particles more easily. With particles sinking and getting remineralized in the deep, POPs are released back into the water. PCBs, for instances, have different levels of chlorination and those with high chlorination degrees have higher tendency to bind to particles. The composition from different PCBs chlorination groups were plotted in Figure 1.4. There was little variation in the PCBs composition along with depth; the only noticeable trend is that penta-chlorinated PCBs decreased towards 250 m and a slight increase of tri-PCBs towards 1800 m. No trend of increasing contribution from higher chlorinated-PCBs was shown, suggesting that particle binding/sinking process does not explain for the 800 m maximum.
The existence of Mediterranean water has been observed in East Atlantic (Martí et al. 2001). Mediterranean water sinks and mixes with Eastern Atlantic Water after it flows out of the Straits of Gibraltar, reaches equilibrium around 1000 m in depth and spreads across the North Atlantic (Knauss, 2005). The implication from our results is that the upper 1800 m of the water column was not well mixed and there were multiple water layers which potentially had different water mass origins, potentially affecting POPs concentrations and profiles.

Depth Profile Shapes
Depth profiles were plotted as in Figure 1.3. All four groups of compounds showed similar depth profiles: a general decrease trend towards the deep and some maximum concentration appearing at 500 m in depth. Decreasing profiles of PCBs were observed in the 1990s in North Atlantic (Schulz-Bull et al., 1998;Schulz et al., 1988), while a nutrient-like profile was shown recently addressing the importance of advective flow-off from the continental shelf (Sobek & Gustafsson, 2014). The distribution pattern observed in this study neither follows a decreasing nor an increasing trend.

Explanations for depth profile
Similar to the discussion above, particle binding and sinking origination was tested by PCBs chlorination composition plotted in Figure 1.4. Compared to the results from the tropical Atlantic site, smaller degrees of chlorination tend to yield larger contribution of the ∑PCBs in the North Atlantic, which means the contribution from each group has the order of tri-> tetra-> penta-> hexa-. This composition pattern of chlorinated groups are in line with (Sobek & Gustafsson, 2014) and (Gioia et al., 2008a). No significant fraction of higher chlorinated PCBs was observed along depth, indicating that particle sinking was not a major contributor to the PCBs in depth in North Atlantic either.
The Fram Strait is the pathway for water exchange between North Atlantic and Arctic Basin at HC is locally topographically steered but does not achieve the cross sill advection (Appen et al., 2015). Unlike the Tropical Atlantic sampling site, it is hard to derive the water mass origins for certain depths at the North Atlantic sampling site due to the complexity and vigorousness in the water mixing processes.

Tropical Atlantic
There was no available SF 6 data for tropical Atlantic close to our deployment period; CFC-12 data were used instead to assess water mass origin and age. CFC-12 derived ventilation age was plotted against the ∑PCBs in Figure S3. Detailed information for ventilation age calculation was given in Table S11. The three sites chosen from CCHDO for the tropical Atlantic gave close CFC-12 data throughout the water column, indicating little variance of water composition around (24 ˚ N, 38 ˚ E). Ventilation age was then derived by averaging out the ventilation age from SITE1 to SITE 3 (Table S11).

North Atlantic
SF6 data was used for deriving the ventilation age of water masses in the North Atlantic.
Large variations occurred between different sites chosen for the SF6 data, resulted from the complexity in bathymetric and water current conditions in this area. The closest sampling SF6 location to our PE sampling site which also covers the whole PE sampling depths was shown in Table S11. Figure S3 indicated a maximum in ∑PCBs at a ventilation age of 10~20 years, younger than the maximum at tropical Atlantic site.
Again, due to the limited data points, it is hard to accurately determine where the ∑PCBs maximum would occur. The whole distribution pattern of ∑PCBs with ventilation age still followed the global emission pattern, with one peak in the middle and decrease on both sides. The shift in the concentration peak of ~20 years could be an oceanic POPs response time lag not captured by the tropical Atlantic site measurement.

Mass balance Implications for POPs in the Ocean
For HCB, the vertical profiles reported here show no significant difference in absolute concentrations across the Atlantic Ocean. Therefore, we assumed a uniform spatial distribution across the Atlantic basin. Pilson estimated the whole surface area of Atlantic as 8.65  10 7 km 2 with average depth of 3,700 m (Pilson, 2013). The upper ocean (0-1,200 m) is loaded with HCB at a concentration of 9.6 ± 3.5 pg L -1 (mean ± standard deviation), while the deep ocean (1,200-3,700 m) has a concentration of 4.4 ± 1.6 pg L -1 , based on the results of this work. The total amount of HCB residing in the Atlantic Ocean is 1,947 ± 709 t; it accounted for 45 ± 16% of the total HCB stored in the ocean if using the estimation (4289 t) (Barber et al.,2005) . The total global production of HCB was estimated as >100,000 t by ATSDR (1997). The contemporary environmental burden of HCB was calculated as 10,000-26,000 t (Barber et al., 2005). Hence, the Atlantic Ocean stores less than 2.6 % of HCB ever produced, but contains 4.8-26 % of the global HCB environmental burdens.

IMPLICATION
Traditionally it is believed that the major inventory of POPs exist in the upper ocean only. However, this study supported previous work in highlighting the important role of deep ocean as a compartment storing POPs. Results from two contrasting sites also imply that the distribution of PCBs in the Oceans is influences by their mobility, with more mobile compounds being more abundant in the North rather than tropical Atlantic Ocean.
The presence of numerous POPs in deeper water suggests that the deep ocean carries a significant mass already, particularly of the legacy POPs. As an example, PBDEs had penetrated the deep water masses of the Atlantic Ocean to a much smaller degree than the legacy PCBs and OCPs.      Table S1). Close to equilibrium state of PCBs and OCPs were found in the oligotrophic Pacific.

TABLES AND
Absolute air-water exchange fluxes were <0.4 ng m -2 day -1 for PCBs and <4.5 ng m -2 day -

INTRODUCTION
Very few studies have been made regarding POPs concentrations in the Pacific Ocean. Iwata et al. (1993) reported the first simultaneous measurement of air and water organochlorines across the global ocean. Results suggested various distribution patterns for different compounds (Iwata et al., 1993). Recently, Zhang & Lohmann (2010).
reported the distribution of PCB and HCB in the Pacific Ocean; Gonzalez-Gaya et al. with only a few regional deposition dominating over volatilization (Gioia et al., 2008b).
To the contrast, net volatilization of PCBs were suggested across an East-West transect near the tropical Atlantic (Lohmann et al., 2012b). From the same cruise, HCB were suggested to be near equilibrium while DDTs were indicated to be depositing into the ocean (Lohmann et al., 2012b). Net deposition of BDEs into the surface waters were implied in the tropical (Lohmann et al., 2013a) and southern Atlantic Ocean . PAHs were found to be undergoing net deposition across the tropical Atlantic Ocean, with conditions closer to equilibrium off the U.S. East Coast (Lohmann et al., 2013b). The only air-seawater exchange study in the Pacific so far reported volatilization of PCBs and near air-water equilibrium for HCB (Zhang & Lohmann, 2010).
The PE air samplers were exposed in inverted stainless steel bowls. The water PE samplers were fixed in a steel pipe connected to the flowing seawater inside the ship's laboratory. Several PEs were deployed inside the ship at several locations including the science lab, galley and engine room.  Table S17.

Sample Analysis
Sample extraction and instrumental analysis were discussed elsewhere (Chapter 1).

Quality assurance/Quality control
QA/QC details were discussed elsewhere (Chapter 1).

Statistics of comparison between the cruises
Mann-Whitney U test were used for comparing the concentrations and fluxes of POPs between the cruises. The null hypothesis is that there is no difference of the distribution of POPs between the two cruises. A significant level of 0.05 was chosen.
For concentration comparison, one representative compound from each group (PCB, OCP, BDE, PAHs) was chosen. They were: PCB-11, HCB, BDE-47 and chrysene/benzo(a)anthracene. They were selected since they were the highest concentration among those detected at both cruises.

Air-Water Exchange Gradient and Flux Calculation
The gradient of air-water exchange is the ratio of the concentration of POPs in the PE at equilibrium with the air (C eqA , ng/g) to the concentration at equilibrium with the water (C eqW , ng/g):  where T is the absolute temperature (K), M air is the average molar mass of air (28.97 g mol -1 ), M i is the chemical's molar mass ( g mol -1 ), p is the gas phase pressure (atm), Vair is the average molar volume of the gases in air (20.1 cm 3 mol -1 ), Vi is the chemical's molar volume (cm 3 mol -1 ), µ is the water viscosity at T (10 -2 g cm -1 s -1 ).

Pacific (2010)
Average concentrations from the present study were based on 4 samples from each cruise HCB was detected at 340 pg m -3 in the air and 12 pg L -1 in the water. HCHs were more dominant in the water than in the air, and α-HCH was detected at higher concentration than ϒ-HCH in both air and water. In the atmosphere, the concentration of detected ∑HCHs (α-HCH, ϒ-HCH) were ~30 times smaller than found by Iwata et al. (1993), while the ∑CHLs (t-chlor, c-chlor, t-nona) were only 10 times smaller. However, the concentration ranges of α-HCH and ϒ-HCH (14-35 and 1.4-4 pg m -3 ) were consistent with more recent studies: (6.5-19 and 1-4.6 pg m -3 ) by Ding et al. (2007) and (26-56 and 10-36 pg m -3 ) by Wu et al. (2010). HCHs were dominant by α-HCH (90%) and followed by ϒ-HCH (10%); this α/ ϒ ratio is higher than detected in Northeast Asia where α-HCH (71%) and ϒ-HCH (23%) were found (Baek et al., 2013) . The reason is that long range atmospheric transport (LRAT) causes an increase in α/ϒ-HCH ratio by either direct partitioning into seawater or through washout by precipitation (Ding et al., 2007). In the dissolved phase, a similar ratio of ∑HCHs to the ∑CHLs was observed -110 from this study compared to the ratio of 120 (Iwata et al., 1993), although the absolute concentration from this study are 10 times smaller than detected 15 years ago.
BDE-47 and 99 were the two dominant congeners in the air (21 pg m -3 and 15 pg m -3 respectively). Concentrations were 40 times lower during a cruise from east Asia to the Arctic . Larger concentration of BDEs was also found of dissolved phase from this study.
Biphenyl, acenaphthene, fluorene, fluoranthene, pyrene, retene, benzo(a)anthracene, chrysene, benzo(b/k)fluoranthene were the detected PAHs, which were also detected during a cruise from east Asia to the Arctic except for biphenyl and retene (Ma et al., 2013). Much higher concentrations in the atmosphere were found from this study; acenaphthene on average of four samples were 50 times higher than reported North Pacific maximum from (Ma et al., 2013), and the other congeners were around 10 times higher from this study (Ma et al., 2013). However, there were great variations between the four air samples in this study as well.  (Zhang & Lohmann, 2010). Results from both studies were much lower than reported by Iwata et al. (1993), in which 130 pg m -3 (atmosphere) and 24 pg L -1 (seawater) were detected for ∑PCB in the Northern Pacific (Iwata et al., 1993).
HCB, α-HCH and ϒ-HCH were detected as OCPs. HCBs dominated in the air (220 pg m -3 ) while HCHs dominated in the dissolved phase (~140 pg L -1 ). HCB from this study (220 pg m -3 and 9 pg L -1 ) were much higher than (61 pg m -3 and 0.9 pg L -1 ) (Zhang & Lohmann, 2010); while close to the upper range for atmosphere background in Japan (137 pg m -3 ) (Murayama et al., 2003) and seawater average for the northern hemisphere (12 pg L -1 ) (Barber et al., 2005). α-HCH and ϒ-HCH were detected at close concentrations, which were ~10 pg m -3 and ~140 pg L -1 in the air and seawater, respectively.    ; while this was not seen in this study.

Concentration comparison between cruises
In general, there were larger variations between samples in the northern cruise, indicating more even POPs distribution along the San Diego-Hawaii transect than along the San Francisco Bay-Hawaii transect.

Air-water exchange gradients
Significant departure from equilibrium was considered as gradient outside of 0.137-3.91 (Ruge, 2013). Any gradients values larger than 3.91 was considered as evaporation; any values smaller than 0.137 were considered as deposition.
Air-water exchange gradients were plotted in Figure 2 A detailed air-water exchange gradients values for each single sample were displayed in Table S17. One thing to note is that A/W-1 and A/W-5 from Table S17 were not used for Also, the sampling location of A/W-4 was close to land emission and could thus be affected by land emissions (Lohmann et al., 2012b).
To the contrast, sampling areas for the other samples were mostly subtropical ocean gyres where productivity is low. There was very limited particle binding process to remove Similarly, OCPs favored equilibrium/evaporation for most samples, with a few exceptions of deposition for cis-chlordane and trans-nonachlor (Table S17). HCBs were at equilibrium for all samples collected from two cruises, in agreement with findings from (Zhang & Lohmann, 2010). HCHs suggested evaporation/equilibrium in this study.
Results for ϒ-HCHs from other studies are a mix. Some were dominated by net deposition such as in the North Atlantic/Arctic region (Harner et al., 1999;Jantunen et al., 2008;Lohmann et al., 2009) ; one study suggested net volatilization in the Pacific/Arctic region (Ding et al., 2007).  Table S17). Highly BDE polluted ocean water could have possibly contributed to this phenomenon. Similar discussion can be applied to PAHs gradients.

Air-water exchange fluxes comparison with literature
Fluxes calculated using a micrometeorological approach was compared with results using the Whitman two-film model in (Wong, 2010). The relative standard deviation (RSD) were estimated as 83-277% for the micrometeorological calculated flux (F M ) and 127-288% for the two-film model calculated flux (F TF ). For all events except for event #14, mean F M was 0.44 ± 0.2 ng m -2 h -1 and mean F TF was 0.29 ± 0.15 ng m -2 h -1 . The good agreement of RSD and mean fluxes calculated from the two methods validated the usage of either method. In the present study, a similar uncertainty 51% (0.15/0.29) for fluxes was adopted as from (Wong, 2010).
The fluxes for BDEs and PAHs in this study could be biased due to the high concentrations detected in the air and water, so that the discussion was moved to Supporting Information.

SUMMARY
This study adds to the current database of POPs distribution in the East Pacific Ocean.

An additional cruise -Atlantic 2010
There was one cruise in the Atlantic that started near Bermuda in mid June 2010 and then headed towards the mid-ocean ridge and finally returned to Bermuda in mid July 2010. The air and water samples collected were marked on Figure 2.1. The numbers on the figure represented the order of samples taken. For each number, there were one air and one water sample collected. In addition, in order to check the background contamination from the ship, PEs were deployed in the lab, engine room and galleys on the ship.

Detected POPs Concentration
Concentrations were biased high, and therefore the discussion of this cruise was not the focus in the main context but only mentioned here. PCB-11 and 118 were the only two congeners detected in the Atlantic cruise, with PCB-11 at much higher concentrations averaging 40 pg m -3 while PCB-118 below 5 pg m -3 . The mean concentration of PCB-118 from this study (0.7 pg m -3 ) is slightly smaller than the measurement (1.5 pg m -3 ) by Jones et al. (2004)  Similar case was found with dissolved BDEs, with concentration from this study at 30 and 20 pg L -1 respectively for BDE-47 and 99; higher than 0.5 pg L -1 (BDE 47) and < 0.1 pg L -1 (BDE 99) by Lohmann (2013a); < 1.05 pg L -1 (BDE 47) and < 0.53 pg L -1 (BDE 99) by . Acenaphthylene, Chrysene/Benzo(a)anthracene and Benzo(b/k)fluoranthene were the only three detected PAHs. The averaged air concentration were 250, 28 and 2 pg m -3 ; dissolved concentration were 80, 30 and 1.5 pg L -1 .

Comparison between Atlantic and Pacific, 2010
There were fewer PCBs congeners and OCPs but more BDEs congeners detected in the Atlantic. Due to the enormous difference in acenaphthene and fluorene from literature data in the Pacific cruise, conclusions were hard to make whether the samples were representative of PAHs in the open Pacific ocean. Table S7 gave the p-value for all the comparisons of both representative compound from each group (PCB-11, HCB, BDE-47 and crysene/benzo(a)znthracene) and the sum of each group (∑PCB, ∑OCP,∑BDE, ∑PAH) between the two basins. The significant differences between the two cruises lied in the same pattern in the atmosphere and water, such that significant difference in the air resulted in also significant difference in the water. This consistency in the POPs distribution within each basin suggesting little contact of air/water between the basins and undergoing air-water exchanges within each basin. Dissolved phase ∑OCP difference was mainly from the ∑HCHs that were only detected in the Pacific. It was pointed out that HCHs were still in use in developing countries such as China and India, while banned for decades in most of the industrialized countries (Kallenborn et al. 2007). Due to the limited transportation between the basins, it is not surprising to see HCHs only being detected in the Pacific.

Gradients
The detailed gradient values for this cruise was given in Table S22. Similar patterns were shown as Pacific 2010 cruise, such that most PCBs and OCPs favored evaporation/equilibrium while BDEs favored deposition and PAHs favored deposition/equilibrium.

Fluxes
The overall air-water exchange fluxes of PCBs for the Atlantic cruise was small (~0.08 ng m -2 day -1 ), smaller range than fluxes in the Atlantic (-7 -0.02 ng m -2 day -1 ) (Gioia et al.,2008b), Pacific Ocean ( 0.5 -30 ng m -2 day -1 ) (Zhang & Lohmann, 2010). An averaged flux for HCB was between 0.8 ng m -2 day -1 . Volatilization/equilibrium of HCB was found in the tropical Atlantic (Lohmann et al., 2012b); equilibrium was found in the North Atlantic (Lohmann et al., 2009) and the Pacific (Zhang & Lohmann, 2010). BDEs favored deposition in all samples. Fluxes in this study were dominated by BDE-47, 99, 100 and were as large as -13 ng m -2 day -1 (BDE-47) due to the large concentration detected either in the air or water. For all congeners, Atlantic samples yielded larger deposition fluxes than Pacific samples. The absolute flux for BDE-47 from the Atlantic cruise (-13 ng m -2 day -1 ) was more than 10 times higher than (<492 ng m -2 day -1 ) , (median: 83 pg m -2 day -1 )  and (~ 320 ng m -2 day -1 ) (Lohmann et al., 2013a). It could not be ruled out that contamination occurred during the Atlantic cruise. The overall PAHs fluxes for the Atlantic cruise were small, with a total flux of -6 ng m -2 day -1 . There were no significant differences in fluxes between the basins.  Figure S7. Averaged air-water exchange fluxes of PCBs, OCPs, BDEs and PAHs from the three cruises.
Error bars represent standards deviation between samples from each cruise.