Spatial Distributions and Seasonal Changes of Current‐Use Pesticides from the North Pacific to the Arctic Oceans

Distributions of eight targeted current‐use pesticides (CUPs) chloroneb, simazine, atrazine, alachlor, dacthal, chlorobenzilate, methoxychlor, and permethrin were investigated in seawater and the atmosphere in a region covering the North Pacific to the Arctic Oceans during the 7th and 8th Chinese National Arctic Research Expedition (2016 and 2017) voyages of the research vessel R/V Xuelong (Snow Dragon in English). Total CUP concentrations in seawater have prominent seasonal and latitudinal trends, with higher concentrations occurring at lower latitudes in early summer. The major contributors of the CUPs (∑8CUP) did not alter over different seasons with dominant chloroneb, alachlor, and atrazine accounting for more than 90%, but their concentrations did have marked seasonal changes. However, the compositions of the eight analyzed CUPs varied indistinctly between seasons indicating a possible combined environmental impact on these compounds rather than the effects of individual chemical properties. Three‐day backward air mass trajectories indicate that atmospheric masses from northeastern China are responsible for the high concentrations of CUPs in East China and Japan Seas, whereas those from the North Atlantic Ocean contribute to the low levels in local area. Fugacity ratios indicate potential volatilization and equilibrium of chloroneb in the Canada Basin of the Arctic Ocean and Japan Sea, respectively, and deposition of other CUPs in both regions. However, atmospheric concentrations are decoupled from those in seawater, which indicates a low exchange rate.


Introduction
Current-use pesticides (CUPs) including fungicides, insecticides, herbicides, and rodenticides (Li, Zeng, et al., 2014;Smalling et al., 2013) have relatively shorter environmental half-lives compared to legacyuse pesticides and are widely used around the globe [http://www.fao.org/statistics/en/]. Compared with legacy persistent organic pollutants banned by the Stockholm Convention, CUPs have similar control efficiencies but relatively low bioaccumulation factors, lower potential for long-range transport, and lower toxicity to non-target biota. As a result, CUPs are used heavily in agricultural activities (Feng et al., 2011;Lammel et al., 2015;Yeo et al., 2003;. For example, the total usage of pesticides in Asia in 2016 was 4.14 million metric tons and up to 1.14 million metric tons in Europe (http://www.fao.org/faostat/en/#data/RP). However, CUPs are not transient or harmless in the environment. These pesticides are being increasingly studied because they have been found to have distinct half-lives in natural conditions from previously expected through laboratory experiments (Gouin, Cousins, et al., 2004;Ruggirello et al., 2010;Thinh et al., 2018) and have been detected in remote areas such as mountains, lakes, groundwater, and polar regions beyond the direct influence of human activities (Guida et al., 2018;Kurt-Karakus et al., 2010;Luek et al., 2017;Muir et al., 2004;Westgate et al., 2013;Yao et al., 2007). Furthermore, the bioaccumulation of CUPs and their passage through food chains pose potential risks to human health (Dittmann et al., 2012;Macdonald et al., 2002). Accordingly, research has become more focused on CUPs sources, environmental behavior, transport, and fate in the environment (He & Balasubramanian, 2009;Macdonald et al., 2003). Jantunen et al. (2015) recently summarized 20 years of study of CUPs in western Arctic seawater and the atmosphere. Seawater and atmosphere are both crucial pathways for the transport of CUPs to high latitudes, with the mode of transport depending on their chemical properties. For example, oceanic currents transport 80-90% of β-hexachlorocyclohexane to the Arctic Ocean, with atmospheric transport being almost negligible (Li et al., 2002). The long-range atmospheric transport (LRAT) model describes a "grasshopper effect" (Wania & Mackay, 1993), which enables the rapid transport of pollutants to remote regions (Bailey et al., 2000;Bigot et al., 2016;Guida et al., 2018;Hung et al., 2016). The model is appropriate for volatile or semivolatile organic compounds with high Henry's law constants. Oceanic currents have huge transport capacities (Li & Macdonald, 2005) and are important in CUP transportation (Lammel et al., 2017) despite their slow movement relative to atmospheric transport. River runoff and groundwater are significant sources of CUPs (Lin et al., 2012;Zhong et al., 2014), and after entering open seas, CUPs will be eventually driven out of the aqueous phase as a result of decreased solubilities caused by salt named the "salting out" effect (Rice et al., 1997).
Seasonal changes in CUP concentrations are dramatic, with temporal and spatial changes varying globally. During spring and summer when the application time is, high concentrations of CUPs in both rivers and coastal seawater are correlated with the intense agricultural practices (Liu et al., 2018;Schottler et al., 1994), although there should be a time lag between agricultural use and peak levels. In the atmosphere, CUPs are volatilized during field spraying (Li, 2012) and enter global atmospheric circulation. This process becomes more significant from spring to summer as temperatures increase (Macdonald et al., 2003), which leads to increasing trends in the air. Afterward, CUPs are transported to remote regions and are precipitated through dry and wet deposition processes by adsorption on aerosol and particles (Ballschmiter, 1992;Wania & Mackay, 1993). Deposition occurs significantly when temperature drops (Xiao et al., 2010), especially obvious in polar regions. Additionally, the polar regions, where agricultural activities are far away, may also receive significant quantities of CUPs released by sea ice melt in summer (Ma et al., 2011). Similar to the behaviors through atmospheric transport, distributions of CUPs in seawater are dynamic and changeable during transportation by ocean currents (Hargrave et al., 1997), with concentrations decreasing over time and distance (Mai et al., 2016) in accordance with microbial activities in water and CUPs' environmental persistence (as characterized by half-lives).
The present study involved a large-scale investigation of the temporal and spatial distributions of CUPs in surface seawater and atmosphere. Previous studies have focused mainly on CUPs at specific regions or locations (Eickmeyer et al., 2016;Li et al., 2016;Ruggirello et al., 2010;Yeo et al., 2004), with few having largescale profiles (Ma et al., 2015). Zhong, Xie, Cai, et al. (2012) studied CUPs during the 4th Chinese National Arctic Research Expedition (CHINARE) in a similar region of the North Pacific Ocean. However, those authors focused mainly on air-sea exchange, with little consideration of seasonal effects. The present study covered two seasons and focused on a different set of CUPs.
The main aims of this study were (1) to determine profiles of eight CUP-chloroneb, simazine, atrazine, alachlor, dacthal, chlorobenzilate, methoxychlor, and permethrin-in surface seawater over an area extending from 125°E/33°N to 167°W/83°N and in the atmosphere from 28°W/61°N to 123°E/31°N; (2) to collect samples in both early and late summer to indicate seasonal differences in CUP levels; and (3) to calculate seawater-atmosphere fugacity ratios (FRs) in the Japan Sea, North Pacific Ocean, and Canadian Basin as indicators of deposition or volatilization.

Sampling in the Field
A new high-volume solid-phase extraction technique coupled with in situ ultrasonic pretreatment ("highvolume SPE ISIU") was used for ultratrace CUP analysis (details are provided in Figures S1 and S2 in the supporting information; Cai, 2017aCai, , 2017bCai, 2018). The 7th and 8th CHINARE cruises took place from 12 July to 23 September 2016 and from 27 July to 7 October 2017, on the R/V Xuelong (Snow Dragon in English). A total of 106 samples, 90 seawater and 16 atmospheric, were collected in 2016 (seawater samples: W01-W74)/ 2017 (seawater samples: W01′-W16′) and 2017 (air samples: A01-A16), respectively. Detailed sampling information is given in Tables S1 and S2 in the supporting information. For seawater sampling,~200 L of surface seawater at the depth of 4 m was collected through a stainless-steel pipeline and concentrated onboard at each station using a high-volume solid-phase extraction sampler fitted with an adsorption column. The core part of the sampler is the adsorption column, which consists of nine components ( Figure S1). There are three grooves to clamp the stainless-steel filters. The top one acts as protection and buffering effect to prevent particles with a diameter of more than 150 μm in case of breaks of glass fiber filter (GF/F) membrane. The second and third stainless-steel filters are used to hold XAD-2/4 materials. Seawater was prefiltered through a GF/F Whatmann filter (0.7-μm pore size, 142-mm diameter) to remove suspended particulates, at a flow rate of 600-1,000 ml/min. After sampling, the adsorption column was dried in a drying oven onboard at 45°C for 96 hr before being sealed in an aluminum foil bag, which was not opened until analyzing. Filters and columns were stored at −20°C pending analysis.
Atmospheric samples were collected every 3 days using an active air sampler equipped with a pump, adsorption columns, and GF/F filters (0.7-μm pore size, 150-mm diameter) from the bottom up. The air adsorption columns are similar with those used for seawater sampling except filled only with XAD-2 material. A GF/F filter was deployed at the top of the column to collect particles and aerosols adsorbed by the column. Both columns and GFF filters were dried at 45°C for 96 hr and sealed in aluminum bags before stored at −20°C as for seawater samples.

Laboratory Sample Pretreatment
All samples were pretreated using the high-volume SPE ISIU technique at the State Oceanic Administration Key Laboratory for Polar Sciences (Shanghai, China). Each column was placed in a stainless-steel tube,~45ml dichloromethane was added in the column, and the column was ultrasonically vibrated for 5 min (90%, break rate 2s/5s) to ensure desorption of CUPs. The dichloromethane would flow out of the column at the bottom into the outside tube, and the level was maintained by further additions to protect the sonicator from being exposed in the air during this treatment. Triplicate extractions were performed, and the eluent (~150 ml) was collected.
The dichloromethane eluent was transferred to a 200-ml evaporation tube (TurboVap II, Biotage, Sweden) and spiked with an internal standard of tetrachloro-m-xylene (TCMX; 100 μl at 2.0 ng/μl). The volume was reduced to 1 ml under a gentle nitrogen stream, followed by the addition of 10-ml n-hexane (to displace dichloromethane), and again concentrated to 1 ml.
Poly-series molecular-imprinted-polymer-polycyclic-aromatic-hydrocarbon solid-phase-extraction cartridges (CNWBOND; ANPEL Laboratory Technologies, Shanghai, China) were used to purify the eluent. The cartridges were preconditioned with 5-ml dichloromethane and 5-ml n-hexane separately at a flow rate of 3 drops/s, before the 1-ml concentrated eluent was added, followed by cleanup with 2-ml n-hexane (in three times). A volume of 3-ml n-hexane was used to wash cartridges twice, then 10-ml dichloromethane was added and the eluent collected in a precleaned evaporation tube (Biotage, Sweden), replaced by 10-ml n-hexane before concentration to 1 ml, again under a gentle nitrogen stream. Finally, 1 ml of concentrated and purified eluent was collected and stored in a sample bottle at −4°C.

GC-MS Analysis
Concentrations of the eight CUPs were determined using a gas chromatography-mass spectrometry (GC-MS; Agilent 6890/5973, CA, USA) system equipped with a DB-5MS capillary column (0.25-mm internal diameter, 30-m length, and 0.25-μm film thickness) and an electron ionization dector. Helium (1 ml/min) was used as the carrier gas, and the GC oven was temperature-programed from 80 to 290°C: held at 80°C for 1 min, increased to 200°C at 12°C/min, increased to 220°C at 1°C/min, increased to 290°C at 15°C/min, and finally held at 290°C for 5 min. CUPs were quantified in the selected-ion mode. GC retention time (min), quantitation ions (m/z) are given in Table S4.

Quality Assurance and Quality control
GF/F filters, glassware, columns, and ultrasonic tubes were baked at 450°C for 4 hr before use. For every 10 samples, 20-L Milli-Q water generated onboard was collected as a field blank to indicate potential contamination during handling. One laboratory blank was analyzed for every 10 samples to check possible background contamination in the clean laboratory. The field and laboratory blanks were 2 to 3 orders of magnitude lower than field samples, which indicated the CUPs background in the ship and laboratory (Table S3). Method detection limits (MDLs) were calculated as the mean plus three times the standard deviation of field blanks (Table S4). TCMX recoveries were 79.8-108.9% (Table S5).
where f w and f a are the fugacities in seawater and atmosphere, respectively; C w is seawater concentration (ng/m 3 ); H w is Henry's law constant (Pa m 3 /mol); C a is atmospheric concentration (ng/m 3 ); R is the gas constant (8.314 Pa m 3 · K -1 · mol -1 ); and T is atmospheric temperature (K). Considering the uncertainties in T, C w , and C a , FR values of 0.3-3 are assumed to be in balance (Bruhn et al., 2003;Lohmann et al., 2009). FR > 3 indicates volatilization, and FR < 0.3 indicates deposition.
R software (version 3.5.1) was used to determine cluster analysis. CUP concentrations were first transferred to proportions at each sample before analysis, then performed using the statement of the cluster in R library. Backward trajectories were performed using HYSPLIT 4 software with dataset downloaded from the National Oceanic and Atmospheric Administration website (ftp://arlftp.arlhq.noaa.gov/pub/archives/reanalysis/). Three elevations, 50, 200, and 500 m, were selected to determine movements of atmospheric masses in a 3-day backward time span with an interval of 12 hr.

CUP Concentrations in the North Pacific Ocean
CUP concentrations in surface seawater of the North Pacific Ocean are shown in Figures 1a-1c and Table S6. Chloroneb was the most abundant at 0.01-89.2 ng/L (mean of 11.4±13.5 ng/L), comparable with the concentration measured in the Somme River, northern France (Net et al., 2015). There have been few studies of chloroneb in seawater (Añasco et al., 2010), and this may be the first report of its large-scale sampling. Alachlor concentrations were in the range 0.01-18.0 ng/L (mean 2.95±4.15 ng/L), much higher than those recorded in other studies (Schottler et al., 1994;Thurman et al., 1996), but lower than that measured in groundwater in a cornfield in the Connecticut River valley, USA (Potter & Carpenter, 1995). Atrazine was the third most abundant CUP, with a mean concentration of 0.73±0.96 ng/L and a range of <MDL-6.35 ng/L, lower than concentrations in the Aegean Sea but higher than those in the Baltic and Northern Adriatic Seas (Nödler et al., 2014). Chlorobenzilate concentrations were in the range <MDL-0.99 ng/L (mean 0.21±0.22 ng/L). ∑Permethrin (including cis and trans isomers) concentrations were in the range <MDL-0.93 ng/L (mean 0.15±0.17 ng/L), with cis/trans ratios of 0.67-3.67. Cis-isomer was abundant in most samples because of being more stable toward chemical and biological degradation (Sharom & Solomon, 1981). Dacthal was below MDL in most samples, and, where it was detected, the mean concentration was 0.02±0.01 ng/L. Dacthal concentrations in the North Pacific Ocean was comparable with or slightly higher than those in melt-ponds and seawater in the Canadian Arctic (Pućko et al., 2016) where the former play a key role in concentrating CUPs. Except for the low frequency of detection, dacthal is consistent with the range of concentrations in ice-core samples from the Devon Island Ice Cap, Nunavut, Canada (Zhang et al., 2013). Simazine and methoxychlor concentrations were in the ranges <MDL-0.76 ng/L (mean 0.13±0.14 ng/L) and <MDL-0.54 ng/L (mean 0.13±0.13 ng/L), respectively. Methoxychlor is an alternative for p,p′-DDT and had been increasingly used from the 1970s (Oh, 2009) until being banned in Europe (2002) and the United States (2003) (Neves et al., 2018). However, methoxychlor can still be detected in some provinces of China, southeastern Asia, and South Africa (Luo et al., 2016;Neves et al., 2018;Zeng et al., 2018). Methoxychlor has been well studied in snow, ice core, air, freshwater, and biota bodies but rarely in seawater (Hoferkamp et al., 2010). The seawater concentrations of methoxychlor measured in the present study were dramatically lower than those measured in ice-core samples in the Arctic between the 1950s and 1990s (Hermanson et al., 2005) when there was great demand for CUPs in North America, Europe, and Russia. The higher concentrations in ice cores also indicate that the predominant pathway of methoxychlor to high latitudes is LRAT. Journal of Geophysical Research: Atmospheres the Chukchi Sea, even though it has been widely reported in high altitude Arctic environments where LRAT had a significant influence (Bidleman et al., 2015;Hermanson et al., 2005;Hoferkamp et al., 2010;Morris et al., 2016;Pućko et al., 2016;Ruggirello et al., 2010;Vorkamp & Rigét, 2014;Yao et al., 2007;Zhang et al., 2013;Zhong, Xie, Cai, et al., 2012). The distribution in the present study indicates a short half-life and rapid degradation of dacthal in the ocean, and varying degrees of LRAT in different environmental conditions of the Arctic (Gouin, Cousins, et al., 2004;van Pul et al., 1998). The remaining five CUPs had comparable concentrations, all lower than those measured in the North Pacific Ocean. Simazine concentrations ranged from <MDL-0.34 ng/L (mean 0.11±0.10 ng/L), chlorobenzilate recorded 0.01-0.61 ng/L (mean 0.18±0.16 ng/L), and atrazine recorded 0.01-0.99 ng/L (mean of 0.26±0.25 ng/L). Chernyak et al. (1996) reported atrazine concentrations in the Chukchi Sea of <9.5 pg/L but with a concentration of 0.4 ng/L in sea ice, comparable with the results of the present study. Methoxychlor concentrations were in the range <MDL-0.38 ng/L (mean 0.15±0.11 ng/L), significantly higher than levels in the Kara Sea (Carroll et al., 2008). ∑Permethrin concentrations were in the range <MDL-0.78 ng/L (mean 0.13±0.16 ng/L) and were higher in early summer in the North Pacific Ocean than in late summer. Permethrin cis/trans isomer ratios ranged from 0.5 to 3, basically higher than the technical ratio of 40:60 or 25:75 for agricultural use (IARC, 1991). The ratio could vary with time and distance due to different rates of hydrolysis and photolysis of the two isomers (Hayes & Laws, 1991). It is reported that oxidation is more extensive for cis-permethrin, while trans-permethrin is more subject to hydrolysis (Shono et al., 1979). Additionally, cis-permethrin is found to be more persistent than trans isomer in aquatic systems (Jordan & Kaufman, 1986;Sharom & Solomon, 1981). Cis/trans ratios in the present study suggest a long existing time of permethrin in most of the study area. However, ratios close to the agricultural use were recorded in open-sea areas in the north of the Chukchi shelf indicating a fresh input from nearby sources.

CUP Concentrations in the Atmosphere
Atmospheric ∑ 8 CUPs concentrations had a range of 0.68-17 ng/m 3 (mean 3.31±4.36 ng/m 3 ; Figure 1d and Table S7). Chloroneb was the most abundant with concentrations of 0.59-15.5 ng/m 3 (mean 2.99±4.02 ng/m 3 ). There are few previous studies of atmospheric chloroneb (Haraguchi et al., 1994;Schummer et al., 2010), and the present study is the first of chloroneb in the Arctic atmosphere. Two herbicides (alachlor and atrazine) concentrations were in the ranges of 0.01-0.28 ng/m 3 (mean 0.06±0.07 ng/m 3 ) and <MDL-0.06 ng/m 3 (mean 0.02±0.02 ng/m 3 ), respectively. Neither of these CUPs has been detected in the atmosphere at high latitudes in the Canadian prairies (Messing et al., 2014). Simazine and dacthal concentrations were in the ranges of <MDL-0.19 ng/m 3 (mean 0.04±0.06 ng/m 3 ) and <MDL, respectively. The low concentrations of simazine may be due to rapid degradation in the atmosphere (Aulagnier et al., 2008) and relatively low degrees of volatilization from soil (<1%; Glotfelty et al., 1989). Dacthal is reported to be transported mainly in the atmosphere (Vorkamp & Rigét, 2014), but it was not detected in air in the present study, although Zhong, Xie, Cai, et al. (2012) reported a mean concentration of 0.14±0.3 ng/m 3 along the same transect in 2010. One possible explanation is that dacthal is not widely used in Asia, considering its higher levels in  (Li, Ma, et al., 2014). The trans-permethrin isomer had higher concentrations than the cis isomer in the ambient Asian atmosphere with cis/trans ratios close to 40:60, but the two were equal in the rest samples. The vapor pressure of cis-permethrin is almost 3 times higher than that of trans-permethrin [https://pubchem.ncbi.nlm.nih.gov/], suggesting easier volatilization for cis isomer. However, the relatively low concentrations in the atmosphere probably indicate a faster photolysis reaction of cis-permethrin (Hayes & Laws, 1991;Shono et al., 1979), especially at low latitudes where solar radiation is stronger than that in polar regions.

CUP Compositions, Distributions, and Sources in Seawater and Atmosphere
The three most abundant CUPs in the North Pacific Ocean were chloroneb, alachlor, and atrazine, together making up >96% of the total (Figures 2a and 2b). Chloroneb made the highest contribution to most samples except at stations W53 and W54 where alachlor was the most abundant. The highest concentrations of chloroneb in seawater were measured at the beginning of sampling off the southern coast of Korea, before decreasing rapidly in the Japan Sea, and varied only slightly from the Soya-Kaikyo strait to the Bering Strait. This trend indicates that China and/or Korea could be the major sources of chloroneb. Low concentrations at high latitudes, far from source regions, indicate dilution by surface water. In contrast to chloroneb, alachlor concentrations were lower in the East China Sea but increased sharply in the Bering Sea, likely due to sources in Eastern Russia. Atrazine had a similar distribution to that of chloroneb with higher concentrations at lower latitudes, and they were almost unchanged from the eastern the Japan Sea to the Bering Sea. The latter uniform trend indicates no other sources and a stable state of atrazine, as supported by Tierney et al. (1999), who reported that atrazine has a long half-life in cold, nutrient-limited regions. The remaining five CUPs made relatively minor contributions with small changes between two sampling seasons.
Cluster analysis shown in Figure 3a provides a clear view of the different sources of CUPs, with samples being divided into three distinct groups: Groups I, II, and III. Group I is located mainly in the Japan Sea and is largely influenced by agricultural activities in surrounding countries as China, Eastern Russia, Korea, and Japan. Samples near the Bering Strait (stations W13 and W51) are included in this group, indicating transport by currents.

Journal of Geophysical Research: Atmospheres
Kamchatka Peninsula appear more similar to Group I than to Group II. A possible explanation is that these samples were subject to the influence of a small region, which did not impact other sampling areas.
Group III includes samples from five stations (W10, W11, W53, W54, and W65) with four in the Bering Basin, which is strongly impacted by the southern warm current through Near Strait. The compositions of samples from four of these stations (W10, W11, W53, and W54) are significantly different from those of Group I and II samples, with high proportions of chloroneb and alachlor (>90%) in the Bering Basin. Sample W65 is located at distance and contains only chloroneb and alachlor similar to Group III.
In the Chukchi Sea, chloroneb was again the dominant CUP, with a higher content (>79%) than in samples from the North Pacific Ocean (Figures 2a and 2b). Alachlor was the second most abundant but contributed only 10.7%. Dacthal was not found in the Chukchi Sea. The remaining five CUPs had low percentages that did not vary substantially. CUPs were abundant in the Chukchi Shelf region but lacked distinct distribution groupings. High concentrations were likely correlated with the melting of sea-ice and melting-ponds, which release significant quantities of CUPs to seawater (Ma et al., 2011). Cluster analysis (Figure 3b) again indicates three compositionally distinct groups, Group IV, V, and VI, although differences are small compared with those between Groups I-III. Group IV includes samples from three stations (W14, W48, and W49) near Bering Strait, which are affected by incoming flows from the North Pacific Ocean and have compositions resembling to those in the Bering Sea. Group V is mainly located in the Chukchi shelf region. Station W25 is an exception as it was located further north in the open sea. Group VI samples were from the broad northern area including the Beaufort Sea, Mendeleev Ridge, and even further north, with the impact of any landmass being small because of the distances involved. Group VI is, however, more susceptible to the effects of sea ice and melting ponds than Group IV and V.
Atmospheric CUP compositions differed from those in seawater (Figure 2d). Chloroneb was again dominant with a higher proportion, reaching a maximum in the Nordic Seas area (Norwegian, Greenland, and Iceland Seas) and decreasing in the central and western Arctic Ocean areas. Methoxychlor and chlorobenzilate had higher concentrations in the atmosphere than in seawater, with the former contributing up to 8.8% over the Bering Sea. However, the two CUPs abundant in seawater, alachlor and atrazine, contributed relatively little to atmospheric samples. Atmospheric CUP concentrations were higher in the coastal regions of Asia, Samples A01 and A13 were both influenced by atmospheric masses from the North Pacific Ocean, but their CUP concentrations differed markedly. This may reflect seasonal changes in the atmosphere during the twomonth sampling interval, with the atmospheric mass moving from the North Pacific Ocean having a major influence on the Bering and Chukchi Seas (samples A02, A10, and A13). Low atmospheric CUP concentrations in the central Arctic Ocean (samples A03 to A07) were influenced mainly by atmospheric masses from the Canadian Arctic. In the North Atlantic Ocean, the atmospheric mass from the Baffin Bay influences the southern area of Iceland (samples A08 and A09), and CUP concentrations there were comparable with those in the central Arctic Ocean. Off the coasts of Alaska, samples A11 and A12 were both impacted by the atmospheric mass originating from the Canadian Basin. FRs (Table S8)  W03′/A11 values >3 indicate strong volatilization, but atmospheric chloeoneb concentration of sample A11 is low compared to the air samples of the other two pairs. A possible explanation is that air-seawater exchange proceeds only slowly. Salinity at W03′ was 28.2, suggesting the significant influence of riverine input or sea-ice melt (Tong et al., 2014) leading to high concentrations of CUPs in seawater, although this does not increase atmospheric concentrations. Furthermore, stations W11′/A14 and W14′/A15 are located in the open sea at relatively low latitudes where such influences are negligible.

Seasonal Changes of CUPs From the North Pacific Ocean to the Arctic
The eight CUPs studied exhibited no obvious seasonal variation in the Chukchi Sea between the two sampling segments over a short interval, but there were significant differences in the North Pacific Ocean. Six CUPs-chloroneb, methoxychlor, ∑permethrin, dacthal, chlorobenzilate, and simazine-had different seasonal peaks in the North Pacific Ocean. In early summer, the peaks occurred at lower latitudes and had decreasing trends with latitudes, consistent with the findings of Zhong, Xie, Cai, et al. (2012) indicating the influence of agricultural activities in eastern Asia during spring. Coastal provinces of China use pesticides heavily in agriculture and have giant industrial production (Tan et al., 2009;Yang et al., 2010;Zheng et al., 2011). Seasonal variations in CUP concentrations have also been documented in relation to agriculture in the North Sea area (Mai et al., 2013). After entry to the sea, CUPs are carried by ocean currents and are degraded by ultraviolet radiation (Beltrán et al., 1993;Ikehata & El-Din, 2006;Konstantinou et al., 2001) and microbial action during the northward transport. The UV radiation could accelerate the rates of CUP degradation in the ocean, although the influences are not equal (Curran et al., 1992;Samanidou et al., 1988). Reductions in agricultural activity also cause a decline in pesticide levels, and CUP concentrations thus decrease significantly in late summer. This is documented by the difference between stations W08 and W57, which were sampled two months apart. The ∑ 8 CUP concentration at W08 (early summer) was almost four times than that at W57 (late summer). In early summer, the frequently agricultural or human activities utilized giant quantities of CUPs for pest and weed control, which had a significant contribution to this high concentration. When it came to late summer, the mass of utilization decreased as a result of few demands on farms. Therefore, ∑ 8 CUP concentration in the ambient sea dropped dramatically. Seasonal changes of dacthal were not as obvious because the concentrations were at or below the MDL. Dacthal is restrictively registered in only a few countries, including the UK, United States, Canada, New Zealand, and Australia (Ruggirello et al., 2010). This CUP may be carried by air masses driven by the dry monsoon in winter and deposit in the ocean. Regardless of its source, the distributions of dacthal in different sampling periods are consistent with the hypothesis of slow air-sea exchange.  Journal of Geophysical Research: Atmospheres concentrations in the East China Sea in late summer may thus be due to input from a remote reservoir acting as a secondary source. The distribution of alachlor was different from that of atrazine, with its concentration not decreasing from early to late summer. Instead, alachlor concentration increased with latitude and peaked in the Bering Sea adjacent to the Sakhalin Peninsula in both sampling periods, before sharply decreasing near the Bering Strait. Alachlor and atrazine are alternative weed-control agents, and their disparate distributions suggest different sources and quantities used. Alachlor may be used throughout the summer in eastern Asia, considering a half-life of 2 weeks (Wauchope et al., 1992), and the Sakhalin Peninsula may be the primary contaminant source.

Pathways of CUPs Transported by Oceanic Currents
Pathways and fluxes of eight CUPs studied are shown in Figure S4 and Table S9, for early and late summer, respectively. In early summer, CUPs are exported into the Japan Sea by Tsushima Warm Current through the Tsushima Strait (Takikawa et al., 2005) and flow out through Soya Strait and Tsugaru Strait, respectively (Fukamachi et al., 2010;Ito et al., 2003). Morimoto and Yanagi (2001) reported that a subpolar gyre strengthens in the Japan Sea in summer, possibly reducing the volume transported through the Soya and Tsugaru Straits and resulting in a significant fraction of the CUPs remaining in the Japan Sea. In winter when the subpolar gyre weakens, the remaining CUPs would be transported out of the Japan Sea to higher latitudes. In eastern Hokkaido, currents flowing out of the Japan Sea separate into two branches (Yoon & Kim, 2009). The northern Soya Warm Current enters the Okhotsk Sea and joins the cyclonic gyre circulation (Andreev & Shevchenko, 2008), while surface water undergoes exchange with the Kamchatka Current through Kuril Strait (Shevchenko et al., 2009). The southern Tsugaru Warm Current turns south with the Oyashio Current and east at the subarctic front (Conlon, 1982;Tomczak & Godfrey, 2003). At the frontal zone, a portion of surface seawater may enter the Bering Sea through the shallow and narrow channels of the Aleutian Islands (Kinney & Maslowski, 2012;Prants et al., 2013;Tabata, 1975). This portion moves to the north, joins the northern Pacific cyclonic gyre (Danielson et al., 2014), and finally turns south at the coast of Kamchatka Peninsula (Panteleev et al., 2012). The Bering Sea shelf is influenced by a combination of the Anadyr Current, Bering Shelf Water, and the Alaska Coastal Current (Coachman & Aagaard, 1966;Woodgate & Aagaard, 2005), which have both terrestrial and seawater signatures (Danielson et al., 2016). In the north, seawater flows into the Arctic through the shallow and narrow Bering Strait which restricts the volume transport of seawater, and, as a result, CUPs flux through the Bering Strait may be limited even though ∑ 8 CUP concentrations are relatively high.

Conclusion
The aim of the present study was to depict distributions of CUPs with seasonal variations in a geographically large scale and explain the decoupling between CUP concentrations and air-sea exchange. The analyzed CUPs exhibit seasonal changes from the North Pacific Ocean to the Arctic Ocean. In early summer, peak concentrations in seawater occur mainly in coastal areas of eastern Asia, indicating a significant influence of agricultural practices there. ∑ 8 CUPs then decrease with distance as a result of dilution and decreasing input by runoff. CUP concentrations decrease significantly in late summer. Relatively abundant CUPs at high latitudes indicate the far impact of human activities. The results of the study provide support for the slow exchange of CUPs between seawater and atmosphere.

Acknowledgments
We wish to express our sincere gratitude to all the members of the 7th and 8th Chinese National Arctic Research Expedition for the access to R/V Xuelong (in English) and for their great support during the in situ sampling collection. This work was funded by the National Natural Science Foundation of China (41776202). All data, including sampling information, samples treatment, and analyzing results, are accessible in the supporting information attached to this article.