Organophosphate ester pollution in the oceans

The large-scale use of organophosphate esters (OPEs) as flame retardants and plasticizers has led to their prevalence in the environment, with still unknown broader impacts. This Review describes the transport and occurrence of OPEs in marine systems and summarizes emerging evidence of their biogeochemical and ecosystem impacts. Long-range environmental transport via the atmosphere and ocean currents distributes OPEs from industrialized regions to the open ocean. OPEs are most prevalent in coastal regions, but notable concentrations are also found in the Arctic and regions far from shore. Air–water interactions are important for the transport of OPEs to remote oceans and polar regions. Processes such as degradation and sinking of particle-bound compounds modulate the properties and fate of OPEs in the water column, where they are potentially a non-accounted source of anthropogenic organic phosphorus for microbial communities. Some OPEs have toxic effects in marine species and are found in measurable quantities in fish and other aquatic organisms. However, there is conflicting evidence on the potential for bioaccumulation and biomagnification of OPEs. Future work must constrain the large-scale impact of OPEs on marine biota and biogeochemistry to support more effective regulation and mitigation. The use of organophosphate esters as flame retardants and plasticizers has increased, leading to their environmental pervasiveness. This Review describes the transport and distribution of these organic pollutants in the ocean and highlights the potential impacts on biogeochemical cycling and ecosystem health. Higher concentrations of organophosphate esters (OPEs) occur in coastal seas near populated and industrial areas than in the open ocean. OPEs are transported to the ocean from terrestrial sources via both atmospheric transport and riverine discharge. Air–water exchange and atmospheric deposition affect the cycling of OPEs. Transport via ocean currents and, potentially, biodegradation control the vertical distribution and sedimentation of OPEs. Re-emission from melting snow and ice in polar regions can impact OPE levels in the High Arctic and the Southern Ocean water columns. OPEs and their transformation products are emerging concerns for marine ecosystems, especially related to their presence in marine mammals and fish. International strategies are needed to manage their environmental emissions. Higher concentrations of organophosphate esters (OPEs) occur in coastal seas near populated and industrial areas than in the open ocean. OPEs are transported to the ocean from terrestrial sources via both atmospheric transport and riverine discharge. Air–water exchange and atmospheric deposition affect the cycling of OPEs. Transport via ocean currents and, potentially, biodegradation control the vertical distribution and sedimentation of OPEs. Re-emission from melting snow and ice in polar regions can impact OPE levels in the High Arctic and the Southern Ocean water columns. OPEs and their transformation products are emerging concerns for marine ecosystems, especially related to their presence in marine mammals and fish. International strategies are needed to manage their environmental emissions.

Organophosphate esters (OPEs) are synthetic organic chemicals used as flame retardants, plasticizers and additives in industry, electronics, household consumer products and personal care products 1 ( Table 1;  Supplementary Table 1). After effective regulation of polybrominated diphenyl ether (PBDE, which were widely used as flame retardants) in 2003, the use of OPEs dramatically increased -global use of OPEs increased from 300 kilotons in 2004 to 620 kilotons in 2013 (refs 2,3 ), accounting for 30% of the global total of flame retardants 2 .
OPEs are highly mobile in the environment, and their broad application and use in plastic products have contributed to their diffusive release through volatilization, leaching and abrasion 4,5 . More than 44.7 million tonnes of electronic waste (e-waste) were generated worldwide in 2016, which would contain non-PBDE brominated flame retardants and OPEs [6][7][8] . Much of this waste was treated in coastal cities, which caused OPEs to be released to the ocean through inadequate disposal procedures or wastewater. Open-air burning of e-waste during recycling potentially releases organic additives, including OPEs, in ambient air 9 . For example, OPE concentrations reached 740-1,000 ng m −3 in ambient air in Canadian e-waste recycling facilities 5 and 3.8-57.7 ng m −3 in a rural e-waste recycling area in South China 10 .
This widespread usage of OPEs, their release to the environment and the potential toxic risk to human beings and ecosystems have spurred research into the broader impacts of OPE usage. As a result, there is now compelling evidence of the widespread occurrence of OPEs in wastewater [11][12][13] , inland surface water 14 and groundwater 15 from Europe, North America and Asia. Concerningly, OPE concentrations are generally two to three orders of magnitude higher than those of brominated flame retardants and other legacy persistent organic pollutants (POPs) in environmental matrices 16,17 . Model predictions have shown that OPEs are persistent and mobile in water 18,19 , implying that riverine run-off plays an important role for the transportation of OPEs from terrestrial sources to the ocean. Despite this prevalence in the environment, there is no international regulation existing to tackle increasing emissions of OPEs, partially related to their persistence, bioaccumulation and potential toxicity in the marine environment.
In this Review, we examine the occurrence and impact of OPEs in marine environments. Major OPE sources and their transport are described, as are environmental concentrations and spatial trends of OPEs Organophosphate ester pollution in the oceans Long-range atmospheric transport. Long-range atmospheric transport (LRAT) 44 is an important pathway for global distribution of POPs 45,46 and other persistent environmental contaminants, but early predictions of atmospheric OPE half-lives were generally below the threshold (2 days) to meet the LRAT criterion of the Stockholm Convention on Persistent Organic Pollutants. For example, the very short half-life proposed by the EU risk assessment in 2008 for TCIPP (8.6 h) led to the incorrect conclusion that TCIPP will not undergo LRAT 47 . Therefore, OPEs were initially considered degradable enough to have a low potential for LRAT 48 . However, the persistence and LRAT potential of OPEs in the gaseous phase might have been underestimated 49 . Moreover, particle-bound OPEs are highly persistent in the atmosphere. For example, based on heterogeneous OH-initiated oxidation of OPEs in air, the approximate atmospheric lifetimes were estimated to be 5.6, 4.3 and 13 days for particle-bound TPhP, TEHP and TDCIPP 49 . This calculated phase lifetime suggested medium-range or long-range transport potential of particle-bound OPEs in the atmosphere.
Improved modelling estimations considering episodic transport, sorption to the particle phase, impact of water mass and the uncertainty of the environmental half-life times show that some of the most commonly used OPEs do travel long distances 19 . These estimates are consistent with field observations showing that OPEs

Key points
• Higher concentrations of organophosphate esters (oPes) occur in coastal seas near populated and industrial areas than in the open ocean. • oPes are transported to the ocean from terrestrial sources via both atmospheric transport and riverine discharge. air-water exchange and atmospheric deposition affect the cycling of oPes. • Transport via ocean currents and, potentially, biodegradation control the vertical distribution and sedimentation of oPes. • Re-emission from melting snow and ice in polar regions can impact oPe levels in the High arctic and the Southern ocean water columns. • oPes and their transformation products are emerging concerns for marine ecosystems, especially related to their presence in marine mammals and fish. • International strategies are needed to manage their environmental emissions. are ubiquitous in the atmosphere globally 16,17,[50][51][52][53][54] and their presence in remote oceanic and polar environments, indicating that they do indeed undergo LRAT 17,53,[55][56][57] . Most observations in the remote areas were associated with chlorinated OPEs (such as TCEP and TCIPP) 54,56 , which implies greater LRAT potential for specific compounds. This occurrence highlights that, as OPEs show a wide range of physical and chemical properties, their atmospheric transport varies between chemicals due to differences in persistence, particle sorption and air-water partitioning.

Air-water exchange and atmospheric dry and wet deposition.
During transport from source regions to remote marine environments, OPEs are subject to exchange at the interfaces between different environmental media 58,59 ( fig. 1). Atmospheric depositional processes play an important role in the environmental fate of OPEs, contribute to aquatic ecosystem burdens and lead to OPE accumulation in marine food webs 54,60 . Several major processes cause OPE atmosphere-ocean interaction, including diffusive air-water exchange between the gaseous and dissolved phases, atmospheric dry deposition of particle-bound OPEs and wet deposition by rain and snow 54,56,57,61 . Dry deposition (F DD ), wet deposition (F WD ) and air-water exchange (F AW ) can be estimated as, where C A , C G , C W and C rain are the chemical's concentrations in particles, gas phase, water (dissolved phase) and rain, respectively. H′ is the dimensionless Henry's law constant, v D is the deposition velocity of the particles, p is the rain precipitation and k AW is the air-water mass transfer coefficient 62 . The magnitude of air-water exchange direction and flux depends non-linearly on wind speed, with enhanced fluxes at high wind speeds 59,63 . For the OPEs with relatively short atmospheric half-lives, volatilization from seawater to air could control the diffusion fluxes. By contrast, dissolved water phase concentrations of hydrophobic OPEs such as TEHP and EHDPP can be lowered by partitioning to particulate matter in the water column, where it can sink and remove OPEs from the surface ocean. Photodegradation and biodegradation can also deplete dissolved phase OPEs, thus, favouring deposition of OPEs in the ocean.
In the North Pacific Ocean and the High Arctic Ocean, air-water exchange fluxes of OPEs range from −0.79 to 0.59 ng m −2 day −1 , with TiBP contributing the most to seawater to air volatilization (0.19 to 0.72 ng m −2 day −1 ) 57 , and TCIPP and TCEP exhibiting net deposition. In the Fram Strait, OPEs were found to be at dynamic equilibrium 61 . In the North Atlantic and European Arctic, net volatilization fluxes of 5 to 1,100 ng m −2 day −1 , 61 to 12,000 ng m −2 day −1 , 12 to 2,000 ng m −2 day −1 and  56 . This range illustrates the uncertainty in the air-water exchange flux of OPEs, owing to poorly constrained H values (Eq. 3). Dry deposition of particle-bound OPEs contributes substantially to total OPE flux to the ocean, including 67 ± 17% of the total OPEs in the European Arctic 54 , 52 ± 23% in the Bohai and Yellow seas 65 and 71-93% in the North Pacific to the Arctic. In absolute values, particle-bound OPEs deposition fluxes of 14 to 94 ng m −2 day −1 have been measured in the northeast Pacific and the Arctic 57 . In the tropical and subtropical Atlantic, Pacific and Indian oceans, the dry deposition fluxes of particle-bound OPEs range from 4 to 140 ng m −2 day −1 , with higher deposition fluxes in the North Pacific and Indian oceans 56 . Overall, the surface waters of the tropical and subtropical oceans receive an estimated yearly integrated amount of ~2 to 13 kilotonnes year −1 of ∑ 14 OPEs from the dry deposition of particle-bound OPEs 56 . Fluxes are somewhat lower in the North Atlantic and European Arctic 54 , where the deposition flux of OPEs was about 2-16 ng m −2 day −1 , which is similar to those estimated for the South China Sea 66,67 . High deposition has been observed in the open Mediterranean (70~880 ng m −2 day −1 ) and Black seas (300~1,100 ng m −2 day −1 ) 52 , the North African coastal Mediterranean (18~180 ng m −2 day −1 ) 31 and the Bohai and Yellow Seas (21-250 ng m −2 day −1 ) 65 . These higher fluxes generally occurred near heavily populated or industrialized areas. In most studies, TCIPP and TCEP dominated the total deposition flux in the oceans because of their high consumption and persistence in the environment.
Wet deposition of OPEs by rain and snow can be quantitatively important in some climatic regions and seasons, as the high water solubility of some OPEs favours wet deposition fluxes 68 . Moreover, snow and rain amplify the concentrations of organic pollutants in the receiving waters 69,70 owing to the high specific surface area of snowflakes and raindrops, which increases the rain-air washout ratios (generally close to 10 5 for OPEs) 69 . Although measurements of OPEs in rainwater are only available for samples collected on land 68,69,71 , the high concentrations of OPEs recorded indicate that wet deposition cannot be ignored, especially for the coastal seas. Snow deposition is an effective scavenger of atmospheric OPEs in the Arctic, the Southern Ocean and the Antarctic 54,72,73 , so snow melting likely represents a flash of OPEs from coastal land to coastal waters, as found with other pollutants [74][75][76] . Anthropogenic warming could increase the rain periods and decrease the snow deposition periods, which would mean that, for some regions, such as the western Antarctica Peninsula, the deposition by rain could become more critical during the coming decades 69 .

Re-emission from melting ice and snow.
As sea ice retreats and snow melts with rising temperatures, chemical contaminants trapped in snow, including OPEs, could be directly discharged into the water column, influencing the relative abundance of the chemical components in coastal seawater and amplifying their seawater-air fugacity gradient 30,77 . For example, the fresh input from melting ice and snow enhanced the OPE concentrations in seawater along the coast of East Greenland, which are 2-5 times higher than those in the Fram Strait 54  www.nature.com/natrevearthenviron of 6.8-19 ng l −1 were measured in the High Arctic Lake Hazen, 5-10 times higher than in the North Atlantic Ocean and the northeast Pacific Ocean 54,57,78 . Elevated atmospheric concentrations of OPEs have been measured along the Antarctic coast, which can be attributed to the re-emission from snow and ice melt 17 .
Ocean current transport. The importance of aqueous transport for OPEs depends on their deposition and subsequent transfer, persistence and mobility 18 . In the marine environment, coastal seas receive large inputs of water-bound OPEs, which leads to higher levels of OPEs in coastal waters compared with those in the open seas 79,80 . Once in the ocean, OPEs can be subject to long-range transport via oceanic circulation. For example, in the tropical North Atlantic, dissolved OPEs (1,300 ng l −1 ) from the Amazon River were transported more than 3,000 km via the North Brazil Current and its retroflection 26 . Chlorinated OPEs, in particular, can be efficiently transported via ocean currents due to their persistence, lower volatility and high solubility 18,19,43 . The higher mobility of the Cl-OPEs versus non-Cl-OPEs is caused by the low degradation rate of Cl-OPE in water, as demonstrated in lakes 81,82 and seawater 26,43,54,83 . Moreover, as OPEs are widely used in plastic and floating debris can transport between continents via ocean circulations 84 , it is speculated that ocean gyres can bring OPEs into the open waters, such as in the Southern Ocean 53 .
Environmental degradation. The environmental degradation of OPEs is an important process, as it determines the persistence of these chemicals in the environment, which, in turn, is a key aspect when assessing the risk of anthropogenic chemicals. Generally, the environmental degradation of organic compounds is efficient in the atmosphere due to the occurrence of OH radicals, among others. However, reaction rates for the OH heterogeneous oxidation of several other OPEs, such as TPhP, TBEP, TEHP and TDCIPP, suggest atmospheric half-lives ranging from a few days to weeks 49 . These relatively long residence times in the atmosphere are enough to explain their potential for LRAT and their occurrence in the oceanic atmosphere. Photodegradation of OPEs in water has been observed in laboratory-based studies 82,85 , which indicate that degradation is more effective for non-chlorinated than chlorinated OPEs. There is in situ evidence of photodegradation of some OPEs in seawater 82 , based on observations of organophosphate diesters -the degradation products of OPEs flame retardants and plasticizers. These diesters have also been detected in fish from marine environments 86 and in polar bears 87 , top predators in the Arctic marine food web. However, the in situ production of diesters in seawater is, so far, unproven. As some OPE diesters are industrially produced 88 , and organophosphate diesters have been found in rivers and lakes 89,90 , it is unclear how much of the OPE diesters are from riverine inputs versus in situ OPE degradation 91 .
OPEs are readily susceptible to biodegradation in rivers by naturally occurring microbial populations and can be biodegraded by activated sludge from domestic sewage treatment plants 92 . Like other organophosphorus triesters, OPEs biodegradation involves the hydrolysis of the phosphorester bonds mediated by phosphotriesterases, then phosphodiesterases and then phosphomonoesterases. The only phosphotriesterase identified so far that mediates TCEP and TDCIPP biodegradation is a haloalkylphosphorus hydrolase (HAD) that differs from the common three families of phosphotriesterases used to degrade aryl dialkyl phosphates, such as parathion and paraoxon [93][94][95][96] .
The biodegradation of OPEs and characterization of the associated microbial communities in the marine environment is poorly studied. Only a few OPEdegrading strains have been cultured to date, mainly from soil habitats. Isolated strains able to degrade and use TCEP and TDCIPP as the sole source of P belong to Sphingobium and Sphingomonas species 97,98 , a Brevi bacillus brevis sp. is able to degrade TCrP 99,100 and the Rhodococcus and Sphingopyxis 101 and Roseobacter 102 are able to degrade TPhP and TCrP.
In phosphorus-limited natural seawater, the consumption of several OPEs has been observed, along with an increase in activity of Flavobacteria 103 . Given the widespread occurrence of Flavobacteria in the global oceans, OPE biodegradation could be a common feature in the upper ocean 104 . Currently, the half-lives of OPEs in marine waters remain unknown but could be dependent on the biogeochemical province. If OPEs are degraded more quickly in P-limited waters, then OPEs would potentially be more persistent when P is not a limiting nutrient. Indeed, phosphodiesterase activities are more abundant than expected in the water column 105 , and they can account for relevant P acquisition by marine bacteria under inorganic P limitation, supporting the link between atmospheric inputs of anthropogenic organic P and its utilization as a nutrient by marine microbiomes.

OPEs in marine environments
OPEs are ubiquitous in the marine environment 106 . This section overviews OPEs in the global ocean and polar regions and their geographic variation owing to regional discharge and long-range transport.  (fig. 2a). These levels were generally two orders of magnitude higher than those of brominated flame retardants in the oceanic atmosphere 107 , which are increasingly phased out.
OPEs are frequently found in the Arctic 108 , despite there often being no substantial local source. In the Canadian Arctic, the level of ∑ 13  The most abundant OPEs were TnBP and TCIPP in air, and chlorinated OPEs (TCEP, TCIPP and TDCIPP) accounted for 51% of total OPEs on average. This level was comparable with those found in Arctic air, suggesting OPE input into the Antarctic via LRAT.
Overall, elevated OPEs in the air are found in nearcoast regions ( fig. 2), especially near urban and industrial areas, followed by mid-latitude open oceans, the Arctic and, finally, the Southern Ocean. This ubiquity of OPEs in the global atmosphere contrasts with previous model predictions of limited LRAT. Such discrepancy is probably related to poor empirical knowledge of the physicochemical properties of OPEs.

OPEs in seawater.
OPEs are present in up to μg l −1 concentrations in wastewater treatment plant effluent 106,112,113 . Consequently, OPEs have been found in the surface waters of lakes and rivers with concentrations ranging from 10 to 1,000 ng l −1 (refs 15,18,25,114,115 ), which can contribute to coastal OPE loads. In a survey in coastal areas of seven European countries 116 , OPEs were detected in all samples, and the ∑ 7 OPEs ranged from 0.43 to 870 ng l −1 in transitional and coastal waters ( fig. 3a). High OPE levels were also found in the seawaters around the UK (∑ 7 OPEs, 280 ± 35 ng l −1 ) and Portugal (∑ 7 OPEs, 550 ± 440 ng l −1 ), correlating with the sampling sites that were closest to urban areas. Comparable levels (∑ 9 OPEs, 240 ± 330 ng l −1 ) have been found in the Bay of Marseille (northwestern Mediterranean Sea) 35 and 5 to 50 ng l −1 ∑ 18 OPEs in the German Bight 41 .
These results are generally consistent with those in the coastal seas of China, which has relatively high levels of OPEs (range 88-3,600 ng l −1 ) near coastal cities 80,115,117 (fig. 3a). Observations in the Pearl River Delta, South China Sea, Yellow River Estuary and Tokyo Bay reveal concentrations of ∑ 14 OPEs ranging from 15 to 1,800 ng l −1 , 1 to 150 ng l −1 , 250 to 1,700 ng l −1 and 110 to 280 ng l −1 , respectively 118 (fig. 3a). The concentrations of ∑ 7 OPEs ranged from 8 to 98 ng l −1 in the Bohai Sea and Yellow Sea 119 . Further offshore, in the western Pacific, the concentration of ∑ 10 OPEs ranged from 3.0 to 48 ng l −1 , with a mean of 25 ± 10 ng l −1 (ref. 83 ).
Concentrations in the ng l −1 range have been measured in high-latitude waters. In Arctic Ocean surface water, concentrations of 0.9 to 17 ng l −1 Σ 3 Cl-OPEs have been recorded 61,120 . Similarly, the concentrations of ∑ 8 OPEs in the North Atlantic and the Arctic ranged from 0.35 to 8.4 ng l −1 . The four highest concentrations were measured at sites near continents 54 , implying anthropogenic inputs into the ocean. In the Canadian Arctic, the mean concentrations of ΣCl-OPEs and Σnon-Cl-OPEs were 10 ng l −1 and 1.3 ng l −1 , respectively, in surface water 43 . OPEs observed in seawater from the northwestern Pacific to the Arctic (8.5 to 140 ng l −1 ) 57 are one order of magnitude higher than those from the European Arctic 54 (fig. 3b). There is no clear spatial trend for OPEs in seawater from China towards the Arctic (fig. 3a). However, a declining trend was noticed from the North Atlantic to the Arctic 54 ( fig. 3b), clearly showing oceanic transport from European seas to the Arctic. Interestingly, relatively high concentrations of ∑ 6 OPEs (<5.0-44 ng l −1 ) were found in seawater of Fildes Peninsula, Antarctica 121 , and 20-9,200 ng l −1 in freshwater from the northern Antarctic Peninsula 122 , which are attributed to local emissions from scientific research stations and tourist vessels in the Antarctic Peninsula.
OPEs in marine sediment. Influenced by the different degrees of human activities, average ΣOPEs concentrations ranging over two orders of magnitude (0.35-71 ng g −1 dry weight (d.w.)) are reported in marine sediments from straits, near-shore and offshore areas ( fig. 4a; Supplementary Table 2). Chlorinated OPEs (especially TCEP and TCIPP) were often the most abundant OPEs detected in ocean sediments ( fig. 4b), consistent with their extensive usage and their low degradation rates. Due to their relatively strong hydrophob icity, TEHP (logK ow 9.5) 36,123 and TCrP (logK ow 5.1) 124 have also been identified as dominant OPEs in sediments. By contrast, TnBP and TBOEP were the most abundant OPEs in sediments from Taiwan Strait, China 125 and Western Scheldt estuary, the Netherlands 126 , respectively, owing to the large usage of these compounds locally.
Marine sediments close to well-populated areas tend to have high concentrations of OPEs. Elevated levels of Σ 18 OPEs are found in sediments along the coast of Korea, with the maximum and average values of 350 and 71 ng g −1 d.w., respectively. The higher concentrations usually occur in harbours 127 . In the Bohai Sea, decreasing levels of ΣOPEs are reported in sediment with increasing distance from shore: Laizhou Bay (6.6-100 ng g −1 d.w. 128 ; 0.10-97 ng g −1 d.w. 129 ) >Bohai Bay (1.7-29 ng g −1 d.w.) 80 >Bohai Sea (0.20-4.6 ng g −1 d.w.) 123 (fig. 4c). A few to tens of ng g −1 d.w. of ΣOPEs are found for sediments from Chinese marginal seas [130][131][132] , the Maizuru Bay in Japan (range: <0.50-56 ng g −1 d.w.) 133 and the San Francisco Bay in the USA (median 23 ng g −1 d.w.) 36 . High concentrations of Σ 9 OPEs (range 4-230 ng g −1 d.w.) are measured in sediments across the Gulf of Lion in the northwestern Mediterranean Sea 38 .
Although based on a limited number of studies, it appears that the levels of OPEs in open-ocean sediment are substantially lower than those found in marginal-sea sediments. For example, the concentrations of Σ 7 OPEs from the North Pacific to the Arctic Ocean ranged from 0.2 to 4.7 ng g −1 d.w. 37 (fig. 4b). This study revealed that the contribution of chlorinated OPEs (typically TCEP and TCIPP) to the total OPEs increased from the Bering Strait to the central Arctic Ocean. Higher levels of chlorinated OPEs (range <0.02-7.4 ng g −1 d.w.) than the non-chlorinated OPEs (range <0.01-2.6 ng g −1 d.w.) were also found in the sediment of Ny-Ålesund 134 . These findings suggested that low temperatures limit the degradation of OPEs in polar oceans 134 . In addition, elevated concentrations of Σ 11 OPEs (range 0.12-57 ng g −1 d.w.) are found in sediment in the Canadian Arctic Ocean, which has been attributed to the local riverine discharge 43 .
Although the deep ocean is commonly considered the final oceanic repository of OPEs, inventory analysis shows that only a small proportion of the OPEs manufactured have been preserved in ocean sediment 43,37,123 . In an OPE inventory in the Canadian Arctic Ocean, water column OPEs even accounted for an estimated ~99% of the total OPE inventory 43 . The transfer of OPEs from surface waters to sediments is mediated by the sorption of OPEs to settling particles, a process especially relevant for the more hydrophobic OPEs. Generally, it seems that the relative roles of sediment and the water column as a final sink of OPEs will depend on the water column biogeochemistry. There, nine OPEs were detected in snow, with the total concentrations ranging from 7.2 to 20 ng l −1 . The ratio of TnBP to TCIPP is similar to those in urban snow and indoor dust 68,135,136 , implying that OPEs in snow at Dome C might partially contribute to local sources.

OPEs in snow in
In the Arctic, OPEs have been measured in surface snow along a transect between East Greenland and Svalbard 54 , with the concentrations of ∑ 8 OPEs ranging from 4.4 to 11 ng l −1 and a mean of 7.8 ng l −1 . The concentrations of OPEs in snow samples collected from coastal sites were two times higher than those from the central Arctic, and the composition of OPEs in snow was comparable with that of OPEs in seawater. These findings demonstrate the prominent role of LRAT and snow deposition in the global distribution of OPEs 78 .
High levels (mean Σ 20 OPEs 1,600 ng g −1 l.w.) were found for fish collected from Laizhou Bay, China 129 . The predominant OPEs in marine organisms were often TCIPP, TBOEP, TnBP , TPhP, TEHP and TDCIPP 130,138 . The differences in the OPE patterns among studies could result from variations in local pollution near shore 137 . Species-dependent differences in OPEs concentrations have been observed in marine organisms, with concentrations substantially lower in birds and  www.nature.com/natrevearthenviron mammals than in fish. For example, nine OPEs were found in capelin collected from Svalbard, whereas <5 OPEs were found in most of the other species, including kittiwake, Brünnich's guillemot, glaucous gull, ringed seal, harbour seal, arctic fox and polar bear (detection frequencies 0-60%) 60 (fig. 5a). Low concentrations of individual OPEs were reported for the peregrine nestlings in the Great Lakes Basin, ranging from 0 to 7.5 ng g −1 wet weight 142 . Only five OPEs were found in the liver and blubber of harbour porpoises from the  UK, with low detection frequencies of 3-44% 143 . Four out of 13 targeted OPEs were found in harbour seal blubber in San Francisco Bay, with median concentrations of <1.0 to 13 ng g −1 l.w. 36 . Five out of 17 targeted OPEs were quantifiable at sub-ppb levels in polar bear fat samples, but with variable and low detection frequencies 138 . These low concentrations contrast with high levels of ΣOPEs detected in the brain (1,500 ng g −1 l.w.), muscle (640 ng g −1 l.w.) and blubber (270 ng g −1 l.w.) in dolphins from the Alboran Sea 144 .
Bioaccumulation. The bioaccumulation and biomagnification potential of OPEs in marine organisms depends on their physicochemical properties, bioavailability and extent of biotransformation 139 . Octanol-water partition coefficient (logK ow ) values of OPEs range from −0.65 (TMP) to 9.49 (TEHP) 106 (Table 1). Significant correlations of bioconcentration factors (BCFs) with log-K ow values of OPEs are observed in marine biota 129,130 . These correlations imply that hydrophobicity plays an important role in the bioaccumulation of OPEs ( fig. 6; Supplementary Table 4), although some work finds no relationship between BCFs and logK ow of OPEs 35 . BCFs higher than the threshold value (5,000 l kg −1 ) used under the Stockholm Convention on Persistent Organic Pollutants to identify bioaccumulative chemicals have been found for TCEP, TDCIPP, TiBP, EHDPP and TEHP in different zooplankton size classes collected in the northwestern Mediterranean Sea 35 . Most OPEs detected in marine organisms do not seem to be correlated with the lipid contents 126,127,137,139 , with the exception of TEP and TPeP reported in fish 139 . Several studies have shown that higher exposure to OPEs was observed in demersal marine organisms than in zooplankton and phytoplankton, indicating that greater accumulation of OPEs occurs in the benthic environment 129,139 .
The extent of biomagnification of OPEs across food webs is still unclear and could be influenced by the differences in local OPE inputs and different metabolic processes among these species. In Manila Bay, OPEs did not biomagnify through the food web, with the exception for TPhP in demersal fish 139 . However, in Western Scheldt estuary, the Netherlands, the levels of TPhP decreased with the increase in trophic levels in both benthic and pelagic food webs, whereas biomagnification of TBOEP, TCIPP and TCEP (trophic magnification factors >1) through the benthic food web was observed 126 . In the food web in Laizhou Bay, China, eight OPEs (including TEP, TnBP, TCIPP, TDCIPP, TBOEP, TEHP, CDPP and TCrP) showed trophic magnification 129 . Different results were obtained in tropical food webs in the South China Sea, China, where OPEs concentrations decreased with the increase of their trophic levels in the following order: phytoplankton (922 ng g −1 d.w.) > zooplankton (660 ng g −1 d.w.) > oysters (309 ng g −1 d.w.) > crabs (225 ng g −1 d.w.) > coral tissues (202 ng g −1 d.w.) > fishes (58.2 ng g −1 d.w.) 145 ( fig. 5b). Lower detection frequencies and concentrations of OPEs were also observed in higher trophic levels (such as birds, seals, arctic foxes and polar bears) than in fish in Arctic biota 60 . As these studies highlight, research on bioaccumulation and biomagnification of OPEs through marine food webs is very limited and   TCEP  TCIPP  TDCIPP  TMP  TEP  TiPrP  TnBP  TiBP  TPeP  THP  TEHP  TBOEP  TPhP  TCrP  www.nature.com/natrevearthenviron is often just performed in select tissue, such as liver, and plasma in marine mammals 60 . A more complete understanding of the bioaccumulation behaviour of OPEs in marine organisms is needed to determine the biomagnification potential of these chemicals.

Impacts.
Negative effects of OPEs on marine species and populations have been confirmed by limited research. In the study on algae, which are sensitive to pollutants [146][147][148] , TDCIPP inhibited the population growth of Phaeoda ctylum tricornutum in a concentration-dependent manner by disrupting photosynthesis 147 . In addition, both TDCIPP and TnBP increased the levels of reactive oxygen species and led to oxidative damage in P. tricornutum cells at the experimental concentrations (2-10 mg l −1 for TDCIPP and 0.2-1.6 mg l −1 for TnBP) 147,148 . TCIPP is also suggested to disturb the immune system of marine mussels, based on measurements of reactive oxygen species, apoptosis, antioxidant system and related gene expressions 149 . Neurotoxicity and developmental and reproductive toxicity of some OPE compounds [150][151][152][153] have been reported for freshwater fish models. However, the biological effects of OPEs in marine fish and higher trophic levels are rarely investigated. In addition, the OPE-associated health risks for bottom-dwelling species are poorly investigated. As the input of OPEs to the marine environment continues, knowledge on potential ecological risks caused by these compounds, especially adverse effects resulting from chronic exposure, antagonistic interactions and biomagnification, is urgently required 36,129,139 .

Summary and future perspectives
OPEs are now prevalent in the marine environment, with unclear consequences for human and ecosystem health. Atmospheric and ocean transport of OPEs is occurring on a global scale, especially for chlorinated OPEs, due to their persistence, lower volatility and high solubility 2,18 . Moreover, in the high latitudes, the input of legacy OPEs from melting ice and snow could alter water column OPE concentrations. The use of OPEs as a source of nutrients by microbial communities suggests a direct link between the environmental occurrence of OPEs and biogeochemistry 154 . Local usage and environmental behaviours of OPEs, as well as discrepancies in habitat, diet and metabolism in different species, likely play important roles in the occurrence and bioaccumulation of OPEs in marine organisms. Although OPEs do not seem to biomagnify like other POPs such as poly chlorinated biphenyls and, to a lesser extent, PBDEs, potential accumulation of hydrophobic OPEs with a high logK ow has been observed in marine organisms.
To better understand and manage the potential environmental risks associated with OPEs, we recommend that the following areas of research be prioritized. First, the transport and fluxes of OPEs to and within the marine environment must be further investigated. Better experimentally derived data on the physicochemical properties of OPEs are needed to constrain the relevance of atmospheric inputs to the ocean. This information is key because the accuracy of physicochemical data can affect the estimation of air-particle portioning process 155 and air-water exchange flux direction and intensity 54 . Sea-spray aerosols have been suggested as an important vector for the regional and long-term transport of organic pollutants, so the importance of this process to OPEs needs to be assessed.
Future research will need to understand the various biogeochemical and geophysical processes under climate change and anthropogenic pressures to be able to predict the environmental fates and the global ocean health impacts of OPEs accurately. New input of OPEs from the cryosphere under warming will change environmental pathways of OPEs in the ocean environment and related health impacts. Further work, therefore, should establish a budget for OPEs present in the Southern Ocean and focus on the oceanic transport from marginal seas to the open ocean, vertical deposition in the water column, microbial degradation and photodegradation, and sinking to deep ocean sediments.
The impacts of OPE inputs on the phosphorus cycle, which are poorly understood, could be of global relevance, as phosphorus can be a limiting nutrient in large oceanic regions. One study suggested that the related organic phosphorus inputs coming from diffusive OPEs fluxes are estimated to potentially trigger up to 1.0% of the reported primary production in the most oligotrophic oceanic regions 56 . Indeed, phosphate esters can account for over 75% of the total dissolved organic phosphorus in marine environments 154 , but the contribution of OPEs and other anthropogenic OPE compounds (such as pesticides) to this pool remains unknown. The use of OPEs as a nutrient in the large oligotrophic oceanic regions should be investigated.
The role of oceanic plastics as a substantial in situ source of OPEs must also be explored. Particularly, the leaching of OPEs from microplastics accumulated in sediments, the occurrence of OPEs at the water-sediment interface and in the deep seafloor, and the potential longer-term or chronic exposure to OPEs and other plastic additives in profound oceanic environments 33 need to be addressed. Similarly, the environmental occurrence of OPEs degradation products in marine environments has been little investigated. Experimental in vivo and in vitro studies have demonstrated that a certain number of tri-OPEs can be transformed to di-ester metabolites (di-OPEs) 156 . Some di-OPEs could induce comparable or more acute toxic effects than their respective triesters 88,157 , but di-OPEs are rarely quantified, especially in marine environments 28 .
Finally, the impacts of OPEs on marine organisms must be better understood to inform future mitigation efforts. Current studies on the occurrence of OPEs in marine organisms mainly focus on near-shore regions. Further investigations in relation to offshore regions could help to clarify the natural behaviours of these compounds and their environmental impacts on the global ocean. The bioaccumulation and biomagnification behaviours of OPEs through food webs, parti cularly the entry mechanisms at the first steps (such as plankton) and their biological effects, need to be further scrutinized. In particular, more attention should be paid to health risks induced by OPEs on benthic species, because benthos shows a greater accumulation of these compounds than pelagic species. At present, toxicological data of OPEs in marine organisms, as well as environmental quality standards, are lacking. These knowledge gaps limit not only the accurate evaluation on the ecotoxicological risk of OPEs to the oceanic ecosystem, especially in the long term, chronic and interactive exposure to OPE pollutants but also our ability to support efficient chemical contamination management of these compounds.
The widespread use of OPEs and the potential resulting impacts have prompted calls for additional regulation of the production and application of OPEs and for the development of safer alternative flame retardants 158,159 . The most prevalent TCEP and TCIPP especially should gain more attention and be considered for inclusion in a global regulatory framework as soon as possible.
Published online 23 March 2022