Anthropogenic 129 I in the South China Sea and coastal waters around Taiwan

In this article, we present a study of seawater 129 I/ 127 I time-series data from several coastal sites in Taiwan, including Yehliu Geopark (north–northeast coast), Kaohsiung (southwest coast), and Zhuwei Fishing Village (northwest coast). The objective of this study was to document how 129 I/ 127 I responds to known seasonal variations in the surface ocean currents that carry 129 I to each of these sites.The responses were shown to be quite distinct. The Zhuwei site, across Mainland China, had elevated spring and summer 129 I/ 127 I values, with abrupt peaks that reflected transient cross-strait currents that carry seawater with elevated 129 I to the northwest coast of Taiwan.The Yehliu site, which receives all of its seawater from the Kuroshio Current, had relatively low and uniform 129 I/ 127 I values year-round.The Kaohsiung site showed a summertime minimum that likely stems from upwelling upstream induced by oceanic eddies associated with the Kuroshio intrusion.To our knowledge,these are the first continuous coastal 129 I/ 127 I time-series data published for Asian waters and the first to show large, abrupt, and regional 129 I/ 127 I changes in the surface seawater. We also documented 129 I/ 127 I values from multiple surface ocean sites in the South China Sea (SCS), including a vertical profile from the South East Asia Time-Series Station that extends to a depth of 3,7 00 m.The 129 I from both coastal Taiwan and surface waters of the SCS is >98% anthropogenic, primarily released to the environment as a byproduct of nuclear fuel reprocessing.


Introduction
Anthropogenic radioisotopes produced by fission of uranium or plutonium are characteristic markers of the Anthropocene (Zalasiewicz et al., 2015). Among these, 129 I occupies a special niche because unlike other fission products, it continues to be released to the environment as a by-product of nuclear fuel reprocessing, and the amount of 129 I in the environment continues to grow. These releases are permitted because 129 I is a relatively benign isotope with a very long half-life (15.7 Â 10 6 year), low specific activity, and relatively low-energy decay products (Edwards, 1962). The first large-scale releases of anthropogenic 129 I into the environment occurred in the late 1950s and early 1960s from aboveground nuclear weapons testing. These produced an estimated 90 + 45 kg of 129 I compared to a prenuclear marine inventory of about 130 kg (Snyder et al., 2010) produced naturally from fission of 238 U and by interactions between cosmic rays and Xe in the atmosphere (Edwards, 1962;Fabryka-Martin et al., 1985). Aboveground nuclear weapons testing produced peak oceanic 129 I/ 127 I concentrations preserved in banded corals (Biddulph et al., 2006;Bautista et al., 2016;Chang et al., 2016;Sakaguchi et al., 2018) and in marine sediments (Fan et al., 2016;Zhao et al., 2019). In the ensuing years, after the era of atmospheric weapons testing, these marine archives also document a second, and much larger, anthropogenic source of 129 I. This second source comes primarily from nuclear fuel reprocessing plants that recover economic isotopes from spent nuclear fuel rods. On a global scale, the two most significant 129 I sources to date are the Sellafield (England) and La Hague (France) commercial nuclear fuel reprocessing plants (Raisbeck and Yiou, 1999;Reithmeier et al., 2006;Hou et al., 2009;Hu et al., 2010b). The amount of 129 I in the world's oceans was recently estimated to have reached 7,310 kg . This estimate implies that as much as 98% of the 129 I in the oceans is anthropogenic and that approximately only 1% of that anthropogenic 129 I came from atmospheric nuclear weapons testing. Hence, the amount of 129 I in the oceans has increased by about a factor of 60 from a prenuclear value of 130 kg, to 7,310 kg. According to the International Atomic Energy Agency (IAEA), from the onset of nuclear power production in 1954 to 2013, approximately 370,000 metric tons of heavy metal (t HM) in spent fuel has been consumed, and about one-third of that (120,300 t HM) has been reprocessed (IAEA, 2018). Most of the remaining spent fuel is stored at the 236 nuclear power stations where it was produced, and the 129 I it contains has not yet been released to the environment (Hou et al., 2009;Hu et al., 2010b).
The increase in the amount of 129 I in the oceans over the past 70 years makes clear that the rate of iodine removal by natural sinks is slower than the rate of its release by reprocessing plants. Anthropogenic marine 129 I measurements from around the world also show that neither the shallow ocean nor the deep ocean is in equilibrium with 129 I (Snyder et al., 2010). As a small subset of this global picture, we examined how seawater 129 I values change with time in the vicinity of Taiwan in 2018 and 2019, with an emphasis on seasonal variations seen in time-series data collected regularly from several sites around the island.
Iodine is a very mobile element in nature that forms few minerals; its geochemistry is dominated by oceanic and atmospheric transport and storage (Fuge and Johnson, 1986). Iodine in seawater occurs as dissolved iodide, iodate, organic iodine, and particulate iodine (Wong, 1991;Santschi and Schwer, 2004). The concentration of dissolved iodine in seawater is about 60 mg L À1 , and although iodine is a very mobile element, 129 I can be used as a semiconservative Àtracer in seawater to determine current pathways, transit times, and water mass mixing. A number of tracer studies of this type have been conducted in the North Atlantic Ocean and the Arctic Ocean, downstream from the release sites at Sellafield and La Hague (Raisbeck and Yiou, 1999;Alfimov et al., 2004;Smith et al., 2005), and in recent years, such studies have been refined by combining 129 I with other anthropogenic isotopes (i.e., 137 Cs, 99 Tc, and 236 U) to better discriminate between point sources and to determine more accurate transit times (Smith et al., 1998(Smith et al., , 2011Yiou et al., 2002;Christl et al., 2015;Casacuberta et al., 2018).
Tracer studies rely on time-series data to determine pathways, rates, and deposition of anthropogenic isotopes. Marine 129 I time series can be derived from seawater , seaweed (Raisbeck and Yiou, 1999;Yiou et al., 2002), corals (Biddulph et al., 2006;Bautista et al., 2016;Chang et al., 2016;Sakaguchi et al., 2018), sediments (Fan et al., 2016;Zhao et al., 2019), or models (Alfimov et al., 2006;Smith et al., 2011). Seaweed and corals allow for time series with sub-annual to annual resolution. The time resolution of seawater samples is limited only by the rate at which samples are collected, which can be tailored to meet specific goals. Here, we chose bimonthly (every 2 weeks) resolution in order to document seasonal 129 I variations. This seasonal variability complements 129 I/ 127 I time-series studies from corals, with annual resolution.  Figure 1). None of these sites is located near a major river. The Zhuwei Fishing Village and Yehliu Geopark sites are only 45 km apart, but their seawater sources are very different. The Zhuwei site faces the northern end of the Taiwan Strait where it connects to the East China Sea (ECS), while the Yehliu site faces the Pacific Ocean. The Kaohsiung site, on the campus of National Sun Yat-Sen University ( Figure  S1), also faces the Taiwan Strait but on the southwest coast of Taiwan ( Figure 1). Open ocean sites were sampled during a cruise of the R/V Legend (LGD-T11) from July 4 to 12, 2018. These included a number of surface sites (5-m depth) and a vertical profile from the South East Asia Time-series Study (SEATS) site ( Figure 1). Figure 2 shows the surface ocean currents that affect the coastal sites selected for this study. The most dominant current affecting Taiwan is the Kuroshio Current, a western boundary current of the North Pacific Ocean, which flows from south to north along the east coast of the Philippines and then along the east coast of Taiwan. The Kuroshio Current has a mean geostrophic volume transport of approximately 15 Sv (1 Sverdrup ¼ 10 6 m 3 s À1 ) as it passes the northern tip of Luzon Island, with maximum transport in spring and minimum transport in fall (Qiu, 2019). When the Kuroshio Current passes this point, a portion of its waters enters the SCS through the Luzon Strait (Chu and Li, 2000;Hu et al., 2000). This portion is known as the  (Schlitzer, 2020 (Hu et al., 2000(Hu et al., , 2010aQiu et al., 2011). Nan et al. (2011) defined a Kuroshio SCS index (KSI) based on geostrophic vorticity to quantify analogous intrusion pathways as a leaking path, looping path, or leaping path. Although the Kuroshio intrusion can follow any of these paths at any time, the leaking and looping paths are most probable in autumn, winter, and spring, and the leaping path is most probable in summer (Nan et al., 2015). This seasonality is associated with the Southeast Asian monsoon. Regardless of the path, the Kuroshio intrusion contributes to the seasonal flow of water through the Taiwan Strait and is responsible for its northward flow year-round. Estimates of the westward volume transport of Kuroshio water through the Luzon Strait range from approximately 3 Sv to >10 Sv (Nan et al., 2015), while the northward volume transport through the Taiwan Strait is approximately 2.3 Sv in summer and approximately 0.8 Sv in winter (Hu et al., 2010a).

Total iodine
The method we describe below extracts dissolved iodine from seawater; in this study, "total iodine" and "total 129 I/ 127 I" are operationally defined as the iodine extracted using our technique. As noted above, dissolved iodine in seawater occurs in three forms: inorganic iodide (I À ), inorganic iodate (IO À 3 ), and organic iodine. Our method extracts iodide, iodate, and organic species where iodine is loosely bound, such as methyl iodide. Many sensitive methods have been devised to extract and quantify individual dissolved inorganic 129 I species (Hou et al., 1997) and dissolved organic species (Schwehr and Santschi, 2003;Zhang et al., 2010). These provide valuable tools to understand 129 I pathways and their kinetics. However, in this study, we are concerned only with the bulk anthropogenic 129 I carried in the seawater. Exchange between various species within a parcel of water will not change the total amount of anthropogenic iodine within it.
We also do not consider stable ( 127 I) iodine concentrations because stable iodine is very uniform in the surface ocean (Fuge and Johnson, 1986;Wong, 1991;Chance et al., 2014). In one large study, for example, surface waters from a vast area of the Pacific (>20 N, 137-163 E) were shown to vary by less than 3% (Tsunogai and Henmi, 1971). This level of variation falls within the errors of our 129 I/ 127 I measurements and cannot explain variations larger than a few percentage.

Sampling
Surface water from open ocean sites was obtained during a cruise of the R/V Legend (LGD-T11) from July 4 to 12, 2018 ( Figure 1). Samples for an 129 I/ 127 I depth profile were collected at the SEATS site (18.120 N, 116.052 E) on July 8, 2018. The depth profile samples were collected with Teflon-lined 12-L Go-Flo bottles mounted on a conductivity-temperature-depth/Rosette frame, which also measured fluorescence (Chlorophyll a) with depth. Upon recovery of the Go-Flo bottles, each open ocean sample was filtered through a 0.22-mm Polycap filter cartridge into a 1-L low-density polyethylene bottle. Surface water samples at 5-m depth were collected from a Teflon pipe in the wet lab on deck using a diaphragm pump. All the water samples were kept under 4 C before shipment to the laboratory. Two liters of seawater were collected from each coastal site ( Figure S1) at 2-week intervals using 1-L high-density polyethylene bottles.

Analysis and uncertainties
Coastal seawater samples were filtered in the laboratory with 0.22-mm nylon membrane filters. Iodine was extracted from the filtered seawater in 1-L batches using the following steps : (1) the sample was placed in a 1-L separatory funnel, (2) it was acidified to pH < 2 using phosphoric acid (H 3 PO 4 ), (3) 2.5 mL of 0.48 M sodium bisulfite (NaHSO 3 ) was added and allowed to react overnight, (4) 5.4 mL of 1.0 M sodium nitrite (NaNO 2 ) was added to the sample, and (5) I 2 was extracted using chloroform in a separatory funnel as an immiscible solvent.
Steps 4 and 5 were repeated 3 times to assure complete removal of the iodine produced using 0.4 mL of NaNO 2 in each successive extraction. The chloroform was decanted into 20-mL vials with about 10 mg of silver powder, which reacted with the I 2 to form AgI. The chloroform was allowed to evaporate, and the AgI was rinsed repeatedly and  dried. No carrier was added to the sample. The reactions that occur during each step of the extraction are given in the Supplemental Material (Text S1) along with a photograph of the chloroform extraction step ( Figure S2). The accelerator mass spectrometric measurements were performed at the Micro Analysis Laboratory, Tandem Accelerator (MALT) facility, University Museum, University of Tokyo, using established measurement protocols (Matsuzaki et al., 2015). AgI samples were mixed with Nb powder and pressed into aluminum target holders for accelerator mass spectrometric analyses. The measurements were conducted at a terminal voltage of 3.48 MV, and particles in the 5þ charge state were selected. 129 I/ 127 I values were calculated relative to the S-Purdue 2 standard with a nominal 129 I/ 127 I ratio of 6.54 Â 10 À11 . The machine blank for 129 I/ 127 I at the MALT facility is below 1.5 Â 10 À14 . Uncertainties were calculated based on counting statistics (Burr et al., 2007) using the number of counts observed in the samples and standards. Machine instabilities were also taken into account.
Chlorophyll a was measured using a fluorescence meter mounted as an auxiliary detector on an SBE 9 plus conductivity-temperature-depth profiling recorder (Seabird Scientific) with a detection resolution of 1 datum/s. The fluorescence technique followed standard protocols for the method (Roesler et al., 2017). The instrument was calibrated at the Taiwan Ocean Research Institute immediately prior to the cruise. Chlorophyll a values and their uncertainties are given in Table S1 based on about 500 measurements at each depth interval.

Results and discussion
We report three types of 129 I/ 127 I results: (1) open ocean surface water (5-m depth; Table 1), (2) a vertical profile from the SEATS site (Table 2), and (3) surface seawater time-series results from the coastal sites ( Table 3). All of the measured values are given in the tables along with site details, collection dates, and number of 129 I particles detected. The Kaohsiung time-series results were reported previously in the Proceedings of the 8th East Asia AMS Symposium  and are included here to facilitate comparison.
The SCS is the largest marginal sea in Southeast Asia with water depths of more than 5,000 m and a very broad continental shelf along its western border. As the basin is largely restricted and the climate is monsoonal, the surface circulation in the SCS is driven by the monsoon, with a seasonal reversal of surface current directions (Wyrtki, 1961). The mean circulation of the upper layer (0-200 m) of the SCS is cyclonic in winter and anticyclonic in summer (Hu et al., 2000). Circulation is also dependent on the Kuroshio intrusion as discussed above. Most of the seawater that replenishes the SCS ultimately comes from the Kuroshio Current, with a net westward flow through the Luzon Strait (Hu et al., 2010a;Nan et al., 2015). In addition to seasonal variations, both the Kuroshio intrusion and Kuroshio Current exhibit significant interannual variability in terms of pathways and volume transport, and the Kuroshio intrusion causes the formation of mesoscale eddies that promote vertical mixing, especially in winter (Caruso et al., 2006;Hu et al., 2010a;Nan et al., 2015;Qiu, 2019). As shown below, 129 I/ 127 I is very sensitive to vertical mixing.

Open ocean results
The locations of the open ocean sites sampled for this study are shown in Figure 1. Taiwan Strait sites (TS a and TS b) receive water from the Kuroshio Loop Current extension in the Penghu Channel ( Figure 2). The Dongsha Island site (DS) and Northern South China Sea site (NSCS) are situated in the path of the SCS Warm Current (Figure 2). The Luzon Strait sites (LZ a and LZ b) are located in the Luzon Strait, where water from the Kuroshio intrusion enters the SCS. Despite their hydrographic differences, the 129 I/ 127 I values at these surface water sites were statistically indistinguishable as shown in Figure S3. That is, all individual measured values overlap with the weighted mean for the group: 129 I/ 127 I ¼ 49.1 + 2.4 Â 10 À12 (2s external variance) during the sample collection period from July 4 to 12, 2018. This average gives us a reference value for the northern SCS, upstream from the Taiwan Strait, to compare with our coastal sites further north, at Kaohsiung and Zhuwei Fishing Village. Our SCS surface ocean results are typical of the open ocean in the region and overlap with southeastern ECS values in the range of 20-60 Â 10 À12 reported elsewhere Wang et al., 2019). They also overlap with surface water values measured at the SEATS site, discussed next.   Figure 3) and from several deep water sites off the northeast coast of Japan (Suzuki et al., 2010(Suzuki et al., , 2018. Our SEATS profile shows that surface water 129 I/ 127 I values in 2019 were enriched by a factor of about 50 above SCS deep waters. Hence, 129 I is a very sensitive upwelling indicator. The 129 I/ 127 I depth profile we observed is typical of marine depth profiles, with most of the 129 I contained in the surface layer of the ocean. The profile observed by Yi et al. (2018), measured from a nearby site 4 years prior to our sampling, is very different with respect to depth in the water column. However, the overall shape of the two profiles is similar, and they have overlapping surface and deep water values. The interannual variability of mesoscale eddies associated with the Kuroshio intrusion north of the site could contribute to the vertical mixing observed in 2014. As these are the only two 129 I/ 127 I depth profiles currently available, conjecturing further would be premature. An interesting feature in our record is an increase in 129 I/ 127 I (from 1.2 to approximately 6 Â 10 À12 ) near the seafloor at 3,500 m depth. This increase suggests remobilization of particulate iodine associated with sinking organic matter or particulate iodine. Based on the work of Brewer et al. (1980), Wong (1991) estimated that about 4% of the particulate iodine in surface ocean waters can settle to the deep ocean, and as much as 95% of that iodine can be dissolved at the water-sediment interface, providing a potential pathway to explain our observed deep water 129 I signal (Figure 3). An inflection point is also observed in our 129 I/ 127 I profile between 65 and 200 m depth (Figure 3). Iodine is a biophilic element known to be scavenged by marine organisms, especially seaweed (Fuge and Johnson, 1986;Hou et al., 1997;Santschi and Schwer, 2004). Our measured peak Chlorophyll a concentrations (approximately 77 m depth) coincide with the 129 I/ 127 I inflection seen in our profile (Figure 3).

Bimonthly time series from coastal sites
Measured 129 I/ 127 I time series from Yehliu Geopark, Kaohsiung , and Zhuwei Fishing Village are given in Table 3. The 1s precision of the measurements was <5% and improved as the study progressed with refinements to the sample processing protocol that included larger sample volumes (up to 2 L) and longer reaction times (overnight) after the addition of sulfite. With reference to the surface ocean currents summarized in Figure 2, we can follow the marine pathways of iodine in the vicinity of Taiwan. The most straightforward of these pathways involves the Yehliu site, which is dominated by the Kuroshio Current as it passes the northern tip of Taiwan. Over the entire observation period (March-November 2019), the Yehliu data are very consistent, with no seasonal trend. At two standard deviations, the average value varies by less than 4%, with a weighted average of 129 I/ 127 I ¼ 48.61 + 1.90 Â 10 À12 (2s external variance). These low values reflect the large volume of water carried  by the Kuroshio Current, which passes no significant anthropogenic 129 I source in its path and is apt to inherit low-129 I upwelled waters from its parent current, the North Equatorial Current, and from mesoscale eddies (Qiu, 2019). Kaohsiung has a direct connection with the Kuroshio Current by way of the Kuroshio Loop extension that flows northward through the Penghu Channel along the southwestern coast of Taiwan (Figure 2). This region is the only part of the west coast of Taiwan where surface currents flow northward year-round, every year (Qiu et al., 2011). Upstream from Kaohsiung is an area where seasonal cold core eddies form, driven by the Kuroshio intrusion (Li et al., 1998). Cold core eddies induce vertical mixing and can dilute surface water 129 I. This effect is what we observed during the summer of 2019 at Kaohsiung, where surface water 129 I/ 127 I dropped from 48.07 + 4.10 Â 10 À12 on June 9 to 38.5 + 4.2 Â 10 À12 (2s) on July 22 during the summer monsoon. This minimum is well below our SCS surface water average value, 49.1 + 2.4 Â 10 À12 ,  and is the lowest value among seven consecutive measurements that define a smooth minimum over a 3-month period (Figure 4). On a longer time scale, our Kaohsiung time-series data are consistent with 129 I/ 127 I measurements from the same site, which ranged from approximately 36 to 53 Â 10 À12 between 2011 and 2015 . At Zhuwei Fishing Village, we recorded the highest 129 I/ 127 I values of any site, as much as 2.4 times the average for Yehliu, which is only 45 km away ( Figure 5). Clearly, this high level of 129 I did not come from the direction of Yehliu because of the low values there, as well as the southern ECS in general . However, Liu et al. (2016) identified high 129 I/ 127 I values (up to approximately 200 Â 10 À12 ) in the Yangtze River Estuary and in coastal waters that extend to the south. The China coastal current flows toward the south, off the coast of Mainland China, and carries water into the Taiwan Strait. During the winter and spring, transient cross-strait surface flow can develop (Liao et al., 2018), shown as the dashed current trajectory in Figure 2. Hence, we have a known source and known pathway that can explain the two   Taiwan Strait (130 km), surface water need not flow directly across to arrive there. Water can cross at any point to the north or south, because although coastal waters along the northwest coast of Taiwan generally flow northward, they have also been observed to flow to the south at times (Qiu et al., 2011). Our 129 I results demonstrate that 129 I can serve as a useful indicator of this cross-strait flow, which was strongest in the spring and summer of 2019 at Zhuwei Fishing Village.

Conclusions
In this study, we have documented surface seawater 129 I/ 127 I from a large area in the SCS and found uniform values that reflect well-mixed waters. We documented a vertical 129 I/ 127 I profile to 3,700 m depth from the SCS SEATS site that shows prenuclear values in deep waters, a typical diffusive curve, and an 129 I/ 127 I enrichment of approximately 50 times in surface waters compared to deep waters. We followed the path of anthropogenic 129 I in ocean currents from the SCS to the east and west coasts of Taiwan. At Yehliu, all of the water comes from the Kuroshio Current and with no distinct seasonal effects. At Kaohsiung, we observed a summertime 129 I minimum likely associated with seasonal cold core eddies induced by the Kuroshio intrusion. At Zhuwei Fishing Village, we observed high spring and summer 129 I/ 127 I values with sporadic short-lived 129 I/ 127 I peaks that record transient periods of cross-strait flow with water likely coming from the China Coastal Current. To our knowledge, these data are the first continuous coastal 129 I/ 127 I time series published for Asian waters and the first to show abrupt changes in surface water 129 I/ 127 I in the coastal waters of Taiwan. In addition to revealing such local changes, studies of this type, and of anthropogenic isotopes in general, are useful for quantifying and understanding human impacts that define the Anthropocene.

Data accessibility statement
A complete list of 129 I/ 127 I results cited in this study is included in the main body of the manuscript (Tables 1-3). Chlorophyll a data measured at the SEATS site are given in Table S1 of the Supplemental Material.

Supplemental files
The supplemental files for this article can be found as follows: Figure S1. Coastal sampling sites. Photographs of the three coastal sites studied in this article: (a) National Sun Yat-Sen University, Kaohsiung, (b) Yehliu Geopark, and (c) Zhuwei Fishing Village. Figure S2. Extraction of iodine from seawater. Immiscible chloroform at the base of a separatory funnel with artificial seawater above. The purple color is from dissolved iodine from 20-mL methyl iodide dissolved in 1 L of artificial seawater. Figure S3. Plot of 129 I/ 127 I values from open ocean sites in the South China Sea. The dashed line is the weighted average of these values, and the shaded error represents the 2s uncertainty of the weighted average (49.1 + 2.4 Â 10 À12 ). All samples overlap with the average within errors. Table S1. Chlorophyll a results. Text S1. Docx.