Introduction

As a global pollutant, mercury (Hg) can be transported over long distances in the atmosphere, reaching even the most remote locations including the polar regions1. Thus elevated Hg levels have been observed in the Antarctic and the Southern Ocean2,3. Following deposition of Hg to oceans, a fraction of Hg is reduced and re-emitted into the atmosphere4,5,6. The remaining Hg can be converted to methylmercury (MeHg)7,8,9,10, which is highly toxic and highly bioavailable. MeHg may then enter the food chain, posing a threat to marine organisms, especially marine predators. For example, high Hg levels up to 73.4 ug g-1 have been reported in the feathers of long-lived birds in the Southern Ocean11.

In the oceanic photic zone, MeHg undergoes photodegradation, contributing to the observed low MeHg concentration in surface seawater2,12,13,14. Therefore, photodemethylation is considered to be an important sink for MeHg in the upper ocean13,15,16. Evidence suggests that many factors can affect the photodemethylation of MeHg, such as light intensity, salinity, dissolved organic matter, etc., among which sunlight plays a dominant role15,16,17,18,19. Interestingly, previous studies suggest that the extent and concentration of sea ice at high Northern Hemisphere latitudes can cause changes in the photodegradation of MeHg by altering the amount of solar radiation that penetrates the seawater20,21. However, this phenomenon has not been demonstrated in the Antarctic and further research is needed.

The past two decades have seen Hg isotopes emerging as a powerful tool for tracing the sources and pathways of Hg in the environment22. Mercury isotopes exhibit mass-dependent fractionation (MDF, represented by δ202Hg) during almost all biotic and abiotic chemical reactions23. In contrast, large magnitude mass-independent fractionation (MIF) for odd-mass-number isotopes (odd-MIF, Δ199Hg and Δ201Hg) only occurs in photochemical reactions23,24. For example, both the photoreduction of inorganic Hg(II) and photodemethylation of MeHg in the aqueous phase with the presence of dissolved organic matter enrich odd isotopes in the residual Hg(II) or MeHg23,24. The odd-MIF produced in aqueous photochemical reactions may be transferred to aquatic organism due to the absence of Hg MIF during trophic transfer and metabolism20,21,25,26,27. Thus, the Hg MIF values in biological tissues allow us to study the photochemical processes experienced by Hg before entering the food web. Both Point et al.21 and Masbou et al.20 used Hg MIF values in biological samples (i.e., seabird eggs and ringed seal tissues) to infer changes in the degree of photodemethylation in seawater and found a strong control of MeHg photodegradation by sea ice. Zheng et al.28 found evidence for enhanced photodemethylation during the Little Ice Age based on the odd-MIF of Hg in Antarctic biological sediments, which was proposed to link to the enlargement of polynyas (open water within sea ice). Therefore, the effect of sea ice change on marine MeHg cycle is well-captured by Hg isotope compositions in biological samples, indicating that Hg isotopes are a promising indicator of sea ice change.

The Antarctic Peninsula is one of the most rapidly warming regions on Earth, with an observed maximum rate of ~0.5 °C per decade since 195029, although some studies suggest that this warming has momentarily declined or that a cooling has been occurring since the late 1990s30,31,32. Higher temperatures are accompanied by rapid changes in the cryosphere and ecosystems of the Antarctic Peninsula, such as: (1) accelerated disintegration of glaciers, with about 52% of the glaciers showing retreat and only about 9% advancing during 1991–201533; (2) reduction in seasonal sea ice extent, with the extent of summer sea ice in the Weddell Sea dropping to a near-record level in the satellite era in 2016/1734,35,36. This rapid climate change is expected to strongly affect the Hg cycle in the Antarctic Peninsula region. The ice-free areas of the Antarctic Peninsula are home to Southern Ocean predators such as penguins and seals, and the biological sediments produced by their activities provide unique archives for studying past climate change and the Antarctic Hg cycle37,38.

In this study we report the Hg isotope compositions of a sediment core (HF4) strongly influenced by seal feces on the Fildes Peninsula, King George Island, Antarctic Peninsula (Fig. 1 and Supplementary Fig. 1)39,40. The objects of this study are to identify the source of Hg and to reconstruct past changes in the degree of photodemethylation and its relationship to climate change (such as changes in sea ice coverage). We show that the source of Hg in the sediment after ~470 CE was dominated by Hg input from seal feces. In particular, changes in Δ199Hg in sediments after ~750 CE recorded changes in the degree of photodemethylation in the ocean, which was strongly controlled by changes in sea ice.

Fig. 1: Location of sampling sites in the Antarctic Peninsula and trends in sea ice concentration in the Antarctic region.
figure 1

a The location of the sampling sites. The red filled circle represents the sediment core HF4 in this study, and the green plus signs represent the JPC-1672, GEBRA-1103 and A3104 sediment cores from previous studies. b Trends of sea ice concentration102 (SIC; red-blue shading surrounding the continent; % per decade) are calculated over 1979–2020. The trends in sea ice concentration plotted in the figure all pass the two-sided t test with a significance level of 0.1. The dotted rectangular box in subpanel (b) represents the geographic extent of subpanel (a).

Full size image

Results and discussion

Sources of Hg to the sediments

Large amounts of seal hair were found in sediment samples above 23 cm depth (Fig. 2). The number of seal hairs varies from 0.18 g–1 to 3734 g–1, with an average of 1359 ± 1059 g–1 (1 SD, n = 38), indicating that the HF4 sedimentary column is strongly influenced by the input of seal remnants (i.e., seal hairs and feces)39,40. According to modern surveys, the main seal species inhabiting the vicinity of our study site are southern elephant seals (Mirounga leonina) and Weddell seals (Leptonychotes weddellii)41. Our sediment core is likely primarily affected by southern elephant seal activity, since this species is dominant in Fildes Peninsula (maximum count ~1383, against ~123 for Weddell seals)42. Sun et al.39 reconstructed seal population changes over the past ~1500 years based on the number of seal hairs in HF4 sediments. Seal populations remained low until ~470 CE, followed by an obvious increase, reaching the first peak at ~750 CE (Fig. 2). Subsequently, seal populations declined markedly, reaching a minimum at ~1000 CE. From ~1000 to ~1300 CE, there was again a notable increase in seal populations. From ~1300 to ~1800 CE, the seal populations decreased first, and then increased to a certain extent. After ~1800 CE, seal populations appeared to decline again. The Hg content in sediments shows a trend consistent with the number of seal hairs (Pearson’s r = 0.66, two-sided t test with p value < 0.001, n = 38; Fig. 2 and Supplementary Fig. 2). Before ~470 CE, the Hg content in the sediments was relatively low, with an average value of 20.2 ± 6.0 ng g–1 (1 SD, n = 14). After ~470 CE, synchronizing with the increase of seal population, the Hg level in the sediments increased to an average of 203.0 ± 80.4 ng g–1 (1 SD, n = 34), much exceeding the soil background value in the Fildes Peninsula (<~13 ng g–1)43. In the following discussion, we further explore the source of Hg to sediments using Hg isotopes.

Fig. 2: The number of seal hairs, Hg concentration and Hg isotopes in the HF4 sediment column.
figure 2

a Changes in seal hair numbers in the sediments, indicating changes in seal populations over historical periods. Variations of Hg concentration (b), Δ199Hg (c), δ202Hg (d), and Δ200Hg (e) in the HF4 sediments. Error bars for Δ199Hg, δ202Hg, and Δ200Hg represent analytical uncertainty (2σ, see Methods for details). The Roman numbers represent different periods.

Full size image

Post-depositional alterations on the Hg isotope signatures in biological sediments are likely negligible, because as previously proposed, Hg in sediments is strongly bound to organic matter and minerals, limiting its mobility and reactivity23,28,44,45,46,47. Therefore, it is reasonable to assume that Hg isotope signals in sediments reflect those of Hg sources. The δ202Hg and Δ199Hg exhibit large variations throughout the HF4 core, whereas Δ200Hg is always near zero with no noticeable variation (Fig. 2 and Supplementary Table 1). Thus, the following discussion will be focusing on δ202Hg and Δ199Hg. Overall, the variation of Hg isotopes can be divided into two periods. Before ~470 CE (hereinafter referred to as period I), δ202Hg and Δ199Hg were relatively low, with average values of –0.76 ± 0.19‰ (1 SD, n = 10) for δ202Hg and 0.12 ± 0.11‰ for Δ199Hg (Fig. 3b), which are both similar to those of geogenic Hg within error23,48. The slightly positive Δ199Hg during period I is also similar to marine sediments receiving primarily background atmospheric Hg deposition (meaning atmospheric Hg that has undergone long-range transport and been well mixed) in non-polar regions27,49,50. Moreover, sediments in period I also show the lowest Hg contents that are close to the local soil background value43, the lowest seal population and extremely low loss on ignition and phosphorous content (indicators of biological inputs) (Supplementary Fig. 3)39,40. Taken together, Hg in sediments during period I was most likely sourced from bedrock weathering and additional contribution from background atmospheric deposition, but the input from biological remnants such as seal feces was likely minor.

Fig. 3: Hg isotope compositions of different natural samples.
figure 3

a Hg isotope signatures of potential non-seal fecal sources (snow, atmospheric deposition, plants, bedrock weathering, volcanic emissions) to the sediments. b Hg isotope compositions of marine biological samples, seal feces, seal feces-dominated sediments, and the HF4 sediments in this study. The gray-filled circles in subpanel (b) represent the Hg isotopes in sediments during Period I (before ~470 CE) in this study, while the red-filled circles represent the Hg isotopes in sediments during Period II (after ~470 CE). “ZSS PHg” represents the atmospheric particulate Hg at Zhongshan Station51, while “GWS TGM” represents atmospheric total gaseous Hg at Great Wall Station in Antarctica52. “Antarctic sediment” represents the coastal sediments of King George Island, Antarctica62. The black arrow in subpanel (b) indicates a shift in the depth of occurrence of North Pacific fish from 200 m to 600 m25. Literature for Hg isotope data for other samples can be found in Supplementary Table 2.

Full size image

After ~470 CE (hereinafter referred to as period II), both δ202Hg and Δ199Hg increased markedly, reaching 1.79 ± 0.24‰ (1 SD, n = 14) and 1.75 ± 0.19‰, respectively. The positive Δ199Hg values of sediments during period II are significantly different from those of atmospheric aerosols in the Antarctic51,52,53, polar snow4,53,54, volcanic emission55,56,57, terrestrial vegetation58,59,60, and soil/bedrock weathering products61,62,63 (two-sided t test, all p values « 0.01), which have either near-zero or negative Δ199Hg (Fig. 3a), suggesting negligible inputs from these sources. Moreover, background atmospheric deposition is also unlikely the dominant Hg source during period II because the Hg concentration of sediments during this period (203.0 ± 80.4 ng g-1) is much higher than that in nearby sediments dominated by atmospheric deposition in the Fildes Peninsula (~2.5 ng g-1)43, indicating that the low atmospheric deposition flux could not explain the high Hg accumulation in sediments. Instead, the strongly positive Δ199Hg values during period II are more similar to those of seal-feces-dominated sediments and modern seal feces in other regions of Antarctica28 (Fig. 3b). Therefore, the source of Hg to sediments during period II was likely dominated by the input of seal feces, as evidenced by the relatively high seal population during this period.

While seal feces input was the dominant source of Hg in sediments during period II, the variations of Δ199Hg and seal populations were not completely synchronized (Fig. 2). The Hg content and Δ199Hg values in sediments were elevated alongside the increase of seal population from ~470 to ~750 CE, suggesting that the increased Hg input from seal feces led to the increase of Δ199Hg during this time interval. However, during the period from ~750 to ~1300 CE, seal populations and Hg concentrations in sediments first decreased and then increased, but Δ199Hg showed an opposite trend, with a distinct peak corresponding to a decline in seal population. After ~1300 CE, although seal population and Hg content of sediments still showed large fluctuations, the Δ199Hg values showed no obvious variation. Therefore, the variation of Δ199Hg after ~750 CE cannot be explained by the change in the fraction of seal feces input to sediments. We suggest that the change in Δ199Hg after ~750 CE is more likely attributable to the change in Δ199Hg of seal feces themselves, which reflects the Δ199Hg of Hg ingested by seals. Below we further discuss the possible causes of the Δ199Hg variation in sediments dominated by seal feces.

The cause of the Δ199Hg variation in seal-feces-dominated sediments

Since trophic transfer and biological metabolism are generally found to not produce Hg MIF21,64,65,66, the Δ199Hg of seal feces should mainly reflect the Δ199Hg characteristics in the prey of seals. Thus, the variation of Δ199Hg in seal-feces-dominated sediments after ~750 CE should reflect changes in either (1) the predatory feeding behavior of seals over the historical period, or (2) the Δ199Hg at the base of the marine food chain.

The change in the predatory feeding behavior of seals over the historical period is unlikely the main cause of the Δ199Hg variation. Previous studies have shown that marine organisms living at deeper depths generally have lower Δ199Hg, because the degree of the photodegradation of MeHg prior to uptake into the food web decreases with depth in the ocean25,67. For example, the Δ199Hg of pilot whales in mesopelagic waters is much lower than the Δ199Hg of fish feeding in upper waters, but close to the Δ199Hg of fish feeding in deeper waters25,68 (Fig. 3b). Therefore, predation depth could strongly affect Δ199Hg. Multiple lines of evidence suggest that Fildes Peninsula has experienced alternating warm and cold periods over the past 2,000 years, such as Medieval Climate Anomalies (MCA, from ~1000 CE to ~1300 CE) and the Little Ice Age (LIA, from ~1300 CE to ~1750 CE) (Fig. 4)69,70,71,72,73,74,75,76. To adapt to this climate change, the predation depth of seals may have changed accordingly. McIntyre et al.77 demonstrated that elephant seals tend to dive to deeper depths when foraging in warmer waters, possibly because the vertical distribution of prey is linked to water temperature. The period from ~1000 CE to ~1300 CE was a period of warm climate, which should have been characterized by warmer waters and reduced sea ice in the study area (Fig. 4). Increased water temperature and light entering the ocean due to reduced sea ice have been found to promote deeper predation for seals77,78. The deepening of the predation depth is expected to lower the Δ199Hg in seals, and eventually lead to the decrease in the Δ199Hg of sediments, which is contradictory to the higher Δ199Hg in sediments during this time interval (Fig. 4). Conversely, ~750–1000 CE and ~1300–1750 CE were periods of colder climates, which should have been accompanied by lower sea temperature and more sea ice, so the predation depth of seals should have been shallower, also inconsistent with the lower Δ199Hg in sediments during these two periods. Therefore, we argue that the change in seal predation depth is not the main cause of Δ199Hg variation after ~750 CE.

Fig. 4: Past climate change and the Hg cycling.
figure 4

a The Southern Annular Mode (SAM) index is plotted according to Abram et al.88, and the dark green line represents the result of “LOWESS” smoothing. b Variation of Δ199Hg in the HF4 sediments after ~750 CE. (c), (d) Changes in the proportion of sea-ice taxa diatoms in the A3 and GEBRA−1 cores103,104. e Changes in total organic carbon (TOC) content in the JPC−16 sedimentary column, indicative of past climate change72. Error bars for Δ199Hg represent analytical uncertainty (2σ, see “Methods” for details). The light red shaded area represents the warm climate period, while the light blue shaded areas represent the cold climate periods.

Full size image

Thus, we hypothesize that the variation of Δ199Hg in seal-feces-dominated sediments reflects the change of Δ199Hg at the base of the marine food chain, which is eventually traced back to the Δ199Hg produced by photochemical processes in seawater. Both the photoreduction of inorganic Hg(II) and the photodegradation of MeHg produce Δ199Hg23,24. However, we argue that the MeHg photodegradation is the main process contributing to the Δ199Hg variation in seal-feces-dominated sediments, whereas the inorganic Hg(II) photoreduction is likely a minor process, for the reasons outlined below. First, the Hg species in the food sources of southern elephant seals are dominated by MeHg. According to the dietary analysis of southern elephant seals (details in Supplementary Note 1), cephalopods and fish are the main foods of southern elephant seals, and the Hg in both preys is dominated by MeHg (average above ~80% of total Hg)62,79,80. Thus, the fraction of inorganic Hg(II) directly contributed by the food sources should be small. Although a recent analysis showed a relatively low MeHg content in seal feces (average below ~20% of total Hg)81, the proportion of Hg(II) in sediments or in seal feces is not equivalent to the proportion of Hg(II) in the food sources ingested by seals, because seals have a strong ability to detoxify MeHg by converting it to Hg(II) in the liver and excreting Hg(II)20,81,82. Since biological metabolic processes do not produce Hg MIF20,21,25,26,27, the inorganic Hg(II) converted from MeHg via in vivo demethylation should inherit the odd-MIF value of the MeHg from food. Therefore, it is more likely that the inorganic Hg(II) in seal feces was mostly converted from MeHg in the food sources, and thus it is reasonable to assume that the Δ199Hg in seal feces mainly reflects the Δ199Hg of MeHg in the food sources, which primarily originated from photodemethylation in the marine environment.

Second, the ratio or regression slope between Δ199Hg and Δ201Hg is an important indicator of photochemical processes22,23. Experimental studies have shown that aqueous photodegradation of MeHg and photoreduction of Hg(II) yield Δ199Hg/Δ201Hg slopes of ~1.3 and ~1.0, respectively23,24. The slope between Δ199Hg and Δ201Hg in our sediment samples is ~1.17 (Supplementary Fig. 4). This slope is similar to Zheng et al.28, who also found a Δ199Hg/Δ201Hg slope between 1.0 and 1.3 (~1.11) in biological sediments dominated by seal feces on Ross Island, and argued that the intermediate Δ199Hg/Δ201Hg slope reflects a combination of residual MeHg that sank from the surface ocean (which should carry a signal of photodemethylation with a Δ199Hg/Δ201Hg slope of ~1.3) and MeHg produced by in situ microbial methylation in intermediate waters (which should maintain the odd-MIF signal of Hg(II) at surface ocean with a Δ199Hg/Δ201Hg slope of ~1.0). However, the Δ199Hg/Δ201Hg slope of our sediment samples is close to the value compiled from global marine fish (~1.2)15,23,25,67,83,84, which is probably more representative of the Δ199Hg/Δ201Hg slope of MeHg photodegradation in actual seawater. Moreover, the Δ199Hg values of the total Hg in global open ocean surface seawater, which is dominated by inorganic Hg(II), are typically less than ~0.4‰50. These Δ199Hg values are too low to explain the much higher Δ199Hg values in the seal-feces-dominated sediments (1.81‰ in average after ~750 CE). Therefore, the odd-MIF in the seal-feces-dominated sediments should be primarily produced by the photodemethylation process in seawater.

Seal-derived Hg MIF records changes in sea ice conditions

Following the argument in Section 2.2, the increase of Δ199Hg between ~1000 and ~1300 CE in period II (corresponding to the MCA) is most likely linked to the change of MeHg Δ199Hg values in seawater, which can be caused by a change in the source of MeHg to seawater or in the degree of marine photodemethylation.

The sources of Hg in seawater in the study area may include snow and ice melt input, upwelling (intrusion of Circumpolar Deep Water, CDW), and atmospheric deposition. Snow/ice melt in polar regions is supposed to have strongly negative Δ199Hg (as low as -5‰) produced by photoreduction on snow/ice surface as shown by both experiments and direct measurements of polar snow samples4,53,54. Thus, during the MCA (~1000 CE to ~1300 CE), the more frequent ice and snow melting events are supposed to lower the Δ199Hg value of marine MeHg, which is inconsistent with the elevated Δ199Hg in sediments during MCA (Fig. 4), suggesting that enhanced snow/ice melting is not the main reason for the increase of Δ199Hg during MCA. Furthermore, the higher Southern Annular Mode index (SAM; the dominant atmospheric circulation mode in the Southern Hemisphere) during the MCA indicates an intensification of westerly winds and therefore stronger upwelling in the study area (Fig. 4)85,86,87,88. However, stronger upwelling is expected to bring MeHg with lower Δ199Hg from deeper seawater, which also cannot explain the increase of Δ199Hg during MCA. At last, the change in atmospheric deposition to ocean could potentially change the initial Δ199Hg of MeHg before undergoing photodegradation in seawater. Thus, there may be a possibility that the increase of Δ199Hg during MCA was due to an increase of the initial Δ199Hg of atmospheric deposition. However, there is no evidence for a sudden change of atmospheric deposition during MCA. Although large variations of temperatures (up to about 12 °C) have been found to affect atmospheric deposition of Hg to snow in East Antarctica (Dome C)89,90, the difference in temperature between MCA and LIA in the Antarctic Peninsula may be within 0.4 degrees91. Thus, the possibility that the increase of Δ199Hg during MCA was due to changes in atmospheric Hg deposition to ocean is very low.

We suggest that the increase of Δ199Hg during MCA mainly reflects the enhanced MeHg photodemethylation in seawater caused by decreased sea ice extent. During the MCA (i.e., ~1000–1300 CE), the decreased sea ice extent can facilitate penetration of solar radiation into seawater, which could enhance the photodegradation of MeHg in seawater, and thus increase the Δ199Hg in the residual MeHg (Fig. 4). As MeHg enters the food chain, the Δ199Hg in seals inherits that of the residual MeHg in seawater, which is eventually recorded in our sediment samples. On the contrary, more sea ice coverage may have inhibited MeHg photodegradation during both cold periods (i.e., ~750–1000 CE and ~1300–1750 CE), leading to lower Δ199Hg in sediments. The effect of sea ice on the photodegradation of marine MeHg has been reported before. Point et al.21 estimated that Δ199Hg in murre eggs increased by ~0.68‰ when moving from the 90% ice-covered Cape Lisburne colony to the 0% ice-covered Gulf of Alaska colonies, corresponding to a ~ 8% increase in the degree of photodemethylation. By analogy, our sediment samples show a difference of 0.39‰ in Δ199Hg between the cold and warm periods, which could account for roughly a 52% difference in sea ice coverage and a 5% difference in the degree of photodemethylation between warm and cold periods.

Our finding of increased Δ199Hg in biological sediments due to enhanced photodemethylation in the Antarctic marine environment is comparable to the study by Zheng et al.28, who also found a similar degree of Δ199Hg increase by about 0.4‰ from ∼1490 CE to ~1650 CE in another sediment core (MB6) strongly influenced by the Adélie penguin on Ross Island, Antarctica. However, the timing of the Δ199Hg increase is different between our study and Zheng et al.28 (Fig. 4). We suggest that this difference may be largely related to differences in regional climate and environment. One of the important differences between the two regions is that the Ross Sea coast is strongly affected by katabatic winds92,93, and large coastal polynyas exist in the Ross Sea area94. In contrast, the Fildes Peninsula area is less affected by katabatic winds and lacks large coastal polynyas. Thus, the colder climate and more sea ice from ~1490 CE to 1650 CE in the Fildes Peninsula area would result in decreased MeHg photodegradation, leading to the lower Δ199Hg in sediments of this study (Fig. 4). However, during the same period, stronger katabatic winds and enhanced upwelling in the Ross Sea area would have increased coastal polynyas28,95,96, which may have promoted MeHg photodemethylation and hence the increase of Δ199Hg in sediments (MB6) on Ross Island28.

Modern climate warming and sea ice loss in the Antarctic Peninsula region are expected to further enhance MeHg photodegradation in seawater, and thus further increase Δ199Hg. There is indeed a weak increasing trend of Δ199Hg approaching the modern times after MCA (Fig. 4). However, the resolution of the current sediment core is not sufficient to resolve the modern changes. Two modern elephant seal feces samples collected from Byers peninsula and Zhongshan Station show lower Δ199Hg than sediments during the cold periods (Fig. 3)28. This may be related to differences in sampling locations and the individual variation of seals. Differences in regional atmospheric and marine environments (e.g., sea ice coverage) may cause large differences in Δ199Hg in seals. Moreover, each sediment sample in the study core represents seal feces input over a long time period, whereas fresh seal feces samples tend to be the excrement of individual seals, and there may be large variations in Hg isotope compositions among individuals (see more details in Supplementary Note 2).

Antarctica is a hotspot of climate change and studying the historical changes in Hg sources and processes will provide important insights for the future changes in Hg cycle under global warming. Using Hg MIF values from a sedimentary column strongly influenced by seal feces input, we reconstructed the change in the degree of photodemethylation in seawater in the study area over the past ~1200 years. The temporal variation of Δ199Hg in sediments during the historical period suggests that sea ice coverage has a substantial influence on the photodegradation of marine MeHg, which is consistent with the conclusions of previous Hg isotope studies in modern Arctic biological samples20,21. From another perspective, our study also demonstrates that Hg isotopes in biological sediments are valuable proxies of past climate change. The ice-free area of Antarctica is rich in biological sediments, which can provide valuable records for studying the temporal and spatial changes of the marine MeHg cycle in historical periods. However, the current work cannot answer whether this change in the degree of photodemethylation can substantially affect Hg levels in organisms, which are likely affected by multiple factors such as primary productivity, trophic levels, and associated atmospheric and oceanic processes. Further research is needed in the future to better understand this complex biogeochemical system. Moreover, Hg isotopes in modern biological samples or biological sediment cores with better temporal resolution may provide more details about the relationship between sea ice and marine Hg cycling across different time periods, and thus warrant further investigation.

Methods

Site description and sample collection

During the Eighteenth Chinese Antarctic Research Expedition (November 2001–March 2002), a 42.5 cm long sediment core, referred to as HF4, was obtained from a terrestrial catchment (62°11'57"S, 59°58'48"W) in a depositional basin situated on the west coast of Fildes Peninsula39. This basin is located within the second marine terrace at an altitude of 8 m (see Fig. 1 and Supplementary Fig. 1). The seals inhabiting around the sampling site were identified as southern elephant seals (Mirounga leonina)39. During sampling, a 12 cm diameter PVC plastic gravity corer was inserted vertically until bedrock was encountered and extracted rapidly. The sediment core was divided into sections in the laboratory, with an interval of 0.5 cm for the top 18 cm and 1.0 cm for the rest of the sediment. There was a substantial quantity of seal hair in the top 18 cm of sediment, whereas the subsequent 5 cm of sediment exhibited a minimal presence of seal hair. No seal hair was found in the sediment below 23 cm. Each section was air-dried and thoroughly mixed. Approximately one-eighth of each bulk sub-sample was further dried for 24 hours at 105°C, and then used to count seal hairs using the methodology described by Hodgson et al.97 The dating of the HF4 core is determined using 137Cs and conventional 14C dating methods, as reported in previous studies39,40. In order to mitigate the marine reservoir effect on 14C dating, only four sediment samples with minimal presence of seal hairs were selected (Supplementary Table 3). By performing a linear extrapolation of 14C-age against depth, an intercept value of 114 years was obtained for the modern surface sediments (confirmed by 137Cs activities)39,40 (Supplementary Fig. 5a). Hence, the value of 114 years is considered to account for the reservoir effect, and used to correct the measured radiocarbon ages by subtracting this value from them. The chronology of the core was established by applying linear interpolation between the corrected 14C ages (Supplementary Fig. 5b).

Hg stable isotope measurements

The determination for Hg concentration in sediments and quality control have been reported in previous work39,40. Prior to isotope analysis, Hg in samples was first extracted using a thermal combustion method98. Briefly, powdered samples (typically <2 g) were combusted within the furnace unit of Hydra IIc Direct Mercury Analyzer at 800 °C under an oxygen flow of 250 ml min-1. The elemental mercury vapor released during thermal decomposition was swept into a trapping solution comprising 0.2% KMnO4 (w/w) and 5% H2SO4 (v/v), where it was oxidized to Hg(II). To monitor Hg yields and the precision of Hg isotope measurement, procedural blanks were processed alongside samples, as was standard reference material (CCRMP TILL-1, Ontario soils). Recoveries of Hg for the procedure standard were 96.1 ± 2% (1 SD, n = 3). Recoveries of Hg for all samples were 94.4 ± 8% (1 SD, n = 24). Mercury concentrations in procedural blanks were typically <0.04 ng g-1, which is less than 1% of the Hg present in samples.

Mercury isotopes were analyzed via multi-collector inductively coupled plasma mass spectrometry (MC-ICP-MS, Neptune Plus, Thermo Scientific) at University of Toronto using previously described techniques98,99. The trapping solutions containing Hg extracted from samples were first neutralized with NH2OH·HCl to reduce KMnO4, and then diluted to 1–2 ng g-1 Hg using a matrix matched pre-neutralized trapping solution. The Hg in the diluted solutions was introduced into the instrument using the online cold vapor generation method99. Instrumental mass bias was corrected using the 205Tl/203Tl ratio of an internal Tl standard (NIST 997) introduced as desolvated aerosol and strict standard-sample-standard bracketing with the Hg standard reference material (SRM) NIST SRM 3133. The bracketing standard was prepared in the same matrix solution as samples and concentrations were matched such that standard signal sizes were matched to within ±10% of sample signal sizes. Isobaric interference from 204Pb was corrected by measuring 206Pb, but was always negligible. On-peak zero corrections determined from a matrix matched blank run prior to the sample were applied to all Hg and Pb masses. Mercury isotope compositions are reported using δ notation:

$${\delta }^{x}{{{{{\rm{Hg}}}}}}(\permil )= \, \left [\frac{{({\!\;}^{x}{{{{{\rm{H}}}}}}{{{{{\rm{g}}}}}}/{\!\;}^{198}{{{{{\rm{H}}}}}}{{{{{\rm{g}}}}}})}_{{{{{{\rm{sample}}}}}}}}{{({\!\;}^{x}{{{{{\rm{H}}}}}}{{{{{\rm{g}}}}}}/{}^{198}{{{{{\rm{H}}}}}}{{{{{\rm{g}}}}}})}_{{{{{{\rm{NIST3133}}}}}}}}\,-\,1\,\right ]\times 1000$$
(1)

where x is the mass number of each Hg isotope, including 199, 200, 201, 202 and 204. MDF is reported as δ202Hg. MIF (ΔxHg) is reported as the difference between measured δxHg and the theoretical values based on kinetic mass dependent fractionation law100:

$${\Delta }^{{{\mbox{x}}}}{{\mbox{Hg}}}={{{{{{\rm{\delta }}}}}}}^{{{\mbox{x}}}}{{\mbox{Hg}}}-{\beta}\times {{{{{{\rm{\delta }}}}}}}^{202}{{\mbox{Hg}}}$$
(2)

where x is the mass number 199, 200, 201 and 204. β is a scaling constant to estimate the theoretical kinetic MDF, and it is 0.2520, 0.5024, 0.7520 and 1.4930 for 199Hg, 200Hg, 201Hg and 204Hg, respectively. Each sample was measured twice and an in-house Hg standard JTBaker was measured 5 − 7 times during each session. The analytical uncertainties (2σ) are reported as either 2 standard error (2SE) of sample replicates or 2 standard deviation (2 SD) of JTBaker (Supplementary Table 1 and 4), whichever is higher. The measured Hg isotope ratios for JTBaker (δ202Hg = –0.63 ± 0.05‰; Δ199Hg = 0.02 ± 0.02‰; Δ200Hg = 0.01 ± 0.03‰, n = 12) and for TILL-1 (δ202Hg = −1.11 ± 0.01‰; Δ199Hg = -0.11 ± 0.01‰; Δ200Hg = 0.01 ± 0.01‰, n = 3) are both consistent with published values98,101.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.