Ocean sediments are the largest sink for mercury (Hg) sequestration and hence an important part of the global Hg cycle¹. Yet accepted global average Hg flux data for deep-ocean sediments (> 200 m depth) are not based on measurements on sediments but are inferred from sinking particulates². Mercury fluxes have never been reported from the deepest zone, the hadal (> 6 km depth). Here we report the first measurements of Hg fluxes from two hadal trenches (Atacama and Kermadec) and adjacent abyssal areas (2–6 km). Mercury concentrations of up to 400 ng g⁻¹ were the highest recorded in marine sediments remote from anthropogenic or hydrothermal sources.

The two trench systems differed significantly in Hg concentrations and fluxes, but hadal and abyssal areas within each system did not. The relatively low recent mean flux at Kermadec was 6–15 times higher than the inferred deep-ocean average^1,3, while the median flux across all cores was 22–56 times higher. Thus, some hadal and abyssal sediments are Hg accumulation hot-spots. The hadal zone comprises only ~ 1% of the deep-ocean area, yet a preliminary estimate based on sediment Hg and particulate organic carbon (POC) fluxes suggests total hadal Hg accumulation may be 12–30% of the estimate for the entire deep-ocean. The few abyssal data show equally high Hg fluxes near trench systems. These results highlight a need for further research into deep-ocean Hg fluxes to better constrain global Hg models.

Introduction

Sediment dynamics in hadal trenches are governed by two processes: (1) continuous pelagic deposition and (2) infrequent mass wasting/deposition events that translocate large quantities of sediment from the trench slope to the trench interior (typically driven by seismic activity^{https://doi.org/10.1038/s41467-017-02504-1 (2017)." href="https://www.nature.com/articles/s41598-021-90459-1#ref-CR27" id="ref-link-section-d39636e818">27,28}). The profiles of sediment core excess ²¹⁰Pb (²¹⁰Pb_ex: defined as the ²¹⁰Pb fraction in secular disequilibrium from ²²⁶Ra daughters derived from the atmosphere and the water column) have clear indications of both processes (Fig. S1). Periods of monotonic exponential decreases of ²¹⁰Pb_ex shown in most core sections reflect relatively constant sediment deposition. Sections of some Atacama cores (At 2, 3, 5, 7, 9 and 10) exhibit short-term ²¹⁰Pb_ex trend reversals and plateaus that reflect dilution of ²¹⁰Pb_ex by episodic mass deposition of low ²¹⁰Pb_ex sediment.

We calculated constant linear sedimentation rates (i.e., the sedimentation rates were unvarying down-core) from the periods of continuous deposition, excluding mass deposition events, and thereby provide a conservative estimate of overall sediment deposition rates at the trench axis. The Atacama cores exhibiting mass deposition events have similar calculated sedimentation rates to the other Atacama sites (Table S1).

For our purposes, we grouped Hg concentration data from the top 5–6 cm (4–5 slices; Fig. S2) of each core, where ²¹⁰Pb_ex was present, as “recent” sediments which, based on the vertical accretion rates, represent material accumulated over about the last 60–190 years (median 120 years). The bottom five slices of each core, including sediments from 10 to 40 cm depth where ²¹⁰Pb_ex was absent (see Fig. S1), were grouped as “historical” sediments. Although the historical samples can not be accurately dated, they occur at core depths well below the point of ²¹⁰Pb_ex extinction, which would typically place their deposition prior to about the middle of the nineteenth century given ²¹⁰Pb’s half-life of 22.3 years.

Atacama sediments overall contained significantly higher Hg concentrations (P < 0.001) than Kermadec (Fig. 2). Sites in the trenches at both locations contained similar recent Hg concentrations to sites on the adjacent abyssal plains and slope (i.e., At1, At7, At9, and Kc7; the difference between trench and abyssal sites is non-significant at P > 0.05). Recent Atacama sediments on average had significantly higher Hg concentrations (188 ± 56 ng g⁻¹; P < 0.001) than historical sediments (137 ± 51 ng g⁻¹).

The exception to this pattern was At2, in which recent sediments had lower Hg concentrations (P < 0.01) than historical. Recent Kermadec sediment Hg concentrations averaged across all sites (44 ± 26 ng g⁻¹) were not significantly different from historical sediments (41 ± 16 ng g⁻¹; P > 0.05), although they were higher at Kc7 (P < 0.001).

Plots of average (A) Hg concentrations and (B) Hg fluxes at the deep-ocean study sites. Left hand panels show concentration and fluxes for recent and historical sediments at individual sites. Right hand panels show concentrations and fluxes at all At and all Kc sites combined, and the concentrations and fluxes in recent and historical sediments of all sites. For individual sites, in (A) bars represent mean ± standard deviation (SD) Hg concentrations in the upper 5 cm (“recent”, N = 5) and bottom 5 slices (“historical”, N = 5) of sediment cores. In (B), bars represent mean ± SD Hg fluxes calculated from the mean and SD concentrations and sediment accumulation rates. Significant differences between recent and historical Hg concentrations and fluxes indicated by: *P < 0.05, **P < 0.01, ***P < 0.001. The Deep-Ocean Average line represents the inferred average flux over the entire deep-ocean based on Outridge et al.¹ and Zhang et al.³.

Full size image

The recent and historical Hg concentrations at Atacama were markedly higher than those in surface sediments in other oceanic regions remote from anthropogenic or hydrothermal sources, where Hg concentrations were usually < 80 ng g^−1 9,15. They were also higher than in other hadal and abyssal sites where mean concentrations ranged from 12 ± 6 to 132 ± 20 ng g⁻¹ (Fig. 3A)^{16,https://doi.org/10.1038/s41467-020-17045-3 (2020)." href="https://www.nature.com/articles/s41598-021-90459-1#ref-CR18" id="ref-link-section-d39636e961">18,19}.

The mean Atacama values were even higher than in some regions affected by anthropogenic activities. For example, the most contaminated deep-ocean site in the Mediterranean Basin exhibited a mean Hg concentration of 77 ng g^−1 13. Maximum concentrations in recent Atacama sediments, which ranged up to 401 ng g⁻¹ (see Fig. S2) are the highest reported from remote locations anywhere including in the Kamchatka Trench and abyssal plains (158 ng g^−1 16) and bathyal areas of the Southern Ocean (87 ng g^−1 15). They approach the maxima reported from some of the most contaminated shelf sites world-wide including the Bohai Sea, China (574 ng g⁻¹²⁹) and the Laurentian Trough, Canada (520 ng g^−1 11).

Plots of sediment Hg concentrations (A) and fluxes (B), and particulate organic carbon (POC) flux in hadal trench and abyssal plain sites world-wide. (Plots show mean ± SD values, for the top few cm of sediment cores or surface sediment grab samples. Study locations and Hg data sources in panel (A): 1—Marianna/Yap Trenches^{https://doi.org/10.1038/s41467-020-17045-3 (2020)." href="https://www.nature.com/articles/s41598-021-90459-1#ref-CR18" id="ref-link-section-d39636e1024">18}, 2—Marianna Trench¹⁹, 3—Kermadec Trench and abyssal plain (this study), 4—New Britain Trench¹⁹, 5—Kuril-Kamchatka Trench and abyssal plain¹⁶, 6—Atacama Trench and abyssal plain and slope (this study). In panel (B), the locations and sources for data points 3 and 6 are the same as in panel A, but for point 5 the flux is from one core on the Kuril Basin abyssal plain^{https://doi.org/10.1016/j.pocean.2019.102235 (2020)." href="https://www.nature.com/articles/s41598-021-90459-1#ref-CR17" id="ref-link-section-d39636e1044">17}. Linear regression equations are shown as dashed lines, and in (B) the regression was forced through the origin. POC flux data are from Longhurst²⁶ as tabulated by Stewart and Jamieson²⁴ and Jamieson^{https://doi.org/10.1017/CBO9781139061384 (2015)." href="https://www.nature.com/articles/s41598-021-90459-1#ref-CR35" id="ref-link-section-d39636e1060">35}.

Full size image

POM scavenging and sedimentation of Hg from the euphotic zone, which is ultimately sourced from atmospheric deposition, is likely the main source of Hg in deep-ocean sediments. This is evidenced by Hg isotope analyses of hadal sediments^{https://doi.org/10.1038/s41467-020-17045-3 (2020)." href="https://www.nature.com/articles/s41598-021-90459-1#ref-CR18" id="ref-link-section-d39636e1076">18}, ocean Hg-primary productivity modeling³, and geochemical associations between Hg and POM especially diatoms in particle traps and sediments^{15,https://doi.org/10.1016/j.pocean.2019.102235 (2020)." href="https://www.nature.com/articles/s41598-021-90459-1#ref-CR17" id="ref-link-section-d39636e1087">17,22}.

An analysis of the few Hg concentration and flux data available for hadal and abyssal sediments showed positive though non-significant correlations between estimated POC flux and sediment Hg concentrations and fluxes (Fig. 3) which are also suggestive of an upper-ocean source of POM-associated Hg.

There are two other possible sources or processes that can contribute to high Hg concentrations in deep-ocean sediments. First, post-depositional redistribution of Hg could occur following the decomposition of Hg-containing POM, releasing dissolved Hg to porewater. Its subsequent diffusion and adsorption onto authigenic Fe and Mn hydroxides in oxic surface sediments can result in Hg enrichment^9,12.

Diagenetic redistribution of Hg is typically negligible in lake and coastal sediments^11,30, but is less studied and could be important in ocean basins due to different sedimentation and mixing regimes^12,13. As porewater Hg concentrations were not measured in this study, we cannot directly estimate the magnitude of redistribution. Significant correlations between concentrations of solid phase Hg and reactive Fe (see Fig. S3), an association suggestive of Hg redistribution¹², were observed in four of nine Atacama cores (At1, At5, At7, At9; Table S2).

To test the effect of possible redistribution, we compared mean recent Hg concentrations with those in the whole cores. Whole core data represent the integrated averages of increased surface sediment concentrations and decreased lower core concentrations due to possible redistribution. Recent sediment Hg concentrations were only 16.7% higher on average than whole core concentrations at Atacama (188 vs 162 ng g⁻¹), and 1.9% higher at Kermadec (44.3 vs 43.5 ng g⁻¹). Higher Hg values in recent Atacama sediments but not in Kermadec could also be the result of elevated deposition of anthropogenic atmospheric Hg into the eastern Pacific Ocean, compared to lower levels of anthropogenic Hg in ocean waters near Kermadec³. While we cannot resolve the two alternate explanations, the absence of significant Hg-reactive Fe correlations at the majority of Atacama sites, and the small increase in recent Hg concentrations relative to whole cores at both trenches, suggests that any effect of redistribution on recent sediment Hg concentrations and fluxes was small.

A second possible Hg source in deep-ocean sediments is seafloor hydrothermal vents associated with mid-ocean ridges. Globally, vents are thought to contribute minor amounts of Hg to oceans, ≤ 100 t a^−1 1,22, but could be locally important. The Atacama study area is several thousand km east of the nearest known hydrothermal field³¹ indicating that, in this area with the highest Hg concentrations, hydrothermal influences are negligible.

The Kermadec trench is located relatively close to a hydrothermal field³¹, but no studies have been conducted on vent releases or seawater Hg concentrations in this region. However, there is little evidence of significant hydrothermal impacts on Hg concentrations in seawater^{https://doi.org/10.1029/2006GL026321 (2006)." href="https://www.nature.com/articles/s41598-021-90459-1#ref-CR32" id="ref-link-section-d39636e1160">32,33,34} or biota^{https://doi.org/10.1038/s41467-020-17045-3 (2020)." href="https://www.nature.com/articles/s41598-021-90459-1#ref-CR18" id="ref-link-section-d39636e1168">18} in the vicinity of vent plumes elsewhere, suggesting that the possibility is unlikely in this study. Overall, therefore, the Hg concentrations in these trench sediments are likely to predominantly reflect POM-associated transport of Hg from the euphotic zone.

Mean total Hg fluxes in recent sediments at Atacama (63.6 ± 30.8 µg m⁻² a⁻¹) were significantly higher (P < 0.001) than in historical sediments (44.4 ± 19.5 µg m⁻² a⁻¹), and also significantly higher (P < 0.001) than recent Kermadec sediments (10.4 ± 7.5 µg m⁻² a⁻¹; see Fig. 2B). There was no significant difference between average fluxes measured in trenches and in abyssal plain and slope sites at either location (P > 0.10). Combined mean recent and historical Hg fluxes were 50.6 ± 36.8 (median 39.1) µg m⁻² a⁻¹ and 35.9 ± 23.8 (median 31.5) µg m⁻² a⁻¹, respectively. A similar range of fluxes (30–44 µg m⁻² a⁻¹) was reported in post-1900 sediments in a dated core from the abyssal Kuril Basin^{https://doi.org/10.1016/j.pocean.2019.102235 (2020)." href="https://www.nature.com/articles/s41598-021-90459-1#ref-CR17" id="ref-link-section-d39636e1204">17}, supporting these findings.

To place these results in context, the relatively low recent Hg flux measured at Kermadec (10.4 µg m⁻² a⁻¹) is 6–15 times higher than the currently estimated deep-ocean average of 0.7–1.8 µg m⁻² a^−1 1,3.  Because the POC flux at Kermadec is similar to the average of all of the world’s trenches > 6.5 km depth²⁶, Kermadec may approximate the global average Hg flux in the hadal. The median recent flux across our study sites (39.1 µg m⁻² a⁻¹) was 22–56 times larger than the inferred deep-ocean average Hg flux, but the median includes Atacama with the highest POC flux in the world, and which therefore may have a substantially higher Hg flux than most other trenches. Overall, however, these results indicate that some hadal trenches and adjacent abyssal areas are intense focal points of Hg accumulation. The question arises as to what impact these high hadal fluxes might have on global deep-ocean Hg accumulation rates. [Emphasis added. ~ Ed.] 

To tentatively calculate the total hadal zone Hg flux, the few available data from hadal and abyssal sites (the present study, and Aksentov and Sattarova^{https://doi.org/10.1016/j.pocean.2019.102235 (2020)." href="https://www.nature.com/articles/s41598-021-90459-1#ref-CR17" id="ref-link-section-d39636e1238">17}) were used to approximate Hg fluxes for each trench system using the much larger global trench POC database. Individual trench estimates were summed to arrive at a total hadal Hg accumulation rate. POC flux data are available for 18 of the deepest trenches > 6.5 km depth, including Atacama and Kermadec^{24,https://doi.org/10.1017/CBO9781139061384 (2015)." href="https://www.nature.com/articles/s41598-021-90459-1#ref-CR35" id="ref-link-section-d39636e1245">35}, representing 92% of the global hadal area below 6.5 km. Applying a Hg:POC regression slope (± standard error, SE) of 16.1 ± 2.9 μg Hg g POC⁻¹ (see Fig. 3B), we arrive at a preliminary estimate of total Hg burial of 16.9 ± 1.0 t a^–1 in this zone (see Table S3). Scaling up to the much larger hadal area bounded by the 6 km isobath^{https://doi.org/10.1016/j.margeo.2014.01.011 (2014)." href="https://www.nature.com/articles/s41598-021-90459-1#ref-CR10" id="ref-link-section-d39636e1260">10}, the total Hg flux is estimated to be 73 ± 4.4 t a^–1. The error terms for the 18 trenches and for the total hadal zone were propagated from the SE of the Hg:POC regression slope applied to each trench’s POC data. We emphasize the uncertainty associated with the 73 t a⁻¹ value because of the few hadal or abyssal sites in total that have been sampled for Hg fluxes (N = 3), the large variation in fluxes between cores within the study trenches, and the relatively large confidence intervals associated with the Hg:POC regression (see Fig. 3B).

This study has substantially expanded the knowledge base concerning Hg flux in the deep-ocean. These results increase by six-fold the number of dated cores analyzed from this zone and provide the first estimates of hadal Hg flux, which is shown to be high at both study sites relative to the currently accepted deep-ocean average flux. Although still limited in number and geographical scope, the results can be placed in the context of the current estimate of deep-ocean Hg accumulation. Our first order estimate of 73 t a⁻¹ for hadal Hg burial represents 12–30% of the putative deep-ocean accumulation of 240–600 t a^−11,3, occurring within an area equal to 1.1% of the total deep-ocean. This estimate omits Hg flux in the much larger abyssal zone which is 85% of total ocean area^{https://doi.org/10.1016/j.margeo.2014.01.011 (2014)." href="https://www.nature.com/articles/s41598-021-90459-1#ref-CR10" id="ref-link-section-d39636e1290">10}. Given the evidence of similarly high Hg fluxes in some abyssal sediments (this study and Ref.^{https://doi.org/10.1016/j.pocean.2019.102235 (2020)." href="https://www.nature.com/articles/s41598-021-90459-1#ref-CR17" id="ref-link-section-d39636e1294">17}), and even higher fluxes in high productivity bathyal areas of the Great Southern Ocean¹⁵, it is possible that the total deep-ocean Hg accumulation is higher than is presently believed. Our results thus call for extensive additional sampling of the deep-ocean to better constrain these preliminary estimates and to improve global Hg modeling.