Sediments of the Adélie Basin were collected during the Integrated Ocean Drilling Program (IODP) Expedition from hole U1357B 318 in 2010. U1357B is located on the continental shelf off Wilkes Land at the Mertz Glacier polynya (region of open water surrounded by sea ice), Antarctica (66∘24.7990" S, 140∘25.5705" E), at about 1021.5 m water depth (Escutia et al., 2011) (Fig. 1). The total length of the recovered core is 170.7 m, corresponding to nearly the entire Holocene (Escutia et al., 2011). The core was sliced by 5 cm3 plastic scoops to 1 cm slices. Samples in the upper core (3.2–25.05 m b.s.f.) were taken at ∼40 cm intervals (a resolution of ∼20 years) and ∼400 cm intervals (a resolution of ∼200 years) in deeper sections (25.05–170.35 m b.s.f.), resulting in a total of 78 samples. Age data and age model (which is based on compound-specific 14C dating) were obtained from Yamane et al. (2014).

The sediment core is characterized by light and dark laminations which are undisturbed by sea-level changes or glacial erosion (Denis et al., 2006; Escutia et al., 2011). Light laminations correspond to spring seasons when light and high nutrient levels promote intense phytoplankton blooms and are mainly composed of biogenic materials (mostly diatom with minor abundance of silicoflagellates, sponge spicules, radiolarians, and foraminifers). Dark layers correspond to the summer/autumn season when sea ice has retreated, and nutrient levels are low. Dark laminations are composed of a mixture of biogenic and terrigenous materials resulting from summer production in open water, with glacial and subglacial inputs, respectively. High levels of primary production in surface water of this region, coupled with rapid fluxes of biogenic debris directly to the seafloor, led to high sedimentation rates of up to 2.0 cm yr-1 during the past 10 000 years (Escutia et al., 2011).

All samples were freeze-dried and ground using a glass pestle prior to geochemical analysis. Total Hg was determined by thermal decomposition followed by preconcentration of Hg on a gold trap and cold vapor atomic absorption spectrometry (CVAAS) Hg detection using a Milestone DMA-80 analyzer (US EPA Method 1998). The quality of the analysis was ensured by including a certified reference material (CRM) (Canmet LKSD-4 =190±17 ng g-1) alongside the analyzed samples. The average measured concentration for LKSD-4 was 197±11 ng g-1. Replicate analyses (n=20) were always within a relative standard deviation (RSD) of 10 % of the certified value.

The samples were analyzed for concentrations of silicon (Si), titanium (Ti), zirconium (Zr), sulfur (S), calcium (Ca), potassium (K), aluminum (Al), yttrium (Y), manganese (Mn), strontium (Sr), iron (Fe), lead (Pb), copper (Cu), zinc (Zn), arsenic (As), bromine (Br), nickel (Ni), chlorine (Cl), and rubidium (Rb) by energy dispersive X-ray fluorescence (ED-XRF). The calibration method, accuracy, and precision are described in detail in Cheburkin and Shotyk (1996). The CRMs (Canmet LKSD-4, NRC/CNRC-PACS-2, NRC/CNR-Mess-3, and NCS-DC75304) and replicates were measured in each set of samples for accuracy and precision control. Repeated analysis of CRMs gave an RSD of less than 10 % for Si, Al, Ca, Y, Sr, Zr, Br, and Rb; 6 %–15 % for Ti, K, Zn, S, Fe, Mn, and Pb; 6 %–19 % for Cl; 10 %–20 % for Ni; 9 %–14 % for Cu; and 14 %–22 % for As.

Principal component analysis (PCA) was applied to the major and trace element concentrations to identify processes controlling the variability of elements in the sediments. When there is a complex set of variables, PCA is used to reduce a large number of variables to a new set of artificial variables, called principal components. Each component includes variables with a similar down-core pattern. The principal components are then interpreted in terms of relevant geochemical processes that can control the variability of the major and trace elements in the sediments. The derived interpretation from PCA was then combined with the Hg data to examine the processes that could affect Hg accumulations. The analysis was performed on the standardized concentration data using Z scores (expressed in terms of standard deviations from their means). Regression analysis of the corresponding residuals was used to establish relationships between the abundance of elements, by considering Si concentration as an independent variable and other element concentrations as dependent variables. Correlation analysis and PCA were performed using the statistical software SPSS 25.0.

Concentration profiles and accumulation rates of Si, Al, K, Ti, S, Ca, Zn, Fe, Br, As, and Cl are shown in Figs. 2–5. The preindustrial geochemical record of Adélie Basin sediments is generally characterized by periodic-like variations in the relative abundance of major and trace elements. The records of element accumulation rates largely follow those of periodic-like variations of concentrations and show no significant trend with depth (except Cl).

Si has the highest concentration of all elements in the sediments. On one hand, Si is associated with the flux of terrestrial derived mineral components and on the other hand with siliceous phytoplankton, protozoans, protists, plant phytoliths, and sponge spicules (Croudace and Rothwell, 2015). Si is mainly biogenic in origin in Adélie Basin sediments, as well as dominated by diatoms (Escutia et al., 2011), and contribution of terrigenous Si is low. Therefore, it is used as a proxy for diatom abundance. The record of Si concentrations shows periodic-like variations by a factor of ∼2 between 21 % and 50 %, with a median of 33 %, corresponding to 70 % SiO2 or biogenic silica. Concentrations of Al, K, and Ti (as indicators of changes in the flux of lithogenic materials) range between ∼1.6 %–7.3 %, ∼0.37 %–1.11 %, and ∼716–1778 mg kg-1, respectively. S and Ca concentrations, which are associated with the biogenic productivity, vary between ∼0.13 %–0.87 % and 0.72 %–1.49 %, respectively. Ca concentration indicates that calcite-producing microorganisms are of minor importance in the Adélie Basin. Concentration of Zn, an important micronutrient for marine phytoplankton (Morel et al., 1994), ranges between ∼96 and 216 mg kg-1. Fe is another essential micronutrient for marine primary production (Smetacek et al., 2012) and biochemical processes of phytoplankton such as photosynthesis, respiration, and nitrogen fixation (Lohan and Tagliabue, 2018). Concentration of Fe varies between ∼1.05 % and 3.46 %, which is similar to other siliceous sediments but lower than the reported concentration in other ocean sediments (Chen et al., 1996). Fe concentrations increase at 66.45 m depth to the top of the core by a factor of ∼1.6 (from a median of ∼1.50 % below the 66.45 m depth to ∼2.40 % above the 66.45 m depth). This is attributed to the upward transport of Fe under anoxic conditions. Chlorine was found to be another major component in these sediments. Cl can go through biological pathways (incorporation into algae) and reach the sediments by the fast-sinking detritus (Leri et al., 2015). Concentrations vary between ∼1.3 % and 19 %, with a median of median 4.5 % and show a decrease from the top to the bottom of the core, which is likely attributed to the increasing mineralization of organic matter with age and the release of chloride through reductive dechlorination.

The PCA resulted in five components, explaining almost 82 % of the total variance (Table 1). The first component (Cp1) explains 33 % of the variance and shows large (>0.7) positive loadings of Mn, Ti, Rb, Zr, K, and Y and moderate positive loading of Fe. The second component (Cp2), which explains 20 % of the variance, is characterized by large positive loadings of Al, Si, S, and Cl and moderate positive loading of K and Ca. The third component (Cp3) explains 17 % of the variance and shows large positive loadings of Zn, Cu, and Ni and moderate positive loading of Fe. The fourth and fifth components (Cp4 and Cp5) account for 7 % and 5 % of the variance, respectively. Cp4 is characterized by high positive loadings for Hg and As and moderate negative loading of Pb. Cp5 shows positive loadings for Sr and Ca.

In general, results of PCA imply that opening and closing of the polynya and biological production are the most important factors influencing sedimentation in the Adélie Basin. This has been shown by loadings of elemental proxies for terrigenous and biological material inputs. Briefly, Cp1, which includes positive loadings of lithogenic elements, represents the variability of terrigenous inputs. Melting of ice releases trapped lithogenic material into the water and leads to the sinking of lithogenic particles and their sedimentation. Cp2 comprises loadings of elements of both biogenic and terrigenous sources. This component appears to reflect phytoplankton blooms and export of biological materials. After ice melt, when the ice is opening, favorable conditions for biological productivity lead to phytoplankton blooms and export of biogenic materials to the seafloor (Denis et al., 2006). Biogenic material is mainly opal because diatoms are a major component of blooms in the Adélie Basin (Escutia et al., 2011). The sinking of diatoms from the surface and their sedimentation can cause scavenging of elements during bloom time. Al shows positive loadings in this component rather than in Cp1. This, other than association with the flux of aluminosilicates material, can also be attributed to the scavenging of Al by diatom particles (Moran and Moore, 1992). Cl also shows loading in Cp2. The possible explanation for the observed covariation is that marine phytoplankton is rich in polyunsaturated lipids and can account as chlorination substrates (Leri et al., 2015). However, the organic C content of Adélie Basin sediments is generally low (between 1 wt % and 2 wt %), and we hypothesize that some of the Cl must be in an inorganic form trapped in sediments owing to high sedimentation rates. Cp3 is mainly characterized by elements that are associated with the organic fraction of diatom cells. This component appears to reflect the remineralization process like decomposition of organic particles during sinking. Trace elements associated with organic parts of cells can be released back into the water column during decomposition. Therefore, cellular locations of elements, i.e., opal frustules of diatoms or organic matter of diatom cells, created different components of Cp2 and Cp3. Cp4 consists of organic particle reactive metals, e.g., Hg and Pb. The possible explanation for not having these two particle reactive metals in Cp2 is that a portion of these two metals begins to enter the system after ice melting, while Cp2 shows scavenging of elements by diatoms that are already present in the water column. Since Pb and Hg are negatively correlated, this component cannot reflect a pollution signal. The reason for that could be the characteristic of Pb that decreases its impact in remote areas such as Antarctica. Atmospheric Pb is associated with particles and therefore is rapidly removed by wet and dry deposition. Moreover, the Southern Ocean circumpolar vortex that isolates Antarctica from the other continental landmasses in the Southern Hemisphere will further limit Pb transport to Antarctica. Covariation of Ca and Sr in Cp5 represents sedimentation of planktonic foraminifera, which appears to be of minor importance here.

Cp1 explains 33 % of the variance and accounts for much of the variability/process which controlled the geochemical composition of these sediments. However, the high concentration of Si and low concentrations of terrigenous elements imply that in an environment like Adélie Basin, with extremely high productivity, input of lithogenic materials is changing while different diatom taxa are always present in the system, e.g., as sea-ice-associated and open-ocean diatom (Escutia et al., 2011).

It has been shown before that seasonal blooms in ocean surface waters result in temporary variable fluxes of elements to the deep ocean (Fowler and Knauer, 1986; Michel et al., 2002; Pilskaln et al., 2004). Although the core was not sampled at a 1-year resolution in our study, we suggest that the fluctuations of elemental concentrations are likely related to the seasonal blooms and variation of sea surface conditions like ice melting and freezing and its subsequent biological or terrestrial material exports. Sampling in light (associated with spring) or dark (associated with summer/autumn) laminae, which contains different amounts of biogenic or terrestrial materials, can cause the observed variations (see Sect. 2.1).

The element concentrations are comparable to other published sediment data, while the accumulation rates are much higher than other reported values. The existence of rapidly settling particles in the Adélie Basin can explain the high element accumulation rates. When nonessential elements and essential elements show high accumulation rates, it is tempting to suggest that most elements in the water column of Adélie Basin are subjected to removal by intense phytoplankton blooms through consumption or scavenging. This agrees with the study of Fowler and Knauer (1986) that demonstrated the role of large particles in the transport of elements through the oceanic water column. Aggregations of diatoms, which create large particles (Turner, 2015), sink to the seafloor and can create a space in which elements can be trapped (Shanks and Trent, 1979). This enhances removal of elements from the water column and their sedimentation as well.

In the preindustrial period, i.e., from the bottom of the core at ∼170 to 2.80 m depth (8600 years ago to ∼1850 CE), the Hg record shows no obvious trend with depth but rather periodic-like variations. Hg concentrations fluctuate by a factor of about 2 between 12.6 and 21.1 µg kg-1 within 170–137 m depth and between 21.7 and 44.6 µg kg-1 within 137–2.80 m depth of the core, with two more pronounced peaks at around 9.99 and 8.20 m depth (Figs. 2 and 3). The lower concentration of Hg within 170–137 m depth of the core is probably attributed to cooler conditions in Adélie Basin (Crosta et al., 2007) and sea ice cover during this period. Hg accumulation rates in the preindustrial period (Figs. 4 and 5) largely follow the Hg concentration record, with periodic-like variations and a median of 556 µg m-2 yr-1, which largely surpass the reported Hg deposition rates to the oceans (Mason and Sheu, 2002).

The high preindustrial Hg accumulation rates in Adélie Basin sediments cannot be explained by preindustrial atmospheric Hg depositions alone, which did not exceed 20 µg m-2 as recorded in an Antarctic ice core (Vandal et al., 1993). Therefore, nonatmospheric sources, such as dissolved water-phase Hg or terrestrial inputs, are needed for the observed high Hg enrichment in these sediments.

To identify driving forces behind the variations in Hg accumulation we used our PCA results. PCA demonstrated that two main processes, i.e., biogenic productivity and lithogenic inputs, controlled the flux of elements to Adélie Basin sediments. The component scores, which illustrate the depth dependency of the extracted components, are characterized by seesaw patterns throughout the entire core. This indicates different contributions of biogenic and terrigenous inputs most likely associated with spring and summer/autumn seasons, respectively.

The variance of Hg was not captured by Cp1, Cp2, or Cp3. Hg instead forms a group on Cp4 together with positive loading of As and negative loading of Pb. The absence of significant loading of Hg on Cp1, Cp2, and Cp3 (Table 1) and the lack of similarity between component scores and Hg concentrations (Figs. 6 and 7), in the preindustrial period, indicate that Hg fluxes are not significantly influenced by changes in lithogenic inputs through ice melting. These results further indicate that changes in the contribution of biogenic material cannot explain the variation in Hg accumulation in the sediments.

Although, the high Hg accumulation rates observed in the Adélie Basin sediments could not be explained by atmospheric Hg deposition, but we expect that the Hg flux from the atmosphere will probably increase during algae blooms attributed to the removal of dissolved Hg through scavenging by algae. Hg removal from the upper water column by diatom organic matter will also likely decrease Hg re-evasion to the atmosphere as previously assumed in model studies (Soerensen et al., 2014, 2016).

We have hypothesized that the high Hg enrichment in Adélie Basin sediments has been caused by scavenging of dissolved water-phase Hg by a large amount of fast-sinking algal debris. To prove this hypothesis we calculated the maximum amount of Hg which could be scavenged by a single bloom event using the Hg concentration of 271±78 pg L-1 in Antarctic bottom water, as determined by Cossa et al. (2011). The amount of Hg in a water column of 1 m2 and 1000 m depth would then amount to 271±78 µg. This means that only about two to three algae blooms and scavenging events per year are necessary to obtain the average Hg accumulation rate in Adèlie Basin diatom ooze sediments, i.e., 556±137 µg m-2 yr-1. This appears to be likely taking into account that bloom events are frequent during Antarctic summer and that the sinking speed of diatom agglomerates at Adélie Basin is high (100–400 m d-1) (Jansen et al., 2018). The annual cycle of water mass transformation beneath the Mertz Glacier polynya system (Williams et al., 2008) and the exchange of water masses can rapidly refill the Hg inventory in the water column after a scavenging event. While algal bloom is a local event at the surface water of the Adèlie Basin, the exchange of water masses, which have not been affected by algae blooms, could refill the Hg inventory in the water column (Fig. 8).

The main reason for not finding any statistical relation between Hg and biogenic materials is that the amount of algal material during blooms is always large and therefore not a limiting factor for the scavenging of Hg. There has always been excess algal material within or passing through the water column to scavenge all water column Hg. Thus, we assume that nearly all Hg in the water column is removed through scavenging during diatom blooms but that Hg scavenging events occur less frequently during winter and summer/autumn seasons when primary productivity is lower and open ice expansion is at its maximum. Similar to other elements, the periodic-like variations observed in the preindustrial Hg record. The influence of periodically climatic changes of phytoplankton activity on periodical changes of Hg content was suggested before for the Caribbean Sea (Kita et al., 2013). It is likely that the periodic-like variations observed in the Adèlie Basin Hg record are also attributed to the seasonal export of biological materials. This could be in line with findings of Hg levels in water (Canário et al., 2017), which suggested that different development stages of a phytoplankton bloom lead to different amounts of dissolved Hg taken up by phytoplankton. However, investigation at seasonal resolution is needed to further confirm our observations and hypotheses.