Introduction
Environmental sources of mercury (Hg) can be either natural or anthropogenic. The natural concentration of Hg in rocks, sediments and soils varies between 0.08 to 0.4 mg kg-1 (Gu et al. Reference Gu, Zhou, Wong and Gan1998). Elemental Hg is rare in nature, although there are several Hg-containing minerals, cinnabar (mercuric sulphide) being the most common in the terrestrial crust (Bowen Reference Bowen1979). Burning fossil fuels is the main anthropogenic source of Hg in the atmosphere (Lacerda Reference Lacerda1997).
Although Antarctica is the most isolated continent, it is not free of contamination from human activities. The contribution of anthropogenic Hg to the continent may originate from distant sources through atmospheric transport across low latitudes or by ocean currents. Some sources of Hg may also arise from human occupation of the continent by incineration of garbage from scientific stations as well as from paints, fuels and sewage (Santos et al. Reference Santos, Silva-Filho, Schaefer, Albuquerque-Filho and Campos2005).
There are few studies concerning the presence of Hg in the Antarctic, with most related to atmospheric transport and its environmental fate (Vandal et al. Reference Vandal, Fitzgerald, Boutron and Candelone1993, Bargagli et al. Reference Bargagli, Monaci and Bucci2007). Springtime Hg depletion events, first observed in the Arctic, were also proven to happen in Antarctica. During these events, elemental Hg, the main species present and transported in the atmosphere, is photochemically oxidized by reactive halogens in the Polar Regions, after which enhanced deposition as ionic Hg species takes place (Ebinghaus et al. Reference Ebinghaus, Temme, Einax, Löwe, Richter, Burrows and Schoeder2002, Witherow & Lyons Reference Witherow and Lyons2008, Nguyen et al. Reference Nguyen, Kim, Shon and Sungmin2009).
Hg and other metals have been extensively studied in biological matrices of marine organisms, but little data from terrestrial Antarctic environments are available. Santos et al. (Reference Santos, Silva-Filho, Schaefer, Sella, Silva, Gomes, Passos and van Ngan2006) quantified the total Hg (HgT) in soils and sediments on King George Island and found very low values, in the range of 15–30 μg kg-1. Crockett (Reference Crockett1998), studying red and gley soils around McMurdo Station (Hut Point Peninsula) reported background levels of Hg < 40 μg kg-1. Sun et al. (Reference Sun, Yin, Liu, Zhu, Xie and Wang2006) also quantified HgT in seal hair from a lake core and proposed that the accumulation of Hg in this material is related to the five main periods of global gold mining in the last two thousand years.
It is well known that the determination of HgT is insufficient for understanding its biogeochemical cycle, whereas research on the environmental risks and on Hg speciation in different matrices is extensive. Sequential chemical extractions, X-ray atomic absorption (Sladek et al. Reference Sladek, Gustin, Biester and Kim2002, Kim et al. Reference Kim, Bloom, Rytuba and Brown2003) and liquid or gas chromatography coupled with many techniques of detection (e.g. mass spectrometry, solid-phase Hg pyrolysis and Cold Vapour Atomic Absorption Spectrometry (CV-AAS)) are examples of techniques currently used to investigate Hg speciation. The identification of elemental Hg by X-ray absorption spectroscopy is difficult (Vandal et al. Reference Vandal, Fitzgerald, Boutron and Candelone1993), as its detection limit is too high (Kim et al. Reference Kim, Bloom, Rytuba and Brown2003). Techniques for determining the specific short-chain organomercury compounds usually involve many steps and sophisticated equipment. According to Sladek et al. (Reference Sladek, Gustin, Biester and Kim2002), solid-phase Hg pyrolysis and CV-AAS is the most appropriate technique for identifying elemental Hg as well as for differentiating it from ionic Hg bound in solid matrices. This technique identifies Hg species by controlled heating and comparison with thermodesorption patterns of known Hg substances and has been commonly used to investigate Hg phases in soils and sediments (Windmöller et al. Reference Windmöller, Wilken and Jardim1996, Biester & Zimmer Reference Biester and Zimmer1998, Biester et al. Reference Biester, Gosar and Covelli2000).
This study aims to provide information about the presence of Hg in soils from Maritime Antarctica. This was accomplished by the determination of HgT in different soil types from the Fildes and Ardley peninsulas (King George Island) and correlating these data with physical and chemical characteristics of the soils. We also studied the metal speciation using solid-phase Hg pyrolysis and CV-AAS, as well as by chemical fractionation.
Materials and methods
Soil sampling and preparation
Eighty-six soil samples (ranging from 0–80 cm in depth) were collected on the Fildes (F samples) and Ardley (A samples) peninsulas on King George Island, South Shetland Islands. Most profiles were collected up to 50 cm due to shallow permafrost occurrence below this depth. The soil was collected and stored in plastic bags at -5°C.
Ornithogenic ecosystems are observed in many areas of the Ardley Peninsula. They are formed by deposition and accumulation of bird excreta, mostly by penguins, known as guano (Simas et al. Reference Simas, Schaefer, Melo, Albuquerque-Filho, Michel, Pereira, Gomes and Costa2007). In these soils, we also find penguin remains including bones, feathers, and decomposing bodies. Hence, ornithogenic soils have high organic matter (OM) content, high phosphate concentrations and show formation of secondary phosphates with metals released from the weathering rock matrix (Kuo Reference Kuo1996, Simas et al. Reference Simas, Schaefer, Melo, Albuquerque-Filho, Michel, Pereira, Gomes and Costa2007, Reference Simas, Schaefer, Albuquerque-Filho, Francelino, Fernandes Filho and Costa2008). Biogenic soils here refers to soils under the influence of other marine animals, such as seals (elephant seals, Weddell seals, crabeaters, fur seals).
The non-ornithogenic soils considered in this study are mainly composed of decomposition products of basalt and andesite-basalt mineral rocks with no apparent contribution of organic matter from birds or marine animals, but show development of mosses on the soil surface.
For comparison, we collected 20 ornithogenic soil profiles, one biogenic soil profile and nine non-ornithogenic soil profiles, as shown in Appendix A. The geographical location of soil profiles are also presented in the same appendix.
Soil samples were air-dried, sieved through a 2 mm sieve and ground in a tungsten mill to produce a sample with < 200 mesh particle size, used for all subsequent analysis. The samples were taken at a depth where it was possible to sample “soil”, below these depths, only fragments of rock or permafrost were present. This is why different sample depths had to be acquired for the profiles. Because it was not possible to visually differentiate the horizons in the soil, we established equal (20 cm) intervals of depth to test, as defined in Appendix B by the letters, a, b, c, d and so on, to ensure the observation of gradual changes within the soil profile.
Soil characterization analysis: total phosphorus, available phosphate and carbon, hydrogen and nitrogen determination
These analyses were performed in all samples. Total phosphorus (Ptotal) was determined by X-ray fluorescence spectrometry using an EDX 720 instrument (Shimadzu, Japan). A certified reference material, GBW07408 (soil), was also analysed to evaluate the accuracy. The experimental result was Ptotal = 0.18% (n = 1), and the certified values is (Ptotal = 0.18 ± 0.01%). Available phosphate (Pavailable) was extracted using a Mehlich-1 solution (H2SO4 0.05 mol l-1 + HCl 0.05 mol l-1) and was quantified by the ascorbic acid method, according to Kuo (Reference Kuo1996). Total C, H and N (CHN) content were determined using a Perkin-Elmer CHNS/O model 2400 CHNS. All results obtained for the measured cystine standard were between 98.9 and 106.73%, showing good accuracy for this method.
HgT determination
The determination of HgT was performed on all collected samples by direct analysis using a Milestone Direct Mercury Analyser (DMA-80). The sample masses ranged between 100 and 300 mg. The sample was subjected to a heating program with the following steps: drying at 200°C, pyrolysis at 650°C where the sample matrix burned and the released Hg is carried by a stream of O2 and retained in a gold trap. Finally the gold trap is heated and the Hg is released into an atomic absorption cell. The great advantage of this equipment is that it precludes a sample extraction step, resulting in less possibility for sample contamination. The reference materials GBW-08301 (river sediment) and IAEA-336 (trace elements in lichens) with Hg concentrations 220 ± 40 ng g-1 and 200 ± 40 ng g-1, respectively, were used to evaluate the accuracy of the method. The certified reference material IAEA-336 was used because we have some soil samples with large amounts of algae and lichens, observed during sample collection. Examples are samples A2a, A2e, and A11a.
The detection limit (LOD) was calculated according to IUPAC (International Union of Pure and Applied Chemistry) recommendation, by taking three times the ratio of the standard deviation of ten independent blanks and the slope of analytical curve. For the quantification limit (LOQ), the same ratio was multiplied by ten. The results of LOD and LOQ were, respectively, 1.04 and 4.32 ng g-1.
The analytical technique used enable differentiation of the different sample types. The equipment used has three cells with different optical paths for readings at different concentration ranges. Therefore it provides three analytical curves, cell 0, cell 1 and cell 2. The data curves were for cell 0, concentration range 0–12 ng, 7 points, equation curve Y = -0.0026 X2 + 0.0907 X + 0.0008 and r 2 = 0.9999; for cell 1, concentration range 0–20 ng, 10 points, equation curve Y = -0.0006 X2 + 0.0421 X - 0.0025 and r 2 = 0.9962; and for cell 2, concentration range 50–1000 ng, 5 points, equation curve Y = -0.0000002 X2 + 0.0008 X + 0.0018 and r 2 = 1. Since the adjustment is not linear the number of points is large enough that the value of r 2 is suitable for quantitative analysis (at least two nines), as can be seen. Virtually all the readings of the samples fell into the equations of cell 0 and cell 1, only the readings of the samples of seal hair fell into the cell 2. For the calculation of LOD and LOQ was made linear fit of the lowest points of the curve of the cell 0, whose slope was 0.07542. Results of recovery for reference materials varied from 99–104%, with an average precision of 4%. The results of the two reference materials were as follows: GBW-08301 (river sediment), certified value 220 ± 40 ng g-1, experimental value 223 ± 4 ng g-1 (n = 3); and IAEA336 (trace elements in lichens) certified value 200 ± 40 ng g-1, experimental value 210 ± 6 ng g-1 (n = 3). The precision obtained was good, the relative standard deviation of all results (Appendix B) were in the range of 0.17–11%, with a mean value of 4.2%.
Statistical data analysis
Data were analysed using STATISTICA, version 6.0, for linear correlations (Pearson correlation) between the studied physico-chemical parameters and Hg contents (variables). Multivariate analysis through hierarchical cluster analysis (HCA) was performed by MINITAB, version 14, to classify the samples according to the values of a set of variables and to generate dendrograms. Ward's linkage method and Euclidean distance were used. The data were mean centered and autoscaled to a variance of 1. Only the samples with all parameters analysed were included in the statistical analysis.
Hg chemical fractionation
Chemical fractionation of Hg was performed on a subset of samples with expressive Hg content: F9a, F11b (also analysed by solid-phase Hg pyrolysis), A28d and A28a, which corresponds to the surface and subsurface layers, primarily consisting of penguin guano. The F11b sample was analysed in two ways: i) fresh, or ii) only seal hair present in the sample. This method, proposed by Bloom et al. (Reference Bloom, Preus, Katon and Hiltner2003), was developed specifically to analyse the mobility of Hg in matrices of soils and sediments. The following extractors were used: (S1) deionized water (30 ml), (S2) 0.1 M CH3COOH + 0.01 M HCl (40 ml), (S3) 1 M KOH (40 ml), (S4) 12 M HNO3 (40 ml) and (S5) aqua regia (1 HNO3: 3 HCl) (10 ml).
An initial sample mass of 3.0 g was used. It was shaken for 18 ± 4 hours in a shaker at a speed of 30 rpm, with each extractant. Each extractation step was centrifuged for 20 mins at 3000 rpm, and the supernatant liquid was analysed by a Direct Mercury Analyser (DMA). The solid residue remaining from each step was washed twice with deionized water before the addition of the next extractant.
It was not possible to analyse samples with Hg low levels (< 100 ng g-1) because the extracts obtained in the fractions showed Hg concentrations below the detection limit for our instruments.
Hg species determination by solid-phase Hg pyrolysis and CV-AAS
The samples were analysed by a solid-phase Hg pyrolysis and CV-AAS system described in previous works (Valle et al. Reference Valle, Santana, Augusti, Egreja-Filho and Windmöller2005, Reference Valle, Santana and Windmöller2006). Sample F9a, (collected at the soil surface) with ornithogenic influence, and sample F11b, a surface A horizon of a profile collected on a marine terrace and containing a visible abundance of seal hair, were chosen because of their high Hg concentrations (215 ± 4 and 256 ± 4 ng g-1, respectively). Sample F11b was analysed in three ways: i) fresh, ii) free of seal hair, and iii) only seal hair. Separation of the seal hair was performed manually with tweezers. An atomic absorption spectrometer (CG Analytica, GBC Model 380) coupled to a thermodesorption furnace was used to evaluate the Hg phases present in the matrix. This system heats the sample from ambient temperature to about 600°C at a constant heating rate (33°C min-1) (Windmöller et al. Reference Windmöller, Wilken and Jardim1996, Valle et al. Reference Valle, Santana, Augusti, Egreja-Filho and Windmöller2005). The vapours released are brought to a cell of an atomic absorption detector by a nitrogen stream. The graphics obtained in units of absorbance as a function of temperature are called thermograms. Sample masses up to 3 g were used in the analysis with at least three replicates of each sample. The differentiation between Hg0 and Hg2+ present in the samples was made by comparison of these profiles with others from previous studies using the same equipment and the same analytical conditions (Valle et al. Reference Valle, Santana and Windmöller2006).
Results and discussion
Soil characterization
The soil samples contained total C, H, N content ranging from 0.09–23.23%, 0.23–6.15% and 0.05–9.12%, respectively. The ornithogenic soil profiles showed the highest C and N values (Appendix B), as expected from previous studies (Michel et al. Reference Michel, Schaefer, Dias, Simas, Benites and Mendonça2006, Simas et al. Reference Simas, Schaefer, Melo, Albuquerque-Filho, Michel, Pereira, Gomes and Costa2007). The heavy use of the terrestrial environment by penguins every year results in high deposition of organic material and nutrients (guano and bird remains) containing especially C, P and N. The total P content can be used to indicate the degree of ornithogenic influence in the local soil environment. In the studied soils, the total P content ranged from 249–89 461 mg kg-1 which reflects soil development, especially with regards to phosphatization (Simas et al. Reference Simas, Schaefer, Albuquerque-Filho, Francelino, Fernandes Filho and Costa2008).The other soil samples, with less ornithogenic influence, showed much lower C, N and P contents, as expected.
The organic matter derived from penguin guano is rich in phosphates and goes through a process of mineralization, enhancing the chemical weathering process by acidic reaction with the underlying rock, resulting in the formation of secondary phosphate minerals (struvite, NH4-taranakite for example) (Simas et al. Reference Simas, Schaefer, Melo, Albuquerque-Filho, Michel, Pereira, Gomes and Costa2007).
The Pavailable contents ranged from 0.2–3182 mg kg-1 (Appendix B). This large variation is due to several factors including ornithogenic composition, type of P-mineral constituents and local soil drainage conditions. The highest Pavailable value found in non-ornithogenic soils was 111.2 mg kg-1 (sample F5b). This value is much lower than values obtained for the ornithogenic profiles and is indicative of decreased phosphorous availability due to low Ptotal content in the soils. In the ornithogenic soils, high levels of P are able to leach into deep layers due to dissolution of primary bone apatite in the topsoil and re-precipitation (Simas et al. Reference Simas, Schaefer, Albuquerque-Filho, Francelino, Fernandes Filho and Costa2008). The degree of P availability also depends on the soil pH and redox potential because NH4+, K+, Mg2+, Ca2+, Al3+ and Fe3+ phosphates have very different solubility constants, with lower solubility following the order presented. Greater P availability can suggest the presence of NH4+. The local drainage is also an important factor in P content, as secondary phosphates are usually concentrated above the permafrost table, which prevents further downward leaching. This phenomenon can be observed in the profiles for samples F13, A2 and A9 (Appendix B). Similar behaviour can be observed at intermediate depths in sample A2 (Ptotal and Pavailable) due to movement of the impermeable ice layer during the thawing season (Fig. 1).
Fig. 1 Concentrations of Ptotal and Pavailable in the profile for sample A2.
Pearson correlation analysis of data from Appendix B showed no significant correlation between Ptotal and N (0.16, P < 0.05 and n = 72), as expected, because the guano present in ornithogenic soils is the main source of the two elements, suggesting that the loss of NH4+ should be more intense than PO43-, which forms insoluble compounds with Al3+ or Fe3+. The low correlation of C with Ptotal (0.43, P < 0.05 and n = 72) suggests an additional input of organic carbon from plant material in addition to the decomposition of dead animals. In some soils (samples F6, F9, A1, A2, A4 and A13) the presence of non-decomposed mosses corroborates input of organic matter from plant sources. Wang et al. (Reference Wang, Wang, Wang and Sun2007) reported that the three major types of vegetation on Ardley Island, namely coprophilic algae, moss and lichen, showed variations in relative abundance according to fluctuations in the penguin populations. A moderate penguin population was observed to be favourable for algae and moss growth, while lichen populations were found to decrease whenever penguin population increased, and vice versa.
HgT determination
The Hg content in soils varied between 4.3 ± 0.2 and 256 ± 4 ng g-1 (Appendix B). These values are considered normal relative to average levels for uncontaminated global soils (200 ng g-1) (Horvat Reference Horvat1996), but they are the highest reported in the literature for soils in Antarctica to date. Siegel et al. (Reference Siegel, Siegel and McMurtry1980) reported values for Hg concentration in soils from Ross Island, Antarctica, from 3.1–7.1 ng g-1. In 2005, Reference Bargagli, Agnorelli, Borghini and MonaciBargagli et al. reported values of Hg content in soils from Victoria Land, Antarctica, in the range of 12–86 (soil fraction < 0.25 mm), and in 2007, they reported values in the range of 3.7–22 ng g-1 (soil fraction < 2 mm). In 2005, Santos et al. reported Hg concentrations in soils from Admiralty Bay, King George Island, close to the levels found in the lower crust (21 ng g-1). However, there has not been a large amount of research measuring Hg concentrations in large numbers of soil samples from Antarctica. Our results may represent a preliminary assessment of natural values for volcanic soils with and without ornithogenic influence in Maritime Antarctica.
The ornithogenic profiles showed a clear trend of higher Hg values (from 6.3 ± 0.4 to 215 ± 4 ng g-1), which may be due to a variety of sources: i) the excreta of seabirds, which feed on fish and krill that bioaccumulate the metal, ii) the decomposition of dead birds, releasing bioaccumulated Hg to the soil, and iii) the atmospheric deposition of Hg transported by air currents from lower latitudes, where the largest anthropogenic emissions occur (springtime depletion events, Schuster Reference Schuster1991, Ebinghaus et al. Reference Ebinghaus, Temme, Einax, Löwe, Richter, Burrows and Schoeder2002, Bargagli et al. Reference Bargagli, Monaci and Bucci2007, Witherow & Lyons Reference Witherow and Lyons2008, Nguyen et al. Reference Nguyen, Kim, Shon and Sungmin2009). Bargagli et al. (Reference Bargagli, Monaci and Bucci2007) analysed sediment cores and did not observe an increase in concentrations of Hg in recent surface sediments from ice-free areas of Victoria Land, which represents deposited Hg from anthropogenic and natural emissions from lower latitudes.
Regardless of the source, the ornithogenic soils contain large amounts of organic carbon, which can complex Hg and lead to its accumulation. The ability of organic matter to complex Hg in soils and sediments is already known and has been observed in many ecosystems (Schuster Reference Schuster1991). This affinity for complexation with OM of both animal and vegetable origin is due to the presence of carbonyl groups and protein-derived material containing nitrogen and sulfur, all of which can bind with the metal.This Hg-OM bond has been considered one of the most important in the case of soils and sediments, especially where clay minerals are present, as they are known to be metal and organic matter retainers (Bloom et al. Reference Bloom, Preus, Katon and Hiltner2003).
Conversely, non-ornithogenic profiles showed lower Hg content (from 4.3 ± 0.2 to 43 ± 3 ng g-1), when compared with global Clarke values of Hg (86 ng g-1) (Ronov & Yaroshevsky Reference Ronov and Yaroshevsky1972) and average values for the continental crust (40 ng g-1) (Wedepohl Reference Wedepohl1995), indicating that there has been no contamination in these soils. The lowest Hg values (4.3 ± 0.2 ng g-1) were found in the shallow lithic soils in samples F12, F14, and F15, all of which are characterized by a very limited degree of weathering.
The exception to the above conclusion is the profile for sample F10, which is not ornithogenic, but contains high organic matter content from seal (elephant seals, Weddell seals, fur seals) excreta and hair. The Hg contents in F11 were the highest value found in all the soils studied (256 ± 4 ng g-1).
Santos et al. (Reference Santos, Silva-Filho, Schaefer, Sella, Silva, Gomes, Passos and van Ngan2006) studied the Hg distribution in different materials from the nearby Keller Peninsula (King George Island) and found low levels of Hg (15–30 ng g-1) in soil and sediments, with very low Hg content in vegetation, invertebrates and fish. The only high Hg concentrations (2060 ng g-1) were found in feathers and mammal hair, indicating biomagnification in the Antarctic ecosystem. These materials have been suggested as possible biomonitors for that region (Santos et al. Reference Santos, Silva-Filho, Schaefer, Sella, Silva, Gomes, Passos and van Ngan2006).
The Pearson correlation between HgT and C was found to be significant but moderate (0.57, P < 0.05 and n = 72), but the graphics in Fig. 2 clearly shows that the profile of Hg and organic matter content is very similar in the ornithogenic soils studied. This correlation is clear in all profiles (except F7) and especially in the profiles with the largest number of samples, such as F9, F11, A2, A6 and A8. For these soils, one can see a decrease in Hg content with increasing depth; when there is a sharp decline in carbon content, there is also a sharp drop in the amount of the metal. The presence of greater amounts of Hg at the surface of ornithogenic soils indicates a recent deposition of the metal, probably from the contribution of guano and from the remains of dead animals that bioaccumulate the metal.
Fig. 2 Mercury and organic carbon concentrations as a function of depth in all ornithogenic profiles and one biogenic profile.
Fig 2 (continued)
This reduction in Hg content with increasing depth was not observed in the non-ornithogenic soils as shown in Fig. 3. In most of these cases, there was an increase in element content with increasing depth, suggesting that the Hg deposited on the surface was leached to deeper soil layers. One cannot exclude that Hg in surface soil is partly released as gaseous Hg. As described by Simas et al. (Reference Simas, Schaefer, Albuquerque-Filho, Francelino, Fernandes Filho and Costa2008), these basaltic or andesitic soils are in general poorly developed, but may be covered by mosses and lichens, which are a source of organic matter, allowing for A horizon formation. Figure 3 shows the correlation of Hg content with organic matter, although the trend is not as clear as was seen for the ornithogenic soils.
Fig. 3 Mercury and organic carbon concentrations as a function of depth in non-ornithogenic profiles.
Fig 3 (continued)
The high correlation observed in ornithogenic soils between Hg and OM indicates that bioaccumulation of Hg in marine animals is probably the main source of Hg in these terrestrial ecosystems. This means that the metal concentration found in these soils indirectly depends on the processes of oxidation from Hg0 to Hg2+, including methylation of the metal in the environment. In the case of ornithogenic soils, this contribution obviously is high, but in non-ornithogenic soils this seems likely to be the predominant Hg source. The contribution due to atmospheric deposition cannot be discounted, but it is a more important contributor in non-ornithogenic soils.
The analysis of seal hair showed a high Hg content of 1795 ± 56 ng g-1, in agreement with values reported by several authors. Santos et al. (Reference Santos, Silva-Filho, Schaefer, Sella, Silva, Gomes, Passos and van Ngan2006), for example, found 2060 ng g-1 of Hg in seal hairs from the Keller Peninsula, and Sun et al. (Reference Sun, Yin, Liu, Zhu, Xie and Wang2006) found 1740 ng g-1 of Hg in the same matrix from the Fildes and Ardley peninsulas. Both attributed these high values to biomagnification.
The statistical analysis of all variables (presented in Appendix B by HCA), using ward linkage and Euclidean distance, showed the formation of two large groups of samples, the non-ornithogenic soils (group A, Fig. 4) and ornithogenic soils (group B, Fig. 4). The branch B′ shown in Fig. 4 grouped ornithogenic soil samples with the highest concentrations of Hg, C and H. These results made clear the importance of physical and chemical processes related to the sea-land transfers promoted by marine animals (penguin and seals) to the geochemistry of Hg in this region.
Fig. 4 Dendrogram of data (samples) from Appendix B.
The Polar Regions are recognized as important sinks for the long-range transport of Hg derived from natural and anthropogenic sources at low latitudes. Bargagli et al. (Reference Bargagli, Monaci and Bucci2007) studied Antarctic ecosystems and suggested that Hg brought from low latitudes is deposited at the poles and accumulates in soils, mosses and lichens in ice-free areas of the Antarctic continent. In the same study the authors reported that due to melting in the summer, the metal is released from the crystalline lattice of ice and interacts in plankton, benthic mats, cyanobacteria and soil OM, resulting in high correlations between Hg and organic carbon.
Hg chemical fractionation in soil and seal hair samples
Results from chemical fractionation can provide information about the availability of Hg to the environment. Extraction of labile phases S1 and S2, which solubilize Hg weakly bound in the matrix, showed low values (< LQ to 4.69 ng g-1), with only one sample (A12a) containing an environmentally significant value over the two phases (15%). A12a is a guano-rich surface sample, in which not enough time had elapsed for complexation and retention of Hg with insoluble OM. Other ornithogenic soils with a greater degree of OM mineralization (F9a) show a greater interaction of Hg with OM (higher Hg extraction in S3, Table I).
Table I Results of Hg (μg kg-1) sequential extraction of the samples F9a, F11b, A12a, A12d and seal hair.
LQ = quantification limit; detection limit = 1.04 ng g-1.
Sequential extraction applied to seal hair samples confirms that the method is effective for the extraction of Hg bound to OM (step S3) because 100% of the Hg was extracted (Table I). The highest percentage of Hg bound to OM was found in the marine terrace soil F11b (88%) because this sample was richest in seal hair, as seals breed at this site during the summer. We believe that OM from seal hair is more difficult to break down and mineralize than that derived from faunal excreta. The mineralized excreta easily reacts with the bedrock, forming insoluble secondary minerals such as Al/Fe phosphate, which probably incorporate Hg into their structures. This phase is well represented by extracting step S4, in which HNO3 solubilizes phosphates. For samples A12a and A12d, deeper layers showed a greater amount of Hg retained in the S4 extract, indicating the formation of insoluble materials that retain Hg in those soils. No sample showed large amounts of residual Hg (S5).
Solid-phase Hg pyrolysis and CV-AAS
The results of solid-phase Hg pyrolysis and CV-AAS (Fig. 5) show that all Hg is present in the form of Hg2+ because the absorbance peaks appeared only above 200°C. Studies with standards of Hg0, Hg22+ and Hg2+ salts, with the same equipment and under the same analytical conditions described here, showed Hg0 release in the range of 20–100°C and Hg22+ release as a narrow peak close to 130°C, whereas other species of Hg2+ release at temperatures higher than 200°C (Fig. 6). It has been found that Hg bound to OM usually releases at higher temperatures (Valle et al. Reference Valle, Santana, Augusti, Egreja-Filho and Windmöller2005). Other authors also found similar results under experimental conditions differing from those used here (Biester & Scholz Reference Biester and Scholz1997). We observed the presence of more than one absorption band in our samples (Fig. 5), indicating at least two types of complexes of Hg with the matrix.
Fig. 5 a. Thermogram of profile F9a. b. Thermogram of profile F11b seal hair. c. Thermogram of profile F11b. d. Thermogram of the profile F11b without seal hair.
Fig. 6 Thermograms for standard samples of Hg compounds (from Valle et al. Reference Valle, Santana and Windmöller2006).
Sample F9a shows the presence of weakly bound Hg compared with sample F11b, which has visible fragments of seal hair. We observed a stronger Hg-OM interaction, as the profile from the deuterium lamp, which is sensitive to gases produced by OM decomposition, follows the profile of the Hg spectrum of the hollow cathode lamp. Sample F11b, containing no seal hair, showed a similar thermogram (Fig. 5b) to sample F9a (Fig. 5a). This indicates a weaker Hg-matrix interaction (temperature release between 220 and 320°C), probably due to Hg bound to highly mineralized OM. A stronger interaction is observed at temperatures from 320–500°C, following the release of Hg bound to OM from the seal hair. These results are in agreement with the results obtained by sequential extraction, with Hg present mainly in association with OM. Thermograms were conducted with larger masses of samples to check for the loss of Hg at low temperatures, characteristics of Hg0, and it was not observed.
Conclusions
The Hg content in soils of the Fildes and Ardley peninsulas in Maritime Antarctica ranged from 4.3 ± 0.2 to 256 ± 4 ng g-1, with higher contents observed for ornithogenic or seal-affected soils. The values found were the highest reported in Antarctic soils to date. The analytical procedure used had an appropriate quantification limit for analysis, as well as good precision and accuracy. There was a clear correlation between Hg and OM associated with nesting birds and breeding seals. Ornithogenic soils showed declining Hg concentrations in deeper soil layers, whereas in non-ornithogenic soils the opposite trend was observed. It can be argued that in the ornithogenic soils the interaction of Hg with OM is dominant over other types of interactions. The possibility that a portion of the Hg present may be derived from atmospheric deposition (springtime depletion events) followed by OM complexation should also be considered, mainly in the non-ornithogenic soils.
The large variation in Ptotal and Pavailable is probably due to varying levels of ornithogenic influence, the types of phosphate minerals present and the conditions of solubilization and local drainage patterns.
All Hg found was in the form of Hg2+ and mainly associated with organic matter, as confirmed by the results of solid-phase Hg pyrolysis coupled to CV-AAS and sequential extraction.
Statistical analysis of all variables made clear the importance of physical and chemical processes related to sea-land transfers promoted by marine animals (penguins, seals) to the geochemistry of Hg in this region. The close correlation observed between Hg and OM in the ornithogenic soils motivates the need for more detailed studies to elucidate these interactions in order to understand the biogeochemistry of Hg in the Antarctic environment.
Acknowledgements
The authors thank the Brazilian Antarctic Program (PROANTAR), CNPq, FAPEMIG, PRPq/UFMG and FEAM and the Brazilian Navy for logistical and financial support. This study received contributions from the International Polar Year ANTPAS Project (Cryosols Project), and is part of the INCT-Criosfera Program, currently supported by the Brazilian Research Agency CNPq. The constructive comments of the reviewers are also gratefully acknowledged.
Appendix A Geographical co-ordinates (UTM) of the sampling points (F = Fildes and A = Ardley).
Appendix B Contents of C, H, N, Ptotal, Pavailable and HgT in soil samples from the Fildes (F) and Ardley (A) peninsulas.
*the letters a, b, c, d, e and f indicate increasing depth.
**standard deviation, n = 3.
***unspecified.
