Mercury in the Urban Topsoil of Athens, Greece

The present study documents the Hg content in 45 urban topsoil samples from the highly urbanized city of Athens, Greece. The Hg concentrations were quantified by applying aqua regia digestion on the <100 μm soil fraction followed by inductively coupled plasma-mass spectrometry (ICP-MS) with a detection limit of 5 μg·kg −1. The median concentration of Hg in Athens soil is 96 μg·kg −1 ; ten out of 45 soil samples were found to contain Hg concentrations higher than 200 μg·kg −1 , which is the maximum concentration value expected to be present in normal uncontaminated soils. Results obtained by multivariate principal component and hierarchical cluster analysis incorporating a large suite of chemical elements were notably effective for elucidating the anthropogenic origin of Hg in the studied soil. The elevated concentrations are most likely related to site-specific point source contamination rather than to the widely documented influences from the vehicular traffic emissions in urban settings. Given the proximity of urban population to the contaminated urban soils, we suggest the implementation of different soil extraction tests with the aim to evaluate the fraction of soil Hg available for absorption by the human body.


Introduction
Urban soil is an integral component of the urban landscape showing a high spatial heterogeneity driven by abrupt lithological and chemical changes.As a result of ongoing urban population growth, urban soil is subject to continuous accumulation of contaminants potentially threatening the health quality of humans living in modern cities [1][2][3].Among the substances closely related to urban development, heavy metals have been frequently investigated because of their non-biodegradable nature and their long residence time in soil [4].Multi-elemental geochemical studies of soil have been conducted in diverse cities around the globe and have addressed specific metals that are useful indicators of anthropogenic contamination, typically including Pb, Zn, Cu and Cd [5][6][7].
Mercury is a widely distributed hazardous contaminant and has received great attention globally because of its persistence in the environment, high toxicity and tendency to form more toxic organic mercury compounds [8].Soil is regarded as an important sink of Hg and plays a vital role in its biogeochemical cycle [9].It is estimated that half of the global Hg contamination over the past few decades has been accumulated in soil [10].The natural loadings of Hg in soil are the result of the composition of parent materials [11] and atmospheric deposition of biogenically derived Hg [10].Nevertheless, anthropogenic sources of Hg emissions have been identified, including coal burning, incineration of municipal solid waste, electronic and paper industries, and pharmaceutical units [12,13].Once in the air, Hg can be widely dispersed and transported thousands of kilometers from emission sources.
The average concentration of Hg is 0.04 mg•kg −1 in the earth's crust [10], whereas, according to Adriano (2001) [14], maximum concentrations of around 0.2 mg•kg −1 are expected to be present in normal uncontaminated soil.In urban soil, Hg concentrations can be higher than the geochemical background values of soil in rural areas [15].Elevated concentrations of a few to several mg of Hg per kg of soil have been recorded in various cities around the world [16][17][18], typically attributed to anthropogenic sources of enrichment [19,20].There is a general paucity of information in terms of Hg concentrations in the urban soil environment.
The existing data for Hg levels in soil all over Greece are very limited.In particular, Haidouti (1991) [21] and Haidouti et al. (1985) [22] reported maximum concentrations of 0.325 mg•kg −1 in surface industrial soils located near the oil refineries of Aspropirgos area, adjacent to the capital of the country, Athens.Furthermore, Rodríguez Martín et al. (2014) [23] claimed that the Hg content of topsoil around the coal-fired power plants of Kozani-Ptolemais Basin is relatively low (maximum concentration of 0.059 mg•kg −1 ).Apart from these cases, Hg soil determinations in the urban environment are not available for any Greek cities.A recent comprehensive systematic survey that was conducted in the Greater Athens and Piraeus area focused on concentration levels and the discrimination of contamination sources for a considerable suite of potentially toxic metals and metalloids [24].Despite the environmental significance of Hg, its loadings were not assessed in that study.
Therefore, the primary purpose of this study is to bridge the gap generated by the lack of knowledge for Hg levels in the Greek soil environment by quantitatively characterizing the Hg content in topsoil samples from the highly urbanized city of Athens, a European Mediterranean city with a very long history.The area has been continuously inhabited for more than 7000 years and provides an example of early urbanization in the ancient world.The city of Athens reached its "Golden Age" during the 5th century B.C. with over 300,000 residents, minted its own coins and had banks, public services and land use planning [25].Unlike most European capitals, the urbanization of modern Athens was not related to the Industrial Revolution.Today, the urban area of Athens (Greater Athens and Greater Piraeus) with a population around 3,000,000 over an area of 412 km 2 concentrates about one third of the Greek population, as well as a major part of the economic and commercial activities of the whole country.Although there is no large-scale industry in Athens, some commercial activities that could contribute to the increasing of Hg content in soil include silver and goldsmiths, dentistry and inappropriate disposal of batteries.
Our aim was to present the first assessment of Hg levels in Athens soil and investigate its relationship with common soil properties, such as pH, total organic carbon (TOC) and soil texture.We also applied multivariate statistical techniques in order to identify possible sources of Hg enrichment.For that reason, we incorporated chemical results of the elements Fe, Mn, Pb, Zn, Cu, Ni, Cr, Co, Sb, Ag, Au that assisted in the source apportionment of Hg in Athens soil.Results of this study will contribute to the international database of investigations on Hg content in soils and provide a stable scientific basis for further investigating the specific sources of Hg accumulation and its geochemical mobility in the urban environment.

Soil Sampling
Forty-five composite soil samples (0-10 cm depth) were selected from the sample database of an earlier systematic soil geochemical survey [24].The criteria for sample selection were the total content of metals (not including Hg) as determined by a strong acid mixture dissolution and the spatial variability of soil chemical composition.In particular, the selected samples included low, medium and high levels of heavy metals of both natural and anthropogenic origin covering both the periphery and the city core of Athens (Figure 1).The land uses of the selected sites were as follows: road verge (19 samples), park and woodland (13 samples), unbuilt space (7 samples) and public square-playground-school yard (6 samples).Sampling was performed using a plastic spatula and samples were placed in plastic bags for transport to the laboratory.Once in the laboratory, the soil samples were air-dried, disaggregated and sieved to <2 mm fraction.Representative portions of each soil sample were further sieved through a nylon 100-μm sieve and stored in room temperature.

Analytical Methods
Major physicochemical properties, including pH, total organic carbon (TOC) and texture (sand, silt, clay) were determined.Soil pH was measured in a soil to deionized water suspension of 1:2.5 (w/v) based on the <2 mm sample fraction [26].Total organic carbon was determined on the <100 μm fraction according to the volumetric method described by Walkley and Black (1934) [27].Grain size distribution in the sand, silt, and clay fractions was determined using the hydrometer sedimentation method [28].
The determination of total Hg concentrations was performed after digestion by aqua regia at the accredited Acme Analytical Laboratories Ltd of Canada.Aqua regia extractable concentrations of Cu, Pb, Zn, Ni, Cr, Co, Mn, Fe, Sb, Au and Ag are also presented in this study, assisting in the source apportionment of Hg.More specifically, a 0.5 g aliquot of each soil was digested by hot (95 °C) aqua regia.Geochemical solutions were analyzed by inductively coupled plasma mass spectrometry (ICP-MS), Perkin Elmer Sciex Elan 9000.To manage possible contamination of the ICP-MS introduction system by Hg and subsequent memory effects, the tubes were rinsed between samples with concentrated acids to avoid carry over from one sample to another sample.Also, to ensure spectral interferences do not affect the measurements of Fe, the results were corrected for mass overlap.
Limits of detection were calculated as three standard deviations of the results by running very lean solutions and were 0.01 mg•kg −1 for Cu and Pb, 0.1 mg•kg −1 for Co, Zn and Ni, 0.02 mg•kg −1 for Sb, 0.5 mg•kg −1 for Cr, 100 mg•kg −1 for Fe, 1 mg•kg −1 for Mn, 5 μg•kg −1 for Hg, 2 μg•kg −1 for Ag and 0.2 μg•kg −1 for Au.Replicates, reagent blanks and in-house reference materials provided by the ACME Analytical Laboratories were analyzed as part of the quality control procedure.Calculation of precision for each pair of duplicates was based on in-house replicates revealing low RPD (relative percent difference) values for all the elements (ranging from 1.01% for Cu to 11.83% for Hg), except for Au, which exhibited a mean RPD value of 32%.The precision was calculated according to the following formula [19]: %RPD = [(SV − DV)/0.5 × (SV + DV)] × 100, where SV = the original sample value and DV = the duplicate sample value.Recoveries for the in-house standard materials ranged from 78% for Au to 113% for Cu verifying the good quality of the geochemical data.The recovery was calculated according to the following formula: Recovery = (X/TV) × 100, where X = laboratory's analysis result for the standard sample and TV = the true value of the standards sample.

Statistical Analysis
The data were summarized using mean values, medians, minimum and maximum values of chemical elements.Principal component analysis (PCA) and hierarchical cluster analysis (HCA) were applied to the data set with the aim of identifying the source of Hg in Athens topsoil.The methods have been widely applied to several urban soil studies aiming to apportion natural versus anthropogenic contributions [5,6,20,29].PCA was performed with Varimax rotation, which facilitates the interpretation of the output results by minimizing the number of variables with a high loading on each component.HCA was developed according to the Ward's method as a linkage rule followed by the Euclidean distance as similarity measurement and the results are reported in the form of a dendrogram providing a visual summary of the clustering processes.High internal (within cluster) homogeneity and high external (between clusters) heterogeneity assist in the interpretation of the output results.Because multivariate analysis is sensitive to outliers and non-normality of geochemical data sets, elemental concentrations were log-transformed prior to implementation of both PCA and HCA.Moreover, all the variables were standardized to their corresponding z-scores, as commonly implemented in multivariate HCA in order to ensure that all chemical parameters are weighed equally [30].

Results
The statistical summary of soil analyses results for the physicochemical properties and elemental content is provided in Table 1.The complete results of the chemical analyses are provided in the Supplementary Material.The pH of soil samples averaged 8.32 reflecting the alkaline nature of Athens soil [24].Total organic carbon content was low, varying between 0.6% and 4.49%, with an average concentration of 2.15%.The selected soil samples exhibited a wide range in clay (10%-27%), silt (15%-58%) and sand (25%-75%) content.A wide range in soil metal concentrations was also observed (Table 1); the large variability in elemental concentration levels has been attributed to both natural and anthropogenic sources of enrichment based on the results of the previous systematic geochemical survey in Athens soil [24].Median values of the studied elements follow the decreasing order of Fe > Mn > Zn > Pb > Ni > Cr > Cu > Co > Sb > Ag > Au > Hg.The Hg concentrations in this study ranged from 10 to 1082 μg•kg −1 with a median value of 96 μg•kg −1 .No differences in Hg levels due to different land uses could be inferred from these 45 soil samples (one way ANOVA, p > 0.05 in all cases).Only seven of the 45 samples exceeded the Dutch Optimum Value of 0.3 mg•kg −1 for Hg in soil [31].All measurements were far below the Dutch intervention value of 10 mg•kg −1 [31] and the Canadian legislative limit value of 6.6 mg•kg −1 for residential/park land uses [32].The Hg concentrations in topsoil samples from Athens were compared with data reported for other cities around the world [5,9,12,19,20,[33][34][35][36][37][38].The results of this comparison are presented in Table 2.The median concentration obtained in the present study was similar to that reported for Changchun, China (118 μg•kg −1 ), Wuhu, China (125 μg•kg −1 ), Aveiro, Portugal (91 μg•kg −1 ) and Trondheim, Norway (90 μg•kg −1 ).Mercury concentration in Athens topsoil was lower than that of Beijing, China (260 μg•kg −1 ), Palermo, Italy (680 μg•kg −1 ), Lisbon, Portugal (180 μg•kg −1 ), Chicago, USA (190 μg•kg −1 ) and Berlin, Germany (190 μg•kg −1 ).In relation to soils of Spain, a typical country of the Mediterranean region, the Hg content encountered in Athens tends to be consistently higher than the median value reported for the Duero river basin, 30 μg•kg −1 [39], the Ebro river basin, 27 μg•kg −1 [40] and the Spanish islands, 31.5 μg•kg −1 [11].
Table 3 presents the correlation coefficients between soil properties and Hg content.A significantly positive correlation was found between TOC and Hg (r = 0.39, p = 0.008).The results of PCA, determined by applying Varimax rotation for metal concentrations in Athens soil, are shown in Table 4. Three principal components (PC) with eigenvalues higher than 1 (before and after rotation) were extracted explaining 81.5% of variance.The first principal component (PC1) explains 43.4% of the total variance and loads heavily on Cu, Pb, Zn, Ag, Au, Hg and Sb.The second principal component (PC2), accounting for 22.8% of total variability, is dominated by Ni, Co and Cr.The third component (PC3) includes Fe, Mn and to a lesser extent Co and explains 15.3% of the total variance.Hierarchical cluster analysis (HCA) is often coupled with PCA to confirm interpretation of geochemical data.The HCA results for the heavy metals studied are shown in Figure 2 as a dendrogram that enabled the identification of two major groups of elements.Group I comprised Cu, Zn, Pb, Sb, Ag, Hg and Au and was clearly distinguished from Group II that consisted of Ni, Cr, Co, Mn and Fe.
The spatial distribution of Hg concentration in soil is presented in Figure 3. Class intervals were defined by natural breaks in the histogram of the original data and concentrations were plotted as circles with size increasing as a function of concentration.The most enriched samples are distributed within a zone running through the city center towards the west and southwest suburbs of Athens, reaching Piraeus.

Discussion
Mercury is an environmentally important trace element that has received limited attention in the urban environment compared to "traditional" metals, particularly Pb, Cu, Zn and Cd, often examined because of their relatively high concentrations and significance as indicators of anthropogenic contamination.In addition, data on Hg concentrations in the soil environment from Greece are scarce.The results of the present study indicate that Hg levels in the highly urbanized city of Athens are relatively low, considering the worldwide range of Hg concentration in uncontaminated soil (10-200 μg•kg −1 , [14]).About 75% of the collected samples exhibited concentrations lower than 177 μg•kg −1 indicating that most soil samples in the studied data set did not generally present high Hg amounts.
Comparing the results from this study to other urban geochemical studies is not so straight forward because of the different sampling and analytical protocols of each survey.However, the comparison with similar studies worldwide indicated that the median value of Hg concentrations in Athens soil is somewhat comparable to the median concentrations determined in other urban settings.Only in Sicily (Italy) and Beijing (China) are the measured Hg concentrations notably higher (maximum values of 6.96 and 9.4 mg•kg −1 , respectively).Those concentrations have been ascribed to existing chemical laboratories [5] and the historical use of Hg in classical gardens [33].On the other hand, and considering soil environments that are not expected to be highly contaminated by Hg, as in the case of European agricultural soils (median value of 0.030 mg•kg −1 , [41]) and agricultural soils from Argolida Basin in Greece (median of 0.036 mg•kg −1 , unpublished data), the concentrations in Athens soil tend to be higher.Furthermore, differences of over 100 times were observed between minimum and maximum concentrations, reflecting a high degree of spatial heterogeneity in Hg content.This feature has been identified in other urban systems [42] and possibly indicates the existence of occasional site-specific anthropogenic contamination [35].
An interesting aspect of Hg in Athens soil is the lack of significant correlation with common soil properties, such as pH, TOC and texture that typically exert control on the retention of metals that mainly occur in cationic forms.In contrast to natural soil settings where a strong correlation of TOC and Hg is anticipated and the fact that clay-rich soils are known to contain elevated Hg amounts [41], the influence of these parameters on Hg loadings in Athens soil is not important.Only for TOC, there is a statistically significant correlation with Hg levels demonstrating that TOC only partially explains Hg variability, in line with findings by Rodrigues et al. (2006) [12] and Tack et al. (2005) [13].Specifically, only 13% of the total variability in Hg content is explained by a linear regression model with total organic carbon as a predictor variable (logHg = 1.78 + 0.796logTOC; p < 0.01).Such observations imply that several mechanisms may govern Hg distribution in Athens soil such as methylation of Hg, binding in soil minerals and adsorption to inorganic solid surfaces [10,43].
There are a great number of researchers that have applied statistical techniques in urban soil geochemical data with the aim of discriminating natural from anthropogenic sources of contamination [44,45].According to the results of PCA for the present Athens soil data set (Table 4), Hg is strongly associated with Cu, Pb, Zn, Ag, Au and Sb.In a previous detailed geochemical survey of Athens soil [24], Pb, Zn, Cu and Sb have been demonstrated to be tracers of anthropogenic contamination showing a strong affinity with urbanization indicators, such as vehicular traffic, population density and the historical advancement of urbanization.Also, the elements included in PC2 and PC3 have been shown to indicate the influence of local geology on the chemistry of urban soil in Athens.Specifically, the contribution of serpentinized ophiolithic rocks is reflected on the elevated concentrations of Ni, Cr and Co, whereas Fe and Mn are associated with the weathering processes of aluminosilicate minerals and pedogenesis.
The spatial distribution of the highest Hg concentrations in the studied samples (Figure 3), further provide evidence of the strong influence of urbanization indicators on the specific element.Maximum Hg concentrations were measured in samples from central and southern suburbs of Athens, which correspond to areas of high population density and represent the oldest parts of the city.On the contrary, the lowest concentrations occurred in samples from the recently developed north and northeastern suburbs where urbanization has an almost exclusive residential character.For a detailed discussion on the contribution of local geology and other urbanization indicators on soil geochemistry in Athens, supported by maps and spatial statistical analysis, the reader is referred to previously published work [24].It may be inferred that Hg distribution in Athens topsoil is controlled by human activities, in accordance with multiple investigations that have shown that Hg is an element that is typically found strongly enriched in urban settings [8,46].Furthermore, the contribution of soil forming parent materials to the Hg loadings in Athens soil is expected to be insignificant since the regional geology does not support the occurrence of Hg-bearing deposits.
Moreover, the results of HCA showed the clustering of Hg with Ag and Au (Figure 2).This group of elements was clearly distinguished from the classical geochemical signals of vehicular traffic influences including Pb, Zn, Cu and Sb that dominate the soil composition in modern cities.Silver and gold are elements that are the least characteristic of the vehicular and roadway particulate matter emissions suggesting a different anthropogenic cause for Hg enrichment in Athens soil.The number and the spatial distribution of the soil samples analyzed in the present study are rather prohibitive for ascribing the observed Hg accumulation to a specific anthropogenic source.Considering the lack of heavy industry in Athens, the anthropogenic overprint could include a variety of different point sources of contamination.For instance, it has been demonstrated that Hg in the urban environment is released during the life cycle of commercial products such as fluorescent lamps, electrical switches (electronics industry), thermostats, pressure sensing devices and blood pressure reading devices [4].A dense sampling grid within the urban net of Athens is required for accurately mapping the spatial distribution of Hg and elucidating specific sources for the Hg anthropogenic soil alteration.

Conclusions
Soil Hg concentrations in the Athens urban area ranged from 10 to 1080 μg•kg −1 , with a median value of 96 μg•kg −1 .Mercury concentration is elevated in the city compared to soils that are not expected to be contaminated by Hg, but is comparable to other urban settings worldwide.The influence of soil properties, including pH and soil texture, on Hg levels was found to be insignificant.Only total organic carbon exerts some control on the Hg accumulation in Athens soil.The Hg enrichment is most likely caused by site-specific anthropogenic influences that result in the notable variability in Hg loadings observed in this study.Application of multivariate principal component and hierarchical cluster analysis verified the anthropogenic origin of Hg and also suggested that its contamination source is different compared to the classical urban indicator elements (Pb, Zn, Cu, Sb) related to vehicular traffic emissions.Further investigations should give emphasis to the geochemical reactive and available pools of this toxic and non-essential element that are strongly related to the chemical forms with which Hg occurs in the soil.plotted the geochemical maps and contributed to writing of the manuscript.Both authors revised the manuscript and answered to the reviewer's comments.Both authors have read and approved the final manuscript.

Figure 1 .
Figure 1.Topographical map showing the soil sampling locations within the urban net of the Greater Athens and Piraeus area (Athens Ring corresponds to the city center where traffic restrictions have been enforced since 1982 by allowing alternatively odd/even plate number vehicles to enter on subsequent days).

Figure 2 .
Figure 2. Dendrogram showing clustering of chemical elements of concern for soils from Athens city.

Figure 3 .
Figure 3. Spatial distribution pattern of Hg concentrations in urban soil in the city of Athens (n = 45).

Table 1 .
Statistical summary of physicochemical properties and aqua regia extractable concentrations of Hg and other elements for the investigated topsoil samples from Athens city (n = 45).

Table 2 .
Literature data on published median concentrations (μg•kg −1 ) of Hg in urban topsoils from various cities around the world.

Table 3 .
Pearson correlation coefficient for the measured parameters and Hg content in topsoils from Athens city.

Table 4 .
Rotated factors loadings, eigenvalues and percentages of variance accounted for the three principal components.Elements with loadings higher than 0.5 are considered as representative of each principal component.