Background limit of Zn and Hg in soils of Eastern Serbia.

The objective of the present study was to point out different approaches for 
 background limits assessment of Zn and Hg. Background limits obtained by 
 graphical methods (cumulative probability plot-CDF and boxplot) were similar 
 for Zn (about 100 mg/kg-1), while background limits for Hg were different 
 (from 0.2 mg kg-1 to 0.3 mg kg-1). For three empirical methods, besides 
 natural data, log-transformed ones were used. Results from log-transformed 
 limits were higher than from natural (except classical methods for Hg). The 
 most of the territory of Eastern Serbia has low content of these elements, 
 whose background limits calculated using log-transformed data by 
 [Median+2MAD] methods (73 mg kg-1 for Zn and 0.15 mg kg-1 for Hg) are the 
 most suitable (MAD-median of the absolute deviations from the data′s median). 
 On the part of area with elevated elements content, background limits were 
 determined using log-transformed data by methods with the highest values: for 
 Zn it was [Mean+2Sdev], and for Hg boxplot (166 mg kg-1 and 0.41 mg kg-1 for 
 Zn and Hg respectively).


Introduction
Risk assessment of trace elements in soils requires prediction of site-specific background limit (Reimann et al., 2005).Several methods for assessing background limits have been used so far.Results of [Mean+2Sdev] method are not always satisfying, thus [Median+2MAD] and Tukey-s boxplot are commonly used, usually with log-transformed values (Gil et al., 2004;Reimann and Garrett, 2005;Galan et al., 2008).Graphical methods, EDA (histogram, boxplot and cumulative probability plot-CDF) enable better insight into data structure, inflection points and background limits (Bech et al., 2010).Results obtained by these procedures are different and subsequent application of those values on geochemical maps enables assessment of background limits more accurately.This paper presents results from

Materials and Methods
In all, 979 surface (0-25cm) soil samples were taken, by grid system at each 3.3 x 3.3 km between 2005-2006 (Figure 1).Content of the hot acid extractable form of Zn was determined after extraction with HNO 3 and H 2 O 2 , while samples for Hg determination were additionally prepared by hydride technique.AAS was used for determination, with VGA system for Hg.SPSS.10.0 was used for statistical analysis and graphs (descriptive statistics, cumulative probability plot-CDF and boxplot).Background limits were determined by several methods, described in Reimann et al. (2005): classic [Mean+2Sd], [Median+2MAD], boxplot-upper whisker (Tukey, 1977) and graphics-CDF and boxplot (Tukey, 1977).For background assessment natural and log-transformed data were used.The software used for mapping was ESRI′ Arc View 8.3.

Results and Discussion
Content and distribution of trace elements Content of Hg is below maximum allowable concentration (2 mg kg -1 ), while Zn concentration exceeds MAC (300 mg kg -1 ) in two samples.Median values are 39.3 mg kg -1 and 0.06 mg kg -1 for Zn and Hg respectively (Table 1).Elements have positive Skewness and Kurtosis coefficients, which means that elements are rightskewed and that observations cluster has longer tails then those in the normal distribution.

Graphical methods
In CDF graph the first inflection point for Zn is at 20 mg kg -1 (Figure 2).The second part of the curve (20 mg kg -1 to 70 mg kg -1 ) is significantly prone, which means that data are homogenous in about 70% of samples.In the part of the curve from 70 mg kg -1 to 100 mg kg -1 curve slope is gentler, e.g.differences in Zn content are significantly higher.Background limit (point on which core data were separated from outliners) is set at 100 mg kg -1 , and above this value the curve remains the same, but points are sparse.CDF for Hg shows that in the first part of the curve, Hg content is increasing slowly, thus in about 95% samples Hg values are up to 0.2 mg kg -1 .In part of the curve, from 0.2 mg kg -1 to 0.3 mg kg -1 , a slope is gentler.Background limit is 0.3 mg kg -1 , and values above these are correspondent to individual points with the highest Hg concentration.Background limits from boxplot for Zn are similar as CDF, while for Hg is 0.2 mg kg -1 .[Median+2MAD] and box-plot-calculated value of upper whisker (Table 2).Values from the [Median+2MAD] procedure for Zn and Hg result in the lowest background limits.In natural simulation classical [Mean+2Sdev] rule presents the highest background limit (large Sdev).In logarithmic simulations the highest values is for boxplot for Hg and for Zn classical methods.Results from antilogarithmic values are higher than from natural, with the exception of classical methods for Hg (the highest dispersion values).
Since background limits are different for different methods, results should be checked on maps in order to determine which methods are the most suitable for local characteristics.

Maps
Maps were drawn according to inflection points from CDF graphs and antilogarithmic background limits obtained by three methods (Figures 3 and 4).
There is elevated Zn concentration in some individual samples from the Kucajsko-Beljanicki and Homoljsko-Neresnicki crystalline complex, in the Cretaceous and Jurassic carbonate rocks with the occurrence of magmatism, and in the zone south from Zajecar (Kalenić et al., 1973).Assuming that this Zn content is of natural origin, the most suitable background limits are those with the highest values e.g.[Mean+2Sd].Higher Hg contents are found in Ramsko-Golubacka sandpit, alluvial sediments of the river Pek, Leptosol of Kljuc area, in southern part in individual samples formed on carbonate substrate and on contact part between carbonated and silicated substrate.On those parts the best fitting background is boxplot, upper whiskers.

Conclusion
The objective of the study was to point out different approaches for assessment of background limits of Zn and Hg in soils of Eastern Serbia.According to graphical methods (CDF and boxplot) background limit for Zn is 100 mg kg -1 , and for Hg is within interval from 0.2 mg kg -1 to 0.3 mg kg -1 .Results from antilogarithmic values are significantly higher than from natural ones (except classical methods for Hg). Background limits from the [Median+2MAD] procedure are the lowest in both simulations.Maps showed that the largest part of territory has low concentrations of elements whose background limits are corresponding with the values obtained by [Median+2MAD] methods.On those parts of territory with elevated content, where there is presumption of natural origin, background limits are corresponding with [Maen+2Sd] for Zn and boxplot, upper whisker for Hg.
We applied different background limits (from different methods) on maps, and according to these maps, we chose method that is the most suitable for the given territory.Investigation shows that there is not a universal method that would be appropriate for all conditions and that the right method is a matter of local natural and anthropogenic characteristics.

Table 2 .
Background concentrations of Zn and Hg determined by different empirical methods (mg kg -1 ).