Metals in some dominant vascular plants, mosses, lichens, algae, and the biological soil crust in various types of terrestrial tundra, SW Spitsbergen, Norway

Arctic environments are commonly considered to be relatively pristine because of minimal local human activity. However, these areas receive air pollution from lower latitude regions. Our goal was to determine concentrations of metals (Cd, Co, Cr, Cu, Fe, Hg, Mn, Ni, Pb, and Zn) in dominant species of vascular plants, mosses, lichens, algae, and in the biological soil crust (BSC), and topsoil (0–3 cm) from various types of tundra in the southwestern part of Spitsbergen, Norway. Results indicate that mosses are more efficient bioaccumulators of Cd, Co, Cr, Cu, Fe, Mn, and Zn than lichens. The highest levels of Co, Cr, Cu, Fe, Hg, Mn, Ni, and Pb were found in the BSC, and the moss species Racomitrium lanuginosum, Sanionia uncinata, and Straminergon stramineum from the polygonal tundra, initial cyanobacteria-moss wet tundra, snow bed cyanobacteria-moss tundra, and flow water moss tundra alimented by melting ice or snow. The observed higher concentrations of Cu and lower concentrations of Hg in mosses, lichens, and vascular plants compared with values observed 20 years earlier were apparently associated with changes in the atmospheric deposition of contaminants over Spitsbergen due to changes in the long-distance transport of anthropogenic emissions from industrialized areas. Prasiola crispa and Salix polaris may be useful bioindicators of Cd and Zn, and the BSC, R. lanuginosum, S. uncinata, and S. stramineum as bioindicators of Co, Cr, Cu, Fe, Hg, Mn, Ni, and Pb. These results may be extrapolated across other areas of Spitsbergen with similar climates.


Introduction
Arctic environments are commonly considered to be relatively pristine and stable because of the absence of intensive local human activities and of significant local atmospheric contamination sources. However, these areas receive air pollution from lower latitude regions (Headley 1996;Bard 1999;Simões and Zagorodnov 2001). Especially in the northern hemisphere, where anthropogenic sources are concentrated in Europe, North America, and Siberia, the natural cycles are strongly influenced by anthropogenic emissions containing toxic elements (Savinov et al. 2003). Ecosystems in the Arctic are highly sensitive to human impacts. Because of relatively simple food webs, even a minor amount of contamination can exert extensive impacts throughout the ecosystem (Gulińska et al. 2003). Svalbard is a unique area in Europe where local atmospheric pollution sources are very restricted. It is far removed from major sources of atmospheric pollution but is recognized to be one of the areas most affected by anthropogenic pollution transported from industrialized areas (Drbal et al. 1992;Birks et al. 2004). These high pollutant concentrations result from specific atmospheric circulation patterns, which bring emissions from oil and coal combustion, non-ferrous metal production, and other sources of anthropogenic pollutants from Europe to Svalbard. Several studies have shown that the atmosphere over Svalbard during winter is heavily loaded with a variety of anthropogenic pollutants, including metals (Heintzenberg et al. 1981;Simões and Zagorodnov 2001). There have also been some indications that polluted air reaches this area during summer (Heintzenberg et al. 1991). Another study demonstrated that contamination of Arctic air led to the bioaccumulation of metals in each trophic level, including terrestrial biota . Bryophytes, as well as lichen and algae, are abundant in the Arctic, particularly in the wet and low tundra (Smith 1982). They live in ecosystems considered to be extremely severe because of low temperatures, low nutrient availability, and a short growing season; hence, abiotic factors are supposed to be largely responsible for the restrictions imposed upon the distribution of populations (Davey 1997;Hoshino et al. 1999). The aim of this study was to investigate the level of metal contaminants (Cd, Co, Cr Cu, Fe, Hg, Mn, Ni, Pb, and Zn) in dominant higher plants, mosses, lichens, algae, and the BSC collected from various types of terrestrial tundra in the southwestern part of Spitsbergen. We tested the hypothesis that terrestrial mosses are the best bioindicators of contamination in comparison with other plant species, lichens, algae, and the BSC in the various types of the terrestrial tundra.

Sampling design
The study site was conducted in the southwestern area of Spitsbergen on Wedel Jarlsberg Land, on the northwest side of Hornsund fjord, in the vicinity of the Polish Polar Station (77°00 0 N; 15°33 0 E), and investigated during the summer of 2011 (Fig. 1). The region is a typical Arctic tundra ecosystem and is located in the Fuglebekken catchment area. A total of 35 sites were selected ( Fig. 1) representing ten types of tundra (Table 1) modified based on Kuc (1998) and Szymański et al. (2013). According to Walker et al. (2005), Svalbard belongs to a bioclimatc subzone of the Arctic, and this part of Spitsbergen is classified as a physiognomic category of sedge/grass moss wetland. We collected vascular plants: At each site within a 25 m 9 25 m square, five replicates per species were randomly collected. Dead material, soil particles, and litter were manually removed from the samples. In addition, topsoil samples (depth of 0-3 cm) in five replicates were collected from each square. Each sample consisted of a mixture of three subsamples. Plant remains and stones were removed from the soil. The total number of soil samples collected was 35 9 5 = 175.

Soil and plant analysis
Soil and plant samples were dried at 50°C until a constant weight was reached. According to Krishna et al. (2003), this temperature is low enough to prevent the loss of mercury. Soil samples were homogenized with a mortar and pestle after the coarse material was removed using a 2-mm sieve. Plant samples were homogenized to a fine powder in an IKA Labortechnik M20 laboratory mill. Dried soil and plant samples (300 mg of dry weight, in triplicate) were digested with 3 mL of nitric acid (ultra pure, 65 %) and 2 mL of perchloric acid (ultra pure, 70 %) in a CEM Mars 5 microwave oven. Samples were then diluted with deionized water to a total volume of 50 mL, and the soil and plant digests were analyzed for Fe, Mn, and Zn using FAAS and Cd, Co, Cr, Cu, Ni, and Pb using ETAAS with Graphite Furnace GF3000 (AVANTA PM Atomic Absorption Spectrophotometry from GBC Scientific Equipment). Mercury was analyzed using an AMA 254 Advanced Mercury Analyzer. All elements were assayed against Atomic Absorption Standard Solution from Sigma Chemical Co. and blanks containing the same matrix as the samples and were processed and analyzed as samples. Results of metal concentrations for the plants were calculated on a dry weight basis. The accuracy of the methods applied for the determination of the metal concentrations in plant and soil samples was checked against certified reference materials: moss M2 and M3 (Finnish Forest Research Institute), DC73348 LGC standards of bush branches and leaves, NCS DC73350 poplar leaves, and RTH 907 Dutch Anthropogenic Soil (Wageningen Evaluating Programs for Analytical Laboratories, WEPAL). The coefficient of variance (CV) was calculated for the measured metal concentrations in the reference materials (Online resource 1). Epilithic moss-lichen tundra Differences among sampling sites in concentrations of elements in soil, higher plants, mosses, lichens, algae, and the BSC were evaluated by nonparametric ANOVA Kruskal-Wallis analysis. Multiple comparisons of mean ranks for soil, higher plants, mosses, lichens, algae, and the BSC were then calculated. The matrix of concentrations of 10 metals (Cd, Co, Cr, Cu, Fe, Hg, Mn, Ni, Pb, and Zn) in plant, lichen, algae, and BSC samples from 35 sites after Box-Cox transformation and standardization was subjected to ordination to reveal possible gradients of element levels by means of the principal component and classification analysis (PCCA). Plots of PCCA ordination of the plant, lichen, algae, and BSC samples, and projection of the concentrations of elements on the factor plane give information about similarities between samples and shows correlations between the original variables and the first two factors. (Legendre and Legendre 1998). All calculations were done with the Statistica 10 program Statsoft (2011).

Results
The ranges of metal concentrations in soil, plant, lichen, algae, and BSC samples are displayed in Tables 2, 3, 4, 5, and 6. The mean concentrations of metals in soil and biota differed significantly (ANOVA, Kruskal-Wallis P \ 0.05).

Soil
Large difference in the carbon percentage (Table 2) in the examined soils depends on the variability of the organic matter content. For example, the geophytic initial dry tundra and initial cyanobacteria-moss wet tundra are growing on a substrate consisting of wet, very fine sand mixed with many rock fragments. However, the epilithic moss-lichen tundra and wet moss tundra are growing on a substrate consisting of dead shoots of mosses on rocky debris (Szymański et al. 2013).
The comparison of metal concentrations observed in the Svalbard soils with pristine soils from the Russian Arctic (in mg kg -1 : Cd 0.2, Cu 23, Pb 9.1, Zn 29) by Zhulidov et al. (1997) and from much drier tundra soils of the maritime lowland Kaffiöyra in western Spitsbergen (in mg kg -1 : Cd 0.1, Co 7.4, Cu 23, Fe 29000, Mn 392, Ni 24, Pb 12, Zn 75) by Plichta and Kuczyńska (1991) revealed that only the concentration of Co and Fe was higher than background concentrations in all types of tundras (Tables 2, 3). Multiple comparisons of mean ranks for soils revealed that the two ornithocoprophilous tundras did not differ in concentrations of each of the examined metals Table 2 Minimum (

Biological soil crust
As we did not find any publications containing data on metal concentrations in the BSC in Arctic regions, we compared our results to the average metal concentrations (mg kg -1 ) in BSC from Hohe Tauern, Austria given by Peer et al. (2010). The BSC from Spitsbergen contained higher concentrations of Cd (0.41), Cu (11.5), and Ni (22.4); and lower concentrations of Pb (107) and Zn (62) than those reported in Peer et al. (2010).
The ratio of metal concentration in biota to metal concentration in soil (Fig. 2) showed that the accumulation of all metals (except Hg and Zn) was highest in the BSC. The Table 6 Minimum (min), maximum (max) values, median (med) of concentrations (mg kg -1 ) of Ni, Pb, and Zn in vascular plants, mosses, lichens, and algae and biological soil crust (BSC) of various types of tundra of Spitsbergen ratio for Hg was highest in R. lanuginosum and for Zn was S. polaris (Fig. 2).

Discussion
Multiple comparisons of mean ranks for all BSC, algae, lichens, mosses, and higher plant samples revealed that P. crispa contained significantly higher (P \ 0.05) concentrations of Cu than all other samples, except for the BSC, S. uncinata, and R. lanuginosum, while the BSC contained significantly higher (P \ 0.01) concentrations of Co and Ni than all other samples, except for P. crispa. These results are in contrast to Drbal et al. (1992) who found lower metal concentrations in algae than in higher plants.
Our results are supported by the PCCA ordination (Fig. 3). The first principal component discriminates between all BSC samples from initial cyanobacteria-moss wet tundra, polygonal tundra, snow bed cyanobacteriamoss tundra, and flow water moss tundra; mosses: R. lanuginosum of geophytic initial dry tundra and polygonal tundra; S. uncinata of geophytic initial dry tundra, initial cyanobacteria-moss wet tundra, snow bed cyanobacteriamoss tundra, and flow water moss tundra; W. sarmentosa of flow water moss tundra; S. stramineum of flow water moss tundra (negative scores); and all moss species, lichens, and higher plants of other sampling sites (positive scores). The second principal component is correlated (positive scores) with all P. crispa and S. polaris. The projection of the variables on the factor plane showed that the BSC, R. lanuginosum, S. uncinata, S. stramineum, and W. sarmentosa with negative scores of factor one were highly correlated with the highest concentrations of Co, Cr, Cu, Fe, Hg, Mn, Ni, and Pb in their tissues. P. crispa and S. polaris from all sites with positive scores of factor two were correlated with the highest concentrations of Cd and Zn. The highest concentrations of metals were found in the BSC, R. lanuginosum, S. uncinata, S. stramineum, and W. sarmentosa from types of tundra which, except for geophytic initial dry tundra, receive water from melting ice or snow. Spring runoff water alimenting these tundras contains elevated concentrations of metals, and soil-water interactions in tundra soils are a potential source of nutrients for surrounding ecosystems (Elberling et al. 2007;Tye and Heaton 2007). Additionally, water from melting glaciers containing an additional portion of metals deposited from the air should be taken into account (Drbal et al. 1992). Of the moss species examined, S. uncinata appeared to be an efficient bioindicator of metal pollution (Samecka-Cymerman et al. 2011). The reason for the elevated concentrations of metals in R. lanuginosum and S. uncinata from the geophytic initial dry tundra became clearer when we consider elevation. Although the difference in altitude between this tundra site and the other sites was only 26 m, S. uncinata and R. lanuginosum at this site were likely more exposed to pollution transported by wind. Svalbard is located very far from major sources of atmospheric pollution. However, some long-range transport of pollutants has been reported, and atmospheric deposition is a significant source of metals in Arctic ecosystems (Bashkin and Howarth 2002;Birks et al. 2004;Samecka-Cymerman et al. 2011). The supply of sea salts and trace metals via precipitation also appears to contribute to high metal concentrations observed in Spitsbergen biota (Bard 1999;Bashkin and Howarth 2002). The high concentrations of metals in the BSC support the suggestion by Beraldi-Campesi et al. (2009) that the BSC plays a role in soil fertility and influences the metal budgets of associated soil and biota. These authors observed that soils underneath the crusts showed depletion of non-biogenic elements. Peer et al. (2010) confirmed that the BSC acts as captors for atmogenic heavy metals. The very high concentrations of metals in the indicator species P. crispa for soils fertilized with wastes (Olech 1996) may be connected with the fact that in this investigation, P. crispa grew only in ornithocoprophilous tundra in the vicinity of colonies of Uria lomvia and Rissa tridactyla feeding on marine fishes and molluscs. As these birds breed on land, they transport contaminants (among them a significant amount of trace metals) to coastal areas in levels that may affect those ecosystems (Grodzińska and Godzik 1991;Dowdall et al. 2005). Seabirds act as a vector for the transport of metals between the marine and terrestrial ecosystems (Headley 1996). Their excretions are the most probable sources of metals to P. crispa that grow in the vicinity of bird colonies (Bargagli et al. 1998). It is important to note, however, that other species growing under the influence of bird colonies, C. groenlandica and C. arcticum, contained low concentrations of metals. The Mann-Whitney U test comparing the concentration of metals in S. polaris indicated that this species collected from both epilithic moss-lichen tundra and prostrate shrub lichen tundra contained significantly higher concentrations of Zn and Cu (P \ 0.05) than S. polaris collected from ornithocoprophilous tundra. This phenomenon may be attributed to metal dilution by higher biomass production. In areas characterized by low primary production and very slow decomposition processes, the availability of soil nutrients is almost negligible. However, additional sources of macroelements, by the presence of numerous seabirds whose excrements fertilize the area around their breeding colonies, affect development of the specific plant communities of ornithocoprophilous tundra receiving extremely high loadings of nitrogen and phosphorus (Olech et al. 2011). According to Baddeley et al. (1994) S. polaris in particular was very responsive to nitrogen and phosphorus inputs, showing an increase in biomass. Concentration (mg kg -1 ) of nitrogen and phosphorus in soil of ornithocoprophilous tundra was as high as 9,000-18,100 and 600-700, respectively, and in S. polaris was as high as 30,000 and 5,630, respectively. In comparison, nitrogen and phosphorus concentrations (mg kg -1 ) in soil of initial tundra were 300-390 and 300-370, respectively. Rate et al. (2004) reported that plant uptake of all elements increased with increasing biosolid application, suggesting that the dilution effect by increased plant biomass was responsible for erratic metal concentrations.

Conclusions
In this study, mosses were better bioindicators of Cd, Co, Cr, Cu, Fe, Mn, and Zn than lichens. PCCA analysis revealed that the BSC, R. lanuginosum, S. uncinata, and S. stramineum accumulated the highest concentrations of Co, Cr, Cu, Fe, Hg, Mn, Ni, and Pb in their tissues growing in polygonal tundra, initial cyanobacteria-moss wet tundra, snow bed cyanobacteria-moss tundra, and flow water moss tundra alimented by melting ice or snow. This implicates the probability that metal contaminants observed to bioaccumulate in these species originate from atmospheric deposits; therefore, a high degree of metal accumulation can be linked to greater wetness of habitats. P. crispa and S. polaris were the best accumulators of Cd and Zn.
Mosses are known to be bioindicators in Spitsbergen, as they accumulate significantly higher concentrations of metals than lichens. These results therefore suggest that the observed higher concentrations of Cu in mosses and lower concentrations of Hg in mosses, lichens, and vascular plants in 2011, compared with values observed in 1992, are associated with changes in the atmospheric deposition of contaminants over Spitsbergen, likely because of changes in the long-distance transport of anthropogenic emissions from industrialized areas.