Introduction

During last two decades the constructed wetlands with horizontal subsurface flow (HF CWs) have increasingly been used in the Czech Republic to treat municipal wastewater (Vymazal 1995, 1996, 2002, 2009; Vymazal and Kröpfelová 2005). The first constructed wetland was put in operation in 1989 and at present, there are about 300 systems in operation. Trace elements are usually not the target of the treatment of municipal wastewater but their concentrations in the sediments within the filtration bed may be the concern when the filtration bed would need to be excavated and disposed. So far only few investigations have been aimed at the heavy metals concentrations in the filtration beds of constructed wetlands with subsurface flow treating sewage (e.g. Obarska-Pempkowiak and Klimkowska 1999; Obarska-Pempkowiak 2001; Vymazal 2003; Vymazal and Krása 2003; Lesage et al. 2007a, b).

Redox potential and pH of the sediment–water system are the major factors known to influence the mobility of trace elements in wetlands (DeLaune et al. 1998; Koretsky et al. 2008). However, in most municipal sewage the pH is around neutral and, therefore, this parameter does not affect mobility and retention of heavy metals in constructed wetland too much. Particularly in wetlands, oxidation and reduction reactions are of prime importance (Du Laing et al. 2008). Under aerobic conditions, the most important process affecting accumulation of heavy metals is precipitation of Fe/Mn hydrous oxides (Singer and Stumm 1970). The most important processes affecting heavy metals accumulation/mobilization under anoxic and anaerobic conditions are creation of hydrogen sulfide via sulfate reduction and dissolution of Fe/Mn hydrous oxides (Khalid et al. 1978; Green et al. 2003; De Volder et al. 2003; Mansfeldt 2004).

Sedimentation has long been recognized as the principle process in removal of heavy metals from wastewater in constructed wetlands. However, it is not a simple straightforward physical reaction and other chemical processes such as precipitation and co-precipitation have to occur first (Yao and Gao 2007). Iron, manganese and also aluminum can form under aerobic conditions insoluble compounds through hydrolysis and/or oxidation. This leads to formation of variety of oxides, oxyhydroxides and hydroxides (Wieder 1989; Batty et al. 2002; Woulds and Ngwenya 2004; Sheoran and Sheoran 2006). Once associated with the particulate phase, these elements become subject to removal from the water via sedimentation. The stability of these inorganic compounds is controlled primarily by the system pH, the solubility of the product, and concentrations of the metals and relevant anions (Gambrell 1994; Sheoran and Sheoran 2006). At near-neutral to slightly alkaline pH levels, metals tend to be effectively immobilized (Gambrell 1994). Co-precipitation is an adsorptive phenomenon in wetland sediments. The concentration and distribution of many elements, such as Ni, Cu, Zn or Cd, in sediments and overlying waters are strongly influenced by adsorption and/or co-precipitation with Fe and Mn oxides (Krauskopf 1956; Jenne 1968; Feely et al. 1983; Ferris et al. 1989). Copper, nickel, zinc and manganese are co-precipitated in Fe oxides and cobalt, iron, nickel and zinc are co-precipitated in manganese oxides (Stumm and Morgan 1981). In addition, zinc is reported to be retained on iron plaques at the surface of plant roots (Otte et al. 1995). However, in filtration beds of HF CWs anoxic/anaerobic conditions prevail (e.g., Dušek et al. 2008) and therefore precipitation of Fe/Mn compounds is not the major retention mechanism as Mn and Fe precipitates dissolute under these conditions (Laanbroek 1990; Jacobson 1994; Lovley 1995; Green et al. 2003; Cooper et al. 2006).

Under reducing conditions, dissimilatory sulfate reduction transforms SO4 2− to H2S during respiration by several genera of strictly anaerobic bacteria by reaction with a variety of organic substrates (Gambrell and Patrick 1978; Laanbroek and Veldkamp 1982; Mandernack et al. 2000; Megonikal et al. 2004). Most of the heavy metals react with hydrogen sulfide to form highly insoluble metal sulfides (Krauskopf 1956; Stumm and Morgan 1981; Kosolapov et al. 2004):

$$ {\text{M}}^{2+} + {\text{H}}_{2} {\text{S}} \to {\text{MS}} \downarrow + 2 {\text{H}}^{ + } $$

where M2+ represents a divalent metal ion such as Fe2+ (pyrite, FeS2; pyrrhotite, FeS), Pb2+ (galena, PbS), Cd2+ (CdS), Cu2+ (covellite, CuS; chalcocite, CuS2; chalcopyrite, CuFeS2), Ni2+ (NiS) or Zn2+ (sphalerite, ZnS). These compounds are very stable and insoluble under anaerobic conditions. However, under oxidized conditions sulfides dissolute and release metals. This may occur, for example, as a consequence of oxygen release from plant roots in the rhizosphere (Engler and Patrick 1975; Gambrell et al. 1980; Jacob and Otte 2003).

Heavy metals may also form carbonates when the bicarbonate concentration in water is high. Although carbonates are less stable than sulfides, they can still perform a significant role in initial trapping of metals (Ramos et al. 1994; Sobolewski 1996; Sheoran and Sheoran 2006; Du Laing et al. 2008). Carbonate precipitation is especially effective for the accumulation of lead and nickel in wetlands (Lin 1995).

Metal complexes with large molecular weight organics tend to be effectively immobilized. There is some evidence that at least some metals are more tightly bound by organics under anoxic or reducing conditions compared with upland conditions because humic material may become structurally less complex under oxic conditions (Gambrell and Patrick 1978; Gambrell et al. 1980; Guo et al. 1997). However, complex formation with soluble and insoluble organic matter under all conditions of pH and oxidation intensity occurs (Verloo and Cottenie 1972).

The purpose of this study was to evaluate the amount of sediments and heavy metal concentration in the sediments in filtration beds of constructed wetlands with horizontal subsurface flow treating municipal sewage with various time of operation.

Materials and methods

Seven constructed wetlands with horizontal subsurface flow (Fig. 1) with time of operation varying between 2 and 16 years (Table 1) were sampled in 2008. In each constructed wetland samples were taken in the inflow, middle and outflow zones (three samples in each zone). Gravel or crushed rock samples were taken using the reinforced stainless steel soil sampler (so called “Russian corer”) which was driven into the filtration substrate to a depth of 60 cm. Samples were divided into surface (0–20 cm) and bottom sections (20–60 cm) in order to evaluate vertical distribution of sediments. In the laboratory, samples were cleaned from roots and freeze-dried under low pressure and temperature. Dried sediment material was weighed, homogenized and passed through a 0.5 mesh sieve. After drying both sediment and filtration material volume were determined in order to calculate the volume ratio between sediment and filtration material. 500 mg of the dry sediment was digested in reverted (Löfelt) aqua-regia (4.5 ml HNO3 and 1.5 ml HCl) under high pressure and temperature in microwave apparatus (MARS-5, CEM, USA) according to the modified U.S. EPA method 3052 (U.S. EPA 1995). After digestion, the sample was filtered in order to obtain a clear sample. Heavy metals were analyzed by ICP-MS (PQ-ExCell, VG-Thermo Elemental, Winsford, Cheshire, UK) according to U.S. EPA method 200.8 (U.S. EPA 1994). For statistical analyses of heavy metal concentrations along the filtration beds paired-sample t-test (P < 0.05) was used. Differences between sediment concentrations in the filtration beds were statistically evaluated through the two-way ANOVA (P < 0.05) for vertical (top and bottom) and horizontal (inflow, middle and outflow) profiles.

Fig. 1
figure 1

Schematic representation of a constructed wetland with horizontal sub-surface flow. 1—distribution zone filled with large stones, 2—surface of the bed, 3—water level in the bed, 4—impermeable liner, 5—medium (e.g., gravel, crushed stones), 6—collection zone filled with large stones, 7—collection drainage pipe, 8—outlet structure for maintaining of water level in the bed. The arrows indicate only a general flow pattern (Vymazal 2001)

Full size image
Table 1 Major design parameters of monitored constructed wetlands
Full size table

Results and discussion

Concentrations of heavy metals in the sediments

Cadmium

Under anoxic conditions, cadmium forms very insoluble compounds with sulfide (CdS) and under slightly reduced to oxidized conditions solid carbonate (CdCO3) is a major control mechanism for cadmium solubility (Khalid (1980). Precipitation of carbonate can be microbially mediated, for example, by Alcaligenes denitrificans (Remacle et al. 1992). Under aerobic conditions, cadmium could be adsorbed or co-precipitated with oxides, hydroxides, and hydrous oxides of Fe, Mn and possible Al (Khalid 1980). Cadmium complexed with the organic fraction may be divided into chelated and organic bound. Chelated Cd is the fraction that is loosely attached to immediately mobile and easily decomposable organic material while organic-bound Cd is the fraction incorporated into the insoluble organic material and can be solubilized only after intense oxidation of the organic matter (Khalid 1980).

The concentration of cadmium in sediments of monitored constructed wetlands varied between 0.095 and 1.35 mg/kg (Fig. 2). In all systems, the concentrations in the top layer did not significantly differed from those found in bottom layers with the exception of Příbraz and Spálené Poříčí where the Cd concentration was significantly higher at the outflow zone in the top layer. The results shown in Fig. 2 indicate that in Břehov, Libníč, Mořina and Spálené Poříčí the concentration of Cd near the inflow was significantly higher than in the middle of the bed and near the outflow. Similar observations were also reported by Lesage et al. (2007a, b) from HF CWs in Zemst and Zevergem, De Pinte, Flanders, Belgium and by Vymazal (2003) from HF CW Nučice in the Czech Republic. Also, the Cd concentrations found in our study were comparable with those reported by Lesage et al. (2007b). The values are slightly lower than those reported by Lesage et al. (2007a), Haberl and Perfler (1990), Samecka-Cymerman et al. (2004), Gschlössl and Stuible (2000) or Zuidervaart (1996) from HF CWs in Belgium, Austria, Poland, Germany and the Czech Republic, respectively. On the other hand, the concentrations were lower than concentrations reported from constructed wetlands for road runoff or mine wastewater treatment (Table 2). The data on sediment concentration in various wetlands (Table 2) indicate that concentrations found in our study are also comparable with those found in natural unpolluted wetlands. It is obvious that the highest Cd concentrations are found in wetlands receiving industrial wastewaters or mine drainage waters.

Fig. 2
figure 2

Concentration of cadmium, nickel and lead in seven horizontal sub-surface flow constructed wetlands. Different letters indicate significant difference at α = 0.05 between the means. Bars without letters are not significantly different

Full size image
Table 2 Concentration of cadmium (mg/kg) in sediments of natural and constructed wetlands
Full size table

Nickel

Under oxic or suboxic conditions, Ni sorbs to Mn oxides and can substitute for Ni in the lattice of some Mn oxides (Green-Pedersen et al. 1997; Tonkin et al. 2004). Under anoxic/anaerobic conditions nickel forms insoluble sulfides (Sobolewski 1999) and is incorporated into pyrite (Morse and Luther 1999). Also carbonates could be an effective sink for nickel (Lin 1995).

The concentration of nickel in sediments of monitored constructed wetlands varied between 7.0 and 111 mg/kg (Fig. 2). In all systems, the concentrations in the top layer did not significantly differed from those found in bottom layers. The results shown in Fig. 2 indicate that in Břehov, and Spálené Poříčí the concentration of Ni near the inflow was significantly higher than in the middle of the bed and near the outflow. In Libníč, Příbraz, Slavošovice and Mořina, the Ni concentration in the sediments did not vary too much. In Čejkovice, the highest Ni concentration was measured at the outflow. The literature results on nickel distribution along the filtration bed also vary. Vymazal (2003) found significantly higher Ni concentration in the inflow zone while Lesage et al. (2007b) observed only a slight decrease along the bed and Lesage et al. (2007a) a slight increase in Ni concentration in the sediments along the bed. Nickel concentrations were comparable with Ni concentrations reported from constructed wetlands treating municipal sewage (Table 3). Also, the Ni concentrations are within the range of Ni concentrations reported from both unpolluted and polluted wetlands. Results presented in Table 3 revealed that by far the highest Ni concentrations are found in wetlands receiving industrial wastewater, mine drainage waters and also road runoff.

Table 3 Concentration of nickel (mg/kg) in sediments of natural and constructed wetlands
Full size table

Lead

Koretsky et al. (2008) pointed out that lead, like Zn and Cu, is a chalcophile that forms discrete sulfide phases and may also bind strongly to organic matter. Also carbonates could be an effective sink for lead (Lin 1995). It has been shown that lead also strongly adsorbs to Fe/Mn oxides and it has been found in association with rhizosphere Fe(III) plaques (Dzombak and Morel 1990). However, it has been concluded that the Pb is not trapped by Fe oxides, but rather is complexed to organic matter either in the rhizosphere solution or on the root surface (Sundby et al. 2005).

The concentration of lead in sediments of monitored constructed wetlands varied between 9.3 and 125 mg/kg but most values were lower than 30 mg/kg (Fig. 2). In all systems, the concentrations in the top layer did not significantly differed from those found in bottom layers. The results shown in Fig. 2 indicate that in Břehov, Libníč, Mořina and Spálené Poříčí the concentration of Pb near the inflow was significantly higher than in the middle of the bed and near the outflow. In Příbraz and Slavošovice the Pb concentration in the sediments did not vary too much and decreased slightly along the bed. In Čejkovice, similarly to Ni, the concentrations gradually increase along the bed. Lesage et al. (2007a, b) reported that lead concentration in the sediment decreased along the bed. Vymazal (2003) found a significant decrease after 16 m of the bed but than the concentration increased again and after 48 m the Pb concentration was only slightly lower as compared to the concentration near the inflow. Lead concentrations found in our study were comparable with Pb concentrations reported from natural unpolluted and lightly polluted wetlands (Table 4) In comparison with the results reported from various constructed wetlands the measured concentrations are slightly lower. The data in Table 4 also clearly indicate that Pb concentrations in sediments of wetlands receiving mining drainage waters and waters affected by smelters are up to two orders of magnitude higher.

Table 4 Concentration of lead (mg/kg) in sediments of natural and constructed wetlands
Full size table

Copper

Copper forms under anoxic conditions very insoluble compounds with sulfur, including both cupric and cuprous sulfides (Sobolewski 1999; Morse and Luther 1999) and may also associate with pyrite (Huerta-Diaz et al. 1993). Copper also forms insoluble hydroxides and carbonates (Morel and Hering 1993) but those are important in presence of sulfides. Copper also forms strong complexes with organic matter and can be bound to Fe/Mn oxides under oxic conditions via formation of ternary complexes with organic matter (Achterberg et al. 1997).

The concentration of copper in sediments of monitored constructed wetlands varied between 6.3 and 139 mg/kg but most values were lower than 75 mg/kg (Fig. 3). The concentrations of Cu in the top layer did not differ from those found in the bottom layers in all seven systems. With the exception of Čejkovice and Příbraz, in all other system the Cu concentration was significantly higher in the inflow zone as compared to the middle and outflow zones. Lesage et al. (2007a, b) observed a steep decrease in Cu sediment concentration in two HF CWS in Belgium. Vymazal and Krása (2003) reported slight decrease in Cu concentration along the filtration bed of a HF CW in the Czech Republic. The copper concentrations found in our study were comparable with higher values found in unpolluted wetlands and with lower end of the range reported for polluted wetlands (Table 5). The copper concentrations shown in Fig. 3 were similar to the concentrations reported from Poland and Italy and also by Zuidervaart (1996) who studied heavy metals in the Czech constructed wetlands more than 10 years ago (Table 5). On the other hand, copper concentrations found in our study were lower than concentrations reported from Belgium (Table 5). The data in Table 5 also clearly indicate that Cu concentrations in sediments of wetlands receiving mining drainage waters and waters affected by smelters are up to two orders of magnitude higher.

Fig. 3
figure 3

Concentration of copper, chromium and zinc in seven horizontal sub-surface flow constructed wetlands. Different letters indicate significant difference at α = 0.05 between the means. Bars without letters are not significantly different

Full size image
Table 5 Concentration of copper (mg/kg) in sediments of natural and constructed wetlands
Full size table

Chromium

Contrary to most heavy metals such as Zn, Cd, Pb or Ni, chromium undergoes a change in oxidation state as a consequence of soil oxidation–reduction conditions (Gambrell 1994). These conditions play a major role in chromium speciation, solubility and mobility with reduction transformations being microbially mediated (Masscheleyn et al. 1992; Cervantes et al. 2001). DeLaune et al. (1998) reported that reduction of Cr(VI) occurs at approximately same redox levels as nitrate reduction. Under oxic and suboxic conditions chromium typically sorbs to Fe, and especially Mn, oxides (Davison 1993; Guo et al. 1997; Achterberg et al. 1997). Under anoxic sediments, reduced chromium is not readily incorporated into sulfides (Huerta-Diaz et al. 1998) but instead tends to associate with organic matter (Otero and Macias 2002). Also Guo et al. (1997) reported that under reducing conditions, the behavior of Cr is controlled primarily by insoluble large molecular humic materials.

The concentration of chromium in sediments of monitored constructed wetlands varied between 13 and 163 mg/kg but in Příbraz, Slavošovice, Mořina and Spálené Poříčí the average Cr concentrations in sediments did not exceed 45 mg/kg (Fig. 3). The concentrations of Cu in the top layer did not differ from those found in the bottom layers in all seven systems. In Břehov and Spálené Poříčí the highest Cr concentrations were recorded in the inflow zone while in Čejkovice and Příbraz the highest concentrations were recorded in the outflow zone. Lesage et al. (2007a, b) observed a slight decrease in Cr sediment concentration in Belgium. The chromium concentrations found in our study were higher as compared to values found in natural unpolluted wetlands and are similar to concentrations found in polluted wetlands (Table 6). The data in Table 6 indicate that Cr concentration in sediments in studied HF CWs was slightly higher than most data reported in the literature from constructed wetlands. The range of concentrations found in our study is comparable with concentrations found in a constructed wetland treating road runoff (Scholes et al. 1998). However, the highest values found in our study were lower than concentrations reported by Gschlössl and Stuible (2000) from Germany.

Table 6 Concentration of chromium (mg/kg) in sediments of natural and constructed wetlands
Full size table

Zinc

Under aerobic conditions zinc is commonly associated with Fe and Mn oxides, hydroxides and oxyhydroxides (Krauskopf 1956; Jenne 1968; Ferris et al. 1989; Bostick et al. 2001). Zinc is also retained in iron plaques on plant root surface (Otte et al. 1995). Under anoxic conditions zinc forms very insoluble sulfides (Huerta-Diaz et al. 1993; Achterberg et al. 1997; Stumm and Morgan 1981; Kosolapov et al. 2004) and carbonates Hansel et al. 2001; Bostick et al. 2001).

The concentration of zinc in sediments of monitored constructed wetlands varied widely between 1.0 and 1,768 mg/kg (Fig. 3). In Příbraz, Slavošovice and Spálené Poříčí significantly more zinc was found in the top layer. In most surveyed constructed wetlands significantly more Zn was found in the inflow zone. The extremely high concentrations of Zn in sediments in Mořina are influenced by naturally high inflow Zn concentrations (Kröpfelová et al. 2009). Very high accumulation of zinc in the inflow zone of HF CWs was also reported by Lesage et al. (2007a, b) from systems in Zemst and Zevergem in Belgium and by Vymazal and Krása (2003) from the HF CW in the Czech Republic. Zinc concentrations found in our study were comparable with higher Zn concentrations reported from natural unpolluted wetlands and with lower range of concentrations reported from polluted wetlands (Table 7). Zinc concentrations found in our study has never reached concentrations reported from wetlands impacted by mining activity (Table 7). In comparison with the results reported from various constructed wetlands the measured Zn concentrations are within the same range with the exception of Zn concentrations reported from a constructed wetland treating mining waters from Pb/Zn mine in China (Table 7).

Table 7 Concentration of zinc (mg/kg) in sediments of natural and constructed wetlands
Full size table

In Table 8, average concentrations of studied heavy metals in seven HF constructed wetlands are shown. The data could be compared with background values and legal limits (Table 9). The data indicate that concentrations of Cd exceeded the Czech limits for light soils in Mořina and Příbraz but in general the concentrations were only slightly elevated as compared to concentrations found in unpolluted soils and sediments (Bowen 1979). Also concentrations of nickel were only slightly elevated as compared to unpolluted soils and sediments and only in Čejkovice the average Ni concentration exceeded the Czech limit for other soils.

Table 8 Average concentrations (mg/kg, SD in parentheses) of studied heavy metals in sediments of monitored constructed wetlands (n = 18)
Full size table
Table 9 Limits for heavy metals concentration (mg/kg) in soils and sediments
Full size table

Concentrations of lead were very low and comparable with unpolluted soils and sediments (Bowen 1979). Also for Cu, concentrations in the sediments were quite low and only in Mořina the average value exceeded slightly the Czech limit for light soils. Concentrations of Cr exceeded slightly the Czech limit for light soils in Čejkovice and Libníč, otherwise the concentrations were low and comparable with unpolluted soils and sediments. Concentrations of zinc showed the greatest variation among studied constructed wetlands. In Mořina and Spálené Poříčí the average values exceeded the Czech limits for other soils.

The results did not show any relationship between the concentration of heavy metals and the time of operation. This is probably a consequence of the sediment build-up in the filtration beds where the sediments are also formed by suspended solids. Haberl and Perfler (1990) documented that concentration of Zn, Cu and Cd remained steady during the 7-year study. While the concentrations do not change substantially during the course of constructed wetland operation, due to increase in the sediment biomass the amount of heavy metals increases. This was also observed in our study.

Concentrations of sediment in the filtration beds

For constructed wetlands in the Czech Republic, washed gravel or crushed stones are used. In the beginning of operation, the amount of sediment is zero and its concentration increases during the time of operation. In Table 10, concentrations of sediment expressed in %DM of the filtration bed material are shown. The results indicate the increase of sediment concentration with increasing time of operation. The amount of sediment was usually greater in the inflow zone as compared to outflow zone but the difference was not always statistically significant (Table 10). In Slavošovice, significantly more sediment mass was found at the bottom layer while in Spálené Poříčí significantly more sediment mass was found in the top layer in the inflow and middle zones. Also, in Břehov and Mořina, more sediment was found in the top layer. This variation is probably affected by the placement of the distribution pipes. While in Slavošovice the distribution pipes are buried near the bottom of the bed, in Spálené Poříčí, Břehov and Mořina, the distribution systems is either laid down on the surface of the filtration bed or it is buried only shallowly bellow the bed surface. Taking into consideration the sediment/filtration material mass ratio it was possible to calculate average heavy metal concentrations in the filtration material including sediments (Table 11). As sediment mass varied between 0.42 and 10.55% of the filtration material, the final heavy metal concentrations are much lower than legal limits (Table 9).

Table 10 Average sediment concentration in inflow, middle and outflow parts of the filtration bed of surveyed CWs (in % DM, SD in parentheses)
Full size table
Table 11 Average concentrations (mg/kg DM, SD in parentheses) of studied heavy metals in filtration material (i.e. sediment + filtration substrate) of monitored constructed wetlands (n = 18)
Full size table

Conclusions

Concentrations of Cd, Ni, Pb, Cu, Cr and Zn were evaluated in seven constructed wetlands with horizontal subsurface flow treating municipal wastewater in the Czech Republic. The time of operation varied between 2 and 16 years among systems. The results revealed that concentrations of heavy metals were low and comparable with concentrations found in unpolluted or lightly polluted natural wetlands. The concentrations were much lower than concentrations found in wetlands receiving mine drainage or industrial wastewaters. The concentrations of heavy metals did not reflect the length of operation but the amount of sediment mass increases with the length of operation. This will result in greater heavy metal mass in the system. The concentrations of heavy metals in the sediment exceeded occasionally the limits for agricultural soils but when filtration material was taken into consideration, the concentrations were well below the limits.