1 Background, aim, and scope

The potential of wetlands for controlling point- and nonpoint-source pollution in surface water has attracted increasing interest over past decades to comply with related regulations and improve the ecological situation. Until now, available information on the performance of wetlands is mainly about general contaminant parameters, such as BOD5, nitrate nitrogen, and ortho-phosphate (U.S. EPA 1993; Belmont and Metcalfe 2003; Bezbaruah and Zhang 2003). However, some recent studies have focused on specific organic contaminants such as herbicides, pesticides, and pharmaceutical and personal care products (Matamoros et al. 2005, 2007, 2008; Matamoros and Bayona 2006; Schulz and Peall 2001; Schulz et al. 2003; Rose et al. 2006). Those studies highlighted the potential of wetlands as a risk-mitigation strategy for many water bodies polluted by organic contaminants. In China, the situation of heavy pollution and scarcity of surface water resources means that any rapid and effective treatment technologies of wastewater are highly welcome.

The mechanisms responsible for removal of organic contaminants in wetlands may include volatilization, adsorption, photolysis, hydrolysis, metabolic degradation and so on (Lewis et al. 1999). These mechanisms can be divided into two classes of processes in the aquatic environment: transport and transformation. Therefore, the removals or fates of organic contaminants in wetlands are the consequences of the two processes. Individual contaminants tend to interact with sediments, suspended particulates, water, and biota (transport processes) (Smit et al. 2008; Weber et al. 2006; Zhou and Maskaoui 2003) and be chemically and biologically transformed (Schulz et al. 2003; Gross et al. 2004) to different extents due to physicochemical properties such as vapor pressure, solubility, hydrophobicity, and biodegradability. Consequently, transport, transformation and fates of organic contaminants in wetlands greatly depend on their physicochemical properties. Organic contaminants could be classified into four groups according to their removal efficiencies: highly efficiently removed, moderately removed, poorly removed, and those recalcitrant to elimination (Matamoros et al. 2008).

Furthermore, one of the fundamental processes controlling the transport, transformation of organic contaminants, especially for hydrophobic organic compounds in aquatic environments is the partitioning and distribution between water, DOC, and POC (Qiao et al. 2008; Servos and Muir 1989; Zhou and Rowland 1997). Contaminants can exist in freely dissolved, DOC-bound, POC-bound, sediment-associated, and biota-associated forms in aquatic systems. Previous studies suggested that POC and DOC played an important role in partitioning processes (Mitra and Dickhut 1999; Chin and Gschwend 1992). Sorption to POC generally retards contaminant’s transport, whereas DOC binding, such as on humic substances, may compete with POC sorption promoting the contaminant transport (Burkhard 2000). However, few studies have been conducted to investigate the effects of POC and DOC on fates and removals of organic contaminants in wetlands.

In this study, fates and removals of 16 priority PAHs, seven OCPs and NPs with different physicochemical properties such as hydrophobicity were examined in a wetland locating along one tributary before its entry into the Guanting reservoir (40°25′N, 115°47′E) in Beijing, China. Agricultural, industrial, and municipal wastewaters are discharged directly into the wetland. This catchment has been known to have relatively constant concentrations of PAHs and OCPs in surface waters and sediments (Xue et al. 2006; Huang et al. 2005). Meantime, nonylphenol ethoxylates (NPEOs) and their degradation products which can cause endocrine disrupting effects have been found widely in wastewater effluents at relatively high concentrations (Gross et al. 2004).

2 Materials and methods

2.1 Chemicals and reagents

The mixture standard solutions of OCPs (α-, β-, γ-, and δ-hexachlorocyclohexane [HCH], p,p′-dichlorodiphenyldichloroethane [DDD], p,p′-dichlorodiphenyldichloroethylene [DDE], p,p′-dichlorodiphenyltrichloroethane [DDT]) and 16 priority PAHs (naphthalene [Nap], acenaphthylene [Acy], acenaphthene [Ace], fluorene [Flu], phenanthrene [Phe], anthracene [Ant], fuoranthene [Fla], pyrene [Pyr], benzo(a)anthracene [BaA], chrysene [Chr], benzo(b)fluoranthene [BbF], benzo(k)fluoranthene [BkF], benzo(a)pyrene [BaP], dibenz(a,h)anthracene [DahA], indeno(1,2,3-cd)pyrene [IcdP], benzo(g,h,i)perylene [BghiP]) and nonylphenols (a mixture of 11 isomers of 4-nonylphenols) were purchased from Supelco (Bellefonte, PA). Standards (from the same vender) of phenanthrene-d 10 , 2,4,5,6-tetrachloro-m-xylene, PCB209, and bisphenol A-d 14 were used as surrogates, and standards of hexamethybenzene, pentachloronitrobenzene and pyrene-d 16 as internal standards. The organic solvents of pesticide grade were obtained from Fisher Scientific (Fair Lawn, NJ).

2.2 Constructed wetland and experiment design

The wetland is a constructed wetland planted with phragmites australis, typha angustifulia and eichnornia crassipes (area, 100 ha.). The hydraulic loading rate and retention time of the wetland are 0.58 m·d−1 and 34 h, respectively.

Water samples were collected at five sites (site 1, 2, 3, 4, 5) from inlet (site 1) to outlet (site 5) of the wetland once a month in summer (from June to August) 2006. Water samples were immediately filtered within 2 days through 0.45 μm glass fiber prefilters (Millipore, Ireland) and enriched by solid-phase extraction (SPE) with HLB cartridges (500 mg/6 mL, Waters Co., Ireland). The filtered particulates were collected as the total suspended particulates (TSPs). At some sites, more water may be needed to filter in order to get enough particulate samples for analysis.

2.3 Analytical methods

Suspended particulates were freeze-dried, ground with a mortar and pestle, and kept at −20°C for further analysis. POC of suspended particulates after removing the inorganic carbon with 21% phosphoric acid was analyzed by total organic carbon (TOC) analyzer (Apollo 9000, Tekmar-Dohrmann, USA). DOC of filtered water samples were analyzed by TOC-5000 (Shimadzu, Japan).

Every HLB cartridge was eluted with 5 mL dichloromethane (DCM)/hexane (1/2, v/v), 10 mL DCM/methanol (9/1, v/v) and 5 mL methanol. The particulate samples extracted with 200 mL hexane/acetone (1/1, v/v) in a Soxhlet apparatus for 24 h. Activated copper was added for desulfurization. All extracts of water and particulate samples were concentrated in a rotary evaporator (Büchi R-200, Switzerland), and solvent-exchanged to hexane for clean-up under a gentle stream of nitrogen gas. The concentrated extracts for PAHs and OCPs were subjected to a glass column (10 mm i.d.) containing 15 g of 1/2 alumina/silica gel. Firstly, the column was eluted with 15 mL of hexane, then was eluted with 70 mL of DCM / hexane (3/7, v/v) and this fraction contained PAHs and OCPs. Eluates were then concentrated followed by adding internal standards into the final extracts prior to GC-MS analysis. Five grams alumina was used to clean up the extracts for NPs, then eluted with 20 mL DCM. The eluates were concentrated, derivatized with the reagent bis(trimethylsilyl)trifluoroacetamide (BSTFA), added internal standard pyrene-d 16 prior to GC-MS analysis.

The PAHs were analyzed using GC 6890N/MSD 5975 inert (Agilent Co., USA) with a HP-5 silica fused capillary column (60 m × 0.25 mm × 0.25 μm) by J&W Scientific (USA). Detection was conducted using selected-ion mode, and the details of instrument conditions were reported elsewhere (Qiao et al. 2006). Quantitation was performed using the internal calibration method based on the five-point calibration curve for individual components. Analysis of OCPs was carried out with an Agilent 6890 series GC equipped with a 63Ni electron-capture detector (Agilent, USA). A 30-m HP-5 column served as the primary column for pesticide identification and quantification, while a 30-m SPB-1701column of Supelco (USA) served as the additional supportive column for pesticide confirmation. Details of the OCPs analysis have been described elsewhere (Xu et al. 2005). Another 30-m HP-5 column was used to analyze NPs using GC 6890N/MSD 5975 inert. The GC oven temperature was programmed from 50°C (2 min) to 100°C (2 min) at 20°C/min, then to 180°C (15 min) at 25°C/min, and finally to 300°C at 20°C/min (5 min). The results were expressed as the sum of the 11 isomers and other procedures of detection and quantitation were similar to those of PAHs.

Laboratory quality-control procedures included analyses of method blanks, spiked blanks (standards spiked into solvent), matrix spikes, and sample duplicates. The average recoveries in spiked water and particulate samples fell within a fairly narrow range from 67 ± 6% to 86 ± 7%, 65 ± 2% to 107 ± 4%, and 71 ± 6% for individual PAHs, OCPs, and total NPs, respectively. Analytical results reported herein were corrected using recoveries of surrogate standards. The method detection limits (MDLs) of PAHs for water and particulate samples were 1.1 to 5.9 ng/L and 0.3 to 1.7 ng/g (dry wt), respectively. MDLs of OCPs for water and particulate samples were 0.10 to 0.16 ng/L and 0.07 to 0.12 ng/g (dry wt), respectively. MDLs of NPs for water and particulate samples were 2.8 to 6.6 ng/L and 1.7 to 5.2 ng/g (dry wt), respectively. Only very small amounts of a few low-molecular-weight PAHs with log K ow <4.5 were detected in method blanks, and neither detectable OCPs nor NPs were found in method blanks.

2.4 Calculation of freely dissolved concentration

Because the bioavailability, fates, and behaviors of organic chemicals in aquatic ecosystems are directly influenced by DOC and POC, freely dissolved concentrations can be estimated using a three-phase partitioning model (Chin and Gschwend 1992; Gschwend and Wu 1984). To examine the behaviors of different forms of contaminants in the wetland, the total concentration in water column was divided into three fractions, i.e., the freely dissolved (f free), the DOC-bound (f DOC) and the POC-bound (f POC). In this study, the concentrations of target contaminants in filtered water were analyzed which did not include the POC-bound phase, so freely dissolved concentrations could be estimated using a two-phase partitioning model:

$$ C_{\text{free}} = \frac{{C_{\text{dissolved}} }}{{1 + K_{\text{DOC}} \times C_{\text{DOC}} }} $$
(1)

where C free is the freely dissolved concentration, C dissolved is the concentration of the contaminant in filtered water including f free and f DOC, C DOC represents concentration of DOC, K DOC is partitioning coefficient of the contaminant between freely dissolved and DOC-bound phases. In present work, K DOC was derived from K DOC = 0.08K ow, as proposed by Burkhard (2000). The ratios of f free/f DOC/f POC were calculated based on the f free which is taken as the unit.

3 Results

3.1 Concentrations and distribution of PAHs, OCPs, and NPs in the wetland

The concentrations of TSPs and DOC, and organic carbon weight fraction (f oc) of the TSPs are listed in Table 1. TSPs could be removed efficiently with a removal rate of 97.4%, and DOC could be moderately removed with a removal rate of 44.7%. The change of DOC along the wetland corresponded well to the change of TSPs (Table 1).

Table 1 Average concentrations (±SD) of total suspended particulates (TSPs) and dissolved organic carbon (DOC), and organic carbon weight fraction (f oc) of the TSPs in the wetland (n = 3)
Full size table

The concentrations of contaminants in filtered water (C dissolved) were detected by solid-phase extraction method, which included freely dissolved (f free) and DOC-bound (f DOC) fractions (Qiao and Farrell 2002; Qiao et al. 2008). Mean concentrations of PAHs, OCPs, and NPs in filtered water were summarized in Table 2. Removal rates of PAHs (excluding BghiP), OCPs, and NPs based on C dissolved ranged from 5.8% to 52.9%, −3.0% to 32.6% and −12.4%, respectively. Correlations were found between removal rates and K ow values of PAHs (r 2 = 0.72, p < 0.01) and OCPs (r 2 = 0.82, p < 0.01).

Table 2 Average concentrations of PAHs, OCPs, and NPs (n = 3, ng/L) in filtered water along the wetland and their removal rates
Full size table

The ratios of f free/f DOC/f POC are listed in Table 3 for the inlet and outlet of the wetland. It was found that the ratios of f free/f DOC for ∑PAHs, ∑OCPs, and NPs were 1.000/0.112, 1.000/0.056, and 1.000/0.016, respectively, at the inlet (Table 3). The f DOC was larger than f free for contaminants with logK ow ≥ 6.0 at the inlet, where DOC and POC concentrations were the highest, and up to 8.2 times for BghiP with logK ow = 6.9. On the other hand, the ratios of f free/f POC were 1.000/0.297, 1.000/0.095, and 1.000/0.129 for ∑PAHs, ∑OCPs, and NPs, respectively. The f POC concentrations of ∑PAHs, ∑OCPs, and NPs accounted for only 21.1%, 8.3%, and 11.3% of total concentration in water column, respectively, at the inlet. However, a good correlation between logK ow and logK oc (dividing organic carbon-normalized concentration in sediment by C dissolved) were observed (r 2 = 0.84 for PAHs and r 2 = 0.86 for OCPs, p < 0.01) (Fig. 1). In the meantime, a better correlation between logK ow and logK′oc (dividing organic carbon-normalized concentration in sediment by C free) was found (r 2 = 0.92 for PAHs and r 2 = 0.91 for OCPs, p < 0.01; Fig. 1). Thus, for contaminants with logK ow > 5.5, the ratios of f free / f POC increased from 1.000/0.854 to 1.000/55.629 with K ow values, and more than 35% of total concentration in water column occurred in f POC, e.g., as much as 90.0% for BghiP (Table 3). Furthermore, we also found that logK oc values (or logK ′ oc values) and ratios of f POC to f free for PAHs were higher than those for OCPs at the same logK ow values (Fig. 1).

Table 3 The ratios of f free/f DOC/f POC and different removals in the wetland
Full size table
Fig. 1
figure 1

Relationships between logK ow, logK oc (organic carbon-normalized partition coefficient calculated by C dissolved), and logK′oc (organic carbon-normalized partition coefficient calculated by C free) of PAHs and OCPs

Full size image

3.2 Removal contributions of different fractions

The total removal rates of PAHs, OCPs, and NPs by the wetland are shown in Table 3, which are based on the sum of f free, f DOC, and f POC. The removal rates ranged from null to 82.0% depending on K ow values. For ∑PAHs, ∑OCPs and NPs, the removal rates of total concentration were 34.3%, 12.7%, and −0.1%, those of f free were 10.2%, 2.6%, and −11.6%, those of f DOC were 4.5%, 2.3%, and 0.5%, and those of f POC were 19.6%, 7.8%, and 11.1%, respectively. For the contaminants with logK ow > 5.0, total removal rates were higher than 35.7%, which was considered as being moderately or efficiently removed according to Matamoros et al. (2008). For these contaminants, more than 50% of the total removal was attributed to the removal of f POC, only a small portion was attributed to the removal of f DOC, and an even smaller portion was attributed to the removal of f free, whereas, for compounds with logK ow < 4.5, the f POC or f DOC only accounted for less than 4.0% of the total, and their removals fell within the range of 15.2% to 25.1%.

4 Discussion

Partitioning of PAHs, OCPs, and NPs between water, DOC, and POC plays a fundamental role in these contaminants’ fates in the wetland, greatly depending on their K ow values. Many studies also showed that K oc values of hydrophobic contaminants in suspended particulates or sediments were closely related to their K ow values (Qiao et al. 2008; Karickhoff 1981; Chu and Chan 2000). It is widely acknowledged that K DOC is also closely related to K ow (Burkhard 2000). Hence, hydrophobic compounds such as PAHs and OCPs with higher K ow would show stronger POC or DOC preference. Hydrophobicity-dependent partitioning of contaminants between different phases was identified as the dominant mechanism for the subsequent trapping of hydrophobic contaminants in natural waters (Mitra and Dickhut 1999; Schulz and Peall 2001). In other words, higher POC or DOC association would occur for the contaminant with higher hydrophobicity, and the ratios of f POC (or f DOC) to f free increased with K ow. Meantime, PAHs exhibited much higher K oc values than OCPs. This effect could result from the enhanced partition of PAHs to suspended particulates, because their planar molecular structures might enable them to gain a closer approach to aromatic components of the suspended particulates and enhance their mutual attractions (Jonker and Smedes 2000; Chiou et al. 1998).

In this study, TSPs (or POC) could be removed efficiently with a removal rate of 97.4%, and DOC was reduced with a rate of 44.7%. That is to say, 97.4% of the POC-bound PAHs, OCPs and NPs, and 44.7% of the DOC-bound were removed with reduction of POC and DOC. Therefore, for the contaminants with logK ow > 5.0, their removal through reducing POC and DOC contributed the overwhelming portion to the total removal in the wetland. The phenomenon that the removal of hydrophobic contaminants was affected by sorption onto the organic matter in wetlands was also observed in many other studies, such as those by Matamoros and Bayona (2006), Matamoros et al. (2008) for galaxolide and tonalide, with logK ow > 5.5, and by Rose et al. (2006) for DDE. As a result, the elimination of organic contaminants attributable to accumulation in suspended particulates which were retained in the wetland was an important removal pathway, especially for organic pollutants of high hydrophobicity in this wetland.

The wetland seemed to be unable to remove NPs. In contrast, concentration of NPs even showed some increase from the inlet to outlet. The first important reason might be that the hydrophobicity of NPs was relatively low (logK ow = 4.2), and the ratios of f free/f POC and f free/f DOC were only 1.000/0.129 and 1.000/0.016, respectively, at the inlet. Thus, the proportion removed by retaining the suspended particulates was very small in the wetland. Secondly, nonylphenol ethoxylates (NPEOs) can be transformed into NPs and other intermediates in the environment (Ahel et al. 1994; Belmont and Metcalfe 2003), and the removal of NPs might be offset by their production from degradation of NPEOs. This could in part explain the increase of NPs in the wetland. Nevertheless, degradation and/or transformation of NPEOs in the wetland were not thorough because their concentrations in filtered water were not reduced efficiently (data not listed in this paper). This phenomenon might be due to the short hydraulic retention time (HRT) of only 34 h in this wetland compared with the HRT of about 1 month in Matamoros et al. (2007). Thus, there might not be enough time or capacity for removing the organic contaminants through chemical transformation and biological degradation in the wetland.

5 Conclusions

POC and DOC played essential roles on the fates and removals of hydrophobic organic contaminants in the wetland. The hydrophobicity of organic contaminants was the most important physicochemical property, and some other properties such as molecular structure and degradability were of secondary importance to the partitioning and distribution processes. The removals of target contaminants with high K ow values should mainly be through association with the suspended particulates which were precipitated and retained in the wetland.

6 Recommendations and perspectives

The influence of hydraulic retention time and some other environmental factors such as temperature on removal of organic contaminants was not yet fully examined in the wetland, and further work should be performed to study these factors. It is also required that behaviors of NPs and their precursors with great environmental impacts in the wetland should be investigated more thoroughly.