The role of aluminium and iron in phosphorus removal by treatment peatlands

Ecological Engineering 86 (2016) 190–201

Contents lists available at ScienceDirect

Ecological Engineering journal homepage: www.elsevier.com/locate/ecoleng

The role of aluminium and iron in phosphorus removal by treatment peatlands Anna-Kaisa Ronkanen a,∗ , Hannu Marttila a , Ahmet Celebi b , Bjørn Kløve a a b

Water Resources and Environmental Engineering Research Group, Faculty of Technology, University of Oulu, PO Box 4300, FIN-90014 Oulu, Finland Department of Environmental Engineering, Sakarya University, Sakarya TR-54187, Turkey

a r t i c l e

i n f o

Article history: Received 28 September 2014 Received in revised form 14 October 2015 Accepted 10 November 2015 Keywords: Phosphorus Peat Constructed wetland Aluminium Iron Diffusive runoff

a b s t r a c t Wetlands are commonly used to treat phosphorus from the effluent of municipal wastewater plants after conventional treatment and wastewater from various diffuse sources, with good results. The long term phosphorous (P) retention capacity of wetland treatment systems is a key research question. This study examined phosphorus retention in wetland (peat) soil columns in order to clarify the role of aluminium (Al) and iron (Fe) concentrations in wastewater on P removal. Since Al and Fe in wastewater could be expected to increase P uptake by increasing peat sorption capacity, laboratory flow-through column experiments were run for almost 700 days in conditions replicating the natural conditions in treatment wetlands. The study set comprised 18 peat columns and five water types from different origins (municipal wastewater, peat extraction runoff, distilled water with phosphate solutions containing 0.1 or 0.4 mg PO4 3− L−1 , and pure distilled water). To study retention of sudden P peak concentrations, a high P peak was injected into the columns after about 500 days of wastewater loading. The results clearly showed that Al and Fe in input water maintained P removal in peat soils, with Al form also affecting retention processes, and P saturation did not occur. Therefore constructed wetlands can in some cases be safely used without the risk of P saturation. Furthermore, in the high P peak test, the additional P was successfully retained in columns with accumulated metals, showing that artificial addition of Al can be used to increase P retention capacity in peat soils with low sorption capacity. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Phosphorus (P) is a major pollutant from different land uses and municipal wastewater treatment plants (Diebel et al., 2008; Son et al., 2013) and its negative impacts on aquatic environments are well understood (USEPA, 2000). Constructed wetlands (CW) are commonly used to treat wastewater and runoff from various diffuse and point sources, with good results (e.g. Silvan et al., 2004; Heal et al., 2005; Vymazal, 2011). However, the processes involved and the stability of the retained P in CWs, especially after long time loading and use, remain unclear for peat-based wetlands, which are commonly used in countries in the northern hemisphere such as Finland. Also influence of fluctuate inflow P concentration on purification performance is poorly know. This is typically a case as runoff from peat extraction, forestry or agriculture is at issue.

∗ Corresponding author. Tel.: +358 500 238 001. E-mail addresses: anna-kaisa.ronkanen@oulu.fi (A.-K. Ronkanen), hannu.marttila@oulu.fi (H. Marttila), [email protected] (A. Celebi), bjorn.klove@oulu.fi (B. Kløve). http://dx.doi.org/10.1016/j.ecoleng.2015.11.011 0925-8574/© 2015 Elsevier B.V. All rights reserved.

In CWs, P is removed from water through chemical precipitation, adsorption and absorption (Heikkinen and Ihme, 1995; Heikkinen et al., 2002; Aslan and Kapdan, 2006; Kadlec, 2006; Lee et al., 2010; Liang et al., 2010; Kõiv et al., 2010), and biological uptake by vegetation (Huttunen et al., 1996; Silvan et al., 2003) and microorganisms (Chen et al., 2005). In treatment peatlands, where water table typically fluctuates from 0 to 30 cm above the peat surface, these processes occur in topmost “free water layer”, in the vegetation/poorly decomposed peat layer (acrotelm) or in the deeper peat layer (catotelm). Biological and physical processes take place in these two first mentioned layers, but in the deeper layer, chemical processes are dominating the removal. Since peat is organic soil, also humic substances have a role in removal processes but as far as they have no capacity to retain P, they can react with P and improve P removal as it has been discussed in some peatland studies (Heikkinen and Ihme, 1995). Generally, different types of wetlands and sorption materials are widely studied and used (Vymazal, 2007; Vohla et al., 2010). Modified biomaterials (e.g. Carvalho et al., 2011) and industrial by-products (e.g. Klimeski et al., 2014) have been tested but their long-term performance for P removal is not well proved. Peat is

A.-K. Ronkanen et al. / Ecological Engineering 86 (2016) 190–201

abundant in the Boreal region offering a cost-efficient way to reduce environmental impact of diffuse and point sources. Chemical sorption is considered to be one of the most important factors in P removal in CWs. In general, the presence of aluminium (Al), iron (Fe) and calcium (Ca) increases the P sorption capacity of filter materials (Karjalainen et al., 2003; Xu et al., 2006; Lai and Lam, 2009), especially oxalate-extractable Al and Fe (Yuan and Lavkulich, 1994; Arias et al., 2006). The importance of Fe and Al to P sorption is typically in acidic soils such as peat whereas Ca bound P is typical in alkaline conditions (pH ≥ 7) (Reddy and DeLaune, 2008). Uptake of P by vegetation has also been noted in treatment peatlands (Huttunen et al., 1996; Uusi-Kämppä et al., 2000) and in peatland buffer areas (Silvan et al., 2004), but the role of vegetation in annual P retention differs depending on site. As biomass production is high in wetlands, accretion of peat can function as a major long-term P sink (Vymazal, 2007). Previous studies have found that saturation of the wetland medium with P can restrict its lifetime and reduce purification efficiency (Heikkinen and Ihme, 1995; Arias et al., 2001), which can increase transport of nutrients to receiving water bodies. Due to this finite capacity to retain P, the life time of treatment peatlands has been estimated by laboratory tests to be around 20 years (Heikkinen et al., 1995). However, there is evidence that the amount of P retained in treatment peatlands can be significantly higher than the estimated maximum (Ronkanen and Kløve, 2009), indicating that the life time could exceed 20 years. However, P sorption capacity of highly loaded filter material or CW medium has not yet been studied. At the moment, we also lack a clear understanding of how metals in runoff waters and wastewaters affect long-term P retention processes in peat material. To study P sorption capacity of filter material is well known to be complicated (Cucarella and Renman, 2009). One crucial factor is contact between filter material and the inflow water which is essential in chemical sorption (Heikkinen et al., 1995; Cucarella and Renman, 2009). Also hydraulic loading rate effect the P binding capacity of filters (Herrmann et al., 2013). This highlight the importance of keeping hydraulic and flow processes in laboratory tests as similar as possible to those in natural conditions. Considering peat as a filter material, the structure of the peat layer and its hydraulic properties (hydraulic conductivity, porosity etc.) must be maintained, as those are important for water flow. Most previous studies have used a batch test procedure, by shaking samples in artificial solutions (e.g. Heikkinen and Ihme, 1995; Gray et al., 2000; Karjalainen et al., 2003; Xu et al., 2006), or column tests for disturbed samples (Seo et al., 2005; Babatunde et al., 2009). In these studies conditions are not close to natural wetland conditions, which can affect the results obtained. There is therefore a clear need for laboratory studies on undisturbed samples. An advantage with using an artificial sample solution (such as KH2 PO4 ) is that it easily provides eligible constant concentrations for the test. However, this approach does not take into account the natural chemical composition of water, which is often complex and can influence sorption processes, especially if metals are present. In this study, column experiments were performed with undisturbed peat samples taken in the natural direction of water flow at the active flow depth in a treatment peatland treating peat extraction runoff for nearly 20 years. The flow velocity used in the experiments was set to be equal to the flow velocity in treatment peatland, in order to maintain natural flow conditions. As changes in environmental condition, such as temperature and pH, are known to influence biological processes and P sorption (e.g. Oliveira et al., 2015), these were kept constant. The column experiments were run for almost 700 days, providing a unique dataset on long-term peat retention processes. In addition, the performance of the peat columns was also studied during a sudden

191

high P concentration imitating extreme situation and clarifying differences between the study columns. The main objective of the present study was to clarify P removal mechanisms in wetlands constructed on pristine peatlands. Four different influent sources were used in the columns and the P removal was analysed as regard the following research questions and hypotheses: i. What is the effect of wastewater and peat extraction runoff composition and the presence of chemical residues such as Al and Fe on the P removal capacity of peat? Our starting hypothesis was that Al and Fe present in the wastewater would maintain sorption in peat and prevent P saturation. We also hypothesised that P in peat extraction runoff containing Fe would be better retained than phosphate in distilled water. ii. What is the response of peat columns when loaded different types of water with a sudden high peak of P concentration? Our starting hypothesis in this case was that peat in columns which had high metal retention in the first experiment would retain more P. 2. Material and methods 2.1. Peat soil samples A total of 18 undisturbed peat samples were taken from the Kompsasuo treatment peatland (65◦ 44 43 N, 25◦ 57 80 E) which had purified peat extraction runoff. The Kompsasuo area, in the mid-boreal region of the southern aapa mire zone of Northern Finland, was drained for peat extraction in 1986–1989 and a wetland treatment is used to purify the runoff after two sedimentation ponds. Prior to sampling in 2005, the wetland had treated runoff from the extraction area of 50 ha for 18 years. The peat samples for the column experiment were taken from the top peat layer (acrotelm layer/active flow depth) of the wetland at a depth of 0.2 m. All samples were taken parallel to the natural flow direction in order to provide as natural conditions and peat structure as possible in the study. The samples were drilled directly into the column containers (∅ 10 cm, height 16 cm) which had a sharp lower edge in order to preserve the natural structure of the peat. The density of the peat samples ranged from 0.11 to 0.24 g cm−3 (mean 0.18 g cm−3 ). 2.2. Column experiment procedure Phosphorus and metal (Al and Fe) removal were studied by vertical columns with upward flow direction (Fig. 1). In preparation, 2.5 cm of peat soil at each end of the column was replaced with inert quartz gravel to ensure uniform water flow over the crosssectional area. A filter paper (0.45 ␮m) was placed between peat and quartz to prevent clogging of peat by organic matter which is common problem in column tests. The filter paper also enabled to study removal of dissolved forms of elements which are most important forms to be transported through the peat layers and to be absorbed to the peat in the treatment peatlands. The columns were sealed at both ends and saturated with inflow water from the Kompsasuo peatland before the experiments in order to keep initial conditions in the columns as similar as possible. A peristaltic pump was installed to pump one type of four test solutions namely (i) humic water, (ii) wastewater, (iii) a 0.1 mg PO4 3− L−1 solution and (iv) a 0.4 mg PO4 3− L−1 P solution to the inlet of three columns (Fig. 1). Two pump sets were used to pump pure Milli-Q water to the control columns with zero P concentration (v). Thus, there were three replicates for solutions (i–iv) and six replicates for Milli-Q water (v) (Table 1).

192

A.-K. Ronkanen et al. / Ecological Engineering 86 (2016) 190–201

Table 1 Experimental setup of the column experiment. Inflow water type

Humic water

Wastewater

0.1 mg PO4 3− L−1

0.4 mg PO4 3− L−1

Zero P

Number of replicates Influent source

3 Peat extraction runoff

3 0.1 mg PO4 3− L−1

3 0.4 mg PO4 3− L−1

6 Milli-Q water

Porosity (%) Volume (L) Dry mass (g) Hydraulic loading rate (L m−2 day−1 ) Residence time (day) Duration of the study (day)

0.88 0.93 172 ± 43 16.6 ± 3.8 6.7 ± 1.3 698

3 Treated municipal wastewater 0.90 0.93 150 ± 20 15.3 ± 2.5 7.3 ± 1.1 652

0.84 0.93 197 ± 11 12.7 ± 2.5 8.1 ± 1.8 698

0.87 0.93 151 ± 28 17.8 ± 3.8 6.0 ± 1.7 652

0.88 0.93 160 ± 27 20.4 ± 2.5 4.3 ± 2.1 652–698

Fig. 1. The used set-up in the study. It was repeated for all inflow water types except for the columns receiving Milli-Q water with zero PO4 3− concentration. For those there were 6 study columns, 2 inflow water canisters and 6 outflow canisters.

The phosphate solutions were produced by adding the appropriate amount of KH2 PO4 to Milli-Q water and stored in 20-L containers. New solutions were made when needed and old containers were replaced with new ones, with no refilling. Municipal treated wastewater was collected in a 40-L container from the outlet before the constructed wetland polishing step at Lakeus municipal wastewater treatment plant, Kempele, Finland. The wastewater contained residues of ferric sulphate (Fe2 (SO4 )3 , PIX) and polyaluminium chloride (PAC), both of which are used in the chemical treatment step in the Lakeus plant. PAC is a highly charged polymer which has a powerful capacity to form flocks and further remove P from water and PIX contains 40% Fe3+ . Both are widely used for P removal from municipal wastewaters (e.g. Gregory and Duan, 2001; Pernitsky and Edzwald, 2006). Peat extraction runoff (named humic water) was collected in two 50-L containers from the Kompsasuo peat extraction area, after the sedimentation ponds but before the wetland treatment. New water was collected when needed. Changes in water quality of different influent waters were monitored monthly. Water quality parameters of the waters tested are presented in Table 2 and Fig. 2. Both particulate and dissolved forms of P, Al and Fe were analysed

although the removal efficiency of dissolved fractions were studied. The variation in the quality of humic water and wastewater was mainly due to natural variation of these waters (a change in concentration was observed when new water was collected from treatment wetland inflow and the municipal wastewater treatment plant for the column runs). The flow velocity of the pumps was set to produce a flow rate of 0.1 L d−1 which is about 13 L m−2 d−1 as a hydraulic flow rate to the columns, corresponding the average flow velocity at 20 cm depth in the treatment peatland (Ronkanen and Kløve, 2005). The hydraulic loading rates of each column are shown in Table 1. The experiments were carried out in a cool, dark room at a constant temperature of 10 ± 1 ◦ C which enable to omit effect of these parameters on removal processes. The temperature controlled room also prevented high temperatures leading to peat decay which could also effect on the results. In order to ensure that the water flows through the peat and did not create a shortcut between the sample and the column wall, tracer tests with sodium chloride (NaCl) were performed in all columns. For this, NaCl concentrations of 0.1 mg L−1 and 0.4 mg L−1 were pumped through the columns for 4 h while the tracer concentrations in outflow were measured as electric conductivity (EC) values. A regression line was used to find the actual concentration values based on a calibration curve made for the outflow water. A quick and sharp response in the outflow indicated a short cut and that kind of samples were omitted. The outflow EC reached background levels before the phosphorus loading experiments started. The outflow water from individual columns was collected. The concentrations of total and dissolved phosphorus (Ptot ), phosphate (PO4 3− –P), Al and Fe were analysed in inflow and outflow using standard (SFS and ISO) analytical methods at the laboratory of the Finnish Environment Institute. Dissolved phosphate concentration is regarded as a reactive form of P and generally termed Dissolved Reactive Phosphorus (DRP). In the first two weeks, the samples were taken at 2–3 day intervals, but later the sampling interval was increased to 1–2 weeks. Dissolved oxygen (DO), pH and EC were measured before the water sampling from the column inflows and outflows. The total duration of the column study was 652–698 days (Table 1) and it was separated into two stages: The first stage (Experiment 1) was performed as a continuous concentration

Table 2 Characteristics of the inflow waters (mean ± standard deviation, except the range for pH) used in the column experiment. In artificial solution only dissolved forms of substances were present. EC, DO and pH were monitored 1 to 4 times in a month during the whole study period (698 days).

Humic water Wastewater 0.1 mg PO4 3− L−1 0.4 mg PO4 3− L−1 Milli-Q water

Dissolved Particulate Dissolved Particulate Dissolved Dissolved Dissolved

PO4 3− (␮g L−1 )

Ptot (␮g L−1 )

Al (␮g L−1 )

Fe (␮g L−1 )

pH

DO (mg L−1 )

EC (mS cm−1 )

35 ± 43 79 ± 80 47 ± 32 1485 ± 3239 33 131 0

46 ± 47 139 ± 129 91 ± 56 2057 ± 4474 33 131 0

125 ± 84 159 ± 94 114 ± 98 4840 ± 6693 0 0 0

1920 ± 1010 6105 ± 5589 136 ± 162 2976 ± 5685 0 0 0

5.2–7.7

10.2 ± 0.51

83 ± 25

3.5–8.3

6.0 ± 1.1

1067 ± 136

5.1–6.6 4.9–5.8 8.3

9.5 ± 0.80 9.1 ± 0.000 9.5 ± 0.75

DO = dissolve oxygen; EC = electric conductivity; n = number of analyses for phosphorus, Al and Fe.

2.8 ± 2.3 2.7 ± 1.3 2.6

n 9 13 7 7 3

A.-K. Ronkanen et al. / Ecological Engineering 86 (2016) 190–201

193

Fig. 2. Dissolved PO4 3− –P (DRP), Fe and Al concentrations of inflow waters from a peat extraction area called humic water (A) and from a municipal wastewater treatment plant called wastewater (B).

column named Normal test, while the second stage (Experiment 2) was a high peak stage (Peak test) where the columns were loaded with a thousand-fold DRP concentration for 4 or 12 days (Table 3). The peak DRP was 34 mg L−1 for all columns except for the columns receiving 0.4 mg PO4 3− L−1 to which 131 mg PO4 3− L−1 was injected. A high DRP pulse was inserted in order to observe clear responses and differences in P removal performance of the study columns.

method is given in Ronkanen and Kløve (2009). The measured fraction of P were (i) loosely bounded P; (ii) Al bounded P and (iii) Fe bounded P and used fractionation method was generally known Chan & Jackson method (Chang and Jackson, 1957; Nieminen and Penttilä, 2004), which consist sequential extraction of 1.0 g peat with 50 ml of 1 M NH4 Cl, 0.5 M NH4 F and 0.1 M NaOH solutions. All peat analyses in this study were carried out in duplicate. 2.4. Mass balance calculation and data analysis

2.3. Peat analysis In order to determine the amount of P sorbed into the peat in different columns, the Ptot , Al and Fe concentrations and also different fraction of P were analysed before and after the tests. In addition, one column from the each influent water groups (humic water, wastewater, 0.1 mg/0.4 mg PO4 3− L−1 and zero P solutions) was divided into four equal layers to measure vertical differences in P and metal concentrations. The Ptot , total Al and total Fe concentrations of peat samples were measured by incinerating 3 g peat at 500 ◦ C in a muffle furnace for 3 h, followed by 6 M HCl digestion (Andersen, 1976). The resulting weight loss was considered the amount of organic matter (OM) in the sample. Furthermore, the dry weight and moisture content of the samples were determined by weighing and drying them at 105 ◦ C overnight. Before the analysis, peat samples were homogenized by hand in a plastic bag following the procedure previously used in peat analysis (Heikkinen et al., 1995; Karjalainen et al., 2003; Ronkanen and Kløve, 2009). The samples were also analysed for oxalate-extractable P, Al and Fe (Pox , Alox and Feox ) using a method based on Niskanen (1989). A detailed description of this

The removed DRP, Al and Fe from inflow waters were determined by two methods: (1) based on in and outflow water analyses, considering the difference between these two concentrations as the removed amount and; (2) based on measured concentration of P, Al and Fe in peat before and after the test, where the difference between initial and end concentrations gives the retained amount. In the method 1, the total removed DRP, Al and Fe (M as mg) during the study were calculated for each column by the following equation: M=



ci ti Q

where ci is concentration measured at water sampling i (mg L−1 ), Q is inflow rate (constant, L d−1 ) and ti is time steps between two consecutive water sampling (d). The removed amount of P, Al and Fe calculated by these two methods were compared. Basic descriptive parameters were obtained using the SPSS 20.0 statistical analysis programme (Armonk, NY, USA). Using correlation analysis, the correlations between the parameters

194

A.-K. Ronkanen et al. / Ecological Engineering 86 (2016) 190–201

Table 3 Removal efficiencies of DRP, Ptotal , Al and Fe based on the inflow and outflow analysis (method 1) in experiments 1 and 2 (mean value ± standard deviation). Inflow water type Duration of test (day) Removed DRP (mg) Removal efficiency Removed Ptotal (mg) Removal efficiency Removed Fe (mg) Removal efficiency Removed Al (mg) Removal efficiency Duration of peak injection (day) Injection DRP (mg L−1 ) Injected amount of DRP (mg) Removed DRP (mg) Removal efficiency Final outflow DRP (mg L−1 ) Time of peak outflow (day) Peak concentration (mg/L) P (mg g−1 dried peat) Fe (mg g−1 dried peat) Al (mg g−1 dried peat)

Humic water

Wastewater

0.1 mg PO4 3− L−1

Experiment 1 (Normal test) 571 252 473 2.87 ± 1.07 3.09 ± 0.42 1.13 ± 0.19 72% 72% 76% 3.18 ± 1.75 5.58 ± 0.76 0.67 ± 0.29 58% 74% 46% 60.7 ± 48.7 48.7 ± 15.7 −141 ± 4.65 18% 52% – 10.8 ± 1.69 287 ± 14.9 −0.72 ± 0.17 80% 100% – Experiment 2 (Peak test) 4 4 4 34 34 34 19.3 16.1 13.6 14.2 ± 3.95 16.1 ± 2.20 9.42 ± 5.86 74% 100% 65% 0.024 0.061 0.002 29–48 13 15 1.5–3.8 0.021–0.14 0.87–5.7 Total amount of elements retained during the study 0.101 0.144 0.051 0.353 0.324 −0.716 0.063 1.91 −0.004

were calculated separately for the Normal and Peak experiments. An analysis of variance test was applied to determine whether there were significant differences between the water type effects in terms of P removal. Where differences between water types in P removal were identified, a post hoc test (Tukey) was applied. 3. Results and discussion 3.1. Retention of P and metals Based on analyses of inflow and outflow water samples, the removal efficiency of DRP was high (≥70%) in all 18 columns treating different types of inflow in the first experiment stage (Experiment 1) (Table 3, Fig. 3). No noticeable differences in removal efficiencies (%) between the columns treating the metalcontaining water (humic water and wastewater) and those treating the metal-free waters (0.1 and 0.4 mg PO4 3− L−1 solutions) were noted. However, the removed amount of DRP from inflow was highest in the column with the highest inflow concentration (0.4 mg PO4 3− L−1 solution) and lowest in the column with the lowest concentration (0.1 mg PO4 3− L−1 solution), as could be expected. The columns flushed with pure Milli-Q water leached the majority of P within the first 150 days. Over the total 600 days of the study (Fig. 4), the DRP amount leached was 0.005–0.019 mg PO4 3− –P g−1 dried peat (Tables 3 and 4), which can also be considered water soluble P. This is higher than the measured loosely bound P in the beginning of the experiments (Table 5) indicating that not only loosely bound P is leached in a long-term test when peat is flushed with pure water. In the beginning of the Experiment 1, a high fluctuation in the outflow concentrations was observed for the humic water columns but also for the columns treating the 0.1 mg PO4 3− L−1 solution (Fig. 3). During this period, DRP was leached from the columns. A steady state was reached in 13–100 days (median 72 days), which was much longer than the residence time of the water in those columns (Table 1). This type of long stabilisation time was not seen in the wastewater and 0.4 mg PO4 3− L−1 columns, where the time to reach stable outflow was less than 5 days (median 0 day) after the tests started. The longer leaching period in humic water and 0.1 mg PO4 3− L−1 columns can indicate that the initial peat P concentration was already close to saturation for the

0.4 mg PO4 3− L−1

Zero P

413 5.69 ± 1.10 70% 5.15 ± 1.02 64% −57.6 ± 25.5 – −0.80 ± 0.04 –

613 −0.11 ± 0.12 – −0.75 ± 0.41 – −252 ± 171 – −2.65 ± 2.08 –

12 131 226 31.3 ± 8.64 14% 0.109 7–21 0.064–0.078 0.242 −0.382 −0.005

−0.005 −1.58 −0.0017

inflow P concentration used (Table 2). On the other hand, the columns with low inflow concentrations performed more consistently than the columns with high inflow concentrations during the remainder of Experiment 1. Although the inflow DRP concentration of the humic water columns declined sharply after 300 days after the inflow canister was changed, no leaching was noted (Fig. 3A). The 0.1 mg PO4 3− L−1 columns also removed DRP steadily (Fig. 3C). In the 0.4 mg PO4 3− L−1 columns, which had the highest constant inflow concentration, little decline in removal efficiency was observed after 300 days. In wastewater columns, some decrease in removal efficiency was observed as the new inflow water used contained less P (Fig. 3B). The removal efficiency for dissolved Al was high in the columns treating humic water (80%) and wastewater (100%), whereas dissolved Fe retention was only 18% and 52%, respectively, for the humic water and wastewater columns (Table 3, Figs. 5 and 6). Although anoxic conditions were not observed in the study columns (mean DOoutflows = 8.1–8.3 mg L−1 ), the columns occasionally leached Fe (Fig. 6), whereas Al was consistently retained in the columns with inflow that included metals (Fig. 5). It is well known that variable oxygen supply and redox conditions significantly affect Fe and increase Fe-bound P mobilisation, which makes it more sensitive from a long-term retention point of view. The results also highlight the importance of redox condition measurements in order to more detail clarify redox sensitive P leaching processes. If high P removal efficiency and long-term retention are to be achieved in treatment peatlands, the Al influx should be considered as an important element for improving P removal. This is especially relevant in treatment peatlands (pristine or constructed wetland), where the water table typically fluctuates and causes anoxic conditions, increasing the risk of leaching of P bound to Fe. The results indicate that even low Al concentration could improve P removal if it is residual of the treatment chemical contains Al as its highly charged polymer form. However, artificially added Al may have negative influences to ecology in downstream waters if it leaches from the wetland or if too high dosing is used leading to Al residues in the outflow. In the water studied the dissolved Al concentration of inflows (humic water and wastewater) was quite similar (Table 2), but the origin of Al was different: natural origin in humic water and remnant chemical in wastewater.

A.-K. Ronkanen et al. / Ecological Engineering 86 (2016) 190–201

195

Fig. 3. Dissolved inflow and outflow PO4 3− –P concentrations (DRP) of the study columns receiving (A) humic water; (B) wastewater; (C) 0.1 mg PO4 3− L−1 solution; (D) 0.4 mg PO4 3− L−1 solution during the Experiment 1. The dashed black line shows the detection limit of DRP (2 ␮g L−1 ) and the bold black line represents the mean cumulative removal for each inflow type.

3.2. Mass balance analysis The Fe bound organic P was the main P pool in the studied peat columns contributing 30–33% of total P in the peat whereas Al bound P fraction was only 3% of all P (Table 5). This was expected as peat taken from the CW purifying peat

extraction runoff contained high amount of Fe (humic water in Table 2). After the column experiment, the accumulation of metals (Al, Fe) and P in peat was examined (Tables 4 and 5) and results were compared to the calculated retention based on inflow and outflow water analysis (Table 3). In the columns with metal-free distilled

Fig. 4. Concentrations of (A) dissolved PO4 3− –P (DRP) and (B) total P (Ptot ) in outflow from the columns treating only with Milli-Q water (Zero P) in the Experiment 1. The bold black line represents the mean cumulative leaching of phosphorus from the columns.

196

A.-K. Ronkanen et al. / Ecological Engineering 86 (2016) 190–201

Table 4 Phosphorus, Al and Fe concentration (mg g−1 dried peat) in the peat before and after the column experiments as well as removal efficiencies (change) calculated by peat analysis (method 2). The number of analyses for the initial values was 18 or 20. Humic water

Wastewater

0.1 mg PO4 3− L−1

0.4 mg PO4 3− L−1

0.961 ± 0.136 0.006 (5%)

1.111 ± 0.060 0.156 (16%)

0.936 ± 0.147 −0.019 (−2%)

7.031 ± 1.774 −0.714 (−9%)

6.680 ± 0.942 −1.065 (−14%)

7.617 ± 3.170 −0.128 (−2%)

1.113 ± 0.135 −0.058 (−5%)

1.060 ± 0.107 −0.111 (−9%)

1.153 ± 0.325 −0.018 (−2%)

Zero P

−1

Final Change Final Change Final Change

P initial value = 0.955 ± 0.108 mg g dried peat 1.075 ± 0.126 0.919 ± 0.106 0.120 (13%) −0.036 (−4%) Fe initial value = 7.745 ± 2.29 mg g−1 dried peat 7.413 ± 1.585 7.128 ± 1.874 −0.332 (−4%) −0.617 (−8%) Al initial value = 1.171 ± 0.435 mg g−1 dried peat 1.190 ± 0.0251 1.722 ± 1.246 0.019 (2%) 0.551 (47%)

Table 5 Average of P fractions and total P concentration (mg kg−1 ) with standard deviation in the analyzed peat columns before and after the phosphorus tests. Loosely bound P Before the phosphorus tests (n = 9) Ptot 1.23 ± 0.32 ± Inorganic P Organic P 0.91 ± After the phosphorus tests (n = 10) 1.00 ± Ptot 0.0000 ± Inorganic P 1.00 ± Organic P

0.47 0.40 0.42 0.52 0.000 0.52

Al bound P

Fe bound P

2.42 ± 4.10 0.50 ± 0.96 1.93 ± 4.37 0.84 ± 1.51 0.0000 ± 0.000 0.84 ± 1.51

Total P

31.4 ± 8.19 0.0000 ± 0.000 31.4 ± 8.19

95.5 ± 10.8

30.1 ± 12.7 0.0000 ± 0.000 30.1 ± 12.7

101.2 ± 13.7

n = number of samples.

water with 0.1/0.4 mg PO4 3− L−1 , the mass balance obtained by the two different methods, gave similar results (e.g. less than 1% difference for Fe). In the humic water columns, only P and Al mass balance results were of same magnitude. Also for the zero P columns the results from two estimation methods of retained elements were similar whereas for the wastewater columns the two methods gave clearly different results. According to peat analysis, the retention of P and Fe from municipal wastewater inflow containing metals was −0.036 mg P g−1 , whereas inflow and outflow analysis showed retention of 0.144 mg P g−1 (Tables 3 and 4). The finding that no P was retained by the peat in the wastewater columns could indicate that P had precipitated with metals,

forming flocs before the inflow entered the peat part of the column (Fig. 1). This would mean that retention processes partly took place in the quartz gravel, as the 0.45 ␮m mesh filter paper between the inert quartz gravel and the peat would allow only dissolved form of elements to be transported into the peat. This is also shown by P fractionation results where no increase of Al bound P were observed in the columns. After the tests, the main P pool was Fe bound organic P (Table 5). Precise changes in P pools could not be observed for individual columns because before and after analysis of the fractions studied were nearly equal. Only loosely bound inorganic P totally leached from all columns during the tests. One reason for no differences could be that the fractionation method

Fig. 5. Dissolved inflow and outflow Al concentrations of the studied columns receiving (A) humic water; (B) wastewater; (C) 0.1 mg PO4 3− L−1 solution; (D) 0.4 mg PO4 3− L−1 solution during the Experiment 1. The bold black line represents the mean cumulative removal of Al in the columns. In the columns treating artificial solutions (C and D), the inflow Al concentration was zero meaning that Al was leached seen in the negative removal in the subfigures (C and D).

A.-K. Ronkanen et al. / Ecological Engineering 86 (2016) 190–201

197

Fig. 6. Dissolved inflow and outflow Fe concentrations of the studied columns receiving (A) humic water; (B) wastewater; (C) 0.1 mg PO4 3− L−1 solution; (D) 0.4 mg PO4 3− L−1 solution during the Experiment 1. The bold black line represents the mean cumulative removal of Fe in the columns. In the columns treating artificial solutions (C and D), the inflow Fe concentration was zero meaning that Fe was leached seen in the negative removal in the subfigures (C and D).

is not well suited for peat although it has been used also in other studies. The differences between calculation methods could also partly be explained by accuracy of calculations and analysis but also by non-homogeneity of the peat causing variations in the initial P values. This led to rather large P, Al and Fe concentration ranges in the columns, and in the most of cases, the observed changes fell within the standard deviation of the initial values (Table 4). Also P retention in the Humic water columns showed differences between calculation methods (23%) (Tables 3 and 4). This can indicate that P input to the columns was mainly in forms, such as humus-metal colloids, which have been retained before the peat whereas only dissolved forms of P such as DRP could flow though the filter paper. These results from wastewater and humic water columns highlight the importance of using natural waters when investigating P removal in filter materials. Artificial solutions are often used in such studies (Dunne et al., 2005; Søvik and Kløve, 2005; Xu et al., 2006) but these cannot give the correct DRP removal efficiency value if waters containing e.g. Al and Fe need to be purified. Because of precipitation, inflow and outflow analysis shows total removal of DRP in the column, whereas peat analysis before and after the study shows that peat adsorbed P. After the column experiments, peat P concentrations were typically highest close to the inlet and lowest near the outlet (Fig. 7A), where the concentrations did not differ clearly from the initial peat P concentration. The highest P concentration was measured in the humic water column. Possible reason for this could be that peat extraction runoff, including high amount of organic Fe colloids, has enhanced P sorption as it is well known that organically bound metals are surface sites for DRP (e.g. Giesler et al., 2005). The opposite finding was for the profile of the column with inflow of 0.4 mg PO4 3− L−1 which showed a high P concentration close to the outlet, with no clear decrease in P in the flow direction. This column was only one where P concentration near the outlet (1.19 mg g−1 dried peat) was clearly higher than the initial peat P concentration (0.96 ± 0.11 mg g−1 dried peat). This column was saturated with DRP in the Experiment 2 (Fig. 8D, Outflow 1) and also resulted in a rather constant P profile (Fig. 7A). An unexpected result was that even close to the inlet, the P concentration of the

wastewater column was within the range of the initial values. This confirms our earlier observation that DRP in the wastewater columns was apparently removed by precipitation processes in the inert quartz gravel between the inlet and the peat, indicating that peat played a minor role in P removal. The Al profiles were similar for humic water and both phosphate solutions. The only exception was the column loaded with municipal wastewater, for which a clear declining trend was found in the Al profile, with the concentration four-fold higher close to the inlet than the outlet (Fig. 7B). In addition, the concentration close to the outlet was equal to that in the phosphate solutions (the inflow water did not contain any metals), indicating that Al was completely retained in the upper end of the columns. The opposite was found for the humic water column, where a slightly elevated concentration was observed at the outlet. This might indicate that different forms of Al give different retention behaviour. The Fe showed no special pattern in accumulation within the column (Fig. 7C). An unexpectedly high Fe concentration was measured 7 cm from the inlet in the column with inflow of 0.4 mg PO4 3− L−1 solution. This can be explained by transport of Fe within the peat, because the inflow did not contain any Fe. Although Fe concentrations varied depending on the distance from the inlet and the inflow water type, near the outlets the Fe concentration was similar in all columns. The Alox and Feox concentrations, as indicators of metals bound to organic matter, increased by 70% and 12% of the initial values in wastewater columns during the experiment (Table 6). In addition, Alox increased (100%) in the humic water columns, but Feox remained constant. In the columns with metal-free inflow, Alox and Feox did not change during the experiment, as could be expected. In the samples where initial Alox concentration was high, Feox was also high (r = 0.803, p < 0.01). The same type of relationship was found for end Alox and Feox concentrations (r = 0.717, p < 0.01). However, Pox slightly increased only in the humic water and zero P columns, indicating active soil processes whereby P is bound to organic matter. The reason why this was not observed in the wastewater columns might be retention of P and metals through chemical precipitation occurring in the gravel layer above the

198

A.-K. Ronkanen et al. / Ecological Engineering 86 (2016) 190–201

Fig. 7. In the end of the column studies, final (A) phosphorus P, (B) aluminum Al and (C) iron Fe concentration profiles (mg g−1 dried peat) in the columns treating different type of water. The initial values of concentrations are shown with light gray lines in each subplot.

columns. However, the values obtained were all very small, which weakens the reliability of the results. The initial metal and P values in the samples were not correlated with amounts removed. This was obvious because the

experimental set-up was designed to provide similar initial conditions in all columns. However, a negative correlation (r = −0.495, p < 0.05) was found between initial Alox concentration and amount of Fe removed.

Fig. 8. Dissolved PO4 3 –P concentration (DRP) of inflows and outflows in the columns treating (A) humic water, (B) wastewater, (C) 0.1 mg PO4 3− L−1 solution and (D) 0.4 mg PO4 3− L−1 solution during the Experiment 2 (peak test). The bold black line represents the mean cumulative removal/leaching of phosphorus for each inflow type.

A.-K. Ronkanen et al. / Ecological Engineering 86 (2016) 190–201

199

Table 6 Mean initial and final concentrations of ammonium oxalate extractable Al (Alox ), Fe (Feox ) and P (Pox ) in the test columns. Alox (mg g−1 ) Initial Humic water Wastewater 0.1 mg PO4 3− L−1 0.4 mg PO4 3− L−1 Zero P

0.16 0.25 0.25 0.24 0.33

± ± ± ± ±

Feox (mg g−1 ) Final

0.13 0.049 0.014 0.045 0.13

0.24 0.38 0.23 0.24 0.23

Initial ± ± ± ± ±

0.046 0.27 0.11 0.087 0.059

1.65 1.53 1.60 1.50 2.11

3.3. Response to a high phosphorus peak The Peak test (Experiment 2) showed that chemical residues (such as Al and Fe) in municipal wastewater maintained high DRP sorption potential in peat. This can be seen from the shape of the response to the P peak (Fig. 8). The columns loaded with municipal wastewater retained DRP completely in the experiment (Table 3). The outflow DRP concentrations also remained at a low level throughout the test, even more than 100 days after the peak in the wastewater columns (Fig. 8B). Low responses were detected in only three replicates 29–48 days after injection, but in other columns the responses were clearly higher. A sudden high phosphate load can cause temporary saturation of peat. This was confirmed by the results from the columns into which the highest phosphate concentration (131 mg PO4 3− –P L−1 ) was injected in the Peak test. The resulting peak was 10 times greater than in the other columns and caused saturation conditions (Fig. 8D). Before the Peak test, these columns were loaded with the metal-free 0.4 mg PO4 3− L−1 solution. After injection, outflow was monitored for 232 days and by the end of this period the outflow DRP concentrations were still higher than the inflow concentration in all three replicate columns. The total amount of retained P was 0.21 mg g−1 dried peat in those columns during the Peak test. This was higher than after the Experiment 1 (Normal test), when the amount of P retained was 0.033 mg g−1 dried peat as total retained P was 0.242 mg P g−1 dried peat in whole study (Table 3). Based on the results, we can conclude that a maximum of 0.2 mg P g−1 dried peat can be retained when no Al and Fe are present and peat P and metal concentrations are similar to those in this study. In a wetland soil in south-east Ireland, maximum P sorption capacity was found to be 0.618–1.464 mg g−1 P (Dunne et al., 2005), whereas in upland soils in Thailand the maximum P sorption range was estimated to be 0.012–1.694 mg g−1 P (Wisawapipa et al., 2009). The results from the present study are within the range of what has been previously reported. 3.4. Role of Al and Fe in P retention in peat The type of inflow waters used in this study differed considerably between the columns and this was reflected in the roles of Al and Fe in the DRP removal mechanism. The Al and Fe in wastewater improved retention notably and the wastewater had a clear impact on the retention of DRP in peat throughout the almost 700 days of the experiment. This was seen as high retention of DRP (up to 100% or 0.14 mg g−1 , Table 3) in the columns receiving municipal wastewater. The columns loaded with humic water also showed high retention, with 74% or 0.10 mg g−1 of P retained. These columns received Fe and Al with the inflow, and removal of dissolved Ptot and DRP has correlation with Al removal (rPtot = 0.527, p < 0.05) and Fe removal (rPtotl = 0.658, p < 0.01; rDRP = 0.496, p < 0.05). These columns (humic water and wastewater) had also the highest capacity to retain a 10 to 100fold higher phosphate inflow in the Peak test. Because there were no metals in the peak solution injected, the high retention capacity

± ± ± ± ±

Pox (␮g g−1 ) Final

0.19 0.098 0.092 0.26 0.84

1.48 1.58 1.47 1.60 1.51

Initial ± ± ± ± ±

0.35 0.44 0.43 0.68 0.58

0.86 1.37 0.54 1.22 0.83

± ± ± ± ±

Final 0.46 0.41 0.000 0.42 0.27

1.13 0.98 0.35 0.51 1.16

± ± ± ± ±

0.61 0.44 0.21 0.61 0.70

can only be explained by metal retention in the columns before the Peak test, in the other words during the preceding Normal test. The ability of a material to remove P typically depends on its physical and chemical properties. It is well known that P removal efficiency is closely related to the content of Fe and Al (Arias et al., 2006; Lai and Lam, 2009) and Ca (Johansson and Gustafsson, 2000). In addition, the shape, particle size and porosity of the material through which the water is infiltrating define the specific surface area and the P sorption capacity (Sakadevan and Bavor, 1998; Drizo et al., 1999). In the present study, the filter material (peat) had similar properties in all 18 columns and only the influent source was changed. Therefore our findings prove that metal concentrations play an essential role in DRP removal. Also possible formation of humus-metal colloids might have a role in DRP removal by treatment peatlands but this has to be more studied in the future. Humus–metal colloids with Fe and Al are typically observed in boreal peatland-dominated catchments, and they form an essential vector for transport of different elements (Ingri et al., 2000; Pokrovsky and Schott, 2002; Ilina et al., 2013). The role of metals in P removal processes has been identified in previous studies, but these studies has used artificial solutions, a batch test procedure or disturbed samples (Arias et al., 2001; Dunne et al., 2005; Xu et al., 2006; Babatunde et al., 2009). Thus in those cases the applicability of the results to the real situation remains unclear. In present study, the role of metals in DRP removal was further demonstrated by the performance of the columns in the Peak test (Experiment 2, Table 3). Based on our knowledge, this is the first time when this kind of stress test (high P peak) has been used and differences in responses have been studied. The columns loaded with metal-free waters (0.1 mg PO4 3− L−1 or 0.4 mg PO4 3− L−1 solution) showed variable retention of DRP. Although the highest amount of retained DRP was achieved in the columns with the highest DRP inflow (those receiving 0.4 mg PO4 3− L−1 ), removal efficiency were clearly the lowest (14%) (Table 3). This low purification is substantially lower than the value for the humic and wasterwater columns despite the fact that the amount of in flowed DRP to 0.4 mg PO4 3− L−1 columns was only double compared to the humic and wastewater columns. This was probably due to saturation of the peat with phosphate in 0.4 mg PO4 3− L−1 columns, as no new metals were transported with the inflow and the inflow DRP was high. Because the peat contained Fe and Al (Table 4), the results also indicate that a high capacity to retain DRP requires continuous feeding with metals. The columns with the lowest inflow DRP concentrations had higher removal efficiency (65–100%), as the peat was not yet saturated with P. Many previous studies have used artificial P solutions for sorption or retention experiments (Cucarella and Renman, 2009). However, the results from the present study show that P sorption experiments in peat must be carried out with real wastewater, as it contains other elements affecting the retention capacity of the material. Thus P can be absorbed to other compounds and transformed during filtration through the material. For example, actual P sorption capacity with wastewater has been noted to be lower than estimated based on artificial solutions (Hedstöm and Rastas, 2006).

200

A.-K. Ronkanen et al. / Ecological Engineering 86 (2016) 190–201

However, in the case of peat material, the long-term retention and sorption of P seems not to be dependent only on ions (Al, Fe) in peat material, but rather also on metals transported with the influent. In wastewater, the metals originate from residues of treatment chemicals whereas in natural humic water, Al and Fe are transported with humus components or colloids. This is important finding, as it reveals that the lifetime of treatment peatlands is reliant not only on the site-specific properties of peat, but also on metals transported in input water. This can increase the effective operating time of wetlands treating P loads originating from point or diffuse sources. 4. Conclusions The results clearly show that metals present in the inflow water increased and maintained the P sorption capacity of the peat. Retention of P was also dependent on the form of Al and Fe in the water and was highest for wastewater containing metals originating from reactive coagulants used in municipal treatment plants. These results were confirmed by good retention of a sudden high P peak in the columns treating wastewater. Thus wetlands with extra metals in the peat matrix will operate well in the long-term, without P saturation, and also e.g. following malfunctions in municipal wastewater treatment plants. Acknowledgements This study was funded by the Academy of Finland, the Multidisciplinary Environmental Graduate School of the University of Oulu (EnviroNet) and Maa-ja vesitekniikan tuki r.y. The authors would also like to thank Dr Pekka Belt, Dr Matti Mottonen and Dr Janne Harkonen for their support in writing this article. We also thank four anonymous reviewers for valuable comments to this manuscript. References Andersen, J.M., 1976. An ignition method for determination of total phosphorus in lake sediments. Water Res. 10, 329–331. Arias, C.A., Del Bubba, M., Brix, H., 2001. Phosphorus removal by sands for use as media in subsurface flow constructed reed beds. Water Res. 35 (5), 1159–1168. ˜ Arias, M., Da Silva-Carballal, J., García-Río, L., Mejuto, J., Núnez, A., 2006. Retention of phosphorus by iron and aluminum oxide-coated quartz particles. J. Colloid Interface Sci. 295, 65–70. Aslan, S., Kapdan, I.K., 2006. Batch kinetics of nitrogen and phosphorus removal from synthetic wastewater by algae. Ecol. Eng. 28 (1), 64–70. Babatunde, A.O., Zhao, Y.Q., Burke, A.M., Morris, M.A., Hanrahan, J.P., 2009. Characterization of aluminium-based water treatment residual for potential phosphorus removal in engineered wetlands. Environ. Pollut. 157 (10), 2830–2836. Carvalho, W.S., Martins, D.F., Gomes, F.R., Leite, I.R., Gustavo da Silva, L., Ruggiero, R., 2011. Phosphate adsorption on chemically modified sugarcane bagasse fibres. Biomass Bioenergy 35, 3913–3919. Chang, S.C., Jackson, M.L., 1957. Fractionation of soil phosphorus. Soil Sci. 84, 133–144. Chen, Y., Liu, Y., Zhou, Q., Gu, G., 2005. Enhanced phosphorus biological removal from wastewater-effect of microorganism acclimatization with different ratios of short-chain fatty acids mixture. Biochem. Eng. J. 27 (1), 24–32. Cucarella, V., Renman, G., 2009. Phosphorus capacity of filter materials used for onsite wastewater treatment determined in batch experiments—a comparative study. J. Environ. Qual. 38, 381–392. Diebel, M.W., Maxted, J.T., Nowak, P.J., Vander Zanden, M.J., 2008. Landscape planning for agricultural non-point source pollution reduction I: A geographical allocation framework. Environ. Manage. 42, 789–802. Drizo, A., Frost, C., Grace, J., Smith, K., 1999. Physico-chemical screening of phosphate-removing substrates for use in constructed wetland systems. Water Res. 33, 3595–3602. Dunne, E.J., Culleton, N., O’Donovan, G., Harrington, R., Daly, K., 2005. Phosphorus retention and sorption by constructed wetland soils in Southeast Ireland. Water Res. 39 (18), 4355–4362. Giesler, R., Andersson, T., Lövgren, L., Persson, P., 2005. Phosphate sorption in aluminium- and iron-rich humus soils. Soil Sci. Soc. Am. J. 69, 77–86. Gray, S., Kinross, J., Read, P., Marland, A., 2000. The nutrient assimilative capacity of Maerl as a substrate in constructed wetland systems for wastewater treatment. Water Res. 34 (8), 2183–2190.

Gregory, J., Duan, J., 2001. Hydrolyzing metal salts as coagulants. Pure Appl. Chem. 73 (12), 2017–2026. Heal, K.V., Dobbie, K.E., Bozika, E., McHaffie, H., Simpson, A.E., Smith, K.A., 2005. Enhancing phosphorus removal in constructed wetlands with ochre from mine drainage treatment. Water Sci. Technol. 51 (9), 275–282. Hedstöm, A., Rastas, L., 2006. Methodological aspects of using blast furnace slag for wastewater phosphorus removal. J. Environ. Eng. 132, 1431–1438. Heikkinen, K., Ihme, R., 1995. Retention of organic Fe–P-colloids from peat mining water in an overland flow wetland treatment system in northern Finland. Arch. Hydrobiol. 134 (4), 547–560. Heikkinen, K., Ihme, R., Osma, A., Hartikainen, H., 1995. Phosphate removal by peat from peat mining drainage water during overland flow wetland treatment. J. Environ. Qual. 24 (4), 597–602. Heikkinen, K., Savolainen, M., Ihme, R., Lakso, E., 2002. Use of the overland flow wetland treatment system for the purification of runoff water from peat mining areas—the service life of the system. In: Mander, U., Jenssen, P.D. (Eds.), Natural Wetlands for Wastewater Treatment in Cold Climates. WIT Press, Southampton, Boston, pp. 29–45. Herrmann, I., Jourak, A., Hedström, A., Lundström, T.S., Viklander, M., 2013. The effect of hydraulic loading rate and influence source on binding capacity of phosphorus filters. PLoS ONE 8 (8), 1–8. Huttunen, A., Heikkinen, K., Ihme, R., 1996. Nutrient retention in the vegetation of an overland flow treatment system in northern Finland. Aquat. Bot. 55 (1), 61–73. Ilina, S.M., Viers, J., Lapitsky, S.A., Mialle, S., Mavromatis, V., Chmeleff, J., Brunet, P., Alekhin, Y.V., Isnard, H., Pokrovsky, O.S., 2013. Stable (Cu, Mg) and radiogenic (Sr, Nd) isotope fractionation in colloids of boreal organic-rich waters. Chem. Geol. 342, 63–75. Ingri, J., Widerlund, A., Land, M., Gustafsson, Ö., Andresson, P., Öhlander, B., 2000. Temporal variations in the fractionation of the rare earth elements in a boreal river: role of colloidal particles. Chem. Geol. 166, 23–45. Johansson, L., Gustafsson, J.P., 2000. Phosphate removal using blast furnace slags and opoka—mechanisms. Water Res. 34, 259–265. Kadlec, R.H., 2006. Free surface wetlands for phosphorus removal: the position of the Everglades Nutrient Removal Project. Ecol. Eng. 27 (4), 361–379. Karjalainen, S.M., Kløve, B., Heikkinen, K., Kettunen, A., Amundsen, C.-E., 2003. Phosphate sorption properties of northern wetland soils. In: Mander, Ü., Vohla, C., Poom, A. (Eds.), Proceeding of the International Conference on Constructed and Riverine Wetlands for Optimal Control of Wastewater at Catchment Scale. September 29–October 2, Tartu, Estonia. Publicationes Instituti Geographici, Universitatis Tartuensis 94, University of Tartu, Tartu, Estonia, pp. 111–115. Klimeski, A., Uusitalo, R., Turtola, E., 2014. Screening of Ca and Fe-rich materials for their applicability as phosphate-retaining filters. Ecol. Eng. 68, 143–154. Kõiv, M., Liira, M., Mander, Ü., Mõtlep, R., Vohla, C., Kirsimäe, K., 2010. Phosphorus removal using Ca-rich hydrated oil shale ash as filter material—the effect of different phosphorus loadings and wastewater compositions. Water Res. 44, 5232–5239. Lai, D.Y.F., Lam, K.C., 2009. Phosphorus sorption by sediments in a subtropical constructed wetland receiving stormwater runoff. Ecol. Eng. 35 (5), 735–743. Lee, M.S., Drizo, A., Rizzo, D.M., Druschel, G., Hayden, N., Twohig, E., 2010. Evaluating the efficiency and temporal variation of pilot-scale constructed wetlands and steel slag phosphorus removing filters for treating dairy wastewater. Water Res. 44 (14), 4077–4086. Liang, H., Liu, J., Wei, Y., Guo, X., 2010. Evaluation of phosphorus removal from wastewater by soils in rural areas in China. J. Environ. Sci. 22 (1), 15–22. Nieminen, M., Penttilä, T., 2004. Inorganic and organic fractions in peat from drained mires in northern Finland. Silva Fenn. 38 (3), 243–251. Niskanen, R., 1989. Extractable aluminium, iron and manganese in mineral soils: II. Extrability by oxalate and pyrohosphate. J. Arric. Sci. Finl. 61, 79–87. Oliveira, M., Araujo, A., Azevedo, G., Pereira, M.F.R., Neves, I.C., Machado, A.V., 2015. Kinetic and equilibrium studied of phosphorus adsorption: effect of physical and chemical properties of adsorption agent. Ecol. Eng. 82, 527–530. Pernitsky, D.J., Edzwald, J.K., 2006. Selection of alum and polyaluminium coagulants: principles and application. J. Water Supply: Res. Technol.—Aqua 55 (2), 121–141. Pokrovsky, O.S., Schott, J., 2002. Iron colloids/organic matter associated transport of major and trace elements in small boreal rivers and their estuaries NW Russia. Chem. Geol. 190, 141–179. Reddy, K.R., DeLaune, R.D., 2008. Biogeochemistry of Wetlands: Science and Applications. CRC Press, Taylor & Francis Group, Boca Raton, FL. Ronkanen, A.-K., Kløve, B., 2005. Hydraulic soil properties of peatlands treating municipal wastewater and peat harvesting runoff. Suo 56 (2), 43–56. Ronkanen, A.-K., Kløve, B., 2009. Long-term phosphorus and nitrogen removal processes and preferential flow paths in Northern constructed peatlands. Ecol. Eng. 35 (5), 843–855. Sakadevan, K., Bavor, H., 1998. Phosphate adsorption characteristics of soils, slags, and zeolites to be used as substrates in constructed wetland systems. Water Res. 32, 393–399. Seo, D.C., Cho, J.S., Lee, H.J., Heo, J.S., 2005. Phosphorous retention capacity of filter media for estimating the longevity of constructed wetland. Water Res. 39 (11), 2445–2447. Silvan, N., Vasander, H., Laine, J., 2004. Vegetation is the main factor in nutrient retention in a constructed wetland buffer. Plant Soil 258 (1), 179–187. Silvan, N., Vasander, H., Karsisto, M., Laine, J., 2003. Microbial immobilisation of added nitrogen and phosphorus in constructed wetland buffer. Appl. Soil Ecol. 24 (2), 143–149.

A.-K. Ronkanen et al. / Ecological Engineering 86 (2016) 190–201 Son, J.H., Crowley, C., Goodwin, S., Arabi, M., Carlson, K.H., 2013. Relative phosphorus load inputs from wastewater treatment plants in a northern Colorado watershed. J. Environ. Qual. 42, 497–506. Søvik, A.K., Kløve, B., 2005. Phosphorus retention processes in shell sand filter systems treating municipal wastewater. Ecol. Eng. 25 (2), 168–182. USEPA (US Environmental Protection Agency), 2000. National water quality inventory report. In: EPA 841/R-02/001. Office of Water. Uusi-Kämppä, J., Braskerud, B., Jansson, H., Syversen, N., Uusitalo, R., 2000. Buffer zones and constructed wetlands as filters for agricultural phosphorus. J. Environ. Qual. 29 (1), 151–158. Vohla, C., Kõiv, M., Bavor, H.J., Chazarenc, F., Mander, Ü., 2010. Filter materials for phosphorus removal from wastewater in treatment wetlands—a review. Ecol. Eng. 37, 70–89.

201

Vymazal, J., 2007. Removal of nutrients in various types of constructed wetlands. Sci. Total Environ. 380 (1–3), 48–65. Vymazal, J., 2011. Constructed wetlands for wastewater treatment: five decades of experience. Environ. Sci. Technol. 45 (1), 61–69. Wisawapipa, W., Kheoruenromne, I., Suddhiprakarn, A., Gilkes, R.J., 2009. Phosphate sorption and desorption by Thai upland soils. Geoderma 153, 408–415. Xu, D., Xu, J., Wu, J., Muhammad, A., 2006. Studies on the phosphorus sorption capacity of substrates used in constructed wetland systems. Chemosphere 63 (2), 344–352. Yuan, G., Lavkulich, L.M., 1994. Phosphate adsorption in relation to extraction iron and aluminium in spodosols. Soil Sci. Soc. Am. J. 58, 343–346.