Biol Fertil Soils (2001) 34:42–48 DOI 10.1007/s003740100369

O R I G I N A L PA P E R

Lars D. Hylander · Gyula Simán

Plant availability of phosphorus sorbed to potential wastewater treatment materials

Received: 24 August 2000 / Published online: 29 May 2001 © Springer-Verlag 2001

Abstract Ecologically engineered wastewater treatment facilities, such as constructed wetlands and infiltration plants, can be further improved in their P retention by using reactive media with a high P-retention capacity. In a sustainable system, the sorbed P should be recycled in agricultural production. The objective of the present study was to determine the plant availability of P sorbed to different P-retention media. The studied media were: crystalline and amorphous blast furnace slag, natural and burned opoka (a bedrock from Poland), limestone, burned lime, soil from a spodic B horizon, and light expanded clay aggregates (LECA). They were soaked in a P solution, rinsed and dried before incorporation into soil. An additional aim was to compare P taken up by barley with amounts chemically extracted for the estimation of plant-available soil P. P sorbed to the crystalline slag was delivered to the barley plants more efficiently than P added in K2HPO4 fertiliser. Soil extraction with acid ammonium lactate correlated well with P taken up by barley and indicated that P bound to Ca is more available to plants than P bound to Al and Fe. The Mg content of the used slag may replace Mg fertilisation in certain soils. It was concluded that among the investigated filter materials, crystalline slag was the most suitable sorbent from an agricultural point of view, since it possessed a large P-sorption capacity and the sorbed P was largely plant available. The heavy metal content of sorption materials must be examined carefully before their application to agricultural soils. Keywords Phosphate sorption · Reactive media · Plant phosphorus uptake · Chemical extraction L.D. Hylander (✉) Department of Limnology, Evolutionary Biology Centre, Uppsala University, Norbyvägen 20, 752 36 Uppsala, Sweden e-mail: [email protected] Tel.: +46-18-4712710, Fax: +46-18-531134 G. Simán Division of Plant Nutrition, Department of Soil Science, Swedish University of Agricultural Sciences, P.O. Box 7014, 750 07 Uppsala, Sweden

Introduction P in the wastewater from households is an important source of pollution in many countries (Swedish Environmental Protection Agency and Ministry of Foreign Affairs 1998; Vymazal et al. 1998). Wetlands have received renewed attention as a low-cost and robust alternative, which are not dependent on advanced engineering, for sewage treatment. The efficiency of reducing the P content of wastewater is often increased by incorporating a suitable P-sorbing material in the soil profile, generally an Al, Fe or Ca compound (Brix 1994; Vymazal et al. 1998; Zhu et al. 1997; Zurayk et al. 1997). A related strategy is to trap P in infiltration beds, by substituting commonly used sand for a material with a large P-sorption capacity (Johansson 1999a; Tofflemire and Chen 1977). The sorbing material has to be replaced when few unoccupied P-sorption sites remain and its sorption efficiency decreases. At present, this P-enriched material is generally dumped at landfill sites. To achieve a sustainable society, P from wastewater should be recycled back to agriculture (Swedish Environmental Protection Agency and Ministry of Foreign Affairs 1998). After taking into consideration the possible content of pathogens, toxic compounds and heavy metals, the P-enriched sorbing materials may be used as fertilisers in agriculture, provided the sorbed P is available to the crop. P-sorbing materials with additional beneficial characteristics may be used, for example, to improve soil structure or as a liming material. The objective of this study was to determine the plant availability of P sorbed to eight different materials compared with P added as K2HPO4. The materials tested were those used in constructed wetlands to retain P from wastewater (Johansson 1999a, 1999b). The availability of soil P was estimated by chemical extraction [acetic acid/ammonium lactate (AL) and 2 M HCl] and determined from the yield response and P content of barley shoots in a pot experiment. Plant P is compared with total soil P and with the two soil-P fractions.

43

Materials and methods The availability of the sorbed P to barley plants was determined in a pot experiment. Mitscherlich pots were filled (4.25 kg dry soil pot–1) with the A horizon (5–20 cm) of a P-depleted agricultural soil under permanent lay from Bjärröd, Scania, Sweden (55°42′N, 13°43′E, altitude 105 m). The soil, classified as a Boroll (USDA) or a Phaeozem (WRB), is a sandy moraine with 14% clay, pHH2O 6.3, C/N 10, and acetic acid/AL- (Egnér et al. 1960) extractable P (P-AL), K, and Mg, i.e. 9, 50, and 66 µg g–1 dry soil, respectively, corresponding to 0.29, 1.28 and 2.72 µmol g–1 dry soil. Each pot received 1.5 g K2SO4, 0.5 g MgSO4, and 1.0 g N as NH4NO3 as basic fertilisation, first incorporated into a small amount of soil, which was then mixed with all the soil of a pot. After the addition of P and other amendments, the soil was watered (80% of water holding capacity) and left to equilibrate before sowing. The soil was mixed with P in the form of K2HPO4, either on its own or sorbed to different materials in quantities corresponding to 0.03 or 0.3 g P pot–1, with each treatment in triplicate (Table 1). Three of the sorbing materials were taken from a column experiment with slag materials, which had received applications of 10 mg P l–1 solution every second hour for 13 months (Johansson 1999b). The other materials were soaked in 1,000 mg P l–1 solution and intermittently shaken for 24 h (material:solution ratio 1:4 for CaO due to a high sorption capacity, 1:2 for the other materials), dried (<35°C), rinsed with distilled water (material: solution 1:2) and dried again. The materials were (Table 1): crystalline slag 0–0.125 mm (CSF); crystalline slag 0.25–4 mm (CSC); amorphous slag 0.25–4 mm (ASC); limestone; burned lime; opoka (a bedrock from Poland, rich in CaCO3 and SiO2 and formed from marine sediments; either burned, natural); the B horizon of a podsolic forest soil; and light expanded clay aggregates (LECA). Total P sorbed to the different materials was determined by extraction with 7 M HNO3 in an autoclave (120°C for 30 min, material:solution 1:20; SIS 1986), followed by the determination of P with an inductively coupled plasma-atomic emission spectrometer (ICP-AES) (Perkin Elmer 1993). Barley, Hordeum vulgare L. cv. Pernilla, was sown on 6 June 1998, and thinned (22 June) to 40 plants pot–1. The plants were kept outdoors to expose them to natural light conditions, and were protected from rain and damage by birds with nets. They were watered once a day or every second day depending on the rate of evapotranspiration, so that soil moisture was kept between 60% and

Table 1 Total element concentration of original sorption materials (Johansson 1999b; Johansson and Hylander 1998), pH in water suspension and P extracted by 7 M HNO3 (P-HNO3) from the materials after sorption of P. OpB Burned opoka, OpN natural opoka, Podsol B horizon of a podsolic forest soil, LECA light expanded clay aggregates, ASC amorphous slag 0.25–4 mm, CSC crystalline Added materials

Al

Ca

Fe

Mg

Mn

Si

80% of water holding capacity. The average air temperature during the 47-day growth period was 14.5°C, and the sun shone for 254 h from a height >5° (Meteorologiska observationer vid Ultuna N 59°49′E 17°39′ 1998). The pots were re-randomised weekly. Mn solution (2.4 g l–1 as SO42–) was sprayed on the leaves twice (7 and 10 July) during the growing season. At harvest, the plants were cut 10 mm above the soil surface on 23 July 1998, dried at 55°C, weighed, milled, and digested in concentrated HNO3 with a block digester set at 125°C for 5.5 h. The Al, B, Ca, Cu, Fe, K, Mg, Mn, P, S, and Zn contents of the digests were determined using an ICP-AES (Perkin Elmer 1993). N was determined by dry combustion on a LECO CNS-2000 analyser (LECO 1994). Chemical extraction for plant-available P was performed on original soil, and on soil mixed with P-sorbing materials after harvest, to estimate plant availability of P at the time of maximal root mass in soil. The extractants used were acetic acid/AL solution (pH 3.75, 0.40 M acetic acid and 0.10 M AL; material:solution 1:20, 90 min shaking; Egnér et al. 1960) and 2 M HCl (material:solution 1:25, 2 h extraction in a boiling-water bath with intermittent shaking; Egnér et al. 1960). P was determined as a molybdate-phosphate complex on a spectrophotometer (Technicon Scandinavia 1976). Soil pH was determined in a 1:2.5 (w:v) soil:water suspension. Data were subjected to ANOVA and to linear regression (Statistix for Windows 1996).

Results and discussion P uptake from different sources by barley The poor P status of the soil was demonstrated by a significantly increased dry matter yield at the two rates of P added, with the exception of the application of P sorbed to CaO and to CaCO3 (Fig. 1). The application of 0.2 µmol P g–1 soil recently sorbed to crystalline slag gave even higher yields than the application of fertiliser P (Fig. 1). The reason for this higher yield may have been the release of P from the slag at a rate corresponding to the P requirements of the growing plants. A sub-

slag 0.25–4 mm, CSF crystalline slag 0–0.125 mm, CSFcol mixed with sand in a 1:1 ratio, P sat. saturation with 1,000 mg 1–1 P solution for 24 h, Col daily applications of 10 mg l–1 P solution for 13 months percolating through columns, Col and P sat. First col then P sat., n.a. not analysed pH

Means of P addition

(mg g–1)

P-HNO3±SE (n=2) (µg g–1)

CaO CaCO3 OpB OpN Podsol LECA ASC

0 0 29 16 69 96 51

710 400 360 200 11 15 250

0 0 13 10 24 43 3

0 0 5 3 5 15 83

0 0 <1 <1 1 1 3

0 0 280 160 n.a. n.a. 166

CSC

60

239

3

86

4

164

CSF

56

241

3

87

4

164

13.0 10.0 12.7 7.9 6.6 9.2 8.9 9.6 9.9 8.1 9.3 10.0 8.2

P sat. P sat. P sat. P sat. P sat. P sat. Col Col and P sat. P sat. Col Col and P sat. P sat. Col

2,290±50 781±71 1,322±20 1,383±59 1,568±18 498±18 990±73 1,750±72 524±56 961±3 1,268±38 620±8 394±34

44 Table 2 Element concentration in shoots of barley at heading stage and recovery of P in shoot after soil addition of P sorbed to different materials compared to fertiliser P. Soil No P-sorbing material; NoP no P-fertilisation; +P 0.2 µmol P g–1 soil; ++P 2.3 µmol P g–1 soil; P sorbed to the indicated materials by Treatmenta Soil No P CaO NoP Soil +P CaO +P CaCO3 +P OpB +P OpN +P Podsol +P LECA +P ASC col +P ASC colP +P CSC +P CSC col +P CSC colP +P CSF +P CSF col +P Soil ++P CaO ++P OpB ++P OpN ++P Podsol ++P LECA ++P CSCcolP ++P Higher limita Lower limita Deficiencya a Reuter

P±SE (µmol g–1) 33±2 47±2 39±0 41±1 32±0 40±1 40±1 38±1 39±3 42±2 41±1 42±2 45±0 39±0 38±1 40±1 71±2 35±2 37±1 39±4 48±3 45±2 68±7 >161 <61 <48

Recovery (%) – – 44 34 19 38 37 31 33 45 52 69 56 51 57 64 23 1 5 8 12 12 21 – – –

Ca±SE (µmol g–1) 208±11 321±19 176±12 228±5 248±10 210±9 218±11 200±6 198±7 180±14 184±4 205±6 196±5 204±7 209±4 207±9 160±4 291±15 218±8 168±14 169±7 166±8 170±13 >300 <75 –

soaking for 24 h in a P solution [1,000 mg l–1 or col daily application of P solution (10 mg l–1) in a column]; CaO NoP soil without P application but limed as in ++P; colP material from a column and then soaked in P-solution (1,000 mg l–1) Mg±SE (µmol g–1) 53±2 44±3 49±2 55±1 60±3 52±1 56±3 52±1 51±1 49±4 53±0 74±0 59±1 59±1 97±1 75±4 58±2 43±4 60±4 55±4 58±2 59±4 88±7 >210 <60 –

K±SE (µmol g–1)

N±SE (µmol g–1)

1,298±36 427±29 1,361±60 1,515±38 1,456±52 1,466±7 1,423±66 1,458±33 1,429±78 1,316±81 1,263±19 1,156±69 1,384±11 1,234±11 1,223±26 1,152±31 888±45 595±41 1,432±31 1,199±121 967±54 917±49 681±67 >800 <400 <300

3,971±61 4,112±30 3,448±137 3,667±13 3,533±56 3,421±82 3,431±39 3,874±24 3,786±62 3,412±70 3,319±71 3,088±112 3,317±17 3,369±71 3,145±112 3,167±69 2,043±55 3,876±19 3,226±19 2,721±84 2,279±48 2,171±71 1,993±178 >2,100 <1,400 <1,100

and Robinson (1986)

Fig. 1 Yield (average and 95% confidence interval) of barley plants (stem and leaves) grown in a soil treated with indicated amounts of P per gram soil sorbed to different materials until saturation (triplicates). Ten times more sorbing material at 2.3 µmol P and at no P sorbed to CaO, compared to 0.2 µmol P. Soil No P-sorbing material, OpB burned opoka, OpN natural opoka, Podsol B horizon of a podsolic forest soil, LECA light expanded clay aggregates, ASC amorphous slag 0.25–4 mm, CSC crystalline slag 0.25–4 mm, CSF crystalline slag 0–0.125 mm, CSFcol CSF mixed with sand in a ratio of 1:1, col the sorbent received 13 months of daily applications of P solution (10 mg l–1) in a column, colP material from a column and then soaked in P solution (1,000 mg l–1)

stantial part of the easily soluble fertiliser P had probably become less plant-available during the growth period due to strong sorption to soil particles. The shoot concentration of Mg was about 85 µmol g–1 dry matter (DM) in plants from all treatments with slag

applied at the high P level, as well as in treatments receiving fine-textured slag at the low P level (Table 2). This was significantly higher than in plants from other treatments, which had <60 µmol g–1 DM (Table 2). It is not probable that Mg deficiency had reduced the yield, since shoot concentrations did not diverge from the lower limit for barley (Reuter and Robinson 1986), because Mg had been applied before sowing (0.5 g MgSO4 pot–1). However, the Mg content of slag (Table 1) will increase yield when applied to soils poor in Mg and not receiving any other Mg fertiliser. The application of P sorbed to the Podsol or LECA resulted in a lower yield than the application of fertiliser P (Fig. 1). This can be explained by P bound to oxides and hydroxides of mainly Al, but also Fe in the Podsol and LECA, according to results from P fractionation (Johansson 1997) and total element content (Table 1). P in sewage sludge from municipal wastewater treatment is generally more available to agricultural crops on neutral to acid soils when P has been precipitated with Ca material than when Al or Fe salts have been used (Jokinen 1990; Soon and Bates 1982; Timmermann et al. 1981). The differences are most pronounced at low P-application rates because the plants then need to use more P than the easily soluble or exchangeable fraction available in sludge precipitated with Al and Fe salts (Jansson 1970). The plant availability of P decreases with time because of crystallisation, among other processes, when

45

monodentate bonds are transformed to bidentate bonds (Marschner 1995). Slags from columns, which had received applications of P for 13 months, were used in parallel treatments (col). As expected, the CSCcol had reduced plant availability, indicated by significantly lower yield in that treatment than in treatment CSC with freshly sorbed P at the lower P rate. However, there was no difference between freshly sorbed CSF and the aged CSFcol, which may be attributed to the finer texture of CSFcol compared to CSCcol. A finer texture increases the contact area between sorption material and soil, which facilitates the processes of roots and micro-organisms, so that sorbed P was possibly desorbed fast enough to eliminate any differences between sorption strength caused by ageing in CSF. The low yields after application of CaCO3 and CaO, and to a smaller degree after application of opoka (Fig. 1), were accompanied by symptoms of grey speck disease at the early development stage. The Mn deficiency was probably caused by the liming effect of many materials (Table 1, Fig. 2), resulting in oxidation and precipitation or adsorption of Mn, and subsequently lowered levels of plant-available Mn2+ (Marschner 1995). The Mn deficiency was probably aggravated by competition between Mn2+ and increased levels of Ca2+ (Marschner 1995). Burned lime (CaO) is a more efficient liming agent than CaCO3 when compared on an equal weight basis, but the symptoms of grey speck disease were more severe after application of CaCO3, since more CaCO3 than CaO was applied to reach the desired P application rate, due to less P sorbed to a given mass of CaCO3 (Table 1). The plants in treatments with slag had fewer or no symptoms of grey speck disease, although application of slag did result in a pH increase (Fig. 2). The main reason for the lower incidence of Mn deficiency with slag is a slower reaction in the soil due to its coarser texture. Even CSF had a coarser texture than CaCO3, CaO, and opoka. The coarse texture and lower intrinsic pH of the slag resulted in a lower soil pH at harvest in most treatments amended with slag compared with those receiving CaCO3, CaO or opoka (Fig. 2). Observations of treatments with grey speck disease during the growth period and the differences in yield at the high P-application rate between treatments with natural and burned opoka (Fig. 1) support the theory of rapidly reacting liming materials as important for worsening the grey speck disease. In treatments with slag, the large Mn content (Table 1) may also have contributed to reducing the occurrence of grey speck disease. Plants in all treatments were sprayed with Mn when the first symptoms of grey speck disease were observed, which resulted in optimal levels of Mn in plants of all treatments at the end of the growth period. The severity of grey speck disease was recorded when the symptoms appeared according to a four-grade scale, where 0 indicated no visible symptoms and 3 severe symptoms. There was a close and negative relation between yield and grey speck disease symptoms, indicating

Fig. 2 Soil pH (average and 95% confidence intervals) at harvest in a soil treated with indicated amounts of P per gram soil sorbed to different materials at planting (triplicates). Abbreviations as in Fig. 1

its hampering effect on yield. The symptoms were present especially in treatments at the low P level, indicating that barley plants may be more vulnerable to Mn deficiency when simultaneously facing deficiency of P. The concentration of P in shoots followed mainly the yield pattern, and was highest in treatments with the largest yield (Table 2). Hence the P uptake from soil was further diverging between different treatments, which is in agreement with results of He et al. (1994). In contrast, the N and K concentrations were generally lower in plants giving the largest yield (Table 2). This is an effect of dilution of the amounts of N and K taken up, but concentrations of N and K in shoots never dropped below adequate concentrations for spring barley (Reuter and Robinson 1986). An important exception from the low K concentrations of plants at high yields was seen in plants receiving the high rate of CaO. Plants in these treatments had the lowest K concentrations in spite of their low yield and the highest Ca concentrations. This indicated that Ca ions from the finely ground, and hence rather easily soluble CaO, had competitively reduced plant uptake of K. Chemical P extraction and estimated plant availability of P P-AL at the high P level was 4–8 times higher when P had been applied as sorbed to materials containing Ca compared to applications of Podsol or LECA with sorbed P (Fig. 3). One reason for this is a reduction of Pbinding sites on Al and Fe (hydr)oxides of soil (Barrow 1983) caused by the increase in pH by the materials containing Ca (Fig. 2). Application of CaO without P increased P-AL more than the high rate of fertiliser P alone, yet far less than a liming agent with sorbed P (Fig. 3). Part of the soil content of Ca phosphates is not easily plant available at pH values >7, where hardly plant-available hydroxyapatite occasionally will be formed (Marschner 1995). Even if the ultimate step to hydroxyapatite is slow in soil, hydroxyapatite appears to be formed within hours during the sorption process of P

46

Fig. 3 P extracted at harvest with acetic acid/ammonium-lactate (AL) (average and 95% confidence interval) in a soil treated with indicated amounts of P per gram soil sorbed to different materials at planting (triplicates). Other abbreviations as in Fig. 1

Fig. 4 P extracted at harvest with 2 M HCl (average and 95% confidence interval) in a soil treated with indicated amounts of P per gram soil sorbed to different materials at planting (triplicates). Abbreviations as in Fig. 1

Fig. 5 P extracted at harvest with acetic acid/AL (P-AL) compared to uptake of P by shoot in a soil treated with no P (● ● ; top panel), 0.2 µmol P g–1 dry soil (■ ■ ; top panel), or 2.3 µmol P g–1 dry soil (▲ ▲ ; bottom panel) sorbed to indicated materials at planting (triplicates). Abbreviations as in Figs. 1 and 3

to some of the filter materials (Johansson and Gustafsson 2000). This is due to the high pH of the P solution and the materials being submerged (Takeuchi and Komada 1998). Subsequently, the AL method overestimated plant-available P in these treatments, because the acid solution extracted hardly plant-available P in addition to plant-available P. Also, the less labile soil-P fraction extracted by HCl increased significantly with the application of large amounts of P sorbed to materials containing Ca, though the relative difference between treatments with P sorbed to materials containing Ca and treatments with fertiliser P or P sorbed to Podsol or LECA was smaller for HCl-extractable P than for P-AL (Fig. 4). Plant yields (Fig. 1) are in agreement with extractable P (Fig. 3), except for treatments with CaO and CaCO3, where grey speck disease largely reduced the yield. The variations in plant yield in the other treatments were explained to 52% by variations in P-AL when including all three P application rates in the regression. Regression with yield data from the P application rate of 0.2 µmol g–1 only, showed that 65% of the variation was explained by the P-AL values. No significant values were obtained for a regression between yield and P-AL values at the highest P-application rate, which may indicate that other

factors than P-AL were important for the differences in yield at this high content of soil P. Soil from all treatments supplied with 0.2 µmol P g–1 soil at planting contained about 0.2 µmol P-AL g–1 soil at harvest (Fig. 5a). However, P taken up by shoots varied from just above 0.04 to close to 0.16 µmol P g–1 soil, depending on with which sorption material the P had been applied, the slags being the most advantageous ones for P uptake (Fig. 5a). The application of lime only, without sorbed P, resulted in an increased soil pH (Fig. 2) and extractable P-AL (Fig. 3), but without increased plant uptake of P (Fig. 5a). This was due to the high extractability of Ca-phosphates with the AL solution and P reactions at high soil pH already discussed, and due to the release of P through oxidation of organic matter (Marschner 1995). At a tenfold higher application rate of P (2.3 µmol P g–1), the P-AL values also increased tenfold in treatments with P sorbed to materials containing Ca, but only about twofold when fertiliser-P or P sorbed to LECA or Podsol was supplied (Fig. 5b). The slags were also superior in releasing P to the plants at the high P-application rate. P-AL correlated with P taken up by shoots when comparing all data. No significant correlation was found

47

between extractable P and P in shoots within a certain P-application rate. This was because large amounts of P were extracted from soil amended with sorption materials containing Ca without any corresponding increase in the P content of shoots compared with the other treatments. In agreement with our results, Johansson and Hylander (1998) showed that P sorbed to Ca in opoka, slag, and limestone, was extracted by AL solution to a higher degree than P bound to Al in LECA or to material of the Podsol, which is rich in Al and contains some Fe (Table 1). Hahlin and Johansson (1977) demonstrated a correlation between P-AL and yield in both acid soils and neutral-alkaline soils in more than 100 field experiments run for 6 years. P-AL increased in all soils after P fertilisation, and in the acid soils it also increased after liming without any P application. Timmermann et al. (1981) found that citric acid extracted >90% of the P in a sewage sludge precipitated with Ca(OH)2, but <70% when the P was precipitated with Al salts. This was probably a pH effect rather than a reaction involving the formation of complexes. The former should anyhow be considered, since plant roots have the capacity to reduce the pH in the rhizosphere, thereby acquiring hardly plant-available Ca-phosphates (Marschner 1995). Sewage sludge P bound to Ca was more soluble in a simulated soil solution than P bound to Al and Fe under aerobic environments (Rydin 1996). The lower plant availability of P bound to Al and Fe (He et al. 1994) indicates that the extremely high P-sorption capacity of certain Al and Fe oxide/hydroxides such as allophane (Hylander 1997; Wada and Gunjigake 1979) should not be used in the proposed wastewater treatment system (Mann and Bavor 1993; Vymazal et al. 1998), since it will reduce the efficiency with which sorbed P can be recycled in agriculture. Feasibility for use in practice The hydraulic conductivity of the different materials is important for their practical uses. A large particle size, like that of CSC and ASC, results in a high hydraulic conductivity (Johansson 1999a), which reduces the risk of clogging. A coarse structure will, on the other hand, reduce the area active in sorbing P, as well as the contact area with soil, and thereby possibly reduce the plant availability of sorbed P when used as fertiliser. The hydraulic conductivity will determine the wastewater flow through a filter; hydraulic conductivity which is too high results in reduced P sorption (Johansson 1999b). The sorption of P was more effective when P was progressively applied at low concentrations than as a single application at high concentration. Of the studied sorbents, blast furnace slags were most suitable for use in practice, since they have a large P-sorption capacity, and readily released sorbed P to plants when applied to soil. There were only small differences in plant availability of sorbed P between the different slag preparations. As when using other easily sol-

uble P fertilisers, the amounts of applied P must balance plant uptake to prevent P run-off and eutrophication of surface waters. P from sewage sludge dumped on landfill sites will become an important source of P leakage in the future (Günter 1999). It is necessary to use materials with low contents of heavy metals, such as those studied in this work, for applications of filter materials to agricultural soils. Any subsequent clean up of heavy metals from soil is expensive. The heavy metal content of the wastewater must also be observed, since certain materials may sorb not only P but also heavy metals. From a sustainability viewpoint, heavy metals should not be added to agricultural soils, even if their present leaching potential is low, because soil microbiological and chemical conditions may increase the leaching potential of heavy metals initially applied in a hardly soluble state. The continuing acidification in humid areas and possibly less resources for liming soils in the future, will increase the leaching and plant uptake of most heavy metals present in soil. Bacteria in wastewater will be killed when treated in calcareous filter materials. The bacteria content of wastewater percolated through filter materials in a pilot plant was reduced by 99.5% in the effluent, which had a pH of 11.2–11.4 (Cucarella Cabañas 2000). A cementation reaction of the slag materials during the sorption processes was observed when using a P solution prepared from laboratory chemicals. This may complicate their practical application. However, no cementation was observed when wastewater percolated through the materials, probably because organic matter present in wastewater reduces cementation. In conclusion, materials such as opoka and slag, which contain Ca, have a high sorption capacity for P and are a cost-efficient way of removing P from wastewater. The pH-increasing effect of calcareous materials reduces the pathogen population in treated wastewater. If sorption materials with negligible contents of heavy metals are used, and if they do not sorb heavy metals when filtering wastewater, they can be used as P fertilisers when their P-sorption capacity has decreased. The present study indicates that P sorbed to blast furnace slag is readily available to plants. The amount of applied sorbent needs, in many cases, to be based on its liming effect rather than on its P content. Further research is needed regarding the plant availability of sorbed P in different soil types under field conditions and after ageing of the filter substrates. Acknowledgements The study was jointly financed by the Swedish Council for Building Research, the Swedish Farmers’ Foundation for Agricultural Research, and VA- Forsk. It is a contribution to the on-going project “Removal of phosphorus from wastewater using reactive filter media and recycling to agriculture”, a joint study with a research group at the Royal Institute of Technology, Stockholm, headed by Associate Prof. Gunno Renman and financed by the Swedish Council for Forestry and Agricultural Research and the Swedish Environmental Protection Agency. SSAB Merox, Oxelösund, Sweden, supplied the blast furnace slag.

48

References Barrow NJ (1983) A mechanistic model for describing the sorption and desorption of phosphate by soil. J Soil Sci 34:733– 750 Brix H (1994) Constructed wetlands for municipal wastewater treatment in Europe. In: Mitsch WJ (ed) Global wetlands: old world and new. Elsevier, Amsterdam, pp 325–333 Cucarella Cabañas VM (2000) Use of sorbents for the removal of phosphorus and bacteria from wastewater. MSc thesis report 2000:16. Division of Land and Water Resources, Royal Institute of Technology, Stockholm, Sweden Egnér H, Riehm H, Domingo WR (1960) Untersuchungen über die chemische Bodenanalyse als Grundlage für die Beurteilung des Nährstoffzustandes der Böden. II. Chemische Extraktionsmethoden zur Phosphor- und Kaliumbestimmung. K Lantbrukshögsk Ann 26:199–215 Günther F (1999) Phosphorus management and societal structure. Hampered effluent accumulation processes (HEAP) in different areas of the Swedish society. Vatten 54:199–208 Hahlin M, Johansson L (1977) The ability of some methods of analysis to describe the phosphorus and potassium status in the soil (in Swedish; English summary). Reports of the Agricultural College of Sweden, A, no. 271. Agricultural College of Sweden, Uppsala He ZL, Yang X, Yuan KN, Zhu ZX (1994) Desorption and plantavailability of phosphate sorbed by some important minerals. Plant Soil 162:89–97 Hylander LD (1997) Comparison of methods for determination of phosphorus retention. In: Ando T, Fujita K, Mae T, Matsumoto H, Mori S, Sekiya J (eds) Plant nutrition – for sustainable food production and environment. Kluwer, Dordrecht, pp 365–366 Jansson SL (1970) Gödslingsförsök med kemiskt fällda slamtyper. Hyg Rev 59:350–356 Johansson L (1997) The use of LECA (light expanded clay aggregates) for the removal of phosphorus from wastewater. Water Sci Technol 35:87–93 Johansson L (1999a) Industrial by-products and natural substrata as phosphorus sorbents. Environ Tech 20:309–316 Johansson L (1999b) Blast furnace slag as phosphorus sorbents – column studies. Sci Total Environ 229:89–97 Johansson L, Gustafsson JP (2000) Phosphate removal from wastewaters using blast furnace slags and opoka – mechanisms. Water Res 34:259–265 Johansson L, Hylander L (1998) Phosphorus removal from wastewater by filter media: retention and estimated plant-availability of sorbed phosphorus. Zesz Proble Poste¸pów Nauk Roln 458:397–409 Jokinen R (1990) Effect of phosphorus precipitation chemicals on characteristics and agricultural value of municipal sewage sludges. Acta Agric Scand 40:123–14 LECO (1994) CNS-2000 elemental analyzer instruction manual, version 4.0. LECO, Chicago, Ill.

Mann RA, Bavor HJ (1993) Phosphorus removal in constructed wetlands using gravel and industrial waste substrata. Water Sci Technol 27:107–113 Marschner H (1995) Mineral nutrition of higher plants, 2nd edn. Academic Press, London Meteorologiska observationer vid Ultuna N 59°49′E 17°39′ (1998) Månadsrapport juni, juli, augusti 1998. Ultuna meteorologiska station, Swedish University of Agricultural Sciences, Uppsala Perkin Elmer (1993) ICP-OES Perkin Elmer Optima 3000 DV hardware guide. Perkin Elmer, Norwalk Reuter DJ, Robinson JB (eds) (1986) Plant analysis: an interpretation manual. Inkata Press, Melbourne Rydin E (1996) Experimental studies simulating potential phosphorus release from municipal sewage sludge deposits. Water Res 30:1695–1701 SIS (1986) Metal content of water, sludge and sediment – determined by flameless absorption spectrometry – electrothermal atomization in a graphite furnace – general principles and guidelines (in Swedish). Svensk standard SS 02 81 83. Standardiseringskommissionen i Sverige, Stockholm Soon YK, Bates TE (1982) Extractability and solubility of phosphate in soils amended with chemically treated sewage sludges. Soil Sci 134:89–96 Statistix for Windows (1996) Analytical software, Tallahassee, Fla. Swedish Environmental Protection Agency and Ministry of Foreign Affairs (1998) Water and wastewater treatment, the Swedish experience. Graphium Norstedts Tryckeri, Stockholm Takeuchi M, Komada M (1998) Phosphorus removal from hoggery sewage using natural calcium carbonate. Jpn Agric Res Q 32:23–30 Technicon Scandinavia (1976) Product labeling for Technicon Autoanalyzer II continuous-flow analytical instrument. Technical publication no. UA4-0170-01. Technicon Instruments, Tarrytown Timmermann F von, Cervenka L, Baran E (1981) Phosphatrückgewinnung aus Abwässern und Einsatz der Fällungsprodukte in der Düngung (P-recovery from waste water and use of the precipitates for fertilization) (in German, English summary). Landwirtsch Forsch, Sonderh 37:176–185 Tofflemire TJ, Chen M (1977) Phosphate removal by sands and soils. Ground Water 15:377–387 Wada K, Gunjigake N (1979) Active aluminum and iron and phosphate adsorption in Ando soils. Soil Sci 128:331–336 Vymazal J, Brix H, Cooper PF, Green MB, Haberl R (eds) (1998) Constructed wetlands for wastewater treatment in Europe. Backhuys, Leiden Zhu T, Jenssen PD, Mæhlum T, Krogstad T (1997) Phosphorus sorption and chemical characteristics of lightweight aggregates (LWA) – potential filter media in treatment wetlands. Water Sci Technol 35:103–108 Zurayk R, Nimah M, Geha Y, Rizk C (1997) Phosphorus retention in the soil matrix of constructed wetlands. Commun Soil Sci Plant Anal 28:521–535