Available online at www.sciencedirect.com SCIENCE @DIRECT• ELSEVIER ENVIRONMENTAL POLLUTION Environmental Pollution 141 (2006) 387-395 www .elsevier.com/locate/envpol Invited paper Woody biomass phytoremediation of contaminated brownfield land Christopher J. French a, Nicholas M. Dickinson a,*, Philip D. Putwain b b "School of Biological and Earth Sciences , Lil'e1pool Jolin Moores Unil·ersity, Byrom Street, Li1•eJpool L3 3AF, UK Ecological Restoration Consultants (ERC), Ness Botanic Gardens, Uni1•ersity of Lil·e1pool, Ness, Cheshire CH64 , UK Received 6 May 2005; received in revised form 17 August 2005; accepted 28 August 2005 Field trials show short-rotation coppice provides effective risk management and remediation solutions to hotspots of residual metal and As contamination of brownfield land. Abstract Economic and environmental regeneration of post-industrial landscapes frequently involves some element of re-afforestation or tree planting. We report field trials that evaluate whether woody biomass production is compatible with managing residual trace element contamination in brownfield soils. Large-scale mapping of contamination showed a heterogenous dispersion of metals and arsenic, and highly localised within-site hotspots. Yields of Sali.r, Populus and Alnus were economically viable, showing that short-rotation coppice has a potentially valuable ro le in community forestry. Mass balance modelling demonstrated that phytoextraction potentially could reduce contamination hotspots of more mobile elements (Cd and Zn) within a 25-30-year life cycle of the crops. Cd and Zn in stems and foliage of Sali.r were 4-13 times higher than EDTA-extractable soil concentrations. Lability of other trace elements (As, Pb, Cu, Ni) was not increased 3 years after planting the coppice; woody biomass may provide an effective reduction of exposure (phytostabi lisation) to these less mobile contaminants. © 2005 Elsevier Ltd. All rights reserved. Keywords: Brownfield ; Heavy metals; Trees; Coppice 1. Introduction Increased tree cover offers low cost and effective solutions to derelict landscapes and degraded environments (Cannell, 1999; Dickinson, 2000b; Paulson et al., 2003). Regeneration and urban greening of former industrial landscapes in lowland England have been achieved through substantial tree planting; in North-West England, the first 15 years of a national Community Forest programme is halfway towards transforming a landscape from 4% to 12% tree cover, with about 250 ha annum - t of trees being planted within the Mersey Forest region that surrounds Liverpool (Anon., 2003; Putwain et al., 2003; Rawlinson et al., 2004). An increasing proportion of this tree planting is taking place on an estimated 11,000 ha of brownfield land in the region, much of which has residual * Corresponding author. Tel. : +44 151 2312190; fax: +44 151 2073224. E-nwil address: n.m .dickinson@livjm.ac.u k (N.M. Di ckin son). 0269-7491/$ - see front matter © 2005 Elsevier Ltd. All rights reserved. doi: 10.10 16/j.envpol.2005.08.065 contamination from former land use (Fig. I). It has been speculated that tree planting may resolve soil contamination issues, either through extraction or immobilisation (Lepp and Dickinson, 1998; Vangronsveld and Cunningham, 1998; Bardos et al., 1999; Dickinson, 2000a; Pulford and Watson, 2003; Berndes et al., 2004). In particular, there is growing evidence that clean-up of cadmium can be achieved using short-rotation coppice (SRC) Salix (willow) clones for phytoextraction (Dickinson and Pulford, 2004). However, despite significant experimental data from laboratory and pot experiments, there is still a real paucity of supportive field evidence. On most brownfield land in the UK, contamination levels are not excessively high and, in fact, may not meet the statutory definition of contaminated land (Environment Agency, 2002). Risk assessment criteria for human health and controlled waters, based on a pollutant- pathway-receptor model, are perceived to be less onerous for soft end-use remediation , compared to situation with more obvious linkages; for example, where land is used to grow food crops (DEFRA, 2002; 388 C.J. French eta/. I Environmental Pollution 141 (2006) 387-395 Fig. I. A typical brownfield site at Cromdale Grove (CRM) with amenity grassland on shallow soil (I 0-30 em depth) overlying a former landfill site that received domestic, industrial and chemical waste (including colliery spoil , ash and rubble) in the 1960s and 1970s, surrounded by residential buildings and light industry and remains of previous industrial activity. The photograph was taken as the two experimental plots in the present study were being established on the landfill site (plots and site are marked on the photograph). DEFRA and EA, 2002). This may be somewhat contradictory because environmental improvements brought about by landscape restoration and urban greening frequently proceed alongside residential redevelopment, leading to increased recreational usage of the land and thus potentially increased human exposure. The actual risk posed from this type of recreational land use cannot be easily assessed under the present system, which may lead practitioners towards using inaccurate or inappropriate guidance. There is also a possibility of harm being caused to humans or other animals, or of pollution of controlled waters, through planting trees on contaminated land; establishment of woodland may alter the lability of soil contaminants, for example due to biological activity associated with leaf fall or the rhizosphere, although there is currently little evidence that this is the case (Dickinson and Lepp, 1997; Dickinson et a!., 2000). The objectives of the current research project are to understand what happens to residual contamination of brownfield land and whether tree planting provides an effective longterm management of risk from soil contaminants. Results presented here are one part of 7 years of experimental fieldwork at more than 50 trial plots at 18 sites in NW England. In this paper, we focus on contamination at five brownfield sites and the potential for (i) site clean-up by tree harvest and (ii) longterm stabilisation by establishing a tree cover. The aim of this experimental work was to quantify dispersion and mobility of metals and As over a 3-year growth cycle of mixed woody biomass planting and harvest. 2. Materials and methods 2.1. Sites Five brownfield sites in the region of Liverpool and St. Helens in NW England were selected on the basis that they were perceived to be contaminated by the Loca l Authority or other landowners, following previous desk studies and Phase I surveys (Table I). The 5 sites had differing former land uses (landfill, industrial waste , sewage-sludge) and thus a likely existence of a varied range of trace element contamination. Nine trial plots (each 30m X 30m or 21m X 21 m, with 1-3 plots per site) were located across the sites and rabbit-proof fencing was installed around the plots, except at MER where 3-wire public-exclusion fencing was used (in the absence of rabbits locally). Soils at 2 sites (SUG and CRM) were crawler ripped to 2 m, in an attempt to resolve perceived drainage problems, and total cultivation was carried out at all sites by turning the soil (using a tracked excavator) to an approximate depth of 30 em , followed by (tractormounted) rotovation and a glyphosate weed-control treatment. Further, weed control using a combination of strimming and herbicides was required during the first 2 years of cultivation. 2.2. Plot characterisation Dispersion patterns of trace elements were measured and mapped, with soil sampling (0-30 em depth , 2.5 em auger) designed to identify 1% hotspots (72 sampling points per 900 nl plot) using a herringbone sampling design (DoE, 1994). Soils were later re-sampled , following harvest of plants after 3 years, from points within the plots identified as hotspots. After sampJjng, soils were air-dried , ground and sieved to < 2 mm. Trace elements were determined as pseudo-total concentrations following HNOiHCl microwave digestion (US-EPA, 1994). Additionally, a sequential extraction procedure was used: 0.0 I M CaC1 2 (3 g soil in 30 ml , 2 h) followed by centrifuging, filtering and C.J. French et a/. I Environmental Pollution I 4 I (2006) 387-395 389 Table I Locations of experimental trials and outline site descriptions Site Abbreviation NGR Details Sugar Brook SUG SJ 392960 Fazakerley FAZ SJ 387963 Kirby Moss KIR SJ 445984 Merton Bank MER SJ 527960 Cromdale Grove CRM SJ 534947 A 3.8-ha urban derelict site adjacent to busy road in a commercial and residential location. Formerly allotments, but neglected grassland and scrub prior to onset of study. Formerly a sewage farm but now adjacent to a modern sewage treatment works. Substantial recent re-engineering of landscape at ca. 15 ha of the site. Sewage-sludge and sewage cake treatments applied to one experimental plot prior to planting. 34 ha landfill site within an agricultural landscape. Livestock grazing on poor grassland ( 15 em topsoil depth) since closure of the landfill site in the 1970s. 6.6 ha of intensively mown amenity grassland, used for public recreation, within a residential and industrial area. Shallow soils (10-30 em). An alkali works from 1873 within an industrial landscape, and subsequently an industrial waste site. 8.8 ha of mown ameruty grassland with shallow soil overlying former landfill site in a residential and industrial area (see Fig. I). Three of the sites were old-style landfills (Dickinson, 2002) with inadequate capping and restoration by modern standards. 0 .05 M EDTA (30 ml , I h). Soil trace elements were then determined using AAS (Year I ) and ICP-OES (Year 3) following standard methodology. Other physico-chemical parameters were detennined from bulked samples for each plot, using standard methods (French, 2005) . Reference Material (NCS DC73349, bushes, branches and leaves) was included in all analytical batches. 3. Results 2.3. Planting and l/lanage111ent Plots were planted with 5 taxa of Salix, 2 Populus hybrids, Alm1s, Bet11ia and Larix (Table 2) in a fully randomised block design (I 0 blocks per taxa, 12 plants per block, double rows 1.5 m X 0.5 m spac ing). Selection of species and varieties was based on UK Forestry Commission guidance (Tabbush and Parfitt, 1999; Tubby and Armstrong, 2002) and earlier work within o ur own project (Punshon and Dickinson , 1999; Rawlinson et al. , 2004). Following the standard practice for management of short-rotation coppice management, Salix and Populus, but not the other species, were manually cut back after I year. Above-ground biomass of all Salix, Populus and Alnus was then manually harvested after a further 2 years. Standardised foliar sampling (top third of the inner 8 trees per block) for metal determination was carried out after the first growing season and at the end of the third year; at the latter time, sampling focussed on trees planted within contamination hotspots and stem samples were also collected at harvest. Plant samples (foliage, and I 0 em section s of stem removed from every 50 em of total stem height) were washed in deionised water, dried at 80 oc and ground prior to trace element determination, using HNO/H 2 0 2 microwave digestion and ICP-OES in all cases. Certified Table 2 Woody species and varieties planted across all plots in a randomised block design Taxa Abbreviation used in text Larix X euro/epis Henry Betula pendula Roth Alnus incana (L.) Moench Populus deltoides X nigra 'Ghoy ' Populus trichocmpa 'Trichobel ' Salix caprea X cinerea X viminalis LAR BET ALN GHY TRI 'Calodendron' Salix vimina/is 'Orm' Salix caprea X vimina/is ' Coles ' Salix bwjatica 'Germany' Salix \'imina/is X schwerinni 'Tora ' CAL ORM CLS GER TOR Salix and Populus were planted as unrooted 25 em pegs; Larix, Betula and Alnus as 20-30 em cell grown stock. Each repli cated block contained 12 plants (2 rows of 6 plants) with 0 .5 m spacing within rows, 1.5 m between rows and 4-10 replicates of each taxa per plot. Blocks were planted contiguously, with no additional spaci ng between blocks. Soils at all of the sites were highly disturbed historically or else derived from various imported substrates through former land uses. They were largely defined as clay loams or silt loams with a range of 5.6-13% organic matter (LOI), 0.14-0.3% total N, pH 6.8-7.4 (except SUG, with soil pH 5.4), and are described more fully elsewhere (French, 2005). There was little evidence of contamination at one of the sites (SUG), apart from 7% of soil samples with marginally elevated Zn (Mean 181 mgkg- t, Range 128-373 mgkg- t) . The other 4 sites showed evidence of contamination with at least one trace element (Table 3). The MER site was highly As-contaminated, with some additional concern over Pb concentrations. Comparison of these data with UK Soil Guideline Values (SGV) suggests that contamination may be somewhat marginal at the other sites, particularly for the more zootoxic elements (Cd and Pb). However, at all sites, there was a wide range of variation for all elements. CLEA guidance (DEFRA, 2002) utilises statistical tools or 'mean value tests' when examining heterogeneously contaminated land. If the calculated upper 95th percentile (US 95 ) is less than the SGV, then contamination may be judged not to be an issue for purposes of human health risk assessment. Using this criterion, all three plots at MER were significantly As-contaminated and one plot was significantly contaminated with Pb. Of the plots at the other 4 sites, two would be judged to be contaminated with Cd, two with Ni and another with Pb. When these contaminated plots are viewed in spatial profile, all but one possessed a highly heterogenous dispersion of trace elements (Fig. 2). Uniform contamination of the land surface appears to be unusual , and the existence of hotspots is typical. Analysis of soil samples collected from identified hotspots showed very low recovery rates using CaC1 2 as an extractant (Table 4) . Substantial proportions of Cd, Zn, Cu and Pb were apparently highly mobile and thus potentially bioavailable in the soils. EDTA extraction of Ni was proportionately low, indicating low mobility of Ni in the hotspots. Obvious care C..!. French eta/. I Environmental Pollution 14 1 (2006) 387-395 390 Table 3 Ranges of soil variables across the 10 plots and 5 sites in the study, and in the context of UK gu idance on Contaminated Land Exposure Assessment (DEFRA and EA , 2002) Variable Cd Zn Cu Ni Pb Asc pH LOI (%) Total N (%) Range in soi l samples ·across all plots (mg kg0-7.9 2.8-1 300 10-880 10-109 45- 1770 4.9-5266 4.9-8.2 1 ) Range of plot means (mgkg 0.3-6.1 121-329 5.9-307 24-50 99-720 312-615 6.1-7.6 5.6-13.1 0.14-0.30 1 ) Soil guidance value (SGV)" (mg kg - 1) Range of the percentage of samples exceeding SGV across all plots Sites where US 95 exceeded SGV 1-8 (300)h ( 130)b 50-75 450 20 0-100 6-72 0-100 1-49 0-59 99-100 FAZ, KJR n/a nla CRM, FAZ FAZ, MER MER n/a - SGYs not published for Zn and Cu, so not applicable. " Soil guideline values from current UK CLEA Guidance (DEFRA and EA, 2002). b Elements not included in current UK CLEA Guidance; former UK Guidance (ICRCL, 1987) thresholds are used here. c Data refer to three plots at one si te (MER) only. FAZ 1 Cd KIR 1 Cd FAZ 1 Ni FAZ 1 Pb CRM 1 Ni Fig. 2. Spatial heterogenei ty of metals (Cd, Ni, Pb) and As (0- 30 em depth) within samp ling plots that were sign ifi cantly contaminated at four of the study si tes (CRM , FAZ, KIR , MER). Sampling points are indi cated on the maps (Note: MER2 and FAZ I are smaller plots of 2 1 m X 2 1 m, with fewer sampling points). Colour differential is arbitrari ly defined by SURFER mapping software, with overall concen trati ons ranges of 0-8 mg Cd kg - 1, 0-180 mg Ni kg - 1, 0-2000 mg Pb kg - I and 0-5000 mg As kg - I. C.J. French eta/. I Environmental Pollution 141 (2006) 387-395 Table 4 Pseudo-total, EDTA-extractable and CaC]z-extractable concentrations of trace elements in soil samples collected from identified hotspots within plots Trace element (site) Cd (KIR) Cd (FAZ) Zn (CRM)" Cu (FAZ)b Ni (CRM) Ni (FAZ) Pb (FAZ) Pb (MER) As (MER)" Mean pseudo-total (mg kg- 1) 2.7 6.4 496 323 76.1 54.1 684 2088 1386 EDTA-extractable Mean (mg kg0.4 2.3 80.4 151.1 2.2 2.9 303.7 645.6 35.4 1 ) CaClrextractable Percentage of pseudototal Mean (mgkg- 14.7 35.1 16.2 46.8 2.9 5.4 44 .4 30.9 2.6 0.1 0.2 4.8 1.1 1.0 0.4 0.1 0.1 3.1 1 ) Percentage of pseudototal 1.9 3.5 1.0 0.3 1.3 0.8 < 0.1 < 0 .1 0.23 Proportions represented by the more readily extractable fractions are shown in italics. 1 a Thirty-eight percentage of samples exceeded 300 mg Zn kg (max . 827 mgZnkg- 1) . b All samples exceeded 130 mg Cu kg- 1 (max. 399 mg Cu kg- 1). c Mean of all three plots at MER. must be taken in relating EDTA-extractable concentrations to assumed bioavailability, particularly for this range of trace elements (Nolan et a!., 2003; Meers et a!., 2005). EDTA acts by complexing cations and, as such, is an inappropriate extractant for As which exists in soils in anionic fmm as arsenate or arsenite. However, more detailed analyses reported earlier for soil from the same site (MER) showed As mobility from short- and long-term leaching tests to be 0.55% and 20% of the total, respectively (Hartley et a!., 2004). Overall mortality of plants, apart from Larix (LAR), was low (mean 11%) and was not related to contamination, except at the site with highly elevated As (MER). LAR mortality was very high (20-77%) at all sites. Mortality of Betula increased from 7% to 29% between Years 1 and 3, visibly due to competition from other faster-growing taxa. Plants yields calculated using data from the Year 3 harvest (Fig. 3) showed considerable variation between the Salix and Populus taxa, and comparably high yields of Alnus. There was also wide variation between sites ( <2-9 tha- 1 annum- 1) and plots (e.g. at MER). Foliar and stem content of Pb (2-5 mg kg- 1 and 02 mg kg- 1 , respectively) was low at all sites, as was Ni in both plant parts (1.5- 3.0 mg kg - 1). Uptake of As by plants at the As-contaminated site (MER) was practically insignificant ( <2 mg kg- 1), with the exception of Larix where up to 14 mg As kg- 1 was recorded in foliage. There was less variability in plant Cu concentrations than of other trace elements, but four of the Salix clones and Larix appeared to have elevated concentrations in stems (Fig. 4). In contrast, Cd and Zn uptake was more vruiable into foliage and stems with notably higher concentrations in some of the Salix clones (CAL, CLS, GER, TOR). For Cu, there was no correlation between plant and soil concentration, but there was a very cleru· relationship in the cases of Cd and Zn. For Cd in particular, uptake into stems proportionately exceeded the amount in soil (Fig. 5). Paired t-test compru·isons of EDTA-extractable trace element concentrations in soil at the onset of the experimental 391 12 )': 'ro 10 .r: 1- 8 8 (j) 2 'iii 6 ~ (/) ~ E r- 4 0 m ffi 2 Q) :2 0 ALN CAL CLS GER GHY ORM TOR TRI Species 12 )': 'ro 10 .r: 10 0 8 (j) Q) ' (3 Q) c. 6 (/) ~ ~ ro E 4 0 m c 2 ro :2 1-n Q) 0 CRM 1 FAZ 1 FAZ 2 KIR 1 MER 1 MER 2 MER 3 SUG 2 SUG 1 Site Fig. 3. Mean biomass in ODT (oven-dry tonnes) ha- 1 yr- 1 of the 8 most productive taxa and across the 9 experimental plots (Abbreviated codes as in Tables 1 and 2). Error bars represent 95 % confidence interval. work and after 3 yeru·s, showed a statistically significant decline in soil hotspots of Cd at KIR (p < 0.05), Zn at CRM (p < 0.05), Cu at CRM (p < 0.05) and As at MER (p < 0.05), although differences in mean concentrations were hru·dly different (French, 2005). There was no evidence of an increase of metal concentrations in soil hotspots at any of the plots following planting with woody biomass. 4. Discussion 4.1. Soil contamination The brownfield sites selected for the present study are typical of urban land in NW England and similar ru·eas with diffuse historic sources of contamination. Detailed knowledge is usually lacking of varied former land uses, waste disposal, spillages, aerial fallout and land disturbance. This has provided brownfield sites that required detailed sampling and field investigation. In some cases perceptions of contamination C.J. French eta/. I Environmental Pollution 141 (2006) 387- 395 392 100 - FoliarYr2 ~ FoliarYr3 80 - stemYr3 '7 Ol .:.:. Ol .s '7 ~ 60 Ol E Q) .:.:. .l9 0. ::::> ::::> 20 "0 0 E 40 $ 0 (f) 10 20 0 25 5 10 Total Soil Cd mg kg- 1 15 5 10 EDTA Soil Cd mg kg- 1 15 20 '7 Ol .:.:. Ol .s 15 Q) .:.:. ro c.::::> '7 Ol 10 .:.:. 20 Ol "0 E 0 "0 5 ~I ni I 0 0 E Q) Ci5 10 1400 1200 '7 1000 .:.:. Ol Ol .s Fig. 5. Relationship between cadmium concentrations in stems of Salix taxa and total soil Cd (upper graph) and EDTA-extractable Cd (lower graph) for TOR (0), ORM (L'>), GER (X), CLS (.).CAL (e) (listing in Table 2). 800 Q) .:.:. .l9 0. 600 ::::> c N 400 . 200 0 m.-Ih [ ~ l T ~ I CAL LAR ALN CLS ORM TRI GHY BET GER TOR Fig. 4. Cu, Cd and Zn concentrations in foliage (Years 2 and 3) and stems (Year 3) within each plant taxa. Values are means for all sites combined. Abbreviations for plant taxa as in Table I. Error bars are 95% Cis. are probably incorrect, particularly if modem risk assessment criteria are properly applied. However, it is clear that highly vaJiable spatial dispersion of metals is revealed by detailed sampling; hotspots are likely to be masked when small numbers of samples are taken and when samples are bulked prior to analysis. Best practice is appropriately discussed elsewhere (Rams~y and Argyraki, 1997; Sastre et al., 2001; CL:AIRE, 2004). It would appear that contamination of many brownfield sites is likely to be restricted to hotspots and this may have considerable relevance to appropriate forms of management, planning of site remediation, treatment and after-use. Low recovery concentrations of metals using CaC1 2 as an extractant were recorded, comparable to the findings of previous studies (Pulford et al., 2002; Pulford and Watson, 2003). Although this may become a standard extractant for contaminated soils in Europe (Pueyo et al., 2003), metals were less easily detectable in soils with lower levels of contamination as others have reported previously (Houba et al., 1996; Laureysens et al., 2004). EDTA was a more useful general extractant for the range of trace elements in the present study, although in some cases (notably Pb) it provided a poor guide to the amount of uptake by plants. Nevertheless, in many cases, EDTA may provide cheaper, easier and equally reliable descriptor of site contamination than total or pseudo-total analysis of trace elements and is probably more relevant to understanding bioavailability and risk, in the absence of more appropriate routine methodology to describe speciation (Nolan et al., 2003). 4.2. Productivity of coppice In the UK it has been considered that economic retum is achieved when woody biomass yields exceed C.J. Fren ch eta/. I Environmental Pollution 141 (2006) 387-395 393 Surface 30cm Surface 1Ocm 6 Ol ..1<: Ol 5 .sc 4 ts 3 0 ::J '0 Q) 0:: 2 '0 0 ·o (f) 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 10 20 30 40 50 60 70 80 90 100 Years 0 10 20 30 40 50 60 70 80 90 100 Years 100 ";' Ol ..1<: Ol 80 0 60 '0 40 .sc t5::J Q) 0:: c N 20 ·o (f) Fig. 6. Number of years to lower concentrations of Cd and Zn in soil at a brownfield site through harvest of Salix (5 taxa; 0 S. X ca/odendron) and Populus (2 hybrids, • <>) stems and foliage [ e, all taxa combined] . Mean total soil concentrations were 6 mg Cd kg - l and 250 mg Zn kg - l Data are for the whole plot without targeting hotspots. Constant off-take throughout the period is assumed (see text). 8-10tha- 1 annum- 1 (Mitchell et al., 1999; Tubby and Armstrong, 2002). In the present study, three sites (CRM, FAZ, KIR), four Salix taxa (CLS, GER, ORM, TOR), and Alnus were either close to, or exceeded, this threshold. Clearly, there is an oppmtunity to increase yield beyond those achieved in the present study, or at least to maintain yields in the longerterm through fertilization and irrigation (Moffat et al., 2001). It has also previously been shown that SRC yields may be substantially higher during subsequent 3-year cutting cycles (Tubby and Armstrong, 2002). In the FAZ plot, a sewage cake treatment (600 tha- 1) increased yield by up to 56% in Salix (GER), 81% in Populus (TRI), 120% in Betula and 6% in Alnus (French, 2005). The reason for poor establishment of Larix across all sites and plots is uncettain but may have been caused by poor quality planting stock or inappropriate cultivation, rather than necessarily being unsuitable for planting on brownfield land. Betula was visibly slower growing than the other broadleaves and was a poor competitor in the dense mixed coppice. Whilst there was considerable yield variation between the Salix and Populus clones, there is likely to be an advantage in planting polyclonal stands rather than monocultures to reduce disease incidence, particularly from Melampsora rusts (McCracken et al., 2001). The wide variation between plots and sites (Fig. 3) was due to local site conditions rather than climatic vruiables. At MER, for example, soils were known to be pruticularly shallow and poor, but aspect and slope of the 3 plots also varied considerably. 4.3. Uptake of metals Lead uptake by plants was low in the present study, as is typical for this relatively immobile metal. Despite a relatively large proportion (20-45%) of soil Pb being extractable in EDTA, very little was available to plants. In soils with a pH above 5.5, as at the present study sites, Pb solubility is controlled by phosphate and carbonate precipitates (Blaylock et al., 1997). Arsenic in the MER soil is likely to be primarily in relatively immobile fmms, particularly arsenate (AsV), and thus persistent. A recent leaching and amendment study on the same soil found that 0.6-2% of the existing As could be leached from MER soils and that mobility was strongly influenced by iron oxides (Hartley et al., 2004). Whilst these authors demonstrated the feasibility of immobilising As using vru·ious Fe amendments, other soil factors such as organic matter, waterlogging and raised pH actually may increase mobility (O'Neill, 1990). Nickel was taken up by plants in small amounts in the present study, although this metal has been found to be taken up by trees in lru·ger quantities from soil where concentrations ru·e much higher (Pulford et al., 2002). In the present study, Cd and Zn were found to be pruticulru·ly mobile in the soil-plant continuum. Both are widespread contaminants that commonly occur in urban environments, but Cd is a much more zootoxic metal and of more concern to human health and to food chains (Adriano, 2001; Dickinson and Pulford, 2004). Salix accumulated Cd in concentrations that substantially exceeded soil concentrations. The taxa CAL 394 C.l. French eta/. I Environmental Pollution 141 (2006) 387- 395 and GER contained 7-9 times more Cd in stems than EDTAextractable soil concentrations, and 9-13 times more in foliage. These data can be combined with biomass yield data in predictive models of longer-term metal off-take (Fig. 6). These are speculative models that assume continued metal lability, consistent soil physico-chemical properties, the same productivity and the same rates of uptake. Obviously, they should be treated with caution. However, over a typical 20year life cycle of the crop this would amount to a reduction of 5.6 mg Cd kg- 1 and 96 mg Zn kg- 1 from the soil by the most efficient taxon (Soli X calodendron). scales for clean-up using phytoextraction are long, but there is now gathering evidence that this can provide a significant contribution to the integrated management of brownfield land. Acknowledgements We are grateful to our collaborative partners, particularly Paul Nolan and the Mersey Forest, Local Authorities and other landowners, and to NWDA for financial support. Professor Nicholas Lepp provided helpful comments on earlier drafts of this manuscript. 5. Conclusions Despite considerable advances in contaminated land exposure assessment in recent years, there are still obvious difficulties in identifying whether or not and under what conditions soil pollution should be a matter of concern. Elevated concentrations of heavy metals do not necessarily equate to an episode or occurrence of contamination, but frequently there is sufficient assessment of risk to justify some degree of cleanup or a reduction of soil concentrations. Disposal of this type of soil to landfill is obviously unnecessary and wasteful, and is not tenable under current regulations in Europe; in situ remediation and risk management are the only realistic options (Fox et al., 1998; Wong and Bradshaw, 2002). The present study has shown that, in the first instance, a satisfactory quantification and mapping of trace element dispersion is required. The occurrence of hotspots may be masked by less detailed sampling and analysis. EDTA extraction of trace elements from soils is as informative as determination of total concentration and is likely to be more cost-effective, but soil guideline values and regulatory limits are still based on the latter. Elevated concentrations of Cd, in particular, frequently cause concern on brownfield land. Phytoextraction using shm1-rotation coppice may provide an efficient and cost-effective method of clean-up. Whilst the same applied to Zn, this metal is less zootoxic and seldom significantly limits plant growth on urban contaminated soils (Dickinson et al., 2000); woody biomass may, however, have relevance to clean-up of sewage-sludge contaminated agricultural soils (Rosselli et al., 2003; Berndes et al., 2004). Potential risks associated with phytoextraction of metals, such as biomass combustion and food chain transfer of metals, have been extensively discussed elsewhere (Dickinson et al., 2000; Pulford and Watson, 2003). Other elements may have little risk associated with human exposure (e.g. Ni and Cu) or are insufficiently mobile in soil-plant systems (e.g. Pb, As). In these situations, the limited evidence from the present study suggests that planting woody biomass does not increase the lability of metals or their mobility to the wider environment, at least during the first 3 years. Ground cover with trees is likely to reduce the reentrainment of particulates and contamination of the wider environment. Thus, woody biomass may provide an effective form of phytostabilisation or monitored natural attenuation. Cultivation of woody plants for biomass provides aesthetic improvement and economic benefits (Paulson et al., 2003). Time References Adriano, D.C., 200 l. Trace Elements in Terrestrial Environments, second ed. Springer-Verlag, 867 pp. Anon. , 2003. Community Forest Programme Evaluation Overview. The Countryside Agency, 7 pp. Bardos, P. , French, C.J., Moffat, A.J. , Nortcliff, S., 1999. Renewable energy, decontamination and recycling: can biomass kill three birds with one stone? Land Contam. Recl am. 7, 174-176. Berndes, G. , Fredrikson, F. , Borjesson, P., 2004. Cadmium accumulation and Salix-based phytoextraction on arable land in Sweden. Agric. Ecosyst. Environ. 103, 207-223. Blaylock, M.J., Salt, D.E., Dushenkov, D.E., Zakharova, S., Gussman, 0 ., Kapulnik, Y., Ensley, B.D., Rasklin, I., 1997. Enhanced accumulation of Pb in Indian Mustard by soil applied chelating agents. Environ. Sci. Techno!. 31, 860-865. Cannell, M.G.R., 1999. Growing trees to sequester carbon in the UK: answers to some common questions. Forestry 72, 237-247. CL:ATRE, 2004. Technical Bulletin TB7 . Improving the Reliability of Contaminated Land Assessment Using Statistical Methods: Part I - Basic Principles and Concepts. CL:AIRE, London . DEFRA, 2002. Contaminated Land Exposure Assessment Model (CLEA): Technical Basis and Algorithms. Report CLRIO. WRc pic, 129 pp. DEFRA and EA, 2002. Soil Guideline Values Reports for Individual Soil Contaminants: Reports CLRIO, SGVI-10. WRc pic, Each report 14-20 pp. Dickinson, N.M ., 2000a. Strategies for sustainable woodlands on contaminated soils. Chemosphere 41 , 259-263. Dickinson, N.M., 2000b. Trees as environmental sentinels. Biologist47, 211-215 . Dickinson, N.M., 2002. The Treatment of Modern Landfills. In: Wong, M.H., Bradshaw, A.D. (Eds.), The Restoration and Management of Derelict Land: Modern Approaches. World Scientific Publishing, New Jersey, pp. 167-180. Dickinson, N.M. , Lepp, N.W., 1997. Metal s and Trees: Impacts, Responses to Exposure and Exploitation of Resistance Traits. In: Prost, R. (Ed.) , Contaminated Soils. INRA, Paris, pp. 247-254. Dickinson, N.M., MacKay, J.M., Goodman, A. , Putwain, P., 2000. Planting trees on contaminated soils: issues and guidelines. Land Contam. Reclam. 8, 87-101. Dickinson, N.M., Pulford, I.D., 2004. Cadmium phytoextraction usin g shortrotation coppice Salix: the evidence trail. Environ. Int. 31 , 609-613. DoE, 1994. Contaminated Land Research Report. Sampling Strategies for Contaminated Land. CLR Report No. 4, Department of the Environment (Prepared by The Centre for Research into the Built Environment, Nottingham Trent University). Environment Agency, 2002. Dealing with contaminated land in England. Environment Agency, 41 pp. Fox, H.R., Moore, H.M., Mcintosh, A.D. (Eds.), 1998. Land Reclam ation: Achieving Sustainable Benefits. A.A. Balkema, Rotterdam. French, C.J., 2005. Tree planting for phytoremediation: the fate of soils contaminants at brownfield sites. Ph.D thesis. Liverpool John Moores University, Liverpool. Hartley, W., Edwards, R., Lepp, N.W., 2004. Arsenic and heavy metal mobility in iron oxide-amended contaminated soils as evaluated by short- and longterm leaching tests. Environ. Pollut. 131, 495-504. C.J. French eta/. I Environmental Pollution 141 (2006) 387-395 Houba, V.J.G. , Lexmond, T.M ., Novozamsky, 1., van der Lee, J.J., 1996. State of the art and future developments in soil analysis for bioavailability assessment. Sci. Total Environ . 178, 21-28 . ICRCL, 1987. Inter-Departmental Committee on the Redevelopment of Contaminated Land : Guidance Notes (59/83), seconded . HMSO, London. Laureysens, 1. , Bogaert, J ., Blust, R., Ceulemans, R. , 2004. Biomass production of 17 poplar c lones in a short-rotation coppice culture on a waste disposal s ite and its relation to soil characteristics. For. Ecol. Manag. 187, 295-309. Lepp , N.W., Dickinson , N.M. , 1998. Biological interactions: the role of woody plants in phytorestoration. In: Vangronsveld , J ., Cunningham, S.D. (Eels .), Metal-Contaminated Soils: In s itu Inactivation and Phytorestoration. R .G . Landes Company, Georgetown , Texas. McCracken , A.R. , Dawson. W.M. , Bowden , G., 2001. Yield responses of willow (Sa/i.r) grown in mixtures in short rotation coppice (SRC). Biomass Bioenergy 21 , 311-319. M eers, E., Ruttens, A ., Hopgood, M.J., Samson, D., Tack, F.M .G., 2005. Comparison of EDTA and EDDS as pot ential soil amendments for enhanced phytoextract ion of heavy metals. Chemosphere 58, I 0 11- 1022. Mitchell, C.P., Stevens, E.A. , Watters, M.P., 1999. Short-rotation forestry: operations. productivity and costs based on the experience in the UK. For. Ecol. Manag. 121 , 123-136. Moffat , A.J. , Armstrong, A.T. , Ockleston , J ., 2001. The optimization of sewage s ludge and efnuent di sposal on energy crops of short rotation hybrid poplar. Biomass Bioenergy 20, 161 - 169. Nolan, A.L. , Lombi, E., McLaughlin, M.J. , 2003. Metal bioaccumulation and toxi c ity in soils - why bother with s pec iation? Aust. J. Chem . 56, 77- 91. O ' Neill , P. , 1990. Arsenic . In: Alloway, B.J. (Ed.), Heavy Metals in Soils. Blackie, London, pp. 83-99. Paulson, M., Bardos, P., Harmsen , J ., Wilczek, J. , Barton , M ., Edwards, D. , 2003. The practical use of short rotation coppice in land restoration. Land Con tam . Reel am. II , 323-338. Pueyo, M ., Lopez-Sanc hez, J.F. , Rauret , G. , 2003 . A ssessment of CaCI 2 , NaN0 3 and NH 4 N0 3 extraction procedures for the study of Cd, Cu, Pb, 395 and Zn extractability in contaminated soils. Anal. Chim. Acta 504, 217-226. Pulford , I. D. , Riddeii-Black, D., Stewart, C., 2002. Heavy metal uptake by willow clones from sewage sludge-treated so il: the potential for phytoremediation. Int. J. Phytorem. 4, 59-72. Pulford, I. D. , Watson, C., 2003 . Phytoremediation of heavy metal-contaminated land by trees - a review. Environ. Int. 29, 529-540. Punshon, T. , Dickinson , N.M ., 1999. Heavy metal resistance and accumulation characteristics in willows. Int. J. Phytorem . I, 361-385. Putwain , P.O., Rawlinson, H.A ., French , C.J. , Dickinson , N.M ., Nolan, P. , 2003. The Mersey forest Brownfield research project. In: Moore, H.M ., Fox , H.R. , Elliott, S . (Eds.), Land Rec lamation: Extending the Boundaries. A.A. Balkema, L isse, The Netherlands, pp. 165-172. Ram sey, M.H ., Argyraki, A., 1997. Estimation of measureme nt uncertainty from field sampling: implication s for the classification of contam inated land . Sci. Total Environ. 198, 243-257. Rawlinson, H ., Dickinson, N., Nolan , P., Putwain, P. , 2004 . Woodland establishment on closed old-style landfill sites in N.W. England . For. Ecol. Manag . 202, 265-280. Rosselli, W. , Keller, C. , Boschi, K., 2003. Phytoextraction capacity of trees g rowing on a meta l-contaminated soil. Plant Soil 256, 265-272. Sastre, J. , Vidal, M. , Rauret, G ., Sauras, T., 2001. A soil sampling s trategy for mapping trace element concentrations in a test area. Sci . Total Environ. 264, 141 - 152. Tabbush, P., Parfitt, R ., 1999. Poplar a nd Willow Varieties for Short Rotation Coppice. Forestry Commission, 4 pp. Tubby, 1. , Armstrong, A. , 2002. Establishment and Management of Short Rotation Coppice . Forestry Commission, 12 pp. US-EPA, 1994. Microwave Assisted Acid Digest ion of Sediments , Sludges, Oi ls and Soils: Method 3051. United States Environmental Protection Agency, 14 pp . Van grons veld, J. , Cunningham , S .D. (Eels .), 1998. Metal-Contam inated Soils: Ln s itu Inact ivation and Phytorestoration. R.G . Landes Company, Georgetown , Texas. Wong, M .H., Bradshaw, A.D. (Eels.), 2002. The Restoration a nd Management of Derel ic t Land: Modern Approaches. World Scientific , New Jersey.