Environment International 27 (2001) 399 – 411 www.elsevier.com/locate/envint

A review of the contamination of soil with lead II. Spatial distribution and risk assessment of soil lead Julie Markus*, Alex B. McBratney Department of Agricultural Chemistry and Soil Science, The University of Sydney, Sydney, NSW, Australia Received 22 March 2001; accepted 30 March 2001

Abstract Contamination of soil with lead has occurred on a global scale. Exposure to lead may cause adverse effects to human health and the environment. It is therefore desirable to obtain a quantitative estimate of the potential risk of lead contamination. Numerous studies have been conducted collecting lead concentration data from both natural and contaminated soil on a range of scales. Very few of these studies have made serious attempts to spatially describe the data. In order to identify contaminated land and to enable development of appropriate environmental guidelines, it is essential to have an understanding of the universal range of lead concentrations. Such data also assists in assessing any potential risk to the environment or human health. This paper reviews the multitude of data collected on soil Pb concentrations. Lead surveys are discussed on the basis of land use, with Australian data presented separately. Data from lead surveys of agricultural, urban and industrial areas, as well as nationwide surveys are summarised. A small but increasing number of studies have employed spatial prediction techniques such as kriging to map the distribution of lead concentrations in soil. These studies are also summarised and a brief description of the basis for their use presented. Finally, environmental and health risk assessment is discussed and some methodologies in use around the world reviewed. D 2001 Elsevier Science Ltd. All rights reserved. Keywords: Lead contamination; Soil lead surveys; Spatial distribution; Spatial prediction; Kriging; Geostatistics; Risk assessment

1. Introduction The adverse health effects caused by low-level exposure to Pb have been extensively documented. Such health effects include neurological impairment and deficits in the functioning of the central nervous system (Needleman, 1983; Needleman et al., 1990). Comprehensive reviews on this subject may be found in Needleman and Landrigan (1981) and Needleman and Bellinger (1991). In order to determine the potential risk to human health of Pb contaminated soil, it is essential to gain an understanding of the universal range of Pb concentrations and how they are spatially distributed. There is a great deal of published data describing the range of Pb concentrations found in soil from both remote locations and those affected by human impact. Soil Pb

* Corresponding author. Present address: NSW Environmental Protection Authority, P.O. Box 29, Lidcombe NSW 1825, Australia.

surveys have been conducted over entire countries, agricultural regions and cities, and have also focussed on specific locations such as industrial sites, parks and along major roads. Sampling schemes employed in these surveys vary from systematic sampling on regular grids to less formal sampling designs where for example samples are collected according to site access. Such surveys have had the particularly useful outcome of generating estimates of the background Pb concentrations in various regions. These provide a reference for assessing the significance of individual values, especially with regard to delineating contaminated soil from ‘‘clean’’ soil (Archer and Hodgson, 1987). These surveys also enable a description of the spatial distribution of a contaminant that may assist in determining its origin and show any patterns in the locations of contaminated sites. Databases of trace metal concentrations enable informed decision making on setting soil protection guidelines and on suitability for particular land uses by assessing potential risk to humans or the environment.

0160-4120/01/$ – see front matter D 2001 Elsevier Science Ltd. All rights reserved. PII: S 0 1 6 0 - 4 1 2 0 ( 0 1 ) 0 0 0 4 9 - 6

400

J. Markus, A.B. McBratney / Environment International 27 (2001) 399–411

2. The spatial distribution of Pb in soil 2.1. Soil Pb surveys of agricultural, urban and industrial areas A summary of the Pb data from all surveys discussed below is given in Table 1. The data has been ordered

according to location and mean Pb concentration. Nationwide surveys of Pb concentrations in soil have been conducted in Northern Ireland (Dickson and Stevens, 1983), England and Wales (McGrath, 1986; McGrath and Loveland, 1992a,b), the USA (Holmgren et al., 1993), Poland (Kabata-Pendias and Dudka, 1991) and Norway (Steinnes et al., 1997). The surveys differ in the intensity of

Table 1 World and Australian soil Pb survey data Location

Number Area of samples (km2)

Range (mg/kg)

National surveys Ireland USA Poland England and Wales England and Wales Norway

1305 3045 132 1521 5692 512

11 700

0.5 – 138.19 < 1 – 135 4.5 – 286.5 4.5 – 2900 3 – 16 338

174

600

Urban and industrial surveys Somerset, UK Great Britain Garden soil Vegetable plot soil Public garden soil Birmingham, UK London, UK Residential Industrial Main road New York/Connecticut, USA Urban Intermediate Rural Aspen, USA Portland, USA High risk Low risk Bangkok, Thailand Pabianice, Poland Lodz City, Poland














Australian surveys Wollongong, NSW Port Macquarie, NSW Mort Bay, NSW Suburban Mt. Lofty Ranges, SA Pasture soil Adelaide, SA Port Pirie, SA Port Pirie, SA Queensland Rural Urban Perth, WA Suburban Tasmania Contaminated sites Orchard soil (Tasmania, SA)









a

Geometric mean.

4126 193 221 86 52 16 24 108

80 30

12.3 18.3a 39.8 – 40

8 – 10 000

183

13 – 14 100 24 – 2560 20 – 1820 56 – 1650

266a 270a 185a 405

8.18 11.0 – 36.8 40

52

656 713 978 2600

65 75 25 30 40 238

Mean Median Standard (Extractant) (mg/kg) (mg/kg) deviation

56

–

798

115 75.8 27.8 172

50 – 10 900 1275 50 – 700 205 12.1 – 269.3 47.8 3 – 902 23 6 – 650 55.7

0.05 M EDTA HNO3 HF/HClO4 HNO3/HClO4 Aqua regia HNO3

Dickson and Stevens, 1983 Holmgren et al., 1993 Kabata-Pendias and Dudka, 1991 Archer and Hodgson, 1987 McGrath and Loveland, 1992a Steinnes et al., 1997

HNO3

Davies and Ballinger, 1990

HNO3

Culbard et al., 1988

HNO3/HClO4/HCl

Davies et al., 1987 Leharne, 1992

8 M HNO3

Pouyat and McDonnell, 1991; Pouyat et al., 1995

AB-DTPA

Boon and Soltanpour, 1992

HNO3 and XRF

Krueger and Duguay, 1989 Wilcke et al., 1998 Czarnowska et al., 1992 Czarnowska and Walczak, 1988

340 270 424

< 181.4

9.2 – 808

24.6 214.5

7.5 –

(Reference)

155

28.9

52.7 15.1 79.8

HNO3/HClO4 20% HCl 20% HCl

0.5 N Acetic acid Beavington, 1973 HF/HNO3/HCl/HClO4 Lottermoser, 1997 Olszowy et al., 1995

1 – 187 6 – 108

9.9 20

11

21

27

14 – 780

256.2

165

233.8

516 69 56 472

0.4 – 74 5 – 1450 4 – 2100

12 97 437

3 – 56

14

Merry and Tiller, 1991

 14

120

7 74 98

0.063

3–9 1 – 396 6 – 550

8.8 49

0.1 0.1 0.1 0.1

14

HNO3

4.7

4

2

91.7 170

50

91

M M M M

EDTA EDTA EDTA EDTA

Tiller, 1992; Tiller et al., 1987 Tiller et al., 1976 Cartwright et al., 1977 Olszowy et al., 1995

Olszowy et al., 1995 Olszowy et al., 1995 Aqua regia

Merry et al., 1983

J. Markus, A.B. McBratney / Environment International 27 (2001) 399–411

the sampling and the sampling design. They also differ in the chemical extractions and analytical techniques used to measure the Pb concentration, requiring care to be exercised when comparing the concentrations reported in each study. A feature common to these surveys was the absence of samples from urbanised areas, or those known to be affected by anthropogenic contamination. With the exception of Kabata-Pendias and Dudka (1991), all of the abovementioned studies made some attempt to spatially describe the Pb concentrations. In general, the objective of all of these studies was to determine the distribution of element concentrations in soil, and in particular, background and anomalous values. In addition, the aim of the geochemical survey of England and Wales was to investigate the relevance of the Pb concentrations to uptake by plants and intake by animals and humans (McGrath and Loveland, 1992b). Holmgren et al. (1993) were specifically interested in soil from major cropping areas of the USA, but like all of these surveys, their purpose was to establish a database of Pb concentrations in soil from across the entire country. In each survey, topsoil samples were collected, and, in most cases, representative samples were comprised of a composite of 10 –20 subsamples at each location. The geochemical survey of England and Wales (McGrath and Loveland, 1992a) and the survey of Northern Ireland (Dickson and Stevens, 1983) were both sampled on regular grids of 5 and 10 km, respectively, whereas the Polish and American surveys were sampled according to soil type. In addition, the sampling design for the USA survey also considered where major crops were grown, avoided sites near main roads or industry, and avoided soil that had received additions of sewage sludge. Dickson and Stevens (1983) found that almost 90% of the samples had EDTA extractable Pb concentrations less than 10 mg/kg. The larger concentrations were found in the south eastern part of the country where previous Pb industries existed. Similarly, the Pb concentration in Polish soil was found to reflect the land use with the rural north eastern part of Poland having smaller mean values of Pb, thought to represent the background value (13.1 mg/kg), than the heavily industrialised southwestern region (20.5 mg/kg) (Kabata-Pendias and Dudka, 1991). The overall mean is given in Table 1. Welsh soil had a larger mean Pb concentration than English soil, reflecting the relatively greater proliferation of metalliferous ore bearing rocks and associated mining sites in Wales (McGrath and Loveland, 1992a,b). Concentrations of Pb in Norwegian soil showed large geographical variations following a north – south gradient with the largest mean Pb concentrations found in the south (111.3 mg/kg) and the smallest in the north (8.5 mg/kg) (Steinnes et al., 1997). Steinnes et al. (1997) explained the large concentrations of Pb in the south of Norway as a result of long range atmospheric transport from other parts of Europe, as no domestic industry or local automotive emissions would have resulted in this geographical distribution of Pb concentrations.

401

Concentrations of Pb found in urban soil are influenced by anthropogenic activity and are therefore likely to be much greater than those found in soil from rural areas. Concern over the health effects of Pb have motivated an increasing number of urban soil Pb surveys. These studies aim to determine the range of Pb concentrations to which the local population is exposed and to assist in quantifying the Pb intakes of young children living in contaminated areas (Davies et al., 1987; Culbard et al., 1988; Krueger and Duguay, 1989; Boon and Soltanpour, 1992). The most extensive study of Pb concentrations in urban soil was a national survey with over 4000 samples collected from 53 locations in Great Britain (Culbard et al., 1988). Sample sites were selected according to geographical location, degree of urban/industrial development and population distribution. Within each location 100 households were sampled on a grid from exposed soil in gardens, parks and vegetable allotments. Large concentrations of soil Pb were found in industrial and mining areas and population size did not always influence Pb concentrations (Culbard et al., 1988). Other smaller urban soil Pb studies have been conducted in Birmingham, UK (Davies et al., 1987), Portland, USA (Krueger and Duguay, 1989), Baltimore, Minneapolis-St. Paul and New Orleans, USA (Mielke, 1991), Aspen, USA (Boon and Soltanpour, 1992) and Bangkok, Thailand (Wilcke et al., 1998). Pb data for these studies are given in Table 1. In contrast to the surveys of agricultural regions discussed previously, the sampling schemes for these urban surveys are variable and none of them are systematically sampled on a regular grid. Sampling designs were based on criteria including distance or direction from the city, proximity to main roads, age of buildings and age of children living in houses. Mielke (1991) describes a number of approaches for sampling urban soil to map Pb concentrations. He suggests that a useful protocol for sampling is to stratify the city according to census tracts, which was successful in the survey of Pb concentrations in New Orleans. With the exception of Mielke (1991) none of the above-mentioned urban surveys attempt to spatially describe the Pb concentrations. Wilcke et al. (1998) state that their objective was to assess the extent and severity of topsoil contamination in urban parts of Bangkok, yet, they only sampled soil at 30 sites, and these were each 2 km apart along four transects through the city. They also compare the Pb concentrations found to those from studies in London, Hamburg and Manila, although a larger number of samples may be required to make these comparisons. The results of Wilcke et al. were unusual in that the Pb concentrations they found were smaller than expected for a large city. Their explanation states that background values of Pb in Bangkok soil are naturally small and that the soil had recently been disturbed by human activity resulting in a relatively short period of time for contaminants to have accumulated. Davies et al. (1987) and Krueger and Duguay (1989) concluded that soil adjacent to older properties or painted structures or those within 500 m of a commercial garage had

402

J. Markus, A.B. McBratney / Environment International 27 (2001) 399–411

larger Pb concentrations than other sites. Pouyat and MacDonnell (1991) and Pouyat et al. (1995) characterised the distribution of Pb concentrations along an urban – rural transect and found Pb concentrations were larger at the urban end of the transect, probably resulting from deposition of Pb emitted in vehicle exhaust. Other examples of urban soil Pb surveys may be found in Purves and MacKenzie (1969), Klein (1972), and Spittler and Feder (1979). In addition to Pb surveys of urban areas, the spatial distribution of Pb concentrations around industrial point sources of contamination such as Pb smelters have also been investigated. Maskall and Thornton (1993) measured Pb concentrations in excess of 30 000 mg/kg at historic smelting sites in England and Wales which are currently either agricultural fields or heather moorland. They sampled each site on a regular grid and both surface soil and subsoil were analysed. Maps of the spatial distribution show decreasing haloes of Pb concentration with distance from the source and the largest Pb concentrations were often found at depth. Davies and Ballinger (1990) sampled soil from North Somerset, England, from a region that included a part of the former Mendip Pb/Zn mining district. They investigated the spatial distribution of Pb concentrations by constructing isoline plots for metal concentration. Pb concentrations and their distribution around the smelter at Avonmouth in England were measured by both Burkitt et al. (1972) and Little and Martin (1972). Their sampling designs were quite different with Burkitt et al. sampling along three transects extending in different directions from the smelter, while Little and Martin (1972) sampled according to a grid with sampling intensity decreasing with distance from the smelter. Little and Martin produced contour maps to illustrate the spatial distribution of Pb concentrations in the vicinity of the smelter. Both studies concluded that the distribution of Pb in topsoil was governed by distance from the smelter and the prevailing wind direction. Linzon et al. (1976) sampled soil at increasing distances from several industrial sites (including a Pb smelter) in up to eight radiating directions from the point source of contamination and report similar findings to Little and Martin (1972). Assessment of Pb contamination of soil adjacent to major roads has received much attention (Singer and Hanson, 1969; Lagerwerff and Specht, 1970; Motto et al., 1970; Albasel and Cottenie, 1985; Leharne, 1992; Hafen and Brinkmann, 1996). The majority of studies have concluded that the concentration of Pb decreases with distance from the road (Singer and Hanson, 1969; Lagerwerff and Specht, 1970; Motto et al., 1970), however, Hafen and Brinkmann (1996) found only a weak relationship between concentration and distance. Microclimatic features including air turbulence and human disturbance of the soil profile were suggested as reasons for the absence of a strong correlation (Hafen and Brinkmann, 1996). Hafen and Brinkmann collected 224 topsoil samples along 32 transects perpendicular to a major interstate highway in Florida, USA. While their results differ from most other investigations of roadside soil, the number of samples collected makes this the most detailed survey of Pb

concentrations surrounding a major roadway. Pb concentrations at distances greater than 15 – 20 m from the road were found to approach constant values and were appreciably smaller than those within a few metres of the road (Singer and Hanson, 1969; Albasel and Cottenie, 1985). Lagerwerff and Specht (1970) and Singer and Hanson (1969) also report a decrease in Pb concentrations with depth in the soil profile and an increase in Pb concentration with traffic density. 2.2. Soil Pb surveys in Australia Published data on Pb contamination of soil in Australia is limited, and most of the data are from South Australian soil. Tiller (1992) reviewed the state of knowledge of urban soil contamination in Australia and presents some unpublished data from contaminated sites and urban areas in each state. It is likely that many commercial surveys of Pb contamination at industrial sites have been conducted, but most of this data is not freely available and little has been published in scientific journals. Urban Pb surveys have been conducted in Wollongong, NSW (Beavington, 1973), Glebe, NSW (Markus and McBratney, 1996), Port Pirie, SA (Tiller et al., 1976; Cartwright et al., 1977) and Adelaide, SA (Tiller et al., 1987). Extractants including acetic acid, EDTA and DTPA were used to measure Pb concentrations in these studies making it difficult to compare concentrations reported in each city. Data from each of the surveys discussed here is summarised in Table 1. Given the limited sample numbers reported in previous studies of urban soil, Beavington (1973) attempted a systematic survey of topsoil in Wollongong. Samples were collected from 398 sites covering an area of 56 km2 including urban, agricultural and industrial land. Data was only obtained for 80 sites, however, because the acetic acid used was not strong enough to extract measurable concentrations of Pb in the other samples. Those samples with the largest Pb concentrations were found either in roadside or industrial soil. Unlike the other Australian surveys discussed here, Beavington did not map the spatial distribution of Pb concentrations in the study area. Tiller et al. (1976) restricted their sampling to soil from home gardens representing different residential areas of Port Pirie, in a survey aiming to delineate areas of land contaminated with material originating from the Port Pirie Pb smelter. Maps of the spatial distribution of Pb in surface soil were constructed by drawing isoconcentration lines, which were hand-drawn assessments based on the available data. Tiller et al. (1976) state that accurate prediction of Pb concentrations in gardens that were not sampled is not possible using maps of this kind. This is because isolated large concentrations, which may occur in areas dominated by many small concentrations are not accounted for. Port Pirie was also the location of a largescale survey of the distribution of metals in the area surrounding a Pb smelter (Cartwright et al., 1977). Samples were collected from 472 sites in a region covering 5000 km2. Maps showing the spatial distribution of Pb concentrations

J. Markus, A.B. McBratney / Environment International 27 (2001) 399–411

were constructed using concentration contours. Concentrations of Pb decreased exponentially with distance from the smelter and were controlled to a large extent by the meteorology and topography of the region. The most extensive study on Pb concentrations in Australian soil was conducted in South Australia to investigate the regional distribution of Pb in soil and the dispersal of urban pollutants (Tiller et al., 1987; Merry and Tiller, 1991). A 20-km-wide transect extending 70 km east of Adelaide through the Mt. Lofty ranges to the Murray plains was surveyed. Tiller et al. (1987) describe sampling on a ‘‘free’’ survey basis where each sample represented an area of approximately 2 km2 and was collected where access and topography allowed. Pb concentrations were larger at the western end of the transect where dispersal of Pb from Adelaide was evident. Merry and Tiller (1991) concluded that Pb contamination of the rural hinterland east of Adelaide from urban sources was present, but of no environmental consequence when compared to the natural Pb concentrations in this soil. Pb concentrations within the Adelaide urban area were highly variable and originated from three main sources, namely historic and modern metal processing industries and automobile exhaust (Tiller et al., 1987). More recently Olszowy et al. (1995) reported background concentrations of Pb in urban and rural soil from around Australia. Their data was based on 320 topsoil samples from Sydney, Melbourne, Adelaide and Brisbane, and 120 samples from rural areas near Brisbane. Samples were collected according to the age of the suburb and its traffic density. Pb concentrations were found to be largest in older suburbs with a higher traffic density (Olszowy et al., 1995). There are also a small number of investigations on Pb concentrations in soil adjacent to major roads around the country. Pb concentrations in soil adjacent to highways were measured by Wylie and Bell (1973) along four highways near Brisbane, Queensland; David and Williams (1975) adjacent to the Hume Highway near Marulan, NSW; Bottomley and Boujos (1975) near the major highway crossing Heirisson Island 3 km east of Perth, WA; and Clift et al. (1983) along the Mulgrave freeway linking Dandenong and Melbourne, Victoria. All data are based on topsoil analyses of samples collected at varying distances along transects perpendicular to the road. All of these studies found that Pb concentrations decreased sharply with distance from the road and traffic density. Most of the Pb was deposited within 25 m of the roadside (Wylie and Bell, 1973; David and Williams, 1975). Clift et al. (1983) monitored the Pb concentration from the opening of the Mulgrave freeway for 5 years. They report an increase in Pb concentration within 2 m from the road from 300 mg/kg shortly after the freeway opened to 1250 mg/kg 5 years later. Pb contamination has also occurred in orchard soil as a result of the use of Pb arsenate insecticides. Merry et al. (1983) investigated the extent of this contamination in a survey of EDTA-extractable Pb concentrations in 98 topsoil samples from Tasmanian and South Australian apple and

403

pear orchards. They found that the surface 10 cm of the profile contained Pb concentrations up to 30 times greater than background concentrations. Pb had not accumulated at depth in any of the soil types sampled, including acid sandy soil found at some Tasmanian orchard sites. They concluded that considerable accumulation of Pb has occurred in Australian orchard soil and that this Pb is not in a mobile form. Other data on soil Pb concentrations in Australia was reported by Lottermoser (1997). He conducted a heavy metal survey in soil in Port Macquarie, NSW focussing on natural heavy metal enrichment from underlying serpentinite, rather than anthropogenic contamination. Concentrations of Pb were found to be randomly distributed and negligible in the study area. 2.3. Use of spatial prediction techniques to map contaminant distribution in soil In order to identify patterns in the spatial distribution of a soil variable such as Pb concentration, it is essential to present soil survey data in the form of a map. Mapping contaminant distribution allows immediate appreciation of the change in the contaminant with space and enables identification of areas that may contain hazardous concentrations. Knowledge of the spatial distribution of a contaminant is essential for site assessment and any subsequent risk assessment. Very few of the surveys discussed in the previous sections mapped the Pb concentrations within the sampled area, although some of these studies did produce contour maps. In general, the large cost of chemical analyses and the time taken to obtain chemical data at enough locations to enable mapping are prohibitive. An alternative to extensive sampling and analysis is to interpolate the point data to estimate Pb concentrations at locations between sampling sites. This may be achieved with the use of geostatistical techniques such as kriging. Geostatistics is based on the theory of regionalized variables developed by Matheron (1971). A regionalized variable is one that varies continuously in space and has some structure in its variation (Webster and Oliver, 1990), for example topsoil Pb concentration. This means that the variable exhibits some spatial dependence where its value at locations close together are more similar than those further apart (Webster and Oliver, 1990). An important part of geostatistical analysis is the semivariogram, which relates the variance between two locations to their separation distance (McBratney and Webster, 1986). It is a model of the spatial dependence and enables the estimation of attribute values at unsampled locations using kriging (Goovaerts, 1997). Kriging is the generic name for a family of generalized least-squares regression algorithms (Goovaerts, 1997). It is a spatial prediction method, which, in its simplest form, is a weighted average of measured data points, estimated at either a point location or over a larger area known as a block (Webster and Oliver, 1990). The method is designed to minimise the estimation variance and has the desirable feature of provid-

404

J. Markus, A.B. McBratney / Environment International 27 (2001) 399–411

ing a measure of the uncertainty associated with predictions (Goovaerts, 1997). Predictions may be either attribute values such as Pb concentrations, or the probability that the attribute will exceed a given threshold. An example of the latter is the probability that the Pb concentration at a particular site is greater than local regulatory guidelines. This is illustrated in Fig. 1 which shows the probability of soil in Glebe exceeding the environmental investigation limit of 300 mg/kg, produced by indicator kriging (Markus and McBratney, 1996). There are many different types of kriging and it is beyond the scope of this review to describe them. Some examples of the

application of this technique to the spatial description of Pb concentrations in soil are summarised in Table 2. An increasing number of surveys have used kriging to describe the spatial distribution of Pb concentrations in soil (e.g., Atteia et al., 1994; Tao, 1995a,b; von Steiger et al., 1996; Bierkens, 1997). Many of these studies do not specify the variogram model, variogram parameters (i.e., range, sill and nugget) or type of kriging used and several do not include the basic Pb data. It should also be noted that the chemical methods employed to extract the Pb vary significantly between surveys and as such Pb concentrations cannot

Fig. 1. Map produced by indicator kriging showing the probability of Pb in Glebe topsoil exceeding the environmental investigation limit (from Markus and McBratney, 1996).

Table 2 ‘Geostatistical’ surveys of Pb concentrations in soil Number Area of samples (km2) 65 50 150 13

1060 75

Fougeres, France Turin, Italy

2.5 kmt 0.375 0.001225 New Mexico, USA 0.001225 0.001225 14.5 Swiss Jura

–

Poland

4334

Shenzhen, China

60

0.0016

57

14 kmt

165 434

Whole island 2.3

204

50

2449

116

125

100

Range (mg g 1) – 30 – 361 12 – 361 500 – 4200

1100 – 7200 400 – 2200 18.7 – 382

0.8 – 929 5 – 191

Mean (mg g

Median (Variance) (Extractant) ) (mg g 1)

21.4

–

1.94a

HF

NA

67.8 49.9 1976

50 40 –

58.3a 45.9a –

Aqua regia

Block Kriging

Portable XRF

Inverse squared distance algorithm

3467 1415 57

46.8

1742

2M HNO3

Ordinary Block Kriging

b

16.41 –

–

–

–

–

Loire estuary, France Sardinia, Italy Jamaica

48 – 453

211

–

6372.8

8 – 240

75

–

56a

41

–

Glebe, Australia

10 – 20 278

854

315

Weinfield, Switzerland Utrecht, Netherlands

6.7 – 110.4

Evin – Malmaison, Nord Pas – de – Calais, France

3 – 1700

–

(Method)

1

23.3

20.3

–

–

Conc HF/ 12N HCl/ 14N HNO3 Aqua regia EDXRF

3 660 956 Aqua regia

166.17

–

–

Conc HNO3/ HClO4 Conc HNO3/ HClO4/HF

2M HNO3 –

–

HF/HClO4/ HCl/HNO3

Inverse distance function Ordinary Block Kriging

Variogram model

Variogram was – flat/chaotic Spherical 498 332 – –

Double spherical – Spherical

Inverse distance weighting Log normal point kriging Indicator kriging Disjunctive Kriging

287

Nugget (mg g 1)

)

–

Zanini and Bonifacio, 1991

–

240 85 –

0.0228L

Atteia et al., 1994

0.0131L –

Piotrowska et al., 1994

390

Tao, 1995 a,b

– 0

3L

–

Spherical

9610

1539.7

–

–

–

Spherical

(Reference) 1

–

9.6  10

2605 – 12 000 major

Sill (mg g

0 0

7500 minor 8700 –

NA

Point Kriging

Range (m)

–

4399 –

Wopereis et al., 1988

Haneberg et al., 1993

Arrouays et al., 1996

Bonifacio et al. (1996) Engel et al., 1996

54.9

0.0001

0.28

Markus and McBratney, 1996

65.1 Pentaspherical 1291

0.0001 0.16L

0.194 0.84L

von Steiger et al., 1996

1

Bierkens, 1997

Stratified Residual Spherical Kriging (Simple block kriging on residuals) Kriging –

750 0.5 (500 and 1000)

–

–

–

J. Markus, A.B. McBratney / Environment International 27 (2001) 399–411

13 13 366

0.01

Location

Frangi and Richard, 1997

a

Standard deviation. Geometric mean. t Transect. L Log-transformed data. b

405

406

J. Markus, A.B. McBratney / Environment International 27 (2001) 399–411

be compared. Geostatistics approaches are more suitable for diffuse contamination, more sophisticated techniques are required for point-source pollution (Walter et al., 2000).

3. Environmental and health risk assessment for soil lead Environmental risk assessment may be defined as the ‘‘process of estimating the potential impact of a chemical or physical agent on a specified ecological system under a specific set of conditions’’ (Australian and New Zealand Environment and Conservation Council (ANZECC)/National Health and Medical Research Council (NH&MRC), 1992). Health risk assessment has been defined as the ‘‘characterisation of the potential adverse health effects of human exposures to environmental hazards’’ (National Academy of Sciences, 1983). With the exception of a small number of risk assessment guidelines, protocols for estimating the environmental or health risk of a particular contaminant are illdefined throughout the world. Methodologies in various stages of development and implementation exist for risk assessment of contaminated sites in the USA, the Netherlands, Germany, Switzerland and the UK. All have some form of soil quality guideline specifying one or more threshold concentrations, which if exceeded, indicate the possibility of adverse effects occurring. They also consider site-specific criteria. Details of the approach and methodology used in Switzerland may be found in Vollmer et al. (1997). Perhaps the most widely recognised methodology for assessing the risk of an environmental contaminant is that developed by the US Environment Protection Agency (USEPA) (USEPA, 1989). The USEPA have devised a sitespecific, quantitative methodology using a modelling approach to quantify human exposure, dose – response and/or plant response relationships, for example, to the substance of concern. This procedure, which aims to produce a quantitative estimate of risk, has received some criticism because many conservative assumptions are made, which may result in an overestimate of the probable level of risk (Langley, 1996). Ryan and Chaney (1997) demonstrate the use of this quantitative methodology when applied to assessing the risk of land application of sewage sludge. Restrictions are placed on the allowable pollutant concentrations in the sludge depending on the final use of the land to which the sludge is applied. Many factors are considered in the determination of this sludge pollutant loading limit. These include identification of exposure pathways to humans, plants, animals and soil biota, identification of dose – response relationships for the receptors and determination of the concentration of contaminant that would protect the most highly exposed individual from adverse effects (Ryan and Chaney, 1997). The Netherlands also have well-developed criteria and guidelines for soil protection and risk assessment. The Dutch policy is based on the idea of soil multifunctionality, which means that ‘‘the number of functions that a soil can have should not be reduced by human activities’’ (Vegter, 1997).

If the risk to humans or an ecosystem is greater than a predefined ‘‘maximum tolerable risk’’, then soil contamination at a particular site is unacceptable (Theelen et al., 1997). For human health risk assessment, the maximum tolerable risk is determined from the Tolerable Daily Intake (TDI) and amount of exposure to the contaminant of concern (Theelen et al., 1997). Environmental risk assessment is based on the soil concentration at which half of the species in an ecosystem are potentially at risk (Theelen et al., 1997). This concentration is known as the HC50 (i.e., the Hazardous Concentration for 50% of species) and is derived from an entire ecosystem. Dutch regulatory guidelines include two soil criteria, namely target values and intervention values. Target values represent ‘‘a level of pollution where the soil (and groundwater) is still considered to be of good quality’’, and intervention values ‘‘indicate a level of pollution where soil quality is severely affected and a cleanup is considered necessary’’ (Vegter, 1997). These values are derived from toxicological references (i.e., TDIs of the contaminant by humans) and ecotoxicological references (i.e., no observed effect concentrations for different species exposed to a given substance) (Vegter, 1997). Theelen et al. (1997) remark on the limitations of Dutch environmental risk assessment stating that the method only provides an indication of potential environmental problems and cannot be used to calculate tolerable levels of risk for different kinds of ecosystems. This is because the ecotoxicological intervention value is not specific for different types of soil, fauna, vegetation or climate for example, it is ‘‘derived from a theoretical complete ecosystem’’. The smaller of the health and environment intervention values is usually selected as the final intervention value to ensure that both humans and the environment are protected. If an intervention value is exceeded then soil cleanup is necessary, although risk assessment will determine whether this remediation is urgent depending on actual exposure to humans, mobility of pollutant and its ecological effects (Theelen et al., 1997; Vegter, 1997). Vegter (1997) states that assessment of soil contamination in the Netherlands began by only considering generic soil criteria whereas the current soil protection guideline has adopted a site specific approach, placing emphasis on assessing exposure to soil contaminants. The risk assessment protocol in Australia is closely related to the USEPA methodology although there are some important differences. A key distinction is the semiquantitative nature of risk assessment in Australia. The ANZECC/ NH&MRC (1992) guidelines state that ‘‘in most situations, a numerical estimate of risk is not feasible because of limitations in toxicological and exposure data’’. The preferred approach is a semiquantitative estimate of risk, assuming there are sufficient data available. Health and Environmental Investigation Levels (HILs and EILs) for a range of soil contaminants were defined in the ANZECC/ NH&MRC guidelines. In the case of Pb, these two investigation levels were the same, equating to 300 mg/kg. HILs are intended to protect humans living on or in close

J. Markus, A.B. McBratney / Environment International 27 (2001) 399–411

proximity to a contaminated site for a lifetime, and take into account the bioavailability of the contaminant, Acceptable Daily Intake (ADI) and background exposure from food for example. Similarly, EILs are intended to protect ecosystems from potential adverse effects of contamination. These investigation levels are comparable to the Dutch intervention values because if exceeded, both the Dutch and Australian values are intended to prompt a site specific risk assessment. The ANZECC/NH&MRC guidelines also introduce the concept of response levels that depend on proposed land use and environmental factors. Response levels are simply concentrations at which a response is required (this may be remediation or just informing occupants that there is a problem) to protect humans and the environment. Adverse effects would not be expected to occur unless concentrations were well in excess of these levels (Langley, 1996). In 1998, the NSW Environment Protection Authority (EPA) introduced new Health-based Investigation Levels (HILs), which differ according to the land use and degree of exposure to soil (NSW Environment Protection Authority, 1998). The new HILs consist of four different threshold concentrations where previously there was only a single investigation level for each contaminant. The original investigation level of 300 mg/kg is still applicable in residential areas with gardens and accessible soil. For residential areas with minimal access to soil, such as high-rise apartments, the HIL is 1200 mg/kg, and for commercial or industrial land, it is 1500 mg/kg. The remaining level of 600 mg/kg applies to parks, recreational open space and playing fields. Guidelines for Pb in soil were proposed by the Society for Environmental Geochemistry and Health, to offer a flexible and scientifically based approach for determining a ‘‘safe’’ soil Pb concentration, applicable to a particular situation (Wixson and Davies, 1994). The use of a singlevalue guideline was rejected, because such guidelines are

407

unable to account for different risk scenarios such as land use and the population likely to be exposed. Instead, the Pb guideline developed is a quantitative relationship between blood-Pb and soil-Pb concentrations (Wixson and Davies, 1994). The model allows for input of updated regulatory criteria as new data become available and enables different environmental situations to be considered. Such guidelines are inherently more useful for assessing risk to human health than a single value guideline. Langley (1996) summarises important issues to consider prior to conducting risk assessment. Amongst other things he states that it is essential that the specific purpose of the risk assessment is clear and the critical receptors identified. For example, it must be determined whether the risk assessment is for the entire population or just the most sensitive individuals and for what use the contaminated site is intended. There may also be more than one contaminant for which risk is being investigated and the possibility of interactions should be considered. The level of acceptable risk must also be determined if not explicitly specified in regulatory guidelines. This may be when adverse effects begin to occur to either the most sensitive receptor likely to be placed at risk, or the maximally exposed individual (Ryan and Chaney, 1997). A basic framework exists through which risk assessments for chemical contamination are typically conducted. This is depicted in Fig. 2. They commence with a detailed appraisal of the site under investigation including site history and collection of samples, followed by the relevant chemical analyses and data evaluation (Langley, 1996). Subsequently, an exposure assessment is conducted to identify the pathways of exposure to individuals at risk, and where possible to quantify the contribution of each of these pathways (ANZECC/NH&MRC, 1992). Exposure assessments invariably involve modelling to generate a risk outcome for particular exposure scenarios. In conjunction with the expo-

Fig. 2. Processes in contaminated site risk assessment.

408

J. Markus, A.B. McBratney / Environment International 27 (2001) 399–411

sure assessment, a toxicity assessment is done. This is required to determine the nature of adverse effects related to exposure and dose – response relationships for these effects (Langley, 1996). Extensive toxicological and epidemiological data is required for this process and in many cases it is not available, hence, the preference for semiquantitative rather than fully quantitative risk assessments in Australia. The final stage in a risk assessment is generally known as the risk characterisation and involves the consideration of all previous stages of the risk assessment taking into account all assumptions and uncertainties to produce a final estimate of the risk (Langley, 1996). The design of an appropriate sampling plan is an important part of the risk assessment process, especially considering the large cost involved with chemical analyses and site remediation. Some guidance is provided by regulatory authorities in Australia, e.g., Contaminated Sites — Sampling Design Guidelines (NSW Environment Protection Authority, 1995). Strategies for sampling contaminated sites vary according to the degree of information about site history and the type of contamination. An efficient way to collect data in an initial site assessment may be the use of judgemental sampling (Laslett and McBratney, 1993; Langley, 1996). A formal, objective sampling scheme, such as stratified random sampling or systematic grid sampling, is more appropriate for detailed site assessment (Laslett and McBratney, 1993). Ferguson (1992) suggests sampling on either a herringbone or triangular grid as the optimal strategy when information on the spatial distribution of contaminants is limited. It is somewhat surprising that geostatistical techniques for spatially interpolating data, such as kriging, have not been employed in standard risk assessment methodologies. A detailed spatial description of contamination may be obtained by kriging to predict the concentration at locations between sampling sites. This may assist in identification of ‘‘hot spots’’, for example. Kriging also has the ability to predict the probability that a contaminant will have a concentration greater than a particular threshold concentration. As an example, Ginevan and Splitstone (1997) use probability kriging as part of an evaluation of the soil-related cancer risk at a site contaminated with hazardous waste. There is some debate in the literature about whether model-based geostatistical sampling schemes are the most appropriate, some authors, e.g., Brus and de Gruijter (1997) advocate a design-based approach. Recently, the USEPA has moved towards adopting Monte Carlo analysis in risk assessment, particularly for exposure assessment (Langley, 1996). Monte Carlo analysis generates a probability distribution of potential risk by simulating a range of randomly selected scenarios to estimate each risk parameter (Langley, 1996). This is more useful than conventional risk assessment methodology, which generates only single estimates and is prone to overestimating actual risk (Langley, 1996). There is also interest in incorporating such methodology in Australian risk assessment guidelines. The ANZECC/NH&MRC

guidelines state that ‘‘risk assessment is based on probabilities rather than absolutes and this should be reflected in decision-making’’. The uncertainty in risk estimates may be significant. It is desirable to identify the source and magnitude of all uncertainties throughout the various stages of a risk assessment in order to assess their cumulative effect (Langley, 1996). Once uncertainties and assumptions are documented, the limitations of a risk assessment may be accounted for so it may be better used in decision making. Langley (1996) recognises the need to define appropriate models for risk assessment and site management which make allowance for some margin of error, without resulting in unnecessarily conservative assessment and management outcomes. He states that the most useful risk assessments ‘‘will provide a range of estimates for different individuals and population subgroups’’ (Langley, 1996). Within a risk assessment, various exposure scenarios may be investigated. For example, the risk to adults and children, or the risk to different species in an ecosystem, may be evaluated separately. Hawley (1985) presents a quantitative methodology for estimating the health risk from exposure to contaminated soil in a residential area. The suggested methodology includes an exposure assessment for young children, older children and adults. Human health risk assessment is likely to focus on the highly exposed, highly susceptible individual. The individual most at risk from Pb pollution is a young child who regularly plays in Pb rich soil, has a large degree of hand to mouth contact, possibly even ‘‘pica’’ for soil, and poor nutrition (Simms and Beckett, 1987; Chaney and Ryan, 1994). Exposure assessments require large amounts of data and it is preferable to use data sourced locally. Data on soil ingestion by children for example is limited in Australia, yet, this is required to undertake effective health risk assessment (Langley, 1996). In the case of Pb, there is some data available of relevance to health risk assessment, as the adverse effects of this contaminant have long been recognised. Blood Pb concentration data for various sectors of the US population are available, although the studies have generally focussed on young children (Moehr et al., 1993; Weitzman et al., 1993; Brody et al., 1994). Surveys measuring human exposure to Pb through blood Pb concentrations have also been carried out in the UK (e.g., Davies et al., 1990). Thomas et al. (1999) present data for the concentration of Pb in blood, gasoline and air, measured before and after the reduction of Pb in gasoline. The data, collated from 17 individual studies conducted in six continents, show a strong linear correlation between blood-Pb and gasoline-Pb concentrations (r = 0.94). As the concentration of Pb in gasoline approached zero, blood-Pb concentrations converged to 3 mg/dl (Thomas et al., 1999). This finding clearly demonstrates that the removal of Pb from gasoline can result in the reduction of population blood-Pb concentrations to acceptable values. The most comprehensive Australian data is for Port Pirie in South Australia, which is a small town surrounding the world’s largest Pb

J. Markus, A.B. McBratney / Environment International 27 (2001) 399–411

smelter, still in operation after 110 years (Langley, 1996). Continuing research on the community living in Port Pirie and its surrounding environment has enabled collection of epidemiological information on the relationship between Pb exposure and neurobehavioural effects and has also involved blood Pb monitoring of children (Langley, 1996). Other Australian blood Pb data has been collected from urban children in central and southern Sydney (Mira et al., 1996). Factors known to influence the risk of soil Pb contamination to humans include the chemical form of Pb, the size of Pb-rich soil particles, nutritional and behavioural factors (Chaney and Ryan, 1994). For example, there is evidence to suggest that the absorption of Pb and its retention in the human gastrointestinal tract is significantly reduced if the Pb is ingested with meals (James et al., 1985). Elwood (1986) presents a lucid review of the contribution of various environmental sources of Pb to blood Pb in humans. Although seldom addressed in regulatory guidelines, bioavailability is one of the single most important concepts in contaminated site assessment and management. The concentration of a contaminant that is in a form able to be absorbed by the body or is mobile and able to be taken up into plants is the bioavailable fraction. This quantity is important in health risk assessment in order to calculate the toxicity of a particular contaminant, because intake is not necessarily the same as absorbed dose (Langley, 1996). The ANZECC/NH&MRC (1992) guidelines state that in time, it is ‘‘conceivable that the concentration of bioavailable metal, for example, will be listed rather than total metal’’. To some degree, this has already occurred. The current guidelines used in Australia in NSW include a Provisional Phytotoxicity-based Investigation Level (PPIL) for a range of contaminants (NSW Environment Protection Authority, 1998). This guideline is currently only applicable to sandy loams and is intended as a screening guide. For sandy loams with a pH of 6 – 8 the PPIL for Pb is 600 mg/kg. An area of current research that may help determine health risk is focussed on the development of Physiologically Based Extraction Tests (PBETs). These in vitro tests aim to simulate conditions such as pH and chemistry, in the stomach and small intestine, to determine the amount of Pb ingested in a soil matrix that would dissolve and become available for absorption (Davis et al., 1992; Ruby et al., 1992, 1993, 1996; Sheppard et al., 1995). This fraction of Pb is known as bioaccessible, as distinct from bioavailable. The PBET only estimates the solubility of Pb in the gastrointestinal tract. It does not replicate the entire physiological process controlling Pb uptake (Ruby et al., 1996). The bioavailable quantity of Pb is likely to be smaller than the bioaccessible fraction as a result of incomplete absorption of solubilised Pb (Ruby et al., 1996). PBETs therefore enable prediction of bioavailability based on the solubility of Pb in the body. The tests also account for factors such as stomach mixing and emptying rates. This has obvious implications for assessing the contribution of the ingested soil pathway in an exposure assessment in a site specific manner (Ruby et al., 1996).

409

4. Conclusions Data describing the range of lead concentrations in soil around the world is abundant. Typical Pb concentrations found in agricultural, urban and industrial soil are well defined. To identify patterns in the location of a contaminant and delineate areas that may be hazardous, it is essential to present survey data in the form of a map. Only a small proportion of studies have attempted to spatially describe Pb concentrations. The use of spatial prediction techniques to describe the spatial distribution of contaminants in soil is increasing. Kriging has been successfully enacted to predict both soil Pb concentrations and the probability of them exceeding an environmental threshold. Scope exists for research into the spatial distribution of Pb in contaminated soil and the application of spatial prediction techniques in contaminated site assessment. Such techniques may then be incorporated into existing risk assessment methodologies. A basic framework exists for environmental and health risk assessment, however, current protocols require further development and more rigorous definition to effectively and accurately estimate the risk posed by a particular contaminant.

References Albasel N, Cottenie A. Heavy metal contamination near major highways, industrial and urban areas in Belgian grassland. Water, Air, Soil Pollut 1985;24:103 – 9. Australian and New Zealand Environment and Conservation Council (ANZECC)/National Health and Medical Research Council (NH&MRC). Australian and New Zealand guidelines for the assessment and management of contaminated sites, ANZECC and NH&MRC, 1992. Archer FC, Hodgson IH. Total and extractable trace element content of soils in England and Wales. J Soil Sci 1987;38:421 – 31. Arrouays D, Mench M, Amans V, Gomez A. Short range variability of fallout Pb in a contaminated soil. Can J Soil Sci 1996;76:73 – 81. Atteia O, Dubois J-P, Webster R. Geostatistical analysis of soil contamination in the Swiss Jura. Environ Pollut 1994;86:315 – 27. Beavington F. Contamination of soil with zinc, copper, lead and cadmium in the Wollongong city area. Aust J Soil Res 1973;11:27 – 31. Bierkens MFP. Using stratification and residual kriging to map soil pollution in urban areas. In: Baafi EY, Schofield NA, editors. Geostatistics Wollongong ’96 vol. 2. The Netherlands: Kluwer Academic Publishers, 1997. pp. 996 – 1007. Bonifacio E, Melis P, Senette C, Zanini E. Spatial dependence of Pb and Cd in some Sardinia soils. Fresenius Environ Bull 1996;5:517 – 22. Boon DY, Soltanpour PN. Lead, cadmium, and zinc contamination of Aspen garden soils and vegetation. J Environ Qual 1992;21:82 – 6. Bottomley GA, Boujos LP. Lead in soil of Heirisson Island Western Australia. Search 1975;6:389 – 90. Brody DJ, Pirkle JL, Kramer RA, Flegal KM, Matte TD, Gunter EW, Paschal DC. Blood lead levels in the US population. J Am Med Assoc 1994;272:277 – 83. Brus DJ, de Gruijter JJ. Random sampling or geostatistical modelling? Choosing between design-based and model-based sampling strategies for soil. Geoderma 1997;80:1 – 44. Burkitt A, Lester P, Nickless G. Distribution of heavy metals in the vicinity of an industrial complex. Nature 1972;238:327 – 8. Cartwright B, Merry RH, Tiller KG. Heavy metal contamination of soils

410

J. Markus, A.B. McBratney / Environment International 27 (2001) 399–411

around a lead smelter at Port Pirie, South Australia. Aust J Soil Res 1977;15:69 – 81. Chaney RL, Ryan JA. Risk based standards for arsenic, lead and cadmium in urban soils. In: Kreysa G, Wiesner J, editors. DECHEMA Deutsche Gesellschaft fur Chemisches Apparatewesen. Frankfurt, Germany: Chemische Technik und Biotechnologie, 1994. Clift D, Dickson IE, Roos T, Collins P, Jolly M, Klindworth A. Accumulation of lead beside the Mulgrave Freeway Victoria. Search 1983; 14:155 – 7. Culbard EB, Thornton I, Watt J, Wheatley M, Moorcroft S, Thompson M. Metal contamination in British urban dusts and soils. J Environ Qual 1988;17:226 – 34. Czarnowska K, Walczak J. Distribution of zinc, lead and manganese in soils of Lodz City. Rocz Glebozn 1988;T39(Z1S):19 – 27. Czarnowska K, Gworek B, Majchrzak B. Spatial distribution of lead, zinc, copper and manganese in Pabianice soils. Ann Warsaw Agric Univ, SGGW Agric 1992;24:27 – 32. David DJ, Williams CH. Heavy metal contents of soils and plants adjacent to the Hume Highway near Marulan, New South Wales. Aust J Exp Agric Anim Husb 1975;15:414 – 8. Davies BE, Ballinger RC. Heavy metals in soils in north Somerset, England, with special reference to contamination from base metal mining in the Mendips. Environ Geochem Health 1990;12:291 – 300. Davies DJA, Watt JM, Thornton I. Lead levels in Birmingham dusts and soils. Sci Total Environ 1987;67:177 – 85. Davies DJA, Thornton I, Watt JM, Culbard EB, Harvey PG, Delves HT, Sherlock JC, Smart GA, Thomas JFA, Quinn MJ. Lead intake and blood lead in two-year-old UK urban children. Sci Total Environ 1990;90:13 – 29. Davis A, Ruby MV, Bergstrom PD. Bioavailability of arsenic and lead in soils from the Butte, Montana, mining district. Environ Sci Technol 1992;26:461 – 8. Dickson EL, Stevens RJ. Extractable copper, lead, zinc and cadmium in Northern Ireland Soils. J Sci Food Agric 1983;34:1197 – 205. Elwood PC. The sources of lead in blood: a critical review. Sci Total Environ 1986;52:1 – 23. Engel B, Lalor GC, Vutchkov MK. Spatial pattern of arsenic and lead distributions in Jamaican soils. Environ Geochem Health 1996; 18:105 – 11. Ferguson CC. The statistical basis for spatial sampling of contaminated land. Ground Eng 1992;25:34 – 8. Frangi J-P, Richard D. Heavy metal soil pollution cartography in northern France. Sci Total Environ 1997;205:71 – 9. Ginevan ME, Splitstone DE. Improving remediation decisions at hazardous waste sites with risk-based geostatistical analysis. Environ Sci Technol 1997;31:92A – 6A. Goovaerts P. Geostatistics for natural resources evaluation New York: Oxford Univ. Press, 1997. Hafen MR, Brinkmann R. Analysis of lead in soils adjacent to an interstate highway in Tampa, Florida. Environ Geochem Health 1996;18:171 – 9. Haneberg WC, Austin GS, Brandvold LA. Soil lead distribution at an abandoned smelter site in Socorro, New Mexico. Environ Geol 1993;21:90 – 5. Hawley JK. Assessment of health risk from exposure to contaminated soil. Risk Anal 1985;5:289 – 302. Holmgren GGS, Meyer MW, Chaney RL, Daniels RB. Cadmium, lead, zinc, copper and nickel in agricultural soils of the United States of America. J Environ Qual 1993;22:335 – 48. James HM, Hilburn ME, Blair JA. Effects of meals and meal times on uptake of lead from the gastrointestinal tract in humans. Hum Toxicol 1985;4:401 – 7. Kabata-Pendias A, Dudka S. Baseline data for cadmium and lead in soils and some cereals of Poland. Water, Air, Soil Pollut 1991;57 – 58:723 – 31. Klein DH. Mercury and other metals in urban soils. Environ Sci Technol 1972;6:560 – 2. Krueger JA, Duguay KM. Comparative analysis of lead in Maine urban soils. Bull Environ Contam Toxicol 1989;42:574 – 81.

Lagerwerff JV, Specht AW. Contamination of roadside soil and vegetation with cadmium, nickel, lead and zinc. Environ Sci Technol 1970; 4:583 – 6. Langley A. Health risk assessment and management of contaminated sites in Australia. In: Naidu R, et al, editor. Contaminants in the soil environment in the Australasia – Pacific region. The Netherlands: Kluwer Academic Publishers, 1996. pp. 281 – 307. Laslett GM, McBratney AB. Planning the sampling of soil that may be contaminated. In: Hazelton PA, Koppi AJ, editors. Soil technology: applied soil science. Sydney, Australia: Australian Society of Soil Science (NSW Branch) and Department of Agricultural Chemistry and Soil Science, The University of Sydney, 1993, pp. 365 – 384. Leharne S. A survey of metal levels in street dusts in an inner London neighbourhood. Environ Int 1992;18:263 – 70. Linzon SN, Chai BL, Temple PJ, Pearson RG, Smith ML. Lead contamination of urban soils and vegetation by emissions from secondary lead industries. J Air Pollut Control Assoc 1976;26:650 – 4. Little P, Martin MH. A survey of zinc, lead and cadmium in soil and natural vegetation around a smelting complex. Environ Pollut 1972;3:241 – 54. Lottermoser BG. Natural enrichment of topsoils with chromium and other heavy metals, Port Macquarie, New South Wales, Australia. Aust J Soil Res 1997;35:1165 – 76. Markus JA, McBratney AB. An urban soil study: heavy metals in Glebe Australia. Aust J Soil Res 1996;34:453 – 65. Maskall J, Thornton I. Metal contamination of soils at historical lead smelting sites. Land Contam Reclam 1993;1:92 – 100. Matheron G. The theory of regionalized variables and its applications Paris: Cahiers du Centre de Morphologie Mathe´matique de Fontainebleu, 1971 (No. 5). McBratney AB, Webster R. Choosing functions for semi-variograms of soil properties and fitting them to sampling estimates. J Soil Sci 1986; 37:617 – 39. McGrath SP. The range of metal concentrations in topsoils of England and Wales in relation to soil protection guidelines. Trace Subst Environ Health 1986;20:242 – 52. McGrath SP, Loveland PJ. The soil geochemical atlas of England and Wales London: Blackie Academic and Professional, 1992a. McGrath SP, Loveland PJ. The geochemical survey of topsoils in England and Wales. In: Beck BD, editor. Trace substances in environmental health vol. 25. 1992b;39 – 51. Merry RH, Tiller KG. Distribution and budget of cadmium and lead in an agricultural region near Adelaide South Australia. Water, Air, Soil Pollut 1991;57 – 58:171 – 80. Merry RH, Tiller KG, Alston AM. Accumulation of copper, lead and arsenic in some Australian orchard soils. Aust J Soil Res 1983;21:549 – 61. Mielke HW. Lead in residential soils: background and preliminary results of New Orleans. Water, Air, Soil Pollut 1991;57 – 58:111 – 9. Mira M, Bawden-Smith J, Causer J, Alperstein G, Karr M, Snitch P, Waller MJ, Fett MJ. Blood lead concentrations of preschool children in Central and Southern Sydney. Med J Aust 1996;164:399 – 402. Moehr AD, Roberts DW, Phillips PE, Evans RG. Childhood lead poisoning near abandoned lead mining and smelting areas: a case study of two affected households. J Environ Health 1993;56:20 – 3. Motto HL, Daines RH, Chilko DM, Motto CK. Lead in soils and plants: its relationship to traffic volume and proximity to highways. Environ Sci Technol 1970;4:231 – 8. National Academy of Sciences. Risk assessment in the federal government: managing the process. Washington, DC: National Academy Press, 1983. p. 191. Needleman HL. Low level lead exposure and neuropsychological performance. In: Rutter M, Russell Jones R, editors. Lead versus health — sources and effects of low level lead exposure. Chichester: Wiley, 1983. pp. 229 – 48. Needleman HL, Landrigan PJ. The health effects of low level exposure to lead. Annu Rev Public Health 1981;2:277 – 98. Needleman HL, Bellinger D. The health effects of low level exposure to lead. Annu Rev Public Health 1991;12:111 – 40.

J. Markus, A.B. McBratney / Environment International 27 (2001) 399–411 Needleman HL, Schell A, Bellinger D, Leviton A, Allred EN. The longterm effects of exposure to low doses of lead in childhood: an 11-year follow-up report. N Engl J Med 1990;322:83 – 8. NSW Environment Protection Authority. Contaminated sites — sampling design guidelines, NSWEPA Sydney, 1995. NSW Environment Protection Authority. Contaminated sites — Guidelines for the NSW site auditor scheme, 1998. Olszowy H, Torr P, Imray P, Smith P, Hegarty J, Hastie G. Trace element concentrations in soils from rural and urban areas of Australia. Contaminated Sites Monograph Series No. 4, South Australian Health Commission, 1995. Piotrowska M, Dudka S, Ponce-Hernandez R, Witek T. The spatial distribution of lead concentrations in the agricultural soils and main crop plants in Poland. Sci Total Environ 1994;158:147 – 55. Pouyat RV, McDonnell MJ. Heavy-metal accumulations in forest soils along an urban – rural gradient in southeastern New York, USA. Water, Air, Soil Pollut 1991;57 – 58:797 – 807. Pouyat RV, McDonnell MJ, Pickett STA. Soil characteristics of oak stands along an urban – rural land-use gradient. J Environ Qual 1995;24:516 – 26. Purves D, Mackenzie EJ. Trace-element contamination of parklands in urban areas. J Soil Sci 1969;20:288 – 90. Ruby MV, Davis A, Kempton JH, Drexler JW, Bergstrom PD. Lead bioavailability: dissolution kinetics under simulated gastric conditions. Environ Sci Technol 1992;26:1242 – 8. Ruby MV, Davis A, Link TE, Schoof R, Chaney R, Freeman G, Bergstrom P. Development of an in vitro screening test to evaluate the in vivo bioaccessibility of ingested mine-waste lead. Environ Sci Technol 1993;27:2870 – 7. Ruby MV, Davis A, Schoof R, Eberle S, Sellstone CM. Estimation of lead and arsenic bioavailability using a physiologically based extraction test. Environ Sci Technol 1996;30:422 – 30. Ryan JA, Chaney RL. Issues of risk assessment and its utility in development of soil standards: the 503 methodology an example. In: Prost R, editor. Contaminated soils: Third International Conference on the Biogeochemistry of Trace Elements, Paris, May 15 – 19, 1995.1997;393 – 413 (colloque 85, INRA editions, Paris, France). Sheppard SC, Evenden WG, Schwartz WJ. Ingested soil: bioavailability of sorbed lead, cadmium, cesium, iodine and mercury. J Environ Qual 1995;24:498 – 505. Simms DL, Beckett MJ. Contaminated land: setting trigger concentrations. Sci Total Environ 1987;65:121 – 34. Singer MJ, Hanson L. Lead accumulation in soils near highways in the twin cities metropolitan area. Soil Sci Soc Am Proc 1969;33:152 – 3. Spittler TM, Feder WA. A study of soil contamination and plant lead uptake in Boston urban gardens. Commun Soil Sci Plant Anal 1979;10:1195 – 210. Steinnes E, Allen RO, Petersen HM, Rambæk JP, Varskog P. Evidence of large scale heavy-metal contamination of natural surface soils in Norway from long-range atmospheric transport. Sci Total Environ 1997;205:255 – 66. Tao S. Spatial structures of copper, lead and mercury contents in surface soil in the Shenzhen area. Water, Air, Soil Pollut 1995a;82:583 – 91. Tao S. Kriging and mapping of copper, lead and mercury contents in

411

surface soil in Shenzhen area. Water, Air, Soil Pollut 1995b;83: 161 – 72. Theelen RMC, Nijhof AG, Bomer H. Dutch methodology for risk assessment of contaminated soils. In: Prost R, editor. Contaminated soils: Third International Conference on the Biogeochemistry of Trace Elements, Paris, May 15 – 19, 1995.1997;425 – 32 (colloque 85, INRA editions, Paris, France). Thomas VM, Socolow RH, Fanelli JJ, Spiro TG. Effects of reducing lead in gasoline: an analysis of the international experience. Environ Sci Technol 1999;33:3942 – 8. Tiller KG. Urban soil contamination in Australia. Aust J Soil Res 1992;30:937 – 57. Tiller KG, de Vries MPC, Spouncer LR, Smith L, and Zarcinas B. Environmental Pollution of the Port Pirie Region: 3. Metal contamination of home gardens in the city and their vegetable produce. CSIRO Australia. Division of Soils, Divisional Report No. 15, 1976. Tiller KG, Smith LH, Merry RH, Clayton PM. The dispersal of automotive lead from metropolitan Adelaide into adjacent rural areas. Aust J Soil Res 1987;25:155 – 66. USEPA. Risk assessment guidance for Superfund. US Environment Protection Agency. Office of Emergency and Remedial Response. Washington DC, Publ 9285.7-01B, 1989. Vegter JJ. Soil protection in the Netherlands: the continuing story. In: Prost R, editor. Contaminated soils: Third International Conference on the Biogeochemistry of Trace Elements, Paris, May 15 – 19, 1995. 1997;433 – 44 (colloque 85, INRA editions, Paris, France). Vollmer MK, Gupta SK, Krebs R. New standards on contaminated soil in Switzerland — comparison with Dutch and German quality criteria. In: Prost R, editor. Contaminated soils: Third International Conference on the Biogeochemistry of Trace Elements, Paris, May 15 – 19, 1995. 1997;445 – 57 (colloque 85, INRA editions, Paris, France). von Steiger B, Webster R, Schulin R, Lehmann R. Mapping heavy metals in polluted soil by disjunctive kriging. Environ Pollut 1996;94:205 – 15. Walter C, McBratney AB, Viscarra Rossel RA, Markus JA. A spatial-pointprocess analysis of lead contamination in Glebe New South Wales. Geoderma. (Submitted). Webster R, Oliver MA. Statistical methods in soil and land resource survey Oxford: Oxford Univ. Press, 1990. Weitzman M, Aschengrau A, Bellinger D, Jones R, Hamlin JS, Beiser A. Lead-contaminated soil abatement and urban childrens blood lead levels. J Am Med Assoc 1993;269:1647 – 54. Wilcke W, Mu¨ller S, Kanchanakool N, Zech W. Urban soil contamination in Bangkok: heavy metal and aluminium partitioning in topsoils. Geoderma 1998;86:211 – 28. Wixson BG, Davies BE. Guidelines for lead in soil: proposal of the Society for Environmental Geochemistry and Health. Environ Sci Technol 1994;28:26A – 31A. Wopereis MC, Gascuel-Odoux C, Bourrie G, Soignet G. Spatial variability of heavy metals in soil on a one hectare scale. Soil Sci 1988;146:113 – 8. Wylie PB, Bell LC. The effect of automobile emissions on the lead content of soil and plants in the Brisbane area. Search 1973;4:161 – 2. Zanini E, Bonifacio E. Lead pollution of soils from a continuous point source: a case study in Italy. J Environ Sci Health, Part A: Environ Sci Eng 1991;26:777 – 96.