Original article

Trace-metal pollution of soils in northern England B. G. Rawlins Æ T. R. Lister Æ A. C. Mackenzie

tion (Reimann and others 2000). At the local scale, site investigations are undertaken to establish the total concentration of a range of trace metals (heavy metals and metalloid elements) in soil which may pose a risk to human health, or cause the pollution of controlled waters. In many countries trace-metal concentrations are measured in areas where sewage sludge is applied to agricultural land to ensure that guideline values are not exceeded. By establishing some form of background concentration for trace metals from a regional soil survey, the results of local, site-specific measurements can be put into context. Environmental geochemical data from soil surveys undertaken at the regional scale generally have non-normal distributions which are often positively skewed (Reimann and Filzmoser 2000). Statistical techniques have been developed to identify thresholds to separate ‘background’ concentrations from sites which have naturally elevated levels or indicate some form of anthropogenic contamination, although there is no universally adopted method (Matschullat and others 2000). Natural or background concentrations of trace metals vary widely, even in undisturbed soils; for example, lead (Pb) concentrations in topsoils throughout rural areas of England and Wales Keywords Diffuse pollution Æ Background range from 20 to approximately 120 mg kg–1 between the levels Æ Trace metals Æ Soil survey Æ Geochemistry Æ 10 and 90th percentiles of the sample distribution reCoal measures spectively (McGrath and Loveland 1992). The primary control on trace-metal contents of undisturbed soil in temperate regions such as the United Kingdom is typically the geochemical composition of the soil parent material (the bedrock geology or Quaternary deposit from which it Introduction formed). Soil geochemistry is determined at a range of spatial scales, In its definitions of soil quality, the International Standards Organisation distinguishes between the natural from national or regional surveys to local, site-specific background concentration, ‘.... derived solely from natural investigations. Regional soil geochemical surveys have recently been used to determine the magnitude and extent sources’, and the background concentration which ‘arises of areas affected by sources of anthropogenic contamina- from both natural sources and non-natural diffuse sources such as atmospheric deposition’ (ISO 1996). Although the latter definition is instructive in limiting the inclusion of contaminants to those from diffuse sources, neither provides any guidance on methods which could be used to Received: 27 June 2001 / Accepted: 30 January 2002 Published online: 6 April 2002 establish the range or upper limit of background values. ª Springer-Verlag 2002 One way of establishing background concentrations of trace metals in soils is to define upper and lower limits using values from the analysis of samples from relatively B.G. Rawlins (&) Æ T.R. Lister Æ A.C. Mackenzie pristine areas with the same parent material, which are British Geological Survey, neither highly mineralised nor subject to significant Keyworth, Nottingham, NG12 5GG, UK E-mail: [email protected] anthropogenic contamination. Elevated trace-metal Abstract Data from a regional geochemical survey of topsoils (n=818) in rural and peri-urban areas over a single parent material (Coal Measures) are used to identify two types of trace-metal pollution – severe local contamination at 20 sites and widespread, diffuse pollution in more densely populated areas. Median concentrations of several trace metals in topsoils were significantly higher in areas of high, compared to low, population density (percentage increases in parenthesis): As (31), Cu (39), Fe (7), Mo (26–36), Ni (29), Pb (20), Sn (40), and Zn (11). Four potential pathways of diffuse trace-metal pollution are postulated: coal-ash dispersal, atmospheric aerosols derived from coal combustion, the historical spreading of sewage waste, and those related to road vehicles. The statistical analysis of geochemical data classified by local, human population density can be an effective means of identifying the magnitude and extent of diffuse pollution, and could help to establish natural background levels.

612

Environmental Geology (2002) 42:612–620

DOI 10.1007/s00254-002-0564-5

Original article

concentrations are often found in soils developed over mineralised bedrock, and methods have been devised to identify thresholds in geochemical data for the purpose of mineral exploration (Sinclair 1974). Several methods have been proposed for establishing the range of background values by removing the highest and lowest values from the sample distribution (Matschullat and others 2000). For example, the ‘natural geochemical baseline’ has been defined by Tidball and Ebens (1976) as encompassing the central 95% of the observed concentrations (i.e. from 2.5 to 97.5% of the distribution). By statistical analysis of a data set, it may be possible to identify a threshold between background values and a limited number of high tracemetal concentrations (resulting from severe contamination or mineralisation). However, in areas with a legacy of atmospheric pollution where trace metals have been deposited over considerable distances, the cumulative distribution may reflect a continuum from almost pristine sites, through sites affected by diffuse contamination, to grossly contaminated sites. In such circumstances, it may be difficult to identify the subtle impact of diffuse pollution on trace-metal contents. It would also require far greater effort to remove the impact of diffuse pollution in establishing natural background concentrations (according to the ISO definition). This study focussed on several trace metals (As, Cu, Cr, Ni, Pb and Zn) which are often cited as being of environmental concern. The aim of this paper was to investigate a data set from a regional soil survey to identify local and diffuse trace-metal contamination, and suggest their potential sources and pathways. The implications of diffuse pollution for establishing natural background concentrations of trace metals are discussed.

Study region The geology of the region consists of Westphalian sequences (ca. 300 million years old) of the Carboniferous Coal Measures (Fig. 1) and has been described by Downie (1960), whilst the soils have been described by the Soil Survey of England and Wales (Soil Survey 1984). The Lower and Middle Coal Measures in this region consist of mudstones, shales and inter-bedded sandstones. The sandstones and shales alternate to form a succession of sandstone, seat earth, coal, marine shales, mudstones and siltstones. Sample sites in the study region occur in a broad range of land-use settings, from semi-natural, rural environments with limited development to the fringes of large, urban conurbations. The main land-cover types in areas around the soil sample locations include tilled land (39%), meadow (21%), suburban (18%), and grazed turf (10%; ITE land cover map; Fuller and others 1994). Elevation across the region varies between around 50 and 250 m above sea level. The soil geochemical atlas of England and Wales shows that the trace-metal content of the soils developed in the region are generally high (McGrath and Loveland 1992). Soils throughout parts of northern England were subject to

considerable atmospheric pollution during the last century, resulting from the deposition of aerosols following the domestic and industrial combustion of coal mined from the underlying Carboniferous Coal Measures. A resume of industrial activity throughout the region, which gives an indication of the magnitude of coal combustion, has been provided by Gilbertson and others (1997). Air pollution in Sheffield (a large conurbation in the region) at the start of the 20th century was widespread and intense, and by the middle of the century had contributed to the establishment of legislation, most notably through the Clean Air Act of 1956. The level of air pollution subsequently declined, as did the size of the steel manufacturing industry, in the 1970s and 1980s.

Soil survey and chemical analysis A regional soil geochemical survey covering the entire Humber-Trent region (Fig. 1) has been undertaken by the British Geological Survey as part of its G-BASE (Geochemical Baseline Survey of the Environment) Programme. As part of this regional survey, a total of 818 topsoil samples were collected over the Lower and Middle Coal Measures Formations, where there were no extensive Quaternary deposits, ensuring that all soils were derived from the same parent material type. The samples described in this paper were collected in the summer months of 1994, 1995 and 1996, and the sites extended across four counties (South Yorkshire, West Yorkshire, Derbyshire, Nottinghamshire), covering an area of approximately 2,100 km2. This gives an average sample resolution of one site per 2.5 km2. Sites were selected on a systematic basis from every second kilometre square of the British National Grid. Site selection in each square was random, subject to the avoidance where possible of roads, tracks, railways, human habitation and other disturbed ground. Samples were not collected in urban areas, but were collected around them (in peri-urban areas). At each sample site, five holes were augered at the corners and centre of a square with a side length of 20 m using a hand auger. There are two main soil sampling approaches for comparing the concentrations of elements across a region – the collection of samples from the same soil horizon in each land-use type or, alternatively, sampling over a specified depth range. Sampling specific soil horizons (which occur at different depths across the landscape) would lead to variations in the amounts of any pollutants present due to their increased attenuation with depth from the surface. It was decided that soil samples would be collected at the same depths at each site, between 0 and 15 cm. Soil samples from each of five holes were combined to form an aggregated sample. At each site, information on the location, catchment geology, contamination, land use and other features required for data interpretation was recorded on a field card and subsequently transferred to an electronic database. All soils were disaggregated following drying and sieved to 2 mm. All samples were coned and quartered, and a

Environmental Geology (2002) 42:612–620

613

Original article

Fig. 1 The study region and sampling locations

50-g subsample ground in an agate planetary ballmill. The total concentration of 24 major and trace elements (listed in Table 1) was determined in each sample by wavelength dispersive XRF (X-ray fluorescence). The lower limits of detection for each element are also shown. The detection limit for Cd (2.5 mg kg–1) by XRF analysis was too high to provide reliable data on its concentration in samples throughout the study region, and the data has therefore not been included in this paper. 614

Environmental Geology (2002) 42:612–620

Throughout the broader regional survey (Fig. 1), two samples were collected at 84 duplicate sites; the duplicates were collected immediately adjacent to the original survey sites. Prior to chemical analysis, the duplicate samples were split into two subsamples, giving a total of 336 separate analyses. This sampling design is suitable for the application of analysis of variance (ANOVA) using a nested design (Snedecor and Cochran 1989). The results of such an analysis provide estimates of the components of variance accounted for by differences between sites

1 0.05 1 1 1 1 1 0.005 26.0 0.3 100 76.8 1.1 8.4 69.8 0.9 2.0 0.1 21.0 8.0 0.5 1.0 21.0 0.4 233 1.3 1,868 159 22.0 161 245 1.2 24.0 0.3 79.0 75.0 0.5 6.0 67.0 0.9 12.2 0.1 101 21.8 1.7 9.4 19.5 0.2 46.9 44.7 100 28.5 148 111 28.0 17.7 7.2 1.9 9.0 0.4 7.1 7.5 2.7 –1.1 107 6.6 127 0.5 68.2 94.7 17.5 0.8 10.4 0.1 35.0 36.4 0.5 3.0 43.0 0.5 46.6 0.7 272 125 5.0 30.1 112 1.1 18 13 20 21 13 21 18 8

1 2.6 0.3 5.9 2.6 0.6 23.1 0.3 2.3 1.4 3.9 18

1 1 1 103 112.2 263 23.0 10.0 99.0 238 1,982 4,320 103 101 256 23.9 85.3 154 23.2 76.0 58.6 0.5 13.9 22.4 2.2 286 590 60.4 44.4 150 153 247 388 20 20 20

(geochemical variance), sampling at-a-site (sampling variance), and analytical precision (analytical variance). The two latter components have been referred to as the technical variance, and it has been suggested that their sum should not be greater than about 20%, with a maximum analytical component of 4% (Ramsey and others 1992). The results of a nested ANOVA using the duplicate and subsample data are presented in Fig. 2 for each element. In each case (with the exception of Co and Sb) the technical variance is less than 20%, and in most cases the analytical variance is well below 4%. We therefore feel justified in asserting that the analytical data meet our requirements for precision.

1 3.9 1.0 43.2 3.3 2.5 64.9 6.6 82.4 1.7 9.3 20

Identifying local contamination

a

Value indicates the number of samples in the dataset above the 97.5 percentile

0.005 0.2 0.001 0.9 0.2 0.1 49.1 2.1 16.6 0.0 0.3 26 0.01 0.1 6.6 0.8 0.3 0.1 18.5 6.3 6.7 0.8 1.7 0.4 25.7 48.1 0.4 6.4 3.8 68.9 3.2 0.3 10.1 1.5 20 15 1 1 92.5 39.2 25.0 4.0 2,534 703 84.0 30.5 97.9 36.0 105 91.8 20.3 9.0 483 144 50.0 12.0 154 126 20 18 1 26.2 3.0 78.0 27.0 6.8 26.1 0.3 4.7 12.0 38.0 18 1 1 0.05 21.6 492 0.8 3.0 126 0.0 101 7,900 10.5 18.0 437 0.6 11.7 397 0.9 53.7 80.7 116 2.3 12.3 5.2 7.4 198 35.8 9.0 273 0.1 55.0 1,018 3.2 19 21 21 0.1 14.1 4.2 22.1 14.2 2.5 17.9 –0.1 0.4 8.9 19.5 20 Detection limit Mean Min. Max. Median SD Coeff. of var. (%) Skewness Kurtosis 2.5%ile 97.5%ile Sites abovea

Sr Sn Sb Rb Pb P2O5 (%) Ni MnO Mo (%) Fe2O3 MgO (%) (%) Cu Cr Co CaO (%) Ba Al2O3 As (%)

Table 1 Statistical summary of the total concentration of major and trace elements in topsoil (n=818). All values are in mg kg–1 (unless otherwise stated)

TiO2 (%)

U

V

Zn

Zr

Original article

A statistical summary is provided in Table 1, and cumulative frequency graphs of each of six trace metals (As, Cr, Cu, Ni, Pb, Zn) are shown in Fig. 3. Each of the distributions of the six trace metals is positively skewed (Table 1). The median concentrations (50th percentile) of Cr, Cu, Pb and Zn from the National Soil Inventory (McGrath and Loveland 1992) are shown as filled arrows in Fig. 3. The median concentrations of each of the trace metals are, with the exception of Ni, higher in the study region than for the rest of England and Wales. The concentrations of Cr and Pb in the study region are particularly high when compared to the median values at the national scale. No survey data are currently available for As at the national scale. There are abrupt increases in trace-metal concentrations in the upper decile of their distributions (Fig. 4), suggesting that a limited number of sites have either been contaminated by local, point sources or reflect local mineralisation. Five of the trace metals have sharp inflections in the upper decile of their distributions (Ni, Cu, Cr, Zn and Pb) whereas As exhibits a more gradual rise (Fig. 4). The 97.5 percentile appears to be below the level at which concentrations rise sharply in each of the distributions. Lead (Pb) and Cr concentrations rise sharply between the two percentiles, whereas the rapid increase in Cu and Ni is above the 99th percentile. This suggests that if it were necessary to establish a uniform upper boundary in each distribution of the data set, and in so doing remove anomalously high values, it should be drawn at the 97.5 percentile. The number of sites with concentrations above the 97.5 percentile is shown in Table 1 for each of the six trace metals. These sites were selected using a GIS to determine whether their locations might explain the elevated concentrations. Many of them were in the periurban areas of medium to large conurbations throughout the study region, including Rotherham, Barnsley, Wakefield, Leeds, Bradford and Halifax. This suggests that many of the sites with concentrations above the 97.5 percentile are the result of local contamination, and we would therefore wish to exclude them in any attempt to define an upper threshold for background values. This finding shows that it is possible to identify and remove highly

Environmental Geology (2002) 42:612–620

615

Original article

developed over the two geological formations, the Lower (LCM: n=415) and Middle Coal Measures (MCM: n=413). As demonstrated in the preceding section, soils at a few sample sites had far greater trace-metal contents than the rest of the distribution, which can significantly increase contaminated sites from a regional survey data set, and the mean value. In such cases, the median is a more robust establish an upper boundary at a specific point in the measure of centrality in a data set. Hence, we performed a distribution to provide an upper limit to background Mood’s median test on the data classified into the two concentrations over a single parent material. separate formations (LCM and MCM) to identify those elements where the distributions were significantly different (Table 2). The Mood’s median test is a non-paratest (highly resistant to the outlying values present Investigating the impact of diffuse metric in the data set) and assumes that the data from each pollution population are independent random samples. Of the elements analysed Ni, Pb, Mg, Rb, Sr, Zr, Mo and Ba To investigate the potential effects of diffuse pollution had significantly different median values in topsoil samthroughout the region, we used human population density ples (Table 2). In subsequent statistical analyses, data for as an indicator of historical, anthropogenic contamination. these elements were analysed within each formation to However, before assessing whether any effect was appar- ensure that geological influences were not responsible for ent, it was first necessary to ensure there were no signifi- significant differences between sites. cant differences between trace-metal contents in soils Population density information was obtained from the 1991 UK Census Small Area Statistics data, part of the UK EDINA database. The data surface models were conFig. 2 Analytical, within- and between-site variance from the analysis of duplicate samples (and their subsamples) from 84 sites throughout northern England using a nested analysis of variance

Fig. 3 Log-normal cumulative percentage distributions for six trace metals Fig. 4 (n=818). The vertical arrows show the median (50th percentile) concentration of each trace metal in topsoils throughout England and The upper decile of Fig. 3. Dashed lines highlight the 97.5 and the 99th percentiles Wales (McGrath and Loveland 1992)

616

Environmental Geology (2002) 42:612–620

9.2 –13.7 0.4 1.2 –7.5 13.5 –1.6 11.8 –18.4 12.3 –6.3 8.5 –13.5 3.4 12.6

a

Significant difference at the 1% confidence level (Mood’s median test)

–1.5 9.6 –12.8 4.4 –6.2 –3.3

23.1 23.5

103 117 243 104 101 242 25.7 107 56.5 –2.9 0.0 –11.3a 2.6 2.6 0.6 0.0 0.8 0.9 0.2 2.8 74.3 72.5 19.7 14.2a 8.4 6.0 10.9 0 1.2 0.5 2.0 0 81.3 79.0 20.1 12.9a 90.4 67.0 80.2 –23.9a 0.3 0.3 0.1 7.7 3.6 27.1 3.0 26.0 1.9 9.1 –16.7a 13.0a 0.87 0.2 0.80 0.2 0.41 0.1 6.7a –6.3 6.6 6.6 1.7 2.7 41.2 31.0 43.8 –3.2 86.9 83.5 32.2 1.8 26.8 27.0 6.4 –3.7 21.0 17.0 11.3 11.8 13.9 14.0 2.6 2.5

544 0.9 464 0.6 511 1.1 17.0a –5.0

103 106 282 101 101 273 22.1 55.6 207 2.6 2.6 0.6 0.9 0.9 0.1 65.4 63.5 18.4 8.5 8.5 6.0 1.0 1.0 0.5 110 72.4 88.0 70.0 117 22.6 0.3 0.3 0.1 4.1 24.9 3.6 23.0 3.0 14.4 0.2 0.2 0.1 0.78 0.75 0.36 6.7 6.8 1.7 98.0 37.2 85.0 30.0 133 26.2 25.6 26.0 7.2 0.7 0.6 0.7 442 397 229 22.3 19.0 12.1 14.4 14.4 2.5

LCM (n=415) Mean Median SD MCM (n=413) Mean Median SD Difference between median values (%) Difference between mean values (%)

Zn V U TiO2 (%) Sr Sn Sb Rb Pb P2O5 (%) Ni MnO Mo (%) Fe2O3 MgO (%) (%) Cu Cr Co CaO (%) Ba Al2O3 As (%)

Table 2 Comparison of major- and trace-element concentrations in topsoils over the Lower (LCM) and Middle Coal Measures (MCM). All values are in mg kg–1 (unless otherwise stated)

Zr

Original article

structed from the population-weighted centroids defined for each 1991 enumeration district (Bracken and Martin 1989). The data set forms the basis for population surface construction, and the counts associated with each centroid are redistributed into the cells of a regular grid (with a cell size of 200 by 200 m). Using a GIS, the total population in each of the cells falling within a one-kilometre radius of each sampling site was calculated. The cumulative distribution of the total population within this radius around each sample site is shown in Fig. 5. Each of the soil samples was classified into one of two groups (high and low population) according to whether they had total populations above or below the median value (a total population of about 1,700 people within a 1-km radius; Fig. 5). For those elements which had no significant differences between their median values over the two formations, a Mood’s median test was applied to determine whether the distributions of values in the high- and low-population classes were significantly different (Table 3). For those elements in which a geological influence was identified in the preceding section, the same analysis was performed on the high- and low-population classes within each formation (Table 4). Trace elements with significantly higher median concentrations in areas of high compared to low population density (percentage increases in parenthesis in Table 3) across both formations include As (31.3), Cu (38.5), Zn (11.1), Sn (40) and Fe (6.5). From the withinformation analysis, soil samples with significantly higher median values of trace elements over the LCM in areas of high population (Table 4) were Ni (28.6), Pb (19.9) and Mo (35.5). For samples over the MCM, significantly higher median values of Pb (19.7) and Mo (25.9) were found in areas of high compared to low population (Table 4). These differences are, with the exception of Fe, much greater than the levels of analytical error (Fig. 2). There were no significant differences between the median values for the other major and trace elements (Cr, V, Co, U, Sb, Al, P, Ca, Ti, Mg, Rb, Sr, Zr and Ba; see Tables 3 and 4). Although correlation coefficients were calculated between population density and trace-metal contents, no consistent relationship could be demonstrated, suggesting that the

Fig. 5 Cumulative percentage of total population within a 1-km radius of each soil sample location (Copyright Crown Copyright Ordnance Survey, an EDINA Digimap/JISC supplied service)

Environmental Geology (2002) 42:612–620

617

Original article

Table 3 Comparison of major- and trace-element concentrations in areas of high and low population density. Elements shown are those with no significant difference in median values over Lower (LCM) and Middle Coal Measures (MCM; see Table 2). All values are in mg kg–1 (unless otherwise stated) Al2O3 As (%) Low population density (n=400) Mean 14.2 18.9 Median 14.3 16.0 SD 2.6 10.0 High population density (n=418) Mean 14.1 24.2 Median 14.1 21.0 SD 2.4 12.6 Increase in median –1.4 31.3a (low to high pop.) % Increase in mean –0.3 28.0 (low to high pop.) % a

CaO (%)

Co

0.8 0.6 1.0

25.5 26.0 7.7

0.8 0.6 0.9 7.1 8.3

Cr

Cu

Fe2O3 MnO (%) (%)

P2O5 (%)

Sb

85.1 82.0 38.7

31.0 26.0 19.9

6.3 6.5 1.8

0.2 0.2 0.1

0.3 0.3 0.1

0.9 0.5 1.2

26.8 27.0 5.9 3.8

99.6 86.0 131 4.9

47.0 36.0 45.1 38.5a

6.9 6.9 1.6 6.5a

0.2 0.2 0.1 –6.3

0.3 0.3 0.1 7.7

5.0

17.1

51.7

8.4

–2.9

10.5

Sn

TiO2 (%)

U

V

Zn

6.4 5.0 5.4

0.9 0.9 0.2

2.6 2.5 0.7

101 101 23.5

100 95.0 41.1

1.4 0.5 2.0 0.0

10.4 7.0 11.7 40.0a

0.9 0.9 0.2 –3.2

2.6 2.6 0.5 4.0

105 105 24.2 4.0

123 105 111 11.1a

56.7

62.8

–2.9

0.8

3.8

23.7

Significant difference at the 1% confidence level (Mood’s median test)

Table 4 Comparison of major- and trace-metal concentrations in areas of high Elements shown are those with a significant difference in median and low population density within each of the formations: the values over LCM and MCM (see Table 2). All values are in mg kg–1 Lower Coal Measures (LCM), and the Middle Coal Measures (MCM). (unless otherwise stated) Ba Lower Coal Measures (LCM) Low population density (n=224) Mean Median SD High population density (n=190) Mean Median SD Increase median: low to high pop. (%) Increase mean: low to high pop. (%) Middle Coal Measures (MCM) Low population density (n=175) Mean Median SD High population density (n=228) Mean Median SD Increase median: low to high pop. (%) Increase mean: low to high pop. (%) a

430 382 240 455 416 215 8.8 5.9 554 462 620 536 470 410 1.7 –3.3

MgO (%)

Mo

Ni

Pb

Rb

Sr

Zr

0.8 0.7 0.4

3.6 3.1 2.1

22.1 21.0 10.1

103 80.5 93.0

72.3 70.0 23.8

63.4 61.5 17.9

273 274 63.3

0.8 0.8 0.2 14.3 3.8

4.8 4.2 3.6 35.5a 35.8

28.3 27.0 17.7 28.6a 28.4

119 96.5 140 19.9a 14.8

72.5 70.5 21.0 0.7 0.3

67.8 65.5 18.7 6.5 7.1

292 273 299 –0.5 6.8

0.9 0.8 0.5

3.1 2.7 1.4

25.8 25.0 8.3

74.7 61.0 56.8

83.2 82.0 21.8

74.2 73.0 18.5

246 245 53.0

0.8 0.8 0.3 0.0 –8.6

4.0 3.4 2.1 25.9a 30.7

28.0 26.0 9.7 4.0 8.4

102 73.0 92.6 19.7b 37.0

79.8 79.0 18.7 –3.7 –4.1

74.3 72.0 20.6 –1.4 0.1

241 241 59.2 –1.6 –1.9

Significant difference at the 1% confidence level (Mood’s median test) Significant difference at the 2% confidence level (Mood’s median test)

b

pattern of diffuse contamination is complex, and only becomes apparent when data for a large number of samples are combined. There are two lines of evidence to suggest that topsoils have been contaminated by diffuse sources in areas of high compared to low population density throughout the study region. First, after accounting for potential differences caused by variations in geological formation (in the case of Mo and Pb), there are consistent, large, statistically significant differences between trace-element concentrations (As, Cu, Mo, Pb, Sn and Zn) in areas of high and low population density (Tables 3 and 4). These elements have 618

Environmental Geology (2002) 42:612–620

close geochemical associations with coal, which was used widely throughout the region. Second, the concentration of major and trace elements which are unlikely to have been affected by human activities have similar concentrations in areas of high and low population density, which suggests that the results do not simply reflect variations in bedrock geochemistry. We believe there are four pathways through which trace elements may have been dispersed around human settlements. The first two relate to the use of coal, which was the primary energy source in domestic dwellings and industry throughout the area during the preceding 150 years. The

Original article

first pathway is the spreading of coal ash, either as a soil amendment or waste disposal, although there is little published information on this practice. The second pathway is the atmospheric deposition of aerosols following domestic (and industrial) coal combustion. Arsenic, Mo and Pb have been shown to be closely associated with pyrite in UK coals (Spears and Zheng 1999), whereas Cu and Zn also form common pyritic minerals. Numerous studies have reported regional heavy-metal contamination of topsoil from aerosol deposition following coal combustion (Kapicka and others 1999; Moon and others 2000). Published data on the fate of trace elements following coal combustion are generally based on the analysis of feed coal and combustion byproducts, through which atmospheric emissions from large power plants can be calculated using a mass-balance approach. Application of this approach for a power plant in Spain demonstrated that the trace elements Cr, Ni, Cu, Zn, As, Mo, Sn and Pb were highly enriched in the finest (<10 lm) fly-ash fractions (Querol and others 1995). There is little published data on the emission of trace elements from domestic coal combustion, which would have been widespread throughout the study region during the last century. However, the correspondence between elements enriched in fine ash particles and those with significantly increased concentrations in densely populated areas suggests that the atmospheric dispersal of aerosols derived from domestic coal combustion is a pathway which could account for much of the observed difference in trace-metal concentrations. The third pathway is the historical spreading of domestic sewage waste, often referred to as ‘night soil’, on land close to urban settlements, whereby human and animal wastes were used as fertilisers in the 18th and 19th centuries. Again, little published information is available, although this practice was common. The fourth pathway is the emission of trace elements from road vehicles, including Pb in petrol and, to a lesser extent, Zn from rubber tyres. In each of these four pathways, solid wastes and aerosols are likely to be deposited at short distances from human settlements, leading to their accumulation in topsoils.

Fig. 6 Cumulative percentage of total arsenic concentrations in topsoils in areas of high and low population density

sample population, as shown for As in Fig. 6. This would be necessary to establish natural background concentrations according to the ISO definition. Likewise, methods based on normalising the data against a conservative variable (one which is not significantly influenced by diffuse contamination) are unlikely to be effective because the magnitude of natural variability is often greater than the subtle changes caused by diffuse pollution. From the analysis presented in this study it is clear that diffuse sources of pollution can increase natural background concentrations by a significant amount, and that environmental geochemical data sets need to be investigated thoroughly to identify such trends. Acknowledgements This paper is published with the permission of the Director of the British Geological Survey (NERC). The authors would like to thank all the BGS staff and volunteer workers involved in the collection and analysis of samples in the G-BASE programme, and two anonymous reviewers for their comments on a preliminary draft of the manuscript.

References Discussion Analysis of the data set indicates that trace-metal contents of topsoils have been affected by severe local pollution at around 20 sites (giving rise to very high concentrations; Table 1), and diffuse pollution which has given rise to median concentrations of between 11 and 40% higher in areas of high population density. Diffuse, trace-metal contamination of topsoils has rarely been demonstrated using data from a regional soil survey in the UK (Ander and others 2000). Most methods of establishing background values in environmental datasets based on the identification of outliers would be effective in identifying the sites affected by severe pollution. However, they cannot account for more diffuse forms of contamination where the concentration of an element has been raised in a significant proportion of the

Ander EL, Smith B, Fordyce FM, Rawlins BG (2000) Trace elements. In: Darcy B, Ferrier R, Ellis JB, Dils RM (eds) Diffuse pollution: economic and environmental impacts in the UK. Lavenham, Suffolk Bracken I, Martin D (1989) The generation of spatial population distributions from census centroid data. Environ Planning A:537–543 Downie C (1960) The area around Sheffield. The Geologists Association, Colchester Fuller RM, Groom GB, Jones AR (1994) The land cover map of Great Britain: an automated classification of Landsat Thematic Mapper data. Photogrammetric Eng Remote Sensing 60:553–562 Gilbertson DD, Grattan JP, Cressey M, Pyatt FB (1997) An airpollution history of metallurgical innovation in iron- and steelmaking: a geochemical archive of Sheffield. Water Air Soil Pollut 100:327–341 ISO (1996) Soil quality – vocabulary, part 1: terms and definitions relating to the protection and pollution of the soil. International Standards Organisation ISO 11074(1):22

Environmental Geology (2002) 42:612–620

619

Original article

Kapicka A, Petrovsky E, Ustjak S, Machackova K (1999) Proxy mapping of fly-ash pollution of soils around a coal-burning power plant: a case study in the Czech Republic. J Geochem Expl 66:291–297 Matschullat J, Ottenstein R, Reimann C (2000) Geochemical background – can we calculate it? Environ Geol 39:990–1000 McGrath SP, Loveland PJ (1992) The soil geochemical atlas of England and Wales. Blackie, Glasgow Moon JW, Moon HS, Woo NC, Hahn JS, Won JS, Song Y, Lin X, Zhao Y (2000) Evaluation of heavy metal contamination and implication of multiple sources from Hunchun basin, northeastern China. Environ Geol 39:1039–1052 Querol X, Fernandez-Turiel JL, Lopez-Soler A (1995) Trace elements in coal and their behaviour during combustion in a large power station. Fuel 74:331–343 Ramsey MH, Thompson M, Hale M (1992) Objective evaluation of precision requirements for geochemical analysis using robust analysis of variance. J Geochem Expl 44:23–36

620

Environmental Geology (2002) 42:612–620

Reimann C, Filzmoser P (2000) Normal and lognormal data distribution in geochemistry: death of a myth. Consequences for the statistical treatment of geochemical and environmental data. Environ Geol 39:1001–1014 Reimann C, Banks D, de Caritat P (2000) Impacts of airborne contamination on regional soil and water quality: the Kola Peninsula, Russia. Environ Sci Technol 34:2727–2732 Sinclair AJ (1974) Selection of thresholds in geochemical data using probability graphs. J Geochem Expl 3:129–149 Snedecor GW, Cochran WG (1989) Statistical methods, 8th edn. Iowa State University Press, Iowa Soil Survey (1984) Soils and their use in northern England. Soil Survey of England and Wales, Harpenden Spears DA, Zheng Y (1999) Geochemistry and origin of elements in some UK coals. Int J Coal Geol 38:161–179 Tidball RR, Ebens RJ (1976) Geochemical baselines for sagebrush and soil, Powder River Basin, Montana, Wyoming. US Geol Surv Rep, pp 6–13