7 Many factors influence whether a metal will be toxic to a particular organism; these include but are not limited to: chemical form, concentration, mobility, behaviour in the environment, bioavailability, acidity, soil texture and mineral composition. The Canadian Council of Ministers of the Environment (CCME) has produced soil quality guidelines for the protection of the environment and human health (Table 1; CCME 2007). These guidelines are recommended concentrations that should result in negligible risk. If the guidelines are exceeded additional work is required to determine whether a significant risk exists. The human health guideline is the soil metal concentration at which no appreciable human health risk is expected upon exposure. The derivation of the guideline takes into account background exposure from air, water, soil, food and consumer products and involves defining scenarios for a sensitive receptor (child or adult) and specific pathways of exposure (CCME 2007). It is adapted for various land use categories, including agricultural, residential, commercial and industrial. The environmental health guideline is based on an assessment of physical and chemical properties, sources and emissions, distribution in the environment, toxicity, and existing guidelines and standards for each metal. Ecological toxicity data are considered to ensure the guideline protects a wide variety of organisms.

8 Metals were chosen for analysis in this study based on the availability of CCME soil quality guidelines (Table 1). The ten metals selected are: As, Ba, Cd, Co, Cr, Cu, Ni, Pb, V, and Zn. As the soil quality guideline is intended to protect both environmental and human health, the lower of the two is the recommended guideline for any land use category. Where the CCME has not specified a human health guideline for a specific metal, the Ministry of Environment and Energy (MOEE) (Government of Ontario) human health guideline for that metal was adopted in this study (Table 1; MOEE 2009).

21 One approach to assessing and visualizing cumulative metal contamination on a landscape is to portray the degree of contamination in each sample using a contamination index (C _d ; Backman et al. 1998). The index sums up the combined influence of the selected metals by sample and when mapped may identify areas of concern or ‘hot spots’ of multi-metal contamination on the landscape. C _d is calculated according to the equation of Backman et al. (1998):

21 where, C _f i is the contamination factor for the i–th metal, C _Ai is the analytical value of the i–th metal, and C _Ni is the guideline value for the i–th metal.

22 The index normalizes each metal against its acceptable guideline value to produce a contamination factor (C _f i ). The C _f i is centered on zero, with values greater than zero exceeding calculation of C _d . The C _f i for each metal in a sample is summed to produce the C _d value for the sample. A metal with a high C _f i will have a larger influence on the index value of a sample than metals with lower C _f i values.

27 The enrichment and contamination indices described above provide an effective means to identify areas that have elevated metal concentrations with reference to background levels or recommended guidelines; however, they do not identify the sources of metal enrichment in soils nor identify a pollution source. Statistical methods that highlight associations between metal occurrences and concentrations can provide a better understanding of the underlying mechanisms responsible for geochemical patterns (Siegel 2002). Those metals that are statistically associated can be compared with the metal content of probable pollution sources.

28 Cluster analysis is a technique commonly used in environmental geochemical studies to group chemical elements (i.e., metals) according to common characteristics. In this study, hierarchical agglomerative cluster analysis grouped individual metals into successively larger clusters based on a measure (correlation coefficient) of their similarity across soil samples (Johnson and Wichern 2007). The decision to group metals followed Ward’s hierarchical clustering method (Swanson et al. 2001; Johnson and Wichern 2007; Templ et al. 2008). A dendrogram is used to graphically display the hierarchical cluster analysis results to visually assess the cohesiveness of the clusters.

29 Standardization of variable measurements is a common strategy in geochemical studies when dealing with widely different concentration ranges (Johnson and Wichern, 2007). The concentration of metals in this study varies by orders of magnitude and hence data standardization was performed using mean and variance values, according to the following equation (Skeppström and Olofsson 2006; Johnson and Wichern 2007):

29 where Z _i is the standardized metal value, X _i the raw metal value, √ _i the arithmetic mean for the metal, and σ _ii the variance of the metal values.

30 Finally, although cluster analysis does not require normally distributed data, it is recommended that heavily skewed data undergo transformation prior to analysis (Templ et al. 2008). A simple logarithm transformation was used to normalize each variable (Templ et al. 2008).

31 The geometric mean metal concentrations for subsurface soil samples from urban locations in St. John’s (n = 22; 0.25– 1.9 m depth range) are at least double the concentration for 5 of 10 metals (As, Cd, Cu, Pb, and Zn) relative to surrounding rural locations (Fig. 3; n = 29; 0.25–1.3 m depth range; Table 4). Samples from sites overlying the Signal Hill Group have the highest number of elevated metals (n = 9) compared with sites from above the St. John’s Group, with Pb and Zn having between 1.5 and 2.5 times higher geometric mean values (Table 4). Based on these results, together with evidence of human disturbance at some urban subsurface sampling sites (Campbell 2008), metal concentrations from urban subsurface soils were not chosen to represent naturally occurring background values in this study. In contrast, rural subsurface samples generated geometric mean values for metal concentrations similar to regional till values (Table 4), and are considered to be broadly representative of background levels in the St. John’s region; consequently, the rural subsurface samples were used in calculations of metal enrichment factors for urban sites.

32 Between 3 and 17% of rural subsurface soil samples exceed CCME/MOEE soil quality guidelines for 5 metals: As, Ba, Cr, Ni, and V (Tables 1 and 5). The human health guidelines for As, Cr and V are exceeded in 7%, 3% and 10% of samples, respectively, whereas between 7 and 17% of samples exceed the environmental health guidelines for As, Ba, Cr and Ni (V does not exceed the environmental health guideline in any samples). These exceedances occur exclusively in samples from above the Conception and Signal Hill groups.

33 Comparison of the geometric mean of urban surface (Table 6) and rural subsurface (background; Table 5) soil geochemistry indicates elevated concentrations for all 10 metals in surface soils in St. John’s, with particularly notable differences for Cd (×5), Cu (×3), Pb (×8.6), and Zn (×3.7). Predictably, all of these metals except Cd (<1%) have significant percentages (30–45%) of samples that exceed CCME/MOEE environmental health guidelines. Similarly, those metals that had concentrations above enviornmental health guidelines (7–17%) in background samples (i.e., As, Ba, Cr, and Ni) also had significant guideline exceedances in surface soil samples (13–32%), except Ni (2%; Tables 5 and 6). Although the geometric mean concentration for Co is elevated by only 20% in urban surface samples compared with background, 6% of St. John’s soil samples exceed the envionmental health guidelines for Co and <1% exceed the human health guidelines. Only three metals – As, Pb, and V – have concentrations that exceed the human health guidelines to any significant degree (<1% of samples from other metals exceeded guidelines) in St. John’s soils; notably, however, the percentage of samples with exceedances were 34%, 47%, and 47%, respectively.

34 The geometric mean C _fi for each metal ranged from 0.0 (Cd, Co, and Ni) to 2.4 (Pb; Table 7). Overall, Pb dominates the C _d for most soil samples but there are significant contributions from As, Cu, and Zn. Contamination index categories were chosen based on the standard deviation (σ) around the mean C _d value (<0.5, 0.5–1.5, 1.5–2.5 and >2.5); the category ranges are 0–6.5, 6.5–14.5, 14.5–22.5, 22.5–105.5. Samples with large standard deviation represent more heavily contaminated soil relative to the other samples. The spatial pattern of C _d plotted by sample location shows that the highest index values are roughly concentrated in the downtown; that is, the urban area marked as developed by 1950 on Fig. 2. In general, this area contains most of the high index samples that comprise the two highest index categories (>14.5). Outside this downtown core, C _d values are generally less than 6.5 or within 0.5σ of the mean (Fig. 2).

35 Scandium was used as the reference element to calculate EF values in this study. There was little or no difference between the geometric mean Sc concentrations in rural subsur-face (background) samples (11.6 ppm; n = 29) and the surface soils of St. John’s (11.0 ppm; n = 997), supporting the choice of Sc as the reference element. The geometric mean for the EF varies from 0.7 (Ni) to 8.5 (Pb), with values for 7 of 10 metals below 3 (Table 8). According to the classification categories of Sutherland (2000) (Table 3), Pb and Cd occur near the lower boundary of the significant enrichment (EF 5–20) category with geometric mean EF values of 8.5 and 5.3, respectively (Table 8). Chromium, Cu and Zn are on average only moderately enriched (geometric mean EF 2-5), whereas the other metals (As, Ba, Co, Ni, and V) are locally deficient to minimally enriched (geometric mean EF <2).

36 The spatial pattern of Pb enrichment factors shows a high concentration of enriched sample sites in the downtown core (Fig. 4), largely mimicking the pattern for contamination index (Fig. 2). Many of these downtown samples are classified as very highly to extremely highly enriched (EF >20) and represent 25% of all soil samples in the study (Table 9, Fig. 4). Large parts of the city outside of the downtown core have moderately enriched sites (EF 2–5). In comparison, the spatial pattern of the C _d enrichment factors also has its highest values in the downtown core, but in general there is a much more even distribution of EF categories across the city with 75% of samples either moderately or significantly enriched (Table 9, Fig. 5).

37 A closer examination of geometric mean EF values for each metal by sample category reveals a spatial pattern of enrichment at the residential property level (Fig. 6). Lead, Cd and Zn have highest enrichment at dripline sites followed by ambient and roadside sites. Chromium and Cu are slightly more enriched at roadside sites followed by dripline and ambient sites. In contrast, As is slightly more enriched in ambient and dripline sites. The remainder of the metals have low EF values that are more or less equally divided among the site categories.

38 Prior to cluster analysis and following logarithmic transformation, Co, Cr, Ni, and V concentrations were symmetrically distributed; As, Ba, Cd, and Cu were only weakly skewed, and Pb and Zn remained strongly skewed. During the hierarchical clustering procedure Pb and Zn and Cr and Ni merged to form two distinct clusters representing a high level of similarity across the samples (Fig. 7). At slightly lower similarity levels, Cu joined the Cr-Ni cluster, and Cd the Pb-Zn cluster to form the two strongest and most distinct metal associations among the group of ten. At lower similarity levels, As and Co form a single cluster that subsequently merges with a grouping that combines the Cr-Cu-Ni cluster with V. (Fig. 7). Barium concentrations have the weakest association with the other metals across all the samples.

39 Residential soils in St. John’s have elevated concentrations above background levels (rural subsurface) for all ten metals with CCME/MOEE soil quality guidelines. Three of the four metals that are the most elevated (Pb>>Cd>Zn>Cu; Table 8) are also among the four metals that contribute most strongly to the contamination index (Pb>>Zn>As>Cu; Table 7). Lead, Cd, Zn, and Cu also have the highest enrichment factors with 60, 53, 30 and 20%, respectively, of surface soil samples in St. John’s classified as significantly to extremely highly enriched. For the most part, soils contaminated with these metals are concentrated in the oldest parts of the city, around the downtown core, where many properties predate the 1930s, though others may be several decades younger. Lead, Zn and Cd are most highly enriched at dripline sites on residential properties, though ambient and roadside sites also have moderate enrichment (Fig. 6).

40 The results of the cluster analysis suggest that two distinct groups of metals are commonly associated together in surface soil samples (Fig. 7). One group that includes Pb, Zn and Cd is commonly found in soils at dripline sites (Fig. 6), especially in areas of the downtown developed prior to the 1950s (Fig. 4), where weathered paint is believed to be a major contaminant source (Bell et al. 2010). The age of these properties corresponds to the period when Pb concentrations were highest in paint, up to 50% by weight (CMHC 2007). Paint may also contain extremely elevated Zn concentrations (Mielke et al. 2000), and Cd is added to paint as a pigment component (CCME 1999b).

41 Lead, Zn, and Cd are also significantly enriched in roadside sites in St. John’s. Similar patterns were also documented for New Orleans, where vehicular traffic, both past and present, is considered to be an important metal source for adjacent soils (Mielke et al. 2000). Lead was used in Canadian gasoline up until the 1980s, Zn is a major component of rubber tires, and Cd is both a trace contaminant of the Zn that is used in tires and part of the alloys used to harden engine parts (Mielke et al. 2000). All three metals also contribute to enriched ambient sites in St. John’s, where coal combustion products and ash dumping are believed to be major sources of soil Pb at sites distal to roads and walls on high-density downtown properties (Bell 2003; Campbell 2008). Coal was a major heating fuel in St. John’s up until the 1960s when oil became more popular (Poole and Cuff 1994). From 1838 to 1948 St. John’s residents burnt almost 80% Cape Breton Island coal, the remainder from the United Kingdom and the United States of America (Newfoundland Customs Returns for the Years Ended 1835–1948). Coal from Cape Breton Island has elevated Pb and Zn compared to UK coal (Table 10).

42 For the second group of clustered metals – Cr, Ni and Cu – Cr and Cu are moderately enriched in residential soils and Ni is minimally enriched. Chromium and Cu are slightly more enriched in roadside samples compared to dripline and ambient sites. All three metals are components of a wide range of alloys that are associated with vehicles; however, their pervasive presence at all residential sites may also in part be associated with their common occurrence in paint and their naturally occurring background concentrations (Tables 4–6).

43 Arsenic does not display strong associations with other metals in the cluster analysis (Fig. 7), which is surprising particularly for ambient sites where Cape Breton Island coal is believed to be a significant source. For instance, Cape Breton coal may contain elevated levels of As, Pb and Zn (Table 10) but only the latter two metals are closely associated in the cluster analysis.

44 Only three metals – As, Pb and V – have concentrations that exceed the human health guidelines in greater than 1% of soil samples taken in St. John’s; however, between one-third and one-half of all soil samples collected in this study exceed environmental health guidelines for these metals (Table 6). Downtown areas of St. John’s have soil-Pb levels well above those from Canadian cities of equal or greater population size (e.g., Ottawa, Ontario, Rasmussen et al. 2001; Victoria, British Columbia, Bowman and Bobrowsky 2003) and closer to much larger North American cities (e.g., New Orleans, Louisiana; Mielke et al. 2000). Although almost half the residential soil sampled in St. John’s has Pb concentrations exceeding the human health guideline of 140 ppm, of greater concern are the exceptionally high concentrations (>1200 ppm) that tend to occur along the dripline of houses built before the 1960s and anywhere on residential properties that date to the 1920s or earlier (Bell et al. 2010). Risk assessment predictions suggest that while the wider St. John’s community may not be at risk of adverse health effects, there may be an increased risk for children living in pre-1960s housing that for the most part is concentrated downtown and in surrounding neighbourhoods (Campbell 2008). It is estimated, for instance, that 8 to 21% of children living in housing built before 1949 in St. John’s may exceed the current recommended blood-Pb level guideline of 10 g/dL (Campbell 2008). Young children are more susceptible to Pb exposure because of hand-to-mouth behaviour and their developing nervous systems, while the neuro-developmental effects of increased body-Pb burden in children are linked to decreased IQ scores and other intelligence and developmental deficits (ATSDR 2007). Because of this public health concern, a comprehensive biomonitoring study is underway in St. John’s to determine the extent of Pb exposure in young children across a range of housing age categories (www.LeadNL.ca).

45 Although As is on average in the deficient to minimally enriched EF category, 34% of all soil samples exceeded the human health guidelines for As in residential soil, mostly concentrated in the downtown core in areas developed prior to 1961 (Table 6, Fig. 8). The human health guideline for As is 12 ppm, 2.4 ppm higher than the geometric mean concentration in surface soils of St. John’s (Table 6) and 5.2 ppm above background levels (Table 5). Till samples from the Avalon Peninsula show that As is naturally elevated (7% of background samples exceed the human health guideline), especially overlying the St. John’s Group, with a geometric mean of 6.9 ppm (maximum value = 246 ppm); (Table 2). On average, 2.3 ppm of As has been added to the soils of St. John’s through anthropogenic processes, based on Equation 3 above, potentially from past coal combustion and ash disposal. Because the mobility and bioavailability of As in soil depends on its form, and the physical and chemical characteristics of the soil (Liu and Cai 2007), more research is needed to assess the exposure risk and potential health impacts of these elevated As levels on pre-1960s properties.

46 The human health guideline for V of 86 ppm is based on the MOEE soil guideline. MOEE guidelines are calculated based on high frequency, high intensity human exposure (both soil ingestion and dermal contact, and including children and pregnant woman) to soil on a variety of non-industrial land uses from residential to agricultural (MOEE 2009). In Ontario, however, where the background soil concentration (86 ppm) of V is above the calculated guideline (39 ppm), the human health guideline defaults to this higher value (86 ppm). The geometric mean for V in rural subsurface soil in St. John’s varies between 67 and 96 ppm, depending on the underlying bedrock type (Table 4). Because the geometric mean V concentration in urban surface soils is 84 ppm, 47% of the soil in St. John’s exceeds the MOEE human health guideline for V.

47 Vanadium is added to special steels (e.g., titanium alloys) and is used as a catalyst in the chemical industry (CCME 1999b), but neither is expected to be a major source to the soil of St. John’s. Food contains low amounts of V, but it represents the major exposure source for the general population. The toxicity of V compounds, however, is low with most toxic effects causing irritation of the eyes and the upper respiratory tract (Barceloux 1999). These effects are mostly observed in people exposed to V in work environments (Barceloux 1999).

48 The remaining metals that have very low (<1%) exceedances of the human health guidelines include Cd, Co, Cr, Cu, and Zn (Table 6). Although enrichment values suggest a significant input of Cd into St. John’s soils, the fact that natural levels of Cd in the soil are so low (geometric mean = 0.06 ppm; Table 5) means that significant enrichment can occur without exceeding the human health guideline of 14 ppm (Table 1). In the case of Zn, and to a lesser degree Cu, the human health guidelines are very high – 5600 and 1100 ppm, respectively – compared with the geometric mean values for St. John’s soils (198 ppm and 50 ppm, respectively) and hence soil is not considered an exposure risk for city residents. Both these metals are essential micronutrients required for optimal functioning of biological processes and organs in humans (Siegel 2002).

49 Five of the ten metals (i.e., As, Ba, Cu, Pb, and Zn) have concentrations exceeding their environmental health guidelines in 20% or more of the samples. In some cases, these metals are essential for both plants and animals (e.g., Cu, Zn; CCME 1999c, e; Bradl 2005), whereas others are considered non-essential (e.g., Pb; CCME 1999d; Bradl 2005) or phytotoxic at relatively low concentrations (e.g., As, as low as 10 ppm; Bradl 2005). The toxicity of these metals in plants is mainly dependent on chemical form, soil properties and environmental conditions (e.g., plant sensitivity to Ba is heightened in the absence of Ca; CCME 1999a) and has largely been measured in the context of agricultural crop yields (Table 11). For example, at soil-Ba concentrations of 500 ppm and 2000 ppm, barley yields decline by 38% and 63%, respectively (Chaudry et al. 1977) and 50% reduction in seed yield of turnips is documented in soil with Zn levels of 50 ppm or more (Sheppard et al. 1993; CCME 1999e). Chlorosis, an abnormally yellow color of plant tissues resulting from partial failure to develop chlorophyll, is an indication of Zn toxicity and may result in depressed plant growth; it occurs at soil concentrations >100 ppm, well below the median value of 178 ppm for St. John’s soils (Table 6, CCME 1999e; Bradl 2005).

50 Lead poses a more significant toxic threat to mammals and birds than plants and exposure occurs through ingestion of soil when birds and herbivores feed (CCME 1999d). Lead is a nonspecific toxin to wildlife and livestock, which inhibits many enzymatic activities and may adversely affect the haematological and central nervous systems and reproduction function (Bradl 2005). In contrast, adverse effects of Zn on wildlife and livestock is less of a concern than on plants but have been observed in sheep at soil concentrations of 750 ppm (Campbell and Mills 1979; CCME 1999f).

51 Although there is limited commercial gardening in downtown areas of St. John’s, a recent survey suggested that roughly 10–15% of homeowners consume garden produce from their own properties (Campbell 2008). Although reduced crop yield and quality may be a concern for some gardeners, the ingestion of metals through locally grown produce and the increased bare soil exposure on residential properties due to phytotoxic effects on plant cover are a potential public health concern for city residents, especially in neighbourhoods where metals exceed environmental health guidelines. The bio-concentration of metals in edible plants would heighten this concern but actual uptake and location of accumulation in different plants and soil concentrations are poorly understood. For instance, plants growing on As-contaminated soil rarely contain greater concentrations than their substrate (i.e., low bio-concentration; CCME 2001), whereas a mean bio-concentration factor of 2.65 was calculated for Cd in different plants and plant tissues (CCME 1996b). A more detailed eco-toxicological assessment is required to properly assess the environmental health impacts of these metals in St. John’s soils.