3.1. Selected Physical and Chemical Soil Properties
The selected physical and chemical characteristics of the examined soils, i.e., the soil texture, the soil pH in an aqueous solution (pH
H2O), the potential acidity (pH
KCl), and the amount of organic carbon (OC) and nitrogen (N), are presented in
Table 1.
Soil texture is a very important factor since a whole range of physical (structure, porosity, water–air, heat) and chemical (adsorption and buffering capacity) properties depend on it [
4]. The studied soils can be classified mainly as clay loam and sandy clay loam, according to the USDA classification [
36]. The lowest total sand content (38.70%) and the highest clay content (38.43%) was measured at the Belgrade II sampling site, while the highest total sand content (53.01%) and the lowest clay content (25.43%) was measured in Smederevo. The silt fraction varied from 21.57% in Smederevo to 27.70% in Obrenovac. Moghtaderi et al. [
53] found urban soils in Shiraz (Southwest Iran) to have a more marked heterogeneity in terms of texture, with the sand content ranging from 21.9–72.6%, silt from 21.57–27.70%, and clay from 7.6–33.5%. It is well known that a fine-grained soil fraction exhibits a higher tendency for PTE adsorption than coarse-grained soils since it contains soil particles with large surface areas such as clay minerals, iron, and manganese oxy-hydroxides, and humic acids [
54]. Based on a specifically built triangle for soil classification according to texture, the dominant soil type at the Pancevo, Obrenovac, and Belgrade II sampling sites was clay loam, while sandy clay loam was identified in Smederevo and at Belgrade I. Dragović et al. [
31] also classified the soils around the ironworks in Smederevo as silty clay loam and silty loam. The pH
H2O of the soil solution fell within a narrow range—between 7.92 at the Belgrade II sampling site and 8.62 at Belgrade I, which classifies them as moderately alkaline [
36]. The pH
KCl solution was relatively uniformly distributed and ranged from 6.85 in Pancevo to 7.10 in Belgrade, classifying these soils as practically neutral [
36]. Soil pH controls the geochemical behaviour of those elements, which are present in the solid and soluble phases of the soil and directly affects the processes of sorption/desorption, uptake and chemical speciation of the elements present in the soil [
55,
56]. Soil alkalinity is a common occurrence in urban soils and is most often the result of using alkalising products such as calcium carbonate and calcium–magnesium carbonate in gravel, cement, and concrete, but is also attributed to the deposition of alkaline atmospheric particles and carbon of anthropogenic origin, from coal combustion [
28,
57,
58]. Rarely, urban soils can also be acid soils since the nature of the geological parent material has a very important role [
26]. High alkalinity is a negative feature of a substrate and often causes a deficiency of essential micronutrients [
28], but it can also have a positive effect in terms of the immobilisation of potentially toxic, labile forms of elements that form permanent complexes with organic matter in the soil [
4,
58].
Organic carbon and nitrogen in soil play a key role in many biogeochemical and pedogenetic processes, as they impact its fertility and have a direct effect on soil-plant interactions [
59,
60]. The origin of OC in soil is twofold: it occurs naturally through the decomposition of organic matter and anthropogenically from various human activities [
4,
61]. The anthropogenic inflow of OC cannot compensate for or replace the natural inflow from the mineralisation of organic matter because in this way the chemical structure of the soil changes [
61]. In this study, the highest OC content was measured at the Belgrade I sampling site (4.05%), which is located near very busy roads in one of the most polluted parts of Belgrade, and can be linked to its anthropogenic inflow. The OC content in the parks in other cities and municipalities was significantly lower and fell within a narrow range—from 2.05% in Obrenovac to 2.80% in Pancevo. Moghtaderi et al. [
53] found high variability in OC levels in urban soils in Shiraz (Southwest Iran), where its content ranged from 0.27–5.5%. Nitrogen in soil originates from the mineralisation of organic matter and its amount depends directly on the inflow of organic matter, which is why soils with a higher percentage of OC have a higher content of nitrogen compounds [
60]. This is confirmed by the results of this study, where the highest percentage of N was measured at the sampling site with the highest content of OC—Belgrade I (0.31%).
3.2. Pseudo-Total Content of Selected PTEs in Examined Urban Soils
Basic statistics and differences between the content of PTEs in the examined soils are shown in
Table 2 and
Table S6. In order to compare the content of PTEs in the examined soils with their content in the upper continental crust, and with the threshold values of PTEs, as set by European and national legislation, the following were used: the composition of the continental crust [
62], Finnish legislation on contaminated soils [
63], and regulations on allowable concentrations of hazardous and harmful substances in the soil [
64]. After examining the options presented by the various approaches and the thresholds they apply, the authors chose the standards set by Finnish legislation on contaminated soils. The Finnish standard values are a good approximation of the mean values of different national systems in Europe [
65].
As concentrations of As and Cd were below the detection level in all of the analysed samples, they were not discussed any further. On the basis of the obtained results, it can be concluded that PTE content at all the investigated sites was below levels in the upper continental crust [
62], threshold values for European soils [
63], and maximum allowable concentrations of PTEs as set by national legislation [
64,
66], except for Ni and Pb (
Table 2). The highest levels of Co, Fe, and Mn were measured in soil samples from Obrenovac, Cr and Ni in Smederevo, and Cu, Pb, Sr, and Zn in Belgrade. Cobalt content at Belgrade II (10.75 mg kg
−1) and Obrenovac (11.75 mg kg
−1) exceeded standard levels of 9 mg kg
−1 set by national legislation [
66]. The content of Co in Obrenovac differs significantly in comparison to all the other sampling sites, with the exception of Belgrade II. However, the lowest level, measured at Belgrade I, is statistically different only from those concentrations measured in Obrenovac and at Belgrade II (
Table 2). The average Cr, Fe, Sr, and Zn levels in the selected soil samples were within the proposed limits according to European and national legislation [
62,
63,
64]. Somewhat elevated Cu concentrations were measured in samples from Smederevo and Belgrade I, which was statistically confirmed (
Table 2), but these were still at a level that does not pose a danger to the environment, given the alkalinity of the soil and slightly higher OC content at these sites [
67]. The average Mn content in the examined soils was higher than its average content in the surface soils of Europe (382 mg kg
−1, ref. [
68]). The content of Ni in soils varies considerably and depends primarily on its content in the parent rock [
69]. In the examined soils, its average content varied from 44.60 mg kg
−1 at the Belgrade I sampling site to 104.05 mg kg
−1 in Smederevo. Statistical analysis showed that Ni content in Smederevo differed significantly in comparison to the other sampling sites. Such results for Ni content in soil are expected, given that numerous studies have confirmed that higher concentrations of Ni are characteristic for the soils of Serbia [
14,
69,
70,
71]. During research spanning several decades, these authors have proven that the main source of Ni in the soil is actually the parent rock, i.e., the minerals it is composed of [
69]. Čakmak et al. [
14] believe that the elevated Ni levels are linked to areas with intensive soil erosion, which occurs due to heavy rainfall, as a result of which it is washed into the basins of larger rivers (the Kolubara and the Morava) and then transported by them [
70,
72].
The results of one-way ANOVA showed that there was a statistically significant difference (p < 0.001, p < 0.01, p < 0.05) in the content of all PTEs at the examined sites, apart from Pb (ns). The lack of significance for Pb occurs due to the high heterogeneity of its content in the examined soils.
A comparison of PTE content in urban soils in some Serbian cities and European urban soils is shown in
Table 3.
When comparing results of PTEs content obtained in this research with results from our previous research [
4,
73], levels of Mn were found to be similar, but Cr, Cu, Ni, Pb, and Zn concentrations were higher. On the other hand, Cr, Cu, Pb, and Zn content in soils in Novi Sad [
74], Lisbon [
79], and Thessaloniki [
73] was similar, but Ni was somewhat lower than the levels measured in this research (
Table 2). Similarities in terms of PTE content were also found in soils from Šabac [
75] and Kragujevac [
76], where concentrations of most of the examined elements were similar to those determined in this research, with the exception of the higher content of Cr in Kragujevac and Mn in Šabac. Soils in Warsaw [
82] and Prague [
81] are characterised by a lower content of Co, Cr, Cu, Mn, Ni, and Pb than in the examined soils in cities in Serbia, while soils in Turin [
83] and Zagreb [
77] have a higher content. In addition, soils in Zagreb [
77], Turin [
83] and Vigo City [
26], are characterised by a higher Zn content than soils in other European cities, and soils in Sopron [
78] by a higher Cu and Pb content. In terms of Sr, its content in the examined soils was lower than in soils in Zagreb [
77], but higher than in Vigo City [
26]. Based on these observations, it can be concluded that urban soils exhibit exceptional heterogeneity in terms of the presence and content of PTEs. This is to be expected, given the fact that urban soils have undergone significant changes and contain varying amounts of different additional materials, such as those used in the construction of buildings or roads, and also the fact that the surface layer (0–20 cm), which is most often the focus of research, is continuously exposed to intensive pollutants from a range of sources.
3.3. Partitioning of PTEs in Examined Urban Soils
Generally, Co content in European soils does not reach levels that could pose a risk to the environment [
84]. Despite the fact that Co in soils originates mainly from parent materials, it can also come from anthropogenic sources, above all from mining and smelting activity, fertiliser use and sewage sludge spreading [
85], which can apply to the samples from Obrenovac. However, the absence of Co extracted in the acid-soluble/exchangeable fraction in the samples from Obrenovac, as well as a significant amount bound to Fe and Mn oxides, points unequivocally to its geological origin, especially if we bear in mind the fact that Co in soil is closely correlated with elements of geological origin [
86], as confirmed by both correlation and PCA in this study: Co-Fe (0.828 **) and Co-Mn (0.810 **). Furthermore, in the alkaline conditions found at all the sampling sites, Co would be expected to immobilise. However, the fractionation profile of Co reveals a similar share in its distribution between the reducible (35–60%) and residual fractions (35–58%), while only a small proportion was bound to organic matter and sulphides (1–12%). A reduction in soil Eh and pH can lead to the solubilisation of precipitated or adsorbed Co [
87].
Chromium can be present in soil in different oxidation states, the most important from the environmental point of view being Cr (VI) and Cr (III), as the most stable and dominant forms of Cr in nature, and its speciation depends on redox potential, soil pH, the presence of potential electron donors and Cr adsorption by colloids [
13,
56,
88]. It is believed that most soil Cr occurs as Cr (III), and is incorporated within the mineral structures. Chromium (III) and its compounds are considered to be very stable and are mainly found bound to organic matter in soils; at pH 5.5 it is almost completely precipitated [
88,
89]. Sequential analysis confirmed that Cr was stable and firmly bound to the crystal structure of the mineral, with most extracted in the residual fraction (68.63–77.70%), while the remainder was bound to organic matter and sulphides (15.71–18.67%) and Fe and Mn oxides (4.33–14.84%). The slightly higher percentage share in the reducible (14.84%) and oxidisable (16.48%) fractions in Smederevo probably represents the combined influence of the geological substrate and the way park surfaces are formed. These results are in line with the results of Wang et al. [
15], who found that in geologic serpentine soils Cr is mainly found in the oxidisable and residual fractions (over 90%), while this share is somewhat lower in anthropogenically contaminated soils. As Cr in soil is mainly present in the insoluble Cr (III) form, it is very immobile, which, together with the alkaline reaction of soil, means the risk to humans is low [
15]. This is also important from the environmental point of view because Cr is a potentially toxic element, yet in this form, it does not pose a serious ecological risk [
6,
56].
Sequential extraction results for Cu indicated a possible anthropogenic impact in Smederevo and at Belgrade I, where approximately 50% of Cu was extracted in the first three fractions, while the rest was bound to primary and secondary soil minerals. At the other sampling sites, Cu was mainly bound to soil minerals (60.68–70.12%), while the rest was bound to organic matter and sulphides (16.31–18.79%) and Fe and Mn oxides (9.47–16.08%), with only a small proportion (3.89–4.43%) extracted in the exchangeable and acid-soluble fraction. It is known that Cu shows high binding affinity for various soil components, particularly organic matter; in addition, the stability and speciation of Cu depend greatly on pH and its mobility decreases with increasing pH [
67]. Thus, it can be concluded that Cu in the examined soils is stable and poorly mobile, while the samples from Smederevo and Belgrade I exhibited slightly higher mobility and bioavailability, with a certain amount of Cu coming from anthropogenic sources (traffic and ironworks). Prolonged exposure to high Cu levels damages the liver, kidneys and brain, which can lead to hepatitis, coronary heart disease, and psychological and neurological symptoms [
87]. Copper, together with Pb and Zn, is a pollutant that is associated with the wear and tear of tyres and brake pads and engine oil decomposition [
35], which is confirmed by the significant positive correlation of Cu-Pb (0.796 **), Cu-Zn (0.841 **), and Pb-Zn (0.750 **).
Given the fact that Fe is one of the most common elements in the lithosphere, its standard, threshold and target values are not defined by European and national legislation [
64,
66,
90]. Iron content in the examined soils ranged from 27.66 g kg
−1 at Belgrade I to 34.64 g kg
−1 in Smederevo. The dynamics of Fe in soil are closely linked to the cycling of O, S, and C, as well as to the Eh–pH system of the external environment and the oxidative phase of the Fe compounds that are present [
91]. In this sense, Fe is characterised by fast and easy changes in its oxidation state and its chemical behaviour is similar to Co and Ni [
87]. Lindsay and Norwell [
92] state that the solubility of inorganic Fe depends on soil pH, with it being lowest in the pH 6.5–8.0 range, which was also the case with the examined soils. In the existing conditions, the solubility and mobility of inorganic Fe is reduced to a minimum, as confirmed by results of sequential extraction, which revealed the strong binding of Fe to silicate and oxide soil minerals (87–92%), indicating the fact that Fe is of geological origin. This is supported by the results of sequential analysis, where the share of Fe in the reducible (3.01–7.54%) and oxidisable (4.25–5.16%) fractions was similar, while it was insignificant in the acid-soluble fraction (0.01–0.02%).
Manganese in urban soils mainly comes from lithogenic sources, but can also be a product of various anthropogenic activities, such as the smelting of metal ores, emissions during steel and iron alloy production, the combustion of fossil fuels, and, to a lesser extent, emissions during the combustion of fuel additives [
93,
94]. Specifically, in the exhaust gases from cars, the organic Mn compound methyl-cyclopentadienyl-manganese-tricarbonyl (MMT) can be found, which is one of the main alternatives to lead additives in gasoline [
35,
94]. In addition, ash from coal combustion in thermal power plants may contain significant amounts of this element [
95], which is probably the cause of its elevated levels in Obrenovac, where the dominant source of pollution is a thermal power plant. Manganese is an essential element, but chronic exposure to Mn can lead to serious disorders, which is why it is considered extremely toxic to human health, and it can have a negative impact on the ecosystem due to its accumulation in the food chain [
95]. Despite this, the positive impact of Mn can be seen in its oxides due to their potential to sorb heavy metals in the soil and to act as oxidants, thus positively affecting soil vitality [
96]. Sequential extraction showed that the largest share of Mn was extracted in the reducible fraction (46.42–63.13%), followed by the exchangeable, acid-soluble (14.89–27.86%), and residual (17.50–23.60%) fractions, while the least was bound to organic matter and sulphides (4.16–5.34%). Such a fractionation profile is expected since Mn deposited in soil tends to be bound to oxides, carbonates and silicates [
35], i.e., in the soil it is mainly in the form of Mn (II), which is its water-soluble and exchangeable form, and in the form of insoluble Mn oxides (Mn (IV)) [
96,
97]. The primary targets of hydroxylamine hydrochloride are specifically Mn oxides and hydroxides [
18]. Keeping in mind the results of the correlation and PCA (
Figure 3), the alkalinity of the examined soils, and the fractionation profile, it can be concluded that Mn in the tested soils is stable and does not pose a danger to the environment. However, any change in pH and redox potential can increase its availability, which would adversely affect the environment.
Elevated Ni concentrations require a more cautious approach given its potential to promote cancer mortality in humans, with this risk depending on its mobility and bioavailability in soil, as only bioavailable fractions can be absorbed by humans [
15]. Sequential analysis of Ni showed that, at the sites in Obrenovac and Belgrade, it is predominantly incorporated in the crystal structure of the mineral (52.74–57.52%), and as such, does not pose a health risk. Of the non-residual fractions, especially in Pancevo and Smederevo, Ni was primarily held in the reducible fraction (the Fe-Mn oxide fraction) (43.02–45.62%). Antić-Mladenović et al. [
71] point out that Fe-Mn (hydro) oxides act as scavengers for soil Ni and that a fractionation profile such as this one represents a characteristic pattern for the lithogenic origin of metals in soils [
98]. The high proportion of Ni in these two fractions can also be linked to Ni-bearing clay minerals, which can be formed during ultramafic rock weathering and can be further transported by rivers [
14,
99]. The smallest part of Ni was extracted within the exchangeable fraction (4.39–6.92%), which implies that under current native conditions, Ni is stable. Fe-Mn oxides are thermodynamically stable in alkaline conditions [
71], which were encountered in our study; however, any change in redox potential, pH, salinity, microbial activity or hypoxic conditions can lead to the release of soluble Ni [
100]. For this reason, most authors do not exclude the possibility of anthropogenic contamination of soil with Ni, along with its natural enrichment. In our study, the geological origin of Ni was also confirmed by correlation analysis whereby strong positive correlations with Co (0.349 *) and Cr (0.954 **) were found, elements for which other researchers also determined significant positive correlations and a predominantly geological origin [
14].
Numerous previous studies on Pb content in urban soils have indicated variations in Pb content depending on the position of the sampling site, with the highest concentrations determined in intensive traffic zones [
35,
101]. Lead is a common pollutant of urban soils and its origin is most often associated with anthropogenic activities, i.e., emissions from traffic and industry, with the anthropogenic impact on soils in cities being more marked due to the higher population density, traffic intensity and proximity to industrial plants. The extremely high content of Pb at Belgrade I is probably the result of past emissions from motor vehicles. Specifically, although the practice of deliberately adding Pb to fuel was halted worldwide long ago [
35], leaded gasoline was produced in Serbia until 2010 [
20,
102] and traces of it can still be found in fuel [
35]. In addition, the low mobility and high adsorption capacity of Pb for different soil components hinders a reduction in its concentrations in the surface layer of soil yet further [
35,
101]. In Smederevo, Pb content (99.11 mg kg
−1) was at the very top end of maximum permitted concentrations prescribed by national legislation (100 mg kg
−1, ref. [
64]), pointing to the obvious impact of emissions from the ironworks, but also from traffic. The results of sequential extraction confirmed the anthropogenic origin of Pb, given that in the soils of Smederevo and Belgrade Pb was predominantly bound to Fe and Mn oxides and hydroxides, i.e., that more than 75% was extracted in the first three fractions, and a significantly smaller part was bound to soil minerals. A fractionation profile such as this indicates that Pb can be easily mobilised if external conditions change, which is a risk to the environment, especially if we bear in mind the fact that ingestion of soil and dust is one of the important pathways of Pb exposure, especially in children [
84,
101]. Chronic effects that occur due to the long-term intake of lead into the body include neurological and gastrointestinal problems, anaemia, damage to the kidneys, endocrine, and immune systems, and disorders in the psychophysical development of children [
1,
7]. At the other sampling sites, Pb was approximately equally distributed between the residual and reducible fractions, with a significant share being extracted at these sites in the first three phases of extraction (>60%), indicating the presence of anthropogenic sources at these sites as well, although their impact is somewhat less pronounced (
Figure S3, provided in
Supplementary Material). The results of the correlation analysis support this assertion regarding the origin of Pb, i.e., there was a significant positive correlation between Cu, Pb and Zn, which are elements that originate from traffic, with Pb-Cu (0.796 **) and Pb-Zn (0.750 **).
Strontium is an alkaline earth element, which can be found in the environment in a wide range of concentrations as a result of natural mineral degradation or the constant impact of anthropogenic pollution [
103,
104]. Due to its marked reactivity, it does not occur in nature in its elemental form, but most often as a mineral, carbonate (strontianite), or sulphate (Celestine). Its natural content in soil depends almost exclusively on the climatic conditions and the composition of the parent rock, while the average content ranges from 175–458 mg kg
−1 [
105]. Strontium is found in small quantities in association with other alkaline earth metals that; have similar properties to Ca, which is why it usually competes with them. From the environmental point of view, this can be a significant problem because this similarity with Ca conditions its fast uptake; it is easy transport to aboveground plant parts, and its involvement in the cycling of matter [
89]. Once it enters the body, Sr easily gets into the bloodstream, mimicking and behaving very much like Ca. It can easily be incorporated into bones, especially in children, where it can create hard bone mineral and lead to problems with bone growth or the weakening of bones. The removal of Sr from the body is a very long and, slow process [
106]. Strontium is known for its mobility and is derived mainly from leachable minerals such as feldspars, apatite, and micas [
68]; hence, why it causes concern in terms of ecosystems [
104]. Existing data from literature indicates that the majority of Sr is retained in the surface soil layer [
20,
107] and is, therefore, mainly associated with the exchangeable fraction in soil, which was confirmed by our study (49.32–76.75%). The second largest fraction is related to Fe and Mn oxides (12.77–25.87%), which indicates the high mobility of Sr in the examined soils and its potential toxicity. The share of Sr in the residual fraction was between 8.48% and 21.32%, while it was least associated with organic matter and sulphides (1.99–4.69%). These results concur with those of Stefanović et al. [
17], who determined that Sr is one of the most easily mobilised elements. Data obtained from correlation and PCA showed that the origin of Sr was different from that of all the other examined elements. Strontium negatively correlated with elements of geological origin such as Co (−0.319 *), Fe (−0.493 **), and Mn (−0.539 **), which indicates its anthropogenic origin (traffic, individual furnaces, the use of various artificial substrates in parks).
The main source of Zn in soil is the parent substrate on which the soil is formed; however, in urban soils, in addition to the parent substrate, Zn can originate from various anthropogenic sources. Copper, Pb, and Zn in urban soils usually have a common origin and are often referred to as ‘urban metals’ [
108], as confirmed by research based on GIS-based and multivariate statistical analyses showing that urban soils are characterised by elevated concentrations of Cu, Pb, and Zn with the dominant pollution sources being traffic, the use of paints and various industrial emissions [
109,
110]. The influence of anthropogenic sources on Zn content, i.e., the impact of traffic and emissions of suspended particles from the ironworks, was determined in this study, with the highest content of this element measured in Smederevo (114.82 mg kg
−1) and at Belgrade I (135.33 mg kg
−1). Sequential extraction also confirmed the anthropogenic impact at these sites, with Zn in the samples from Smederevo and Belgrade I being mainly exchangeable and associated with carbonates (14.0–20.31%), and Mn and Fe oxides primary and secondary soil minerals (56.43–64.75%), followed by its binding to Mn and Fe oxides (24.30–29.16%), while only a small portion was extracted in the oxidisable (8.07–9.52%) and acid-soluble/exchangeable fractions (2.88–4.88%). The significant portion of Zn in the reducible fraction indicates the importance of Fe and Mn oxides for the accumulation of this element, which is confirmed by the similar fractionation profile of these two elements. Based on all the above, it can be concluded that the anthropogenic impact is most pronounced in Smederevo and at Belgrade I, and also that Cu, Pb, and Zn are of common origin.
3.5. Health Risk Assessment
Non-carcinogenic risk values through ingestion (HQ
ing) for children and adults decreased in the following order: Fe > Pb ≥ Co > Mn > Cr > Ni > Cu > Zn > Sr and were less than 1 for all elements, indicating that there was no significant non-carcinogenic risk. The exception is HQ
ing values for Pb at Belgrade I, where this element presents a significantly higher non-carcinogenic risk for children than the other tested PTEs (
Table 5). This result can be explained by the elevated Pb concentrations at this site, as well as by its high intoxication and respective RfD value [
40]. Similar high HQ
ing values for Pb were also obtained by Jia et al. [
7]. This result deserves special attention due to the fact that the amount of potentially available Pb at this site is greater than 75%. However, despite the fact that the HQ
ing values for each element were classified as ‘acceptable’, their cumulative values (CHQ
ing) for children were significantly greater than 1 at all sites, which indicates unequivocally that the investigated PTEs in the studied soils pose a significant non-carcinogenic risk through the ingestion exposure route. The highest CHQ
ing values for both children and adults were determined in soil samples from Belgrade I (
Table 5 and
Table 6). The values for the non-carcinogenic risk through dermal contact (HQ
der) decreased in the order of Mn ≥ Cr > Ni > Fe ≥ Pb ≥ Co > Cu > Zn > Sr for both children and adults, while values for the non-carcinogenic risk through inhalation (HQ
inh) followed the order: Mn > Co > Ni > Cr > Pb > Cu > Zn for both children and adults. Unlike CHQ
ing, values for both individual HQ
der and HQ
inh and for cumulative CHQ
der and CHQ
inh were less than 1, indicating that PTEs in soil pose no significant non-carcinogenic risk when it comes to the dermal and inhalation exposure routes. When considering the results for the non-carcinogenic risk for the three exposure pathways, it was observed that ingestion has the higher potential risk for children and adults, which is in accordance with previous research by Jia et al. [
7], Wu et al. [
113] and Jadoon et al. [
9]. Health index (HI) values for children and adults decreased in the order: Fe > Pb > Co > Mn > Cr > Ni > Cu > Zn > Sr. However, although the individual HI values for children and adults were less than 1, their total, cumulative values (CHI) for children were above the maximum acceptable values at all sampling sites, with the non-carcinogenic risk through ingestion being the main contributing factor (HQ
ing) at over 90%, which is in accordance with previous findings [
113,
114]. The highest CHI values were at Belgrade I. Based on these results, it is clear that residents are exposed to the greatest risk through ingestion and that of all the examined PTEs, Pb potentially represents the greatest risk. Adding to this risk is the fact that a significant portion of Pb in the examined soils is in a potentially available form (
Figure 2). Comparing the HI values for children and adults, it was noted that those for children were almost 10 times higher, which points to the greater sensitivity of children to the detrimental health effects of PTEs in soils compared to adults. Numerous previous studies have also highlighted the increased sensitivity of children to the effects of PTEs [
7,
10,
42,
113,
114], with their specific hygiene-related habits during play further increasing this impact [
2,
9].
The carcinogenic risk (CR) values for Co, Cr, Ni, and Pb and the cumulative carcinogenic risk (CCR) through ingestion, dermal contact, and inhalation pathways were in the ‘acceptable’ range (10
−4—10
−6, ref. [
50]), indicating that there is no danger of cancer,
Table 7. The CCR values for all three exposure pathways decreased in the following order: CCR
ingestion > CCR
dermal > CCR
inhalation, which is in accordance with the results of Varol et al. [
10]. CCR
ing accounts for the largest share of the total cumulative cancer risk (CTCR), being as much as 10 times higher than CCR
der and almost 100 times higher than CCR
inh. In terms of the results of the carcinogenic risk by sampling site, it was noticed that although the values were ‘acceptable’, the residents of Smederevo are at potentially the greatest risk of getting the some type of illness. It was further observed that Cr accounts for the largest share of CTCR by ingestion, dermal contact, and inhalation, which is in accordance with the findings of Varol et al. [
10]. However, sequential analysis showed that Cr in the examined soils is stable and firmly bound to the crystal structure of the mineral, which reduces its potentially negative effects.
3.6. Determining the Origin and Relationship between PTEs
Based on the correlation analysis results, the examined elements were divided into three groups. The first group consisted of Co, Fe, and Mn, which exhibited a significant positive correlation with each other (
Table 8), but also a significant negative correlation with Sr and Zn. The second group included Cu, Pb, and Zn, which were significantly positively correlated with each other, with only Zn being significantly negatively correlated with Co, Fe, and Mn. The third group consisted of positively correlated Cr and Ni, with Ni also positively correlated with Co. Strontium displayed behaviour that was opposite to the other examined elements and it could not be put into any of these groups, i.e., it exhibited a significant negative correlation with Co, Fe, and Mn. A similar relationship between the elements in urban soils has been found in five European cities with a significant positive correlation between Cu, Pb, and Zn, and a positive correlation between Cr and Ni [
108]. These authors revealed the anthropogenic origin of Pb, Zn, and Cu and the natural origin of Cr and Ni. Other studies have yielded similar results [
115,
116].
PCA results were in accordance with correlation analysis results, i.e., three components were singled out, which together explained 85.09% of the variability (
Table 9,
Figure 3a). The first component (41.92%) represented the influence of the geological substrate and exhibited a significant positive correlation with Co, Fe, and Mn, and a significant negative one with Sr. The second component (26.09%) represented the anthropogenic impact, i.e., the impact of traffic, given that the wear and tear of brake pads, tyres, and various parts of the car leads to the release of Cu, Pb, and Zn into the environment [
2,
35], and it is precisely these elements that Component II had the strongest influence on. The third component (17.08%) represented the combined influence of the geological substrate and the way park surfaces are formed, with Cr and Ni correlating with it. Elevated Cr and Ni levels is one of the specific features of alluvial soils [
2,
14], which were probably used for the filling and formation of park areas due to their wide availability in the investigated localities. Based on the results of correlation and PCA, it was established that Sr did not belong to any of the determined groups of elements and that this element is influenced by an additional factor that is not statistically significant (eigenvalues < 1), but which indicates that Sr is of different origin to the other examined elements, especially when compared to those elements of geological origin (Co, Fe, and Mn), with which it correlates negatively (
Table 8).
Spatially speaking, samples from Obrenovac, Pancevo, and Belgrade II stand out as being under the influence of the first component (PC1) in particular, while in Obrenovac the influence of PC3 is also present (
Figure 3b). Samples from Belgrade are strongly impacted by PC2, while samples from Smederevo are influenced by PC3 and PC2. Hence, it can be concluded that the anthropogenic influence is dominant at Belgrade I and, to some extent, in Smederevo, while in Pancevo, Obrenovac and at Belgrade II the dominant impact is geological. The way parks are formed impacts the samples from Smederevo and partly from Obrenovac, too. Park soils are formed for that specific purpose by using substrates of different origins and composition; in addition, they are under the constant influence of anthropogenic factors, which makes them very heterogeneous and explains the differences that existed between samples collected from the same site.