Assessment of Heavy Metal Concentrations in Urban Soil of Novi Sad: Correlation Analysis and Leaching Potential

Abstract

Soil samples from the urban area of Novi Sad were analyzed to determine the total concentrations of heavy metals including Cr, Pb, Cu, Zn, As, Mn, Ni, Co, Cd and Fe. In addition, leaching tests according to CEN 12457-2—Milli-Q deionized leaching procedure and ISO/TS 21268-2—CaCl₂ solution leaching procedure were conducted to assess the mobility of these metals. Multivariate statistical methods, including Pearson’s correlation, Principal Component Analysis (PCA) and Cluster Analysis, were applied to identify pollution sources and grouping patterns among elements. The results revealed a distinct clustering of Pb and Zn, separate from other metals, indicating their predominant origin from anthropogenic activities. Contamination Factor (CF), Pollution Load Index (PLI), and Geoaccumulation Index (Igeo) were calculated to evaluate the degree of pollution. Combining total concentration, mobility, and multivariate analyses offers a more comprehensive insight into the extent and origin of pollution in the urban area of Novi Sad. The results obtained are valuable for evaluating the soil conditions in the Western Balkans, which have been recognized as a necessity by the EU.

1. Introduction

A clean environment is essential for maintaining a healthy global population. Sustainability, which refers to practices that meet present needs without compromising the ability of future generations to meet their own needs, was formally recognized with the adoption of the Green Agenda 2030 [1]. This agenda established 17 goals for sustainable development. Healthy soil plays a central role in achieving several of these goals, including zero hunger, good health and well-being, clean water and sanitation, life below water, and life on land.

The significance and relevance of soil pollution investigations are evident in the continuously growing number of published papers in this field. For instance, a search on Scopus for “heavy metals in soil” reveals over 64,000 results, indicating a substantial upward trend in global awareness and academic focus on soil quality issues. Until 2007, the number of results was fewer than 1000. However, by 2014, it had surpassed 2000, and by 2018, it exceeded 3000. The most significant increase occurred in 2024, with more than 6000 papers published on this subject.

Soil pollution leads to soil degradation and the loss of essential soil functions, posing significant threats to ecosystems and human health. Exposure to elevated concentrations of heavy metals is associated with various health risks [2]. Numerous industrial activities, such as mining, power generation, the application of fertilizers and pesticides, wastewater discharge, hazardous waste disposal, and exhaust emissions, are major sources of heavy metal contamination [3]. In addition to these anthropogenic sources, natural processes can also contribute to elevated heavy metal levels in soils, particularly in regions with naturally high geochemical backgrounds.

The most common heavy metals found in soil include copper, cobalt, chromium, cadmium, nickel, lead, manganese, zinc, and iron [4]. Hence, it is important to determine their concentrations and distinguish naturally occurring levels from those originating from anthropogenic activities in order to assess the environmental risk. In this context, measuring and mapping the total metal concentrations in soils becomes essential for several reasons. First, it helps to identify areas with elevated contamination, which is a critical step in risk assessment and remediation planning. Moreover, such data serve as a baseline for predicting contaminant spread, guiding land management decisions, and safeguarding agricultural productivity. Second, by combining measured values with mathematical models, it is also possible to estimate the concentrations of toxic elements in untested parts of the study area [5,6,7].

To address these issues, many legislative authorities worldwide have adopted regulations establishing guideline values for the assessment and remediation of contaminated sites. These guidelines often focus on the total metal content in the soil. A notable aspect of developing soil quality standards is the consideration of multiple input factors. These include the subject and object of standardization, the assessment of potential negative effects, regional soil and geochemical backgrounds, and the implications for both the ecosystem and landowners if a standard is exceeded [8]. However, this approach may inadvertently lead to an overestimation of potential risks, resulting in unnecessary and costly soil remediation efforts [9]. I was shown that soil-dwelling organisms are only negatively impacted by the biologically available (bioavailable) fraction of the total metal content in the soil, rather than by the portions that are sequestered or irreversibly bound to the soil matrix [10,11,12]. Thus, determination of toxic elements’ mobility and bioavailability is also crucial [13]. Chemical species can bond with soil components, minerals, and organic matter. The nature of these bonds affects how contaminants move through soil and water, as well as their bioavailability-the ability of contaminants to enter and accumulate in plant species within a given area [10]. Mobility and bioavailability can be evaluated using chemical methods that involve extracting contaminants with a chemical agent. Often used methods include the CEN 12457-2—Milli-Q deionized leaching procedure [14] and the ISO/TS 21268-2—CaCl₂ solution leaching procedure [15]. The Milli-Q deionized water leaching method was developed to assess the mobility of inorganic constituents from solid waste materials and to estimate their potential bioavailability in the environment, while the CaCl₂ solution leaching method was designed to evaluate the release of inorganic and organic constituents from soil samples and to predict the ecotoxicological effects of the resulting eluates on microorganisms, fauna, and flora.

To identify the origins, loading conditions, and spatial distribution of heavy metals, statistical methods such as Pearson’s correlation coefficient and Principal Component Analysis (PCA) are frequently used. PCC identifies relationships among different heavy metals in soil, while PCA determines the sources of heavy metal contamination in the study area [16]. PCA reduces multidimensional data by converting correlations between variables into a representative two-dimensional graph [17].

Assessing total heavy metal content in soil alone is insufficient for comprehensive evaluation [18]. Comparing measured concentrations with literature reference values provides only a preliminary indication of contamination and does not fully characterize soil quality [19,20]. To more effectively evaluate heavy metal enrichment and its relationship with soil properties, computational tools and pollution indices are increasingly utilized [18,21,22]. Computational methods are valuable for managing large datasets, uncovering hidden patterns, and establishing quantitative relationships between heavy metals and soil properties. However, these statistical approaches do not capture the full complexity of soil pollution. Pollution indices offer a more comprehensive assessment by integrating multiple variables into a single metric that reflects environmental risks, degradation levels, and potential health impacts [23]. These indices also facilitate differentiation between contamination from natural processes and that from anthropogenic activities.

For example, extensive research has already been conducted in the Handan region, where Positive Matrix Factorization (PMF) was utilized to analyze the contributions of various pollution sources [24]. Similar investigations have also been carried out in urban areas [6,25,26,27], mining areas [7,28,29,30,31], and agricultural soils [32,33,34]. Studies from Bangladesh focused on contamination patterns, potential sources, and associated health risks [35]. The research encompassed spatial and temporal variations, pollution source identification, statistical analyses, and evaluations of ecological and health risks [36,37].

This study investigated the metal content and mobility in soil samples from the Novi Sad municipality. Novi Sad has been identified as a city with multiple sources of pollution, including industrial activity and heavy traffic. Its location on a flat plain allows for the easy movement of pollutants by the wind, as there are no geographical features to obstruct or redirect their transport. The primary objectives were to identify contaminated locations, determine the sources of contamination, and evaluate the associated risks. In order to achieve the given goals, it is necessary to:

• Define the total amounts of heavy metals in the soils and their interrelationships.
• Determine the availability and mobility of potentially toxic elements, including Cr, Pb, Cu, Zn, As, Co, Cd, and Ni, through leaching studies.
• Apply statistical tools to classify and differentiate the observed samples.
• Calculate the pollution indices.

The study underscores the importance of evaluating total concentrations, mobility and bioavailability parameters simultaneously. This integrated approach provides a more accurate evaluation of environmental risks and supports the development of sustainable land-use and industrial policies.

Arias-Navarro et al. [38] highlighted that countries in the Western Balkans are relying on outdated soil data for assessments and monitoring. This reliance has led to a significant gap between the region’s soil health and that of EU countries. Therefore, there is an urgent need to establish a robust soil monitoring system that can provide reliable data for updating soil policies. The findings presented in this work will contribute to filling in missing data and addressing knowledge gaps regarding the state of soil in the Western Balkans.

2. Materials and Methods

2.1. Sampling Locations and Sample Preparation

In 2020, urban soil sampling was conducted in the city of Novi Sad [39]. This urban area is situated on a geological foundation of alluvial sediments from the Holocene geological period. These alluvial deposits comprise Holocene sandbars, sandy clays, silt, and sand [40].

Samples were randomly collected in the areas with potential different pollution sources (Figure 1, Table S1):

• Near oil refinery (S13–S16).
• Near road in Avijatičarsko quarter (S2, S3)
• Bistrica quarter (S8, S11, S12)
• Detelinara quarter (S5, S7)
• Jugovićevo quarter (S1, S4, S6)
• Liman quarter (S9)
• The city center (S17)
• Southwest of the city center, along the road to Veternik settlement (S19)
• Danube Park (S21) and Futog Park (S18).
• Sunny Quay located in the forest (S10) and Sunny Quay next to the river (S20).

Soil samples were collected from the surface layer at a depth of 0 to 5 cm.

The pebbles and debris were removed. Following collection, the samples were dried at 105 °C and subsequently sieved.

Figure 1. Sampling locations.

Figure 1. Sampling locations.

2.2. X-Ray Fluorescence Spectroscopy (XRF)

Total amounts of Cr, Pb, Cu, Zn, As, Mn, Co, Cd, Ni and Fe were analyzed by X-ray fluorescence spectroscopy (XRF) using a Niton Xl3t Goldd+ analyzer (Thermo Fisher Scientific, Waltham, MA, USA) which features a 50 kV radiation source and a high-resolution detector. The instrument is mounted on a specially designed test stand with a fixed holder, ensuring precise and repeatable positioning between the radiation source, sample, and detector. This setup provides stable measurement conditions, minimizes external influences, and reduces signal noise.

Each sample was analyzed five times, for 180 s in Soil mode. Measuring time was chosen in order to achieve limits of detection (LOD) for each element. Defined LOD values were: 15.0 mg/kg for Cr, 4,6 mg/kg for Pb, 8.7 mg/kg for Cu, 6.9 mg/kg for Zn and Cd, 4.0 mg/kg for As, 37.6 mg/kg for Mn, 25.0 mg/kg for Co and 17.3 for Ni. Before measurements, XRF analyzer was calibrated by performing an auto-calibration system check. Before, during and after XRF measurements, the instrument precision was evaluated by measurement of standard reference material (NIST, SRM 2711a).

2.3. Statistical Evaluation

The correlation analysis was used to determine the relationships between the total amounts of heavy metals in the studied samples. The Pearson coefficients (r) were calculated with a significance level of α = 95%. Multivariate statistics, PCA and Cluster analysis was performed using MATLAB R2024b Software.

2.4. Pollution Indices

Commonly used indices, CF (Contamination Factor), PLI (Pollution Load Index) and, I_geo (Geoaccumulation Index) [21] were calculated.

The CF represents the degree of enrichment of a given metal in soil relative to its background concentration. It provides a straightforward measure of how much the concentration of a particular metal exceeds natural values and can be calculated using Equation (1).

$$CF={\displaystyle \frac{C_i}{C_{i,background}}}$$

(1)

where c_i is metal concentration in the soil sample, and c_i,background is natural background metal concentration.

The PLI provides an overall assessment of heavy metal pollution in soil by integrating the contamination factors (CFs) of all analyzed metals. It is calculated as the geometric mean of CF values for each determined element (CF1 is CF calculated for Cr, CF2 is CF calculated for Pb, etc.) (Equation (2)):

$$PLI=\sqrt[n]{CF1\times CF2\dots \times CFn}$$

(2)

Igeo index evaluates the degree of soil contamination by accounting for both current concentrations and natural background variability (Equation (3)). It is widely applied to classify pollution severity across different environments.

$$I_{geo}=\mathit{log}2\left({\displaystyle \frac{C}{1.5\times C_{\mathrm{background}}}}\right)$$

(3)

2.5. Mobility and Bioavailability of Elements of Interest

In addition to evaluating the total elemental content, the mobility and potential bioavailability of these elements were also examined.

The total concentration of metals in the soil does not always accurately reflect the actual ecotoxicological risk, as the availability of these elements depends on their chemical forms and their migration ability. To address this, the standardized leaching tests based on CEN 12457-2 and ISO/TS 21268-2 were conducted. These tests enable the assessment of the proportion of elements capable of transitioning into a mobile form. To evaluate the mobility and bioavailability in the soil samples studied, the following extraction or leaching tests were performed:

• CEN 12457-2—Milli-Q deionized leaching procedure [14]: 10 g of each soil sample with an accuracy of ±0.0001 g and a particle size < 4 mm were shaken with 100 mL of Milli-Q deionized water (without pH adjustment) at room temperature on a rotary shaker at 10 rpm for 24 h. Deionized water was prepared by Barnstead GenPure water purification system (Thermo Fisher Scientific, USA).
• ISO/TS 21268-2—CaCl₂ leaching procedure [15]: 10 g of each soil sample with an accuracy of ±0.0001 g and a particle size < 4 mm were shaken with 100 mL of 0.001 M calcium chloride (p.a. Merck, Darmstadt, Germany) solution (without pH adjustment) at room temperature on a rotary shaker at 10 rpm for 24 h.

At the end of each procedure, the suspensions were centrifuged for 15 min at 8000 revolutions per minute (rpm) in a Heraeus Megafuge 16 centrifuge (Thermo Fisher Scientific, USA), filtered through 0.45 mM membrane filters, and analyzed within 24 h. All leaching experiments were performed in triplicate, and the results were finally expressed as mean values.

The concentrations of the elements of interest were measured using an inductively coupled plasma spectrometer ICP-OES Avio 200 (Perkin Elmer, Shelton, CT, USA), according to EPA [41]. Before measurements, instrument was calibrated using solutions prepared using a multicomponent standard solution containing 28 elements in HNO₃ 5% for ICP (CPA chem Ltd., Stara Zagora, Bulgaria). Each sample was measured three times and the relative standard deviation in these measurements was less than 8%. Quality control of the analytical process, carried out using the certified reference material EPA Method 200.7 LPC Solution (CPA chem Ltd., Bulgaria), demonstrated that the resulting concentrations were within the range of 91–110%. Limits of quantification of ICP-OES were 0.004 mg/L for As, Cr, Mn, Ni, Zn and Cd, 0.001 mg/L for Co and Pb, and 0.005 mg/L for Cu.

In addition, the pH values of the filtrates obtained according to CEN 12457-2 were determined. pH was measured according to SRPS EN ISO 10523:2016 [42] using the InoLab pH meter (WTW, Weilheim, Germany) equipped with the glass electrode SenTix-81. The electrical conductivities (ECs) were determined according to SRPS EN 27888:2009 [43] with the conductometer Cond 3100 equipped with TetraCon^® 325 electrode (WTW, Germany). Both instruments were regularly calibrated with WTW standard solutions.

3. Results and Discussion

3.1. Total Concentrations of Metals

The total concentrations of metals in soil samples were measured using XRF analysis, whereas the descriptive statistics of the obtained data is given in

Table 1. Descriptive statistics of concentrations of the elements (mg/kg) defined using XRF analysis.

Variable	Mean	SE Mean	StDev	CoefVar	Minimum	Maximum
Cr	30.2	2.999	13.74	45.4	15.0	54.5
Pb	24.3	3.904	17.89	73.72	10.96	84.2
Cu	28.0	4.246	19.46	69.59	8.67	83.6
Zn	29.5	8.306	38.10	129.13	6.94	120
As	8.43	0.4756	2.179	25.84	5.24	13.1
Mn	406	25.04	114.7	28.26	256	705
Ni	26.9	2.700	12.37	45.97	17.3	56.6
Fe	18,936	736.3	3374	17.82	14,473	26,193

. All measured samples had Cd and Co values below detection limits. Therefore, these elements were excluded from further analysis.

The measured values varied in the wide range for Cr, Pb, Cu, Zn, Mn, Fe and Ni which standard deviations and coefficients of variations suggested large spatial variation and possible contaminated sites. On the basis on mean values, the concentration changed in the following sequence: As < Pb < Ni < Cu < Zn < Cr < Mn < Fe. Found concentrations for Fe and Mn are comparable with the background values [44]. Measured values of Cr, Pb, Zn, As, Co and Cd are lower than maximally allowed in the Republic of Serbia [45]. Eight samples contain higher amounts than maximally allowed of Ni and Cu, but lower than remediation values and all of them are on the same, south–north direction. The obtained results are similar to already reported studies for this territory [40,46], with the exception that no significant amounts of Co were found in soil samples. Notably, high concentrations of arsenic have been reported in the Province of Vojvodina, where Novi Sad is located [47].

It has been observed that the concentrations of Cr, As, Ni and Zn are higher in the city of Kragujevac, Serbia, compared to Novi Sad. Meanwhile, the levels of Pb and Cu are similar in both cities [48]. Additionally, soil samples from Belgrade show significantly higher concentrations of nickel (Ni) and zinc (Zn) [49].

Table 2. Threshold values for heavy metals established in specific regulations > (1) maximally allowed, (2) remediation level, (3) Intervention levels, (4), lower guideline values, (5) risk values limits for agricultural land pH > 7.5 and (6) limited values.

	Serbia [45]		Dutch [50]	Finland [51]	China [52]	EU [53]
	(1)	(2)	(3)	(4)	(5)	(6)
Cr	100	380	180 Cr (III) 78 Cr (VI)	200	250	-
Pb	85	530	530	200	170	50–300
Cu	36	190	190	150	100	50–140
Zn	140	720	720	250	300	150–300
As	29	55	76	50	25	-
Ni	35	210	100	100	190	30–75
Cd	0.8	12	13	10	0.6	1–3
Co	9	240	190	100	-	-

presents limits for heavy metals as outlined in various regulations. Overall, the regulations regarding soil quality are relatively consistent across the different laws.

3.2. Statistical Analysis

To define the relationships between heavy metal content in the examined soils, a correlation analysis was conducted, calculating Pearson’s coefficients for each pair of heavy metals (Figure 2 and

Table 3. Pearson’s correlation coefficients and probability values between different elements. Correlations that are statistically significant at * 99% and ** 95% (null hypothesis test).

	Cr	Pb	Cu	Zn	As	Mn	Ni
Pb	0.538 ** p = 0.012
Cu	0.779 * p < 0.001	0.434 p = 0.049
Zn	0.618 ** p = 0.003	0.890 ** p < 0.001	0.419 p = 0.058
As	0.563 ** p = 0.008	0.289 p = 0.204	0.699 ** p < 0.001	0.310 p = 0.172
Mn	0.787 ** p < 0.001	0.385 p = 0.085	0.809 ** p < 0.001	0.523 * p = 0.015	0.766 * p < 0.001
Ni	0.697 ** p < 0.001	0.571 ** p = 0.007	0.720 ** p < 0.001	0.620 ** p = 0.003	0.536 ** p = 0.012	0.682 * p = 0.001
Fe	0.573 ** p = 0.007	0.005 p = 0.984	0.644 ** p = 0.002	0.181 p = 0.431	0.860 * p < 0.001	0.820 * p < 0.001	0.481 p = 0.027

). The results of linear regression for each element pair are shown in Figure 2, while the regression coefficients along with their respective p-values are displayed in Table 3.

Statistically significant coefficients were identified for the following pairs: Pb-Cr, Zn-Cr, As-Cr, Mn-Cr, Pb-Zn, Pb-Ni, Cu-As, Cu-Mn, Cu-Fe, Zn-Mn, Zn-Ni, and As-Mn. The results presented in Table 2 suggest that As, Ni, Mn, Cr, Cu and Fe may share a similar natural origin.

The low correlation coefficients observed between Pb-Cu, Pb-As, Cu-Zn, As-Zn, Mn-Pb, Fe-Pb, Fe-Ni, and Fe-Zn indicate that these elements likely originate from different sources. This suggests that the pollution of Pb and Zn may be driven by human activities, approved with strong correlation between Pb and Zn, with probability p < 0.001. Various identification techniques used to determine pollution sources in cropland soils across China yielded similar findings [54].

PCA was utilized to extract valuable information from data related to element concentrations. The choice of PCA is based on its ability to reveal hidden structures within highly correlated environmental data and to group heavy metals in the soil according to their common sources [55]. PCA simplifies complex datasets by projecting them into fewer dimensions, making patterns easier to visualize [56]. The aim of PCA is to identify a reduced set of features that adequately represent the original data in a lower-dimensional subspace, minimizing the loss of information [57]. Several prerequisites were considered prior to applying PCA in this study. First, the dataset included ten heavy metal variables measured on a continuous scale, which is sufficient for dimensionality reduction. Second, a reasonable degree of linear correlation among the variables was confirmed, as shown in Table 2, since PCA relies on Pearson-type relationships. Third, the suitability of the dataset for PCA was assessed using the Kaiser–Meyer–Olkin (KMO) measure, which yielded a value of 0.717, indicating that the dataset was appropriate for PCA [58]. To ensure the validity of the analysis, cobalt and cadmium were excluded due to their zero variance. Additionally, Bartlett’s test of sphericity was significant (p < 0.05), confirming that the correlation matrix was suitable for factor extraction [59]. Outliers were examined, and no extreme values were found. Factor extraction was conducted using varimax rotation with a maximum of 25 iterations to enhance interpretability, while eigenvalue decomposition was applied to the correlation matrix to maximize the explained variance of the initial components.

The constructed scree plot (Figure 3a) illustrates the relationship between eigenvalues and factors. Two factors are sufficient to describe the dataset based on eigenvalues greater than 1. Consequently, PCA was performed by extracting two components that account for 84.1% of the data’s variability (PC1 explains 48.7%, whereas PC2 explains 35.4%).

According to the values of rotated factor coefficients, presented in

Table 4. Factor coefficients of heavy metals for the first two components.

Variable	PC1	PC2
Fe	0.959	0.063
As	0.886	−0.140
Mn	0.872	−0.364
Cu	0.794	−0.403
Cr	0.664	−0.580
Pb	0.063	−0.956
Zn	0.175	−0.929
Ni	0.572	−0.631

and Figure 3b, it can be concluded that PC1 is strongly connected with amounts of Fe, As, Mn, Cr and Cu whereas PC2 is strongly connected with Pb and Zn concentrations. Ni is almost equally connected to PC1 and PC2. PCA was already successfully used for the investigation of pollutants in soil and for detection of their sources [40,54,60].

Investigating soil samples from Novi Sad, Mihailovic et al. [40] also extracted two principal components from the datasets but with strong connections of As, Co, Cr, Mn, and Ni in PC1 and Cu, Pb, and Zn in PC2.

Since the examined elements showed a clear association through the principal components, cluster analysis was subsequently applied. Cluster analysis is a technique used to group similar objects or data points into subgroups, known as clusters, based on their similarities. The objective is to maximize the similarity within each cluster while maximizing the differences between clusters. A dendrogram visually represents how variables or parameters come together to form clusters with similar characteristics. In this study, cluster analysis was conducted for two purposes: (1) to identify contamination locations and to group similar sampling sites and (2) to identify similar groups between contaminants of interest.

As illustrated in Figure 4, all sampling sites are categorized into two statistically significant groups based on Pearson’s distance and Ward linkage. Cluster 1 includes sites S1–S12, S18, S19, and S21, while Cluster 2 consists of sites S13–S17 and S20. In general, Cluster 2 contains sites with higher heavy metal concentrations, with mean values of 40.3 mg Pb/kg, 44.8 mg Cr/kg, 50.9 mg Cu/kg, 68.6 mg Zn/kg, 10.8 mg As/kg, and 42.5 mg Ni/kg. These sampling sites are located around the oil refinery, at the center of the town, and near the river in Sunny Quay, all organized in a northeast–southwest direction. The dispersion of heavy metals from the refinery to the city center is plausible, especially considering that the wind occasionally blows in that direction, though this occurs less than 10% of the time annually [61]. Additionally, heavy metals contribute to environmental pollution due to the intense traffic in the city center.

The average element concentrations at locations in Cluster 1 are 17.8 mg Pb/kg, 24.4 mg Cr/kg, 18.7 mg Cu/kg, 13.8 mg Zn/kg, 7.47 mg As/kg, and 20.7 mg Ni/kg. Among the points in Cluster 1, Danube Park has the highest concentrations of heavy metals, likely due to its location and the impact of nearby traffic. This green space is surrounded by heavily trafficked roads that often experience congestion and slow-moving vehicles. Notably, to the southeast of the park lies the busy Mihajlo Pupin Boulevard, and the southeast wind is the most prevalent in this city. Points S21 and S10 are positioned in the same direction as the points in Cluster 2; however, they belong to Cluster 1 due to their lower heavy metal concentrations. This is likely because both locations are forested, indicating that trees may help mitigate the influence of pollution.

Furthermore, the Cluster analysis applied in order to group contaminants (Figure 5) yielded the same answers as correlation analysis and PCA. Two dendrogram branches were found: one containing Cr, Cu, Mn, Ni, As, Fe and the other containing Pb and Zn.

Novi Sad has several well-developed branches of industry, including the oil refinery, food industry, electric cable industry, and chemical industry. Pb and zinc Zn can be emitted as a result of fuel combustion from traffic and the oil refinery [40]. While leaded gasoline was banned in EU countries in 2000, it remained in use in Serbia until 2011. Zinc is frequently incorporated into motor oil formulations and subsequently combusted in older engines. This practice has increased in prevalence over the past several decades.

3.3. Pollution Indices

In urban environments, where multiple anthropogenic sources contribute to heavy metal contamination, pollution indices provide an effective means to quantify contamination levels and differentiate natural impacts from those induced by humans. These indices are widely used in environmental studies because they integrate complex chemical data into simple, interpretable measures of pollution intensity, soil quality, and ecological risk.

The soil in the Novi Sad municipality has a sandy texture, with most samples classified as sandy loam. There are much lower amounts of loamy sand, sand, clay loam, sandy clay loam, and loam [40]. The contamination factor was calculated for each element and soil sample according to Equation (1) using the background values specific to this region [40]. In general, for CF < 1, the soil is uncontaminated, for 1 ≤ CF < 3 low contamination; for 3 ≤ CF < 6 → moderate contamination; and for CF ≥ 6 → high contamination. Average CF values were found to be 1.00, 1.41, 1.63, 0.49, 3.83 and 1.81 for Cr, Pb, Cu, Zn, As and Ni, respectively.

According to this, there is low contamination level for Pb, Cu, Zn and Ni, as well as moderate As contamination. However, from Figure 6 can be seen that contamination factors varied in a wide range for each metal, indicating that contamination was not found in each soil sample. For 48% of samples showed CF < 1 for Cr, and 52% with 1 < CF < 3. When lead was considered a contaminant, 47.7% of samples were uncontaminated, 42.8% low, and 9.5% moderately contaminated. In addition, 43% showed low Cu and 9% moderate contamination. Low Zn contamination was found in 3 soil samples, whereas 15 soil samples were moderately contaminated with As. Ni contamination was low in 20 samples.

To assess the general pollution levels of each soil sample, PLI values were calculated. The PLI acts as a cumulative indicator of soil quality by accounting for multiple pollutants. [62]. Equation (2) was used to calculate the PLI for each soil sample by multiplying all the CF values for each metal. The CF values were determined for Cr, Pb, Cu, Zn, As, and Ni. The PLI is then calculated as the sixth root of the product of these CF values.

Calculated PLI values showed that 52.4% of samples poses PLI < 1, indicating that soils are unpolluted (Figure 7).

The Igeo is used to assess the degree of soil and sediment contamination with metals in relation to their natural background concentration. The values of this index enable the classification of pollution into several categories, each with distinct risk implications. When Igeo ≤ 0, the sample is considered uncontaminated and poses no additional environmental risk, as the metal concentrations are within natural limits. Within the range of 0 to 1, the soil is described as mildly polluted, indicating a low but potential risk, particularly to sensitive organisms. Values between 1 and 2 indicate moderate pollution, implying a moderate risk that may require monitoring or management. A further increase, from 2 to 3, marks the transition between moderate and heavy pollution, with the potential for significant ecological risk. The highest values, when Igeo exceeds 5, indicate extreme pollution, which signals very high anthropogenic impact and substantial environmental risk that may necessitate urgent intervention.

Based on the calculated mean metal concentrations, the values of Igeo show an increasing trend in the following order: Zn (−1.61) < Cr (−0.58)< Cu (−0.094) < Pb (−0.08) < Ni (0.27) < As (1.35). This arrangement indicates that zinc and chromium are in the non-pollution zone (Igeo < 0), while copper and lead are approaching the limit value and can be considered unpolluted to slightly polluted. In contrast, nickel already shows a slight degree of pollution. The highest value is recorded by arsenic, which is classified as moderate pollution, making it a special element of importance for assessing potential environmental risk.

3.4. Mobility and Bioavailability of Elements of Interest

The obtained results of the pollution indices indicate elevated levels of certain elements. However, the concentration itself in the solid phase does not necessarily imply a high risk; therefore, their mobility was also examined. Figure 8 presents the acidity (pH) and conductivity (EC) values of the solution obtained after leaching with deionized Milli-Q water (obtained by Barnstead GenPure water purification system, Thermo Fisher Scientific, USA) according to the CEN 12457-2 procedure [14]. These parameters indicate the potential environmental impact of the leachate.

The pH values measured in the water indicate the active acidity of the soil and were found in the range from 8.3 to 8.9, with an average value of 8.4. The soils in Novi Sad have a high carbonate content, with measured pH values in KCl exceeding 7, indicating their alkaline nature [40]. Soils with an alkaline pH of 8.5 or higher are considered unfavorable, as they impede nutrient uptake and the application of fertilizers, ultimately negatively impacting potential remediation efforts.

Table 5. Descriptive statistics of the leaching results (mg/kg) according to the CEN 12457-2 procedure.

Variable	Minimum	Maximum	Mean	Standard Deviation	Limit Value for Inert Waste [14]
Cr	0.010	0.010	0.010	0	0.5
Pb	0.010	0.010	0.010	0	0.5
Cu	0.070	0.580	0.210	0.105	2.0
Zn	0.025	0.080	0.027	0.012	4.0
As	0.005	0.620	0.097	0.158	0.5
Mn	0.005	9.73	3.05	3.268	-
Co	0.010	0.010	0.010	0	-
Cd	0.005	0.005	0.005	0	0.02
Ni	0.010	0.080	0.038	0.030	0.40

presents the concentrations of leached elements of interest, as determined by the CEN 12457-2 [14] procedure. The leaching results show that the concentrations of all tested elements were within acceptable limits, with some being significantly below the prescribed thresholds [14]. From this perspective, the amounts of leached contaminants do not appear to be harmful, due to their low water solubility. Therefore, it can be concluded that the functional properties of the investigated soils are intact, indicating that sustainable soil quality has been achieved.

The results of leaching by ISO/TS 21268-2, with a CaCl₂ solution with a concentration of 0.001 M [5], are shown in

Table 6. Descriptive statistics of leaching results (mg/kg) according to ISO/TS 21268-2 procedure.

Variable	Minimum	Maximum	Mean	StDev
Cr	0.010	0.010	0.010	0
Pb	0.010	0.010	0.010	0
Cu	0.200	1.37	0.411	0.297
Zn	0.025	0.120	0.036	0.030
As	0.015	0.660	0.232	0.223
Mn	0.130	36.4	10.4	8.53
Co	0.005	0.005	0.005	0
Cd	0.005	0.005	0.005	0
Ni	0.010	0.230	0.088	0.059

. These values are the same or higher than those obtained after leaching in water due to the fact that CaCl₂ solution could desorb elements bonded by ion-exchange on soil matrix [63].

In general, based on the results of both leaching tests, it can be concluded that mobility of investigated elements is very low. Accordingly, it can be assumed that the ecological and health risk is not as high as would be predicted based on the total concentrations.

4. Conclusions

The conducted analysis of soil samples in the region of Novi Sad, Serbia, provides a comprehensive assessment of heavy metal contamination resulting from industrial activities. Although the majority of measured concentrations fall within the limits prescribed by national regulations, elevated levels of nickel and copper were observed in eight samples distributed along a distinct south–north axis.

Multivariate statistical methods—including Pearson’s correlation, cluster analysis, and principal component analysis—indicate that lead and zinc predominantly originate from anthropogenic sources such as oil refineries and traffic-related emissions, whereas other elements (Fe, Mn, Cu As, Cr and Ni) are largely of geogenic origin. Despite the presence of contaminated locations, the low mobility of most heavy metals implies limited bioavailability, thereby reducing their immediate ecological risk.

Nonetheless, the persistence of heavy metals in soil and their potential for long-term accumulation necessitate continuous monitoring and the implementation of targeted remediation strategies for further pollution prevention. The study underscores the importance of evaluating both total concentrations and mobility parameters to accurately assess environmental risks and inform sustainable land-use and industrial policies.