The Adaase community is found in the northern part of the Obuasi Municipality. Obuasi is in the Ashanti region of Ghana, it is located at latitude 6˚12'00" north, longitude 1˚40'00" west. The Obuasi municipality can be accessed by paved roads from Accra, Accra-Takoradi and the Yamoranza-Asante-Bekwai inland roads. The location and accessibility of the study area are shown in Figure 1.

Figure 1. Location and accessibility map of the study area.

The climate of Obuasi is classified as a tropical wet and dry climate under the Köppen-Geiger system. This type of climate is warm throughout the year, with an average annual temperature of 25.5˚C. Obuasi receives 1259 mm to 1750 mm of rainfall yearly, and the humidity is usually high, averaging 75% to 80% during the rainy seasons ([6]). These weather conditions are good for farming but can also cause problems like flooding when there are heavy rains and water shortages during dry seasons ([11]). In the past, the area was covered by thick tropical rainforest, but mining has caused a lot of deforestation and land degradation. Measures are being put in place to restore the vegetation. However, potentially toxic elements have been released into the soil through mining, which makes it harder for plants to grow. These metals can accumulate in crops and disturb human health. This stays a big challenge for farming and vegetation in the area ([17]; [11]). Regardless of these challenges, the vegetation in Obuasi is still essential for keeping the environment balanced, supporting farming and reducing the effects of climate change ([4]). Vegetation in the area stays critical for maintaining ecological balance, supporting livelihoods, and mitigating the impacts of climate change.

Figure 2. Geological map of the study area.

The geology of Obuasi and its geological features, as shown in Figure 2, is shaped by natural processes that led to one of the richest gold deposits’ formations, globally. The area belongs to the Birimian Supergroup, a large rock formation in West Africa known for its gold deposits. These rocks were formed about 2.1 - 2.2 billion years ago and they include phyllites, greywackes, volcanic rocks and gold-bearing granites ([45]).

One most important characteristic of the area is the Obuasi Shear Zone. A large crack in the Earth’s crust that acted like a pathway for these gold-rich fluids. These fluids, which were hot and carried dissolved gold, were transported through the cracks and interacted with the surrounding rocks. As the fluids cooled, the gold settled in the rocks, majority of them in quartz veins and areas rich in sulphide minerals such as arsenopyrite and pyrite ([38]; [40]). The host rocks, mainly greywackes and phyllites, were originally sediments deposited by rivers and weathering of older rocks. Over time, these sediments were buried, squeezed and turned into harder rocks. The area is full of intersecting faults, folds, and cracks, which created a system for the fluids to flow through. Over time, as tectonic forces continued to act, these cracks became more open, allowing the fluids to flow and deposit gold in multiple stages ([39]).

The field visit was a key part of this research, allowing direct observation and data collection from the study areas. The visit included an artisanal mining site, a nearby farm, a school, and residential areas to collect soil samples for analysis. A total of 10 topsoil samples were collected from five (5) locations around the artisanal mining sites at a depth of 50 cm using a spade. These locations include two separate abandoned mining sites, a farm nearby, a school and a residence. Other details of the sampling sites can be found in Figure 3 and Table 1. These sites were chosen for sample collection to ascertain the level of degradation of the land in the community. Food crops from farms and homes would have to be guaranteed of a good level of quality for the inhabitants. To prevent contamination of the samples during transport, the samples were stored in sealed airtight bags and transported to the lab for further analysis.

Table 1. Locations for sample collection.

Figure 3. Locations for sample collection in Obuasi-Adaase community.

Ten samples were collected from the sampling locations in study area and analysed separately, three duplicates per location were analyzed at the Ghana Geological Survey Authority Laboratory in Accra in May 2025. The values reported throughout this paper are the means of the three subsamples at each location, and n = 10 accordingly refers to location means rather than to individual analyses. Soil samples were air-dried, sieved to ensure homogeneity, and split into two portions for pH determination and X-Ray Fluorescence (XRF) analysis, conducted at the Environmental Laboratory in the University of Mines and Technology, Ghana and the Ghana Geological Survey Authority, respectively.

For XRF analysis, samples were oven-dried for 24 hours and sieved to 90 µm. A 4.0 g aliquot of each sample was homogenised with 0.9 g of Hoechst wax binder and pressed into a 32 mm pellet under 15 tons of pressure using a hydraulic press. Multi-element composition was determined using an energy-dispersive polarising XRF spectrometer (Olympus Vanta M Series, 50 kV, 0.2 mA). To minimise contamination, gloves were worn throughout sample handling, mixing equipment was cleaned with acetone between samples, and pellet surfaces were not touched prior to analysis.

For pH determination, 30 g of soil was mixed with 150 mL of deionised water in a 1:5 soil-to-water ratio, shaken for 45 min on an orbital shaker, and allowed to settle for approximately 3 hours. Soil pH was then measured using a calibrated pH meter, with readings recorded once stable.

3.3.1. Descriptive Statistics

Descriptive statistics, including mean, median, minimum, maximum, range, variance, standard deviation, skewness, and kurtosis, were computed to characterise the distribution and variability of each soil property ([5]). Standard deviation was used to assess data dispersion about the mean, while skewness described deviation from a normal distribution -positive values indicating a right-skewed distribution, negative values indicating a left-skewed distribution, and values near zero indicating symmetry ([31]; [34]).

3.3.2. Multivariate Statistical Analysis

Of the 21 elements determined by XRF, nine were retained for multivariate analysis: eight potentially toxic elements (Cr, Mn, Fe, Ni, Cu, Zn, As and Pb) together with Al, which was included as a conservative lithogenic reference element to help distinguish geogenic from anthropogenic associations. Restricting the variable set in this way also limited the number of variables relative to the ten available cases. A two-tailed Spearman’s correlation matrix was used to assess relationships among the retained elements. R-mode Factor Analysis (FA), using Principal Component Analysis (PCA) with varimax rotation, was applied to identify groupings of related variables. Sampling adequacy was assessed using the Kaiser-Meyer-Olkin (KMO) measure and Bartlett’s test of sphericity. The KMO value was 0.470, below the 0.50 minimum generally regarded as acceptable for factor analysis ([28]), while Bartlett’s test was significant (chi-square = 83.550, df = 36, p < 0.001), confirming that the correlation matrix differed from an identity matrix. The factor solution is accordingly reported as exploratory. Factor analysis was used to infer whether soil contamination originated from anthropogenic or natural sources; only factors with eigenvalues greater than one were retained, yielding five principal factors ([44]). Hierarchical Cluster Analysis (HCA), presented as a dendrogram, was performed to further examine inter-elemental relationships and validate the factor analysis results. All statistical analyses were conducted using SPSS version 27.

The pollution status and environmental risk of Potentially Toxic Elements (PTEs) in soils were evaluated using the Enrichment Factor (EF), Geo-accumulation Index (Igeo), Contamination Factor (CF), and Pollution Load Index (PLI). No published pristine pre-mining background dataset exists for Obuasi covering the elements in the study area, so a global lithological reference is used throughout for internal consistency. Average shale after [48] is the most widely used background reference for Igeo, CF, EF and PLI in the soil and sediment pollution literature and has been applied in the Obuasi area ([17]; [5]). All the 21 elements used in the estimation of the various PTEs used background values from [48], average shale.

3.4.1. Enrichment Factor

The Enrichment Factor (EF) was used to normalise and assess metal concentrations relative to a conservative reference element using Equation (1) ([2]; [22]; [46]).

where (M/Al) sample is the metal to Al ratio in the sample, and (M/Al) background is the corresponding background ratio. Aluminium was adopted as the conservative reference element in preference to iron. Iron is not conservative in this dataset: it has a coefficient of variation of 58.8% against 21.1% for aluminium, and it loads 0.93 on the first principal component alongside Cu and Ni while pairing most closely with Cu in the cluster analysis, indicating that it forms part of the enriched metal association rather than an inert lithogenic reference. The choice is supported by agreement with the other indices: across the ten locations, the aluminium-normalised EF for arsenic tracks its contamination factor at r = 0.933 and its geoaccumulation index at r = 0.883, whereas the iron-normalised EF tracks the same two indices at only r = 0.334 and r = 0.170. EF values were interpreted using the five-tier classification of [46] and [22], as shown in Table 2.

Table 2. Enrichment factor classes ([46]; [22]).

3.4.2. GEO-Accumulation Index (Igeo)

The Geo-chemical Index (Igeo) was introduced by [36] as a method to identify and assess metal contamination in soils by comparing current PTEs concentrations to pre-industrial levels. The Igeo is calculated using Equation (2).

where Bn is the background value of the metal, taken as the average shale composition of [48] and applied uniformly to all 21 elements, Cn is the metal n’s measured concentration in the soil, and factor 1.5 is applied to correct for lithological variations in the background data. Igeo was computed separately for each element at each of the ten sample locations, and the values reported are the mean, minimum, maximum and standard deviation of those per-location indices. Because the logarithm is non-linear, the mean of the per-location indices differs slightly from the index computed from the pooled mean concentration, and the per-location convention is applied throughout so that the reported ranges and the count of locations exceeding Igeo zero refer to the same quantity. The Igeo values are divided into seven categories: Class 0 (Igeo ≤ 0) indicates that the area is practically unpolluted. Class 1 (0 < Igeo < 1) represents areas that are unpolluted to moderately polluted. Class 2 (1 < Igeo < 2) indicates moderately polluted. Class 3 (2 < Igeo < 3) indicates moderate to strongly polluted areas. Class 4 (3 < Igeo < 4) indicates strongly polluted. Class 5 (4 < Igeo ≤ 5) indicates strong to very strongly polluted, and Class 6 (Igeo > 5) signifies severe pollution ([24]; [36]), as shown in Table 3.

Table 3. Geo-accumulation index classes ([24]; [36]).

3.4.3. Contamination Factor (CF)

The Contamination Factor (CF) is calculated as the ratio of the concentration of a specific metal in sediments to its background value, as in Equation (3) ([22]; [25]).

The Contamination Factor (CF) levels were interpreted as follows: CF < 1 indicates no or low contamination, 1 ≤ CF < 3 represents moderate contamination, 3 ≤ CF < 6 indicates strong contamination, and CF ≥ 6 signifies very strong contamination ([22]; [25]), as shown in Table 4.

Table 4. Contamination factor pollution classes ([22]; [25]).

3.4.4. Pollution Load Index (PLI)

The Pollution Load Index (PLI) was calculated using the CFs in Equation (3) as shown in Equation (4) ([47]).

The numerical values in the above equation provide a simple framework for estimating the level of PTE contamination in the soils. A PLI value greater than 1 (PLI > 1) indicates the presence of PTE contamination while a PLI value less than 1 (PLI < 1) shows that there is no contamination, as shown in Table 5.

Table 5. Pollution load index description and classes ([12]; [47]).

The Potential Ecological Risk Index method was developed by Swedish scientist Hakanson in 1980 and it is widely used to evaluate the harmful effects of heavy metals in sediments and assess their potential ecological risks ([25]). The RI method is used to assess the potential for ecological harm from contaminants, particularly heavy metals, in soil and sediments. It considers both the concentration of the contaminant and its toxicity ([32]). The RI was calculated using Equation (5) ([25]).

where Eri is the potential ecological risk factor of an individual element i, Tri is the toxic response factor of element i, and CF i is the contamination factor of element i obtained from Equation (3).

An RI < 150 indicates low-grade ecological risk. 150 ≤ RI < 300 indicates moderate ecological risk. 300 ≤ RI < 600 indicates severe ecological risk, and an RI > 600 indicates serious ecological risk, as shown in Table 6.

Table 6. Scope of RI and general level of RI ([25];[26]).

Elemental analysis of soil samples was performed using a handheld Olympus Vanta VMR X-Ray Fluorescence (XRF) analyser (serial No. 803781, GeoChem mode, 40 kV, graphene detector window) at the Ghana Geological Survey Authority Laboratory, following field-portable XRF protocols for soil and sediment analysis ([49]; [29]). The analyser was factory-calibrated using fundamental parameters calibration and periodically verified against Certified Reference Material NIST SRM 2711a (Montana II Soil; [37]), consistent with recommended handheld-XRF QA/QC practice ([49]). Instrument detection limits (3σ, 99.7% confidence, silica-blank basis) for the elements of principal interest were: As 1 mg/kg, Pb 2 mg/kg, Zn 1 mg/kg, Cu 2 mg/kg, Cr 8 mg/kg, Ni 4 mg/kg, Mn 5 mg/kg, and Fe 12 mg/kg ([41]). Concentrations below the respective LOD were recorded as non-detects.

Table 7. Analytical precision for selected elements from triplicate XRF analysis of soil samples.

Ten samples (A1, A2, B1, B2, C1, C2, D1, D2, E1, E2) were each analysed in triplicate, and precision was expressed as Relative Standard Deviation (RSD) in Table 7. Major elements (Al, Fe) showed excellent reproducibility (RSD < 1%), and As, Zn, and Cu showed good precision (mean RSD < 10%). Occasional elevated RSD for Mn, Pb, and Cr (notably one Cr replicate at sample B1) reflects proximity to instrument detection limits rather than sample heterogeneity, a recognised limitation of field-portable XRF near the LOD ([29]).

The multivariate statistical analyses (PCA and HCA) were conducted on a limited sample size (n = 10) relative to the number of variables analysed (nine elements), which is below conventional recommendations for factor analysis (typically a minimum 5:1 case-to-variable ratio). The sampling-adequacy diagnostics reflect this constraint: the Kaiser-Meyer-Olkin measure was 0.470, below the 0.50 minimum generally regarded as acceptable, although Bartlett’s test of sphericity was significant (chi-square = 83.550, df = 36, p< 0.001), indicating that the correlation matrix was factorable in principle and that the binding limitation is the number of cases rather than an absence of inter-elemental correlation. Consequently, the factor structure and inferred source groupings should be interpreted strictly as exploratory and hypothesis-generating rather than statistically confirmatory, and individual factor loadings should not be treated as stable estimates. It is noted, however, that the element groupings recovered by the Spearman correlation matrix, the factor analysis and the hierarchical cluster analysis were mutually concordant, which lends confidence to the broad source associations inferred, even though the number of factors retained, the variance explained and the individual loadings remain provisional. Because all three procedures derive from the same correlation matrix, this concordance reflects internal consistency rather than independent replication. Future studies should expand the sample size to improve the robustness and generalisability of the multivariate results.

A further limitation applies to the pollution indices. The geoaccumulation index, contamination factor, enrichment factor, pollution load index and potential ecological risk index are all computed against the same set of background values, so a change in the reference set propagates to every classification reported. The sensitivity is greatest for arsenic, whose mean geoaccumulation index of 1.89 lies close to the boundary between Müller classes 2 and 3, such that a modest revision of the arsenic background value would alter its reported class. The value adopted, 13 ppm, has been verified against the primary source, so the residual sensitivity lies in the choice of reference composition rather than in the accuracy of the figure taken from it. Classifications should therefore be read as relative to the reference composition adopted in Section 3.3 rather than as absolute statements of contamination.

During the field visit to artisanal mining sites in the Adaase community, several observations relevant to potential PTE contamination were made. Farms were observed near the mining sites, raising concerns about the potential contamination of agricultural soils with Potentially Toxic Elements (PTEs). This proximity increases the likelihood of crop uptake of harmful metals, posing potential health risks to the local population through dietary exposure. Interaction with a local resident also revealed that gold extraction is not conducted on-site, a deliberate measure by miners to prevent robbery, meaning mercury is not used at the site, thereby reducing the likelihood of mercury contamination in the immediate environment.

Table 8displays the concentrations and statistical summaries of elements in soil samples collected from around mining sites in the study area (Adaase). The mean concentrations (ppm) of the potentially toxic elements decrease in the following order: Fe > Mn > Cr > As > Zn > Cu > Ni > Pb. Pb and Ni recorded the lowest mean values, whereas Fe and Mn recorded the highest. Mean concentrations of Mn, Cr, As, Zn, Cu, Ni and Pb were compared against the Canadian Council of Ministers of the Environment ([15]) agricultural soil quality guidelines, only arsenic and chromium exceeded these thresholds. Arsenic (80.90 ppm) exceeded the CCME limit of 12 ppm by a factor of ~6.7. Chromium (83.03 ppm) exceeded the CCME guideline of 64 ppm by a factor of 1.3. Manganese (123.20 ppm), zinc (37.13 ppm), copper (26.33 ppm), nickel (21.90 ppm) and lead (13.87 ppm) all fell below their respective guidelines, with nickel and lead more than threefold under. Soils around the Adaase mining sites are therefore contaminated with respect to arsenic, and marginally so with respect to chromium under the more conservative CCME criterion; the remaining elements pose no guideline-based concern. Guideline exceedance and geochemical enrichment are treated as distinct here, the pollution indices reported below are calculated against the average shale composition of [48] and thus reflect enrichment relative to crustal background rather than breach of a health- or ecology-based threshold.

The pH of the soil ranged from 4.80 to 6.96 with a mean value of 5.80. This indicates that the soil is generally acidic in this area. The acidic nature of the soil may increase the mobility and toxicity of PTEs, posing potential risks to plants and organisms. All eight potentially toxic elements, ranging from 0.03 for Cr to 1.86 for Zn, have positive skewness, indicating that for these parameters, most of the values are concentrated on the lower side of the range, with a few higher values pulling the tail to the right. The effect is most pronounced for Zn (1.86) and Mn (1.25), which also record the highest kurtosis values (3.26 and 1.70 respectively), reflecting a small number of markedly enriched samples. Aluminium, by contrast, is negatively skewed (−0.41), consistent with its behaviour as a lithogenic reference element rather than a contaminant.

Table 8.Descriptive statistics of elemental concentrations in soil samples.

Box and whisker plots of all 21 elements determined by XRF are presented in Figure 4 to visualise the distribution of elemental concentrations in the soil samples collected. Concentrations are plotted on a logarithmic axis in ppm to accommodate the five orders of magnitude spanned by the data set, from below 1 ppm for U to approximately 280,000 ppm for Si. On this scale, the height of each box represents proportional rather than absolute variability, so boxes may be compared directly across elements of very different abundance. This was done to assess data spread, identify outliers, compare data depths and assess soil contamination ([44]).

Figure 4. Box and Whisker plots of the elements in the study area, concentrations in ppm on a logarithmic scale (dash lines in the boxes are the means whereas the solid lines represent the median).

The boxplots for Ni, Pb, Cr and Cu show that the data are tightly grouped with small ranges, their coefficients of variation being 16.6%, 21.6%, 26.0% and 26.1% respectively. The middle values (medians) are close to the centre of the boxes, which means the data is evenly spread on both sides. The whiskers extend to the minimum and maximum observed values rather than to a fitted outlier threshold, and for these elements, they are short, indicating that the data do not vary much between samples. The similar levels of these elements in all the samples suggest they are naturally present in the environment, with no signs of pollution or unusual increases in their amounts ([14]). The major matrix elements Si, Al, Ti and V likewise plot as narrow boxes, Si being the least variable element in the data set at a coefficient of variation of 6.8%, which is consistent with a uniform lithogenic background across the sampling area.

The boxplots for As and Fe indicate the presence of a larger range of data when compared with Ni, Pb, Cr and Cu, with coefficients of variation of 48.3% and 58.8% respectively. The middle values (medians) are near the middle of the boxes, showing that the data is equally spread on either side of the median. Whiskers are longer, indicating that the concentrations of these elements are more variable. This is a moderate variation, indicating that the concentrations of As and Fe are not equal in all the samples. The differences may be attributed to natural variation in the soil or to the presence of increased levels of these elements in some soils ([8]).

The boxplots for Mn and Zn indicate that these elements have the greatest relative variability of all the potentially toxic elements, with coefficients of variation of 70.6% and 54.9% respectively. As the concentration axis is logarithmic, the boxes express proportional rather than absolute spread: Mn ranges from 35.00 to 316.67 ppm, a ninefold difference between the lowest and highest samples, and Zn from 22.33 to 86.00 ppm. Of the two, Zn is the more strongly bottom-weighted: its median of 27.50 ppm lies close to the lower quartile of 24.33 ppm against a maximum of 86.00 ppm, so most values cluster at the low end with a few samples reaching much higher concentrations. For Mn, the median of 112.67 ppm sits nearer the centre of the interquartile range of 52.33 to 151.67 ppm, and the long upper whisker is driven by the highest sample at 316.67 ppm. The high skewness and kurtosis of both elements (1.25 and 1.70 for Mn, 1.86 and 3.26 for Zn) confirm that these distributions are shaped by a small number of markedly enriched samples rather than a general elevation across the study area.

4.4.1. Spearman’s Correlation Matrix

The Spearman’s correlation coefficients for the elements of the data set are presented in Table 9. Spearman’s correlation is a measure of monotonic relationships between two variables. The values are from −1 to 1, with −1 indicating a perfect negative correlation, 1 indicating a perfect positive correlation and 0 indicating no correlation. Coefficients marked * and ** are significant at the 0.05 and 0.01 levels respectively, and coefficients of ±0.79 or above are shown in bold. The strongest correlation recorded among the potentially toxic elements is between Fe and Cu (0.92**), and the correlation between Mn and As is also strong and positive (0.79**), suggesting that they are related by similar natural or anthropogenic processes such as mining or shared mineral sources. There is also a strong positive correlation between Cu and As (0.73*), which could indicate that they are derived from a shared source or influenced by similar environmental conditions. The strong associations between Cu and Ni (0.74*), Fe and Ni (0.74*) and Mn and Fe (0.70*) suggest that these elements may share a common source of origin. There is a moderate positive correlation between Cr and Pb (0.65*), and a weaker positive correlation between Zn and Pb (0.60), indicating some link between these elements. There is a moderate correlation between Cu and Cr (0.55), indicating that they may share similar sources or environmental activity. There is effectively no correlation between Ni and Zn (0.02), suggesting that they derive from different sources or processes. The correlation between Mn and Cr is very weak and positive (0.15), implying only a minor relationship between the two elements. Al, retained as a lithogenic reference element, correlates negatively with Mn (−0.73*) and Zn (−0.61), indicating that these two elements do not track the aluminosilicate background and are more plausibly controlled by a non-lithogenic input.

Table 9. Spearman’s correlation matrix.

Note: **: Correlation is significant at the 0.01 level (2-tailed). *: Correlation is significant at the 0.05 level (2-tailed).

4.4.2. PCA with Varimax Rotation

Principal Component Analysis (PCA) was performed by extraction on the correlation matrix, with orthogonal Varimax rotation. Five components met the Kaiser criterion (eigenvalue > 1), together accounting for 94.85% of total variance. The complete loading matrix is reported, with loadings ≥ 0.70 shown in bold to mark component-defining variables and those below 0.40 greyed out for clarity; no loadings were suppressed, so squared loadings across each row sum to the reported communality. Negative loadings reflect the opposing pole of a bipolar component, on PC1, for instance, Fe, Ni, and Cu vary inversely with Si, Ti, V, and Sr. Cr, Mn, and Zn carry appreciable cross-loadings and are not cleanly assignable to a single component.

In Table 10, the component resolved the elements into five components. Each component represents a group of elements that vary together, and the accompanying percentage is the share of the total variance it accounts for. PC1 accounts for 30.17% of the total variance and is bipolar: Cu (0.95), Fe (0.93), Ni (0.80) and As (0.65) load positively, while Si (-0.84), Ti (−0.83), Sr (−0.78) and V (−0.70) load negatively. The positive pole groups elements governed by a common source, while the negative pole reflects their dilution by the primary silicate matrix. The grouping indicates that PC1 is strongly influenced by the local geology, in particular the Birimian rocks that underlie the area. The Birimian rocks are known to contain elevated concentrations of Fe, Ni and As, the elements that define this component. Fe and Ni are frequently found in mafic and ultramafic rocks, which are abundant in the Birimian Supergroup. These elements occur naturally, released by the weathering of arsenopyrite and of the iron and nickel-bearing sulphides and silicates of the mafic units. Their concentrations, however, can be raised by human activities (for example, mining) and by environmental factors (for example, soil acidity).

PC2 accounts for 26.75% of the variance and is dominated by Rb (0.99), K (0.95) and Al (0.87). It is interpreted as the aluminosilicate component of the soils, reflecting the distribution of clay and mica phases derived from weathering of the underlying rocks. Its negative pole is carried by Y (−0.78) and Zr (−0.69), so PC2 separates soils dominated by weathered clay and mica from those in which resistant detrital minerals are concentrated. Ni and Cu, which were grouped together in the earlier solution, now load with Fe on PC1.

PC3 explains 14.53% of the total variance and is defined by Th (0.91) against negative loadings for As (−0.72), Mn (−0.67) and Zn (−0.57), separating the resistant thorium-bearing minerals from the more redox-mobile metals. PC4, which explains 11.55% of the variance, is the component most plausibly attributable to anthropogenic activity in the study area. It is defined by Pb (0.96) and Cr (0.74), with a secondary contribution from Zn (0.52), which suggests that these elements are from a shared source or are affected by similar environmental processes. The primary ores of lead and zinc are galena (PbS) and sphalerite (ZnS), respectively, which are exposed in the environment by mining activities. This exposure hastens their oxidation and releases lead and zinc to the soil. They can also be found with gold-bearing rocks, with further details as to their location in the mining areas. PC5 accounts for the remaining 11.85% of the variance and comprises U (0.82), Nb (0.72), Mo (0.62), Zr (0.54) and Y (0.50), a resistant heavy-mineral association of limited environmental significance in this setting. Communalities range from 0.79 for Ni to 0.99 for several elements, indicating that the five-component solution reproduces almost all of the variance in each element. Cr, Mn and Zn nevertheless load appreciably on more than one component and cannot be attributed to a single source.

Table 10.Varimax-rotated component matrix and variance explained for the elements in soils.

4.4.3. Hierarchical Cluster Analysis

Figure 5 shows the clustering of elements based on their similarities or relationships. It uses Ward’s minimum-variance linkage applied to Euclidean distances between the standardised (z-scored) element concentrations, with distances rescaled to a 0 - 25 range. A cut at approximately 13 on that rescaled scale yields the five-element clusters described below. Distance is labelled on the horizontal axis and refers to the rescaled distance, or how similar or different the elements are. The closer elements are together on the dendrogram, the more similar they are together. The clustering pattern broadly aligns with the factor groupings described above, which lends some mutual support to both analyses, though (as with the PCA) this agreement should be interpreted cautiously, given the small sample size, rather than treated as independent confirmation.

Figure 5. Dendrogram from R-mode hierarchical cluster analysis using Ward’s linkage on euclidean distances of standardised data (n = 10), showing five element clusters: Cluster 1 (Al-K-Rb-V-Sr), Cluster 2 (Si-Ti-Nb-Y-Zr-U), Cluster 3 (Mn-Zn-As), Cluster 4 (Fe-Ni-Cu) and Cluster 5 (Cr-Pb-Mo-Th). Distances are rescaled to 0 - 25.

Cluster 1 groups Al, K, Rb, V and Sr, and represents the feldspar and mica aluminosilicate matrix of the soils. K and Rb are the closest pair in the entire distance matrix, joining at a rescaled distance of 1.79, with Al attaching at 5.68, while V and Sr pair separately at 4.71 and merge with the Al-K-Rb subgroup at 11.07. The cluster is interpreted as the detrital silicate framework of the soils and corresponds to the positive pole of PC2.

Cluster 2 comprises Si, Ti, Nb, Y, Zr and U, a resistant heavy-mineral suite of quartz, rutile, zircon and monazite-type phases. Si and Ti join at 3.05 and Y and Zr at 3.58, with Nb attaching to Si-Ti at 6.75 and U to Y-Zr at 7.33, the two subgroups merging at 11.76; these are physically and chemically resistant minerals of geogenic origin. Cluster 3 is formed by Mn, Zn and As, with Mn and As joining at 3.78 and Zn added at 5.72, and is interpreted as an association of As and Zn scavenged by manganese oxides. The Birimian rocks are thought to contain considerable amounts of Mn and As, which are grouped together in this cluster. These are naturally occurring and result from the weathering of arsenopyrite and manganese oxides in the Birimian rocks, but concentrations are probably higher because of the mining operations, which bring the minerals into contact with the environment and the acidic environment that results.

Cluster 4 groups Fe, Ni and Cu, an association hosted by iron oxides and sulphide phases. Fe and Cu form the second closest pair in the distance matrix, joining at 2.55, with Ni added at 6.66. This cluster corresponds exactly to the positive pole of PC1 and to the Fe-Cu correlation of 0.92** in Table 9. Cluster 5 contains Cr, Pb, Mo and Th. Cr and Pb join at 5.89 and Mo and Th at 7.58, the two pairs merging at 10.93. The Cr-Pb pairing is interpreted as a discrete input, probably anthropogenic, distinct from both the silicate matrix and the iron-oxide association, and it reproduces the Pb-Cr grouping recovered as PC4 in the factor analysis.

At higher levels of the dendrogram, Clusters 4 and 5 merge at 12.14 and then join Cluster 3 at 16.16 to form a broad metal-bearing supergroup, whereas Clusters 1 and 2 remain separate from that supergroup until rescaled distances of 23.71 and 25.00. The primary division in the data is therefore between the lithogenic silicate and heavy-mineral assemblage and the metal-bearing assemblage. Because the analysis rests on ten samples, cluster membership at the margins, particularly for Mo, Th, U and Nb, would be sensitive to the addition of further samples. The dendrogram is accordingly best read as indicative of element associations rather than as a definitive classification.

4.5.1. Results of Geo-Accumulation Index (Igeo)

The results of the geoaccumulation index are presented in Table 11. Seventeen of the twenty-one elements return negative mean Igeo values and fall under Class 0, among them Zn (−2.09), Mn (−3.70), Ni (−2.24), Cu (−1.40), Pb (−1.14), Fe (−1.17) and Cr (−0.75), indicating that their concentrations are below pollution thresholds. This means that the soil is not polluted with respect to these metals, and their levels are within natural background ranges. Zr (0.33), Nb (0.28) and Mo (0.55) return slightly positive mean values and fall under Class 1, unpolluted to moderately polluted. All three are associated with the resistant heavy-mineral phases identified in the cluster analysis rather than with any anthropogenic source, so this marginal enrichment is most likely lithogenic. As has the highest Igeo value among the metals. It falls under Class 2, moderately polluted. Arsenic is the only element to reach a polluting classification, and it does so at every sample location, with individual values ranging from 0.68 to 2.98 and the maximum, at location D1, falling in Class 3, moderately to heavily polluted. This suggests that arsenic levels are significantly higher than natural background levels, likely due to contamination from activities such as mining. Mean Igeo values greater than zero are shown in bold in Table 11, and classes are assigned from the mean across the ten sample locations. Thorium at one location and uranium at six locations returned concentrations below the detection limit of the instrument; Igeo is undefined for a zero concentration, so those samples are excluded from the statistics for these two elements and they should be read as below detection rather than as unpolluted.

Table 11. Geo-accumulation Index (Igeo) of elements determined in soils of the study.

4.5.2. Results of Enrichment Factor (EF)

The enrichment factor for the PTEs is presented in Table 12. Nineteen of the twenty-one elements return mean EF values below 2, indicating deficiency to minimal enrichment. Their concentrations are close to natural background levels relative to aluminium and they do not pose a significant pollution risk. Molybdenum is the exception among the remaining elements, with a mean EF of 2.38, placing it in the moderate category, and it exceeds EF 2 at six of the ten locations. Arsenic has a mean EF of 6.39, placing it in the significant enrichment category, and it exceeds EF 2 at nine of the ten locations, rising to 15.80 at location D1. No element reaches EF 20 anywhere in the study area, so none is very highly or extremely enriched. This is likely from exposure of arsenopyrite to the atmosphere because of mining.

Table 12. Aluminium-normalised enrichment factors of elements determined in soils of the study area.

4.5.3. Results of Contamination Factor (CF)

In Table 13, Zn, Mn, Cu, Pb, Ni and Cr all have mean CF values below 1 (0.39, 0.14, 0.59, 0.69, 0.32 and 0.92 respectively), indicating low contamination. Their concentrations are within natural background levels, and they do not pose any significant pollution risk. Neither Pb nor Ni exceeds CF 1 at any location, and Cr does so at four of the ten locations without its mean reaching the moderate class. As has a CF value greater than 6, indicating very high contamination. This suggests significant pollution, likely from exposure of arsenopyrite to the atmosphere because of mining. Arsenic is the only element in the very high class, with a mean of 6.22 and a maximum of 11.82 at location D1, and it exceeds CF 1 at all ten locations. Ti, Zr and Nb also exceed CF 1 at every location but remain within the moderate class and are interpreted as lithogenic, being resistant heavy-mineral constituents that group together in the cluster analysis; Mo is moderate at nine of the ten locations. Al, Si and Y return mean CF values only slightly above unity (1.08, 1.03 and 1.18). These are matrix constituents rather than pollutants, and values close to 1 indicate that the soils are compositionally similar to average shale, which supports the use of average shale as the reference composition for this study area.

Table 13.Summary of contamination factor for the elements.

4.6.1. Pollution Load Index

The pollution load index of the study area (Table 14) gives a study-area pollution load index of 0.599, calculated as the geometric mean of the ten site values. Every location returns a PLI below unity, ranging from 0.397 at A2 to 0.897 at E1, so no site exceeds the baseline value of 1.0 and the soils are classed as unpolluted on this index throughout the study area. While most elements contribute minimally to the pollution, As is the primary contributor and requires further investigation and management to reduce its environmental impact.

4.6.2. Potential Ecological Risk Index

The single-element risk factors and site-level indices are presented in Tables 14(a)-(b). The ecological risk index averages 72.61 across the ten locations and ranges from 33.38 at A2 to 129.28 at D1 (Table 14(b)). All ten values fall below the conventional threshold of 150 for low risk; on a scale adjusted for the sum of the toxic-response factors, which accounts for the seven elements assessed here rather than the eight of the original formulation. Hakanson’s conventional boundaries of 150, 300 and 600 were calibrated for eight pollutants with a toxic-response factor sum of 133; the seven elements assessed here sum to 29, so those thresholds cannot be reached by construction and would return low risk at every location. Rescaling them in proportion gives boundaries of 32.7, 65.4 and 130.8, on which the four locations in the A and B transects are of moderate risk and the remaining six, including D1 at 129.28, are of considerable risk. Arsenic accounts for 85.7% of the total risk on average. Arsenic is by far the dominant contributor, with a risk factor of 62.23, and is accordingly the element of principal concern in the study area. It should be noted that cadmium and mercury were not among the elements determined by XRF in this study. Both carry the highest toxic-response factors in Hakanson’s formulation, 30 and 40 respectively against 10 for arsenic, so the risk index reported here rests on seven elements only and represents a lower bound on the true ecological risk rather than a complete assessment. The next largest contributions come from Pb (3.47), Cu (2.93), Cr (1.85) and Ni (1.61), all of which are an order of magnitude below arsenic. Zn (0.39) and Mn (0.14) contribute negligibly to the overall risk.

Table 14.Summary of PLI and RI for the elements.

Arsenic contamination in the study area using multi-pollution index assessment is presented in Tables 15(a)-(b). The soils of Adaase are polluted with respect to a single element. Arsenic averages 80.90 ppm against an average-shale background of 13 ppm (range 31.33 - 153.67 ppm, SD 39.07), giving a mean contamination factor of 6.22 (“very high”), a mean geoaccumulation index of 1.89 (Class 2, moderately polluted) and a mean Al-normalised enrichment factor of 6.39 (“significant”). Contamination factor and Igeo place As above threshold at all ten locations, EF at nine and the ecological risk factor at seven; arsenic contributes 72.2% - 91.4% (mean 85.7%) of the ecological risk index at every site, and no other element exceeds the low-risk band anywhere. The apparent conflict between indices is a property of their formulations, Igeo compresses the scale through a 1.5 background correction and a log₂ transform, and PLI dilutes one enriched element across the geometric mean of eight, rather than a contradiction; the low pooled PLI of 0.599 should therefore not be read as evidence of an unpolluted area.

Table 15.Arsenic in soils of the study area: pollution indices and statistical corroboration.

Geochemistry separates lithogenic from anthropogenic control. Arsenic correlates positively with Mn (+0.79), Cu (+0.73) and Fe (+0.64) and negatively with the resistant silicate suite Ti (−0.73), Sr (−0.71), Nb (−0.70) and V (−0.65), and it defines the negative pole of PC3 (loading −0.720; communality 0.990) alongside Mn and Zn, a component independent of the two lithogenic components PC1 and PC2. Hierarchical clustering recovers the same Mn-Zn-As group (As joining Mn at a Ward distance of 1.477), so the association is stable to the choice of method. A coefficient of variation near 50% with positive skew (+0.47) and a single pronounced maximum at D1 (CF 11.82, Igeo 2.98, EF 15.80, Er 118.21) against minima at A1/A2 indicates a point-source-influenced, heterogeneous distribution rather than uniform geogenic enrichment. This is consistent with the geological setting: the Birimian metavolcanic and metasedimentary rocks of the Ashanti belt host gold in arsenopyrite-bearing sulphide ore, so weathering of mineralised bedrock supplies a naturally elevated baseline, while ore processing, tailings and artisanal workings redistribute As onto surface soils where it is scavenged by secondary Mn-Fe oxide phases, the non-detrital host implied by the correlation and factor structure.

The pH-arsenic relationship observed here reflects a between-area rather than a within-site control. Areas A and B are acidic (pH 4.80 - 5.04) and uniformly low in As (31 - 53 ppm), whereas Areas C, D and E are less acidic (pH 5.18 - 6.96) and uniformly higher in As (76 - 154 ppm). This pattern is consistent with the behaviour of arsenic as a predominantly anionic species in soil: unlike cationic trace elements, whose sorption to sesquioxide surfaces weakens as pH falls. Arsenate mobility and availability increase as pH rises, because the positive surface charge of Fe/Mn/Al oxides, and hence their capacity to retain anionic As diminishes at higher pH ([35]). This anion-specific response distinguishes arsenic from the cationic metals discussed above (Mn, Zn, Cu, Ni, Pb), which remained below guideline values across all areas and would be expected to show the opposite pH sensitivity. Elevated pH in the less acidic zones is also consistent with proximity to reclaimed or historically disturbed tailings material. [13] reported that ex-tailings soils at the AngloGold Obuasi concession carried substantially higher pH (7.6 - 8.4) than undisturbed control soil (pH 6.0), reflecting the neutralising effect of tailings substrate and reclamation amendments. That mining disturbance in the wider Obuasi area is associated with elevated, spatially structured arsenic contamination is further corroborated at the groundwater interface by [7], who linked arsenic exceedances in groundwater at Sanso, within the Obuasi municipal district, to mining-related sources. These studies support the interpretation that pH and arsenic respond jointly to a common spatial factor, proximity to mining disturbance, with pH acting mechanistically on the stability of the oxide phases that host arsenic, rather than reflecting a direct within-site causal control.

The contribution of this study is methodological as much as descriptive. Rather than reporting a single index, it applies five (concentration, CF, Igeo, Al-normalised EF, Er/RI) to the same locations on a common background and demonstrates that they agree on the spatial ranking while diverging in absolute class, and it explains that divergence analytically, providing a template for reconciling indices that are routinely reported as though interchangeable. It further couples the index results to correlation, PCA and hierarchical clustering so that the arsenic anomaly is corroborated by three independent lines of statistical evidence, and it quantifies arsenic’s dominance of ecological risk at the location scale rather than as a study-area average. The limits are stated: Cd and Hg were not determined by XRF and carry the highest toxic-response factors (30 and 40 against 10 for As), so RI is a lower bound; with n = 10, the multivariate results are exploratory, and the agreement of PCA and HCA reflects internal consistency of one correlation structure rather than independent replication.