Geosystems and Geoenvironment 4 (2025) 100396 Contents lists available at ScienceDirect Geosystems and Geoenvironment journal homepage: www.elsevier.com/locate/geogeo Source apportionment of potentially toxic elements in soils from an urbanising region: Insights from multivariate analysis in Singida, Tanzania Raymond Webrah Kazapoe a , ∗, Benatus Norbert Mvile b , John Desderius Kalimenze c , d , Daniel Kwayisi e , f , Samuel Dzidefo Sagoe g , Kwabina Ibrahim f , Obed Fiifi Fynn h a Department of Geological Engineering, University for Development Studies, Nyankpala, Ghana b Department of Physics, College of Natural and Mathematical Sciences, University of Dodoma, P. O. Box 259, Dodoma, Tanzania c Department of Geography and Geology, University of Turku, FI-20014, Turku, Finland d Geological Survey of Tanzania (GST), P.O. Box 903, Dodoma, Tanzania e Department of Geology, University of Johannesburg, Auckland Park Kingsway Campus, South Africa f Department of Earth Science, University of Ghana, Legon-Accra, Ghana g Department of Environment and Sustainability Sciences, University for Development Studies, Tamale, Northern region, Ghana h Department of Geological Sciences, University of Energy and Natural Resources, Sunyani, Ghana a r t i c l e i n f o Article history: Received 1 January 2025 Revised 28 March 2025 Accepted 31 March 2025 Handling Editor: Dr. S Sanzhong Li Keywords: Soil pollution Potentially toxic elements Self-organising maps Urbanisation and pollution hazards a b s t r a c t This study evaluates the spatial distribution and geochemical characteristics of potentially toxic elements (PTEs) in soil samples across the Singida area, Central Tanzania, highlighting the environmental impli- cations of rapid urbanisation and contributing to a deeper understanding of soil pollution in urbanis- ing landscapes. A total of 1884 soil samples were analysed with an Inductively Coupled Plasma Mass Spectrometer (ICP-MS). The results of the study show that the background concentrations of the PTEs exceeded their corresponding Upper Continental Crustal (UCC) values in this order; Pb (86.25 %) > Ba (65.23 %) > As (45.65 %) > Cr (15.92 %) > Zn (15.18 %) > V (8.60 %) > Co (7.86 %) > Cu (5.68 %). However, only Cu (17 samples), Pb (2 samples), and Zn (1 sample) reached contaminant thresholds of 200 mg/kg, 200 mg/kg and 150 mg/kg, respectively in some samples. Agricultural practices and soil conditions are possible explanations for the high Cu values, which may be combined with other factors. This study also found that the Co, Cr, Ba and V concentrations vary greatly and even in some samples exceed the recom- mended levels. The principal component analysis, hierarchical cluster analysis, self-organising maps and positive matrix factorisation analysis revealed two main clusters: Ba, Zn and Pb (Factor 1) and Co, Cu, As, Cr and V (Factor 2). Cluster 1 is more prominent across most of the area, particularly the south. Cluster 2 is shown to be more prominent in the Northern part of the area such as Sekenke, Shelui, Lambi, Mtinko and New Kiomboi. Due to the growing rate of urbanisation, these areas have become relatively populous and have a high level of anthropogenic activities, such as gold mining, sunflower oil milling and agri- cultural activities which have been shown in the study to influence the spatial patterns of PTEs in the area. The level of anthropogenic influence on the PTEs calls for remediation and educative measures to be implemented. © 2025 The Author(s). Published by Elsevier Ltd on behalf of Ocean University of China. This is an open access article under the CC BY-NC-ND license ( http://creativecommons.org/licenses/by-nc-nd/4.0/ ) 1 s u a c d t 2 p G h 2 ( . Introduction Soil is a very important element in the delivery of ecosystem ervices to human beings and their environment. Soil resource tilisation as an economic factor, such as urbanisation, industri- lisation, and intensive agricultural production systems, has be- ∗ Corresponding author. E-mail address: rkazapoe@yahoo.com (R.W. Kazapoe) . r t 2 ttps://doi.org/10.1016/j.geogeo.2025.100396 772-8838/© 2025 The Author(s). Published by Elsevier Ltd on behalf of Ocean University http://creativecommons.org/licenses/by-nc-nd/4.0/ ) ome an issue of concern. This is due to the myriad of effects the egradation has on the environment including the release of po- entially toxic elements (PTEs) ( Kumar et al., 2020 ; Ahmad et al., 022 ). PTEs pollution in soils is of particular concern due to its ersistence, concealment and potential irreversibility ( Sethi and upta, 2020 ). In addition, these pollutants are spatially and tempo- ally heterogeneous which depends on parameters such as indus- rial discharge, agricultural activities and urbanisation ( Yang et al., 022 ). of China. This is an open access article under the CC BY-NC-ND license https://doi.org/10.1016/j.geogeo.2025.100396 http://www.ScienceDirect.com http://www.elsevier.com/locate/geogeo http://creativecommons.org/licenses/by-nc-nd/4.0/ mailto:rkazapoe@yahoo.com https://doi.org/10.1016/j.geogeo.2025.100396 http://creativecommons.org/licenses/by-nc-nd/4.0/ R.W. Kazapoe, B.N. Mvile, J.D. Kalimenze et al. Geosystems and Geoenvironment 4 (2025) 100396 m i a e A c a t t f t 2 h a m i c ( c r c t m o r s f r s a m b M c a a n m w c S t Z t p a d n T o r 2 a t t t 2 a c o s h P 2 v c s F t d t i t b s e m 2 m i n h u p a 2 p i m a n s i s i P e i s a f p f f s e u 2 2 ( t d G m m s w a f T m Globally, PTE contamination in soil is attributed to two pri- ary sources: natural (geogenic) inputs and anthropogenic activ- ties. For natural inputs, background concentrations vary between nd within regions, although they are derived mainly from the par- nt material of the soil ( Kabata-Pendias et al., 2017 ; Kazapoe and rhin, 2021 ). In contrast, anthropogenic sources are varied and onsist of industrial production waste, agricultural chemicals, and tmospheric deposition from transportation and energy produc- ion ( Akhtar et al., 2021 ). However, they often overlap, leading o such complex contamination patterns that a robust analytical ramework is needed to identify sources and implement mitiga- ion strategies ( Alloway, 2012 ; Kowalska et al., 2018 ; Kwayisi et al., 024 ). The rapid urbanisation in Tanzania during the last few decades as led to the emergence of pollution hazards often associ- ted with ineffective environmental management. The country re- ains exposed to a significant level of industrialisation, urban- sation, energy production, transportation, and mechanised agri- ultural expansion zones, which have promoted PTE pollution Mng’ong’o, 2022 ; Nyika and Dinka, 2023 ). Urbanization has be- ome a substantial trend throughout Singida region during the ecent decades as its urban districts continue to grow. Statisti- al data from the 2022 Population and Housing Census indicates hat Singida region holds a total population of 2008,058 while its ain urban district contains 232,459 residents ( National Bureau f Statistics (NBS), 2022 ). The population of Singida urban district ose significantly between 2012 and 2022 according to the cen- us data, increasing from 150,379 to 232,459 people, accounting or an annual growth rate of 4.4 % ( NBS, 2022 ). The district cur- ently experiences accelerating urbanisation because of population hifts from inside the country as well as infrastructure expansion nd new economic opportunities located in the area. These accompanying anthropogenic coupled with the extensive ineral deposits throughout Tanzania’s geological landscape com- ine to produce background concentrations of PTEs in its soils. afic-ultramafic rocks in Central Tanzania regions make significant ontributions to soil concentrations of manganese (Mn), zinc (Zn) nd chromium (Cr), according to Mvile et al. (2023) . Small-scale rtisanal mining operations have grown throughout central Tanza- ia producing high levels of PTEs in adjacent soils. Research docu- ents excessive levels of lead (Pb), arsenic (As), and cadmium (Cd), hich exceed safety standards because of unsafe mining activities ombined with flawed waste disposal methods ( Mvile et al., 2023 ). tudies have also shown that soil contamination rises when fer- ilizers and pesticides containing trace metals are used. Zaller and aller (2020) found evidence that agrochemical exposure leads soil o accumulate hazardous PTEs which then threaten agricultural roducts as well as people who consume them. Dar es Salaam, long with other urban centres, experiences environmental degra- ation due to industrial operations releasing untreated waste into atural resources and non-compliant waste management practices. he identified industrial practices produce elevated concentrations f PTEs in both urban soils and aquatic systems which in turn pose isks to nearby agricultural areas during irrigation ( Kibassa et al., 013 ). The sources of these heterogeneities and their interrelations re effectively analysed by employing both geostatistical and mul- ivariate methodologies. These methods mainly analyse the spa- ial distribution but fail to provide comprehensive information on he types of heterogeneity across PTE pollutants ( Kazapoe et al., 021b ). Additionally, the use of soil pollution indices is essential in ddressing many of these challenges. Numerous studies have suc- essfully applied various pollution indices to evaluate the degree f soil contamination ( Kowalska et al., 2018 ; Baran, 2022 ). For in- tance, PTE contamination in soil and associated ecological risks ave been assessed using pollution indices ( Weissmannová and 2 avlovský, 2017 ; Mahvi et al., 2022 ; Xiang et al., 2021 ; Hoque et al., 023 ). To identify sources and PTEs interactions in sediments, ad- anced statistical and modeling approaches are viable methods in- luding principal component analysis (PCA), positive matrix factori- ation (PMF), and the UNMIX model ( Gulgundi and Shetty, 2019 ). or example, PMF quantifies source contributions from concentra- ions and uncertainty of chemical species; PCA is used to reduce ata dimensionality to facilitate the interpretation of pollution pat- erns ( Mas et al., 2010 ). However, these methodologies are increas- ngly complemented by ecological risk assessment methodologies o assess the potential human and environmental impacts of soil- ound PTEs ( Kazapoe et al., 2022 ; Liao et al., 2021 ). Although ignificant progress has been made, the spatial distribution and cosystem risk assessment of PTEs often remain regionally frag- ented and rarely studied on a comprehensive scale ( Ding et al., 018 ; Zhang et al., 2019 ; Gan et al., 2023 ; Wang, 2023 ). The frag- ented nature of current research and the importance of conduct- ng large-scale studies to develop effective mitigation strategies ecessitates a study of this nature. Additionally, temporal trends ighlight the dynamic nature of PTE pollution over time, yet the nderlying drivers, such as urbanisation and shifts in industrial ractices, require further investigation to inform sustainable man- gement strategies ( Sharley et al., 2016 ; Li et al., 2020 ; Pan et al., 021 ; Kazapoe et al., 2023, 2024 ). Understanding these dynamics is articularly critical in regions including Singida and its surround- ngs, where PTE contamination poses a potential threat to environ- ental quality and public health. The hypothesis of this study is that soils in the Singida region re contaminated with potentially toxic elements (PTEs) from both atural and human-induced sources, with anthropogenic activities uch as urbanisation, small-scale mining, and agriculture contribut- ng significantly to elevated levels. Therefore, the objectives of this tudy are to PTE pollution to determine (i) concentrations of PTE n the soil of the study area, (ii) sources and interactions of these TEs, (iii) spatial distribution, and (iv) ecological risk related to el- ments. This study is framed within the context of rapid urbanisation n Tanzania, a perspective rarely emphasized in similar PTE as- essments. The study uses a relatively high-resolution sampling pproach and combines SOM with standard multivariate methods or enhanced source identification as well as detects new urban ollution sources. This research serves as a PTE baseline study or Singida region through which policymakers can use findings or environmental management and urban policy design. The re- earch contributions unify geochemical data analysis with social- nvironmental findings to enhance existing knowledge about an nderdeveloped yet growing vulnerable area. . Materials and methods .1. Area of study The study area is situated in the core of the Tanzania Craton Fig. 1 ) that is concealed by Neoarchaean granitoids, metasedimen- ary, and metavolcanic rocks. The Neoarchaean magmatism was iscontinuous and took 160 million years (from 2775 to 2612 Ma; ST, 2015). Between 2712–2683 Ma, magmatic activity was ter- inated due to crustal uplift and late intrusive rocks were re- oved resulting in siliciclastic deposits and volcanic activity, pos- ibly in an island arc setting. Granite volcan-sedimentary stages ere the last to form, being intruded by large granite plutons nd their associated vein system, syenite plutons were the last ormed in the Neoarchaean magmatic event ( Mvile et al., 2023 ). he invading granites were affected by deformation and low-grade etamorphism of the volcano-sedimentary rocks as evidenced R.W. Kazapoe, B.N. Mvile, J.D. Kalimenze et al. Geosystems and Geoenvironment 4 (2025) 100396 Fig. 1. Geology and Location map of the study area. b m s i ( c M c t a W c a i t b l 2 o t S g i 5 t s t p a a s t s f D s A m s t d h fi y folded granite veins. However, the duration of the tectono- etamorphic event is not clear. Syenite plutons may in places have ome slight signs of deformation or metamorphism, whereby mak- ng the tectono-metamorphic event pre-dated at about 2612 Ma Abu et al., 2024 ). Ferruginous sandstones and conglomerates of Ndago are lo- ated above a peneplained granite surface that can be the regional iocene Surface ( Eades and Reeve, 1938 ). The clastic and chemi- al layer of Quaternary sediments indicating unending erosion due o the tectonic activities in East African Rift System. The highland reas are characterised by residual soils and the lowlands and the embere Mbuga occur in deposited, mainly alluvium. The chemi- al sediments are confined to small calcrete and silcrete layers in lluvium and ferricrete in soil. The study area and Singida region n particular have active artisanal mining for gold, copper, and de- rital zircons-rare earth concentration. Additionally, the mining of uilding materials including granite and clay brick production from ocal clays is the customary activities ( Kalimenze and Mvile, 2024 ). .2. Soil sampling A systematic grid-based sampling design enabled the collection f 1884 soil samples to examine potentially toxic elements dis- ribution and sources along with their concentrations across the 3 ingida region of Central Tanzania. The sampling locations used a eographic coordinate system for placement and followed a set of ntervals between points, which resulted in an average distance of .5 km. The method enabled space consistency while acquiring ex- ensive area coverage that supported upcoming spatial analysis and ource identification examinations. Surface soil samples were ex- racted at each sampling location through a composite collection rocess starting from the top 0–20 cm layer using hand augers nd stainless-steel shovels to characterise agriculturally important nd human-exposure areas. The collection of at least three sub- amples using a triangular layout with 50 to 100-meter spacing be- ween points helped to overcome micro-level variability and boost ampling representativeness. The collected sub-samples were care- ully combined to form a single 5-kilogram bulk sample. The High- ensity Polyethylene (HDPE) bags were used to store individual amples, which received distinct pre-printed identification codes. fter collection, the researchers used field sieving with a 2 mm esh to eliminate stones, roots and coarse organic matter from the amples. The cleaned soil fractions were conveyed to the base sta- ion immediately after sealing them before storage under ambient ry conditions until laboratory analysis began. A set of control samples stemmed from locations lacking uman-caused impacts (outside mining areas and agricultural elds) utilizing land use data together with local expertise. The R.W. Kazapoe, B.N. Mvile, J.D. Kalimenze et al. Geosystems and Geoenvironment 4 (2025) 100396 Table 1 A description of the indices used in this study. Equations Definition of terms References Cf = Ci Cn where Ci is the mean concentration of pollutants and Cn is the pre-industrial reference value. Cf is the contamination factor. ( Hakanson, 1980 ) Cd = n ∑ i =1 Ci f where Ci is the mean concentration of pollutants and Cn is the pre-industrial reference value. Cd is the degree of contamination. ( Hakanson, 1980 ) mCd = 1 n n ∑ i =1 Ci f Cf is the contamination factor, n represents the number of pollutants considered, and mCd is the modified contamination factor. ( Abrahim and Parker, 2008 ) PLI = n √ CF1 × CF2 × CF3 × . . . . . . CFn Cf is the contamination factor, n represents the number of pollutants considered, and PLI is the pollution Load Index. ( Rodrigue et al., 2016 ) PI = Cn GB Cn represents the concentration of heavy metals and GB represents the geochemical background values, and PI is the pollution index. ( Gong et al., 2008 ) PIsum = n ∑ i =1 PI PI is the pollution index, PIsum is the sum of pollution index, and n represents the number of heavy metals considered. ( Kowalska et al., 2018 ) P Ia v erage = 1 n n ∑ i =1 P I PI is the pollution index, PIaverage is the average of pollution index, and n represents the number of heavy metals considered. ( Qingjie et al., 2008 ; Inengite et al., 2015 ) r d r i r w e l p g 2 a t a a r T I r s h t w t y t t t e l p s y a a s A 2 i t d f 2 s c m ( o d w c f r s p a d p t 2 K c p i t a a l T w b r t t T fl s t i l i w B d t esearchers handled the control samples exactly like they han- led the typical grid samples. During the sampling operation pe- iod, the team conducted rigorous quality assurance and qual- ty control (QA/QC) procedures. The sampling process incorpo- ated five percent duplicate sample collection from field stations as ell as blank tests for contamination monitoring and official refer- nce standards during lab-based examinations. The procedure fol- owed for collecting geochemical data achieved reliability and re- roducibility standards that match best practices for environmental eochemistry. .3. Analytical procedures For chemical digestion and elemental analysis, representative liquots (0.25 g) of the < 75 μm fraction were sent to the in- ernationally accredited ACME Laboratories in Canada. The four- cid mixture of hydrofluoric acid (HF) and perchloric acid (HClO4 ), long with nitric acid (HNO3 ) and deionised water operated at a atio of 2:1:1:2 performed a near-total digestion on the samples. he chemical digestion follows the U.S. Geological Survey’s (USGS) CMPS81 procedure alongside total digestion methods, which is egularly used in high-precision environmental geochemistry re- earch (e.g., USGS Open-File Report 03–024). A hot block applied eat to samples, which received 50 % hydrochloric acid (HCl) reatment before complete solid dissolution. Dilute HCl solutions ere used to prepare test tubes which contained cooled solu- ions before the final dilution step to standard volume. The anal- sis was conducted on a 0.25 g split of the samples. The Induc- ively Coupled Plasma Mass Spectrometry (ICP-MS) used to de- ermine the main elements also had detection limits of 0.0 0 01 % o 0.01 % for oxides and 0.002 mg/kg to 2 mg/kg for trace el- ments. Duplicate samples were used to ensure the accuracy of aboratory analytical methods and the results. This research com- rised 1884 soil samples and 209 replicates. The original and sub- tantiation samples exhibited an outstanding correlation. The anal- sed original and replicate samples showed an acceptable vari- tion of 1.2–9.3 %, thereby reflecting good quality control. The nalysis procedure and sampling protocol were conducted as de- cribed by Kalimenze et al. (2023) , Kazapoe et al. (2021a) and bu et al. (2024) . .4. Assessment of PTE pollution in the soil Five indices were analysed to assess the extent of PTE pollution n the study area ( Table 1 ). These included the pollutant accumula- ion index (PGI), PTE enrichment index (HMEI), ecological risk in- ex (ERI), and PTE pollution load index (HMPLI). Table 3 shows the ormula for each index. 4 .5. Statistical analysis The R software was used for the geostatistical analysis. De- criptive statistics helped determine the distribution of elemental oncentrations through calculation of minimum, maximum, mean, edian and standard deviation (SD) and coefficient of variation CV) as well as kurtosis and skewness. The research concentrated n nine potential toxic elements (PTEs). A study has been con- ucted which looks at the correlations between these elements as ell as their relationships with As, Ba, Co, Cr, Cu, Pb and V geo- hemical factors. Python version 3.10.12 was predominantly used or the statistical analysis in the study. This encompasses the cor- elation between parameters and the heatmap generated for its vi- ualisation, the self-organising map (SOM) and its associated com- onent planes, principal component analysis (PCA) and the hier- rchical cluster analysis (HCA), as well as the chord diagram and endrogram heatmap used to visualise them. Excel 2010 was em- loyed for the summary statistics and EPA 5.0 was used to conduct he positive matrix factorisation analysis (PMF). .6. Self-organising map (SOM) The self-organising map (SOM) algorithm, developed by ohonen (1982) , was employed in this study as a powerful tool for lustering and visualising relationships within the dataset. By em- loying a two-dimensional grid topology, the SOM algorithm facil- tates the identification of patterns, groups, and similarities among he variables ( Kohonen, 2001 ). This unsupervised machine learning pproach enables the organisation of high-dimensional data into n interpretable low-dimensional representation, effectively high- ighting inherent clusters and correlations between the input data. he SOM model is trained on the input data, the model updates its eight vectors iteratively based on the input data, using a neigh- ourhood function to ensure that adjacent neurons in the grid rep- esent similar data points. Once the SOM training was complete, he non-hierarchical K-means classification algorithm was applied o refine and finalise the cluster assignments ( Astel et al., 2007 ). his process facilitates the grouping of data into clusters that re- ect their interdependencies, spatial distributions, and potential ources. To ensure the robustness of the clustering process, quan- itative evaluation metrics were incorporated. The Davies-Bouldin ndex (DBI) was used to optimise the SOM parameters, such as the earning rate, neighbourhood size, and grid dimensions. This helps dentify the parameter set that minimised within-cluster variance hile maximising the separation between clusters ( Davies and ouldin, 1979 ). Additionally, Silhouette analysis was employed to etermine the optimal number of clusters by evaluating the consis- ency and compactness of the resulting groupings. Higher Silhou- R.W. Kazapoe, B.N. Mvile, J.D. Kalimenze et al. Geosystems and Geoenvironment 4 (2025) 100396 Table 2 Statistical summary of the results from the Singida area. Cu(mg/kg) Pb(mg/kg) Zn(mg/kg) Co(mg/kg) As(mg/kg) Cr(mg/kg) Ba(mg/kg) V(mg/kg) Min 1.1 2.1 2 0.6 0.5 5 30 1 Max 246.8 302 424 68 84 679 4048 440 Median 11.8 22.7 38 8.7 1 48 539 47 Average 19.25 25.32 44.01 11.25 1.85 62.55 575.06 61.18 SD 20.84 17.38 29.79 8.99 3.03 50.45 330.83 49.94 Skewness 3.06 5.71 2.85 2.06 13.28 2.84 1.96 2.1 Kurtosis 15.06 62.27 20.86 5.75 309.89 16.18 10.71 6.06 CV (%) 108.24 68.63 67.7 79.89 183.03 80.66 57.53 81.62 Exceedance (%) 5.68 86.25 15.18 7.86 45.65 15.92 65.23 8.6 TBS 200 200 100 CCME 63 70 200 12 64 USEPA 1600 400 3100 0.39 0.3 VROM 36 85 140 29 100 ∗TBS: Tanzanian Bureau of Standards contaminant levels for heavy metals. ∗CCME: Canadian Council for Ministers of the Environment standards for Agricultural soils. ∗USEPA: The United States Environmental Protection Agency Regional Screening Level for residential soils. ∗VROM: Circular on target values and intervention values for soil remediation of The Netherlands’s Ministry of Housing, Spatial Planning and Environment. e g 2 p r t d t r w s r 3 3 E p a y ( A v m > c s t a o 1 W V ( o t C t ( a t t 2 t h A v t E s t H 2 t r s n H o t t T j c ( c a w r o i b o o K r c B o h t a p a 6 tte scores indicated well-separated and cohesive clusters, which uided the final selection of clusters. .7. Positive matrix factorisation (PMF) The EPA PMF 5.0 software ( Song et al., 2006 ) was used to com- lete source apportionment analysis. The normalised data with its espective uncertainties were able to input to properly represent he model. Many possible factors were experimented with in or- er to find the most appropriate number to reflect the order of he data. The model performance was done by the Q-value and esidual analysis ( Karakas et al., 2017 ). The resulting PMF model as run with multiple initial conditions to ensure stability and the olution of lowest Q value and consistent factor profiles between uns was selected as the final solution. . Results and discussions .1. Pollutants and elemental concentrations and spatial patterns A statistical summary of the soil quality is presented in Table 2 . xceedance as shown in Table 2 , represents the percentage of sam- les which are above the Upper Continental Crustal (UCC) aver- ges for the PTEs. The average concentration in mg/kg of the anal- sed PTEs in decreasing order are Ba (575.06) > Cr (62.55) > V 61.18) > Zn (44.01) > Pb (25.32) > Cu (19.25) > Co (11.25) > s (1.65) as seen in Fig. 2 a and b. This marks a significant de- iation from the findings of Rudnick and Gao (2003) , who docu- ented the elemental order for the UCC as Ba > V > Cr > Zn Cu > Co > Pb, particularly differing from the results of this urrent study in the relative positions of Cr versus V, Pb ver- us Cu, and Cu versus Co. This relative change in the order of he natural abundance of the PTEs is suggestive of a degree of nthropogenic influence. This is corroborated by the coefficient f variation (CV %) recorded for the PTEs in this study (57.53– 83.03 %) which are classified as high (i.e., 51–100 %) following ang et al. (2024) . The CV % in decreasing order are As > Cu > > Cr > Co > Pb > Zn > Ba. Ba (57.53 %), Pb (68.63 %) and Zn 67.7 %) have high CV % which suggests a mixed influence of ge- genic and anthropogenic factors on their spatial variance across he area. V (81.62 %), Cr (80.66 %) and Co (79.89 %) have very high V % which implies a more pronounced anthropogenic control on heir variance within the area. The significantly high CV % of Cu 108.24 %) and As (183.03 %) indicate that anthropogenic factors re largely responsible for the variability of these metals across 5 he area. CV % > 100 % are typically classified as denoting ex- rinsic sources ( Wu et al., 2014 ; Zhu et al., 2023 ; Adimalla et al., 024 ; Cai et al., 2024 ; Gao et al., 2024 ). As concentrations across he area recorded a maximum value of 84 mg/kg. This relatively igh average value of As (1.85 mg/kg) is identical to the UCC of s which is 1.8 mg/kg. Nearly half (45.65 %) of all samples had As alues exceeding the UCC threshold. This value for As fell below he threshold for both the Canadian Council for Ministers of the nvironment ( CCME, 2007 ; NBS, 2013 ) standards for agricultural oils of 12 mg/kg and the circular on target values and interven- ion values for soil remediation of the Netherlands’s Ministry of ousing, Spatial Planning and Environment (VROM, 20 0 0 ) value of 9 mg/kg. However, it was above the United States Environmen- al Protection Agency’s (USEPA, 2023 ) Regional Screening Level for esidential soils of 0.39 mg/kg. The determined As values in this tudy were much lower than values determined in Geita district of orthwestern Tanzania (63.8 ± 53.2 mg/kg) by Kaaya et al. (2025) . otspots of As covers a significant portion of the Northern part f the area from Sekenke to Lambi ( Fig. 3 a). Ba concentration in he area ranged from 30 to 4840 mg/kg. The average concentra- ion of Ba was 1.3 times more than the UCC of Ba (425 mg/kg). his high value is consistent for most of the samples, with a ma- ority of them (65.23 %) exceeding this threshold. Hotspots of Ba an be found at Ikungi, Malandala, Kinyanguri, Iguguno and Lambi Fig. 3 b). The high spatial variability of Ba is consistent with geo- hemical heterogeneity that is largely driven by both geogenic and nthropogenic factors. Ba may be elevated as a consequence of eathering of Ba-rich Neoarchaean granitoid and syenites, tectonic edistribution of sediments or artisanal mining activities. The area f Ba hotspots corresponds to geological complex areas with min- ng activity, indicating local enrichment. Cu concentration ranges etween 1.10 to 246.80 mg/kg (avg. 19.25). All the samples except ne were below the contaminant level set by the Tanzanian Bureau f Standards ( TBS, 20 07a, 20 07b ) limit of 200 mg/kg as reported in ibassa et al. (2013) . These values are relatively higher than those eported by Kibassa et al. (2013) (8.98 mg/kg) for a study that was arried out in Dar-es-Salaam city. The results also mirror studies by anzi et al. (2015) and Mng’ong’o et al. (2021) who reported values f 8.70 mg/kg and 3.34 mg/kg in Southern Tanzania. The relatively igh concentration of Cu in the area may be because Cu concen- ration generally increases with Cu application, soil pH, soil salinity nd exchangeable Na and Ca ( Wightwick et al., 2006 ). The results resented in Table 2 show that Cr ranges from 5 to 679 mg/kg with n average concentration of 62.55 mg/kg which shows that about 0 % of the samples exceeded the threshold of 80.74 mg/kg that R.W. Kazapoe, B.N. Mvile, J.D. Kalimenze et al. Geosystems and Geoenvironment 4 (2025) 100396 Fig. 2. Box and Whisker plots showing the concentrations of (a) Cu, Pb, Co and As in the study area and (b) the concentrations of Zn, Cr, Ba and V in the study area. Fig. 3. Spatial distribution maps of (a) As (b) Ba (c) Co (d) Cr (e) Cu (f) Pb (g) V and (h) Zn across the Singida area. 6 R.W. Kazapoe, B.N. Mvile, J.D. Kalimenze et al. Geosystems and Geoenvironment 4 (2025) 100396 Fig. 4. Correlation matrix of the elements in the study area. K 3 t T t f w ( t t c s t M t s T a a U t v s w t t ( r a h f t 3 t a 2 r t e w s ( F t p ( C g t a a a t a M t P c t P l p azapoe and Arhin (2021) set in West Africa. This also shows that 01 samples representing 15.98 % of the samples exceed the con- aminant threshold for Cr set by the TDS and VROM (100 mg/kg). he average Cr value for this study (62.55 mg/kg) was nearly iden- ical to the CCME value of 64 mg/kg but above the USEPA value oe residential soils of 0.39 mg/kg. This value for Cr is lower than hat was determined for Nyarugusu in northwestern Tanzania 204.53 mg/kg) by Kaaya et al. (2025) . V has a mean concentra- ion of 61.18 mg/kg. The study showed that the mean concentra- ion of V is below crustal values as set by Taylor (1964) which ould be explained by the geology of the underlying rocks in the tudy area. Co, Cr, Cu and V show similar spatial dispersion pat- erns with hotspots around Shelui, Sekenke, Kinyangui and Lambi tinko in the northern part of the area ( Fig. 3 c-g). Pb concentra- ion ranged from 2.10 to 302.00 mg/kg (Avg. 25.32) with only 2 amples exceeding the guideline value for contaminants set by the BS for Pb. Machiwa (2010) also recorded similar mean values for study that was done in the Lake Victoria Basin of Tanzania. The verage Pb concentration fell below values for CCME (70 mg/kg), SEPA (400 mg/kg) and VROM (85 mg/kg). However, a majority of he samples (86.25 %) exceed the UCC of Pb (12.5 mg/kg). These alues can be found across the entire area with the most intense pots found in the eastern part of the area around Malandala to- ards Puma and Mtinko ( Fig. 3 f). The results of Zn varied be- ween 2.00 and 424.00 mg/kg (44.01 mg/kg). The results show hat 17 of the samples were above the set contaminant level of Zn 150 mg/kg) for Tanzania. Similarly, high values of Zn have been eported for Tazara Mchichani and Temeke Wailes (57.10 mg/kg nd 46.82 mg/kg) in southern Tanzania ( Kibassa et al., 2013 ). Zn otspots are sparse, comprising 15.18 % of the samples, and can be ound around Lambi in the northeast and Ikungi and Mkurusi in he southern part of the area ( Fig. 3 h). 7 .2. Relationships among the pollutant and trace elements The covariance-matrix provides a means to assess the associa- ion between the various elements considered in the study. It is measure of the closeness of dissimilar variables ( Mugheri et al., 019 ). In the output, only positive (direct) and negative (inverse) elationships were highlighted. Out of all the PTE considered in his study, Ba and Pb showed no strong correlation with the other lements. Ba was weakly correlated with Pb (0.29), Zn (0.08) and eakly negatively correlated with As (−0.09) and Cr (−0.15). Zn is trongly associated with V (0.58) and Cr (0.75) and less so with Cr 0.45). Cr on the other hand is strongly correlated with V (0.72). rom Fig. 4 , it is clear that the correlation of elemental concen- rations could be grouped in three categories: high ( r = over 0.5, < 0.05), medium ( r = positive but < 0.5, p < 0.05) and low r = negative values). Those in the first category includes: Zn-Cu, r-Cu, Cr-Co, V-Cu, V-Zn, Co-Zn, Cu-Co, V-Co, and V-Cr. The first roup signals PTEs with likely common origins. This in combina- ion with the summary and spatial characteristics suggests they re likely linked to anthropogenic sources, which are agricultural nd mining activities. These elements are commonly associated nd occur naturally in the environment more closely in associa- ion with the mafic gneisses, metagabbros and anorthosites char- cteristic of the neighbouring areas such as New Kiomboi and kalama. This suggests they may have originated from elemen- al mobility processes. Those in the second category include Zn- b, As-Cu, As-Zn, Cr-Zn, Ba-Pb, Ba-Zn, and V-As. The last group onsist of the correlation between Ba and Pb on one hand and he other PTEs, such as, Pb-Cu, As-Pb, Cr-Pb, Ba-Cu, Ba-Cr, V- b and V-Ba. The third group suggests an association with the ocal geology, influenced to a greater degree by anthropogenic rocesses. R.W. Kazapoe, B.N. Mvile, J.D. Kalimenze et al. Geosystems and Geoenvironment 4 (2025) 100396 Fig. 5. The loadings of the analysed potentially toxic elements (PTEs) in the study onto the first two principal components derived from the dataset. Table 3 Results for the KMO and Bartlett’s test. KMO and Bartlett’s test 0.857 KMO measure of sampling adequacy Approx. Chi-square 8351.361 Bartlett’s test of sphericity Sig 0.000 T a m t t m V t T s t w e a c t w c a p f a o w c a ( 2 t s h t 5 1 n t d a S o f s r ( In terms of the principal component analysis, results from able 3 present the outcomes of the KMO measure of sampling dequacy and Bartlett’s test of sphericity. The sampling adequacy easure was determined to be 0.857, exceeding the required hreshold of > 0.5. Additionally, the significance value for Bartlett’s est was 0.0 0 0, meeting the condition of ( p < 0.001). Fig. 5 and Table 3 show the loadings of potentially toxic ele- ents (PTEs) analysed in the study (Cu, Pb, Zn, Co, As, Cr, Ba and ) onto the first two principal components (PC1, PC2) derived from he dataset. Contributions to PC1 are shown in blue, PC2 in red. he thickness and direction of the connecting chords represent the trength and nature of the relationships between the metals and he components. PC1 accounts for the largest variance (48.80 %), ith Cu, Co, Cr and V having the most significant connections. PC2 xplains 17.01 % of the variance, with notable connections to Pb nd Ba. Factor analysis results support those of the element asso- iation in cluster 1 ( Fig. 4 ) and hence suggest lithological control of he elements present in the samples, particularly those associated ith mafic rocks. The HCA analysis ( Fig. 6 ) outlines 2 main clusters. The first is omposed only of Ba while the second has As, Co, Cu. Cr, Pb, V 8 nd Zn. The high concentration of Ba may be linked to agricultural ractices as well as the surrounding country rocks which include erruginous sandstones and conglomerates, which dominate the rea, and often contain Ba-bearing minerals such as barite (BaSO4 ) r Ba-rich feldspars. These minerals release Ba into the soil during eathering processes. The ferruginous components of these rocks, haracterised by their high iron content and large reactive surface rea, further enhance Ba retention through adsorption mechanisms Tarawneh et al., 2011 ; Jones et al., 2023 ; Jiménez-Vázquez et al., 025 ). The second cluster could be linked to anthropogenic activi- ies which induce localised concentration of these PTEs in the soil uch as mining and agricultural practices Fig. 7 . Leveraging the Davies-Bouldin index (Best DBI: 0.9607) and Sil- ouette score for the identification of the best set of parame- ers (grid dimensions 10, 10, learning rate 0.5, sigma 2, iterations, 0,0 0 0) and the number of clusters (2). The SOM is composed of 00 hexagons each representing a neuron, the number within each euron helps identify the samples within each hexagon thus iden- ifying the clusters the samples belong to as well. The image shows ifferent number of clusters ranging from 2 to 10 and its associ- ted Silhouette score ( Fig. 8 ). The number of clusters with the best ilhouette score is the best for the SOM model. The highest score ccurs at 2 clusters suggesting that 2 is the best number of clusters or the SOM model. Fig. 9 is a component plane for each PTE considered in the tudy. Each square within each component plane represents a neu- on, neurons with high values are represented by bright colours yellow) and neurons with low values, dark colours (blue). Ar- R.W. Kazapoe, B.N. Mvile, J.D. Kalimenze et al. Geosystems and Geoenvironment 4 (2025) 100396 Fig. 6. Dendogram showing the main clusters of PTEs from the study. Fig. 7. Self-organizing map (SOM) d -matrix visualizing clusters based on input data. Sample IDs are displayed within each node, illustrating the distribution of data points across the SOM. e s l T t T C i t t v m P A c u w i K n L r t a L m p h p 9 as with high values suggest high activation indicating the as- ociated neuron is strongly represented by the PTE, areas with ow values suggest less activation thus the PTE is less prominent. he results indicate that Co, Cu, As, Cr and V have high activa- ions in relatively similar regions. Ba, Zn and Pb are the outliers. hese findings mirror the results from Fig. 9 , where Co, Cu, As, r, and V all find themselves in one cluster (1) and Ba, Zn, Pb n another (2). The SOM analysis corroborates the results from he PCA, HCA and spatial analysis in outlining these two clus- ers. Both clusters show the impact of anthropogenic activities to arying degrees. The influence of anthropogenic activities, chiefly ining and agricultural practices, on the concentrations of these TEs in that part of Tanzania has been attested by Mihale (2019) , bu et al. (2024) and Karungamye et al. (2023) . Fig. 10 shows luster 1 covering most of the area, particularly the south. This nderscores the mixed nature of the sources for these elements, here they naturally occur in these areas but have been signif- cantly influenced by human-induced activities ( Abu et al., 2021 ; alimenze et al., 2023 ). Cluster 2 is shown to be more promi- ent in the Northern part of the area such as Sekenke, Shelui, ambi, Mtinko and New Kiomboi. These areas are known to be elatively populous and have high level of anthropogenic activi- ies such as gold mining, sunflower oil milling and agricultural ctivities ( Kalimenze and Mvile, 2024 ). Studies by Manya (2012) , awley et al. (2014) and Henckel et al. (2016) demonstrate gold ineralisation patterns in the area that establish mafic rocks as rimary hosts for gold deposits. The designation of these areas as ubs for Artisanal and Small-Scale Mining (ASSM) has further am- lified their importance as gold production centre. Kalimenze and R.W. Kazapoe, B.N. Mvile, J.D. Kalimenze et al. Geosystems and Geoenvironment 4 (2025) 100396 Fig. 8. Figure showing the results from the Silhouette analysis. Fig. 9. Component planes for each of the PTEs. M t p t t e s G a ( t y g e f t P t T i P B ( t t d a t vile (2024) report that the presence of these deposits has at- racted numerous miners to Singida causing ASSM activities to ex- and quickly. The absence of proper regulations in these activi- ies creates major environmental problems. Rising mining activi- ies have increased concerns as this may introduce PTEs into the nvironment and thereby threaten local ecological systems. Re- earch in similar gold-mining territories including Lake Victoria oldfields shows that mining operations release toxic metals such s arsenic along with Pb and Hg into surrounding natural systems Van Straaten, 20 0 0 ). This aligns with the presence of these clus- ers centred around the known mining areas. The EPA PMF model (V. 5) was applied with the elements anal- sed in the soil samples to identify and quantify the probable ori- ins of the PTEs in the study area as well as the effect of every el- ment. In the present research, the PMF model was used 20 times 10 or which the chosen factors were 2 or 3. In this study, two fac- ors were chosen according to the level of pollutant for each of the TEs. The value of R2 (fitness) of predicted and observed concen- ration for values above 0.94 mark out the fitness of the model. he concentration and contribution rate of every factor are shown n Figs. 11 and 12 for the PTEs. The results from the PMF analysis aligns with the results of the CA ( Table 4 ). Factor 1 shows high loadings by Pb (94.4 %) and a (94.2 %) as the sole members. The CV % of Pb (68.63 %) and Ba 57.53 %) as well as their spatial characteristics shown in Fig. 3 dis- inguishes them from the other PTEs. These characteristics outlines hat these elements are sourced from the local geology but their ispersal across the area have been influenced by anthropogenic ctivities. Consistent with findings by Mvile et al. (2023) , we infer hat the mining and agricultural practices in central Tanzania play R .W . K a za p o e, B .N . M v ile, J.D . K a lim en ze et a l. G eo sy stem s a n d G eo en v iro n m en t 4 (2 0 2 5 ) 10 0 3 9 6 F ig . 1 0 . S p a tia l re p re se n ta tio n o f S O M U -m a trix a cro ss th e stu d y a re a . T a b le 4 Fa cto r lo a d in g s fo r C o m p o n e n ts 1 a n d 2 . P C 1 P C 2 E ig e n v a lu e s 3 .9 0 6 1 .3 6 1 P e rce n ta g e o f v a ria n ce (% ) 4 8 .8 0 1 7 .0 1 C u m u la tiv e p e rce n ta g e (% ) 4 8 .8 0 6 5 .8 1 C u 0 .4 6 2 0 .0 3 3 P b 0 .0 6 1 0 .6 6 6 Z n 0 .3 3 8 0 .3 1 1 C o 0 .4 7 3 0 .0 2 5 A s 0 .2 0 9 0 .1 7 2 C r 0 .4 2 6 0 .0 2 3 B a 0 .0 6 2 0 .6 5 1 V 0 .4 6 3 0 .0 5 4 a ro le in so il p o llu tio n w ith P T E s. Fa cto r 2 h a s stro n g lo a d in g s fo r C u (9 6 .1 % ), V (9 0 .9 % ), C r (9 0 .5 % ), C o (8 4 .8 % , A s (7 2 .7 % ) a n d Z n (7 0 .4 % ). T h e se P T E s sh o w sim ila r v a ria n ce a n d sp a tia l ch a ra cte r- istics in th e a re a . A d d itio n a lly, th e ir v e ry h ig h C V % v a lu e s (6 7.7 – 1 8 3 .0 3 % ) su g g e sts a n th ro p o g e n ic a ctiv itie s lin k e d to th e g o ld m in - in g a n d fa rm in g p ra ctice s. S im ila r a sso cia tio n s w e re d o cu m e n te d in th e S in g id a re g io n b y H e rm a n a n d K ih a m p a (2 0 1 5 ) w ith sm a ll- sca le g o ld m in in g su sp e cte d to b e th e so u rce o f e le v a te d le v e ls o f P T E s in so ils a n d w a te r. Table 5 Summary of the results of the indices. Index Min Max Mean Range Class Number of Samples (%) Degree of Contamination (Cd) ( Hakanson, 1980 ) 0.851 23.997 6.070 < 8 8 ≤Cd < 16 16 ≤Cd < 32 Low Moderate Considerable 1506 (79.94 %) 368 (19.53 %) 10 (0.53 %) Modified Degree of Contamination (mCd) ( Abrahim and Parker, 2008 ) 0.085 2.340 0.607 < 1.5 1.5 ≤ mCd < 2 2 ≤ mCd < 4 4 ≤ mCd < 8 8 ≤ mCd < 16 16 ≤ mCd < 32 mCd ≥ 32 nil to very low low moderate high very high extremely high ultra-high 1868 (99.15 %) 15 (0.80 %) 1 (0.05 %) Pollution Load Index (PLI) ( Rashed, 2010 ; Rai et al., 2019 ) 0 79.49065 0.150969 0 1 2 3 4 5 6 None None to medium Moderate Moderate to strong Strongly polluted Strong to very strong Very strong 1850 (98.20 %) 15 (0.80 %) 7 (0.40 %) 3 (0.20 %) 2 (0.11 %) 1 (0.05 %) 6 (0.32 %) Sum of Pollution Index (PIsum) ( Haque et al., 2022 ) 0.925756 53.88584 10.19577 < 1 1 < Pisum < 3 3 < Pisum < 6 > 6 Low Moderate High Very high 2 (0.11 %) 105 (5.57 %) 488 (25.90 %) 1289 (68.42 %) Average of Pollution Index (PIaverage) ( Gong et al., 2008 ) 0.077146 4.490487 0.849647 < 1 > 1 High quality soil Low quality soil 1271 (67.46 %) 631 (33.49 %) 11 R.W. Kazapoe, B.N. Mvile, J.D. Kalimenze et al. Geosystems and Geoenvironment 4 (2025) 100396 Fig. 11. PTEs profile source and contribution from PMF (a) Factor 1 and (b) Factor 2. Fig. 12. PTEs source and factor fingerprint from PMF. 3 n o a s e j i S o l a t u a t p .3. Potentially toxic element contamination in singida urban and on-urban areas A summary of the results of the indices used to assess the level f contamination or pollution in the soil samples of the study area re shown in Table 5 . The calculation for degree of contamination hows that 19.53 % and 0.53 % of the samples are classed as “mod- rate” to “considerable” contaminated respectively, while the ma- ority of the samples (79.94 %) of the samples are classed as “low” n terms of contamination. These results fall more in line with 12 ivakumar et al. (2016) in coastal sediment from South East Coast f Tamilnadu. The values determined in this study are however ower than what was determined by Mihale (2019) , who found that ll samples from Mtoni estuary in Dar es Sallam were severely con- aminated (DC > 48). Similarly, Table 5 shows that most of the samples (99.15 %) fall nder the “nil to very low” depicting insignificant contamination mong the soil samples in the area. The PIsum analysis indicates hat most samples (68.42 %) are polluted. However, only two sam- les (0.11 %) record low pollution. This is in contrast to the PIaver- R.W. Kazapoe, B.N. Mvile, J.D. Kalimenze et al. Geosystems and Geoenvironment 4 (2025) 100396 a T o 2 l t a d e C r t t i n c C a a o a d m p s B 4 p s p v t P ( ( s i t A o r t a c C e s l r c t v a n L r s t m t a i l t e p v i a t t e s m m f t p D c i C i v B – t W y i W K d w R A A A ge which shows that 67.46 % of the samples are of high quality. he Nemerow Pollution Index reflects a mixed picture where 33 % f the samples are classed under the Heavy Polluted class while 2.56 % and 18.95 % are classed in the Slight and Moderate Pol- uted categories respectively. The HMPLI records the majority of he samples (69.43 %) under the “control” classification. 28.77 % re reported under the baseline level with only 1.8 % reported un- er continuous degradation class. The HMEI analysis identified no nrichment in Cu (91.14 %), Zn (95.28 %), Co (85.56 %), As (99.10 %), r (81.48 %) and V (90.34 %). The analysis also shows moderate en- ichment in Pb (58.12 %) and Ba (43.84 %). The PGI analysis shows hat the significant majority of the samples record no pollution in he order of Ni > Zn > Co > Cu > Sr > Cr > Mn > Pb. Sim- lar findings have been reported by studies from central Tanza- ia by Mvile et al. (2023) which showed that some parts of the oastal zone had elevated concentrations of PTEs such as Pb and r ( Mvile et al., 2023 ). Similarly, Abu et al. (2021) identified As, Cd nd Pb as polluted in Singida. The authors assert that these PTEs re predominantly linked to geogenic source related to the mafic re bearing with some input from the mining processes within the rea. The moderate enrichment shown by the PTE enrichment in- ex (HMEI) could be explained by anthropogenic activities such as ining and industrial processes that have been reported in other arts of Tanzania. Studies have shown, for example, the role of mall-scale gold mining in soil PTEs contamination such as Pb and a ( Mnali, 2001 ). . Conclusions The present study is a comprehensive risk assessment of PTE ollution in Singida and its surrounding areas. The study was de- igned to investigate the environmental quality in relation to PTE ollution as follows: pollutants, concentrations and sources, inter- ening patterns, distributions and ecological risks. The results of he study show that although the background concentrations the TEs exceeded their corresponding UCC values in this order: Pb 86.25 %) > Ba (65.23 %) > As (45.65 %) > Cr (15.92 %) > Zn 15.18 %) > V (8.60 %) > Co (7.86 %) > Cu (5.68 %); only Cu (17 amples), Pb (2 samples), and Zn (1 sample) had reached contam- nant thresholds of 20 0 mg/kg, 20 0 mg/kg and 150 mg/kg respec- ively in some samples. Nearly half (45.65 %) of all samples had s values exceeding the UCC threshold. The average concentration f Ba was 1.3 times more than the UCC of Ba (425 mg/kg). While esults for Cr shows that about 60 % of the samples exceeded the hreshold of 80.74 mg/kg. Agricultural practices and soil conditions re possible explanations for the high-Cu values, which may be ombined with other factors. This research has found that the Co, r, Ba andVconcentrations vary greatly and even in some samples xceed the recommended levels. The covariance-matrix suggests trong possibilities for some of these factors having the same geo- ogical origins or anthropogenic impacts based on their strong cor- elation. The PCA, HCA, SoM and PMF analysis revealed two main luster; Ba, Zn and Pb (Factor 1) and Co, Cu, As, Cr, and V (Fac- or 2). Both clusters show the impact of anthropogenic activities to arying degrees. Cluster 1 is more prominent across most of the rea particularly the south. cluster 2 is shown to be more promi- ent in the Northern part of the area such as Sekenke, Shelui, ambi, Mtinko and New Kiomboi. These areas are known to be elatively populous and have high level of anthropogenic activities uch as gold mining, sunflower oil milling and agricultural activi- ies. Results of contamination assessment indices indicate that the ajority of soil samples possess low nor negligible levels of con- amination, 79.94 % fall under the ’low’ degree of contamination nd 99.15 % in the ’nil to very low’ contamination categories. Var- ous indices show certain differences, indicating that 68.42 % Pol- 13 uted samples, according to Pollution Index (PIsum), mixed pollu- ion from Nemerow pollution index and PTEs enrichment in gen- ral absent except slight enrichment in Pb and Ba. Based on the above, the following is recommended: • Implement phytoremediation using metal-absorbing plants. Use soil washing and stabilisation to reduce PTE concentrations. • Establish health surveillance programs in high-risk areas. En- sure access to safe drinking water with filtration systems or al- ternative sources. • Strengthen environmental laws to align with international PTE standards. Mandate periodic environmental assessments with transparent reporting. • Conduct awareness campaigns on PTE pollution risks and pre- vention. Engage communities in remediation effort s, such as phytoremediation projects. • Deploy advanced monitoring systems such as GIS and real-time sensors. Analyse data using machine learning for pollution pre- diction, and optimised remediation. Future research should prioritise longitudinal monitoring of PTE ollution to capture seasonal variations and long-term trends, pro- iding a more comprehensive means of environmental monitor- ng. Researchers may also need to study the bioavailable fraction nd mobility of PTEs within soil systems to better evaluate poten- ial risk impacts on human health and ecosystems. It is also vital o utilise advanced source apportionment techniques to determine xactly how industrial activities and agricultural practices lead to oil contamination when designing targeted interventions. Further- ore, to combat pollution and guarantee sustainable soil manage- ent, stakeholders in the area need to develop and evaluate eco- riendly remediation methods which fit specific local environmen- al conditions and socio-economic situations for the most affected arts of the area. eclaration of competing interest The authors declare that they have no known competing finan- ial interests or personal relationships that could have appeared to nfluence the work reported in this paper. RediT authorship contribution statement Raymond Webrah Kazapoe: Writing – review & editing, Writ- ng – original draft, Visualization, Supervision, Methodology, In- estigation, Formal analysis, Data curation, Conceptualization. enatus Norbert Mvile: Writing – review & editing, Writing original draft, Validation, Supervision, Methodology, Investiga- ion, Data curation, Conceptualization. John Desderius Kalimenze: riting – review & editing, Writing – original draft, Formal anal- sis, Data curation. Daniel Kwayisi: Writing – review & edit- ng, Writing – original draft, Methodology. Samuel Dzidefo Sagoe: riting – review & editing, Writing – original draft, Methodology. wabina Ibrahim: Writing – review & editing, Writing – original raft. Obed Fiifi Fynn: Writing – original draft, Visualization, Soft- are. eferences brahim, G.M.S., Parker, R.J., 2008. Assessment of heavy metal enrichment factors and the degree of contamination in marine sediments from Tamaki Estuary, Auckland, New Zealand. Environ. Monit. Assess. 136 (1), 227–238 . bu, M., Kalimenze, J., Mvile, B.N., Kazapoe, R.W., 2021. Sources and pollution as- sessment of trace elements in soils of the central, Dodoma region, East Africa: implication for public health monitoring. Environ. Technol. Innov. 23, 101705 . bu, M., Mvile, B.N., Kalimenze, J.D., 2024. Provenance studies of Au-bearing stream sediments and performance assessment of machine learning-based models: insight from whole-rock geochemistry central Tanzania, East Africa. Environ. Earth. Sci. 83 (3), 105 . http://refhub.elsevier.com/S2772-8838(25)00046-9/optrUYRiAdyG1 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0001 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0002 R.W. Kazapoe, B.N. Mvile, J.D. Kalimenze et al. Geosystems and Geoenvironment 4 (2025) 100396 A A A A A B W C C D D E G G G G H H H H H I J J K K K K K K K K K K K K K K K K K K L L L M M M M M M M M M M N N dimalla, N., Gao, Y., Wang, Z., Qian, H., 2024. Spatial distribution, contamination characteristics, and potential ecological risk assessment of trace metals in sur- face soils of south-central stretch of India. Water, Air, & Soil Pollution 235 (8), 522 . hmad, F., Saeed, Q., Shah, S.M.U., Gondal, M.A., Mumtaz, S., 2022. Environmental sustainability: challenges and approaches. Nat. Resources Conservation and Adv. Sustain. 243–270 . khtar, N., Syakir Ishak, M.I., Bhawani, S.A., Umar, K., 2021. Various natural and an- thropogenic factors responsible for water quality degradation: a review. Water. (Basel) 13 (19), 2660 . lloway, B.J., 2012. Heavy Metals in soils: Trace metals and Metalloids in Soils and Their Bioavailability. Springer Science & Business Media . stel, A., Tsakovski, S., Barbieri, P., Simeonov, V., 2007. Comparison of self-organizing maps classification approach with cluster and principal components analysis for large environmental data sets. Water Res. 41 (19), 4566–4578 . anzi, F. P., Msaki, P. K., Mohammed, N. K., 2015. Distribution of heavy metals in soils in the vicinity of the proposed Mkuju uranium mine in Tanzania. Environ Pollut. 4 (3), 42 . ieczorek, J., Baran, A., 2022. Pollution indices and biotests as useful tools for the evaluation of the degree of soil contamination by trace elements. J. Soils Sedi- ments. 1–18 . ai, S., Shen, Z., Zhou, S., Wang, Q., Cheng, J., Yan, X., Tan, M., Tu, G., Cen, Y., 2024. Health risk assessment and potential sources of metals in riparian soils of the Wujiang River, China. Environ. Geochem. Health 46 (3), 106 . CME, 2007. Canadian Soil Quality Guidelines for the Protection of Environmental and Human Health. Canadian Council of Ministers of the Environment, Win- nipeg . avies, D.L., Bouldin, D.W., 1979. A cluster separation measure. IEEE Trans. Pattern Anal. Mach. Intell. 1, 224–227 . ing, X., Ye, S., Yuan, H., Krauss, K.W., 2018. Spatial distribution and ecological risk assessment of heavy metals in coastal surface sediments in the Hebei Province offshore area, Bohai Sea, China. Mar. Pollut. Bull. 131, 655–661 . ades, N.W., Reeve, W.H., 1938. Explanation of the geology of degree sheet No. 29 (Singida). Tanganyika Department of Lands and Mines. Geological Division . an, T., Zhao, H., Ai, Y., Zhang, S., Wen, Y., Tian, L., Mipam, T.D., 2023. Spatial dis- tribution and ecological risk assessment of heavy metals in alpine grasslands of the Zoige Basin, China. Front. Ecol. Evol. 11, 1093823 . ao, Z., Tan, M., Liu, J., Zhang, Y., Niu, Y., Jiang, B., 2024. Characterization of soil trace metal pollution, source identification, and health risk assessment in the middle reaches of the Guihe River Basin. Environ. Monit. Assess. 196 (2), 122 . ong, Q., Deng, J., Xiang, Y., Wang, Q., Yang, L., 2008. Calculating pollution indices by heavy metals in ecological geochemistry assessment and a case study in parks of Beijing. J. China Univ. Geosci. 19 (3), 230–241 . ulgundi, M.S., Shetty, A., 2019. Source apportionment of groundwater pollution us- ing unmix and positive matrix factorization. Environ. Processes 6, 457–473 . akanson, L., 1980. An ecological risk index for aquatic pollution control. A sedi- mentological approach. Water Res. 14 (8), 975–1001 . aque, S. E., Shahriar, M. M., Nahar, N., Haque, M. S., 2022. Impact of brick kiln emissions on soil quality: A case study of Ashulia brick kiln cluster, Bangladesh. Environ. Chall. 9, 100640 . enckel, J., Poulsen, K. H., Sharp, T., Spora, P., 2016. Lake victoria goldfields. Episodes 39 (2), 135–154 . erman, A., Kihampa, C., 2015. Heavy metals contamination in soils and water in the vicinity of small scale gold mines at londoni and sambaru, Singida region, Tanzania. Int. J. Environ. Monitoring and Analysis 3 (6), 397 . oque, M. M., Islam, A ., Islam, A . R. M. T., Pal, S. C., Mahammad, S., Alam, E., 2023. Assessment of soil heavy metal pollution and associated ecological risk of agri- culture dominated mid-channel bars in a subtropical river basin. Sci. Rep. 13 (1), 11104 . nengite, A. K., Abasi, C. Y., Walter, C., 2015. Application of pollution indices for the assessment of heavy metal pollution in flood impacted soil. Int. Res. J. Pure Appl. Chem 8 (3), 175–189 . iménez-Vázquez, A., Jaimes-López, R., Morales-Bautista, C.M., Pérez-Rodríguez, S., Gochi-Ponce, Y., Estudillo-Wong, L.A., 2025. Catalytic applications of natural iron oxides and hydroxides: a review. Catalysts. 15 (3), 236 . ones, K.L., Ziegler, B.A., Davis, A.M., Cozzarelli, I.M., 2023. Attenuation of barium, strontium, cobalt, and nickel plumes formed during microbial iron reduction in a crude-oil-contaminated aquifer. ACS Earth and Space Chemistry 7 (7), 1322–1336 . aaya, N.I., Vegi, M.R., Macheyeki, A.S., 2025. Health risks of geogenic contaminants in gold mining areas in Geita, Tanzania. J. Trace Elem. Med. Biol., 100222 . abata-Pendias, A., Dudka, S., Chlopecka, A., 2017. Background levels and environ- mental influences on trace metals in soils of the temperate humid zone of Eu- rope. In: Biogeochemistry of Trace Metals. CRC Press, pp. 73–96 . alimenze, J.D., Mvile, B., 2024. The investigations of ore and tailing samples for im- proving extracting methods and enhancing gold recovery of artisanal and smal- l-scale miners: a case study of Singida Region in Tanzania, East Africa. J. Mines Geosci. 1, 44–72 . alimenze, J.D., Abu, M., Mvile, B.N., 2023. Soil geochemistry and multivariate statis- tical assessment of copper–Gold-PGEs mineralization in parts of Singida Region of the Tanzania Craton, Tanzania, East Africa. Arab. J. Geosci. 16 (1), 59 . arakas, F., Imamoglu, I., Gedik, K., 2017. Positive matrix factorization dynamics in fingerprinting: a comparative study of PMF2 and EPA-PMF3 for source appor- tionment of sediment polychlorinated biphenyls. Environ. Pollution 220, 20–28 . arungamye, J., Rwiza, M., Selemani, J., Marwa, J., 2023. Geochemistry of Potentially 14 Toxic Elements in Soil and Sediments of a Tanzanian Small-Scale Gold Mining Area . azapoe, R., Arhin, E., 2021. Determination of local background and baseline val- ues of elements within the soils of the Birimian terrain of the Wassa Area of Southwest Ghana. Geology, Ecology, and Landscapes 5 (3), 199–208 . azapoe, R.W., Amuah, E.E.Y., Dankwa, P., Ibrahim, K., Mville, B.N., Abubakari, S., Bawa, N., 2021a. Compositional and source patterns of potentially toxic ele- ments (PTEs) in soils in southwestern Ghana using robust compositional con- tamination index (RCCI) and k-means cluster analysis. Environ. Challenges 5, 100248 . azapoe, R.W., Arhin, E., Amuah, E.E.Y., 2021b. Known and anticipated medical ge- ology issues in Ghana. Ecofeminism and Climate Change 2 (4), 169–184 . azapoe, R.W., Amuah, E.E.Y., Dankwa, P., 2022. Sources and pollution assessment of trace elements in soils of some selected mining areas of southwestern Ghana. Environ. Technol. Innov. 26, 102329 . azapoe, R. W., Addai, M. O., Amuah, E. E. Y., Dankwa, P., 2023. Hydrogeochemi- cal characterization of groundwater in the Wassa Amenfi East and Prestea-Huni Valley areas of southern Ghana using GIS-based and multivariate statistical techniques. Sustain. Water Resour. Manag. 9 (5), 141 . azapoe, R.W., Amuah, E.E.Y., Dankwa, P., Fynn, O.F., Addai, M.O., Berdie, B.S., Douti, N.B., 2024. Fluoride in groundwater sources in Ghana: a multifaceted and country-wide review. Heliyon. 10 (13) . ibassa, D., Kimaro, A . A ., Shemdoe, R. S., 2013. Heavy metals concentrations in se- lected areas used for urban agriculture in Dar es Salaam, Tanzania. Sci. Res. Es- says. 8 (27), 1296–1303 . ohonen, T., 1982. Self-organized formation of topologically correct feature maps. Biol. Cybern. 43 (1), 59–69 . ohonen, T., 2001. Self-Organizing Maps, 3rd ed Springer, Berlin/Heidelberg, Ger- many . owalska, J.B., Mazurek, R., Gąsiorek, M., Zaleski, T., 2018. Pollution indices as useful tools for the comprehensive evaluation of the degree of soil contamination–A review. Environ. Geochem. Health 40, 2395–2420 . umar, V., Singh, J., Kumar, P. (Eds.), 2020. Environmental Degradation: Causes and Remediation Strategies (Vol. 1). Agro Environ Media, Publication Cell of AESA, Agriculture and Environmental Science Academy . wayisi, D., Kazapoe, R.W., Alidu, S., Sagoe, S.D., Umaru, A.O., Amuah, E.E.Y., Kpiebaya, P., 2024. Exploring soil pollution patterns in Ghana’s northeastern mining zone using machine learning models. J. Hazard. Mater. Adv. 16, 100480 . awley, C. J. M., Selby, D., Condon, D., Imber, J., 2014. Palaeoproterozoic orogenic gold style mineralization at the Southwestern Archaean Tanzanian cratonic mar- gin, Lupa Goldfield, SW Tanzania: Implications from U–Pb titanite geochronol- ogy. Gondwana Res. 26 (3–4), 1141–1158 . i, C., Sanchez, G.M., Wu, Z., Cheng, J., Zhang, S., Wang, Q., Li, F., Sun, G., Meen- temeyer, R.K., 2020. Spatiotemporal patterns and drivers of soil contamina- tion with heavy metals during an intensive urbanization period (1989–2018) in southern China. Environ. Poll. 260, 114075 . iao, J., Cui, X., Feng, H., Yan, S., 2021. Environmental background values and eco- logical risk assessment of heavy metals in watershed sediments: a comparison of assessment methods. Water, 14 (1), 51 . achiwa, J.F., 2010. Heavy metal levels in paddy soils and rice (Oryza sativa (L)) from wetlands of Lake Victoria Basin, Tanzania. Tanz. J. Sci. 36 . ahvi, A. H., Eslami, F., Baghani, A. N., Khanjani, N., Yaghmaeian, K., Mansoorian, H. J., 2022. Heavy metal pollution status in soil for different land activities by con- tamination indices and ecological risk assessment. Int. J. Environ. Sci. Technol. 19 (8), 7599–7616 . anya, S., 2012. SHRIMP zircon U-Pb dating of the mafic and felsic intrusive rocks of the Saza area in the Lupa goldfields, southwestern Tanzania: Implication for gold mineralization. Nat. Sci. 4 (9), 724–730 . as, S., de Juan, A., Tauler, R., Olivieri, A.C., Escandar, G.M., 2010. Application of chemometric methods to environmental analysis of organic pollutants: a review. Talanta 80 (3), 1052–1067 . ihale, M. J., 2019. Multivariate assessment of metal contamination in mtoni sedi- mentary environment, Tanzania. Tanz. J. Sci. 45 (2), 158–172 . nali, S.R., 2001. Assessment of heavy metal pollution in the lupa gold field, SW Tanzania. Tanzania J. Sci. 27 (2), 15–22 . ng’ong’o, M., Comber, S., Munishi, L. K., Ndakidemi, P. A., Blake, W., Hutchin- son, T. H., 2021. Potentially toxic elements status and distribution in Us- angu agroecosystem–Tanzania. Environ. Chall. 4, 10 020 0. doi: 10.1016/j.envc. 2021.10 020 0 . ng’ong’o, M., 2022. Comparative assessment of soil phosphate status, water eu- trophication, and potentially toxic metal accumulation in Usagu agro-ecosystem (Doctoral dissertation, NM-AIST). ugheri, M.H., Pathan, A .M., Sayed, M.A ., Maira, M., Soomro, D.B., Ali amur, S., Soomro, N.A., 2019. Assessment of drinking water quality district Jamshoro Sindh Pakistan: A case study. Int. J. Curr. Res. 11 (3), 1812–1816 . vile, B.N., Abu, M., Kalimenze, J.D., 2023. Assessment of heavy metals concentra- tion in soils in the central parts of Tanzania using pollution indices and multi- variate statistical approach: implication on the source and health. J. Sediment. Environ. 8 (3), 457–469 . ational Bureau of Statistics, 2013. 2012 Population and Housing Census: Population distribution By Administrative Areas. Government of Tanzania, Dar es Salaam, Tanzania Retrieved from. https://www.nbs.go.tz . ational Bureau of Statistics, 2022. 2022 Population and housing census: Population distribution by administrative units. Government of Tanzania, Dodoma, Tanzania Retrieved from. https://www.nbs.go.tz . http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0003 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0004 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0005 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0007 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0008 http://refhub.elsevier.com/S2772-8838(25)00046-9/opt0FLVjfKP9w http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0015a1 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0010 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0048 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0013 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0014 http://refhub.elsevier.com/S2772-8838(25)00046-9/optxrXHvrrOav http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0016 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0017 http://refhub.elsevier.com/S2772-8838(25)00046-9/opt5HKOcSdWdt http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0018 http://refhub.elsevier.com/S2772-8838(25)00046-9/optAioYBmtmF5 http://refhub.elsevier.com/S2772-8838(25)00046-9/optHKBjlm4Tm2 http://refhub.elsevier.com/S2772-8838(25)00046-9/optjBxZCNQTHl http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0019 http://refhub.elsevier.com/S2772-8838(25)00046-9/optV12CD2XIMT http://refhub.elsevier.com/S2772-8838(25)00046-9/optQHP6TvpCXa http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0020 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0021 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0022 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0023 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0024 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0025 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0026 http://refhub.elsevier.com/S2772-8838(25)00046-9/optkNUpsg8hu0 http://refhub.elsevier.com/S2772-8838(25)00046-9/optkNUpsg8hu0 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0032 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0029 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0031 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0027 http://refhub.elsevier.com/S2772-8838(25)00046-9/optWvoC0vJ55J http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0028 http://refhub.elsevier.com/S2772-8838(25)00046-9/optjUJKDhkn1T http://refhub.elsevier.com/S2772-8838(25)00046-9/optl1dDlrE5Cc http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0036 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0037 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0038 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0039 http://refhub.elsevier.com/S2772-8838(25)00046-9/optIHydmewZY8 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0040 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0041 http://refhub.elsevier.com/S2772-8838(25)00046-9/opt51LyGzBzns http://refhub.elsevier.com/S2772-8838(25)00046-9/opte7UeNlCK2t http://refhub.elsevier.com/S2772-8838(25)00046-9/optIkaTyLihMB http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0043 http://refhub.elsevier.com/S2772-8838(25)00046-9/optNzkLmVaaWb http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0044 https://doi.org/10.1016/j.envc.2021.100200 http://refhub.elsevier.com/S2772-8838(25)00046-9/optBSm5sKrZbf http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0046 https://www.nbs.go.tz https://www.nbs.go.tz R.W. Kazapoe, B.N. Mvile, J.D. Kalimenze et al. Geosystems and Geoenvironment 4 (2025) 100396 N P Q R R R R S S S S T T T T U V V W W W W W X Y Z Z Z yika, J., Dinka, M.O. (Eds.), 2023. Global Industrial Impacts of Heavy Metal Pollu- tion in Sub-Saharan Africa. IGI Global . an, Y., Ding, L., Xie, S., Zeng, M., Zhang, J., Peng, H., 2021. Spatiotemporal simula- tion, early warning, and policy recommendations of the soil heavy metal envi- ronmental capacity of the agricultural land in a typical industrial city in China: case of Zhongshan City. J. Clean. Prod. 285, 124 84 9 . ingjie, G., Jun, D., Yunchuan, X., Qingfei, W., Liqiang, Y., 2008. Calculating pollution indices by heavy metals in ecological geochemistry assessment and a case study in parks of Beijing. J. China Univ. Geosci. 19 (3), 230–241 . ai, R., Sharma, S., Gurung, D. B., Sitaula, B. K., 2019. Heavy metal contamination in sediments from vehicle washing: a case study of Olarong Chhu Stream and Paa Chhu River, Bhutan. Int. J. Environ. Stud. 76 (1), 66–83 . ashed, M. N., 2010. Monitoring of contaminated toxic and heavy metals, from mine tailings through age accumulation, in soil and some wild plants at Southeast Egypt. J. Hazard. Mater. 178 (1–3), 739–746 . odrigue, K. A., Yao, B., Trokourey, A., Kopoin, A., 2016. Assessment of heavy metals contamination in sediments of the Vridi Canal (Côte d’Ivoire). Journal of Geo- science and Environment Protection 4 (10), 65–73 . udnick, R.L., Gao, S., 2003. Composition of the continental crust. Treatise on Geo- chemistry vol. 3, 1–64 . ethi, S., Gupta, P., 2020. Soil contamination: a menace to life. Soil Contamination. IntechOpen . harley, D.J., Sharp, S.M., Bourgues, S., Pettigrove, V.J., 2016. Detecting long-term temporal trends in sediment-bound trace metals from urbanised catchments. Environ. Poll. 219, 705–713 . ivakumar, S., Chandrasekaran, A., Balaji, G., Ravisankar, R., 2016. Assessment of heavy metal enrichment and the degree of contamination in coastal sediment from South East Coast of Tamilnadu, India. J. Heavy Metal Toxicity Diseases 1 (2), 1–8 . ong, Y., Xie, S., Zhang, Y., Zeng, L., Salmon, L.G., Zheng, M., 2006. Source apportion- ment of PM2. 5 in Beijing using principal component analysis/absolute principal component scores and UNMIX. Sci. Total Environ. 372 (1), 278–286 . anzania Bureau of Standards (2007). Soil quality limits for Soil contaminants in habitat and agriculture TZS 972:2007 (E). anzania Bureau of Standards. (20 07). TZS 972:20 07: soil quality – Limits for soil contaminants in habitat and agriculture. Dar es Salaam, Tanzania. arawneh, K., Alnawafleh, H., Harahsheh, M., Radaideh, N., Shreideh, S., Moumani, K., Tarawneh, B., Abdelghafoor, M., 2011. Mineralogy, geochemistry and genesis of the ferruginous sandstone in Batn Al Ghul Area/Southern Jordan. Ann Univ Min- ing Geol 54 (I), 89–94 . 15 aylor, S.R., 1964. Abundance of chemical elements in the continental crust: a new table. Geochim. Cosmochim. Acta. 28 (8), 1273–1285 . SEPA, 2023. Regional Screening Levels (RSLs) – Generic Tables (May 2023). U.S. Environmental Protection Agency, Office of Superfund Remediation and Tech- nology Innovation . an Straaten, P., 20 0 0. Mercury contamination associated with small-scale gold mining in Tanzania and Zimbabwe. Sci. Total Environ. 259 (1–3), 105–113 . ROM, 20 0 0. Circular On Target Values and Intervention Values For Soil Remedia- tion. Ministry of Housing, Spatial Planning and Environment, Netherlands . ang, N., Liu, Z., Sun, Y., Lu, N., Luo, Y., 2024. Analysis of soil fertility and toxic metal characteristics in open-pit mining areas in northern Shaanxi. Sci. Rep. 14 (1), 2273 . ang, Y., 2023. Ecological risk identification and assessment of land remediation project based on GIS technology. Environ. Sci. Poll. Res. 30 (27), 70493–70505 . eissmannová, H.D., Pavlovský, J., 2017. Indices of soil contamination by heavy met- als–methodology of calculation for pollution assessment (minireview). Environ. Monit. Assess. 189 (12), 616 . ightwick, A., Mollah, M., Smith, J., MacGregor, A., 2006. Sampling considerations for surveying copper concentrations in Australian vineyard soils. Soil Res. 44 (7), 711–717 . u Jin, W.J., Teng YanGuo, T.Y., Lu SiJin, L.S., Wang YeYao, W.Y., & Jiao XuDong, J.X. (2014). Evaluation of soil contamination indices in a mining area of Jiangxi, China. iang, M., Li, Y., Yang, J., Lei, K., Li, Y., Li, F., Zheng, D., Fang, X., Cao, Y., 2021. Heavy metal contamination risk assessment and correlation analysis of heavy metal contents in soil and crops. Environ. Poll. 278, 116911 . ang, Z., Zou, L., Xia, J., Qiao, Y., Bai, F., Wang, Q., Cai, D., 2022. Spatiotemporal vari- ation characteristics and source identification of water pollution: insights from urban water system. Ecol. Indic. 139, 108892 . aller, J. G., Zaller, J. G., 2020. Pesticide impacts on the environment and humans. Daily poison: pesticides-an underestimated . danger, pp. 127–221 . hang, Y., Hou, D., O’Connor, D., Shen, Z., Shi, P., Ok, Y.S., Tsang, D.C., Wen, Y., Luo, M., 2019. Lead contamination in Chinese surface soils: source identification, spatial-temporal distribution and associated health risks. Crit. Rev. Environ. Sci. Technol. 49 (15), 1386–1423 . hu, H., Liu, X., Wang, Q., Zhang, B., Xu, C., Wang, Z., Chen, H., 2023. Heavy metals pollution of soil in central plains urban agglomeration (CPUA), China: human health risk assessment based on Monte Carlo simulation. Environ. Geochem. Health 45 (11), 8063–8079 . http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0050 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0051 http://refhub.elsevier.com/S2772-8838(25)00046-9/opt86xlA367Ab http://refhub.elsevier.com/S2772-8838(25)00046-9/optfhBsXpNdzO http://refhub.elsevier.com/S2772-8838(25)00046-9/optNGTxXtstGP http://refhub.elsevier.com/S2772-8838(25)00046-9/optNnFXk7fSeW http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0015a http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0053 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0054 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0055 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0057 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0061 http://refhub.elsevier.com/S2772-8838(25)00046-9/optLEQfi0RxvJ http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0064 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0065 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0066 http://refhub.elsevier.com/S2772-8838(25)00046-9/optJG9LO4xldP http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0067 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0068 http://refhub.elsevier.com/S2772-8838(25)00046-9/optC6pFfsjrKo http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0070 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0071 http://refhub.elsevier.com/S2772-8838(25)00046-9/opt0UPqnHCjHa http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0072 http://refhub.elsevier.com/S2772-8838(25)00046-9/sbref0073 Source apportionment of potentially toxic elements in soils from an urbanising region: Insights from multivariate analysis in Singida, Tanzania 1 Introduction 2 Materials and methods 2.1 Area of study 2.2 Soil sampling 2.3 Analytical procedures 2.4 Assessment of PTE pollution in the soil 2.5 Statistical analysis 2.6 Self-organising map (SOM) 2.7 Positive matrix factorisation (PMF) 3 Results and discussions 3.1 Pollutants and elemental concentrations and spatial patterns 3.2 Relationships among the pollutant and trace elements 3.3 Potentially toxic element contamination in singida urban and non-urban areas 4 Conclusions Declaration of competing interest CRediT authorship contribution statement References