Baseline Concentrations of 11 Elements as a Function of Land uses in Surface Soils of the Katangese Copperbelt Area (D.R. Congo)

Biogeochemistry and Ecology of Tropical Soils and Ecosystems (BESET), Faculty of Agricultural Sciences, University of Lubumbashi, Lubumbashi B.P: 1825, DR. Congo Département des Sciences de l’Environnement (DES) Faculté des Sciences des Aliments et de l’Environnement (FSAE), Université Nouveaux Horizons, Lubumbashi, Haut-Katanga, DR. Congo Department of Renewable Natural Resources Management, Faculty of Agricultural Sciences, University of Lubumbashi, Lubumbashi B.P: 1825, DR. Congo Department of Chemical Engineering, University of Lubumbashi, Lubumbashi, Democratic Republic of the Congo


Introduction
Ecosystem pollution with metals is of great concern due to the inherent adverse effects on living organisms including humans (Herselman, 2007). In the prospect to preventing humans' exposure, assessing the environmental quality of soils represents a crucial step, for which understanding the patterns of trace elements concentrations is of critical importance (Albright, 1998).
The behaviour of metals in soils is constrained by various geochemical processes and pH is one of the most important factors determining the forms of metal occurrence and influencing metal solubility and speciation and thus toxicity (Temminghoff et al., 1998). The first approach for soil contamination evaluation is the determination of the total available trace element content (Ballesta et al., 2010). In this regard, the natural background concentrations of trace elements are suggested as a reference to identify polluted soils and eventually to propose a plan for their proper management (Herselman, 2007;Ballesta et al., 2010).
Referred to as the natural range of concentration prior to any anthropogenic influence, the concept of geochemical background concentrations is scarcely considered however, and related data are quasi-inexistant for most environments. Conventionally, 95% of the expected range of background concentrations is designated as baseline concentrations, which is most often considered as reference to evaluate soil contamination levels (Herselman, 2007).
Several sources affecting the natural background concentrations of metals in soils are listed including mining, agricultural and other industrial activities (Baize and Sterckeman, 2001;Zhao et al., 2007). In the The Katangese Copperbelt Area (KCA), located south-eastern of the Democratic Republic of Congo, significant deposits of Copper (Cu), Cobalt (Co) and other metals are found . Hence, most environmental studies have revealed widespread soil pollution in the region (Banza et al.,2009;Lubalega et al., 2015;Katemo et al., 2010;Mpundu et al. 2013). In most cases, these studies expressed the metal contents in soil referring to the so-called universal concentrations of trace elements. A reference study by Ngoy et al. (2018) provided an interesting overview of metal contents in the region, but did not present the related baseline concentration. Moreover, none of these studies established the lower and upper limits of elements contents based on the baseline calculation that can be used as soil quality reference values for the KCA. This makes it difficult to adequately establish quality standards to detect and monitor sites affected by metals contamination in the region. Yet the concentrations of trace elements in soils depends on the mineralogical composition of the parent material and is influenced by weathering processes; therefore, it appears inappropriate to use universal reference concentrations (Ballesta et al., 2010;Tack et al., 1997). Alternatively, a more appropriate approach would imply establishing baseline concentrations on a regional basis to be used as a reference in the identification of contaminated soils (Herselman, 2007). Unfortunately, no data on baseline concentrations of trace elements exist for the Katangese Copperbelt Area (KCA).
The present study was initiated to (i) establish baseline concentrations of 11 potentially toxic trace elements (i.e., Mn, Zn, Cu, Co, Cr, Pb, Cd, Ti, Ni, Al, Fe) in soils of the Katangese Copperbelt Area (KCA); and (ii) investigate their variation under 3 different land uses (croplands, forest and mining areas). We collected topsoils (0-20) across different land uses and measured element content using the Inductively coupled plasma -optical emission spectrometry. The results of this research can be used as soil quality reference values in assessing anthropogenic vs. natural levels of trace elements in the KCA.

Study Area
This study was conducted in the KCA extending from Sakania to Kolwezi (Haut-Katanga and Lualaba provinces, D.R. Congo) (Fig. 1). Climate of the KCA is humid subtropical (Köppen climate classification Cwa) with one dry season (May to September) and one wet season (November to April). Annual mean rainfall is about 1300 mm, the majority falling during the wet season. The temperature fluctuates from 15-17°C (early dry season) to 31-33°C (late dry season), with the mean annual temperature averaging 20°C (Leblanc and Malaisse, 1978;Saad et al., 2012). The bedrock of the KCA consists of ~10,000 m thick sedimentary succession that belongs to Neoproterozoic Katangan Supergroup (Kampunzu et al., 2000;Kampunzu et al., 2009). Based on the regional occurrence of two diamictites (the 'Grand Conglomérat' and 'Petit Conglomérat'), the Katangan sedimentary succession is subdivided in three lithostratigraphic groups i.e., from base to top, the Roan, Nguba and Kundelungu Groups (Cailteux et al., 2005;Mujinya et al., 2014). The Roan Group comprises siliciclastic and dolomitic rocks, whereas the Nguba and Kundelungu Groups largely consist of siliciclastic rocks, with one major carbonate unit occurring as part of the Nguba Group (Kakontwe Limestone) (Batumike et al., 2007;Mujinya et al., 2014). The primary vegetation of the region is miombo woodland (Malaisse, 1974). Strong anthropogenic degradation of the vegetation resulted in woodland being substituted by a secondary grass savannah in the peri-urban zones of the KCA (Malaisse, 1973).

Sample Collection and Laboratory Analyses
Surface soils (0-20 cm) used in this study were collected in Lubumbashi and its surroundings between 2015 and 2021. The sampling area covers 3 land uses with a total of 490 samples ( Fig. 1): Agricultural land (450 samples), mining sites (24 samples) and miombo woodland (16 samples).
Analyses were performed on air-dried soil fractions (<2 mm). The soil pH was measured potentiometrically in a 1:2.5 (W/V) suspension of 1 M KCl. Total trace element contents were measured using the Inductively Coupled Plasma-Optical Emission Spectrometry (ICP-OES: Optima 8300 078S1509233; detection limits: 0.5 -10 ppm). Briefly, the trace elements analyzed in the soil samples were: Mn, Zn, Cu, Co, Cr, Pb, Cd, Ti, Ni, Al and Fe. Digestion of the sample was performed by wet extraction with EDTA at pH 4.65 following the protocol described in the soil analysis manual (Van Ranst et al., 1999). All analyses were performed in triplicate. Initially, a wider set of trace elements was considered, of which 11 were retained based on their threshold for detection by the spectrometer used for elements quantification.

Data Processing and Statistics
Principal Component Analysis (PCA) was performed using the package Facto Miner in R software (4.0.1) to characterize studied land uses based on measured elements (Team, 2013). The Arithmetic Mean (AM), Arithmetic Standard Deviation (ASD), Geometric Mean (GM), Geometric Standard Deviation (GSD), minimum, maximum and median were calculated to compare trace element contents and the pH across land uses. The AM and ASD are best estimates of geochemical abundance of an element, whereas the GM and GSD are better maximum likelihood estimators for most geochemical data. Baseline concentrations of the 11 trace elements were calculated using GM/GSD 2 and GM x GSD 2 of the samples, respectively as the lower and the upper limit of the established baseline (Dudka and Markert, 1992). To understand relationships between the pH and studied trace elements, we transformed our data using best Normalize package in R to meet the normal distribution requirement of data before calculation of Pearson correlation coefficient (Webster and Lark, 2019).

Element Contents and pH Across Land uses
The Principal Component Analysis (ACP) explained over 55% of variability within the dataset (Fig. 2). The largest variability in metal contents was observed in croplands. The PCA results in two groups of metals positively correlated with each other. The first group consisted of Cd, Ti, Fe and Zn, while the second group consisted of Ni, Co, Cu, Mn and Cr. The pH was genitively correlated with Pb along the dimension 1. Al was strongly associated with samples from forests and mining areas.
The pH in cropland varied between 3.9 and 7.3 ( Table 1). The mean pH in investigated cropland was 5.3 and 5.4 for arithmetic and geometric mean respectively. The minimum Al and Fe concentrations were 16.6 and 10.7 mg. kg -1 respectively, whereas their respective maximum concentrations were 282.9 and 215.7 mg.kg -1 . Al, Fe and Mn were the most abundant elements in studied cropland soils, with respectively 63.4, 42.2 and 57.6 mg. kg -1 as average concentration in soil. The mean Cu and Co concentrations were 5.4 and 1.0 mg.kg -1 , respectively. The following elements were found in trace concentrations (maximum concentration <5 mg.kg -1 ), Cd, Cr, Pb, Zn and Ni.
In forests, pH varied between 4.0 and 5.2 with an average of 4.3 ( Table 2). The most abundant elements were Al, Fe and Mn, with respectively 222.2, 162.6 and 163.0 mg.kg -1 on average. Mean Cu concentration was higher (14.6 mg.kg -1 ) than that of Co (2.7 mg.kg -1 ). Up to 31.4 mg. kg -1 Cu and 4.5 mg. kg -1 Co were found. The maximum concentrations in Cd, Cr, Ni, Pb, Zn and Ti were lower than 10 mg.kg -1 . In contrast, croplands presented Cr soil contents up to a maximum of 44.9 mg.kg -1 with an average of 6.1 mg.kg -1 .
In the mining area, pH varied between 4.2 and 6.8 with an average of 5.8. Cu was the most abundant metal, with a maximum of 1828.0 mg.kg-1 and an average of 424 mg.kg-1 (Table 3). Following Cu, Zn concentrations were found up to a maximum of 1410.0 mg. kg -1 and an average of 408.2 mg.kg -1 . Mean Al, Fe and Mn concentrations were respectively 60.5, 105.3, 32.4 mg. kg -1 . Cr, Ni and Ti were found in lower concentrations, with a maximum of 1.0, 1.7, 1.2 mg.kg -1 respectively.

Baseline Concentrations of Elements and pH Across Different Land uses
The lower and upper limits of pH in croplands were 3.8-8.0 (Fig. 3). The minimum expectable Al content (baseline) in cropland was 18.4 mg.kg -1 while the maximum was 162 mg. kg -1 . Minimum and maximum baseline Fe concentrations was 4.7-234.0 mg.kg -1 . Baseline Mn concentrations varied between 3.5-576.0 mg.kg -1 . The minimum baseline Cu and Co concentrations were 0.1 and 0.8 mg.kg -1 while the maximum was 4.0 and 18.0 mg.kg -1 .
In the forests, minimum-maximum baseline pH was 3.6-5.0 (Fig. 4). Minimum baseline Al and Fe concentrations were respectively 18.8 and 86.4 mg.kg -1 while the maximum baseline concentrations were 1167 and 283 mg.kg -1 . Minimum and maximum baseline Mn concentrations varied between 4.9 and 1539 mg.kg -1 . Obtained baseline Cu concentrations was 3.6-43.0 mg.kg -1 while that of Co was 0.2-18.0 mg.kg -1 .

Correlations between pH and Studied Elements
In croplands, pH was negatively correlated with Co, Fe and Zn but positively correlated with Cu, Al and Zn. However, none of these correlations was statistically significant (Fig. 6 a, d, g). In mining areas, pH was negatively correlated with Al while positively significantly correlated with Fe and Zn. A positive but non-statistically significant correlations of pH with Cu and Co were observed. In the forests, pH was negatively and significantly correlated with Mn and Co, but positively and significantly correlated with Al.

Discussion
Soils across investigated land uses were acidic with a mean pH lower than 6, which is in accordance with previous studies in the regions (Kasongo et al., 2013). However, the baseline pH in studied land uses varied from acidic to alkaline soils, with 3.8-8.0 for croplands, 3.6-5.0 for forests and 4.5-7.0 for mining areas (Fig. 4-5). These results suggest that although studied soils are globally characterized as acidic, the pH at some locations can be neutral or alkaline as observed for mining areas and croplands respectively. Surprisingly, soils in the mining areas and croplands showed higher pH compared to forests, which would come from the difference between soil units. The possibility that the use of lime in mining areas during waste treatment and cropland could be raised to explain the elevated pH. Additionally, we cannot rule out the possibility that organic matter decomposition and the inherent organic acidic production in forests soils may lower the pH. Compared to other land uses, croplands showed the largest range from lower and upper limit of reported pH baseline (3.8-8.0). This is an indication that pH in croplands can be largely variable as a result of differences in soil management and cropping systems. Though, the variation may simply be due to differences in soil units. Cropland

Mining area Forest
More interestingly, results delineated a considerable variability of element concentrations across different land use allocations (Table 1-3). The most abundant elements in croplands were Al, Fe and Mn with 63.4, 42.2 and 57.6 mg.kg -1 on average respectively. One of the typical characteristics of tropical soils is their marked alteration with rather lower pH values (Upadhyay and Raghubanshi, 2020). Coincidentally, Al, Fe and Mn were also the most abundant elements in forests, with values averaging 222.2, 162.6 and 163.0 mg.kg -1 respectively. In mining areas on contrary, soils were dominated by Cu (424 mg.kg -1 ) and Zn (408.2 mg.kg -1 ). This accumulation could be linked to the complexation of these cations with organic acids in forest soils.
We suppose that these element concentrations do not reflect their natural abundance since ongoing human activities related to agriculture and mining are likely to alter natural distribution and concentration patterns of metals in comparison to their concentrations in the underlying parent materials. However, baseline element concentrations presented here, reflect their common occurrence in the study area and can thus be used as reference values for soil quality. The high concentrations of trace elements in the soils of different sites in the Katangese Copperbelt area have already been reported by several authors (Malaisse, 1999;Mpundu et al., 2013). The upper limits of Cu and Co concentrations in forest soils were higher compared to cropland soils. This is an indication that in the Katangese Copperbelt area, even non-disturbed areas can contain significantly higher Cu and Co without a direct human activity. Nevertheless, Cu, Co, Pb and Zn contents found in forest and cropland soils were significantly lower than those reported by Ngoy et al. (2018) as upper limits of their natural geochemical background in the Lubumbashi region (400, 50, 120 and 200 mg.kg-1 for total Cu, Co, Pb and Zn respectively). Moreover, the ranges of trace element concentrations reported in this study are significantly wider than those reported in several studies carried out in other regions (Kaminski and Landsberger, 2000;Tarvainen and Kallio, 2002;Krishna and Govil, 2004;Massas et al., 2013;Paulette et al., 2015). These discrepancies could be related to differences in sample size, sampling strategies sampling areas and analytical methods. Indeed, contrast to Ngoy et al. (2018) who used data reported in studies that focused in potentially polluted areas with sample size lower than 50, in this study, 490 samples were used, thus covering a larger set of soils with potentially varied properties.
In the Katanga Copperbelt area in general and the Lubumbashi area in particular, mining activities have possibly modified natural metal contents in soils, making comparisons with international or regional reference values inappropriate for environmental studies. It was reported that in all cases in Lubumbashi soils, total Cu, Co, Zn and Pb concentrations are higher in surface layers than the deeper ones (Ngoy et al., 2018). It is therefore possible that the element concentrations observed across land uses in this study are a result of combination of the soil geochemical background that is naturally rich in these elements (Mpundu, 2010;Kaya et al., 2015) and the intensification of anthropogenic activities (Gustin et al., 2008).
We observed that pH was negatively correlated with Co, Fe and Zn but positively correlated with Cu, Al and Zn in croplands. However, none of these correlations was statistically significant (Fig. 6 a, d, g). In mining areas, pH was negatively correlated with Al, but positively correlated with Fe and Zn. In the forests, pH was negatively and significantly correlated with Mn and Co, but positively and significantly correlated with Al. These relationships suggest a lack of a direct and strong pH-dependence of measured elements. As pH-KCl measured in this study reflects total soil acidity rather than the actual acidity, the concentrations of metals would likely be strongly related to pH-H2O. Nevertheless, our results showed contrasting and poor relationship of pH with metals depending on the land use. The lack of strong correlation between metals and pH was also observed by Tume et al. (2006) who argued that their observation was possibly due to the very wide range of pH, having little influence on elemental behaviour.

Conclusion
This study has provided crucial information regarding lower and upper limits of trace element concentrations expected in surface soils of the Katangese Copperbelt Area (KCA). The identified baseline data did not reflect the natural geochemical abundance of elements, suggesting a predominant influence of ongoing anthropogenic activities in the region (i.e., Agriculture and Mining) over natural processes. Overall, mining activities were found to induce the greatest influence on the baseline concentrations of elements, particularly for Cu and Zn. Comparatively, Agriculture generated mild effects on the natural background concentrations of metals, as a similar trend of Al and Mn distribution was recorded in croplands and forests. Results also provided indication of possible variability in natural background concentrations of metals in the region, with forests presenting significantly higher Cu and Co contents than croplands.

Data Availability Statement
Data used in this study can be obtained from the corresponding author upon reasonable request.