Critically raw materials as potential emerging environmental contaminants, their distribution patterns, risks and behaviour in floodplain soils contaminated by heavy metals

The expanding demand for new critical raw materials can lead to their increased release to the environment in the form of emerging environmental contaminants (EECs). However, there has never been a comprehensive study that takes into account the total EEC content, the content of various EEC fractions, their behaviour in floodplain soils, and potential ecological and human health risks. The occurrence, fractions, and influencing factors of the seven EECs (Li, Be, Sr, Ba, V, B, Se) originating from historical mining in floodplain soils of various ecosystems (arable lands, grasslands, riparian zones, contaminated sites) were investigated. Based on the evaluation of the overall levels of EECs (potentially toxic elements) in comparison to the soil guideline values set by European legislation for Be, Ba, V, B, and Se, it was found that only Be did not exceed the recommended limits. Among the elements analyzed, Li had the highest average contamination factor (CF) of 5.8, followed by Ba with 1.5 and B with 1.4. Particularly concerning was the discovery of a potential serious health risk associated with Li exposure for children, as indicated by hazard quotients ranging from 0.128 to 1.478. With the exception of Be and Se, the partitioning of the EECs into the different fractions revealed that the EECs are primarily bound with the residual fraction. Be (13.8%) had the highest percentage of exchangeable fraction as the most bioavailable in the first soil layer, followed by Sr (10.9%), Se (10.2%), Ba (10.0%), and B (2.9%). The most frequently observed correlations were between EEC fractions and pH/KCl, followed by soil organic carbon and manganese hydrous oxides. Variance analyses confirmed the impact of different ecosystems on EEC total content and fractions.

Sampling and field measurements. We collected a total of 42 soil samples from two layers (0-10 and 20-30 cm) from two contaminated sites (CS, classified as Ekranic Technosols), 1 reference site (RS, classified as Haplic Cambisol), and 12 alluvial sites (AS, classified as Haplic Fluvisols). The 12 AS were located along the continuous elevation gradient of the SR, from the highest elevation (659 m a.s.l.) down to an elevation of 123 m a.s.l. (Fig. 1, Table 1).
The sampling points were located on the river bank within the riparian zones (RZ) in the Štiavnické vrchy Mountains, and on arable land (AL) in the Podunajská pahorkatina Hills. In the case of a larger floodplain, we also collected samples distant from the water flow at the end of the floodplain, that were used as permanent grasslands (GL) in the Štiavnické vrchy Mountains, and as arable land in the Podunajská pahorkatina Hills (at 5 of 12 alluvial sites). Outside the river floodplain, we selected the two contaminated sites (CS), which are poorly developed on metallurgical slag and shallow soils. As CS sites, we selected the area of the Maximilian shaft that is registered in the Register of environmental burdens of the Slovak republic as probable environmental burden (PEB), and the hard rock mine tailing Lintich that is registered as a confirmed environmental burden (CEB), where hazardous substances caused by mining activities can pose a risk to human health or to the environment, soil and groundwater. The Maximilian shaft area was located approximately 1 km and the Lintich mine tailing approximately 100 m from the Štiavnica River.
The RS site represents a site with a 'normal' level of contaminants and arising from a combination of natural and diffuse pollution contributions. The RS site was used to determine the background values. The RS site was located about 400 m away from the Štiavnica River outside the floodplain on a slope and is used as a meadow. The soil type was classified as Haplic Cambisol.
Soil samples were collected in the study area in September 2018. Five subsamples were taken with a stainlesssteel blade, and then mixed thoroughly to get a composite soil sample. The collected soil samples were stored in a polyethylene bag, transported to the laboratory where they were air-dried at ambient temperature, crushed, and sieved through a 2-mm mesh.
The volumetric soil moisture (SM in %) and salinity (SAL in mS/m) were measured at the same time and places of soil samples collection with WET Sensor (type HH2, Delta-T Devices Ltd., Cambridge, UK), by taking the average of five measurements. Similarly, the redox potential (Eh) was measured by the Oxygen Diffusion meter (Ejkelkamp). The map was generated using 3D Map Generator, ver. 1.5, https:// www. 3d-map-gener ator. com/ 3d-map-gener ator-terra in/. www.nature.com/scientificreports/ Soil pollution assessment. We evaluated soil contamination by EECs using the contamination factor (CF) according to Varol 11 and the pollution load index (PLI) according to Tomlinson et al. 12 . The CF was calculated according to the formula: where Ci is the measured content of EEC in soil and Cb is the background value of the EEC content of the reference site. According to Varol 11 , CF < 1 means low contamination, 1-3 is moderate contamination, 3-6 is considerable contamination, and CF > 6 means very high contamination. The PLI was calculated according to the formula: where n is the number of elements. The PLI > 1 indicates pollution.
Human health risk assessment. Potential human health risk was also evaluated. In the study, ingestion of soils contaminated with EECs was considered as the main pathways for risk assessment. We calculated the average daily dose (ADD, mg element kg −1 bodyweight day −1 ) for children, adult male and adult females: where C is total soil element content; IR is soil ingestion rate (children: 200; adults: 100 mg dust day −  www.nature.com/scientificreports/ lowest observed adverse effect level (LOAEL) from the therapeutically used lithium of 2.1 mg/kg-day for adverse effects in several organs and systems. Dividing the LOAEL of 2.1 mg/kg-day by an uncertainty factor of 1000 yields a subchronic and chronic pRfD of 0.002 mg/kg-day 17 . We think that this value can be used for health risk assessment of soil because exposure through environmental compartments should be minimal to zero. Then we calculated hazard index (HI) as the sum of HQ values of all EECs. Values of HQ and HI of > 1 indicate high probability of the occurrence of adverse health effects. Statistical analysis. The study analyzed the data using the Pearson correlations, principal component analysis (PCA), one-way analysis of variance (ANOVA), and hierarchical cluster analysis (HCA). The SPSS Statistics 28 software was used.
Soil samples collection. In Slovakia, it is allowed to collect soil samples for research purposes.

Results
Soil properties, soil contamination indices and potential health risk assessment indices. The basic soil properties are reported in Table 2.
The total content of EECs in the two soil layers of the SR floodplain together with soil guideline values (SGV) as a range in European legislation 18 is presented in Table 3. At least one lower SGV limit set up for 5 selected EECs (Be, Ba, V, B, Se) was exceeded in the case of all selected EECs with the exception of Be. Some exceedances were recorded in the CS, AS and RS localities, in both layers, and in all ecosystem types (CS, AL, GL and RZ). For Ba and Se, the upper SGV limit was even exceeded, but only in one locality. The contamination risk in the CS and AS localities was confirmed by the CF and PLI along the whole SR floodplain for 5 EECs (Li, Sr, Ba, V, B) ( Table 4). The CF and PLI were not calculated for 2 EECs (Be, Se) due to the appearance of zero values. We measured the lowest content of these two elements in soil with no SGV exceedance in the case of Be and only a few exceedances in the case of Se. In the CS localities, the average CF showed moderate contamination by Li. At AS localities, the average CF showed considerable contamination by Li, moderate contamination by Ba and B. The highest CF was quantified for Li (15.6 in 13AS-AL2), Ba (2.6 in 15AS-AL2) and B (2.3 in 8AS-RZ). In many AS localities, contamination by EECs was similar to or even higher than in CS localities. The PLI indicated that all AS sites were contaminated by EECs (PLI > 1) except one AS locality (3AS-RZ).
The potential health risk was assessed by considering oral topsoil ingestion for 7 EECs (Table 5). We discovered possible serious potential health risks only in children, a vulnerable population. The two adult groups are not under the potential health risk by EECs. In the two adult groups, the mean value of HQ for individual EECs ranged from 0.000 to 0.043 for males, and from 0.000 to 0.050 for females. The resultant mean HI values for female persons were 0.040 at CS localities and 0.080 at AS localities. The mean HI values for male persons were 0.034 at CS localities and 0.068 at AS localities. The mean value of HQ for individual EECs ranged from 0.002 to 0.545 for children. The resultant mean HI value for children was 0.435 at CS localities and 0.868 at AS localities. In children, the mean HQ (0.545) at AS localities was found high for Li. Lithium was followed by V (0.238), and Ba (0.070); all the rest of the EECs were much lower and ranged from 0.002 (Sr, B) to 0.005 (Be, Se). The risk for www.nature.com/scientificreports/ children indicated by HQ for Li was significant at 5 of 15 AS localities where HQ was found high (ranging from 1.085 to 1.478). Two of these 5 localities are used as grasslands, two as arable land and one is a riparian zone.
EEC fractionation. The content of seven EECs leached in the sequential extraction procedure in the CS, AS, and RS localities and in both depths is summarized in Fig. 2a-e. We recorded different distribution patterns among all four fractions between different elements and related chemical groups of EECs. Between the two different layers among the AS and RS sites, more similar patterns occurred at the AS sites. V and B showed the highest percentage in the residual fraction (IV) in all CS, AS, and RS sties, and in both depths. This fraction may be considered as biologically unavailable. On the other hand, among all the elements observed, only one element, Sr, showed the highest percentage in the first extractable fraction, which was considered the most mobile in the RS site at the second depth. The third (oxidisable) fraction occurred in the highest percentage in the cases of Li (at CS and in the first depth at RS), Ba (in the first layer at AS and in the second Table 3. Total content (T) of emerging environmental contaminants in the depth of 0-10 and 20-30 cm in contaminated, alluvial and reference sites with basic statistical characteristics compared to soil guideline values (mg/kg). AS alluvial site, CS contaminated site, RS reference site, RZ riparian zone, AL arable land, GL grassland, PEB probable environmental burden, CEB confirmed environmental burden, SGV soil guideline value according to Reimann et al. 18 . Normal letter-below soil guideline value; bold letter-above soil guideline value. Statistical values are in italics. Table 4. Contamination factor (CF) and pollution load index (PLI) for contaminated and alluvial sites in the depth of 0-10 cm. AS alluvial site, CS contaminated site, RS reference site, RZ riparian zone, AL arable land, GL grassland, PEB probable environmental burden, CEB confirmed environmental burden. The CF < 1 means low contamination, 1-3 is moderate contamination, 3-6 is considerable contamination, and CF > 6 means very high contamination.  Alkali earth metals (Be, Sr, and Ba) reflected the most equable distribution between fractions together with B and Se in some cases. The distribution of the alkali earth metals in the different fractions mostly associated with the first two most mobile fractions (I and II) indicated that the mobility declined on the order of Ba, Sr, and Be, at CS sites; Ba, Be, and Sr at AS sites; Be, Sr, and Ba at the RS site in the first depth and Sr, Be, and Ba in the second depth, respectively.
At AS localities in the first depth, the average content of Ba represented 10.0% in exchangeable, 22.5% in reducible, and 9.6% in oxidisable fraction; the content of V represented 0.2% in exchangeable, 2.7% in reducible, and 8.8% in oxidisable fraction; the content of B represented 2.9% in exchangeable, 1.6% in reducible, and 7.8% in oxidisable fraction; and the content of Se represented the same.
EEC fractions relationships with soil properties, ecosystem types, and soil depth. The correlation coefficient matrix for the EEC fractions with the chemical and physical parameters of the soil is shown in Table 6. The most numerous significant correlations with chemical properties occurred in the case of B (17), while the least occurred in the case of Li (6). The most numerous significant correlations with physical properties occurred in the case of Sr (12)  The PCA showed that the first two factors can explain more than 43% of the total variance for all EECs, from 43.2% in the case of Be to 50.9 in the case of B (Fig. 3). Salinity and pH/H 2 O were positively and Eh negatively  To evaluate the significance of the difference between mean values of each EEC fraction, one-way ANOVA was used. The independent variables were two different ecosystem types (riparian zone and agricultural ecosystem represented by arable land and grassland) and soil depths (0-10 and 20-30 cm), while the dependent variables were the EEC fractions. The result of ANOVA test is presented in Table 7. The statistical analysis showed significant differences between the ecosystem types and all Ba fractions, including the total content. Significant differences appeared between the different ecosystem types for Li and Sr in fractions III, IV and total content,   www.nature.com/scientificreports/ for B and Se in fractions II and III, and for V in fraction I. The ANOVA showed significant differences between the soil depth and the EEC fractions only in the case of Li (I).
The HCA (Fig. 4) has highlighted two large clusters (A and B). Cluster A contains sites that are primarily used as agricultural land (GL and AL). Two clusters of lower hierarchical order are recognised in B (B1 and B2). Cluster B1 includes predominantly RZ sites (16 out of 19). Cluster B2 includes five sites, two of which are contaminated sites.

Discussion
Based on the available SGVs, we discovered four EECs (Ba, V, B, Se) contaminating the entire SR floodplain. The SGV value exceeded was not only reported in the case of Be. The SGVs were not available for Li and Sr. According to the calculated contamination indices for five EECs (Li, Sr, Ba, V, and B), Li is the most serious EEC of the SR floodplain, followed by Ba and B. Based on the calculated contamination indices for 5 EECs (Li, Sr, Ba, V, B), mainly Li and fewer Ba, B are the most serious EECs of the SR floodplain. Based on the content of selected EECs in soil and calculated contamination and pollution indices, many AS localities are more contaminated by selected EECs than the confirmed environmental burden (4CS-CEB). In addition, some of them are used intensively for agricultural production. Although the contents of the selected EECs exceeded some SGVs, the contamination indices indicate mainly moderate contamination. Therefore, they are probably not of interest for further industrial  Table 6. Pearson's correlation coefficient matrix of EECs in various fractions and chemical and physical properties of the soil (n = 42). I exchangeable fraction, II reducible fraction, III oxidisable fraction, IV residual fraction, T total content, Eh redox potential, SOC soil organic carbon, HA/FA ratio of humic acid carbon to fulvic acid carbon, o extracted by oxalate, SAL salinity, SM soil moisture; *r-values shown in bold are significant at P < 0.05; **r-values shown in bold are significant at P < 0.01; N.C. no correlation. www.nature.com/scientificreports/ use in the sense of soil mining. The demand for new raw materials is growing very quickly, and other raw or natural materials are already being used today for the extraction of CRM. In recent times, many CRMs are seldom mined as the main ore. They are usually coproducts of the principal ore. Therefore, historical mining regions may be of such interest 19 . Many tailings did not meet the requirements of further industrial production and were only piled up as waste 20 . In addition, tailing deposits have become potential contamination source 21 . Thus, they impacted the surrounding environmental compartments. The proof of this is the SR floodplain contaminated www.nature.com/scientificreports/ by heavy metals as reported by Kanianska et al. 22 and other pollutants. The HMs together with EECs released during mining processes were transported and spread along the floodplain. Rivers and alluvial soils are affected by discharges from mines and tailings 23 . In addition, polluted aquifers act as long-term pollution source to the surrounding gaining rivers and soils, even after the mining activities were stopped 24 . Despite the considerable scale, there are many such territories with limited data availability 25 . However, they can pose ecological and health risks with respect to the cumulative effect of possible additional contamination from existing and potential new pollution sources. In our study, Li exhibited the highest mean contamination factor with a value of 5.8, followed by Ba with 1.5 and B with 1.4. We also found possible potential health risk for children in the case of Li at 5 of 15 alluvial localities with HQ higher than 1. Different contaminants with different chemical properties can contribute to these risks and interact with each other under the simultaneous effect of the soil environment and its specific properties. Therefore, research on the most serious EEC behaviour in soil in relation to soil properties is one of the key factors to prevent or decrease ecological and health risks. The BCR SEP method allowed us to study the mobility and availability of EECs. This method is accepted by many researchers dealing with toxic metals in mining waste (e.g., Fernandez-Ondono et al. 26 ). Although there are authors who believe that the application of this procedure can lead to ambiguous and misleading interpretation 27 . This method is still being increasingly applied in EEC analyses (e.g. Chen et al. 28 , Mikavica et al. 29 , Xiong et al. 30 ). Table 7. Results of ANOVA for the EEC fractions by ecosystem type and soil depth (n = 40). P-values highlighted in bold are highly statistically significant (P < 0.001) and statistically significant (P < 0.05). www.nature.com/scientificreports/ Lithium and its behaviour in floodplain contaminated soils. Lithium does not belong to the essential nutrients group. But there is evidence that low Li levels have beneficial effects on living organisms 31 . Lithium possesses unique essential and neurophysiological characteristics in biota and humans 32 . However, at high doses, it can be toxic to humans 33 . The Li content in the soil is variable from place to place. Upper layers usually contain less Li than the underlying layers 34 , as markedly the case at the RS locality. In the AS localities, the total Li content was mostly higher in the first depth, but at 7 out of 19 localities it was the opposite. The average Li content in European agricultural soils is 11.4 mg/kg, with a predominantly low concentration (median 6.4 mg/kg) in northern Europe and higher values (median 15 mg/kg) in southern Europe 35 . We recorded a total Li content of 14.78 mg/kg in the RS locality and 29.29 mg/kg as a mean value in the CS localities. A significantly higher total Li content was recorded in AS localities used in agriculture with a mean value of 152.52 mg/kg compared to localities in the riparian zone with a mean value of 36.46 mg/kg. Such content can pose an environmental risk. In the CS localities, the average CF (2.0) showed moderate contamination by Li. At AS localities, the average CF (5.8) showed considerable contamination by Li.

F-value P-value F-value P-value
We also discovered serious potential health risks for children, a vulnerable population. The mean HQ (0.545) at AS localities was found to be high for Li. The risk for children indicated by HQ for Li was significant at 5 of 15 AS localities where HQ was found high (ranging from 1.085 to 1.478). Lithium toxicity results in abnormalities and dysfunctions in several metabolic pathways, causing serious problems for human health 3 .
The toxicity of EECs is also related to the distribution of their fractions. Li was heterogeneously distributed in fractions among different ecosystem types. Exchangeable and reducible fractions showed up in all land use types and in both layers with the lowest Li content, accounting for ˂ 1% Li, which is in line with the findings of other authors 36 . The residual fraction demonstrated the highest Li content at the AS localities in both depths and at the RS locality in the deeper layer. At the CS and in the first depth at the RS locality, Li demonstrated the highest content in the oxidisable fraction. This was confirmed by correlation analyses, where we found positive correlations between the reducible and oxidizable Li fractions and the SOC, similarly as with pH/KCl. Motowicka-Terelek et al. 37 also found correlations between soil Li content and SOC, but there are also authors who did not 36,38 . From the physical properties, the fine granulometric fraction (silt) showed a positive correlation with www.nature.com/scientificreports/ the residual fraction and the total Li content. Kabata-Pendias and Mukherjee 39 refer to the texture, respectively, the fine clay granulometric fraction as the most important factors controlling the Li status of soil. They consider SOC and pH to be factors of lesser importance.
Barium and its behaviour in floodplain contaminated soils. Barium is not an essential nutrient for humans. But it probably fulfils biochemical functions important for animals and plants 40 . The content of barium in food does not usually pose a health risk. Its long-term impact on human health is questionable 41 . Barium toxicity is related to the solubility of the compound 42 .
The general threshold and guideline values for Ba have not been developed. Only a few countries in the world have adopted them. The Ba limit for food safety has not yet been established 43 . Furthermore, geographic variation is little known 44 . Anthropogenic releases are primarily associated with industrial activities, including mining, and pesticide application in agriculture 45 . Barium compounds are quickly dissolved and can be widely spread, as was also the case with the SR.
In AS localities, the total Ba content was mostly higher in the first depth, but at 6 out of 19 localities, it was the opposite. The median content of Ba in European agricultural soil is 62 mg/kg, with predominantly low concentrations (median 43 mg/kg) in Northern Europe and higher values (median 74 mg/kg) in Southern Europe 18 . We recorded a total Ba content of 255.34 mg/kg in the RS locality and 66.91 mg/kg as a mean value in the CS localities. A higher total Ba content was recorded in AS localities with a mean value of 384.33 mg/kg (median value of 394.47 mg/kg), ranging from 108.34 to 659.82 mg/kg in the first soil layer. González-Valoys et al. 46  The oxidisable fraction demonstrated the lowest Ba content in all types of land use and in both layers, accounting for 4.14-10.81% Ba. The Ba is thought to be incapable of forming organic ligands or complexes with fulvic and humic acids 47 . This was also confirmed by correlation analysis, where we did not find correlations between the Ba fractions and the SOC. We recorded only negative correlations between the oxidisable Ba fraction and the HA/FA ratio. In the AS and RS localities, the residual fraction showed the highest content, accounting for 51.7 to 85.44% Ba. Ba was heterogeneously distributed in extractable and reducible fractions between different types of land use and soil layers. The reducible and extractable fractions showed the highest Ba content in CS localities. The extractable fraction demonstrated on average 10.0% Ba at AS localities in the first layer and 11.6% in the second layer. Cappuyns 44 found low Ba mobility in alluvial soils of the Schelter and Demer rivers in Flanders (˂ 3% of the total content). Ba adsorbs into Fe hydroxides 48 , and we also found a positive correlation between the reducible Ba fraction and Fe o . The pH of the soil can influence barium behaviour and is more mobile under acidic conditions. The pH showed negative correlations with the total Ba content, oxidisable fractions, and residual fractions. In contrast to Frohne et al. 49 , we did not find correlations with Eh. Similarly, we did not find correlations with clay particles, as was found by Atun and Bascetin 50 . The reducible and oxidisable Ba fractions were positively correlated with the content of the sand particles. Only the IV Ba fraction was positively correlated with the silt content and negatively correlated with the content of the sand particles.
Boron and its behaviour in floodplain contaminated soils. Boron at low concentrations is probably an essential element for plants 51 humans and animals, while it is toxic at high concentrations 52 . But recently, B was assumed to be rather toxic to plant cells at low levels 53 . Boron has an extremely narrow range between deficiency and toxicity 54 .
In AS localities, the total B content was mostly higher in the first depth, but at 6 out of 19 localities it was the opposite. The median content of B in European agricultural soil is 2.42 mg/kg, with a predominantly low concentration (median 1.7 mg/kg) in northern Europe and higher values (median 3 mg/kg) in southern Europe 18 . Near a large boron mine in Turkey, Alan and Kara 55 measured the total B content in soil from 36.5 to 820.5 mg/ kg. We recorded a total B content of 27.49 mg/kg at the RS locality and 25.56 mg/kg as the mean value at the CS localities. In AS localities, the mean value was 37.23 mg/kg (median value 32.46 mg/kg), ranging from 20.27 to 64.57 mg/kg in the first soil layer.
Although the total B content exceeds the soil guideline value in most AS localities, sequential extraction showed the highest percentage of residual fraction (IV) in all CS, AS and RS sties, and at both depths. On the other hand, the percentage of mobile and other fractions was low (2.9% in exchangeable, 1.6% in reducible and 7.8% in oxidisable fraction). Similarly, Alan and Kara 55 found the highest B content in the residual fraction. Therefore, most of the boron that occurs in the soil is unavailable to plants 56 .
We found the most correlations between the B fractions and the physicochemical properties. However, the Eh was negatively correlated with extractable and reducible fractions of B. The pH, SOC, and Mn o oxides were positively correlated with the extractable, reducible, and oxidisable fractions of B. The extractable B fraction was also positively correlated with HA/FA. Hou et al. 57 also found positive correlations between readily soluble B and pH and between the oxidisable fraction and SOC. From the physical properties, the extractable fraction correlated positively with the clay content of the soil in the study area.

Conclusions
We detected some level of EEC contamination, particularly by Li, Ba, and B in the floodplain soils of the SR. This contamination could originate simultaneously with HM contamination, mainly from historical mining. The downstream transport could be a source of EECs as accessories to HMs for distant soils. www.nature.com/scientificreports/ Sequential extraction technique was used to assess the environmental risk of EECs. The partitioning of the EECs showed that the EECs are mainly associated with the most stable (biologically unavailable) residual fraction, except for Be and Se. The risk of these two elements is reduced by their low (even zero) total contents in the soil. The exchangeable fraction as the most bioavailable occurred in the first soil layer at the AS localities in the highest percentage for Be (13.8%) followed by Sr (10.9%), Se (10.2%), Ba (10.0%) and B (2.9%). The Li and V showed the lowest percentage in the fraction (less than 0.3%). However, the potentially available fractions represented by the II and III fractions represented a higher percentage in the case of Ba (22.5% in reducible and 9.6% in the oxidizable fraction) and Li (25.5% in the oxidizable fraction).
The soil parameters were selectively correlated with the EEC fractions, and we most often observed correlations with pH/KCl, followed by SOC and manganese hydroxides.
The highest mean value of the contamination factor was calculated for Li (5.8), followed by Ba (1.5) and B (1.4). We also found a possible potential health risk for children in the case of Li (HQ ranged from 0.128 to 1.478). The results underscore the significance of recognizing the interconnections between environmental and human health. Such integrating framework of environmental risk assessment posed by EECs comprising chemical and ecotoxicological lines of evidence can improve decision-making processes and help to manage contaminated sites.