Introduction

Chile is a country located along the southwestern border of South America (Fig. 1A) and its main economic activity corresponds to copper mining, which nearly accounts for 13% of the national global domestic product (GDP; SERNAGEOMIN 2021). Following Spanish conquest of the national territory, anthropogenic activities were performed without proper care of the environment; it was only until 30 years ago (March 1994) that the first Chilean Environment Law (19,300) was enacted as an initial measure to control environmental pollution (Tapia et al. 2019).

Fig. 1
figure 1

A General location of Chile in South America; B Location of the capital city of Chile (RM, Santiago) and sacrifice zones of the country; C Industrial activity locations within Tocopilla. Small mining operations are related to operations performed by pirquineros (Spanish word referred to people who extract minerals using artisanal techniques). 1. Thermoelectric ash deposits; 2. Ex-Municipal Landfill; 3. First ash deposits; 4. Former Cu-processing Plant; 5. Unregistered Landfills; 6. Corpesca S.A.; 7. Tocopilla Benefactor Society (SBT); 8. Tocopilla Mining Company (CMT); 9. SQM potassium warehouse; 10. AES Gener Thermoelectric; 11. Engie S.A. Thermoelectric or CTT; 12. Lipesed Cu-refining company

Full size image

Historical anthropogenic activities performed in Chile have given rise to numerous environmental sacrifice zones throughout the country; although it is not a legal denomination, it has been used for years to describe highly contaminated zones (VID and CITRID 2020). These areas are segregated and stigmatized locations in which, despite the economic profits previously obtained, inhabitants live in poverty and do not receive the benefits related to those anthropogenic activities (de Souza 2021). Known sacrifice zones of Chile correspond to: Coronel in Southern Chile (Tume et al. 2021), Ventanas in Central Chile (Oyarzo-Miranda et al. 2020), Huasco (Pérez et al. 2019) in North-Central Chile, and Mejillones (Guiñez et al. 2015) and Tocopilla (Jorquera 2009) in Northern Chile (Fig. 1B).

The assessment of soil quality in Chile is fairly new, however, the following studies have been developed: (i) contaminated and/or industrialized areas such as Arica (Tume et al. 2018c) and Taltal (Reyes et al. 2020) in northern Chile, Talcahuano and Coronel in south-central Chile (Tume et al. 2018a, b, 2021), and also (ii) in schoolyard and playground soils (e.g., Rodríguez-Oroz et al. 2018; Tume et al. 2021; Vega et al. 2022). Despite historic mining activities (Jorquera 2009) and reported health issues (Viviani et al. 2021), the quality of Tocopilla soils has not been fully addressed yet or alleviated. Tocopilla is a coastal town located nearly 1600 km north of Santiago, the capital of Chile (Fig. 1B). This city is inhabited by 25,186 people in a total area of 4039 km2 (BCN 2020). Numerous anthropogenic activities are (and were) developed in Tocopilla; the most significant activities correspond to thermoelectric energy production, treatment and refining of copper ores (Lipesed), potassium transportation and storage (SQM), and industrial fisheries (Fig. 1C). Due to these intense current and historic activities, Tocopilla has been declared a particulate matter (PM) saturated zone, exceeding the annual guideline of 50 µg/m3 of breathable 10 µm PM (PM10; BCN 2007; DICTUC 2006). For instance, the average concentration of PM10 ranged between 60.8 and 117 µg/m3 in diverse areas of the city during a 5-week air quality assessment performed in 2006 (DICTUC 2006). Indeed, Tocopilla has shown significant sources of atmospheric pollution (Jorquera 2009) and its inhabitants have been affected by respiratory disease such as asthma and bronchiolitis (Viviani et al. 2021).

Given the historic anthropogenic setting of Tocopilla, a soil sampling campaign was performed in 2014 to obtain a geochemical baseline of potential contaminants related to mining activities (Cu, Zn and Pb) and coal thermoelectric residues (Co, Cr and Ni) in the top (2 cm) and deep (20 cm) portion of the soil. To complement this study and compare the results of 2014, topsoil samples of Tocopilla were obtained in concentric circles around the thermoelectric companies in 2020–2021. The results of both studies are compared and discussed to better understand the environmental issues related to inorganic contaminants in Tocopilla.

Study area

Physiography and geology

The city of Tocopilla resides on the physiographic unit known as the Coastal Plain (Fig. 2A) and above an emerging marine platform (Quezada et al. 2010). Tocopilla is limited to the east by the Coastal Cordillera (Fig. 2A), a N–S range of 25–45 km width, with elevations ranging between 1 and 2 km. The marine platform is located on the base of the western coastal cliff (~ 40° W) of the Coastal Cordillera (González et al. 2003; Quezada et al. 2010). Immediately to the east of the Coastal Cordillera, another physiographic feature known as the Central Depression is found (Fig. 2A).

Fig. 2
figure 2

A Physiography at Tocopilla latitude and main gorges; 1. Iquine Gorge; 2. Ancha Gorge; 3. Barril or Seca Gorge; 4. Tres Higueras Gorge; 5. de Tames Gorge. B Geology of the study area. C Compass rose, which was made using WRPLOT with Cerro Moreno Station (150 km south of Tocopilla) data from January to June 2021 (https://climatologia.meteochile.gob.cl/application/historico/datosDescarga/230)

Full size image

The Coastal Cordillera at the latitude of Tocopilla is constituted by basaltic-andesitic volcanic rocks of Lower to Upper Jurassic age which are intruded by dioritic to granitic magmatic bodies (184–147 Ma). Both the volcanic and intrusive outcrops constitute the Mesozoic basement of this area (López et al. 2018; Fig. 2B). The Mesozoic basement is covered by unconsolidated alluvial and colluvial deposits (González and Niemeyer 2005; Fig. 2B). Tocopilla is located above alluvial and colluvial deposits which cover the Mesozoic basement rocks (Quezada et al. 2010; Fig. 2B).

Regarding mineral ores, this region is associated with the Middle Jurassic to Lower Cretaceous metallogenic belts, which are mainly related to porphyry copper deposits and skarns, iron oxide copper gold (IOCG), and stratabound deposits (Sillitoe 2012, 2003; Fig. 2B).

Climate and hydrometeorological events

Tocopilla is located within an extremely arid zone of northern Chile, and following the Koppen classification, its climate corresponds to a cold desert (BWk; Beck et al. 2018; Kottek et al. 2006). This aridity is produced by: (i) the interaction of regional scale atmospheric systems (such as the South Pacific Anticyclone) related to atmospheric stability, and (ii) the Andes Cordillera, which acts as a barrier that blocks the entry of moisture from the Amazon and Atlantic; both factors inhibit the development of precipitation and facilitate arid conditions (Rutllant et al. 2003; Schulz et al. 2012). Average precipitation of Tocopilla between 01/01/2010 and 11/30/2020 was negligible (0.0046 mm/year) based on data from the National Water Directorate (DGA) of Chile (https://snia.mop.gob.cl/BNAConsultas/reportes).

In the area of study, the preferential daily and nightly wind directions are SW (Fig. 2C; 33.8% of the time), and wind speed ranges between 0.5 m s−1 and 2.1 m s−1. No wind conditions represent 3.43% of the day.

Another climatic characteristic of the area is the low elevation fog, known as camanchaca. It is a frequent characteristic of the coast, producing a significant relative humidity (~ 60%; Algoritmos 2019) and low-homogeneous temperatures that range from 12° to 19 °C and 16° to 27 °C in the winter and summer, respectively (Aguas Antofagasta and GHD 2015). The camanchaca is halted by the Coastal Cordillera and occasionally is related to low intensity drizzles (Luebert et al. 2007; Rutllant et al. 1998). In addition, the coastal desert of northern Chile is sporadically affected by moderate to strong El Niño Southern Oscillation (ENSO) events that produce short and intense rains (Vargas et al. 2006, 2000). Due to its location and geomorphological features, moderate to heavy rain events are related to massive landslides. Significant debris flow events in Tocopilla have been registered during 1912, 1940, 1991, and 2015 which caused fatalities, damage to the infrastructure, and mining related economic issues (Galaz-Mandakovic 2018).

Hydrogeology and hydrologic systems

The study area is found in the Coastal Basin hydrographic zone along the boundary between Iquine Gorge (992 km2) and Ancha Gorge (1441 km2; Fig. 2A) sub-basins (Muñoz et al. 2014). These sub-basins present a pluvial exoreic regime and drain into the Pacific Ocean, with no evidence of perennial flows, but rather occasional runoff as a result of scarce precipitation (Ferrando 1992).

In the coastal area, rocks have very low permeability and there are no available studies indicating the presence of a freshwater aquifer in Tocopilla, therefore it has been assumed that this resource is largely absent in the city (DGA 1986). There are some punctual exceptions at 70 and 140 km south of Tocopilla, where evidence indicates the existence of groundwater related to secondary porosity of volcanic rocks (Herrera et al. 2018; Herrera and Custodio 2014). Additionally, the Coastal Cordillera operates as a barrier preventing the groundwater flow to the west; therefore, potential hydrogeologic connections between the Central Depression and the Coastal Cordillera are scarce and occur in areas cut by significant faults at the Loa River outlet (75 km north of Tocopilla), or at Las Vertientes spring in La Negra Gorge (175 km south of Tocopilla; Herrera et al. 2018).

Anthropogenic activities

Numerous anthropogenic activities have been present in Tocopilla during its history, the most significant of which correspond to: mining and ore processing companies, thermoelectric companies, fertilizer companies, and fisheries (Fig. 1C). These activities have been installed within the city, close to schools, health centers, and parks.

Mining activities and ore processing companies. At the beginning of the twentieth century, the Cu concentrate plant and smelter called Tocopilla Benefactor Society (SBT) was operating (Fig. 1C). After its closure around the 1920s, this area was known as “El Escorial” (slag) and it currently operates as a residential zone (Galaz-Mandakovic 2018). In contrast, the Tocopilla Mining Company (CMT; Fig. 1C) started its operations in 1918 on the grounds of a former Cu smelting company and began treating high grade Cu ore. After the installation of a treatment plant in 1960, the CMT abandoned its facility in the urban area of Tocopilla (Fig. 1C). It was inferred from Galaz-Mandakovic (2018) that no remediation efforts were made to clean this area. In 1987, Lipesed S.A. construction started, an ore processing plant 4 km south of Tocopilla (Fig. 1C), and operations began in 1988. After 20 years, Lipesed S.A. finished its operations, and since 2018, the remaining pilled material has been transported to another processing plant (30 km south of Tocopilla).

Thermoelectric and energy generation companies. In 1915, the first thermoelectric plant currently known as Central Termoeléctrica Tocopilla (Engie or CTT; Fig. 1C) was installed in the urban area of Tocopilla to supply Chuquicamata, one of the largest and oldest copper deposits exploited in Chile (Galaz-Mandakovic 2021). This facility has used diverse raw materials (oil, bituminous coal, sub-bituminous coal, and petcoke) as fuel over time (Galaz-Mandakovic 2018). It currently produces 268 Mw by coal and 398 Mw by gas. In 1995, the Nueva Tocopilla Thermoelectric plant, currently known as AES Gener (Fig. 1C) began operating (Galaz-Mandakovic 2021); this thermoelectric plant uses coal as a fuel and produces 277 Mw. The ashes generated from the burning of fuels were deposited to the north of the city (Fig. 1C). In 2022 the CTT stopped using coal and presently its fuel only corresponds to gas.

Fertilizer companies. Since 1988, the SQM company has operated a potassium warehouse (Fig. 1C) intended for the distribution of fertilizers and storage of potassium sulfate and potassium salts, which at least until 2008 were stored in open areas (Jorquera 2009).

Fisheries. In 1959, the extractive activities of the Chilean fishery company known as Corpesca S.A. started (www.corpesca.cl; Fig. 1C). This company processed fish to produce fishmeal and fish oil. In 2012 the company closed its operations in Tocopilla (Galaz-Mandakovic 2018); currently the fisheries are only artisanal.

The historic and current anthropogenic activities of Tocopilla, which mainly include thermoelectric plants and mining companies, might be related to the presence of significant amounts of breathable PM10, exceeding for three consecutive years the annual guideline of 50 µg/m3 (BCN 2007). Considering this pollutant, the city was declared a PM10 saturated zone in 2007 (BCN 2007).

Work methodology

Sampling

Two sampling campaigns were performed in the soils of Tocopilla, one in 2014 and the other in 2020–2021 (Fig. 3A). A total of 88 superficial samples and 70 deep samples were taken, which were divided into 10 baseline superficial and deep samples in 2014 (Fig. 3B, C), 60 superficial and deep samples in 2014 (Fig. 3B), and 18 superficial samples in 2020–2021 (Fig. 3D). For clarity and simplification purposes, throughout the text the superficial sampling of 2014 is called S-2014, the deep sampling of 2014 is called D-2014, and the superficial sampling of 2020–2021 is called S-2020.

Fig. 3
figure 3

Sampling sites. A General sample locations; B Northern baseline samples; C Southern baseline samples; D Urban Tocopilla sampling sites of 2014 (S-2014 and D-2014) and 2020 (S-2020)

Full size image

The 2014 sampling campaign was exploratory and was performed between July and November 2014 in the upper 2 cm or superficial (S-2014; Supplementary Material 1A) and 20 cm deep (D-2014; Supplementary Material 1B) soils of Tocopilla. During this sampling campaign, 10 superficial and 10 deep soil samples representing the geology of supposedly unimpacted soils, were collected to the north (Fig. 3B) and south (Fig. 3C) of Tocopilla. These samples were obtained in areas protected from the wind to avoid anthropogenic influence coming from Tocopilla and were sampled following a N–S transect (Fig. 3A). The samples obtained to the north (8 superficial and 8 deep) represent soils associated with granitic rocks (northern baseline, Fig. 3B), and the ones obtained to the south (2 superficial and 2 deep) represent soils related to andesitic to basaltic volcanic rocks (southern baseline, Fig. 3C). From the city of Tocopilla, 60 superficial and 60 deep samples were collected (Fig. 3D). These samples were obtained from a random point on a 1 m2 surface, from a 400 m × 400 m regular grid (Fig. 3D). These samples were collected using previously acid-cleansed plastic shovels. Nearly 1 kg of soil was obtained for each sample and weas stored in plastic bags until analysis.

The S-2020 sampling campaign was carried out between November 2020 and April 2021. Due to the fact that during the 2014 sampling campaign it was observed that Tocopilla soils were relatively homogeneous in the upper 10 cm, the upper 5 cm were selected for sampling during the S-2020 campaign (Supplementary Material 1C). A total of 18 samples were taken at an approximate radius of 250 m, 500 m, 750 m, 1000 m, 2500 m, and 5000 m from the AES Gener Thermoelectric plant, as the central point of the sampling circumferences (Fig. 3D). The central point was chosen because this thermoelectric plant was associated with significant concentrations of Cu (between 1025 and 1401 mg/kg) and Zn (between 406 and 653 mg/kg) in the 2014 sampling campaign. For each radius, samples were taken in the north, east, and south cardinal points (Fig. 3D), where nearly 1 kg of soil samples were collected using a previously acid-cleansed plastic shovel or stainless-steel shovel (depending on the hardness of the soil). Each sample represented a composite of 3 vertices of an equilateral triangle within an area corresponding to 2 m in diameter, at each vertex 300 to 400 g of sample were taken from the uppermost 5 cm of the urban soil.

Physicochemical characteristics of soils

In-situ soil properties

For the S-2020 sampling campaign, pH, temperature, conductivity, and oxidation–reduction potential were measured for each soil sample. These parameters were quantified in a soil sample diluted in distilled water (1 soil: 1 water) using a field multiparameter (HANNA HI-98194).

Metal concentrations

Prior to analyses, the composite soil samples were dried, homogenized with an agate mortar, and divided into two portions using a stainless-steel sample cutter. For the S-2014 and D-2014 sampling campaign, the wet soil samples were dried using an oven at 60 °C and were sieved to a size lower than 75 µm applying a shaker and stainless-steel sieve, following the protocols of the Chilean National Survey of Geology and Mining (Lacassie et al. 2014). For the S-2020 sampling, the wet soil samples were dried using an oven at 50 °C and were sieved to a size less than 63 µm using a shaker and stainless-steel sieve. The decrease in drying temperature was made to avoid changes in the composition of organic matter (OM), although in the Atacama Desert, the maximum accumulation of OM occurs between 40 and 150 cm depth (Mörchen et al. 2019) which is below the sampling depth. This implies that no significant changes in the soil composition should be observed. The change in sieving size was done to consider the potential contaminants related to the clay-silt size, which tends to concentrate metallic contaminants (Zhang et al. 2013).

The soils obtained during the S-2014, D-2014, and S-2020 sampling campaigns were digested using reverse aqua regia i.e., 4 mL HCl + 12 mL HNO3 + 300 mg of soil sample. These digestions were performed using a hot plate at 100 °C for 1 h in the CEITSAZA Laboratory at Universidad Católica del Norte (UCN). The digested residue was quantified for cobalt (Co), chromium (Cr), nickel (Ni), copper (Cu), zinc (Zn), and lead (Pb). In addition, iron (Fe) was analyzed to normalize other element concentrations. The quantification was done using a Perkin Elmer OPTIMA 7000 DV ICP-OES at the CEITSAZA Laboratory. Quality control (QC) and quality assurance (QA) of the instrument were checked by blanks and calibration curves obtained using a multielement solution measured over the analytical time. To study the QA/QC of the analyses, 23% of the total samples were analyzed in duplicates to check precision, and a certified soil material (ERM-CC141) was run every 25 samples to check accuracy. The detection (DL) and quantification (QL) limits were obtained according to the IUPAC recommendations (Danzer and Currie 1998). All of the studied elements presented a good repeatability calculated between the sample and the duplicate (R2 ≥ 0.98; Supplementary Material 2), in addition, the analyses of the certified material were in general within 10% when values were above the quantification limit (Table 1). One exception was the low recovery of Cu from the certified material likely due to the acid reagents used for the digestion. Previous reports indicated that the partial digestion of Cu is related to the incomplete breakdown of oxides, silicates or phosphates to which Cu could be bound (Lo and Sakamoto 2005).

Table 1 Element detection limit (DL), quantification limit (QL), duplicate correlations (DC), certified values (CV) obtained in this study, and certified values (ERM-CC141)
Full size table

Statistics

The univariate statistics and numerical indices were calculated with the raw data (Supplementary Material 1). To obtain principal component analyses and interpolation maps the values below the detection limit were recalculated as the DL divided by 2 (e.g., Tapia et al. 2018a).

Basic statistics

Basic statistics such as mean, absolute and relative standard deviation, median, median absolute deviation (MAD), minimum, maximum, and percentiles (5, 25, 50, 75, and 95) were calculated. For graphic visualization of certain statistical values, histograms and boxplots were made using OriginPro 2021. For boxplots, the concentrations 1.5 times higher than the third quartile were considered outliers (e.g., Aggarwal 2017).

Element correlations were obtained using the Spearman correlation coefficient (r), which gives more robust results than Pearson when datasets exhibit outliers (Spearman 1961; Zimmerman and Williams 1997). The correlations significance was addressed considering the following absolute values: very high (r > 0.8); high (0.6 < r ≤ 0.8); moderate (0.4 < r ≤ 0.6), and low (r ≤ 0.4; Williams 1986).

Principal component analysis

The principal component analysis (PCA) is a statistical multivariate method aiming at explaining most of the variance of a dataset with the minimum number of correlated variables (Anderson 2003). In this study, the PCA was used to identify correlated groups of elements. Since the data distribution presents positive asymmetries, before the PCA analysis, the concentrations were transformed using the centered log ratio (clr; Eq. 1; Bern et al. 2019; Reimann et al. 2011); due to the fact that the clr transformation is a less subjective method when compared to other log-transformations (Bern et al. 2019; Reimann et al. 2011).

$$clr \left(x\right)=log\left(\frac{x}{g(x)}\right)$$
(1)

In this transformation \(x\) represents the concentration of the element in the sample and \(g(x)\) represents the geometric mean of the dataset; for this transformation all the concentrations must be in the same unit (Bern et al. 2019). The PCA was carried out with standardized variables and varimax rotation (e.g. Mishra et al. 2019; Zhu et al. 2020). For the test determination of the suitability of PCA, the Kaiser–Mayer–Olkin (KMO) test and Barlett sphericity test were performed. For PCA appropriateness, KMO test must be greater than 0.5, while significance level must be < 0.05 for Bartlett's test (Vural and Cicek 2022).

Spatial and numerical analysis

Shallow versus deep soil

Trace metals are naturally contained in soils by rock weathering. Nevertheless, trace metals derived from anthropogenic activities tend to accumulate in the topsoil (Desaules 2012). The difference of concentration in different soil horizons have been studied in the past by the United States Geological Survey-USGS (Smith et al. 2014). The rate between superficial and deep soil horizons might aid in the constraining of anthropogenic and geogenic sourced elements, because it is assumed that metals of anthropogenic origin are deposited and accumulated mostly on the soil surface (e.g., Chiprés et al. 2009; Desaules 2012).

In this study, a rate between the concentration of the studied elements in superficial soil versus deep soil was performed (S-2014/D-2014). Following McIlwaine et al. (2017), rates above 1.5 were considered to present significant anthropogenic influence. Box-plots and interpolations (see the section "Distribution and interpolation maps") were performed to visualize the rates of each element.

Distribution and interpolation maps

Distribution maps were made using the Geostatistical Analyst extension of ESRI ArcMap 10.8.2 to visualize spatial variations in the elemental concentrations.

Due to a regular grid, interpolation maps using ordinary kriging were applied to the S-2014 and D-2014 datasets. In addition, interpolation maps using the values obtained between the ratio S-2014/D-2014 were performed. Before the interpolation, the concentrations were transformed using clr (Eq. 1). The clr transformation underestimated the most extreme outliers. For the S-2014 and D-2014 datasets, a natural break (Jenks) classification was used in the map legends.

Due to the concentric grid, no interpolations were made for the S-2020 dataset and metal concentrations are shown as crosses in the maps. To compare both samplings, classification limits shown in the maps are the same for the S-2014 and S-2020 datasets.

Numerical analysis

To infer the degree of contamination and the metal enrichment in the soils of Tocopilla the geoaccumulation index (Igeo; Muller 1979) and enrichment factor (EF; Zoller et al. 1974; Eq. 3) were calculated. The geoaccumulation index was calculated following Eq. 2.

$${I}_{{\text{geo}}}= {{\text{log}}}_{2} \frac{{C}_{n}}{1.5\times {B}_{n}}$$
(2)

Cn represents the concentration of the studied element in the sample and Bn corresponds to the concentration of the same element in a background material; a factor of 1.5 is applied to consider the lithological variability (Muller 1979; Nowrouzi and Pourkhabbaz 2014). Müller (1979) defined seven classes of geoaccumulation index ranging from class 0 (Igeo = 0, unpolluted) to class 6 (Igeo > 5, extremely polluted). The highest class (class 6) reflects at least a 100- fold enrichment factor above background values.

To calculate the enrichment factor, Eq. 3 was used.

$${\text{EF}}= \frac{{M}_{m}}{{({\text{Fe}},{\text{Al}},{\text{Ti}})}_{m}}/\frac{{M}_{b}}{{({\text{Fe}},{\text{Al}};{\text{Ti}})}_{b}}$$
(3)

Mm corresponds to the concentration of the element studied in the sample, (Fe, Al, Ti)m is the concentration of Fe, Al, or Ti in the sample, and Mb and (Fe, Al, Ti)b correspond to the concentration of the element and Fe, Al, or Ti in the selected background material. For this index, values close to 1 indicate no enrichment and that the soil was probably derived from the background material used for the calculation. Values of EF > 1.5 indicate that a significant portion of metal was not delivered from natural weathering processes, rather by other sources. On the other hand, EFs greater than 1 indicate that metals are enriched in the soil surface either from geogenic or anthropogenic sources (e.g. Nolting et al. 1999). These factors were categorized as follows: EF < 1 indicates no enrichment; EF < 3 indicates minor enrichment; EF = 3–5 indicates moderate enrichment; EF between 5 and 10 corresponds to moderately severe enrichment; EF between 10 and 25 indicates severe enrichment; EF values ranging from 25 to 50 are related to very severe enrichment; and EF above 50 is considered extremely severe enrichment (Chen et al. 2007 and references therein).

Both Igeo and EF depend on the background value used; some authors suggest that environmental indices should be quantified using regional backgrounds (Rubio et al. 2000). Given the close presence of metallic ore deposits in Tocopilla, different background values were considered: (i) local volcanic and intrusive rocks of the Coastal Cordillera near Tocopilla (Rogers and Hawkesworth 1989; Oliveros et al. 2007); (ii) soil samples collected to the north and south of Tocopilla (this study; Fig. 3); and (iii) the average metal concentrations of the upper continental crust (UCC; Hu and Gao 2008; Rudnick and Gao 2003). Iron was used as a conservative element (Table 2), which has been previously used in studies of Antofagasta, 170 km south of Tocopilla (Tapia et al. 2018b).

Table 2 Background concentrations for studied metals and conservative elements in soils. All concentrations are in mg/kg
Full size table

Results

Soil characteristics

In the studied zones, B horizons were not observed and the superficial materials show variable colors and textures, where they mostly correspond to a sand granulometry (Table 3). These soils show in general circumneutral pH, are oxidizing, and present variation in temperatures throughout the day. A summary of the measured physicochemical parameters is shown in Table 3.

Table 3 General characteristics of Tocopilla soils
Full size table

According to the taxonomic classification for soils of the United States Department of Agriculture (USDA; Soil Survey Staff 1999), the soils of Tocopilla correspond to aridisols. This type of soil is characterized by scarce to non-existent available humidity, preventing the growth of most plant species. In this soil type, the arid environment produces a disequilibrium between precipitation and evapotranspiration, resulting in an accumulation of salts (Soil Survey Staff 1999).

Univariate statistical analyses

A summary of statistical parameters obtained for the studied metals is shown in Table 4. For S-2014 and D-2014 datasets, the metals show a higher relative standard deviation (σ%) than S-2020, except for Cu which maintains similar σ%  for both sampling campaigns. In D-2014, the σ% for Pb and Zn was higher than 200%, while Ni, Cr, and Co show the lowest percentual standard deviations (except for Cr and Co from S-2014; Table 4).

Table 4 Summary of main univariate statistical parameters of Tocopilla soils
Full size table

Figure 4 shows boxplots and histograms of the studied metals at different depths and sampling campaigns. All elements in S-2020 show a narrower distribution range of concentrations and the lowest values are higher than those of S-2014 and D-2014. It was observed that Ni, Cr, and Co do not show significant changes in their concentrations during the sampling times and at different depths, while Pb, Zn, and to a lesser extent Cu show significant differences (Fig. 4). Average Pb concentrations for S-2020 are nearly three times higher than S-2014 and D-2014, whereas the average concentrations of superficial Zn and Cu (S-2014 and S-2020) exceed that of the deep soil (D-2014) where Zn concentrations are nearly double.

Fig. 4
figure 4

Box-plots and histograms for the studied metals in surface soil of 2014 (S-2014), 20 cm depth soil of 2014 (D-2014), and surface soil of 2020 (S-2020)

Full size image

For all of the studied elements and sampling campaigns, histograms show positive asymmetries (Fig. 4). Outlier values are more abundant in S-2014 and D-2014, and Cu and Cr exhibit outlier values in S-2020 (Fig. 4).

Bivariate and multivariate statistics

Spearman correlation coefficient

The Spearman correlation coefficients obtained for the studied metals in S-2014, D-2014, and S-2020 datasets are shown in Table 5. The S-2014 dataset shows a moderate correlation between Zn–Cu (r = 0.54) and a high correlation between Cr–Zn (r = 0.61). The D-2014 dataset indicates a very high correlation between Zn–Pb (r = 0.84), and moderate correlations between Cr–Ni (r = 0.59) and Cu–Zn (r = 0.52). As in D-2014, the S-2020 dataset exhibits a very high correlation between Zn–Pb (r = 0.84), while Cr–Ni and Co–Ni present high correlations (r = 0.65 and 0.71, respectively), and Ni–Pb present a moderate negative correlation (r = – 0.53). Only Pb and Zn (D-2014 and S-2020) present very high (> 0.8) correlation coefficients (Table 5).

Table 5 Spearman’s correlation coefficients
Full size table

Principal component analysis (PCA)

The results of the PCA made using rotated component matrix are shown in Table 6. For S-2014 and D-2014, 82% and 83% of the accumulated variance is explained by three principal components (PC1, PC2, and PC3), while for the S-2020, 91% of accumulated variance is explained by three principal components (PC1, PC2, and PC3; Supplementary Material 3). The KMO test is greater than 0.6 for D-2014 and S-2020, and is also higher than 0.5 for S-2014. Furthermore, Barlett’s test of sphericity indicated significance levels < 0.001 for all sampling campaigns (Supplementary Material 3).

Table 6 Rotated component matrix for different sampling campaigns
Full size table

In the PCA analysis, the following relations were found in the different datasets: (i) S-2014 Zn–Cr and to a lesser extent Cu to PC1, Co–Cu to PC2, and Ni–Pb to PC3; (ii) D-2014 Zn–Pb and to a lesser extent Cu to PC1, Ni–Cr to PC2, and Co–Cu to PC3; and (iii) S-2020 Zn–Pb to PC1, Ni–Cr and to a lesser extent Co to PC2, and Cu–Co to PC3 (Table 6).

Spatial analysis

The interpolated maps obtained for S-2014 and D-2014 datasets present similar patterns for Pb–Cu and lesser for Cr and Zn, these maps show the highest concentrations for Pb, Cu, and Cr in the central and northern areas of Tocopilla. In addition, Cu shows significant concentrations in the southern zone outside of the city in S-2014 (Fig. 5) and D-2014 (up to 9747 mg/kg; Supplementary Material 4). Nickel in the S-2014 dataset shows the highest concentrations in the northernmost and central areas of Tocopilla, whereas in the D-2014 dataset the highest concentrations are found in the downtown portion of the city. For the S-2014 and D-2014 datasets, Co exhibits little variability on its distribution, yet the southern and central (near AES Thermoelectric) areas show the highest values for S-2014 and D-2014, respectively. The distribution of S-2020 dataset is similar to S-2014 for Cu, Cr, Ni, and Co (Fig. 5); nevertheless, Ni and Cr concentrations are significantly higher in southern Tocopilla, while Pb concentrations are higher in close proximity to the port compared to 2014 (Fig. 5 and Supplementary Material 4).

Fig. 5
figure 5

Spatial distribution of metals in superficial soils of Tocopilla. Chromium (Cr); cobalt (Co); nickel (Ni); copper (Cu); zinc (Zn); lead (Pb). S-2014 concentrations are represented by interpolation using ordinary kriging. S-2020 concentrations are represented by crosses. Legend applies to both sampling campaigns. *Outlies concentrations in S-2014 were underestimated by ordinary kriging

Full size image

The ratio between S-2014/D-2014 was interpolated (Fig. 6). Most of the elements present higher superficial concentrations of soils when compared to the deep soils (Fig. 6). Of the studied elements, Co, Cr, and Ni show most of their rates below 1.5; Cu and Pb exhibit nearly half of their values above this rate, while Zn presents most of its rates above 1.5. High S-2014/D-2014 ratios of Cu were observed in the southern area of Tocopilla, while the highest ratios of Pb were observed in the northern and southern limits of the sampling area, whereas Ni showed higher rates in the northern area of the city (Fig. 6).

Fig. 6
figure 6

Top: Interpolation of S-2014/D-2014 using ordinary kriging. Chromium (Cr); cobalt (Co); nickel (Ni); copper (Cu); zinc (Zn); lead (Pb). Bottom: Boxplots showing the rate between the concentration of superficial soil (S-2014) and deep soil (D-2014). The dotted line shows the 1.5 value, which has been proposed as the limit of anthropogenic influence (McIlwaine et al. 2017); the straight lines represent the normal distribution of each element

Full size image

Geoaccumulation index and enrichment factor

Table 7A shows the results of Igeo calculated with different background materials. The elements Cu, Pb, and Zn show higher values of Igeo when compared to Ni, Cr, and Co. In fact, the latter elements were generally classified as non-soil contaminants. For Cu, Pb, and Zn, Igeo shows a significant variability depending on the background material used and can change from non-contaminant to strong contaminant (Table 7A). The D-2014 and S-2014 datasets show a similar contamination degree, although Cu and Zn exhibit lower and Pb higher values in D-2014. The contamination degree for Cu is similar during 2014 and 2020, whereas Pb and Zn increase their level in 2020 (Table 7A).

Table 7 (A) Average Geoaccumulation Index (Igeo) calculated with different background materials. (B) Average Enrichment Factor (EF) calculated using different background materials
Full size table

A summary of EFs is shown in Table 7B. Iron normalization shows relatively uniform values, which can be associated with fine solid surfaces and to the fact that its geochemistry is similar to that of many trace metals (Bhuiyan et al. 2010). In general, the decreasing order of enrichment obtained for soils of Tocopilla corresponds to: Cu > Pb or Zn > Ni > Cr > Co. Copper is the metal with the highest enrichment in soils for both years and depths, showing moderate to extremely severe enrichment (Table 7B). In general, EFs were lower when using a local baseline (Cu, Pb, Zn and Ni), and local volcanic rocks show lower indices than local intrusive rocks (Cu, Pb, and Zn).

Discussion

Physicochemical and geochemical environment of Tocopilla soils

The studied soils correspond to aridisols which are present under circumneutral pH and oxidizing conditions (Table 3). In addition, they are predominantly constituted by the sand fraction, whereas the clay-silt grain size presents proportions ranging from 0.8 to 11.1% (Table 3). The dominant grain size implies that the soils of Tocopilla are associated with a low adsorption/absorption and good infiltration capacity (Aprisal et al. 2019). However, aridisols are characterized by the accumulation and concentration of soluble and insoluble weathering products (Soil Survey Staff 1999); moreover, some authors indicated that in arid and hyper-arid zones, similar to Tocopilla, soils are a reservoir of chemical components and trace elements (Corradini et al. 2017). This suggests that soils of the area might concentrate and accumulate metals, which could be benefited from the circumneutral pH that prevents metal mobilization (Zhai et al. 2018).

Regarding rock geochemistry of the study area, volcanic and intrusive rocks near Tocopilla are characterized by an andesitic to basaltic composition. The main oxides are present in the following ranges: SiO2 51.1 to 56.4%, Al2O3, 14.6 to 18.6%, FeOtot 5.2 to 10.3%, MnO 0.1 to 0.3%, MgO 2.97 to 5%, CaO 2.9 to 10.3%, Na2O 3.3 to 4.8%, K2O 0.5 to 2.9%, TiO2 0.8 to 1.4%, P2O5 0.1 to 0.41% (Oliveros et al. 2007). In relation to trace elements, the Coastal Cordillera of Tocopilla contains important Cu deposits (Sillitoe 2003) which are usually hosted by La Negra Formation (Fig. 2B), and show increased natural concentrations of Cu (up to 348 mg/kg) and other elements such as Zn (up to 186 mg/kg; Oliveros et al. 2007) when compared to the UCC. Based on the spatial distribution of metal concentrations coupled with numerical indices such as Igeo and EF, the geochemistry of the studied trace elements was classified into two groups: (i) one related to Ni–Cr–Co, elements with a narrower range of values, lower standard deviations (Table 4), and lower indices (Table 7); and (ii) a second group related to Cu–Pb–Zn, which presents a wide range of concentrations, higher standard deviations (Table 4), and enhanced indices (Table 7). Multivariate statistics (e.g., Ni–Cr in PC2 of D-2014 dataset; Ni–Cr and lesser Co in PC2 of S-2020 dataset; Zn–Pb–Cu in PC1 of D-2014 dataset; Zn–Pb in the PC1 of S-2020 dataset; Table 6) and interpolation maps (Fig. 5) also indicate the existence of two groups in the soils of Tocopilla. The Cu–Pb–Zn group coincides with zones in close proximity to anthropogenic emissions, and the EFs ratio indicates that their concentrations cannot be explained only by the weathering of the rocks present in Tocopilla (Table 7B). The existence of positive outlier values in both groups (Fig. 4) suggests that natural concentrations are probably affected by external sources (Tume et al. 2018b).

Anthropogenic versus geogenic metals in soils of Tocopilla

In relation to the previously described groups, because of their distribution and numerical indices, Ni–Cr–Co and Cu–Pb–Zn were classified as mostly geogenic and anthropogenic in origin, respectively. In addition, the S-2014/D-2014 rate indicated that values elevated above 1.5 were present in the following (decreasing) order: Zn (67%) > Pb (52%) > Cu (48%) > Ni–Cr (37%) > Co (10%). This classification is supported by the fact that the Cu–Pb–Zn group presents nearly half or more than half of their values above 1.5 (McIlwaine et al. 2017).

Geogenically sourced elements (Cr–Co–Ni)

The lowest indices found for Cr, Co, and Ni suggested a general geogenic origin, although in some cases anthropogenic sources cannot be excluded. For instance, the highest values of Cr were found in the central area of Tocopilla (Fig. 5), a zone where mining activities were performed in the 1900s (Galaz-Mandakovic 2018). The highest values of Co are observed in areas where present and past mining operations were found (Fig. 5), which might be explained by relatively enhanced natural concentrations of Co (21–29 mg/kg; Oliveros et al. 2007) in the host rock of treated ores. In the case of Ni, the superficial soil concentration of the northernmost baseline (Fig. 3) is five times higher than deep soils (Table 2) where dominant SW wind directions (Fig. 2C) would allow for the transport of metal-rich material from Tocopilla to sectors far from its emission sources. Positive anomalies of Ni within the city were found in the fly ash deposits in northern Tocopilla (Figs. 1, 5, and 6). From 1983 to 2001, bituminous and subbituminous coal were the main fuel in thermoelectric plants of Tocopilla; subsequently (from 2001 to 2007), coal was replaced by petcoke (Galaz-Mandakovic 2018, 2021). Petcoke has Ni concentrations higher than coal (e.g., Hower et al. 2005; Langmyhr and Aadalen 1980) and this element tends to concentrate in fly ash during the combustion processes (e.g., Álvarez-Ayuso et al. 2006; Ibrahim 2015). Therefore, fly ash produced by petcoke burning might be related to higher Ni content. In addition, a ratio of 0.36 between Ni and V has been observed in PM of the city, which is characteristic of coal and residual oil combustion (Jorquera 2009). Accordingly, and although most of Tocopilla does not present Ni contamination, the Ni-enriched areas are probably related to thermoelectric plant residues and wind transport.

For Cr, Co, and Ni, important changes between the 2014 and 2020 concentrations were not observed (Fig. 4 and Table 4).

Anthropogenically sourced elements (Cu–Zn–Pb)

Copper, Zn, and Pb showed the highest indices and rates suggesting that their enrichment in soils is largely due to anthropogenic sources. Lower indices for these elements were obtained using local volcanics when compared to intrusives as background (Table 7); these lower indices might reflect Cu deposits hosted by volcanics in this zone (Zambra et al. 2007). Additionally, in the vicinity of Tocopilla, volcanics show enriched concentrations of Cu and Zn when compared to the UCC (Table 2).

Copper is an ubiquitous element of old Chilean mine tailings (Alcalde et al. 2018), which are present in Tocopilla (e.g., CMT). Contamination and enrichment of Cu in the soils of this city was obtained using different background materials (Table 7), suggesting that an important anthropogenic origin exists for this metal. Copper tends to adhere to soil particles better than other trace metals, lowering its mobility and increasing its accumulation in the soil (Alloway 2013). A previous environmental assessment made in particulate matter lower than 10 µm (PM10) and 2.5 µm (PM2.5) of Tocopilla established that the source of Cu in this environmental matrix corresponded to the Lipesed ore Cu processing plant (Fig. 1; Jorquera 2009). However, high Cu concentrations were not observed in the central portion of Tocopilla (as in this study; Fig. 5) which might be explained by the fact that Lipesed was active during the Jorquera (2009) assessment (while downtown emissions were inactive), not favoring the production of PM10 and PM2.5. Regarding Cu concentrations between 2014 and 2020, no important changes were observed (Fig. 4 and Table 4), which might be explained by mining activities and the remaining Cu from ancient mining.

In Tocopilla soils, the positive correlation between Zn and Pb (0.28 for S-2014, 0.75 for D-2014, and 0.84 for S-2020; Table 5) indicated a probable common origin. Despite presenting a lower contamination and enrichment degree in soils when compared to Cu, environmental indices for Zn and Pb suggest a significant anthropogenic origin (Table 7). Zinc and Pb concentrations in the central and northern Tocopilla overlapped with the Cu anomalies suggesting a potential relationship of both metals with mining activities (Fig. 5). Potentially toxic metals such as Zn and Pb are normally found in Cu-mine tailings of Chile and their concentrations were higher in historical mine tailings when compared to present time mine tailings (Rubinos et al. 2021). In addition, in the vicinity of mining activities these metals tend to persist in the soils after mine closure (e.g., Khalid et al. 2017; Wang et al. 2014). Both old tailings and metal persistence could explain the enhanced concentrations of Cu, Zn, and Pb nearby CMT at city downtown. Furthermore, significant concentrations of Pb (S-2014 and S-2020) in the northernmost part of the city could be related to fly ash from thermoelectric activities (Fig. 5). A previous study showed that contaminant elements such as Pb are preferably absorbed on the surface of fly ashes (e.g., Sharma and Tripathi 2008). Also, Jorquera (2009) established that coal and oil combustion were a source of Pb in PM2.5 in Tocopilla. Moreover, a significant correlation between Pb (0.78) and Ni (0.84) in PC3 (S-2014; Table 6) might suggest that fly ash contributes to increased Pb concentrations in the northernmost soils of the study area. Regarding Zn and Pb, the increased levels found in the last sampling was likely associated with the enactment in 2010 of the national law to reduce atmospheric pollution (PM10) in Tocopilla. While the emissions of thermoelectric companies such as AES Gener and Engie S.A. were considered to be the main sources of PM10 and they were limited, no measures were applied to specifically control Pb and Zn emissions (BCN 2010). In addition, increased concentrations might be related to the redistribution of these elements during hydrometeorological events that affected this city (e.g., 2015 and 2017 mudslides; see the section "Redistribution of metals in the soils of Tocopilla").

Redistribution of metals in the soils of Tocopilla

Heterogeneous soil profiles exhibit lithologic discontinuities and pedogenic processes that could redistribute metals (Desaules 2012). In Tocopilla, the soils are characterized by scarce to non-existent available humidity, and pedogenetic processes are expected to be minimal (Soil Survey Staff 1999). Nevertheless, in this particular area of the Atacama Desert heavy rains occur associated with ENSO events (Vargas et al. 2000); these hydrometeorological events increase the mobilization of trace elements from arid mountainous areas (such as the Coastal Cordillera, eastern limit of Tocopilla) to populated (flat) zones through debris flow (e.g., Copiapó landslide; Corradini et al. 2017). In Tocopilla, the outlet of gorges located in the central and northern portions of the city are typically affected by landslides (Galaz-Mandakovic 2018), which coincide with the highest concentrations of Pb and Zn in deep soils (Supplementary Material 4). Deep soils distribution might suggest historic debris flow events of Tocopilla were likely responsible of redistribution of soil trace elements content. In this regard, the use of the rate S-2014/D-2014 is not a good indicator of geogenic vs anthropogenic origin. However, S-2014/D-2014 rate may reflect the influence of modern anthropogenic activities on Cu, Pb, and Zn deposition since a ratio ≥ 1.5 (Fig. 6) indicates a significant accumulation in superficial rather than deep soils.

Additionally, metal redistribution occurred in the coastal areas. Most of the historic mining activities were developed on the shores and significant metal contents were found in beach sediments throughout the city (e.g., El Salitre, El Arenal, and El Remanso beaches; Figs. 1 and 6). Between 1960 and 1980, CMT (Fig. 1C) dumped tailings directly into the sea (Galaz-Mandakovic 2018, 2021 and references therein), staining the sands with black tones. Due to this historic contamination, El Salitre Beach (Fig. 1C) cleaning was completed in 2018. The contaminated sand was used to fill a geotube (highly resistant fabric), covered with rock-blocks to make a breakwater and attenuate the waves of this artificial beach. Despite the fact that the geotube is classified as a resistant material, its weathering may cause damage to marine ecosystems through the slow release of metallic components into the ocean (e.g., Carneiro et al. 2018).

Global comparisons to Tocopilla soils

As observed in the previous sections, the urban soils of Tocopilla were and are affected by anthropogenic contaminants. Table 8 shows different cities around the world where the metal content in soils were studied. In cities with high population density and tourism such as Bangkok (Thailand) and Hong Kong (Lee et al. 2006; Wilcke et al. 1998) in Asia, Galway (Ireland), Palermo, Naples and Salerno (Italy) in Europe (Cicchella et al. 2020; Imperato et al. 2003; Manta et al. 2002; Zhang 2006), or Baltimore (United States) in North America (Yesilonis et al. 2008), high Pb concentration was associated with vehicle transport, which also might contribute to Cu and Zn enrichment (e.g., Galway; Zhang 2006). In Dublin, Ireland, the topsoil Pb concentration was related to diffuse atmospheric emissions from traffic and Pb-rich (UK) coal combustion in homes and industries (Glennon et al. 2014). Despite the fact that leaded gasoline was banned on April 2001 in Chile (Tchernitchin et al. 2006), various industries are close to residential areas and the highway (Ruta 1; Fig. 1) which traverse Tocopilla; thus, both the traffic and close proximity to industrial activities could explain the increased concentrations of Pb in these urban soils. When comparing Tocopilla urban soils to another northern Chile cities, higher values of Pb were related to the storage of Pb concentrate in the port of Antofagasta (170 km south of Tocopilla; Tapia et al. 2018b) and the storage of hazardous toxic mining waste in Arica (575 km north of Tocopilla; the waste was imported by Promel Chile from the Swedish Boliden Metals); these materials were stored in suburbs of Arica between 1983 and 1985 where later a residential area was built and inhabitants were exposed for more than 10 years (Tchernitchin et al. 2006). Furthermore, in Bangkok, Hong Kong, Oslo (Norway), Naples, and Lisbon (Portugal) significant concentrations of Cu, Pb, and Zn in soils are found near different types of industrial activities (Cachada et al. 2013; Imperato et al. 2003; Lee et al. 2006; Tijhuis et al. 2002; Wilcke et al. 1998). At Tocopilla, Cu (and to a lesser extent Zn) concentrations are the highest when compared to other industrialized cities of Chile (Tume et al. 2019, 2018a, b, 2018c), China (Lee et al. 2006; Li et al. 2013), or Portugal (Cachada et al. 2013), suggesting that Cu mining emitted significant amounts of these metals.

Table 8 Published data of median concentrations of trace metals in superficial urban soils of different cities around the world
Full size table

Elements showing mostly a geogenic signature, such as Co and Cr, exhibited lower concentrations in Tocopilla when compared to soils of southern Chile (e.g., Talcahuano; Tume et al. 2018a) where more basaltic rocks are found along the coastal areas (Western Metamorphic Series; Hervé et al. 2017). Similarly to Athens, Greece, the presence of serpentinized ophiolitic outcrops were classified as a source of these metals (Argyraki and Kelepertzis 2014 and references therein).

Conclusions

The main conclusions of this research include the following:

  • Studied soils of Tocopilla present metals such as Co, Cr, and to a lesser extent Ni, which are mostly related to geogenic sources. On the contrary, Pb, Zn, and Cu are significantly sourced from anthropogenic contributions.

  • The weathering of mineral deposits of the Coastal Cordillera is believed to represent the main source of geogenic sourced elements.

  • Historic and current mining activities might be related to the highest concentrations of Cu, Zn, and Pb. In addition, thermoelectric plant residues and wind transport might be related to Ni enrichment and to a lesser degree Pb contamination.

  • Only Zn and Pb have increased their concentration between the 2014 and 2020 sampling campaigns.

  • The urban soils of Tocopilla present the highest concentrations of Cu (and to a lesser extent Zn) when compared to industrial, touristic, and densely populated cities worldwide.

The results of this research are concerning to the population of Tocopilla, which is chronically exposed to Cu, Zn, Pb, and to a lower degree Ni. In this regard, globally, environmental sacrifice zones with this level of contamination should not be designated as inhabitable locations. In particular, to safeguard the health of Tocopilla inhabitants, regulators must exert more control over inorganic contaminant emissions, improve mine closure of abandoned small scale mining operations, and ensure continuous monitoring of particulate matter and soils in this high-risk zone. Future studies should focus on health issues derived from the anthropogenic activities present in this city and main redistribution processes controlling metal enrichment and inorganic contaminants.