Research article Geochemistry: Exploration, Environment, Analysis

Published online April 24, 2019 [Link] | Vol. 19 | 2019 | pp. 431–447

Geochemical mapping and estimation of background

concentrations in soils of Carajás mineral province,
eastern Amazonian Craton, Brazil
Gabriel Negreiros Salomão1,2*, Roberto Dall’Agnol1,2, Rômulo S. Angélica1,
Marco Aurélio Figueiredo2,3, Prafulla K. Sahoo2, Carlos A. de Medeiros Filho4 &
Marlene F. da Costa5
1
Programa de Pós-graduação em Geologia e Geoquímica, Instituto de Geociências, Universidade Federal do Pará, Rua
Augusto Corrêa, 1, Belém 66075-110, PA, Brazil
2
Instituto Tecnológico Vale, Rua Boaventura da Silva, 955, Belém 66055-090, PA, Brazil
3
Terrativa, Rua Ministro Orozimbo Nonato, 102 Tore A Sala 705, Nova Lima 34000-000, MG, Brazil
4
VALE S/A – Exploração Mineral, Rua Grajáu, 63, Carajás, 68515-000 Parauapebas, PA, Brazil
5
Gerência de Meio Ambiente – Minas de Carajás, Departamento de Ferrosos Norte, Estrada Raymundo Mascarenhas,
S/N Mina de N4, 68516-000, Parauapebas, Pará, Brazil
G.N.S., 0000-0003-3729-7840; R.D., 0000-0003-1754-295X; R.S.A., 0000-0002-3026-5523
* Corresponding author: [Link]@[Link]

Abstract: Companhia de Pesquisa de Recursos Minerais (CPRM, the Brazilian geological survey) has carried out geochemical
surveys focused on the Carajás region, the largest mineral province in Brazil. One of these surveys covered the area of Canaã dos
Carajás, where 257 soil samples, including 32 duplicate samples, were collected from the B-horizon of soils within an area of
c. 3500 km2. The <0.177 mm fraction of these samples was ground and digested with aqua regia and then analysed for
53 elements by inductively coupled plasma atomic emission spectrometry and inductively coupled plasma mass spectrometry.
The statistical treatment of the data allowed the development of multi-element geochemical maps based on interpolation
techniques and the establishment of geochemical background values in soils. Different methods of obtaining the geochemical
threshold values are presented and compared. The geochemical behaviour of 43 elements, including potentially toxic elements,
is discussed. Geochemical background values were established using different statistical methods and the most significant
results were obtained from the iterative 2σ technique. The influence of the geogenic contribution in the study area is largely
dominant over anthropogenic effects. There is no conclusive evidence of contamination related to human activity at the time
when the geochemical survey was carried out. This paper provides multi-element geochemical maps and background values in
the most prominent Brazilian mining province.
Keywords: B-horizon; cluster analysis; heavy metals; natural contamination; Itacaiúnas River watershed; eastern Amazon.
Received 27 August 2018; revised 26 November 2018; accepted 21 March 2019

Geochemical mapping has become increasing important in those influenced by anthropogenic activities. The statistical
environmental studies over the last few decades (Darnley & methods that have been widely applied to calculate background
Garrett 1990; Darnley et al. 1995; Plant et al. 2001) and has been values are the iterative 2σ technique, the calculated distribution
used to discriminate between geochemical impacts related to function, the cumulative frequency curve and boxplot representa-
anthropogenic or geogenic sources (Albanese et al. 2007; Gałuszka tions (Reimann et al. 2005, 2018; Peh et al. 2009; Teng et al. 2009;
2007). Geochemical maps are considered to be an important tool in Mrvić et al. 2010, 2011; Zgłobicki et al. 2011; Ander et al. 2013;
territorial management and have been used not only to monitor the Cembranel et al. 2017).
effects of anthropogenic activities in environmental and pollution Geochemical mapping of soils has rarely been carried out in
research, but also in agriculture, forestry, geomedical and animal Brazil, particularly in the Amazon region. However, Companhia de
health studies (Yuan et al. 2013; Cheng et al. 2014; Ripin et al. Pesquisa de Recursos Minerais (CPRM, the Brazilian geological
2014; Costa et al. 2015). They have also been widely used in areas survey) has executed some projects in the Carajás region (Araújo &
contaminated by toxic elements to evaluate the effects of pollution Maia 1991; Vasquez et al. 2008; CPRM 2012, 2013), which is the
on the environment and to plan remediation strategies (Thornton largest mineral province in Brazil and contains several active mines.
et al. 2008; Guan et al. 2014; Rezaei et al. 2015). The area has experienced an accelerated expansion of human
Background values for chemical elements have been discussed occupation with profound effects on land use and the environment
by many researchers (Matschullat et al. 2000; Reimann & Garrett (Souza-Filho et al. 2016).
2005; Reimann et al. 2005; Gałuszka 2007; Mrvić et al. 2011; The present study is based in a dataset obtained in a soil
Zhang et al. 2011; Gałuszka & Migaszewski 2012; Yuan et al. geochemical survey of the Canaã dos Carajás – Parauapebas area
2013; Cheng et al. 2014). Natural background values can be defined reported by CPRM (2012). This dataset is available to the public on
as the concentrations of chemical elements in an environment that GeoSGB, CPRM’s online geoscience platform ([Link]
has not been influenced by anthropogenic activities (NFESC 2002; [Link]/). Our study is associated with a large geochemical mapping
Reimann & Garrett 2005). They are generally used as a reference project being executed by the Instituto Tecnológico Vale in the
when distinguishing the natural concentrations of elements from Itacaiúnas River watershed (IRW) (Fig. 1). Multivariate statistics are

© 2019 The Author(s). Published by The Geological Society of London for GSL and AAG. All rights reserved. For permissions: [Link]
permissions. Publishing disclaimer: [Link]/pub_ethics

Downloaded from [Link]

by The Lib. East China Geol Inst. user
 432 G. N. Salomão et al.

Fig. 1. Location area and sampling sites. (a) Area covered by the Itacaiúnas River watershed showing the location of the study area, the main cities and
towns, mines, indigenous lands and environmental protection areas; the land cover and land use of the study area in 2013 (Souza-Filho et al. 2016,
simplified) is also shown. (b) Location of the sampling sites in the study area.

used for the first treatment and interpretation of soil geochemical Geological setting
data from this part of the Carajás region. The results will be useful
for comparison with the data now being obtained over the whole The study area is located in the Amazonian craton, an old nucleus of
IRW, as well as for the assessment of future contamination. The the South American plate that became tectonically stable at the
aims of this study were: (1) to analyse geochemical maps of the beginning of the Neoproterozoic (Alkmim 2015). The Amazonian
distribution of elements and to infer possible geogenic, anthropo- craton is divided into seven tectonic provinces (Fig. 2a; Santos et al.
genic or mixed anomalies; (2) to estimate the background values of 2000). The Archean Carajás Province is located in the southeastern
geochemical elements in soils in the study area; and (3) to set up a part of the craton (Machado et al. 1991; Macambira & Lafon 1995;
database for use as a reliable record of soil geochemistry in the study Feio et al. 2013; Moreto et al. 2014). It contains active Fe, Cu
area at the time of sampling (the end of 2011). (both world-class deposits), Mn and Ni mines (Fig. 1a; Docegeo
1988; Moreto et al. 2014) and is limited to the north by the
Archean–Paleoproterozoic Bacajá Domain of the Transamazonas
Location and general characteristics of the study area Province and to the east by the Neoproterozoic Araguaia Belt
The study was carried out in the Carajás region, located c. 700 km (Fig. 2a).
south of Belém, the capital of Pará State in Brazil. The study The Carajás Province is divided into four domains: the
area covers c. 3500 km2. It is entirely located in the IRW and Mesoarchean Rio Maria domain in the south; the Mesoarchean to
encompasses parts of Parauapebas, Curionópolis and Canaã dos Neoarchean Sapucaia and Canaã dos Carajás domains in the centre-
Carajás counties (Fig. 1a). The area was originally covered by the northern area of the province; and the Neoarchean Carajás Basin in
Amazon rainforest, but is now dominated by pastureland with local the north (Dall’Agnol et al. 2017). The study area is located in the
remnants of tropical forest (Souza-Filho et al. 2016). It has a northeastern part of Carajás Province and transcends the boundary
monsoon climate (Alvares et al. 2013) with a mean precipitation of of the province with the Araguaia Belt to the east (Fig. 2). The
c. 1550 mm in the rainy season (November–April) and a mean southern part of the study area is located in the Canaã dos Carajás
precipitation of c. 350 mm in the dry season (May–October) (Silva Domain, which is composed dominantly of Mesoarchean
Júnior et al. 2017). (3.0–2.85 Ga) calc-alkaline granites with subordinate tonalites
In terms of relief, the Carajás region is marked by the Serra dos and trondhjemites and local remnants of greenstone belts
Carajás plateau, which has an altitude ranging from 400 to 900 m. (Feio et al. 2013; Moreto et al. 2014). The crystalline basement
By contrast, the area of soil sampling (Fig. 1b) and the adjacent of the Carajás Basin in the central part of the area is also composed
lands in general have altitudes between 80 and 300 m. Five main of similar orthogneisses and migmatites (c. 2850 Ma; Machado
types of land cover and land use were distinguished in the study area et al. 1991). For the purposes of the present study, we can assume
(Fig. 1b; based on Souza-Filho et al. 2016). Tropical rainforest and that this Mesoarchean basement is of essentially granitoid
montane savannah dominate the pristine landscape, but there has composition.
been a rapid increase in the area of pastureland, which currently The granitoid basement in the Carajás Basin is overlain by the
occupies most of the landscape in the study area and in the IRW Neoarchean Grão Pará and Rio Novo groups (Gibbs et al. 1986;
as a whole. Docegeo 1988; Machado et al. 1991; Vasquez et al. 2008). The

Downloaded from [Link]

by The Lib. East China Geol Inst. user
 Geochemical mapping and background concentrations 433

Fig. 2. Geological setting. (a) Major provinces of the Amazonian craton in northern South America (Santos et al. 2000, modified). (b) Simplified
geological map of the study area (Vasquez et al. 2008 and references cited therein).

Grão Pará Group is metamorphosed in the greenschist facies and Materials and methods
is composed of mafic–felsic volcanic sequences (the Parauapebas
Formation) and banded iron formations (BIFs; the Carajás Original geochemical dataset and sampling procedures
Formation). The Rio Novo Group is similar lithologically to the The dataset used in this study was retrieved from the CPRM
Grão Pará Group (Fig. 2b). database related to the Projeto avaliação do potencial dos recursos
The Mesoarchean and Neoarchean units are intruded by minerais estratégicos do Brasil (CPRM 2012) conducted during
syntectonic Neoarchean granitoids and layered mafic–ultramafic November and December 2011 in the extreme NE of Carajás
intrusions. The Neoarchean granitoids (2.76–2.73 Ga) are asso- Province (Fig. 2a) in the Parauapebas and Canaã dos Carajás areas
ciated with charnockitic rocks and have affinities with A-type (Fig. 1b). A total of 257 soil samples (5 kg per sample), including
granites (Barros et al. 2009; Feio et al. 2012). They are present as 32 duplicate samples, were collected from the B-horizon of soils
distinct units in the study area: the Igarapé Gelado Granite; the within an area of c. 3500 km2 (grid spacing 4 km× 4 km) (Fig. 1b)
Estrela Granite Complex (Barros et al. 2009); the Serra do Rabo with distinct types of vegetation cover (forest, pasture and montane
Granite (Sardinha et al. 2006); and the Planalto Granite Suite (Feio savannah) and the effects of anthropogenic occupation (e.g. urban
et al. 2012). The layered mafic–ultramafic intrusions are composed areas and active mines). Duplicate samples were collected
of peridotites, serpentinites, pyroxenites, norites and gabbros, separately immediately after the collection of the soil samples at a
represented in the study area by the Luanga (2763 ± 7 Ma, Machado rough rate of one duplicate for each set of eight samples collected.
et al. 1991; Teixeira et al. 2015) and Vermelho complexes. The The residual standard deviation of the duplicate analysis was
Santa Inês Gabbro is possibly a similar mafic body with an acceptable (<15%; Sinclair & Blackwell 2004; Eppinger et al.
undefined age (Vasquez et al. 2008). The sandstones and pelitic 2015).
sedimentary rocks of the Águas Claras Formation (Araújo & Maia
1991) partially cover the Archean sequences of the Carajás Basin
Analytical methods
(Fig. 2b) and are probably of Paleoproterozoic age. The Cigano and
Gogó da Onça Paleoproterozoic anorogenic A-type granite plutons Soil samples were dried at 60°C in the laboratory, sieved through an
(Machado et al. 1991; Dall’Agnol et al. 2005; Teruiya et al. 2008; 80 mesh grid (<0.177 mm) and ground in the <0.075 mm fraction.
Teixeira et al. 2017) cross-cut the previously described units Pulp samples weighing 200 g were digested with aqua regia and
(Fig. 2b). analysed for 53 elements (Table 1). Major and minor elements were
The Bacajá Domain of the Transamazonas Province is not determined by inductively coupled plasma atomic emission
exposed in the study area. The Araguaia Belt only occurs in a narrow spectrometry and trace elements by inductively coupled plasma
zone in the eastern part of the area (Fig. 2b). It is represented by the mass spectrometry by ALS Brasil LTDA, Belo Horizonte, Minas
Neoproterozoic Couto Magalhães Formation, which is composed of Gerais. The lower limit of detection (LLD), the proportion of
phyllites and slates interlayered with minor amounts of quartzite, samples with concentrations below this value and the number of
meta-arkose and metalimestone (Hasui et al. 1984; Dall’Agnol et al. outliers for each analysed element are given in Table 1. Details
1988). about the sampling, analysis and quality control of samples and

Downloaded from [Link]

by The Lib. East China Geol Inst. user
 434 G. N. Salomão et al.

Table 1. Detection limit and proportion of samples with contents below the lower limit of detection, samples within the detection range, residual standard
deviation and number of outliers for each analysed element in soil samples of the Carajás region (n = 225)

Data below the lower Samples within Residual standard No. of outliers (Grubbs’ test)
Element Detection limit limit of detection (%) detection range deviation (%) (Grubbs 1969)
Aga 0.01 69.8 68 3.93 2
Al 0.01 225 8.57 1
Asa 1.0 37.8 140 7.48 10
Ba 5.0 4.0 216 8.56 5
Be 0.1 11.6 199 10.88 4
Bia 0.02 46.2 121 9.54 10
Ca 0.01 9.8 203 7.76 8
Cda 0.01 44.0 126 8.75 2
Ceb 0.02–1000.0 224 5.35 14
Co 0.1 225 8.08 9
Cr 1.0 225 7.65 14
Cs 0.05 4.0 216 10.38 5
Cu 0.2 225 9.59 7
Feb 0.03–15.0 208 8.74 0
Ga 0.1 225 8.83 2
Hf 0.05 20.9 178 9.90 12
Hga 0.01 26.2 166 11.51 6
Ina 0.02 30.7 156 8.58 1
K 0.01 5.3 213 11.78 12
La 0.2 225 10.03 10
Lia 1.0 32.9 151 11.54 17
Mg 0.01 8.0 207 11.78 19
Mn 5.0 225 10.47 2
Mo 0.05 12.9 196 8.20 13
Nb 0.05 4.4 215 8.14 8
Ni 0.2 225 10.78 20
P 50.0 5.3 213 6.16 15
Pb 0.2 1.8 221 6.02 9
Rb 0.2 0.9 223 9.23 11
Sa 0.01 44.4 125 5.70 5
Sba 0.05 56.9 97 9.62 2
Sc 0.5 17.8 185 6.52 0
Sn 0.2 225 9.64 3
Sr 0.2 225 7.77 7
Th 0.2 225 6.39 28
Ti 0.01 9.8 203 11.16 5
Tl 0.02 23.1 173 10.18 7
U 0.1 225 7.75 13
V 1.0 225 5.80 1
Wa 0.1 51.6 109 8.05 9
Y 0.1 225 7.84 9
Zn 1.0 5.8 212 10.39 10
Zr 0.5 15.1 191 11.61 9
−1
Detection limit for Al, Ca, Fe, K, Mg, S and Ti is expressed in wt% and for other elements in mg kg . Blank entries indicate not data below the lower limit of detection.
a
Elements with a large proportion of samples below the lower limit of detection (between 25 and 70% of the overall data).
b
Elements with concentrations greater than the upper limit of detection.

regional distribution maps for all elements in different sample Tests for normality (Lilliefors and Kolmogorov–Smirnov) were also
materials have been given in reports and documents provided by the applied. Spearman’s correlation matrix was used to identify the
Division of Geochemistry, CPRM (2012, 2013). relationship between elements. The complete linkage method was
used to construct a dendrogram. The 20 elements (Ag, As, Au, B,
Bi, Cd, Ge, Hg, In, Li, Na, Pd, Pt, Re, S, Sb, Se, Ta, Te and W) that
Statistical analysis
represented >25% of the overall data below the LLD (cf. Table 1)
Analytical values for each element that were below the detection were not used in the computation of the correlation coefficient
limit were replaced by a single value of one-half of the detection matrix nor in the cluster analysis. All statistical analyses were
limit and analytical values above the upper detection limit (UDL) processed using STATISTICA 7.0, StatSoft. Palo Alto, California.
were replaced by the UDL values multiplied by 1.5 for each element
(cf. Reimann & Caritat 2017). Descriptive statistics of element
Geochemical mapping
concentrations were determined, the mean (X) and standard
deviation (SD) values were calculated for both the original dataset The geochemical data were georeferenced with the datum World
and the dataset after removing X ± 2SD. Outliers in the dataset Geodetic System 1984. The spatial interpolation process was based
(Table 1) were identified according to Grubb’s test (Grubbs 1969). on classical algorithms of kriging or inverse distance weighting

Downloaded from [Link]

by The Lib. East China Geol Inst. user
 Geochemical mapping and background concentrations 435

interpolation extended by the application of some principles of Results

fractal geometry (Lima et al. 2003; Cheng et al. 2014). ArcGIS 10.2
software was used to perform the interpolation. Contouring and The descriptive statistics for the 33 elements which have
colouring of all elements were performed based on cumulative concentrations >75% above the LLD (Table 1) are summarized in
frequencies of 5, 15, 25, 40, 50, 65, 75, 85, 90, 91, 93, 95, 97 and Table 2, both before and after eliminating outliers. The mean and
98% (Cheng et al. 2014; Lee et al. 2016). SD values of the concentrations of these elements and the location
of the samples were used to determine the spatial trend and variation
of the data. Statistical analysis of ten additional elements (Ag, As,
Geochemical background determination techniques Bi, Cd, Hg, In, Li, S, Sb and W) was also carried out, but it did not
The geochemical background values were determined using four give consistent responses, probably due to the relatively large
different statistical methods: the boxplot representation (Tukey’s number of analyses with values below the LLD (Table 1). For this
inner fences); cumulative frequency curves; the iterative 2σ reason, these elements were not included in the correlation matrix
technique; and the frequency distribution (Reimann et al. 2005, (Table 3) and cluster analysis (Fig. 3). These elements are not
2018; Nakić et al. 2007; Ander et al. 2013; Urresti-Estala et al. discussed further in this paper.
2013; Cembranel et al. 2017; Reimann & Caritat 2017). The An inter-element correlation matrix of the 33 elements was
boxplot function is most informative if the true number of outliers is calculated. Sinclair & Blackwell (2004) suggested that for
<10% (Reimann et al. 2005) and the calculation of the background Spearman’s correlation values at a significance level of 0.01, in
value by the cumulative curve method requires a minimum number samples with a large number of variables, values with r ≥ 0.65 are
of samples (n > 50; Matschullat et al. 2000). For the iterative the most significant. Based on this suggestion, two reference values
technique and distribution function, the bin (class interval) was were assumed to describe all the correlation values presented in the
calculated using Sturge’s rule and the estimation of background Spearman correlation matrix (Table 3): r ≥ 0.65, represents the most
values was carried out using Excel-based VB Background freeware significant correlations; 0.65 < r ≥ 0.57 represents significant
(Nakić et al. 2007). correlations; and r < 0.57 represents weak correlations.

Table 2. Statistical data on the 33 selected chemical elements analysed in soils of the Carajás region

Before removing After removing

X ± 2SD X ± 2SD

Element Xmin 5th Q1 (25th) Median (50th) Q3 (75th) 95th Xmax X SD X SD

Al 0.13 0.82 2.18 3.41 5.19 9.21 14.07 3.90 2.39 3.55 1.90
Ba <5.0 6.0 18.0 50.0 109.0 223.4 1167.0 81.45 105.18 70.61 61.59
Be <0.1 <0.1 0.3 0.5 0.9 2.2 3.4 0.80 0.66 0.68 0.46
Ca <0.01 <0.01 0.02 0.05 0.09 0.25 0.64 0.08 0.09 0.07 0.05
Ce 1.33 5.87 24.40 48.38 92.82 326.51 >1000.0 80.63 104.11 59.66 51.19
Co 0.3 0.8 3.1 9.5 25.8 67.6 197.0 19.09 28.25 14.76 15.92
Cr 3.0 6.0 16.0 39.0 85.0 288.2 2562.0 106.51 286.18 63.54 72.19
Cs <0.05 0.06 0.19 0.34 0.61 1.557 3.17 0.51 0.51 0.42 0.31
Cu 2.50 7.53 31.20 61.40 151.95 433.35 1856.0 132.52 202.66 105.27 111.95
Fe 0.55 1.25 3.28 5.83 9.13 >15.0 >15.0 6.06 3.55 5.79 3.28
Ga 1.30 4.03 9.90 15.70 21.25 30.19 50.40 16.08 8.47 15.27 7.20
Hf <0.05 <0.05 0.06 0.13 0.25 0.64 1.38 0.23 0.20 0.19 0.12
K <0.01 <0.01 0.03 0.07 0.15 0.43 1.64 0.13 0.19 0.10 0.09
La 0.30 1.73 4.90 11.70 31.30 107.47 426.0 30.43 51.70 21.49 25.26
Mg <0.01 <0.01 0.02 0.03 0.06 0.41 6.59 0.14 0.59 0.08 0.14
Mn 19.0 74.0 189.0 452.0 927.5 2707.8 6634.0 766.93 869.09 631.02 598.09
Mo <0.05 <0.05 0.38 0.79 1.36 4.19 18.41 1.41 1.87 1.11 0.87
Nb <0.05 0.053 0.33 0.74 2.16 5.35 8.76 1.60 1.80 1.29 1.29
Ni 1.40 2.83 7.65 16.90 38.45 123.96 3894.70 58.14 275.84 33.50 59.49
P <50.0 <50.0 99.50 191.00 286.50 725.40 2832.0 266.26 324.71 219.50 147.87
Pb <0.2 2.49 4.90 7.10 11.10 23.44 74.10 9.87 8.87 8.51 5.01
Rb <0.2 2.40 8.80 16.80 36.35 110.30 172.40 29.41 32.53 22.42 19.29
Sc <0.5 <0.5 2.10 7.90 20.05 39.72 58.10 15.31 13.02 13.48 10.37
Sn 0.30 0.70 1.30 2.10 3.10 6.07 9.10 2.48 1.65 2.21 1.20
Sr 0.70 1.20 2.55 4.10 6.70 14.55 43.0 5.51 4.99 4.84 3.16
Th 0.60 2.23 6.30 10.0 17.05 77.22 195.90 19.38 26.91 13.64 12.50
Ti <0.01 <0.01 0.02 0.05 0.1 0.2 0.47 0.08 0.07 0.07 0.05
Tl <0.02 <0.02 0.02 0.07 0.13 0.32 0.74 0.12 0.11 0.11 0.08
U 0.11 0.36 0.87 1.63 3.53 12.57 30.68 3.34 4.69 2.55 2.71
V 4.0 7.0 27.0 71.0 164.5 340.4 665.0 113.80 113.64 99.49 92.53
Y 0.20 0.66 2.23 7.49 20.45 75.35 232.75 18.31 28.43 13.52 16.79
Zn <1.0 <1.0 6.0 12.0 24.5 71.7 135.0 22.15 24.86 17.60 15.21
Zr <0.5 <0.5 1.65 4.40 9.90 23.37 63.60 8.92 10.17 7.52 6.28
−1
Concentrations of Al, Ca, Fe, K, Mg and Ti are given as wt% and the other elements as mg kg .
Xmin (minimum concentration), 5th percentile, 25th percentile (first quartile, Q1), 50th percentile (median), 75th percentile (third quartile, Q3), 95th percentile of the dataset, Xmax
(maximum concentration), arithmetic mean (X) and the standard deviation (SD) were calculated before and after removing X ± 2SD.

Downloaded from [Link]

by The Lib. East China Geol Inst. user
 436
Table 3. Spearman correlation matrix of selected chemical elements of the soils of the Carajás region

Al Ba Be Ca Co Cr Cu Fe Ga Hf K La Mn Ni P Rb Sc Sn Sr Th U V Y Zr
Al 1.00
Ba 0.06 1.00
Be 0.25 0.43 1.00
Ca −0.04 0.49 0.39 1.00
Co 0.06 0.50 0.53 0.59b 1.00
Cr 0.46 0.26 0.23 0.26 0.43 1.00
Cu 0.26 0.19 0.45 0.42 0.68 a 0.41 1.00
Fe 0.58b 0.17 0.34 0.23 0.45 0.43 0.64b 1.00

G. N. Salomão et al.
Ga 0.83 a 0.14 0.49 0.08 0.25 0.44 0.43 0.68 a 1.00
Hf 0.75 a −0.13 0.05 −0.23 −0.06 0.24 0.18 0.43 0.65 a 1.00
K 0.12 0.46 0.20 0.15 0.04 0.05 −0.02 −0.06 −0.03 0.05 1.00
La −0.06 0.45 0.61b 0.55 0.55 0.17 0.36 0.12 0.11 −0.15 0.19 1.00
Mn 0.06 0.56 0.49 0.56 0.82 a 0.18 0.48 0.42 0.28 −0.10 0.05 0.48 1.00
Ni 0.38 0.36 0.42 0.44 0.64b 0.80 a 0.64b 0.51 0.42 0.14 0.07 0.40 0.41 1.00
P 0.44 0.14 0.39 0.32 0.45 0.31 0.70 a 0.80 a 0.58b 0.31 0.02 0.16 0.43 0.47 1.00
Rb −0.01 0.62b 0.47 0.36 0.38 −0.03 0.22 0.06 0.04 −0.10 0.71 a 0.49 0.47 0.08 0.16 1.00
Sc 0.64b 0.20 0.21 −0.01 0.22 0.35 0.29 0.66 a 0.63b 0.49 0.00 0.06 0.32 0.42 0.45 0.03 1.00
Sn 0.49 0.12 0.42 0.08 0.29 0.20 0.43 0.52 0.67 a 0.54 0.05 0.31 0.32 0.29 0.44 0.21 0.55 1.00
Sr −0.08 0.54 0.21 0.74 a 0.40 0.33 0.24 0.03 −0.05 −0.34 0.14 0.33 0.36 0.43 0.13 0.27 −0.08 −0.07 1.00
Th 0.21 −0.08 0.26 −0.02 −0.02 0.06 0.10 0.15 0.45 0.28 −0.06 0.21 0.07 0.01 0.12 0.03 0.12 0.42 −0.16 1.00
U 0.27 0.05 0.58b 0.17 0.22 0.13 0.28 0.27 0.53 0.26 0.05 0.37 0.24 0.20 0.33 0.16 0.13 0.54 −0.05 0.71 a 1.00
V 0.39 0.08 0.17 0.19 0.50 0.45 0.67 a 0.75 a 0.48 0.29 −0.14 −0.01 0.36 0.58b 0.60b −0.09 0.58b 0.25 0.10 −0.03 0.07 1.00
Y −0.01 0.39 0.69 a 0.56 0.60b 0.16 0.37 0.19 0.21 −0.09 0.16 0.86 a 0.54 0.40 0.22 0.51 0.10 0.41 0.26 0.20 0.50 0.08 1.00
Zr 0.59b −0.18 0.04 −0.16 −0.12 0.10 0.24 0.45 0.57b 0.82 a −0.06 −0.04 −0.12 0.08 0.31 −0.12 0.52 0.62b −0.33 0.31 0.23 0.26 −0.05 1.00
a
At a significance level of 0.01, coefficients ≥0.65 are the most significant correlations and are shown in bold type.
b
At a significance level of 0.01, coefficients 0.65 < r ≥ 0.57 are significant (Sinclair & Blackwell 2004).

Downloaded from [Link]

 Geochemical mapping and background concentrations 437

Fig. 3. Dendrogram derived from the

hierarchical cluster analysis of the
33 selected elements. Amalgamation
linkage = complete linkage. Distance
measure = 1 − Pearson r. Mo was
extracted from the dendrogram because
it overpassed the grouping threshold
guide (linkage distance 0.8).

Nine elements (Ce, Cs, Mg, Mo, Nb, Pb, Ti, Tl, and Zn) shown by Fe, V and Sc are commonly observed in geochemical
presented low correlation values. The remaining elements displayed maps (Salminen et al. 1998). The similar behaviour shown by these
19 correlations with coefficients ≥0.65 (Table 3) and an additional three elements with Cu and P is discussed in later in this paper.
18 correlations with acceptable values (0.575 ≤ r < 0.65).The
results obtained via the cluster analysis of 33 elements are shown
Ni–Cr–Mg group
in a dendrogram (Fig. 3). The degree of correlation between groups
of elements is represented by the distance axis – for instance, the The areal distributions of Ni and Cr in the study area are fairly
lower the value on the axis, the more significant the association. similar, with higher concentrations in the northern and southwestern
According to the Pearson’s cluster method, 33 of the selected parts (Fig. 5). These elements are concentrated in mafic–ultramafic
elements were classified and merged into six different clusters rocks with high MgO contents and their association with Mg is
with a maximum linkage distance of 0.8 (Fig. 3). The elements predictable. Ni and Cr show a strong correlation with each other and
comprising the six clusters are shown in Figure 3. A comparison of Ni shows a moderate correlation with Cu and Co (Table 3). Mg does
the background levels of the studied elements in soils in the study not show any relevant correlation with other elements.
area obtained from various statistical methods discussed herein
are presented in Table 4. Al–Ga–Hf–Zr–Sn–Nb group
The highest Al concentrations are observed in the western part of the
Geochemical maps study area (Fig. 6) and a similar distribution pattern is shown by Ga.
The presentation of geochemical maps follows the six clusters or These elements show a strong correlation with each other (0.83;
groups of elements defined in the dendrogram of Figure 3. Table 3). Zr and Hf also show a strong correlation with each other
(0.82; Table 3) and their spatial distribution is fairly similar (Fig. 6),
with stronger anomalies in the western part of the area and localized
Mn–Co–Zn group anomalies in the north. These two elements also show a strong
Anomalous concentrations of elements in this group occur in the correlation with Al, Ga and Sn (Table 3). Ga and Sn also show a
northern and northeastern parts of the study area and, in the case of strong correlation with each other. The areal distribution of Sn (not
Co, scattered in the south (Fig. 4). Mn and Co are strongly correlated shown in Fig. 6) approaches that of Al, but it shows specific
(0.82; Table 3). Zn does not show significant correlations with the features, such as a high anomalous concentration in the southeastern
other elements, but its areal distribution is similar to that of Mn part of the study area. Al and Ga also have significant correlations
(Fig. 4). Although they are in different groups, Co and Cu show a with Fe, an element in another distinct group.
strong correlation (0.68, Table 3) and there are similarities in their
areal distribution.
Ba–K–Rb–Cs–Ca–Sr group
This group puts together the alkaline and alkaline earth elements
Fe–Cu–P–Sc–V–Ti group and there is, in general, a strong or significant correlation among all
Higher concentrations of Fe, Cu and P are observed in the soils of these elements (Table 3). The areal distribution of the elements of
the central-northern, central-southern and western parts of the study this group is illustrated by Ba and Rb. Both show anomalous values
area (Fig. 5). Fe and Cu have a fairly strong correlation with each in the southeastern, southwestern and central-eastern parts of the
other (0.64; Table 3), P and V have a strong correlation with both Fe mapped region (Fig. 6). Ba shows a significant positive correlation
and Cu, but Sc only has a strong correlation with Fe. Ti does not with Ca, K, Rb and Sr (Table 3), indicating that these elements have
show any significant correlation. The similarities in distribution a similar spatial distribution.

Downloaded from [Link]

by The Lib. East China Geol Inst. user
 438 G. N. Salomão et al.

Table 4. Background values obtained by different statistical methods in soils from the Carajás region

Element Boxplot representation Cumulative frequency curve Distribution function Iterative 2σ technique Proposed background values
a
Ag <0.01 <0.01 <0.01
Al 1.45–5.92 1.53–6.33 0.25–6.57 0.04–6.46 0.04–6.46
Asa <1.0–2.38 <1.0–1.83 <1.0–1.83
Ba <5.00–130.50 <5.00–134 <5.00–113.18 <5.00–39.04 <5.00–39.04
Be 0.15–1.05 <0.1–1.07 <0.1–0.95 <0.1–0.95
Bia <0.02–0.14 <0.02–0.31 <0.02–0.07 <0.02 <0.02
Ca <0.01–0.11 <0.01–0.31 <0.01–0.11 <0.01–0.09 <0.01–0.09
Cda <0.01–0.04 <0.01–0.02 <0.01–0.02 <0.01–0.02
Ce 7.62–108.90 5.47–199.86 <0.02–102.96 <0.02–74.97 <0.02–74.97
Co <0.1–31.33 0.7–29 <0.1–22.17 <0.1–5.55 <0.1–5.55
Cr <1.0–102.25 6.0–244.0 <1.0–85.10 <1.0–34.3 <1.0–34.3
Cs 0.09–0.72 0.05–0.74 <0.05–0.71 <0.05–0.64 <0.05–0.64
Cu <0.2–181.83 <0.2–362.9 <0.2–117 <0.2–66.0 <0.2–66.0
Fe 1.86–10.53 1.15–9.35 0.09–11.57 0.06–9.49 0.06–9.49
Ga 7.10–23.90 3.7–29.7 1.39–30.01 2.52–26.32 2.52–26.32
Hf 0.01–0.30 <0.05–0.38 <0.05–0.28 <0.05–0.2 <0.05–0.2
Hga <0.01–0.12 <0.01–0.16 <0.01–0.12
Ina <0.02–0.09 <0.02–0.11 <0.02–0.06 <0.02–0.06
K <0.01–0.17 <0.01–0.33 <0.01–0.15 <0.01–0.12 <0.01–0.12
La <0.2–37.78 1.3–52.30 <0.2–25.45 <0.2–11.22 <0.2–11.22
Lia <1.0–3.63 <1.0–13.0 <1.0–1.80 <1.0–3.05 <1.0–3.05
Mg <0.01–0.07 <0.01–0.23 <0.01–0.06 <0.01–0.05 <0.01–0.05
Mn 5.75–1105.25 64.5–1119.5 <5.0–982.81 14.38–354.00 14.38–354.00
Mo 0.14–1.59 <0.05–2.54 <0.05–1.79 <0.05–1.4 <0.05–1.4
Nb <0.05–2.61 <0.05–2.33 <0.05–0.91 <0.05–0.91
Ni 0.03–46.08 <0.2–140.9 <0.2–36.31 <0.2–15.62 <0.2–15.62
P 54.75–332.25 <50.0–372 <50.0–394.67 <50.0–338.53 <50.0–338.53
Pb 3.38–12.53 <0.2–17.5 1.39–12.81 2.28–9.62 2.28–9.62
Rb 2.05–43.15 1.95–47.25 <0.2–35.32 <0.2–21.75 <0.2–21.75
Sa <0.01–0.02 <0.01–0.02
Sba <0.05–0.25 <0.05–0.16 <0.05–0.12 <0.05–0.12
Sc <0.5–24.35 <0.5–35.25 <0.5–8.83 <0.5–8.83
Sn 0.85–3.55 0.65–4.5 0.28–3.92 0.20–3.52 0.20–3.52
Sr 1.58–7.73 0.95–10 0.33–7.87 <0.2–6.85 <0.2–6.85
Th 1.55–28.5 0.99–19.01 2.09–13.93 2.09–13.93
Ti <0.01–0.12 <0.01–0.17 <0.01–0.09 <0.01–0.09
Tl <0.02–0.16 <0.02–0.25 <0.02–0.2 <0.02–0.15 <0.02–0.15
U 0.22–4.18 0.29–4.62 <0.1–3.26 <0.1–2.31 <0.1–2.31
V <1.0–194.25 <1.0–228 <1.0–159.94 <1.0–71.01 <1.0–71.01
Wa <0.1–0.65 <0.1 <0.1 <0.1
Y <0.1–24.99 0.62–38.84 <0.1–17.68 <0.1–3.76 <0.1–3.76
Zn 1.50–28.50 <1.0–26.5 <1.0–26.17 <1.0–21.18 <1.0–21.18
Zr <0.5–11.95 <0.5–10.6 <0.5–6.06 <0.5–6.06
−1
The Al, Ca, Fe, K, Mg, S and Ti concentrations are expressed as wt% values and the other elements in mg kg . Blank entries are where the technique was not applicable to the analysed
dataset. The geochemical background values for Au, B, Ge, Na, Pd, Pt, Re, Se, Ta and Te were not calculated.
a
Elements presenting from 25 to 70% of the overall dataset below the lower limit of detection (see Table 1). The geochemical background values proposed in this study were mainly
based on the iterative 2σ technique, except for As (distribution function technique) and Hg and S (boxplot representation).

U–Th–La–Ce–Be–Y–Pb–Tl group remaining elements of this group suggest that the distribution of
This group puts together the radiogenic (U and Th) and rare earth these two elements is influenced by additional factors.
elements (REEs, exemplified by La and Ce), in addition to Be, Y, Pb
and Tl (Fig. 7). The geochemical maps show that U, Th, La, Ce, Be
Establishment of geochemical background values
and Y have a similar distribution in the study area. All these
elements are concentrated in the central-northern part of the area, As highlighted in previous studies (Reimann et al. 2005; Gałuszka
with scattered high anomalous values in other places. The 2007; Urresti-Estala et al. 2013), there are several approaches to
distribution of Pb and Tl is not strictly coincident with those of calculating the geochemical background in soils, stream sediments
the other elements in this group. This tendency is also observed in and water bodies. For the dataset of soils of the Carajás region, the
the correlation matrix (Table 3), where U–Th, Y–Be and Y–La background values were obtained using the four techniques
show strong correlations and La–Be and U–Be show significant described previously (Table 4). When the differences in the
correlations. Ce, Pb and Tl do not show any significant correlation background values between the applied methods were significant,
among them, but Ce follows the general geochemical behaviour of we chose the most appropriate technique.
REEs and its distribution is effectively similar to that of La, as It was not possible to estimate the background values for Au, B,
observed in their geochemical maps (Fig. 7). This aspect and the Ge, Na, Pd, Pt, Re, Se, Ta and Te, which have a large amount of data
lack of a consistent correlation between Pb and Tl and the six (≥90%; Table 1) below the detection limit. However, it can be

Downloaded from [Link]

by The Lib. East China Geol Inst. user
 Geochemical mapping and background concentrations 439

Fig. 4. Geochemical maps based on the inverse distance weighting interpolation method showing the distribution of Mn, Co and Zn in soils of the study
area. Urban areas are indicated by black dots and the geological contacts between different units are shown as grey lines (see Fig. 2b). The crosses located
in the upper right-hand corner and at the lower left-hand corner indicate latitudes of 6° and 6°30′ S and longitudes 49° 30′ and 50° W, respectively.

concluded that the threshold values for these elements are lower observe their distance from the main mass of the data, which should
than their detection limits. The remaining data for these elements correspond to the background values (Reimann et al. 2005). In
can be considered as outliers and are hence not representative of the addition, values below the LLD (here set to half the LLD) and
background concentrations in the study area. A similar behaviour values above the ULD (here set to 1.5 times the ULD) are easily
is shown by Ag, As, Bi, Cd, Hg, In, Li, S, Sb and W, which also detectable by this technique (Fig. 8). However, recognizing the
display many analytical values below the LLD (Tables 1 & 4). For threshold in this technique is a complex task because multiple
this reason, these elements are not discussed further in this paper. inflection or break points can be identified. For instance, it is
All the techniques applied in this study have different advantages possible to identify threshold levels in Figure 8 as a major boundary
and disadvantages in establishing the geochemical background and that separates possible outliers from the main population.
threshold values (Peh et al. 2003, 2009; Rodrigues et al. 2013; The boxplot and cumulative frequency curve diagram are useful
Urresti-Estala et al. 2013). For instance, the boxplot representation in clarifying the geochemical behaviour of different elements in the
is a simple and accurate method (Matschullat et al. 2000; Reimann context of the study area. However, the calculated distribution
& Caritat 2005, 2017; Reimann & Garrett 2005; Reimann et al. function and iterative 2σ technique are powerful tools with which to
2005). However, it is consistent only if the true number of outliers is discriminate between anomalous and background populations and
<10% (Reimann et al. 2005). For this reason, this technique was not provide more satisfactory results for the establishment of background
used for Th, for which 12.4% of samples were anomalous (Table 1). values (Nakić et al. 2007, 2010; Urresti-Estala et al. 2013). In
The cumulative frequency curve diagram has advantages over the general, both techniques are applicable, but one can be more
other techniques because it is able to display each single data point, appropriate than the other depending on the particular statistical
the outliers are easily detectable as single values and it is possible to characteristics of the parameters (Nakić et al. 2007; Urresti-Estala

Fig. 5. Geochemical maps based on inverse distance weighting interpolation method showing the distribution of Fe, Cu, P, V, Ni, Cr and Mg in soils of the
study area. The location of known mineral deposits in the area is indicated (see Fig. 2b). Urban areas are indicated by black dots and the geological contacts
between different units are shown as grey lines (see Fig. 2b). The crosses located in the upper right-hand corner and at the lower left-hand corner indicate
latitudes of 6° and 6°30′ S and longitudes 49° 30′ and 50° W, respectively.

Downloaded from [Link]

by The Lib. East China Geol Inst. user
 440 G. N. Salomão et al.

Fig. 6. Geochemical maps based on inverse distance weighting interpolation method showing the distribution of Al, Ga, Hf, Zr, Ba and Rb in soils of the
study area. Urban areas are indicated by black dots and the geological contacts between different units are shown as grey lines (see Fig. 2b). The crosses
located in the upper right-hand corner and at the lower left-hand corner indicate latitudes of 6° and 6°30′ S and longitudes 49° 30′ and 50° W, respectively.

Fig. 7. Geochemical maps based on inverse distance weighting interpolation method showing the distribution of U, Th, La, Ce, Be, Y, Pb and Tl in soils of
the study area. Urban areas are indicated by black dots and the geological contacts between different units are shown as grey lines (see Fig. 2b). The crosses
located in the upper right-hand corner and at the lower left-hand corner indicate latitudes of 6° and 6°30′ S and longitudes 49° 30′ and 50° W, respectively.

Downloaded from [Link]

by The Lib. East China Geol Inst. user
 Geochemical mapping and background concentrations 441

Fig. 8. Four selected cumulative probability plots for Fe, Cu, Cs and Pb. The arrows indicate multiple inflections or break points suggesting possible
thresholds (see text and Table 4). The points displayed vertically in the lower left-hand corner and in the top right-hand corner of the plots register,
respectively, the data below the lower limit of detection (set to LLD/2) and the data above the upper limit of detection (set to 1.5 × ULD).

et al. 2013), the frequency distribution curves presented by the data defining the background values. When only one of these techniques
series and the nature of the parameters to be estimated. Commonly, was consistent according to the normality test, the results obtained by
both methods passed the normality test and, in some cases, the the validated technique were considered to be the background level.
background levels obtained (Fig. 9) were similar (e.g. for Al) and Occasionally, both techniques were rejected by the normality test
both methods could be used, or the results were significantly and, in these cases, the results obtained by the boxplot representation
different (e.g. for Fe) and other issues needed to be considered in were assumed to be the background value (e.g. for Hg and S).

Fig. 9. Comparison between the calculation of background levels for Al and Fe in soils of the Carajás region using the distribution function and iterative 2σ
technique.

Downloaded from [Link]

by The Lib. East China Geol Inst. user
 442 G. N. Salomão et al.

According to Urresti-Estala et al. (2013), there are basically two slightly from 19.01 to 17.16 mg kg−1. By contrast, the results
scenarios for deciding which method is more appropriate to obtained for Th by the iterative 2σ technique indicate no change in
determine background levels. The first scenario is when the the background levels, showing that the calculation of the
dataset presents a wide distribution, but with a large number of background was not influenced by the extraction of the outliers.
smaller concentration values, which may possibly represent the Hence it was concluded that the iterative 2σ technique is a more
background values. In this case, the use of the distribution function stable and reliable tool for the calculation of the background in the
is more appropriate. The second scenario is when the prevailing context of the study area.
natural conditions are responsible for anomalous and high The threshold levels calculated for each element by the iterative
concentrations of parameters. In this case, it is more appropriate 2σ technique lie mostly between the median and the 75th percentile.
to use the iterative 2σ technique. In other words, to decide the most In contrast with the distribution function technique, the character-
suitable method for defining the background, it is important to istics of this distribution can, in some cases, favour an overesti-
evaluate whether natural geogenic features or anthropogenic mation of the background values, as observed for Ba, Ce, Co, Cr,
influences have a more important role in the geochemical Cu, Fe, Ga, La, Mn, Ni, Pb, Th, V and Y (Table 4).
distribution of elements in the study area.
For this reason, some approaches were conducted to determine
the influence of extreme values in a certain dataset for the Discussion
calculation of background levels by both techniques. In the first Influence of geological setting and lithology on the soil
attempt, the original dataset, for the 43 elements studied, was
geochemical signature
submitted to a straightforward statistical analysis in VB Background
(Nakić et al. 2007). For Pb and Th, even when their data were As can be seen in the geochemical maps (Figs 4–7), there are three
accepted by the normality test in both methods, the graphical main domains of multi-element positive anomalies in the soils of
visualization was compromised due to the dispersed distribution of eastern Carajás. The first domain is located in the north, in the form
the data. This characteristic of the distribution seriously affects the of a corridor oriented in an east–west direction. The second domain
construction of histograms (Fig. 10, upper panels). The second is located in the central-northern region of the area and is oriented
attempt was based on the extraction of the outliers, followed by WNW–ESE, following the dominant regional Archean trend
statistical analysis in VB Background. As a result, the graphical (Barros et al. 2009; Dall’Agnol et al. 2017). The third domain is
visualization of the data was remarkably improved in both methods in the southern part of the study area. It is lies along an arc segment
and changes in the background levels were observed (Fig. 10, lower from the SW towards the south and is largely coincident with the
panels). For some elements, such as Th, the threshold level Tarzan and Bocaina serras, which comprise the recently created
calculated by the distribution function technique was changed Ferruginous Fields National Park (Fig. 1). The Ferruginous Fields

Fig. 10. Differences observed in the calculation of the background levels for Th in soils of the Carajás region based on the calculated distribution function
and the iterative 2σ technique by (upper panel) considering the original dataset and (lower panel) excluding outliers from the dataset (see Table 1).

Downloaded from [Link]

by The Lib. East China Geol Inst. user
 Geochemical mapping and background concentrations 443

National Park is less affected by anthropogenic activities than the The lithological group of undifferentiated Archean granitoids and
dominant deforested areas of the IRW. The lowest concentrations of gneisses and associated rocks is less homogeneous geochemically
almost all the studied elements occur in soils derived from Archean because the granitoids include varied lithologies, ranging from
granitoid rocks, located mainly in the southwestern and eastern classical tonalite–trondhjemite associations to high-Mg sanuki-
regions of the study area, and from the Proterozoic (meta) toids, passing through several varieties of calc-alkaline leucogra-
sedimentary Águas Claras and Couto Magalhães formations. nodiorites and granites (Feio et al. 2013; Costa Silva et al. 2015).
The Carajás Province is the main mining region in the Amazonian These rocks show significant geochemical contrasts depending on
craton and there are important IOCG-type deposits of Cu, generally the dominant crustal or mantle origin of their magmas. Minor
with Au as a by-product, in the Canaã dos Carajás area (Fig. 2b; occurrences of gabbroic rocks and remnants of greenstone belts,
Cristalino, Bacaba, Castanha, Estrela and Serra Verde; Moreto et al. which could exert a local influence on the soil geochemistry, are
2015) and the Sossego Cu mine is located immediately to the west. associated with the granitoids. These Archean rocks were also
All these deposits are in a hydrothermal corridor located in the affected by strong hydrothermal alteration in some areas and their
southern part of the area and oriented in a roughly east–west original geochemistry was altered, with enrichment in some
direction following the dominant Archean trend (Moreto et al. incompatible and siderophile elements. As a result, the geochemical
2015). The Serra Leste Fe mine is located in the northeastern part of signature observed in the soils of these lithologies clearly differ
the area and large Fe deposits also exist in the Serra Sul de Carajás from the A-type Archean and Paleoproterozoic granitoids and
(S5 and S6) and in the Serra do Buriti to the south (Fig. 2). BIFs are approach, in some respects, that of metamorphosed mafic–
scattered in the different occurrences of the Grão Pará Group and intermediate volcanic rocks. Nevertheless, the more common
ferruginous crusts are commonly developed over different kinds feature in the areas where these rocks are dominant is the absence
of rocks. Both constitute an additional source of Fe enrichment. Ni or scarcity of strong anomalies in the geochemical elements
and Cr deposits are associated with mafic–ultramafic intrusions considered (Figs 4–7).
(Vermelho in the south and Luanga complex in the NE). In the past, Soils derived from Mg-enriched mafic–ultramafic rocks show
large Au deposits were exploited in the Serra Pelada area. anomalous values of Ni, Cr and Co, with subordinate influences
In general, there is strong evidence that the patterns observed on from Cu and V. This signature is clearly seen in the occurrences of
the geochemical maps of Canaã dos Carajás intrinsically reflect the the Luanga complex in the north and the Vermelho complex in the
geology of the area. Most of the positive anomalies observed on the south of the study area (Fig. 5). These geological units are mainly
maps (Figs 4–7) are clearly related to a specific lithological group composed of dunite, gabbro, harzburgite, peridotite, norite,
and the main contrasts in the distribution of chemical elements can pyroxenite and serpentinites, with subordinate anorthosite and
be largely explained by the local dominant lithology. Soils chromitite (Machado et al. 1991; Vasquez et al. 2008; Mansur &
developed on the four lithological groups all have their own Ferreira Filho 2017).
characteristic geochemical signature, as observed on the maps.
The soils developed on Neoarchean and Paleoproterozoic A-type
Preliminary evaluation of the anthropogenic impact
granites show anomalous values of U, Th, Zr, Hf, Y and Be and
constitute a distinct geochemical group (Fig. 7). Pb and Tl were also The region discussed in this paper has experienced major changes
statistically associated with this group, but their geochemical link in land cover and land use, with intense deforestation and the
with the A-type granites is less conclusive. The A-type granites are formation of new landscapes dominated by pastureland
enriched in U, Th, Zr, Hf and Y (|Dall’Agnol et al. 2005, 2017; (Souza-Filho et al. 2016). However, agricultural production in the
Barros et al. 2009) relative to most of the local Archean granitoids region remains limited and industrial activities are concentrated
(Feio et al. 2013). Be is commonly found in pegmatites around the main cities and active mines.
(Č erný 2002) that occur associated with the Estrela complex in Mn, Al and heavy metals (e.g. Ni, Pb, Cd, As, Cr, Cu, Co, Zn and
the central-northern part of the area (Barros et al. 2009). In addition Hg) can seriously affect ecosystems and cause environmental
to the Estrela complex, these six elements also show anomalous problems (Dumčius et al. 2011; Zgłobicki et al. 2011; Yuan et al.
values in soils sampled in areas where the Igarapé Gelado, Serra 2013). These elements can be released to soils as a result of
do Rabo, Cigano and Gogó da Onça granites and the Planalto anthropogenic influences, but may also be affected by lithological
Suite occur. or geogenic factors (Ogundiran & Osibanjo 2009; Khalid et al.
The soils formed on clastic sedimentary and metasedimentary 2016). Anthropogenic sources of soil contamination by heavy
rocks corresponding to the Águas Claras and Couto Magalhães metals include mining activities, the industrial refining of ores,
formations (Fig. 2b) are generally devoid of anomalous positive pesticides and fertilizers related to agriculture, farming and industry,
values for the chemical elements considered (Figs 4–7). It can be solid waste disposal and atmospheric deposition (Ross 1994; Niazi
concluded that the soils derived from these lithologies are et al. 2011, 2015; Khalid et al. 2016). These metals occur naturally
geochemically depleted relative to those derived from other in rocks as minerals and, rarely, as ores, often in form of sulphides
dominant lithologies. and oxides (Fuge et al. 1991).
Metamorphosed mafic–intermediate volcanic rocks and BIFs are Reimann & Caritat (2017) assumed As, Ba, Cd, Co, Cr, Cu, Hg,
represented in the area by the Grão Pará and Rio Novo groups Mo, Ni, Pb, Sn, Zn, Mn and V to be potentially toxic elements
(Fig. 2b). These lithologies are the sources of soils enriched in Cu, presenting a relatively high risk to the environment. They concluded
P, Fe, Sc, V and, possibly, Ti (Fig. 5). The good correlation shown that Australian surface soils show only minor local contamination
by Fe, V and Sc is a reflection of their geochemical affinity and is from these elements, with positive anomalies indicating mineral
commonly observed in other regions (Salminen et al. 1998). A exploration potential rather than the effects of pollution. It is
similar areal distribution of Cu and P (Fig. 5) can be explained by important to establish the geochemical background of these
the fact that apatite is an accessory phase in mafic volcanic rocks and elements to distinguish their natural concentrations from those
that Cu sulphides with subordinate apatite are concentrated in the resulting from anthropogenic contamination (Reimann & Caritat
hydrothermal rocks responsible for Cu–Au mineralization in the 2017). Following this reasoning, some elements typically associated
Canaã dos Carajás area (Fig. 2b) along the hydrothermal corridor with human activities, including Ba, Mn, P, Pb, Sr and Zn, were
(Moreto et al. 2015). The main anomalous Fe concentrations are evaluated in this study (Figs 4–7), but we have not been able to
typically associated with BIFs of the Carajás Formation or with detect a clear anthropogenic influence in the soils at the mapping
laterites derived by supergenic alteration. scale used in this study. Our preliminary conclusion is that, in this

Downloaded from [Link]

by The Lib. East China Geol Inst. user
 444 G. N. Salomão et al.

Fig. 11. Boxplots for nine potentially toxic elements (Ba, Co, Cr, Cu, Mn, Mo, Ni, Pb and Zn) in soils of the Carajás region grouped according to the
geology and lithological groups of the sampling location (see text). Group 1, A-type Archean and Paleoproterozoic granites; group 2, clastic sedimentary
and metasedimentary rocks; group 3, Archean granitoids, gneisses and associated rocks; and group 4, metamorphosed mafic and intermediate volcanic rocks
and banded iron formations. The green horizontal line in the boxplots represents the threshold values calculated for each element based on the iterative 2σ
technique. The box indicates the 25th, 50th and 75th percentiles; outliers are defined according to: (upper whisker, lower whisker) = (upper hinge, lower
hinge) ± 1.5 × hinge width.

study area, the geological setting has a more important role in Geological influences on the establishment of
determining the geochemical soil signature than human activities. background values
However, the Australian study sampled surface soils, in contrast
with this research where only data from the subsoil were considered. Four main lithological associations occur in the study area (Fig. 2b):
Preliminary unpublished data from the Itacaiúnas project indicate (1) A-type Neoarchean and Paleoproterozoic granites (39 soil
that, in general, there is no significant contrast in composition samples); (2) clastic sedimentary and metasedimentary rocks
between the surface soil (0–20 cm deep) and deeper soil samples (28 samples); (3) undifferentiated Archean granitoids and gneisses
(30–50 cm deep). and associated rocks (70 samples); and (4) metamorphosed mafic

Downloaded from [Link]

by The Lib. East China Geol Inst. user
 Geochemical mapping and background concentrations 445

and intermediate volcanic rocks and BIFs (80 samples). In addition References
to these lithological associations, there are also stratified mafic and Albanese, S., De Vivo, B., Lima, A. & Cicchella, D. 2007. Geochemical
ultramafic rocks represented by eight soil samples. These could not background and baseline values of toxic elements in stream sediments of
be evaluated statistically, but can exert a local influence on the Campania region (Italy). Journal of Geochemical Exploration, 93, 21–34,
[Link]
geochemical signature of the soils. Alkmim, F.F. de. 2015. Geological background: a tectonic panorama of Brazil.
The analysed samples were grouped into these lithological In: Hermelin, M. (ed.) Landscapes and Landforms of Colombia. Springer,
groups and boxplots for nine potentially toxic elements (Ba, Co, Cr, Dordrecht, 1–210, [Link]
Alvares, C.A., Stape, J.L., Sentelhas, P.C., Moraes Gonsalves, J.L. & Sparovek,
Cu, Mn, Mo, Ni, Pb and Zn) were constructed and compared with G. 2013. Koppen’s climate classification map for Brazil. Meteorologische
the calculated threshold values in the study area (Fig. 11). The Zeitschrift, 22, 711–728, [Link]
results indicate that it is not appropriate to establish a single Ander, E.L., Johnson, C.C., Cave, M.R., Palumbo-Roe, B., Nathanail, C.P. &
Lark, R.M. 2013. Methodology for the determination of normal background
threshold value to identify unusually high element concentrations. concentrations of contaminants in English soil. Science of the Total
For instance, the background values of Co in the study area range Environment, 454–455, 604–618, [Link]
from <0.10 to 5.55 mg kg−1, roughly the same values as for soils 03.005
derived from lithological group 2 (1.67–6.02 mg kg−1), whereas the Araújo, O.J.B. & Maia, R.G. 1991. Programa Levantamentos Geologicos
Basicos do Brasil: Projeto Especial Mapas de Recursos Minerais, de Solos e
obtained values are 3–6.5 times higher in soils derived from groups de Vegetacao Para a Area do Programa Grande Carajas – Subprojeto
1 (3.90–15.90 mg kg−1), 3 (2.78–29.38 mg kg−1) and 4 (4.05– Recursos Minerais – Serra dos Carajas – Folha SB.22-Z-A – Estado do Para.
36.35 mg kg−1) (Fig. 11). According to Reimann & Caritat (2017), Departamento Nacional da Produção Mineral–Companhia de Recursos
Minerais, Brasília.
these large differences do not allow the definition or use of a single Barros, C.E.M., Sardinha, A.S., Barbosa, J.P.O., Macambira, M.J.B., Barley, P.
threshold value. Nevertheless, background values are commonly & Boullier, A. 2009. Structure, petrology, geochemistry and zircon U/Pb and
used in legislation to distinguish potentially contaminated regions Pb/Pb geochronology of the synkinematic Archean (2.7 Ga) A-type granities
from pristine areas without any consideration of the specific from the Carajás Metallogenic Province, northern Brazil. The Canadian
Mineralogist, 47, 1423–1440, [Link]
geological setting and this remains an important contradiction in Cembranel, A.S., Sampaio, S.C., Remor, M.B., Gotardo, J.T. & Dalla Rosa, P.M.
environmental regulations. 2017. Geochemical background in an oxisol. Engenharia Agricola, 37,
565–573, [Link]
Č erný, P. 2002. Mineralogy of beryllium in granitic pegmatites. Reviews in
Conclusions Mineralogy and Geochemistry, 50, 405–444, [Link]
2002.50.10
The geochemical maps presented in this paper can be used to show Cheng, Z., Xie, X., Yao, W., Feng, J., Zhang, Q. & Fang, J. 2014. Multi-element
geochemical mapping in southern China. Journal of Geochemical
the background variations in elements on a regional scale. The Exploration, 139, 183–192, [Link]
geochemical background associated with these maps has suggested Costa, R. de V. da, Leite, M.G.P., Mendonça, F.P.C. & Junior, H.A.N. 2015.
essentially natural geogenic enrichment for these elements in the Geochemical mapping of arsenic in surface waters and stream sediments of the
Quadrilátero Ferrífero, Brazil. Geosciences, 68, 43–51, [Link]
study area. The background levels will be a useful reference for 1590/0370-44672015680077
future evaluations of anthropogenic activity in this region. Costa Silva, A.R., Villas, R.N.N., Lafon, J.M., Craveiro, G.S. & Ferreira, V.P.
Among the different methods applied to calculate and determine 2015. Stable isotope systematics and fluid inclusion studies in the Cu–Au
Visconde deposit, Carajás Mineral Province, Brazil: implications for fluid
the geochemical background, the iterative 2σ technique provided
source generation. Mineralium Deposita, 50, 547–569, [Link]
the most satisfactory results for the establishment of background 1007/s00126-014-0558-8
values in soils in the study area. Prior adjustment of the data is not CPRM (Serviço Geológico do Brasil) 2012. Projeto avaliação do potencial dos
required in this technique, although its major advantage is related to recursos minerais estratégicos do Brasil: área Carajás Oriental, [Link]
[Link]/ [last accessed 13 November 2018].
the improvement in graphic visualization of a certain dataset by CPRM (Serviço Geológico do Brasil) 2013. Projeto metalogenia das províncias
the extraction of outliers without changing the background. The minerais do Brasil: área Sapucaia, [Link] [last accessed 13
threshold levels calculated for each element by the iterative 2σ November 2018].
Dall’Agnol, R., Teixeira, N.P., Macambira, J.B., Kotschoubey, B., Gorayeb, P.S.S.
technique mostly lie between the median and the 75th percentile, & Santos, M.D. 1988. Petrologia dos gnaisses e micaxistos da porção norte da
but some values also occur beyond this range. In contrast with other Faixa de Dobramentos Araguaia, Goiás-Brasil. In: 7th Latino-American
methods, which, in some cases, could overestimate the background Geological Congress. Sociedade Brasileira de Geologia, Belém, 1–19.
values, the characteristics of the iterative 2σ technique tend to avoid Dall’Agnol, R., Teixeira, N.P., Rämö, O.T., Moura, C.A.V., Macambira, M.J.B.
& de Oliveira, D.C. 2005. Petrogenesis of the Paleoproterozoic rapakivi
this problem. A-type granites of the Archean Carajás metallogenic province, Brazil. Lithos,
Many anomalies of the analysed elements were identified in the 80, 101–129, [Link]
study area. However, by evaluating the spatial distribution of the Dall’Agnol, R., Cunha, I.R.V. da, Guimarães, F.V., Oliveira, D.C. de, Teixeira,
M.F.B., Feio, G.R.L. & Lamarão, C.N. 2017. Mineralogy, geochemistry, and
elements and the geochemical background values obtained, it is petrology of Neoarchean ferroan to magnesian granites of Carajás Province,
concluded that there is strong evidence of a geogenic influence and Amazonian Craton: The origin of hydrated granites associated with
no conclusive evidence of contamination related to human activity charnockites. Lithos, 277, 3–32, [Link]
in the soils of the Canaã dos Carajás region. Darnley, A.G., Bjiirklund, A. et al. 1995. A Global Geochemical Database
for Environmental and Resource Management – Recommendations for
International Geochemical Mapping, [Link]
pf0000101010
Acknowledgements We acknowledge Lúcia Travassos Rosa-Costa and Darnley, A. & Garrett, R.G. (eds). 1990. International Geochemical Mapping
César Lisboa Chaves for support in accessing the CPRM database and general Special Issue. Journal of Geochemical Exploration, 39.
information about geochemical mapping in the Canaã dos Carajás area. Docegeo (Rio Doce Geologia e Mineração) 1988. Revisão litoestratigráfica da
Província Mineral de Carajás. In: 35 Congresso Brasileiro de Geologia.
CVRD/SBG, Belém, 11–59.
Funding This work was supported by the Conselho Nacional de Dumčius, A., Paliulis, D. & Kozlovska-Kędziora, J. 2011. Selection of
Desenvolvimento Científico e Tecnológico (CNPq) (grants 130.789/2017-8 to investigation methods for heavy metal pollution on soil and sediments of
GNS, 306.108/2014-3 to RD, 305.392/2014-0 to RSA, 443.247/2015-3 Project water basins and river bottoms: a review. Ekologija, 57, 30–38, [Link]
coordinated by RD with DTI scholarship to MABMF/GNS – process 381.570/ 10.6001/ekologija.v57i1.1307
2016-9), the Geochemical Background project, currently under execution at Eppinger, R.G., Giles, S.A., Lee, G.K. & Smith, S.M. 2015. Second Projet de
Vale Institute of Technology Sustainable Development (ITVDS) with Renforcement Institutionnel Du Secteur Minier de la Republique Islamique de
Vale funding (GABAN-DIFN), the Institute of Geoscience of the Federal Phase II : Geochemistry – Database Creation, Data Quality Assessment, and
University of Pará and the Fundação de Amparo e Desenvolvimento da Pesquisa Geochemical Maps – Final Report on Compilation and Validation of
(FADESP). Geochemical Data. United States Geological Survey, Reston, VA.
Feio, G.R.L., Dall’Agnol, R. et al. 2012. Geochemistry, geochronology, and
Scientific editing by John Carranza origin of the Neoarchean Planalto granite suite, Carajás, Amazonian craton:

Downloaded from [Link]

by The Lib. East China Geol Inst. user
 446 G. N. Salomão et al.

A-type or hydrated charnockitic granites? Lithos, 151, 57–73, [Link] geostatistics. Environmental Science and Technology, 45, 10463–10470,
10.1016/[Link].2012.02.020 [Link]
Feio, G.R.L., Dall’Agnol, R., Dantas, E.L., Macambira, M.J.B., Santos, J.O.S., Niazi, N.K., Singh, B. & Minasny, B. 2015. Mid-infrared spectroscopy and
Althoff, F.J. & Soares, J.E.B. 2013. Archean granitoid magmatism in the partial least-squares regression to estimate soil arsenic at a highly variable
Canaã dos Carajás area: implications for crustal evolution of the Carajás arsenic-contaminated site. International Journal of Environmental Science
province, Amazonian craton, Brazil. Precambrian Research, 227, 157–185, and Technology, 12, 1965–1974, [Link]
[Link] Ogundiran, M.B. & Osibanjo, O. 2009. Mobility and speciation of heavy metals
Fuge, R., Glover, S.P., Pearce, N.J.G. & Perkins, W.T. 1991. Some observations in soils impacted by hazardous waste. Chemical Speciation and
on heavy metal concentrations in soils of the Mendip region of north Somerset. Bioavailability, 21, 59–69, [Link]
Environmental Geochemistry and Health, 13, 193–196, [Link] Peh, Z., Miko, S. & Bukovec, D. 2003. The geochemical background in Istrian
1007/BF01758636 soils. Natura Croatica, 12, 195–233, [Link]
Gałuszka, A. 2007. Different approaches in using and understanding the term Peh, Z., Miko, S. & Hasan, O. 2009. Geochemical background in soils: a linear
‘geochemical background’ - practical implications for environmental studies. process domain? An example from Istria (Croatia). Environmental Earth
Polish Journal of Environmental Studies, 16, 389–395, [Link] Sciences, 59, 1367–1383, [Link]
Different-Approaches-in-Using-and-Understanding-r-nthe-Term-Geochemical- Plant, J., Smith, D., Smith, B. & Williams, L. 2001. Environmental geochemistry
Background,88000,0,[Link] at the global scale. Applied Geochemistry, 16, 1291–1308, [Link]
Gałuszka, A. & Migaszewski, Z.M. 2012. Geochemical background – an 1016/S0883-2927(01)00036-1
environmental perspective. Mineralogia, 42, 7–17, [Link] Reimann, C. & Caritat, P. 2005. Distinguishing between natural and
v10002-011-0002-y anthropogenic sources for elements in the environment: regional geochemical
Gibbs, A.K., Wirth, K.R., Hirata, K.H. & Olszewski, W.J., Jr 1986. Age and surveys versus enrichment factors. Science of the Total Environment, 337,
composition of the Grao Para Group volcanics, Serra dos Carajas, Brazil. 91–107, [Link]
Revista Brasileira de Geociencias, 16, 201–211, [Link] Reimann, C. & Caritat, P. 2017. Establishing geochemical background variation
0375-7536.1986201211 and threshold values for 59 elements in Australian surface soil. Science of the
Grubbs, F.E. 1969. Procedures for detecting outlying observations in samples. Total Environment, 578, 633–648, [Link]
Technometrics, 11, 1–21, [Link] 11.010
Guan, Y., Shao, C. & Ju, M. 2014. Heavy metal contamination assessment and Reimann, C. & Garrett, R.G. 2005. Geochemical background – concept and
partition for industrial and mining gathering areas. International Journal of reality. Science of the Total Environment, 350, 12–27, [Link]
Environmental Research and Public Health, 11, 7286–7303, [Link] [Link].2005.01.047
10.3390/ijerph110707286 Reimann, C., Filzmoser, P. & Garrett, R.G. 2005. Background and threshold:
Hasui, Y., Costa, J.B.S. & Abreu, F.A.M. 1984. Província Tocantins-Setor critical comparison of methods of determination. Science of the Total
setentrional. In: Almeida, F.F.A. & Hasui, Y. (eds) O Precambriano No Environment, 346, 1–16, [Link]
Brasil. Edgard Blucher, São Paulo, 137–204. Reimann, C., Fabian, K. et al. 2018. GEMAS: establishing geochemical background
Khalid, S., Shahid, M., Niazi, N.K., Murtaza, B., Bibi, I. & Dumat, C. 2016. A and threshold for 53 chemical elements in European agricultural soil. Applied
comparison of technologies for remediation of heavy metal contaminated Geochemistry, 88, 302–318, [Link]
soils. Journal of Geochemical Exploration, 182, 247–268, [Link] Rezaei, A., Shayestehfar, M., Hassani, H. & Mohammadi, M.R.T. 2015.
1016/[Link].2016.11.021 Assessment of the metals contamination and their grading by SAW method: a
Lee, G.K., Yager, D.B., Mauk, J.L., Granitto, M., Denning, P.D., Wang, B. & case study in Sarcheshmeh copper complex, Kerman, Iran. Environmental
Werdon, M.B. 2016. The Geochemical Atlas of Alaska, 2016. US Geological Earth Sciences, 74, 3191–3205, [Link]
Survey Data Series 908, [Link] Ripin, S.N.M., Hasan, S. & Kamal, M. 2014. Environmental geochemical
Lima, A., De Vivo, B., Cicchella, D., Cortini, M. & Albanese, S. 2003. mapping on distribution of metal contamination in topsoils Perlis, Malaysia.
Multifractal IDW interpolation and fractal filtering method in environmental Journal of Medical and Bioengineering, 3, 277–281, [Link]
studies: an application on regional stream sediments of (Italy), Campania jomb.3.4.277-281
region. Applied Geochemistry, 18, 1853–1865, [Link] Rodrigues, [Link] L., Malafaia, G., Costa, A.T. & Nalini Júnior, H.A. 2013.
S0883-2927(03)00083-0 Background values for chemical elements in sediments of the Gualaxo do
Macambira, M.J.B. & Lafon, J.M. 1995. Geocronologia da Província Mineral de Norte River Basin, MG, Brazil. Revista de Ciências Ambientais, 7, 15–32,
Carajás. Síntese dos dados e novos desafios. Boletim do Museu Paraense Emílio [Link]
Goeldi, 7, 263–288, [Link] Ross, S.M. 1994. Sources and forms of potentially toxic metals in soil–plant
000216&pid=S0001-3765200700010001400016&lng=pt systems. In: Ross, S.M. (ed.) Toxic Metals in Soil–Plant Systems. Wiley,
Machado, N., Lindenmayer, Z., Krogh, T.H. & Lindenmayer, Z.G. 1991. U-Pb Chichester.
geochronology of Archaean magmatism and basement reactivation in the Santos, J.O.S., Hartmann, L.A. et al. 2000. A New Understanding of the
Carajás area, Amazon shield, Brazil. Precambrian Research, 49, 329–354, Provinces of the Amazon Craton Based on Integration of Field Mapping and
[Link] U-Pb and Sm-Nd Geochronology. Gondwana Research, 3, 453–488, https://
Mansur, E.T. & Ferreira Filho, C.F. 2017. Chromitites from the Luanga Complex, [Link]/10.1016/S1342-937X(05)70755-3
Carajás, Brazil: stratigraphic distribution and clues to processes leading to Salminen, R., Tarvainen, T. et al. 1998. FOREGS Geochemical Mapping, Field
post-magmatic alteration. Ore Geology Reviews, 90, 110–130, [Link] Manual. Geological Survey of Finland, Espoo.
10.1016/[Link].2017.03.016 Sardinha, A.S., Barros, [Link] M. & Krymsky, R. 2006. Geology, geochemistry,
Matschullat, J., Ottenstein, R. & Reimann, C. 2000. Geochemical background – and U–Pb geochronology of the Archean (2.74 Ga) Serra do Rabo granite stocks,
can we calculate it? Environmental Geology, 39, 990–1000, [Link] Carajás Metallogenetic Province, northern Brazil. Journal of South American
1007/s002549900084 Earth Sciences, 20, 327–339, [Link]
Moreto, C.P.N., Monteiro, L.V.S. et al. 2014. Timing of multiple hydrothermal Silva Júnior, R.O., de Souza, E.B., Tavares, A.L., Mota, J.A., Ferreira, D.B.S.,
events in the iron oxide–copper–gold deposits of the Southern Copper Belt, Souza-Filho, P.W.M. & da Rocha, E.J.P. 2017. Three decades of reference
Carajás Province, Brazil. Mineralium Deposita, 50, 517–546, [Link] evapotranspiration estimates for a tropical watershed in the eastern Amazon.
10.1007/s00126-014-0549-9 Anais da Academia Brasileira de Ciencias, 89, 1985–2002, [Link]
Moreto, C.P.N., Monteiro, L.V.S. et al. 2015. Neoarchean and Paleoproterozoic 1590/0001-3765201720170147
iron oxide–copper–gold events at the Sossego deposit, Carajás Province, Sinclair, A.J. & Blackwell, G.H. 2004. Applied Mineral Inventory Estimation.
Brazil: Re–Os and U–Pb geochronological evidence. Economic Geology, 110, Cambridge University Press, Cambridge.
809–835, [Link] Souza-Filho, P.W.M., Souza, E.B. et al. 2016. Four decades of land-cover, land-use
Mrvić, V., Kostic-Kravljanac, L., Zdravkovic, M., Brebanovic, B., Cakmak, D., and hydroclimatology changes in the Itacainas River watershed, southeastern
Sikiric, B. & Saljnikov, E. 2010. Background limit of Zn and Hg in soils of Amazon. Journal of Environmental Management, 167, 175–184, [Link]
Eastern Serbia. Journal of Agricultural Sciences, Belgrade, 55, 157–163, org/10.1016/[Link].2015.11.039
[Link] Teixeira, A.S., Ferreira Filho, C.F., Giustina, M.E.S.D., Araújo, S.M. & da Silva,
Mrvić, V., Kostić-Kravljanac, L., Zdravković, M., Koković, N., Perović, V., H.H.A.B. 2015. Geology, petrology and geochronology of the Lago Grande
Č akmak, D. & Nikoloski, M. 2011. Methods for assessment of background layered complex: evidence for a PGE-mineralized magmatic suite in the
limit of Ni and Cr in soils of eastern Serbia. Ratarstvo i povrtarstvo, 48, Carajás Mineral Province, Brazil. Journal of South American Earth Sciences,
189–194, [Link] 64, 116–138, [Link]
Nakić, Z., Posavec, K. & Bačani, A. 2007. A Visual Basic spreadsheet macro for Teixeira, M.F.B., Dall’Agnol, R., Santos, J.O.S., Sousa, [Link] & Lafon, J.-
geochemical background analysis. Ground Water, 45, 642–647, [Link] M. 2017. Geochemistry, geochronology and Nd isotopes of the Gogó da Onça
org/10.1111/j.1745-6584.2007.00325.x Granite: a new Paleoproterozoic A-type granite of Carajás Province, Brazil.
Nakić, Z., Posavec, K. & Parlov, J. 2010. Model-based objective methods for the Journal of South American Earth Sciences, 80, 47–65, [Link]
estimation of groundwater geochemical background. AQUAmundi, 1, 65–72, 1016/[Link].2017.09.017
[Link] Teng, Y., Ni, S., Wang, J. & Niu, L. 2009. Geochemical baseline of trace elements
NFESC 2002. Guidance for Environmental Background Analysis, Vol. 1: Soil. in the sediment in Dexing area, South China. Environmental Geology, 57,
Naval Facilities Engineering Command, Washington DC. 1649–1660, [Link]
Niazi, N.K., Bishop, T.F.A. & Singh, B. 2011. Evaluation of spatial variability of Teruiya, R.K., Paradella, W.R., Dos Santos, A.R., Dall’Agnol, R. & Veneziani, P.
soil arsenic adjacent to a disused cattle-dip site, using model-based 2008. Integrating airborne SAR, Landsat TM and airborne geophysics data for

Downloaded from [Link]

by The Lib. East China Geol Inst. user
 Geochemical mapping and background concentrations 447

improving geological mapping in the Amazon region: the Cigano Granite, Yuan, G.L., Sun, T.H., Han, P. & Li, J. 2013. Environmental geochemical
Caraja’s Province, Brazil. International Journal of Remote Sensing, 29, mapping and multivariate geostatistical analysis of heavy metals in topsoils of
3957–3974, [Link] a closed steel smelter: Capital Iron & Steel Factory, Beijing, China. Journal of
Thornton, I., Farago, M.E. et al. 2008. Urban geochemistry: research strategies to Geochemical Exploration, 130, 15–21, [Link]
assist risk assessment and remediation of brownfield sites in urban areas. 02.010
Environmental Geochemistry and Health, 30, 565–576, [Link] Zgłobicki, W., Lata, L., Plak, A. & Reszka, M. 2011. Geochemical and statistical
1007/s10653-008-9182-9 approach to evaluate background concentrations of Cd, Cu, Pb and Zn (case
Urresti-Estala, B., Carrasco-Cantos, F., Vadillo-Pérez, I. & Jiménez-Gavilán, P. study: Eastern Poland). Environmental Earth Sciences, 62, 347–355, https://
2013. Determination of background levels on water quality of groundwater [Link]/10.1007/s12665-010-0529-z
bodies: a methodological proposal applied to a Mediterranean River basin Zhang, Q., Zhang, Y., Wang, X. & Luo, J. 2011. Comparison of sequential
(Guadalhorce River, Málaga, southern Spain). Journal of Environmental chemical extraction procedure and DGT technique to measure effective
Management, 117, 121–130, [Link] arsenic in sediment of Lake Taihu, China. Journal of Agro-Environmental
Vasquez, M.L., Sousa, C.S. & Carvalho, K.M.A. 2008. Geologia eRecursos Science, 30, 1398–1403, [Link]
Minerais do Estado do Pará. Belém. [Link]?file_no=11987&year_id=2011&quarter_id=7&falg=1

Downloaded from [Link]

by The Lib. East China Geol Inst. user