-
Seediscussions,stats,andauthorprofilesforthispublicationat:http://www.researchgate.net/publication/257673746
MethodsforExtractingHeavyMetalsinSoilsfromtheSouthwesternAmazon,BrazilARTICLEinWATERAIRANDSOILPOLLUTIONFEBRUARY2013ImpactFactor:1.69DOI:10.1007/s11270-012-1430-z
DOWNLOADS22
VIEWS74
2AUTHORS:
SabrinaNovaesdosSantos-AraujoUniversityofSoPaulo4PUBLICATIONS35CITATIONS
SEEPROFILE
LuisAlleoniUniversityofSoPaulo121PUBLICATIONS575CITATIONS
SEEPROFILE
Availablefrom:SabrinaNovaesdosSantos-AraujoRetrievedon:23June2015
-
Methods for Extracting Heavy Metals in Soilsfrom the
Southwestern Amazon, Brazil
Sabrina Novaes dos Santos &Lus Reynaldo Ferracci Alleoni
Received: 20 August 2012 /Accepted: 18 December 2012 /Published
online: 19 January 2013# Springer Science+Business Media Dordrecht
2013
Abstract Heavy metals occur naturally in soil, atconcentrations
that depend on the parent material fromwhich the soil was formed,
the processes of formation,and the composition and the proportion
of the compo-nents of its solid phase. Quantifying these
concentra-tions is important for environmental studies of
soilcontamination and pollution, and choosing the meth-ods for
doing so is a key step in establishing heavymetal contents in soil
samples. We evaluated twodigestion methods (aqua regia and EPA
3051, bothmicrowave oven-assisted) for assessing
pseudo-totalconcentrations of Cd, Co, Cr, Cu, Ni, Pb and Zn inthe
surface layer (020 cm) of soil samples from theBrazilian
agricultural frontier in the southwesternAmazon. Nineteen composite
samples of the mostrepresentative soil classes for the states of
MatoGrosso and Rondnia were collected under nativevegetation
undisturbed by human intervention.Canonical discriminant analysis
and principal compo-nent analysis were used for multivariate
exploration ofthe data. Aqua regia extracted higher amounts of
Co,
Ni, Pb, and Zn than EPA 3051, while levels of Cr andCu did not
differ between methods. In general, aquaregia recovered more of the
metals when compared toreference soil samples.
Keywords Aqua regia . EPA 3051 . Backgroundconcentrations . Soil
contamination . Acid extractors
1 Introduction
Heavy metals occur naturally in soils, at concentra-tions that
depend on the soil parent material, on theprocesses by which the
soil formed, and on the com-position and proportion of the
components of its solidphase. Quantifying these concentrations is
importantfor studies of environmental pollution, and a key
steptowards establishing soil quality reference values.
Environmental agencies require reference indica-tors in order to
monitor environmental impacts anddetermine pollution levels, as
part of their mandate toenforce environmental legislation. These
indicatorsare obtained by comparing concentrations of toxicelements
in soils with those observed in natural(nonpolluted) soils, or with
reference values.
To date, a number of studies have been carried outto quantify
natural concentrations and reference back-ground values of heavy
metals in Brazilian soils.However, no studies have described
natural concen-trations of heavy metals in the soils of Mato
Grossoand Rondnia, Brazilian states in the southwestern
Water Air Soil Pollut (2013) 224:1430DOI
10.1007/s11270-012-1430-z
S. N. dos Santos (*)Graduate Student in Soil Science and Plant
Nutrition,University of Sao Paulo (ESALQ/USP),Av. Pdua Dias, 11,
C.P. 9,Piracicaba, So Paulo 13418-900, Brazile-mail:
[email protected]
L. R. F. AlleoniDepartment of Soil Science, ESALQ/USP,Av. Pdua
Dias, 11, C.P. 9,Piracicaba, So Paulo 13418-900, Brazil
-
Amazon, which represent one of the largest agricul-tural
frontiers on Earth.
Deciding which method to prepare and extract con-centrations of
elements in soil samples is a crucial stepin the process of
describing the environmental condi-tions of a given area. There are
several different meth-ods of acidic digestion, ranging from aqua
regia (AR)(3:1 HCl/HNO3, v/v), with varying quantities of acidsand
various digestion times and temperatures, in anopen system, and
sometimes even hydrofluoric acid isused in a closed system. This
latter digestion is con-sidered total, due to the destruction of
the silicatematrices (Chen and Ma 1998; Caires 2009).
A large number of studies (e.g., Akker and Delft1991; Chen and
Ma 1998; Tam and Yao 1999; Chenand Ma 2001; Campos et al. 2003;
Tighe et al. 2004;Chander et al. 2008; Caires 2009; Nemati et al.
2010)have documented marked differences in the amountsof metals
extracted by these different methods. Thismeans that government
agencies must establish normsregarding methods to extract naturally
occurring con-centrations in soils, in order to permit rigorous
com-parisons with pre-established values.
In Brazil, the analytic methods adopted in legislationfor
quantifying total concentrations of inorganic substan-ces (apart
frommercury) in soil samples are twomethodsof the United States
Environmental Protection Agency(USEPA), known as 3050B (HNO3 +
H2O2) and 3051(HNO3) (USEPA, 2007a, b, c), which are widely used
todigest samples of soils, sediments, and wastes.
In the 3051 method, the oxidation of organic matteris performed
by nitric acid without the solubilizationof the silicate fraction
(pseudo-total). Conversely, inthe 3052 method the use of nitric and
hydrofluoricacids promotes a full dissolution of the
sample.However, hydrofluoric acid requires very careful han-dling,
because it can cause severe burns in contactwith skin (Costa et al.
2008). It can also damageanalytical instruments (e.g., it attacks
silicate materi-als, especially glass) and promote undesirable
forma-tion of insoluble fluoride precipitation with Al, Ca, Fe,and
Mg, and coprecipitation with Rb, Sr, Y, Cs, Ba,Pb, Th, and U
(Yokoyama et al. 1999; Vieira et al.2005). The use of the 3052
method is hardly justifiedin highly weathered tropical soils
because mineralreserves are not impressive and their silicate
fractioncontains low amounts of heavy metals.
Microwave ovens have been used in soil chemistrylaboratories
since the 1980s (Jassie and Kingston
1988) and provide fast, secure, and efficient digestion(Chen and
Ma 2001).
Microwave heating is employed in the 3050B,3051, and 3052
methods. Its advantages include ashorter sample digestion, a more
complete dissolutionof samples, and a smaller loss of volatile
elements, inaddition to a lower risk of contamination than
othermethods (Abreu et al. 2006).
AR is also very efficient at extracting pseudo-totalcontents of
heavy metals in soils. It is the standardmethod for certifying soil
samples in Britain andFrance (Prez et al. 1997). Nitric acid
oxidizes hydro-chloric acid, giving rise to various oxidation
productssuch as molecular chlorine and nitrosyl chloride (3HCl
+HNO3 = 2H2O + Cl2 + NOCl) (Chen and Ma 2001).This property, along
with the presence of chloride(complexing), makes AR a highly
efficient extractor fordissolving heavy metals (Costa et al.
2008).
In this study we assessed the AR and EPA 3051methods of
pseudo-total digestion for extracting Cd,Co, Cr, Cu, Ni, Pb, and
Zn. Concentrations weresubsequently correlated with physical and
chemicalproperties of soils from the states of Mato Grossoand
Rondnia, in Brazils southwestern Amazonregion.
2 Materials and Methods
The territories of Mato Grosso and Rondnia stateswere first
divided into 11 ecoregions or biogeoclimaticmacrozones intended to
reflect large-scale similaritiesand differences in soil, climatic,
and managementconditions. This classification was performed with
anArcView 9.0 Geographical Information System bysuperimposing maps
of soil conditions (scale of 1:5million), native vegetation,
geology, climate, and to-pography, with the goal of identifying
homogeneousareas that would allow our measurements of soil
attrib-utes at the study sites to be extrapolated across theentire
region (Maia et al. 2009). These homogeneouszones illustrate
biogeoclimatic conditions free of an-thropogenic disturbances such
as deforestation, agri-culture, or ranching (Table 1).
Soils were sampled in June and July 2007. Twotownships in each
of the regions were selected atrandom, for a total of 19 townships.
These yielded19 soil samples, with three replicates each,
collect-ed under native vegetation free of anthropogenic
1430, Page 2 of 16 Water Air Soil Pollut (2013) 224:1430
-
disturbance and considered to be representative ofsoils in Mato
Grosso and Rondnia. Within eachtownship ranches with areas of
native vegetationin which to collect soils were selected (Fig.
1).
Compound samples were formed from samplestaken at five
collection sites within a 100100 mplot at each ranch: one site at
the center of theplot and the other four on the square
corners.Samples were collected at a depth of 020 cm.After being
collected, samples were air-dried, ho-mogenized, and sieved through
2-mm mesh. Mostof the soils collected belonged to the
followingorders: Oxisols (74 %), Inceptsols (16 %), andEntisols (10
%). More details regarding the studyare found in the work of Santos
and Alleoni(2012).
The contents of available phosphorus and potassi-um were
extracted with Mehlich-I (Mehlich 1953),and calcium and magnesium
were extracted with1 M KCl (Anderson and Ingram 1992).
Phosphoruscontent was determined by colorimetry, Ca and Mgwere
quantified in an atomic absorption spectropho-tometer (AAS), and K
in a flame photometer.Exchangeable aluminum was removed with 1
moll1
KCl (Anderson and Ingram 1992), and determined bytitration with
0.025 moll1 NaOH. Total acidity (H +Al) was extracted with a 1 M
calcium acetate (pH7buffered) solution and determined by titration
with0.025 M ammonium hydroxide.
pH was determined potentiometrically for suspen-sions of
air-dried fine earth in 1 M KCl, H2O, and0.01 M CaCl2 1:2.5
(Anderson and Ingram 1992). The
Table 1 Characteristics of macroregions in the states of Mato
Grosso and Rondnia, southwestern Amazon region, Brazil(Mello,
2007)
Region Topography Soil Climate and averageannual rainfall
(mm)
Vegetation
Alto Xingu Flat with large plateaus Oxisols Amia (1,7502,250)
Seasonal semi-deciduousforest to open Amazonforest
Paran Basin Flat with large plateaus Oxisols Quartzipsamments
Am,b Cwac (1,2501,750) Cerrado sensu stricto
Parecis Plateau Flat with large plateaus Quartzipsamments
Oxisols Ami (1,5002,250) Cerrado sensu stricto andseasonal
semi-deciduousforest
Araguaia Depression Flat areas and rollinghills;
frequentlywaterlogged
Entisols AquentEntisols
Ami (1,2502,000) Open cerrado (dominatedby grasses) and
cerradosensu stricto
Cuiab Depression Flat areas and rolling hills Inceptisols
Entisols Am (1,5001,750) Cerrado sensu stricto
Guapor Depression Mostly flat Oxisols, UltisolsEntisols
Ami (1,7502,250), Ami,Am (1,5001,750)
Open Amazon forest(north) and seasonalsemi-deciduous forestto
Cerrado (south)
NortheasternMato Grosso
Irregular with rolling hills Ultisols Amid (2,0002,500) Cerrado
to seasonalsemi-deciduous forest
NorthernMato Grosso
Irregular with rolling hills Ultisols, OxisolsInceptisols
Awi, Ami (2,0002,750) Open Amazon forest toseasonal
semi-deciduousforest
Northern Rondnia Predominately flat Oxisols Awi (2,0002,500)
Open Amazon forest
Central Rondnia Varies between flatareas and rolling hills
Ultisols, Oxisols Awi, Ami (1,7502,250) Open Amazon forest
Pantanal Mostly flat; frequentlywaterlogged
Entisols, Alfisols Am (1,5001,750) Open Cerrado and
seasonalsemi-deciduous forest
a Ami = very hot tropical monsoon climateb Am = tropical monsoon
climatec Cwa = wet subtropical climated Awi = tropical savanna
climate
Water Air Soil Pollut (2013) 224:1430 Page 3 of 16, 1430
-
difference between pH1 M KCl and H2O was used toverify the sign
of the net charge (Mekaru and Uehara1972). Based on the contents of
exchangeable cations,the sum of exchangeable bases, the cation
exchangecapacity (CEC) at pH7.0, and base and Al saturationswere
then calculated. Granulometric analysis was per-formed by the
densimeter method (Gee and Or 2002)and organic carbon was
determined with a LECO CN-2000 dry combustion elemental
analyzer.
The oxides (expressed as SiO2, Al2O3, Fe2O3,TiO2, and MnO) were
extracted using 9 M H2SO4.Contents of Fe, Mn and Al were determined
usingatomic absorption spectrophotometry (AAS), Ti bycolorimetry
(Vettori 1969), and Si by gravimetry.The Ki weathering index was
calculated by the molarrelation method where Ki =
(%SiO2/60)/(%Al2O3/102). Sodium citratebicarbonatedithionite
(Na-CBD) solution was used to extract the free iron oxides(Mehra
and Jackson 1960) and the concentration ofthe poorly crystalline
oxides of Fe, Al, and Mn were
determined by the method described by Loeppert andInskeep
(1996).
Pseudo-total concentrations of metals wereobtained via digestion
by (1) the AR method(Mcgrath and Cunliffe 1985) and (2) with
concentrat-ed acid in a microwave oven, under controlled
tem-perature and pressure, following the EPA 3051 method(USEPA,
1996).
AR: 0.5 g of samples sieved through 100 mesh wastransferred to
Teflon tubes. Each tube received 9 ml ofconcentrated HCl (32 %) and
3 ml of concentratedHNO3 (65 %) (3:1) of analytical-level purity,
and themixture was left undisturbed for 12 h of pre-digestion.The
tubes were subsequently placed in a closed systemin a microwave
oven (Mars Xpress, CEM Corporation)for 20 min at 180 C subsequently
passing the temper-ature ramp. After cooling, samples were
transferred tocertified 25-ml flasks (NBR ISO/IEC), the flask
volumewas completed with ultrapure water, and the extractswere
filtered. All analyses were carried out with three
Fig. 1 Map of southwestern Amazonian Brazil, showing townships
selected for soil sampling (RO Rondnia, MT Mato Grosso)
1430, Page 4 of 16 Water Air Soil Pollut (2013) 224:1430
-
replicates. Analysis quality control was performed withcertified
soil samples (NIST SRM 2709 San Joaquinsoil and DO65-540 Metals in
Soil EnvironmentalResource Associates).
Calibration curves to determine metal concen-trations were
prepared based on 1,000 mg l1
standards (TITRISOL, Merck), using ultrapurewater for dilution,
cleaning, and decontaminationof the glassware, which was kept in 10
% nitricacid solution for 24 h and rinsed with distilledwater and
ultrapure water. Concentrations of Co,Cr, Cu, Ni, Pb, and Zn were
determined via atom-ic absorption spectrophotometry (FAAS) and
Cdconcentrations in a graphite oven.
The AR method is normally carried out with diges-tion in an open
system. However, methods in which aclosed system is heated with a
microwave oven arewell established and widely employed
(Nieuwenhuizeet al. 1991; Chen and Ma 1998, 2001; Nemati et
al.2010; Sakan et al. 2011). The advantages of the closedsystem
compared to other methods are: (1) ashorter digestion time for
samples, (2) a lower riskof pollution, (3) a more complete
dissolution ofsamples, and (4) lower losses of volatile
elements(Sakan et al. 2011).
EPA 3051: 0.5 g of samples sieved through 100mesh was
transferred to Teflon tubes, to which 10 mlof concentrated HNO3 (65
%) of analytical purity wereadded. Samples were placed in a closed
system micro-wave oven (Mars Xpress, CEM Corporation) for4 min and
30 s at 175 C, subsequently passingthe temperature ramp. After
cooling, samples weretransferred to certified 25-ml flasks (NBR
ISO/IEC), with flask volume completed with ultrapurewater, and the
extracts were filtered with slowfilter paper. All analyses were
carried out in threereplicates, and sample blanks were performed
si-multaneously. Analysis quality control was per-formed with
certified soils (NIST SRM 2709 SanJoaquin soil and DO65-540 Metals
in Soil Environmental Resource Associates). The NISTrecommends
comparing methods that do not usehydrofluoric acid (3050, 3051, and
their updatedversions) with recovery based on leachable
con-centrations (NIST, 2002), since certified concentrationsare
determined based on methods to determine totalconcentrations.
Calibration curves and the determi-nation of metal concentrations
were carried out asdescribed in the previous paragraph.
Results were examined via analysis of variance(ANOVA) (F Cu. The
rangesof recovery were within those recommended for theelements
under study (USEPA, 1996).
Water Air Soil Pollut (2013) 224:1430 Page 5 of 16, 1430
-
3.2 Digestion Methods
The two methods of pseudo-total digestion yieldeddifferent
levels of naturally occurring heavy metalsin soils of Mato Grosso
and Rondnia (Table 4). Ingeneral, pseudo-total concentrations of
metalsobtained via digestion by both methods were wellbelow the
maximum levels considered acceptable forsoils worldwide (Alloway
1990; Reimann et al. 2000).Mean concentrations observed in soils of
Rondnia
and Mato Grosso were also generally lower than thosereported in
the literature for other Brazilian states andfor other countries
(MINEROPAR 2005; Salonen andKorkka-Niemi 2007; Paye et al. 2010;
Caires 2009;Bini et al. 2011; Shah et al. 2012).
The low natural levels obtained by the two methodsmay be related
to the physical and chemical soil attrib-utes and to the parent
material of soils in the region.Oxisols and Ultisols are highly
weathered soils, andtheir clay fraction is dominated by kaolinite,
gibbsite,
Table 3 Average contents ofheavy metals recovered for cer-tified
reference materials (Biondi2010)
aValue of metal content for bothcertified soilsbPercentage of
metals recoveredfor DO65 and percentage of met-als recovered in
relation to theleachate 2709acUsed to extract metals with
ex-tractor EPA 3051dUsed to extract metals with theaqua regia
extractor
Element Certified soil Recovery level(mg kg1)
Certified valuea
(mg kg1)Recovery(determined)b (%)
Leachrecovery (%)
Cd 2709ac 0.3 0.370.002 104 110
DO65d 84.7 91.0 93
Co 2709a 9.3 12.800.2 73 81
DO65 174.4 190.0 91.8
Cr 2709a 52.0 130.09.0 40 41
DO65 140.4 144.0 97
Cu 2709a 24.4 33.90.5 72 81
DO65 251.9 237.0 106
Ni 2709a 59.5 852 70 77
DO65 214.0 200.0 107
Pb 2709a 8.1 17.30.1 47 53
DO65 111.2 104.0 107
Zn 2709a 71.0 1034 69 77
DO65 314.4 341.0 107
Table 2 Selected physical andchemical attributes of soil
sam-ples from the states of Rondoniaand Mato Grosso, Brazil
(n=19)
CECe effective cation exchangecapacity, CECt totalcation
ex-change capacity, V base satura-tion, m saturation for
aluminum
Variable Mean Minimum Maximum Standard Deviation
pH (H2O) 4.6 3.5 7.4 0.9
pH (0.01 M CaCl2) 4.3 3.2 7.2 0.9
pH (1 M KCl) 4.2 3.2 6.9 0.9
mmolc kg1
CECe 19.0 2.3 96.4 17.9
CECt 55.4 17.8 158.0 29.0
%
V 24 0.2 99 26.9
m 46 0.1 98 34.6
g kg1
Org. C 13.3 4.4 35.1 7.1
Sand 654 271 953 189
Silt 38 10 92 21
Clay 307 25 651 174
1430, Page 6 of 16 Water Air Soil Pollut (2013) 224:1430
-
goethite, and iron and aluminum oxides (Fontes andWeed 1991).
They are also typically acidic (pH in H2Ovarying from 4.3 to 6.2)
and have very low concen-trations of heavy metals. Parent material
is one of theprincipal determinants in the distribution of
heavymetals in soils. Even under severe weathering condi-tions
(pedogenesis) such as those present in the wettropics, parent
material may play an important role indetermining a large part of
the heavy metal content ofsoils (Baize and Sterckeman 2001).
Cd concentrations were below the detection levelfor both
methods. Co concentrations varied from 7.2to 38.9 mgkg1 for
digestion with AR and from 16.6to 39.0 mgkg1 for EPA 3051
digestion. Mean Coconcentrations were higher for the EPA 3051
extrac-tion than for AR. Mean Cr concentrations were higherfor the
AR digestion (47.9 mgkg1) than for EPA 3051(39.4 mgkg1). However,
there was significant varia-tion within methods, with AR showing
results from20.4 to 142.1 mgkg1 and EPA 3051 from 19.2 to98.8
mgkg1. Mean Cu concentrations were similarfor the two methods: 18.3
and 16.6 mgkg1 for ARand EPA 3051, respectively.
AR extracted much higher quantities of Ni, Pb, andZn than EPA
3051. Mean values varied from 0.2, 5.2,and 1.2 mgkg1 to 24.3, 25.9,
and 100.9 mgkg1 forthe AR method, and from 0, 2.7, and 0 mgkg1 to
5.6,15.8, and 69.6 mgkg1 for the EPA 3051 method,respectively. The
methods did not differ (p
-
especially so for elements that are involved in thesilicate
matrix, such as Cr, Ni, and Pb. The variationin Ni, Pb and Zn
contents in this study probablyoccurred because the amount and the
nature of thealuminosilicate matrix also varied.
AR is a mixture of the acids HNO3 and HCl.They react to form
nitrosyl chloride (NOCl) andmolecular chlorine (Cl2), which are
highly reac-tive, have a high oxidizing power, and are capableof
dissolving even noble metals, but do not totallydissolve silicates.
In the European Union, AR isthe most commonly used solution for
extractingmetals from polluted soils (Gleyzes et al. 2002;Grotti et
al. 2002; Quevauviller 2002), and is thestandard method for
certifying soil samples inGreat Britain and France (Sakan et al.
2011).The EPA 3051 method uses nitric acid (HNO3),an oxidizing
agent commonly used to extract met-als, which are converted into
soluble nitrates.Hydrogen peroxide is used as an auxiliary agentin
the oxidation of samples with a high content oforganic carbon,
which reacts to produce H2O.Oliveira et al. (2008) evaluated the
effectivenessof three methods of extracting total metal
concen-trations (AR, EPA 3051, and EPA 3052), andconcluded that AR
was the most appropriate.Although it does not extract total
concentrationsof heavy metals, AR does provide a reasonableestimate
of the maximum amount that can poten-tially become available to
plants or be leachedinto groundwater (Diaz-Barrientos et al.
1991).
In our study, AR was also the most effective meth-od for
extracting naturally occurring heavy metals.This may be due to the
association of nitric acid andhydrochloric acid in a closed system,
creating a dis-solving mixture that is extremely efficient for
heavymetals (Costa et al. 2008) and increases the extractionpower
of AR. Extraction procedures with strong acidssuch as AR with HNO3
and/or HCl typically seek toreflect a pollutants potential
availability and mobility,factors which are related to its transfer
from soils toplants (Rauret 1998). While it did not extract
higherconcentrations of heavy metals from Mato Grosso andRondnia
soils, the EPA 3051 method is widely usedin the United States and
by many environmental agen-cies worldwide, and its use is mandated
underBrazilian law by CONAMA.
The CDA model explained the variation betweenthe methods, with
Canonical Discriminant Function
(CDF1) (AR) responsible for 100 % of the difference(Fig. 3). One
important insight provided by the CDAis the value of the parallel
discrimination rate (PDR),which is the product of the standardized
canonicalcoefficients (SCC) and the correlation (r). The param-eter
r provides univariate information for each metal(i.e., its
individual contribution), independent of themethod used. The PDR is
more efficient when theresearcher wants to discriminate between
areas(Cruz-Castillo et al. 1994; Baretta et al. 2005) ormethods, as
in this study. Positive values of thePDR coefficient indicate a
separation effect be-tween the methods, while negative values
indicatesimilarities between methods for a given attribute(Baretta
et al. 2005).
Within the CDF1 the methods varied depending onthe heavy metals
studied (Table 3). According to thePDR values within the CDF1, Co,
Ni, Pb, and Zn werethe elements that contributed the most to
distinguish-ing between the digestion methods (0.21, 0.27, 0.37,and
0.17, respectively), and showed excellent po-tential as indicators
(Fig. 3 and Table 5). Pbshowed the highest PDR and was the element
thatcontributed the most to distinguishing between themethods. In
the CDF2, Ni had the highest PDRvalue (0.26) of all the heavy
metals analyzed. Cuwas less sensitive, had a lower PDR value,
andcontributed less (0.04 and 0.06, respectively) inthe two
functions (CDF1 and CDF2) to distin-guishing between the areas
studied (Table 5).
Fig. 3 Relationship between the first and second
discriminantcanonical function (DCF1 and DCF2) on the mean
(centroid) ofthe standardized canonical coefficients (SCC) for Zn,
Ni, Pb, Cr,Co, and Cu in soils of Mato Grosso and Rondnia,
Brazil
1430, Page 8 of 16 Water Air Soil Pollut (2013) 224:1430
-
3.3 Correlations
Correlations between the heavy metal concentrationsin the 19
soil samples and the 34 physical and chem-ical attributes of those
soils are presented in Table 6.Metal concentrations extracted with
AR, which had ahigher extraction power of the pseudo-total
concen-trations of heavy metals, were used in the correlations.With
the exception of Cr, there was generally a posi-tive correlation
between the concentrations of Co, Cu,Ni, Pb, and Zn (Table 6).
Biondi (2010) reportedsimilar results in a study of soils from
Pernambuco,Brazil, where Cu concentrations were positively
cor-related with concentrations of Zn (r=0.78; r=0.81),Co (r=0.81;
r=0.76), and Ni (r=0.73; r=0.71) in thesurface and subsurface
horizons, respectively (p0.7*). Lee et al. (1997)studied soils in
Oklahoma, USA collected nearthe parent material, underlying
sedimentary rocks and concluded that the most influential attribute
in thedistribution of heavy metals in the soil profile was thesum
of silt and clay contents. In our study, that sumwas also
positively correlated with metal concentra-tions, independent of
extraction method, and showed acorrelation coefficient varying from
0.31 to 0.73.
In general, Fe and Mn oxides were the variablesthat showed the
strongest correlations with heavy met-als. Various authors have
noted the importance of ironoxides as determinants of heavy metal
mobility insoils under a tropical climate (Fontes et al.
2000;Fontes and Weed 1991; Alloway 1990; Covelo et al.2007). Oxides
have varying capacities for adsorbingmetallic cations. For a given
type of oxide, the capac-ity for adsorbing cations varies with the
degree ofcrystallization of the oxide, and can vary as the
weath-ering process causes changes in the form of the crystal,in
the surface area, and in the chemical properties ofthe surface of
the oxides (Yu 1997). In general, poorlycrystallized substances
have a large specific surfacearea and a high capacity for adsorbing
metals. Bycontrast, the activity of well-crystallized substancesis
comparatively lower (Yu 1997).
Mn oxides and hydroxides are rare in soils and, forthis reason,
less studied than Fe oxides. Nonetheless,they are efficient
sorbents of heavy metals because oftheir small size and large
specific surface area(McKenzie 1979). Like Fe oxides, Mn oxides are
verystable (i.e., they have low levels of solubility; Stummand
Morgan 1996), their surfaces are highly reactive,and they
precipitate as tiny poorly crystallized oramorphous crystals
(Fortin et al. 1993; Tessier et al.1996).
CO concentrations were not correlated with theconcentrations of
any metal in this study (p
-
Table6
Pearson
correlationmatrixbetweennaturallevelsof
heavymetals(aquaregiamethod)
andsoilattributes,with
significantvalues
(p