lable at ScienceDirect

Journal of Environmental Management 92 (2011) 2069e2075

Contents lists avai

Journal of Environmental Management

journal homepage: www.elsevier .com/locate/ jenvman

Natural attenuation of residual heavy metal contamination in soils affected by theAznalcóllar mine spill, SW Spain

Saúl Vázquez, América Hevia, Eduardo Moreno, Elvira Esteban, Jesús M. Peñalosa, Ramón O. Carpena*

Department of Agricultural Chemistry, Universidad Autónoma de Madrid, 28049 Madrid, Spain

a r t i c l e i n f o

Article history:Received 24 August 2010Received in revised form24 February 2011Accepted 24 March 2011Available online 29 April 2011

Keywords:Spatial-temporal variationAvailabilityHeavy metalsArsenicAcidified soilsPyritic sludge

* Corresponding author. Tel./fax: þ34 914973938.E-mail address: ramon.carpena@uam.es (R.O. Carp

0301-4797/$ e see front matter � 2011 Elsevier Ltd.doi:10.1016/j.jenvman.2011.03.030

a b s t r a c t

Non-amended soils affected by pyritic sludge residues were monitored for 7 years to assess the long-term natural attenuation ability of these soils. The decrease in both the total concentration ofelements (particularly As) and (NH4)2SO4-extractable fractions of Mn, and Zn, below the maximumpermissible levels indicate a successful natural ability to attenuate soil pollution. Soil acidification bypyrite oxidation and rainfall-enhanced leaching were the largest contributors to the reduction of metalsof high (Mn, Cu, Zn and Cd) and low (Fe, Al, and As) availability. Periodic use of correlation and spatialdistribution analysis was useful in monitoring elemental dispersion and soil property/elementrelationships.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

A serious arsenic and heavy metal contamination incidentoccurred after the widely-publicized pyrite mine reservoir accidentin Aznalcóllar, Spain in 1998 (Aguilar et al., 2007). National andregional environmental authorities initiated a clean-up program byfirst mechanically removing toxic sludge and surface soil, thenapplying soil amendments to prevent dispersion of the contami-nants. Despite that, the affected area remained contaminated withan irregular distribution of pollutants resulting from residualsurface sludge being inadvertently buried during removal (Álvarez-Ayuso et al., 2008). Aerobically-enhanced pyrite oxidation in thesludge-contaminated soil mobilized metallic and metalloidelements through acidification. Both the level and availability of soiltrace elements increased, incurring a risk of further contamination.

Reports describing the evolution of contamination from theAznalcóllar spill have examined leaching across the soil profile(Kraus and Wiegand, 2006; Aguilar et al., 2007; Álvarez-Ayusoet al., 2008), temporal and spatial variations in the pollutantconcentrations (Burgos et al., 2006; Vanderlinden et al., 2006;Ordóñez et al., 2007), the effects of soil amendments (Walker

ena).

All rights reserved.

et al., 2004), and phytoremediation techniques (Bernal et al.,2007; Clemente et al., 2005; Vázquez et al., 2006).

The dynamic behavior of trace elements in soil is complex, beinginfluenced by factors such as pH, organic matter, texture, redoxpotential, and temperature (Alloway, 1995). Natural attenuation isa term used to describe a collection of in-situ physical, chemical,and biological processes that, under favorable conditions, actwithout human intervention to reduce the mass, toxicity, mobility,volume, or concentration of contaminants in soil or groundwater(EPA 1999). To our knowledge, the effect of two years of naturalattenuation on the Aznalcóllar soil contamination has only beendiscussed in Clemente et al. (2006). In this study we sought toassess the long-term natural attenuation ability of non-amendedsoils affected by residual contamination from the Aznalcóllarspill. Correlations between soil properties and element concen-trations, correlations between the concentrations of differentelements, and spatial evolution of elemental profiles were alsoexamined.

2. Material and methods

2.1. Experimental site and design

The study was carried out at the 1000 m2 B2 experimental plotin the “El Vicario” area (37�260210 0N 06�1300000W) located in San-lúcar la Mayor, SW Spain. The physico-chemical properties of the

S. Vázquez et al. / Journal of Environmental Management 92 (2011) 2069e20752070

soil were highly heterogeneous due to the residual sludge content,so the research area was divided into 32 (4 � 8) subplots of 25 m2

each (discarding the last 200 m2). The Typic Xerofluvent soil waspreviously characterized (Vázquez et al., 2006). Soil monitoringconsisted of extracting five surface soil (<20 cm depth) samples (in2000, 2001, 2003, 2005, and 2006) from each of the 32 subplots.The pH, electrical conductivity (EC), organic matter content (OM),iron oxide content, and total, available, and soluble concentrationsof Fe, Mn, Cu, Zn, Al, As, and Cd were measured.

2.2. Analytical determinations

The soil samples were air-dried and sieved at the minor than2 mm. Quantitative paper-filtered (Filter-Lab, S A) 1:5 soil/waterextracts made bymixing air-dried soil and deionizedwater, shakingat 180 rpm for 1 h and waiting for 30 min prior to using a digital pHmeter (Model 8155SC; ORION Research Inc., Boston, USA) anda conductivity meter (Crisol CM 2200) (MAPA, 1994). The organicmatter content (OM) was analyzed using wet oxidation withdichromate and titration with ferrous ammonium sulfate (Walkeyand Black, 1934). Total OM was calculated using the 1.724 Wask-mann coefficient (Spanish official methods of analysis, (MAPA,1994). The Fe oxide content was obtained from the citrate-bicarbonate-dithionite (CBD)-extractable Fe fraction (Mehra andJackson, 1960). Soluble fractions were extracted from 2 g sampleswith 20 mL of 0.1 M (NH4)2SO4 (Wenzel et al., 2001); and availablesoil fractionswere extracted using 20mL of 0.02M ethylenediaminetetraacetic acid (EDTA) in 0.5 M acetic acid/ammonium acetatebuffer, pH 4.5 (Lakanen and Erviö, 1971). The extractions wereperformed in 60-mL plastic flasks shaken at 180 rpm for 4 h at roomtemperature. The total concentration of elements (pseudo-total)

Fig. 1. Temporal variation of pH (a), CE (b), OM (c) and iron oxides (d) in the soil. Diagrams redifferences among years are indicated by different characters (P < 0.05; n ¼ 32).

was assessed using hydrochloric-nitric acid (3:1) microwave-assisted digestion (Vázquez et al., 2008). The samples werefiltered through quantitative filter paper (Filter-Lab, S.A.) and thefiltrates were analyzed for metal and As content using atomicabsorption (PerkineElmer Analyst 800) and atomic fluorescencespectrometry (PS Analytical 10.055 Millennium Excalibur system).

2.3. Statistical analysis

The dataset contained 32 samples per year (n ¼ 32), corre-sponding to the 32 subplots. Despite the subdivision of the plot, thedata show a highly dispersed pattern and produced a non-Gaussiandistribution. Because of this, non-parametric statistics was used asfollows: box diagrams representing the median (line), percentiles(boxes and bars), and outliers (circles) were used to report valuesinstead of mean and standard error. Wilcoxon’s test was usedinstead of Duncan’s test to identify significant differences (a¼ 0.05)between years (represented by different letters in Figs.1 and 2). ThePearson coefficient was used to determine correlations betweenvariables (Supplementary Table 1). The spatial distribution of thevariables is depicted using diagrams resembling level curves(Fig. 4), in which relative values are plotted to facilitate comparisonbetween years and variables. The statistical analyses were per-formed using the SPSS v15.0 software package.

3. Results

3.1. Temporal variation

The pH remained quite constant (4.1e4.6), with the lowestvalues of 4.1 and 4.2 occurring in the years 2003 and 2005 (Fig. 1a).

present the median (line), percentiles (boxes and bars) and outliers (circles). Significant

Fig. 2. Temporal variation of total, EDTA-extractable and (NH4)2SO4-extractable Fe (aec), Mn (def), Cu (gei), Zn (jel), Al (meo), Cd (per) and As (seu) concentrations (mg kg�1) in the soil.Diagrams represent the median (line), percentiles (boxes and bars) and outliers (circles). Significant differences among years are indicated by different characters (P < 0.05; n ¼ 32).

S. Vázquez et al. / Journal of Environmental Management 92 (2011) 2069e2075 2071

Fig. 3. Percentage of (NH4)2SO4-extractable and EDTA-extractable element concentrations respect to the total ones.

S. Vázquez et al. / Journal of Environmental Management 92 (2011) 2069e20752072

The electrical conductivity (EC) decreased significantly with time,reaching 50% of 2000 levels by 2005 and 2006 (Fig. 1b). Organicmatter content (OM) did not display a clear trend with time; butseemed to increase toward the end of the experiment witha significant increase in 2005 (Fig. 1c). The iron oxide contentincreased significantly over time andwas 1.2-fold higher during thelast two years than in the first year (Fig. 1d).

All iron fractions (total, EDTA-extractable, and (NH4)2SO4-extractable) decreased with time, particularly toward the end of the

experiment (Fig. 2aec). Manganese (Fig. 2def), Cu (Fig. 2gei), Zn(Fig. 2jel), and As (Fig. 2seu) followed amore obvious trend than Fe,with all fractions decreasing with time, except for NH4)2SO4-extractable Cu, and the highest reductions occurring in totalconcentration (35% for Mn, 27% for Cu, 53% for Zn, and 47% for As). Incontrast, the total Cd concentration increased significantlywith time,though the behavior was different for the EDTA-extractable and(NH4)2SO4-extractable fractions (Fig. 2per). The Al concentration didnot display any clear trend with time for any fraction (Fig. 2meo).

Fig. 4. Spatial distribution of the most correlated variables (pH and (NH4)2SO4-extractable fractions of Zn, Fe, Al and As). Relative values are represented in a thermographic scale, inwhich hot colors (red) mean low pH and high levels of elements and cold ones (blue) mean the opposite.

S. Vázquez et al. / Journal of Environmental Management 92 (2011) 2069e2075 2073

S. Vázquez et al. / Journal of Environmental Management 92 (2011) 2069e20752074

3.2. Availability of elements in the soil

The availability of the elements in the soil is indicated in Fig. 3 asthe percentage of (NH4)2SO4-extractable and EDTA-extractableconcentrations with respect to the total concentration for thatelement. The elements may be sorted into two groups based ontheir availability, withMn, Cu, Zn, and Cd classified as high availableelements and Fe, Al, and As as low available elements. Theconcentrations of high available elements ((NH4)2SO4-extractable)decreased with time, with the highest reductions occurring in 2006with respect to the 2000 concentrations (50% Mn, 70% Cu, 32% Zn,and 34% Cd).

3.3. Correlation analysis

Relationships between soil properties and elemental soilconcentrations and relationships between the concentrations ofvarious elements are listed in Table 1(Supplementarymaterial). SoilpH and EC were the variables most strongly correlated to the totalconcentrations of most of the elements analyzed, particularly Fe.Also, significant correlation (P < 0.01) was obtained for pH and the(NH4)2SO4-extractable concentrations of all the elements analyzed.The correlation between element concentrations and pH wasnegative, while the correlation between EC and element concen-trations was positive. Moderate to low correlations were observedbetween the element concentration and both Fe oxide content andOM. The Fe concentration was strongly and positively correlated toAs and Al, with the highest values of Pearson coefficients corre-sponding to 0.909 for FeeAs (EDTA-extractable fractions) and 0.888for FeeAl (EDTA-extractable fractions). High correlation valueswere also observed between Fe and Cu ((NH4)2SO4-extractablefractions)and (NH4)2SO4-extractable Zn and Fe. A remarkablepositive correlation was found between Mn and Zn concentrations,particularly for the EDTA and (NH4)2SO4-extractable fractions.

3.4. Spatial variation

The most highly correlated variables (high Pearson coefficientsin Table 1, Supplementary material, mentioned above) were plottedto compare their spatial distribution in the soil and their changewith time (Fig. 4). The same patterns of spatial distribution wereobserved for pH and (NH4)2SO4-extractable Zn for each of the yearsanalyzed. Also, similar pattern of spatial distribution was observedfor the other three correlated variables, the (NH4)2SO4-extractablefractions of Fe, Al, and As. The evolution of soil pH values suggeststhat an increase in soil acidity took place over time. Likewise,a dispersion of zones containing high levels of (NH4)2SO4-extract-able Zn seems to have simultaneously occurred. For (NH4)2SO4-extractable fractions of Fe, Al, and As this pattern of evolution withtime was not so well-defined.

4. Discussion

Soil pH changed slightly with time (Fig. 1a), whereas spatially thesoil pH decreased with time (Fig. 4). This behavior could beexplained by the effect of seasonal rainfalls on soil pyrite oxidation.Under field conditions, oxidation of pyrites preferably occurs in dryseasons (samples obtained in the summers of 2000 and 2001)whereas during wet seasons (2003, 2005, and 2006 autumn andwinter samples) pyrite oxidation is hindered due to a lowerconcentration of oxygen in the soil (Vanderlinden et al., 2006). Onthe other hand, the annual precipitation was 616-494 mm between2000 and 2002 and 716e793 mm from 2003e2006 (except for2004, 322 mm, which was a very dry year), and rainfall coulddissolve sulfateminerals formed during the previous dry season. The

significant decrease in EC over time (Fig.1b) could be due to leachingof salts by the frequent rainfall occurring in the years 2003e2006,though sorption phenomenamay also have contributed (see below).Rainfall also enhances wild plant or weed growth (shoot biomassproduction has been estimated at 150g FW per m2 with a third morefor root biomass; unpublished results), which could explain theslight increase in OM content toward the end of the monitoringperiod, especially in 2005 (Fig. 1c). In any case, the existing OM-richtopsoil layers were mechanically removed together with the sludgeduring the remediation efforts in 1998, and the remaining top soilcontaining normal levels of OM (2%) is very shallow. The lower soilhorizons display a sharp decrease in OM content (Álvarez-Ayusoet al., 2008). For this reason, no correlations with OM wereobserved (Supplementary Table 1).

The increase in Fe oxides over time (Fig. 1d) indicates a contin-uous release of Fe from pyrite after oxidation and subsequentprecipitation. Effects such as natural sludge weathering thatpromote continuous oxidation of pyritic minerals and the subse-quent penetration of the oxidized products into the topsoil havebeen widely described (Clemente et al., 2006; Aguilar et al., 2007;Álvarez-Ayuso et al., 2008). Iron oxides including goethite andferrihydrite are important in influencing the mobility of As in soils.Studies using soil and pure Fe hydroxides have demonstrated thatAs solubility increases with pH within the range commonlymeasured in soil (pH 3e8) (Fitz and Wenzel, 2002). At low pH, thesolubility of As decreases due to surface charges on the Fe oxides.However, the availability of As in soils can increase under acidicconditions (mainly below 5) as the increased solubility of the Feoxides releases more immobilized As (Alloway, 1995). Due to thesesoil processes affecting the availability of As, moderate correlationsbetween Fe oxide content and As were observed (SupplementaryTable 1).

Both the strong negative correlation of pH and the strongpositive correlation of EC with the elemental composition of thesoil (Supplementary Table 1) indicate that the general decrease inmetal concentrationwith time (except for Cd and Al, Fig. 2) could bedue to continuous leaching of elements from the most availablefractions, and that the available concentration increases at lowerpH. This behavior is clearer for the total levels than for the availableand soluble fractions because the elements in these fractions arecontinuously replaced fromnon-available residual fractions as soonas the element is leached (Alloway, 1995). Aluminum is not presentin the pyritic sludge but rather is an acidity-induced pollutant. Asthe pH decreases, large amounts of Al are released through disso-lution of Al oxides (Alloway, 1995). This trait of Al could be a reasonfor the unclear pattern of concentration with time (Fig. 2meo). Theincrease in total soil Cd concentration in later years (Fig. 3p) couldbe a consequence of increasing soil acidity. This would be inagreement with Álvarez-Ayuso et al. (2008) who reported a verticaldistribution of elements across the soil profile, with the greatestlevels of Cd and Zn occurring in deeper soil layers (50e60 cm) buthigher leachable Cd and Zn levels in the upper layers (10e20 cmand 20e30 cm). Therefore, environmental monitoring of adjacentareas and groundwater is still strongly recommended.

In 2000 all trace elements except As were below the maximumpermissible levels. The general trend of decreasing levels of metalsexcept for Cd and Al over time seems to indicate a natural atten-uation ability for metals in these soils. In the last year, the totalconcentration of As was decreased by 47% with respect to 2000level, reaching amedian value of 99.8mg kg�1, that is just under themaximum permissible level (Supplementary Table 2).

In addition to the total concentration, the availability of a givenelement is essential to assessing the potential risk. The high (Mn,Cu, Zn and Cd) and low (Fe, Al and As) availability of elements aredetermined by the actual soil conditions such as pH, soil type, etc.

S. Vázquez et al. / Journal of Environmental Management 92 (2011) 2069e2075 2075

In this study, leaching processes are important for Mn, Cu, Zn, andCd (high availability elements) as the amount capable of beingmobilized is greater than for low availability elements (Fe, Al, As).This behavior has beenwidely reported (Kraus andWiegand, 2006;Vanderlinden et al., 2006; Aguilar et al., 2007). There were alsosignificant reductions in the (NH4)2SO4-extractable percentages ofMn, and Zn, more fluctuating for Cu and Cd during the later years(Fig. 3b, c, d and f), indicating that the natural attenuation effect notonly acts on the total concentration but also on element availability.Similar decreases in availability were observed by Clemente et al.(2006), though the total concentration of As was not reducedduring their experiments.

The spatial distributions may be organized into two groups, pH-(NH4)2SO4-extractable Zn and (NH4)2SO4-extractable fractions ofFeeAleAs (Fig. 4). The variables comprising these groups were alsothose displaying higher Pearson coefficients (SupplementaryTable 1). Therefore, the spatial distribution was closely related tothe correlation analysis. The results pertaining to the pH-(NH4)2SO4-extractable Zn group were in accordance with Vanderlinden et al.(2006), who described an inverse relationship between pH and Zndue to precipitation and further dissolution of minerals formed fromthe products of pyrite oxidation. The high correlation and similarspatial distribution observed for Fe, Al, and As (SupplementaryTable 1; Fig. 4) are due to the close relationship of these elementsand the soil chemistry, particularly under acidic conditions(Clemente et al., 2006; Aguilar et al., 2007; Álvarez-Ayuso et al.,2008). Evolution of the spatial distribution with time indicateda progressive and continuous soil acidification, together witha significant dispersionof available elements similar to thatdescribedfor Zn. In contrast, the evolution of the dispersion of non-availableelements such as Fe, Al, and As was not as clear (Fig. 4). This wouldsuggest that elemental availability is the main factor controllingmobility in soils, and highly available elements are themost sensitiveto dispersion-driving forces (acidification and leaching) in sludge-contaminated soil. Therefore, correlation and spatial distributionanalysis are complementary studies allowing us to establish rela-tionships between variables (soil properties and elements) and toassess the evolution of these variables with time. This in turn enablesus to study the dispersion of contamination in a given area.

5. Conclusions

A capacity for natural attenuation was identified in the pyriticsludge-contaminated soils of the Vicario area (Guadiamar basin),responsible for reductions in both total element concentration(particularlyAs)belowmaximumpermissible levels, and (NH4)2SO4-extractable percentages of Mn, Cu, Zn, and Cd. Soil acidification bypyrite oxidation and leaching by rainfall were the main factorscontributing to temporal and spatial evolution of thehighly availableelements Mn, Cu, Zn and Cd, and of the low availability elements Fe,Al, and As to a lesser extent. Periodic use of correlation and spatialdistribution analysis proved to be useful tools in studying thedispersion of elements/contaminants in pyritic-contaminated areas.

Acknowledgments

Financial support from the Spanish MICINN (CTM 2007-66401-CO2-02/TECNO; CTM 2004-06715-CO2-01), and from Comunidad

de Madrid (EIADES S2009/AMB-1478) is acknowledged. Authorsare grateful for the experimental plot cession in “El Vicario” by“Consejería de Medio Ambiente, Junta de Andalucía.”

Appendix. Supplementary data

Supplementary data related to this article can be found onlineat doi:10.1016/j.jenvman.2011.03.030.

References

Aguilar, J., Dorronsoro, C., Fernández, E., Fernández, J., García, I., Martín, F., Sierra, M.,Simón, M., 2007. Arsenic contamination in soils affected by a pyrite-mine spill(Aznalcóllar, SW Spain). Water Air Soil Pollut. 180, 271e281.

Alloway, B.J., 1995. Soil processes and the behaviour of metals. In:Alloway, B.J. (Ed.), Heavy Metals in Soils. Blackie Academic & Profesional,London, pp. 11e36.

Álvarez-Ayuso, E., García-Sánchez, A., Querol, X., Moyano, A., 2008. Trace elementmobility in soils seven years after the Aznalcóllar mine spill. Chemosphere 73,1240e1246.

Bernal, M.P., Clemente, R., Vázquez, S., Walker, D.J., 2007. Aplicación de la fitorre-mediación a los suelos contaminados por metales pesados en Aznalcóllar.Ecosistemas 2, 68e82.

Burgos, P., Madejón, E., Pérez-de-Mora, A., Cabrera, F., 2006. Spatial variability of thechemical characteristics of a trace-element-contaminated soil before and afterremediation. Geoderma 130, 157e175.

Clemente, R., Walker, D.J., Bernal, M.P., 2005. Uptake of heavy metals and As byBrassica juncea grown in a contaminated soil in Aznalcóllar (Spain): the effect ofsoil amendments. Environ. Pollut. 138, 46e58.

Clemente, R., Almela, C., Bernal, M.P., 2006. A remediation strategy based on activephytoremediation followed by natural attenuation in a soil contaminated bypyrite waste. Environ. Pollut. 143, 397e406.

EPA, 1999. Use of Monitored Natural Attenuation at Superfund, Rcra CorrectiveAction, and Underground Storage Tank Sites. Oswer Directive. 9200.4e17P.

Fitz, W.J., Wenzel, W.W., 2002. Arsenic transformations in the soil-rhizosphere-plant system: fundamentals and potential application to phytoremediation.J. Biotechnol. 99, 259e278.

Kraus, U., Wiegand, J., 2006. Long-term effects of the Aznalcóllar mine spill-heavymetal content and mobility in soils and sediments of the Guadiamar rivervalley (SW Spain). Sci. Total Environ. 367, 855e871.

Lakanen, E., Erviö, R., 1971. A comparison of eight extractants for the determinationof plant available micronutrients in soils. Acta Agric. Scand. 17, 131e139.

MAPA (Ministerio de Agricultura Pesca y Alimentación), 1994. Métodos oficiales deanálisis (spanish), third ed.. SGT-MAPA, Madrid, Spain.

Mehra, O.P., Jackson, M.L., 1960. Iron oxide removal from soils and clays bya dithionite-citrate system buffered with sodium bicarbonate. Clays Clay Miner7, 317e327.

Ordóñez, R., Giráldez, J.V., Vanderlinden, K., Carbonell, R., González, P., 2007.Temporal and spatial monitoring of the pH and heavy metals in a soil pollutedby mine spill. Post cleaning effects. Water Air Soil Pollut. 178, 229e243.

Vanderlinden, K., Polo, M.J., Ordóñez, R., Giraldez, J.V., 2006. Spatiotemporalevolution of soil pH and zinc after the Aznalcóllar mine spill. J. Environ. Qual.35, 37e49.

Vázquez, S., Agha, R., Granado, A., Sarro, M.J., Esteban, E., Peñalosa, J.M.,Carpena, R.O., 2006. Use of white lupin plant for phytostabilization of Cd and Aspolluted acid soil. Water Air Soil Pollut. 177, 349e365.

Vázquez, S., Moreno, E., Carpena, R.O., 2008. Bioavailability of metals and As fromacidified multi-contaminated soils: validation of several extraction methods bylupin plants. Environ. Geochem. Health 30, 193e198.

Walker, D.J., Clemente, R., Bernal, M.P., 2004. Contrasting effects of manure andcompost on soil pH, heavy metal availability and growth of Chenopodiumalbum L. in a soil contaminated by pyritic mine waste. Chemosphere 57,215e224.

Walkey, A., Black, I.A., 1934. An examination of the Degtjareff method for deter-mining soil organic matter and proposed modification of the chromic acidtitration method. Soil Sci. 37, 29e38.

Wenzel, W.W., Kirchbaumer, N., Prohaska, T., Stingeder, G., Lombi, E., Adriano, D.C.,2001. Arsenic fractionation in soils using an improved sequential extractionprocedure. Anal. Chim. Acta 436, 309e323.