This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

Author's personal copy

Mercury contamination in the vicinity of a chlor-alkali plantand potential risks to local population

Ana Teresa Reis⁎, Sónia Morais Rodrigues, Clarisse Araújo, João Pedro Coelho,Eduarda Pereira, Armando C. DuarteCESAM and Chemistry Department, University of Aveiro, Campus de Santiago, 3810-193, Aveiro, Portugal

A R T I C L E D A T A A B S T R A C T

Article history:Received 12 May 2008Received in revised form27 October 2008Accepted 31 October 2008Available online 11 February 2009

A mercury-cell chlor-alkali plant operated in Estarreja (North-western Portugal) for 50 yearscausing widespread environmental contamination. Although production by this processceased in 2002, mercury contamination from the plant remains significant. The mainobjective of this study was to investigate mercury impact on the nearby environment andpotential risks to local population. To assess the level of contamination soil samples werecollected from agricultural fields in the vicinity of the plant, extending the study by takingsamples of the predominant vegetation suitable for animal and human consumption, watersamples, and fish species from a nearby coastal lagoon, to gain a preliminary insight into thepotential for contamination of the terrestrial and aquatic food web. To determinepopulation exposure to mercury, hair samples were collected from local residents. Totalmercury concentration in the 0–15 cm layer of soil was found to be highly variable, rangingbetween 0.010 and 91 mg kg−1, although mercury contamination of soils was found to berestricted to a confined area. Lolium perenne roots contained between 0.0070 and 2.0 mg kg−1,and there is evidence that root systems uptake mercury from the soil. Levels of mercury inthe aerial parts of plants ranged between 0.018 and 0.98 mg kg−1. It appears that plants withhigher mercury concentration in soils and roots also display higher mercury concentrationin leaves.Total mercury concentration in water samples ranged between 12 and 846 ng L−1, allsamples presenting concentrations below the maximum level allowable for drinking waterdefined in the Portuguese law (1.0 μg L−1).Mercury levels in fish samples were below themaximum limit defined in the Portuguese law(0.5 mg kg−1), ranging from 0.0040 to 0.24 mg kg−1. Vegetables collected presentedmaximummercury concentration of 0.17 mg kg−1. In general, food is not contaminated and should notbe responsible for major human exposure to the metal.Mercury determined in human hair samples (0.090–4.2 mg kg−1; mean 1.5 mg kg−1) can beconsidered within normal limits, according to WHO guidelines suggesting that it is notaffecting the local population. Despite being subject to decades of mercury emissions,nowadays this pollutant is only found in limited small areas and must not constitute a riskfor human health, should these areas be restricted and monitored.Considering the present data, it appears that the population from Estarreja is currently notbeing affected by mercury levels that still remain in the environment.

© 2008 Elsevier B.V. All rights reserved.

Keywords:Total mercuryChlor-alkali plantEnvironmental pollutionPublic health

S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 9 ) 2 6 8 9 – 2 7 0 0

⁎ Corresponding author. Tel.: +351 234 370 737; fax: +351 234 370 084.E-mail address: teresareis@ua.pt (A.T. Reis).

0048-9697/$ – see front matter © 2008 Elsevier B.V. All rights reserved.doi:10.1016/j.scitotenv.2008.10.065

ava i l ab l e a t www.sc i enced i rec t . com

www.e l sev i e r. com/ loca te / sc i to tenv

Author's personal copy

1. Introduction

Environmental contamination by mercury (Hg) is a majorconcern because of the impact of human activities on itscycling, toxicity and bioconcentration potential (Lindberg,1998). Natural sources of mercury include volcanic eruptions,crustal degassing, and emissions from forest, lakes andoceans (Dommergue et al., 2002). It has often been suggestedthat anthropogenic emissions (industry and mining, amongothers) are leading to a general increase in mercury on local,regional and global scales.

Being a potent neurotoxin, mercury is particularly hazar-dous to developing fetuses, infants and young children, withresulting effects including delays in learning and motorfunctions (walking, talking and speaking). Mercury builds upin the body over time, causing impaired peripheral vision,disturbances in sensations (tingling, numbness) usually in thehands and feet and sometimes around the mouth, impairedspeech, hearing, walking, writing, as well as mental distur-bances. If additional exposure does not occur, it can beexcreted from the body over a period of four to five months(USEPA, 1999).

Edible plants and vegetables constitute an important partof the human diet, as they provide carbohydrates and proteinsas well as vitamins, minerals and trace elements. Leafyvegetables such as spinach, lettuce, chards and beet greensare classified as high elemental uptake crops (USEPA, 1983). Inturn, they can affect humanhealth through intake via the foodweb, particularly if one considers the persistence of heavymetals in the environment.

The industrial cluster north from the city of Estarreja,western coast of Portugal, dates back to 1950 (Inácio et al.,1998) and is home to a large chlor-alkali plant which used toproduced chlorine and caustic soda by the mercury cellprocess, where liquid elemental mercury is utilized as acathode in the electrolysis of a saturated brine solution(Ullrich et al., 2007). As many studies show (Lacerda andSalomons, 1998; Ullrich et al., 2007) mercury-cell chlor-alkaliplants have been identified asmajor sources of mercury to theenvironment. Although the plant started to change theproduction process in 1994 and completely ceased the use ofmercury in 2002 (Ospar Commission, 2006), mercury emittedfrom the existing plant still remains significant in thesurrounding environment. This problem becomes even amajor issue because until 1975 liquid effluents produced bythese industrial units were discharged directly into severalnearby streams that cross agricultural fields. These effluentscontainedmany different types of contaminants (Batista et al.,2002), including mercury, that may reach the crops and,consequently, animals and humans through the food web.The industrial units also dumped a large amount of solidwaste, which was stockpiled, directly on the surrounding land(Batista et al., 2002). Additionally, it is known that through theyears, large amounts of mercury have reached Ria de Aveiro, anearby coastal lagoon, due to wastewater discharges. Giventhat Ria is regularly used by fishermenwho catch fish both fortheir own consumption and for selling it on local markets, thiscontamination raises potentially serious health concerns, asmercury can easily accumulate and biomagnify in aquatic

biota and enter the food web (Clarkson, 1998). The consump-tion of contaminated fish is also generally regarded as themain route of human exposure to methylmercury (Clarkson,2003; USEPA, 2005).

Due to high levels of mercury found in Ria de Aveiro, thisarea and its surroundings have been extensively studied, withparticular focus on sediments (Hall et al., 1987; Ramalhosaet al., 2006b) and aquatic organisms (Coelho et al., 2006; Válegaet al., 2006). An effort has been made to monitor economicallyimportant estuarine resources, both invertebrate (Coelho etal., 2006; Pereira et al., 2006) and vertebrate species (Abreu etal., 2000) directly consumed by local populations, in anattempt to clarify the inherent health hazards. But, althoughthe aquatic environment is well characterized there's still lackof information related to mercury contamination on theterrestrial environment, which may also be affectingpopulation.

The main objective of this study was to assess the presentimpact of decades of mercury emissions from the chlor-alkaliplant on local populations. In order to do so, a study on thesurrounding environment took place, considering agriculturalfields, predominant vegetation suitable for animal and humanconsumption, water samples from wells used to water thefields, and fish from Ria de Aveiro. This holistic approachprovides an insight into the potential for contamination of theterrestrial and aquatic food webs and the associated level ofrisk for human consumers. Mercury levels of residents weredetermined in hair, since human hair has been considered auseful bio-indicator (Lacerda and Salomons, 1998).

2. Materials and methods

2.1. Site description

This study was conducted in Estarreja, a small town with apopulation of around 7800 inhabitants that is located in thecentre of Portugal, 14.5 km North East from Aveiro (Fig. 1). AChemical Industrial Complex is situated 1 km north from thecentre of the town and, among other industries, this includes achlor-alkali plant. Until 1975, the liquid effluents from thisplant were discharged directly into manmade effluentstreams (“Vala de S. Filipe”, “Vala da Breja” and “Vala doCanedo”). Consequently, the pollutants were transported forseveral kilometres through the agricultural fields surroundingthe Complex, mainly to the river branch named Esteiro deEstarreja (Costa and Jesus-Rydin, 2001). Pollution caused bythese effluents became a major concern, since they weredirectly discharged in Ria de Aveiro, a nearby coastal lagoon,without any previous treatment. Although after 1975impermeable pipes were constructed, the streams (no longerused for effluent transport) are still present in fields that aremostly used for agricultural and livestock purposes. It is alsocommon to find wells, since in this region every householdused to have its own well for drinking water and agriculturalsupply that rarely exceeded the depth of 20–25 m. The highpermeability of the sandy deposits and shallow groundwatertable made the aquifer extremely vulnerable to agronomic,urban and industrial pollution. Eventually, the presence ofnitrates, chlorates, heavy metals and organic compounds made

2690 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 9 ) 2 6 8 9 – 2 7 0 0

Author's personal copy

the domestic supply from this source inadvisable. Nevertheless,despite some casualties among cattle, even nowadays the use ofwater from shallow wells for agricultural and even drinkingpurposes is not uncommon (Costa and Jesus-Rydin, 2001).

Sampling in this study took place only in agricultural fieldsand it was divided into two distinct areas (Fig. 2) that distance,at least, 2 km from each other. Area of Study 1 is located inagricultural fields located in the immediate vicinity of thechlor-alkali plant (9 sampling sites), whereas area of Study 2 isplaced in the centre of the city (6 sampling sites).

2.2. Sampling and sample processing

2.2.1. SoilA total of 139 soil samples were collected in 15 locations alongthe northern, eastern, southern and western area around thechlor-alkali plant. In small fields, the samples were randomlycollected, but in larger fields a common grid (25×25 m) wasused. Sampling was performed using a plastic spatula andsamples were placed in plastic bags during transport tothe laboratory for pre-treatment. The soil sampling depthwas 0–15 cm. Once in the laboratory, soil samples were airdried at room temperature, until constant weight. Stones wereremoved and soil aggregates were crushed and homogenised,during the drying stage. The dried samples were sieved tob2 mm using a nylon sieve.

Parameters such as soil pH (CaCl2) (WTW pH meter-538,according to ISO 10390:1994 method), total and organic carbonpercentage (Elemental Analysis, LECO CNH-2000, according toISO 10694:1995) were determined on the b2 mm fraction forgeneral characterization of these soil samples.

2.2.2. FeedstuffA total of 94 samples of one species of forage plant (Loliumperenne, the predominant forage plant available on location)were collected in fields surrounding the chlor-alkali plant andplaced in plastic bags during transport to the laboratory. Theywere thoroughly cleaned with tap water and a brush toremove all traces of dust and soil from the roots and leaves,rinsed with distilled water and dried, first at room tempera-ture, and later at 40 °C in an oven until constant weight. Thedried samples were grinded and stored in plastic bags, at roomtemperature, until analysis.

2.2.3. WaterA total of six water samples were taken from six different wellscurrently only used for irrigation purposes. Bottles used forsampling were rinsed three timeswith sample water before thefinal sample was collected. In addition, one water sample wastaken from a small water stream using the same procedure.

Water samples were filtered with 0.45 μm Millipore filters.The filtrate was then acidified with HNO3 (Merck, mercury-

Fig. 1 –Map showing sample distribution in Estarreja.

2691S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 9 ) 2 6 8 9 – 2 7 0 0

Author's personal copy

free) to pH b2 and stored at 4 °C until analysis. Water sampletreatment and analysis were performed using ultra-cleanprotocols (adapted from Bloom, 1995). Ultra-pure water wasobtained from a Millipore Milli-Q model 185 system.

2.2.4. VegetablesSamples of common vegetable edible species were purchasedin Estarreja's farmer's market and collected in fields aroundthe chlor-alkali plant, then transported to the laboratory inplastic bags. Samples analysed included Brassica oleraceacapitata (cabbage, n=6), Spinacia oleracea (spinach, n=5), Latucasativa (lettuce, n=3), Nasturtium officinale (watercress, n=1),Brassica napus (rapeseed, n=2) and Brassica oleracea acephala(collards, n=4). The 21 samples collected were thoroughlywashedwith tapwater and a brush to remove all traces of dust

and dirt, rinsed with distilled water and dried at 40 °C untilconstant weight. The dried samples were grinded into powderand stored in plastic bags, at room temperature, until analysis.

2.2.5. FishFish samples, in a total of 30, were purchased in Estarreja'smarket, fisherman and local supermarkets. Species analysedwere Chelon labrosus (Thicklip grey mullet, n=3), Platichthysflesus (Flounder, n=8), Scophthalmus rhombus (Brill, n=1), Soleasolea (Common sole, n=1), Trachurus trachurus (Atlantic horsemackerel, n=8), Sparus aurata (Gilthead seabream, n=1),Scomber scombrus (Atlantic mackerel, n=3), Trisopterus luscus(Pouting, n=4) and Alosa fallax (Twaite shad, n=1). All ediblemuscle tissue was removed for analysis with a stainless steelknife, freeze dried and later homogenized before analysis.

Fig. 2 –Study areas and mercury levels found in soil samples.

2692 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 9 ) 2 6 8 9 – 2 7 0 0

Author's personal copy

2.2.6. HairFor this study 45 donors were considered. Hair from theoccipital portion of the head was cut close to the scalp, placedin an identified plastic bag and stapled to prevent cross-contamination of samples. Each donor was given a ques-tionnaire to gather the most possible information concerningwork, food habits, age, health condition and hair treatments.

Hair samples were thoroughly washed in Triton X-100 (1%)for 1 h, then left in distilled water for 1 h and dried at roomtemperature. A lock of 20 cm, maximum, was analysed.

2.3. Analytical techniques

2.3.1. Dissolved mercuryDissolved reactive mercury concentrations in water sampleswere measured by Cold Vapour Atomic Fluorescence Spectro-metry (CV-AFS; PSA model Merlin 10.023 equipped with adetector PSA model 10.003) using tin(II) chloride as a reducingagent. The same method was used for total dissolved mercury,after addition of 500 μL of a saturated solution of potassiumpersulfate to 50 mL of filtered water and irradiation with a UVlamp (1000 W) for 30 min. Following irradiation, the excessoxidant was reduced with 37.5 μL of 12% hydroxylaminesolution (w/v) prior to analysis (Mucci et al., 1995). Theequipment was calibrated every day with acidified (HNO3

mercury-free) standard solutions prepared from a 1000 mg L−1

solution (BDH). The detection limit of the CV-AFS technique fordissolved reactivemercury was 1.4 ng L−1, and for totalmercurywas 2.3 ng L−1. Procedure blankswere always runwith samples,being their contribution corrected when necessary.

2.3.2. Total mercury in solid samplesThe determination of total mercury concentrations wascarried out by thermal decomposition atomic absorptionspectroscopy with gold amalgamation (LECO model AMA-254), a rapid total mercury determination method (Costleyet al., 2000). This technique has particular advantages formercury analysis, as determinations are performed directly onsolid samples without sample pre-treatment or digestion,avoiding mercury losses and contamination during digestion.It is also free from matrix interferences. The method is basedon the thermal decomposition of the solid sample in acombustion tube at 750 °C, in an oxygen atmosphere;collection of elementalmercury vapour in a gold amalgamatorand detection by atomic absorption spectrometry (AAS). Atleast three soil replicate measurements were carried out foreach sample.

2.3.3. Quality assuranceAnalytical quality control was performed by using CertifiedReferenceMaterials (CRMs)BCR-141R (average recovery: 95±10%at 0.05 significance level) and Vector Soil3 (average recovery:114±2.6% at 0.05 significance level) for soils, BCR-60 (averagerecovery: 99±9.0% at 0.05 significance level) for feedstuff andvegetables, TORT-2 (average recovery: 106±4.9% at 0.05significance level) and DORM-2 (average recovery: 101±3.0%at 0.05 significance level) for fish and BCR-397 for human hair(average recovery: 93±4.2% at 0.05 significance level). AllCertified Reference Materials had recoveries within thequantitative acceptance range of 80–120%.

3. Results

3.1. Soil

Mercury content in the 139 soil samples collected in Estarrejawas found to be highly variable, ranging between 0.01 and90.8 mg kg−1 (mean 5.4 mg kg−1; median 0.41 mg kg−1). Theseresults show the significant spatial variability occurringin this type of sites. The low median mercury concentrationof 0.41 mg kg−1 indicates that only a comparatively smallnumber of samples show a high enrichment in mercury.Differences of almost 103 times were obtained betweenmaximum and minimum concentrations. This confirmsthat total mercury concentrations are highly variable andthat these agricultural soils are very heterogeneous media.Fig. 2 shows an overall distribution of samples and theirmercury concentration in areas of Study 1 and 2.

Area of Study 1 is situated in the immediate surroundingsof the industrial complex. This area is characterized byagricultural fields, mainly used for cattle grazing. For manyyears an “open air” stream (“Vala de S. Filipe”) that crossesthese fields was used to carry the effluents from the chlor-alkali plant without any previous treatment (Ferreira, 1993).Although the effluents are no longer transported in thestreams and discharged in Ria de Aveiro, due to mercurypersistence in soils, high concentrations of the metal can stillbe found. Results show that mercury concentrations in soilnear the stream are much higher than any other location. Infact, it can be easily observed that there is a distinct area ofhigh concentration parallel to the stream (Fig. 2). Mercuryconcentrations in these samples ranged from 4.6 to 91 mgkg−1, the mean and median value being 27 and 22 mg kg−1,respectively. On the other hand, concentrations lower than1.0 mg kg−1 can be found just 25 m away from the stream.These results are highlighted in area of Study 1 in Fig. 2.

Samples taken in agricultural lands in the urban area of thecity (Area of Study 2, Fig. 2) show a median mercuryconcentration of 0.54 mg kg−1. Although this value is higherthan the world soil reference median (0.05 mg kg−1) (Reimann,1998) it is quite similar to the one found in urban areas of manycities around the world (Table 1), leading to the conclusion thatthis can be considered a normal value for mercury concentra-tion in an industrialized urban area. The town of Aveiro islocated 15 km South West and is probably the biggest urbancentre near Estarreja. Table 1 shows that the median mercuryconcentration in Estarreja is about 5 times higher than the onefound in Aveiro (0.091 mg kg−1), reflecting the influence of thechlor-alkali plant in its immediate surroundings.

The pH of all samples varied from acidic to slightly alkaline(3.9–7.0), although the median pH was 4.8 and only 7% of thevalues had pHN6. Considering that explaining the variation ofsoil pH was not a goal for this study, the authors found noreason to further explore this aspect. However, the relationbetween pH and mercury concentration was verified with nocorrelation between the two variables being found. These soilsmay also be consideredpoor in organicmatter since themedianpercentage of total carbon was 2.6%. Also, the medianpercentage of organic carbon was 2.1%, clearing showing thelow carbonate content of these soils.

2693S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 9 ) 2 6 8 9 – 2 7 0 0

Author's personal copy

Values obtained for organic carbon and total mercury arenot significantly correlated. In fact, sampling points showinghigher total mercury values do not correspond to high organiccarbon content, as one could expect. From these data it can beconcluded that the streams used to transport the effluents arethe primary reason for mercury contamination in soilssurrounding the chlor-alkali plant, rather than soil character-istics, but the influence of atmospheric deposition in theseresults can not be disregard, since atmosphere plays anessential role in mercury cycle and exchanges with terrestrialcompartment.

Comparing these results with the ones previously found byInácio et al. (1998) (range 0.12–49 mg kg−1; mean 0.82 mg kg−1;median 0.59 mg kg−1) in a study at the same area, but using aregular grid with an interval of 500 m near the industrializedcore and 1000m in the area outside, it can be observed that themercury range in soils from the present study is broader thanthe one determined by these authors. Median mercuryconcentrations are similar, but the higher levels found in the

present study result from a more intensive collection ofsamples around the “Vala de S. Filipe”, themost contaminatedspot in the area near the industrial core and which was notidentified in past studies using the 500 m grid.

Data on soil mercury concentration are rather scarce inliterature. Table 2 reports values found in soils in the vicinityof chlor-alkali plants around the world. The mercury levelsfound in Estarreja agricultural soils are lower than the onesfound by Bernaus et al. (2006) in Holland, but, in general, arehigher than concentrations reported in the other studies,especially in what concerns maximum values.

3.2. Feedstuff

Food of animal origin is the greatest source of humanexposure to mercury and animal feeds may be an importantsource of contamination to livestock. In this case, grass-fedanimals in highly contaminated areas may give rise to unsafefood products. Therefore, plant materials, like Lolium perenne,growing in areas with high levels of mercury that are used asfeed may also lead to unacceptably high levels of contamina-tion in food products of animal origin (Speedy, 2001). Mercuryincorporated into feed at any stage of production up to thepoint of feeding, either intentionally or unintentionally, canresult in hazards in food of animal origin (Speedy, 2001) andconsequently in humans, which is why it is important tomonitor mercury levels since early stages of production.

Mercury concentration in Lolium perenne root systemranged widely between 0.0070 and 2.0 mg kg− 1 (mean0.38 mg kg−1 and median 0.063 mg kg−1), while in aerial part(leaves) concentration ranged from 0.018 to 0.98 mg kg−1

(mean 0.16 mg kg−1 and median 0.067 mg kg−1)-Table 3. Themost likely explanation for the presence of high levels of

Table 2 – Total mercury (mg kg−1) concentrations in soilsin the vicinity of chlor-alkali plants from other studies

Location Range Mean Reference

Estarreja, Portugal 0.010–91 5.4 This studyEstarreja, Portugal 0.18–49 0.82 Inácio et al. (1998)Holland, 2004 4.3–1150 – Bernaus et al. (2006)Lake Balkyldak,Kazakhstan

0.22–5.72 2.7 Ullrich et al. (2007)

River Irtysh,Kazakhstan

Ullrich et al. (2007)

Core 1: 0–10 cm 0.11–0.32 –Core 2: 0–10 cm – 0.13

Livorno, Italy 0.070–0.61 0.38 Maserti and Ferrara (1991)USA 0.70–9.4 4.7 Southworth et al. (2004)

Table 1 – Total mercury concentrations (mg kg−1) in soilsfrom other studies

Soil type/land use

Location Mean/Median

Range Reference

Soil worldmedian

Worldwide 0.05b – Reimann and Caritat(1998)

Urban soil Palermo,Italy

0.68b 0.04–6.96

Manta et al. (2002)

Urban soil Pittsburg,USA

0.51a – Carey et al. (1980)

Urban soil Aveiro,Portugal

0.091b 0.015–0.50

Rodrigues et al.(2006b)

Urban soil Glasgow,Scotland

1.6a; 1.8b 0.31–5,2

Rodrigues et al.(2006a)

Urban soil Ljubljana,Slovenia

0.41a;0.38b

0.15–0.86

Rodrigues et al.(2006a)

Urban soil Sevilla,Spain

0.42a;0.30b

0.11–1.3

Rodrigues et al.(2006a)

Urban soil Torino,Italy

0.48a;0.47b

0.21–0.90

Rodrigues et al.(2006a)

Urban soil Uppsala,Sweden

0.35a;0.25b

0.015–1.2

Rodrigues et al.(2006a)

Agriculturalsoil

Canada 0.05b 0.005–0.13

Reimann and Caritat(1998)

⁎ aMean; bMedian

Table 3 – Hg levels (mg kg−1, dry weight) in soil, Loliumperenne, vegetables, fish and hair samples in Estarreja

Sample n Mean Median stddev

Min Max

Soil 139 5.4 0.41 14 0.010 91Area of Study 1

(N25 m away fromthe stream)

96 0.41 0.29 0.40 0.010 2.0

Area of Study 1(b25 m away fromthe stream)

26 27 22 22 4.6 91

Area of Study 2 17 0.51 0.52 0.21 0.12 0.88Lolium perenne 94 0.28 0.060 0.62 0.0070 4.8Complete (root

and aerial part)26 0.30 0.040 0.97 0.014 4.8

Roots 32 0.38 0.063 0.56 0.0070 2.0Aerial part 36 0.16 0.067 0.22 0.018 0.98

Vegetables 21 0.079 0.070 0.053 0.010 0.25Field 6 0.10 0.085 0.078 0.020 0.25Market 15 0.069 0.070 0.039 0.010 0.17

Fish 30 0.092 0.050 0.095 0.010 0.42Ria de Aveiro 12 0.12 0.060 0.13 0.010 0.42Ocean 18 0.080 0.050 0.063 0.030 0.070

Hair 45 1.5 1.3 0.91 0.090 4.2Male 19 1.4 1.3 0.90 0.19 3.4Female 26 1.6 1.4 0.92 0.090 4.2

2694 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 9 ) 2 6 8 9 – 2 7 0 0

Author's personal copy

mercury in the root system of Lolium perenne is direct uptakefrom contaminated soil (Maserti and Ferrara, 1991). From Fig. 3an uptake of mercury by the plant roots depending on theconcentration of the metal in the soil appears to be evident,confirming the Maserti and Ferrara's hypothesis, with acorrelation factor equal to 0.57 (p=0.009). The uptake by theroots also seems to be restricted by the concentration andavailability of the metal in soil and accumulation depends ontissues age, with older tissues having higher mercury levels(Cooper and Gillespie, 2001). The plants in this study wereundeveloped which could be the reason why, generally,low levels of mercury were found, even in plants growing onhighly contaminated soils. Anthropogenic mercury dischargedfrom chlor-alkali plant appears to accumulate in the parts thatinteract with contaminated soil, but translocation to the aerialparts in some species appears to be insignificant. However thepossibility of atmospheric assimilation and deposition ofmercury onto the leaves of plants should not be excluded(Caille et al., 2004; Ericksen et al., 2003; Lindberg et al.,1992, Maserti and Ferrara, 1991), because atmospheric transport

ofmercurymight be relevant to the study area due to proximityof the anthropogenic source (Pio et al., 1986; Southworth et al.,2004). Nevertheless, in the present case, when plottingmercuryconcentrations in root system against aerial part of the samesample (Fig. 3) it appears evident that there is some relationbetween the two (r=0.68; p=1.2×10−4). High mercury concen-tration in roots implies higher mercury concentration in leavesand these concentrations are also related to soil levels. Thismay be due to efficient translocation of mercury to the aerialparts in Lolium perenne species. On the other hand, values foundin aerial parts of the plants could be a result of atmosphericdeposition. In this case, mercury concentration in the corre-sponding surface soil sample could partially be a result of thisprocess, since atmospheric deposition can too contribute tolocal soil contamination. To justify the presence of mercury inleaves, one should also not ignore the possibility of metal re-emission from soil (Biester et al., 2002b) and consequentabsorption of that re-emitted mercury by leaves.

The European Commission (2002) has proposed a max-imummercurycontent of 0.1mgkg−1 in feedstuffs intended for

Fig. 3 –Total mercury levels (mg kg−1) in Lolium perenne roots and aerial parts at different mercury levels in soil.

2695S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 9 ) 2 6 8 9 – 2 7 0 0

Author's personal copy

animal nutrition with a relative moisture content of 12%(equivalent to 0.11 mg kg−1 dry weight), and 0.2 mg kg−1 forcomplementary feedstuffs. Of the Lolium perenne samplesanalysed in the current study, some exceed maximum allow-able levels ofmercury in animal feedstuffs. Given the direct linkbetween feed safety and safety of foods of animal origin, it isessential that feed production and manufacture be consideredas an integral part of the food production web (Speedy, 2001).Feed production must therefore be subject, in the same way asfood production, to quality assurance in order to guarantee theconsumer's safety. Therefore, locals where high mercuryconcentrations can be found should be prohibited for feedstuffgrowth, in order to avoid input of the metal in the soil-plant-herbivore-carnivore food chain. In the case of Estarreja, landclose to “Vala de S. Filipe”, where high mercury levels werefound, should be considered unsuitable for growing cultures.

3.3. Water

Total and reactive mercury concentrations in water samplesare shown in Table 4. Total mercury concentration rangedbetween 12 and 847 ng L−1, all samples presenting mercuryconcentrations below the maximum level allowable for drink-ing water defined in the Portuguese law (1.0 μg L−1). Althoughresults obtained for water samples in Estarreja are not as highas the ones found by Ullrich et al. (2007) in water samples fromLake Balkyldak, Northern Kazakhstan (0.11–1.39 μg L−1), theyare comparable to the ones found in other studies regardingchlor-alkali plants as mercury source (Ullrich et al., 2007).

The values found in this study may be a result of years ofmercury emissions to both water systems and air in combina-tion with leaching of heavymetals from the contaminated soilsto groundwater. This area is mainly agricultural and most ofthese wells are used for field irrigation, which can become aproblem when mercury species dissolved in water reach thecrops and accumulate through the food web. Well 2 has thehighest total and reactivemercury concentrations, which couldbe explained by it's proximity to the stream “Vala de S. Filipe”.Mercury present in effluents may have reached groundwaterthrough permeable soil polluting the water system.

3.4. Vegetables

During this study, 21 samples of the following species werecollected: Brassica oleracea capitata, Spinacia oleracea, Latucasativa, Nasturtium officinale, Brassica napus and Brassica oleracea

acephala. All species mentioned are extensively cultivatedand highly consumed in Portugal and are generally present ineveryday's meals. Mercury levels found varied between 0.010and 0.25 mg kg−1, with 0.079 mg kg−1 as mean value and0.070 mg kg−1 as median value (Table 3).

One intention was also to compare if vegetables bought inthe local market (which are produced in small villages nearEstarreja) differ from vegetables cultivated in lands surround-ing the chlor-alkali plant. Mean mercury concentrations invegetables from the first and the latter were 0.069 and 0.10 mgkg−1, respectively. Mann-Whitney test results reveal that thereis no significant difference between the two (p=0.17), sodistance to themercury source does not seem to be relevant inthis case.

Considering the relatively low levels found, there is noreason to believe that vegetables growing in this area pose arisk for consumers, even the ones from highly contaminatedsoils. Although mercury concentration in these samples islow, the presence of the metal can be due to atmosphericdeposition, to the presence of mercury in soil, since leafyvegetables can easily uptake elements from it, and irrigationwith contaminated waters.

Corn (Zea mays) is extensively produced in Estarreja areaand is mainly used as feedstuff and to produce flour. Despitethe proximity to the plant, all samples analysed (n=5) provedto be below the detection limit (0.05 ng Hg).

3.5. Fish

Fish samples were purchased at different locations in Estarreja.These locations were chosen after a preliminary survey onpopulation'sdietaryhabits.Accordingtothe informationcollected,themajority of people buy fish in local supermarkets or fisherman(fishwithorigin inAtlanticOcean) and fewat the localmarket (fishwith both origin in Atlantic Ocean and in Ria de Aveiro). Differentspecies of fish were purchased in all these locations to guaranteethat all fish that reaches the population was considered in thisstudy. Results obtained are shown in Table 3.

No significant differences (Mann-Whitney test, p=0.32)were found between fish captured in the Atlantic Ocean andsamples from the Ria de Aveiro. The slightly higher statisticalmean from the latter may be related with species-specifictraits, since most specimens collected in Ria de Aveiro arebottom-feeders (Solea spp, Pleuronectes platessa), which usuallyhave higher contaminant loads.

The mercury fish concentrations never exceed the thresh-old concentration the Portuguese law establishes for humanconsumption (0.5 mg kg−1, fresh weight), neither the FDAguideline of 1.0 mg kg−1. Previous studies in the same systemhave reportedmercury levels in fish to exceed 1mg kg−1 in themost contaminated area of Ria de Aveiro (Laranjo bay) (Abreuet al., 2000; Pereira et al., 1997). Nowadays, the fish stock atLaranjo bay is decaying and therefore hardly any fish is caughtthere, which means that it is not likely to be reaching andaffecting the local population. Data from other zones of Ria deAveiro show lower mercury concentration suggesting thatcontamination is mainly restricted to that area (Abreu et al.,2000). Nevertheless, permanent mercury monitoring of fishspecies is considered a fundamental tool to guaranteenutritional quality and safety (Olivero et al., 1997).

Table 4 – Total and reactive mercury (mean±std, ng L−1)found in water samples

Sample Total mercury Reactive mercury

Wells 1 526±1.4 1.5±0.312 847±76 14±0.373 3.9±0.10 2.7±0.404 24±0.83 b1.45 32±1.1 b1.46 26±1.4 2.3±0.27

Water stream 7 12±1.4 b1.4Limit of detection 2.3 1.4

2696 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 9 ) 2 6 8 9 – 2 7 0 0

Author's personal copy

3.6. Human hair

The main goal of this study was to infer if Estarreja'sinhabitants are currently being affected by mercury releasedto the environment by the chlor-alkali plant. Up to this point,data from this study revealed that mercury is not presenteither in the aquatic or terrestrial food web in high levels, butto realize if despite that the population shows concerningmercury levels that may be affecting their health, localresidents' hair samples were analysed.

A mercury concentration range of 0.090–4.2 mg kg−1 (mean1.5 mg kg−1, median 1.3 mg kg−1) was found in the 45 samplesconsidered. For males (nf19), a mean value of 1.4 mg kg−1 andmedian value of 1.3 mg kg−1 were found, whereas for females(nf26), mean and median values found were 1.6 mg kg−1 and1.4 mg kg−1, respectively (Table 3). Regardless of marked sexdifferences found in other studies (Williams et al., 2000), nosignificant difference of exposure is evident between the maleand female groups among Estarreja's population (Mann-Whitney test, pf0.85).

Biomonitoring studies generally find higher levels of mer-cury associated with advancing age as mercury levels accumu-late over time (Hac and Krechniak, 1996; Hornig, 1999). In thisstudy, volunteers' age ranged between 15 and 74 years, meanvalue being 41 years. Correlation between age andmercury levelinhair, althoughpositive, is veryweak (rf0.36;pf0.015), leadingto the conclusion that there is not a significant influence of agein mercury levels found in Estarreja's population.

As previously mentioned, consumption of fish is normallyconsidered the main contamination pathway of mercury tohuman beings (Agusa et al., 2005; Malm et al., 1995). Eventhough fish samples analysed in this study didn't showupsetting levels of mercury, one could still expect that higherlevels in hair were associated with people who consumemorefish. Nevertheless, correlation factor between the two vari-ables (rf0.38; pf0.011), although positive, is not strongenough to establish a clear influence of fish consumption inhair mercury levels of Estarreja inhabitants. These results can,however, be justified by the low number of fish servings perweek, an average of five.

Currently, a consensus on exposure level guidelines doesnot seem to exist. In 2001, the U.S. Environmental ProtectionAgency (USEPA) adopted a revised reference dose (RfD) formethylmercury of 0.1 μg/kg body weight per day. The RfD isbased on neurologic developmental effects measured inchildren associated with exposure in utero to methylmercuryfrom the maternal diet and is related to a hair mercuryconcentration of 1.0 mg kg−1 (USEPA, 2005). In this study 36%(n=16) of all samples were below this reference concentration.The percentages of the samples whose mercury concentra-tions ranged 1–2 mg kg−1 (n=19) and exceeded 2 mg kg−1

(n=10) were 42 and 22%, respectively. On the other hand, theJoint FAO/World Health Organization (WHO) Expert Commit-tee on Food Additives (JECFA) (FAO/WHO, 2006) based itsevaluation on the studies in the Seychelles and Faroe Islandsto establish a reference concentration. The Committee used14 μg g−1 in maternal hair-mercury as the average from thetwo studies. This value is an estimate of the level in maternalhair reflecting exposures that would be without appreciableadverse effects in the offspring in these two study popula-

tions. In accordance with these guidelines, none of thesubjects exceeded this threshold.

4. Discussion

Although several studies have been developed in the last twodecades to evaluate mercury contamination of the aquaticenvironment in Ria de Aveiro system, this is the firstintegrated assessment of mercury contamination consideringdifferent environmental compartments and providing insightinto potential risks to local population performed in theterrestrial environment surrounding the Estarreja chlor-alkaliplant.

So far, Portugal still does not have defined a contaminatedland management strategy or a nation specific approach toassess risks arising from contaminated soils. Nevertheless,issues related to the assessment of significant risk to humanhealth and ecosystems from contaminated land rousedrelevant discussions in varied countries (such as the UK, theNetherlands, Spain or Canada and theUSA) particularly duringthe last decade. Petts et al. (1997) describe the Source-Pathway-Receptor framework as a key concept which is essential toeffective understanding and management of the risks ofcontaminated land and suggest its use to guide action at thesite level. The pathway is the route (direct and indirect) in theenvironment by which the contaminant(s) (mercury, in thepresent case) may be transferred to the receptor (or potentialtargets) of concern (in this study humans were considered thekey “receptor”). According to Petts et al. (1997), even ifcontaminants are present on a site above a backgroundconcentration or a guideline value of acceptability, there willnot be a risk if there is neither a pathway by which thecontaminant could reach a sensitive receptor, nor any targetson or in the vicinity of the site. Nathanail et al. (2005) describe adeterministic method for deriving site-specific human healthassessment criteria for contaminants in soil that also adoptsthis risk-based Source-Pathway-Receptor pollutant linkageframework.

The Source-Pathway-Receptor conceptual framework thatis described in Fig. 4 was developed as a preliminaryassessment of potential risks to Estarreja human populationarising from mercury contamination. Human exposure path-ways considered by the present study are direct ingestion ofsoil and dust, inhalation of soil vapors, drinking of contami-natedwater, consumption of both home grown and purchasedvegetables, and ingestion of fish (available to local populationfrom local markets). Although not considered in this study,exposure associated to the ingestion of meat is also a humanexposure pathway and could arise from the intake of meatfrom locally grown animals contaminated bymercury presentin feedstuff. Human hair was selected as an indicator ofmercury levels actually reaching the receptor which is thehuman population from Estarreja.

From Fig. 4 it is clear that a significant contamination sourcehas emitted several tons of mercury during decades, and thatalthough the mercury-cells technology has been replaced in2002, mercury remains accumulated in both aquatic andterrestrial systems (historical contamination). The emission of

2697S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 9 ) 2 6 8 9 – 2 7 0 0

Author's personal copy

liquid effluents, particularly prior to 1975, is likely to have beenone of themost relevant pathways ofmercury contamination ofthe Ria de Aveiro system. The mercury released through thesestreams is nowadays accumulated in sediments from thislagoon, predominantly in the Laranjo bay area. The presentstudy revealed that despite contamination, the consumption offish from the Laranjo bay area is no longer a significant pathwayfor humans as the fish being currently consumed by theEstarreja population is no longer originated from this area.The fish fromRiadeAveiro that iseffectivelybeing consumedbyhumans has mercury levels comparable to fish from distantareas at the North East Atlantic. However, this situation shouldbemonitored to ensure that in the future and even if the stocksincrease, fish from Laranjo bay will not be made available tolocal population. In the same way, the ingestion of vegetablesdoes not appear to be a significant exposure pathway for localpopulation asmercury levels of locally grownvegetables are lowandcomparable to those fromother areas. This is likely tooccur,because although mercury soil contamination that has beenreported in the agricultural fields under study is considerable,vegetables don't seem to uptake the metal.

This study shows that feedstuff is clearly accumulatingmercury from contaminated soils in the areas surrounding pastmanmade effluent streams, and in this case this couldeffectively be a pathway for mercury accumulation throughthe food web and human exposure. Given the low valuesof mercury observed in human hair from local inhabitants,it appears that the population from Estarreja is currently not

being affected by mercury levels that still remain in theenvironment. Despite living close to a formermercury emissionspot, and possibly subject to occupational mercury exposure inthe past, this study reveals no evidence of this influence in localpopulation. Nevertheless, it is strongly advised that agriculturalpractices in the highly contaminated area of the “Vala de S.Filipe” are banned in the close future and that the access to thisstream is prohibited. A remediation solution for this contami-nated soil should also be considered. Moreover, the possibleimpacts of this contaminated stream on the urban area ofEstarreja as well as the possible impacts of further past effluentstreams on soil quality should also be investigated.

To our knowledge, there is no study developed for theassessment of the effects of atmospheric mercury emissionson the human health of local populations during the time thatmercury-cells were being used in the production process.These atmospheric emissions should not be significant in thepresent days and in what concerns their effects on theterrestrial environment it appears that in terms of both soilsand plants quality the impact of liquid effluents contamina-tion in the past was more significant than airborne mercurycontamination. Nevertheless, air quality considering mercuryand its relative impact in relation to other pathways should befurther investigated. Given the concentrations of mercuryobserved both in soil and plant samples, additional researchon site-specific determinations of contaminant bioaccessibil-ity or of plant-to-soil concentration factors should alsobe developed as this could provide valuable information for

Fig. 4 –Conceptual Source-Pathway-Receptor framework and simplified scheme of most relevant pathways for mercurycontamination, plant uptake and Human intake in the area of Estarreja.

2698 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 9 ) 2 6 8 9 – 2 7 0 0

Author's personal copy

future mercury plant uptake evaluations in other contami-nated areas. In addition, in the future, risk assessments tobe developed in this area could also take into account othertarget receptors rather than humans, as for example, sensitiveecosystems.

5. Summary and conclusions

It is estimated thatduring thedecades fromthe1950suntil theendof the90sabout 27 tonsofmercuryweredischarged into theRiadeAveiroandthatapproximately74%of this isconfined inthenearbylagoon, andmostparticularly at Laranjobay (Nuneset al., 2007).Asexpected in an area that has been subjected tomercury emissionsin the past, the soils of Estarreja are also generally slightlycontaminated. Moreover, a specific area of very high mercurycontent in agricultural soils has been clearly identified anddelineated. It was observed that high mercury contamination isrestricted to a confined area close to an old stream used totransport liquid effluents from the chlor-alkali plant. Therefore aneffort should bemade to identify and remediate all “hotspots” andagricultural practices in this locale should be restricted in order toavoidplantmercuryuptakeand theconsequent inputof themetalin the food web. With the exception of this high mercurycontaminated area, a sharp decrease of mercury levels in soilsand inLoliumperenneplantswasobservedwith increasingdistancefrom the past mercury source.

One of the main goals of this work was to infer clearrelationships between mercury concentrations in the foodweb and its human influence. Results suggest that there is nosignificant risk arising from food products, since highmercurylevels were only found in soils and, in a few cases, feedstuff.Fish and vegetables do not show mercury concentrations thatcould be considered hazardous for consumers. In summary,mercury does not seem to be reaching population, asconfirmed by hair analysis.

R E F E R E N C E S

Abreu SN, Pereira E, Vale C, Duarte AC. Accumulation of mercuryin sea bass from a contaminated lagoon (Ria de Aveiro,Portugal). Mar Pollut Bull 2000;40:293–7.

Agusa T, Kunito T, IwataH,Monirith I, Tana TS, Subramanian A, et al.Mercury contamination in human hair and fish from Cambodia:levels, specific accumulation and risk assessment. Environ Pollut2005;134:79–86.

Batista A, Silva EF, Azevedo M, Sousa A, Fonseca EC. Soil dataanalysis from central Portugal by Principal ComponentAnalysis and geostatistical techniques. Geochem ExplorEnviron Anal 2002;2:15–25.

Bernaus A, Gaonaa X, Reeb DV, Valiente M. Determination ofmercury in polluted soils surrounding a chlor-alkali plant.Direct speciation by X-ray absorption spectroscopy techniquesand preliminary geochemical characterisation of the area. AnalChim Acta 2006;565:73–80.

Biester H, Müller G, Schöler HF. Binding and mobility of mercuryin soils contaminated by emissions from chlor-alkali plants.Sci Total Environ 2002;284:191–203.

Bloom NS. Mercury as a case-study of ultraclean sample handlingand storage in aquatic trace metal research. Environ Lab1995:20–5 March/April.

Caille N, Vauleon C, Leyval C, Morel JL. Metal transfer to plantsgrown on a dredged sediment: use of radioactive isotope 203Hgand titanium. Sci Total Environ 2004;341:227–39.

Carey A, Gowen J, Forehan T, Tai H, Wiersma G. Heavy metalconcentrations in soils of 5 United State cities, 1972, urban soilsmonitoring program. Pestic Monit J 1980;13:150–4.

Clarkson T. Human toxicology of mercury. J Trace Elem Exp Med1998;11:303–17.

Clarkson T. Human exposure to mercury: the three moderndilemmas. J Trace Elem Exp Med 2003;16:321–43.

Coelho JP, Rosa M, Pereira E, Duarte A, Pardal MA. Pattern andannual rates of Scrobicularia plana mercury bioaccumulation ina human induced mercury gradient (Ria de Aveiro, Portugal).Estuar Coast Shelf Sci 2006;69:629–35.

Cooper CM, Gillespie W. Arsenic and mercury concentrations inmajor landscape components of an intensively cultivatedwatershed. Environ Pollut 2001;111:67–74.

Costa C, Jesus-Rydin C. Site investigation on heavy metalscontaminated ground in Estarreja-Portugal. Eng Geol2001;60:39–47.

Costley CT, Mossop KF, Dean JR, Garden LM, Marshall J, Carroll J.Determination of mercury in environmental and biologicalsamples using pyrolysis atomic absorption spectrometry withgold amalgamation. 2000;405:179–183.

Dommergue A, Ferrari C, Planchon F, Boutron C. Influence ofanthropogenic sources on total gaseous mercury variability inGrenoble suburban air (France). Sci Total Environ2002;297:203–13.

Ericksen JA, Gustin MS, Schorran DE, Johnson DW, Lindberg SE,Coleman JS. Accumulation of atmospheric mercury in forestfoliage. Atmos Environ 2003;37:1613–22.

European Commission. Amended proposal for a directive of theEuropeanParliament andof theCouncil on undesirable substancesand products in animal nutrition. Official Journal C 096 E. 2002;346.

FAO/WHO, Summary and conclusions of the sixty-seventhmeetingof the Joint FAO/WHO Expert Committee on Food Additives(JECFA), 20–29 June 2006. JECFA/61/SC 2006. http://www.chem.unep.ch/mercury/Report/JECFA-PTWI.htm. April, 2008.

Ferreira,M.M.S.I.Mercúrio emsolos daáreadocomplexoquímicodeEstarreja. Master Thesis. Geochemistry Department. Universityof Aveiro, Aveiro. 1993.

Hac E, Krechniak J. Mercury content in human kidney and hair.Toxicol Lett 1996;88:56–7.

Hall A, AC AD, Caldeira M, Lucas M. Source and sinks ofmercury inthe costal lagoon of Aveiro, Portugal. Sci Total Environ1987;64:75–87.

Hornig JF. Social and environmental impacts of the James Bayhydroelectric project. McGill-Queen's University Press; 1999.p. 169.

Inácio M, Pereira V, Pinto M. Mercury contamination in sandy soilssurrounding an industrial emission source (Estarreja, Portugal).Geoderma 1998;85:325–39.

Lacerda L, Salomons W. Mercury from gold and silver mining: achemical time bomb? Berlin: Springer-Verlag; 1998. p. 146.

Lindberg S. Introduction: the fourth international conference onmercury as a global pollutant. Atmos Environ 1998;32:807–8.

Lindberg SE, Meyers TP, Taylor GE, Turner RR, Schroeder WH.Atmosphere-surface exchange ofmercury in a forest: results ofmodeling and gradient approaches. J Geophys Res D Atmos1992;97:2519–28.

Malm O, Branches FJP, Akagi H, Castro MB, Pfeiffer WC, Harada M,et al. Mercury andmethylmercury in fish and human hair fromthe Tapajós river basin, Brazil. Sci Total Environ1995;175:141–50.

Manta D, Angelone M, Bellanca A, Neri R, Sprovieri M. Heavymetals in urban soils: a case study from the city of Palermo(Sicily), Italy. Sci Total Environ 2002;300:229–43.

Maserti BE, Ferrara R. Mercury in plants, soil and atmosphere neara chlor-alkali complex. Water Air Soil Pollut 1991;56:15–20.

2699S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 9 ) 2 6 8 9 – 2 7 0 0

Author's personal copy

Mucci A, Lucotte M, Montgomery S, Plourde Y, Pichet P, Tra HV.Mercury remobilization from flooded soils in a hydroelectricreservoir of northern Quebec, La Grande-2: results of a soilresuspension experiment. Can J Fish Aquat Sci1995;52:2507–17.

Nathanail P, McCaffrey C, Earl N, Foster ND, Gillett AG, Ogden R. Adeterministic method for deriving site-specific human healthassessment criteria for contaminants in soil. Hum Ecol RiskAssess 2005;11:389–410.

Nunes T, Casco P, Pereira ME, Duarte AC, Figueira E. Mercury fluxevaluation to the atmosphere in a contaminated area (Esteirode Estarreja). Geophys Res Abstr 2007;9:10978.

Olivero J, Navas V, Perez A, Solano B, Salas R, Vivas R, et al. Ninthinternational symposium on trace elements in man andanimals, vol. 9. Canada: National Research Council; 1997. p. 171.

Ospar Commission. Mercury losses from the chlor-alkali industryin 2004. Hazardous Substances Series; 2006. p. 37.

Pereira ME, Duarte AC, Millward GE, Abreu SN, ReisMC. Distributionof mercury and other heavy metals in the Ria de Aveiro. QuímAnal I 1997;16:S31–5.

Pereira E, Abreu SN, Coelho JP, Lopes CB, Pardal MA, Vale C, et al.Seasonal fluctuations of tissue mercury contents in theEuropean shore crab Carcinus maenas from low and highcontamination areas (Ria de Aveiro, Portugal). Mar Pollut Bull2006;52:1450–7.

Petts J, Cairney T, Smith M. Risk-based contaminated landinvestigation and assessment. Chichester: Wiley; 1997. p. 334.

Pio CA, Costa MJ, Borrego CA. Atmospheric mercury in anindustrialized region. Proceedings of the internationalconference on chemicals in the environment. Lisbon, Portugal;1986. p. 123–8.

Ramalhosa E, Pato P, Monterroso P, Pereira E, Vale C, Duarte AC.Accumulation versus remobilization of mercury in sedimentsof a contaminated lagoon. Mar Pollut Bull 2006b;52:353–6.

Reimann, Caritat. Chemical elements in the environment: factsheets for the geochemist and the environmental scientist.Berlin: Springer; 1998. p. 396.

Rodrigues S, Pereira ME, Duarte AC, Ajmone-Marsan F, DavidsonCM, Grčman H, et al. Mercury in urban soils: a comparison oflocal spatial variability in six European cities. Sci Total Environ2006a;368:926–36.

Rodrigues S, Pereira ME, Sarabando L, Lopes L, Cachada A, DuarteA. Spatial distribution of total Hg in urban soils from anAtlantic coastal city (Aveiro, Portugal). Sci Total Environ2006b;368:40–6.

Southworth GR, Lindberg SE, Zhang H, Anscombe FR. Fugitivemercury emissions from a chlor-alkali factory: sources andfluxes to the atmosphere. Atmos Environ 2004;38:597–611.

Speedy AW. FAO and pre-harvest food safety in the livestock andanimal feed industry. WHO Consultation on Pre-Harvest FoodSafety; 2001.

USEPA. Land application of municipal sludge process designmanual. MER laboratory. Cincinnati, OH, US GovernmentPrinting Office: Washington DC; 1983.

Ullrich S, Ilyushchenko M, Kamberov I, Tanton T. Mercurycontamination in the vicinity of a derelict chlor-alkali plant.Part I: sediment and water contamination of Lake Balkyldakand the River Irtysh. Sci Total Environ 2007;381:1–16.

USEPA. The problem with mercury. Clean Air Network Fact Sheet;1999.

USEPA.Water quality criterion for the protection of human health.Methylmercury; 2005. http://www.epa.gov/waterscience/criteria/methylmercury/document.html. April, 2008.

Válega M, Abreu S, Pato P, Rocha L, Gomes A, Pereira M, et al.Determination of organic mercury in biota, plants andcontaminated sediments using a thermal atomic absorptionspectrometry method. Water Air Soil Pollut 2006;174:223–34.

Williams TM, Apostol AN, Miranda CR. Assessment by hairanalysis of mercury exposure amongmining impactedcommunities of Mindanao and Palawan, The Philippines.Environ Geochem Health 2000;22:19–31.

2700 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 9 ) 2 6 8 9 – 2 7 0 0