Mercury contamination in Lavras do Sul, south Brazil: a ... - UQAM

3 downloads 3814 Views 1MB Size Report
left by CRM (Companhia Riograndense de Miner- acao) mining company ...... Maron M.A.C., Silva A.R.B. Perfil Analıtico do ouro, boletim. ´. No. 57, Ministerio ...
The Science of the Total Environment 307 (2003) 125–140

Mercury contamination in Lavras do Sul, south Brazil: a legacy from past and recent gold mining M.H.D. Pestanaa,*, M.L.L. Formosob ¸ ˜ Estadual de Protecao ¸ ˜ Ambiental Henrique Luıs ´ Roessler FEPAM, Rua Carlos Chagas 55, Fundacao Porto Alegre Rio Grand do Sul, Porto Alegre 90030 020, Brazil b ´ ˆ Centro de Estudos em Petrologia e Geoquımica (CPGq), Instituto de Geociencias, ¸ Universidade Federal do Rio Grande do Sul (UFRGS), Av. Bento Goncalves, 9500 CEP Porto Alegre 91509 000, Brazil a

Received 31 October 2000; accepted 18 October 2002

Abstract An attempt is made, in this work, to establish approximate gold production and, consequently, mercury emission rates in Lavras do Sul during the 20th century, after a description of the historical background of the study area. The identification of two heavily polluted sites (‘hot spots’) shows the persistence of Hg contamination originated in the early 1900s until the 1950s, as well as more recent soil pollution, from the 1980s. The evaluation of natural and anthropogenic residual contamination is approached by the study of Hg concentrations in mineralized rock samples, in soil samples neighboring mining wastes and milling facilities and in stream sediments. Anthropogenic contamination in soil samples reached 110 000 ngyg Hg in bulk samples and 506 000 ngyg Hg in the silt–clay fraction, of which 82–83% as Hg0, and 16–18% associated to the sulfideyresidual fraction, according to complementary speciation analyses. The association of Hg with base metal sulfides may be contributing to local background concentrations varying from 140 to 207 ngyg in stream sediments. 䊚 2003 Elsevier Science B.V. All rights reserved. Keywords: Mercury contamination; Gold mining; Sediments; Soils; Rocks

1. Introduction Mercury can be readily re-emitted to the atmosphere from any site in which it has been deposited, due to its volatility (Kim et al., 1993; Nriagu, 1994; Schroeder, 1995). The study of Hg contaminated sites by historic gold mining wastes can be useful in the prediction of mercury mobilization, re-emission and persistence in the environment, and also forms the basis for site remediation. *Corresponding author. Tel.: q55-51-225-1588; fax: q5551-325-4215. E-mail address: [email protected] (M.H.D. Pestana).

The knowledge of the different stages in which mercury is applied, during amalgamation processes, to extract gold and silver has been used to evaluate rates of environmental mercury losses during present and past mining operations. Emission factors calculated according to the efficiency of the applied method made it possible to estimate mercury loss rates from gold production rates, especially when the exact amount of mercury used is unknown. The importance of mercury emitted during colonial times from gold and silver mining in the

0048-9697/03/$ - see front matter 䊚 2003 Elsevier Science B.V. All rights reserved. PII: S 0 0 4 8 - 9 6 9 7 Ž 0 2 . 0 0 5 3 5 - 1

126

M.H.D. Pestana, M.L.L. Formoso / The Science of the Total Environment 307 (2003) 125–140

Americas has been only recently evaluated, and its re-cycling may be partly responsible for present high background concentrations in the global environment (Lacerda and Salomons, 1998; Nriagu, 1994). The average of 612 tyyear of mercury were lost in the silver mines of Spanish America between 1580 and 1900, and an average of 1360 tyyear of mercury was produced and imported into the US between 1850 and 1900 (Nriagu, 1994). In Brazil, the Hg loss from gold mining was ;2.5 tyyear from 1800 to 1960 (Lacerda, 1997). In the 18th century, Brazil produced 862 t of gold (40% of a world production of 2154 t) (Maron and Silva, 1984), when only gravimetric procedures were used to extract gold from high concentration ores, until 1850 (Lacerda, 1997). The exhaustion of the rich, easily extractable gold deposits gave place to low graded ores, propitiating the use of Hg amalgamation techniques. By the end of the 1700s, gold production from Minas Gerais, Southeast Brazil, had decreased from a yearly average of 11025 kg yeary1, in the period 1736–1751, to 3675 kg yeary1, in the period 1788–1801 (Prieto, 1976). Contrasting with colonial times, a second gold rush started in Brazil after the 1968 increase in the international gold prices, and was characterized by the extensive use of mercury amalgamation, mostly by illegal miners (garimpeiros). In 1983, illegal mining (garimpos) was responsible for 85% (47.5 t) of the total Brazilian gold production (53.7 t). However, from 1972 to 1983, approximately 149 t of gold produced by the garimpos were not included in the official Brazilian production, due to the lack of proper inspection (Maron and Silva, 1984). The second gold rush reached its peak during 1990–1993, and started decreasing since 1994. Concerning Hg amalgamation, this gold rush can be regarded, in a minor proportion, as a similar event to what happened in the Spanish American colonies. According to Nriagu (1994), the emission factor (EF) in colonial times (Spanish America) varied from 0.85 to 4.1 kg Hg kgy1 Au for, respectively, poor and rich ores, and is similar to the range 1.3–1.7 kg Hg kgy1 Au estimated by Pfeiffer and Lacerda (1988) for the recent Brazilian gold rush in the Amazon region. Present-day

annual input of mercury to the environment from gold mining sites in the Brazilian Amazon is estimated to be 180 tyyear (Lacerda and Salomons, 1998). Lavras do Sul is one of the areas in Brazil where both gold rushes are registered through evidences of environmental mercury contamination in, at least, two different time periods: since the early 1900s until ;1950; and, more recently, during the second half of the 1980s and the 1990s. This study area is also characterized by some relatively high Hg concentrations due to natural sources, like the local bedrock and the occurrences of other metal sulfides containing Hg (Pestana et al., 2000). This work has two main purposes. The first is to establish soil emission rates, calculated from estimates of gold production in Lavras do Sul during the 20th century. To do this, a survey of the historical background of gold mining in the area was proved necessary, due to the scarcity of official records. The second purpose is to identify sites of Hg anthropogenic contamination (past and recent) and distinguish them from those contaminated by natural sources, using spatial distribution of concentration values in different geologic materials (rocks, soils, sediments), ranked in classes defined by factors of the natural background, and plotted on the lithologic map of the area. The knowledge of the location of old mills and mining facilities, and speciation results from contaminated soil and sediment samples provided complementary information on the identification of ‘hot spots’, the persistence of Hg contamination and its potential mobility in the area. The expression ‘hot spot’ is used here in the environmental sense, according to Stigliani’s general definition ‘regions of high inputs of toxic materials’ (Stigliani et al., 1991). In a more specific sense, the term is used following the sugges´ (1992) who refers to ‘hot tion proposed by Skei spot’ sediments as being those where ‘concentrations (of a toxic substance) are a minimum of 50– ´ 100 times background concentrations’ (Skei, 1992). The importance of the identification and mapping of ‘hot spots’ is the fact that their harmful effects on the environment may happen very long

M.H.D. Pestana, M.L.L. Formoso / The Science of the Total Environment 307 (2003) 125–140

after the chemical loading of soils and sediments occurred. 2. Materials and methods 2.1. Sampling Compound samples of stream sediments were taken, at each 6-month period, in eight sites from the upper reaches of the Camaqua˜ River, in the Au mining area of Lavras do Sul (Fig. 1), during the period 1992–1996 (Projects FEPAMFAPERGS and FEPAM-FNMAyMMA). Seven sampling sites were distributed among three trib´ utaries (Lavras Stream, Jaques Stream and Hilario Stream), located upstream and downstream of recent mining sites, or at alluvial sections, where individual miners operated in the 1990s. Another site (C1) is located on the Camaqua˜ River, downstream from the tributaries confluence, to evaluate pollutants transfer to the eastern part of the basin. The 6-month period frequency during the 4 years were meant for evaluation of temporal trends and monitoring of the water and sediments of the upper Camaqua˜ River basin, which are the objective of other works. For the purposes of this work, mean values of Hg concentrations at each sampling site will be used for spatial considerations. In order to evaluate the influence of historic mining sites or milling facilities in Hg contamination in Lavras do Sul, the above program was complemented by the selected sampling of rocks, soil and stream sediments, thus permitting the identification of ‘hot spots’ and natural Hg sources. In October 1997, two surface soil samples were collected at the following sites: CRM7, located nearby an abandoned gold amalgamation structure left by CRM (Companhia Riograndense de Miner¸ ˜ mining company; and Afl.10, close to the acao) ruins of Chiapettas’s mill, which was active from 1909 until the late 1930s. In the same period, 10 additional stream sediment samples were collected, most of them at sites located near historic mining sites and milling facilities. Four selected rock samples containing different kinds of sulfide minerals were analyzed for various metals including Hg, since previous speciation results (Pestana et al., 2000) suggested sulfide

127

association with Hg in the area. The samples were collected from different mineralized lithologies in the Lavras do Sul area: BB3—a hydrothermally weathered granite near Bloco do Butia´ mine; 14VT—a vein quartz with pyrite, near Valdo Teixeira mine; RMSJ—weathered granitic wastes ˜ Jose´ mine; 13CRM—an andesite from the Sao with Cu sulfides from the CRM mining area. Two other samples were analyzed only for Hg: 15CR1, a weathered andesitic tuff; and 15CR2, a fresh andesite with quartz vein and hematite, both from outcrops located near the Cerro Rico mine. The location of these rock, soil and sediment samples and their ranges of Hg concentrations can be seen on the map in Fig. 2, Section 4.3. 2.2. Soil, rock and sediment analyses Hg in soil and sediment samples were analyzed solely in the fine (silt–clay) fraction, except the soil samples Afl.10 and CRM7, in which total Hg concentrations in bulk samples were also determined. After wet sieving with deionized water and a polyethylene and nylon sieve, the -63-mm fraction was air-dried at room temperature, following procedure recommended by Ure (1994), except for samples collected before August 1994, which were oven-dried at 40 8C (Pestana et al., 1997). Rocks were ground to a fine powder and later digested with the same total extraction procedures for total metals (Fe, Mn, Cu, Cd, Pb, Zn), used for sediments: a hot mixture of HFyHNO3 y HClO4 (1:1:1), repeated several times until the complete digestion of the samples, as described in Pestana et al. (1997, 2000). Metals were determined by conventional AAS at the UFRGS-CPGq Laboratory. Sediment samples collected in 1996 were analyzed by ICP at the FEPAM Laboratory Division. Mercury determinations in sediments and rocks were conducted at the UFRGS Soil Laboratory, by AAS (cold vapor technique), after digestion with hot HNO3 and HClO4 acids in a closed system (samples collected after January 1993), as described by Pestana et al. (1997, 2000). Speciation (water soluble, exchangeable, elemental and sulfide) analyses in the bulk soil samples were done at the Nevada Bureau of Mines and Geology

128

M.H.D. Pestana, M.L.L. Formoso / The Science of the Total Environment 307 (2003) 125–140

Fig. 1. Map location of the study area in Rio Grande do Sul state (RS), south Brazil (Fig. 1a); and location of the eight stream sediments sampling sites (period 1992–1996), in relation to the main recent Au mining areas in the west region of the Camaqua˜ River basin (Fig. 1b).

M.H.D. Pestana, M.L.L. Formoso / The Science of the Total Environment 307 (2003) 125–140

(NBMG), according to Lechler et al. (1997). These analyses complement previous speciation data obtained for other samples from the Lavras do Sul area (Pestana et al., 2000). Reference materials were used to ensure analytical quality: USGS Sco-1 for the rock samples and RM-8407 for the Hg contaminated soil samples. Reproducibility errors in rocks were F30% for all the analyzed metals, and were 33% for Hg in soil samples. Precision errors were evaluated by comparing analytical results in duplicate (rocks and soils) and triplicate (sediments) subsamples. In sediments, 54–67% of the Hg analyses showed precision errors F10%, and 64–92% in a class F15%. In rocks and soil samples, approximately 67% of the Hg analyses presented precision errors F10%. Iron, Mn, Pb, Cd and Zn in rock samples had average precision errors F10%, and Cu had F13%. 3. Study area 3.1. Land-uses and geologic features The Camaqua˜ River basin drains the shield region of Rio Grande do Sul state, south Brazil, and is part of the Southeastern Mountain Range. Lavras do Sul is located in the western part of the basin. Fig. 1 shows the location of the study area in the State of Rio Grande do Sul, and the sites where stream sediments were sampled in the period 1992–1996. Gold and copper mining activities in the area are presently interrupted, except for scarce individual gold mining during summer. Land-uses are restricted to cattle and sheep raising, with incipient agriculture, since the predominant soil in the area is litholic over andesitic substrate, shallow and susceptible to erosion (Brasil, 1973). Most of the sampling sites in Fig. 1 are located in the area of occurrence of this litholic soil, except for sites H2 and C1. Upstream from site H2, a red–yellow– albic (bleached elluvial horizon) podzol occurs locally, developed over the Santa Barbara Formation sandstone (Brasil, 1973). The most representative lithology in area, occupying all the central part of the study area, is a volcanic sequence of shoshonitic composition

129

formed by trachyandesitic lavas (in this work, referred simply as ‘andesites’), tuffs and volcanic ´ Formation), which, togethconglomerates (Hilario er with the nucleus of the Lavras do Sul Granitic Complex (LSGC) (Nardi and Lima, 1988) constitute the Lavras do Sul Shoshonitic Association (LSSA) (Lima, 1995). Overlying these Cambrian–Ordovician rocks, are the conglomerates and sandstones of the Santa Barbara Formation (Eopaleozoic) and fluvial Quaternary deposits. ´ The intrusion of the LSGC on the Hilario andesites produced fractures and a complex hydrothermal system on the two lithologies (Mexias et al., 1990). In both, the sulfidic mineralization concentrated Au and Fe (in pyrites and chalcopyrites), Ag (in galenas, especially at the Merita area), Cu (bornites, chalcopyrites), Pb and Zn (galena and sphalerites, respectively). Molibdenite was observed in fractures of the nucleus of the LSGC. The Cu–Fe sulfides containing Au predominate in the granites and in the nearby andesites, while the Ag-bearing Pb and Zn sulfides predominate far from the contact with the granitic intrusion, on the upper portions of the volcanic sequence (Nardi and Lima, 1988). The lithologic map of the area can be seen in Fig. 2, Section 4.3, and includes other lithologies not described here. 3.2. Gold mining in Lavras do Sul—historical background In 1799, the gold deposits of Lavras do Sul were discovered in south Brazil, in a historical context of smuggling of precious metals, of gold decline in the southeast Minas Gerais, of several frontier fights between Portugal and Spain (Pesavento, 1997). Therefore, it is not surprising that records of gold production from this period are practically inexistent. After the second half of the 19th century and early 1900s, local exploitation has been marked by successive cycles in which individual miners and mining companies of different nationalities (English, Belgium or Brazilian) operated in Lavras do Sul. From the early 1900s until the 1950s, approximately 37 mining sites were exploited in an area of approximately 60 km2 (Teixeira, 1992).

130

M.H.D. Pestana, M.L.L. Formoso / The Science of the Total Environment 307 (2003) 125–140

Most of the ore extracted during the first half of the 1900s was mainly processed in three mills ˜ and Cerro Rico), which used (Chiapetta’s, Paredao mercury amalgamation and were active for different time periods. The first reference to the use of mercury amalgamation reports to 1909, at Chiapetta’s mill (Teixeira, 1992). Other mills operated in the area in the 1930s or 1940s, but left no present visual vestige. Hg amalgamation was also used nearby the mines, although at least one miner used cyanide process to extract gold at the Bloco do Butia´ mine in the 1930s (Gavronski, 2000 personal communication). More recently, in the 1980s, another gold rush started in Lavras do Sul, probably stimulated by the socio-economic conditions that produced the Amazon gold rush in the north of Brazil. It lasted from 1992 to 1993. During this period, CRM ¸˜ (Companhia Riograndense de Mineracao), a state mining company, exploited alluvial gold deposits of the Lavras Stream at the site called ‘Volta Grande’, from 1983 to 1989. The company also used mercury amalgamation, having installed a field laboratory and ore extracting facilities at the mining site. With the decrease of the ore content in the alluvial deposits, CRM ceased its activities in the area in 1989. From the end of the 1980s until approximately 1993 individual ‘garimpo’ miners operated at different sites nearby the ´ Jaques, Hilario and Lavras Streams, or at their tributaries. Presently, the occurrence of ‘garimpeiros’ in the area is rare. 4. Results 4.1. Estimates of gold production and mercury fluxes in Lavras do Sul A brief summary of the gold mining history of Lavras do Sul and the available data on gold production is shown in Table 1. Most of the information was obtained from Teixeira (1992). The present calculation for the first half of the century is almost three times higher than the value initially estimated by Pestana et al. (2000), which was based solely on the information on gold production provided by Nardi and Lima (1988), and can still be further modified if more exact

information is obtained on past gold production in the study area. The annual input of mercury, based on the presently available data (Table 1) is approximately 1 t in 82 years. This corresponds to an average annual input to the environment of 0.01 tyyear, considering that the first known use of mercury amalgamation started at Chiapetta’s mill in 1909, and the latest gold production reported for individual miners (garimpeiros) refers to 1991. 4.2. Metal concentrations in rocks and soils Metal concentrations in six selected rock samples from five different mining sites are in Table 2. Considering the samples in which other metals’ concentrations were also analyzed, the higher values in Hg concentrations can be observed in the two samples that also presented Cu and Fe enrichment (13CRM and 14VT), both containing visible sulfide minerals. Although the number of rock samples analyzed is too small for any generalization, these results suggest a link between Hg and CuyFe sulfides in the study area. A natural Hg association with sulfides was evidenced by Pestana et al. (2000) in previous speciation analyses of sediments from the Lavras do Sul area. Those results showed that the relative importance of the sulfide species reached 30% of the total Hg distribution in overbank sediment samples from site L3, located near the CRM mining area. However, the highest percentage of Hg sulfide (55%) was found in a stream sediment sample collected upstream from the Cerro Rico mill, in a creek draining andesites and tuffs of the ´ Hilario Formation (sample C1H-c). Total Hg concentrations in a weathered andesitic tuff (Table 2, sample 15CR1) from the same watershed was also relatively high (473 ngyg), while the fresh andesite sample containing oxide instead of sulfide (15CR2) showed only 53 ngyg. Mercury concentrations in soil samples collected at a distance -5 m from the ruins of the Chiapeta’s mill (Afl.10) and from the milling facilities of CRM mining company (CRM7) are shown in Table 3. Besides the enrichment of approximately five - or six - fold in the fine fractions of both samples as compared to bulk samples, speciation

Period

Mine

Gold production (kg)

Mill or mining facilites

Mining company

Reference

1860

Various

xxx

1898

xxx

)20

Individual miners, etc. Lopes & Tallouard

1900–1915

´ Aurora, Virgınia, Cerro Rico, Dourada, alluvial deposits Bloco do Butia´ ´ ˜ Jose´ Virgınia, Sao

xxx

D. Laut; J.F. Souza water mills (monjolos) Lopes & Tallouard water mills (monjolos) No mills

Teixeira (1992), p. 80–88 Teixeira (1992), p. 90–91 CRM (1979) Kaul and Rheinheimer (1974)

Compagnies de Mines d’Or du Cerrito (Belgium) ChiapettaqBelgium Co.

Brazilian Goldfield Ltd. (BrazilianyEnglish)

1909–1910

Caneleira

200

˜ Joao, ˜ same site of Sao Lopes & Tallouard, later sold to Mr Chiapetta Chiapetta’s

1909–1912

V. Alegre, C. Rico, Aurora and others Bloco do Butia´

xxx

Chiapetta’s

Volta Grande, Saraiva and others Cerro Rico Cerro Rico, Bloco do Butia´ Alluvial deposits Alluvial deposits, Volta Grande Volta Grande Alluvial deposits

200

˜ mill Paredao

Brazilian Goldfield Limited Aurora Gold Syndicate ‘Pedro Mata e Cia.’; ¸ ˜ de ouro ‘Mineracao Bloco do Butia´ Ltda’ Various

xxx 4.48 6.5ymonth=12s78 0.8ymonth=12s9.6

Cerro Rico mill ˜ de Souza, Serapiao Pedro Mata None None

Cia. Jose´ H. de Souza ˜ de Souza, Serapiao Pedro Mata Individual miners Individual miners

Teixeira (1992), p. 103 Teixeira (1992), p. 115

93.756 0.2ymonth=12s2.4

CRM none

CRM Individual miners

CRM (1987) Saraiva (1992), p. 19

1901–1908

1935–1939

1935–1947 1938–1942 1949 1939 1952 1983–1988 1991

xxx

80

Subtotal gold production (1898–1852): ;592.08 kg

Chiapetta’s, Pedro Mata’s

Kaul and Rheinheimer (1974) Teixeira (1992), p. 95 Gavronski (1976), p. 45 Teixeira (1992), p. 97 Gavronski (1976), p. 46–47 Teixeira (1992), p. 98–100 CRM (1998) Gavronski, 2000 (pers. com.) Kaul and Rheinheimer (1974) Teixeira (1992), p. 102 Kaul and Rheinheimer (1974) Gavronski (1976), p. 49

Total gold production (1898–1991): 592.08q93.76q2.4s ;688.24 kg

M.H.D. Pestana, M.L.L. Formoso / The Science of the Total Environment 307 (2003) 125–140

Table 1 Approximate gold production in Lavras do Sul in the 19th and 20th centuries (‘xxx’ is unavailable or inexistent data)

131

132

M.H.D. Pestana, M.L.L. Formoso / The Science of the Total Environment 307 (2003) 125–140

results are also presented. The relative importance of the sulfide and elemental species in the range of 16–18% and 82–83%, respectively, shown in both samples, are very close to the Hg distribution previously determined for sample C1H-p (Pestana et al., 2000). This sample, a subaqueous sediment from a pond containing wastes from the Cerro Rico mill, was considered by the authors as a ‘hot spot’ in the environmental sense. ´ Following the criterion proposed by Skei (1992), the Hg concentration on the fine fraction of sample CRM7 greatly surpasses the limit of 50–100 times the adopted background value of 180 ngyg mean shale (MS). Sample AFL10, whose concentration is below the 50= MS can, however, be considered a hot spot according with the general definition of Stigliani et al. (1991). The difference in time of pollution between these two samples, which possibly accounts for the difference in Hg concentrations is further discussed. In addition, the three ‘hot spot’ samples, although one is a subaqueous sediment and two are soil samples have in common: a sulfidic (probably natural) association in the range of 16– 18% of the total Hg distribution; a relatively high (82–83%) contribution of elemental Hg, part of which from a probable anthropogenic origin; and relatively high total Hg concentrations, as compared to other soil samples in the area: L12B and CRM3. Actually, sample L12B is located at what was ˜ mill, but its total left of the ruins of the Paredao Hg concentration is not high enough to consider it

Table 3 Total Hg concentrations in the silt–clay fraction of four soil samples from Lavras do Sul; and Hg speciation (Lechler et al., 1997) and total Hg concentration in bulk samples in two selected soil samples (‘;’snot analyzed) Sample Total Hg (ng.g)

CRM7 Afl 10 L12B CRM3

Hg speciation (%)

Silt–clay fraction

Bulk sample

Water Exch. soluble

Elem Sulfide

506 000 2100 294 195

110 000 0.33 0.67 83.1 318 -0.10 -0.10 82.0 ; ; ; ; ; ;

16* 18* ; ;

* Obs: sulfide fraction was extracted with the residual fraction.

a ‘hot spot’. This can be, because the site is located too near to the river bank and most of the contaminated soil is likely to have been already washed away during floods. Another possible explanation is the fact that the building of the old mill on this site was destructed by the town hall a few years before the soil sample was collected, which means that the surface soil has been probably disturbed, and contamination dispersed. The present results bring additional information on the identification of ‘hot spots’ in the study area, so the number of local ‘hot spots’ in Lavras do Sul now currently sums three: C1H-p (Pestana et al., 2000), Afl10 and CRM7. 4.3. Hg concentrations in stream sediments The mean concentrations, S.D. and ranges for total Hg in stream sediment samples of the Lavras

Table 2 Hg and other metal concentrations in six rock samples from Lavras do Sula Sample

Sample description

Fe (mgykg)

Mn (mgykg)

Pb (mgykg)

Cu (mgykg)

Cd (mgykg)

Zn (mgykg)

Hg (ngyg)

BB3 14VT RMSJ 13CRM 15CR1 15CR2

h.a.gr. qz.v.qpy w.m.w. and.qCu sul. w.a.t. f.a.qqz.v.qhem

22 750 167 750 30 500 69 500 ; ;

4965 15 14 1249 ; ;

343 108 971 481 ; ;

22 96 323 6 7667 ; ;

-2.5 -2.5 5 -2.5 ; ;

124 90 1383 214 ; ;

76 1800 294 322 473 53

a Abbreviations: h.a.g.shydrothermally altered granite; qz. v.qpysquartz vein with pyrite; w.m.w.sweathered mine waste; and qCu sul.sandesite with Cu sulfides; w.a.t.sweathered andesitic tuff; f.a.qqz. v.qhem.sfresh andesite with quartz vein and hematite; ‘;’snot analyzed.

M.H.D. Pestana, M.L.L. Formoso / The Science of the Total Environment 307 (2003) 125–140

133

Table 4 Total Hg concentrations in stream sediments of the Lavras do Sul area, Camaqua˜ River basin (period 1992–1996) S.D.sstandard deviation; nsnumber of samples Site

Basic statistics

L1

Mean"S.D. Range Mean"S.D. Range Mean"S.D. Range Mean"S.D. Range

L2 L3 JQ1

(n) (n) (n) (n)

Total Hg (mgykg)

Site

Basic statistics

155"110 (6) 31–344 252"139 (7) 135–467 139"44 (6) 100–230 207"101 (3) 73–324

JQ2

Mean"S.D. Range Mean"S.D. Range Mean"S.D. Range Mean"S.D. Range

do Sul area, collected during the period 1992– 1996, are shown in Table 4. Maximum concentrations higher than MS occurred in all sampling sites, except H2 and C1, which are the farthest from possible Hg sources in the area, both anthropogenic and natural. In comparison with other gold areas in Brazil, these Hg concentrations in Lavras do Sul are many orders of magnitude lower than those reported for the Madeira River in the Amazon region (Malm, 1990; Pfeiffer and Lacerda, 1988), but are twice or more than those reported by Lacerda et al. (1991) for the Pocone´ region. Mean Hg concentrations were above the MS value (Bowen, 1979) at sites JQ1 and L2. The former site also presented the highest mean values in the area for Fe, Pb, Zn and Cu concentrations (Pestana et al., 1997; Pestana, unpublished data), which suggests the occurrence, at site JQ1, of a local natural Hg source associated with other metal sulfides. Site L2, however, receives anthropogenic contributions from urban sewage and is located downstream of sites where individual gold miners used to work in the beginning of the 1990s.

H1 H2 C1

Total Hg (mgykg) (n) (n) (n) (n)

139"79 (7) 65–223 112"79 (6) 56–280 71"27 (7) 43–120 87"33 (4) 48–140

4.4. Interpretation of spatial trends, anthropogenic and natural Hg sources Hg concentrations in the different geologic matrices and the spatial relations among them were investigated as a tool to distinguish anthropogenic from natural Hg contributions, assuming as ‘anthropogenic’ the concentrationsydeposits resulting from man’s interference in the environment (in this case, mining), and ‘natural’ as those resulting exclusively from processes free of such interference (mineralizations, pre-mining ores, bedrock, etc.). In order to evaluate and compare Hg concentrations from different matrices, the range of concentrations found in the area was classified in ranks, which are multiples of the Bowen’s MS value for Hg (180 ppb). This criterion was chosen because most of the soil and sediment samples were analyzed in the silt–clay fraction. The ranking of the classes proposed are listed in Table 5. All the present data from Hg concentrations in the different geologic matrices analyzed in the Lavras do

Table 5 Ranking criteria for classification of Hg contamination levels according to ranges of concentration divided in classes which are multiples of the Bowen’s mean shale value: 180 ngyg (Bowen, 1979) Rank Class Class Class Class

1 2 3 4

Hg concentration range (ngyg)

MS range

Rank

Hg concentration range (ngyg)

MS range

-90 91–180 181–360 361–1080

-0.5 MS 0.5–1 MS 1–2 MS 2–6 MS

Class 5 Class 6 Class 7

1081–1800 1801–5400 )5400

6–10 MS 10–30 MS )30 MS

134

M.H.D. Pestana, M.L.L. Formoso / The Science of the Total Environment 307 (2003) 125–140

Table 6 Mercury concentrations (ngyg) in different geologic matrices from Lavras do Sula Site description Camaqua˜ Chico Str. Valdo Teixeira h.m.s. Bloco Butia´ h.m.s. Boa Vista h.m.s. Chiapeta’s h.mill ˜ Jose´ h.m.s. Sao Lavras Str. up.urb.; r.g.s. ˜ h.mill; urb. Paredao Lavras Str. down. urb. CRM r.m.s. CRM r.m.s. Lavras Str. r.g.s. Jaques Str. r.g.s. Jaques Str. r.g.s. Cerro Rico h.m.s.; h. mill Cerro Rico h.m.s.; h. mill ´ Hilario Str. up. Cerro Rico ´ Hilario Stream down. C.R.

Sediments 1992–1996

Sediments 1997 CAC–157 (2) VT1–194 (3) BB1–542 (4) BV8–125 (2) L10–172 (2)

Soils 1997

Rocks 1997 14VT–1800 (5) BB3–76 (1)

Afl.10–2100 (6) RMSJ–294 (3)

L1–155 (2) L12–384 (4)

L12B–294 (4)

CRM4–174 (2)

CRM3–195 (3) CRM7–506000 (7)

L2–252 (3) L3–139 (2) JQ1–207 (3) JQ2–139 (2) C1Hc–120 (2)* C1Hp–5233 (6)* H1–112 (2) H2–71 (1)

13CRM–322 (3)

L5–174 (2) 15CR1–53 (1) 15CR1–473 (4)

a

Ex.: ‘CAC-157 (2)’: sample identification—Hg concentration (‘contamination’ class). Abbreviations: Str.sstream; h.m.s.s historic (before 1950s) mining site; urb.surban area; down. C.R.sdownstream from Cerro Rico mining area. Sediment and soil results refer to the silt–clay fraction, except for two bulk samples marked ‘*’, whose results are from Pestana et al. (2000); sampling periods are indicated: ‘1992–1996’.

Sul area are summarized in Table 6, which also includes the results for the stream sediments collected in October 1997, the mean concentrations from the sampling period 1992–1996, and two samples from the Cerro Rico mining area collected in September 1996 (Pestana et al., 2000). The ranks of Hg concentrations in soils, sediments and rocks from potentially contaminated and uncontaminated sites were identified by different symbols and plotted on the lithologic map of the area (Fig. 2). Sample C1 (stream sediment collected during the 1992–1996 period) is not seen in Fig. 2 because its location (Fig. 1, Section 2.1) is not compatible with the map scale. The spatial distribution of Hg concentrations in Fig. 2 shows a great proximity among samples of very different contamination levels. This can be explained by the low solubilityymobility of Hg0, which is the main species in the contaminated soil samples (Table 3) and, according to previous work in the area (Pestana et al., 2000), also predominates in the overbank and stream sediments, favored by present Eh and pH conditions. How-

ever, the disparity in Hg concentrations in nearby samples of rocks, soils and stream sediments suggest anthropogenic contamination of sites Afl.10 (soil, Chiapetta’s mill), CRM7 (soil, CRM mining site) and C1H-p (sediment from tailings pond, Cerro Rico mill), since local highly concentrated sites is also a characteristic of ‘hot spots’. In contrast, soil sample L12B, although collect˜ mill, presents ed nearby the ruins of the Paredao the same Hg concentration (294 ngyg), of granite ˜ Jose´ mine (Tables 2 wastes from the nearby Sao and 3). So, apparently this soil site has presently a ‘natural’ level of contamination, although it might have been polluted in the past by milling activity. Another example of an apparently natural Hg content in a potentially polluted site is the soil sample CRM3, from CRM mining area, which is in the same class of ‘contamination’ of rock sample 13CRM, although with a lower Hg concentration (Table 6). Spatial proximity with comparatively low concentration samples also permits to interpret sediment samples BB1 and L12 as being

M.H.D. Pestana, M.L.L. Formoso / The Science of the Total Environment 307 (2003) 125–140 Fig. 2. Levels of Hg concentration in different geologic matrices from the Lavras do Sul area, Rio Grande do Sul, Brazil, plotted on map adapted from Brasil (1989) and Lima (1995). 135

136

M.H.D. Pestana, M.L.L. Formoso / The Science of the Total Environment 307 (2003) 125–140

Fig. 3. Mercury concentrations in three ‘hot spot’ soil samples collected in Lavras do Sul, showing exponential increase from the oldest to the more recently polluted site.

anthropogenically contaminated. The high Hg concentration of sediment sample BB1 (542 ppb), collected from a pond formed over mining wastes of the Bloco do Butia´ mining area, is not consistent with the concentrations -180 ppb observed in the nearby rock and creek sediment samples (Fig. 3). Supposing that individual miners have worked in the vicinities (Matuella, personal communication) and that amalgamation processes in the past may have been carried on near the Bloco do Butia´ area (Gavronski, personal communication), then this relatively high Hg concentration is likely to have been anthropogenically originated. The same class 4 contamination of sediment sample BB1 was observed in the stream sediment sample L12 (Fig. 3). The enhancement, as compared to the concentrations from nearby sites (L1, L10, soil sample L12B, rock sample RMSJ), could be explained by anthropogenic Hg contaminated ˜ mill having been soil particles from the Paredao washed away by floods (or by partial collapse of the stream bank) and deposited at site L12. The mean Hg concentrations from the stream sediment samples collected during 1992–1996, indicate sites L2 and JQ1 as the more contaminated: class 3. Stream sediment VT1, collected in October 1997, presented this same class of Hg contamination. All the other samples mean Hg concentrations lie in classes 2 or 1. So, according to present data, the Bowen’s mean shale value

(180 ppb) can be considered an appropriate criterion as an ‘environmental threshold’ of Hg contamination in the stream sediments of watersheds drained by granites and andesites in Lavras do Sul. The mean Hg concentration at site L2, although in the same range of the neighboring rock and soil samples, is possibly of anthropogenic origin, at least in part. Data presented by Pestana et al. (2000) showed that most of the sites sampled in 1992–1996 presented concentrations )MS, in at least one of the sampling periods, except H2 and C1. Only site L2 reached concentrations )2= MS in two periods, January 1993 and August 1994, suggesting pulse contamination. These periods coincide with years of greater ‘garimpo’ activities in the area. The origin of the high mean Hg concentration at site JQ1, however, is not so obvious. Although this site also presented a significant decrease in Hg concentrations after 1994, probably related to the weakening of the ‘garimpos’ activity in the whole area of Lavras do Sul (Pestana et al., 2000), other aspects suggest a natural local anomaly: (1) it is the only site where other chalcophile metals (Cu, Fe, Pb, Cd) also presented concentrations higher than the MS (Pestana et al., 1997); (2) it is located very close to a dike (andesite, monzonite or rhyolite) and to a quartz monzonite intrusion, and, for different reasons, both types of lithologic bodies are susceptible of presenting sulfide deposition (Lima, 1995). Following the same approach, the class 3 contamination level of stream sediment VT1 (Fig. 3) can also be regarded as being of natural, rather than anthropogenic origin, taking into account the relatively high Hg concentration of the nearby rock sample 14VT (Table 2, Section 4.1). Finally, a tentative regional background for Hg in stream sediments of watersheds draining the andesites and granites of the Lavras do Sul area can be estimated by an average value of all the samples in classes 1 and 2, which represent 73% of the total stream sediment sampling done in the area (Table 6) from 1992 to 1997: 140 ngyg. The samples in class 3 can be regarded as a product of either anthropogenic contamination (L2) or sulfide-rich rocks (JQ1and VT1). The average Hg concentration for this set of samples is 218 ngyg,

M.H.D. Pestana, M.L.L. Formoso / The Science of the Total Environment 307 (2003) 125–140

which represents 20% of the total spatial distribution, while the Hg concentration of sample L12 (383 ngyg class 4; apparently anthropogenically contaminated) represents 7% of the total number of stream sediment sampling sites. In summary, the comparison of Hg concentrations in different geologic materials and their spatial relations in the Lavras do Sul area showed that stream sediment samples whose Hg concentrations are below 180 ngyg (class 2; Bowen’s MS) can be considered free of anthropogenic contamination, although the background for stream sediments according to the presented criteria is 140 ngyg. The concentrations within class 3 or above it can be either of anthropogenic or natural origin, depending on the proximity to past or recent mining areas or to mineralized lithologies. Besides the criterion of regional background, the distance from milling facilities (past or recent) or to sulfide occurrences has proved useful in the interpretation of natural or anthropogenic origin of the sites contaminated by Hg in Lavras do Sul. 5. Discussion Changes in factors affecting the storage capacities of soils and sediments can produce sudden and often unexpected mobilization of toxic chemicals (in this case, Hg) to the environment, characterizing the phenomena described as CTC, or ‘chemical time bomb’ (Stigliani et al., 1991; Nriagu, 1994; Lacerda and Salomons, 1998). The CTC concept has brought a new insight on the understanding of the function of soils and sediments as sinks, and its dependence on the relativity of their capacity in storing and immobilizing toxic chemicals. The Lavras do Sul average annual input of Hg to the environment from 1909 to 1991, estimated as 0.01 tyyear, is at least 1 order of magnitude less than the annual inputs for highly contaminated areas impacted by gold mining activities worldwide in different time periods, and is quite low if compared to the input of 180 tyyear from the Brazilian Amazon since 1979 until ‘present-day’ (Lacerda and Salomons, 1998). However, taking into account the enormous difference in area comprising the Amazon and the Lavras do Sul regions,

137

then the environmental impact from gold mining activities in this study area, may have been of great importance, as shown by its still present effects (Pestana et al., 2000; this work). Therefore, the annual inputs of Hg to the environment from different regions must not be compared in absolute terms, but should be expressed in terms of area dimensions. In this way, the contributions of small abandoned mining sites throughout the world can constitute a mosaic of information which may sum up to a significant amount to be considered in the evaluation of the global Hg pool. Considering that two of the identified ‘hot spot’ sites in Lavras do Sul were contaminated until the early 1950s (Afl10 and C1H), they indicate Hg pollution persistence in the area to be as old as approximately 50 years, at least. The total Hg concentrations increase exponentially (Fig. 3) from the oldest to the most recently contaminated site: 318 ngyg (Afl10 from 1909 to 1930s), 5220 ngy g (C1Hp) from the 1950s) and 110 000 ngyg (CRM7 from the 1980s) (all concentrations in bulk samples). Obviously, not only time is a factor in the enhancement of concentrations in the more recently contaminated sites, but also the intensity of gold production (and, consequently, of the amalgamation processes) and the local characteristics of each site (organic matter content, Eh, pH, FeyMn oxides and hydroxides, sulfides, Hg0, etc.) are decisive in the maintenance of Hg in relatively undisturbed, or ‘inert’ conditions. Hg volatilization from the older sites during a longer period of time may also account for their present depletion in Hg concentrations. In addition, Pestana et al. (2001) measured Hg fluxes over different geologic substrates from sites Afl10, L12B, CRM7, C1Hp and CR2 and observed linear correlation with substrate concentrations. All these aspects have to be considered if clean up measures are to be adopted. In the Lavras do Sul area, the predominance of litholic soils underlain by andesites, as well as the occurrence of disseminated pyrites and chalcopyrites, are the most probable natural sources of Fe. According to Brasil (1973), these litholic soils are enriched in Fe. The importance of Fe oxides and hydroxides in the adsorption and accumulation of Hg in mineral horizons of ferralitic soils from

138

M.H.D. Pestana, M.L.L. Formoso / The Science of the Total Environment 307 (2003) 125–140

tropical rain forest areas (French Guiana) was reported by Roulet and Lucotte (1995), and podzolic soils were considered a major control in Hg ´ River Basin (Amazon, geochemistry in the Tapajos Brazil) by Roulet et al. (1998). However, the occurrence of podzolic soils does not seem to be related with Hg concentrations in stream sediments of Lavras do Sul, since the restricted occurrence of this kind of soil would more likely influence site H2, where the least mean Hg concentrations were found (Fig. 3, Table 4). Sequential extractions performed in stream sediments collected from selected sites in Lavras do Sul showed Fe distribution in both phases sulfidicy organic (preferentially) and oxidesyhydroxides (Pestana and Formoso, in preparation). Both associations can be expected, considering the occurrence of pyrites and other sulfide deposits as primary sources, and the interaction of many factors leading to the formation of oxy-hydroxides in soils and their exportation to the water system. Such factors are: the pH 5.7 characteristic of the predominant litholic soil (Brasil, 1973), propitiating Fe and Al residual enrichment (both metal hydroxides present least solubilities at pH;6); the absence of the B horizon and the susceptibility to erosion. Since this litholic soil is described as being rich in Fe and quartz (Brasil, 1973), a trend to sandification may be in initial stages. Desertification process is already common in the southeastern part of Rio Grande do Sul state, mainly due to soil erosion (Melfi et al., 1999). Sandification, which is also a final step in podzolization process after the dispersion of the B horizon (Melfi et al., 1999), could lead to Fe and Hg enrichment in the particulate matter of the local water system, in a similar process as that described by Roulet et al. (1998). So, no matter what is the source of Hg, if natural or anthropogenic, once it enters the Hg cycle it is conditioned to the biogeochemical processes that control the fluxes among soily sediments, water and atmospheric compartments. The anthropogenic interference, however, is not restricted to the generation of artificial Hg sources, but also influences the rate of these processes, as it accelerates, for example, soil erosion. In Lavras do Sul, erosion is being stimulated not only by the type of soil existent but also by anthropogenic

deforestation and fire settings to grassland, which have been a historical practice in the area since the last century. According to the present data from Lavras do Sul, no relation concerning Fe oxides in the litholic soils as a ‘secondary’ natural Hg source can be suggested. Instead, the role of Fe oxidesyhydroxides in the Hg cycle of Lavras do Sul can be more related to the transport in the water system as Hg carriers in suspended particles. If so, the continued soil erosion may lead to future Hg accumulations downstream from the local sources. 6. Conclusions Natural sources of Hg in Lavras do Sul are associated with Cu and Fe sulfides, as shown by the analysis of mineralized rock samples from different lithologies, and by the spatial proximity between some of these samples and the relatively high Hg concentrations in nearby stream sediments. Anthropogenically contaminated soils by past and recent milling amalgamation are secondary, but are important Hg sources to the local atmospheric compartment and to the Lavras stream, especially near gold mining sites. Their importance rely on the high Hg concentrations in the silt–clay fraction, which can be easily transported to the river system, especially if erosion rates are enhanced, and on the predominance of the volatile elemental Hg species, that can be emitted to the atmosphere or leached downward to the water table. The persistence of contamination lasting for 50 years or more and the exponential decrease in Hg concentrations from the oldest to the more recently contaminated sites give a faint idea of the amount of Hg that has been lost to the environment during the XXth century in Lavras do Sul, and that is still being released at the more contaminated sites. However, the lack of information on Au production before the 1980s makes the average emission rate of 0.01 tyyear still fairly low. This work permitted the identification of heavily anthropogenically contaminated sites (‘hot spots’ at Chiapetta’s and CRM) and other less contaminated sediments and soils near mining sites or milling facilities, such as Bloco do Butia´ (BB1),

M.H.D. Pestana, M.L.L. Formoso / The Science of the Total Environment 307 (2003) 125–140

˜ (L12) and CRM (CRM3). Natural Hg Paredao sources were found to contribute to Hg concentrations above MS (180 ngyg) and above the background for stream sediments (established as 140 ngyg) at a site from the Jaques Stream (JQ1) and near the Valdo Teixeira mine (VT1). A more detailed study should be carried on in the near future, aiming remedial clean-up actions of the three sites considered ‘hot spots’ in the Lavras do Sul area, namely AFL10, CRM7 (this work) and C1Hp (Pestana et al., 2000). Acknowledgments This work was financially supported by the Brazilian Ministry of Environment FNMAyMMA Hg speciation analyses in soil samples were done by Dr Paul Lechler at the Nevada Bureau of Mines and Geology, USA. The geoprocessing of the map in Fig. 2 was accomplished by Prof. L. Guasselli (CEPSRMyUFRGS). The authors are also thankful to Dr M.C.P. Gastal, Geologist I. Mattuella and to C. Peixoto for their help during the 1997 sampling of rocks and soils. References Bowen HJM. Environmental chemistry of the elements. New York: Academic Press, 1979. p. 333. ´ Brasil, Ministerio da Agricultura, DPP. Levantamento de Reconhecimento dos Solos do Estado do Rio Grande do ´ Sul. Boletim Tecnico No 30, Recife. (1973). pp. 431. ´ ´ Brasil, Ministerio das Minas e Energia, DNPM. Mapa Geolo´ gico do Estado do RGS, escala 1:600 000. Brasılia, DF, 1989. ´ CRM. Relatorio de pesquisa-janeiroy1979 (Vista Alegre e ´ Fazenda da Chacara, Lavras do Sul). Companhia Riogran´ ¸ ˜ Porto Alegre, Relatorio dense de Mineracao, avulso, 1979. ´ ¸ ˜ sobre o distrito auro cuprıfero Gavronski E.F. Informacao de ˆ Vista Alegre—Lavras do Sul—RS, com enfase especial para ´ a jazida da ‘Mina do Cerro Rico’—Relatorio No. 7. Porto Alegre, DOCEGEO. (1976). pp. 127. Kaul P.F.T., Rheinheimer D. Projeto ouro no Rio Grande do ´ final-volume 1. Porto Alegre, Sul e Santa Catarina; Relatorio DNPMyCPRM, (1974). pp. 290. Kim K-H., Lindberg S.E., Hanson P.H., Meyers T.P., Owens J. Application of micrometeorological methods to measurements of mercury emissions over contaminated soils. Proceedings of the International Conference on Heavy Metals in the Environment, CEP Consultants Ltd, Toronto, Canada, Vol. 1. (1993) pp. 328–331.

139

Lacerda LD. Evolution of mercury contamination in Brazil. Water Air Soil Pollut 1997;97:247 –255. Lacerda LD, Salomons W. Mercury from gold and silver mining: a chemical time bomb?. Berlin: Springer-Verlag, 1998. p. 146. Lacerda LD, Pfeiffer WC, Bastos WR. Mercury dispersal in the Pocone´ region, State of Mato Grosso, Central Brazil. ˆ Ciencia e Cultura 1991;43(4):317 –320. Lechler P, Miller JR, Hsu L, Desilets MO. Mercury mobility at the Carson River Superfund Site, west-central Nevada, USA: interpretation of mercury speciation results from mill tailings, soils, and sediments. J Geochem Explor 1997;58:259 –267. ˆ Lima E.F. Petrologia das Rochas Vulcanicas e Hipabissais da ´ ¸ ˜ Shoshonıtica Associacao de Lavras do Sul-ASLS, RS, Dr thesis, Univ. Fed. Rio Grande do Sul, Porto Alegre. (1995). pp. 338. Malm O. Mercury pollution due to gold mining in the Madeira River basin, Brazil. Ambio 1990;19(1):11 –15. ´ Maron M.A.C., Silva A.R.B. Perfil Analıtico do ouro, boletim ´ No. 57, Ministerio das Minas e Energia, Departamento ´ ¸ ˜ Mineral, Brasılia. Nacional da Producao (1984). pp. 143. Melfi A, Cerri CC, Fritsch E, Formoso MLL. Tropical soils: genesis, distribution and degradation of lateritic pedological systems. In: Formoso MLL, Cerri CC, editors. Workshop on tropical soils. Rio de Janeiro, Brasil: Academia Brasileira ˆ de Ciencias, 1999. p. 1 –30. Mexias AS, Formoso MLL, Meunier A, Beaufort D. O sistema ´ hidrotermal fossil de Volta Grande, Lavras do SulyRS Parte 1-petrografia do hidrotermalismo. Geochim Brasil 1990;4(2):139 –157. ´ Nardi LVS, Lima E. Hidrotermalismo no complexo Granıtico ˆ Lavras e vulcanicas associadas, RS. Revista Brasileira Geoˆ ciencias 1988;18(30):369 –375. Nriagu JO. Mercury pollution from the past mining of gold and silver in the Americas. Sci Total Environ 1994;149:167 – 181. ´ ˜ 1. Pesavento S.J. Historia do Rio Grande do Sul. Revisao, Mercado Aberto, Porto Alegre. (1997). pp. 141. Pestana MHD, Formoso MLL, Teixeira EC. Heavy Metals in stream sediments from copper and gold mining areas in southern Brazil. J Geochem Explor 1997;58:133 –143. Pestana MHD, Lechler P, Formoso MLL, Miller J. Mercury in sediments from gold and copper exploitation areas in the Camaqua˜ River basin, southern Brazil. J S Am Earth Sci 2000;13:537 –547. ¸ ˜ de Pestana M.H.D., Lechler P., Formoso M.L.L. Volatilizacao ¸ ˜ de ouro Hg a partir de solos contaminados por mineracoes ´ historicas e recentes em Lavras do Sul, RS, Brasil, Actas ´ ´ do VI Congresso Brasileiro de Geoquımica dos Paıses de ´ Lıngua Portuguesa, vol. 1. (2001). pp. 477–480. Pfeiffer WC, Lacerda LD. Mercury imputs into the Amazon Region, Brazil. Environ Technol Lett 1988;9:325 –330. ˜ Paulo. ¸ ˜ e o Novo Mundo. Cultrix, Sao Prieto C.A. Mineracao (1976). pp. 225. Roulet M, Lucotte M. Geochemistry of mercury in pristine and flooded ferralitic soils of a tropical rain forest in French

140

M.H.D. Pestana, M.L.L. Formoso / The Science of the Total Environment 307 (2003) 125–140

Guiana, South America. Water Air Soil Pollut 1995;80:1079 –1088. ´ Roulet M, Lucotte M, Saint-Aubin A, Tran S, Rheault I, Farella N, Silva EJ, Dezencourt J, Passos CJS, Soares GS, ˜ JRD, Mergler D, Amorim M. The Geochemistry Guimaraes of mercury in central Amazonian soils developed on the ˜ formation of the lower Tapajos ´ River Valley, Alter-do-Chao Para´ state, Brazil. Sci Total Environ 1998;223:1 –24. ´ Saraiva J.A. A face Geologica do Lavras do Sul. Casa de ´ da Silveira, Prefeitura de Lavras do Sul, Cultura Jose´ Neri Lavras do Sul. (1992). pp. 44. Schroeder W.H. Environmental measurements and results of mercury volatilization rates (‘fluxes’) from soil and lake surfaces. Canadian Mercury Network Workshop. Proceedings. 1995.

´ JM. A review of assessment and remediation strategies Skei for hot spot sediments. Hydrobiologia 1992;235y236:629 – 638. Stigliani WM, Doelman P, Salomons W, Schulin R, Smidt GRB, Van der Zee SEATM. Chemical time bombs: predicting the unpredictable. Environment 1991;33(4):4 –30. Teixeira E. Lavras do Sul na Bateia do Tempo, Vol. 1. Casa ´ da Silveira, Prefeitura de Lavras do de Cultura Jose´ Neri Sul, Lavras do Sul. (1992). pp. 529. Ure A.M. The effects of drying on element concentrations and speciation in soils and sediments. In: Quevauvieller, P. and Rauret, G., editors. Workshop on Quality Assurance of Sampling and Sample Pre-treatment for Sediment and Soil ´ Analysis, 8–10 May 1994, Barcelona, Spain, 1994; Quımica ´ Analıtica, 13 (suppl. 1): S64–S69.