Increase in mercury contamination recorded in lacustrine ... - UQAM

3 downloads 0 Views 679KB Size Report
Blandford Forum, Dorset, England. Rev. Geomorphol. Dyn. 32, 95–99. Bruland, K.W., Bertine, K., Koide, M., Windom, H., 1974. His- tory of metal pollution in ...
Chemical Geology 165 Ž2000. 243–266 www.elsevier.comrlocaterchemgeo

Increase in mercury contamination recorded in lacustrine sediments following deforestation in the central Amazon 1 M. Roulet a

a,)

, M. Lucotte a , R. Canuel a , N. Farella a , M. Courcelles a , J.-R.D. Guimaraes ˜ b, D. Mergler c , M. Amorim d

Chaire de Recherche en EnÕironnement, H-Q r CRSNGr UQAM and GEOTOP, UniÕersite´ du Quebec ´ a` Montreal, ´ B.P. 8888, succ. Centre Ville, Montreal, Quebec, Canada H3C3P8 b Laboratorio Inst. de Biofısica, UniÕersidade Federal do Rio de Janeiro, Bloco G-CCS, Rio de Janeiro, ´ de Radioisotopos, ´ ´ CEP 21949-900, Brazil c CINBIOSE, UniÕersite´ du Quebec ´ a` Montreal, ´ Montreal, Quebec, Canada d UniÕersidade Federal do Para, ´ Belem, ´ Brazil Received 19 October 1998; accepted 11 August 1999

Abstract The geochemical study of surface sediments Žvertical profiles of 30–80 cm. from lentic ecosystems of the Tapajos ´ River permit the observation of environmental changes responsible for the mercury contamination of aquatic systems exploited by the human riverine population. The Arapiuns and Amazon rivers are compared. Measurements of mercury, textural indicators Žwater content and dry density., mineralogic indicators Žiron and aluminum associated with oxyhydroxides and aluminosilicates., and organic indicators Žcarbon, nitrogen, atomic CrN ratio. were performed over the full length of the cores. The results demonstrate that soil erosion is responsible for an increase in surficial sediment mercury concentrations in the different aquatic systems of the Tapajos ´ and Arapiuns rivers. This increase is the result of the relative enrichment of the sedimentary deposits in fine particulates rich in aluminosilicates, oxyhydroxides and mercury, transported in suspension in the water column. The oxyhydroxides of iron and aluminum associated with fine, clayey particles seem to control the accumulation of heavy metals in the sediments of the Tapajos, ´ Arapiuns and Amazon rivers. Overall, the mercury levels in the sediments studied have the same relationship with the aluminosilicates and the texture of the sediments. The quantity of aluminosilicates permits the evaluation of diagenetic effects, the influence of the clay content and the matrix effect on the levels of mercury in sediments. The activity of lead-210 measured in two cores suggests that the superficial sediments originate from eroded soils. A preliminary dating using the constant initial concentration model indicates that the environmental changes recorded in the Tapajos ´ River sediments would have been initiated some time between the 1950s and 1970s. They would then coincide with the important colonization of the Brazilian Amazon during this period. The results presented have important implications for the geochemical interpretation of anthropogenic disturbances in the Amazon. They demonstrate that the recent colonization of the drainage basins and the growing exploitation of new parcels of land in the

)

Corresponding author. Tel.: q1-514-987-3000 4080a; fax: q1-514-987-3635; e-mail: [email protected]@er.uqam.ca The present investigation is part of an ongoing study, the CARUSO project ŽCRDI-UFPa-UQAM., initiated to determine the sources, fate and health effects of the presence of MeHg in the area of the Lower Tapajos. ´ 1

0009-2541r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved. PII: S 0 0 0 9 - 2 5 4 1 Ž 9 9 . 0 0 1 7 2 - 2

244

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

central Amazon disturb the mineral and organic matter cycles, as well as that of mercury. The ensuing result is increased exportation of fine particulate mercury from the surface of soils to drainage waters that transport them to fluvio-lacustrine systems where they finally settle out. q 2000 Elsevier Science B.V. All rights reserved. Keywords: Mercury; Amazon; Deforestation; Sediment; Erosion; Normalization

1. Introduction In the Tapajos ´ River numerous studies have shown that higher than normal levels of mercury ŽHg. are found in fish and hair ŽCleary et al., 1994; Akagi et al., 1995; Malm et al., 1995, 1997a; Lebel et al., 1996, 1997; Kehrig et al., 1997.. The concentrations are sufficiently elevated to reveal early signs of sensory and motor deficiencies ŽLebel et al., 1996; Lebel et al., 1998. in the riverine populations as a result of exposure to methylmercury ŽMeHg. from the consumption of fish products ŽLebel et al., 1997.. The presence of MeHg was initially attributed to Hg wastes produced by gold mining activities along the upper Tapajos ´ River ŽAkagi et al., 1994, 1995; Cleary et al., 1994; Malm et al., 1995, 1997a.. More recently, however, a series of detailed studies on forest ecosystems, soils and the water column along the Tapajos, ´ from the valley to the lower reaches, indicated ongoing leaching and erosion of Hg present in the soils ŽRoulet et al., 1996, 1998a,b, 1999a,b.. Along the river no Hg concentration gradient was observed with increasing distance from the gold mining regions ŽRoulet et al., 1998a.. The levels of Hg in the pedologic cover are, however, relatively elevated ŽRoulet et al., 1998b.. The soils constitute an important reservoir of naturally accumulated Hg. In comparison, the inputs from gold mining would represent less than 3% of the Hg present in the first 20 cm of the soil ŽRoulet et al., 1999a.. These studies suggest that throughout the river valley, erosion of fine particles from the soil cover is responsible for the increased export of terrestrial Hg to aquatic ecosystems. The principal objective of the present investigation is to provide temporal evidence of the effect erosion has on Hg deposited on the bottom of lentic systems of the Tapajos ´ River. High resolution vertical profiles of the sediments were produced for quantitative and qualitative analysis of geochemical

indicators of the mineral and organic matter. The importance of the sediment geochemistry on the accumulation and behaviour of Hg is discussed.

2. Materials and methods 2.1. Region, sites and the context of the study The Tapajos ´ is one of the principal tributaries of the Amazon River ŽFig. 1.. It is a clear water river that drains the central Brazilian shield on its upper part and the plateau of the Cretaceous-aged ŽPutzer, 1984. Alter-do-Chao ˜ sedimentary formation on its lower part. In the region of Santarem ´ Ž02825X S, X 54842 W. the maximum seasonal water level variation is 6 m. The average annual precipitation is 2096 mm with regular temperatures of 21.9 to 33.18C for an annual average of 268C ŽInemet, 1979, in Salati and Marques, 1984.. The climate is equatorial with a marked dry season. The section of the Tapajos ´ studied is composed of two distinct aquatic ecosystems, corresponding to two successive physiographic zones in the lower part of the river ŽFig. 1.. The section upstream from Aveiro Ž rio . is formed by a river–floodplain system. The river is relatively narrow Ž2–4 km wide. and characterized by a strong, central advection current. The section immediately downstream from Aveiro, where the Tapajos ´ suddenly widens and the current slows, constitutes a large sedimentation zone characterized by numerous islands and channels ŽSioli, 1984.. From this zone until its meeting with the Amazon, a river lake Ž ria or mouthbay-lake. 8–15 km wide and having a weak current and is characteristic of several of the major tributaries of the Amazon. As a means of comparing the Hg geochemistry in the sediments of different drainage basins, the ria of the Arapiuns River and a varzea ´ of the Amazon River were also studied ŽFig. 1.. The Arapiuns River,

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

245

Fig. 1. Map of the study region and sampling sites.

the major tributary of the Tapajos ´ and characterized by very clear water, is located upstream from the point where the Tapajos ´ flows into the Amazon. In the Tapajos, ´ four lakes were chosen in the upstream zone. They represent two lake types characteristic of Amazonian floodplains ŽHamilton and Lewis, 1990; Sippel et al., 1992.. The channel lakes, such as the lago Piranga ŽAM43, on the edge of the river. and the lago da Ilha Itapena ŽAM-60, in the middle of an island., are small, elongate lakes Ž; 0.02–0.04 km2 . that are fed directly from advection of the river. Plankton development is negligible during all seasons. During the rainy season, these lakes are completely flooded by high water flow in the river. The low residence time of the water in this type of lake implies that the water column is directly influenced by inputs of material from the rio. In contrast, the channel lakes are completely isolated during the dry season. Wind action at the surface of these lakes is seriously restricted by their size as well as by the canopy of the flood plain forest Ž igapo´ . along the river banks. No resuspension of bottom sediments occurs during the dry season.

The lago Enseada Grande ŽAM-32. and the lago do Tavio ŽAM-65. are the largest lakes Ž; 0.3–1 km2 . and are more or less circular Žsimilar to the dish-lakes as described by Hamilton and Lewis, 1990.. They are continuously linked to the river by access canals but are less directly influenced by water of the Tapajos. ´ Water inflow is principally via the canal, even during the rainy season. The current is very weak and the residence time is longer than that of the channel lakes. Inundation due to overflow of the river banks occurs only when maximum water levels are attained. These lakes are always less turbid than the river during the rainy season and are practically without current, permitting plankton development. They represent large, wind exposed surfaces susceptible to sediment resuspension Že.g., in Enseada Grande during the dry season.. The lago do Arrozalzinho ŽAM-102., situated on Ituqui Island downstream from Santarem ´ ŽFig. 1., can also be included in this last lake category, although much larger Ž3–4 km2 .. It is directly and uniquely influenced by the Amazon during the rainy season, but remains completely isolated during the dry season.

246

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

The Tapajos ´ River ria is crossed, season long, by waters coming from upstream. In the ria, two stations were chosen in deep regions Ž10–20 m., the first one lying just beyond the sedimentation zone ŽAM-73. and the second located in the middle of the ria ŽAM-80.. In the Arapiuns, only the downstream section was sampled. The rias are characterized by a weak current and significant phytoplankton and zooplankton development ŽSchmidt, 1982; Sioli, 1984.. At the inlet of the ria, the Tapajos ´ loses approximately 50% of its fine particulate load and nearly all of the coarse material ŽRoulet et al., 1998a.. Since the beginning of the century, the drainage basins of the Amazon have been affected by numerous anthropogenic inputs. During the 1960s and 1970s, the initiation of large-scale development projects in the Amazonian frontier of Brazil promoted unprecedented colonization and exploitation in the region ŽKohlhepp, 1984.. In the Tapajos ´ Valley, these projects comprise trans-Amazonian routes ŽTransamazonica BR 163, Cuiaba-Santarem ˆ ´ ´ PA 150., installation of large agricultural colonies, the colonization of the river banks, mining and logging activities ŽKohlhepp, 1984.. From the end of the 1970s until the beginning of the 1990s, gold rush have greatly affected the region upstream from the river ŽBidone et al., 1997; Malm et al., 1997a.. In the late 1980s, the region above the Tapajos ´ was the largest gold mining center but between 1990 and 1992 experienced a considerable decrease in activity ŽMalm et al., 1997b.. In comparison, the Arapiuns river is practically free of mining activities and its banks are relatively unpopulated. 2.2. Sampling methods All of the sediment samples were collected during 1995. Two sampling methods were employed. Short cores Ž25–30 cm. and floc from the sediment–water interface were collected manually by scuba divers. The cores were retrieved by gently inserting a PVC tube Ž30 cm long and 15 cm in diameter. into the sediments. The floc samples were collected directly from the interface with 50 cc plastic syringes. Long cores Ž80–100 cm. were collected using a Mackereth corer connected to a scuba tank and controlled from the surface.

The two coring methods allow for sampling with a minimum of compaction or disturbance to the sediments. Following retrieval, a jack is used to extract the sediments from the corer at 0.5 cm intervals. Consecutive slices are carefully cut and transferred to polyethylene centrifuge tubes and then immediately placed in a freezer. Prior to sub-sampling the sediments, redox potential Ž E h . profiles were determined by inserting a small platinum electrode at numerous depths along the length of the core. 2.3. Laboratory analyses and calculation All of the sediment samples were freeze dried. The samples were weighed before and after drying to determine water content Ž% water.. The dry densities Ž d . were obtained by measuring the volume of a known weight of sediment. The volumes were measured directly in the tubes following centrifugation of the unfrozen samples. The percent water content and the dry density measurements ŽFig. 3. were used to evaluate textural variations of the sediments and changes in the deposition history ŽSmith and Walton, 1980; Eakins et al., 1983; Flower et al., 1984; Droppo and Stone, 1994; Foster and Walling, 1994.. The Hg analyses are performed using cold vapour atomic fluorescence spectrometry ŽCVAFS. following a modification ŽPichet et al., 1999. of the method developed by Bloom and Fitzgerald Ž1988.. Briefly, the sediments were digested in glass test tubes with a 10:1 mixture of nitric acid and hydrochloric acid ŽHNO 3rHCl. and heated at 1208C. Carbon ŽC. and nitrogen ŽN. were determined in all of the samples using a Carlo-Erba elemental analyzer. The iron and aluminum oxyhydroxides were measured using the protocol developed by Lucotte and d’Anglejan Ž1985., a chemical extraction with a citrate–dithionite–bicarbonate Žcdb. buffer. Total iron ŽFe tot. . and total aluminum ŽAl tot. . were extracted from samples in Teflon bombs in a 6:1 mixture of aqua regia and concentrated hydrofluoric acid ŽHF. heated in a microwave. The concentrations of Fe cdb , Al cdb , Fe tot., and Al tot. were determined by atomic absorption spectrometry. The fraction associated with the silicate matrix ŽFe si. and Al si. . was

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

calculated by subtraction of the cdb fraction from the corresponding total quantity ŽFontes, 1995; Jonsson, 1997.. The cdb extraction is specific for Fe oxyhydroxides and corresponds to crystalline and amorphous forms ŽJonsson, 1997.. Al cdb represents Al substituted for Fe in Fe oxyhydroxides, as well as amorphous Al. The cdb method also seems to be able to extract Al and Fe associated with organic matter ŽParfitt and Childs, 1988; Ross and Wang, 1993.. Al si. corresponds to Al atoms that are incorporated in clay minerals of the sediments, while Fe si. represents Fe atoms substituted for Al in the same minerals ŽJonsson, 1997.. The analysis of lead-210 Ž210 Pb. by alpha spectrometry is based on the postulate of a secular radioactive equilibrium between 210 Pb and its daughter, polonium-210 Ž210 Po.. The samples were analyzed more than 6 months following their collection in order to respect this equilibrium. Approximately 1 g of dry sediment was initially treated according to a protocol developed by Courcelles Ž1998.. Sample deposits on a silver disk are dried and then placed in one of the vacuum detection chambers of the alpha spectrometer ŽEGG-Ortec, type 576A.. The excess 210 Pb activities Ž210 Pb u . by the parent 226 Ra are calculated by assuming that the activity of 210 Pb in equilibrium is uniform in each core. In this case, the constant activity of 210 Pb with depth represents the 210 Pb in equilibrium with 226 Ra. The 210 Pb u can also be estimated by subtracting the total stable activity measured at the base of the core profile from the activity measured closer to the surface. The two alternative geochronological models used are the constant initial concentration ŽC.I.C.. model and the constant rate of supply ŽC.R.S.. model ŽGoldberg, 1963; Appleby and Oldfield, 1978; Robbins, 1978; Appleby and Oldfield, 1983.. If the C.I.C. conditions are satisfied, the 210 Pb u activities vary with depth according to the following equation: C s C0 eyk t , where C0 is the sediment 210 Pb u activity at the sediment–water interface and k is the radioactive decay constant for 210 Pb Ž0.03114 yry1 .. The age of sediment activity C is given by: t s 1rk lnŽ C0rC .. If the C.R.S. conditions are satisfied, it can be presumed that the cumulate residual 210 Pb u activity, A, below sediments of age, t, varies according to the following formula: A s A 0 eyk t , where A 0

247

is the total accumulated residual activity in the sediment column. A and A 0 are calculated by the direct numerical integration of 210 Pb u profiles as a function of the cumulate mass. The age at a particular sediment depth is given by: t s 1rk lnŽ ArA 0 ..

3. Results 3.1. Vertical profiles in the sediments In all of the cores, the Hg profiles ŽFig. 2. show a more or less marked increase at the surface. For several centimeters below this increase, the concentration profiles show less elevated and more stable Hg levels. The increases in Hg concentrations correspond to the changes in the physical quality of the sediments ŽFig. 3.. Aside from the AM-80 core, all of the profiles for C also show a superficial increase. Several of the profiles delimit, more or less, a leveling off at the surface ŽFig. 4.. The CrNatom ratio varies inversely with C. In all of the profiles Žexcept AM-102., the cdb fractions for Fe and Al follow the variation of the silicate-associated Al fraction ŽFig. 5.. In the floodplain lakes ŽAM-60, AM-43, AM-32 and AM-65., the surface concentrations vary little from one core to another, ranging from 200 to 250 ngrg. The cores AM-43 and AM-60, corresponding to lakes more directly influenced by the river Ži.e., channel lakes ., demonstrate a drastic change from the general Hg concentration profiles, with maximum increases of 5–20 times greater than basal concentrations. The marked density variations in the sediments do not appear to be the result of simple progressive compression under the weight of overlying sediments. Visually, we noted important variations in the proportion of fine material at the surface as compared to a preponderance of silty andror sandy material mixed with fine matter Žand, hence, denser. with depth. These observations suggest a change in the relative quality of deposited sediments and corresponds to an increase in flocculated fine material. The amplitude of the superficial C and CrNatom variations in vertical profiles is significant. The quantities of Fe cdb and Al cdb increase in the surficial zone where Eh show lower values.

248

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

Fig. 2. Profiles of mercury concentration Žngrg dry weight. in sediments.

The core profiles from the two large circular lakes Ž dish-lakes AM-32 and AM-65. show a progressive and less pronounced increase in Hg concentration than in channel lakes, with the surface sediments being only 1.2–2 times greater than those at depth.

The variations in water content and density with depth are less evident and, with visual inspection, do not appear to correspond to changes in the granulometry. The surface sediments are more flocculated than those at depth where they become clayey,

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

249

Fig. 3. Profiles of water content and dry density in sediments.

clumped and compact. In superficial sediments, the Fe cdb and Al cdb profiles do not follow the Eh variations. The AM-32 profile, however, demonstrates a much greater increase in oxyhydroxides as compared to aluminosilicates.

In floodplain lakes, the Fe cdb and Al cdb contents more closely follow aluminosilicate contents than the redox conditions. The increase of oxyhydroxides at the surface does not appear to be the effect of remobilization and reprecipitation of these redox

250

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

Fig. 4. Profiles of carbon content and atomic CrN ratio in sediments.

sensitive elements. In following the siliceous fraction we can account for the increase in the quantity of clay and associated oxyhydroxides in the sediments. In the Tapajos ´ ria ŽAM-73 and AM-80., the superficial increase of Hg levels is greater at the inlet

ŽAM-73. of the system than in the middle ŽAM-80. where the change in concentration is more progressive for the same sediment thickness. The surface concentrations correspond to increases of 3 and 1.5 times those of the basal levels for AM-73 and AM-80,

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

respectively. The marked density variations at the inlet of the Tapajos ´ ria profile ŽAM-73. illustrates the presence of sediments that are finer and richer in water than those for channel lakes ŽAM-43 and AM-60.. The AM-80 core shows very weak and progressive changes at the surface. The sediments of the core comprise fine, flocculated material which, over the length of the core, do not appear to have been affected by compression. At the beginning of the Tapajos ´ ria the surface levels of C increase gradually but distinctly. The AM-80 core, which shows an exceptional decrease in C at the surface, has relatively high C Ž) 3%. in comparison to other cores. The two cores from the ria show lower Eh values just below the surface. The general shape of the Fe cdb profile is not changed but a net impoverishment in oxyhydroxides, as compared with Al si., is observed at this depth. It is possible that a portion of the Fe oxyhydroxides is remobilized in this reducing zone of the sediments. The effect of this reduction on the vertical profiles does not, however, alter the absolute increase in concentrations of Fe cdb and Al cdb corresponding to an enrichment of clayey material, as indicated by the increase in Al si. concentrations. At the end of the Arapiuns ria ŽAM-94., the Hg profile shows a net progressive increase of 1.3 times for the surface sediments. For more than 60 cm below this increase, the Hg levels are relatively stable. The sediments of the core are particularly rich in water Ž85%–95%. and show imperceptible variations in the density. These sediments have C contents of ) 4.8% and show the least variability with depth. In the Arapiuns ŽAM-94., the levels of oxyhydroxides progressively increase at the surface where the sediments have lower E h values. The sediment–water interface clearly shows a favourable effect on the precipitation of oxyhydroxides in the floc. As was observed for the sediments of the Tapajos ´ rio, the progressive increase in the cdb fraction follows that of the silicate fraction for over 20 cm below the interface. The depth profile of Hg in a Õarzea lake of the ´ Amazon ŽAM-102. shows the lowest concentrations with two distinct levels. Sediments from the surface contain more Hg, with a maximum of 130 ngrg between 8 and 12 cm, while below this, for more than 25 cm, the Hg data do not vary. The surface result indicates an increase of 1.3 times, with a

251

maximum just below of 2 times the basal concentrations. Although, the dry densities for the sediments of the Õarzea ´ lake are not available, the % water results are similar to those of the circular lakes of the Tapajos. ´ This core is the only one to yield highly modified oxyhydroxide profiles, indicative of diagenetic effects. The profiles demonstrate significant Fe cdb and Al cdb precipitation at a depth corresponding to the base of surficial C enrichment ŽFig. 5.. The sediments have higher Eh values at this depth. The C increase corresponds to a slight decrease in Al si. and Fe si. in the superficial part of the profiles and appears to be independent of the oxyhydroxides. There seems to be a relationship between the superficial accumulation of C, the oxyhydroxide geochemistry and the redox potential in these sediments. 3.2. Hg geochemistry of the sediments The Hg concentrations measured in all of the cores collected in the basins of the Tapajos ´ and Amazon Rivers are compiled according to different geochemical and textural indices ŽFig. 6.. Inter-indicator relationships in the sediments were investigated by linear regression analysis, with the resulting correlation matrix shown in Table 1. The core from the Arapiuns ŽAM-94. was excluded from the calculation of the regression parameters because the interindicator relationships for this core always indicate it to be an outlier as compared to the other sediments. The sediments from the end of the Arapiuns ria are clearly distinguished from those of the Tapajos ´ River by their relatively low densities despite similar concentrations of Hg, their lower Hg levels as compared with C and oxyhydroxides and their slight enrichment of Hg as compared to Al si.. In the sediments from the Tapajos ´ and Amazon rivers, Hg is significantly correlated with all indicators, particularly with Al si., water content and N. Although a strong relationship exists between C and N, the correlation between Hg and N is better than that with C. The relationships observed are in contrast to those previously noted for soils of the region where no correlation was found between Hg concentrations and C and N ŽRoulet et al., 1998b.. The most marked deviations in C as compared to Hg are noted in core AM-43 and appear to correspond to the presence of coarse plant debris observed in sedi-

252

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

Fig. 5. Profiles of the different fractions of iron and aluminum, and redox potentials in sediments.

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

Fig. 5 Žcontinued..

253

254

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

Fig. 6. Relationships between Hg concentration and water content, dry density, carbon, nitrogen, iron and aluminum oxyhydroxides and aluminosilicate concentrations in the different sediments.

M. Roulet et al.r Chemical Geology 165 (2000) 243–266 Table 1 Correlation matrix of Hg concentrations, geochemical and textural indicators for the Tapajos ´ and Amazon river sediments p- 0.0001 except U ps 0.0285; UU ps 0.0937; UUU ps 0.8127; UUUU ps 0.0025; UUUUU ps 0.0083. % eau d Hg 0.81 % eau d C N Fe cdb Fe si Al cdb

C e

N

0.65 0.63 0.78 0.71 0.73 0.77 0.63 d 0.67 c 0.92

Fe cdb Fe si 0.52 0.59 0.60 b 0.23 0.38

Al cdb Al si UUUUU

0.05 0.078UUUU 0.18 0.00UUU 0.02UU 0.30

0.58 0.57 0.39 0.31 0.42 0.60 0.04U

0.87 0.74 0.62 a 0.39 0.55 0.46 0.15 0.40

Note: best fit with 2nd degree polynom. a 2 r s 0.71, p- 0.0001. b 2 r s 0.66, p- 0.0001. c 2 r s 0.75, p- 0.0001. d 2 r s 0.74, p- 0.0001.

ments Žarrows in Fig. 6c.. The debris probably originated from the Igapo´ Žseasonally flooded forest. surrounding this little lake. C and N were only weakly correlated with the mineral matter indices ŽFe cdb , Al cdb and Al si . but show stronger relationships with textural indices Ž d and % water.. Inter-relationships between Hg, d, % water, Al cdb , Fe cdb and Al si show that all of these indices are significantly correlated with each other ŽTable 1.. Between the different cores and over their lengths, the Hg concentration profiles demonstrate the same relationship with the % water and d. The only noted

255

deviation is with the deep sediments collected from the circular lakes of the Tapajos ´ plain which, below the flocculated surface, have a clayey, lumpy and compact texture. The relationships of Hg with mineral matter demonstrate that the Al si. inputs coincide with those of Hg over the length of the profiles and among all sediments of the Tapajos ´ and Amazon rivers. Weaker relationships between Hg and Fe cdb and Al cdb are observed. They present several deviations that are probably related to diagenesis, as well as the specific oxyhydroxide mineralogy Ži.e., quality and size of the crystals, Al substitution, interaction with organic matter.. All of these factors greatly influence the adsorption capacity and the availability of adsorption surfaces. The quantity of oxyhydroxides in the clayey matrix are equally variable, depending on the texture of the particles and the type of soil in the Tapajos ´ Valley ŽRoulet et al., 1998b.. The profiles of the Õarzea, showing a decoupling between the cdb and ´ siliceous fractions, demonstrate a similar relationship between Hg and the quantity of Al si.. 3.3. Profiles of

210

Pb and geochronological models

Profiles of the total 210 Pb activity as a function of the dry material accumulated in the long cores from the flood plain, lago Piranga ŽAM-43., and the ria ŽAM-73. seem to follow the increase of aluminosilicates, oxyhydroxides, C, as well as the textural variations noted at the surface ŽFig. 7.. The profiles

Fig. 7. Profiles of 210 Pb activities in AM-43 and AM-73 sediments as a function of cumulative dry matter weight.

256

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

Fig. 8. Relationships between time and depth in AM-43 and AM-73 sediments calculated with C.R.S. and C.I.C. geochronological models.

of excess 210 Pb Ž210 Pb u . are not regular and therefore indicate that the classic radioactive decay conditions are not respected. In such a case, the geochronology cannot be established using a simple constant fluxconstant sedimentation rate model ŽCF:CS, Robbins, 1978.. The C.I.C. model ŽGoldberg, 1963; Robbins, 1978. was developed for systems where an increase in sedimentation is associated with an increase in 210 Pb u inputs, be they the result of erosion of surface material containing appreciable quantities of 210 Pb u or due to a significant adsorption of available 210 Pb by particles in the water column. Another model, the C.R.S. model is applicable to systems where, even before a perturbation event, the transfer of 210 Pb, via particulate matter, from the water column to the sediments is an efficient route. For this model, an increase in particulate inputs cannot provoke a depositional increase in the isotope. The levels of residual 210 Pb are comparable to the atmospheric flux as well as to those of other cores from the same region. The 210 Pb flux is considered to be constant. The application of the two models to the surface sediments yields different chronologies ŽFig. 8..

4. Discussion 4.1. Normalization of the quantities of Hg in the sediments It has already been demonstrated that even if clays do not play a direct role in the sequestration of

trace metals, they may act as a physical substrate for the transport, precipitation and flocculation of organic matter and oxyhydroxides which more strongly absorb metals ŽWindom et al., 1989; Ravichandran et al., 1995.. Numerous studies have demonstrated that As, Co, Cr, Cu, Fe, Pb, Mn, Ni, and Zn significantly correlate with Al tot., Fe tot. and grain size, suggesting that natural aluminosilicates are the support of natural trace metals ŽBruland et al., 1974; Goldberg et al., 1979; Trefry et al., 1985; Sinex and Wright, 1988; Windom et al., 1989; Schropp et al., 1990; Alexander et al., 1993; Morse et al., 1993; Daskalakis and O’Connor, 1995; Shine et al., 1995.. The Fe tot. and Al tot. concentrations are also frequently used for the normalization of naturally occurring metals to their anthropogenic counterparts. In these studies, however, Cd and, especially, Hg were generally found to not correlate with these parameters as a result of the greater natural contribution of organic matter to their adsorption by sediments. In the present study, since the mineral indicators are linearly inter-correlated, they can be normalized to each other ŽTable 2.. Where data permitted, the same normalizations for soils ŽRoulet et al., 1998b. and suspended particulate matter ŽRoulet et al., 1999b., as well as the mean HgrAl atomic ratios for the earth’s crust, sediments and soils ŽBowen, 1979; Salomons and Forstner, 1984. were also compiled. ¨ HgrAl si. atomic ratios in the Tapajos ´ sediments are higher than the mean for the earth’s crust but similar to overall mean for sediments, shells and soils. Values for soils from the Tapajos ´ Valley are higher than the mean earth soils. Along the Tapajos ´ Valley,

Table 2 Elemental atomic ratios in sediments, suspended particles and soils from the lower Tapajos ´ Valley and in the earth’s surface HgrAl si. Ž=10 6 . surf. 0.26 0.21 0.23 0.22 0.18 0.2 0.33 0.18

Soils Oxisol AM-78 a Oxisol AM-114a Oxisol AM-25a Ultisol AM-09 a Organic horizons c

deep 0.35 0.16 0.19 0.19 0.15 0.2 0.32 0.12

0.74 1.31 1.21 1.16 1.20 1.00 1.03 1.50 1.14

Fe cdb rAl si. Ž=10. surf. 0.68 0.84 0.84 0.82 0.79 0.71 2.1 0.85

deep 1 0.76 0.48 0.73 1.3 0.71 1.9 1

EF

0.68 1.11 1.75 1.12 0.61 1.00 1.11 0.85 1.03

Al cdb rAl si. Ž=10. surf. 0.18 0.22 0.25 0.16 0.26 0.15 0.72 0.1

deep 0.45 0.23 0.16 0.13 0.3 0.14 0.65 0.14

EF

0.40 0.96 1.56 1.23 0.87 1.07 1.11 0.71 0.99

HgrFe cdb Ž=10 6 . surf. 3.82 2.65 2.73 2.65 1.29 2.82 1.52 2.15

deep 3.63 2.15 4.72 2.65 1.25 2.79 1.65 2.13

EF

1.05 1.23 0.58 1.00 1.03 1.01 0.92 1.01 0.98

HgrAl cdb Ž=10 6 . surf. 15.4 9.98 9 13.9 7.14 13.7 4.56 18.4

deep 9.54 7.06 13 14.8 4.69 13.7 4.7 10.1

EF

1.61 1.41 0.69 0.94 1.52 1.00 0.97 1.82 1.25

HgrC surf. 2.97 2.31 3.39 3.59 3.53 4.45 1.54 1.99

EF deep 3.74 1.88 2.98 3.7 2.87 3.01 13.7 3.33

0.79 1.23 1.14 0.97 1.23 1.48 0.11 0.60 0.94

AlrAlqFe cdb CrN surf. 0.2 0.21 0.23 0.16 0.16 0.17 0.29 0.11

deep 0.28 0.24 0.28 0.15 0.22 0.17 0.26 0.15

surf. 13.7 16.2 12.9 12.9 13.0 15.1 12.5 14.2

deep 16.2 27.9 13.4 13.8 19.1 15.9 14.9 16.5

0.25

0.66

0.42

3.74

6.02

5.42

0.39

16.8

0.33

2.26

0.49

1.40

6.27

6.50

0.18

20.76

0.51

1.47

0.75

3.59

7.09

8.10

0.34

23.52

0.68

2.14

1.11

3.18

6.57

5.24

0.27

13.24

4.6–17.8

5.9–34.5

0.19–0.79

0.30–0.48

17.0–21.8

3.13 6.26

6.67 49.05

3.75 1.72

0.33 0.12

12.2 14.3

Suspended particles FPM CPM

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

Sediments AM-60 AM-43 AM-32 AM-65 AM-73 AM-80 AM-94 AM-102 Mean

EF

Mean earthb Crust 0.082 Sediment 0.20–0.35 Shale 0.28–0.30 Soils 0.11–0.20 a

Mineral horizons. Ž1984.. HgrAl after data compiled by Bowen Ž1979. and Salomons and Forstner ¨ c Range of mean values. b

257

258

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

HgrAl si., HgrFe cdb , HgrAl cdb , and HgrC atomic ratios in sediments are similar or slightly lower than those of soils and fine particulate matter ŽFPM, - 63 mm.. This indicates that sediments have the tendency to be less enriched in Hg than soils and suspended particles. This apparent impoverishment of Hg could be due to a desorption during transport or sedimentation, or early diagenesis, but could also be the result of a difference in the quantity of oxyhydroxides relative to aluminosilicates, as shown by the variation of Fe cdbrAl si. and Al cdbrAl si. atomic ratios ŽTable 2.. HgrFe cdb , HgrAl cdb and HgrC atomic ratios in soils and FPM are similar, indicating the pedological source and the natural geochemical equilibrium of the Hg associated with particles in the river ŽRoulet et al., 1999b.. An enrichment factor ŽEF. for Hg and Fe cdb and Al cdb in the surface Žsurf.. relative to the subsurface Ždeep. sediment normalized to Al si. can be calculated according to the formula ŽHorovitz, 1991.: EF s ŽHg surf.rAl si. surf..rŽHg deeprAl si. deep. ŽTable 2.. EFs for Hg relative to Fe cdb , Al cdb and C are also given. In all cores, EFs are close to 1 indicating the absence of an anthropogenic component in the sediment Hg load. When EF values for Hg relative to Al si. are slightly higher than 1, a similar trend of the EF values for Fe cdb and Al cdb relative to Al si. is noted in most of the cores. EFs of Hg relative to Fe cdb are very close to 1 in the majority of the cores suggesting that along sediment profiles, as with soils ŽRoulet et al., 1998b., Hg is in equilibrium of with the Fe oxyhydroxide phase. The variation of these ratios, and hence the variation of EFs, may be influenced by diagenesis, change in oxyhydroxide adsorption surfaces, as well as, changes in the quality and type of oxyhydroxides. Another factor of variation could be the presence of more or less Al and Fe linked to organic matter in the cdb extract. These geochemical factors do not sufficiently alter the relationship between Hg and textural and aluminosilicate indicators in the Tapajos ´ sediments. In the soils of the Tapajos ´ Valley, Hg follows Al cdb and, in a less pronounced manner, Fe cdb ŽRoulet et al., 1998b.. The profiles of these elements in the Tapajos ´ sediments show the same trend. The relationship between Hg and C and N is principally attributed to the effect of a clayey matrix. In the profiles, Hg does not follow C but the concentrations

increase with the different mineral indicators considered, and C follows textural indicators rather than mineral indicators. As in the soils of the region ŽRoulet et al., 1998b., as well as the suspended matter of the Tapajos ´ river ŽRoulet et al., 1999b., it is the oxyhydroxides incorporated into the aluminosilicate matrix that appear to be responsible for the adsorption of Hg by the sediments ŽGibbs, 1977.. However, even if oxyhydroxides are excellent absorbents of Hg in soils, they are not the best candidates for the normalization of natural Hg levels Žoriginating from the alteration and erosion of the soil cover. as compared to other Hg inputs Žanthropogenic contamination. to the sediments. They, themselves are subject to redox andror geochemical transformation processes during their transport and following their deposition. The fact that cdb also extracts Al and Fe associated with organic matter introduces another possible factor of variation. These processes can significantly modify the initial or apparent adsorption capacity of the material of concern. Aluminum is the element most commonly used to trace the natural and anthropogenic origin of metals accumulated in sediments ŽRavichandran et al., 1995.. Even if it is preferable to use Al si., the use of Al tot. yields a satisfactory estimation of the quantity of aluminosilicates since the cdb-extracted fraction of Al is often low as compared to the siliceous fraction Ž- 10%.. Al si. compensates for the effects of grain size and composition variations, redox processes and sedimentation on the concentration of metals in the sediments. The Al si. concentrations represent the quantity of aluminosilicates; the natural adsorption matrix for Hg originating from the alteration andror surficial erosion of soils in the drainage basin of the Tapajos ´ River. 4.2. Change of sediment source In the case of colonized drainage basins such as that of the Tapajos, ´ it is important to take into consideration numerous possible ground cover alterations and their effects on the quantity and quality of the drained material. The development of mines, forestry practices, crop and pasture land, and roads promotes the loss of forest cover and is followed by the progressive exploitation of the underlying soils. As a result the soils are subjected to increased and

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

more invasive erosion. The bottom sediments of aquatic systems may also be subjected to resuspension as a result of dredging carried out by gold miners during the dry season. Aluminum concentrations and size of the deposited particles are often used as indicators of erosion of fine clayey particles from cultivated soils ŽMackereth, 1966; Binford, 1983; Engstrom and Wright, 1983; Engstrom et al., 1985; Foster and Walling, 1994; De Boer, 1997.. An increase in the proportion of fine sediments andror suspended particles may lead to increased flocculation ŽDroppo and Stone, 1994; Milligan and Loring, 1997.. The increase in Al si. in the surface sediments is a probable indication of such effects on the finer, less dense and more water-rich surface sediments. Additional evidence of soil erosion processes was observed by the variation of organic matter quality in some of the cores ŽFarella et al., 1996.. These researchers observed that variations in the quantity of C and the CrNatom ratio correspond to a complete change in lignin-derived phenols as a result of erosion processes affecting the organic and mineral horizons of the drainage basin soils. The indicators suggest that the geochemical and textural changes observed in the surface sediments are the result of an increase in fine particles from the soils of the drainage basin. These changes correspond, in turn, to increased overland flow as a result of clear cutting. The absence of forest cover and the development of agriculture encourages the surface erosion of fine, mineral particles rich in Hg ŽRoulet et al., 1998b.. Erosion during the rainy season increases the quantity of fine terrigenous suspended particles transported by the river ŽRoulet et al., 1999b.. The subsequent sedimentation of these fine particles increases the Hg concentrations measured in the surface sediments of the flood plain and ria of the Tapajos ´ river. 4.3. Sedimentation of Hg in the flood plains and rias The turbidity of the Tapajos ´ in the zone upstream from Aveiro ŽFig. 1. is always relatively elevated in comparison to the ria ŽIrion, 1984; Roulet et al., 1998a; Roulet et al., 1999b.. In the turbid systems, the water of the alluvial flood plains undergoes progressive decanting as a function of current and

259

geomorphology ŽBrown, 1983; Sioli, 1984; Pizzuto, 1987; Marriott, 1992; Asselman and Middelkoop, 1995; Walling et al., 1996.. The sedimentation is generally greater, and the sediments deposited coarser, in the zones close to the channel of the river, as compared to those further away ŽSioli, 1984.. These conditions explain the variations in texture and quantity of aluminosilicates observed between the different cores of the flood plain. The small channel lakes ŽAM-60 and AM-43., as compared to the circular lakes ŽAM-32, AM-65 and AM-102., are characterized, at depth, by coarser sediments relatively poor in Al si. and Hg. Similar effects are anticipated downstream ŽHe and Walling, 1997.. In the ria, the current determines the sedimentation at the inlet Žsedimentation zone. as well as along the length of the system via its influence on the load of organo-mineral material transported and progressively decanted ŽIrion, 1984; Roulet et al., 1998a, 1999b.. The sedimentation in the middle of the ria ŽAM-80. is richer in Hg and fine Al si. than at the inlet ŽAM-73.. The greater levels of C in the Tapajos ´ ŽAM-80. and Arapiuns ŽAM-94. rias suggest coagulation of the mineral particles with the colloidal organic matter ŽChauvel et al., 1996; Roulet et al., 1999b., particularly in the Arapiuns. This effect is suggested by suspended particles which appear to be enriched in C, Al cdb , Fe cdb and Hg ŽRoulet et al., 1999b.. In the AM-94 core, the relationships between the different indicators are visibly modified as compared to those of the other sediment types ŽFig. 6.. The effect of erosion on Hg geochemistry in sediments depends upon the sedimentation conditions prevailing prior to the system perturbation. Increased sedimentation of fine material provokes a significant increase in superficial Hg concentrations in sediments of the channel lakes and at the beginning of the ria. In the circular lakes and at the end of the ria, the sediments are already richer in fine, clayey material as evidenced by Al si. at the base of the cores. In these lakes, erosion has a relatively less pronounced effect on the variation of Hg concentrations and on the quality of surface sediments. The resuspension of bottom sediments in the circular lakes during the dry season may also contribute to the mixing or homogenization of the sediments. In the long term, these perturbations attenuate the effect

260

Table 3 Comparative data on the use of

210

Pb in tropical America and the world

Western Europe core satisfying the C.R.S. criteria Bolivian Ande Valley, C.R.S. sediment with detritic input bog

Total residual unsupported 210 Pb content A 0 Ždpm cmy2 . 11.1–95

6.7–15.5

0.54

Direct atmospheric flux Ž198–208S. Titicaca Lake, Bolivia C.R.S. low depth- 5 m

Unsupported Pb concentration at surface C 0 Ždpm gy1 .

210

5.2–81.5

28.3–31.2

210

Pb flux equivalent to 210 Pb residual Ždpm cmy2 yry1 .

Pb flux equivalent to 210 Pb at surface Ždpm cmy2 yry1 .

References

0.27–2.95

Appleby and Oldfield, 1983

0.88–0.97

Pourchet et al., 1995

7.8

0.25

6.1–7.4

0.19–0.23

12.1

210

Krishnaswami and Lal, 1978

0.34

0.66

Pourchet et al., 1994

2.3

Forsberg et al., 1989

Brazilian Amazon Floodplain lake, Jamari River, C.R.S.

9.9

85.4

2.5

2.66

AM-43 AM-73

2.48 4.33

160.84 120.15

18.63 12.84

5.01 3.74

this study this study

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

Mean concentration of supported 210 Pb Ždpm gy1 .

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

of variations on the accumulation of sediments in this type of lake. As indicated by higher concentrations of Hg and Al si. in surface sediments, erosion promotes an increase in Hg contamination in aquatic environments, but the effect of sedimentation on the texture of sediments dominates the apparent amplitude of this increase. 4.4. History of mercurial contamination in the lower Tapajos ´ Few studies have used 210 Pb or other such dating methods in the central Amazonian basin. Forsberg et al. Ž1989. employed the C.R.S. model to interpret the effect of colonization of a drainage basin on the sedimentation rates in Rondonia ˆ Žsouthern Amazon.. While they estimated the residual 210 Pb, they did not compare numerous cores, thereby greatly limiting the validity of the model used ŽBinford et al., 1993.. In the AM-43 and AM-73 cores, the accumulated residual 210 Pb activities of 120 and 160 dpm cmy2 are elevated compared to other studies and the observed atmospheric fluxes below those for low latitudes ŽTable 3.. The AM-43 sediments show the greater accumulated residual activity and correspond to greater sedimentation of eroded material in com-

261

parison to the AM-73 core, as defined by the physiography of the river ŽRoulet et al., 1998a.. The increase in 210 Pb can be explained by an increase in superficial erosion of soils containing appreciable quantities of excess 210 Pb ŽEakins et al., 1983; McCall et al., 1984.. For our case, the C.I.C. model is the most appropriate for surface sediments ŽEakins et al., 1983. but it necessitates an elimination of sedimentation zones where the activities increase. The efficient modeling of such a system implicates, however, an evaluation of the relative importance of inputs derived from the surface of soils as compared to those coming from the atmosphere, as well as the behaviour and fate of 210 Pb in the water column ŽOldfield and Appleby, 1984; He and Walling, 1996.. In view of the considerable incertitude in chronologies determined via theoretical modeling of 210 Pb fluxes, Oldfield and Appleby Ž1983; 1984. proposed an empirical approach in which the theoretical model is tested using geochemical, magnetic, palynological and textural factors. If the C.I.C. model is valid, the increase in Hg at the surface would, according to the AM-73 core, be dated at least 25 years Ži.e., since 1960–1970. and at more than 30 years Ži.e., before 1960. for the AM-43 cores. The AM-73 core is, however, irregular below the superfi-

Fig. 9. Population growth of the Santarem, and Cuiaba Municipalities. ´ Itaituba, AveirorRuropolis ´

262

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

cial increase and so it is difficult to determine if the profiles are comparable in terms of chronology. As a means of testing the chronology we can attempt to use historical or geographical indicators of colonization in the region. Fig. 9 shows the population growth in Santarem, ´ Itaituba and AveirorRuropolis ´ Municipalities in the lower Tapajos ´ region and another Amazonian Municipality, Cuiaba ŽMato Grosso.. The growth between 1940 and 1991 is perfectly exponential. The most important increase occurred since the beginning of the 1970s and corresponds to the heavy colonization in the region following development of the trans-amazonian highway ŽWalker et al., 1997.. The observed erosion typically corresponds to the effects of colonization accompanied by exponential population growth on the drainage basins ŽMcCall et al., 1984; Forsberg et al., 1989; De Boer, 1997.. The history of the region indicates that the increased transport and deposition of terrigenous Hg was significantly accelerated during the 1950s, 1960s and 1970s.

5. Conclusions Our sediment results indicate, for the first time, an excellent relationship between Hg and Al si. and other geochemical parameters associated with aluminosilicates. The relationship between Al si. and Hg over the length of, and between, all of the profiles for a particular physiographic zone raises numerous questions regarding interpretations documented for previous studies in the Amazon region which did not take into consideration mineral or organic matter geochemistry Že.g., Lacerda and Salomons, 1991; Lacerda et al., 1991; Pfeiffer et al., 1993; Hylander et al., 1994; Gomes et al., 1996; Filho and Maddock, 1997.. The use of independent geochemical indicators appears to be indispensable for verifying whether the increase of Hg levels in surface sediments is indeed the result of gold mining activities in the Amazon. The systematic use of Al si. concentrations opens an interesting route toward the normalization of Hg derived from the alteration andror erosion of terrestrial regions and measured in the sediments since, in one determination, it accounts for granulometry and the concentration of clays and oxyhydroxides.

The geochronology is imprecise and the geo-chronological indicators insufficient for determining, with certainty, the perturbation history observed in the lacustrine sediments of the Tapajos ´ basin. In all of the aquatic environments studied in the Tapajos, ´ the Arapiuns and the Amazon, a significant increase Ž1.2- to 25-fold depending on the type of sediment and sedimentation. in Hg concentrations in recent or superficial sediments is observed. This increase is significantly correlated with a change in the quality of deposited sediments and is related to a relative modification of the source of surface sediments. They are finer and richer in Hg, oxyhydroxides, aluminosilicates, and C than deeper sediments. The variations in concentrations of the different elements studied are similar in all of the lentic systems and suggest an overall enrichment of the matrix of recent sediments by fine clays. These observations provide geochemical evidence for the effects of the mechanical erosion of the surface soils of the Tapajos ´ Valley ŽRoulet et al., 1998b. on the increased transport Žduring the rainy season. of fine particles rich in Hg ŽRoulet et al., 1999b.. The origin of recently accumulated Hg demonstrates that continual colonization, deforestation and agricultural exploitation of virgin Amazonian forest are responsible for a major perturbation of the cycling of Hg in soils of the central Amazon. This perturbation, in turn, results in increased Hg contamination of lentic systems of the lower Tapajos. ´ The bioavailability of sediment-bound Hg would be expected to be low given its strong association with sediment particles. Due to the importance of MeHg in the trophic transfer and toxicity of Hg, inorganic Hg associated with sediment particles also needs to be methylated before ingestion. The bioavailability of MeHg to filter-feeders is generally greater than that for inorganic Hg. Gagnon and Fisher Ž1997. show that the adsorption of sedimentbound MeHg by mussels is rapid and efficient, even from uncoated inorganic particles with low nutritional value. They conclude that MeHg is assimilable from all major sedimentary components. Favourable methylation conditions were observed in the Tapajos ´ flood plain ŽGuimaraes ˜ et al., 1996; Roulet et al., 1999c.. We need now to evaluate the influence of erosion on nutrients ŽDorich et al., 1984. and organic matter availability and to consider its effects on the

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

production and food transfer of MeHg in the Tapajos ´ aquatic ecosystems.

Acknowledgements Financial support for this project was provided by Canada’s federal government through the International Development and Research Center ŽIDRC. and as part of the CARUSO project Žcontract a961052-01r001300-01 and scholarship to the first author.. Supplementary financial assistance, in the form of a doctoral scholarship, was provided to the first author by the Quebec ´ government via the Fond pour la formation de Chercheurs et l’Aide a` la Recherche ŽFCAR.. We would like to thank the Universidade Federal do Para´ ŽUFPa. and the UFPa campus in Santarem ´ for their assistance with financial, logistical and technical aspects. The project would never have seen the light of day without the collaboration of Aldo Queiroz Gomes, coordinator of the Santarem ´ campus and Cristovan Waverly Diniz, rector of UFPa. We would also like to thank Louise Cournoyer, Gerson da Costa da Oliveira, Jadson Dezincourt, Erinaldo de Jesus da Silva, Jean Lebel, Everaldo Portela, Isabelle Rheault, Sewbert Rodrigues Jati, Gil Serique, Delaine Sampaio da Silva, Persio Scavone de Andrade, Nicolas Soumis, Silmara Sousa de Morales, Carlos-Jose´ Sousa Pasos, Sophie Tran, and Ronilson Vasconcelos Barbosa for their help in the field and laboratory. The analytical methods used for Hg were developed by Professor P. Pichet for the laboratory of the Chaire de Recherche en Environnement H-QrCRSNGrUQAM and GEOTOP. Special thanks to Shelagh Montgomery for help in the english translation of the manuscript. [JD]

References Akagi, H., Malm, O., Branches, F.J.P., Kinjo, Y., Kashima, Y., Guimaraes, ˜ J.R.D., Oliveira, R.B., Haraguchi, K., Pfeiffer, W.C., Takizawa, Y., Kato, H., 1994. Human exposure to mercury due to goldmining in the Tapajos river basin, Amazon, Brazil: speciation of mercury in human hair, blood and urine. Water, Air, Soil Pollut. 80, 85–94.

263

Akagi, H., Malm, O., Kinjo, Y., Harada, M., Branches, F.J.P., Pfeiffer, W.C., Kato, H., 1995. Methylmercury pollution in Amazon, Brazil. Sci. Total Environ. 175, 85–95. Alexander, C.R., Smith, R.G., Calder, F.D., Schropp, S.J., Windom, H.L., 1993. The historical record of metal enrichment in two Florida estuaries. Estuaries 16, 627–637. Appleby, P.G., Oldfield, F., 1978. The calculation of lead-210 dates assuming a constant rate of supply of unsupported 210 Pb to the sediment. Catena 5, 1–8. Appleby, P.G., Oldfield, F., 1983. The assessment of 210 Pb data from sites with varying sediment accumulation rates. Hydrobiologia 103, 29–35. Asselman, N.E.M., Middelkoop, H., 1995. Floodplain sedimentation: quantities, patterns and processes. Earth Surf. Processes Landforms 20, 481–499. Bidone, E.D., Castilhos, Z.C., Cid de Souza, T.M., Lacerda, L.D., 1997. Fish contamination and human exposure to mercury in the Tapajos ´ river basin, Para´ state, Amazon, Brazil: a screening approach. Bull. Environ. Contam. Toxicol. 59, 194–201. Binford, M.W., 1983. Paleolimnology of the Peten Lake district, Guatemala: I. Erosion and deposition of inorganic sediment as inferred from granulometry. Hydrobiologia 103, 199–203. Binford, M.W., Kahl, J.S., Norton, S.A., 1993. Interpretation of 210 Pb profiles and verification of the CRS dating model in PIRLA project lake sediment cores. J. Paleolimnol. 9, 275– 296. Bloom, N., Fitzgerald, W.F., 1988. Determination of volatile mercury species at the picogram level by low-temperature gas chromatography with cold-atomic fluorescence detection. Anal. Chim. Acta 208, 151–161. Bowen, H.J.M., 1979. Environmental Chemistry of the Elements. Academic Press, London. Brown, A.G., 1983. An analysis of overbank deposit of a flood at Blandford Forum, Dorset, England. Rev. Geomorphol. Dyn. 32, 95–99. Bruland, K.W., Bertine, K., Koide, M., Windom, H., 1974. History of metal pollution in southern California coastal zone. Environ. Sci. Technol. 8, 425–432. Chauvel, A., Walker, I., Lucas, Y., 1996. Sedimentation and pedogenesis in a Central Amazonian Black water basin. Biogeochemistry 33, 77–95. Cleary, D., Thornton, I., Brown, N., Kazantzis, G., Delves, T., Worthinston, S., 1994. Mercury in Brazil. Nature 369, 613– 614. Courcelles, M., 1998. Enregistrement sedimentaire des flux recents ´ ´ de metaux lourds ŽPb, Hg. et d’isotopes a` courte periode ´ ´ 210 137 228 Ž Pb, Cs et Th. dans un lac sub-arctique a` faible ŽLac Jobert, Quebec vitesse de sedimentation ´ ´ .. PhD Thesis, Universite´ du Quebec a` Montreal, ´ ´ Canada. Daskalakis, K.D., O’Connor, T.P., 1995. Normalization and elemental sediment contamination in the coastal United States. Environ. Sci. Technol. 29, 470–477. De Boer, D.H., 1997. Changing contributions of suspended sediment sources in small basins resulting from European settlement on the Canadian Prairies. Earth Surf. Processes Landforms 22, 623–639. Dorich, R.A., Nelson, D.W., Sommers, L.E., 1984. Availability of

264

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

phosphorus to algae from eroded soil fractions. Agric. Ecosys. Environ. 11, 2543–2640. Droppo, I.G., Stone, M., 1994. In-channel surficial fine-grained sediment laminae: Part I. Physical characteristics and formational processes. Hydrol. Proc. 8, 101–111. Eakins, J.D., Cambray, R.S., Chambers, K.C., Lally, A.E., 1983. The transfer of natural and artificial radionuclides to brothers water from its catchment. In: Harworth, E.W., Lung, J.W.G. ŽEds.., Lake Sediments and Environment History. University of Minnesota Press, Minneapolis, pp. 125–144. Engstrom, D.R., Wright, H.E., 1983. Chemical stratigraphy of lake sediments as record of environmental change. In: Harworth, E.W., Lung, J.W.G. ŽEds.., Lake Sediments and Environment History. University of Minnesota Press, Minneapolis, pp. 11–67. Engstrom, D.R., Swain, E.B., Kingston, J.C., 1985. A palaeolimnological record of human disturbance from Harvey’s Lake, Vermont: geochemistry, pigments and diatoms. Freshwater Biol. 15, 261–288. Farella, N., Lucotte, M., Louchouarn, P., Roulet, M., Canuel, R., Tran, S., 1996. Use of lignin-derived phenols as tracers of soil erosion in Amazonian rivers. Proceedings of the 4th International Symposium on the Geochemistry of the Earth’s Surface, 22–28 July 1996, Ilkley, UK, pp. 406–408. Filho, S.R., Maddock, J.E.L., 1997. Mercury pollution in two gold mining areas of the Brazilian Amazon. J. Geochem. Explor. 58, 231–240. Flower, R.J., Dearing, J.A., Nawas, R., 1984. Sediment supply and accumulation in a small Moroccan lake: an historical perspective. Hydrobiologia 112, 81–92. Fontes, M.P.F., 1995. Comments on the effects of soil properties on the differential sorption by semiarid soils from northeast Brazil. Soil Sci. 159, 74–75. Forsberg, B., Godoy, J.M., Victoria, R., Martinelli, L.A., 1989. Development and erosion in the Brazilian Amazon: a geochronological case study. Geojournal 19, 402–405. Foster, I.D.L., Walling, D.E., 1994. Using reservoir deposit to reconstruct changing sediment yields and sources in the catchment of Old Mill Reservoir, South Devon, UK, over past 50 years. Hydrol. Sci. J. 39, 347–368. Gagnon, C., Fisher, N.S., 1997. Bio-availability of sediment-bound methyl and inorganic mercury to a marine bivalve. Environ. Sci. Technol. 31, 993–998. Gibbs, R.J., 1977. Transport phase of transition metals in the Amazon and Yukon rivers. Geol. Soc. Am. Bull. 88, 829–843. Goldberg, E.D., 1963. Geochronology with 210 Pb. Radioactive Dating. Int. Atom. Energy Ag., Vienna, pp. 121–131. Goldberg, E.D., Griffin, J.J., Hodge, V., Koide, M., Windom, H., 1979. Pollution history of the Savannah river Estuary. Environ. Sci. Technol. 13, 588–594. Gomes, M.R., Ayres, G.A., Lacerda, L.D., 1996. Atmospheric mercury deposition rates in Alta Floresta, southern Amazon. Book of Abstracts, 4th Int. Conf. Mercury as a Global Pollutant, Hamburg, Germany, 78 pp. Guimaraes, ˜ J.R.D., Roulet, M., Lucotte, M., 1996. Seasonal and spatial variations of Hg net methylation in a floodplain lake of the Tapajos ´ river, Brazil. Book of Abstracts, 4th Int. Conf. Mercury as a Global pollutant, Hamburg, Germany, 418 pp.

Hamilton, S.K., Lewis, W.M.J., 1990. Physical characteristics of the fringing floodplain of Orinoco river, Venezuela. Interciencia 15, 491–500. He, Q., Walling, D.E., 1996. Use of fallout Pb-210 measurements to investigate longer-term rates and patterns of overbank sediment deposition on the floodplains of lowland rivers. Earth Surf. Processes Landforms 21, 141–154. He, Q., Walling, D.E., 1997. Spatial variability of the particle size composition of overbank floodplain deposit. Water, Air, Soil Pollut. 99, 71–80. Horovitz, A., 1991. A Primer on Sediment-Trace Element Chemistry. Lewis Publishers, Chelsea. Hylander, L.D., Silva, E.C., Oliveira, L.J., Silva, S.A., Kuntze, E.K., Silva, D.X., 1994. Mercury levels in Alto Pantanal: a screening study. Ambio 23, 478–484. Irion, G., 1984. Sedimentation and sediments of Amazonian rivers and evolution of the Amazonian landscape since Pliocene times. In: Sioli, H. ŽEd.., The Amazon — Limnonology and Landscape Ecology of a Mighty Tropical River and its Basin. Dr. W. Junk Publishers, Dordrecht, pp. 201–214. Jonsson, A., 1997. Fe and Al sedimentation and their importance as carriers for P, N, and C in large humic lake in northern Sweden. Water, Air, Soil Pollut. 99, 283–295. Kehrig, H., Malm, O., Akagi, H., 1997. Methylmercury in hair samples from different riverine groups, Amazon, Brazil. Water, Air, Soil Pollut. 97, 17–29. Kohlhepp, G., 1984. Development planning and practices of economic exploitation in Amazonia. Recent trends in spatial organization of a tropical frontier region in Brazil Ž1966–1981.. In: Sioli, H. ŽEd.., The Amazon — Limnonology and Landscape Ecology of a Mighty Tropical River and its Basin. Dr. W. Junk Publishers, Dordrecht, pp. 649–674. Krishnaswami, S., Lal, D., 1978. Radionuclide limnochronology. In: Lerman, A. ŽEd.., Lakes Chemistry, Geology and Physics. Springer-Verlag, NY, pp. 153–177. Lacerda, L.D., Salomons, W., 1991. Mercury in the Amazon. A chemical time bomb? Dutch Ministry of Housing, Physical Planning and the Environment, Haren. Lacerda, L.D., Salomons, W., Pfeiffer, W.C., Bastos, W.R., 1991. Mercury distribution in sediment profiles from lakes of the high Pantanal, Mato Grosso State, Brazil. Biogeochemistry 14, 91–97. Lebel, J., Mergler, D., Lucotte, M., Amorim, M., Dolbec, J., Miranda, D., Arantes, G., Rheault, I., Pichet, P., 1996. Evidence of early nervous system dysfunction in Amazonian populations exposed to low-levels of methylmercury. Neurotoxicology 17, 157–168. Lebel, J., Roulet, M., Mergler, D., Lucotte, M., Laribe, F., 1997. Fish diet and mercury exposure in a riparian Amazonian population. Water, Air, Soil Pollut. 97, 31–44. Lebel, J., Mergler, D., Branches, F., Lucotte, M., Amorim, M., Larribe, F., Dolbec, J., 1998. Neurotoxic effects of low-level methylmercury contamination in the Amazonian basin. Environ. Res. 79, 20–32. Lucotte, M., d’Anglejan, B., 1985. A comparison of several methods for the determination of iron hydroxides and associated orthophosphates in estuarine particulate matter. Chem. Geol. 48, 257–264.

M. Roulet et al.r Chemical Geology 165 (2000) 243–266 Mackereth, F.J.H., 1966. Some chemical observations on post-glacial lake sediment. Philos. Trans. R. Soc. London, Ser. B 250, 165–213. Malm, O., Branches, F.J.P., Akagi, H., Castro, M.B., Pfeiffer, W.C., Harada, M., Bastos, W.R., Kato, H., 1995. Mercury and methylmercury in fish and human hair from the Tapajos ´ river basin, Brazil. Sci. Total Environ. 175, 141–150. Malm, O., Guimaraes, ˜ J.-R.D., Castro, M.B., Bastos, W.R., Viana, J.P., Branches, F.J.P., Silveira, E.G., Pfeiffer, W.C., 1997a. Follow-up of mercury levels in fish, human hair and urine in the Madeira and Tapajos ´ basins, Amazon, Brazil. Water, Air, Soil Pollut. 97, 45–51. Malm, O., Guimaraes, J.-R.D., Castro, M.B., Bastos, W.R., ˜ Branches, F.J.P., Pfeiffer, W.C., 1997b. Mercurio na ´ Amazonia: evoluc¸ao ˆ ˜ da contaminac¸ao ˜ ambiental e humana. Ciencia Hoje 22, 16–23. Marriott, S., 1992. Textural analysis and modeling of a flood deposit: river Severn, UK. Earth Surf. Processes Landforms 17, 687–697. McCall, P.L., Robbins, J.A., Matisoff, G., 1984. 137Cs and 210 Pb transport and geochronologies in urbanized reservoirs with rapidly increasing sedimentation rates. Chem. Geol. 44, 33–65. Milligan, T.G., Loring, D.H., 1997. The effect of flocculation on the size distributions of bottom sediment in Coastal Inlets: implications for contaminant transport. Water, Air, Soil Pollut. 99, 33–42. Morse, J.W., Presley, B.J., Taylor, R.J., Benoit, G., Santschi, P.H., 1993. Trace metal chemistry of Galveston Bay: water, sediments and biota. Mar. Environ. Res. 36, 1–37. Oldfield, F., Appleby, P.G., 1983. Empirical testing of 210 Pb-dating models for lake sediments. In: Harworth, E.W., Lung, J.W.G. ŽEds.., Lake Sediments and Environment History. University of Minnesota Press, Minneapolis, pp. 91–124. Oldfield, F., Appleby, P.G., 1984. A combined radiometric and mineral magnetic approach to recent geochronology in lake affected by catchment disturbance and sediment redistribution. Chem. Geol. 44, 67–83. Parfitt, R.L., Childs, C.W., 1988. Estimation of forms of Fe and Al: a review, and analysis of contrasting soils by dissolution and Mossbauer methods. Aust. J. Soil Res. 26, 121–144. Pfeiffer, W.C., Lacerda, L.D., Salomons, W., Malm, O., 1993. Environmental fate of mercury from gold mining in the Brazilian Amazon. Environ. Rev. 1, 26–37. Pichet, P., Morrison, K., Rheault, I., Tremblay, A., 1999. Analysis of total mercury and methylmercury in environmental samples. In: Lucotte, M., Schetagne R., Therien, N., Langlois, C., ´ Tremblay, A. ŽEds.., Mercury in the Biogeochemical Cycle. Springer, Berlin, pp. 41–52. Pizzuto, J.E., 1987. Sediment diffusion during overbank flows. Sedimentology 34, 301–317. Pourchet, M., Mourguiart, P., Pinglot, J.F., Preiss, N., Argollo, J., Wirrmann, D., 1994. Sedimentation recente dans le lac Titi´ ´ caca ŽBolivie.. C. R. Acad. Sci. Paris 319, 535–541, Serie ´ II. Pourchet, M., Mourguiart, P., Pinglot, J.F., Preiss, N., Argollo, J., ´ Wirrmann, D., 1995. Evaluation des vitesses de sedimentation ´ recente dans les hautes vallees ´ ´ des Andes boliviennes. Son interet atmospherique. ´ ˆ dans l’estimation des paleo-pollutions ´ ´ C. R. Acad. Sci. Paris 320, 477–482, Serie ´ IIa.

265

Putzer, H., 1984. The geological evolution of the Amazon basin and its mineral resources. In: Sioli, H. ŽEd.., The Amazon — Limnonology and Landscape Ecology of a Mighty Tropical River and its Basin. Dr. W. Junk Publishers, Dordrecht, pp. 15–46. Ravichandran, M., Baskaran, M., Santschi, P.H., Bianchi, T.S., 1995. History of trace metal pollution in Sabine-Neches Estuary, Beaumont, TX. Environ. Sci. Technol. 29, 1495–1503. Robbins, J.A., 1978. Geochemical and geophysical application of radio lead. In: Nriagu, J.O. ŽEd.., Biogeochemistry of Lead in the Environment. Elsevier, Holland, pp. 285–393. Ross, G.F., Wang, C., 1993. Extractable Al, Fe, Mn, and Si. In: Carter, M.R. ŽEd.., Soil Sampling and Methods. Lewis Publishers, Chelsea, pp. 239–246. Roulet, M., Lucotte, M., Rheault, I., Tran, S., Farella, N., Canuel, R., Mergler, D., Amorim, M., 1996. Mercury in Amazonian soils: accumulation and release. Proceedings of the 4th International Symposium on the Geochemistry of the Earth’s Surface, 22–28 July 1996, Ilkley, UK, pp. 453–457. Roulet, M., Lucotte, M., Canuel, R., Rheault, I., Tran, S., De Freitos Gogh, Y.G., Farella, N., Souza do Valle, R., Sousa Passos, C.J., De Jesus da Silva, E., Mergler, D., Amorim, M., 1998a. Distribution and partition of total mercury in waters of the Tapajos ´ river basin, Brazilian Amazon. Sci. Total Environ. 213, 203–211. Roulet, M., Lucotte, M., Saint-Aubin, A., Tran, S., Rheault, I., ´ Farella, N., De Jesus da Silva, E., Dezencourt, J., Sousa Passos, C.J., Santos Soares, G., Guimaraes, ˜ J.R.D., Mergler, D., Amorim, M., 1998b. The geochemistry of Hg in central Amazonian soils developed on the Alter-do-Chao ˜ formation of the lower Tapajos ´ river valley, Para´ State, Brazil. Sci. Total Environ. 223, 1–24. Roulet, M., Lucotte, M., Farella, N., Serique, G., Coelho, H., Sousa Passos, C.J., de Jesus da Silva, E., Scavone de Andrade, P., Mergler, D., Guimaraes, ˜ J.-R.D., Amorim, M., 1999a. Effects of recent human colonization on the presence of mercury in Amazonian ecosystems. Water, Air, Soil Pollut. 112, 297–313. Roulet, M., Lucotte, M., Canuel, R., Farella, De Freitos Goch, Y.G., Pacheco Peleja, J.R., Guimaraes, ˜ J.-R.D., Mergler, D., Amorim, M., 1999b. Spatio-temporal geochemistry of Hg in waters of the Tapajos ´ and Amazon rivers, Brazil. Limnol. Oceanogr., submitted. Roulet, M., Guimaraes, ˜ J.R.D., Lucotte, M., 1999c. Methylmercury production and accumulation in sediments and soils of an Amazonian floodplain — effect of seasonal inundation. Water, Air, Soil Pollut., submitted. Salati, E., Marques, J., 1984. Climatology of the Amazon region. In: Sioli, H. ŽEd.., The Amazon, Limnonology and Landscape Ecology of a Mighty Tropical River and its Basin. Dr. W. Junk Publishers, Dordrecht, pp. 85–126. Salomons, W., Forstner, U., 1984. Metals in the Hydrocycle. ¨ Springer-Verlag, New York. Schmidt, G.W., 1982. Primary production of phytoplankton in the three types of Amazonian waters: V. Some investigations on the phytoplankton and its primary productivity in the clear water of the lower Rio Tapajos ´ ŽPara, Brazil.. Amazoniana 7, 335–348.

266

M. Roulet et al.r Chemical Geology 165 (2000) 243–266

Schropp, S.J., Lewis, F.g., Windom, H.L., Ryan, J.D., Calder, F.D., Burneys, L.C., 1990. Interpretation of metal concentrations in estuarine sediments of Florida using aluminium as reference element. Estuaries 3, 227–235. Shine, J.P., Ika, R.V., Ford, T.E., 1995. Multivariate statistical examination of spatial and temporal patterns of heavy metal contamination in New Bedford marine sediments. Environ. Sci. Technol. 29, 1781–1788. Sinex, S.A., Wright, D.A., 1988. Distribution of trace metals in the sediments and biota of Chesapeake Bay. Mar. Pollut. Bull. 19, 425–431. Sioli, H., 1984. The Amazon and its main affluents: hydrography, morphology of the river courses, and river types. In: Sioli, H. ŽEd.., The Amazon — Limnology and Landscape Ecology of a Mighty Tropical River and its Basin. Dr. W. Junk Publishers, Dordrecht, pp. 127–165. Sippel, S., Hamilton, S.K., Melack, J.M., 1992. Inundation area and morphometry of lakes on the Amazon river floodplain, Brazil. Arch. Hydrobiol. 123, 385–400. Smith, J.N., Walton, A., 1980. Sediment accumulation rates and

geochronologies measured in the Saguenay Fjord using the Pb-210 dating method. Geochim. Cosmochim. Acta 44, 225– 240. Trefry, J.H., Metz, S., Trocine, R.P., Nelson, T.A., 1985. A decline in lead transport by the Mississippi river. Science 230, 439–441. Walker, R.T., Homma, A.K.O., Conto, A.J., Carvalho, R.A., Ferreira, C.A.P., Santos, A.I.M., Rocha, A.C.P.N., Oliveira, P.M., Pedraza, C.D.R., 1997. As contradic¸oes ˜ do processo de desenvolvimento agrıcola na Transamazonica. Embrapa, Doc´ ˆ umentos, no. 93. Walling, D.E., He, Q., Nicholas, A.P., 1996. Floodplains as suspended sediment sinks. In: Anderson, M., Walling, D.E., Bates, P. ŽEds.., Floodplain Processes. Wiley, Chichester, UK, pp. 399–440. Windom, H.L., Schropp, S.J., Calder, F.D., Ryan, J.D., Smith, R.G., Burney, L.C., Lewis, F.G., Rawlinson, C.H., 1989. Natural trace metal concentrations in estuarine and coastal marine sediments of the southeastern United States. Environ. Sci. Technol. 23, 314–320.