Distribution and Partition of Trace Metals in the Amazon Basin

HYDROLOGICAL PROCESSES Hydrol. Process. 17, 1345–1361 (2003) Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/hyp.1288

Distribution and partition of trace metals in the Amazon basin

Patrick T. Seyler1* and Geraldo R. Boaventura2 1 Institut de Recherche pour le Développement (IRD), Laboratoire des Mécanismes de Transfert en Géologie (LMTG), Université Paul Sabatier, 38 rue des trente-six ponts, 31400 Toulouse, France 2 Instituto de Geociências, Universidade de Bras´ılia, CEP 70910-900 Bras´ılia, DF, Brazil

Abstract: The distribution of trace metals (V, Cr, Mn, Co, Cu, Zn, As, Rb, Sr, Mo, Cd, Sb, Cs, Ba, U) was investigated in surface waters and associated particulates in the Amazon mainstream (Solimoes˜ and Amazon rivers). Dissolved V, Cu, As, Sr, Ba, U correlate with major ions and appear to be predominantly derived from soluble rocks occurring in the Amazon upper basin. These elements appear conservative in waters and are progressively diluted by less-concentrated waters coming from the lowland and shield areas. A monthly time series obtained at the Obidos´ gauging station shows that temporal variability of trace element concentrations reflects the source, remobilization and/or biological processes occurring in the channel or in the surrounding floodplain lakes. The trace element concentrations in the particulate matter show a clear relationship with the location of the samples. V, Co, Cr, Mn, Sr, Cs and Ba concentrations are higher in the Solimoes˜ and the Rio Negro is enriched in Fe, Al and Zn. In the Rio Solimoes,˜ V, Cr, Mn, Co, Ni, Zn, Cs and Pb are almost entirely carried by the river particulate matter; Cu, Rb, Sr, Ba and U are transported mainly by the suspended particles, but a dissolved phase contributes to the transport. In the Rio Negro, the proportion of elements transported by the dissolved phase is higher for the whole set of elements. The implications of these results allow us to compute the fluxes from the Amazon River to the Atlantic Ocean. Copyright  2003 John Wiley & Sons, Ltd.

KEY WORDS Amazon basin; trace metal geochemistry; trace metal ﬂuxes

INTRODUCTION Trace metallic pollutants, such as heavy metals (Cd, Pb, Hg, Cr, Zn, Cu, etc.) are a major issue for the quality of continental waters and of great concern in terms of pressure on both ecosystems and human health. In contrast to organic pollutants, metals do not degrade or eliminate; their physical and chemical forms change and these elements are readily remobilized into the environment by natural transformation mechanisms. An understanding of the biogeochemical processes involved in the transfer, accumulation and mobility of trace metals in pristine environments is required in evaluating the capacity of receiving waters to accommodate wastes without detrimental effects. The Amazon River system, which covers 6 ð 106 km2 and supplies up to 20% of all the river water discharged to the ocean (Molinier et al., 1997) is relatively free of industrial and agricultural interference. Due to its magnitude, the Amazon River is a prominent link in trace element cycles and quantifying its exportation ﬂuxes to the ocean is of considerable importance. A limited number of papers deal with the distribution and partition of trace metals in the Amazon River. Martin and Meybeck (1979) and Martin and Gordeev (1986) gave a global tabulation of trace metal concentrations in particulate matter of major rivers including the Amazon, and Palmer and Edmond (1992)

* Correspondence to: Patrick T. Seyler, Institut de Recherch pour le Developpement,´ Laboratoire des Mecanismes´ de Transfert en Geologie´ (LMTG), Universite´ Paul Sabatier, 38 rue des trente-six ponts, 31400 Toulouse, France. E-mail: [email protected] Received 5 January 2002 Copyright  2003 John Wiley & Sons, Ltd. Accepted 22 July 2002 1346 P. T. SEYLER AND G. R. BOAVENTURA measured dissolved Fe, Al and Sr concentrations in the Amazon mainstream and a number of its tributaries. Konhauser et al. (1994) reported the trace and rare earth elemental composition of sediments, soils and waters, mainly in the region of Manaus. Gaillardet et al. (1997) published an extensive data set for major and trace concentrations in the Amazon region extending between Manaus and Santarem. More recently, Seyler et al. (1999), Elbaz-Poulichet et al. (1999) and Seyler and Boaventura (2001) presented a comprehensive survey of trace metal forms in the Madeira and Amazon river basins. Our aim in the present work is to gain new insights into weathering and transport processes controlling the fate and ﬂux of trace metals (V, Cr, Mn, Co, Cu, Zn, As, Rb, Sr, Mo, Cd, Sb, Cs, Ba, U) of the major streams in the Amazonian basin.

MAJOR ENVIRONMENTAL FEATURES OF THE AMAZON BASIN The drainage basin of the Amazon River is characterized by a high diversity of geological formations. Three major geological provinces can be distinguished: (i) the Guyana and Brazilian shields (44% of the basin area) with metamorphic and crystalline rocks; (ii) the Andean cordillera (11% of the basin area) which consists of the Precambrian basement, formed by sediments, igneous and metamorphic rocks, overlain by Paleozoic and Mesozoic red clays and dark shales, and by carbonates and evaporites; (iii) the Amazon trough (45% of the basin area) ﬁlled with a massive layer of ﬂuviolacustrine sediments ranging in age from Paleozoic to Pleistocene (Figure 1). The average air temperature in the Amazon is rather uniform, ranging from 24 to 26 °Cinthemajorpartof the basin, and the mean annual precipitation for the basin is 2400 mm. The Amazon climatology was described extensively by Marengo and Nobre (2001). The whole Amazon basin is covered by tropical rainforest (71%) and savannas (29%; Sioli, 1984). The soils of the Amazon basin belong mainly to the red ferralitic soil family. Their mineralogy is dominated by quartz, Al and Fe oxides and kaolinite, with a few accessory minerals such as anatase and zircon (Bernoux et al., 2001 and references cited therein). Compared to crustal abundances, these soils are more siliceous and

Figure 1. Major geomorphological features of the Amazon River basin

Copyright  2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1345–1361 (2003) TRACE METALS IN THE AMAZON BASIN 1347 aluminous with considerably lower levels of major cations. Numerous podzol zones exist on the central plain (Konhauser et al., 1994). The major tributaries in the upper Amazon basin have their sources in the Andes (Solimoes,˜ I¸ca, Japura and Madeira rivers), in the sub-Andean trough (Jurua´ and Purus rivers) or in the Guyana shield (Negro and Trombetas rivers). The main tributaries of the lower course, the Tapajos´ and Xingu´ rivers, drain the Brazilian shield (Figure 1). During the period 1965–1990, the mean annual water discharge of Solimoes˜ at Manacapuru (confluence with the Rio Negro) was estimated at 103 000 m3 s1, the discharge of the Negro River at 28 000 m3 s1 and the discharge of the Madeira at 31 200 m3 s1 (Molinier et al., 1997). At Obidos,´ which is the ultimate gauging station on the Amazon river upstream of the marine influence, the mean discharge was estimated as 168 000 m3 s1. The proportion of water originated from the Solimoes,˜ Negro and Madeira rivers accounts for 95% of the total discharge at Obidos´ and varies with total discharge during the annual cycle (Guyot et al., 1999). Regarding the whole basin, up to 209 000 m3 s1 is discharging to the Atlantic Ocean (Molinier et al., 1997). The low inter-annual variability of rainfall, the size of the basin, the lag between tributary inflows and the storage capacity of the extensive floodplain (varzeas)´ are responsible for the low inter-annual variability of the Amazon hydrograph. Concerning sediment transport, the more recent results obtained by Filizola et al. (1999) give a mean annual discharge of suspended sediment close to 770 ð 106 tatObidos,´ where 97% is due to the contribution of Andean tributaries (62% from the Solimoes˜ and 35% from the Madeira). The contributions of the Negro, Trombetas, Tapajos´ and Xingu´ account for less than 3%. An examination of the suspended sediment concentrations (SSM) and discharge versus time at Obidos´ indicates that plots of the relations between sediment discharge and water discharge will form loops rather than a straight line. During the hydrological period, SSM concentrations show high frequency variations (10 days) and the sediment peak discharge precedes by three months the maximum water discharge (Guyot et al., 1999).

SAMPLING AND METHODS The Solimoes,˜ as the Amazon mainstream is called between the Colombia and Brazil boundary and its confluence with the Rio Negro and the Amazon, downstream, were sampled on board R/V C. Dario, in October/November 1995 and in April/May 1997, corresponding to high water stage (HW) and low water stage (LW), respectively. Moreover, a monthly time series covering a whole hydrological cycle was obtained at the Obidos´ gauging station during the 1997 year. The sampling locations (Figure 2) were mostly the DNAEE/ANEEL stations, where water discharge was measured systematically with an acoustic Doppler current profiler (ADCP-RDI, 300 kHz) with a precision of over š5%. Suspended sediment concentrations were determined at the same time by filtering 5 l of river water, collected on three or five vertical profiles (depending on the section width) across the river section. In order to avoid contamination of the main vessel, water samples for major and trace element determination were taken from a small boat upwind the vessel, at approximately 0Ð5 m depth in the middle of the river section. Samples were collected in acid-washed polyethylene containers. The samples were filtered immediately through 0Ð22-µm Millipore filters, under a laminar air flow bench in the laboratory of the research vessel according to the procedure described elsewhere (Seyler and Elbaz-Poulichet, 1996). Conductivity, pH and alkalinity were measured on board, using routing conventional techniques. Major cations and anions were analysed at the IRD laboratory by ion chromatography (HPLC, Dionex 4000i) and 1 2C C 2C SiO2 by plasma spectrophotometry (ICP-AES). Detection limits were 0Ð01 mmol l for Ca ,K ,Mg , C 1 2 Na ,SiO2 and 0Ð005 mmol l for SO4 and Cl . Trace elements analysis was performed by ICP-MS using a VG-Elemental PQ2 instrument (Geochemical Laboratory, Montpellier University, France). Concentrations were established by calibrating peak intensity

10 Santarem 15 20 8 Manaus 5 6 12 16 4 79 11 14 18 21 2 Tabatinga 17 1 3 13

Figure 2. Location of the sampling points. Numbers correspond to the station names reported in Tables I–III acquired in scan mode with multi-elemental standard solutions and using indium as internal standard. The details of analytical procedures are given in Seyler and Elbaz-Poulichet (1996). Prior to analysis, the suspended particulate matter filters were dried at 105 °C and digested in sealed Teflon beakers with HNO3 –HF–HClO4 ultrapure acids, according to standard procedures. The accuracy and precision of the analytical methods have been established using appropriate certified reference materials (Elbaz-Poulichet et al., 1999).

RESULTS AND DISCUSSION Dissolved material transport Early on (Sioli, 1984 and references cited therein), the Amazon basin waters were recognized from visually observable features and classified into whitewaters, clearwaters and blackwaters. Whitewaters are yellow coloured due to high sediment content, as for example the Solimoes˜ and Madeira river waters which flow from the Andes cordillera. Blackwaters are tea coloured due to a high content of dissolved organic compounds and generally present very low concentrations in suspended solids, a prime example being the Rio Negro which drains the central basin typically covered by the evergreen rainforest. Clearwaters, such as the Tapajos´ and Trombetas, have high transparencies due to low concentrations of suspended sediment and drain the highly weathered shield areas. Although this classification was based on visual characteristics of the waters, subsequent studies based on analysis of dissolved and particulate constituents (Stallard, 1980; Stallard and Edmond, 1983, 1987; Richey et al., 1989) have shown that the three water types have different biochemical signatures. Our data on major dissolved element concentrations in the Amazon River mainstream is in good agreement with that already published and we do not discuss in detail the geochemical classification of the Amazon basin rivers. The major constituent data (Table I) allows us to distinguish the three groups of river waters previously mentioned. The first type has a low pH (<6Ð25) and low alkalinity (<0Ð25 meq l1) and is represented by the Negro basin rivers and the lower Amazon tributaries (Tapajos,´ Trombetas and Xingu´ rivers). The second group of rivers is constituted by the right-bank Solimoes˜ tributaries (Javari, Itaquai, Jutai) which have pH values higher than 5Ð5 and alkalinities lower than 0Ð25 meq l1. The other river group is dominated by carbonate ion (Alk >0Ð2meql1 and pH > 6Ð25) and is represented by the Amazon and Madeira rivers, the Solimoes˜ and its left-bank tributaries, I¸ca and Japura (Figure 3). The trace elements measured in the dissolved phase areV,Cr,Mn,Co,Ni,Cu,Zn,Rb,Sr,Mo,Cd,Sb,Ba, U and are given in Table II. Previous dissolved concentration values have been reported by Furch (1984) for the restricted region of Manaus, by Moore and Edmond (1984) for Ba concentrations in the whole basin, by

Copyright  2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1345–1361 (2003) TRACE METALS IN THE AMAZON BASIN 1349 ) 1 4 003 019 013 023 049 014 002 018 008 027 025 004 005 039 002 007 058 005 005 065 003 050 009 114 000 077 011 062 007 029 093 012 034 009 092 009 010 107 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð SO (mmol l ) 1 012 0 036 0 026 0 041 0 015 0 041 0 008 0 044 0 008 0 070 0 075 0 009 0 009 0 106 0 008 0 012 0 158 0 009 0 007 0 176 0 095 0 210 0 106 0 093 0 012 0 382 0 032 0 245 0 033 0 055 0 357 0 040 0 125 0 027 0 410 0 008 0 008 0 434 0 Cl Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð (mmol l ) 1 037 0 082 0 056 0 089 0 082 0 075 0 005 0 081 0 034 0 112 0 113 0 016 0 053 0 149 0 015 0 029 0 202 0 025 0 022 0 223 0 023 0 199 0 062 0 161 0 024 0 357 0 134 0 274 0 056 0 391 0 391 0 039 0 168 0 055 0 435 0 037 0 027 0 500 0 Na Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð (mmol l ) 1 022 0 022 0 018 0 024 0 036 0 017 0 003 0 022 0 020 0 022 0 021 0 008 0 023 0 025 0 012 0 011 0 032 0 015 0 012 0 031 0 020 0 025 0 029 0 039 0 013 0 029 0 036 0 028 0 015 0 044 0 031 0 020 0 023 0 020 0 036 0 027 0 019 0 037 0 K Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð (mmol l ) 1 023 0 039 0 029 0 046 0 069 0 032 0 002 0 039 0 022 0 050 0 049 0 008 0 034 0 062 0 007 0 013 0 081 0 015 0 015 0 087 0 024 0 064 0 012 0 133 0 006 0 073 0 053 0 075 0 017 0 104 0 098 0 006 0 042 0 017 0 103 0 012 0 012 0 115 0 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð centrations in the rivers of the Amazon basin Mg (mmol l ) 1 034 0 127 0 085 0 148 0 109 0 133 0 002 0 160 0 055 0 226 0 223 0 017 0 124 0 300 0 009 0 027 0 431 0 043 0 051 0 381 0 024 0 218 0 017 0 224 0 007 0 262 0 117 0 264 0 044 0 354 0 312 0 007 0 138 0 034 0 392 0 035 0 036 0 404 0 Ca Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð (mmol l ) 1 185 0 364 0 249 0 415 0 374 0 352 0 456 0 198 0 598 0 592 0 085 0 398 0 813 0 048 0 048 0 100 0 147 0 161 0 770 0 097 0 551 0 067 0 443 nd nd nd nd nd nd 669 0 000 0 760 0 530 0 750 0 220 0 350 0 860 0 090 0 450 0 190 0 030 0 130 0 110 0 005 0 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð (mmol l 23 0 38 0 31 0 34 0 32 0 18 0 29 nd 0 38 0 88 0 50 0 56 0 67 0 29 0 80 0 95 0 23 0 79 1 13 0 09 0 88 0 10 0 90 0 30 0 40 0 00 0 90 0 30 0 32 0 44 0 07 0 70 1 34 0 68 0 02 0 67 0 31 1 64 0 82 0 07 1 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð pH Alk. ) 1 16 36 33 96 96 24 86 65 36 56 75 16 96 94 56 86 66 76 16 06 06 05 07 07 05 07 07 07 07 07 07 06 07 06 07 06 06 07 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Cond. Scm µ ( ) 1 s meters (conductivity, pH) and major element con 3 (m ca 4 28/10/95 24 251 212 ca 6 10/29/1995 32 539 78 ˜ ˜ ´ ´ ao 21 31/05/97 13 120 15 ao 21 11/18/1995 6 027 15 a 9 03/05/97 10 120 40 a 9 11/4/1995 1 045 147 ´ a jusante 13 11/9/1995 2 534 68 ´ ua 12 08/05/97 99 200 72 ˜ ˜ ao Paulo de Olivenca 4 28/04/97 53 400 132 ao Paulo de Oliven¸ Table I. Water discharge, physicochemical para ˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ oes Manacapuru 14 12/05/97 133 880 52 oes Itape oes Tefe 11 07/05/97 90 800 71 oes Fonte Boa 8 03/05/97 71 810 95 oes S oes Tabatinga 1 26/04/97 51 830 156 oes Manacapuru 14 06/11/95 52 477 123 oes Itapeua 12 05/11/95 46 847 124 oes Fonte Boa 8 04/11/95 34 333 154 oes Santo Antonio do I¸ oes S oes Tabatinga 1 26/10/95 20 115 289 ´ ´ os Alter do Ch ´ os Alter do Ch a Jacitara 10 11/4/1995 10 264 29 ´ ´ a Foz do Juru a Foz do Juru ca Ipiranga 5 30/04/97 7 740 17 ca Ipiranga 5 10/31/1995 5 354 29 Tapaj Amazonas Obidos 20 28/05/98 204 490 43 Trombetas Oriximina 19 26/05/97 9 690 26 Amazonas Itacoatiara 18 21/05/97 199 950 50 Madeira Foz 17 22/05/97 49 780 49 6 Amazonas Jatuarana 16 21/05/97 147 130 42 Negro Paricatuba 15 18/05/97 34 360 13 Solim Purus Aruma Jusante 13 10/05/97 25 290 21 Solim Solim Japura Jacitara 10 04/05/97 9 810 8 Juru Solim Jutai Porto Antunes 7 02/05/97 4 130 6 I¸ Solim Itaquai Foz do Javari 3 26/04/97 1 660 16 Javari Foz do Itaquai 2 26/04/97 2 220 18 Solim Tapaj Amazonas Obidos 20 08/11/95 81 090 91 Trombetas Oriximina 19 11/16/1995 1 258 17 Amazonas Itacoatiara 18 11/15/1995 75 017 99 Madeira Foz 17 11/15/1995 5 132 103 Negro Paricatuba 15 13/11/95 12 311 15 Solim Purus Arum Solim Japur Juru Solim Jutai Porto Antunes 7 11/3/1995 1 143 27 Solim I¸ Solim Itaquai Foz do Javari 3 10/27/1995 793 26 Javari Foz do Itaquai 2 10/27/1995 1565 22 Sample riverSolim Station No. Date Discharge nd: not determined.

Copyright  2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1345–1361 (2003) 1350 P. T. SEYLER AND G. R. BOAVENTURA ) 1 092 215 150 210 148 171 097 197 145 191 199 124 122 247 062 109 343 182 186 356 164 050 134 130 050 130 155 113 155 101 252 567 088 122 424 071 223 244 370 U Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð l (nmol 10 20 50 20 00 40 00 00 30 10 50 40 80 90 90 10 90 90 00 ) Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð 1 Ba l (nmol ) 1 08 147 0 50 154 0 31 113 0 56 165 0 97 175 0 31 128 0 10 46 0 54 181 0 21 151 0 64 198 0 57 175 0 09 46 0 17 232 0 95 237 0 06 61 0 15 59 0 23 272 0 18 113 0 19 87 0 33 276 0 26 196 05 125 06 111 12 205 52 236 02 45 07 242 37 240 38 232 22 67 69 257 63 409 07 38 69 144 55 283 17 80 39 93 28 91 60 303 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Sb l (nmol ) 1 344 0 594 0 348 0 588 0 910 0 935 0 235 0 503 0 204 0 227 0 928 0 966 0 502 0 795 0 543 0 607 0 791 1 752 0 554 0 559 1 143 1 098 0 420 0 125 2 170 4 125 0 152 1 018 0 134 1 054 0 152 1 089 0 045 0 098 0 045 0 071 1 116 0 143 0 080 1 Cd Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð l (nmol ) 1 58 1 82 1 35 1 91 1 24 1 30 0 86 2 95 1 90 1 60 1 96 0 47 0 99 1 44 0 24 0 18 0 53 0 77 0 71 0 81 0 88 0 36 0 46 0 47 0 04 0 11 0 67 0 34 0 33 0 95 0 33 0 69 0 14 0 27 0 60 0 28 0 90 0 54 0 81 0 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Mo l (nmol ) 90 71 31 01 72 81 40 51 30 62 71 90 20 13 00 11 74 10 30 34 82 30 00 43 84 30 63 92 33 50 94 33 00 62 61 46 60 20 74 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð 1 Sr l (nmol ) 1 86 106 03 298 04 195 37 339 05 261 24 290 99 31 37 403 49 133 35 544 39 546 27 63 61 233 70 782 00 31 18 109 56 961 70 78 69 76 19 1028 32 561 93 76 55 110 94 613 32 620 60 49 74 766 17 783 19 171 15 989 62 869 88 39 69 468 87 174 24 1276 95 92 62 88 60 1444 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Rb l (nmol ) 1 49 31 86 20 42 24 55 18 78 20 81 18 54 13 15 21 64 24 68 20 96 18 83 12 76 28 45 20 39 22 69 14 76 19 34 23 82 20 87 18 65 23 74 21 54 45 11 20 63 22 23 20 21 20 83 20 53 21 97 34 80 27 38 17 06 17 36 21 85 25 20 21 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Zn l (nmol ) 1 9154 6 84 65 25 25 20 92 148 04 8 07 10 40 2 52 5 55 1 21 2 98 4 46 6 31 0 37 16 83 21 51 0 53 19 84 9 08 5 03 8 84 12 36 55 47 9 30 18 13 15 39 11 06 nd nd 385 81 5 92 nd 15 14 19 94 2 45 11 39 5 31 4 19 15 14 nd 33 00 17 59 32 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Cu l (nmol ) 1 1783 22 3 05 11 71 25 22 29 78 15 08 3 23 23 85 15 90 23 12 22 62 9 80 10 51 26 31 3 05 13 38 26 58 14 67 14 68 28 80 3 97 19 93 13 63 3 51 22 05 21 27 9 71 26 86 14 97 5 02 18 85 13 61 21 15 22 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ni l (nmol ) 1 8767 20 11 04 16 57 10 69 15 25 12 79 19 85 19 59 22 97 15 55 13 50 6 16 21 08 16 52 10 12 8 69 17 20 13 15 12 47 12 22 nd 2 64 nd 18 93 2 86 1 79 1 16 2 95 3 81 nd 11 02 3 44 4 09 2 84 2 86 3 47 7 87 2 11 nd 23 91 10 02 nd 25 Co Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð l (nmol ) 6915 1 0 48 1 61 1 27 1 87 1 83 2 39 1 18 2 90 1 16 1 38 1 42 2 10 2 00 6 48 1 21 1 66 1 95 1 46 1 70 86 0 40 10 80 91 46 0 70 77 1 30 0nd6 06 1 40 00 50 70 0 12 52 22 1 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð 1 Mn l (nmol ) 7887 451 37 13 85 80 443 68 360 94 232 01 193 86 291 36 251 20 292 50 267 62 202 44 239 58 373 06 618 08 191 54 308 41 165 16 115 77 258 81 6 12 53 15 112 85 85 42 119 69 135 48 101 85 33 69 123 85 65 35 156 31 105 88 182 81 93 1 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Cr l (nmol ) 7986 13 13 64 14 92 12 40 15 36 11 54 17 54 12 06 14 38 11 65 9 55 11 41 13 68 10 82 13 02 12 05 9 56 14 05 12 92 7 73 2 12 4 80 5 24 8 08 4 16 9 80 5 61 6 33 6 47 nd 76 29 3 77 6 57 6 92 14 10 nd 99 11 16 51 nd 445 55 nd 474 86 nd 114 1 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð l (nmol Table II. Trace element dissolved concentrations in the rivers of the Amazon basin ca 4 44 ca 6 23 ˜ ´ ˜ ´ ao 21 6 a910 ao 21 6 a920 ´ a jusante 13 12 ´ ua 12 17 ˜ ˜ ao Paulo de Olivenca 4 28 ao Paulo de Oliven¸ ˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ oes Manacapuru 14 15 oes Itape oes Tefe 11 16 oes Fonte Boa 8 22 oes S oes Tabatinga 1 29 oes Manacapuru 14 26 oes Itapeua 12 30 oes Fonte Boa 8 34 oes Santo Antonio do I¸ oes S oes Tabatinga 1 46 ´ ´ os Alter do Ch ´ os Alter do Ch a Jacitara 10 11 ´ ´ a Foz do Juru a Foz do Juru ca Ipiranga 5 12 ca Ipiranga 5 14 Tapaj Amazonas Obidos 20 13 Trombetas Oriximina 19 9 Amazonas Itacoatiara 18 12 Madeira Foz 17 8 Amazonas Jatuarana 16 12 Negro Paricatuba 15 9 Solim Purus Aruma Jusante 13 11 Solim Solim Japura Jacitara 10 8 Juru Solim Jutai Porto Antunes 7 6 I¸ Solim Itaquai Foz do Javari 3 14 Javari Foz do Itaquai 2 10 Solim Tapaj Amazonas Obidos 20 19 Trombetas Oriximina 19 11 Amazonas Itacoatiara 18 21 Madeira Foz 17 9 Negro Paricatuba 15 10 Solim Purus Arum Solim Japur Juru Solim Jutai Porto Antunes 7 6 Solim I¸ Solim Itaquai Foz do Javari 3 13 Javari Foz do Itaquai 2 12 Sample riverSolim Station No. V Nd: not determined.

1.2

Amazon R. 1 Solimoes R. Madeira R. Lowland Solimoes Tributaries 0.8 Lower Amazon Tributaries Negro 0.6

Alk (mmol.l-1) 0.4

0.2

0 3.50 4.00 4.50 5.00 5.50 6.00 6.50 7.00 7.50 8.00 pH

Figure 3. Relation of alkalinity to pH for dissolved material in surface waters of the Amazon River and its main tributaries

Konhauser et al. (1994) for the Rio Negro and Solimoes˜ (also near Manaus) and by Gaillardet et al. (1997). Our data is in excellent agreement with that already published, but shows some discrepancies with the data from Konhauser et al. (1994). The Cr, Co and Zn concentrations measured in the Rio Negro and Zn in the Rio Solimoes˜ are one order of magnitude less than those reported by Konhauser et al. (1994). In the Rio Negro, V, Mn and U are between two and four times lower that those reported by these authors. Otherwise most trace elements are present with concentrations in the range established as normal for other large rivers; differences may however be recognized for Zn, Sr, Ba and U in the Upper Solimoes˜ and Japura rivers, which have significantly higher content compared to other rivers of the Amazonian basin. The variations of dissolved concentrations of major cations along the Amazon mainstream (Solimoes˜ and Amazon rivers) are reported in Figure 4. As already observed by Stallard and Edmond (1983), concentrations decrease downstream from Tabatinga to Obidos.´ Conservative patterns are usually found for all of those elements in the Amazon River and in its tributaries (Ferreira et al., 1988). The concentration decrease is a consequence of the dilution of the high-concentration waters coming from the Andes by the low-concentration waters originating in drainage basins of the lowland and shield areas. Downstream distributions of dissolved trace element concentrations are reported in Figure 5, which compares low-flow and high-flow distributions. V, Sr, Ba, U and Sb closely mirror the decreasing pattern observed for the major cations, reflecting the high percentage of Andean whitewater upstream and continuous dilution downstream by blackwater and clearwater tributaries. Major cations and trace element concentrations are systematically higher during low-flow periods than high-flow periods, and this pattern is in agreement with the hypothesis of a dilution process. At the upstream stations during low-water period, V and Sr concentrations are between 25 and 40% higher than during high-water period. Cu, Mn, Mo show a general downriver decreasing, but lowland tributaries inputs influence the mainstream concentration profile: the slightly higher values measured for U and Mo at station 4 (Sao˜ Paulo de Oliven¸ca) and for Cu, Sb and Co at station 8 (Fonte Boa) are not related with the inputs of tributaries. The only parameter

Copyright  2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1345–1361 (2003) 1352 P. T. SEYLER AND G. R. BOAVENTURA Tefe Obidos Itapeua Parintins Tabatinga 0.6 Fonte Boa 250,000 Manacapuru Discharge LW Sao Paulo de O. Discharge HW

0.5 Jatarana/Itacoatiara Ca (LW) 200,000

Ca (HW) 0.4 Mg (LW) 150,000 Mg (HW) 0.3 K (LW) K (HW) 100,000 0.2 Na (LW) Conc. (mmol.l-1) Na (HW) Discharge (m3.s-1) 50,000 0.1

0 0 3500 3000 2500 2000 1500 1000 500 0 Distance from the mouth (km) Figure 4. Downstream distributions of major cations along the Solimoes˜ River and Amazon mainstream, between Tabatinga (2900 km from the mouth) and Obidos´ (800 km from the mouth). Distribution for each element given for both high-water and low-water stages that changes in the river reach is the DOC (Moreira-Turcq et al., in press), decreasing from 26Ð5mgl1 at station 1 (Tabatinga, 3000 km upstream the mouth) to 6Ð55 mg l1 at station 8 (Fonte Boa, 2300 km upstream the mouth). We may suppose that a fraction of these elements was adsorbed on the particulate organic matter and that desorption processes occurred when the Solimoes˜ waters mixed with Rio I¸ca waters. Ni, Cr, Rb, Co and Cd are relatively uniform throughout the mainstream. These observed distribution patterns reﬂect variations in both sources and transport of the dissolved material during the different stages of the hydrograph. Downstream the Negro and Solimoes˜ junction, the breakdown in the Cu, Rb, Co, Cd, Ni distribution patterns is due to the input of less concentrated waters coming from the Rio Negro, which remains unmixed with the Solimoes˜ waters for more than 25 km downwards the junction (Laraque et al., 1999). While it is possible to see seasonal signals in the downstream distributions of some trace elements, a clearer seasonal picture is available from time series data covering a whole hydrological cycle. The time series was obtained at the Obidos´ gauging station, situated 800 km upstream the mouth. The temporal variation of trace elements is reported in Figure 6. Several patterns are shown:

ž Elements for which the concentrations decrease with increasing discharge: this is the case for Sb, Mo, Cu, Sr, Ba and V. ž Elements for which the concentrations increase with increasing discharge, but which have their maximum concentration during a decreasing stage, one or two months after the peak discharge: this is the case for Co, Cd and Mn. ž Elements showing little variation with discharge: U, Rb, Ni, Cr.

Variations of river chemistry may reﬂect variations of the sources. As previously noted the ‘shield’ rivers (Negro and Tapajos)´ have typically depleted concentrations in Sb, V, Sr, Ba, Mo, Cu and V as compared with Andean rivers. The increased proportion of waters from these less solute-rich rivers during the high discharge period of the Amazon contributes to the observed decrease of concentrations.

Copyright  2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1345–1361 (2003) TRACE METALS IN THE AMAZON BASIN 1353 Tefe Tefe Obidos Itapeua

1800 Parintins 60 Obidos Tabatinga Itapeua Fonte Boa Parintins Manacapuru Tabatinga 1600 Fonte Boa Manacapuru Sao Paulo de O. 50 Jatarana/Itacoatiara 1400 Sao Paulo de O. V (LW) 25.0 25.0 Jatarana/Itacoatiara 1200 40 Madeira V (HW) Negro Sr (LW) 1000 20.0 20.0 30 Sr (HW) Ni (LW) Ba (LW) Ni (HW) 800 15.0 15.0 Ba (HW) Cr (LW) 600 20 Mn (LW) Cr (HW)

V Conc. (nmoll-1) 10.0 Mn (HW) 10.0 Rb (LW)

Sr & Ba Conc. (nmol/l) 400 Rb (HW) 10 5.0 5.0 200

0 0 Cr, Ni & Rb Conc. (nmol.I-1) 0.0 0.0 3500 3000 2500 2000 1500 1000 500 0 3500 3000 2500 2000 1500 1000 500 0 Distance from the mouth (km) Distance from the mouth (km)

35.0 8.0 30.0 7.0 2.00 2.50 25.0 6.0 Mo (LW) 5.0 1.80 20.0 Mo (HW) 4.0 Cu (LW) 1.60 2.00 15.0 3.0 Cu (HW) 1.40 10.0 Co (LW) 2.0 1.20 1.50 5.0 Co (HW) 1.0 1.00 Cu Conc. (nmol.I-1) Cd (LW) 0.0 Mo Conc. (nmoI.I-1) 0.0 0.80 1.00 35003000 2500 2000 1500 1000 500 0 Cd (HW) 0.60 Distance from the mouth (km) 0.40 0.50 Cd Conc. (nmol/1) Co Conc. (nmoll-1) 2.0 0.5 0.20 1.8 0.5 0.00 0.00 1.6 0.4 3500 3000 2500 2000 1500 1000 500 0 1.4 0.4

g.I-1) U (LW) Distance from the mouth (km) 1.2 0.3 µ U (HW) 1.0 0.3 Sb (LW) 0.8 0.2 0.6 0.2 Sb (HW) 0.4 0.1 U Conc (

Sb Conc. (nmol/I) 0.2 0.1 0.0 0.0 4000 3000 2000 1000 0 Distance from the mouth (km)

Figure 5. Downstream distributions of the trace element dissolved concentrations along the Solimoes˜ River and Amazon mainstream, between Tabatinga (2900 km from the mouth) and Obidos´ (800 km from the mouth). Distribution for each element given for both high-water and low-water stages. For graphical reasons, the elements are regrouped according to their concentrations

Elements such as Co, Cd and Mn are mainly transported by the flood flows. Mn is known to be concentrated in lateric (ferricrete) soils which represent 80% in the Amazon basin, suggesting that this element is washed away in solution during the high discharge. Moreover, these elements can be stored in the surrounding floodplain areas (varzea).´ At high water as much as 20 000 m3 s1 of water enter the varzeas,´ as compared to a whole discharge of about 168 000 m3 s1 at Obidos´ (Richey et al., 1989). There is a direct exchange of suspended sediment between the varzea´ and the main river through the processes of entrainment and deposition (Dunne et al., 1998). The deposition/resuspension cycle as well as the exchange rate between floodplain and mainstream channel may control at least partially the temporal variation of a redox-sensitive element such as Mn. With regard to Co and Cd, these elements show similar concentrations in the Solimoes,˜ Negro and Madeira rivers, and their variation at Obidos´ cannot be explained by the variations of sources. Therefore, processes occurring in the riverbed itself or in the floodplain lakes (with subsequent exchange with the river) seem likely and would help to explain seasonal Co and Cd distributions. Dissolved trace metals are not necessarily conservative upon mixing, since a large percentage of the reactive forms of some of those elements are adsorbed. The different physicochemical conditions occurring in the varzeas,´ as compared to the mainstream (for instance pH is about 2 units more acid in the varzeas´ than the Solimoes;˜ Piedade et al., 2001), may lead to some desorption when the Solimoes˜ waters mix with the more acidic waters of the

1000 300,000

250,000 Sb nM 100 200,000 Mo nM

Cu nM 10 150,000 Sr (nm

100,000 Ba nM Conc. (nM)

1 discharge (m3.s-1) V nm 50,000 Discharge

0.1 0

15/02/97 06/04/97 26/05/97 15/07/97 03/09/97 23/10/97 12/12/97 Time

U nM Rb (nM) Co nM Cd nM Ni nM Mn nM Cr nM Discharge Discharge 100 300,000 1000 300,000

250,000 250,000 100 10 200,000 200,000

150,000 10 150,000

Conc. (nM) 100,000 1 Conc. (nM) 100,000

1 discharge (m3.s-1) discharge (m3.s-1) 50,000 50,000

0.1 0 0.1 0 06/04/97 23/10/97 15/02/97 26/05/97 15/07/97 12/12/97 15/02/97 06/04/97 26/05/97 15/07/97 03/09/97 23/10/97 12/12/97 03/09/97 Time Time

Figure 6. Monthly variations of trace element concentrations from March to December 1997 at Obidos´ station varzeas.´ Moreover, the inﬂuence of biological processes occurring in the varzeacouldalsoplayanimportant´ role in the behaviour of nutrient-like elements such as Cd and Co (Collier and Edmond, 1984; Shiller and Hebert, 1998). Concerning the third type of pattern, U, Ni and Cr are relatively more depleted in the Negro waters than in the Solimoes˜ waters. These elements have a very limited solubility and are transported mainly in the particulate form. Following Gaillardet et al. (1997), pH strongly controls the transport phase of these elements. Adsorption/desorption processes and coagulation mechanisms may explain their low temporal variations. Rb is more soluble, but the narrow concentration ranges obtained in the major tributaries (Table II) are comparable with those found at Obidos´ station.

Figure 7. Comparison of the trace element suspended matter composition of the major Amazon tributaries

Transport of the particulate trace elements Major (Fe, Al) and trace elements of suspended particulate material are given in Table III. The trace element concentrations show a clear relationship with the location of the samples (Table III and Figure 7). For instance, V, Co, Cr, Mn, Sr, Ba concentrations are higher in the Solimoes˜ left-bank tributaries than in its right-bank tributaries. Left-bank tributaries come from the Andean cordillera (I¸ca and Japura rivers), while right-bank tributaries (Javari, Jurua, Purus) drain the soils and sediments of the sub-Andean trough and of the central plain. The Rio Solimoes˜ presents the highest concentrations for all the elements studied, with a few exceptions: Zn and Fe concentrations are more elevated in the Rio Negro, Co and Ni are in the same range in the Solimoes,˜ Negro and Madeira rivers. The composition of suspended sediments in the Amazon mainstream reflects the mixing of both, but is closer to the composition of its Andean tributaries. All the trace elements are highly autocorrelated. The correlation coefficients are always higher than 0Ð87 except for Fe, which is less strongly correlated with the others (coefficients >0Ð62). Similar observations have been made in many rivers (Meybeck and Helmer, 1989). As the trace element composition of the suspended matter of these rivers is highly dependent on the grain size distribution, a normalization with Al will be used (Figure 8). Flat patterns are observed for the samples from the Solimoes,˜ Madeira and Amazon mainstream. By contrast, the Rio Negro is enriched in Fe and impoverished in Mn, Co and Ni. The Trombetas show enrichment for Mn and to a lesser extent for Zn. The trace content in the suspended load of the Tapajos´ River is abnormally high for the entire set of data. The Tapajos´ River was probably sampled during a high primary productivity period and the Al content of its suspended sediment (3Ð17%) might be diluted by the particulate organic matter (POC value 15Ð4%; Moreira-Turcq et al., in press).

Partitioning between dissolved and particulate phases In order to compare the dissolved and suspended trace element loads of the different types of river of the Amazon basin, we computed the mass of each element in 1 l of river water. Several observations are apparent from Figure 9:

Copyright  2003 John Wiley & Sons, Ltd. Hydrol. Process. 17, 1345–1361 (2003) 1356 P. T. SEYLER AND G. R. BOAVENTURA 9 3 4 3 1 1 0 1 6 0 1 3 9 2 4 8 5 2 0 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð U (ppm) 58 72 62 019 01 23 63 82 02 17 42 32 51 52 72 92 92 62 32 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ba (ppm) 8 2319 3 532 3 410 0 4154 1 312 8 510 8 593 6 498 0 571 9 1531 1 542 0 603 3 491 5 634 7 704 2 713 7 433 9 434 6 621 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Sr (ppm) 8 408 346 7 2657 1 198 7 411 2 507 4 370 6 503 8 506 0 340 7 546 3 310 9 570 6 583 8 549 3 262 8 292 9 576 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Rb (ppm) 0 nd 217 266 846 9 469 251 933 368 972 570 357 876 465 034 463 448 170 751 752 660 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Zn (ppm) 1 130 4 131 3 249 2 934 3 101 7 228 7 144 1 137 8 131 5 111 7 126 7 124 565 6 118 8 103 2 144 5 106 8 109 5 129 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Cu (ppm) 829 035 622 9 253 521 225 441 629 543 254 255 641 423 844 642 650 924 822 341 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ni (ppm) 0 227 770 335 9 257 957 335 077 379 381 087 080 173 818 329 426 142 974 921 857 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Co (ppm) 857 914 713 098 28 911 617 717 216 630 315 914 610 813 712 816 514 412 817 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Mn (ppm) iment matter in the rivers of the Amazon basin 6 1272 1 618 9 1043 1 3673 9 439 5 343 9 755 6 676 4 693 0 802 0 1293 6 802 2 488 2 747 6 1036 2 816 4 473 4 448 2 427 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Cr (ppm) 261 244 8 415 839 256 466 265 766 3 132 553 357 247 359 450 8 125 261 254 965 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð V (ppm) 17 nd 204 60 132 53 85 70 902 70 57 73 56 94 142 32 119 89 144 88 135 30 119 30 135 47 79 33 143 07 132 74 139 55 124 34 108 32 129 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Al (%) 07 3 25 9 73 8 16 68 78 5 56 10 56 9 06 10 82 10 54 11 98 10 29 9 03 9 80 10 85 10 36 11 28 10 72 9 93 8 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Fe (%) ) 1 51 24 83 131 32 931 14 65 74 55 34 94 54 04 44 55 34 63 53 Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð (mg l Table III. Concentrations in the suspended sed ˜ ´ ao 3 a56 ´ a jusante 38 ˜ ao Paulo de O. 74 ˜ ˜ ˜ ˜ ˜ ˜ oes Manacapuru 127 oes Itapeua 63 oes Fonte Boa 60 oes Santo Antonio do I. 46 oes S oes Tabatinga 166 ´ ´ os Alter do Ch a Jacitara 28 ´ a Foz do Juru ca Ipiranga 41 9Juru 8 Solim 7 Jutai Porto Antunes 13 6 Solim 5I¸ 4 Solim 3 Itaquai Foz do Javari 148 2 Javari Foz do Itaquai 127 1 Solim 21 Tapaj 20 Amazonas Obidos 44 19 Trombetas Oriximina 14 18 Amazonas Itacoatiara 46 17 Madeira Foz 21 15 Negro Paricatuba 8 14 Solim 13 Purus Arum 12 Solim 10 Japur Sample river Station SPM nd: not determined SPM: suspended particulate matter concentrations.

Figure 8. Variations of the trace element composition of suspended particulate matter along the Solimoes˜ River and Amazon mainstream. Elemental concentrations are normalized with Al content of suspended matter

ž For the Solimoes˜ River, V, Cr, Mn, Co, Ni and Zn are almost entirely carried by the river particulate matter; Cu, Rb, Sr, Ba and U are transported mainly by the suspended particles, but a dissolved phase contributes to the transport. ž For the Negro River, the proportion of elements transported by the dissolved phase is higher for the whole set of elements; the proportion of V, Cr, Mn, Rb, Sr, Ba and U associated with the dissolved phase accounts for more than half of the total transport. ž For the Amazon River at Obidos,´ the proportion between the dissolved and particulate transport clearly shows an intermediate pattern. According to Martin and Meybeck (1979) and Martin and Whitﬁeld (1983), the increase of the particulate transport observed from the Rio Negro to the Solimoes˜ can be explained by the degree of mobility for a given element during the weathering processes. As already emphasized, and concerning at least the tropical weathering type, Cu, Rb, Sr, Ba and U are the most easily washed trace elements, whereas V, Cr, Mn, Co, Ni and Zn are the less ‘mobile’.

SUMMARY AND CONCLUSIONS This study provides a comprehensive survey of trace elements in the mainstream Amazon River. Concen- trations of most elements are comparable to those of major pristine rivers in the word. The Andean part of

Solimóes at Manacapuru 100%

80%

60% part diss 40%

20%

0% V CrMnCoNiCuZnRbSrBaU Elements

Negro river 100%

80%

60% part diss 40%

20%

0% V CrMnCoNiCuZnRbSrBaU Elements

Madeira river 100%

80%

60% part diss 40%

20%

0% V CrMnCoNiCuZnRbSrBaU Elements

Figure 9. Mass proportions of trace element transported as dissolved and particulate forms in the Solimoes˜ River at Manacapuru, in the Negro River at Paricatuba (Manaus), in the Madeira River before the conﬂuence with the Amazon and in the Amazon River at Obidos´

Aamazon River at Óbidos 100%

80%

60% part diss 40%

20%

0% VCrMnCoCuZnRbSrBaU Elements Figure 9. (Continued)

Figure 10. Trace element exportation fluxes from the Amazon River. Error bars represent the sum of errors assuming the following uncertainties: water discharge (10%), solid discharge (10%), analytical errors of dissolved concentrations (10%) and of particulate concentrations (5%) the basin stands out as the dominant source area for most trace elements transported by the river. Tempo- ral variability of trace element concentrations reflects the source, remobilization and/or biological processes occurring in the channel or in the surrounding floodplain lakes. In the Amazon mainstream, trace metals are predominantly transported in the particulate phase.

In order to estimate the riverine fluxes from the Amazon system to the Atlantic Ocean, we calculate the amount of trace elements as particulate and dissolved forms transported at Obidos.´ The variability of dissolved metal concentrations in the Amazon River discussed above indicates that single samples are not representative of fluvial trace element concentrations. Therefore, the systematic relationships obtained between dissolved trace elements and discharge (Figure 6) are used to compute the dissolved river fluxes. The particulate flux of the set of trace elements is estimated by multiplying the trace element composition of the suspended load by the total suspended discharge corresponding to the same sampling period. The results are shown in Figure 10. As is evident, for all the elements, riverine particulate fluxes are higher than dissolved fluxes. Only for the most mobile ones (Sr, Ba, Cu) does the dissolved flux contribute a part of the total flux. These values can be considered as the riverine fluxes of trace metals entering the Amazon estuarine zone. As the chemical, physical and biological processes that take place in the river/ocean mixing zone provide an important control on the riverine fluxes (Boyle et al., 1982 and references cited therein), further studies on the trace metals in the Amazon estuarine zone should be undertaken to assess the net exportation fluxes of the Amazon River to the Atlantic Ocean.

ACKNOWLEDGEMENTS This study was carried out within the framework of the HiBAM project (Hydrology and Geochemistry of the Amazonian Basin) under the auspices of Brazilian and French organizations: Conselho Nacional de Desenvolvimento Cientif´ıco e Tecnologico (CNPq) and the Institut de recherche pour le developpement´ (IRD). We would like to thank the ANNEL technical group and the crew of R/V C. Dario for their help during the cruises.

REFERENCES Bernoux M, Graca PMA, Cerri CC, Fearnside PM, Feig BJ, Piccolo MC. 2001. Carbon storage in biomass and soils. In The Biogeochemistry of the Amazon Basin and its Role in a Changing World, McClain M, Victoria RL, Richey JE (eds). Oxford University Press: Oxford; 165–184. Boyle EA, Huested S, Grant B. 1982. The chemical mass balance of the Amazon plume. II. Copper, nickel, and cadmium. Deep-Sea Research 29(11A): 1355–1364. Collier R, Edmond J. 1984. The trace element geochemistry of marine biogenic particulate matter. Progress in Oceanography 13: 113–199. Dunne T, Mertes LAK, Meade RH, Richey JE, Forsberg BR. 1998. Exchanges of sediment between the ﬂood plain and channel of the Amazon River in Brazil. Geological Society of America Bulletin (V) 110: 450–467. Elbaz-Poulichet F, Seyler P, Maurice-Bourgoin L, Guyot JL, Dupuy C. 1999. Trace element geochemistry in the upper Amazon drainage basin (Bolivia). Chemical Geology 157: 319–334. Ferreira JR, Devol AH, Martinelli L, Forsberg BR, Victoria R, Richey JE, Mortatti J. 1988. Chemical composition of the Madeira River: seasonal trends and total transport. Mitbouwn Geology-Palaeontology Institut Universitad Hamburg 66: 63–75. Filizola N, Guyot JL, Baoventura G. 1999. Fluxo de sedimentos em suspensao˜ na Amazonia;ˆ uma analise´ a partir da base de dados da ANEEL. In International Symposium on Hydrological and Geochemical Processes in Large Scale River Basins. Manaus, November 15–19, 1999. IRD-HiBam Publication: Brazil (CD-ROM). Furch K. 1984. Seasonal variation of the major cation content of the varzea-lake´ Lago Camaleao, middle Amazon, Brazil, in 1981 and 1982. International Verein Limnology 22: 1288–1293. Gaillardet J, DupreB,All´ egre` CJ, Negrel´ P. 1997. Chemical and physical denudation in the Amazon River Basin. Chemical Geology 142: 141–173. Guyot JL, Filizola N, Laraque A, Seyler P. 1999. La variabilite´ saisonniere` des ﬂux sedimentaires´ dans le basin de l’Amazone. In International Symposium on Hydrological and Geochemical Processes in Large Scale River Basins. Manaus, November 15–19, 1999. IRD-HiBam Publication: Brazil (CD-ROM). Konhauser KO, Fyfe WS, Kronberg BI. 1994. Multi-element chemistry of some Amazonian waters and soils. Chemical Geology 111: 155–175. Laraque A, Guyot JL, Seyler P, Filizola NP. 1999. Dynamique hydrologique et geochimique de la rencontre des rios Solimoes˜ et Negro dans le bassin de l’Amazone. In International Symposium on Hydrological and Geochemical Processes in Large Scale River Basins. Manaus, November 15–19, 1999. IRD-HiBam Publication: Brazil (CD-ROM). Marengo JA, Nobre CA. 2001. General characteristics and variability of climate in the Amazon basin and its links to the global climate system. In The Biogeochemistry of the Amazon Basin and its Role in a Changing World, McClain M, Victoria RL, Richey JE (eds). Oxford University Press: Oxford; 17–41.

Martin JM, Gordeev VV. 1986. River input to ocean system; a reassessment. In Estuarine Processes: Application to the Tagus Estuary. Unesco and CNA: Lisbon; 203–240. Martin JM, Meybeck M. 1979. Elemental mass-balance of material carried by major world rivers. Marine Chemistry 7: 173–206. Martin JM, Whitfield M. 1983. The significance of the river input of chemical elements to the ocean. In Trace Metals in Sea Water, Wong CS, Boyle EA, Bruland KW, Burton JD, Goldberg ED (eds). Plenum: New York; 265–296. Meybeck M, Helmer R. 1989. The quality of rivers: from pristine stage to global pollution. Palaeogeography, Palaeoclimatology, Palaeoecology 75: 283–309. Molinier M, Guyot JL, Oliveira E, Guimaraes˜ V. 1997. Les regimes´ hydrologiques de l’Amazone et de ses affluents. In L’hydrologie tropicale: géoscience et outil pour le développement, Chevallier P, Pouyaud B (eds). AIHS 238: 209–222. Moore WS, Edmond JM. 1984. Radium and barium in the Amazon River system. Journal of Geophysical Research 89(C2): 2061–2065. Moreira-Turcq P, Seyler P, Guyot JL, Etcheber H. 2003. Exportation of organic carbon from the Amazon river and its main tributaries. Hydrological Processes 17: 1329–1344. Palmer MR, Edmond JM. 1992. Control over the strontium isotopic composition of river water. Geochimica Cosmochimica Acta 56: 2099–2112. Piedade MT, Worbes M, Junk WJ. 2001. Geo-ecological controls on elemental fluxes in communities of higher plants in Amazonian floodplains. In The Biogeochemistry of the Amazon Basin and its Role in a Changing World, McClain M, Victoria RL, Richey JE (eds). Oxford University Press: Oxford; 209–234. Richey JE, Mertes LAK, Dunne T, Victoria R, Forsberg BR, Tancredi ACFNS, Oliveira E. 1989. Source and routing of the Amazon River flood wave. Global Biogeochemical Cycles 3(3): 191–204. Seyler P, Boaventura G. 2001. Trace metals in the mainstem river. In The Biogeochemistry of the Amazon Basin and its Role in a Changing World , McClain M, Victoria RL, Richey JE (eds). Oxford University Press: Oxford; 307–327. Seyler P, Elbaz-Poulichet F. 1996. Biogeochemical control on the temporal variability of the trace element concentrations in the Oubangui River (Central African Republic). Journal of Hydrology 180: 319–332. Seyler P, Guyot JL, Maurice-Bourgoin L, Sondag F, Elbaz-Poulichet F, Etcheber H, Quintanilla J. 1999. Origin of trace elements in the Bolivian Amazonian drainage basin. In Hydrology in the Humid Tropic Environment, Johnson AI, Fernandez-Jauregui CA (eds). IAHS 245: 55–63. Shiller AM, Hebert TL. 1998. Biogeochemical controls on dissolved trace elements in rivers. Mineralogical Magazine 62A: 1391–1392. Sioli H. 1984. The Amazon and its main affluents: hydrography, morphology, river courses and river types. Monography in Biology 65: 85–127. Stallard RF. 1980. Major element chemistry of the Amazon River system. PhD dissertation, MIT/WHOI, Cambridge, MA; 362. Stallard RF, Edmond JM. 1983. Geochemistry of the Amazon. 2. The influence of geology and weathering environment on the dissolved load. Journal of Geophysical Research 88(C14): 9671–9688. Stallard RF, Edmond JM. 1987. Geochemistry of the Amazon. 3. Weathering chemistry and limits to dissolved inputs. Journal of Geophysical Research 92: 8293–8302.