Ž .The Science of the Total Environment 261 2000 9]20

Fish mercury concentration in the Alto Pantanal,Brazil: influence of season and water parameters

Lars D. Hylandera,U, Fernando N. Pintob, Jean R.D. Guimaraesb,˜Markus Meili1c, Lazaro J. Oliveirad, Edinaldo de Castro e Silvae´

aDepartment of Limnology, E¨olutionary Biology Centre, Uppsala Uni ersity, Norby agen 20, S-752 36, Uppsala, Sweden¨b ´Laboratorio de Radioisotopos, Inst. de Biofisica, Uni ersidade Federal do Rio de Janeiro, Bloco G CCS, Ilha do Fundao,´ ˜

21 949-900 Rio de Janeiro, RJ, BrazilcDepartment of Earth Sciences, Uppsala Uni ersity, Villa¨agen 16, S-752 36 Uppsala, Sweden¨

dDepartamento de Ciencias Biologicas, UNEMAT-Uni ersidade do Estado de Mato Grosso, A¨. Sao Joao srn, Campus˜ ˜Uni ersitario-Bairro Ca¨alhada, 78200]000 Caceres, MT, Brazil´

eDepartamento de Quımicar Pos Graduacao em Saude e Ambiente-Uni ersidade Federal de Mato Grosso, 78090-900´ ´ ˜ ´Cuiaba, MT, Brazil´

Received 27 September 1999; accepted 25 February 2000

Abstract

Ž .The tropical flood plain Pantanal is one of the world’s largest wetlands and a wildlife sanctuary. Mercury Hgemissions from some upstream gold mining areas and recent findings of high natural Hg levels in tropical oxisolsmotivated studies on the Hg cycle in the Pantanal. A survey was made on total Hg in the most consumed piscivorous

Ž .fish species from rivers and floodplain lakes in the north Caceres and Barao de Melgaco and in the south part of´ ˜Ž .Alto Pantanal around the confluence of the Cuiaba and Paraguai rivers . Samples were collected in both the rainy´

Ž . Ž . Žand dry seasons March and August 1998 and included piranha Serrasalmus spp. , and catfish Pseudoplatystoma.coruscans, pintado, and Pseudoplatystoma fasciatum, cachara or surubim . There was only a small spatial variation in

Hg concentration of the 185 analyzed fish samples from the 200=200 km large investigation area, and 90%Ž y1.contained total Hg concentration below the safety limit for regular fish consumption 500 ng g . Concentration

above this limit was found in both Pseudoplatystoma and Serrasalmus samples from the Baıa Sia Mariana, the only´ ´acid soft-water lake included in this study, during both the rainy and dry seasons. Concentration above this limit wasalso found in fish outside Baıa Sia Mariana during the dry season, especially in Rio Cuiaba in the region of Barao de´ ´ ´ ˜Melgaco. The seasonal effect may be connected with decreasing water volumes and changing habitat during the dry

U Corresponding author. Tel.: q46-18-4712710; fax: q46-18-531134.Ž .E-mail address: lars.hylander@ebc.uu.se L.D. Hylander .

1 Ž .Present address: Institute of Applied Environmental Research ITM , Stockholm University, S-106 91 Stockholm, Sweden.

0048-9697r00r$ - see front matter Q 2000 Elsevier Science B.V. All rights reserved.Ž .PII: S 0 0 4 8 - 9 6 9 7 0 0 0 0 5 9 1 - X

( )L.D. Hylander et al. r The Science of the Total En¨ironment 261 2000 9]2010

season. The results indicate that fertile women should restrict their consumption of piscivorous fishes from the RioCuiaba basin during the dry season. Measures should be implanted to avoid a further deterioration of fish Hg levels.´Q 2000 Elsevier Science B.V. All rights reserved.

Keywords: Soft-water; Catfish; Mercury; Piranha; Season; White-water

1. Introduction

The tropical flood plain Pantanal is one of theworld’s largest wetlands and a wildlife sanctuary.Fishing contributes substantially to the peopleliving in the Pantanal wetland, either directly forconsumption or indirectly as an attraction for agreat number of tourists visiting the region. Anexample is the community Barao de Melgaco˜Ž .6000 inhabitants . It was founded around sugarand alcohol production from sugar cane, laterdepended on beef and leather as the sugar canefields were eroded and replaced by pastures, andpresently gets its major income from marketed

Ž .fish Mattos, 1997 . Actions reducing the quantityor quality of fish hence will seriously affect theliving conditions of the Pantanal population.

Erosion will reduce the fish stock due to deteri-orated feeding and mating conditions as an effectof reduced visibility and a lower primary produc-tion, due to less photosynthesis. Also the mercuryŽ .Hg concentration in the river water is expectedto increase as an effect of erosion and leaching ofHg from soil without a protective vegetation layerŽ .Roulet et al., 1998a,b .

Artisanal gold-mining has been suspected to beresponsible for the release of much of the mer-cury encountered in tropical floodplains such as

Žthe Amazon and the Pantanal in Brazil e.g. Lac-.erda et al., 1990 . During the gold-rush of the

past decades, an estimated 2000 t of Hg has beenŽreleased in the Amazon region Veiga et al.,

.1995 . Furthermore, the Spaniards brought an-other estimated 260 000 t of Hg into Latin Amer-ica between 1550 and 1930, mostly used for silver

Ž .amalgamation Nriagu, 1993; Lacerda, 1997 . Onthe other hand, it has been concluded that mostHg in remote soils and sediments is supplied by

Ž .atmospheric deposition Fitzgerald et al., 1998 .Large quantities of Hg can be accumulated in

some soils and can eventually reach surface wa-

ters, in particular at improper management.Ž .Roulet et al. 1998a,b suggested that Hg entering

Amazonian soils is initially firmly bound to soilconstituents like sesquioxides and crystalline Fe

Ž .and Al oxy-hydroxides. Forsberg et al. 1999 esti-Žmated that the soils in the Rio Negro basin a

.tributary to the Amazon river contains consider-ably more Hg than could possibly be of anthro-pogenic origin, in spite of three times higherglobal Hg deposition nowadays compared to pre-industrial deposition. This Hg may leach from thesoil by organic acids and can cause Hg concentra-tion in fish to exceed the limit for safe consump-

Ž .tion Nakazono et al., 1999 .Mercury emissions from some upstream gold

mining areas and recent findings of high naturalHg levels in tropical oxisols motivated studies onthe Hg cycle in the Pantanal. The contribution ofanthropogenic Hg to the fish Hg content is oflarge importance in many temperate areas with

Žboreal forests like Northern Europe e.g. Meili,.1997 , while the effect on fish Hg content in

tropical regions is at debate. Nevertheless fishesare an important food resource and useful asbioindicators for areas with toxic levels of Hg ofanthropogenic as well as of natural origin. Manyfactors influencing fish Hg levels in the tropicsare only poorly understood, e.g. seasonal changes,organic carbon dissolved in water and the largevariety of species, which form complex food webs.

The aim of the present study was to survey theHg levels in commonly consumed, piscivorousfishes in rivers and floodplain lakes of the north-

Ž .ern high part of the Pantanal to verify earlierreports of Hg concentration in fish exceeding the

Žlimit for consumption Vieira, 1991; Hylander et.al., 1994 . Another aim was to identify relation-

ships between the fish Hg concentration and wa-ter parameters of the different areas and sites. Athird aim was to compare samples from both

( )L.D. Hylander et al. r The Science of the Total En¨ironment 261 2000 9]20 11

Ž .rainy and dry seasons. Total Hg HgTot contentwas determined in all fish samples and methyl-HgŽ .MeHg was determined in two species from dif-

Žferent locals during both seasons in all 23 sam-.ples to determine any local or seasonal diver-

gence from the general MeHg-percentage infish. The presented data are part of aSwedish]Brazilian collaboration project with theaim to study the accumulation and transforma-tion of mercury in tropical rivers and wetlands inthe Brazilian Pantanal.

2. Study area

The savanna wetland Pantanal is a 137 000 km2

large alluvial plain in central-western Brazil fac-ing the border of Bolivia and Paraguay, receivingwater from a surrounding upland drainage basin

2 Ž .that occupies 359 000 km Hamilton et al., 1996 .The area is hydrologically divided into the AltoŽ . Ž .high , Medio and Baixo low Pantanal. Flood-´plains make up approximately 20% of the up-

Ž .stream watershed Alto Pantanal and most of theŽ .area in the Baixo Pantanal Alho et al., 1988 .

The rain falling from October until March resultsin an annual flooding of usually more than halfand often more than 90% of the alluvial plain,while on the other hand permanent open-water

Ž .areas are minimal Hamilton et al., 1996 . Theannual fluctuation of the inundation area is mostextreme in the Alto Pantanal, where the inunda-tion is maximal in March]April and minimal in

Ž .October]November Hamilton et al., 1996 . TheŽ .Alto Pantanal Fig. 1 is the northernmost part of

the Rio Paraguai river basin, including the tribu-Žtary Rio Cuiaba approx. 15.58]17.58S and´

.558]598W . The rivers run southwards, and are tothe north separated from the Amazon basin bythe Serra dos Parecis and Serra Azul mountain

Žchains Estudos hidrologicos da bacıa do Alto´ ´.Paraguai, 1974 . Rio Paraguai has within the stud-

ied area a slope of 0.03]0.15 m kmy1, while theeast]west slope for the plain is 0.3]0.5 m kmy1

Ž .Zeilhofer, 1996 , causing Rio Cuiaba to flow´faster than Rio Paraguai through the plain, be-fore they join in the southern part of the studyarea.

The Alto Pantanal houses gold miners, sinceveins of gold-containing coarser material are

Ž . Ž . Ž . Ž .Fig. 1. Location of sampling sites in the Alto Pantanal Brazil : 1 Rio Cuiaba at Cuiaba; 2 Rio Cuiaba at Barao de Melgaco; 3´ ´ ´ ˜Ž . Ž . Ž . Ž .Rio Mutum and Baıa Sia Mariana; 4 Rio Cuiaba at Porto Jofre and downstream; 5 Baıa do Burro; 6 Baıa Amolar; 7 Baıa´ ´ ´ ´ ´ ´

Ž . Ž . Ž .Paraıso; 8 Rio Paraguai from Baıa Cachorada to Acurizal; 9 Rio Paraguai South and West of Caceres; and 10 Rio Paraguai at´ ´ ´Baıa Ximbuva.´

( )L.D. Hylander et al. r The Science of the Total En¨ironment 261 2000 9]2012

found in the clay layers at some places on theŽ .margin of the plain Rodrigues Filho, 1995 . This

material is excavated in huge open cast mines aswell as small pits and processed for gold extrac-

Ž .tion. During this process mercury Hg is used asan amalgamating agent, which results in Hg losses

Žto the air and the water of the wetland Lacerdaet al., 1991; Veiga et al., 1991; Hylander et al.,1994; Rodrigues Filho, 1995; Tumpling et al.,¨

.1995; Alho and Vieira, 1997 .Another source of Hg to the flood plain is Hg

dissolved in river water or sorbed to soil particlesand brought there by the rivers, eventually withincreased Hg burdens as a result of the extensiveerosion. The water, falling as rain on the sur-rounding Guimaraes Plateau, carries a large bur-˜den of suspended material on its route down tothe plain. The erosion is mainly caused by large-scale agriculture and by cutting trees and busheson the river banks and other areas sensitive to

Ž .erosion Goncalves, 1996 . Also uncontrolled sec-ondary gold mining and dredging for constructionsand from the river bottom contribute to theerosion, as well as the intense traffic of largebarges, promoting sediment resuspension andriver bank erosion by waves or simply by friction,in narrow or curved river areas.

3. Material and methods

Four fish species, two catfish species and twopiranha species, were studied: Pseudoplatystoma

Ž .fasciatum, Linnaeus 1766 , locally named‘cachara’ and ‘surubim’; Pseudoplatystoma corus-

Ž .cans, Agassiz 1829 , locally named ‘pintado’; thebig red and black piranha Serrasalmus nattereri orPygocentrus nattereri, ZUEC No 1580; and thesomewhat smaller Serrasalmus spilopleura, ZUEC

Ž .No 1602 Sazima and Machado, 1990 . They werecollected during two campaigns, one at the end ofthe rainy season in March 1998, and the other at

Žthe end of the dry season in August 1998 thefishes from Caceres were collected 2 months later´

.due to logistics . Water pH, dissolved oxygen,temperature, conductivity and turbidity weremeasured in the field with a multipurpose field

Ž .instrument Horiba U-10 during both campaigns.

The piranhas and half of the catfishes werecaught by the authors, while the rest was obtainedfrom local fishermen found fishing simultaneouslyalong the rivers. The fishes were weighed and

Ž .measured standard length in the field. Approxi-mately 10-g fillet samples were removed andstored frozen until the Hg analysis. The useddigestion methodology for determination of Hg-Tot corresponded mainly to that of Bastos et al.Ž .1998 , who modified a method of Agemian and

Ž . Ž .Cheam 1978 . Hydrogen peroxide 1 ml H O2 2Ž .was added to 0.1]1 g typically 0.5 g fresh weight

of homogenized fish muscle, each sample in tripli-cate. After 10 min, 3 ml of a solution of concen-

Ž .trated H SO and HNO 1:1 were added to the2 4 3sample recipients that were kept in an ice bath.

Ž .The recipients were placed in a water bath 608Cfor 30 min, then cooled down in an ice bath. Afteraddition of 5 ml of 5% KMnO solution, the4samples were left overnight, after which somedrops of a solution of 12% hydroxyl ammonium

Ž .chloride HONH Cl were added until the oxidiz-3Ž .ing agent KMnO was neutralized. The HgTot4

content was then determined by a atomic absorp-tion spectrometry instrument, equipped with acold vapor generator accessory with sodium

Žborohydride as a reducing agent AA-FIMSrPer-.kin Elmer-FIAS 400 .

Fish MeHg content was determined accordingto a method developed by Akagi and NishimuraŽ . Ž .1991 . Fish fillet samples 0.5 g in duplicate were

Ž .digested under heat 1008C with a solution ofKOH and ethanol, followed by dithizone-benzeneextraction. The benzene phase was washed withNaOH before electron-capture detection gas

Ž .chromatography ECD-GC for determination ofMeHg. Further details in Kehrig and MalmŽ .1999b .

Distilled and de-ionized water and laboratorygrade chemicals were used for preparing solu-tions. Accuracy of the HgTot and MeHg analyseswas checked by comparison with certified refer-

Žence materials dogfish, tuna, and mussel sam-.ples . Precision was verified by internal standards

and replicated analyses. The detection limit ofy1 Ž .HgTot was below 10 ng g fresh wt. fw fish

muscle when using a 0.5 g sample and 10 ml ofŽextractant 3 S.D. for 34 blanks run at 12 days of

( )L.D. Hylander et al. r The Science of the Total En¨ironment 261 2000 9]20 13

Table 1ŽTotal Hg concentration encountered in certified reference samples and in an internal reference sample AFPX5130, fillet of

.Pseudoplatystoma coruscans from Alta Floresta

y1 y1Ž . Ž .Sample Total Hg ng g Methyl-Hg ng g

Certif. value Obtained S.D. R.S.D. n Certif. value Obtained S.D. R.S.D. nŽ . Ž . Ž . Ž .95% C.I. value % 95% C.I. value %

Ž .DORM-1 dogfish 798 755 46 6.1 3 ] ] ] ] ]Ž .724]872

Ž .DORM-2 dogfish 4640 4428 ] ] 1 ] ] ] ] ]Ž .4380]4900

Ž .IAEA 350 tuna 4680 4297 343 6.6 3 3650 3390 130 3.9 8Ž . Ž .4400]4960 3300]4000

Ž .IAEA 142 mussel ] ] ] ] ] 47 46.3 2.8 6.0 2Ž .43]51

aŽ .AFPX5130 catfish 13 320 14 403 1182 8.2 8 ] ] ] ] ]Ž .11 180]15 460

a Ž .Value obtained from inter-laboratory comparisons Bastos, 1997 .

.extraction . Typical mean values of triplicatedsamples had a standard error of approximately3.6%. The relative standard deviation of analyzedHgTot concentration values in the internal refer-

Ž .ence sample was below 9% Table 1 . Valuesobtained when analyzing certified reference sam-ples were with one exception within the 95%

Ž .confidence interval Table 1 .The detection limit of MeHg was 0.5 ng gy1 fw

fish muscle, determined by diluting series of MeHgstandard in benzene until no peak of MeHg wasobserved in the chromatogram. Typical mean val-ues of duplicated samples had a standard error ofapproximately 5.2%. Values of MeHg obtainedwhen analyzing certified reference samples were

Ž .within the 95% confidence interval Table 1 .

4. Results

Surface waters with different hydrological char-acteristics were found in the Alto Pantanal. TheRio Cuiaba is a white-water river with a high load´

Žof suspended material, causing high turbidity Ta-.ble 2 . The tributary Rio Mutum has acid soft-

water with signs of oxygen depletion, and with aninsignificant load of suspended material, so thedissolved organic matter will dominate its water

Ž .characteristics Table 2 . The gently flowing Rio

Paraguai has a lower load of particulate matter inŽ .suspension than Rio Cuiaba Table 2 .´

The HgTot concentration in the four studiedfish species ranged from 41 to above 2000 ng Hggy1 fw; the mean concentration was 265 ng Hggy1 fw, the median 194, the first quartile 130 and

y1 Ž .the third quartile 345 ng Hg g fw Table 3 .The concentration varied widely among individu-als, especially during the dry season, and no sta-tistically significant difference in concentrationwas found between the studied species. The con-centration varied among seasons and sites. Fishcaught in March, at the end of the rainy season,had approximately 50% lower HgTot concentra-

Ž y1 .tion average 190 ng HgTot g fw than fishŽcaught during the dry season in August average

y1 .345 ng HgTot g fw . The fish HgTot concentra-tions were generally below 300 ng HgTot gy1 fwin March, except for fish caught in Baıa Sia´ ´Mariana and a few P. fasciatum, caught in the

Ž .region of Porto Jofre Fig. 2 . Fish HgTot concen-tration increased during the dry season at all sitesexcept for Caceres. Most fish caught in the Rio´Cuiaba river basin in August contained more´than 300 ng HgTot gy1 fw and 20% containedHgTot concentration above the consumption limit

y1 Ž .of 500 ng HgTot g fw Brasil, 1975 .Fish samples analyzed for MeHg contained in

Ž .average 92% MeHg of HgTot Table 4 . This is a

( )L.D. Hylander et al. r The Science of the Total En¨ironment 261 2000 9]2014

Table 2Ž . Ž .Turbidity, pH, conductivity, dissolved oxygen, and dissolved organic carbon in the water of floodplain lakes Baıa and in rivers Rio´

aŽ .of Alto Pantanal Brazil in August 1998

Sampling site Turbidity pH Conductivity DO DOCy1 y1 y1Ž . Ž . Ž . Ž .NTU mS cm mg l mg l

Rio Paraguai, Ximbuva 24 6.1 22 7.3 3.9Rio Paraguai, W Caceres 24 6.1 35 7.7 3.4´Rio Paraguai, S Caceres 28 6.1 34 7.7 3.3´Rio Paraguai, Cachorada 22 6.7 59 6.9 6.4Rio Paraguai, Acurizal 22 6.9 58 6.7 7.3Rio Paraguai, Paraıso 19 6.3 58 6.8 8.1´

Ž . Ž . Ž . Ž .Baıa Paraıso 3 24 6.6 6.1 59 48 2.7 5.4 7.7´ ´Rio Paraguai, Amolar 32 6.4 45 6.5 6.5

Ž . Ž . Ž . Ž .Baıa do Amolar 26 20 6.4 6.0 47 42 6.5 3.7 6.7´Rio Cuiaba, Cuiaba 8 6.9 77 7.4 3.9´ ´Rio Cuiaba, Jofre 35 6.1 27 6.3 3.1´Rio Cuiaba, Burro 47 6.5 27 7.0 3.3´

Ž . Ž . Ž . Ž .Baıa do Burro 38 33 6.4 6.0 31 41 7.1 5.1 5.6´Rio Mutum 27 4.5 3 5.6 3.7

Ž . Ž . Ž . Ž .Baıa Sia Mariana 23 1 5.3 4.2 4 16 6.5 4.2 5.1´ ´a Values from March, where available, within parentheses.

commonly encountered percentage in fish not af-fected by exceptionally large exposure to metallic

Ž .Hg Mikac et al., 1985; Malm et al., 1995 . Therewas no statistically significant difference inMeHg-percentage between seasons, sites or

Ž .species Table 4 .

5. Discussion

At the onset of the rainy season in the end ofSeptember or October, the rivers rise and waterflows slowly and freely outside the river beds,resulting in inundation of large areas and a mix-ing of waters from areas drained by differentwater courses during the dry season. Hence thedistinct characteristics of different sites at the endof the dry season are partly levelled out duringthe rainy season, which is evident from the vari-ability of HgTot concentration in fish from dif-

Ž .ferent locals Table 3 . The inundation also re-sults in a many times larger habitat for fish, whichduring the dry season are gathered in the rivercourses or depressions forming shallow flood plainlakes. The expansion leads to an increasing avail-ability of food, in particular inundated terrestrialvegetation with comparatively low HgTot concen-

tration. The contraction is likely to be accom-panied by a higher degree of predation, longerfood chains, and probably also a bioconcentrationof Hg in many animals as a result of lower growthefficiency due to starvation. The onset of therainy season and accompanied spawning also re-sults in extensive migration, especially by the cat-fish, while the piranhas are more stationaryŽ .Sazima and Machado, 1990 . Many of the catfish,who during the dry season are living on the bot-tom in the deep rivers, are during the rainy sea-

Žson moving into the flood plain lakes F.A..Machado, personal communication, 1998 . The

mixing of water, the habitat fluctuation, and thefish migration patterns complicate the evaluationof the influence by water parameters on fishHgTot concentration.

Baıa Sia Mariana contained mainly acid soft-´ ´water at the sampling in March. Although thedissolved organic carbon is in parity with theother waters, the reduced amount of suspendedparticulate matter as well as dissolved ions causesdissolved organic matter to determine much ofthe water characteristics in Baıa Sia Mariana.´ ´This type of water, as well as black-water with ahigher organic carbon content, is known to in-crease the mobility and probably also the

( )L.D. Hylander et al. r The Science of the Total En¨ironment 261 2000 9]20 15

Table 3Lowest, highest, median values, lower and upper quartile of total Hg concentration encountered in fillet of Pseudoplatystoma

Ž . Ž . Ž .fasciatus cachara , Pseudoplatystoma coruscans pintado , Serrasalmus spilopleura and Serrasalmus nattereri piranhas caught ataŽ . Ž . Ž . Ž .different sites see Fig. 1 in the Alto Pantanal Brazil in March rainy season and August dry season 1998, and average

standard length "S.D. of analyzed fishes

y1Ž .Species, season and LengthHgTot ng gŽ .sampling site cmMin. 25% Median 75% Max. n

quartile quartile

P. fasciatus, rainy seasonBaıa Sia Mariana 118 238 272 512 518 5 77"6´ ´Barao de Melgaco 107 130 263 370 446 5 78"5˜Downstream Porto Jofre 278 342 420 3 71"33

P. fasciatus, dry seasonBaıa Sia Mariana 362 1114 2 64"23´ ´Barao de Melgaco 91 371 429 502 515 5 78"6˜Downstream Porto Jofre 193 259 377 403 843 5 78"6Acurizal 186 298 306 3 79"4

P. coruscans, rainy seasonBaıa Sia Mariana 127 201 211 687 724 5 78"6´ ´Barao de Melgaco 55 64 140 152 168 5 85"12˜Downstream Porto Jofre 47 85 136 193 264 9 90"20Acurizal 60 65 96 170 202 7 110"20Caceres 121 135 364 3 80"6´

P. coruscans, dry seasonBaıa Sia Mariana 403 1 54´ ´Barao de Melgaco 132 301 320 350 371 5 113"9˜Downstream Porto Jofre 137 169 263 352 761 11 96"21Acurizal 98 99 290 302 812 5 96"17Caceres 41 55 76 83 137 5 101"5´

Serrasalmus spp., rainy seasonBaıa Sia Mariana 188 222 255 315 374 3 24"4´ ´Barao de Melgaco 138 163 235 238 245 5 22"2˜Downstream Porto Jofre 53 114 155 240 530 13 23"2Caceres 310 378 2 24"0´Acurizal 48 71 82 114 187 9 21"3Baıa Paraıso 48 97 140 162 326 22 23"2´ ´

Serrasalmus spp., dry seasonBaıa Sia Mariana 282 355 423 813 2048 9 15"1´ ´Barao de Melgaco 324 432 475 540 711 8 26"2˜Downstream Porto Jofre 105 106 148 195 212 4 21"3

bCaceres 205 208 212 279 345 3 n.a.´Acurizal 115 173 224 332 463 17 23"4Baıa do Amolar 71 159 224 290 704 6 21"4´

All fishes, rainy season 47 104 152 238 724 96 86"19r22"3All fishes, dry season 41 182 302 403 2048 89 89"20r21"5All fishes, rainyqdry season 41 130 194 345 2048 185 87"19r22"4

a Fish from Caceres caught in May and October.´b Not analyzed.

( )L.D. Hylander et al. r The Science of the Total En¨ironment 261 2000 9]2016

Ž .Fig. 2. Total Hg concentration in muscle tissue of Pseudoplatystoma fasciatum cachara or surubim, striped or barred catfish ,Ž .Pseudoplatystoma coruscans pintado, spotted catfish , and the piranha species Serrasalmus spilopleura and Serrasalmus nattereri,

Ž . Ž . Ž .also Pygocentrus nattereri . Fish caught in March rainy season and in August dry season of 1998. Fish from Caceres caught in´May and October. Vertical bars indicate "1 S.D. and figures indicate N.

Table 4Ž . Ž .Total and methyl-mercury concentration average"S.D. encountered in fillet of Pseudoplatystoma fasciatum cachara and

Ž . Ž . Ž . Ž .Pseudoplatystoma coruscans pintado caught at three sites see Fig. 1 in the Alto Pantanal Brazil in March rainy season andŽ .August dry season 1998, weight and standard length of analyzed fishes

Sampling site Species Season n HgTot MeHg MeHg of Weight Lengthy1 y1Ž . Ž . Ž . Ž . Ž .ng g ng g HgTot % kg cm

Sia Mariana Cachara Dry 2 738"532 577"426 78"2 5.6"6.2 64"23´Sia Mariana Pintado Dry 1 403 343 85 1.8 54´B. Melgaco Pintado Dry 4 281"103 252"94 89"5 15.8"3.5 113"10Acurizal Pintado Dry 4 376"305 331"224 97"24 13.0"5.7 100"17Sia Mariana Pintado Rainy 4 437"312 416"299 95"5 4.9"1.1 78"6´B. Melgaco Pintado Rainy 4 103"50 89"40 90"19 5.9"1.4 80"5Acurizal Pintado Rainy 4 72"17 68"16 95"1 14.9"3.7 101"7

All, dry season 11 410"297 348"224 90"15 11.6"6.5 94"25All, rainy season 12 204"239 191"229 93"10 8.6"5.1 86"12Dryqrainy season 23 302"282 266"236 92"13 10.0"5.9 90"20

( )L.D. Hylander et al. r The Science of the Total En¨ironment 261 2000 9]20 17

Ž .bioavailability of Hg Meili, 1991, 1997 . Silva-Ž .Forsberg et al. 1999 reported increased fish Hg

levels at lower pH and at increased levels ofdissolved organic carbon in the Rio Negro basin,Amazon. The HgTot concentration found inMarch was highest in P. coruscans caught in BaıaSia Mariana. In August only one individual of this´specie was caught, which had a HgTot concentra-tion of the same level as the average for the rainy

Ž y1 .season 400 ng HgTot g fw . The size of thisŽ .individual was quite small 1.75 kg and it is

expected that larger individuals, generally usedfor consumption, would have a higher HgTotconcentration in parity with that of P. fasciatum

Ž .and piranhas from the same site Fig. 2 . InAugust, there was a visible boundary between

Žwhite-water from an adjacent lake Baıa Cha-´.corore , entering to the southern part of Baıa Sia´ ´ ´

Mariana, and soft-water in the northern part ofBaıa Sia Mariana, received from its main tribu-´ ´tary Rio Mutum. This complicates the evaluationof the influence of the soft-water, although thepiranhas were caught in the soft-water part of thelake. However, the water color differed amongdifferent creeks also in that part, possibly as aneffect of organic acids released during the decom-position of organic matter including the hugemats of water hyacinths. The piranha with the

Ž y1 .highest HgTot concentration 2048 ng Hg g fwwas caught in a creek with darker water thanfurther out in the lake. The floating vegetationmats also increase the net methylation of Hg in

Ž .their root zone Guimaraes et al., 2000 .˜The poor mixing of water makes the Hg con-

centration in fish with restricted mobility varyconsiderably within the lakes, over distances asshort as a few hundred meters. The HgTot con-centration among individuals from the same sitevaried more in the present study compared to a

Žsimilar survey performed in September 1992 Hy-.lander et al., 1994 , when all fish were caught by

the researchers in geographically restricted areasalong the river courses and not in any lakesvisited in the present study. This may have con-tributed to lower variation between fish caught atthe same site in 1992. The previous finding thatthe HgTot concentration of fish increased whenmoving downstream Rio Cuiaba from its sources´

at the Guimaraes Plateau down to Porto Jofre,˜the sampling point furthest downstream, raisedthe question whether even higher fish Hg concen-tration would be found further downstream. Thepresent study indicates that no further increase offish Hg concentration occurs downstream PortoJofre. The present study also indicates that theHgTot concentration in fish around Barao de˜Melgaco, both in Rio Cuiaba and in the adjacent´lakes, are higher than in fish caught in RioParaguai between Caceres and the confluence´with Rio Cuiaba.´

Mercury in sewage and day water from Cuiabamay contribute to higher Hg concentration in fishin Rio Cuiaba than in Rio Paraguai. Cuiaba is a´ ´

Žlarge city 624 000 inhabitants including Varzea.Grande with many potential Hg contamination

sources like gold retailers, dentists, thermome-ters, electrical instruments, dry cells, button cells,etc. Caceres, on the other hand is a small town´Ž .less than 80 000 inhabitants , without any major

Ž .industries IBGE, 1997 . In addition the waterflow in Rio Paraguai is approximately three timesthat in Rio Cuiaba, resulting in a larger dilution.´Leaching of Hg from abandoned and still activegold mining sites along Rio Cuiaba may also´contribute to the higher HgTot levels.

Another possibility to explain the higher HgTotconcentration in fish caught around Barao de˜Melgaco could be the larger load of suspendedparticulate matter in Rio Cuiaba than in Rio´Paraguai, due to erosion caused by agriculture,dredging and clear cutting of the riverbanksŽ .Goncalves, 1996 . Mercury is absorbed to this

Žmaterial Roulet et al., 1998a; Hylander et al.,. Ž2000 , and may be released to the water Roulet

.et al., 1998b and transformed to bioavailableŽ .MeHg e.g. Guimaraes et al., 2000 , in particular˜

in waters rich in dissolved organic matter butpoor in dissolved inorganic matter. Release of Hgabsorbed to soils and particulate matter is en-

Žhanced by organic acids Meili, 1991, 1997; Fors-.berg et al., 1999; Nakazono et al., 1999 . On the

other hand, enhanced soil erosion may also dilutethe Hg in relevant particulate phases. Finally, thefate of atmospheric Hg deposition differs spatiallyand temporally by directly entering large watersurfaces during the rainy season and remaining in

( )L.D. Hylander et al. r The Science of the Total En¨ironment 261 2000 9]2018

the restricted water bodies when they shrink dur-ing the dry season.

The seasonal differences in fish HgTot levelsfound in Alto Pantanal are not reported from the

ŽAmazon basin Kehrig and Malm, 1999a; Roulet.et al., 1999 , where fish HgTot levels often are

Ž .higher Malm, 1998 . Geochemical characteristicsof the Amazon differ from those of Pantanal, e.g.Hg-totrAl and Hg-totrFe ratios were lower in

Ž .Pantanal Hylander et al., 2000 than those pre-Ž .sented by Roulet et al. 1998a for soils and

sediments in the Tapajos basin in the Amazon.´This suggests that the Hg cycle in the Pantanalmay be governed by other parameters than in theAmazon basin, where the soils are younger and ofdifferent origin than in Pantanal. This may alsoinfluence the seasonal changes.

Further studies are needed to study the reasonsto fish HgTot fluctuations during the year inPantanal. Increased information about when thefish HgTot levels peak and the duration of highlevels is also needed. This could be performed bylocal authorities on a restricted budget with a

Ž .robust methodology Yallouz and Campos, 1999 .Cost advice should then be issued on receivedresults, with special attention paid to pregnant orbreast-feeding women, who should not consumeindicated species at seasons with expected HgTotlevels above the limit where mental retardation of

Žthe fetus or child is probable Grandjean et al.,.1999a,b .

6. Conclusions

The HgTot concentration in fish caught in andnear the upstream area of Rio Cuiaba around´Barao de Melgaco was higher than in fish from˜Rio Cuiaba downstream Porto Jofre or from Rio´Paraguai. Highest mean values were recorded inthe acid soft-water lake Baıa Sia Mariana. Fish´ ´caught in the river course during the dry season,and with a low probability of originating from theacid soft-water area, also contained HgTot con-centration often exceeding the limit for humanconsumption. The encountered HgTot concentra-tion was not as high as in an earlier alarming

Ž .report Vieira, 1991 , but in fish from soft-waterareas they were often above the safety limit of500 ng HgTot gy1 fw. Fish with HgTot levelsabove the safety limit were found also in white-water areas during the dry season. Relevant stepsshould be taken to reduce the fish Hg concentra-tion in order to minimize health impacts as wellas negative effects on the fishing and wildlifetourism that has become essential to the localeconomy in the Pantanal.

Acknowledgements

ŽWork supported by SidarSAREC SwedishInternational Development Co-operation

.AgencyrDepartment for Research Co-operationand the Federal University of Mato Grosso, Brazil.We want to thank all institutions and individualswho have contributed to this study, in particularEcotropica for logistical support, Marco Maya for

Žboats and lodging, Angelika Junker, AMEC As-¨.sociacao Ecologica Melgassense , the Coopera-˜ ´ ´

tives of fishermen in Barao de Melgaco and in˜Cuiaba, and Mario for fish samples, Paulo Al-´varez for boat, fish and guidance, Projeto GranPantanal at UFMT for boat, FEMA and IBAMAfor permits and fishing net, and Helena Kehrigfor analysis of MeHg.

References

Agemian H, Cheam V. Simultaneous extraction of mercuryand arsenic from fish tissues, and an automated determina-tion of arsenic by atomic absorption spectrometry. AnalChim Acta 1978;101:193]197.

Akagi H, Nishimura H. Speciation of mercury in the environ-ment. In: Suzuki T, Nobumassa I, Clarkson TW, editors.Advances in mercury toxicology. New York, USA: Plenum,1991:53]76.

Alho CJR, Vieira LM. Fish and wildlife resources in thePantanal wetlands of Brazil and potential disturbancesfrom the release of environmental contaminants. EnvironToxicol Chem 1997;16:71]74.

Alho CJR, Lacher TE, Goncalves HC. Environmental degra-Ž .dation in the Pantanal ecosystem. BioScience 1988;38 3 :

164]171.Bastos WR. Metodos de digestao utilizando microondas para´ ˜

determinacao automatizada de Hg em amostras ambientais˜

( )L.D. Hylander et al. r The Science of the Total En¨ironment 261 2000 9]20 19

e humanas: implantacao de laboratorios e avaliacao da˜ ´ ˜Ž .qualidade analıtica in Portuguese . MSc thesis, Universi-´

dade Federal do Rio de Janeiro, Rio de Janeiro, Brazil,1997:102 pp.

Bastos WR, Malm O, Pfeiffer WC, Cleary D. Establishmentand analytical quality control of laboratories for Hg de-termination in biological and geological samples in the

Ž .Amazon, Brazil. Cienc Cultura 1998;50 4 :255]260.Brasil Ministerio da Saude, Resolucao No. 18r75 da Comissao´ ´ ˜ ˜

ŽNacional de Normas e Padroes para Alimentos in Por-˜.tuguese . In: Diario Oficial da Uniao. Brasilia 19 de´ ˜

Dezembro de 1975. Secao 1, p. 16378.˜Estudos hidrologicos da bacıa do Alto Paraguai, Ministerio do´ ´ ´

Interior, Departamento Nacional de Obras de Saneamento.ŽRel. Tec. UNESCOrPNUD. BRA-66.521. Vol. I in Por-´

.tuguese . Brasılia, Brazil, 1974:284 pp.´Fitzgerald WF, Engstrom DR, Mason RP, Nater EA. The case

for atmospheric mercury contamination in remote areas.Environ Sci Technol 1998;32:1]7.

Forsberg BR, Jardim WF, Zeidemann VK et al. The bio-Žgeochemistry of mercury in the Negro river basin Brazilian

.Amazon . In: Barbosa J, Melamed R, Villas Boas R, edi-ˆtors. Mercury as a global pollutant } Fifth InternationalConference, 23]27 May 1999, Rio de Janeiro, Brazil. Riode Janeiro, Brazil: CETEM-Center for Mineral Tech-nology, 1999:153.

Goncalves LL. Diagnostico das influencias hidro-´ ˆsedimentologicas das Sub Bacıas dos Rio Arica-Acu e´ ´ ´ ´Arica-Mirim na Bacıa do Cuiaba River, no trecho Com-´ ´ ´preendido entre Santo Antonio do Leverger-Barao Melgacoˆ ˜Ž .in Portuguese . Cuiaba, Brazil: UFMT. Depto Engenharia´Sanitaria e Ambiental, 1996:134 pp.qannex.´

Grandjean P, Budtz-Jørgensen E, White RF et al. Methylmer-cury exposure biomarkers as indicators of neurotoxicity inchildren. In: Barbosa J, Melamed R, Villas Boas R, editors.ˆMercury as a global pollutant } Fifth International Con-ference, 23]27 May 1999, Rio de Janeiro, Brazil. Rio deJaneiro, Brazil: CETEM-Center for Mineral Technology,1999a:414.

Grandjean P, White RF, Nielsen A, Cleary D, Santos E.Neurobehavioral deficits in Amazonian children exposed tomethylmercury. In: Barbosa J, Melamed R, Villas Boas R,ˆeditors. Mercury as a global pollutant } Fifth Internatio-nal Conference, 23]27 May 1999, Rio de Janeiro, Brazil.Rio de Janeiro, Brazil: CETEM-Center for Mineral Tech-nology, 1999b:415.

Guimaraes JRD, Meili, M., Hylander LD et al. Net mercury˜methylation in five tropical flood plain regions of Brazil:high in the root zone of floating macrophyte mats but lowin surface sediments and flooded soils. Sci Total Environ

Ž .2000 in press .Hamilton SK, Sippel SJ, Melack JM. Inundation patterns in

the Pantanal wetland of South America determined frompassive microwave remote sensing. Arch Hydrobiol1996;137:1]23.

Hylander LD, Silva EC, Oliveira LJ, Silva SA, Kuntze EK,

Silva DX. Mercury levels in Alto Pantanal } a screeningŽ .study. Ambio 1994;23 8 :478]484.

Hylander LD, Meili M, Oliveira LJ et al. Relationship ofmercury with aluminum, iron and manganese oxy-hydrox-ides in sediments from the Alto Pantanal, Brazil. Sci Total

Ž .Environ 2000 in press .IBGE, Instituto Brasileira de Geografia e Estatıstica, Censo´

Demograficor1991. In: Anuario Estatıstico e Agroindus-´ ´ ´Ž .trial de Mato Grosso in Portuguese . Empresa Mato-gros-

sense de Pesquisa, Assistencia e Extensao Rural-EM-ˆ ˜Ž .PAER-MT. Ano 1 1978]1996 , Cuiaba-MT, Brazil,´

1997:429 pp.Kehrig HA, Malm O. Methylmercury in fish from Brazilian

Amazon. In: Barbosa J, Melamed R, Villas Boas R, editors.ˆMercury as a global pollutant } Fifth International Con-ference, 23]27 May 1999, Rio de Janeiro, Brazil. Rio deJaneiro, Brazil: CETEM-Center for Mineral Technology,1999a:208.

Kehrig HD, Malm O. Methylmercury in fish as a tool forunderstanding the Amazon mercury contamination. ApplOrganometal Chem 1999b;13:689]696.

Lacerda LD. Global mercury emissions from gold and silvermining. Water Air Soil Pollut 1997;97:209]221.

Lacerda LD, Pfeiffer WC, Bastos WR. Mercury dispersal inwater and sediments draining a gold mining tailing depositin Pocone, Brazil. Univ. Fed. Fluminense and Univ. Fed. do´Rio de Janeiro, Brazil, 1990.

Lacerda LD, Salomons W, Pfeiffer WC. Mercury distributionin sediment profiles from lakes of the high Pantanal, MatoGrosso State, Brazil. Biogeochemistry 1991;14:91]97.

Malm O. Gold mining as a source of mercury exposure in theBrazilian Amazon. Environ Res 1998;77:73]78.

Malm O, Branches FJP, Akagi H et al. Mercury and meth-ylmercury in fish and human hair from the Tapajos riverbasin, Brazil. Sci Total Environ 1995;175:141]150.

Mattos JP. Industria do Turismo como Fonte de Renda´Ž . Ž .Municıpio de Barao de Melgaco-MT in Portuguese . BSc´ ˜thesis. Universidade Federal de Mato Grosso, Departa-mento de Ciencias Contabeis, Cuiaba, Brazil, 1997.ˆ ´ ´

Meili M. Mercury in boreal forest lake ecosystems. ActaUniversitatis Upsaliensis, Comprehensive Summaries ofUppsala Dissertations from the Faculty of Science, Vol.336. Sweden: Uppsala University, 1991.

Meili M. Mercury in lakes and rivers. In: Sigel A, Sigel H,editors. Mercury and its effects on environment and bi-ology. Metal ions in biological systems, vol. 34. New York,USA: Marcel Dekker Inc, 1997:21]51.

Mikac N, Picer M, Stegnar P, Tusektnidarit. Mercury distribu-tion in a polluted marine area: ratio of total mercury,methyl mercury and selenium in sediments, mussels andfish. Water Res 1985;19:1387]1392.

Nakazono EM, Forsberg BR, Martins AE, Padovani CR.Factors controlling mercury contamination in tucunaresŽ .Cichla sp. from the Negro and Tapajos river basins,´Brazilian Amazon. In: Barbosa J, Melamed R, Villas Boasˆ

( )L.D. Hylander et al. r The Science of the Total En¨ironment 261 2000 9]2020

R, editors. Mercury as a global pollutant } Fifth Interna-tional Conference, 23]27 May 1999, Rio de Janeiro, Brazil.Rio de Janeiro, Brazil: CETEM-Center for Mineral Tech-nology, 1999:212.

Nriagu JO. Mercury pollution from silver mining in colonialSouth America. In: Abrao J, Wasserman JC, Silva-Filho˜EV, editors. Proc. Int. Symp. Perspect. Environ. Geochem.Trop. Countries, Niteroi, Brazil, 1993:365]368.´

Rodrigues Filho S. Metais pesados nas sub-bacıas hidrograficas´ ´Žde Pocone e Alta Floresta in Portuguese; English sum-´

.mary . Serie Tecnologia Ambiental 10, CETEMrCNPq,´Rio de Janeiro, Brazil, 1995:90 pp.

Roulet M, Lucotte M, Saint-Aubin A et al. The geochemistryof mercury in central Amazonian soils developed on theAlter-do-Chao formation of the lower Tapajos River Val-˜ ´ley, Para state, Brazil. Sci Total Environ 1998a;223:1]24.´

Roulet M, Lucotte M, Canuel R et al. Distribution andpartition of mercury in waters of the Tapajos River basin,´Brazilian Amazon. Sci Total Environ 1998b;213:203]211.

Roulet M, Lucotte M, Dolbec J et al. Mercury bioaccumula-tion in two fish communities of the Tapajos River, Brazilian´Amazon. In: Barbosa J, Melamed R, Villas Boas R, editors.ˆMercury as a global pollutant } Fifth International Con-ference, 23]27 May 1999, Rio de Janeiro, Brazil. Rio deJaneiro, Brazil: CETEM-Center for Mineral Technology,1999:232.

Sazima I, Machado FA. Underwater observations of piranhasin western Brazil. Environ Biol Fishes 1990;28:17]31.

Silva-Forsberg MC, Forsberg BR, Zeidemann VK. Mercurycontamination in humans linked to river chemistry in the

Ž .Amazon basin. Ambio 1999;28 6 :519]521.von Tumpling Jr W, Wilken R-D, Einax J. Mercury contami-¨

nation in the northern Pantanal region Mato Grosso, Brazil.J Geochem Explor 1995;52:127]134.

Veiga MM, Fernandes FRC, Farid LH et al. Pocone: Um´Žcampo de estudos do impacto ambiental do garimpo in

.Portuguese . Tecnologia Ambiental 1, CETEMrCNPq, Riode Janeiro, Brazil, 1991:113 pp.

Veiga MM, Meech JA, Hypolito R. Educational measures toaddress mercury pollution from gold-mining activities inthe Amazon. Ambio 1995;24:371]375.

Vieira LM. Avaliacao dos nıveis de mercurio na cadeia trofica˜ ´ ´ ´como indicador de sua biomagnificacao em ambientes˜

Ž .aquaticos da regiao do Pantanal in Portuguese . PhD´ ˜thesis. Univ. Fed. de Sao Carlos, Sao Paulo, Brazil, 1991.˜ ˜

Yallouz AY, Campos RC. A low-cost semiquantitative methodto determine mercury in fish and sediments samples. In:Barbosa J, Melamed R, Villas Boas R, editors. Mercury asˆa global pollutant } Fifth International Conference, 23]27May 1999, Rio de Janeiro, Brazil. Rio de Janeiro, Brazil:CETEM-Center for Mineral Technology, 1999:33.

Zeilhofer P. Geookologische Charakterisierung des nordlichen¨ ¨Pantanal von Mato Grosso, Brasilien, anhand multitem-

Ž .poraler Landsat Thematic Mapper-Daten in German .Munchen: Herbert Utz Verlag Wissenschaft, 1996.¨