Emissions and environmental implications of mercury from artisanal gold mining in north Sulawesi, Indonesia

The Science of the Total Environment 302(2003) 227–236

0048-9697/03/$ - see front matter� 2002 Elsevier Science B.V. All rights reserved.PII: S0048-9697Ž02.00397-2

Emissions and environmental implications of mercury fromartisanal gold mining in north Sulawesi, Indonesia

Daniel Limbong , Jeims Kumampung , Joice Rimper , Takaomi Arai *,a a a b,

Nobuyuki Miyazakib

Faculty of Fisheries and Marine Science, Sam Ratulangi University, Manado 95115, Indonesiaa

Otsuchi Marine Research Center, Ocean Research Institute, The University of Tokyo, Akahama, Otsuchi, Iwate 028-1102, Japanb

Received 1 June 2002; accepted 11 September 2002

Abstract

In artisanal gold mining practiced in North Sulawesi Island, Indonesia, gold is separated from ore by the use ofmercury, which forms an amalgam with gold. All related processes are undertaken with a low level of technicalknowledge and skills, no regulation, and with disregard for the safety of human and environment health. The situationis generating serious potential health and environmental risks in the area. As part of an ongoing monitoring program,total mercury concentrations were examined in water, bottom sediment and fish samples from three main rivers inTalawaan Watershed, which receives drainage from gold mining practices. Monitoring began in May–June 2000,almost 2 years after artisanal gold mining had begun. At that time, the mercury concentration in the sediment wasgenerally low, except in places close to the gold processing plants. In the present study, a more systematic samplingand analysis was conducted in May–June 2001. Bottom surface sediments, water, and fish samples were collected at12 sites along the three main rivers in the watershed. In addition, one site outside the watershed was sampled toserve as a control. Sample collections were conducted in three phases in duplicate, with two-week intervals betweeneach phase. The mercury concentration observed in this study indicated that an increase took place along the threemain rivers in the watershed. Solutions to this problem must be formulated as soon as possible in order to avoid amajor health, economic, and ecological disaster arising from the continuing discharge of Hg. The present studyproposes that mercury dispersion occur downstream of the mining.� 2002 Elsevier Science B.V. All rights reserved.

Keywords: Mercury; Bioaccumulation; Food chain; Artisanal Gold mining; Indonesia

1. Introduction

Artisanal gold mining is an activity that reliesmainly on manual labor and makes use of simplemethods. It offers poor people an important meansof livelihood and has served as a safety net in

*Corresponding author. Tel.:q81-193-42-5611; fax:q81-193-42-3715.

E-mail address: [email protected](T. Arai).

times of economic distress, especially during theextended economic crisis of Indonesia of the lastfive years. Approximately 10 000 workers areactively involved in artisanal mining and process-ing of gold deposits in the North Sulawesi Prov-ince of Indonesia. This relatively large numberindicates the great importance of artisanal goldmining as an employment opportunity in this ruralarea.

228 D. Limbong et al. / The Science of the Total Environment 302 (2003) 227–236

Over 100 years ago, the Dutch complained ofartisanal gold mining in the nearby Ratatotokregion in the North Sulawesi Province of Indone-sia, and illegal miners still operate in that region.In 1997, the awarding of a gold mine concessionto the Aurora Mining Co. of Australia in theDimembe Sub-district in the North Sulawesi Prov-ince, northeast of Manado City, gave rise to a goldrush of artisanal miners to the area and this rushhas expanded to include thousands of miners.The amalgamation method of artisanal gold

mining causes mercury(Hg) emissions to leakinto the environment in several different ways. Forexample, when Hg is unintentionally spilled ontothe ground. Atmospheric transport and depositionat normal temperature is the pathway deliveringHg to many of the word’s rivers, lakes and oceans.Moreover, Hg is often discharged together withother wastes into inadequate tailings ponds, or isdisposed of directly into rivers and waterways. Yetanother means of introducing Hg into the environ-ment take place when purifying the amalgam byburning and vaporized Hg is released into theatmosphere.For several decades, Hg has been recognized as

an environmental pollutant(Akagi and Nishimura,1991; Clark, 1997). The Hg pollution problem inthe Talawaan Watershed has generated severalpotential threats to the province to date(Martens,2000). The immediate health threat of the use ofHg for gold extraction affects those who work orlive in areas around processing plants. As regardswater pollution, some of the metallic Hg dis-charged into rivers and waterways is transformedinto methyl Hg, by microorganisms then eaten byaquatic species, which are in turn consumed byhumans. Like bioaccumulation of many environ-mental contaminants, that of Hg accumulates alongthe food chain of aquatic organisms(Lodenius andMalm, 1998; Veiga, et al., 1999). Fish and otherwildlife from various ecosystems commonly attainHg levels of toxicological concern when directlyaffected by Hg emissions from human-initiatedactivities (Samoiloff, 1989). Solutions to theseproblems must be implemented as soon as possiblein order to avoid a major health, economic, andecological disaster arising from the continued dis-charge of Hg.

This study addresses the question of whether ornot variations in Hg concentration in river systemscan reflect the influence of Hg emission from goldprocessing plants using amalgamation. Weassessed the levels of Hg concentration in water,sediment, and biological samples such as fishes ina watershed of North Sulawesi Island, Indonesia.

2. Materials and methods

2.1. Study area

The Talawaan Watershed drains from the peakof Mount Klabat into the western coast of thepeninsula of North Sulawesi(Figs. 1 and 2). Thehighest point of this watershed is the peak ofMount Klabat (1995 m), and the total area isestimated to be 34 400 ha(Martens, 2000). Thedistance from the peak of Klabat to the sea isapproximately 20 km. The area includes the drain-age basins of the Talawaan, Kima, and BailangRivers. Talawaan and Bailang Rivers flow throughthe main center of the mining area(Figs. 1 and2).Land use in the Talawaan Watershed is primarily

agricultural and is dominated by plantations ofcoconut, clove, and nutmeg. There are also asso-ciated areas of irrigated rice cultivation and fish-ponds. Cattle, pigs, goats, chickens and ducks, arereared in the region. There is no important industrylocated along the banks of the three main rivers,which potentially exaggerates the effects of theHg pollutant (Martens, 2000). The coastal envi-ronment of the watershed includes mangrove areasand coral reefs. Fishing is carried out in the coastalareas, and crabs and molluscs are also collected inthe area for human consumption. There is also asmall aquaculture around brackish water in thearea.The population of the study area is estimated to

be approximately 150 000(Martens, 2000). Themajority of households in the area are dependentupon agriculture as their main source of incomeand sustenance, but the number of individualsinvolved in gold mining has increased rapidly since1998. At present, approximately ten thousand indi-viduals are estimated being directly involved inthe mining activities. Most of those involved in

229D. Limbong et al. / The Science of the Total Environment 302 (2003) 227–236

Fig.1.Study

area

inNorthSulawesiIsland,Indonesia.

230 D. Limbong et al. / The Science of the Total Environment 302 (2003) 227–236

Fig. 2. Sampling sites in the Talawaan watershed, North Sulawesi Island, Indonesia.

mining have come from outside of Dimembe sub-district (Talawaan Watershed).The mining areas are located in the villages of

Tatelu, Warukapas, Rondor, and Talawaan of theDimembe sub-district. By June 2001, there wereapproximately 400 gold processing plants in thearea. The processing plants are mostly built at theside of the river, because water is necessary forthe processing of ore. As depicted in Fig. 2, theprocessing plants are primarily located in the upperpart of the watershed. This mining area is moreaccessible than other mining areas in the NorthSulawesi Province, because it is located in anagricultural area near the villages, and is also closeto Manado, the capital of the province. This maybe the main reason that many experienced minersfrom other mining sites of the province come to

seek their fortune in the Dimembe gold mine area(Martens, 2000).

2.2. Gold processing plants as a possible pollutionsource

The mining process begins in vertical shaft andtunnels, which are dug by hand to a maximumdepth of approximately 30 m, where gold-bearingdeposits are mined. Crude assay methods are usedto estimate gold recovery and to guide the directionof underground excavation. The ore pulled out ofthese shafts is then packed in sacks and transport-ed, usually by oxen, to a processing plant. In theprocessing plant, the first step is to crush the orein order to reduce the processing time at the mills.This is usually done manually using hammers and,

231D. Limbong et al. / The Science of the Total Environment 302 (2003) 227–236

as platforms, large rocks or iron blocks are used.Some processing plants now use a homemademechanical crushers to reduce manual labor.Crushed ore fed into a facility for grinding theore. Approximately 30–40 kg of ore are loadedinto a ball mill per batch. Water, hard rock, andsometimes lime are added into the ball mill andgrinding commences to break down the ore to finesand, thus releasing gold grains. After it has rotatedfor 3 to 4 h, it is turned off and 1 kg or more Hgis added. Then, the mill is rotated for another halfan hour for amalgamation, whereby fine gold willattach itself to the Hg. Afterwards, water is chan-neled into the ball mill for a period of time inorder to pour out the slurry.All mill content is then dumped into a large

basin, and the heavy metal alloy is allowed tosettle down to the bottom. Water is again addedto remove the slurry, leaving behind the amalga-mated Hg through the panning process. This isthen collected and placed in a fine-woven cloth,and by twisting the cloth the excess Hg is separatedfrom amalgam, which remains in the cloth. Thus,the amalgam still contains major of Hg. To separatethe gold from Hg, the amalgam is then heated ina circular clay pot in order to vaporize the Hg.Borax is added to the burning amalgam as acleaning material for the removal of impuritiesfrom the final product, which is gold bullion.Amalgamation is popular in artisanal gold min-

ing since it is a simple process and it requires onlylow investment. The production of gold Hg amal-gam through manual manipulation leads to theinitial contamination of the environment.

2.3. Sampling, preparation and total mercuryanalysis

Twelve sampling sites were selected: five siteswere located along the Talawaan River, four at theBailang River, and three Kima River. One locationoutside of the watershed at Poigar River wasserving as a control site. The locations are illus-trated in Fig. 1, and details regarding those siteswithin the Talawaan Watershed are depicted inFig. 2. Sites T1 and B1 were upper most sites ofthe Talawaan and Bailang Rivers. These points arelocated upstream of the gold processing plants.

Samples from these points and from the controlsite at the Poigar River might reflect the Hgbackground levels in the area. The other samplingpoints were located downstream of the gold proc-essing plants. Water, sediment, and fish sampleswere collected at the sampling sites. Sample col-lection was conducted at three times and in dupli-cate except for fish sample in order to account forany short-term fluctuations within the watershed.Two-week intervals between three observations ateach site were maintained whenever possible. Thefirst phase was successfully carried out duringMay 1–12, 2001, and the second phase from May14 to 22, 2001. The last phase took place betweenMay 29 and June 6, 2001.Water samples were taken from the main stream

at each sampling site. A pre-labeled and precleaned250 ml Nalgene polyethylene bottle was immersedapproximately 10 cm below the water surface, andthen its cap was opened for 1–2 min in order toallow for water overflow to ensure the eliminationof air bubbles. Approximately 2 ml of HNO was3

added into each sample bottle before the cap wasinserted in order to preserve the water.Sediment samples were taken down to 10 cm

below the sediment surface using a sediment coresampler. For each site, sediment samples weretaken from three representative points, then weremixed and packaged as a composite sample in apre-labeled 250 ml plastic sample bottle.Fish samples at each sampling site were then

packed in pre-labeled plastic bags. Total lengthsof the fish samples were measured, and speciesidentified. Fish were analyzed by dissecting muscletissue from each individual. When many fish werecaught, or when the size of fish was large enough,then a single species sample was made. When theamount of tissues was too small, composite sam-ples were used.

2.4. Mercury analysis

Water samples were filtered through a disposa-ble 0.45mm polycarbonate portable filtration unit.After filtration, water samples were treated in thecold-oxidation of acidified samples using brominemonochloride prior to the reduction of the sampleswith stannous chloride. Sediment samples were

232 D. Limbong et al. / The Science of the Total Environment 302 (2003) 227–236

digested using a 1:1 ratio of nitric acid andhydrochloric acid, along with hotplate or micro-wave heating. Fish samples were digested usingnitric acid and hydrogen peroxide along withhotplate heating, and the Hg concentrations of 13fish species were examined. Total Hg for allsamples were determined by the cold vapor tech-nique using a Sansou Automatic Mercury AnalyzerModel HG-3000, according to the method sug-gested by Akagi and Nishimura(1991). Detectionlimits for water, sediment, and fish samples were10 mgyl, 1 mgykg dry weight, and 1mgykgy2

wet weight, respectively. Quality control was mon-itored for all chemical analyses. Instrument cali-bration was verified by analyzing certifiedcalibration solutions during each instrumental run.These external reference standards were generallywithin 92 to 103% of the nominal concentrations.All of the sample spikes for water, sediment, andfish samples were within 90 to 113% recovery.Preparation blanks were prepared to detect poten-tial contamination during the digestion procedure.These preparation blanks generally measuredbelow the detection limit.

3. Results and discussion

3.1. Distribution of mercury in the river system

Hg levels among the three sampling phases inthe year 2001 were variable at sampling sites T2,T3, T4, T5(Talawaan River), B2, B3, B4(BailangRiver), and K1(Kima River), as shown in Fig. 3and Fig. 4. Vastly fluctuating Hg levels at onesampling site within a relative short time intervalmay reflect the influence of respective environ-mental conditions such as dilution rate, discharge,effluent mixing, and other weather-related effects.In such case, the average of a sequence observationprovided a more representative picture. Comparedto the concentrations of Hg in sediments, those inwater were very low. This finding suggests thatHg rapidly attaches to suspended solid particlesand organic matter, and that Hg thus can rapidlyslump to the river bottom. The US EPA mercurywater quality criterion for protection of freshwaterin 12 ng Hgyl, and for seawater it is 100 ngyl(USEPA 1985). The water quality criterion for

mercury proposed for Minnesota’s freshwater is 7ngyl while a value of 2 ngyl has been establishedfor Wisconsin waters(Glass et al., 1990). Themercury concentrations found in this study werealmost higher than both the US EPA tolerancelimit and those established in Minnesota andWisconsin.The increased Hg level both in water and sedi-

ment samples at sites T2 and B2, with subsequentvariations at high levels, were present into the restof the downstream sites; this result was interesting,because it clearly indicated existing influence ofthe gold processing plants. As shown in Fig. 2,sites T1 and B1 were located upstream, whereassites T2 and B2 were downstream of the processingplants. At the Kima River, gold processing plantswere located at the uppermost sampling site of theriver, and therefore the impact of a high Hg levelwas observed at site K1. There was little variationobserved at sites T1, B1 and P, as well as smallmean total Hg levels; both of these results indicatethat the sites are not affected by the processingplants.As depicted in Fig. 2, gold processing plants

are denser along the Talawaan River, followed bythe Bailang River, and the Kima River. This factis well correlated with the general Hg levels inwater and sediment at those rivers(Figs. 3 and 4).At almost sites of downstream of the processingplants along the Talawaan River(T2–T5), the Hglevels were more than three times over the safetylevel for drinking water(1 mgyl) (WHOyICPS,1990), and Hg levels exceeded threefold the safetylevel of 2 mgykg for sediment (WHOyICPS,1990). Hg levels in sediment at all sites below theprocessing plants at the Bailang River(B2–B4)were moderately high, whereas at the Kima River,rather low levels were found. These findings dem-onstrate that the emission of metallic Hg fromthese gold processing plants has been dispersed toa few downstream through the river system.A preliminary explanation of the accumulation

rate of the emitted Hg in the river system wasevaluated by comparing Hg levels in the sedimentafter monitoring the levels in the year 2000(Mar-tens, 2000) and in 2001 at sites T1, T2, T5, B2and K1 (Fig. 5). Site T1 that was free from theimpact of gold processing; however, a large

233D. Limbong et al. / The Science of the Total Environment 302 (2003) 227–236

Fig. 3. Total mercury in water samples from sampling sites along the three rivers in Talawaan Watershed and one site at PoigarRiver, Indonesia.

Fig. 4. Total mercury in sediment at sampling sites along the three rivers in the Talawaan Watershed and at one site at Poigar River,Indonesia.

increase in the Hg levels occurred within only oneyear at the other sites. Since there was no otherpotential source of Hg emissions in the study area,the increasing Hg levels were attributed to theexisting gold processing plants. Attention shouldbe given to the estuary of the Talawaan River(T5), where Hg levels in the sediment reached

approximately 7 times that of the previous year’slevel.

3.2. Bioaccumulation and biomagnification

Bioaccumulation and biomagnification are con-sidered to be important factors in the overall

234 D. Limbong et al. / The Science of the Total Environment 302 (2003) 227–236

Fig. 5. Comparison of mercury levels in sediment at sampling sites in the TalawaanWatershed in the years 2000 and 2001.

Fig. 6. Mercury levels in fish muscle from the Talawaan Water-shed and Poigar River.

assessment of Hg pollutants, since these processesenhance the probability of long-term toxic effectsby contamination of human food sources(Akagiand Nishimura 1991; Clark, 1997). Food chain(biomagnification) can be several orders of mag-nitude higher than in ambient water. In order toindicate the rate of bioaccumulation and biomag-nification in the aquatic environment of the Tala-waan Watershed, the results of 47 fish musclesamples of 13 species were examined(Fig. 6).Nine of the samples contained high Hg levelswhich were up to 6.3 times that of the safety levelof the international human consumption advisorylimit of fish (0.5 mgykg wet weight) (WHOyICPS, 1990). All of the samples were from theTalawaan River, the most contaminated river of allof those included in the present study. The highestlevel (3.14 mgykg) was observed in a relativelysmall fish; total length, 5.3 cm,Caranx sp. (acarnivorous species) caught in the estuary.Sediment plays an important role in the bioac-

cumlation process(Samoiloff, 1989; Akagi andNishimura, 1991). A clear indication of the processwas found in this study by plotting the mean levelsof Hg contents in fish muscle at each samplingsite against their counterpart of mean Hg levels inthe sediments(ys0.07xq0.09, rs0.748). It wasclearly observed that the Hg contents in the fish

muscle were positively correlated with those in thesediments. The increasing Hg levels in fish tissuetended to follow those in sediment samples. Thesefindings agree with those of Ikingura and Akagi(1996), who discovered that fish and other wildlifefrom various ecosystems commonly attain Hglevels of toxicological concern when directlyaffected by Hg emissions from human-related

235D. Limbong et al. / The Science of the Total Environment 302 (2003) 227–236

Fig. 7. Mercury levels in fish muscle samples according to foodchain groups.

activities. In order to provide some indication ofthe biomagnification process, all fish samples weredivided into three groups according to their foodhabit (Fig. 7). A trend of increasing Hg levels wasseen among the carnivorous species in the Tala-waan and Kima Rivers. The same phenomena wasobserved by Lodenius and Malm(1998), whoanalyzed fish in rivers and reservoirs impacted by

gold mining activities, namely, they found thehighest Hg concentrations in carnivorous fish,intermediate values in planktivorous and omnivo-rous fish, and the lowest concentrations in herbiv-orous fish. These considerations all led to theconclusion that Hg accumulation occurs along thefood chain and that increased Hg levels mightaffect human health in such areas.

4. Conclusions

The fluctuation of Hg levels in water and sedi-ment in relation to the sampling sites and goldprocessing plant locations within the TalawaanWatershed provide insight into the pathway of Hgdispersion from gold processing plants throughoutthe river system. Increasing Hg levels in fishsamples provided strong indication of a high bioac-cumulation within this contaminated area. Thepresent study has shown that environmental con-tamination by Hg from artisanal gold miningactivities is elevated and that Hg has accumulatedto acute levels. Therefore, reduction of Hg emis-sion from the processing plants is of immediateconcern. A regular monitoring program is neces-sary in order to better elucidate the rate of bioac-cumulation and biomagnification. Such a programwould also facilitate a more detailed risk assess-ment regarding human health issues.

References

Akagi H, Nishimura H. Speciation of mercury in the environ-ment. Toxicology. In: Suzuki T, Imura N, editors. Advancesin mercury. New York: Plenum Press, 1991. p. 53–76.

Clark RB. Marine Pollution. Fourth Edition. UK: OxfordUniversity Press, 1997. 161. pp.

Glass GE, Sorensen JA, Schmidt KW, Rapp GR. New sourceidentification of mercury contamination in the Great Lakes.Environ Sci Technol 1990;24:1059–1069.

Ikingura JR, Akagi H. Monitoring of fish and human exposureto mercury due to gold mining in the Lake Victoria Gold-fields, Tanzania. Sci Total Environ 1996;191:59–68.

Lodenius M, Malm O. Mercury in the Amazon. EnvironContam Toxicol 1998;157:25–52.

Martens H, Mercury Pollution Awareness and Control. A casestudy at small-scale gold mining sites in Ratatotok andTatelu Districts of North Sulawesi, Indonesia. CollaborativeEnvironmental Project in Indonesia(CEPI). 2000; 64 pp.

236 D. Limbong et al. / The Science of the Total Environment 302 (2003) 227–236

Samoiloff MR. Toxicity testing of sediments: Problems, trendsand solutions. In: Nriagu JO, Llakshminarayana JSS, editors.Aquatic Toxicology and water quality. Canada: John Wileyand Sons, Inc, 1989. p. 143–152.

USEPA, Water quality for Mercury. 440y5y84-026, Office ofWater, US Environmental Protection Agency, Washington,DC, 1985.

Veiga MM, Hinton J, Lilly C. Mercury in Amazon: AComprehensive Review with Special Emphasis on Bioac-cumulation and Bioindicators. Proceeding of National Insti-tute for Minamata Disease Japan. 1999: 19–39.

WHOyICPS, Environmental Health Criteria 101, Methylmer-cury. WHOyICPS, Geneva. 1990.