Ž .The Science of the Total Environment 228 1999 95]109

Mercury contamination associated with artisanal goldmining on the island of Mindanao, the Philippines

J.D. Appletona,U, T.M. Williamsa, N. Brewarda, A. Apostolb, J. Miguelb,C. Mirandab

aBritish Geological Sur ey, Keyworth, Nottingham NG12 5GG, UKbMines and Geosciences Bureau, Quezon City, Philippines

Received 27 October 1998; accepted 14 December 1998

Abstract

The Agusan River basin of eastern Mindanao, the Philippines, hosts several centres of artisanal gold mining, themost important of which, Diwalwal, is a significant gold producer in global terms. An investigation of theenvironmental impacts of artisanal mining in the Agusan system, with particular reference to mercury contamination,was initiated in 1995 following reports of several incidents of human Hg poisoning in the province of Davao delNorte. Results show drainage downstream of Diwalwal is characterised by extremely high levels of Hg both in

Ž . Ž .solution maximum 2906 mgrl and in bottom sediments )20 mgrkg . Filtered surface water Hg levels exceed theWHO Drinking Water guideline and the US-EPA Water Quality Criteria for the Protection of Aquatic Life for adownstream distance of more than 14 km, including channel sections utilised for fishing and potable water supply.The Environment Canada sediment quality Hg Toxic Effect Threshold for the Protection of Aquatic Life is exceededfor a downstream distance of 20 km. Hair Hg data indicate that ballmill and CIP plant operators processing Hgcontaminated tailings at eastern Mindanao’s principal gold beneficiation centre, Apokon, may be subject to enhancedoccupational Hg exposure. It appears that the wider population of this area has not been affected.Q 1999 NERC. Published by Elsevier Science B.V. All rights reserved.

Keywords: Mercury; Contamination; Artisanal gold mining; Mindanao; Philippines; ‘Dissolved’ mercury; Suspended particulatematter; Bottom sediment; Heavy mineral concentrates; Hair; Dispersion; Water quality; Sediment quality

U Corresponding author.

0048-9697r99r$ - see front matter Q 1999 NERC. Published by Elsevier Science B.V. All rights reserved.Ž .PII: S 0 0 4 8 - 9 6 9 7 9 9 0 0 0 1 6 - 9

( )J.D. Appleton et al. r The Science of the Total En¨ironment 228 1999 95]10996

1. Introduction

Amalgamation with mercury has been used as amethod of gold and silver beneficiation since Ro-man times. The total global release of Hg into theenvironment through this process prior to 1930has been estimated as over 260 000 t, after whichemissions declined with the introduction of

Žcyanidation processing technology Lacerda and.Salomons, 1998 . In the 1970s high gold prices

triggered a major artisanal gold rush in manycountries of the Southern Hemisphere, possiblyinvolving more than 10 million people. The atten-dant resurgence in the use of Hg for Au process-ing which occurred at this time has continuedlargely unabated to the present day, and maycurrently account for 10% of the total anthro-pogenic flux of Hg into the global environment.

Problems of Hg contamination arising fromgold mining in the Amazon River basin of Brazil

Žhave been widely documented e.g. Lacerda and.Salomons, 1998 , yet there remains relatively little

quantitative data for other major artisanal miningareas world-wide. Here the results are reported ofa study of the magnitude and extent of mining-re-lated Hg contamination of waters and sedimentsof the Agusan River Basin, eastern Mindanao,the Philippines, undertaken as part of a Tech-nology Development and Research Programmefunded by the UK Department for International

Ž .Development DFID and the Philippine Depart-ment of Environment and Natural ResourcesŽ .DENR . The potential toxicological significanceof Hg contamination is evaluated with referenceto water and sediment quality guidelines. Hair Hgdata are also presented to indicate human expo-sure among inhabitants of Apokon, eastern Min-danao’s principal centre for the processing of goldore.

2. Background

2.1. De¨elopment of artisanal mining acti ity

Several previously undocumented alluvial andbedrock gold occurrences were discovered by arti-sanal miners in eastern Mindanao during the1980s, prompting a series of largely uncontrolled

gold rushes and the development of several min-Žing communities and )100 000 inhabitants Fig.

.1 . Two of the most important gold mining locali-ties, Diwalwal and Mainit, are located within themountainous east Mindanao Ridge, from whichdeeply incised rivers flow north and west into thelow-lying terrain of the Agusan]Davao trough. AtDiwalwal, an epithermal system with a total Aureserve of several million ounces was discoveredby artisanal prospectors in 1983. At its peak inthe mid-1980s the Diwalwal community num-bered over 100 000. Since then, it has declined toapproximately 50 000 partly as a consequence ofmajor landslides in 1985 and 1988. The focus ofmining activity at Diwalwal lies in the headwatersof a series of minor tributaries of the Mamunga

Ž .River known locally as the Naboc River , whichsubsequently joins the north-flowing Agusan RiverŽ .Fig. 1 . Veins are worked via a series of adits anddrives, and the ore is crushed in rod mills. Muchof the processing activity during the 1980s in-volved amalgamation, but the predominance ofthis procedure has declined during the 1990s fol-lowing the installation of a co-operative carbon-

Ž .in-pulp CIP cyanidation plant. The Mainit goldmining district incorporates at least eight epither-mal systems and two auriferous porphyry and

Ž .stockwork deposits Mitchell and Leach, 1991 .Alluvial gold from Mainit is worked by artisanalminers using gravimetric methods, with subse-quent amalgamation of Au from heavy mineralconcentrates.

2.2. E¨idence of en¨ironmental impacts

The poorly-controlled use of Hg for gold amal-gamation in eastern Mindanao resulted in numer-ous poisoning incidents among miners during thelate 1980s, the first of which, involving 11 injuriesand one fatality, was reported from Tagum, Davao

Ž .del Norte, in 1987 Torres, 1992 . In 1988, theŽ .Philippine Mines and Geosciences Bureau MGB

observed the use of Hg for gold beneficiation in53 mining localities on Mindanao, and subse-quently estimated that some 140 t of Hg mayhave been released into the Agusan River catch-ment during the period 1986]1988.

Ž .Enhanced Hg levels up to 2.30 mgrkg wet wt.

( )J.D. Appleton et al. r The Science of the Total En¨ironment 228 1999 95]109 97

Fig. 1. Location of artisanal gold mining areas in eastern Mindanao, Philippines. Inset map shows the location of the main maparea within the Philippines; dashed box indicates area of Fig. 2 and Fig. 3; small filled circles indicate sampling sites on lower sectorof Agusan River.

have been reported for fish flesh from the AgusanRiver and tributary systems draining Diwalwaland Mainit, and from the estuarine outflow of the

Ž .Agusan at Butuan Bay Torres, 1992 . In Davaodel Norte, air Hg concentrations of 42]1664 mgmy3 were recorded in four different goldcommercialisation shops, with 65% of samples

Ž .exceeding the WHO 1976 industrial exposurey3 Ž .limit of 50 mg m Torres, 1992 . However, only

6% of 230 potentially exposed workers sampledŽ .by Torres 1994 showed elevated blood levels,

and all had urine Hg below 50 mg ly1. A survey in1996 by the Philippine Department of HealthŽ .DOH of blood Hg levels in Apokon residents

( )J.D. Appleton et al. r The Science of the Total En¨ironment 228 1999 95]10998

Žrevealed blood burdens of up to 20 ngrml Wil-.liams, 1997 , interpreted by the DOH as consis-

tent with excessive residential andror occupatio-nal exposure. Detoxification of residents with Hglevels )10 ngrml was proposed and instigated.While blood mercury concentrations in the range10]20 ngrml are common among many commu-

Žnities with a high dietary fish intake Paccagnellaand Pratti, 1974; Riodolfi, 1977; Health and Wel-fare of Canada, 1979; Lacerda and Salomons,

.1998 and fall below the critical threshold typi-cally specified as hazardous to human healthŽWHO, 1976; Piotrowski and Inskip, 1981; Lac-

.erda and Salomons, 1998 , national media cover-age of the DOH action in Apokon during 1996prompted public and political pressure for thecompensation of residents, and for the closure ofall gold processing operations.

3. Sample collection and analysis

3.1. Water and sediment samples

The Agusan basin survey reported here wasexecuted by staff of the British Geological SurveyŽ .BGS and the Philippine Mines and Geosciences

Ž .Bureau MGB . A reconnaissance study of sev-eral artisanal gold mining districts of eastern

ŽMindanao was undertaken in July 1995 Williams.et al., 1995 followed by systematic sampling of

the Agusan drainage basin during the periodŽNovember 1995 to February 1996 Breward et al.,.1996; Appleton and Williams, 1998 . Samples wereŽ .collected along a 200-km length approx. of the

Agusan River, plus several sub-catchments di-rectly impacted by contemporary mining and ap-propriate control stations in unmined tributaries.Stream water pH, temperature, Eh and conductiv-ity were determined in the field using a series oftemperature-compensated electrodes and meters.Water samples for chemical analysis were filteredthrough 25-mm diameter, 0.45-mm MilliporeTM

cellulose acetate membranes into 30-ml Sterilintubes or acid-washed 30-ml HPDE bottlesŽ TM .Nalgene . Coarse prefilters were used in con-junction with the 0.45-mm cellulose disks on allobviously turbid samples. At each site, the suite

Ž .of water samples collected included: a 30 ml

Ž .preserved with 1% vrv HNO ARISTAR for3determination of major and trace cations by in-ductively-coupled plasma emission spectrometryŽ . Ž . 2yICP-ES ; b 30 ml unacidified water for SO ,4NOy and Cly analysis by ion chromatography;3

Ž .and c 30 ml preserved with 0.3 ml conc. HNO3q0.3 ml 0.2 vol.% K CrO for total Hg analysis2 7by cold vapour atomic fluorescence spectroscopyŽ .CVAFS to a practical detection limit of 30 ngrl.

Ž .Bottom sediment BS samples, each of 100]200g, were collected by wet-screening of river orstream-bed detritus through a -150-mm sieve,using a minimal amount of water to avoid the lossof fine silt and clay fractions. Samples were sealedin securitainers to avoid evaporative losses and

Ž .oxidation. Suspended particulate matter SPMsamples were obtained by filtering a stream ofwater through 25-mm diameter, 0.45-mm Milli-poreTM cellulose filters. The filter membraneswere carefully removed to avoid contaminationand stored in 30-ml Sterilin tubes. Heavy mineral

Ž .concentrate HMC samples were collected byscreening approximately 2 kg of -2 mm streambottom sediment, and removing the lighter frac-tions in a conventional prospecting pan. Mercuryanalyses of BS and HMC samples were carriedout by CVAFS, using 1-g milled sub-samples,digested at -508C in aqua regia. SPM sampleswere dried and the sediment digested with thecellulose filter membrane in aqua regia at -508C.Mercury was then determined by CVAFS to apractical detection limit for solid samples of 0.02mgrkg. Major oxide and trace metals were de-termined for BS samples collected in the recon-

Ž .naissance study by X-ray fluorescence XRFanalysis of 12-g pellets made following disaggre-gation, ignition of organic matter and milling ofsediment to -63 mm.

3.2. Hair

Hair samples were collected from close to theŽ .scalp from 1 eight workers from a CIP plant

which processes tailings previously treated withŽ . Ž .Hg; 2 two ballmill operators using Hg; 3 a

blow-torch operator involved in burning amal-Ž .gam; 4 17 people living within 50 m of the CIP

Ž .plants; and 5 19 residents from four communi-

( )J.D. Appleton et al. r The Science of the Total En¨ironment 228 1999 95]109 99

ties located up to 3 km from the main goldprocessing plants at Apokon Tagum, Davao del

Ž .Norte Williams, 1997 . Samples of 2-g mass werewashed repeatedly in distilled water plus a surfac-tant to remove dust and other surficial contami-nants, which is similar to the preparation proce-dure used for a certified hair powder reference

Ž .material Okamoto et al., 1985 . Although a widerange of hair preparation methods have been

Ž .reported e.g. Chittleborough, 1980 , Cargnello etŽ .al. 1995 observed that no obvious differences

could be detected between five wash methodstested in their ability to remove surface debris.Whereas all methods produced reductions in sur-face residues, none was able to remove contami-nants fully. In this study, all hair samples were air

Ž .dried at -408C , cut into 1-cm lengths andŽ .digested sequentially in a 10 mg V O q5 ml2 5

Ž . Ž .HNO ARISTAR and b H SO for subse-3 2 4quent Hg analysis by CVAFS. Eight independentanalyses of certified hair reference material GSH-Ž .1 certified Hg concentration 0.36"0.05 mgrkg

fell in the range 0.32]0.34 mgrkg. Seven analysesŽof CRM BCR-397 certified Hg concentration

.12.3"0.5 mgrkg fell in the range 10.06]12.1mgrkg; indicating an average recovery of 92%across the entire concentration range.

4. Results and discussion

4.1. Mercury in filtered water

Extremely high ‘dissolved’ Hg concentrations ofŽup to 2900 mgrl were recorded in filtered -45

.mm water samples collected from the headwatersof Mamunga River tributaries draining the Diwal-

Ž .wal mining centre Fig. 2 . Concentrations declineŽ .rapidly along the 25-km approx. channel section

feeding the Agusan River, but remain at least twoorders of magnitude above the natural back-ground in the mid-reaches of the Mamunga ap-proximately 14 km downstream. A slightly en-hanced Hg concentration of 0.1 mgrl wasrecorded near the Mamunga]Agusan confluence.

The peak ‘dissolved’ Hg concentrationsrecorded near Diwalwal during both the recon-

Žnaissance study and the principal survey 2900.mgrl and 1540 mgrl, respectively are particularly

noteworthy given the inherent insolubility of mostinorganic Hg compounds in natural waters. Suchvalues are in excess of those documented for anyother mining-impacted waters world-wide. Datapublished for gold mining impacted rivers of Northand South America fall in the range 0.2]19.8

Žmgrl e.g. Pfeiffer et al., 1989, 1991; Lacerda and.Salomons, 1998 , while the typical natural back-

ground for unpolluted freshwaters is 0.003]0.005mgrl.

Ž .Williams et al. 1995 have reported high totalŽ .organic carbon TOC levels in surface waters at

Ž .Diwalwal up to 40 mgrkg , much of which com-prises humic organic compounds derived from thedischarge of untreated sewage from the approxi-mately 50 000 resident settlement. The incorpora-tion of Hg2q into soluble organo-metallic com-plexes, as previously described for TOC-rich wa-

Ž .ters e.g. Lacerda and Salomons, 1998 , is there-fore considered a probable mechanism of Hgmobilisation. Although not directly determined inthis study, the discharge into surface drainage ofcyanic process fluids from the Diwalwal CIP planthas been inferred from the prevalence of high pH

Ž . Žlevels )9 and dissolved Na concentrations a.product of sodium cyanide feed . The formation

of soluble cyanometallic compounds may thusalso influence Hg mobility. Thus the extremelyhigh Hg concentrations in filtered water samplescollected close to Diwalwal probably reflect acombination of true dissolved mercury together

Ž .with mercury in organic humic and cyanic com-plexes.

4.2. Suspended particulate matter, bottom sedimentand hea¨y mineral concentrates

Sediment-bound Hg concentrations in surfacedrainage characteristically reflect contamination

Ž .via a range of pathways including 1 depositionand inwash of Hg formerly mobilised as vapour

Ž .during the ‘torching’ of amalgams; 2 particulateHg inputs derived from the inwash of contami-

Ž .nated tailings; and 3 Hg scavenged from solu-

( )J.D. Appleton et al. r The Science of the Total En¨ironment 228 1999 95]109100

Fig. 2. Distribution of Hg in filtered stream water, upper Agusan, Mindanao.

Žtion by surface adsorption Lacerda and Salo-.mons, 1998 . In the Agusan basin, elevated sedi-

mentary Hg concentrations occur close to allknown point source areas, including Diwalwal,

Ž .Bango, Mainit, Pantukan and Pasian Fig. 3 . Thepeak level of sedimentary Hg contamination near

ŽDiwalwal is high maximum values of 34 mgrkg,40 mgrkg and 62 mgrkg in BS, SPM and HMC,

.respectively relative to all other sites in easternŽ .Mindanao Williams et al., 1995 . Such concentra-

tions are also higher than recorded in BS samplesfrom most other mining contaminated regions.

Ž .Callaham et al. 1994 report Hg concentrationsof up to 7.4 mgrkg in North Carolina mining

Ž .districts, and Lacerda and Salomons 1998 havedocumented levels to 25 mgrkg in artisanal min-ing areas of the Brazilian Amazon. Malm et al.Ž .1990 have, however, reported much higher sedi-ment concentrations of up to 157 mgrkg Hg inthe Madeira River, Brazil.

( )J.D. Appleton et al. r The Science of the Total En¨ironment 228 1999 95]109 101

Fig. 3. Distribution of Hg in bottom sediment, upper Agusan, Mindanao.

4.3. Dispersion trends and mechanisms

The trends of Hg dispersion recognised in theeastern Mindanao survey are summarised below:

Ž .1. In the Mamunga River draining Diwalwalthe rate of downstream decline in Hg is, afterlarge fluctuations in the first 2]5 km, broadly

Ž .analogous in SPM, W and HMC Fig. 4 . TheHg gradient in BS is less acute with relativelyhigh concentrations of approximately 20

Ž .mgrkg maintained for almost 15 km Fig. 4 .The Manat-Agusan profiles, downstream ofthe much smaller Pantukan and Mainit min-

Ž .ing areas Fig. 6 , are characterised by steeplydeclining Hg concentrations in all media for20 km, after which values rise sharply at the

Žconfluence with the Mamunga River whichintroduces high levels of contamination from

.the Diwalwal area .2. The decline of aqueous Hg concentrations by

Žmore than two orders of magnitude from

( )J.D. Appleton et al. r The Science of the Total En¨ironment 228 1999 95]109102

Fig. 4. Downstream dispersion of Hg in the Mamunga River below the Diwalwal mining region.

1540 mgrl to 7 mgrl; November 1995 survey.data; Fig. 4 within 5 km of Dilwalwal is

accompanied by a discharge increment factorof 2]3. Such adjustments to the ‘dissolved’Hg load are thus not explicable solely interms of dilution, and secondary adsorptionandror volatilisation processes must additio-nally be significant.

3. In the main Agusan channel, continued dilu-tion of Hg-contaminated water and SPM isindicated by a progressive decline of concen-

Ž .trations by approx. one order of magnitudeover a distance of approximately 140 km to-

Ž .wards the Butuan Bay estuary Fig. 5 . Thistrend is not evident in bottom sediments, inwhich a 10-fold elevation of Hg occurs in BS

Fig. 5. Downstream dispersion of Hg in the Mamunga and Agusan Rivers below the Diwalwal mining region.

( )J.D. Appleton et al. r The Science of the Total En¨ironment 228 1999 95]109 103

Fig. 6. Downstream dispersion of Hg in the Manat]Agusan River system below the Pantukan]Mainit mining areas.

from 0.17 mgrkg in a sample taken from theMamunga, where it crosses the floodplain ofthe Agusan immediately above the Ma-munga]Agusan confluence, to 1.5 mgrkg ap-proximately 60 km downstream of the con-

Ž .fluence Figs. 3 and 5 . It is plausible that Hgin the former sample had been strongly di-

Ž .luted by low Hg sediment 0.1 mgrkg Hgwhich characterises the upper reaches of theAgusan floodplain. The rise to 1.5 mgrkgsuggests a greater contribution of contami-nated sediment from Diwalwal, adsorption ofHg onto organic matter in the lower hydraulicregime of the lower section of the Agusanandror progressive downstream translocationof Hg from a contaminant source of decliningsignificance, reflecting the known reductionof Hg use at Diwalwal in the last decade.

4. Mass transport of Hg is dominated by theŽSPM by a factor of 1000 relative to ‘dis-

.solved’ Hg . HMC Hg is generally less thanHg in bottom sediment suggesting that muchof the Hg dispersed in the aquatic environ-ment is adsorbed onto fine mineral particlesderived from mineral processing rather thandiscrete particles of metallic mercury oramalgam.

5. The progressive increase of the BSrHMCratio downstream from Diwalwal reflects sedi-mentation of the high specific gravity metallic

mercury and Au]Hg amalgam particles closeŽ .to the source of contamination Fig. 7 . SPM

Hg declines fairly rapidly downstream as therelatively heavy Hg contaminated particles arepreferentially deposited close to the source of

Ž .contamination Fig. 4 . In contrast, Hg in BSdeclines less rapidly reflecting more extensivecontamination created during periods of max-imum river flow and also the adsorption, overan extended period, of Hg onto BS. Hencethe BSrSPM ratio also increases progres-

Ž .sively downstream from Diwalwal Fig. 7 . Itis likely that the BSrSPM ratio would notincrease so rapidly downstream at times ofmaximum flow because the relatively heavyHg contaminated particles would be trans-ported further in a more dynamic hydraulicregime. The extent and magnitude of down-stream dispersion reflects the magnitude ofthe contaminant flux. A broad correlationexists between the number of active miners

Žand the level of Hg contamination in BS Fig..8 . The closeness of the relationship may in

part reflect the similar topographic and hy-draulic regimes in the areas studied and asimilar relationship may not apply to con-taminated rivers in the Amazon River Basin,for example. Hg in BS 5]8 km downstream ofthe main mining areas is used as an indicatorof the magnitude of contamination because

( )J.D. Appleton et al. r The Science of the Total En¨ironment 228 1999 95]109104

Fig. 7. Relative concentrations of Hg in bottom sediment, suspended particulate matter and heavy mineral concentrate in theMamunga]Agusan drainage system

Hg concentrations in the immediate vicinityof the mining areas are subject to relativelylarge spatial and temporal variance.

4.4. Temporal ¨ariation

The extent of temporal variation to which wa-ter and sediment geochemical data are subject iscritical to the interpretations placed on any sur-vey of contaminant behaviour in fluvial systems.Temporal variations in the Hg content of BS,HMC and W samples were evaluated at five sitessampled in November 1995 and February 1996.At all sites, the concentration of Hg in BS andHMC samples was found to increase over this

period, while ‘dissolved’ Hg values typically re-mained analogous or declined slightly. At onestation close to Diwalwal, a substantial reductionof Hg concentrations from 2900 to 0.86 ppb was

Ž .recorded Fig. 9 . The relatively high Hg concen-trations recorded in BS and HMC samples duringthe February 1996 campaign reflect enhanced

Žtransport of contaminated sediment including.eroded tailings from the Diwalwal and Mainit

source areas as a consequence of extremeŽ .typhoon climatic conditions and high attendantrunoff across much of eastern Mindanao duringearly 1996. Within-site variance is negligible com-pared with between-site and temporal varianceŽ .Appleton and Williams, 1998 .

Fig. 8. Relationship between the estimated number of people actively involved in gold extraction and the level of Hg contaminationŽ . Žin bottom sediment BS . D: Diwalwal; P: Ponce Enrıquez, Ecuador; N: Nambija, Ecuador; G: Gango; data for Gango can be´

.found in Williams et al., 1995; data for Ponce Enrıquez and Nambija in Ecuador from Appleton and Williams, 1998 .´

( )J.D. Appleton et al. r The Science of the Total En¨ironment 228 1999 95]109 105

Ž . Ž .Fig. 9. Temporal variation in a Hg in filtered river water; bŽ .Hg in bottom sediment; and c Hg in heavy mineral concen-

Ž .trate for the Diwalwal Diwa and Mainit mining areas, Min-danao, Philippines.

4.5. Data e¨aluation in relation to water qualitycriteria

The remarkably high aqueous Hg concentra-

tions recorded in Diwalwal drainage warrant par-ticular scrutiny with respect to their potentialhuman and ecotoxicological impact. In the imme-diate vicinity of Diwalwal, the drainage is ex-tremely turbid, anoxic and probably incapable ofsupporting aquatic life irrespective of the ambient

ŽHg loading maxima 2900 mgrl and 1540 mgrlfor the July 1995 and November 1995 surveys,

.respectively . Approximately 10 km downstreamthe Mamunga River is, however, used both forpotable supply and for fishing. Values of 103mgrl recorded 8 km downstream of Diwalwal,and 7 mgrl at 14 km, clearly exceed both theWHO Drinking Water Guideline value and theUS-EPA Water Quality Criteria for the Protec-

Ž .tion of Aquatic Life Table 1 . Mercury data forwaters in the vicinity of the Bango, Pasian, NewBataan, Mainit and Pantukan artisanal miningareas in the Agusan basin are compliant withinternational water quality criteria.

Data for additional potentially toxic elementsindicate that, in contrast to many gold and base-metal mining localities characterised by aciddrainage, Diwalwal drainage holds a relatively lowbase-metal content. This reflects both the rela-tively low sulphidation character of the Diwalwalepithermal system, plus the strongly-buffered

Ž .regime pH 7.1]8.5 which inhibits mobilisationof metals, such as Pb, Zn and Cu. Filtered watersamples analysed during this study did, however,yield five Cu values in excess of the US-EPAFreshwater threshold for the Protection of

Ž .Aquatic Life Table 1 . This may reflect a combi-nation of true dissolved Cu together with Cu in

Ž .organic humic and cyanic complexes.

4.6. Sediment quality

The Hg Toxic Effect Threshold for the Protec-Žtion of Aquatic Life Environment Canada, 1992;

. Ž .Table 2 is exceeded in bottom sediments BSalong a 20-km section of the Mamunga River.The bioavailability of this Hg load, and hence itstrue toxicity is, however, unknown. Such highsedimentary concentrations of Hg may constitutea long-term source of contamination, which willinevitably become more widely dispersed duringperiods of high flow. With progressive methyla-

( )J.D. Appleton et al. r The Science of the Total En¨ironment 228 1999 95]109106

Table 1Comparison of potentially harmful element concentrations in filtered water samples from the Diwalwal artisanal mining area with

Ž .water quality criteria concentrations in mgrl

WHO EC drinking-Protection of aquatic Diwalwala drinking water MAClife } freshwater

waterbMax. conc. Cont. Max. 8 km downstreamcconc.

Cu 17 11 2000 100 344 73dHg 2.1 0.012 1 1 1539 103

Zn 110 100 5000 na 81 12

Notes. Max., maximum concentration at or immediately downstream of mining area; 8 km downstream, concentration approximately8 km downstream of mining area; na, not available; EC, European Community; MAC, Maximum Admissible Concentration; WHO,World Health Organisation.a Ž . ŽEPA Section 304 a Criteria for the Protection of Aquatic Life from Priority Toxic Pollutants US Clean Water Act, February 5,

.1993; Part 131-Water Quality Standards, Sec. 131.36 proposed amendment April 1995 .b Ž .Criteria maximum concentration CMC , the highest concentration of a pollutant to which aquatic life can be exposed for a short

Ž .period of time 1-h average without deleterious effects.c Ž .Criteria continuous concentration CCC , the highest concentration of a pollutant to which aquatic life can be exposed for an

Ž .extended period of time 4 days without deleterious effects.d If the CCC exceeds 0.012 mgrl more than once in a 3-year period in the ambient water, the edible portion of aquatic species ofconcern must be analysed to determine whether the concentration of methyl mercury exceeds the FDA action level of 1.0 mgrkg.

tion in the anoxic sediments which characteriseparts of the lower Mamunga and Agusan rivers,

Žfood chain translocation is also plausible Lacerda.and Salomons, 1998 .

With the exception of Hg, sediments analysedfrom the gold mining areas of eastern Mindanaoare generally characterised by relatively low con-

Žcentrations of toxic trace elements including As,.Cu, Pb, Zn . Sediment Quality Criteria for the

ŽProtection of Aquatic Life Toxic Effects Thresh-.old, Environment Canada, 1992 are not ex-

ceeded.

4.7. Hair sur ey

Results for 48 hair samples collected from thevicinity of the Apokon ore processing operationsshow a total range of 0.31]13 mgrkg. Using anestablished empirical hairrblood transformation

Žratio of 250r1 Piotrowski and Inskip, 1981; Lac-.erda and Salomons, 1998 , a blood concentration

range of approximately 1.2]52.0 ngrml may beindicated. Only two subjects yielded values equalto or in excess of the WHOrFAO hair Hg refer-

Table 2Ž .Comparison of potentially harmful element concentrations in stream bottom sediments BS from the Diwalwal area with Sediment

Ž .Quality Criteria for Protection of Aquatic Life Environment Canada, 1992 . All concentrations are given in mgrkg

aSediment quality criteria Diwalwal

No effects Minimal effects Toxic effects Max. 8 km downstreamthreshold threshold threshold

As 3 7 17 11 naCu 28 28 86 67 naHg 0.05 0.2 1 32 20Pb 23 53 170 65 naZn 100 150 540 114 na

Notes. Max., maximum concentration at or immediately downstream of mining area; 8 km downstream, concentration approximately8 km downstream of mining area; na, not available.a Ž .Sediment Quality Criteria for Protection of Aquatic Life Environment Canada, 1992 quoted in Haines et al., 1994 .

( )J.D. Appleton et al. r The Science of the Total En¨ironment 228 1999 95]109 107

Žence dose of 7 mgrkg blood equivalent 28.ngrml , and only one exceeded the 10 mgrkg

Ž .value proposed by Barbosa et al. 1995 as poten-tially associated with abnormal infantile develop-ment.

Analysis of the Apokon hair survey data bysub-group revealed a clear tendency for enhancedHg burdens among workers from CIP processing

Ž .plant and ballmills Fig. 10 . The only amalgamblow-torcher sampled revealed, however, a sur-prisingly low Hg burden. The median value for

Ž .the CIP plant workers 3.5 mgrkg is close to thatrecorded for gold prospectors in the Madeira

ŽRiver area, Brazil 4.6 mgrkg; Lacerda and Salo-.mons, 1998 , however, the peak values in the

ŽApokon workers are substantially lower 7.mgrkg . A sub-group of residents living within 50

m of the main Apokon ore processing plantyielded hair Hg burdens which are indistinguish-able from those of people living at greater dis-tances.

With the exception of three hair valuesrecorded in the range 5]13 mgrkg, the Apokonhair data are essentially consistent with the re-sults of the Philippine DOH blood survey in 1996,showing blood burdens to be generally less than20 ngrml. Interviews conducted with subjects in-cluded in the Apokon hair sampling survey re-vealed an almost universal consumption of fish ata daily frequency. Fish constitutes the principal

source of methyl Hg in all communities not other-wise prone to residential or occupational expo-

Ž .sure, and the WHO 1976 and Piotrowski andŽ .Inskip 1981 report mean hair Hg concentrations

Ž .of up to 5 mgrkg blood equivalent 20 ngrml ascommon in fish-eating communities of theMediterranean and the Americas. While slightoccupational exposure among ore-processingplant operators is recognised, any wider humanhealth impact of the Apokon ore-processing plantis difficult to confirm.

5. Conclusions

This survey has highlighted the marked impactof artisanal gold mining on the flux of Hg into theAgusan basin of eastern Mindanao in the Philip-pines. In the vicinity of Diwalwal, this impact isparticularly notable on account of the localisedprevalence of ‘dissolved’ Hg at concentrationshitherto undocumented in artisanal gold mininglocalities world-wide. The bulk of the contami-nant Hg load in the aquatic environment has,however, been shown to be transported predomi-

Ž .nantly in association with the suspended SPMload of the rivers. Extensive dispersion of Hg inbottom sediments has been also been highlighted.

Ž .Hg in filtered -0.45 mm river water 14 kmdownstream of Diwalwal exceeds both the WHODrinking Water Guideline value and the USEPA

Ž .Fig. 10. Hg mgrkg in hair results for Apokon human impact assessment. 50 m: people living within 50 m of CIP processing plants;BM: ball mill operators using mercury; BT: blow torch operator; CIP: workers at CIP processing plant; O: other people notincluded in previous groups.

( )J.D. Appleton et al. r The Science of the Total En¨ironment 228 1999 95]109108

Water Quality Criteria for the Protection ofAquatic Life, and is thus a potential health haz-ard. The Hg Sediment Quality Toxic EffectThreshold for the Protection of Aquatic Life isexceeded in a 20-km section of the MamungaRiver, below Diwalwal. Progressive release of Hgthrough methylation may pose a long-term hazardto aquatic biota.

Water and SPM samples indicate the currentflux of contamination to the drainage system butare very susceptible to temporal variations relatedto short-term fluctuations in discharges from pro-cessing plants and to rainfall-related dilution ef-fects. BS provides a more stable indication of theextent and magnitude of contamination whereasHMC samples indicate the amount of particulateHg metal and Au]Hg amalgam in the river sedi-ment. BS Hg indicates the likely hazard to biotafrom remobilisation of Hg in bottom sediments asa result of methylation processes. Water and SPMHg provide a more relevant indicator of contami-nant fluxes at the time of sampling and hence thepotential hazards to biota and humans via thisexposure route. More detailed monitoring is re-quired to verify the level, frequency and durationof high contaminant fluxes in stream water andsuspended particulate matter and to establish fac-tors, such as rates of sedimentary Hg methylationand subsequent bioassimilation.

Hair Hg data indicate that ballmill operatorsand workers at the CIP plants processing Hgcontaminated tailings at eastern Mindanao’s prin-cipal gold beneficiation centre, Apokon, may besubject to enhanced occupational Hg exposure.However, even these workers have Hg hair con-centrations which fall within the range found infishing communities where there is no exposureto Hg derived from gold processing sources. Itappears that the wider population of this area hasnot been impacted significantly by Hg contamina-tion-related to gold beneficiation.

Acknowledgements

The research described in this paper was car-ried out as part of a UK Department for Interna-

Ž .tional Development DFID funded KnowledgeŽ .and Research KAR project concerned with the

theme: ‘Identification and amelioration of miner-als-related and other toxic hazards’. The purposeof the research is to help mitigate Hg contamina-tion, health and ecological hazards associated withartisanal alluvial gold mining. The programmeforms part of the British Government’s provisionof aid to developing countries. Logistic supportand field guidance in the Philippines was providedby staff of the Philippine Department of Environ-

Žment and Natural Resources Mines and Geo-.sciences Bureau , particularly Resty Gomez. Bar-

Ž .bara Vickers BGS was responsible for the CV-AFS Hg determinations, and Mark Cave and

Ž .Linda Ault BGS for the ICP-AES analyses. Thepaper is published with the permission of the

Ž .Director, British Geological Survey NERC .

References

Appleton JD, Williams MW. Mitigation of mining-relatedmercury pollution hazards: Summary Report. British Geo-logical Survey Technical Report WCr98r34, 1998.

Barbosa AC, Boischio AAP, East GA, Ferrari I, Goncalves A,Silva PRM, Cruz TME. Mercury contamination in theBrazilian Amazon. Environmental and occupational as-pects. Water Air Soil Pollut 1995;80:109]121.

Breward N, Apostol A, Appleton JD, Gomez R, Miguel J.Mercury and other heavy-metal contamination associatedwith gold mining in the Agusan river catchment, Mindanao,the Philippines. British Geological Survey Technical Re-port WCr96r01. 2nd ed., 1996.

Callaham JE, Miller JW, Craig JR. Mercury pollution as aresult of gold extraction in North Carolina, USA. ApplGeochem 1994;9r2:235]241.

Cargnello JA, Powell JJ, Thompson RPH, Crocker PR, WattF. Elemental hair analysis using nuclear microscopy andX-ray energy dispersive spectroscopy. Analyst 1995;120:783]787.

Chittleborough G. A chemist’s view of the analysis of humanhair for trace elements. Sci Total Environ 1980;14:53]75.

Haines ML, Brydges K, MacDonald MJ, Smith SL, MacDo-nald DD. A review of environmental quality criteria andguidelines for priority substances in the Fraser River Basin.Supporting Documentation. DOE FRAP 1994-31. Environ-ment Canada, Vancouver, 1994.

Health and Welfare of Canada. Task force on organic mer-cury in the environment. Grassy Narrows and White Dogreserves, Ontario. Department of National Health andWelfare, Ottowa, Canada, 1979.

Lacerda LD, Salomons, W. Mercury from gold and silvermining: a chemical time bomb? Berlin: Springer, 1998:146.

Malm O, Pfeiffer WC, Souza CMM, Reuther A. Mercurypollution due to gold mining in the Madeira River basin,Brazil. Ambio 1990;19r1:11]15.

( )J.D. Appleton et al. r The Science of the Total En¨ironment 228 1999 95]109 109

Mitchell AHG, Leach TM. Epithermal gold in the Philippines:island arc metallogenesis, geothermal systems and geology.New York: Academic Press, 1991:457.

Okamoto K, Morita M, Quan H, Uehiro T, Fuwa K. Prepara-tion and certification of human hair powder referencematerial. Clin Chem 1985;31r10:1592]1597.

Paccagnella B, Pratti L. Total mercury in blood and hair ofItalian people. Ig Mod 1974;67:369]370.

Pfeiffer WC, Lacerda DL, Malm O, Souza CMM, De SilvaEG, Bastos WR. Mercury concentrations in inland watersof gold mining areas in Rondonia, Brazil. Sci Total Environ1989;78:233]240.

Pfeiffer WC, Malm O, Souza CMM, Lacerda DL, De SilvaEG, Bastos WR. Mercury in the Madeira River ecosystem,Rondonia, Brazil. Forest Ecol Manage 1991;38:239]245.

Piotrowski JK, Inskip MJ. Health effects of methylmercury.Monitoring and Assessment Research Centre, University ofLondon, London, 1981:82.

Riodolfi M. Further epidemiological study of Hg levels in fishand human blood and hair. Ig Mod 1977;70:169]186.

Torres EB. Environmental and health surveillance of mercuryuse in small scale gold processing industries in the Philip-pines. In: Proceedings of International Symposium Epi-demiological Studies of Environmental Pollution andHealth Effects of Methylmercury. National Institute forMinamata Disease, Kumamoto, 1992:56]65.

Torres EB. Epidemiological investigation of mercury exposureand health effects in the Philippines. In: Proceedings ofInternational Workshop on Environmental Mercury Pollu-tion in the Amazon River Basin. National Institute forMinamata Disease, Rio de Janeiro, 1994:3]9.

Williams TM. Visit to the Philippines and Thailand. BritishGeological Survey Technical Report WPr97r1, 1997.

Williams TM, Apostol A, Miguel J. Mercury contamination inartisanal gold mining areas of eastern Mindanao, Philip-pines: a preliminary assessment. British Geological SurveyTechnical Report WCr95r72rR, 1995.

WHO. Environmental health criteria document 1: mercury.World Health Organisation, Geneva, 1976.