Geolink_mitigasi.pdf

DFID Department for international Development

BRITISH GEOLOGICAL SURVEY TECHNICAL REPORT WC/98/34

Overseas Geology Series

Mitigation of mining-related mercury pollution hazards: Project Summary Report J D Appleton' and T M Williams' with contributions from: A Apostol', N Breward', M Carrasco5,EJ Evans', R Maldonado3,J Migue12,C Miranda2,CJ Mitchell', H Orbea3,MT Styles', J Weekes6 'British Geological Survey, Keyworth, Nottingham, NG12 5GG, UK.

'Mines and Geosciences Bureau, Quezon City, Philippines 3

CODIGEM, Quito, Ecuador DINAMI, Machala, Ecuador 'DINAPA, Quito, Ecuador ?nstitute of Terrestrial Ecology, NERC, UK

4

This report is an output from a project funded by the UK Department for International Development (DFID). The views expressed in this report are not necessarily those of the DFID.

DFID classification Subsector: Geoscience Theme: G2 Identify and ameliorate minerals-related and other geochemical toxic hazards Project title: Mitigation of mining-related mercury pollution hazards. Project Reference: R6226 Bibliographic reference:

J D Appleton and MW William, 1998. Mitigation of mining-related mercury pollution hazards: Project Summary Report. British Geological Survey, Overseas Geology Series Technical Report WC/96/55, Keyworth, Nottingham, UK. Key words: Ecuador, gold mining, mercury Cover illustration: Artisanal miner using mercury to amalgamategold in alluvial heavy mineral concentrate, Mainit, eastern Mindanao, Philippines.

0 NERC 1998

Keyworth, Nottingham, British Geological Survey, 1998

EXECUTIVE SUMMARY

This report summarises the objectives, results and relevance to sustainable development of a study of the environmental impacts of mining-related mercury (Hg) contamination. The purpose of the research is to help mitigate the health and ecological hazards associated with artisanal gold mining. The research was carried out during the period 1 April 1995 to 31 December 1998 as part of the UK Department for International Development (DFID) Engineering Division Technology Development and Research (TDR) Programme concerned with the theme: "Identification and amelioration of minerals-related and other toxic hazards". This programme forms part of the British Government's provision of aid to developing countries. Mercury has been widely used in the mining industry since Roman times for the amalgamation and concentration of gold and silver. The total global release of Hg into the environment through this process prior to 1930 has been estimated as over 260,000 t, after which emissions declined with the introduction of cyanidation processing technology. In the 1970's high gold prices triggered a major artisanal gold rush in many countries of the Southern Hemisphere, possibly involving more than 10 million people. The attendant resurgence in the use of Hg for Au processing which occurred at this time has continued largely unabated to the present day, and may currently account for 10% of the total anthropogenic emissions of Hg into the global environment. The amalgamation process requires the use of 2 to 3 tons of mercury for each ton of gold recovered. Small-scale miners produce thousands of tons of gold each year so environmental contamination by mercury release is of considerable concern. A number of developing countries in Africa, Asia and Latin America have experienced a rapid growth of the small-scale gold mining sector during the last two decades. This has reached huge proportions on the island of Mindanao, in the Philippines, where more than 100,000 people have been engaged in largely uncontrolled artisanal mining. There was a requirement in the Philippines in particular, for a detailed evaluation to be carried out of the environmental impact on both the miners and local communities of mercury contamination derived from mining. Smaller magnitude artisanal mining had also developed in southern Ecuador. As good collaborative links had been established with organisations in the Philippines and Ecuador that were able to provide both logistic and scientific support, these two contrasting examples of artisanal gold mining were selected as case study areas in which to fulfil the project objectives: 1. Provision of a generic protocol for monitoring the sources and fates of contaminant mercury in gold mining regions. 2. Assessment of human and ecotoxicological impacts of mercury pollution and calculation of humadenvironmental risk thresholds. 3. Promotion of improved methods of gold recovery with applicability to the artisanal sector. 4. Design and implementation of practical legislation for abatement of mining-related mercury contamination. It was originally intended to restrict monitoring and impact assessment studies to mercury contamination associated with small-scale gold mining. However, the main collaborating organisation in the Philippines, the Mines and Geoscience Bureau (MGB) was obliged to reappraise its priorities in the light of widespread media reports of human mercury poisoning in an area of former cinnabar mining on the island of Palawan. In response to a directive from the Philippines President, the MGB commissioned an inter-agency investigation into the Palawan mercury scare, and a formal request for the diversion of BGS-DFID funds from Mindanao to Palawan was received by BGS. Although falling strictly beyond the remit of TDR project R6226 (which is primarily concerned with Hg contamination associated with artisanal gold mining), the request was considered by BGS and DFID to constitute a strong case for demand-led TDR expenditure. Accordingly, a preliminary investigation of the extent of the Palawan Hg problem was carried out in collaboration with the UK Institute of Terrestrial Ecology (ITE).In view of the paucity of geo-

1

chemical, ecotoxicological or epidemiological data depicting the extent and magnitude of Hg hazards in central-eastem Palawan, the principal objectives of the study were to: Establish the spatial extent, magnitude and temporal trends of Hg deposition in the marine environment of Honda Bay. 0 Assess the risk posed to inhabitants of the Sitio Honda Bay jetty and the cinnabar mine locality as a consequence of living on a mine-waste substrate. Assess the extent of Hg bioassimilation and attendant toxicological stress in marine biota (fish and shellfish) at various localities in Honda Bay. Magnitude of mercury contamination.

Monitoring studies highlighted the marked impact of artisanal gold mining on the flux of mercury into the aquatic environment. In the vicinity of the Diwalwal artisanal gold mining area on the island of Mindanao this impact is particularly notable because of the localised prevalence of ‘dissolved’ Hg at concentrations hitherto undocumented in artisanal gold mining localities worldwide. The magnitude and extent of contamination by mercury, derived from both mine tailings and mineral processing activities, and other potentially harmful elements (such as arsenic, copper and lead) derived from sulphide minerals is highly variable, but is generally a predictable function of geological, environmental, social and technological factors. Protocol for monitoring contamination

The results of this study have shown that water and suspended particulate matter samples indicate the current flux of contamination to the drainage system but are very susceptible to temporal variations related to short term fluctuations in discharges from processing plants and to rainfallrelated dilution effects. Bottom sediment provides a more stable indication of the extent and magnitude of contamination whereas heavy mineral concentrate samples indicate the amount of particulate Hg metal and Au-Hg amalgam in the river sediment. Mercury concentrations in bottom sediments indicate the likely hazard to biota from remobilization of Hg as a result of methylation processes. Mercury in filtered water and suspended particulate matter provides a more relevant indicator of contaminant fluxes at the time of sampling and hence the potential hazards to biota and humans via this exposure route. Detailed monitoring is required to verify the level, frequency and duration of high contaminant fluxes in stream water and suspended particulate matter and to establish factors such as rates of sedimentary Hg methylation and subsequent bioassimilation. In near-shore marine environments, mercury data for sediments may be used to evaluate the impact of mercury contamination. In the Sitio Honda Bay case study such data demonstrated quite clearly that contamination is strictly localised in the near-shore zone. Mineralogical and chemical characterisation of mine and mineral processing waste piles, tailings ponds, and acid mine drainage is a valuable method for assessing the potential hazards associated with these contaminant sources. The Palawan case study demonstrated that very high mercury concentrations found in the Sitio Honda Bay waste substrate are held predominantly in non-sulphide alteration products of probable low bioavailability. The contribution to total human exposure among the Sitio Honda Bay population (through hand to mouth ingestion and particulate or volatile phase respiration) is therefore likely to be modest. Environmental agencies (EAs) in developing countries need to be able evaluate both continuous as well as sudden, potentially catastrophic, pollution incidents, such as may be associated with the collapse of a tailings dam. The EA will need to determine when the pollution occurred, as well as its source and magnitude. In such incidents, the release of contaminants may well have ceased by the time the EA is able to react to the incident and start its investigations. For this reason, whilst filtered water or suspended particulate matter samples will be effective for the continuous or periodic monitoring of aquatic environments that are being subject to relatively continuous contamination by effluent from mineral processing activities, these media would generally be ineffective for detecting sporadic pollution incidents. To assist with its function of pol-

11

A range of changes to gold processing procedures used by small-scale miners would reduce the amount of mercury contamination. These include: (a) Improved sluice box design and increased use of effective gravity separation methods (such as the shaking table) might reduce the use of mercury on sluices. This could be achieved by demonstration of the financial benefits of increased gold recuperation and hence the amount of income generated. (b) Development of a retort system which retains the Hg but produces a higher quality gold product uncontaminated by Fe and As (which reduces the value of the Au). In Ecuador, for example, small-scale gold processors frequently have retorts but they are rarely used. It is not entirely clear whether artisanal miners do not use the retorts because (1) they are suspicious that they are losing some of the gold in the retort, (2) the processing takes longer or (3) the product is lower quality and less marketable. If 80-90% of Hg loss is related to burning off the mercury from amalgam then the biggest effort should clearly be put into improving this aspect of the process. Some NGOs have tried this approach but, in general, this appears to have been unsuccessful. There is clearly a requirement for further educational campaigns designed to demonstrate the positive health and economic aspects of using retorts. (c) Increasing the use of cyanidation in an environmentally acceptable way would reduce the need for amalgamation. This would require better control on the location of cyanidation plants and management procedures. (d) Perhaps the greatest potential for reducing environmental impact would be through the adoption of an effective system of co-operative production. This should lead to better management of mineral processing plants with reduced environmental impact while still leaving the means of production in the hands of the poorer sector of the population. In theory it should permit a more effective and economical production system to be adopted through upsizing production systems and providing access to capital for larger investments (e.g. shaking tables and efficiently run cyanidation plants). Legislative control of mercury contamination.

In the artisanal gold mining sector, the over-riding concern is with mercury pollution. Potential items of significance in a regulatory framework include both controlling the supply of mercury as well as the imposition of practical (attainable and enforceable) legal standards for environmental mercury concentrations in both water and sediment and also discharges. These need to be complemented by the inclusion of monitoring guidelines within all new legislation. For the miners, a legislative framework and standards could be set to ensure that Hg was used in smaller amounts; that tromol mills were replaced by gravity separation etc.; and that tailings of ore treated with Hg should pass through a series of settling ponds before water containing suspended particulate matter was allowed to pass into water courses. One option that might be used in the control of mercury would be to licence buyers and sellers within a legislative framework and to have a register of mercury users. However, imposition of mercury controls or bans in many developing countries is likely to be totally ineffective as the majority of artisanal miners operate illegally. Moreover, there is little financial incentive for miners to stop using mercury as it often costs only 3 to 4 US dollars a kilogram. The sustainable development of artisanal gold mining is severely constrained by: (1) environmental, health and safety problems such as those linked to the use of mercury; (2) use of inefficient methods and equipment; (3) lack of appropriate legal, regulatory and institutional frameworks. The World Bank has identified the need to: (1) set reliable environmental standards that are practical to apply and appropriate to local circumstances; (2) strengthen environmental impact assessment processes; (3) encourage the establishment of comprehensive environmental management systems. Attempts to bring about pollution prevention in the mining sector has generally been through legislation (e.g. US EPA Pollution Prevention Act 1990) and specific regulations for mining ac-

V

tivities (e.g. in Bolivia and Ecuador). Pollution control is achieved through waste assessment procedures such as Environmental Impact Analysis and Environmental Audits but also includes environmental standards for exploration, mineral extraction, and processing. In Ecuador, there are additional standards applicable to the artisanal sector. In many developing countries, enforcement is generally extremely restricted and is almost totally concentrated on the larger polluters. It had been anticipated that recommendations for maximum permissible environmental mercury levels in the Philippines would evolve from the biological and human impact assessments and that these would be incorporated into new legislative controls on the small-scale mining sector being prepared by the MGB at the time the project started. As the authority responsible for the control of mining activities and related environmental impacts, the MGB needed to formulate legislation to regulate mercury usage in small-scale mining activities with the aim of preventing further damage to the environment and human health. In the event, the new Philippines mining laws were pushed through much faster than had been envisaged, effectively precluding BGS’ involvement in this aspect of the project programme. Insufficient information has been compiled through the current project to establish maximum permissible environmental levels (based on impact assessmenddose response data). Legislative approaches have serious limitations, especially as a mechanism for controlling the small-scale, artisanal mining sector. The use of legislation to control working conditions among artisanal miners is commendable in theory but somewhat unrealistic within the context of most artisanal gold mining operations. Legislation for the artisanal sector is in many respects futile as such activities usually operate beyond the law. For example, at Diwalwal in the Philippines, up to 100,OOO people work a world class gold deposit which is officially held by four major international companies. In the absence of the political will to enforce the law, legislative controls will be ineffective. Having a legislative framework does not necessarily mean that it will be adhered to. Alternatives to legislative control.

Alternative approaches to bring about better working conditions and a reduction in emissions and waste include: direct incentives or financial return to the producer; 0 cleaner production technologies which demonstrably save the producer money and/or increase his income (e.g. by increasing recuperation of gold, for example, by the use of more efficient mineral processing technologies) use of ‘voluntary’ compliance methods to encourage improved environmental performance. A ‘stakeholder’ community-based approach involving producers, local authorities and smallscale miners associations needs to be promoted to share information about pollution impacts, to develop environmental awareness and to bring about a reduction of the negative environmental impacts of small-scale mining. Drivers to bring about change and adoption of safer practices must be economic rather than regulatory. Artisanal miners must be provided with more cost-effective methods of gold beneficiation from which they can gain tangible economic benefits. These benefits must have immediate impact, as the pre-eminent concern of people involved in the artisanal mining sector is to eam a living. Human health and financial security over the long-term are much lesser concerns, as is the impact of their activities on the environment. Development agencies, and especially NGOs, have a critical role to play in supporting this approach. The implementation of cleaner technologies and the ‘stakeholder’ approach are important foci of NGO led projects in Bolivia, Ecuador and Zimbabwe.

vi

CONTENTS 1. PURPOSE AND SCOPE OF REPORT

1

2. MERCURY CONTAMINATION 2.1 Chemistry of mercury 2.2 Toxicity and human exposure 2.3 Mercury contamination 2.4 A chemical time bomb ? 3. PROJECT OBJECTIVES (OUTPUTS)

4

4. WORK PLAN

5

5 . COLLABORATIVEFRAMEWORK

6

6. MINING RELATED MERCURY CONTAMINATION 6.1 Philippines 6.1.1 Gold mining areas: Mindanao 6.1.2 Mercury mining - Palawan 6.2 Ecuador 6.2.1 Nambija 6.2.2 Ponce Enriquez

7 7

7. MONITORING THE DISPERSION OF MERCURY CONTAMINATION 7.1 Sampling and analytical methods 7.1.1 Drainage surveys

10 10 10 10 11 11 11 11 11 12 12 15 15 16 19 22 22 24 24 24 25 25

Water (W) Stream bottom sediment (BS) Suspended particulate matter (SPM) Heavy mineral concentrates (HMC) 7.1.2 Marine sediment survey : Palawan 7.2 Monitoring results for gold mining areas 7.2.1 With-site variance 7.2.2 Temporal variance 7.2.3 Downstream dispersion of contamination from artisanal gold mining areas Water Suspended particulate matter (SPM), bottom sediment and heavy mineral concentrates Downstream dispersion trends Relative merits of sample media 7.2.4 Magnitude of mercury contamination 7.3 Monitoring results for mercury mining area: Palawan 7.3.1 Hydrogeochemical survey 7.3.2 Stream sediment 7.3.3 Marine sediment: Honda Bay 7.3.4 Sitio Honda Bay - geochemical and mineralogical results 8. WATER AND SEDIMENT QUALITY 8.1 Water quality 8.1.1 Mercury 8.1.2 Arsenic 8.1.3 Copper

7 8 9

9 10

26 26 26 26 27

8.2 Sediment quality 8.2.1 Mercury 8.2.2 Arsenic 8.2.3 Copper 8.2.4 Cadmium 8.2.5 Discussion and recommendations 9. HUMAN AND ECOTOXICOLOGICAL IMPACT ASSESSMENT 9.1 Human exposure to mercury 9.2 Palawan, Philippines 9.2.1 Background 9.2.2 Fish and shell-fish Sampling and analysis Neutral-red biomarker assessment. Previous work Results of present study Fish Mussels Neutral-red biomarker assessment results 9.2.3 Assessment of human Hg burdens Background Hair sampling and analysis Hair Results Hair-blood relationships Dietary and residential Hg exposure Toxicological implications of the recorded Hg burdens Amelioration strategies 9.3 Tagum, Eastern Mindanao, Philippines 9.3.1 Background 9.3.2 Hair sampling 9.3.3 Hair results and toxicological implications 9.4 Ponce Enriquez, Ecuador 9.5 Cyanide 10. PROMOTION OF IMPROVED METHODS OF GOLD RECOVERY WITH APPLICABILITY TO THE ARTISANAL SECTOR 10.1 Introduction 10.2 Review of gold recovery methods 10.2.1 Gold particle size 10.2.2 Mercury amalgamation 10.2.3 Cyanidation 10.2.4 Costs and environmental impact 10.2.5 Gold recovery flow sheets Alluvial gold processing Hard rock gold processing 10.3 Recommendations for improved recovery of fine gold not involving amalgamation 10.3.1 Improvements to current sluice box practice Optimise clean-out Appropriate configuration Appropriate operating practice 10.3.2 Process modifications Washing prior to sluicing Processing coarse and fine separately Expanded metal and angle iron sluice rifles

28 28 28 28 29 29 29 29 30 30 30 30 31 31 31 31 32 33 33 33 33 34 36 36 36 37 37 37 38 38 38 39 39 39 40 40 40 41 41 42 42 42 42 43 43 43 43 43 43 44 44

10.3.3 Efficient alternatives

Replace sluices with jigs Introduce shaking tables and bowl concentrators to concentrate fine gold 10.4 Laboratory evaluation of BGS-designed gravity separator for gold recovery 10.4.1 Introduction 10.4.2 Sample collection for beneficiation tests Acupan Benguet Gold Operation (BGO) concession. Kias Creek Gango 10.4.3 Characterisation of gold-bearing samples 10.4.4 BGS-designed shaking table 10.4.5 Results 10.5 Field testing and promotion of BGS gravity separator 10.5.1 Existing practices 10.5.2 Kias Area Field tests Laboratory tests on products 10.5.3 Acupan Benquet Mine area Field tests Laboratory Tests 10.5.4 Miners reaction to the shaking table trials 10.6 Conclusions 10.7 Recommendations

44 44 44 44 44 45 45 45 45 46 46 47 48 49 49 49 49 49 49 49 50

50 50

11. LEGISLATION AND ENVIRONMENTAL STANDARDS 1 1.1 Legislative control of mercury contamination 11.2 Alternatives to legislative controls

52 52 53

12. DISSEMINATION WORKSHOPS

54

13. CONCLUSIONS AND RECOMMENDATIONS

57

14. TAKE-UP

57

15. ACKNOWLEDGEMENTS

57

16. REFERENCES

57

APPENDIX 1

PROJECT TECHNICAL REPORTS

61

APPENDIX 2

OTHER PROJECT PUBLICATIONS

61

I. PURPOSE AND SCOPE OF REPORT

This report summarises the objectives, results and relevance to sustainable development of a study of the environmental impacts of mining-related mercury (Hg) contamination. The purpose of the research is to help mitigate mercury contamination, health and ecological hazards associated with artisanal alluvial gold mining. The research was carried out during the period 1 April 1995 to 31 December 1998 as part of the DFID Engineering Division Technology Development and Research Programme concerned with the theme: "Identification and amelioration of minerals-related and other toxic hazards". The programme forms part of the British Government's provision of aid to developing countries. Amalgamation with mercury has been used as a method of gold and silver beneficiation since Roman times. The total global release of Hg into the environment through this process prior to 1930 has been estimated as over 260,000 t, after which emissions declined with the introduction of cyanidation processing technology (Lacerda and Salomons, 1998). In the 1970's high gold prices triggered a major artisanal gold rush in many countries of the Southern Hemisphere, possibly involving more than 10 million people. The attendant resurgence in the use of Hg for Au processing which occurred at this time has continued largely unabated to the present day, and may currently account for 10%of the total anthropogenic flux of Hg into the global environment. The report brings together the key results and recommendations of a series of BGS Technical Reports (Appendix A) which describe in detail the results of (a) six site monitoring and impact assessment studies carried out in the Philippines and Ecuador; (b) a review of gold recovery methods; (c) laboratory and field testing of a gravity separator designed to improve gold recovery methods and reduce the need for the use of mercury. Dissemination of the results of the programme was attained through workshops held in Bolivia and Ecuador, and a symposium in Hong Kong. This summary technical report is intended to convey the most important results in a form which is accessible and readily understandable by government, agency and mining company personnel including policy makers, scientists, planners, engineers and environmental health specialists. The results of the research programme are particularly timely insofar that the World Bank International Conference on Development, Environment and Mining (World Bank, 1994) and the International Roundtable on Artisanal Mining (World Bank, 1996) have recently identified the urgent need to: (1) set reliable environmental standards that are practical to apply; (2) strengthen environmental impact assessment processes; (3) encourage the establishment of comprehensive environmental management systems. It is recognised that many countries need assistance with developing appropriate procedures and methodologies for setting environmental standards that are appropriate to their particular circumstances. Artisanal gold mining, in particular, is considered to be primarily a poverty issue which requires a holistic approach if it is to be transformed from an activity with major negative social and environmental side effects into environmentally sustainable small-scale mining. However, the sustainable development of artisanal gold mining is severely constrained by: (1) environmental, health and safety problems such as those linked to the use of mercury; (2) use of inefficient methods and equipment; (3) lack of appropriate legal, regulatory and institutional frameworks. A strategy for assisting artisanal mining within the context of World Bank supported reform programmes includes components aimed at improving the environmental, living and working conditions of artisanal miners together with the establishment of enabling conditions, including appropriate legal and regulatory frameworks (World Bank, 1996). A pre-requisite for improving environmental health and safety conditions and effective environmental management is the establishment of an environmental baseline, together with monitoring environmental pollution, and an assessment of health risks and occupational hazards. There is also a need to establish appropriate environmental control measures, including regulatory mechanisms, as well as the introduction of environmentally sound mining and processing techniques. The World Bank identified the need for donor organisations to support baseline studies, conferences and seminars that aim to find workable solutions to these important issues. The technology development and research programme summarised in this report deals directly with some of these objectives.

1

2. MERCURY CONTAMINATION 2. I Chemistry of mercury

Mercury has a density of 13.53 g/cm3 and an unusually low melting point of -38.9"C. It is distinguished from all other metals by its persistence in a liquid state under most surface-environment conditions. Mercury possesses three oxidation states (MO- M+2)of which Hg2+compounds and elemental Hg are most common. Mercury is among the rarest of naturally occurring elements, with an average abundance in the earth's crust of 0.08 pg/g (Greenwood and Earnshaw, 1984). It is characteristically enriched in acid relative to mafic igneous rocks (average 0.08 pg/g in granites and 0.01 pg/g in basalts), with concentrations frequently one to two orders of magnitude higher in argillaceous and organic-rich sedimentary rocks (Wedepohl, 1978). The principal economic ore mineral is cinnabar (HgS), from which Hg is recovered by flotation and subsequent roasting of the crushed ore to yield HgO and ultimately free mercury. Mercury occurs in the atmosphere as a result of the natural degassing of the earths crust through the release of volcanic gases and evaporation from the oceans. Many activities of man account for substantial releases of mercury into the environment. One of these is the mining and smelting of mercury. In particular, the dumping of mine tailings or process wastes into lakes and rivers has resulted in the high levels of mercury in these bodies of water, as well as in sediments and biota. Hg in uncontaminated freshwater systems ranges from 5 to 50 ng/l rising to 0.1-60 mgA in geothermal waters (Fergusson, 1990) and 0.2 to 10.0 mg/l in mining contaminated surface waters (Pfeiffer et al., 1989; 1991). At moderate or high redox potentials, inorganic Hg is predominantly stabilised as Hg(OH)2 at pH>6, while at lower pH, HgOH' or chloro-complexes predominate depending on the ambient concentration of chloride. Free Hg2+forms a stable dissolved phase in acid waters at redox potentials of around 0.5V whilst free Hg and HgS phases are stabilised at redox potentials of <0.3V across a wide pH range. Mercury is a very persistent contaminant. In the metallic form it may persist in sediments for many years. Inorganic Hg in sediments is usually strongly adsorbed onto mineral particles and is not readily bioavailable. Methylation of inorganic Hg in lake, river and stream sediments is a key stage in the transfer of Hg into aquatic biota. Aerobic or anaerobic bacteria present in sediments are commonly involved in the methylation process with the highest rates of methylation occurring at relatively low redox potentials of +100 - -200 mV. The methyl species formed is strongly pH dependent, with methyl mercury ((CH3)Hg) most common under acid conditions and dimethyl mercury ((CH&Hg) in alkaline systems. The alkyVmethy1 species of Hg commonly constitute 0.5-3% of the total dissolved Hg in natural waters. Relatively small amounts of inorganic mercury are methylated by bacterial activity and may enter the food chain in this organic form, or even as inorganic Hg2+salts. Mercury also accumulates in the food chain and so can be concentrated in higher organisms, especially in fish which may be an important human food source.

2.2 Toxicity and human exposure

Mercury is a potent toxin with no known beneficial role in human metabolic processes. It's toxicity is strongly species-dependent,generally increasing through the sequence:- phenyl Hg salts > Hg2+salts > free Hg vapour > alkyl species. Of the methyl species, (CH3)Hg is considerably more toxic than dimethyl mercury. Adverse effects of Hg ingestion are evident with respect to both physiological and neurological processes, of which the degeneration of the central nervous system by (CH3)Hgis most pronounced. Human exposure pathways vary considerably, with both occupational and dietary factors potentially significant. Data compiled by Fergusson (1990) indicate typical Hg concentration ranges of 5-20 ng/g in most grain crops and 1-40 pg/g in vegetables. Methyl Hg is prone to gross bioaccumulation in aquatic systems, with relative levels for lake water, macrophytes, algae and fish reported by Craig (1982) to be 1: 1OOO: 2800: 23200 respectively. The risk associated with consumption of shellfish and fish is thus disproportionately high, due to the typical presence of >80% total Hg in methyl form. The average daily intake of Hg derived from food (30 pg/d) is an order of magnitude greater than that derived from water (Fergusson, 1990), for which a potable supply standard of 1 pgA is applied internationally. 2.3 Mercury contamination The human and ecotoxicological hazard associated with mercury (Hg) contamination of terrestrial and aquatic systems has been recognised globally since the 1950s, when the Minamata poisoning episode in Japan (arising from human exposure to methyl Hg in fish) triggered a tightening of legislative controls on mercury discharges 2

across Europe, North America and parts of south-east Asia. Whereas most mercury contamination previously resulted from hydrocarbon combustion and manufacturing processes in industrialised regions (Tables 1 and 2), significant quantities of mercury are now released into the environment as a result of its use for amalgamation by small scale gold producers. Table 1. Current global mercury additions to the biosphere (excluding the atmosphere) from natural and anthropogenic sources (adapted from Lacerda and Salomons, 1998) Annual additions to biosphere (tyr-')

Source Natural Agriculture and forestry Electricity generation and coal ash Solid wastes and weathering Urban waste

900 1960 4400 2780 1070

Table 2. Current global mercury emissions to the atmosphere from natural and anthropogenic sources (adapted from Lacerda and Salomons, 1998) Source Soil Sea spray Volcanic Forest fires Biogenic sources

AM& emissions (tyi')

50 20

loo0 20 1400

Natural sources (Total) Energy production Metallic ore refining and smelting Waste incineration Anthropogenic sources (Total)

2490

2260 130 1160 6040

Mercury has been used in gold and silver mining since Roman times and it is estimated that mercury released to the biosphere due to this activity may have reached over 260,000 t from 1550 to 1930 (Table 3), when mercury amalgamation was replaced by the more efficient cyanidation process. The 1970s gold-rush, sparked off principally by high gold prices, involved more than 10 million people in all continents and most of these were in the developing countries of the southern hemisphere. The mercury amalgamation process is used as a major technique for gold production in South America, China, Southeast Asia and in some African countries. It is estimated that mercury loss to the environment from amalgamation may be as much as 460 tonnes per year (Table 4). Gold mining is presently responsible for approximately 10% of the global Hg emissions from anthropogenic sources. Much of the estimated 300,000 tonnes of mercury released to the environment as a result of gold and silver mining during the last 500 years is still important in the global mercury Hg cycle insofar that it is susceptible to remobilization from abandoned mine waste and other land contaminated with mercury. Table 3. Estimated annual and total emissions to the environment from historic gold and silver production (adapted from Lacerda and Salomons, 1998)

Spanish colonial era Rio Carson, Nevada, USA California, USA North America (All) Bendigo Fields, Australia Brazil, Colonial Gwynedd, Wales, UK

El3

AM& emissions (tyr-')

1554-1880 1850-1900 1840-1900 1840-1900 1850-1930 1800-1960 1860-1916

292 - 1085 200 17 1000 11 5 0.1

3

Total emissions (tyi') 196,000 7000 1000 60,000 900 400 6

Table 4. Estimated emissions of mercury to the environment from present-day gold mining areas (adapted from Lacerda and Salomons, 1998) Gold mining area

Year production started

Annual emissions (ty i ’ )

Amazon, Brazil Mindanao, Philippines Rio Puyango, Peru Canada COCO, Colombia Pando, Bolivia N. Sulawesi, Indonesia Dixing, China Guyanesa Shield, Venezuela

1979 1985 1987 1976 1987 1979 1988 1992 1988

180 26 3 0.5 30 7-30 10-20 120 40-50

Total emissions (ty r-’1 3000 200 23 10 240 300 120 480 360

The amalgamation process requires the use of about 2 tons of mercury for each ton of gold recovered. Smallscale gold miners produce thousands of tons per annum so environmental contamination by mercury release is of considerable concern in many less developed countries. The most serious contamination occurs in Brazil where it is estimated that the annual release of mercury into the environment is approximately 180 tonnes which constitutes more than 80 % of total anthropogenic mercury emissions in Brazil. Approximately 136 t.yr(-1) of mercury from gold mining sources is emitted directly to the atmosphere.

2.4 A chemical time bomb?

At Minamata, the onset of severe human impacts more than a decade after the peak in environmental mercury pollution has led to the ‘Chemical Time Bomb’ concept. The concept has been evaluated critically by Lacerda and Salomons (1998) who emphasise the limited capacity of soils and sediments to act as permanent sinks for Hg. Contaminant mercury accumulated over long periods (ranging from decades to millennia), may be mobilised and bioassimilated in response to minor adjustment of soil or sediment organic content, redox state or acidity (which in turn regulate mercury methylation). Once mobilised, the ‘time-bomb’ effect is accentuated by Hg biomagnification during food-chain translocation, as described in section 2.2 (above). The credibility of the ‘Chemical Time Bomb’ concept within the context of mercury contamination from artisanal gold mining areas remains to be fully established. An insight into the potential for such deferred toxicological impacts can, however, be gained by characterising Hg dispersal trends and fates (i.e. sinks) in these settings. It is in this specific respect that the results of the current project may contribute to the wider debate.

3. PROJECT OBJECTIVES (OUTPUTS)

A number of developing countries in Africa, Asia and Latin America have experienced a rapid growth of the small-scale gold mining sector during the last two decades. This has reached huge proportions in the Philippines where more than 100,OOO people have been engaged in largely uncontrolled artisanal mining. There was a requirement in the Philippines in particular, for a detailed evaluation to be carried out of the environmental impact of mercury contamination derived from the mining both on the miners and the local communities. Smaller magnitude artisanal mining had also developed in southern Ecuador in the Nambija and Ponce Enriquez areas. As good collaborative links had been established with organisations in the Philippines and Ecuador that were able to provide both logistic and scientific support, these two contrasting examples of artisanal gold mining were selected as case study areas in which to fulfil the following project objectives (outputs):1. Provision of a generic protocol for monitoring the sources and fates of contaminant mercury in gold mining regions. 2. Assessment of human and ecotoxicological impacts of mercury pollution and calculation of humadenvironmental risk thresholds. 3. Promotion of improved methods of gold recovery with applicability to the artisanal sector. 4. Design and implementation of practical legislation for abatement of mining-related mercury contamination. It was originally intended to restrict monitoring and impact assessment studies to mercury contamination associated with small-scale gold mining in Mindanao, the Philippines and to the Ponce Enriquez and Nambija

4

areas in Ecuador. However, in August 1995, the main collaborating organisation in the Philippines, the Mines and Geoscience Bureau (MGB) was obliged to reappraise their priorities in the light of widespread media reports of human mercury poisoning in an area of former cinnabar mining on the island of Palawan. In response to a directive from Philippines President Fidel Ramos, the MGB commissioned an inter-agency investigation into the Palawan scare, and a formal request for the diversion of BGS-DFID funds from Mindanao to Palawan was received by BGS in September 1995. Although falling strictly beyond the remit of TDR project R6226 (which is primarily concerned with Hg contamination associated with artisanal gold mining), the request was considered by BGS and DFID to constitute a strong case for demand-led TDR expenditure. Accordingly, a preliminary BGS-MGB investigation of the extent of the Palawan Hg problem was sanctioned in December 1995, and executed with assistance from the UK Institute of Terrestrial Ecology (PE). In view of the paucity of geochemical, ecotoxicological or epidemiological data depicting the extent and magnitude of Hg hazards in central-eastem Palawan, the-principal objectives of the study were:1. Establish the spatial extent and magnitude of Hg contamination within the sediments of Honda Bay 2. Evaluate temporal trends of Hg deposition in the marine environment, with particular reference to flux adjustments associated with the onset of mining and the construction of the Sitio Honda Bay jetty. 3. Assess the risk posed to inhabitants of Sitio Honda Bay jetty and the PQMI mine locality as a consequence of living on a mine-waste substrate and hence the need to re-house the people at a less contaminated site. 4. Assess the extent of Hg bioassimilation and attendant toxicological stress in marine biota (fish and shellfish) at various localities in Honda Bay, and appraise the significance of Sitio Honda Bay as a contaminant source for biota. 5. Assess alternative potential sources of Hg exposure (e.g. potable water) 6. Prepare recommendations for remedial action, if appropriate.

4. WORK PLAN A work programme incorporating the following five major phases was designed to address the project objectives : Phase 1 : Collation of existing data and bibliographic review. A thorough bibliographic review was carried out at the beginning of the project in order to obtain relevant background information on monitoring protocols and for the human and ecotoxicological impact assessment. The review was updated prior to compilation of this report. Phase 2: Design and testing of monitoring protocol to identify mercury sources, transport pathways and fates. Monitoring in each of the selected case study areas was performed to quantify (1) the magnitude and extent of mining related mercury contamination in drainage sediments and water, and (2) the ultimate fate of mercury contamination. At the project outset, Eastern Mindanao was jointly identified by staff of the BGS and the Philippines Mines and Geosciences Bureau (MGB) as an appropriate focus for research due to the extensive gold-rush which has occurred in the region during the last two decades, with widespread utilisation of Hg for gold amalgamation. The sampling campaigns in the Ponce Enriquez and Nambija areas of Ecuador were executed as precursors to more comprehensive environmental monitoring scheduled under the World Bank funded PRODEMINCA project, sub-component 3.1 (Monitoreo de 10s impactos ambientales, socioeconomicos y sobre la salud, relacionados con las actividades mineras). In the Honda Bay (Palawan) case study, monitoring covered both the terrestrial and marine environments. Phase 3: Human and ecotoxicological impact assessment. The likely toxicological significance of mercury contamination was evaluated by comparing monitoring data with relevant water and sediment quality thresholds for human health, drinking water, aquatic biota and shellfish. In Palawan, the extent of mercury bioassimilation and attendant toxicological stress in marine biota (fish and shellfish) was assessed. The impact of mercury contamination on the local inhabitants in the Honda Bay area of Palawan and the Tagum area of Mindanao was evaluated through determination of mercury in hair samples. Phase 4: Evaluation of alternative gold recovery methods. Recommendations for improving the efficiency of fine gold recovery evolved from a comprehensive literature review. Gravity separation was recommended as an efficient alternative to mercury amalgamation for the concentration of fine gold. A simple BGS-designed gravity separator was constructed and tested under laboratory conditions using Au-bearing material (ore) collected from three sites in the Philippines. Following design modifications, the gravity separator was tested under field conditions and donated to the Philippines MGB as a model for the development and construction of full-scale production equipment.

5

Phase 5: Recommendationsfor environmental control legislation relating to mercury pollution. It had been anticipated that recommendations for maximum permissible environmental mercury levels in the Philippines would evolve from the biological and human impact assessments (Phases 2 and 3) and that these would be incorporated into new legislative controls on the small-scale mining sector being prepared by the MGB at the time the project started. It was also intended to recommend that the Governments of other developing countries afflicted by the impacts of small-scale gold mining should adopt these standards. Recommendations for increased restriction of amalgamation andor alluvial mining in environmentally sensitive areas (such as localities in which discharges potentially affect economically important fisheries in Ecuador, for example) were planned to be made to government authorities where appropriate and where substantiated by monitoring data (Phase 2). Practical (i.e. attainable and enforceable) legal standards for environmental mercury levels and discharges need to be complemented by the inclusion of monitoring guidelines within all new legislation. In the event, the new Ecuadorian and Philippine mining and environmental protection regulations and laws were pushed through much faster than had been envisaged, effectively precluding BGS’ involvement in this aspect of the project programme. A summary of the case studies that comprise Phases 2-4 of the workplan together with their associated output (Technical reports) are summarised in Table 5. Table 5 Project case studies and Technical Reports Country

Area

Site

Philippines

Eastern Mindanao Eastern Mindanao

Gango, Mainit, Diwalwal

Philippines

Upper Agusan (Diwalwal, Bango, Pasian, Mainit) Honda Bay

Date

Type of study

July 1995

Orientation survey

Output Tech. Report No. wc/95/72/R

2

November 1995

Monitoring

WC/96/61/R

2-3

December 1995 February 1996 July 1996 November 1996 January 1997

Monitoring and impact assessment Monitoring

Wc/96/31/R

Work Plan Phase 2

Philippines

Palawan

Philippines

Eastern Mindanao Ponce Enriquez Nambija

Agusan (N. sector)

2

Bella Rica, Rio Siete Nambija, Campanilla

2 2

Philippines

Luzon; Mindanao

4

Philippines

Mindanao

Acupan & Kias creek (Baguio; Luzon); Cagayan de Or0 (Gango, Mindanao) TagudAcupan , Davao del Norte

Philippines

BGS Laboratories

Philippines

Mindanao

Ecuador Ecuador

2-3

4 Benquet artisanal miningarea,-Baguio

4

April 1997 MarchDecember 1997 February 1998

WC/96/61/R

Monitoring Monitoring Collection of bulk samples for mineral beneficiation studies

WFV97/1

Monitoring and impact assessment Laboratory testing of gravity separator

No report

Field demonstration and testing of gravity separator

WC/97/61

WC/97/61

5. COLLABORATIVE FRAMEWORK

All aspects of the research programme were co-ordinated and managed by the BGS. Specialist support for the ecotoxicological impact assessment studies in Palawan was provided by the Institute of Terrestrial Ecology (RE), a component body of the UK Natural Environment Research Council (NERC). Field investigations in the Philippines were carried out in close collaboration with the Philippines Department of Environment and Natural Resources (Mines and Geoscience Bureau), the Puerto Princesa Hospital, Palawan and the University of the Philippines. Access to small-scale mining and processing operations was facilitated through the cooperation of the mine and processing-plant owners. In Ecuador the principal collaborators were CODIGEM (Corporacidn de Desarrollo e Investigacidn Geolbgico-Minero Metalrirgica), DINAPA (Departamento Nacional de Proteccidn Ambiental), DINAMI (Direccidn Nacional de Mineria) and the Swedish consultants of the World Bank PRODEMINCA project. Advice of the significance of mercury levels in humans was provided by toxicologists from Guy’s and St Thomas’ Hospital, London.

6

6. MINING RELATED MERCURY CONTAMINATION 6. I Philippines 6. I. I Gold mining areas: Mindanao

Since the discovery of the alluvial gold deposits in Eastern Mindanao during the 1980%the area has been subject to a series of gold rushes by small scale miners, with substantial communities (totalling > 1 0 0 , ~ ) centred on Diwalwal (Figure 1) and Compostella. The poorly-controlled use of mercury for gold amalgamation in these areas has resulted in numerous poisoning incidents amongst miners, the first of which was reported from Tagum, Davao del Norte, in 1987 and involved 11 injuries and 1 fatality. In the late 1980s, 53 gold and panning areas in 24 municipalities used mercury to process and recover gold. Studies on mercury contamination in eastern Mindanao were conducted by the Small Scale Mining Unit of the Mines and Geosciences Bureau (MGB) in 1989 in response to reports of mercury poisoning in the gold rush areas. It was shown that more than 90% of the total mercury pollution in the region derived from small-scale producers in Diwalwal, Compostela, Boringot, Tagum and Masara, all in Davao Province. The MGB estimated that 140 tons of mercury was released into the Agusan River catchment during the period 1986-88 (the peak of the Mindanao gold rush), after which discharges declined slightly due to a decline in artisanal mining activity, coupled with the increased use of cyanidation technology. The wider socio-economic and ecological implications of mercury use have been highlighted more recently, through the publication of data showing grossly enhanced mercury levels (>2 ppm) in fish from the Ngan and Agusan rivers, and in the estuarine environment of Batuan Bay, confirmed the ecological and potential human health hazards associated mercury use.

Figure 1. Location of artisanal gold mining areas in eastern Mindanao, Philippines (inset map shows the location of the main map area within the Philippines; dashed box indicates area of Figure 11; small filled circles indicate sampling sites on lower sector of Agusan River)

The major gold mining localities of Mainit and Diwalwal are located within the mountainous East Mindanao Ridge from which deeply incised rivers flow north and west into the low-lying gently undulating terrain of the Agusan-Davao trough to the west. The Mainit gold locality includes at least eight epithermal Au and three porphyry Cu prospects situated in the upper reaches of the Manat River catchment, plus an alluvial mining area some 5-10 km downstream. Au is recovered by sinking trenches into active stream alluvium (Plate l), into which sluice boards are emplaced. The heavy mineral residues on the sluice boards are subsequently removed and refined in a prospecting pan, with the addition of Hg to amalgamate gold (Cover illustration). 7

At Diwalwal (Plate 2), a world-class gold deposit was discovered by casual prospectors in 1983. At the height of the Diwalwal gold rush, the locality was occupied by over 100,OOO small-scale miners, but has since declined to approximately 50,000 as a consequence of major landslides in 1985 and 1988. The artisanal mining sector has control of one of the world’s most important gold occurrences while major mining companies are unable to gain access to the deposit (Guardian, UK, February 28, 1998). The mining area lies in the headwaters of a series of minor tributaries of the Mamunga River (known locally as the Naboc River), which subsequently joins the Agusan River (Figure 1) The gold veins are worked via a series of adits and drives, and the ore is crushed at co-operative rod mills. Much of the processing activity during the 1980s involved amalgamation. In the early 1990s, a centralised carbon-in-pulp (CIP) lime-cyanide leaching plant was established. The extraction-plant’s efficiency was claimed to give better than 95% recovery. Mercury, often present in older discarded tailings which are now being bought in and reprocessed by the CIP operators, is recovered at various stages in the processing and sold back to the miners for re-use. The recovery efficiency for mercury is not known. The total ‘official’ gold production for each of the last 10 years was of the order of 0.5 million oz (DENR, unpublished data). Actual production is conservatively estimated to be more than twice this figure giving a total gold production over the last 10 years of about 280 t. Assuming half the production involved amalgamation, this implies that nearly 300 t of mercury will have been used of which approximately 10% (30 t) will have been discharged to the rivers. Mining of placer deposits was initiated at Gango in 1960, but underground working of high-grade narrow vein systems started as recently as 1988. A minor gold rush into the locality was witnessed at this time, and a community of approximately 1400 was dependent on small-scale mining in 1990. Gold production in 1990 was estimated at 153 g/day, with 18 rod-mills operative for the processing of ore. Until the late 1980s, gold was amalgamated by the direct addition of Hg during milling in a tromol mill (James, 1994) after which the residues were discarded. More recently, small-scale CIP cyanidation plants have been developed locally, and are utilised on a co-operative basis. On account of the low-recoveries attained by amalgamation, tailings previously treated in this manner (and hence containing Hg-residue) are now being widely reprocessed by CIP operators.

-

6. I .2 Mercury mining Palawan

Cinnabar mining and tailings disposal in Honda Bay, Palawan, Philippines (Figure 2) led to reports of human mercury poisoning. Three villages located in the low-lying coastal area bordering on Honda Bay were reported to be at risk. Mercury, mainly in the form of cinnabar (mercury sulphide, HgS) had been extracted principally at a mine operated by the Palawan Quicksilver Mining Inc. (PQMI) located 3 km inland from Honda Bay from 1953-1976. Ore was extracted from an open pit and in excess of one million tons of tailings from the PQMI roasting plant was discarded along the Honda Bay coastline during the 1960s, and was ultimately used for the construction of a 400 m long jetty, Sitio Honda Bay (Figure 2), which provided a landing point for the fuel tankers which supplied the PQMI ore-roaster. Following the closure of the mine in 1976, a substantial community (approximately 200 inhabitants) settled on the Sitio Honda Bay jetty, mainly engaging in fishing.

sfi

Bacungan R.

HONDA BAY

Figure 2. Location of Sitio Honda Bay (SHB) Jetty, Palawan and sampling locations (grey diamonds) visited during the MGB Explorer gravity coring survey of Honda Bay in September 1995. Stations 2,7,9 and 25 (black diamonds) were subsequently re-sampled by BGS for high-resolution downcore Hg analysis. (PQMI = former Palawan Quicksilver Mining Inc. cinnabar mine).

8

6.2 Ecuador

In Ecuador, mercury pollution from gold mining, both within and along the coastal margin of the Western Cordillera, has attracted considerable concern with respect to downstream pollution, especially of the economically important coastal shellfish production areas north of Machala. An important DFID-SIDA-World Bank funded "Mining and Environment" programme was initiated in 1994 with the objective of developing and expanding an "environmentally responsible" minerals sector. The gold mining sector in Ecuador provides employment for a large number of people, the majority of whom are artisanal miners. 6.2. I Nambija

The major development of the informal mining sector in southern Ecuador took place in the 1980s in Nambija where up to 20,000 people adopted relatively inefficient mining and mineral beneficiation processes which involved the widespread use of mercury amalgamation. This resulted in extensive environmental degradation. The gold occurrences and associated mining settlements of Nambija and Campanilla lie approximately 25 km east of the southern Ecuadorian town of Zamora at an altitude of approximately 1800 m (Figure 3). Gold occurs principally in quartz-carbonate veins and stringers that follow NE-trending faults within a N-S trending skarnfield of Jurassic age. Gold mineralization appears to be associated with a later phase of hydrothermal activity that is focused along pre-existing structural trends within the skamified rocks. Common accessory phases include magnetite, sphalerite and chalcopyrite. Artisanal mining at Nambija and the neighbouring Campanilla settlement are focused within the Q. Calixto and Q. Cambana sub-catchment areas, both of which form headwater tributaries of the Rio Nambija. The drainage is deeply incised with channels to interfluve relief averaging 800 m. Climatic conditions are temperate, with an average diurnal range of 9-20°C and annual precipitation of approximately 2000 mm. Exploitation of alluvial and bedrock gold from Nambija can be traced to pre-colonial times (Navarro, 1986), however the present-day settlement was established as a consequence of a gold rush during the 1980s (Litherland et al., 1994). Gold-bearing skarn pockets are worked via a series of adits. Beneficiation is carried out primarily within the major river valleys, where a series of sluices have been constructed to process milled ore. Smaller-scale alluvial mining operations, mostly dependent on mercury amalgamation as a beneficiation method, have been developed at several locations downstream of the main artisanal settlement within the Quebrada Calixto and throughout the mid- and lower reaches of the Rio Nambija. Progressively falling production during the 1990s has prompted a decline in the Nambija population and the community currently numbers approximately 8,000. I

J.,

Figure 3. Location of the Ponce Em'quez and Nambija artisanal gold mining areas, Ecuador.

9

6.2.2 Ponce Enriquez The Ponce Enriquez mining district (Figure 3) was developed by artisanal miners during the mid-l980s, although the number of people involved and the social and environmental impacts are correspondingly less than at Nambija. It has been estimated that more than 2,000 people live in the Ponce Enriquez mining settlements of which about 1000 are involved in mining the Bella Rica sector. Most of the gold and associated minerals in the high temperature hydrothermal vein mineralisation are extracted by hard rock mining methods and the gold is concentrated gravimetrically following grinding in Chilean mills (trupiches).Gold is recuperated from heavy mineral concentrates by amalgamation, either in tromol mills (Plate 3) or from gravity concentrates. Cyanidation plants (Plate 4) have recently been installed at various sites, principally adjacent to the Rio Siete, to treat gravity tailings.

7. MONITORING THE DISPERSION OF MERCURY CONTAMINATION

The principal objective of the monitoring case studies was to fulfil output (1) ‘Provision of a generic protocol for monitoring the sources and fates of contaminant mercury in gold mining regions.’ Monitoring in each of the selected case study areas was performed to quantify (1) the magnitude and extent of mining related mercury contamination in drainage sediments and water, and (2) the ultimate fate of contaminant mercury. Potentially harmful element contamination derived from ore minerals associated with the gold or mercury mineralisation, including arsenic and copper, are also considered. The objective of these studies was to design and test a monitoring protocol which could be used in a range of climatic, topographic and geological environments to identify mercury sources, transport pathways and fates. This section of the report comprises a review of sampling and analytical protocols; an evaluation of within-site and temporal variation; and a description of the extent and magnitude of downstream dispersion in relation to the magnitude of the contaminant flux. 7. I Sampling and analytical methods

7. I. I Drainage surveys Sample sites were selected to obtain information on the magnitude and extent of downstream dispersion of Hg and other contaminants derived from mining and mineral processing activities. Filtered stream water, stream bottom sediment and heavy mineral concentrate samples were collected at each site. Wuter (w) Stream water pH, temperature, Eh and conductivity were determined in the field using appropriate temperaturecompensated electrodes and meters. Water samples for chemical analysis were filtered through 25 mm diameter, 0.45 pm Millipore (TM) cellulose acetate membranes into 30 ml Sterilin tubes or into acid-washed 30 ml HPDE bottles (NalgeneTM).The suspended particulate load of the mining contaminated rivers was very high at some sites and this caused severe problems with water filtration at some sites. Coarse prefilters were used, ahead of the normal 0.45 mm cellulose disks, on all obviously turbid samples and the filters retained for analysis in the Upper Agusan case study. At each site, the suite of water samples collected included: - (a) 30 ml spiked with 1% v/v HN03(Analytical Grade) for determination of major and trace cations (including K, Na, Ca, Mg, Sr, Cd, Ba, Si, Fe, P, B, V, MO,Al, Be, Zn, Cu, Pb, Li, Zr, CO, Ni, Y, La, Cr) by inductively-coupled plasma emission spectrometry (ICP-ES) and, for the Ponce Enriquez samples, high precision hydride generation A A S analysis of As, (b) 30 ml unacidified water for Sod, NO3 and C1 analysis by ion chromatography and (c) 30 ml spiked with 0.3 ml conc. HN03 + 0.3 ml 0.2 vol.% K2Cr07for total Hg analysis by Cold Vapour Atomic Fluorescence Spectroscopy (CVAFS). Blank water samples, made using distilled water (provided by the BGS laboratories) and appropriate aciddreagents, were submitted in blind fashion to the BGS ICP-ES and CVAFS laboratories in conjunction with field samples. Results for these blanks confirmed that there was no detectable background enhancement caused by the addition of aciddreagents to the field samples with respect to the elements of major interest to the present survey: As, Cd, Cu, and Hg. On Palawan, Total Hg analyses were conducted following pre-treatment of one aliquot with a brominating agent to oxidise all organo-mercury compounds to inorganic species. The difference between the Hg value recorded for this sub-sample and that for the corresponding non-brominated aliquot was assumed to constitute

10

organo-Hg. For both aliquots, the oxidant preservative (K2Cr07) was reduced using a dilute solution of NH2.0H.HCl immediately prior to analysis. Stream bottom sediment (BS)

Bottom sediment samples of 100 to 200 g mass were collected by wet-screening active stream-bed sediment through a e100 BSI mesh (e150 pm) sieve, using a minimal amount of water to avoid the loss of fine silt and clay fractions. Samples were then sealed in securitainers to avoid evaporative losses and oxidation. Suspended particulate matter (SPM)

In the Ponce Enriquez survey, suspended particulate matter samples were obtained by filtering a fixed volume of stream water (250 or 500 ml depending upon the amount of SPM) through 45 mm diameter, 0.45 pm MilliporeTMcellulose acetate membranes using a hand operated Mityvacll vacuum pump. Filter membranes were carefully removed to avoid contamination and stored in 30 ml Sterilin tubes. In the Agusan drainage survey, SPM samples were collected on the smaller 25 mm. diameter filters used to filter water samples. Heavy mineral concentrates (HMC)

Heavy mineral concentrate samples were collected by screening approximately 2 kg of <2mm stream bottom sediment, and removing the lighter fractions in a conventional prospecting pan. Mercury analyses of stream bottom sediments and HMC were carried out by CVAFS, using 1 g milled subsamples digested at 4 0 ° C in aqua regia. SPM samples were dried and the sediment digested with the cellulose filter membrane in aqua regia at 4 0 ° C and Hg determined by CVAFS. The practical CVAFS limit of determination for Hg in solid samples is 0.02 mgkg. Multi-element analysis of bottom sediment and SPM samples was carried out by ICP-AES using an aliquot of the aqua regia digestate. For bottom sediments collected during the Nambija survey and the Mindanao pilot study, major oxide and trace metal characterisation was carried out by X-ray fluorescence (XRF) analysis of 12 g pellets made following disaggregation, ignition of organic matter and milling of sediment to e63 mm. In the Nambija survey, heavy mineral concentrates were visually examined in the field to establish the presence of metallic Hg-Au amalgams and additional features of environmental significance. No quantitative analyses were performed on these samples. 7. I .2 Marine sediment survey: Palawan

Marine sediment cores from deep-water sites within Honda Bay were sampled by the MGB using a gravity corer and estuarine and nearshore sites using a hand auger. All cores were initially sub-sampled at 10 cm intervals. Cores from shallow water sites near Sitio Honda Bay and in the Tagburos estuary (i.e. the major postulated sources of Hg contamination) were re-sampled at 1-2 cm resolution together with a core from a Honda Bay site located approximately 20 km offshore which was selected to provide an indication of bay-wide impacts. Core from a control site, which receives sediment from unexploited cinnabar deposits, was also sampled. If a typical nearshore rate of 1 &yr. is assumed, the uppermost 10 cm sub-sample analysed from each coring station could represent 100 years sedimentation. This served to provide both a detailed record of Hg fluxes to Honda Bay during the past century, and also an independent corroboration of existing MGB analytical data. Core sub-samples of approximately 4 g mass were analysed for total and inorganic Hg following the preparation and analytical methods used for bottom sediments. 7.2 Monitoring results for gold mining areas

In this section, the relative merits of a range of sample media, including stream water, bottom sediment, heavy mineral concentrates derived from bottom sediment, and suspended particulate matter, are considered with respect to: within-site and temporal variance in contaminant concentrations; sensitivity to contamination including the extent of the dispersion detected by each sample medium; the extent and magnitude of downstream dispersion in relation to the magnitude of the contaminant flux.

11

The magnitude of within-site variance of contaminant concentrations is compared with temporal variation related to (1) changes in water flow and (2) variations in the fluxes of soluble and suspended contaminants, including both mercury derived from amalgamation processing as well as other potentially harmful contaminants derived from ore processing including As and Cu. The relationship between within-site, between-site and temporal variance is important in any assessment of the relative merits of the sample media for detecting and mapping the extent and magnitude of contamination. 7.2. I With-site variance

Within site variance was tested only in the Ponce Enriquez area (Appleton et al., 1996) where within-site variance is very small compared with between-site variance (Table 6). The greatest within-site variance was recorded for Hg in HMC, which reflects the low Hg concentrations and the ‘nugget’ effect caused by the occurrence of Hg in discrete particles of Au-Hg amalgam. Table 6. Analytical data for sample pairs taken at the same site, Ponce Em’quez area, Ecuador (Appleton et al., 1996)

6 7

0.02 0.01

1.09 1.38

3.00 3.00

3.00 3.00

16332 22626

3389 4383

2.56 2.98

42435 46049

8327 9134

12 13

0.01 0.01

0.01 0.02

0.69 0.83

1.00 0.09

427 25 1

460 436

0.18 0.17

129 157

523 601

19 20

0.01 0.01

0.32 0.26

0.85 1 .00

0.1 1 0.16

2660 2624

976 945

0.18 0.19

1354 1539

1423 1508

31

32

0.02 0.02

1.61 1.39

7.00 8.00

0.32 0.94

5647 5400

1363 1326

1.oo 0.83

7780 7096

4266 4307

35

0.01

36

0.01

0.32 0.48

0.86 0.58

0.14 0.08

27 1 296

1024 1121

0.94 0.92

282 23 1

1651 1455

7.2.2 Temporal variance

Temporal variation of contaminant concentrations in river water is particularly problematic in the case of anthropogenic sources. The baseflow dominated hydrogeochemical signatures which make data robust at variable discharges for lithospherically derived elements (e.g. Simpson et al., 1993) have not been recorded in the case study areas. There are, as would be expected, complex dilution effects related both to seasonal and daily variations in rainfall as well as to temporal, short and long-term variations in fluxes derived from mineral processing plants. Ideally, sampling should be carried out at constant discharge rates, which should, if possible be quantified. Rivers that are subject to strong fluctuations in discharge rates related to rainfall will show equally strong fluctuations in contaminant concentrations, especially in water and SPM samples. The relationship is further complicated by the initial flushing of contaminants at the onset of heavy rain following prolonged dry periods. Hence, the initial rise in discharge may cause an elevated dissolved contaminant level, which then rapidly tails off. This is not significant with respect to Hg, but may be with respect to elements such as As, Cd, Cu and Zn. Temporal variability with respect to bottom sediments is more complex. Large temporal variation would undermine all geochemical reconnaissance and contaminant monitoring surveys and there is a general assumption in natural systems that a steady state prevails. Again, the situation is rather more complex for contaminants from mining sources as the discharge rate tends to be erratic and bottom sediment concentrations may reflect changes in contaminant fluxes. Variability for suspended particulate matter (SPM) is totally discharge related so that monitoring results may be extremely difficult to compare.

12

Thus, in theory, stream bottom sediments should be subject to the least short-term temporal variation in contaminant concentrations, although major changes in contaminant fluxes are likely to affect concentrations in this sample type. In the Ponce Enriquez area, there is some evidence of short-term temporal variation in the contaminant load of water and suspended particulate matter. This would, of course, be expected, as the release of overflow water from the mineral processing and cyanidation plants is not constant. As sampling was executed over a period of days, fluctuations in the load of suspended particulate matter reflect the level of discharges from grinding and processing plants at the time of sampling (see Figure 4). Such fluctuations are also mirrored in contaminant concentrations in both the river water and in the SPM. At 8 km above the shrimp ponds on the Rio Siete, for example, there is a peak of SPM load which is also reflected most strongly by As and Cu in SPM, as well as As, Cu and Na in water. The peak is not seen in the bottom sediment or HMC contaminant concentrations (see also Figs. 13 and 14).

6 5

1

P

a

0 0

2

4

6

8

10

12

14

16

18

Mstrnca above shrlmp ponds (lon)

Figure 4. SPM load - Rio Siete, Ponce Enriquez area, Ecuador

Longer term temporal variations may be evaluated at five sites in Mindanao (two from the Mainit and three from the Diwalwal mining areas) which were sampled in November 1995 and February 1996. At all sites, Hg in BS and HMC increased from November to February whereas Hg in water was the same or slightly lower, apart from one site where the Hg concentrations had reduced from 2906 to 0.86 ppb (Figure 5a-d). No SPM data are available for these sites. Increased Hg in bottom sediment and heavy mineral concentrates suggests that the amount of Hg contamination had increased over the period of 4 months. Substantially higher concentrations in HMCs suggest that much of the Hg is in metallic form or Au-Hg amalgam. Sampling in February took place after a period of very heavy rains which may have resulted in much mercury contaminated material from tailings dumps being washed into the drainage system, thereby increasing contaminant levels in bottom sediments. C1, Na and SO4 at the two Mainit sites are an order of magnitude higher in the February samples compared with the November samples (Figure 5d). This implies that at the time of sampling in February, the rainfall component of the surface drainage water flow was lower compared with the flux from the hot springs at Pantukan. No marked season variations in rainfall characterise Eastern Mindanao that is subject to periods of very heavy rainfall after which both contaminant and base flow signatures in water and SPM will be strongly diluted. ~

~~~~~

~

~~~~~

. . .

.

1

Figure 5. Temporal variation in (a) Hg and (b) C1 in river water; (c) Hg in bottom sediment and (d) Hg in heavy mineral concentrates for the Diwalwal and Mainit mining areas, Mindanao, Philippines.

13

Concentrations of Cu in water samples tended to be lower in samples collected in a period of high rainfall and river flow (March-April) compared with samples collected in the drier month of July in the Ponce Enriquez area (Figure 6) although other elements demonstrated a less consistent relationship reflecting the high temporal variance in contaminant fluxes in water. Hg and Cu in bottom sediments collected at the same sites for different sampling programmes agree quite well in the both Ponce Enriquez (Figure 7) and Zaruma (Figure 8) areas. However, other elements showed very disparate concentrations reflecting the large differences in contaminant concentrations resulting from fluctuations in the amounts of finely ground sulphide minerals deposited from the outflow water of the mineral processing plants. Replicate sampling at the same site on the same day indicates low within-site variance. However, replicate sampling at the same site at different times clearly results in greater variance in contaminant concentrations for some contaminants in both water and bottom sediments.

0

1

2

3

4

5

BGS

Figure 6. Cu in river water collected in high flow (March 1996) and low flow (July 1996) conditions, Ponce Enriquez, Ecuador.

&

10000 8000 6000 4000

0

I

0

0

10000

5000 BGS

Figure 7. Hg and Cu in bottom sediments collected at the same sites (GIMP 1995 and this study, July 1996); Ponce Enriquez.

I 0

0.5

1

1.5

2

2.5

3

3.5

1996

Figure 8. Hg in bottom sediments collected at the same sites in 1995 (Maldonado, 1995) and 1996 (this study); Zaruma, Ecuador.

14

7.2.3 Downstream dispersion of contamination from artisanal gold mining areas

In the case study areas, sampling was carried out in drainage catchments directly impacted by contemporary mining together with hydrochemically appropriate control stations within unmined tributaries. The most obvious indicator of mining activity is the grey, highly turbid nature of the rivers draining the mining areas, in contrast to the clear waters of undisturbed streams. In the Ponce Enriquez area, Ecuador, large volumes of gravity tailings which contain very high concentrations of copper (chalcopyrite) and arsenic (arsenopyrite), in addition to being contaminated with mercury, are stored on the steep slopes of the Bella Rica mining sector. The tailings dumps are prone to erosion by heavy rainfall that washes the material into the Rio Siete and Estero Guanache drainage systems (see Figure 12 for Ponce Enriquez drainage plan). Most of the surface drainage water in the area, including the Rio Siete, is very turbid due to the high load of very fine-grained gravity tailings. Acid mine drainage with attendant iron oxide precipitation is less common (Plate 5). Suspended particulate matter loads of up to 6.2 g/1 were recorded during the present survey, decreasing to 1.0 g/l in the Rio Siete at the Pan-American highway and 0.02 g/1 immediately above the shrimp ponds (Figure 4). Similar grey, highly turbid waters characterise rivers in Mindanao where logging and other physical disturbance also causes turbidity, though this is usually buff-brown in colour due to eroded soil. Water

In the two case study areas in Ecuador, mobilisation of Hg in the aqueous phase is negligible and even in the much larger Diwalwal artisanal mining area in Mindanao, Hg in filtered samples, although locally extremely high, has a relatively short downstream persistence, probably as a result of dilution and adsorption. In the Ponce Enriquez area, Hg exceeded the detection limit (0.02 pgA) in only two filtered stream water samples from the Rio Siete that drains the main mining and mineral processing area of Bella Rica. The background range for fresh (unpolluted) waters is 0.005 to 0.05 pg/l; thus none of the water samples from the Rio Siete exceed this range. Hg in stream water below this anomalous site decreases to ~ 0 . 0 2pg/l within a distance of 2 km.Recent working of alluvial gravels at one site gave rise to the only high Hg value in filtered water (0.9 pg/l). Generally, low Hg concentrations in filtered river water also characterise the Nambija area with occasional higher concentrations of up to 0.1 pg/l reflecting local influx of contamination (Figure 9). The only strongly anomalous value (2.84 pg/l) in the Gango area (Mindanao) was recorded in outflow water from a CIP plant where the high Hg concentration can probably be ascribed to the presence of Hg in the processing feed, much of which has previously been subjected to treatment by amalgamation. The chemistry of Au-Hg amalgams in cyanic solutions requires further investigation, but initial indications are that Hg is effectively mobilised following the formation of AuCN complexes from formerly amalgamated Au. Regulations and guidelines to control the Hg hazard associated with CIP discharges from plants treating tailings previously subjected to amalgamation processing may therefore be warranted. 100

I"

I 0 1 1

,

~

' --4 , , ,\,-/

0 01

Figure 9. Distribution of Hg in bottom sediment and water in rivers draining the Nambija and Campanilla artisanal gold mining areas, Ecuador.

As would be expected, much higher concentrations occur in rivers draining the major Diwalwal gold mining area reaching a maximum concentration of more than 2000 pg/l immediately below the mining area but de15

clining rapidly to about 5 pg/l some 10 km downstream and to 0.1 pg/l near the confluence with the Agusan River. Thus Hg in water at Diwalwal is much higher than typical concentrations of Hg in waters polluted by mining and mineral processing activities in North and South America which fall in the range 0.2 to 13 p d l (Pfeiffer et al. 1989, 1991; Lacerda and Salomons, 1998). Sulphate produced as a consequence of the oxidation of sulphide minerals associated with the gold mineralisation dominates the ionic balance, especially at Ponce Enriquez and Nambija. Elevated levels of NOs in stream and river water collected close to the mining settlements of Diwalwal, Bella Vista (Ponce Enriquez) and Nambija reflect high inputs of organic material (sewage) and eutrophication. TOC is also elevated in the immediate vicinity of Diwalwal but levels return to background a couple of kilometres downstream. High Hg solubility is attributed to a substantial humic organic loading to the drainage (emanating from an artisanal settlement of approximately 50,000 inhabitants) and the consequent incorporation of Hg2+into soluble organo-metallic complexes. The formation of Hg-CN complexes in cyanic drainage water may also be significant. The determination of TIC, TOC, humic organics, CN’ in water would aid the interpretation of Hg concentration in filtered water. At some sites in Mindanao and Ecuador, Na derived from CIP plant effluent is the dominant cation. Suspended prticulate matter (SPM), bottom sediment and heavy mineral concentrates

Hg concentrations in drainage sediment are a function of deposition and inwash of Hg formerly mobilised as vapour during ‘torching’ of amalgams (Lacerda and Salomons, 1998) as well as Hg inputs derived from tailings from amalgamation processing of gold concentrates and tromol mills. Hence, the highest concentrations are observed closest to these operations in all the study areas with particularly high concentrations observed especially close to ongoing or recent processing of alluvial gravels. In the Ponce Enriquez area, for example, Hg in suspended sediments, stream bottom sediments and heavy mineral concentrates range from 0.01 to 9.61 mgkg, 0.1 to 13 mgkg and 0.01 to 5.0 mgkg, respectively. Slightly higher concentrations characterise bottom sediments in the Nambija area (Table 7 and Figure 10). The level of Hg contamination is much higher in the artisanal gold mining areas of Eastern Mindanao where maximum values of 34 mgkg, 40 mgkg and 62 mgkg have been recorded in bottom sediment, suspended sediment and heavy mineral concentrates, respectively (Table 7). Much lower concentrations occur in the smaller Bango and Mainit mining areas, Mindanao (Figure 11). Background concentrations in uncontaminated areas typically range from 0.1 - 0.4 mgkg in bottom sediments whereas in other mining contaminated regions Hg ranges up to 7.4 mgkg in N. Carolina (Callahan et al., 1994) and 25 mgkg in Brazil (Lacerda and Salomons, 1998; Reuther, 1994).

I

4kn

I

Figure 10. Distribution of Hg in bottom sediment in rivers draining the Nambija and Campanilla artisanal gold mining areas, Ecuador.

16

Downstream variation of Hg in stream bottom sediments at Ponce Enriquez is erratic ranging from 3 mgkg immediately below the main Bella Rica mining and mineral processing area to 3 mgkg some 15 km downstream of Bella Rica, at a site which is only 2 km upstream of the eastern margin of the commercially important shrimp ponds (Figure 12). Hg in suspended sediments and heavy mineral concentrates decrease more progressively over the same transect (Figs. 12 and 13). Hg in suspended sediments decreases by a factor of nearly 100 over a distance of 15 km, suggesting that the contaminant load at the time of sampling was low. However, the relatively high Hg concentration (3 mgkg) in stream bottom sediment 15 km below Bella Rica suggests that the average level of contaminant transfer may have been higher, presumably during periods of higher rainfall.

PASIAN

BATAAN MAlNlT

PANTUKAN

0

5xm

U ,12566'

,12616

Figure 11.Distribution of Hg in bottom sediment, Upper Agusan, Eastern Mindanao.

Observation of Hg-Au amalgam particles in heavy mineral concentrates indicates that much of the Hg dispersion is in an inorganic state, especially in the highly siliceous and aluminous, apparently low TOC, sediments

17

of the Nambija and Ponce Enriquez areas (Appleton et al., 1996; Williams and Orbea, 1997). Bottom sediments in both these locations are dominated by recently deposited, finely ground tailings material from the mining operations (Plate 6) which have been mechanically transported along the river bed or deposited from material transported in suspension (SPM). High concentrations of Hg in suspended particulate matter (SPM) indicate that Hg adsorbed onto mineral surfaces will make a significant contribution to the Hg load of bottom sediments. In some areas, there is a close correlation between Hg in SPM and water and also between Hg in BS and HMC (Figure 15 and 16).

I

I I

I

LShrimp ponds \

--

,

- .-., . .-----,-----___I 8.

,-_---I

I

,

t

3km

-%Shrimp \ \ \

10-15 mg/kg

0 0 0 0

5.0 - 9.9 mg/kg

-

2.0 4.9 mgncg

0.5- 1.9 mgkg

c0.5mykg

Figure 12. Hg in bottom sediment, Ponce Enriquez.

Distance below Bella Rica (km)

Figure 13. Downstream dispersion of contaminants in bottom sediment and heavy mineral concentrates, Rio Siete, Ponce Enriquez.

18

+SWFe(%) ~~~

1

0.1 0.01

0

2

6

4

8

10

12

14

16

Distance below BeWa Wca (krn)

Figure 14. Downstream dispersion of contaminants in suspended particulate matter (SPM), Rio Siete, Ponce Enriquez. 100

* *

0.001 J 0.01

100

1

loo00

w-Hs (PPm) Figure 15. Relationship between Hg in SPM and water, Upper Agusan, Mindanao.

1 00

+ *** ***

0.01

4

0.001 0.01

** 0.1

1

10

100

Figure 16. Relationship between Hg in BS and SPM, Upper Agusan, Mindanao.

Downstream dispersion trends

Hg and associated contaminants such as As and Cu have approximately logarithmic downstream dispersion patterns in many of the case study areas (Figs. 13 and 14). In the Mamunga River draining the Diwalwal mining area, the rate of downstream decline in Hg is, after large fluctuations in the first 2-5 km,almost equal in 19

SPM, water and HMC whereas Hg in bottom sediments declines much less steeply maintaining a relatively high concentration of about 20 mgkg for nearly 15 km (Figure 17). Extremely high concentrations of Hg in solution, reaching a maximum of 1539 ppb, decline rapidly to 7 ppb in the first 5 kilometres below Diwalwal as a result of the combined effects of dilution, sorption onto SPM and bottom sediment and volatilisation. The bulk of the contaminant load is in the SPM, being a factor of loo0 times more than the Hg in solution. It is assumed that Hg is either fine particulate Hg or more likely is sorbed onto silt and clay particles. Historic contamination is indicated at those sites where Hg in water and SPM are extremely low whereas high levels occur in the bottom sediment. The last sample site on the Mamunga River is actually within the floodplain of the Agusan River, which possibly explains why Hg in bottom sediment falls so dramatically at this point where contaminated sediment from Diwalwal is diluted by the relatively uncontaminated sediments in the Agusan River. Hg in SPM and water decline progressively downstream in the Agusan River although Hg is bottom sediment increases slightly at the next sample site 50 km downstream before declining gradually over the next 70 km (Figure 18). It is possible that Hg is enhanced as a result of adsorption onto organic matter in the bottom sediments within the relatively slow-flow regime of this section of the Agusan drainage basin. Alternatively, Hg may be slightly enhanced as a result of historic contamination. The downstream dispersion pattern on the Manat-Agusan River system below the much smaller Pantukan and Mainit mining areas declines steeply over the first 20 km before rising sharply below the confluence with the Mamunga River which introduces high levels of contamination from the Diwalwal area (Figure 19). The relationship between Hg in bottom sediment with Hg in SPM and HMC provides some indication of the downstream sedimentation processes. Hg in bottom sediment accumulates as a result of direct sorption from solution and by sedimentation of SPM derived from mineral processing activities. The ratio between Hg in bottom sediment compared with SPM and HMC increases progressively in both the Mamunga-Agusan drainage system in Mindanao and the Rio Siete in Ecuador (Figs. 20 and 21). HMC Hg, mainly in the form of particles of Au-Hg amalgam, is generally high close to the source of contamination, as it is heavy and not readily transported by the river. In the Rio Siete area, HMC is generally less than Hg in bottom sediment suggesting that much of the Hg is adsorbed onto fine mineral particles derived from mineral processing rather than as discrete particles of Au-Hg amalgam. Adsorption onto organic matter may be a factor, although no TOC data is available to assess this. At the time of sampling, which was characterised by relatively low river flow, SPM in these low flow conditions tends to decline fairly rapidly downstream as the water becomes clearer as a result of sedimentation. In contrast, the Hg concentration in bottom sediment declines less rapidly downstream reflecting more extensive contamination created during periods of maximum river flow and also the adsorption, over a long period, of Hg onto bottom sediments. It is assumed that the ratio between Hg is bottom sediment and SPM would not increase so rapidly downstream at times of maximum flow as the relatively heavy SPM contaminated with Hg would be transported further in a more dynamic hydraulic regime. Thus the relatively high concentrations of contaminants in bottom sediment in the Rio Siete some 2 km above the shrimp ponds indicates that contaminated sediment is carried further during periods of high flow. In the Rio Siete, the BS/SPM ratio for Hg is significantly higher for Hg compared with As, Cu and Fe (Figure 21) reflecting the differences in specific gravity of contaminant particles. In addition, the amplitude of fluctuations in both element and ratio dispersion patterns is greatest for Hg reflecting the generally low concentrations as well as the likely erratic distribution of Hg-rich particles within the bottom sediment. 4 W=1539ppm 50

40

30

t” 20

10

0 0

4

8

12

16

20

24

aStwm (km) below Diwalwal mining mntre

Figure 17. Downstream dispersion of Hg in the Mamunga River below the Diwalwal mining region.

20

1000

100

10

2

1

0.1

0.01

0.001 0

40 60 80 100 120 Distance (km) below Diwalwal mining area

20

140

160

Figure 18. Downstream dispersion of Hg in the Mamunga and Agusan Rivers below the Diwalwal mining region.

0

20

40

60 80 100 120 140 Distance (km) below Mainit mining area

160

180

Figure 19. Downstream dispersion in the Manat-Agusan river system below the Pantukan-Mainit mining areas

I

100

10

1

0.1

0

10

20

30

40

50

60

70

80

90

100

110

Distance (km) below Dlwalwal mining centre

Figure 20. Relative concentrations of Hg in bottom sediment (BS), suspended particulate matter (SPM) and heavy mineral concentrate (HMC) in the Mamunga-Agusan drainage system

21

0 0

2

6

4

8

10

12

14

16

Distanm below Blla Rica (km)

Figure 21. Relative concentrations of Hg, As, Cu and Fe in bottom sediment (BS) and suspended particulate matter (SPM) in the Rio Siete, Ponce Enriquez area.

Relative merits of sample media

Water and SPM samples indicate the current flux of contamination to the drainage system but are very susceptible to temporal variations related to short term fluctuations in discharges from processing plants and to rainfall-related dilution effects. Stream bottom sediments provide a more stable indication of the extent and magnitude of contamination whereas HMC samples, although subject to greater within site sampling variance, provide an indication of the amount of particulate Hg metal and Au-Hg amalgam in the bottom sediment, as well as of heavy minerals, such as arsenopyrite, chalcopyrite and galena which are associated with the gold mineralisation in some areas. Of all the sample media, the fine fraction of the bottom sediment provides the most sensitive and robust sample medium for evaluating the extent of contamination and likely hazards to biota from remobilization of Hg in bottom sediments as a result of methylation processes. Water and SPM provide a more relevant indicator of contaminant fluxes at the time of sampling and hence the potential hazards to biota and humans via this exposure route. 7.2.4 Magnitude of mercury contamination The extent and magnitude of downstream dispersion reflects the magnitude of the contaminant flux. This is clearly demonstrated by data presented in Table 7 and Figure 22 based on information for Diwalwal and Gango in Mindanao and Ponce Enriquez and Nambija in Ecuador, for which a relatively reliable estimate may be made of the number of miners involved. The relationship between number of miners and contamination is based on measurements made approximately 10 km downstream of the main mining areas as concentrations in the immediate vicinity of the mining areas are subject to large spatial and temporal variance. There is a broad correlation between the estimated amount of contamination and Hg concentrations in both water and bottom sediment (Figure 22) although the closeness of the relationship may in part reflect the similar topographic and hydraulic regimes in the four areas.

15

5

0 0

2.000

4,000

6,000

8,000

10,000

Umber of peopb actively invoked in gold e x t r u l i n

Figure 22. Relationship between estimated number of people actively involved in gold extraction and the level of Hg contamination in bottom sediment (BS). (D: Diwalwal; P: Ponce Enriquez, Ecuador; N: Nambija, Ecuador; G: Gango)

22

Table 7. Summary statistics for Hg in sample media compared with estimated number of people involved in gold mining and processing Diwalwal, Mindanao

Gango, Mindanao

Nambija, Ecuador

10,000

350

2,000

Max (av.) 10 km

34 (20) 20

25

34 (8) 5

13 (4.5) 3

Max (av.) 10 km

1539 (100) 0.5

2.6 <0.02

0.1 (0.025)

0.01

0.9 (0.04) 0.01

Max (av.) 10 km

40 (40) 8

na na

na na

9.6 (5) 0.8

Max (av.) 10 km

62 (18) 5

5 0.01

na na

5 (2) 0.3

Max (av.) 10 km

na na

nd nd

na na

470 (320) 280

Max (av.) 10 km

344 (200) 45

nd nd

3.5 <3

7277 (3900) 800

Max (av.) 10 km

11

333 26

34 (30) 25

44OOo (25000)

<2

Max (av.) 10 km

67 24

188 78

409 (350) 230

9134 (7600) 2850

Max (av.) 10 km

65 23

1078 49

42 (30) 20

666 (400) 230

MINING AREA

1

Ponce Enrfquez, Ecuador 1,000

5500

Note: BS = bottom sediment; W = river water; SPM = suspended particulate matter; na = no data available; nd = not detected

Mercury concentrations in bottom sediments and river water are generally higher than those recorded from Brazil, especially at Diwalwal. For example, near PoconC in the Mat0 Grosso area where about 5000 workers have been employed in 130 gold mines since the 1980s, Hg reaches a maximum of only 160 ng/l in filtered water and 198 ng/g (Tumpling et al., 1995) to 1.85 mgkg (Rodrigues and Maddock, 1993, 1997) in bottom sediment. In the Camaquii River Basin, Rio Grande do Sul state, Brazil, an area of abandoned gold working where mercury had been used extensively for amalgamation, sediment concentrations average 0.15 mgkg with a maximum of 0.43 mgkg (Pestana et al., 1997). These values are low compared to corresponding values for 2 19.8 pg/g in sedisome other gold mining areas in the Amazon basin ( ~ 0 . 0 1- 12.9 pg/l in water and ~ 0 . 0 to ments; Lacerda and Salomons, 1998). Malm et al. (1990) reported much higher sediment concentrations of up to 157 mgkg Hg in the Madeira River, Brazil where Hg had been use extensively for amalgamation in gold mining. However, data for numbers of people involved in mining and mineral processing are not readily available for these areas so the amount of Hg contamination in relation to mining activity can not be compared to the data for the present study. However, the generally low levels of Hg contamination in the Amazon drainage system where over 600,000 workers are reported to be involved in small scale gold production (Lacerda and Salomons, 1998) compared with values recorded in the Philippines and Ecuador, where much smaller numbers of miners are involved, may reflect the more mountainous topography in these two countries compared with the mature topography of the Amazon region. Dispersion in a fluvial environment would tend to be greater in fast flowing rivers. For comparison, bottom sediment concentration of about 2.5 mgkg have been recorded in the Carson River 50 km downstream of the Comstock load historic Au-Ag mining area where an estimated 7000 short tons of Hg were released to the Carson River between 1860 and 1900 (Lechler et al, 1997). Higher concentrations of Hg are associated with mercury mines of Monte Amiata in Italy (228 mgkg), Suplja Stena near Belgrade, in former Yugoslavia (up to 6000 mgkg) and Idrija near Ljubljana, Slovenia (up to 1000 mgkg falling to 5-300

23

mgkg some 5 km downstream of the mine and smelter; Gosar et al. 1997). Mine spoil mercury concentrations of >loo0 mgkg decline to only 8 mgkg some 4 km downstream of cinnabar mine in Palawan (Williams et al., 1996). These lower values in Palawan may reflect the smaller size and shorter life of the Palawan Hg deposit compared to the Idrija mines which produced 105,000 tonnes over a period of 500 years (Gosar et al., 1997). In comparison, pollution of rivers by other industrial processes has produced average and maximum sediment concentrations of 20 mgkg and 157 mgkg, respectively, in the Elbe River (Wilken et al., 1990), and 0.7 to 20 mgkg in sediments contaminated by effluent from a chlor-alkali plant near Rio de Janeiro, Brazil (Moriera and Pivetta, 1997). Extremely high concentrations of As and Cu characterise the drainage systems in the Ponce Enriquez area because of the large amounts of arsenopyrite and chalcopyrite associated with the gold mineralisation. In contrast to both Ponce Enriquez and mining localities characterised by acid drainage, conditions downstream of Nambija, Diwalwal and Gango (Table 7) are the predictable function of a low toxic trace element component within the mineralised assemblage (and hence the process waste) coupled with a strongly-buffered regime which inhibits mobilisation of metals such as Pb, Zn and Cu.

7.3 Monitoring results for mercury mining area: Palawan 7.3. I Hydrogeochemical survey

Analyses of aquifer- and stream water samples from the study area provided no evidence of Hg contamination, with values typically below 40 ng/l. Mercury was in all instances found to be present as >99% inorganic species. The role of potable water as a source of human Hg exposure is thus likely to be negligible. Previously reported high concentrations of Hg in the range 2-4 kg/l can probably be explained by the fact that the samples were unfiltered, and thus probably held variable suspended loads containing Hg. 7.3.2 Stream sediment

Extremely high Hg levels (1533 mgkg) in sediments emanating from the PQMI mercury mine tailings dump (Plate 7) indicate that erosion from the unvegetated surface of the former mine workings may locally increase the flux of Hg into the Tagburos River. However this influence is not very evident 2 km downstream where sediment Hg concentrations of 20 mgkg are closely analogous to those prevailing upstream of the PQMI mine or in mangrove mud samples from the mouth of the Tagburos River where Hg has declined to 8 mgkg (Figure 23). This value is substantially higher than recorded in offshore sediment, reflecting an ultrafine (claydominated) granulometry and high capacity for Hg adsorption, but it has to be taken into account that the nearshore samples (with a maximum of 0.4 mgkg, were not sieved). Qualitative field observations of the flow regime of the Tagburos river suggest that the total sediment discharge (and hence the total Hg flux) from this system into Honda Bay is unlikely to be substantial. In addition, the lower reaches of the river are characterised by extensive marshy areas and mangroves, which act as an effective filter for Hg-enriched suspended sediment, thus preventing significant contamination of sediments in Honda Bay.

1o()oo

l _ l l . . " l l . l * . " . . . l . l l l . l l l l . l . . l . . l

I

lll~*l-."..l---

+Bottom sedimnt +w

f-

100

cn

I

1 -

0

1

2 Distance (km)

3

4

Figure 23. Downstream dispersion of Hg from tailings piles (0 km) to the sea (4 km) in Tagburos River, Palawan.

24

7.3.3 Marine sediment: Honda Bay

A study of the geochemistry of sediments in close proximity to Sitio Honda Bay (Benoit et al., 1994)provided evidence that the Hg concentration of the jetty substrate, constructed of mine waste, may be higher than previously envisaged (up to 560 mgkg). A substantial Hg flux into the marine environment was indicated by the prevalence of distinct gradients in surficial sediments southward and eastward from the jetty. An inverse relationship between distance from source and the percentage of Hg held as cinnabar was also demonstrated (using a series of leach-tests), and interpreted as consistent with the transformation of detrital sulphides to more reactive (methyl) Hg phases during transport. A preliminary survey of Hg in marine sediment in 1995 by the MGB involving sediment coring at 12 sites (Figure 2) revealed Hg concentrations in the global background range (c40 pgkg) in the surficial 10 cm of sediments at most offshore stations, with no significant downcore variation. Coastal sediments routinely yielded similar values, although enrichment by approximately an order of magnitude (approximately 400 pgkg) was recorded at the mouth of the Tagburos River. Marine sediment cores were used to assess spatial and temporal trends of Hg deposition in Honda Bay. The average Hg concentration in surficial sediment at offshore sampling stations throughout the study area was found to be approximately 40 pgkg, and is thus within the global background range. Detailed re-sampling of selected cores (sampling stations 2, 7,9 and 25, Figure 2) showed evidence of Hg enrichment at or near the sediment-water interface, most conspicuously in core 2 in which a two-fold increase of concentration to a maximum of 95 pgkg occurs within the uppermost 5 cm (Figure 24). While such concentration adjustments occur in sediments deposited during the last 10 - 50 years, it is unlikely that mining or any anthropogenic activity is the causal factor. The role of mining can be specifically discounted, given the location of this coring station in the un-mined basin of Puerto Princesa Bay. In all analysed cores, total organic content (as defined by LOI) was found to increase significantly in the uppermost 2-4cm. Preferential incorporation of Hg into this organic sediment fraction thus provides the most plausible explanation.

0

5

A

0 0

10

4

15

fi

20

0./@

q-O 9

9 0 T

PI

CORE 7

O \

/.

0

I

25

0

30 10 20 30 40 50 6070 80 90 10 20 3040 50 60 70 80 90 10 20 30 40 50 60 70 80 9(1 10 20 30 40 50 6070 80 90

Concentration (pg/g) Figure 24. Downcore Hg profiles through the upper 26 cm of sediment at selected Honda Bay coring stations.

-

7.3.4 Sitio Honda Bay geochemical and mineralogicalresults Earlier studies (Benoit et al., 1994)indicated Hg concentrations of 560 mgkg in the waste used to construct the Sitio Honda site and led to the proposal that the 200 residents should be re-settled as the structure was considered too contaminated for human habitation. Geochemical and mineralogical analyses of mine waste from the Sitio Honda Bay jetty structure (Plate 8) were undertaken to establish the total concentration and bioavailability of Hg. Profiles through the waste characteristically display a depthward reduction of Hg concentration, from surficial values of up to 340 mgkg to basal concentrations of c40 mgkg. It is notable that garden topsoil at one site holds a high Hg concentration (179 mgkg) relative to most analysed samples of Sitio Honda Bay minewaste. This soil was imported specifically for use as an uncontaminated matrix for growing garden vegetables. It is possible that organic matter in this soil has assimilated Hg from the underlying waste, with potentially important implications for it's long term utility for domestic vegetable production.

25

Gravimetric partitioning of Hg in waste samples indicated that only 2-35% of Hg is held in the heavy mineral fraction. This implies that cinnabar within the waste must be predominantly fine-grained and prone to rapid weathering. Any cinnabar originally present in waste forming these horizons must, therefore, have been comprehensively degraded and the Hg repartitioned into clays, and this is confirmed by the high Hg (up to 550 mgkg) in the fine fraction of the tailings. Microprobe analysis of the e20 pm fraction indicated that the Hg in the clay-dominated assemblage is primarily held as an inorganic impurity in (or sorbed to) hydrous Fe oxide phases such as goethite and ferrihydrite. The solid-phase speciation of Hg in the e2mm fraction of the waste is dominated by inorganic non-sulphide Hg-phases (generally constituting >90% of the total Hg mass balance). Seaward adjustments to the Hg totaVHgS ratio as reported by Benoit et al. (1994) may be caused by gravimetric sorting of cinnabar and secondary Hg carriers in the fine-silt and clay size range following the erosion of waste from the jetty by wave action. These phases are of typically low bioavailability. Human Hg exposure through particulate inhalation or hand-mouth ingestion is therefore considered unlikely to be significant.

8. WATER AND SEDIMENT QUALITY

As a precursor to direct assessment of human and ecotoxicological impact it is conventional to evaluate potential toxicological hazards to aquatic biota and humans of contaminants in water and sediment by comparison with sediment and water quality thresholds for human health, drinking water, aquatic biota and shellfish. 8. I Water quality 8.1.1 Mercury

Of the artisanal gold mining and the mercury mining case studies, the Diwalwal area is characterised by extraordinarily high concentrations of Hg in water (approximately 2000 pgA) which decreases to about 5 pg/l some 10 km downstream, which is still a very high concentration. Whereas the river water in the immediate vicinity of Diwalwal is extremely turbid and not abstracted for drinking, 10 km downstream the water is no longer turbid and is used by the local inhabitants for routine domestic purposes. With a Hg load of 5 pgA in solution, this is clearly a significant health hazard. Such high values clearly exceed both the WHO Drinking Water Guideline value and the USEPA Water Quality Criteria for the Protection of Aquatic Life (Table 8). In the other areas studied, Hg in river water is below these water quality criteria and the Human Health Criteria (carcinogenic risk from consumption of aquatic organisms) so have no toxicological significance. 8. I .2 Arsenic

Whereas the WHO Drinking Water Guideline (10 pgA) and the EC Directive Maximum Admissible Concentration in Drinking Water (50 pg/l) are clearly exceeded in several instances in the Ponce Enriquez area, the maximum water safety level for aquatic life (400 pg/l; Fergusson, 1990) is exceeded in only one sample. No samples exceed the value of 3000 pg/l As adopted by the UK-DOE (NRA, 1994) to comply with the EEC Shellfish Waters Directive (79/923/EEC) but the more stringent USEPA Water Quality Criteria for fresh water (190 pg/l) is exceeded at five sites (Appleton et al., 1996). Marine invertebrates and fish usually contain high concentrations of arsenic in the range of about 1 to 100 mgkg dry weight (Neff, 1997) with wet weight concentrations for prawns typically being in the range of 1 to 18 mgkg (mean 5 mgkg; Sadiq et al., 1995). The arsenic is largely in the form of organoarsenic compounds of which arsenobetaine is the major component. Fortunately, although organoarsenic compounds are bioaccumulated by human consumers of seafood, the arsenic is mostly excreted. Furthermore, the arsenic is mainly arsenobetaine which is neither toxic nor carcinogenic to mammals (Neff, 1997). The USEPA human health criterion (fish consumption) value of 0.0175 pgA is difficult to understand as the average concentration of arsenic in the Ocean is 1.7 pg/l (Neff, 1997). There is reported to be a potential human health risk (based on 10-6risk for carcinogens) associated with the consumption of fish or shrimps living in waters with As concentrations >0.14 pg/1 (Table 8). This level is clearly exceeded in the lower reaches of the Rio Siete immediately above the shrimp farms but its toxicological significance is not clear. A detailed study of arsenic in the shrimps and fish both in the lower reaches of the Rio Siete and adjacent shrimp ponds may help to determine the toxicological significance of the high concentrations of arsenic in the river water as well as in the SPM and bottom sediments (see section 8.2.2, below).

26

8. I .3 Copper

In the Ponce Enriquez area, copper in nine water samples exceeds the EEC Directive Guide Level for Drinking water (100 yg/l) and 5 samples exceed the WHO Drinking Water Guideline (2000 pg/l). A sample taken about 3.5 km upstream of the shrimp farms on the Rio Siete contained 807 pgA which significantly exceeds the UKDOE standard (10 yg/l) adopted to comply with the EC Shellfish Waters Directive (79/923/EEC; NRA, 1994) and also the USEPA Freshwater Criteria for the Protection of Aquatic Life (11 pg/l; Table 8). If the temporal variations related to changes in contaminant fluxes and hydraulic regime are common, then higher concentrations may sometimes occur near the shrimp farms. More detailed monitoring is required to verify the level, frequency and duration of high contaminant fluxes in stream water. All Nambija, Diwalwal and Gango, water samples analysed during this study yielded Cu and As values which are compliant with WHO standards for potable water (Table 8). Table 8. Comparison of potentially harmful element concentrationsin filtered water samples from 2 km (MA-28) and 4 km (MA-30) above the Rio Siete shrimp farms (Ponce Enriquez) with maximum concentrations in the Ponce Enriquez, Nambija, Diwalwal and Gango artisanal mining areas and Water Quality Criteria (concentrationsin &I) Freshwater’

As Cd Cu Pb Hg Ni Zn

Max. Conc.2 360 3.7 17 65 2.1 1400 110

Saltwater’

Cont. Max. ~ o n c . Conc.’ ~ 190 69 1 42 11 2.4 2.5 210 0.0126 1.8 160 74 100 90

Cont. ~onc.~ 36 3000 9.3 330 2.4 10 8.1 100 0.025 1 8.2 100 81 10

Ponce Enriquez, Ecuador MA28 77 <4 c u 17 Pb na Hg ~ 0 . 0 2 Ni <10 Zn 49

As Cd

MA30 109 <4 809 na 0.03 <10 46

Shellfish Human Directive health’,’ (79/923/E EC)4

Mm. 470 9 7277 na 0.9 165 82 1

Nambija, Ecuador Ma. na na 3.5 na 0.1 na na

0.14

0.15 4600

WHO Drinking Water

EEC Drinking Water MAC

10 5 2000 50 1

50 5 100 50 1

50 5000

Diwalwal, Mindanao Max. na na 344 nd 1539 na 81

Gango, Mindanao Max. nd na nd na 0.1 nd na

Footnotes: na = not available; nd = not detected I EPA Section 304 (a) Criteria for the Protection of Aquatic Life from Priority Toxic Pollutants (US Cleanwater Act, February 5, 1993; Part 131-Water Quality Standards, Sec. 131.36 proposed amendment April 1995). The criteria refer to the inorganic form only. Freshwater aquatic life criteria vary with total hardness and pollutant’s water effect ratio (WER). Values quoted here correspond to total hardness of 100 mgA and a WER of 1.O 2 Criteria maximum concentration (CMC) = the highest concentration of a pollutant to which aquatic life can be exposed for a short period of time (I-hour average) without deleterious effects. Criteria continuous concentration (CCC) = the highest concentration of a pollutant to which aquatic life can be exposed for an extended period of time (4 days) without deleterious effects UK-DOE “I” (imperative) values recommended for compliance with EC Shellfish Waters Directive (791923EEC) (NRA, 1994) Human health ( 10-6risk for carcinogens) for consumption of aquatic organisms living in waters. For 10” risk, move decimal point one place to the right. If the CCC exceeds 0.012 pgfl more than once in a 3 year period in the ambient water, the edible portion of aquatic species of concern must be analysed to determine whether the concentration of methyl mercury exceeds the FDA action level of 1.O mgkg.

’

27

8.2 Sediment quality’ 8.2.1 Mercury

In the Diwalwal area maximum values of 34 mgkg and 40 mgkg have been recorded in stream bottom sediment and suspended sediment and similar maximum values have been recorded in the other case study areas (Table 9). In the Ponce Enriquez area, the relatively high Hg concentration (3 mgkg) in stream bottom sediment 15 km below Bella Rica implies that the average level of contaminant transfer was higher historically, probably during the previous period of high rainfall. The bioavailability of Hg in stream bottom sediments in the lower reaches of the Rio Siete, immediately above the shrimp farms, merits further investigation as Hg exceeds the Toxic Effect Threshold for the Protection of Aquatic Life (Table 9). The Toxic Effects Threshold is exceeded over a lengthy section of the Mamunga River and also along short sections of river draining the Nambija and Gango areas. In the nearshore zone of the Sitio Honda Bay site, Hg in the top layer of marine sediment is unlikely to be a significant toxicological hazard (Williams et al., 1996). High concentrations of Hg in bottom sediments is a long term source of contamination which may be dispersed by high flow rates and also progressively release Hg through methylation processes. Such contamination has been referred to as a ‘Chemical Time Bomb’ (Lacerda and Salomons, 1998; see also Section 2.4 above). Table 9. Comparison of potentially harmful element concentrations in suspended particulate matter (SPM) and stream bottom sediments (BS) from above the Rio Siete shrimp farms (MA-28) at Ponce Em’quez with maximum concentrations in the Ponce Em’quez, Nambija, Diwalwal and Gango areas and Sediment Quality Criteria for Protection of Aquatic Life (concentrations in mgkg). Sediment Quality Criteria

Ponce Enriquez

Nambija

BS

SPM

Diwalwal

Gango

BS

BS

BS

No effects threshold

Minimal effects threshold

Toxic effects threshold

MA28

Max.

MA28

Max.

Max.

Max.

Max.

As Cd cu

3 0.2 28

17 3 86

Hg

0.05

Pb

23

Zn

100

7 0.9 28 0.2 53 150

957 2 387 0.1 123 108

22626 18 6437 9.61 1061 743

7258 12 2912 3 175 347

46049 24 9134 13 666 924

34 3 409 34 42 23 1

11 <3 67 34 65 114

333 c3 188 0.13 1078 1211

1

170 540

Note: SPM = suspended particulate matter; BS = bottom sediment; Sediment Quality Criteria for Protection of Aquatic Life (Environment Canada, 1992 quoted in MacDonald, 1994).

8.2.2 Arsenic

All arsenic concentrations in stream bottom sediments in the Rio Siete exceed the Sediment Quality Criteria (Toxic Effects Threshold) for the Protection of Aquatic Life by a factor of up to 2700 (Table 9). It should be realised that very high As concentrations (possibly up to loo0 mgkg) would occur naturally in this area of arsenical mineralisation. The other areas are characterised by much lower concentrations and the threshold is exceeded by a significant margin only at Gango. 8.2.3 Copper

Copper in the <150pm fraction of stream bottom declines gradually from 9,134 mgkg in the headwaters of the Rio Siete to 2912 mgkg at a site immediately above the shrimp farms (Table 9). In the corresponding suspended sediment samples, copper concentrations decrease from 6437 mgkg to 387 mgkg. All copper concentrations in stream bottom sediments in the Rio Siete exceed Sediment Quality Criteria (Toxic Effects 1

It should also be taken into account that contaminant concentrations in all case study areas were determined in the fine (c150pm) fraction of bottom sediment whereas Sediment Quality Criteria are concentrations in unsieved sediment. Most of the contaminants will tend to be concentrated in the fine fraction as a result of adsorption onto fine particles, clay minerals and organic material.

28

Threshold) for the Protection of Aquatic Life by a factor of up to 106 (Table 9). As with arsenic, high ‘natural’ levels would have characterised the area prior to the initiation of mining activities. 8.2.4 Cadmium

Cadmium in stream sediment at the site immediately above the shrimp farms on the Rio Siete exceed Sediment Quality Criteria (Toxic Effects Threshold) for the Protection of Aquatic Life by a factor of 4 reaching a maximum of 16 in the headwaters of the Rio Siete (Appleton et al., 1996). Pb and Zn in stream sediment from the lower section of the Rio Siete are either lower, or do not significantly exceed, the Sediment Quality Criteria for the Protection of Aquatic Life (Table 9).

8.2.5 Discussion and recommendations Although both As and Cu in stream bottom sediment from the Rio Siete immediately above the shrimp farms exceed Sediment Quality Criteria (Toxic Effects Threshold) for the Protection of Aquatic Life by factors of about 400 and 30, respectively, this does not necessarily imply that sediment in the shrimp farm ponds is contaminated or that these contaminants are either affecting the development of shrimps or being absorbed by the shrimps. In the Ponce Enriquez area, it is recommended that that the composition of sediment and water in the shrimp ponds, and the shrimps themselves, should be carefully assessed. Considerable care should be taken to ensure that contaminated sediment does not enter the shrimp ponds as the high concentrations of As and Cu, as well as Hg, may negatively affect both the productivity and quality of the shrimps. Base metal sulphide minerals are not a major component of gold mineralisation in the Diwalwal, Gango and Nambija areas so only a few samples exceed the Toxic Effects Thresholds by a relatively small margin (Table 9). In any aquatic system subject to inorganic Hg, As and Cu contamination, the attendant toxicological risk (with respect to magnitude and temporal duration) cannot readily be determined from data depicting total sedimentary load. Additional geochemical and ecotoxicological studies are required to establish factors such as rates of sedimentary Hg methylation and subsequent bioassimilation. With regard to human exposure, detailed epidemiological information, and body burden data would be critical precursors to any meaningful assessment. Such studies have been carried out in the Palawan and Tagum case study areas in the Philippines in conjunction with toxicologists. The results of these studies are described in the following section. 9 HUMAN AND ECOTOXICOLOGICAL IMPACT ASSESSMENT

9. I Human exposure to mercury

Mercury adversely effects physiological and neurological processes in humans. The toxicity of Hg generally increases through the sequence: - phenyl Hg salts > Hg2+salts > free Hg vapour > alkyvmethyl species. Methyl mercury is considered to be 100-loo0 times more toxic than inorganic Hg so the rate and extent of methylation is clearly important. Methyl mercury, which is considerably more toxic than dimethyl mercury, causes marked degeneration of the central nervous system. A major environmental concern in all gold mining districts where mercury amalgamation is used is the potential exposure of workers to mercury because of its use in the recovery of gold. Mercury vapour lost during the amalgamation process presents a health risk to workers and also contaminates soils and vegetation. Humans may also be exposed through the ingestion of Hg contaminated food. Whereas Hg is generally low in grain (5 - 20 ng/g) and vegetables (1-40mg/g; Fergusson, 1990), methyl mercury tends to accumulate in aquatic biota. Hg in fish may be more than 20,000 times the concentration of Hg in lake waters (Craig, 1982). Consumption of Hg contaminated shellfish and fish is especially hazardous because more than 80% of the total Hg occurs as the highly toxic methyl species. Additional information on the geochemistry and toxicology of mercury is presented by Fergusson (1990). In Palawan, the extent of mercury bioassimilation and attendant toxicological stress in marine biota (fish and mussels) was assessed. Bivalves are excellent indicators of heavy metal contamination, due to both their capacity to accumulate metals from their environment (mainly in particulate form), and their widespread distribution throughout the world. In many regions, bivalves also form an important source of human nutrition, and may therefore constitute a significant pathway for human metal exposure. On account of the biomagnification of metals which typically occurs during assimilation by shellfish, any small increase in ambient metal concentration resulting from pollution will typically be reflected by a distinct increase in mussel tissue concentrations. 29

The impact of mercury contamination on the local inhabitants in the Honda Bay area of Palawan and the Tagum area of Mindanao was evaluated through determination of mercury in hair samples. 9.2 Palawan, Philippines 9.2. I Background

In November 1994, the Philippines Environmental Monitoring Bureau (EMB) undertook preliminary analyses of fish and shellfish from Honda Bay and concluded that several tissue concentrations fell within or above the range recorded at Minamata, Japan. In March 1995, a Department of Health (DOH) survey of blood Hgconcentrations in 42 subjects from the three barangays located in closest proximity to the PQMI operation highlighted a 25%failure to meet the global threshold for 'normal' blood Hg levels of 20 ng/ml. Media reports of the results of these preliminary biological and medical studies were, inevitably, interlinked with the geochemical data of Benoit et a1 (1994), showing a major elevation of Hg (to 560 mgkg) on and around the Sitio Honda Bay jetty which is constructed of about a million tons of Hg contaminated roaster waste. Consequently, the situation portrayed by the media in August-September 1995 was one of unacceptable human Hg exposure as a direct result of the contamination of Honda Bay and it's fishkhellfish reserves by detrital Hg fluxes from Sitio Honda Bay, with subsequent methylation within the bay-floor sediment column. This hypothesis formed the basis of a request by Palawan health officials for the granting of Disaster Area status to central Palawan, and the immediate provision of central government funds to support one or more of the following actions: - (i) detoxification of approximately 25% of the local population (based on the 25% failure of subjects in the preliminary DOH survey to meet the 'normal' blood threshold of 20 ng/ml), (ii) provision of a laboratory facility at Puerto Princesa Provincial Hospital for the analysis of Hg in blood, (iii) removal of the Sitio Honda Bay jetty and safe disposal of the contaminated tailings, (iv) resettlement of the approximately 200 population of Sitio Honda Bay, (v) dredging of sediments from contaminated sectors of Honda Bay. By late 1995, the widespread publicity surrounding the Palawan mercury scare had imposed a significant economic impact on the island, notably its fishing industry as a consequence of concern over product quality amongst wholesale purchasers in Manila. Prior to the BGS/ITE survey, however, little quantitative data existed to show the genuine extent of ecotoxicological risk to populations living close to the PQMI mine, on the Sitio Honda Bay jetty, or those consuming fish from Honda Bay. Particular uncertainty may have arisen through the misrepresentation of the geochemical data of Benoit et a1 (1994) in the media which, in failing to emphasise the extremely localised nature of the study, inferred that anomalous Hg concentrations were characteristic of the entire Honda Bay floor. In reality, the gradients reported by Benoit et al (1994) show concentrations declining rapidly from 560 mgkg on the jetty, to 38 mgkg at 25 m distance, and 2.3-18.8mgkg at a distance of 200 m. Data for control sites 7-10 km offshore indicate the prevalence of conditions within the global background range (0.03 - 0.2 mgkg). Uncertainties in the preliminary biological and ecotoxicological datasets collated by the EMB and DOH primarily reflect the small sample populations involved (Williams et al., 1996a). Of the 12 people who yielded blood Hg values exceeding the 20 ng/ml threshold, 50% were former miners or roasting plant operators. Data for these individuals do not therefore infer any wider exposure of the population via the food chain. Within this sub-group, the highest recorded value (25 ng/ml) can be regarded as unexceptional for individuals subject to long-term occupational exposure. Although clinical symptoms of Hg poisoning have potentially been observed in a limited number of Honda Bay miners, it is unlikely that these could have been caused by blood Hg levels of this magnitude. Reference data from a comprehensive UNEPNHO-endorsed Monitoring and Assessment Research Centre (MARC) study indicate that the lifetime exposure threshold required to induce pronounced symptoms of clinical Hg poisoning is 80 ng/ml, with effects typically absent at levels of <200 ng/ml. Clinical damage in miners could, however, have been caused by short-term acute exposure (no longer reflected in blood) up to several decades previously. 9.2.2 Fish and shell-fish

Samphg and analytis Fish muscle samples from ten individuals of each species were dried at 50°C within 6 hr of collection (to facilitate transport to the UK without substantial decomposition) and the weight loss assessed in each case. All samples were subsequently digested in HN03, diluted to an appropriate volume and analysed for Hg by CVAFS using the instrumentation described in Section 7.1 above. Samples of the species Perna viridis (green

30

mussel) were collected from two locations. The first was situated in shallow water approximately 10 m off the southern margin of the Sitio Honda Bay jetty (Figure 2). The second was located close to the western shore of Canon island, approximately 6 km off the Tagburos - Sitio Honda Bay coast. All samples were returned to the haematology laboratory of Puerto Princesa Hospital for preparation and biomarker assessment within 8 hours of collection. Samples were purged in clean water prior to sub-sampling. The soft tissues of mussels were airdried at 50°C prior to transport to the UK for final weight determination and analysis. The dry tissues were partially digested in cold HN03 at ITE (Monks Wood) and forwarded to BGS for further reflux digestion and total Hg analysis by CVAFS. Neutral-red biomarker assessment. An ecotoxicological field test based on the neutral-red retention (NRR) capacity of invertebrate cells has been

successfully utilised to assess metal-induced stress in a number of marine and terrestrial settings (e.g. Weeks and Williams, 1995). In the present study, haemolymph (0.02-0.05 ml) was extracted from mussel specimens and mixed with an equal volume of temperature-adjusted physiological ringer using a 1 ml hypodermic syringe. Each haemolymph suspension was transferred then to a siliconized Eppendorf (0.5 ml) for subsequent NRR analysis. A neutral-red stock solution, comprising 20 mg of neutral-red dye dissolved in 1 ml of dimethyl sulphoxide, was freshly prepared. Subsequently, 10 pl of the stock solution was diluted with 2.5 ml of physiological ringer, giving a working concentration of 80 pg/ml. To avoid crystallisation of the neutral-red dye, the working solution was renewed every hour during the measurement process. Haemolymph samples of 20 pl were placed on a microscope slide, and the cells allowed to adhere to the slide surface for 3 minutes before the application of the neutral-red working solution (20 pl) and a cover slip. Each slide was continuously scanned at random (by rapid random repositioning) under a microscope (at constant magnification) to observe any temporal adjustments to the condition of the cells. Each visualisation was divided into 3 minute intervals, from which the numbers of cells with fully stained and unstained cytosol were determined. Observation was stopped at the interval when the ratio of stainedunstained cytosol was greater than 50% of the total number of cells counted. The midpoint of the interval was noted as the NRR time. Previous work

It had been proposed that contamination by mining activities of the Honda Bay fish stocks may account for the prevalence of elevated Hg levels in some individuals who had not worked at the mine or resided near the minewaste accumulation at Sitio Honda Bay. Previous studies had compared dry-weight Hg concentrations ranging up to 1 1 mgkg with wet-weight values for fish and shell-fish at Minamata (0.4-30 mgkg and 1.3-14 mgkg respectively) and concluded that there was evidence of serious contamination of the Honda Bay fish stocks. However, recalculation of these values on a wet-weight basis reduces the concentrations to a more moderate level (<0.0005 to 2.0 mgkg). Comparison of the wet weight Hg burdens in Honda Bay fish with threshold values for marketed fish in the USA (0.5 mgkg, US-EPA), and with global average data for species of varying ecologies (e.g. IRPTC, 1980; Table 1 1 indicates that the mean and, more significantly, the median levels prevailing in the Honda Bay samples are not exceptional. Separation and subsequent statistical analysis of data for individual specimens caught in the inner and outer sectors of Honda Bay yielded no significant differentiation. Realistically, all sampled species are extremely motile; hence significant differences across a continuous area of open water could not be expected. A subsequent study of fish Hg burdens of 9 species with widely varying ecologies in Honda Bay conducted by the University of the Philippines showed only two species to carry mean burdens in excess of the US-EPA marketing threshold of 0.5 mgkg Hg. Highest concentrations (mean 0.924 mg/kg) were found to prevail in a predatory species (talakitok), for which such values could be considered normal (Piotrowski and Inskip, 1981). Resutes of present study

Fish. The present study showed that mercury concentrations in six species of fish from Honda Bay fall within the ranges typically encountered for analogous species world-wide (Tables 10 and 11). The fish types collected are considered to be representative of the full range of detrital, algal and carnivorous food chain pathways. They are known only by local name: malakapas, salmonete, tuko, taba-taba, sap-sap and mackerel. Hg values for all samples apart from one of Salmonete lie within the US-EPA marketing threshold of 0.5 mgkg (Figure 25).

31

Table 10. Mercury burdens in Honda Bay fish (mgkg). Species

Range Wet

Mean Wet wt. Hg rngilcg

Salrnonete Malakapas Tuko Taba Taba Sap Sap Mackerel

Mean Dry wt. Hg

wt. Hg rngkg 0.07-1.07 0.05-0.42 0.03-0.21 0.05-0.32 0.19-0.41 0.18-0.64

0.19 0.15 0.09 0.09 0.3 1 0.33

0.94 0.74 0.43 0.45 1.55

1.70

Table 11. Approximate average wet-weight mercury levels (mgkg) in muscle tissues of marine fish in major oceans worldwide. Species Tuna Mackerel Sardine Other non-predatory

Atlantic

Pacific

Indian

Mediterranean

0.30 - 0.80 0.07 - 0.20 0.03 - 0.06 0.08 - 0.27

0.30 0.16 - 0.25 0.03 0.07 - 0.09

0.06 - 0.40 0.005 0.006 0.02 - 0.16

1.20 0.24 0.15 0.10 0.30

-

0.2

0.33

-

0.09 0.19

a *

0.4

0.3

-

Honda Bay

* *

t.

a

' a

1:

0.1

Mackerel

Malakapa Salmonete Sap Sap Taba Taba

Tuko

Figure 25. Hg in fish from Honda Bay, Palawan compared with the USEPA marketing threshold value (dashed line = 0.5 mgkg).

Mussels. The average wet weight value (2.13 mgkg) and range (0.86 - 4.37 mgkg) established for samples of Perna Viridis (green mussel) collected from within 10-20 m of the Sitio Honda Bay jetty is almost an order of magnitude greater than that for Canon Island (mean 0.34 mgkg; Figure 26), located 7 km offshore. The Sitio Honda Bay mean value falls within the lower quartile of the Minamata shellfish range (1.3 - 14.0 mg/kg), indicating that mussels and analogous filter feeders from the immediate vicinity of the jetty may be unfit for human consumption. The contamination of biota in this locality is entirely accordant with the high Hg concentrations observed in sediment from the same vicinity (see Benoit et al., 1994; or MGB survey data, September, 1995), and lateral concentration gradients for the two media are probably analogous. Accordingly, it is plausible that bivalve Hg tissue burdens decline to background within a few hundred metres of the Sitio Honda Bay source. The maximum Hg concentration recorded in Perna viridis from Canon Island is within the international background range for shellfish (< 3 mgkg dry weight), signifying that bivalves from the wider Honda Bay may be suitable for human consumption.

32

1

E

m 1

0

0

H asel

't

0

l

i

r

1

Figure 26. Hg in green mussel (Perna Viridis) from close to the Sitio Honda Bay jetty (Honda) and from Canon Island (Canon).

Neutral-red biomarker assessment results A statistically significant difference (P
The extent to which the populations of Santa Lourdes and Sitio Honda Bay are exposed to Hg through residence on, or near, a mine waste substrate can be inferred from bioavailability data derived from mineralogical studies of Hg speciation in the waste material (section 7.3.4 above). Such studies must be supplemented by direct monitoring of the potentially affected populations (alongside one or more appropriate control groups) to produce a comprehensive risk assessment. Whereas some toxicologists now consider that hair analysis is unsuitable for virtually any exposure studies, mainly due to the problems of effectively removing surface contaminants, hair analysis is a pragmatic approach which requires little expertise. Hair mercury is accepted by some international authorities to be a suitable indicator for the assessment of contamination in human populations exposed to methyl-mercury (e.g. WHO, 1981; Lacerda and Salomons, 1998). The method holds advantages over blood Hg analysis as it is non-invasive, and results are not prone to short-term dietary influences. Hair Hg burdens are a direct function of average blood concentration in a ratio of approximately 250:l (see below also) and can thus be used to estimate blood mercury concentrations. Hair sampling and analysis

Hair samples were collected from the rear of the scalp (Plate 9) from representative sub-groups totalling 35% of the Sitio Honda Bay population (a community living on a mine-waste substrate and smaller sample from Tagburos (a coastal fishing community 1 km south of Sitio Honda Bay) and Santa Lourdes (a community living at the margin of the PQMI site; see Figure 2 for location of these sites). A small control population from Manila was also sampled for comparative purposes. For data analysis purposes, the samples were classified into five groups: (i) Sitio Honda Bay residents, (ii) Tagburos residents, (iii) Santa Lourdes residents living adjacent to the PQMI site, (iv) other residents of Santa Lourdes (mainly living west of the PQMI site), (v) exmineworkers from PQMI. Approximately 2 g of hair was prepared for analysis by washing repeatedly in distilled water to remove dust and other surficial contaminants. They were then air-dried; cut into 1 cm lengths using nylon scissors, weighed 33

and transferred to 50 ml graduated test tubes. Ten mg of V205and 5 ml of HN03 was added to each tube, and the samples were left under air-reflux overnight. The following morning all samples were heated to 140°C for 5 minutes, cooled and 2 ml of H2S04added. After a further heating period of 15 minutes, all solid material was digested and the solutions were cooled and diluted to an appropriate volume. An aliquot of each solution was used to determine total Hg content by CVAFS. Analysis of certified reference hair samples (0.36 and 12.3 mgkg Hg) indicated an average recovery of 92% (Williams et al., 1996a) Hair Resuks

The results indicate that all Palawan residents are subject to high Hg exposure, relative to a control population from Manila (Figure 27). Within the Palawan groups, only the Santa Lourdes group is statistically different, possibly because this inland group is likely to consume less fish than the Sitio Honda Bay and Tagburos groups and is also less exposed to Hg than the Santa Lourdes mine group. Fishermen of Tagburos and Sitio Honda Bay (Figure 28) yielded a mean hair Hg concentration of 6.49 mgkg, compared to a mean of 2.73 g mgkg for nonfishermen (of >16 yr. of age in the same localities). This trend has been observed frequently in coastal populations elsewhere (e.g. Italy, Paccagnella et al., 1973) and is typically equated with fish consumption. The average hair Hg concentration recorded in the ex-miner group (3.3 mgkg) is slightly lower than recorded for the populations of Tagburos and Sitio Honda Bay as a whole, although the median values are almost identical (Figure 27). All individuals within this group were aged >65 and two have been diagnosed as displaying Hg poisoning symptoms. These individuals, however, yielded the lowest hair Hg burdens within the group (<2 mgkg). Clinical damage in response to exposure at such levels is, on the basis of worldwide epidemiological data, extremely unlikely. It is therefore probable that clinical damage was induced by a historical exposure of a much greater magnitude (i.e. during employment at the PQMI mine). The emergence of Hg poisoning symptoms in subjects with low hair or blood Hg burdens is common under such circumstances, reflecting the long post-exposure latency period (approximately 10 years) which characterises several Hgresponsive human conditions. Hg in adult males (median 4.52 mgkg) is statistically higher than in adult females (median 2.67 mgkg) (Figure 29), although these differences are unrelated to body weight and most likely reflect the higher occupational exposure of males in the fishing sector (Williams et al., 1996a). An evaluation of the Hg status of juvenile populations from Sitio Honda Bay and Tagburos showed that age and body mass had little effect on Hg burdens. However, male children have slightly higher hair Hg than female children (Figure 29) do. Comparative data on hair mercury concentrations in populations from artisanal gold mining sites in Brazil and Colombia are presented in Table 12. The high Hg concentrations in hair of riverine populations from the Tucurui reservoir and the Rio Negro reflect the high daily consumption (0.5 kg/day) of fish containing >2 mg Hg g-’, which is approximately double that believed to be potentially harmful to human health (Lacerda and Salomons, 1998). Table 12. Mercury concentrations in hair samples from prospectors and riverine populations, for which fish is a major dietary component, of gold mining and non-gold mining sites in Brazil and Colombia (adapted from Lacerda and Salomons, 1998) Median Hg (pg g-9

Population Gold mining areas Prospectors, Madeira R., Rondonia, Brazil Riverine population, Madeira R., Brazil Riverine population, Tapajos R., Brazil Indian groups living close to Amazon mining sites Nariiio Province, Colombia Non-gold minim areas Riverine populations, Negro R., Brazil Riverine population, TUCUN~ reservoir, Brazil na. = not available

34

Range Hg (pg 8’)

4.6 7.9 na. na. na.

0.2 - 24.1 0.5 - 71.3 0.74 - 17.7 1.42 - 8.1 ~ 0 . 0 1- 17.1

75.5 47.0

5.76 - 171.4 4.0 - 241 .O

12,

I

8 -

I

E

-B

7 -

P6L .-

a

5-

I

4-

321

I CON

EX

SLM

SL

TAG

Figure 27. Hg in hair samples, Palawan. (CON : Manila control group; EX : ex-miners; SHB : Sitio Honda Bay; SL : Santa Lourdes; SLM : Santa Lourdes Hg mine; TAG : Tagburos).

Children

Omen

Houwkenpws

Rdennen

SloreK-

Figure 28. Hg in hair samples from fishermen compared with non-fishermen >16 years of age (Housekeepers, Others, Storekeepers) and children (<16 years of age) living in Tagburos and Sitio Honda Bay, Palawan.

18 16

-

14

-

ZJ12

-

10

-

2 L

I

S 6 4

2

CF

CM

F

M

Figure 29. Hg in hair from female (CF) and male (CM) children (e16 years age) and adult females (F) and males (M), Palawan.

35

Hair-blood relationships

The relationship between hair and blood Hg concentrations, and hence the reliability with which data intercomparisons for the two media can be attempted, has been examined extensively through statistical analysis of global epidemiological datasets (Piotrowski and Inskip, 1981). A reproducible empirical constant for hair is 250 x the simultaneous blood Hg concentration (e.g. Kershaw et al., 1980). Mean blood Hg values inferred from hair data for the five sample groups ranged from 8.8 - 17.6 ng/ml, with a maximum individual value of 74.1 ng/ml. Such blood Hg concentrations values are typical of populations consuming fish at a daily frequency. Dietary and residential H g exposure

Derivation of the relative importance of dietary, occupational and residential contributions to total Hg exposure is critical for the accurate assessment (and, if necessary, amelioration) of toxicological risk. From the data presented in this study, there is no evidence to suggest that residential factors significantly influence the Hg body burdens of any Palawan population groups. In particular, any exposure of the Sitio Honda Bay population as a consequence of their residence on a mine-waste substrate can be considered negligible, given the statistical comparability of hair Hg burdens for this group with those determined for the adjacent coastal barangay of Tagburos. The limited significance of mining activities or mine waste deposits on present-day human Hg exposure is equally evident with respect to the Santa Lourdes community living adjacent to the PQMI site, for which a mean hair Hg burden substantially lower than that of the Tagburos coastal community has been derived. These trends are, in turn, consistent with the mineralogical data for mine waste presented in Section 7.3.4, indicating that Hg in mine waste at the Sitio Honda Bay site is both predominantly inorganic, and held in species with extremely low bioavailability. Viewed in conjunction, the mineralogical and human body burden data currently available suggest that there is no immediate requirement to re-locate the Sitio Honda Bay population (or to otherwise modify the structure of the jetty) on grounds of Hg exposure limitation. In contrast to the limited influence of residential factors, there is strong evidence that diet is the major control on Hg exposure within the Palawan population. The predominance of fish as a source of methyl Hg in the human diet has been recognised for several decades (WHO, 1976), to the extent that up to 90% of spatial variations of Hg burdens world-wide are explicable by reference to fish consumption (Piotrowski and Inskip, 1981). This reflects the intense biomagnification of Hg in aquatic foodchains, often producing enrichment factors of >30,000 in top-carnivores (shark, barracuda etc.) relative to ambient water Hg concentrations. While the global average Hg concentration in hair is of the order of 2 mgkg (blood equivalent of 8 ng/ml), the ‘normal’ blood levels for high fish consuming populations (including Italy, southern France, indigenous Canadian and American Indians) can range from 20-80 ng/ml (e.g. Paccagnella and Prati, 1974; Riodolfi, 1977; Clarkson, 1975; Health and Welfare of Canada, 1979). The WHO has provided guidelines for estimating likely hair Hg burdens in populations with varying levels of (non-contaminated) fish consumption. These are:- consumption once monthly = 1.4 mgkg, once weekly = 2.5 mgkg, and once daily = 11-6 mgkg. Interviews with participants in the Palawan survey indicated that fish consumption is generally of a daily frequency. Of the 130 subjects included in the Palawan survey, 128 displayed hair concentrations falling below the anticipated daily consumption threshold of 11.6 mgkg. The inferred body burdens of virtually all subjects can thus be considered a predictable and direct function of diet. There is clearly some discrepancy between what the WHO accept as a realistic Hg dose associated with daily fish consumption and their 7 mgkg ‘acceptable’ hair threshold. Toxicological implications of the recorded Hg burdens

There is no international consensus regarding the practical ‘risk’ threshold for methyl Hg in humans. The precise blood concentration beyond which clinical damage occurs (often following a long latent period) is dependent on a range of factors including age, exposure duration and nutrient status (with respect to elements such as Se). It has been concluded on the basis of over 300 dose-response studies that an increased risk of mild non-specific neurological damage may result from life-long exposure beyond a blood Hg threshold of 80 ng/ml (WHO, 1981). Assuming shorter exposure, risk is generally apparent beyond a higher threshold of 200 ng/ml. Pregnant women constitute a special case, for whom a threshold of <100 ng/ml may be more appropriate. A risk threshold of 50 mgkg in hair (blood equivalent = 200 ng/ml) is routinely utilised. The adoption of a threshold lower than 80 ng/l might have significant cost-benefit implications as it would fall within the ‘normal’ blood concentration range for sizeable populations in Europe and North America in which no toxicological effects have been observed (e.g. Paccagnella and Prati, 1974; Riodolfi, 1977; Clarkson, 1975; Health and Welfare of Canada, 1979). 36

Blood concentrations in inhabitants of mining settlements in the SE Amazon region of Brazil are in the range 90-149 ng/ml methyl mercury which corresponds to 72-99% of the total Hg in the blood (Lacerda and Salomons, 1998). Lacerda and Salomons (1998) note that most toxicological studies indicate that concentrations are almost always 4 0 n g / d in unexposed individuals, with symptoms typically developing in the range 20-50 ng/ml. With sensitive equipment it is possible to detect ‘shake’ at exposures of 20 ng/ml but the 50 n g / d would be the level at which to become concerned (pers. comm. 1997, Guy’s and St Thomas’ toxicologists). This is equivalent to 12.5 mgkg in hair, which is in excess of the WHO ‘acceptable’ threshold level (7 mgkg). The data obtained in the present study do not fully substantiate the claims made in the Philippine media during August and September 1995 regarding the occurrence of a major mercury-poisoning episode in Palawan. Total mercury body burdens are not considered to reflect geological or mining influences as statistical analysis of data for five Palawan sub-groups failed to significantly discriminate between those living on a mine-waste substrate and other populations. However, it should be noted that these conclusions are based largely on a single field sampling programme and longer-term monitoring is recommended to gain a more comprehensive understanding of the sources, environmental behaviour and toxicity of Hg on the island of Palawan and elsewhere. Amelioration strategies

Detoxification procedures (typically involving the use of a sulphydryl receptor) are suited to the treatment of occupational rather than dietary exposure, and is unlikely to provide a long-term mechanism for reducing the impact of dietary Hg exposure in the coastal communities of Palawan. No record has been encountered of cases where clinical detoxification has not been proposed, or applied, for the treatment of individuals/populations exposed via fish consumption. A more common amelioration strategy under such circumstances involves progressive modification of the diet (notably methods of fish preparation) through community education. Programmes of this type have been successfully deployed among native American Indian populations in Canada and the USA.

9.3 Tagum, Eastern Mindanao, Philippines 9.3. I Background Apokon, near Tagum, is the principal centre for processing gold ore from the major mining areas, including Diwalwal. Although C P is now the dominant gold extraction method, amalgamation continues to be used, especially through the direct addition of mercury to ball and rod mills. Tailings from the amalgamation process are then treated by cyanidation so cyanometallic complexes could have an important role in mercury mobilisation. Enhanced Hg levels (up to 2.30 mgkg wet weight) have been reported for fish flesh from the Agusan River and tributary systems draining Diwalwal and Mainit, and from the estuarine outflow of the Agusan at Butuan Bay (Torres, 1992). In Davao del Norte, air Hg concentrations of 42 - 1664 pg m-3were recorded in four different gold commercialisation shops, with 65% of samples exceeding the WHO (1976) industrial exposure limit of 50 pg m-3(Torres, 1992). However, only 6% of 230 potentially exposed workers sampled by Torres (1994) showed elevated blood levels, and all had urine Hg below 50 pg 1-’. An initial survey in 1996 by the Philippines Department of Health (DOH) of blood Hg levels in Apokon residents revealed that the maximum was 20 ng/ml (Williams, 1997). Clear clinical indication of Hg poisoning was reported and detoxification of residents with >10 ng/ml was proposed and instigated. Media coverage prompted public and political pressure for compensation of the Apokon residents and for the closure of the gold processing operations. The response of the DOH to the blood Hg data was unusual because values of up to 20 ng/ml are fairly common in fish-eating communities such as Apokon. It is possible, therefore that the Apokon residents Hg burden derived principally from fish and was not related to Hg used in the gold processing operations. Interviews with 50 residents from Apokon revealed that the majority consumed fish at least 7 times a week. The 10 ng/ml action level used by the DOH for detoxification could be considered clinically unsound and arguably unethical; clinical symptoms are virtually impossible to detect at 10 ng/ml and sophisticated instrumentation is required to detect nerve response suppression at 20 ng/ml. An increased risk of mild non-specific neurological damage may result from life-long exposure beyond a blood Hg threshold of 80 ng/l (Williams et al., 1996a) or above 200 ngA for shorter exposure. A short-term exposure threshold of 50 n g / d may be more realistic for the general population whereas 100 ngA may be more appropriate for pregnant women.

37

9.3.2 Hair sampling

In order to re-assess the toxicological impact, if any, of mercury from the gold processing plants on human Hg status in the Apokon area, a survey of Hg in the hair of 48 subjects was carried out by the BGS and the MGB in April 1997. Samples were taken from (1) eight Kimpro workers involved in the CIP processing of tailings previously treated with mercury; (2) two ballmill operators using mercury; (3) a blow-torch operator involved in burning amalgam; (4)17 people living within 50 m of the Kimpro, Godenez and Bulbscor processing plants; and (5) 19 residents from Bar Hall, Omandac, Edgehill and Dabaweno communities located 100-200 m from the main processing plants. 9.3.3 Hair results and toxicological implications

The hair Hg results are consistent with the DOH blood data insofar that they indicate blood Hg is generally less than 20 ng/ml (based on the WHO conversion factor of 1 mgkg in hair being equivalent to 4 n g / d in blood). Only one Kimpro worker (7 mgkg Hg in hair) and one school child (13 mgkg) exceed this level. As might be expected, hair Hg for the Kimpro and Hg ballmill workers is in the upper part of the range recorded (Figure 30) whereas the blow-torcher has a surprisingly low Hg burden. People living very close to the Kimpro, Godonez and Bulbscor processing plants have Hg hair levels that cannot be distinguished from those of people living at a greater distance to the plants. All the hair Hg data fall within the range reported for fishing communities and for most of the Apokon population (apart from the Kimpro workers and ballmill operators) do not indicate abnormal exposure to Hg derived from gold processing activities. Even the Kimpro workers have Hg hair concentrations which fall within the range found in fishing communities where there is no exposure to mercury derived from gold processing sources. Whereas these conclusions should be treated with some caution as they are based on a relatively small sample size, they do confirm earlier results from the DOH survey of blood Hg. The Hg data do not indicate the need for Hg detoxification of any of the subjects assessed, However, it is clear that ballmill operators and workers at the Kimpro plant (and presumably also at other plants processing Hg contaminated tailings) are subject to enhanced occupational exposure to Hg which may be harmful to their health. Appropriate strategies are required to reduce this occupational exposure.

12

e

-

1

2c

18 ° ~ 6

Figure 30. Hg (mgkg) in hair results for Apokon human impact assessment (50m : people living within 50m of CIP processing plants; BM : ball mill operators using mercury; BT : blow torch operator; CIP : workers at CIP processing plant; 0 : other people not included in previous groups).

9.4 Ponce Enriquez, Ecuador

Contamination by mercury, arsenic and associated elements in solution and suspension is carried downstream to the coastal plain where commercially important banana plantations and shrimp farms are located. It has been estimated that about 2 tonnes of Hg is released to the environment each year in the Ponce Enriquez area (Land-

38

ner, 1991; CODIGEM & DINAPA, pers. comm.). Part of this mercury (estimated to be about 100 kg per year) directly contaminates streams and rivers leading to a gradual increase in Hg concentrations in the aquatic environment. Methyl mercury is more readily taken up by biota and less slowly released than inorganic mercury. Thus bioconcentration factors for methyl mercury are 30 to 900 times greater than for inorganic mercury (WHO, 1981). Bioconcentration is likely to be further enhanced at the high ambient temperatures prevailing in hot tropical regions, such as the Ponce Enriquez area. Methylation of inorganic Hg in organic rich stream sediments would magnify the environmental hazard because methyl mercury is readily absorbed by aquatic organisms. The potential risk to the economically important shrimp farms located downstream from the Ponce Enriquez mining district is thus of particular concern. Conflicting reports were received during the case study survey regarding the use of water from the Rio Siete for irrigation of the banana plantations and for pumping into the shrimp ponds. Although only a very limited amount of data is available, there are indications that fish from the estuary of the Rio Siete have Hg concentrations in the range 0.31-0.36 mgkg (wet weight) whereas the few samples of shrimps examined from shrimp farms adjacent to the Rio Siete contained lower Hg concentrations (0.01-0.04 mgkg, wet weight; unpublished data). These Hg concentrations in fish do not exceed the human health safety threshold value for marketed fish in the USA (0.5 mgkg; USEPA) and may fall close to the average concentrations for marine fish in the Pacific (0.16-0.30 mgkg in tuna and mackerel; 0.07-0.09 mgkg for other non-predatory fish; IRPTC, 1980, quoted in Williams et al., 1996a). Additional sampling and analysis of both fish and shrimps has been carried out as part of the World bank funded PRODEMINCA (Sub-components 3.1,3.2) project, but the results have not yet been reported. 9.5 Cyanide

Although the use of cyanide leaching does carry some environmental risk and also tends to raise somewhat emotive opposition because cyanides are such well-known poisons, in fact its use is more environmentally benign than that of mercury. This is because although cyanides are highly toxic, they have a very limited survival time in natural waters where they are very quickly oxidised to harmless compounds such as bicarbonates, nitrates and nitrogen gas. Thus a cyanide spill may have an immediate and dramatic toxic effect on river life, but the toxin will be completely removed by natural processes within a matter of a few days. At Diwalwal, Ponce Enriquez and Nambija there is evidence from elevated concentrations of Na and C1 in water that gold cyanidation processes are impacting surface drainage. This is not surprising as most of the cyanidation plants are located close to rivers (see Figure 12). However, the presence of Na and C1 in river waters need not necessarily indicate that cyanide concentrations are also elevated. Cyanide was not measured during the case studies reported here.

10. PROMOTION OF IMPROVED METHODS OF GOLD RECOVERY WITH APPLICABILITY TO THE ARTISANAL SECTOR 10. I Introduction

The third major objective of the project was to evaluate alternative gold recovery methods that are particularly applicable to the artisanal sector and promote their use within the case study areas. This objective was addressed by carrying out a comprehensive literature review which led to the recommendation that gravity separation should be evaluated and promoted as an efficient alternative to mercury amalgamation for the concentration of fine gold within the small-scale gold mining sector. A simple BGS-designed gravity separator was constructed and tested under laboratory conditions using Au-bearing material (ore) collected from three sites in the Philippines. Following design modifications, the gravity separator was tested under field conditions and donated to the Philippines MGB as a model for the development and construction of full-scale production equipment. The overall aim of the use of improved gold recovery methods was to help reduce mercury pollution hazards and their negative impacts on both the environment and human health. The widespread dissemination of such technologies is beyond the remit of the current TDR project, but is the central objective of a BGS project which started in April 1998 (R7120: ‘Recovering the lost gold of the developing world’).

39

10.2 Review of gold recovery methods

10.2. I Gold particle size Alluvial gold has a particle-size range typically between 5 mm and 1 0 0 pm although there is little information about the amount of gold finer than 100 pm. Hard rock gold has a wide range of particle-size, depending upon the nature of the ore - in some cases this can range down to less than 5 pm. In the large-scale gold mining (non-artisanal) sector, gold is usually recovered by a combination of physical and chemical processing methods. Commonly riffled sluice boxes and jigs are used to pre-concentrate heavy minerals (including gold) from gold-bearing ore. Shaking tables, spirals, rotating cones and bowl concentrators are used to upgrade this to a heavy mineral concentrate. Gold is commonly extracted from fine-grained concentrates by Carbon-in-pulp (CIP) cyanidation and from coarse-grained concentrates by mercury amalgamation. The efficiency of gold recovery depends upon the processing method and the proficiency of operation. A well operated modern sluice box can potentially recover up to 98%of gold coarser than 100 pm. However gold finer than 100 pm will be lost unless a fine processing method (e.g. shaking table or bowl concentrator) is used to reprocess the fine tailings. The particle-size range and typical efficiencies of gravity and chemical gold recovery methods are summarised in Table 13. Table 13. Particle-size range and typical efficiencies of gravity and chemical gold recovery methods Method Effective size range Recovery efficiency Sluice boxes(+ Reichert cones) 2500 to 100 p n As low as 20%for - 4 0 0 p n gold to 96% for cl000 p n gold Jigs 2500 to 75 p n As low as 50%for 100 pm gold to 98% for 1000 pm gold Shaking tables 3000 to 15 pm As low as 20%for 20 to 40 pm gold up to 90% for gold A 0 pm 3000 to 75 pm 65 to 80 % Spirals Rotating cones and bowl concen6000 to 30 pm u p to 99% trators Amalgamation 1500to 70 pm As low as 65% for (15 pm gold up to 98% for 4 0 0 p n gold Cyanidation Finer than 200 pm At least 80%to 99% Note. Recovery data are specific to particle-size, ore nature and processing operation.

10.2.2 Mercury amalgamation

Significant quantities of mercury are released into the environment as a result of its use during various stages of the gold beneficiation process (Figure 3 1). Gold is commonly extracted from process concentrates using mercury, which combines with gold to form an 'amalgam'. The gold is removed from the amalgam by evaporation of the mercury. Mercury is commonly added to sluice box riffles and also to grinding mills (Subasinghe & Maru, 1994). Also gold can be recovered from fine-grained tailings by washing them over a copper plate covered with mercury. Mercury amalgamation is effective for the recovery of gold from material in the size range 1.5 mm to 70 pm. Gold recovery efficiency falls for grains finer than 70 pm and typically only 65% of free gold grains finer than 75 pm are recovered. A recovery of up to 98%is quoted for one operation in Papua New Guinea (Eltham, 1984). Mercury is occasionally poured between the riffles on a sluice box in an attempt to capture fine-grained gold. However the contact time between the mercury and the gold is not sufficient to allow amalgamation to occur. Often fine gold remains suspended in the flow of material above the riffles and does not come into contact with the mercury. Up to 30% of the mercury used in sluices in Papua New Guinea finds its way directly into local rivers. Passing the tailings over 'amalgamation units' or through mercury filled columns has been recommended as a method of recovering this fine gold. However, these are ultimately unsatisfactory as they still pose a threat to the environment (Subasinghe & Maru, 1994).

40

Mine

Crusher

I Cyanidation 1

Final tailings

Figure 31. Flow chart for gold beneficiation processes employed by artisanal mining sector in the Philippines and Ecuador. Stages at which Hg is lost to the environment are indicated by the shaded boxes.

10.2.3 Cyanidation Cyanidation is the process whereby gold is recovered using a cyanide solution. Gold is dissolved using the cyanide solution and the resulting complex, Au(CN)2, can be removed from solution by various methods (Deschenes, 1986): i) The “Merill-Crowe” process, is used to remove the gold from the cyanide by cementation with powdered zinc. ii) Activated carbon absorption (otherwise known as C-I-P, carbon-in-pulp) is used for the processing of ores with a high slimes content which are difficult to treat by the Merrill-Crowe process. Typical gold recovery efficiencies for cyanidation range from 95 to 99% (Marsden h Fuerstenau, 1993). Cyanidation is more effective for finer grained gold as it relies on dissolution and is particularly effective for the recovery of gold finer than 200 pm. Free coarse gold is often recovered prior to cyanidation to avoid the cost of grinding and of the extra reagents required. 10.2.4 Costs and environmental impact

Gold recovery methods differ in terms of their financial cost (capital and operating) and their potential impact on the environment (Table 14). Operating costs will, of course, depend on scale and local prices for capital equipment and reagents. The actual environmental impact will depend on how effectively the mineral beneficiation processes are carried out and on the precautionary measures taken to ensure that cyanide, froth flotation reagents or mercury do not enter the drainage systems, for example. Cyanidation and froth floatation, if properly managed, are not as inherently hazardous as the use of Hg for amalgamation.

41

Table 14. Relative* cost and potential environmental impact of gold recovery methods Recovery method

Relative* cost

Sluice box Jig Shaking table Spiral Rotating cone Bowl Drum Magnetic separation Electrostatic separation Hydrocycloning Froth flotation Amalgamation (mercury) Cyanidation

1 3 2 3 2

Relative* potential environmental impact 1 1

4 4 4 4 2

1 1 1 1 1 1 1 1

3-4 2 3-4

3 4 3

Note * Qualitative ranking from low (1) to high (4).

10.2.5 Gold recovery flowsheets

Typical flowsheets for large-scale gold processing of alluvial and hard rock gold comprise the following stages: Alluvial gold processing Extraction - Typically by hydraulic sluicing or dredging. Feedpreparation - Drum scrubbing & trommel screening. Coarse material (>10 mm) to tailings. Fines dewatered by hydrocyclone. Fine screening (e.g. 500 pm) optional. Primary processing - Riffled sluice or jig concentration. Coarse tailings removed. Secondary processing - Shaking table or bowl concentrator. Fine tailings removed. Fine processing - Spiral and shaking table concentration to process fine material. Gold production - Amalgamation of concentrates with mercury to extract gold. Hard rock gold processing Extraction - Open cast bench mining or deep mining. Feedpreparation - Primary jaw or gyratory crusher (e.g. to e150 mm) followed by secondary milling (e.g. ball,rod- or semi-autogenous milling). Size classification by screening or hydrocycloning. Coarse processing (>500 pm) - Jig and shaking table or bowl concentration. Fine processing ( d o 0 pm) - Reichert cone concentration followed by spiral andor shaking table concentration. Froth flotation of very fine material (e.g. e180 pm). Gold production - Coarse concentrate by amalgamation. Fine concentrate by milling and CIP cyanidation. 10.3 Recommendationsfor improved recovery of fine gold not involving amalgamation

The review of separation techniques showed that several methods are effective for the recovery of gold coarser than approximately 100 pm but for gold finer than approximately 50 pm only chemical methods are very efficient. The efficient recovery of gold coarser than 50 pm should be possible using simple gravity methods and some recommendations for possible implementation are given below. If there is evidence that there is significant gold finer than 50 pm a cyanide treatment of the fine tailings is necessary for recovery. Recommendations for the recovery of fine gold include improving current operating practice, modifying the process used, and introducing alternatives that are more efficient. These recommendations apply to gold recovery by sluice box and the main recommendations for their use in small-scale gold mining are summarised in Figure 32: 1. Wet screening to remove material coarser than 500 pm (which should be passed over a sluice box to recover gold coarser than 500 pm). 2. Material finer than 500 pm to be passed over a second sluice, to recover gold coarser than 200 pm. 3. Tailings from Stage 2 to be wet screened to remove material coarser than 200 pm (ideally free of gold) and then passed over a shaking table (to recover gold down to 50 pm).

42

Trommel Screen 500pm

Sluice (Au

1

c500pm Coarse Au; > 500pm Tailings <500pm

Sluice (Au 200 -500pm)

Screen 200pm Fine Au;

fl

Tailings <200pm; Au c40pm Figure 32. Flow chart for recommended small-scale sluice box processing

10.3. I Improvements to current sluice box practice Optimise clean-out

The time interval between clean out of sluice box riffles is dependent upon the nature of the material processed and the operating conditions applied. It is recommended that the time interval should be short enough to enhance the recovery of fine gold (that would otherwise be lost due to solids packed around the riffles). However, this should be tempered by the requirement to maintain a relatively high throughput, which if reduced by too much cleaning will ultimately reduce gold production. Appropriate configuration

It is recommended that the tilt of the sluice box should increase with increasing particle-size. Typically (in the Yukon, Canada) for material finer than 1 mm a tilt of 7 to 12" is used. The tilt angle is increased to 12 - 14"for material coarser than lmm. Appropriate operating practice

Appropriate feed and wash water rates are dependent upon the nature of the material being processed, especially the particle-size but also the clay and /or heavy mineral contents. It is recommended that the feed and wash water rates are high enough to enable the efficient separation of coarse-grained gold without excessive loss of fine-grained gold. 10.3.2 Process modifications Washing prior to sluicing

It is recommended that gold ore with a significant proportion of clay-bound and weakly-cemented material be washed ('scrubbed') and screened prior to sluicing. This will enable the liberation of gold trapped in clay.

43

Processing coarse and fine separately

Operating conditions appropriate for the recovery of coarse-grained gold are different to those for fine-grained gold. Typically, material containing coarse and fine-grained gold will be processed on the same sluice. This will invariably lead to a loss of gold. Ideally coarse gold should be recovered on a steep sluice using high feed and wash rates, whereas fine gold should be recovered on a shallower sluice using lower feed and wash rates. It is recommended that ore is screened prior to sluicing and the resulting coarse and fine streams are diverted down different sluices. This will improve overall gold recovery. Expanded metal and angle iron sluice riffres

The use of alternative riffles will enable a higher recovery of gold. Expanded metal riffles are recommended for gold finer than 1 mm and angle iron riffles for gold coarser than 1 mm. 10.3.3. Efficient alternatives Replace sluices with jigs

Jigs are efficient in the recovery of gold down to 75 pm and have been used to replace sluices, with a resultant increase in gold recovery. Introduce shaking tables and bowl concentrators t o concentrate fine gold

Shaking tables and bowl concentrators are efficient in the recovery of gold down to at least 40 pm and are used to process fine gold-bearing ore. It is recommended that these be considered for the recovery of fine gold that is not recovered by sluice box. If small-scale operators can use the recommended methods at efficiencies close to that achieved by large companies, and in the laboratory, this will be as good as, and often superior, to mercury amalgamation. 10.4 Laboratory evaluation of BGS-designed gravity separator for gold recovery 10.4. I Introduction

The aim of this work was to construct a simple device, using technology appropriate for less developed countries, for the recovery of fine-grained gold (e100 pm) from alluvial deposits or crushed bed rock ore. This device would be, to a large extent, an alternative to the environmentally damaging use of mercury for gold recovery. Previous work indicated that the most appropriate method of recovering fine-grained gold is through the use of a shaking table type gravity separator (Mitchell et al, 1997). Shaking tables are one of the most common forms of gravity separators in use. They consist of a flat ‘deck’ with parallel riffles that ‘trap’ gold during operation. During operation the deck is vibrated longitudinally, with an ‘end-knock’ induced at the end of the table where the concentrate will form. The feed is introduced as a slurry, along with wash water, from the upslope side. The ‘end-knock’ encourages heavy minerals to migrate along the riffles to the end of the deck and light minerals are washed over the riffles to the downslope side of the deck. Shaking tables are effective in the processing of material in the size range 3 mm to 50 pm and are capable of recovering up to 90 wt.% of gold present in that size range. In the absence of shaking tables, sluice boxes are the most commonly used means of recovering gold from alluvial gravels and crushed ores in less developed countries. Efficiently operated sluices can achieve gold recoveries of up to 80 wt.%, whereas the makeshift sluices operated by small-scale miners recover less than 50 wt.%. These low gold recoveries are in part related to the particle-size of the gold. Gold finer than 200 pm will invariably be lost to the tailings. Other factors leading to poor recoveries include inappropriate sluice box design and operation, high feed and wash water rates and long time intervals between riffle clean-out to remove accumulated heavy minerals. The shaking table constructed as part of this work was designed to be used in Stage 3 of the recommended gold recovery scheme (Figure 32). A set of samples collected from the Philippines were first characterised and then used as test materials for the shaking tables.

44

10.4.2 Sample collection for beneficiation tests

A pre-requisite for beneficiation tests was a supply of feed material (hand crushed or milled ore, alluvium), representative of that currently processed by artisanal miners in the Philippines. On account of the widely variable granulometry of gold in both primary and alluvial deposits, samples for experimental beneficiation studies were obtained from a range of settings within two independent gold provinces of the Philippines. These comprised the small-scale processing operations at Acupan and Kias Creek centred on Baguio (Luzon) and the Gango mining area near Cagayan de Oro, Mindanao. On account of the nugget effects which characteristically bias gold assay data for small ore or alluvium samples, a minimum of 20 kg of crushed or milled material was collected from each site. Acupan Benguet Gold Operation (BGO) concession. The concession, located 15 km east of Baguio, forms part of the extensive property of the Benguet Gold Operation (BGO), who exploited the high-grade veins commercially until 1993 when flooding of several adits forced closure. Approximately 200 small-scale miners are now active within the area. Ore is hand-crushed and milled at several small balVrod mill plants. Coarse Au is then recovered gravimetrically using a felt-bed sluice. On the basis of information provided by local miners, recoveries gained by this method may reach 80 g/t. Prior to sampling, a brief assessment of a approximately 3 kg milled ore sample was undertaken using a conventional prospecting pan, and revealed abundant coarse, leafy Au. Overflow tailings from the Acupan mills are sold to BGO, who transport the material to their Antomok operation for CIP cyanidation. The use of mercury by smallscale miners in this area has been severely restricted by BGO, who monitor activities rigorously. Some mercury usage does, however, persist. A 20 kg milled ore sample (BGl) was collected from a holding tank into which ball-mill output from a single operation was emptied prior to sluicing. Tailings overspill (BG2), of value for assessing percentage recovery achieved by the sluice, was recovered by placing a large pan directly under the sluice discharge, and then decanting all fluid after a suitable settlement period. A small sample of unmilled feed chippings (BG4) was retained, including at least one fragment bearing free gold. Kias Creek

Small-scale workings along Kias Creek, approximately 1 km from the northern end of the Philex mine access road, are focused along a N-S trending structure within which silicic fault breccia carries up to 100 g/t Au. Samples were collected from two adjacent millinghecovery plants. The first operated an identical system to that encountered at Acupan, although overspill tailings were reportedly discarded rather than sold. The second did not utilise a sluicing system. Instead, the coarse (>150 pm) milled output was panned and amalgamated, while the fine fraction was fed to an agitation tank containing a sodium cyanide solution. Both milled ore (BG5) and tailings overspill (BG6) samples were collected from the first of these operations. Due to the nature of the process described, milled ore only was available from the second (BG7). Gango

At Gango, near Cagayan de Oro, Mindanao, Au mineralisation occurs in narrow veins in phyllites close to the contact with serpentinised peridotite. Mining of placer deposits was initiated in 1960 and gold production in 1990 was estimated at 153 g/day, with 18 rod-mills operative for the processing of ore. Until the late 1980s, gold was amalgamated by the direct addition of Hg during milling, after which the residues were discarded. More recently, small-scale CIP cyanidation plants have been developed locally, and are utilised on a cooperative basis. Because the low-recoveries attained by amalgamation, tailings previously treated in this manner (and hence containing Hg-residue) are now being widely reprocessed by CIP operators. Samples (approximately 15 kg each) of hand- crushed material with an estimated grade of 40 gh were obtained from two independent adits.

45

10.4.3 Characterisation of gold bearing samples

The mineralogy, gold grade and gold size distribution of the test samples is summarised in Table 15 and Figure 33. Table 15. Gold assay & mineralogy of gold-bearing samples ‘as received’, Philippines Sample

Gold assay Mineralogy (dt) Acupan Benguet Gold Operation (BGO)concession, Baguio, Luzon BG 1 96.7 Quartz ****, pyrite **, anglesite * BG 2 25.4 Quartz ****, pyrite *, anglesite * BG 4 46.4 Quartz ****, pyrite *, ? dolomite * Kias Creek, Baguio, Luzon Quartz ****, pyrite **, chlorite **, albite *, ? dolomite * 12.1 BG 5 Quartz ****, albite *, chlorite *, pyrite *, ? dolomite * 9.8 BG 6 Quartz ****, ? albite *, ? mica * 3.0 BG 7 Gango, Cagayan de Oro, Mindanao Quartz ****, dolomite **, pyrite *, chlorite *, muscovite * 12.6 Sample 1 Quartz ****, dolomite ***, pyrite **, chlorite * 8.6 Sample 2 Note. g/t = grams per tonne. **** = dominant (>50 wt.%),

*** = major (20-50 wt.%), ** = minor (7-20 wt.%) and * = trace (c7 wt.%).

Acupan (BG1)

Kiss Creek (BGS)

Figure 33. Gold grain size (pm) distribution in sluice box feed.

10.4.4 BGS-designed shaking table

The BGS-designed shaking table adopted technology appropriate for less developed countries, as outlined in the introduction (Section 10.4.1). Therefore, the design was restricted to that of a traditional riffleddeck using widely available materials and a manual drive mechanism (Figure 34). The frame and supporting base of the separator were constructed out of hard wood, with a Formica deck surface (slightly roughened with wet and dry abrasive paper). The drive mechanism consists of bicycle gear wheels and chains, with an appropriate gearing ratio to step up the manual drive input. The drive is powered via a hand-operated crank. The manual drive input is stepped up via a two-stage gearing process, producing an overall gearing ratio of approximately 5 to 1. This rotational motion is translated into reciprocal motion by the use of an eccentric cam, which is in turn attached, via a universal ball joint, to the shaking table deck. When the handle is turned at 1 revolutiordsecond (a comfortable speed for the operator), this produces a longitudinal motion of about 300 strokeshinute, which is suited to the separation of fine particles. The eccentric cam causes stretching of a strong rubber band (made from old car tyre inner tube), which releases suddenly to produce the ‘end-knock‘ effect. The slope of the deck is controlled by appropriate ‘wedging’ of the deck sub-base. Wash water is supplied via perforated plastic piping. The tailings and concentrate collectors consist of plastic drainpipe, suitably partitioned to achieve correct collection of products. This prototype table was produced with a hand cranked drive mechanism but this could readily be modified to be powered by a bicycle, or an electric motor after appropriate alterations to the gearing mechanism.

46

Tailingscollector

Drive mechanismCam-

p t l

Aain

,

7I

o[ t i I0 I

A

Concentrat collector

--

Riffies

.

I

Reciprocal motion

1 4

~

7

Water in

I

Or

1 0 Gearing ratio & rpm 5tep-u~ A --> D 5.123 One turn of crank eqlnvalent to 5 badc & forth mouements of deck. Crank tumed atapprox 1 rewlutions per semnd which S equivalent to 300 rpm at the table.

U

Figure 34. BGS-designed shaking table

10.4.5 Results

The mineral processing trials conducted using the gold-bearing samples from the Philippines confirmed the effectiveness of shaking tables for recovery of gold, down to a particle-size of approximately 50 pm. Both of the samples used for the large-scale trials contain a significant proportion of gold finer than 50 pm. Sample BG 5 is coarser than BG 1 with a higher proportion of gold between 125 and 500 pm (Figure 33). This accounts for the dramatic difference in the processing results. The concentrate produced from BG 1 has a relatively low gold assay and also only represents a tenth of the gold in the sample. This is probably due to its fine grain size; the small gold grains were probably entrained within the tailings and swept over the riffles. In contrast, the concentrate produced from BG 5 has a much higher gold recovery, albeit at a lower gold assay, due mainly to its coarser particle-size (Table 16). Table 16. Laboratory results: Gravimetric concentration of gold from ground rock using shaking tables Au (g/t) original

AcupanWBGl AcupanBBG1 Kias CreekWBG5

97 97

12

Yield (%) 2.8 2.5 5.0

Au (dt)concentrate

1867 2305 360

Gold recovery (wt.%) 57 62 85

W = Wiljley laboratory shaking table; B = BGS shaking table

In artisanal gold beneficiation operations, sluice box concentrations are upgraded by panning. This process was simulated in the laboratory tests by using a ‘Superpanner’ to upgrade concentrates from the shaking tables. Processing of BG1 using both the Wilfley and BGS-designed shaking tables produced markedly different results. The Wilfley table concentrate has a gold assay of 25%, but this only represents 10% of the gold present in the sample; whereas the BGS-designed table concentrate has a gold assay of only 10%, which represents 20% of the gold present in the sample (Table 17). The BGS-designed table has therefore removed a higher proportion of gold from the original samples, albeit at a lower gold content, but this indicates that it is more effective than the Wilfley table.

47

Table 17. Laboratory results: Gravimetric concentration of gold from ground rock (using shaking tables with concentrates upgraded using a ‘Superpanner’)

Au (g/t) original AcupanWBGI AcupanBBG1 Kias Creekw BG5

97 97

12

Yield(%) 0.003 0.02 0.02

Au (%) concentrate

25.1 10.0 7.5

Gold recovery (wt.%) 10 21 60

W = Wilfley laboratory shaking table; B = BGS shaking table

Another means of comparing the performance of the two shaking tables is to ignore the results of the final upgrading by Superpanner. The performance of the BGS-designed shaking table is then seen to be comparable to, if not slightly better than, the Wilfley shaking table (see Table 16, results for Acupan BG1). The gold assays and recoveries prior to final upgrading by ‘Superpanner’ can be seen in Table 16. The combined concentrate produced using the BGS-designed shaking table and ‘Superpanner’ resulted in a product with a slightly higher gold assay and recovery than that produced by the Wilfley shaking table (Table 17). However, despite these small differences the results are effectively identical. This indicates that for gravity separation the BGS-designed shaking table is as effective as the Wilfley shaking table. Processing of the tailings samples (BG2 & BG6), using the ‘Superpanner’, yielded similar results to those produced by processing of the corresponding feed samples (BGl and BG5 respectively) using the Wilfley table. The proportion of gold recovered from BG6 is approximately four times that recovered from BG2. This is probably due, as with the difference between BG5 and BGl, to the coarser grain size of the gold present in the samples from Kias Creek. However, the relatively high proportion (68 wt.%) of gold recovered from BG6 demonstrates that simple methods of gravity separation can be used to extract fine-grained gold from tailings, provided it is coarser than 50 pm (Table 18). The results of the investigation have shown that a relatively high proportion of gold can be recovered using simple gravity methods. However, the gold content of the products has generally been low, less than 1%. This is a function of the amount of material available for processing, as after each processing stage the amount of material available for subsequent processing stages decreases. In these studies, a Superpanner was used to make higher concentrations with around 10% gold and this might be possible using the table if sufficient material were available. Processing larger amounts of material would enable further processing stages to be carried out and production of higher-grade gold concentrates. For example at a mine, the concentrates produced during a whole day could be reprocessed. There is still a need to establish whether the table could be used to clean the concentrate to a purity suitable for sale as if not another method for this final stage is still required. Table 18. Laboratory results: Gravimetric concentration of gold from tailings

Acupan BG2 Kias Creek BG6

Au (g/t) original 25 10

Yield (%) 0.7

2.1

Au (g/t) concentrate 66 1 434

Gold recovery (wt.%) 17 68

W = Wilfleylaboratory shaking table, B = BGS shaking table

The findings of the investigation have reinforced the value of characterisation prior to mineral processing. Mineralogy, particle-size and texture all have an influence on the performance of mineral processing trials. For example, the proportion of gold finer than 50 pm in an ore has a significant effect on the proportion of gold recoverable using simple gravity methods. Therefore, a determination of the particle-size distribution of gold could be used to indicate the likely success of using gravity methods for the recovery of gold from a given ore. 10.5 Field testing and promotion of BGS gravity separator

The BGS shaking table was shipped to the Philippines and taken to Baguio, the mining area from which some of the samples for the laboratory tests had been obtained. Field trials (Plate 10) were carried out in collaboration with counterpart staff from the headquarters in Manila and the Baguio Regional Office of the Mines and Geoscience Bureau (MGB). The trials were held in two areas with the participation the local Small-Scale Miners Associations.

48

10.5. I Existing practices

Ore from small mines is carried in sacks to the ‘processing plant’ where the ore is crushed in a ball mill POWered by a diesel engine. After wet milling for a long period (2-3 hours), the slurry is emptied into a ‘washing’ tank (3m by 2m and 50 cm deep), where the slurry is agitated by a miner with a wooden paddle to separate the mud from the heavier minerals. After washing for about an hour the muddy water and ‘washed’ milled ore are passed over a shallow sluice, with no riffles, covered with blanket cloth. A heavy, gold bearing, concentrate collects on the cloth and this is removed periodically by washing the cloths in a large basin. The tailings were collected for further processing. At Kias the miners no longer use mercury but are now experimenting with cyanide extraction, both by heap leach and CIF’ separation in an oil drum. The tailings from cyanidation were disposed of by dumping them in the creek. At Acupan, the main procedures were the same as at Kias, except that the tailings are sold to Benquet Mining Company. 10.5.2 Kias Area Field tests

Tests were carried out on the tailings material from which a good separation of heavy minerals was achieved. Approximately 2 kg of tailings was processed in a period of 1 hour. The resulting concentrate was then evaluated by hand panning which revealed a significant quantity of very fine gold. Samples of the concentrates and starting material were collected for laboratory examination in the UK. Laboratory tests on products

The concentrates from the table were further separated using a Superpanner to upgrade the gold to make examination easier. The concentrate contained a considerable amount of gold, with a few ‘coarse’ grains around 0.2-0.4 mm, which are generally flattened and flaky due to the crushing. Most of the grains were much finer with the dominant population around 30-50pm (Plate 1l), but grains as small as lOpm had also been recovered. This is quite remarkable as most of the grains recovered are in the size range where shaking tables are generally not considered effective. The shape of the gold grains may be an important factor as it can be seen from the photograph that most of the fine gold is equant, which makes separation more effective compared to flaky grains. It appears that very small gold grains do not get flattened during the crushing process. The concentrates only contained a very small proportion of coarser grains because the material processed was tailings and most of the coarser gold would have been removed on the sluice by the miners. 10.5.3 Acupan Benquet Mine area Field tests

Considerable difficulty was encountered with setting up the table due to the absence of a suitably flat and stable surface at the trial site. The milled ore tested here was much muddier than the tailings processed at Kias and this also makes separation more difficult. Concentrates were collected from both freshly milled ore and tailings from previously sluiced material. When the concentrates were briefly examined on site they did not appear to contain much gold. The trials here demonstrated very clearly how crucial stable conditions are for the table to work effectively. Valuable insight was gained about the operation of the table and features needing design improvements were identified. Laboratory Tests

The field concentrates were further processed with the Superpanner. The concentrate from ore material contained gold with a wide range of grain sizes, with several around 0.5 mm, many around 0.1-0.2 mm, much very fine gold around 50pm and some down to 20pm. The concentrates were not as rich as those from Kias but contained far more than was expected from the field examination. The concentrate from tailings contained a few ‘coarse’ grains around 0.2 mm but most was less than 60pm with many around 30pm (Plate 12). The trials at Acupan though not as successful as at Kias had actually been far better than was realised at the time. This is possibly because in the field it was difficult to see the gold due to Fe-oxide coatings on the coarser grains and the finer grains being hidden in the fine muddy material.

49

10.5.4 Miners reaction to the shaking table trials

The miners at Kias were very impressed by the trials as they had seen fine gold recovered from the tailings from their current operation. Although they no longer use mercury to try to recover this gold, they were aware that cyanidation also has associated problems and hazards and could see that a simple gravity method that only used water was preferable. This was both from a pollution point of view and also because they did not need to buy relatively expensive reagents. They studied the table in detail and made sketches of it, and asked if it could be made larger to increase the throughput, which it can. They suggested some modifications to the design and in particular thought that it could be powered by a belt drive from the ball mill shaft so that it could be run continuously for long periods without a need for hand winding. Once we had started the trial, several other miners from nearby mines brought samples of their tailings for testing and similar good results were obtained. The miners would be very keen to have a table in regular use if it could be produced at a price they could afford. The miners at Acupan had been involved in the problems of setting-up the shaking table and were unfortunately unable to appreciate the effectiveness of the table, that was only revealed by laboratory examination of the field concentrates. They were interested to try it further, particularly due to the reports of successful trials at Kias the previous day. They were unsure whether this might violate their concession agreement that definitely precluded the use of chemical extraction. The MGB staff who were involved in the trials were impressed by the simple construction but effective operation of the table. It was unfortunate that we were unable to demonstrate it to a greater number of staff in Manila due to the water shortage. The shaking table was donated to MGB and they plan to carry out further trials and particularly in the mining areas further from Manila where mercury amalgamation is widely used. For time and logistical reasons, it was not possible to test the table in these areas during a short visit. 10.6 Conclusions

1. Laboratory characterisation of gold bearing ores and tailings from the Philippines established the grade and grain size distribution of gold within the samples. This showed that they contained dominantly fine-grained gold with one of the main samples studied having 60% <63pm. References in the literature suggested that shaking tables are only effective on grain sizes down to about 50pm. 2. Laboratory separation test using a commercially produced laboratory-scale Wilfley table recovered only 20% of the gold from the finest-grained sample but nearly 80% from one that was slightly coarser. 3. A simple, hand-powered, shaking table was designed and constructed of cheap and simple materials that are available in developing countries. It was particularly aimed at the recovery of fine-grained gold. Laboratory trials showed that this table was as good as and probably slightly more effective for the separation of fine-grained gold than the commercial Wilfley table. 4. The simple shaking table was taken to the Philippines for field trials. Field trials demonstrated that it is remarkably good at recovering fine-grained gold when properly set-up and recovered a significant quantity of gold from miners tailings where the grain size was only around 30pm. The trials also showed how important and sometimes difficult it can be to adjust the table to the correct settings. 5. The small-scale miners were impressed by the performance of the table and were keen to test it further and possibly produce a larger version for regular use as an alternative to cyanidation. 6 . A simple shaking table can be very effective for the recovery of fine-grained gold down to a grain-size of around 20pm and hence could play an important role in providing an alternative to the use of mercury in the extraction of gold from bulk samples. The heavy mineral concentrates produced contain several weight percent gold and still require further processing to remove the remaining impurities. This might be possible with the table but this has not yet been demonstrated. Carefully controlled amalgamation followed by roasting to release the gold and condensation of the mercury vapour using a retort might still be required.

10.7 Recommendations 1. The use of a shaking table can increase the amount of fine-grained gold recovered during processing of ore.

Figure 35 illustrates the recommended sequence of process stages, incorporating a shaking table, for the recovery of fine gold.

50

2. It is assumed that most artisanal gold miners have little knowledge of the particle size distribution of the gold present in their ore. Therefore, the most practical method of establishing the effectiveness of the shaking table for the recovery of gold from a particular ore, would be to perform trials. 3. It is also unlikely that the use of mercury amalgamation for the final concentration of gold from gravity concentrates will be easily replaced. However, its use could be restricted to this stage only. The use of mercury in tromol mills, sluice beds and other process equipment could be significantly reduced or eliminated by demonstrating the effectiveness of simple gravity techniques.

Ore extraction

v

(
0 microns)

Washing I fines removal I

v Gold concentrate (coarser than 500 microns)

Coarse-grained gold

recovery (sluice box stage)

‘Short-circuit’ (ore finer than 500 microns)

9 Final concentration of

Tailings (containing gold 4 0 0 microns)

I

Gold concentrate (finer than 500 microns)

Fine-grainedgold recovery

I

(shaking table stage)

I

Cyanidation (if significant proportion of gold <30 microns)

Tailings

Figure 35. Recommended sequence of process stages, incorporating a shaking table, for the enhanced recovery of finegrained gold.

51

I I. LEGISLATION AND ENVIRONMENTAL STANDARDS

I I. I Legislative control of mercury contamination

The sustainable development of artisanal gold mining is severely constrained by: (1) environmental, health and safety problems such as those linked to the use of mercury; (2) use of inefficient methods and equipment; (3) lack of appropriate legal, regulatory and institutional frameworks. The World Bank International Conference on Development, Environment and Mining (World Bank, 1994) and the International Roundtable on Artisanal Mining (World Bank, 1996) identified the need to: (1) set reliable environmental standards that are practical to apply and appropriate to local circumstances; (2) strengthen environmental impact assessment processes; (3) encourage the establishment of comprehensive environmental management systems2. Regulation that promotes pollution control is based on the assumption that emissions and waste materials are an inevitable part of minerals and metal production processes. Regulation is designed to reduce these impacts using Best Available Technology (BAT) (e.g. improved, less contaminating, mineral processing procedures). Attempts to bring about pollution prevention has generally been through legislation (e.g. US EPA Pollution Prevention Act 1990) and specific regulations for mining activities (e.g. in Bolivia (Gaceta Oficial de Bolivia, 1997) and Ecuador (PRODEMINCA, 1998)). Pollution control is achieved mainly through waste assessment procedures such as Environmental Impact Analysis and Environmental Audits but also includes environmental standards for exploration, mineral extraction, and processing. In Ecuador, there are additional standards applicable to the artisanal sector (PRODEMINCA, 1998). In many developing countries, enforcement is generally extremely restricted and is almost totally concentrated on the larger polluters. In general, environmental legislation has been seen more as a major cost-burden on the mining sector rather than providing an incentive to develop and use processing techniques that will reduce environmental impacts. In the artisanal gold mining sector, the over-riding concern is with mercury pollution. Potential items of significance in a regulatory framework include both controlling the supply of mercury as well as the imposition of practical (attainable and enforceable) legal standards for both water and sediment quality and also discharges. A monitoring protocol needs to provide guidance on which media need to be sampled and also suggest threshold levels that can take into consideration sampling and temporal variance. There is little financial incentive for miners to stop using mercury as it often costs only 3 to 4 US dollars a kilogram. One option that might be used in the control of mercury would be to licence buyers and sellers within a legislative framework and to have a register of mercury users. Taxing Hg imports to discourage its use might help but would probably lead to illegal importation of Hg. Imposition of mercury bans in developing countries is likely to be totally ineffective as many artisanal miners operate illegally. Much of the mercury is reported to be supplied to artisanal miners by the gold dealers to ensure that the dealers get the gold produced. The gold buyers who re-heat the gold to remove that last vestiges of Hg get the highest Hg doses, so legislation might be used in an attempt to control their working conditions in order to ensure that they (and their neighbours) inhaled less Hg. For the miners, a legislative framework and standards could be set to ensure that Hg was used in smaller amounts; that tromol mills were replaced by gravity separation etc.; and that tailings of ore treated with Hg should pass through a series of settling ponds before water containing suspended particulate matter was allowed to pass into water courses. Settling ponds would need to be cleaned out regularly. There also remains the problem of disposal of contaminated tailings. If tailings are dumped, they are still likely to be washed into rivers; especially artisanal gold mining areas characterised by steep topography and narrow valleys, such as the Diwalwal, Ponce Enriquez and Nambija areas. It had been anticipated that recommendations for maximum permissible environmental mercury levels in the Philippines would evolve from the biological and human impact assessments (Phases 2 and 3 of this project) and that these would be incorporated into new legislative controls on the small-scale mining sector being prepared by the MGB at the time the project started. It was also intended to recommend that the Governments of other developing countries afflicted by the impacts of small-scale gold mining should adopt these standards. Practical (i.e. attainable and enforceable) legal standards for environmental mercury levels and discharges need to be complemented by the inclusion of monitoring guidelines within all new legislation. As the authority responsible for the control of mining activities and related environmental impacts, the MGB needed to formulate legislation to regulate mercury usage in small-scale mining activities with the aim of preventing further Some of the strategies and policies for the small-scale (artisanalandor informal) mining sectors have been evaluated in Mining and EnvironmentalResearch Network (MERN) Working Papers (Holloway, 1996; Echavania, 1996; Hanai, 1996;Warhurst and Thomas, 1998).

52

damage to the environment and human health. Prior to drafting appropriate legislation it was necessary to (a) establish practical targets for environmental mercury levels, (b) provide economically viable alternatives to mercury amalgamation procedures used by the small-scale mining sector and (c) design an appropriate monitoring protocol to assist in enforcing environmental controls. In the event, the new Philippines mining laws were pushed through much faster than had been envisaged, effectively precluding BGS’ involvement in this aspect of the project programme. Insufficient information has been compiled through the current project to establish maximum permissible environmental levels (based on impact assessmentldose response data). One specific problem with regard to environmental standards incorporated in legislation is the lack of suitable Certified Reference Materials (CRMs) certified for their content of specific chemical forms (species) of elements like mercury and arsenic. The lack of CRMs for speciation analysis hampers quality control, although CRMs for methyl mercury in biological matrices and sediments are reported to be available. Legislative approaches have serious limitations, especially as a mechanism for controlling the small-scale, artisanal mining sector. The use of legislation to control working conditions among artisanal miners is commendable in theory but somewhat unrealistic within the context of most artisanal gold mining operations (Cleary and Thornton, 1994). Legislation for the artisanal sector is in many respects futile as such activities usually operate beyond the law. For example, at Diwalwal in the Philippines, up to 100,OOO people work a world class gold deposit which is officially held by four major international companies. In the absence of the political will to enforce the law, legislative controls will be ineffective. Having a legislative framework does not necessarily mean that it will be adhered to. I I .2 Alternatives to legislative controls

Scott (1998) evaluated the environmental impacts of various small-scale industries in the Third World and concluded that sole reliance on the enforcement of any statutory regulations and standards to control and reduce environmental damage by small-scale industries (including artisanal gold mining) is unlikely to work and that alternative approaches are needed. Whereas statutory regulations and standards are required to protect the environment, public health, and to provide guidance to producers and those affected by pollution, Scott (1998) concluded that regulations are, in general ineffective. Alternative approaches need to be considered in order to bring about better working conditions and a reduction in emissions and waste. These include: direct incentives or financial return to the producer; cleaner production technologies which demonstrably save the producer money andor increase his income (e.g. by increasing recuperation of gold, for example, by the use of more efficient mineral processing technologies) use of ‘voluntary’ compliance methods to encourage improved environmental performance. A ‘stakeholder’ community-based approach involving producers, local authorities and small-scale miners associations needs to be promoted to share information about pollution impacts, to develop environmental awareness and to bring about a reduction of the negative environmental impacts of small-scale mining. Drivers to bring about change and adoption of safer practices must be economic rather than regulatory. Artisanal miners must be provided with more cost-effective methods of gold beneficiation from which they can gain tangible economic benefits. These benefits must have immediate impact, as the pre-eminent concern of people involved in the artisanal mining sector is to earn a living. Human health and financial security over the long-term is a much lesser concern, as is the impact of their activities on the environment. In order to bring about the adoption of cleaner technologies and the stakeholder approach to enforcement, there is a need for educational, training and awareness-raising programmes (e.g. Eve et. al., 1996; Camara et. al., 1997; Veiga et. al., 1995a, 1995b). Development agencies, and especially NGOs, have a critical role to play in supporting this approach (Scott, 1998). The implementation of cleaner technologies and the ‘stakeholder’ approach have been the important foci of NGO led projects in Bolivia (Manejo Zntegrado del Medio Ambiente en la Pequeria Mineria; Appleton, 1998b), Ecuador (Proyecto Mineria sin Contaminacion) and Zimbabwe (Shamva Mining Centre: run by Intermediate Technology and the Small-scale Miners Association of Zimbabwe). The projects in Bolivia and Ecuador are advised by PROJEKT-CONSULT GMBHand partly financed by COTESU(Swiss Government Technical Co-operation)

53

12. DISSEMINATION WORKSHOPS The project results were promoted at two dissemination workshops held in Quito, Ecuador and La Paz, Bolivia, and also at the First Asia Pacific Symposium on Environmental Geochemistry held in Hong Kong. The Quito workshop sessions were attended by up to 73 participants from the state, parastatal, mining and environmental protection organisations, the university/polytechnic sector, and a wide range of NGO environmental action group. Participants found the workshop both useful and educational and especially highlighted the wider application of geochemical data for planning, agriculture and other environmental studies. Requests were received for DFID KAR project reports - these were distributed through the British Geological Mission in Quito. The La Paz (Bolivia) workshop provided a forum for discussions between professionals and academics involved in mining, mineral exploration and environmental monitoring as well as representatives of the main environmental NGO’s in Bolivia. The workshop sessions were attended by up to 95 participants. Participants were highly complimentary about the workshop, particularly with regard to its practical application of geochemistry to mining potential, environmental contamination and humadanimal health-related problems. 13. CONCLUSIONS AND RECOMMENDATIONS Magnitude of contamination

Monitoring studies highlighted the marked impact of artisanal gold mining on the flux of mercury into the aquatic environment. In the vicinity of the Diwalwal artisanal gold mining area on the island of Mindanao, in the Philippines, this impact is particularly notable because of the localised prevalence of ‘dissolved’ Hg at concentrations hitherto undocumented in artisanal gold mining localities world-wide. The magnitude and extent of contamination by mercury, derived from both mine tailings and mineral processing activities, and potentially harmful elements (such as arsenic, copper and lead) derived from sulphide mineralisation is highly variable, but is generally a predictable function of geological, environmental, social and technological factors. For example, this project has identified how mangrove systems can act as natural filters for particulate contaminants and prevent mercury contaminated sediments from reaching the marine system of Honda Bay, PalaWan. Protocol for monitoring contamination

Environmental agencies (EAs) in developing countries need to be able evaluate both continuous as well as sudden, potentially catastrophic, pollution incidents, such as may be associated with the collapse of a tailings dam. The EA will need to determine when the pollution occurred, as well as its source and magnitude. In such incidents, the release of contaminants may well have ceased by the time the EA is able to react to the incident and start its investigations. For this reason, whilst filtered water or suspended particulate matter samples will be effective for the continuous or periodic monitoring of aquatic environments that are being subject to relatively continuous contamination by effluent from mineral processing activities, these media would be ineffective sample media for sporadic pollution incidents. To assist with its function of pollution monitoring, the EA would require baseline data for the full range of sample media used in the case studies (i.e. filtered water, suspended particulate matter and especially bottom sediments). The results of this study have shown that water and suspended particulate matter samples indicate the current flux of contamination to the drainage system but are very susceptible to temporal variations related to short term fluctuations in discharges from processing plants and to rainfall-related dilution effects. Bottom sediment provides a more stable indication of the extent and magnitude of contamination whereas heavy mineral concentrate samples indicate the amount of particulate Hg metal and Au-Hg amalgam in the river sediment. Mercury concentrations in bottom sediments indicate the likely hazard to biota from remobilization of Hg as a result of methylation processes. Mercury in filtered water and suspended particulate matter provides a more relevant indicator of contaminant fluxes at the time of sampling and hence the potential hazards to biota and humans via this exposure route. More detailed monitoring is required to verify the level, frequency and duration of high contaminant fluxes in stream water and suspended particulate matter and to establish factors such as rates of sedimentary Hg methylation and subsequent bioassimilation. In near-shore marine environments, mercury data for sediments may be used to evaluate the impact of mercury contamination. In the Sitio Honda Bay case

54

study, for example, such data demonstrated quite clearly that contamination is strictly localised in the nearshore zone. Single pollution episodes involving non-persistent contaminants are difficult to detect and quantify in the absence of continuous monitoring data. This is a very common problem with respect to cyanide, for example, and nitrate residues have been used in the USA to infer contamination of the aquatic environment. Periodic releases of contaminants in solution or as suspended matter may be detectable only for a relatively short period until the contaminant flux passes through the drainage system. Bottom or interfacial sediments are often the best sample media for non-biological sampling. However, biomarkers, such as those used in the Palawan case study, are generally more effective because stress induced by short-term exposure is detectable long after the ambient contaminant level has declined. This is particularly true where species death is involved. From doseresponse data it is possible to tentatively infer exposure magnitude and/or time. Mineralogical and chemical characterisation of mine and mineral processing waste piles, tailings ponds, and acid mine drainage is a valuable method for assessing the potential hazards associated with these contaminant sources. The Palawan case study demonstrated that very high mercury concentrations (in the range 31 to 339 mgkg) found in the Sitio Honda Bay waste substrate are held predominantly in non-sulphide alteration products of probable low bioavailability. The contribution to total human exposure among the Sitio Honda Bay population (through hand to mouth ingestion and particulate or volatile phase respiration) is therefore likely to be modest. Water and sediment quality criteria

As a precursor to direct assessment of human and ecotoxicological impact it is conventional to evaluate potential toxicological hazards to aquatic biota and humans of contaminants in water and sediment by comparison with sediment and water quality thresholds for human health, drinking water, aquatic biota and shellfish. Monitoring data demonstrated that Hg in filtered (<0.45pm) river water exceeds the WHO Drinking Water Guideline value and the USEPA Water Quality Criteria for the Protection of Aquatic Life downstream of the Diwalwal mining district, where it is a potential hazard to the local people who use the Mamunga River as a source of drinking water. Water quality criteria for other potentially harmful elements are exceeded by a wide margin in the Ponce Enriquez (As, Cu, Zn)district, although only dissolved Cu exceeds the criteria for Protection of Aquatic Life in samples taken immediately upstream of the commercial shrimp ponds. The Hg Sediment Quality Toxic Effect Threshold for the Protection of Aquatic Life is exceeded in a 20 km section of the Mamunga River, below Diwalwal, where progressive release of Hg through methylation may pose a long term hazard to aquatic biota. Arsenic and Cu in bottom sediment collected 2 km upstream of the shrimp farms exceed Sediment Quality Criteria (Toxic Effects Threshold) for the Protection of Aquatic Life by factors of about 400 and 30 respectively. This study was unable to collect samples from the banana plantations and shrimp ponds. Until such data become available, it is recommended that care should be taken to ensure that contaminated sediment and water do not enter the banana plantations and shrimp ponds. In any aquatic system subject to inorganic As, Cu and Hg contamination, the attendant toxicological risk (with respect to magnitude and temporal duration) cannot readily be determined from data depicting total sedimentary loads. Additional geochemical and ecotoxicological studies are required to establish factors such as the rates of sedimentary Hg methylation and subsequent bioassimilation. With regard to human exposure, detailed epidemiological information, and body burden data would be critical precursors to any meaningful assessment. Ecotoxicologicaland human impacts In two of the areas investigated, namely Sitio Honda Bay, Palawan and Apokom, Mindanao, mercury pollution

had become an emotive issue following widespread media coverage of human mercury poisoning. The results of this project have been instrumental in providing a balanced input to the debate and have materially affected how and where people live. Tissue Hg concentrations of Honda Bay fish are less than the US-EPA marketable threshold. Thus, earlier media reports of ‘Minamata range’ Hg concentrations in fish from Honda Bay remain essentially unsubstantiated. Hair Hg concentrations within the Sitio Honda Bay population are similar to those reported for the neighbouring fishing community of Tagburos confirming that residential exposure to the Sitio Honda Bay community arising from the Hg-rich substrate must be negligible relative to alternative pathways. Over 98% of the 130 people investigated in the Palawan case-study yielded hair Hg concentrations considered to be a pre-

55

dictable function of diet. Theses data do not provide clear evidence of abnormal food chain contamination by mercury andor human exposure via additional pathways, such as hand-to-mouth ingestion of contaminated soil. Hair Hg data indicate that ballmill operators and workers at the cyanidation plants processing Hg contaminated tailings at eastern Mindanao’s principal gold beneficiation centre, Apokon, may be subject to enhanced occupational Hg exposure. However, even these workers have Hg hair concentrations which fall within the range found in fishing communities where there is no exposure to Hg derived from gold processing sources. Hair Hg data indicate that the wider population of the Apokon area has not been impacted significantly by Hg contamination related to gold beneficiation. With regard to human exposure, detailed epidemiological information, and body burden data would be critical precursors to a full impact assessment. Such studies are now required in the Ponce Enriquez, Nambija and Diwalwal areas. Alternative gold-recovery methods designed to reduce mercury contamination

The results of the laboratory investigations demonstrated that a relatively high proportion of gold can be recovered using simple gravity methods, although the gold content of the products was generally low (less than 1%). A demonstration model of a simple, hand-powered, shaking table was designed and constructed by the BGS of cheap and simple materials that are available in developing countries. It was particularly aimed at the recovery of fine-grained gold. Laboratory trials showed that the BGS shaking table was as good as and probably slightly more effective for the separation of fine-grained gold than the commercial Wilfley laboratory-scale shaking table. The BGS shaking table was taken to the Philippines for field trials at which it was demonstrated to be remarkably efficient at recovering fine-grained gold when properly set-up. The table recovered a significant quantity of gold from miners tailings where the grain size was only around 30pm. The field trials showed how important and sometimes difficult it can be to adjust the table to the correct operational settings. The small-scale miners were impressed by the performance of the table and were keen to test it further and possibly produce a larger version for regular use as an alternative to cyanidation. The findings of this study have reinforced the value of characterisation prior to mineral processing. Mineralogy, particle-size and texture all have an influence on the performance of mineral processing trials. For example, the proportion of gold finer than 50 pm in an ore has a significant effect on the proportion of gold recoverable using simple gravity methods. Therefore, a determination of the particle-size distribution of gold could be used to indicate the likely success of using gravity methods for the recovery of gold from a given ore. It is assumed that most artisanal gold miners have little knowledge of the particle size distribution of the gold present in their ore. Therefore, the most practical method of establishing the effectiveness of the shaking table for the recovery of gold from a particular ore would be to perform trials. A simple shaking table can be very effective for the recovery of fine-grained gold down to a grain-size of around 20pm and hence could play an important role in providing an alternative to the use of mercury in the extraction of gold from bulk samples. The heavy mineral concentrates produced contain several weight percent gold but still require further processing to remove the remaining impurities. This might be possible with the shaking table but this has not yet been demonstrated and it is unlikely that an alternative will be found to mercury amalgamation for the final concentration of gold from gravity concentrates. Carefully controlled amalgamation followed by roasting to release the gold and condensation of the mercury vapour using a retort will still be required in most cases. The use of mercury could be restricted to this final stage only and its use in tromol mills, sluice beds and other process equipment could be significantly reduced or eliminated through the demonstration of the effectiveness of gravity separation techniques. The following changes to gold processing procedures used by small-scale miners would reduce the amount of mercury contamination: 1. Improved sluice box design and increased use of effective gravity separation methods (such as the shaking table) might reduce the use of mercury on sluices. This would be achieved by demonstration of the financial benefits of increased gold recuperation and hence the amount of income generated. 2. Development of a retort system which retains the Hg but produces a higher quality gold product uncontaminated by Fe and As (which reduces the value of the Au). It is not entirely clear whether artisanal miners do not use the retorts because (1) they are suspicious that they are loosing some of the gold in the retort, (2) the processing takes longer or (3) the product is lower quality and less marketable. If 80-90% of Hg loss is related to burning off the mercury then the biggest effort should clearly be put into improving this aspect 56

of the process. Numerous NGOs (such as Intermediate Technology) have tried this approach which in general appears to have been unsuccessful. In Ecuador, for example, small-scale gold processors frequently have retorts but they are never used. 3. Increasing the use of cyanidation in an environmentally acceptable way would reduce the need for amalgamation. This would require better control on the location of cyanidation plants and management procedures. 4. Perhaps the greatest potential for reducing environmental impact would be through the adoption of an effective system of co-operative production. This should lead to better management of mineral processing plants with reduced environmental impact while still leaving the means of production in the hands of the poorer sector of the population. In theory it should permit a more effective and economical production system to be adopted through upsizing production systems and providing access to capital for larger investments (e.g. shaking tables and efficiently run cyanidation plants). An alternative outcome in areas with major gold deposits currently being worked by artisanal miners but suitable for large scale exploitation (e.g. Mindanao) would be to let the mining rights to a major company, evict the artisanal miners and for the Government to use tax revenues for the benefit of the rural economy. Finally, coal-oil agglomeration is a novel process that might merit further evaluation. It is reported to have many advantages compared with conventional amalgamation and cyanidation procedures such as lower operational and capital costs and less environmental impact. (Marciano et al., 1994; Zhao et al, 1997). I4.TAKE-UP

The TDR project provided data of direct interest and concern to the Philippines Government and were used to inform the policy formulation process. In both Palawan and Mindanao the results of the project have materially affected how and where people live. More important, the project has been instrumental in providing a balanced input to emotive issues - namely the supposed poisoning of populations on Palawan and Mindanao. There is a vast array of press cuttings available that highlight the environmental problems associated with mercury use in the Philippines. The work in the Philippines has been acknowledged as one of BGS's major successes in the field of environmental geochemistry.

IS. ACKNOWLEDGEMENTS Logistic support and field guidance in the Philippines was provided by staff of the Philippines Department of Environment and Natural Resources (Mines and Geosciences Bureau), particularly Tony Apostol, Resty Gomez, Juliet Miguel, and Carlos Miranda. In Mindanao, the mine and processing-plant owners and operators for permission to examine the mining operations and carry out the sampling, and the local people who acted as guides to some of the more obscure areas. In Palawan, University of the Philippines researchers Ms. Grace Doming0 and Mr Mike Reyes participated enthusiastically in all aspects of the Palawan field programme and the additional logistic support was provided by staff of the Puerto Princesa Hospital,. Discussions with Dr G.S. Jacinto (University of the Philippines) and Dr A. Socrates (Provincial Health Officer, Palawan) were invaluable throughout both the design and execution of the case study. In Ecuador, the collaboration, advice and logistic support provided by Ing. Hug0 Orbea of CODIGEM (Corporacidn de Desarrollo e Investigacidn Geoldgico-Minero Metalu'rgica), Ing. Milton Carrasco of DINAPA (Departamento Nacional de Proteccidn Ambiental), Ing. Roque Maldonado of DINAMI (Direccidn Nacional de Mineria), Swedish Mission consultants, Reider Hoffner, Goran Fredrikson and Bo Erikson (PRODEMINCA Subcomponents 3.1 and 3.2) and Dr John Aspden and all other members of the resident Mision Britanica team in Quito is gratefully acknowledged. The assistance and cooperation of Guido Villegran (CODIGEM driver) during the Ponce Enriquez field campaign is gratefully acknowledged. Barbara Vickers (BGS) was responsible for the CV-AFS Hg determinations, and Mark Cave and Linda Ault (BGS) for the ICP-AES analyses. Clive Mitchell, Mike Styles and Ellie Evans (BGS) carried out all the gold mineral processing laboratory and field investigations. Jason Weekes of the Institute of Terrestrial Ecology (ITE)carried out the ecotoxicological studies in Palawan. 16. REFERENCES APPLETON, J.D. CARASCO, M., ORBEA,H. AND MALDONADO, R. 1996. Assessment of mercury contamination in the Ponce Enriquez artisanal gold mining area, Ecuador. British Geological Survey Overseas Geology Series report WC/96/55. BGS, Keyworth, Nottingham, UK.

57

APPLETON, J.D., WILLIAMS T M, BREWARD, N, APOSTOL,T, MIGUEL,J. AND MIRANDA,c. 1999a. Mercury contamination associated with artisanal gold mining on the island of Mindanao, the Philippines. Science of the Total Environment. (in press). APPLETON, J.D. WILLIAMS TM, ORBEAH, CARRASCO, M., AND MALDONADO, R. 1999b. Fluvial contamination associated with artisanal gold mining in the Ponce Enriquez, Portovelo-Zaruma and Nambija areas, Ecuador. Water, Air and Soil Pollution. (submitted for publication) BENOIT,G., SCHWANTES, J.M., JACINTO, G.S. AND GOUD-COLLINS, M.R. 1994: Preliminary study of the redistribution and transformation of HgS from cinnabar mine tailings deposited in Honda Bay, Palawan, Philippines. Marine Pollution Bulletinl. 12,754-759. CALLAHAN J E, MILLER,J W AND CRAIGJ R, 1994. Mercury pollution as a result of gold extraction in North Carolina, USA. Applied Geochemistry, Vol. 9, No. 2,235-241. CAMARA V DE M, FILHOTE,MI DE F, LIMAMIM, ALHEIRAFV,MARTINS MS, DANTASTO AND Lurz RR. Strategies for preventing adolescent mercury exposure in Brazilian gold mining areas. Toxicology and Industrial Health, Vol. 13, NOS. 213, 285-297. CLARKSON T.W. 1975: Exposure to methyl Hg in Grassy Narrows and White Dog reserves. Interim-report., US-EPA. CLEARY D AND THORNTON I. 1994. The environmental impact of gold mining in the Brazilian Amazon. In: Hester RE and Harrison RM. Mining and environmental impact. Issues in Environmental Science and Technology NO. 1, Cambridge, The Royal Society of Chemistry. 17-29. CRAIG,P.J. 1982. Environmental aspects of organometallic chemistry. In Wilkinson, G., Stone, F.G.A. and Abel., E.W. (Eds) Comprehensive Organometallic Chemistry. 2.

DESCHENES, G (1986) Literature survey on the recovery of gold from thiourea solutions and the comparison with cyanidation. CIM Bulletin, 79, No. 895, p. 76 - 83. ECHAVARRIA C. 1996. Strategies and policies for sustainable small-scale mining in Colombia. Mining & Environmental Research Network Working Paper No. 76, University of Bath, 30 pp. ELTHAM, JA (1984) Mining and processing of a low-grade gold ore body in the Wau Valley, Papua New Guinea. Proceedings of AusIMM conference, Darwin, Australasian Institute of Mining & Metallurgy, p. 427-434. Eve E, Oliveira EF and Eve C. 1996. The mercury problem and diets in the Brazilian Amazon: planning a solution. Environmental Conservation 23 (2) 133-139.

,J E. 1990. The heavy elements: chemistry, environmental impact and health effects. (Oxford: Pergamon) FERGUSSON GACETA OFICIAL DE BOLIVIA 1997: Reglamento ambiental para actividades mineras. Decreto Supremo No. 24782. Gaceta Oficial de Bolivia, La Paz, 47 pp.

S AND BIDOVEC M, 1997. Mercury in the Idrijca River sediments as a reflection of mining and smelting activities of the Idrija mercury mine. Journal of Geochemical Exploration 58: 125-131.

GOSAR M, PIRC

GREENWOOD, N.N. AND EARNSHAW, A: 1984: Chemistry of the elements. Pergamon Press. HANAI M.. 1996. Formal and ‘Garimpo’ gold mining and the environment in Brazil. Mining & Environmental Research Network Working Paper No. 88, University of Bath, 15 pp.

HEALTHAND WELFARE OF CANADA.1979: Task force on organic mercury in the environment. Grassy Narrows and White Dog reserves, Ontario. Dept. National Health and Welfare, Ottawa, Canada. HOLLOWAY J. 1996. Lessons from Zimbabwe for best practice for small and medium scale mines. Mining & Environmental Research Network Working Paper No. 62, University of Bath, 30 pp. IRPTC 1980 JAMESLP. 1994. The mercury tromol mill: an innovative gold recovery technique and a possible environmental concern. Journal of Geochemical Exploration 50: 493-500. KERSHAW, T.G., CLARKSON, T.W. AND DHAHIR, P.H. 1980: The relationship between blood levels and dose of methylmercury in man. Arch. Environ. Health, 35, 28-36.

LACERDA LD AND SALOMONS, W. 1998. Mercury from gold and silver mining: a chemical time bomb ? (Berlin:Springer), 146 pp. LANDNER, L. 1991. Environmental impact and measures for environmental control: Study N0.2, Vol. 1. Preparatory Studies for PRODEMINCA (unpublished).

58

LECHLER PJ, MILLERJR, Hsu L-C AND DESILETSMO, 1997. Mercury mobility at the Carson River Superfund Site, westcentral Nevada, USA: interpretation of mercury speciation data in mill tailings, soils and sediments. Journal of Geochemical Exploration, 58: 259-268. R.A. 1994. The metamorphic belts of Ecuador. British Geological Survey, LITHERLAND, M. ASPDEN,J.A. AND JEMIELITA, Overseas Memoir No. 1 1. BGS, Keyworth, Nottingham, UK. MACDONALD, D D. 1994. A review of environmental quality criteria and guidelines f o r priority substances in the Fraser River Basin. DOE FRAP 1994-30. MALDONADO, R.C.:1996. AvaliapTo geoquimica da contaminapio por merclirio dos sedimentos de coreintes e solos do distrito mineiro de Zaruma-Port0 Velho, Repliblica do Ecuador. M.Sc. thesis, Centro de GeociCncias, Universidade Federal do Par& Belem, Brasil. M m 0, P F E ~WC, R SOUZA CMM, REUTHERA. 1990. Mercury pollution due to gold mining in the Madeira River basin, Brazil.Ambio 194: 11-15. MARCIANO A, COSTALSN AND LINSFF. 1994. Utilization of coal-oil agglomerates to recover gold particles. Minerals Engineering, Vol. 7, No. 11, 1401-1409. MARSDEN, JO & FUERSTENAU, MC (1993) Comparison of Merrill Crowe precipitation and carbon absorption for precious metal recovery. Proceedings of XVIZZ International Mineral Processing Congress, Volume 5: Gold processing, hydrometallurgy & dewatering and miscellaneous, Australasian Institute of Mining and Metallurgy, p. 1189-1194. MT 1997a. A review of gold particle-size and recovery methods. British Geological MITCHELL, CJ, EVANS,EJ & STYLES, Survey Technical Report WCl97114. MKHELLCJ, EVANSEJ AND STYLESMT. 1997b. The design, construction and testing of a simple shaking table f o r gold recovery: laboratory testing and3eld trials. British Geological Survey Overseas Geology Series Technical Report WC/97/61. MOREIRA JC AND PIVE'ITAF, 1997. Human and environmental contamination by mercury from industrial uses in Brazil. Water, Air and Soil Pollution 97: 241-246. H & WARE,AR (1991) The geology and mineralisation of the Ladolam gold deposit, MOYLE,AJ, DOYLE,BJ, HOOGVLIET, Lihir island, Papua New Guinea. Proceedings of PNG geology, exploration and mining conference, Rabaul, Australasian Institute of Mining and Metallurgy, p.101-111. NAVARRO, M. 1986: Investigacih histbrica de la mineria en e1 Ecuador. Quito Instituto Ecuatoriana de Mineria. NEW, J.M.: 1997, Ecotoxicology of arsenic in the marine environment. Environmental Toxicology and Chemistry 16 (3, 9 17-927.

NRA, 1994. Implementation of the EC Shelljish Waters Directive. NRA Water Quality Series No. 16. PACCAGNEU, B. AND PRATTI,L: 1974: Total mercury in blood and hair of Italian people. Zg. Mod. 67,369-30. PACCAGNEU, B., PRATI,L. AND BIGONI,A. 1973: Studio epidemiologico sur mercurio nei pesci e la salute unama in un isola Italiana del Metiterraneo. Ig. Mod. 66,479-503. PESTANAMHD, FORMOSO MLL AND TEIXEIRA EC, 1997. Heavy metals in stream sediments from copper and gold mining areas in southern Brazil. Journal of Geochemical Exploration, 58: 133-144. PFEIFFER, W C, LACERDA D L, MALM,0, SOUZA C M M, DE SILVAE G AND BASTOS W R. 1989. Mercury concentrations in inland waters of gold mining areas in Rondonia, Brazil. Science of the Total Environment, Vol. 78, 233-240.

0, SOUZA C M M, LACERDA D L, DE SILVA E G AND BASTOSW R. 1991. Mercury in the Madeira River ecosystem, Rondonia, Brazil. Forest Ecology and Management, Vol. 38,239-245.

PFEIFFER, W C, MAN,

PIOTROWSKI, J.K. AND INSKIP,M.J. 1981: Health efsects of methylmercury. MARC, University of London. 82 pp. PRODEMINCA. 1998 Reglamento para actividades mineras en la Rep6blica del Ecuador. PRODEMINCA, Quito, 91pp. REUTHER R, 1994. Mercury accumulation in sediment and fish from rivers affected by alluvial gold mining in the Brazilian Madeira River Basin, Amazon. Environmental Monitoring and Assessment, Vol. 32, No., 3, 239-258.

RIODOLFI. M. 1977: Further epidemiological study of Hg levels in fish and human blood and hair. Zg. Mod. 70, 169-186. RODRICUES S AND MADDOCK JEL, 1993. A survey of heavy metal distribution in sediments and water from the gold prospecting region of PoconC (MT), Brazil. In: Abriio JJ, Wassennan JC, Silva-Filho EV (eds). Proceedings International Symposium Perspectives f o r Environmental Geochemistry in Tropical Countries, pp 46 1-466. RODRIGUES S AND MADDOCK JEL, 1997. Mercury pollution in two gold mining areas of the Brazilian Amazon. Journal of Geochemical Exploration, 58: 23 1-240.

59

SCOV, A. 1998. The environmental impact of small-scale industries in the Third World. Global Environmental Change Programme Briefing No. 19. ESRC Global Environmental Change Programme Office, University of Sussex, 4 pp. SIMPSON,PR, EDMUNDS,WM,BREWARD,N, COOK JM, FLIGHTD, HALL GEM AND LISTERTR. 1993. Geochemical mapping of stream water for environmental studies and mineral exploration in the UK. Journal of Geochemical Exploration, 49, 63-8. SUBASINGHE, GKN AND MARU,Y (1994) The use of mercury in small-scale gold mining operations: effectiveness and pollution aspects. In: Rogerson R (ed.) Proceedings of PNG geology, exploration and mining conference, Lae, Papua New Guinea, Australasian Institute of Mining and Metallurgy, p.264-270. TORRESE B, 1992. Environmental and health surveillance of mercury use in small scale gold processing industries in the Philippines. In: Proceedings of International Symposium Epidemiological Studies of Environmental Pollution and Health Effects of Methylmercury, National Institute for Minamata Disease, Kumamoto, pp. 56-65. TORRES,E B, 1994. Epidemiological investigation of mercury exposure and health effects in the Philippines. In: Proceedings of International Workshop on Environmental Mercury Pollution in the Amazon River Basin, National Institute for Minamata Disease, Rio de Janeiro, pp. 3-9. TUMPLING WV, WILKENRD AND EINAXJ. 1995. Mercury contamination in the northern Pantanal region, Mat0 Grosso, Brazil. Journal of Geochemical Exploration, 52: 127-134. VEIGA MM, MEECHJA AND HYPOLITOR. 1995a. Educational measures to address mercury pollution from gold-mining activities in the Amazon. Ambio, Vol. 24, No. 4,216-220. VElGA MM, AND MEECHJA . 1995b. Gold mining activities in the Amazon: clean-up techniques and remdial procedures for mercury pollution. Ambio, Vol. 24, No. 6, 371-375. WARHURST A AND THOMAS R. 1998. A p r e h h a r y review of the literature regarding the environmental and social issues as they affect small-scale mining. Mining & Environmental Research Network Working Paper No. 118, University of Bath, 16 PP. WEDEPOHL, 1978. Handbook of Geochemistry. (Berlin-Heidelberg:Springer-Verlag). WEEKS, J. AND WlLLIAMS, T.M: 1995: A simple ecotoxicological field test to determine mining-related heavy toxic trace element stress. Ecotoxicology-(in press).

H AND EDIGHAUS R. 1990. Biologische Quecksilberumsetzung in der Elbe. Wasser, 74: 383WILKEN RD, HINTELMANN 392. WILLIAMS, TM, 1997 Visit to the Philippines and Thailand. British Geological Survey Technical Report WPl97l1. WILLIAMS, T M, APOSTOL,A, AND MIGUELJ. 1995. Mercury contamination in artisanal gold mining areas of eastern Mindanao, Philippines: a preliminary assessment. British Geological Survey Technical Report WCl95172lR C. 1996a. Assessment of mercury toxicity hazard associated WILLIAMS, T M, WEEKS, J M, APOSTOL,A, AND MIRANDA, with former cinnabar mining and tailings disposal in Honda Bay, Palawan, Philippines. British Geological Survey Technical Report WCl96131lR WILLIAMS T M, BREWARD, N, APOSTOL, T AND MIGUEL,J. 1 9 9 6 ~Assessment . of mercury contamination associated with contrasting mining activities on the islands of Mindanao and Palawan, Philippines. Proceedings of the 2nd International Symposium on the Environmental Geochemistry of Tropical Countries, Cartagena, Colombia, November, 1996. WlLLIAMS, T.M. AND ORBEA,H.: 1997, Dispersal of mercury and ore-derived metals in surface drainage of the Nambija artisanal miningfield, Ecuador. British Geological Survey Technical Report WCl97/37R, 21 pp. WORLDBANK1994, Development, Environment and Mining Enhancing the Contribution of the Mineral Industry to Sustainable Development International Conference on Development, Environment and Mining,Washington, DC,June 1-3, 1994 WORLDBANK,1996. Proceedings of the International Roundtable on Artisanal Mining Organised by the World Bank, Washington, D.C. May 17-19, 1995 Industry and Energy Department Occasional Paper No. 6 Mamadou Barry, editor April 1996 WORLDHEALTH ORGANISATION 1976: Environmental Health Criteria Document I : Mercury. ZHAO B, LU L AND XIE H. 1997. Transactions Nonferrous Met. Soc. China. Vol. 7, No. 4, 152-155.

60

APPENDIX I :PROJECTTECHNICAL REPORTS

APPLETON,J.D. CARRASCO, M., ORBEA,H. AND MALDONADO, R. 1996. Assessment of mercury contamination in the Ponce Enriquez artisanal gold mining area, Ecuador. British Geological Survey Overseas Geology Series Technical Report WC/96/55. APPLETON JD. 1998. Report on the dissemination workshop: Taller sobre geoquimica y medio ambiente con referencia especial a la contaminacidn minera. 6-7Mayo de 1998, Quito, Ecuador. BGS Technical Report Overseas Geology Series, WC/98132. APPLETON JD. 1998. Report on the dissemination workshop: Taller sobre geoquimica y medio ambiente con referencia especial a la contaminacidn minera. 5 de Octubre de 1998, La Paz, Bolivia. BGS Technical Report Overseas Geology Series, WC/98/57. BREWARD N, APOSTOLA, APPLETON JD, GOMEZR, MIGUEL J. 1996. Mercury and other heavy-metal contamination associated with gold mining in the Agusan river catchment, Mindanao, the Philippines. British Geological Survey Overseas Geology Series Technical Report WC/96/01 (2nd edition) MITCHELL CJ, EVANSEJ AND STYLES MT. 1997. A review of gold particle-size and recovery methods. British Geological Survey Overseas Geology Series Technical Report WC/97/14. MITCHELL CJ, EVANS EJ AND STYLES MT. 1997. The design, construction and testing of a simple shaking table for gold recovery: laboratory testing andfield trials. British Geological Survey Overseas Geology Series Technical Report WC/97/6 1. WILLIAMS, TM, 1995. Visit to Thailand and the Philippines. British Geological Survey Overseas Geology Series Technical Report WP/95/15. WILLIAMS, TM, 1997. Visit to the Philippines and Thailand. British Geological Survey Overseas Geology Series Technical Report WP/97/1. W W S , T M, APOSTOL,A, AND MIGUELJ. 1995. Mercury contamination in artisanal gold mining areas of eastern Mindanao, Philippines: a preliminary assessment. British Geological Survey Overseas Geology Series Technical Report wc/95/72m. C. 1996. Assessment of mercury toxicity hazard associated WILLIAMS, T M, WEEKS,J M, APOSTOL,A, AND MIRANDA, with former cinnabar mining and tailings disposal in Honda Bay, Palawan, Philippines. British Geological Survey Overseas Geology Series Technical Report WC/96/3 lm WULIAMS, T.M. AND ORBEA, H.: 1997, Dispersal of mercury and ore-derived metals in surface drainage of the Nambija artisanal mining field, Ecuador. British Geological Survey Overseas Geology Series Technical Report WC/97/37R.

APPENDIX 2: OTHER PROJECTPUBLICATIONS APPLETON, J.D., WUIAMs T M, BREWARD, N, APOSTOL,T, MIGUEL,J. AND MIRANDA, c. 1999. Mercury contamination associated with artisanal gold mining on the island of Mindanao, the Philippines. Science of the Total Environment. (in press). M., AND MALDONADO, R. (submitted) Fluvial contamination associAPPLETON, J.D. WILLWS TM, ORBEAH, CARRASCO, ated with artisanal gold mining in the Ponce Enriquez, Portovelo-Zaruma and Nambija areas, Ecuador. Water, Air and Soil Pollution (in press). WILLIAMS T M, BREWARD, N, APOSTOL,T AND MIGUEL,J. 1 9 9 6 ~Assessment . of mercury contamination associated with contrasting mining activities on the islands of Mindanao and Palawan, Phillipines. Proceedings of the 2nd International Symposium on the Environmental Geochemistry of Tropical Countries, Cartagena, Colombia, November, 1996. WILLIAMS TM, APPLETON JD AND ORBEA 0. 1998. Assessment of mercury contamination associated with artisanal gold mining in the Nambija and Ponce Em’quez areas of Ecuador. Proceedings of the First Latin American Conference on Contaminated Soil and Water, Quito, May 1997, p.26-33. WILLmS T M, WEEKS, JM, APOSTOL, AND MIRANDA, c. 1999. Assessment of mercury contamination and human exposure associated with coastal disposal of waste from a cinnabar mining operation, Honda Bay, Palawan, Philippines. Environmental Geology (in press).

61

62