Mine Water Management
Strategies and Solutions

At Rum Jungle in Australia's Northern Territory, the rehabilitation bill exceeded the total revenue the mine ever generated. These are not rare disasters. They are the ordinary outcome of a system that treats mine water as a problem to be addressed later, by someone else, with money that has not been set aside.

The technical literature on mine water management is enormous. The INAP Global Acid Rock Drainage Guide runs to hundreds of pages. Canada's MEND program generated decades of field research. None of this knowledge is hidden. The question that matters is why so much of it fails to influence decisions at the point where decisions are made, which is during feasibility and early operations, by people whose jobs depend on production metrics.

The groundwater model at a mine site is supposed to predict where water will come from, how much, and what it will carry. The model gets built during feasibility by a consulting firm working from whatever borehole data and pump test results the project has generated. The budget for this work is typically small relative to the resource drilling program. A large open pit feasibility study might spend $20 million on resource definition drilling and under $2 million on hydrogeological characterization. The resource gets a geostatistical model with confidence classifications. The groundwater gets a numerical model built in MODFLOW or FEFLOW, calibrated to a few pump tests, and submitted as part of the environmental impact assessment.

The calibration problem is worth dwelling on because it is central to why models fail. A numerical groundwater model has adjustable parameters: hydraulic conductivity in each zone, recharge rates, boundary conditions. The modeler adjusts these until the model reproduces observed water levels and pump test drawdowns. With enough zones and enough parameters, almost any model can be made to match the calibration targets. This says nothing about whether the model will predict correctly when the mine goes deeper, when a new fault zone is encountered, when the pit intersects a different hydrostratigraphic unit. The model matched the data it was built to match. It has not been tested against conditions it was not built to match, because those conditions do not exist yet. They will exist in year five or year ten of mining, when the investment is committed and the pit geometry cannot be changed.

Consultants know this. The good ones include uncertainty analysis, sensitivity runs, and explicit statements about what the model cannot predict. These caveats appear in the appendices of the hydrogeological report. The executive summary, which is what decision-makers read, presents the base case. The base case becomes the plan.

Structural geology determines whether a given section of pit wall will produce ten liters per second or a thousand. Clay gouge in a fault means low permeability. Calcite fill or open voids in the same fault can transmit enormous volumes. The difference between these conditions cannot be determined from surface mapping or geophysics with any reliability. It requires direct observation of the fault plane, which means underground exposure or oriented core drilling with detailed logging. This kind of work is expensive and slow. It competes for budget with resource drilling, which has a direct and visible connection to project economics. The hydrogeological field program loses this competition at most sites.

I keep coming back to the Mufulira inrush because it is the starkest example of the consequence. In 1970, 89 miners died when workings in the Zambian Copperbelt breached a water-bearing zone. The geological structure was known to exist in a general sense. Its capacity to deliver a catastrophic inflow was not established before the workings reached it. The investigation reports are still taught in mining hydrogeology courses. The lesson that site characterization budgets should reflect the consequence of getting it wrong has been absorbed unevenly.

Vertical hydraulic conductivity is something I want to spend more time on than most treatments of this subject do, because it connects to a failure mode that is common and poorly recognized. Pump tests measure horizontal transmissivity well. Vertical leakage between aquifer layers, which controls whether dewatering one horizon will drain the one above it or below it, is almost never adequately measured. The typical approach is to assign a vertical-to-horizontal conductivity ratio of 0.1 or 0.01 based on literature values for the rock type. In a real geological sequence, a two-meter siltstone interbed can reduce vertical conductivity by four orders of magnitude. A set of subvertical fractures cutting through that siltstone can restore connectivity almost entirely. The spatial variation of vertical conductivity across a mine site is large, site-specific, and unmapped at nearly every operation. Dewatering models that treat vertical conductivity as a uniform ratio produce results that match early monitoring data (because the early data reflect conditions close to the calibration boreholes) and diverge when the mine develops into areas where the vertical connectivity is different.

This is a topic that gets almost no space in the mine water literature and deserves more than it gets, because it connects dewatering, slope stability, and drilling-and-blasting practice in a way that most mine organizational structures cannot handle.

Every production blast that breaks rock at the pit wall generates a damage halo behind the excavation surface. The halo extends one to three meters into the rock mass in hard rock operations. Within the halo, existing fractures are opened and new fractures are created. The permeability of this zone exceeds intact rock permeability by one to three orders of magnitude. That damaged zone then becomes the primary pathway for water to enter the pit and for pore pressure changes to propagate into the slope.

Controlled blasting techniques, pre-splitting, and trim blasting reduce the damage halo. They produce a cleaner wall with less fracture connectivity. They also cost more per cubic meter of rock broken, use more drill holes per meter of face, and slow the rate of advance. The cost difference shows up in the drill-and-blast budget, which belongs to the mining department. The savings show up in the dewatering budget and the geotechnical risk profile, which belong to other departments.

The result is that most open pits use controlled blasting on final walls only, where the geotechnical team has enough influence to insist on it, and production blasting everywhere else. The cumulative permeability increase from years of production blasting on interim walls creates water management problems that the dewatering design did not anticipate because the design was based on intact rock properties.

I have not seen a published case study that quantifies the total cost of blast-induced permeability increase over the life of an open pit, accounting for incremental pumping energy, incremental treatment volume, and the geotechnical risk premium associated with elevated pore pressures in the blast damage zone. The individual components have been studied. Peter Stacey and John Read's book on open pit slope design addresses pore pressure management in detail. The International Society of Explosives Engineers' literature covers blast damage assessment. The integration of the two into a single cost-benefit framework that could change how mines allocate blast engineering budgets does not appear to exist in published form.

The geochemistry of pyrite oxidation has been laid out thoroughly by Nordstrom and Alpers, by Jambor et al. in their environmental mineralogy reviews, and by Blowes, Ptacek, Jambor, and Weisener in their contributions to the Treatise on Geochemistry. The reaction pathway is not in dispute. What continues to be underweighted in operational practice is the dominance of microbial catalysis.

The abiotic oxidation rate of ferrous iron to ferric iron at pH values below 3.5 is extremely slow. At pH 2, the abiotic half-life for this reaction is on the order of several hundred days. Acidithiobacillus ferrooxidans and Leptospirillum ferrooxidans catalyze the same reaction at rates five to six orders of magnitude faster. This is not a minor acceleration. It is the difference between a waste rock dump that generates nuisance-level acidity over decades and one that produces aggressively acidic drainage within years. Kirk Nordstrom's group measured negative pH values in the underground workings at Iron Mountain in California. That level of acid intensity is sustained by microbial iron cycling at rates that abiotic chemistry cannot approach.

Attempts to suppress the microbial population directly, through surfactant application or organic acid treatments that disrupt bacterial cell membranes, have been tested in laboratory columns and in some field pilot studies. The results in controlled conditions are encouraging. Sodium dodecyl sulfate applied to waste rock columns at concentrations in the tens of milligrams per liter reduced iron-oxidizing bacterial populations and acid generation rates substantially. The gap between laboratory results and commercial adoption is wide, and it is wide for a specific reason: nobody has figured out who pays for field-scale application on a waste rock dump that will need retreatment at intervals for decades. The mine operator has a finite interest in the site. The government agencies responsible for long-term management have limited budgets. The surfactant manufacturer has no mechanism for recurring revenue from a product applied to an abandoned waste pile.

The lag time between sulfide exposure and the onset of acidic drainage is where planning most often goes wrong. Wheal Jane illustrates the timeline. Pumping at this Cornish tin mine ceased in 1991. Fourteen months later, in January 1992, a massive discharge of acidic water carrying zinc, cadmium, iron, and arsenic reached the Carnon River and turned the Fal Estuary orange-brown. The discharge chemistry was worse than prior assessments had predicted, and the volume was higher. The carbonate minerals in the surrounding geology had buffered the acidity during the mine's operating life and for some months after closure. When the buffering capacity was locally exhausted along the primary flow paths, the transition from neutral to acidic drainage happened over a period of weeks.

The prediction tool for this kind of lag behavior is kinetic testing, typically humidity cell tests run according to ASTM D5744. The standard test duration is 20 weeks, though some practitioners run cells for 40 or more weeks. Kevin Morin, through his MDAG consulting practice and publications, has been one of the most persistent voices pointing out the limitations of short-duration kinetic tests for predicting long-term drainage chemistry. A 20-week humidity cell is being asked to predict behavior over centuries. The minerals that control buffering and acid generation are present in finite quantities. A 20-week test may not exhaust the fast-reacting carbonate minerals, much less provide data on what happens after they are gone. Extending tests to two or three years helps, and costs more, and does not solve the fundamental problem of extrapolation across timescales that differ by orders of magnitude.

The comparison with resource estimation standards keeps coming up in professional discussions and it keeps being uncomfortable. JORC requires that a Competent Person classify resources based on geological confidence, data density, and estimation methodology. The Competent Person sign-off carries legal weight. Geochemical predictions that will determine environmental outcomes for centuries have no equivalent standard. The same mining company that would never publish an Inferred resource as Measured will publish a 200-year drainage chemistry prediction based on 20 weeks of humidity cell data and a reactive transport model calibrated to those 20 weeks, and submit it to regulators as the basis for closure planning and financial assurance calculation.

I want to spend disproportionate space on the Elk Valley situation in British Columbia because it illustrates several things at once: how neutral drainage evades pH-centric regulatory frameworks, how treatment technology development lags behind problem recognition, and how the gap between permitted discharge and ecological effect can persist for years while regulatory and corporate systems negotiate a response.

Teck Resources operates five large open pit metallurgical coal mines in the Elk Valley of southeastern British Columbia. Coal mining in the valley generates waste rock and tailings with elevated selenium content. Selenium leaches from waste rock dumps and tailings impoundments as selenate, the fully oxidized form, at concentrations ranging from tens to hundreds of micrograms per liter. The discharge pH is unremarkable.

Selenate is a problem because of its behavior in aquatic food chains. It substitutes for sulfur in amino acid synthesis by organisms that cannot distinguish between the two elements. Selenium bioaccumulates from water through periphyton, through invertebrates, and into fish, with bioconcentration factors at each trophic step. The result is that waterborne selenium concentrations in the low single-digit micrograms per liter range produce tissue concentrations in fish eggs sufficient to cause embryo deformity and reproductive failure. The federal and provincial water quality guidelines for selenium have been revised downward repeatedly as the understanding of chronic toxicity has improved. The most recent Canadian Water Quality Guideline for selenium protection of aquatic life is 1 microgram per liter, a concentration that the Elk Valley discharges exceed by a large margin.

Teck's permitted discharge limits, set under provincial environmental assessment certificates and federal Fisheries Act authorizations, allowed selenium concentrations well above the levels at which ecological effects were being documented in monitoring programs. This gap between permitted discharge and ecological effect persisted for years. Environment Canada's monitoring programs and university researchers, particularly at the University of Lethbridge and the University of Saskatchewan, published data showing selenium accumulation in fish tissues and reproductive effects in westslope cutthroat trout populations in the Elk River and its tributaries.

Treatment for selenate is not simple. Lime neutralization, the standard mine water treatment approach, does not remove selenate. The charge and size of the selenate oxyanion allow it to pass through a lime treatment plant essentially unaffected. Effective selenate removal requires either biological reduction, in which bacteria convert selenate to elemental selenium under anaerobic conditions, or membrane separation, which removes selenate along with other dissolved constituents at the cost of producing a brine.

Teck invested in saturated rock fill bioreactors as the primary treatment technology. These are large engineered cells filled with gravel-sized rock media through which selenium-laden water flows under saturated, anaerobic conditions. An organic carbon source is added to sustain the microbial community responsible for selenate reduction. Pilot-scale results were promising. Scaling from pilot to the flow rates required to treat the combined discharge from five large coal operations proved more difficult and more expensive than the pilot data suggested. Temperature sensitivity is a factor; the Elk Valley has cold winters and the bioreactor kinetics slow substantially at low water temperatures.

By the mid-2010s, the British Columbia provincial government had issued a ministerial order requiring Teck to meet specific selenium and nitrate reduction targets on a defined timeline, with a designated Area-Based Management Plan. The company committed capital measured in billions of dollars. Construction of full-scale treatment facilities began. Meeting the targets has required a treatment infrastructure buildout with few precedents in the coal mining industry.

The Elk Valley situation is not unique in kind. Coal mines and some metal mines in other jurisdictions discharge selenate-bearing neutral drainage. Phosphate mining operations in Idaho produce selenium-laden drainage from waste rock and overburden. Coal mines in the Appalachian region generate selenium at lower concentrations that are still ecologically significant in small receiving streams. What makes the Elk Valley distinctive is the scale of the operation, the quality of the monitoring dataset, and the regulatory response. In jurisdictions with less monitoring infrastructure or less regulatory capacity, comparable selenium discharges go undocumented.

Lime treatment plants at closed mines run perpetually, or until someone decides the cost is no longer bearable and begins looking for alternatives. The Britannia mine south of Squamish, British Columbia, discharged acidic drainage into Howe Sound from the early twentieth century until a comprehensive remediation project was completed in 2005 at a cost of about $50 million, with funding from the provincial and federal governments. Before remediation, the province had been operating a water treatment plant at the site after the mining company became insolvent. The remediation project included a drainage collection tunnel, a water treatment plant, and source area management. The Howe Sound marine environment showed measurable recovery after the copper and zinc loading was reduced.

Faro is a different scale. The mine operated from 1969 to 1998, producing lead, zinc, and silver. Curragh Resources, the last operator, went bankrupt. The Yukon and Canadian governments assumed responsibility. The site has 70 million tonnes of tailings, 320 million tonnes of waste rock, and several open pits. Water management, including perpetual treatment of pit water and seepage, accounts for the largest share of the long-term cost estimate, which has been revised upward at every reassessment. The most recent public figures place the total liability above $1 billion. The mine generated roughly $500 million in revenue during its operating life, before accounting for the operating costs and profits distributed to shareholders. The remediation cost exceeds the lifetime revenue.

How perpetual liabilities get reduced to manageable-looking numbers in closure cost estimates involves discount rate selection. A 5% discount rate converts $2 million per year in perpetuity to a present value of $40 million. A 3% rate gives $67 million. The difference between these two numbers is larger than the capital cost of many passive treatment systems. The choice of discount rate is an assumption that is made by the financial analyst preparing the closure estimate, reviewed by the regulator assessing the closure bond, and reported in the company's annual report as a provision for rehabilitation. At no point in this chain does anyone with geological or geochemical expertise evaluate whether the discount rate appropriately reflects the physical persistence of the liability. The financial convention is to apply a positive discount rate. The physical reality is that the water will be acidic for centuries, and no financial instrument matures on that timescale. These two facts coexist in the same closure plan without being reconciled.

Robert Hedin, Robert Nairn, and George Watzlaf, working at what was then the U.S. Bureau of Mines and later through affiliated academic positions, built the empirical foundation for passive mine water treatment design in the Appalachian coalfields during the late 1980s and 1990s. Their sizing criteria for constructed wetlands, anoxic limestone drains, and successive alkalinity producing systems are still the starting point for most passive treatment designs worldwide.

The systems built during that era are now old enough to provide data on questions that could not be answered at the time of construction. How long does the organic substrate in an anaerobic wetland sustain sulfate-reducing bacteria? When does limestone armoring reduce alkalinity generation below the rate needed to neutralize incoming acidity? How much metal can wetland sediments accumulate before breakthrough occurs?

The answers, based on monitoring data from Pennsylvania, West Virginia, and Ohio systems now approaching 20 to 25 years of age, suggest that the term "passive" needs qualification. Organic substrate consumption rates vary with influent chemistry and climate, and substrates are showing depletion in systems that were expected to last longer. Limestone in anoxic limestone drains develops armoring coatings of iron and aluminum hydroxides that progressively reduce dissolution rates. Metal accumulation in wetland sediments has been tracked at some sites and shows a trajectory toward capacity limits, though the exact timeline depends on influent loading rates.

The Nickel Rim site near Sudbury, Ontario, has one of the longer-monitored organic carbon permeable reactive barriers in a mine water application. Installed in 1995, the barrier used municipal compost and leaf mulch as the organic carbon source in a trench across the contaminated groundwater plume. Monitoring by David Blowes' research group at the University of Waterloo tracked sulfate reduction performance and metal removal over an extended period. The barrier demonstrated sustained treatment capacity for well over a decade, providing field-scale validation of the concept. Performance changes observed over time gave empirical constraints on longevity that laboratory column tests could not have provided at the same confidence level.

This point is established in the literature and is not the interesting question. The interesting question is why, given that the point is established, source control remains underinvested at most mines.

The answer has to do with who decides and what they are measured on. The environmental manager who advocates for selective waste handling is asking the mining department to incur additional cost, additional truck movements, additional geological staff at the face, additional scheduling complexity, in exchange for a benefit that will materialize after the mine closes. The mining superintendent's bonus is tied to tonnes moved per shift. The environmental manager's advocacy, however well-supported technically, is asking the superintendent to accept lower performance against the metric that determines the superintendent's compensation, in exchange for a benefit that the superintendent will never see because the superintendent will be at a different mine by the time it matters.

This dynamic is not specific to mining. It occurs in any industry where the costs of preventive action and the benefits of preventive action fall on different people at different times. What makes mining distinctive is the timescale. The water quality consequences of waste rock management decisions made in 2025 may not become fully apparent until 2055 or 2075. No manager at the site in 2025 will be evaluated on the water quality in 2055. No mechanism exists within most mining company management systems to create that evaluation link.

The Equity Silver mine in northern British Columbia provides a case where source control worked. Reactive tailings were placed under a water cover at closure, maintaining oxygen exclusion conditions. Post-closure monitoring spanning more than two decades has confirmed effective suppression of acid generation. The water cover must be maintained through dam management and water balance control, so the solution is not zero cost. It is a manageable, predictable cost, fundamentally different from the open-ended treatment liabilities at sites where source control was not implemented.

Rum Jungle is the counter-case. The uranium-copper mine in the Northern Territory closed in 1971 with no source control. Waste rock dumps were left exposed to tropical wet-dry climate conditions that are close to ideal for pyrite oxidation: hot, with strongly seasonal rainfall that alternately wets and dries the dump surfaces. Acid drainage entered the East Branch of the Finniss River and persisted for over 40 years. A rehabilitation effort in the 1980s, funded by the Australian Government, installed a cover system that underperformed because it did not adequately limit oxygen diffusion. A second, far more comprehensive rehabilitation commenced in the 2010s after additional decades of monitoring and research. The second effort involves waste consolidation, multi-layer engineered covers designed using current best practice, and landform redesign to a geomorphically stable configuration. The total cost of the two rehabilitation programs exceeds the value of the uranium and copper the mine produced. Research conducted at Rum Jungle by groups at Charles Darwin University and through the Cooperative Research Centre for Contamination Assessment and Remediation of the Environment has contributed substantially to the understanding of cover system performance in tropical environments. The scientific value of this work is genuine. The cost of generating the field conditions that made the research necessary was borne by a river and the people who depended on it.

Selective waste handling works when it is sustained. The mines that sustain it have a common characteristic that is organizational rather than technical. Someone senior enough to push back against production pressure took ownership of the program and stayed at the site long enough to embed it. When that person left, the program either survived because it had become standard operating procedure or eroded because it had depended on individual persistence rather than institutional commitment. This is an observation, not a solution. Converting individual persistence into institutional commitment requires changes to management systems, incentive structures, and succession planning that most mining companies have not made.

The Mijnwater project in Heerlen, Netherlands, has been operating since 2008. It uses flooded coal mine workings beneath the city as a thermal source for heating and a thermal sink for cooling. Heat pump systems exchange energy with the mine water through wellbores and surface heat exchangers. The system provides climate control to over 500,000 square meters of buildings. The mine water temperature is stable year-round at approximately 28°C in the deeper workings and 16°C in the shallower workings, providing a temperature differential that heat pumps exploit efficiently.

The Coal Authority in England has been assessing mine water geothermal potential across former coalfield areas. Pilot installations exist at several sites. The Seaham Garden Village development in County Durham is planning to use mine water heat as a primary thermal source for a new residential community. The water in the former coal workings is already being pumped and treated by the Coal Authority to prevent uncontrolled discharge of ferruginous water to surface watercourses. Adding a heat exchange step to the existing pumping circuit requires incremental capital investment, not a new system.

The Berkeley Pit in Butte, Montana, has been accumulating water since the Anaconda Company ceased pit dewatering in 1982. The pit contains approximately 190 billion liters of water with pH around 2.5 and elevated concentrations of copper, zinc, cadmium, arsenic, and sulfate. Copper has been recovered from the water using cementation (iron displacement) since the 1990s. Research at Montana Tech and by companies exploring the pit as a mineral resource has examined recovery of other metals. The water is pumped and treated to prevent it from reaching the critical water level at which it would flow into the alluvial aquifer connected to Silver Bow Creek. The treatment obligation exists regardless of whether metal recovery is pursued, so the incremental cost of adding recovery processes is the relevant economic comparison, not the full cost of water extraction and treatment.

When dewatering stops, water levels rise through rock that has been draining into an open atmosphere for years or decades. The accumulated oxidation products on fracture surfaces and in rock pore spaces go into solution. The first pulse of water to reach the discharge point carries the stored contamination load in concentrated form. How concentrated depends on how much oxidation product accumulated during the dewatered period and how fast the water table rises through the oxidized zone. Slow rebound through heavily oxidized workings produces a sustained, high-concentration first flush. Rapid rebound in a moderately oxidized system produces a shorter, less extreme pulse. Predicting the difference requires three-dimensional data on oxidation product distribution that is almost never collected before closure, because collecting it would require access to the mine void during the final years of operation or after closure, at a point when access is being eliminated for safety reasons.

The Witwatersrand Basin in South Africa presents the connected-mine rebound problem at its most challenging. Gold mining on the Witwatersrand started in 1886. Over a century of mining created an interconnected network of underground workings extending over thousands of square kilometers at depths exceeding 3,500 meters. As mines closed from the 1990s onward, dewatering ceased progressively. Water levels in the western basin workings rose and reached the surface around 2002, decanting acid mine drainage into streams flowing through the western suburbs of Johannesburg.

The individual mining companies responsible for specific mines either no longer existed, had been through multiple ownership changes, or argued that their responsibility was limited to their specific lease areas, not to the interconnected basin system that their workings had become part of. The South African government, through the Department of Water and Sanitation, implemented emergency pumping and lime neutralization in the western basin. The cost falls on the public. The eastern and central basins continue to fill. Predictive modeling by the Council for Geoscience and the Water Research Commission indicates that similar decant events will occur in those basins, with timing dependent on the rate of rebound and the elevation of the lowest surface discharge points. The chemistry of the decanting water will depend on the interaction between rising water and decades of oxidation products stored in dewatered workings and backfilled stopes.

The Witwatersrand problem cannot be solved mine by mine because the mines are no longer separate hydrological entities. Regional management, coordinated pumping to control water levels across the basin, centralized treatment at locations with adequate infrastructure, long-term monitoring of the interconnected void system, is the only approach that matches the scale of the problem. The institutions to implement regional management at this scale, with funding mechanisms that persist across decades, do not exist. Building them requires agreement among national government departments, provincial authorities, remaining mining companies, municipal governments, and affected communities. The technical components are understood. The governance is not built.

Appalachian coalfields have the same problem at a different scale and with different institutional context. The Abandoned Mine Land Reclamation Program funded by the Surface Mining Control and Reclamation Act has financed treatment and remediation at individual coal mine discharge sites since 1977. Watershed-level coordination, where multiple discharge points within a single drainage basin are prioritized and treated as a system, has been pursued by state agencies in Pennsylvania, West Virginia, and Ohio, and by watershed organizations like the Foundation for Pennsylvania Watersheds and the Appalachian Regional Commission. These efforts have achieved measurable water quality improvements in specific watersheds through strategic treatment of the highest-loading discharge points. The work depends on sustained funding, volunteer effort, and institutional continuity that is vulnerable to budget cycles and political changes.

Permit limits are set for individual parameters: pH must be between 6 and 9, iron below 3 mg/L, copper below some fraction of a milligram per liter. Each parameter is evaluated independently. The receiving ecosystem does not experience parameters independently. It experiences the combined effect of elevated sulfate, elevated selenium, reduced base flow due to consumptive water use, altered temperature regime from pit lake discharge, and whatever other stressors are present. The cumulative and interactive effects of these stressors are not captured by single-parameter permit limits.

The Elk Valley is the case where this gap has been most thoroughly documented. For years, the coal operations met their permit conditions. The fish in the rivers showed reproductive effects from selenium bioaccumulation. The permit system was not designed to detect or respond to chronic, sub-lethal, food-chain-mediated impacts from a constituent that was not on the original permit. The regulatory response, when it came, required a new regulatory instrument (the Area-Based Management Plan and ministerial order) that went beyond the standard permitting framework.

Whether other jurisdictions have comparable situations going undetected is not something that can be answered from published data, because the monitoring programs that would detect them do not exist at most mine sites. Selenium is the test case because it bioaccumulates efficiently, produces effects at low waterborne concentrations, and has been studied intensively in a few locations. Other constituents with sub-lethal chronic effects at low concentration, molybdenum affecting ruminant livestock, sulfate affecting benthic invertebrate communities, manganese accumulating in sediments, may be producing effects at mine sites where the monitoring program was designed to check permit compliance rather than to assess ecological condition.

Community engagement is either specific and responsive to material concerns or it is a formality. The communities downstream of a mine use the water for drinking, irrigation, livestock, recreation, and cultural practice. Their interest in mine water management is practical and sustained. An operation that provides monitoring data in accessible format, responds to specific questions with specific answers, and follows through on commitments builds a relationship that can absorb the stress of incidents and changing conditions. An operation that provides quarterly summary reports in technical language, holds community meetings where the answers to difficult questions are "we're looking into it," and treats engagement as a regulatory requirement rather than a relationship, does not build that resilience. When something goes wrong, and something always goes wrong eventually over a multi-decade mine life, the accumulated trust or accumulated suspicion determines whether the community and the operation can work through it or whether the relationship breaks down into adversarial proceedings.