Brazil Tapajos Telmer UNIDO chapter

Submitted in February, 2003 as a chapter of the UNIDO sponsored book titled “Mercury in the
Tapajós Basin” to come out in 2003.
Mining Induced Emissions of Sediment and Mercury in the Tapajós
River Basin Pará, Brazilian Amazon, 1985-1998, determined from the
Ground and from Space
Kevin Telmer1, Maycira Costa2, Rômulo Simões Angélica3, Eric S. Araujo4 and Yvon Maurice5
1
School of Earth and Ocean Sciences, University of Victoria, Victoria, BC, Canada
Instituto Nacional de Pesquisa Espacial – Brazilian Institute for Space Research, São Jose dos
Campos, SP, Brasil
3
Departmento de Geochimica e Petrologia, Centro de Geociências - Universidade Federal do Pará,
Belém, Pará, Brasil
4
Companhia de Pesquisas e Recursos Minerais – Geological Survey of Brazil, Goiánia, Goias,
Brasil
5
Geological Survey of Canada, 601 Booth St., Ottawa, Ontario, K1A 0E8, Canada
2
Abstract
Dissolved mercury concentrations in waters draining mining operations in the Reserva
Garimpeira do Tapajós are elevated when compared to concentrations in pristine rivers and the
Tapajós River. However, essentially all dissolved mercury concentrations fall below 0.1 ppb.
Mercury bound to suspended sediment is roughly 600 and 200 times the concentration of dissolved
mercury per litre of water in impacted areas and pristine areas, respectively, and thus represents the
major pathway of river-borne mercury. Median concentrations of Hg in suspended load from both
impacted and pristine waters are 145 ppb and 80% of samples are below 300ppb - in the range of
naturally occurring surficial materials world-wide.
Regionally, aqueous Hg fluxes are proportional to the concentration of total suspended
solids suggesting that the dominant source of mercury is the sediment itself and not mercury
discharge from garimpos. The calculated annual export of mercury from the Crepori River is 4
tonnes for the year 1998. This mass is difficult to account for by garimpo discharge alone
suggesting that the regional riverine mercury problem is dominantly caused by the enhanced
physical erosion of garimpo sluicing and dredging operations and not by direct mercury discharge.
This has major implications for remediation and prevention. Halting the use of mercury will reduce
local pollution and is surely a step in the right direction but regional mercury pollution will only be
reduced when the dredging is stopped or contained.
Satellite imagery is shown to be very useful in reconstructing historical mining activity,
locating impacted areas, and estimating historical mercury fluxes. It is also clearly a useful tool for
effective monitoring of present and future mining activity. The spectral characteristics of satellite
images have been matched to the suspended sediment load of rivers and their associated mercury
concentration. This illustrates the spatial extent of the impact of mining on the Tapajós basin. To
compare current activities with those in the past, the relationships constructed from 1998 imagery
and water chemistry are applied to images from 1985. this shows that mining induced sediment
plumes have been contaminating the Tapajós River system for decades and that the intensity of
these plumes was greater in 1985 than they were in 1998. It also documents the shifts in location
and intensity of mining operations through time. For example tributaries on the west bank of the
Tapajós were actively being mined in 1985 but had been abandoned in 1998. Historical mining
emissions of sediment and associated Hg and in-basin storage of these emissions, therefore must
also be considered in any Hg source apportioning theories.
Introduction
The Amazon is the world’s largest tropical rainforest, fed by the world’s largest river and
home to the richest and most diverse ecosystem on earth. The Amazon River carries a staggering
20% of the earth’s river water. It has 1100 large tributaries, 10 of which carry more water than the
Mississippi. One of those ten, the Tapajós River, boasts a drainage basin of 520,000 square
kilometres, representing 8% of the Amazon Basin (Figure 1). Apart from being one of the most
prominent of the Amazon’s tributaries, the Tapajós is well known for being the focal point for
Brazil’s modern gold rush. Since the 1950s, the Tapajós River Basin has been mined for placer
(sediment-based) gold. Through the 1960s, gold mining activity increased with the discovery of the
rich ores in the tributary Marupa (Figure 1). A combination of events in the 1970s then led to the
modern gold rush: (1) the federal government attempted to tame the Amazon with an ambitious
infrastructure program that built the 3000 km long Transamazonica highway. This provided
transportation routes for heavy equipment and access to previously remote regions of the Amazon;
(2) in 1978 the price of gold skyrocketed to over $800 US an ounce; (3) at the same time, drought
struck Brazil’s most impoverished region - the Northeast - driving thousands of poor farmers into
the Amazon in search of a basic livelihood. These farmers turned gold miners, are called
garimpeiros.
After the price of gold skyrocketed, perhaps 200,000 garimpeiros were working the Marupa
and Creporí Rivers (Veiga, 1997). The boomtown of Creporizão arose at the junction of these two
rivers while Itaituba on the banks of the Tapajós River became the gold shop capital of the Amazon.
During these high times, rumours abounded of extraction rates by single workers of up to 2kg/of
gold per day.
When the price of gold peaked at $800 an ounce in 1978, so too did mining activity. During
this era, 90 tonnes of gold were extracted by ~80,000 garimpeiros from the infamous Sera Pelada
deposit in the Carajás region of Pará. Social turmoil and tax evasion prompted the government to
attempt to establish a system of regulation, but this failed primarily because garimpos (the small
scale gold mines) were still constitutionally illegal. The antiquated Brazilian constitution gave
farmers ownership of the land surface only, while the subsurface remained the exclusive property of
the government. For example, a farmer could not sell water from a spring on his land without
governmental permission, which could be arbitrarily denied.
As the rich gold deposits were exhausted, garimpeiros migrated to new and more remote
areas leading to the establishment of over 300 airstrips in the Tapajós basin. Under such conditions
the government agency responsible for regulating mining activity DNPM (Departmento Nacional
Producão Mineral) realised that existing regulation systems were futile and so in an attempt to bring
the garimpeiros back into contact with the government, legal status for garimpos was considered. In
1988 a constitutional change allowed the introduction of the PLG (permissão lavra garimpeira),
which allowed garimpeiros to establish and operate garimpos legally for the first time. Under these
conditions the Reserva Garimpeira do Tapajós and nine other regions in the legal Amazon were
established to protect the land holdings of garimpeiros. Within these special areas, mining
companies cannot stake claims without first negotiating with the affected garimpeiros as was
formerly allowed. This has stabilized both the society of the garimpeiros and their relationship with
the Brazilian government sufficiently to allow extensive collaboration on addressing the
environmental problems associated with garimpo style mining, namely, river siltation and mercury
pollution.
The Use of Mercury
The use of mercury is a common, easy-to-use, and cost-effective method for mining gold. It
has been employed since Roman times, and more recently in California’s gold rush and Canada’s
Klondike near the end of the 19th century. Because gold and mercury easily form amalgam, mercury
can be used to sponge up the fine particles of gold found in river sediments. To recover the gold, the
amalgam is simply heated, vapourizing the volatile mercury, leaving a precious metal residue. Its
like using water to extract sugar from a mixture of sugar and sand. Water is added to the sugar and
sand mixture to dissolve the sugar; the liquid is separated from the solids by filtering; and the sugar
is re-constituted by evaporating the water. The Hg extraction method is the dominant one used by
garimpeiros to extract the very fine gold particles found in modern and paleo-river sediments in the
Tapajós basin. Consequently, toxic accumulations of Hg in the food chain including humans have
been reported (Akagi, 1995) and minamata disease is a grave concern (Lebel et al., 1996).
Between 1550 and 1930, roughly 250,000 tonnes of mercury were released to the
environment globally through this process (Lacerda, 1997). Beginning in the 1930s, mercury-based
technologies were replaced by more efficient cyanide based methods. However, today, due to its
low cost and ease of use, mercury amalgamation has once again become the dominant technology
used by an estimated one million artisanal miners active in Latin America (Veiga, 1997). Many of
these work in the Reserva Garimpeira do Tapajós, at 29,000 square kilometres, the largest artisanal
mining reserve in Latin America. Today, despite the fact that the scarcity of easily exploitable ores
and the huge drop in the price of gold has left many garimpos and boomtowns like Creporizão in
decline, mining activity continues.
Extraction Methods
Throughout the Reserva Garimpeira do Tapajos, three types of Garimpo operations can be
found, all of which use mercury to amalgamate gold. The first, the bausa, a small-scale dredge, is
manned by 2-3 workers and a boss – a Dono. A simple pump/suction system is guided by crudely
equipped divers to vacuum sediment from the river bottom and pump it through a small riffled
sluice box lined with coarse synthetic carpets. Because gold is so heavy, gold grains and even very
fine gold particles will rapidly sink and get trapped in the carpet fibres while common minerals like
quartz, feldspar, and clays, because they are lighter, are swept away by the swift current in the
sluices. At day’s end, the “gravity concentrate” trapped in the carpets is shaken free and collected.
Mercury is mixed in with the gravity concentrate to amalgamate any gold. The amalgam is then
squeezed through fabric - often an old T-shirt - to remove any excess mercury.
The remaining amalgam is placed in a spoon or pan and heated with a butane torch to
vapourize the mercury leaving behind a precious metal residue called “sponge gold”. Mercury
recovered by squeezing the t-shirt is re-used. The vapourized mercury is only rarely captured by
using a retort and so generally, it is escapes to the atmosphere as gaseous mercury. Its fate is not
well known as some is certainly redeposited locally within 1-2 kilometres (Lindberg et al. 1992),
while another fraction could potentially remain in the atmosphere for a long time as the average
residence time for Hg in the atmosphere is estimated to be around 1 year (Mason et al., 1992). How
much is deposited locally and how much enters the global Hg cycle is unknown but is probably
controlled by weather and other environmental factors. Roughly 5% of the amalgamated mercury
remains in the sponge gold product – the low temperature butane torches being too cool to release
100% of the mercury. The sponge gold is taken to town where it is further refined, removing the
final 5% of mercury. To accomplish this, the sponge gold is mixed with borate which acts as a flux
to assist with both melting and with trapping silica based impurities such as quartz. The mixture is
heated to roughly 1000 oC, producing two separate immiscible phases – one, a heavy metallic liquid
consisting primarily of gold but also silver and traces of other heavy metals such as copper, lead,
and platinum. The other, a silicate glass composed mainly of borate plus silicate impurities. This
procedure releases any remaining mercury which is exhausted out the chimney in the middle of
town – sometimes first through a water retort. Highly elevated Hg concentrations in the immediate
vicinity of gold shops have been reported by de Andrade et al., 1988. The gold is poured into a
mould and quenched, producing a small ingot. Only the gold is paid for. To determine the quantity
of gold versus other metals in the ingot, it is weighed first in air and next in water. The two weights
are entered into a formulae and the percentage gold is determined. 80 - 95% gold is typical but it is
unpredictable. Some of the product is crafted and sold locally at bargain prices but most is taken to
São Paulo where the gold and other metals are separated into pure forms by sophisticated electrochemical methods. These are then sold on international markets.
At the Bausa, profits are split - 30% to be divided amongst the divers and 70% for the Dono.
The Dono must maintain the operation, buy fuel for pumps, and secure mining rights for his
operation’s position on the river. A typical salary for a garimpeiro diver is 40g of sponge gold per
month. That translates into just over 1 ounce of pure gold. At current prices, this is roughly $300
US a month. This is used to buy food and pay the camp cook and laundry washer. She typically
receives 1g/day or 30g/month, a payment shared equally among the divers. Garimpeiros typically
work six and a half days a week minus one weekend per month.
The second type of extraction method – called “Barranco” or River Bank – is a larger, more
expansive and more ecologically destructive method of mining gold. River alluvium is blasted into
a soupy mixture by high pressure hydraulic cannons. The soup is then pumped out of a pit and
through sluice boxes much larger than those on the Bausas. The cost of more powerful equipment
and the extra fuel needs for running the larger barranco operations is considerably higher than the
bausas, and so more gold must be recovered to pay the overhead – economies of scale. To
accomplish this, work can be continuous 24 hours a day, 2 shifts of 12 hours.
The gold is most concentrated in pebble beds and associated iron crusts - products of
millions of years of tropical weathering and subsequent reworking by rivers. The concentration
mechanisms and source of the gold are still poorly understood. Regardless, garimpeiros use little
geological knowledge, relying instead on intuition and luck and trial and error to lead them to ores.
After passing through the sluice box, tailings flow into the river. Due to the finer nature of
river bank sediments and their higher proportion of clays, the suspended sediment load and its
residence time in the water column of rivers of this waste is much higher than that of the Bausas.
The “draga” is the third and most destructive mining method used by garimpeiros. It is a
large industrial scale dredge. The draga is really a combination of the bausa and the barranco, but
also much more automated. They use huge engines and hydraulic cranes to suck up enormous
amounts of both river and riverbank sediment. Operating typically in 40 hour cycles, dragas can
excavate small lakes in the flood plains of rivers. They typically recover 100 - 500 grams of gold
per cycle. They need to recover a minimum of 100 grams per cycle to cover costs versus about 50
grams for the bausas. Nevertheless, due to their greater throughput, dragas are usually more
profitable than the other operations. The main disadvantage of the draga is its lack of mobility. Such
large equipment cannot exploit ore in smaller, more inaccessible rivers and riverbanks.
From 1980 - 1993, fleets of dragas scoured not just the Creporí and Marupa rivers, but also
the Tapajós River proper. Long time captains of riverboats used for transporation say that during
the height of the gold rush, the clear blue water of the Tapajós was turned muddy brown for a
decade. The satellite imagery discussed later confirms this observation. Dragas were outlawed on
the Tapajós proper in 1992 and since that time, the Tapajós has begun to recover. However, it still
receives huge sediment inputs from its tributaries where dragas, bausas and barranco operations
continue to operate.
Study Area
The Tapajós River Basin (Figure 1) resides within the southern Amazon Basin in Brazil.
With a mean discharge of 7,200 m3/s (Gleick, 1993) and a drainage area of 520,000 km2
(determined by GIS) it is one of the larger tributaries of the Amazon River. The Crepori River is a
medium sized tributary of the Tapajós River with a drainage area of roughly 50,000 km2
(determined by GIS) and a mean discharge of 700 m3/s (minimal estimate using the area/discharge
relationship for the Tapajós Basin). The Crepori River begins in pristine uplands in the central
Tapajós basin, flows north bisecting the area of garimpo activity known as Creporizão and then
continues on to conflue with the Tapajós River some 250 km south of Itaituba, the largest town on
the Tapajós River and the centre for the Amazon's gold economy.
Method
Water samples from the Reserva Garimpeira do Tapajós, the Tapajós River and Pristine
Streams were taken along the length of rivers and from the mouths of their major tributaries at as
close to one instant in time as possible - in this case 2 weeks - twice, once during low water stand in
October 1997 and during high water stand in May of 1998. 51 samples were collected per
campaign, not including duplicates, 29 from a 1:50,000 area surrounding the garimpeiro community
of Creporizão and 21 from a larger 1:250,000 scale area also centred around Creporizão. This
strategy was adotpted to capture both detailed (impacted) and regional scale (background)
geochemistry.
All samples are filtered to separate the dissolved from suspended load and are then treated
and stored at 4oC (in iced coolers in the field) for chemical analyses. Water for the determination of
Hg is processed immediately to avoid any Hg loss by volatolization. Water is filtered by syringe
pressure and preserved with BrCl to stabilize and oxidize all Hg species to Hg2+ as per USEPA
method 1631 (Environmental Protection Agency, 1999). The membranes from the cation aliquots
and another for the determination of particulate organic matter are retained and collected in small
polyethylene vials for the determination of total suspended solids and the Hg contained therein.
Further details on sampling methodology can be found in Telmer et al. (2002) and Telmer and
Desjardins (2003).
Some garimpos are no longer working on placer ores but have switched to directly mining
gold bearing veins in saprolites or in unaltered bedrock. This gold is often associated with sulphides
and so these are ground to relaese the gold. To amalgamate and capture the gold a slurry is typically
passed over a Hg coated copper plate. The grinding and dissolution of sulphides produces acid (acid
mine drainage) and so waters being discharged from such operations are very low pH. To
understand the fate of Hg added to the environment in such operations, samples were taken of the
effluent from Garimpo Joel. Samples were taken directly from the discharging mill and of the same
effluent 100m further away. As well, the samples were taken 1 year later after the operation had
been shut down for ~1 year.
Disolved Hg for these samples was analysed at the Geological Survey of Canada by hydride
generation and ICP-MS or by Cold Vapor Atomic Fluorescence Spectrophotometry for very low
concentration samples. For determination of Hg in the suspended load, a fixed volume of sample is
pushed through 47mm pre-ignited glass fibre membrane. The membrane and its load are then
analysed for Hg directly by ignition on a Milestone AMA 254 Mercury Analyser. An evaluation of
Hg analyses capability using the Milestone AMA 254 can be found in Hall and Pelchat (1997).
Results are shown for waters and suspended load in Figure 3 and for the direct milling operation at
Garimpo Joel in Figure 4.
To quantify Hg in potential source materials, a depth profile of supergene materials (latosol)
was sampled and analysed. Samples were taken about 1km from a garimpo (Garimpo Mamoal),
separated into 4 size fractions, and analysed by cold vapour atmoic adsorption after digestion with a
3:1 mix of HCl and HNO3 (aqua regia). Results are shown in Figure 5.
Satellite image processing was done as follows. LANSAT 5 satellite images from 1998 and
1985 were geometrically corrected using the polynomial method available in the PCI/GCPworks
software package. This method uses ground control points chosen from a geometrically corrected
reference (maps) and equivalent points in the uncorrected images. In this case the UTM projection
(row M, Zone 21) and the SAD69 South American datum were the cartographic parameters used.
The images were corrected with a minimum of 30 GCPs and a second order polynomial and cubic
convolution interpolator. th efinal accuracy of the corrected images is estimated at 2.7 pixels or 81
meters. Better accuracuy is possible with better base maps but they are not available for this region.
A land mask was generated so that the processing of the water alone could be accomplished. A
linear enhancement was applied to LANDSAT band 2 to highlight the different optical water
masses. A more robust enhancement across multiple bands and and a calibration to field measured
total suspended solids is possible allowing the classification of water masses into discrete entities
with known concentrations of suspended solids. This is currently uunderway. The results of the
image processing are shown in Figures 7 and 8.
Discussion
Source and Fate of Hg
Dissolved mercury concentrations are elevated in waters affected by mining operations
when compared to concentrations in pristine rivers (Figure 3). However, essentially all dissolved
mercury concentrations fall below the commonly used aquatic life limit of 0.1 ppb. Mercury bound
to suspended sediment tells another story. Particulate Hg fluxes average 600 and 200 times those of
dissolved mercury in impacted and pristine waters, respectively, and thus, particulate bound Hg
represents the major pathway of river-borne mercury.
This is clearly demonstrated by the results from Garimpo Joel. The immediate discharge
waters from Garimpo Joel are very low pH and high in suspended solids. The low pH is caused by
the oxidation of sulphide minerals found in the ore. These waters are extremely high in dissolved
Hg (300 ng/g versus an average of 0.045 ng/g for other contaminated waters in the area). Within
100 meters of the milling and amalgamating operation, the concentration of dissolved Hg drops by
300 times as it is adsorbed by suspended sediments – as illustrated by the crossover in the graph.
After gold recovery dropped, Garimpo Joel was abandoned. Suspended particles in waters draining
the tailings remain highly elevated in Hg. It should be noted though that this kind of operation is
rare and contaminated discharges from it are insignificant on a regional basis. It is shown here not
to illustrate source but to illustrate the behaviour of Hg in this environment.
The results of the depth profile in surficial materials (Figure 5) show that Hg concentrations
decrease with depth and degree of weathering from about 300 to 100 ng/g dry weight with the
uppermost, most weathered and smallest size fractions reaching a maximum of 360 ng/g. This trend
is similar for many other elements such as Au, As, B, W, Cu, Sn, and others in supergene material
in the region as demonstrated by Costa et al., (1999). The profile is shown here to demonstrate that
supergene materials such as latosols and saprolites are a natural source of Hg that is available for
riverine transport if the these materials are washed into rivers.
As shown by Figure 3, total Hg in impacted waters average 15 times greater than those of
pristine waters but can be up to 100 times greater. This indicates clearly that mining has a profound
impact on transport rates of riverine Hg. However, this does not elaborate the source of the Hg.
Figure 6a shows that Hg fluxes are proportional to the concentration of total suspended solids, and
the histogram of Hg concentration in total suspended solids (inset, figure 6b) shows that the
majority of samples, regardless of source, either impacted or pristine, have a consistent
concentration of Hg. The outliers are from samples immediately proximal (within 100 m) to mining
operations such as Garimpo Joel. This strongly suggests that the source of mercury is the sediment
itself and not mercury discharge from Garimpos. This is supported by the observed Hg
concentrations in the supergene depth profile (Figure 5) and by the fact that these concentrations
fall within the normal range of surficial materials world-wide. Figure 6c shows a weaker
proportionality between loss on ignition (LOI), a proxy for particulate organic carbon (POC), and
particulate Hg. The best-fit lines for 6a and 6c suggest concentrations of Hg in suspended sediments
of 134 ng/g, and in POC of 1620 ng/g.
Further evidence suggesting that the majority of Hg emanating from the Creporí River into
the Tapajós comes from accelerated erosion and not direct Hg inputs is provided from a mass
balance calculation. Using an annual flux of water from the Creporí of 700 m3/s and multiplying by
the concentration of suspended sediment at the mouth of the Creporí River of 540 mg/L and
multiplying by the mean concentration of Hg in the sediment of 134 ng/g produces an annual flux
of Hg of 1.6 tonnes. This is a very conservative estimate (a minimum) as the average concentration
of Hg in TSS for the region was used, not the concentration measured in suspended sediments at the
mouth of the creporí River, and as well, it is likely that the concentration of suspended sediments in
the Creporí is considerably higher in the rainy season due to enhanced erosion of the areas exposed
by mining. The lower value of Hg was chosen to reflect the estimated regional concentration and to
ensure the estimate is robust rather than using a value based on 2 samples from the mouth of the
Creporí. An unconservative and perhaps more realistic estimate would be closer to 5 tonnes of Hg.
Presently, Hg released by garimpeiros is almost exclusively atmospheric as mercury is generally no
longer used directly in sluice boxes. This practice has been eliminated largely through education
programs and the fact that this does not increase yields. Although a portion of the Hg released to the
atmosphere may find its way back into the river, it is unlikely that much of this atmospheric release
finds its way back into the river quickly – Hg does not have the geochemical behaviour of Chloride.
Rather, it is likely to get adsorbed by mineral phases such as Fe-oxyhydroxides in soils or by
organic matter. Hg captured by soils in this manner would take hundreds or thousands of years to
get leached into rivers. This makes the calculated annual export of mercury from the Creporí River
difficult to account for by Garimpo discharge alone and again points to enhanced physical erosion
of garimpo sluicing and dredging operations as the main source of riverine Hg. This is not to say
that direct discharge of mercury is always insignificant. Analyses of river sediments shows
localized hot spots of contamination.
Noise around the best fit lines is likely caused by two factors: (i) changing the mix of source
materials that contain different Hg concentrations. Higher levels in the soil profile contain more Hg
(perhaps due to more organic carbon) and therefore more Hg enters waters when these materials are
eroded by mining than when recent river alluvium or deeper supergene materials are eroded.
Outliers strongly enriched in Hg are the product of Hg pollution from gold mining but as noted
above these are present only in very close proximity to the mining operations and are secondary in
terms of regional Hg flux. All of these high values come from small heavily worked “igarapes”
(creeks) very close to operations and none of them come from the Creporí or Tapajós River proper.
Background concentrations of suspended solids in the Tapajós waters upstream of the
Creporí plume are ~7 mg/l. The Creporí has concentrations of up to 500 mg/L but once flowed as
clear as the Tapajós. The intense plume of sediment emitted by Rio Creporí hugs the east bank of
the Tapajós river for many kilometers but eventually becomes homogenously distributed from bank
to bank as the river winds around corners and goes through rapids (Figure 7). Because surficial
materials in the tropics are highly weathered, they are very rich in clays. Figure 9 is an SEM
micrograph of suspended material from the Creporí plume collected on a filter membrane and
magnified 2000 times. The suspended material is agglomerations of superfine clay particles. The
small particle size of clays causes suspended sediments to remain in the water column for very long
distances as can be seen in the satellite images. This allows sediment emitted by the Creporí to be
transported very long distances, perhaps all the way to the Amazon River and beyond. Within the
Tapajós basin similar scenarios exist for other smaller tributaries (such as Rio Rato shown in Figure
7) and also on a large scale in the Alta Floresta area in the upper Tapajós basin.
Roulet et al. (1998) also identified burdens of riverine Hg to be controlled dominantly by
particulate matter. However, they assert that the Hg content of Tapajós waters is independent from
upstream gold mining activities and is in fact dominantly caused by deforestation. The study area
for their work however, is not in the gold mining areas. They sampled the lower Tapajós river from
roughly from Itaituba to Santarem – an area several hundred kilometers downstream of the Reserva
do Garimpeira do Tapajós. We conclude here however, that, on the contrary, gold mining activities
are the greatest source of particulate matter to the Tapajós system. Landuse changes such as
deforestation may be an important secondary source of particulate matter to the Tapajós river,
particularly during the wet season, but as illustrated by satellite imagery (Figure 7), the suspended
sediment emitted from the Rio Creporí is its dominant source into the Tapajós.
In terms of the fate of suspended sediment hosted Hg, it is possible, as described above, that
it makes it to the Amazon and beyond due to the superfine nature of the particles. However, the
mouth of the Tapajós river is very wide and lake like due to damming by the Amazon river and so
flow is greatly reduced and settling of particles would be enhanced there. It is possible that this
forms a sink for sediment hosted Hg which then finds its way into fish after methylation at the
sediment water interface possibly explaning elevated fish Hg in the lower Tapajós. But this is
speculation and it is important to keep in mind that elevated fish Hg has been recorded in many
other areas of the Amazon basin where there are few or no mining activities (Silva-Forsberg et al.,
1999).
Extent and History of Emissions
Mining induced sediment plumes must have been contaminating the Tapajós system for 20
years or more – since the beginning of the gold rush. The satellite imagery shown in Figures 7 and 8
from 1998 and 1985 clearly show this to be the case but are also useful for other purposes. The
images document that the footprint from the mining operations extends hundreds of kilometers
downstream. The comparison of activities in 1998 with those in 1985 shows that mining induced
sediment plumes have been contaminating the Tapajós River system for decades and that the
intensity of these plumes was greater in 1985 than they were in 1998. It also documents the shifts in
location and intensity of mining operations through time. For example tributaries on the west bank
of the Tapajós were actively being mined in 1985 but had been abandoned in 1998. And tributaries
of the Rio Jamaxim were being more aggressively mined in 1985 than in 1998. An important
conclusion of these results is that, investigating the source and impact of placer mining operations
cannot be done simply by taking samples of current conditions. Historical mining emissions of
sediment and its associated Hg, and the in-basin storage of these emissions, must also be considered
in any Hg source apportioning theories.
Currently, we can only document and verify that the locale and intensity of mining has
varied significantly through time but cannot yet quantify the history of mining in the Tapajos
region. To do so, a larger database of imagery and a more intensive processing effort is required.
We have begun this task in the hope of providing a quantitative history of gold mining activities in
the Tapajós extending from the 1970s through to the present. Imagery is also clearly a useful tool
for effective monitoring of present and future mining activity. And perhaps more importantly, it can
be used to observe and monitor the healing of wounded landscapes and communities.
Prevention and Remediation
As desirable as it may seem, Garimpo operations cannot simply be stopped. They directly
employ almost half a million people in the Amazon and countless others indirectly. The large
majority of garimpeiros are subsistence workers, just scraping by from pay-day to pay-day. They
have no savings and they have no other skill. Brazil has no social safety net and the capacity of
retraining programs are insufficient. Many who do quit end up in the favelas of the large
Amazonian cities of Manaus or Belém, where it is difficult to escape social marginalization.
Farming is a viable option for some garimpeiros as many came from farms in northeast
Brazil. But the same land ownership and unfair profit sharing problems exist in Amazonia as those
they initially escaped from. Rich landowners, poor labourers. As well, the fruits of labour on the
farm are slow and predictable. The attraction of getting rich quick, even if it never happens on the
garimpo, does not exist at all on the farm. For many, farming is resigning oneself to a life of poverty
and labour.
Better technology and improved use of existing technology could reduce the ecological
impact of garimpos. Presently, garimpos lose roughly 50% of their gold due to poor sluicing
practices and because fine gold is trapped inside other minerals. Gravity concentrators, sluice
operation education, and milling of the ore can increase the yield but unfortunately do not reduce
river siltation nor necessarily the use of mercury. Some of these technologies are in practice in
Brazil but due to investment costs, field logistics and ignorance, their use is not widespread.
Containment ponds could hugely reduce river siltation and therefore mercury contamination at the
same time. Plankton would flourish in clear flowing rivers which could again feed a thriving
fishery. However, containment represents a lot of additional labour and planning but produces no
additional gold. Given the cost/benefit analyses of simple containment it is doubtful such an
approach would be embraced.
If containment were combined with technologies which yielded more gold, such as cyanide
heap leaching, it might sound attractive but it may not be viewed as practical as it requires much
greater initial investment, a higher level of expertise to operate, and a more organized labour pool;
rarities in garimpeiro culture. Distrust of authority, rugged individualism, frontiersmanship and
ignorance, work against the implementation of more organized labour and more sophisticated
technology. As well, use of cyanide has its own serious environmental hazards and although its
impacts are more localized than those of Hg, they are unacceptably risky in remote and unregulated
areas and the health consequences for unsafe practices (death) are far more servere and immediate
that the use of Hg.
And so without practical intervention from government, it is likely that garimpeiros will
continue to practice inefficient and destructive but simple and attainable mercury amalgamation
methods. Perhaps basic education and health care are good first steps in changing the legacy of the
gold rush from one of ignorance and destruction into something more sustainable.
Currently, garimpeiro culture has no vision of the future. Extract the land’s wealth today
without weighing the consequences of tomorrow. Fundamental changes are needed to prevent a
legacy of self-destruction and poverty from being passed onto future generations. Long-term
government investment and goals are needed to bring a sustainable economy to the goldrush zone in
the Amazon.
References
Akagi, H., Malm, O., Kinjo, Y., Harada, M., Branches, F.J.P., Pfeiffer, W.C., and Kato, H. (1995)
Methylmercury pollution in the Amazon, Brazil. The Science of the Total Environment,
175, no. 2, pp. 85-95.
Costa, M.L., Angélica, R.S., Costa, N.C. 1999. The geochemical association Au-As-B-W-Cu-Sn in
latosol, colluvium, lateritic iron crust and gossan in Carajás, Brazil: its importance for
identification of primary ore. J. Geochem. Explor., 67(1-3): 33-49
Environmental Protection Agency, United States. (1999) Method 1631, Revision B: Mercury in
Water by Oxidation, Purge and Trap, and Cold Vapour Atomic Flourescence Spectrometry
33 pp
Gleick P.H. (1993) Water in Crisis, A Guide to the World's Fresh Water Resources. New York,
Oxford University Press.
Hall, G.E.M. and Pelchat, P. (1997) Evaluation of a direct solid sampling atomic absorption
spectrometer for the trace determination of mercury in geological samples. The Analyst,
122, pp. 921-924.
Lacerda, L.D. (1997) Global mercury emissions from gold and silver mining. Water, Air, and Soil
Pollution, 97, pp. 209-221.
Lebel, J., Mergler, D., Lucotte, M., Amorim, M., Dolbec, J., Miranda, D., Arantes, G., Rheault, I.,
and Pichet, P. (1996) Evidence of early nervous system dysfunction in Amazonian
populations exposed to low-levels of methylmercury. NeuroToxicolgy, 17, no. 1, pp. 157168.
Lindberg, S.E., Meyers, T.P., Taylor Jr., G.E., Turner, R.R., and Schroeder, W.H. (1992)
Atmosphere-surface exchange of mercury in a forest: Results of modeling and gradient
approaches. Journal of Geophysical Research, 97, no. D2, pp. 2519-2528.
Mason, R.P., Fitzgerald, W.F., and Vandal, G.M. (1992) The sources and composition of mercury
in Pacific Ocean basin. Journal of Atmospheric Chemistry, 14, pp. 489-500.
Roulet, M., Lucotte, M., Canuel, R., Rheault, I., Tran, S., De Freitos Gog, Y.G., Farella, N., Souza
do Vale, R., Sousa Passos, C.J., De Jesus da Silva, E., Mergler, D., and Amorim, M. (1998)
Distribution and partition of total mercury in waters of the Tapajos River Basin, Brazilian
Amazon. The Science of the Total Environment, 213, no. 1, p. 203.
Silva-Forsberg, M.C., Forsberg, B.R., and Zeidemann, V.K. (1999) Mercury Contamination in
Humnas Linked to River Chemistry in the Amazon Basin. Ambio, 28, no. 6, pp. 519-521.
Telmer K. and Desjardins M. (2003) Mercury in Lake Sediments and Porewaters: Multiple
Evidence for Remobilization. In: A Multidisciplinary Study of Mercury Cycling in a
Wetland Dominated Ecosystem: Kejimkujik Park, Nova Scotia. Eds. Andrew Rencz and
Nelson O'Driscoll. Society of Environmental Toxicology and Chemistry.
Telmer, K., Pass, H., and Cook, S. (2002) Combining dissolved, suspended and bed load stream
geochemistry to explore for volcanic massive sulfide deposits; Big Salmon Complex,
northern British Columbia, Journal of Geochemical Exploration Vol. 75, p. 107-121.
Veiga, M.M., Meech, J.A., and Onate, N. (1994) Mercury pollution from deforestation. Nature,
368, pp. 816-817.
Figures
Tapaj
ó
s
Amazon
Brazil
Tapajós River Basin
Km 250 0
1000
Figure 1: the Tapajós river Basin, Pará, Brazil, with the Creporí River basin illustrated in the inset.
Syringe
membrane
filtration
River
Water
Syringe
membrane
filtration
Add BrCl
Liquid
Chromatog.
Water
Zinc
Reduction Equlibration
Unit
-
Al
F
Ba
Cl
Ca
N O 3Fe(total)
NO2
K
Br
2Mg
SO4
Mn
Na
Rb
Si
Sr
Trace Elements
Atomic
Flourecence
DIC
Preparation
Line
Sulphate
Distillation
+
Hg
Mass Spectrometry
Gas Source
Mass Spectrometer
Solid Source
87
86
Sr/ Sr
Figure 2: Field and Laboratory Methodology
BaSO4
δD
δ18O
δ34S
δ13C
Field
Turbidity
Laboratory
pH
Conductivity
Temperature
Transport
Transport
ICP-ES
ICP-MS
Alkalinity
O2
Transport
Nutrients
Turbidometer
Transport
Conductivity
Meter
Photometer
Filtrate (BaSO4)
Filtration
Filtered Water
Add HNO3
Add NaN2
35
Particulate Bound Hg
30
Dissolved Hg
ug/L Partculate Hg (ppb)
ng/L Dissolved Hg (ppt)
25
20
15
10
5
0
Impacted
Recovering
Distantly
Impacted
Municipal
Agriculture
Pristine
Type of Surface Water
Figure 3: Burdens of Hg per litre of water in the dissolved and suspended load of water per source.
1000
Dissolved Hg
Particulate Hg
ng/l
100
10
1
Direct discharge
Tailings drainage
Tailings drainage one year later
Pathway
Figure 4: Hg transformation from dissolved to suspended transport. The immediate discharge waters
from Garimpo Joel are very low pH and high in suspended solids. The low pH is caused by the
oxidation of sulphide minerals found in the ore. These waters are extremely high in dissolved Hg
(300ppb versus an average of 0.045ppb for other contaminated waters in the area). Within 100
meters of the milling and amalgamating operation, the concentration of dissolved Hg drops by 300
times as it is uptaken by the suspended sediments – illustrated by the crossover in the graph. In 1997,
after gold recovery dropped to uneconomical levels, Garimpo Joel was abandoned. Suspended
sediments in waters draining the tailings remain highly elevated in Hg.
0
Latossolo
60
120
Hg (ppb)
180
b
240
300
a
360
c
Saprólito
"veio" no
saprólito
Saprólito
"veio" no
saprólito
a (Amostra Total)
b ( >80# )
c ( <80# - >200# )
d ( <200# )
Saprólito
sobre
granito
Figure 5: Depth profile of Hg in different size fractions of supergene materials.
d
1000
100
30
10
Frequency
Hg ng/L
35
25
20
15
10
5
1
1500
1000
500
0
0
ng/g Hg in dry TSS
Linear Regression r2 = 0.60
y = 0.134x; p<0.01
0
1
10
100
1000
10000
TSS mg/L
1000
Hg ng/L
100
10
1
y = 1.6192x
Linear Regression r2 = 0.47
p<0.01
0
1
10
100
LOI mg/L
Figure 6: a. The proportional relationship between total suspended solids (TSS) and particulate Hg
in surface waters of the Tapajós River basin. b. Histogram of Hg concentration in dry suspended
sediment for 87 samples. c. The weaker proportionality between loss on ignition (LOI), a proxy for
particulate organic carbon (POC), and particulate Hg. The best-fit lines suggest roughly constant
concentrations of Hg in suspended sediments of 134 ppb, and in POC of 1620 ppb. Noise in the data
is caused by changing the mix of source materials with different Hg concentrations (see Figure 4).
Higher levels in the soil profile contain more organic carbon and therefore more Hg enters waters
when these materials are eroded. Outliers strongly enriched in Hg are the product of Hg pollution
from gold mining but these are present only in very close proximity to the mining operations (see
figure 2) and are secondary in terms of regional Hg flux.
Figure 7: main: Landsat TM image from 1997 showing the Creporí river discharging its enormous
mining-induced sediment load into the Tapajós River. Inset: aerial photograph of the Creporí plume
15 km downstream from the confluence with the Tapajós, taken during this project.
Figure 8: Landsat TM image from 1985 showing the Creporí river discharging its mining-induced
sediment load into the Tapajós River. The land is masked in this image to enhance the visibility of
the sediment plume. Its intensity is greater than in the 1997 image in Figure 7. For example the
plume entirely engulfs the large island about 15 km downstream from the confluence of the Creporí
and the Tapajós. The image also shows a sediment plume tracking the west bank of the Tapajós that
is not visible in 1997 illustrating the capabilities of satellite imagery to reconstruct the locale of
historical mining operations and their intensity.
Figure 9: clay particles from the suspended sediment load of the Tapajós River captured on a glass
fibre membrane by filtering. Magnification is 2170 times.