Is mercury from small-scale gold mining prevalent in the southeastern Peruvian Amazon?

Is mercury from small-scale gold mining prevalent in the southeastern Peruvian Amazon?

Environmental Pollution 218 (2016) 150e159 Contents lists available at ScienceDirect Environmental Pollution journal homepage: www.elsevier.com/loca...

1MB Sizes 4 Downloads 35 Views

Environmental Pollution 218 (2016) 150e159

Contents lists available at ScienceDirect

Environmental Pollution journal homepage: www.elsevier.com/locate/envpol

Is mercury from small-scale gold mining prevalent in the southeastern Peruvian Amazon?*  nica Moreno-Brush a, b, *, Johan Rydberg a, 1, Nadia Gamboa c, Ilse Storch b, Mo Harald Biester a €t Braunschweig, Langer Kamp 19c, 38106, Braunschweig, Germany €kologie, AG Umweltgeochemie, Technische Universita Institut für Geoo €t Freiburg, Tennenbacherstr. 4, 79106, Freiburg, Germany €kologie und Wildtiermanagement, Universita Professur für Wildtiero c n Química; GRIDESePUCP, Pontificia Universidad Cato lica del Perú e PUCP, Av. Universitaria 1801, San Departamento Acad emico de Ciencias, Seccio Miguel, Lima 32, Peru a

b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 6 March 2016 Received in revised form 12 July 2016 Accepted 11 August 2016

There is an ongoing debate on the fate of mercury (Hg) in areas affected by artisanal and small-scale gold mining (ASGM). Over the last 30 years, ASGM has released 69 tons of Hg into the southeastern Peruvian Amazon. To investigate the role of suspended matter and hydrological factors on the fate of ASGM-Hg, we analysed riverbank sediments and suspended matter along the partially ASGM-affected MalinowskiTambopata river system and examined Hg accumulation in fish. In addition, local impacts of atmospheric Hg emissions on aquatic systems were assessed by analysing a sediment core from an oxbow lake. Hg concentrations in riverbank sediments are lower (20e53 ng g1) than in suspended matter (~400 e4000 ng g1) due to differences in particle size. Elevated Hg concentrations in suspended matter from ASGM-affected river sections (~1400 vs. ~30e120 ng L1 in unaffected sections) are mainly driven by the increased amount of suspended matter rather than increased Hg concentrations in the suspended matter. The oxbow lake sediment record shows low Hg concentrations (64e86 ng g1) without evidence of any ASGM-related increase in atmospheric Hg input. Hg flux variations are mostly an effect of variations in sediment accumulation rates. Moreover, only 5% of the analysed fish (only piscivores) exceed WHO recommendations for human consumption (500 ng g1). Our findings show that ASGM-affected river sections in the Malinowski-Tambopata system do not exhibit increased Hg accumulation, indicating that the released Hg is either retained at the spill site or transported to areas farther away from the ASGM areas. We suspect that the fate of ASGM-Hg in such tropical rivers is mainly linked to transport associated with the suspended matter, especially during high water situations. We assume that our findings are typical for ASGM-affected areas in tropical regions and could explain why aquatic systems in such ASGM regions often show comparatively modest enrichment in Hg levels. © 2016 Elsevier Ltd. All rights reserved.

Keywords: Amazon Gold mining ASGM Mercury transport Lake sediments Fish

1. Introduction Artisanal and small-scale gold mining (ASGM) occurs in over 70 countries (Telmer and Veiga, 2009) and is the largest anthropogenic source of mercury (Hg) in the environment (UNEP, 2013). In the Brazilian Amazon alone, 2000e3000 tons of Hg has been released

*

This paper has been recommended for acceptance by David Carpenter. € kologie e Abt. * Corresponding author. TU Braunschweig, Institut für Geoo Umweltgeochemie., Langer Kamp 19c, 38106, Braunschweig, Germany. E-mail address: [email protected] (M. Moreno-Brush). 1 Present address: Department of Ecology and Environmental Science, Umeå University, Umeå, Sweden. http://dx.doi.org/10.1016/j.envpol.2016.08.038 0269-7491/© 2016 Elsevier Ltd. All rights reserved.

into the environment since the 1970s (Malm, 1998; Pfeiffer et al., 1993). In the Peruvian Amazon, the Madre de Dios (MDD) region accounted for ~70% of the national ASGM-gold production in 1990 (Kuramoto, 2002), and since then the land area used for ASGM activities has increased by 400% (Asner et al., 2013). Based on data from the Peruvian National Institute of Statistics and Informatics, it is estimated that ASGM in the MDD region used 273 tons of Hg during the period 1990e2003 (based on a Hg-to-gold ratio of 2:1) and that over 40 tons of Hg was used in 2006 alone (based on a Hgto-gold ratio of 2.8:1; Brack et al. (2011)). These estimates, together with reports of high Hg levels in commercial fish species and environmental and human samples from ASGM regions (Ashe, 2012; Barbieri, 2004; Brack et al., 2011; CAMEP, 2013; Deza

M. Moreno-Brush et al. / Environmental Pollution 218 (2016) 150e159

Arroyo, 1996; Diringer et al., 2015; Fernandez and Gonzalez, 2009; Gammons et al., 2006; Wade, 2013), have raised serious concerns among the local population and MDD authorities in terms of the potential ecological and public health issues related to the use of Hg in ASGM activities. ASGM miners often use metallic Hg (Hg0) for extracting gold particles by amalgamation. However, due to the lack of efficient recovery techniques, Hg is released into the environment both during the amalgam process (to soils, tailing, and water systems) and amalgam roasting (to the atmosphere). It is, however, uncertain to what extent Hg is actually being lost to the environment during these processes. Estimated emission factors per kilo of extracted gold vary from 0.1 to 10 depending on the type of ore and the extraction process (Pfeiffer et al., 1993). The fate of the released Hg is also uncertain, e.g., Pfeiffer and de Lacerda (1988) estimated that 45% of the losses go to the hydrosphere and 55% go to the atmosphere, while Pfeiffer et al. (1993) reported that 65e83% of the total Hg losses go to the atmosphere. Considering the large amount of Hg that is estimated to have been released by ASGM activities in the MDD region, both to the hydrosphere and the atmosphere, it should be possible to track the released Hg in soils and sediments. However, no apparent signs of elevated Hg levels have been found either locally in the MDD region (Diringer et al., 2015) or in the southeastern Andes (Beal et al., 2013). The scenario of ASGM-Hg release in the MDD region is similar to the one from preindustrial silver/gold mining in Spanish Colonial America; even though it has been estimated that 100,000e140,000 tons of Hg was released into the environment from 1570 to 1900 (Nriagu, 1994, 1993; Streets et al., 2011), there is little evidence of this massive amount of Hg in geoarchives from both hemispheres (Beal et al., 2014, 2013; Biester et al., 2003; Cooke et al., 2011, 2009; Engstrom et al., 2014; Hermanns and Biester, 2013; Conaway et al., 2012; Lamborg et al., 2002). This discrepancy is a manifestation of our lack of understanding regarding how much Hg is released by ASGM activities and what its fate is. In particular, the final sink of ASGM-Hg losses in aquatic environments is challenging to track not only because of the informal (sometimes illegal) character of the activity but also because Hg dispersion mechanisms, especially in the tropics, are still not well understood. Moreover, most of the theoretical estimates of Hg release have neglected the influence of hydrological and geochemical factors. For the MDD region, Diringer et al. (2015) recently reported higher Hg concentrations in river sections downstream from ASGM activities compared to river sections upstream of ASGM activities. This result suggests that Hg concentrations can be directly linked to ASGM activities, and riverine communities located downstream from ASGM areas are exposed to an elevated risk through fish in their diet. Nevertheless, the Hg levels reported by Diringer et al. (2015), both in river sediments and suspended matter from upstream and downstream ASGM activities, are lower in comparison with findings in other ASGM-affected areas (Gammons et al., 2006; Lechler et al., 2000; Ouboter et al., 2012). It is also worth noting that ASGM is not the only possible source of Hg in the MDD region or in the rest of the Amazon; deforestation as well as forest burning and clearing e and important regional activity for different purposes including but not limited to ASGM e are responsible for perturbation and erosion of Hg-rich natural surface soils (Roulet et al., 2000; 2001). The currently reported Hg concentrations in the aquatic systems of the MDD region do not agree with estimates based on regional ASGM-Hg losses. Thus, we chose to re-examine the area to characterize the local dispersion of Hg and elucidate the factors that control the distribution and transport of Hg in the system. We investigated riverbank sediments and suspended matter from the

151

Malinowski-Tambopata river system, and compared ASGMaffected river sections with unaffected tributaries. In addition, we investigated atmospheric deposition of ASGM-related Hg on local lakes using down-core lake sediment profiles with the aim of ascertaining the effects of amalgam roasting and related atmospheric pollution on the local aquatic systems. Moreover, we investigated the Hg concentration in local fish species in both rivers and oxbow lakes of the Malinowski-Tambopata river system. 2. Materials and methods 2.1. Study area The study focused on a 228 km stretch of the MalinowskiTambopata river system, which forms the border between the Tambopata National Reserve (TNR) and its buffer zone (a managed forest used for agriculture, farming, ecotourism and forestry; Fig. 1). The studied river system flows into the heavily ASGM-affected MDD River. The three largest ASGM areas in the MDD region e Huepetuhe, Delta-1, and Guacamayo e are located on the Colorado and Inambari rivers, two major tributaries of the MDD River (Asner et al., 2013). Our studied river system is located south and east from those three ASGM areas. Annual precipitation ranges from 1600 to 2400 mm, with July being the driest (~30 mm) month and January being the wettest month (~260 mm). Annual average air temperature is 26  C (10e38  C), and the predominant wind direction is from northeast (Sernanp, 2011). Both Malinowski and Tambopata rivers have a high load of suspended material (SI-Fig. 1). In Tambopata River, discharge is reported to be between 1132 and 2313 m3 s1 (MINAG, 2010). Annual average values for total suspended solids (TSS) in the Malinowski vary from 10 to 40 mg L1 (Chang, 1998), but concentrations as high as 3342 mg L1 have been recorded after a rain event (Barbieri, 2004). Such extreme values are likely a consequence of siltation caused by the intense bank erosion and sediment dredging related to ASGM activities along the Malinowski River. The Malinowski and Tambopata River lower section (i.e., downstream the confluence with the Malinowski) have been affected by limited ASGM activities in the past, i.e., prior to 1990 (Damonte et al., 2013). However; since ~2010, ASGM has drastically expanded with the easier accessibility after paving the Interoceanic Highway. Today, intense ASGM activities are confined to the Malinowski River, while in the Tambopata River lower section it is restricted to just downstream of the confluence with the Malinowski and close to Puerto Maldonado, capital city of the MDD region. Most of the regional gold shops, where amalgam roasting takes place, are located in Puerto Maldonado. Hereinafter, we define ASGM-affected based on if ASGM occurs along the river stretch, not based on upstream activities. 2.2. Sample collection Field work was carried out between July and September (dry season) in 2012 and 2013. In 2012, sampling was also extended into the heavily ASGM-affected MDD River (Fig. 1). In 2013, the middle and upper sections of the Malinowski, as well as its tributaries, i.e., Manuani, Azul, Malinowquillo, and Quebrada Yarinal, could not be accessed due to ongoing protests from ASGM miners that made the area unsecure at the time of the field campaign. Sampling within the TNR was approved by research permits SERNANP-JEG 036e2011 and 013e2013. Riverbank sediments were collected from the top 10 cm on each side of the river using a plastic shovel in 18 locations (Fig. 1), and total suspended solids in river water were sampled in triplicate at eight locations by filtering known volumes of river water

152

M. Moreno-Brush et al. / Environmental Pollution 218 (2016) 150e159

Fig. 1. Sampling locations along the Tambopata-Malinowski river system (flowing eastward). The extents of the Tambopata National Reserve and Bahuaja-Sonene National Park are marked in green. Riverine locations are indicated with squares or diamonds, while circles represent lakes. A black outline indicate sediment sampling (riverbank sediments in riverine systems and down-core sediment profiles in lakes), while a white outline indicates that no sediment samples were collected. Blue colour indicates that a fishing effort was made, while brown indicates no fishing effort. The diamonds are used to indicate locations where total suspended solids of river water were sampled. The area where ASGMactivities occur is indicated by the hatched area, and the black line represents the Interoceanic Highway. The map is partially based on material provided by the Servicio Nacio nal de Areas Protegidas por el Estado (Sernanp). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

(150e500 mL) through 0.45-mm cellulose acetate filters. To assess temporal variations in Hg deposition, a short ~30-cm-long sediment profile, thereafter sub-sampled at 1-cm resolution, was retrieved from the centre of oxbow lake Cocococha using an Uwitec gravity corer (www.uwitec.com). Cocococha is located within the primary forest of the TNR ~20 km downwind from Puerto Maldonado where extensive amalgam burning takes place (Fraser, 2009) (see location in Fig. 1). It is a shallow (3.5e4 m maximum depth), slightly acidic (pH 6.5e7), well oxygenated (dissolved oxygen ~5e8 mg L1) lake, with warm surface waters (~28  C), low conductivity (20e30 mS cm1), and with no connection to the Tambopata River. The spatial distribution of mercury in Cocococha was assessed by additional surface sediment samples (0e2, 2e4, and 4e6 cm) collected along linear transects perpendicular to the shoreline (11 sampling points). Fish were caught in both the Malinowski-Tambopata river system and four oxbow lakes, i.e., Cocococha, Tres Chimbadas, Sachavacayoc, and Valencia (Fig. 1), using either gillnetting or a hook and line. Fish species were determined using identification charts prepared by Araújo-Flores (2015). The weight and length of each specimen were recorded, and bone free dorsal muscle tissue was sampled using a stainless steel knife. 2.3. Sample treatment All samples were put in Whirlpak bags, Tub-Ex Rilsan bags, or amber-glass jars with PTFE-lined caps (previously washed with Milli-Q water and concentrated HNO3), and were kept dark and cold (~4  C) in the field. All collected samples were first shipped to lica del Perú (PUCP) in Lima (Peru), Pontificia Universidad Cato €t (TU) frozen, repacked, and then sent to Technische Univerista Braunschweig (Germany) for preparation and analysis. Riverbank

sediments were oven dried <40  C, while all other samples were freeze dried. After drying, all riverbank samples were fractionated into >90-, 90-63-, and <63-mm fractions using plastic sieves to differentiate geochemically active (Hg binding/adsorption) grain fractions. In addition, the <63-mm fraction was further divided into 63-20- and <20-mm fractions by wet sieving one sample per location. The freeze dried suspended matter on the filters was washed into centrifuge tubes with Milli-Q water, deep frozen, and freeze dried again. The dry samples were microwave digested in bi-distilled concentrated HNO3 for 1 h in closed vessels (Mars 5 Microwave System, CEM, USA). To assure sufficient sample amounts for the analyses, triplicates were pooled per location (except from Malinowski) before the analysis. 2.4. Chemical analysis 2.4.1. Sediments, soil, and fish The mercury (Hg) content in sediments, soil, and fish was analysed by thermal decomposition followed by pre-concentration of Hg on a gold trap and cold vapour atomic absorption spectrophotometry (CV-AAS) Hg detection using a DMA-80 direct mercury analyser (Milestone, USA; EPA Method 7473 (EPA, 1998)). Hg concentrations in dry sediments and soils are reported per dry weight (dw), and Hg concentrations in fish are reported per wet weight (ww). Total concentrations of Titanium (Ti) and Zirconium (Zr) e lithogenic elements proxies for catchment erosion e in lake sediments were analysed by energy-dispersive X-ray fluorescence (mED-XRF) according to the method of Cheburkin and Shotyk (1996, 2005). Content of organic matter (OM) e the most known Hg sorbent e was determined in the <63-mm riverbank sediment fractions and soils as loss-on-ignition at 550  C (LOI550). In brief, 0.4 g of

M. Moreno-Brush et al. / Environmental Pollution 218 (2016) 150e159

sample was oven-dried at 105  C for 2 h and then ashed at 550  C for 2 h. For lake sediment samples, total carbon (C) and nitrogen (N) were analysed using a EuroEA 3000 elemental analyser (Hekatech GmbH, Germany). The C:N ratio is an indicator of organic matter sources (Meyers and Ishiwatari, 1993). Carbonates were assumed negligible due to highly weathered soils in this area and acidic water pH (6.5e6.9). This assumption is supported by the low carbonate content (LOI950 <2.5%) in lake sediment samples. Hence, total C was assumed to equal the total organic C, and this value was multiplied by a factor of two to obtain the OM content (Pribyl, 2010). C:N ratios are reported on a molar basis.

153

test was used to test the normal distribution of the data. Bi-variate correlation coefficients (r) between geological samples parameters (Hg, concentrations, lithogenic element concentrations, OM content, and percentage of fine-grained material) were calculated as 2tailed Spearman correlations (p). Median Hg concentrations in fish were compared using a Kruskal-Wallis ANOVA test, whereas median Hg concentrations in non-piscivore and piscivore fish species were compared using a Mann-Whitney test. A significance level of 0.05 was used for all statistical analysis. Values are reported as the median ± one SD unless otherwise denoted. 3. Results

2.4.2. Lake sediment core dating Freeze-dried samples (n ¼ 9) from the Cocococha sediment profile were analysed for Lead-210 (Pb-210) and Radium 226 (Ra226) activity at Flett Research Ltd (Winnipeg Manitoba, Canada). Based on the unsupported Pb-210 activity (derived by subtracting the Ra-226 activity from the total Pb-210 activity) the age of each sediment section was modelled by assuming a constant rate of supply of Pb-210 according to Appleby et al. (1979) (SI-Fig. 2). This model gives sediment accumulation rates ranging from 0.023 to 0.074 g cm2$yr1, with an average of 0.050 g cm2$yr1. It should be noted that the two upper samples (0e1 and 3e4 cm) had a relatively similar Pb-210 activity, which indicates that the surface sediments have been mixed (SI-Fig. 2a). The slightly higher accumulation rates at the surface should be an artefact of the sediment mixing. The fact that sediment mixing tends to introduce errors in the chronology (Appleby, 2008) is not a large issue in the present study because the purpose of the dating was only to establish that the sediment profile from Cocococha pre-dates the expansion in ASGM activity in the 1970s. Dry mass sediment accumulation rate (g cm2$yr1) was calculated as the product of the mean bulk density of a given interval (g cm3) and the interval represented by the given section (cm), divided by the time span covered by the interval (yr). The average sedimentation rate was calculated as weighted average (because each section represents different intervals). Hg fluxes at each sediment interval (mg m2$yr1) were calculated as described by Beal et al. (2013, 2014), i.e., as the product of Hg concentrations (ng g1) and sedimentation rate (g cm2$yr1). 2.4.3. Suspended matter Total suspended solids per volume (TSS in mg L1, i.e., >0.45 mm) were calculated by dividing the dry mass on the filter by the volume of the filtered river water. Digested TSS samples were analysed for Hg content by CV-AAS (STM, Germany), from which the dry mass TSS Hg concentration (TSS-Hg, ng g1) was calculated. TSS Hg concentrations per litre of water (TSS-HgL, ng L1) were then calculated as the product of TSS-Hg (ng g1) and TSS (mg L1). 2.4.4. Analytical accuracy and reproducibility In all analyses, certified reference materials (CRMs) were included along with the samples to assess the accuracy and precision (SI-Table 1). For m-ED-XRF analyses, further details on accuracy and precision can be found in Cheburkin and Shotyk (1996, 2005). Sample replicates from the m-ED-XRF analyses were within 3% RSD for Zr (n ¼ 6) and within 10% RSD for Ti (n ¼ 9). All sample replicates from the DMA analyses (fish n ¼ 22, sediments and soils n ¼ 51), CNS analyses (n ¼ 4), and CVAAS analyses (n ¼ 6) were within ±10% RSD. 2.5. Statistical analysis All statistical analyses was performed using OriginPro, version 9.0.0 academic (www.originlab.com). A post hoc Anderson-Darling

3.1. Mercury in the Malinowski-Tambopata river system 3.1.1. Mercury distribution in riverbank sediments Hg concentrations in bulk riverbank sediments from the Malinowski-Tambopata river system (n ¼ 72) showed large variations without any apparent pattern that can be directly linked to ASGM activities (Fig. 2, SI-Table 2). For the sampled ASGM-affected river sections, i.e., Malinowski, Manuani, and Quebrada Yarinal, median Hg concentrations in bulk riverbank sediments were 24 ± 4.3, 29 ± 8, and 9 ± 4.6 ng g1, respectively. In river sections unaffected by ASGM activities, median Hg concentrations were 27 ± 14, 21 ± 0.3, 28 ± 14, 14 ± 10.5 and 31 ± 9.6 ng g1 for Azul, Malinowsquillo, Puesto de Control (PC.) Malinowski, La Torre, and Quebrada Aguas Negras, respectively. Downstream of the Malinowski-Tambopata confluence e i.e., where the waters of the unaffected Tambopata River upper section are mixed with the ASGM-affected Malinowski River e median Hg concentrations in bulk riverbank sediments varied from 37 ± 3.2 ng g1 at Filadelfia, to 53 ± 41 ng g1 at Middle-Tambopata, and finally 30 ± 25 ng g1 at PC. La Torre. In the MDD River, the median Hg concentration in bulk riverbank sediments was 98 ± 20 ng g1 upstream of the MDD-Palma Real confluence (Lower-MDD River), which is about twice as high as the highest median value found in the MalinowskiTambopata river system. However, samples collected downstream of the same confluence (PC. Huisene) had a median Hg concentration of 28 ± 32 ng g1, which is within the range of the Malinowski-Tambopata river system. The bulk riverbank sediments of the unaffected Palma Real River also had median Hg concentrations within the range found in the Malinowski-Tambopata river system (18 ± 16 ng g1). When fractionating the riverbank sediments into different particle size fractions, the <63-mm (n ¼ 42) fraction had equal or higher Hg concentrations than the bulk samples. In addition, the <20-mm fraction (n ¼ 17) had equal or higher concentrations than the <63-mm fraction (Fig. 2). The low median Hg concentration in bulk samples from the ASGM-affected tributary Manuani (9 ± 4.6 ng g1) is likely due to the high content of coarser grain sizes, i.e., ~90% sand (SI-Table 2). When excluding the tributaries (e.g., Azul, La Torre, Quebrada Aguas Negras), there is a general increase in the Hg concentrations for the <20-mm fraction when moving downstream from the Malinowski River (37e66 ng g1), to the Tambopata River lower section (77e181 ng g1) and out into the MDD River (179e597 ng g1). However, the unaffected tributary La Torre has Hg concentrations in the <20-mm fraction that are almost as high as the ASGM affected MDD River (517 ng g1). Similarly, the unaffected tributary Quebrada Aguas Negras has Hg concentrations in the <63-mm fraction (92 ± 63 ng g1) that are comparable to the lower section of the Tambopata River. The percentage of material in the <63-mm fractions varied from 5 to 82% and in the <20-mm fractions from 3 to 54%. The LOI550 for the <63-mm fractions were in the range of 3e4% in all samples (SITable 2). Neither the grain size distribution nor LOI550 showed any

154

M. Moreno-Brush et al. / Environmental Pollution 218 (2016) 150e159

Fig. 2. Mercury (Hg) concentrations in riverbank sediments from the Malinowski-Tambopata river system and the Madre de Dios River (MDD River). Data are shown from bulk (grey boxes) and fractionated riverbank samples (<63 mm in white boxes and <20 mm in circles) from unaffected and ASGM-affected river stretches (see locations in Fig. 1). Whiskers indicate outlier values. ASGM-affected areas are defined based on whetherASGM occurs along the river stretch, not based on the upstream activities.

systematic variations among the different river sections or any relationship with the ASGM-activities (SI-Table 2). 3.1.2. Mercury in suspended matter 3.1.2.1. Total suspended solids. Total suspended solids (TSS) in the ASGM-affected Malinowski River (751 mg L1) were much higher than in the Tambopata River lower section (median 63 mg L1), the Tambopata River upper section (73 and 19 mg L1 in UpperTambopata and PC. Malinowski, respectively), and in the tributary rivers La Torre (34 mg L1) and Quebrada Aguas Negras (16 mg L1). 3.1.2.2. Hg concentration in TSS (TSS-Hg). TSS-Hg in the ASGMaffected Malinowski was 2045 ng g1, while in the unaffected Tambopata upper section, it was 1148 and 1581 ng g1 at UpperTambopata and PC. Malinowski, respectively. In the tributaries La Torre and Quebrada Aguas Negras, TSS-Hg was 3488 and 2408 ng g1, respectively, while TSS-Hg levels in the locations in the lower section of the Tambopata River were 429, 3488, and 4019 ng g1 at Filadelfia, PC. La Torre, and Lower-Tambopata, respectively. When compared to the riverbank sediments, the median TSS-Hg in the lower section of the Tambopata River was ~50-fold higher than the median Hg concentration in the <63-mm fraction, and ~30-fold higher than the median Hg concentration in the <20-mm fraction. 3.1.2.3. Hg concentration in river water (TSS-HgL). TSS-HgL in the ASGM affected Malinowski River (1391 ng L1) was ~23-fold higher than the median TSS-HgL in the tributaries of the Tambopata River: Tambopata upper section (83 and 30 ng L1, in Upper-Tambopata and PC. Malinowski, respectively), La Torre (119 ng L1), and Quebrada Aguas Negras (38 ng L1). In the Tambopata lower section, TSS-HgL levels were 28, 126, and 254 ng L1 in Filadelfia, PC. La Torre, and Lower-Tambopata, respectively. 3.2. Mercury in lake sediments Hg content in surface sediments, i.e., 0e6 cm sediment depth, from Cocococha ranged from 66 to 165 ng g1, with a whole-lake median of 91 ng g1 (Table 1). Sediment Hg concentrations increased by a 2.5-fold from the easternmost to the westernmost sampling location. OM varied from 6 to 35% (median 15%) and also showed a westward increase, while elements associated with silicate minerals, i.e., Ti and Zr, decreased towards the western end of the lake. C:N ratios ranged from 9.9 to 18.6, with a whole-lake median of 15.6 (Table 1). The lowest C:N ratio was found in the

centre of Cocococha, while the highest was found at the eastern end of the lake. In the Cocococha sediment core, Hg concentrations range from 64 to 86 ng g1 (median 72 ng g1), without any increasing trend towards the sediment surface (Fig. 3). OM content ranged from 4 to 15% (median 10.4), also without any clear trend. C:N ratios varied between 9.5 and 11.4, with a median value of 10.3. Based on modelling of the unsupported Pb-210 activities (SIFig. 2), a depth of 22 cm (the last sediment section with unsupported Pb-210) correspond to ~102 years and a depth of 14 cm corresponds to ~39 years, i.e., approximately A.D. 1970 when ASGM activities started to increase in the MDD region (SI-Fig. 2b). Sediment accumulation rates ranged from 0.023 to 0.074 g cm2$yr1 (mean 0.050 g cm2$yr1). The slightly higher accumulation rates at the surface should be an artefact of the sediment mixing as previously described (SI-Fig. 2a). An increase in the sediment accumulation rates occurs in ~A.D. 1974, followed by relatively stable values (~0.070 g cm2$yr1) until the most recent sediment in A.D. 2013. Hg fluxes ranged from 15.2 to 54.4 mg m2$yr1, with an average of 38.0 mg m2$yr1 (Fig. 3). 3.3. Mercury in fish A total of 192 fish specimens from 24 different species (most of them important local commercial fish species) were sampled and analysed for Hg (SI-Table 3). The majority of fish specimens were caught in either oxbow lakes (i.e., Cocococha, Tres Chimbadas, Sachavacayoc, and Valencia) or unaffected tributaries (i.e., Azul, Quebrada Aguas Negras, Upper-Tambopata, and Palma Real) (see locations in Fig. 1). A smaller number of fish (n ¼ 10) was caught in the ASGM-affected Malinowski River (i.e., Upper-Malinowski and Lower-Malinowski) and Tambopata River (i.e., PC. Malinowski, and Filadelfia). No fish could be caught in the ASGM-affected Quebrada Yarinal and Middle-Malinowski, or in the unaffected La Torre. No fishing effort was made in Manuani, Malinowsquillo, MiddleTambopata, PC. La Torre, Lower-Tambopata or MDD River. Hg concentrations in bone free dorsal fish muscle ranged from 8 to 1215 ng g1 ww (median 182 ng g1). The highest Hg concentration was found in a chambira (a strict piscivore), and the lowest concentration was found in a palometa (an omnivore) caught in the Palma Real River. For 10 specimens (5% of all of the caught fish) e all strict piscivores e the Hg concentration surpassed the World Health Organization (WHO), European Union, and Brazilian Hg health-guidelines for human consumption (i.e., 500 ng g1 ww for non-predatory fish and 1000 ng g1 ww for predatory fish; WHO, 2007; EC, 2006; Amaro et al., 2014). Of the seven fish specimens

M. Moreno-Brush et al. / Environmental Pollution 218 (2016) 150e159

155

Table 1 Concentrations of mercury (Hg), organic matter (OM), and lithogenic elements (Ti and Zi) in bottom sediments (0-6 cm depth) from Cocococha. C:N ratios are also presented. Element

Unit

Whole-lake values Min

Hg OM Ti Zr C:N

ng g-1 % mg kg-1 mg kg-1 molar ratio

Max

West end Median

Transect west Shore N

Center

Transect middle Shore S

Shore N

Center

Transect east Shore S

Shore N

Center

East end Shore S

(n¼11)

(n¼11)

(n¼11)

C11

C10

C9

C8

C7

C1

C2

C5

C4

C3

C6

66 5.7 2381 84 9.9

165 35.1 5951 263 18.6

91 15.4 3913 118 15.6

165 29.4 2906 101 13.2

119 35.1 2381 84 15.0

106 15.4 3810 118 13.8

102 19.2 3301 111 13.9

83 15.6 3913 139 15.6

72 12.7 4504 106 9.9

95 23.8 3181 116 16.0

91 12.2 4568 192 16.5

68 6.9 5951 212 17.1

76 9.9 5759 185 16.0

66 5.7 5900 263 18.6

Fig. 3. Down-core profiles for mercury concentrations, organic matter content, C:N-ratio, lithogenic elements (Ti and Zr) concentrations, and mercury accumulation rates (from left to right). ASGM expanded in the Madre de Dios region in the 1960e1970s.

that were caught in the ASGM-affected Malinowski, two (a chambira, and a pez perro) surpassed the WHO Hg guidelines; however, Hg concentrations in these fish were not higher than those found in specimens of the same species from unaffected river sections. For the nine fish species where more than five specimens were caught (n ¼ 156), piscivores (omnivores and strict piscivores) showed higher (Mann-Whitney Test, p < 0.001) Hg concentrations than non-piscivore species (detritivores and herbivores) (Fig. 4). Because few if any fish specimens were caught in the ASGMaffected river sections, the data do not allow for statistical comparisons between ASGM-affected and unaffected locations. 4. Discussion 4.1. Mercury distribution in the Malinowski-Tambopata river system We found no pattern in Hg concentration variations in either bulk or fractionated (<63 and < 20 mm) riverbank sediment samples that could be linked to ASGM activities. Locations with the highest ASGM activity, i.e., Malinowski and Manuani, show equal or lower concentrations when compared to unaffected locations, i.e., Azul, La Torre, and Quebrada Aguas Negras. The variability of Hg concentrations in bulk riverbank samples is partly explained by variations in the percentage of material in the <63mm fraction (r ¼ 0.50, p ¼ 0.0007), but it is not related to the percentage of material in the <20-mm fraction (r ¼ 0.47, p ¼ 0.06).

There is also a relatively strong correlation between the Hg concentrations in bulk sediments and Hg concentrations in the <63mm and the <20-mm fractions (r ¼ 0.76 and 0.63; p < 0.01, respectively). Hg concentrations in riverbank samples are also found to be weakly correlated with LOI550 (r ¼ 0.37, p ¼ 0.03); nevertheless, LOI550 values < 2e4% might not represent only OM due to dewatering of clay minerals (Heiri et al., 2001; Santisteban et al., 2004). These results suggest that Hg concentrations in bulk sediment samples partially depend on the following: i) the relative amount of fine-grained material at each specific location, ii) the grain size distribution of the finest fraction, i.e., whether the <63or <20-mm-fractions mainly consist of material in the upper or lower range of grain sizes, and iii) on the differences in the amount of Hg that is associated with the mineral material. These factors could, in turn, be related to either differences in the supply of Hg from the atmosphere or geological sources or be a result of the fine particle's chemical composition that could affect its sorption capacity for Hg, e.g., due to the presence of aluminosilicates (Roulet et al., 2000) and hydrous ferric oxides (Richard et al., 2016; Roulet and Lucotte, 1995). The median Hg concentrations in bulk sediments from Upper Malinowski (21 ng g1) and Lower Malinowski (26 ng g1) sampled during the dry season are in line with the average Hg concentration range recently reported by Diringer et al. (2015) for bulk river sediments from the MDD River (9e31 ng g1). However, our values are ~10 times lower than those reported by Barbieri (2004) for the same river sections.

156

M. Moreno-Brush et al. / Environmental Pollution 218 (2016) 150e159

Fig. 4. Mercury (Hg) concentration in dorsal muscle of the nine fish species where more than five specimens were caught (n ¼ 156). Fish were caught in the watersheds of the river system Malinowski-Tambopata and Madre de Dios River (sampling locations in Fig. 1). Populations are significantly different according to a Kruskal-Wallis ANOVA test at the 0.05 significance level. The dotted line indicates the advisory guidelines for human consumption (500 ng of Hg g1) specified by the WHO and the European Union.

4.2. Particulate mercury in the Malinowski-Tambopata river system During the amalgamation process in ASGM, Hg is released into surrounding soils and aquatic systems either as Hg0 or Hg0amalgam. In the aquatic system, Hg0 is relatively unreactive, has low solubility (60 mg L1 at 25  C), and has a high ionization energy, density, and surface tension. Being unreactive and heavy, it is likely that some of the metallic Hg accumulates as droplets close to the amalgamation sites. Such Hg droplets would then be stabilized by mineral particles as they are progressively buried (Dominique et al., 2007). They could also be transported further downstream, especially during high current. In the aquatic environment, Hg0 is likely to be sorbed onto suspended particles, such as hydrous ferric oxides or organic-mineral agglomerates, because of its slow oxidation rate in freshwater environments (Amyot et al., 2005; Meech et al., 1998). Nevertheless, both oxidation and dissolution of Hg0 would be enhanced in oxygenated environments (Eh above 0.4 V) and in presence of dissolved organic matter (DOM) (Meech et al., 1998; Melamed et al., 2000). Manganese oxides would also favour chemical interactions that might lead to the formation of a soluble coating of HgO crystals on the surface of Hg0 droplets, which would enhance the Hg mobility because HgO is three orders of magnitude more soluble than Hg0 (Miller et al., 2015). Different from other aquatic systems, the OM content in the Malinowski-Tambopata system is low and bottom sediments are exclusively oxic due to the high current. Hence, precipitation of mineral or organic Hg sulfides is unlikely. Based on our data, ASGM activities in the Malinowski drastically increase the TSS load in the river water (751 mg TSS L1). However, even if ASGM activities and their associated increase in riverbank erosion are the most important contributors to the temporal variations in TSS loads, variations in river discharge will also affect the TSS load. Tropical rivers, such as Malinowski and Tambopata, are susceptible to changes in river discharge and thus in water levels. These changes, can induce large temporal variability in TSS loads over a short time scale (Thieme et al., 2007). Such variability can be

observed by comparing the TSS load in the Tambopata River upper section. The Upper-Tambopata location (73 mg TSS L1) was sampled immediately following an intense rain event, causing the water level to rise approximately 1 m overnight, while location PC. Malinowski (19 mg TSS L1) was sampled two days later when the water level had dropped again. The reduction in TSS loads as moving downstream from Malinowski into the Tambopata lower section is likely a result of a dilution effect, i.e., the water of the Malinowski mixed with the water of the Tambopata upper section. Within the lower section of the Tambopata River, there was still a large variability in the TSS loads (36e65 mg TSS L1) without any clear trend, likely due to inflowing streams, changes in water level, and variable human activities along the river course. In rivers, the mobility of Hg is related to a large extent to the mobility of the suspended matter (Gabriel and Williamson, 2004; Roulet et al., 2000, 2001). This is also true for the MalinowskiTambopata river system, where the high Hg concentration in the water phase of the ASGM-affected Malinowski River (TSSHgL ¼ 1391 ng L1) is a consequence of the higher TSS load rather than higher Hg concentrations in the TSS (TSS-Hg) of the ASGMaffected rivers compared to the unaffected river sections (1148, 1581, and 2045 ng g1 at the Upper-Malinowski, PC. Malinowski, and Malinowski, respectively). Brack et al. (2011) reported dissolved Hg concentrations of less than 2 ng L1 for these studied rivers. Thus, mobilization and transport of Hg bound to suspended matter in the Malinowski-Tambopata river system, and most likely also in other tropical rivers, is predominantly controlled by hydrological characteristics, erosion of natural soils (highly increased by ASGM activities from vegetation cover removal and soil disturbance), and to a lesser extent by the release of Hg during the ASGM process. Furthermore, the overall TSS-Hg values are ~5e30 times higher than the Hg concentrations in the <20-mm riverbank fractions, which could be taken as an indicator that the TSS consists mainly of small particles that are not retained in the riverbanks. The higher TSS-Hg found in the unaffected Tambopata tributaries, Quebrada Aguas Negras and La Torre (3488 and 2408 ng g1, respectively), could be a response to differences in the TSS grain size, i.e., that TSS grain size in these locations is smaller than in the Tambopata upper section and Malinowski. However, it could also be due to differences in the geochemical composition of the material that affect both the Hg concentrations in the source material and the Hg sorption capacity of the TSS. Additionally, if the surrounding soils are richer in natural Hg accumulation (associated with iron oxides or OM), a greater amount of Hg would be mobilized into the river by runoff or soil erosion after heavy rain or flooding events (Roulet and Lucotte, 1995; Roulet et al., 2001). Comparing our data to previous studies in the area, we find that our dry season TSS value in the Malinowski River (751 mg$TSS L1) is higher than the TSS values reported by Diringer et al. (2015) for the mid-MDD River e where most regional ASGM and deforestation occurs e both in the dry and wet season (228 and 642 mg TSS$L1, respectively). Although our TSS in the Malinowski does not differ greatly from the wet season TSS in the mid-MDD River, our TSS-HgL in the Malinowski River (1391 ng L1) was found to be ~80 times higher than the wet season TSS-HgL values reported by Diringer et al. (2015). Similarly, our TSS-HgL at Lower-Tambopata (254 ng L1), which is close to Puerto Maldonado, is two orders higher than the dry season and wet season TSS-HgL values reported by Diringer et al. (2015) for that same location. 4.3. Atmospheric mercury input to lake sediments 4.3.1. Spatial mercury variations in Cocococha Hg concentrations in surface sediments of Cocococha were similar to those found in the <63-mm riverbank sediment fractions

M. Moreno-Brush et al. / Environmental Pollution 218 (2016) 150e159

from the Tambopata River lower section (medians: 91 and 79 ng g1, respectively). Hg concentrations in bottom sediments from Cocococha are 40-fold lower than TSS-Hg (3446 ng g1), indicating that it is unlikely that TSS from the Tambopata River contributes to the lake (e.g., during flooding). Surface sediments from the western end of the lake have both higher Hg and OM concentrations, whereas samples from the eastern side are richer in minerogenic material and lower in Hg and OM. This spatial pattern is likely an effect of the predominant easterly winds, which promote the transport of light organic material to the western side of the lake. From the high positive correlation between OM and Hg (r ¼ 0.87, p ¼ 0.0004), it is also apparent that the main driver behind the observed spatial variability in Hg concentrations are variations in the OM concentration. This observation is further supported by the significant negative correlation between Hg and Ti and Zr (r ¼ 0.90 and 0.75, p ¼ 0.0001 and 0.008, respectively), indicating that spatial variations in Hg are not associated with the distribution of lithogenic elements. From a whole-lake perspective, the OM in the surface sediments of Cocococha is a mixture of algae and terrestrial OM (median C:N 15.6 ± 2.3) because a C:N ratio from 4 to 10 indicates algae material (in-lake OM production) while terrestrial organic matter generally has a C:N ratio 20 (Meyers and Teranes, 2001). The lower C:N ratio in the centre of the lake basin indicates that the input of terrestrial organic matter is more important in near shore locations and that sediments in the centre of the basin are mostly from algae. The limited input of terrestrial OM, which is an important vector for Hg transport from the catchment to the lake, indicates that the main Hg input to Cocococha is atmospheric deposition. This result, together with the relatively small spatial variability that shows that the centre basin is a representative sampling location (Rydberg et al., 2012), makes Cocococha a suitable lake for tracking changes in Hg atmospheric deposition. 4.3.2. Historical Hg concentrations in Cocococha Hg concentrations in Cocococha are relatively constant along the whole length of the sediment profile. Additionally, the concentrations of OM, Ti, and Zr, as well as the C:N ratios, show minimal variations with time, indicating that the entire record reflects a period of relatively stable conditions without any major shifts in the sources of the sediment material. Hg concentrations were not correlated with either OM, Zr, or Ti (r ¼ 0.27, 0.25, and 0.10; p ¼ 0.1, 0.2, and 0.6, respectively), indicating that down-core Hg variations are independent of lithogenic input from the catchment. This lack of correlation with compounds that have been suggested to be important transport vectors for Hg e i.e., OM (Gabriel and Williamson, 2004; Ravichandran, 2004) and oxyhydroxides, and aluminosilicates (Roulet et al., 2000) e further supports the idea that the down-core record from Cocococha reflects mainly variations in the atmospheric Hg deposition. The Cocococha down-core Hg profile shows relatively low Hg concentrations and no signs of increasing Hg levels during the last 30 years when ASGM expanded in the MDD region. Being Hg concentrations relative stable along the whole sediment core, variations in the Hg flux profile are largely an effect of variations in sediment accumulation rates. This said, the increase in the Hg flux observed in A.D. 1974 (Fig. 3) is unlikely related to the increase of ASGM activities in the region, but responds to an increase of the sediment input to the lake. The same variations are observed when calculating the fluxes of titanium and zirconium. It is not clear why in Cocococha the sedimentation rate increases in A.D. 1974, but it is not related to changes in the sediment composition (i.e., there are no changes in the down-core profiles of OM, C:N-ratio, or lithogenic elements). We presume that changes in seasonal flooding events could have had affluence, but it could also be due to errors in the

157

chronology (sediment mixing). We unfortunately do not have information regarding anthropogenic activities conducted in the surroundings of Cocococha prior to 1990, i.e., when the TambopataCandamo Reserved Zone was created, but they could have had an influence in the increase of sediment input to the lake. 4.4. Mercury bioaccumulation in fish During both sampling campaigns, very few fish specimens were caught in ASGM-affected river sections, i.e., Quebrada Yarinal and Malinowski, which was presumably related to the very high TSS loads in these river sections. Siltation and turbidity have adverse effects on aquatic species susceptible to in-water light climate changes (Mol and Ouboter, 2004), making these river sections largely uninhabitable for many of the fish species that normally inhabit these rivers. Because comparable fish samples in sufficient numbers could not be systematically collected from both nonaffected and ASGM-affected rivers, we regretfully could not make a comparison between the Hg content in fish relative to ASGM activities. We found that Hg enrichment in fish depends mainly on the feeding habits of each species, i.e., lower tissue Hg concentrations were found in non-piscivores than in piscivores (Fig. 4, SI-Table 3). Most of the fish species represented in Fig. 4, i.e., palometa, bocabalo, yahuarachi, huasaco, corvina, and piran ~ a blanca, have chico, sa been previously assessed for Hg concentrations in the Peruvian Amazon. Our overall Hg concentrations in these fish are similar to those previously reported e i.e., below or slightly above the WHO guidelines of 500 ng g1 e in both ASGM-affected and unaffected locations (Barbieri, 2004; CAMEP, 2013; Deza Arroyo, 1996; Diringer et al., 2015; Fernandez and Gonzalez, 2009; Gutleb et al., 2002, 1997; Roach et al., 2013). Both in this and previous studies, Hg concentrations in chambira (a large piscivore species) are above WHO guidelines (in some cases, even >1000 ng g1). This result is typical of many large piscivore species found in both ASGMaffected and unaffected waters, as well as in local fish markets from Puerto Maldonado (Barbieri, 2004; CAMEP, 2013; Deza Arroyo, 1996; Diringer et al., 2015; Fernandez and Gonzalez, 2009; Gutleb et al., 1997; Roach et al., 2013). The considerable variability between sites and the often small sample sizes (1e2 specimens per location) e both in this study and in previously published studies e make it difficult to determine whether ASGM activities have had any significant effect on Hg concentrations in fish. The difficulty in accurately assessing the effect of ASGM-Hg releases also stems from the lack of any data regarding “natural” baseline Hg concentrations for Amazonian fish species. We found no contaminated sediments and suspended matter in our investigated rivers and no sign of a recent increase in Hg concentrations in our studied lake. Thus, we conclude that there is no indication that fish collected in this study are affected by Hg from local ASGM, and these results can be considered to represent modern baseline levels for this region. We speculate that Hg enrichment in the aquatic food web in these rivers is limited because of the lack of fine-grained organic-rich bottom sediments with high Hg concentrations that could act as a source for the Hg and facilitate methylation (Ullrich et al., 2001). However, the transfer of Hg from the organic-rich sediments in oxbow lakes to fish seems to be relatively small. 4.5. Where does the mercury go? Between 1995 and 2007, ASGM in the MDD region emitted over 300 tons of Hg to the atmosphere. From this estimate, approximately 13.2 tons was emitted by ASGM conducted in the Malinowski River watershed (Mosquera et al., 2009). Assuming these

158

M. Moreno-Brush et al. / Environmental Pollution 218 (2016) 150e159

estimations are correct and that atmospheric Hg emissions represent up to 83% of the total Hg losses (Pfeiffer et al., 1993), we estimate that 2.7 tons of Hg was directly released into the Malinowski River over the same period. Although the variable hydrology makes it challenging to assess Hg transport, neither our findings nor previous studies have been able to show geochemical evidence that ASGM-Hg is accumulating in the local river system. Therefore, we suggest that most ASGM-Hg either remains close to the emission source in the form of metallic Hg droplets or is transported downstream e presumably adsorbed to suspended matter e and accumulates in sediment sinks distant from the source. Moreover, our findings do not indicate any substantial increase in atmospheric Hg deposition. This observation could be either because deposition only occurs close to the amalgam roasting sites or because most of the emitted Hg is subject to long range-transport and dispersion. 5. Conclusions Findings from our study of the Malinowski-Tambopata river system demonstrate that Hg concentrations in riverbank sediments are within the range of background values and cannot be linked to the location of ASGM activities or to ASGM-Hg pollution. Riverine Hg transport is predominantly driven by suspended matter loads, which in turn are strongly controlled by hydrological conditions. These findings demonstrate that both Hg distribution and transport are affected by multiple geochemical variables, and it is essential to consider the system hydrology when tracking the final sink of water-borne ASGM-Hg in tropical environments. Furthermore, from the absence of any increase in Hg concentrations in the last decades in the down-core lake sediment record, we assume that atmospheric ASGM-Hg releases are not deposited locally in this area in substantial amounts. Finally, we were unable to compare Hg levels in fish from ASGM-affected and unaffected river stretches because of insufficient fish found in the affected river stretches. However, based on our geochemical findings, we infer that our findings in fish represent typical baseline Hg concentrations for non-impacted locations in the MDD region. Notes The authors declare no competing financial interest. Acknowledgments This research was financially supported by PUCP (Grant No. DGI70242.2015 and DGI-70243.0102), AG Umweltgeochemie e TU €nder-Dienst Braunschweig, and Katholischer Akademischer Ausla  (KAAD). We thank the logistic support of Servicio Nacional de Areas n para la Investigacio n y Protegidas por el Estado (Sernanp), Asociacio Desarrollo Integral (Aider), Frankfurt Zoological Society, La Casa del Curandero, and the touristic lodges Explorer's Inn, Sachavacayoc Center, and Posada Amazonas. We also thank Tub-Ex for kindly donating the Rilsan bags for the sampling. Thanks to Vito Bolivar, Hernan Huallahua, Betty Flores, Fabian Limonchi, and the TNR forest rangers for their assistance in the field. We especially thank Ernesto Fernandez, Boris and Dora Zlatar, and Eric Cosio for their help organizing the field excursions, as well as Julio Araújo Flores n Ortega for their advice on the fishing techniques and and Herna species identification. Three anonymous reviewers provided valuable critiques and helped to enhance this manuscript. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.envpol.2016.08.038.

References Amyot, M., Morel, F.M.M., Ariya, P.A., 2005. Dark oxidation of dissolved and liquid elemental mercury in aquatic environments. Environ. Sci. Technol. 39, 110e114.  n de la Biodiversidad Acua tica en la Cuencia Araújo-Flores, J.M., 2015. Caracterizacio nica de Madre de Dios - Perú (Caracterization of the aquatic Andino-Amazo biodiversity in the Andean-Amazonian Basin of Madre de Dios - Peru). Universidad de Huelva, Dep. de Biología Ambiental y Salud Pública, Huelva - Spain (Thesis ((Ph.D) (in Spanish). Amaro, C.S.O., Rodrigues Junior, D., Silva, M.C.F., Lima, A.A.S., Santos, G.F.S., Pinheiro, M.C.N., 2014. Concentraç~ ao de mercúrio total (Hg-T) em peixes comercializados em diferences períodos sazonais no Mercado do Ver-o-Peso, m, Estado do Para , Brasil (Total mercury (Hg-T) concentration in marketBele m, Para  State, able fish in different seasonal periods at Ver-o-Peso Market, Bele Brazil). Rev. Pan Amaz. Saúde 5, 53e60 (in Portuguese). Appleby, P.G., Oldfield, F., Thompson, R., Huttunen, P., Tolonen, K., 1979. Pb-210 dating of annually laminated lake-sediments from Finland. Nature 280, 53e55. Appleby, P.G., 2008. Three decades of dating recent sediments by fallout radionuclides: a review. Holocene 18, 83e93. Ashe, K., 2012. Elevated mercury concentrations in humans of Madre de Dios, Peru. PLoS One 7, e33305. Asner, G.P., Llactayo, W., Tupayachi, R., Luna, E.R., 2013. Elevated rates of gold mining in the Amazon revealed through high-resolution monitoring. Proc. Natl. Acad. Sci. U. S. A. 110, 18454e18459. n de mercurio total en peces, agua y sedimento en el Río Barbieri, G., 2004. Evaluacio Malinowski, Departmento de Madre de Dios, causada por la minería aurífera aluvial (Assessment of total mercury in fish, water, and sediment from the Malinowski River, Department of Madre de Dios, caused by artisanal and smallscale gold mining). Universidad Nacional Agraria La Molina, Facultad de Pesquería, Lima - Peru (Thesis (Engineer)) (in Spanish). Beal, S.A., Jackson, B.P., Kelly, M.A., Stroup, J.S., Landis, J.D., 2013. Effects of historical and modern mining on mercury deposition in southeastern Peru. Environ. Sci. Technol. 47, 12715e12720. Beal, S.A., Kelly, M.A., Stroup, J.S., Jackson, B.P., Lowell, T.V., Tapia, P.M., 2014. Natural and anthropogenic variations in atmospheric mercury deposition during the Holocene near Quelccaya Ice Cap, Peru. Global Biogeochem. Cycles 28, 437e450. Biester, H., Martinez-Cortizas, A., Birkenstock, S., Kilian, R., 2003. Effect of peat decomposition and mass loss on historic mercury records in peat bogs from patagonia. Environ. Sci. Technol. 37, 32e39. Brack, A., Ipenza, C., Alvarez, J., Sotero, V., 2011. Minería Aurífera en Madre de Dios y n con Mercurio - Una Bomba de Tiempo (Gold mining in Madre Contaminacio de Dios and Mercury Pollution - A time bomb). Ministerio del Ambiente, Lima. ISBN: 978-612-45818-7-8 (in Spanish). CAMEP (Carnegie Amazon Mercury Ecosystem Project), 2013. Mercury in Madre de Dios - Mercury concentrations in fish and humans in Puerto Maldonado. Research Brief #1 [Online]. Standford: CAMEP. https://dge.stanford.edu/ research/CAMEP/CAMEP%20Research%20Brief%20-%20Puerto%20Maldonado% 20English%20-%20FINAL.pdf [21 January 2016]. Chang, F., 1998. Fishes of the tambopata-candamo reserved zone. Southeast. Peru. Rev. Per. Biol. 5, 17e27. Cheburkin, A.K., Shotyk, W., 2005. Energy-dispersive XRF spectrometer for Ti determination (TITAN). X-Ray Spectrom. 34, 69e72. Cheburkin, A.K., Shotyk, W., 1996. An Energy-dispersive Miniprobe Multielemental Analyzer (EMMA) for direct analysis of Pb and other trace elements in peats. Fresenius J. Anal. Chem. 354, 688e691. Conaway, C.H., Swarzenski, P.W., Cohen, A.S., 2012. Recent paleorecords document rising mercury contamination in. Lake Tanganyika. Appl. Geochem 27, 352e359. Cooke, C.A., Balcom, P.H., Kerfoot, C., Abbott, M.B., Wolfe, A.P., 2011. Pre-colombian mercury pollution associated with the smelting of argentiferous ores in the bolivian Andes. Ambio 40, 18e25. Cooke, C.A., Balcom, P.H., Biester, H., Wolfe, A.P., 2009. Over three millennia of mercury pollution in the Peruvian Andes. Proc. Natl. Acad. Sci. U. S. A. 106, 8830e8834. vez Quijada, M., Flores, A., De Damonte, G., Bueno de Mezquita, M., Pachas, V.H., Cha Echave C aceres, J., 2013. Small-scale gold mining and social and environmental conflict in the Peruvian Amazon. In: Cremers, L., Kolen, J., De Theije, M. (Eds.), Small-scale Gold Mining in the Amazon. The Cases of Bolivia, Brazil, Colombia, Peru and Suriname. Centre for Latin American Studies and Documentation, Amsterdam, pp. 68e84. ISBN 978-90-70280-18-5. Deza Arroyo, N.E., 1996. Mercury accumulation in fish from Madre de Dios, a goldmining area in the Amazon Basin, Perú. (Thesis (M.Sc.)). Oregon State University, Dept. of Fisheries and Wildlife, Oregon - U.S.A. Diringer, S.E., Feingold, B.J., Ortiz, E.J., Gallis, J.A., Araújo-Flores, J.M., Berky, A., Pan, W.K.Y., Hsu-Kim, H., 2015. River transport of mercury from artisanal and small-scale gold mining and risks for dietary mercury exposure in Madre de Dios. Peru. Environ. Sci. Process. Impacts 17, 478e487. Dominique, Y., Muresan, B., Duran, R., Richard, S., Boudou, A., 2007. Simulation of the chemical fate and bioavailability of liquid elemental mercury drops from gold mining in amazonian freshwater systems. Environ. Sci. Technol. 41, 7322e7329. EC (European Comission), 2006. No. 1881/2006 of 19 December 2006. Setting maximum levels for certain contaminants in foodstuffs. Off. J. Eur. Union, L 364, 5e24. Engstrom, D.R., Fitzgerald, W.F., Cooke, C.A., Lamborg, C.H., Drevnick, P.E.,

M. Moreno-Brush et al. / Environmental Pollution 218 (2016) 150e159 Swain, E.B., Balogh, S.J., Balcom, P.H., 2014. Atmospheric Hg emissions from preindustrial gold and silver extraction in the Americas: a reevaluation from lake-sediment archives. Environ. Sci. Technol. 48, 6533e6543. EPA (U.S. Environmental Protection Agengy), 1998. Method 7473 (SW-846): Mercury in Solids and Solutions by Thermal Decomposition, Amalgmation, and Atomic Absorption Spectrophotometry. U.S. EPA, Washington, DC. Fernandez, L.E., Gonzalez, V.H., 2009. Niveles de mercurio en peces de Madre de Dios (Mercury levels in fish from Madre de Dios). Carnegie Institution for Science [Online PowerPoint presentation]. (in Spanish). http://www.minam.gob. pe/mineriailegal/wp-content/uploads/sites/43/2013/10/Carnegie-mercurioMadre-de-Dios.pdf [26 November 2015]. Fraser, B., 2009. Peruvian gold rush threatens health and the environment. Environ. Sci. Technol. 43, 7162e7164. Gabriel, M.C., Williamson, D.G., 2004. Principal biogeochemical factors affecting the speciation and transport of mercury through the terrestrial environment. Environ. Geochem. Health 26, 421e434. Gammons, C.H., Slotton, D.G., Gerbrandt, B., Weight, W., Young, C.A., McNearny, R.L., mac, E., Caldero n, R., Tapia, H., 2006. Mercury concentrations of fish, river Ca water, and sediment in the Río Ramis-Lake Titicaca watershed. Peru. Sci. Total Environ. 368, 637e648. Gutleb, A.C., Helsberg, A., Mitchell, C., 2002. Heavy metal concentrations in fish from a pristine rainforest valley in Peru: a baseline study before the start of oildrilling activities. Bull. Environ. Contam. Toxicol. 69, 523e529. Gutleb, A.C., Schenck, C., Staib, E., 1997. Giant otter (Pteronura brasiliensis) at risk? total mercury and methylmercury levels in fish and otter scats, Peru. Ambio 26, 511e514. Hermanns, Y.-M., Biester, H., 2013. Anthropogenic mercury signals in lake sediments from southernmost Patagonia, Chile. Sci. Total Environ. 445e446, 126e135. Heiri, O., Lotter, A.F., Lemcke, G., 2001. Loss on ignition as a method for estimating organic and carbonate content in sediments: reproducibility and comparability of results. J. Paleolim 25, 101e110. Kuramoto, J.R., 2002. La Minería artesanal e informal en el Perú (Artisanal and informal mining in Peru). Grupo de An alisis para el Desarrollo (GRADE) for Mining, Minerals, and Sustainable Development (MMSD), London (in Spanish). http://pubs.iied.org/pdfs/G00731.pdf [21 January 2016]. Lamborg, C.H., Fitzgerald, W.F., Damman, A.W.H., Benoit, J.M., Balcom, P.H., Engstrom, D.R., 2002. Modern and historic atmospheric mercury fluxes in both hemispheres: global and regional mercury cycling implications. Glob. Biogeochem. Cycles 16, 1104. Lechler, P.J., Miller, J.R., Lacerda, L.D., Vinson, D., Bonzongo, J.C., Lyons, W.B., Warwick, J.J., 2000. Elevated mercury concentrations in soils, sediments, water, and fish of the Madeira River basin, Brazilian Amazon: a function of natural enrichments? Sci. Total Environ. 260, 87e96. Malm, O., 1998. Gold mining as a source of mercury exposure in the Brazilian Amazon. Environ. Res. 77, 73e78. Meech, J.A., Veiga, M.M., Tromans, D., 1998. Reactivity of mercury from gold mining activities in darkwater ecosystems. Ambio 27, 92e98. ^ as, R.C., 2000. The effect of humic acid on Melamed, R., Trigueiro, F.E., Villas Bo mercury solubility and complexation. Appl. Organomet. Chem. 14, 473e476. Meyers, P.A., Ishiwatari, R., 1993. Lacustrine organic geochemistry - an overview of indicators of organic matter sources and diagenesis in lake sediments. Org. Geochem. 20, 867e900. Meyers, P.A., Teranes, J.L., 2001. Sediment organic matter. In: Last, W.M., Smol, J.P. (Eds.), Tracking Environmental Changes Using Lake Sediments e Volume II: Physical and Chemical Techniques. Kluwer, Dordrecht, pp. 239e269.  stico Hidrolo  gico de la MINAG (Ministerio de Agricultura), 2010. Estudio Diagno Cuenca Madre de Dios (Hydrological Diagnostic Study of the Madre de Dios Basin). Ministerio de Agricultura, Lima (in Spanish). http://www.ana.gob.pe/ sites/default/files/publication/files/diagnostico_hidrologico_-madre_de_dios_0. pdf [21 January 2016]. Miller, C.L., Watson, D.B., Lester, B.P., Howe, J.Y., Phillips, D.H., He, F., Liang, L., Pierce, E.M., 2015. formation of soluble mercury oxide coatings: transformation of elemental mercury in soils. Environ. Sci. Technol. 49, 120105e212111. Mol, J.H., Ouboter, P.E., 2004. Downstream effects of erosion from small-scale gold mining on the instream habitat and fish community of a small neotropical rainforest stream. Conserv. Biol. 18, 201e214. vez, M.L., Pachas, V.H., Moschella, P., 2009. Estudio diagno  stico de Mosquera, C., Cha

159

la actividad minera artesanal en Madre de Dios (Diagnostic Study of artisanal  n Conservacio n Internacional, mining activities in Madre de Dios). Fundacio Lima. ISBN: 978-9972-865-07-7(in Spanish). Nriagu, J.O., 1994. Mercury pollution from the past mining of gold and silver in the. Am. Sci. Total Environ. 149, 167e181. Nriagu, J.O., 1993. Legacy of mercury pollution. Nature 363, 589. Ouboter, P.E., Landburg, G.A., Quik, J.H.M., Mol, J.H.A., Van Der Lugt, F., 2012. Mercury levels in pristine and gold mining impacted aquatic ecosystems of Suriname, south America. Ambio 41, 873e882. Pfeiffer, W.C., de Lacerda, L.D., 1988. Mercury inputs into the Amazon region, Brazil. Environ. Technol. Lett. 9, 325e330. Pfeiffer, W.C., Lacerda, L.D., Salomons, W., Malm, O., 1993. Environmental fate of mercury from gold mining in the Brazilian Amazon. Environ. Rev. 1, 26e37. Pribyl, D.W., 2010. A critical review of the conventional SOC to SOM conversion factor. Geoderma 156, 75e83. Ravichandran, M., 2004. Interactions between mercury and dissolved organic matterea review. Chemosphere 55, 319e331. Richard, J.-H., Bischoff, C., Ahrens, C.G.M., Biester, H., 2016. Mercury (II) reduction and co-precipitation of metallic mercury on hydrous ferric oxide in contaminated groundwater. Sci. Total Environ. 539, 36e44. Roach, K.A., Jacobsen, N.F., Fiorello, C.V., Stronza, A., Winemiller, K.O., 2013. Gold mining and mercury bioaccumulation in a floodplain lake and main channel of the Tambopata River, Perú. J. Environ. Prot. Irvine., Calif. 4, 51e60. Roulet, M., Lucotte, M., 1995. Geochemistry of mercury in pristine and flooded ferralitic soils of a tropical rain forest in French Guiana, South America. Water. Air. Soil Pollut. 80, 1079e1088. ~es, J.R.D., Roulet, M., Lucotte, M., Canuel, R., Farella, N., Courcelles, M., Guimara Mergler, D., Amorim, M., 2000. Increase in mercury contamination recorded in lacustrine sediments following deforestation in the central. Amaz. Chem. Geol. 165, 243e266. Roulet, M., Lucotte, M., Canuel, R., Farella, N., De Freitos Goch, Y.G., Pacheco Peleja, J.R., Guimaraes, J.-R.D., Mergler, D., Amorim, M., 2001. Spatio-Temporal  s and Amazon rivers, Brazil. geochemistry of mercury in waters of the Tapajo Limnol. Oceanogr. 46, 1141e1157. n, P., Lambertsson, L., De Vleeschouwer, F., Tomasdotter, S., Rydberg, J., Rose Bindler, R., 2012. Assessment of the spatial distributions of total- and methylmercury and their relationship to sediment geochemistry from a whole-lake perspective. J. Geophys. Res. 117, G04005. pez-Pamo, E., Dabrio, C.J., Ruiz Zapata, M.B., Gil Santisteban, J.I., Mediavilla, R., Lo ~ o, S., Martínez-Alfaro, P.E., 2004. Loss on ignition: a qualiGarcía, M.J., Castan tative or quantitative method for organic matter and carbonate mineral content in sediments? J. Paleolimnol. 32, 287.  Sernanp (Servicio Nacional de Areas Protegidas por el Estado), 2011. Reserva Nacional Tambopata, Plan Maestro 2011-2016 (Tambopata National Reserve,  Management Plan 2011-2016). Servicio Nacional de Areas Protegidas por el Estado, Lima. ISBN: 978-612-46157-7-1 (in Spanish). Streets, D.G., Devane, M.K., Lu, Z., Bond, T.C., Sunderland, E.M., Jacob, D.J., 2011. Alltime releases of mercury to the atmosphere from human activities. Environ. Sci. Technol. 45, 10485e10491. Telmer, K.H., Veiga, M.M., 2009. Mercury fate and transport in the global atmosphere. In: Mason, R., Pirrone, N. (Eds.), Mercury Fate and Transport in the Global Atmosphere: Emissions, Measurements and Models. Springer, US, Boston, MA, pp. 131e172. Thieme, M., Lehner, B., Abell, R., Hamilton, S.K., Kellndorfer, J., Powell, G., Riveros, J.C., 2007. Freshwater conservation planning in data-poor areas: an example from a remote Amazonian basin (Madre de Dios River, Peru and Bolivia). Biol. Conserv. 135, 500e517. Ullrich, S.M., Tanton, T.W., Abdrashitova, S.A., 2001. Mercury in the aquatic environment: a review of factors affecting methylation. Crit. Rev. Environ. Sci. Technol 31, 241e293. UNEP (United Nations Environment Programme), 2013. Global Mercury Assessment 2013: Sources, Emissions, Releases, and Environmental Transport. UNEP Chemicals Branch, Geneva. Wade, L., 2013. Gold's dark side. Science 341, 1448e1449. WHO (World Health Organization), 2007. Evaluation of Certain Food Additives and Contaminants. Sixty-seventh Report of the Joint FAO/WHO Expert Committee on Food Additives; WHO Technical Report Series, No. 940. WHO, Geneva. ISBN: 92-4-120940-2.