Environmental and Anthropogenic Factors Influencing MercuryDynamics During the Past Century in Floodplain Lakesof the Tapajos River, Brazilian Amazon

Jordan Sky Oestreicher1,2 • Marc Lucotte2,3 • Matthieu Moingt2,3 •

Emilie Belanger2 • Christine Rozon2 • Robert Davidson4 • Frederic Mertens1 •

Christina A. Romana5,6

Received: 7 June 2016 / Accepted: 2 November 2016

� Springer Science+Business Media New York 2016

Abstract In the Tapajos River region of the Brazilian

Amazon, mercury (Hg) is a prevalent contaminant in the

aquatic ecosystem. Few studies have used comprehensive

chronological analyses to examine the combined effects of

environmental and anthropogenic factors on Hg accumu-

lation in sediments. Total mercury (THg) content was

measured in sediments from eight floodplain lakes andPb210 isotope analysis was used to develop a timeline of

THg accumulation. Secondary data representing environ-

mental and anthropogenic factors were analyzed using geo-

spatial analyses. These include land-cover change,

hydrometeorological time-series data, lake morphology,

and watershed biophysical characteristics. The results

indicate that THg accumulation and sedimentation rates

have increased significantly at the surface of most sediment

cores, sometimes doubling since the 1970s. Human-driven

land-cover changes in the watershed correspond closely to

these shifts. Tropical deforestation enhances erosion,

thereby mobilizing the heavy metal that naturally occurs in

soils. Environmental factors also contribute to increased

THg content in lacustrine sediments. Climate shifts since

the 1980s are further compounding erosion and THg

accumulation in surface sediments. Furthermore, variations

in topography, soil types, and the level of hydrological

connectivity between lakes and the river explain observed

variations in THg fluxes and sedimentation. Although

connectivity naturally varies among sampled lakes, defor-

estation of sensitive floodplain vegetation has changed

lake–river hydrology in several sites. In conclusion, the

results point to a combination of anthropogenic and envi-

ronmental factors as determinants of increased THg accu-

mulation in tropical floodplain sediments in the Tapajos

region.

Human mercury (Hg) intoxication in the Tapajos River

region (Fig. 1) is among the highest in the world (Berzas

Nevado et al. 2010; Mergler et al. 2007), attributed to the

frequent consumption of Hg-loaded fish (Dolbec et al.

2001; Lebel et al. 1997; Passos et al. 2007). In the past

decades, researchers have sought to identify the sources of

Hg in the aquatic ecosystem (Berzas Nevado et al. 2010;

Hacon et al. 2008). Indeed, there are gold mining opera-

tions in the region that use Hg as an amalgam (Fig. 1)

(Boas 1997; Malm 1998). However, the metal is also a

naturally occurring element in the region’s soils (Grimaldi

et al. 2008; Roulet et al. 1998b). Studies have found that

deforestation and agricultural activities accelerate soil

erosion and leaching of Hg (Farella et al. 2006; Roulet

et al. 2000). Bound to clay minerals and organic matter, the

metal is then transported to the floodplain and is deposited

in lake and river sediments (Lacerda et al. 2012; Maurice-

Bourgoin et al. 2002; Roulet et al. 1998b).

Much research in the Amazon has focused on recent

anthropogenic change (e.g., agriculture, deforestation,

& Jordan Sky Oestreicher

jsoestreicher@gmail.com

1 Centro de Desenvolvimento Sustentavel (CDS), Universidade

de Brasılia, Campus Universitario Darcy Ribeiro, Gleba A,

Asa Norte, Brasılia, DF 70910-900, Brazil

2 Institut des sciences de l’environnement (ISE), Universite du

Quebec a Montreal, Montreal, Canada

3 Centre de recherche GEOTOP, Universite du Quebec a

Montreal, Montreal, Canada

4 Biodome de Montreal, Montreal, Canada

5 Centro de Desenvolvimento Sustentavel (CDS), Universidade

de Brasılia, Brasılia, Brazil

6 Universite Paris Descartes, Paris, France

123

Arch Environ Contam Toxicol

DOI 10.1007/s00244-016-0325-1

mining) and its impact on Hg inputs to the aquatic

ecosystem (Almeida et al. 2005; Lacerda et al. 2012;

Roulet et al. 1999, 2000 Sampaio da Silva et al. 2009).

Comparatively few studies have considered Hg dynamics

before the onset of these activities or in undisturbed

ecosystems—except for a few notable works (Laperche

et al. 2014; Maurice-Bourgoin et al. 2002). In fact, there is

a lack of substantial longitudinal research to document how

terrestrial Hg transfers have changed overtime. Moreover,

little attention has been given to natural variation in envi-

ronmental factors, such as soil type, topography, watershed

size, and hydroclimatic cycling, even though they explain

Hg transfer in other regions (Babiarz et al. 1998; Hurley

et al. 1995; Maia et al. 2009; Moingt et al. 2013).

To address these gaps, this paper presents a longitudinal

study of total Hg (THg) loadings in sediments sampled in

eight floodplain lakes in the Tapajos River region. To

examine the combined influence of human activities and

Fig. 1 Tapajos River region and basins selected for this study. The river flows towards the Amazon at the estuary near Santarem

Arch Environ Contam Toxicol

123

environmental factors, THg profiles are analyzed in rela-

tion to hydrological and climate variables, watershed bio-

physical characteristics, lake morphology, and land-cover

change.

This unique study combines geo-spatial analyses with

the first substantial geo-chronological dating of multiple

sediment cores in the region. Lead-210 (Pb210) isotope

techniques are used to establish a timeline of THg accu-

mulation. The system can be compared before and after

anthropogenic activities and the impact of climate events

can be examined (Dearing 2008; Dearing et al. 2010).

Moreover, with multiple sampling sites, highly degraded

watersheds are compared with a pristine forested site.

Materials and Methods

Study Area: Tapajos River Region

The Tapajos River (Fig. 1) is a large tributary of the

Amazon River. It is a floodplain system characterized by a

strong seasonal flood pulse (Junk et al. 1989). Riverbanks

and shoreline lakes are inundated during the rainy season,

from December to May, when there is up to 300 mm of

precipitation per month (INMET 2012). During the

remainder of the year, water levels decrease and lakes are

partially or completely desiccated.

Shoreline lakes can be classified into two morphological

types, building on (Sippel et al. 1992). Lateral-levee lakes

(dish and channel lakes) are elliptical in shape and fed

directly from advection of the river. As such, there is a

short hydrological residency time, currents are at times

lotic, and limnology is primarily influenced by fluvial

forces. These lakes are considered to have a high connec-

tivity with the main river channel (Amoros and Bornette

2002). Blocked-valley lakes, on the other hand, are den-

dritic in shape and are usually linked to the river by canals.

These lakes generally exhibit weak currents (lentic), are

less influenced by the river, and have a higher residency

time. As such, these lakes have a lower connectivity with

the main river channel.

Regional vegetation is predominantly dry upland forest

(EMBRAPA 2007); however, crop and pasture cover has

increased in the past 40 years with agrarian development

and urbanization (Rozon et al. 2015). Riparian zones and

lake outlets are typically occupied by two types of vege-

tation: (1) diverse alluvial forest stands (igapos) that reduce

the flux of water and contaminants during the rainy season

(EMBRAPA 2007; Martinez and Le Toan 2007), or (2)

macrophytes: semi-aquatic, herbaceous reed beds that form

a permeable boundary during flooding (Dunne et al. 1998;

Junk 1997; Tockner et al. 2010). As such, the type of

floodplain vegetation growing at the lake–river boundary

also determines the level of connectivity (Amoros and

Bornette 2002).

Watershed Selection and Characterization

Eight floodplain lakes were selected for sediment sampling

(Fig. 1) according to the following criteria: (1) the extent

of deforestation in the watershed, (2) lake morphology and

connectivity, and (3) the logistical feasibility of sampling.

The geographic limits of each watershed was delimited

using GRASS—an open-source GIS platform (GRASS

Development Team 2012)—and a 30-m digital elevation

model (DEM) (Valeriano 2004) reconstituted from the

Shuttle Radar Topography Mission (SRTM) 90-m resolu-

tion model (USGS 2004).

Morphological and biophysical characteristics of

watersheds were then measured using the DEM and two

satellite images: a 2009 Landsat 7 TM image (available

from the U.S. Geological Survey: www.eros.usgs.gov) and

a 2008 SPOT image (graciously provided by the SEAS-

Guyane Project: www.seas-guyane.org). Variables mea-

sured included slope steepness, lake size, channel length,

and width of vegetation at the lake–river boundary. Soils

were classified according to the Brazilian national classi-

fication system (EMBRAPA 2006) using a map (scale

1:25000) developed by the Zoneamento Ecologico-Eco-

nomico (ZEE) project (Venturieri 2007).

Core Sampling and Sample Preparation

Between January and February 2009, sediments were sam-

pled near the outlet of each lake using a Wildco� hand core

sediment sampler (120 cm in length and 5 cm in diameter)

lined with PVC tubing. During the transition from the dry to

the wet season, water levels are low (around 3 m in depth)

and disturbance of the sediments and water column are

minimal, allowing for easy extraction and handling of sed-

iments. Local fishers also shared their knowledge of the lake

systems to help identify adequate sampling sites—areas of

the lake that are inundated, even in the peak of the dry

season. The sediments were then extracted from the PVC

tube and subsampled at 1-cm intervals. Because the action

of coring can cause mixing of sedimentary layers at the outer

edges of the core, only the center of each 1-cm slice was

retained for chemical analysis; the edges were removed and

put aside for physical analyses. Subsamples were transferred

to glass tubes and were immediately frozen for transporta-

tion to the laboratory.

Laboratory Analysis: THg, C, and N Measurements

All subsamples were freeze-dried. Sample weight was

taken before and after drying to measure water content.

Arch Environ Contam Toxicol

123

Density was measured at every 10 cm along cores by

drying a known volume of sediment (1 cm3) at 60 �C for

24 h. The particle size distribution of five subsamples from

the Araipa Lake were measured using a Beckman Coulter

Laser Diffraction Particle Size Analyzer (LS13320) and the

results were synthesized with the program Gradistat v.4

(Blott and Pye 2001). This lake was chosen over others,

because, as detailed below, it has the lowest sedimentation

rate at thus provides deeper insights into the past.

The following subsamples from each core were chosen

for chemical analysis: the first 15 cm followed by a sub-

sample every 5 cm (20 cm, 25 cm, 30 cm, etc.). If present,

minute traces of leaves, sticks, or other organic material

were removed. Freeze-dried subsamples were then ground

to a homogenous powder with an agate mortar and pestle.

THg was extracted using hydrochloric acid digestion

and the total content was measured using cold vapor atomic

fluorescence spectrometry (Pichet et al. 1999). This method

has a detection limit of 0.1 ng/g for a 250-mg sample. The

accuracy of the method was verified using a Mess-3 cer-

tified standard. Replicates were run on all samples to

ensure precision of THg measurements, with a standard

error ranging from 0 to 4.5.

Total carbon (C) and total nitrogen (N) measurements

were carried out. Samples were combusted in a Carlo Erba

(NA-1500) elementary analyzer attached to a Thermo

Scientific Delta V Advantage isotope ratio mass spec-

trometer, with a relative precision of ±5% (1r) corrected

for atomic weight (Verardo et al. 1990). Replicates were

run on all samples to ensure precision, with a standard error

ranging from 0.019 to 0.50 for C and 0.002 to 0.12 for N

measurements. Atomic C/N ratios were then calculated by

dividing the weight percentage of C to the weight per-

centage of N.

The Anthropogenic Sedimentary Enrichment Factor

(ASEF) was calculated for each of the THg profiles (Lu-

cotte et al. 1995). It is the ratio of the baseline THg content

to the surface level content. The baseline is defined as the

part of the profile before which there is an important and

continual increase in THg content towards the surface.

A Kruskal–Wallis nonparametric ANOVA test, designed to

identify significant differences in population medians, was

performed using the Kruskal.test function in the stats

package of R, an open-source data analysis software

(R Core Team 2012).

Radioisotope Dating

To estimate sedimentation rates and model sediment age,

four cores were selected for 210Pb radioisotope analysis.

These cores were chosen because their geochemical pro-

files exhibited minimal disturbance, which is important

for accurate geochronological modeling. Radiometric

measurements of 210Pb activity were carried out using the

method described by (Moingt et al. 2014). Chronology was

established using downcore unsupported 210Pb and a con-

stant rate of supply (CRS) model (Ghaleb 2009). The CRS

model was chosen, because it provides more accurate

results in Brazilian hydrographic basins compared with

other geochronological models (Bonotto and Garcıa-

Tenorio 2014). To validate sediment age predictions,

Cesium-137 radioisotope (137Cs) measurements also were

taken (Ali et al. 2008). Sediment accumulation rates were

then calculated using the CRS sedimentation rate (cm/year)

and the porosity (g/cm3) of subsamples.

THg accumulation rates were also calculated using the

sediment accumulation rate and THg content. For sub-

samples beyond the level of supported 210Pb activity, THg

accumulation was estimated using the oldest CRS accu-

mulation rate for a given sample. While recognizing that

chronologies using 210Pb are uncertain for ages over

130 years, we nonetheless extrapolate deeper into time for

visual and illustrative purposes.

Land-Cover Change

Land-cover change in the eight watersheds was calculated

using data and methods from (Rozon et al. 2015). In their

study, Landsat images from three non-consecutive years

(1986, 2001, and 2009) were classified into the following

classes: (1) slightly altered old growth forest, (2) advanced

secondary forest growth, (3) initial secondary forest growth,

(4) crop and pasture lands, (5) highly degraded pasture lands,

(6) exposed soils/urban areas, and (7) humid/flooded zones.

To improve the temporal resolution of land-cover change

data, the same procedure was applied to two additional

satellite images captured in 1975 (Landsat 2 MSS) and 1997

(Landsat 7 TM). For each watershed, the proportion of non-

forested area (sum of classes 3–7) was calculated.

To calculate the extent of deforestation in the greater

Tapajos River watershed, a compilation of yearly defor-

estation data from the Sistema de Deteccao de Desmata-

mento em Tempo Real (DETER) program (INPE 2012) and

data from the Brazilian Ministry of the Environment

(MMA/IBGE 2012) were used. All procedures were per-

formed using (GRASS Development Team 2012).

Climate and Hydrological Data

Data were acquired from a hydrological station (048280S,

558990W) and a weather station (048170S, 568000W) located

near the city of Itaituba (Fig. 1). These include: monthly

average precipitation from 1968 to 2010 (INMET 2012),

monthly average stream flow and water-levels of the

Tapajos River from 1972 to 2011 (Cochonneau et al. 2006),

and daily suspended sediment and flow from 1997 to 2012

Arch Environ Contam Toxicol

123

(Cochonneau et al. 2006). These stations offered the most

consistent measurements, are in closest proximity to the

sampling sites, and are downstream from all sampled lakes.

The loess model, a robust technique to extract patterns

in seasonal and cyclical datasets, was used to decompose

the climatological and hydrological data into seasonal,

trend, and irregular components (Cleveland et al. 1990).

Because residuals (irregularities) were randomly dis-

tributed and demonstrated no temporal trends, the additive

model was deemed an appropriate method for removing the

Table 1 Structural characteristics of sampled lake and their drainage basin

Lake Lake

size

(km2)a

Watershed

size (km2)

Watershed

to lake

ratio

Soil

classes

Mean

slope

steepness

(%)

Proportion of

slopes\10%

incline

Lake–river

ConnectivitybLake

Shape

Vegetation at

lake outlet

(widthc) (m)

Canal

lengthc

(m)

Araipa 3.0 217.79 69.47 58%

argisol

8.64 0.36 Medium–

high

Dendritic 2–5 (river) –

42%

latosol

10%

neosol

Bom

Intento

0.27 3.33 12.33 100%

latosol

10.28 0.44 Low Elliptical 5–8 (canal) 65

Brasilia 1.97 527.38 268.59 79%

argisol

3.73 0.63 Medium Elliptical 3–7 (river) –

11%

latosol

10%

neosol

Cupu 1.97 81.21 40.32 42%

argisol

2.98 0.33 Medium Dendritic 7–9 (canal) 26

58%

latosol

Ipanema 0.08 16.48 206.97 100%

latosol

8.06 0.34 Low Elliptical 1 (canal) 88

Ipaupixuna 1.27 91.37 71.98 80%

argisol

2.60 0.16 Medium–

high

Dendritic 1–4 (river) –

11%

latosol

9%

gleysol

Piracana 2.39 702.79 297.47 81%

argisol

3.26 0.43 Medium Elliptical 3–9 (river) –

17%

latosol

2%

gleysol

Torbias 0.21 115.08 553.19 70%

latosol

3.19 0.38 High Elliptical 0 –

27%

argisol

3%

neosol

a Mean lake size, calculated from multiple dry season measurements using classified satellite imagery (Rozon et al. 2015)b Connectivity with the main river channel at the time of sampling based on shoreline lake classifications by (Sippel et al. 1992). Connectivity is

a function of floodplain vegetation at the lake–river boundary and lake morphology. Note that these variables change over time due to natural

ecosystem successionsc Measured using dry season satellite imagery (Rozon et al. 2015). ‘‘River’’ indicates that the vegetated area separating the main river channel

from the lake was measured. ‘‘Canal’’ indicates that the lake is connected to the river by a canal, and so the vegetated area between the canal and

the main river channel was measured

Arch Environ Contam Toxicol

123

seasonal component from the original time-series datasets.

Suspended sediment and flow data were only available at

irregular time-steps, sometimes with months or years

missing between measurements; as such, trend analysis

was not feasible. All data manipulations and analyses were

undertaken with the zoo and stats packages in (R Core

Team 2012; Zeileis and Grothendieck 2005).

Results

Watershed Characteristics

The small and contiguous watersheds of Ipanema and Bom

Intento (Fig. 1) have some of the steepest slopes (average

steepness of 8.06 and 10.28%, respectively; Table 1).

Ipaupixuna and Cupu, on the other hand, have the flattest

topography (average steepness of 2.60 and 2.98%

respectively).

In most watersheds in the study region, Argisols are the

predominant soil class (similar to Ultisols of the U.S.

system) (Table 1). These are highly weathered soils that

exhibit leaching of the fine clay fraction to lower horizons.

However, in the Ipanema, Bom Intento, and Torbias

watersheds, soils are almost exclusively Latosols. These

soils tend to have higher Hg content due to elevated levels

of iron and aluminum sesquioxydes (Roulet et al. 1998b).

Lakes were categorized by their level of connectivity

with the main Tapajos River channel (Table 1). Bom

Intento Lake has low connectivity due to its dense vege-

tation at the lake outlet and long canal, which reduce the

flow of matter and energy between the lake and the river.

As such, Bom Intento is primarily fed from the surrounding

watershed and there is a high hydrological residence time.

Brasilia Lake does not have a canal and less vegetation at

the lake–river boundary. It was thus classified as having

mid-range connectivity with the Tapajos River (Table 1).

Conversely, Torbias Lake is directly linked to the Tapajos

River, separated only by a few macrophyte communities at

the lake–river boundary. With a highly permeable vegeta-

tive barrier, Torbias Lake is predominantly fed from

advection of the river and has a shorter residence time. The

lake was thus classified as having the highest level of

connectivity (Table 1).

Land-Cover Change

In 1975, all watersheds had close to 100% forest cover

(Fig. 2a). Since then, most watersheds have undergone

extensive forest loss, sometimes by up to 50%. However,

the Torbias watershed does not follow these trends. Instead,

forest cover has remained relatively constant over the past

40 years (Fig. 2a).

Deforestation was most acute in the Ipaupixuna water-

shed, followed closely by Piracana (Fig. 2a). These two

watersheds are located near the city of Itaituba (Fig. 1),

where urbanization has been ongoing since the early 1980s

(Rozon et al. 2015). Notably, however, Ipaupixuna’s land

cover stabilized after 1997. Bom Intento was deforested at

a faster rate and earlier relative to other watersheds in close

proximity (e.g., Cupu, Araipa, and Ipanema; Figs. 1, 2a).

Proportionally, the deforested area in Bom Intento parallels

that of Ipaupixuna; however, the net deforested area is

actually different between the two watersheds (2 and

32 km2 for Bom Intento and Ipaupixuna, respectively).

Bom Intento Lake’s relative rapid forest loss is thus a

function of its small size (Table 1).

Fig. 2 Total deforested area for selected watersheds (a) and for the

Tapajos River watershed (b) from the 1970s to present day

Arch Environ Contam Toxicol

123

Vegetation at the lake–river boundary has changed

overtime in some lakes. In the case of Araipa Lake, a

sparse barrier of pioneering trees was observed at the outlet

at the time of sampling. However, satellite imagery from

1975 shows that vegetation was minimal and discontinuous

in the past—comparable to the macrophyte barrier

observed in present day Lake Torbias. Similar transitions

were observed at the boundary of Cupu and Brasilia lakes.

Vegetation is thicker in present day satellite images,

pointing towards an increase in mature flooded forest cover

at these lake–river boundaries. The opposite, however, was

observed for Ipanema Lake. In 1975, the lake–river

boundary was thicker and in mature flooded forest cover,

while in 2009 this barrier was thinned (Table 1).

Cumulative deforested area in the greater Tapajos River

watershed (Fig. 2b) generally parallels the early temporal

trajectories in sampled watersheds. The region was mostly

forested before 1970, and there was a rapid increase in

deforestation between 1985 and 2004.

THg, C, and C/N Profiles

In general, a common trend is observed across THg profiles

(Fig. 3). Deeper within most profiles, THg content is rel-

atively constant, remaining around a general baseline value

ranging between 100 and 150 ng/g. At a depth of

40–60 cm, depending on the sample, there is a rapid

change in THg loadings, tending towards surface enrich-

ment. While there are fluctuations in THg content deeper

along some cores, for example Bom Intento, Torbias, and

Araipa lakes, these variations do remain about the baseline.

In the case of Piracana Lake, a baseline could not be

established, as the core is shorter (65 cm in length). It was

not possible to collect a deeper sample in this lake.

Fig. 3 Hg profiles of sampled sediment cores. Standard error bars are shown where there was discrepancy among replicates

Arch Environ Contam Toxicol

123

Fig. 3 continued

Table 2 Mean THg content,

ASEF values, and associated

statistics

Lake THg l ± q (ng/g) Baseline depth (cm) AEF F-stat ANOVA (p value)

Araipa 217 ± 26 40 1.14 56.52 \0.0001

Bom Intento 153 ± 37 45 1.88 39.70 \0.0001

Brasilia 142 ± 43 40 1.85 367.27 \0.0001

Cupu 109 ± 30 40 1.71 64.85 \0.0001

Ipanema 200 ± 24 30 1.09 7.25 0.0195

Ipaupixuna 1 98 ± 20 40 1.59 283.11 \0.0001

Ipaupixuna 2 98 ± 20 40 1.18 20.98 0.0007

Piracana 183 ± 26 40 1.54 112.3 \0.0001

Torbias 160 ± 14 40 1 0.000 1.000

The Kruskal–Wallis nonparametric ANOVA test is for the null hypothesis that enrichment is not signifi-

cant; that is, there is no statistical difference between the baseline and surface THg. The depth at which the

baseline begins is indicated. Two ASEF values were calculated for Lake Ipaupixuna because there are

changes in the THg trends at the surface: at 40 cm and again at 13 cm. Ipaupixuna 2 is calculated as the

THg for the all surface subsamples (from 35 to 1 cm), whereas Ipaupixuna 1 only considers THg loadings

only from the 35–13 cm subsamples

Arch Environ Contam Toxicol

123

In most sediment profiles, there is significant THg

enrichment towards the surface, with levels of 200 to

250 ng/g that are typical of surface sediments in the region

(Roulet et al. 2000) (Table 2). In Brasilia, Cupu and Bom

Intento lakes, surface THg content has nearly doubled from

baseline levels (Table 2; Fig. 3). On the other hand, there is

no significant increase for the Torbias profile (p = 1.000;

Table 2). Surface THg enrichment in Ipanema and

Ipaupixuna lakes is significant, but not statistically strong

(Table 2).

The C and C/N profiles also indicate common trends

among most cores (Fig. 4). While variations deeper in the

cores vary, all profiles have pronounced surface level

enrichment of C content. Interestingly, Araipa, Torbias,

Brasilia, Cupu and Ipanema cores demonstrate strikingly

similar C profiles. Variations are most extreme in the

Torbias sample. This trend corresponds to a decrease in C

approximately 40–60 cm and a marked increase again after

20–10 cm. In all cases, the C/N ratios follow a similar

trajectory to C profiles. Two of the largest lakes,

Fig. 4 Proportion of total C and C/N atomic ratios for sampled cores. Solid circles represent %C and squares represent C/N ratios. Standard

error bars are shown for replicates. Note different scales are sometimes used to facilitate interpretation and visualization of the profiles

Arch Environ Contam Toxicol

123

Ipaupixuna and Piracana (Table 1), however, exhibit

smaller variations in C content, tending to increase at the

surface. Their C/N ratios remain relatively constant with a

slight downward trend towards the surface. This is due to

an increase in C content while N remains constant. The C

profile of Bom Intento Lake is unique. There is a constant

and increasing trend along the core, tending towards very

high values at the surface (Fig. 4).

Water content remains fairly constant in Bom Intento,

Piracana, and Ipaupixuna samples, followed by a sharp

increase at the surface (Fig. 5). In Araipa, Torbias, Ipa-

nema, and Brasilia lakes, both dry density and water con-

tent vary considerably along cores, although these two

variables consistently maintain an inverse relationship

(Fig. 5). Visual inspection of these samples indicated that

sediments were finer and more clayey in Torbias, Araipa,

and Ipanema lakes at a mid-depth of about 20–50 cm

(depending on the core). At this depth, water content is

very low and density is high. Sediments became notably

more granular towards the surface, where densities are

lower and water content is higher.

Granulometric analysis of five subsamples from Araipa

Lake confirms that there were important texture changes

along the core (Fig. 6). These changes correspond to

fluctuations in density and water content (Fig. 5). For

example, clay particle content is highest at 30 cm, when

Fig. 4 continued

Arch Environ Contam Toxicol

123

density is highest and water content is low. Deeper along

the core, at 80 cm, there is markedly less clay. This cor-

responds to low density and high water content. Given the

links between granulometric changes and density and water

content in the Araipa sample, it is assumed that similar

textural changes occurred in sediment samples from Ipa-

nema, Brasilia, and Torbias lakes.

Sedimentation Rates and Sediment Age

The results of the radioisotope analysis show that 210Pb

supported activity is reached at about 20 cm for both Cupu

and Brasilia cores. On the other hand, supported activity

was measured much deeper in the Bom Intento core

(50 cm) and closer to the surface in Araipa (15 cm; Fig. 7).210Pb activities generally follow a classic radioactive decay

curve. There are, however, signs of disturbance regimes in

sediment samples. This is likely due to natural (hydrolog-

ical) resuspension or bioperturbation processes. Such dis-

turbances rendered dating other samples of interest, notably

the Torbias core, unfeasible.

Sedimentation rates are high at the surface of sediment

cores (Fig. 7). This corresponds to the onset of significant

increases in THg content (Fig. 3; Table 2). In Araipa Lake,

the sedimentation rate doubled between 15 cm and the

surface (0.25–0.59 cm/year). In contrast, the sedimentation

rate in Cupu Lake varied slightly over time (0.21 cm/year

at a depth of 40 cm, 0.23 cm/year at the surface). In both

Fig. 5 Density and water content profiles of sampled cores. Circles represent density measurements and squares represent percent water content

Arch Environ Contam Toxicol

123

Brasilia and Bom Intento lakes, sedimentation rates

increased by nearly fourfold (0.11 cm/year at a depth of

40 cm, 0.41 cm/year at the surface) and fivefold (between

0.23 cm/year at a depth of 65 cm, 1.25 cm/year at the

surface) respectively.

To correct for the effect of changing sedimentation rates

over time, THg accumulation was calculated (Fig. 7). The

four lakes show a similar timeline of events with respect to

THg accumulation. Although variations were observed in

the 1920s–1960s, it is not until after the 1970s that there

was a notable shift in THg fluxes to lacustrine sediments.

This is most pronounced in Brasilia and Bom Intento Lakes

(Fig. 7). In Cupu Lake, annual accumulation follows this

temporal trend; however, the net amount of THg input is

relatively lower compared to other lakes.

Hydro-Climatological Trends

Figure 8 shows monthly hydro-climatological data with the

regular seasonal component removed. Within the time

period analyzed, no overall increasing or decreasing trends

are observed for monthly flow and water level data

(Fig. 8). Notwithstanding, the effect of El Nino Southern

Oscillation (ENSO) (cycle with 3- to 5-year periodicity)

(Botta et al. 2002; Marengo 2009) is apparent in these

hydrological datasets (Fig. 8). In particular, there was a

Fig. 5 continued

Arch Environ Contam Toxicol

123

period of frequent El Nino events from the late 1980s until

the beginning of the 21st century (Marengo 2004, 2009),

resulting in very low water-levels (Fig. 8).

The monthly precipitation trend shows a shift in 1982

(Fig. 8). This corresponds to a period of low rainfall

between the 1950s and 1980s in the Amazon, attributed to

long-term hydroclimatic cycles with a 28-year periodicity

(Botta et al. 2002; Marengo 2009). Although there are

reports of an overall drying trend in the greater Amazon

basin, regional variations in rainfall are notable (Duffy

et al. 2015; Foley et al. 2002), and there is evidence of a

modest positive trend in precipitation in the northern

region1 (Marengo 2009).

Daily flow measurements in the Tapajos River are highly

cyclical, due to seasonal flooding (Fig. 9). Peak sediment

loadings were measured in 2009–2010, whereas flow

remained within the seasonal range (Fig. 9). With long gaps

in the time-series data, it could not be determined if sediment

loadings significantly increased between 1997 and 2012.

Continuous data from 2005 to 2012, however, indicates that

suspended sediment measurements remained relatively low

until 2010, when a single event increased loadings.

Discussion

There has been an important shift in the geochemical

composition of lacustrine sediments in the Tapajos River

region in the last century. A significant enrichment of THg

content was observed at the surface of all sampled sedi-

ments, with the exception of the Torbias Lake core (Fig. 3;

Table 2). Most saliently, THg fluxes began increasing after

the 1970s (Fig. 7). This appears to be linked to the onset of

land-cover changes in the region (Fig. 2) as well as shifts in

climate (Fig. 8).

In the past 40 years, there has been extensive defor-

estation of dryland and seasonally flooded forests (igapos)

for ranching and agriculture (Oestreicher et al. 2014;

Rozon et al. 2015). Land-cover changes accelerate erosion

and leaching (Foley et al. 2005; Lal 2001), mobilizing soil

particles and naturally occurring elements, such as Hg

(Farella et al. 2006; Roulet et al. 1999, 2000). The heavy

metal then accumulates in lake sediments, as observed in

this study (Fig. 7). Although there was a brief gold rush in

the 1980s/1990s in the Tapajos region (Bezerra et al.

1996), profiles show no peak in THg at this time (Fig. 7).

This indicates that the effect of mining on THg fluxes is

likely minimal compared with land-cover change (Roulet

et al. 1998a, 1999).

Indeed, in the Torbias watershed, where land-cover

change has been negligible (Fig. 2), no significant surface

enrichment was observed (Table 2). Under primary forest

cover, raindrop impact on soils is limited and rainwater

tends to percolate through soil horizons (Fostier et al.

2000). As such, runoff, erosion, and overall Hg loss is

significantly lower on forested soils compared with agri-

cultural areas (Almeida et al. 2005; Beliveau et al., in

press).

Increased THg fluxes also correspond to a shift in pre-

cipitation in the 1980s and the onset of more frequent El

Nino events in the past 25 years (Fig. 8) (Marengo 2009).

Consequently, there have been more severe droughts

(Fig. 8) and more frequent and extreme short duration

rainfall events in this region (Marengo et al. 2009; Val-

verde and Marengo 2014). With stronger raindrop impact,

particularly after prolonged dry seasons, coarse particles

are more likely to be eroded (Beliveau et al., in press). In

the highly deforested Araipa watershed, for example,

coarse (sand) particles were measured only at 10 cm and

no other depth (Fig. 6). This corresponds to the 1980s

according to geochronological modeling (Fig. 7), when the

area deforested increased by more than 20% in a short time

(Fig. 2). Coarse sediments were similarly found at the

surface of the Ipanema and Cupu cores. Intensification of

drought periods also results in forest loss due to fires and

tree mortality (Malhi et al. 2008), further increasing the

risk of erosion. This indicates that a combination of both

land-cover change and climate variability are likely driving

increased THg loadings in sediments in the Tapajos River

region over the past decades.

The surface enrichment of THg in recent years (Fig. 3)

also may be explained by the complex feedbacks between

Fig. 6 Granulometry of selected sub-samples from the Lake Araipa

core

1 When the effect of increasing El Nino events is removed.

Arch Environ Contam Toxicol

123

climate, land-cover, and the biogeochemical cycling of

mercury (Roland et al. 2012). For example, deforestation

and climate change also increase the pH and temperature of

the aquatic environment (Batalha et al. 2014; Lacerda et al.

2012; Marengo et al. 2009). Such a shift in physiochemical

conditions promotes the sedimentary sequestration of Hg

from the suspended or dissolved fractions of the water

column in the tropics (Chakraborty and Babu 2015).

However, it is likely that deposition of terrestrial matter is

the primary mechanism of Hg accumulation in sediments,

given that sedimentation rates and THg content have sub-

stantially increased over time (Fig. 7). Nonetheless, Hg

isotopic analysis of mass-dependent and mass-independent

fractionations would complement these findings and help

to understand better the behavior, fate, and origin of the

metal in the floodplain system (Blum et al. 2014).

In addition to climate and land-cover change, watershed

biophysical characteristics explain THg accumulation

profiles (Fig. 7). Sediment yield and hydrologic response

are scale-dependent processes (Chang 2003; Merten et al.

Fig. 7 Total Pb210 activity for the four lakes and the CRS age and estimate Hg flux. Standard error bars are shown where there was discrepancy

among replicates

Arch Environ Contam Toxicol

123

2016), such that smaller watersheds, like Bom Intento or

Ipanema (Table 1), can accumulate sediments (and asso-

ciated Hg) more quickly (Fig. 7). As such, the response to

land use change, and even climate variability, may be

amplified (Krishnaswamy et al. 2001) (Fig. 7). In larger

watersheds, such as Araipa, Ipaupixuna, and Piracana

(Table 1), hydrologic flow paths are longer and soil parti-

cles are trapped in gullies, bottom slopes, or other sedi-

mentary sinks in the basin (Hamilton and King 1983;

Krishnaswamy et al. 2001). As such, the response to land

use change is dampened in larger basins (Fig. 7). Similarly,

erosion is more pronounced on steep slopes, so farming on

these areas can contribute substantially to Hg accumulation

in sediments (Belanger 2012; Beliveau et al., in press). The

Bom Intento watershed has some of the steepest slopes and

Latosol soils (Table 1), known to have high Hg loadings

(Roulet et al. 1998b). Accordingly, some of the highest

levels of sedimentary THg were measured in this lake

(Fig. 3; Table 2).

In this sense, the spatial arrangement of the landscape is

important, including where deforestation occurs. Distur-

bance of riparian vegetation can increase the flow of soil

particles and contaminants, because it acts as a buffer

between the terrestrial to aquatic systems (Fennessy and

Fig. 7 continued

Arch Environ Contam Toxicol

123

Cronk 1997). In Bom Intento Lake, one-third of seasonally

flooded forests in the riparian zone were removed between

1997 and 2001, corresponding to the most important

increase in THg accumulation rates in this sample (Fig. 7).

In fact, approximately 75% of this sensitive ecotone had

been deforested or degraded in all watersheds at the time of

the study—with the exception of Torbias Lake.2 On the

other hand, the preservation of riparian zones can limit

sediment and THg fluxes to waterbodies (Celentano et al.

2016).

Vegetation at the lake outlet also will affect sedimentary

dynamics, because it regulates lake–river connectivity

(Amoros and Bornette 2002). For example, in the late

1980s, the seasonally flooded forest at the lake–river

boundary of Ipanema Lake was slashed and burned; this

caused extreme flood pulses in the years that followed, as

the lake abruptly transitioned to an open system. Indeed,

there is evidence of this in the Ipanema core: a decrease in

organic C content and increase in density at around

30–40 cm (Figs. 4, 5), which corresponds to textural

changes (coarser particles) and the presence of charcoal in

the sample. These physiochemical shifts appear to be due

to the fact that in seasonally flooded forests, sedimentary C

content is elevated owing to high ecosystem productivity

(Kalliola et al. 1991) and water flow is lower compared to

permeable and connected systems (Dunne et al. 1998; Junk

1997; Tockner et al. 2010). Evidence of a similar transition

in lake–river connectivity is found deep in the core of Lake

Araipa (40–60 cm; Figs. 5, 6). Because there is no satellite

imagery or hydroclimatic data from this time, it is unclear

whether this was due to human activity or natural events

(e.g., extreme floods, fire) (Junk and Piedade 2011; Salo

et al. 1986; Worbes et al. 1992). Notably, Cupu and Bra-

silia lakes have strikingly similar C and density profiles

(Figs. 4, 5), suggesting that lake–river connectivity also

may have changed over time.

Connectivity also explains the unique variations in the

Torbias Lake THg profile (Fig. 3). With its elliptical

shape and macrophytes (floating reed beds) growing at the

outlet, the lake–river boundary is highly permeable (Junk

1997; Junk and Piedade 2011; Sippel et al. 1992). As

such, this open lake system is strongly influenced by the

Tapajos River (Amoros and Bornette 2002), leading to

seasonal resuspension of sediments and mixing of par-

ticulate matter (Tockner et al. 2010; Tockner and Stanford

2002). Furthermore, there is considerable seasonal varia-

tion of THg levels in the waters of the Tapajos (Roulet

et al. 1998a). Given this, the unique THg profile in this

core is likely due to the fluctuations of flow and suspended

sediments in the main river channel (Fig. 9) (Aalto and

Nittrouer 2012; Maurice-Bourgoin et al. 2007). In par-

ticular, the frequency of El Nino events (Fig. 8) is an

important driver of episodic deposition in floodplains, the

effect of which has been observed in stratigraphic records

elsewhere in the Amazon (Aalto et al. 2003).

While lake–river connectivity clearly affects sedimen-

tary dynamics, it seems to have little effect on THg accu-

mulation. Deeper in many cores, THg content and

accumulation rates fluctuate but remain close to baseline

values. This provides further evidence that recent land-use

change is the primary source of Hg deposition in surface

Fig. 8 Long-term hydrological and rainfall data from the Itaituba

station (Fig. 1) decomposed into the estimated trend and the

seasonally adjusted values. Dotted lines represent the mean and

standard deviation (1r) for the datasets. Arrows indicate strong El

Nino years

2 The riparian zone is considered to be a 100-m buffer zone around

the lake limits. Calculated using data from (Rozon et al. 2015).

Arch Environ Contam Toxicol

123

sediments and that the metal originates from the local

drainage basin (Belanger 2012).

Conclusions

In areas where land-cover changes have occurred, a sig-

nificant enrichment of THg was documented at the surface

of lacustrine sediments in the Tapajos River. This is a

serious concern for environmental and human health in the

region. New road developments and hydroelectric dam

projects are planned for the near future, and they will

promote further deforestation of upland forests. Moreover,

disturbance of vegetation in sensitive areas merits consid-

eration in management strategies and development plan-

ning—this includes riparian zones and the lake–river

boundary with its unique seasonally flooded ecosystem.

These areas are key determinants of THg accumulation and

sedimentation dynamics, as they regulate the connectivity

between terrestrial and aquatic ecosystems and between the

lake and the river. Given this, future research should

examine the spatial configuration of land-cover change to

better quantify the extent to which these areas regulate the

flux of contaminants in tropical floodplain systems. Hydro-

climatic cycles and events also influence THg accumula-

tion and sedimentary dynamics in the floodplain. Because

shifts in climate variables occurred at the same time as

land-cover changes, this study cannot determine the rela-

tive contribution of anthropogenic disturbances and climate

factors on THg. Nonetheless, it seems important for future

research to include climate change in study designs, par-

ticularly if rainfall anomalies and seasonal flooding

become more extreme or if river chemistry shifts.

Acknowledgements This study was performed as part of the Poor

Land Use, Poor Health Project (PLUPH 2012), a research endeavor

that examined the interconnections between environmental change

and human health in the Brazilian Amazon in 2008–2012. This

research was made possible through funding by the Global Health

Research Initiative (GHRI), the International Development Research

Centre (IDRC), the GEOTOP strategic cluster, the Canadian National

Research council (NSERC), and the Quebec Research Foundation

Fig. 9 Recent daily flow and

sediment loadings of the

Tapajos River from the Itaituba

station (Fig. 1)

Arch Environ Contam Toxicol

123

(FQRNT). The authors thank Sophie Chen, Bassam Ghaleb, Leandra

Fatorelli, Annie Beliveau, Stephane Tremblay, Jose Raimundo Serra

Nogueira (Ray), Chiquinho, Naldo, Marie Tissot, Joevin, Pierre

Taillardat, and Carlos Jose Passos. They graciously assisted with data

collection and analysis, provided their local and specialized knowl-

edge, or gave key insights, suggestions, and logistical support that

made this paper possible. Many thanks to the community members

who participated in this study.

References

Aalto R, Nittrouer CA (2012) 210Pb geochronology of flood events in

large tropical river systems. Phil Trans Royal Soc A Math Phys

Eng Sci 370(1966):2040–2074. doi:10.1098/rsta.2011.0607

Aalto R, Maurice-Bourgoin L, Dunne T, Montgomery D, Nittrouer C,

Guyot J (2003) Episodic sediment accumulation on Amazonian

flood plains influenced by El Nino/Southern Oscillation. Nature

425(6957):493–497

AliAA Ghaleb B, Garneau M, Asnong H, Loisel J (2008) Recent peat

accumulation rates in minerotrophic peatlands of the Bay James

region, Eastern Canada, inferred by 210Pb and 137Cs radiometric

techniques. Appl Radiat Isotopes 66(10):1350–1358

Almeida MD, Lacerda LD, Bastos WR, Herrmann JC (2005) Mercury

loss from soils following conversion from forest to pasture in

Rondonia, Western Amazon, Brazil. Environ Pollut 137:179–186

Amoros C, Bornette G (2002) Connectivity and biocomplexity in

waterbodies of riverine floodplains. Freshw Biol 47(4):761–776.

doi:10.1046/j.1365-2427.2002.00905.x

Babiarz CL, Benoit JM, Shafer MM, Andren AW, Hurley JP, Webb

DA (1998) Seasonal influences on partitioning and transport of

total and methylmercury in rivers from contrasting watersheds.

Biogeochemistry 41(3):237–257

Batalha SSA, Martorano LG, Biase AG, Morales GP, Pontes AN,

Santos LSd (2014) Physico-chemical and biological conditions

in surface waters of the Tapajos River and the conservation of

the National Forest in the Amazon, Brazil. Revista Ambiente

Agua 9(4):647–663

Belanger E (2012) Biomarqueurs de la lignine dans les vegetaux

dominants et les sols de milieux boises et perturbes de

l’Amazonie bresilienne. (M.Sc.). Universite du Quebec a

Montreal, Montreal

Beliveau A, Lucotte M, Davidson R, Paquet S, Mertens Fdr, Passos

CJS, Romana CA (in press) Soil erosion and mercury losses in

agroforestry systems compared to cultivated fields in the

Brazilian Amazon. J Environ Management

Berzas Nevado JJ, RodrIguez MartIn-Doimeadios RC, Guzm�nBernardo FJ, JimEnez Moreno M, Herculano AM, do Nasci-

mento JLM, Crespo-LUpez ME (2010) Mercury in the Tapajos

River basin Brazilian Amazon: a review. Environ Int

36(6):593–608

Bezerra O, Verıssimo A, Uhl C (1996) The regional impacts of small-

scale gold mining in Amazonia. Natural Res Forum

20(4):305–317

Blott SJ, Pye K (2001) GRADISTAT: a grain size distribution and

statistics package for the analysis of unconsolidated sediments.

Earth Surf Processes Landforms 26(11):1237–1248

Blum JD, Sherman LS, Johnson MW (2014) Mercury isotopes in

earth and environmental sciences. Ann Rev Earth Planet Sci

42:249–269

Boas RV (1997) The mercury problem in the Amazon due to gold

extraction. J Geochem Explor 58(2):217–222

Bonotto D, Garcıa-Tenorio R (2014) A comparative evaluation of the

CF: CS and CRS models in 210Pb chronological studies applied

to hydrographic basins in Brazil. Appl Radiat Isotopes 92:58–72

Botta A, Ramankutty N, Foley J (2002) Long-term variations of

climate and carbon fluxes over the Amazon basin. Geophys Res

Lett 29(9):31–33

Celentano D, Rousseau GX, Engel VL, Zelarayan M, Oliveira EC,

Araujo ACM, Moura EG (2016) Degradation of riparian forest

affects soil properties and ecosystem services provision in

eastern amazon of brazil. Land Degradation Develop

Chakraborty P, Babu PR (2015) Environmental controls on the

speciation and distribution of mercury in surface sediments of a

tropical estuary, India. Mar Pollut Bull 95(1):350–357

Chang H (2003) Basin hydrologic response to changes in climate and

land use: the Conestoga River basin, Pennsylvania. Phys Geogr

24(3):222–247

Cleveland RB, Cleveland WS, McRae JE, Terpenning I (1990) STL: a

seasonal-trend decomposition procedure based on loess. J Off

Stat 6(1):3–73

Cochonneau G, Sondag F, Guyot J-L, Geraldo B, Filizola N, Fraizy P,

Noriega L (2006) The Environmental Observation and Research

project, ORE HYBAM, and the rivers of the Amazon basin.

Climate Variability Change 44–50

Dearing JA (2008) Landscape change and resilience theory: a

palaeoenvironmental assessment from Yunnan, SW China.

Holocene 18(1):117–127. doi:10.1177/0959683607085601

Dearing JA, Braimoh AK, Reenberg A, Turner BL, van der Leeuw S

(2010) Complex land systems: the need for long time perspec-

tives to assess their future. Ecol Soc 15(4):21

Dolbec J, Mergler D, Larribe F, Roulet M, Lebel J, Lucotte M (2001)

Sequential analysis of hair mercury levels in relation to fish diet

of an Amazonian population, Brazil. Sci Total Environ

271(1–3):87–97

Duffy PB, Brando P, Asner GP, Field CB (2015) Projections of future

meteorological drought and wet periods in the Amazon. Proc

Natl Acad Sci 112(43):13172–13177

Dunne T, Mertes LAK, Meade RH, Richey JE, Forsberg BR (1998)

Exchanges of sediment between the flood plain and channel of

the Amazon River in Brazil. Geol Soc Am Bull

110(4):450–467

EMBRAPA (2006) Sistema brasileiro de classificacao de solos.

Brazilian Agricultural Research Corporation, Brasılia, DF

EMBRAPA (2007) Mapa da Cobertura Vegetal do Brasil

Farella N, Lucotte M, DavidsonR Daigle S (2006) Mercury release

from deforested soils triggered by base cation enrichment. Sci

Total Environ 368:19–29

Fennessy M, Cronk J (1997) The effectiveness and restoration

potential of riparian ecotones for the management of nonpoint

source pollution, particularly nitrate. Crit Rev Environ Sci Tech

27(4):285–317

Foley JA, Botta A, Coe MT, Costa MH (2002) El Nino-Southern

oscillation and the climate, ecosystems and rivers of Amazonia.

Global Biogeochem Cycles 16(4):1132

Foley JA, DeFries R, Asner GP, Barford C, Bonan G, Carpenter SR,

Gibbs HK (2005) Global consequences of land use. Science

309(5734):570–574

Fostier A-H, Forti MC, Guimaraes JRD, Melfi AJ, Boulet R, Santo

CME, Krugg FJ (2000) Mercury fluxes in a natural forested

Amazonian catchment (Serra do Navio, Amapa State, Brazil).

Sci Total Environ 260:201–211

Ghaleb B (2009) Overview of the methods for the measurement and

interpretation of short-lived radioisotopes and their limits. IOP

Conf Ser 5:01007

GRASS Development Team (2012) Geographic Resources Analysis

Support System (GRASS) Software. Open Source Geospatial

Foundation Project. Accessed at http://grass.osgeo.org/

Grimaldi C, Grimaldi M, Guedron S (2008) Mercury distribution in

tropical soil profiles related to origin of mercury and soil

processes. Sci Total Environ 401(1–3):121–129

Arch Environ Contam Toxicol

123

Hacon S, Barrocas PR, Vasconcellos ACSd, Barcellos C, Wasser-

man JC, Campos RC, Azevedo-Carloni FB (2008) An

overview of mercury contamination research in the Amazon

basin with an emphasis on Brazil. Cadernos Saude Publ

24(7):1479–1492

Hamilton LS, King PN (1983) Tropical forested watersheds: hydro-

logic and soils response to major uses or conversions. Westview

Press, Boulder

Hurley JP, Benoit JM, Babiarz CL, Shafer MM, Andren AW, Sullivan

JR, Webb DA (1995) Influences of watershed characteristics on

mercury levels in Wisconsin rivers. Environ Sci Tech

29(7):1867–1875

INMET (2012) BDMEP—Banco de Dados Meteorologicos para

Ensino e Pesquisa

INPE (2012) Sistema de Deteccao de Desmatamento em Tempo Real

(DETER)

Junk WJ (1997) General aspects of floodplain ecology with special

reference to amazonian floodplains. In: Junk WJ (ed) The central

Amazon floodplain: ecology of a pulsing system, vol 126.

Springer, New York

Junk WJ, Piedade MT (2011) An introduction to South American

wetland forests: distribution, definitions and general character-

ization Amazonian Floodplain Forests. Springer, Netherlands,

pp 3–25

Junk WJ, Bayley PB, Sparks RE (1989) The flood pulse concept in

river-floodplain systems. Can Spec Publ Fish Aquat Sci

106:110–127

Kalliola R, Salo J, Puhakka M, Rajasilta M (1991) New site formation

and colonizing vegetation in primary succession on the Western

Amazon floodplains. J Ecol 79(4):877–901. doi:10.2307/

2261087

Krishnaswamy J, Halpin PN, Richter DD (2001) Dynamics of

sediment discharge in relation to land-use and hydro-climatology

in a humid tropical watershed in Costa Rica. J Hydrol

253(1):91–109

Lacerda LD, Bastos WR, Almeida MD (2012) The impacts of land

use changes in the mercury flux in the Madeira River, Western

Amazon. Anais Acad Bras Ciencias 84(1):69–78

Lal R (2001) Soil degradation by erosion. Land Degrad Dev

12(6):519–539

Laperche V, Hellal J, Maury-Brachet R, Joseph B, Laporte P, Breeze

D, Blanchard F (2014) Regional distribution of mercury in

sediments of the main rivers of French Guiana (Amazonian

basin). SpringerPlus 3(1):1

Lebel J, Roulet M, Mergler D, Lucotte M, Larribe F (1997) Fish diet

and mercury exposure in a riparian Amazonian population.

Water Air Soil Pollut 97(1):31–44

Lucotte M, Mucci A, Hillaire-Marcel C, Pichet P, Grondin A

(1995) Anthropogenic mercury enrichment in remote lakes of

northern Quebec (Canada). Water Air Soil Pollut 80(1–4):

467–476

Maia PD, Maurice L, Tessier E, Amouroux D, Cossa D, Perez M,

Rheault I (2009) Mercury distribution and exchanges between

the Amazon River and connected floodplain lakes. Sci Total

Environ 407(23):6073–6084

Malhi Y, Roberts JT, Betts RA, Killeen TJ, Li W, Nobre CA (2008)

Climate change, deforestation, and the fate of the Amazon.

Science 319(5860):169–172. doi:10.1126/science.1146961

Malm O (1998) Gold mining as a source of mercury exposure in the

Brazilian Amazon. Environ Res 77(2):73–78

Marengo JA (2004) Interdecadal variability and trends of rainfall

across the Amazon basin. Theor Appl Clim 78(1–3):79–96.

doi:10.1007/s00704-004-0045-8

Marengo JA (2009) Long-term trends and cycles in the hydromete-

orology of the Amazon basin since the late 1920s. Hydrol Proc

23(22):3236–3244. doi:10.1002/hyp.7396

Marengo JA, Jones R, Alves L, Valverde M (2009) Future change of

temperature and precipitation extremes in South America as

derived from the PRECIS regional climate modeling system. Int

J Clim 29(15):2241–2255

Martinez J-M, Le Toan T (2007) Mapping of flood dynamics and

spatial distribution of vegetation in the Amazon floodplain using

multitemporal SAR data. Remote Sensing Environ

108(3):209–223

Maurice-Bourgoin L, Aalto R, Guyot JL (2002) Sediment-associated

mercury distribution within a major Amazon tributary: century-

scale contamination history and importance of flood plain

accumulation. IAHS Publication, pp. 161–168

Maurice-Bourgoin L, Bonnet M-P, Martinez J-M, Kosuth P, Cochon-

neau G, Moreira-Turcq P, Seyler P (2007) Temporal dynamics of

water and sediment exchanges between the Curuaı floodplain

and the Amazon River, Brazil. J Hydrol 335(1–2):140–156

Mergler D, Anderson HA, Chan LHM, Mahaffey KR, Murray M,

Sakamoto M, Stern AH (2007) Methylmercury exposure and

health effects in humans: a worldwide concern. Ambio

36(1):3–11

Merten GH, Welch HL, Tomer M (2016) Effects of hydrology,

watershed size, and agricultural practices on sediment yields in

two river basins in Iowa and Mississippi. J Soil Water Conserv

71(3):267–278

MMA/IBGE (2012) Dados geograpicos da Amazonia Legal Brasileira

Moingt M, Lucotte M, Paquet S, Beaulne J (2013) The influence of

anthropogenic disturbances and watershed morphological char-

acteristics on Hg dynamics in Northern Quebec large boreal

lakes. Adv Environ Res 2(2):81–98

Moingt M, Lucotte M, Paquet S, Ghaleb B (2014) Deciphering the

impact of land-uses on terrestrial organic matter and mercury

inputs to large boreal lakes of central Quebec using lignin

biomarkers. Appl Geochem 41:34–48

Oestreicher JS, Farella N, Paquet S, Davidson R, Lucotte M, Mertens

F, Saint-Charles J (2014) Livelihood activities and land-use at a

riparian frontier of the Brazilian Amazon: quantitative charac-

terization and qualitative insights into the influence of knowl-

edge, values, and beliefs. Hum Ecol 42(4):521–540

Passos CJS, Mergler D, Lemire M, Fillion M, Guimares JR (2007)

Fish consumption and bioindicators of inorganic mercury

exposure. Sci Total Environ 373(1):68–76

Pichet P, Morrison K, Rheault I, Tremblay A (1999) Analysis of total

mercury and methylmercury in environmental samples Mercury

in the biogeochemical cycle. Springer, Berlin, pp 41–52

PLUPH (2012) Poor land use, poor health project. Details available

at: http://www.pluph.uqam.ca/

R Core Team (2012) R A Language and Environment for Statistical

Computing (Version version 0.97.316). Vienna, Austria: R

Foundation for Statistical Computing. Accessed at http://www.

r-project.org/

Roland F, Huszar V, Farjalla V, Enrich-Prast A, Amado A, Ometto J

(2012) Climate change in Brazil: perspective on the biogeo-

chemistry of inland waters, Brazil. J Biol 72:709–722

Roulet M, Lucotte M, Canuel R, Rheault I, Tran S, De Freitos Gog Y,

Jesus De, da Silva E (1998a) Distribution and partition of total

mercury in waters of the Tapajos River Basin, Brazilian

Amazon. Sci Total Environ 213(1):203–211

Roulet M, Lucotte M, Saint-Aubin A, Tran S, Rheault I, Farella N,

Amorim M (1998b) The geochemistry of mercury in central

Amazonian soils developed on the Alter-do-Chao formation of

the lower Tapajos River Valley, Para state, Brazil. Sci Total

Environ 223:1–24

Roulet M, Lucotte M, Farella N, Serique G, Coelho H, Passos CJS,

Amorim M (1999) Effects of recent human colonization on the

presence of mercury in Amazonian ecosystems. Water Air Soil

Pollut 112:297–313

Arch Environ Contam Toxicol

123

Roulet M, Lucotte M, Canuel R, Farella N, Courcelles M, Guimaraes

J-RD, Amorim M (2000) Increase in mercury contamination

recorded in lacustrine sediments following deforestation in the

central Amazon. Chem Geol 165:243–266

Rozon C, Lucotte M, Davidson R, Paquet S, Oestreicher JS, Mertens

F, Romana CA (2015) Spatial and temporal evolution of family-

farming land use in the Tapajos region of the Brazilian Amazon.

Acta Amazon 45:203–214

Salo J, Kalliola R, Hakkinen I, Makinen Y, Niemela P, Puhakka M,

Coley PD (1986) River dynamics and the diversity of Amazon

lowland forest. Nature 322(6076):254–258

Sampaio da Silva D, Lucotte M, Paquet S, Davidson R (2009)

Influence of ecological factors and of land use on mercury levels

in fish in the Tapajos River basin, Amazon. Environ Res

109(4):432–446

Sippel S, Hamilton S, Melack J (1992) Inundation area and

morphometry of lakes on the Amazon River floodplain, Brazil.

Arch Hydrobiol Stuttgart 123(4):385–400

Tockner K, Stanford JA (2002) Riverine flood plains: present state

and future trends. Environ Conserv 29(3):308–330

Tockner K, Pusch M, Borchardt D, Lorang MS (2010) Multiple

stressors in coupled river–floodplain ecosystems. Freshw Biol

55:135–151. doi:10.1111/j.1365-2427.2009.02371.x

USGS (2004) Shuttle Radar Topography Mission, 1 Arc Second scene

SRTM_u03_n008e004, Unfilled Unfinished 2.0, Global Land

Cover Facility

Valeriano MdM (2004) Modelo digital de elevacao com dados SRTM

disponıveis para a America do Sul INPE

Valverde MC, Marengo JA (2014) Extreme rainfall indices in the

hydrographic basins of Brazil. Open J Mod Hydrol 4:10–26

Venturieri AV (2007) Resumo dos diagnosticos tematicos do

zoneamento ecologico-economico na area de influencia da

rodovia BR 163 (Cuiaba-Santarem), vol 1. Embrapa Amazonia

Oriental, Belem, Brazil

Verardo DJ, Froelich PN, McIntyre A (1990) Determination of

organic carbon and nitrogen in marine sediments using the Carlo

Erba NA-1500 Analyzer. Deep Sea Research Part A. Oceanogr

Res Pap 37(1):157–165

Worbes M, Klinge H, Revilla JD, Martius C (1992) On the dynamics,

floristic subdivision and geographical distribution of varzea

forests in Central Amazonia. J Veg Sci 3(4):553–564. doi:10.

2307/3235812

Zeileis A, Grothendieck G (2005) zoo: S3 infrastructure for regular

and irregular time series. J Stat Soft 14(i06):1–27

Arch Environ Contam Toxicol

123