|
International Journal of Environment Science and Technology, Vol. 8, No. 1, 2010, pp. 1-18 Urbanization impact on metals mobility in riverine suspended sediment: Role of metal oxides *C. Priadi; S. Ayrault; S. Pacini; P. Bonte Laboratoire des Sciences du Climat et de l’Environnement (LSCE/IPSL, CEA-CNRS-UVSQ) avenue de la terrasse, 91198 Gif sur Yvette cedex, France *Corresponding Author Email: cindy.priadi@lsce.ipsl.fr Tel.: +33 16982 4362; Fax: +33 1698 24362 Received 22 June 2010; revised 4 October 2010; accepted 25 November 2010 Code Number: st11001 ABSTRACT: Spatial and temporal fractionation of trace metals and major elements in suspended particulate matter in the Seine River was investigated to study the impact of the increasing urbanization in the Greater Paris Region. Suspended sediments in the Seine River were collected between December 2008 to August 2009 upstream and downstream of Paris. They were subjected to total digestion and sequential extraction procedure certified by the Bureau Communautaire de Référence and trace metals along with major elements were analyzed with inductively coupled plasma mass spectroscopy. Metal enrichment factors increased up to eight folds after the Seine River downstream of the Greater Paris Region showing a significant contribution of urbanization. Enrichment of copper, lead and zinc downstream of Paris are followed by the increase of their reducible fraction of at least 10% implicating an increase in metals associated with iron oxides. The exchangeable fraction, which includes the carbonate-associated metals, is only significant for cadmium, nickel and zinc (more than 2 %) while the oxidisable fraction accounts for less than 20 % for the anthropogenic metals downstream except for copper. The metals can be divided to (a) "reducible" group including cadmium, lead, and zinc, associated with more than 60 % of the total Bureau Communautaire de Référence extractable metals to the reducible fraction containing mostly iron oxide phases for the downstream sites. (b) A "distributed" group including chromium, copper, and nickel that are associated to at least 3 different phase-groups: (1) oxides, (2) organic matter and sulphides and (3) mineral phases. Keywords: Carbonate; Enrichment factor; Pollution; Reducible fraction;
Sequential extraction INTRODUCTION In the last decade, metal contamination in urban continental aquatic system has been a growing concern (Sekabira et al., 2010; Mohiuddin et al., 2010). Impacts of anthropogenic activities on metal contamination in a watershed are far from being insignificant (Horowitz et al., 1999; Davis et al., 2001; Taylor and Owens, 2009; Igbinosa and Okoh, 2009). They are known to generate a considerable amount of metal to the environment through various pathways including atmospheric particles (Azimi et al., 2005; Zhu et al., 2009), urban runoff (Gromaire-Mertz et al., 1999; Igwe et al., 2008), industrial and wastewater effluents (Buzier et al., 2006; Nwuche, 2008; Shah et al., 2009). Urbanization impacts on metal contamination concern watersheds inhabiting more than 50 % of the world population (Meybeck, 2003). The Seine River watershed is home to 25-30 % of French industries, 23 % of French constantly increasing urban and agricultural activities contribute to metallic aquatic contamination. In the last two decades, many institutions carried out actions in the Seine River to understand pollutant behaviour with parallel decontamination. Since then, the concentration of trace elements such as the Cd, Cr, Cu, Ni, Pb, and Zn showed a significant decline measured in dated sediment cores (Le Cloarec et al., 2009). Nevertheless, due to very high anthropogenic pressures and very limited dilution power, the Seine River downstream of Greater Paris is still among the world's most contaminated rivers (Meybeck et al., 2007). Despite the decreasing concentrations in the solid phases, Elbaz-Poulichet et al. (2006) still found moderate contamination of dissolved metals in the water column of the Marne and Seine Rivers. Various studies contributed to the understanding of metal behaviour in different compartments due to the dynamic metal distribution between different phases in the water column. In the Seine River, the value and dynamics of dissolved Mn, Cu, Cd and Mo were partly attributed to variation in redox condition (Elbaz-Poulichet et al., 2006). Short-time extreme variation of dissolved Zn is also observed using high-definition sampling (Pepe et al., 2008). Depending on the element, the solid fraction holds 50-90 % of the anthropogenic metal stock (Cd, Cu, Cr, Zn, and Pb) in the Seine River water column (Thévenot et al., 2007). Furthermore, floodplain and bed sediments contain a large stock of historical deposit and contamination. Therefore, this solid fraction plays an important role in metal contamination as it may release heavy metal to the water column, as well as scavenging them. It is then necessary to understand the behaviour and distribution of trace elements in the different solid phases and their mobility towards the dissolved fraction. The latter, more specifically the labile fraction, represent the bio-available fraction. Moreover, the solid fraction may hold evidences of metal sources and formation processes. The study of metal speciation in the solid fraction may therefore help us understand the contribution of different sources and biogeochemical processes in the formation and mobility of metal in the solid phase. Metal speciation studies such as the sequential extraction are often performed on mining-impacted rivers (Da Silva et al. , 2002; Galan et al., 2003; Audry et al., 2006; Lesven et al., 2009) but rarely on non-mining urban watershed even though urban anthropogenic activities can generate a specific characteristic of metal speciation in a water course (Garnaud, 1999; Dali-Youcef et al., 2004; Carter et al., 2006; Sutherland and Tack, 2007). Interpretation are also rarely compared by simultaneous extraction of major elements originating from the metal-bearing particles including Ca, Fe, Mn, and Mg, except for a few studies (Gagnon et al., 2009; Li et al., 2009; Vieira et al., 2009). Tongtavee et al. (2005) have shown that the analysis of extracted major elements is of great interest to understand lead speciation in soils affected by mining activities. Despite the Seine River basin's great economic importance
and the many articles published on contents and behaviour in dissolved and
labile fraction of trace metals in the Seine River (Elbaz-Poulichet et al., 2006; Tusseau-Vuillemin et
al., 2007; Chen et al., 2009; Jouvin et al., 2009),
trace metal speciation on solid fraction were only conducted on average suspended
matter from 3 sites (Taconet, 1996) and urban
source-related samples (Garnaud, 1999). According to these studies, metal
mobility can be ranked as follows:
Cu< MATERIALS AND METHODS
Location
Seven monthly samplings from December 2008 Sampling and sample treatment
Materials and sample handling were done in a systematic clean
method. All bottles and containers were soaked in 2 N HNO3 during
at least 3 days. Afterwards they were rinsed thoroughly 3 times with
de-ionised water. All bottles and containers were rinsed 3
times with river water before handling the samples.
To obtain enough SPM for sequential extraction purposes, a
sediment trap was installed in each site. It consisted of a 2 L PVC water bottle
hanged from at least 1 m from the river bank at mid-depth. Two holes of a diameter
of around 4 cm were carved on two sides on the upper side of the bottle with
the two holes posed creating a parallel axe to the river flow. This method
was previously successfully used by Tessier and Bonté (2002) to collect the Seine SPM.
SPM from sediment trap was emptied in a polyethylene bottle on field along
with uplying water. Samples were transported and stored in the dark at 4 °C
before analysis. Storage duration ranged from 2 days to one month before sample
treatment.
Samples were then centrifuged in the laboratory at 3500 rpm
for 20 min for several cycles until all the water are centrifuged and all the
SPM are recovered. It was then freeze-dried for at least 48 h and homogenized
with agate mortar. Samples were then stored in acid-cleaned glass jars in the
dark before analysis.
Solid phase analyses
Bulk digestion
To obtain total metal contents, 0.1 g of SPM was digested
in Teflon vessels under laminar flow hood using a heating block (Digiprep,
SCP Science). The first attack uses 10 mL aqua regia (HNO3 65
%: HCl 30 %, 3:1) for 3 days at room temperature. Excess nitrate and chloride
were evaporated at 90°C and underlying liquid was separated ensuring the
least quantity of HCl remaining in the vessels through rinsing 3 times with
10 mL 0.5 N HNO3. Digestion continued with 10 mL of a HF 48.9 %:
HNO3 65 % mixture (1:1) to attack siliceous minerals during 24h
at room temperature. Sample was evaporated to dryness at 100°C to eliminate
hexafluorosilicic acid. The solid residue was then attacked with 12 mL of a
HNO3 65 % : HClO4 69-72 % (1:1) mixture heated at 120 °C
during 5 days. Final solutions were evaporated near to dryness. 1 mL of 65
% HNO3 was added to the remaining solution that was then evaporated
near dryness. This step was repeated three times. The solutions were then brought
into a 50 mL 0.5 N HNO3 solution. This method was adapted for the
Seine River carbonated SPM, and allows complete SPM digestion. The 3 day aqua
regia step allowed dissolution of the abundant amount of carbonates in
samples. Afterwards, underlying liquid was separated
to avoid re-precipitation of the Ca and Mg fluorides
in HF solution. All solutions were ultrapure reagents
to assure minimum contamination (HNO3 and HCl Normatom grade, VWR
France, and HF and
HClO4 "for trace metal analyses", Baker, from Sodipro France).
BCR sequential extraction
Each BCR sequential extraction was performed on duplicates
of 0.25 g of sediment following Revised BCR with extra rinsing to overcome
the difficulty due to the smaller amount of extracting solution used. Extraction
protocol is summarized in Table 1. Due to a lack of end-to-end shaker, shaking was
performed using a platform orbital shaker during 16 h at 300 rpm (Heidolph
vibramax 100). The speed was chosen to keep the samples well in suspension
without shaking excessively to avoid over-extraction. Due to the relatively
low quantity of samples used, separating the extracted solution without removing
the sample was a difficult task. A second rinsing was applied using half the
normal volume of the same extracting solution before rinsing samples with water.
Each extraction batch was accompanied by a duplicate of the BCR 701 (Bureau
Communautaire de Recherches, Gent, Belgium) certified lake sediment for sequential
extraction.
Analytical procedure
Major and trace metal concentration were determined in total
and sequential extraction samples using Inductively Coupled Plasma Quadrupolar
Mass Spectrometry (ICP-QMS) (X-Series, CCT II+ ThermoElectron, France). ICP-QMS
spectrometer was calibrated using standard solutions and checked for with certified
river water (SRM 1640, National Institute for Science and Technology, Gaithersburg,
USA) routinely during analysis, at least once per day of analysis and once
for every 20 samples at the most. Instrumental drifts and plasma fluctuations
were corrected using internal standards (Re, Rh, and In (SPEX, SCP Science,
France)) for all the metals studied, and Ge for major elements including Ca,
Al, and Mg. To minimise interferences, analysis with the Collision Cell Technology
(CCT) introducing a supplementary gas mixture of H2 (7 %) and He
(93 %) was applied for Fe, Mn, and the 6 metals studied (Cd, Cr, Cu, Ni, Pb,
Zn). Major elemental bulk analysis including Ca, Si, Al, Fe, Mg, S, K, P, and
Ti were performed with micro (50 µm) X-ray fluorescence (XRF) (Microfocus
x-ray source IFG X-1) through measurements of pressed pellets (diameter 0.5
cm,
average weight 0.02 g). Measurements were
calibrated with at least 5 reference materials (USGS Mn
Nodule A1, USGS Marine mud MAG-1, USGS jasperoid
GXR-1, IAEA lake sediment SL1 and IAEA Soil-7) and
sample analysis was done in duplicate to compensate
for possible sample heterogeneity. Particulate
organic carbon (POC) and nitrogen (PON) contents
were measured in 0.25 mg samples previously
decarbonated by 3.4 mL of 1 N HCl. Decarbonation comprises of
4 cycles of addition of HCl solution, 20 mn of 300
rpm orbital shaking, 3000 rpm centrifugation,
solution separation. Samples were weighed precisely
and analysed through a Carbon Hydrogen and
Nitrogen (CHN) analyser (Thermoflash EA 1112 series).
Enrichment factors
Enrichment factors (EF) were calculated by normalizing concentrations
to Al and using local background values established for the Seine river watershed
by (Thévenot et al., 2002) through
measurements of selected river mouth values based on an Al content of 33000
mg/kg. Assembled background values are shown on Table
2 and formulae is shown in Equation 1. The EF values
should be evaluated keeping in mind the shortcomings of this approach (Reimann and de Caritat, 2005; Karbassi et al., 2008).
Here, local background values were preferred to continental crust concentrations,
due to the specificity of the carbonaceous Seine River basin geology.
Where EF where [Me]sample = metal concentration
in sample, [Al]sample = aluminum concentration in sample,
[Me] background = metal concentration in background
value and [Al] background = aluminum concentration
in sample.
Scanning electron microscopy analyses
Cartography of suspended sediment were collected at 6-7 kV
using in backscattered electrons imaging mode on a Zeiss ULTRA scanning electron
microscopy (SEM) coupled with field emission gun (FEG) at the IMPMC, Paris,
France. Energy Dispersive X-ray Spectroscopy (EDXS) data were collected at
the same electron beam-voltage using a BRUKER AXS Si-drift detector. A supplementary
image coupled with its EDS spectrum is also demonstrated. This latter image
was obtained with JEOL JSM 840 SEM coupled to an X-ray microanalysis system
from Princeton Gamma Tech (PGT) at LSCE.
RESULTS AND DISCUSSION
Extraction recoveries, analytical uncertainties and limit
of detection
Extraction and analytical processes were validated using various
standard materials. Extraction recovery (recovery in Table
4) was calculated by comparing the average concentration
of our BCR sequential extraction on 0.25 g (n=4) with the BCR certified values
comprising of Zn, Cd, Pb, Cr, Ni and Cu in the exchangeable, reducible and
oxidisable fraction. Recovery value is accompanied with standard deviation
(SD in Table 4). Analytical detection limits (DL in Table
4) are two times the blank value of sample that
went to the same preparation, extraction, and analytical treatment. Speciation
values were also validated by comparing the sum of the 4 extracted fractions
with
the total digestion results (Table 5). For the trace elements, SD values showed that
Cd and Cu shows a low reproducibility with SD ranging from 13-19 % for Cd and
13-16 % for Cu. Cd concentrations are relatively low, which may be prone to contamination
during the extraction. This is reflected in our recovery levels which are higher
than total digestion values
in downstream sites, 99+12 and 93+19 at Bougival and Triel, respectively.
Ca presented difficulties as it is present in extremely high quantity in the
exchangeable fraction and may often pose difficult
analytical problems.
Nevertheless, the recovery level between the 4 extraction
phases of the BCR and our total bulk extraction depends on the metal's properties
and the site's location. For most cases, more than 90 % of Cd and Pb are extracted
by the BCR extraction, except for Pb in Marnay which has a slightly lower recovery,
at 79+7 %. More than 80 % of Cu, Zn, and Ca are extracted with the BCR sequential
extraction except for Zn at Marnay where recovery compared to the total digestion
is relatively lower, 65+6 %. Zn, Pb, Cr, Ni and Al extracted by the BCR are
significantly higher at Bougival than at Marnay. This shows that Zn, Pb, Cr,
and Ni could be
from natural sources, associated to Al which would
be more resistant to chemical extraction. Cr and Ni
are probably found in the crystalline phases because
they are only half extracted by the BCR extraction, 52+2,
62+6, 58+5 for Ni, and even lower for Cr 29+3 41+5 37+3
at Marnay, Bougival and Triel, respectively. For Cd, Cr,
Cu, Ni, Pb and Zn, these results agree with most
studies using the BCR sequential extraction, although
none specified the uncertainty and detection limits for
each extraction phase and each element. Larner et al. (2008) stated an extraction efficiency
of 70-115 %. Nevertheless, these values were incomparable with those from other
studies because our total digestion protocol attacks almost 100% of the bulk
sediment. Our recovery would then be underestimated compared to recoveries values
in other studies where pseudo-total digestion
were performed (i.e. Larner et al., 2008).
The major elements showed different extractability by the BCR extraction compared
to the total bulk digestion, reflecting the different mineral forms in which
the major element take form. The order of extractability from the element less
extracted by the BCR to the most extracted is Al< Mg < Fe < Mn < Ca.
Enrichment factors
Enrichment factors (EF) of the 7 values at each site were
averaged and shown on Fig. 3.
Values showed significant increase in downstream samples for Zn, Cd, Pb and
Cu. On the other hand, Cr and Ni show no spatial
increase in EF, keeping a relatively steady value of around 2 from Marnay to
Bougival and Triel. These two
elements represent metals with the solid phase
non-enriched by urban activities. In the case of the
Seine River, the evolution of the concentration of the
trace elements is also accompanied by an increase in most
of the major elements (Table
3). The trend of metal enrichment in suspended sediments is similar with
recent studies of metals in the Seine River through analysis of sediment cores
and estuarine mussels (Meybeck et al., 2007; Le
Cloarec et al., 2009). This shows that SPM collected in this study
are representative to the Seine watershed and that these four metals represent
metals affected by anthropogenic sources. Urban activities may increase EF in
two or even seven folds in the downstream sites, depending on the metals. In
average, Cd is the metal most enriched while Zn, Pb and Cu present EF of 6-7.
Metal concentrations at Bougival were not found to be lower than at Triel (Mann-Whitney,
p = 0.983), despite of the Seine Aval WWTP outlet situated between the two sites.
Considering the possible contribution of the WWTP outgoing flow to the metal
contamination in the water column, one could have expected that metal concentrations
in SPM would be higher at Triel than
at Bougival. In 2002, Thévenot et al.
(2002) calculated that particulate flux of Cd, Cu and Pb from the Seine-Aval
WWTP contributed from 2-6 % in the river mouth despite of the relatively low
SPM concentration in the
WWTP outlet water of 28-49 mg/L ((Estebe et al., 1998;
Meybeck et al., 1998; Thévenot et
al., 2007). Through calculations of influx, WWTP treatment efficiency
and
sludge recovery, Thévenot et al. (2007) established
an outflux of 26, 25.5 and 90
t.y-1 for Cu, Pb, and Zn respectively. Buzier et al.,
(2006) also found that WWTP outflux may have a significant impact on the
labile metal flux downstream of the outlet in low flow conditions. Ever since
these studies were published, an effort was made through construction of numerous
WWTP to distribute and relieve the wastewater load of Seine Aval. This finding
indicates that improved WWTP are efficient in reducing incoming metal load to
the Seine River. The analysis of the major elements is an important aspect in
the interpretation of sequential extraction data as these major phases are likely
to be the carrier phase of the trace elements. In order to confirm whether our
extraction corresponds to the operationally-defined fraction, the extracted major
elements for each step
are presented (Fig. 4). At Bougival
and Triel, Ca is more than 90 % extracted on the exchangeable fraction, leaving
less than 10
% extracted in the reducible step and almost no Ca is found in the oxidisable
and residual
phase. Upstream, the Seine shows a slightly
different behaviour where more Ca is found in the
reducible fraction. Unlike Ca, Fe and Al that varies between
the three sites, there is no significant difference in
the distribution of Mg and Mn at the three sites. Mn
takes on a slightly similar behaviour to Ca where around
70 % are found in the exchangeable form while Mg contains more oxidisable and
residual form. Even though Fe and Mn exist as oxides, the distribution
of the two elements is significantly different. Fe is
not released in the exchangeable fraction but rather
during reducible and residual extraction for Bougival and
Triel and mostly in the residual fraction for Marnay.
The domination of Ca at Marnay with an average of 26.5
% reflects the geological condition of the Seine
watershed where 95 % being underlain by carbonate
rocks (Meybeck et al.,
1999). Further downstream, Ca
concentration decreases to an average of 18.5 %
and 15.4 % for Bougival and Triel, respectively, and
replaced by the general increase of other major elements such
as Si, Fe, Mg, P, S, Mn, K and Ti. For example, Si
concentration increases from 10.9 % at Marnay to 17.4 % and 20.6 %,
Fe average concentration increases from 2.1 % to 3.3 %
in the downstream sites. A dramatic increase is observed
for P and S concentrations at Bougival and Triel compared
to Marnay. The impacts of anthropogenic activities on
P and S are less well documented but some studies mentioned sources mostly coming
from
wastewater (Houhou et
al., 2009). The increase of P and S may be due to the increase of
organic matter. Even though on the average, there is no significant difference
between POC at Marnay, Bougival and Triel, but on a monthly basis, Bougival contains
higher POC than Marnay and Triel (Mann Whitney Test for 7 months of samples α=
0.05).
Nevertheless, it is also possible that P and S increase is
due to increasing phosphate, sulfate or sulfide species. The evolution of Al
concentrations in the 3 sites underlines the importance of also interpreting
data normalized with Al through enrichment factor.
Speciation
The average distribution for the 7 monthly sampling of the
6 metals is shown in Fig. 5. The distribution is different for each metal and
site. Detailed speciation of the environmentally relevant trace metals are
presented in Fig. 6. Sediment traps integrate
one month of SPM during which the sample may be subjected to evolution of speciation.
In the course of a month, SPM sedimenting on the bottom of sediment trap may
create reducing conditions, favourizing the oxidation of organic matter and
the formation of reduced forms. This possibility is to be taken into account
during the lecture and interpretation of data. A spatial evolution in the Al
speciation is also observable mostly for the reducible phase. This evolution
of Al would mean that there is an increase of allochthonous Al coming from
the Greater Paris Region. This is also the case for Fe where the reducible
proportion increases in the downstream sites. The similarity of Fe and Al where
both elements tend to be present mostly in the reducible fraction at Bougival
and Triel may mean a similar anthropogenic source. Indeed, Fe and Al are the
two most abundant metals (Luoma and Rainbow, 2008)
thus becoming the two mostly utilized metals worldwide (USGS,
2010). In 2004, 1370 kt of Al were used in France, mostly in the transportation,
construction and packaging sector (AFA). Nevertheless, this anthropogenic Al
possibly present in the reducible fraction only accounts for 30-40 % of total
BCR extractable Al (Fig. 4), which in turn only contributes to 14-15 % of
the total Al in Seine SPM (Online Resource Table
2).
This small proportion of reducible and possibly anthropogenic Al would only
be around 6 % of the total Al and with this uncertainty it is still considered
safe to calculate enrichment factors through normalization by Al. As a whole,
an evolution from upstream to downstream is significant enough while the difference
between Bougival and Triel is less pronounced (Mann Whitney, p= 59 %, 49 %
85 % between Marnay to Bougival, Marnay to Triel and Bougival and Triel, respectively,
with H0 hypothesis is proportion site 1 = proportion site
2). Regarding temporal variation, the distribution at a given site is
relatively stable with deviation of around 5 % for most metal in certain fractions
(n=7), and with a maximum of 10-15 % for some metals.
Speciation evolution from Marnay to Bougival and Triel was
observed for Zn, Cd, Pb and Cu but not for Cr and Ni. Moreover, these four
metals are those demonstrating a significantly higher EF downstream. This implicates
that anthropogenic contamination is likely to bring in material with different
metal distribution and that variation in the physico-chemical condition downstream
leads to a different type of speciation. The general observation allows us
to divide the six metals into 2 distinct groups. The first group constituting
of Zn, Cd, and Pb, averages more than 60 % in the reducible fraction and even
reaching up to 90 %. The reducible fraction for these three elements is always
significantly higher than the 3 other fractions (t test, α=0.05). These metals
are known to be
preferably associated with iron oxy-hydroxides (Luoma and Bryan, 1981) which would explain the high proportion
found in the reducible fraction. The increasing proportion of these reducible
metals rises significantly from upstream to downstream, marking a possible anthropogenic
impact. This idea is also supported with the increasing EF as the water reaches
Bougival and Triel. The increase in EF is accompanied with the increase in the
reducible fraction for Zn, Pb and Cu but in a lesser extent for Cd which shows
that for this metal, the metal physico-chemical properties also play an important
role in determining the distribution. The second group consists of Cr, Ni, and
Cu with higher fractions of the residual phase than the first group, averaging
around 35 %. Compared to the first group, the metals in the second group are
distributed relatively equal where no fraction holds more than 60 % for a given
metal, except for residual Cu at Marnay. While in a first approach, the metals
can be divided into two groups, each metal shows a typically different behaviour
and it will be interesting to discuss the distribution of each particular metal.
From the first group, Zn seems to stand out from Cd and Pb in terms of association
to the exchangeable fraction (Fig. 5). The proportion of Zn associated to this phase
remains around 20 % from Marnay to Triel. Despite the stability in its proportion,
this means an increase of 4-6 folds in concentration going from an average of
9 mg/kg
at Marnay to an average of 63 and 45 mg/kg at Bougival and Triel respectively.
Moreover, the relative stability of the proportion of Zn associated to the exchangeable
phase is seen in a smaller proportion for oxidisable Zn. The proportion remains
around 10 % while the concentration increases 2-4 folds in most cases. Except
for July 2009, the average proportion of exchangeable and oxidisable Zn is relatively
stable from Marnay to
Triel. Distinct changes are observable with
Zn associated to the reducible phase (student test confidence interval 99 % between
Marnay and the two downstream sites). Average reducible Zn
increased from upstream to downstream with 52+9 % at
Marnay to 67+6 % at Bougival and 67+8 % at Triel. This is
the result of a 4 to 7 time increase of concentration
from Marnay to Bougival and Triel. Consequently,
the reducible fraction accounts for at least 50 % of
BCR extractable Zn at Marnay and up to 75 % for
Bougival and Triel. Garnaud et al. (1999) also found a majority of
reducible and exchangeable Zn particularly in SPM sampled during a rain period
in the le Marais catchment outlet of the Seine River. A weak increase of Zn in
the residual form from upstream to downstream also marks a relatively steady
amount of residual Zn transported in the Seine River which could originate from
the
Seine's geological background. It is difficult to compare Seine
Zn distribution with other urban river studies
because Zn in the Seine river watershed is particular
compared to other urban rivers in the world. Unlike
most watersheds, roof runoff constitutes one of the
major sources of Zn contamination in the Greater Paris
Region, where it reaches roughly 40 % of all roofing
surfaces (Robert-Sainte et
al., 2009). 85-100 % of Zn washed from roof surfaces are in its ionic
form (Heijerick et al.,
2002). Zn has a high affinity for solid particles and so it is easily
adsorbed mostly to carbonates, hydrous
iron oxide and silicate minerals (Fujiyoshi et al.,
1994). Nevertheless, these Zn-bearing particles undergo physico-chemical
evolution and once found in the sewer system, 70 % of Zn is found to be associated
with organic matter and biofilm (Rocher et al.,
2004). However, our analysis showed that Zn in the SPM is far more likely
to be associated with the reducible form, with an average of more than two thirds
of the total
BCR extractable Zn for Bougival and Triel. This means that
the sediment undergoes further transformation until reaching the water body.
Our result is comparable to the lower median percentage of acid-soluble Zn
found in the urban streams of Prague (Hnatukova et al.,
2009). The dominating reducible phase of Zn is to a further extent similar
for Cd. Cd is found to be associated up to 80 % with the reducible phase increasing
the concentration to 10 folds going from 200 µg/kg of
Cd to around 2000 µg/kg from upstream to downstream. Although the relative
distribution of the 4 phases remains relatively stable, the absolute concentration
increases to 4-6 times from Marnay to Triel, and increasing more labile Cd in
the exchangeable and reducible fraction simultaneously. The exchangeable proportion
is higher in Bougival and Triel, in the 5-10 % range. Nevertheless, along with
Zn and Ni, Cd still represents one of the metals with the highest proportion
associated to the exchangeable phase making it relatively more mobile to the
environment than other metals, because the exchangeable phase is the phase most
likely to go into the dissolved phase thus becoming more available to the environment.
The high proportion of reducible Cd and Zn in the Seine River SPM differs from
previous urban river studies
which showed mostly acid-soluble fraction. Hnatukova et al.,
(2009) found 38-64 and 15-43 % acid-soluble Cd and Zn, respectively, in
the urban streams of Prague as opposed to the acid-soluble Cd and Zn in Seine
measuring at 3-11 % and 8-28 % respectively. This makes Cd one of the most mobile
element compared to the 6 studied elements showing the highest proportion associated
to the mobile phase (exchangeable+ reducible+ oxidisable), with an average of
2 % associated to the residual phase. Comparing the six studied metals, Cd shows
the highest EF increase from upstream to downstream, increasing 3 to 5 folds
compared to an average EF increase of 2 to 3 folds for Zn and Pb in the downstream
sites. This significant Cd increase in the bulk concentration is not accompanied
with an evolution of the average distribution
from upstream to downstream (Fig. 5). This is not the case for two urban watersheds,
the St Lawrence
River, Canada (Gagnon et
al., 2009)and the Loura River, Spain (Filgueiras et al.,
2004) which exhibited significant spatial variation of Cd solid speciation.
Evolution for the Seine River solid SPM is only observable once comparing monthly
variations when the proportion of exchangeable Cd along with exchangeable Zn
increases
in July 2009. This phenomenon could be attributed to the low
flow occurring during the SPM accumulation for this sample (22 June to 22 July,
2009), illustrated
in Fig. 2, but also the transport of first flush
runoff from the storm rain taking place between the 16 and 17 July increasing
the river flow rate from 126 to 189
m3/s. This runoff discharge may consist of particles with different
speciation as opposed to the low flow-particles which would explain the abrupt
increase of exchangeable Cd and Zn. The high EF of Cd and the relatively high
proportion of Cd in the mobile fraction make Cd an important element to monitor
in the environment. Nevertheless, despite of this high enrichment of the Cd downstream
of the Seine River, Cd shows a stable distribution in the 4 fractions from upstream
to downstream. The speciation of Cd found downstream is not similar to Cd speciation
of solid matters collected in the basin such as traffic aerosols containing 75
%
of exchangeable Cd (Lebreton and Thevenot,
1992)or road dusts containing more than 30 % of oxidisable
Cd (Thévenot et
al., 2002). This implies that urban Cd would be mobile and solid Cd
entering the river would be directly remobilized to its preferable fractions
in the solid phase, despite of its original form. This would make speciation-based
source tracing in the solid form unadoptable for Cd in the Seine River. Pb showed
the highest increase of the reducible fraction relative content from upstream
to downstream (20-30 %) and a 10-15 fold the concentration increase. This is
by far the highest spatial increase of the reducible fraction compared to Zn
and Cd. Compared to Zn and Cd, there does not seem to be a significant proportion
of exchangeable Pb compared to the four other fractions. In the downstream sites,
the three remaining fractions, exchangeable, oxidisable and residual only makes
up 10 % of the total extractable Pb. This means that further studies on Pb contamination
has to be focused on the reducible fraction, containing mostly Pb associated
with iron and manganese oxides. Nevertheless, our results remain similar to other
studies of metal fractionation in urban watersheds. The preference of Pb for
the reducible
phase is also observed by Sutherland and Tack
(2007) and Carter et
al. (2006). They found a high association of Pb with the Mn oxide
and to a lesser extent with the
Fe oxide. Hnatukova et
al. (2009) also found Pb to be mainly bound to the reducible fraction. Jain et al. (2008)
also found only 1-3 % of exchangeable Pb in the sediments of the River Narmada,
India. The weak association of Pb with
the exchangeable fraction is equally observed in
the study by Carter et al. (2006). Nevertheless,
the Seine oxidisable Pb proportion seems to be underestimated compared to their
study as fractionation of Pb is mainly dominated by the reducible fraction. The
study of Hassellov and von der Kammer (2008) strongly suggests that iron-oxide
Pb bearing particles are in the form of nano-colloids. Consequently, Pb may be
efficiently transported to the estuary. The Seine River is located in a carbonated
basin. Consequently, SPM contains abundant carbonates onto which the metals could
be adsorbed. Nevertheless, this is not the case for the Seine River where the
absence of exchangeable Cu, along with Cr and Pb, is notably similar to the distribution
in
the Aire River (Carter et
al., 2006). The predominant species in the range of pH of the Seine
River measured during the campaign (between 7.5-8.3) would be
CuCO3(aq) and
Cu(CO3)22- (Stumm and Morgan, 1981)which means
that CuCO3(s) is not present even with abundant
CaCO3 in the system. Along with Zn, Cu seems to be the element with
a mobile phase evolving considerably from upstream to downstream, where its proportion
could reduce 20 % the proportion of the residual phase. This would mean a higher
mobility downstream, and it would also mean a high anthropogenic contribution.
Similar to Cd, Ni is distributed steadily from Marnay to Triel, around 15, 20,
30, and 40 % for the exchangeable, oxidisable, reducible and residual phases
respectively. Ni displays a minimum spatial increase in absolute concentration
from upstream to downstream, reflected by the values of Ni EF. These two evidences
may mean that Ni sources mostly originate from lithogenic sources rather than
anthropogenic contamination. Ni seems to be the metal containing in average the
highest metal proportion in the residual fraction ranging around 31-47 %. The
relatively steady proportion of the residual Ni from upstream to downstream would
signify the steady contribution of lithogenic background, reflected by the steady
values of enrichment factor from upstream to downstream. The relatively high
exchangeable Ni is somehow comparable to that of Zn but while reducible and residual
Zn varies considerably from upstream to downstream, reducible and residual Ni
remain stable and do not show significant spatial variation. Therefore, the high
exchangeable Ni cannot be contributed to anthropogenic sources, but more to the
typical geochemistry of Ni to the solid phase. Among the elements found in the
second group, Ni shows the highest exchangeable phase, which is not the case
for other fractions. Similarly to Cd and Zn, Ni is found to be already significantly
associated with
the exchangeable phase beginning from the upstream
site. The presence of a significant exchangeable phase in
Ni, Cd and Zn is not at all apparent in Pb, Cr, and Cu.
The grouping of Ni, Cd, and Zn was observed by (Tusseau-Vuillemin et al.,
2005)on Seine River SPM. Based on a multi-elementary study, they found
a correlation between the ratio of the dissolved and solid
fraction (Kd) of Ni, Cd and Zn. This would indicate similar adsorption-desorption
behaviour for Ni, Cd and Zn. As mentioned above, a recent study observed a high
variation of dissolved Zn in the Seine. Zn similar behaviour with Ni and Cd would
imply that pulsating concentration of dissolved Ni and Cd could also be a problem
in the Seine River. This should be further investigated as Cd and Ni are considered
even a more toxic element and regulated by the European Water Directive. Similar
to Ni, Cr displays a constant EF for the 3 sites. Nevertheless, what distinguishes
Cr from Ni is that Cr seems to even be less mobile with relatively no exchangeable
phase present. This is also supported by the low BCR-extractability of Cr, representing
more than 60 % associated with the non-extractable phase indicated by the difference
of the total Cr of the four BCR extracted fractions and the total Cr obtained
by the bulk extraction. This would indicate as Cr being mostly incorporated in
mineral particles, relatively difficult to extract. Chromite is regularly found
in the Seine SPM through analysis by Scanning Electron Microscopy (SEM) (unpublished
work) and this may be a possible mineral form of Cr. This is an evidence of the
importance of completing total digestion with metal speciation study to understand
its mobility to the environment. Compared to the 5 other elements, Cr seems to
be an element that is mostly associated with the oxidisable phase, averaging
about 40-50 %. The strong preference of Cr with organic matter was mostly observed
in an anoxic estuary (Du Laing et al. 2009). The high capacity of Cr complexation
with
the organic matter is also observed in bed sediments (Lin and Chen, 1998).This
would mean that Cr could be ingested by organisms consuming the organic matter.
Despite the stable enrichment factor
from upstream to downstream indicating possible lithogenic origin of Cr, a shift
of distribution, is apparent from Marnay towards Bougival and Triel. There is
a 10 % decrease of Cr in the residual phase, replaced by an increase in the reducible
form. This shows that Cr evolves more considerably than Ni inside the solid phase
and is more labile to physico-chemical changes. The abundance in the oxidisable
phase is also apparent at a lesser extent for
Cu. The oxidisable Cu represents 30 % as compared to Cr going
up to 50 %. A concentration increase of 5-20 times is observed, the highest
upstream-downstream increase of oxidisable Cu compared to other
metals. Luoma and Rainbow
(2008) noted the strong affinity of Cu to organic ligands. More recently,
various studies in the Seine River also noted this characteristic which would
be due to the presence of urban dissolved organic matter. It presents different
complexing capacity as compared to the natural organic matter, thus creating
a higher affinity for the dissolved copper (Pernet-Coudrier et al.,
2008). Such colloidal organic matters could possibly be collected in a
monthly trap. The formation of bio-film inside the trap could also be an equally
important Cu-collecting mechanism. As for Zn and Pb, reducible Cu also shows
an increase in proportion going from 12 % at Marnay to 30-40 % at Bougival and
Triel. This abrupt apparition of reducible Cu could be attributed to a source
related phenomenon where average EF increases 3-5 folds from Marnay to Bougival
and Triel. Cu introduced to the river could be associated with the reducible
fraction, especially the iron and Mn
oxides. Luoma and Bryan
(1981) noted Cu preference to iron oxy-hydroxides, although in our cases,
the proportion of reducible Cu is relatively low as compared to reducible Zn,
Cd and Pb ranging about 40 %. A higher reducible fraction is observed at Triel
on February 2009 where it reaches up to 70 %. This sudden increase of reducible
Cu on February 2009 coincides with the increase of reducible Zn, Cd, and Pb for
the same period. During this period, the Seine River flow rate at Bougival more
than tripled in 11 days from 157 to 528
m3/s. The flow decreased by an average of 200
m3/s during the next 9 days and re-increased to 505
m3/s in 7 days. These increasing flow rate episodes due to rain runoff
would include urban runoff with specific source-related characteristics. As during
this period, a significant increase in reducible Zn, Cd, Pb and Cu is observed,
the 2 rain episodes were likely to transport urban pollutants associated in the
reducible fraction, mostly iron oxides. Images of the Triel August 2008 suspended
sediment sample using the SEM-FEG demonstrated iron oxide as a cluster with a
geometrical crystal structure (Fig. 7a inset). The Ca detected behind the iron
oxide cluster could indicate a calcite-hosted iron-oxide growth as suggested
by the
elemental cartography (Fig. 7a). The similar composition of iron oxide with Ca
was also observed by SEM-EDS on bed sediment collected
on 2001 (Tessier and Bonté,
2002). Iron oxide particles were
found as cohesive particles and proved to be
an effective metal scavenger (Hochella and White, 1990; Morin et
al., 2009; Sekabira et al., 2010) found to be associated with
Cu, Pb, Sb and Zn.
Sulfidic species
The grouping of Zn, Cd, and Pb along with relatively mobile
Cu would support the hypothesis of possible sulfidic forms in the samples.
These four elements are known as chalcophile elements in the Geochemical Classification
of Goldschmidt and preferably bond with sulphur rather than oxygen. Indeed,
according to Larner et al. (2008), sequential
extraction of samples containing sulfidic phases in oxic conditions would lead
to redistribution of Cd, Zn, Cu and Pb to a lesser extent from the oxidisable
to the reducible phase. This study explained that during the exchangeable extraction,
sulfidic phases may be oxidised and redistributed to the reducible phase. No
previous studies showed a significant amount of sulfidic phases in oxic riverine
SPM non-affected by mining activities (Taylor and Owens,
2009). Therefore no special care was taken in preserving oxidation state
during the sequential extraction procedure. This study showed that with the
minor oxidisable phase, and the extremely high reducible fraction, a considerable
amount of the studied metals could be associated with the sulfidic phases.
Urban impacts on metal speciation
The proportion of the reducible fraction seems to reflect
the enrichment factor of the metal. Metals affected by anthropogenic activities
having a relatively high enrichment factor (EF>3), seems to prefer association
with the reducible form. Obviously, the proportion of the residual fraction
reflects the contribution of the background. The oxidisable fraction also seems
to show, to a lesser extent, contribution by natural sources as it is exceptionally
high for Cr, reaching up to 50 % compared to the other fractions. Ni, a non-enriched
metal in the Seine River, also contains a relatively high oxidisable fraction,
ranging around 23 % at Marnay, and 11-23 % at the downstream sites. It seems
that the immediate mobility is minimized for most of the metals, reflected
by their small proportion in the exchangeable form. Two downstream sites were
chosen in order to measure impacts of Greater Paris at Bougival and of the
wastewater treatment plant (WWTP) Seine Aval at Triel. Even though at Bougival
the metal concentrations are
significantly higher than at Triel, no significant
difference is remarkable in their distribution between the
four phases for any of the six metals. The addition of
numerous WWTP and the redistribution of incoming
wastewater to other plants in the course of the Seine River
may prove to be effective in reducing the metal load
flowing in the Seine River at Bougival. Nevertheless, the
issue of reducing the metal load is still important for
Greater Paris where enrichment from Marnay to Bougival
remains considerably high.
CONCLUSION
In this study, applying the BCR sequential extraction to suspended
particulate matter sampled monthly in the Seine River, a river impacted by
human activities, the highly urban Greater Paris Region is found to modify
not only metal concentration but also metal speciation as observed in speciation
evolution between Marnay
vs Bougival and Triel. As Cd, Cu, Pb and Zn
are enriched in the suspended matter, their
reducible fraction is found to greatly increase. Based on
the speciation behaviour, the analysed metals can
be divided into two groups; the first group contains
the anthropogenic metals, Cd, Pb, and Zn, in which
the reducible fraction accounts for more than 60 % of
the total BCR extractable metals for the downstream
Paris sites. The second group includes Cr, Cu, and Ni that
are associated with at least 15 % in three of the four
defined BCR fraction in the downstream Paris sites.
Exchangeable fraction is only significant for Cd, Ni and Zn while
the oxidisable fraction accounts for less than 20 % for
the anthropogenic metals downstream except for Cu.
The enrichment of Zn, Pb and Cu by the Greater Paris
Region seen at Bougival and Triel is accompanied by
the increasing distribution of metals on more mobile
phases including the exchangeable and reducible phases. This
fraction is more mobile as the metals can be released to the
dissolved phase easily with pH variation, thus making the metal more bio-available.
Temporal variation in the speciation is found to be related with discharge
variations. No impact from the waste water treatment plant was observed, neither
on the trace metal
concentration, enrichment factors nor speciation.
This study suggests that a considerable amount of the metal
studied could be associated with sulfidic phases which will be investigated
in further studies. Nevertheless, the possible formation of sulfidic phases
in the SPM accumulating during one month in the trap must be considered.
ACKNOWLEDGEMENTS
The authors would like to thank A. Bourgeault and the HBAN
team in Cemagref Antony for sampling assistance, E. Robin for XRF analysis,
N. Tisnerat-Laborde for decarbonatation discussion, C. Gautier and C. Hatté for
POC and PON measurements, E. Douville for ICP-MS assistance, D. Thévenot
for advice and corrections and J.-M. Mouchel, G. Morin and C. Quantin for helpful
discussions. The Marnay-sur-Seine botanical garden and the SIAAP are thanked
for giving access to the sampling sites located on their ground. For the acquisition
of SEM data, we would like to thank A. Elmaleh and I. Estève at IMPMC
and E. Robin at LSCE. We are also grateful to R. Hochreutener (MSc student)
and to her research advisor G. Morin (IMPMC) as for organizing and preparing
SEM analysis. The study is a contribution of the interdisciplinary research
program on the environment PIREN-Seine and was also funded by the the Continental
and Coastal Ecosphere French research program (EC2CO INSU). Cindy Priadi acknowledges
the French Foreign Affairs Ministry and PIREN Seine for her PhD scholarship.
This is LSCE contribution number 4055.
REFERENCES
© IRSEN, CEERS, IAU |
|