Article pubs.acs.org/est
Spatial Heterogeneity and Kinetic Regulation of Arsenic Dynamics in Mangrove Sediments: The Sundarbans, Bangladesh Mahmud Hossain,† Paul N. Williams,‡ Adrien Mestrot,† Gareth J. Norton,† Claire M. Deacon,† and Andrew A. Meharg*,† †
School of Biological Sciences, Cruickshank Building, University of Aberdeen, St Machar Drive, Aberdeen AB24 3UU, United Kingdom. ‡ Lancaster Environment Centre, Lancaster University, Lancaster LA1 4YQ, United Kingdom S Supporting Information *
ABSTRACT: The biogeochemistry of arsenic (As) in sediments is regulated by multiple factors such as particle size, dissolved organic matter (DOM), iron mobilization, and sediment binding characteristics, among others. Understanding the heterogeneity of factors affecting As deposition and the kinetics of mobilization, both horizontally and vertically, across sediment depositional environments was investigated in Sundarban mangrove ecosystems, Bengal Delta, Bangladesh. Sediment cores were collected from 3 different Sundarbans locations and As concentration down the profiles were found to be more associated with elevated Fe and Mn than with organic matter (OM). At one site chosen for field monitoring, sediment cores, pore and surface water, and in situ diffusive gradients in thin films (DGT) measurements (which were used to model As sediment pore−water concentrations and resupply from the solid phase) were sampled from four different subhabitats. Coarse-textured riverbank sediment porewaters were high in As, but with a limited resupply of As from the solid phase compared to fine-textured and high organic matter content forest floor sediments, where porewater As was low, but with much higher As resupply. Depositional environment (overbank verses forest floor) and biological activity (input of OM from forest biomass) considerably affected As dynamics over very short spatial distances in the mosaic of microhabitats that constitute a mangrove ecosystem.
■
INTRODUCTION
OM influences As speciation and, therefore, mobility, including As sorption/complexation11,12 and As biomethylation.13 This study focuses on the Sundarbans, the world’s largest mangrove ecosystem,1 to ascertain spatial patterns (vertically down the profile and horizontally across the surface) in As deposition and dynamics, and how these parameters are associated with known drivers of aresnic's biogeochemical cycle (C, N, P, Fe, and Mn), as well as particle size analysis. The study comprises two sections, one focusing on the characterization of As deposition and mobilization at 3 geographically separate Bangladesh sites, and the other focusing on a more detailed assessment of arsenic's spatial distribution and mobilization at one location. Kinetics of As mobilization in the field at this one site were assessed by deploying diffusion gradients in thin-films (DGT) and Rhizon porewater samplers, with subsequent As speciation characterization of porewaters, in
Mangrove ecosystems, as they are often situated on the fringe of deltaic floodplains, are a potential large depositional store of sediment-associated arsenic (As), and it is important to assess the mobility of As in such environments.1−3 As mangrove ecosystems are highly biologically active,4 with considerable spatial heterogeneity with respect to subhabitats (overbank sediments, deep forest, forest fringe, network of tributaries),1 as well as subject to seasonal changes in salinity and diurnal flooding,2,3 it is expected that Mangrove As sediment dynamics will be spatially and temporally complex.1−3,5−7 Furthermore, sediment particle size distribution, which relates to sediment surface area, is known to be a key driver in sediment As dynamics, along with organic matter (OM) and iron geochemical cycles.8 Generally, it is supposed that the oxidation of OM with concomitant reduction of As-bearing iron (Fe) minerals causes mobilization of As and Fe in sediments.8,9 Mangrove ecosystems have considerable OM inputs,4 as well as complex grading of sediment particle size,10 both of which could profoundly affect As cycling.8,9 Moreover, © 2012 American Chemical Society
Received: Revised: Accepted: Published: 8645
April 4, 2012 July 24, 2012 July 27, 2012 July 27, 2012 dx.doi.org/10.1021/es301328r | Environ. Sci. Technol. 2012, 46, 8645−8652
Environmental Science & Technology
Article
Figure 1. Total As, TOC, TIC, total Fe, total Mn, and total P in sediments plotted down the profile at 3 sampling sites of the Sundarbans, Bangladesh. Data are means and ± standard error (SE) (n = 2).
subhabitats of the swamp representing overbank, forest floor, and tributary sediments, as DGT and porewater measurements, along with dissolved organic matter (DOM) content of those porewaters, have previously been shown to effectively model (porewater equilibrium concentrations and solid phase resupply) As sediment dynamics.11 The overall objective of this study was to investigate the factors regulating As loadings and dynamics in the important Sundarbans sediment depositional environment.
The cores were then cut into 5-cm lateral slices and air-dried. Cores were located on the forest floor, 5 m in from the treeline. Detailed Characterization of One Site, Mongla. A sampling campaign was conducted in December 2009 at Koromjal, Mongla at 22°25'56″ N 89°35'43″ E (Figure SI 2). This site was the same forest reserve sampled by Meharg et al.1 and near the Mongla site reported in Figure SI 1. Sampling locations within the study site were chosen to cover a range of environments: (i) main river bank (Pasur River), (ii) a major tributary ∼10 m across, (iii) a minor tributary ∼1 m across, and (iv) forest floor, flooded by tidal water up to ∼15 cm diurnally during December, with Sundari trees (Heritiera fomes) as a dominant species). Rhizon samplers (details in SI) were inserted vertically into the ground to collect porewater from 0−5 and 5−15 cm depths throughout the four sampling zones in triplicate with samplers deployed 2 m apart using a triangular sampling design with the same distance among the replicates. DGT, core, and sediment
■
MATERIALS AND METHODS Multisite Survey and Experiments. Sediment cores were collected from 3 Sundarban locations: Mongla, Bagerhat at 22°23′54.8″ N, 89°39′29.1″ E; Dacope, Khulna at 22°26′13.3″ N, 89°35′20.1″ E; and Shyamnagar, Satkhira at 22°11′23.3″ N, 89°04′51.9″ E (Supporting Information (SI) Figure SI 1). Two undisturbed sediment cores were collected from each sampling site to a depth of 80 cm using PVC pipe (5 cm in diameter). 8646
dx.doi.org/10.1021/es301328r | Environ. Sci. Technol. 2012, 46, 8645−8652
Environmental Science & Technology
Article
(changing from 2.6 mg/kg As at the surface to 3.0 mg/kg As at 35 cm).3 Fine grain sediment content (fine silt + clay, < 6 μm, method described in Text SI) had significant correlations with sediment As concentration in all 3 locations (SI Figures 3 and 4). Such clay/silt−As concentration relationships are a consequence of large surface areas with which to bind As.3,8 Average TOC content of the sediments from the 3 locations ranged from 0.4 to 0.7% (Figure 1), consistent with studies on the Indian Sundarban.3 Analysis of variance showed significant variation in TOC between cores and between depth, with the location × depth term also being significant (p < 0.001). This significant interaction term is due to TOC decreasing with depth at Mongla (r = −0.532; p = 0.002) and Dacope (r = −0.621; p < 0.001), while increasing with depth at Shyamnagar (r = 0.338; p = 0.067). Total inorganic carbon (TIC) concentrations increased with depth in Mongla (r = 0.309; p = 0.086) but decreased in Dacope (r = −0.621; p < 0.001) and Shyamnagar (r = −0.595; p = 0.001) (Figure 1). Arsenic concentration down the profile had a negative correlation with C:N at Dacope (r = −0.445; p = 0.011) and Shyamnagar (r = −0.34; p = 0.066), but weakly positive for Mongla (r = 0.214). Arsenic was not significantly correlated with TOC, TIC, or TN at any site (Table SI 1). This contrasts with the positive correlations for TOC seen in deeper aquifer cores from floodplain sediments.1 Negative relationships between total As and organic carbon were reported for four sites in the Sundarbans, West Bengal.3 Organic carbon content is not a strong predictor of total As in Sundarban surface sediments. Analysis of variance found that Fe and Mn concentration in sediment varied significantly with sampling depth and location but their interaction term was not significant (Figure 1). A strong positive correlation between As and Fe for all 3 locations was found (SI Table 1), indicating that As concentrations are controlled by Fe diagenesis and suggesting its oxides are retaining As in the Sundarban sediment as observed widely for Holocene Bengal Sediments.8,9 There was also a positive significant association between As and Mn in the core sediments in Mongla (r = 0.578; p = 0.001) and Dacope (r = 0.726; p < 0.001), but not in Shyamnagar (r = 0.117). Sediment P concentrations varied significantly with core location but not with sediment depth or interaction term. Phosphorus had a negative correlation with As at Mongla (r = −0.37; p = 0.037) and Dacope (r = −0.158), yet the relationship was positive at Shyamnagar (r = 0.311). To explore As−Fe relationships within sediments in more detail, sequential extraction was conducted (method described in SI) and found that nonspecifically adsorbed and specifically adsorbed As together represented 12 ± 0.7% of the total sediment As (Figure SI 5). Amorphous Fe oxyhydroxide (FeOOH) bound-As is the dominant fraction, constituting 53 ± 2.0% of total sediment As, and crystalline FeOOH bound As was 19 ± 2.0%. FeOOH bound As is readily mobilized during sediment reduction.9 The BGS8 concluded that As in Bengal delta aquifer sediment porewaters is derived from iron oxides. Total As and Fe measurements, as well as sequential extractions of As,, indicated that Fe associations were the primary driver for As cycling in Sundarban sediments. Spatial Distribution of As at Mongla. For the detailed study at Mongla, total As concentrations in sediment varied significantly (p < 0.001) among the four sampling zones, but there was no significant (p > 0.05) trend in total As down the profile (Figure 2); again unlike the Indian Sundarban survey of
trap locations were also based on this triangulated design, with all sample types collected within proximity to each other. Porewater sampling was carried out every 24 h for 2 d using 20-mL plastic syringes. Surface water was collected at the end of low tide from the river and major tributary and from the forest floor and minor tributary from ditches therein. Tidal seawater was collected from the same point in the river at high and low tides for 2 d. All the surface and tidal seawater samples were 0.45-μm filtered and divided into batches. One batch was acidified to 1% HNO3 , EDTA was added to a final concentration of 5 mM for metal analysis, and transferred into centrifuge tubes wrapped with aluminum foil. The other batch was preserved as such for dissolved organic carbon (DOC) analysis. Water samples were kept in a cool box while in the field and then transferred to a freezer. DGT devices with a ferrihydrite binding phase formed by in situ precipitation within a hydrogel14 are proven to be effective for measuring average fluxes and concentrations (hours to days) of As in soils.11 Here the cylindrical DGTs (exposure window 2.54 cm2) were pushed gently and smoothly into the sediment to 5-cm depth and covered with sediment, then removed 24 h later.15 This was the first in situ deployment of DGT in Bangladesh soil. Synthetic grass turfs (30 cm ×30 cm) were placed on the riverbank and minor tributary in triplicate to trap suspended particles in tidal water and collected for 2 tidal cycles. Sediment cores were collected using PVC pipe (5 cm in diameter) from each of the four sampling sites. Sediment cores were cut into 5cm lateral slices and air-dried along with other sediment samples. Laboratory Analysis. Full analytical details are given in the SI. Briefly, soils samples were digested in concentrated nitric acid using a block digester. Resulting digests, river water, seawaters, and pore waters were analyzed for As and other elements by ICP-MS with collision cell in operation to remove any ArCl interferences. As speciation was conducted by anion exchange HPLC-ICP-MS. A subset of soil samples were digested for total P analysis using a sulphuric acid, lithium sulfate, hydrogen peroxide mixture, with total P determined by flow injection colorimetry. A further subset of sediments was ball-milled (Retsch, Germany) and analyzed for total carbon (TC), total organic carbon (TOC), and total nitrogen (TN) by flash combustion. An aqueous C analyzer determined DOC in water samples. Total Fe and Mn in water samples were analyzed by flame-AAS. DGT analysis and calculation equations are given in the SI, as are the statistical approaches used.
■
RESULTS AND DISCUSSION Comparison of the Three Sites. Despite only modest differences in As concentrations, two-way analysis of variance showed the variation among the 3 locations to be significant (p < 0.001). Average As concentrations in Dacope (5.6 ± 0.3 mg/ kg) and Shyamnagar (5.5 ± 0.1 mg/kg) sediments were significantly higher than those in Mongla sediments (4.4 ± 0.2 mg/kg). In contrast, Mandal et al.2 and Chatterjee et al.8 reported a large spatial variability in sediment concentration, 0.6−18.3 mg/kg As, in the Indian Sundarban. Here, interaction of sampling depth and the depth by location was not significant; core As profiles are shown in Figure 1. This contrasts with four sites in the West Bengal (India), Sundarbans where As decreased down the sediment profile (0−35 cm), though this increase was relatively low 8647
dx.doi.org/10.1021/es301328r | Environ. Sci. Technol. 2012, 46, 8645−8652
Environmental Science & Technology
Article
Figure 2. Total As, TOC, TIC, total Fe, and total Mn in sediments plotted down the profile at four sampling sites at Koromjal. Data are means and ±SE (n = 3).
Mandal et al.3 Here, the minor tributary sediments, on average, contained the highest As (10.1 ± 0.2 mg/kg), followed by forest floor (8.5 ± 0.2 mg/kg As), major tributary overbank (8.0 ± 0.1 mg/kg As) and main riverbank (6.1 ± 0.1 mg/kg As) sediment. Sediment total As concentrations at this site showed general consistency with a previous study also located here, though note that the previous study was only on bulked surface (0−10 cm) samples.1 Suspended sediment particles that settled on the forest floor (collected after 2 tidal cycles), had higher total As concentrations (13.1 ± 1.0 mg/kg) compared to sediments deposited on the riverbank (7.3 ± 0.3 mg/kg) (Table SI 2), indicating either (a) differential sediment sources or likely (b) differential settling of inputted sources, i.e. finer grained sediments have higher arsenic concentrations as
discussed below. The settled sediment may be entirely fresh or mixed with suspended particles from previous deposits. There was a significant change in TOC, on average, between sampling zones, increasing from nonvegetated riverbank to vegetated forest floor and minor tributary sediments (p < 0.001). The minor tributary sediment contained the highest TOC (2.04 ± 0.14%), while the forest floor sediment TOC (1.66 ± 0.08%) was higher than the riverbank (1.1 ± 0.05%) and major tributary sediment (0.81 ± 0.03%) (Figure 2); all are higher than for other Sundarban sites where 0.5% organic carbon is typical.3 Moreover, suspended particles deposited onto the forest floor had 3-fold higher TOC (2.08 ± 0.10%) compared to sediment particles deposited on the river bank (0.61 ± 0.02%), again suggesting different sediment sources/ settling/mixing (Table SI 2). Differences in the total As and 8648
dx.doi.org/10.1021/es301328r | Environ. Sci. Technol. 2012, 46, 8645−8652
Environmental Science & Technology
Article
Figure 3. Arsenic, DOC, Fe, and Mn concentrations in porewater collected from four sites at Koromjal. Data are means and ± standard error (SE) (n = 3).
TOC concentrations in trapped deposited sediment may also be due to particle size as forest floor sediments have a finer particle size distribution than overbank sediments (Figure SI 6). Unfortunately, not enough sediment was trapped on the synthetic turf to allow particle size analysis. TOC concentration was found to increase with sampling depth (p = 0.001) at riverbank, whereas, no such relationship was apparent in the other 3 sites (Figure 2, Table SI 3). Data from all four sampling sites show no clear correlation between sediment As and TOC in cores sampled up to 80 cm depth, which is further confirmed by regression analysis (Table SI 3) and supports the finding for the wider geographical core study (Figure 1). The carbonate (TIC) concentrations were negatively correlated with sediment depth in riverbank sediments (p = 0.002) but in the other 3 sites the correlations
were inconsistent. One-way ANOVA showed significant variation in sediment TIC concentration among the four sampling zones (p < 0.001) with the order being major tributary (0.39 ± 0.03%) > river bank (0.29 ± 0.05%) > forest floor (0.20 ± 0.04%) > minor tributary (0.08 ± 0.02%). The relationship between TOC and TIC was negative and significant in the sites except for the forest floor sediment where there was no significant correlation (Table SI 3). Fe and Mn minerals efficiently scavenge As.8,9 The four sampling sites differed significantly in sediment Fe concentration (p < 0.001). Riverbank (p = 0.001) and forest floor (p < 0.001) sediments showed significant positive correlations with sampling depth for Fe (Figure 2, Table SI 3), but for the other 2 sites no such relationship was evident. Correlation matrix analysis confirmed a significant positive correlation between Fe 8649
dx.doi.org/10.1021/es301328r | Environ. Sci. Technol. 2012, 46, 8645−8652
Environmental Science & Technology
Article
Table 1. Sediment, Porewater, and DGT Characteristics of Koromjal at 0−5 cm Depth (Data are Means and ±SE (n = 3)) sample
river bank
major tributary
forest floor
minor tributary
sediment (μg As/kg) porewater (μg As/L) surface water (μg As/L) CDGT (μg As/L) R (ratio) Kd (kg As/L) porewater DOC (mg/L)
6243 ± 376 23.4 ± 13.6 1.58 ± 0.1 2.7 ± 1.2 0.14 474 19.5 ± 2.6
7422 ± 223 8.6 ± 3.0 1.79 ± 0.1 3.3 ± 0.6 0.42 1115 26.0 ± 1.7
7929 ± 202 6.63 ± 3.4 1.03 ± 0.1 5.4 ± 2.8 1.13 5421 27.7 ± 3.2
9126 ± 385 4.08 ± 1.7 2.6 ± 0.2 4.3 ± 0.2 1.37 2982 28.3 ± 1.0
down the profile in riverbank and forest floor sediment, but not for the major tributary and minor tributary sediments. Fe concentrations were highest for the vegetated forest floor (Figure 2). As for Fe, the Mn concentration in sediments showed significant variation between the four sites (p = 0.001) and down the core profile (p < 0.001). Turf-deposited sediment from the forest floor had almost double the concentration of Fe and Mn compared to deposited riverbank sediments (Table SI 2), again indicating variation in setting and digenesis. Particle size distribution in sediments showed significant positive correlations with total As, TOC, and Fe (Figures SI 6 and 7), which is in agreement with studies on Bengal Holocene aquifer sediments.8 Clay and fine clay particulates (6 μm) had significant negative correlations with total As, TOC, and Fe. Riverbank and major tributary sediments (>70%) were found dominated by coarse particles, whereas finer particles (50%) of forest floor and minor tributary sediments. The banks of the river and major tributary are characterized by the absence of vegetation and by coarse sediments, while the forest floor and the minor tributary have dense vegetation and finer sediments. As a consequence, forest floor and minor tributary sediments had higher As, TOC, and Fe concentrations. The high TOC of vegetated sediments/environments may also be due to higher plant- and animal-derived C-inputs (i.e., leaf deposition and root turnover dominating), as well as having finer sediments that are associated with higher TOC. Moreover, mangrove surface sediments have a high tree root density.7 Kirby et al.6 reported that fine roots in mangrove ecosystems can accrete As to elevated concentrations, ∼10 times higher than bulk sediment concentration. Root iron plaque formation is widely observed for plants growing in anaerobic sediments such as coastal swamps6 and freshwater wetlands.12 Indeed, enrichment of As in iron root plaque layers could explain why As concentrations in the bulk soils of the forest floor were slightly lower than those in the minor tributary despite the high As content of the deposited sediments. Arsenic in Porewaters. Porewater collected from the four different sampling locations at Mongla showed significant variation among the locations (p < 0.001) (Figure 3). The riverbank had the highest porewater As concentration (63.8 ± 17.0 μg/L) at 5−15 cm depth > major tributary (13.9 ± 3.0 μg/L) > forest floor (13.6 ± 4.0 μg/L) > minor tributary (2.0 μg/L ± 0.3). At 0−5 cm depth the porewater As concentrations were highest at river bank (20.6 ± 8.0 μg/L) andlowest at minor tributary (3.0 ± 1.0 μg/L) (Figure 3). Porewater at 5−15 cm depth (24.3 ± 6.7 μg/L), averaged across the four sites, was significantly higher (p = 0.013) than that at 0−5 cm (9.8 ± 2.6 μg/L). The depth by location interaction term was also significant (p = 0.022), indicating that As mobilization is regulated by location and sampling depth.
Sampling for 2 consecutive days did not show any variation in As concentrations in porewater. To evaluate the factors affecting As mobilization in porewater, other geochemical parameters of porewater and their relation with As were also evaluated. Significant variations in DOC concentrations (p < 0.001) at the four sampling sites of Mongla were found with minor tributary (36.2 ± 4.0 mg/L) > forest floor (28.1 ± 2.7 mg/L) > major tributary (19.9 ± 1.7 mg/L) > river bank (19.5 ± 1.9 mg/L) at 5−15 cm depth. The upper layer (0−5 cm) followed the same order minor: tributary (28.3 ± 2.6 mg/L) > forest floor (27.4 ± 3.0 mg/L) > major tributary (24.2 ± 1.7 mg/L) > riverbank (20.7 ± 1.5 mg/L) (Figure 3). Variation for DOC concentration in porewater between sampling day and depth of sampling was not significant. Sampling sites in Mongla were not significantly different in porewater Fe concentrations but a significant difference was observed between the sampling depths (p = 0.035) (Figure 3). Fe concentration in porewater (average across all sites) at 5−15 cm depth (2.2 ± 0.4 mg/L) was more than 2-fold higher than at 0−5 cm depth (1.1 ± 0.2 mg/L). Mn concentrations in porewater showed significant spatial variations (p = 0.005). At 5−15 cm depth the highest Mn concentration in porewater was found at the river bank (3.6 ± 1.5 mg/L) which was significantly higher than for the major tributary (1.7 ± 0.7 mg/L) but not significant to minor tributary (3.0 ± 0.1 mg/L) and forest floor (3.2 ± 1.3 mg/L). At 0−5 cm depth the highest Mn was observed at forest floor (2.3 ± 0.6 mg/L) and the lowest was at the major tributary (1.3 ± 0.2 mg/L). Variation in porewater Mn with depth was significant (p = 0.04), but the sampling day and location by depth interaction was not. Mn concentration in porewater (average across sites) was found higher at 5−15 cm (2.9 ± 0.3 mg/L) compared to that at 0−5 cm sampling depth (2.0 ± 0.2 mg/L). The regression analysis between As and DOC, Fe, and Mn (Figure SI 8) revealed a significant negative association between As and DOC (r = −0.335; p < 0.05). Arsenic in porewater was found strongly correlated with Fe (r = 0.653; p < 0.001) and Mn (r = 0.521; p < 0.001). DOC did not have a significant effect on Fe and Mn mobilization in porewater and the relationship between Fe and Mn was not significant. In contrast, Williams et al.11 reported a strong positive correlation between As and DOC in paddy field porewaters at sites with a similar range of porewater As but higher DOC concentrations; a feature consistent with rice farmers' use of cow dung as the principal fertilizer. Standing surface water analysis had the same pattern for DOC as porewater for the four sites, but differed for As with the highest surface water As found at the minor tributary (Figure SI 9). The correlation between total As and DOC in surface water was significant (r = 0.724, p = 0.01). Mean surface water As concentrations in the four sites were
