Water Resources Research

Hydrological and chemical connectivity dynamics in a groundwater-dependent ecosystem impacted by acid sulfate soils

Authors

  • B. Nath,

    1. School of Environmental Systems Engineering, The University of Western Australia, Crawley, Western Australia, Australia
    2. Now at School of Geosciences, University of Sydney, Sydney, New South Wales, Australia
    Search for more papers by this author
  • A. M. Lillicrap,

    1. School of Environmental Systems Engineering, The University of Western Australia, Crawley, Western Australia, Australia
    2. Centre of Excellence for Ecohydrology, The University of Western Australia, Nedlands, Western Australia, Australia
    3. Department of Agriculture and Food, Albany, Western Australia, Australia
    Search for more papers by this author
  • L. C. Ellis,

    1. School of Environmental Systems Engineering, The University of Western Australia, Crawley, Western Australia, Australia
    Search for more papers by this author
  • D. D. Boland,

    1. School of Environmental Systems Engineering, The University of Western Australia, Crawley, Western Australia, Australia
    2. Now at School of Civil and Environmental Engineering, University of New South Wales, Sydney, New South Wales, Australia
    Search for more papers by this author
  • C. E. Oldham

    Corresponding author
    • School of Environmental Systems Engineering, The University of Western Australia, Crawley, Western Australia, Australia
    Search for more papers by this author

Corresponding author: C. E. Oldham, School of Environmental Systems Engineering, The University of Western Australia, Crawley, Western Australia 6009, Australia. (carolyn.oldham@uwa.edu.au)

Abstract

[1] Groundwater-dependent ecosystems (GDEs) in arid and semiarid environments play significant ecological roles, and, yet in many parts of the world, these ecosystems have been drained for agricultural use. In wetlands containing acid sulfate soils, the altered hydrology may trigger acidification and subsequent trace metal release. Quantifying shifts in hydrological regime and connectivity dynamics across wetlands is critical for understanding the resilience of these GDEs to anthropogenic impacts. Seasonal water balances for a wetland severely impacted by drainage and acidification were combined with laboratory geochemical data and field observations to develop a conceptual model describing hydrological connectivity across the wetland. The data indicated that, with the onset of the dry season, the superficial aquifer was lowered, exposing sulfides that oxidized to form sulfuric acid and dissolving metal salts. The following dry season enhanced capillary action causing upwelling of oxidized products to the surface where evaporative precipitation created acidity scalds. Subsequent winter rainfall and infiltration caused groundwater levels to rise, intersect with the ground surface, and form disconnected acidic pools. As the wet season progressed, connectivity was established between the pools, resulting in metal-rich acid discharge from the wetland. The degree of acid fluxes and metal release was controlled by the physicochemical characteristics of the soils, its exposure to the seasonally variable wetland hydrology, antecedent hydrological conditions, hydrological connectivity (both vertical and horizontal), and the resulting biogeochemical conditions.

1. Introduction

[2] The ecological significance of groundwater-dependent ecosystems (GDEs) has historically been neglected, and these ecosystems continue to be at risk from human-induced changes [Eamus et al., 2006; Mackay, 2006]. Over the past decade, there has been increasing recognition of their ecological significance, particularly in arid and semiarid landscapes [Boulton and Hancock, 2006; Contreras et al., 2011; Murray et al., 2003]. In such landscapes, intermittent hydrological connectivity and disconnectivity between GDEs and other catchment elements (e.g., subcatchments, riparian zones, wetlands, and streams) play an important role in the healthy functioning of the ecosystems. In GDEs, local groundwater-surface water interactions control soil pH, redox conditions, dissolved organic matter availability, and plant growth [Du Laing et al., 2009]. Seasonal and decadal hydrological variability affects the mobilization of trace metals [Hrachowitz et al., 2010]. However, a detailed understanding of hydrological processes and the resulting biogeochemical dynamics remains challenging and unanswered.

[3] The retention, mobility, and transport of material (e.g., chemicals and biota) across the landscape depend on transport pathways and hydraulic residence times. For example, small-scale and large-scale spatial heterogeneity creates multiple transport pathways and a residence time distribution (RTD) [Botter et al., 2011; Hrachowitz et al., 2010]. The RTD has been used as a fundamental characteristic in aquatic systems with free surfaces [Carleton, 2002; Werner and Kadlec, 2000] and in catchments [Bevan, 2001; McDonnell et al., 2010; McGuire and McDonnell, 2006]. These authors note the importance of the residence and transit times for the export of dissolved and particulate constituents or pollutants from lakes and catchments. Given the specific interest in pollutant transformation, we note that a more pertinent parameter is the exposure timescale, inline image, which is the timescale over which the dissolved and particulate constituents or pollutants have the opportunity to be transformed during transport. The concept of an exposure timescale has been previously used in surface renewal theory [Dankwerts, 1951] and chemical engineering [Asarita, 1967].

[4] The exposure timescale, inline image, of a hydrologically connected landscape element (for example, a riparian zone discharging to a river) is conceptually understood as follows:

display math(1)

where V is the wetted volume of the landscape element (m3), and Q is the volume flux (m3 s−1). For wetlands dominated by surface water-groundwater interactions, quantification of exposure timescales is challenging because of temporal and spatial variability in flow pathways.

[5] When the landscape is made up of multiple disconnected elements, e.g., oxbow lakes, isolated surface ponding, or seasonally perched wetlands, the isolated waters provide opportunity for chemical reaction [Cendón et al., 2010; Eamus and Froend, 2006; Sophocleous, 2002]. Once reflooded, the landscape becomes hydrologically connected, and accumulated materials can be transported across the landscape. The exposure timescales of intermittently disconnected landscape elements can no longer be conceptualized as (1), because Q is not defined in a disconnected landscape. However, the isolation timescale, inline image, provides an equivalent period of opportunity for chemical reaction. Physicochemical processes that occur across isolation timescales create antecedent conditions, which subsequently control the export of chemicals from catchments, particularly in arid and semiarid environments where disconnectivity is a key seasonal feature of the landscape [Bertrand et al., 2012; Cendón et al., 2010; Eamus and Froend, 2006]. Thus, an intermittently connected landscape may be characterized by both its isolation timescales and its residence times.

[6] When groundwater-dependent wetlands are impacted by the exposure of acid sulfate soils (ASS), the dynamics of hydrological connectivity and disconnectivity can control acidity and the release of contaminants [Dent, 1986; Sammut et al., 1995, 1996b; White et al., 1997]. The export of acidity products during hydrologically connected periods depends on the opportunity for accumulation and/or consumption of acidity products during the disconnected period. This antecedent state in turn depends on the balance between the isolation and reaction timescales, and, once connected, export of chemicals depends on the balance between transport and reaction timescales.

[7] Climate or human-induced perturbations of the groundwater table may create time lags between vertical and horizontal connectivity of years to decades. For example, prolonged drought may cause the lowering of groundwater tables and result in the isolation and deposition of acidic products in the soil profile. These acidic products could remain in the soil profile for decades until disturbance by land-use change or unusual rainfall/flooding episodes that create sufficient vertical and/or horizontal connectivity to flush accumulated acidity products from the soil matrix.

[8] The processes described above require intermittent hydrological connectivity in both the vertical and horizontal directions [Johnston et al., 2004, 2009; Kinsela and Melville, 2004]. However, these processes have rarely been described in terms of isolation timescales; most ASS and wetland research focuses on periods of hydrological connectivity. In this paper, we use a simple water balance for an acidic wetland and field geochemical data to test this conceptual model of seasonal connectivity dynamics. We also identify isolation timescales to assess the impact of connectivity dynamics on geochemical processes within the wetland.

2. Methods

2.1. Site Description

[9] The Muddy Lakes wetland (approximately 30 ha) is a GDE located on the Swan Coastal Plain, 180 km south of Perth, Western Australia (Figure 1). The area experiences hot, dry summers (16°C–30°C) and cool, wet winters (7°C–17°C). The mean annual rainfall over the last 15 years has been approximately 700 mm, the majority of which occurs between May and September, whereas the mean annual potential evaporation is approximately 1500 mm. The site sits between aeolian sand dune deposits of Pliocene to Holocene age [Hirschberg, 1987]. Most notable of these is Tamala limestone, an aeolian calcarinite that formed elongated dunes parallel to the present coastline [Hirschberg, 1987]. The area was drained prior to the 1930s for agricultural development. The major drain runs north to south, parallel to the coastline on the western fringe of the wetland, and is fed by groundwater and other farm drains. The site is a seasonally inundated depression in the landscape, drying up in midsummer and wetting up in the late autumn when the major drain starts to flow. The major drain continues to flow for approximately 3 km south of the study area before discharging into Geographe Bay, Indian Ocean.

Figure 1.

The study area showing location of the eight boreholes (bores 1–8) and two gauging stations (SWN and SWS). The elevations shown are from the LiDAR DEM. Surface runoff and groundwater discharge to the major drain (flowing north to south) that flows to another wetland whose levels are controlled by floodgates to the Indian Ocean. Elevation units are meters.

[10] Apart from barren acid scalds, the wetland vegetation has been disturbed by introduced kikuyu grass Pennisetum clandestinum and in the margins by the salt and acid tolerant, native couch grasses Paspalum vaginatum and Cynodon dactylon [Semple et al., 2004]. Despite its degraded state, the wetland retains a high ecological value, sustaining a threatened ecological community (Quinalup Dune damplands) [Environmental Protection Authority, Western Australia (EPAWA), 2003] and providing habitat for three Australian nationally listed endangered species: Western Ringtail Possum (Pseudocheirus occidentalis), Quokka (Setonix brachyurus), and Baudin's Black Cockatoo (Calyptorhynchus baudinii) [EPAWA, 2003].

2.2. Installation of Monitoring Boreholes, Gauging Station, and Field Measurements

[11] Eight groundwater-monitoring boreholes were installed across the site using hollow flight augers. The boreholes were cased in 50 mm Class 18 polyvinyl chloride (PVC) pipes and were constructed to different depths depending on soil stability. The lowest 2 m of each borehole was screened with 50 mm Class 18 slotted PVC pipes. The screens were covered with a geotextile filter sock to prevent ingress of fines. The screens were packed with clean sand, and a bentonite seal was placed above the screen. The rest of the borehole was backfilled with cuttings and grouted at the surface. Pronounced stratigraphic layering was observed in the boreholes across the top 50–100 cm depth. However, the borehole slots were positioned within relatively homogenous fine to medium sands. Soil/sediment samples were collected to test for field pH and later chemical analysis in the laboratory. Immediately upon collection, the soil/sediment samples were placed on ice and stored at 4°C until further analysis. Field pH was measured on a sediment-distilled water slurry (pHF) and after reaction with peroxide (pHFOX) [American Public Health Association (APHA), 1992].

[12] Two gauging stations were also installed at the study site at the northern (SWN) and southern (SWS) ends of the major drain extending through the site (Figure 1). Water level and velocity (ISCO 750 AV Flow Module, Teledyne ISCO Inc., Lincoln, USA), pH, electrical conductivity (EC), and temperature (YSI 6583 probes, YSI Inc., Yellow Springs, USA) were measured every 15 min. Flow rates were estimated from the water level and velocity measurements. Each gauging station was also equipped with a Teledyne ISCO 6712 automatic water sampler that collected 200 mL of drain water every 6 h; four samples were combined into a daily sample. Bottles were acid-washed and rinsed with deionized water three times before being placed into the autosamplers. Manual measurements of pH and EC of the water samples collected daily from SWS confirmed the continued performance of the YSI probe.

[13] Hydraulic conductivity of the soil formation was determined using the Hvorslev slug test [Fetter, 1994]. Approximately 1 L of water was poured down the boreholes to provide an instantaneous head (h0) above the normal water level (h); the decline in h0 with time was tracked using a capacitance probe (Scott Parsons Electronics, Albany, Western Australia). The hydraulic conductivity was calculated as follows:

display math(2)

where Ks is the hydraulic conductivity (m d−1), r is the radius of the borehole (m), L is the length of the borehole screen (m), and T0 is the time taken for h0 to drop to 37% of the original induced head (days). The value of T0 was determined by plotting the relative head change (h0/h) on a log scale against the elapsed time.

2.3. Soil Sample Preparation and Laboratory Analysis

[14] Soil samples were air-dried at 25°C for 48 h and crushed gently to pass through a 2 mm sieve. To assess the presence of ASS in the soil profile, the suspension peroxide oxidation combined acidity and sulfate (SPOCAS) suite of analyses were conducted [Ahern et al., 2004]. A subsample was extracted with KCl solution for the determination of soluble and absorbed sulfur (nonsulfidic, SKCl, %), measured using inductively coupled plasma atomic emission spectrometry (ICP-AES). The pH (pHKCl) and titratable actual acidity (TAA) of the extracted solution were also measured. Another subsample was oxidized with H2O2 and analyzed for sulfur using ICP-AES, which represented total soluble, absorbed, and sulfidic species (SP, %). The titratable potential acidity (TPA) and pH (pHOX) were also measured. When pHOX of the solution was greater than 6.5, the samples were analyzed for ‘excess’ acid neutralizing capacity (ANCE) through titration with HCl (to pH 4). The peroxide oxidizable sulfur (SPOS, %) was calculated as the difference between SP (%) and SKCl (%), giving a measure of the amount of sulfides in the sample. Soluble ions, metals, acidity, and alkalinity were also measured in surface soil slurries (one part soil:five parts water) [Rayment and Higginson, 1992].

[15] Mineralogical studies using X-ray diffraction (XRD) were carried out using Cu-Kα radiation, operating at 40 kV and 30 mA, and a post-diffraction graphite monochromator. The crystalline mineral phases were quantified using SIROQUANTTM software (Sietronics Pty. Ltd., Canberra, Australia). Total carbon (TC) and total sulfur (TS) were analyzed using an Elementar vario MAX CNS analyzer (Elementar, Hanau, Germany). Total organic carbon (TOC) was analyzed by wet oxidation [Heanes, 1984]. The major elemental analyses involved preparation of a homogeneous glass bead of 40 mm in diameter and analyzed in a Bruker-AXS S4 Pioneer (Bruker AXS GmbH, Karlsruhe, Germany) X-ray fluorescence (XRF) spectrometer against a reference standard (Andesite, AGV-2, silicates general). The data were processed through Spectra PLUS software (Pioneer Hill Software LLC, USA). The trace element analyses were carried out on a pressed pellet prepared from the bulk powdered soil samples and compared against a reference standard (Basalt, BHVO-2).

2.4. Water Sampling and Laboratory Analysis

[16] Groundwater samples were collected 12 times (on a monthly basis) between 11 September 2008 and 11 September 2009. Groundwater level (relative to local datum) was measured prior to sample collection. Three borehole casing volumes of water were removed prior to sample collection. The pH, EC, temperature, and Eh were measured on-site using a TPS multimeter (TPS 90FL-T, TPS Australia). The samples were then filtered through mixed cellulose (0.2 µm) filter papers. The filtered samples were divided into two; half of each sample was acidified with a few drops of concentrated HNO3 for subsequent analysis of major cations and trace elements, whereas nonacidified samples were analyzed for acidity, alkalinity, inline image, and Cl. All samples were stored on ice, transferred to the laboratory, and then stored in the dark at 4°C until analysis. Water samples were analyzed in the laboratory for acidity (by titration with 0.01 M NaOH to pH 8.3, APHA 2310 B [APHA, 1992], including the peroxide oxidation step), alkalinity (by acid titration using 0.05 M HCl to pH 4.5, APHA 2320 B [APHA, 1992]), inline image and Cl using ion chromatography, major cations using ICP-AES, and trace element concentrations using inductively coupled plasma mass spectroscopy.

2.5. Water Balance Calculations

[17] A wetland water balance was attempted to (a) estimate the fraction of the total wetland water volume that was above versus below the ground surface and (b) to estimate net groundwater flows into and out of the wetland. The water balance also indicated how both of these dynamics changed with season. The undulating terrain at the site intersected with a rising and falling groundwater table, creating a series of ephemeral pools about 0.5 m in depth and a few meters in width that were intermittently connected over winter, both via surface flows and the high groundwater table. This hydrological dynamic is similar to that observed in peat wetlands, with their hummocks and hollows [Frei et al., 2012], and is significantly different from the standing waters typically observed in many wetlands and on which most wetland water balances are based [see, for example, Jolly et al., 2008; Krasnostein and Oldham, 2004]. This hydrological dynamic creates challenges for our water balance and quantification of connectivity in the wetland.

[18] Initially, the water balance was classically conceptualized as follows:

display math(3)

where ΔV is the change in storage of the wetland with time, P is the direct precipitation experienced by the wetland, GWIN is the groundwater inflow to the wetland, R is the surface runoff into the wetland, ET is the evapotranspiration, GWOUT is the groundwater outflow, and QOUT is the wetland outflow. All units are m3 d−1.

[19] Meteorological data were sourced from the SILO Data Drill service provided by the Queensland Department of Natural Resources [QNRM, 2009]. Daily rainfall data were used to estimate the total volume of precipitation (P) received by the wetland. Extensive drainage of adjacent agricultural lands ensured negligible surface runoff into the wetland; no surface flow pathways to the wetland were observed over 4 years of monitoring. Evapotranspiration was estimated using the Morton [1983] areal evapotranspiration model that was multiplied by the wetland surface area to estimate the ET volume on a daily time step.

[20] A digital elevation model (DEM) was developed using Light Detection and Ranging (LiDAR), with a horizontal resolution of 1 m × 1 m and a vertical accuracy of 0.15 m Australian height datum (AHD). The horizontal positional accuracy was 0.6 m. The DEM was used to obtain AHD elevations of the boreholes, and then groundwater and drain surface water height data were converted to AHD. The monthly water elevation data were converted to a raster grid (water elevation grid) via kriging interpolation using a variable search radius and a spherical semivariogram model. For increased accuracy, no limit was set on the maximum data points, and z tolerance was set to 0.01. The raster calculator in the ArcGIS Spatial Analyst Tools (ESRI Inc., CA, USA) was used to subtract the DEM grid from the water elevation grid; all values greater than zero represented expression of surface water. The ArcGIS 3D Analyst Tool then allowed calculation of surface water volumes. The areas of surface water were converted to polygon shape files, and the number and area of surface water pools analyzed statistically for each month. The total wetted volume was calculated from the water elevation grid using 0 m AHD as the lower boundary condition. Changes in surface water storage with time, ΔVs, were calculated as the difference between surface water volumes at consecutive sampling periods divided by the number of days between sampling periods.

[21] To estimate the volumes of subsurface water the surface water volume was subtracted from the total wetted volume; the remainder was multiplied by the specific yield. A constant specific yield (in both time and space) of 0.2 was used for this calculation [Davidson and Yu, 2008]. The lithostratigraphy was dominated by fine to medium sands; measured local hydraulic conductivities (KSAT) ranged from 0.3 to 6 m d−1. The classification developed by Rawls et al. [1982] indicates for these soil types; we expect hydraulic conductivity from 1 to 6 m d−1 and specific yield in the range of 0.3–0.5. CSIRO [2010] calibrated their South West Aquifer Modeling System (SWAMS, CSIRO, Australia) using specific yield (among other parameters) and validated results against local piezometer data. For the superficial sandy aquifer of the region, specific yield values of 0.2–0.5 produced reliable modeling outcome. Specific yield varies as a function of depth to water table [Childs, 1960] and, at our site, may therefore vary with time in response to the rising and falling water table. Johnston et al. [2009] showed that saturated hydraulic conductivity, Ksat, varied strongly in the sulfuric horizons found in ASS wetlands; we expect specific yield to show similar heterogeneity. However, in the absences of high-resolution conductivity or yield data at our site, we assumed a conservative value at the lower end of the SWAMS range and will discuss the implications of our choice below. Finally, the changes in subsurface storage with time, ΔVss, were calculated as the difference between groundwater volumes at consecutive sampling periods divided by the number of days between sampling periods.

[22] The wetland outflow, QOUT, was defined as the volumetric discharge through the major drain at SWS. The volumes were calculated by multiplying the flow velocity measured at SWS gauging station by the wetted cross-sectional area. The estimates of groundwater flow into, and out of, the wetland were complicated by possible contributions from both regional and local groundwater flows, so we combined these terms into an unknown net groundwater inflow:

display math(4)

[23] The resulting water balance for the wetland is given as follows:

display math(5)

2.6. Flux Estimates

[24] Estimates of surface fluxes out of the wetland via the major drain did not rely on the water balance but could be calculated directly from outflow measurements at SWS. The fluxes of Al, inline image, and titratable acidity (mol of H+) from the wetland through the drain were estimated during the period from 11 September 2008 to 11 September 2009. The dissolved metals and acidity products were likely to be generated within the wetland during surface rainwater flushing and/or subsurface sediment-water interactions [Johnston et al., 2004]. The fluxes were estimated as follows:

display math(6)

where FD is the flux estimated on a daily basis, QOUT is the volumetric daily discharge through the major drain at SWS, and C is the concentration of relevant chemical parameters at SWS.

[25] Total salt fluxes, inline image, were determined from EC values (after converting to TDS) measured daily at SWS. Monthly chemical concentrations at SWS were transformed into daily data, using linear interpolation or linear averaging, as appropriate. The calculated fluxes of major ions were summed to provide an additional estimate of total salt fluxes, inline image. The two estimates of total salt fluxes agreed well (Figure 2) and gave confidence to our interpolation from monthly to daily chemical concentrations and, therefore, our estimates of chemical fluxes from the study site.

Figure 2.

Comparison of salt fluxes estimated from (a) measured EC values (after converting to TDS) and (b) major ion concentrations. A strong positive correlation was observed between the two estimates, and the seasonal dynamic of measured and calculated salt fluxes agreed well.

3. Results

3.1. Groundwater Levels and Wetland Water Balance

[26] The groundwater levels in the boreholes exhibited strong seasonality, decreasing during summer and increasing during winter (Figure 3). The difference between summer and winter water levels ranged from 0.82 to 1.1 m. In autumn, the local groundwater gradient is directly toward the sea, i.e., toward the west (Figure 4). This is the direction of regional flow [Commander, 1984]; however, the local groundwater gradient shifts toward the southwest over winter and spring. There is an ephemeral lake to the north that creates a localized groundwater mound and is the likely driver of the southwest groundwater gradient (Figure 4). Note that the ephemeral lake is disconnected from the study site with respect to surface water.

Figure 3.

Seasonal trends in water levels across east-west borehole transects. (a) The northern transect (bores 8 and 7) was dominated by surface water over winter with groundwater being recharged. (b) In the middle transect (bores 1–3), the groundwater gradients were essentially flat over winter/spring. (c) In the southern transect (bores 4–6), there was a groundwater gradient toward the drain.

Figure 4.

Groundwater levels (m) in boreholes for March 2009 (black-dotted contours) and September 2009 (dark-dotted gray contours). The extent of surface water for September 2009, as calculated from the analysis of LiDAR DEM and groundwater height measurements, is shown in light gray shades. The location of boreholes is shown as crosses, and the two gauging stations are shown as stars. The groundwater levels have a westerly gradient in autumn (March 2009) and a southwesterly gradient during late spring.

[27] The variability in water levels and changing hydrological dynamics created seasonally and spatially variable ponding across the wetland (Figure 5). The horizontal connectivity of the ponded areas, and the vertical connectivity between ponded areas and groundwater, determined the discharge to drains.

Figure 5.

Seasonal ponding and connectivity of surface water at the study site. Images are derived from the analysis of LiDAR DEM and groundwater measurements, from September 2008 to October 2009. Red dots indicate boreholes, and green dots indicate drain gauging stations.

[28] During the winter months, the higher groundwater levels in the northern end of the wetland (bores 7 and 8, Figure 3a) and the limited horizontal connectivity of surface pools (Figure 5) suggest that recharge to groundwater occurred to the southwest. Such “flow-through” wetland dynamics are typical for GDEs on the sandy Swan Coastal Plain [Jolly et al., 2008]. Through the middle section of the wetland (bores 1–3), the groundwater gradient was essentially flat, with little or no groundwater discharge to the major drain (Figure 3b). Across the southern section, ponded areas with limited horizontal connectivity at the surface (Figure 5) provided “windows” to a groundwater gradient (Figure 3c) suggesting that recharge of groundwater occurred to the southwest.

[29] Precipitation occurred during winter to early spring; summer and autumn were generally dry except for intermittent storms (Figure 6a). Evapotranspiration was greatest during early summer when ponded surface water was present; it declined from the beginning of autumn until the end of winter (Figure 6a). There was no measured wetland outflow over summer, and it then peaked after winter rains (Figure 6a), with a maximum of 3315 m3 of water (runoff depth ≈ 10 mm) on 19 July 2009.

Figure 6.

Seasonal variation in the water balance terms. (a) Measured daily drain outflow, precipitation and evapotranspiration, and calculated wetland water volume using specific yield = 0.2. All units are in m3. (b) Changes in the volume of measured drain outflow, precipitation, evapotranspiration, and calculated net groundwater outflow (indicated for Sy = 0.2 and 0.5). All units are in m3 d−1.

[30] The wetland storage (ΔV) showed large seasonal fluctuations associated with periods of high precipitation or high evapotranspiration (Figure 6a); however, at all times, it was dominated by superficial groundwater (97%–99%). Note the 3 to 4 month lag between (a) the time of maximum precipitation and maximum wetland storage and (b) the time of maximum evapotranspiration and minimum wetland storage.

[31] Net groundwater inflow to the wetland (QIN) occurred from throughout the winter months until early summer (August-January; Figure 6b). Negative groundwater inflow (i.e., groundwater outflow) occurred during February (Figure 6b) in response to the cessation of drain outflow during the period of maximum wetland volume (cf., Figure 6a). Groundwater inflow peaked again in June with the onset of winter rains but before the drain started to flow. We note that the estimated superficial groundwater volume and, therefore, the net groundwater inflow are dependent on our assumed value of specific yield (0.2), which is at the lower end of the range typical for sands. If we assume Sy = 0.5, the dynamics of the net groundwater inflow remain similar, though the magnitude of peaks is adjusted (Figure 6b).

3.2. Sediment Geochemical Characteristics and Hydrogeochemical Patterns

[32] The field measurement of the soil pH (pHF and pHOX) provides a preliminary indication of the presence of potential and actual acid sulfate soils, which was later confirmed by SPOCAS analysis of selected samples [Ahern et al., 2004]. The SPOCAS results showed the presence of measurable quantity of TPA (bores 1, 5, and 7) and TAA (bore 7; Table 1). The soils in bore 1 at 24 to 92 cm depth contained high sulfide concentrations (12% of TS) including the presence of pyrite, as determined by XRD analysis (Table 2). XRF analysis indicated the soil samples contained high levels of iron, carbon, and sulfur as well as measurable elevated As concentrations (Table 2). The XRD analysis indicated the dominant minerals were quartz, feldspar, and calcite. The chemical analysis of extracts from surface precipitate slurries (one part soil:five parts water) indicated high concentrations of Al (up to 11 mg L−1), titratable acidity (up to 592 mg L−1), Fe (up to 251 mg L−1), and inline image (up to 1230 mg L−1; Table 3).

Table 1. SPOCAS Results From Selected Soil Samples Collected at Different Depths of the Boreholesa
SitesDepth (cm)pHFpHOXpHKClSP (%)SKCl (%)SPOS (%)ANCE (mol H+/t)TPA (mol H+/t)TAA (mol H+/t)
  1. a

    pHF, pH measured in the field after suspended the soils in distilled water; pHOX, pH of the soils after reaction with H2O2; pHKCl, pH of the soils after reaction with KCl; SP, total soluble and absorbed sulfide species; SKCl, total nonsulfidic sulfur; SPOS, peroxide oxidizable sulfur (difference between SP and SKCl); ANCE, “excess” acid neutralizing capacity; TPA, titratable potential acidity; TAA, titratable actual acidity.

Bore 13006.82.06.72.360.112.25<21650<2
Bore 23257.87.69.10.080.010.074480<2<2
Bore 31007.87.79.30.01<0.010.01220<2<2
Bore 44008.08.49.50.440.050.39860<2<2
6258.68.29.60.040.010.035100<2<2
Bore 51253.77.29.00.040.030.0183<2<2
2006.12.87.70.250.040.21<297<2
Bore 6258.38.89.50.060.040.022520<2<2
1255.08.09.80.020.010.01580<2<2
2006.27.19.50.070.020.0561<2<2
Bore 7503.06.46.00.050.05<0.015<28
1504.02.44.80.520.030.49<231052
Bore 85007.28.39.70.350.040.31956<2<2
Table 2. Elemental Composition of Selected Soils Collected at Different Locationsa
SitesDepth (cm)Soil DescriptionsSiO2 (%)TiO2 (%)Al2O3 (%)Fe2O3 (%)MnO (%)MgO (%)CaO (%)Na2O (%)K2O (%)P2O5 (%)LOI (%)As (mg kg−1)TOC (%)TC (%)TS (%)Minor Minerals
  1. a

    XRD data indicated that all sites were dominated by quartz, potassium feldspar, calcite, and minor minerals as shown in the final column. bdl, below detection limit; LOI, loss on ignition; TOC, total organic carbon; TC, total carbon; TS, total sulfur. N/A, not available.

Bore 1SurfaceBlack, organic-rich loam321.20.872.60.051.7230.260.360.2237bdl11160.05N/A
 8–14Brown, fine sand380.820.941.60.032.0250.240.440.1630bdl3.99.50.02N/A
 14–24Dark gray organic-rich loam130.310.325.30.020.212.10.140.110.33784.026330.22N/A
 24–92Black organic-rich loam3.00.490.29170.050.381.20.040.090.06766.0212412Pyrite
Bore 2SurfaceBlack, organic-rich sandy loam441.71.21.60.061.3220.220.510.22252.02.67.30.01N/A
Bore 3SurfaceOrganic-rich black sand830.542.60.710.020.110.440.381.70.069.78.04.44.10.01N/A
Bore 4SurfaceBlack, organic-rich sandy loam543.71.73.30.130.91150.250.690.24204.04.17.30.003N/A
Bore 5SurfaceBlack, organic-rich sandy loam540.361.71.10.081.16.70.261.10.42322.012150.07N/A
Bore 6SurfacePeaty, black top soil660.321.80.610.040.33130.301.10.15162.02.55.30.001N/A
Bore 7SurfaceDark, organic-rich sandy loam640.623.85.30.030.221.00.311.10.16221510100.06Goethite
 12–25Dark coarse sand, little clay880.703.01.90.020.120.170.391.60.023.16.00.680.68bdlN/A
 25–90Gray coarse sand, Fe mottling920.612.80.820.010.120.110.411.80.010.93160.260.250.01N/A
Bore 8SurfaceDark black top soil571.41.71.40.050.99170.260.830.16192.02.26.3bdlN/A
SWNSurfaceBrownish yellow Fe crusts100.640.56100.040.180.520.130.160.77759.015158.3Pyrite
 10–100Organic-rich, peaty material284.61.3210.180.300.720.180.591.4405.033381.9Pyrite
SWSSurfaceDark brown, Fe crusts291.60.94380.260.140.790.230.493.723bdl6.97.80.29N/A
Table 3. Geochemical Data of Chemical Extracts From Surface Soil Samples Collected at Three Locations in June 2009a
SitespHFAcidity (mg L−1)Alkalinity (mg L−1) inline image (mg L−1)Cl (mg L−1)Ca2+ (mg L−1)Mg2+ (mg L−1)Na+ (mg L−1)K+ (mg L−1)Al (mg L−1)As (µg L−1)Fe (mg L−1)Mn (µg L−1)Cd (µg L−1)Pb (µg L−1)Zn (µg L−1)Ni (µg L−1)
  1. a

    N/A, not available.

SWN2.6592<1123011912316525.9≤0.55.562.34480.060.18685.94
Bore 16.6<124615870140517526118.7225169800.393.5439836
SWS4.0N/A23835582120360.47≤0.52.674675100.150.23867.7

[33] The groundwater showed two distinct geochemical characteristics: bores 3, 5, and 7 were enriched with sulfates, whereas bores 1, 2, and 6 were more enriched with bicarbonates (Figure 7). The pH, Eh, and depth to the groundwater showed strong seasonality, with alternating periods of oxic and anoxic conditions (Figures 8 and 9). During late winter and early spring, waterlogging associated with a rising water table created reducing conditions in the organic-rich sediments, subsequent reduction of Fe-oxide and Mn-oxide, and an increase in As, Fe, and Mn concentrations in soil water (Table 4, see bore 3). During this time, we observed the formation of iron monosulfides in the drain water close to the SWS monitoring station. The formation of iron monosulfides is common in ASS environments and typically accumulates at the bottom of drains [Smith and Melville, 2004]. During summer, high rates of evaporation and evapotranspiration caused a decline in the water table, facilitating reoxidation of the surface sediments. When the water table intersected the borehole slots (for example, at bore 3 during drier periods of the year), acidic conditions were detected. In contrast, the groundwater at bore 6 was near pH-neutral and anoxic during much of the study period (Figure 9). At bore 6, the water table remained above the slots throughout the year and thus exhibited low concentrations of acidity, Al, Fe, and inline image, and high concentrations of As and Mn. The redox conditions at this bore were independent of water level.

Figure 7.

Piper diagram illustrating the averaged hydrochemistry of the surface water- and groundwater-monitoring sites.

Figure 8.

Seasonal variations of (a) water level (m below surface), (b) Eh (mV), (c) As (µg L−1), (d) pH, (e) Al (mg L−1), (f) Fe (mg L−1), (g) acidity (mg L−1), (h) inline image (mg L−1), and (i) Mn (µg L−1) concentrations observed in the groundwater bore 3. When the water table intersected the borehole slotting zones (0.75–2.75 m), fluctuating groundwater levels were observed with variable Eh-pH conditions.

Figure 9.

Seasonal variations of (a) water level (m below surface), (b) Eh (mV), (c) As (µg L−1), (d) pH, (e) Al (mg L−1), (f) Fe (mg L−1), (g) acidity (mg L−1), (h) inline image (mg L−1), and (i) Mn (µg L−1) concentrations observed in the groundwater bore 6. The water table remained above the borehole slotting zones (2.0–4.0 m) and stable Eh-pH conditions were observed.

Table 4. Geochemistry of Water Samples Collected From Boreholes (3 and 6) and Gauging Stations (SWN and SWS)a
SitesSampling DateWater Level (m Below Surface)pHEh (mV)EC (µS cm−1)Acidity (mg L−1)Alkalinity (mg L−1)Ca2+ (mg L−1)Mg2+ (mg L−1)Na+ (mg L−1)K+ (mg/L) inline image (mg L−1)Cl (mg L−1) inline image (mg L−1)Al (mg L−1)As (mg L−1)Fe (mg L−1)Mn (mg L−1)Zn (mg L−1)
  1. a

    N/A, not available; bdl, below detection limit.

Bore 311 September 20081.225.85140176290<118413488.4<1728460.830.0332500.210.06
Depth: 2.75 m below surface1 October 20081.166.5914311201612517315416.0153484340.410.009430.160.05
Slotting depth: 0.75–2.75 m24 October 20081.266.5894137055018115357.761596400.520.0211200.170.06
13 November 20081.306.5−73151992519916351231656630.670.0181200.190.05
4 December 20081.366.31−89142555019015291161655560.690.0241000.190.05
22 December 20081.426.26−102143282518613321431625850.960.0371100.140.05
14 January 20091.565.94−6515476<1250183517<1678481.500.0411400.160.08
28 February 20091.764.87163118982<11049.13614<1665176.300.0461400.080.07
27 March 20091.824.43181121050<1788.93511<171447120.0471400.070.13
23 May 20091.623.93193152665<1733.82913<167470330.0411830.070.15
11 July 20090.984.819772688<1645.3393.5<1442609.090.003310.030.07
11 September 20091.066.5651603N/A120733.9294.81424471110.0139.510.050.07
Bore 611 September 20080.676.56477515215889.74227262581550.820.066100.100.02
Depth: 4.0 m below surface1 October 20080.656.431007471615087114320183591694.800.027210.110.37
Slotting depth: 2.0–4.0 m24 October 20080.786.76256835250212124121305594420.990.090170.090.06
13 November 20080.886.57−75811515079124417183571910.340.030210.080.05
4 December 20080.966.53−607355150689.94716183561610.080.036150.080.02
22 December 20081.066.61−777375125758.84419153561550.190.03360.070.02
14 January 20091.26.57−959113310090154520122582530.100.031260.140.02
28 February 20091.366.65−566845125689.84320153551770.060.052170.070.05
27 March 20091.466.79−96815402301218.94520281541760.100.0345.90.050.07
23 May 20091.266.47−9409N/A74383.027137436620.580.0314.690.050.06
11 July 20090.596.64−326371112735.34722112571150.040.044120.090.07
11 September 20090.496.8810744N/A134866.84723134601563.160.129320.180.10
SWN gauging station11 September 2008SW2.9N/A17115<11032410123<11435000.280.001121.000.01
Surface water (SW)1 October 2008SW3.0N/A14815<183207923<11194100.090.00140.770.01
13 November 2008SW3.1N/A12211<157166521<1893090.070.0015.60.490.01
22 December 2008SW3.7N/A76.29<141125515<1612080.010.0010.780.17<0.01
14 January 2009SW3.0N/A2018<1179278411<1806970.020.0017.20.850.01
15 June 2009SW2.863693890902<1279359280<11541430120.0012443.280.09
11 July 2009SW3.054803640666<12964518033<128013904.370.0243115.100.09
11 September 2009SW3.632563190749<13044613734<126014400.070.0174534.140.06
SWS gauging station11 September 2008SW2.6N/A26573<11353813719<12007791.900.001852.100.04
Surface water (SW)1 October 2008SW2.7N/A23830<11143211619<11816391.200.001691.800.02
13 November 2008SW2.9N/A1629<159197819<11083730.320.001381.000.01
22 December 2008SW2.8N/A19124<152219722<11513850.180.001550.770.01
14 January 2009SW2.5N/A32397<1923213740<12136400.880.0011001.500.04
15 June 2009SW2.935013560784<12043511242<119010906.630.0022353.170.05
11 July 2009SW3.054253850206<12304520339<132512205.750.0012034.210.04
11 September 2009SW3.014612800514<11742716126<11899802.340.0011702.670.03

4. Discussion

4.1. Hydrological Connectivity Dynamics

[34] The conceptual model of seasonal hydrological connectivity across the wetland and the subsequent release and transport of acidity products offsite are shown in Figure 10. Evapotranspiration during summer drives capillary action, which transports groundwater or soil water upward until it reaches the evaporation front [Rose et al., 2005]. The superficial soil water becomes hydrologically disconnected or isolated from the underlying groundwater (i.e., vertical disconnectivity). This process leads to the precipitation and accumulation of dissolved acidity products in superficial soil horizons [Minh et al., 1997]. The accumulation of surficial salts, creating bare surface scalds, has been documented for acid-mine-drainage-affected sites [Hammarstrom et al., 2005] and ASS sites [Rosicky et al., 2004]. In wetland dominated by sandy soils, initial winter rains infiltrate and drive a rise in groundwater levels that dissolve acid salts previously precipitated in the upper soil profile.

Figure 10.

Conceptual model of acidification process at the study site.

[35] When the groundwater levels intersect the ground surface levels, ponding of Fe(II)-rich water occurs at the surface, and Fe(II) oxidation and acidification commence. Although these surface ponds may be vertically connected to the groundwater, they may still be horizontally disconnected from surface flows to the drain, and thus acidity would not be discharged from the wetland (Figures 3 and 5). Once the ponded water is hydraulically connected to the drain, a first flush mechanism is initiated. Surface waters are continuously connected horizontally and, once the water table rises and intersects the drain, provide ongoing discharge to coastal waters. We note that this conceptual model does not include a first flush event in which the first major rainfall after a prolonged dry period washes surface acid and salt deposits into adjacent waters [Kinsela and Melville, 2004; Lin and Melville, 1993]. In this wetland, the rising groundwater table provides the connectivity that flushes acid out of the system.

[36] Over spring-early summer, outflow from the wetland ceases and the surface water retreats back to a series of disconnected pools before completely drying out. These ephemeral standing waters display strong patchiness in both space and time. The patchiness of the surface pools provides an indication of the “exposure” length and timescales. However, the time lag between the onset of rains and the first flush event would be strongly dependent on antecedent conditions, e.g., the length and intensity of the preceding dry period [Wilson et al., 1999]. The longevity of the pools provides opportunity for Fe oxidation and subsequent production of acidity products. The patchiness of the pools was quantified by analysis of the total surface water volume (provided by the water balance) and the surface elevations (provided by the DEM). Comparison of ponding variability over seasons (Figure 11) highlights the lag between cessation of rainfall (in October) and the minimum number and surface area of ponds (in May). It also illustrates the variable connectivity between ponds and between the ponding surface water and the drain. These ephemeral pools, containing high concentrations of dissolved acidity products, discharge only once they are connected to the drain.

Figure 11.

Size distribution of the ponds over different seasons; the date of each snapshot is given in Table 5.

[37] The oxidation of Fe(II) can be modeled using the rate law given by the following equation [Stumm and Lee, 1961]:

display math(7)

where inline image is the partial pressure of oxygen (0.21 atm), and inline image is the rate constant (1.5 × 1013 L2 mol−2 atm−1 min−1 at 25°C). This rate law predicts extremely low rates of oxidation under acidic conditions.

[38] The rate constant k in equation (7) can be converted into a pseudo-first-order rate constant for ambient temperature and pressure conditions and for a predefined pH as follows:

display math(8)

yielding k′ = 3 × 10−2 min−1 and 3 × 10−8 min−1, at pH 7 and 4, respectively; these values will be used for the analyses below.

[39] The size distribution of the ponds and how this varies over seasons (Figure 11 and Table 5) allow us to quantify the range of pool sizes and, therefore, the distribution of exposure length and timescales. The characteristic length and timescales are physically constrained, and the maximum duration of ponding at this site is approximately 6 months, τE ≈ 180 days (approximately 1.5 × 107 s). Then using

display math(9)

we estimate the extent of Fe(II) oxidation over the 6 months of surface ponding to be 100% in the circum-neutral pools and approximately 1% in the pools with pH 4. These simple analyses, using the water balance and DEM to provide estimates of inline image, together with estimated rate constants, provide some insight into causes of geochemical heterogeneity observed within both the soil profiles and the surrounding soil water. These predictions require verification by high-resolution field measurements and also via the virtual experiments explored by Frei et al. [2012].

Table 5. Changes in the Number, Maximum Surface Area, and Total Surface Area of Pools, Calculated From the Analysis of the LiDAR DEM and Groundwater Height Measurements From July 2008 to October 2009, Including the Dates Shown in Figure 5
EventDateNumber of PoolsSurface Area of Largest Pool (m2)Total Surface Area of Pools (m2)
14 July 2008155920043,900
215 August 2008153930043,700
311 September 2008153980043,800
41 October 2008165990049,300
524 October 200816110,70037,000
613 November 200815410,10029,200
74 December 2008133970023,300
822 December 200887370014,000
916 January 20094320005800
1023 May 2009356003100
1115 June 200966360012,100
1211 July 200918529,10063,000
131 August 200918730,00066,000
1422 August 20092741,60083,700
1511 September 200919137,40076,200
161 October 200918412,90059,000

4.2. Links Between Seasonal Hydrological Connectivity and Transport of Acidic Salts

[40] Two dominant geochemical conditions, i.e., acid forming and carbonate buffering, occur at the site and interact with seasonal hydrology to add complexity to the geochemical signatures. Where groundwater acidification was greater than the buffering capacity of the formation, the groundwater became acidic and vice versa. The concentrations of Al, As, Fe, and inline image showed high-temporal variability consistent with the observed changes in pH and Eh conditions. The strong positive correlation (r2 = 0.77) observed between Fe and inline image concentrations in groundwater and surface waters (Figure 12) suggests the oxidative breakdown of sulfidic minerals (e.g., iron monosulfides), releasing Fe, inline image, and H+ ions to the soil water (Table 4). The subsequent precipitation of these dissolved minerals on the wetland surface over summer resulted in acid scalds (Figure 13). Iron coatings were observed on plant roots during summer aerobic conditions; this has been previously suggested as a significant sink for As [Keon and Hemond, 2002]. Waterlogging over the subsequent winter may create reducing conditions and slowly release this As to the soil water (Figures 8 and 9).

Figure 12.

The correlation between Fe and inline image concentrations in boreholes 3 and 6 and gauging stations (SWN and SWS).

Figure 13.

Example of surface acid scald at the study site.

[41] The fluxes of Al, inline image, and titratable acidity (i.e., mol of H+) from the wetland showed a very similar dynamic to the measured/estimated salt fluxes (Figure 2a). The fluxes were very low or insignificant from October 2008 to June 2009, during low discharge of drain water, and peaked in December 2008 and then from June to September 2009 coinciding with rainfall events. The observed peaks indicated early hydrological controls on mass fluxes out of the wetland. The decline in H+ fluxes over late winter suggests a successive depletion of surface species and dilution of discharged solutes following extended periods of rainfall. With the onset of the dry summer, increasing evapotranspiration triggered a rapid decline in the volumetric discharge concomitant with the loss of water from the ponding surface. Evaporative precipitation deposited orange and yellow acidity products (up to 38% Fe2O3 and 8.3% TS, Table 2) along the drain bed and in the wetland depressions. These precipitates, in turn, became one of the sources of acidity during the subsequent wet season.

4.3. Environmental Impacts of Acidic Metal-Rich Discharges

[42] The discharge of acidic metal-rich waters from ASS-affected catchments has contributed to ecological damage [Green et al., 2006; Rosicky et al., 2004; Sammut et al., 1995] by directly inhibiting plant growth and affecting the availability of plant nutrients [Rosicky et al., 2006]. The formation of acidic bare surface scalds may kill off or exclude vegetation [Rosicky et al., 2004].

[43] Acid discharge from the major drain was calculated to determine the potential danger to aquatic life downstream [Sammut et al., 1996a]. The measured pH value of the discharged waters at SWS during the study period was used to conservatively estimate the production of sulfuric acid from the wetland study area. This conservative estimate showed that the major drain discharged approximately 720 kg ha−1 yr−1 of sulfuric acid, leading to a release of 23.5 t from 11 September 2008 to 11 September 2009. This discharge rate is significantly higher than previously published estimates of coastal ASS floodplain discharge, 100–200 kg H2SO4 ha−1 yr−1 for the Tweed River floodplain and 200–300 kg H2SO4 ha−1 yr−1 for the Richmond River [White et al., 1997].

[44] Freshwater quality guidelines were used to assess the potential impact of the acid metal-rich discharge on downstream ecosystems. Dissolved Al is highly toxic to aquatic and terrestrial organisms; for example, high Al concentrations can cause damage to fish gills [Dussault et al., 2003]. The trigger value for Al in freshwater environments with pH <6.5 is 0.8 µg L−1 [Australian and New Zealand Environment Conservation Council and Agriculture and Resource Management Council of Australia and New Zealand (ANZECC and ARMCANZ), 2000]. The Al concentrations in the wetland discharge waters were mostly >1.2 mg L−1 (mean: 2.4 mg L−1). The study site is surrounded by agricultural and urban development and provides an important refuge to endangered and ecologically important species that may be impacted by the acid metal-rich waters of the wetland. In addition, the presence of acidic stagnant water benefits acid-tolerant mosquitoes, which has implications for the spread of mosquito-borne diseases such as Ross River virus [Flexman et al., 1998; Ljung et al., 2009].

4.4. Potential Management Options Controlling Acidic Discharges

[45] The most cost-effective solutions to manage acid discharge from ASS-contaminated wetlands are drainage redesign or reflooding [Johnston et al., 2004; Sammut et al., 1996a; White et al., 1997]. Reflooding or inundation aims to re-establish reducing conditions to generate alkalinity and counter acidification [Dent, 1986; White et al., 1997, 2007]. The method depends on the supply of dissolved organic carbon for the growth of sulfate reducing bacteria and the generation of sufficient alkalinity to neutralize acidic waters [Johnston et al., 2005]. Our sediment data indicated abundant TOC (between 0.26% and 33%, Table 2) suitable for bacterial growth and the reduction of iron and sulfate as described by the following chemical reactions [Dent, 1986]:

Microbial Fe reduction:

display math

Microbial sulfate reduction:

display math

[46] Acidity is consumed during both iron and sulfate reduction. The generated sulfide can react with Fe2+ and precipitate as secondary pyrite, FeS2, which neither produces nor consumes hydrogen ions, or iron monosulfides, which produces hydrogen ions [Kalin et al., 2006]:

display math

[47] The precipitation of FeS or FeS2 removes metal acidity from the system.

[48] A potential complication associated with re-flooding of Muddy Lakes may arise due to the large seasonal variation in groundwater levels. Under these conditions, it may be difficult to keep the site permanently inundated. Also, the alternative drying and wetting conditions induced by the fluctuating groundwater table may destabilize the iron monosulfides causing a rerelease of acidity and adsorbed contaminants [Bush et al., 2004; Smith and Melville, 2004]. Control strategies should be adopted to minimize groundwater losses and evaporation during late summer. The application of thick mulches, such as straw, onto the scalds can create capillary breaks and reduce the evaporative concentration of acid salts at the land surface [Minh et al., 1997; Rosicky et al., 2006]. Mulching in conjunction with ridging and once-off liming may encourage the re-establishment of native vegetation. Mulching combined with re-flooding may also provide additional sources of labile organic carbon to support microbial reduction of sulfate and the production of alkalinity. During dry periods (i.e., October-March), the presence of mulch may also limit the exposure of monosulfide layers to oxygen and thus minimizes the cyclic buildup of acid salts.

[49] During reflooding, the changing redox conditions could lead to the reductive dissolution of minerals resulting in the enrichment of pore water with As, Fe2+, and other redox-sensitive elements [Johnston et al., 2011]. Our data demonstrated that the groundwater underlying the wetland exhibited seasonal changes in redox status, and increased concentrations of redox-sensitive species were observed. Such redox-triggered release of toxic metal(oid)s must be monitored through a pilot-scale study at the site before implementing a flooding strategy.

5. Conclusions

[50] Our conceptual model highlights interactions between seasonal hydrological processes and biogeochemical characteristics of wetland. The model indicates the time lag among acidity generation in the subsurface, precipitation of acidity products at the surface, and the dissolution and transport of acidic salts offsite through the major drain. The release of acidity products and trace metals threatens the survival of endangered species and communities living within the wetland as well as aquatic ecosystems in adjoining waterways. The conceptual understanding of the acidification processes and the seasonal hydrological connectivity will facilitate improved management of the wetland ecosystems and the protection of downstream waters.

Acknowledgments

[51] The authors would like to express thanks to Azra Mat Daud, Gary Newman, Paul Livsey, and many others for their assistance in the collection of field data. This project was funded by the Australian Federal Government through the Natural Heritage Trust Regional Competitive Component project 53454.

Ancillary