Water Air Soil Pollut (2012) 223:801–818 DOI 10.1007/s11270-011-0903-9

Performance of Pilot-Scale Sulfate-Reducing Bioreactors Treating Acidic Saline Water Under Semi-Arid Conditions Brad P. Degens

Received: 14 February 2011 / Accepted: 25 July 2011 / Published online: 9 August 2011 # Springer Science+Business Media B.V. 2011

Abstract Groundwater drains used to manage saline watertables in the semi-arid zone of south-western Australia can discharge acidic saline water with high concentrations of metals to waterways. Mitigating the acidity impacts of the waters requires sulfate-reducing bioreactors capable of functioning under semi-arid conditions with limited source materials. Two simple pilot-scale bioreactor designs using straw and sheep manure mixtures were evaluated over several years. The bioreactors increased pH from <3.5 to >5.5 for 125– 260 days, with concurrent evidence of sulfate reduction, >85% reductions in net acidity and >90% reductions in Al and most trace elements (e.g. Pb, Cu, Ni, Zn, Ce and La). When outflow pH<5.5 (remaining greater than inflows), reduction in net acidity was 10–80% but concentrations of Pb, Cu and Ni remained >80% reduced over periods of 250 to >700 days. Rates of alkalinity generation initially exceeded 10 g CaCO3/m2/ day in both bioreactors thereafter decreasing to >1– 2 CaCO3/m2/day. Al and Fe retention was implicated in trace metal removal when pH<5.5, mediated by biological alkalinity generation. High evaporation rates limited bioreactor function by restricting outflows with no benefits to alkalinity generation rates. This experiment showed that simple bioreactors can neutralise acidic waters and remove metals for short durations B. P. Degens (*) Department of Water, P.O. Box K822, Perth, WA, Australia 6892 e-mail: [email protected]

and show capacity for sustained reduction in acidity and metal concentrations over several years despite low alkalinity generation. Keywords Sulfate-reducing bioreactor . Acidity treatment . Metal precipitation . Acid drains . Salt water . Climate effects

1 Introduction Clearing of native vegetation has long been recognised to have altered the hydrology of landscapes in south-western Australia causing rising watertables and secondary salinisation (Hatton et al. 2003). The rising saline groundwater in many low-relief valley floors has also recently been found to be acidic (pH< 4.5) with evidence that this has resulted in acidification of lakes, waterways and floodplains (Shand and Degens 2008). The threat of continued rise in watertables and expansion of land salinisation has increased farmer interest in using deep leveed drains to manage saline groundwater discharge, particularly in low-relief valley floors (Hatton et al. 2003). Drains can accelerate discharge of acidic water to surface environments (Degens and Shand 2010) and have heightened demand for practical options to treat and manage these discharges (Degens 2009). Acidic saline water discharging from drains has been characterised as being similar to that of acid mine drainage (AMD), containing high concentrations

802

of Fe, Al and Mn, although more saline, frequently exceeding concentrations of salt occurring in seawater (Degens and Shand 2010). The waters are typically highly oxidised (Eh>400 mV) and can also contain high concentrations of trace elements such as Pb, Ce, Cu, La, Ni and Zn (Shand and Degens 2008), although often less than that encountered with AMD. These trace elements, along with the acidity of the waters present a clear hazard to aquatic ecosystems in inland waterways and lakes (Degens 2009; Stewart et al. 2009). The highly oxidised, iron and aluminium-rich acidic drain waters discharging from leveed drains in flat landscapes are ideally suited to sulfate-reducing bioreactors such as composting treatment wetlands prescribed for similar quality AMD (Hedin et al. 1994; Younger et al. 2002). These are also attractive to farmers in that the systems are low maintenance and passive. Rate of acidity treatment (or conversely alkalinity generation) per unit area or volume is a fundamental design parameter for simple bioreactors such as anaerobic composting wetlands and other bioreactor designs such as permeable reactive barriers (Younger et al. 2002). Treatment of acidic metal rich waters in these systems is primarily driven by rates of bacterial sulfate reduction which is dependent on the mix of cellulosic and non-cellulosic organic materials used in construction (Logan et al. 2005; Neculita et al. 2007). The availability of these in many parts of south-western Australia is limited to straw residues from extensive cereal production systems and low volumes of manures from the few intensive animal production systems. However, bench-top bioreactor investigations found that combinations of straw with local non-cellulosic materials showed promise in supporting sulfate reduction at rates that achieved treatment of acidic saline drain waters (Santini et al. 2010). The semi-arid climate of inland south-western Australia presents challenges in designing sulfatereducing bioreactors, which are mostly based on experiences in cooler, wetter climates in the northern hemisphere with comparatively fresh waters (e.g. Hedin et al. 1994; Younger et al. 2002; Ziemkiewicz et al. 2003). Use of sulfate-reducing bioreactors such as anaerobic composting wetlands in semi-arid and arid climates may be restricted by high evaporation rates (Tyrell et al. 1997). The climate of inland southwestern Australia characteristically has highly sea-

Water Air Soil Pollut (2012) 223:801–818

sonal rainfall and high evaporation rates (1,800– 2,400 mm/annum) with mild winters and warm summers. These semi-arid conditions are likely to reduce water flow through bioreactors (and continuity of flow) and increase water salinity and temperatures. The combined effects of increased salinity and temperature on alkalinity production (and therefore acidity treatment) driven by bacterial sulfate reduction are poorly known (Tyrell et al. 1997). However, these would be expected to benefit rates of alkalinity production provided the bioreactors remain flooded. Rates of sulfate reduction can be greater with increasing salinity, mostly attributed to increased sulfate concentrations (Kerkar and LokaBharathi 2007). Similar effects can occur with increased water temperatures (Faulwetter et al. 2009; Nedwell and Abram 1979), although are dependent on sulfate and the supply of organic substrates being non-limiting (Nedwell and Abram 1979). This study presents results of several years of evaluation of the first pilot-scale sulfate-reducing bioreactors to treat saline acidic drain waters under cool semi-arid zone conditions. Alkalinity generation rates and reduction in acidity were determined and compared with metal retention and hydraulic limitations introduced by the semi-arid climate. The bioreactors consisted of a sulfate-reducing bed within the base of a leveed groundwater drain (in-drain) to treat acidity at the point of groundwater discharge and an open sulfate-reducing pit to treat acidic water collected within a storage basin (end-of-drain). These simple designs using simple organic mixtures defined the lower bench-mark in the application of these systems.

2 Methods 2.1 Bioreactor Design and Construction 2.1.1 In-Drain Bioreactor Bed The in-drain bioreactor was constructed in the base of a 200-m section of leveed deep (2.5 m) drain forming the upper part of a 9.6-km drainage network near the town of Doodlakine, 200 km east of Perth, Western Australia. The drain intercepted shallow acidic saline groundwater (<2 metres below ground level) with levee banks designed to exclude surface-water runoff

Water Air Soil Pollut (2012) 223:801–818

803

into the channel. Flow and acidity were monitored over 3 months and, with indicative treatment rates from bench-top trials (Santini et al. 2010), used to broadly guide the initial design of the bioreactor bed (the volume of organic materials likely to achieve treatment per square metre of the drain base). There were three phases in the construction and operation of the in-drain bioreactor, corresponding with initial commissioning and subsequent additions of non-cellulosic organic materials. Phase 1 consisted of 1% by volume of composted sheep manure (noncellulosic material) sandwiched between wheat straw (cellulosic material) to create a bed of 0.35–0.45-m thickness (Fig. 1a). The composted sheep manure contained 2.3% N and 0.7% P with C/N of 12.6 while the wheat straw contained less than 0.4% N with C/N of 90:1. Water level in the bioreactor was raised to cover the organic bed by 25 mm (Fig. 1a) and enabled measurement of outflow using a flow meter. ContinFig. 1 Plan diagrams of the (a) in-drain bioreactor bed (showing the cross-section view) and (b) end-of-drain pit bioreactor (top view)

uous discharge of groundwater to the system prevented the common practice of pre-conditioning the bioreactor to allow sulfate-reducing processes to establish before operation (Younger et al. 2002). Phase 2 started at day 206 when additional composted sheep manure was mixed into the bioreactor to rectify initial poor performance. This material was combined into two sections (6 and 10 m length) and increased the overall non-cellulosic material volume to 5% of the total organic volume. Phase 3 started at day 602 when waste wheat grain (2.1% N, 0.35% P and a C/N of 20) was mixed into six 2-m sections increasing the non-cellulosic proportion to 6.2% with the feed grain being 1.2% by volume. 2.1.2 End-of-Drain Pit Bioreactor A pilot pit bioreactor was constructed to treat water from a storage basin containing over 100

(a)

0m

Flow meter

Bunds with water level control pipes

Organic mix with water cover

Flow direction 2.5m Groundwater discharge 220 m

110m

0m

(b) Outflow sump drain Slotted pipe with riser outlet Flow direction through organic mix Inflow pool

Outflow pool Inflow Drain water storage basin (source water)

804

million litres of acidic saline water (Fig. 1b) collected from 25 km of groundwater drains near Beacon, 260 km north-east of Perth, Western Australia. A simple single-cell design (100 m long, 3 m wide and 1 m deep) lined with 0.5-mm polythene plastic sheeting was constructed at right angles to the local land gradient so that water flowed along the floor of the cell (Fig. 1b). The organic mixture was loaded into the reactor as a layer of composted sheep manure overlaid with 0.5 m of wheat straw to make a final mixture consisting of 4.7% (by final volume) composted sheep manure with wheat straw. These materials were similar to the initial materials used in the in-drain bioreactor. The simple layered construction was the lowest cost construction approach requiring minimal preprocessing or handling of the organic materials. Sections of the pit at the inflow and outflow points (1.5 m) were retained free of organic materials to serve as inflow distributor and outflow collection pools. Water levels in the bioreactor were controlled at the outflow point by a riser placed on an outflow pipe (Fig. 1b). The pit bioreactor was filled with acidic saline water (pH 3.2, 560 mg CaCO3/L net acidity, 17 mg Fe/L, 70 mg Al/L, 55.6 g total dissolved salts/L) from the adjacent drain water storage basin (Fig. 1b) and pre-conditioned for 9 weeks to allow sulfatereducing processes to establish. Following this, a continuous inflow of acidic source water was maintained to the inflow cell of the bioreactor using a battery powered pump, except for one brief period at 232 days. Flows to the pit bioreactor were reduced on occasion to address rising net acidity concentrations in outflow waters. Throughout the experiment, water levels were kept near the top of the organic bed along most of the bioreactor and not ponded above to limit surface oxidation, reduce water loss by evaporation and prevent flow shortcircuiting over the organic bed.

Water Air Soil Pollut (2012) 223:801–818

an electromagnetic flow meter (EmFlux 2020 electromagnetic flow meter with I300 transmitter) with an error of less than 1% volume. Outflow from the pit bioreactor was measured at the time of sampling but not continuously recorded. Groundwater discharging to the in-drain bioreactor was monitored at 91, 380 and 890 days using twelve 4-m deep groundwater wells (with 3 m screens) constructed in two rows of six at 8–12 m either side of the drain. Samples were obtained with a peristaltic pump using low-flow sampling techniques. Water quality measurements included in situ measurement of pH, temperature, oxidationreduction potential (using Ag/AgCl probe) and dissolved oxygen (DO). Samples collected for analysis were filtered through 0.45-μm membrane filters, acidified with ultrapure nitric acid (to 0.1%) to preserve trace elements or stored at 5°C. The concentrations of major cations (Na, Ca, K and Mg) and elements such as Al, Fe and Mn were determined by inductively coupled plasma optical emission spectrometry. The trace metals Co, Cd, Cu, Ni, Pb and Zn and elements such as La and Ce were determined by inductively coupled plasma mass spectrometry (ICP-MS). Arsenic was also determined using the hydride generation coupled to ICP-MS. Selected anion analyses were conducted using ion chromatography (SO42−) or colourimetic methods using a segmented flow analyser (NO3−, NH4+, PO43−). Alkalinity for samples with pH>4.5, was determined by dilute H2SO4 titration to pH 4.5. Total nitrogen (TN) was determined using a Formacs HT TOC/TN Analyser with chemi-luminescence detector. Dissolved organic C (DOC) was determined directly by non-dispersive infrared detector after removal of CO2, HCO3− and H2CO3 by acidification and purging with oxygen. Total dissolved salts (TDS) were calculated as the sum of major cations and anions. Only samples verified for charge balance (<5%) are reported or used in geochemical modelling.

2.2 Monitoring 2.3 Data Analyses Measurements and sampling of outflow water quality and flow were made prior to construction and after construction of the in-drain bioreactor and on inflow and ouflow to the pit bioreactor. Outflow from the in-drain bioreactor and inflow to the pit bioreactor was continuously recorded using

Net acidity was calculated using the formula, net aciditycalculated =acidcalculated −alkalinitymeasured (Kirby and Cravotta 2005). Aciditycalculated was calculated from metal concentrations and using Eq. 1 (modified from Kirby and Cravotta 2005) assuming all dissolved

Water Air Soil Pollut (2012) 223:801–818

805

Fe was present as Fe2+ with negligible Fe3+(based on speciation calculations using PHREEQC). A ¼ 50; 000  10pH þ 1:79  Fe2þ þ 2:68  Fe3þ þ 5:56  Al3þ þ 1:82  Mn2þ

ð1Þ

Where A Fe2+ Fe3+ Al3+ Mn2+

is calculated acidity (milligrams per litre CaCO3 equivalent) is Fe2+ concentration (milligrams per litre) is Fe3+ concentration (milligrams per litre) is Al3+ concentration (milligrams per litre) is Mn2+ concentration (milligrams per litre).

The hydraulic residence time (HRT) of outflow water from the pit bioreactor was estimated to indicate the average time that water had remained in the treatment system. This was not carried out for the indrain bioreactor because of the uncertainty in point of inflow (via groundwater discharge). It was assumed that piston flow occurred through the pit bioreactor with water lost or gained only by evaporation or rainfall. Calculation of HRT differentiated between when outflow represented continuous piston flow (HRTcf) from when outflow contained water trapped during periods of no outflow (HRTrw and HRTnof). Average HRTcf was calculated between sampling times (tx to ty) by Eq. 2. HRTcf ¼

Vw  KET Qtytx

ð2Þ

where VW = dead-water volume in the bioreactor (estimated at 60 m3), Qty-tx =average flow rate (cubic metre per day) during estimation period (tx to ty) and KET =correction for evaporation/rainfall between tx and ty using proportional change in [Cl−]. HRT of the discharge after high evaporation periods was estimated by Eqs. 3 and 4 where outflow was initially water retained since the last period of outflow (HRTrw) followed by water that had accumulated in the bioreactor while no outflow occurred (HRTnof). HRTrw ¼ HRTtco þ ðtrcf  tco Þ

ð3Þ

HRTnof ¼ trcf  tco

ð4Þ

where HRTtco =HRT of water in the bioreactor when outflow ceased (tco) and trcf − tco = time between outflow ceasing (tco) and restarting (trcf).

Rates of alkalinity production per square metre bioreactor (being equivalent to rates of acidity treatment per square metre) were calculated by difference between inflow and outflow net acidity load (Ziemkiewicz et al. 2003) for periods of time greater than the calculated HRT. This was deemed more conservative than estimating treatment per unit bioreactor volume. Net acidity load was estimated between water quality sampling times from the product of flow volume and net acidity assuming that measurements reflected the quality since the previous sample. For the in-drain bioreactor, outflow volumes were calculated as metered outflow volumes minus rainfall inputs to the drain since water-quality measurements reflected base-flow conditions. Rainfall inputs were calculated from interpolated daily rainfall data (Bureau of Meteorology 2010) assuming 100% input to the drain and 10% runoff from within the drain levees. Inflow of acidic water (being groundwater discharge) was estimated as outflow volumes (as above) plus evaporation loss between sampling times estimated from daily interpolated pan evaporation (PE) data for the site (Bureau of Meteorology 2010). Actual evaporation was assumed to be 0.7×PE with negligible effect of water salinity on evaporation (since salinity<100 g/L). Net acidity in groundwater inflow to the bioreactor was assumed stable at 330 mg CaCO3/L (average of monitoring wells across all sample dates) based on annual average groundwater net acidity being 336 to 348 mg CaCO3/L (with standard deviations<170 mg CaCO3/L). Rates of alkalinity production in the end-of-drain pit bioreactor were calculated for time intervals of at least 40 days by difference between net acidity load in inflow and outflow. The calculation interval corresponded with the average HRT estimated for the bioreactor. Loads were calculated for each sampling interval and summed for periods of >40 days but only during periods when outflow occurred. Inflow volumes for the load calculations were obtained from flow meter recordings whereas outflow volumes were calculated by adding rainfall and subtracting evaporation as conducted for the in-drain bioreactor. The solubility of selected minerals and speciation of selected dissolved elements in the bioreactors was evaluated using outflow water chemistry with PHREEQC version 2.13.2 (Parkhurst and Apello 1999). The phreeqc.dat thermodynamic database was used with log K values for selected phases obtained

806

Water Air Soil Pollut (2012) 223:801–818

from the minteq.v4 database (Parkhurst and Apello 1999). Sulfate removal was assessed using SO4/Cl, assuming Cl as a conservative ion, to avoid confounding effects of changes in concentrations of SO42− ions due to evaporation.

three sampling times for individual wells (standard deviation<45 mg CaCO3/L) and no specific pattern of spatial variation. Following construction of the bioreactor, outflow water exhibited a pattern of increasing pH and decreasing net acidity over the first 48–100 days of each phase of the experiment followed by gradual decreases in pH and increases in net acidity (Fig. 3a). Throughout phase 1, outflow pH rarely exceeded pH 5 and net acidity varied 179–321 mg CaCO3/L (Fig. 3a) with redox potential rarely reaching less than 0 mV (Fig. 3b). In contrast, pH steadily increased to a plateau of 6.5–6.6 during the first 48 days of phase 2 with net acidity decreasing to less than −400 mg CaCO3/L and redox potential approaching −200 mV (Fig. 3b). This trend was not sustained with pH decreasing to 3.6–5 and net acidity increasing over the subsequent 350 days, although remained less than groundwater inflow (Fig. 3a). During the final phase, outflow pH reached a plateau of 6.5–7.0 lasting more than 180 days with net acidity decreasing to less than 0 mg CaCO 3 /L (Fig. 3a) and redox potential approaching −200 mV (Fig. 3b). However, net acidity did not stabilise (Fig. 3a) and subsequently increased gradually to vary between 107 and 230 mg CaCO3/L with outflow pH falling below 5.5 (but not less than 4; Fig. 3a). During phase 3, pH remained greater than 5.5 for 260 days. Extensive sulfidic precipitates were observed to form on the surface of the bioreactor during the early part of each phase while pH<5.5. TDS concentrations and temperature in outflow water from the bioreactor followed a pattern reflecting seasonal conditions. TDS concentrations varied from

3.1 In-Drain Bioreactor 3.1.1 Hydraulics and Performance Parameters

Fig. 2 Groundwater (GW) inflow and total outflow in the in-drain bioreactor with corresponding groundwater acidity load and rate of alkalinity generation (seasons indicated as S-A=summer–autumn and W-S=winter–spring)

Acidity load/alkalinity generation rate (g CaCO3/m2/day)

Groundwater discharge comprised over 95% of the inflow to the bioreactor and seasonally varied from less than 3 kL/day in late autumn to more than 5 kL/ day in late winter (Fig. 2). This reflected variation in watertable levels in the soil adjacent to the reactor of between 0.1 and 0.5 m above the bioreactor bed. Although HRT was not possible to calculate, it was estimated to be less than 11 days (based on groundwater inflows and a dead-water volume of 40 m3). Untreated waters prior to bioreactor construction (Fig. 3) were saline (36–40 g TDS/L), acidic (pH 2.8– 3.8; net acidity 440–541 mg CaCO3/L) and highly oxidising (Eh 600–660 mV). This reflected the quality of the shallow groundwater discharge to the drain as indicated in 12 monitoring wells adjacent to the drain. Shallow groundwater was similarly saline (14–38 g TDS/L), acidic (pH 3.2–5.0; net acidity 100–616 mg CaCO3/L) and generally highly oxidising (Eh 254–573 mV). There was little variation in the net acidity of shallow groundwater between the

Phase 1

40

Phase 2

9

Phase 3

8

35

6 5

30

3 25

2 Acidity inflow load

20

Alkalinity generation rate Total discharge

15

GW inflow

10 5 S-A

0 0

W-S

S-A

200

S-A

W-S

400

600

W-S

800

Days after establishment

S-A

1000

1200

0

Discharge (kL/day)

3 Results

Water Air Soil Pollut (2012) 223:801–818

(a)900

Phase 1

9

Phase 2

Phase 3

600

8

300

7

0

6

-300

5

-600

4

-900

3

pH

Net acidity (mg CaCO3/L)

Fig. 3 Evolution of (a) net acidity and pH and (b) redox potential (Eh), total dissolved salts (TDS) and temperature of outflow waters from the in-drain bioreactor (seasons indicated as per Fig. 2)

807

-1200

Net acidity

pH

-1500 -1800 -100

2 1

S-A

W-S

0

100

200

W-S

S-A

300

400

500

S-A

600

700

W-S

800

S-A

0

900 1000 1100 1200

Days after establishment

(b)

60

Phase 2

Phase 3

50 40

Eh (mV)

30 600

20

400

10

0

200 Eh Salinity Temperature

0 -200 -100

S-A

W-S

0

100

200

S-A

300

400

W-S

500

W-S

S-A

600

700

800

S-A

-10

Salinity (g TDS/L) and temperature (°C)

Phase 1

-20

900 1000 1100 1200

Days after establishment

less than 29 g TDS/L in winter and spring when evaporation rates were least to more than 52 g/L in summer and autumn when evaporation rates were greatest (Fig. 3b). A similar pattern was evident in water temperatures (reflecting day–time temperature) where outflow water temperatures were generally mild in winter (13–20°C) but could be warm during the remainder of the year (25–30°C). While outflow temperatures were probably greater than the temperature within the bioreactor bed for most of the time, these were likely to vary in direct proportion. 3.1.2 Sulfate, Dissolved Elements and Nutrients There were periodically large decreases in the SO4/Cl of outflow water compared with source groundwater with no similar trends in Na/Cl, Mg/Cl or K/Cl (data not shown). Outflow SO4/Cl tended to be similar to that of source groundwater during phase 1 (Fig. 4),

but was significantly lower during the early part of phase 2 (days 200–300) and phase 3 (days 650–850) indicating depletion of sulfate. Similarly, outflow SO4/Cl tended to be consistently less than that of the source groundwater towards the end of phase 3 (day 850 onwards; Fig. 4). Outflow Ca/Cl was generally similar to source groundwater during phases 1 and 3, but was elevated for approximately 100 days during the early part of phase 2 (Fig. 4). Untreated waters (pre-construction and source groundwater) contained generally high concentrations of Fe and Al and a range of trace elements mostly at average concentrations less than 370 μg/L (Table 1). Drain water prior to treatment contained an average of 104 mg Fe/L and 43 mg Al/L (Table 1) which for Fe was similar to the shallow groundwater (115 mg/L) but almost double that of Al concentrations in groundwater (Table 1). Of the trace elements, the lanthanides elements (Ce and La) generally occurred

808 Phase 1

Phase 2

0.030

Phase 3 Groundwater SO4:Cl

0.12

0.025

0.10 Outflow SO4:Cl

0.020

0.08 0.015 0.06

0.005

0.02 S-A 0.00 -100 0

0.010

Outflow Ca:Cl

0.04

Ratio of Ca:Cl

0.14

Ratio of SO4:Cl

Fig. 4 Ratios of SO4/Cl and Ca/Cl in outflow waters from the in-drain bioreactor (bold long dash line) with average (solid line), 10th and 90th percentile (dash line) ratios for shallow source groundwater (seasons indicated as per Fig. 2)

Water Air Soil Pollut (2012) 223:801–818

Groundwater Ca:Cl W-S

S-A

W-S

S-A

W-S

S-A

0.000

100 200 300 400 500 600 700 800 900 1000 1100 1200

Days after establishment

in greatest concentrations in untreated waters, followed by Co>Pb>Ni>Zn>Cu (Table 1). Concentrations of Cr, As and Cd were very low, often being at or less than detection limits of 0.3 μg/L. A variety of metals and trace elements were removed from the drain water depending on the pH of outflows. Results were summarised depending on whether water pH was greater than or less than 5.5, which corresponded with when net acidity was generally less than 0 mg CaCO3/L. Average concentrations of Al and Fe of outflows were more than 80% less (P<0.05) than either pre-treatment water or source groundwater when pH>5.5 (Table 1), whereas Mn increased (P>0.01). Under these conditions, average concentrations of trace elements such as Co, Ce, Ni, La and Pb were more than 90% less (often to below detection limits) than in pre-treatment water or source groundwater (P<0.02; Table 1). Decreases in concentrations of Zn and Cu of at least 80% were also evident (P<0.05; Table 1).When pH decreased to less than 5.5 average outflow concentrations of Al remained decreased by more than 75% (P<0.01), but not for Fe (P>0.05). Reduction in trace element concentrations continued for some elements while pH<5.5 with average Cu, Ni and Pb remaining at least 80% less than in pre-treatment water or source groundwater (P< 0.01; Table 1). These reductions were sustained over periods of more than 250 days. Concentrations of PO43− and DOC in bioreactor outflows were increased to the greatest extent of all nutrients compared with untreated waters (pre-treatment water and the source groundwater) (Table 2). The increases were greatest (almost twofold) while outflow pH>5.5 and diminished when pH<5.5. In

contrast, TN, NO32− and NH4+ remained similar to that of untreated waters (Table 2). 3.1.3 Rates of Alkalinity Generation Rates of alkalinity generation varied significantly during each phase being characteristically greater during the early part (after addition of organic materials) and thereafter decreasing to lower plateau levels. During phase 1, rates of alkalinity generation ranged from 2.1 to 8.8 g CaCO3/m2/day (Fig. 2) which was calculated to have neutralised half of the groundwater acidity load. During phases 2 and 3, these rates exceeded 10 g CaCO3/m2/day for 100–175 days, but were not sustained and decreasing to between 2 and 3 g CaCO3/m2/day (Fig. 2). Maximum alkalinity generation rates during peak activity were more than double that of net acidity loading in groundwater inflow (Fig. 2). Decreases in alkalinity generation rates with time were not gradual and appeared to follow patterns of groundwater acidity loading more than temperature or TDS concentrations. 3.2 End-of-Drain Pit Bioreactor 3.2.1 Hydraulics and Performance Parameters Inflow to the pit bioreactor averaged 1.7 kL/day, with some variation during the first spring (83–205 days) due to initial difficulties with the pumping system (Fig. 5). Discharge from the bioreactor was more variable, frequently less than half of inflows and often ceased during summer–autumn periods due to high

35 (24–64) 35 (17–105)

Ni

Zn

10.4

38

52

143

9.3

369

85

0.4

Below analytical detection limits

Number of samples on which summary values are based

Mean and range data from 3-year average of 12 monitoring wells

69 (16–114) 22 (5–38)

La

15.7 (4–65)

Cu

Pb

66 (51–82) 141 (30–234)

Co

Ce

0.7 (0.2–2.2)

Mn

104

43

Outflow (pre-treatment) (n=3)b

2.0* (<0.4 –6) c

13 (2–124)

95** (6–180)

17* (<0.1c–130)

11* (<0.8c–29)

2432 (288–1936)

10* (3–361)

2.5* (<0.5c–10)

1183** (564–4181)

192* (<41–459)

10** (4–18)

16 (1–70)

12* (2–60)

Outflow when pH<5.5 (n=26)b

0.2* (<0.1c–2)

<2c* (<2c)

20* (3–515)

28* (10–55)

7.0 (5–10)

17 (2.5–37)

5.1* (2.5–6)

Outflow when pH≥5.5 (n=5)b

16 (<3 –80) c

160 (70–228)

1400 (474–2097)

1436 (507–2120)

22 (2–51)

3576 (1310–4918)

220 (93–308)

6.4 (2.5–8)

15 (9–20)

65 (45–127)

(n=41)b

Inflow

Pit bioreactor

125 (14–359)

4.4* (<0.4 –25)

**P<0.05 level of significance of values from inflow or pre-treatment/inflow groundwater

c

2.1* (<0.5c–5)

10 (<0.4c–76)

8.2* (<0.5c–32)

2.6* (<0.5c–11)

8.5* (<2 –60) 0.6* (<0.5c–1)

c

1.4* (<1c–9)

273 (10–932)

38* (3.4–83)

1.3* (0.1–2.9)

91 (32–178)

10* (1–28)

Outflow when pH<5.5 (n=23)b

1.3* (<1c–11)

16* (2.6–117)

3.9* (0.4–23)

1.8* (0.9–4.2)

21** (0.03–95)

1.1* (0.05–5)

Outflow when pH≥5.5 (n=29)b

*P<0.01 level of significance of values (two-tailed, unequal variance paired sample T tests) from inflow or pre-treatment/inflow groundwater

c

b

a

Trace elements (μg/L)

24 (1.9–51) 115 (29–216)

Al

Major metals (mg/L)

Shallow groundwatera (n=12)b

Fe

Element

Grouping

In-drain bioreactor bed

Table 1 Mean and range concentrations of major metals and trace elements in inflow and outflow water for the pit bioreactor and in-drain bioreactor bed

Water Air Soil Pollut (2012) 223:801–818 809

**P<0.05 level of significance of values from inflow or pre-treatment/inflow groundwater

*P<0.01 level of significance of values (two-tailed, unequal variance paired sample T tests) from inflow or pre-treatment/inflow groundwater

Number of samples on which summary values are based

Below analytical measurement limits c

Mean and range data from 3-year average of 12 monitoring wells

12 (7–19)

b

2.8 2.7 (1.3–3.8) TN

DOC

14

2.1 2.7 (1.3–3.8)

a

2.7 (0.47–8.5)

19** (12–38) 44* (11–112)

5.2* (0.53–22) 5.2 (2.9–7.8)

16 (9–26) 43** (13–67)

8.1** (3.6–19) 34* (10–59)

4.2 (1.1–5.4) 6.7 (1.2–17) 4.9 (2.1–9.6)

344* (60–600)

2.4 (<0.1c–12) 3.8 (<0.1 –22)

0.1 (<0.01c–0.2)

c

<0.01 1.6 (0.3–5.2)

0.61** (<0.01 –6.8) <0.01 *

0.01 (<0.01 –0.30)

<0.01

0.42* (<0.05c–3.1) 0.046 (<0.05c–0.17)

c c

0.042 (<0.05c–0.15) 3.0* (0.03–12)

c c

NO3− NH4+

c

0.009 (<0.05c–0.05) PO43−

<0.05c

(n=41)b

Outflow when pH≥5.5 (n=29)b Inflow

Shallow groundwatera (n=12)b Constituent (mg/L)

Outflow (pre-treatment) (n=3)b

Outflow when pH≥5.5 (n=5)b

Outflow when pH<5.5 (n=26)b

Pit bioreactor In-drain bioreactor bed

Table 2 Mean and range concentrations of nutrients and dissolved organic C in inflow and outflow water for the pit bioreactor and in-drain bioreactor bed

0.033** (<0.05c–0.3)

Water Air Soil Pollut (2012) 223:801–818

Outflow when pH<5.5 (n=23)b

810

evaporation rates (Fig. 5; indicated by breaks in outflow lines on the figures). The variations in flows resulted in HRT varying between 9 and 45 days during the first winter–spring (up to 160 days) and thereafter increasing to consistently greater than 20 days (Fig. 5). HRT varied between 41 and 76 days in the second winter–spring period and between 19 and 51 days during the third winter–spring period with intermittent high HRT caused by cessation of outflows during high evaporation conditions in summer–autumn periods (Fig. 5). This resulted in HRT increasing to more than 100 days, but only for periods less than 45 days while water was trapped during the ceased outflow was flushed from the bioreactor (Fig. 5). During periods of continuous through flow, evaporation increased HRT by up to 60% compared with expected HRT in the absence of any evaporation (calculated as dead volume divided by average daily inflows). Over the 882-day monitoring period, inflow water to the pit bioreactor exhibited relatively stable pH ranging from pH 2.8–3.5 (Fig. 6a), however the net acidity ranged from 500 to 900 mg CaCO3/L (Fig. 6a). The net acidity of the inflow steadily increased from 500 to 673 mg CaCO3/L during winter–spring (0–120 days) to over 800 mg CaCO3/L in summer–autumn (Fig. 6a), but gradually decreased thereafter to less than 500 mg CaCO3/L. Outflow water from the pit bioreactor had a decreasing pH trend throughout the 882 days of monitoring, contrasting with net acidity where greatest change occurred during the first 161 days (Fig. 6a). Outflow pH steadily decreased from 6.7 to 5.6 during the first 125 days, although after this, pH decreased sharply to less than 4.5 and generally varied 3.6–4.8 thereafter with a slight decreasing trend (Fig. 6a). Outflow pH appeared to gradually converge on the inflow pH, however was always remained at least 0.4 units greater than inflow pH (Fig. 6a). Similarly net acidity of outflow was consistently less than inflow net acidity despite the sometimes significant evaporation occurring in the bioreactor (evident in TDS concentrations; Fig. 6b). During the first 124 days, outflow net acidity was more than 1,000 mg CaCO3/L less than inflow net acidity with this difference diminishing thereafter to generally range 160–250 mg CaCO3/L less than that in inflows (Fig. 6a). Outflow water after periods of

Water Air Soil Pollut (2012) 223:801–818 250

HRT

200

Daily net evap.

150

Inflow rate Outflow rate

100 50

10 8

0 W-S

W-S

S-A

W-S

S-A

6

4 2 0

100

300

400 500 Days

600

700

0 900

800

was continuous. This applied to outflow on days 198– 205, 359–380 and 587–707. The redox potential (Eh) of inflow and outflow did not follow a seasonal pattern and tended to corre-

ceased flow (with substantially increased HRT; Fig. 5), did not exhibit any discernable increase in pH or decrease in net acidity relative to outflow waters during subsequent periods where through flow

(a)

300 0

Inflow pH Outflow pH Inflow net acidity Outflow net acidity

7

-600 -900

6

pH

-300

Net Acidity (mg CaCO3/L)

900 600

No outflow

5

No outflow

No outflow

4 3 S-A

W-S

2

0

100

200

400

W-S

S-A

W-S

300

500

600

700

800

900

(b)

Inlet Eh

Outlet Eh

Inflow TDS

Outflow TDS

250

200 150 100

50

800

0

600 400 200 0 S-A

W-S

-200 0

100

200

S-A

W-S

300

400

500

Days

600

W-S

700

800

900

Total dissolved salts (g/L)

Days

Eh (mV)

Fig. 6 Temporal pattern of (a) pH, net acidity and (b) Eh, total dissolved salts (TDS) in inflow and outflow water for the pit bioreactor (with seasons indicated as per Fig. 2)

200

Discharge (kL/day)

HRT (days) & daily net evap. (mm)

Fig. 5 Hydraulic residence time (HRT) of outflow (filled squares: HRTcf as per Eq. 2 and open squares: HRTrw and HRTnof as per Eqs. 3 and 4), average daily inflow and estimated outflow and daily net evaporation (evaporation–rainfall) for the pit bioreactor (with seasons indicated as per Fig. 2)

811

812

spond with variation in net acidity (Fig. 6b). The inflow waters were consistently highly oxidised with an Eh varying from 493 to 688 mV. In contrast, the Eh of outflow discharge was generally less than 0 during the first 124 days and corresponded with observation of extensive formation of black precipitates on the surface of the bioreactor. Thereafter, outflow Eh steadily increased to at level marginally less than in inflows (Fig. 6b). Outflow Eh probably reflected conditions in the outflow pool rather than the Eh of the bioreactor matrix, which was generally at least 200 mV less in direct measurements (data not shown). The salt concentrations of inflow water varied between 53 g TDS/L and 89 g TDS/L and were generally lower during the winter–spring periods compared with summer–autumn periods (Fig. 6b). Water discharging from the bioreactor often contained TDS concentrations more than 10% greater than that of inflow water; generally increasing towards the end of the winter–spring period (corresponding with greater temperatures; Fig. 6b). TDS concentrations in bioreactor outflows were more than 150% that of inflows during these times. The patterns of TDS concentrations were weakly correlated with net evaporation rates (r2 >0.2; data not shown). Inflow and outflow water temperatures (sampled during daylight hours) were generally similar and varied from 14 to 25°C in winter–spring periods to generally 17–33°C in summer–autumn periods. 3.2.2 Sulfate, Dissolved Elements and Nutrients Depletion of SO4 was evident as decreases in SO4/Cl of outflows at various times during the experiment. The SO4/Cl in inflows were less than outflows by more than 0.01 during the first 100 days with smaller differences of 0.004–0.02 intermittently evident thereafter (e.g. days 471–510, 624–653 and 775–810; Fig. 7). Periods when outflow SO4/Cl was greater than inflow (at the same time) were generally associated with flow after periods of ceased flow (evident as gaps in the lines in Fig. 7), but were often less than SO4/Cl of previous inflows (Fig. 7). These differences occurred against a varying background where inflow SO4/Cl ranged between 0.14 and 0.18. Ratios of Na/Cl, Mg/Cl and K/Cl in outflow water closely followed that of inflow water, except for initially elevated K/Cl during the first 93 days (data not shown). Similarly, outflow Ca/Cl was generally

Water Air Soil Pollut (2012) 223:801–818

similar to inflow Ca/Cl except during the first 40– 55 days of flows after periods of ceased outflow, probably mediated by dissolution of gypsum formed during high evaporation periods. The pit bioreactor reduced the concentrations of metals such as Al and many trace elements in inflow water for longer than 700 days, despite evaporation having an intermittently large effect on outflow salt concentrations. Untreated waters contained generally high concentrations of Al (average 65 mg/L) and to a lesser extent Fe and Mn concurrent with a range of trace elements mostly at average concentrations less than 3,500 μg/L (Table 1). Outflow Al, Co, Cu, Ce, Ni, La, Pb and Zn were reduced by more than 85% compared with inflows (P<0.01) while pH remained above 5.5 (Table 1), in some cases to below detection limits (e.g. Cu). In contrast, concentrations of Fe and Mn were not generally reduced in outflows compared with inflows (Table 1). Low concentrations of elements such as As and Cd prevented clear determination of differences between inflow and outflow. Concentrations of Al, Cu and Pb continued to be reduced by more than 80% (P<0.01) when outflows had pH<5.5 with lesser (but still statistically significant) reductions in Ni and Ce (Table 1). The outflow concentrations of most elements retained while pH> 5.5 did not rise above inflow concentrations when outflow pH decreased. As with the in-drain bioreactor, concentrations of PO43−, TN and DOC in bioreactor outflows were increased to the greatest extent of all nutrients measured compared with inflows (Table 2). The increases were greatest (more than an order of magnitude) while outflow pH>5.5 and diminished when pH<5.5. In contrast, while NH4+ remained similar to that of inflow waters irrespective of pH, NO32− was decreased in outflows, but only when pH> 5.5 (Table 2). 3.2.3 Rates of Alkalinity Generation Rates of alkalinity generation varied from 5 to 11 g CaCO3/m2/day during the first 110 days and decreased thereafter to 3–4 g CaCO3/m2/day (Fig. 7), after which outflows ceased (at 161 days). These rates occurred during a period of high and erratic acid loading rates that were subsequently reduced to mitigate overloading. When outflows recommenced in the second winter–spring (400–510 days) rates of

16

0.18

14

0.16 0.14

12

0.12

10

Net acidity load (sampling interval) Rate of alkalinity generation

8

0.08

Inflow SO4:Cl

6

0.1

Outflow SO4:Cl

SO4:Cl

Fig. 7 Net acidity loading (for sampling intervals) and rate of alkalinity generation in the pit bioreactor after establishment with corresponding SO4/Cl in inflow and outflow (seasons indicated as per Fig. 2)

813

Net acidity load & rate of alkalinity generation (g CaCO3/m2/day)

Water Air Soil Pollut (2012) 223:801–818

0.06

4

0.04

2

0.02 W-S

0 0

100

W-S

S-A

200

300

400

W-S

S-A

500

600

700

800

0 900

Days

alkalinity generation had decreased to 0.6–1 g CaCO3/m2/day and in the following winter–spring period (720–882 days) was initially less than this but recovered to 1.5–1.8 g CaCO3/m2/day thereafter (Fig. 7). Although rates of alkalinity generation were not determined during periods when no outflow occurred (most summer–autumn periods), this would not have contributed to an underestimation of generation rates because the net acidity of outflow waters reflecting high HRT periods were not significantly reduced relative to inflows.

4 Discussion 4.1 Rates of Alkalinity Generation Basic bioreactors with simple mixes of local organic materials showed a sustained capacity to reduce acidity and metals in highly acidic saline waters under semi-arid conditions. Less than 5% composted sheep manure layered with wheat straw in the pit bioreactor was sufficient to neutralise acidity and remove most metals over the first 90 days. After this time the bioreactor exhibited the capacity to continue reducing acidity and removing some trace elements to more than 882 days. Neutralisation of acidity and removal of most metals was also achieved for periods up to 183 days with a similar organic mix in the indrain bioreactor. This bioreactor also showed longterm capacity to reduce acidity and remove some metals continuing for more than 400 days (the limit of reporting). These results bench-mark the lower performance thresholds for simple bioreactors since any

increase in the proportion of non-cellulosic materials or alteration of construction techniques (such as milling and mixing of organic materials) will likely achieve much greater treatment capacity. Both bioreactors exhibited a pattern of initially high rates of alkalinity generation declining to a lower plateau level representing long-term rates. Initial rates of alkalinity generation in the pit bioreactor (6–12 g/ m2 CaCO3/day) and at various stages in the in-drain bioreactor (10–25 g/m2 CaCO3/day) were comparable with those in more than 5-year-old horizontal flow wetland bioreactors using much higher quality organic mixtures in milder climates (Hedin et al. 1994; Ziemkiewicz et al. 2003). However, long-term rates of alkalinity generation (after 1 year) in the bioreactors generally ranged between 1 and 3 g/m2 CaCO3/day and were much less than typical rates for systems elsewhere in the world (Hedin et al. 1994; Ziemkiewicz et al. 2003). Despite the low rates of alkalinity generation, both bioreactors showed the capacity to reduce drain water acidity and remove metals using low cost approaches that would otherwise only be possible with more expensive, less attractive options. In practical terms, drains with multiples of similar in-drain bioreactors may ultimately require smaller treatment systems at the end-of-drains (Degens 2009). High rates of alkalinity generation were sustained for longer with small increases in the proportion of some non-cellulosic materials suggesting that manipulating these may achieve greater rates over the long term. Increasing the proportion of starch-based noncellulosic material (waste wheat grain) to 1% of the in-drain bioreactor volume-enabled rates of alkalinity generation of >10 g/m2 CaCO3/day to be sustained

814

Water Air Soil Pollut (2012) 223:801–818

for double the time (190 days) than was achieved when the volume of non-cellulosic composted sheep manure was increased from 1% to 5% (phase 2). This suggests that the relative benefits of non-cellulosic materials in mixtures are not equal. Sustained treatment is increasingly considered to be due to the cellulosic component of organic mixtures, with synergistic effects commonly achieved by mixing with non-cellulosic materials (Neculita and Zagury 2008; Waybrant et al. 1998). This effect is attributed to non-cellulosic materials supporting a microbial population that enhances decomposition of cellulosic materials and increases carbon availability for sulfate reduction (Logan et al. 2005). The present experiment indicated that increasing the proportion of waste wheat grain as the non-cellulosic component in the bioreactor mixes by tenfold (to at least 10% by volume) will achieve far greater improvements in long-term rates of alkalinity generation than similar increases in composted sheep manure. 4.2 Alkalinity Generation Mechanisms Sulfate reduction was evident during periods of peak acidity reduction in both bioreactors indicating that this was a significant mechanism of alkalinity generation. Evidence of sulfate reduction during these times included visual evidence of sulfide precipitates in the bioreactors, reductions in sulfate concentrations of outflow waters observed as decreases in SO4/Cl relative to inflow and the redox potential (Eh) approaching −200 mV (after Faulwetter et al. 2009). The large decreases in SO4/Cl ratios (relative to source groundwater) in outflow waters from the indrain bioreactor during the early part of phases 2 and 3 were consistent with sulfate reduction and corresponded with the observed magnitude of decreases in acidity. Divergence of SO4/Cl from 0.12 (the average of inflow groundwater) to less than 0.10 during phases 2 and 3 could account for generation of alkalinity in the order of at least 500–600 mg CaCO3/L assuming all removal was due to sulfate reduction (with the range depending on outflow Clvarying 23–28 g/L, corresponding with TDS varying 40–50 g/L). This was based on the common equation for microbial sulfate reduction according to Eq. 5 (Hedin et al. 1994; Neculita et al. 2007).  2CH2 O þ SO2 4 ! H2 S þ 2HCO3

ð5Þ

where reduction of 1 mol of sulfate generates 2 mol of bicarbonate that has an equivalent alkalinity of 1 mol of CaCO3. Similarly, in the pit bioreactor, initial decreases in SO4/Cl from more than 0.16 in inflows to less than 0.15 in outflows could account for a reduction in acidity in outflows in the order of 300–350 mg CaCO3/L (with outflow Cl− ranging 28–33 g/L). Given inflows contained net acidity of at least 500 mg CaCO3/L, neutralisation was occurring as depletion of stored alkalinity that had probably accumulated during the pre-incubation prior to operation. After periods of peak alkalinity generation in both bioreactors, there was limited evidence of sulfate reduction. However, biological generation of alkalinity was evident as marginal increases in pH and decreases in acidity sustained in the outflows of both bioreactors over hundreds of days. In the in-drain bioreactor, this was supported by decreases in the outflow SO4/Cl of at least 0.01 (most evident toward the end of phases 2 and 3), which were consistent with sulfate reduction that would result in reduction of acidity by at least 200 mg CaCO3/L (based on Cl− concentrations at the time). Similar decreases in outflow SO4/Cl were less distinguishable in the pit bioreactor and intermittently evident in only winter– spring periods. The much greater increase in salt concentrations (and therefore background sulfate concentrations) as water flowed through the pit bioreactor limited the sensitivity of SO 4/Cl to decreases in sulfate concentrations. Furthermore, rainfall-dissolution of gypsum in the pit bioreactor contributed to periodic confounding variations in SO4/Cl. Greater quantification and clarity of the contribution of sulfate reduction to acidity treatment in both bioreactors is likely to be achieved by analysis of changes in stable sulfur isotope ratios (Waybrant et al. 2002). 4.3 Metal Retention Retention of Al was evident in the bioreactors, but not Fe or Mn. Average concentrations of Al in acidic waters were commonly reduced to near or below detection limits when outflow pH exceeded 5.5 with evidence that this continued to occur when outflow pH fell less than 5.5. These effects occurred despite evidence of evapo-concentration of waters as these passed through the bioreactors. Aluminium retention

Water Air Soil Pollut (2012) 223:801–818

in bioreactors and composting wetlands is generally as precipitated hydroxide or hydroxysulfate minerals mediated by raised pH (Gusek and Wildeman 2002; Neculita et al. 2007). In this experiment, PHREEQC predicted that hydroxysulfate minerals such as alunite (KAl 3 (SO 4 ) 2 (OH) 6 ), basaluminite (Al 4 (SO 4 ) (OH)10.5H20) and to a lesser extent a jurbanite-like mineral (AlSO4OH) as well as hydroxides such as gibbsite (Al(OH)3) had formed in the bioreactors when pH<5.5. These minerals consistently exhibited positive saturation indices for outflow waters at pH 4– 5.5. With no evidence that K/Cl depletion occurred in the bioreactors, indicating minimal removal of K, it was unlikely that alunite precipitation occurred indicating that most removal probably occurred as either basaluminite or jurbanite-like minerals. The bioreactors did not reduce Fe concentrations when pH<5.5 probably because of recycling of iron oxyhydroxides within the bioreactors (Johnson and Hallberg 2005). However, this only indicates that net retention of iron is poor in the bioreactors and does not exclude active Fe precipitation occurring in oxidised surface layers, particularly as hydroxysulfates or oxyhydroxides (Caraballo et al. 2010). Mineral saturation modelling for outflow waters of both bioreactors using PHREEQC indicated that at pH<5.5 precipitation of iron oxyhydroxides similar to akaganéite (Fe(OH)2.7Cl0.3) were likely, but not hydroxysulfate minerals such as schwertmannite (Fe8O8(OH)4.8(SO4)1.6). However, localised microbial oxidation of Fe2+ to Fe3+ observed as precipitate accumulation within surface layers of the bioreactor indicates that localised formation of hydroxysulfate minerals such as schwertmannite cannot be ruled out. The absence of reductions in Mn concentrations in bioreactor outflows is consistent with previous work (Santini et al. 2010) and the widely acknowledged limitations of Mn retention by sulfides or carbonate precipitation in treatment bioreactors (Neculita et al. 2007) and treatment wetlands (Younger et al. 2002). High removal of trace elements by the bioreactors was achieved for a range of elements (e.g. Ce, Co, La, Ni, Co and Pb) while outflow pH exceeded 5.5 that continued for extended periods for some elements (e. g. Ce, Cu, Ni and Pb) when pH decreased below 5.5. As with Al, these reductions occurred despite evaporation likely increasing the concentrations of metals in water flowing through the bioreactors. The reduction in divalent metal concentrations (e.g. Cu, Ni, Pb

815

and Zn) when pH>5.5 corresponded with times when there was strong evidence of sulfate reduction occurring. Metal retention in the bioreactors was therefore probably mediated by sulfide precipitation. This mechanism tends to dominate anaerobic treatment wetlands and bioreactors with generally limited contribution often attributed to adsorption by organic materials after start-up (Machemer and Wildeman 1992; Neculita et al. 2007). However, adsorption or complexing of metals cannot be discounted in high hydraulic residence time systems operating over longer periods of time. Reductions in metal concentrations of water flowing through the bioreactors when pH<5.5 corresponded with limited evidence of sulfate reduction, therefore metal retention mechanisms were unlikely to have been dominated by sulfide precipitation. The mechanisms of retention of the rare earth element such as La and Ce appear to be highly sensitive to pH with retention occurring only when pH>5.5. Possible mechanisms include pH-dependent absorption on residual iron oxides (Verplank et al. 2004) or organic matter (Tang and Johannesson 2003). Precipitation of the rare earths as phosphates seems unlikely given that high concentrations of La and Ce persisted despite detectable concentrations of phosphate (>0.05 mg/L) in outflow waters when pH<5.5. Retention of Pb, Cu and Ni when pH<15.5 was most likely due to adsorption to precipitated aluminium or iron minerals. Adsorption of metals onto oxyhydroxide minerals has been reported in sulfate-reducing bioreactors at pH>6 (Neculita et al. 2007; 2008), but there are no reports that these function in bioreactors where pH falls below 5.5. As discussed previously, the high removal of Al from the acidic waters suggests that large amounts of Al-dominated minerals have precipitated over time, most likely as hydroxysulfates. Similarly, Fe may have been recycled and precipitated as an iron oxyhydroxide similar to akaganéite and possibly hydroxysulfate minerals such as schwertmannite. These minerals can adsorb Pb, Cu, Zn and Ni (in that general order) from solutions at pH 4–5 with the process enhanced by the formation of sulfate complexes (Munk et al. 2002; Sánchez-España et al. 2006). However, speciation modelling of Cu and Pb for outflow waters from the bioreactors using PHREEQC indicated that sulfate complexes may be less important in the saline waters or form only at the surface of the minerals. While Cu was predicted to be

816

predominantly present as either Cu2+ followed by Cu– SO4 ion complexes, Pb was present as Pb–Cl ion complexes (e.g. PbCl3−, PbCl+, PbCl42−). Formation of adsorption complexes with organic matter may have been an additional mechanism contributing to metal retention, but may be limited under the high Cl− concentrations in the bioreactors treating saline acidic water. Neculita et al. (2008) reported that formation of metal-organic matter complexes for Cd, Ni and Zn were an important long-term retention mechanism in a bioreactor. However, it was not clear whether these mechanisms were likely to occur when through flow pH is less than 5.5. Furthermore, formation of metal-chloride complexes in the saline waters in the present study may have limited adsorption of divalent metals onto organic matter, as has been reported to occur for organic matter in estuarine water (Lores and Pennock 1998). The significance of Al mineral precipitation in sulfate-reducing bioreactors for trace metal retention has received little attention in previous investigations (e.g. Johnson and Hallberg 2005; Machemer and Wildeman 1992; Neculita et al. 2008; Waybrant et al. 2002). This has mostly arisen because of the focus on sulfide precipitation mechanisms with little attention to aluminium-rich acidic waters. However, in this and other studies (e.g. Gusek and Wildeman 2002; Caraballo et al. 2010) precipitation of Al minerals can be significant in sulfate-reducing systems with feed-waters containing Al. Under conditions where biological alkalinity generation can achieve sustained lowering of the pH of acidic aluminium-rich waters, the formation of such precipitates may provide a significant mechanism for metal retention that has previously been overlooked. 4.4 Nutrient Losses The bioreactors significantly increased concentrations of nutrients in the drain waters that have been overlooked in most previous investigations. While high concentrations of DOC, NH4+ and in some cases PO43− have been reported in some effluents from sulfatereducing bioreactors (Neculita and Zagury 2008; Waybrant et al. 2002), the implications for discharge to surface-water environments have been given little consideration. Concentrations of PO43− and DOC in the present investigation were increased more than an order of magnitude after passage through the bioreac-

Water Air Soil Pollut (2012) 223:801–818

tors and may stimulate excessive bacterial and algal growth in receiving environments. Selection of organic materials with low available P content would reduce the risk of high phosphate concentrations in outflows, although total P content in organic materials can poorly correspond with available P content (Waybrant et al. 1998). Discharge of Fe from the bioreactors may mitigate loss of PO43− if oxidised in oxidation ponds and retained prior to discharge since iron precipitates can be significant in adsorbing PO43− from neutralised waters (Simmons 2010). 4.5 Climatic Influence on Bioreactor Function The function of the bioreactors in the semi-arid climate did not benefit from elevated temperatures or large hydraulic residence times induced by high seasonal evaporation rates. Outflows immediately following periods of ceased outflow (with estimated HRT>100 days) did not exhibit discernibly increased pH or decreased net acidity compared with subsequent outflows. This indicated that treatment was more dependent on hydraulics of water movement and less dependent on residence time, where greater treatment generally occurs with greater residence time (Faulwetter et al. 2009; Younger et al. 2002). The high residence times also contributed to degradation of water quality because of evaporation-induced increases in the salinity of waters. This problem is not encountered in temperate climates where evaporation is low and rainfall dilution can improve water quality (Hedin et al. 1994). In design terms, this can be overcome by reducing the area exposed to evaporation by lowering the water level within the organic bed or covering the bioreactor. Seasonal variation in temperature or salinity appeared to have little effect on rates of alkalinity generation by the bioreactors. Although warmer temperatures and increased sulfate concentrations can stimulate rates of sulfate reduction, this is dependent on available carbon substrates being nonlimiting (Faulwetter et al. 2009; Nedwell and Abram 1979). These conditions were probably not met in the present investigation. In the pit bioreactor, rates of alkalinity generation appeared to decrease with increasing temperature and salinity towards the end of the first spring and thereafter responded little to seasonal variations in temperature from <15°C to

Water Air Soil Pollut (2012) 223:801–818

>30°C or salinity from <70 to >120 g/L. The effects of slowed through flow and greater residence time caused by high evaporation rates may also have masked any effects of temperature and salinity on alkalinity generation rates. Similarly, alkalinity generation rates in the in-drain bioreactor tended to respond to additions of organic matter with gradual declines thereafter. Variations in water temperature and salinity during this time did not appear to contribute to varying rates of alkalinity generation. Periods of increasing water salinity and temperature had no clear stimulating effect when rates were low (e.g. days 100– 200 or 500–600) and did not counter declining trends in alkalinity generation rates (e.g. days 900–1,000).

5 Conclusions Evaluation of two bioreactors over several years found that biological alkalinity generation could achieve short-term neutralisation and removal of most trace elements from saline acidic drain waters with longer-term reduction in metal concentrations sustained under subsequent lower rates of alkalinity generation. Metal retention when pH>5.5 was consistent with sulfide precipitation supported by evidence of sulfate reduction activity in the bioreactors. However, when bioreactor pH fell to less than 5.5, metal retention mechanisms were likely associated with formation of aluminium and iron hydroxysulfate precipitates. This indicated that even when biological alkalinity reduction is not sufficient to neutralise drain water, metal removal can still be achieved by other non-sulfide mechanisms. This experiment shows that there were no benefits afforded by environmental conditions to bioreactor function in semi-arid climates. Higher hydraulic residence times induced by high evaporation rates and greater temperatures hindered bioreactor function with no benefit to water treatment under low rates of alkalinity generation. The climatic conditions also had the effect of reducing the quality of water discharging from the bioreactors by increasing salt concentrations. Acknowledgements The author would like to thank Clare Wright, Adrian Beech and Julie Smith at the CSIRO laboratories in Adelaide for water analyses; Peter Geste and David Rowlands for technical assistance in sampling and construction; Harold Beagley and Ashley Bonser for providing their land and machinery; Glenice Batchelor for

817 negotiating support with local communities and Wheatbelt Natural Resource Management Inc. (formerly Avon Catchment Council Inc.) for funding this project through Project 04A1-04 (IWM 005). The helpful comments of anonymous reviewers were also greatly appreciated.

References Bureau of Meteorology (2010). Climate data online for Station 010040 (Doodlakine) and Beacon (010004). http://www. bom.gov.au/climate/data/weather-data.shtml. Accessed June 2010. Caraballo, M. A., Santofimia, E., & Jarvis, A. P. (2010). Metal retention, mineralogy, and design considerations of a mature permeable reactive barrier (PRB) for acidic mine water drainage in Northumberland, U.K. American Mineralogist, 95, 1642–1649. Degens, B. (2009). Proposed guidelines for treating acidic drain water in the Avon catchment: adapting acid mine drainage treatment systems for saline acidic drains, Western Australia. Perth, Western Australia: Department of Water, Salinity and Land Use Impacts Series, SLUI 54. Degens, B., & Shand, P. (2010). Assessment of acidic saline groundwater hazard in the Western Australian Wheatbelt: Yarra Yarra, Blackwood and South Coast. Adelaide: CSIRO: Water for a Healthy Country National Research Flagship. Faulwetter, J. L., Gagnon, V., Sundberg, C., Chazarenc, F., Burr, M. D., Brisson, J., et al. (2009). Microbial processes influencing performance of treatment wetlands: A review. Ecological Engineering, 35, 987–1004. Gusek, J. J., & Wildeman, T. R. (2002). Passive treatment of aluminium-bearing acid rock drainage. In Proceedings of the 23rd West Virginia surface mine drainage taskforce symposium. Morgantown, West Virginia: West Virginia Surface Mine Drainage Taskforce. Hatton, T. J., Ruprecht, J., & George, R. J. (2003). Preclearing hydrology of the Western Australian wheatbelt: Target for the future. Plant and Soil, 257, 341–356. Hedin, R. S., Nairn, R. W., & Kleinmann, R. L. P. (1994). Passive treatment of coal mine drainage. Pittsburgh, PA: United States Department of the Interior, Bureau of Mines Information Circular 9389. Johnson, D. B., & Hallberg, K. B. (2005). Biogeochemistry of the compost bioreactor components of a composite acid mine drainage passive remediation system. Science of the Total Environment, 338, 81–93. Kerkar, S., & LokaBharathi, P. A. (2007). Stimulation of sulphate reducing activity at salt-saturation in the salterns of Ribandar, Goa. Indian Geomicrobiology Journal, 24, 101–110. Kirby, C. S., & Cravotta, C. A. (2005). Net alkalinity and net acidity 1: Theoretical considerations. Applied Geochemistry, 20, 1920–1940. Logan, M. V., Reardon, K. F., Figueroa, L. A., McLain, J. E. T., & Ahmann, D. M. (2005). Microbial community activities during establishment, performance and decline of benchscale passive treatment systems for mine drainage. Water Research, 39, 4537–4551.

818 Lores, E. M., & Pennock, J. R. (1998). The effect of salinity on binding of Cd, Cr, Cu and Zn to dissolved organic matter. Chemosphere, 37, 861–874. Machemer, S. D., & Wildeman, T. R. (1992). Adsorption compared with sulfide precipitation as metal removal processes from acid mine drainage in a constructed wetland. Journal of Contaminant Hydrology, 9, 115–131. Munk, L., Faure, G., Pride, D. E., & Bigham, J. M. (2002). Sorption of trace metals to an aluminum precipitate in a stream receiving acid rock-drainage; Snake River, Summit County, Colorado. Applied Geochemistry, 17, 421–430. Neculita, C. M., & Zagury, G. J. (2008). Biological treatment of highly contaminated acid mine drainage in batch reactors: Long-term treatment and reactive mixture characterisation. Journal of Hazardous Materials, 157, 358–366. Neculita, C. M., Zagury, G. J., & Bussiere, B. (2007). Passive treatment of acid mine drainage in bioreactors using sulfate-reducing bacteria: Critical review and research needs. Journal of Environmental Quality, 36, 1–16. Neculita, C. M., Zagury, G. J., & Bussière, B. (2008). Effectiveness of sulphate-reducing passive bioreactors for treatment of highly contaminated acid mine drainage: II. Metal removal mechanisms and potential mobility. Applied Geochemistry, 23, 3545–3560. Nedwell, D. B., & Abram, J. W. (1979). Relative importance of temperature and electron donor and electron acceptor concentrations on bacterial sulfate reduction in salt marsh sediment. Microbial Ecololgy, 5, 67–72. Parkhurst, D. L., & Appelo, C. A. J. (1999). User’s Guide to PHREEQC (Version 2)—A computer program for speciation, batch-reaction, one-dimensional transport, and inverse geochemical calculations. Denver: United States Geological Survey. Water Resources investigations report 99–4259 Sánchez-España, J., López-Pamo, E., Santofimia-Pastor, E., Reyes-Andrés, J., & Martín-Rubí, J. A. (2006). The removal of dissolved metals by hydroxysulphate precipitates during oxidation and neutralisation of acid mine waters, Iberian Pyrite Belt. Aquatic Geochemistry, 12, 269–298. Santini, T., Degens, B. P., & Rate, A. W. (2010). Organic substrates in bioremediation of acidic saline drainage

Water Air Soil Pollut (2012) 223:801–818 waters by sulfate-reducing bacteria. Water, Air, and Soil Pollution, 209, 251–268. Shand, P., & Degens, B. (2008). Avon catchment acid groundwater: Geochemical risk assessment. Perth, Western Australia: Cooperative Research Centre for Landscape Environments and Mineral Exploration, Open File Report 191. Simmons, J. A. (2010). Phosphorus removal by sediment in streams contaminated with acid mine drainage. Water, Air, and Soil Pollution, 209, 123–132. Stewart, B., Strehlow, K., & Davis, J. (2009). Impacts of deep open drains on water quality and biodiversity of receiving waterways in the Wheatbelt of Western Australia. Hydrobiologia, 619, 103–118. Tang, J., & Johannesson, K. H. (2003). Speciation of rare earth elements in natural terrestrial waters: Assessing the role of dissolved organic matter from the modelling approach. Geochimica et Cosmochimica Acta, 67, 2321–2339. Tyrell, W. R., Mulligan, D. R., Sly, L. I., & Bell, L. C. (1997). Trialing wetlands to treat coal mining wastes in a low rainfall, high evaporation environment. Water Science and Technology, 35, 293–299. Verplanck, P. L., Nordstrom, D. K., Taylor, H. E., & Kimball, B. A. (2004). Rare earth element partitioning between hydrous ferric oxides and acid mine water during iron oxidation. Applied Geochemistry, 19, 1339–1354. Waybrant, K. R., Blowes, D. W., & Ptacek, C. J. (1998). Selection of reactive mixtures for use in permeable reactive walls for treatment of acid mine drainage. Environmental Science and Technology, 32, 1972–1979. Waybrant, K. R., Ptacek, C. J., & Blowes, D. W. (2002). Treatment of mine drainage using permeable reactive barriers: Column experiments. Environmental Science and Technology, 36, 1349–1356. Younger, P. L., Banwart, S. A., & Hedin, R. S. (2002). Mine water: Hydrology, pollution, remediation. London: Kluwer Academic Publishers. Ziemkiewicz, P. F., Skousen, J. G., & Simmons, J. (2003). Long-term performance of passive acid mine drainage treatment systems. Mine Water and the Environment, 22, 118–123.