Environmental Pollution 284 (2021) 117318

Contents lists available at ScienceDirect

Environmental Pollution
journal homepage: www.elsevier.com/locate/envpol

Saline mine-water alters the structure and function of prokaryote

communities in shallow groundwater below a tropical stream☆
Lisa Chandler a, b, *, Andrew J. Harford b, Grant C. Hose c, Chris L. Humphrey b,
Anthony Chariton c, Paul Greenfield d, Jenny Davis a
a
Research Institute for the Environment and Livelihoods, College of Engineering, IT & Environment, Charles Darwin University, Darwin, Northern Territory, Australia
b
Supervising Scientist Branch, Department of Agriculture, Water and the Environment, Darwin, Northern Territory, Australia
c
Department of Biological Sciences, Macquarie University, Sydney, New South Wales, Australia
d
Commonwealth Scientific and Industrial Research Organisation (CSIRO), Canberra, Australian Capital Territory, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: Bacteria and archaea (prokaryotes) are vital components for maintaining healthy function of groundwater
Freshwaters ecosystems. The prokaryotic community composition and associated putative functional processes were exam­
Low ionic strength ined in a shallow sandy aquifer in a wet-dry tropical environment. The aquifer had a contaminated gradient of
Hyporheic
saline mine-water, which primarily consisted of elevated magnesium (Mg2+) and sulfate (SO2− 4 ), although other
Metagenomics/community genomics
Microbial ecology
major ions and trace metals were also present. Groundwaters were sampled from piezometers, approximately 2
m in depth, located in the creek channel upstream and downstream of the mine-water influence. Sampling
occurred during the dry-season when only subsurface water flow was present. Next generation sequencing was
used to analyse the prokaryote assemblages using 16S rDNA and metabolic functions were predicted with
FAPROTAX. Significant changes in community composition and functional processes were observed with
exposure to mine-waters. Communities in the exposed sites had significantly lower relative abundance of
methanotrophs such as Methylococcaceae and methanogens (Methanobacteriaceae), but higher abundance in
Nitrososphaeraceae, associated with nitrification, indicating potentially important changes in the biogeochem­
istry of the exposed sites. The changes were most strongly correlated with concentrations of SO2−
4 , Mg
2+
and Na+.
This knowledge allows an assessment of the risk of mine-water contamination to groundwater ecosystem
function and aids mine-water management.

1. Introduction biodiversity, thus altering various biogeochemical cycling processes

(Hemme et al., 2015). The response of groundwater prokaryote com­
Groundwater is a valuable global resource that is under increasing munities to contamination has been the focus of a number of studies (e.g.
pressure from disturbances such as extraction and contamination hydrocarbons, (Grösbacher et al., 2016; Mouser et al., 2005; Stephenson
(Griebler et al., 2019). Groundwaters in uncontaminated aquifers are et al., 2013; Wright et al., 2017); uranium and metals, (He et al., 2018;
often oligotrophic, with little or no dissolved oxygen, limited carbon, Hemme et al., 2010, 2015); general agricultural/industrial pollutants
nutrients and energy resources, and the prokaryote communities in (Korbel et al., 2013; Smith et al., 2012, 2018); and salinity (Sang et al.,
these systems are generally dependent on heterotrophy and lithoauto­ 2018)). Changes in physico-chemical conditions result in changes in
trophy (Goldscheider et al., 2006; Taubert et al., 2019). Prokaryotes are community composition, e.g. Stephenson et al. (2013) and richness, e.g.
involved in various biogeochemical cycling processes in groundwater Hemme et al. (2015), as well as decreases in functional diversity, e.g. He
ecosystems, particularly playing an important role in carbon and et al. (2018). Understanding the ecology of groundwater prokaryote
nutrient cycling in groundwaters (Griebler and Leuders, 2009). communities and the impacts of groundwater contamination on these
Groundwater contamination not only affects the quality of groundwater communities is critical for management and protection of all ground­
directly but causes changes in prokaryote communities, including loss of water values.

☆
This paper has been recommended for acceptance by Charles Wong.
* Corresponding author. Supervising Scientist Branch, Department of Agriculture, Water and the Environment, Darwin, Northern Territory, Australia.
E-mail addresses: [email protected], [email protected] (L. Chandler).

https://doi.org/10.1016/j.envpol.2021.117318
Received 24 November 2020; Received in revised form 30 April 2021; Accepted 3 May 2021
Available online 12 May 2021
0269-7491/Crown Copyright © 2021 Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
L. Chandler et al. Environmental Pollution 284 (2021) 117318

The discharge of saline mine-waters has significantly impacted the coarse sands (Roberts, 1991). The average thickness of the sands (i.e.
quality of many surface- and groundwater ecosystems around the globe depth to original stream bed) has been reported as 5–12 m in sections
(Lottermoser, 2010), resulting in the modification of hydrogeochemistry (Ahmad et al., 1982; Nanson et al., 1993; Roberts, 1991). The sand
that can impact on biological communities (Cañedo-Argüelles et al., channel has been described as a strip, unconfined aquifer and has high
2016). Saline mine-waters arise through the dissolution and mobi­ transmissivities, ranging between 400 and 800 m2 day− 1 (Ahmad and
lisation of metals, non-metals and metalloids, and can, depending on the Green, 1986). To access the groundwater during the dry season, PVC
solutes, range from acidic to alkaline (Nordstrom et al., 2015). Knowl­ piezometers (Enviroequip 50 mm, 1.5 m PVC Bore casings and screens
edge of salinity effects on prokaryote communities in groundwater (Thermofisher Scientific)) were installed to a depth of approximately 2
(often measured as total dissolved solids (TDS) or electrical conductivity m within the sand channel (Supplementary Figure 2). A series of pie­
(EC)), has mostly been derived from studies of communities along gra­ zometers were installed both laterally and longitudinally along the creek
dients in coastal or estuarine systems, e.g. Héry et al. (2014) and Sang from approximately 3 km upstream of Ranger uranium mine to
et al. (2018). These studies observed changes in bacterial community approximately 4 km downstream (Supplementary Figure 1). A 300 m
composition associated with changes in salinity due to seawater intru­ longitudinal section of the shallow groundwaters in Magela Creek is
sion, for example Héry et al. (2014) noted increases in relative abun­ currently contaminated with mine-waters, most likely arising from
dance of Gammaproteobacteria associated with increased salinity. groundwater moving through contaminated sediments from the adja­
However, sodium chloride (NaCl), the key constituent of seawater, is cent Coonjimba Billabong. This mine-water source typically contains
rarely a major component of mine-waters (Nordstrom et al., 2015), and high concentrations of Mg2+, SO42− (the dominant contamination
input of higher salinity waters with different ionic composition can have signature in the mine-waters) along with elevated levels of Mn, U and
different effects on freshwater biota (Cañedo-Argüelles et al., 2016). ammonia (Supplementary Table 1) (Harford et al., 2015; Hart et al.,
To date, little is known of the groundwater prokaryote community in 1992; Noller, 1991; Trenfield et al., 2019).
seasonally-flowing sandy creek channels that are common in the wet-dry
tropics. Moreover, there is little understanding of the community 2.2. Sample collection
response to salinisation and/or other associated contaminants of po­
tential concern (COPCs) from anthropogenic sources such as mine- Samples of groundwater for water chemistry and molecular analysis
waters. The impacts of a uranium mine (Ranger) located in the Austra­ were collected monthly from 17 sites, during the dry season (August to
lian wet-dry tropics, on downstream surface water communities of November), on 3 occasions in the first sampling year (2017) and 4 oc­
Magela Creek, have been well documented (Humphrey and Chandler, casions in the second year (2018) (Supplementary Table 2). A peristaltic
2018; McCullough, 2006; Mooney et al., 2020; van Dam et al., 2010), pump (Series II, Geotech Environmental Equipment, Inc), was used to
but the impacts on groundwater communities have not been studied. purge piezometers prior to sample collection by pumping at least three
Although the mine has ceased production (in January 2021), and the site casing volumes (Sundaram et al., 2009). In-situ measurements of
is undergoing rehabilitation (to 2026), the site will continue to be a physico-chemical variables (Supplementary Table 3) were taken using a
source of salts and other contaminants of potential concern (COPCs) multiparameter sonde and flow-through cell (YSI EXO1, Ohio USA).
through surface water runoff and exfiltrating groundwater. The major Once purged, and the water quality parameter readings had stabilised,
contaminants are magnesium (Mg2+) and sulfate (SO2− 4 ), derived from samples for water chemistry and molecular analysis were collected.
the waste-rock landform and pit capping. The broad aim of this study Samples for water chemistry were collected in 1 L HDPE bottles,
was to examine changes in the structure and function of prokaryote ensuring that there was no head space in the bottle. Water samples, for
communities in the shallow groundwater of Magela Creek associated molecular analysis, were collected in sterile 2 L plastic bottles and both
with inputs of contaminated mine-waters, with two specific objectives: were stored on ice until they were processed back in the laboratory.
(1) to determine whether mine-water contaminants (specifically Mg2+
and SO2−4 ) altered the composition of the prokaryote community; and 2.3. Chemical analysis of groundwater samples
(2) to assess the implications of any community composition changes on
ecosystem function. Prokaryote communities in the groundwater were A portion of the 1 L water sample was filtered within 4 h of collection
characterised using next generation sequencing technology. Commu­ using a 150 mL Nalgene™ Rapid-Flow™ Sterile Single Use Vacuum
nities were viewed in terms of taxonomic and functional composition, Filter unit (0.45 μm), for analysis of filtered metals and major ions. The
and changes to assemblages were related to water quality associated remaining sample was decanted into separate bottles for analysis of
with mine-water discharge. alkalinity, nutrients and dissolved organic carbon (Supplementary
Table 4). Samples were analysed were analysed for a suite of metals and
2. Methods major ions (Supplementary Table 3) by inductively coupled plasma mass
spectrometry, inductively coupled plasma atomic emission spectrometry
2.1. Site description and sampling and FIA methods, by a NATA-registered laboratory (EnviroLab, Sydney,
Australia). Total and dissolved organic carbon (TOC/DOC) samples
The study was conducted in a temporary sand-bed stream arising in were analysed in-house using a high-temperature combustion method
Arnhem Land, with the mid-lower reaches flowing through Kakadu 5310B (TOC-VCSH, Shimadzu Scientific Instruments, Oceania Pty. Ltd,
National Park (KNP), Australia (Supplementary Figure 1). The study SD < 0.1 mg L− 1, maximum CV of 2%).
sites were located along a 7 km reach of Magela Creek in the Ranger
Project Area (RPA i.e. in a mine lease excluded from KNP). The tradi­ 2.4. DNA extraction and amplification
tional owners of land on which the study was conducted are the Mirrar
people and they rely on the downstream water resources for sustenance. Water samples were filtered within 4 h of collection, through sterile
The region is characterised by a wet-dry tropical climate with clearly 0.22 μm mixed cellulose membranes using vacuum filtration, 100 mL
defined summer wet and winter dry seasons, and a mean annual rainfall increments were added until a total of 1000 mL had been filtered. The
of 1550 mm. The wet season prior to the 2018 sampling period had a filter membranes were placed in sterile petri dishes and stored at − 80 ◦ C
higher average mean rainfall (1919 mm) than the wet season preceding until DNA extraction. For DNA extraction, membranes were sliced into
the 2017 sampling period (1701 mm). During the dry season, surface- thin strips, using a maximum of 75% of each membrane. The weight of
flow in the stream ceases for several months (July–December) but per­ sediment and filter membrane was recorded, as the recommended 0.25
manent groundwater flows through sand channel throughout the year. mg of sediment was generally not collected on the filters. Extractions
The sand channel has been filled since the Holocene with medium- were performed using the Qiagen DNeasy Powersoil kit, following the

2
L. Chandler et al. Environmental Pollution 284 (2021) 117318

manufacturers protocol with the following modifications; increase in plume (Upstream Reference), sites laterally adjacent to contamination
time and intensity for bead beating. 30 min incubation period, 4 min plume, in the central channel of the creek (Central Reference), sites
centrifuge period, and only 50 μL of the final C6 solution was used to downstream of contamination plume (Downstream Reference), and sites
concentrate the final volume of DNA extracted. within the plume (Exposed).
Polymerase chain reaction (PCR) was conducted on all samples using Patterns amongst the environmental characteristics of the shallow
the primer set 515F (50–3’: GTGYCAGCMGCCGCGGTAA) (Parada et al., groundwater samples were examined with Principal Components
2016) and 806R (50–3’: GGACTACNVGGGTWTCTAAT) (Apprill et al., Analysis (PCA), based on Euclidean distances using Primer v7 (Clarke
2015) for the V4 region of the 16S rRNA gene. Amplification used a and Gorley, 2015). Data were visualised using draftsman plots, and a
modified Earth Microbiome Project protocol v4.13 (Caporasa et al., log-transformation applied if the data were strongly skewed, before
2012). The sample (2 μL) was added to each reaction, with 15 μL of being normalised to account for different measurement scales.
AmpliTaq Gold 360 Master Mix (Applied Biosystems), 10 μL of ultrapure Differences in prokaryote taxonomic and functional communities
water (UltraPure™ DNase/RNase-Free Distilled Water, Invitrogen) and and environmental variables were compared amongst exposure groups
3 μL of a unique combination of forward and reverse tag primers (2 μM) using the Kruskal-Wallis test.
for total PCR reaction volume of 30 μL. A sample specific Golay barcode Multivariate statistical analyses were conducted on Hellinger trans­
was included in the 806R reverse primer. Amplification included both a formed OTU data, and square root transformed functional group data
positive control (Supplementary Figure 3) and negative controls using Primer v7 (Clarke and Gorley, 2015). Non-metric multidimen­
(DNA-free water). Thermal cycling included a simplified hot start at sional scaling (nMDS) was used to visualise the community assemblages
95 ◦ C and an initial denature cycle at 95 ◦ C for 10 min followed by of Bray-Curtis resemblance matrices for both the OTU and functional
annealing for 35 cycles of 95 ◦ C for 30 s, 50 ◦ C for 30 s and 72 ◦ C for 1 group datasets. PERMANOVA was used to test whether community
min, followed by elongation at 72 ◦ C for 7 min and a final hold at 4 ◦ C. composition and function differed between year and a priori “exposure
Amplicon quality and quantity was determined by Quant-it Picogreen groups”. Sample sites were included as a random factor nested in
assay (Thermofisher Scientific) and agarose gel electrophoresis. PCR exposure groups. Differences in data dispersion between exposure
products of all samples were pooled in equimolar amounts before being groups, sites and years were tested with PermDISP and the association
purified, using AMPure XP kit (Beckman Coulter, cat#A63880). The between environmental variables and prokaryote community composi­
final purified pool was sent to Ramaciotti Centre for Gene Function tion and function was determined using DISTLM and dbRDA. Highly
Analysis (Sydney, Australia) for paired-end sequencing on the Illumina correlated environmental variables (r > 0.90) were removed from the
MiSeq using the 250 bp v2.2 process. dataset prior to running DISTLM. DISTLM was used with stepwise se­
lection and Akaike information criterion (AIC). The significance level (α)
2.5. Bioinformatics for all inferential tests was 0.05.
A negative binomial model applied to the non-rarefied OTU data
A custom pipeline was used to process the amplicon sequence data with the DESeq2 implementation in phyloseq (McMurdie and Holmes,
(Greenfield Hybrid Amplicon Pipeline, GHAP, v2.4) as detailed in Gissi 2014) was used to test for differences between exposure groups. P-values
et al. (2019). Usearch v11.0.667 and RDP Classifier (v2.12) were used to were adjusted for multiple tests with the False Discovery Rate
classify OTU sequences supplemented with a curated reference set Benjamini-Hochberg method.
(RefSeq 16S, downloaded September 2019) for the purposes of
species-level classification. Complete with taxonomic classifications, 3. Results
species assignments and counts for each sample were generated in OTU
tables. 3.1. Overview of chemistry
After processing through the GHAP pipeline, and prior to statistical
analysis, data were further processed using the R library phyloseq The data set included exposed sites (n = 5) within a contamination
(McMurdie and Holmes, 2013). Rare OTUs that occurred in <2 samples plume on the mine-lease area, as well as reference sites that were up­
were excluded (10.2% of 37647), as were OTUs that were not classified stream (n = 6), downstream (n = 3) and in the central creek channel (n
to a domain (0.19%), and OTUs that matched Eukaryote organelles (i.e. = 3) adjacent to the exposed sites (Supplementary Figure 1). The natural
chloroplasts). OTUs that had reads less than 0.01% of total number of shallow groundwater, observed at the reference sites during the dry
sequences were also removed (95.2%) to obtain a total of 1606 different season, was generally acidic (pH 4.8–6.4), with low electrical conduc­
OTUs. Sequences were rarefied to the lowest common number of se­ tivity (EC < 50 μS cm− 1), alkalinity (as HCO−3 <10 mg L− 1) and con­
quences (7766 sequences/sample) to minimise batch effect due to dif­ centrations of major ions and metals (Supplementary Figure 4). Mine-
ferences in the numbers of total sequences present in the two sampling water exposed sites had significantly higher concentrations (Kruskal-
years. All subsequent analyses, except the DESeq2 analysis, were based Wallis, P < 0.001) of known mine-water contaminants, SO2− 2+
4 , Mg , and
on these rarefied sequences. Mn, significantly higher concentrations of Na+, Cl− and Fe (Supple­
mentary Figure 4) and lower concentrations of Ca2+. The exposed sites
2.6. Functional analysis also had generally negative redox values (cf. generally positive redox
measures for reference sites), and a slightly lower range of pH
Functional group prediction was performed using FAPROTAX (Louca (4.54–5.93) compared to reference sites. All sites, reference and
et al., 2016). Limitations to this approach include: 1) the FAPROTAX exposed, had low concentrations of dissolved oxygen (<1 mg L− 1),
database was constructed primarily to analyse functional profiles of indicating suboxic or anoxic conditions, and were high in temperature
marine and lentic prokaryote communities; and (2) the database is (27–32 ◦ C). The sites were typically low in nitrogen with NO−2 and NO−3
limited with the result that only a small percentage of OTUs in this study concentrations generally below the detection limit (<0.005 mg L− 1).
could be assigned to at least one functional group. A heatmap illus­ Dissolved organic carbon (DOC) was highly variable (0.84–998 mg L− 1),
trating the FAPROTAX was generated using the heatmap() function in R. with a median value of 15 mg L− 1.
Principal Components Analysis (PCA) showed distinct separation of
2.7. Statistical analyses the environmental characteristics of exposed sites and reference sites
(Fig. 1). The first PCA axis explained 26.4% of the variation in site
Samples were classified a priori according to position in the study environmental variables, and the second axis explained 16%. Separation
reach around a contamination plume (Supplementary Figure 1). Four of samples along the first axis was positively correlated with Mg2+ and
exposure groups were defined: all sites upstream of contamination EC, along with Mn (Pearson correlation coefficient, r > 0.80). The

3
L. Chandler et al. Environmental Pollution 284 (2021) 117318

Fig. 1. Principal components analysis for samples

based on selected environmental data. To aid visual
interpretation, only vectors with correlations >0.5 are
displayed. Metals and major ions were log trans­
formed. DPW = distance to permanent surface water;
SWL = standing water level. Sites were classified a
priori based on contamination concentrations and the
location of the site within the study reach around the
contaminated (exposed) area (i.e. upstream of
contamination, downstream of contamination, and
adjacent to in the central channel where no contami­
nation was observed).

second axis separated samples, based on Ca2+ (r > 0.7). Mg concentra­ archaeal phyla with 12 genera across 4 classes. Most OTUs with Eur­
tion and EC have a strong linear relationship. yarchaeota were assigned to the genus Methanobacterium (37%).
A number of chemical variables had lower concentrations in 2018 The mine-water exposed sites had a significantly higher proportion
compared to 2017 (Supplementary Figure 4) presumably due to the of unclassified Bacteria (Fig. 2), compared to reference sites. Methyl­
slightly higher antecedent rainfall in 2018 and hence dilution available ococcaceae, as well as other methylotrophic families were absent (or
from higher recessional flows. However, these differences were not had < 1% relative abundance) at the exposed sites, as was the Meth­
marked and the PCA is notable for the interspersion of 2017 and 2018 anobacteriaceae and Clostridiaceae 1 families. Helicobacteraceae were
data points within each of the exposure groups indicating relatively low absent or had <1% relative abundance at the upstream reference sites
interannual variability in water quality. but were present at both the downstream and central reference sites and
the exposed sites.
3.2. Taxonomic composition
3.3. Community composition
In the final dataset, 1606 unique OTUs were identified across 17
sites, with Bacteria representing 83% (1332 OTUs) and Archaea 17% Ordination plots of the prokaryote communities showed a distinct
(274 OTUs). OTUs were further assigned to lower taxonomic ranks and separation of communities between the exposed and the reference site
relative abundance for each rank was estimated across different groups. groups (Fig. 3 A), with PERMANOVA showing significant differences (p
Across all the sites a total of 31 prokaryote phyla were assigned, con­ < 0.01) amongst the exposure groups, sampling years and sites (Sup­
sisting of 66 classes, 107 orders, 165 families and 255 genera. Of the 255 plementary Table 5). Exposure groups, site and the residual were the
genera groups 43% were not classified into a known genus. A large largest contributors to variation, and year, although statistically signif­
portion of the reads (~27%) were unclassified at phylum level (24.9% icant (p = 0.01) was less important than the other factors (Supplemen­
for Bacteria and 2.5% for Archaea). Excluding the unclassified Bacteria, tary Table 5). Pairwise comparisons for exposure groups, indicated that
the dominant phyla groups across the study were Proteobacteria the exposed sites were significantly different from each reference group,
(35.6%), Acidobacteria (5.4%), Thaumarchaeota (5%), Firmicutes and reference groups did not significantly differ from each other. Data
(4.9%), Euryarchaeota (4.5%) and Crenarchaeota (4.2%) accounting for dispersion (PermDISP p > 0.05) did not significantly differ between
approximately 60% of the total reads. exposure groups or years (Supplementary Table 5), suggesting that the
The Bacteria were represented by 25 known phyla groups, and the results were not influenced by differences in heterogeneity between
predominant groups included Proteobacteria (43.9%), unclassified these groups. However, data dispersion differed significantly amongst
Bacteria phyla (30.7%), Acidobacteria (6.6%) and Firmicutes (6.1%). sites (PermDISP p = 0.016), indicating the PERMANOVA results were
Proteobacteria had the highest diversity at genus level (117 assigned also influenced by differences in dispersion (and Supplementary
genera in total), with Methylomonas having highest relative abundance Table 5), indicating spatial heterogeneity, which is also shown in a
(11%). Acidobacteria were represented by 13 classes with Geothrix the cluster analysis where samples from each site tended to cluster together
most abundant genus (27%). The phylum Firmicutes had OTUs divided (Supplementary Figure 6).
between 4 classes, with Clostridium sensu stricto the most abundant genus Examination of the association between community structure and
(25%). environmental variables showed the same strong separation between
Five archaeal phyla were represented, with most OTUs within the exposed and reference groups (Fig. 3 B). Samples from the exposed
Archaea assigned to the Thaumarchaeota (9%), Euryarchaeota (29%) group with higher contaminant concentrations were separated along the
and Crenarchaeota (16%). Nitrososphaera was the dominant genus first axis (to the right). Na+, SO2−
4 and Mg
2+
concentrations accounted
(99%) within Thaumarchaeota. Euryarchaeota was the most diverse for the largest portion of variation in the community structure, (12.4,

4
L. Chandler et al. Environmental Pollution 284 (2021) 117318

Fig. 2. The relative abundances (%) of dominant prokaryote families in the different exposure groups. Taxa that had <1% of total community abundance were
grouped into “Others”.

11.8 and 11.2% of the variation, respectively, Supplementary Table 6). response of individual sites between years (Site(Exp) x Year p = 0.001).
Differential abundance analysis indicated OTUs differed significantly Pairwise tests examining the exposure groups showed that each refer­
(adjusted p < 0.01) in relative abundance between exposed and refer­ ence group differed significantly from the exposed group. Examination
ence sites (Fig. 4). The reference sites had significantly more OTUs (and of associations between functional groups and environmental factors
in greater relative abundance) assigned to an unclassified Betaproteo­ (Fig. 6 B) indicated that Na+ and SO2−4 explained most (15 and 13%,
bacteria family group, Methanobacteriaceae, Methylococcaceae and respectively) of the variation observed in the functional composition,
Methylocystaceae compared to the exposed sites. Whereas the exposed followed by Mg2+ (10%) and Cl− (9%) (Supplementary Table 9).
sites had significantly more OTUs, and in greater relative abundance,
associated with Nitrososphaeraceae, an unclassified Thermoproteales 4. Discussion
family group, Rhodocyclaceae and an unclassified Campylobacterales
family group (Fig. 4). We found changes in groundwater prokaryote community assem­
blages and functional groups strongly correlated with contamination
arising from saline expressions of mine waste waters.
3.4. Prokaryote functional annotation
Samples from the exposed sites had significantly elevated concen­
trations of Mg2+, SO2− 4 , Na , Cl , Fe and Mn, compared to concentra­
+ −
FAPROTAX assigned 479 out 1606 bacterial OTUs (29.8%) to at least
tions in reference sites, adjacent to and upstream and downstream of the
one predicted functional group. A total of sixty-nine functional groups
contaminated area. Elevated concentrations of Mg2+ and SO2− 4 are
were identified and most groups were represented by more than one
indicative of mine-derived water contaminants. However, while Mn is
OTU (Supplementary Table 7). Chemoheterotrophy was the most
elevated in some mine-waters, it also occurs naturally in groundwater
dominant function across all samples (17.7% of total OTUs). OTUs
and samples from one reference site (MCP03) had consistently elevated
assigned to chemoheterotrophy were also associated with other pro­
concentrations of Mn (340–410 μg L− 1). In the reference sites, suboxic
cesses, such as methylotrophy, hydrocarbon degradation, and aromatic
redox conditions and low pH may have favoured elevated concentra­
compound degradation. Exposed sites had significantly (Kruskal-Wallis,
tions of free Mn2+, but there may have also been microbially mediated
p < 0.001) lower relative abundance of OTUs associated with chemo­
dissolution from the sands. The exposed sites also had high concentra­
heterotrophy, hydrocarbon degradation, methylotrophy and methano­
tions of Fe (predominantly as Fe2+, unpublished data), a contaminant
trophy, compared to reference sites (Fig. 5), and significantly (Kruskal-
not associated with mine-waters, and we observed precipitation of iron
Wallis, p < 0.001) higher representation of groups involved with nitri­
oxyhydroxides within the sand channel around these exposed sites,
fication, ureolysis (Fig. 5) and reduction of sulfur compounds along with
indicating activity of iron oxidising bacteria. Therefore, a combination
groups involved in the sulfur cycle (i.e. reduction of sulfur compounds).
of input from mine-waters and the influence of local microbial activity
Multivariate analysis of the functional data separated the exposed
and redox conditions may have created the elevated concentrations of
sites from reference sites (Fig. 6 A), although not as strongly as observed
Mn and Fe at the exposed sites.
for the community composition data. PERMANOVA showed significant
Increases in Mg2+ concentration greater than 3 mg L− 1 have been
difference among the exposure groups, sites and between sampling years
associated with toxic effects on local tropical surface water species
(Supplementary Table 8). Exposure groups, site and the residual
(Hogan et al., 2013; Prouse et al., 2015; van Dam et al., 2010). While
contributed the largest components of variation in the analysis (Sup­
Mg2+ is strongly associated with SO2− 2−
4 in mine-waters, the SO4 ion has
plementary Table 8). Patterns among exposure groups did not vary be­
been shown to have low toxicity to surface water organisms (van Dam
tween years (Exp x Year, p = 0.49) but there was variability in the

5
L. Chandler et al. Environmental Pollution 284 (2021) 117318

Fig. 3. nMDS ordination (A) of the prokaryote communities. Dashed line circles represent significant clusters identified by a SIMPROF test for 45% similarity. dbRDA
ordination (B) showing relationship between prokaryote communities and environmental variables. Vector overlay of environmental variables identified as
significantly correlated using a stepwise DISTLM with AIC.

et al., 2010). The sensitivity of Mg2+ has been attributed to the concentrations at which these effects occurred were markedly higher
extremely soft waters of Magela Creek (van Dam et al., 2010). The high than observed in our samples, indicating direct toxicity from Mg2+ isn’t
toxicity and high mobility of Mg2+ places this major ion as the toxicant likely to be occurring in these waters. SO2−4 is used by a number of
of most concern in mine-waters released, or dispersed, to the Magela prokaryotes as an energy source, through electron donation, and
Creek system. However, elevated concentrations of SO2− 4 were more increasing SO2−
4 has been linked to changes in prokaryote community
likely to be influencing the groundwater prokaryote communities in this structure and function (discussed further below). While specific mech­
system than Mg2+ concentrations, although the two major ions were anisms of influence of increased concentrations of Mg2+ and SO2−4 were
highly correlated. not examined in this study, we hypothesise that the most likely mech­
Increasing solutes such as Mg2+ and SO2−4 can have diverse mecha­ anism for community changes in the shallow groundwater of Magela
nisms of action on prokaryote organisms, which will be dependent on Creek is through preferential growth of species responding to sulfate,
the concentration. Increased levels of Mg2+ have been reported to which probably leads to complex biogeochemical processes with other
decrease survival of Staphylococcus aureus cells by disrupting membrane species in synergistic relationships.
function (Xie and Yang, 2016), as well as decreasing adherence of bac­ Elevated SO2−
4 concentrations have been correlated with changes in
teria to surfaces, impairing biofilm assembly (Demishtein et al., 2019), prokaryote community structure in field (Flynn et al., 2012, 2013) and
which can decrease resistance to external stressors. However, the laboratory (Bethke et al., 2008; Raskin et al., 1996) studies, and our

6
L. Chandler et al. Environmental Pollution 284 (2021) 117318

Fig. 4. The 26 most abundant prokaryote families showing a significant differential abundance between the reference and exposed sites, based on the mean
abundance change in log2fold (p < 0.01). The log2fold change has been adjusted for multiple tests and it is based on a negative binomial model using DESeq2. Circle
size reflects the mean base abundance of OTUs aggregated to family-level and the grey scale indicates the number of different OTUS within the corresponding family.

results concur with these findings. OTUs associated with sulfate chemoheterotrophy and those involved with methane oxidation (i.e.
-reducing bacteria (SRB) (i.e. Desulfovibrionaceae, Desulfobulbaceae methanotrophy) and oxidation of other carbon compounds (i.e. meth­
and Desulfobacteraceae) were more prevalent in the exposed sites ylotrophy), and these functions were also lower at the exposed site.
(Supplementary Figure 7), however they represented only a small per­ Redox potential is one of the most important controls of methane
centage (<0.5%) of the overall relative read abundance in the samples. oxidation (Chowdhury and Dick, 2013), therefore the exposed sites
Flynn et al. (2013) observed that SRB were 4x more abundant in as­ having generally negative Eh (ORP) values) would not favour growth of
semblages attached to sediments within an aquifer, compared to as­ methanotrophs.
semblages suspended in the groundwater fraction. In this study, sulfate In addition to Mg2+ and SO2−4 , we identified Na as a possible driver,
+

reduction was obviously occurring based on observations of strong H2S influencing the differences observed between exposed and reference
odour associated with the exposed sites, but organisms associated with prokaryote communities and functions. Na+ is highly correlated with
the process were not in high abundances and thus were not highlighted Mg2+ and SO2− 4 (r > 0.80), and elevated concentrations of Na in the
+

as major discriminators in analyses between the exposed and reference shallow groundwater of the exposed sites in Magela Creek appeared to
sites. However, this study only examined filtered groundwater samples be mine-water related, as concentrations were higher than reference
and other attached biota may have been involved in the reduction of sites. However, Na+ is not elevated in surface-water discharges from the
sulfur compounds. Future studies aimed at establishing the contribution mine and has not been identified as a Contaminant of Potential Concern
of attached microbes in these processes would be valuable. for the Ranger mine surface water monitoring program. Hence, there has
Samples from our exposed sites had reduced relative abundances of been limited site-specific toxicity testing and it is unknown at what
OTUs assigned to taxa associated with methanogenesis (e.g. Meth­ concentration Na+ may result in adverse effects to local surface water
anobacteriaceae). Flynn et al. (2013) suggested that the relatively low organisms. If elevated concentrations of Na+ were discharged to the
abundance of methanogens they observed in wells with greater than surface-waters of the Magela Creek, it is possible the ecosystem may be
0.03 mM of SO2− − 1
4 (~3 mg L ) could be explained by methanogens being sensitive to the additions due the extremely soft waters of the stream (i.
unable to respire quickly enough to maintain a viable population in the e. <5 mg L− 1 CaCO3). There is also limited information on the influence
presence of active SO2−
4 reduction. Raskin et al. (1996) indicated addi­ of Na+ on prokaryote communities, more broadly. Two studies that have
tion of 30 mg L− 1 of SO2− 4 resulted in decreased concentrations of focused on the effect of Na+ (as NaCl) (i.e. (Baldwin et al., 2006;
methanogens (from 25% initial concentration to 8%) and these low Beyer-Robson, 2014) on sediment prokaryote communities in fresh­
concentrations persisted for at least a year. This could explain why, even water ecosystems applied NaCl experimentally at up to 100 times higher
with measured SO2−4 concentrations decreasing in the second year of our concentrations than those observed in the Magela Creek shallow
study, these groups were still absent, or in very low abundances at sites groundwaters. Baldwin et al. (2006) observed shifts in community
within the exposed group. structure in wetland sediment bacterial communities only at the highest
OTUs assigned to taxa involved with oxidation of methane (i.e. concentration of NaCl (equivalent to approximately 2352 mg L− 1 Na+).
methanotrophs) were also poorly represented in samples from the However, they observed significant changes with increasing NaCl con­
exposed sites, particularly taxa in the Methylococcaceae family. These centrations in the archaeal communities and noted decreased produc­
taxa were associated with several putative functional groups including tion of methane even at low Na+ levels (57 mg L− 1). The decreased

7
L. Chandler et al. Environmental Pollution 284 (2021) 117318

Fig. 5. A heatmap showing distribution of the top 35 functional groups predicted by FAPROTAX. Sites are grouped by exposure, with reference group sites ordered
from the most upstream site (MCP02) to most downstream (MCP15) site. Exposed group sites are ordered by mean concentration of SO42-, with MCP11 having the
lowest concentration and MCP08 the highest. The square root of the relative abundance of each functional group is represented by colour intensity according to the
legend. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

production of methane was attributed to inhibition of acetoclastic considerable spatial heterogeneity in the prokaryote community struc­
methanogenesis, a process undertaken by archaeal genera Meth­ ture within the shallow groundwater of the sand channel that was not
anosarcina and Methanothrix. In our study we noted these taxa had necessarily associated with water chemistry. The study sites were spread
generally reduced relative abundance at the exposed sites compared to over a 7 km reach of the creek channel, with minimum distance of 10 m
reference sites but they were in low abundances (<1%) across all sam­ between the closest sites (Supplementary Figure 1). Although the sand
ples, indicating that acetoclastic methanogenesis was not a dominant channel comprises medium-coarse sands (Roberts, 1991) with similarly
process in the shallow groundwater ecosystem of Magela Creek, and any high hydraulic conductivities along the reach (unpublished data), there
responses to Na+ are presumably associated with other taxa and pro­ is likely to be localised physical differences at each site associated with
cesses. Increased salinity (associated with NaCl) has also been impli­ factors such as differing porosity, and organic content including buried
cated in reduction of nitrification activity (i.e. (Nelson et al., 2016) but leaf litter and riparian plant roots. Brad et al. (2008) observed location
at concentrations of Na+ much higher than those measured in our specific grouping of bacteria communities in sediment samples from an
samples. In laboratory studies, using saturated sandy sediments from aquifer impacted by landfill leachate, even over sampling depth dis­
ephemeral creeks in north Queensland, Australia, Beyer-Robson (2014), tances of 1 m. Lin et al. (2012) noted spatial heterogeneities, both
noted reduced potential rates of nitrification processes with salinities of laterally and vertically in prokaryote communities collected from bores
10000 μS cm− 1 (NaCl concentration not reported). In our study, we approximately 30 m apart in an unconfined aquifer, and hypothesised
observed the opposite, with taxa associated with inferred nitrification (i. the variation was associated with habitat heterogeneities such as voids
e. Nitrososphaera sp.) having higher relative abundance in the exposed between gravelly cobble within the aquifer, the study also observed a
sites (mean of 150 μS cm− 1) compared to the reference sites. It could also strong temporal variation governed by fluctuations in groundwater
be that the high correlations amongst the major ions in our samples are elevation which influenced both geophysical and hydrochemical
confounding results and reflect a statistical, rather than a true causal variables.
effect. Nevertheless, Na+ and Mg2+ are present in equivalent concen­ While PERMANOVA showed a small but significant contribution of
trations in our samples and could be exerting effects independently, sampling year to variation in the prokaryote community composition,
working together or could be antagonising any effects observed on the these year-to-year differences were not marked in ordination space
prokaryote communities. The roles of different ions are complex, and it (Fig. 3). The ordination depicted a general pattern of interspersion of
would require further studies to tease out the effects of elevated con­ 2017 and 2018 data points within each of the exposure groups, indi­
centrations of the different ions on the prokaryote communities in these cating relatively low interannual variability in community composition.
shallow groundwater systems. The interannual variability in communities, may be due to changes in
Sampling location also had a strong influence on community as­ water chemistry with differences particularly evident in sites within the
semblages, and to a lesser extent functional groups, indicating there was contamination plume. These exposed group sites had mean

8
L. Chandler et al. Environmental Pollution 284 (2021) 117318

Fig. 6. nMDS (A) and dbRDA (B) ordinations of the predicted functional groups. Dashed line circles represent significant clusters identified by a SIMPROF test for
80% similarity. Vector overlays on the on the nMDS (A) are predicted functional groups with >0.5 correlation, and similar functional processes have been grouped to
primary function (indicated by bold font) for clarity. dbRDA has a vector overlay of the environmental variables identified as significantly correlated using a stepwise
DISTLM with AIC.

concentrations of contaminants such as Mg2+ and SO2− 4 at least 50% depth of samples between the two years (Supplementary Figure 5),
lower in the second year of the study. This difference was likely due to which, although rarefaction was done to minimise this batch effect, may
the greater rainfall and longer duration of the wet season that preceded have also contributed to the statistical difference between years.
the first sampling, which led to larger and longer surface flow, and There was considerable taxonomic variability within the putative
would have contributed to longitudinal dilution of the contaminant functional groups, and a number of OTUs were assigned to four or more
plume. Alternatively, there was a considerable difference in the read different functional groups indicating a functional redundancy within

9
L. Chandler et al. Environmental Pollution 284 (2021) 117318

these shallow ground water communities. This might explain why the Appendix A. Supplementary data
separation observed between exposure groups was not as strong in the
functional data compared to the community composition data. How­ Supplementary data to this article can be found online at https://doi.
ever, much of the diversity that was observed in the prokaryote com­ org/10.1016/j.envpol.2021.117318.
munity ordinations was associated with OTUs that were unclassified
beyond phyla level, and FAPROTAX ignored OTUs that were not iden­ Author contribution
tified at species or genus level because many functions are only
conserved at this level. In studies of prokaryote communities in sediment Lisa Chandler: Conceptualization, Methodology, Project adminis­
samples from floodplains (Nelson et al., 2016) and a billabong (Sutcliffe tration, Formal analysis, Investigation, Visualisation, Writing - Original.
et al., 2017) within Kakadu National Park large proportions of unclas­ Andrew J Harford: Supervision, Conceptualization, Methodology, Proj­
sified sequences were also reported, and it was suggested that these ect administration, Writing -Reviewing and Editing. Grant C Hose: Su­
sequences may be unique to tropical systems. Sutcliffe et al. (2017) pervision, Conceptualization, Methodology, Writing -Reviewing and
compared shotgun metagenomics with 16S rRNA amplicon data and Editing. Chris L Humphrey: Supervision, Conceptualization, Methodol­
observed that shotgun metagenomics identified a greater number of ogy, Writing -Reviewing and Editing. Anthony Chariton: Resources,
genera groupings than the 16S amplicon data. Future studies examining Methodology, Writing -Reviewing and Editing. Paul Greenfield: Soft­
groundwater communities in these understudied regions would benefit ware, Formal analysis, Writing -Reviewing and Editing. Jenny Davis:
from undertaking whole metagenome sequencing. Supervision, Conceptualization, Methodology, Writing -Reviewing and
Given this is the first report, to our knowledge, examining ground­ Editing.
water prokaryote communities in the region, it provides a baseline
characterisation of these communities for sand channel streams, as well References
as empirical evidence that mine-waters with elevated levels of Mg2+ and
SO42− changed important biogeochemical processes such as (increased) Ahmad, M., Green, D.C., 1986. Groundwater regimes and isotopic studies, Ranger mine
area, Northern Territory. Aust. J. Earth Sci. 33 (4), 391–399. https://doi.org/
nitrification and (decreased) methane oxidation, albeit in a localised 10.1080/08120098608729379.
area. Groundwater prokaryotes are not widely used in assessment of Ahmad, M., Salas, G., Scott, F., 1982. Stream Bed Underflow Studies in the Magela Creek,
mining impacts, despite having a direct link to changes in biogeo­ Northern Territory Geological Survey Technical Report. Department of Mines and
Energy, Darwin.
chemistry, and this study shows they can provide a useful line of evi­ Apprill, A., McNally, S., Parsons, R., Weber, L., 2015. Minor revision to V4 region SSU
dence to inform whole of mine-water management. rRNA 806R gene primer greatly increases detection of SAR11 bacterioplankton.
Aquat. Microb. Ecol. 75 (2), 129–137. https://doi.org/10.3354/ame01753.
Baldwin, D.S., Rees, G.N., Mitchell, A.M., Watson, G., Williams, J., 2006. The short-term
5. Conclusion effects of salinization on anaerobic nutrient cycling and microbial community
structure in sediment from a freshwater wetland. Wetlands 26 (2), 455–464.
Despite a considerable history of work investigating impacts of Bethke, C.M., Ding, D., Jin, Q., Sanford, R.A., 2008. Origin of microbiological zoning in
groundwater flows. Geology 36 (9), 739–742. https://doi.org/10.1130/G24859A.1.
mining in the Alligator Rivers Region, NT, this is the first study of re­
Beyer-Robson, J., 2014. Microbial Communities in an Ephemeral Stream System and the
sponses of groundwater prokaryotes in the region to mine-water Implications of Saline Mine Discharge. Sustainable Minerals Institute. The University
contamination. The study demonstrated clear differences in prokary­ of Queensland.
ote community structure and function in response to a contamination Brad, T., van Breukelen, B.M., Braster, M., van Straalen, N.M., Röling, W.F.M., 2008.
Spatial heterogeneity in sediment-associated bacterial and eukaryotic communities
plume on the Ranger Project Area. The contamination was associated in a landfill leachate-contaminated aquifer. FEMS Microbiol. Ecol. 65 (3), 534–543.
primarily with Mg2+ and SO2− 4 , although Na was identified as the
+
https://doi.org/10.1111/j.1574-6941.2008.00533.x.
variable most strongly correlated with prokaryote structure and func­ Cañedo-Argüelles, M., Hawkins, C.P., Kefford, B.J., Schäfer, R.B., Dyack, B.J., Brucet, S.,
Buchwalter, D.B., Dunlop, J.E., Frör, O., Lazorchak, J., Coring, E., Fernandez, H.R.,
tion. This knowledge assists in the assessment of potential mining im­ Goodfellow, W., González Achem, A.L., Hatfield-Dodds, S., Karimov, B.K.,
pacts on groundwater ecosystems and will aid ongoing mine-water Mensah, P., Olson, J.R., Piscart, C., Prat, N., Ponsá, S., Schulz, C.-J., Timpano, A.J.,
management. 2016. Saving freshwater from salts: ion-specific standards are needed to protect
diversity. Science 351 (6276), 914–916. https://doi.org/10.1126/science.aad3488.
Caporasa, J.G., Lauber, C.L., Walters, W.A., Berg-Lyons, D., Huntley, J., Fierer, N.,
Declaration of competing interest Owens, S.M., Betley, J., Fraser, L., Bauer, M., Gormley, N., Gilbert, J.A., Smith, G.,
Knight, R., 2012. Ultra-high-throughput microbial community analysis on the
Illumina HiSeq and MiSeq platforms. ISME J. 6, 1621–1624. https://doi.org/
The authors declare that they have no known competing financial 10.1038/ismej.2012.8.
interests or personal relationships that could have appeared to influence Chowdhury, T.R., Dick, R.P., 2013. Ecology of aerobic methanotrophs in controlling
the work reported in this paper. methane fluxes from wetlands. Appl. Soil Ecol. 65, 8–22. https://doi.org/10.1016/j.
apsoil.2012.12.014.
Clarke, K.R., Gorley, R.N., 2015. PRIMER V7: User Manual/Tutorial. Primer-E Ltd,
Acknowledgements Plymouth, UK.
Demishtein, K., Reifen, R., Shemesh, M., 2019. Antimicrobial properties of magnesium
We pay our respects to all Traditional Owners of Kakadu National open opportunities to develop healthier food. Nutrients 11 (10), 2363. https://doi.
org/10.3390/nu11102363.
Park where we conducted this study and acknowledge Elders past, Flynn, T.M., Sanford, R.A., Ryu, H., Bethke, C.M., Levine, A.D., Ashbolt, N.J., Santo
present and emerging. This study was funded by the Supervising Sci­ Domingo, J.W., 2013. Functional microbial diversity explains groundwater
entist Branch (SSB) of the Department of Agriculture, Water and the chemistry in a pristine aquifer. BMC Microbiol. 13 (1), 146. https://doi.org/
10.1186/1471-2180-13-146.
Environment and through an Australian Government Research Training Flynn, T.M., Sanford, R.A., Santo Domingo, J.W., Ashbolt, N.J., Levine, A.D., Bethke, C.
Program Scholarship awarded to Lisa Chandler. We would like to thank M., 2012. The active bacterial community in a pristine confined aquifer. Water
Andrew Jansen, Julie Hanley, Costi Buccella, Ian Douglas, Matt Piamnok Resour. Res. 48 (9), W09510. https://doi.org/10.1029/2011WR011568.
Gissi, F., Reichelt-Brushett, A.J., Chariton, A.A., Stauber, J.L., Greenfield, P.,
and Sam Walker (SSB) for assistance in the field and collection of sam­ Humphrey, C., Salmon, M., Stephenson, S.A., Cresswell, T., Jolley, D.F., 2019. The
ples. We are grateful to Brodie Sutcliffe for her support and guidance effect of dissolved nickel and copper on the adult coral Acropora muricata and its
with the DNA extraction and amplification process, along with other microbiome. Environ. Pollut. 250, 792–806. https://doi.org/10.1016/j.
envpol.2019.04.030.
members of the Macquarie University EGEEL Laboratory. We’d also like Goldscheider, N., Hunkeler, D., Rossi, P., 2006. Review: microbial biocenoses in pristine
to thank Mirjam Kaestli (Charles Darwin University) for advice with aquifers and an assessment of investigative methods. Hydrogeol. J. 14 (6), 926–941.
statistical analysis. The authors declare that they have no conflicts of https://doi.org/10.1007/s10040-005-0009-9.
Griebler, C., Avramov, M., Hose, G., 2019. Groundwater ecosystems and their services:
interest.
current status and potential risks. In: Schröter, M., Bonn, A., Klotz, S., Seppelt, R.,
Baessler, C. (Eds.), Atlas of Ecosystem Services. Springer, Cham, pp. 197–203.

10
L. Chandler et al. Environmental Pollution 284 (2021) 117318

Griebler, C., Leuders, T., 2009. Microbial biodiversity in groundwaters. Freshw. Biol. 54 Australia. Sediment. Geol. 83, 277–302. https://doi.org/10.1016/0037-0738(93)
(4), 649–677. https://doi.org/10.1111/j.1365-2427.2008.02013.x. 90017-Y.
Grösbacher, M., Spicher, C., Bayer, A., Obst, M., Karwautz, C., Pilloni, G., Nelson, T.M., Streten, C., Gibb, K.S., Chariton, A.A., 2016. Bacteria in tropical floodplain
Wachsmann, M., Scherb, H., Griebler, C., 2016. Organic contamination versus soils are sensitive to changes in saltwater. Mar. Freshw. Res. 69, 1110–1123. https://
mineral properties: competing selective forces shaping bacterial community doi.org/10.1071/MF16033.
assembly in aquifer sediments. Aquat. Microb. Ecol. 76 (3), 243–255. https://doi. Noller, B.N., 1991. Non-radiological contaminants from uranium mining and milling at
org/10.3354/ame01781. ranger, jabiru, northern territory, Australia. Environ. Monit. Assess. 19 (1–3),
Harford, A.J., Hogan, A.C., Jones, D.R., van Dam, R.A., 2015. Ecotoxicology of highly 383–400. https://doi.org/10.1007/BF00401327.
treated mine waters: lessons from an Australian mine. Mine Water Environ. 34 (1), Nordstrom, D.K., Blowes, D.W., Ptacek, C.J., 2015. Hydrogeochemistry and microbiology
75–86. https://doi.org/10.1007/s10230-014-0282-8. of mine drainage: an update. Appl. Geochem. 57, 3–16. https://doi.org/10.1016/j.
Hart, B.T., Noller, B.N., LeGras, C., Currey, N., 1992. Manganese speciation in Magela apgeochem.2015.02.008.
creek, northern Australia. Mar. Freshw. Res. 43 (2), 421–431. https://doi.org/ Parada, A.E., Needham, D.M., Fuhrman, J.A., 2016. Every base matters: assessing small
10.1071/MF9920421. subunit rRNA primers for marine microbiomes with mock communities, time series
He, Z., Zhang, P., Wu, L., Rocha, A.M., Tu, Q., Shi, Z., Wu, B., Qin, Y., Wang, J., Yan, Q., and global field samples. Environ. Microbiol. 18 (5), 1403–1414. https://doi.org/
Curtis, D., Ning, D., Van Nostrand, J.D., Wu, L., Yang, Y., Elias, D.A., Watson, D.B., 10.1111/1462-2920.13023.
Adams, M.W.W., Fields, M.W., Alm, E.J., Hazen, T.C., Adams, P.D., Arkin, A.P., Prouse, A.E., Hogan, A.C., Harford, A.J., van Dam, R.A., Nugegoda, D., 2015. Hydra
Zhou, J., 2018. Microbial functional gene diversity predicts groundwater viridissima (green Hydra) rapidly recovers from multiple magnesium pulse exposures.
contamination and ecosystem functioning. mBio 9 (1). https://doi.org/10.1128/ Environ. Toxicol. Chem. 34 (8), 1734–1743. https://doi.org/10.1002/etc.2966.
mBio.02435-17 e02435-02417. Raskin, L., Rittmann, B.E., Stahl, D.A., 1996. Competition and coexistence of sulfate-
Hemme, C.L., Deng, Y., Gentry, T.J., Fields, M.W., Wu, L., Barua, S., Barry, K., Green- reducing and methanogenic populations in anaerobic biofilms. Appl. Environ.
Tringe, S., Watson, D.B., He, Z., Hazen, T.C., Tiedje, J.M., Rubin, E.M., Zhou, J., Microbiol. 62 (10), 3847. https://doi.org/10.1128/aem.62.10.3847-3857.1996.
2010. Metagenomic insights into evolution of a heavy metal contaminated Roberts, R.G., 1991. Sediment Budgets and Quaternary History of the Magela Creek
groundwater microbial community. ISME J. 4 (5), 660–672. https://doi.org/ Catchment, Tropical Northern Australia. Department of Geography. University of
10.1038/ismej.2009.154. Wollongong.
Hemme, C.L., Tu, Q., Shi, Z., Qin, Y., Chain, P.S.G., Tringe, S.G., Fields, M.W., Rubin, E. Sang, S., Zhang, X., Dai, H., Hu, B.X., Ou, H., Sun, L., 2018. Diversity and predictive
M., Tiedje, J.M., Hazen, T.C., Arkin, A.P., Zhou, J., 2015. Comparative metabolic pathways of the prokaryotic microbial community along a groundwater
metagenomics reveals impact of contaminants on groundwater microbes. Front. salinity gradient of the Pearl River Delta, China. Sci. Rep. 8 (1), 17317. https://doi.
Microbiol. 6, 1205. https://doi.org/10.3389/fmicb.2015.01205. org/10.1038/s41598-018-35350-2.
Héry, M., Volant, A., Garing, C., Luquot, L., Elbaz Poulichet, F., Gouze, P., 2014. Smith, R.J., Jeffries, T.C., Roudnew, B., Fitch, A.J., Seymour, J.R., Delphin, M.W.,
Diversity and geochemical structuring of bacterial communities along a salinity Newton, K., Brown, M.H., Mitchell, J.G., 2012. Metagenomic comparison of
gradient in a carbonate aquifer subject to seawater intrusion. FEMS Microbiol. Ecol. microbial communities inhabiting confined and unconfined aquifer ecosystems.
90 (3), 922–934. https://doi.org/10.1111/1574-6941.12445. Environ. Microbiol. 14 (1), 240–253. https://doi.org/10.1111/j.1462-
Hogan, A.C., Trenfield, M.A., Harford, A.J., van Dam, R.A., 2013. Toxicity of magnesium 2920.2011.02614.x.
pulses to tropical freshwater species and the development of a duration-based water Smith, R.J., Paterson, J.S., Wallis, I., Launer, E., Banks, E.W., Bresciani, E., Cranswick, R.
quality guideline. Environ. Toxicol. Chem. 32 (9), 1969–1980. https://doi.org/ H., Tobe, S.S., Marri, S., Goonan, P., Mitchell, J.G., 2018. Southern South Australian
10.1002/etc.2251. groundwater microbe diversity. FEMS Microbiol. Ecol. 94 (10), fiy158. https://doi.
Humphrey, C.L., Chandler, L., 2018. Use of Field-Effects Information to Inform Surface org/10.1093/femsec/fiy158.
Water Guideline Values for Magnesium Sulfate in Magela Creek, Supervising Stephenson, S., Chariton, A.A., Holley, M.P., O’Sullivan, M.M., Gillings, M.R., Hose, G.C.,
Scientist Report 212. Commonwealth of Australia, Darwin, NT, Australia, p. 234. 2013. Changes in prokaryote and eukaryote assemblages along a gradient of
Korbel, K.L., Lim, R.P., Hose, G.C., 2013. An inter-catchment comparison of groundwater hydrocarbon contamination in groundwater. Geomicrobiol. J. 30 (7), 623–631.
biota in the cotton-growing region of north-western New South Wales. Crop Pasture https://doi.org/10.1080/01490451.2012.746408.
Sci. 64 (12), 1195–1208. https://doi.org/10.1071/CP13176. Sundaram, B., Feitz, A.J., de Caritat, P., Plazinska, A., Brodie, R.S., Coram, J.,
Lin, X., McKinley, J., Resch, C.T., Kaluzny, R., Lauber, C.L., Fredrickson, J., Knight, R., Ransley, T., 2009. Groundwater Sampling and Analysis – A Field Guide, p. 104.
Konopka, A., 2012. Spatial and temporal dynamics of the microbial community in Geoscience Australia.
the Hanford unconfined aquifer. ISME J. 6 (9), 1665–1676. https://doi.org/ Sutcliffe, B., Chariton, A.A., Harford, A.J., Hose, G.C., Greenfield, P., Elbourne, L.D.H.,
10.1038/ismej.2012.26. Oytam, Y., Stephenson, S., Midgley, D.J., Paulsen, I.T., 2017. Effects of uranium
Lottermoser, B., 2010. Mine Wastes: Characterization, Treatment and Environmental concentration on microbial community structure and functional potential. Environ.
Impacts. Springer, Berlin, Heidelberg, Germany. Microbiol. 19 (8), 3323–3341. https://doi.org/10.1111/1462-2920.13839.
Louca, S., Wegener Parfrey, L., Doebeli, M., 2016. Decoupling function and taxonomy in Taubert, M., Stähly, J., Kolb, S., Küsel, K., 2019. Divergent microbial communities in
the global ocean microbiome. Science 353 (6305), 1272–1277. https://doi.org/ groundwater and overlying soils exhibit functional redundancy for plant-
10.1126/science.aaf4507. polysaccharide degradation. PloS One 14 (3), e0212937. https://doi.org/10.1371/
McCullough, C.D., 2006. A Multi-Scale Assessment of the Ecological Risk of Magnesium journal.pone.0212937.
Sulfate to Aquatic Biota of Magela Creek, Northern Territory, Faculty of Education, Trenfield, M.A., Harford, A.J., Mooney, T., Ellis, M., Humphrey, C., van Dam, R.A., 2019.
Health and Science. Charles Darwin University, Darwin. Integrating laboratory and field studies to assess impacts of discharge from a
McMurdie, P.J., Holmes, S., 2013. phyloseq: an R package for reproducible interactive uranium mine and validate a water quality guideline value for magnesium.
analysis and graphics of microbiome census data. PloS One 8 (4), e61217. https:// Integrated Environ. Assess. Manag. 15 (1), 64–76. https://doi.org/10.1002/
doi.org/10.1371/journal.pone.0061217. ieam.4098.
McMurdie, P.J., Holmes, S., 2014. Waste not, want not: why rarefying microbiom data is van Dam, R.A., Hogan, A.C., McCullough, C.D., Houston, M.A., Humphrey, C.L.,
inadmissible. PLoS Comput. Biol. 10 (4), e1003531 https://doi.org/10.1371/ Harford, A.J., 2010. Aquatic toxicity of magnesium sulfate, and the influence of
journal.pcbi.1003531. calcium, in very low ionic concentration water. Environ. Toxicol. Chem. 29 (2),
Mooney, T.J., McCullough, C.D., Jansen, A., Chandler, L., Douglas, M., Harford, A.J., van 410–421. https://doi.org/10.1002/etc.56.
Dam, R., Humphrey, C., 2020. Elevated magnesium concentrations altered Wright, J., Kirchner, V., Bernard, W., Ulrich, N., McLimans, C., Campa, M.F., Hazen, T.,
freshwater assemblage structures in a mesocosm experiment. Environ. Toxicol. Macbeth, T., Marabello, D., McDermott, J., Mackelprang, R., Roth, K.,
Chem. 39, 1973–1987. https://doi.org/10.1002/etc.4817. Lamendella, R., 2017. Bacterial community dynamics in Dichloromethane-
Mouser, P.J., Rizzo, D.M., Röling, W.F.M., van Breukelen, B.M., 2005. A multivariate contaminated groundwater undergoing natural attenuation. Front. Microbiol. 8,
statistical approach to spatial representation of groundwater contamination using 2300. https://doi.org/10.3389/fmicb.2017.02300.
hydrochemistry and microbial community profiles. Environ. Sci. Technol. 39 (19), Xie, Y., Yang, L., 2016. Calcium and magnesium ions are membrane-active against
7551–7559. https://doi.org/10.1021/es0502627. stationary-phase Staphylococcus aureus with high specificity. Sci. Rep. 6 (1), 20628.
Nanson, G.C., East, T.J., Roberts, R.G., 1993. Quaternary stratigraphy, geochronology https://doi.org/10.1038/srep20628.
and evolution of the Magela Creek catchment in the monsoon tropics of northern

11