The spatial legacy of Australian mercury contamination in the sediment of the Molonglo River

European colonies were first established in Australia in 1788. The discovery of gold and subsequent gold rushes in the 1850s brought migrants to the country and developed mining into a full-fledged industry (Blainey, 1993; Davies and Oliver, 2018). The prosperity of this industry formed an economic backbone of early European Australia, and the country enjoys the legacy of that prosperity to this day. Much of this mining industry, however, has also left a legacy of contamination. Given the lack of metal contamination control in these historical mines, a comprehensive environmental assessment is needed to comprehend the extent and nature of mining legacy impacts on ecosystems and communities across Australia. It is estimated that more than 50,000 legacy mines in the country require environmental assessments to avoid ongoing and future environmental issues (Pepper et al., 2014).

Captains Flat, in New South Wales, is home to the Lake George Mine, a legacy mine site that predates the implementation of any environmental protection legislation in New South Wales (Zillig et al., 2015). The earliest relevant legislation in the state, the Mining Act 1992, was introduced 30 years after the mine closed (Zillig et al., 2015). Mining in Captains Flat originally sought gold, but later targeted galena, sphalerite, chalcopyrite and pyrite deposits in the orebody, seeking lead, zinc, copper and iron (Pryke, 1995). Hg amalgamation techniques in Captains Flat were used to extract gold, but the sulfide ores targeted in later mining were also known to contain Hg by-products (Pryke, 1995; Rytuba, 2003). These mining operations contaminated the Molonglo River, which feeds Lake Burley Griffin, an artificial lake in Canberra, the Australian capital. The lake was completed in 1963 after the Molonglo River was dammed (Caitcheon et al., 1988). The lake is located in the approximate geographic centre of the capital and is a popular location for many recreational uses; including swimming, rowing, fishing, and sailing.

While Lake Burley Griffin has social and heritage importance, it is a focus of study here because it acts as a sink for sediment transported by the Molonglo River (Caitcheon et al., 1988). Previous studies have shown significant contamination of metals in both the Molonglo River and Lake Burley Griffin (Craze, 1980; Jacobson and Sparksman, 1988; Maher et al., 1992; Wadige et al., 2016; Weatherley et al., 1967) but no study has considered Hg or the environmental and physical factors playing a role in Hg metal distribution in the Molonglo-Lake Burley Griffin system.

Here we provide the first assessment of Hg contamination in the Molonglo River and Lake Burley Griffin. Analyses of Hg in sediment cores were used to assess the temporal Hg contamination in Captains Flat and Lake Burley Griffin. We also evaluated the influence of distance from the mine, total organic matter (TOM) and iron (Fe) abundance in sediment, and water depth on the distribution of metals in Lake Burley Griffin. Distance from the mine, and hence separation from the likely source of Hg, was considered the primary factor that would determine the mine’s influence on the Molonglo River system. Total organic matter (TOM), which consists of dissolved and particulate organic matter, is a significant physical factor for the environmental distribution of Hg. TOM typically comprises macromolecular humic and fulvic acids that complex dissolved metals and form heterogenous colloidal species with iron oxides and oxyhydroxides (Huber et al., 2011). Iron (Fe) was also analysed in this study as it plays a key role in the adsorption of Hg to particulates and their precipitation in sediments (Tukura et al., 2007). Most Fe in this area is expected to be Fe oxides and oxyhydroxides, as these are common components of natural gossans, soils and mine waste (Bowell, 1994). Fe sulphides are also expected, given the high sulfur content of mine waste and seepage on the Lake George Mine. These phases are prevalent in natural environments as particles or surface coatings, and have a high affinity for metal ions, allowing them to control metal contaminant adsorption and mobility (Tukura et al., 2007). Previous work in the field of lake-borne metal contamination had found depth to be a significant factor in the distribution of metals. Water slows with increasing depth, and hence its capacity to transport particles in suspension reduces. This extra deposition can increase the density of metal particles in sediment, hence its consideration here (Schneider et al., 2016).

The aims of the study were: (i) to determine the extent, both historical and spatial, of Hg contamination in Captains Flat and Lake Burley Griffin (ii) provide information on Hg transport from Lake George Mine to Lake Burley Griffin; and (iii) to determine physical and chemical environmental factors affecting Hg distribution in surface samples in the Molonglo River system. As Hg may be transported in association with particles or as dissolved inorganic species (Hsu-Kim et al., 2018), physical and chemical factors controlling sediment and Hg dispersion need to be considered in order to fully understand its distribution throughout the system (Schneider et al., 2016).

The results of this paper will provide additional information on the state of the abandoned Lake George Mine to be considered in future rehabilitation, and provides an environmental health assessment of Lake Burley Griffin to support the Australian Commonwealth Government to protect and manage a site of heritage values to Australians.

The Molonglo River originates upstream of the township Captains Flat, New South Wales, approximately 45 kilometres southeast of Canberra, Australian Capital Territory (Figure 1). Impounded by the town’s dam, the river then flows past the former mine site, through farm country and into the regional centre of Queanbeyan. Here the Queanbeyan River, a major tributary, feeds into the Molonglo River, and flows further downstream into Lake Burley Griffin (Figure 1).

The bedrock underlying the Molonglo River mostly comprises Silurian and Ordovician sedimentary rock, Silurian volcanic rock, and Quaternary alluvial deposits (NSW Department of Planning and Environment, 2019). The Captains Flat township is located in a cool temperate zone, with an annual mean maximum temperature at Captains Flat of 19.3°C, and a mean monthly minimum of 6.1°C (Bureau of Meterology, 2019). The township typically receives an average of 742 mm of rainfall a year, with the maximum in November and the minimum in July (Bureau of Meterology, 2019).

Lake Burley Griffin is an artificial recreational water body located in the centre of Canberra, with an approximate length of 9 km and a width that varies between 300 and 1200 metres (National Capital Authority, 2019) (Figure 2A). Lake Burley Griffin has an average depth of 4 m and a maximum depth of about 17.6 m, where its outflow and water levels are regulated by the 33 m tall Scrivener Dam, at the south-western end of the Lake (Figure 2A) (National Capital Authority, 2019). The Lake is divided into three primary basins, separated by major road-bridges (Figure 2A). The East and Central Basins, featuring more lakeside development, are both commonly used for lakeside recreation purposes like walking and exercise, while the less-built-up West Basin tends to be used more for water-based recreation, such as sailing, rowing and fishing. Given the reduction in capacity that decreases in flow rates can have on sediment dispersion and transport, it is expected that more material from Captains Flat will be present in the East Basin, likely around the Kingston Foreshore area (Figure 2A). Sullivans Creek and Jerrabomberra Creek, flowing into the West Basin from the north and the East Basin from the south, respectively, are also notable potential sources of material for the Lake system.

Mining in the Captains Flat area, though beginning with alluvial gold mining, was formalised with the establishment of two mines in 1886, Koh-i-noor to the north of the township and El Capitan to the south. They were merged into one company ten years later. This combined venture, the Lake George Mine, is the main source of heavy metal contamination in the contemporary Molonglo River (Maher et al., 1992).

Captains Flat mines targeted the Silurian-era sulfide orebody across its two primary operational periods: the original period between 1896 and 1899, and its reopening between 1925 and 1962 (McGowan and Grinbergs, 2006). The mining operation released Hg in two primary ways: first, as a waste product as part of amalgamation gold extraction during the first mining period; secondly, as a sulfide by-product from the extraction of lead, zinc, copper, and iron from the galena, sphalerite, chalcopyrite, and pyrite in the orebody during the second operational period.

Two tailing dumps were used to store waste product extracted from both the mine and the smelters, one south of the mine and one north of it (Figure 2B). These dumps each failed on at least one occasion (the southern in 1942 and the northern in 1945) delivering mine waste material into the Molonglo River (Brooks, 1998). Although these events are not the only known failures of the tailings dumps, on the basis of these alone the dumps have been identified as major sources of heavy metal contamination (Wadige et al., 2016).

The Captains Flat Dam, the main reservoir for the township, was constructed adjacent to the southern tailings dump in 1936 (Pryke, 1995). The failing of the southern dump in 1942 deposited waste material directly into this reservoir (Brooks, 1998), and may have posed risk of contamination in the township’s drinking water supply.

The Lake George Mine also had three integrated smelting stoves to extract metals from mined ore (McGowan and Grinbergs, 2006). When the mine was in operation, the primary smelters were located adjacent to the southern tailings dump (Figure 2B). The predominant smelting process was the direct pyritic process developed by Robert Carl Sticht, previously adopted by the Mt Lyell mine in Queenstown, Tasmania (McGowan and Grinbergs, 2006). Essentially, the pyritic smelting technique treats ore using heat generated by its own oxidation reaction; hence saving on fuel and labour costs in the treatment process. While adoption of these techniques increased the mine’s output (Mainwaring, 2011), the oxidation of these ores also releases several waste products, including sulfur, and in the case of the Lake George Mine, Hg. These smelters were used until the mine’s closure in 1962 (Mainwaring, 2011).

In addition to smelting, the “selective flotation” process was adopted in the late 1930s for the handling of complex ores (McGowan and Grinbergs, 2006). In this process the ore is mixed with chemical reagents and oil rather than heating the ore as in smelting or other processes. This extracts the target metals – lead, zinc, copper and iron – in phases by the combination of agitators that froth the mixture and a horizontal “skimming fan” that removes the froth – containing extracted metals – from the top (McGowan and Grinbergs, 2006). The remainder of the mixture was contained in the northern tailings dump, near to the treatment plant site (Figure 2B).

A sediment core was collected from the Captains Flat Dam in 2014 (Figure 2B, Table 1). The 43 cm-core was collected at 5 m-water depth, using a single drive universal corer in a polycarbonate tube. This core was sliced every 1 cm and stored at 4°C in plastic bags at the Palaeoworks Laboratory at the Department of Archaeology and Natural History, Australian National University, Canberra, until analyses.

Four sediment cores were retrieved from Lake Burley Griffin (Figure 2A), also using a single drive universal corer in a polycarbonate tube. One core was collected in 2011 in the Lake’s west basin, while three more were collected in the east basin in 2016 (Table 1).

A total of 42 sediment samples were collected during 2018 (Figures 1 and 3). The first collection, in April, focused upon Lake Burley Griffin totalling 25 sediment samples (Figure 3A). The October collection focused on the Molonglo River and Captains Flat Township, totalling 17 samples (Figure 3B). All samples were collected using a zinc-plated Petersen grab and stored in Falcon 50-mL high-density polyethylene centrifuge tubes (Thermo Fisher, Sydney, Australia) within a cooler with ice, and stored on return to the laboratory in a 4°C cool room at the Australian National University. Additionally, in Lake Burley Griffin, water depth was measured at each sampling point using a Norcross Hawkeye H22PX Handheld Digital Sonar.

Before analyses, sediment samples were freeze-dried for 72 hours using a Christ Alpha 1–2 LDplus lyophilizer (John Morris Scientific, Sydney, Australia), and stored in refrigerated conditions until analysis. Additionally, sediment samples were sorted with a sieve (<63µm) using a Fritsch Analysette 3 Spartan vibratory sieve shaker (John Morris Scientific, Sydney, Australia). This was in order to remove the effect of grain size on metal concentrations, as different grain sizes can have differing absorptive capacities for metal (Zoumis et al., 2001).

The four sediment cores were analysed at the Australian Nuclear Science and Technology Organisation (ANSTO). Samples from the LBG1, LBG3 and CF sediment cores were analysed using gamma ray spectrometry for the determination of lead-210 (²¹⁰Pb), radium-226 (²²⁶Ra) and caesium-137 (¹³⁷Cs) activities (Schneider et al., 2014) (Figure S1.). Samples from the LBG4 sediment core were analysed by alpha particle spectrometry for the determination of polonium-210 (²¹⁰Po) and ²²⁶Ra activities (Atahan et al., 2015). The unsupported ²¹⁰Pb activity was calculated by subtracting the activity of ²²⁶Ra from ²¹⁰Pb or ²¹⁰Po for each sample. Based on the calculated unsupported ²¹⁰Pb activities from each core, the Constant Flux Constant Sedimentation (CFCS) model was used to determine sediment ages (Krishnaswamy et al., 1971).

Total organic matter (TOM) analyses were performed using the weight-loss-on-ignition method described by Wang et al. (2011). Samples were sieved to a <2 mm sieve and approximately one gram of each sample was weighed out and heated in a muffle furnace (model CEMLL-SD; Labec, Sydney, Australia) at 550° for six hours. Once cooled to room temperature, samples were weighed, and the difference was reported as organic content in mass/mass percentages.

Prior to Fe analyses, sediment samples were homogenized by intensive manual mixing of sediment. Approximately 0.2 g of freeze-dried material was weighed into a 60-mL polytetrafluoroacetate (PFA) closed digestion vessel (Mars Express) and 2 mL of concentrated nitric acid (Aristar, BDH Australia) and 1 mL of 30% concentrated hydrochloric acid (Merck Suprapur, Germany) added (Telford et al., 2008). Each PFA vessel was then capped, placed into an 800-W microwave oven (CEM model MDS-81, Indian Trail, NC, USA), and samples heated at 120°C for 15 min. The digests were cooled to room temperature and diluted to 50 mL with deionised water (Sartorius). The tubes were then centrifuged at 5000 rpm for 10 min. One mL of the digest was transferred into a 10-mL centrifuge tube and then diluted to 10 mL with an inductively coupled plasma mass spectrometer internal standard (Li⁶, Y¹⁹, Se⁴⁵, Rh¹⁰³, In¹¹⁶, Tb¹⁵⁹ and Ho¹⁶⁵). Digests were stored (0–5°C) until analysis.

Samples were analysed for total Fe using an inductively coupled plasma mass spectrometer (Perkin Elmer DRC-e) with an AS-90 autosampler (Maher et al., 2001). The certified reference NIST- 2711a (Buffalo River sediment) and Climate Change Canada WQB-1 Lake Ontario were used as controls to check the quality and traceability of metal results. Measured concentrations were in agreement with certified values (Table S1.).

Mercury analyses were conducted using a Milestone Direct Mercury Analyzer (DMA-80 Tri-Cell; Milestone, Bergamo, Italy). The instrument was calibrated with aqueous Hg standard for AAS Sigma Aldrich TraceCERT® (1000 mg/L Hg in nitric acid). A calibration curve was constructed by plotting the absorbances of standards against Hg concentrations in nanograms, which was considered valid for an R² value of 0.99 or higher.

The DMA-80 determines total Hg concentration through a sequence of thermal decomposition, amalgamation, and atomic absorption spectrometry. Samples were analysed using the USEPA method 7473 (USEPA, 1998); two blanks and two Standard Reference Materials (SRMs) were analysed for every 40 samples. Approximately 100 mg of sample was weighed in nickel boats. After every tenth sample, a replicate sample was analysed. When replicate recovery exceeded a variance of 10% compared to the original sample, a third replicate was run. Certified sediment reference materials NIST-2711a (Montana River Sediment) and SECCC WQB-1 (Lake Ontario) were analysed, and results were in agreement with published values (Table S1.).

To understand the main drivers of Hg spatial distribution in sediments of the Molonglo River and Lake Burley Griffin, the correlation between Hg and distance from the mine was assessed with the following factors: distance from the source point, TOM, and total Fe concentration. Additionally, water depth was considered as a significant factor for Hg distribution for Lake Burley Griffin, due to its influence identified in studies of other lakes (Schneider et al., 2016). In the Molonglo River, depth was not considered as a predictor for Hg distribution as the river depth is relatively consistent along its transect between Captains Flat and Lake Burley Griffin.

All computations investigating these correlations were performed using the software R 3.6.0 (R Development Core Team, 2008). Age-depth models were performed using the R package ‘rbacon’ (https://cran.r-project.org/web/packages/rbacon/rbacon.pdf) (Blaauw and Christen, 2011). Stratiplot graphs were built using the R package ‘analogue’ using the function Stratiplot (https://cran.r-project.org/web/packages/analogue/index.html).

Prior to running parametric statistics, assumptions were checked, and data were transformed to the natural logarithm (ln). A multiple regression with backward-stepwise selection was performed to identify the chemical and physical factors that most significantly explain metal distribution in sediments in the Molonglo River and Lake Burley Griffin. These chemical and physical factors included distance from the mouth of the river (for Lake Burley Griffin samples) and distance from the mine (for Molonglo River samples), organic matter (% mass/mass), water depth (for Lake Burley Griffin) and Fe oxides and oxyhydroxides.

Two principal component analyses (PCA) were performed, one for the Lake Burley Griffin samples and one for the Molonglo River samples. The function prcomp() was used to run the PCA and the ‘ggplot2’ R package (https://cran.r-project.org/web/packages/ggplot2/index.html) was used to create a PCA plot using the function ggbiplot. Samples MR01, MR02 and MR06 have not been included as they are from sites upstream of the mine. Additionally, LBG19 has not been included in the PCA because there was insufficient sample to conduct all analyses required.

For the sediment core collected in Captains Flat, unsupported ²¹⁰Pb activities do not exhibit a decay profile with depth (Table S2.) and the activity is still relatively high at 36 cm depth. The deviation from a decay profile is likely the result of leached contaminants, including ²¹⁰Pb, from the mine’s southern tailings dump. Additionally, retention failures in the tailings dumps, especially the southern dump’s collapse in 1942 (Brooks, 1998), could have introduced contaminated sediment from much earlier in the mine’s history, altering the natural accumulation of ²¹⁰Pb in the dam (Table S2.). All these events affect the application of the ²¹⁰Pb dating method as its accuracy depends on the estimation of unsupported (i.e., excess or atmospherically derived) ²¹⁰Pb activity at contiguous depths throughout the sediment core. Our dating results corroborate studies reporting mining leaching as creating special difficulties for ²¹⁰Pb dating, as disequilibrium between ²²⁶Ra and supported ²¹⁰Pb creates problems for dating sediment deposits (Brenner et al., 2004; Mudd, 2008).

Although ²¹⁰Pb activities were not useful to date this core, ¹³⁷Cs activities were only detected in the top 9 cm of this core, indicating that the sediments above this depth refer to deposition after 1950, the commencement of atomic bomb testing in Australia. Our interpretation for this core, therefore, relies on the time before and after 1950 (Figure S3.). Based on the ¹³⁷Cs activities found in the top 9cm of the core, the sedimentation rate at the Captains Flat Dam is estimated to be 0.14cm/year.

A strong correlation between Hg and S, Fe, Cu, Zn and Pb concentration/depth profiles in the Captains Flat core (Spearman rank correlation >0.6, p < 0.01, unpublished data) indicates that accumulated sediments in Captains Flat Dam were highly influenced by erosional inputs of mine tailings enriched in these elements before and after 1950 (Craze, 1980).

For Lake Burley Griffin, the core collected in 2011 in the West Basin exhibited a decay profile of unsupported ²¹⁰Pb activities with depth (Table S3.) and was successfully dated down to 15 cm. Sediment ages were calculated between 0 and 15 cm using the CFCS dating model, finding the 15 cm depth correlated to the beginning of the 1960s. Below 15 cm, the unsupported ²¹⁰Pb activities were not included in the calculations, as the activities no longer exhibited a decreasing profile with depth (Figure S1.). Other proxies (charcoal and pollen) have assisted the chronology in this study and results for core LBG1 are detailed in Pritchard et al. (2019).

The three sediment cores collected in 2015 on the East Basin were not suitable for dating, as ²¹⁰Pb activities did not exhibit a decay and ¹³⁷Cs activities were low and not reliable for dating (Tables S4. and S5.). This dating issue is likely a consequence of the original construction of the Lake as well as the recent development of the Kingston Foreshore (Figure 2A), a lakeside urban area. Formerly industrial, the area has been undergoing development into a recreational/commercial/residential district since the turn of the century, with construction of multiple apartment complexes, and landscaping of a promenade and harbour area on the Lake (ACT Government Suburban Land Agency, 2019). This landscaping process involved lakebed dredging work in Lake Burley Griffin, which dispersed and redistributed the historical sedimentation present in the East Basin. Even so, the lake’s original construction involved scouring top-soil and vegetation from the existing landscape. This could have impaired the suitability of the sediment for dating long before the Kingston Foreshore development.

Given that the profile of Hg concentration/depth for the Captains Flat Dam core (Figure 4A, Table S6.) is dependent on ¹³⁷Cs dating, rather than ²¹⁰Pb, it is best analysed in two parts: the pre-1950 and post-1950 parts of the core. The pre-1950 period has an early phase with a Hg concentration of approximately 135 ng/g, followed by an increase in Hg concentration closer to 170 ng/g (Figure 4A, Table S6.). While this increase occurs in uncertain chronology (at the depth of 15 cm), it does predate 1950 and can be surmised to correlate to mining activity during the Lake George Mine’s second operational period, particularly the collapses of tailings dumps that occurred in the early 1940s (Brooks, 1998). Mercury concentrations in the post-1950 period decrease slightly to an approximate average of 160 ng/g. This decrease is likely an effect of the Lake George Mine’s closure, especially closer to the surface of the core.

The sediment core collected at Lake Burley Griffin was successfully dated back to the construction of the lake in 1960. Unsupported PB-210 activities below this depth were not considered in the age-depth model, as they no longer exhibited a decreasing profile with depth (Figure S1b.) The Hg concentration/depth profile in the Lake Burley Griffin core (Figure 4B, Table S7.) shows concentrations increasing prior to the construction of the lake, which is likely linked to the mining activity in Captains Flat. The largest peak in this core (55 ng/g) can be seen to occur around the late 1960s, likely a result of mercury leaching from the tailings dumps after the closure of the mine in 1962.

This leaching prompted the New South Wales government to undertake an AUD2.5-million remediation effort in the mid-seventies. This remediation effort included stabilisation of mine waste dumps to prevent ongoing erosion and leaching to minimise the risk of major collapse, as well as drainage modifications to a local creek to minimise the inflow of water into the mine. The subsequent decrease in Hg concentration in sediments of Lake Burley Griffin is likely a result of these remediation efforts, as well as the lack of continued direct input from mine operations.

In this study, the site MR06, 7.5-km upriver of Captains Flat, was used as reference for background concentration. Mercury concentration for this site was analysed as 24 ng/g. The mean concentration of the samples taken in the immediate area around Captains Flat (sites MR03 – MR05, MR07 and MR08) had a mean Hg concentration of 646 ng/g ± 345 ng/g.

The highest Hg concentrations in this study were recorded in sediments collected at the Lake George Mine (1021 ng/g; site MR03, Table 2) and at a significant source of Hg leaching into the river through an old adit (1065 ng/g; Site MR05, Table 2). While originally constructed to supply air to miners, the adits now act as drainage for rainwater and groundwater flowing through the mine and hence represent point sources for contaminants.

Along the Molonglo River, Hg concentrations decrease relatively quickly with distance. Sediment Hg concentrations at site MR12 (Figure 1 and Table 2, 14-km downstream of the mine) were found to be 124 ng/g (one fifth of the readings taken in Captains Flat township site MR07, 665 ng/g). Site MR 13, another 18-km downstream of site MR12, is lower still at 37 ng/g (MR13). These lower values are likely driven by the physical and chemical attributes of the river system. Physically, the moderate temperatures and rainfall at Captains Flat (Bureau of Meterology, 2019) result in low flows with relatively little transport capacity for much of the year. Additionally, some areas of the river system, act as deposition zones for sediment transported from the upper Molonglo River. Deposition of material in sites, such as the Carwoola Flats (upstream of Queanbeyan), is likely play a part in reducing the total load of Hg contaminated material transported down the river (Wadige et al., 2016).

Chemically, the abundance of sulfide and Fe oxyhydroxides in the mine site’s tailings deposits restrict the transport of Hg through adsorption. Sulfides, oxyhydroxides and organic matter are well documented as important sinks for Hg and other metals and contribute to the permanent burial of trace elements (Nimick and Moore, 1991; Wallschläger et al., 1998). This capacity to remove Hg from transport is likely an important mechanism to maintain Lake Burley Griffin within safe levels of contamination. The Queanbeyan River (MR15) sediment, a possible input source for the Lake, was determined to have a concentration of 21 ng/g, lower than any site on the main course of the Molonglo River. Consequently, the Queanbeyan River system was determined to be unlikely to be an additional source of Hg for the lake.

Sample MR16, located at the site immediately upstream of Lake Burley Griffin, had a concentration of 12 ng/g, half of the upstream reference site MR06 (24 ng/g). This result reflects the site currently being used as an unsanctioned dumping ground for concrete and other construction materials. This sample, likely to represent concentrations of external debris, was considered an outlier and removed from interpretations in this study.

The sediments in Lake Burley Griffin had a wide range of Hg concentrations, with the lake’s maximum Hg value occurring in the Central Basin (74 ng/g, site LBG12, Figure 2B, Table 2) and the minimum Hg value in the West Basin (6.6 ng/g, site LBG24, Figure 2B, Table 2). The East Basin, where the Molonglo River enters the lake (Figure 2B), had the highest Hg concentrations, with a mean of 44 ± 18 ng/g (sites LBG1 to LBG6). The western stretch of the lake (sites LBG14 to LBG25), the furthest from the Molonglo River entrance into the lake, had a lower mean Hg concentration of 20 ± 13 ng/g. The Central Basin (LBG7 to LBG13) registered a mean of 44 ± 15 ng/g.

Similar to sample MR16, both LBG23 and LBG24 demonstrate unexpectedly low values. These values are likely the result of dilution induced by numerous stormwater drainage and overland flow inputs into the lake, or from the sites being outside of the channel of enhanced depth present in the West Basin’s bathymetry (National Capital Authority, 2015). There is no clear explanation for these values without a more detailed investigation.

Mercury concentrations in these sediments demonstrate that Hg has been deposited along the river, resulting in lower Hg contamination in Lake Burley Griffin than in the Molonglo River. Lake Burley Griffin’s East Basin, at the entrance of the Molonglo River, has noticeably higher Hg concentrations relative to the West Basin (Table 2).

The two minor streams flowing into Lake Burley Griffin – Sullivans and Jerrabomberra Creeks – were both found to be unlikely to meaningfully influence Hg in the lake, given their relative concentrations. Sullivans Creek (LBG15) had an Hg concentration of 38 ng/g, exceeded by nearby values, and the Jerrabomberra Creek’s (LBG03) concentration of 19 ng/g is 20 ng/g less than the next lowest concentration in the East Basin.

In the Molonglo River, distance from Lake George Mine was the main predictor of Hg concentrations in sediments, followed by TOM (Table 3, Figure S2.). This indicates the primary drivers of Hg distribution to be fluvial transport of Hg sorted to sedimentary organic matter in runoff from heavy rainfall and its subsequent deposition along the river.

For Lake Burley Griffin, distance from where the river enters the lake was the main predictor of Hg concentrations in sediments, followed by Fe concentration (Table 3, Figure S3.). This indicates that Lake Burley Griffin is still affected by sediment transport from the Lake George Mine, which is to be expected given its position in the Molonglo River system.

The importance of distance as the main predictor of Hg distribution in sediments of both Captains Flat and Lake Burley Griffin can be further described in the PCA plots (Figure 5A and B) and spatially using GIS mapping (Figure 6). For Captains Flat, surface sediment samples within or near mining areas are at the high Hg concentration zone of Axis 1 (PC1) along with Fe and organic matter (Figure 5A). Samples at the low Hg concentration zone of Axis 1 (PC1) are the ones with longest distance from the mining site. Samples MR01, MR02 and MR06 were removed from statistical analyses as they are upstream of the mining site, hence not contaminated, and would therefore bias analysis results if included.

For Lake Burley Griffin, samples near the entrance of the Molonglo River are at the high Hg concentration zone of Axis 1 (PC1) (Figure 5B), while samples far from the entrance of the Molonglo River are at the low Hg concentration zone of Axis 1 (PC1).

The high correlation in both PCAs between concentration and distance strongly suggests that the Lake George Mine is the primary source of contamination. Aside from the mine itself, much of the landscape surrounding Captains Flat is forested or pastoral and hence not considered a significant source of Hg. Likewise, the strong correlation between distance and Lake Burley Griffin suggests that landscape uses are unlikely to be a notable source of Hg.

In Australia, the Australian and New Zealand Sediment Quality Guidelines (ANZG, 2018) outline the criteria for assessing site contamination and determining the requirements for further investigation, or assessment of risk to determine if any further action is required. This document advises further investigation if sediment metal concentrations are above the default guidelines value (DGV) of 150 ng/g and guidelines value- high (GV-high) level of 1000 ng/g Hg.

In the Molonglo River, most samples near the mine sites are above the DGV threshold and two samples are above the GV-high high threshold (Table 4, Figure 7). Of greatest consideration is the Captains Flat Dam which supplies water to the town. At the dam, surface sediment concentration had a Hg concentration of 157 ng/g, which exceeds the DGV sediment quality guideline by 7ng/g (Table 4, Figure 7) (ANZG, 2018).

The New South Wales Department of Trade and Investment, Resources and Energy began a Derelict Mines Program in 1974, with the aim of rehabilitating “abandoned mines”. According to the Strategic Framework for Managing Abandoned Mines (MCMPR and MCA, 2010), “Abandoned mines” are defined as “mines where mining leases or titles no-longer exist, and responsibility for rehabilitation cannot be allocated to any individual, company or organisation responsible for the original mining activities” (p 6). The Lake George Mine at Captains Flat lies within this definition of an “abandoned mine”.

Currently, information on biodiversity, ecosystem integrity, public health and safety remains elusive for many of the abandoned mines across Australia. Legacy and abandoned mines are a state and territory responsibility with little national coordination and leadership (Pepper et al., 2014). The Australian states and territories have differing views about mining legacies, with different solutions, funding arrangements and government agencies to manage the mining legacies issue.

Despite this, the New South Wales government has taken responsibility for the Lake George Mine and undertaken remedial action. This action has been quite substantial, from a AUD2.5-million operation stabilising the tailings dumps in 1976 to another AUD 1.01 million spent on a variety of works from 1993 through to 2015 (NSW Department of Resources and Geoscience, 2019). As the analysis of sediment cores shows (Figure 4), these rehabilitation efforts resulted in significant decreases in Hg released in Lake Burley Griffin. Given the rapidly expanding scale of mining in Australia, the limited study and environmental assessment of abandoned mines hinders both an understanding of the associated environmental risks and impacts and the increased capacity to rehabilitate modern mines. The results from the rehabilitation of the Lake George Mine show that – even 20 years after the fact – rehabilitation and intervention can still have an impact in reducing the potential environmental contamination from legacy mine sites.

Outside of Captains Flat, the results from Lake Burley Griffin show the Lake’s East Basin should be the focus of further Hg targeted studies (Figure 6). Hg concentrations should not be considered a past problem “sunk in the lake”, as recent building development and dredging activities in the Kingston Foreshore region has demonstrated through the ²¹⁰Pb analyses that sediments in the Lake can be remobilised. Remobilising historical Hg deposited on the lakebed can expose organisms to the risks of uptake and bioaccumulation. The remobilisation of sediment has been demonstrated by the three cores collected in this basin; all three had mixed sediments that impeded temporal analyses in this basin. Given this remobilisation, as well as the potential bioavailability of Hg, the Kingston Foreshore Redevelopment area would be an important area to direct future studies on metals in Lake Burley Griffin’s East Basin.

One of the most important considerations of Hg contamination is the possible impact on human health. From the results in this study, the Hg concentrations present 14-km downstream of the mining site, in both the Molonglo River and Lake Burley Griffin, are unlikely to pose a risk to local residents. Much of the waterways around Captains Flat, however – especially including the township’s dam – surpasses the ANZG (2018) DGV. Further study is necessary to understand the level of Hg exposure to residents in that area. Additionally, there may be small-scale ecosystem impacts as a result of metal contamination. Given that several of the mine-adjacent sites (MR03, MR04 and MR05 in particular) reported water pH values around 2 and featured an Fe oxide algal sludge, effects of mine leachate on the surrounding area appears to be a worthy focus of future studies.

The environmental contamination impacts of legacy mining at Captains Flat are clearly still present along the Molonglo River System to this day. While the Hg concentration localised around Captains Flat still exceeds safe levels, Lake Burley Griffin appears to be relatively unaffected by contamination from the mine site. This is likely due to a combination of the government’s rehabilitation efforts and the physical and chemical factors of the Molonglo River – including its flow regime and the occurrence of sedimentary organic matter and Fe oxyhydroxides around the mine site – enhancing deposition rates and hence reducing the amount of Hg transported to Lake Burley Griffin.

Although the river sediment around the Lake George Mine are still concerningly high in Hg contamination, the New South Wales government’s efforts at rehabilitation have clearly reduced the impact of Hg leaching from the mine site. Given the number of legacy mines around Australia, this is an indicator that rehabilitation of abandoned mines is possible and can have significant positive results.

The following datasets were generated for this study:

• – surficial sample mercury and iron concentrations, total organic matter (Table 2)

• – sediment core concentration/depth profiles for Captains Flat Dam (Table S6) and Lake Burley Griffin (Table S7)

These data are uploaded as online supporting information as part of this manuscript.

Australian Institute of Nuclear Science and Engineering provided funding for sediment dating Award-R2 2015, granted to Larissa Schneider.

Mercury analyses were funded by a The Discovery Early Career Researcher Award (DE180100573), awarded to Larissa Schneider.

The authors have no competing interests to declare.

• Conception and Research Design: DS, LS, SB, SH, JS

• Acquisition of Data: DS, LS, SB, JS, SH, OF

• Analysis and Interpretation of Data: DS, LS, AZ, OF, SH

• Drafting and Revising Article: DS, LS, SB, AZ, OF, BM

• Approval of Final Version: LS, SB, BM, OF, AZ