Geochemistry, spatial distribution, and sources of trace element pollution in the surface sediments of Port Hacking, southern Sydney, Australia

,


Introduction
Aquatic coastal environments are vulnerable ecosystems, widely distributed along the eastern seaboard of Australia. Assessment of such aquatic environments indicates that most contamination arises from anthropogenic activities. Sediments are important for the transportation and accumulation of trace elements emanating from both natural and anthropogenic activities (Morelli et al. 2012;Zhao et al. 2019). Sediment has the potential to affect water quality because it is directly connected to the water column. Moreover, polluted sediments are considered to provide a long-term record of trace element dispersion, even when water contamination has declined. In other words, sediments provide time-integrated mean pollutant concentrations compared to more rapidly variable concentrations in the overlying water column (Mao et al. 2017;Aghadadashi et al. 2019).
Over the past century, numerous anthropogenic sources have discharged into coastal and marine aquatic systems, such as bays, estuaries, lagoons, rivers, and lakes. These sources include discharge from urban development, sewage, industrialisation, municipal wastewater, domestic waste, waste incineration, oil pollution, agricultural pollutants, consumption of fossil fuels, mining, and atmospheric emissions from smelters (Morelli et al. 2012;Yolcubal et al. 2016;Dijkstra et al. 2017;Mandaric et al. 2017;Sarı et al. 2018;El-Alfy et al. 2019). High concentrations of trace elements, such as lead (Pb), cadmium (Cd), arsenic (As), nickel (Ni), copper (Cu), mercury (Hg), and zinc (Zn), are considered to be detrimental to the diversity of marine life (Tarique et al. 2012;Zhang et al. 2012). Thus, high trace element concentrations have become a critical problem in many aquatic ecosystems around the world.
Trace elements are dispersed in aquatic habitats and then deposited in aqueous environments in combination with sediments and soils by various processes such as adsorption, absorption, ion exchange, metal substitution, and dissolution (Shahidul Islam and Tanaka 2004;Abrahim and Parker 2008;Turner 2010). Moreover, mud deposits with high percentages of clay minerals are considered to be a significant sink for the accumulation of aquatic trace elements. This is because clay minerals have large surface areas since they occur as very fine particles (<2 μm), they have variable crystal structures (tetrahedral and octahedral), and variable aluminosilicate compositions that commonly results in them being negatively charged at the end of the silicon-oxygen chains. Thus positively charged trace elements can enter the crystal structure and (or) be absorbed onto their surface coating leading to the accumulation of trace elements (Singh et al. 2005;Mendiguchía et al. 2006;Kalo et al. 2013;Alyazichi et al. 2017).
Trace element concentrations in sediments and soils can also play a critical role in the contamination of aquatic environments. This is due to the toxicity, persistence, slow degradation, and rapid accumulation of trace elements (Dural et al. 2007;Hu et al. 2011). These trace elements can be released again into the water column from sediments and soils as free ions and (or) complex compounds. This release can result from influx of new sediment, physical disturbance, chemical diagenetic processes such as changes in pH and redox potential (Eh), bioturbation, and organic degradation (Schulz-Zunkel et al. 2015;Cantera et al. 2018;Ho et al. 2019). Increasing concentrations of trace elements can be deleterious for marine flora and fauna, such as surface dwelling organisms, fish, and microorganisms like foraminifera and ostracoda, that may experience reduced growth because of impaired reproduction, decline in species diversity, and death (Li et al. 2015;Xi et al. 2017). Trace elements may also enter into human bodies through the food chain, with high accumulations resulting in serious health problems such as brain damage, weakness, and various other illnesses (Olawoyin et al. 2012;Zhao et al. 2012;Alves et al. 2013;Huang et al. 2013).
The purpose of the present study is to examine the spatial distribution of total trace element pollution in sediments in the Port Hacking estuary, to determine the status of the environmental ecosystems in the estuary, and to define areas that have high contamination and require further investigation to determine the bioavailability of the trace elements.

Study area
The Port Hacking estuary is located approximately 30 km south of the central business district of Sydney (33°55′S, 151°11′E; Fig. 1). It was scoured into Triassic strata of the Sydney Basin during the Last Glacial Maximum (LGM) low stand of sea level. The head of Port Hacking begins in the tidal upper regions of the Hacking River and it extends about 27 km until it joins the main Port Hacking estuary. Tides within the main estuary are strong, with the average tidal range in Port Hacking being 1 m leading to a substantial exchange of water with the Tasman Sea (Smith et al. 1990). As a result, sediments within the lower estuary tend to be clean, well sorted sand locally forming flood tide deltas (Fig. 1). The estuary covers approximately 11 km 2 with a catchment area of about 208 km 2 (Sherwin and Holmes 1986), approximately 60% of which is covered by natural bushland. The Royal National Park occupies the southern and south-western parts of the catchment, although it does have construction of marine craft in the lower reaches of the Hacking River and small fringing urban settlements along the southern shore of the estuary. Surface runoff has continued to bring sediment into the upper estuary and this has created fluvial deltas at the mouth of the Hacking River and at the heads of the elongate side bays. In these side bays, small amounts of sediment (especially silt and clay) have been deposited in the mud basins between the small fluvial deltas and the sandy flood tide delta at the entrance of the bay. Therefore, some of the bays have remained almost at their original LGM depths of up to 21 m (Albani and Cotis 2004), whereas the river sites are shallow (0.5-2.9 m). Consequently, these deeper bays form effective sediment traps for fine and very fine particles and organic detritus that are discharged from the rivers and creeks.
This study investigated the main inner bays, some shallow rivers and creeks, and other selected sub-environments in the Port Hacking estuary including Gunnamatta Bay, Burraneer Bay, Yowie Bay, Gymea Bay, Mansion Bay, Hacking River, North West Arm, and South West Arm (Fig. 1). Sampling was not undertaken in the outer part of the estuary where clean sand forms the dominant substrate.

Sample collection and analyses
Surface sediment samples (233) were collected using a grab sampler in water depths of 0.5-21 m (Fig. 1). In each case, the top 0-5 cm of the subaqueous estuarine sediment was sampled. A variety of sites within each bay were sampled to obtain a realistic distribution of the sediments and trace elements in the bays and the river delta areas. Additional samples were collected in the vicinity of stormwater channel discharge points, to provide details of possible sources of the trace elements (Supplementary Tables S1 and S2 1 ).
Laser particle size analysis (Mastersizer 2000), following 2 min of ultrasonic disaggregation, was carried out on all the sediment samples. The relative measurement error for grain size parameters was within 3%. The mineralogical composition of each sample was determined by X-ray diffraction using a Philips (PW 1771/100) goniometer, Spellman DF3 controller, and Cu Kα radiation at 1 kW, followed by analysis using Traces and Siroquant software. Trace element concentrations were determined by X-ray florescence analysis using the method of Norrish and Chappell (1977). These analyses used energy dispersive X-ray fluorescence spectrometry (XRF) on a Spectro Xepos instrument. International standards were used to calibrate concentrations of trace elements (accuracy and precision) in the study samples (Table 1). Detection limits for the trace elements ranged between 0.5 and 4 mg/kg, but vary according to the matrix of the samples. All these analyses were conducted at the School of Earth, Atmospheric and Life Sciences in the University of Wollongong, Australia.
Pearson bivariate correlation coefficients and their significance levels were calculated on log-transformed data for the trace elements, percentages of sand, silt and clay, and water depths from all 233 samples.
ArcGIS desktop software (version 10.4.1) was used to create boundaries around the collected sediment samples in each bay. A buffer was created around the samples, and it was clipped to the land boundary where the layers intersected. This was completed for each of the variables analysed, e.g., water depth, sediment characteristics, and trace element concentrations (Li and Heap 2008). Spline with Barriers was chosen as the interpolation method for this analysis due to the ability to input a barrier feature when interpolating a raster surface from points using a minimum curvature spline technique (Li et al. 2018).

Sediment grain size
Most of the surface sediment samples from Port Hacking are dominated by fine-to medium-grained sand and silty sand (Figs. 2 and 3a). As shown in Fig. 3b, coarse-grained Silt and clay percentages Alyazichi et al. 5 sand is located in the main estuary channel, the rivers and deltas, the bay mouths, at stormwater discharge points, and in patches along the bay shorelines. Mud and very fine-grained sediments are concentrated in the inner and middle parts of the bays. The higher wave energy conditions near the shoreline resuspend and transport most fine material from the shoreline towards these deeper areas (Fig. 3c).
The percentages of sand in Port Hacking embayments and rivers varied from 0.87% to 100% with an average of 72.1% (Fig. 2). High percentages of sand were concentrated around the shorelines and at the mouths of bays such as Burraneer Bay and Gunnamatta Bay. This was because these sites have sand barriers produced by the flood tidal delta at the mouths of the bays where water depths are shallow (<2 m, Fig. 3d) and waves and currents are more active. In addition, the Hacking River and South West Arm also have high percentages of sand (Fig. 3b) because they have shallow water depths (Fig. 3d) and are periodically scoured by fluvial currents during flood events. Some of the river deltas and stormwater outlets also have coarse grains derived from the Hawkesbury Sandstone and road materials. The current velocities at discharge points and tidal flows both have the capability to transport fine particles. As a result, the fine to very fine particles, including associated trace elements, can be disturbed and moved from edges and shorelines to the middle of the bays where they are gradually deposited (Mantovanelli et al. 2004;Shi et al. 2010; Fig. 3c). As illustrated in Fig. 3c, silt-and clay-sized particles predominantly accumulated in the inner and middle parts of the bays where water depths were >7 m (Figs. 3a and 3d). Also, any waves and currents become less active in the deeper areas compared to the edges and mouths of the bays (Gong et al. 2014;Alyazichi et al. 2015).

Spatial distribution of trace elements
All the bays analysed in this study, apart from the Port Hacking River, have local drainage from local catchments in the Hawkesbury Sandstonea unit composed of quartz and minor kaolinite that has extremely low trace element concentrations ( Table 2). The Port Hacking River mainly drains the Hawkesbury Sandstone but also cuts down into the Narrabeen Group within the Royal National Park. As see in Fig. 4, it is also characterised by very low (background) trace element concentrations. No data of actual contributions from individual sources are available for the area. Conventional methods look at the residual product in the sampled sediments compared to the natural background levels in adjacent areas relatively unaffected by human activity. Since the trace element concentrations analysed in this study are all derived from local sources rather than regional contributions, and the aim of this study was to show the actual distribution of the trace elements, there was no point normalising the concentrations based on grain size or conservative elements (cf. Sun et al. 2018).
The spatial distributions of analysed trace elements in Port Hacking all show similar patterns of concentration in the inner parts of bays and close to river deltas and direct stormwater outlets, as in Gunnamatta Bay ( Fig. 4; Table 2). The higher concentrations of trace elements in these locations are because they are surrounded by residential areas with stormwater runoff. Trace elements may also be derived from the many boats and boatyards in the northern bays. The stormwater and local river discharge from the residential areas have brought in trace element pollutants mainly from iron roofs, copper water pipes, vehicles and vehicle exhausts, and historical discharge from sewerage. The trace elements have accumulated in areas with high concentrations of fine to very fine particles (80%-100% silt and clay; Fig. 3c) at depths >5 m (Fig. 3d) where waves, currents, and tidal activity do not The concentration of rubidium (Rb) in the muddy embayments also increased (Fig. 5b) and provides a good geochemical proxy for clay minerals such as kaolinite, chlorite, illite, and mixed-layers illite-smectite (Ackermann 1980;Clark et al. 2000). In the Port Hacking area, kaolinite and illite are the dominant clay minerals. Furthermore, organic matter is also concentrated in the inner parts of the bays and this leads to high bromine (Br, Fig. 5a) concentrations (Förstner 1989), ranging between 100 and 500 mg/kg at most sites in these areas. The common presence of clay minerals and organic matter has resulted in greater amounts of trace elements accumulating at these sites through processes such as adsorption and ion exchange (Zhu et al. 2012;Yang et al. 2017).
All the presented trace elements show strong positive correlations that are significant at the 99% confidence level. The trace elements also show a significant positive relationship with the fine grain size fractions (silt and clay percentagesmuddy particles) and negative relationships with mean grain size and of sand percentages, indicating that grain size is one of the main factors controlling trace element distributions (Table 3). Other factors include the amount of organic matter and the proximity to discharge points. This is illustrated in Fig. 6 where the best correlation of lead (Pb) is with silt + clay but considerable scatter remains, even when considering the data from a single bay. Values above the average trend are typically organic-rich samples from creek deltas or near stormwater discharge points, while low values are associated with shallow samples subject to current or wave reworking. The trace elements, silt, and clay are also positively correlated with water depth showing the preferential occurrence of all these components in the low-energy deeper parts of the bays. The negative correlation with sand is also clearly illustrated by the cluster analysis (Fig. 7).
The lowest concentrations of trace elements occurred in the South West Arm, Hacking River, and at the mouths of the bays and represent much lower contamination compared with other sites in the Port Hacking area. This is because these areas are dominated by coarser sand size sediments (>75% sand; Fig. 3b). Also it is partly due to a lack of moored watercraft, residential areas, or municipal waste discharge points in these areas (Figs. 4a-4e). Moreover, fluvial and tidal currents and waves are more active in the shallow water areas at the bay entrances and rivers where the sediments are reworked. In these situations, the muddy particles are disturbed and moved by the waves and currents to be  (Table 2). This is because much of the catchment at Port Hacking drains from national park areas free from residential development and stormwater discharge. Port Hacking also has higher percentages of sand and low percentages of mud (silt and clay) than in similar embayments in the Georges River (Table 4; Alyazichi et al. 2017).
Approximately 95% of the total pollutant discharge into the Georges River (PAB 1992) comes from 786 discharge points, marine moorings, and stormwater outlets. Point sources in the Georges River also include waste dumps and large numbers of watercraft and boatyards (Alyazichi et al. 2017). Natural runoff provides a minor amount of non-point source pollution. George River sediments show a higher percentage of clay minerals (23%) and have lower percentages of quartz and carbonate minerals in comparison with those at Port Hacking (Table 4; Fig. 8). Clay minerals such as kaolinite, chlorite, illite, and illite smectite, as well as pyrite (Fig. 8), can play an important role in the accumulation of trace elements, especially under anoxic conditions that favour trapping of trace elements (Alyazichi et al. 2017;Houben et al. 2017;Jones et al. 2019). The trace elements can be incorporated into surface layers via ion exchange, absorption, and adsorption of these minerals (Gunawardana et al. 2014;Alyazichi et al. 2015). The trace element components in Port Hacking were also compared with the Australian and New Zealand interim sediment quality guidelines (ANZECC/ARMCANZ 2000; Simpson et al. 2013) in Table 5 to assess the sediment quality and its ability to support healthy benthic ecological communities. According to the protocols of ANZECC/ARMCANZ (2000) and Simpson et al. (2013), the anthropogenic pollution of trace elements such as Cr, Ni, Cu, Zn, As, Co, and Pb within some of the sediment sites in the study area were above the interim sediment quality guideline ISQG-low, but the concentrations were below the ISQG-high guideline. Only one sediment site, located within Gunnamatta Bay in the Port Hacking estuary, had Cu and Zn concentrations above the ISQG-high guideline (Figs. 4c and 4d;398 and 417 mg/kg,respectively). This site is located close to mooring areas where there are large numbers of moored watercraft with potential contaminant spills and where leaching, anti-fouling, and fuel spillages add to the pollutant sources. Moreover, slipways, moored boats, as well as settled aerosol dust, galvanised iron roofs, and plumbing, are  considered to be sources of both zinc and copper pollution (Fujita et al. 2014;Huang et al. 2014;Song et al. 2014).
The trace element concentrations in the Port Hacking study area were also compared with selected polluted and non-polluted estuaries in Australia. The results have been summarised in Table 5. The mean values of trace element pollution for Cu, Zn, and Pb in the surface sediments of the study area had lower concentrations compared with Sydney Harbour, Port Kembla, and the Derwent estuary (Table 5). This is because the surface sediments in Sydney Harbour (Irvine and Birch 1998), Derwent River (Jones et al. 2003a), and Port Kembla (Jones et al. 2019) have been subject to large amounts of contamination from urban, industrial, and commercial activities. These include zinc and copper refineries, stormwater runoff, mine discharges, sewage effluent, and watercraft, as well as leaching from reclamation areas. Secondly, the analysed surface sediments in these harbours and ports were dominated by fine to medium silt and clay fractions, along with high percentages of organic matter, that retain trace elements. In contrast, even the muddy surface sediments in Port Hacking are much less contaminated because more than half of its catchment consists of national park and the remainder just has residential development along its shoreline.
A more constructive comparison of trace element concentrations is with the estuarine intermittently closed and open lakes and lagoons (ICOLLs) farther south in the Sydney Basin where their source areas are still usually dominated by Permian quartz-rich sandstone and siltstone strata. Examples such as Burrill Lake (Jones et al. 2003b), Lake Conjola (Packwood 1999), and the background sediments in Lake Illawarra before European settlement (Payne et al. 1997;Gillis and Birch 2006). While the catchment of Lake Illawarra has now been extensively developed for agriculture, urban settlement, and industry, the deeper   sediments in the lake record the natural trace element concentrations when most of the catchment was forested. Burrill Lake and Lake Conjola have only minor residential development around small parts of their margins and their catchments are mainly forested with some minor clearing for agriculture. These coastal lagoons have very similar trace element concentrations as those recorded in the Port Hacking estuary (see data from Burrill Lake in Table 5). The slightly higher Pb and Zn concentrations in Port Hacking estuary probably reflect the greater proportion of residential development along the northern margin of the estuary. Overall, the Port Hacking estuary can be considered as a relatively unpolluted estuary on the southern fringe of the Sydney metropolitan area.

Conclusions
Sediments in marine and coastal environments are sensitive indicators for spatial monitoring of trace metal pollutants. Most of the catchment area around the Port Hacking estuary consists of national park with limited urban development with only a few drainage and stormwater outlets along the northern shore. In contrast, Georges River contains higher concentrations of trace elements than Port Hacking because the catchment areas of Georges River are predominantly urbanised with some light industry and many discharge points and stormwater outlets. In both areas, concentrations of environmentally significant trace elements were found to be higher close to contamination sources such as discharge points and stormwater outlets, and they declined markedly towards the shoreline and mouths of bays and rivers where the percentages of sand particles are higher. This is due to concentrations of fine-and very fine-grained sediments, including clays and organic matter, in deeper areas where trace elements can accumulate by absorption and ion exchange. The Port Hacking estuary has trace element concentrations that are much more similar to those in relatively pristine coastal lagoons such as Burrill Lake. Thus, overall, the Port Hacking estuary can be considered as a relatively unpolluted estuary on the southern fringe of the Sydney metropolitan area.