Environmental Toxicology and Chemistry, Vol. 34, No. 3, pp. 608–617, 2015 # 2014 SETAC Printed in the USA
MODELING FOOD WEB STRUCTURE AND SELENIUM BIOMAGNIFICATION IN LAKE MACQUARIE, NEW SOUTH WALES, AUSTRALIA, USING STABLE CARBON AND NITROGEN ISOTOPES LARISSA SCHNEIDER,*yz WILLIAM A. MAHER,y JAIMIE POTTS,yx ANNE M. TAYLOR,y GRAEME E. BATLEY,k FRANK KRIKOWA,y ANTHONY A. CHARITON,y# and BERND GRUBERy
yInstitute for Applied Ecology, University of Canberra, Canberra, Australian Capital Territory, Australia zArchaeology and Natural History, College of Asia and the Paciﬁc, The Australian National University, Canberra, Australian Capital Territory, Australia xNew South Wales Ofﬁce of Environment and Heritage, Lidcombe, New South Wales, Australia kCSIRO Land and Water, Lucas Heights, New South Wales, Australia #CSIRO Oceans and Atmosphere, Lucas Heights, New South Wales, Australia (Submitted 29 July 2014; Returned for Revision 2 December 2014; Accepted 3 December 2014) Abstract: As a consequence of coal-ﬁred power station operations, elevated selenium concentrations have been reported in the sediments and biota of Lake Macquarie (New South Wales, Australia). In the present study, an ecosystem-scale model has been applied to determine how selenium in a seagrass food web is processed from sediments and water through diet to predators, using stable isotopes (d13C and d15N) to establish the trophic position of organisms. Trophic position, habitat, and feeding zone were examined as possible factors inﬂuencing selenium bioaccumulation. Selenium concentrations ranged from 0.2 mg/g dry weight in macroalgae species to 12.9 mg/g in the carnivorous ﬁsh Gerres subfasciatus. A mean magniﬁcation factor of 1.39 per trophic level showed that selenium is biomagnifying in the seagrass food web. Habitat and feeding zone inﬂuenced selenium concentrations in invertebrates, whereas feeding zone was the only signiﬁcant factor inﬂuencing selenium concentrations in ﬁsh. The sediment–water partitioning coefﬁcient (Kd) of 4180 showed that partitioning of selenium entering the lake to particulate organic material (POM) is occurring, and consequently availability to food webs from POM is high. Trophic transfer factors (invertebrate ¼ 1.9; ﬁsh ¼ 1.2) were similar to those reported for other water bodies, showing that input source is not the main determinant of the magnitude of selenium bioaccumulation in a food web, but rather the initial partitioning of selenium into bioavailable POM. Environ Toxicol Chem 2015;34:608–617. # 2014 SETAC Keywords: Ecosystem-scale selenium modeling
The primary mode of selenium assimilation in aquatic food webs is through diet . Plant roots and microorganisms biotransform inorganic selenium into selenomethionine, which is then assimilated by primary consumers and transferred through food webs or converted into selenocysteine . Selenium concentrations usually increase between successive trophic levels [3,5], but not always . The reason for this discrepancy is that selenium enrichment is dependent on both the bioavailability of selenium in food and the assimilation efﬁciency of the consumer . The magnitude of enrichment can differ among species within a given trophic level because of different diets and varying assimilation and elimination efﬁciencies . Other factors that are generally understudied and may signiﬁcantly inﬂuence the intake of selenium include the habitat (sediment or water column) and the feeding zone (benthos or water column) used by organisms [3,7]. These factors may be important in environments with long water residence times where selenium concentrations are higher in the sediments than in the water column. Organisms capable of obtaining selenium directly from sediments via ingestion of associated algae and microorganisms will be more likely than pelagic species to accumulate selenium to high concentrations [8,9]. Organisms living in sediments and feeding on benthic organisms are, therefore, expected to have higher selenium concentrations than pelagic biota. Food web studies improve our understanding of pathways of exposure, giving insights into how selenium moves through aquatic ecosystems, and potentially permit identiﬁcation of
Coal-based power generation produces large amounts of coal ash that contain elevated concentrations of many trace elements that can pose health risks to aquatic ecosystems . Coal combustion during power generation is the major anthropogenic source of selenium to the environment, either directly during combustion as ﬂy ash or indirectly from leaching of coal ash stored in ash dams . In Australia, selenium is an element of major concern in aquatic systems . Although it is an essential micronutrient in animals, it is toxic at elevated concentrations . Toxicity may arise through multiple biochemical pathways, but most often manifests as reproductive impairment or teratogenicity . Selenium exists in a variety of chemical forms that affect its chemical and biological reactivity and thus its capacity to enter into food webs. In oxygenated waters, dissolved selenium exists as either selenate (Se[VI]) or selenite (Se[IV]), together with minor amounts of organoselenium forms . Selenate and Se(IV) can be found in sediments, as adsorbed or coprecipitated selenite and selenite or as elemental selenium (Se). Once incorporated into cells, selenium is reduced to organic selenides and incorporated mainly as selenium in proteins .
All Supplemental Data may be found in the online version of this article. * Address correspondence to [email protected]
Published online 11 December 2014 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/etc.2847 608
Selenium biomagnification in Lake Macquarie
vulnerable components of the ecosystem. In the present study, the enrichment and trophic transfer factors of selenium were examined in a seagrass ecosystem in Lake Macquarie (New South Wales, Australia). Stable isotopes (d13C and d15N) were used to establish the trophic positions of organisms, and an ecosystem-scale selenium model  was used to fully describe selenium transfer in this system and allow comparisons with other systems. We then investigated the hypothesis that habitat and feeding zone are important factors inﬂuencing selenium bioaccumulation in organisms. MATERIALS AND METHODS
Environ Toxicol Chem 34, 2015
character because of minimal freshwater dilution from the 2 main ﬂuvial inputs . The present study was conducted in the southern part of Lake Macquarie, where 2 of 3 original coal-ﬁred power stations are still in operation: the Vales Point Power Station, which started operations in 1963, and the Eraring Power Station, which started in 1981 . The sampling site for the present study was selected based on a previous study  showing that Morisset Bay, Lake Macquarie, a shallow bay with extensive seagrass meadows, is a deposition zone for sediments containing selenium and other contaminants (Figure 1).
Lake Macquarie is situated approximately 90 km north of Sydney and close to the city of Newcastle (Figure 1). The lake extends approximately 22 km in a north–south direction and has a maximum width of approximately 10 km and a maximum depth of approximately 11 m, with an average depth of 8 m . The lake is separated from the ocean by a narrow entrance channel and sand bars at Swansea. The tidal range in Lake Macquarie is small, with the spring tidal range estimated at 0.15 m at the western end of the tidal channel, decreasing with distance from the entrance to 0.06 m. The lake has a marine
To maximize the suite of organisms collected, sampling was undertaken in early September 2011 (during spring) and in late February 2012 (during summer). A variety of organisms was collected (with 10 replicates where possible) to obtain representation of each trophic level of the seagrass food web, including both benthic and pelagic habitats, classiﬁed as autotrophs, herbivores/ﬁlter feeders, detritivores, and carnivores. Water. Water for selenium analysis was collected after plankton was ﬁltered through a glass-ﬁber ﬁlter (Whatman 1.2 mm pore size). Water was frozen and stored in Falcon 50-mL
Figure 1. Lake Macquarie (New South Wales, Australia). The black circled area is the location of the seagrass ecosystem biota collection site. An X indicates the locations of Eraring and Vales Point Power Stations. Arrows point to the associated ash dams: Whiteheads Lagoon in the north and Mannering Bay in the south.
Environ Toxicol Chem 34, 2015
centrifuge tubes until analysis. Water samples were thawed and diluted 1:10 with deionized water prior to analysis. Plankton. Plankton samples were collected in daylight hours by dragging a net (mesh size 150 mm) approximately 4 m behind a boat at a speed of less than 1 knot for at least 5 min. A weight was placed on the opening of the net to perform horizontal and oblique hauls to collect samples from all depths. Samples were then placed in acid-washed 100-mL screw-cap glass bottles with lake water for 12 h to depurate and then ﬁltered through a glass-ﬁber ﬁlter paper (1.2 mm pore size). Although it was not possible to separate zooplankton from phytoplankton, the d15N signal showed that the sample was predominantly composed of phytoplankton. Detritus and mangroves. Detritus was randomly collected within the seagrass bed by hand, using gloves. Samples were transferred to clean polystyrene sampling vials. Mangrove leaves (Avicennia marina) were obtained from trees along the shore. Macroalgae, seagrasses, and epiphytes. Seagrasses and macroalgae species were collected by hand, using gloves, to a maximum depth of 1 m. Samples were removed from the substrate, rinsed with lake water to remove particulate material, and placed in polyethylene bags. Epiphytes were obtained by scraping seagrass leaves using a stainless-steel blade and transferred to clean polystyrene sample vials. Although epiphytes can be composed of sessile animals, the d15N signal showed that the sample was predominantly composed of plant material. Benthic microalgae. Samples of benthic microalgae were collected by scraping the top layer of sediment from 30-cm cores to separate the surﬁcial layer from woody debris, rocks, and most of the underlying sediment. The samples consisted of cyanobacteria and microalgae in a sediment–water slurry. Invertebrates. Amphipods were collected from frozen detritus using tweezers. Crabs were collected from under rocks. The ghost shrimp (Trypaea australiensis), the emerald shrimp (Metapenaeus insolitus), and a number of polychaete taxa were obtained by using a yabbie pump (a cylindrical metal tube and plunger used by ﬁsherman for collecting burrowing crustaceans) to collect sediment and then sieving through a 5-mm stainless steel sieve. The molluscs Anadara trapezia, Batillaria australis, Paphia undulata, Soletellina alba, Ostrea angasi, Corbula truncata, and a species from the Achidae family were collected by sieving sediment obtained from seagrass beds using a stainless steel shovel. The molluscs Trichomya hirsuta, Saccostrea glomerata, Bembicium auratum, and Nerita sp. were collected by hand using gloves. Fish. Fish samples were collected using seine nets (length, 100 m; drop, 1.5 m; mesh, 5 mm). Nets were set from the bay, extending out at a 908 angle to shore. Juvenile ﬁsh were collected to minimize the inﬂuence of size, age, and change of migration patterns on selenium concentrations (Supplemental Data, Table S1). Fish were euthanized with benzocaine, stored individually in sealed polyethylthene bags, and then chilled on site using ice. The species collected comprised Hyporhamphus regularis, Gerres subfasciatus, Monacanthus chinensis, Mugil cephalus, Girella tricuspidata, Rhabdosargus sarba, Dicotylichthys punctulatus, Pelates sexlineatus, Tylosurus gavialoides, Sillago ciliata, Pomatomus saltatrix, and Sillago maculata. All samples were placed on ice in a cooler for transport to the laboratory and stored at -208C. Sample preparation and analysis
Benthic microalgae were separated from sediments following the method of Hamilton et al. . Sediments were washed
L. Schneider et al.
through a 200-mm mesh to remove infauna, coarse sediments, and detritus. Collected material was placed in a 30-mL centrifuge tube containing 21 mL of colloidal silica (density ¼ 1.3) and 9 mL of deionized water, and then centrifuged at 1100 rpm for 7 min. Material that settled at the silica/deionized water interface was pipetted onto a 47-mm glass-ﬁber ﬁlter (Whatman 1.2 mm pore size), and the process was repeated until the ﬁlter clogged. Benthic microalgae and ﬁlters were weighed into 7-mL polytetraﬂuoroacetate (PFA) tubes, and the ﬁlter mass was subtracted. Two blanks were produced using clean ﬁlters to check for possible cross-contamination in the benthic microalgae. Invertebrates were removed from their shells, and the whole body tissue was used for metal analysis. Total ﬁsh lengths were recorded, and the whole body muscle tissue was removed. Tissues were freeze-dried at –808C for 48 h (Labconco FreeZone plus 6) and ground to a ﬁne powder with an Ika A11 Basic Analytical Mill (Staufen). Ground samples were placed into sterile 30-mL screw-cap vials and stored in a desiccator prior to analysis. Approximately 0.07 g of freeze-dried material was weighed into a 7-mL PFA closed digestion vessel, and 1 mL of concentrated nitric acid (Aristar; BDH) was added. Each 7mL PFA vessel was then capped and tightened to 2.3 Nm, and placed into larger 120-mL screw-top pressure vessels, tightened to 16.3 Nm. A model MDS-81 (CEM; Indian Trail) microwave oven rated at 600 W was used for all digestions, with the microwave procedure consisting of 3 steps: 2 min at 600 W, 2 min at 0 W, and 45 min at 450 W, as described in Schneider et al. . After digestion, the 7-mL PFA vessels were allowed to cool at room temperature for 20 min, and then the contents were diluted to 10 mL with deionized water. Digests were stored in labeled polyethylene vials in a cool room (58C) until analysis. Samples were analyzed for selenium using inductively coupled plasma–mass spectrometry (DRC-e with an AS-90 auto-sampler; PerkinElmer). Selenium concentrations were cross-checked using a PerkinElmer 5100 Zeeman graphite furnace atomic absorption spectrometer with AS-60 autosampler, using a mixture of palladium and magnesium as the matrix modiﬁers for selenium. Samples used for d13C and d15N analysis were weighed into tin capsules and analyzed using an ANCA GSL2 elemental analyzer interfaced with a Hydra 20–22 continuous-ﬂow isotope ratio mass spectrometer (Sercon). The precision was 0.1‰ for d13C and 0.2‰ for d15N (standard deviation [SD], n ¼ 5). Stable isotope data are expressed in the delta notation (d13C and d15N), relative to the stable isotopic ratio of Vienna Pee Dee Belemnite standard (RVPDB ¼ 0.0111797) for carbon and atmospheric nitrogen (RAir ¼ 0.0036765) for nitrogen. Food web structure
Classiﬁcation of samples into trophic groups, habitat, and feeding zones. Carbon and nitrogen isotope results and stomach contents reported in the literature for invertebrates and ﬁsh species were used to classify organisms to appropriate trophic levels (Table 1; Supplemental Data, Table S3). Invertebrates were classiﬁed into 3 categories reﬂecting their niche: pelagic invertebrates feeding from the water column, benthic invertebrates feeding from the water column, and benthic invertebrates feeding on other benthos. Fish were placed into the following categories: pelagic ﬁsh feeding from the water column, pelagic ﬁsh feeding from benthos, and benthic ﬁsh feeding on other benthos. In the present study, no pelagic invertebrates that feed on benthos and no benthic ﬁsh that feed
7 11 2 23 2 10 10 6 10 10 8 10 5 5 2 10 1 4 2 10
Dicotylichthys punctulatus Pelates sexlineatus Tylosurus gavialoides Sillago ciliata Pomatomus saltatrix Sillago maculata
2 10 7 10 7 3 18
5 6 6 10 12 6 5 15 10 12
NI NI Helograpsus haswellianus Metapenaeus insolitus NI Paphia undulata Batillaria australis Soletellina alba Trypaea australiensis Hyporhamphus regularis Gerres subfasciatus Monacanthus chinensis Mugil cephalus Girella tricuspidata Rhabdosargus sarba
Nerita sp. Bembicium auratum Corbula truncata Trichomya hirsuta Ostrea angasi Anadara trapezia Saccostrea glomerata
NI Cystoseira trinodis NI Gracilaria sp. Laurencia sp. Sargassum sp. Chaetomorpha sp. Microdictyon umbilicatum Halophila sp. Avicennia marina Zostera capricornii NI
NI NI NI
3.2 3.0 3.7 3.2 4.0 —
2.10 1.4 1.9 — 2.0 1.4 1.4 1.4 2.1 2.6 3.3 2.7 3.0 2.9 2.9
1.8 1.7 1.3 1.3 2.0 2 1.9
0.3 _ 1.4 _ _ 1.5 _ _ 0.5 1.4 1.0 1.4
— — — — — — — — — — — Water
— — — — — — — — — — — Pelagic
Benthic Benthic Pelagic Benthic Pelagic Benthic
Pelagic Benthic Benthic Pelagic Benthic Benthic Benthic Benthic Benthic Pelagic Pelagic Pelagic Pelagic Pelagic Pelagic Benthos Benthos Water Benthos Water Benthos
Water Benthos Water Benthos Benthos Water Benthos Benthos Benthos Water Benthos Benthos Benthos Benthos Benthos
Benthos Benthos Water Water Water Water Water
— — —
— — —
Benthic Benthic Benthic Benthic Benthic Benthic Benthic
–14.7) –13.7) –21.9) –18.6) –18.7) –19.0) –18.2)
5.4 0.1 (5.4–5.5) 3.9 0.9 (2.5–5.5) 3.6 5.3 1.4 (4.4–7.4) 3.6 1.0 (2.9–4.3) 6.8 1.1 (4.8–8.1)
1.2 0.1 (1.1–1.4) 5.6 2.2 (1.2–8.1) 2.1 0.7 (1.5–3.9) 3.9 6.2 1.8 (1.5–9.0) 3.9 0.3 (3.7–4.2) 5.5 0.6 (4.8–6.4) 7.8 0.9 (6.4–9.4) 2.4 0.2 (2.2–2.8) 1.4 0.4(0.9–2.1) 7.3 2.7 (3.7–12.9) 4.3 1.0 (2.7–5.6) 3.3 0.6 (2.4–4.3) 4.2 3.3 (0.9–8.7) 3.7 1.7 (2.5–6.6)
2.2 2.3 (0.6–3.8) 1.9 0.7 (0.9–3.0) 2.5 0.2 (2.3–3.0) 4.9 1.6 (3.2–8.0) 3.6 1.5 (0.4–4.6) 4.8 2.2 (3.4–7.4) 2.9
1.37 0.3 0.1 (0.3–0.4) 1.1 0.4 (0.6–1.8) 0.2 0.4 (BDL–0.9) 0.2 0.1 (0.1–0.3) 0.7 0.2 (0.5–1.2) 0.3 0.2 (0.2–0.8) 1.6 0.6 (0.6–2.0) 0.4 0.2 (0.1–0.7) 0.3 0.2 (0.1–0.8) 0.7 0.2 (0.2–1.1) 2.2 0.3 (2.0–2.5)
0.1 0.0003 mg/L 1.7 0.2 (1.4 to –2.0) 1.8
Mean Se SD (min–max) (mg/g dry wt)
Environ Toxicol Chem 34, 2015
to to to to to to to
–14.1 0.1 (–14.0 to –14.7 0.7 (–15.7 to –15.5 –14.1 1.0 (–15.0 to –17.7 0.1 (–17.8 to
(–16.4 (–14.5 (–22.8 (–21.4 (–19.3 (–19.6 (–22.8
11.0 1.5 (10.0–12.1) 10.2 0.8 (9.1–11.4) 12.8 11.1 0.3 (10.8–11.4) 13.6 0.3 (13.4–13.8)
–15.4 0.9 –13.9 0.3 –22.4 0.3 –20.4 1.0 –19.1 0.2 –19.3 0.3 –21.2 1.3
–17.5 1.0 (–19.0 to –16.7) –22.3 1.8 (–23.5 to –18.5) –16.3 0.9 (–19.0 to –15.4) — –18.6 3.3 (–30.4 to –15.0) –23.0 0.4 (–23.3 to –22.7) –15.0 0.7 (–15.7 to –14.2) –19.0 0.2 (–19.9 to –19.2) –17.3 0.7 (–17.8 to –16.3) –13.3 0.9 (–14.5 to –11.7) –15.1 0.9 (–16.4 to –13.6) –15.3 1.2 (–17.1 to –14.1) –16.7 2.5 (–19.5 to –12.0) –15.7 1.8 (–17.8 to –13.5) –14.8 1.4 (–15.6 to –2.3)
(5.6–7.3) (5.5–6.4) (4.3–4.7) (3.7–5.6) (6.6–7.3) (6.5–7.2) (5.8–7.5)
6.4 1.2 6.0 0.3 4.5 0.0 4.6 0.7 7.0 0.2 6.9 0.4 6.6 0.5
–13.8 0.1 (–13.8 to –13.7) –27.1 0.1 (–28.5 to –28.3) –10.0 10.0 (–11.0 to –8.1) –22.6 0.7 (–23.1 to –22.1)
7.2 0.2 (6.9–7.3) 4.7 0.3 (4.2–5.0) 6.6 0.8 (5.4–7.7) — 6.9 1.2 (4.9–9.4) 5.0 0.2 (4.9–5.1) 4.8 0.2 (4.5–4.9) 4.7 0.3 (4.1–5.1) 7.3 0.1 (7.1–7.4) 8.9 0.6 (8.3–10.4) 11.2 0.6 (10.4–12.2) 9.1 0.4 (8.7–9.8) 10.4 1.3 (8.6–12.9) 9.9 0.9 (8.9–11) 10.1 0.1 (9.8–10.2)
(1.8–1.9) (4.8–5.3) (1.5–5.1) (4.6–5.0)
1.9 0.0 4.7 5.0 3.7 1.3 4.8 0.3
5.1 0.9 (3.8–6.4) c
–16.6 1.8 (–20.4 to –14.6)
–14.9 3.2 (–17.2 to –9.2)
4.7 0.8 (3.7–5.4) c
— –16 –23.2
Mean d13C SD (min–max.) (‰)
— 3.4 0.6
Mean d15N SD (min–max.) (‰)
Trophic level (TL) determined relative to BMA as the baseline, as described in Equation 1 in Materials and Methods. Composite samples. Insufﬁcient tissue or not analyzed. SOM ¼ suspended organic material; BMA ¼ benthic microalgae; NI ¼ not identiﬁed to species level; BDL ¼ below the detection limit.
Water Detritus SOM Autotrophs BMA Brown alga Epiphytes Gracilaria Red algae Sargassum Spaghetti algae Velley Halophila Grey mangrove Zostera Plankton Herbivores: suspension feeder Grazing snail Grazing snail Basket shell Hairy mussel Southern mud oyster Sydney cockle Sydney rock oyster Detritivores Amphipods Achidae Grapsid crab Greasyback prawn Polychaete Short necked clams Small mud whelk White sunset shell Yabbie Garfish Common silver belly Fan-belly leather jacket Flathead mullet Luderick Tarwhine Carnivores Three-bar porcupine fish Eastern striped trumpeter Long tom Sand whiting Tailor Trumpeter whiting
Table 1. Stable isotopes (d15N and d13C) and selenium in the seagrass ecosystem of Morisset Bay, Lake Macquarie (NSW, Australia)
Selenium biomagnification in Lake Macquarie 611
Environ Toxicol Chem 34, 2015
from the water column were collected. These 2 combinations were therefore excluded from further statistical analysis. Trophic levels and trophic magniﬁcation factor. Trophic levels were determined from stable nitrogen isotope ratios (d15N) calculated from tissue ratios of d15N/d13N in food webs, which represent average food web biomagniﬁcation . Calculation methods are presented in the Supplemental Data. The ability of a contaminant to biomagnify can be expressed in terms of a trophic magniﬁcation factor, where a trophic magniﬁcation factor greater than 1 indicates biomagniﬁcation . The trophic magniﬁcation factor is applied to determine biomagniﬁcation of selenium throughout the food web . These calculations are also presented in the Supplemental Data. Selenium trophic transfer model. The trophic transfer of selenium through the Lake Macquarie food web was examined using a model outlined by Presser and Luoma . This model calculates the environmental partitioning between particulate and dissolved phases (Kd) and assesses the transfer of selenium between trophic levels, the trophic transfer function. In the present study, particulate organic matter (POM) is the base of the food web and consists of benthic microalgae, plankton, detritus, and suspended organic matter (SOM). Seagrass and mangrove leaves were excluded from POM, as d13C analysis showed they did not contribute signiﬁcantly to the POM of the food web (Figure 2). Sediment was also excluded from POM, as few organisms directly ingest sediments, and selenium concentrations of sediment (2 mg/kg dry wt) were similar to POM (x ¼ 1.9 mg/g; range, 1.1– 2.2 mg/kg). Both Kd and trophic transfer function calculations are presented in the Supplemental Data. Statistical analysis
Data were analyzed using the statistical package R  with p 0.05 as the level of statistical signiﬁcance. The assumptions of normality and homogeneity of variances were checked by
L. Schneider et al.
examining plots of residuals. Where data were not normal, they were log-transformed, and parametric tests were used. A one-way analysis of variance (ANOVA) was used to test for a signiﬁcant difference in log-transformed selenium concentrations between trophic levels (source, autotrophs, herbivores and ﬁlter feeders, detritivores, and carnivores). Post hoc Tukey tests were performed to identify trophic groups in which selenium concentration differed signiﬁcantly (p < 0.05). An ANOVA was also performed to check for differences in d15N between trophic levels. A linear regression model was used to determine the relationship between trophic level and d15N for log-transformed selenium concentrations. Prior to these regressions to calculate trophic magniﬁcation factors, the prey–predator relationships of food web organisms were checked using published literature, and any organisms that do not rely on others were removed prior to performing regressions . The effects of habitat and feeding zone on selenium concentrations of invertebrates and ﬁsh were examined using an analysis of covariance (ANCOVA). The logarithm of selenium concentration (log Se) was used as the dependent variable, habitat and feeding zone as predictor variables, and d15N as a continuous covariate. Invertebrates and ﬁsh were analyzed separately because trophic magniﬁcation factors in invertebrates and ﬁsh are, to different degrees, inﬂuenced by their different metabolisms, direct uptake across respiratory surfaces, and the relative food versus sediment versus water exposure . All ANCOVA assumptions were checked prior to analysis: the normality of residuals, homogeneity of variances, homogeneity of regression slopes, linearity of regression, and independence of error terms. The third assumption, concerning the homogeneity of the different treatment regression slopes, is particularly important in evaluating the appropriateness of the ANCOVA model. In the case that this assumption could not be met and the main effects interaction was signiﬁcant, a regression
Figure 2. Seagrass ecosystem food web, Morisset Bay, Lake Macquarie. Dashed lines indicate subjectively identiﬁed general boundaries of carbon sources based on d13C signal (black square ¼ autotroph/source; red circle ¼ herbivore/suspension feeder; green triangle ¼ detritivore; blue square ¼ carnivore). BMA ¼ benthic microalgae; SOM ¼ suspended organic matter.
Selenium biomagnification in Lake Macquarie
Environ Toxicol Chem 34, 2015
of covariate against the dependent variable was run for each of the habitat/feeding zone categories independently to check whether the covariate effect was signiﬁcant. RESULTS AND DISCUSSION
The range of d13C values in the present study (Table 1; Supplemental Data, Table S4) revealed that invertebrates had 3 dietary sources: from benthic microalgae or/and SOM (plankton, C. truncata, Achidae, P. undulata, T. hirsuta, S. glomerata); from detritus (O. angasi, A. trapezia, T. australiensis, amphipods, S. alba, B. australis, Helograpsus haswellianus, and Nerita sp.); and from seagrass-associated components (B. auratum) (Figure 2). This division is, however, subjective, and organisms close to the boundary lines use several food sources. For example, O. angasi and A. trapezia, although classiﬁed in the detritus zone, are ﬁlter feeders and feed on SOM . Because they also ingest detritus particles, however, their d13C signal placed them on the boundary between detritus and SOM/benthic microalgae. Bembicium auratum are located in the seagrass zone, but feed on microalgae on the surface of the hard substrate or the sediment on which they live ; therefore the d13C signal placed this species in the boundary between seagrass and detritus. These results illustrate the importance of considering the diet, habitat, and characteristics of an organism in addition to its d13C signal. Invertebrates had a large range of d13C and d15N, reﬂecting their individual niche (habitat and food sources). Saccostrea glomerata, T. hirsuta, A. trapezia, and O. angasi, for example, are all ﬁlter feeders, but the ﬁrst 2 species listed had lower d13C as a result of ﬁlter feeding in their niche in the water column containing SOM, whereas the other 2 had higher d13C from ﬁlter feeding water from their niche in sediment containing detrital material (Figure 2). All ﬁsh had diets reﬂective of detritus and associated epiphytes (d13C range of –12 to –18; Table 1). Fish had clear values of d15N, with herbivores showing a relatively depleted d15N signature reﬂecting grazing on detritus and algae. Carnivores showed a relatively enriched d15N signature reﬂecting feeding on animals (Figure 2 and Table 2). Mangrove leaves (Avicennia marina) had d13C values (–27.9 0.1) clearly showing that these were not being substantially incorporated into the diet of the organisms in the seagrass food web. Live Zostera capricornii had a d13C of –10.0 0.8 and was not being directly consumed, whereas detritus, which contained mostly decomposing seagrass, had a d13C of –16 and was being consumed (Figure 2).
species to 12.9 mg/g dry weight in the ﬁsh silver belly (G. subfasciatus), with an overall mean among all organisms of 3 2 mg/g. Mean selenium concentrations were different between trophic levels, in the following order: autotrophs < herbivores/suspension feeders < detritivores–omnivores < carnivores (Figure 3). There were signiﬁcant differences in selenium concentrations between the groups (F3,297 ¼ 160.8, p < 0.001), with the autotrophs, herbivores, and suspensionfeeders having signiﬁcantly lower selenium concentrations than detritivores and carnivores (Supplemental Data, Table S5). Selenium tissue concentrations (Table 1) were generally higher than those in the organisms from other Australian estuaries not receiving inputs from coal-ﬁred power stations [8,15,16], and similar to those from Lake Wallace (New South Wales) , an Australian freshwater lake receiving inputs from a coal-ﬁred power station (plankton, 4.3 mg/g; seagrass epiphytes, 1.3 mg/g; benthic microalgae, 8.0 mg/g; detritus, 10.0 mg/g; invertebrates 3.9 mg/g; whole ﬁsh, 7.6 mg/g dry wt). Comparison with the tissue data of Barwick and Maher  conﬁrmed that selenium concentrations have, in general, decreased in food-web elements of Lake Macquarie since 1999, which may be a consequence of the changed ash handling procedures adopted by the Vales Power Station . Although selenium concentrations in Lake Macquarie biota are higher than in biota from other Australian estuaries, they are considerably lower than in organisms from freshwater lakes receiving coal-ﬁred power station inputs in the United States . These results are explained by the lower selenium concentration found in Australian coals than in international coals and the lower water column selenium concentration , making selenium emissions from coal-ﬁred power stations low relative to countries such as the United States and Canada. In the United States, for example, Hyco Lake in North Carolina provides cooling water for a coal-ﬁred power station and receives selenium-laden efﬂuent from an ash settling pond. Selenium concentrations reached 14 mg/L in water, 5 mg/g to 60 mg/g in phytoplankton, 30 mg/g to 88 mg/g in benthic invertebrates, and 22 mg/g to 68 mg/g dry weight in bluegill tissues . Similarly, the Savannah River in South Carolina
Selenium concentrations in water, sediment, detritus, and organisms are shown in Table 1 and Supplemental Data, Table S4. Water concentrations were below 0.1 mg/L. Biota concentrations ranged from 0.2 mg/g dry weight in macroalgal Table 2. Stable isotope (d15N) values per trophic level and increments up the seagrass food chain of Lake Macquarie Trophic level Autotroph Herbivore: Suspension feeder Detritivore Carnivore Average increment
d15N d15N Average Increment min. (%) max. (%) (%) (%) 1.1 4.54 4.73 10.23
5.14 6.96 11.25 13.6
3.12 5.75 7.99 11.92
– 2.63 2.24 3.93 2.93
Figure 3. Selenium concentrations in trophic levels of the seagrass food web of Morisset Bay, Lake Macquarie. Bar whiskers show the 10th and 90th percentiles, the box shows the 25th and 75th percentiles, and the solid line in the box shows the median selenium concentration for each trophic group. The open circles indicate values that are more extreme than 1.5 interquartile range. Letters shared in common between or among the groups indicate no signiﬁcant difference.
Environ Toxicol Chem 34, 2015
receives selenium inputs from a power plant after slurried ﬂy ash is discharged into a series of large settling basins before surface water is expelled into Beaver Creek, a tributary of the Savannah River. This power plant has been operating for more than 50 yr and, as a result, selenium concentrations reached 110 mg/L in portions of the drainage system, 6 mg/g in macrophytes, 7 mg/g in periphyton, 20 mg/g in bivalves, and 15 mg/g dry weight in ﬁsh tissues . In the above examples, selenium contamination resulted in health issues for the food-web organisms. To our knowledge, no acute or chronic effects have been observed or reported for the selenium exposed organisms in Lake Macquarie. Nonetheless, laboratory studies exposing the species A. trapezia and Tellina deltoidalis to sediment selenium doses of 5 mg/g and 20 mg/g dry weight  found that they were able to detoxify only a small percentage of accumulated selenium, which suggests a limited detoxiﬁcation and storage capacity for this element. Trophic level and biomagnification
The linear regression of log selenium concentrations and d15N (slope ¼ 0.14, adjusted R2 ¼ 0.26, F1,233 ¼ 66.94, p < 0.001) was signiﬁcant (Figure 4). Based on the concept that a trophic magniﬁcation factor greater than 1 indicates that biomagniﬁcation is occurring, selenium is being biomagniﬁed through the food web, as an increase of 1.39 times (95% conﬁdence interval 1.29–1.51) in selenium is occurring per trophic level. It is important to note that the slope of the regression is biased by deposit feeders such as polychaetes and the bivalve S. alba, which have higher selenium concentrations than other invertebrates in the same or close trophic levels. Selenium biomagniﬁcation has been reported for many food webs and in laboratory studies [3,17,18]; however, no evidence of selenium biomagniﬁcation was found in other studies such as that of the Savannah River ecosystem . Differences in biomagniﬁcation between systems will be dependent on the initial partitioning of selenium from water and/or sediment to POM and initial consumers, as has been found for mercury bioaccumulation in aquatic ecosystems .
L. Schneider et al.
Relationship between selenium concentration, trophic level, habitat, and feeding zone
Invertebrates. For invertebrates, there was no signiﬁcant relationship between d15N and selenium concentration (F1,128 ¼ 5.34; p > 0.05), because most invertebrates were from a similar trophic level. There was a signiﬁcant effect of habitat and feeding zone on selenium tissue concentration (F2,128 ¼ 24.80; p < 0.001). There was a signiﬁcant difference between selenium tissue concentrations of invertebrates from different habitats (Figure 5). Pelagic invertebrates feeding from the water column had lower selenium concentrations than benthic invertebrates feeding from the water column (p < 0.001), which in turn had lower selenium concentrations than benthic invertebrates feeding on benthos (p < 0.001). These results show that both habitat and feeding zone are important factors inﬂuencing selenium concentrations in invertebrates (Figure 5). Benthic invertebrates with close association with sediments had signiﬁcantly higher selenium concentrations than pelagic invertebrates. Ingestion of sediment-associated algae and bacteria by invertebrates will be an important source of selenium . This is also evident for Lake Macquarie invertebrates, in which selenium concentrations in ﬁlter-feeding invertebrates that consume particulates from the water column are lower compared with epibenthic deposit feeders (Supplemental Data, Table S1). Deposit-feeding invertebrates such as polychaetes and the bivalve mollusc S. alba had higher mean selenium concentrations (6 1 mg/g and 8 1 mg/g, respectively) than zooplankton and the ﬁlter-feeding bivalve mollusc C. truncata (2.0 0.2 mg/g and 3.0 0.1 mg/g dry wt, respectively). Vertebrates (ﬁsh). For ﬁsh, there was a signiﬁcant relationship between d15N and selenium concentrations (F1,73 ¼ 10.9; p < 0.01), indicating that the feeding group (i.e., herbivore or carnivore) with which ﬁsh were associated was an important factor inﬂuencing selenium accumulation. There was also a signiﬁcant effect of habitat/feeding zone on selenium concentration (F2,73 ¼ 14.41; p < 0.001). Pelagic ﬁsh feeding from the
Figure 4. Relationship between selenium concentrations and d15N in the seagrass ecosystem of Morisset Bay, Lake Macquarie. Red dashed line indicates the regression line, and blue dashed lines indicate 95% conﬁdence intervals.
Selenium biomagnification in Lake Macquarie
Environ Toxicol Chem 34, 2015
Figure 5. Effects of habitat, feeding zone, and trophic level (as indicated by d15N) on selenium concentrations of invertebrates from Morisset Bay, Lake Macquarie. Red, orange, and blue lines are the slopes of each habitat/feeding zone category determined by the analysis of covariance procedure.
water column had signiﬁcantly lower selenium concentrations than pelagic ﬁsh feeding on benthos (p < 0.001; Figure 6). There was no difference in selenium concentrations between pelagic ﬁsh feeding on benthos and benthic ﬁsh feeding on benthos (p > 0.05). Fish that are more mobile had selenium concentrations more inﬂuenced by feeding zone than by habitat exposure. Fish that fed on benthos had signiﬁcantly higher mean selenium concentrations (5 2 mg/g dry wt) than those that were pelagic
(3 2 mg/g), that is, feeding from the water column (Figure 6 and Table 1). The inﬂuence of feeding zone is clear; for example, the pelagic carnivorous ﬁsh T. gavialoides and P. saltatrix had lower selenium concentrations (3.6–4.0 mg/g) than the benthic carnivorous ﬁsh S. maculata and Dicotylychthys punctulatus, with selenium concentrations of 7 1 mg/g and 5.4 0.1 mg/g dry weight, respectively. The diet (i.e., feeding zone) is more important than the trophic level in determining the selenium concentrations in Lake Macquarie ﬁsh. As illustrated
Figure 6. Effects of habitat, feeding zone, and trophic level (as indicated by d15N) on selenium concentrations of ﬁsh from Morisset Bay, Lake Macquarie (New South Wales, Australia). Red, green, and blue lines are the slopes of each habitat/feeding zone category determined by the analysis of covariance procedure.
Environ Toxicol Chem 34, 2015
in the present study, trophic level may not be the overriding factor contributing to selenium concentrations in invertebrates and ﬁsh; other factors such as habitat and feeding zone also require consideration when examining the biomagniﬁcation of selenium in food webs. Selenium trophic transfer model
To apply the trophic transfer model of Presser and Luoma , it is ﬁrst necessary to establish the food web operating in the aquatic system of interest. In the present study, we have satisﬁed this criterion by using d13C and d15N isotopes and other published work on the stomach content of organisms (Figure 2 and Table 1). The selenium trophic transfer model established for the Morisset Bay seagrass ecosystem (Figure 7) is in agreement with the models presented by Luoma and Presser  for a number of different aquatic ecosystems. The water residence time in Lake Macquarie is long, and there are large expanses of shallow areas and seagrass beds; thus it was expected that selenium would partition to sediments and seagrasses. This was conﬁrmed by the low selenium concentrations in the water column (mostly