Abstract
Metal risk assessment of industrialized harbors and coastal marine waters requires the application of robust water quality guidelines to determine the likelihood of biological impacts. Currently there is no such guideline available for aluminium in marine waters. A water quality guideline of 24 µg total Al/L has been developed for aluminium in marine waters based on chronic 10% inhibition or effect concentrations (IC10 or EC10) and no‐observed‐effect concentrations (NOECs) from 11 species (2 literature values and 9 species tested including temperate and tropical species) representing 6 taxonomic groups. The 3 most sensitive species tested were a diatom Ceratoneis closterium (formerly Nitzschia closterium; IC10 = 18 µg Al/L, 72‐h growth rate inhibition) < mussel Mytilus edulis plannulatus (EC10 = 250 µg Al/L, 72‐h embryo development) < oyster Saccostrea echinata (EC10 = 410 µg Al/L, 48‐h embryo development). Toxicity to these species was the result of the dissolved aluminium forms of aluminate (Al(OH4−) and aluminium hydroxide (Al(OH)30) although both dissolved, and particulate aluminium contributed to toxicity in the diatom Minutocellus polymorphus and green alga Dunaliella tertiolecta. In contrast, aluminium toxicity to the green flagellate alga Tetraselmis sp. was the result of particulate aluminium only. Four species, a brown macroalga (Hormosira banksii), sea urchin embryo (Heliocidaris tuberculata), and 2 juvenile fish species (Lates calcarifer and Acanthochromis polyacanthus), were not adversely affected at the highest test concentration used. Environ Toxicol Chem 2015;34:141–151. © 2014 SETAC
INTRODUCTION
Elevated aluminium concentrations in the waters of industrialized harbors such as Port Curtis, Queensland, Australia [1], are a potential threat to aquatic ecosystem health. Sources of aluminium include discharges associated with alumina and aluminium production activities [2] as well as natural sources such as urban runoff and acid sulfate soils along the coastal plains of northeast Australia [3]. In spite of these known inputs of aluminium to coastal environments, a search of the international literature revealed no reliable water quality guideline for aluminium in marine waters. The Australian and New Zealand water quality guidelines [4] provide only a low‐reliability environmental concern level of 0.5 µg Al/L. This is because insufficient toxicity data were available that met the required quality assurance criteria to derive a proper guideline.
Published toxicity data for aluminium in marine waters that were considered when deriving the environmental concern level are summarized in the Supplemental Data, Table S1. Of the 7 toxicity studies, 6 used acute toxicity values and all used mortality as the effect rather than using chronic sublethal endpoints. Only 3 taxonomic groups were represented instead of the recommended 4 [4]. All the toxicity values were derived using nominal aluminium concentrations rather than measured, and none of the studies considered speciation of aluminium in marine waters. Because the minimum data requirements were not met, an assessment factor of 200 was applied to the lowest acute toxicity value of 97 µg Al/L for a polychaete species, to derive the 0.5 µg Al/L environmental concern level. This created an overly conservative value that was of little use in regulation.
Pristine oceanic dissolved aluminium concentrations are near 0.019 µg Al/L in the Southern Ocean [5], whereas relatively uncontaminated coastal waters such as those in Port Hacking (New South Wales, Australia) have concentrations in the 1.2 µg to 1.8 µg Al/L range (J. King, 2013, Honours thesis, University of Wollongong, Wollongong, Australia), which are well above the environmental concern level. There is therefore a need for a reliable guideline value.
The chemical speciation of aluminium is an important consideration in determining toxicity to marine organisms. At a typical seawater pH of 8.1, ion interaction modeling calculations have shown that dissolved inorganic aluminium comprises aluminate (Al(OH)4−; 68%) and a noncolloidal neutral hydroxide species (Al(OH)30; 32%) [6]. As the pH becomes more alkaline, the proportion of neutral species decreases, whereas the anionic aluminate species increases. Given the low concentration of dissolved organic carbon (DOC) in seawater, organic complexation of aluminium in seawater will not be discussed in the present study. It is noteworthy that cationic species that dominate aluminium speciation in acidic waters are negligible at the pH of seawater. In the context of interactions with living organisms, bioaccumulation may result from the uptake of either the anionic or neutral species (or both). It is currently not possible to distinguish the relative contribution of these species to toxicity or the mechanisms of toxicity involved. Further information on aluminium speciation in marine systems has been provided by B.M. Angel et al. (unpublished manuscript).
The significance of metal solubility is often overlooked in aquatic toxicity tests. If the amount of metal added results in solution concentrations exceeding the solubility limit, then precipitation is predicted to occur, although the kinetics of precipitation are often unpredictable. Practically, this introduces many problems in understanding the concentration that organisms receive. Both dissolved and particulate aluminium species may be toxic, and thus the contributions of dissolved and particulate species to toxicity will likely change with time. Aluminium solubility in seawater has recently been investigated by B.M. Angel et al. (unpublished manuscript) under controlled conditions. Briefly, that study found that, under equilibrium conditions, the solubility of aluminium in seawater (pH 8.2) was approximately 500 µg/L. However, the solution chemistry is dynamic, and transient dissolved aluminium concentrations above this solubility limit may persist for a number of days before precipitation and establishment of equilibrium conditions. For instance, when seawater solutions were spiked with high concentrations of total aluminium (e.g., >10 000 µg/L), precipitation occurred almost immediately, but a pulse of dissolved aluminium as high as 1700 µg/L was sustained for several days before decreasing to the equilibrium solubility concentration. This has a bearing on toxicity studies that are typically conducted over periods of 2 d to 5 d. Over these timescales, the dissolved aluminium concentrations are likely to vary considerably, reflecting precipitation of aluminium hydroxide (B.M. Angel et al., unpublished manuscript). It is therefore apparent that at high concentrations of total aluminium (>500 µg/L), test organisms will be exposed to a mixture of both dissolved and particulate forms of aluminium. The extent to which each form contributes to the toxicity is an important consideration.
The objective of the present study was to develop a water quality guideline for aluminium in marine waters that could be categorized as “high reliability” [4]. A toxicity dataset (11 species from 6 taxonomic groups) was generated from a battery of relevant chronic toxicity tests and recent literature data, and the protocols for guideline development that currently apply in Australia and New Zealand were applied to the dataset [4, 7, 8]. Particular emphasis was placed on characterizing the dissolved and particulate forms of aluminium in solution and the dynamic changes in these forms over the time course of the bioassays.
MATERIALS AND METHODS
Preparation and analysis of bioassay solutions
Unless otherwise stated, all glass and plasticware used in the present study were soaked in 10% (v/v) concentrated HNO3 (BDH Analytical Reagent grade) for a minimum of 24 h and rinsed with deionized water (18 MΩ/cm, Milli‐Q, Millipore) prior to use. Chemical reagents were of analytical reagent grade or better. Aluminium bioassay solutions were prepared with clean seawater (34 PSU, pH 8.2, DOC 1 mg/L) collected from a rock platform at Cronulla, New South Wales, Australia (34°04′13.35′′S, 151°09′25.69′′E) using plastic containers (5 L), filtered (0.45‐µm Sartobran P MidiCaps, Sartorius), and then stored (at 4 °C) in the dark until time of use.
The dynamic nature of aluminium chemistry in seawater posed special concerns as to how best to prepare stock aluminium solutions that could be diluted to the required concentrations and pH for toxicity testing. In a separate study (B.M. Angel et al. unpublished manuscript), an appropriate protocol with which to prepare stable solutions of aluminium to cover the expected concentration ranges needed for chronic toxicity testing was developed. To limit changes in chemical speciation, aluminium solutions were prepared no more than 24 h prior to toxicity test initiation. Aluminium stock solutions of 1 g and 10 g Al/L were prepared in 30‐mL polycarbonate containers by adding 0.179 g and 0.895 g of AlCl3·6H2O (Ajax Finechem, UNIVAR, AR grade) to 20 mL of 0.108 M NaOH (Ajax Finechem, UNIVAR, AR grade) and 10 mL of 1.08 M NaOH, respectively (B.M. Angel et al., unpublished manuscript). A third stock solution of 0.01 g Al/L was prepared by diluting 0.1 mL of 1 g Al/L stock in a final volume of 10 mL 0.02 M NaOH. A white precipitate of aluminium hydroxide was observed to form in the stock solutions as the solubility limit was exceeded; however, this quickly dissolved once stocks were diluted below the limit of solubility to form the test solutions at pH 8.2, and preliminary tests demonstrated the absence of colloidal aluminium (B.M. Angel et al., unpublished manuscript) at the time that test organisms were added. Aluminium test solutions (10 µg Al/L–100 000 µg Al/L nominal) were prepared in polycarbonate or low‐density polyethylene containers by adding appropriate aliquots (≤1% of final volume to maintain pH and salinity) of the stock solutions to filtered seawater. The pH of the test solutions was measured, and fine adjustments to the pH were made by adding 0.108 M or 1.08 M NaOH dropwise to reach a final pH of 8.2 ± 0.05. Volumes (5 mL–500 mL) of the test solutions were then dispensed into test vessels specific to the bioassay being conducted.
Physicochemical properties (pH, salinity, conductivity, dissolved oxygen) were measured in the bioassay solutions at the time of test initiation and termination. Total and filtered (0.45‐µm syringe filter, Minisart, Sartorius) water samples (5 mL) for analysis of total and operationally defined dissolved aluminium, respectively, were collected from each replicate at test initiation (in the case of the microalgal bioassays) or the test solutions (in the case of all other bioassays) and from each replicate at test termination (in the case of the microalgal bioassays) or from 1 to 2 replicates of each concentration (in the case of all other bioassays). Additional water samples were collected on day 3 of the 7‐d fish tests, and ammonia was measured at test termination. Filtered and unfiltered water samples were acidified with 2% or 5% (v/v) HNO3 (Tracepur, Merck), respectively, in 5‐mL polycarbonate vials and stored (at 4 °C in the dark) prior to analysis.
Samples were analyzed for both dissolved and total aluminium using inductively coupled plasma atomic emission spectrometry (ICPAES; Varian 730‐ES). Multimetal standards (2007‐LPC4R, QDC Analysts) were prepared in matrix‐matched filtered seawater and preserved with 2% or 5% (v/v) HNO3 (Tracepur, Merck). The limit of detection (LOD) for aluminium was 2.5 µg/L (3× the standard deviation [SD] of the mean blank, n = 132). Values less than the LOD were substituted as half of the LOD for subsequent data analysis. There are no certified reference seawater samples available for aluminium analysis.
Copper reference toxicant solutions were prepared from CuSO4·5H2O (Pronalys, Selby Biolab, AR grade) at nominal concentrations of 0.125 µg to 2000 µg Cu/L. These samples were analyzed for dissolved copper by ICPAES with a LOD of 1.5 µg/L.
Bioassay species and test conditions
Bioassays were conducted with a range of Australian coastal aquatic species representing different trophic levels and temperature zones. A total of 11 species from 6 taxonomic groups (as defined by the Australia and New Zealand Environment and Conservation Council/Agricultural and Resource Management Council of Australia and New Zealand [ANZECC/ARMCANZ] [4], Appendix 5) representing marine microalgal diatoms and green algae, brown macroalgae, molluscs (mussels and oysters), echinoderms (sea urchin), and fish were exposed to aluminium in filtered seawater (34 PSU, pH 8.2, DOC 1 mg/L), under static nonrenewal conditions. With the exception of the juvenile fish bioassays, all bioassays were defined as having chronic toxicity values [7]. Effects on cell growth inhibition rate were assessed with bioassays using microalgae (Bacillariophyceae [diatoms]): Ceratoneis closterium (formerly Nitzschia closterium [Ehrenb.] W. Smith [strain CS‐5, temperate, collected from Port Hacking]) and Minutocellus polymorphus (Hargraves and Guillard) Hasle, Von Stosch and Syvertsen; the green alga (Chlorophyceae): Dunaliella tertiolecta (Butcher); and the green flagellate (Prasinophyceae): Tetraselmis sp. The effects of germination success were assessed with bioassays using macroalgae (brown algae; Phaeophyceae) Ecklonia radiata and Hormosira banksii. Effects on embryo development were assessed with bioassays using a mussel (Mollusca) Mytilus edulis plannulatus, a tropical oyster (Mollusca) Saccostrea echinata, and a sea urchin (Echinodermata) Heliocidaris tuberculata. Bioassays using tropical damselfish (Acanthochromis polyacanthus) and barramundi (Lates calcarifer) determined acute effects on juvenile fish swimming imbalance and growth.
With the exception of the chronic H. banksii test and the acute fish tests, all bioassays were performed on at least 2 separate occasions with at least 2 concentrations common between repeats. The number of test concentrations for each species ranged from 5 to 14, with 3 to 4 replicates per concentration. Reference toxicity testing using copper sulfate was conducted on all species except fish, and the toxicity value was compared with in‐house historical data where available, to ensure consistent sensitivity of the test organisms. Reference toxicity testing was not conducted with the fish because of low availability of animals. Microalgal tests were conducted with laboratory‐cultured organisms at the Commonwealth Scientific and Industrial Research Organisation (CSIRO) and the remaining bioassays were conducted at Ecotox Services Australasia using field‐collected or hatchery‐harvested organisms, with the exception of the damselfish, which were also laboratory cultured.
Microalgal tests
Microalgal species were sourced and cultured in media according to the method of Levy et al. [9]. Bioassay methods were based on Test Guideline 201 of the Organization for Economic Co‐operation and Development [10] and the protocol of Stauber et al. [11], with modifications summarized by Franklin et al. [12]. These bioassays use highly sensitive flow cytometric detection procedures for cell quantification purposes, which allow the use of low algal cell densities. Bioassays were conducted in acid‐washed 250‐mL borosilicate glass Erlenmeyer flasks (with lids) coated with Coatasil silanizing solution (APS Ajax Finechem) to reduce adsorption of metals to the flask walls.
Bioassay test solutions were prepared in triplicate using filtered seawater (50 mL) with nutrients added (15 mg/L NO3− as NaNO3, 1.5 mg/L PO43− as KH2PO4) and incubated for 72 h at 21 °C in 12:12‐h light (140 µmol photons/m2/s photosynthetically active radiation):dark. Cells were harvested from cultures in their exponential growth phase (5–6 d old) and rinsed 3 times in 0.45‐µm filtered seawater using centrifugation (1188 g, 7 min, rotor radius 17 cm, Spintron GT‐175BR) to remove residual culture media. Tetraselmis sp. was rinsed once (190 g, 4 min) to prevent lysis of the larger more fragile cells.
Initial test algal concentrations were 2 × 104 cells/mL to 4 × 104 cells/mL for C. closterium (for direct comparison with Harford et al. [13]) and 2 × 103 cells/mL to 4 × 103 cells/mL for the other species. Flasks were shaken manually twice a day and positioned randomly to account for light and temperature gradients. The algal cell concentration in each flask was measured daily using flow cytometry (BD‐FacsCalibur, Becton Dickinson BioSciences) whereby cells were excited with blue light (488 nm), and chlorophyll a autofluorescence was measured in the 660‐nm to 700‐nm band (FL3). The C. closterium cells were gently homogenized to reduce clumping of cells prior to counting by flow cytometry. Inspection of cells under the microscope before and after homogenizing confirmed no damage to cells and reduced clumping. Cell homogenization was not necessary for the other microalgal species. Further details are given in Franklin et al. [12]. Growth rate was calculated as the slope of the linear regression of log10 algal cell concentration versus time and was used in deriving the chronic effect value for algal growth inhibition. The test was acceptable if the control growth rate was >1 doubling/d and the control growth rate coefficient of variation was <10% [11], and the inhibitory concentration, 50% (IC50) for the copper reference test was within 2 SDs of the mean (C. closterium IC50 = 16 µg Cu/L, 2 SD = 3.0 µg–26 µg Cu/L, n = 44; M. polymorphus IC50 = 1.0 µg Cu/L, 2 SD = 0.94 µg–1.1 µg Cu/L, n = 1 [see Levy et al. [8] for the method of calculating IC50 when less than LOD]; D. tertiolecta IC50 = 170 µg Cu/L, 2 SD = 97 µg–240 µg Cu/L, n = 1; Tetraselmis sp. IC50 = 70 Cu/L, 2 SD = 56 µg/L–84 µg/L, n = 1).
Macroalgal tests
The macroalgal species H. banksii and E. radiata were collected from coastal reef sites in New South Wales and Tasmania, Australia, respectively, and were used in bioassays within 24 h of collection. Methods were based on those of Kevekordes and Clayton [14] and Bidwell et al. [15]. Bioassays were conducted in acid‐washed 20‐mL soda‐glass Petri dishes (15‐mm depth, 60‐mm diameter, Schott) with lids. Test concentrations were prepared in quadruplicate using filtered seawater (5 mL) and incubated for 72 h at 18 °C in 16:8‐h light (80 µmol photons/m2/s):dark. Several plants were spawned using warm (30–40 °C) seawater (in the case of H. banksii) and cold (15 °C) seawater (in the case of E. radiata), and the released gametes or zoospores were concentrated by allowing settling and removal of excess seawater.
The H. banksii eggs were fertilized by the addition of sperm (1:200 egg:sperm) and were dispensed to obtain a final density of 100 fertilized eggs/mL. The E. radiata zygotes were dispensed into replicates to reach a final density of between 20 000 organisms/mL and 75 000 organisms/mL. The E. radiata were initially dispensed onto a glass coverslip (24 mm × 40 mm) in a Petri dish containing seawater to act as an artificial substratum for zoospore adhesion. The seawater was replaced with test solution at test initiation. After 72 h, the number of germinated embryos out of 100 was counted by microscopy (×100 magnification) and used to derive the inhibition of germination chronic effect value. The test was acceptable if there was >70% control germination and the EC50 for the copper reference test was within 2 SD of the mean (H. banksii EC50 = 133 µg Cu/L, 2 SD = 55 µg–381 µg Cu/L, n = 20; E. radiata EC50 = 431 µg Cu/L, 2 SD = 294 µg–615 µg Cu/L, n = 18).
Mussel, oyster, and sea urchin tests
Mussels (M. plannulatus) were sourced from aquaculture farms in Tasmania, Australia; oysters (S. echinata) and sea urchins (H. tuberculata) were field collected subtidally from Queensland and New South Wales, Australia, respectively. Bioassays were performed within 48 h of receiving the animals according to the methods of Krassoi [16] for the oyster, Doyle et al. [17] for the sea urchin, and the United States Environmental Protection Agency (USEPA) [18] for the mussel.
Bioassays were conducted in new unwashed disposable borosilicate glass test tubes (13 mm × 100 mm, Kimble Chase). Test solutions were prepared in quadruplicate using filtered seawater (5 mL), and the tubes were covered in plastic film and incubated for 72 h at 20°C in 16:8‐h light (80 µmol photons/m2/s):dark in the case of the mussels and sea urchins and for 48 h at 29°C in 12:12‐h light (80 µmol photons/m2/s):dark in the case of the oysters. Gametes from several mussels were collected using incremental (10 °C/20 °C) water temperature shock and gametes from several oysters were strip collected. Sea urchins were spawned by injecting 0.5 M KCl solution into the oral cavity. Sperm and eggs from individuals were collected separately in seawater, checked for quality by microscopy and pooled accordingly. Gamete density was determined by microscopy and sperm were added to the egg solution to achieve an egg:sperm ratio of 1:100. When fertilisation was greater than 80%, an aliquot of egg:sperm solution was added to each replicate to achieve 100 fertilised eggs/mL for the mussel and oyster tests or 80 fertilised eggs/mL for the urchin larval test.
The duration of the bioassays was 72 h for the mussel and sea urchin and 48 h for the oyster. The tests were terminated by the addition of 10% (v/v) buffered formalin and embryo development in the first 100 individuals was assessed by microscopy (100× magnification) using a Sedgewick‐Rafter counting chamber. Normal development produced the D‐shaped veliger larva in the mussels and oysters or the pluteus larva in the sea urchins whereas abnormal individuals failed to develop from the zygote or trochophore stages or were misshapen. The percentage of normal embryo development relative to the control was used to derive the chronic endpoint. The test was acceptable if there was >70% normal development in the controls and the EC50 for the reference toxicant (copper) was within 2 SD of the mean (S. echinata EC50 = 16 µg Cu/L, 2 SD = 12 µg–21 µg Cu/L, n = 20; M. edulis plannulatus EC50 = 11 µg Cu/L, 2 SD = 9 µg–14 µg Cu/L, n = 20; H. tuberculata EC50 = 11 µg Cu/L, 2 SD = 7 µg–18 µg Cu/L, n = 20).
Fish tests
Juvenile barramundi (5 wk old, 12.2 ± 5.8 mg dry wt, mean ± SD) were hatchery harvested (WBA Hatcheries), and damselfish (7 wk old, 9.9 ± 6.0 mg dry wt, mean ± SD) were taken from laboratory cultures (Ecotox Services Australasia) for use in bioassays according to a modified method of the USEPA [19]. Bioassays were conducted in acid‐washed glass beakers (600 mL). Test solutions were prepared in triplicate or quadruplicate (barramundi and damselfish, respectively) using filtered seawater (500 mL) and gently aerated (∼60 bubbles/min) by pumping air through a glass Pasteur pipette positioned in each beaker. The beakers were covered with plastic film and incubated for 7 d at 25 °C in 16:8‐h light (100 µmol photons/m2/s):dark. Five fish were randomly assigned to each replicate beaker and were fed daily with live brine shrimp (800 live shrimp/fish/d).
Dry weight was determined on a subsample of fish sacrificed at test initiation by drying the fish at 60 °C for at least 24 h to get a constant dry weight. Daily observations were made of fish behavior, and swimming imbalance was recorded as the inability of a fish to remain upright. Fish with imbalance were subsequently removed and euthanized with isoeugenol (clove oil derivative, 54 mg/L, Aqui‐S). After 7 d, surviving fish were euthanized and dried at 60 °C for at least 24 h, and a constant dry body weight was recorded. Acute sublethal toxicity values for imbalance and growth were then derived. The test was acceptable if there was >80% survival in the control and the dry weight of the controls was >120% of the initial dry weight at the beginning of the test.
Statistical analysis
Concentration–response curves were constructed using the biological effect as a function of the measured aluminium concentration (both total and dissolved) data from the bioassays. The concentration of total measured aluminium at the time of bioassay initiation causing a 10% inhibition (IC10) or effect (EC10) for each organism was determined by nonlinear regression using R package drc (Ver 2.3‐0) [20, 21]. The package uses 2‐ to 5‐parameter log‐logistic and Weibull models to describe the concentration–response relationship. Model fit was assessed visually in addition to the minimum mean residual error of the fit. The 95% confidence limits on the toxicity values were generated using the delta method of estimating the asymptotic standard error and the appropriate t‐distribution [20]. Data for each replicate were entered as the response relative to the control, and data from multiple tests with the same species were pooled after control responses were found to be similar between tests. Bioassay concentrations were selected to define the whole concentration–response curve and also to reduce variability at the lower end of the curve where the EC10/IC10 was derived. An effect level of 10% was chosen to provide a conservative level of protection. The measured total aluminium concentration at the time of test initiation was used instead of a time‐weighted average concentration over the exposure period to provide a conservative exposure scenario. Where possible, the EC10 or IC10 value for each species was used to create a species sensitivity distribution using the BurrliOZ software (Ver 2.0, CSIRO). For species in which no adverse effects were observed over the range of concentrations tested, the highest test concentration was used as a surrogate for a no‐observed‐effect concentration (NOEC) in the species sensitivity distribution. When there was more than 1 chronic effect concentration for the same species, the geometric mean of the effect concentrations was used. The BurrliOZ software fits 2 nonlinear regression models (Burr type III, or, if fewer than 8 data points, a log‐logistic regression) and calculates the guideline value that protects a given percentage of species (typically 95%) with a 50% level of confidence.
RESULTS AND DISCUSSION
Forms of aluminium in the bioassay test solutions
Figure 1 shows the dissolved aluminium data plotted against measured total concentrations for the start and end of each bioassay (replicate bioassays included; fish tests have additional sampling times). Specific data for individual test biota as a function of total and dissolved aluminium at the start of each bioassay are shown in Figures 2 through 5 and at the end of each bioassay in Supplemental Data, Figure S1. The most interesting feature of Figure 1 is the almost linear 1:1 relationship between dissolved and total aluminium up to approximately 1000 µg/L total aluminium. Aluminium below these concentrations is predominantly in dissolved forms. At higher concentrations, a combination of dissolved and particulate forms coexists. In this range, even higher dissolved aluminium concentrations, up to 1700 µg/L, were seen at the start of the tests irrespective of the total aluminium concentration, which was varied up to 100 000 µg/L. Figure 1 also shows how dissolved aluminium concentrations decline with time over the course of the bioassays. Note also that in the presence of excess aluminium precipitate (>10 000 µg total Al/L), dissolved concentrations may fall below the solubility limit of 500 µg/L. This is likely related to adsorption of aluminate onto newly formed aluminium hydroxide flocs and is discussed in more detail by B.M. Angel et al. (unpublished manuscript).
Concentrations of dissolved and total aluminium (µg/L) in bioassay solutions over time during the bioassays (A and B are higher and lower concentration ranges of total aluminium, respectively). The area enclosed within the dotted line and axes represents the range of total aluminium for which dissolved aluminium is close to 100%. Outside of that area, dissolved and particulate aluminium forms coexist. Taxonomic groups in the bioassays are represented by symbols. ▪ = diatoms, ♦ = green algae, ▴ = brown algae, ● = molluscs (mussels and oysters), × (for beginning of bioassay) or 
(for end of bioassay) = echinoderm (sea urchin), + (for beginning of bioassay) or – (for end of bioassay) = fish. Closed and open symbols represent aluminium measurements at the beginning and end of the bioassay, respectively.
The tests on the microalgae M. polymorphus, D. tertiolecta, and Tetraselmis sp. had higher dissolved aluminium in the first few hours after preparation of the test solutions at high total aluminium concentrations (>1000 µg Al/L) than in the study by B.M. Angel et al. (unpublished manuscript), in which test organisms were absent. This finding suggested that in the presence of these microalgal species, the solubility of aluminium was increased. This behavior and the potential role of algal exudates is the subject of ongoing mechanistic studies.
For each species, there were decreases in dissolved aluminium over the test duration. For the 72‐h exposures (i.e., the duration of the majority of the bioassays), these decreases averaged 32 ± 28%, n = 206 (mean ± standard deviation [SD]) where total aluminium was less than 1000 µg/L (Supplemental Data, Figure S1). For bioassays at higher concentrations (≥1000 µg total Al/L) and for longer durations (168 h), dissolved aluminium declined further (54 ± 29%, n = 14), and was in many cases similar to those observed in the absence of biota (B.M. Angel et al. unpublished manuscript). In the bioassays (using brown macroalgae, mussels, oysters, and sea urchins) carried out at greater than 10 000 µg/L total aluminium concentrations in small‐volume glass containers, the measured total aluminium concentrations at the end of the bioassay were variable (39 ± 22% of nominal, n = 14) because of the large variability as the settled particulate material was sampled. Adsorptive losses of dissolved aluminium to container walls and particulate aluminium also occurred. The changes in the forms of aluminium with time and concentration demonstrated the importance of quantifying the total and dissolved concentrations at the beginning and end of the bioassay.
Physicochemical conditions over all the bioassay test solutions containing aluminium remained stable with pH 8.2 (±0.1 SD, n = 330; Supplemental Data, Figure S2) and dissolved oxygen at 100% saturation (±4% SD, n = 306). Temperature remained within the required operating specifications of the incubators used. Total ammonia concentrations in the fish bioassays ranged from 0 mg to 10 mg NH3‐N/L. The coefficient of variation in the response of bioassay controls ranged from 0% to 7%, with a notable variation of 100% in the mean percentage growth of control damselfish because of 1 replicate with high growth.
Concentration–response curves and toxicity values for test organisms
Relationships between aluminium (total and dissolved) at the time of test initiation and chronic toxicity in the 11 marine species tested are shown in Figures 2, 3, 4, 5. The concentration–response data as a function of total and dissolved aluminium at the beginning and end of the bioassay are shown in Supplemental Data, Figure S1. These plots are of value in determining the relative importance of dissolved and particulate species in eliciting toxicity.
Chronic toxicity of total (filled circles) and dissolved (open circles) aluminium (µg/L) to microalgal species: the diatom Ceratoneis closterium (A); the diatom Minutocellus polymorphus (B); the green alga Dunaliella tertiolecta (C); and the green flagellate Tetraselmis sp. (D). Lines are modeled concentration–response relationships for total (solid line) and dissolved (dotted line) aluminium.
Chronic toxicity of total (filled circles) and dissolved (open circles) aluminium (µg/L) to embryo development in: the mussel Mytilus edulis plannulatus (A); the tropical oyster Saccostrea echinata (B); and the sea urchin Heliocidaris tuberculata (C). Lines are modeled concentration–response relationships for total (solid line) and dissolved (dotted line) aluminium.
Chronic toxicity of total (filled circles) and dissolved (open circles) aluminium (µg/L) to brown macroalgal species: Hormosira banksii (A); and Ecklonia radiata (B). Lines are modeled concentration–response relationships for total (solid line) and dissolved (dotted line) aluminium.
Acute toxicity of total (filled circles) and dissolved (open circles) aluminium (µg/L) to growth in the tropical marine fish: barramundi Lates calcarifer (A); and damselfish Acanthochromis polyacanthus (B). Lines are modeled concentration–response relationships for total (solid line) and dissolved (dotted line) aluminium.
Chronic, IC10 or EC10 toxicity values were determined for 7 of the species (2 diatoms, 2 green algae, 1 brown macroalga, a mussel, and an oyster), and the highest test concentration was used as a surrogate NOEC for 2 species (a brown macroalga and a sea urchin) where no effects were observed at any of the tested concentrations (Table 1, Figures 3 and 4). Additional effect concentrations (IC or EC) at the 20% and 50% effect level for all species tested are presented in Supplemental Data, Table S2. Because of the need to provide protection from all (both dissolved and particulate) forms of aluminium, the total measured aluminium was used in the calculation of the toxicity values. Total aluminium values are the most commonly measured and reported, thus making the toxicity values directly comparable with much of the literature.
Chronic toxicity data used to derive the water quality guideline value for total aluminum in marine waters
| Taxonomic group | Species | Duration (h) | Temperature (°C) | Endpoint | Toxicity measure | Toxicity value (µg/L)1 | Model | Residual standard error of model |
| Microalgae: diatom | Ceratoneis closterium | 72 | 21 | Growth rate inhibition | IC10 | 16 (3.2–78)2 | ||
| Mollusc: oyster | Saccostrea glomerata3 | 48 | 24 | Embryo development | NOEC | 100 –4 | ||
| Mollusc: mussel | Mytilus edulis plannulatus | 72 | 20 | Embryo development | EC10 | 250 (240–260) | Log logistic, 3‐parameter | 3.7 |
| Mollusc: oyster | Saccostrea echinata | 48 | 29 | Embryo development | EC10 | 410 (340–480) | Log logistic, 3‐parameter | 8.4 |
| Microalgae: diatom | Minutocellus polymorphus | 72 | 21 | Growth rate inhibition | IC10 | 690 (580–800) | Log logistic, 3‐parameter | 5.5 |
| Cnidarian: coral | Acropora tenuis | 18 | 32 | Larval metamorphosis | EC10 | 1300 (860–1700)5 | ||
| Microalgae: green alga | Dunaliella tertiolecta | 72 | 21 | Growth rate inhibition | IC10 | 1400 (820–1900) | Weibull Type 2, 4‐parameter | 7 |
| Microalgae: green flagellate | Tetraselmis sp. | 72 | 21 | Growth rate inhibition | IC10 | 3200 (1100–5400) | Log logistic, 3‐parameter | 13 |
| Macroalgae: brown alga | Ecklonia radiata | 72 | 18 | Germination success | IC10 | 6800 (5400–8200) | Weibull type 2, 4‐parameter | 8.3 |
| Macroalgae: brown alga | Hormosira banksii | 72 | 18 | Germination success | ND | 9800 | ||
| Echinoderm: sea urchin | Heliocidaris tuberculata | 72 | 20 | Embryo development | ND | 28000 |
| Taxonomic group | Species | Duration (h) | Temperature (°C) | Endpoint | Toxicity measure | Toxicity value (µg/L)1 | Model | Residual standard error of model |
| Microalgae: diatom | Ceratoneis closterium | 72 | 21 | Growth rate inhibition | IC10 | 16 (3.2–78)2 | ||
| Mollusc: oyster | Saccostrea glomerata3 | 48 | 24 | Embryo development | NOEC | 100 –4 | ||
| Mollusc: mussel | Mytilus edulis plannulatus | 72 | 20 | Embryo development | EC10 | 250 (240–260) | Log logistic, 3‐parameter | 3.7 |
| Mollusc: oyster | Saccostrea echinata | 48 | 29 | Embryo development | EC10 | 410 (340–480) | Log logistic, 3‐parameter | 8.4 |
| Microalgae: diatom | Minutocellus polymorphus | 72 | 21 | Growth rate inhibition | IC10 | 690 (580–800) | Log logistic, 3‐parameter | 5.5 |
| Cnidarian: coral | Acropora tenuis | 18 | 32 | Larval metamorphosis | EC10 | 1300 (860–1700)5 | ||
| Microalgae: green alga | Dunaliella tertiolecta | 72 | 21 | Growth rate inhibition | IC10 | 1400 (820–1900) | Weibull Type 2, 4‐parameter | 7 |
| Microalgae: green flagellate | Tetraselmis sp. | 72 | 21 | Growth rate inhibition | IC10 | 3200 (1100–5400) | Log logistic, 3‐parameter | 13 |
| Macroalgae: brown alga | Ecklonia radiata | 72 | 18 | Germination success | IC10 | 6800 (5400–8200) | Weibull type 2, 4‐parameter | 8.3 |
| Macroalgae: brown alga | Hormosira banksii | 72 | 18 | Germination success | ND | 9800 | ||
| Echinoderm: sea urchin | Heliocidaris tuberculata | 72 | 20 | Embryo development | ND | 28000 |
Toxicity value derived from measured initial total aluminum concentration, reported to 2 significant figures; 95% confidence limits shown in parentheses.
Geometric mean of 10% inhibitory concentration (IC10) from this study (18 µg Al/L, 11 µg Al/L–26 µg Al/L 95% confidence limits; model used in R package “drc” is Weibull type 2, 4‐parameter; residual standard error is 4.7) and from Harford et al. [13] of 14 µg Al/L (3 µg Al/L–25 µg Al/L 95% confidence limits).
Saccostrea glomerata was formerly known as Saccostrea commercialis.
Taken from Wilson and Hyne [27] where NOEC is the no‐observed‐effect concentration.
Taken from Negri et al. [33].
ND = no adverse effects were observed in the test range of aluminum concentrations; therefore the highest measured total Al was used as the NOEC in the species sensitivity distribution.
Chronic toxicity data used to derive the water quality guideline value for total aluminum in marine waters
| Taxonomic group | Species | Duration (h) | Temperature (°C) | Endpoint | Toxicity measure | Toxicity value (µg/L)1 | Model | Residual standard error of model |
| Microalgae: diatom | Ceratoneis closterium | 72 | 21 | Growth rate inhibition | IC10 | 16 (3.2–78)2 | ||
| Mollusc: oyster | Saccostrea glomerata3 | 48 | 24 | Embryo development | NOEC | 100 –4 | ||
| Mollusc: mussel | Mytilus edulis plannulatus | 72 | 20 | Embryo development | EC10 | 250 (240–260) | Log logistic, 3‐parameter | 3.7 |
| Mollusc: oyster | Saccostrea echinata | 48 | 29 | Embryo development | EC10 | 410 (340–480) | Log logistic, 3‐parameter | 8.4 |
| Microalgae: diatom | Minutocellus polymorphus | 72 | 21 | Growth rate inhibition | IC10 | 690 (580–800) | Log logistic, 3‐parameter | 5.5 |
| Cnidarian: coral | Acropora tenuis | 18 | 32 | Larval metamorphosis | EC10 | 1300 (860–1700)5 | ||
| Microalgae: green alga | Dunaliella tertiolecta | 72 | 21 | Growth rate inhibition | IC10 | 1400 (820–1900) | Weibull Type 2, 4‐parameter | 7 |
| Microalgae: green flagellate | Tetraselmis sp. | 72 | 21 | Growth rate inhibition | IC10 | 3200 (1100–5400) | Log logistic, 3‐parameter | 13 |
| Macroalgae: brown alga | Ecklonia radiata | 72 | 18 | Germination success | IC10 | 6800 (5400–8200) | Weibull type 2, 4‐parameter | 8.3 |
| Macroalgae: brown alga | Hormosira banksii | 72 | 18 | Germination success | ND | 9800 | ||
| Echinoderm: sea urchin | Heliocidaris tuberculata | 72 | 20 | Embryo development | ND | 28000 |
| Taxonomic group | Species | Duration (h) | Temperature (°C) | Endpoint | Toxicity measure | Toxicity value (µg/L)1 | Model | Residual standard error of model |
| Microalgae: diatom | Ceratoneis closterium | 72 | 21 | Growth rate inhibition | IC10 | 16 (3.2–78)2 | ||
| Mollusc: oyster | Saccostrea glomerata3 | 48 | 24 | Embryo development | NOEC | 100 –4 | ||
| Mollusc: mussel | Mytilus edulis plannulatus | 72 | 20 | Embryo development | EC10 | 250 (240–260) | Log logistic, 3‐parameter | 3.7 |
| Mollusc: oyster | Saccostrea echinata | 48 | 29 | Embryo development | EC10 | 410 (340–480) | Log logistic, 3‐parameter | 8.4 |
| Microalgae: diatom | Minutocellus polymorphus | 72 | 21 | Growth rate inhibition | IC10 | 690 (580–800) | Log logistic, 3‐parameter | 5.5 |
| Cnidarian: coral | Acropora tenuis | 18 | 32 | Larval metamorphosis | EC10 | 1300 (860–1700)5 | ||
| Microalgae: green alga | Dunaliella tertiolecta | 72 | 21 | Growth rate inhibition | IC10 | 1400 (820–1900) | Weibull Type 2, 4‐parameter | 7 |
| Microalgae: green flagellate | Tetraselmis sp. | 72 | 21 | Growth rate inhibition | IC10 | 3200 (1100–5400) | Log logistic, 3‐parameter | 13 |
| Macroalgae: brown alga | Ecklonia radiata | 72 | 18 | Germination success | IC10 | 6800 (5400–8200) | Weibull type 2, 4‐parameter | 8.3 |
| Macroalgae: brown alga | Hormosira banksii | 72 | 18 | Germination success | ND | 9800 | ||
| Echinoderm: sea urchin | Heliocidaris tuberculata | 72 | 20 | Embryo development | ND | 28000 |
Toxicity value derived from measured initial total aluminum concentration, reported to 2 significant figures; 95% confidence limits shown in parentheses.
Geometric mean of 10% inhibitory concentration (IC10) from this study (18 µg Al/L, 11 µg Al/L–26 µg Al/L 95% confidence limits; model used in R package “drc” is Weibull type 2, 4‐parameter; residual standard error is 4.7) and from Harford et al. [13] of 14 µg Al/L (3 µg Al/L–25 µg Al/L 95% confidence limits).
Saccostrea glomerata was formerly known as Saccostrea commercialis.
Taken from Wilson and Hyne [27] where NOEC is the no‐observed‐effect concentration.
Taken from Negri et al. [33].
ND = no adverse effects were observed in the test range of aluminum concentrations; therefore the highest measured total Al was used as the NOEC in the species sensitivity distribution.
Microalgae growth rate inhibition
The toxicity of aluminium to the chronic growth rate of marine microalgae as quantified by the IC10 was species specific, with greater effects on diatoms relative to green algae by as much as 200 times (Table 1, Figure 2, and Supplemental Data, Figure S1). The same observation has been made regarding differences in aluminium toxicity to freshwater diatoms and desmids compared with green microalgal taxa [22]. The microalgal species ranked in sensitivity to aluminium as follows: the pennate diatom C. closterium (IC10 = 18 µg/L) > the centric diatom M. polymorphus (IC10 = 690 µg/L) > the green alga D. tertiolecta (IC10 = 1400 µg/L) > the green flagellate Tetraselmis sp. (IC10 = 3200 µg/L; Table 1).
Ceratoneis closterium was the most sensitive to aluminium of all 11 species that were tested. The concentration–response curve differed from those of the other microalgae tested, in that it extended over 3 orders of magnitude of total aluminium from 10 µg/L to 10 000 µg/L (Figure 2) and was the same as that observed for the tropical strain of C. closterium [13]. By comparison, the same curve for the copper reference toxicant spans less than 2 orders of magnitude (Supplemental Data, Figure S3). The concentration–response curve at 10 µg to 250 µg total Al/L had a sharp decline in growth rate to reach approximately 50% of control. Thereafter, there was a less pronounced decline in growth rate up to concentrations of approximately 1200 µg total Al/L. Toxicity appears to be more strongly related to dissolved aluminium with potential contributions from particulate aluminium at total concentrations of >1000 µg AL/L.
The shape of the concentration–response curves for the other 3 microalgal species differed markedly from that of C. closterium (Figure 2 and Supplemental Data, Figure S1). The concentration range of declining growth rate for the other diatom M. polymorphus was much narrower (only 1 order of magnitude) than that for C. closterium, with the onset of declining growth rate occurring at above 660 µg total Al/L. There was a gradual decrease in growth rate to 80% at approximately 800 µg total Al/L and then a very sharp drop in growth rate (80% to less than 10%) at approximately 900 µg/L coincident with the appearance of particulate aluminium. At 1000 µg total Al/L, particulate aluminium was approximately 120 µg/L. It is most likely that both dissolved and particulate aluminium contributed to toxicity in this diatom species.
For D. tertiolecta, there was a sharp drop in algal growth rate above 890 µg total Al/L. Growth rate reached 50% of the control where dissolved aluminium reached a maximum of 1700 µg/L and total aluminium concentrations exceeded 4000 µg/L (Figure 2 and Supplemental Data, Figure S1). This region of the concentration–response curve was dominated by particulate aluminium and there was little change in dissolved aluminium concentrations with increasing total aluminium concentration. Growth rate declined further to 30% of the control when total aluminium reached 43 000 µg/L. Growth rates showed the best relationship with total aluminium, suggesting that a combination of particulate and dissolved forms contributed to toxicity.
For Tetraselmis sp., there were large differences between total and dissolved aluminium concentrations because of the very high aluminium concentrations used (predominance of particulate aluminium). A gradual decline in growth rate with total aluminium concentrations was observed that was not related to changes in dissolved aluminium concentration (Figure 2 and Supplemental Data, Figure S1). When total aluminium exceeded 1000 µg/L at the end of the bioassay, growth rate declined and yet dissolved aluminium concentrations were in the same range as those where there was no effect on growth rate (Supplemental Data, Figure S1). Therefore, toxicity was the result of the increasing particulate aluminium concentration with increasing total aluminium (>1000 µg total Al/L) rather than the dissolved aluminium concentration. Out of all the toxicity tests conducted, this species showed the strongest relationship with particulate aluminium.
There is little published information available on the relationship between aluminium speciation (i.e., filtered versus unfiltered) and toxicity to algae in marine waters to compare with the data in the present study. Harford et al. [13] determined 72‐h IC10 and IC50 values of 14 µg Al/L (3 µg/L–25 µg/L, 95% confidence limits) and 190 µg Al/L (140 µg/L–280 µg/L, 95% confidence limits) respectively, using the tropical strain of C. closterium at 32 °C. The 72‐h IC10 concurred with that derived in the present study (Table 1) using the temperate strain of the same species. Information on aluminium speciation and effects on algae in freshwaters [22, 23] is more readily available but has focused on situations of lower pH where aluminium speciation is dominated by the monomeric trivalent forms (Al3+), hydroxy forms (Al(OH)2+, Al(OH)2+), and organic complexes. These forms are not comparable with aluminium speciation and toxicity at the pH, DOC, and ionic composition of seawater.
Aluminium in marine waters may be directly toxic to the algae by adsorption to the cell wall followed by internalization, transformation, and interference with cellular metabolic processes, as is speculated to occur with freshwater algae [24]. Aluminate could be transported into the cell via anion membrane channels, but the mechanism by which the neutral but polar dissolved Al(OH)30 form may interact with the cell membrane is unknown. The increased diffusive barrier formed via smothering of algal cells by aluminium hydroxide in particulate forms may reduce the transport of nutrients and gases and act as a form of photosynthetic shade or interfere with the movement of flagella (in the case of the flagellate green alga Tetraselmis sp.). Such a diffusion barrier may be formed if excessive clumping of algal cells were to occur by flocculation and deposition of cells in the presence of aluminium. However, examination of the sensitive diatom C. closterium by microscope found no correlation between cell clumping and aluminium concentration. In addition, indirect mechanisms of aluminium toxicity are possible whereby aluminate competes for binding and uptake of anionic nutrients such as nitrate or phosphate. However, aluminate outcompeting nutrient uptake is unlikely given that the algal bioassays were enriched with nitrate and phosphate to ensure nutrients were not limiting. Effects of aluminate on diatom growth rate via depletion of the essential nutrient silica as a result of precipitation with aluminum also seem unlikely. Silica was present in molar excess at low aluminium concentrations (<100 µg Al/L) where diatom growth was affected and there was no marked loss of dissolved aluminium as would be expected if precipitation with silica had occurred. However, in the presence of particulate aluminium, silicates may coprecipitate, resulting in silica depletion at the higher end of the concentration–response curve thereby having no impact at the 10% effect level. The mechanism of aluminium toxicity to marine algae and in particular to the more sensitive marine diatoms is the subject of ongoing research.
Mussel, oyster, and sea urchin embryo development
A distinctive feature of the mussel and oyster embryo development tests was the steepness of the concentration–response curve, with the appearance and dominance of embryo abnormalities occurring over an extremely narrow concentration range (250 µg–320 µg total Al/L for the mussel and 240 µg–760 µg total Al/L for the oyster; Figure 3). The toxic response for both mussel and oyster embryos occurred well within the solubility range for aluminium in seawater (B.M. Angel et al., unpublished manuscript) indicating that it was the dissolved forms of aluminium affecting embryo development with a minor contribution from particulate aluminium. The 72‐h IC10 for the mussel was 250 µg Al/L, and the 48‐h IC10 for the oyster was 410 µg Al/L. After the diatom, these molluscs were the next most sensitive organisms to aluminium of all the species tested (Table 1).
In contrast, the sea urchin embryo development was not adversely affected at the highest aluminium concentration tested, and therefore 28 000 µg/L was used as a surrogate for the NOEC in the species sensitivity distribution. Although no effects to sea urchin development were observed in the present study, embryo development of other sea urchin species has been affected by exposure to aluminium (2.7 µg–2700 µg total Al/L [25, 26]. These studies did not report their method of preparing the test solutions or measurements of pH in the test solutions, making it difficult to confirm whether aluminium concentrations or declining pH as a result of aluminium hydrolysis was responsible for the biological effects measured. Differences in species sensitivity may also explain the difference in results from the present study. These issues also relate to the observed toxicity to mussel embryo development at much lower aluminium concentrations (nominal NOEC = 2.7 µg Al/L) than in the present study, as recorded by Pagano et al. [26]. A nominal chronic 48‐h NOEC of 100 µg Al/L for embryo development of the oyster Saccostrea glomerata (formerly known as Saccostrea commercialis) was comparable with the present study [27]. Although a nominal concentration was used, the authors noted that the nominal concentrations were within 10% of the measured concentration.
Mechanisms of aluminium toxicity to bivalve embryo development have not been determined. Aluminate may be transported into cells via anion membrane channels and disrupt normal cellular activity, but the mechanism by which the dissolved aluminium hydroxide disrupts normal embryo development is unknown.
Macroalgal germination success
The germination success of the macroalgal species H. banksii was not significantly affected by aluminium over the concentration range tested (Table 1 and Figure 4). Therefore, the highest measured test concentration of 9800 µg total Al/L was used as the surrogate for the NOEC for H. banksii in the species sensitivity distribution. The germination success data for E. radiata were quite variable; however, some decrease in response was observed at aluminium concentrations in excess of 2000 µg/L, indicating the role of particulate aluminium in determining toxicity (Figure 4). A 10% inhibition of germination was calculated to occur at 6800 µg total Al/L. This value was used in the species sensitivity distribution.
Juvenile fish swimming imbalance and growth
Aluminium was not significantly toxic to either fish species tested. All 5 damselfish in a single replicate at 12 µg Al/L died on day 6 of the 7‐d test. This was considered an anomalous result as no other mortality in the test occurred, and physicochemical conditions were within the physiological limits of the fish. Juvenile fish growth in both species was highly variable, but the average fish growth increased by factors of 1.3 and 3.0 in the barramundi and damselfish, respectively, relative to the control fish by day 7 (Figure 5). No adverse effects on either swimming imbalance or growth were observed at the highest measured aluminium test concentration (10 000 µg total Al/L) for either fish species. Because the duration of the bioassay was less than 21 d, it was defined as an acute exposure rather than chronic [7] and was therefore not used to derive the guideline value but provides useful toxicity data in the context of an acute exposure. The absence of toxicity to fish in marine waters is in contrast to the aluminium toxicity to fish in more acidic freshwaters that has been widely observed [22, 28, 29]. The mechanisms of toxicity include interference with ionoregulatory cell function at the gill and decreased respiratory function possibly as a result of aluminium precipitation and excess mucus production at the gill surface [22]. Aluminium in estuarine systems has also been shown to be of concern to fish with the formation on fish gills of polymeric aluminium that has been mobilized from aluminium colloids when acidic freshwater meets the higher pH and ionic strength of seawater [30]. However, toxicity of aluminium to fish in full‐strength seawater has not previously been identified in the literature [22, 29] as an issue of concern for fish and this was supported by the results for 7‐d exposures for juvenile barramundi and damselfish in the present study.
Aluminium guideline derivation
The derivation of the water quality guideline involved the generation of a species sensitivity distribution from the chronic test data [4]. Although chronic NOECs have traditionally been used, their limitations are increasingly recognized, and IC10 or EC10 values are now preferred [31, 32]. The chronic toxicity dataset generated in the present study comprised 9 species representing 5 taxonomic groups. Relevant literature toxicity values for 3 species (1 existing species and 2 new species, 1 of which was from a new taxonomic group) were added to the database, as follows.
First, as already discussed, Harford et al. [13] determined a chronic 72‐h IC10 of 14 µg Al/L for growth rate inhibition for a tropical strain of the diatom C. closterium (the same species used in the present study). As prescribed by ANZECC/ARMCANZ [4], for deriving guideline trigger values, the geometric mean (16 µg Al/L) of the IC10 values for both strains was used in the species sensitivity distribution (Table 1).
Second, Negri et al. [33] determined a chronic 18‐h EC10 of 1300 µg Al/L for metamorphosis of the coral Acropora tenuis representing the cnidarian taxonomic group.
Third, the NOEC value (100 µg Al/L) reported by Wilson and Hyne [27] for the oyster S. glomerata was valuable in that it provided an additional toxicity value at the lower end of the species sensitivity distribution.
The final dataset used in the species sensitivity distribution comprising 11 species from 6 taxonomic groups (Table 1) was used to derive guideline values to protect 90%, 95%, and 99% of species with 50% confidence (Table 1, Figure 6).
Species sensitivity distribution of toxicity data for 11 marine species from 6 taxonomic groups as a function of total measured aluminium (at bioassay initiation). Data were fitted using a Burr type III model derived from BurrliOZ V2 software.
A number of statistical models are available for describing species sensitivity distributions that include both parametric and nonparametric methods [34, 35, 36, 37, 38]. Commonly used parametric methods that have been used include: logistic, log‐logistic, log‐normal, Burr type III, Gompertz, and log‐triangular. In Australia and New Zealand, the Burr type III distribution was adopted in 2000 as the preferred distribution to be fitted to the toxicity data [38]. The recommended 3‐parameter Burr type III statistical distribution was therefore used to fit the data to a species sensitivity distribution, and a visual assessment of the goodness of fit was made [4, 7, 8].
A value of 24 µg/L total aluminium was derived for aluminium in marine waters for 95% species protection (Figure 6). For 99% and 90% species protection, the values were 2.1 µg total Al/L and 69 µg total Al/L, respectively. The species protection levels are defined by ANZECC/ARMCANZ [4] as the 95% protection value that applies to slightly to moderately disturbed waters such as harbors and coastal waters impacted by anthropogenic activities, whereas the 99% protection value is for pristine waters or those of high conservation value. The 90% value represents a highly disturbed ecosystem.
The dataset exceeded the recommended criterion minimum of 8 species from 4 taxonomic groups to be adopted in the revised Australian and New Zealand guidelines [7] and is consistent with dataset requirements adopted worldwide. For example, the Canadian water quality guidelines require studies on fish, invertebrates, and plants for both fresh and marine waters [38]. From a tabulated list of minimum data required for specific taxonomic groups, it was anticipated that “generally at least 10 to 15 data points should be available,” although it was acknowledged that on occasions fewer data may be acceptable to produce an adequate curve. A goodness of fit test was included as a requirement [38].
Under this new guideline value, all species except the most sensitive diatom C. closterium (both tropical and temperate strains) would be protected by the value of 24 µg/L derived using the full dataset. The 95% confidence limits on the individual and geometric mean IC10 toxicity values for temperate and tropical C. closterium overlap with the new guideline value (Table 1), suggesting protection equivalent to a 10% inhibition effect on growth rate for this species. The 99% species protection value of 2.1 µg/L is close to the analytical detection limit for aluminium in the present study but is still above background aluminium concentrations in relatively uncontaminated coastal water systems (J. King, 2013, Honours thesis, University of Wollongong, Wollongong, Australia). The 99% protection value is derived from the lower end of the species sensitivity distribution where it is a less than reliable extrapolation. If a more reliable 99% protection value is required, then additional site‐specific testing might be required using sensitive species.
To derive this water quality guideline, care was taken to acknowledge the importance of aluminium speciation in seawater and how it relates to toxic effects on marine biota over a broad sensitivity distribution. In particular, the present study recognizes the significance of metal solubility in seawater by demonstrating that the dissolved forms of aluminium dominate speciation below approximately 500 µg total Al/L, however, between 500 µg/L and approximately 1000 µg/L, particulate aluminium hydroxide is also present and becomes increasingly dominant at total concentrations greater than 1000 µg/L. Of the 11 organisms tested, 3 species (C. closterium, M. edulis plannulatus, and S. echinata) were affected at the 10% effect level by dissolved aluminium only (Supplemental Data, Table S2), 2 species (M. polymorphus and D. tertiolecta) were affected by a combination of dissolved and particulate aluminium, 2 species (Tetraselmis sp. and E. radiata) were affected by particulate aluminium only, and 4 species (H. banksii, H. tuberculata, A. polyacanthus, and L. calcarifer) were not affected by the highest aluminium concentration tested (9800 µg/L–28 000 µg/L). With just 3 species affected by dissolved aluminium only, it was not possible to use a species sensitivity distribution to derive a water quality guideline for dissolved aluminium. By using all of the toxicity data based on measured total aluminium, a guideline value was derived that protects 95% of the species. At this concentration, all of the aluminium is dissolved. In this way a conservative level of protection is afforded from both forms of aluminium in seawater. In reality, aluminium precipitates are not likely to be present unless catastrophically high levels of aluminium are released into seawater, for example, in the case of a spill.
Other particulate forms of aluminium exist in seawater as mineralized clay, soil, and sediment in suspension that are assumed to be less toxic than the aluminium precipitates observed in the present study. To prevent a comparison of these forms of particulate aluminium with the water quality guideline, it is recommended that a 0.45‐µm filtered seawater sample that removes the mineralized aluminium be compared with the derived aluminium guideline.
An alternative approach to the development of a water quality guideline based on a species sensitivity distribution would be the application of a predictive model such as the biotic ligand model, which incorporates metal speciation in the external media and applies a mechanistic understanding of metal toxicity. This approach is attractive in the case of aluminium in marine waters as it would potentially allow separate consideration of the toxicity of dissolved and precipitated aluminium species (i.e., by incorporating 2 or more biotic ligands with different mechanisms of toxicity). To advance this approach, however, there is a need to address knowledge gaps in terms of understanding the mechanisms of aluminium toxicity in marine waters.
It is perhaps surprising that aluminium appears more toxic in marine waters than in freshwaters at pH > 6.5, based on the derived freshwater value of 54 µg Al/L [4]. This freshwater guideline value was derived from acute data converted to chronic using an arbitrary acute to chronic ratio of 10, so it is only considered to be of moderate reliability. There is a need for further research to improve the freshwater dataset and investigate toxicity mechanisms and related aluminium speciation.
CONCLUSIONS
A new guideline value of 24 µg total Al/L for 95% species protection has been derived based on chronic effect values from 11 species from 6 taxonomic groups and a visual assessment of the model fit. This value is significantly higher than the low‐reliability environmental concern level of 0.5 µg/L [4]. The application of this new guideline value will permit more reliable risk‐based management of aluminium toxicity in marine waters. By using toxicity values in the species sensitivity distribution based on total aluminium rather than dissolved aluminium, the 2 possible phases of aluminium speciation (dissolved and particulate) that could cause toxicity are accounted for.
Toxicity to the most sensitive species, the diatom C. closterium, the mussel M. edulis plannulatus, and the oyster S. echinata, was shown to be the result of dissolved aluminium (aluminate and Al(OH)30), whereas toxicity to the diatom M. polymorphus and green alga D. tertiolecta was attributable to both dissolved and particulate forms of aluminium. In contrast, aluminium toxicity to the green flagellate Tetraselmis sp. was the result of particulate aluminium only. The mechanisms of toxicity to these species tested were not identified. Further research into the mechanisms of toxicity is especially important for the diatom C. closterium, as this species proved to be the most sensitive to aluminium in marine waters.
SUPPLEMENTAL DATA
Tables S1–S2.
Figures S1–S3. (267 KB PDF).
Acknowledgment
We gratefully acknowledge J. King and C. Jarolimek, CSIRO Land and Water, for their assistance with chemical analyses. We also thank M. Adams and S. Simpson, CSIRO Land and Water, for their helpful comments on the draft manuscript.
