Environmental Toxicology and Chemistry, Vol. 34, No. 12, pp. 2856–2863, 2015 Published 2015 SETAC Printed in the USA

MANGANESE TOXICITY TO TROPICAL FRESHWATER SPECIES IN LOW HARDNESS WATER ANDREW J. HARFORD,* THOMAS J. MOONEY, MELANIE A. TRENFIELD, and RICK A. VAN DAM Department of the Environment, Environmental Research Institute of the Supervising Scientist, Darwin, Northern Territory, Australia (Submitted 21 April 2015; Returned for Revision 7 June 2015; Accepted 22 June 2015) Abstract: Elevated manganese (Mn) is a common contaminant issue for mine water discharges, and previous studies have reported that its toxicity is ameliorated by Hþ, Ca2þ, and Mg2þ ions. In the present study, the toxicity of Mn was assessed in a high risk scenario, that is, the slightly acidic, soft waters of Magela Creek, Kakadu National Park, Northern Territory, Australia. Toxicity estimates were derived for 6 tropical freshwater species (Chlorella sp., Lemna aequinoctialis, Amerianna cumingi, Moinodaphnia macleayi, Hydra viridissima, and Mogurnda mogurnda). Low effect chronic inhibition concentration (IC10) and acute lethal concentration (LC05) values ranged between 140 mg L–1 and 80 000 mg L–1, with 3 of the species tested (M. macleayi, A. cumingi, and H. viridissima) being more sensitive to Mn than all but 1 species in the international literature (Hyalella azteca). A loss of Mn was observed on the final day for 2 of the H. viridissima toxicity tests, which may be a result of the complex speciation of Mn and biological oxidation. International data from toxicity tests conducted in natural water with a similar physicochemistry to Magela Creek water were combined with the present study’s data to increase the sample size to produce a more reliable species sensitivity distribution. A 99% protection guideline value of 73 mg L–1 (33466 mg L–1) was derived; the low value of this guideline value reflects the higher toxicity of Mn in slightly acidic soft waters. Environ Toxicol Chem 2015;34:2856–2863. # 2015 Commonwealth of Australia. Published by Wiley Periodicals, Inc. on behalf of SETAC. Keywords: Uranium mining

Aquatic toxicology

Water quality guideline

Metal toxicity

biota is low. This was reflected in the relatively high default 95% protection guideline value reported by the Australian and New Zealand Environment Conservation Council (ANZECC) and the Agriculture and Resource Management Council of Australia and New Zealand (ARMCANZ), of 1900 mg L–1 [7]. However, more recent studies have reported some particularly sensitive species, such as Hyalella azteca, with a low-effect chronic inhibition concentration (IC10) of 96 mg L–1 Mn [8,9]. A review of Mn toxicity in freshwaters for the United Kingdom Environment Agency recommended a predicted no effect concentration (PNEC) of 62 mg L–1 to 123 mg L–1 [9], which was based on a species sensitivity distribution of 12 toxicity estimates. The calculated hazardous concentration predicted to affect 5% of species (HC5; equivalent to a 95% guideline value) was 246 mg L–1. The PNECs were derived by applying application factors of 2 to 4 to the HC5. The use of an application factor is mandatory for deriving PNECs that are to be used as European environmental quality standards [10]. However, this led to an environmental quality standard that was too stringent for many waterways but was considered relevant to conditions of high bioavailability—that is, low pH, hardness, alkalinity, and dissolved organic carbon (DOC). To address this issue, the European Commission developed a biotic ligand model for Mn, which allowed researchers to adjust the environmental quality standard under different physicochemical conditions [11]. The biotic ligand model developed by Peters et al. [11] described the toxicity of dissolved Mn as a function of water quality, in particular competition between major ions and Mn for binding sites on or in organisms. They found that Ca2þ cations ameliorate the toxicity of Mn to fish and invertebrates, whereas Mg2þ cations ameliorate Mn toxicity to invertebrates but not to the extent of Ca2þ. In addition, increasing Hþ ions primarily ameliorated the toxicity of Mn to algae. The dependence of Mn toxicity on water hardness and pH is also consistent with what is known for other metals [11]. The results of Peters et al. [11] highlighted the higher risk of Mn toxicity to

INTRODUCTION

Manganese (Mn) is a ubiquitous element in the earth’s mantle and is present in most rocks and soil types [1]. It is an important constituent of several enzymes and cofactors, making trace amounts essential for many species [2]. It is commonly associated with the production of steel and batteries and is an important ingredient in products such as ceramics, varnishes, fertilizers, and pesticides. High concentrations of Mn are commonly found in discharges from mining, smelting, and other industrial processes [3]. The aquatic chemistry of Mn is a complex function of the pH and redox micro-environment, with Mn primarily existing as soluble Mn(II) and insoluble Mn(IV) oxidation states, the latter being less bioavailable and toxic. Increasing pH and redox of a solution generally results in precipitate formation as a result of the oxidation of Mn(II) to form Mn(III)/Mn(IV) oxyhydroxides [4]. These reactions are slow in the absence of a catalyst [4] but many aquatic bacteria use Mn(II) as a terminal electron acceptor during respiration, which results in the production of insoluble Mn(IV) oxides in the environment [5]. Richardson et al. [6] also reported that microalgae can form micro-environments of high pH and high oxygen, which promotes the formation of insoluble Mn oxide colloids. Hence, compared with other metals, the complex speciation of Mn may present challenges in determining the exposure conditions during laboratory toxicity tests, and detailed chemical analyses are essential to accurately measure exposure concentrations. Historically, results of laboratory toxicity testing have indicated that Mn is less of an environmental hazard than many other metals, with evidence from the literature suggesting that the acute and chronic toxicity of Mn to many freshwater All Supplemental Data may be found in the online version of this article. * Address correspondence to [email protected] Published online 26 June 2015 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/etc.3135 2856

Mn toxicity to tropical freshwater species

fish and invertebrates in acidic, Ca-deficient waters, because these conditions can markedly enhance the uptake and toxicity of metals, including Mn. Such water types, although not common, are distributed globally, particularly throughout northern Australia, Scandinavia, eastern North America, and Southeast Asia [12,13]. Manganese is present in substantial concentrations in mine waters at the Ranger Uranium Mine, with concentrations of 2000 mg L–1 in process water, which is never discharged untreated, and up to 160 mg L–1 in the release waters, which are discharged to the environment [14,15]. The mine is located in the wet–dry tropics of northern Australia and is surrounded by the World Heritage-listed Kakadu National Park. Consequently, it operates under strict environmental regulations to ensure the aquatic ecosystems downstream of the mine are protected [16]. Evaporation and land irrigation practices are used onsite to reduce the mine water inventory during the dry season. However, releases of mine water containing solutes to an adjacent waterway, Magela Creek, are required each wet season (December to April) to manage the water inventory effectively. The slight acidity (pH 5.5–6.5) and low ionic strength (water hardness, 3–6 mg L–1 as CaCO3; alkalinity, 5–10 mg L–1 as CaCO3; and electrical conductivity, 5–20 mS cm–1) of Magela Creek water indicated the need to develop a site-specific Mn guideline value, because these conditions fall outside the scope of the Mn biotic ligand model (i.e., pH of 5.7–8.7 and Ca of 6.4– 200 mg L–1). Moreover, Mn toxicity data for tropical soft freshwater species were not sufficiently available to conclude, with high confidence, that no adverse effects would occur to the aquatic ecosystem from mine water discharges into Magela Creek. To ensure the environmental protection of this unique ecological region, an assessment of the toxicity of Mn was conducted with 2 aims: to assess the toxicity of Mn in low pH, low hardness water to 6 tropical freshwater species and to derive guideline values for Mn in low pH, low hardness freshwater, which could be incorporated into the water quality objectives for the creeks adjacent to the Ranger Uranium Mine. METHODS

Permissions and ethics statement

The Northern Land Council granted permission for water collections at Bowerbird Billabong (latitude 128460 1500 , longitude 1338020 2000 ). No endangered or protected organisms were used in the present study, and approval for the ethical use of the fish, M. mogurnda, and the use of the survival endpoint was granted through the Charles Darwin University’s Animal Ethics Committee (license number A12028). Ethics approval for the use of the other organisms was not required. General laboratory procedures

All equipment that test organisms or media came in contact with or were exposed to was made of chemically inert materials (e.g., Teflon, glass, or polyethylene). All plastics and glassware were washed by soaking in 5% (v/v) nitrix acid (HNO3) for 24 h before being washed with a nonphosphate detergent (neodisher1 LaboClean FLA; Dr. Weigert) in a laboratory dishwasher operated with reverse osmosis/deionized water (Elix; Millipore). Glassware used in the toxicity tests was silanized with 2% dimethyldichlorosilane in 1,1,1-trichloroethane (Coatasil; AJAX) to reduce Mn adsorption to the glass. All reagents used were analytical grade, and stock solutions were made up in ultrapure water (18 MV, Milli-Q Element; Millipore).

Environ Toxicol Chem 34, 2015

2857

Test diluent

Natural Magela Creek water was used as the control treatment and diluent in all tests and was obtained upstream of the Ranger Uranium Mine at 2 sites: 1 site close to the mine (latitude 128400 2800 , longitude 1328550 5200 ) during the wet season, and 1 site further upstream, at Bowerbird Billabong (latitude 128460 1500 , longitude 1338020 2000 ), during the dry season. The natural waters were collected in 20-L acid-washed plastic containers and transported by road (2.5 h) to the laboratory at ambient temperature. At the laboratory, they were filtered within 2 d of collection (3.0 mm, Sartopure PP2 depth filter MidiCaps, Sartorius). The waters were stored at 4  1 8C and kept for up to 1 mo following collection. Given the high volumes of water required for the Amerianna cumingi tests, the Magela Creek water diluent used in those tests was not prefiltered. This had the potential to introduce coarse particulates and wild zooplankton into the test but all test solutions were visibly free of such issues. Test solutions were subsampled for physicochemical analyses. Specifically, pH, dissolved oxygen, electrical conductivity, and DOC were measured in-house. Additional subsamples were sent to an environmental chemistry laboratory (Envirolab) to measure alkalinity (APHA2320B; Supplemental Data, Table S1) and a limited metal and major ion suite (totals only; Al, Cd, Co, Cr, Cu, Fe, Mn, Ni, Pb, Se, U, Zn, Ca, Mg, Na, and SO4 [inferred from S]) by inductively coupled plasma–mass spectrometry (ICP-MS) and inductively coupled plasma–atomic emission spectroscopy (ICP-AES; Supplemental Data, Table S2). Toxicity tests

The toxicity of Mn was assessed using 6 Australian tropical freshwater species: the unicellular green alga (Chlorella sp.), the duckweed (Lemna aequinoctialis), the green hydra (Hydra viridissima), the cladoceran (Moinodaphnia macleayi), the aquatic snail (Amerianna cumingi), and the Northern trout gudgeon (Mogurnda mogurnda). All organisms were isolated from soft surface waters in Kakadu National Park and have been cultured continuously in the laboratory for 10 yr to 25 yr, depending on the species. The test methods are described in detail by Riethmuller et al. [17] and for A. cumingi by Houston et al. [18]. Key details of each test are provided in Table 1. With the exception of the M. mogurnda test, which measured larval fish survival, the tests examined sublethal endpoints over chronic or subchronic exposure periods. For the acute M. mogurnda survival test, the fish were monitored every 24 h. At the end of the test, the animals were sacrificed using anaesthetic MS-222 (Sigma-Aldrich). For the L. aequinoctialis and Chlorella sp. tests, nutrients (nitrate: NO3– and phosphate: PO43–) were added at the minimum concentrations that would sustain acceptable growth (Table 1). The Magela Creek water used in the Chlorella sp. tests also had 1 mM HEPES (N-2hydroxyethylpiperazine-N0-2-ethanesulfonic acid) buffer added to maintain stable pH. The Magela Creek water diluents were spiked with Mn using stock solutions of either 52.5 mg L–1 or 26.2 g L–1 Mn sulfate (MnSO4H2O; Sigma-Aldrich). Concentrations of dissolved Mn (0.1 mm filtered) were measured before and after the test exposure through ICP-MS analysis (Supplemental Data, Table S3). A minimum of 2 valid toxicity tests was completed for each species. Fate of manganese in the H. viridissima test

Due to observed losses of Mn in the first H. viridissima toxicity tests, extensive chemical analyses were conducted during the second test in an attempt to recover Mn from the test

2858

Environ Toxicol Chem 34, 2015

A.J. Harford et al.

Table 1. Details of toxicity tests for the 6 Australian tropical freshwater species used to assess the toxicity of manganesea Species (common name)

Test duration and endpoint

Control response acceptability criterion

Temperature; light intensity; photoperiod

Feeding/ nutrition

Chlorella sp. 72-h population 1.4  0.3 doublings day–1; 29  1 8C; 14.5 mg L–1 NO3 (unicellular green growth rate %CV < 20% 100–150 mmol m–2 sec–1; 0.14 mg L–1 PO4 alga) 12:12h 29  1 8C; 3 mg L–1 NO3 Lemna aequinoctialis 96-h surface Mean surface area growth (tropical duckweed) area growth rate rate (k, cm2 day–1)  0.40; 100–150 mmol m–2 sec–1; 0.3 mg L–1 PO4 %CV < 20% 12:12h Hydra viridissima 96-h population Mean population growth 27  18C; 3–4 Artemia (green hydra) growth rate rate (k, day –1)  0.27; % 30–100 mmol m–2 sec–1; nauplii day–1 CV < 20% 12:12h 30 ml FFVc and Moinodaphnia 3-brood Mean adult survival 27  1 8C; macleayi (120–144 h)  80%; mean neonates 30–100 mmol m–2 sec–1; 6  106 cells of 12:12h Chlorella sp. d–1 (cladoceran) reproduction per adult  30 2 cm2 lettuce Amerianna cumingi 96-h Mean eggs per snail pair 30 8C; 30–100 mmol m–2; sec–1; 12:12h disc per snail per (Aquatic snail) reproduction 100; %CV < 30% day Nil Mogurnda mogurnda 96-h survival Mean larval survival 27  1 8C; (Northern trout  80%; %CV < 20% 30–100 mmol m–2 sec–1; 12:12h gudgeon)

Test No. replicates (individuals per volume b (mL) replicate)

Water changes

2 (3  104 cells mL–1)

50

Static

2 (4 with 3 fronds)

100

Static

2 (10)

30

Daily renewals

5 or 10 (1)d

30

Daily renewals

3 (12)

1750

Daily renewals

2 (10)

30

Daily renewals

a

Full details of the methods are provided in Riethmuller et al. [17] and Houston et al. [18]. Replication was reduced for all tests but the snail and cladoceran in order to increase the number of treatments, but the control had 3 replicates. c Fermented food with vitamins (FFV) represents an organic and bacterial suspension prepared by method described in Riethmuller et al. [17]. d The first M. macleayi was a modified design with 5 replicates but the second had 10 replicates. %CV ¼ percent coefficient of variation. b

system and determine where it was being lost. This involved analyzing total Mn from the end of test solutions, from the tissue of H. viridissima, and from the petri dishes following a 5% HNO3 rinse. The results of these analyses were inconclusive, however, and an additional experiment was conducted to further assess the fate of Mn in the hydra test system. This also involved investigating if Mn loss could be reduced through “priming” the Petri dishes with a 24-h pre-exposure to a solution of 250 mg L–1 Mn. The Mn fate test was conducted at 3 Mn concentrations in Magela Creek water (background, 250 mg L–1 and 600 mg L–1), and additional primed Petri dish treatments were included for each of these Mn concentrations. Measurements of total and 0.1 mm filtered Mn were made on the following components of the test system: test solutions from the Petri dishes at test commencement and every 24 h prior to test solution renewal, until the end of the test (96 h); test solutions from the 5-L test solution storage bottles at the commencement and end of the test; the surface of the test Petri dishes, following rinsing with 5% HNO3 (total Mn only); and hydra tissue at the end of the test (total Mn in all hydra). Hydra tissues were digested in HNO3 prior to ICP-MS analysis. Concentrations of Mn could only be analyzed as total Mn recovered from the tissue. Quality control

Manganese chemistry. Water samples for chemical analyses (total and 0.1 mm filtered) were collected and analyzed at both the start and end of all toxicity tests to determine the fate of the added Mn. Filtration through 0.1-mm membranes, rather than the conventional 0.45-mm filtration, was used specifically to provide increased ability to identify whether Mn was present in colloidal form. Water chemistry. For each test, blanks, procedural blanks (ultrapure water that has been exposed to all components of the test system) and control waters (Magela Creek water) were analyzed for a limited metal and major ion suite (Al, Cd, Co, Cr, Cu, Fe, Mn, Ni, Pb, Se, U, Zn, Ca, Mg, Na, SO4 [inferred from S]) by ICP-MS and ICP-AES (Supplemental Data, Table S2).

Chemistry data for the blanks and procedural blanks were assessed initially by searching for analyte concentrations higher than detection limits. Where these concentrations were greater than 1 mg L–1 and above background levels of Magela Creek water, duplicate procedural blank samples were reanalyzed or the control water concentrations were compared with those in tests without blank contamination to determine if the contamination was limited to the 1 sample bottle or experienced throughout the test. General water quality. For each test, data were considered acceptable if the recorded temperature of the incubator remained within the prescribed limits (see test description, Table 1), the recorded pH was within  1 unit of value at test commencement (i.e., day 0), the electrical conductivity for each test solution was within 10% (or 5 mS cm–1 for samples with low conductivity) of the values at test commencement, and the dissolved oxygen concentration was greater than 70% throughout the test. The occurrence of any significant water quality changes were investigated and discussed on a case-by-case basis. Statistical analysis

Toxicity estimate calculations. For each species, individual tests were pooled and the raw data analyzed. Non-linear regressions (3-parameter log-logistic) were used to determine point estimates of inhibitory concentrations that reduced endpoint responses by 10% and 50% (IC10 and IC50, respectively) relative to the control responses (CETIS Ver 1.8.7.4; Tidepool Scientific Software). For the acute M. mogurnda lethality test, the standard 50% effect/lethal concentration (LC50) and a more conservative 5% effect/lethal concentration (LC05) were estimated. A more conservative approach was required for the M. mogurnda larval survival test because it measures an acute response. Guideline value derivation. Using the calculated IC10 values, the species sensitivity distribution method (BurliOz 2.0, CSIRO) was used to derive 99%, 95%, and 80% protection

Mn toxicity to tropical freshwater species

guideline values for Mn in low pH, low hardness waters. To improve the fit of the distribution, 3 extra toxicity estimates from international studies conducted in waters with physicochemical conditions closely related to Magela Creek were added to the local species dataset. Specifically, toxicity estimates from the temperate, northern hemisphere species, Pseudokirchneriella subcapitata (alga), Ceriodaphnia dubia (cladoceran), and Pimephales promelas (fish) were added to the species sensitivity distribution [11], which increased the sample size from 6 to 9 species. These toxicity tests were conducted at 25 8C in a natural soft water (hardness ¼ 12 mg L–1 CaCO3, Ca ¼ 4 mg L–1) with a pH of 6.7 and 12 mg L–1 DOC. Although the DOC was 4 times higher than that in Magela Creek water (typically A. cumingi > M. macleayi >> L. aequinoctialis > Chlorella sp. >> M. mogurnda. Fate of manganese in the H. viridissima toxicity test

The loss of Mn from the first H. viridissima toxicity test was characterized during the second toxicity test. Manganese measurement in the test solutions at the end of the 96-h exposure period showed 0.1 mm, which had dissolved by the following day. Observed Mn loss was greatest on day 4, when measured on a day-by-day basis. Approximately 20% more Mn was recovered from the primed dishes of the 600 mg L–1 treatments compared with the unprimed dishes. However, priming the test dishes had no discernible effect in the 250 mg L–1 treatments (Figure 3). The higher loss of Mn on day 4 coincided with the observation of a floating precipitate (presumably a form of Mn-oxyhydroxide, although this precipitate was not characterized), particularly in the 600 mg L–1 treatment. There was a relationship between measured Mn concentrations in the hydra tissue and nominal Mn concentrations, which was consistent between the 2 tests where this was analyzed. There was, on average, 3.0 mg Mn in the tissue of all hydra exposed to 250 mg L–1 Mn and 4.2 mg Mn in those exposed to 600 mg L–1 Mn. However, the total amount of Mn recovered from hydra was not high enough to account for the observed loss from the system. Derivation of guideline values for low pH, low hardness waters

Combining the toxicity estimates from the present study (Table 2) with the international data [11] produced a species sensitivity distribution that appeared to fit the Burr type III statistical distribution (Figure 4). This species sensitivity distribution was used to derive a 99% protection guideline value of 73 mg L–1 Mn (33–466 mg L–1 Mn). The calculated 95% and 80% protection guideline values were 153 mg L–1 Mn (75–724 mg L–1 Mn) and 443 mg L–1 Mn (201–1909 mg L–1 Mn), respectively. DISCUSSION

The results from the present study found that 3 of the tropical species tested—M. macleayi, A. cumingi, and H. viridissima— were more sensitive to Mn than most of the species in the

2860

Environ Toxicol Chem 34, 2015

A.J. Harford et al.

Figure 1. Manganese concentration–response relationships for the 6 species tested. Data points represent the mean  standard error of 2 to 3 replicates, except for Moinodaphnia macleayi (n ¼ 5 to 10 replicates). Symbols indicate the first (circle), second (square), and third (diamond) toxicity tests. Toxicity estimates for all species were determined using 3-parameter logistic models.

international literature. Only 1 other species—the amphipod, Hyalella azteca—was more sensitive [8,9]. The high sensitivity of the 3 species may be attributed at least partially to the low pH and hardness of Magela Creek water diluent. Previous studies have specifically demonstrated the amelioration of Mn toxicity by increasing water hardness [19]. Research involving the development of a biotic ligand model for Mn also reported that

there is competition between Mn and cations in solution, primarily Hþ and Ca2þ [11]. The mechanism(s) of Mn toxicity in aquatic invertebrates are not well understood, but studies have demonstrated the metal’s ability to inhibit various Camediated physiological processes in mammals [20] and invertebrates [21]. Considering this, it would be of interest to determine if there is a common mechanism of action, which

Mn toxicity to tropical freshwater species

Environ Toxicol Chem 34, 2015

2861

a

Table 2. Summary of the manganese toxicity estimates to 6 local freshwater species in Magela Creek Water Species Chlorella sp. Lemna aequinoctialis Hydra viridissima Moinodaphnia macleayi Amerianna cumingi Mogurnda mogurndab

IC10 (mg L–1) 12103 (10–14  103) 2200 (910–3400) 140 (100–180) 610 (500–690) 340 (80–920)

IC50 (mg L–1) 60103 11103 1380 1100 5660

(55–70  103) (9–13  103) (1200–1560) (1030–1170) (2830–12660)

LC05 (mg L–1)

LC50 (mg L–1)

80  103 (40–110  103)

240  103 (200–320  103)

The toxicity estimates were based on Mn concentrations calculated by averaging the before and after test 0.1-mm filtered Mn concentrations in the test solutions. Raw data and statistical analyses are in Supplemental Data, Table S5. Toxicity estimates for M. mogurnda are LC05 and LC50. IC10 ¼ concentration that results in a 10% reduction in growth rate relative to controls; IC50 ¼ concentration that results in a 50% reduction in growth rate relative to controls; LC05 ¼ concentration that results in 5% reduction in the survival of the fish; LC50 ¼ concentration that results in 50% reduction in the survival of the fish. a

b

specific guideline value that was more applicable when compared against the background environmental data and was also comparable to the recently derived European environmental quality standard for Mn. The environmental quality standard developed for Mn in European water bodies was problematic due to regular exceedance in European waters. The European Commission’s solution was to recommend that the environmental quality standard was useful only in situations where Mn was of highest

a) Total Manganese 700 Unprimed plates Primed plates

600

Manganese (µg L 1)

might involve the inhibition of Mg and Ca mediated physiological processes. A species sensitivity distribution using toxicity estimates from the 6 local species tested in the present study resulted in a 99% guideline value of 4.1 mg L–1 Mn (1.0–182 mg L–1 Mn). Applying this guideline value in Magela Creek was impossible because it was lower than 50% of the Mn measurements reported at upstream (background) and downstream monitoring sites in the creek (Figure 5). The low guideline value was a result of the large difference in toxicity estimates between species used in the species sensitivity distribution, where the high toxicity estimates in the distribution skewed the lower tail of the model to lower concentrations. Regardless, it is well documented that small sample sizes can yield unreliable estimates and that the inclusion of extra toxicity estimates increases the reliability of a guideline value [22]. The European Commission and Australia now recommend that a minimum of 8 toxicity estimates are needed for a high reliability guideline value [10,23]. Inclusion of relevant nonlocal data to site-specific data is a method recommended by ANZECC and ARMCANZ [7], provided that the toxicity tests were conducted under relevant physicochemical conditions. In this case, 3 toxicity estimates for nonlocal species were identified as being conducted in natural water with sufficiently low hardness (12 mg L–1 CaCO3, Ca ¼ 4 mg L–1) and a temperature similar to that used for the site specific species (25 8C compared with 27–29 8C). Inclusion of the toxicity estimates from P. subcapitata, C. dubia, and P. promelas to the SSD with the 6 local species yielded a site-

500

400

300

200

100

0 0

20

40

60

80

100

120

Time (h) b) 0.1 µm Filtered Manganese 700 Unprimed plates Primed plates

Manganese (µg L 1)

600

500

400

300

200

100

0 0

20

40

60

80

100

120

Time (h)

Figure 2. Percent recovery of manganese from test solutions, Petri dishes, and hydra at the end of the second Hydra viridissima toxicity test. Samples from each replicate were pooled for chemical analysis.

Figure 3. Loss of total Mn (a) and dissolved Mn (b) from the test solutions during the 96-h exposure manganese fate test.

2862

Environ Toxicol Chem 34, 2015

Figure 4. Species sensitivity distribution of manganese toxicity estimates for the 6 local species (Chlorella sp., Lemna aequinoctialis, Amerianna cumingi, Moinodaphnia macleayi, Hydra viridissima, and Mogurnda mogurnda) and including 3 toxicity estimates from international datasets (Pseudokirchneriella subcapitata, Ceriodaphnia dubia, and Pimephales promelas [10]).

bioavailability. To predict Mn toxicity for waters with other physicochemical conditions, a biotic ligand model was developed [9,11]. Unfortunately, the biotic ligand model could not be used for Magela Creek, because the model is not validated for the pH and hardness characteristics of the creek. Mn loss during testing

Potential sources of Mn loss from the toxicity test included adsorption to the test solution bottles or the test containers, precipitation, or adsorption/absorption by the test animals. Attempts to recover Mn from the H. viridissima tests were unproductive. The unrecoverable Mn may have been bound to

A.J. Harford et al.

the Petri dishes, and the 5% HNO3 acid-extraction may have been insufficient to extract the Mn bound to the dishes. However, if this Mn was adsorbed to the test dishes, a decrease in daily loss of the metal over the test period would have been expected [24]; yet the greatest loss was observed on day 4. Priming the test Petri dishes with Mn reduced the loss in the 600 mg L–1 treatment by approximately 20%. However, this had no discernible effect in the 250 mg L–1 treatment. Measured Mn in the hydra tissues at the end of the tests showed a good relationship between nominal Mn concentration and hydra tissue concentrations, but the amount recovered did not account for the missing proportion of Mn. This suggests that the Mn was not bound to the dishes or absorbed/adsorbed by the hydra. Extensive chemical analysis of the old waters (i.e., those exposed to the hydra for 24 h) showed that the loss was markedly higher on the last day of the test. This time also coincided with the appearance of a white floating precipitate that was suspected to be an oxy-hydroxide of Mn. The sudden loss of the Mn on the final day suggests that the reaction may be biologically catalyzed. Indeed, H. viridissima contain a symbiotic Chlorella sp. that may be producing Mn-oxidizing microenvironments described by Richardson et al. [6]. In addition, Mn-oxidizing bacteria are well known, are reported to be ubiquitous in freshwater environments, and are also credited for the majority of Mn oxidation [25,26]. Bacterial oxidation does not explain why the loss occurred only in the hydra toxicity tests, however, because the same batch of water was used for tests with other species. The most likely explanation was that the hydra played a role in a speciation change and that the subsampling of the test solutions was insufficient to accurately sample the species of Mn (e.g., the floating precipitate). Nonetheless, further experiments would be needed to determine if the loss was due to hydra-catalyzed Mn speciation changes. Further experimentation was not possible in this instance because a loss in Mn was not observed for the following 2 tests. Although it is possible that the Mn loss was due to sampling or analytical errors in the first 2 hydra tests and the Mn fate testing, the pattern of Mn loss was repeatable, and the quality control

Figure 5. Comparison of environmental manganese chemistry (