Deriving a Chronic Guideline Value for Nickel in Tropical and Temperate Marine Waters

Abstract The absence of chronic toxicity data for tropical marine waters has limited our ability to derive appropriate water quality guideline values for metals in tropical regions. To aid environmental management, temperate data are usually extrapolated to other climatic (e.g., tropical) regions. However, differences in climate, water chemistry, and endemic biota between temperate and tropical systems make such extrapolations uncertain. Chronic nickel (Ni) toxicity data were compiled for temperate (24 species) and tropical (16 species) marine biota and their sensitivities to Ni compared. Concentrations to cause a 10% effect for temperate biota ranged from 2.9 to 20 300 µg Ni/L, with sea urchin larval development being the most sensitive endpoint. Values for tropical data ranged from 5.5 to 3700 µg Ni/L, with copepod early–life stage development being the most sensitive test. There was little difference in temperate and tropical marine sensitivities to Ni, with 5% hazardous concentrations (95% confidence interval) of 4.4 (1.8–17), 9.6 (1.7–26), and 5.8 (2.8–15) µg Ni/L for temperate, tropical, and combined temperate and tropical species, respectively. To ensure greater taxonomic coverage and based on guidance provided in Australia and New Zealand, it is recommended that the combined data set be used as the basis to generate a jurisdiction‐specific water quality guideline of 6 µg Ni/L for 95% species protection applicable to both temperate and tropical marine environments. Environ Toxicol Chem 2020;39:2540–2551. © 2020 The Authors. Environmental Toxicology and Chemistry published by Wiley Periodicals LLC on behalf of SETAC.


INTRODUCTION
Nickel (Ni) exposure in marine waters occurs from many anthropogenic and natural sources, of which mining is a particular concern (Hédouin et al. 2016). Nickel is predominantly mined from 2 main ore types, magmatic sulfides, typical of colder climates (e.g., Russia and Canada), and lateritic ores, which are formed from the extensive chemical and physical weathering of ultramafic rock, common in tropical regions (Mudd 2010). Recent estimates show that 60% of the world's Ni reserves are contained in laterite deposits (US Geological Survey 2019). In 2018, 48% of the world's Ni production came sediments into the coastal marine environment (Hunter and Walton 2008).
Concentrations of Ni in unimpacted surface marine coastal waters are typically <0.2 µg/L (Apte et al. 2018). For unimpacted European waters, Heijerick and Van Sprang (2008) reported 90th percentile monitoring values of 3.3 and 0.3 µg Ni/L for estuarine/coastal and open ocean waters, respectively. In some regions, such as New Caledonia, Ni concentrations in soils and aquatic systems are naturally enriched; but mining of lateritic Ni ores can result in additional input of metals into the surrounding coastal system (Hedouin et al. 2009). Dissolved Ni concentrations in seawater in New Caledonia have been reported in the range <0.1 to 11 µg Ni/L (Moreton et al. 2009).
The question arises as to whether water quality guideline values derived for temperate regions will be applicable (or protective) to tropical marine ecosystems. Several studies have investigated the differences in sensitivities of saltwater species to chemical contaminants across different climatic regions. Chapman et al. (2006) showed that tropical marine invertebrates were no more or less sensitive to 4 metals (copper, cadmium, zinc, and lead) than their temperate counterparts based on differences in species sensitivity distributions (SSDs) of acute toxicity data. Similarly, Wang et al. (2014) found only small differences in the acute toxicity of chemicals between tropical and temperate marine biota. For Ni, they showed that hazardous concentrations for 10% of species (HC10) were 658 (95% CI 557-767) µg Ni/L for temperate species (n = 49) and 1560 (95% CI 366-3060) µg Ni/L for tropical species (n = 8). Although this suggests that tropical species may be less acutely sensitive to Ni than temperate species, there is considerable uncertainty (overlap of the 95% CI). There were also significant differences in the amount of toxicity data available for temperate and tropical regions; 49 temperate species versus 8 tropical species .
A recent study by Peters et al. (2019) compared the chronic toxicity of Ni for temperate (n = 19) and tropical (n = 16) freshwater species. They found little difference in the sensitivities of temperate and tropical freshwater species to Ni and recommended combining temperate and tropical data sets to include the most diverse range of taxa possible to ensure the protection of sensitive species across both regions. Similar comparisons with chronic marine Ni data have not been published, and only recently has the toxicity of Ni to key unique tropical taxa such as corals been reported (Gissi et al. 2017Wang et al. 2020).
The scarcity of chronic toxicity data for marine waters (Gissi et al. 2016) has limited our ability to conduct robust risk assessments or to derive appropriate guideline values for tropical regions. More usually, temperate data are simply applied to meet environmental management needs. Differences in physicochemical parameters, such as temperature, rainfall, and irradiance, between temperate and tropical systems make such extrapolations highly uncertain. Therefore, the aim of the present study was to compile existing and newly generated chronic Ni toxicity data for temperate and tropical marine organisms, to determine if there were significant differences in sensitivity to Ni between these 2 geographical groups and to derive a guideline value for Ni in marine waters for either or both climatic regions.

Definition of temperate and tropical species
For the present study, temperate biota were defined as species isolated from temperate regions and/or having a natural geographical distribution outside of the Tropics of Cancer and Capricorn, and toxicity tests were conducted at temperatures <25°C. Tropical biota were defined as species isolated from tropical regions and/or having a natural geographical distribution between the Tropics of Cancer and Capricorn, and toxicity tests were conducted at temperatures ≥25°C. Data were considered marine and included in the compilation if the species was found in, and the test was conducted in, salinities ≥25‰ (Warne et al. 2018).

Compilation and quality check of data
Existing Ni toxicity data for tropical and temperate species were compiled from the literature (up to 2018) following searches in databases including Web of Science, Scopus, and Google Scholar. The data preference was for chronic 10% effect concentration (EC10) and 10% inhibition concentration (IC10) values rather than no-observed-effect concentration (NOEC) values. In some instances, only EC50 and IC50 values (the concentration that causes a 50% effect or inhibition relative to the control) and lowest-observed-effect concentration (LOEC) values were available. Chronic toxicity is defined as an adverse effect that occurs after exposure for a substantial portion of the organism's life span (usually > 10%) or an adverse sublethal effect on a sensitive early life stage (e.g., fertilization over 5 h; Batley et al. 2018;Warne et al. 2018). The species and life stage of the test organisms, exposure duration, and test endpoint (e.g., survival, growth, fertilization) were recorded. In addition, key water quality parameters such as test temperature, pH, dissolved organic carbon (DOC) and salinity were compiled for each test.
Prior to use in the SSDs, all data were quality-checked using a data quality checklist (Australian and New Zealand Governments 2018;Warne et al. 2018). We followed guidance provided by Warne et al. (2018) and included data that scored ≥50% in the SSDs. To achieve a score ≥50%, criteria to be met in the checklist of Warne et al. (2018) included, but were not limited to, use of appropriate controls, replication of controls and contaminant concentrations, inclusion of reference toxicant, stated test acceptability criteria, description of test organism (e.g., life stage, length, mass, age), measurement of contaminant concentrations, measurement of water quality parameters, and use of appropriate statistical method to determine toxicity.

SSDs
Data were fitted to SSDs using the Burrlioz 2.0 software (Australian and New Zealand Governments 2018). If insufficient chronic EC, IC, and lethal concentration (LC; 10-20) and NOEC data were available, chronic LOEC and EC/IC/LC50 data were converted to chronic NOEC values by dividing by factors of 2.5 and 5, respectively, according to the method of Warne et al. (2018). These conversions were required; otherwise, EC50 and LOEC data would need to be excluded from the chronic SSDs. Where there was more than one value reported for the same species and endpoint, the geometric mean was calculated and included in the SSD. Where there were multiple values for the same species using different endpoints, the most sensitive endpoint (i.e., the lowest toxicity value) was used. Chronic toxicity data sets with 8 or more data were fitted to a Burr type III distribution, and the HC1, HC5, HC10, and HC20 values determined. In Australia and New Zealand, the HC5 (i.e., 95% protective concentration [PC95]) value applies to slightly to moderately disturbed systems (i.e., most systems), whereas the HC1 (i.e., PC99) is applied to systems of high ecological value (Australian and New Zealand Governments 2018).

Comparison of species sensitivity
Pairwise temperate and tropical SSD comparisons were conducted by a combination of statistical tests and visual inspection . Analysis of covariance (ANCOVA; SPSS) was conducted to compare the slopes of the 2 SSDs based on a log-normal distribution. A visual comparison of either congruence or discrepancy of the temperate and tropical distributions was also used by plotting the cumulative distribution of the Ni toxicity data in Excel (Supplemental Data, Figure S1). If the 95% CIs of the 2 HC5 values (e.g., temperate vs tropical SSDs) did not overlap, then the 2 SSDs were considered significantly different ). The toxicity data for each phylum from temperate and tropical regions were displayed in box plots, created in the statistical package NCSS (2007, Ver 07.1.21). Toxicity data for crustacea from temperate and tropical regions were compared using analysis of variance (ANOVA) in the statistical software package NCSS.
The formula of Litchfield and Wilcoxon (1949) was also used to compare the HC5 values derived from each SSD (temperate, tropical, and combined), to determine if there were any significant differences in the values calculated from the different data sets. Ratios between the HC5 values from temperate, tropical, and combined SSDs were calculated to compare the HC5 values calculated from each SSD. The upper limit of the 95% CI of the HC5 value calculated from each SSD (temperate, tropical, and combined) was also used to calculate the F ratio. If the ratio of the HC5 values was less than the F ratio, there was no significant difference in the HC5 values.

Temperate and tropical marine toxicity data
Of the chronic toxicity data that scored ≥50% and were deemed applicable for water quality guideline development, there were 24 temperate species and 16 tropical species, giving a combined data set of 40 species representing 14 taxonomic groups (based on phyla; Warne et al. 2018) including diatoms, green algae, a dinoflagellate, a brown-golden alga, cyanobacteria, copepods, a brown macroalga, a red macroalga, polychaetes, crustaceans, bivalve mollusks, gastropod mollusks, echinoderms, cnidarians (corals, sea anemone), and fish (Table 1). The tropical data set did not have data for macroalgae and bivalves, but it had 3 values for corals which mainly occur in tropical regions. In general, the species compositions between the 2 regions were comparable, with 6 out of 9 (67%) taxonomic groups in common.

Temperate species sensitivity to Ni
The sensitivity of temperate marine species to Ni varied widely across different taxonomic groups, with threshold and no effects (EC10/NOEC values) observed between 2.9 and 20 300 µg Ni/L ( Table 2). The most sensitive species were echinoderms (sea urchins; EC10/NOEC values 2.9-500 µg Ni/L), polychaetes (reproduction inhibited by 10% at 23 µg Ni/L), gastropods (EC10/NOEC values for growth inhibition 21-36 µg Ni/L), and mysid shrimps (neonate survival [48 h] inhibited by 10% at 45 µg Ni/L; Table 2 and Figure 1A and B; Supplemental Data, Table S1). For sea urchins, the most sensitive endpoint was larval development, with EC10 values between 2.9 and 335 µg Ni/L (Hwang et al. 2012;DeForest and Schlekat 2013;Blewett et al. 2016). Fertilization tended to be less sensitive, with NOEC and EC50 values of 500 and 217 µg Ni/L, respectively for the urchin Paracentrotus lividus (Novelli et al. 2003;Pagano 2007). These studies used different species and different test conditions, so caution must be taken when making comparisons. Fish and green algae were the least sensitive taxa to Ni ( Figure 1B), with 10% effects on fish larval survival and algal growth above 17 000 µg Ni/L. All temperate toxicity data, including values used in the SSDs are presented in Table 2. The complete data set, including details on test medium and physicochemical parameters (temperature, salinity, pH), is given in Supplemental Data, Table S1. Of the  entire data set, 24 values (i.e., species) representing 10 taxonomic groups were selected for input into the SSDs (Figure 2A).

Tropical species sensitivity to Ni
Concentrations of Ni that caused a 50% effect to tropical species ranged from 6.6 to 22 500 µg Ni/L. Similar to the temperate data set, sea urchins and crustaceans were the most sensitive taxa to Ni, whereas algae were the least sensitive ( Figure 1A and B). The most sensitive tropical marine species was the copepod Acartia sinjiensis, with an EC10 for development of 5.5 (5.0-6.0) µg Ni/L ). The EC10/NOEC values for other crustacea, including another species of copepod and a barnacle, ranged from 29 to 99.8 µg Ni/L. Endpoints included mortality, development/ metamorphosis, and intrinsic rate of population increase Wang et al. 2020). Only one species of tropical sea urchin has been reported in the literature, with a NOEC for normal larval development of 23 µg Ni/L (Rosen et al. 2015). Gastropods (snails) were also relatively sensitive to Ni, with EC10 values for growth rate inhibition at concentrations of 33 to 64 µg Ni/L Wang et al. 2020). Cyanobacteria, diatoms, dinoflagellates, and a brown alga were some of the least sensitive species ( Figure 1B), with EC10 values between 330 and 3700 µg Ni/L (Alquezar and Anastasi 2013; Gissi 2018). Coral was one of the least sensitive species, with an EC10 for fertilization success of 2000 µg Ni/L (Gissi et al. 2017). All tropical toxicity data, including values used in the SSDs, are presented in Table 3. The complete data set, including details on test medium and physicochemical parameters (temperature, salinity, pH) is presented in Supplemental Data, Table S2. Of the entire data set, 16 values (i.e., species) were compiled for input into the SSDs representing 10 taxonomic groups ( Figure 2B).

Comparison of species sensitivity
There was little difference in the sensitivities of temperate and tropical species to Ni ( Figure 1A and B). Comparisons are limited to phyla where toxicity data were available for more than one species for each group. For example, the ranges of toxicity values for crustacea (6 temperate species, 4 tropical species) overlap in the box plots ( Figure 1A), and there was no significant difference in the range of toxicity values (ANOVA, p = 0.22). Temperate and tropical microalgae also have a similar range of sensitivities ( Figure 1B). Because of the uneven representation of taxonomic groups and species across the temperate and tropical data sets, it is difficult to make direct comparisons of species sensitivities to Ni across different climatic regions as was done for freshwaters by Peters et al. (2019). There were more chronic Ni toxicity data available for temperate freshwater species (31 species; Peters et al. 2019) compared to 24 temperate marine species. The number of tropical data was similar, with 13 freshwater tropical species used by Peters et al. (2019) and 16 tropical marine species in the present study.   Tables 2 and 3, were log-transformed for graphical representation; when n > 3, box = median and interquartile range, whiskers = maximum and minimum values; when n ≤ 3, box = mean and maximum and minimum values-refer to Table 1. Te = temperate; Tr = tropical.   Previously known as Nitzschia closterium, grown in G2 media (Loeblich and Smith 1968). b Previously known as Nitzschia closterium, grown in F2 media (Guillard and Ryther 1962 Guideline values for Ni in marine waters-Environmental Toxicology and Chemistry, 2020;39:2540-2551 The log-normal and Burr type III distributions were fitted to the SSDs, and the resulting HCx values calculated from these distributions were compared (Supplemental Data, Table S3). There was no significant difference in the HC values calculated from either distribution. This is further supported by the study of Wang et al. (2014), which showed little difference in a range of different distributions applied to acute marine toxicity data for several different toxicants. Requirements around SSDs and the preference for model selection are jurisdiction-dependent. The present study has followed guidance described in Warne et al. (2018) and Batley et al. (2018), and the objective was to develop a guideline value for the region; therefore, the Burr type III distribution was ultimately selected to derive the final HCx values.
Tropical SSDs ( Figure 2B) were found to be significantly different from the temperate and the combined SSDs, as indicated by significantly different slope parameters (ANCOVA p < 0.05; Supplemental Data, Table S4) and the crossover of the distributions (Supplemental Data, Figure S1; Wang et al. 2014). The temperate SSD was also found to be different from the combined SSD (ANCOVA p < 0.05 for different slopes; Supplemental Data, Table S4 and Figure S1). However, tropical and temperate species shared a similar sensitivity to Ni (Figure 1). In addition, there was no significant difference in the HC5 values derived from each SSD, temperate, tropical, or combined (Figure 2A-C) as reflected by the overlapping 95% CIs (Table 4). This was supported by calculations using the Litchfield-Wilcoxon formula, which showed that there was no statistically significant difference between the HC5 values obtained from all SSDs. For all 3 comparisons, the ratios of the HC5 values were less than the F ratio (temperate vs tropical, 2.13 < 3.3; temperate vs combined, 1.29 < 3.2; tropical vs combined, 1.66 < 2.3).
A range of factors other than geographical distribution and number of taxa limits our ability to compare the sensitivities of tropical and temperate species to Ni. Differences in temperature between temperate and tropical tests may affect both toxicokinetics and toxicodynamic processes , although species are likely adapted to the temperature conditions of their environment. Temperate species were typically tested in the range of 15 to 24°C, whereas tropical species were tested between 25 and 30°C. It is important to note that species were tested at temperatures to which they had been acclimated. For this reason, it is unlikely that temperature plays a significant role in Ni toxicity, and the response of organisms to Ni exposure at temperatures outside of their normal range may not be as we expect. For example, Pereira et al. (2017) showed that Ni toxicity to the freshwater flea Daphnia magna increased as temperature decreased.
Nickel is known to be an essential nutrient for microorganisms and terrestrial plants, but essentiality in aquatic animals has not been confirmed (Muyssen et al. 2004;Moreton et al. 2009). In freshwater temperate biota, Ni is thought to be a respiratory toxicant (acute exposures to fish and some invertebrates), an ionoregulatory toxicant (invertebrates), and a promoter of oxidative stress (Brix et al. 2017). Mechanisms of Ni toxicity, particularly chronic effects, in the marine environment, however, are not well understood, with only limited studies on temperate killifish, copepods, mussels, and the green shore crab, looking at acute effects at high Ni concentrations (Blewett and Leonard 2017). Only 2 mechanistic studies have investigated tropical species, and both only examined acute toxicity to freshwater species. Nath and Kumar (1989) found effects of Ni on the gills of the tropical perch at very high Ni concentrations (13 mg Ni/L) over a 96-h exposure at 24°C, whereas Palmero et al. (2015) found that Ni at 2.5 mg/L affected antioxidant defenses in the freshwater fish Prochilodus lineatus (but at 20°C). Consequently, there are insufficient data to determine whether Ni has a different mode of action to tropical species at higher temperatures than to temperate marine species.
Water quality parameters, such as salinity and DOC, can also influence the toxicity of Ni to marine biota, although the effects of DOC may be small (Blewett et al. 2018). Blewett et al. (2016Blewett et al. ( , 2018 showed that Ni toxicity to the urchin Evechinus chloroticus and the mussel Mytilus edulis was influenced by both DOC quantity and quality. However, Ni toxicity varied by less than a factor of 2 among different natural water sources. No clear influence of DOC was found for the mussel Mytilus galloprovincialis for DOC in the range 1.2 to 2.7 mg/L or for the diatom Selenastrum costatum for DOC in the range 0.2 to 2.7 mg/L (Deforest and Schlekat 2013). Because the effect of DOC on Ni toxicity is limited, marine HC values have not been corrected for bioavailability.

SSDs and HC values
Relative to many other chemicals, there is now a large data set for Ni toxicity to both temperate and tropical species. Using this data set, the derived marine HC values for different levels of ecosystem protection are shown in Table 4 (Australian and New Zealand Governments 2018). It is the HC5 value that would mostly be applied in slightly to moderately disturbed systems in Australia and New Zealand ( Figure 2C and Table 4). Warne et al. (2018) provided guidance for assessing the reliability of HC values derived from SSD methods. This is based on the sample size (number of species for which toxicity data are available), the type of data (chronic, chronic and acute, or converted values), and visual assessment of the fit of the SSD to the toxicity data (i.e., good or poor). The fit of the SSD was good (Figure 2A and C) for both the temperate and combined temperate plus tropical data sets, so, using the classification outlined by Warne et al. (2018), the derived HC values were classified as being of very high reliability ( Table 5). The fit of the tropical data set ( Figure 2B) was poor, particularly at higher Ni concentrations; and the derived HC values were considered to be of moderate reliability (Table 5). It is recommended that the combined data set and HC5 value, 5.8 µg Ni/L (rounded up to 6 µg/L), be used as the basis for a jurisdiction-specific guideline for Ni for both temperate and tropical marine waters. There was little difference in the overall sensitivity of temperate and tropical species to Ni and in the resultant HC5 values calculated from each SSD. The combined SSD utilizes a larger data set and includes a broader range of species, 40 in total. It is inevitable that the guideline will not be protective of all species. In this instance, 1 sensitive sea urchin and 1 copepod were below the HC5 value, although other sea urchin and copepod species were protected. In addition, this value is above the typically reported background concentrations of Ni in seawater (<5 µg/L; DeForest and Schlekat 2013; Apte et al. 2018).
The HC5 value reported in the present study (6 µg/L) is similar to other guideline values previously reported in Australia and overseas. The current Australian and New Zealand water quality guideline value for Ni (based on temperate marine data) is 70 µg Ni/L, based on 15 species from 5 taxonomic groups. However, because this was insufficiently protective of some species, the default guideline for slightly to moderately disturbed systems was set at 7 µg Ni/L (99% species protection; Australian and New Zealand Environment and Conservation Council and Agriculture and Resource Management Council of Australia and New Zealand 2000). The US Environmental Protection Agency derived a similar chronic Ni guideline for saltwater of 8.2 µg/L. The Ni environmental quality standard (EQS) under the European Union's Water Framework Directive for coastal marine waters is 8.6 µg Ni/L (Nickel Institute 2012). DeForest and Schlekat (2013) undertook further toxicity testing with temperate marine species and provided 2 additional data. In their study, the most sensitive species to Ni was a tropical species of a long-spined sea urchin (Diadema antillarum) from the Caribbean region, which had an EC10 value of 2.9 µg Ni/L. However, this toxicity test was carried out at 20°C so was not included in our tropical compilation. DeForest and Schlekat (2013) derived a marine Ni HC5 value of 3.9 µg Ni/L (including this tropical sea urchin) and 21 µg Ni/L (when the sea urchin data were excluded because of lack of relevance to European marine waters).

CONCLUSION
The present study compiled and quality-checked chronic Ni toxicity data for temperate and tropical marine species. There was little difference in the range of sensitivities of temperate and tropical marine species to Ni and in the resultant HC values that were calculated from the SSDs. As such, and to ensure greater taxonomic coverage, it is recommended that the combined temperate and tropical data set be used to generate a guideline value of 6 µg Ni/L, to ensure the protection of marine species in both temperate and tropical environments.