Empirical Bioavailability Corrections for Nickel in Freshwaters for Australia and New Zealand Water Quality Guideline Development

Abstract

Bioavailability-based approaches have been developed for the regulation of metals in freshwaters in several countries. Empirical multiple linear regression (MLR) models have been developed for nickel that can be applied to aquatic organisms. The MLR models have been compared against the use of previously developed biotic ligand models (BLMs) for the normalization of an ecotoxicity dataset compiled for the derivation of a water quality guideline value that could be applied in Australia and New Zealand. The MLR models were developed from data for a number of specific species and were validated independently to confirm their reliability. An MLR modeling approach using different models for algae, plants, invertebrates, and vertebrates performed better than either a pooled MLR model for all taxa or the BLMs, in terms of its ability to correctly predict the results of the tests in the ecotoxicity database based on their water chemistry and a fitted species-specific sensitivity parameter. The present study demonstrates that MLR approaches can be developed and validated to predict chronic nickel toxicity to freshwater ecosystems from existing datasets. The MLR approaches provide a viable alternative to the use of BLMs for taking account of nickel bioavailability in freshwaters for regulatory purposes.

Application of Bioavailability Models to Derive Chronic Guideline Values for Nickel in Freshwaters of Australia and New Zealand

“…This may be due to the large datasets for algae and invertebrates, and the fact that an additional local species, H. viridissima, was used in the MLR development, compared with its absence in the European Union BLM development. The relative errors for the trophic levelspecific MLRs were lower than for the pooled MLR and the BLMs, providing further support for selection of the trophic level-specific MLR for bioavailability normalization (Peters et al 2021).…”

“…A bias was noted by Peters et al (2021) for both the algal and invertebrate MLRs, with overprediction of the EC10 values for the more sensitive EC10 values (i.e., less conservative) and underestimation of the EC10 values for the less sensitive EC10 values (more conservative). The tendency for overestimation of toxicity under insensitive conditions was greater than the tendency for underestimation of toxicity in sensitive conditions.…”

“…One disadvantage of the trophic level‐specific MLR is that it does require each species in the nickel SSD to be corrected for its matching trophic‐level algorithm, compared with the use of the pooled MLR (e.g., as developed by Brix et al 2017 for copper). However, the pooled MLR scored poorly in the model comparisons, and had poor adjusted and predictive r 2 values and higher relative errors, and only 66% of the data were within a factor of 2 of the observed EC10 values (Peters et al 2021). For these reasons, we do not recommend use of the pooled MLR for nickel bioavailability normalization.…”

“…MLRs. The MLRs for 4 different trophic levels (algae, aquatic plants, invertebrates, and fish), as well as a pooled MLR model that combined the trophic-level MLRs (Peters et al 2021) were used to normalize the ecotoxicity data for the combined tropical/temperate dataset only (26 species). These MLRs are shown in Table 1.…”

“…Until recently, no MLRs had been published and validated for nickel in freshwaters. Therefore, we developed trophic level‐specific MLRs for a freshwater alga P. subcapitata , an aquatic plant ( Lemna minor ), 3 invertebrates ( D. magna, C. dubia and an Australian clone of Hydra viridissma ), and one fish Pimephales promelas , as well as a pooled MLR (Peters et al 2021). In contrast to North American MLRs for other metals, these MLR models were based on 10% effect concentration (EC10) values rather than median effect concentration (EC50) values, because EC10 values are used in Australia and New Zealand in SSDs to derive guideline values.…”