Development of a multiple linear regression model for chronic nickel toxicity to ceriodaphnia dubia: performance of bicarbonate vs. pH as toxicity modifying factors

Multiple linear regression models were developed to predict chronic nickel (Ni) toxicity to Ceriodaphnia dubia under an expanded range of conditions relative to published datasets. Test conditions were expanded to study Ni toxicity under very high hardness (up to 1,020 mg/L as CaCO3) and bicarbonate (up to 366 mg/L HCO3) by conducting tests in synthetic waters and mine-influenced site waters. Toxicity modifying factors (TMFs) identified by the models were hardness, dissolved organic carbon, and either pH or bicarbonate. Because high pH and high bicarbonate co-occur in some mine-influenced waters and their relative importance as TMFs for Ni is unclear, we compared the performance of models with each candidate TMF. The model with bicarbonate performed better than the model with pH and showed closer alignment between site-specific and published datasets, supporting bicarbonate as a TMF in both datasets. The model with bicarbonate performed well across the expanded range of conditions and is expected to be more robust than previous Ni models under the high hardness and bicarbonate conditions studied. Comparison of TMF effects on Ni toxicity to other invertebrates indicated stronger support for a bicarbonate TMF effect than pH for some species, including C. dubia. Further support came from site-specific testing that indicated bicarbonate is not toxic to C. dubia at the concentrations in our dataset. These findings suggest bicarbonate may play an important role in modifying chronic Ni toxicity to C. dubia in alkaline waters. More work is needed to understand the mechanism for bicarbonate and pH TMF effects and why TMF effects are species-specific to reduce uncertainties associated with collinearity in model datasets and in applying models across species.