The acute toxicity of nickel to Daphnia magna: predictive capacity of bioavailability models in artificial and natural waters.

The effects of Ca, Mg, Na and pH on the acute toxicity of Ni to Daphnia magna were investigated in a series of 48-h immobilization assays in synthetic test solutions. Both Ca and Mg reduced Ni toxicity, while Na did not. Ni toxicity was not affected in the pH range of 5.7-7.5, but a further increase of pH up to 8.1 resulted in an increase of toxicity of the free Ni2+ ion. Based on the results of these experiments, a biotic ligand model (BLM) was developed in which the effects of Ca and Mg were modeled as single-site competition effects. Stability constants representing the binding strength between Ca2+ and Mg2+ and the biotic ligand (BL) were logK(CaBL)=3.10 and logK(MgBL)=2.47, respectively. The effect of pH could not be appropriately described by single-site competition between Ni2+a nd H+. Since the overall variation of toxicity within the tested pH range was relatively small, we decided not to incorporate the effect of pH in the current model. The model was able to predict 48-h EC50s in all synthetic test solutions by an error less than factor 2. The model's predictive capacity was also evaluated using results of toxicity tests in Ni-spiked natural surface waters. For 15 out of 16 tested waters, 48-h EC50s were predicted by an error less than factor 2. Additionally, after calibration to account for interclonal or interspecies sensitivity differences, the model was able to accurately predict earlier published 48-h EC50s for another D. magna clone as well as for Ceriodaphnia dubia. Finally, the predictive capacity of the model was demonstrated to be better than that of previously proposed models that include a logK(NaBL), a logK(HBL) and a logK(CaBL), but did not incorporate a logK(MgBL). An in-depth comparison of these models learned that (i) there is no need to incorporate a logK(NaBL), (ii) it is important to recognize the protective effect of Mg, and (iii) the incorporation of a logK(HBL) does not adequately describe the effect of pH. Although our model seems very promising, further research, especially into the effects of elevated pH and alkalinity levels, is needed to allow further refinement.