Related Experiment Video
Updated: Jun 24, 2026
10:20
Quantification of Heavy Metals and Other Inorganic Contaminants on the Productivity of Microalgae
Published on: July 10, 2015
16.0K
Development and Validation of Multiple Linear Regression Models for Predicting Chronic Zinc Toxicity to Freshwater
Gwilym A V Price1,2, Jenny L Stauber2,3, Dianne F Jolley4
1Faculty of Science, University of Technology Sydney Broadway, Ultimo, New South Wales, Australia.
Environmental Toxicology and Chemistry
|September 20, 2023
Summary
Multiple linear regression models for predicting chronic zinc toxicity to Chlorella sp. failed to accurately predict toxicity in natural waters. Models consistently overpredicted toxicity, questioning the applicability of lab-based models to real-world environmental conditions.
Area of Science:
- Environmental Toxicology
- Ecotoxicology
- Aquatic Toxicology
Background:
- Chronic zinc toxicity to freshwater microalgae is a significant environmental concern.
- Predictive models are crucial for assessing ecological risks, but their accuracy in natural waters is often limited.
- Understanding the influence of water chemistry on toxicity is key to developing robust ecotoxicological models.
Purpose of the Study:
- To develop and validate multiple linear regression (MLR) models for predicting chronic zinc toxicity to Chlorella sp.
- To investigate the influence of toxicity-modifying factors (TMFs) including pH, hardness, and dissolved organic carbon (DOC) on zinc toxicity.
- To assess the predictive performance of MLR models in diverse Australian natural waters.
Main Methods:
- Development of MLR models using pH, hardness, and DOC as TMFs, including interactive effects.
- Models were constructed for three effect concentration (EC) levels: EC10, EC20, and EC50.
- Independent validation of models using six different zinc-spiked natural waters with varying water chemistries.
Main Results:
- Hardness was identified as a consistently influential TMF, retained in all final models.
- pH, DOC, and interactive terms showed variable influence and were included in some models.
- Models generally predicted toxicity well in autovalidation but performed poorly during independent validation, consistently overpredicting toxicity in natural waters.
Conclusions:
- MLR models developed from synthetic waters showed poor predictive accuracy for zinc toxicity in natural Australian waters.
- Consistent overprediction suggests unaccounted TMFs or limitations in applying laboratory-derived models to complex natural environments.
- Further research is needed to improve the applicability of predictive ecotoxicological models to natural water systems.
