Estimating species sensitivity distributions on the basis of readily obtainable descriptors and toxicity data for three species of algae, crustaceans, and fish

Abstract

Estimation of species sensitivity distributions (SSDs) is a crucial approach to predicting ecological risks and water quality benchmarks, but the amount of data required to implement this approach is a serious constraint on the application of SSDs to chemicals for which there are few or no toxicity data. The development of statistical models to directly estimate the mean and standard deviation (SD) of the logarithms of log-normally distributed SSDs has recently been proposed to overcome this problem. To predict these two parameters, we developed multiple linear regression models that included, in addition to readily obtainable descriptors, the mean and SD of the logarithms of the concentrations that are acutely toxic to one algal, one crustacean, and one fish species, as predictors. We hypothesized that use of the three species’ mean and SD would improve the accuracy of the predicted means and SDs of the logarithms of the SSDs. We derived SSDs for 60 chemicals based on quality-assured acute toxicity data. Forty-five of the chemicals were used for model fitting, and 15 for external validation. Our results supported previous findings that models developed on the basis of only descriptors such as log KOW had limited ability to predict the mean and SD of SSD (e.g., r2 = 0.62 and 0.49, respectively). Inclusion of the three species’ mean and SD, in addition to the descriptors, in the models markedly improved the predictions of the means and SDs of SSDs (e.g., r2 = 0.96 and 0.75, respectively). We conclude that use of the three species’ mean and SD is promising for more accurately estimating an SSD and thus the hazardous concentration for 5% of species in cases where limited ecotoxicity data are available.

Ecotoxicological risk assessment of organic compounds against various aquatic and terrestrial species: application of interspecies i-QSTTR and species sensitivity distribution techniques

“…Most of the collected works on SSDs targeted a small dataset size or a restricted group of species. 27–32 A few notable attempts made to assess the SSD profile of a chemical using QSAR predictions using a limited number of compounds and species were by Tetko et al , 2013 and others, 33–35 or using additional parameters to already defined SSD models as observed with works of Bejarano et al , 2017 36 and Bejarano 2019, 37 or without incorporating any experimental ecotoxicity data (based on predictions). 38 Again, some of the SSD works incorporated non-linear machine learning tools such as artificial neural networks which offer limited use from the regulatory perspective.…”

To Split or Not to Split: Characterizing Chemical Pollution Impacts in Aquatic Ecosystems with Species Sensitivity Distributions for Specific Taxonomic Groups

“…Laboratory-toxicity data-based SSDs are practically used for regulatory purposes and Life Cycle Impact Assessment (LCIA), e.g., to derive protective standards (threshold concentrations) or expected impact levels of ambient chemical pollution. , Recently, their use has expanded to the comprehensive diagnosis of the role of chemical pollution as a driver for biodiversity loss in polluted ecosystems by using SSD-based mixture toxic pressure information (expressed as msPAF, the multisubstance Potentially Affected Fraction of species) as pressure metric, as this resulted in reduced parameters numbers and thus improved statistical power in diagnostic analyses. ,, The choice of required input data and the statistical distribution methods vary among jurisdictions. Models commonly used to fit SSDs include log-normal, log–logistic, or other models that fit the available data well, and commonly, confidence intervals or other metrics of variability and uncertainty are reported. , Crucial to acknowledge is that SSDs are commonly fitted to all available test data per chemicalfollowing the principles developed by the earliest userswhere it is assumed that the SSD describes the exposure-impact relationship for whole field species assemblages.…”

“…Laboratory-toxicity data-based SSDs are practically used for regulatory purposes and Life Cycle Impact Assessment (LCIA), e.g., to derive protective standards (threshold concentrations) or expected impact levels of ambient chemical pollution. 5 , 6 Recently, their use has expanded to the comprehensive diagnosis of the role of chemical pollution as a driver for biodiversity loss in polluted ecosystems by using SSD-based mixture toxic pressure information (expressed as msPAF, the multisubstance Potentially Affected Fraction of species) as pressure metric, as this resulted in reduced parameters numbers and thus improved statistical power in diagnostic analyses. 2 , 7 , 8 The choice of required input data and the statistical distribution methods vary among jurisdictions.…”

Potential for Machine Learning to Address Data Gaps in Human Toxicity and Ecotoxicity Characterization

“…However, assessing the growing number of marketed chemicals across different consumer products, populations, and environments is increasingly challenging. , Characterizing chemical toxicity impacts, including aspects on environmental fate, exposure, and (eco-)­toxicity effects, is essential across a wide range of chemical-related decision support tools, such as risk assessment , and screening, life cycle impact assessment (LCIA), , chemical footprinting, , chemical substitution, benchmarking chemical pollution against local-to-global boundaries, , and safe-and-sustainable-by-design (SSbD) assessments . The application of chemical-related decision support tools to the >100,000 marketed chemicals and the wide range of product uses is currently limited by a lack of structured, high-quality input data needed to characterize toxicity for millions of chemical–product combinations. , Obtaining new data from experimental tests is cost- and time-consuming, and confidential or nontransparent reporting hinder access to existing data. − To address data gaps, scientists have been developing quantitative structure–activity relationships (QSAR) for decades by creating quantitative links between chemical structures and various target properties, including input parameters for characterizing chemical toxicity. , With increasing data availability and computing power, QSAR evolved from simple regressions on small sets of congeneric compounds to applying advanced statistical and machine learning (ML) techniques on large chemical sets with diverse molecular structures, boosting their predictive performance and applicability for a broader realm of chemicals. , Several advanced chemical data prediction models are readily accessible through public modeling suites providing predictions for multiple chemical properties , and many more have been documented in the scientific literature for individual chemical properties, including dissociation constants, , root concentration factors, − and ecotoxicity end points. − While the development of ML-based approaches has been an active field of research, a systematic adaptation for chemical toxicity characterization is still limited. The main challenges relate to a lack of oversight into required input parameters which could support the systematic development of ML-based approaches and a lack of transparency about whether such approaches can robustly predict parameter...…”