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

Oginah, Susan Anyango; Posthuma, Leo; Hauschild, Michael; Slootweg, Jaap; Kosnik, Marissa; Fantke, Peter

doi:10.1021/acs.est.3c04968

Cited by 4 publications

(3 citation statements)

References 38 publications

(80 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It serves to identify which percentage of tested species is at risk at a certain concentration. Decades after the introduction, SSDs are still the subject of active research 47,48 . Here, we produced SSDs for compounds that have been tested on at least 15 different fish species to investigate how well the model predictions match the species sensitivity of the original biological data.…”

Section: Resultsmentioning

confidence: 99%

“…The experimental features describe the experimental conditions, specifically, observation duration, media type, exposure type, and concentration type, which we used as the four experimental features in the models. The observation duration is ordinal with four levels (24,48,72,96 tion included in the dataset that we do not deem relevant, and for some features, we argue in Ref. 23 against using them for modeling.…”

Section: Description and Processing Of Modeling Featuresmentioning

confidence: 99%

See 1 more Smart Citation

Machine learning-based prediction of fish acute mortality: Implementation, interpretation, and regulatory relevance

Gasser,

Schür,

Perez-Cruz

et al. 2024

Preprint

View full text Add to dashboard Cite

Regulation of chemicals requires knowledge of their toxicological effects on a large number of target species. Traditionally, this knowledge has been acquired throughin vivotesting. The recent effort to find alternatives based on machine learning, however, has not focused on guaranteeing transparency, comparability and reproducibility, which makes it difficult to assess advantages and disadvantages of these methods. Also, comparable baseline performances are needed. In this study, we trained regression models on the ADORE “t-F2F” challenge proposed in [Schüret al.,Nature Scientific data, 2023] to predict acute mortality, measured as LC50 (lethal concentration 50), of organic compounds on fishes. We trained LASSO, random forest (RF), XGBoost, Gaussian process (GP) regression models, and found a series of aspects that are stable across models: (i) using mass or molar concentrations does not affect performances; (ii) the performances are only weakly dependent on how a chemical is represented, but (iii) strongly on how the data is split. Overall, the tree-based models RF and XGBoost performed best and we were able to predict the log10-transformed LC50 with a root mean square error of 0.90, which corresponds to an order of magnitude on the original LC50 scale. On a local level, on the other hand, the models are not able to predict the toxicity of single chemicals accurately enough. Predictions for single chemicals are dominated by a few chemical properties and taxonomic traits are not captured sufficiently by the models. As a consequence, the models are not yet applicable in the regulatory process. Nevertheless, this work contributes to the ongoing discussions on how machine learning can be integrated in the regulatory process.Environmental significanceConventional environmental hazard assessment in its current form will not be able to adapt to the growing need for toxicity testing. Alternative methods, such as toxicity prediction through machine learning, could fulfill that need in an economically and ethically sound manner. Proper implementation, documentation, and the integration into the regulatory process are prerequisites for the usability and acceptance of these models.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Description and Processing Of Modeling Featuresmentioning

confidence: 99%

Machine learning-based prediction of fish acute mortality: Implementation, interpretation, and regulatory relevance

Gasser,

Schür,

Perez-Cruz

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

“…It serves to identify the percentage of tested species at risk at a certain concentration. Decades after the introduction, SSDs are still the subject of active research 44,45 . Here, we produced SSDs for compounds that have been tested on at least 15 different fish species to investigate how well the model predictions match the species sensitivity of the original biological data.…”

Section: Species Sensitivitymentioning

confidence: 99%

Machine learning-based prediction of fish acute mortality: Implementation, interpretation, and regulatory relevance

Gasser,

Schür,

Perez-Cruz

et al. 2024

Environ. Sci.: Adv.

View full text Add to dashboard Cite

show abstract

Evaluating toxicity impacts of local chemical emissions in Life Cycle Assessment

Cappucci,

Neri,

Ferrari

et al. 2024

Int J Life Cycle Assess

View full text Add to dashboard Cite

Purpose The local dimension of toxicity effect on humans and ecosystem from chemical emissions into rural environments is currently not considered in impact characterization models underlying the Life Cycle Assessment (LCA) methodology. The aim of the present study was to understand the relevance of considering a local exposure environment for the magnitude of damage related to human toxicity and ecotoxicity impacts associated with chemical emissions into a local environment. Method Unit and realistic European emission scenarios were considered in a proposed multimedia nested box model, based on an existing framework modified with the inclusion of a local scale, and tested for ten chemicals in an illustrative case study. Results A substantial damage increase in terms of characterization factors in the proposed model was found for human health, mainly for local freshwater emissions (up to three orders of magnitude compared to emissions into continental freshwater for naphthalene) and soil emissions (up to two orders of magnitude, in particular for emissions into natural soil for 1,2-dichloroethane). Conclusions Based on our results, we suggest to use the proposed framework in LCA applications, when more specific information about the local emission environment is known. With respect to considering the local-scale, however, spatialized models might be preferable over nested box models to properly capture local phenomena in rather highly densely populated areas.

show abstract

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

Cited by 4 publications

References 38 publications

Machine learning-based prediction of fish acute mortality: Implementation, interpretation, and regulatory relevance

Machine learning-based prediction of fish acute mortality: Implementation, interpretation, and regulatory relevance

Machine learning-based prediction of fish acute mortality: Implementation, interpretation, and regulatory relevance

Evaluating toxicity impacts of local chemical emissions in Life Cycle Assessment

Contact Info

Product

Resources

About