By integrating molecular descriptors and life-stage information, the models accurately forecast acute and chronic toxicity without relying on traditional testing. This approach enables rapid ecological risk assessment, highlighting pollutants that could pose long-term hazards, and provides a powerful tool to guide evidence-based conservation and pollution management in vulnerable aquatic ecosystems.
Aquatic ecosystems worldwide are under mounting pressure from industrialization, urbanization, and chemical pollution. Rare and endangered species, like G. rarus, are especially vulnerable due to limited habitat range, small populations, and high sensitivity to environmental stressors. Traditional toxicity testing for these species is often impractical or ethically problematic, leaving a critical knowledge gap. While QSAR models have been used for widely distributed species, conventional approaches struggle with complex, high-dimensional chemical structures and nonlinear toxicity responses. Recent advances in machine learning provide an opportunity to overcome these limitations, offering precise predictions for species with sparse experimental data.
A study (DOI: 10.48130/newcontam-0026-0010) published in New Contaminants on 30 April 2026 by Ying Wang’s & Wenhong Fan’s team, Beihang University, demonstrates that ML-QSAR models can successfully predict acute and chronic toxicity for pollutants, including PFASs and conazoles, in different life stages of the species. Life stage dominates acute toxicity responses, whereas molecular interaction descriptors primarily drive chronic toxicity predictions.
The researchers employed microbiological, electrochemical, spectroscopic, and genomic methods to investigate the bidirectional extracellular electron transfer (EET) and carbon fixation potential of Fundidesulfovibrio terrae SG127. Extracellular respiration was first assessed using ferrihydrite as a solid Fe(III) electron acceptor under anaerobic conditions. Fe(II) production showed phase-dependent kinetics, reaching 68.3% reduction in seven days, and occurred without soluble electron shuttles, indicating direct membrane-bound electron transfer. To evaluate bidirectional EET, a single-chamber bioelectrochemical system with graphite electrodes was constructed. In anodic mode, pyruvate served as the electron donor, producing a maximum current density of 27.50 μA/cm², while in cathodic mode with sulfate as the electron acceptor, the peak cathodic current reached 28.75 μA/cm². Scanning electron microscopy confirmed dense biofilm formation on both electrodes. Cyclic voltammetry revealed surface-controlled, quasi-reversible electron transfer with distinct oxidation and reduction peaks. Respiratory chain inhibitors, including rotenone, antimycin A, DCCD, and dicoumarol, were applied to probe the contributions of complex I, complex III, and the quinone pool, showing differential sensitivity for inward and outward EET. UV-visible spectroscopy detected characteristic c-type cytochrome absorption, and genome analysis identified key genes encoding membrane-associated redox proteins (MacA, MtrD, MtrC) and type IV pili, suggesting a cytochrome–pilus network mediating electron flow. Finally, under sulfate-free autotrophic conditions in a microbial electrosynthesis system, F. terrae used electrode-derived electrons and CO₂ as the sole carbon source, producing acetate at 11.05 mM. These results demonstrate that F. terrae harnesses direct, cytochrome-mediated bidirectional electron transfer to support both extracellular respiration and electroautotrophic carbon fixation, highlighting its potential for microbial electrosynthesis and bioelectrochemical applications.
The findings offer a blueprint for applying machine learning in ecological risk assessment, enabling rapid, data-driven evaluation of chemical pollutants in habitats where experimental testing is limited. This approach can be extended to other rare or endangered species, providing policymakers and conservationists with actionable insights to mitigate chemical hazards and safeguard biodiversity in fragile aquatic ecosystems. Long-term monitoring strategies and targeted risk management for persistent pollutants, particularly PFASs, are recommended to ensure the ongoing protection of G. rarus and its habitat.
###
References
DOI
10.48130/newcontam-0026-0010
Original Source URL
https://doi.org/10.48130/newcontam-0026-0010
Funding Information
This work was supported by the National Key R&D Program of China (Grant No. 2022YFC3204800), the National Natural Science Foundation of China (Grant No. 42177240), and the Beijing Natural Science Foundation (Grant No. 8242033).
About New Contaminants
New Contaminants is a multidisciplinary platform for communicating advances in fundamental and applied research on emerging contaminants. It is dedicated to serving as an innovative, efficient and professional platform for researchers in the field of new contaminants research around the world to deliver findings from this rapidly expanding field of science.