Cover story: This graphic illustrates an integrated computational framework for identifying aquatic toxic chemicals across six ecotoxicological endpoints. Molecular structures are encoded by a multi-task graph convolutional network (MTL-GCN), which jointly predicts acute and chronic toxicity in fish, crustaceans, and algae by exploiting shared information among related endpoints. Attention-based learning further captures toxicity-relevant structural features, while applicability domain (AD) characterization defines the reliable chemical space for model predictions. Applied to 783,681 inventory chemicals, the framework achieved broad screening coverage, with 94.7% of chemicals falling within the AD of at least one endpoint, supporting efficient large-scale ecological hazard assessment.

Ji, S.; Liu, W.; Bi, G.; et al. Multi-Task Graph Convolutional Network Model with Improved Performance and Broad Applicability Domains for Identifying Aquatic Toxic Chemicals. Glob. Environ. Sci. 2026, 2(3), 278–294.

Back cover story: This review highlights how single-atom catalysts enable low-temperature oxidation of volatile organic compounds by coupling isolated metal centers with tailored supports and interfacial microenvironments. The graphic centers on an atomically dispersed active site, illustrating how anchoring motifs, oxygen vacancies, and metal–support interactions regulate oxygen activation, C–H bond cleavage, and intermediate evolution. Representative VOCs, including formaldehyde, benzene, alkanes, and chlorinated aromatics, converge toward complete oxidation to CO₂ and H₂O. The graphic also emphasizes realistic challenges—humidity, poisoning, and sintering—and outlines emerging design principles for building stable, efficient, and practically deployable single-atom catalysts.

Sun, C.; Huang, Z.; Liu, H.; et al. Design Principles and Structure-Activity Relationships of Single-Atom Catalysts for Low-Temperature VOCs Oxidation and Related Probe Reactions. Glob. Environ. Sci. 2026, 2(3), 312–345.