Researchers in Beijing have generated a transcriptomic atlas comparing human fetal tooth germs from the upper and lower jaws at the critical cap stage of development. Previous studies of tooth development mainly focused on mandibular teeth, while this study uncovers gene expression patterns of maxillary teeth and their differences from mandibular tooth germs.

The instability of anode catalysts during the oxygen evolution reaction (OER) is a central obstacle to commercializing proton exchange membrane (PEM) electrolyzers. In the highly oxidative and acidic anode environment, catalysts suffer from dissolution, mechanical detachment, and impurity-driven degradation—failure modes that are tightly interconnected and cannot be solved through material optimization alone. This perspective evaluates these coupled degradation pathways and the limitations of current material, structural, and system-level strategies. We argue that durable acidic OER requires mechanistic insight under realistic operating conditions and the coordinated advancement of catalyst design, operando characterization, engineering improvements, and data-driven modeling. Such an integrated framework is essential for developing stable anodes and enabling large-scale, long-lifetime PEM electrolyzers.

Electronic devices face dual challenges of electromagnetic wave (EMW) interference and heat accumulation, yet achieving simultaneous EMW absorption and thermal conductivity in hexagonal boron nitride (h-BN) remains difficult due to its electrical insulation. Here, a simple and scalable mechanochemical strategy is developed to modify inert h-BN flakes (BNFs) with liquid metal (LM), activating their surface to generate abundant interfacial polarization centers. The optimized H-BNF@LM composite delivers outstanding EMW absorption with a minimum reflection loss of -48.4 dB and an effective absorption bandwidth of 5.76 GHz. Moreover, when integrated into an aramid nanofiber (ANF) matrix, the composite film exhibits a thermal conductivity nearly five times higher than that of pure ANF film. Beyond superior EMW absorption and thermal management, the films demonstrate excellent flexibility and remarkable flame retardancy, ensuring reliable operation even under harsh conditions. This work provides an efficient route for designing multifunctional composites suitable for next generation electronics.

A research team from Lanzhou University, China, has improved tree-ring simulations of a widely used forest growth model, 3-PG, by adding a carbon storage component. The new model version significantly enhances the model’s ability to simulate variations in both tree-ring widths and stable carbon isotope (δ¹³C). The upgrade addresses a key limitation in previous versions and provides a more physiologically accurate picture of how trees grow and store carbon over time.

Published in Forest Ecosystems, a seven-year study of loblolly pine plantations shows that crowded forests favor big trees in diameter growth, while smaller trees grow faster in height. Thinning rows and removing weaker trees slowed this dominance, letting smaller trees catch up and creating a more balanced forest. This shift also boosted overall wood production, offering insights for smarter forest management.

A new study of European beech trees reveals that their root systems respond more to short-term changes in soil water than to the long-term wetness of their growing sites. During dry periods, beech trees grow thinner, longer roots with more tips, enhancing water absorption, while wetter conditions lead to shorter, thicker roots. This seasonal root flexibility allows the trees to adapt rapidly to fluctuating soil moisture, highlighting the importance of monitoring short-term water availability for understanding tree resilience to drought.

Researchers have introduced a statistical method that allows accurate forest monitoring using satellite images with missing data. The hybrid estimator works directly with flawed data, bypassing the need for complex and uncertain data repair processes. This approach achieved over 90% sampling precision, meeting national forest inventory standards, and performed as well as techniques requiring complete satellite imagery. This provides a cost-effective way to leverage decades of archived satellite data for reliable forest and carbon stock assessment, supporting vital climate and conservation efforts.

Autonomous driving systems increasingly rely on data-driven approaches, yet many still struggle with reasoning, handling rare scenarios, and transparently explaining their actions. A new study introduces DriveMLM, a multi-modal large language model framework that aligns language-based reasoning with structured behavioral planning states, enabling full closed-loop driving in realistic simulators. By integrating multi-view images, LiDAR inputs, traffic rules, and natural-language instructions, DriveMLM generates both driving decisions and human-readable explanations that map directly to vehicle control. The system significantly improves safety, adaptability, and interpretability, demonstrating how large language models (LLMs) can advance the next generation of autonomous driving technology.