Researchers at the University of Michigan have introduced a machine learning algorithm that automates the extraction of leaf traits from digital herbarium specimens, significantly advancing the scale and precision of plant ecological analysis. Developed by William Weaver, a Schmidt AI in Science fellow, and embedded within software designated as LeafMachine2, the algorithm efficiently measures leaf area and petiole width across extensive digital datasets. The tool resolves previous limitations in automating petiole measurement, a morphological parameter requiring precise geometric calculations, thereby enabling rapid processing of specimens held in collections such as the U-M Herbarium, which houses records dating to the 1800s. Published in New Phytologist, the research team applied the algorithm to analyze more than 22,000 leaves representing 1,580 species of woody angiosperms. The analysis revealed that predicted leaf mass per area, a proxy for structural investment, correlates more strongly with temperature and solar radiation than with precipitation variables on a global scale. The findings confirm that plants in tropical regions generally develop thicker leaves to support year-round retention, whereas seasonal species allocate resources differently. These results reinforce the utility of automated phenotypic measurement for modeling trait-environment relationships across diverse biomes. This work contributes to the 2026 edition of the State of the World's Plants and Fungi report, issued by the Royal Botanic Gardens, Kew. The report emphasizes the growing role of digital infrastructure in addressing global biodiversity and climate challenges. By leveraging digitized herbarium archives, scientists can identify critical knowledge gaps and prioritize conservation actions. The automation of trait extraction allows for the synthesis of historical data, facilitating the detection of phenotypic shifts associated with climate change and informing strategies to safeguard plant and fungal diversity. Beyond contemporary ecology, the algorithm supports paleoclimatology by establishing reliable proxies for fossil plant analysis. Applied paleobotanist Aly Baumgartner highlights that petiole thickness scales with leaf blade mass due to biomechanical requirements, offering a method to estimate leaf mass in fossil assemblages where direct measurement is unfeasible. This proxy approach enables researchers to reconstruct past climate conditions based on morphological patterns derived from extant species. Co-authors Thais Vasconcelos, Aly Baumgartner, Zoë Bugnaski, and James Boyko underscore that systematic botanical collection over centuries provides a unique historical baseline. Integrating artificial intelligence with these archives transforms static specimens into dynamic data assets, enhancing the scientific community's ability to monitor environmental change and support evidence-based conservation efforts.