🌿 COVER STORY — Mangroves Rebound: Four Decades of Global Recovery

Research Article · pp. 1082–1087 · DOI: 10.1126/science.aec9773

Global mangrove forests have disappeared rapidly because of deforestation but have also regrown through natural regeneration and restoration. Yet their long-term trends in extent and canopy cover remained poorly quantified. By developing a global dataset, Zhang et al. reveal that four decades of satellite observations show that mangrove forests globally have recently shifted from net loss to net gain, with reduced degradation and continued regrowth and expansion, suggesting early signs of conservation success. The study draws on remote sensing and AI-assisted image analysis across hundreds of millions of satellite pixels — a model for how computational earth observation is reshaping conservation science. ScienceScience

🤖 A Narrowing Window to Understand AI — Eric Horvitz & Robert West (Commentary, p. 1003)

DOI: 10.1126/science.aei3167

As capabilities of artificial intelligence advance rapidly, human understanding of these systems is increasingly falling behind. Several trends are converging to make AI systems harder to understand just as they become more consequential. Without deliberate countervailing efforts, the window for building AI systems that we can meaningfully understand and guide may close beyond recovery. Understanding AI need not mean grasping every line of code or every neural-network parameter — just as we study human behavior at multiple levels, from neuroscience to psychology to sociology, AI principles and operations can be explored and understood at varying levels. Written by Microsoft's chief scientific officer and an EPFL researcher, this is a pointed call to action for interpretability research before it becomes too late. Science

🔭 Will AI Replace Astronomers? — Joshua Sokol (News Feature, Vol. 392, Issue 6802)

https://www.science.org/content/article/amid-flood-ai-advances-astrophysicists-are-questioning-soul-their-field

Researchers discussed how they planned to apply machine learning to problems in astronomy — observing an interstellar comet, discerning wispy filaments of galaxies at the universe's largest scales. None of the scientists interviewed seemed confident projecting whether, in a few months or a year, their concerns about AI's role will seem embarrassingly overblown or prescient. This wide-ranging feature captures growing anxiety in the astrophysics community about whether AI is a powerful tool or an existential threat to the human practice of science. Science

🦠 Molecular Mimicry in Plant Immunity — Banfield, Talbot, Moscou et al. (Research Article, pp. 1050–1055)

DOI: 10.1126/science.aef9946

Plants and animals respond to pathogen attack by mounting innate immune responses that require intracellular nucleotide-binding leucine-rich repeat (NLR) proteins. These immune receptors detect pathogen infection by sensing virulence effector proteins. However, the mechanisms by which receptors evolve new recognition specificities remain poorly understood. This structural and computational study reveals that a barley NLR receptor (MLA3) mimics the plant host target of a rice blast fungal effector — providing a blueprint for rationally engineering disease resistance in crops. bioRxiv

🐝 Bumble Bees Demonstrate Spontaneous Problem-Solving — Bhambore, Akmeşe, Loukola et al. (Research Article)

DOI: 10.1126/science.ady1618

What makes this behavior especially remarkable is that the bees had never been trained to roll the ball. This was a completely new challenge. Their behavior appeared goal-directed, with successful individuals showing more directed movement patterns. Spontaneous problem-solving is something that has never been shown in any invertebrate before. Vertebrates including chimpanzees and parrots can problem solve on their own — the new finding adds yet another skill to bumblebee repertoires. With implications for understanding minimal neural substrates of cognition, this study is also a reference point for AI research on goal-directed behavior in systems with constrained resources. Phys.orgInteresting Engineering

❤️ Obesity Rewires Heart Muscle in HFpEF — Jani, Rhodehamel, Kass et al. (Research Article)

DOI: 10.1126/science.adz7118

Heart failure with preserved ejection fraction (HFpEF) causes substantial morbidity and mortality and has few effective therapies. Its phenotype has changed over time, with morbid obesity and metabolic defects supplanting hypertension and cardiac hypertrophy. Scientists analyzed heart muscle cells from people with HFpEF and identified abnormalities including weakened force production in muscle cells obtained from the hearts of very obese people. In a subgroup who underwent weight loss therapy, those who lost the most weight had better contraction of their heart muscle cells, suggesting that some molecular changes may be reversible. The findings caution against certain existing cardiomyopathy drugs in this population and point toward weight-loss-centered therapies. Unbound MedicineEurekAlert!

🥗 Ultraprocessed Foods and Obesity: Evidence Reconsidered — Magkos, Forde & Robinson (Commentary, pp. 1020–1022)

DOI: 10.1126/science.aef3495

Although several randomized controlled trials have explored the effects of ultraprocessed foods on obesity — leading to widespread claims that ultraprocessing is inherently harmful — the authors argue that the evidence does not support this conclusion. Based on the design of existing clinical trials, it is difficult to attribute any negative health effects directly to ultraprocessing itself. Instead, these outcomes are more likely driven by common nutritional characteristics of ultraprocessed foods, including soft texture, which may lead people to eat more and more quickly. A significant reframing of a hotly contested public health debate. Medical Xpress

Issue theme in brief: This issue pairs a landmark AI-interpretability essay with a striking news feature on AI displacing scientific practice — making it the most AI-focused issue of the three covered. On the life sciences side, the heart failure sarcomere study, the plant immunity structure, and the bumblebee cognition paper are the standout research articles.