Reproduce “Feasibility to detect rapid change and disappearance of seagrass: Lessons from nearly 80 years of vegetation change in the Ako, Seto Inland Sea, Japan” on OpenResearch

About this paper

Takehisa Yamakita, Yoji Igarashi, Akira Eto, Ken Ishida, Masaaki Iiyama

Researchers developed an integrated remote sensing workflow that combines nearly 80 years of historical aerial photographs with contemporary satellite imagery and deep learning (YOLOv8) to reconstruct and monitor seagrass distribution. This method successfully detected a rapid, near-complete disappearance of seagrass at the Ako tidal flat in 2025, demonstrating the feasibility of identifying both long-term trends and acute ecosystem collapses, while also providing practical recommendations for global biodiversity monitoring frameworks.

Key insights

• Seagrass meadows are dynamic ecosystems with long-term fluctuations driven by both anthropogenic and natural factors, rather than stable, ancient habitats.
• Integrated remote sensing and deep learning can detect not only gradual long-term trends but also rapid, acute ecosystem collapses, distinguishing them from normal variability.
• The workflow provides actionable recommendations for global biodiversity monitoring frameworks (e.g., TNFD's SoN metrics), including guidelines for baseline definition, seasonal standardization, and interpreting extreme anomalies.

Summary by alphaXiv.