Inferring seagrass meadow resilience from self-organized spatial patterns
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
Abstract Assessing ecosystem resilience at large spatial scales remains a major challenge in ecology and conservation. While resilience is typically inferred from temporal dynamics or perturbation experiments, self-organized ecosystems may encode resilience-related information directly in spatial structure. Here, we combine a mechanistic model of Posidonia oceanica self-organization with convolutional neural networks trained on synthetic seascapes to infer meadow condition from single cartographic snapshots. The networks classify model-defined pattern states and estimate an effective mortality parameter along a deterioration gradient. Applied to habitat cartography across the Balearic Islands, the framework revealed regional variation in meadow condition. Cover-preserving randomization altered both classifications and inferred effective mortality, demonstrating dependence on spatial organization beyond vegetation cover alone. Networks trained exclusively on synthetic data transferred to real meadows without site-specific calibration. Our results establish a transferable strategy for integrating ecological theory and machine learning to infer resilience-related ecosystem condition when temporal observations are sparse or unavailable.
Fields of Study & Subjects
Marine and coastal plant biology
Ecosystem dynamics and resilience
Underwater Acoustics Research
Publication details
- Journal / source
- bioRxiv (Cold Spring Harbor Laboratory)
- Publication date
- 26 March 2026
- Language
- EN
- Document type
- Preprint