Ecosystem condition assessments: A context-specific workflow to integrate local expert knowledge and remote sensing

Despite decades of conservation science, we still struggle with a deceptively simple question: how do we know if an ecosystem is in good or poor condition? We present a reproducible, six-step workflow for assessing ecosystem condition using remote sensing, ecological knowledge, and expert validation. The approach is designed to be applied consistently across diverse biomes, while remaining sensitive to ecosystem-specific dynamics. It suggests steps for the detection of gradual changes in ecosystem condition driven by key pressures such as livestock farming and ranching, invasive species, or altered disturbance regimes, that traditional land cover classifications may miss. We demonstrate the workflow in Albany Thicket vegetation (South Africa) using high-dimensional satellite embeddings and a deviation-from-reference approach, capturing spatial gradients in ecosystem condition consistent with known degradation patterns and enabling continuous mapping for conservation and restoration planning and prioritization. By integrating expert input with scalable remote sensing metrics, the workflow provides ecologically meaningful and policy-relevant outputs. Importantly, it is aligned with both the IUCN Red List of Ecosystems and UN System of Environmental-Economic Accounting Ecosystem Accounting, thereby bridging two global standards and supporting national reporting towards the Kunming-Montreal Global Biodiversity Framework and Land Degradation Neutrality targets. This operational framework offers a transparent, adaptable, locally relevant and globally comparable approach for assessing ecosystem condition along a continuum of change, strengthening the evidence base for conservation, restoration, and emerging biodiversity markets.