High-resolution species distribution modelling to guide eelgrass (Zostera marina) management

Eelgrass (Zostera marina) meadows are declining globally despite their critical role as foundational ecosystems and water quality indicators, e.g. under the EU Water Framework Directive (WFD). With the EU Nature Restoration Regulation mandating recovery targets, robust habitat distribution maps are urgently needed to support evidence-based restoration. Traditional surveys provide detailed local estimates but lack the spatial coverage required for comprehensive mapping.<br/>We developed a high-resolution (50 × 50 m) species distribution model (SDM) for Z. marina across Denmark's ∼7000 km coastline using comprehensive national monitoring data (NOVANA, 2011-2022). A Random Forest model trained on ∼30,000 field observations and validated on 7500 independent samples achieved strong predictive performance (accuracy: 80.3%, 95% CI: 79.4–81.2%), estimating current coverage at ∼2000 km2.<br/>Critically, the model moves beyond traditional habitat suitability mapping to also predict depth distribution limits (Zmax), achieving strong agreement with independent WFD field observations of Zmax for 85 of 109 water bodies (r = 0.779, slope = 0.99, mean error &lt;1 m). This accuracy provides actionable restoration targets aligned with EU mandates.<br/>Accumulated Local Effects (ALE) analysis identified light availability at depth (Iz) as the most influential predictor (ALE range = 0.83), followed by Maximum Exposure Index (MEI, 0.70) and 95th percentile temperature (T95, 0.68), all showing nonlinear relationships with eelgrass presence.<br/>Beyond mapping current distribution, the model enables scenario-based assessments of restoration feasibility under improved water clarity and climate change. The methodology is scalable to other species and regions, offering a robust framework for adaptive management and supporting EU restoration goals.