Projections of current and future European potential vegetation types

The extent and condition of natural ecosystems is a key factor enabling species populations to thrive. However, the distribution of ecosystems is changing owing to both climatic and anthropogenic factors. Recently negotiated European policy directives, such as the Nature Restoration Regulation, argue for the restoration of natural ecosystems. Yet to determine what is to be restored the range of possible outcomes should be explored, also with regards to future climatic conditions. Here the concept of potential natural vegetation (PNV) is applied and mapped in a data-driven manner at European extent, exploring where PNV transitions are most likely to happen under contemporary and future conditions. Specifically, I predict current and future potential coverage of six natural vegetation types at 1 km 2 grain using Bayesian machine learning approaches, relying on a range of contemporary vegetation type records and climate and soil data for prediction. Most current land cover and land use could develop towards no single, but multiple PNV states. Results also indicate that suitable areas for some vegetation types, such as wetlands, might become rarer under future climatic conditions. Furthermore, the challenge of transitioning to PNV was found to be particularly high for current intensively cultivated landscapes. Overall data-driven PNV mapping holds considerable promise for assessing land potentials and supporting restoration assessments. Future work should expand the thematic grain of vegetation maps and also consider feedback with biotic factors. • Potential vegetation estimates are useful to delineate the upper potential of restoring natural land. • A Bayesian machine learning framework is used to infer the probability of potential natural vegetation across Europe. • Spatial-explicit probabilities for current and future potential natural vegetation types are made available. • Further advances in continental-scale natural vegetation modelling are outlined and how it can support regional planning.