Plant trait responses to herbivore type managed as domestic or as wild

The urgency of restoring ecosystems over vast areas has placed rewilding using wild herbivores at the forefront. However, few scientific studies address its effects on biodiversity and ecosystem functioning compared to more traditional conservation interventions with domestic herbivores. Equus ferus przewalskii horses introduced 30 years ago in the National Park of Cévennes, France - as a step of a conservation program of the species for its reintroduction in Mongolia - now occur as a semi-wild horse population, socially natural (i.e., management is based on natural selection). This introduction allowed us to compare this management system ('as wild') with other breeding systems: domestic sheep or domestic horses. Rewilding projects especially focus on the restoration of ecological functions, we therefore sought to find out if there are differences in the functional traits of the vegetation depending on the grazer and its management (domestic vs 'as wild') with the aim to generalize our results to other rewilding projects. Two levels of organization were tested (1) plant communities - using the TRY database and botanical surveys and (2) plant populations - by selecting four indicator species with traits measured in situ. Our results show that at equivalent grazing pressure, domestic horses or 'as wild' horses preserve the same plant functional types as in the traditional reference ecosystem grazed by sheep. Whatever the grazer and its management, grassland plant communities are characterized and dominated by perennial polycarpic plants, with oligotrophic to mesotrophic nutrient requirement, hemicryptophytes, graminoids, plants with entire leaf blades, not spiny, erect and mainly forming tussock. Nevertheless, some interesting significant differences were measured between the plant communities growing under the three types of grazing (e.g., lower functional richness and more plants with entire leaf blade under sheep grazing). Even between horse sub-species, differences were significant for some traits (e.g., more chamaephyte species growing under 'as wild' horses), which could be linked to physiological and morphological differences between the two horses sub-species (e.g., nutritional requirements, herbivore size) and by herd management. However, the intraspecific variability on the four quantitative traits measured on four indicator species do not show a general pattern. Although these results are valid at a given time and in a particular ecosystem, the vegetation functional traits approach can help in decision-making regarding ecosystem management choices and highlight the fact that the 'as wild' management adds an interesting alternative to conservation.