Understanding what governs grassland biodiversity across different spatial scales is crucial for effective conservation and management. However, current evidence often focuses on single sampling grain sizes, leaving the mechanisms of biodiversity drivers and their scale-dependency unclear. Here, we investigated the impact of climate, soil properties, abiotic disturbance, and land use on plant diversity across fine spatial scales in various grassland types. We collected spatially explicit data on species presence, relative cover, and total community cover at two grain sizes (α- and γ-diversity) to assess the mechanisms driving scale-dependent diversity patterns (β-diversity). In our study, the most influential factors of plant diversity at both scales (grain sizes) were climate variables, followed by soil humus content, litter cover, and soil pH. The effects of soil and litter were primarily driven by the response of rare species, while climate and grazing effects were driven by locally common species. The strength of most of these effects varied between spatial scales and therefore affected β-diversity. We identified three key mechanisms through which these drivers affect the scale-dependency of biodiversity: total plant cover, species relative cover (commonness or rarity of species and species evenness in the community), and species intraspecific aggregation. Climate effects operated through changes in species relative cover and intraspecific aggregation. Soil humus influenced β-diversity by altering the total cover of the plant community and by increasing intraspecific aggregation, resulting in stronger effects of soil productivity on plant diversity at larger than smaller spatial scales. Microhabitat patchiness by litter altered distributions in the relative cover of species due to reduced asymmetric competition, and affected the total cover of the plant community. Our results underscore the importance of incorporating the scale-dependency of biodiversity drivers in conservation efforts, management strategies, and analyses of global change impacts, which would enhance our ability to predict potential biodiversity change.

Changing climatic conditions and unsustainable land use are perceived as major threats to savannas worldwide. In the past, land use in African savannas was dominated by livestock-farming as one of the major economic products, which led to degraded, shrub encroached pastures in many regions. One response to this widespread degradation is a shift from land use dominated by cattle to strategies characterized by animal compositions with more mixed feeding regimes and higher browser proportions. However, the consequences for ecosystem properties and processes remain so far largely unclear. We used the ecohydrological, spatially explicit savanna model EcoHyD to assess the impacts of two contrasting, herbivore-related land use strategies on a Namibian savannah: grazing versus browsing herbivores. We varied the densities of grazers and browsers and determined the resulting composition and diversity of the plant community, total vegetation cover, soil moisture and water use by plants. Our results show that properties making plants less attractive to herbivores were best adapted to different densities of grazing (cattle) or browsing (pure browsing wildlife) animals. Also, properties leading to a competitive advantage under limited water availability were among the dominant ones. Overall, the results are in line with our expectations: we found heavy shrub encroachment with a loss of the perennial grass matrix under high stocking rates of cattle. A novel and unexpected result was that regardless of the density of browsers, grass cover and plant functional diversity were significantly higher in wildlife scenarios. This increased grass cover, but also the higher total cover improved water uptake by plants. We conclude that in contrast to grazers, browsers even in high densities do not lead to ecosystem degradation, but rather sustain a diverse vegetation with high cover of perennial grasses over a long time, implying also a lower erosion risk and higher provision of ecosystem services.