Questions: Studying dry grasslands in a previously unexplored region, we asked: (a) which environmental factors drive the diversity patterns in vegetation; (b) are taxonomic groups (vascular plants, bryophytes, lichens) and functional vascular plant groups differently affected; and (c) how is fine-grain beta diversity affected by environmental drivers?.
Location: Northwestern and Central Bulgaria.
Methods: We sampled environmental data and vascular plant, terricolous bryophyte and lichen species in 97 10-m² plots and 15 nested-plot series with seven grain sizes (0.0001-100 m²) of ten grassland sites within the two regions. We used species richness as measure of alpha-diversity and the z-value of the power-law species–area relationship as measure of beta-diversity. We analysed effects of landscape, topographic, soil and land-use variables on the species richness of the different taxonomic and functional groups. We applied generalised linear models (GLMs) or, in the presence of spatial autocorrelation, generalised linear mixed-effect models (GLMMs) in a multi-model inference framework.
Results: The main factors affecting total and vascular plant species richness in 10-m² plots were soil pH (unimodal) and inclination (negative). Species richness of bryophytes was positively affected by rock cover, sand proportion and negatively by inclination. Inclination and litter cover were also negative predictors of lichen species richness. Elevation negatively affected phanerophyte and therophyte richness, but positively that of cryptophytes. A major part of unexplained variance in species richness was associated with the grassland site. The z-values for total richness showed a positive relationship with elevation and inclination.
Conclusions: Environmental factors shaping richness patterns strongly differed among taxonomic groups, functional vascular plant groups and spatial scales. The disparities between our and previous findings suggest that many drivers of biodiversity cannot be generalised but rather depend on the regional context. The large unexplained variance at the site level calls for considering more site-related factors such as land-use history.