Discussion
Our results show strikingly consistent patterns in the response of ɑ-diversity of soil microbial and soil fauna groups to cessation of grazing. A number of key soil fauna and microbial groups, with the exception of soil bacteria, showed a marked decrease in ɑ-diversity following grazer exclusion, which also coincided with a marked decline in local plant species richness. The decline in plant species richness with grazer exclusion is consistent with previous studies showing that extensive grazing generally has a positive effect on local plant diversity (Olff & Ritchie 1998; Bakker et al. 2006; Epeldeet al. 2017). However, we found no evidence that plant species richness itself was a prominent proximate driver of changes in belowground α-diversity. Rather, our analysis showed that other factors, especially soil pH, carbon concentration, soil moisture, bulk density, litter biomass and aboveground biomass, were the most important determinants of grazer exclusion-induced changes in belowground α-diversity, largely consistent with previous studies showing that habitat characteristics are important determinants for the effect of grazer exclusion on soil communities in grasslands (e.g. (Bardgettet al. 1993; Bardgett et al. 1997; Bardgett et al.2001; Epelde et al. 2017; Oggioni et al. 2020). Nevertheless, we speculate that, although plant communities were not identified as a direct driver of soil fauna communities, they may be one of the ultimate drivers underlying these patterns in α-diversity. For example, we observed a shift towards fern- or dwarf shrub-dominated vegetation in many of the abandoned plots on acid soils, which is often associated with reduced soil pH and increased litter mass (Johnson-Maynard et al. 1998). Such vegetation-induced changes in soil biotic and abiotic conditions would then be the proximate driver of the observed changes in soil communities. This suggests that, rather than changes in plant species diversity per se , shifts in plant species composition ultimately drive the observed patterns in local belowground richness.
At the local scale, effects of cessation of grazing on belowground species were even more pronounced for relatively rare than for relatively widespread and common species. These results are consistent with McKinney & Lockwood’s (1999) original idea of ‘biotic homogenization’ and (scenario A & F in Figure 1): relatively rare species suffer more from land use change than relatively common or widespread species. We speculate that a mix of drivers may be responsible for this pattern. First, an important driver may be the loss of certain (rare) plant species from the areas where grazing was halted. Different plant species are known to selectively influence community composition in their rhizosphere (Bezemer et al. 2010; Leffet al. 2018). For example, most short grass species (e.g.,Cynosurus cristatus, Agrostis stolonifera ), legumes (e.g.,Trifolium repens ) and short herbs (e.g., Bunium bulbocastanum ) decreased strongly or disappeared altogether after grazing exclusion. In addition, the selective loss of relatively rare species might result from a decrease in local heterogeneity after grazing exclusion as a result of a lack of small scale trampling (Sørensen et al. 2009) or a lack of local defecation (Augustine & Frank 2001). Because of a lack of ecological information about specific plant-microbe and plant-fauna relationships and changes in patterns of local heterogeneity, it is impossible to provide conclusive evidence for each of the two hypotheses. As both processes typically coincide with removal of grazing (Adler et al. 2001; Pykälä 2005), it is likely that the resulting pattern can be generalized to other systems: cessation of grazing results in the loss of belowground species richness through the combined effect of a loss of local heterogeneity and local plant species richness.