Effects of ecological restoration on soil microbial diversity in a temperate grassy woodland

Soil microbial communities are often overlooked in the context of ecological restoration. Given their central role in a broad range of ecosystem processes, however, understanding their response to restoration activities is critical to predicting restoration trajectories. In this study, we quantified the response of soil bacterial and fungal communities to restoration treatments, variation in microhabitat elements and vegetation in a critically endangered Australian box-gum grassy woodland ecosystem. Restoration treatments included the addition of coarse woody debris (CWD) and reduced grazing pressure. Four years after applying restoration treatments, we found no significant effect of CWD addition on soil microbial diversity, while reduced grazing significantly affected composition of the fungal, but not the bacterial, communities. Both bacterial and fungal communities responded to microhabitat element (open ground vs. old logs and trees), overlying vegetation and soil edaphic properties, and strong aboveground-belowground linkages were observed. Plant alpha diversity was positively correlated to soil bacterial, but not fungal, alpha diversity and plant community composition was a good predictor of both soil bacterial and fungal beta diversity. Co-occurrence network analysis identified numerous complex, non-linear associations between soil bacteria, fungi, edaphic properties and overlying plants. Soil microbes affected by restoration treatments included fungal saprotrophs and Actinobacteria, likely involved in litter breakdown, as well as bacteria likely involved in soil N cycling. Although the directions of the observed plant-microbe relationships remain unclear, we demonstrated the possibility of inducing changes to soil microbial communities to enhance restoration outcomes in box-gum grassy woodland ecosystems.Funding was provided by the Australian Capital Territory Government and an Australian Research Council Linkage Grant (LP110100126). ADM was supported by an Australian Research Council Future Fellowship (FT100100358)