In the years following the initial reestablishment of natural habitat, ecological restorations often require active management such as prescribed burns, supplemental seeding, and removal of invasive species in order to maintain important ecological functions and diverse biological communities. Though these management methods often target plants and vegetation structure, they have the potential to strongly affect a diverse set of other restored organisms both directly, by affecting their survival and reproduction, as well as indirectly, by altering their interactions with the plant community. The aim of our study was to identify methods of restoration management that affect the abundance and diversity of an important functional group, wild bees, and to separate out the direct effects of these methods from those that are mediated through the available floral resources. We therefore conducted a two year observational study of wild bees and flowering plants across 14 restored patches of tallgrass prairie habitat in an intensively managed prairie preserve in north-central Illinois, USA. We used piecewise structural equation modeling in order to examine the causal relationships between our variables of interest: restoration management factors, floral community variables, and bee community variables.
Though we found that management factors had a significant effect on restored communities of flowering plants, we did not find any evidence that these factors in turn affected bee communities, either directly or through their floral resources. Using the full hypothesized path model (Fisher’s C= 2.22, p= 0.33), we found that the abundance of floral resources was negatively affected by the age of the restoration (estimate= -0.096, p< 0.001) and positively affected by prescribed burning (estimate= 0.316, p= 0.003), and floral diversity was only negatively affected by restoration age (estimate= -0.047, p< 0.001). However, the sampled bee communities showed no response to restoration management or to these altered floral resources, contrary to our expectations. Instead, the bee communities only responded to year of sampling, both in terms of abundance (estimate= -0.419, p= 0.006) and diversity (estimate= -0.196, p= 0.017). We conclude that though methods of restoration management significantly impacted plant communities, observed variation in bee communities across restored sites was likely due to a variable not included in our preliminary analysis, such as landscape context.