The structure and state of an ecological community is dependent both upon its comprising species and species interactions. One approach for studying such complex systems is an ecological network framework, which emphasizes the importance of the structure of species interactions at the community level and investigates relationships between community-level structure and system resilience. Here, we investigate changes in the network structure of communities of interacting pollinating insects and flowering plants in grassland habitats with different fire and grazing disturbance regimes. Communities were sampled at 12 watersheds in summer 2014 on Konza Prairie Biological Station (KS, USA) subjected to three fire interval treatments (1, 4, and 20 year intervals) and factorially crossed with two grazer treatments (presence or absence of the large ungulate grazer Bison bison). A stepwise partial redundancy analysis resulted in a top model containing grazer treatment, historic burn interval and conditional on sampling date. The final model explained 30 % of the variance in network structural properties with the largest proportion of variance explained by grazing treatment (20.9%). Understanding relationships between environmental drivers and interaction structure improves our ability to predict changes in ecological communities under changing habitat conditions.