Modularity maximization with different null models can reveal distinct types of community structures in granular networks. In a network representation in which particles are represented by nodes and contact forces are represented by weighted edges, (a) using the NG null model helps uncover contiguous domains in a granular system, and (b) using a geographical null model on the same network helps detect chain-like structures that are reminiscent of force chains. In both panels, nodes (i.e., particles) of the same colour are assigned to the same community.