Bacteria are ubiquitous in both our body and the environment, and most often exist in complex communities. These bacterial communities can serve positive purposes, such as in bioremediation, in which bacteria can find and degrade groundwater contaminants. They can also serve negative purposes, such as implantable device-related infection, in which biofilms grow on the surface of implanted medical devices. In all such relevant applications, bacteria develop a heterogeneous, complex community that varies spatially. However, we typically study these bacterial communities in well-mixed, liquid conditions, thereby eliminating spatial complexities. Here, we explore how spatial and temporal heterogeneity in chemical gradients, bacterial species, and bacterial phenotypes influences community function and behavior using a combination of simulations and experiments. Specifically, we consider ``pockets" of nutrients and investigate how the spacing between sources impacts bacterial motility. Additionally, we consider how a self-generated oxygen gradient can support multi-species communities of aerobic bacteria, which consume oxygen, and anaerobic bacteria, which die in the presence of oxygen. Finally, we develop the first system of equations that captures the phenotypic transition from free-swimming, planktonic bacteria to a sessile biofilm community. In all such studies, we develop minimal biophysical models to not only predict bacterial community composition, but also to design communities for future applications in human health and our environment.