This thesis combines methods from statistical physics and nonlinear dynamics to advance research on the pattern formation in active fluids in several directions. In particular, it focuses on mesoscale turbulence, a state observed in microswimmer suspensions, which is characterized by the emergence of dynamic vortex patterns. The first major contribution concerns the bottom-up derivation of a frequently used continuum model of mesoscale turbulence from a set of particle-resolved stochastic equations. Utilizing the model, mesoscale turbulence is shown to induce nontrivial transport properties including a regime of optimal diffusion. The thesis then explores possible strategies of control. One of these relies on an external field that leads to stripe-like structures and can even suppress patterns entirely. The other involves geometric confinement realized by strategically placed obstacles that can reorganize the flow into a variety of ordered vortex structures. The turbulence transition inside an obstacle lattice is shown to have an intriguing analogy to an equilibrium transition in the Ising universality class. As a whole, this thesis provides important contributions to the understanding and control of turbulence in active fluids, as well as outlining exciting future directions, including applications. It includes a substantial introduction to the topic, which is suitable for newcomers to the field.