Non-equilibrium physical systems, be they biological or otherwise, are powered by differences in intensive thermodynamic variables, which result in flows of matter and energy through the system. This thesis is concerned with the response of physical systems and ecosystems to complex types of boundary conditions, where the flows and intensive variables are constrained to be functions of one another. I concentrate on what I call negative feedback boundary conditions, where the potential difference is a decreasing function of the flow. Evidence from climate science suggests that, in at least some cases, systems under these conditions obey a principle of maximum entropy production. Similar extremum principles have been suggested for ecosystems. Building on recent work in theoretical physics, I present a statisticalmechanical argument in favour of this principle, which makes its range of application clearer. Negative feedback boundary conditions can arise naturally in ecological scenarios, where the difference in potential is the free-energy density of the environment and the negative feedback applies to the ecosystem as a whole. I present examples of this, and develop a simple but general model of a biological population evolving under such conditions. The evolution of faster and more efficient metabolisms results in a lower environmental energy density, supporting an argument that simpler metabolisms could have persisted more easily in early environments. Negative feedback conditions may also have played a role in the origins of life, and specifically in the origins of individuation, the splitting up of living matter into distinct organisms, a notion related to the theory of autopoiesis. I present simulation models to clarify the concept of individuation and to back up this hypothesis. Finally I propose and model a mechanism whereby systems can grow adaptively under positive reinforcement boundary conditions by the canalisation of fluctuations in their structure.