Non-equilibrium thermodynamics has had a significant historic influence on the development
of theoretical ecology, even informing the very concept of an ecosystem. Much of this influence
has manifested as proposed extremal principles. These principles hold that systems will tend
to maximise certain thermodynamic quantities, subject to the other constraints they operate
under. A particularly notable extremal principle is the maximum entropy production principle
(MaxEPP); that systems maximise their rate of entropy production. However, these principles
are not robustly based in physical theory, and suffer from treating complex ecosystems in
an extremely coarse manner. To address this gap, this thesis derives a limited but physically
justified extremal principle, as well as carrying out a detailed investigation of the impact of
non-equilibrium thermodynamic constraints on the assembly of microbial communities. The extremal
principle we obtain pertains to the switching between states in simple bistable systems,
with switching paths that generate more entropy being favoured. Our detailed investigation
into microbial communities involved developing a novel thermodynamic microbial community
model, using which we found the rate of ecosystem development to be set by the availability
of free-energy. Further investigation was carried out using this model, demonstrating the way
that trade-offs emerging from fundamental thermodynamic constraints impact the dynamics of
assembling microbial communities. Taken together our results demonstrate that theory can be
developed from non-equilibrium thermodynamics, that is both ecologically relevant and physically
well grounded. We find that broad extremal principles are unlikely to be obtained, absent
significant advances in the field of stochastic thermodynamics, limiting their applicability to
ecology. However, we find that detailed consideration of the non-equilibrium thermodynamic
mechanisms that impact microbial communities can broaden our understanding of their assembly
and functioning.