Disordered and Partially Structured Models in Community Ecology: What are they? And how do we use them?

Community ecology describes how species interact with each other and with their environment. In nature, processes can be very complex because they involve hundreds to thousands of species interacting with each other in complex environmental landscapes. Classical approaches that have provided key insights have largely focused on the study of tractable subsets of species and patches, but these do not always adequately address the wider scope of natural complexity. Alternate approaches that use specific parameters and/or that use simulations to study such highly diverse systems are problematic because they can become very detailed, system-specific, and easily divorced from general principles. Finally, ’minimal’ models to explain data exist (e.g. null models, ’neutral theory’ and ’entropy based’ models), but they often do not provide adequate connections to experimental or mechanistic studies and results. Here we describe and discuss an alternate approach that seeks to link basic processes of community assembly (environmental heterogeneity, species interactions, dispersal, and stochasticity) with each other using ‘disordered systems models’ to make robust predictions about community structure, albeit without the detail of more system-specific approaches. We describe the logic of the approach, outline the methods involved, and identify important limitations. We also describe how this approach can be expanded to better incorporate additional nonrandom structure (such as intercorrelated parameters) in these basic processes and leading to ’partially structured models’, and we introduce the idea that this could also be applied to metacommunities. Although implementing this approach in empirical studies will still be quite challenging, these approaches reduce the complexity of the overall problem by orders of magnitude, making it a promising approach to improve the study of biodiversity in realistic landscapes