Predictive models in spatially and temporally variable freshwater systems

This paper discusses the relationships between scaling and predictability in ecosystems. The logical basis of ecosystem modelling is explored using ideas first developed in complexity theory and analogies with the behaviour of complex adaptive systems. Any ecological model is a scale-dependent entity and both empirical and dynamic models of freshwater systems have their strengths and weaknesses. The logical basis of modelling using functional groups is explored. I conclude that such an approach can be justified and that such models have predictive power. Any predictive model of freshwater systems must take the major scales of external (atmospheric and catchment) forcing into account as well as the scales of key processes in the ecosystem itself. The importance of so-called 'pink noise' spectra, which arise both from external forcing and the internal dynamics of dynamic systems, is noted. The key scales of pattern and process in freshwater ecosystems are discussed in relation to the properties of the major functional groups. In order to have predictive power, I conclude that models of freshwater systems must include sediment exchanges and the properties of aquatic macrophytes as well as water column interactions and the pelagic components. When viewed at the scale of functional groups and the major biogeochemical processes, freshwater ecosystems may not be as complex as is often assumed.