Structure

The process of simplification described above typically reduces an ecosystem to a small number of state variables.Depending upon the system under study，these may represent ecological components in terms of numbers of discrete individuals or quantify the component more continuously as a measure of the total biomass of all organisms of that type，often using a common model currency.

The components are then linked together by mathematical functions that describe the nature of the relationships between them.For instance，in models which include predator-prey relationships，the two components are usually linked by some function that relates total prey captured to the populations of both predators and prey.Deriving these relationships is often extremely difficult given habitat heterogeneity，the details of component behavioral ecology（including issues such as perception，foraging behavior），and the difficulties involved in unobtrusively studying these relationships under field conditions.

Typically relationships are derived statistically or heuristically.For example，some standard functional forms describing these relationships are linear，quadratic，hyperbolic or sigmoid functions.The latter two are known in ecology as typeⅡand typeⅢresponses，named by C.S.Holling in early，groundbreaking work on predation in mammals.Both describe relationships in which a linkage between components saturates at some maximum rate（e.g.above a certain concentration of prey organisms，predators cannot catch any more per unit time）.Some ecological interactions are derived explicitly from the biochemical processes that underlie them；for instance，nutrient processing by an organism may saturate because of either a limited number of binding sites on the organism's exterior surface or the rate of diffusion of nutrient across the boundary layer surrounding the organism.

After establishing the components to be modelled and the relationships between them，another important factor in ecosystem model structure is the representation of space used.Historically，models have often ignored the confounding issue of space，utilizing zero-dimensional approaches，such as ordinary differential equations.With increases in computational power，models which incorporate space are increasingly used（e.g.partial differential equations，cellular automata）.This inclusion of space permits dynamics not present in non-spatial frameworks，and illuminates processes that lead to pattern formation in ecological systems.