Processes cause changes of the state variables and, in case of individuals, also of the number of entities. Examples of processes in IBMs include growth, dispersal, foraging, habitat selection, mating, habitat dynamics, disturbance events, phenology, mortality, reproduction, etc. The modeler has to decide which process to represent explicitly, and in which detail. This decision is linked to the decisions on state variables and scales. If, for example, temporal resolution is 1 year, the process offoraging may not need to be represented explicitly. If, on the other hand, habitat selection in response to short-term environmental changes is considered important for the problem addressed with the model, a temporal resolution of 1 day or even less might be required, and in turn an explicit representation of the individual's decision of where to move in the next time step.
If processes are not considered important enough to be included explicitly, they usually are represented by constant parameters. Mortality, for example, can explicitly depend on foraging and interaction with predators; or, mortality is represented by just a constant parameter, which in the model is implemented as the probability of dying in a certain time step.
The submodels implementing the model processes are often formulated as a combination of mathematical expressions and IF-THEN conditions. Growth, for example, can be represented by a growth equation, whereas habitat choice will require probabilistic IF-THEN rules: ''IF the best habitat within my perception range has higher quality then my current habitat AND IF predation risk is low THEN with a certain probability I will move to the new habitat.''
All design decisions of the modeler regarding entities, processes, scheduling, etc., are experimental, that is, they have to be tested and analyzed and, as a result, usually be modified.
Was this article helpful?