Abstract

How can we model complex ecosystems with stochastic interactions between individuals at different spatial and temporal scales? Some case studies

Ecological systems are hugely complex; and therefore prediciting how they behave raises many issues for modelling. In ecology, models can be either general, precise but inaccurate, or specific, accurate but imprecise, as the more detail they include to make them accurate ties them more closely to specific empirical systems. Further, the more processes they contain, the more imprecise they are likely to become (by creating variation). I will illustrate the issues with two empirical examples of ecological systems. First, a well-understood soil mite – where (a) the environment-phenotype map is very complex due to lagged effects on individual development, leading to considerable “noise” in the dynamics, and (b) the system evolves rapidly, such that parameters are not stationary. A natural approach to modelling this system is via individual-based models, or trait-based integral-projection models. The second case study is biodiversity dynamics in agricultural land. Here different taxa respond to the environment in context-dependent and scale-dependent ways. We need tools to understand how best to design landscapes for optimal delivery of ecosystem services. So the question arises as to how best to model complex ecosystems with stochastic interactions between individuals at different spatial and temporal scales. This is a classic question in systems biology: how do you take the necessary information across scales in an efficient way.