Extending ecological network analysis measures to dynamic ecosystem models

Abstract

Ecological network analysis measures such as cycling index, indirect effects, and storage analysis provide insightful information on ecosystem organization and function, which can be extremely useful for environmental management and control. These system-wide measures focus on indirect relations among system compartments, providing a holistic approach. Unfortunately, the application of these useful measures are restricted to steady state models. Seasonal changes, environmental impacts, and climate shifts are not accommodated by the current methodology, which greatly limits their application. The novel methodology introduced in this paper extends the application of these useful but limited measures to dynamic compartmental models. This method relies on network particle tracking simulation, which is an agent based algorithm, whereas the current methods utilize steady-state flow rates and compartment storage values. We apply this new methodology to storage analysis, which quantifies how much storage is generated at any compartment within the system by a unit external input into another compartment. Also called compartmental mean residence time, this measure is widely used in environmental sciences, pharmacokinetics and nutrition, to assess the interaction between system boundary (e.g. drug intake, pollution, feeding) and internal compartments (e.g. tissues, crops, species). Storage analysis is chosen for demonstration because it is applicable to a limited class of dynamic models (linear and donor-controlled), which gives us an opportunity to verify our new method. The methodology introduced here is also applicable to Finn's cycling index, indirect effects index, throughflow analysis, and possibly other network analysis based indicators as well.