Network analysis: Foundations, extensions, and applications of a systems theory of the environmen

The goal of this dissertation is to show that environmental influences, which are propagated by the direct and indirect interactions of the internal and external network organization, significantly affect system behavior. It focuses on four emergent properties or hypotheses: amplification (Patten et al., 1990), homogenization (Patten et al., 1990), synergism (Patten, 1991), and indirect effects (Higashi and Patten, 1986). These properties give insight into the behavior of holistic network interactions. This research is broken into two areas, theoretical development and application of these properties.;The first of three theoretical advances is a formal description of the path analysis and a methodology to create random flow matrices from connection matrices (function from structure). Second, network homogenization is made quantitative and testable. Third, network synergism is used to show that all networks are on the whole dominated by positive, mutualistic relations.;Four applications of network properties are given. Utility analysis is a control measure because it simultaneously accounts for both top-down and bottom-up processes. As a result, ecosystem control is distributed among many components. When the network properties are tested on large-scale systems, homogenization and indirect effects increase with size, synergism decreases, and amplification does not occur. These results indicate that homogenization and indirect effects are underestimated in small-scale, empirically based models because real ecosystems are generally composed of many components. Third, utility analysis is used as a conditional explanation for the observation of low transfer efficiency between components. A limitation exists because of the constraint in computing the utility analysis. Fourth, utility analysis is introduced as an ecological goal function parameter. The benefit to using utility analysis is its dual determination by both throughflow and net flow because it is maximized when these two processes are optimized. The developments and applications presented here support network analysis as a useful way to investigate system structure and function, and the results reinforce the importance of considering direct and indirect environmental contributions to overall system behavior.