The objective of this research is to characterize the dynamics of ecological networks. The work presented here takes a systems approach. It establishes a total system view of ecosystem behaviors and behavioral relationships before drilling down to details. The research is a blend of synthesis and analysis. We synthesize a function-structure-process framework that provides the high-level context for a full range of ecological system functions and their implementation. The implementation architecture of ecosystems is the network. Within this context, and based on an extensive review of the complex systems and network literature, we synthesize a view of ecological network dynamics. This view, in turn, is the basis for the central hypothesis of the research: ecological networks are ever-changing networks with propagation dynamics that are punctuated, fractal, and enabled by indirect effects. At this point, analysis becomes the focus of the work. We define, design, and develop an ecological network dynamics model to analyze and fully test the hypothesis. Our innovative modeling approach seeks to emulate features of real-world ecological networks. The approach does not make a priori assumptions about ecological network dynamics, but rather lets the dynamics develop as the model simulation runs. The model software development effort is substantial and includes not only a comprehensive implementation of ecological network processes but also a full complement of analysis capabilities and graphics generation procedures. Model analysis results corroborate our hypothesis. We see that ecological networks exhibit fractal behavior in space and time. Network events have punctuated time series and power-law/fractal distributions. We see that ecological networks flicker. (Network structure is not static and network flows are not steady state.) We see that indirect effects play a prominent role in ecological network dynamics. When observing the total behavioral picture, we can glimpse a general equivalence a universality in ecological network dynamics. We notice that these dynamics exhibit fundamentally the same form of behavioral statistics across spatial and temporal scales and even across processes. The same, apparently universal, principles seem to apply.