What are these ecological systems that are disassembled by human disturbances and subsequently need restoration?  Can basic understanding of such systems be applied to predict the consequences of species loss and invasion as well as aid restoration?  We present a general theory of ecological systems as complex networks of coupled populations nonlinearly fluctuating in time that might answer these questions.  Several of the theory’s analytical and empirical successes will be described along with emerging insights for conservation biology.  Successes include the theory’s close fit to the central tendencies of ecological network structure and the empirically corroborated parameter space needed for many species to persist despite chaotic network dynamics.  Successes also include predictive hypotheses emerging from computational explorations of species loss within biotically and abiotically variable environments.  Limits to further insights will also be discussed including the need for more data to distinguish network structures remaining after disassembly from naturally assembled networks.  Limits to better understanding of network dynamics include the need for richer sets of functional responses that describe how feeding depends on myriad aspects of nature.  Finally, progress towards and prospects for understanding spatially explicit structure and dynamics will also be discussed.