The main theme of this dissertation is a challenge to the traditional paradigm of optimal control for the management of renewable resources. That is not to say that optimization should not be the goal but rather, previous modeling and policy have presumed that managers have greater control and knowledge of marine systems than is the case. All models make simplifying (and false) assumptions but assuming control and knowledge in marine systems is not benign. In a dynamic control problem, one must actually control the variable of interest and know or learn the system's parameters. I discuss reasons why managers may not control harvests and cannot know system parameters and consider possible remedies to the historically sub-optimal management of marine resources. Marine resources are observed imperfectly and are often held as common property. In this dissertation I explore the feasibility of management plans when naturalcapital stock dynamics are unobservable and when political structures constrain the implementation of optimal management. Resource managers are faced with conflicting user groups and limited information. In three chapters I study constrained management In my first chapter, "Jobs or Resources?" I consider the political economy implications of technological under various management scenarios. I show that typical management targets will require the retirement of inputs as technology progresses. This is particularly problematic for fisheries in which labor is involved in management decisions. This can lead to false inference about the health of the stock. For the second chapter, "Natural Resource Collapse: Technological Change and Biased Estimation", I show that unexpected fisheries collapse may be linked to unobserved technological change. Unexpected collapse of natural resources is of great concern to policy makers. The literature and popular press have attributed collapse to the lack of well-defined property rights and policies that pay inadequate attention to random environmental variability. Both the literature and policy makers have overlooked how unobserved technological change can obscure the depletion of natural- capital stocks. The paper shows that even if property rights are well-defined and random fluctuations are small, modest increases in technical efficiency conceal the depletion of stocks. Using the most general model of surplus production in a single-species fishery I show analytically that proportional growth of the fish stock is overestimated when even one period of technological change is ignored. Through simulations, I find that standard statistical tests overestimate the productivity of the fish stock. I show that collapse is inevitable if technology increases without bound and that the path to collapse is not observed until stocks are low and declining rapidly. In the third chapter, "Marine protected areas as a risk management tool" I consider a potential fix to the inference problems highlighted in the second chapter and in other work such as Carson and Murray (2005). When parameter uncertainty is significant, I show that even in an otherwise deterministic world, expected payoffs can be increased by using simple spatial closures. Though optimal fleet size and reserve size combinations exist, a spatial closure can increase expected payoff even if the fleet-size is chosen to be too large or too small. The benefit of closures is not limited to hedge against stock collapse but is of value even when stock size is large and steady-state catches are relatively high