To implement ecosystem-based approaches to fisheries management, decision makers need insight on the potential costs and benefits of the policy options available to them. In the Southern Ocean, two such options for addressing trade-offs between krill-dependent predators and the krill fishery include "feedback management" (FBM) strategies and marine protected areas (MPAs); in theory, the first adjusts to change, while the latter is robust to change. We compared two possible FBM options to a proposed MPA in the Antarctic Peninsula and Scotia Sea given a changing climate. One of our feedback options, based on the density of Antarctic krill (Euphasia superba), projected modest increases in the abundances of some populations of krill predators, whereas outcomes from our second FBM option, based on changes in the abundances of penguins, were more mixed, with some areas projecting predator population declines. The MPA resulted in greater increases in some, but not all, predator populations than either feedback strategy. We conclude that these differing outcomes relate to the ways the options separate fishing and predator foraging, either by continually shifting the spatial distribution of fishing away from potentially vulnerable populations (FBM) or by permanently closing areas to fishing (the MPA). For the krill fishery, we show that total catches could be maintained using an FBM approach or slightly increased with the MPA, but the fishery would be forced to adjust fishing locations and sometimes fish in areas of relatively low krill density-both potentially significant costs. Our work demonstrates the potential to shift, rather than avoid, ecological risks and the likely costs of fishing, indicating trade-offs for decision makers to consider.