ABSTRACT
Localized hypoxic habitats were created in Delta Marsh, Manitoba, Canada to determine the potential of regions of moderate hypoxia to act as refuges for forage fishes from piscine predators. Minnow traps and giving-up density (GUD) plates (plexiglas plates covered with trout crumble and fine gravel) were used to assess habitat use and perceived habitat quality for forage fishes, respectively, while passive integrated transponder tags provided data on habitat use by predator species to assess the level of predation risk. Data were collected both before and after a hypoxia manipulation (2-3 mg l(-1) dissolved oxygen, DO) to create a before-after control-effect style experiment. Fathead minnows Pimephales promelas were more abundant and consumed more food from GUD plates in hypoxic bays after the DO manipulation, indicating hypoxic locations were perceived as higher quality, lower-risk habitats. The frequency of predator visits was not consistently affected. The duration of visits, and therefore the total time spent in these habitats, however, was significantly shorter. These predator data, combined with the prey information, are consistent with the hypothesis that hypoxic regions function as predator refuges. The refuge effect is not the result of predator exclusion, however; instead predators are rendered less capable of foraging and pose less of a threat in hypoxic locations.