Abstract
Predation can have substantial and long-term effects on the population dynamics of ecologically important prey. Diverse predator assemblages, however, may produce stabilizing (i.e., portfolio) effects on prey mortality when consumption varies asynchronously among predators. We calculated spatiotemporal variation in predation on a dominant forage species to quantify synchrony and portfolio effects in a food web context and better understand diversity-stability relationships in a large marine ecosystem that has undergone considerable changes in community composition. We selected Walleye Pollock (Gadus chalcogrammus) as our case study because they support some of the largest, most valuable commercial fisheries in the world and serve as essential prey for an array of economically and culturally important species. Thus, there are sufficient data for Pollock with which to test ecological theories in an empirical setting. Spatially explicit predation indices accounted for annual variation in predator biomass, bioenergetics-based rations, and age-specific proportions of Pollock consumed by a suite of groundfishes in the Gulf of Alaska (1990-2015). We found that Arrowtooth Flounder (Atheresthes stomias) was, by far, the dominant Pollock predator (proportional consumption: 0.74 ± 0.14). We also found synchronous trends in consumption among predator species, indicating a lack of portfolio effects at the basin scale. This combination of a single dominant predator and synchronous consumption dynamics suggests strong top-down control over Pollock in the Gulf of Alaska, though the degree of synchrony was highly variable at all spatial scales. Whereas synchrony generally increased in the western subregion, consumption in the central Gulf of Alaska became less synchronous through time. This suggests diminished trophic stability in one area and increased stability in another, thereby emphasizing the importance of spatiotemporal heterogeneity in maintaining food web structure and function. Finally, total Pollock consumption was highly variable (ranging from 1.87 to 7.63 Tg) and often exceeded assessment-based estimates of productivity. We assert that using our holistic and empirically derived predation index as a modifier of assumed constant natural mortality would provide a practical method for incorporating ecological information into single-species stock assessments.