We investigated the relative roles of bottom-up and top-down factors in limiting productivity of an upper trophic level marine predator. Our primary working hypothesis was that the reproductive success of black-legged kittiwakes (Rissa tridactyla) a piscivorous, colonial-nesting seabird, was most limited by the abundance, distribution, and species composition of surface-schooling forage fishes. A secondary working hypothesis was that reproductive loss to kittiwake nest predators was greatest during years of reduced prey availability. We report on a broad-scale, integrated study of kittiwakes and their prey in Prince William Sound, Alaska. Our study spanned five breeding seasons (1995-1999) and focused on three colonies that differed in size (ranging from ca. 220 to ca. 7000 breeding pairs) and proximity to each other (50-135 km apart). Kittiwakes in PWS encountered a variety of aquatic habitats, creating a complex foraging environment for breeding birds. We measured kittiwake reproductive success and foraging activities, while simultaneously measuring the abundance of surface schooling forage fishes throughout the foraging range of breeding kittiwakes. The abundance of primary prey species for kittiwakes (Pacific herring Clupea pallasi, Pacific sand lance Ammodytes hexapterus, and capelin Mallotus villosus) varied both annually and regionally, with no one region consistently having the greatest abundance of prey. Likewise, kittiwake reproductive success varied considerably among colonies and years. We found that bottom-up, top-down, timing mismatch, and colony-specific effects were all important to kittiwake productivity. Although bottom-up effects appeared to be strongest, they were not evident in some cases until other effects, such as geographic location (proximity of colony to prey concentrations) and top-down predation, were considered. Important bottom-up effects on kittiwake reproductive success were not only total prey abundance and distribution, but also species, age composition, and chronology of prey occurrence (match/mismatch of timing with critical brood-rearing periods); these effects varied by colony. Top-down effects of predation on kittiwake nest contents (independent of prey abundance) confounded seabird-forage fish relationships. Ultimately, when confounding factors were minimized, non-linear asymptotic relationships were identified between kittiwakes and their prey, with an asymptotic threshold of fish school surface area density of ca. 5 m2/km2, beyond which top-down, physiological, or phylogenetic constraints likely restrict further reproductive output. The integrated approach of our investigations provided a more thorough understanding of the mechanisms underlying predator-prey relationships in the complex marine environment. However, such mechanistic theories can only be tested and refined through long-term research and monitoring of much greater duration than the 5-year study reported herein. ?? 2006 Elsevier Ltd. All rights reserved.