Abstract
Global-scale studies of marine food webs are rare, despite their necessity for examining and understanding ecosystem level effects of climate variability. Here we review the progress of an international collaboration that compiled regional diet datasets of multiple top predator fishes from the Indian, Pacific and Atlantic Oceans and developed new statistical methods that can be used to obtain a comprehensive ocean-scale understanding of food webs and climate impacts on marine top predators. We loosely define top predators not as species at the apex of the food web, but rather a guild of large predators near the top of the food web. Specifically, we present a framework for world-wide compilation and analysis of global stomach-contents and stable-isotope data of tunas and other large pelagic predatory fishes. To illustrate the utility of the statistical methods, we show an example using yellowfin tuna in a “test†area in the Pacific Ocean. Stomach-contents data were analyzed using a modified (bagged) classification tree approach, which is being prepared as an R statistical software package. Bulk δ15N values of yellowfin tuna muscle tissue were examined using a Generalized Additive Model, after adjusting for spatial differences in the δ15N values of the baseline primary producers predicted by a global coupled ocean circulation-biogeochemical-isotope model. Both techniques in tandem demonstrated the capacity of this approach to elucidate spatial patterns of variations in both forage species and predator trophic positions and have the potential to predict responses to climate change. We believe this methodology could be extended to all marine top predators. Our results emphasize the necessity for quantitative investigations of global-scale datasets when evaluating changes to the food webs underpinning top ocean predators under long-term climatic variability.