Species interactions are widely thought to be strongest in the tropics, potentially contributing to the greater number of species at lower latitudes. Yet, empirical tests of this "biotic interactions" hypothesis remain limited and often provide mixed results. Here, we analyze 55 years of catch per unit effort data from pelagic longline fisheries to estimate the strength of predation exerted by large predatory fish in the world's oceans. We test two central tenets of the biotic interactions hypothesis: that predation is (1) strongest near the equator, and (2) positively correlated with species richness. Counter to these predictions, we find that predation is (1) strongest in or near the temperate zone and (2) negatively correlated with oceanic fish species richness. These patterns suggest that, at least for pelagic fish predation, common assumptions about the latitudinal distribution of species interactions do not apply, thereby challenging a leading explanation for the latitudinal gradient in species diversity.