Adrian C. Stier

How strongly predators and prey interact is both notoriously context dependent and difficult to measure. Yet across taxa, interaction strength is strongly related to predator size, prey size and prey density, suggesting that general cross‐taxonomic relationships could be used to predict how strongly individual species interact. Here, we ask how accurately do general size‐scaling relationships predict variation in interaction strength between specific species that vary in size and density across space and time? To address this question, we quantified the size and density dependence of the functional response of the California spiny lobster Panulirus interruptus , foraging on a key ecosystem engineer, the purple sea urchin Strongylocentrotus purpuratus , in experimental mesocosms. Based on these results, we then estimated variation in lobster–urchin interaction strength across five sites and 9 years of observational data. Finally, we compared our experimental estimates to predictions based on general size‐scaling relationships from the literature. Our results reveal that predator and prey body size has the greatest effect on interaction strength when prey abundance is high. Due to consistently high urchin densities in the field, our simulations suggest that body size—relative to density—accounted for up to 87% of the spatio‐temporal variation in interaction strength. However, general size‐scaling relationships failed to predict the magnitude of interactions between lobster and urchin; even the best prediction from the literature was, on average, an order of magnitude (+18.7×) different than our experimental predictions. Harvest and climate change are driving reductions in the average body size of many marine species. Anticipating how reductions in body size will alter species interactions is critical to managing marine systems in an ecosystem context. Our results highlight the extent to which differences in size‐frequency distributions can drive dramatic variation in the strength of interactions across narrow spatial and temporal scales. Furthermore, our work suggests that species‐specific estimates for the scaling of interaction strength with body size, rather than general size‐scaling relationships, are necessary to quantitatively predict how reductions in body size will alter interaction strengths.