The impacts of altered biogeochemical cycles on ecological systems are likely to vary with trophic level. Predicting how these changes will affect ecological food webs is further complicated by human activities which are simultaneously altering the availability of macronutrients like nitrogen (N) and phosphorus (P), and micronutrients such as sodium (Na). Here we contrast three hypotheses that predict how increasing nutrient availability will shape grassland food webs. We conducted a distributed factorial fertilization experiment (N and P crossed with NaCl) across four North American grasslands, quantifying the responses of aboveground plant biomass and volume, plant tissue and soil elemental concentrations, as well as the abundance of five arthropod functional groups. Fertilization with N and P increased plant biomass and foliar N and P concentrations in grasses but not forbs. Fertilization with Na had no effect on plant biomass but increased foliar Na concentrations. Consistent with the Nutrient Limitation Hypothesis we found strong evidence of nutrient limitation for insect herbivores across the four sites with sucking (phloem and xylem feeding) herbivores increasing in abundance with NP fertilization and chewing herbivores increasing in response to both Na and NP fertilization, and a trend for increased response of arthropods to lower plant nutrient availability. We found no evidence for an interaction of NaCl and NP on arthropod abundance as predicted by the Serial Co-Limitation Hypothesis. Finally, consistent with the Ecosystem Size Hypothesis, predator and parasitoid abundances increased with plant volume, but not fertilization. Our results suggest these functional group-specific responses to changes in plant nutrients and structure are key to predicting the future of grassland food webs in an era with increasing use of N and P fertilizers, and increasing terrestrial inputs of Na from road salt, saline irrigation water, and aerosols due to rising sea levels.

Meta-data includes variable names, descriptions, and data type.