1.Decades of grazing studies have identified a number of key plant and soil processes affected by large herbivores and how those grazer effects vary among different grassland types. However, there remains little mechanistic understanding about how the effects of grazers on plants and soils may be biogeochemically linked in regulating grassland processes.

        2.Here we measured monthly plant and soil variables, including soil moisture, soil nitrogen (N) availability, plant biomass, shoot N concentration and plant production, in grazed and ungrazed (fenced) grasslands during the 2012-2014 growing seasons. Measurements were used to assess direct and indirect biogeochemical pathways by which grazers influenced net aboveground plant production (NAP) in dry and mesic grasslands in Yellowstone National Park (YNP).

        3.Herbivores only had direct effects on plant variables at the dry grassland compared to direct and indirect effects on both plant and soil variables at the mesic grassland. By enhancing leaf N content at both grasslands, grazers shifted the resource controlling NAP from N in ungrazed grassland to moisture, and potentially phosphorus and/or other soil nutrients, in grazed grassland.

        4.Synthesis. These results indicate the mechanistic linkage between top-down (herbivore) and bottom-up (soil resource) control of grassland production. Changing the resources that limit NAP likely has a profound impact on how grazed vs ungrazed YNP grasslands respond to environmental (e.g., climate, atmospheric N deposition) variability. Because grazing enhances leaf N among many types of grasslands, increasing the sensitivity of plant production to the availability of moisture and nutrients other than N may be a general response of grasslands to grazing.