The interplay between top-down and bottom-up processes determines ecosystem productivity. Yet, the factors that mediate the balance between these opposing forces remain poorly understood. Furthering this challenge, complex and often cryptic factors like ecosystem engineering and trait-mediated interactions may play major roles in mediating the outcomes of top-down and bottom-up interactions. In semi-arid grasslands of northeastern China, we conducted a large-scale, three-year experiment to evaluate how soil engineering by ants and plasticity in plants independently and jointly influenced the top-down effects of grazing by a ubiquitous herbivore (cattle) on aboveground standing biomass of the dominant perennial grass, Leymus chinensis. Herbivory had strong top-down effects, reducing L. chinensis AB by 25% relative to baseline levels without cattle or ants. In contrast, soil engineering by ants facilitated weak bottom-up effects in the absence of herbivory. However, in the presence of herbivory, soil engineering effects were strong enough to fully offset herbivore removal of aboveground biomass. This outcome was mediated by L. chinensis’s plasticity in reallocating growth from below- to aboveground biomass, a result linked to additive effects of engineers and herbivores increasing soil N availability and engineering effects improving soil structure. Soil engineering increased soil N by 12%, promoting aboveground biomass. Herbivores increased soil N by 13% via defecation, but this increase failed to offset their reductions in aboveground biomass in isolation. However, when combined, engineers and herbivores increased soil N by 26% and engineers improved soil bulk density, facilitating L. chinensis to shift resource allocations from below- to aboveground biomass sufficiently to fully offset herbivore suppression of aboveground biomass. Our results demonstrate that soil engineering and trait-mediated effects of plant plasticity can strongly mediate the outcome of top-down and bottom-up interactions. These cryptic but perhaps ubiquitous processes may help to explain the long-debated phenomenon of plant compensatory responses to large grazers.